{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996631862579993,
  "eval_steps": 100,
  "global_step": 1484,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006736274840013472,
      "grad_norm": 3.42309308052063,
      "learning_rate": 1.0067114093959731e-07,
      "loss": 0.4257,
      "step": 1
    },
    {
      "epoch": 0.0013472549680026945,
      "grad_norm": 3.701201915740967,
      "learning_rate": 2.0134228187919462e-07,
      "loss": 0.4285,
      "step": 2
    },
    {
      "epoch": 0.0020208824520040417,
      "grad_norm": 4.045602321624756,
      "learning_rate": 3.0201342281879193e-07,
      "loss": 0.4232,
      "step": 3
    },
    {
      "epoch": 0.002694509936005389,
      "grad_norm": 3.859919786453247,
      "learning_rate": 4.0268456375838924e-07,
      "loss": 0.4029,
      "step": 4
    },
    {
      "epoch": 0.003368137420006736,
      "grad_norm": 4.171447277069092,
      "learning_rate": 5.033557046979866e-07,
      "loss": 0.4158,
      "step": 5
    },
    {
      "epoch": 0.0040417649040080834,
      "grad_norm": 3.556626796722412,
      "learning_rate": 6.040268456375839e-07,
      "loss": 0.3945,
      "step": 6
    },
    {
      "epoch": 0.004715392388009431,
      "grad_norm": 3.78082537651062,
      "learning_rate": 7.046979865771813e-07,
      "loss": 0.372,
      "step": 7
    },
    {
      "epoch": 0.005389019872010778,
      "grad_norm": 3.459005355834961,
      "learning_rate": 8.053691275167785e-07,
      "loss": 0.392,
      "step": 8
    },
    {
      "epoch": 0.006062647356012125,
      "grad_norm": 3.6338694095611572,
      "learning_rate": 9.060402684563759e-07,
      "loss": 0.4221,
      "step": 9
    },
    {
      "epoch": 0.006736274840013472,
      "grad_norm": 3.6951706409454346,
      "learning_rate": 1.006711409395973e-06,
      "loss": 0.4504,
      "step": 10
    },
    {
      "epoch": 0.00740990232401482,
      "grad_norm": 2.708463668823242,
      "learning_rate": 1.1073825503355705e-06,
      "loss": 0.3848,
      "step": 11
    },
    {
      "epoch": 0.008083529808016167,
      "grad_norm": 2.9015040397644043,
      "learning_rate": 1.2080536912751677e-06,
      "loss": 0.3936,
      "step": 12
    },
    {
      "epoch": 0.008757157292017514,
      "grad_norm": 3.133338212966919,
      "learning_rate": 1.3087248322147651e-06,
      "loss": 0.3844,
      "step": 13
    },
    {
      "epoch": 0.009430784776018861,
      "grad_norm": 2.1770880222320557,
      "learning_rate": 1.4093959731543626e-06,
      "loss": 0.3474,
      "step": 14
    },
    {
      "epoch": 0.010104412260020209,
      "grad_norm": 1.66121244430542,
      "learning_rate": 1.5100671140939598e-06,
      "loss": 0.3646,
      "step": 15
    },
    {
      "epoch": 0.010778039744021556,
      "grad_norm": 1.725306510925293,
      "learning_rate": 1.610738255033557e-06,
      "loss": 0.3739,
      "step": 16
    },
    {
      "epoch": 0.011451667228022903,
      "grad_norm": 1.5393097400665283,
      "learning_rate": 1.7114093959731544e-06,
      "loss": 0.2967,
      "step": 17
    },
    {
      "epoch": 0.01212529471202425,
      "grad_norm": 1.6653029918670654,
      "learning_rate": 1.8120805369127518e-06,
      "loss": 0.3491,
      "step": 18
    },
    {
      "epoch": 0.012798922196025598,
      "grad_norm": 1.4329285621643066,
      "learning_rate": 1.912751677852349e-06,
      "loss": 0.3438,
      "step": 19
    },
    {
      "epoch": 0.013472549680026945,
      "grad_norm": 1.1590880155563354,
      "learning_rate": 2.013422818791946e-06,
      "loss": 0.2907,
      "step": 20
    },
    {
      "epoch": 0.014146177164028292,
      "grad_norm": 1.4018336534500122,
      "learning_rate": 2.1140939597315434e-06,
      "loss": 0.3504,
      "step": 21
    },
    {
      "epoch": 0.01481980464802964,
      "grad_norm": 1.201278805732727,
      "learning_rate": 2.214765100671141e-06,
      "loss": 0.3176,
      "step": 22
    },
    {
      "epoch": 0.015493432132030987,
      "grad_norm": 1.14249849319458,
      "learning_rate": 2.3154362416107382e-06,
      "loss": 0.3079,
      "step": 23
    },
    {
      "epoch": 0.016167059616032334,
      "grad_norm": 1.0337632894515991,
      "learning_rate": 2.4161073825503354e-06,
      "loss": 0.3039,
      "step": 24
    },
    {
      "epoch": 0.016840687100033683,
      "grad_norm": 0.9944117665290833,
      "learning_rate": 2.516778523489933e-06,
      "loss": 0.297,
      "step": 25
    },
    {
      "epoch": 0.017514314584035028,
      "grad_norm": 0.946663498878479,
      "learning_rate": 2.6174496644295303e-06,
      "loss": 0.3315,
      "step": 26
    },
    {
      "epoch": 0.018187942068036377,
      "grad_norm": 1.056069254875183,
      "learning_rate": 2.7181208053691275e-06,
      "loss": 0.3274,
      "step": 27
    },
    {
      "epoch": 0.018861569552037723,
      "grad_norm": 0.9784092903137207,
      "learning_rate": 2.818791946308725e-06,
      "loss": 0.3399,
      "step": 28
    },
    {
      "epoch": 0.01953519703603907,
      "grad_norm": 1.07163667678833,
      "learning_rate": 2.9194630872483223e-06,
      "loss": 0.3361,
      "step": 29
    },
    {
      "epoch": 0.020208824520040417,
      "grad_norm": 0.9870592951774597,
      "learning_rate": 3.0201342281879195e-06,
      "loss": 0.3026,
      "step": 30
    },
    {
      "epoch": 0.020882452004041766,
      "grad_norm": 0.9180539846420288,
      "learning_rate": 3.120805369127517e-06,
      "loss": 0.2716,
      "step": 31
    },
    {
      "epoch": 0.02155607948804311,
      "grad_norm": 0.8827613592147827,
      "learning_rate": 3.221476510067114e-06,
      "loss": 0.2623,
      "step": 32
    },
    {
      "epoch": 0.02222970697204446,
      "grad_norm": 0.8390945196151733,
      "learning_rate": 3.3221476510067116e-06,
      "loss": 0.2792,
      "step": 33
    },
    {
      "epoch": 0.022903334456045806,
      "grad_norm": 0.8577262163162231,
      "learning_rate": 3.4228187919463088e-06,
      "loss": 0.2906,
      "step": 34
    },
    {
      "epoch": 0.023576961940047155,
      "grad_norm": 0.7939577102661133,
      "learning_rate": 3.523489932885906e-06,
      "loss": 0.2569,
      "step": 35
    },
    {
      "epoch": 0.0242505894240485,
      "grad_norm": 0.8591914772987366,
      "learning_rate": 3.6241610738255036e-06,
      "loss": 0.3072,
      "step": 36
    },
    {
      "epoch": 0.02492421690804985,
      "grad_norm": 0.8437011241912842,
      "learning_rate": 3.724832214765101e-06,
      "loss": 0.3002,
      "step": 37
    },
    {
      "epoch": 0.025597844392051195,
      "grad_norm": 0.8370192646980286,
      "learning_rate": 3.825503355704698e-06,
      "loss": 0.2693,
      "step": 38
    },
    {
      "epoch": 0.026271471876052544,
      "grad_norm": 0.7814688086509705,
      "learning_rate": 3.926174496644295e-06,
      "loss": 0.2816,
      "step": 39
    },
    {
      "epoch": 0.02694509936005389,
      "grad_norm": 0.8348170518875122,
      "learning_rate": 4.026845637583892e-06,
      "loss": 0.251,
      "step": 40
    },
    {
      "epoch": 0.02761872684405524,
      "grad_norm": 0.7987892627716064,
      "learning_rate": 4.12751677852349e-06,
      "loss": 0.2637,
      "step": 41
    },
    {
      "epoch": 0.028292354328056584,
      "grad_norm": 0.8840119242668152,
      "learning_rate": 4.228187919463087e-06,
      "loss": 0.315,
      "step": 42
    },
    {
      "epoch": 0.028965981812057933,
      "grad_norm": 0.7633718848228455,
      "learning_rate": 4.328859060402685e-06,
      "loss": 0.2988,
      "step": 43
    },
    {
      "epoch": 0.02963960929605928,
      "grad_norm": 0.7476988434791565,
      "learning_rate": 4.429530201342282e-06,
      "loss": 0.2856,
      "step": 44
    },
    {
      "epoch": 0.030313236780060628,
      "grad_norm": 0.7812101244926453,
      "learning_rate": 4.530201342281879e-06,
      "loss": 0.2622,
      "step": 45
    },
    {
      "epoch": 0.030986864264061973,
      "grad_norm": 0.7842202186584473,
      "learning_rate": 4.6308724832214765e-06,
      "loss": 0.3366,
      "step": 46
    },
    {
      "epoch": 0.03166049174806332,
      "grad_norm": 0.7462322115898132,
      "learning_rate": 4.731543624161074e-06,
      "loss": 0.2825,
      "step": 47
    },
    {
      "epoch": 0.03233411923206467,
      "grad_norm": 0.7542632818222046,
      "learning_rate": 4.832214765100671e-06,
      "loss": 0.2549,
      "step": 48
    },
    {
      "epoch": 0.033007746716066017,
      "grad_norm": 0.7200729250907898,
      "learning_rate": 4.932885906040269e-06,
      "loss": 0.2593,
      "step": 49
    },
    {
      "epoch": 0.033681374200067365,
      "grad_norm": 0.8686407208442688,
      "learning_rate": 5.033557046979866e-06,
      "loss": 0.2828,
      "step": 50
    },
    {
      "epoch": 0.03435500168406871,
      "grad_norm": 0.734254002571106,
      "learning_rate": 5.134228187919463e-06,
      "loss": 0.2841,
      "step": 51
    },
    {
      "epoch": 0.035028629168070056,
      "grad_norm": 0.7483602166175842,
      "learning_rate": 5.2348993288590606e-06,
      "loss": 0.2956,
      "step": 52
    },
    {
      "epoch": 0.035702256652071405,
      "grad_norm": 0.7722125053405762,
      "learning_rate": 5.335570469798658e-06,
      "loss": 0.2817,
      "step": 53
    },
    {
      "epoch": 0.036375884136072754,
      "grad_norm": 0.7247833609580994,
      "learning_rate": 5.436241610738255e-06,
      "loss": 0.2589,
      "step": 54
    },
    {
      "epoch": 0.037049511620074096,
      "grad_norm": 0.8258161544799805,
      "learning_rate": 5.536912751677853e-06,
      "loss": 0.2929,
      "step": 55
    },
    {
      "epoch": 0.037723139104075445,
      "grad_norm": 0.784130334854126,
      "learning_rate": 5.63758389261745e-06,
      "loss": 0.2639,
      "step": 56
    },
    {
      "epoch": 0.038396766588076794,
      "grad_norm": 0.8519976735115051,
      "learning_rate": 5.738255033557047e-06,
      "loss": 0.2611,
      "step": 57
    },
    {
      "epoch": 0.03907039407207814,
      "grad_norm": 0.7617088556289673,
      "learning_rate": 5.838926174496645e-06,
      "loss": 0.3038,
      "step": 58
    },
    {
      "epoch": 0.039744021556079485,
      "grad_norm": 0.7174592018127441,
      "learning_rate": 5.939597315436242e-06,
      "loss": 0.2451,
      "step": 59
    },
    {
      "epoch": 0.040417649040080834,
      "grad_norm": 0.7933776378631592,
      "learning_rate": 6.040268456375839e-06,
      "loss": 0.2979,
      "step": 60
    },
    {
      "epoch": 0.04109127652408218,
      "grad_norm": 0.7308351993560791,
      "learning_rate": 6.140939597315437e-06,
      "loss": 0.2547,
      "step": 61
    },
    {
      "epoch": 0.04176490400808353,
      "grad_norm": 0.8782221674919128,
      "learning_rate": 6.241610738255034e-06,
      "loss": 0.2948,
      "step": 62
    },
    {
      "epoch": 0.042438531492084874,
      "grad_norm": 0.7220450043678284,
      "learning_rate": 6.342281879194631e-06,
      "loss": 0.2397,
      "step": 63
    },
    {
      "epoch": 0.04311215897608622,
      "grad_norm": 0.8042862415313721,
      "learning_rate": 6.442953020134228e-06,
      "loss": 0.2963,
      "step": 64
    },
    {
      "epoch": 0.04378578646008757,
      "grad_norm": 0.6996918320655823,
      "learning_rate": 6.543624161073825e-06,
      "loss": 0.2542,
      "step": 65
    },
    {
      "epoch": 0.04445941394408892,
      "grad_norm": 0.7606627941131592,
      "learning_rate": 6.644295302013423e-06,
      "loss": 0.285,
      "step": 66
    },
    {
      "epoch": 0.04513304142809026,
      "grad_norm": 0.8591688275337219,
      "learning_rate": 6.74496644295302e-06,
      "loss": 0.2671,
      "step": 67
    },
    {
      "epoch": 0.04580666891209161,
      "grad_norm": 0.8488709330558777,
      "learning_rate": 6.8456375838926175e-06,
      "loss": 0.2751,
      "step": 68
    },
    {
      "epoch": 0.04648029639609296,
      "grad_norm": 0.7567676305770874,
      "learning_rate": 6.946308724832215e-06,
      "loss": 0.301,
      "step": 69
    },
    {
      "epoch": 0.04715392388009431,
      "grad_norm": 0.7121560573577881,
      "learning_rate": 7.046979865771812e-06,
      "loss": 0.2557,
      "step": 70
    },
    {
      "epoch": 0.04782755136409565,
      "grad_norm": 0.7666682600975037,
      "learning_rate": 7.147651006711409e-06,
      "loss": 0.2582,
      "step": 71
    },
    {
      "epoch": 0.048501178848097,
      "grad_norm": 0.7414038181304932,
      "learning_rate": 7.248322147651007e-06,
      "loss": 0.262,
      "step": 72
    },
    {
      "epoch": 0.04917480633209835,
      "grad_norm": 0.8357811570167542,
      "learning_rate": 7.348993288590604e-06,
      "loss": 0.2591,
      "step": 73
    },
    {
      "epoch": 0.0498484338160997,
      "grad_norm": 0.7933549880981445,
      "learning_rate": 7.449664429530202e-06,
      "loss": 0.282,
      "step": 74
    },
    {
      "epoch": 0.05052206130010104,
      "grad_norm": 0.7420201301574707,
      "learning_rate": 7.5503355704698e-06,
      "loss": 0.2469,
      "step": 75
    },
    {
      "epoch": 0.05119568878410239,
      "grad_norm": 0.7670828104019165,
      "learning_rate": 7.651006711409396e-06,
      "loss": 0.295,
      "step": 76
    },
    {
      "epoch": 0.05186931626810374,
      "grad_norm": 0.722752571105957,
      "learning_rate": 7.751677852348993e-06,
      "loss": 0.2301,
      "step": 77
    },
    {
      "epoch": 0.05254294375210509,
      "grad_norm": 0.7430191040039062,
      "learning_rate": 7.85234899328859e-06,
      "loss": 0.2875,
      "step": 78
    },
    {
      "epoch": 0.05321657123610643,
      "grad_norm": 0.6979767084121704,
      "learning_rate": 7.953020134228188e-06,
      "loss": 0.2326,
      "step": 79
    },
    {
      "epoch": 0.05389019872010778,
      "grad_norm": 0.7197319269180298,
      "learning_rate": 8.053691275167785e-06,
      "loss": 0.2413,
      "step": 80
    },
    {
      "epoch": 0.05456382620410913,
      "grad_norm": 0.7689131498336792,
      "learning_rate": 8.154362416107382e-06,
      "loss": 0.2868,
      "step": 81
    },
    {
      "epoch": 0.05523745368811048,
      "grad_norm": 0.7233304381370544,
      "learning_rate": 8.25503355704698e-06,
      "loss": 0.2586,
      "step": 82
    },
    {
      "epoch": 0.05591108117211182,
      "grad_norm": 0.8464373350143433,
      "learning_rate": 8.355704697986576e-06,
      "loss": 0.2998,
      "step": 83
    },
    {
      "epoch": 0.05658470865611317,
      "grad_norm": 0.8020244240760803,
      "learning_rate": 8.456375838926174e-06,
      "loss": 0.3323,
      "step": 84
    },
    {
      "epoch": 0.05725833614011452,
      "grad_norm": 0.9260913729667664,
      "learning_rate": 8.55704697986577e-06,
      "loss": 0.3353,
      "step": 85
    },
    {
      "epoch": 0.057931963624115866,
      "grad_norm": 0.824252188205719,
      "learning_rate": 8.65771812080537e-06,
      "loss": 0.2778,
      "step": 86
    },
    {
      "epoch": 0.05860559110811721,
      "grad_norm": 0.7277565598487854,
      "learning_rate": 8.758389261744967e-06,
      "loss": 0.2863,
      "step": 87
    },
    {
      "epoch": 0.05927921859211856,
      "grad_norm": 0.7575395107269287,
      "learning_rate": 8.859060402684564e-06,
      "loss": 0.2192,
      "step": 88
    },
    {
      "epoch": 0.059952846076119906,
      "grad_norm": 0.7741091251373291,
      "learning_rate": 8.959731543624161e-06,
      "loss": 0.2808,
      "step": 89
    },
    {
      "epoch": 0.060626473560121255,
      "grad_norm": 0.7291881442070007,
      "learning_rate": 9.060402684563759e-06,
      "loss": 0.2624,
      "step": 90
    },
    {
      "epoch": 0.0613001010441226,
      "grad_norm": 0.7662385106086731,
      "learning_rate": 9.161073825503356e-06,
      "loss": 0.2803,
      "step": 91
    },
    {
      "epoch": 0.061973728528123946,
      "grad_norm": 0.7009522914886475,
      "learning_rate": 9.261744966442953e-06,
      "loss": 0.26,
      "step": 92
    },
    {
      "epoch": 0.06264735601212529,
      "grad_norm": 0.8707520365715027,
      "learning_rate": 9.36241610738255e-06,
      "loss": 0.3179,
      "step": 93
    },
    {
      "epoch": 0.06332098349612664,
      "grad_norm": 0.8629103302955627,
      "learning_rate": 9.463087248322147e-06,
      "loss": 0.3065,
      "step": 94
    },
    {
      "epoch": 0.06399461098012799,
      "grad_norm": 0.8592970371246338,
      "learning_rate": 9.563758389261745e-06,
      "loss": 0.2574,
      "step": 95
    },
    {
      "epoch": 0.06466823846412934,
      "grad_norm": 0.8038861751556396,
      "learning_rate": 9.664429530201342e-06,
      "loss": 0.2699,
      "step": 96
    },
    {
      "epoch": 0.06534186594813068,
      "grad_norm": 0.7168505787849426,
      "learning_rate": 9.765100671140939e-06,
      "loss": 0.2507,
      "step": 97
    },
    {
      "epoch": 0.06601549343213203,
      "grad_norm": 0.7545929551124573,
      "learning_rate": 9.865771812080538e-06,
      "loss": 0.2922,
      "step": 98
    },
    {
      "epoch": 0.06668912091613338,
      "grad_norm": 0.7718814611434937,
      "learning_rate": 9.966442953020135e-06,
      "loss": 0.254,
      "step": 99
    },
    {
      "epoch": 0.06736274840013473,
      "grad_norm": 0.8245450854301453,
      "learning_rate": 1.0067114093959732e-05,
      "loss": 0.2869,
      "step": 100
    },
    {
      "epoch": 0.06736274840013473,
      "eval_loss": 0.2735002040863037,
      "eval_runtime": 104.2064,
      "eval_samples_per_second": 47.982,
      "eval_steps_per_second": 3.004,
      "step": 100
    },
    {
      "epoch": 0.06803637588413607,
      "grad_norm": 0.9367719888687134,
      "learning_rate": 1.016778523489933e-05,
      "loss": 0.3273,
      "step": 101
    },
    {
      "epoch": 0.06871000336813742,
      "grad_norm": 0.7697410583496094,
      "learning_rate": 1.0268456375838927e-05,
      "loss": 0.2493,
      "step": 102
    },
    {
      "epoch": 0.06938363085213876,
      "grad_norm": 0.7449803948402405,
      "learning_rate": 1.0369127516778524e-05,
      "loss": 0.2558,
      "step": 103
    },
    {
      "epoch": 0.07005725833614011,
      "grad_norm": 0.7809726595878601,
      "learning_rate": 1.0469798657718121e-05,
      "loss": 0.3079,
      "step": 104
    },
    {
      "epoch": 0.07073088582014146,
      "grad_norm": 0.8014216423034668,
      "learning_rate": 1.0570469798657718e-05,
      "loss": 0.2774,
      "step": 105
    },
    {
      "epoch": 0.07140451330414281,
      "grad_norm": 0.7782856225967407,
      "learning_rate": 1.0671140939597316e-05,
      "loss": 0.2874,
      "step": 106
    },
    {
      "epoch": 0.07207814078814416,
      "grad_norm": 0.7489345669746399,
      "learning_rate": 1.0771812080536913e-05,
      "loss": 0.2618,
      "step": 107
    },
    {
      "epoch": 0.07275176827214551,
      "grad_norm": 0.7881894111633301,
      "learning_rate": 1.087248322147651e-05,
      "loss": 0.2914,
      "step": 108
    },
    {
      "epoch": 0.07342539575614684,
      "grad_norm": 0.8149965405464172,
      "learning_rate": 1.0973154362416109e-05,
      "loss": 0.2904,
      "step": 109
    },
    {
      "epoch": 0.07409902324014819,
      "grad_norm": 0.8088157176971436,
      "learning_rate": 1.1073825503355706e-05,
      "loss": 0.3084,
      "step": 110
    },
    {
      "epoch": 0.07477265072414954,
      "grad_norm": 0.804861843585968,
      "learning_rate": 1.1174496644295303e-05,
      "loss": 0.2919,
      "step": 111
    },
    {
      "epoch": 0.07544627820815089,
      "grad_norm": 0.7035599946975708,
      "learning_rate": 1.12751677852349e-05,
      "loss": 0.2971,
      "step": 112
    },
    {
      "epoch": 0.07611990569215224,
      "grad_norm": 0.8036991357803345,
      "learning_rate": 1.1375838926174498e-05,
      "loss": 0.2857,
      "step": 113
    },
    {
      "epoch": 0.07679353317615359,
      "grad_norm": 0.6793683767318726,
      "learning_rate": 1.1476510067114095e-05,
      "loss": 0.2742,
      "step": 114
    },
    {
      "epoch": 0.07746716066015494,
      "grad_norm": 0.7865248918533325,
      "learning_rate": 1.1577181208053692e-05,
      "loss": 0.3156,
      "step": 115
    },
    {
      "epoch": 0.07814078814415629,
      "grad_norm": 0.6990460157394409,
      "learning_rate": 1.167785234899329e-05,
      "loss": 0.281,
      "step": 116
    },
    {
      "epoch": 0.07881441562815762,
      "grad_norm": 0.7218809723854065,
      "learning_rate": 1.1778523489932886e-05,
      "loss": 0.2584,
      "step": 117
    },
    {
      "epoch": 0.07948804311215897,
      "grad_norm": 0.6970985531806946,
      "learning_rate": 1.1879194630872484e-05,
      "loss": 0.2438,
      "step": 118
    },
    {
      "epoch": 0.08016167059616032,
      "grad_norm": 0.7687243819236755,
      "learning_rate": 1.1979865771812081e-05,
      "loss": 0.2846,
      "step": 119
    },
    {
      "epoch": 0.08083529808016167,
      "grad_norm": 0.6929764151573181,
      "learning_rate": 1.2080536912751678e-05,
      "loss": 0.2611,
      "step": 120
    },
    {
      "epoch": 0.08150892556416302,
      "grad_norm": 0.729848325252533,
      "learning_rate": 1.2181208053691277e-05,
      "loss": 0.3007,
      "step": 121
    },
    {
      "epoch": 0.08218255304816437,
      "grad_norm": 0.7301985025405884,
      "learning_rate": 1.2281879194630874e-05,
      "loss": 0.2847,
      "step": 122
    },
    {
      "epoch": 0.08285618053216572,
      "grad_norm": 0.7333296537399292,
      "learning_rate": 1.2382550335570471e-05,
      "loss": 0.2674,
      "step": 123
    },
    {
      "epoch": 0.08352980801616706,
      "grad_norm": 0.7411990165710449,
      "learning_rate": 1.2483221476510069e-05,
      "loss": 0.2777,
      "step": 124
    },
    {
      "epoch": 0.08420343550016841,
      "grad_norm": 0.6465498805046082,
      "learning_rate": 1.2583892617449664e-05,
      "loss": 0.2579,
      "step": 125
    },
    {
      "epoch": 0.08487706298416975,
      "grad_norm": 0.6950599551200867,
      "learning_rate": 1.2684563758389261e-05,
      "loss": 0.3164,
      "step": 126
    },
    {
      "epoch": 0.0855506904681711,
      "grad_norm": 0.6696597337722778,
      "learning_rate": 1.2785234899328858e-05,
      "loss": 0.2564,
      "step": 127
    },
    {
      "epoch": 0.08622431795217245,
      "grad_norm": 0.6537868976593018,
      "learning_rate": 1.2885906040268456e-05,
      "loss": 0.2375,
      "step": 128
    },
    {
      "epoch": 0.0868979454361738,
      "grad_norm": 0.7363224029541016,
      "learning_rate": 1.2986577181208053e-05,
      "loss": 0.2589,
      "step": 129
    },
    {
      "epoch": 0.08757157292017514,
      "grad_norm": 0.7354284524917603,
      "learning_rate": 1.308724832214765e-05,
      "loss": 0.3049,
      "step": 130
    },
    {
      "epoch": 0.0882452004041765,
      "grad_norm": 0.6521575450897217,
      "learning_rate": 1.3187919463087247e-05,
      "loss": 0.2385,
      "step": 131
    },
    {
      "epoch": 0.08891882788817784,
      "grad_norm": 0.6530443429946899,
      "learning_rate": 1.3288590604026846e-05,
      "loss": 0.2588,
      "step": 132
    },
    {
      "epoch": 0.08959245537217919,
      "grad_norm": 0.7331404089927673,
      "learning_rate": 1.3389261744966443e-05,
      "loss": 0.3061,
      "step": 133
    },
    {
      "epoch": 0.09026608285618053,
      "grad_norm": 0.7427138090133667,
      "learning_rate": 1.348993288590604e-05,
      "loss": 0.3513,
      "step": 134
    },
    {
      "epoch": 0.09093971034018188,
      "grad_norm": 0.6774203181266785,
      "learning_rate": 1.3590604026845638e-05,
      "loss": 0.2639,
      "step": 135
    },
    {
      "epoch": 0.09161333782418322,
      "grad_norm": 0.6679060459136963,
      "learning_rate": 1.3691275167785235e-05,
      "loss": 0.2503,
      "step": 136
    },
    {
      "epoch": 0.09228696530818457,
      "grad_norm": 0.6390411853790283,
      "learning_rate": 1.3791946308724832e-05,
      "loss": 0.2298,
      "step": 137
    },
    {
      "epoch": 0.09296059279218592,
      "grad_norm": 0.7115532159805298,
      "learning_rate": 1.389261744966443e-05,
      "loss": 0.255,
      "step": 138
    },
    {
      "epoch": 0.09363422027618727,
      "grad_norm": 0.6546367406845093,
      "learning_rate": 1.3993288590604027e-05,
      "loss": 0.2623,
      "step": 139
    },
    {
      "epoch": 0.09430784776018862,
      "grad_norm": 0.7526003122329712,
      "learning_rate": 1.4093959731543624e-05,
      "loss": 0.2701,
      "step": 140
    },
    {
      "epoch": 0.09498147524418997,
      "grad_norm": 0.7417687773704529,
      "learning_rate": 1.4194630872483221e-05,
      "loss": 0.2488,
      "step": 141
    },
    {
      "epoch": 0.0956551027281913,
      "grad_norm": 0.6994727849960327,
      "learning_rate": 1.4295302013422818e-05,
      "loss": 0.2861,
      "step": 142
    },
    {
      "epoch": 0.09632873021219265,
      "grad_norm": 0.7503766417503357,
      "learning_rate": 1.4395973154362415e-05,
      "loss": 0.3002,
      "step": 143
    },
    {
      "epoch": 0.097002357696194,
      "grad_norm": 0.6777353882789612,
      "learning_rate": 1.4496644295302014e-05,
      "loss": 0.2548,
      "step": 144
    },
    {
      "epoch": 0.09767598518019535,
      "grad_norm": 0.8131176829338074,
      "learning_rate": 1.4597315436241612e-05,
      "loss": 0.2736,
      "step": 145
    },
    {
      "epoch": 0.0983496126641967,
      "grad_norm": 0.6841787099838257,
      "learning_rate": 1.4697986577181209e-05,
      "loss": 0.2647,
      "step": 146
    },
    {
      "epoch": 0.09902324014819805,
      "grad_norm": 0.673572838306427,
      "learning_rate": 1.4798657718120806e-05,
      "loss": 0.2414,
      "step": 147
    },
    {
      "epoch": 0.0996968676321994,
      "grad_norm": 0.6950225234031677,
      "learning_rate": 1.4899328859060403e-05,
      "loss": 0.268,
      "step": 148
    },
    {
      "epoch": 0.10037049511620075,
      "grad_norm": 0.7058023810386658,
      "learning_rate": 1.5e-05,
      "loss": 0.2682,
      "step": 149
    },
    {
      "epoch": 0.10104412260020208,
      "grad_norm": 0.7642398476600647,
      "learning_rate": 1.4999979233262118e-05,
      "loss": 0.2871,
      "step": 150
    },
    {
      "epoch": 0.10171775008420343,
      "grad_norm": 0.7045179605484009,
      "learning_rate": 1.4999916933163468e-05,
      "loss": 0.2589,
      "step": 151
    },
    {
      "epoch": 0.10239137756820478,
      "grad_norm": 0.6908326148986816,
      "learning_rate": 1.499981310004906e-05,
      "loss": 0.2727,
      "step": 152
    },
    {
      "epoch": 0.10306500505220613,
      "grad_norm": 0.7265616655349731,
      "learning_rate": 1.4999667734493901e-05,
      "loss": 0.3177,
      "step": 153
    },
    {
      "epoch": 0.10373863253620748,
      "grad_norm": 0.630407452583313,
      "learning_rate": 1.4999480837302995e-05,
      "loss": 0.2636,
      "step": 154
    },
    {
      "epoch": 0.10441226002020883,
      "grad_norm": 0.6864127516746521,
      "learning_rate": 1.4999252409511335e-05,
      "loss": 0.3013,
      "step": 155
    },
    {
      "epoch": 0.10508588750421018,
      "grad_norm": 0.7556886076927185,
      "learning_rate": 1.4998982452383916e-05,
      "loss": 0.279,
      "step": 156
    },
    {
      "epoch": 0.10575951498821153,
      "grad_norm": 0.7267988324165344,
      "learning_rate": 1.4998670967415701e-05,
      "loss": 0.2528,
      "step": 157
    },
    {
      "epoch": 0.10643314247221286,
      "grad_norm": 0.6894652843475342,
      "learning_rate": 1.4998317956331634e-05,
      "loss": 0.2833,
      "step": 158
    },
    {
      "epoch": 0.10710676995621421,
      "grad_norm": 0.7065450549125671,
      "learning_rate": 1.4997923421086613e-05,
      "loss": 0.3159,
      "step": 159
    },
    {
      "epoch": 0.10778039744021556,
      "grad_norm": 0.6692951321601868,
      "learning_rate": 1.49974873638655e-05,
      "loss": 0.2747,
      "step": 160
    },
    {
      "epoch": 0.10845402492421691,
      "grad_norm": 0.589299738407135,
      "learning_rate": 1.4997009787083088e-05,
      "loss": 0.2436,
      "step": 161
    },
    {
      "epoch": 0.10912765240821826,
      "grad_norm": 0.6986613869667053,
      "learning_rate": 1.49964906933841e-05,
      "loss": 0.2893,
      "step": 162
    },
    {
      "epoch": 0.1098012798922196,
      "grad_norm": 0.6756588220596313,
      "learning_rate": 1.4995930085643173e-05,
      "loss": 0.3076,
      "step": 163
    },
    {
      "epoch": 0.11047490737622095,
      "grad_norm": 0.6988603472709656,
      "learning_rate": 1.4995327966964838e-05,
      "loss": 0.2646,
      "step": 164
    },
    {
      "epoch": 0.1111485348602223,
      "grad_norm": 0.6961201429367065,
      "learning_rate": 1.4994684340683506e-05,
      "loss": 0.2984,
      "step": 165
    },
    {
      "epoch": 0.11182216234422364,
      "grad_norm": 0.7064459323883057,
      "learning_rate": 1.4993999210363444e-05,
      "loss": 0.3186,
      "step": 166
    },
    {
      "epoch": 0.11249578982822499,
      "grad_norm": 0.6374897360801697,
      "learning_rate": 1.4993272579798773e-05,
      "loss": 0.2833,
      "step": 167
    },
    {
      "epoch": 0.11316941731222634,
      "grad_norm": 0.6672942638397217,
      "learning_rate": 1.4992504453013422e-05,
      "loss": 0.2891,
      "step": 168
    },
    {
      "epoch": 0.11384304479622769,
      "grad_norm": 0.6631248593330383,
      "learning_rate": 1.499169483426112e-05,
      "loss": 0.2512,
      "step": 169
    },
    {
      "epoch": 0.11451667228022903,
      "grad_norm": 0.7132297158241272,
      "learning_rate": 1.4990843728025367e-05,
      "loss": 0.2988,
      "step": 170
    },
    {
      "epoch": 0.11519029976423038,
      "grad_norm": 0.6612878441810608,
      "learning_rate": 1.4989951139019425e-05,
      "loss": 0.283,
      "step": 171
    },
    {
      "epoch": 0.11586392724823173,
      "grad_norm": 0.6382921934127808,
      "learning_rate": 1.4989017072186267e-05,
      "loss": 0.2597,
      "step": 172
    },
    {
      "epoch": 0.11653755473223308,
      "grad_norm": 0.5888513922691345,
      "learning_rate": 1.498804153269856e-05,
      "loss": 0.243,
      "step": 173
    },
    {
      "epoch": 0.11721118221623442,
      "grad_norm": 0.7310932874679565,
      "learning_rate": 1.498702452595865e-05,
      "loss": 0.2871,
      "step": 174
    },
    {
      "epoch": 0.11788480970023577,
      "grad_norm": 0.6769680380821228,
      "learning_rate": 1.4985966057598512e-05,
      "loss": 0.2896,
      "step": 175
    },
    {
      "epoch": 0.11855843718423711,
      "grad_norm": 0.7013587355613708,
      "learning_rate": 1.4984866133479729e-05,
      "loss": 0.2913,
      "step": 176
    },
    {
      "epoch": 0.11923206466823846,
      "grad_norm": 0.7067077159881592,
      "learning_rate": 1.4983724759693456e-05,
      "loss": 0.2931,
      "step": 177
    },
    {
      "epoch": 0.11990569215223981,
      "grad_norm": 0.6384806632995605,
      "learning_rate": 1.498254194256039e-05,
      "loss": 0.2433,
      "step": 178
    },
    {
      "epoch": 0.12057931963624116,
      "grad_norm": 0.733525276184082,
      "learning_rate": 1.4981317688630729e-05,
      "loss": 0.314,
      "step": 179
    },
    {
      "epoch": 0.12125294712024251,
      "grad_norm": 0.6598628759384155,
      "learning_rate": 1.4980052004684146e-05,
      "loss": 0.281,
      "step": 180
    },
    {
      "epoch": 0.12192657460424386,
      "grad_norm": 0.616263210773468,
      "learning_rate": 1.4978744897729741e-05,
      "loss": 0.2616,
      "step": 181
    },
    {
      "epoch": 0.1226002020882452,
      "grad_norm": 0.6175768971443176,
      "learning_rate": 1.4977396375006006e-05,
      "loss": 0.2624,
      "step": 182
    },
    {
      "epoch": 0.12327382957224654,
      "grad_norm": 0.676030695438385,
      "learning_rate": 1.4976006443980785e-05,
      "loss": 0.287,
      "step": 183
    },
    {
      "epoch": 0.12394745705624789,
      "grad_norm": 0.6331183314323425,
      "learning_rate": 1.4974575112351235e-05,
      "loss": 0.2647,
      "step": 184
    },
    {
      "epoch": 0.12462108454024924,
      "grad_norm": 0.656204104423523,
      "learning_rate": 1.497310238804378e-05,
      "loss": 0.2755,
      "step": 185
    },
    {
      "epoch": 0.12529471202425058,
      "grad_norm": 0.6582143306732178,
      "learning_rate": 1.4971588279214065e-05,
      "loss": 0.2774,
      "step": 186
    },
    {
      "epoch": 0.12596833950825193,
      "grad_norm": 0.6152216792106628,
      "learning_rate": 1.4970032794246918e-05,
      "loss": 0.2694,
      "step": 187
    },
    {
      "epoch": 0.12664196699225327,
      "grad_norm": 0.5943458676338196,
      "learning_rate": 1.4968435941756303e-05,
      "loss": 0.2698,
      "step": 188
    },
    {
      "epoch": 0.12731559447625462,
      "grad_norm": 0.7527596354484558,
      "learning_rate": 1.496679773058526e-05,
      "loss": 0.2996,
      "step": 189
    },
    {
      "epoch": 0.12798922196025597,
      "grad_norm": 0.6229069828987122,
      "learning_rate": 1.4965118169805868e-05,
      "loss": 0.275,
      "step": 190
    },
    {
      "epoch": 0.12866284944425732,
      "grad_norm": 0.620919406414032,
      "learning_rate": 1.4963397268719198e-05,
      "loss": 0.2956,
      "step": 191
    },
    {
      "epoch": 0.12933647692825867,
      "grad_norm": 0.6090366244316101,
      "learning_rate": 1.4961635036855249e-05,
      "loss": 0.258,
      "step": 192
    },
    {
      "epoch": 0.13001010441226002,
      "grad_norm": 0.5942346453666687,
      "learning_rate": 1.4959831483972901e-05,
      "loss": 0.266,
      "step": 193
    },
    {
      "epoch": 0.13068373189626137,
      "grad_norm": 0.6019350290298462,
      "learning_rate": 1.4957986620059866e-05,
      "loss": 0.256,
      "step": 194
    },
    {
      "epoch": 0.13135735938026272,
      "grad_norm": 0.6708882451057434,
      "learning_rate": 1.4956100455332623e-05,
      "loss": 0.2924,
      "step": 195
    },
    {
      "epoch": 0.13203098686426407,
      "grad_norm": 0.7132793068885803,
      "learning_rate": 1.4954173000236369e-05,
      "loss": 0.3174,
      "step": 196
    },
    {
      "epoch": 0.13270461434826542,
      "grad_norm": 0.602311909198761,
      "learning_rate": 1.495220426544496e-05,
      "loss": 0.2388,
      "step": 197
    },
    {
      "epoch": 0.13337824183226676,
      "grad_norm": 0.5862560868263245,
      "learning_rate": 1.495019426186085e-05,
      "loss": 0.2382,
      "step": 198
    },
    {
      "epoch": 0.1340518693162681,
      "grad_norm": 0.6618714332580566,
      "learning_rate": 1.4948143000615028e-05,
      "loss": 0.2654,
      "step": 199
    },
    {
      "epoch": 0.13472549680026946,
      "grad_norm": 0.6195774078369141,
      "learning_rate": 1.4946050493066965e-05,
      "loss": 0.2696,
      "step": 200
    },
    {
      "epoch": 0.13472549680026946,
      "eval_loss": 0.2768155038356781,
      "eval_runtime": 105.0569,
      "eval_samples_per_second": 47.593,
      "eval_steps_per_second": 2.979,
      "step": 200
    },
    {
      "epoch": 0.1353991242842708,
      "grad_norm": 0.5954621434211731,
      "learning_rate": 1.4943916750804537e-05,
      "loss": 0.2625,
      "step": 201
    },
    {
      "epoch": 0.13607275176827213,
      "grad_norm": 0.610717236995697,
      "learning_rate": 1.494174178564398e-05,
      "loss": 0.2953,
      "step": 202
    },
    {
      "epoch": 0.13674637925227348,
      "grad_norm": 0.6930943727493286,
      "learning_rate": 1.4939525609629809e-05,
      "loss": 0.2774,
      "step": 203
    },
    {
      "epoch": 0.13742000673627483,
      "grad_norm": 0.6402983069419861,
      "learning_rate": 1.4937268235034754e-05,
      "loss": 0.2814,
      "step": 204
    },
    {
      "epoch": 0.13809363422027618,
      "grad_norm": 0.6476616859436035,
      "learning_rate": 1.4934969674359698e-05,
      "loss": 0.2829,
      "step": 205
    },
    {
      "epoch": 0.13876726170427753,
      "grad_norm": 0.6163775324821472,
      "learning_rate": 1.49326299403336e-05,
      "loss": 0.2682,
      "step": 206
    },
    {
      "epoch": 0.13944088918827888,
      "grad_norm": 0.6615155935287476,
      "learning_rate": 1.4930249045913437e-05,
      "loss": 0.2656,
      "step": 207
    },
    {
      "epoch": 0.14011451667228023,
      "grad_norm": 0.6666435599327087,
      "learning_rate": 1.4927827004284117e-05,
      "loss": 0.2972,
      "step": 208
    },
    {
      "epoch": 0.14078814415628157,
      "grad_norm": 0.6047382950782776,
      "learning_rate": 1.4925363828858407e-05,
      "loss": 0.2527,
      "step": 209
    },
    {
      "epoch": 0.14146177164028292,
      "grad_norm": 0.6405648589134216,
      "learning_rate": 1.4922859533276882e-05,
      "loss": 0.2589,
      "step": 210
    },
    {
      "epoch": 0.14213539912428427,
      "grad_norm": 0.6201145648956299,
      "learning_rate": 1.4920314131407817e-05,
      "loss": 0.2419,
      "step": 211
    },
    {
      "epoch": 0.14280902660828562,
      "grad_norm": 0.6683364510536194,
      "learning_rate": 1.4917727637347132e-05,
      "loss": 0.2973,
      "step": 212
    },
    {
      "epoch": 0.14348265409228697,
      "grad_norm": 0.5999878644943237,
      "learning_rate": 1.4915100065418302e-05,
      "loss": 0.2714,
      "step": 213
    },
    {
      "epoch": 0.14415628157628832,
      "grad_norm": 0.6046174764633179,
      "learning_rate": 1.491243143017229e-05,
      "loss": 0.2841,
      "step": 214
    },
    {
      "epoch": 0.14482990906028967,
      "grad_norm": 0.6034740209579468,
      "learning_rate": 1.4909721746387454e-05,
      "loss": 0.2896,
      "step": 215
    },
    {
      "epoch": 0.14550353654429102,
      "grad_norm": 0.6835145354270935,
      "learning_rate": 1.4906971029069473e-05,
      "loss": 0.2778,
      "step": 216
    },
    {
      "epoch": 0.14617716402829237,
      "grad_norm": 0.6769616603851318,
      "learning_rate": 1.490417929345126e-05,
      "loss": 0.2697,
      "step": 217
    },
    {
      "epoch": 0.1468507915122937,
      "grad_norm": 0.6558434367179871,
      "learning_rate": 1.4901346554992879e-05,
      "loss": 0.2708,
      "step": 218
    },
    {
      "epoch": 0.14752441899629504,
      "grad_norm": 0.6363021731376648,
      "learning_rate": 1.489847282938146e-05,
      "loss": 0.297,
      "step": 219
    },
    {
      "epoch": 0.14819804648029639,
      "grad_norm": 0.6437724828720093,
      "learning_rate": 1.4895558132531112e-05,
      "loss": 0.2827,
      "step": 220
    },
    {
      "epoch": 0.14887167396429773,
      "grad_norm": 0.6295124292373657,
      "learning_rate": 1.4892602480582836e-05,
      "loss": 0.2998,
      "step": 221
    },
    {
      "epoch": 0.14954530144829908,
      "grad_norm": 0.634768545627594,
      "learning_rate": 1.4889605889904426e-05,
      "loss": 0.2686,
      "step": 222
    },
    {
      "epoch": 0.15021892893230043,
      "grad_norm": 0.624239981174469,
      "learning_rate": 1.4886568377090396e-05,
      "loss": 0.3161,
      "step": 223
    },
    {
      "epoch": 0.15089255641630178,
      "grad_norm": 0.6285136342048645,
      "learning_rate": 1.4883489958961875e-05,
      "loss": 0.3089,
      "step": 224
    },
    {
      "epoch": 0.15156618390030313,
      "grad_norm": 0.6140178442001343,
      "learning_rate": 1.4880370652566516e-05,
      "loss": 0.2888,
      "step": 225
    },
    {
      "epoch": 0.15223981138430448,
      "grad_norm": 0.5987722873687744,
      "learning_rate": 1.4877210475178403e-05,
      "loss": 0.2586,
      "step": 226
    },
    {
      "epoch": 0.15291343886830583,
      "grad_norm": 0.6315680146217346,
      "learning_rate": 1.487400944429796e-05,
      "loss": 0.2876,
      "step": 227
    },
    {
      "epoch": 0.15358706635230718,
      "grad_norm": 0.6932382583618164,
      "learning_rate": 1.487076757765184e-05,
      "loss": 0.2886,
      "step": 228
    },
    {
      "epoch": 0.15426069383630853,
      "grad_norm": 0.5736963748931885,
      "learning_rate": 1.4867484893192847e-05,
      "loss": 0.2524,
      "step": 229
    },
    {
      "epoch": 0.15493432132030988,
      "grad_norm": 0.6102257370948792,
      "learning_rate": 1.4864161409099814e-05,
      "loss": 0.2518,
      "step": 230
    },
    {
      "epoch": 0.15560794880431122,
      "grad_norm": 0.5340930819511414,
      "learning_rate": 1.4860797143777526e-05,
      "loss": 0.2466,
      "step": 231
    },
    {
      "epoch": 0.15628157628831257,
      "grad_norm": 0.6170995831489563,
      "learning_rate": 1.4857392115856597e-05,
      "loss": 0.2588,
      "step": 232
    },
    {
      "epoch": 0.15695520377231392,
      "grad_norm": 0.5439332127571106,
      "learning_rate": 1.4853946344193386e-05,
      "loss": 0.2377,
      "step": 233
    },
    {
      "epoch": 0.15762883125631524,
      "grad_norm": 0.6084430813789368,
      "learning_rate": 1.4850459847869866e-05,
      "loss": 0.2514,
      "step": 234
    },
    {
      "epoch": 0.1583024587403166,
      "grad_norm": 0.6239585280418396,
      "learning_rate": 1.4846932646193554e-05,
      "loss": 0.2892,
      "step": 235
    },
    {
      "epoch": 0.15897608622431794,
      "grad_norm": 0.6361899375915527,
      "learning_rate": 1.4843364758697371e-05,
      "loss": 0.264,
      "step": 236
    },
    {
      "epoch": 0.1596497137083193,
      "grad_norm": 0.5994705557823181,
      "learning_rate": 1.4839756205139555e-05,
      "loss": 0.2756,
      "step": 237
    },
    {
      "epoch": 0.16032334119232064,
      "grad_norm": 0.6532281041145325,
      "learning_rate": 1.4836107005503543e-05,
      "loss": 0.3262,
      "step": 238
    },
    {
      "epoch": 0.160996968676322,
      "grad_norm": 0.6311124563217163,
      "learning_rate": 1.483241717999786e-05,
      "loss": 0.3137,
      "step": 239
    },
    {
      "epoch": 0.16167059616032334,
      "grad_norm": 0.5731788873672485,
      "learning_rate": 1.4828686749056007e-05,
      "loss": 0.2476,
      "step": 240
    },
    {
      "epoch": 0.1623442236443247,
      "grad_norm": 0.5689460039138794,
      "learning_rate": 1.4824915733336355e-05,
      "loss": 0.2717,
      "step": 241
    },
    {
      "epoch": 0.16301785112832604,
      "grad_norm": 0.6340669989585876,
      "learning_rate": 1.4821104153722023e-05,
      "loss": 0.2756,
      "step": 242
    },
    {
      "epoch": 0.16369147861232738,
      "grad_norm": 0.6497682929039001,
      "learning_rate": 1.4817252031320766e-05,
      "loss": 0.3197,
      "step": 243
    },
    {
      "epoch": 0.16436510609632873,
      "grad_norm": 0.6404630541801453,
      "learning_rate": 1.481335938746485e-05,
      "loss": 0.2641,
      "step": 244
    },
    {
      "epoch": 0.16503873358033008,
      "grad_norm": 0.5862687230110168,
      "learning_rate": 1.480942624371095e-05,
      "loss": 0.261,
      "step": 245
    },
    {
      "epoch": 0.16571236106433143,
      "grad_norm": 0.6154356598854065,
      "learning_rate": 1.4805452621840015e-05,
      "loss": 0.2856,
      "step": 246
    },
    {
      "epoch": 0.16638598854833278,
      "grad_norm": 0.7411592602729797,
      "learning_rate": 1.4801438543857154e-05,
      "loss": 0.2838,
      "step": 247
    },
    {
      "epoch": 0.16705961603233413,
      "grad_norm": 0.6304882764816284,
      "learning_rate": 1.479738403199152e-05,
      "loss": 0.3102,
      "step": 248
    },
    {
      "epoch": 0.16773324351633548,
      "grad_norm": 0.5838252305984497,
      "learning_rate": 1.479328910869617e-05,
      "loss": 0.3074,
      "step": 249
    },
    {
      "epoch": 0.16840687100033683,
      "grad_norm": 0.6592857241630554,
      "learning_rate": 1.4789153796647957e-05,
      "loss": 0.2482,
      "step": 250
    },
    {
      "epoch": 0.16908049848433815,
      "grad_norm": 0.6678220629692078,
      "learning_rate": 1.4784978118747404e-05,
      "loss": 0.2858,
      "step": 251
    },
    {
      "epoch": 0.1697541259683395,
      "grad_norm": 0.7072235345840454,
      "learning_rate": 1.4780762098118564e-05,
      "loss": 0.317,
      "step": 252
    },
    {
      "epoch": 0.17042775345234085,
      "grad_norm": 0.6481045484542847,
      "learning_rate": 1.4776505758108901e-05,
      "loss": 0.3074,
      "step": 253
    },
    {
      "epoch": 0.1711013809363422,
      "grad_norm": 0.573128342628479,
      "learning_rate": 1.477220912228916e-05,
      "loss": 0.2421,
      "step": 254
    },
    {
      "epoch": 0.17177500842034354,
      "grad_norm": 0.5758487582206726,
      "learning_rate": 1.4767872214453241e-05,
      "loss": 0.2874,
      "step": 255
    },
    {
      "epoch": 0.1724486359043449,
      "grad_norm": 0.5688092112541199,
      "learning_rate": 1.4763495058618056e-05,
      "loss": 0.2897,
      "step": 256
    },
    {
      "epoch": 0.17312226338834624,
      "grad_norm": 0.607288658618927,
      "learning_rate": 1.4759077679023406e-05,
      "loss": 0.2707,
      "step": 257
    },
    {
      "epoch": 0.1737958908723476,
      "grad_norm": 0.6363064646720886,
      "learning_rate": 1.4754620100131838e-05,
      "loss": 0.2977,
      "step": 258
    },
    {
      "epoch": 0.17446951835634894,
      "grad_norm": 0.6312716007232666,
      "learning_rate": 1.475012234662852e-05,
      "loss": 0.2794,
      "step": 259
    },
    {
      "epoch": 0.1751431458403503,
      "grad_norm": 0.6589624285697937,
      "learning_rate": 1.4745584443421097e-05,
      "loss": 0.3483,
      "step": 260
    },
    {
      "epoch": 0.17581677332435164,
      "grad_norm": 0.5797691345214844,
      "learning_rate": 1.4741006415639555e-05,
      "loss": 0.3013,
      "step": 261
    },
    {
      "epoch": 0.176490400808353,
      "grad_norm": 0.5717487335205078,
      "learning_rate": 1.473638828863608e-05,
      "loss": 0.2725,
      "step": 262
    },
    {
      "epoch": 0.17716402829235434,
      "grad_norm": 0.6161592602729797,
      "learning_rate": 1.4731730087984924e-05,
      "loss": 0.3049,
      "step": 263
    },
    {
      "epoch": 0.17783765577635569,
      "grad_norm": 0.6334370970726013,
      "learning_rate": 1.4727031839482251e-05,
      "loss": 0.2844,
      "step": 264
    },
    {
      "epoch": 0.17851128326035703,
      "grad_norm": 0.576859176158905,
      "learning_rate": 1.472229356914601e-05,
      "loss": 0.244,
      "step": 265
    },
    {
      "epoch": 0.17918491074435838,
      "grad_norm": 0.6241918802261353,
      "learning_rate": 1.4717515303215776e-05,
      "loss": 0.2838,
      "step": 266
    },
    {
      "epoch": 0.1798585382283597,
      "grad_norm": 0.5989061594009399,
      "learning_rate": 1.4712697068152619e-05,
      "loss": 0.2984,
      "step": 267
    },
    {
      "epoch": 0.18053216571236105,
      "grad_norm": 0.5685368180274963,
      "learning_rate": 1.4707838890638941e-05,
      "loss": 0.2787,
      "step": 268
    },
    {
      "epoch": 0.1812057931963624,
      "grad_norm": 0.6349403262138367,
      "learning_rate": 1.4702940797578345e-05,
      "loss": 0.3078,
      "step": 269
    },
    {
      "epoch": 0.18187942068036375,
      "grad_norm": 0.6529637575149536,
      "learning_rate": 1.4698002816095473e-05,
      "loss": 0.307,
      "step": 270
    },
    {
      "epoch": 0.1825530481643651,
      "grad_norm": 0.5679558515548706,
      "learning_rate": 1.4693024973535863e-05,
      "loss": 0.25,
      "step": 271
    },
    {
      "epoch": 0.18322667564836645,
      "grad_norm": 0.5999310612678528,
      "learning_rate": 1.4688007297465796e-05,
      "loss": 0.259,
      "step": 272
    },
    {
      "epoch": 0.1839003031323678,
      "grad_norm": 0.6034629344940186,
      "learning_rate": 1.4682949815672146e-05,
      "loss": 0.3071,
      "step": 273
    },
    {
      "epoch": 0.18457393061636915,
      "grad_norm": 0.610670268535614,
      "learning_rate": 1.467785255616221e-05,
      "loss": 0.2913,
      "step": 274
    },
    {
      "epoch": 0.1852475581003705,
      "grad_norm": 0.628016471862793,
      "learning_rate": 1.4672715547163584e-05,
      "loss": 0.2839,
      "step": 275
    },
    {
      "epoch": 0.18592118558437185,
      "grad_norm": 0.6297721862792969,
      "learning_rate": 1.4667538817123977e-05,
      "loss": 0.3403,
      "step": 276
    },
    {
      "epoch": 0.1865948130683732,
      "grad_norm": 0.540552020072937,
      "learning_rate": 1.4662322394711067e-05,
      "loss": 0.2454,
      "step": 277
    },
    {
      "epoch": 0.18726844055237454,
      "grad_norm": 0.513788640499115,
      "learning_rate": 1.4657066308812342e-05,
      "loss": 0.233,
      "step": 278
    },
    {
      "epoch": 0.1879420680363759,
      "grad_norm": 0.6221415996551514,
      "learning_rate": 1.4651770588534937e-05,
      "loss": 0.2969,
      "step": 279
    },
    {
      "epoch": 0.18861569552037724,
      "grad_norm": 0.5859697461128235,
      "learning_rate": 1.4646435263205475e-05,
      "loss": 0.2771,
      "step": 280
    },
    {
      "epoch": 0.1892893230043786,
      "grad_norm": 0.5720670819282532,
      "learning_rate": 1.4641060362369904e-05,
      "loss": 0.2758,
      "step": 281
    },
    {
      "epoch": 0.18996295048837994,
      "grad_norm": 0.5609393119812012,
      "learning_rate": 1.4635645915793333e-05,
      "loss": 0.256,
      "step": 282
    },
    {
      "epoch": 0.19063657797238126,
      "grad_norm": 0.5734854340553284,
      "learning_rate": 1.4630191953459862e-05,
      "loss": 0.3233,
      "step": 283
    },
    {
      "epoch": 0.1913102054563826,
      "grad_norm": 0.570590615272522,
      "learning_rate": 1.4624698505572432e-05,
      "loss": 0.2757,
      "step": 284
    },
    {
      "epoch": 0.19198383294038396,
      "grad_norm": 0.623126208782196,
      "learning_rate": 1.4619165602552637e-05,
      "loss": 0.2964,
      "step": 285
    },
    {
      "epoch": 0.1926574604243853,
      "grad_norm": 0.5599439144134521,
      "learning_rate": 1.4613593275040572e-05,
      "loss": 0.2582,
      "step": 286
    },
    {
      "epoch": 0.19333108790838666,
      "grad_norm": 0.5614957809448242,
      "learning_rate": 1.4607981553894654e-05,
      "loss": 0.27,
      "step": 287
    },
    {
      "epoch": 0.194004715392388,
      "grad_norm": 0.5625648498535156,
      "learning_rate": 1.4602330470191453e-05,
      "loss": 0.2751,
      "step": 288
    },
    {
      "epoch": 0.19467834287638935,
      "grad_norm": 0.5504026412963867,
      "learning_rate": 1.4596640055225521e-05,
      "loss": 0.2429,
      "step": 289
    },
    {
      "epoch": 0.1953519703603907,
      "grad_norm": 0.5794048309326172,
      "learning_rate": 1.4590910340509224e-05,
      "loss": 0.2882,
      "step": 290
    },
    {
      "epoch": 0.19602559784439205,
      "grad_norm": 0.550942599773407,
      "learning_rate": 1.4585141357772554e-05,
      "loss": 0.2604,
      "step": 291
    },
    {
      "epoch": 0.1966992253283934,
      "grad_norm": 0.6088408827781677,
      "learning_rate": 1.4579333138962966e-05,
      "loss": 0.2993,
      "step": 292
    },
    {
      "epoch": 0.19737285281239475,
      "grad_norm": 0.6309805512428284,
      "learning_rate": 1.4573485716245193e-05,
      "loss": 0.297,
      "step": 293
    },
    {
      "epoch": 0.1980464802963961,
      "grad_norm": 0.6433154344558716,
      "learning_rate": 1.456759912200108e-05,
      "loss": 0.2919,
      "step": 294
    },
    {
      "epoch": 0.19872010778039745,
      "grad_norm": 0.6373067498207092,
      "learning_rate": 1.456167338882938e-05,
      "loss": 0.2719,
      "step": 295
    },
    {
      "epoch": 0.1993937352643988,
      "grad_norm": 0.5514649748802185,
      "learning_rate": 1.4555708549545607e-05,
      "loss": 0.2638,
      "step": 296
    },
    {
      "epoch": 0.20006736274840015,
      "grad_norm": 0.5804110169410706,
      "learning_rate": 1.4549704637181827e-05,
      "loss": 0.2828,
      "step": 297
    },
    {
      "epoch": 0.2007409902324015,
      "grad_norm": 0.5397315621376038,
      "learning_rate": 1.4543661684986484e-05,
      "loss": 0.2712,
      "step": 298
    },
    {
      "epoch": 0.20141461771640282,
      "grad_norm": 0.6435424089431763,
      "learning_rate": 1.4537579726424221e-05,
      "loss": 0.3095,
      "step": 299
    },
    {
      "epoch": 0.20208824520040417,
      "grad_norm": 0.5241397023200989,
      "learning_rate": 1.453145879517569e-05,
      "loss": 0.2635,
      "step": 300
    },
    {
      "epoch": 0.20208824520040417,
      "eval_loss": 0.2736159861087799,
      "eval_runtime": 107.1602,
      "eval_samples_per_second": 46.659,
      "eval_steps_per_second": 2.921,
      "step": 300
    },
    {
      "epoch": 0.20276187268440551,
      "grad_norm": 0.5774008631706238,
      "learning_rate": 1.4525298925137362e-05,
      "loss": 0.2752,
      "step": 301
    },
    {
      "epoch": 0.20343550016840686,
      "grad_norm": 0.5994575619697571,
      "learning_rate": 1.4519100150421343e-05,
      "loss": 0.3073,
      "step": 302
    },
    {
      "epoch": 0.2041091276524082,
      "grad_norm": 0.5691470503807068,
      "learning_rate": 1.4512862505355195e-05,
      "loss": 0.2846,
      "step": 303
    },
    {
      "epoch": 0.20478275513640956,
      "grad_norm": 0.5722606182098389,
      "learning_rate": 1.450658602448172e-05,
      "loss": 0.2549,
      "step": 304
    },
    {
      "epoch": 0.2054563826204109,
      "grad_norm": 0.632279634475708,
      "learning_rate": 1.45002707425588e-05,
      "loss": 0.3197,
      "step": 305
    },
    {
      "epoch": 0.20613001010441226,
      "grad_norm": 0.5538962483406067,
      "learning_rate": 1.449391669455918e-05,
      "loss": 0.2656,
      "step": 306
    },
    {
      "epoch": 0.2068036375884136,
      "grad_norm": 0.5925297737121582,
      "learning_rate": 1.4487523915670286e-05,
      "loss": 0.2821,
      "step": 307
    },
    {
      "epoch": 0.20747726507241496,
      "grad_norm": 0.6299713850021362,
      "learning_rate": 1.448109244129403e-05,
      "loss": 0.3116,
      "step": 308
    },
    {
      "epoch": 0.2081508925564163,
      "grad_norm": 0.6114513874053955,
      "learning_rate": 1.447462230704661e-05,
      "loss": 0.285,
      "step": 309
    },
    {
      "epoch": 0.20882452004041765,
      "grad_norm": 0.5723987817764282,
      "learning_rate": 1.4468113548758313e-05,
      "loss": 0.278,
      "step": 310
    },
    {
      "epoch": 0.209498147524419,
      "grad_norm": 0.5769573450088501,
      "learning_rate": 1.4461566202473322e-05,
      "loss": 0.2892,
      "step": 311
    },
    {
      "epoch": 0.21017177500842035,
      "grad_norm": 0.6040593981742859,
      "learning_rate": 1.4454980304449506e-05,
      "loss": 0.3123,
      "step": 312
    },
    {
      "epoch": 0.2108454024924217,
      "grad_norm": 0.5362566113471985,
      "learning_rate": 1.4448355891158235e-05,
      "loss": 0.24,
      "step": 313
    },
    {
      "epoch": 0.21151902997642305,
      "grad_norm": 0.560070812702179,
      "learning_rate": 1.4441692999284159e-05,
      "loss": 0.2663,
      "step": 314
    },
    {
      "epoch": 0.21219265746042437,
      "grad_norm": 0.6649965047836304,
      "learning_rate": 1.443499166572502e-05,
      "loss": 0.3441,
      "step": 315
    },
    {
      "epoch": 0.21286628494442572,
      "grad_norm": 0.5337359309196472,
      "learning_rate": 1.4428251927591445e-05,
      "loss": 0.253,
      "step": 316
    },
    {
      "epoch": 0.21353991242842707,
      "grad_norm": 0.6185274720191956,
      "learning_rate": 1.4421473822206729e-05,
      "loss": 0.305,
      "step": 317
    },
    {
      "epoch": 0.21421353991242842,
      "grad_norm": 0.5125210881233215,
      "learning_rate": 1.4414657387106646e-05,
      "loss": 0.2774,
      "step": 318
    },
    {
      "epoch": 0.21488716739642977,
      "grad_norm": 0.5758813619613647,
      "learning_rate": 1.4407802660039226e-05,
      "loss": 0.2484,
      "step": 319
    },
    {
      "epoch": 0.21556079488043112,
      "grad_norm": 0.5220269560813904,
      "learning_rate": 1.4400909678964556e-05,
      "loss": 0.2399,
      "step": 320
    },
    {
      "epoch": 0.21623442236443247,
      "grad_norm": 0.5919392704963684,
      "learning_rate": 1.4393978482054561e-05,
      "loss": 0.2924,
      "step": 321
    },
    {
      "epoch": 0.21690804984843381,
      "grad_norm": 0.5359899997711182,
      "learning_rate": 1.4387009107692808e-05,
      "loss": 0.2493,
      "step": 322
    },
    {
      "epoch": 0.21758167733243516,
      "grad_norm": 0.568356454372406,
      "learning_rate": 1.4380001594474267e-05,
      "loss": 0.2877,
      "step": 323
    },
    {
      "epoch": 0.2182553048164365,
      "grad_norm": 0.5183501243591309,
      "learning_rate": 1.4372955981205127e-05,
      "loss": 0.262,
      "step": 324
    },
    {
      "epoch": 0.21892893230043786,
      "grad_norm": 0.5353648662567139,
      "learning_rate": 1.436587230690256e-05,
      "loss": 0.269,
      "step": 325
    },
    {
      "epoch": 0.2196025597844392,
      "grad_norm": 0.5863710641860962,
      "learning_rate": 1.4358750610794522e-05,
      "loss": 0.2933,
      "step": 326
    },
    {
      "epoch": 0.22027618726844056,
      "grad_norm": 0.5193360447883606,
      "learning_rate": 1.4351590932319506e-05,
      "loss": 0.2539,
      "step": 327
    },
    {
      "epoch": 0.2209498147524419,
      "grad_norm": 0.521597146987915,
      "learning_rate": 1.4344393311126367e-05,
      "loss": 0.24,
      "step": 328
    },
    {
      "epoch": 0.22162344223644326,
      "grad_norm": 0.5621289014816284,
      "learning_rate": 1.4337157787074063e-05,
      "loss": 0.2647,
      "step": 329
    },
    {
      "epoch": 0.2222970697204446,
      "grad_norm": 0.6134183406829834,
      "learning_rate": 1.432988440023146e-05,
      "loss": 0.2846,
      "step": 330
    },
    {
      "epoch": 0.22297069720444593,
      "grad_norm": 0.5819990634918213,
      "learning_rate": 1.4322573190877091e-05,
      "loss": 0.2725,
      "step": 331
    },
    {
      "epoch": 0.22364432468844728,
      "grad_norm": 0.6009438037872314,
      "learning_rate": 1.4315224199498952e-05,
      "loss": 0.2507,
      "step": 332
    },
    {
      "epoch": 0.22431795217244863,
      "grad_norm": 0.5484105944633484,
      "learning_rate": 1.4307837466794258e-05,
      "loss": 0.2715,
      "step": 333
    },
    {
      "epoch": 0.22499157965644997,
      "grad_norm": 0.5025244951248169,
      "learning_rate": 1.4300413033669241e-05,
      "loss": 0.2257,
      "step": 334
    },
    {
      "epoch": 0.22566520714045132,
      "grad_norm": 0.5583484172821045,
      "learning_rate": 1.4292950941238898e-05,
      "loss": 0.3015,
      "step": 335
    },
    {
      "epoch": 0.22633883462445267,
      "grad_norm": 0.5975006222724915,
      "learning_rate": 1.4285451230826783e-05,
      "loss": 0.2924,
      "step": 336
    },
    {
      "epoch": 0.22701246210845402,
      "grad_norm": 0.6017155051231384,
      "learning_rate": 1.4277913943964763e-05,
      "loss": 0.2928,
      "step": 337
    },
    {
      "epoch": 0.22768608959245537,
      "grad_norm": 0.5619384050369263,
      "learning_rate": 1.4270339122392808e-05,
      "loss": 0.2744,
      "step": 338
    },
    {
      "epoch": 0.22835971707645672,
      "grad_norm": 0.576554536819458,
      "learning_rate": 1.4262726808058735e-05,
      "loss": 0.3019,
      "step": 339
    },
    {
      "epoch": 0.22903334456045807,
      "grad_norm": 0.5621641874313354,
      "learning_rate": 1.4255077043117994e-05,
      "loss": 0.2801,
      "step": 340
    },
    {
      "epoch": 0.22970697204445942,
      "grad_norm": 0.5104705095291138,
      "learning_rate": 1.424738986993343e-05,
      "loss": 0.2572,
      "step": 341
    },
    {
      "epoch": 0.23038059952846077,
      "grad_norm": 0.5731213688850403,
      "learning_rate": 1.4239665331075048e-05,
      "loss": 0.2545,
      "step": 342
    },
    {
      "epoch": 0.23105422701246212,
      "grad_norm": 0.6381127238273621,
      "learning_rate": 1.4231903469319772e-05,
      "loss": 0.3023,
      "step": 343
    },
    {
      "epoch": 0.23172785449646346,
      "grad_norm": 0.5358138680458069,
      "learning_rate": 1.4224104327651213e-05,
      "loss": 0.2597,
      "step": 344
    },
    {
      "epoch": 0.2324014819804648,
      "grad_norm": 0.5517827272415161,
      "learning_rate": 1.4216267949259437e-05,
      "loss": 0.2669,
      "step": 345
    },
    {
      "epoch": 0.23307510946446616,
      "grad_norm": 0.5380638241767883,
      "learning_rate": 1.4208394377540712e-05,
      "loss": 0.2706,
      "step": 346
    },
    {
      "epoch": 0.23374873694846748,
      "grad_norm": 0.6162987351417542,
      "learning_rate": 1.4200483656097278e-05,
      "loss": 0.2721,
      "step": 347
    },
    {
      "epoch": 0.23442236443246883,
      "grad_norm": 0.6142714619636536,
      "learning_rate": 1.4192535828737102e-05,
      "loss": 0.3158,
      "step": 348
    },
    {
      "epoch": 0.23509599191647018,
      "grad_norm": 0.6231828331947327,
      "learning_rate": 1.4184550939473644e-05,
      "loss": 0.3022,
      "step": 349
    },
    {
      "epoch": 0.23576961940047153,
      "grad_norm": 0.5371239185333252,
      "learning_rate": 1.4176529032525584e-05,
      "loss": 0.2372,
      "step": 350
    },
    {
      "epoch": 0.23644324688447288,
      "grad_norm": 0.5987442135810852,
      "learning_rate": 1.4168470152316624e-05,
      "loss": 0.2856,
      "step": 351
    },
    {
      "epoch": 0.23711687436847423,
      "grad_norm": 0.5490831732749939,
      "learning_rate": 1.41603743434752e-05,
      "loss": 0.2352,
      "step": 352
    },
    {
      "epoch": 0.23779050185247558,
      "grad_norm": 0.5611885786056519,
      "learning_rate": 1.415224165083426e-05,
      "loss": 0.2763,
      "step": 353
    },
    {
      "epoch": 0.23846412933647693,
      "grad_norm": 0.5451275706291199,
      "learning_rate": 1.4144072119431e-05,
      "loss": 0.2725,
      "step": 354
    },
    {
      "epoch": 0.23913775682047828,
      "grad_norm": 0.5789247155189514,
      "learning_rate": 1.413586579450662e-05,
      "loss": 0.2604,
      "step": 355
    },
    {
      "epoch": 0.23981138430447962,
      "grad_norm": 0.6164606213569641,
      "learning_rate": 1.4127622721506087e-05,
      "loss": 0.2932,
      "step": 356
    },
    {
      "epoch": 0.24048501178848097,
      "grad_norm": 0.5564325451850891,
      "learning_rate": 1.4119342946077864e-05,
      "loss": 0.2735,
      "step": 357
    },
    {
      "epoch": 0.24115863927248232,
      "grad_norm": 0.6473014950752258,
      "learning_rate": 1.4111026514073657e-05,
      "loss": 0.2808,
      "step": 358
    },
    {
      "epoch": 0.24183226675648367,
      "grad_norm": 0.5950415730476379,
      "learning_rate": 1.4102673471548186e-05,
      "loss": 0.2819,
      "step": 359
    },
    {
      "epoch": 0.24250589424048502,
      "grad_norm": 0.576295793056488,
      "learning_rate": 1.4094283864758896e-05,
      "loss": 0.2818,
      "step": 360
    },
    {
      "epoch": 0.24317952172448637,
      "grad_norm": 0.5290201306343079,
      "learning_rate": 1.4085857740165727e-05,
      "loss": 0.2731,
      "step": 361
    },
    {
      "epoch": 0.24385314920848772,
      "grad_norm": 0.5469079613685608,
      "learning_rate": 1.4077395144430845e-05,
      "loss": 0.2533,
      "step": 362
    },
    {
      "epoch": 0.24452677669248907,
      "grad_norm": 0.553629457950592,
      "learning_rate": 1.4068896124418383e-05,
      "loss": 0.2784,
      "step": 363
    },
    {
      "epoch": 0.2452004041764904,
      "grad_norm": 0.5426369905471802,
      "learning_rate": 1.4060360727194188e-05,
      "loss": 0.2687,
      "step": 364
    },
    {
      "epoch": 0.24587403166049174,
      "grad_norm": 0.5466113686561584,
      "learning_rate": 1.4051789000025555e-05,
      "loss": 0.2721,
      "step": 365
    },
    {
      "epoch": 0.2465476591444931,
      "grad_norm": 0.5685258507728577,
      "learning_rate": 1.4043180990380968e-05,
      "loss": 0.283,
      "step": 366
    },
    {
      "epoch": 0.24722128662849444,
      "grad_norm": 0.5648797154426575,
      "learning_rate": 1.4034536745929835e-05,
      "loss": 0.2579,
      "step": 367
    },
    {
      "epoch": 0.24789491411249578,
      "grad_norm": 0.5363840460777283,
      "learning_rate": 1.4025856314542223e-05,
      "loss": 0.2577,
      "step": 368
    },
    {
      "epoch": 0.24856854159649713,
      "grad_norm": 0.5171375870704651,
      "learning_rate": 1.40171397442886e-05,
      "loss": 0.2351,
      "step": 369
    },
    {
      "epoch": 0.24924216908049848,
      "grad_norm": 0.646500825881958,
      "learning_rate": 1.4008387083439554e-05,
      "loss": 0.3039,
      "step": 370
    },
    {
      "epoch": 0.24991579656449983,
      "grad_norm": 0.5827479362487793,
      "learning_rate": 1.3999598380465552e-05,
      "loss": 0.2913,
      "step": 371
    },
    {
      "epoch": 0.25058942404850115,
      "grad_norm": 0.5602329969406128,
      "learning_rate": 1.3990773684036636e-05,
      "loss": 0.2822,
      "step": 372
    },
    {
      "epoch": 0.2512630515325025,
      "grad_norm": 0.5731973648071289,
      "learning_rate": 1.3981913043022187e-05,
      "loss": 0.2638,
      "step": 373
    },
    {
      "epoch": 0.25193667901650385,
      "grad_norm": 0.6127945780754089,
      "learning_rate": 1.397301650649063e-05,
      "loss": 0.314,
      "step": 374
    },
    {
      "epoch": 0.2526103065005052,
      "grad_norm": 0.5554071664810181,
      "learning_rate": 1.396408412370918e-05,
      "loss": 0.2575,
      "step": 375
    },
    {
      "epoch": 0.25328393398450655,
      "grad_norm": 0.5913053750991821,
      "learning_rate": 1.3955115944143558e-05,
      "loss": 0.2669,
      "step": 376
    },
    {
      "epoch": 0.2539575614685079,
      "grad_norm": 0.6104479432106018,
      "learning_rate": 1.3946112017457715e-05,
      "loss": 0.2575,
      "step": 377
    },
    {
      "epoch": 0.25463118895250925,
      "grad_norm": 0.6109972596168518,
      "learning_rate": 1.393707239351357e-05,
      "loss": 0.3141,
      "step": 378
    },
    {
      "epoch": 0.2553048164365106,
      "grad_norm": 0.605560302734375,
      "learning_rate": 1.3927997122370724e-05,
      "loss": 0.2869,
      "step": 379
    },
    {
      "epoch": 0.25597844392051194,
      "grad_norm": 0.5215985774993896,
      "learning_rate": 1.3918886254286182e-05,
      "loss": 0.2464,
      "step": 380
    },
    {
      "epoch": 0.2566520714045133,
      "grad_norm": 0.5480206608772278,
      "learning_rate": 1.3909739839714081e-05,
      "loss": 0.2713,
      "step": 381
    },
    {
      "epoch": 0.25732569888851464,
      "grad_norm": 0.5150758028030396,
      "learning_rate": 1.3900557929305408e-05,
      "loss": 0.2537,
      "step": 382
    },
    {
      "epoch": 0.257999326372516,
      "grad_norm": 0.606860876083374,
      "learning_rate": 1.3891340573907715e-05,
      "loss": 0.2929,
      "step": 383
    },
    {
      "epoch": 0.25867295385651734,
      "grad_norm": 0.5383312106132507,
      "learning_rate": 1.3882087824564841e-05,
      "loss": 0.2778,
      "step": 384
    },
    {
      "epoch": 0.2593465813405187,
      "grad_norm": 0.5356404185295105,
      "learning_rate": 1.3872799732516635e-05,
      "loss": 0.2318,
      "step": 385
    },
    {
      "epoch": 0.26002020882452004,
      "grad_norm": 0.5665723085403442,
      "learning_rate": 1.386347634919866e-05,
      "loss": 0.2898,
      "step": 386
    },
    {
      "epoch": 0.2606938363085214,
      "grad_norm": 0.5390300750732422,
      "learning_rate": 1.3854117726241922e-05,
      "loss": 0.2789,
      "step": 387
    },
    {
      "epoch": 0.26136746379252274,
      "grad_norm": 0.5479271411895752,
      "learning_rate": 1.3844723915472568e-05,
      "loss": 0.2552,
      "step": 388
    },
    {
      "epoch": 0.2620410912765241,
      "grad_norm": 0.6038428544998169,
      "learning_rate": 1.3835294968911615e-05,
      "loss": 0.3018,
      "step": 389
    },
    {
      "epoch": 0.26271471876052543,
      "grad_norm": 0.5380761027336121,
      "learning_rate": 1.3825830938774653e-05,
      "loss": 0.2683,
      "step": 390
    },
    {
      "epoch": 0.2633883462445268,
      "grad_norm": 0.5072317719459534,
      "learning_rate": 1.3816331877471562e-05,
      "loss": 0.2728,
      "step": 391
    },
    {
      "epoch": 0.26406197372852813,
      "grad_norm": 0.5953329205513,
      "learning_rate": 1.3806797837606206e-05,
      "loss": 0.2644,
      "step": 392
    },
    {
      "epoch": 0.2647356012125295,
      "grad_norm": 0.5941304564476013,
      "learning_rate": 1.3797228871976162e-05,
      "loss": 0.2841,
      "step": 393
    },
    {
      "epoch": 0.26540922869653083,
      "grad_norm": 0.6646502614021301,
      "learning_rate": 1.378762503357242e-05,
      "loss": 0.2966,
      "step": 394
    },
    {
      "epoch": 0.2660828561805322,
      "grad_norm": 0.545456051826477,
      "learning_rate": 1.377798637557908e-05,
      "loss": 0.2481,
      "step": 395
    },
    {
      "epoch": 0.26675648366453353,
      "grad_norm": 0.5886520147323608,
      "learning_rate": 1.3768312951373076e-05,
      "loss": 0.2735,
      "step": 396
    },
    {
      "epoch": 0.2674301111485349,
      "grad_norm": 0.5731514096260071,
      "learning_rate": 1.3758604814523863e-05,
      "loss": 0.2953,
      "step": 397
    },
    {
      "epoch": 0.2681037386325362,
      "grad_norm": 0.5029922723770142,
      "learning_rate": 1.3748862018793131e-05,
      "loss": 0.228,
      "step": 398
    },
    {
      "epoch": 0.2687773661165376,
      "grad_norm": 0.557115375995636,
      "learning_rate": 1.3739084618134502e-05,
      "loss": 0.2861,
      "step": 399
    },
    {
      "epoch": 0.2694509936005389,
      "grad_norm": 0.5246098041534424,
      "learning_rate": 1.3729272666693235e-05,
      "loss": 0.2705,
      "step": 400
    },
    {
      "epoch": 0.2694509936005389,
      "eval_loss": 0.2706840932369232,
      "eval_runtime": 105.373,
      "eval_samples_per_second": 47.451,
      "eval_steps_per_second": 2.97,
      "step": 400
    },
    {
      "epoch": 0.2701246210845403,
      "grad_norm": 0.5355361104011536,
      "learning_rate": 1.371942621880592e-05,
      "loss": 0.249,
      "step": 401
    },
    {
      "epoch": 0.2707982485685416,
      "grad_norm": 0.5726237297058105,
      "learning_rate": 1.3709545329000187e-05,
      "loss": 0.2849,
      "step": 402
    },
    {
      "epoch": 0.27147187605254297,
      "grad_norm": 0.5560792088508606,
      "learning_rate": 1.3699630051994395e-05,
      "loss": 0.2397,
      "step": 403
    },
    {
      "epoch": 0.27214550353654426,
      "grad_norm": 0.509462833404541,
      "learning_rate": 1.3689680442697332e-05,
      "loss": 0.2412,
      "step": 404
    },
    {
      "epoch": 0.2728191310205456,
      "grad_norm": 0.5348261594772339,
      "learning_rate": 1.3679696556207913e-05,
      "loss": 0.2588,
      "step": 405
    },
    {
      "epoch": 0.27349275850454696,
      "grad_norm": 0.5228528380393982,
      "learning_rate": 1.3669678447814871e-05,
      "loss": 0.2482,
      "step": 406
    },
    {
      "epoch": 0.2741663859885483,
      "grad_norm": 0.5533547401428223,
      "learning_rate": 1.3659626172996459e-05,
      "loss": 0.2581,
      "step": 407
    },
    {
      "epoch": 0.27484001347254966,
      "grad_norm": 0.538163959980011,
      "learning_rate": 1.3649539787420126e-05,
      "loss": 0.2444,
      "step": 408
    },
    {
      "epoch": 0.275513640956551,
      "grad_norm": 0.6091170907020569,
      "learning_rate": 1.3639419346942227e-05,
      "loss": 0.2963,
      "step": 409
    },
    {
      "epoch": 0.27618726844055236,
      "grad_norm": 0.5507506728172302,
      "learning_rate": 1.3629264907607709e-05,
      "loss": 0.2835,
      "step": 410
    },
    {
      "epoch": 0.2768608959245537,
      "grad_norm": 0.5167334079742432,
      "learning_rate": 1.361907652564979e-05,
      "loss": 0.2751,
      "step": 411
    },
    {
      "epoch": 0.27753452340855506,
      "grad_norm": 0.6182762384414673,
      "learning_rate": 1.3608854257489656e-05,
      "loss": 0.2953,
      "step": 412
    },
    {
      "epoch": 0.2782081508925564,
      "grad_norm": 0.6356998085975647,
      "learning_rate": 1.3598598159736155e-05,
      "loss": 0.2586,
      "step": 413
    },
    {
      "epoch": 0.27888177837655775,
      "grad_norm": 0.5957326889038086,
      "learning_rate": 1.358830828918547e-05,
      "loss": 0.283,
      "step": 414
    },
    {
      "epoch": 0.2795554058605591,
      "grad_norm": 0.5173368453979492,
      "learning_rate": 1.3577984702820811e-05,
      "loss": 0.2403,
      "step": 415
    },
    {
      "epoch": 0.28022903334456045,
      "grad_norm": 0.5449368357658386,
      "learning_rate": 1.3567627457812107e-05,
      "loss": 0.2641,
      "step": 416
    },
    {
      "epoch": 0.2809026608285618,
      "grad_norm": 0.6340479850769043,
      "learning_rate": 1.355723661151567e-05,
      "loss": 0.3286,
      "step": 417
    },
    {
      "epoch": 0.28157628831256315,
      "grad_norm": 0.49671491980552673,
      "learning_rate": 1.3546812221473898e-05,
      "loss": 0.2585,
      "step": 418
    },
    {
      "epoch": 0.2822499157965645,
      "grad_norm": 0.5974727272987366,
      "learning_rate": 1.3536354345414944e-05,
      "loss": 0.2674,
      "step": 419
    },
    {
      "epoch": 0.28292354328056585,
      "grad_norm": 0.5984825491905212,
      "learning_rate": 1.35258630412524e-05,
      "loss": 0.2548,
      "step": 420
    },
    {
      "epoch": 0.2835971707645672,
      "grad_norm": 0.5152942538261414,
      "learning_rate": 1.3515338367084975e-05,
      "loss": 0.2323,
      "step": 421
    },
    {
      "epoch": 0.28427079824856855,
      "grad_norm": 0.5210486054420471,
      "learning_rate": 1.3504780381196178e-05,
      "loss": 0.2538,
      "step": 422
    },
    {
      "epoch": 0.2849444257325699,
      "grad_norm": 0.6852086782455444,
      "learning_rate": 1.3494189142053988e-05,
      "loss": 0.3409,
      "step": 423
    },
    {
      "epoch": 0.28561805321657124,
      "grad_norm": 0.5637288689613342,
      "learning_rate": 1.3483564708310535e-05,
      "loss": 0.2435,
      "step": 424
    },
    {
      "epoch": 0.2862916807005726,
      "grad_norm": 0.565467357635498,
      "learning_rate": 1.3472907138801775e-05,
      "loss": 0.2699,
      "step": 425
    },
    {
      "epoch": 0.28696530818457394,
      "grad_norm": 0.6443371176719666,
      "learning_rate": 1.346221649254716e-05,
      "loss": 0.3226,
      "step": 426
    },
    {
      "epoch": 0.2876389356685753,
      "grad_norm": 0.5877301096916199,
      "learning_rate": 1.3451492828749317e-05,
      "loss": 0.2626,
      "step": 427
    },
    {
      "epoch": 0.28831256315257664,
      "grad_norm": 0.635368824005127,
      "learning_rate": 1.3440736206793717e-05,
      "loss": 0.2808,
      "step": 428
    },
    {
      "epoch": 0.288986190636578,
      "grad_norm": 0.5623096823692322,
      "learning_rate": 1.3429946686248346e-05,
      "loss": 0.2583,
      "step": 429
    },
    {
      "epoch": 0.28965981812057934,
      "grad_norm": 0.5355499386787415,
      "learning_rate": 1.341912432686338e-05,
      "loss": 0.2425,
      "step": 430
    },
    {
      "epoch": 0.2903334456045807,
      "grad_norm": 0.5870991349220276,
      "learning_rate": 1.3408269188570837e-05,
      "loss": 0.2638,
      "step": 431
    },
    {
      "epoch": 0.29100707308858204,
      "grad_norm": 0.5296127796173096,
      "learning_rate": 1.3397381331484273e-05,
      "loss": 0.2587,
      "step": 432
    },
    {
      "epoch": 0.2916807005725834,
      "grad_norm": 0.5635933876037598,
      "learning_rate": 1.3386460815898427e-05,
      "loss": 0.2966,
      "step": 433
    },
    {
      "epoch": 0.29235432805658473,
      "grad_norm": 0.5246622562408447,
      "learning_rate": 1.3375507702288894e-05,
      "loss": 0.2513,
      "step": 434
    },
    {
      "epoch": 0.2930279555405861,
      "grad_norm": 0.6050205826759338,
      "learning_rate": 1.3364522051311793e-05,
      "loss": 0.3016,
      "step": 435
    },
    {
      "epoch": 0.2937015830245874,
      "grad_norm": 0.5831138491630554,
      "learning_rate": 1.3353503923803424e-05,
      "loss": 0.312,
      "step": 436
    },
    {
      "epoch": 0.2943752105085887,
      "grad_norm": 0.5354754328727722,
      "learning_rate": 1.3342453380779939e-05,
      "loss": 0.2743,
      "step": 437
    },
    {
      "epoch": 0.2950488379925901,
      "grad_norm": 0.6059128642082214,
      "learning_rate": 1.3331370483437e-05,
      "loss": 0.2836,
      "step": 438
    },
    {
      "epoch": 0.2957224654765914,
      "grad_norm": 0.6208754181861877,
      "learning_rate": 1.332025529314944e-05,
      "loss": 0.3069,
      "step": 439
    },
    {
      "epoch": 0.29639609296059277,
      "grad_norm": 0.5791683197021484,
      "learning_rate": 1.3309107871470922e-05,
      "loss": 0.2904,
      "step": 440
    },
    {
      "epoch": 0.2970697204445941,
      "grad_norm": 0.5765690803527832,
      "learning_rate": 1.3297928280133606e-05,
      "loss": 0.3015,
      "step": 441
    },
    {
      "epoch": 0.29774334792859547,
      "grad_norm": 0.5978572368621826,
      "learning_rate": 1.3286716581047791e-05,
      "loss": 0.2827,
      "step": 442
    },
    {
      "epoch": 0.2984169754125968,
      "grad_norm": 0.5690959692001343,
      "learning_rate": 1.3275472836301592e-05,
      "loss": 0.2819,
      "step": 443
    },
    {
      "epoch": 0.29909060289659817,
      "grad_norm": 0.5888264775276184,
      "learning_rate": 1.3264197108160582e-05,
      "loss": 0.297,
      "step": 444
    },
    {
      "epoch": 0.2997642303805995,
      "grad_norm": 0.566338837146759,
      "learning_rate": 1.3252889459067452e-05,
      "loss": 0.2703,
      "step": 445
    },
    {
      "epoch": 0.30043785786460087,
      "grad_norm": 0.5249893665313721,
      "learning_rate": 1.3241549951641663e-05,
      "loss": 0.252,
      "step": 446
    },
    {
      "epoch": 0.3011114853486022,
      "grad_norm": 0.6007825136184692,
      "learning_rate": 1.3230178648679102e-05,
      "loss": 0.2696,
      "step": 447
    },
    {
      "epoch": 0.30178511283260356,
      "grad_norm": 0.5482873916625977,
      "learning_rate": 1.3218775613151737e-05,
      "loss": 0.2523,
      "step": 448
    },
    {
      "epoch": 0.3024587403166049,
      "grad_norm": 0.6056886315345764,
      "learning_rate": 1.3207340908207258e-05,
      "loss": 0.2616,
      "step": 449
    },
    {
      "epoch": 0.30313236780060626,
      "grad_norm": 0.5885447859764099,
      "learning_rate": 1.319587459716874e-05,
      "loss": 0.2976,
      "step": 450
    },
    {
      "epoch": 0.3038059952846076,
      "grad_norm": 0.5747894644737244,
      "learning_rate": 1.318437674353428e-05,
      "loss": 0.2898,
      "step": 451
    },
    {
      "epoch": 0.30447962276860896,
      "grad_norm": 0.569401741027832,
      "learning_rate": 1.3172847410976658e-05,
      "loss": 0.3104,
      "step": 452
    },
    {
      "epoch": 0.3051532502526103,
      "grad_norm": 0.5612210631370544,
      "learning_rate": 1.3161286663342972e-05,
      "loss": 0.2825,
      "step": 453
    },
    {
      "epoch": 0.30582687773661166,
      "grad_norm": 0.5914261937141418,
      "learning_rate": 1.3149694564654295e-05,
      "loss": 0.2781,
      "step": 454
    },
    {
      "epoch": 0.306500505220613,
      "grad_norm": 0.5259233713150024,
      "learning_rate": 1.3138071179105314e-05,
      "loss": 0.2542,
      "step": 455
    },
    {
      "epoch": 0.30717413270461436,
      "grad_norm": 0.5168178081512451,
      "learning_rate": 1.3126416571063972e-05,
      "loss": 0.2514,
      "step": 456
    },
    {
      "epoch": 0.3078477601886157,
      "grad_norm": 0.5078200101852417,
      "learning_rate": 1.3114730805071123e-05,
      "loss": 0.2422,
      "step": 457
    },
    {
      "epoch": 0.30852138767261705,
      "grad_norm": 0.5727274417877197,
      "learning_rate": 1.3103013945840166e-05,
      "loss": 0.2809,
      "step": 458
    },
    {
      "epoch": 0.3091950151566184,
      "grad_norm": 0.5502845048904419,
      "learning_rate": 1.309126605825668e-05,
      "loss": 0.2552,
      "step": 459
    },
    {
      "epoch": 0.30986864264061975,
      "grad_norm": 0.5696067214012146,
      "learning_rate": 1.3079487207378084e-05,
      "loss": 0.2959,
      "step": 460
    },
    {
      "epoch": 0.3105422701246211,
      "grad_norm": 0.5644879341125488,
      "learning_rate": 1.3067677458433258e-05,
      "loss": 0.2713,
      "step": 461
    },
    {
      "epoch": 0.31121589760862245,
      "grad_norm": 0.5638664364814758,
      "learning_rate": 1.3055836876822196e-05,
      "loss": 0.2687,
      "step": 462
    },
    {
      "epoch": 0.3118895250926238,
      "grad_norm": 0.5337838530540466,
      "learning_rate": 1.3043965528115625e-05,
      "loss": 0.2238,
      "step": 463
    },
    {
      "epoch": 0.31256315257662515,
      "grad_norm": 0.5844706892967224,
      "learning_rate": 1.3032063478054666e-05,
      "loss": 0.268,
      "step": 464
    },
    {
      "epoch": 0.3132367800606265,
      "grad_norm": 0.6730402112007141,
      "learning_rate": 1.3020130792550456e-05,
      "loss": 0.2976,
      "step": 465
    },
    {
      "epoch": 0.31391040754462785,
      "grad_norm": 0.5756520628929138,
      "learning_rate": 1.3008167537683776e-05,
      "loss": 0.2859,
      "step": 466
    },
    {
      "epoch": 0.3145840350286292,
      "grad_norm": 0.5855886340141296,
      "learning_rate": 1.2996173779704704e-05,
      "loss": 0.2997,
      "step": 467
    },
    {
      "epoch": 0.3152576625126305,
      "grad_norm": 0.5359857082366943,
      "learning_rate": 1.2984149585032237e-05,
      "loss": 0.2814,
      "step": 468
    },
    {
      "epoch": 0.31593128999663184,
      "grad_norm": 0.5448024868965149,
      "learning_rate": 1.2972095020253912e-05,
      "loss": 0.2681,
      "step": 469
    },
    {
      "epoch": 0.3166049174806332,
      "grad_norm": 0.518844723701477,
      "learning_rate": 1.296001015212547e-05,
      "loss": 0.2538,
      "step": 470
    },
    {
      "epoch": 0.31727854496463453,
      "grad_norm": 0.5422329306602478,
      "learning_rate": 1.2947895047570446e-05,
      "loss": 0.2346,
      "step": 471
    },
    {
      "epoch": 0.3179521724486359,
      "grad_norm": 0.5567420721054077,
      "learning_rate": 1.2935749773679833e-05,
      "loss": 0.259,
      "step": 472
    },
    {
      "epoch": 0.31862579993263723,
      "grad_norm": 0.5199055671691895,
      "learning_rate": 1.2923574397711684e-05,
      "loss": 0.2273,
      "step": 473
    },
    {
      "epoch": 0.3192994274166386,
      "grad_norm": 0.552947461605072,
      "learning_rate": 1.291136898709076e-05,
      "loss": 0.2541,
      "step": 474
    },
    {
      "epoch": 0.31997305490063993,
      "grad_norm": 0.537124752998352,
      "learning_rate": 1.2899133609408146e-05,
      "loss": 0.2709,
      "step": 475
    },
    {
      "epoch": 0.3206466823846413,
      "grad_norm": 0.5493146777153015,
      "learning_rate": 1.2886868332420873e-05,
      "loss": 0.2838,
      "step": 476
    },
    {
      "epoch": 0.32132030986864263,
      "grad_norm": 0.6109126806259155,
      "learning_rate": 1.2874573224051556e-05,
      "loss": 0.3088,
      "step": 477
    },
    {
      "epoch": 0.321993937352644,
      "grad_norm": 0.5879717469215393,
      "learning_rate": 1.2862248352388005e-05,
      "loss": 0.282,
      "step": 478
    },
    {
      "epoch": 0.3226675648366453,
      "grad_norm": 0.5227838754653931,
      "learning_rate": 1.2849893785682852e-05,
      "loss": 0.2646,
      "step": 479
    },
    {
      "epoch": 0.3233411923206467,
      "grad_norm": 0.4744933545589447,
      "learning_rate": 1.2837509592353181e-05,
      "loss": 0.2219,
      "step": 480
    },
    {
      "epoch": 0.324014819804648,
      "grad_norm": 0.508222758769989,
      "learning_rate": 1.2825095840980133e-05,
      "loss": 0.2698,
      "step": 481
    },
    {
      "epoch": 0.3246884472886494,
      "grad_norm": 0.5351443290710449,
      "learning_rate": 1.2812652600308544e-05,
      "loss": 0.2617,
      "step": 482
    },
    {
      "epoch": 0.3253620747726507,
      "grad_norm": 0.5842475295066833,
      "learning_rate": 1.2800179939246552e-05,
      "loss": 0.2496,
      "step": 483
    },
    {
      "epoch": 0.32603570225665207,
      "grad_norm": 0.5165258646011353,
      "learning_rate": 1.2787677926865216e-05,
      "loss": 0.2399,
      "step": 484
    },
    {
      "epoch": 0.3267093297406534,
      "grad_norm": 0.5721768736839294,
      "learning_rate": 1.2775146632398142e-05,
      "loss": 0.2754,
      "step": 485
    },
    {
      "epoch": 0.32738295722465477,
      "grad_norm": 0.47171083092689514,
      "learning_rate": 1.2762586125241093e-05,
      "loss": 0.2107,
      "step": 486
    },
    {
      "epoch": 0.3280565847086561,
      "grad_norm": 0.5318099856376648,
      "learning_rate": 1.2749996474951603e-05,
      "loss": 0.2422,
      "step": 487
    },
    {
      "epoch": 0.32873021219265747,
      "grad_norm": 0.5478540062904358,
      "learning_rate": 1.2737377751248598e-05,
      "loss": 0.2634,
      "step": 488
    },
    {
      "epoch": 0.3294038396766588,
      "grad_norm": 0.4972551167011261,
      "learning_rate": 1.2724730024012002e-05,
      "loss": 0.232,
      "step": 489
    },
    {
      "epoch": 0.33007746716066017,
      "grad_norm": 0.6141415238380432,
      "learning_rate": 1.2712053363282363e-05,
      "loss": 0.2998,
      "step": 490
    },
    {
      "epoch": 0.3307510946446615,
      "grad_norm": 0.5177733302116394,
      "learning_rate": 1.2699347839260448e-05,
      "loss": 0.2574,
      "step": 491
    },
    {
      "epoch": 0.33142472212866286,
      "grad_norm": 0.5531916618347168,
      "learning_rate": 1.268661352230687e-05,
      "loss": 0.2719,
      "step": 492
    },
    {
      "epoch": 0.3320983496126642,
      "grad_norm": 0.5089963674545288,
      "learning_rate": 1.2673850482941687e-05,
      "loss": 0.2508,
      "step": 493
    },
    {
      "epoch": 0.33277197709666556,
      "grad_norm": 0.557072103023529,
      "learning_rate": 1.2661058791844016e-05,
      "loss": 0.2823,
      "step": 494
    },
    {
      "epoch": 0.3334456045806669,
      "grad_norm": 0.6557756662368774,
      "learning_rate": 1.2648238519851644e-05,
      "loss": 0.2821,
      "step": 495
    },
    {
      "epoch": 0.33411923206466826,
      "grad_norm": 0.5633836984634399,
      "learning_rate": 1.2635389737960632e-05,
      "loss": 0.2576,
      "step": 496
    },
    {
      "epoch": 0.3347928595486696,
      "grad_norm": 0.594456136226654,
      "learning_rate": 1.262251251732492e-05,
      "loss": 0.2985,
      "step": 497
    },
    {
      "epoch": 0.33546648703267096,
      "grad_norm": 0.5753186345100403,
      "learning_rate": 1.2609606929255942e-05,
      "loss": 0.2775,
      "step": 498
    },
    {
      "epoch": 0.3361401145166723,
      "grad_norm": 0.6262162327766418,
      "learning_rate": 1.259667304522222e-05,
      "loss": 0.3254,
      "step": 499
    },
    {
      "epoch": 0.33681374200067365,
      "grad_norm": 0.5529574155807495,
      "learning_rate": 1.2583710936848977e-05,
      "loss": 0.2711,
      "step": 500
    },
    {
      "epoch": 0.33681374200067365,
      "eval_loss": 0.2681807279586792,
      "eval_runtime": 104.7062,
      "eval_samples_per_second": 47.753,
      "eval_steps_per_second": 2.989,
      "step": 500
    },
    {
      "epoch": 0.33748736948467495,
      "grad_norm": 0.6187270283699036,
      "learning_rate": 1.2570720675917734e-05,
      "loss": 0.3082,
      "step": 501
    },
    {
      "epoch": 0.3381609969686763,
      "grad_norm": 0.5153407454490662,
      "learning_rate": 1.2557702334365916e-05,
      "loss": 0.26,
      "step": 502
    },
    {
      "epoch": 0.33883462445267765,
      "grad_norm": 0.5447744727134705,
      "learning_rate": 1.2544655984286451e-05,
      "loss": 0.2641,
      "step": 503
    },
    {
      "epoch": 0.339508251936679,
      "grad_norm": 0.5450101494789124,
      "learning_rate": 1.253158169792738e-05,
      "loss": 0.276,
      "step": 504
    },
    {
      "epoch": 0.34018187942068034,
      "grad_norm": 0.6855320930480957,
      "learning_rate": 1.2518479547691437e-05,
      "loss": 0.3589,
      "step": 505
    },
    {
      "epoch": 0.3408555069046817,
      "grad_norm": 0.52507483959198,
      "learning_rate": 1.250534960613567e-05,
      "loss": 0.2489,
      "step": 506
    },
    {
      "epoch": 0.34152913438868304,
      "grad_norm": 0.5259436964988708,
      "learning_rate": 1.2492191945971028e-05,
      "loss": 0.2568,
      "step": 507
    },
    {
      "epoch": 0.3422027618726844,
      "grad_norm": 0.5746189951896667,
      "learning_rate": 1.2479006640061958e-05,
      "loss": 0.2878,
      "step": 508
    },
    {
      "epoch": 0.34287638935668574,
      "grad_norm": 0.5484218001365662,
      "learning_rate": 1.2465793761426005e-05,
      "loss": 0.3059,
      "step": 509
    },
    {
      "epoch": 0.3435500168406871,
      "grad_norm": 0.5747763514518738,
      "learning_rate": 1.24525533832334e-05,
      "loss": 0.2505,
      "step": 510
    },
    {
      "epoch": 0.34422364432468844,
      "grad_norm": 0.5692996382713318,
      "learning_rate": 1.2439285578806678e-05,
      "loss": 0.3077,
      "step": 511
    },
    {
      "epoch": 0.3448972718086898,
      "grad_norm": 0.5282084345817566,
      "learning_rate": 1.2425990421620235e-05,
      "loss": 0.2763,
      "step": 512
    },
    {
      "epoch": 0.34557089929269114,
      "grad_norm": 0.4825171232223511,
      "learning_rate": 1.241266798529995e-05,
      "loss": 0.2423,
      "step": 513
    },
    {
      "epoch": 0.3462445267766925,
      "grad_norm": 0.5359032154083252,
      "learning_rate": 1.239931834362277e-05,
      "loss": 0.2796,
      "step": 514
    },
    {
      "epoch": 0.34691815426069383,
      "grad_norm": 0.473827600479126,
      "learning_rate": 1.2385941570516297e-05,
      "loss": 0.2531,
      "step": 515
    },
    {
      "epoch": 0.3475917817446952,
      "grad_norm": 0.4639384150505066,
      "learning_rate": 1.2372537740058382e-05,
      "loss": 0.2326,
      "step": 516
    },
    {
      "epoch": 0.34826540922869653,
      "grad_norm": 0.5909863710403442,
      "learning_rate": 1.2359106926476714e-05,
      "loss": 0.2824,
      "step": 517
    },
    {
      "epoch": 0.3489390367126979,
      "grad_norm": 0.5261175036430359,
      "learning_rate": 1.234564920414841e-05,
      "loss": 0.2757,
      "step": 518
    },
    {
      "epoch": 0.34961266419669923,
      "grad_norm": 0.577748715877533,
      "learning_rate": 1.2332164647599599e-05,
      "loss": 0.2619,
      "step": 519
    },
    {
      "epoch": 0.3502862916807006,
      "grad_norm": 0.5614107251167297,
      "learning_rate": 1.2318653331505015e-05,
      "loss": 0.2928,
      "step": 520
    },
    {
      "epoch": 0.35095991916470193,
      "grad_norm": 0.5660324692726135,
      "learning_rate": 1.2305115330687585e-05,
      "loss": 0.2797,
      "step": 521
    },
    {
      "epoch": 0.3516335466487033,
      "grad_norm": 0.5362821817398071,
      "learning_rate": 1.2291550720117997e-05,
      "loss": 0.2931,
      "step": 522
    },
    {
      "epoch": 0.3523071741327046,
      "grad_norm": 0.5424318909645081,
      "learning_rate": 1.2277959574914317e-05,
      "loss": 0.2709,
      "step": 523
    },
    {
      "epoch": 0.352980801616706,
      "grad_norm": 0.5283873081207275,
      "learning_rate": 1.226434197034154e-05,
      "loss": 0.2478,
      "step": 524
    },
    {
      "epoch": 0.3536544291007073,
      "grad_norm": 0.5451403260231018,
      "learning_rate": 1.2250697981811195e-05,
      "loss": 0.2684,
      "step": 525
    },
    {
      "epoch": 0.3543280565847087,
      "grad_norm": 0.5320309400558472,
      "learning_rate": 1.2237027684880914e-05,
      "loss": 0.2678,
      "step": 526
    },
    {
      "epoch": 0.35500168406871,
      "grad_norm": 0.558335542678833,
      "learning_rate": 1.2223331155254026e-05,
      "loss": 0.2715,
      "step": 527
    },
    {
      "epoch": 0.35567531155271137,
      "grad_norm": 0.5011473298072815,
      "learning_rate": 1.220960846877913e-05,
      "loss": 0.2535,
      "step": 528
    },
    {
      "epoch": 0.3563489390367127,
      "grad_norm": 0.5432257056236267,
      "learning_rate": 1.2195859701449672e-05,
      "loss": 0.2802,
      "step": 529
    },
    {
      "epoch": 0.35702256652071407,
      "grad_norm": 0.5836246013641357,
      "learning_rate": 1.2182084929403531e-05,
      "loss": 0.3088,
      "step": 530
    },
    {
      "epoch": 0.3576961940047154,
      "grad_norm": 0.5858445167541504,
      "learning_rate": 1.2168284228922597e-05,
      "loss": 0.2751,
      "step": 531
    },
    {
      "epoch": 0.35836982148871677,
      "grad_norm": 0.556725800037384,
      "learning_rate": 1.2154457676432344e-05,
      "loss": 0.2693,
      "step": 532
    },
    {
      "epoch": 0.35904344897271806,
      "grad_norm": 0.5822067260742188,
      "learning_rate": 1.2140605348501409e-05,
      "loss": 0.3145,
      "step": 533
    },
    {
      "epoch": 0.3597170764567194,
      "grad_norm": 0.5754439830780029,
      "learning_rate": 1.212672732184117e-05,
      "loss": 0.3009,
      "step": 534
    },
    {
      "epoch": 0.36039070394072076,
      "grad_norm": 0.5826534032821655,
      "learning_rate": 1.2112823673305317e-05,
      "loss": 0.3112,
      "step": 535
    },
    {
      "epoch": 0.3610643314247221,
      "grad_norm": 0.5259435176849365,
      "learning_rate": 1.209889447988943e-05,
      "loss": 0.2572,
      "step": 536
    },
    {
      "epoch": 0.36173795890872346,
      "grad_norm": 0.5303089022636414,
      "learning_rate": 1.2084939818730554e-05,
      "loss": 0.2745,
      "step": 537
    },
    {
      "epoch": 0.3624115863927248,
      "grad_norm": 0.4945959150791168,
      "learning_rate": 1.2070959767106762e-05,
      "loss": 0.2624,
      "step": 538
    },
    {
      "epoch": 0.36308521387672615,
      "grad_norm": 0.5212944149971008,
      "learning_rate": 1.2056954402436743e-05,
      "loss": 0.2367,
      "step": 539
    },
    {
      "epoch": 0.3637588413607275,
      "grad_norm": 0.5474100708961487,
      "learning_rate": 1.2042923802279356e-05,
      "loss": 0.2922,
      "step": 540
    },
    {
      "epoch": 0.36443246884472885,
      "grad_norm": 0.5586138963699341,
      "learning_rate": 1.2028868044333218e-05,
      "loss": 0.2779,
      "step": 541
    },
    {
      "epoch": 0.3651060963287302,
      "grad_norm": 0.4587612450122833,
      "learning_rate": 1.2014787206436256e-05,
      "loss": 0.2291,
      "step": 542
    },
    {
      "epoch": 0.36577972381273155,
      "grad_norm": 0.5979660749435425,
      "learning_rate": 1.200068136656529e-05,
      "loss": 0.2663,
      "step": 543
    },
    {
      "epoch": 0.3664533512967329,
      "grad_norm": 0.5004269480705261,
      "learning_rate": 1.1986550602835595e-05,
      "loss": 0.2325,
      "step": 544
    },
    {
      "epoch": 0.36712697878073425,
      "grad_norm": 0.5056456327438354,
      "learning_rate": 1.1972394993500466e-05,
      "loss": 0.2691,
      "step": 545
    },
    {
      "epoch": 0.3678006062647356,
      "grad_norm": 0.5447576642036438,
      "learning_rate": 1.1958214616950794e-05,
      "loss": 0.272,
      "step": 546
    },
    {
      "epoch": 0.36847423374873695,
      "grad_norm": 0.5720804929733276,
      "learning_rate": 1.1944009551714623e-05,
      "loss": 0.2651,
      "step": 547
    },
    {
      "epoch": 0.3691478612327383,
      "grad_norm": 0.5342965722084045,
      "learning_rate": 1.1929779876456713e-05,
      "loss": 0.2681,
      "step": 548
    },
    {
      "epoch": 0.36982148871673964,
      "grad_norm": 0.5355931520462036,
      "learning_rate": 1.191552566997812e-05,
      "loss": 0.2504,
      "step": 549
    },
    {
      "epoch": 0.370495116200741,
      "grad_norm": 0.6217589378356934,
      "learning_rate": 1.1901247011215733e-05,
      "loss": 0.2704,
      "step": 550
    },
    {
      "epoch": 0.37116874368474234,
      "grad_norm": 0.6108464002609253,
      "learning_rate": 1.1886943979241874e-05,
      "loss": 0.2995,
      "step": 551
    },
    {
      "epoch": 0.3718423711687437,
      "grad_norm": 0.5349010229110718,
      "learning_rate": 1.187261665326382e-05,
      "loss": 0.2873,
      "step": 552
    },
    {
      "epoch": 0.37251599865274504,
      "grad_norm": 0.5306320786476135,
      "learning_rate": 1.1858265112623388e-05,
      "loss": 0.2546,
      "step": 553
    },
    {
      "epoch": 0.3731896261367464,
      "grad_norm": 0.5984854102134705,
      "learning_rate": 1.18438894367965e-05,
      "loss": 0.3019,
      "step": 554
    },
    {
      "epoch": 0.37386325362074774,
      "grad_norm": 0.5498750805854797,
      "learning_rate": 1.1829489705392727e-05,
      "loss": 0.2702,
      "step": 555
    },
    {
      "epoch": 0.3745368811047491,
      "grad_norm": 0.5973288416862488,
      "learning_rate": 1.1815065998154849e-05,
      "loss": 0.2947,
      "step": 556
    },
    {
      "epoch": 0.37521050858875044,
      "grad_norm": 0.5865532755851746,
      "learning_rate": 1.180061839495843e-05,
      "loss": 0.3207,
      "step": 557
    },
    {
      "epoch": 0.3758841360727518,
      "grad_norm": 0.5075846314430237,
      "learning_rate": 1.1786146975811359e-05,
      "loss": 0.2474,
      "step": 558
    },
    {
      "epoch": 0.37655776355675313,
      "grad_norm": 0.5501227378845215,
      "learning_rate": 1.1771651820853417e-05,
      "loss": 0.274,
      "step": 559
    },
    {
      "epoch": 0.3772313910407545,
      "grad_norm": 0.5292581915855408,
      "learning_rate": 1.1757133010355821e-05,
      "loss": 0.2546,
      "step": 560
    },
    {
      "epoch": 0.37790501852475583,
      "grad_norm": 0.5926501750946045,
      "learning_rate": 1.1742590624720796e-05,
      "loss": 0.2847,
      "step": 561
    },
    {
      "epoch": 0.3785786460087572,
      "grad_norm": 0.5264430046081543,
      "learning_rate": 1.1728024744481117e-05,
      "loss": 0.2634,
      "step": 562
    },
    {
      "epoch": 0.37925227349275853,
      "grad_norm": 0.5014563798904419,
      "learning_rate": 1.171343545029967e-05,
      "loss": 0.2301,
      "step": 563
    },
    {
      "epoch": 0.3799259009767599,
      "grad_norm": 0.48584073781967163,
      "learning_rate": 1.1698822822969001e-05,
      "loss": 0.2482,
      "step": 564
    },
    {
      "epoch": 0.38059952846076117,
      "grad_norm": 0.5884197354316711,
      "learning_rate": 1.1684186943410867e-05,
      "loss": 0.286,
      "step": 565
    },
    {
      "epoch": 0.3812731559447625,
      "grad_norm": 0.556430459022522,
      "learning_rate": 1.16695278926758e-05,
      "loss": 0.2496,
      "step": 566
    },
    {
      "epoch": 0.38194678342876387,
      "grad_norm": 0.5392268300056458,
      "learning_rate": 1.165484575194264e-05,
      "loss": 0.2786,
      "step": 567
    },
    {
      "epoch": 0.3826204109127652,
      "grad_norm": 0.5491148233413696,
      "learning_rate": 1.1640140602518102e-05,
      "loss": 0.2289,
      "step": 568
    },
    {
      "epoch": 0.38329403839676657,
      "grad_norm": 0.5565954446792603,
      "learning_rate": 1.162541252583631e-05,
      "loss": 0.2614,
      "step": 569
    },
    {
      "epoch": 0.3839676658807679,
      "grad_norm": 0.5307971239089966,
      "learning_rate": 1.1610661603458363e-05,
      "loss": 0.2577,
      "step": 570
    },
    {
      "epoch": 0.38464129336476927,
      "grad_norm": 0.5446802377700806,
      "learning_rate": 1.159588791707187e-05,
      "loss": 0.292,
      "step": 571
    },
    {
      "epoch": 0.3853149208487706,
      "grad_norm": 0.5837084054946899,
      "learning_rate": 1.1581091548490505e-05,
      "loss": 0.2771,
      "step": 572
    },
    {
      "epoch": 0.38598854833277196,
      "grad_norm": 0.5611515045166016,
      "learning_rate": 1.156627257965355e-05,
      "loss": 0.2602,
      "step": 573
    },
    {
      "epoch": 0.3866621758167733,
      "grad_norm": 0.5338358879089355,
      "learning_rate": 1.155143109262544e-05,
      "loss": 0.2573,
      "step": 574
    },
    {
      "epoch": 0.38733580330077466,
      "grad_norm": 0.4791894853115082,
      "learning_rate": 1.1536567169595316e-05,
      "loss": 0.2411,
      "step": 575
    },
    {
      "epoch": 0.388009430784776,
      "grad_norm": 0.5701311826705933,
      "learning_rate": 1.1521680892876563e-05,
      "loss": 0.2973,
      "step": 576
    },
    {
      "epoch": 0.38868305826877736,
      "grad_norm": 0.4976153075695038,
      "learning_rate": 1.1506772344906356e-05,
      "loss": 0.2716,
      "step": 577
    },
    {
      "epoch": 0.3893566857527787,
      "grad_norm": 0.5492983460426331,
      "learning_rate": 1.1491841608245204e-05,
      "loss": 0.2621,
      "step": 578
    },
    {
      "epoch": 0.39003031323678006,
      "grad_norm": 0.5490813255310059,
      "learning_rate": 1.1476888765576493e-05,
      "loss": 0.2687,
      "step": 579
    },
    {
      "epoch": 0.3907039407207814,
      "grad_norm": 0.5402075052261353,
      "learning_rate": 1.1461913899706025e-05,
      "loss": 0.3112,
      "step": 580
    },
    {
      "epoch": 0.39137756820478276,
      "grad_norm": 0.5017600059509277,
      "learning_rate": 1.1446917093561564e-05,
      "loss": 0.2242,
      "step": 581
    },
    {
      "epoch": 0.3920511956887841,
      "grad_norm": 0.5590758919715881,
      "learning_rate": 1.1431898430192375e-05,
      "loss": 0.2569,
      "step": 582
    },
    {
      "epoch": 0.39272482317278545,
      "grad_norm": 0.5497624278068542,
      "learning_rate": 1.1416857992768764e-05,
      "loss": 0.3114,
      "step": 583
    },
    {
      "epoch": 0.3933984506567868,
      "grad_norm": 0.5833696126937866,
      "learning_rate": 1.1401795864581616e-05,
      "loss": 0.2999,
      "step": 584
    },
    {
      "epoch": 0.39407207814078815,
      "grad_norm": 0.5114924907684326,
      "learning_rate": 1.1386712129041937e-05,
      "loss": 0.2428,
      "step": 585
    },
    {
      "epoch": 0.3947457056247895,
      "grad_norm": 0.5477609038352966,
      "learning_rate": 1.1371606869680388e-05,
      "loss": 0.2722,
      "step": 586
    },
    {
      "epoch": 0.39541933310879085,
      "grad_norm": 0.5121515393257141,
      "learning_rate": 1.1356480170146826e-05,
      "loss": 0.2376,
      "step": 587
    },
    {
      "epoch": 0.3960929605927922,
      "grad_norm": 0.502560019493103,
      "learning_rate": 1.1341332114209838e-05,
      "loss": 0.2737,
      "step": 588
    },
    {
      "epoch": 0.39676658807679355,
      "grad_norm": 0.5239719748497009,
      "learning_rate": 1.1326162785756281e-05,
      "loss": 0.2563,
      "step": 589
    },
    {
      "epoch": 0.3974402155607949,
      "grad_norm": 0.5645294189453125,
      "learning_rate": 1.131097226879081e-05,
      "loss": 0.308,
      "step": 590
    },
    {
      "epoch": 0.39811384304479625,
      "grad_norm": 0.5425258278846741,
      "learning_rate": 1.1295760647435424e-05,
      "loss": 0.2388,
      "step": 591
    },
    {
      "epoch": 0.3987874705287976,
      "grad_norm": 0.5374796390533447,
      "learning_rate": 1.1280528005928988e-05,
      "loss": 0.2774,
      "step": 592
    },
    {
      "epoch": 0.39946109801279894,
      "grad_norm": 0.5628758072853088,
      "learning_rate": 1.1265274428626775e-05,
      "loss": 0.2689,
      "step": 593
    },
    {
      "epoch": 0.4001347254968003,
      "grad_norm": 0.5226148366928101,
      "learning_rate": 1.125e-05,
      "loss": 0.2713,
      "step": 594
    },
    {
      "epoch": 0.40080835298080164,
      "grad_norm": 0.5630069971084595,
      "learning_rate": 1.1234704804635342e-05,
      "loss": 0.3279,
      "step": 595
    },
    {
      "epoch": 0.401481980464803,
      "grad_norm": 0.508704423904419,
      "learning_rate": 1.1219388927234482e-05,
      "loss": 0.2623,
      "step": 596
    },
    {
      "epoch": 0.40215560794880434,
      "grad_norm": 0.5345742702484131,
      "learning_rate": 1.1204052452613638e-05,
      "loss": 0.2865,
      "step": 597
    },
    {
      "epoch": 0.40282923543280563,
      "grad_norm": 0.5258358120918274,
      "learning_rate": 1.1188695465703092e-05,
      "loss": 0.2721,
      "step": 598
    },
    {
      "epoch": 0.403502862916807,
      "grad_norm": 0.5306556820869446,
      "learning_rate": 1.1173318051546713e-05,
      "loss": 0.2753,
      "step": 599
    },
    {
      "epoch": 0.40417649040080833,
      "grad_norm": 0.49859175086021423,
      "learning_rate": 1.1157920295301498e-05,
      "loss": 0.2594,
      "step": 600
    },
    {
      "epoch": 0.40417649040080833,
      "eval_loss": 0.2652011811733246,
      "eval_runtime": 105.8884,
      "eval_samples_per_second": 47.22,
      "eval_steps_per_second": 2.956,
      "step": 600
    },
    {
      "epoch": 0.4048501178848097,
      "grad_norm": 0.558407723903656,
      "learning_rate": 1.114250228223709e-05,
      "loss": 0.256,
      "step": 601
    },
    {
      "epoch": 0.40552374536881103,
      "grad_norm": 0.508040726184845,
      "learning_rate": 1.1127064097735315e-05,
      "loss": 0.2575,
      "step": 602
    },
    {
      "epoch": 0.4061973728528124,
      "grad_norm": 0.5474634766578674,
      "learning_rate": 1.1111605827289698e-05,
      "loss": 0.2805,
      "step": 603
    },
    {
      "epoch": 0.4068710003368137,
      "grad_norm": 0.519263505935669,
      "learning_rate": 1.1096127556505e-05,
      "loss": 0.2534,
      "step": 604
    },
    {
      "epoch": 0.4075446278208151,
      "grad_norm": 0.5802994966506958,
      "learning_rate": 1.1080629371096738e-05,
      "loss": 0.2756,
      "step": 605
    },
    {
      "epoch": 0.4082182553048164,
      "grad_norm": 0.5730322599411011,
      "learning_rate": 1.1065111356890712e-05,
      "loss": 0.2888,
      "step": 606
    },
    {
      "epoch": 0.4088918827888178,
      "grad_norm": 0.5447918176651001,
      "learning_rate": 1.1049573599822537e-05,
      "loss": 0.2848,
      "step": 607
    },
    {
      "epoch": 0.4095655102728191,
      "grad_norm": 0.5072281360626221,
      "learning_rate": 1.1034016185937149e-05,
      "loss": 0.2972,
      "step": 608
    },
    {
      "epoch": 0.41023913775682047,
      "grad_norm": 0.6098499298095703,
      "learning_rate": 1.1018439201388346e-05,
      "loss": 0.299,
      "step": 609
    },
    {
      "epoch": 0.4109127652408218,
      "grad_norm": 0.594445526599884,
      "learning_rate": 1.1002842732438301e-05,
      "loss": 0.2778,
      "step": 610
    },
    {
      "epoch": 0.41158639272482317,
      "grad_norm": 0.5406931638717651,
      "learning_rate": 1.0987226865457091e-05,
      "loss": 0.2948,
      "step": 611
    },
    {
      "epoch": 0.4122600202088245,
      "grad_norm": 0.5487210750579834,
      "learning_rate": 1.0971591686922211e-05,
      "loss": 0.256,
      "step": 612
    },
    {
      "epoch": 0.41293364769282587,
      "grad_norm": 0.5063245296478271,
      "learning_rate": 1.0955937283418104e-05,
      "loss": 0.2481,
      "step": 613
    },
    {
      "epoch": 0.4136072751768272,
      "grad_norm": 0.5232447981834412,
      "learning_rate": 1.0940263741635678e-05,
      "loss": 0.2436,
      "step": 614
    },
    {
      "epoch": 0.41428090266082856,
      "grad_norm": 0.5449836254119873,
      "learning_rate": 1.092457114837182e-05,
      "loss": 0.2621,
      "step": 615
    },
    {
      "epoch": 0.4149545301448299,
      "grad_norm": 0.5582854151725769,
      "learning_rate": 1.090885959052892e-05,
      "loss": 0.2885,
      "step": 616
    },
    {
      "epoch": 0.41562815762883126,
      "grad_norm": 0.5433541536331177,
      "learning_rate": 1.0893129155114396e-05,
      "loss": 0.2659,
      "step": 617
    },
    {
      "epoch": 0.4163017851128326,
      "grad_norm": 0.5937801599502563,
      "learning_rate": 1.0877379929240198e-05,
      "loss": 0.2968,
      "step": 618
    },
    {
      "epoch": 0.41697541259683396,
      "grad_norm": 0.4904331564903259,
      "learning_rate": 1.0861612000122341e-05,
      "loss": 0.2508,
      "step": 619
    },
    {
      "epoch": 0.4176490400808353,
      "grad_norm": 0.5370484590530396,
      "learning_rate": 1.0845825455080411e-05,
      "loss": 0.2564,
      "step": 620
    },
    {
      "epoch": 0.41832266756483666,
      "grad_norm": 0.535376250743866,
      "learning_rate": 1.0830020381537088e-05,
      "loss": 0.2796,
      "step": 621
    },
    {
      "epoch": 0.418996295048838,
      "grad_norm": 0.5508119463920593,
      "learning_rate": 1.0814196867017656e-05,
      "loss": 0.281,
      "step": 622
    },
    {
      "epoch": 0.41966992253283936,
      "grad_norm": 0.525283694267273,
      "learning_rate": 1.079835499914952e-05,
      "loss": 0.2306,
      "step": 623
    },
    {
      "epoch": 0.4203435500168407,
      "grad_norm": 0.5157189965248108,
      "learning_rate": 1.078249486566173e-05,
      "loss": 0.2679,
      "step": 624
    },
    {
      "epoch": 0.42101717750084205,
      "grad_norm": 0.6008614301681519,
      "learning_rate": 1.0766616554384477e-05,
      "loss": 0.2815,
      "step": 625
    },
    {
      "epoch": 0.4216908049848434,
      "grad_norm": 0.5147749185562134,
      "learning_rate": 1.0750720153248626e-05,
      "loss": 0.2587,
      "step": 626
    },
    {
      "epoch": 0.42236443246884475,
      "grad_norm": 0.5508129596710205,
      "learning_rate": 1.073480575028521e-05,
      "loss": 0.2788,
      "step": 627
    },
    {
      "epoch": 0.4230380599528461,
      "grad_norm": 0.5465036034584045,
      "learning_rate": 1.0718873433624966e-05,
      "loss": 0.2606,
      "step": 628
    },
    {
      "epoch": 0.42371168743684745,
      "grad_norm": 0.5761625170707703,
      "learning_rate": 1.070292329149782e-05,
      "loss": 0.3149,
      "step": 629
    },
    {
      "epoch": 0.42438531492084874,
      "grad_norm": 0.5194136500358582,
      "learning_rate": 1.0686955412232419e-05,
      "loss": 0.2305,
      "step": 630
    },
    {
      "epoch": 0.4250589424048501,
      "grad_norm": 0.5823161602020264,
      "learning_rate": 1.0670969884255636e-05,
      "loss": 0.2495,
      "step": 631
    },
    {
      "epoch": 0.42573256988885144,
      "grad_norm": 0.5550847053527832,
      "learning_rate": 1.0654966796092073e-05,
      "loss": 0.2539,
      "step": 632
    },
    {
      "epoch": 0.4264061973728528,
      "grad_norm": 0.5327949523925781,
      "learning_rate": 1.0638946236363578e-05,
      "loss": 0.2655,
      "step": 633
    },
    {
      "epoch": 0.42707982485685414,
      "grad_norm": 0.5146956443786621,
      "learning_rate": 1.0622908293788758e-05,
      "loss": 0.2599,
      "step": 634
    },
    {
      "epoch": 0.4277534523408555,
      "grad_norm": 0.5790160894393921,
      "learning_rate": 1.0606853057182481e-05,
      "loss": 0.298,
      "step": 635
    },
    {
      "epoch": 0.42842707982485684,
      "grad_norm": 0.5627730488777161,
      "learning_rate": 1.059078061545538e-05,
      "loss": 0.2622,
      "step": 636
    },
    {
      "epoch": 0.4291007073088582,
      "grad_norm": 0.619365394115448,
      "learning_rate": 1.0574691057613376e-05,
      "loss": 0.2905,
      "step": 637
    },
    {
      "epoch": 0.42977433479285954,
      "grad_norm": 0.5521032810211182,
      "learning_rate": 1.0558584472757167e-05,
      "loss": 0.2705,
      "step": 638
    },
    {
      "epoch": 0.4304479622768609,
      "grad_norm": 0.5045711398124695,
      "learning_rate": 1.0542460950081747e-05,
      "loss": 0.2289,
      "step": 639
    },
    {
      "epoch": 0.43112158976086223,
      "grad_norm": 0.5129411816596985,
      "learning_rate": 1.0526320578875909e-05,
      "loss": 0.2572,
      "step": 640
    },
    {
      "epoch": 0.4317952172448636,
      "grad_norm": 0.5294272899627686,
      "learning_rate": 1.0510163448521747e-05,
      "loss": 0.2702,
      "step": 641
    },
    {
      "epoch": 0.43246884472886493,
      "grad_norm": 0.5448393225669861,
      "learning_rate": 1.0493989648494165e-05,
      "loss": 0.2808,
      "step": 642
    },
    {
      "epoch": 0.4331424722128663,
      "grad_norm": 0.5107436776161194,
      "learning_rate": 1.0477799268360384e-05,
      "loss": 0.248,
      "step": 643
    },
    {
      "epoch": 0.43381609969686763,
      "grad_norm": 0.5598347187042236,
      "learning_rate": 1.0461592397779435e-05,
      "loss": 0.2342,
      "step": 644
    },
    {
      "epoch": 0.434489727180869,
      "grad_norm": 0.5707139372825623,
      "learning_rate": 1.0445369126501676e-05,
      "loss": 0.2764,
      "step": 645
    },
    {
      "epoch": 0.4351633546648703,
      "grad_norm": 0.48345211148262024,
      "learning_rate": 1.0429129544368283e-05,
      "loss": 0.2215,
      "step": 646
    },
    {
      "epoch": 0.4358369821488717,
      "grad_norm": 0.5131022930145264,
      "learning_rate": 1.0412873741310763e-05,
      "loss": 0.2423,
      "step": 647
    },
    {
      "epoch": 0.436510609632873,
      "grad_norm": 0.5428949594497681,
      "learning_rate": 1.0396601807350452e-05,
      "loss": 0.2331,
      "step": 648
    },
    {
      "epoch": 0.4371842371168744,
      "grad_norm": 0.47753867506980896,
      "learning_rate": 1.038031383259801e-05,
      "loss": 0.2552,
      "step": 649
    },
    {
      "epoch": 0.4378578646008757,
      "grad_norm": 0.48779332637786865,
      "learning_rate": 1.0364009907252937e-05,
      "loss": 0.2499,
      "step": 650
    },
    {
      "epoch": 0.4385314920848771,
      "grad_norm": 0.4910006523132324,
      "learning_rate": 1.0347690121603047e-05,
      "loss": 0.2498,
      "step": 651
    },
    {
      "epoch": 0.4392051195688784,
      "grad_norm": 0.5575456023216248,
      "learning_rate": 1.0331354566024005e-05,
      "loss": 0.2503,
      "step": 652
    },
    {
      "epoch": 0.43987874705287977,
      "grad_norm": 0.5806515216827393,
      "learning_rate": 1.0315003330978799e-05,
      "loss": 0.254,
      "step": 653
    },
    {
      "epoch": 0.4405523745368811,
      "grad_norm": 0.5564923882484436,
      "learning_rate": 1.0298636507017241e-05,
      "loss": 0.2804,
      "step": 654
    },
    {
      "epoch": 0.44122600202088247,
      "grad_norm": 0.5716164708137512,
      "learning_rate": 1.0282254184775473e-05,
      "loss": 0.2844,
      "step": 655
    },
    {
      "epoch": 0.4418996295048838,
      "grad_norm": 0.5606719255447388,
      "learning_rate": 1.0265856454975473e-05,
      "loss": 0.2576,
      "step": 656
    },
    {
      "epoch": 0.44257325698888517,
      "grad_norm": 0.5467285513877869,
      "learning_rate": 1.0249443408424535e-05,
      "loss": 0.2782,
      "step": 657
    },
    {
      "epoch": 0.4432468844728865,
      "grad_norm": 0.569665253162384,
      "learning_rate": 1.0233015136014773e-05,
      "loss": 0.272,
      "step": 658
    },
    {
      "epoch": 0.44392051195688786,
      "grad_norm": 0.5965842604637146,
      "learning_rate": 1.021657172872262e-05,
      "loss": 0.3023,
      "step": 659
    },
    {
      "epoch": 0.4445941394408892,
      "grad_norm": 0.5759636163711548,
      "learning_rate": 1.0200113277608326e-05,
      "loss": 0.2621,
      "step": 660
    },
    {
      "epoch": 0.44526776692489056,
      "grad_norm": 0.5999960899353027,
      "learning_rate": 1.0183639873815448e-05,
      "loss": 0.2976,
      "step": 661
    },
    {
      "epoch": 0.44594139440889186,
      "grad_norm": 0.5440315008163452,
      "learning_rate": 1.0167151608570346e-05,
      "loss": 0.2889,
      "step": 662
    },
    {
      "epoch": 0.4466150218928932,
      "grad_norm": 0.4932374358177185,
      "learning_rate": 1.0150648573181685e-05,
      "loss": 0.2271,
      "step": 663
    },
    {
      "epoch": 0.44728864937689455,
      "grad_norm": 0.5871284604072571,
      "learning_rate": 1.0134130859039921e-05,
      "loss": 0.3202,
      "step": 664
    },
    {
      "epoch": 0.4479622768608959,
      "grad_norm": 0.5287674069404602,
      "learning_rate": 1.0117598557616796e-05,
      "loss": 0.2486,
      "step": 665
    },
    {
      "epoch": 0.44863590434489725,
      "grad_norm": 0.588444709777832,
      "learning_rate": 1.0101051760464837e-05,
      "loss": 0.2555,
      "step": 666
    },
    {
      "epoch": 0.4493095318288986,
      "grad_norm": 0.5376453399658203,
      "learning_rate": 1.0084490559216843e-05,
      "loss": 0.2506,
      "step": 667
    },
    {
      "epoch": 0.44998315931289995,
      "grad_norm": 0.5496957898139954,
      "learning_rate": 1.006791504558538e-05,
      "loss": 0.2616,
      "step": 668
    },
    {
      "epoch": 0.4506567867969013,
      "grad_norm": 0.523008406162262,
      "learning_rate": 1.0051325311362278e-05,
      "loss": 0.2597,
      "step": 669
    },
    {
      "epoch": 0.45133041428090265,
      "grad_norm": 0.5686816573143005,
      "learning_rate": 1.0034721448418105e-05,
      "loss": 0.2665,
      "step": 670
    },
    {
      "epoch": 0.452004041764904,
      "grad_norm": 0.5065593719482422,
      "learning_rate": 1.0018103548701688e-05,
      "loss": 0.2566,
      "step": 671
    },
    {
      "epoch": 0.45267766924890535,
      "grad_norm": 0.5687103867530823,
      "learning_rate": 1.0001471704239577e-05,
      "loss": 0.2628,
      "step": 672
    },
    {
      "epoch": 0.4533512967329067,
      "grad_norm": 0.5782075524330139,
      "learning_rate": 9.984826007135544e-06,
      "loss": 0.2732,
      "step": 673
    },
    {
      "epoch": 0.45402492421690804,
      "grad_norm": 0.5679803490638733,
      "learning_rate": 9.968166549570075e-06,
      "loss": 0.2664,
      "step": 674
    },
    {
      "epoch": 0.4546985517009094,
      "grad_norm": 0.5293748378753662,
      "learning_rate": 9.951493423799866e-06,
      "loss": 0.2498,
      "step": 675
    },
    {
      "epoch": 0.45537217918491074,
      "grad_norm": 0.5444015264511108,
      "learning_rate": 9.934806722157294e-06,
      "loss": 0.2549,
      "step": 676
    },
    {
      "epoch": 0.4560458066689121,
      "grad_norm": 0.5367648601531982,
      "learning_rate": 9.918106537049921e-06,
      "loss": 0.2623,
      "step": 677
    },
    {
      "epoch": 0.45671943415291344,
      "grad_norm": 0.5820662975311279,
      "learning_rate": 9.901392960959983e-06,
      "loss": 0.2771,
      "step": 678
    },
    {
      "epoch": 0.4573930616369148,
      "grad_norm": 0.5573861598968506,
      "learning_rate": 9.884666086443862e-06,
      "loss": 0.2614,
      "step": 679
    },
    {
      "epoch": 0.45806668912091614,
      "grad_norm": 0.6296043992042542,
      "learning_rate": 9.867926006131597e-06,
      "loss": 0.3102,
      "step": 680
    },
    {
      "epoch": 0.4587403166049175,
      "grad_norm": 0.5795363187789917,
      "learning_rate": 9.851172812726344e-06,
      "loss": 0.3059,
      "step": 681
    },
    {
      "epoch": 0.45941394408891884,
      "grad_norm": 0.48046785593032837,
      "learning_rate": 9.834406599003885e-06,
      "loss": 0.2323,
      "step": 682
    },
    {
      "epoch": 0.4600875715729202,
      "grad_norm": 0.4878872036933899,
      "learning_rate": 9.817627457812105e-06,
      "loss": 0.2467,
      "step": 683
    },
    {
      "epoch": 0.46076119905692153,
      "grad_norm": 0.5333375334739685,
      "learning_rate": 9.800835482070479e-06,
      "loss": 0.2282,
      "step": 684
    },
    {
      "epoch": 0.4614348265409229,
      "grad_norm": 0.543725848197937,
      "learning_rate": 9.784030764769553e-06,
      "loss": 0.2427,
      "step": 685
    },
    {
      "epoch": 0.46210845402492423,
      "grad_norm": 0.5145445466041565,
      "learning_rate": 9.76721339897044e-06,
      "loss": 0.2291,
      "step": 686
    },
    {
      "epoch": 0.4627820815089256,
      "grad_norm": 0.5099066495895386,
      "learning_rate": 9.75038347780429e-06,
      "loss": 0.245,
      "step": 687
    },
    {
      "epoch": 0.46345570899292693,
      "grad_norm": 0.5599386096000671,
      "learning_rate": 9.73354109447179e-06,
      "loss": 0.2994,
      "step": 688
    },
    {
      "epoch": 0.4641293364769283,
      "grad_norm": 0.5298258662223816,
      "learning_rate": 9.716686342242632e-06,
      "loss": 0.231,
      "step": 689
    },
    {
      "epoch": 0.4648029639609296,
      "grad_norm": 0.5349884033203125,
      "learning_rate": 9.69981931445501e-06,
      "loss": 0.2436,
      "step": 690
    },
    {
      "epoch": 0.465476591444931,
      "grad_norm": 0.5078858137130737,
      "learning_rate": 9.682940104515097e-06,
      "loss": 0.2735,
      "step": 691
    },
    {
      "epoch": 0.4661502189289323,
      "grad_norm": 0.5433405637741089,
      "learning_rate": 9.666048805896524e-06,
      "loss": 0.2472,
      "step": 692
    },
    {
      "epoch": 0.4668238464129337,
      "grad_norm": 0.5337989926338196,
      "learning_rate": 9.649145512139876e-06,
      "loss": 0.2815,
      "step": 693
    },
    {
      "epoch": 0.46749747389693497,
      "grad_norm": 0.491817831993103,
      "learning_rate": 9.632230316852153e-06,
      "loss": 0.2712,
      "step": 694
    },
    {
      "epoch": 0.4681711013809363,
      "grad_norm": 0.5814330577850342,
      "learning_rate": 9.615303313706271e-06,
      "loss": 0.2931,
      "step": 695
    },
    {
      "epoch": 0.46884472886493767,
      "grad_norm": 0.5358330607414246,
      "learning_rate": 9.598364596440534e-06,
      "loss": 0.2546,
      "step": 696
    },
    {
      "epoch": 0.469518356348939,
      "grad_norm": 0.5111145377159119,
      "learning_rate": 9.581414258858116e-06,
      "loss": 0.2607,
      "step": 697
    },
    {
      "epoch": 0.47019198383294036,
      "grad_norm": 0.5266521573066711,
      "learning_rate": 9.564452394826538e-06,
      "loss": 0.2554,
      "step": 698
    },
    {
      "epoch": 0.4708656113169417,
      "grad_norm": 0.5091780424118042,
      "learning_rate": 9.54747909827716e-06,
      "loss": 0.2723,
      "step": 699
    },
    {
      "epoch": 0.47153923880094306,
      "grad_norm": 0.5414915680885315,
      "learning_rate": 9.530494463204646e-06,
      "loss": 0.2577,
      "step": 700
    },
    {
      "epoch": 0.47153923880094306,
      "eval_loss": 0.26179420948028564,
      "eval_runtime": 105.0708,
      "eval_samples_per_second": 47.587,
      "eval_steps_per_second": 2.979,
      "step": 700
    },
    {
      "epoch": 0.4722128662849444,
      "grad_norm": 0.505789577960968,
      "learning_rate": 9.513498583666456e-06,
      "loss": 0.2448,
      "step": 701
    },
    {
      "epoch": 0.47288649376894576,
      "grad_norm": 0.46454617381095886,
      "learning_rate": 9.496491553782314e-06,
      "loss": 0.221,
      "step": 702
    },
    {
      "epoch": 0.4735601212529471,
      "grad_norm": 0.5358849763870239,
      "learning_rate": 9.479473467733697e-06,
      "loss": 0.2872,
      "step": 703
    },
    {
      "epoch": 0.47423374873694846,
      "grad_norm": 0.5496987700462341,
      "learning_rate": 9.462444419763306e-06,
      "loss": 0.2464,
      "step": 704
    },
    {
      "epoch": 0.4749073762209498,
      "grad_norm": 0.5485591292381287,
      "learning_rate": 9.445404504174546e-06,
      "loss": 0.2695,
      "step": 705
    },
    {
      "epoch": 0.47558100370495116,
      "grad_norm": 0.5437228679656982,
      "learning_rate": 9.42835381533101e-06,
      "loss": 0.2823,
      "step": 706
    },
    {
      "epoch": 0.4762546311889525,
      "grad_norm": 0.5094515085220337,
      "learning_rate": 9.411292447655948e-06,
      "loss": 0.2401,
      "step": 707
    },
    {
      "epoch": 0.47692825867295385,
      "grad_norm": 0.5395442843437195,
      "learning_rate": 9.394220495631744e-06,
      "loss": 0.2659,
      "step": 708
    },
    {
      "epoch": 0.4776018861569552,
      "grad_norm": 0.4930800795555115,
      "learning_rate": 9.377138053799399e-06,
      "loss": 0.2383,
      "step": 709
    },
    {
      "epoch": 0.47827551364095655,
      "grad_norm": 0.5237337350845337,
      "learning_rate": 9.360045216758008e-06,
      "loss": 0.2527,
      "step": 710
    },
    {
      "epoch": 0.4789491411249579,
      "grad_norm": 0.5243161916732788,
      "learning_rate": 9.342942079164223e-06,
      "loss": 0.2515,
      "step": 711
    },
    {
      "epoch": 0.47962276860895925,
      "grad_norm": 0.5414012670516968,
      "learning_rate": 9.325828735731747e-06,
      "loss": 0.275,
      "step": 712
    },
    {
      "epoch": 0.4802963960929606,
      "grad_norm": 0.547073245048523,
      "learning_rate": 9.308705281230796e-06,
      "loss": 0.276,
      "step": 713
    },
    {
      "epoch": 0.48097002357696195,
      "grad_norm": 0.49008458852767944,
      "learning_rate": 9.291571810487584e-06,
      "loss": 0.246,
      "step": 714
    },
    {
      "epoch": 0.4816436510609633,
      "grad_norm": 0.5415433645248413,
      "learning_rate": 9.27442841838379e-06,
      "loss": 0.2658,
      "step": 715
    },
    {
      "epoch": 0.48231727854496464,
      "grad_norm": 0.5856931209564209,
      "learning_rate": 9.257275199856032e-06,
      "loss": 0.2675,
      "step": 716
    },
    {
      "epoch": 0.482990906028966,
      "grad_norm": 0.5154370665550232,
      "learning_rate": 9.24011224989535e-06,
      "loss": 0.2422,
      "step": 717
    },
    {
      "epoch": 0.48366453351296734,
      "grad_norm": 0.5306107401847839,
      "learning_rate": 9.222939663546677e-06,
      "loss": 0.2687,
      "step": 718
    },
    {
      "epoch": 0.4843381609969687,
      "grad_norm": 0.4880635142326355,
      "learning_rate": 9.2057575359083e-06,
      "loss": 0.2276,
      "step": 719
    },
    {
      "epoch": 0.48501178848097004,
      "grad_norm": 0.6055603623390198,
      "learning_rate": 9.18856596213135e-06,
      "loss": 0.2907,
      "step": 720
    },
    {
      "epoch": 0.4856854159649714,
      "grad_norm": 0.5602757930755615,
      "learning_rate": 9.171365037419272e-06,
      "loss": 0.2511,
      "step": 721
    },
    {
      "epoch": 0.48635904344897274,
      "grad_norm": 0.5492405295372009,
      "learning_rate": 9.15415485702729e-06,
      "loss": 0.246,
      "step": 722
    },
    {
      "epoch": 0.4870326709329741,
      "grad_norm": 0.6091371178627014,
      "learning_rate": 9.136935516261887e-06,
      "loss": 0.3003,
      "step": 723
    },
    {
      "epoch": 0.48770629841697544,
      "grad_norm": 0.5400590300559998,
      "learning_rate": 9.119707110480272e-06,
      "loss": 0.2576,
      "step": 724
    },
    {
      "epoch": 0.4883799259009768,
      "grad_norm": 0.5183984041213989,
      "learning_rate": 9.10246973508985e-06,
      "loss": 0.2519,
      "step": 725
    },
    {
      "epoch": 0.48905355338497813,
      "grad_norm": 0.5791885256767273,
      "learning_rate": 9.08522348554771e-06,
      "loss": 0.269,
      "step": 726
    },
    {
      "epoch": 0.48972718086897943,
      "grad_norm": 0.5196906328201294,
      "learning_rate": 9.067968457360073e-06,
      "loss": 0.2681,
      "step": 727
    },
    {
      "epoch": 0.4904008083529808,
      "grad_norm": 0.5393977165222168,
      "learning_rate": 9.050704746081779e-06,
      "loss": 0.2487,
      "step": 728
    },
    {
      "epoch": 0.4910744358369821,
      "grad_norm": 0.5441868305206299,
      "learning_rate": 9.033432447315751e-06,
      "loss": 0.2603,
      "step": 729
    },
    {
      "epoch": 0.4917480633209835,
      "grad_norm": 0.4999203383922577,
      "learning_rate": 9.016151656712473e-06,
      "loss": 0.2569,
      "step": 730
    },
    {
      "epoch": 0.4924216908049848,
      "grad_norm": 0.5059922933578491,
      "learning_rate": 8.998862469969452e-06,
      "loss": 0.2428,
      "step": 731
    },
    {
      "epoch": 0.4930953182889862,
      "grad_norm": 0.5794141292572021,
      "learning_rate": 8.981564982830683e-06,
      "loss": 0.2901,
      "step": 732
    },
    {
      "epoch": 0.4937689457729875,
      "grad_norm": 0.5344904065132141,
      "learning_rate": 8.964259291086141e-06,
      "loss": 0.278,
      "step": 733
    },
    {
      "epoch": 0.49444257325698887,
      "grad_norm": 0.5577378273010254,
      "learning_rate": 8.946945490571227e-06,
      "loss": 0.2753,
      "step": 734
    },
    {
      "epoch": 0.4951162007409902,
      "grad_norm": 0.48888590931892395,
      "learning_rate": 8.92962367716625e-06,
      "loss": 0.2565,
      "step": 735
    },
    {
      "epoch": 0.49578982822499157,
      "grad_norm": 0.5605798363685608,
      "learning_rate": 8.912293946795895e-06,
      "loss": 0.274,
      "step": 736
    },
    {
      "epoch": 0.4964634557089929,
      "grad_norm": 0.5351974964141846,
      "learning_rate": 8.894956395428685e-06,
      "loss": 0.259,
      "step": 737
    },
    {
      "epoch": 0.49713708319299427,
      "grad_norm": 0.530037522315979,
      "learning_rate": 8.877611119076454e-06,
      "loss": 0.2468,
      "step": 738
    },
    {
      "epoch": 0.4978107106769956,
      "grad_norm": 0.5955355763435364,
      "learning_rate": 8.860258213793819e-06,
      "loss": 0.2702,
      "step": 739
    },
    {
      "epoch": 0.49848433816099696,
      "grad_norm": 0.5594556927680969,
      "learning_rate": 8.842897775677645e-06,
      "loss": 0.2796,
      "step": 740
    },
    {
      "epoch": 0.4991579656449983,
      "grad_norm": 0.5318235158920288,
      "learning_rate": 8.825529900866507e-06,
      "loss": 0.2721,
      "step": 741
    },
    {
      "epoch": 0.49983159312899966,
      "grad_norm": 0.6066297888755798,
      "learning_rate": 8.808154685540164e-06,
      "loss": 0.2814,
      "step": 742
    },
    {
      "epoch": 0.500505220613001,
      "grad_norm": 0.520949125289917,
      "learning_rate": 8.790772225919031e-06,
      "loss": 0.2479,
      "step": 743
    },
    {
      "epoch": 0.5011788480970023,
      "grad_norm": 0.532832682132721,
      "learning_rate": 8.77338261826364e-06,
      "loss": 0.2717,
      "step": 744
    },
    {
      "epoch": 0.5018524755810037,
      "grad_norm": 0.4917290210723877,
      "learning_rate": 8.755985958874096e-06,
      "loss": 0.2331,
      "step": 745
    },
    {
      "epoch": 0.502526103065005,
      "grad_norm": 0.6336959004402161,
      "learning_rate": 8.73858234408957e-06,
      "loss": 0.3059,
      "step": 746
    },
    {
      "epoch": 0.5031997305490064,
      "grad_norm": 0.5722649693489075,
      "learning_rate": 8.72117187028774e-06,
      "loss": 0.2682,
      "step": 747
    },
    {
      "epoch": 0.5038733580330077,
      "grad_norm": 0.47712576389312744,
      "learning_rate": 8.70375463388427e-06,
      "loss": 0.2468,
      "step": 748
    },
    {
      "epoch": 0.504546985517009,
      "grad_norm": 0.49866771697998047,
      "learning_rate": 8.68633073133228e-06,
      "loss": 0.2609,
      "step": 749
    },
    {
      "epoch": 0.5052206130010104,
      "grad_norm": 0.5410306453704834,
      "learning_rate": 8.6689002591218e-06,
      "loss": 0.2733,
      "step": 750
    },
    {
      "epoch": 0.5058942404850117,
      "grad_norm": 0.5518447160720825,
      "learning_rate": 8.651463313779241e-06,
      "loss": 0.2525,
      "step": 751
    },
    {
      "epoch": 0.5065678679690131,
      "grad_norm": 0.5311466455459595,
      "learning_rate": 8.634019991866863e-06,
      "loss": 0.275,
      "step": 752
    },
    {
      "epoch": 0.5072414954530144,
      "grad_norm": 0.5381631255149841,
      "learning_rate": 8.61657038998224e-06,
      "loss": 0.275,
      "step": 753
    },
    {
      "epoch": 0.5079151229370158,
      "grad_norm": 0.48526835441589355,
      "learning_rate": 8.599114604757716e-06,
      "loss": 0.2431,
      "step": 754
    },
    {
      "epoch": 0.5085887504210171,
      "grad_norm": 0.5347431302070618,
      "learning_rate": 8.581652732859887e-06,
      "loss": 0.2731,
      "step": 755
    },
    {
      "epoch": 0.5092623779050185,
      "grad_norm": 0.5098583102226257,
      "learning_rate": 8.56418487098905e-06,
      "loss": 0.294,
      "step": 756
    },
    {
      "epoch": 0.5099360053890198,
      "grad_norm": 0.499496191740036,
      "learning_rate": 8.54671111587867e-06,
      "loss": 0.2294,
      "step": 757
    },
    {
      "epoch": 0.5106096328730212,
      "grad_norm": 0.5586072206497192,
      "learning_rate": 8.529231564294858e-06,
      "loss": 0.2506,
      "step": 758
    },
    {
      "epoch": 0.5112832603570225,
      "grad_norm": 0.5203363299369812,
      "learning_rate": 8.51174631303581e-06,
      "loss": 0.2505,
      "step": 759
    },
    {
      "epoch": 0.5119568878410239,
      "grad_norm": 0.5142697095870972,
      "learning_rate": 8.494255458931304e-06,
      "loss": 0.2456,
      "step": 760
    },
    {
      "epoch": 0.5126305153250252,
      "grad_norm": 0.4652908444404602,
      "learning_rate": 8.476759098842129e-06,
      "loss": 0.2085,
      "step": 761
    },
    {
      "epoch": 0.5133041428090266,
      "grad_norm": 0.5014703273773193,
      "learning_rate": 8.459257329659571e-06,
      "loss": 0.239,
      "step": 762
    },
    {
      "epoch": 0.5139777702930279,
      "grad_norm": 0.5147262215614319,
      "learning_rate": 8.441750248304872e-06,
      "loss": 0.2727,
      "step": 763
    },
    {
      "epoch": 0.5146513977770293,
      "grad_norm": 0.564335823059082,
      "learning_rate": 8.424237951728689e-06,
      "loss": 0.2983,
      "step": 764
    },
    {
      "epoch": 0.5153250252610306,
      "grad_norm": 0.5217107534408569,
      "learning_rate": 8.406720536910568e-06,
      "loss": 0.238,
      "step": 765
    },
    {
      "epoch": 0.515998652745032,
      "grad_norm": 0.529780924320221,
      "learning_rate": 8.389198100858385e-06,
      "loss": 0.271,
      "step": 766
    },
    {
      "epoch": 0.5166722802290333,
      "grad_norm": 0.5005664229393005,
      "learning_rate": 8.371670740607833e-06,
      "loss": 0.265,
      "step": 767
    },
    {
      "epoch": 0.5173459077130347,
      "grad_norm": 0.4695169925689697,
      "learning_rate": 8.354138553221869e-06,
      "loss": 0.225,
      "step": 768
    },
    {
      "epoch": 0.518019535197036,
      "grad_norm": 0.6260945200920105,
      "learning_rate": 8.336601635790184e-06,
      "loss": 0.2725,
      "step": 769
    },
    {
      "epoch": 0.5186931626810374,
      "grad_norm": 0.5363501310348511,
      "learning_rate": 8.319060085428664e-06,
      "loss": 0.2631,
      "step": 770
    },
    {
      "epoch": 0.5193667901650387,
      "grad_norm": 0.5340143442153931,
      "learning_rate": 8.301513999278851e-06,
      "loss": 0.2829,
      "step": 771
    },
    {
      "epoch": 0.5200404176490401,
      "grad_norm": 0.5355620384216309,
      "learning_rate": 8.283963474507402e-06,
      "loss": 0.2675,
      "step": 772
    },
    {
      "epoch": 0.5207140451330414,
      "grad_norm": 0.5030906796455383,
      "learning_rate": 8.266408608305555e-06,
      "loss": 0.2243,
      "step": 773
    },
    {
      "epoch": 0.5213876726170428,
      "grad_norm": 0.5517938137054443,
      "learning_rate": 8.248849497888598e-06,
      "loss": 0.2554,
      "step": 774
    },
    {
      "epoch": 0.5220613001010441,
      "grad_norm": 0.47788354754447937,
      "learning_rate": 8.231286240495305e-06,
      "loss": 0.2258,
      "step": 775
    },
    {
      "epoch": 0.5227349275850455,
      "grad_norm": 0.550268828868866,
      "learning_rate": 8.213718933387438e-06,
      "loss": 0.2586,
      "step": 776
    },
    {
      "epoch": 0.5234085550690468,
      "grad_norm": 0.5247451066970825,
      "learning_rate": 8.196147673849165e-06,
      "loss": 0.2491,
      "step": 777
    },
    {
      "epoch": 0.5240821825530482,
      "grad_norm": 0.49666067957878113,
      "learning_rate": 8.17857255918655e-06,
      "loss": 0.2501,
      "step": 778
    },
    {
      "epoch": 0.5247558100370495,
      "grad_norm": 0.5575336217880249,
      "learning_rate": 8.160993686727015e-06,
      "loss": 0.3047,
      "step": 779
    },
    {
      "epoch": 0.5254294375210509,
      "grad_norm": 0.5327598452568054,
      "learning_rate": 8.143411153818773e-06,
      "loss": 0.289,
      "step": 780
    },
    {
      "epoch": 0.5261030650050522,
      "grad_norm": 0.4978947043418884,
      "learning_rate": 8.125825057830323e-06,
      "loss": 0.2817,
      "step": 781
    },
    {
      "epoch": 0.5267766924890536,
      "grad_norm": 0.5068449378013611,
      "learning_rate": 8.108235496149892e-06,
      "loss": 0.2549,
      "step": 782
    },
    {
      "epoch": 0.5274503199730549,
      "grad_norm": 0.5815426111221313,
      "learning_rate": 8.090642566184896e-06,
      "loss": 0.3215,
      "step": 783
    },
    {
      "epoch": 0.5281239474570563,
      "grad_norm": 0.528716504573822,
      "learning_rate": 8.073046365361404e-06,
      "loss": 0.2405,
      "step": 784
    },
    {
      "epoch": 0.5287975749410576,
      "grad_norm": 0.5129048824310303,
      "learning_rate": 8.0554469911236e-06,
      "loss": 0.2696,
      "step": 785
    },
    {
      "epoch": 0.529471202425059,
      "grad_norm": 0.5234351754188538,
      "learning_rate": 8.037844540933245e-06,
      "loss": 0.2608,
      "step": 786
    },
    {
      "epoch": 0.5301448299090603,
      "grad_norm": 0.531194269657135,
      "learning_rate": 8.020239112269131e-06,
      "loss": 0.2826,
      "step": 787
    },
    {
      "epoch": 0.5308184573930617,
      "grad_norm": 0.5546161532402039,
      "learning_rate": 8.002630802626538e-06,
      "loss": 0.2635,
      "step": 788
    },
    {
      "epoch": 0.531492084877063,
      "grad_norm": 0.5576707124710083,
      "learning_rate": 7.985019709516714e-06,
      "loss": 0.2591,
      "step": 789
    },
    {
      "epoch": 0.5321657123610644,
      "grad_norm": 0.5075989961624146,
      "learning_rate": 7.967405930466305e-06,
      "loss": 0.2751,
      "step": 790
    },
    {
      "epoch": 0.5328393398450657,
      "grad_norm": 0.547538161277771,
      "learning_rate": 7.94978956301685e-06,
      "loss": 0.2767,
      "step": 791
    },
    {
      "epoch": 0.5335129673290671,
      "grad_norm": 0.6105408072471619,
      "learning_rate": 7.932170704724202e-06,
      "loss": 0.3202,
      "step": 792
    },
    {
      "epoch": 0.5341865948130684,
      "grad_norm": 0.517285943031311,
      "learning_rate": 7.914549453158025e-06,
      "loss": 0.2497,
      "step": 793
    },
    {
      "epoch": 0.5348602222970698,
      "grad_norm": 0.5324558615684509,
      "learning_rate": 7.896925905901223e-06,
      "loss": 0.2804,
      "step": 794
    },
    {
      "epoch": 0.5355338497810711,
      "grad_norm": 0.5467241406440735,
      "learning_rate": 7.879300160549423e-06,
      "loss": 0.274,
      "step": 795
    },
    {
      "epoch": 0.5362074772650725,
      "grad_norm": 0.5673408508300781,
      "learning_rate": 7.86167231471042e-06,
      "loss": 0.2681,
      "step": 796
    },
    {
      "epoch": 0.5368811047490738,
      "grad_norm": 0.5435929298400879,
      "learning_rate": 7.844042466003643e-06,
      "loss": 0.2456,
      "step": 797
    },
    {
      "epoch": 0.5375547322330751,
      "grad_norm": 0.5365129113197327,
      "learning_rate": 7.826410712059607e-06,
      "loss": 0.2433,
      "step": 798
    },
    {
      "epoch": 0.5382283597170765,
      "grad_norm": 0.556115984916687,
      "learning_rate": 7.808777150519384e-06,
      "loss": 0.2723,
      "step": 799
    },
    {
      "epoch": 0.5389019872010778,
      "grad_norm": 0.6075104475021362,
      "learning_rate": 7.791141879034055e-06,
      "loss": 0.3197,
      "step": 800
    },
    {
      "epoch": 0.5389019872010778,
      "eval_loss": 0.25853946805000305,
      "eval_runtime": 105.3349,
      "eval_samples_per_second": 47.468,
      "eval_steps_per_second": 2.971,
      "step": 800
    },
    {
      "epoch": 0.5395756146850792,
      "grad_norm": 0.5173077583312988,
      "learning_rate": 7.773504995264167e-06,
      "loss": 0.2458,
      "step": 801
    },
    {
      "epoch": 0.5402492421690805,
      "grad_norm": 0.5317369699478149,
      "learning_rate": 7.755866596879203e-06,
      "loss": 0.2535,
      "step": 802
    },
    {
      "epoch": 0.5409228696530819,
      "grad_norm": 0.5028438568115234,
      "learning_rate": 7.738226781557024e-06,
      "loss": 0.2558,
      "step": 803
    },
    {
      "epoch": 0.5415964971370832,
      "grad_norm": 0.4917846918106079,
      "learning_rate": 7.720585646983346e-06,
      "loss": 0.2567,
      "step": 804
    },
    {
      "epoch": 0.5422701246210846,
      "grad_norm": 0.5413616299629211,
      "learning_rate": 7.702943290851183e-06,
      "loss": 0.3068,
      "step": 805
    },
    {
      "epoch": 0.5429437521050859,
      "grad_norm": 0.5557405352592468,
      "learning_rate": 7.685299810860319e-06,
      "loss": 0.2807,
      "step": 806
    },
    {
      "epoch": 0.5436173795890872,
      "grad_norm": 0.5536317229270935,
      "learning_rate": 7.667655304716762e-06,
      "loss": 0.2535,
      "step": 807
    },
    {
      "epoch": 0.5442910070730885,
      "grad_norm": 0.6285427808761597,
      "learning_rate": 7.650009870132202e-06,
      "loss": 0.2687,
      "step": 808
    },
    {
      "epoch": 0.5449646345570899,
      "grad_norm": 0.5142940282821655,
      "learning_rate": 7.632363604823466e-06,
      "loss": 0.2328,
      "step": 809
    },
    {
      "epoch": 0.5456382620410912,
      "grad_norm": 0.5419033765792847,
      "learning_rate": 7.614716606511986e-06,
      "loss": 0.2687,
      "step": 810
    },
    {
      "epoch": 0.5463118895250926,
      "grad_norm": 0.5078312158584595,
      "learning_rate": 7.597068972923254e-06,
      "loss": 0.2429,
      "step": 811
    },
    {
      "epoch": 0.5469855170090939,
      "grad_norm": 0.5140127539634705,
      "learning_rate": 7.579420801786278e-06,
      "loss": 0.2358,
      "step": 812
    },
    {
      "epoch": 0.5476591444930953,
      "grad_norm": 0.5336434841156006,
      "learning_rate": 7.561772190833041e-06,
      "loss": 0.2561,
      "step": 813
    },
    {
      "epoch": 0.5483327719770966,
      "grad_norm": 0.4892539978027344,
      "learning_rate": 7.544123237797967e-06,
      "loss": 0.2447,
      "step": 814
    },
    {
      "epoch": 0.549006399461098,
      "grad_norm": 0.5128865838050842,
      "learning_rate": 7.526474040417368e-06,
      "loss": 0.2305,
      "step": 815
    },
    {
      "epoch": 0.5496800269450993,
      "grad_norm": 0.5284186601638794,
      "learning_rate": 7.508824696428914e-06,
      "loss": 0.2665,
      "step": 816
    },
    {
      "epoch": 0.5503536544291007,
      "grad_norm": 0.49982714653015137,
      "learning_rate": 7.491175303571087e-06,
      "loss": 0.2361,
      "step": 817
    },
    {
      "epoch": 0.551027281913102,
      "grad_norm": 0.5274138450622559,
      "learning_rate": 7.473525959582631e-06,
      "loss": 0.2542,
      "step": 818
    },
    {
      "epoch": 0.5517009093971034,
      "grad_norm": 0.5714825987815857,
      "learning_rate": 7.4558767622020345e-06,
      "loss": 0.287,
      "step": 819
    },
    {
      "epoch": 0.5523745368811047,
      "grad_norm": 0.5137256979942322,
      "learning_rate": 7.438227809166959e-06,
      "loss": 0.2416,
      "step": 820
    },
    {
      "epoch": 0.5530481643651061,
      "grad_norm": 0.5832123756408691,
      "learning_rate": 7.4205791982137215e-06,
      "loss": 0.2589,
      "step": 821
    },
    {
      "epoch": 0.5537217918491074,
      "grad_norm": 0.6384348273277283,
      "learning_rate": 7.402931027076746e-06,
      "loss": 0.3011,
      "step": 822
    },
    {
      "epoch": 0.5543954193331088,
      "grad_norm": 0.5485447645187378,
      "learning_rate": 7.385283393488017e-06,
      "loss": 0.2596,
      "step": 823
    },
    {
      "epoch": 0.5550690468171101,
      "grad_norm": 0.5725424885749817,
      "learning_rate": 7.367636395176536e-06,
      "loss": 0.278,
      "step": 824
    },
    {
      "epoch": 0.5557426743011115,
      "grad_norm": 0.49892446398735046,
      "learning_rate": 7.349990129867802e-06,
      "loss": 0.2308,
      "step": 825
    },
    {
      "epoch": 0.5564163017851128,
      "grad_norm": 0.5304402709007263,
      "learning_rate": 7.332344695283239e-06,
      "loss": 0.2661,
      "step": 826
    },
    {
      "epoch": 0.5570899292691142,
      "grad_norm": 0.5314590334892273,
      "learning_rate": 7.314700189139683e-06,
      "loss": 0.2545,
      "step": 827
    },
    {
      "epoch": 0.5577635567531155,
      "grad_norm": 0.5156052112579346,
      "learning_rate": 7.297056709148819e-06,
      "loss": 0.2513,
      "step": 828
    },
    {
      "epoch": 0.5584371842371169,
      "grad_norm": 0.5569677352905273,
      "learning_rate": 7.279414353016655e-06,
      "loss": 0.2701,
      "step": 829
    },
    {
      "epoch": 0.5591108117211182,
      "grad_norm": 0.5068705081939697,
      "learning_rate": 7.261773218442978e-06,
      "loss": 0.2578,
      "step": 830
    },
    {
      "epoch": 0.5597844392051196,
      "grad_norm": 0.5413905382156372,
      "learning_rate": 7.244133403120797e-06,
      "loss": 0.2657,
      "step": 831
    },
    {
      "epoch": 0.5604580666891209,
      "grad_norm": 0.5509982109069824,
      "learning_rate": 7.226495004735833e-06,
      "loss": 0.2421,
      "step": 832
    },
    {
      "epoch": 0.5611316941731223,
      "grad_norm": 0.5037456750869751,
      "learning_rate": 7.208858120965949e-06,
      "loss": 0.2366,
      "step": 833
    },
    {
      "epoch": 0.5618053216571236,
      "grad_norm": 0.45753926038742065,
      "learning_rate": 7.191222849480618e-06,
      "loss": 0.2295,
      "step": 834
    },
    {
      "epoch": 0.562478949141125,
      "grad_norm": 0.5005747079849243,
      "learning_rate": 7.1735892879403955e-06,
      "loss": 0.2431,
      "step": 835
    },
    {
      "epoch": 0.5631525766251263,
      "grad_norm": 0.6139580607414246,
      "learning_rate": 7.155957533996361e-06,
      "loss": 0.2954,
      "step": 836
    },
    {
      "epoch": 0.5638262041091276,
      "grad_norm": 0.4900098443031311,
      "learning_rate": 7.1383276852895805e-06,
      "loss": 0.2472,
      "step": 837
    },
    {
      "epoch": 0.564499831593129,
      "grad_norm": 0.5588510632514954,
      "learning_rate": 7.120699839450578e-06,
      "loss": 0.2963,
      "step": 838
    },
    {
      "epoch": 0.5651734590771303,
      "grad_norm": 0.45477819442749023,
      "learning_rate": 7.103074094098776e-06,
      "loss": 0.2459,
      "step": 839
    },
    {
      "epoch": 0.5658470865611317,
      "grad_norm": 0.5369901061058044,
      "learning_rate": 7.085450546841977e-06,
      "loss": 0.2378,
      "step": 840
    },
    {
      "epoch": 0.566520714045133,
      "grad_norm": 0.5580633878707886,
      "learning_rate": 7.0678292952757986e-06,
      "loss": 0.2466,
      "step": 841
    },
    {
      "epoch": 0.5671943415291344,
      "grad_norm": 0.5392370223999023,
      "learning_rate": 7.050210436983152e-06,
      "loss": 0.2847,
      "step": 842
    },
    {
      "epoch": 0.5678679690131357,
      "grad_norm": 0.5429926514625549,
      "learning_rate": 7.032594069533694e-06,
      "loss": 0.2589,
      "step": 843
    },
    {
      "epoch": 0.5685415964971371,
      "grad_norm": 0.529365062713623,
      "learning_rate": 7.0149802904832865e-06,
      "loss": 0.2692,
      "step": 844
    },
    {
      "epoch": 0.5692152239811384,
      "grad_norm": 0.5019341707229614,
      "learning_rate": 6.997369197373462e-06,
      "loss": 0.2501,
      "step": 845
    },
    {
      "epoch": 0.5698888514651398,
      "grad_norm": 0.5088992714881897,
      "learning_rate": 6.979760887730873e-06,
      "loss": 0.2741,
      "step": 846
    },
    {
      "epoch": 0.5705624789491411,
      "grad_norm": 0.5390922427177429,
      "learning_rate": 6.962155459066755e-06,
      "loss": 0.2653,
      "step": 847
    },
    {
      "epoch": 0.5712361064331425,
      "grad_norm": 0.5300227403640747,
      "learning_rate": 6.9445530088764015e-06,
      "loss": 0.2356,
      "step": 848
    },
    {
      "epoch": 0.5719097339171438,
      "grad_norm": 0.5471487641334534,
      "learning_rate": 6.926953634638598e-06,
      "loss": 0.2434,
      "step": 849
    },
    {
      "epoch": 0.5725833614011452,
      "grad_norm": 0.49165770411491394,
      "learning_rate": 6.909357433815104e-06,
      "loss": 0.2539,
      "step": 850
    },
    {
      "epoch": 0.5732569888851465,
      "grad_norm": 0.5154786705970764,
      "learning_rate": 6.891764503850109e-06,
      "loss": 0.2525,
      "step": 851
    },
    {
      "epoch": 0.5739306163691479,
      "grad_norm": 0.5185630321502686,
      "learning_rate": 6.874174942169674e-06,
      "loss": 0.2709,
      "step": 852
    },
    {
      "epoch": 0.5746042438531492,
      "grad_norm": 0.5015746355056763,
      "learning_rate": 6.856588846181228e-06,
      "loss": 0.2522,
      "step": 853
    },
    {
      "epoch": 0.5752778713371506,
      "grad_norm": 0.5378702282905579,
      "learning_rate": 6.839006313272989e-06,
      "loss": 0.2634,
      "step": 854
    },
    {
      "epoch": 0.5759514988211519,
      "grad_norm": 0.5816572308540344,
      "learning_rate": 6.82142744081345e-06,
      "loss": 0.3396,
      "step": 855
    },
    {
      "epoch": 0.5766251263051533,
      "grad_norm": 0.5909308791160583,
      "learning_rate": 6.803852326150838e-06,
      "loss": 0.2834,
      "step": 856
    },
    {
      "epoch": 0.5772987537891546,
      "grad_norm": 0.5006569623947144,
      "learning_rate": 6.786281066612564e-06,
      "loss": 0.212,
      "step": 857
    },
    {
      "epoch": 0.577972381273156,
      "grad_norm": 0.5730767846107483,
      "learning_rate": 6.768713759504694e-06,
      "loss": 0.2998,
      "step": 858
    },
    {
      "epoch": 0.5786460087571573,
      "grad_norm": 0.5159865617752075,
      "learning_rate": 6.751150502111406e-06,
      "loss": 0.2685,
      "step": 859
    },
    {
      "epoch": 0.5793196362411587,
      "grad_norm": 0.5225328803062439,
      "learning_rate": 6.733591391694444e-06,
      "loss": 0.2404,
      "step": 860
    },
    {
      "epoch": 0.57999326372516,
      "grad_norm": 0.540481686592102,
      "learning_rate": 6.7160365254926005e-06,
      "loss": 0.265,
      "step": 861
    },
    {
      "epoch": 0.5806668912091614,
      "grad_norm": 0.5876161456108093,
      "learning_rate": 6.698486000721151e-06,
      "loss": 0.2758,
      "step": 862
    },
    {
      "epoch": 0.5813405186931627,
      "grad_norm": 0.5269771218299866,
      "learning_rate": 6.680939914571336e-06,
      "loss": 0.2497,
      "step": 863
    },
    {
      "epoch": 0.5820141461771641,
      "grad_norm": 0.5683711171150208,
      "learning_rate": 6.663398364209817e-06,
      "loss": 0.2895,
      "step": 864
    },
    {
      "epoch": 0.5826877736611654,
      "grad_norm": 0.5690784454345703,
      "learning_rate": 6.645861446778131e-06,
      "loss": 0.2927,
      "step": 865
    },
    {
      "epoch": 0.5833614011451668,
      "grad_norm": 0.4923837184906006,
      "learning_rate": 6.628329259392169e-06,
      "loss": 0.2294,
      "step": 866
    },
    {
      "epoch": 0.5840350286291681,
      "grad_norm": 0.5871672630310059,
      "learning_rate": 6.610801899141618e-06,
      "loss": 0.2883,
      "step": 867
    },
    {
      "epoch": 0.5847086561131695,
      "grad_norm": 0.5314139127731323,
      "learning_rate": 6.593279463089433e-06,
      "loss": 0.2698,
      "step": 868
    },
    {
      "epoch": 0.5853822835971708,
      "grad_norm": 0.4713616967201233,
      "learning_rate": 6.575762048271311e-06,
      "loss": 0.2551,
      "step": 869
    },
    {
      "epoch": 0.5860559110811722,
      "grad_norm": 0.5604876279830933,
      "learning_rate": 6.558249751695129e-06,
      "loss": 0.2507,
      "step": 870
    },
    {
      "epoch": 0.5867295385651735,
      "grad_norm": 0.5332925319671631,
      "learning_rate": 6.54074267034043e-06,
      "loss": 0.2921,
      "step": 871
    },
    {
      "epoch": 0.5874031660491748,
      "grad_norm": 0.5870206356048584,
      "learning_rate": 6.523240901157874e-06,
      "loss": 0.305,
      "step": 872
    },
    {
      "epoch": 0.5880767935331761,
      "grad_norm": 0.5209013223648071,
      "learning_rate": 6.505744541068696e-06,
      "loss": 0.2504,
      "step": 873
    },
    {
      "epoch": 0.5887504210171774,
      "grad_norm": 0.5347055196762085,
      "learning_rate": 6.488253686964189e-06,
      "loss": 0.26,
      "step": 874
    },
    {
      "epoch": 0.5894240485011788,
      "grad_norm": 0.5568848252296448,
      "learning_rate": 6.470768435705146e-06,
      "loss": 0.2506,
      "step": 875
    },
    {
      "epoch": 0.5900976759851801,
      "grad_norm": 0.4880235493183136,
      "learning_rate": 6.45328888412133e-06,
      "loss": 0.2549,
      "step": 876
    },
    {
      "epoch": 0.5907713034691815,
      "grad_norm": 0.5328478217124939,
      "learning_rate": 6.435815129010952e-06,
      "loss": 0.2892,
      "step": 877
    },
    {
      "epoch": 0.5914449309531828,
      "grad_norm": 0.5507891178131104,
      "learning_rate": 6.418347267140113e-06,
      "loss": 0.295,
      "step": 878
    },
    {
      "epoch": 0.5921185584371842,
      "grad_norm": 0.5917878150939941,
      "learning_rate": 6.400885395242284e-06,
      "loss": 0.2775,
      "step": 879
    },
    {
      "epoch": 0.5927921859211855,
      "grad_norm": 0.5396655201911926,
      "learning_rate": 6.383429610017763e-06,
      "loss": 0.2601,
      "step": 880
    },
    {
      "epoch": 0.5934658134051869,
      "grad_norm": 0.5640776753425598,
      "learning_rate": 6.3659800081331375e-06,
      "loss": 0.2532,
      "step": 881
    },
    {
      "epoch": 0.5941394408891882,
      "grad_norm": 0.5693733096122742,
      "learning_rate": 6.348536686220761e-06,
      "loss": 0.276,
      "step": 882
    },
    {
      "epoch": 0.5948130683731896,
      "grad_norm": 0.49299901723861694,
      "learning_rate": 6.331099740878201e-06,
      "loss": 0.2197,
      "step": 883
    },
    {
      "epoch": 0.5954866958571909,
      "grad_norm": 0.5112996697425842,
      "learning_rate": 6.3136692686677204e-06,
      "loss": 0.2685,
      "step": 884
    },
    {
      "epoch": 0.5961603233411923,
      "grad_norm": 0.5770703554153442,
      "learning_rate": 6.2962453661157305e-06,
      "loss": 0.2439,
      "step": 885
    },
    {
      "epoch": 0.5968339508251936,
      "grad_norm": 0.5604544878005981,
      "learning_rate": 6.2788281297122605e-06,
      "loss": 0.2603,
      "step": 886
    },
    {
      "epoch": 0.597507578309195,
      "grad_norm": 0.5164006948471069,
      "learning_rate": 6.261417655910432e-06,
      "loss": 0.2419,
      "step": 887
    },
    {
      "epoch": 0.5981812057931963,
      "grad_norm": 0.5085450410842896,
      "learning_rate": 6.244014041125906e-06,
      "loss": 0.2714,
      "step": 888
    },
    {
      "epoch": 0.5988548332771977,
      "grad_norm": 0.5820232629776001,
      "learning_rate": 6.226617381736361e-06,
      "loss": 0.2909,
      "step": 889
    },
    {
      "epoch": 0.599528460761199,
      "grad_norm": 0.5919815301895142,
      "learning_rate": 6.209227774080969e-06,
      "loss": 0.3283,
      "step": 890
    },
    {
      "epoch": 0.6002020882452004,
      "grad_norm": 0.5612049102783203,
      "learning_rate": 6.191845314459836e-06,
      "loss": 0.2623,
      "step": 891
    },
    {
      "epoch": 0.6008757157292017,
      "grad_norm": 0.5206785798072815,
      "learning_rate": 6.174470099133495e-06,
      "loss": 0.2391,
      "step": 892
    },
    {
      "epoch": 0.6015493432132031,
      "grad_norm": 0.5109294652938843,
      "learning_rate": 6.157102224322357e-06,
      "loss": 0.2435,
      "step": 893
    },
    {
      "epoch": 0.6022229706972044,
      "grad_norm": 0.5124114155769348,
      "learning_rate": 6.13974178620618e-06,
      "loss": 0.2508,
      "step": 894
    },
    {
      "epoch": 0.6028965981812058,
      "grad_norm": 0.538691520690918,
      "learning_rate": 6.1223888809235475e-06,
      "loss": 0.2742,
      "step": 895
    },
    {
      "epoch": 0.6035702256652071,
      "grad_norm": 0.4782629609107971,
      "learning_rate": 6.105043604571319e-06,
      "loss": 0.215,
      "step": 896
    },
    {
      "epoch": 0.6042438531492085,
      "grad_norm": 0.48708873987197876,
      "learning_rate": 6.087706053204106e-06,
      "loss": 0.2685,
      "step": 897
    },
    {
      "epoch": 0.6049174806332098,
      "grad_norm": 0.5199108719825745,
      "learning_rate": 6.070376322833751e-06,
      "loss": 0.2522,
      "step": 898
    },
    {
      "epoch": 0.6055911081172112,
      "grad_norm": 0.5264055728912354,
      "learning_rate": 6.053054509428774e-06,
      "loss": 0.2702,
      "step": 899
    },
    {
      "epoch": 0.6062647356012125,
      "grad_norm": 0.5014949440956116,
      "learning_rate": 6.035740708913861e-06,
      "loss": 0.2592,
      "step": 900
    },
    {
      "epoch": 0.6062647356012125,
      "eval_loss": 0.255189448595047,
      "eval_runtime": 106.7863,
      "eval_samples_per_second": 46.823,
      "eval_steps_per_second": 2.931,
      "step": 900
    },
    {
      "epoch": 0.6069383630852139,
      "grad_norm": 0.5549845695495605,
      "learning_rate": 6.01843501716932e-06,
      "loss": 0.2676,
      "step": 901
    },
    {
      "epoch": 0.6076119905692152,
      "grad_norm": 0.5285577178001404,
      "learning_rate": 6.001137530030551e-06,
      "loss": 0.287,
      "step": 902
    },
    {
      "epoch": 0.6082856180532166,
      "grad_norm": 0.5555633306503296,
      "learning_rate": 5.983848343287529e-06,
      "loss": 0.27,
      "step": 903
    },
    {
      "epoch": 0.6089592455372179,
      "grad_norm": 0.4878551661968231,
      "learning_rate": 5.966567552684248e-06,
      "loss": 0.2132,
      "step": 904
    },
    {
      "epoch": 0.6096328730212193,
      "grad_norm": 0.5712552070617676,
      "learning_rate": 5.949295253918223e-06,
      "loss": 0.264,
      "step": 905
    },
    {
      "epoch": 0.6103065005052206,
      "grad_norm": 0.5029177665710449,
      "learning_rate": 5.932031542639929e-06,
      "loss": 0.2327,
      "step": 906
    },
    {
      "epoch": 0.610980127989222,
      "grad_norm": 0.5280793309211731,
      "learning_rate": 5.914776514452292e-06,
      "loss": 0.2666,
      "step": 907
    },
    {
      "epoch": 0.6116537554732233,
      "grad_norm": 0.5493948459625244,
      "learning_rate": 5.897530264910151e-06,
      "loss": 0.2747,
      "step": 908
    },
    {
      "epoch": 0.6123273829572247,
      "grad_norm": 0.5202659964561462,
      "learning_rate": 5.880292889519733e-06,
      "loss": 0.2648,
      "step": 909
    },
    {
      "epoch": 0.613001010441226,
      "grad_norm": 0.5150463581085205,
      "learning_rate": 5.863064483738114e-06,
      "loss": 0.2465,
      "step": 910
    },
    {
      "epoch": 0.6136746379252274,
      "grad_norm": 0.5893501043319702,
      "learning_rate": 5.845845142972711e-06,
      "loss": 0.258,
      "step": 911
    },
    {
      "epoch": 0.6143482654092287,
      "grad_norm": 0.5318591594696045,
      "learning_rate": 5.828634962580728e-06,
      "loss": 0.2566,
      "step": 912
    },
    {
      "epoch": 0.6150218928932301,
      "grad_norm": 0.5880672335624695,
      "learning_rate": 5.811434037868652e-06,
      "loss": 0.2776,
      "step": 913
    },
    {
      "epoch": 0.6156955203772314,
      "grad_norm": 0.536673903465271,
      "learning_rate": 5.794242464091703e-06,
      "loss": 0.2655,
      "step": 914
    },
    {
      "epoch": 0.6163691478612328,
      "grad_norm": 0.53472501039505,
      "learning_rate": 5.777060336453324e-06,
      "loss": 0.2465,
      "step": 915
    },
    {
      "epoch": 0.6170427753452341,
      "grad_norm": 0.5193040370941162,
      "learning_rate": 5.75988775010465e-06,
      "loss": 0.2597,
      "step": 916
    },
    {
      "epoch": 0.6177164028292355,
      "grad_norm": 0.5033950209617615,
      "learning_rate": 5.742724800143967e-06,
      "loss": 0.2564,
      "step": 917
    },
    {
      "epoch": 0.6183900303132368,
      "grad_norm": 0.479815274477005,
      "learning_rate": 5.725571581616212e-06,
      "loss": 0.2359,
      "step": 918
    },
    {
      "epoch": 0.6190636577972382,
      "grad_norm": 0.6129284501075745,
      "learning_rate": 5.708428189512418e-06,
      "loss": 0.2789,
      "step": 919
    },
    {
      "epoch": 0.6197372852812395,
      "grad_norm": 0.5521009564399719,
      "learning_rate": 5.691294718769205e-06,
      "loss": 0.2605,
      "step": 920
    },
    {
      "epoch": 0.6204109127652409,
      "grad_norm": 0.5177654027938843,
      "learning_rate": 5.674171264268255e-06,
      "loss": 0.2519,
      "step": 921
    },
    {
      "epoch": 0.6210845402492422,
      "grad_norm": 0.5175455808639526,
      "learning_rate": 5.657057920835781e-06,
      "loss": 0.2247,
      "step": 922
    },
    {
      "epoch": 0.6217581677332435,
      "grad_norm": 0.5852333903312683,
      "learning_rate": 5.639954783241994e-06,
      "loss": 0.2767,
      "step": 923
    },
    {
      "epoch": 0.6224317952172449,
      "grad_norm": 0.508068323135376,
      "learning_rate": 5.622861946200602e-06,
      "loss": 0.2584,
      "step": 924
    },
    {
      "epoch": 0.6231054227012462,
      "grad_norm": 0.5253134965896606,
      "learning_rate": 5.605779504368256e-06,
      "loss": 0.2479,
      "step": 925
    },
    {
      "epoch": 0.6237790501852476,
      "grad_norm": 0.5303956866264343,
      "learning_rate": 5.588707552344052e-06,
      "loss": 0.2445,
      "step": 926
    },
    {
      "epoch": 0.624452677669249,
      "grad_norm": 0.5583487749099731,
      "learning_rate": 5.571646184668989e-06,
      "loss": 0.2703,
      "step": 927
    },
    {
      "epoch": 0.6251263051532503,
      "grad_norm": 0.48656296730041504,
      "learning_rate": 5.5545954958254535e-06,
      "loss": 0.22,
      "step": 928
    },
    {
      "epoch": 0.6257999326372516,
      "grad_norm": 0.5838629603385925,
      "learning_rate": 5.537555580236696e-06,
      "loss": 0.2995,
      "step": 929
    },
    {
      "epoch": 0.626473560121253,
      "grad_norm": 0.4895997643470764,
      "learning_rate": 5.520526532266303e-06,
      "loss": 0.2508,
      "step": 930
    },
    {
      "epoch": 0.6271471876052543,
      "grad_norm": 0.5736584663391113,
      "learning_rate": 5.503508446217687e-06,
      "loss": 0.2738,
      "step": 931
    },
    {
      "epoch": 0.6278208150892557,
      "grad_norm": 0.507853627204895,
      "learning_rate": 5.486501416333547e-06,
      "loss": 0.2342,
      "step": 932
    },
    {
      "epoch": 0.628494442573257,
      "grad_norm": 0.5749799013137817,
      "learning_rate": 5.469505536795354e-06,
      "loss": 0.2505,
      "step": 933
    },
    {
      "epoch": 0.6291680700572584,
      "grad_norm": 0.5327485203742981,
      "learning_rate": 5.452520901722843e-06,
      "loss": 0.2444,
      "step": 934
    },
    {
      "epoch": 0.6298416975412597,
      "grad_norm": 0.5296816229820251,
      "learning_rate": 5.435547605173464e-06,
      "loss": 0.2369,
      "step": 935
    },
    {
      "epoch": 0.630515325025261,
      "grad_norm": 0.568265974521637,
      "learning_rate": 5.4185857411418856e-06,
      "loss": 0.2668,
      "step": 936
    },
    {
      "epoch": 0.6311889525092623,
      "grad_norm": 0.571258544921875,
      "learning_rate": 5.401635403559467e-06,
      "loss": 0.2651,
      "step": 937
    },
    {
      "epoch": 0.6318625799932637,
      "grad_norm": 0.5336675047874451,
      "learning_rate": 5.384696686293728e-06,
      "loss": 0.2571,
      "step": 938
    },
    {
      "epoch": 0.632536207477265,
      "grad_norm": 0.5422372221946716,
      "learning_rate": 5.367769683147849e-06,
      "loss": 0.2474,
      "step": 939
    },
    {
      "epoch": 0.6332098349612664,
      "grad_norm": 0.5360538363456726,
      "learning_rate": 5.350854487860127e-06,
      "loss": 0.2612,
      "step": 940
    },
    {
      "epoch": 0.6338834624452677,
      "grad_norm": 0.5526731014251709,
      "learning_rate": 5.333951194103476e-06,
      "loss": 0.291,
      "step": 941
    },
    {
      "epoch": 0.6345570899292691,
      "grad_norm": 0.501751720905304,
      "learning_rate": 5.317059895484905e-06,
      "loss": 0.2305,
      "step": 942
    },
    {
      "epoch": 0.6352307174132704,
      "grad_norm": 0.5227620005607605,
      "learning_rate": 5.300180685544992e-06,
      "loss": 0.2425,
      "step": 943
    },
    {
      "epoch": 0.6359043448972718,
      "grad_norm": 0.4993986189365387,
      "learning_rate": 5.28331365775737e-06,
      "loss": 0.2426,
      "step": 944
    },
    {
      "epoch": 0.6365779723812731,
      "grad_norm": 0.5128391981124878,
      "learning_rate": 5.266458905528214e-06,
      "loss": 0.2635,
      "step": 945
    },
    {
      "epoch": 0.6372515998652745,
      "grad_norm": 0.5762251615524292,
      "learning_rate": 5.2496165221957105e-06,
      "loss": 0.2652,
      "step": 946
    },
    {
      "epoch": 0.6379252273492758,
      "grad_norm": 0.48678985238075256,
      "learning_rate": 5.232786601029562e-06,
      "loss": 0.2518,
      "step": 947
    },
    {
      "epoch": 0.6385988548332772,
      "grad_norm": 0.5538300275802612,
      "learning_rate": 5.215969235230447e-06,
      "loss": 0.2489,
      "step": 948
    },
    {
      "epoch": 0.6392724823172785,
      "grad_norm": 0.5497295260429382,
      "learning_rate": 5.199164517929521e-06,
      "loss": 0.2454,
      "step": 949
    },
    {
      "epoch": 0.6399461098012799,
      "grad_norm": 0.49725213646888733,
      "learning_rate": 5.182372542187895e-06,
      "loss": 0.2555,
      "step": 950
    },
    {
      "epoch": 0.6406197372852812,
      "grad_norm": 0.533641517162323,
      "learning_rate": 5.165593400996114e-06,
      "loss": 0.2927,
      "step": 951
    },
    {
      "epoch": 0.6412933647692826,
      "grad_norm": 0.5489848256111145,
      "learning_rate": 5.148827187273657e-06,
      "loss": 0.2801,
      "step": 952
    },
    {
      "epoch": 0.6419669922532839,
      "grad_norm": 0.5017451643943787,
      "learning_rate": 5.132073993868406e-06,
      "loss": 0.264,
      "step": 953
    },
    {
      "epoch": 0.6426406197372853,
      "grad_norm": 0.5372846722602844,
      "learning_rate": 5.115333913556137e-06,
      "loss": 0.2721,
      "step": 954
    },
    {
      "epoch": 0.6433142472212866,
      "grad_norm": 0.5356566309928894,
      "learning_rate": 5.098607039040019e-06,
      "loss": 0.2608,
      "step": 955
    },
    {
      "epoch": 0.643987874705288,
      "grad_norm": 0.5957320928573608,
      "learning_rate": 5.081893462950079e-06,
      "loss": 0.2601,
      "step": 956
    },
    {
      "epoch": 0.6446615021892893,
      "grad_norm": 0.5288376212120056,
      "learning_rate": 5.0651932778427074e-06,
      "loss": 0.2587,
      "step": 957
    },
    {
      "epoch": 0.6453351296732907,
      "grad_norm": 0.5290555953979492,
      "learning_rate": 5.048506576200137e-06,
      "loss": 0.2756,
      "step": 958
    },
    {
      "epoch": 0.646008757157292,
      "grad_norm": 0.5248770117759705,
      "learning_rate": 5.031833450429925e-06,
      "loss": 0.2451,
      "step": 959
    },
    {
      "epoch": 0.6466823846412934,
      "grad_norm": 0.5826844573020935,
      "learning_rate": 5.0151739928644585e-06,
      "loss": 0.2619,
      "step": 960
    },
    {
      "epoch": 0.6473560121252947,
      "grad_norm": 0.5782036185264587,
      "learning_rate": 4.998528295760426e-06,
      "loss": 0.2751,
      "step": 961
    },
    {
      "epoch": 0.648029639609296,
      "grad_norm": 0.5718022584915161,
      "learning_rate": 4.981896451298311e-06,
      "loss": 0.2754,
      "step": 962
    },
    {
      "epoch": 0.6487032670932974,
      "grad_norm": 0.5494672060012817,
      "learning_rate": 4.965278551581896e-06,
      "loss": 0.2612,
      "step": 963
    },
    {
      "epoch": 0.6493768945772987,
      "grad_norm": 0.5269261002540588,
      "learning_rate": 4.948674688637724e-06,
      "loss": 0.2498,
      "step": 964
    },
    {
      "epoch": 0.6500505220613001,
      "grad_norm": 0.5737338662147522,
      "learning_rate": 4.932084954414619e-06,
      "loss": 0.2512,
      "step": 965
    },
    {
      "epoch": 0.6507241495453014,
      "grad_norm": 0.6112325191497803,
      "learning_rate": 4.915509440783158e-06,
      "loss": 0.2436,
      "step": 966
    },
    {
      "epoch": 0.6513977770293028,
      "grad_norm": 0.5490378737449646,
      "learning_rate": 4.898948239535162e-06,
      "loss": 0.2666,
      "step": 967
    },
    {
      "epoch": 0.6520714045133041,
      "grad_norm": 0.49087807536125183,
      "learning_rate": 4.882401442383205e-06,
      "loss": 0.2307,
      "step": 968
    },
    {
      "epoch": 0.6527450319973055,
      "grad_norm": 0.5699329972267151,
      "learning_rate": 4.865869140960081e-06,
      "loss": 0.2788,
      "step": 969
    },
    {
      "epoch": 0.6534186594813068,
      "grad_norm": 0.5976101756095886,
      "learning_rate": 4.8493514268183154e-06,
      "loss": 0.295,
      "step": 970
    },
    {
      "epoch": 0.6540922869653082,
      "grad_norm": 0.5223735570907593,
      "learning_rate": 4.8328483914296545e-06,
      "loss": 0.2524,
      "step": 971
    },
    {
      "epoch": 0.6547659144493095,
      "grad_norm": 0.521709680557251,
      "learning_rate": 4.816360126184552e-06,
      "loss": 0.256,
      "step": 972
    },
    {
      "epoch": 0.6554395419333109,
      "grad_norm": 0.6724926829338074,
      "learning_rate": 4.799886722391676e-06,
      "loss": 0.3489,
      "step": 973
    },
    {
      "epoch": 0.6561131694173122,
      "grad_norm": 0.5592423677444458,
      "learning_rate": 4.783428271277383e-06,
      "loss": 0.2486,
      "step": 974
    },
    {
      "epoch": 0.6567867969013136,
      "grad_norm": 0.5178484320640564,
      "learning_rate": 4.766984863985229e-06,
      "loss": 0.231,
      "step": 975
    },
    {
      "epoch": 0.6574604243853149,
      "grad_norm": 0.5621674060821533,
      "learning_rate": 4.750556591575467e-06,
      "loss": 0.286,
      "step": 976
    },
    {
      "epoch": 0.6581340518693163,
      "grad_norm": 0.5048483610153198,
      "learning_rate": 4.734143545024527e-06,
      "loss": 0.2308,
      "step": 977
    },
    {
      "epoch": 0.6588076793533176,
      "grad_norm": 0.5381827354431152,
      "learning_rate": 4.7177458152245286e-06,
      "loss": 0.262,
      "step": 978
    },
    {
      "epoch": 0.659481306837319,
      "grad_norm": 0.6153239607810974,
      "learning_rate": 4.701363492982763e-06,
      "loss": 0.2889,
      "step": 979
    },
    {
      "epoch": 0.6601549343213203,
      "grad_norm": 0.5119926333427429,
      "learning_rate": 4.684996669021202e-06,
      "loss": 0.2313,
      "step": 980
    },
    {
      "epoch": 0.6608285618053217,
      "grad_norm": 0.5575754046440125,
      "learning_rate": 4.668645433975994e-06,
      "loss": 0.2926,
      "step": 981
    },
    {
      "epoch": 0.661502189289323,
      "grad_norm": 0.5210109949111938,
      "learning_rate": 4.652309878396955e-06,
      "loss": 0.2567,
      "step": 982
    },
    {
      "epoch": 0.6621758167733244,
      "grad_norm": 0.5215359330177307,
      "learning_rate": 4.635990092747066e-06,
      "loss": 0.2542,
      "step": 983
    },
    {
      "epoch": 0.6628494442573257,
      "grad_norm": 0.5880634188652039,
      "learning_rate": 4.619686167401991e-06,
      "loss": 0.3099,
      "step": 984
    },
    {
      "epoch": 0.6635230717413271,
      "grad_norm": 0.6169697046279907,
      "learning_rate": 4.603398192649549e-06,
      "loss": 0.3095,
      "step": 985
    },
    {
      "epoch": 0.6641966992253284,
      "grad_norm": 0.5105169415473938,
      "learning_rate": 4.5871262586892365e-06,
      "loss": 0.2439,
      "step": 986
    },
    {
      "epoch": 0.6648703267093298,
      "grad_norm": 0.5112780928611755,
      "learning_rate": 4.5708704556317195e-06,
      "loss": 0.2843,
      "step": 987
    },
    {
      "epoch": 0.6655439541933311,
      "grad_norm": 0.5523808002471924,
      "learning_rate": 4.554630873498325e-06,
      "loss": 0.2779,
      "step": 988
    },
    {
      "epoch": 0.6662175816773325,
      "grad_norm": 0.49872297048568726,
      "learning_rate": 4.538407602220566e-06,
      "loss": 0.2385,
      "step": 989
    },
    {
      "epoch": 0.6668912091613338,
      "grad_norm": 0.4888902008533478,
      "learning_rate": 4.522200731639616e-06,
      "loss": 0.2541,
      "step": 990
    },
    {
      "epoch": 0.6675648366453352,
      "grad_norm": 0.5053279995918274,
      "learning_rate": 4.506010351505834e-06,
      "loss": 0.2465,
      "step": 991
    },
    {
      "epoch": 0.6682384641293365,
      "grad_norm": 0.5656309723854065,
      "learning_rate": 4.489836551478254e-06,
      "loss": 0.2878,
      "step": 992
    },
    {
      "epoch": 0.6689120916133379,
      "grad_norm": 0.5291764736175537,
      "learning_rate": 4.473679421124092e-06,
      "loss": 0.2803,
      "step": 993
    },
    {
      "epoch": 0.6695857190973392,
      "grad_norm": 0.5425894260406494,
      "learning_rate": 4.457539049918253e-06,
      "loss": 0.2758,
      "step": 994
    },
    {
      "epoch": 0.6702593465813406,
      "grad_norm": 0.5237170457839966,
      "learning_rate": 4.441415527242835e-06,
      "loss": 0.2615,
      "step": 995
    },
    {
      "epoch": 0.6709329740653419,
      "grad_norm": 0.48956528306007385,
      "learning_rate": 4.425308942386624e-06,
      "loss": 0.2502,
      "step": 996
    },
    {
      "epoch": 0.6716066015493433,
      "grad_norm": 0.5325567722320557,
      "learning_rate": 4.409219384544621e-06,
      "loss": 0.2663,
      "step": 997
    },
    {
      "epoch": 0.6722802290333446,
      "grad_norm": 0.5196683406829834,
      "learning_rate": 4.3931469428175195e-06,
      "loss": 0.2785,
      "step": 998
    },
    {
      "epoch": 0.672953856517346,
      "grad_norm": 0.5358217358589172,
      "learning_rate": 4.377091706211243e-06,
      "loss": 0.2701,
      "step": 999
    },
    {
      "epoch": 0.6736274840013473,
      "grad_norm": 0.5513088703155518,
      "learning_rate": 4.3610537636364256e-06,
      "loss": 0.2583,
      "step": 1000
    },
    {
      "epoch": 0.6736274840013473,
      "eval_loss": 0.2523915767669678,
      "eval_runtime": 104.4369,
      "eval_samples_per_second": 47.876,
      "eval_steps_per_second": 2.997,
      "step": 1000
    },
    {
      "epoch": 0.6743011114853485,
      "grad_norm": 0.599454402923584,
      "learning_rate": 4.345033203907931e-06,
      "loss": 0.3127,
      "step": 1001
    },
    {
      "epoch": 0.6749747389693499,
      "grad_norm": 0.5195282697677612,
      "learning_rate": 4.329030115744368e-06,
      "loss": 0.2336,
      "step": 1002
    },
    {
      "epoch": 0.6756483664533512,
      "grad_norm": 0.5394783616065979,
      "learning_rate": 4.313044587767581e-06,
      "loss": 0.2266,
      "step": 1003
    },
    {
      "epoch": 0.6763219939373526,
      "grad_norm": 0.502860963344574,
      "learning_rate": 4.297076708502179e-06,
      "loss": 0.2226,
      "step": 1004
    },
    {
      "epoch": 0.6769956214213539,
      "grad_norm": 0.5646010637283325,
      "learning_rate": 4.281126566375035e-06,
      "loss": 0.2612,
      "step": 1005
    },
    {
      "epoch": 0.6776692489053553,
      "grad_norm": 0.5330033898353577,
      "learning_rate": 4.265194249714788e-06,
      "loss": 0.27,
      "step": 1006
    },
    {
      "epoch": 0.6783428763893566,
      "grad_norm": 0.5152742266654968,
      "learning_rate": 4.249279846751376e-06,
      "loss": 0.2522,
      "step": 1007
    },
    {
      "epoch": 0.679016503873358,
      "grad_norm": 0.5699672698974609,
      "learning_rate": 4.233383445615524e-06,
      "loss": 0.3023,
      "step": 1008
    },
    {
      "epoch": 0.6796901313573593,
      "grad_norm": 0.5329395532608032,
      "learning_rate": 4.21750513433827e-06,
      "loss": 0.2413,
      "step": 1009
    },
    {
      "epoch": 0.6803637588413607,
      "grad_norm": 0.4887201488018036,
      "learning_rate": 4.201645000850481e-06,
      "loss": 0.24,
      "step": 1010
    },
    {
      "epoch": 0.681037386325362,
      "grad_norm": 0.49501362442970276,
      "learning_rate": 4.1858031329823445e-06,
      "loss": 0.2288,
      "step": 1011
    },
    {
      "epoch": 0.6817110138093634,
      "grad_norm": 0.48089247941970825,
      "learning_rate": 4.169979618462912e-06,
      "loss": 0.2311,
      "step": 1012
    },
    {
      "epoch": 0.6823846412933647,
      "grad_norm": 0.5128735899925232,
      "learning_rate": 4.154174544919591e-06,
      "loss": 0.2342,
      "step": 1013
    },
    {
      "epoch": 0.6830582687773661,
      "grad_norm": 0.5249293446540833,
      "learning_rate": 4.13838799987766e-06,
      "loss": 0.2799,
      "step": 1014
    },
    {
      "epoch": 0.6837318962613674,
      "grad_norm": 0.5358514785766602,
      "learning_rate": 4.122620070759805e-06,
      "loss": 0.2569,
      "step": 1015
    },
    {
      "epoch": 0.6844055237453688,
      "grad_norm": 0.4961945712566376,
      "learning_rate": 4.106870844885606e-06,
      "loss": 0.2856,
      "step": 1016
    },
    {
      "epoch": 0.6850791512293701,
      "grad_norm": 0.5068737268447876,
      "learning_rate": 4.091140409471082e-06,
      "loss": 0.247,
      "step": 1017
    },
    {
      "epoch": 0.6857527787133715,
      "grad_norm": 0.5845658779144287,
      "learning_rate": 4.0754288516281805e-06,
      "loss": 0.3199,
      "step": 1018
    },
    {
      "epoch": 0.6864264061973728,
      "grad_norm": 0.5782644152641296,
      "learning_rate": 4.05973625836432e-06,
      "loss": 0.277,
      "step": 1019
    },
    {
      "epoch": 0.6871000336813742,
      "grad_norm": 0.4946306347846985,
      "learning_rate": 4.044062716581894e-06,
      "loss": 0.2596,
      "step": 1020
    },
    {
      "epoch": 0.6877736611653755,
      "grad_norm": 0.4905906915664673,
      "learning_rate": 4.02840831307779e-06,
      "loss": 0.243,
      "step": 1021
    },
    {
      "epoch": 0.6884472886493769,
      "grad_norm": 0.5413460731506348,
      "learning_rate": 4.012773134542911e-06,
      "loss": 0.2787,
      "step": 1022
    },
    {
      "epoch": 0.6891209161333782,
      "grad_norm": 0.5733334422111511,
      "learning_rate": 3.997157267561701e-06,
      "loss": 0.2473,
      "step": 1023
    },
    {
      "epoch": 0.6897945436173796,
      "grad_norm": 0.5300191044807434,
      "learning_rate": 3.981560798611655e-06,
      "loss": 0.2451,
      "step": 1024
    },
    {
      "epoch": 0.6904681711013809,
      "grad_norm": 0.5103181600570679,
      "learning_rate": 3.965983814062852e-06,
      "loss": 0.2519,
      "step": 1025
    },
    {
      "epoch": 0.6911417985853823,
      "grad_norm": 0.6034629344940186,
      "learning_rate": 3.950426400177465e-06,
      "loss": 0.2702,
      "step": 1026
    },
    {
      "epoch": 0.6918154260693836,
      "grad_norm": 0.5429885387420654,
      "learning_rate": 3.934888643109288e-06,
      "loss": 0.2549,
      "step": 1027
    },
    {
      "epoch": 0.692489053553385,
      "grad_norm": 0.5191195011138916,
      "learning_rate": 3.919370628903266e-06,
      "loss": 0.263,
      "step": 1028
    },
    {
      "epoch": 0.6931626810373863,
      "grad_norm": 0.5467904210090637,
      "learning_rate": 3.903872443495005e-06,
      "loss": 0.2502,
      "step": 1029
    },
    {
      "epoch": 0.6938363085213877,
      "grad_norm": 0.5506168007850647,
      "learning_rate": 3.888394172710305e-06,
      "loss": 0.2731,
      "step": 1030
    },
    {
      "epoch": 0.694509936005389,
      "grad_norm": 0.5254278182983398,
      "learning_rate": 3.872935902264689e-06,
      "loss": 0.2547,
      "step": 1031
    },
    {
      "epoch": 0.6951835634893904,
      "grad_norm": 0.5569198131561279,
      "learning_rate": 3.857497717762911e-06,
      "loss": 0.2644,
      "step": 1032
    },
    {
      "epoch": 0.6958571909733917,
      "grad_norm": 0.5069310069084167,
      "learning_rate": 3.8420797046985024e-06,
      "loss": 0.2643,
      "step": 1033
    },
    {
      "epoch": 0.6965308184573931,
      "grad_norm": 0.539691686630249,
      "learning_rate": 3.826681948453288e-06,
      "loss": 0.259,
      "step": 1034
    },
    {
      "epoch": 0.6972044459413944,
      "grad_norm": 0.49122515320777893,
      "learning_rate": 3.8113045342969083e-06,
      "loss": 0.2326,
      "step": 1035
    },
    {
      "epoch": 0.6978780734253958,
      "grad_norm": 0.5575307011604309,
      "learning_rate": 3.7959475473863624e-06,
      "loss": 0.262,
      "step": 1036
    },
    {
      "epoch": 0.6985517009093971,
      "grad_norm": 0.48936426639556885,
      "learning_rate": 3.7806110727655185e-06,
      "loss": 0.2561,
      "step": 1037
    },
    {
      "epoch": 0.6992253283933985,
      "grad_norm": 0.5563534498214722,
      "learning_rate": 3.76529519536466e-06,
      "loss": 0.2706,
      "step": 1038
    },
    {
      "epoch": 0.6998989558773998,
      "grad_norm": 0.567457377910614,
      "learning_rate": 3.750000000000002e-06,
      "loss": 0.2453,
      "step": 1039
    },
    {
      "epoch": 0.7005725833614012,
      "grad_norm": 0.5170741677284241,
      "learning_rate": 3.7347255713732236e-06,
      "loss": 0.2199,
      "step": 1040
    },
    {
      "epoch": 0.7012462108454025,
      "grad_norm": 0.5691453218460083,
      "learning_rate": 3.7194719940710135e-06,
      "loss": 0.2831,
      "step": 1041
    },
    {
      "epoch": 0.7019198383294039,
      "grad_norm": 0.5637477040290833,
      "learning_rate": 3.7042393525645793e-06,
      "loss": 0.2747,
      "step": 1042
    },
    {
      "epoch": 0.7025934658134052,
      "grad_norm": 0.5031242966651917,
      "learning_rate": 3.689027731209191e-06,
      "loss": 0.2321,
      "step": 1043
    },
    {
      "epoch": 0.7032670932974066,
      "grad_norm": 0.6389548182487488,
      "learning_rate": 3.6738372142437223e-06,
      "loss": 0.2598,
      "step": 1044
    },
    {
      "epoch": 0.7039407207814079,
      "grad_norm": 0.5508800148963928,
      "learning_rate": 3.6586678857901624e-06,
      "loss": 0.2607,
      "step": 1045
    },
    {
      "epoch": 0.7046143482654093,
      "grad_norm": 0.5346333980560303,
      "learning_rate": 3.6435198298531762e-06,
      "loss": 0.2484,
      "step": 1046
    },
    {
      "epoch": 0.7052879757494106,
      "grad_norm": 0.5504537224769592,
      "learning_rate": 3.6283931303196123e-06,
      "loss": 0.2751,
      "step": 1047
    },
    {
      "epoch": 0.705961603233412,
      "grad_norm": 0.5008198618888855,
      "learning_rate": 3.6132878709580612e-06,
      "loss": 0.235,
      "step": 1048
    },
    {
      "epoch": 0.7066352307174133,
      "grad_norm": 0.5508736371994019,
      "learning_rate": 3.5982041354183843e-06,
      "loss": 0.2627,
      "step": 1049
    },
    {
      "epoch": 0.7073088582014146,
      "grad_norm": 0.5439531207084656,
      "learning_rate": 3.583142007231235e-06,
      "loss": 0.2524,
      "step": 1050
    },
    {
      "epoch": 0.707982485685416,
      "grad_norm": 0.5159290432929993,
      "learning_rate": 3.5681015698076254e-06,
      "loss": 0.2323,
      "step": 1051
    },
    {
      "epoch": 0.7086561131694173,
      "grad_norm": 0.5393319725990295,
      "learning_rate": 3.5530829064384378e-06,
      "loss": 0.2732,
      "step": 1052
    },
    {
      "epoch": 0.7093297406534187,
      "grad_norm": 0.5629071593284607,
      "learning_rate": 3.5380861002939764e-06,
      "loss": 0.2651,
      "step": 1053
    },
    {
      "epoch": 0.71000336813742,
      "grad_norm": 0.4863939881324768,
      "learning_rate": 3.523111234423509e-06,
      "loss": 0.2181,
      "step": 1054
    },
    {
      "epoch": 0.7106769956214214,
      "grad_norm": 0.4968824088573456,
      "learning_rate": 3.508158391754798e-06,
      "loss": 0.22,
      "step": 1055
    },
    {
      "epoch": 0.7113506231054227,
      "grad_norm": 0.49191296100616455,
      "learning_rate": 3.493227655093645e-06,
      "loss": 0.246,
      "step": 1056
    },
    {
      "epoch": 0.7120242505894241,
      "grad_norm": 0.526577353477478,
      "learning_rate": 3.4783191071234387e-06,
      "loss": 0.2494,
      "step": 1057
    },
    {
      "epoch": 0.7126978780734254,
      "grad_norm": 0.5475011467933655,
      "learning_rate": 3.463432830404685e-06,
      "loss": 0.2609,
      "step": 1058
    },
    {
      "epoch": 0.7133715055574268,
      "grad_norm": 0.5295203328132629,
      "learning_rate": 3.448568907374563e-06,
      "loss": 0.2494,
      "step": 1059
    },
    {
      "epoch": 0.7140451330414281,
      "grad_norm": 0.5042027831077576,
      "learning_rate": 3.4337274203464523e-06,
      "loss": 0.2266,
      "step": 1060
    },
    {
      "epoch": 0.7147187605254295,
      "grad_norm": 0.5058079957962036,
      "learning_rate": 3.4189084515094974e-06,
      "loss": 0.2344,
      "step": 1061
    },
    {
      "epoch": 0.7153923880094308,
      "grad_norm": 0.5442999601364136,
      "learning_rate": 3.40411208292813e-06,
      "loss": 0.2545,
      "step": 1062
    },
    {
      "epoch": 0.7160660154934322,
      "grad_norm": 0.578435480594635,
      "learning_rate": 3.3893383965416355e-06,
      "loss": 0.2534,
      "step": 1063
    },
    {
      "epoch": 0.7167396429774335,
      "grad_norm": 0.5491860508918762,
      "learning_rate": 3.37458747416369e-06,
      "loss": 0.3073,
      "step": 1064
    },
    {
      "epoch": 0.7174132704614348,
      "grad_norm": 0.49808141589164734,
      "learning_rate": 3.3598593974818997e-06,
      "loss": 0.2254,
      "step": 1065
    },
    {
      "epoch": 0.7180868979454361,
      "grad_norm": 0.5253027081489563,
      "learning_rate": 3.345154248057359e-06,
      "loss": 0.2227,
      "step": 1066
    },
    {
      "epoch": 0.7187605254294375,
      "grad_norm": 0.5097954273223877,
      "learning_rate": 3.3304721073242004e-06,
      "loss": 0.2159,
      "step": 1067
    },
    {
      "epoch": 0.7194341529134388,
      "grad_norm": 0.5558974146842957,
      "learning_rate": 3.3158130565891347e-06,
      "loss": 0.2458,
      "step": 1068
    },
    {
      "epoch": 0.7201077803974402,
      "grad_norm": 0.529330849647522,
      "learning_rate": 3.3011771770310014e-06,
      "loss": 0.2666,
      "step": 1069
    },
    {
      "epoch": 0.7207814078814415,
      "grad_norm": 0.5007720589637756,
      "learning_rate": 3.286564549700333e-06,
      "loss": 0.2415,
      "step": 1070
    },
    {
      "epoch": 0.7214550353654429,
      "grad_norm": 0.6243747472763062,
      "learning_rate": 3.271975255518884e-06,
      "loss": 0.291,
      "step": 1071
    },
    {
      "epoch": 0.7221286628494442,
      "grad_norm": 0.5337501168251038,
      "learning_rate": 3.2574093752792068e-06,
      "loss": 0.2675,
      "step": 1072
    },
    {
      "epoch": 0.7228022903334456,
      "grad_norm": 0.6054463982582092,
      "learning_rate": 3.2428669896441833e-06,
      "loss": 0.3009,
      "step": 1073
    },
    {
      "epoch": 0.7234759178174469,
      "grad_norm": 0.5312137007713318,
      "learning_rate": 3.228348179146586e-06,
      "loss": 0.2513,
      "step": 1074
    },
    {
      "epoch": 0.7241495453014483,
      "grad_norm": 0.510999858379364,
      "learning_rate": 3.2138530241886403e-06,
      "loss": 0.2454,
      "step": 1075
    },
    {
      "epoch": 0.7248231727854496,
      "grad_norm": 0.5202507972717285,
      "learning_rate": 3.199381605041571e-06,
      "loss": 0.2348,
      "step": 1076
    },
    {
      "epoch": 0.725496800269451,
      "grad_norm": 0.5304343700408936,
      "learning_rate": 3.18493400184515e-06,
      "loss": 0.2677,
      "step": 1077
    },
    {
      "epoch": 0.7261704277534523,
      "grad_norm": 0.5709294676780701,
      "learning_rate": 3.1705102946072746e-06,
      "loss": 0.2855,
      "step": 1078
    },
    {
      "epoch": 0.7268440552374537,
      "grad_norm": 0.5579668879508972,
      "learning_rate": 3.156110563203498e-06,
      "loss": 0.2858,
      "step": 1079
    },
    {
      "epoch": 0.727517682721455,
      "grad_norm": 0.6169936060905457,
      "learning_rate": 3.141734887376612e-06,
      "loss": 0.2939,
      "step": 1080
    },
    {
      "epoch": 0.7281913102054564,
      "grad_norm": 0.556327223777771,
      "learning_rate": 3.127383346736184e-06,
      "loss": 0.2797,
      "step": 1081
    },
    {
      "epoch": 0.7288649376894577,
      "grad_norm": 0.4888077974319458,
      "learning_rate": 3.1130560207581275e-06,
      "loss": 0.2147,
      "step": 1082
    },
    {
      "epoch": 0.729538565173459,
      "grad_norm": 0.568587601184845,
      "learning_rate": 3.098752988784268e-06,
      "loss": 0.2786,
      "step": 1083
    },
    {
      "epoch": 0.7302121926574604,
      "grad_norm": 0.5443982481956482,
      "learning_rate": 3.084474330021882e-06,
      "loss": 0.2445,
      "step": 1084
    },
    {
      "epoch": 0.7308858201414618,
      "grad_norm": 0.4714532494544983,
      "learning_rate": 3.070220123543288e-06,
      "loss": 0.2044,
      "step": 1085
    },
    {
      "epoch": 0.7315594476254631,
      "grad_norm": 0.5746622085571289,
      "learning_rate": 3.0559904482853808e-06,
      "loss": 0.2627,
      "step": 1086
    },
    {
      "epoch": 0.7322330751094644,
      "grad_norm": 0.5493502616882324,
      "learning_rate": 3.041785383049206e-06,
      "loss": 0.2564,
      "step": 1087
    },
    {
      "epoch": 0.7329067025934658,
      "grad_norm": 0.5477399826049805,
      "learning_rate": 3.027605006499536e-06,
      "loss": 0.252,
      "step": 1088
    },
    {
      "epoch": 0.7335803300774671,
      "grad_norm": 0.507681667804718,
      "learning_rate": 3.013449397164407e-06,
      "loss": 0.246,
      "step": 1089
    },
    {
      "epoch": 0.7342539575614685,
      "grad_norm": 0.5245915651321411,
      "learning_rate": 2.99931863343471e-06,
      "loss": 0.2374,
      "step": 1090
    },
    {
      "epoch": 0.7349275850454698,
      "grad_norm": 0.526141345500946,
      "learning_rate": 2.985212793563745e-06,
      "loss": 0.2358,
      "step": 1091
    },
    {
      "epoch": 0.7356012125294712,
      "grad_norm": 0.5303114652633667,
      "learning_rate": 2.971131955666782e-06,
      "loss": 0.232,
      "step": 1092
    },
    {
      "epoch": 0.7362748400134725,
      "grad_norm": 0.5322446227073669,
      "learning_rate": 2.957076197720644e-06,
      "loss": 0.2536,
      "step": 1093
    },
    {
      "epoch": 0.7369484674974739,
      "grad_norm": 0.6054010987281799,
      "learning_rate": 2.9430455975632593e-06,
      "loss": 0.2825,
      "step": 1094
    },
    {
      "epoch": 0.7376220949814752,
      "grad_norm": 0.4822597801685333,
      "learning_rate": 2.9290402328932374e-06,
      "loss": 0.2158,
      "step": 1095
    },
    {
      "epoch": 0.7382957224654766,
      "grad_norm": 0.5092408061027527,
      "learning_rate": 2.9150601812694477e-06,
      "loss": 0.2434,
      "step": 1096
    },
    {
      "epoch": 0.7389693499494779,
      "grad_norm": 0.48755231499671936,
      "learning_rate": 2.901105520110569e-06,
      "loss": 0.2489,
      "step": 1097
    },
    {
      "epoch": 0.7396429774334793,
      "grad_norm": 0.5457514524459839,
      "learning_rate": 2.887176326694684e-06,
      "loss": 0.269,
      "step": 1098
    },
    {
      "epoch": 0.7403166049174806,
      "grad_norm": 0.5498961210250854,
      "learning_rate": 2.8732726781588325e-06,
      "loss": 0.2446,
      "step": 1099
    },
    {
      "epoch": 0.740990232401482,
      "grad_norm": 0.5210698246955872,
      "learning_rate": 2.859394651498592e-06,
      "loss": 0.2447,
      "step": 1100
    },
    {
      "epoch": 0.740990232401482,
      "eval_loss": 0.25004303455352783,
      "eval_runtime": 104.4563,
      "eval_samples_per_second": 47.867,
      "eval_steps_per_second": 2.996,
      "step": 1100
    },
    {
      "epoch": 0.7416638598854833,
      "grad_norm": 0.517212986946106,
      "learning_rate": 2.8455423235676586e-06,
      "loss": 0.252,
      "step": 1101
    },
    {
      "epoch": 0.7423374873694847,
      "grad_norm": 0.5591882467269897,
      "learning_rate": 2.8317157710774066e-06,
      "loss": 0.2567,
      "step": 1102
    },
    {
      "epoch": 0.743011114853486,
      "grad_norm": 0.5390484929084778,
      "learning_rate": 2.8179150705964713e-06,
      "loss": 0.2752,
      "step": 1103
    },
    {
      "epoch": 0.7436847423374874,
      "grad_norm": 0.5284495949745178,
      "learning_rate": 2.8041402985503294e-06,
      "loss": 0.2248,
      "step": 1104
    },
    {
      "epoch": 0.7443583698214887,
      "grad_norm": 0.5000547170639038,
      "learning_rate": 2.7903915312208696e-06,
      "loss": 0.2352,
      "step": 1105
    },
    {
      "epoch": 0.7450319973054901,
      "grad_norm": 0.5302792191505432,
      "learning_rate": 2.7766688447459735e-06,
      "loss": 0.2328,
      "step": 1106
    },
    {
      "epoch": 0.7457056247894914,
      "grad_norm": 0.5096173286437988,
      "learning_rate": 2.762972315119088e-06,
      "loss": 0.2408,
      "step": 1107
    },
    {
      "epoch": 0.7463792522734928,
      "grad_norm": 0.5064148902893066,
      "learning_rate": 2.7493020181888058e-06,
      "loss": 0.2385,
      "step": 1108
    },
    {
      "epoch": 0.7470528797574941,
      "grad_norm": 0.508243203163147,
      "learning_rate": 2.735658029658461e-06,
      "loss": 0.2482,
      "step": 1109
    },
    {
      "epoch": 0.7477265072414955,
      "grad_norm": 0.5063915848731995,
      "learning_rate": 2.7220404250856833e-06,
      "loss": 0.2661,
      "step": 1110
    },
    {
      "epoch": 0.7484001347254968,
      "grad_norm": 0.5426428318023682,
      "learning_rate": 2.7084492798820035e-06,
      "loss": 0.2527,
      "step": 1111
    },
    {
      "epoch": 0.7490737622094982,
      "grad_norm": 0.5647068023681641,
      "learning_rate": 2.6948846693124188e-06,
      "loss": 0.2906,
      "step": 1112
    },
    {
      "epoch": 0.7497473896934995,
      "grad_norm": 0.5713849663734436,
      "learning_rate": 2.681346668494985e-06,
      "loss": 0.258,
      "step": 1113
    },
    {
      "epoch": 0.7504210171775009,
      "grad_norm": 0.5479983687400818,
      "learning_rate": 2.6678353524004027e-06,
      "loss": 0.2393,
      "step": 1114
    },
    {
      "epoch": 0.7510946446615022,
      "grad_norm": 0.5582695007324219,
      "learning_rate": 2.654350795851593e-06,
      "loss": 0.2351,
      "step": 1115
    },
    {
      "epoch": 0.7517682721455036,
      "grad_norm": 0.5141502618789673,
      "learning_rate": 2.640893073523286e-06,
      "loss": 0.2587,
      "step": 1116
    },
    {
      "epoch": 0.7524418996295049,
      "grad_norm": 0.5738133788108826,
      "learning_rate": 2.6274622599416197e-06,
      "loss": 0.2719,
      "step": 1117
    },
    {
      "epoch": 0.7531155271135063,
      "grad_norm": 0.5252017974853516,
      "learning_rate": 2.614058429483703e-06,
      "loss": 0.2979,
      "step": 1118
    },
    {
      "epoch": 0.7537891545975076,
      "grad_norm": 0.5406326055526733,
      "learning_rate": 2.600681656377229e-06,
      "loss": 0.2803,
      "step": 1119
    },
    {
      "epoch": 0.754462782081509,
      "grad_norm": 0.5155901312828064,
      "learning_rate": 2.587332014700051e-06,
      "loss": 0.2645,
      "step": 1120
    },
    {
      "epoch": 0.7551364095655103,
      "grad_norm": 0.49665388464927673,
      "learning_rate": 2.5740095783797656e-06,
      "loss": 0.2482,
      "step": 1121
    },
    {
      "epoch": 0.7558100370495117,
      "grad_norm": 0.585488498210907,
      "learning_rate": 2.560714421193323e-06,
      "loss": 0.3037,
      "step": 1122
    },
    {
      "epoch": 0.756483664533513,
      "grad_norm": 0.5360546708106995,
      "learning_rate": 2.547446616766597e-06,
      "loss": 0.2697,
      "step": 1123
    },
    {
      "epoch": 0.7571572920175144,
      "grad_norm": 0.5727128982543945,
      "learning_rate": 2.534206238573997e-06,
      "loss": 0.2627,
      "step": 1124
    },
    {
      "epoch": 0.7578309195015157,
      "grad_norm": 0.5131103992462158,
      "learning_rate": 2.5209933599380443e-06,
      "loss": 0.2576,
      "step": 1125
    },
    {
      "epoch": 0.7585045469855171,
      "grad_norm": 0.5608295798301697,
      "learning_rate": 2.507808054028972e-06,
      "loss": 0.2851,
      "step": 1126
    },
    {
      "epoch": 0.7591781744695184,
      "grad_norm": 0.47002115845680237,
      "learning_rate": 2.4946503938643306e-06,
      "loss": 0.2293,
      "step": 1127
    },
    {
      "epoch": 0.7598518019535198,
      "grad_norm": 0.5585823655128479,
      "learning_rate": 2.4815204523085656e-06,
      "loss": 0.2893,
      "step": 1128
    },
    {
      "epoch": 0.7605254294375211,
      "grad_norm": 0.5366945266723633,
      "learning_rate": 2.4684183020726213e-06,
      "loss": 0.2358,
      "step": 1129
    },
    {
      "epoch": 0.7611990569215223,
      "grad_norm": 0.5874181985855103,
      "learning_rate": 2.4553440157135496e-06,
      "loss": 0.2795,
      "step": 1130
    },
    {
      "epoch": 0.7618726844055237,
      "grad_norm": 0.4842762351036072,
      "learning_rate": 2.442297665634085e-06,
      "loss": 0.2238,
      "step": 1131
    },
    {
      "epoch": 0.762546311889525,
      "grad_norm": 0.5179473161697388,
      "learning_rate": 2.4292793240822682e-06,
      "loss": 0.236,
      "step": 1132
    },
    {
      "epoch": 0.7632199393735264,
      "grad_norm": 0.5912408232688904,
      "learning_rate": 2.4162890631510233e-06,
      "loss": 0.2599,
      "step": 1133
    },
    {
      "epoch": 0.7638935668575277,
      "grad_norm": 0.541069746017456,
      "learning_rate": 2.4033269547777788e-06,
      "loss": 0.2805,
      "step": 1134
    },
    {
      "epoch": 0.7645671943415291,
      "grad_norm": 0.5376386046409607,
      "learning_rate": 2.3903930707440584e-06,
      "loss": 0.2604,
      "step": 1135
    },
    {
      "epoch": 0.7652408218255304,
      "grad_norm": 0.5389772057533264,
      "learning_rate": 2.3774874826750796e-06,
      "loss": 0.2417,
      "step": 1136
    },
    {
      "epoch": 0.7659144493095318,
      "grad_norm": 0.49994543194770813,
      "learning_rate": 2.364610262039369e-06,
      "loss": 0.237,
      "step": 1137
    },
    {
      "epoch": 0.7665880767935331,
      "grad_norm": 0.5099679827690125,
      "learning_rate": 2.351761480148358e-06,
      "loss": 0.2376,
      "step": 1138
    },
    {
      "epoch": 0.7672617042775345,
      "grad_norm": 0.5313609838485718,
      "learning_rate": 2.3389412081559842e-06,
      "loss": 0.2559,
      "step": 1139
    },
    {
      "epoch": 0.7679353317615358,
      "grad_norm": 0.5471706390380859,
      "learning_rate": 2.326149517058314e-06,
      "loss": 0.2667,
      "step": 1140
    },
    {
      "epoch": 0.7686089592455372,
      "grad_norm": 0.4816801846027374,
      "learning_rate": 2.313386477693131e-06,
      "loss": 0.2245,
      "step": 1141
    },
    {
      "epoch": 0.7692825867295385,
      "grad_norm": 0.5714917182922363,
      "learning_rate": 2.3006521607395516e-06,
      "loss": 0.3004,
      "step": 1142
    },
    {
      "epoch": 0.7699562142135399,
      "grad_norm": 0.5020681619644165,
      "learning_rate": 2.2879466367176393e-06,
      "loss": 0.2477,
      "step": 1143
    },
    {
      "epoch": 0.7706298416975412,
      "grad_norm": 0.5577126145362854,
      "learning_rate": 2.275269975987998e-06,
      "loss": 0.2691,
      "step": 1144
    },
    {
      "epoch": 0.7713034691815426,
      "grad_norm": 0.481965035200119,
      "learning_rate": 2.262622248751405e-06,
      "loss": 0.2481,
      "step": 1145
    },
    {
      "epoch": 0.7719770966655439,
      "grad_norm": 0.5549229979515076,
      "learning_rate": 2.250003525048398e-06,
      "loss": 0.2568,
      "step": 1146
    },
    {
      "epoch": 0.7726507241495453,
      "grad_norm": 0.48651793599128723,
      "learning_rate": 2.2374138747589086e-06,
      "loss": 0.2255,
      "step": 1147
    },
    {
      "epoch": 0.7733243516335466,
      "grad_norm": 0.5344985723495483,
      "learning_rate": 2.224853367601858e-06,
      "loss": 0.2485,
      "step": 1148
    },
    {
      "epoch": 0.773997979117548,
      "grad_norm": 0.5281147360801697,
      "learning_rate": 2.212322073134783e-06,
      "loss": 0.2634,
      "step": 1149
    },
    {
      "epoch": 0.7746716066015493,
      "grad_norm": 0.5449070930480957,
      "learning_rate": 2.199820060753449e-06,
      "loss": 0.2726,
      "step": 1150
    },
    {
      "epoch": 0.7753452340855507,
      "grad_norm": 0.5562757253646851,
      "learning_rate": 2.187347399691457e-06,
      "loss": 0.2837,
      "step": 1151
    },
    {
      "epoch": 0.776018861569552,
      "grad_norm": 0.5350236892700195,
      "learning_rate": 2.1749041590198664e-06,
      "loss": 0.2456,
      "step": 1152
    },
    {
      "epoch": 0.7766924890535534,
      "grad_norm": 0.5433318614959717,
      "learning_rate": 2.1624904076468215e-06,
      "loss": 0.2465,
      "step": 1153
    },
    {
      "epoch": 0.7773661165375547,
      "grad_norm": 0.5183177590370178,
      "learning_rate": 2.1501062143171506e-06,
      "loss": 0.245,
      "step": 1154
    },
    {
      "epoch": 0.7780397440215561,
      "grad_norm": 0.5763646960258484,
      "learning_rate": 2.137751647611997e-06,
      "loss": 0.2403,
      "step": 1155
    },
    {
      "epoch": 0.7787133715055574,
      "grad_norm": 0.47587332129478455,
      "learning_rate": 2.125426775948446e-06,
      "loss": 0.2331,
      "step": 1156
    },
    {
      "epoch": 0.7793869989895588,
      "grad_norm": 0.4680344760417938,
      "learning_rate": 2.113131667579127e-06,
      "loss": 0.2246,
      "step": 1157
    },
    {
      "epoch": 0.7800606264735601,
      "grad_norm": 0.5854060649871826,
      "learning_rate": 2.1008663905918553e-06,
      "loss": 0.3072,
      "step": 1158
    },
    {
      "epoch": 0.7807342539575615,
      "grad_norm": 0.48982059955596924,
      "learning_rate": 2.088631012909242e-06,
      "loss": 0.2257,
      "step": 1159
    },
    {
      "epoch": 0.7814078814415628,
      "grad_norm": 0.522206723690033,
      "learning_rate": 2.0764256022883174e-06,
      "loss": 0.2607,
      "step": 1160
    },
    {
      "epoch": 0.7820815089255642,
      "grad_norm": 0.524056077003479,
      "learning_rate": 2.0642502263201687e-06,
      "loss": 0.2478,
      "step": 1161
    },
    {
      "epoch": 0.7827551364095655,
      "grad_norm": 0.49944427609443665,
      "learning_rate": 2.052104952429555e-06,
      "loss": 0.2371,
      "step": 1162
    },
    {
      "epoch": 0.7834287638935669,
      "grad_norm": 0.5648651719093323,
      "learning_rate": 2.0399898478745307e-06,
      "loss": 0.2341,
      "step": 1163
    },
    {
      "epoch": 0.7841023913775682,
      "grad_norm": 0.5346094965934753,
      "learning_rate": 2.027904979746088e-06,
      "loss": 0.2694,
      "step": 1164
    },
    {
      "epoch": 0.7847760188615696,
      "grad_norm": 0.5461552739143372,
      "learning_rate": 2.0158504149677643e-06,
      "loss": 0.2178,
      "step": 1165
    },
    {
      "epoch": 0.7854496463455709,
      "grad_norm": 0.5188043713569641,
      "learning_rate": 2.003826220295295e-06,
      "loss": 0.2449,
      "step": 1166
    },
    {
      "epoch": 0.7861232738295723,
      "grad_norm": 0.5148183703422546,
      "learning_rate": 1.9918324623162253e-06,
      "loss": 0.2381,
      "step": 1167
    },
    {
      "epoch": 0.7867969013135736,
      "grad_norm": 0.550039529800415,
      "learning_rate": 1.979869207449545e-06,
      "loss": 0.2633,
      "step": 1168
    },
    {
      "epoch": 0.787470528797575,
      "grad_norm": 0.5604125261306763,
      "learning_rate": 1.9679365219453337e-06,
      "loss": 0.2605,
      "step": 1169
    },
    {
      "epoch": 0.7881441562815763,
      "grad_norm": 0.5368652939796448,
      "learning_rate": 1.9560344718843746e-06,
      "loss": 0.2725,
      "step": 1170
    },
    {
      "epoch": 0.7888177837655777,
      "grad_norm": 0.5414903163909912,
      "learning_rate": 1.9441631231778063e-06,
      "loss": 0.2505,
      "step": 1171
    },
    {
      "epoch": 0.789491411249579,
      "grad_norm": 0.4991462528705597,
      "learning_rate": 1.932322541566743e-06,
      "loss": 0.2455,
      "step": 1172
    },
    {
      "epoch": 0.7901650387335803,
      "grad_norm": 0.4789801836013794,
      "learning_rate": 1.920512792621917e-06,
      "loss": 0.2252,
      "step": 1173
    },
    {
      "epoch": 0.7908386662175817,
      "grad_norm": 0.572302520275116,
      "learning_rate": 1.908733941743322e-06,
      "loss": 0.2803,
      "step": 1174
    },
    {
      "epoch": 0.791512293701583,
      "grad_norm": 0.5717800259590149,
      "learning_rate": 1.8969860541598358e-06,
      "loss": 0.2782,
      "step": 1175
    },
    {
      "epoch": 0.7921859211855844,
      "grad_norm": 0.5871665477752686,
      "learning_rate": 1.885269194928876e-06,
      "loss": 0.2884,
      "step": 1176
    },
    {
      "epoch": 0.7928595486695857,
      "grad_norm": 0.5179949402809143,
      "learning_rate": 1.8735834289360281e-06,
      "loss": 0.2484,
      "step": 1177
    },
    {
      "epoch": 0.7935331761535871,
      "grad_norm": 0.5230392813682556,
      "learning_rate": 1.8619288208946858e-06,
      "loss": 0.244,
      "step": 1178
    },
    {
      "epoch": 0.7942068036375884,
      "grad_norm": 0.4939616024494171,
      "learning_rate": 1.850305435345704e-06,
      "loss": 0.2202,
      "step": 1179
    },
    {
      "epoch": 0.7948804311215898,
      "grad_norm": 0.5704658031463623,
      "learning_rate": 1.8387133366570284e-06,
      "loss": 0.2999,
      "step": 1180
    },
    {
      "epoch": 0.7955540586055911,
      "grad_norm": 0.5412735342979431,
      "learning_rate": 1.8271525890233412e-06,
      "loss": 0.254,
      "step": 1181
    },
    {
      "epoch": 0.7962276860895925,
      "grad_norm": 0.5522796511650085,
      "learning_rate": 1.8156232564657204e-06,
      "loss": 0.256,
      "step": 1182
    },
    {
      "epoch": 0.7969013135735938,
      "grad_norm": 0.5172164440155029,
      "learning_rate": 1.8041254028312604e-06,
      "loss": 0.2408,
      "step": 1183
    },
    {
      "epoch": 0.7975749410575952,
      "grad_norm": 0.5235007405281067,
      "learning_rate": 1.792659091792742e-06,
      "loss": 0.2455,
      "step": 1184
    },
    {
      "epoch": 0.7982485685415965,
      "grad_norm": 0.5019668340682983,
      "learning_rate": 1.781224386848265e-06,
      "loss": 0.2212,
      "step": 1185
    },
    {
      "epoch": 0.7989221960255979,
      "grad_norm": 0.5576637387275696,
      "learning_rate": 1.7698213513208983e-06,
      "loss": 0.2655,
      "step": 1186
    },
    {
      "epoch": 0.7995958235095992,
      "grad_norm": 0.5962572693824768,
      "learning_rate": 1.758450048358339e-06,
      "loss": 0.2673,
      "step": 1187
    },
    {
      "epoch": 0.8002694509936006,
      "grad_norm": 0.5175941586494446,
      "learning_rate": 1.7471105409325507e-06,
      "loss": 0.2609,
      "step": 1188
    },
    {
      "epoch": 0.8009430784776019,
      "grad_norm": 0.5470423698425293,
      "learning_rate": 1.7358028918394187e-06,
      "loss": 0.2781,
      "step": 1189
    },
    {
      "epoch": 0.8016167059616033,
      "grad_norm": 0.5484721660614014,
      "learning_rate": 1.7245271636984072e-06,
      "loss": 0.2503,
      "step": 1190
    },
    {
      "epoch": 0.8022903334456046,
      "grad_norm": 0.5539147257804871,
      "learning_rate": 1.7132834189522075e-06,
      "loss": 0.2697,
      "step": 1191
    },
    {
      "epoch": 0.802963960929606,
      "grad_norm": 0.5356603860855103,
      "learning_rate": 1.7020717198663948e-06,
      "loss": 0.2343,
      "step": 1192
    },
    {
      "epoch": 0.8036375884136073,
      "grad_norm": 0.5115563273429871,
      "learning_rate": 1.690892128529078e-06,
      "loss": 0.2507,
      "step": 1193
    },
    {
      "epoch": 0.8043112158976087,
      "grad_norm": 0.5782252550125122,
      "learning_rate": 1.6797447068505604e-06,
      "loss": 0.2993,
      "step": 1194
    },
    {
      "epoch": 0.8049848433816099,
      "grad_norm": 0.4831259548664093,
      "learning_rate": 1.6686295165630005e-06,
      "loss": 0.2095,
      "step": 1195
    },
    {
      "epoch": 0.8056584708656113,
      "grad_norm": 0.5409023761749268,
      "learning_rate": 1.6575466192200609e-06,
      "loss": 0.2591,
      "step": 1196
    },
    {
      "epoch": 0.8063320983496126,
      "grad_norm": 0.510886013507843,
      "learning_rate": 1.6464960761965773e-06,
      "loss": 0.2221,
      "step": 1197
    },
    {
      "epoch": 0.807005725833614,
      "grad_norm": 0.5136117935180664,
      "learning_rate": 1.635477948688209e-06,
      "loss": 0.2306,
      "step": 1198
    },
    {
      "epoch": 0.8076793533176153,
      "grad_norm": 0.5284943580627441,
      "learning_rate": 1.624492297711106e-06,
      "loss": 0.2497,
      "step": 1199
    },
    {
      "epoch": 0.8083529808016167,
      "grad_norm": 0.547935426235199,
      "learning_rate": 1.6135391841015749e-06,
      "loss": 0.237,
      "step": 1200
    },
    {
      "epoch": 0.8083529808016167,
      "eval_loss": 0.24800407886505127,
      "eval_runtime": 104.1907,
      "eval_samples_per_second": 47.989,
      "eval_steps_per_second": 3.004,
      "step": 1200
    },
    {
      "epoch": 0.809026608285618,
      "grad_norm": 0.5304118990898132,
      "learning_rate": 1.6026186685157299e-06,
      "loss": 0.2637,
      "step": 1201
    },
    {
      "epoch": 0.8097002357696194,
      "grad_norm": 0.5001785755157471,
      "learning_rate": 1.591730811429165e-06,
      "loss": 0.2512,
      "step": 1202
    },
    {
      "epoch": 0.8103738632536207,
      "grad_norm": 0.5340592265129089,
      "learning_rate": 1.5808756731366246e-06,
      "loss": 0.2356,
      "step": 1203
    },
    {
      "epoch": 0.8110474907376221,
      "grad_norm": 0.5753365755081177,
      "learning_rate": 1.5700533137516538e-06,
      "loss": 0.2411,
      "step": 1204
    },
    {
      "epoch": 0.8117211182216234,
      "grad_norm": 0.5412161946296692,
      "learning_rate": 1.559263793206282e-06,
      "loss": 0.259,
      "step": 1205
    },
    {
      "epoch": 0.8123947457056248,
      "grad_norm": 0.5372704267501831,
      "learning_rate": 1.5485071712506836e-06,
      "loss": 0.2583,
      "step": 1206
    },
    {
      "epoch": 0.8130683731896261,
      "grad_norm": 0.5177714228630066,
      "learning_rate": 1.5377835074528396e-06,
      "loss": 0.2566,
      "step": 1207
    },
    {
      "epoch": 0.8137420006736275,
      "grad_norm": 0.5761350989341736,
      "learning_rate": 1.5270928611982252e-06,
      "loss": 0.2748,
      "step": 1208
    },
    {
      "epoch": 0.8144156281576288,
      "grad_norm": 0.4516087770462036,
      "learning_rate": 1.5164352916894639e-06,
      "loss": 0.2042,
      "step": 1209
    },
    {
      "epoch": 0.8150892556416302,
      "grad_norm": 0.5373425483703613,
      "learning_rate": 1.5058108579460117e-06,
      "loss": 0.2473,
      "step": 1210
    },
    {
      "epoch": 0.8157628831256315,
      "grad_norm": 0.5737300515174866,
      "learning_rate": 1.4952196188038232e-06,
      "loss": 0.2378,
      "step": 1211
    },
    {
      "epoch": 0.8164365106096328,
      "grad_norm": 0.5578494071960449,
      "learning_rate": 1.4846616329150252e-06,
      "loss": 0.2455,
      "step": 1212
    },
    {
      "epoch": 0.8171101380936342,
      "grad_norm": 0.5392588376998901,
      "learning_rate": 1.4741369587476023e-06,
      "loss": 0.2587,
      "step": 1213
    },
    {
      "epoch": 0.8177837655776355,
      "grad_norm": 0.5129286050796509,
      "learning_rate": 1.4636456545850584e-06,
      "loss": 0.2269,
      "step": 1214
    },
    {
      "epoch": 0.8184573930616369,
      "grad_norm": 0.5656298398971558,
      "learning_rate": 1.4531877785261032e-06,
      "loss": 0.25,
      "step": 1215
    },
    {
      "epoch": 0.8191310205456382,
      "grad_norm": 0.5242322087287903,
      "learning_rate": 1.4427633884843321e-06,
      "loss": 0.2213,
      "step": 1216
    },
    {
      "epoch": 0.8198046480296396,
      "grad_norm": 0.5498111248016357,
      "learning_rate": 1.432372542187895e-06,
      "loss": 0.2842,
      "step": 1217
    },
    {
      "epoch": 0.8204782755136409,
      "grad_norm": 0.5031629204750061,
      "learning_rate": 1.42201529717919e-06,
      "loss": 0.2404,
      "step": 1218
    },
    {
      "epoch": 0.8211519029976423,
      "grad_norm": 0.5014514327049255,
      "learning_rate": 1.4116917108145318e-06,
      "loss": 0.2447,
      "step": 1219
    },
    {
      "epoch": 0.8218255304816436,
      "grad_norm": 0.6042701005935669,
      "learning_rate": 1.4014018402638454e-06,
      "loss": 0.2935,
      "step": 1220
    },
    {
      "epoch": 0.822499157965645,
      "grad_norm": 0.4977283179759979,
      "learning_rate": 1.3911457425103444e-06,
      "loss": 0.2463,
      "step": 1221
    },
    {
      "epoch": 0.8231727854496463,
      "grad_norm": 0.5383139252662659,
      "learning_rate": 1.3809234743502109e-06,
      "loss": 0.2432,
      "step": 1222
    },
    {
      "epoch": 0.8238464129336477,
      "grad_norm": 0.4999409019947052,
      "learning_rate": 1.3707350923922915e-06,
      "loss": 0.2427,
      "step": 1223
    },
    {
      "epoch": 0.824520040417649,
      "grad_norm": 0.5174874067306519,
      "learning_rate": 1.3605806530577725e-06,
      "loss": 0.2475,
      "step": 1224
    },
    {
      "epoch": 0.8251936679016504,
      "grad_norm": 0.545793890953064,
      "learning_rate": 1.3504602125798742e-06,
      "loss": 0.26,
      "step": 1225
    },
    {
      "epoch": 0.8258672953856517,
      "grad_norm": 0.5483867526054382,
      "learning_rate": 1.340373827003543e-06,
      "loss": 0.2454,
      "step": 1226
    },
    {
      "epoch": 0.8265409228696531,
      "grad_norm": 0.5197309851646423,
      "learning_rate": 1.3303215521851303e-06,
      "loss": 0.2109,
      "step": 1227
    },
    {
      "epoch": 0.8272145503536544,
      "grad_norm": 0.5258607864379883,
      "learning_rate": 1.3203034437920889e-06,
      "loss": 0.2473,
      "step": 1228
    },
    {
      "epoch": 0.8278881778376558,
      "grad_norm": 0.5303134322166443,
      "learning_rate": 1.3103195573026708e-06,
      "loss": 0.2348,
      "step": 1229
    },
    {
      "epoch": 0.8285618053216571,
      "grad_norm": 0.6092815399169922,
      "learning_rate": 1.3003699480056073e-06,
      "loss": 0.3257,
      "step": 1230
    },
    {
      "epoch": 0.8292354328056585,
      "grad_norm": 0.5898075103759766,
      "learning_rate": 1.2904546709998153e-06,
      "loss": 0.2941,
      "step": 1231
    },
    {
      "epoch": 0.8299090602896598,
      "grad_norm": 0.519887387752533,
      "learning_rate": 1.2805737811940814e-06,
      "loss": 0.2452,
      "step": 1232
    },
    {
      "epoch": 0.8305826877736612,
      "grad_norm": 0.5605584979057312,
      "learning_rate": 1.2707273333067675e-06,
      "loss": 0.254,
      "step": 1233
    },
    {
      "epoch": 0.8312563152576625,
      "grad_norm": 0.5284910202026367,
      "learning_rate": 1.2609153818654983e-06,
      "loss": 0.2709,
      "step": 1234
    },
    {
      "epoch": 0.8319299427416639,
      "grad_norm": 0.5385324358940125,
      "learning_rate": 1.2511379812068683e-06,
      "loss": 0.2483,
      "step": 1235
    },
    {
      "epoch": 0.8326035702256652,
      "grad_norm": 0.5575158596038818,
      "learning_rate": 1.2413951854761364e-06,
      "loss": 0.2434,
      "step": 1236
    },
    {
      "epoch": 0.8332771977096666,
      "grad_norm": 0.4902471601963043,
      "learning_rate": 1.231687048626925e-06,
      "loss": 0.2183,
      "step": 1237
    },
    {
      "epoch": 0.8339508251936679,
      "grad_norm": 0.4895704984664917,
      "learning_rate": 1.22201362442092e-06,
      "loss": 0.2555,
      "step": 1238
    },
    {
      "epoch": 0.8346244526776693,
      "grad_norm": 0.5114362835884094,
      "learning_rate": 1.2123749664275823e-06,
      "loss": 0.2474,
      "step": 1239
    },
    {
      "epoch": 0.8352980801616706,
      "grad_norm": 0.5052047371864319,
      "learning_rate": 1.2027711280238396e-06,
      "loss": 0.2158,
      "step": 1240
    },
    {
      "epoch": 0.835971707645672,
      "grad_norm": 0.549039900302887,
      "learning_rate": 1.1932021623937954e-06,
      "loss": 0.2728,
      "step": 1241
    },
    {
      "epoch": 0.8366453351296733,
      "grad_norm": 0.5076087117195129,
      "learning_rate": 1.1836681225284401e-06,
      "loss": 0.248,
      "step": 1242
    },
    {
      "epoch": 0.8373189626136747,
      "grad_norm": 0.5593048930168152,
      "learning_rate": 1.1741690612253455e-06,
      "loss": 0.2778,
      "step": 1243
    },
    {
      "epoch": 0.837992590097676,
      "grad_norm": 0.557877242565155,
      "learning_rate": 1.1647050310883855e-06,
      "loss": 0.2744,
      "step": 1244
    },
    {
      "epoch": 0.8386662175816774,
      "grad_norm": 0.571789562702179,
      "learning_rate": 1.155276084527435e-06,
      "loss": 0.2623,
      "step": 1245
    },
    {
      "epoch": 0.8393398450656787,
      "grad_norm": 0.532217800617218,
      "learning_rate": 1.1458822737580804e-06,
      "loss": 0.2604,
      "step": 1246
    },
    {
      "epoch": 0.8400134725496801,
      "grad_norm": 0.488652765750885,
      "learning_rate": 1.1365236508013396e-06,
      "loss": 0.2302,
      "step": 1247
    },
    {
      "epoch": 0.8406871000336814,
      "grad_norm": 0.48154351115226746,
      "learning_rate": 1.1272002674833668e-06,
      "loss": 0.2292,
      "step": 1248
    },
    {
      "epoch": 0.8413607275176828,
      "grad_norm": 0.5416498184204102,
      "learning_rate": 1.1179121754351587e-06,
      "loss": 0.2675,
      "step": 1249
    },
    {
      "epoch": 0.8420343550016841,
      "grad_norm": 0.5223404169082642,
      "learning_rate": 1.1086594260922873e-06,
      "loss": 0.2495,
      "step": 1250
    },
    {
      "epoch": 0.8427079824856855,
      "grad_norm": 0.4795687198638916,
      "learning_rate": 1.0994420706945922e-06,
      "loss": 0.2405,
      "step": 1251
    },
    {
      "epoch": 0.8433816099696868,
      "grad_norm": 0.528590202331543,
      "learning_rate": 1.0902601602859192e-06,
      "loss": 0.234,
      "step": 1252
    },
    {
      "epoch": 0.8440552374536882,
      "grad_norm": 0.5488812327384949,
      "learning_rate": 1.0811137457138195e-06,
      "loss": 0.2309,
      "step": 1253
    },
    {
      "epoch": 0.8447288649376895,
      "grad_norm": 0.48816734552383423,
      "learning_rate": 1.0720028776292775e-06,
      "loss": 0.2252,
      "step": 1254
    },
    {
      "epoch": 0.8454024924216909,
      "grad_norm": 0.4672640860080719,
      "learning_rate": 1.0629276064864315e-06,
      "loss": 0.2241,
      "step": 1255
    },
    {
      "epoch": 0.8460761199056922,
      "grad_norm": 0.6129331588745117,
      "learning_rate": 1.053887982542286e-06,
      "loss": 0.2933,
      "step": 1256
    },
    {
      "epoch": 0.8467497473896936,
      "grad_norm": 0.4707486033439636,
      "learning_rate": 1.0448840558564437e-06,
      "loss": 0.2263,
      "step": 1257
    },
    {
      "epoch": 0.8474233748736949,
      "grad_norm": 0.5238653421401978,
      "learning_rate": 1.0359158762908206e-06,
      "loss": 0.2251,
      "step": 1258
    },
    {
      "epoch": 0.8480970023576961,
      "grad_norm": 0.5185546278953552,
      "learning_rate": 1.0269834935093692e-06,
      "loss": 0.2423,
      "step": 1259
    },
    {
      "epoch": 0.8487706298416975,
      "grad_norm": 0.551475465297699,
      "learning_rate": 1.0180869569778146e-06,
      "loss": 0.252,
      "step": 1260
    },
    {
      "epoch": 0.8494442573256988,
      "grad_norm": 0.5508469343185425,
      "learning_rate": 1.0092263159633643e-06,
      "loss": 0.2689,
      "step": 1261
    },
    {
      "epoch": 0.8501178848097002,
      "grad_norm": 0.5417614579200745,
      "learning_rate": 1.000401619534449e-06,
      "loss": 0.2693,
      "step": 1262
    },
    {
      "epoch": 0.8507915122937015,
      "grad_norm": 0.5418084859848022,
      "learning_rate": 9.91612916560445e-07,
      "loss": 0.2355,
      "step": 1263
    },
    {
      "epoch": 0.8514651397777029,
      "grad_norm": 0.49578985571861267,
      "learning_rate": 9.828602557114017e-07,
      "loss": 0.2373,
      "step": 1264
    },
    {
      "epoch": 0.8521387672617042,
      "grad_norm": 0.5037760734558105,
      "learning_rate": 9.741436854577778e-07,
      "loss": 0.2109,
      "step": 1265
    },
    {
      "epoch": 0.8528123947457056,
      "grad_norm": 0.5789549946784973,
      "learning_rate": 9.654632540701663e-07,
      "loss": 0.2314,
      "step": 1266
    },
    {
      "epoch": 0.8534860222297069,
      "grad_norm": 0.5235623717308044,
      "learning_rate": 9.568190096190321e-07,
      "loss": 0.2648,
      "step": 1267
    },
    {
      "epoch": 0.8541596497137083,
      "grad_norm": 0.5276992917060852,
      "learning_rate": 9.482109999744456e-07,
      "loss": 0.2422,
      "step": 1268
    },
    {
      "epoch": 0.8548332771977096,
      "grad_norm": 0.5321447253227234,
      "learning_rate": 9.396392728058129e-07,
      "loss": 0.2257,
      "step": 1269
    },
    {
      "epoch": 0.855506904681711,
      "grad_norm": 0.5687943696975708,
      "learning_rate": 9.311038755816187e-07,
      "loss": 0.2534,
      "step": 1270
    },
    {
      "epoch": 0.8561805321657123,
      "grad_norm": 0.5804548263549805,
      "learning_rate": 9.226048555691583e-07,
      "loss": 0.2888,
      "step": 1271
    },
    {
      "epoch": 0.8568541596497137,
      "grad_norm": 0.603203535079956,
      "learning_rate": 9.141422598342745e-07,
      "loss": 0.2695,
      "step": 1272
    },
    {
      "epoch": 0.857527787133715,
      "grad_norm": 0.49607640504837036,
      "learning_rate": 9.057161352411055e-07,
      "loss": 0.2329,
      "step": 1273
    },
    {
      "epoch": 0.8582014146177164,
      "grad_norm": 0.5365235209465027,
      "learning_rate": 8.973265284518168e-07,
      "loss": 0.2576,
      "step": 1274
    },
    {
      "epoch": 0.8588750421017177,
      "grad_norm": 0.4797859191894531,
      "learning_rate": 8.889734859263429e-07,
      "loss": 0.2337,
      "step": 1275
    },
    {
      "epoch": 0.8595486695857191,
      "grad_norm": 0.5661630034446716,
      "learning_rate": 8.806570539221378e-07,
      "loss": 0.2612,
      "step": 1276
    },
    {
      "epoch": 0.8602222970697204,
      "grad_norm": 0.5369821190834045,
      "learning_rate": 8.723772784939132e-07,
      "loss": 0.2509,
      "step": 1277
    },
    {
      "epoch": 0.8608959245537218,
      "grad_norm": 0.6320977807044983,
      "learning_rate": 8.641342054933799e-07,
      "loss": 0.2837,
      "step": 1278
    },
    {
      "epoch": 0.8615695520377231,
      "grad_norm": 0.605116069316864,
      "learning_rate": 8.559278805690027e-07,
      "loss": 0.3332,
      "step": 1279
    },
    {
      "epoch": 0.8622431795217245,
      "grad_norm": 0.5639594793319702,
      "learning_rate": 8.477583491657404e-07,
      "loss": 0.267,
      "step": 1280
    },
    {
      "epoch": 0.8629168070057258,
      "grad_norm": 0.5377295613288879,
      "learning_rate": 8.396256565247987e-07,
      "loss": 0.2624,
      "step": 1281
    },
    {
      "epoch": 0.8635904344897272,
      "grad_norm": 0.5511677265167236,
      "learning_rate": 8.315298476833749e-07,
      "loss": 0.2253,
      "step": 1282
    },
    {
      "epoch": 0.8642640619737285,
      "grad_norm": 0.5488938093185425,
      "learning_rate": 8.234709674744156e-07,
      "loss": 0.2745,
      "step": 1283
    },
    {
      "epoch": 0.8649376894577299,
      "grad_norm": 0.538172721862793,
      "learning_rate": 8.154490605263592e-07,
      "loss": 0.2664,
      "step": 1284
    },
    {
      "epoch": 0.8656113169417312,
      "grad_norm": 0.46900901198387146,
      "learning_rate": 8.074641712628963e-07,
      "loss": 0.2223,
      "step": 1285
    },
    {
      "epoch": 0.8662849444257326,
      "grad_norm": 0.5262213349342346,
      "learning_rate": 7.995163439027223e-07,
      "loss": 0.2444,
      "step": 1286
    },
    {
      "epoch": 0.8669585719097339,
      "grad_norm": 0.5740031003952026,
      "learning_rate": 7.916056224592899e-07,
      "loss": 0.3013,
      "step": 1287
    },
    {
      "epoch": 0.8676321993937353,
      "grad_norm": 0.48303380608558655,
      "learning_rate": 7.837320507405633e-07,
      "loss": 0.2352,
      "step": 1288
    },
    {
      "epoch": 0.8683058268777366,
      "grad_norm": 0.5323200821876526,
      "learning_rate": 7.758956723487872e-07,
      "loss": 0.2453,
      "step": 1289
    },
    {
      "epoch": 0.868979454361738,
      "grad_norm": 0.5278131365776062,
      "learning_rate": 7.680965306802288e-07,
      "loss": 0.227,
      "step": 1290
    },
    {
      "epoch": 0.8696530818457393,
      "grad_norm": 0.4973738491535187,
      "learning_rate": 7.603346689249515e-07,
      "loss": 0.2294,
      "step": 1291
    },
    {
      "epoch": 0.8703267093297407,
      "grad_norm": 0.4885808825492859,
      "learning_rate": 7.526101300665692e-07,
      "loss": 0.2251,
      "step": 1292
    },
    {
      "epoch": 0.871000336813742,
      "grad_norm": 0.5182288289070129,
      "learning_rate": 7.44922956882006e-07,
      "loss": 0.2279,
      "step": 1293
    },
    {
      "epoch": 0.8716739642977434,
      "grad_norm": 0.6276636123657227,
      "learning_rate": 7.37273191941267e-07,
      "loss": 0.2574,
      "step": 1294
    },
    {
      "epoch": 0.8723475917817447,
      "grad_norm": 0.514959454536438,
      "learning_rate": 7.296608776071931e-07,
      "loss": 0.2344,
      "step": 1295
    },
    {
      "epoch": 0.873021219265746,
      "grad_norm": 0.5535395741462708,
      "learning_rate": 7.220860560352365e-07,
      "loss": 0.2702,
      "step": 1296
    },
    {
      "epoch": 0.8736948467497474,
      "grad_norm": 0.538652241230011,
      "learning_rate": 7.145487691732194e-07,
      "loss": 0.2414,
      "step": 1297
    },
    {
      "epoch": 0.8743684742337487,
      "grad_norm": 0.4960603415966034,
      "learning_rate": 7.070490587611014e-07,
      "loss": 0.2188,
      "step": 1298
    },
    {
      "epoch": 0.8750421017177501,
      "grad_norm": 0.6888505220413208,
      "learning_rate": 6.995869663307588e-07,
      "loss": 0.2467,
      "step": 1299
    },
    {
      "epoch": 0.8757157292017514,
      "grad_norm": 0.5374876856803894,
      "learning_rate": 6.921625332057413e-07,
      "loss": 0.2615,
      "step": 1300
    },
    {
      "epoch": 0.8757157292017514,
      "eval_loss": 0.24685746431350708,
      "eval_runtime": 105.528,
      "eval_samples_per_second": 47.381,
      "eval_steps_per_second": 2.966,
      "step": 1300
    },
    {
      "epoch": 0.8763893566857528,
      "grad_norm": 0.5609838962554932,
      "learning_rate": 6.847758005010493e-07,
      "loss": 0.2512,
      "step": 1301
    },
    {
      "epoch": 0.8770629841697541,
      "grad_norm": 0.5215060114860535,
      "learning_rate": 6.774268091229097e-07,
      "loss": 0.2401,
      "step": 1302
    },
    {
      "epoch": 0.8777366116537555,
      "grad_norm": 0.5238656401634216,
      "learning_rate": 6.701155997685413e-07,
      "loss": 0.2291,
      "step": 1303
    },
    {
      "epoch": 0.8784102391377568,
      "grad_norm": 0.5399895906448364,
      "learning_rate": 6.628422129259371e-07,
      "loss": 0.2594,
      "step": 1304
    },
    {
      "epoch": 0.8790838666217582,
      "grad_norm": 0.5661953687667847,
      "learning_rate": 6.556066888736334e-07,
      "loss": 0.2781,
      "step": 1305
    },
    {
      "epoch": 0.8797574941057595,
      "grad_norm": 0.5317832827568054,
      "learning_rate": 6.484090676804927e-07,
      "loss": 0.2365,
      "step": 1306
    },
    {
      "epoch": 0.8804311215897609,
      "grad_norm": 0.5083217024803162,
      "learning_rate": 6.412493892054802e-07,
      "loss": 0.251,
      "step": 1307
    },
    {
      "epoch": 0.8811047490737622,
      "grad_norm": 0.5453861951828003,
      "learning_rate": 6.341276930974377e-07,
      "loss": 0.2472,
      "step": 1308
    },
    {
      "epoch": 0.8817783765577636,
      "grad_norm": 0.524014949798584,
      "learning_rate": 6.270440187948734e-07,
      "loss": 0.2392,
      "step": 1309
    },
    {
      "epoch": 0.8824520040417649,
      "grad_norm": 0.5500375628471375,
      "learning_rate": 6.19998405525734e-07,
      "loss": 0.2429,
      "step": 1310
    },
    {
      "epoch": 0.8831256315257663,
      "grad_norm": 0.4858246445655823,
      "learning_rate": 6.129908923071933e-07,
      "loss": 0.2301,
      "step": 1311
    },
    {
      "epoch": 0.8837992590097676,
      "grad_norm": 0.524882972240448,
      "learning_rate": 6.060215179454379e-07,
      "loss": 0.265,
      "step": 1312
    },
    {
      "epoch": 0.884472886493769,
      "grad_norm": 0.47017255425453186,
      "learning_rate": 5.990903210354456e-07,
      "loss": 0.2178,
      "step": 1313
    },
    {
      "epoch": 0.8851465139777703,
      "grad_norm": 0.5531392097473145,
      "learning_rate": 5.921973399607738e-07,
      "loss": 0.2613,
      "step": 1314
    },
    {
      "epoch": 0.8858201414617717,
      "grad_norm": 0.5758329033851624,
      "learning_rate": 5.853426128933548e-07,
      "loss": 0.2408,
      "step": 1315
    },
    {
      "epoch": 0.886493768945773,
      "grad_norm": 0.5558485984802246,
      "learning_rate": 5.78526177793271e-07,
      "loss": 0.271,
      "step": 1316
    },
    {
      "epoch": 0.8871673964297744,
      "grad_norm": 0.5643607974052429,
      "learning_rate": 5.717480724085564e-07,
      "loss": 0.2524,
      "step": 1317
    },
    {
      "epoch": 0.8878410239137757,
      "grad_norm": 0.5513696670532227,
      "learning_rate": 5.650083342749796e-07,
      "loss": 0.271,
      "step": 1318
    },
    {
      "epoch": 0.8885146513977771,
      "grad_norm": 0.5440685749053955,
      "learning_rate": 5.583070007158425e-07,
      "loss": 0.2397,
      "step": 1319
    },
    {
      "epoch": 0.8891882788817784,
      "grad_norm": 0.5702263712882996,
      "learning_rate": 5.516441088417665e-07,
      "loss": 0.2512,
      "step": 1320
    },
    {
      "epoch": 0.8898619063657798,
      "grad_norm": 0.5015043020248413,
      "learning_rate": 5.450196955504946e-07,
      "loss": 0.2414,
      "step": 1321
    },
    {
      "epoch": 0.8905355338497811,
      "grad_norm": 0.5015976428985596,
      "learning_rate": 5.384337975266789e-07,
      "loss": 0.2394,
      "step": 1322
    },
    {
      "epoch": 0.8912091613337825,
      "grad_norm": 0.5432824492454529,
      "learning_rate": 5.318864512416871e-07,
      "loss": 0.2451,
      "step": 1323
    },
    {
      "epoch": 0.8918827888177837,
      "grad_norm": 0.550553023815155,
      "learning_rate": 5.253776929533898e-07,
      "loss": 0.229,
      "step": 1324
    },
    {
      "epoch": 0.8925564163017851,
      "grad_norm": 0.49185919761657715,
      "learning_rate": 5.1890755870597e-07,
      "loss": 0.2408,
      "step": 1325
    },
    {
      "epoch": 0.8932300437857864,
      "grad_norm": 0.5123654007911682,
      "learning_rate": 5.124760843297144e-07,
      "loss": 0.2529,
      "step": 1326
    },
    {
      "epoch": 0.8939036712697878,
      "grad_norm": 0.5548482537269592,
      "learning_rate": 5.060833054408206e-07,
      "loss": 0.2219,
      "step": 1327
    },
    {
      "epoch": 0.8945772987537891,
      "grad_norm": 0.5485543608665466,
      "learning_rate": 4.997292574412019e-07,
      "loss": 0.281,
      "step": 1328
    },
    {
      "epoch": 0.8952509262377905,
      "grad_norm": 0.49521514773368835,
      "learning_rate": 4.934139755182801e-07,
      "loss": 0.2481,
      "step": 1329
    },
    {
      "epoch": 0.8959245537217918,
      "grad_norm": 0.5502017140388489,
      "learning_rate": 4.871374946448077e-07,
      "loss": 0.2141,
      "step": 1330
    },
    {
      "epoch": 0.8965981812057932,
      "grad_norm": 0.5505282878875732,
      "learning_rate": 4.808998495786577e-07,
      "loss": 0.2761,
      "step": 1331
    },
    {
      "epoch": 0.8972718086897945,
      "grad_norm": 0.5089839696884155,
      "learning_rate": 4.747010748626404e-07,
      "loss": 0.2086,
      "step": 1332
    },
    {
      "epoch": 0.8979454361737959,
      "grad_norm": 0.556952953338623,
      "learning_rate": 4.685412048243118e-07,
      "loss": 0.2672,
      "step": 1333
    },
    {
      "epoch": 0.8986190636577972,
      "grad_norm": 0.5195740461349487,
      "learning_rate": 4.6242027357577903e-07,
      "loss": 0.2466,
      "step": 1334
    },
    {
      "epoch": 0.8992926911417986,
      "grad_norm": 0.528290867805481,
      "learning_rate": 4.5633831501351616e-07,
      "loss": 0.24,
      "step": 1335
    },
    {
      "epoch": 0.8999663186257999,
      "grad_norm": 0.5216066241264343,
      "learning_rate": 4.5029536281817386e-07,
      "loss": 0.2312,
      "step": 1336
    },
    {
      "epoch": 0.9006399461098012,
      "grad_norm": 0.473034143447876,
      "learning_rate": 4.442914504543924e-07,
      "loss": 0.2353,
      "step": 1337
    },
    {
      "epoch": 0.9013135735938026,
      "grad_norm": 0.5582137703895569,
      "learning_rate": 4.3832661117061993e-07,
      "loss": 0.2657,
      "step": 1338
    },
    {
      "epoch": 0.901987201077804,
      "grad_norm": 0.5023413300514221,
      "learning_rate": 4.3240087799892357e-07,
      "loss": 0.2335,
      "step": 1339
    },
    {
      "epoch": 0.9026608285618053,
      "grad_norm": 0.5657601952552795,
      "learning_rate": 4.2651428375480694e-07,
      "loss": 0.2517,
      "step": 1340
    },
    {
      "epoch": 0.9033344560458066,
      "grad_norm": 0.6110347509384155,
      "learning_rate": 4.206668610370362e-07,
      "loss": 0.2873,
      "step": 1341
    },
    {
      "epoch": 0.904008083529808,
      "grad_norm": 0.5331605076789856,
      "learning_rate": 4.14858642227447e-07,
      "loss": 0.2316,
      "step": 1342
    },
    {
      "epoch": 0.9046817110138093,
      "grad_norm": 0.5356966853141785,
      "learning_rate": 4.090896594907767e-07,
      "loss": 0.2124,
      "step": 1343
    },
    {
      "epoch": 0.9053553384978107,
      "grad_norm": 0.5137231349945068,
      "learning_rate": 4.033599447744785e-07,
      "loss": 0.2475,
      "step": 1344
    },
    {
      "epoch": 0.906028965981812,
      "grad_norm": 0.5301010012626648,
      "learning_rate": 3.9766952980854755e-07,
      "loss": 0.2319,
      "step": 1345
    },
    {
      "epoch": 0.9067025934658134,
      "grad_norm": 0.5305171012878418,
      "learning_rate": 3.9201844610534667e-07,
      "loss": 0.2307,
      "step": 1346
    },
    {
      "epoch": 0.9073762209498147,
      "grad_norm": 0.5359605550765991,
      "learning_rate": 3.8640672495942777e-07,
      "loss": 0.2406,
      "step": 1347
    },
    {
      "epoch": 0.9080498484338161,
      "grad_norm": 0.534532904624939,
      "learning_rate": 3.8083439744736296e-07,
      "loss": 0.263,
      "step": 1348
    },
    {
      "epoch": 0.9087234759178174,
      "grad_norm": 0.5522202253341675,
      "learning_rate": 3.75301494427569e-07,
      "loss": 0.257,
      "step": 1349
    },
    {
      "epoch": 0.9093971034018188,
      "grad_norm": 0.5170401334762573,
      "learning_rate": 3.6980804654013794e-07,
      "loss": 0.2534,
      "step": 1350
    },
    {
      "epoch": 0.9100707308858201,
      "grad_norm": 0.542862594127655,
      "learning_rate": 3.643540842066692e-07,
      "loss": 0.2502,
      "step": 1351
    },
    {
      "epoch": 0.9107443583698215,
      "grad_norm": 0.5424035787582397,
      "learning_rate": 3.5893963763009713e-07,
      "loss": 0.2531,
      "step": 1352
    },
    {
      "epoch": 0.9114179858538228,
      "grad_norm": 0.517169177532196,
      "learning_rate": 3.5356473679452524e-07,
      "loss": 0.2209,
      "step": 1353
    },
    {
      "epoch": 0.9120916133378242,
      "grad_norm": 0.536840558052063,
      "learning_rate": 3.482294114650639e-07,
      "loss": 0.2681,
      "step": 1354
    },
    {
      "epoch": 0.9127652408218255,
      "grad_norm": 0.6323632001876831,
      "learning_rate": 3.4293369118765794e-07,
      "loss": 0.3221,
      "step": 1355
    },
    {
      "epoch": 0.9134388683058269,
      "grad_norm": 0.5508329272270203,
      "learning_rate": 3.3767760528893356e-07,
      "loss": 0.2675,
      "step": 1356
    },
    {
      "epoch": 0.9141124957898282,
      "grad_norm": 0.5150956511497498,
      "learning_rate": 3.324611828760241e-07,
      "loss": 0.2383,
      "step": 1357
    },
    {
      "epoch": 0.9147861232738296,
      "grad_norm": 0.5284437537193298,
      "learning_rate": 3.272844528364161e-07,
      "loss": 0.2326,
      "step": 1358
    },
    {
      "epoch": 0.9154597507578309,
      "grad_norm": 0.4829590618610382,
      "learning_rate": 3.221474438377903e-07,
      "loss": 0.2247,
      "step": 1359
    },
    {
      "epoch": 0.9161333782418323,
      "grad_norm": 0.5343140363693237,
      "learning_rate": 3.1705018432785673e-07,
      "loss": 0.2778,
      "step": 1360
    },
    {
      "epoch": 0.9168070057258336,
      "grad_norm": 0.5202133655548096,
      "learning_rate": 3.1199270253420397e-07,
      "loss": 0.262,
      "step": 1361
    },
    {
      "epoch": 0.917480633209835,
      "grad_norm": 0.6042024493217468,
      "learning_rate": 3.069750264641369e-07,
      "loss": 0.3138,
      "step": 1362
    },
    {
      "epoch": 0.9181542606938363,
      "grad_norm": 0.4585079550743103,
      "learning_rate": 3.0199718390452825e-07,
      "loss": 0.1988,
      "step": 1363
    },
    {
      "epoch": 0.9188278881778377,
      "grad_norm": 0.5287424921989441,
      "learning_rate": 2.9705920242165565e-07,
      "loss": 0.2417,
      "step": 1364
    },
    {
      "epoch": 0.919501515661839,
      "grad_norm": 0.5599202513694763,
      "learning_rate": 2.9216110936105906e-07,
      "loss": 0.2709,
      "step": 1365
    },
    {
      "epoch": 0.9201751431458404,
      "grad_norm": 0.5495063066482544,
      "learning_rate": 2.8730293184738105e-07,
      "loss": 0.2546,
      "step": 1366
    },
    {
      "epoch": 0.9208487706298417,
      "grad_norm": 0.529340386390686,
      "learning_rate": 2.8248469678422346e-07,
      "loss": 0.2454,
      "step": 1367
    },
    {
      "epoch": 0.9215223981138431,
      "grad_norm": 0.5447036027908325,
      "learning_rate": 2.7770643085399004e-07,
      "loss": 0.2703,
      "step": 1368
    },
    {
      "epoch": 0.9221960255978444,
      "grad_norm": 0.5069358348846436,
      "learning_rate": 2.729681605177492e-07,
      "loss": 0.2529,
      "step": 1369
    },
    {
      "epoch": 0.9228696530818458,
      "grad_norm": 0.5129917860031128,
      "learning_rate": 2.6826991201507724e-07,
      "loss": 0.2237,
      "step": 1370
    },
    {
      "epoch": 0.9235432805658471,
      "grad_norm": 0.5351532697677612,
      "learning_rate": 2.636117113639194e-07,
      "loss": 0.2592,
      "step": 1371
    },
    {
      "epoch": 0.9242169080498485,
      "grad_norm": 0.5014567375183105,
      "learning_rate": 2.589935843604452e-07,
      "loss": 0.2112,
      "step": 1372
    },
    {
      "epoch": 0.9248905355338498,
      "grad_norm": 0.5409959554672241,
      "learning_rate": 2.54415556578903e-07,
      "loss": 0.259,
      "step": 1373
    },
    {
      "epoch": 0.9255641630178512,
      "grad_norm": 0.547943115234375,
      "learning_rate": 2.4987765337148e-07,
      "loss": 0.262,
      "step": 1374
    },
    {
      "epoch": 0.9262377905018525,
      "grad_norm": 0.5011503100395203,
      "learning_rate": 2.453798998681625e-07,
      "loss": 0.2436,
      "step": 1375
    },
    {
      "epoch": 0.9269114179858539,
      "grad_norm": 0.5170352458953857,
      "learning_rate": 2.4092232097659486e-07,
      "loss": 0.2529,
      "step": 1376
    },
    {
      "epoch": 0.9275850454698552,
      "grad_norm": 0.5595082640647888,
      "learning_rate": 2.3650494138194257e-07,
      "loss": 0.2843,
      "step": 1377
    },
    {
      "epoch": 0.9282586729538566,
      "grad_norm": 0.5302537679672241,
      "learning_rate": 2.3212778554675766e-07,
      "loss": 0.2382,
      "step": 1378
    },
    {
      "epoch": 0.9289323004378579,
      "grad_norm": 0.5083282589912415,
      "learning_rate": 2.277908777108387e-07,
      "loss": 0.2587,
      "step": 1379
    },
    {
      "epoch": 0.9296059279218593,
      "grad_norm": 0.512374222278595,
      "learning_rate": 2.2349424189109984e-07,
      "loss": 0.2337,
      "step": 1380
    },
    {
      "epoch": 0.9302795554058606,
      "grad_norm": 0.5485064387321472,
      "learning_rate": 2.192379018814372e-07,
      "loss": 0.263,
      "step": 1381
    },
    {
      "epoch": 0.930953182889862,
      "grad_norm": 0.5666427612304688,
      "learning_rate": 2.150218812525953e-07,
      "loss": 0.2683,
      "step": 1382
    },
    {
      "epoch": 0.9316268103738633,
      "grad_norm": 0.5534345507621765,
      "learning_rate": 2.1084620335204225e-07,
      "loss": 0.2069,
      "step": 1383
    },
    {
      "epoch": 0.9323004378578647,
      "grad_norm": 0.5635040402412415,
      "learning_rate": 2.0671089130383152e-07,
      "loss": 0.3081,
      "step": 1384
    },
    {
      "epoch": 0.932974065341866,
      "grad_norm": 0.5579578280448914,
      "learning_rate": 2.0261596800848132e-07,
      "loss": 0.2694,
      "step": 1385
    },
    {
      "epoch": 0.9336476928258673,
      "grad_norm": 0.5865366458892822,
      "learning_rate": 1.9856145614284616e-07,
      "loss": 0.261,
      "step": 1386
    },
    {
      "epoch": 0.9343213203098687,
      "grad_norm": 0.5374801754951477,
      "learning_rate": 1.9454737815998546e-07,
      "loss": 0.258,
      "step": 1387
    },
    {
      "epoch": 0.9349949477938699,
      "grad_norm": 0.4848352372646332,
      "learning_rate": 1.9057375628905112e-07,
      "loss": 0.2121,
      "step": 1388
    },
    {
      "epoch": 0.9356685752778713,
      "grad_norm": 0.5470321774482727,
      "learning_rate": 1.8664061253514997e-07,
      "loss": 0.2556,
      "step": 1389
    },
    {
      "epoch": 0.9363422027618726,
      "grad_norm": 0.4898010194301605,
      "learning_rate": 1.8274796867923578e-07,
      "loss": 0.2436,
      "step": 1390
    },
    {
      "epoch": 0.937015830245874,
      "grad_norm": 0.5198950171470642,
      "learning_rate": 1.788958462779766e-07,
      "loss": 0.2523,
      "step": 1391
    },
    {
      "epoch": 0.9376894577298753,
      "grad_norm": 0.5553786754608154,
      "learning_rate": 1.750842666636443e-07,
      "loss": 0.2764,
      "step": 1392
    },
    {
      "epoch": 0.9383630852138767,
      "grad_norm": 0.5620520114898682,
      "learning_rate": 1.7131325094399352e-07,
      "loss": 0.249,
      "step": 1393
    },
    {
      "epoch": 0.939036712697878,
      "grad_norm": 0.5561796426773071,
      "learning_rate": 1.6758282000214202e-07,
      "loss": 0.2581,
      "step": 1394
    },
    {
      "epoch": 0.9397103401818794,
      "grad_norm": 0.5772345066070557,
      "learning_rate": 1.6389299449645734e-07,
      "loss": 0.2718,
      "step": 1395
    },
    {
      "epoch": 0.9403839676658807,
      "grad_norm": 0.5675050020217896,
      "learning_rate": 1.6024379486044517e-07,
      "loss": 0.2668,
      "step": 1396
    },
    {
      "epoch": 0.9410575951498821,
      "grad_norm": 0.517271876335144,
      "learning_rate": 1.5663524130262867e-07,
      "loss": 0.2287,
      "step": 1397
    },
    {
      "epoch": 0.9417312226338834,
      "grad_norm": 0.5633882284164429,
      "learning_rate": 1.5306735380644698e-07,
      "loss": 0.2676,
      "step": 1398
    },
    {
      "epoch": 0.9424048501178848,
      "grad_norm": 0.5311648845672607,
      "learning_rate": 1.4954015213013427e-07,
      "loss": 0.2269,
      "step": 1399
    },
    {
      "epoch": 0.9430784776018861,
      "grad_norm": 0.4805348813533783,
      "learning_rate": 1.4605365580661668e-07,
      "loss": 0.2116,
      "step": 1400
    },
    {
      "epoch": 0.9430784776018861,
      "eval_loss": 0.24632865190505981,
      "eval_runtime": 107.1175,
      "eval_samples_per_second": 46.678,
      "eval_steps_per_second": 2.922,
      "step": 1400
    },
    {
      "epoch": 0.9437521050858875,
      "grad_norm": 0.4763931632041931,
      "learning_rate": 1.4260788414340226e-07,
      "loss": 0.2256,
      "step": 1401
    },
    {
      "epoch": 0.9444257325698888,
      "grad_norm": 0.5097639560699463,
      "learning_rate": 1.3920285622247454e-07,
      "loss": 0.2468,
      "step": 1402
    },
    {
      "epoch": 0.9450993600538902,
      "grad_norm": 0.5467292666435242,
      "learning_rate": 1.3583859090018664e-07,
      "loss": 0.2254,
      "step": 1403
    },
    {
      "epoch": 0.9457729875378915,
      "grad_norm": 0.52953040599823,
      "learning_rate": 1.3251510680715562e-07,
      "loss": 0.2675,
      "step": 1404
    },
    {
      "epoch": 0.9464466150218929,
      "grad_norm": 0.5016205906867981,
      "learning_rate": 1.2923242234816003e-07,
      "loss": 0.2179,
      "step": 1405
    },
    {
      "epoch": 0.9471202425058942,
      "grad_norm": 0.5073173642158508,
      "learning_rate": 1.2599055570204077e-07,
      "loss": 0.2698,
      "step": 1406
    },
    {
      "epoch": 0.9477938699898956,
      "grad_norm": 0.6035676002502441,
      "learning_rate": 1.2278952482159628e-07,
      "loss": 0.2815,
      "step": 1407
    },
    {
      "epoch": 0.9484674974738969,
      "grad_norm": 0.5148323178291321,
      "learning_rate": 1.196293474334842e-07,
      "loss": 0.2466,
      "step": 1408
    },
    {
      "epoch": 0.9491411249578983,
      "grad_norm": 0.5295473337173462,
      "learning_rate": 1.1651004103812479e-07,
      "loss": 0.234,
      "step": 1409
    },
    {
      "epoch": 0.9498147524418996,
      "grad_norm": 0.5339107513427734,
      "learning_rate": 1.1343162290960357e-07,
      "loss": 0.2453,
      "step": 1410
    },
    {
      "epoch": 0.950488379925901,
      "grad_norm": 0.5701354146003723,
      "learning_rate": 1.1039411009557465e-07,
      "loss": 0.2803,
      "step": 1411
    },
    {
      "epoch": 0.9511620074099023,
      "grad_norm": 0.5201859474182129,
      "learning_rate": 1.0739751941716585e-07,
      "loss": 0.2405,
      "step": 1412
    },
    {
      "epoch": 0.9518356348939037,
      "grad_norm": 0.48217666149139404,
      "learning_rate": 1.0444186746888795e-07,
      "loss": 0.2175,
      "step": 1413
    },
    {
      "epoch": 0.952509262377905,
      "grad_norm": 0.5258702039718628,
      "learning_rate": 1.0152717061854056e-07,
      "loss": 0.2438,
      "step": 1414
    },
    {
      "epoch": 0.9531828898619064,
      "grad_norm": 0.5111593008041382,
      "learning_rate": 9.865344500712137e-08,
      "loss": 0.2394,
      "step": 1415
    },
    {
      "epoch": 0.9538565173459077,
      "grad_norm": 0.5637338161468506,
      "learning_rate": 9.582070654874126e-08,
      "loss": 0.2796,
      "step": 1416
    },
    {
      "epoch": 0.954530144829909,
      "grad_norm": 0.5105905532836914,
      "learning_rate": 9.302897093052765e-08,
      "loss": 0.2336,
      "step": 1417
    },
    {
      "epoch": 0.9552037723139104,
      "grad_norm": 0.5599192380905151,
      "learning_rate": 9.027825361254626e-08,
      "loss": 0.2682,
      "step": 1418
    },
    {
      "epoch": 0.9558773997979118,
      "grad_norm": 0.5528537631034851,
      "learning_rate": 8.756856982771122e-08,
      "loss": 0.2374,
      "step": 1419
    },
    {
      "epoch": 0.9565510272819131,
      "grad_norm": 0.5219756960868835,
      "learning_rate": 8.489993458169837e-08,
      "loss": 0.2489,
      "step": 1420
    },
    {
      "epoch": 0.9572246547659145,
      "grad_norm": 0.535754382610321,
      "learning_rate": 8.227236265286958e-08,
      "loss": 0.2348,
      "step": 1421
    },
    {
      "epoch": 0.9578982822499158,
      "grad_norm": 0.5138925909996033,
      "learning_rate": 7.968586859218363e-08,
      "loss": 0.2354,
      "step": 1422
    },
    {
      "epoch": 0.9585719097339171,
      "grad_norm": 0.5310443043708801,
      "learning_rate": 7.714046672311791e-08,
      "loss": 0.2452,
      "step": 1423
    },
    {
      "epoch": 0.9592455372179185,
      "grad_norm": 0.558163046836853,
      "learning_rate": 7.46361711415927e-08,
      "loss": 0.2499,
      "step": 1424
    },
    {
      "epoch": 0.9599191647019198,
      "grad_norm": 0.4975683391094208,
      "learning_rate": 7.21729957158862e-08,
      "loss": 0.2202,
      "step": 1425
    },
    {
      "epoch": 0.9605927921859212,
      "grad_norm": 0.5497083067893982,
      "learning_rate": 6.975095408656373e-08,
      "loss": 0.2575,
      "step": 1426
    },
    {
      "epoch": 0.9612664196699225,
      "grad_norm": 0.5261396765708923,
      "learning_rate": 6.737005966639953e-08,
      "loss": 0.2744,
      "step": 1427
    },
    {
      "epoch": 0.9619400471539239,
      "grad_norm": 0.5656693577766418,
      "learning_rate": 6.503032564030342e-08,
      "loss": 0.2331,
      "step": 1428
    },
    {
      "epoch": 0.9626136746379252,
      "grad_norm": 0.5382882952690125,
      "learning_rate": 6.273176496524674e-08,
      "loss": 0.2358,
      "step": 1429
    },
    {
      "epoch": 0.9632873021219266,
      "grad_norm": 0.5217492580413818,
      "learning_rate": 6.047439037019236e-08,
      "loss": 0.2605,
      "step": 1430
    },
    {
      "epoch": 0.9639609296059279,
      "grad_norm": 0.571123480796814,
      "learning_rate": 5.82582143560198e-08,
      "loss": 0.2446,
      "step": 1431
    },
    {
      "epoch": 0.9646345570899293,
      "grad_norm": 0.5425116419792175,
      "learning_rate": 5.6083249195463536e-08,
      "loss": 0.2603,
      "step": 1432
    },
    {
      "epoch": 0.9653081845739306,
      "grad_norm": 0.5614632964134216,
      "learning_rate": 5.394950693303646e-08,
      "loss": 0.3064,
      "step": 1433
    },
    {
      "epoch": 0.965981812057932,
      "grad_norm": 0.5277864336967468,
      "learning_rate": 5.18569993849724e-08,
      "loss": 0.2733,
      "step": 1434
    },
    {
      "epoch": 0.9666554395419333,
      "grad_norm": 0.5147943496704102,
      "learning_rate": 4.9805738139150346e-08,
      "loss": 0.2231,
      "step": 1435
    },
    {
      "epoch": 0.9673290670259347,
      "grad_norm": 0.5468779802322388,
      "learning_rate": 4.7795734555039496e-08,
      "loss": 0.2465,
      "step": 1436
    },
    {
      "epoch": 0.968002694509936,
      "grad_norm": 0.5027517080307007,
      "learning_rate": 4.5826999763630973e-08,
      "loss": 0.2335,
      "step": 1437
    },
    {
      "epoch": 0.9686763219939374,
      "grad_norm": 0.5079773664474487,
      "learning_rate": 4.389954466737789e-08,
      "loss": 0.2536,
      "step": 1438
    },
    {
      "epoch": 0.9693499494779387,
      "grad_norm": 0.5721175074577332,
      "learning_rate": 4.201337994013538e-08,
      "loss": 0.246,
      "step": 1439
    },
    {
      "epoch": 0.9700235769619401,
      "grad_norm": 0.488193154335022,
      "learning_rate": 4.016851602709898e-08,
      "loss": 0.2314,
      "step": 1440
    },
    {
      "epoch": 0.9706972044459414,
      "grad_norm": 0.5057032704353333,
      "learning_rate": 3.836496314475135e-08,
      "loss": 0.2331,
      "step": 1441
    },
    {
      "epoch": 0.9713708319299428,
      "grad_norm": 0.5356112718582153,
      "learning_rate": 3.660273128080149e-08,
      "loss": 0.2551,
      "step": 1442
    },
    {
      "epoch": 0.9720444594139441,
      "grad_norm": 0.5309187769889832,
      "learning_rate": 3.4881830194131435e-08,
      "loss": 0.2541,
      "step": 1443
    },
    {
      "epoch": 0.9727180868979455,
      "grad_norm": 0.5499880909919739,
      "learning_rate": 3.32022694147413e-08,
      "loss": 0.2915,
      "step": 1444
    },
    {
      "epoch": 0.9733917143819468,
      "grad_norm": 0.5452831387519836,
      "learning_rate": 3.156405824369768e-08,
      "loss": 0.2894,
      "step": 1445
    },
    {
      "epoch": 0.9740653418659482,
      "grad_norm": 0.529076874256134,
      "learning_rate": 2.9967205753081153e-08,
      "loss": 0.2786,
      "step": 1446
    },
    {
      "epoch": 0.9747389693499495,
      "grad_norm": 0.5517397522926331,
      "learning_rate": 2.8411720785935514e-08,
      "loss": 0.2531,
      "step": 1447
    },
    {
      "epoch": 0.9754125968339509,
      "grad_norm": 0.5285606384277344,
      "learning_rate": 2.6897611956221148e-08,
      "loss": 0.22,
      "step": 1448
    },
    {
      "epoch": 0.9760862243179522,
      "grad_norm": 0.4788746237754822,
      "learning_rate": 2.5424887648765048e-08,
      "loss": 0.2387,
      "step": 1449
    },
    {
      "epoch": 0.9767598518019536,
      "grad_norm": 0.5199988484382629,
      "learning_rate": 2.3993556019215045e-08,
      "loss": 0.2461,
      "step": 1450
    },
    {
      "epoch": 0.9774334792859549,
      "grad_norm": 0.5488206744194031,
      "learning_rate": 2.260362499399482e-08,
      "loss": 0.266,
      "step": 1451
    },
    {
      "epoch": 0.9781071067699563,
      "grad_norm": 0.4709354639053345,
      "learning_rate": 2.125510227025895e-08,
      "loss": 0.1913,
      "step": 1452
    },
    {
      "epoch": 0.9787807342539575,
      "grad_norm": 0.5103728771209717,
      "learning_rate": 1.9947995315853784e-08,
      "loss": 0.2363,
      "step": 1453
    },
    {
      "epoch": 0.9794543617379589,
      "grad_norm": 0.4883621037006378,
      "learning_rate": 1.868231136927162e-08,
      "loss": 0.2278,
      "step": 1454
    },
    {
      "epoch": 0.9801279892219602,
      "grad_norm": 0.4752127230167389,
      "learning_rate": 1.7458057439611597e-08,
      "loss": 0.2011,
      "step": 1455
    },
    {
      "epoch": 0.9808016167059616,
      "grad_norm": 0.5171606540679932,
      "learning_rate": 1.6275240306544703e-08,
      "loss": 0.2515,
      "step": 1456
    },
    {
      "epoch": 0.9814752441899629,
      "grad_norm": 0.52321857213974,
      "learning_rate": 1.5133866520271322e-08,
      "loss": 0.247,
      "step": 1457
    },
    {
      "epoch": 0.9821488716739643,
      "grad_norm": 0.49326056241989136,
      "learning_rate": 1.4033942401487921e-08,
      "loss": 0.2492,
      "step": 1458
    },
    {
      "epoch": 0.9828224991579656,
      "grad_norm": 0.5408816933631897,
      "learning_rate": 1.2975474041349577e-08,
      "loss": 0.2642,
      "step": 1459
    },
    {
      "epoch": 0.983496126641967,
      "grad_norm": 0.4778939485549927,
      "learning_rate": 1.1958467301440013e-08,
      "loss": 0.2245,
      "step": 1460
    },
    {
      "epoch": 0.9841697541259683,
      "grad_norm": 0.5252901911735535,
      "learning_rate": 1.0982927813735777e-08,
      "loss": 0.287,
      "step": 1461
    },
    {
      "epoch": 0.9848433816099696,
      "grad_norm": 0.5245988368988037,
      "learning_rate": 1.0048860980575447e-08,
      "loss": 0.272,
      "step": 1462
    },
    {
      "epoch": 0.985517009093971,
      "grad_norm": 0.536003053188324,
      "learning_rate": 9.156271974632146e-09,
      "loss": 0.2393,
      "step": 1463
    },
    {
      "epoch": 0.9861906365779723,
      "grad_norm": 0.5178948640823364,
      "learning_rate": 8.305165738881903e-09,
      "loss": 0.251,
      "step": 1464
    },
    {
      "epoch": 0.9868642640619737,
      "grad_norm": 0.5255127549171448,
      "learning_rate": 7.495546986578671e-09,
      "loss": 0.238,
      "step": 1465
    },
    {
      "epoch": 0.987537891545975,
      "grad_norm": 0.49453413486480713,
      "learning_rate": 6.72742020122602e-09,
      "loss": 0.2145,
      "step": 1466
    },
    {
      "epoch": 0.9882115190299764,
      "grad_norm": 0.47628140449523926,
      "learning_rate": 6.000789636554649e-09,
      "loss": 0.2091,
      "step": 1467
    },
    {
      "epoch": 0.9888851465139777,
      "grad_norm": 0.5566253662109375,
      "learning_rate": 5.315659316495747e-09,
      "loss": 0.2696,
      "step": 1468
    },
    {
      "epoch": 0.9895587739979791,
      "grad_norm": 0.5432794094085693,
      "learning_rate": 4.6720330351635034e-09,
      "loss": 0.237,
      "step": 1469
    },
    {
      "epoch": 0.9902324014819804,
      "grad_norm": 0.5230391621589661,
      "learning_rate": 4.069914356827631e-09,
      "loss": 0.2453,
      "step": 1470
    },
    {
      "epoch": 0.9909060289659818,
      "grad_norm": 0.5614542961120605,
      "learning_rate": 3.509306615900043e-09,
      "loss": 0.2528,
      "step": 1471
    },
    {
      "epoch": 0.9915796564499831,
      "grad_norm": 0.5609020590782166,
      "learning_rate": 2.9902129169123717e-09,
      "loss": 0.2472,
      "step": 1472
    },
    {
      "epoch": 0.9922532839339845,
      "grad_norm": 0.5249729752540588,
      "learning_rate": 2.512636134500146e-09,
      "loss": 0.2489,
      "step": 1473
    },
    {
      "epoch": 0.9929269114179858,
      "grad_norm": 0.5459350943565369,
      "learning_rate": 2.07657891338614e-09,
      "loss": 0.2335,
      "step": 1474
    },
    {
      "epoch": 0.9936005389019872,
      "grad_norm": 0.5332579016685486,
      "learning_rate": 1.6820436683670503e-09,
      "loss": 0.2527,
      "step": 1475
    },
    {
      "epoch": 0.9942741663859885,
      "grad_norm": 0.5601089596748352,
      "learning_rate": 1.3290325842976736e-09,
      "loss": 0.2471,
      "step": 1476
    },
    {
      "epoch": 0.9949477938699899,
      "grad_norm": 0.5015659332275391,
      "learning_rate": 1.0175476160834142e-09,
      "loss": 0.2318,
      "step": 1477
    },
    {
      "epoch": 0.9956214213539912,
      "grad_norm": 0.4812217056751251,
      "learning_rate": 7.475904886636298e-10,
      "loss": 0.2066,
      "step": 1478
    },
    {
      "epoch": 0.9962950488379926,
      "grad_norm": 0.5097085237503052,
      "learning_rate": 5.191626970066366e-10,
      "loss": 0.2859,
      "step": 1479
    },
    {
      "epoch": 0.9969686763219939,
      "grad_norm": 0.5650218725204468,
      "learning_rate": 3.322655060988833e-10,
      "loss": 0.2942,
      "step": 1480
    },
    {
      "epoch": 0.9976423038059953,
      "grad_norm": 0.5577125549316406,
      "learning_rate": 1.8689995093912338e-10,
      "loss": 0.2726,
      "step": 1481
    },
    {
      "epoch": 0.9983159312899966,
      "grad_norm": 0.5873110294342041,
      "learning_rate": 8.306683653175329e-11,
      "loss": 0.2565,
      "step": 1482
    },
    {
      "epoch": 0.998989558773998,
      "grad_norm": 0.5006901621818542,
      "learning_rate": 2.0766737883481934e-11,
      "loss": 0.2353,
      "step": 1483
    },
    {
      "epoch": 0.9996631862579993,
      "grad_norm": 0.5085142254829407,
      "learning_rate": 0.0,
      "loss": 0.2232,
      "step": 1484
    }
  ],
  "logging_steps": 1,
  "max_steps": 1484,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.85361893151395e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}