{
  "best_metric": 1.6065778732299805,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.1172504763300601,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005862523816503004,
      "grad_norm": 71.38330078125,
      "learning_rate": 1e-05,
      "loss": 13.8691,
      "step": 1
    },
    {
      "epoch": 0.0005862523816503004,
      "eval_loss": 3.616081714630127,
      "eval_runtime": 225.3349,
      "eval_samples_per_second": 12.75,
      "eval_steps_per_second": 3.191,
      "step": 1
    },
    {
      "epoch": 0.0011725047633006009,
      "grad_norm": 78.53181457519531,
      "learning_rate": 2e-05,
      "loss": 13.9133,
      "step": 2
    },
    {
      "epoch": 0.0017587571449509014,
      "grad_norm": 69.6111831665039,
      "learning_rate": 3e-05,
      "loss": 12.7798,
      "step": 3
    },
    {
      "epoch": 0.0023450095266012018,
      "grad_norm": 54.87360382080078,
      "learning_rate": 4e-05,
      "loss": 9.8777,
      "step": 4
    },
    {
      "epoch": 0.0029312619082515023,
      "grad_norm": 25.765483856201172,
      "learning_rate": 5e-05,
      "loss": 8.2299,
      "step": 5
    },
    {
      "epoch": 0.003517514289901803,
      "grad_norm": 8.265965461730957,
      "learning_rate": 6e-05,
      "loss": 7.8503,
      "step": 6
    },
    {
      "epoch": 0.004103766671552103,
      "grad_norm": 6.195957183837891,
      "learning_rate": 7e-05,
      "loss": 7.067,
      "step": 7
    },
    {
      "epoch": 0.0046900190532024036,
      "grad_norm": 6.622790336608887,
      "learning_rate": 8e-05,
      "loss": 7.2812,
      "step": 8
    },
    {
      "epoch": 0.005276271434852704,
      "grad_norm": 3.7051756381988525,
      "learning_rate": 9e-05,
      "loss": 7.1102,
      "step": 9
    },
    {
      "epoch": 0.005862523816503005,
      "grad_norm": 5.186803340911865,
      "learning_rate": 0.0001,
      "loss": 6.4383,
      "step": 10
    },
    {
      "epoch": 0.006448776198153305,
      "grad_norm": 3.68558406829834,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.9717,
      "step": 11
    },
    {
      "epoch": 0.007035028579803606,
      "grad_norm": 5.164979457855225,
      "learning_rate": 9.997266286704631e-05,
      "loss": 6.6645,
      "step": 12
    },
    {
      "epoch": 0.007621280961453906,
      "grad_norm": 4.284131050109863,
      "learning_rate": 9.993849845741524e-05,
      "loss": 6.5131,
      "step": 13
    },
    {
      "epoch": 0.008207533343104206,
      "grad_norm": 2.9592626094818115,
      "learning_rate": 9.989068136093873e-05,
      "loss": 5.9796,
      "step": 14
    },
    {
      "epoch": 0.008793785724754507,
      "grad_norm": 3.119295120239258,
      "learning_rate": 9.98292246503335e-05,
      "loss": 6.2774,
      "step": 15
    },
    {
      "epoch": 0.009380038106404807,
      "grad_norm": 3.1393046379089355,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.7729,
      "step": 16
    },
    {
      "epoch": 0.009966290488055108,
      "grad_norm": 2.9545459747314453,
      "learning_rate": 9.966546331768191e-05,
      "loss": 6.6589,
      "step": 17
    },
    {
      "epoch": 0.010552542869705408,
      "grad_norm": 3.1731271743774414,
      "learning_rate": 9.956320346634876e-05,
      "loss": 6.385,
      "step": 18
    },
    {
      "epoch": 0.011138795251355709,
      "grad_norm": 2.918433666229248,
      "learning_rate": 9.944739353007344e-05,
      "loss": 6.4173,
      "step": 19
    },
    {
      "epoch": 0.01172504763300601,
      "grad_norm": 3.2693846225738525,
      "learning_rate": 9.931806517013612e-05,
      "loss": 6.4684,
      "step": 20
    },
    {
      "epoch": 0.01231130001465631,
      "grad_norm": 3.0860283374786377,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.4741,
      "step": 21
    },
    {
      "epoch": 0.01289755239630661,
      "grad_norm": 3.0630736351013184,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.2077,
      "step": 22
    },
    {
      "epoch": 0.013483804777956911,
      "grad_norm": 2.8980870246887207,
      "learning_rate": 9.884934153917997e-05,
      "loss": 6.3289,
      "step": 23
    },
    {
      "epoch": 0.014070057159607212,
      "grad_norm": 3.3431448936462402,
      "learning_rate": 9.86663298624003e-05,
      "loss": 6.7361,
      "step": 24
    },
    {
      "epoch": 0.014656309541257512,
      "grad_norm": 2.7981321811676025,
      "learning_rate": 9.847001329696653e-05,
      "loss": 6.539,
      "step": 25
    },
    {
      "epoch": 0.015242561922907813,
      "grad_norm": 2.992576837539673,
      "learning_rate": 9.826044551386744e-05,
      "loss": 6.6618,
      "step": 26
    },
    {
      "epoch": 0.01582881430455811,
      "grad_norm": 3.071791887283325,
      "learning_rate": 9.803768380684242e-05,
      "loss": 7.1825,
      "step": 27
    },
    {
      "epoch": 0.016415066686208412,
      "grad_norm": 3.268291711807251,
      "learning_rate": 9.780178907671789e-05,
      "loss": 6.6932,
      "step": 28
    },
    {
      "epoch": 0.017001319067858713,
      "grad_norm": 3.145251989364624,
      "learning_rate": 9.755282581475769e-05,
      "loss": 6.7342,
      "step": 29
    },
    {
      "epoch": 0.017587571449509013,
      "grad_norm": 3.0199079513549805,
      "learning_rate": 9.729086208503174e-05,
      "loss": 6.8416,
      "step": 30
    },
    {
      "epoch": 0.018173823831159314,
      "grad_norm": 3.4400506019592285,
      "learning_rate": 9.701596950580806e-05,
      "loss": 6.8802,
      "step": 31
    },
    {
      "epoch": 0.018760076212809614,
      "grad_norm": 3.107358455657959,
      "learning_rate": 9.672822322997305e-05,
      "loss": 6.5541,
      "step": 32
    },
    {
      "epoch": 0.019346328594459915,
      "grad_norm": 3.601229429244995,
      "learning_rate": 9.642770192448536e-05,
      "loss": 7.4063,
      "step": 33
    },
    {
      "epoch": 0.019932580976110215,
      "grad_norm": 3.697211503982544,
      "learning_rate": 9.611448774886924e-05,
      "loss": 7.2312,
      "step": 34
    },
    {
      "epoch": 0.020518833357760516,
      "grad_norm": 3.537655830383301,
      "learning_rate": 9.578866633275288e-05,
      "loss": 6.8599,
      "step": 35
    },
    {
      "epoch": 0.021105085739410816,
      "grad_norm": 3.5358612537384033,
      "learning_rate": 9.545032675245813e-05,
      "loss": 6.5161,
      "step": 36
    },
    {
      "epoch": 0.021691338121061117,
      "grad_norm": 3.3208799362182617,
      "learning_rate": 9.509956150664796e-05,
      "loss": 6.7477,
      "step": 37
    },
    {
      "epoch": 0.022277590502711418,
      "grad_norm": 3.670466184616089,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.4318,
      "step": 38
    },
    {
      "epoch": 0.022863842884361718,
      "grad_norm": 3.622748374938965,
      "learning_rate": 9.43611409721806e-05,
      "loss": 6.2738,
      "step": 39
    },
    {
      "epoch": 0.02345009526601202,
      "grad_norm": 3.5902013778686523,
      "learning_rate": 9.397368756032445e-05,
      "loss": 6.6068,
      "step": 40
    },
    {
      "epoch": 0.02403634764766232,
      "grad_norm": 3.878711700439453,
      "learning_rate": 9.357421218136386e-05,
      "loss": 6.3687,
      "step": 41
    },
    {
      "epoch": 0.02462260002931262,
      "grad_norm": 3.7746715545654297,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.6814,
      "step": 42
    },
    {
      "epoch": 0.02520885241096292,
      "grad_norm": 3.956824779510498,
      "learning_rate": 9.273963562927695e-05,
      "loss": 7.4852,
      "step": 43
    },
    {
      "epoch": 0.02579510479261322,
      "grad_norm": 3.892331123352051,
      "learning_rate": 9.230476262104677e-05,
      "loss": 6.944,
      "step": 44
    },
    {
      "epoch": 0.02638135717426352,
      "grad_norm": 3.8866968154907227,
      "learning_rate": 9.185832391312644e-05,
      "loss": 6.8041,
      "step": 45
    },
    {
      "epoch": 0.026967609555913822,
      "grad_norm": 4.013768196105957,
      "learning_rate": 9.140044155740101e-05,
      "loss": 6.9016,
      "step": 46
    },
    {
      "epoch": 0.027553861937564122,
      "grad_norm": 4.0931172370910645,
      "learning_rate": 9.093124073433463e-05,
      "loss": 7.0825,
      "step": 47
    },
    {
      "epoch": 0.028140114319214423,
      "grad_norm": 4.9015116691589355,
      "learning_rate": 9.045084971874738e-05,
      "loss": 8.2827,
      "step": 48
    },
    {
      "epoch": 0.028726366700864724,
      "grad_norm": 5.217071533203125,
      "learning_rate": 8.995939984474624e-05,
      "loss": 7.0103,
      "step": 49
    },
    {
      "epoch": 0.029312619082515024,
      "grad_norm": 5.247965335845947,
      "learning_rate": 8.945702546981969e-05,
      "loss": 8.1012,
      "step": 50
    },
    {
      "epoch": 0.029312619082515024,
      "eval_loss": 1.8406819105148315,
      "eval_runtime": 226.081,
      "eval_samples_per_second": 12.708,
      "eval_steps_per_second": 3.18,
      "step": 50
    },
    {
      "epoch": 0.029898871464165325,
      "grad_norm": 36.602195739746094,
      "learning_rate": 8.894386393810563e-05,
      "loss": 8.9474,
      "step": 51
    },
    {
      "epoch": 0.030485123845815625,
      "grad_norm": 11.299726486206055,
      "learning_rate": 8.842005554284296e-05,
      "loss": 7.4255,
      "step": 52
    },
    {
      "epoch": 0.031071376227465926,
      "grad_norm": 3.2682106494903564,
      "learning_rate": 8.788574348801675e-05,
      "loss": 6.0846,
      "step": 53
    },
    {
      "epoch": 0.03165762860911622,
      "grad_norm": 3.1484243869781494,
      "learning_rate": 8.73410738492077e-05,
      "loss": 6.1174,
      "step": 54
    },
    {
      "epoch": 0.03224388099076653,
      "grad_norm": 2.4846739768981934,
      "learning_rate": 8.678619553365659e-05,
      "loss": 6.3133,
      "step": 55
    },
    {
      "epoch": 0.032830133372416824,
      "grad_norm": 2.7919976711273193,
      "learning_rate": 8.622126023955446e-05,
      "loss": 6.8546,
      "step": 56
    },
    {
      "epoch": 0.03341638575406713,
      "grad_norm": 2.343735456466675,
      "learning_rate": 8.564642241456986e-05,
      "loss": 5.8857,
      "step": 57
    },
    {
      "epoch": 0.034002638135717425,
      "grad_norm": 2.1155941486358643,
      "learning_rate": 8.506183921362443e-05,
      "loss": 6.0379,
      "step": 58
    },
    {
      "epoch": 0.03458889051736773,
      "grad_norm": 2.987092971801758,
      "learning_rate": 8.44676704559283e-05,
      "loss": 6.7259,
      "step": 59
    },
    {
      "epoch": 0.035175142899018026,
      "grad_norm": 2.7514383792877197,
      "learning_rate": 8.386407858128706e-05,
      "loss": 6.472,
      "step": 60
    },
    {
      "epoch": 0.03576139528066833,
      "grad_norm": 2.2778918743133545,
      "learning_rate": 8.32512286056924e-05,
      "loss": 5.8649,
      "step": 61
    },
    {
      "epoch": 0.03634764766231863,
      "grad_norm": 2.639725685119629,
      "learning_rate": 8.262928807620843e-05,
      "loss": 5.6167,
      "step": 62
    },
    {
      "epoch": 0.03693390004396893,
      "grad_norm": 2.694910764694214,
      "learning_rate": 8.199842702516583e-05,
      "loss": 6.3341,
      "step": 63
    },
    {
      "epoch": 0.03752015242561923,
      "grad_norm": 2.3939948081970215,
      "learning_rate": 8.135881792367686e-05,
      "loss": 6.3729,
      "step": 64
    },
    {
      "epoch": 0.03810640480726953,
      "grad_norm": 2.7309372425079346,
      "learning_rate": 8.07106356344834e-05,
      "loss": 6.0694,
      "step": 65
    },
    {
      "epoch": 0.03869265718891983,
      "grad_norm": 2.5090363025665283,
      "learning_rate": 8.005405736415126e-05,
      "loss": 5.8858,
      "step": 66
    },
    {
      "epoch": 0.039278909570570134,
      "grad_norm": 2.2945783138275146,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.8615,
      "step": 67
    },
    {
      "epoch": 0.03986516195222043,
      "grad_norm": 2.4047927856445312,
      "learning_rate": 7.871643313414718e-05,
      "loss": 6.5312,
      "step": 68
    },
    {
      "epoch": 0.040451414333870735,
      "grad_norm": 3.418168544769287,
      "learning_rate": 7.803575286758364e-05,
      "loss": 6.8929,
      "step": 69
    },
    {
      "epoch": 0.04103766671552103,
      "grad_norm": 2.7161736488342285,
      "learning_rate": 7.734740790612136e-05,
      "loss": 6.786,
      "step": 70
    },
    {
      "epoch": 0.04162391909717133,
      "grad_norm": 2.410762310028076,
      "learning_rate": 7.66515864363997e-05,
      "loss": 6.373,
      "step": 71
    },
    {
      "epoch": 0.04221017147882163,
      "grad_norm": 2.6828157901763916,
      "learning_rate": 7.594847868906076e-05,
      "loss": 6.5563,
      "step": 72
    },
    {
      "epoch": 0.04279642386047193,
      "grad_norm": 2.8595144748687744,
      "learning_rate": 7.52382768867422e-05,
      "loss": 6.213,
      "step": 73
    },
    {
      "epoch": 0.043382676242122234,
      "grad_norm": 2.469182252883911,
      "learning_rate": 7.452117519152542e-05,
      "loss": 6.4558,
      "step": 74
    },
    {
      "epoch": 0.04396892862377253,
      "grad_norm": 2.560044050216675,
      "learning_rate": 7.379736965185368e-05,
      "loss": 6.2229,
      "step": 75
    },
    {
      "epoch": 0.044555181005422835,
      "grad_norm": 2.5314645767211914,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.9446,
      "step": 76
    },
    {
      "epoch": 0.04514143338707313,
      "grad_norm": 2.5443999767303467,
      "learning_rate": 7.233044034264034e-05,
      "loss": 5.976,
      "step": 77
    },
    {
      "epoch": 0.045727685768723436,
      "grad_norm": 2.6876840591430664,
      "learning_rate": 7.158771761692464e-05,
      "loss": 6.4755,
      "step": 78
    },
    {
      "epoch": 0.04631393815037373,
      "grad_norm": 2.7290542125701904,
      "learning_rate": 7.083909302476453e-05,
      "loss": 6.3939,
      "step": 79
    },
    {
      "epoch": 0.04690019053202404,
      "grad_norm": 2.7984707355499268,
      "learning_rate": 7.008477123264848e-05,
      "loss": 6.9784,
      "step": 80
    },
    {
      "epoch": 0.047486442913674334,
      "grad_norm": 2.883815288543701,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.9441,
      "step": 81
    },
    {
      "epoch": 0.04807269529532464,
      "grad_norm": 2.8060102462768555,
      "learning_rate": 6.855986244591104e-05,
      "loss": 6.3532,
      "step": 82
    },
    {
      "epoch": 0.048658947676974935,
      "grad_norm": 2.948848247528076,
      "learning_rate": 6.778969234612584e-05,
      "loss": 6.5472,
      "step": 83
    },
    {
      "epoch": 0.04924520005862524,
      "grad_norm": 2.9612081050872803,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.4263,
      "step": 84
    },
    {
      "epoch": 0.04983145244027554,
      "grad_norm": 3.068837881088257,
      "learning_rate": 6.623497346023418e-05,
      "loss": 6.961,
      "step": 85
    },
    {
      "epoch": 0.05041770482192584,
      "grad_norm": 3.183027744293213,
      "learning_rate": 6.545084971874738e-05,
      "loss": 6.8196,
      "step": 86
    },
    {
      "epoch": 0.05100395720357614,
      "grad_norm": 3.0777058601379395,
      "learning_rate": 6.466250186922325e-05,
      "loss": 6.4177,
      "step": 87
    },
    {
      "epoch": 0.05159020958522644,
      "grad_norm": 3.085439920425415,
      "learning_rate": 6.387014543809223e-05,
      "loss": 6.9312,
      "step": 88
    },
    {
      "epoch": 0.05217646196687674,
      "grad_norm": 3.265589475631714,
      "learning_rate": 6.307399704769099e-05,
      "loss": 6.9438,
      "step": 89
    },
    {
      "epoch": 0.05276271434852704,
      "grad_norm": 3.1998724937438965,
      "learning_rate": 6.227427435703997e-05,
      "loss": 7.1417,
      "step": 90
    },
    {
      "epoch": 0.05334896673017734,
      "grad_norm": 3.108302354812622,
      "learning_rate": 6.147119600233758e-05,
      "loss": 6.7236,
      "step": 91
    },
    {
      "epoch": 0.053935219111827644,
      "grad_norm": 3.1540513038635254,
      "learning_rate": 6.066498153718735e-05,
      "loss": 6.192,
      "step": 92
    },
    {
      "epoch": 0.05452147149347794,
      "grad_norm": 3.446075916290283,
      "learning_rate": 5.985585137257401e-05,
      "loss": 7.1851,
      "step": 93
    },
    {
      "epoch": 0.055107723875128245,
      "grad_norm": 3.8707261085510254,
      "learning_rate": 5.90440267166055e-05,
      "loss": 7.0413,
      "step": 94
    },
    {
      "epoch": 0.05569397625677854,
      "grad_norm": 3.4619266986846924,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 6.969,
      "step": 95
    },
    {
      "epoch": 0.056280228638428846,
      "grad_norm": 4.064024925231934,
      "learning_rate": 5.74131823855921e-05,
      "loss": 6.9526,
      "step": 96
    },
    {
      "epoch": 0.05686648102007914,
      "grad_norm": 3.898376226425171,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 7.1065,
      "step": 97
    },
    {
      "epoch": 0.05745273340172945,
      "grad_norm": 4.27237606048584,
      "learning_rate": 5.577423184847932e-05,
      "loss": 7.2459,
      "step": 98
    },
    {
      "epoch": 0.058038985783379744,
      "grad_norm": 4.235079288482666,
      "learning_rate": 5.495227651252315e-05,
      "loss": 7.8494,
      "step": 99
    },
    {
      "epoch": 0.05862523816503005,
      "grad_norm": 4.2886481285095215,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 7.0301,
      "step": 100
    },
    {
      "epoch": 0.05862523816503005,
      "eval_loss": 1.6863446235656738,
      "eval_runtime": 225.8828,
      "eval_samples_per_second": 12.719,
      "eval_steps_per_second": 3.183,
      "step": 100
    },
    {
      "epoch": 0.059211490546680345,
      "grad_norm": 9.578201293945312,
      "learning_rate": 5.330452921628497e-05,
      "loss": 8.3987,
      "step": 101
    },
    {
      "epoch": 0.05979774292833065,
      "grad_norm": 6.767113208770752,
      "learning_rate": 5.247918773366112e-05,
      "loss": 7.6868,
      "step": 102
    },
    {
      "epoch": 0.060383995309980946,
      "grad_norm": 2.6064271926879883,
      "learning_rate": 5.165316846586541e-05,
      "loss": 6.4807,
      "step": 103
    },
    {
      "epoch": 0.06097024769163125,
      "grad_norm": 2.001678705215454,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 6.3408,
      "step": 104
    },
    {
      "epoch": 0.06155650007328155,
      "grad_norm": 1.9707530736923218,
      "learning_rate": 5e-05,
      "loss": 5.6117,
      "step": 105
    },
    {
      "epoch": 0.06214275245493185,
      "grad_norm": 2.2305493354797363,
      "learning_rate": 4.917330276168208e-05,
      "loss": 6.9079,
      "step": 106
    },
    {
      "epoch": 0.06272900483658214,
      "grad_norm": 2.1230199337005615,
      "learning_rate": 4.834683153413459e-05,
      "loss": 5.7214,
      "step": 107
    },
    {
      "epoch": 0.06331525721823245,
      "grad_norm": 2.1910359859466553,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 5.9372,
      "step": 108
    },
    {
      "epoch": 0.06390150959988275,
      "grad_norm": 2.100861072540283,
      "learning_rate": 4.669547078371504e-05,
      "loss": 5.7909,
      "step": 109
    },
    {
      "epoch": 0.06448776198153305,
      "grad_norm": 2.5007755756378174,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 6.4156,
      "step": 110
    },
    {
      "epoch": 0.06507401436318334,
      "grad_norm": 2.4833149909973145,
      "learning_rate": 4.504772348747687e-05,
      "loss": 6.4771,
      "step": 111
    },
    {
      "epoch": 0.06566026674483365,
      "grad_norm": 2.2103281021118164,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 6.2594,
      "step": 112
    },
    {
      "epoch": 0.06624651912648395,
      "grad_norm": 2.138319253921509,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 6.0618,
      "step": 113
    },
    {
      "epoch": 0.06683277150813426,
      "grad_norm": 2.277013063430786,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 6.5713,
      "step": 114
    },
    {
      "epoch": 0.06741902388978455,
      "grad_norm": 2.24790096282959,
      "learning_rate": 4.17702704859633e-05,
      "loss": 6.1821,
      "step": 115
    },
    {
      "epoch": 0.06800527627143485,
      "grad_norm": 2.2910306453704834,
      "learning_rate": 4.095597328339452e-05,
      "loss": 6.149,
      "step": 116
    },
    {
      "epoch": 0.06859152865308515,
      "grad_norm": 2.2897682189941406,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 6.5942,
      "step": 117
    },
    {
      "epoch": 0.06917778103473546,
      "grad_norm": 2.400585174560547,
      "learning_rate": 3.933501846281267e-05,
      "loss": 6.2582,
      "step": 118
    },
    {
      "epoch": 0.06976403341638575,
      "grad_norm": 2.39587140083313,
      "learning_rate": 3.852880399766243e-05,
      "loss": 6.8109,
      "step": 119
    },
    {
      "epoch": 0.07035028579803605,
      "grad_norm": 2.3303301334381104,
      "learning_rate": 3.772572564296005e-05,
      "loss": 6.3897,
      "step": 120
    },
    {
      "epoch": 0.07093653817968636,
      "grad_norm": 2.4612951278686523,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 6.4337,
      "step": 121
    },
    {
      "epoch": 0.07152279056133666,
      "grad_norm": 2.48661732673645,
      "learning_rate": 3.612985456190778e-05,
      "loss": 6.6579,
      "step": 122
    },
    {
      "epoch": 0.07210904294298695,
      "grad_norm": 2.777845859527588,
      "learning_rate": 3.533749813077677e-05,
      "loss": 6.4186,
      "step": 123
    },
    {
      "epoch": 0.07269529532463725,
      "grad_norm": 2.5977935791015625,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 6.8135,
      "step": 124
    },
    {
      "epoch": 0.07328154770628756,
      "grad_norm": 2.726835012435913,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 6.3536,
      "step": 125
    },
    {
      "epoch": 0.07386780008793786,
      "grad_norm": 2.549349546432495,
      "learning_rate": 3.298534127791785e-05,
      "loss": 6.6661,
      "step": 126
    },
    {
      "epoch": 0.07445405246958815,
      "grad_norm": 2.7201812267303467,
      "learning_rate": 3.221030765387417e-05,
      "loss": 6.3219,
      "step": 127
    },
    {
      "epoch": 0.07504030485123846,
      "grad_norm": 2.7335641384124756,
      "learning_rate": 3.144013755408895e-05,
      "loss": 6.1523,
      "step": 128
    },
    {
      "epoch": 0.07562655723288876,
      "grad_norm": 2.6927502155303955,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 6.1037,
      "step": 129
    },
    {
      "epoch": 0.07621280961453906,
      "grad_norm": 2.471005439758301,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.5268,
      "step": 130
    },
    {
      "epoch": 0.07679906199618936,
      "grad_norm": 3.0687620639801025,
      "learning_rate": 2.916090697523549e-05,
      "loss": 6.3406,
      "step": 131
    },
    {
      "epoch": 0.07738531437783966,
      "grad_norm": 2.914454698562622,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 6.5744,
      "step": 132
    },
    {
      "epoch": 0.07797156675948996,
      "grad_norm": 3.0881731510162354,
      "learning_rate": 2.766955965735968e-05,
      "loss": 6.9825,
      "step": 133
    },
    {
      "epoch": 0.07855781914114027,
      "grad_norm": 2.821725606918335,
      "learning_rate": 2.693294185106562e-05,
      "loss": 6.9401,
      "step": 134
    },
    {
      "epoch": 0.07914407152279056,
      "grad_norm": 2.757154703140259,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 5.9439,
      "step": 135
    },
    {
      "epoch": 0.07973032390444086,
      "grad_norm": 2.9801981449127197,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.0921,
      "step": 136
    },
    {
      "epoch": 0.08031657628609117,
      "grad_norm": 3.1656322479248047,
      "learning_rate": 2.476172311325783e-05,
      "loss": 6.8298,
      "step": 137
    },
    {
      "epoch": 0.08090282866774147,
      "grad_norm": 3.065058469772339,
      "learning_rate": 2.405152131093926e-05,
      "loss": 6.2826,
      "step": 138
    },
    {
      "epoch": 0.08148908104939176,
      "grad_norm": 3.4856159687042236,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 6.7237,
      "step": 139
    },
    {
      "epoch": 0.08207533343104206,
      "grad_norm": 3.0992062091827393,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 6.0419,
      "step": 140
    },
    {
      "epoch": 0.08266158581269237,
      "grad_norm": 3.462188243865967,
      "learning_rate": 2.196424713241637e-05,
      "loss": 6.8798,
      "step": 141
    },
    {
      "epoch": 0.08324783819434266,
      "grad_norm": 3.5840885639190674,
      "learning_rate": 2.128356686585282e-05,
      "loss": 6.7436,
      "step": 142
    },
    {
      "epoch": 0.08383409057599296,
      "grad_norm": 3.518064022064209,
      "learning_rate": 2.061073738537635e-05,
      "loss": 6.9207,
      "step": 143
    },
    {
      "epoch": 0.08442034295764327,
      "grad_norm": 3.564720869064331,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 6.9566,
      "step": 144
    },
    {
      "epoch": 0.08500659533929357,
      "grad_norm": 3.536327838897705,
      "learning_rate": 1.928936436551661e-05,
      "loss": 6.8501,
      "step": 145
    },
    {
      "epoch": 0.08559284772094386,
      "grad_norm": 3.493380308151245,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 7.0708,
      "step": 146
    },
    {
      "epoch": 0.08617910010259416,
      "grad_norm": 3.687791109085083,
      "learning_rate": 1.800157297483417e-05,
      "loss": 7.4765,
      "step": 147
    },
    {
      "epoch": 0.08676535248424447,
      "grad_norm": 3.894880533218384,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 7.198,
      "step": 148
    },
    {
      "epoch": 0.08735160486589477,
      "grad_norm": 3.8245627880096436,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 7.3763,
      "step": 149
    },
    {
      "epoch": 0.08793785724754506,
      "grad_norm": 4.5974650382995605,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 8.3387,
      "step": 150
    },
    {
      "epoch": 0.08793785724754506,
      "eval_loss": 1.633521556854248,
      "eval_runtime": 226.0772,
      "eval_samples_per_second": 12.708,
      "eval_steps_per_second": 3.18,
      "step": 150
    },
    {
      "epoch": 0.08852410962919537,
      "grad_norm": 4.9135870933532715,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.4894,
      "step": 151
    },
    {
      "epoch": 0.08911036201084567,
      "grad_norm": 5.033265590667725,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 7.1238,
      "step": 152
    },
    {
      "epoch": 0.08969661439249597,
      "grad_norm": 5.139833927154541,
      "learning_rate": 1.435357758543015e-05,
      "loss": 6.9575,
      "step": 153
    },
    {
      "epoch": 0.09028286677414626,
      "grad_norm": 3.8156111240386963,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 6.4162,
      "step": 154
    },
    {
      "epoch": 0.09086911915579657,
      "grad_norm": 2.8289899826049805,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 6.0572,
      "step": 155
    },
    {
      "epoch": 0.09145537153744687,
      "grad_norm": 2.1409289836883545,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 5.5625,
      "step": 156
    },
    {
      "epoch": 0.09204162391909718,
      "grad_norm": 2.184295892715454,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 6.7061,
      "step": 157
    },
    {
      "epoch": 0.09262787630074747,
      "grad_norm": 2.0775392055511475,
      "learning_rate": 1.157994445715706e-05,
      "loss": 5.9066,
      "step": 158
    },
    {
      "epoch": 0.09321412868239777,
      "grad_norm": 2.113314390182495,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 5.0104,
      "step": 159
    },
    {
      "epoch": 0.09380038106404807,
      "grad_norm": 2.2511730194091797,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 5.6316,
      "step": 160
    },
    {
      "epoch": 0.09438663344569838,
      "grad_norm": 2.060756206512451,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 5.4911,
      "step": 161
    },
    {
      "epoch": 0.09497288582734867,
      "grad_norm": 2.036313772201538,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.7948,
      "step": 162
    },
    {
      "epoch": 0.09555913820899897,
      "grad_norm": 2.3644955158233643,
      "learning_rate": 9.068759265665384e-06,
      "loss": 6.0811,
      "step": 163
    },
    {
      "epoch": 0.09614539059064928,
      "grad_norm": 2.277362108230591,
      "learning_rate": 8.599558442598998e-06,
      "loss": 6.5315,
      "step": 164
    },
    {
      "epoch": 0.09673164297229958,
      "grad_norm": 2.248805046081543,
      "learning_rate": 8.141676086873572e-06,
      "loss": 6.1857,
      "step": 165
    },
    {
      "epoch": 0.09731789535394987,
      "grad_norm": 2.33545184135437,
      "learning_rate": 7.695237378953223e-06,
      "loss": 6.2232,
      "step": 166
    },
    {
      "epoch": 0.09790414773560017,
      "grad_norm": 2.517578363418579,
      "learning_rate": 7.260364370723044e-06,
      "loss": 5.994,
      "step": 167
    },
    {
      "epoch": 0.09849040011725048,
      "grad_norm": 2.34413480758667,
      "learning_rate": 6.837175952121306e-06,
      "loss": 6.0814,
      "step": 168
    },
    {
      "epoch": 0.09907665249890078,
      "grad_norm": 2.5742177963256836,
      "learning_rate": 6.425787818636131e-06,
      "loss": 5.9977,
      "step": 169
    },
    {
      "epoch": 0.09966290488055107,
      "grad_norm": 2.3905839920043945,
      "learning_rate": 6.026312439675552e-06,
      "loss": 6.0969,
      "step": 170
    },
    {
      "epoch": 0.10024915726220138,
      "grad_norm": 2.851008653640747,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 6.1431,
      "step": 171
    },
    {
      "epoch": 0.10083540964385168,
      "grad_norm": 2.5609467029571533,
      "learning_rate": 5.263533508961827e-06,
      "loss": 6.3342,
      "step": 172
    },
    {
      "epoch": 0.10142166202550199,
      "grad_norm": 2.339158058166504,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.8788,
      "step": 173
    },
    {
      "epoch": 0.10200791440715228,
      "grad_norm": 2.593463182449341,
      "learning_rate": 4.549673247541875e-06,
      "loss": 6.2099,
      "step": 174
    },
    {
      "epoch": 0.10259416678880258,
      "grad_norm": 2.5745880603790283,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 6.0579,
      "step": 175
    },
    {
      "epoch": 0.10318041917045288,
      "grad_norm": 2.524670362472534,
      "learning_rate": 3.885512251130763e-06,
      "loss": 6.5299,
      "step": 176
    },
    {
      "epoch": 0.10376667155210317,
      "grad_norm": 2.679459571838379,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 6.6853,
      "step": 177
    },
    {
      "epoch": 0.10435292393375348,
      "grad_norm": 2.920398235321045,
      "learning_rate": 3.271776770026963e-06,
      "loss": 7.0185,
      "step": 178
    },
    {
      "epoch": 0.10493917631540378,
      "grad_norm": 2.8493683338165283,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 7.0858,
      "step": 179
    },
    {
      "epoch": 0.10552542869705409,
      "grad_norm": 2.814725875854492,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 6.5342,
      "step": 180
    },
    {
      "epoch": 0.10611168107870438,
      "grad_norm": 2.7952964305877686,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 6.7515,
      "step": 181
    },
    {
      "epoch": 0.10669793346035468,
      "grad_norm": 2.735405445098877,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 6.7435,
      "step": 182
    },
    {
      "epoch": 0.10728418584200498,
      "grad_norm": 2.869135618209839,
      "learning_rate": 1.962316193157593e-06,
      "loss": 6.0154,
      "step": 183
    },
    {
      "epoch": 0.10787043822365529,
      "grad_norm": 2.9325292110443115,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.7613,
      "step": 184
    },
    {
      "epoch": 0.10845669060530558,
      "grad_norm": 2.87384033203125,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.0645,
      "step": 185
    },
    {
      "epoch": 0.10904294298695588,
      "grad_norm": 3.3462395668029785,
      "learning_rate": 1.333670137599713e-06,
      "loss": 6.9597,
      "step": 186
    },
    {
      "epoch": 0.10962919536860619,
      "grad_norm": 3.02093505859375,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 6.4318,
      "step": 187
    },
    {
      "epoch": 0.11021544775025649,
      "grad_norm": 3.3400521278381348,
      "learning_rate": 9.810017062595322e-07,
      "loss": 7.2103,
      "step": 188
    },
    {
      "epoch": 0.11080170013190678,
      "grad_norm": 3.268441677093506,
      "learning_rate": 8.247462563808817e-07,
      "loss": 7.0968,
      "step": 189
    },
    {
      "epoch": 0.11138795251355708,
      "grad_norm": 3.1540441513061523,
      "learning_rate": 6.819348298638839e-07,
      "loss": 6.5756,
      "step": 190
    },
    {
      "epoch": 0.11197420489520739,
      "grad_norm": 3.2289986610412598,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.8902,
      "step": 191
    },
    {
      "epoch": 0.11256045727685769,
      "grad_norm": 3.514127492904663,
      "learning_rate": 4.367965336512403e-07,
      "loss": 7.2681,
      "step": 192
    },
    {
      "epoch": 0.11314670965850798,
      "grad_norm": 3.438389301300049,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 6.7378,
      "step": 193
    },
    {
      "epoch": 0.11373296204015829,
      "grad_norm": 3.536512851715088,
      "learning_rate": 2.458548727494292e-07,
      "loss": 6.8209,
      "step": 194
    },
    {
      "epoch": 0.11431921442180859,
      "grad_norm": 3.4927186965942383,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 6.9012,
      "step": 195
    },
    {
      "epoch": 0.1149054668034589,
      "grad_norm": 3.606351613998413,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 6.784,
      "step": 196
    },
    {
      "epoch": 0.11549171918510918,
      "grad_norm": 3.8921163082122803,
      "learning_rate": 6.150154258476315e-08,
      "loss": 7.7776,
      "step": 197
    },
    {
      "epoch": 0.11607797156675949,
      "grad_norm": 3.770442008972168,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 6.9755,
      "step": 198
    },
    {
      "epoch": 0.11666422394840979,
      "grad_norm": 5.291163921356201,
      "learning_rate": 6.834750376549792e-09,
      "loss": 7.4909,
      "step": 199
    },
    {
      "epoch": 0.1172504763300601,
      "grad_norm": 4.994363784790039,
      "learning_rate": 0.0,
      "loss": 7.679,
      "step": 200
    },
    {
      "epoch": 0.1172504763300601,
      "eval_loss": 1.6065778732299805,
      "eval_runtime": 226.1075,
      "eval_samples_per_second": 12.706,
      "eval_steps_per_second": 3.18,
      "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": 2.966802063603794e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}