{
  "best_metric": 10.224454879760742,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7130124777183601,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0035650623885918,
      "grad_norm": 0.4096238911151886,
      "learning_rate": 1e-05,
      "loss": 10.3758,
      "step": 1
    },
    {
      "epoch": 0.0035650623885918,
      "eval_loss": 10.376816749572754,
      "eval_runtime": 1.0347,
      "eval_samples_per_second": 457.123,
      "eval_steps_per_second": 115.005,
      "step": 1
    },
    {
      "epoch": 0.0071301247771836,
      "grad_norm": 0.40438687801361084,
      "learning_rate": 2e-05,
      "loss": 10.3767,
      "step": 2
    },
    {
      "epoch": 0.0106951871657754,
      "grad_norm": 0.4014440178871155,
      "learning_rate": 3e-05,
      "loss": 10.3767,
      "step": 3
    },
    {
      "epoch": 0.0142602495543672,
      "grad_norm": 0.38748297095298767,
      "learning_rate": 4e-05,
      "loss": 10.3742,
      "step": 4
    },
    {
      "epoch": 0.017825311942959002,
      "grad_norm": 0.3970891237258911,
      "learning_rate": 5e-05,
      "loss": 10.373,
      "step": 5
    },
    {
      "epoch": 0.0213903743315508,
      "grad_norm": 0.392198383808136,
      "learning_rate": 6e-05,
      "loss": 10.3758,
      "step": 6
    },
    {
      "epoch": 0.024955436720142603,
      "grad_norm": 0.40189436078071594,
      "learning_rate": 7e-05,
      "loss": 10.3749,
      "step": 7
    },
    {
      "epoch": 0.0285204991087344,
      "grad_norm": 0.39497140049934387,
      "learning_rate": 8e-05,
      "loss": 10.3734,
      "step": 8
    },
    {
      "epoch": 0.03208556149732621,
      "grad_norm": 0.3877145051956177,
      "learning_rate": 9e-05,
      "loss": 10.369,
      "step": 9
    },
    {
      "epoch": 0.035650623885918005,
      "grad_norm": 0.42304643988609314,
      "learning_rate": 0.0001,
      "loss": 10.367,
      "step": 10
    },
    {
      "epoch": 0.0392156862745098,
      "grad_norm": 0.41205286979675293,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.3631,
      "step": 11
    },
    {
      "epoch": 0.0427807486631016,
      "grad_norm": 0.4002147912979126,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.3571,
      "step": 12
    },
    {
      "epoch": 0.04634581105169341,
      "grad_norm": 0.41098731756210327,
      "learning_rate": 9.993849845741524e-05,
      "loss": 10.3615,
      "step": 13
    },
    {
      "epoch": 0.049910873440285206,
      "grad_norm": 0.42324918508529663,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.3605,
      "step": 14
    },
    {
      "epoch": 0.053475935828877004,
      "grad_norm": 0.470659464597702,
      "learning_rate": 9.98292246503335e-05,
      "loss": 10.349,
      "step": 15
    },
    {
      "epoch": 0.0570409982174688,
      "grad_norm": 0.4759865701198578,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.3489,
      "step": 16
    },
    {
      "epoch": 0.06060606060606061,
      "grad_norm": 0.48057541251182556,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.3457,
      "step": 17
    },
    {
      "epoch": 0.06417112299465241,
      "grad_norm": 0.5113264918327332,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.3409,
      "step": 18
    },
    {
      "epoch": 0.0677361853832442,
      "grad_norm": 0.5823430418968201,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.3392,
      "step": 19
    },
    {
      "epoch": 0.07130124777183601,
      "grad_norm": 0.5749707221984863,
      "learning_rate": 9.931806517013612e-05,
      "loss": 10.3347,
      "step": 20
    },
    {
      "epoch": 0.0748663101604278,
      "grad_norm": 0.5930309295654297,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.331,
      "step": 21
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 0.6082383990287781,
      "learning_rate": 9.901899829374047e-05,
      "loss": 10.3285,
      "step": 22
    },
    {
      "epoch": 0.08199643493761141,
      "grad_norm": 0.6586599349975586,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.3196,
      "step": 23
    },
    {
      "epoch": 0.0855614973262032,
      "grad_norm": 0.6275491714477539,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.3154,
      "step": 24
    },
    {
      "epoch": 0.08912655971479501,
      "grad_norm": 0.6618029475212097,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.3076,
      "step": 25
    },
    {
      "epoch": 0.09269162210338681,
      "grad_norm": 0.6224701404571533,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.3068,
      "step": 26
    },
    {
      "epoch": 0.0962566844919786,
      "grad_norm": 0.6258916258811951,
      "learning_rate": 9.803768380684242e-05,
      "loss": 10.305,
      "step": 27
    },
    {
      "epoch": 0.09982174688057041,
      "grad_norm": 0.587232232093811,
      "learning_rate": 9.780178907671789e-05,
      "loss": 10.2989,
      "step": 28
    },
    {
      "epoch": 0.10338680926916222,
      "grad_norm": 0.5430893301963806,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.2914,
      "step": 29
    },
    {
      "epoch": 0.10695187165775401,
      "grad_norm": 0.5052014589309692,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.2911,
      "step": 30
    },
    {
      "epoch": 0.11051693404634581,
      "grad_norm": 0.4604130983352661,
      "learning_rate": 9.701596950580806e-05,
      "loss": 10.2871,
      "step": 31
    },
    {
      "epoch": 0.1140819964349376,
      "grad_norm": 0.4616841971874237,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.281,
      "step": 32
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.4156523048877716,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.2794,
      "step": 33
    },
    {
      "epoch": 0.12121212121212122,
      "grad_norm": 0.37444058060646057,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.2723,
      "step": 34
    },
    {
      "epoch": 0.12477718360071301,
      "grad_norm": 0.3438158631324768,
      "learning_rate": 9.578866633275288e-05,
      "loss": 10.2793,
      "step": 35
    },
    {
      "epoch": 0.12834224598930483,
      "grad_norm": 0.30453529953956604,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.2719,
      "step": 36
    },
    {
      "epoch": 0.1319073083778966,
      "grad_norm": 0.3030957579612732,
      "learning_rate": 9.509956150664796e-05,
      "loss": 10.2726,
      "step": 37
    },
    {
      "epoch": 0.1354723707664884,
      "grad_norm": 0.28578367829322815,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.2729,
      "step": 38
    },
    {
      "epoch": 0.13903743315508021,
      "grad_norm": 0.26348865032196045,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.2706,
      "step": 39
    },
    {
      "epoch": 0.14260249554367202,
      "grad_norm": 0.2681392431259155,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.2685,
      "step": 40
    },
    {
      "epoch": 0.14616755793226383,
      "grad_norm": 0.22420889139175415,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.2673,
      "step": 41
    },
    {
      "epoch": 0.1497326203208556,
      "grad_norm": 0.22650201618671417,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.2656,
      "step": 42
    },
    {
      "epoch": 0.1532976827094474,
      "grad_norm": 0.21033430099487305,
      "learning_rate": 9.273963562927695e-05,
      "loss": 10.2624,
      "step": 43
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 0.1877707690000534,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.2636,
      "step": 44
    },
    {
      "epoch": 0.16042780748663102,
      "grad_norm": 0.16589908301830292,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.2683,
      "step": 45
    },
    {
      "epoch": 0.16399286987522282,
      "grad_norm": 0.1690380573272705,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.2546,
      "step": 46
    },
    {
      "epoch": 0.16755793226381463,
      "grad_norm": 0.13539306819438934,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.2673,
      "step": 47
    },
    {
      "epoch": 0.1711229946524064,
      "grad_norm": 0.17839652299880981,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.2612,
      "step": 48
    },
    {
      "epoch": 0.1746880570409982,
      "grad_norm": 0.16410034894943237,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.2611,
      "step": 49
    },
    {
      "epoch": 0.17825311942959002,
      "grad_norm": 0.16085289418697357,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.2613,
      "step": 50
    },
    {
      "epoch": 0.17825311942959002,
      "eval_loss": 10.259688377380371,
      "eval_runtime": 1.0383,
      "eval_samples_per_second": 455.551,
      "eval_steps_per_second": 114.61,
      "step": 50
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.14522618055343628,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.2608,
      "step": 51
    },
    {
      "epoch": 0.18538324420677363,
      "grad_norm": 0.13988983631134033,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.2584,
      "step": 52
    },
    {
      "epoch": 0.1889483065953654,
      "grad_norm": 0.14809903502464294,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.261,
      "step": 53
    },
    {
      "epoch": 0.1925133689839572,
      "grad_norm": 0.11754299700260162,
      "learning_rate": 8.73410738492077e-05,
      "loss": 10.2579,
      "step": 54
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 0.1315905749797821,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.2609,
      "step": 55
    },
    {
      "epoch": 0.19964349376114082,
      "grad_norm": 0.10722599178552628,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.2576,
      "step": 56
    },
    {
      "epoch": 0.20320855614973263,
      "grad_norm": 0.13075731694698334,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.2577,
      "step": 57
    },
    {
      "epoch": 0.20677361853832443,
      "grad_norm": 0.10768215358257294,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.2616,
      "step": 58
    },
    {
      "epoch": 0.2103386809269162,
      "grad_norm": 0.11184946447610855,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.2594,
      "step": 59
    },
    {
      "epoch": 0.21390374331550802,
      "grad_norm": 0.10637281090021133,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.2569,
      "step": 60
    },
    {
      "epoch": 0.21746880570409982,
      "grad_norm": 0.11345820873975754,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.257,
      "step": 61
    },
    {
      "epoch": 0.22103386809269163,
      "grad_norm": 0.10244770348072052,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.2608,
      "step": 62
    },
    {
      "epoch": 0.22459893048128343,
      "grad_norm": 0.10883080214262009,
      "learning_rate": 8.199842702516583e-05,
      "loss": 10.2544,
      "step": 63
    },
    {
      "epoch": 0.2281639928698752,
      "grad_norm": 0.08807890117168427,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.2572,
      "step": 64
    },
    {
      "epoch": 0.23172905525846701,
      "grad_norm": 0.10900051146745682,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.2554,
      "step": 65
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.11238571256399155,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.2595,
      "step": 66
    },
    {
      "epoch": 0.23885918003565063,
      "grad_norm": 0.09217486530542374,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.2576,
      "step": 67
    },
    {
      "epoch": 0.24242424242424243,
      "grad_norm": 0.10687471926212311,
      "learning_rate": 7.871643313414718e-05,
      "loss": 10.2565,
      "step": 68
    },
    {
      "epoch": 0.24598930481283424,
      "grad_norm": 0.10869158059358597,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.2547,
      "step": 69
    },
    {
      "epoch": 0.24955436720142601,
      "grad_norm": 0.09704132378101349,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.2582,
      "step": 70
    },
    {
      "epoch": 0.2531194295900178,
      "grad_norm": 0.100535549223423,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.26,
      "step": 71
    },
    {
      "epoch": 0.25668449197860965,
      "grad_norm": 0.1033538430929184,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.2566,
      "step": 72
    },
    {
      "epoch": 0.26024955436720143,
      "grad_norm": 0.10201014578342438,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.2498,
      "step": 73
    },
    {
      "epoch": 0.2638146167557932,
      "grad_norm": 0.10933349281549454,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.2543,
      "step": 74
    },
    {
      "epoch": 0.26737967914438504,
      "grad_norm": 0.12258835136890411,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.2532,
      "step": 75
    },
    {
      "epoch": 0.2709447415329768,
      "grad_norm": 0.10582207143306732,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.2525,
      "step": 76
    },
    {
      "epoch": 0.27450980392156865,
      "grad_norm": 0.1255902647972107,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.2535,
      "step": 77
    },
    {
      "epoch": 0.27807486631016043,
      "grad_norm": 0.10968287289142609,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.2556,
      "step": 78
    },
    {
      "epoch": 0.2816399286987522,
      "grad_norm": 0.10734967142343521,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.2489,
      "step": 79
    },
    {
      "epoch": 0.28520499108734404,
      "grad_norm": 0.11152880638837814,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.2515,
      "step": 80
    },
    {
      "epoch": 0.2887700534759358,
      "grad_norm": 0.11925005167722702,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.2474,
      "step": 81
    },
    {
      "epoch": 0.29233511586452765,
      "grad_norm": 0.11574631184339523,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.2483,
      "step": 82
    },
    {
      "epoch": 0.29590017825311943,
      "grad_norm": 0.12162072956562042,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.2523,
      "step": 83
    },
    {
      "epoch": 0.2994652406417112,
      "grad_norm": 0.12150608003139496,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.247,
      "step": 84
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 0.13030201196670532,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.2459,
      "step": 85
    },
    {
      "epoch": 0.3065953654188948,
      "grad_norm": 0.10662204772233963,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.2516,
      "step": 86
    },
    {
      "epoch": 0.31016042780748665,
      "grad_norm": 0.11084286123514175,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.2453,
      "step": 87
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 0.12849649786949158,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.2479,
      "step": 88
    },
    {
      "epoch": 0.3172905525846702,
      "grad_norm": 0.11266584694385529,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.2479,
      "step": 89
    },
    {
      "epoch": 0.32085561497326204,
      "grad_norm": 0.11476273834705353,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.2469,
      "step": 90
    },
    {
      "epoch": 0.3244206773618538,
      "grad_norm": 0.12081189453601837,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.2419,
      "step": 91
    },
    {
      "epoch": 0.32798573975044565,
      "grad_norm": 0.12849433720111847,
      "learning_rate": 6.066498153718735e-05,
      "loss": 10.2435,
      "step": 92
    },
    {
      "epoch": 0.3315508021390374,
      "grad_norm": 0.11772958189249039,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.2464,
      "step": 93
    },
    {
      "epoch": 0.33511586452762926,
      "grad_norm": 0.10807912796735764,
      "learning_rate": 5.90440267166055e-05,
      "loss": 10.2441,
      "step": 94
    },
    {
      "epoch": 0.33868092691622104,
      "grad_norm": 0.10709314048290253,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.244,
      "step": 95
    },
    {
      "epoch": 0.3422459893048128,
      "grad_norm": 0.11932533234357834,
      "learning_rate": 5.74131823855921e-05,
      "loss": 10.2433,
      "step": 96
    },
    {
      "epoch": 0.34581105169340465,
      "grad_norm": 0.13316868245601654,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 10.2412,
      "step": 97
    },
    {
      "epoch": 0.3493761140819964,
      "grad_norm": 0.14504730701446533,
      "learning_rate": 5.577423184847932e-05,
      "loss": 10.2448,
      "step": 98
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.1346348226070404,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.2428,
      "step": 99
    },
    {
      "epoch": 0.35650623885918004,
      "grad_norm": 0.18154430389404297,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.2411,
      "step": 100
    },
    {
      "epoch": 0.35650623885918004,
      "eval_loss": 10.240640640258789,
      "eval_runtime": 1.0336,
      "eval_samples_per_second": 457.605,
      "eval_steps_per_second": 115.127,
      "step": 100
    },
    {
      "epoch": 0.3600713012477718,
      "grad_norm": 0.10482309758663177,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.2441,
      "step": 101
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.10016144067049026,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.2435,
      "step": 102
    },
    {
      "epoch": 0.3672014260249554,
      "grad_norm": 0.10192058980464935,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.242,
      "step": 103
    },
    {
      "epoch": 0.37076648841354726,
      "grad_norm": 0.10803036391735077,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.2376,
      "step": 104
    },
    {
      "epoch": 0.37433155080213903,
      "grad_norm": 0.10726811736822128,
      "learning_rate": 5e-05,
      "loss": 10.2394,
      "step": 105
    },
    {
      "epoch": 0.3778966131907308,
      "grad_norm": 0.11886053532361984,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.2438,
      "step": 106
    },
    {
      "epoch": 0.38146167557932265,
      "grad_norm": 0.10632043331861496,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.2442,
      "step": 107
    },
    {
      "epoch": 0.3850267379679144,
      "grad_norm": 0.09809351712465286,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.2411,
      "step": 108
    },
    {
      "epoch": 0.38859180035650626,
      "grad_norm": 0.10108325630426407,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.241,
      "step": 109
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.09720364958047867,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.2439,
      "step": 110
    },
    {
      "epoch": 0.39572192513368987,
      "grad_norm": 0.09525421261787415,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.2394,
      "step": 111
    },
    {
      "epoch": 0.39928698752228164,
      "grad_norm": 0.09292629361152649,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.2428,
      "step": 112
    },
    {
      "epoch": 0.4028520499108734,
      "grad_norm": 0.0957150086760521,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.2447,
      "step": 113
    },
    {
      "epoch": 0.40641711229946526,
      "grad_norm": 0.09626445174217224,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.237,
      "step": 114
    },
    {
      "epoch": 0.40998217468805703,
      "grad_norm": 0.10777674615383148,
      "learning_rate": 4.17702704859633e-05,
      "loss": 10.2413,
      "step": 115
    },
    {
      "epoch": 0.41354723707664887,
      "grad_norm": 0.10807076096534729,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.2394,
      "step": 116
    },
    {
      "epoch": 0.41711229946524064,
      "grad_norm": 0.09537796676158905,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.2356,
      "step": 117
    },
    {
      "epoch": 0.4206773618538324,
      "grad_norm": 0.08085369318723679,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.2377,
      "step": 118
    },
    {
      "epoch": 0.42424242424242425,
      "grad_norm": 0.09614592790603638,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.2349,
      "step": 119
    },
    {
      "epoch": 0.42780748663101603,
      "grad_norm": 0.10015296936035156,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.231,
      "step": 120
    },
    {
      "epoch": 0.43137254901960786,
      "grad_norm": 0.10398237407207489,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.2316,
      "step": 121
    },
    {
      "epoch": 0.43493761140819964,
      "grad_norm": 0.101392462849617,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.2415,
      "step": 122
    },
    {
      "epoch": 0.4385026737967914,
      "grad_norm": 0.10035762935876846,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.2318,
      "step": 123
    },
    {
      "epoch": 0.44206773618538325,
      "grad_norm": 0.10666773468255997,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.2335,
      "step": 124
    },
    {
      "epoch": 0.44563279857397503,
      "grad_norm": 0.10492396354675293,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.2322,
      "step": 125
    },
    {
      "epoch": 0.44919786096256686,
      "grad_norm": 0.110710009932518,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.2408,
      "step": 126
    },
    {
      "epoch": 0.45276292335115864,
      "grad_norm": 0.09907089918851852,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.2326,
      "step": 127
    },
    {
      "epoch": 0.4563279857397504,
      "grad_norm": 0.11085660010576248,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.2288,
      "step": 128
    },
    {
      "epoch": 0.45989304812834225,
      "grad_norm": 0.09669133275747299,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.2354,
      "step": 129
    },
    {
      "epoch": 0.46345811051693403,
      "grad_norm": 0.10574474930763245,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.2354,
      "step": 130
    },
    {
      "epoch": 0.46702317290552586,
      "grad_norm": 0.10318287461996078,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.2337,
      "step": 131
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.10769189894199371,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 10.2317,
      "step": 132
    },
    {
      "epoch": 0.4741532976827095,
      "grad_norm": 0.10050246119499207,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.2353,
      "step": 133
    },
    {
      "epoch": 0.47771836007130125,
      "grad_norm": 0.09134748578071594,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.2344,
      "step": 134
    },
    {
      "epoch": 0.48128342245989303,
      "grad_norm": 0.12340895086526871,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.2336,
      "step": 135
    },
    {
      "epoch": 0.48484848484848486,
      "grad_norm": 0.12569357454776764,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.2361,
      "step": 136
    },
    {
      "epoch": 0.48841354723707664,
      "grad_norm": 0.12218425422906876,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.2369,
      "step": 137
    },
    {
      "epoch": 0.4919786096256685,
      "grad_norm": 0.11096305400133133,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.2325,
      "step": 138
    },
    {
      "epoch": 0.49554367201426025,
      "grad_norm": 0.10852594673633575,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.2355,
      "step": 139
    },
    {
      "epoch": 0.49910873440285203,
      "grad_norm": 0.1262076050043106,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.2248,
      "step": 140
    },
    {
      "epoch": 0.5026737967914439,
      "grad_norm": 0.11646734178066254,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.233,
      "step": 141
    },
    {
      "epoch": 0.5062388591800356,
      "grad_norm": 0.12405558675527573,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.2348,
      "step": 142
    },
    {
      "epoch": 0.5098039215686274,
      "grad_norm": 0.11200746893882751,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.2285,
      "step": 143
    },
    {
      "epoch": 0.5133689839572193,
      "grad_norm": 0.11780711263418198,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.2294,
      "step": 144
    },
    {
      "epoch": 0.5169340463458111,
      "grad_norm": 0.12805044651031494,
      "learning_rate": 1.928936436551661e-05,
      "loss": 10.2325,
      "step": 145
    },
    {
      "epoch": 0.5204991087344029,
      "grad_norm": 0.12423862516880035,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 10.2306,
      "step": 146
    },
    {
      "epoch": 0.5240641711229946,
      "grad_norm": 0.14587782323360443,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.2335,
      "step": 147
    },
    {
      "epoch": 0.5276292335115864,
      "grad_norm": 0.13238057494163513,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 10.2271,
      "step": 148
    },
    {
      "epoch": 0.5311942959001783,
      "grad_norm": 0.13622386753559113,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.2299,
      "step": 149
    },
    {
      "epoch": 0.5347593582887701,
      "grad_norm": 0.16942785680294037,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 10.2286,
      "step": 150
    },
    {
      "epoch": 0.5347593582887701,
      "eval_loss": 10.227557182312012,
      "eval_runtime": 1.0347,
      "eval_samples_per_second": 457.149,
      "eval_steps_per_second": 115.012,
      "step": 150
    },
    {
      "epoch": 0.5383244206773619,
      "grad_norm": 0.10314422100782394,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.2339,
      "step": 151
    },
    {
      "epoch": 0.5418894830659536,
      "grad_norm": 0.11178354918956757,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.2326,
      "step": 152
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.1087479516863823,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.2322,
      "step": 153
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 0.10929717123508453,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.228,
      "step": 154
    },
    {
      "epoch": 0.5525846702317291,
      "grad_norm": 0.09741606563329697,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.2316,
      "step": 155
    },
    {
      "epoch": 0.5561497326203209,
      "grad_norm": 0.10917097330093384,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 10.2306,
      "step": 156
    },
    {
      "epoch": 0.5597147950089126,
      "grad_norm": 0.10162121802568436,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.2282,
      "step": 157
    },
    {
      "epoch": 0.5632798573975044,
      "grad_norm": 0.10842300951480865,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.2298,
      "step": 158
    },
    {
      "epoch": 0.5668449197860963,
      "grad_norm": 0.1043349876999855,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.2295,
      "step": 159
    },
    {
      "epoch": 0.5704099821746881,
      "grad_norm": 0.10424104332923889,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.2289,
      "step": 160
    },
    {
      "epoch": 0.5739750445632799,
      "grad_norm": 0.08852940797805786,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.2352,
      "step": 161
    },
    {
      "epoch": 0.5775401069518716,
      "grad_norm": 0.10525081306695938,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.2296,
      "step": 162
    },
    {
      "epoch": 0.5811051693404634,
      "grad_norm": 0.09818146377801895,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.2311,
      "step": 163
    },
    {
      "epoch": 0.5846702317290553,
      "grad_norm": 0.10265008360147476,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.2306,
      "step": 164
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.10339802503585815,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.2283,
      "step": 165
    },
    {
      "epoch": 0.5918003565062389,
      "grad_norm": 0.10667331516742706,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.2338,
      "step": 166
    },
    {
      "epoch": 0.5953654188948306,
      "grad_norm": 0.13101749122142792,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.228,
      "step": 167
    },
    {
      "epoch": 0.5989304812834224,
      "grad_norm": 0.0895804613828659,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.233,
      "step": 168
    },
    {
      "epoch": 0.6024955436720143,
      "grad_norm": 0.10891091078519821,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.2269,
      "step": 169
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 0.10129300504922867,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.2291,
      "step": 170
    },
    {
      "epoch": 0.6096256684491979,
      "grad_norm": 0.11627360433340073,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.2261,
      "step": 171
    },
    {
      "epoch": 0.6131907308377896,
      "grad_norm": 0.10390149056911469,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.2345,
      "step": 172
    },
    {
      "epoch": 0.6167557932263814,
      "grad_norm": 0.10863722115755081,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.2228,
      "step": 173
    },
    {
      "epoch": 0.6203208556149733,
      "grad_norm": 0.10054612159729004,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.2294,
      "step": 174
    },
    {
      "epoch": 0.6238859180035651,
      "grad_norm": 0.11975911259651184,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.2323,
      "step": 175
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 0.11578824371099472,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.2271,
      "step": 176
    },
    {
      "epoch": 0.6310160427807486,
      "grad_norm": 0.12156493961811066,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.2253,
      "step": 177
    },
    {
      "epoch": 0.6345811051693404,
      "grad_norm": 0.12006792426109314,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.225,
      "step": 178
    },
    {
      "epoch": 0.6381461675579323,
      "grad_norm": 0.109780453145504,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.2255,
      "step": 179
    },
    {
      "epoch": 0.6417112299465241,
      "grad_norm": 0.11165507882833481,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.2298,
      "step": 180
    },
    {
      "epoch": 0.6452762923351159,
      "grad_norm": 0.10968225449323654,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.2255,
      "step": 181
    },
    {
      "epoch": 0.6488413547237076,
      "grad_norm": 0.1103392243385315,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 10.2258,
      "step": 182
    },
    {
      "epoch": 0.6524064171122995,
      "grad_norm": 0.12329904735088348,
      "learning_rate": 1.962316193157593e-06,
      "loss": 10.2245,
      "step": 183
    },
    {
      "epoch": 0.6559714795008913,
      "grad_norm": 0.10597586631774902,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.223,
      "step": 184
    },
    {
      "epoch": 0.6595365418894831,
      "grad_norm": 0.10912919789552689,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.2304,
      "step": 185
    },
    {
      "epoch": 0.6631016042780749,
      "grad_norm": 0.1188606396317482,
      "learning_rate": 1.333670137599713e-06,
      "loss": 10.2236,
      "step": 186
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.1207171306014061,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.2197,
      "step": 187
    },
    {
      "epoch": 0.6702317290552585,
      "grad_norm": 0.12332808971405029,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.2275,
      "step": 188
    },
    {
      "epoch": 0.6737967914438503,
      "grad_norm": 0.10755743831396103,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.2248,
      "step": 189
    },
    {
      "epoch": 0.6773618538324421,
      "grad_norm": 0.14832819998264313,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.2286,
      "step": 190
    },
    {
      "epoch": 0.6809269162210339,
      "grad_norm": 0.12695495784282684,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.226,
      "step": 191
    },
    {
      "epoch": 0.6844919786096256,
      "grad_norm": 0.15086950361728668,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.2189,
      "step": 192
    },
    {
      "epoch": 0.6880570409982175,
      "grad_norm": 0.11901029199361801,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.2266,
      "step": 193
    },
    {
      "epoch": 0.6916221033868093,
      "grad_norm": 0.14885728061199188,
      "learning_rate": 2.458548727494292e-07,
      "loss": 10.2286,
      "step": 194
    },
    {
      "epoch": 0.6951871657754011,
      "grad_norm": 0.12078335136175156,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.2157,
      "step": 195
    },
    {
      "epoch": 0.6987522281639929,
      "grad_norm": 0.1310369372367859,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 10.2213,
      "step": 196
    },
    {
      "epoch": 0.7023172905525846,
      "grad_norm": 0.13711321353912354,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.2242,
      "step": 197
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.13360749185085297,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.2279,
      "step": 198
    },
    {
      "epoch": 0.7094474153297683,
      "grad_norm": 0.13220982253551483,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.2308,
      "step": 199
    },
    {
      "epoch": 0.7130124777183601,
      "grad_norm": 0.17126797139644623,
      "learning_rate": 0.0,
      "loss": 10.2259,
      "step": 200
    },
    {
      "epoch": 0.7130124777183601,
      "eval_loss": 10.224454879760742,
      "eval_runtime": 1.0332,
      "eval_samples_per_second": 457.791,
      "eval_steps_per_second": 115.174,
      "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": 43412304691200.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}