{
  "best_metric": 0.32313525676727295,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.030456852791878,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01015228426395939,
      "grad_norm": 16.641149520874023,
      "learning_rate": 1e-05,
      "loss": 3.7755,
      "step": 1
    },
    {
      "epoch": 0.01015228426395939,
      "eval_loss": 3.967555046081543,
      "eval_runtime": 12.4653,
      "eval_samples_per_second": 13.317,
      "eval_steps_per_second": 3.369,
      "step": 1
    },
    {
      "epoch": 0.02030456852791878,
      "grad_norm": 17.681974411010742,
      "learning_rate": 2e-05,
      "loss": 3.524,
      "step": 2
    },
    {
      "epoch": 0.030456852791878174,
      "grad_norm": 10.733755111694336,
      "learning_rate": 3e-05,
      "loss": 2.6168,
      "step": 3
    },
    {
      "epoch": 0.04060913705583756,
      "grad_norm": 11.69923210144043,
      "learning_rate": 4e-05,
      "loss": 3.0268,
      "step": 4
    },
    {
      "epoch": 0.050761421319796954,
      "grad_norm": 9.91382884979248,
      "learning_rate": 5e-05,
      "loss": 2.6983,
      "step": 5
    },
    {
      "epoch": 0.06091370558375635,
      "grad_norm": 8.325790405273438,
      "learning_rate": 6e-05,
      "loss": 2.6117,
      "step": 6
    },
    {
      "epoch": 0.07106598984771574,
      "grad_norm": 8.814705848693848,
      "learning_rate": 7e-05,
      "loss": 1.7984,
      "step": 7
    },
    {
      "epoch": 0.08121827411167512,
      "grad_norm": 6.311023712158203,
      "learning_rate": 8e-05,
      "loss": 1.6477,
      "step": 8
    },
    {
      "epoch": 0.09137055837563451,
      "grad_norm": 9.39177417755127,
      "learning_rate": 9e-05,
      "loss": 1.5077,
      "step": 9
    },
    {
      "epoch": 0.10152284263959391,
      "grad_norm": 7.047640323638916,
      "learning_rate": 0.0001,
      "loss": 1.1966,
      "step": 10
    },
    {
      "epoch": 0.1116751269035533,
      "grad_norm": 7.027691841125488,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.2913,
      "step": 11
    },
    {
      "epoch": 0.1218274111675127,
      "grad_norm": 7.180845737457275,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.3658,
      "step": 12
    },
    {
      "epoch": 0.1319796954314721,
      "grad_norm": 5.794748306274414,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.0353,
      "step": 13
    },
    {
      "epoch": 0.14213197969543148,
      "grad_norm": 6.452200412750244,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.065,
      "step": 14
    },
    {
      "epoch": 0.15228426395939088,
      "grad_norm": 4.539834976196289,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.7424,
      "step": 15
    },
    {
      "epoch": 0.16243654822335024,
      "grad_norm": 3.665280818939209,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.5421,
      "step": 16
    },
    {
      "epoch": 0.17258883248730963,
      "grad_norm": 5.097992420196533,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6278,
      "step": 17
    },
    {
      "epoch": 0.18274111675126903,
      "grad_norm": 4.158536434173584,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.6224,
      "step": 18
    },
    {
      "epoch": 0.19289340101522842,
      "grad_norm": 6.671938419342041,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9401,
      "step": 19
    },
    {
      "epoch": 0.20304568527918782,
      "grad_norm": 5.329474449157715,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.0451,
      "step": 20
    },
    {
      "epoch": 0.2131979695431472,
      "grad_norm": 9.060242652893066,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.967,
      "step": 21
    },
    {
      "epoch": 0.2233502538071066,
      "grad_norm": 5.815667629241943,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7141,
      "step": 22
    },
    {
      "epoch": 0.233502538071066,
      "grad_norm": 7.216561794281006,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1294,
      "step": 23
    },
    {
      "epoch": 0.2436548223350254,
      "grad_norm": 2.3890023231506348,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6643,
      "step": 24
    },
    {
      "epoch": 0.25380710659898476,
      "grad_norm": 10.254504203796387,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1532,
      "step": 25
    },
    {
      "epoch": 0.2639593908629442,
      "grad_norm": 6.4724812507629395,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8319,
      "step": 26
    },
    {
      "epoch": 0.27411167512690354,
      "grad_norm": 4.96103572845459,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9737,
      "step": 27
    },
    {
      "epoch": 0.28426395939086296,
      "grad_norm": 3.7735328674316406,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.7217,
      "step": 28
    },
    {
      "epoch": 0.29441624365482233,
      "grad_norm": 5.711489200592041,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.747,
      "step": 29
    },
    {
      "epoch": 0.30456852791878175,
      "grad_norm": 3.950021505355835,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.5511,
      "step": 30
    },
    {
      "epoch": 0.3147208121827411,
      "grad_norm": 3.105947494506836,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.5659,
      "step": 31
    },
    {
      "epoch": 0.3248730964467005,
      "grad_norm": 3.8298869132995605,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.6114,
      "step": 32
    },
    {
      "epoch": 0.3350253807106599,
      "grad_norm": 3.8430259227752686,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.6099,
      "step": 33
    },
    {
      "epoch": 0.34517766497461927,
      "grad_norm": 4.849984645843506,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.7488,
      "step": 34
    },
    {
      "epoch": 0.3553299492385787,
      "grad_norm": 7.143765449523926,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.7675,
      "step": 35
    },
    {
      "epoch": 0.36548223350253806,
      "grad_norm": 3.5498955249786377,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.6389,
      "step": 36
    },
    {
      "epoch": 0.3756345177664975,
      "grad_norm": 3.149752140045166,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.426,
      "step": 37
    },
    {
      "epoch": 0.38578680203045684,
      "grad_norm": 2.729262590408325,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.4492,
      "step": 38
    },
    {
      "epoch": 0.39593908629441626,
      "grad_norm": 2.9004666805267334,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.4228,
      "step": 39
    },
    {
      "epoch": 0.40609137055837563,
      "grad_norm": 2.2637741565704346,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.3618,
      "step": 40
    },
    {
      "epoch": 0.41624365482233505,
      "grad_norm": 1.6494472026824951,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3304,
      "step": 41
    },
    {
      "epoch": 0.4263959390862944,
      "grad_norm": 1.4239898920059204,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.3048,
      "step": 42
    },
    {
      "epoch": 0.4365482233502538,
      "grad_norm": 4.939732074737549,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.4393,
      "step": 43
    },
    {
      "epoch": 0.4467005076142132,
      "grad_norm": 5.306370258331299,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.7552,
      "step": 44
    },
    {
      "epoch": 0.45685279187817257,
      "grad_norm": 2.6042845249176025,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.4825,
      "step": 45
    },
    {
      "epoch": 0.467005076142132,
      "grad_norm": 3.2139954566955566,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5494,
      "step": 46
    },
    {
      "epoch": 0.47715736040609136,
      "grad_norm": 3.111391305923462,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.6597,
      "step": 47
    },
    {
      "epoch": 0.4873096446700508,
      "grad_norm": 8.200444221496582,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.6122,
      "step": 48
    },
    {
      "epoch": 0.49746192893401014,
      "grad_norm": 5.513930320739746,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.8008,
      "step": 49
    },
    {
      "epoch": 0.5076142131979695,
      "grad_norm": 3.412057876586914,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4439,
      "step": 50
    },
    {
      "epoch": 0.5076142131979695,
      "eval_loss": 0.6668925285339355,
      "eval_runtime": 12.674,
      "eval_samples_per_second": 13.098,
      "eval_steps_per_second": 3.314,
      "step": 50
    },
    {
      "epoch": 0.5177664974619289,
      "grad_norm": 2.4831855297088623,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.4458,
      "step": 51
    },
    {
      "epoch": 0.5279187817258884,
      "grad_norm": 2.6279590129852295,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.5793,
      "step": 52
    },
    {
      "epoch": 0.5380710659898477,
      "grad_norm": 2.8039629459381104,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.5588,
      "step": 53
    },
    {
      "epoch": 0.5482233502538071,
      "grad_norm": 2.5664989948272705,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6391,
      "step": 54
    },
    {
      "epoch": 0.5583756345177665,
      "grad_norm": 3.4372622966766357,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4629,
      "step": 55
    },
    {
      "epoch": 0.5685279187817259,
      "grad_norm": 1.9379569292068481,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.4363,
      "step": 56
    },
    {
      "epoch": 0.5786802030456852,
      "grad_norm": 2.2308216094970703,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.4166,
      "step": 57
    },
    {
      "epoch": 0.5888324873096447,
      "grad_norm": 1.8501418828964233,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3284,
      "step": 58
    },
    {
      "epoch": 0.5989847715736041,
      "grad_norm": 2.6845145225524902,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.4415,
      "step": 59
    },
    {
      "epoch": 0.6091370558375635,
      "grad_norm": 3.932295083999634,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.6329,
      "step": 60
    },
    {
      "epoch": 0.6192893401015228,
      "grad_norm": 3.1955435276031494,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.512,
      "step": 61
    },
    {
      "epoch": 0.6294416243654822,
      "grad_norm": 2.334563732147217,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.4117,
      "step": 62
    },
    {
      "epoch": 0.6395939086294417,
      "grad_norm": 1.953082799911499,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.2918,
      "step": 63
    },
    {
      "epoch": 0.649746192893401,
      "grad_norm": 1.385120153427124,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.2518,
      "step": 64
    },
    {
      "epoch": 0.6598984771573604,
      "grad_norm": 5.226515293121338,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3715,
      "step": 65
    },
    {
      "epoch": 0.6700507614213198,
      "grad_norm": 2.116763114929199,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.344,
      "step": 66
    },
    {
      "epoch": 0.6802030456852792,
      "grad_norm": 3.010911703109741,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3712,
      "step": 67
    },
    {
      "epoch": 0.6903553299492385,
      "grad_norm": 4.351038455963135,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5257,
      "step": 68
    },
    {
      "epoch": 0.700507614213198,
      "grad_norm": 2.020462989807129,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3541,
      "step": 69
    },
    {
      "epoch": 0.7106598984771574,
      "grad_norm": 2.0711519718170166,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.5343,
      "step": 70
    },
    {
      "epoch": 0.7208121827411168,
      "grad_norm": 3.301103115081787,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.7541,
      "step": 71
    },
    {
      "epoch": 0.7309644670050761,
      "grad_norm": 1.9120086431503296,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5172,
      "step": 72
    },
    {
      "epoch": 0.7411167512690355,
      "grad_norm": 24.710140228271484,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3123,
      "step": 73
    },
    {
      "epoch": 0.751269035532995,
      "grad_norm": 4.699228763580322,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.6601,
      "step": 74
    },
    {
      "epoch": 0.7614213197969543,
      "grad_norm": 3.6631758213043213,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5681,
      "step": 75
    },
    {
      "epoch": 0.7715736040609137,
      "grad_norm": 2.7861716747283936,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.4676,
      "step": 76
    },
    {
      "epoch": 0.7817258883248731,
      "grad_norm": 1.8518892526626587,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.4754,
      "step": 77
    },
    {
      "epoch": 0.7918781725888325,
      "grad_norm": 2.3046631813049316,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4604,
      "step": 78
    },
    {
      "epoch": 0.8020304568527918,
      "grad_norm": 1.768671989440918,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.4007,
      "step": 79
    },
    {
      "epoch": 0.8121827411167513,
      "grad_norm": 2.345764398574829,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.4764,
      "step": 80
    },
    {
      "epoch": 0.8223350253807107,
      "grad_norm": 7.554089546203613,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.6176,
      "step": 81
    },
    {
      "epoch": 0.8324873096446701,
      "grad_norm": 5.878147602081299,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.4675,
      "step": 82
    },
    {
      "epoch": 0.8426395939086294,
      "grad_norm": 2.7786145210266113,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.4993,
      "step": 83
    },
    {
      "epoch": 0.8527918781725888,
      "grad_norm": 2.411370277404785,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4866,
      "step": 84
    },
    {
      "epoch": 0.8629441624365483,
      "grad_norm": 1.3441894054412842,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3481,
      "step": 85
    },
    {
      "epoch": 0.8730964467005076,
      "grad_norm": 1.3834481239318848,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.3466,
      "step": 86
    },
    {
      "epoch": 0.883248730964467,
      "grad_norm": 1.2694226503372192,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4229,
      "step": 87
    },
    {
      "epoch": 0.8934010152284264,
      "grad_norm": 1.4985469579696655,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3354,
      "step": 88
    },
    {
      "epoch": 0.9035532994923858,
      "grad_norm": 0.8708801865577698,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3356,
      "step": 89
    },
    {
      "epoch": 0.9137055837563451,
      "grad_norm": 1.079602837562561,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4371,
      "step": 90
    },
    {
      "epoch": 0.9238578680203046,
      "grad_norm": 1.9462090730667114,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.4388,
      "step": 91
    },
    {
      "epoch": 0.934010152284264,
      "grad_norm": 1.000754952430725,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.319,
      "step": 92
    },
    {
      "epoch": 0.9441624365482234,
      "grad_norm": 1.6465628147125244,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4011,
      "step": 93
    },
    {
      "epoch": 0.9543147208121827,
      "grad_norm": 2.106436252593994,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.4945,
      "step": 94
    },
    {
      "epoch": 0.9644670050761421,
      "grad_norm": 2.4169046878814697,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.5072,
      "step": 95
    },
    {
      "epoch": 0.9746192893401016,
      "grad_norm": 2.873663902282715,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.5012,
      "step": 96
    },
    {
      "epoch": 0.9847715736040609,
      "grad_norm": 5.384432315826416,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.7218,
      "step": 97
    },
    {
      "epoch": 0.9949238578680203,
      "grad_norm": 5.002354621887207,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.6874,
      "step": 98
    },
    {
      "epoch": 1.0050761421319796,
      "grad_norm": 3.0200672149658203,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.6584,
      "step": 99
    },
    {
      "epoch": 1.015228426395939,
      "grad_norm": 2.190378189086914,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3692,
      "step": 100
    },
    {
      "epoch": 1.015228426395939,
      "eval_loss": 0.4789780378341675,
      "eval_runtime": 12.6623,
      "eval_samples_per_second": 13.11,
      "eval_steps_per_second": 3.317,
      "step": 100
    },
    {
      "epoch": 1.0253807106598984,
      "grad_norm": 2.6336112022399902,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3425,
      "step": 101
    },
    {
      "epoch": 1.0355329949238579,
      "grad_norm": 1.9283448457717896,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3778,
      "step": 102
    },
    {
      "epoch": 1.0456852791878173,
      "grad_norm": 1.7850037813186646,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3455,
      "step": 103
    },
    {
      "epoch": 1.0558375634517767,
      "grad_norm": 2.4555656909942627,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4618,
      "step": 104
    },
    {
      "epoch": 1.0659898477157361,
      "grad_norm": 2.6725573539733887,
      "learning_rate": 5e-05,
      "loss": 0.445,
      "step": 105
    },
    {
      "epoch": 1.0761421319796955,
      "grad_norm": 2.472411632537842,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.4347,
      "step": 106
    },
    {
      "epoch": 1.0862944162436547,
      "grad_norm": 2.2673721313476562,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.38,
      "step": 107
    },
    {
      "epoch": 1.0964467005076142,
      "grad_norm": 2.717268705368042,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4329,
      "step": 108
    },
    {
      "epoch": 1.1065989847715736,
      "grad_norm": 2.6598331928253174,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4417,
      "step": 109
    },
    {
      "epoch": 1.116751269035533,
      "grad_norm": 1.8067651987075806,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3309,
      "step": 110
    },
    {
      "epoch": 1.1269035532994924,
      "grad_norm": 1.8983044624328613,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2791,
      "step": 111
    },
    {
      "epoch": 1.1370558375634519,
      "grad_norm": 0.9728879928588867,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.2618,
      "step": 112
    },
    {
      "epoch": 1.1472081218274113,
      "grad_norm": 1.8422456979751587,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3218,
      "step": 113
    },
    {
      "epoch": 1.1573604060913705,
      "grad_norm": 1.2632497549057007,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.2827,
      "step": 114
    },
    {
      "epoch": 1.16751269035533,
      "grad_norm": 0.8235652446746826,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2807,
      "step": 115
    },
    {
      "epoch": 1.1776649746192893,
      "grad_norm": 1.4295490980148315,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3327,
      "step": 116
    },
    {
      "epoch": 1.1878172588832487,
      "grad_norm": 0.9367296695709229,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3271,
      "step": 117
    },
    {
      "epoch": 1.1979695431472082,
      "grad_norm": 1.0259267091751099,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2927,
      "step": 118
    },
    {
      "epoch": 1.2081218274111676,
      "grad_norm": 1.3881288766860962,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3519,
      "step": 119
    },
    {
      "epoch": 1.218274111675127,
      "grad_norm": 2.625000476837158,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.4099,
      "step": 120
    },
    {
      "epoch": 1.2284263959390862,
      "grad_norm": 2.4155685901641846,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.5656,
      "step": 121
    },
    {
      "epoch": 1.2385786802030456,
      "grad_norm": 1.8987289667129517,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.5914,
      "step": 122
    },
    {
      "epoch": 1.248730964467005,
      "grad_norm": 1.2853312492370605,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3792,
      "step": 123
    },
    {
      "epoch": 1.2588832487309645,
      "grad_norm": 1.9290783405303955,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.4234,
      "step": 124
    },
    {
      "epoch": 1.2690355329949239,
      "grad_norm": 1.9319968223571777,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3708,
      "step": 125
    },
    {
      "epoch": 1.2791878172588833,
      "grad_norm": 1.707262635231018,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3728,
      "step": 126
    },
    {
      "epoch": 1.2893401015228427,
      "grad_norm": 1.531927466392517,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.351,
      "step": 127
    },
    {
      "epoch": 1.299492385786802,
      "grad_norm": 1.3563967943191528,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.393,
      "step": 128
    },
    {
      "epoch": 1.3096446700507614,
      "grad_norm": 3.904257297515869,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.418,
      "step": 129
    },
    {
      "epoch": 1.3197969543147208,
      "grad_norm": 1.139366865158081,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3997,
      "step": 130
    },
    {
      "epoch": 1.3299492385786802,
      "grad_norm": 1.3162213563919067,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.3768,
      "step": 131
    },
    {
      "epoch": 1.3401015228426396,
      "grad_norm": 1.5603002309799194,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.4503,
      "step": 132
    },
    {
      "epoch": 1.350253807106599,
      "grad_norm": 1.1917487382888794,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.403,
      "step": 133
    },
    {
      "epoch": 1.3604060913705585,
      "grad_norm": 0.8971186876296997,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3198,
      "step": 134
    },
    {
      "epoch": 1.3705583756345177,
      "grad_norm": 0.9901673793792725,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.2977,
      "step": 135
    },
    {
      "epoch": 1.380710659898477,
      "grad_norm": 0.7619752287864685,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2398,
      "step": 136
    },
    {
      "epoch": 1.3908629441624365,
      "grad_norm": 1.069899559020996,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1968,
      "step": 137
    },
    {
      "epoch": 1.401015228426396,
      "grad_norm": 1.2586966753005981,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.2597,
      "step": 138
    },
    {
      "epoch": 1.4111675126903553,
      "grad_norm": 1.2916020154953003,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.2951,
      "step": 139
    },
    {
      "epoch": 1.4213197969543148,
      "grad_norm": 1.2144585847854614,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3118,
      "step": 140
    },
    {
      "epoch": 1.4314720812182742,
      "grad_norm": 1.1343133449554443,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.3382,
      "step": 141
    },
    {
      "epoch": 1.4416243654822334,
      "grad_norm": 1.4946447610855103,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.3771,
      "step": 142
    },
    {
      "epoch": 1.451776649746193,
      "grad_norm": 1.5617653131484985,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.4045,
      "step": 143
    },
    {
      "epoch": 1.4619289340101522,
      "grad_norm": 2.2020020484924316,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.4597,
      "step": 144
    },
    {
      "epoch": 1.4720812182741116,
      "grad_norm": 2.495849609375,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.5254,
      "step": 145
    },
    {
      "epoch": 1.482233502538071,
      "grad_norm": 1.796535611152649,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.5499,
      "step": 146
    },
    {
      "epoch": 1.4923857868020305,
      "grad_norm": 1.5350875854492188,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.454,
      "step": 147
    },
    {
      "epoch": 1.50253807106599,
      "grad_norm": 1.765089750289917,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.3348,
      "step": 148
    },
    {
      "epoch": 1.512690355329949,
      "grad_norm": 1.3076599836349487,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.2877,
      "step": 149
    },
    {
      "epoch": 1.5228426395939088,
      "grad_norm": 1.2580403089523315,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2763,
      "step": 150
    },
    {
      "epoch": 1.5228426395939088,
      "eval_loss": 0.3739900588989258,
      "eval_runtime": 12.6587,
      "eval_samples_per_second": 13.114,
      "eval_steps_per_second": 3.318,
      "step": 150
    },
    {
      "epoch": 1.532994923857868,
      "grad_norm": 1.5276485681533813,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.3218,
      "step": 151
    },
    {
      "epoch": 1.5431472081218274,
      "grad_norm": 1.4127485752105713,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.3527,
      "step": 152
    },
    {
      "epoch": 1.5532994923857868,
      "grad_norm": 1.2763543128967285,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3405,
      "step": 153
    },
    {
      "epoch": 1.5634517766497462,
      "grad_norm": 1.2161738872528076,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3169,
      "step": 154
    },
    {
      "epoch": 1.5736040609137056,
      "grad_norm": 0.8805043697357178,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2952,
      "step": 155
    },
    {
      "epoch": 1.5837563451776648,
      "grad_norm": 1.123992681503296,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3274,
      "step": 156
    },
    {
      "epoch": 1.5939086294416245,
      "grad_norm": 1.3950940370559692,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3414,
      "step": 157
    },
    {
      "epoch": 1.6040609137055837,
      "grad_norm": 2.3260064125061035,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3885,
      "step": 158
    },
    {
      "epoch": 1.614213197969543,
      "grad_norm": 0.8243120312690735,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.3201,
      "step": 159
    },
    {
      "epoch": 1.6243654822335025,
      "grad_norm": 0.7181909680366516,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3112,
      "step": 160
    },
    {
      "epoch": 1.634517766497462,
      "grad_norm": 1.1954624652862549,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3385,
      "step": 161
    },
    {
      "epoch": 1.6446700507614214,
      "grad_norm": 0.6222485303878784,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2508,
      "step": 162
    },
    {
      "epoch": 1.6548223350253806,
      "grad_norm": 0.6330597400665283,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2653,
      "step": 163
    },
    {
      "epoch": 1.6649746192893402,
      "grad_norm": 0.671626091003418,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2619,
      "step": 164
    },
    {
      "epoch": 1.6751269035532994,
      "grad_norm": 0.9608164429664612,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2914,
      "step": 165
    },
    {
      "epoch": 1.6852791878172588,
      "grad_norm": 1.1137410402297974,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.3301,
      "step": 166
    },
    {
      "epoch": 1.6954314720812182,
      "grad_norm": 1.2798115015029907,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3989,
      "step": 167
    },
    {
      "epoch": 1.7055837563451777,
      "grad_norm": 1.5007836818695068,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.4485,
      "step": 168
    },
    {
      "epoch": 1.715736040609137,
      "grad_norm": 2.147376298904419,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.5758,
      "step": 169
    },
    {
      "epoch": 1.7258883248730963,
      "grad_norm": 2.042949676513672,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.5126,
      "step": 170
    },
    {
      "epoch": 1.736040609137056,
      "grad_norm": 1.5477014780044556,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3637,
      "step": 171
    },
    {
      "epoch": 1.7461928934010151,
      "grad_norm": 0.8590633273124695,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.2695,
      "step": 172
    },
    {
      "epoch": 1.7563451776649748,
      "grad_norm": 0.7962092161178589,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.1969,
      "step": 173
    },
    {
      "epoch": 1.766497461928934,
      "grad_norm": 0.5653324723243713,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2178,
      "step": 174
    },
    {
      "epoch": 1.7766497461928934,
      "grad_norm": 0.6608909368515015,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.2408,
      "step": 175
    },
    {
      "epoch": 1.7868020304568528,
      "grad_norm": 0.5651311874389648,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2586,
      "step": 176
    },
    {
      "epoch": 1.796954314720812,
      "grad_norm": 0.6455914974212646,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.2546,
      "step": 177
    },
    {
      "epoch": 1.8071065989847717,
      "grad_norm": 0.6791276335716248,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2584,
      "step": 178
    },
    {
      "epoch": 1.8172588832487309,
      "grad_norm": 0.6901379227638245,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.269,
      "step": 179
    },
    {
      "epoch": 1.8274111675126905,
      "grad_norm": 0.9977598786354065,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.2988,
      "step": 180
    },
    {
      "epoch": 1.8375634517766497,
      "grad_norm": 0.8531815409660339,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2955,
      "step": 181
    },
    {
      "epoch": 1.8477157360406091,
      "grad_norm": 0.8325989842414856,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.337,
      "step": 182
    },
    {
      "epoch": 1.8578680203045685,
      "grad_norm": 0.7359166741371155,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.273,
      "step": 183
    },
    {
      "epoch": 1.868020304568528,
      "grad_norm": 0.6772722005844116,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.2762,
      "step": 184
    },
    {
      "epoch": 1.8781725888324874,
      "grad_norm": 0.5276232361793518,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.2595,
      "step": 185
    },
    {
      "epoch": 1.8883248730964466,
      "grad_norm": 0.5787997841835022,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2377,
      "step": 186
    },
    {
      "epoch": 1.8984771573604062,
      "grad_norm": 0.6214790344238281,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2268,
      "step": 187
    },
    {
      "epoch": 1.9086294416243654,
      "grad_norm": 0.6290463209152222,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2465,
      "step": 188
    },
    {
      "epoch": 1.9187817258883249,
      "grad_norm": 0.6768893599510193,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2728,
      "step": 189
    },
    {
      "epoch": 1.9289340101522843,
      "grad_norm": 0.7624281048774719,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.2848,
      "step": 190
    },
    {
      "epoch": 1.9390862944162437,
      "grad_norm": 0.7929137349128723,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.3134,
      "step": 191
    },
    {
      "epoch": 1.9492385786802031,
      "grad_norm": 1.205314040184021,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.348,
      "step": 192
    },
    {
      "epoch": 1.9593908629441623,
      "grad_norm": 1.4788498878479004,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.4731,
      "step": 193
    },
    {
      "epoch": 1.969543147208122,
      "grad_norm": 1.6881470680236816,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.4911,
      "step": 194
    },
    {
      "epoch": 1.9796954314720812,
      "grad_norm": 1.1335477828979492,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.3226,
      "step": 195
    },
    {
      "epoch": 1.9898477157360406,
      "grad_norm": 0.4012901484966278,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2269,
      "step": 196
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6436210870742798,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.5313,
      "step": 197
    },
    {
      "epoch": 2.010152284263959,
      "grad_norm": 0.9164676070213318,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2824,
      "step": 198
    },
    {
      "epoch": 2.020304568527919,
      "grad_norm": 0.7725908160209656,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.1912,
      "step": 199
    },
    {
      "epoch": 2.030456852791878,
      "grad_norm": 0.600430965423584,
      "learning_rate": 0.0,
      "loss": 0.2108,
      "step": 200
    },
    {
      "epoch": 2.030456852791878,
      "eval_loss": 0.32313525676727295,
      "eval_runtime": 12.6639,
      "eval_samples_per_second": 13.108,
      "eval_steps_per_second": 3.317,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 3.0123645116350464e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}