{
  "best_metric": 1.5060151815414429,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.06749915626054674,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003374957813027337,
      "grad_norm": 3.6639511585235596,
      "learning_rate": 1e-05,
      "loss": 2.505,
      "step": 1
    },
    {
      "epoch": 0.0003374957813027337,
      "eval_loss": 3.0136096477508545,
      "eval_runtime": 146.8649,
      "eval_samples_per_second": 33.977,
      "eval_steps_per_second": 8.498,
      "step": 1
    },
    {
      "epoch": 0.0006749915626054674,
      "grad_norm": 3.381553888320923,
      "learning_rate": 2e-05,
      "loss": 2.5705,
      "step": 2
    },
    {
      "epoch": 0.0010124873439082012,
      "grad_norm": 3.0773203372955322,
      "learning_rate": 3e-05,
      "loss": 2.4593,
      "step": 3
    },
    {
      "epoch": 0.0013499831252109348,
      "grad_norm": 3.1308059692382812,
      "learning_rate": 4e-05,
      "loss": 2.5445,
      "step": 4
    },
    {
      "epoch": 0.0016874789065136687,
      "grad_norm": 3.124946355819702,
      "learning_rate": 5e-05,
      "loss": 2.3191,
      "step": 5
    },
    {
      "epoch": 0.0020249746878164025,
      "grad_norm": 2.9933860301971436,
      "learning_rate": 6e-05,
      "loss": 2.3489,
      "step": 6
    },
    {
      "epoch": 0.002362470469119136,
      "grad_norm": 3.1016297340393066,
      "learning_rate": 7e-05,
      "loss": 2.0804,
      "step": 7
    },
    {
      "epoch": 0.0026999662504218697,
      "grad_norm": 2.968693494796753,
      "learning_rate": 8e-05,
      "loss": 1.8521,
      "step": 8
    },
    {
      "epoch": 0.0030374620317246033,
      "grad_norm": 2.330232620239258,
      "learning_rate": 9e-05,
      "loss": 1.623,
      "step": 9
    },
    {
      "epoch": 0.0033749578130273373,
      "grad_norm": 2.491941213607788,
      "learning_rate": 0.0001,
      "loss": 1.4373,
      "step": 10
    },
    {
      "epoch": 0.003712453594330071,
      "grad_norm": 2.9340157508850098,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.784,
      "step": 11
    },
    {
      "epoch": 0.004049949375632805,
      "grad_norm": 2.1506173610687256,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.7149,
      "step": 12
    },
    {
      "epoch": 0.004387445156935539,
      "grad_norm": 2.4897687435150146,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.5804,
      "step": 13
    },
    {
      "epoch": 0.004724940938238272,
      "grad_norm": 1.9792346954345703,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.4783,
      "step": 14
    },
    {
      "epoch": 0.005062436719541006,
      "grad_norm": 1.7974101305007935,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.5524,
      "step": 15
    },
    {
      "epoch": 0.005399932500843739,
      "grad_norm": 1.7417069673538208,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7657,
      "step": 16
    },
    {
      "epoch": 0.005737428282146473,
      "grad_norm": 1.973816990852356,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5573,
      "step": 17
    },
    {
      "epoch": 0.006074924063449207,
      "grad_norm": 1.8495699167251587,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.7699,
      "step": 18
    },
    {
      "epoch": 0.00641241984475194,
      "grad_norm": 1.7429189682006836,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.5887,
      "step": 19
    },
    {
      "epoch": 0.006749915626054675,
      "grad_norm": 1.7235437631607056,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.7197,
      "step": 20
    },
    {
      "epoch": 0.007087411407357408,
      "grad_norm": 1.7469090223312378,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6879,
      "step": 21
    },
    {
      "epoch": 0.007424907188660142,
      "grad_norm": 1.53190016746521,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3991,
      "step": 22
    },
    {
      "epoch": 0.0077624029699628755,
      "grad_norm": 1.7227346897125244,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.481,
      "step": 23
    },
    {
      "epoch": 0.00809989875126561,
      "grad_norm": 1.6486566066741943,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6684,
      "step": 24
    },
    {
      "epoch": 0.008437394532568343,
      "grad_norm": 1.7526777982711792,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.7031,
      "step": 25
    },
    {
      "epoch": 0.008774890313871077,
      "grad_norm": 1.7023621797561646,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.6891,
      "step": 26
    },
    {
      "epoch": 0.00911238609517381,
      "grad_norm": 1.6384953260421753,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4134,
      "step": 27
    },
    {
      "epoch": 0.009449881876476544,
      "grad_norm": 1.6035668849945068,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.3459,
      "step": 28
    },
    {
      "epoch": 0.009787377657779277,
      "grad_norm": 1.809048056602478,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5018,
      "step": 29
    },
    {
      "epoch": 0.010124873439082012,
      "grad_norm": 1.7149184942245483,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.4484,
      "step": 30
    },
    {
      "epoch": 0.010462369220384746,
      "grad_norm": 1.6432181596755981,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4212,
      "step": 31
    },
    {
      "epoch": 0.010799865001687479,
      "grad_norm": 1.670172929763794,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.806,
      "step": 32
    },
    {
      "epoch": 0.011137360782990213,
      "grad_norm": 1.539548635482788,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5579,
      "step": 33
    },
    {
      "epoch": 0.011474856564292946,
      "grad_norm": 1.515500545501709,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4572,
      "step": 34
    },
    {
      "epoch": 0.01181235234559568,
      "grad_norm": 1.740642786026001,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.6059,
      "step": 35
    },
    {
      "epoch": 0.012149848126898413,
      "grad_norm": 1.6643253564834595,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5727,
      "step": 36
    },
    {
      "epoch": 0.012487343908201148,
      "grad_norm": 1.803133487701416,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.7534,
      "step": 37
    },
    {
      "epoch": 0.01282483968950388,
      "grad_norm": 1.6213222742080688,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3928,
      "step": 38
    },
    {
      "epoch": 0.013162335470806615,
      "grad_norm": 1.5043983459472656,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.7375,
      "step": 39
    },
    {
      "epoch": 0.01349983125210935,
      "grad_norm": 1.5561853647232056,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4826,
      "step": 40
    },
    {
      "epoch": 0.013837327033412082,
      "grad_norm": 2.0976758003234863,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.8666,
      "step": 41
    },
    {
      "epoch": 0.014174822814714817,
      "grad_norm": 2.003178119659424,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5683,
      "step": 42
    },
    {
      "epoch": 0.01451231859601755,
      "grad_norm": 1.6325174570083618,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.8515,
      "step": 43
    },
    {
      "epoch": 0.014849814377320284,
      "grad_norm": 1.5984230041503906,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.6549,
      "step": 44
    },
    {
      "epoch": 0.015187310158623017,
      "grad_norm": 1.905388593673706,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6744,
      "step": 45
    },
    {
      "epoch": 0.015524805939925751,
      "grad_norm": 1.7882310152053833,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.609,
      "step": 46
    },
    {
      "epoch": 0.015862301721228485,
      "grad_norm": 1.782175898551941,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5189,
      "step": 47
    },
    {
      "epoch": 0.01619979750253122,
      "grad_norm": 1.7065168619155884,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4361,
      "step": 48
    },
    {
      "epoch": 0.01653729328383395,
      "grad_norm": 1.6321532726287842,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.6779,
      "step": 49
    },
    {
      "epoch": 0.016874789065136685,
      "grad_norm": 1.9198822975158691,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5329,
      "step": 50
    },
    {
      "epoch": 0.016874789065136685,
      "eval_loss": 1.5430727005004883,
      "eval_runtime": 147.0853,
      "eval_samples_per_second": 33.926,
      "eval_steps_per_second": 8.485,
      "step": 50
    },
    {
      "epoch": 0.01721228484643942,
      "grad_norm": 1.4228336811065674,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.6601,
      "step": 51
    },
    {
      "epoch": 0.017549780627742154,
      "grad_norm": 1.322295069694519,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2419,
      "step": 52
    },
    {
      "epoch": 0.017887276409044885,
      "grad_norm": 1.445678472518921,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.6227,
      "step": 53
    },
    {
      "epoch": 0.01822477219034762,
      "grad_norm": 1.3230798244476318,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.4315,
      "step": 54
    },
    {
      "epoch": 0.018562267971650354,
      "grad_norm": 1.3651715517044067,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5568,
      "step": 55
    },
    {
      "epoch": 0.01889976375295309,
      "grad_norm": 1.269474983215332,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5887,
      "step": 56
    },
    {
      "epoch": 0.019237259534255823,
      "grad_norm": 1.4516477584838867,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4599,
      "step": 57
    },
    {
      "epoch": 0.019574755315558554,
      "grad_norm": 1.282692313194275,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4515,
      "step": 58
    },
    {
      "epoch": 0.01991225109686129,
      "grad_norm": 1.5570868253707886,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5617,
      "step": 59
    },
    {
      "epoch": 0.020249746878164023,
      "grad_norm": 1.2447081804275513,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.368,
      "step": 60
    },
    {
      "epoch": 0.020587242659466758,
      "grad_norm": 1.3168991804122925,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4655,
      "step": 61
    },
    {
      "epoch": 0.020924738440769492,
      "grad_norm": 1.3223047256469727,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5246,
      "step": 62
    },
    {
      "epoch": 0.021262234222072223,
      "grad_norm": 1.4290637969970703,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.6008,
      "step": 63
    },
    {
      "epoch": 0.021599730003374958,
      "grad_norm": 1.4363245964050293,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5222,
      "step": 64
    },
    {
      "epoch": 0.021937225784677692,
      "grad_norm": 1.4145311117172241,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3268,
      "step": 65
    },
    {
      "epoch": 0.022274721565980427,
      "grad_norm": 1.368954062461853,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3751,
      "step": 66
    },
    {
      "epoch": 0.022612217347283158,
      "grad_norm": 1.4794690608978271,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.604,
      "step": 67
    },
    {
      "epoch": 0.022949713128585892,
      "grad_norm": 1.5086948871612549,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4682,
      "step": 68
    },
    {
      "epoch": 0.023287208909888626,
      "grad_norm": 1.3436882495880127,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5351,
      "step": 69
    },
    {
      "epoch": 0.02362470469119136,
      "grad_norm": 1.5566866397857666,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6446,
      "step": 70
    },
    {
      "epoch": 0.023962200472494095,
      "grad_norm": 1.3160170316696167,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.3528,
      "step": 71
    },
    {
      "epoch": 0.024299696253796826,
      "grad_norm": 1.5016075372695923,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.4308,
      "step": 72
    },
    {
      "epoch": 0.02463719203509956,
      "grad_norm": 1.5386384725570679,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5839,
      "step": 73
    },
    {
      "epoch": 0.024974687816402295,
      "grad_norm": 1.502233624458313,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.6412,
      "step": 74
    },
    {
      "epoch": 0.02531218359770503,
      "grad_norm": 1.5245769023895264,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4053,
      "step": 75
    },
    {
      "epoch": 0.02564967937900776,
      "grad_norm": 1.3348147869110107,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.279,
      "step": 76
    },
    {
      "epoch": 0.025987175160310495,
      "grad_norm": 1.5596102476119995,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.6353,
      "step": 77
    },
    {
      "epoch": 0.02632467094161323,
      "grad_norm": 1.695809245109558,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.5378,
      "step": 78
    },
    {
      "epoch": 0.026662166722915964,
      "grad_norm": 1.4951751232147217,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.7345,
      "step": 79
    },
    {
      "epoch": 0.0269996625042187,
      "grad_norm": 1.2774806022644043,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3024,
      "step": 80
    },
    {
      "epoch": 0.02733715828552143,
      "grad_norm": 1.4456425905227661,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.5536,
      "step": 81
    },
    {
      "epoch": 0.027674654066824164,
      "grad_norm": 1.423951506614685,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.5473,
      "step": 82
    },
    {
      "epoch": 0.0280121498481269,
      "grad_norm": 1.4260379076004028,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.5229,
      "step": 83
    },
    {
      "epoch": 0.028349645629429633,
      "grad_norm": 1.3661842346191406,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4443,
      "step": 84
    },
    {
      "epoch": 0.028687141410732364,
      "grad_norm": 1.4436497688293457,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5462,
      "step": 85
    },
    {
      "epoch": 0.0290246371920351,
      "grad_norm": 1.3736367225646973,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4586,
      "step": 86
    },
    {
      "epoch": 0.029362132973337833,
      "grad_norm": 1.6109968423843384,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.6427,
      "step": 87
    },
    {
      "epoch": 0.029699628754640568,
      "grad_norm": 1.4181476831436157,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4003,
      "step": 88
    },
    {
      "epoch": 0.030037124535943302,
      "grad_norm": 1.3104472160339355,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.4153,
      "step": 89
    },
    {
      "epoch": 0.030374620317246033,
      "grad_norm": 1.6174646615982056,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.5468,
      "step": 90
    },
    {
      "epoch": 0.030712116098548767,
      "grad_norm": 1.6859798431396484,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.5389,
      "step": 91
    },
    {
      "epoch": 0.031049611879851502,
      "grad_norm": 1.3607832193374634,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2407,
      "step": 92
    },
    {
      "epoch": 0.03138710766115423,
      "grad_norm": 1.3353404998779297,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3828,
      "step": 93
    },
    {
      "epoch": 0.03172460344245697,
      "grad_norm": 1.4945926666259766,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5571,
      "step": 94
    },
    {
      "epoch": 0.0320620992237597,
      "grad_norm": 1.5455503463745117,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.5421,
      "step": 95
    },
    {
      "epoch": 0.03239959500506244,
      "grad_norm": 1.3872216939926147,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.5206,
      "step": 96
    },
    {
      "epoch": 0.03273709078636517,
      "grad_norm": 1.7368088960647583,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.7135,
      "step": 97
    },
    {
      "epoch": 0.0330745865676679,
      "grad_norm": 1.4623812437057495,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.437,
      "step": 98
    },
    {
      "epoch": 0.03341208234897064,
      "grad_norm": 1.4175429344177246,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2106,
      "step": 99
    },
    {
      "epoch": 0.03374957813027337,
      "grad_norm": 2.094141960144043,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.578,
      "step": 100
    },
    {
      "epoch": 0.03374957813027337,
      "eval_loss": 1.5353269577026367,
      "eval_runtime": 147.1368,
      "eval_samples_per_second": 33.914,
      "eval_steps_per_second": 8.482,
      "step": 100
    },
    {
      "epoch": 0.03408707391157611,
      "grad_norm": 1.4249728918075562,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.293,
      "step": 101
    },
    {
      "epoch": 0.03442456969287884,
      "grad_norm": 1.43645179271698,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.5217,
      "step": 102
    },
    {
      "epoch": 0.03476206547418157,
      "grad_norm": 1.549728512763977,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.6349,
      "step": 103
    },
    {
      "epoch": 0.03509956125548431,
      "grad_norm": 1.386811375617981,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5743,
      "step": 104
    },
    {
      "epoch": 0.03543705703678704,
      "grad_norm": 1.32630455493927,
      "learning_rate": 5e-05,
      "loss": 1.385,
      "step": 105
    },
    {
      "epoch": 0.03577455281808977,
      "grad_norm": 1.1966615915298462,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1981,
      "step": 106
    },
    {
      "epoch": 0.03611204859939251,
      "grad_norm": 1.3795210123062134,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.4195,
      "step": 107
    },
    {
      "epoch": 0.03644954438069524,
      "grad_norm": 1.3812130689620972,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3496,
      "step": 108
    },
    {
      "epoch": 0.03678704016199798,
      "grad_norm": 1.5323227643966675,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.6077,
      "step": 109
    },
    {
      "epoch": 0.03712453594330071,
      "grad_norm": 1.5685880184173584,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5661,
      "step": 110
    },
    {
      "epoch": 0.03746203172460344,
      "grad_norm": 1.2461915016174316,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.3674,
      "step": 111
    },
    {
      "epoch": 0.03779952750590618,
      "grad_norm": 1.4735571146011353,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4838,
      "step": 112
    },
    {
      "epoch": 0.03813702328720891,
      "grad_norm": 1.2594475746154785,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2772,
      "step": 113
    },
    {
      "epoch": 0.038474519068511646,
      "grad_norm": 1.355596899986267,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.5189,
      "step": 114
    },
    {
      "epoch": 0.03881201484981438,
      "grad_norm": 1.5534967184066772,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.478,
      "step": 115
    },
    {
      "epoch": 0.03914951063111711,
      "grad_norm": 1.2867431640625,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3196,
      "step": 116
    },
    {
      "epoch": 0.039487006412419846,
      "grad_norm": 1.7533015012741089,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9202,
      "step": 117
    },
    {
      "epoch": 0.03982450219372258,
      "grad_norm": 1.3565618991851807,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.4292,
      "step": 118
    },
    {
      "epoch": 0.040161997975025315,
      "grad_norm": 1.397993803024292,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.6807,
      "step": 119
    },
    {
      "epoch": 0.040499493756328046,
      "grad_norm": 1.3319307565689087,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.5143,
      "step": 120
    },
    {
      "epoch": 0.04083698953763078,
      "grad_norm": 1.321202278137207,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3522,
      "step": 121
    },
    {
      "epoch": 0.041174485318933515,
      "grad_norm": 1.4271525144577026,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.5424,
      "step": 122
    },
    {
      "epoch": 0.041511981100236246,
      "grad_norm": 1.3564966917037964,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.7958,
      "step": 123
    },
    {
      "epoch": 0.041849476881538984,
      "grad_norm": 1.2295968532562256,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4234,
      "step": 124
    },
    {
      "epoch": 0.042186972662841715,
      "grad_norm": 1.4020897150039673,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.4671,
      "step": 125
    },
    {
      "epoch": 0.042524468444144446,
      "grad_norm": 1.62809419631958,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.8034,
      "step": 126
    },
    {
      "epoch": 0.042861964225447184,
      "grad_norm": 1.267198085784912,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4299,
      "step": 127
    },
    {
      "epoch": 0.043199460006749915,
      "grad_norm": 1.31877601146698,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.551,
      "step": 128
    },
    {
      "epoch": 0.043536955788052646,
      "grad_norm": 1.3092563152313232,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4684,
      "step": 129
    },
    {
      "epoch": 0.043874451569355384,
      "grad_norm": 1.5711746215820312,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2988,
      "step": 130
    },
    {
      "epoch": 0.044211947350658115,
      "grad_norm": 1.3491564989089966,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.4796,
      "step": 131
    },
    {
      "epoch": 0.04454944313196085,
      "grad_norm": 1.4088948965072632,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4395,
      "step": 132
    },
    {
      "epoch": 0.044886938913263584,
      "grad_norm": 1.2712147235870361,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4685,
      "step": 133
    },
    {
      "epoch": 0.045224434694566315,
      "grad_norm": 1.2743107080459595,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.3902,
      "step": 134
    },
    {
      "epoch": 0.04556193047586905,
      "grad_norm": 1.3566430807113647,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.518,
      "step": 135
    },
    {
      "epoch": 0.045899426257171784,
      "grad_norm": 1.4612895250320435,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.5552,
      "step": 136
    },
    {
      "epoch": 0.04623692203847452,
      "grad_norm": 1.4626133441925049,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.5718,
      "step": 137
    },
    {
      "epoch": 0.04657441781977725,
      "grad_norm": 1.3682875633239746,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.3619,
      "step": 138
    },
    {
      "epoch": 0.046911913601079984,
      "grad_norm": 1.4876656532287598,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5829,
      "step": 139
    },
    {
      "epoch": 0.04724940938238272,
      "grad_norm": 1.537940502166748,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.5279,
      "step": 140
    },
    {
      "epoch": 0.04758690516368545,
      "grad_norm": 1.0990314483642578,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.1876,
      "step": 141
    },
    {
      "epoch": 0.04792440094498819,
      "grad_norm": 1.5663527250289917,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.7232,
      "step": 142
    },
    {
      "epoch": 0.04826189672629092,
      "grad_norm": 1.3516157865524292,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.5107,
      "step": 143
    },
    {
      "epoch": 0.04859939250759365,
      "grad_norm": 1.4804091453552246,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4559,
      "step": 144
    },
    {
      "epoch": 0.04893688828889639,
      "grad_norm": 1.5756516456604004,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5922,
      "step": 145
    },
    {
      "epoch": 0.04927438407019912,
      "grad_norm": 1.319180965423584,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4902,
      "step": 146
    },
    {
      "epoch": 0.04961187985150186,
      "grad_norm": 1.4037140607833862,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.3511,
      "step": 147
    },
    {
      "epoch": 0.04994937563280459,
      "grad_norm": 1.6963108777999878,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5755,
      "step": 148
    },
    {
      "epoch": 0.05028687141410732,
      "grad_norm": 1.6080647706985474,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.6222,
      "step": 149
    },
    {
      "epoch": 0.05062436719541006,
      "grad_norm": 1.8582013845443726,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.5842,
      "step": 150
    },
    {
      "epoch": 0.05062436719541006,
      "eval_loss": 1.5126420259475708,
      "eval_runtime": 147.1818,
      "eval_samples_per_second": 33.904,
      "eval_steps_per_second": 8.479,
      "step": 150
    },
    {
      "epoch": 0.05096186297671279,
      "grad_norm": 1.1610660552978516,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4331,
      "step": 151
    },
    {
      "epoch": 0.05129935875801552,
      "grad_norm": 1.2772867679595947,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.5064,
      "step": 152
    },
    {
      "epoch": 0.05163685453931826,
      "grad_norm": 1.210777759552002,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3817,
      "step": 153
    },
    {
      "epoch": 0.05197435032062099,
      "grad_norm": 1.3157647848129272,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.6171,
      "step": 154
    },
    {
      "epoch": 0.05231184610192373,
      "grad_norm": 1.3022435903549194,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4746,
      "step": 155
    },
    {
      "epoch": 0.05264934188322646,
      "grad_norm": 1.1724581718444824,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.3533,
      "step": 156
    },
    {
      "epoch": 0.05298683766452919,
      "grad_norm": 1.2420326471328735,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.4836,
      "step": 157
    },
    {
      "epoch": 0.05332433344583193,
      "grad_norm": 1.197049617767334,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.3554,
      "step": 158
    },
    {
      "epoch": 0.05366182922713466,
      "grad_norm": 1.3101555109024048,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3777,
      "step": 159
    },
    {
      "epoch": 0.0539993250084374,
      "grad_norm": 1.3743592500686646,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.6413,
      "step": 160
    },
    {
      "epoch": 0.05433682078974013,
      "grad_norm": 1.3461825847625732,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5547,
      "step": 161
    },
    {
      "epoch": 0.05467431657104286,
      "grad_norm": 1.2572141885757446,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4351,
      "step": 162
    },
    {
      "epoch": 0.0550118123523456,
      "grad_norm": 1.2942315340042114,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.6654,
      "step": 163
    },
    {
      "epoch": 0.05534930813364833,
      "grad_norm": 1.208166480064392,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4456,
      "step": 164
    },
    {
      "epoch": 0.055686803914951066,
      "grad_norm": 1.3001453876495361,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.6141,
      "step": 165
    },
    {
      "epoch": 0.0560242996962538,
      "grad_norm": 1.3971080780029297,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.5421,
      "step": 166
    },
    {
      "epoch": 0.05636179547755653,
      "grad_norm": 1.2448586225509644,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4295,
      "step": 167
    },
    {
      "epoch": 0.056699291258859266,
      "grad_norm": 1.258288025856018,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.3398,
      "step": 168
    },
    {
      "epoch": 0.057036787040162,
      "grad_norm": 1.4159022569656372,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4889,
      "step": 169
    },
    {
      "epoch": 0.05737428282146473,
      "grad_norm": 1.2871901988983154,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.435,
      "step": 170
    },
    {
      "epoch": 0.057711778602767466,
      "grad_norm": 1.4442795515060425,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.534,
      "step": 171
    },
    {
      "epoch": 0.0580492743840702,
      "grad_norm": 1.3893766403198242,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.6793,
      "step": 172
    },
    {
      "epoch": 0.058386770165372935,
      "grad_norm": 1.2522118091583252,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.4804,
      "step": 173
    },
    {
      "epoch": 0.058724265946675666,
      "grad_norm": 1.1943901777267456,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3358,
      "step": 174
    },
    {
      "epoch": 0.0590617617279784,
      "grad_norm": 1.4615720510482788,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.5894,
      "step": 175
    },
    {
      "epoch": 0.059399257509281135,
      "grad_norm": 1.5168812274932861,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.6026,
      "step": 176
    },
    {
      "epoch": 0.059736753290583866,
      "grad_norm": 1.3174370527267456,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.4032,
      "step": 177
    },
    {
      "epoch": 0.060074249071886604,
      "grad_norm": 1.3193551301956177,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.4026,
      "step": 178
    },
    {
      "epoch": 0.060411744853189335,
      "grad_norm": 1.5137648582458496,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.7196,
      "step": 179
    },
    {
      "epoch": 0.060749240634492066,
      "grad_norm": 1.473146677017212,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.6671,
      "step": 180
    },
    {
      "epoch": 0.061086736415794804,
      "grad_norm": 1.3389666080474854,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.6202,
      "step": 181
    },
    {
      "epoch": 0.061424232197097535,
      "grad_norm": 1.4197286367416382,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5409,
      "step": 182
    },
    {
      "epoch": 0.06176172797840027,
      "grad_norm": 1.6522517204284668,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.7478,
      "step": 183
    },
    {
      "epoch": 0.062099223759703004,
      "grad_norm": 1.316051959991455,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.2369,
      "step": 184
    },
    {
      "epoch": 0.062436719541005735,
      "grad_norm": 1.2479242086410522,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.3954,
      "step": 185
    },
    {
      "epoch": 0.06277421532230847,
      "grad_norm": 1.4464031457901,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.317,
      "step": 186
    },
    {
      "epoch": 0.0631117111036112,
      "grad_norm": 1.4399216175079346,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.501,
      "step": 187
    },
    {
      "epoch": 0.06344920688491394,
      "grad_norm": 1.2375539541244507,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.539,
      "step": 188
    },
    {
      "epoch": 0.06378670266621667,
      "grad_norm": 1.3177428245544434,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.3098,
      "step": 189
    },
    {
      "epoch": 0.0641241984475194,
      "grad_norm": 1.4911314249038696,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.5175,
      "step": 190
    },
    {
      "epoch": 0.06446169422882214,
      "grad_norm": 1.189093828201294,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.174,
      "step": 191
    },
    {
      "epoch": 0.06479919001012488,
      "grad_norm": 1.3943027257919312,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.5491,
      "step": 192
    },
    {
      "epoch": 0.0651366857914276,
      "grad_norm": 1.5117491483688354,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.7054,
      "step": 193
    },
    {
      "epoch": 0.06547418157273034,
      "grad_norm": 1.346320629119873,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3933,
      "step": 194
    },
    {
      "epoch": 0.06581167735403308,
      "grad_norm": 1.259371042251587,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.4043,
      "step": 195
    },
    {
      "epoch": 0.0661491731353358,
      "grad_norm": 1.426383137702942,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.5046,
      "step": 196
    },
    {
      "epoch": 0.06648666891663854,
      "grad_norm": 1.5296632051467896,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.4275,
      "step": 197
    },
    {
      "epoch": 0.06682416469794128,
      "grad_norm": 1.358502984046936,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.4823,
      "step": 198
    },
    {
      "epoch": 0.067161660479244,
      "grad_norm": 1.2938090562820435,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.3648,
      "step": 199
    },
    {
      "epoch": 0.06749915626054674,
      "grad_norm": 1.559105634689331,
      "learning_rate": 0.0,
      "loss": 1.5589,
      "step": 200
    },
    {
      "epoch": 0.06749915626054674,
      "eval_loss": 1.5060151815414429,
      "eval_runtime": 147.1799,
      "eval_samples_per_second": 33.904,
      "eval_steps_per_second": 8.479,
      "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": 5.449698231975936e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}