{
  "best_metric": 2.252807140350342,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.05133140840551813,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00025665704202759064,
      "grad_norm": 0.3664189875125885,
      "learning_rate": 1e-05,
      "loss": 2.3495,
      "step": 1
    },
    {
      "epoch": 0.00025665704202759064,
      "eval_loss": 2.9002344608306885,
      "eval_runtime": 735.0568,
      "eval_samples_per_second": 8.929,
      "eval_steps_per_second": 2.232,
      "step": 1
    },
    {
      "epoch": 0.0005133140840551813,
      "grad_norm": 0.41134214401245117,
      "learning_rate": 2e-05,
      "loss": 2.5502,
      "step": 2
    },
    {
      "epoch": 0.0007699711260827719,
      "grad_norm": 0.44699937105178833,
      "learning_rate": 3e-05,
      "loss": 2.6686,
      "step": 3
    },
    {
      "epoch": 0.0010266281681103626,
      "grad_norm": 0.382760226726532,
      "learning_rate": 4e-05,
      "loss": 2.565,
      "step": 4
    },
    {
      "epoch": 0.0012832852101379532,
      "grad_norm": 0.33547085523605347,
      "learning_rate": 5e-05,
      "loss": 2.4906,
      "step": 5
    },
    {
      "epoch": 0.0015399422521655437,
      "grad_norm": 0.376569002866745,
      "learning_rate": 6e-05,
      "loss": 2.4825,
      "step": 6
    },
    {
      "epoch": 0.0017965992941931343,
      "grad_norm": 0.3572351336479187,
      "learning_rate": 7e-05,
      "loss": 2.5394,
      "step": 7
    },
    {
      "epoch": 0.002053256336220725,
      "grad_norm": 0.37440285086631775,
      "learning_rate": 8e-05,
      "loss": 2.5917,
      "step": 8
    },
    {
      "epoch": 0.0023099133782483157,
      "grad_norm": 0.37039896845817566,
      "learning_rate": 9e-05,
      "loss": 2.4376,
      "step": 9
    },
    {
      "epoch": 0.0025665704202759063,
      "grad_norm": 0.4337705671787262,
      "learning_rate": 0.0001,
      "loss": 2.4979,
      "step": 10
    },
    {
      "epoch": 0.002823227462303497,
      "grad_norm": 0.41142916679382324,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.4636,
      "step": 11
    },
    {
      "epoch": 0.0030798845043310875,
      "grad_norm": 0.43371912837028503,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.5277,
      "step": 12
    },
    {
      "epoch": 0.003336541546358678,
      "grad_norm": 0.47285693883895874,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.5497,
      "step": 13
    },
    {
      "epoch": 0.0035931985883862687,
      "grad_norm": 0.43652060627937317,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.4335,
      "step": 14
    },
    {
      "epoch": 0.0038498556304138597,
      "grad_norm": 0.5904943346977234,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.4055,
      "step": 15
    },
    {
      "epoch": 0.00410651267244145,
      "grad_norm": 0.40625137090682983,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.4177,
      "step": 16
    },
    {
      "epoch": 0.00436316971446904,
      "grad_norm": 0.501611053943634,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.4404,
      "step": 17
    },
    {
      "epoch": 0.0046198267564966315,
      "grad_norm": 0.4470362067222595,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4248,
      "step": 18
    },
    {
      "epoch": 0.004876483798524222,
      "grad_norm": 0.42792126536369324,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.4861,
      "step": 19
    },
    {
      "epoch": 0.005133140840551813,
      "grad_norm": 0.44940832257270813,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.4652,
      "step": 20
    },
    {
      "epoch": 0.005389797882579404,
      "grad_norm": 0.4557294547557831,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.2774,
      "step": 21
    },
    {
      "epoch": 0.005646454924606994,
      "grad_norm": 0.4595181345939636,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3819,
      "step": 22
    },
    {
      "epoch": 0.005903111966634585,
      "grad_norm": 0.441147118806839,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.3821,
      "step": 23
    },
    {
      "epoch": 0.006159769008662175,
      "grad_norm": 0.4458269774913788,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.4026,
      "step": 24
    },
    {
      "epoch": 0.006416426050689766,
      "grad_norm": 0.4557860493659973,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.411,
      "step": 25
    },
    {
      "epoch": 0.006673083092717356,
      "grad_norm": 0.46162599325180054,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.3747,
      "step": 26
    },
    {
      "epoch": 0.006929740134744947,
      "grad_norm": 0.48158013820648193,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.4596,
      "step": 27
    },
    {
      "epoch": 0.007186397176772537,
      "grad_norm": 0.4900691509246826,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.4275,
      "step": 28
    },
    {
      "epoch": 0.007443054218800128,
      "grad_norm": 0.514424204826355,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3373,
      "step": 29
    },
    {
      "epoch": 0.007699711260827719,
      "grad_norm": 0.5473058819770813,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.3755,
      "step": 30
    },
    {
      "epoch": 0.00795636830285531,
      "grad_norm": 0.4835929870605469,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.3049,
      "step": 31
    },
    {
      "epoch": 0.0082130253448829,
      "grad_norm": 0.48676031827926636,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.2379,
      "step": 32
    },
    {
      "epoch": 0.00846968238691049,
      "grad_norm": 0.5120344758033752,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.353,
      "step": 33
    },
    {
      "epoch": 0.00872633942893808,
      "grad_norm": 0.5717011094093323,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.3785,
      "step": 34
    },
    {
      "epoch": 0.008982996470965673,
      "grad_norm": 0.5246809124946594,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2561,
      "step": 35
    },
    {
      "epoch": 0.009239653512993263,
      "grad_norm": 0.6063455939292908,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.3989,
      "step": 36
    },
    {
      "epoch": 0.009496310555020853,
      "grad_norm": 0.513489305973053,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.2743,
      "step": 37
    },
    {
      "epoch": 0.009752967597048443,
      "grad_norm": 0.5698059797286987,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.3314,
      "step": 38
    },
    {
      "epoch": 0.010009624639076035,
      "grad_norm": 0.5455905795097351,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.2257,
      "step": 39
    },
    {
      "epoch": 0.010266281681103625,
      "grad_norm": 0.5889667868614197,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.3345,
      "step": 40
    },
    {
      "epoch": 0.010522938723131215,
      "grad_norm": 0.6319016814231873,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.4437,
      "step": 41
    },
    {
      "epoch": 0.010779595765158807,
      "grad_norm": 0.5464950799942017,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.1975,
      "step": 42
    },
    {
      "epoch": 0.011036252807186397,
      "grad_norm": 0.5818386673927307,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.2918,
      "step": 43
    },
    {
      "epoch": 0.011292909849213988,
      "grad_norm": 0.6758567094802856,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.2487,
      "step": 44
    },
    {
      "epoch": 0.011549566891241578,
      "grad_norm": 0.6282801032066345,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.1841,
      "step": 45
    },
    {
      "epoch": 0.01180622393326917,
      "grad_norm": 0.6689670085906982,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.1653,
      "step": 46
    },
    {
      "epoch": 0.01206288097529676,
      "grad_norm": 0.7425477504730225,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.1696,
      "step": 47
    },
    {
      "epoch": 0.01231953801732435,
      "grad_norm": 0.7808821201324463,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.31,
      "step": 48
    },
    {
      "epoch": 0.01257619505935194,
      "grad_norm": 0.7420758605003357,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.2314,
      "step": 49
    },
    {
      "epoch": 0.012832852101379532,
      "grad_norm": 0.8017773032188416,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.0436,
      "step": 50
    },
    {
      "epoch": 0.012832852101379532,
      "eval_loss": 2.351137638092041,
      "eval_runtime": 739.2463,
      "eval_samples_per_second": 8.878,
      "eval_steps_per_second": 2.22,
      "step": 50
    },
    {
      "epoch": 0.013089509143407122,
      "grad_norm": 1.151561975479126,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.389,
      "step": 51
    },
    {
      "epoch": 0.013346166185434712,
      "grad_norm": 0.8239222168922424,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.4156,
      "step": 52
    },
    {
      "epoch": 0.013602823227462304,
      "grad_norm": 0.29609519243240356,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.326,
      "step": 53
    },
    {
      "epoch": 0.013859480269489894,
      "grad_norm": 0.3043610751628876,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.3231,
      "step": 54
    },
    {
      "epoch": 0.014116137311517485,
      "grad_norm": 0.3249354660511017,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.3588,
      "step": 55
    },
    {
      "epoch": 0.014372794353545075,
      "grad_norm": 0.3120054602622986,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.3624,
      "step": 56
    },
    {
      "epoch": 0.014629451395572667,
      "grad_norm": 0.3138444721698761,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3748,
      "step": 57
    },
    {
      "epoch": 0.014886108437600257,
      "grad_norm": 0.3040318787097931,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.3719,
      "step": 58
    },
    {
      "epoch": 0.015142765479627847,
      "grad_norm": 0.3202269971370697,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.4432,
      "step": 59
    },
    {
      "epoch": 0.015399422521655439,
      "grad_norm": 0.33661895990371704,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.3332,
      "step": 60
    },
    {
      "epoch": 0.015656079563683027,
      "grad_norm": 0.30936452746391296,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.2801,
      "step": 61
    },
    {
      "epoch": 0.01591273660571062,
      "grad_norm": 0.3115326762199402,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.3606,
      "step": 62
    },
    {
      "epoch": 0.01616939364773821,
      "grad_norm": 0.3484363257884979,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.3108,
      "step": 63
    },
    {
      "epoch": 0.0164260506897658,
      "grad_norm": 0.32701075077056885,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.3629,
      "step": 64
    },
    {
      "epoch": 0.01668270773179339,
      "grad_norm": 0.3241425156593323,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.3118,
      "step": 65
    },
    {
      "epoch": 0.01693936477382098,
      "grad_norm": 0.3512122333049774,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3702,
      "step": 66
    },
    {
      "epoch": 0.01719602181584857,
      "grad_norm": 0.35062772035598755,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.3604,
      "step": 67
    },
    {
      "epoch": 0.01745267885787616,
      "grad_norm": 0.36103612184524536,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.381,
      "step": 68
    },
    {
      "epoch": 0.017709335899903755,
      "grad_norm": 0.3503100574016571,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.2989,
      "step": 69
    },
    {
      "epoch": 0.017965992941931345,
      "grad_norm": 0.36472007632255554,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.377,
      "step": 70
    },
    {
      "epoch": 0.018222649983958936,
      "grad_norm": 0.38080689311027527,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.3344,
      "step": 71
    },
    {
      "epoch": 0.018479307025986526,
      "grad_norm": 0.37981727719306946,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.4158,
      "step": 72
    },
    {
      "epoch": 0.018735964068014116,
      "grad_norm": 0.38469165563583374,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.3487,
      "step": 73
    },
    {
      "epoch": 0.018992621110041706,
      "grad_norm": 0.3890038728713989,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.2574,
      "step": 74
    },
    {
      "epoch": 0.019249278152069296,
      "grad_norm": 0.40403974056243896,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.2977,
      "step": 75
    },
    {
      "epoch": 0.019505935194096886,
      "grad_norm": 0.4120301902294159,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.3054,
      "step": 76
    },
    {
      "epoch": 0.01976259223612448,
      "grad_norm": 0.439554899930954,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.276,
      "step": 77
    },
    {
      "epoch": 0.02001924927815207,
      "grad_norm": 0.5092542767524719,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.3156,
      "step": 78
    },
    {
      "epoch": 0.02027590632017966,
      "grad_norm": 0.4996526539325714,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.3464,
      "step": 79
    },
    {
      "epoch": 0.02053256336220725,
      "grad_norm": 0.4822778105735779,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.3555,
      "step": 80
    },
    {
      "epoch": 0.02078922040423484,
      "grad_norm": 0.4535146653652191,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.2672,
      "step": 81
    },
    {
      "epoch": 0.02104587744626243,
      "grad_norm": 0.4873507022857666,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.2601,
      "step": 82
    },
    {
      "epoch": 0.02130253448829002,
      "grad_norm": 0.4486106336116791,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.2825,
      "step": 83
    },
    {
      "epoch": 0.021559191530317615,
      "grad_norm": 0.47596490383148193,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.29,
      "step": 84
    },
    {
      "epoch": 0.021815848572345205,
      "grad_norm": 0.5172278881072998,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.3019,
      "step": 85
    },
    {
      "epoch": 0.022072505614372795,
      "grad_norm": 0.532233476638794,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.1487,
      "step": 86
    },
    {
      "epoch": 0.022329162656400385,
      "grad_norm": 0.4997048079967499,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.2716,
      "step": 87
    },
    {
      "epoch": 0.022585819698427975,
      "grad_norm": 0.49617281556129456,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.275,
      "step": 88
    },
    {
      "epoch": 0.022842476740455565,
      "grad_norm": 0.4903511106967926,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.0744,
      "step": 89
    },
    {
      "epoch": 0.023099133782483156,
      "grad_norm": 0.4909917116165161,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.2088,
      "step": 90
    },
    {
      "epoch": 0.02335579082451075,
      "grad_norm": 0.5346497893333435,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.2042,
      "step": 91
    },
    {
      "epoch": 0.02361244786653834,
      "grad_norm": 0.5274331569671631,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.2326,
      "step": 92
    },
    {
      "epoch": 0.02386910490856593,
      "grad_norm": 0.5872141122817993,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.2687,
      "step": 93
    },
    {
      "epoch": 0.02412576195059352,
      "grad_norm": 0.5494994521141052,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.0829,
      "step": 94
    },
    {
      "epoch": 0.02438241899262111,
      "grad_norm": 0.6368223428726196,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.2188,
      "step": 95
    },
    {
      "epoch": 0.0246390760346487,
      "grad_norm": 0.5976985692977905,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.2644,
      "step": 96
    },
    {
      "epoch": 0.02489573307667629,
      "grad_norm": 0.5988062620162964,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.0768,
      "step": 97
    },
    {
      "epoch": 0.02515239011870388,
      "grad_norm": 0.6056662201881409,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.1045,
      "step": 98
    },
    {
      "epoch": 0.025409047160731474,
      "grad_norm": 0.6570653319358826,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.1288,
      "step": 99
    },
    {
      "epoch": 0.025665704202759064,
      "grad_norm": 0.7626357078552246,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.0981,
      "step": 100
    },
    {
      "epoch": 0.025665704202759064,
      "eval_loss": 2.29904842376709,
      "eval_runtime": 738.3568,
      "eval_samples_per_second": 8.889,
      "eval_steps_per_second": 2.223,
      "step": 100
    },
    {
      "epoch": 0.025922361244786654,
      "grad_norm": 0.6772850155830383,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.3253,
      "step": 101
    },
    {
      "epoch": 0.026179018286814244,
      "grad_norm": 0.7162484526634216,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.3223,
      "step": 102
    },
    {
      "epoch": 0.026435675328841834,
      "grad_norm": 0.6093961596488953,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.4454,
      "step": 103
    },
    {
      "epoch": 0.026692332370869425,
      "grad_norm": 0.4366320073604584,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.2957,
      "step": 104
    },
    {
      "epoch": 0.026948989412897015,
      "grad_norm": 0.2959626019001007,
      "learning_rate": 5e-05,
      "loss": 2.3535,
      "step": 105
    },
    {
      "epoch": 0.02720564645492461,
      "grad_norm": 0.30287203192710876,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.3967,
      "step": 106
    },
    {
      "epoch": 0.0274623034969522,
      "grad_norm": 0.3101503551006317,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.23,
      "step": 107
    },
    {
      "epoch": 0.02771896053897979,
      "grad_norm": 0.31346189975738525,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.2443,
      "step": 108
    },
    {
      "epoch": 0.02797561758100738,
      "grad_norm": 0.3116395175457001,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.353,
      "step": 109
    },
    {
      "epoch": 0.02823227462303497,
      "grad_norm": 0.32071176171302795,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.3093,
      "step": 110
    },
    {
      "epoch": 0.02848893166506256,
      "grad_norm": 0.32827410101890564,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.3656,
      "step": 111
    },
    {
      "epoch": 0.02874558870709015,
      "grad_norm": 0.3322199583053589,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.3787,
      "step": 112
    },
    {
      "epoch": 0.029002245749117743,
      "grad_norm": 0.3370152413845062,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.3795,
      "step": 113
    },
    {
      "epoch": 0.029258902791145333,
      "grad_norm": 0.33916231989860535,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.2443,
      "step": 114
    },
    {
      "epoch": 0.029515559833172923,
      "grad_norm": 0.3768143951892853,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.4304,
      "step": 115
    },
    {
      "epoch": 0.029772216875200513,
      "grad_norm": 0.36799487471580505,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.307,
      "step": 116
    },
    {
      "epoch": 0.030028873917228104,
      "grad_norm": 0.3804920017719269,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.343,
      "step": 117
    },
    {
      "epoch": 0.030285530959255694,
      "grad_norm": 0.37816157937049866,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.2698,
      "step": 118
    },
    {
      "epoch": 0.030542188001283284,
      "grad_norm": 0.37779971957206726,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.2893,
      "step": 119
    },
    {
      "epoch": 0.030798845043310877,
      "grad_norm": 0.40218082070350647,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.3404,
      "step": 120
    },
    {
      "epoch": 0.031055502085338468,
      "grad_norm": 0.38081374764442444,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.3014,
      "step": 121
    },
    {
      "epoch": 0.031312159127366054,
      "grad_norm": 0.3967231512069702,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.2787,
      "step": 122
    },
    {
      "epoch": 0.031568816169393644,
      "grad_norm": 0.39703407883644104,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.3104,
      "step": 123
    },
    {
      "epoch": 0.03182547321142124,
      "grad_norm": 0.4294150769710541,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.2901,
      "step": 124
    },
    {
      "epoch": 0.03208213025344883,
      "grad_norm": 0.40045395493507385,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.2867,
      "step": 125
    },
    {
      "epoch": 0.03233878729547642,
      "grad_norm": 0.3969850540161133,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.2952,
      "step": 126
    },
    {
      "epoch": 0.03259544433750401,
      "grad_norm": 0.41207626461982727,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.2793,
      "step": 127
    },
    {
      "epoch": 0.0328521013795316,
      "grad_norm": 0.4165365993976593,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.2544,
      "step": 128
    },
    {
      "epoch": 0.03310875842155919,
      "grad_norm": 0.43575000762939453,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.2314,
      "step": 129
    },
    {
      "epoch": 0.03336541546358678,
      "grad_norm": 0.4352191984653473,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.3342,
      "step": 130
    },
    {
      "epoch": 0.03362207250561437,
      "grad_norm": 0.439800500869751,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.3201,
      "step": 131
    },
    {
      "epoch": 0.03387872954764196,
      "grad_norm": 0.4738902449607849,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.2236,
      "step": 132
    },
    {
      "epoch": 0.03413538658966955,
      "grad_norm": 0.4956141710281372,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.2705,
      "step": 133
    },
    {
      "epoch": 0.03439204363169714,
      "grad_norm": 0.5049107670783997,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2448,
      "step": 134
    },
    {
      "epoch": 0.03464870067372473,
      "grad_norm": 0.48702457547187805,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.0647,
      "step": 135
    },
    {
      "epoch": 0.03490535771575232,
      "grad_norm": 0.5005754232406616,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.2033,
      "step": 136
    },
    {
      "epoch": 0.035162014757779914,
      "grad_norm": 0.4950476884841919,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.1552,
      "step": 137
    },
    {
      "epoch": 0.03541867179980751,
      "grad_norm": 0.5001799464225769,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.1576,
      "step": 138
    },
    {
      "epoch": 0.0356753288418351,
      "grad_norm": 0.547107458114624,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.2176,
      "step": 139
    },
    {
      "epoch": 0.03593198588386269,
      "grad_norm": 0.5037182569503784,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.3269,
      "step": 140
    },
    {
      "epoch": 0.03618864292589028,
      "grad_norm": 0.5539939403533936,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.18,
      "step": 141
    },
    {
      "epoch": 0.03644529996791787,
      "grad_norm": 0.5221781730651855,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1957,
      "step": 142
    },
    {
      "epoch": 0.03670195700994546,
      "grad_norm": 0.5297254920005798,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.2451,
      "step": 143
    },
    {
      "epoch": 0.03695861405197305,
      "grad_norm": 0.5830360651016235,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.2241,
      "step": 144
    },
    {
      "epoch": 0.03721527109400064,
      "grad_norm": 0.5511125326156616,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.2624,
      "step": 145
    },
    {
      "epoch": 0.03747192813602823,
      "grad_norm": 0.6714289784431458,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.2162,
      "step": 146
    },
    {
      "epoch": 0.03772858517805582,
      "grad_norm": 0.6063690781593323,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.0669,
      "step": 147
    },
    {
      "epoch": 0.03798524222008341,
      "grad_norm": 0.6806104183197021,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.1953,
      "step": 148
    },
    {
      "epoch": 0.038241899262111,
      "grad_norm": 0.6930797100067139,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.2899,
      "step": 149
    },
    {
      "epoch": 0.03849855630413859,
      "grad_norm": 0.8597152233123779,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.2818,
      "step": 150
    },
    {
      "epoch": 0.03849855630413859,
      "eval_loss": 2.26181960105896,
      "eval_runtime": 738.479,
      "eval_samples_per_second": 8.887,
      "eval_steps_per_second": 2.222,
      "step": 150
    },
    {
      "epoch": 0.03875521334616618,
      "grad_norm": 0.3422974646091461,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.2933,
      "step": 151
    },
    {
      "epoch": 0.03901187038819377,
      "grad_norm": 0.4707757830619812,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.3283,
      "step": 152
    },
    {
      "epoch": 0.03926852743022137,
      "grad_norm": 0.4521164894104004,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.3406,
      "step": 153
    },
    {
      "epoch": 0.03952518447224896,
      "grad_norm": 0.44188806414604187,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.2193,
      "step": 154
    },
    {
      "epoch": 0.03978184151427655,
      "grad_norm": 0.502597451210022,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.3505,
      "step": 155
    },
    {
      "epoch": 0.04003849855630414,
      "grad_norm": 0.450508713722229,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.3172,
      "step": 156
    },
    {
      "epoch": 0.04029515559833173,
      "grad_norm": 0.3949015736579895,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.3929,
      "step": 157
    },
    {
      "epoch": 0.04055181264035932,
      "grad_norm": 0.3617224097251892,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.3032,
      "step": 158
    },
    {
      "epoch": 0.04080846968238691,
      "grad_norm": 0.36014440655708313,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.2744,
      "step": 159
    },
    {
      "epoch": 0.0410651267244145,
      "grad_norm": 0.3576134443283081,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.3233,
      "step": 160
    },
    {
      "epoch": 0.04132178376644209,
      "grad_norm": 0.31801852583885193,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.2295,
      "step": 161
    },
    {
      "epoch": 0.04157844080846968,
      "grad_norm": 0.3260203003883362,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.3959,
      "step": 162
    },
    {
      "epoch": 0.04183509785049727,
      "grad_norm": 0.31979161500930786,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.1995,
      "step": 163
    },
    {
      "epoch": 0.04209175489252486,
      "grad_norm": 0.3350762128829956,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.3218,
      "step": 164
    },
    {
      "epoch": 0.04234841193455245,
      "grad_norm": 0.33540862798690796,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.3084,
      "step": 165
    },
    {
      "epoch": 0.04260506897658004,
      "grad_norm": 0.33332133293151855,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.2475,
      "step": 166
    },
    {
      "epoch": 0.04286172601860763,
      "grad_norm": 0.33638399839401245,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.275,
      "step": 167
    },
    {
      "epoch": 0.04311838306063523,
      "grad_norm": 0.38856881856918335,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2398,
      "step": 168
    },
    {
      "epoch": 0.04337504010266282,
      "grad_norm": 0.3757133185863495,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.3224,
      "step": 169
    },
    {
      "epoch": 0.04363169714469041,
      "grad_norm": 0.4081624448299408,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.3078,
      "step": 170
    },
    {
      "epoch": 0.043888354186718,
      "grad_norm": 0.3590552806854248,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.2531,
      "step": 171
    },
    {
      "epoch": 0.04414501122874559,
      "grad_norm": 0.38641491532325745,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.268,
      "step": 172
    },
    {
      "epoch": 0.04440166827077318,
      "grad_norm": 0.396953821182251,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.2258,
      "step": 173
    },
    {
      "epoch": 0.04465832531280077,
      "grad_norm": 0.400569349527359,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.2044,
      "step": 174
    },
    {
      "epoch": 0.04491498235482836,
      "grad_norm": 0.394077330827713,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.2371,
      "step": 175
    },
    {
      "epoch": 0.04517163939685595,
      "grad_norm": 0.38696473836898804,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.1553,
      "step": 176
    },
    {
      "epoch": 0.04542829643888354,
      "grad_norm": 0.4218221604824066,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.3042,
      "step": 177
    },
    {
      "epoch": 0.04568495348091113,
      "grad_norm": 0.4155186414718628,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.2698,
      "step": 178
    },
    {
      "epoch": 0.04594161052293872,
      "grad_norm": 0.4689392149448395,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.2759,
      "step": 179
    },
    {
      "epoch": 0.04619826756496631,
      "grad_norm": 0.4387744963169098,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2866,
      "step": 180
    },
    {
      "epoch": 0.0464549246069939,
      "grad_norm": 0.45942121744155884,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.2913,
      "step": 181
    },
    {
      "epoch": 0.0467115816490215,
      "grad_norm": 0.4631074368953705,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.2984,
      "step": 182
    },
    {
      "epoch": 0.04696823869104909,
      "grad_norm": 0.47069424390792847,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.1726,
      "step": 183
    },
    {
      "epoch": 0.04722489573307668,
      "grad_norm": 0.49852561950683594,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.2916,
      "step": 184
    },
    {
      "epoch": 0.04748155277510427,
      "grad_norm": 0.5104568004608154,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.2821,
      "step": 185
    },
    {
      "epoch": 0.04773820981713186,
      "grad_norm": 0.47923555970191956,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2352,
      "step": 186
    },
    {
      "epoch": 0.04799486685915945,
      "grad_norm": 0.5244100093841553,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.1826,
      "step": 187
    },
    {
      "epoch": 0.04825152390118704,
      "grad_norm": 0.5392053127288818,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.3235,
      "step": 188
    },
    {
      "epoch": 0.04850818094321463,
      "grad_norm": 0.5164491534233093,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.2133,
      "step": 189
    },
    {
      "epoch": 0.04876483798524222,
      "grad_norm": 0.5362449884414673,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.247,
      "step": 190
    },
    {
      "epoch": 0.04902149502726981,
      "grad_norm": 0.5653479695320129,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.2328,
      "step": 191
    },
    {
      "epoch": 0.0492781520692974,
      "grad_norm": 0.572383463382721,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.243,
      "step": 192
    },
    {
      "epoch": 0.04953480911132499,
      "grad_norm": 0.6023736000061035,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.2308,
      "step": 193
    },
    {
      "epoch": 0.04979146615335258,
      "grad_norm": 0.606105387210846,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.2906,
      "step": 194
    },
    {
      "epoch": 0.05004812319538017,
      "grad_norm": 0.6184576749801636,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.3062,
      "step": 195
    },
    {
      "epoch": 0.05030478023740776,
      "grad_norm": 0.6309393048286438,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.255,
      "step": 196
    },
    {
      "epoch": 0.05056143727943536,
      "grad_norm": 0.6904622316360474,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.2363,
      "step": 197
    },
    {
      "epoch": 0.05081809432146295,
      "grad_norm": 0.7422359585762024,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.2325,
      "step": 198
    },
    {
      "epoch": 0.05107475136349054,
      "grad_norm": 0.7413392066955566,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.2729,
      "step": 199
    },
    {
      "epoch": 0.05133140840551813,
      "grad_norm": 0.8726180195808411,
      "learning_rate": 0.0,
      "loss": 2.0746,
      "step": 200
    },
    {
      "epoch": 0.05133140840551813,
      "eval_loss": 2.252807140350342,
      "eval_runtime": 736.7405,
      "eval_samples_per_second": 8.908,
      "eval_steps_per_second": 2.227,
      "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": 2.7344078169440256e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}