{
  "best_metric": 1.253917932510376,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.21580547112462,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0060790273556231,
      "grad_norm": 1.761592984199524,
      "learning_rate": 1e-05,
      "loss": 1.9276,
      "step": 1
    },
    {
      "epoch": 0.0060790273556231,
      "eval_loss": 2.0459976196289062,
      "eval_runtime": 30.9989,
      "eval_samples_per_second": 8.936,
      "eval_steps_per_second": 2.258,
      "step": 1
    },
    {
      "epoch": 0.0121580547112462,
      "grad_norm": 1.9396612644195557,
      "learning_rate": 2e-05,
      "loss": 1.9162,
      "step": 2
    },
    {
      "epoch": 0.0182370820668693,
      "grad_norm": 1.9415793418884277,
      "learning_rate": 3e-05,
      "loss": 2.0918,
      "step": 3
    },
    {
      "epoch": 0.0243161094224924,
      "grad_norm": 1.6954610347747803,
      "learning_rate": 4e-05,
      "loss": 2.1201,
      "step": 4
    },
    {
      "epoch": 0.030395136778115502,
      "grad_norm": 1.7460685968399048,
      "learning_rate": 5e-05,
      "loss": 2.0101,
      "step": 5
    },
    {
      "epoch": 0.0364741641337386,
      "grad_norm": 1.6817845106124878,
      "learning_rate": 6e-05,
      "loss": 1.8639,
      "step": 6
    },
    {
      "epoch": 0.0425531914893617,
      "grad_norm": 1.6025727987289429,
      "learning_rate": 7e-05,
      "loss": 1.7984,
      "step": 7
    },
    {
      "epoch": 0.0486322188449848,
      "grad_norm": 1.9342198371887207,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 8
    },
    {
      "epoch": 0.0547112462006079,
      "grad_norm": 2.2323875427246094,
      "learning_rate": 9e-05,
      "loss": 1.7929,
      "step": 9
    },
    {
      "epoch": 0.060790273556231005,
      "grad_norm": 1.7553737163543701,
      "learning_rate": 0.0001,
      "loss": 1.6588,
      "step": 10
    },
    {
      "epoch": 0.0668693009118541,
      "grad_norm": 1.6046030521392822,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5522,
      "step": 11
    },
    {
      "epoch": 0.0729483282674772,
      "grad_norm": 1.7818574905395508,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.6286,
      "step": 12
    },
    {
      "epoch": 0.0790273556231003,
      "grad_norm": 1.5256266593933105,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4941,
      "step": 13
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": 1.517819881439209,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5374,
      "step": 14
    },
    {
      "epoch": 0.0911854103343465,
      "grad_norm": 1.4997202157974243,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4758,
      "step": 15
    },
    {
      "epoch": 0.0972644376899696,
      "grad_norm": 1.592433214187622,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6169,
      "step": 16
    },
    {
      "epoch": 0.1033434650455927,
      "grad_norm": 1.3706495761871338,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6543,
      "step": 17
    },
    {
      "epoch": 0.1094224924012158,
      "grad_norm": 1.4116281270980835,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.6443,
      "step": 18
    },
    {
      "epoch": 0.11550151975683891,
      "grad_norm": 1.6152551174163818,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.8082,
      "step": 19
    },
    {
      "epoch": 0.12158054711246201,
      "grad_norm": 1.4601709842681885,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5163,
      "step": 20
    },
    {
      "epoch": 0.1276595744680851,
      "grad_norm": 1.6320929527282715,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6623,
      "step": 21
    },
    {
      "epoch": 0.1337386018237082,
      "grad_norm": 1.6018469333648682,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.7458,
      "step": 22
    },
    {
      "epoch": 0.1398176291793313,
      "grad_norm": 1.3183369636535645,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.505,
      "step": 23
    },
    {
      "epoch": 0.1458966565349544,
      "grad_norm": 1.3536615371704102,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.2991,
      "step": 24
    },
    {
      "epoch": 0.1519756838905775,
      "grad_norm": 1.2052907943725586,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.3043,
      "step": 25
    },
    {
      "epoch": 0.1580547112462006,
      "grad_norm": 1.3558847904205322,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.3127,
      "step": 26
    },
    {
      "epoch": 0.1641337386018237,
      "grad_norm": 1.2103129625320435,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2627,
      "step": 27
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 1.1874868869781494,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.2123,
      "step": 28
    },
    {
      "epoch": 0.1762917933130699,
      "grad_norm": 1.2453527450561523,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2254,
      "step": 29
    },
    {
      "epoch": 0.182370820668693,
      "grad_norm": 1.1241514682769775,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.2046,
      "step": 30
    },
    {
      "epoch": 0.1884498480243161,
      "grad_norm": 1.121284008026123,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.254,
      "step": 31
    },
    {
      "epoch": 0.1945288753799392,
      "grad_norm": 1.3332960605621338,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2834,
      "step": 32
    },
    {
      "epoch": 0.2006079027355623,
      "grad_norm": 1.2418055534362793,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2806,
      "step": 33
    },
    {
      "epoch": 0.2066869300911854,
      "grad_norm": 1.1326993703842163,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.1825,
      "step": 34
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 1.0901497602462769,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.2682,
      "step": 35
    },
    {
      "epoch": 0.2188449848024316,
      "grad_norm": 1.10700523853302,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1842,
      "step": 36
    },
    {
      "epoch": 0.22492401215805471,
      "grad_norm": 1.1828606128692627,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1569,
      "step": 37
    },
    {
      "epoch": 0.23100303951367782,
      "grad_norm": 1.1258238554000854,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3126,
      "step": 38
    },
    {
      "epoch": 0.23708206686930092,
      "grad_norm": 1.2246427536010742,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.3259,
      "step": 39
    },
    {
      "epoch": 0.24316109422492402,
      "grad_norm": 1.1289308071136475,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.2666,
      "step": 40
    },
    {
      "epoch": 0.24924012158054712,
      "grad_norm": 1.2298989295959473,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3318,
      "step": 41
    },
    {
      "epoch": 0.2553191489361702,
      "grad_norm": 0.9683639407157898,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.587,
      "step": 42
    },
    {
      "epoch": 0.2613981762917933,
      "grad_norm": 1.4129326343536377,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.4743,
      "step": 43
    },
    {
      "epoch": 0.2674772036474164,
      "grad_norm": 0.9456275105476379,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5002,
      "step": 44
    },
    {
      "epoch": 0.2735562310030395,
      "grad_norm": 1.1622868776321411,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.5013,
      "step": 45
    },
    {
      "epoch": 0.2796352583586626,
      "grad_norm": 1.09859037399292,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.6183,
      "step": 46
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 1.108465552330017,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4265,
      "step": 47
    },
    {
      "epoch": 0.2917933130699088,
      "grad_norm": 1.0654693841934204,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3385,
      "step": 48
    },
    {
      "epoch": 0.2978723404255319,
      "grad_norm": 1.1229150295257568,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3933,
      "step": 49
    },
    {
      "epoch": 0.303951367781155,
      "grad_norm": 1.1337101459503174,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4113,
      "step": 50
    },
    {
      "epoch": 0.303951367781155,
      "eval_loss": 1.367661952972412,
      "eval_runtime": 31.4477,
      "eval_samples_per_second": 8.808,
      "eval_steps_per_second": 2.226,
      "step": 50
    },
    {
      "epoch": 0.3100303951367781,
      "grad_norm": 1.1549537181854248,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4435,
      "step": 51
    },
    {
      "epoch": 0.3161094224924012,
      "grad_norm": 1.0424610376358032,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2615,
      "step": 52
    },
    {
      "epoch": 0.3221884498480243,
      "grad_norm": 1.1868759393692017,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4175,
      "step": 53
    },
    {
      "epoch": 0.3282674772036474,
      "grad_norm": 1.2601715326309204,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.4801,
      "step": 54
    },
    {
      "epoch": 0.3343465045592705,
      "grad_norm": 1.286606788635254,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4707,
      "step": 55
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 1.3452922105789185,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5311,
      "step": 56
    },
    {
      "epoch": 0.3465045592705167,
      "grad_norm": 1.329443335533142,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4985,
      "step": 57
    },
    {
      "epoch": 0.3525835866261398,
      "grad_norm": 1.4766901731491089,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.7028,
      "step": 58
    },
    {
      "epoch": 0.3586626139817629,
      "grad_norm": 1.523435354232788,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5525,
      "step": 59
    },
    {
      "epoch": 0.364741641337386,
      "grad_norm": 1.4292651414871216,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.6289,
      "step": 60
    },
    {
      "epoch": 0.3708206686930091,
      "grad_norm": 1.48844575881958,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4409,
      "step": 61
    },
    {
      "epoch": 0.3768996960486322,
      "grad_norm": 1.2734694480895996,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1666,
      "step": 62
    },
    {
      "epoch": 0.3829787234042553,
      "grad_norm": 1.9541734457015991,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.168,
      "step": 63
    },
    {
      "epoch": 0.3890577507598784,
      "grad_norm": 1.0450496673583984,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1791,
      "step": 64
    },
    {
      "epoch": 0.3951367781155015,
      "grad_norm": 1.1312540769577026,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.2601,
      "step": 65
    },
    {
      "epoch": 0.4012158054711246,
      "grad_norm": 1.2040770053863525,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.4096,
      "step": 66
    },
    {
      "epoch": 0.4072948328267477,
      "grad_norm": 1.1914122104644775,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1566,
      "step": 67
    },
    {
      "epoch": 0.4133738601823708,
      "grad_norm": 1.09154212474823,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1598,
      "step": 68
    },
    {
      "epoch": 0.4194528875379939,
      "grad_norm": 1.154236078262329,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3208,
      "step": 69
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 1.0120573043823242,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2509,
      "step": 70
    },
    {
      "epoch": 0.4316109422492401,
      "grad_norm": 1.1286239624023438,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.222,
      "step": 71
    },
    {
      "epoch": 0.4376899696048632,
      "grad_norm": 1.0104598999023438,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1168,
      "step": 72
    },
    {
      "epoch": 0.44376899696048633,
      "grad_norm": 1.1920275688171387,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2207,
      "step": 73
    },
    {
      "epoch": 0.44984802431610943,
      "grad_norm": 1.0564591884613037,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1974,
      "step": 74
    },
    {
      "epoch": 0.45592705167173253,
      "grad_norm": 1.029231071472168,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1208,
      "step": 75
    },
    {
      "epoch": 0.46200607902735563,
      "grad_norm": 1.0670363903045654,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.2225,
      "step": 76
    },
    {
      "epoch": 0.46808510638297873,
      "grad_norm": 1.1883869171142578,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2314,
      "step": 77
    },
    {
      "epoch": 0.47416413373860183,
      "grad_norm": 1.1696826219558716,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.2794,
      "step": 78
    },
    {
      "epoch": 0.48024316109422494,
      "grad_norm": 1.2501018047332764,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.166,
      "step": 79
    },
    {
      "epoch": 0.48632218844984804,
      "grad_norm": 1.0672279596328735,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.207,
      "step": 80
    },
    {
      "epoch": 0.49240121580547114,
      "grad_norm": 1.131777048110962,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.2141,
      "step": 81
    },
    {
      "epoch": 0.49848024316109424,
      "grad_norm": 1.188834309577942,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3121,
      "step": 82
    },
    {
      "epoch": 0.5045592705167173,
      "grad_norm": 0.940176248550415,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.3942,
      "step": 83
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 0.9767402410507202,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5384,
      "step": 84
    },
    {
      "epoch": 0.5167173252279635,
      "grad_norm": 0.9623183012008667,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.412,
      "step": 85
    },
    {
      "epoch": 0.5227963525835866,
      "grad_norm": 1.0471317768096924,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3955,
      "step": 86
    },
    {
      "epoch": 0.5288753799392097,
      "grad_norm": 1.049780249595642,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4862,
      "step": 87
    },
    {
      "epoch": 0.5349544072948328,
      "grad_norm": 1.3101881742477417,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3828,
      "step": 88
    },
    {
      "epoch": 0.541033434650456,
      "grad_norm": 1.1504623889923096,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2833,
      "step": 89
    },
    {
      "epoch": 0.547112462006079,
      "grad_norm": 1.1060081720352173,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2397,
      "step": 90
    },
    {
      "epoch": 0.5531914893617021,
      "grad_norm": 1.1813318729400635,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.277,
      "step": 91
    },
    {
      "epoch": 0.5592705167173252,
      "grad_norm": 1.1772738695144653,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2138,
      "step": 92
    },
    {
      "epoch": 0.5653495440729484,
      "grad_norm": 1.1979900598526,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3073,
      "step": 93
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.2181065082550049,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.2804,
      "step": 94
    },
    {
      "epoch": 0.5775075987841946,
      "grad_norm": 1.2531542778015137,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.3294,
      "step": 95
    },
    {
      "epoch": 0.5835866261398176,
      "grad_norm": 1.4513905048370361,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3925,
      "step": 96
    },
    {
      "epoch": 0.5896656534954408,
      "grad_norm": 1.177195429801941,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.2597,
      "step": 97
    },
    {
      "epoch": 0.5957446808510638,
      "grad_norm": 1.451488971710205,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.5541,
      "step": 98
    },
    {
      "epoch": 0.601823708206687,
      "grad_norm": 1.3667665719985962,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2283,
      "step": 99
    },
    {
      "epoch": 0.60790273556231,
      "grad_norm": 1.5578776597976685,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.4027,
      "step": 100
    },
    {
      "epoch": 0.60790273556231,
      "eval_loss": 1.291818380355835,
      "eval_runtime": 31.4819,
      "eval_samples_per_second": 8.799,
      "eval_steps_per_second": 2.223,
      "step": 100
    },
    {
      "epoch": 0.6139817629179332,
      "grad_norm": 1.443122148513794,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4089,
      "step": 101
    },
    {
      "epoch": 0.6200607902735562,
      "grad_norm": 1.6079407930374146,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.4846,
      "step": 102
    },
    {
      "epoch": 0.6261398176291794,
      "grad_norm": 1.2146152257919312,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.2541,
      "step": 103
    },
    {
      "epoch": 0.6322188449848024,
      "grad_norm": 1.0780636072158813,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.155,
      "step": 104
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 1.0256733894348145,
      "learning_rate": 5e-05,
      "loss": 1.2241,
      "step": 105
    },
    {
      "epoch": 0.6443768996960486,
      "grad_norm": 1.025854468345642,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2551,
      "step": 106
    },
    {
      "epoch": 0.6504559270516718,
      "grad_norm": 0.9595382809638977,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.1257,
      "step": 107
    },
    {
      "epoch": 0.6565349544072948,
      "grad_norm": 1.0922917127609253,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.1945,
      "step": 108
    },
    {
      "epoch": 0.662613981762918,
      "grad_norm": 1.0498374700546265,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1911,
      "step": 109
    },
    {
      "epoch": 0.668693009118541,
      "grad_norm": 1.0221478939056396,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1414,
      "step": 110
    },
    {
      "epoch": 0.6747720364741642,
      "grad_norm": 1.0440903902053833,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2226,
      "step": 111
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 1.2824653387069702,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1579,
      "step": 112
    },
    {
      "epoch": 0.6869300911854104,
      "grad_norm": 1.0503424406051636,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2229,
      "step": 113
    },
    {
      "epoch": 0.6930091185410334,
      "grad_norm": 0.9603865146636963,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.1739,
      "step": 114
    },
    {
      "epoch": 0.6990881458966566,
      "grad_norm": 1.053389549255371,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.0727,
      "step": 115
    },
    {
      "epoch": 0.7051671732522796,
      "grad_norm": 0.9794132709503174,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1693,
      "step": 116
    },
    {
      "epoch": 0.7112462006079028,
      "grad_norm": 1.194525957107544,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.1384,
      "step": 117
    },
    {
      "epoch": 0.7173252279635258,
      "grad_norm": 1.0075552463531494,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.1697,
      "step": 118
    },
    {
      "epoch": 0.723404255319149,
      "grad_norm": 1.3113508224487305,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0993,
      "step": 119
    },
    {
      "epoch": 0.729483282674772,
      "grad_norm": 1.0296517610549927,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.2017,
      "step": 120
    },
    {
      "epoch": 0.7355623100303952,
      "grad_norm": 1.151802897453308,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2173,
      "step": 121
    },
    {
      "epoch": 0.7416413373860182,
      "grad_norm": 1.101464033126831,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2155,
      "step": 122
    },
    {
      "epoch": 0.7477203647416414,
      "grad_norm": 1.2130014896392822,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2716,
      "step": 123
    },
    {
      "epoch": 0.7537993920972644,
      "grad_norm": 0.8958890438079834,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.3998,
      "step": 124
    },
    {
      "epoch": 0.7598784194528876,
      "grad_norm": 1.0223948955535889,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3954,
      "step": 125
    },
    {
      "epoch": 0.7659574468085106,
      "grad_norm": 1.014959692955017,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4315,
      "step": 126
    },
    {
      "epoch": 0.7720364741641338,
      "grad_norm": 1.0947637557983398,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4246,
      "step": 127
    },
    {
      "epoch": 0.7781155015197568,
      "grad_norm": 0.9514850974082947,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2555,
      "step": 128
    },
    {
      "epoch": 0.78419452887538,
      "grad_norm": 1.2168844938278198,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4294,
      "step": 129
    },
    {
      "epoch": 0.790273556231003,
      "grad_norm": 1.093896746635437,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2719,
      "step": 130
    },
    {
      "epoch": 0.7963525835866262,
      "grad_norm": 2.7117221355438232,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.5384,
      "step": 131
    },
    {
      "epoch": 0.8024316109422492,
      "grad_norm": 1.3037480115890503,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3691,
      "step": 132
    },
    {
      "epoch": 0.8085106382978723,
      "grad_norm": 1.3540011644363403,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4782,
      "step": 133
    },
    {
      "epoch": 0.8145896656534954,
      "grad_norm": 1.2777421474456787,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.3177,
      "step": 134
    },
    {
      "epoch": 0.8206686930091185,
      "grad_norm": 1.3406814336776733,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.1462,
      "step": 135
    },
    {
      "epoch": 0.8267477203647416,
      "grad_norm": 1.2267261743545532,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.465,
      "step": 136
    },
    {
      "epoch": 0.8328267477203647,
      "grad_norm": 1.4003666639328003,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2814,
      "step": 137
    },
    {
      "epoch": 0.8389057750759878,
      "grad_norm": 1.2393686771392822,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2385,
      "step": 138
    },
    {
      "epoch": 0.8449848024316109,
      "grad_norm": 1.312151312828064,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.2368,
      "step": 139
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 1.6161844730377197,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.2925,
      "step": 140
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 1.444501280784607,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3931,
      "step": 141
    },
    {
      "epoch": 0.8632218844984803,
      "grad_norm": 1.3983210325241089,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3165,
      "step": 142
    },
    {
      "epoch": 0.8693009118541033,
      "grad_norm": 1.3116635084152222,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3285,
      "step": 143
    },
    {
      "epoch": 0.8753799392097265,
      "grad_norm": 1.119754433631897,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2094,
      "step": 144
    },
    {
      "epoch": 0.8814589665653495,
      "grad_norm": 1.1628637313842773,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0616,
      "step": 145
    },
    {
      "epoch": 0.8875379939209727,
      "grad_norm": 1.0161055326461792,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1605,
      "step": 146
    },
    {
      "epoch": 0.8936170212765957,
      "grad_norm": 1.0418037176132202,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1572,
      "step": 147
    },
    {
      "epoch": 0.8996960486322189,
      "grad_norm": 0.9632786512374878,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.0861,
      "step": 148
    },
    {
      "epoch": 0.9057750759878419,
      "grad_norm": 1.1108787059783936,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.2152,
      "step": 149
    },
    {
      "epoch": 0.9118541033434651,
      "grad_norm": 0.9935660362243652,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.2279,
      "step": 150
    },
    {
      "epoch": 0.9118541033434651,
      "eval_loss": 1.2594267129898071,
      "eval_runtime": 31.213,
      "eval_samples_per_second": 8.875,
      "eval_steps_per_second": 2.243,
      "step": 150
    },
    {
      "epoch": 0.9179331306990881,
      "grad_norm": 1.0456987619400024,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.2057,
      "step": 151
    },
    {
      "epoch": 0.9240121580547113,
      "grad_norm": 0.9676578640937805,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0679,
      "step": 152
    },
    {
      "epoch": 0.9300911854103343,
      "grad_norm": 0.9445816278457642,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.0741,
      "step": 153
    },
    {
      "epoch": 0.9361702127659575,
      "grad_norm": 0.9686632752418518,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.1458,
      "step": 154
    },
    {
      "epoch": 0.9422492401215805,
      "grad_norm": 1.1513304710388184,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.2101,
      "step": 155
    },
    {
      "epoch": 0.9483282674772037,
      "grad_norm": 1.3059872388839722,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.084,
      "step": 156
    },
    {
      "epoch": 0.9544072948328267,
      "grad_norm": 1.0212550163269043,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.2212,
      "step": 157
    },
    {
      "epoch": 0.9604863221884499,
      "grad_norm": 1.06368088722229,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1434,
      "step": 158
    },
    {
      "epoch": 0.9665653495440729,
      "grad_norm": 0.990797758102417,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.079,
      "step": 159
    },
    {
      "epoch": 0.9726443768996961,
      "grad_norm": 1.0803565979003906,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.149,
      "step": 160
    },
    {
      "epoch": 0.9787234042553191,
      "grad_norm": 1.047892689704895,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.1347,
      "step": 161
    },
    {
      "epoch": 0.9848024316109423,
      "grad_norm": 1.0211436748504639,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2288,
      "step": 162
    },
    {
      "epoch": 0.9908814589665653,
      "grad_norm": 1.1094372272491455,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.035,
      "step": 163
    },
    {
      "epoch": 0.9969604863221885,
      "grad_norm": 1.3595134019851685,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.2756,
      "step": 164
    },
    {
      "epoch": 1.0030395136778116,
      "grad_norm": 1.6225723028182983,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.8043,
      "step": 165
    },
    {
      "epoch": 1.0091185410334347,
      "grad_norm": 0.8201916217803955,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.1228,
      "step": 166
    },
    {
      "epoch": 1.0151975683890577,
      "grad_norm": 0.8471580743789673,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1311,
      "step": 167
    },
    {
      "epoch": 1.0212765957446808,
      "grad_norm": 0.9013404250144958,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.1765,
      "step": 168
    },
    {
      "epoch": 1.027355623100304,
      "grad_norm": 0.96379554271698,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.2927,
      "step": 169
    },
    {
      "epoch": 1.033434650455927,
      "grad_norm": 0.9141067862510681,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0763,
      "step": 170
    },
    {
      "epoch": 1.0395136778115501,
      "grad_norm": 0.9531221985816956,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.1042,
      "step": 171
    },
    {
      "epoch": 1.0455927051671732,
      "grad_norm": 0.9676122665405273,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.0336,
      "step": 172
    },
    {
      "epoch": 1.0516717325227964,
      "grad_norm": 0.9848034381866455,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0514,
      "step": 173
    },
    {
      "epoch": 1.0577507598784195,
      "grad_norm": 0.9543176293373108,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0644,
      "step": 174
    },
    {
      "epoch": 1.0638297872340425,
      "grad_norm": 0.9698459506034851,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9047,
      "step": 175
    },
    {
      "epoch": 1.0699088145896656,
      "grad_norm": 0.9518846273422241,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9306,
      "step": 176
    },
    {
      "epoch": 1.0759878419452888,
      "grad_norm": 1.0438520908355713,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9377,
      "step": 177
    },
    {
      "epoch": 1.082066869300912,
      "grad_norm": 1.1060409545898438,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.0116,
      "step": 178
    },
    {
      "epoch": 1.088145896656535,
      "grad_norm": 1.0859452486038208,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.9415,
      "step": 179
    },
    {
      "epoch": 1.094224924012158,
      "grad_norm": 1.0959041118621826,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.0567,
      "step": 180
    },
    {
      "epoch": 1.1003039513677813,
      "grad_norm": 1.2742414474487305,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0596,
      "step": 181
    },
    {
      "epoch": 1.1063829787234043,
      "grad_norm": 1.243722677230835,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1104,
      "step": 182
    },
    {
      "epoch": 1.1124620060790273,
      "grad_norm": 1.2263867855072021,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.9401,
      "step": 183
    },
    {
      "epoch": 1.1185410334346504,
      "grad_norm": 1.1445238590240479,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0591,
      "step": 184
    },
    {
      "epoch": 1.1246200607902737,
      "grad_norm": 1.1770378351211548,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1894,
      "step": 185
    },
    {
      "epoch": 1.1306990881458967,
      "grad_norm": 1.0755008459091187,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.0064,
      "step": 186
    },
    {
      "epoch": 1.1367781155015197,
      "grad_norm": 0.9145883917808533,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.8981,
      "step": 187
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.8915071487426758,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0023,
      "step": 188
    },
    {
      "epoch": 1.148936170212766,
      "grad_norm": 0.8272641897201538,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8643,
      "step": 189
    },
    {
      "epoch": 1.155015197568389,
      "grad_norm": 0.9100205302238464,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.8841,
      "step": 190
    },
    {
      "epoch": 1.1610942249240122,
      "grad_norm": 1.0280265808105469,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.9846,
      "step": 191
    },
    {
      "epoch": 1.1671732522796352,
      "grad_norm": 0.8951881527900696,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.8748,
      "step": 192
    },
    {
      "epoch": 1.1732522796352582,
      "grad_norm": 0.8505797386169434,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9897,
      "step": 193
    },
    {
      "epoch": 1.1793313069908815,
      "grad_norm": 0.808299720287323,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.842,
      "step": 194
    },
    {
      "epoch": 1.1854103343465046,
      "grad_norm": 0.8740032315254211,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8655,
      "step": 195
    },
    {
      "epoch": 1.1914893617021276,
      "grad_norm": 0.9953221082687378,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.0702,
      "step": 196
    },
    {
      "epoch": 1.1975683890577509,
      "grad_norm": 1.0144661664962769,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9131,
      "step": 197
    },
    {
      "epoch": 1.203647416413374,
      "grad_norm": 0.8584089875221252,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.892,
      "step": 198
    },
    {
      "epoch": 1.209726443768997,
      "grad_norm": 0.874872088432312,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9343,
      "step": 199
    },
    {
      "epoch": 1.21580547112462,
      "grad_norm": 0.8324904441833496,
      "learning_rate": 0.0,
      "loss": 0.914,
      "step": 200
    },
    {
      "epoch": 1.21580547112462,
      "eval_loss": 1.253917932510376,
      "eval_runtime": 31.4871,
      "eval_samples_per_second": 8.797,
      "eval_steps_per_second": 2.223,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 4.603794878135009e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}