{
  "best_metric": 0.773449718952179,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.46893317702227433,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023446658851113715,
      "grad_norm": 23.57974624633789,
      "learning_rate": 1e-05,
      "loss": 5.4807,
      "step": 1
    },
    {
      "epoch": 0.0023446658851113715,
      "eval_loss": 1.620888352394104,
      "eval_runtime": 52.78,
      "eval_samples_per_second": 13.604,
      "eval_steps_per_second": 3.41,
      "step": 1
    },
    {
      "epoch": 0.004689331770222743,
      "grad_norm": 30.60625648498535,
      "learning_rate": 2e-05,
      "loss": 5.8935,
      "step": 2
    },
    {
      "epoch": 0.007033997655334115,
      "grad_norm": 25.76280975341797,
      "learning_rate": 3e-05,
      "loss": 5.8374,
      "step": 3
    },
    {
      "epoch": 0.009378663540445486,
      "grad_norm": 23.49654769897461,
      "learning_rate": 4e-05,
      "loss": 5.3307,
      "step": 4
    },
    {
      "epoch": 0.011723329425556858,
      "grad_norm": 24.191038131713867,
      "learning_rate": 5e-05,
      "loss": 5.3859,
      "step": 5
    },
    {
      "epoch": 0.01406799531066823,
      "grad_norm": 16.579008102416992,
      "learning_rate": 6e-05,
      "loss": 4.4844,
      "step": 6
    },
    {
      "epoch": 0.016412661195779603,
      "grad_norm": 16.126588821411133,
      "learning_rate": 7e-05,
      "loss": 4.2335,
      "step": 7
    },
    {
      "epoch": 0.01875732708089097,
      "grad_norm": 14.067370414733887,
      "learning_rate": 8e-05,
      "loss": 4.0868,
      "step": 8
    },
    {
      "epoch": 0.021101992966002344,
      "grad_norm": 12.361042976379395,
      "learning_rate": 9e-05,
      "loss": 3.8966,
      "step": 9
    },
    {
      "epoch": 0.023446658851113716,
      "grad_norm": 11.765653610229492,
      "learning_rate": 0.0001,
      "loss": 4.0382,
      "step": 10
    },
    {
      "epoch": 0.02579132473622509,
      "grad_norm": 12.464676856994629,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.9801,
      "step": 11
    },
    {
      "epoch": 0.02813599062133646,
      "grad_norm": 10.50444507598877,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.9215,
      "step": 12
    },
    {
      "epoch": 0.03048065650644783,
      "grad_norm": 16.138383865356445,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.227,
      "step": 13
    },
    {
      "epoch": 0.032825322391559206,
      "grad_norm": 9.515331268310547,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.3742,
      "step": 14
    },
    {
      "epoch": 0.035169988276670575,
      "grad_norm": 12.985983848571777,
      "learning_rate": 9.98292246503335e-05,
      "loss": 3.7959,
      "step": 15
    },
    {
      "epoch": 0.03751465416178194,
      "grad_norm": 10.050265312194824,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.6855,
      "step": 16
    },
    {
      "epoch": 0.03985932004689332,
      "grad_norm": 9.893367767333984,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.4816,
      "step": 17
    },
    {
      "epoch": 0.04220398593200469,
      "grad_norm": 11.19422721862793,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.801,
      "step": 18
    },
    {
      "epoch": 0.044548651817116064,
      "grad_norm": 8.433975219726562,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.2922,
      "step": 19
    },
    {
      "epoch": 0.04689331770222743,
      "grad_norm": 9.601617813110352,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.1675,
      "step": 20
    },
    {
      "epoch": 0.0492379835873388,
      "grad_norm": 9.019408226013184,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.5708,
      "step": 21
    },
    {
      "epoch": 0.05158264947245018,
      "grad_norm": 9.71084976196289,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.5677,
      "step": 22
    },
    {
      "epoch": 0.053927315357561546,
      "grad_norm": 10.199828147888184,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.4368,
      "step": 23
    },
    {
      "epoch": 0.05627198124267292,
      "grad_norm": 9.20158863067627,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.0773,
      "step": 24
    },
    {
      "epoch": 0.05861664712778429,
      "grad_norm": 14.455418586730957,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.9972,
      "step": 25
    },
    {
      "epoch": 0.06096131301289566,
      "grad_norm": 9.237549781799316,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.4514,
      "step": 26
    },
    {
      "epoch": 0.06330597889800703,
      "grad_norm": 10.045668601989746,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.303,
      "step": 27
    },
    {
      "epoch": 0.06565064478311841,
      "grad_norm": 11.65676212310791,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.2973,
      "step": 28
    },
    {
      "epoch": 0.06799531066822978,
      "grad_norm": 8.70138931274414,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.0374,
      "step": 29
    },
    {
      "epoch": 0.07033997655334115,
      "grad_norm": 8.465364456176758,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.9405,
      "step": 30
    },
    {
      "epoch": 0.07268464243845252,
      "grad_norm": 9.00812816619873,
      "learning_rate": 9.701596950580806e-05,
      "loss": 3.0825,
      "step": 31
    },
    {
      "epoch": 0.07502930832356389,
      "grad_norm": 8.821876525878906,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.0537,
      "step": 32
    },
    {
      "epoch": 0.07737397420867527,
      "grad_norm": 9.36876106262207,
      "learning_rate": 9.642770192448536e-05,
      "loss": 3.2912,
      "step": 33
    },
    {
      "epoch": 0.07971864009378664,
      "grad_norm": 9.534127235412598,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.9173,
      "step": 34
    },
    {
      "epoch": 0.08206330597889801,
      "grad_norm": 8.567021369934082,
      "learning_rate": 9.578866633275288e-05,
      "loss": 3.3396,
      "step": 35
    },
    {
      "epoch": 0.08440797186400938,
      "grad_norm": 8.149921417236328,
      "learning_rate": 9.545032675245813e-05,
      "loss": 3.0252,
      "step": 36
    },
    {
      "epoch": 0.08675263774912075,
      "grad_norm": 10.303092002868652,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.282,
      "step": 37
    },
    {
      "epoch": 0.08909730363423213,
      "grad_norm": 9.2345552444458,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.2862,
      "step": 38
    },
    {
      "epoch": 0.0914419695193435,
      "grad_norm": 14.739202499389648,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.2216,
      "step": 39
    },
    {
      "epoch": 0.09378663540445487,
      "grad_norm": 7.720669269561768,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.3479,
      "step": 40
    },
    {
      "epoch": 0.09613130128956623,
      "grad_norm": 7.961097717285156,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.2198,
      "step": 41
    },
    {
      "epoch": 0.0984759671746776,
      "grad_norm": 7.674463272094727,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.83,
      "step": 42
    },
    {
      "epoch": 0.10082063305978899,
      "grad_norm": 7.735635280609131,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.7617,
      "step": 43
    },
    {
      "epoch": 0.10316529894490035,
      "grad_norm": 8.600885391235352,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.0858,
      "step": 44
    },
    {
      "epoch": 0.10550996483001172,
      "grad_norm": 8.574167251586914,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.5804,
      "step": 45
    },
    {
      "epoch": 0.10785463071512309,
      "grad_norm": 8.848064422607422,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.6468,
      "step": 46
    },
    {
      "epoch": 0.11019929660023446,
      "grad_norm": 8.83186149597168,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.4967,
      "step": 47
    },
    {
      "epoch": 0.11254396248534584,
      "grad_norm": 8.947534561157227,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.4509,
      "step": 48
    },
    {
      "epoch": 0.11488862837045721,
      "grad_norm": 10.69256591796875,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.1117,
      "step": 49
    },
    {
      "epoch": 0.11723329425556858,
      "grad_norm": 11.915793418884277,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.8231,
      "step": 50
    },
    {
      "epoch": 0.11723329425556858,
      "eval_loss": 1.075331449508667,
      "eval_runtime": 53.6232,
      "eval_samples_per_second": 13.39,
      "eval_steps_per_second": 3.357,
      "step": 50
    },
    {
      "epoch": 0.11957796014067995,
      "grad_norm": 24.798919677734375,
      "learning_rate": 8.894386393810563e-05,
      "loss": 6.8674,
      "step": 51
    },
    {
      "epoch": 0.12192262602579132,
      "grad_norm": 18.361974716186523,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.8898,
      "step": 52
    },
    {
      "epoch": 0.1242672919109027,
      "grad_norm": 10.277095794677734,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.2936,
      "step": 53
    },
    {
      "epoch": 0.12661195779601406,
      "grad_norm": 9.56214714050293,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.5164,
      "step": 54
    },
    {
      "epoch": 0.12895662368112543,
      "grad_norm": 9.94495964050293,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.725,
      "step": 55
    },
    {
      "epoch": 0.13130128956623682,
      "grad_norm": 8.306497573852539,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.9273,
      "step": 56
    },
    {
      "epoch": 0.1336459554513482,
      "grad_norm": 7.849757194519043,
      "learning_rate": 8.564642241456986e-05,
      "loss": 4.0073,
      "step": 57
    },
    {
      "epoch": 0.13599062133645956,
      "grad_norm": 9.636812210083008,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.9613,
      "step": 58
    },
    {
      "epoch": 0.13833528722157093,
      "grad_norm": 7.855923175811768,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.9367,
      "step": 59
    },
    {
      "epoch": 0.1406799531066823,
      "grad_norm": 7.7075300216674805,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.5982,
      "step": 60
    },
    {
      "epoch": 0.14302461899179367,
      "grad_norm": 9.229376792907715,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.6426,
      "step": 61
    },
    {
      "epoch": 0.14536928487690504,
      "grad_norm": 9.280519485473633,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.7981,
      "step": 62
    },
    {
      "epoch": 0.1477139507620164,
      "grad_norm": 8.732070922851562,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.6093,
      "step": 63
    },
    {
      "epoch": 0.15005861664712777,
      "grad_norm": 8.52567195892334,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.6317,
      "step": 64
    },
    {
      "epoch": 0.15240328253223914,
      "grad_norm": 8.398834228515625,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.482,
      "step": 65
    },
    {
      "epoch": 0.15474794841735054,
      "grad_norm": 7.95924186706543,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.834,
      "step": 66
    },
    {
      "epoch": 0.1570926143024619,
      "grad_norm": 6.904319763183594,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.486,
      "step": 67
    },
    {
      "epoch": 0.15943728018757328,
      "grad_norm": 7.4533185958862305,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.2828,
      "step": 68
    },
    {
      "epoch": 0.16178194607268465,
      "grad_norm": 8.788702011108398,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.8091,
      "step": 69
    },
    {
      "epoch": 0.16412661195779601,
      "grad_norm": 6.397895812988281,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.2773,
      "step": 70
    },
    {
      "epoch": 0.16647127784290738,
      "grad_norm": 8.078803062438965,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.948,
      "step": 71
    },
    {
      "epoch": 0.16881594372801875,
      "grad_norm": 7.596654415130615,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.989,
      "step": 72
    },
    {
      "epoch": 0.17116060961313012,
      "grad_norm": 7.248602867126465,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.0982,
      "step": 73
    },
    {
      "epoch": 0.1735052754982415,
      "grad_norm": 6.818299293518066,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.2279,
      "step": 74
    },
    {
      "epoch": 0.17584994138335286,
      "grad_norm": 6.427083969116211,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.1889,
      "step": 75
    },
    {
      "epoch": 0.17819460726846426,
      "grad_norm": 7.547183513641357,
      "learning_rate": 7.30670581489344e-05,
      "loss": 3.099,
      "step": 76
    },
    {
      "epoch": 0.18053927315357562,
      "grad_norm": 7.652503490447998,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7142,
      "step": 77
    },
    {
      "epoch": 0.182883939038687,
      "grad_norm": 8.411200523376465,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.0264,
      "step": 78
    },
    {
      "epoch": 0.18522860492379836,
      "grad_norm": 8.300537109375,
      "learning_rate": 7.083909302476453e-05,
      "loss": 3.1136,
      "step": 79
    },
    {
      "epoch": 0.18757327080890973,
      "grad_norm": 7.834685325622559,
      "learning_rate": 7.008477123264848e-05,
      "loss": 3.2318,
      "step": 80
    },
    {
      "epoch": 0.1899179366940211,
      "grad_norm": 6.925619125366211,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.5868,
      "step": 81
    },
    {
      "epoch": 0.19226260257913247,
      "grad_norm": 7.67003870010376,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.9803,
      "step": 82
    },
    {
      "epoch": 0.19460726846424384,
      "grad_norm": 7.498819351196289,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.7582,
      "step": 83
    },
    {
      "epoch": 0.1969519343493552,
      "grad_norm": 7.7189412117004395,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.937,
      "step": 84
    },
    {
      "epoch": 0.19929660023446658,
      "grad_norm": 7.78687858581543,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.1227,
      "step": 85
    },
    {
      "epoch": 0.20164126611957797,
      "grad_norm": 9.510988235473633,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.2845,
      "step": 86
    },
    {
      "epoch": 0.20398593200468934,
      "grad_norm": 9.355915069580078,
      "learning_rate": 6.466250186922325e-05,
      "loss": 3.2586,
      "step": 87
    },
    {
      "epoch": 0.2063305978898007,
      "grad_norm": 7.125824928283691,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.9818,
      "step": 88
    },
    {
      "epoch": 0.20867526377491208,
      "grad_norm": 8.271867752075195,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.9082,
      "step": 89
    },
    {
      "epoch": 0.21101992966002345,
      "grad_norm": 7.700958251953125,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.7223,
      "step": 90
    },
    {
      "epoch": 0.21336459554513482,
      "grad_norm": 7.786774635314941,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.1735,
      "step": 91
    },
    {
      "epoch": 0.21570926143024619,
      "grad_norm": 7.868963718414307,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.3055,
      "step": 92
    },
    {
      "epoch": 0.21805392731535755,
      "grad_norm": 7.272830009460449,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.8501,
      "step": 93
    },
    {
      "epoch": 0.22039859320046892,
      "grad_norm": 7.141085624694824,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.8644,
      "step": 94
    },
    {
      "epoch": 0.2227432590855803,
      "grad_norm": 7.3042402267456055,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.5173,
      "step": 95
    },
    {
      "epoch": 0.2250879249706917,
      "grad_norm": 7.959392547607422,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.7805,
      "step": 96
    },
    {
      "epoch": 0.22743259085580306,
      "grad_norm": 7.983731269836426,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.9661,
      "step": 97
    },
    {
      "epoch": 0.22977725674091443,
      "grad_norm": 6.660508632659912,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.2967,
      "step": 98
    },
    {
      "epoch": 0.2321219226260258,
      "grad_norm": 8.27916145324707,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.2433,
      "step": 99
    },
    {
      "epoch": 0.23446658851113716,
      "grad_norm": 7.3086466789245605,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4027,
      "step": 100
    },
    {
      "epoch": 0.23446658851113716,
      "eval_loss": 0.9120429158210754,
      "eval_runtime": 53.6028,
      "eval_samples_per_second": 13.395,
      "eval_steps_per_second": 3.358,
      "step": 100
    },
    {
      "epoch": 0.23681125439624853,
      "grad_norm": 12.511914253234863,
      "learning_rate": 5.330452921628497e-05,
      "loss": 5.1524,
      "step": 101
    },
    {
      "epoch": 0.2391559202813599,
      "grad_norm": 11.650274276733398,
      "learning_rate": 5.247918773366112e-05,
      "loss": 4.6861,
      "step": 102
    },
    {
      "epoch": 0.24150058616647127,
      "grad_norm": 8.172693252563477,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.8242,
      "step": 103
    },
    {
      "epoch": 0.24384525205158264,
      "grad_norm": 6.815614700317383,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 4.1423,
      "step": 104
    },
    {
      "epoch": 0.246189917936694,
      "grad_norm": 6.612921237945557,
      "learning_rate": 5e-05,
      "loss": 3.5026,
      "step": 105
    },
    {
      "epoch": 0.2485345838218054,
      "grad_norm": 8.0675687789917,
      "learning_rate": 4.917330276168208e-05,
      "loss": 4.0927,
      "step": 106
    },
    {
      "epoch": 0.2508792497069168,
      "grad_norm": 7.350432395935059,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.6943,
      "step": 107
    },
    {
      "epoch": 0.2532239155920281,
      "grad_norm": 6.199204921722412,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.3317,
      "step": 108
    },
    {
      "epoch": 0.2555685814771395,
      "grad_norm": 6.378794193267822,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.6218,
      "step": 109
    },
    {
      "epoch": 0.25791324736225085,
      "grad_norm": 6.892790794372559,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.4969,
      "step": 110
    },
    {
      "epoch": 0.26025791324736225,
      "grad_norm": 6.6119279861450195,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.4233,
      "step": 111
    },
    {
      "epoch": 0.26260257913247365,
      "grad_norm": 6.906591415405273,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.5802,
      "step": 112
    },
    {
      "epoch": 0.264947245017585,
      "grad_norm": 7.080374717712402,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 4.0366,
      "step": 113
    },
    {
      "epoch": 0.2672919109026964,
      "grad_norm": 6.781357288360596,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.6486,
      "step": 114
    },
    {
      "epoch": 0.2696365767878077,
      "grad_norm": 6.406153678894043,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.1465,
      "step": 115
    },
    {
      "epoch": 0.2719812426729191,
      "grad_norm": 6.648326396942139,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.3809,
      "step": 116
    },
    {
      "epoch": 0.27432590855803046,
      "grad_norm": 6.58186149597168,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 3.2072,
      "step": 117
    },
    {
      "epoch": 0.27667057444314186,
      "grad_norm": 6.234082221984863,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.1483,
      "step": 118
    },
    {
      "epoch": 0.2790152403282532,
      "grad_norm": 6.033302307128906,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.2893,
      "step": 119
    },
    {
      "epoch": 0.2813599062133646,
      "grad_norm": 6.187047004699707,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.0994,
      "step": 120
    },
    {
      "epoch": 0.283704572098476,
      "grad_norm": 6.225713729858398,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.1627,
      "step": 121
    },
    {
      "epoch": 0.28604923798358733,
      "grad_norm": 7.438437461853027,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.6515,
      "step": 122
    },
    {
      "epoch": 0.28839390386869873,
      "grad_norm": 6.751818656921387,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.3384,
      "step": 123
    },
    {
      "epoch": 0.29073856975381007,
      "grad_norm": 6.50659704208374,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0012,
      "step": 124
    },
    {
      "epoch": 0.29308323563892147,
      "grad_norm": 5.966840744018555,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.8828,
      "step": 125
    },
    {
      "epoch": 0.2954279015240328,
      "grad_norm": 6.445148944854736,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.8949,
      "step": 126
    },
    {
      "epoch": 0.2977725674091442,
      "grad_norm": 5.643779277801514,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6265,
      "step": 127
    },
    {
      "epoch": 0.30011723329425555,
      "grad_norm": 7.064511775970459,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.8426,
      "step": 128
    },
    {
      "epoch": 0.30246189917936694,
      "grad_norm": 6.810649871826172,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 3.3855,
      "step": 129
    },
    {
      "epoch": 0.3048065650644783,
      "grad_norm": 6.587100505828857,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.8196,
      "step": 130
    },
    {
      "epoch": 0.3071512309495897,
      "grad_norm": 6.349030494689941,
      "learning_rate": 2.916090697523549e-05,
      "loss": 3.1042,
      "step": 131
    },
    {
      "epoch": 0.3094958968347011,
      "grad_norm": 6.581236839294434,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 3.157,
      "step": 132
    },
    {
      "epoch": 0.3118405627198124,
      "grad_norm": 6.690001964569092,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.7037,
      "step": 133
    },
    {
      "epoch": 0.3141852286049238,
      "grad_norm": 6.236543655395508,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.4435,
      "step": 134
    },
    {
      "epoch": 0.31652989449003516,
      "grad_norm": 7.735386848449707,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.9265,
      "step": 135
    },
    {
      "epoch": 0.31887456037514655,
      "grad_norm": 6.0608601570129395,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.7175,
      "step": 136
    },
    {
      "epoch": 0.3212192262602579,
      "grad_norm": 6.784418106079102,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.8534,
      "step": 137
    },
    {
      "epoch": 0.3235638921453693,
      "grad_norm": 6.770670413970947,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.7636,
      "step": 138
    },
    {
      "epoch": 0.32590855803048063,
      "grad_norm": 7.257603645324707,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 3.0434,
      "step": 139
    },
    {
      "epoch": 0.32825322391559203,
      "grad_norm": 7.267916679382324,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.7816,
      "step": 140
    },
    {
      "epoch": 0.3305978898007034,
      "grad_norm": 7.654088020324707,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.2745,
      "step": 141
    },
    {
      "epoch": 0.33294255568581477,
      "grad_norm": 7.405394077301025,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.7475,
      "step": 142
    },
    {
      "epoch": 0.33528722157092616,
      "grad_norm": 8.093865394592285,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.7861,
      "step": 143
    },
    {
      "epoch": 0.3376318874560375,
      "grad_norm": 7.341494560241699,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.6269,
      "step": 144
    },
    {
      "epoch": 0.3399765533411489,
      "grad_norm": 6.422662734985352,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.3371,
      "step": 145
    },
    {
      "epoch": 0.34232121922626024,
      "grad_norm": 8.928483009338379,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.8811,
      "step": 146
    },
    {
      "epoch": 0.34466588511137164,
      "grad_norm": 6.97425651550293,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.2997,
      "step": 147
    },
    {
      "epoch": 0.347010550996483,
      "grad_norm": 6.842998504638672,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.3682,
      "step": 148
    },
    {
      "epoch": 0.3493552168815944,
      "grad_norm": 7.386614799499512,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.4438,
      "step": 149
    },
    {
      "epoch": 0.3516998827667057,
      "grad_norm": 8.465286254882812,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.5444,
      "step": 150
    },
    {
      "epoch": 0.3516998827667057,
      "eval_loss": 0.83348149061203,
      "eval_runtime": 53.6368,
      "eval_samples_per_second": 13.386,
      "eval_steps_per_second": 3.356,
      "step": 150
    },
    {
      "epoch": 0.3540445486518171,
      "grad_norm": 13.089689254760742,
      "learning_rate": 1.553232954407171e-05,
      "loss": 4.7924,
      "step": 151
    },
    {
      "epoch": 0.3563892145369285,
      "grad_norm": 12.273059844970703,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 4.2913,
      "step": 152
    },
    {
      "epoch": 0.35873388042203985,
      "grad_norm": 10.967082023620605,
      "learning_rate": 1.435357758543015e-05,
      "loss": 4.0261,
      "step": 153
    },
    {
      "epoch": 0.36107854630715125,
      "grad_norm": 9.32818603515625,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.7357,
      "step": 154
    },
    {
      "epoch": 0.3634232121922626,
      "grad_norm": 8.205735206604004,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 3.591,
      "step": 155
    },
    {
      "epoch": 0.365767878077374,
      "grad_norm": 7.317920207977295,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.7667,
      "step": 156
    },
    {
      "epoch": 0.36811254396248533,
      "grad_norm": 8.465678215026855,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.5215,
      "step": 157
    },
    {
      "epoch": 0.3704572098475967,
      "grad_norm": 6.183009624481201,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.3378,
      "step": 158
    },
    {
      "epoch": 0.37280187573270807,
      "grad_norm": 6.24606990814209,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.4539,
      "step": 159
    },
    {
      "epoch": 0.37514654161781946,
      "grad_norm": 6.455911636352539,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.2931,
      "step": 160
    },
    {
      "epoch": 0.37749120750293086,
      "grad_norm": 6.238929271697998,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.8424,
      "step": 161
    },
    {
      "epoch": 0.3798358733880422,
      "grad_norm": 7.027613162994385,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.6846,
      "step": 162
    },
    {
      "epoch": 0.3821805392731536,
      "grad_norm": 6.125659942626953,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.1225,
      "step": 163
    },
    {
      "epoch": 0.38452520515826494,
      "grad_norm": 7.045245170593262,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.4809,
      "step": 164
    },
    {
      "epoch": 0.38686987104337633,
      "grad_norm": 6.9122209548950195,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.2121,
      "step": 165
    },
    {
      "epoch": 0.3892145369284877,
      "grad_norm": 6.833375453948975,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.3358,
      "step": 166
    },
    {
      "epoch": 0.39155920281359907,
      "grad_norm": 6.130532264709473,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.2976,
      "step": 167
    },
    {
      "epoch": 0.3939038686987104,
      "grad_norm": 6.679653167724609,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.1948,
      "step": 168
    },
    {
      "epoch": 0.3962485345838218,
      "grad_norm": 7.021723747253418,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.4097,
      "step": 169
    },
    {
      "epoch": 0.39859320046893315,
      "grad_norm": 6.679131031036377,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.3376,
      "step": 170
    },
    {
      "epoch": 0.40093786635404455,
      "grad_norm": 6.9932169914245605,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.4275,
      "step": 171
    },
    {
      "epoch": 0.40328253223915594,
      "grad_norm": 5.966531276702881,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.0404,
      "step": 172
    },
    {
      "epoch": 0.4056271981242673,
      "grad_norm": 6.2008867263793945,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.96,
      "step": 173
    },
    {
      "epoch": 0.4079718640093787,
      "grad_norm": 7.096470832824707,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.1843,
      "step": 174
    },
    {
      "epoch": 0.41031652989449,
      "grad_norm": 6.362069129943848,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 3.03,
      "step": 175
    },
    {
      "epoch": 0.4126611957796014,
      "grad_norm": 6.271614074707031,
      "learning_rate": 3.885512251130763e-06,
      "loss": 3.0083,
      "step": 176
    },
    {
      "epoch": 0.41500586166471276,
      "grad_norm": 5.450413703918457,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.5152,
      "step": 177
    },
    {
      "epoch": 0.41735052754982416,
      "grad_norm": 5.825325965881348,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.6462,
      "step": 178
    },
    {
      "epoch": 0.4196951934349355,
      "grad_norm": 6.524065017700195,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.7689,
      "step": 179
    },
    {
      "epoch": 0.4220398593200469,
      "grad_norm": 6.079909801483154,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.8035,
      "step": 180
    },
    {
      "epoch": 0.4243845252051583,
      "grad_norm": 6.251575469970703,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 3.1259,
      "step": 181
    },
    {
      "epoch": 0.42672919109026963,
      "grad_norm": 6.5589752197265625,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.5653,
      "step": 182
    },
    {
      "epoch": 0.42907385697538103,
      "grad_norm": 5.842008113861084,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.6723,
      "step": 183
    },
    {
      "epoch": 0.43141852286049237,
      "grad_norm": 6.21817684173584,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 3.0308,
      "step": 184
    },
    {
      "epoch": 0.43376318874560377,
      "grad_norm": 7.044606685638428,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.7926,
      "step": 185
    },
    {
      "epoch": 0.4361078546307151,
      "grad_norm": 6.023564338684082,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.7825,
      "step": 186
    },
    {
      "epoch": 0.4384525205158265,
      "grad_norm": 6.771507740020752,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.5944,
      "step": 187
    },
    {
      "epoch": 0.44079718640093785,
      "grad_norm": 6.0420823097229,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.5171,
      "step": 188
    },
    {
      "epoch": 0.44314185228604924,
      "grad_norm": 6.126768589019775,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9916,
      "step": 189
    },
    {
      "epoch": 0.4454865181711606,
      "grad_norm": 6.734144687652588,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.0855,
      "step": 190
    },
    {
      "epoch": 0.447831184056272,
      "grad_norm": 6.718818187713623,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.7487,
      "step": 191
    },
    {
      "epoch": 0.4501758499413834,
      "grad_norm": 7.570540428161621,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.0669,
      "step": 192
    },
    {
      "epoch": 0.4525205158264947,
      "grad_norm": 7.8883209228515625,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.0002,
      "step": 193
    },
    {
      "epoch": 0.4548651817116061,
      "grad_norm": 6.0179443359375,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.0846,
      "step": 194
    },
    {
      "epoch": 0.45720984759671746,
      "grad_norm": 6.500763893127441,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.2252,
      "step": 195
    },
    {
      "epoch": 0.45955451348182885,
      "grad_norm": 7.733562469482422,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.5796,
      "step": 196
    },
    {
      "epoch": 0.4618991793669402,
      "grad_norm": 6.960104942321777,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.8474,
      "step": 197
    },
    {
      "epoch": 0.4642438452520516,
      "grad_norm": 8.806565284729004,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.6906,
      "step": 198
    },
    {
      "epoch": 0.46658851113716293,
      "grad_norm": 9.187260627746582,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.5102,
      "step": 199
    },
    {
      "epoch": 0.46893317702227433,
      "grad_norm": 9.199540138244629,
      "learning_rate": 0.0,
      "loss": 2.4954,
      "step": 200
    },
    {
      "epoch": 0.46893317702227433,
      "eval_loss": 0.773449718952179,
      "eval_runtime": 53.6086,
      "eval_samples_per_second": 13.393,
      "eval_steps_per_second": 3.358,
      "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": 2.876309235396772e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}