{
  "best_metric": 0.38313207030296326,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.061354398343431246,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003067719917171562,
      "grad_norm": 3.867211103439331,
      "learning_rate": 1e-05,
      "loss": 1.4165,
      "step": 1
    },
    {
      "epoch": 0.0003067719917171562,
      "eval_loss": 0.5137439370155334,
      "eval_runtime": 431.309,
      "eval_samples_per_second": 12.729,
      "eval_steps_per_second": 3.183,
      "step": 1
    },
    {
      "epoch": 0.0006135439834343124,
      "grad_norm": 4.187973976135254,
      "learning_rate": 2e-05,
      "loss": 1.491,
      "step": 2
    },
    {
      "epoch": 0.0009203159751514687,
      "grad_norm": 3.8121917247772217,
      "learning_rate": 3e-05,
      "loss": 1.5326,
      "step": 3
    },
    {
      "epoch": 0.0012270879668686248,
      "grad_norm": 3.2435965538024902,
      "learning_rate": 4e-05,
      "loss": 1.6677,
      "step": 4
    },
    {
      "epoch": 0.0015338599585857811,
      "grad_norm": 3.5780789852142334,
      "learning_rate": 5e-05,
      "loss": 1.7373,
      "step": 5
    },
    {
      "epoch": 0.0018406319503029374,
      "grad_norm": 3.797482967376709,
      "learning_rate": 6e-05,
      "loss": 1.7688,
      "step": 6
    },
    {
      "epoch": 0.0021474039420200934,
      "grad_norm": 3.7806077003479004,
      "learning_rate": 7e-05,
      "loss": 1.8273,
      "step": 7
    },
    {
      "epoch": 0.0024541759337372497,
      "grad_norm": 3.55958890914917,
      "learning_rate": 8e-05,
      "loss": 1.7633,
      "step": 8
    },
    {
      "epoch": 0.002760947925454406,
      "grad_norm": 3.5978844165802,
      "learning_rate": 9e-05,
      "loss": 1.8491,
      "step": 9
    },
    {
      "epoch": 0.0030677199171715622,
      "grad_norm": 3.2147061824798584,
      "learning_rate": 0.0001,
      "loss": 1.8737,
      "step": 10
    },
    {
      "epoch": 0.0033744919088887185,
      "grad_norm": 3.405806303024292,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.0367,
      "step": 11
    },
    {
      "epoch": 0.0036812639006058748,
      "grad_norm": 3.477496385574341,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8873,
      "step": 12
    },
    {
      "epoch": 0.003988035892323031,
      "grad_norm": 3.2102620601654053,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8225,
      "step": 13
    },
    {
      "epoch": 0.004294807884040187,
      "grad_norm": 3.2617945671081543,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.9135,
      "step": 14
    },
    {
      "epoch": 0.004601579875757343,
      "grad_norm": 3.2378807067871094,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.7631,
      "step": 15
    },
    {
      "epoch": 0.004908351867474499,
      "grad_norm": 3.1695950031280518,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.884,
      "step": 16
    },
    {
      "epoch": 0.005215123859191656,
      "grad_norm": 3.0632283687591553,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.777,
      "step": 17
    },
    {
      "epoch": 0.005521895850908812,
      "grad_norm": 3.3818905353546143,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.8708,
      "step": 18
    },
    {
      "epoch": 0.005828667842625968,
      "grad_norm": 3.155194044113159,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.0549,
      "step": 19
    },
    {
      "epoch": 0.0061354398343431244,
      "grad_norm": 3.059497356414795,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.8732,
      "step": 20
    },
    {
      "epoch": 0.006442211826060281,
      "grad_norm": 2.9313485622406006,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.8737,
      "step": 21
    },
    {
      "epoch": 0.006748983817777437,
      "grad_norm": 2.914853811264038,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.8744,
      "step": 22
    },
    {
      "epoch": 0.007055755809494593,
      "grad_norm": 3.045229196548462,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.7995,
      "step": 23
    },
    {
      "epoch": 0.0073625278012117495,
      "grad_norm": 3.194204568862915,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6646,
      "step": 24
    },
    {
      "epoch": 0.007669299792928906,
      "grad_norm": 2.955012083053589,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.7695,
      "step": 25
    },
    {
      "epoch": 0.007976071784646061,
      "grad_norm": 3.1171529293060303,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.7475,
      "step": 26
    },
    {
      "epoch": 0.008282843776363218,
      "grad_norm": 3.0841450691223145,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.7644,
      "step": 27
    },
    {
      "epoch": 0.008589615768080374,
      "grad_norm": 3.0407214164733887,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.8768,
      "step": 28
    },
    {
      "epoch": 0.00889638775979753,
      "grad_norm": 2.9777040481567383,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.7495,
      "step": 29
    },
    {
      "epoch": 0.009203159751514686,
      "grad_norm": 3.0118660926818848,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.7803,
      "step": 30
    },
    {
      "epoch": 0.009509931743231843,
      "grad_norm": 3.0673904418945312,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.72,
      "step": 31
    },
    {
      "epoch": 0.009816703734948999,
      "grad_norm": 2.9070444107055664,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.6515,
      "step": 32
    },
    {
      "epoch": 0.010123475726666156,
      "grad_norm": 3.0477380752563477,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.7649,
      "step": 33
    },
    {
      "epoch": 0.010430247718383311,
      "grad_norm": 3.129366159439087,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5294,
      "step": 34
    },
    {
      "epoch": 0.010737019710100468,
      "grad_norm": 3.196800708770752,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.6907,
      "step": 35
    },
    {
      "epoch": 0.011043791701817624,
      "grad_norm": 3.3101892471313477,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5625,
      "step": 36
    },
    {
      "epoch": 0.011350563693534781,
      "grad_norm": 3.2586722373962402,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.7031,
      "step": 37
    },
    {
      "epoch": 0.011657335685251936,
      "grad_norm": 3.1446077823638916,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.4559,
      "step": 38
    },
    {
      "epoch": 0.011964107676969093,
      "grad_norm": 3.4737548828125,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.5921,
      "step": 39
    },
    {
      "epoch": 0.012270879668686249,
      "grad_norm": 3.509357452392578,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.3701,
      "step": 40
    },
    {
      "epoch": 0.012577651660403406,
      "grad_norm": 3.5204527378082275,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4898,
      "step": 41
    },
    {
      "epoch": 0.012884423652120561,
      "grad_norm": 3.7146849632263184,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.4411,
      "step": 42
    },
    {
      "epoch": 0.013191195643837717,
      "grad_norm": 4.295719623565674,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5573,
      "step": 43
    },
    {
      "epoch": 0.013497967635554874,
      "grad_norm": 4.345009803771973,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.7055,
      "step": 44
    },
    {
      "epoch": 0.01380473962727203,
      "grad_norm": 4.270930290222168,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3713,
      "step": 45
    },
    {
      "epoch": 0.014111511618989186,
      "grad_norm": 4.617498874664307,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5265,
      "step": 46
    },
    {
      "epoch": 0.014418283610706342,
      "grad_norm": 4.915421485900879,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.6459,
      "step": 47
    },
    {
      "epoch": 0.014725055602423499,
      "grad_norm": 5.67216157913208,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6361,
      "step": 48
    },
    {
      "epoch": 0.015031827594140654,
      "grad_norm": 5.974374771118164,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.5773,
      "step": 49
    },
    {
      "epoch": 0.015338599585857812,
      "grad_norm": 12.7064790725708,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.978,
      "step": 50
    },
    {
      "epoch": 0.015338599585857812,
      "eval_loss": 0.46886593103408813,
      "eval_runtime": 434.9161,
      "eval_samples_per_second": 12.623,
      "eval_steps_per_second": 3.157,
      "step": 50
    },
    {
      "epoch": 0.01564537157757497,
      "grad_norm": 3.6866564750671387,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2687,
      "step": 51
    },
    {
      "epoch": 0.015952143569292122,
      "grad_norm": 2.6937808990478516,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.233,
      "step": 52
    },
    {
      "epoch": 0.01625891556100928,
      "grad_norm": 2.3204245567321777,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4214,
      "step": 53
    },
    {
      "epoch": 0.016565687552726437,
      "grad_norm": 2.3629627227783203,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.5954,
      "step": 54
    },
    {
      "epoch": 0.016872459544443594,
      "grad_norm": 2.4090921878814697,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7069,
      "step": 55
    },
    {
      "epoch": 0.017179231536160747,
      "grad_norm": 2.4678914546966553,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7505,
      "step": 56
    },
    {
      "epoch": 0.017486003527877905,
      "grad_norm": 2.441708564758301,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.7319,
      "step": 57
    },
    {
      "epoch": 0.01779277551959506,
      "grad_norm": 2.4955708980560303,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.8441,
      "step": 58
    },
    {
      "epoch": 0.01809954751131222,
      "grad_norm": 2.365328550338745,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.789,
      "step": 59
    },
    {
      "epoch": 0.018406319503029372,
      "grad_norm": 2.52234148979187,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.849,
      "step": 60
    },
    {
      "epoch": 0.01871309149474653,
      "grad_norm": 2.5644991397857666,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.8863,
      "step": 61
    },
    {
      "epoch": 0.019019863486463687,
      "grad_norm": 2.4594507217407227,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.7247,
      "step": 62
    },
    {
      "epoch": 0.01932663547818084,
      "grad_norm": 2.370521068572998,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.8515,
      "step": 63
    },
    {
      "epoch": 0.019633407469897998,
      "grad_norm": 2.463268518447876,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.7623,
      "step": 64
    },
    {
      "epoch": 0.019940179461615155,
      "grad_norm": 2.445389986038208,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.7943,
      "step": 65
    },
    {
      "epoch": 0.020246951453332312,
      "grad_norm": 2.4769937992095947,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.8089,
      "step": 66
    },
    {
      "epoch": 0.020553723445049465,
      "grad_norm": 2.480973720550537,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.8798,
      "step": 67
    },
    {
      "epoch": 0.020860495436766623,
      "grad_norm": 2.4982151985168457,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.8022,
      "step": 68
    },
    {
      "epoch": 0.02116726742848378,
      "grad_norm": 2.5915822982788086,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.7638,
      "step": 69
    },
    {
      "epoch": 0.021474039420200937,
      "grad_norm": 2.6000962257385254,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.8611,
      "step": 70
    },
    {
      "epoch": 0.02178081141191809,
      "grad_norm": 2.654924154281616,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.8758,
      "step": 71
    },
    {
      "epoch": 0.022087583403635248,
      "grad_norm": 3.0606656074523926,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.903,
      "step": 72
    },
    {
      "epoch": 0.022394355395352405,
      "grad_norm": 2.6482133865356445,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.8505,
      "step": 73
    },
    {
      "epoch": 0.022701127387069562,
      "grad_norm": 2.6382734775543213,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.7878,
      "step": 74
    },
    {
      "epoch": 0.023007899378786716,
      "grad_norm": 2.641846179962158,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.8739,
      "step": 75
    },
    {
      "epoch": 0.023314671370503873,
      "grad_norm": 2.55903959274292,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.7638,
      "step": 76
    },
    {
      "epoch": 0.02362144336222103,
      "grad_norm": 2.5169951915740967,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7313,
      "step": 77
    },
    {
      "epoch": 0.023928215353938187,
      "grad_norm": 2.58994722366333,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7086,
      "step": 78
    },
    {
      "epoch": 0.02423498734565534,
      "grad_norm": 2.512345790863037,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.5469,
      "step": 79
    },
    {
      "epoch": 0.024541759337372498,
      "grad_norm": 2.697726249694824,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.7046,
      "step": 80
    },
    {
      "epoch": 0.024848531329089655,
      "grad_norm": 2.663831949234009,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.7107,
      "step": 81
    },
    {
      "epoch": 0.025155303320806812,
      "grad_norm": 2.6148416996002197,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.5426,
      "step": 82
    },
    {
      "epoch": 0.025462075312523966,
      "grad_norm": 2.8210105895996094,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.7408,
      "step": 83
    },
    {
      "epoch": 0.025768847304241123,
      "grad_norm": 2.7916059494018555,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.6582,
      "step": 84
    },
    {
      "epoch": 0.02607561929595828,
      "grad_norm": 2.778249979019165,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5864,
      "step": 85
    },
    {
      "epoch": 0.026382391287675434,
      "grad_norm": 2.8042855262756348,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4477,
      "step": 86
    },
    {
      "epoch": 0.02668916327939259,
      "grad_norm": 2.8042662143707275,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4565,
      "step": 87
    },
    {
      "epoch": 0.026995935271109748,
      "grad_norm": 2.8745734691619873,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4609,
      "step": 88
    },
    {
      "epoch": 0.027302707262826905,
      "grad_norm": 3.0955867767333984,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5444,
      "step": 89
    },
    {
      "epoch": 0.02760947925454406,
      "grad_norm": 3.0098795890808105,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2125,
      "step": 90
    },
    {
      "epoch": 0.027916251246261216,
      "grad_norm": 3.138331651687622,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.4283,
      "step": 91
    },
    {
      "epoch": 0.028223023237978373,
      "grad_norm": 3.676004648208618,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4786,
      "step": 92
    },
    {
      "epoch": 0.02852979522969553,
      "grad_norm": 3.189359426498413,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2687,
      "step": 93
    },
    {
      "epoch": 0.028836567221412684,
      "grad_norm": 3.4852240085601807,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4256,
      "step": 94
    },
    {
      "epoch": 0.02914333921312984,
      "grad_norm": 4.142465114593506,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.5008,
      "step": 95
    },
    {
      "epoch": 0.029450111204846998,
      "grad_norm": 4.0958781242370605,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.4943,
      "step": 96
    },
    {
      "epoch": 0.029756883196564155,
      "grad_norm": 3.7917978763580322,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.1717,
      "step": 97
    },
    {
      "epoch": 0.03006365518828131,
      "grad_norm": 4.897753715515137,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4985,
      "step": 98
    },
    {
      "epoch": 0.030370427179998466,
      "grad_norm": 5.293882846832275,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4607,
      "step": 99
    },
    {
      "epoch": 0.030677199171715623,
      "grad_norm": 7.581045150756836,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.771,
      "step": 100
    },
    {
      "epoch": 0.030677199171715623,
      "eval_loss": 0.4151594042778015,
      "eval_runtime": 434.6836,
      "eval_samples_per_second": 12.63,
      "eval_steps_per_second": 3.159,
      "step": 100
    },
    {
      "epoch": 0.03098397116343278,
      "grad_norm": 2.395158290863037,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.1422,
      "step": 101
    },
    {
      "epoch": 0.03129074315514994,
      "grad_norm": 2.009828567504883,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.1197,
      "step": 102
    },
    {
      "epoch": 0.03159751514686709,
      "grad_norm": 1.9227544069290161,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3121,
      "step": 103
    },
    {
      "epoch": 0.031904287138584245,
      "grad_norm": 2.0807535648345947,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5416,
      "step": 104
    },
    {
      "epoch": 0.0322110591303014,
      "grad_norm": 2.0546822547912598,
      "learning_rate": 5e-05,
      "loss": 1.6535,
      "step": 105
    },
    {
      "epoch": 0.03251783112201856,
      "grad_norm": 2.0844359397888184,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.6739,
      "step": 106
    },
    {
      "epoch": 0.032824603113735716,
      "grad_norm": 2.180797576904297,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.6795,
      "step": 107
    },
    {
      "epoch": 0.03313137510545287,
      "grad_norm": 2.171038866043091,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.751,
      "step": 108
    },
    {
      "epoch": 0.03343814709717003,
      "grad_norm": 2.2308526039123535,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.7422,
      "step": 109
    },
    {
      "epoch": 0.03374491908888719,
      "grad_norm": 2.2218360900878906,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.653,
      "step": 110
    },
    {
      "epoch": 0.03405169108060434,
      "grad_norm": 2.2783052921295166,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.7694,
      "step": 111
    },
    {
      "epoch": 0.034358463072321495,
      "grad_norm": 2.218576431274414,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.7232,
      "step": 112
    },
    {
      "epoch": 0.03466523506403865,
      "grad_norm": 2.349517822265625,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.7462,
      "step": 113
    },
    {
      "epoch": 0.03497200705575581,
      "grad_norm": 2.320361375808716,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.7257,
      "step": 114
    },
    {
      "epoch": 0.035278779047472966,
      "grad_norm": 2.2596595287323,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.7402,
      "step": 115
    },
    {
      "epoch": 0.03558555103919012,
      "grad_norm": 2.3461170196533203,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.8138,
      "step": 116
    },
    {
      "epoch": 0.03589232303090728,
      "grad_norm": 2.3731675148010254,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.8307,
      "step": 117
    },
    {
      "epoch": 0.03619909502262444,
      "grad_norm": 2.3603556156158447,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.8327,
      "step": 118
    },
    {
      "epoch": 0.03650586701434159,
      "grad_norm": 2.417180061340332,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9131,
      "step": 119
    },
    {
      "epoch": 0.036812639006058745,
      "grad_norm": 2.32993483543396,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.7673,
      "step": 120
    },
    {
      "epoch": 0.0371194109977759,
      "grad_norm": 2.3519070148468018,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.7153,
      "step": 121
    },
    {
      "epoch": 0.03742618298949306,
      "grad_norm": 2.558466672897339,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.8604,
      "step": 122
    },
    {
      "epoch": 0.037732954981210216,
      "grad_norm": 2.472303867340088,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.7699,
      "step": 123
    },
    {
      "epoch": 0.038039726972927373,
      "grad_norm": 2.4697232246398926,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8037,
      "step": 124
    },
    {
      "epoch": 0.03834649896464453,
      "grad_norm": 2.4891624450683594,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.7886,
      "step": 125
    },
    {
      "epoch": 0.03865327095636168,
      "grad_norm": 2.4804000854492188,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7454,
      "step": 126
    },
    {
      "epoch": 0.03896004294807884,
      "grad_norm": 2.548565149307251,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.7038,
      "step": 127
    },
    {
      "epoch": 0.039266814939795995,
      "grad_norm": 2.496056079864502,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.7823,
      "step": 128
    },
    {
      "epoch": 0.03957358693151315,
      "grad_norm": 2.51379656791687,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.5432,
      "step": 129
    },
    {
      "epoch": 0.03988035892323031,
      "grad_norm": 2.5729219913482666,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.5727,
      "step": 130
    },
    {
      "epoch": 0.040187130914947466,
      "grad_norm": 2.469094753265381,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.5351,
      "step": 131
    },
    {
      "epoch": 0.040493902906664624,
      "grad_norm": 2.5991036891937256,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.5525,
      "step": 132
    },
    {
      "epoch": 0.04080067489838178,
      "grad_norm": 2.5172080993652344,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.436,
      "step": 133
    },
    {
      "epoch": 0.04110744689009893,
      "grad_norm": 2.893340826034546,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.7322,
      "step": 134
    },
    {
      "epoch": 0.04141421888181609,
      "grad_norm": 2.680251121520996,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.3377,
      "step": 135
    },
    {
      "epoch": 0.041720990873533245,
      "grad_norm": 2.8206801414489746,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4483,
      "step": 136
    },
    {
      "epoch": 0.0420277628652504,
      "grad_norm": 2.802145004272461,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3801,
      "step": 137
    },
    {
      "epoch": 0.04233453485696756,
      "grad_norm": 2.9905385971069336,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.4374,
      "step": 138
    },
    {
      "epoch": 0.04264130684868472,
      "grad_norm": 3.0047194957733154,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.4918,
      "step": 139
    },
    {
      "epoch": 0.042948078840401874,
      "grad_norm": 2.988581657409668,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.3255,
      "step": 140
    },
    {
      "epoch": 0.04325485083211903,
      "grad_norm": 2.9595563411712646,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.211,
      "step": 141
    },
    {
      "epoch": 0.04356162282383618,
      "grad_norm": 3.2183542251586914,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2759,
      "step": 142
    },
    {
      "epoch": 0.04386839481555334,
      "grad_norm": 3.3673641681671143,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4227,
      "step": 143
    },
    {
      "epoch": 0.044175166807270495,
      "grad_norm": 3.56329607963562,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4078,
      "step": 144
    },
    {
      "epoch": 0.04448193879898765,
      "grad_norm": 3.6367719173431396,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1758,
      "step": 145
    },
    {
      "epoch": 0.04478871079070481,
      "grad_norm": 3.6097664833068848,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.3056,
      "step": 146
    },
    {
      "epoch": 0.04509548278242197,
      "grad_norm": 4.355531215667725,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.408,
      "step": 147
    },
    {
      "epoch": 0.045402254774139124,
      "grad_norm": 4.689583778381348,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5061,
      "step": 148
    },
    {
      "epoch": 0.045709026765856274,
      "grad_norm": 5.361886978149414,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.4443,
      "step": 149
    },
    {
      "epoch": 0.04601579875757343,
      "grad_norm": 9.398231506347656,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.5383,
      "step": 150
    },
    {
      "epoch": 0.04601579875757343,
      "eval_loss": 0.3894903063774109,
      "eval_runtime": 435.2651,
      "eval_samples_per_second": 12.613,
      "eval_steps_per_second": 3.154,
      "step": 150
    },
    {
      "epoch": 0.04632257074929059,
      "grad_norm": 1.657464623451233,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.9781,
      "step": 151
    },
    {
      "epoch": 0.046629342741007745,
      "grad_norm": 1.7459784746170044,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.1395,
      "step": 152
    },
    {
      "epoch": 0.0469361147327249,
      "grad_norm": 1.993617057800293,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3266,
      "step": 153
    },
    {
      "epoch": 0.04724288672444206,
      "grad_norm": 2.078782796859741,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4872,
      "step": 154
    },
    {
      "epoch": 0.04754965871615922,
      "grad_norm": 2.079587459564209,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.488,
      "step": 155
    },
    {
      "epoch": 0.047856430707876374,
      "grad_norm": 2.1271324157714844,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5552,
      "step": 156
    },
    {
      "epoch": 0.048163202699593524,
      "grad_norm": 2.21760630607605,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.6819,
      "step": 157
    },
    {
      "epoch": 0.04846997469131068,
      "grad_norm": 2.348280906677246,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.7581,
      "step": 158
    },
    {
      "epoch": 0.04877674668302784,
      "grad_norm": 2.139024257659912,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.6533,
      "step": 159
    },
    {
      "epoch": 0.049083518674744996,
      "grad_norm": 2.2002735137939453,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.723,
      "step": 160
    },
    {
      "epoch": 0.04939029066646215,
      "grad_norm": 2.226425886154175,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.7674,
      "step": 161
    },
    {
      "epoch": 0.04969706265817931,
      "grad_norm": 2.141286849975586,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6402,
      "step": 162
    },
    {
      "epoch": 0.05000383464989647,
      "grad_norm": 2.11076021194458,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.6625,
      "step": 163
    },
    {
      "epoch": 0.050310606641613624,
      "grad_norm": 2.16029691696167,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.7238,
      "step": 164
    },
    {
      "epoch": 0.050617378633330774,
      "grad_norm": 2.2302353382110596,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.8069,
      "step": 165
    },
    {
      "epoch": 0.05092415062504793,
      "grad_norm": 2.1031105518341064,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.5298,
      "step": 166
    },
    {
      "epoch": 0.05123092261676509,
      "grad_norm": 2.209357738494873,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6816,
      "step": 167
    },
    {
      "epoch": 0.051537694608482246,
      "grad_norm": 2.4503467082977295,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.9451,
      "step": 168
    },
    {
      "epoch": 0.0518444666001994,
      "grad_norm": 2.226034164428711,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.6469,
      "step": 169
    },
    {
      "epoch": 0.05215123859191656,
      "grad_norm": 2.2792415618896484,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7356,
      "step": 170
    },
    {
      "epoch": 0.05245801058363372,
      "grad_norm": 2.3297059535980225,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.66,
      "step": 171
    },
    {
      "epoch": 0.05276478257535087,
      "grad_norm": 2.3574416637420654,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.762,
      "step": 172
    },
    {
      "epoch": 0.053071554567068024,
      "grad_norm": 2.3705716133117676,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.7512,
      "step": 173
    },
    {
      "epoch": 0.05337832655878518,
      "grad_norm": 2.3903427124023438,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.756,
      "step": 174
    },
    {
      "epoch": 0.05368509855050234,
      "grad_norm": 2.40191650390625,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6934,
      "step": 175
    },
    {
      "epoch": 0.053991870542219496,
      "grad_norm": 2.3922252655029297,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.5961,
      "step": 176
    },
    {
      "epoch": 0.05429864253393665,
      "grad_norm": 2.55488657951355,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.802,
      "step": 177
    },
    {
      "epoch": 0.05460541452565381,
      "grad_norm": 2.489356279373169,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.6271,
      "step": 178
    },
    {
      "epoch": 0.05491218651737097,
      "grad_norm": 2.5332062244415283,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.7849,
      "step": 179
    },
    {
      "epoch": 0.05521895850908812,
      "grad_norm": 2.5501859188079834,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.7045,
      "step": 180
    },
    {
      "epoch": 0.055525730500805275,
      "grad_norm": 2.522325038909912,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.6855,
      "step": 181
    },
    {
      "epoch": 0.05583250249252243,
      "grad_norm": 2.615895986557007,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5042,
      "step": 182
    },
    {
      "epoch": 0.05613927448423959,
      "grad_norm": 2.595858335494995,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.5491,
      "step": 183
    },
    {
      "epoch": 0.056446046475956746,
      "grad_norm": 2.6438980102539062,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.5476,
      "step": 184
    },
    {
      "epoch": 0.0567528184676739,
      "grad_norm": 2.7027549743652344,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.5726,
      "step": 185
    },
    {
      "epoch": 0.05705959045939106,
      "grad_norm": 2.6841559410095215,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4902,
      "step": 186
    },
    {
      "epoch": 0.05736636245110821,
      "grad_norm": 2.7947614192962646,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3637,
      "step": 187
    },
    {
      "epoch": 0.05767313444282537,
      "grad_norm": 2.7080700397491455,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.339,
      "step": 188
    },
    {
      "epoch": 0.057979906434542525,
      "grad_norm": 2.653451442718506,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.0903,
      "step": 189
    },
    {
      "epoch": 0.05828667842625968,
      "grad_norm": 3.005455255508423,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.4023,
      "step": 190
    },
    {
      "epoch": 0.05859345041797684,
      "grad_norm": 3.1648271083831787,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.4136,
      "step": 191
    },
    {
      "epoch": 0.058900222409693996,
      "grad_norm": 3.284635305404663,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3972,
      "step": 192
    },
    {
      "epoch": 0.05920699440141115,
      "grad_norm": 3.19057035446167,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.3037,
      "step": 193
    },
    {
      "epoch": 0.05951376639312831,
      "grad_norm": 3.4141805171966553,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.403,
      "step": 194
    },
    {
      "epoch": 0.05982053838484546,
      "grad_norm": 3.3018360137939453,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.2452,
      "step": 195
    },
    {
      "epoch": 0.06012731037656262,
      "grad_norm": 3.7016236782073975,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3085,
      "step": 196
    },
    {
      "epoch": 0.060434082368279775,
      "grad_norm": 3.894061326980591,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2464,
      "step": 197
    },
    {
      "epoch": 0.06074085435999693,
      "grad_norm": 4.873144149780273,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.494,
      "step": 198
    },
    {
      "epoch": 0.06104762635171409,
      "grad_norm": 4.605802536010742,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1683,
      "step": 199
    },
    {
      "epoch": 0.061354398343431246,
      "grad_norm": 6.5521039962768555,
      "learning_rate": 0.0,
      "loss": 1.0239,
      "step": 200
    },
    {
      "epoch": 0.061354398343431246,
      "eval_loss": 0.38313207030296326,
      "eval_runtime": 433.9966,
      "eval_samples_per_second": 12.65,
      "eval_steps_per_second": 3.164,
      "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": 2.972901709719798e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}