{
  "best_metric": 1.516052007675171,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.0230179028132993,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005115089514066497,
      "grad_norm": 9.48886775970459,
      "learning_rate": 1e-05,
      "loss": 5.5047,
      "step": 1
    },
    {
      "epoch": 0.005115089514066497,
      "eval_loss": 7.651984214782715,
      "eval_runtime": 23.2283,
      "eval_samples_per_second": 14.207,
      "eval_steps_per_second": 3.573,
      "step": 1
    },
    {
      "epoch": 0.010230179028132993,
      "grad_norm": 7.729719638824463,
      "learning_rate": 2e-05,
      "loss": 5.8985,
      "step": 2
    },
    {
      "epoch": 0.015345268542199489,
      "grad_norm": 8.052858352661133,
      "learning_rate": 3e-05,
      "loss": 6.0288,
      "step": 3
    },
    {
      "epoch": 0.020460358056265986,
      "grad_norm": 6.788303852081299,
      "learning_rate": 4e-05,
      "loss": 5.5813,
      "step": 4
    },
    {
      "epoch": 0.02557544757033248,
      "grad_norm": 4.8062262535095215,
      "learning_rate": 5e-05,
      "loss": 5.5641,
      "step": 5
    },
    {
      "epoch": 0.030690537084398978,
      "grad_norm": 3.997166872024536,
      "learning_rate": 6e-05,
      "loss": 5.9237,
      "step": 6
    },
    {
      "epoch": 0.03580562659846547,
      "grad_norm": 3.928126811981201,
      "learning_rate": 7e-05,
      "loss": 5.4307,
      "step": 7
    },
    {
      "epoch": 0.04092071611253197,
      "grad_norm": 5.754164695739746,
      "learning_rate": 8e-05,
      "loss": 4.9893,
      "step": 8
    },
    {
      "epoch": 0.04603580562659847,
      "grad_norm": 4.404843807220459,
      "learning_rate": 9e-05,
      "loss": 5.0629,
      "step": 9
    },
    {
      "epoch": 0.05115089514066496,
      "grad_norm": 4.053828716278076,
      "learning_rate": 0.0001,
      "loss": 4.5917,
      "step": 10
    },
    {
      "epoch": 0.056265984654731455,
      "grad_norm": 3.8155884742736816,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.2516,
      "step": 11
    },
    {
      "epoch": 0.061381074168797956,
      "grad_norm": 3.4340007305145264,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.174,
      "step": 12
    },
    {
      "epoch": 0.06649616368286446,
      "grad_norm": 2.701197385787964,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.8316,
      "step": 13
    },
    {
      "epoch": 0.07161125319693094,
      "grad_norm": 3.004122734069824,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.987,
      "step": 14
    },
    {
      "epoch": 0.07672634271099744,
      "grad_norm": 2.845054864883423,
      "learning_rate": 9.98292246503335e-05,
      "loss": 3.6807,
      "step": 15
    },
    {
      "epoch": 0.08184143222506395,
      "grad_norm": 3.3714535236358643,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.0097,
      "step": 16
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 2.416005849838257,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.2052,
      "step": 17
    },
    {
      "epoch": 0.09207161125319693,
      "grad_norm": 2.8669612407684326,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.2715,
      "step": 18
    },
    {
      "epoch": 0.09718670076726342,
      "grad_norm": 2.7934770584106445,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.0802,
      "step": 19
    },
    {
      "epoch": 0.10230179028132992,
      "grad_norm": 2.592581272125244,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.1131,
      "step": 20
    },
    {
      "epoch": 0.10741687979539642,
      "grad_norm": 3.4438283443450928,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.175,
      "step": 21
    },
    {
      "epoch": 0.11253196930946291,
      "grad_norm": 3.016294479370117,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.5276,
      "step": 22
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 2.802727222442627,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.9135,
      "step": 23
    },
    {
      "epoch": 0.12276214833759591,
      "grad_norm": 2.832228422164917,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.1054,
      "step": 24
    },
    {
      "epoch": 0.1278772378516624,
      "grad_norm": 2.6756863594055176,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.9628,
      "step": 25
    },
    {
      "epoch": 0.1329923273657289,
      "grad_norm": 2.733081817626953,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.9537,
      "step": 26
    },
    {
      "epoch": 0.13810741687979539,
      "grad_norm": 2.7606492042541504,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.0507,
      "step": 27
    },
    {
      "epoch": 0.1432225063938619,
      "grad_norm": 2.96380877494812,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.7823,
      "step": 28
    },
    {
      "epoch": 0.1483375959079284,
      "grad_norm": 3.0976133346557617,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.9072,
      "step": 29
    },
    {
      "epoch": 0.1534526854219949,
      "grad_norm": 3.3771605491638184,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.009,
      "step": 30
    },
    {
      "epoch": 0.1585677749360614,
      "grad_norm": 3.442176580429077,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.603,
      "step": 31
    },
    {
      "epoch": 0.1636828644501279,
      "grad_norm": 3.1501970291137695,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.0066,
      "step": 32
    },
    {
      "epoch": 0.16879795396419436,
      "grad_norm": 3.4222052097320557,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.802,
      "step": 33
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 3.5420730113983154,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.795,
      "step": 34
    },
    {
      "epoch": 0.17902813299232737,
      "grad_norm": 3.4514718055725098,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.4177,
      "step": 35
    },
    {
      "epoch": 0.18414322250639387,
      "grad_norm": 4.618533134460449,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.9886,
      "step": 36
    },
    {
      "epoch": 0.18925831202046037,
      "grad_norm": 4.481081485748291,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0961,
      "step": 37
    },
    {
      "epoch": 0.19437340153452684,
      "grad_norm": 4.901217460632324,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.8523,
      "step": 38
    },
    {
      "epoch": 0.19948849104859334,
      "grad_norm": 3.705735921859741,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.8015,
      "step": 39
    },
    {
      "epoch": 0.20460358056265984,
      "grad_norm": 1.5407553911209106,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.1171,
      "step": 40
    },
    {
      "epoch": 0.20971867007672634,
      "grad_norm": 2.0642812252044678,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3588,
      "step": 41
    },
    {
      "epoch": 0.21483375959079284,
      "grad_norm": 2.8660378456115723,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.2585,
      "step": 42
    },
    {
      "epoch": 0.21994884910485935,
      "grad_norm": 2.2530696392059326,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.3681,
      "step": 43
    },
    {
      "epoch": 0.22506393861892582,
      "grad_norm": 2.2244601249694824,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.0182,
      "step": 44
    },
    {
      "epoch": 0.23017902813299232,
      "grad_norm": 0.11490941047668457,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0033,
      "step": 45
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.13940373063087463,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.0032,
      "step": 46
    },
    {
      "epoch": 0.24040920716112532,
      "grad_norm": 3.6119070053100586,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.1022,
      "step": 47
    },
    {
      "epoch": 0.24552429667519182,
      "grad_norm": 1.867747187614441,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0344,
      "step": 48
    },
    {
      "epoch": 0.2506393861892583,
      "grad_norm": 56.889617919921875,
      "learning_rate": 8.995939984474624e-05,
      "loss": 12.0252,
      "step": 49
    },
    {
      "epoch": 0.2557544757033248,
      "grad_norm": 51.18042755126953,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.018,
      "step": 50
    },
    {
      "epoch": 0.2557544757033248,
      "eval_loss": 4.55251932144165,
      "eval_runtime": 23.5332,
      "eval_samples_per_second": 14.023,
      "eval_steps_per_second": 3.527,
      "step": 50
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 33.81939697265625,
      "learning_rate": 8.894386393810563e-05,
      "loss": 6.0675,
      "step": 51
    },
    {
      "epoch": 0.2659846547314578,
      "grad_norm": 8.315890312194824,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.3208,
      "step": 52
    },
    {
      "epoch": 0.2710997442455243,
      "grad_norm": 6.041251182556152,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.0021,
      "step": 53
    },
    {
      "epoch": 0.27621483375959077,
      "grad_norm": 5.480395793914795,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.5547,
      "step": 54
    },
    {
      "epoch": 0.2813299232736573,
      "grad_norm": 4.745906352996826,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.1183,
      "step": 55
    },
    {
      "epoch": 0.2864450127877238,
      "grad_norm": 4.063253402709961,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.7754,
      "step": 56
    },
    {
      "epoch": 0.2915601023017903,
      "grad_norm": 2.9666452407836914,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.6057,
      "step": 57
    },
    {
      "epoch": 0.2966751918158568,
      "grad_norm": 2.41210675239563,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.9152,
      "step": 58
    },
    {
      "epoch": 0.30179028132992325,
      "grad_norm": 2.5263257026672363,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.5977,
      "step": 59
    },
    {
      "epoch": 0.3069053708439898,
      "grad_norm": 2.23828125,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.6563,
      "step": 60
    },
    {
      "epoch": 0.31202046035805625,
      "grad_norm": 2.502732276916504,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.5823,
      "step": 61
    },
    {
      "epoch": 0.3171355498721228,
      "grad_norm": 2.560762882232666,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.6343,
      "step": 62
    },
    {
      "epoch": 0.32225063938618925,
      "grad_norm": 2.3925020694732666,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.7233,
      "step": 63
    },
    {
      "epoch": 0.3273657289002558,
      "grad_norm": 2.2098705768585205,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.2702,
      "step": 64
    },
    {
      "epoch": 0.33248081841432225,
      "grad_norm": 2.209908962249756,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.1663,
      "step": 65
    },
    {
      "epoch": 0.3375959079283887,
      "grad_norm": 2.2588908672332764,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.2931,
      "step": 66
    },
    {
      "epoch": 0.34271099744245526,
      "grad_norm": 2.3859808444976807,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5883,
      "step": 67
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 2.467388391494751,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.2226,
      "step": 68
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 2.762615442276001,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.5631,
      "step": 69
    },
    {
      "epoch": 0.35805626598465473,
      "grad_norm": 2.483009099960327,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.3403,
      "step": 70
    },
    {
      "epoch": 0.3631713554987212,
      "grad_norm": 2.626396417617798,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.5624,
      "step": 71
    },
    {
      "epoch": 0.36828644501278773,
      "grad_norm": 2.52079439163208,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.0151,
      "step": 72
    },
    {
      "epoch": 0.3734015345268542,
      "grad_norm": 2.606149196624756,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2943,
      "step": 73
    },
    {
      "epoch": 0.37851662404092073,
      "grad_norm": 2.5749869346618652,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.2601,
      "step": 74
    },
    {
      "epoch": 0.3836317135549872,
      "grad_norm": 2.8876609802246094,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.3571,
      "step": 75
    },
    {
      "epoch": 0.3887468030690537,
      "grad_norm": 2.7740166187286377,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.2734,
      "step": 76
    },
    {
      "epoch": 0.3938618925831202,
      "grad_norm": 2.7462379932403564,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.246,
      "step": 77
    },
    {
      "epoch": 0.3989769820971867,
      "grad_norm": 2.99219012260437,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.1732,
      "step": 78
    },
    {
      "epoch": 0.4040920716112532,
      "grad_norm": 3.3825836181640625,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.2592,
      "step": 79
    },
    {
      "epoch": 0.4092071611253197,
      "grad_norm": 3.214312791824341,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.4639,
      "step": 80
    },
    {
      "epoch": 0.4143222506393862,
      "grad_norm": 3.0671980381011963,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.4281,
      "step": 81
    },
    {
      "epoch": 0.4194373401534527,
      "grad_norm": 3.599956512451172,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.6475,
      "step": 82
    },
    {
      "epoch": 0.42455242966751916,
      "grad_norm": 3.5274112224578857,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.9389,
      "step": 83
    },
    {
      "epoch": 0.4296675191815857,
      "grad_norm": 3.325232982635498,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.2739,
      "step": 84
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 2.8733978271484375,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2412,
      "step": 85
    },
    {
      "epoch": 0.4398976982097187,
      "grad_norm": 2.7152678966522217,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3306,
      "step": 86
    },
    {
      "epoch": 0.44501278772378516,
      "grad_norm": 1.1031886339187622,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.2063,
      "step": 87
    },
    {
      "epoch": 0.45012787723785164,
      "grad_norm": 1.2320791482925415,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.1947,
      "step": 88
    },
    {
      "epoch": 0.45524296675191817,
      "grad_norm": 2.740642786026001,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.7391,
      "step": 89
    },
    {
      "epoch": 0.46035805626598464,
      "grad_norm": 1.454041838645935,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0163,
      "step": 90
    },
    {
      "epoch": 0.46547314578005117,
      "grad_norm": 0.48821330070495605,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0089,
      "step": 91
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 1.3230316638946533,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1831,
      "step": 92
    },
    {
      "epoch": 0.47570332480818417,
      "grad_norm": 1.6790002584457397,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0349,
      "step": 93
    },
    {
      "epoch": 0.48081841432225064,
      "grad_norm": 0.4050397574901581,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0087,
      "step": 94
    },
    {
      "epoch": 0.4859335038363171,
      "grad_norm": 1.697842001914978,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.2403,
      "step": 95
    },
    {
      "epoch": 0.49104859335038364,
      "grad_norm": 0.3642440736293793,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.0087,
      "step": 96
    },
    {
      "epoch": 0.4961636828644501,
      "grad_norm": 9.899507522583008,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.3783,
      "step": 97
    },
    {
      "epoch": 0.5012787723785166,
      "grad_norm": 8.864073753356934,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.3658,
      "step": 98
    },
    {
      "epoch": 0.5063938618925832,
      "grad_norm": 5.942741394042969,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.8918,
      "step": 99
    },
    {
      "epoch": 0.5115089514066496,
      "grad_norm": 4.304798126220703,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.6705,
      "step": 100
    },
    {
      "epoch": 0.5115089514066496,
      "eval_loss": 2.1948964595794678,
      "eval_runtime": 23.5471,
      "eval_samples_per_second": 14.014,
      "eval_steps_per_second": 3.525,
      "step": 100
    },
    {
      "epoch": 0.5166240409207161,
      "grad_norm": 2.9918038845062256,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.7033,
      "step": 101
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 2.4725492000579834,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.5573,
      "step": 102
    },
    {
      "epoch": 0.5268542199488491,
      "grad_norm": 2.1304585933685303,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.1512,
      "step": 103
    },
    {
      "epoch": 0.5319693094629157,
      "grad_norm": 2.5665347576141357,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1427,
      "step": 104
    },
    {
      "epoch": 0.5370843989769821,
      "grad_norm": 2.6651530265808105,
      "learning_rate": 5e-05,
      "loss": 2.2871,
      "step": 105
    },
    {
      "epoch": 0.5421994884910486,
      "grad_norm": 2.7729225158691406,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.3458,
      "step": 106
    },
    {
      "epoch": 0.5473145780051151,
      "grad_norm": 2.565452814102173,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.1206,
      "step": 107
    },
    {
      "epoch": 0.5524296675191815,
      "grad_norm": 2.531663656234741,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.2027,
      "step": 108
    },
    {
      "epoch": 0.5575447570332481,
      "grad_norm": 2.5573108196258545,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0738,
      "step": 109
    },
    {
      "epoch": 0.5626598465473146,
      "grad_norm": 2.2978596687316895,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.2811,
      "step": 110
    },
    {
      "epoch": 0.5677749360613811,
      "grad_norm": 2.3859710693359375,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.3055,
      "step": 111
    },
    {
      "epoch": 0.5728900255754475,
      "grad_norm": 2.604112148284912,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.115,
      "step": 112
    },
    {
      "epoch": 0.578005115089514,
      "grad_norm": 2.5339314937591553,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.3342,
      "step": 113
    },
    {
      "epoch": 0.5831202046035806,
      "grad_norm": 2.5405259132385254,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.1979,
      "step": 114
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 2.4363160133361816,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.2799,
      "step": 115
    },
    {
      "epoch": 0.5933503836317136,
      "grad_norm": 2.7355687618255615,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.3658,
      "step": 116
    },
    {
      "epoch": 0.59846547314578,
      "grad_norm": 2.3593525886535645,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.0906,
      "step": 117
    },
    {
      "epoch": 0.6035805626598465,
      "grad_norm": 2.9883930683135986,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3255,
      "step": 118
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 2.663736581802368,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.5815,
      "step": 119
    },
    {
      "epoch": 0.6138107416879796,
      "grad_norm": 2.20088529586792,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.8641,
      "step": 120
    },
    {
      "epoch": 0.618925831202046,
      "grad_norm": 2.3998143672943115,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.1038,
      "step": 121
    },
    {
      "epoch": 0.6240409207161125,
      "grad_norm": 2.7294936180114746,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.3393,
      "step": 122
    },
    {
      "epoch": 0.629156010230179,
      "grad_norm": 2.7451772689819336,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.3762,
      "step": 123
    },
    {
      "epoch": 0.6342710997442456,
      "grad_norm": 2.3972485065460205,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8208,
      "step": 124
    },
    {
      "epoch": 0.639386189258312,
      "grad_norm": 2.821089029312134,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.2598,
      "step": 125
    },
    {
      "epoch": 0.6445012787723785,
      "grad_norm": 2.72977352142334,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.3335,
      "step": 126
    },
    {
      "epoch": 0.649616368286445,
      "grad_norm": 2.9516425132751465,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.2865,
      "step": 127
    },
    {
      "epoch": 0.6547314578005116,
      "grad_norm": 2.9510819911956787,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.4788,
      "step": 128
    },
    {
      "epoch": 0.659846547314578,
      "grad_norm": 2.730337619781494,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.9114,
      "step": 129
    },
    {
      "epoch": 0.6649616368286445,
      "grad_norm": 3.1395456790924072,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.2471,
      "step": 130
    },
    {
      "epoch": 0.670076726342711,
      "grad_norm": 3.3875725269317627,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.2192,
      "step": 131
    },
    {
      "epoch": 0.6751918158567775,
      "grad_norm": 3.5486505031585693,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4071,
      "step": 132
    },
    {
      "epoch": 0.680306905370844,
      "grad_norm": 2.0727181434631348,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.645,
      "step": 133
    },
    {
      "epoch": 0.6854219948849105,
      "grad_norm": 2.7907888889312744,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1249,
      "step": 134
    },
    {
      "epoch": 0.690537084398977,
      "grad_norm": 1.239582896232605,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0448,
      "step": 135
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.10967634618282318,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0033,
      "step": 136
    },
    {
      "epoch": 0.7007672634271099,
      "grad_norm": 1.6341110467910767,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.4667,
      "step": 137
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.01786043681204319,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0006,
      "step": 138
    },
    {
      "epoch": 0.710997442455243,
      "grad_norm": 0.048320382833480835,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0015,
      "step": 139
    },
    {
      "epoch": 0.7161125319693095,
      "grad_norm": 0.3364434540271759,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0049,
      "step": 140
    },
    {
      "epoch": 0.7212276214833759,
      "grad_norm": 0.11003550887107849,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0025,
      "step": 141
    },
    {
      "epoch": 0.7263427109974424,
      "grad_norm": 0.17777171730995178,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0019,
      "step": 142
    },
    {
      "epoch": 0.731457800511509,
      "grad_norm": 3.2541725635528564,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.8271,
      "step": 143
    },
    {
      "epoch": 0.7365728900255755,
      "grad_norm": 0.059293996542692184,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0025,
      "step": 144
    },
    {
      "epoch": 0.7416879795396419,
      "grad_norm": 5.065552711486816,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.8666,
      "step": 145
    },
    {
      "epoch": 0.7468030690537084,
      "grad_norm": 4.6517815589904785,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.091,
      "step": 146
    },
    {
      "epoch": 0.7519181585677749,
      "grad_norm": 4.511196613311768,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.2543,
      "step": 147
    },
    {
      "epoch": 0.7570332480818415,
      "grad_norm": 4.343446254730225,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.3404,
      "step": 148
    },
    {
      "epoch": 0.7621483375959079,
      "grad_norm": 3.1797878742218018,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.0131,
      "step": 149
    },
    {
      "epoch": 0.7672634271099744,
      "grad_norm": 2.811389684677124,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.287,
      "step": 150
    },
    {
      "epoch": 0.7672634271099744,
      "eval_loss": 1.606002926826477,
      "eval_runtime": 23.5262,
      "eval_samples_per_second": 14.027,
      "eval_steps_per_second": 3.528,
      "step": 150
    },
    {
      "epoch": 0.7723785166240409,
      "grad_norm": 2.290623188018799,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.0731,
      "step": 151
    },
    {
      "epoch": 0.7774936061381074,
      "grad_norm": 2.233295202255249,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.2991,
      "step": 152
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 2.1524369716644287,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9883,
      "step": 153
    },
    {
      "epoch": 0.7877237851662404,
      "grad_norm": 2.0260424613952637,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.804,
      "step": 154
    },
    {
      "epoch": 0.7928388746803069,
      "grad_norm": 1.8963541984558105,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.1641,
      "step": 155
    },
    {
      "epoch": 0.7979539641943734,
      "grad_norm": 1.8538708686828613,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.8249,
      "step": 156
    },
    {
      "epoch": 0.80306905370844,
      "grad_norm": 2.0748133659362793,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8712,
      "step": 157
    },
    {
      "epoch": 0.8081841432225064,
      "grad_norm": 2.037038803100586,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.2869,
      "step": 158
    },
    {
      "epoch": 0.8132992327365729,
      "grad_norm": 1.9473636150360107,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.1583,
      "step": 159
    },
    {
      "epoch": 0.8184143222506394,
      "grad_norm": 2.130141258239746,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.2796,
      "step": 160
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 2.16231369972229,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.1266,
      "step": 161
    },
    {
      "epoch": 0.8286445012787724,
      "grad_norm": 2.157963752746582,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.8107,
      "step": 162
    },
    {
      "epoch": 0.8337595907928389,
      "grad_norm": 2.0424017906188965,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.8851,
      "step": 163
    },
    {
      "epoch": 0.8388746803069054,
      "grad_norm": 2.6570191383361816,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.2362,
      "step": 164
    },
    {
      "epoch": 0.8439897698209718,
      "grad_norm": 2.5424201488494873,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.7362,
      "step": 165
    },
    {
      "epoch": 0.8491048593350383,
      "grad_norm": 2.377540111541748,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.986,
      "step": 166
    },
    {
      "epoch": 0.8542199488491049,
      "grad_norm": 2.2673635482788086,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.9098,
      "step": 167
    },
    {
      "epoch": 0.8593350383631714,
      "grad_norm": 2.72878098487854,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1336,
      "step": 168
    },
    {
      "epoch": 0.8644501278772379,
      "grad_norm": 2.5567519664764404,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0403,
      "step": 169
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 2.5409350395202637,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.3521,
      "step": 170
    },
    {
      "epoch": 0.8746803069053708,
      "grad_norm": 2.605874538421631,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.2019,
      "step": 171
    },
    {
      "epoch": 0.8797953964194374,
      "grad_norm": 2.6873183250427246,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.1183,
      "step": 172
    },
    {
      "epoch": 0.8849104859335039,
      "grad_norm": 3.1060965061187744,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.5017,
      "step": 173
    },
    {
      "epoch": 0.8900255754475703,
      "grad_norm": 3.0001893043518066,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.3379,
      "step": 174
    },
    {
      "epoch": 0.8951406649616368,
      "grad_norm": 2.7274348735809326,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.3073,
      "step": 175
    },
    {
      "epoch": 0.9002557544757033,
      "grad_norm": 2.838489055633545,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.0906,
      "step": 176
    },
    {
      "epoch": 0.9053708439897699,
      "grad_norm": 3.029648542404175,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.2339,
      "step": 177
    },
    {
      "epoch": 0.9104859335038363,
      "grad_norm": 3.465585708618164,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.372,
      "step": 178
    },
    {
      "epoch": 0.9156010230179028,
      "grad_norm": 3.2866809368133545,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.1481,
      "step": 179
    },
    {
      "epoch": 0.9207161125319693,
      "grad_norm": 3.1852052211761475,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.6515,
      "step": 180
    },
    {
      "epoch": 0.9258312020460358,
      "grad_norm": 2.4479782581329346,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.9021,
      "step": 181
    },
    {
      "epoch": 0.9309462915601023,
      "grad_norm": 2.6004116535186768,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.8208,
      "step": 182
    },
    {
      "epoch": 0.9360613810741688,
      "grad_norm": 0.04456498846411705,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0012,
      "step": 183
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 2.1983840465545654,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3821,
      "step": 184
    },
    {
      "epoch": 0.9462915601023018,
      "grad_norm": 2.3192474842071533,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.4354,
      "step": 185
    },
    {
      "epoch": 0.9514066496163683,
      "grad_norm": 0.11937044560909271,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0019,
      "step": 186
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.19781145453453064,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0034,
      "step": 187
    },
    {
      "epoch": 0.9616368286445013,
      "grad_norm": 1.5076892375946045,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2181,
      "step": 188
    },
    {
      "epoch": 0.9667519181585678,
      "grad_norm": 0.09216476231813431,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0028,
      "step": 189
    },
    {
      "epoch": 0.9718670076726342,
      "grad_norm": 0.14573799073696136,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0031,
      "step": 190
    },
    {
      "epoch": 0.9769820971867008,
      "grad_norm": 1.7069849967956543,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.351,
      "step": 191
    },
    {
      "epoch": 0.9820971867007673,
      "grad_norm": 1.4825915098190308,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0484,
      "step": 192
    },
    {
      "epoch": 0.9872122762148338,
      "grad_norm": 1.7542392015457153,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.7002,
      "step": 193
    },
    {
      "epoch": 0.9923273657289002,
      "grad_norm": 2.4199843406677246,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.3144,
      "step": 194
    },
    {
      "epoch": 0.9974424552429667,
      "grad_norm": 2.448477268218994,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0675,
      "step": 195
    },
    {
      "epoch": 1.0025575447570332,
      "grad_norm": 0.9669291973114014,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9477,
      "step": 196
    },
    {
      "epoch": 1.0076726342710998,
      "grad_norm": 1.7305530309677124,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.6039,
      "step": 197
    },
    {
      "epoch": 1.0127877237851663,
      "grad_norm": 1.5814672708511353,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.9185,
      "step": 198
    },
    {
      "epoch": 1.0179028132992327,
      "grad_norm": 1.9270161390304565,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.6387,
      "step": 199
    },
    {
      "epoch": 1.0230179028132993,
      "grad_norm": 1.5292868614196777,
      "learning_rate": 0.0,
      "loss": 1.486,
      "step": 200
    },
    {
      "epoch": 1.0230179028132993,
      "eval_loss": 1.516052007675171,
      "eval_runtime": 23.5337,
      "eval_samples_per_second": 14.022,
      "eval_steps_per_second": 3.527,
      "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": 2.6836619050038067e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}