{
  "best_metric": 1.6063930988311768,
  "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": 71.74431610107422,
      "learning_rate": 1e-05,
      "loss": 13.8691,
      "step": 1
    },
    {
      "epoch": 0.0005862523816503004,
      "eval_loss": 3.616081714630127,
      "eval_runtime": 225.8,
      "eval_samples_per_second": 12.724,
      "eval_steps_per_second": 3.184,
      "step": 1
    },
    {
      "epoch": 0.0011725047633006009,
      "grad_norm": 78.9216079711914,
      "learning_rate": 2e-05,
      "loss": 13.9133,
      "step": 2
    },
    {
      "epoch": 0.0017587571449509014,
      "grad_norm": 69.60846710205078,
      "learning_rate": 3e-05,
      "loss": 12.7652,
      "step": 3
    },
    {
      "epoch": 0.0023450095266012018,
      "grad_norm": 55.5763053894043,
      "learning_rate": 4e-05,
      "loss": 9.8582,
      "step": 4
    },
    {
      "epoch": 0.0029312619082515023,
      "grad_norm": 25.325239181518555,
      "learning_rate": 5e-05,
      "loss": 8.2207,
      "step": 5
    },
    {
      "epoch": 0.003517514289901803,
      "grad_norm": 15.54898738861084,
      "learning_rate": 6e-05,
      "loss": 7.8553,
      "step": 6
    },
    {
      "epoch": 0.004103766671552103,
      "grad_norm": 6.679100513458252,
      "learning_rate": 7e-05,
      "loss": 7.1132,
      "step": 7
    },
    {
      "epoch": 0.0046900190532024036,
      "grad_norm": 6.3928022384643555,
      "learning_rate": 8e-05,
      "loss": 7.3109,
      "step": 8
    },
    {
      "epoch": 0.005276271434852704,
      "grad_norm": 3.7989559173583984,
      "learning_rate": 9e-05,
      "loss": 7.1344,
      "step": 9
    },
    {
      "epoch": 0.005862523816503005,
      "grad_norm": 5.412967681884766,
      "learning_rate": 0.0001,
      "loss": 6.4554,
      "step": 10
    },
    {
      "epoch": 0.006448776198153305,
      "grad_norm": 4.018998622894287,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.984,
      "step": 11
    },
    {
      "epoch": 0.007035028579803606,
      "grad_norm": 5.7428107261657715,
      "learning_rate": 9.997266286704631e-05,
      "loss": 6.6835,
      "step": 12
    },
    {
      "epoch": 0.007621280961453906,
      "grad_norm": 4.269711017608643,
      "learning_rate": 9.993849845741524e-05,
      "loss": 6.5248,
      "step": 13
    },
    {
      "epoch": 0.008207533343104206,
      "grad_norm": 2.978231906890869,
      "learning_rate": 9.989068136093873e-05,
      "loss": 5.9893,
      "step": 14
    },
    {
      "epoch": 0.008793785724754507,
      "grad_norm": 3.280388593673706,
      "learning_rate": 9.98292246503335e-05,
      "loss": 6.2859,
      "step": 15
    },
    {
      "epoch": 0.009380038106404807,
      "grad_norm": 3.077812910079956,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.7801,
      "step": 16
    },
    {
      "epoch": 0.009966290488055108,
      "grad_norm": 3.040489673614502,
      "learning_rate": 9.966546331768191e-05,
      "loss": 6.6661,
      "step": 17
    },
    {
      "epoch": 0.010552542869705408,
      "grad_norm": 3.116974115371704,
      "learning_rate": 9.956320346634876e-05,
      "loss": 6.3934,
      "step": 18
    },
    {
      "epoch": 0.011138795251355709,
      "grad_norm": 3.060584783554077,
      "learning_rate": 9.944739353007344e-05,
      "loss": 6.4177,
      "step": 19
    },
    {
      "epoch": 0.01172504763300601,
      "grad_norm": 3.299783229827881,
      "learning_rate": 9.931806517013612e-05,
      "loss": 6.4753,
      "step": 20
    },
    {
      "epoch": 0.01231130001465631,
      "grad_norm": 3.0765109062194824,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.4795,
      "step": 21
    },
    {
      "epoch": 0.01289755239630661,
      "grad_norm": 3.0756661891937256,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.211,
      "step": 22
    },
    {
      "epoch": 0.013483804777956911,
      "grad_norm": 2.891641616821289,
      "learning_rate": 9.884934153917997e-05,
      "loss": 6.336,
      "step": 23
    },
    {
      "epoch": 0.014070057159607212,
      "grad_norm": 3.514195680618286,
      "learning_rate": 9.86663298624003e-05,
      "loss": 6.739,
      "step": 24
    },
    {
      "epoch": 0.014656309541257512,
      "grad_norm": 2.8506174087524414,
      "learning_rate": 9.847001329696653e-05,
      "loss": 6.5377,
      "step": 25
    },
    {
      "epoch": 0.015242561922907813,
      "grad_norm": 2.9577016830444336,
      "learning_rate": 9.826044551386744e-05,
      "loss": 6.6655,
      "step": 26
    },
    {
      "epoch": 0.01582881430455811,
      "grad_norm": 3.104137420654297,
      "learning_rate": 9.803768380684242e-05,
      "loss": 7.1886,
      "step": 27
    },
    {
      "epoch": 0.016415066686208412,
      "grad_norm": 3.215820074081421,
      "learning_rate": 9.780178907671789e-05,
      "loss": 6.6964,
      "step": 28
    },
    {
      "epoch": 0.017001319067858713,
      "grad_norm": 3.065380096435547,
      "learning_rate": 9.755282581475769e-05,
      "loss": 6.735,
      "step": 29
    },
    {
      "epoch": 0.017587571449509013,
      "grad_norm": 3.0146641731262207,
      "learning_rate": 9.729086208503174e-05,
      "loss": 6.8454,
      "step": 30
    },
    {
      "epoch": 0.018173823831159314,
      "grad_norm": 3.3363120555877686,
      "learning_rate": 9.701596950580806e-05,
      "loss": 6.8841,
      "step": 31
    },
    {
      "epoch": 0.018760076212809614,
      "grad_norm": 3.1610114574432373,
      "learning_rate": 9.672822322997305e-05,
      "loss": 6.5538,
      "step": 32
    },
    {
      "epoch": 0.019346328594459915,
      "grad_norm": 3.559464454650879,
      "learning_rate": 9.642770192448536e-05,
      "loss": 7.4027,
      "step": 33
    },
    {
      "epoch": 0.019932580976110215,
      "grad_norm": 3.6785998344421387,
      "learning_rate": 9.611448774886924e-05,
      "loss": 7.2259,
      "step": 34
    },
    {
      "epoch": 0.020518833357760516,
      "grad_norm": 3.6969072818756104,
      "learning_rate": 9.578866633275288e-05,
      "loss": 6.8629,
      "step": 35
    },
    {
      "epoch": 0.021105085739410816,
      "grad_norm": 3.533625364303589,
      "learning_rate": 9.545032675245813e-05,
      "loss": 6.5136,
      "step": 36
    },
    {
      "epoch": 0.021691338121061117,
      "grad_norm": 3.2969021797180176,
      "learning_rate": 9.509956150664796e-05,
      "loss": 6.7494,
      "step": 37
    },
    {
      "epoch": 0.022277590502711418,
      "grad_norm": 3.5021414756774902,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.4376,
      "step": 38
    },
    {
      "epoch": 0.022863842884361718,
      "grad_norm": 3.596881628036499,
      "learning_rate": 9.43611409721806e-05,
      "loss": 6.2761,
      "step": 39
    },
    {
      "epoch": 0.02345009526601202,
      "grad_norm": 3.4956023693084717,
      "learning_rate": 9.397368756032445e-05,
      "loss": 6.6019,
      "step": 40
    },
    {
      "epoch": 0.02403634764766232,
      "grad_norm": 3.799384117126465,
      "learning_rate": 9.357421218136386e-05,
      "loss": 6.3625,
      "step": 41
    },
    {
      "epoch": 0.02462260002931262,
      "grad_norm": 3.76020884513855,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.6828,
      "step": 42
    },
    {
      "epoch": 0.02520885241096292,
      "grad_norm": 3.880293607711792,
      "learning_rate": 9.273963562927695e-05,
      "loss": 7.4852,
      "step": 43
    },
    {
      "epoch": 0.02579510479261322,
      "grad_norm": 3.8231923580169678,
      "learning_rate": 9.230476262104677e-05,
      "loss": 6.9405,
      "step": 44
    },
    {
      "epoch": 0.02638135717426352,
      "grad_norm": 4.1071038246154785,
      "learning_rate": 9.185832391312644e-05,
      "loss": 6.8187,
      "step": 45
    },
    {
      "epoch": 0.026967609555913822,
      "grad_norm": 3.8722639083862305,
      "learning_rate": 9.140044155740101e-05,
      "loss": 6.8968,
      "step": 46
    },
    {
      "epoch": 0.027553861937564122,
      "grad_norm": 4.049685001373291,
      "learning_rate": 9.093124073433463e-05,
      "loss": 7.0898,
      "step": 47
    },
    {
      "epoch": 0.028140114319214423,
      "grad_norm": 4.8560357093811035,
      "learning_rate": 9.045084971874738e-05,
      "loss": 8.2837,
      "step": 48
    },
    {
      "epoch": 0.028726366700864724,
      "grad_norm": 5.141579627990723,
      "learning_rate": 8.995939984474624e-05,
      "loss": 7.0076,
      "step": 49
    },
    {
      "epoch": 0.029312619082515024,
      "grad_norm": 5.118967056274414,
      "learning_rate": 8.945702546981969e-05,
      "loss": 8.1159,
      "step": 50
    },
    {
      "epoch": 0.029312619082515024,
      "eval_loss": 1.837236762046814,
      "eval_runtime": 227.1315,
      "eval_samples_per_second": 12.649,
      "eval_steps_per_second": 3.166,
      "step": 50
    },
    {
      "epoch": 0.029898871464165325,
      "grad_norm": 35.49456787109375,
      "learning_rate": 8.894386393810563e-05,
      "loss": 8.8848,
      "step": 51
    },
    {
      "epoch": 0.030485123845815625,
      "grad_norm": 10.488907814025879,
      "learning_rate": 8.842005554284296e-05,
      "loss": 7.3923,
      "step": 52
    },
    {
      "epoch": 0.031071376227465926,
      "grad_norm": 3.1175405979156494,
      "learning_rate": 8.788574348801675e-05,
      "loss": 6.0721,
      "step": 53
    },
    {
      "epoch": 0.03165762860911622,
      "grad_norm": 2.8991098403930664,
      "learning_rate": 8.73410738492077e-05,
      "loss": 6.1196,
      "step": 54
    },
    {
      "epoch": 0.03224388099076653,
      "grad_norm": 2.571641206741333,
      "learning_rate": 8.678619553365659e-05,
      "loss": 6.3167,
      "step": 55
    },
    {
      "epoch": 0.032830133372416824,
      "grad_norm": 2.7571520805358887,
      "learning_rate": 8.622126023955446e-05,
      "loss": 6.8518,
      "step": 56
    },
    {
      "epoch": 0.03341638575406713,
      "grad_norm": 2.2308900356292725,
      "learning_rate": 8.564642241456986e-05,
      "loss": 5.8886,
      "step": 57
    },
    {
      "epoch": 0.034002638135717425,
      "grad_norm": 2.1759371757507324,
      "learning_rate": 8.506183921362443e-05,
      "loss": 6.0425,
      "step": 58
    },
    {
      "epoch": 0.03458889051736773,
      "grad_norm": 3.0100531578063965,
      "learning_rate": 8.44676704559283e-05,
      "loss": 6.7173,
      "step": 59
    },
    {
      "epoch": 0.035175142899018026,
      "grad_norm": 2.506157159805298,
      "learning_rate": 8.386407858128706e-05,
      "loss": 6.4661,
      "step": 60
    },
    {
      "epoch": 0.03576139528066833,
      "grad_norm": 2.425422191619873,
      "learning_rate": 8.32512286056924e-05,
      "loss": 5.8649,
      "step": 61
    },
    {
      "epoch": 0.03634764766231863,
      "grad_norm": 2.6796298027038574,
      "learning_rate": 8.262928807620843e-05,
      "loss": 5.6195,
      "step": 62
    },
    {
      "epoch": 0.03693390004396893,
      "grad_norm": 2.490652322769165,
      "learning_rate": 8.199842702516583e-05,
      "loss": 6.3357,
      "step": 63
    },
    {
      "epoch": 0.03752015242561923,
      "grad_norm": 2.3136560916900635,
      "learning_rate": 8.135881792367686e-05,
      "loss": 6.3827,
      "step": 64
    },
    {
      "epoch": 0.03810640480726953,
      "grad_norm": 2.845165491104126,
      "learning_rate": 8.07106356344834e-05,
      "loss": 6.0737,
      "step": 65
    },
    {
      "epoch": 0.03869265718891983,
      "grad_norm": 2.4424049854278564,
      "learning_rate": 8.005405736415126e-05,
      "loss": 5.8786,
      "step": 66
    },
    {
      "epoch": 0.039278909570570134,
      "grad_norm": 2.281189203262329,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.859,
      "step": 67
    },
    {
      "epoch": 0.03986516195222043,
      "grad_norm": 2.4373369216918945,
      "learning_rate": 7.871643313414718e-05,
      "loss": 6.5304,
      "step": 68
    },
    {
      "epoch": 0.040451414333870735,
      "grad_norm": 3.3283677101135254,
      "learning_rate": 7.803575286758364e-05,
      "loss": 6.8858,
      "step": 69
    },
    {
      "epoch": 0.04103766671552103,
      "grad_norm": 2.647940158843994,
      "learning_rate": 7.734740790612136e-05,
      "loss": 6.7807,
      "step": 70
    },
    {
      "epoch": 0.04162391909717133,
      "grad_norm": 2.4119699001312256,
      "learning_rate": 7.66515864363997e-05,
      "loss": 6.3713,
      "step": 71
    },
    {
      "epoch": 0.04221017147882163,
      "grad_norm": 2.6970508098602295,
      "learning_rate": 7.594847868906076e-05,
      "loss": 6.5585,
      "step": 72
    },
    {
      "epoch": 0.04279642386047193,
      "grad_norm": 2.729961633682251,
      "learning_rate": 7.52382768867422e-05,
      "loss": 6.2155,
      "step": 73
    },
    {
      "epoch": 0.043382676242122234,
      "grad_norm": 2.4872384071350098,
      "learning_rate": 7.452117519152542e-05,
      "loss": 6.465,
      "step": 74
    },
    {
      "epoch": 0.04396892862377253,
      "grad_norm": 2.584740161895752,
      "learning_rate": 7.379736965185368e-05,
      "loss": 6.2198,
      "step": 75
    },
    {
      "epoch": 0.044555181005422835,
      "grad_norm": 2.4839394092559814,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.9417,
      "step": 76
    },
    {
      "epoch": 0.04514143338707313,
      "grad_norm": 2.538069725036621,
      "learning_rate": 7.233044034264034e-05,
      "loss": 5.9775,
      "step": 77
    },
    {
      "epoch": 0.045727685768723436,
      "grad_norm": 2.699746608734131,
      "learning_rate": 7.158771761692464e-05,
      "loss": 6.4739,
      "step": 78
    },
    {
      "epoch": 0.04631393815037373,
      "grad_norm": 2.685896873474121,
      "learning_rate": 7.083909302476453e-05,
      "loss": 6.3922,
      "step": 79
    },
    {
      "epoch": 0.04690019053202404,
      "grad_norm": 2.727782726287842,
      "learning_rate": 7.008477123264848e-05,
      "loss": 6.9726,
      "step": 80
    },
    {
      "epoch": 0.047486442913674334,
      "grad_norm": 2.8873696327209473,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.9436,
      "step": 81
    },
    {
      "epoch": 0.04807269529532464,
      "grad_norm": 2.8404133319854736,
      "learning_rate": 6.855986244591104e-05,
      "loss": 6.3638,
      "step": 82
    },
    {
      "epoch": 0.048658947676974935,
      "grad_norm": 2.9060897827148438,
      "learning_rate": 6.778969234612584e-05,
      "loss": 6.5394,
      "step": 83
    },
    {
      "epoch": 0.04924520005862524,
      "grad_norm": 2.985032320022583,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.4281,
      "step": 84
    },
    {
      "epoch": 0.04983145244027554,
      "grad_norm": 3.1234097480773926,
      "learning_rate": 6.623497346023418e-05,
      "loss": 6.9645,
      "step": 85
    },
    {
      "epoch": 0.05041770482192584,
      "grad_norm": 3.1589691638946533,
      "learning_rate": 6.545084971874738e-05,
      "loss": 6.8246,
      "step": 86
    },
    {
      "epoch": 0.05100395720357614,
      "grad_norm": 3.082123041152954,
      "learning_rate": 6.466250186922325e-05,
      "loss": 6.4129,
      "step": 87
    },
    {
      "epoch": 0.05159020958522644,
      "grad_norm": 3.1589043140411377,
      "learning_rate": 6.387014543809223e-05,
      "loss": 6.932,
      "step": 88
    },
    {
      "epoch": 0.05217646196687674,
      "grad_norm": 3.284149169921875,
      "learning_rate": 6.307399704769099e-05,
      "loss": 6.9355,
      "step": 89
    },
    {
      "epoch": 0.05276271434852704,
      "grad_norm": 3.1623196601867676,
      "learning_rate": 6.227427435703997e-05,
      "loss": 7.1432,
      "step": 90
    },
    {
      "epoch": 0.05334896673017734,
      "grad_norm": 3.01210880279541,
      "learning_rate": 6.147119600233758e-05,
      "loss": 6.7221,
      "step": 91
    },
    {
      "epoch": 0.053935219111827644,
      "grad_norm": 3.1389036178588867,
      "learning_rate": 6.066498153718735e-05,
      "loss": 6.1897,
      "step": 92
    },
    {
      "epoch": 0.05452147149347794,
      "grad_norm": 3.4790282249450684,
      "learning_rate": 5.985585137257401e-05,
      "loss": 7.1784,
      "step": 93
    },
    {
      "epoch": 0.055107723875128245,
      "grad_norm": 3.912142276763916,
      "learning_rate": 5.90440267166055e-05,
      "loss": 7.0445,
      "step": 94
    },
    {
      "epoch": 0.05569397625677854,
      "grad_norm": 3.462918281555176,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 6.9702,
      "step": 95
    },
    {
      "epoch": 0.056280228638428846,
      "grad_norm": 4.0536627769470215,
      "learning_rate": 5.74131823855921e-05,
      "loss": 6.9405,
      "step": 96
    },
    {
      "epoch": 0.05686648102007914,
      "grad_norm": 3.9330832958221436,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 7.1096,
      "step": 97
    },
    {
      "epoch": 0.05745273340172945,
      "grad_norm": 4.333833694458008,
      "learning_rate": 5.577423184847932e-05,
      "loss": 7.2487,
      "step": 98
    },
    {
      "epoch": 0.058038985783379744,
      "grad_norm": 4.501316070556641,
      "learning_rate": 5.495227651252315e-05,
      "loss": 7.852,
      "step": 99
    },
    {
      "epoch": 0.05862523816503005,
      "grad_norm": 4.297638893127441,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 7.0243,
      "step": 100
    },
    {
      "epoch": 0.05862523816503005,
      "eval_loss": 1.6908631324768066,
      "eval_runtime": 226.9683,
      "eval_samples_per_second": 12.658,
      "eval_steps_per_second": 3.168,
      "step": 100
    },
    {
      "epoch": 0.059211490546680345,
      "grad_norm": 10.180007934570312,
      "learning_rate": 5.330452921628497e-05,
      "loss": 8.4237,
      "step": 101
    },
    {
      "epoch": 0.05979774292833065,
      "grad_norm": 7.255772590637207,
      "learning_rate": 5.247918773366112e-05,
      "loss": 7.707,
      "step": 102
    },
    {
      "epoch": 0.060383995309980946,
      "grad_norm": 2.8105390071868896,
      "learning_rate": 5.165316846586541e-05,
      "loss": 6.4831,
      "step": 103
    },
    {
      "epoch": 0.06097024769163125,
      "grad_norm": 2.000762939453125,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 6.3403,
      "step": 104
    },
    {
      "epoch": 0.06155650007328155,
      "grad_norm": 1.9792354106903076,
      "learning_rate": 5e-05,
      "loss": 5.6166,
      "step": 105
    },
    {
      "epoch": 0.06214275245493185,
      "grad_norm": 2.2282116413116455,
      "learning_rate": 4.917330276168208e-05,
      "loss": 6.9028,
      "step": 106
    },
    {
      "epoch": 0.06272900483658214,
      "grad_norm": 2.1197075843811035,
      "learning_rate": 4.834683153413459e-05,
      "loss": 5.7175,
      "step": 107
    },
    {
      "epoch": 0.06331525721823245,
      "grad_norm": 2.1858396530151367,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 5.9439,
      "step": 108
    },
    {
      "epoch": 0.06390150959988275,
      "grad_norm": 2.093513011932373,
      "learning_rate": 4.669547078371504e-05,
      "loss": 5.7939,
      "step": 109
    },
    {
      "epoch": 0.06448776198153305,
      "grad_norm": 2.5304839611053467,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 6.4176,
      "step": 110
    },
    {
      "epoch": 0.06507401436318334,
      "grad_norm": 2.635596513748169,
      "learning_rate": 4.504772348747687e-05,
      "loss": 6.478,
      "step": 111
    },
    {
      "epoch": 0.06566026674483365,
      "grad_norm": 2.2483584880828857,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 6.2564,
      "step": 112
    },
    {
      "epoch": 0.06624651912648395,
      "grad_norm": 2.1168601512908936,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 6.0574,
      "step": 113
    },
    {
      "epoch": 0.06683277150813426,
      "grad_norm": 2.2489209175109863,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 6.5691,
      "step": 114
    },
    {
      "epoch": 0.06741902388978455,
      "grad_norm": 2.27750563621521,
      "learning_rate": 4.17702704859633e-05,
      "loss": 6.1844,
      "step": 115
    },
    {
      "epoch": 0.06800527627143485,
      "grad_norm": 2.2860987186431885,
      "learning_rate": 4.095597328339452e-05,
      "loss": 6.1519,
      "step": 116
    },
    {
      "epoch": 0.06859152865308515,
      "grad_norm": 2.290987968444824,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 6.5961,
      "step": 117
    },
    {
      "epoch": 0.06917778103473546,
      "grad_norm": 2.3825812339782715,
      "learning_rate": 3.933501846281267e-05,
      "loss": 6.2614,
      "step": 118
    },
    {
      "epoch": 0.06976403341638575,
      "grad_norm": 2.404409408569336,
      "learning_rate": 3.852880399766243e-05,
      "loss": 6.8128,
      "step": 119
    },
    {
      "epoch": 0.07035028579803605,
      "grad_norm": 2.317049503326416,
      "learning_rate": 3.772572564296005e-05,
      "loss": 6.3945,
      "step": 120
    },
    {
      "epoch": 0.07093653817968636,
      "grad_norm": 2.479034900665283,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 6.4291,
      "step": 121
    },
    {
      "epoch": 0.07152279056133666,
      "grad_norm": 2.488262414932251,
      "learning_rate": 3.612985456190778e-05,
      "loss": 6.6535,
      "step": 122
    },
    {
      "epoch": 0.07210904294298695,
      "grad_norm": 2.7070815563201904,
      "learning_rate": 3.533749813077677e-05,
      "loss": 6.4235,
      "step": 123
    },
    {
      "epoch": 0.07269529532463725,
      "grad_norm": 2.623577833175659,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 6.815,
      "step": 124
    },
    {
      "epoch": 0.07328154770628756,
      "grad_norm": 2.7124648094177246,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 6.3577,
      "step": 125
    },
    {
      "epoch": 0.07386780008793786,
      "grad_norm": 2.5578460693359375,
      "learning_rate": 3.298534127791785e-05,
      "loss": 6.6705,
      "step": 126
    },
    {
      "epoch": 0.07445405246958815,
      "grad_norm": 2.7071213722229004,
      "learning_rate": 3.221030765387417e-05,
      "loss": 6.3228,
      "step": 127
    },
    {
      "epoch": 0.07504030485123846,
      "grad_norm": 2.724385976791382,
      "learning_rate": 3.144013755408895e-05,
      "loss": 6.1514,
      "step": 128
    },
    {
      "epoch": 0.07562655723288876,
      "grad_norm": 2.6133148670196533,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 6.0927,
      "step": 129
    },
    {
      "epoch": 0.07621280961453906,
      "grad_norm": 2.5122416019439697,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.5315,
      "step": 130
    },
    {
      "epoch": 0.07679906199618936,
      "grad_norm": 3.1386067867279053,
      "learning_rate": 2.916090697523549e-05,
      "loss": 6.3387,
      "step": 131
    },
    {
      "epoch": 0.07738531437783966,
      "grad_norm": 2.9565436840057373,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 6.5669,
      "step": 132
    },
    {
      "epoch": 0.07797156675948996,
      "grad_norm": 3.0717828273773193,
      "learning_rate": 2.766955965735968e-05,
      "loss": 6.9754,
      "step": 133
    },
    {
      "epoch": 0.07855781914114027,
      "grad_norm": 2.854853630065918,
      "learning_rate": 2.693294185106562e-05,
      "loss": 6.9432,
      "step": 134
    },
    {
      "epoch": 0.07914407152279056,
      "grad_norm": 2.7369515895843506,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 5.9499,
      "step": 135
    },
    {
      "epoch": 0.07973032390444086,
      "grad_norm": 2.9582712650299072,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.0952,
      "step": 136
    },
    {
      "epoch": 0.08031657628609117,
      "grad_norm": 3.1780998706817627,
      "learning_rate": 2.476172311325783e-05,
      "loss": 6.8335,
      "step": 137
    },
    {
      "epoch": 0.08090282866774147,
      "grad_norm": 3.0727474689483643,
      "learning_rate": 2.405152131093926e-05,
      "loss": 6.2819,
      "step": 138
    },
    {
      "epoch": 0.08148908104939176,
      "grad_norm": 3.8813369274139404,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 6.7358,
      "step": 139
    },
    {
      "epoch": 0.08207533343104206,
      "grad_norm": 3.075451612472534,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 6.041,
      "step": 140
    },
    {
      "epoch": 0.08266158581269237,
      "grad_norm": 3.4926958084106445,
      "learning_rate": 2.196424713241637e-05,
      "loss": 6.8879,
      "step": 141
    },
    {
      "epoch": 0.08324783819434266,
      "grad_norm": 3.591356039047241,
      "learning_rate": 2.128356686585282e-05,
      "loss": 6.7446,
      "step": 142
    },
    {
      "epoch": 0.08383409057599296,
      "grad_norm": 3.528189182281494,
      "learning_rate": 2.061073738537635e-05,
      "loss": 6.9325,
      "step": 143
    },
    {
      "epoch": 0.08442034295764327,
      "grad_norm": 3.581657886505127,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 6.967,
      "step": 144
    },
    {
      "epoch": 0.08500659533929357,
      "grad_norm": 3.52889347076416,
      "learning_rate": 1.928936436551661e-05,
      "loss": 6.8557,
      "step": 145
    },
    {
      "epoch": 0.08559284772094386,
      "grad_norm": 3.518007755279541,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 7.0728,
      "step": 146
    },
    {
      "epoch": 0.08617910010259416,
      "grad_norm": 3.710296869277954,
      "learning_rate": 1.800157297483417e-05,
      "loss": 7.4821,
      "step": 147
    },
    {
      "epoch": 0.08676535248424447,
      "grad_norm": 3.8919053077697754,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 7.2119,
      "step": 148
    },
    {
      "epoch": 0.08735160486589477,
      "grad_norm": 3.9789481163024902,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 7.3836,
      "step": 149
    },
    {
      "epoch": 0.08793785724754506,
      "grad_norm": 4.608246803283691,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 8.3462,
      "step": 150
    },
    {
      "epoch": 0.08793785724754506,
      "eval_loss": 1.631693720817566,
      "eval_runtime": 227.0982,
      "eval_samples_per_second": 12.651,
      "eval_steps_per_second": 3.166,
      "step": 150
    },
    {
      "epoch": 0.08852410962919537,
      "grad_norm": 4.644277572631836,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.4804,
      "step": 151
    },
    {
      "epoch": 0.08911036201084567,
      "grad_norm": 4.785318851470947,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 7.111,
      "step": 152
    },
    {
      "epoch": 0.08969661439249597,
      "grad_norm": 4.977895259857178,
      "learning_rate": 1.435357758543015e-05,
      "loss": 6.9439,
      "step": 153
    },
    {
      "epoch": 0.09028286677414626,
      "grad_norm": 3.7211222648620605,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 6.414,
      "step": 154
    },
    {
      "epoch": 0.09086911915579657,
      "grad_norm": 2.7649683952331543,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 6.0524,
      "step": 155
    },
    {
      "epoch": 0.09145537153744687,
      "grad_norm": 2.1293816566467285,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 5.562,
      "step": 156
    },
    {
      "epoch": 0.09204162391909718,
      "grad_norm": 2.197725772857666,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 6.7056,
      "step": 157
    },
    {
      "epoch": 0.09262787630074747,
      "grad_norm": 2.0989270210266113,
      "learning_rate": 1.157994445715706e-05,
      "loss": 5.9067,
      "step": 158
    },
    {
      "epoch": 0.09321412868239777,
      "grad_norm": 2.1401162147521973,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 5.0098,
      "step": 159
    },
    {
      "epoch": 0.09380038106404807,
      "grad_norm": 2.345038414001465,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 5.6253,
      "step": 160
    },
    {
      "epoch": 0.09438663344569838,
      "grad_norm": 2.064814567565918,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 5.4931,
      "step": 161
    },
    {
      "epoch": 0.09497288582734867,
      "grad_norm": 2.0150678157806396,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.7947,
      "step": 162
    },
    {
      "epoch": 0.09555913820899897,
      "grad_norm": 2.36637282371521,
      "learning_rate": 9.068759265665384e-06,
      "loss": 6.0822,
      "step": 163
    },
    {
      "epoch": 0.09614539059064928,
      "grad_norm": 2.2767763137817383,
      "learning_rate": 8.599558442598998e-06,
      "loss": 6.5392,
      "step": 164
    },
    {
      "epoch": 0.09673164297229958,
      "grad_norm": 2.239847421646118,
      "learning_rate": 8.141676086873572e-06,
      "loss": 6.1831,
      "step": 165
    },
    {
      "epoch": 0.09731789535394987,
      "grad_norm": 2.3377020359039307,
      "learning_rate": 7.695237378953223e-06,
      "loss": 6.2192,
      "step": 166
    },
    {
      "epoch": 0.09790414773560017,
      "grad_norm": 2.487506628036499,
      "learning_rate": 7.260364370723044e-06,
      "loss": 5.9929,
      "step": 167
    },
    {
      "epoch": 0.09849040011725048,
      "grad_norm": 2.350316286087036,
      "learning_rate": 6.837175952121306e-06,
      "loss": 6.0822,
      "step": 168
    },
    {
      "epoch": 0.09907665249890078,
      "grad_norm": 2.542607069015503,
      "learning_rate": 6.425787818636131e-06,
      "loss": 6.0007,
      "step": 169
    },
    {
      "epoch": 0.09966290488055107,
      "grad_norm": 2.3959882259368896,
      "learning_rate": 6.026312439675552e-06,
      "loss": 6.0989,
      "step": 170
    },
    {
      "epoch": 0.10024915726220138,
      "grad_norm": 2.9211270809173584,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 6.1464,
      "step": 171
    },
    {
      "epoch": 0.10083540964385168,
      "grad_norm": 2.5673844814300537,
      "learning_rate": 5.263533508961827e-06,
      "loss": 6.3327,
      "step": 172
    },
    {
      "epoch": 0.10142166202550199,
      "grad_norm": 2.360288619995117,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.8837,
      "step": 173
    },
    {
      "epoch": 0.10200791440715228,
      "grad_norm": 2.5725009441375732,
      "learning_rate": 4.549673247541875e-06,
      "loss": 6.207,
      "step": 174
    },
    {
      "epoch": 0.10259416678880258,
      "grad_norm": 2.560316801071167,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 6.0591,
      "step": 175
    },
    {
      "epoch": 0.10318041917045288,
      "grad_norm": 2.5278682708740234,
      "learning_rate": 3.885512251130763e-06,
      "loss": 6.532,
      "step": 176
    },
    {
      "epoch": 0.10376667155210317,
      "grad_norm": 2.692450761795044,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 6.6783,
      "step": 177
    },
    {
      "epoch": 0.10435292393375348,
      "grad_norm": 2.8818061351776123,
      "learning_rate": 3.271776770026963e-06,
      "loss": 7.0181,
      "step": 178
    },
    {
      "epoch": 0.10493917631540378,
      "grad_norm": 2.8357906341552734,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 7.0833,
      "step": 179
    },
    {
      "epoch": 0.10552542869705409,
      "grad_norm": 2.8222532272338867,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 6.5395,
      "step": 180
    },
    {
      "epoch": 0.10611168107870438,
      "grad_norm": 2.79948353767395,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 6.7513,
      "step": 181
    },
    {
      "epoch": 0.10669793346035468,
      "grad_norm": 2.742783546447754,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 6.7368,
      "step": 182
    },
    {
      "epoch": 0.10728418584200498,
      "grad_norm": 2.8662562370300293,
      "learning_rate": 1.962316193157593e-06,
      "loss": 6.0122,
      "step": 183
    },
    {
      "epoch": 0.10787043822365529,
      "grad_norm": 2.9705612659454346,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.7583,
      "step": 184
    },
    {
      "epoch": 0.10845669060530558,
      "grad_norm": 2.8730576038360596,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.0586,
      "step": 185
    },
    {
      "epoch": 0.10904294298695588,
      "grad_norm": 3.270841121673584,
      "learning_rate": 1.333670137599713e-06,
      "loss": 6.9656,
      "step": 186
    },
    {
      "epoch": 0.10962919536860619,
      "grad_norm": 3.0216994285583496,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 6.4309,
      "step": 187
    },
    {
      "epoch": 0.11021544775025649,
      "grad_norm": 3.350586414337158,
      "learning_rate": 9.810017062595322e-07,
      "loss": 7.2117,
      "step": 188
    },
    {
      "epoch": 0.11080170013190678,
      "grad_norm": 3.269277811050415,
      "learning_rate": 8.247462563808817e-07,
      "loss": 7.1121,
      "step": 189
    },
    {
      "epoch": 0.11138795251355708,
      "grad_norm": 3.1367297172546387,
      "learning_rate": 6.819348298638839e-07,
      "loss": 6.5725,
      "step": 190
    },
    {
      "epoch": 0.11197420489520739,
      "grad_norm": 3.2576162815093994,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.8892,
      "step": 191
    },
    {
      "epoch": 0.11256045727685769,
      "grad_norm": 3.5447022914886475,
      "learning_rate": 4.367965336512403e-07,
      "loss": 7.2643,
      "step": 192
    },
    {
      "epoch": 0.11314670965850798,
      "grad_norm": 3.4383511543273926,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 6.7315,
      "step": 193
    },
    {
      "epoch": 0.11373296204015829,
      "grad_norm": 3.5435619354248047,
      "learning_rate": 2.458548727494292e-07,
      "loss": 6.8162,
      "step": 194
    },
    {
      "epoch": 0.11431921442180859,
      "grad_norm": 3.544842004776001,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 6.9079,
      "step": 195
    },
    {
      "epoch": 0.1149054668034589,
      "grad_norm": 3.6281445026397705,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 6.7825,
      "step": 196
    },
    {
      "epoch": 0.11549171918510918,
      "grad_norm": 3.8765647411346436,
      "learning_rate": 6.150154258476315e-08,
      "loss": 7.7639,
      "step": 197
    },
    {
      "epoch": 0.11607797156675949,
      "grad_norm": 3.739752769470215,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 6.9701,
      "step": 198
    },
    {
      "epoch": 0.11666422394840979,
      "grad_norm": 5.229872226715088,
      "learning_rate": 6.834750376549792e-09,
      "loss": 7.4868,
      "step": 199
    },
    {
      "epoch": 0.1172504763300601,
      "grad_norm": 4.8796491622924805,
      "learning_rate": 0.0,
      "loss": 7.6656,
      "step": 200
    },
    {
      "epoch": 0.1172504763300601,
      "eval_loss": 1.6063930988311768,
      "eval_runtime": 227.0847,
      "eval_samples_per_second": 12.652,
      "eval_steps_per_second": 3.166,
      "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
}