{
  "best_metric": 1.3743650913238525,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.08122652045892984,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004061326022946492,
      "grad_norm": 1.3459707498550415,
      "learning_rate": 1e-05,
      "loss": 1.5815,
      "step": 1
    },
    {
      "epoch": 0.0004061326022946492,
      "eval_loss": 2.1705708503723145,
      "eval_runtime": 307.0362,
      "eval_samples_per_second": 13.507,
      "eval_steps_per_second": 3.377,
      "step": 1
    },
    {
      "epoch": 0.0008122652045892984,
      "grad_norm": 1.445604681968689,
      "learning_rate": 2e-05,
      "loss": 1.7408,
      "step": 2
    },
    {
      "epoch": 0.0012183978068839476,
      "grad_norm": 1.4932692050933838,
      "learning_rate": 3e-05,
      "loss": 1.828,
      "step": 3
    },
    {
      "epoch": 0.0016245304091785967,
      "grad_norm": 1.3266582489013672,
      "learning_rate": 4e-05,
      "loss": 1.8319,
      "step": 4
    },
    {
      "epoch": 0.0020306630114732462,
      "grad_norm": 1.1805715560913086,
      "learning_rate": 5e-05,
      "loss": 1.7692,
      "step": 5
    },
    {
      "epoch": 0.0024367956137678953,
      "grad_norm": 1.1116411685943604,
      "learning_rate": 6e-05,
      "loss": 1.6888,
      "step": 6
    },
    {
      "epoch": 0.0028429282160625444,
      "grad_norm": 1.1109813451766968,
      "learning_rate": 7e-05,
      "loss": 1.6346,
      "step": 7
    },
    {
      "epoch": 0.0032490608183571934,
      "grad_norm": 1.1213299036026,
      "learning_rate": 8e-05,
      "loss": 1.5795,
      "step": 8
    },
    {
      "epoch": 0.003655193420651843,
      "grad_norm": 1.517636775970459,
      "learning_rate": 9e-05,
      "loss": 1.6551,
      "step": 9
    },
    {
      "epoch": 0.0040613260229464924,
      "grad_norm": 0.8936739563941956,
      "learning_rate": 0.0001,
      "loss": 1.5318,
      "step": 10
    },
    {
      "epoch": 0.0044674586252411415,
      "grad_norm": 0.8239598274230957,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5003,
      "step": 11
    },
    {
      "epoch": 0.004873591227535791,
      "grad_norm": 0.8810031414031982,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5238,
      "step": 12
    },
    {
      "epoch": 0.00527972382983044,
      "grad_norm": 0.762035071849823,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4975,
      "step": 13
    },
    {
      "epoch": 0.005685856432125089,
      "grad_norm": 0.6892722249031067,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.4336,
      "step": 14
    },
    {
      "epoch": 0.006091989034419738,
      "grad_norm": 0.7215124368667603,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4898,
      "step": 15
    },
    {
      "epoch": 0.006498121636714387,
      "grad_norm": 0.6996978521347046,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5738,
      "step": 16
    },
    {
      "epoch": 0.006904254239009037,
      "grad_norm": 0.7445999383926392,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5072,
      "step": 17
    },
    {
      "epoch": 0.007310386841303686,
      "grad_norm": 0.6681432127952576,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4469,
      "step": 18
    },
    {
      "epoch": 0.007716519443598335,
      "grad_norm": 0.7040372490882874,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.4469,
      "step": 19
    },
    {
      "epoch": 0.008122652045892985,
      "grad_norm": 0.7005623579025269,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.4524,
      "step": 20
    },
    {
      "epoch": 0.008528784648187633,
      "grad_norm": 0.7213242053985596,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4383,
      "step": 21
    },
    {
      "epoch": 0.008934917250482283,
      "grad_norm": 0.7225236296653748,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4274,
      "step": 22
    },
    {
      "epoch": 0.009341049852776931,
      "grad_norm": 0.6702327132225037,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.4264,
      "step": 23
    },
    {
      "epoch": 0.009747182455071581,
      "grad_norm": 0.809352695941925,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.386,
      "step": 24
    },
    {
      "epoch": 0.01015331505736623,
      "grad_norm": 0.7973009943962097,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4209,
      "step": 25
    },
    {
      "epoch": 0.01055944765966088,
      "grad_norm": 0.8504800200462341,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.4312,
      "step": 26
    },
    {
      "epoch": 0.01096558026195553,
      "grad_norm": 0.8228721022605896,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4466,
      "step": 27
    },
    {
      "epoch": 0.011371712864250177,
      "grad_norm": 0.8575668334960938,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.3192,
      "step": 28
    },
    {
      "epoch": 0.011777845466544827,
      "grad_norm": 0.8672021627426147,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.557,
      "step": 29
    },
    {
      "epoch": 0.012183978068839476,
      "grad_norm": 0.8064315319061279,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3426,
      "step": 30
    },
    {
      "epoch": 0.012590110671134126,
      "grad_norm": 0.9522659778594971,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.488,
      "step": 31
    },
    {
      "epoch": 0.012996243273428774,
      "grad_norm": 0.8255912065505981,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3874,
      "step": 32
    },
    {
      "epoch": 0.013402375875723424,
      "grad_norm": 0.8663228750228882,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.3202,
      "step": 33
    },
    {
      "epoch": 0.013808508478018074,
      "grad_norm": 0.9034584760665894,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3119,
      "step": 34
    },
    {
      "epoch": 0.014214641080312722,
      "grad_norm": 0.8480425477027893,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.1869,
      "step": 35
    },
    {
      "epoch": 0.014620773682607372,
      "grad_norm": 0.9139435887336731,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.2952,
      "step": 36
    },
    {
      "epoch": 0.01502690628490202,
      "grad_norm": 0.9090644121170044,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3284,
      "step": 37
    },
    {
      "epoch": 0.01543303888719667,
      "grad_norm": 0.9656711220741272,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.366,
      "step": 38
    },
    {
      "epoch": 0.015839171489491318,
      "grad_norm": 0.943183958530426,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.2349,
      "step": 39
    },
    {
      "epoch": 0.01624530409178597,
      "grad_norm": 1.0320135354995728,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.2602,
      "step": 40
    },
    {
      "epoch": 0.016651436694080618,
      "grad_norm": 1.0319671630859375,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3153,
      "step": 41
    },
    {
      "epoch": 0.017057569296375266,
      "grad_norm": 0.944151759147644,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3717,
      "step": 42
    },
    {
      "epoch": 0.017463701898669914,
      "grad_norm": 0.9808139801025391,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.2681,
      "step": 43
    },
    {
      "epoch": 0.017869834500964566,
      "grad_norm": 1.066519021987915,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.2371,
      "step": 44
    },
    {
      "epoch": 0.018275967103259214,
      "grad_norm": 1.1237547397613525,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3255,
      "step": 45
    },
    {
      "epoch": 0.018682099705553862,
      "grad_norm": 0.9697301387786865,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1256,
      "step": 46
    },
    {
      "epoch": 0.019088232307848514,
      "grad_norm": 1.1894623041152954,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5102,
      "step": 47
    },
    {
      "epoch": 0.019494364910143162,
      "grad_norm": 1.3104721307754517,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3807,
      "step": 48
    },
    {
      "epoch": 0.01990049751243781,
      "grad_norm": 2.2278523445129395,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.5591,
      "step": 49
    },
    {
      "epoch": 0.02030663011473246,
      "grad_norm": 3.040036916732788,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.7211,
      "step": 50
    },
    {
      "epoch": 0.02030663011473246,
      "eval_loss": 1.9163098335266113,
      "eval_runtime": 309.92,
      "eval_samples_per_second": 13.381,
      "eval_steps_per_second": 3.346,
      "step": 50
    },
    {
      "epoch": 0.02071276271702711,
      "grad_norm": 2.5100488662719727,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9305,
      "step": 51
    },
    {
      "epoch": 0.02111889531932176,
      "grad_norm": 2.1271471977233887,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.8833,
      "step": 52
    },
    {
      "epoch": 0.021525027921616407,
      "grad_norm": 1.6900986433029175,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.8019,
      "step": 53
    },
    {
      "epoch": 0.02193116052391106,
      "grad_norm": 1.3168630599975586,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6667,
      "step": 54
    },
    {
      "epoch": 0.022337293126205707,
      "grad_norm": 0.7889320254325867,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5371,
      "step": 55
    },
    {
      "epoch": 0.022743425728500355,
      "grad_norm": 0.8151615262031555,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5597,
      "step": 56
    },
    {
      "epoch": 0.023149558330795003,
      "grad_norm": 0.9233412146568298,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5154,
      "step": 57
    },
    {
      "epoch": 0.023555690933089655,
      "grad_norm": 0.8535129427909851,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4912,
      "step": 58
    },
    {
      "epoch": 0.023961823535384303,
      "grad_norm": 0.736276388168335,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.542,
      "step": 59
    },
    {
      "epoch": 0.02436795613767895,
      "grad_norm": 0.733509361743927,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5942,
      "step": 60
    },
    {
      "epoch": 0.024774088739973603,
      "grad_norm": 0.726263165473938,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4467,
      "step": 61
    },
    {
      "epoch": 0.02518022134226825,
      "grad_norm": 0.6740058660507202,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5055,
      "step": 62
    },
    {
      "epoch": 0.0255863539445629,
      "grad_norm": 0.6746672987937927,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4322,
      "step": 63
    },
    {
      "epoch": 0.025992486546857548,
      "grad_norm": 0.6926741003990173,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5049,
      "step": 64
    },
    {
      "epoch": 0.0263986191491522,
      "grad_norm": 0.6756122708320618,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.499,
      "step": 65
    },
    {
      "epoch": 0.026804751751446847,
      "grad_norm": 0.6307368874549866,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3963,
      "step": 66
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.6332759261131287,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3711,
      "step": 67
    },
    {
      "epoch": 0.027617016956036147,
      "grad_norm": 0.7248176336288452,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.438,
      "step": 68
    },
    {
      "epoch": 0.028023149558330795,
      "grad_norm": 0.764191210269928,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4231,
      "step": 69
    },
    {
      "epoch": 0.028429282160625444,
      "grad_norm": 0.7459575533866882,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4424,
      "step": 70
    },
    {
      "epoch": 0.028835414762920092,
      "grad_norm": 0.7871569395065308,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4006,
      "step": 71
    },
    {
      "epoch": 0.029241547365214744,
      "grad_norm": 0.7043696641921997,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3879,
      "step": 72
    },
    {
      "epoch": 0.029647679967509392,
      "grad_norm": 0.7912059426307678,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4553,
      "step": 73
    },
    {
      "epoch": 0.03005381256980404,
      "grad_norm": 0.7687498927116394,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4772,
      "step": 74
    },
    {
      "epoch": 0.03045994517209869,
      "grad_norm": 0.7192708253860474,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.3879,
      "step": 75
    },
    {
      "epoch": 0.03086607777439334,
      "grad_norm": 0.6956753730773926,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4224,
      "step": 76
    },
    {
      "epoch": 0.03127221037668799,
      "grad_norm": 0.803547739982605,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3905,
      "step": 77
    },
    {
      "epoch": 0.031678342978982636,
      "grad_norm": 0.7080717086791992,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.2997,
      "step": 78
    },
    {
      "epoch": 0.032084475581277284,
      "grad_norm": 0.7270420789718628,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3861,
      "step": 79
    },
    {
      "epoch": 0.03249060818357194,
      "grad_norm": 0.7509323358535767,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3482,
      "step": 80
    },
    {
      "epoch": 0.03289674078586659,
      "grad_norm": 0.8757091164588928,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3756,
      "step": 81
    },
    {
      "epoch": 0.033302873388161236,
      "grad_norm": 0.8361058831214905,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3912,
      "step": 82
    },
    {
      "epoch": 0.033709005990455884,
      "grad_norm": 0.8529960513114929,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2062,
      "step": 83
    },
    {
      "epoch": 0.03411513859275053,
      "grad_norm": 0.9276316165924072,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3142,
      "step": 84
    },
    {
      "epoch": 0.03452127119504518,
      "grad_norm": 0.851285994052887,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3927,
      "step": 85
    },
    {
      "epoch": 0.03492740379733983,
      "grad_norm": 0.9086071252822876,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2213,
      "step": 86
    },
    {
      "epoch": 0.035333536399634484,
      "grad_norm": 0.9446242451667786,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.3843,
      "step": 87
    },
    {
      "epoch": 0.03573966900192913,
      "grad_norm": 0.8604442477226257,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1815,
      "step": 88
    },
    {
      "epoch": 0.03614580160422378,
      "grad_norm": 0.8646284937858582,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3559,
      "step": 89
    },
    {
      "epoch": 0.03655193420651843,
      "grad_norm": 0.9388222098350525,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3209,
      "step": 90
    },
    {
      "epoch": 0.03695806680881308,
      "grad_norm": 0.8788503408432007,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3472,
      "step": 91
    },
    {
      "epoch": 0.037364199411107725,
      "grad_norm": 0.9233901500701904,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.263,
      "step": 92
    },
    {
      "epoch": 0.03777033201340237,
      "grad_norm": 0.9628087282180786,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.203,
      "step": 93
    },
    {
      "epoch": 0.03817646461569703,
      "grad_norm": 0.9851207733154297,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.2775,
      "step": 94
    },
    {
      "epoch": 0.038582597217991677,
      "grad_norm": 0.9901912808418274,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.2643,
      "step": 95
    },
    {
      "epoch": 0.038988729820286325,
      "grad_norm": 1.051145315170288,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.282,
      "step": 96
    },
    {
      "epoch": 0.03939486242258097,
      "grad_norm": 0.9764732718467712,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3819,
      "step": 97
    },
    {
      "epoch": 0.03980099502487562,
      "grad_norm": 1.2044464349746704,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.5645,
      "step": 98
    },
    {
      "epoch": 0.04020712762717027,
      "grad_norm": 1.6172761917114258,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.5574,
      "step": 99
    },
    {
      "epoch": 0.04061326022946492,
      "grad_norm": 3.0102949142456055,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.3643,
      "step": 100
    },
    {
      "epoch": 0.04061326022946492,
      "eval_loss": 1.7166489362716675,
      "eval_runtime": 309.9898,
      "eval_samples_per_second": 13.378,
      "eval_steps_per_second": 3.345,
      "step": 100
    },
    {
      "epoch": 0.04101939283175957,
      "grad_norm": 1.9809935092926025,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7555,
      "step": 101
    },
    {
      "epoch": 0.04142552543405422,
      "grad_norm": 1.9369913339614868,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.7646,
      "step": 102
    },
    {
      "epoch": 0.04183165803634887,
      "grad_norm": 1.674850344657898,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.7719,
      "step": 103
    },
    {
      "epoch": 0.04223779063864352,
      "grad_norm": 1.4067672491073608,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.705,
      "step": 104
    },
    {
      "epoch": 0.042643923240938165,
      "grad_norm": 1.117656946182251,
      "learning_rate": 5e-05,
      "loss": 1.5836,
      "step": 105
    },
    {
      "epoch": 0.043050055843232814,
      "grad_norm": 0.7025765776634216,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.5923,
      "step": 106
    },
    {
      "epoch": 0.04345618844552746,
      "grad_norm": 0.8131675720214844,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5141,
      "step": 107
    },
    {
      "epoch": 0.04386232104782212,
      "grad_norm": 0.9580807685852051,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.598,
      "step": 108
    },
    {
      "epoch": 0.044268453650116765,
      "grad_norm": 0.8206263780593872,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.5433,
      "step": 109
    },
    {
      "epoch": 0.04467458625241141,
      "grad_norm": 0.697608470916748,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.4524,
      "step": 110
    },
    {
      "epoch": 0.04508071885470606,
      "grad_norm": 0.6470988988876343,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.4266,
      "step": 111
    },
    {
      "epoch": 0.04548685145700071,
      "grad_norm": 0.6238524913787842,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4186,
      "step": 112
    },
    {
      "epoch": 0.04589298405929536,
      "grad_norm": 0.5966043472290039,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.3564,
      "step": 113
    },
    {
      "epoch": 0.046299116661590006,
      "grad_norm": 0.5739925503730774,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.4389,
      "step": 114
    },
    {
      "epoch": 0.04670524926388466,
      "grad_norm": 0.5564324259757996,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.4113,
      "step": 115
    },
    {
      "epoch": 0.04711138186617931,
      "grad_norm": 0.5870735049247742,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3268,
      "step": 116
    },
    {
      "epoch": 0.04751751446847396,
      "grad_norm": 0.5620031952857971,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.3327,
      "step": 117
    },
    {
      "epoch": 0.047923647070768606,
      "grad_norm": 0.5979235768318176,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3915,
      "step": 118
    },
    {
      "epoch": 0.048329779673063254,
      "grad_norm": 0.6064242720603943,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.4333,
      "step": 119
    },
    {
      "epoch": 0.0487359122753579,
      "grad_norm": 0.631511926651001,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3669,
      "step": 120
    },
    {
      "epoch": 0.04914204487765255,
      "grad_norm": 0.6458702683448792,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.4258,
      "step": 121
    },
    {
      "epoch": 0.049548177479947206,
      "grad_norm": 0.6667675971984863,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.4117,
      "step": 122
    },
    {
      "epoch": 0.049954310082241854,
      "grad_norm": 0.6369161605834961,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.3541,
      "step": 123
    },
    {
      "epoch": 0.0503604426845365,
      "grad_norm": 0.6958560347557068,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.292,
      "step": 124
    },
    {
      "epoch": 0.05076657528683115,
      "grad_norm": 0.7173919081687927,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3786,
      "step": 125
    },
    {
      "epoch": 0.0511727078891258,
      "grad_norm": 0.7203386425971985,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4363,
      "step": 126
    },
    {
      "epoch": 0.05157884049142045,
      "grad_norm": 0.7156944274902344,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.3003,
      "step": 127
    },
    {
      "epoch": 0.051984973093715095,
      "grad_norm": 0.7624905109405518,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.3995,
      "step": 128
    },
    {
      "epoch": 0.05239110569600975,
      "grad_norm": 0.8011417984962463,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4126,
      "step": 129
    },
    {
      "epoch": 0.0527972382983044,
      "grad_norm": 0.7700819969177246,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.3715,
      "step": 130
    },
    {
      "epoch": 0.05320337090059905,
      "grad_norm": 0.8159439563751221,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.4165,
      "step": 131
    },
    {
      "epoch": 0.053609503502893695,
      "grad_norm": 0.8066980838775635,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3141,
      "step": 132
    },
    {
      "epoch": 0.05401563610518834,
      "grad_norm": 0.928561270236969,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2663,
      "step": 133
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 0.8653779029846191,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4253,
      "step": 134
    },
    {
      "epoch": 0.05482790130977764,
      "grad_norm": 0.8156696557998657,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2435,
      "step": 135
    },
    {
      "epoch": 0.055234033912072295,
      "grad_norm": 0.8434756398200989,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.2682,
      "step": 136
    },
    {
      "epoch": 0.05564016651436694,
      "grad_norm": 0.9540610909461975,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3011,
      "step": 137
    },
    {
      "epoch": 0.05604629911666159,
      "grad_norm": 0.8743776679039001,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.3464,
      "step": 138
    },
    {
      "epoch": 0.05645243171895624,
      "grad_norm": 0.8367823362350464,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.2398,
      "step": 139
    },
    {
      "epoch": 0.05685856432125089,
      "grad_norm": 0.9489424824714661,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.2853,
      "step": 140
    },
    {
      "epoch": 0.057264696923545536,
      "grad_norm": 0.8658900856971741,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2019,
      "step": 141
    },
    {
      "epoch": 0.057670829525840184,
      "grad_norm": 0.8773536682128906,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1453,
      "step": 142
    },
    {
      "epoch": 0.05807696212813484,
      "grad_norm": 1.0234720706939697,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3279,
      "step": 143
    },
    {
      "epoch": 0.05848309473042949,
      "grad_norm": 0.9431840181350708,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2179,
      "step": 144
    },
    {
      "epoch": 0.058889227332724135,
      "grad_norm": 0.927308976650238,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1919,
      "step": 145
    },
    {
      "epoch": 0.059295359935018783,
      "grad_norm": 0.9419561624526978,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1926,
      "step": 146
    },
    {
      "epoch": 0.05970149253731343,
      "grad_norm": 1.0439989566802979,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.2789,
      "step": 147
    },
    {
      "epoch": 0.06010762513960808,
      "grad_norm": 1.4132856130599976,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.4584,
      "step": 148
    },
    {
      "epoch": 0.06051375774190273,
      "grad_norm": 1.8904759883880615,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.2134,
      "step": 149
    },
    {
      "epoch": 0.06091989034419738,
      "grad_norm": 3.5856516361236572,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.8681,
      "step": 150
    },
    {
      "epoch": 0.06091989034419738,
      "eval_loss": 1.4577025175094604,
      "eval_runtime": 310.1093,
      "eval_samples_per_second": 13.373,
      "eval_steps_per_second": 3.344,
      "step": 150
    },
    {
      "epoch": 0.06132602294649203,
      "grad_norm": 1.1017037630081177,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.5434,
      "step": 151
    },
    {
      "epoch": 0.06173215554878668,
      "grad_norm": 1.2320129871368408,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.5389,
      "step": 152
    },
    {
      "epoch": 0.06213828815108133,
      "grad_norm": 1.2117505073547363,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.5758,
      "step": 153
    },
    {
      "epoch": 0.06254442075337598,
      "grad_norm": 1.1583046913146973,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.5657,
      "step": 154
    },
    {
      "epoch": 0.06295055335567062,
      "grad_norm": 1.0914018154144287,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4844,
      "step": 155
    },
    {
      "epoch": 0.06335668595796527,
      "grad_norm": 1.020070195198059,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.4348,
      "step": 156
    },
    {
      "epoch": 0.06376281856025992,
      "grad_norm": 0.9296380877494812,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.4725,
      "step": 157
    },
    {
      "epoch": 0.06416895116255457,
      "grad_norm": 0.8589698672294617,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.5322,
      "step": 158
    },
    {
      "epoch": 0.06457508376484922,
      "grad_norm": 0.7975783348083496,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.4754,
      "step": 159
    },
    {
      "epoch": 0.06498121636714388,
      "grad_norm": 0.7380327582359314,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4245,
      "step": 160
    },
    {
      "epoch": 0.06538734896943853,
      "grad_norm": 0.7264203429222107,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5668,
      "step": 161
    },
    {
      "epoch": 0.06579348157173318,
      "grad_norm": 0.6353253126144409,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4047,
      "step": 162
    },
    {
      "epoch": 0.06619961417402782,
      "grad_norm": 0.601177990436554,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.4177,
      "step": 163
    },
    {
      "epoch": 0.06660574677632247,
      "grad_norm": 0.6236414313316345,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4479,
      "step": 164
    },
    {
      "epoch": 0.06701187937861712,
      "grad_norm": 0.6228318810462952,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.3862,
      "step": 165
    },
    {
      "epoch": 0.06741801198091177,
      "grad_norm": 0.6326121091842651,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.2904,
      "step": 166
    },
    {
      "epoch": 0.06782414458320642,
      "grad_norm": 0.6374157071113586,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4229,
      "step": 167
    },
    {
      "epoch": 0.06823027718550106,
      "grad_norm": 0.6612967252731323,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5289,
      "step": 168
    },
    {
      "epoch": 0.06863640978779571,
      "grad_norm": 0.6351006031036377,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4034,
      "step": 169
    },
    {
      "epoch": 0.06904254239009036,
      "grad_norm": 0.6338084936141968,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.4047,
      "step": 170
    },
    {
      "epoch": 0.06944867499238501,
      "grad_norm": 0.6329898238182068,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.37,
      "step": 171
    },
    {
      "epoch": 0.06985480759467966,
      "grad_norm": 0.6494223475456238,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.3661,
      "step": 172
    },
    {
      "epoch": 0.0702609401969743,
      "grad_norm": 0.6772523522377014,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.3983,
      "step": 173
    },
    {
      "epoch": 0.07066707279926897,
      "grad_norm": 0.690122127532959,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4469,
      "step": 174
    },
    {
      "epoch": 0.07107320540156362,
      "grad_norm": 0.6646182537078857,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.351,
      "step": 175
    },
    {
      "epoch": 0.07147933800385826,
      "grad_norm": 0.7394789457321167,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.3264,
      "step": 176
    },
    {
      "epoch": 0.07188547060615291,
      "grad_norm": 0.6804371476173401,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.342,
      "step": 177
    },
    {
      "epoch": 0.07229160320844756,
      "grad_norm": 0.7120791673660278,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.2843,
      "step": 178
    },
    {
      "epoch": 0.07269773581074221,
      "grad_norm": 0.8053888082504272,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.3445,
      "step": 179
    },
    {
      "epoch": 0.07310386841303686,
      "grad_norm": 0.8388939499855042,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.4256,
      "step": 180
    },
    {
      "epoch": 0.0735100010153315,
      "grad_norm": 0.7814505696296692,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.4238,
      "step": 181
    },
    {
      "epoch": 0.07391613361762615,
      "grad_norm": 0.7952286601066589,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.3003,
      "step": 182
    },
    {
      "epoch": 0.0743222662199208,
      "grad_norm": 0.8597591519355774,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2381,
      "step": 183
    },
    {
      "epoch": 0.07472839882221545,
      "grad_norm": 0.9671509861946106,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3072,
      "step": 184
    },
    {
      "epoch": 0.0751345314245101,
      "grad_norm": 1.1733289957046509,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.278,
      "step": 185
    },
    {
      "epoch": 0.07554066402680475,
      "grad_norm": 0.884924054145813,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.2248,
      "step": 186
    },
    {
      "epoch": 0.0759467966290994,
      "grad_norm": 0.849631667137146,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2509,
      "step": 187
    },
    {
      "epoch": 0.07635292923139406,
      "grad_norm": 0.8570610284805298,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0759,
      "step": 188
    },
    {
      "epoch": 0.0767590618336887,
      "grad_norm": 1.0926884412765503,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.3148,
      "step": 189
    },
    {
      "epoch": 0.07716519443598335,
      "grad_norm": 0.997616708278656,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.3344,
      "step": 190
    },
    {
      "epoch": 0.077571327038278,
      "grad_norm": 0.9728947281837463,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1879,
      "step": 191
    },
    {
      "epoch": 0.07797745964057265,
      "grad_norm": 1.0188639163970947,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3021,
      "step": 192
    },
    {
      "epoch": 0.0783835922428673,
      "grad_norm": 1.0173429250717163,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.2605,
      "step": 193
    },
    {
      "epoch": 0.07878972484516195,
      "grad_norm": 1.1394449472427368,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3249,
      "step": 194
    },
    {
      "epoch": 0.0791958574474566,
      "grad_norm": 1.1786574125289917,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.2217,
      "step": 195
    },
    {
      "epoch": 0.07960199004975124,
      "grad_norm": 1.216571569442749,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3072,
      "step": 196
    },
    {
      "epoch": 0.08000812265204589,
      "grad_norm": 1.325181245803833,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2424,
      "step": 197
    },
    {
      "epoch": 0.08041425525434054,
      "grad_norm": 1.5836933851242065,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.4076,
      "step": 198
    },
    {
      "epoch": 0.08082038785663519,
      "grad_norm": 1.8556915521621704,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.3504,
      "step": 199
    },
    {
      "epoch": 0.08122652045892984,
      "grad_norm": 3.47357177734375,
      "learning_rate": 0.0,
      "loss": 1.5353,
      "step": 200
    },
    {
      "epoch": 0.08122652045892984,
      "eval_loss": 1.3743650913238525,
      "eval_runtime": 310.3069,
      "eval_samples_per_second": 13.364,
      "eval_steps_per_second": 3.342,
      "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": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}