{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 216015,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006943962224845497,
      "grad_norm": 2.206239700317383,
      "learning_rate": 0.00029930699256996036,
      "loss": 4.6213,
      "step": 500
    },
    {
      "epoch": 0.013887924449690994,
      "grad_norm": 1.519214153289795,
      "learning_rate": 0.0002986125963474758,
      "loss": 4.5637,
      "step": 1000
    },
    {
      "epoch": 0.02083188667453649,
      "grad_norm": 2.3049542903900146,
      "learning_rate": 0.0002979182001249913,
      "loss": 4.514,
      "step": 1500
    },
    {
      "epoch": 0.027775848899381988,
      "grad_norm": 2.0959904193878174,
      "learning_rate": 0.00029722380390250673,
      "loss": 4.4586,
      "step": 2000
    },
    {
      "epoch": 0.034719811124227486,
      "grad_norm": 2.667476177215576,
      "learning_rate": 0.0002965294076800222,
      "loss": 4.4977,
      "step": 2500
    },
    {
      "epoch": 0.04166377334907298,
      "grad_norm": 1.8514024019241333,
      "learning_rate": 0.00029583501145753765,
      "loss": 4.4347,
      "step": 3000
    },
    {
      "epoch": 0.04860773557391848,
      "grad_norm": 1.47382390499115,
      "learning_rate": 0.00029514061523505306,
      "loss": 4.4555,
      "step": 3500
    },
    {
      "epoch": 0.055551697798763976,
      "grad_norm": 1.5439454317092896,
      "learning_rate": 0.0002944462190125685,
      "loss": 4.4132,
      "step": 4000
    },
    {
      "epoch": 0.06249566002360947,
      "grad_norm": 2.2414743900299072,
      "learning_rate": 0.000293751822790084,
      "loss": 4.4148,
      "step": 4500
    },
    {
      "epoch": 0.06943962224845497,
      "grad_norm": 2.522726058959961,
      "learning_rate": 0.00029305742656759944,
      "loss": 4.3732,
      "step": 5000
    },
    {
      "epoch": 0.07638358447330046,
      "grad_norm": 2.1512997150421143,
      "learning_rate": 0.0002923630303451149,
      "loss": 4.3563,
      "step": 5500
    },
    {
      "epoch": 0.08332754669814596,
      "grad_norm": 1.3137620687484741,
      "learning_rate": 0.00029166863412263035,
      "loss": 4.3309,
      "step": 6000
    },
    {
      "epoch": 0.09027150892299146,
      "grad_norm": 2.2317123413085938,
      "learning_rate": 0.0002909742379001458,
      "loss": 4.335,
      "step": 6500
    },
    {
      "epoch": 0.09721547114783696,
      "grad_norm": 1.4936091899871826,
      "learning_rate": 0.0002902798416776612,
      "loss": 4.3287,
      "step": 7000
    },
    {
      "epoch": 0.10415943337268245,
      "grad_norm": 1.4944448471069336,
      "learning_rate": 0.0002895854454551767,
      "loss": 4.2728,
      "step": 7500
    },
    {
      "epoch": 0.11110339559752795,
      "grad_norm": 2.103372573852539,
      "learning_rate": 0.00028889104923269214,
      "loss": 4.2945,
      "step": 8000
    },
    {
      "epoch": 0.11804735782237345,
      "grad_norm": 1.6695603132247925,
      "learning_rate": 0.0002881966530102076,
      "loss": 4.3126,
      "step": 8500
    },
    {
      "epoch": 0.12499132004721894,
      "grad_norm": 2.2535531520843506,
      "learning_rate": 0.00028750225678772306,
      "loss": 4.3014,
      "step": 9000
    },
    {
      "epoch": 0.13193528227206444,
      "grad_norm": 2.5216352939605713,
      "learning_rate": 0.0002868078605652385,
      "loss": 4.233,
      "step": 9500
    },
    {
      "epoch": 0.13887924449690994,
      "grad_norm": 2.3059794902801514,
      "learning_rate": 0.0002861134643427539,
      "loss": 4.2875,
      "step": 10000
    },
    {
      "epoch": 0.14582320672175544,
      "grad_norm": 2.7461228370666504,
      "learning_rate": 0.0002854190681202694,
      "loss": 4.2872,
      "step": 10500
    },
    {
      "epoch": 0.15276716894660092,
      "grad_norm": 1.7328039407730103,
      "learning_rate": 0.00028472467189778484,
      "loss": 4.2354,
      "step": 11000
    },
    {
      "epoch": 0.15971113117144642,
      "grad_norm": 1.904403805732727,
      "learning_rate": 0.0002840302756753003,
      "loss": 4.2339,
      "step": 11500
    },
    {
      "epoch": 0.16665509339629192,
      "grad_norm": 1.4885430335998535,
      "learning_rate": 0.00028333587945281576,
      "loss": 4.2214,
      "step": 12000
    },
    {
      "epoch": 0.17359905562113742,
      "grad_norm": 1.5213295221328735,
      "learning_rate": 0.0002826414832303312,
      "loss": 4.1642,
      "step": 12500
    },
    {
      "epoch": 0.18054301784598292,
      "grad_norm": 1.5499557256698608,
      "learning_rate": 0.0002819470870078466,
      "loss": 4.2083,
      "step": 13000
    },
    {
      "epoch": 0.18748698007082842,
      "grad_norm": 1.7087458372116089,
      "learning_rate": 0.0002812526907853621,
      "loss": 4.192,
      "step": 13500
    },
    {
      "epoch": 0.19443094229567393,
      "grad_norm": 1.4186779260635376,
      "learning_rate": 0.00028055829456287754,
      "loss": 4.1686,
      "step": 14000
    },
    {
      "epoch": 0.2013749045205194,
      "grad_norm": 1.7402822971343994,
      "learning_rate": 0.000279863898340393,
      "loss": 4.1988,
      "step": 14500
    },
    {
      "epoch": 0.2083188667453649,
      "grad_norm": 1.821722149848938,
      "learning_rate": 0.00027916950211790846,
      "loss": 4.1941,
      "step": 15000
    },
    {
      "epoch": 0.2152628289702104,
      "grad_norm": 1.207729458808899,
      "learning_rate": 0.0002784751058954239,
      "loss": 4.1868,
      "step": 15500
    },
    {
      "epoch": 0.2222067911950559,
      "grad_norm": 1.804909110069275,
      "learning_rate": 0.0002777807096729393,
      "loss": 4.1657,
      "step": 16000
    },
    {
      "epoch": 0.2291507534199014,
      "grad_norm": 2.126279830932617,
      "learning_rate": 0.0002770863134504548,
      "loss": 4.1083,
      "step": 16500
    },
    {
      "epoch": 0.2360947156447469,
      "grad_norm": 1.3076670169830322,
      "learning_rate": 0.00027639191722797024,
      "loss": 4.1446,
      "step": 17000
    },
    {
      "epoch": 0.24303867786959238,
      "grad_norm": 2.152813196182251,
      "learning_rate": 0.0002756975210054857,
      "loss": 4.1444,
      "step": 17500
    },
    {
      "epoch": 0.24998264009443788,
      "grad_norm": 1.9263052940368652,
      "learning_rate": 0.00027500312478300116,
      "loss": 4.1581,
      "step": 18000
    },
    {
      "epoch": 0.2569266023192834,
      "grad_norm": 1.7251839637756348,
      "learning_rate": 0.0002743087285605166,
      "loss": 4.1468,
      "step": 18500
    },
    {
      "epoch": 0.2638705645441289,
      "grad_norm": 1.2336386442184448,
      "learning_rate": 0.0002736143323380321,
      "loss": 4.1253,
      "step": 19000
    },
    {
      "epoch": 0.2708145267689744,
      "grad_norm": 1.2728581428527832,
      "learning_rate": 0.0002729199361155475,
      "loss": 4.0935,
      "step": 19500
    },
    {
      "epoch": 0.2777584889938199,
      "grad_norm": 1.6062270402908325,
      "learning_rate": 0.00027222553989306294,
      "loss": 4.1383,
      "step": 20000
    },
    {
      "epoch": 0.2847024512186654,
      "grad_norm": 1.4133198261260986,
      "learning_rate": 0.0002715311436705784,
      "loss": 4.1044,
      "step": 20500
    },
    {
      "epoch": 0.2916464134435109,
      "grad_norm": 1.421473741531372,
      "learning_rate": 0.00027083674744809386,
      "loss": 4.103,
      "step": 21000
    },
    {
      "epoch": 0.2985903756683564,
      "grad_norm": 1.92391836643219,
      "learning_rate": 0.0002701423512256093,
      "loss": 4.0771,
      "step": 21500
    },
    {
      "epoch": 0.30553433789320184,
      "grad_norm": 2.6180472373962402,
      "learning_rate": 0.0002694479550031248,
      "loss": 4.0713,
      "step": 22000
    },
    {
      "epoch": 0.31247830011804734,
      "grad_norm": 2.3145902156829834,
      "learning_rate": 0.0002687535587806402,
      "loss": 4.0287,
      "step": 22500
    },
    {
      "epoch": 0.31942226234289284,
      "grad_norm": 1.9222602844238281,
      "learning_rate": 0.00026805916255815565,
      "loss": 4.0714,
      "step": 23000
    },
    {
      "epoch": 0.32636622456773834,
      "grad_norm": 1.6106317043304443,
      "learning_rate": 0.0002673647663356711,
      "loss": 4.0375,
      "step": 23500
    },
    {
      "epoch": 0.33331018679258384,
      "grad_norm": 1.6297123432159424,
      "learning_rate": 0.00026667037011318657,
      "loss": 4.0668,
      "step": 24000
    },
    {
      "epoch": 0.34025414901742934,
      "grad_norm": 2.0038726329803467,
      "learning_rate": 0.000265975973890702,
      "loss": 4.0266,
      "step": 24500
    },
    {
      "epoch": 0.34719811124227484,
      "grad_norm": 1.7728261947631836,
      "learning_rate": 0.0002652815776682175,
      "loss": 4.0499,
      "step": 25000
    },
    {
      "epoch": 0.35414207346712034,
      "grad_norm": 1.9266184568405151,
      "learning_rate": 0.0002645871814457329,
      "loss": 4.0055,
      "step": 25500
    },
    {
      "epoch": 0.36108603569196585,
      "grad_norm": 3.5189244747161865,
      "learning_rate": 0.00026389278522324835,
      "loss": 4.004,
      "step": 26000
    },
    {
      "epoch": 0.36802999791681135,
      "grad_norm": 1.9975138902664185,
      "learning_rate": 0.0002631983890007638,
      "loss": 3.9998,
      "step": 26500
    },
    {
      "epoch": 0.37497396014165685,
      "grad_norm": 3.087763547897339,
      "learning_rate": 0.00026250399277827927,
      "loss": 4.0146,
      "step": 27000
    },
    {
      "epoch": 0.38191792236650235,
      "grad_norm": 1.297499179840088,
      "learning_rate": 0.0002618095965557947,
      "loss": 4.0206,
      "step": 27500
    },
    {
      "epoch": 0.38886188459134785,
      "grad_norm": 1.4603493213653564,
      "learning_rate": 0.0002611152003333102,
      "loss": 4.0337,
      "step": 28000
    },
    {
      "epoch": 0.3958058468161933,
      "grad_norm": 1.5912282466888428,
      "learning_rate": 0.0002604208041108256,
      "loss": 3.9878,
      "step": 28500
    },
    {
      "epoch": 0.4027498090410388,
      "grad_norm": 1.4256983995437622,
      "learning_rate": 0.00025972640788834105,
      "loss": 4.0107,
      "step": 29000
    },
    {
      "epoch": 0.4096937712658843,
      "grad_norm": 1.6172006130218506,
      "learning_rate": 0.0002590320116658565,
      "loss": 4.0011,
      "step": 29500
    },
    {
      "epoch": 0.4166377334907298,
      "grad_norm": 2.2637939453125,
      "learning_rate": 0.00025833761544337197,
      "loss": 4.0057,
      "step": 30000
    },
    {
      "epoch": 0.4235816957155753,
      "grad_norm": 1.6595959663391113,
      "learning_rate": 0.00025764321922088743,
      "loss": 4.0112,
      "step": 30500
    },
    {
      "epoch": 0.4305256579404208,
      "grad_norm": 1.7675671577453613,
      "learning_rate": 0.0002569488229984029,
      "loss": 3.9575,
      "step": 31000
    },
    {
      "epoch": 0.4374696201652663,
      "grad_norm": 1.9230527877807617,
      "learning_rate": 0.00025625442677591835,
      "loss": 3.9734,
      "step": 31500
    },
    {
      "epoch": 0.4444135823901118,
      "grad_norm": 1.6587070226669312,
      "learning_rate": 0.00025556003055343375,
      "loss": 3.973,
      "step": 32000
    },
    {
      "epoch": 0.4513575446149573,
      "grad_norm": 1.8445744514465332,
      "learning_rate": 0.0002548656343309492,
      "loss": 3.9792,
      "step": 32500
    },
    {
      "epoch": 0.4583015068398028,
      "grad_norm": 2.5224626064300537,
      "learning_rate": 0.00025417123810846467,
      "loss": 3.9501,
      "step": 33000
    },
    {
      "epoch": 0.4652454690646483,
      "grad_norm": 1.8237272500991821,
      "learning_rate": 0.00025347684188598013,
      "loss": 3.9696,
      "step": 33500
    },
    {
      "epoch": 0.4721894312894938,
      "grad_norm": 3.2028214931488037,
      "learning_rate": 0.0002527824456634956,
      "loss": 3.9386,
      "step": 34000
    },
    {
      "epoch": 0.47913339351433926,
      "grad_norm": 1.7500147819519043,
      "learning_rate": 0.00025208804944101105,
      "loss": 3.9328,
      "step": 34500
    },
    {
      "epoch": 0.48607735573918476,
      "grad_norm": 1.8961389064788818,
      "learning_rate": 0.00025139365321852645,
      "loss": 3.9132,
      "step": 35000
    },
    {
      "epoch": 0.49302131796403026,
      "grad_norm": 1.3247839212417603,
      "learning_rate": 0.0002506992569960419,
      "loss": 3.9139,
      "step": 35500
    },
    {
      "epoch": 0.49996528018887576,
      "grad_norm": 1.6709811687469482,
      "learning_rate": 0.00025000486077355737,
      "loss": 3.9646,
      "step": 36000
    },
    {
      "epoch": 0.5069092424137213,
      "grad_norm": 1.717537760734558,
      "learning_rate": 0.00024931046455107283,
      "loss": 3.9328,
      "step": 36500
    },
    {
      "epoch": 0.5138532046385668,
      "grad_norm": 2.0001790523529053,
      "learning_rate": 0.0002486160683285883,
      "loss": 3.9522,
      "step": 37000
    },
    {
      "epoch": 0.5207971668634123,
      "grad_norm": 1.53634774684906,
      "learning_rate": 0.00024792167210610375,
      "loss": 3.8908,
      "step": 37500
    },
    {
      "epoch": 0.5277411290882578,
      "grad_norm": 1.6681393384933472,
      "learning_rate": 0.00024722727588361916,
      "loss": 3.9057,
      "step": 38000
    },
    {
      "epoch": 0.5346850913131033,
      "grad_norm": 1.4480671882629395,
      "learning_rate": 0.0002465328796611346,
      "loss": 3.9144,
      "step": 38500
    },
    {
      "epoch": 0.5416290535379488,
      "grad_norm": 2.7067551612854004,
      "learning_rate": 0.0002458384834386501,
      "loss": 3.9379,
      "step": 39000
    },
    {
      "epoch": 0.5485730157627943,
      "grad_norm": 1.919639229774475,
      "learning_rate": 0.00024514408721616553,
      "loss": 3.8559,
      "step": 39500
    },
    {
      "epoch": 0.5555169779876398,
      "grad_norm": 1.9291149377822876,
      "learning_rate": 0.000244449690993681,
      "loss": 3.8821,
      "step": 40000
    },
    {
      "epoch": 0.5624609402124853,
      "grad_norm": 2.4123694896698,
      "learning_rate": 0.00024375529477119642,
      "loss": 3.9055,
      "step": 40500
    },
    {
      "epoch": 0.5694049024373308,
      "grad_norm": 1.5772641897201538,
      "learning_rate": 0.00024306089854871186,
      "loss": 3.9045,
      "step": 41000
    },
    {
      "epoch": 0.5763488646621763,
      "grad_norm": 2.2179367542266846,
      "learning_rate": 0.00024236650232622732,
      "loss": 3.9022,
      "step": 41500
    },
    {
      "epoch": 0.5832928268870218,
      "grad_norm": 1.1816768646240234,
      "learning_rate": 0.00024167210610374278,
      "loss": 3.8836,
      "step": 42000
    },
    {
      "epoch": 0.5902367891118673,
      "grad_norm": 1.7630631923675537,
      "learning_rate": 0.00024097770988125823,
      "loss": 3.8657,
      "step": 42500
    },
    {
      "epoch": 0.5971807513367128,
      "grad_norm": 1.777095913887024,
      "learning_rate": 0.00024028331365877367,
      "loss": 3.8529,
      "step": 43000
    },
    {
      "epoch": 0.6041247135615583,
      "grad_norm": 1.7181869745254517,
      "learning_rate": 0.00023958891743628913,
      "loss": 3.8781,
      "step": 43500
    },
    {
      "epoch": 0.6110686757864037,
      "grad_norm": 1.855504035949707,
      "learning_rate": 0.00023889452121380459,
      "loss": 3.8391,
      "step": 44000
    },
    {
      "epoch": 0.6180126380112492,
      "grad_norm": 1.7183347940444946,
      "learning_rate": 0.00023820012499132002,
      "loss": 3.836,
      "step": 44500
    },
    {
      "epoch": 0.6249566002360947,
      "grad_norm": 1.8155463933944702,
      "learning_rate": 0.00023750572876883548,
      "loss": 3.8238,
      "step": 45000
    },
    {
      "epoch": 0.6319005624609402,
      "grad_norm": 1.4531205892562866,
      "learning_rate": 0.00023681133254635094,
      "loss": 3.8348,
      "step": 45500
    },
    {
      "epoch": 0.6388445246857857,
      "grad_norm": 2.0805277824401855,
      "learning_rate": 0.00023611693632386637,
      "loss": 3.8441,
      "step": 46000
    },
    {
      "epoch": 0.6457884869106312,
      "grad_norm": 2.2948648929595947,
      "learning_rate": 0.00023542254010138183,
      "loss": 3.8511,
      "step": 46500
    },
    {
      "epoch": 0.6527324491354767,
      "grad_norm": 1.958189606666565,
      "learning_rate": 0.0002347281438788973,
      "loss": 3.8678,
      "step": 47000
    },
    {
      "epoch": 0.6596764113603222,
      "grad_norm": 1.9029563665390015,
      "learning_rate": 0.00023403374765641272,
      "loss": 3.8204,
      "step": 47500
    },
    {
      "epoch": 0.6666203735851677,
      "grad_norm": 1.6925806999206543,
      "learning_rate": 0.00023333935143392818,
      "loss": 3.861,
      "step": 48000
    },
    {
      "epoch": 0.6735643358100132,
      "grad_norm": 2.417433023452759,
      "learning_rate": 0.00023264495521144364,
      "loss": 3.8151,
      "step": 48500
    },
    {
      "epoch": 0.6805082980348587,
      "grad_norm": 1.94263756275177,
      "learning_rate": 0.00023195055898895907,
      "loss": 3.8351,
      "step": 49000
    },
    {
      "epoch": 0.6874522602597042,
      "grad_norm": 2.0970757007598877,
      "learning_rate": 0.00023125616276647453,
      "loss": 3.8728,
      "step": 49500
    },
    {
      "epoch": 0.6943962224845497,
      "grad_norm": 1.8286621570587158,
      "learning_rate": 0.00023056176654399,
      "loss": 3.8262,
      "step": 50000
    },
    {
      "epoch": 0.7013401847093952,
      "grad_norm": 1.3233591318130493,
      "learning_rate": 0.00022986737032150542,
      "loss": 3.8186,
      "step": 50500
    },
    {
      "epoch": 0.7082841469342407,
      "grad_norm": 1.760081171989441,
      "learning_rate": 0.00022917297409902088,
      "loss": 3.8268,
      "step": 51000
    },
    {
      "epoch": 0.7152281091590862,
      "grad_norm": 2.040560722351074,
      "learning_rate": 0.00022847857787653634,
      "loss": 3.8425,
      "step": 51500
    },
    {
      "epoch": 0.7221720713839317,
      "grad_norm": 2.493685007095337,
      "learning_rate": 0.00022778418165405177,
      "loss": 3.8362,
      "step": 52000
    },
    {
      "epoch": 0.7291160336087772,
      "grad_norm": 1.7292836904525757,
      "learning_rate": 0.00022708978543156723,
      "loss": 3.8317,
      "step": 52500
    },
    {
      "epoch": 0.7360599958336227,
      "grad_norm": 2.2565951347351074,
      "learning_rate": 0.0002263953892090827,
      "loss": 3.7835,
      "step": 53000
    },
    {
      "epoch": 0.7430039580584682,
      "grad_norm": 1.6440356969833374,
      "learning_rate": 0.00022570099298659812,
      "loss": 3.8079,
      "step": 53500
    },
    {
      "epoch": 0.7499479202833137,
      "grad_norm": 1.8633214235305786,
      "learning_rate": 0.00022500659676411358,
      "loss": 3.803,
      "step": 54000
    },
    {
      "epoch": 0.7568918825081592,
      "grad_norm": 2.401519775390625,
      "learning_rate": 0.00022431220054162904,
      "loss": 3.7996,
      "step": 54500
    },
    {
      "epoch": 0.7638358447330047,
      "grad_norm": 1.3482192754745483,
      "learning_rate": 0.0002236178043191445,
      "loss": 3.7957,
      "step": 55000
    },
    {
      "epoch": 0.7707798069578502,
      "grad_norm": 2.4375321865081787,
      "learning_rate": 0.00022292340809665993,
      "loss": 3.8011,
      "step": 55500
    },
    {
      "epoch": 0.7777237691826957,
      "grad_norm": 1.4207526445388794,
      "learning_rate": 0.0002222290118741754,
      "loss": 3.7809,
      "step": 56000
    },
    {
      "epoch": 0.7846677314075411,
      "grad_norm": 2.166013717651367,
      "learning_rate": 0.00022153461565169085,
      "loss": 3.8016,
      "step": 56500
    },
    {
      "epoch": 0.7916116936323866,
      "grad_norm": 1.4218441247940063,
      "learning_rate": 0.00022084021942920628,
      "loss": 3.7636,
      "step": 57000
    },
    {
      "epoch": 0.7985556558572321,
      "grad_norm": 1.5661506652832031,
      "learning_rate": 0.00022014582320672174,
      "loss": 3.7958,
      "step": 57500
    },
    {
      "epoch": 0.8054996180820776,
      "grad_norm": 1.311798095703125,
      "learning_rate": 0.0002194514269842372,
      "loss": 3.7593,
      "step": 58000
    },
    {
      "epoch": 0.8124435803069231,
      "grad_norm": 1.3802398443222046,
      "learning_rate": 0.00021875703076175264,
      "loss": 3.7758,
      "step": 58500
    },
    {
      "epoch": 0.8193875425317686,
      "grad_norm": 1.7688322067260742,
      "learning_rate": 0.0002180626345392681,
      "loss": 3.7685,
      "step": 59000
    },
    {
      "epoch": 0.8263315047566141,
      "grad_norm": 1.8496917486190796,
      "learning_rate": 0.00021736823831678355,
      "loss": 3.8083,
      "step": 59500
    },
    {
      "epoch": 0.8332754669814596,
      "grad_norm": 1.2840275764465332,
      "learning_rate": 0.00021667384209429899,
      "loss": 3.7912,
      "step": 60000
    },
    {
      "epoch": 0.8402194292063051,
      "grad_norm": 1.4152112007141113,
      "learning_rate": 0.00021597944587181445,
      "loss": 3.75,
      "step": 60500
    },
    {
      "epoch": 0.8471633914311506,
      "grad_norm": 2.001692771911621,
      "learning_rate": 0.0002152850496493299,
      "loss": 3.7748,
      "step": 61000
    },
    {
      "epoch": 0.8541073536559961,
      "grad_norm": 2.6924116611480713,
      "learning_rate": 0.00021459065342684534,
      "loss": 3.7576,
      "step": 61500
    },
    {
      "epoch": 0.8610513158808416,
      "grad_norm": 1.5775929689407349,
      "learning_rate": 0.0002138962572043608,
      "loss": 3.7357,
      "step": 62000
    },
    {
      "epoch": 0.8679952781056871,
      "grad_norm": 2.122657060623169,
      "learning_rate": 0.00021320186098187626,
      "loss": 3.7513,
      "step": 62500
    },
    {
      "epoch": 0.8749392403305326,
      "grad_norm": 1.8863738775253296,
      "learning_rate": 0.0002125074647593917,
      "loss": 3.7259,
      "step": 63000
    },
    {
      "epoch": 0.8818832025553781,
      "grad_norm": 1.46346914768219,
      "learning_rate": 0.00021181306853690715,
      "loss": 3.7851,
      "step": 63500
    },
    {
      "epoch": 0.8888271647802236,
      "grad_norm": 2.3657708168029785,
      "learning_rate": 0.0002111186723144226,
      "loss": 3.7523,
      "step": 64000
    },
    {
      "epoch": 0.8957711270050691,
      "grad_norm": 1.5897114276885986,
      "learning_rate": 0.00021042427609193804,
      "loss": 3.7526,
      "step": 64500
    },
    {
      "epoch": 0.9027150892299146,
      "grad_norm": 1.8869891166687012,
      "learning_rate": 0.0002097298798694535,
      "loss": 3.7406,
      "step": 65000
    },
    {
      "epoch": 0.9096590514547601,
      "grad_norm": 1.891735315322876,
      "learning_rate": 0.00020903548364696896,
      "loss": 3.7229,
      "step": 65500
    },
    {
      "epoch": 0.9166030136796056,
      "grad_norm": 1.4305230379104614,
      "learning_rate": 0.0002083410874244844,
      "loss": 3.7712,
      "step": 66000
    },
    {
      "epoch": 0.9235469759044511,
      "grad_norm": 1.571385145187378,
      "learning_rate": 0.00020764669120199985,
      "loss": 3.739,
      "step": 66500
    },
    {
      "epoch": 0.9304909381292966,
      "grad_norm": 1.64103102684021,
      "learning_rate": 0.0002069522949795153,
      "loss": 3.7291,
      "step": 67000
    },
    {
      "epoch": 0.9374349003541421,
      "grad_norm": 1.683289647102356,
      "learning_rate": 0.00020625789875703077,
      "loss": 3.7349,
      "step": 67500
    },
    {
      "epoch": 0.9443788625789876,
      "grad_norm": 1.9319536685943604,
      "learning_rate": 0.0002055635025345462,
      "loss": 3.7298,
      "step": 68000
    },
    {
      "epoch": 0.9513228248038331,
      "grad_norm": 1.2139371633529663,
      "learning_rate": 0.00020486910631206166,
      "loss": 3.7077,
      "step": 68500
    },
    {
      "epoch": 0.9582667870286785,
      "grad_norm": 2.366407871246338,
      "learning_rate": 0.00020417471008957712,
      "loss": 3.7156,
      "step": 69000
    },
    {
      "epoch": 0.965210749253524,
      "grad_norm": 1.2618952989578247,
      "learning_rate": 0.00020348031386709255,
      "loss": 3.6964,
      "step": 69500
    },
    {
      "epoch": 0.9721547114783695,
      "grad_norm": 1.3639748096466064,
      "learning_rate": 0.000202785917644608,
      "loss": 3.7082,
      "step": 70000
    },
    {
      "epoch": 0.979098673703215,
      "grad_norm": 1.7581994533538818,
      "learning_rate": 0.00020209152142212347,
      "loss": 3.7073,
      "step": 70500
    },
    {
      "epoch": 0.9860426359280605,
      "grad_norm": 2.42798113822937,
      "learning_rate": 0.00020139712519963887,
      "loss": 3.7116,
      "step": 71000
    },
    {
      "epoch": 0.992986598152906,
      "grad_norm": 1.6432359218597412,
      "learning_rate": 0.00020070272897715436,
      "loss": 3.6997,
      "step": 71500
    },
    {
      "epoch": 0.9999305603777515,
      "grad_norm": 1.079620361328125,
      "learning_rate": 0.00020000833275466982,
      "loss": 3.6916,
      "step": 72000
    },
    {
      "epoch": 1.0,
      "eval_loss": 3.4631764888763428,
      "eval_rouge1": 0.04941977533882873,
      "eval_rouge2": 0.009172388840507199,
      "eval_rougeL": 0.04889121785661922,
      "eval_rougeLsum": 0.04914022943695945,
      "eval_runtime": 3964.6307,
      "eval_samples_per_second": 4.036,
      "eval_steps_per_second": 2.018,
      "step": 72005
    },
    {
      "epoch": 1.006874522602597,
      "grad_norm": 1.7696194648742676,
      "learning_rate": 0.00019931393653218523,
      "loss": 3.6153,
      "step": 72500
    },
    {
      "epoch": 1.0138184848274425,
      "grad_norm": 1.091058611869812,
      "learning_rate": 0.00019861954030970068,
      "loss": 3.6146,
      "step": 73000
    },
    {
      "epoch": 1.020762447052288,
      "grad_norm": 1.8638333082199097,
      "learning_rate": 0.00019792514408721617,
      "loss": 3.609,
      "step": 73500
    },
    {
      "epoch": 1.0277064092771335,
      "grad_norm": 2.0278666019439697,
      "learning_rate": 0.00019723074786473158,
      "loss": 3.6129,
      "step": 74000
    },
    {
      "epoch": 1.034650371501979,
      "grad_norm": 2.3207359313964844,
      "learning_rate": 0.00019653635164224704,
      "loss": 3.601,
      "step": 74500
    },
    {
      "epoch": 1.0415943337268245,
      "grad_norm": 1.4969432353973389,
      "learning_rate": 0.00019584195541976252,
      "loss": 3.604,
      "step": 75000
    },
    {
      "epoch": 1.04853829595167,
      "grad_norm": 1.567814588546753,
      "learning_rate": 0.00019514755919727793,
      "loss": 3.5989,
      "step": 75500
    },
    {
      "epoch": 1.0554822581765155,
      "grad_norm": 1.7745885848999023,
      "learning_rate": 0.00019445316297479339,
      "loss": 3.5884,
      "step": 76000
    },
    {
      "epoch": 1.062426220401361,
      "grad_norm": 1.9692455530166626,
      "learning_rate": 0.00019375876675230885,
      "loss": 3.6422,
      "step": 76500
    },
    {
      "epoch": 1.0693701826262065,
      "grad_norm": 1.5432820320129395,
      "learning_rate": 0.00019306437052982428,
      "loss": 3.5908,
      "step": 77000
    },
    {
      "epoch": 1.076314144851052,
      "grad_norm": 1.8240996599197388,
      "learning_rate": 0.00019236997430733974,
      "loss": 3.6187,
      "step": 77500
    },
    {
      "epoch": 1.0832581070758975,
      "grad_norm": 1.2577378749847412,
      "learning_rate": 0.0001916755780848552,
      "loss": 3.537,
      "step": 78000
    },
    {
      "epoch": 1.090202069300743,
      "grad_norm": 1.969597339630127,
      "learning_rate": 0.00019098118186237063,
      "loss": 3.6229,
      "step": 78500
    },
    {
      "epoch": 1.0971460315255885,
      "grad_norm": 2.464380979537964,
      "learning_rate": 0.0001902867856398861,
      "loss": 3.6009,
      "step": 79000
    },
    {
      "epoch": 1.104089993750434,
      "grad_norm": 2.1590375900268555,
      "learning_rate": 0.00018959238941740155,
      "loss": 3.5834,
      "step": 79500
    },
    {
      "epoch": 1.1110339559752795,
      "grad_norm": 1.2929880619049072,
      "learning_rate": 0.000188897993194917,
      "loss": 3.605,
      "step": 80000
    },
    {
      "epoch": 1.117977918200125,
      "grad_norm": 2.7584567070007324,
      "learning_rate": 0.00018820359697243244,
      "loss": 3.5863,
      "step": 80500
    },
    {
      "epoch": 1.1249218804249705,
      "grad_norm": 2.8105475902557373,
      "learning_rate": 0.0001875092007499479,
      "loss": 3.5646,
      "step": 81000
    },
    {
      "epoch": 1.131865842649816,
      "grad_norm": 1.6573877334594727,
      "learning_rate": 0.00018681480452746336,
      "loss": 3.6216,
      "step": 81500
    },
    {
      "epoch": 1.1388098048746615,
      "grad_norm": 1.3650224208831787,
      "learning_rate": 0.0001861204083049788,
      "loss": 3.6155,
      "step": 82000
    },
    {
      "epoch": 1.145753767099507,
      "grad_norm": 1.3206992149353027,
      "learning_rate": 0.00018542601208249425,
      "loss": 3.62,
      "step": 82500
    },
    {
      "epoch": 1.1526977293243525,
      "grad_norm": 2.5373497009277344,
      "learning_rate": 0.0001847316158600097,
      "loss": 3.595,
      "step": 83000
    },
    {
      "epoch": 1.159641691549198,
      "grad_norm": 1.503808856010437,
      "learning_rate": 0.00018403721963752514,
      "loss": 3.6213,
      "step": 83500
    },
    {
      "epoch": 1.1665856537740436,
      "grad_norm": 2.725497007369995,
      "learning_rate": 0.0001833428234150406,
      "loss": 3.6006,
      "step": 84000
    },
    {
      "epoch": 1.173529615998889,
      "grad_norm": 1.50645112991333,
      "learning_rate": 0.00018264842719255606,
      "loss": 3.6263,
      "step": 84500
    },
    {
      "epoch": 1.1804735782237346,
      "grad_norm": 1.8035708665847778,
      "learning_rate": 0.0001819540309700715,
      "loss": 3.5539,
      "step": 85000
    },
    {
      "epoch": 1.18741754044858,
      "grad_norm": 1.9327325820922852,
      "learning_rate": 0.00018125963474758695,
      "loss": 3.5822,
      "step": 85500
    },
    {
      "epoch": 1.1943615026734253,
      "grad_norm": 1.14821195602417,
      "learning_rate": 0.0001805652385251024,
      "loss": 3.6075,
      "step": 86000
    },
    {
      "epoch": 1.201305464898271,
      "grad_norm": 1.6110094785690308,
      "learning_rate": 0.00017987084230261784,
      "loss": 3.5879,
      "step": 86500
    },
    {
      "epoch": 1.2082494271231163,
      "grad_norm": 1.9771841764450073,
      "learning_rate": 0.0001791764460801333,
      "loss": 3.5976,
      "step": 87000
    },
    {
      "epoch": 1.215193389347962,
      "grad_norm": 1.6200594902038574,
      "learning_rate": 0.00017848204985764876,
      "loss": 3.5788,
      "step": 87500
    },
    {
      "epoch": 1.2221373515728073,
      "grad_norm": 1.9356050491333008,
      "learning_rate": 0.0001777876536351642,
      "loss": 3.5611,
      "step": 88000
    },
    {
      "epoch": 1.229081313797653,
      "grad_norm": 2.2079310417175293,
      "learning_rate": 0.00017709325741267965,
      "loss": 3.5828,
      "step": 88500
    },
    {
      "epoch": 1.2360252760224983,
      "grad_norm": 1.6722863912582397,
      "learning_rate": 0.0001763988611901951,
      "loss": 3.5509,
      "step": 89000
    },
    {
      "epoch": 1.2429692382473438,
      "grad_norm": 1.2027846574783325,
      "learning_rate": 0.00017570446496771054,
      "loss": 3.5512,
      "step": 89500
    },
    {
      "epoch": 1.2499132004721893,
      "grad_norm": 3.1451849937438965,
      "learning_rate": 0.000175010068745226,
      "loss": 3.5704,
      "step": 90000
    },
    {
      "epoch": 1.2568571626970348,
      "grad_norm": 1.64677095413208,
      "learning_rate": 0.00017431567252274146,
      "loss": 3.5879,
      "step": 90500
    },
    {
      "epoch": 1.2638011249218803,
      "grad_norm": 1.2925798892974854,
      "learning_rate": 0.0001736212763002569,
      "loss": 3.5499,
      "step": 91000
    },
    {
      "epoch": 1.2707450871467258,
      "grad_norm": 1.2768690586090088,
      "learning_rate": 0.00017292688007777235,
      "loss": 3.5538,
      "step": 91500
    },
    {
      "epoch": 1.2776890493715714,
      "grad_norm": 2.6654281616210938,
      "learning_rate": 0.0001722324838552878,
      "loss": 3.5431,
      "step": 92000
    },
    {
      "epoch": 1.2846330115964169,
      "grad_norm": 1.8698071241378784,
      "learning_rate": 0.00017153808763280327,
      "loss": 3.5646,
      "step": 92500
    },
    {
      "epoch": 1.2915769738212624,
      "grad_norm": 1.6036585569381714,
      "learning_rate": 0.0001708436914103187,
      "loss": 3.5595,
      "step": 93000
    },
    {
      "epoch": 1.2985209360461079,
      "grad_norm": 1.4960416555404663,
      "learning_rate": 0.00017014929518783416,
      "loss": 3.549,
      "step": 93500
    },
    {
      "epoch": 1.3054648982709534,
      "grad_norm": 1.5892603397369385,
      "learning_rate": 0.00016945489896534962,
      "loss": 3.5144,
      "step": 94000
    },
    {
      "epoch": 1.3124088604957989,
      "grad_norm": 1.2791684865951538,
      "learning_rate": 0.00016876050274286506,
      "loss": 3.5494,
      "step": 94500
    },
    {
      "epoch": 1.3193528227206444,
      "grad_norm": 4.176353454589844,
      "learning_rate": 0.00016806610652038052,
      "loss": 3.5705,
      "step": 95000
    },
    {
      "epoch": 1.3262967849454899,
      "grad_norm": 1.6479995250701904,
      "learning_rate": 0.00016737171029789597,
      "loss": 3.5644,
      "step": 95500
    },
    {
      "epoch": 1.3332407471703354,
      "grad_norm": 1.6295863389968872,
      "learning_rate": 0.0001666773140754114,
      "loss": 3.5614,
      "step": 96000
    },
    {
      "epoch": 1.3401847093951809,
      "grad_norm": 2.4127180576324463,
      "learning_rate": 0.00016598291785292687,
      "loss": 3.5488,
      "step": 96500
    },
    {
      "epoch": 1.3471286716200264,
      "grad_norm": 2.4236507415771484,
      "learning_rate": 0.00016528852163044233,
      "loss": 3.5188,
      "step": 97000
    },
    {
      "epoch": 1.3540726338448719,
      "grad_norm": 1.2415298223495483,
      "learning_rate": 0.00016459412540795776,
      "loss": 3.5843,
      "step": 97500
    },
    {
      "epoch": 1.3610165960697174,
      "grad_norm": 2.39335298538208,
      "learning_rate": 0.00016389972918547322,
      "loss": 3.5445,
      "step": 98000
    },
    {
      "epoch": 1.3679605582945629,
      "grad_norm": 1.481112003326416,
      "learning_rate": 0.00016320533296298868,
      "loss": 3.5725,
      "step": 98500
    },
    {
      "epoch": 1.3749045205194084,
      "grad_norm": 1.8762099742889404,
      "learning_rate": 0.0001625109367405041,
      "loss": 3.5422,
      "step": 99000
    },
    {
      "epoch": 1.3818484827442539,
      "grad_norm": 1.4844539165496826,
      "learning_rate": 0.00016181654051801957,
      "loss": 3.5469,
      "step": 99500
    },
    {
      "epoch": 1.3887924449690994,
      "grad_norm": 1.776289701461792,
      "learning_rate": 0.00016112214429553503,
      "loss": 3.5329,
      "step": 100000
    },
    {
      "epoch": 1.3957364071939449,
      "grad_norm": 1.566076636314392,
      "learning_rate": 0.00016042774807305046,
      "loss": 3.5146,
      "step": 100500
    },
    {
      "epoch": 1.4026803694187904,
      "grad_norm": 1.8773123025894165,
      "learning_rate": 0.00015973335185056592,
      "loss": 3.5298,
      "step": 101000
    },
    {
      "epoch": 1.4096243316436359,
      "grad_norm": 1.92935049533844,
      "learning_rate": 0.00015903895562808138,
      "loss": 3.5298,
      "step": 101500
    },
    {
      "epoch": 1.4165682938684814,
      "grad_norm": 1.807790994644165,
      "learning_rate": 0.0001583445594055968,
      "loss": 3.5586,
      "step": 102000
    },
    {
      "epoch": 1.4235122560933269,
      "grad_norm": 1.67229425907135,
      "learning_rate": 0.00015765016318311227,
      "loss": 3.5202,
      "step": 102500
    },
    {
      "epoch": 1.4304562183181724,
      "grad_norm": 1.2355769872665405,
      "learning_rate": 0.00015695576696062773,
      "loss": 3.5161,
      "step": 103000
    },
    {
      "epoch": 1.4374001805430179,
      "grad_norm": 1.7655647993087769,
      "learning_rate": 0.00015626137073814316,
      "loss": 3.5094,
      "step": 103500
    },
    {
      "epoch": 1.4443441427678634,
      "grad_norm": 1.4021390676498413,
      "learning_rate": 0.00015556697451565862,
      "loss": 3.5533,
      "step": 104000
    },
    {
      "epoch": 1.4512881049927089,
      "grad_norm": 1.7360609769821167,
      "learning_rate": 0.00015487257829317408,
      "loss": 3.4831,
      "step": 104500
    },
    {
      "epoch": 1.4582320672175544,
      "grad_norm": 1.5841504335403442,
      "learning_rate": 0.00015417818207068954,
      "loss": 3.5008,
      "step": 105000
    },
    {
      "epoch": 1.4651760294423999,
      "grad_norm": 1.7698231935501099,
      "learning_rate": 0.00015348378584820497,
      "loss": 3.4889,
      "step": 105500
    },
    {
      "epoch": 1.4721199916672454,
      "grad_norm": 1.3631160259246826,
      "learning_rate": 0.00015278938962572043,
      "loss": 3.5313,
      "step": 106000
    },
    {
      "epoch": 1.4790639538920909,
      "grad_norm": 1.3617082834243774,
      "learning_rate": 0.0001520949934032359,
      "loss": 3.5399,
      "step": 106500
    },
    {
      "epoch": 1.4860079161169364,
      "grad_norm": 1.367946743965149,
      "learning_rate": 0.00015140059718075132,
      "loss": 3.545,
      "step": 107000
    },
    {
      "epoch": 1.492951878341782,
      "grad_norm": 1.3500925302505493,
      "learning_rate": 0.00015070620095826678,
      "loss": 3.5544,
      "step": 107500
    },
    {
      "epoch": 1.4998958405666274,
      "grad_norm": 3.89847731590271,
      "learning_rate": 0.00015001180473578224,
      "loss": 3.4885,
      "step": 108000
    },
    {
      "epoch": 1.5068398027914727,
      "grad_norm": 2.2299306392669678,
      "learning_rate": 0.00014931740851329767,
      "loss": 3.4893,
      "step": 108500
    },
    {
      "epoch": 1.5137837650163184,
      "grad_norm": 2.350405693054199,
      "learning_rate": 0.00014862301229081313,
      "loss": 3.4992,
      "step": 109000
    },
    {
      "epoch": 1.5207277272411637,
      "grad_norm": 1.3148006200790405,
      "learning_rate": 0.00014792861606832856,
      "loss": 3.5146,
      "step": 109500
    },
    {
      "epoch": 1.5276716894660094,
      "grad_norm": 1.533084750175476,
      "learning_rate": 0.00014723421984584402,
      "loss": 3.5149,
      "step": 110000
    },
    {
      "epoch": 1.5346156516908547,
      "grad_norm": 1.3361338376998901,
      "learning_rate": 0.00014653982362335948,
      "loss": 3.4956,
      "step": 110500
    },
    {
      "epoch": 1.5415596139157004,
      "grad_norm": 1.581416130065918,
      "learning_rate": 0.00014584542740087492,
      "loss": 3.5074,
      "step": 111000
    },
    {
      "epoch": 1.5485035761405457,
      "grad_norm": 1.6259821653366089,
      "learning_rate": 0.00014515103117839037,
      "loss": 3.4961,
      "step": 111500
    },
    {
      "epoch": 1.5554475383653914,
      "grad_norm": 1.981719732284546,
      "learning_rate": 0.00014445663495590583,
      "loss": 3.5192,
      "step": 112000
    },
    {
      "epoch": 1.5623915005902367,
      "grad_norm": 1.2760684490203857,
      "learning_rate": 0.00014376223873342127,
      "loss": 3.5265,
      "step": 112500
    },
    {
      "epoch": 1.5693354628150824,
      "grad_norm": 7.409369468688965,
      "learning_rate": 0.00014306784251093673,
      "loss": 3.5335,
      "step": 113000
    },
    {
      "epoch": 1.5762794250399277,
      "grad_norm": 2.6644575595855713,
      "learning_rate": 0.00014237344628845218,
      "loss": 3.4924,
      "step": 113500
    },
    {
      "epoch": 1.5832233872647734,
      "grad_norm": 1.3111194372177124,
      "learning_rate": 0.00014167905006596764,
      "loss": 3.5339,
      "step": 114000
    },
    {
      "epoch": 1.5901673494896187,
      "grad_norm": 4.329043388366699,
      "learning_rate": 0.00014098465384348308,
      "loss": 3.4945,
      "step": 114500
    },
    {
      "epoch": 1.5971113117144644,
      "grad_norm": 1.5919106006622314,
      "learning_rate": 0.00014029025762099854,
      "loss": 3.5081,
      "step": 115000
    },
    {
      "epoch": 1.6040552739393097,
      "grad_norm": 1.7565652132034302,
      "learning_rate": 0.000139595861398514,
      "loss": 3.5337,
      "step": 115500
    },
    {
      "epoch": 1.6109992361641554,
      "grad_norm": 1.8960776329040527,
      "learning_rate": 0.00013890146517602943,
      "loss": 3.4879,
      "step": 116000
    },
    {
      "epoch": 1.6179431983890007,
      "grad_norm": 1.8651204109191895,
      "learning_rate": 0.0001382070689535449,
      "loss": 3.483,
      "step": 116500
    },
    {
      "epoch": 1.6248871606138464,
      "grad_norm": 2.5513360500335693,
      "learning_rate": 0.00013751267273106035,
      "loss": 3.4763,
      "step": 117000
    },
    {
      "epoch": 1.6318311228386917,
      "grad_norm": 1.5704069137573242,
      "learning_rate": 0.00013681827650857578,
      "loss": 3.4932,
      "step": 117500
    },
    {
      "epoch": 1.6387750850635374,
      "grad_norm": 1.619181513786316,
      "learning_rate": 0.00013612388028609124,
      "loss": 3.5177,
      "step": 118000
    },
    {
      "epoch": 1.6457190472883827,
      "grad_norm": 1.3884799480438232,
      "learning_rate": 0.0001354294840636067,
      "loss": 3.5247,
      "step": 118500
    },
    {
      "epoch": 1.6526630095132282,
      "grad_norm": 1.5763874053955078,
      "learning_rate": 0.00013473508784112213,
      "loss": 3.4879,
      "step": 119000
    },
    {
      "epoch": 1.6596069717380737,
      "grad_norm": 1.2959508895874023,
      "learning_rate": 0.00013404069161863756,
      "loss": 3.4924,
      "step": 119500
    },
    {
      "epoch": 1.6665509339629192,
      "grad_norm": 3.481456756591797,
      "learning_rate": 0.00013334629539615305,
      "loss": 3.4886,
      "step": 120000
    },
    {
      "epoch": 1.6734948961877647,
      "grad_norm": 6.646812438964844,
      "learning_rate": 0.00013265189917366848,
      "loss": 3.5136,
      "step": 120500
    },
    {
      "epoch": 1.6804388584126102,
      "grad_norm": 1.200080394744873,
      "learning_rate": 0.00013195750295118394,
      "loss": 3.5017,
      "step": 121000
    },
    {
      "epoch": 1.6873828206374557,
      "grad_norm": 1.4992713928222656,
      "learning_rate": 0.0001312631067286994,
      "loss": 3.4533,
      "step": 121500
    },
    {
      "epoch": 1.6943267828623012,
      "grad_norm": 2.5522916316986084,
      "learning_rate": 0.00013056871050621483,
      "loss": 3.48,
      "step": 122000
    },
    {
      "epoch": 1.7012707450871467,
      "grad_norm": 1.5243773460388184,
      "learning_rate": 0.0001298743142837303,
      "loss": 3.5058,
      "step": 122500
    },
    {
      "epoch": 1.7082147073119922,
      "grad_norm": 1.4201898574829102,
      "learning_rate": 0.00012917991806124572,
      "loss": 3.4667,
      "step": 123000
    },
    {
      "epoch": 1.7151586695368377,
      "grad_norm": 1.7786469459533691,
      "learning_rate": 0.00012848552183876118,
      "loss": 3.4931,
      "step": 123500
    },
    {
      "epoch": 1.7221026317616832,
      "grad_norm": 3.3978912830352783,
      "learning_rate": 0.00012779112561627664,
      "loss": 3.4855,
      "step": 124000
    },
    {
      "epoch": 1.7290465939865287,
      "grad_norm": 4.56933069229126,
      "learning_rate": 0.00012709672939379207,
      "loss": 3.4691,
      "step": 124500
    },
    {
      "epoch": 1.7359905562113742,
      "grad_norm": 2.483752489089966,
      "learning_rate": 0.00012640233317130753,
      "loss": 3.4761,
      "step": 125000
    },
    {
      "epoch": 1.7429345184362197,
      "grad_norm": 1.5909661054611206,
      "learning_rate": 0.000125707936948823,
      "loss": 3.5052,
      "step": 125500
    },
    {
      "epoch": 1.7498784806610652,
      "grad_norm": 1.670730471611023,
      "learning_rate": 0.00012501354072633842,
      "loss": 3.4758,
      "step": 126000
    },
    {
      "epoch": 1.7568224428859107,
      "grad_norm": 1.2424238920211792,
      "learning_rate": 0.00012431914450385388,
      "loss": 3.4405,
      "step": 126500
    },
    {
      "epoch": 1.7637664051107562,
      "grad_norm": 1.6950280666351318,
      "learning_rate": 0.00012362474828136934,
      "loss": 3.5296,
      "step": 127000
    },
    {
      "epoch": 1.7707103673356017,
      "grad_norm": 2.1729133129119873,
      "learning_rate": 0.00012293035205888477,
      "loss": 3.4733,
      "step": 127500
    },
    {
      "epoch": 1.7776543295604472,
      "grad_norm": 1.6061240434646606,
      "learning_rate": 0.00012223595583640023,
      "loss": 3.4562,
      "step": 128000
    },
    {
      "epoch": 1.7845982917852927,
      "grad_norm": 2.095271587371826,
      "learning_rate": 0.0001215415596139157,
      "loss": 3.4618,
      "step": 128500
    },
    {
      "epoch": 1.7915422540101382,
      "grad_norm": 2.206932306289673,
      "learning_rate": 0.00012084716339143114,
      "loss": 3.4577,
      "step": 129000
    },
    {
      "epoch": 1.7984862162349837,
      "grad_norm": 1.7425895929336548,
      "learning_rate": 0.0001201527671689466,
      "loss": 3.4979,
      "step": 129500
    },
    {
      "epoch": 1.8054301784598292,
      "grad_norm": 2.1199731826782227,
      "learning_rate": 0.00011945837094646204,
      "loss": 3.4951,
      "step": 130000
    },
    {
      "epoch": 1.8123741406846747,
      "grad_norm": 1.4702428579330444,
      "learning_rate": 0.00011876397472397749,
      "loss": 3.4778,
      "step": 130500
    },
    {
      "epoch": 1.81931810290952,
      "grad_norm": 1.5938681364059448,
      "learning_rate": 0.00011806957850149295,
      "loss": 3.4782,
      "step": 131000
    },
    {
      "epoch": 1.8262620651343657,
      "grad_norm": 1.5015869140625,
      "learning_rate": 0.0001173751822790084,
      "loss": 3.4367,
      "step": 131500
    },
    {
      "epoch": 1.833206027359211,
      "grad_norm": 1.8470075130462646,
      "learning_rate": 0.00011668078605652385,
      "loss": 3.4463,
      "step": 132000
    },
    {
      "epoch": 1.8401499895840567,
      "grad_norm": 2.0054242610931396,
      "learning_rate": 0.0001159863898340393,
      "loss": 3.4445,
      "step": 132500
    },
    {
      "epoch": 1.847093951808902,
      "grad_norm": 1.4376716613769531,
      "learning_rate": 0.00011529199361155473,
      "loss": 3.4449,
      "step": 133000
    },
    {
      "epoch": 1.8540379140337477,
      "grad_norm": 2.702432870864868,
      "learning_rate": 0.0001145975973890702,
      "loss": 3.4499,
      "step": 133500
    },
    {
      "epoch": 1.860981876258593,
      "grad_norm": 1.6058772802352905,
      "learning_rate": 0.00011390320116658565,
      "loss": 3.4567,
      "step": 134000
    },
    {
      "epoch": 1.8679258384834387,
      "grad_norm": 3.2056682109832764,
      "learning_rate": 0.00011320880494410108,
      "loss": 3.4845,
      "step": 134500
    },
    {
      "epoch": 1.874869800708284,
      "grad_norm": 1.6341466903686523,
      "learning_rate": 0.00011251440872161656,
      "loss": 3.4699,
      "step": 135000
    },
    {
      "epoch": 1.8818137629331297,
      "grad_norm": 1.4821678400039673,
      "learning_rate": 0.00011182001249913199,
      "loss": 3.453,
      "step": 135500
    },
    {
      "epoch": 1.888757725157975,
      "grad_norm": 2.204435110092163,
      "learning_rate": 0.00011112561627664743,
      "loss": 3.3845,
      "step": 136000
    },
    {
      "epoch": 1.8957016873828207,
      "grad_norm": 2.4527945518493652,
      "learning_rate": 0.0001104312200541629,
      "loss": 3.4595,
      "step": 136500
    },
    {
      "epoch": 1.902645649607666,
      "grad_norm": 1.7573269605636597,
      "learning_rate": 0.00010973682383167834,
      "loss": 3.4554,
      "step": 137000
    },
    {
      "epoch": 1.9095896118325117,
      "grad_norm": 2.0512332916259766,
      "learning_rate": 0.00010904242760919379,
      "loss": 3.4537,
      "step": 137500
    },
    {
      "epoch": 1.916533574057357,
      "grad_norm": 2.056835174560547,
      "learning_rate": 0.00010834803138670924,
      "loss": 3.441,
      "step": 138000
    },
    {
      "epoch": 1.9234775362822027,
      "grad_norm": 4.340169906616211,
      "learning_rate": 0.00010765363516422469,
      "loss": 3.4806,
      "step": 138500
    },
    {
      "epoch": 1.930421498507048,
      "grad_norm": 2.0016748905181885,
      "learning_rate": 0.00010695923894174015,
      "loss": 3.4363,
      "step": 139000
    },
    {
      "epoch": 1.9373654607318938,
      "grad_norm": 1.8290444612503052,
      "learning_rate": 0.0001062648427192556,
      "loss": 3.4428,
      "step": 139500
    },
    {
      "epoch": 1.944309422956739,
      "grad_norm": 1.6090489625930786,
      "learning_rate": 0.00010557044649677104,
      "loss": 3.4514,
      "step": 140000
    },
    {
      "epoch": 1.9512533851815848,
      "grad_norm": 1.5943214893341064,
      "learning_rate": 0.0001048760502742865,
      "loss": 3.425,
      "step": 140500
    },
    {
      "epoch": 1.95819734740643,
      "grad_norm": 1.9066240787506104,
      "learning_rate": 0.00010418165405180195,
      "loss": 3.4637,
      "step": 141000
    },
    {
      "epoch": 1.9651413096312758,
      "grad_norm": 1.718125820159912,
      "learning_rate": 0.00010348725782931739,
      "loss": 3.4784,
      "step": 141500
    },
    {
      "epoch": 1.972085271856121,
      "grad_norm": 1.8587770462036133,
      "learning_rate": 0.00010279286160683285,
      "loss": 3.4058,
      "step": 142000
    },
    {
      "epoch": 1.9790292340809668,
      "grad_norm": 2.709913492202759,
      "learning_rate": 0.0001020984653843483,
      "loss": 3.4442,
      "step": 142500
    },
    {
      "epoch": 1.985973196305812,
      "grad_norm": 1.3006025552749634,
      "learning_rate": 0.00010140406916186374,
      "loss": 3.394,
      "step": 143000
    },
    {
      "epoch": 1.9929171585306575,
      "grad_norm": 1.3316704034805298,
      "learning_rate": 0.0001007096729393792,
      "loss": 3.4307,
      "step": 143500
    },
    {
      "epoch": 1.999861120755503,
      "grad_norm": 2.0763180255889893,
      "learning_rate": 0.00010001527671689465,
      "loss": 3.4408,
      "step": 144000
    },
    {
      "epoch": 2.0,
      "eval_loss": 3.239363193511963,
      "eval_rouge1": 0.055521599130116214,
      "eval_rouge2": 0.013459250920580113,
      "eval_rougeL": 0.054963256550744514,
      "eval_rougeLsum": 0.05513512231622069,
      "eval_runtime": 3954.943,
      "eval_samples_per_second": 4.046,
      "eval_steps_per_second": 2.023,
      "step": 144010
    },
    {
      "epoch": 2.0068050829803488,
      "grad_norm": 2.819636106491089,
      "learning_rate": 9.932088049441011e-05,
      "loss": 3.3738,
      "step": 144500
    },
    {
      "epoch": 2.013749045205194,
      "grad_norm": 1.5194923877716064,
      "learning_rate": 9.862648427192555e-05,
      "loss": 3.3729,
      "step": 145000
    },
    {
      "epoch": 2.0206930074300398,
      "grad_norm": 1.9506241083145142,
      "learning_rate": 9.7932088049441e-05,
      "loss": 3.3605,
      "step": 145500
    },
    {
      "epoch": 2.027636969654885,
      "grad_norm": 2.37030029296875,
      "learning_rate": 9.723769182695646e-05,
      "loss": 3.354,
      "step": 146000
    },
    {
      "epoch": 2.0345809318797308,
      "grad_norm": 1.798161506652832,
      "learning_rate": 9.65432956044719e-05,
      "loss": 3.3393,
      "step": 146500
    },
    {
      "epoch": 2.041524894104576,
      "grad_norm": 1.7773220539093018,
      "learning_rate": 9.584889938198735e-05,
      "loss": 3.3564,
      "step": 147000
    },
    {
      "epoch": 2.0484688563294218,
      "grad_norm": 1.383130669593811,
      "learning_rate": 9.515450315950281e-05,
      "loss": 3.3493,
      "step": 147500
    },
    {
      "epoch": 2.055412818554267,
      "grad_norm": 1.8702272176742554,
      "learning_rate": 9.446010693701825e-05,
      "loss": 3.3706,
      "step": 148000
    },
    {
      "epoch": 2.0623567807791128,
      "grad_norm": 2.419377326965332,
      "learning_rate": 9.37657107145337e-05,
      "loss": 3.369,
      "step": 148500
    },
    {
      "epoch": 2.069300743003958,
      "grad_norm": 1.842032551765442,
      "learning_rate": 9.307131449204916e-05,
      "loss": 3.3367,
      "step": 149000
    },
    {
      "epoch": 2.0762447052288033,
      "grad_norm": 1.890652060508728,
      "learning_rate": 9.23769182695646e-05,
      "loss": 3.3245,
      "step": 149500
    },
    {
      "epoch": 2.083188667453649,
      "grad_norm": 1.6845180988311768,
      "learning_rate": 9.168252204708005e-05,
      "loss": 3.3037,
      "step": 150000
    },
    {
      "epoch": 2.0901326296784943,
      "grad_norm": 1.8656178712844849,
      "learning_rate": 9.098812582459551e-05,
      "loss": 3.3236,
      "step": 150500
    },
    {
      "epoch": 2.09707659190334,
      "grad_norm": 1.933976173400879,
      "learning_rate": 9.029372960211096e-05,
      "loss": 3.3425,
      "step": 151000
    },
    {
      "epoch": 2.1040205541281853,
      "grad_norm": 1.2388135194778442,
      "learning_rate": 8.959933337962642e-05,
      "loss": 3.3654,
      "step": 151500
    },
    {
      "epoch": 2.110964516353031,
      "grad_norm": 2.0735113620758057,
      "learning_rate": 8.890493715714186e-05,
      "loss": 3.329,
      "step": 152000
    },
    {
      "epoch": 2.1179084785778763,
      "grad_norm": 1.460747241973877,
      "learning_rate": 8.821054093465731e-05,
      "loss": 3.3325,
      "step": 152500
    },
    {
      "epoch": 2.124852440802722,
      "grad_norm": 1.381603479385376,
      "learning_rate": 8.751614471217277e-05,
      "loss": 3.3549,
      "step": 153000
    },
    {
      "epoch": 2.1317964030275673,
      "grad_norm": 2.0302138328552246,
      "learning_rate": 8.682174848968821e-05,
      "loss": 3.3099,
      "step": 153500
    },
    {
      "epoch": 2.138740365252413,
      "grad_norm": 1.4594683647155762,
      "learning_rate": 8.612735226720366e-05,
      "loss": 3.3213,
      "step": 154000
    },
    {
      "epoch": 2.1456843274772583,
      "grad_norm": 1.5398012399673462,
      "learning_rate": 8.543295604471912e-05,
      "loss": 3.3918,
      "step": 154500
    },
    {
      "epoch": 2.152628289702104,
      "grad_norm": 1.6090102195739746,
      "learning_rate": 8.473855982223456e-05,
      "loss": 3.3241,
      "step": 155000
    },
    {
      "epoch": 2.1595722519269493,
      "grad_norm": 1.5653716325759888,
      "learning_rate": 8.404416359975001e-05,
      "loss": 3.3673,
      "step": 155500
    },
    {
      "epoch": 2.166516214151795,
      "grad_norm": 1.3755892515182495,
      "learning_rate": 8.334976737726547e-05,
      "loss": 3.3085,
      "step": 156000
    },
    {
      "epoch": 2.1734601763766404,
      "grad_norm": 2.4337687492370605,
      "learning_rate": 8.265537115478091e-05,
      "loss": 3.3441,
      "step": 156500
    },
    {
      "epoch": 2.180404138601486,
      "grad_norm": 1.5850881338119507,
      "learning_rate": 8.196097493229637e-05,
      "loss": 3.3193,
      "step": 157000
    },
    {
      "epoch": 2.1873481008263314,
      "grad_norm": 2.200465679168701,
      "learning_rate": 8.126657870981182e-05,
      "loss": 3.3384,
      "step": 157500
    },
    {
      "epoch": 2.194292063051177,
      "grad_norm": 2.115725517272949,
      "learning_rate": 8.057218248732725e-05,
      "loss": 3.3447,
      "step": 158000
    },
    {
      "epoch": 2.2012360252760224,
      "grad_norm": 1.528279423713684,
      "learning_rate": 7.987778626484272e-05,
      "loss": 3.3155,
      "step": 158500
    },
    {
      "epoch": 2.208179987500868,
      "grad_norm": 1.796518087387085,
      "learning_rate": 7.918339004235816e-05,
      "loss": 3.3501,
      "step": 159000
    },
    {
      "epoch": 2.2151239497257134,
      "grad_norm": 2.301734685897827,
      "learning_rate": 7.84889938198736e-05,
      "loss": 3.3132,
      "step": 159500
    },
    {
      "epoch": 2.222067911950559,
      "grad_norm": 1.7091695070266724,
      "learning_rate": 7.779459759738908e-05,
      "loss": 3.3373,
      "step": 160000
    },
    {
      "epoch": 2.2290118741754044,
      "grad_norm": 1.247341275215149,
      "learning_rate": 7.710020137490451e-05,
      "loss": 3.3235,
      "step": 160500
    },
    {
      "epoch": 2.23595583640025,
      "grad_norm": 3.0286383628845215,
      "learning_rate": 7.640580515241995e-05,
      "loss": 3.3162,
      "step": 161000
    },
    {
      "epoch": 2.2428997986250954,
      "grad_norm": 1.669455885887146,
      "learning_rate": 7.571140892993541e-05,
      "loss": 3.2822,
      "step": 161500
    },
    {
      "epoch": 2.249843760849941,
      "grad_norm": 1.5718942880630493,
      "learning_rate": 7.501701270745086e-05,
      "loss": 3.3213,
      "step": 162000
    },
    {
      "epoch": 2.2567877230747864,
      "grad_norm": 1.445328950881958,
      "learning_rate": 7.432261648496632e-05,
      "loss": 3.3117,
      "step": 162500
    },
    {
      "epoch": 2.263731685299632,
      "grad_norm": 2.205052614212036,
      "learning_rate": 7.362822026248176e-05,
      "loss": 3.3613,
      "step": 163000
    },
    {
      "epoch": 2.2706756475244774,
      "grad_norm": 2.702474594116211,
      "learning_rate": 7.293382403999721e-05,
      "loss": 3.3368,
      "step": 163500
    },
    {
      "epoch": 2.277619609749323,
      "grad_norm": 2.0145928859710693,
      "learning_rate": 7.223942781751267e-05,
      "loss": 3.3162,
      "step": 164000
    },
    {
      "epoch": 2.2845635719741684,
      "grad_norm": 1.5684378147125244,
      "learning_rate": 7.154503159502811e-05,
      "loss": 3.3532,
      "step": 164500
    },
    {
      "epoch": 2.291507534199014,
      "grad_norm": 1.6014593839645386,
      "learning_rate": 7.085063537254357e-05,
      "loss": 3.3014,
      "step": 165000
    },
    {
      "epoch": 2.2984514964238594,
      "grad_norm": 1.8588491678237915,
      "learning_rate": 7.015623915005902e-05,
      "loss": 3.353,
      "step": 165500
    },
    {
      "epoch": 2.305395458648705,
      "grad_norm": 1.7128487825393677,
      "learning_rate": 6.946184292757447e-05,
      "loss": 3.31,
      "step": 166000
    },
    {
      "epoch": 2.3123394208735504,
      "grad_norm": 1.5588475465774536,
      "learning_rate": 6.876744670508992e-05,
      "loss": 3.3193,
      "step": 166500
    },
    {
      "epoch": 2.319283383098396,
      "grad_norm": 1.8460384607315063,
      "learning_rate": 6.807305048260537e-05,
      "loss": 3.3158,
      "step": 167000
    },
    {
      "epoch": 2.3262273453232414,
      "grad_norm": 1.634889006614685,
      "learning_rate": 6.737865426012082e-05,
      "loss": 3.3089,
      "step": 167500
    },
    {
      "epoch": 2.333171307548087,
      "grad_norm": 2.4914541244506836,
      "learning_rate": 6.668425803763628e-05,
      "loss": 3.328,
      "step": 168000
    },
    {
      "epoch": 2.3401152697729324,
      "grad_norm": 1.2565484046936035,
      "learning_rate": 6.598986181515172e-05,
      "loss": 3.2968,
      "step": 168500
    },
    {
      "epoch": 2.347059231997778,
      "grad_norm": 2.460926055908203,
      "learning_rate": 6.529546559266717e-05,
      "loss": 3.353,
      "step": 169000
    },
    {
      "epoch": 2.3540031942226234,
      "grad_norm": 2.0668790340423584,
      "learning_rate": 6.460106937018263e-05,
      "loss": 3.3118,
      "step": 169500
    },
    {
      "epoch": 2.360947156447469,
      "grad_norm": 2.0417802333831787,
      "learning_rate": 6.390667314769807e-05,
      "loss": 3.3098,
      "step": 170000
    },
    {
      "epoch": 2.3678911186723144,
      "grad_norm": 4.113938808441162,
      "learning_rate": 6.321227692521352e-05,
      "loss": 3.3298,
      "step": 170500
    },
    {
      "epoch": 2.37483508089716,
      "grad_norm": 1.9370335340499878,
      "learning_rate": 6.251788070272896e-05,
      "loss": 3.3009,
      "step": 171000
    },
    {
      "epoch": 2.3817790431220054,
      "grad_norm": 2.2328431606292725,
      "learning_rate": 6.182348448024442e-05,
      "loss": 3.3067,
      "step": 171500
    },
    {
      "epoch": 2.3887230053468507,
      "grad_norm": 1.4481734037399292,
      "learning_rate": 6.112908825775988e-05,
      "loss": 3.3408,
      "step": 172000
    },
    {
      "epoch": 2.3956669675716964,
      "grad_norm": 1.1176230907440186,
      "learning_rate": 6.043469203527532e-05,
      "loss": 3.3586,
      "step": 172500
    },
    {
      "epoch": 2.402610929796542,
      "grad_norm": 2.566291332244873,
      "learning_rate": 5.9740295812790774e-05,
      "loss": 3.2948,
      "step": 173000
    },
    {
      "epoch": 2.4095548920213874,
      "grad_norm": 1.9052931070327759,
      "learning_rate": 5.9045899590306226e-05,
      "loss": 3.3025,
      "step": 173500
    },
    {
      "epoch": 2.4164988542462327,
      "grad_norm": 1.9683756828308105,
      "learning_rate": 5.835150336782167e-05,
      "loss": 3.3348,
      "step": 174000
    },
    {
      "epoch": 2.4234428164710784,
      "grad_norm": 3.174572706222534,
      "learning_rate": 5.7657107145337125e-05,
      "loss": 3.329,
      "step": 174500
    },
    {
      "epoch": 2.430386778695924,
      "grad_norm": 2.3149008750915527,
      "learning_rate": 5.696271092285258e-05,
      "loss": 3.2889,
      "step": 175000
    },
    {
      "epoch": 2.4373307409207694,
      "grad_norm": 1.6363497972488403,
      "learning_rate": 5.626831470036803e-05,
      "loss": 3.3258,
      "step": 175500
    },
    {
      "epoch": 2.4442747031456147,
      "grad_norm": 1.2083947658538818,
      "learning_rate": 5.5573918477883476e-05,
      "loss": 3.3456,
      "step": 176000
    },
    {
      "epoch": 2.4512186653704604,
      "grad_norm": 2.0313804149627686,
      "learning_rate": 5.487952225539893e-05,
      "loss": 3.3154,
      "step": 176500
    },
    {
      "epoch": 2.458162627595306,
      "grad_norm": 1.4387134313583374,
      "learning_rate": 5.418512603291438e-05,
      "loss": 3.303,
      "step": 177000
    },
    {
      "epoch": 2.4651065898201514,
      "grad_norm": 1.9549680948257446,
      "learning_rate": 5.349072981042983e-05,
      "loss": 3.309,
      "step": 177500
    },
    {
      "epoch": 2.4720505520449967,
      "grad_norm": 1.7797924280166626,
      "learning_rate": 5.279633358794527e-05,
      "loss": 3.3338,
      "step": 178000
    },
    {
      "epoch": 2.4789945142698424,
      "grad_norm": 1.5036529302597046,
      "learning_rate": 5.2101937365460725e-05,
      "loss": 3.3122,
      "step": 178500
    },
    {
      "epoch": 2.4859384764946877,
      "grad_norm": 2.212462902069092,
      "learning_rate": 5.140754114297618e-05,
      "loss": 3.2841,
      "step": 179000
    },
    {
      "epoch": 2.4928824387195334,
      "grad_norm": 3.3562114238739014,
      "learning_rate": 5.071314492049162e-05,
      "loss": 3.3017,
      "step": 179500
    },
    {
      "epoch": 2.4998264009443787,
      "grad_norm": 1.547029733657837,
      "learning_rate": 5.0018748698007076e-05,
      "loss": 3.3182,
      "step": 180000
    },
    {
      "epoch": 2.5067703631692244,
      "grad_norm": 1.6302788257598877,
      "learning_rate": 4.932435247552253e-05,
      "loss": 3.2912,
      "step": 180500
    },
    {
      "epoch": 2.5137143253940697,
      "grad_norm": 2.3086509704589844,
      "learning_rate": 4.862995625303798e-05,
      "loss": 3.2617,
      "step": 181000
    },
    {
      "epoch": 2.5206582876189154,
      "grad_norm": 1.8361918926239014,
      "learning_rate": 4.7935560030553427e-05,
      "loss": 3.3025,
      "step": 181500
    },
    {
      "epoch": 2.5276022498437607,
      "grad_norm": 2.591750144958496,
      "learning_rate": 4.724116380806888e-05,
      "loss": 3.3148,
      "step": 182000
    },
    {
      "epoch": 2.5345462120686064,
      "grad_norm": 1.2800657749176025,
      "learning_rate": 4.654676758558433e-05,
      "loss": 3.3176,
      "step": 182500
    },
    {
      "epoch": 2.5414901742934517,
      "grad_norm": 1.7381142377853394,
      "learning_rate": 4.585237136309978e-05,
      "loss": 3.2937,
      "step": 183000
    },
    {
      "epoch": 2.5484341365182974,
      "grad_norm": 1.6898601055145264,
      "learning_rate": 4.515797514061523e-05,
      "loss": 3.2955,
      "step": 183500
    },
    {
      "epoch": 2.5553780987431427,
      "grad_norm": 1.9518952369689941,
      "learning_rate": 4.446357891813068e-05,
      "loss": 3.3093,
      "step": 184000
    },
    {
      "epoch": 2.5623220609679884,
      "grad_norm": 1.4132803678512573,
      "learning_rate": 4.3769182695646135e-05,
      "loss": 3.3142,
      "step": 184500
    },
    {
      "epoch": 2.5692660231928337,
      "grad_norm": 1.304002046585083,
      "learning_rate": 4.307478647316158e-05,
      "loss": 3.2875,
      "step": 185000
    },
    {
      "epoch": 2.5762099854176794,
      "grad_norm": 3.1700334548950195,
      "learning_rate": 4.2380390250677033e-05,
      "loss": 3.3445,
      "step": 185500
    },
    {
      "epoch": 2.5831539476425247,
      "grad_norm": 1.5128083229064941,
      "learning_rate": 4.1685994028192486e-05,
      "loss": 3.3174,
      "step": 186000
    },
    {
      "epoch": 2.5900979098673704,
      "grad_norm": 2.8518435955047607,
      "learning_rate": 4.099159780570794e-05,
      "loss": 3.2766,
      "step": 186500
    },
    {
      "epoch": 2.5970418720922157,
      "grad_norm": 1.2636958360671997,
      "learning_rate": 4.0297201583223384e-05,
      "loss": 3.3032,
      "step": 187000
    },
    {
      "epoch": 2.6039858343170614,
      "grad_norm": 2.1779420375823975,
      "learning_rate": 3.960280536073884e-05,
      "loss": 3.3202,
      "step": 187500
    },
    {
      "epoch": 2.6109297965419067,
      "grad_norm": 1.8627387285232544,
      "learning_rate": 3.890840913825429e-05,
      "loss": 3.2831,
      "step": 188000
    },
    {
      "epoch": 2.6178737587667524,
      "grad_norm": 2.000037670135498,
      "learning_rate": 3.821401291576973e-05,
      "loss": 3.3431,
      "step": 188500
    },
    {
      "epoch": 2.6248177209915977,
      "grad_norm": 1.3186124563217163,
      "learning_rate": 3.751961669328519e-05,
      "loss": 3.3161,
      "step": 189000
    },
    {
      "epoch": 2.6317616832164434,
      "grad_norm": 1.7129555940628052,
      "learning_rate": 3.682522047080064e-05,
      "loss": 3.2901,
      "step": 189500
    },
    {
      "epoch": 2.6387056454412887,
      "grad_norm": 1.727120041847229,
      "learning_rate": 3.6130824248316086e-05,
      "loss": 3.3028,
      "step": 190000
    },
    {
      "epoch": 2.6456496076661344,
      "grad_norm": 2.223973512649536,
      "learning_rate": 3.543642802583154e-05,
      "loss": 3.3096,
      "step": 190500
    },
    {
      "epoch": 2.6525935698909797,
      "grad_norm": 1.908118486404419,
      "learning_rate": 3.4742031803346984e-05,
      "loss": 3.2971,
      "step": 191000
    },
    {
      "epoch": 2.6595375321158254,
      "grad_norm": 1.4966055154800415,
      "learning_rate": 3.404763558086244e-05,
      "loss": 3.3041,
      "step": 191500
    },
    {
      "epoch": 2.6664814943406707,
      "grad_norm": 1.5852959156036377,
      "learning_rate": 3.335323935837789e-05,
      "loss": 3.3396,
      "step": 192000
    },
    {
      "epoch": 2.6734254565655164,
      "grad_norm": 1.956778883934021,
      "learning_rate": 3.2658843135893335e-05,
      "loss": 3.3153,
      "step": 192500
    },
    {
      "epoch": 2.6803694187903617,
      "grad_norm": 1.6665066480636597,
      "learning_rate": 3.196444691340879e-05,
      "loss": 3.3051,
      "step": 193000
    },
    {
      "epoch": 2.6873133810152074,
      "grad_norm": 1.5020607709884644,
      "learning_rate": 3.127005069092424e-05,
      "loss": 3.2956,
      "step": 193500
    },
    {
      "epoch": 2.6942573432400527,
      "grad_norm": 1.861676573753357,
      "learning_rate": 3.057565446843969e-05,
      "loss": 3.3266,
      "step": 194000
    },
    {
      "epoch": 2.701201305464898,
      "grad_norm": 1.6980831623077393,
      "learning_rate": 2.988125824595514e-05,
      "loss": 3.2896,
      "step": 194500
    },
    {
      "epoch": 2.7081452676897437,
      "grad_norm": 3.289989709854126,
      "learning_rate": 2.9186862023470588e-05,
      "loss": 3.2898,
      "step": 195000
    },
    {
      "epoch": 2.7150892299145895,
      "grad_norm": 1.6162209510803223,
      "learning_rate": 2.849246580098604e-05,
      "loss": 3.2698,
      "step": 195500
    },
    {
      "epoch": 2.7220331921394347,
      "grad_norm": 4.295835018157959,
      "learning_rate": 2.779806957850149e-05,
      "loss": 3.2747,
      "step": 196000
    },
    {
      "epoch": 2.72897715436428,
      "grad_norm": 2.021383762359619,
      "learning_rate": 2.7103673356016942e-05,
      "loss": 3.2866,
      "step": 196500
    },
    {
      "epoch": 2.7359211165891257,
      "grad_norm": 5.153833389282227,
      "learning_rate": 2.640927713353239e-05,
      "loss": 3.3057,
      "step": 197000
    },
    {
      "epoch": 2.7428650788139715,
      "grad_norm": 1.4507516622543335,
      "learning_rate": 2.5714880911047844e-05,
      "loss": 3.2617,
      "step": 197500
    },
    {
      "epoch": 2.7498090410388167,
      "grad_norm": 1.0930436849594116,
      "learning_rate": 2.5020484688563293e-05,
      "loss": 3.3113,
      "step": 198000
    },
    {
      "epoch": 2.756753003263662,
      "grad_norm": 2.5461559295654297,
      "learning_rate": 2.4326088466078745e-05,
      "loss": 3.3073,
      "step": 198500
    },
    {
      "epoch": 2.7636969654885077,
      "grad_norm": 1.3845510482788086,
      "learning_rate": 2.363169224359419e-05,
      "loss": 3.2805,
      "step": 199000
    },
    {
      "epoch": 2.7706409277133535,
      "grad_norm": 1.5489321947097778,
      "learning_rate": 2.293729602110964e-05,
      "loss": 3.2863,
      "step": 199500
    },
    {
      "epoch": 2.7775848899381987,
      "grad_norm": 1.2201488018035889,
      "learning_rate": 2.2242899798625093e-05,
      "loss": 3.2959,
      "step": 200000
    },
    {
      "epoch": 2.784528852163044,
      "grad_norm": 1.8551557064056396,
      "learning_rate": 2.1548503576140542e-05,
      "loss": 3.3299,
      "step": 200500
    },
    {
      "epoch": 2.7914728143878897,
      "grad_norm": 2.28908371925354,
      "learning_rate": 2.0854107353655995e-05,
      "loss": 3.2962,
      "step": 201000
    },
    {
      "epoch": 2.7984167766127355,
      "grad_norm": 2.0773096084594727,
      "learning_rate": 2.0159711131171444e-05,
      "loss": 3.2798,
      "step": 201500
    },
    {
      "epoch": 2.8053607388375807,
      "grad_norm": 2.611323833465576,
      "learning_rate": 1.9465314908686896e-05,
      "loss": 3.3228,
      "step": 202000
    },
    {
      "epoch": 2.812304701062426,
      "grad_norm": 2.0584192276000977,
      "learning_rate": 1.8770918686202346e-05,
      "loss": 3.3112,
      "step": 202500
    },
    {
      "epoch": 2.8192486632872718,
      "grad_norm": 3.291172981262207,
      "learning_rate": 1.8076522463717795e-05,
      "loss": 3.3141,
      "step": 203000
    },
    {
      "epoch": 2.8261926255121175,
      "grad_norm": 2.5255441665649414,
      "learning_rate": 1.7382126241233247e-05,
      "loss": 3.2951,
      "step": 203500
    },
    {
      "epoch": 2.8331365877369628,
      "grad_norm": 1.3763819932937622,
      "learning_rate": 1.6687730018748696e-05,
      "loss": 3.294,
      "step": 204000
    },
    {
      "epoch": 2.840080549961808,
      "grad_norm": 1.9026843309402466,
      "learning_rate": 1.599333379626415e-05,
      "loss": 3.3003,
      "step": 204500
    },
    {
      "epoch": 2.8470245121866538,
      "grad_norm": 1.6121410131454468,
      "learning_rate": 1.5298937573779598e-05,
      "loss": 3.3096,
      "step": 205000
    },
    {
      "epoch": 2.8539684744114995,
      "grad_norm": 2.3993430137634277,
      "learning_rate": 1.4604541351295047e-05,
      "loss": 3.3075,
      "step": 205500
    },
    {
      "epoch": 2.8609124366363448,
      "grad_norm": 1.6766456365585327,
      "learning_rate": 1.3910145128810498e-05,
      "loss": 3.2873,
      "step": 206000
    },
    {
      "epoch": 2.86785639886119,
      "grad_norm": 2.1749913692474365,
      "learning_rate": 1.3215748906325947e-05,
      "loss": 3.2432,
      "step": 206500
    },
    {
      "epoch": 2.8748003610860358,
      "grad_norm": 1.5734447240829468,
      "learning_rate": 1.2521352683841398e-05,
      "loss": 3.2501,
      "step": 207000
    },
    {
      "epoch": 2.881744323310881,
      "grad_norm": 1.3672767877578735,
      "learning_rate": 1.1826956461356849e-05,
      "loss": 3.292,
      "step": 207500
    },
    {
      "epoch": 2.8886882855357268,
      "grad_norm": 3.4438602924346924,
      "learning_rate": 1.11325602388723e-05,
      "loss": 3.2755,
      "step": 208000
    },
    {
      "epoch": 2.895632247760572,
      "grad_norm": 3.021101474761963,
      "learning_rate": 1.043816401638775e-05,
      "loss": 3.2599,
      "step": 208500
    },
    {
      "epoch": 2.9025762099854178,
      "grad_norm": 1.2644829750061035,
      "learning_rate": 9.743767793903202e-06,
      "loss": 3.291,
      "step": 209000
    },
    {
      "epoch": 2.909520172210263,
      "grad_norm": 1.7406469583511353,
      "learning_rate": 9.04937157141865e-06,
      "loss": 3.2871,
      "step": 209500
    },
    {
      "epoch": 2.9164641344351088,
      "grad_norm": 1.8715460300445557,
      "learning_rate": 8.354975348934102e-06,
      "loss": 3.3359,
      "step": 210000
    },
    {
      "epoch": 2.923408096659954,
      "grad_norm": 1.7464805841445923,
      "learning_rate": 7.660579126449552e-06,
      "loss": 3.263,
      "step": 210500
    },
    {
      "epoch": 2.9303520588847998,
      "grad_norm": 1.6525601148605347,
      "learning_rate": 6.966182903965002e-06,
      "loss": 3.2654,
      "step": 211000
    },
    {
      "epoch": 2.937296021109645,
      "grad_norm": 2.9503705501556396,
      "learning_rate": 6.2717866814804524e-06,
      "loss": 3.2797,
      "step": 211500
    },
    {
      "epoch": 2.9442399833344908,
      "grad_norm": 1.702645182609558,
      "learning_rate": 5.5773904589959024e-06,
      "loss": 3.291,
      "step": 212000
    },
    {
      "epoch": 2.951183945559336,
      "grad_norm": 1.7340339422225952,
      "learning_rate": 4.882994236511353e-06,
      "loss": 3.3051,
      "step": 212500
    },
    {
      "epoch": 2.9581279077841818,
      "grad_norm": 1.9832649230957031,
      "learning_rate": 4.188598014026803e-06,
      "loss": 3.2941,
      "step": 213000
    },
    {
      "epoch": 2.965071870009027,
      "grad_norm": 2.080734968185425,
      "learning_rate": 3.4942017915422537e-06,
      "loss": 3.3281,
      "step": 213500
    },
    {
      "epoch": 2.972015832233873,
      "grad_norm": 1.7874020338058472,
      "learning_rate": 2.799805569057704e-06,
      "loss": 3.2567,
      "step": 214000
    },
    {
      "epoch": 2.978959794458718,
      "grad_norm": 1.32713782787323,
      "learning_rate": 2.1054093465731546e-06,
      "loss": 3.3156,
      "step": 214500
    },
    {
      "epoch": 2.985903756683564,
      "grad_norm": 2.1903092861175537,
      "learning_rate": 1.4110131240886048e-06,
      "loss": 3.2949,
      "step": 215000
    },
    {
      "epoch": 2.992847718908409,
      "grad_norm": 1.8510949611663818,
      "learning_rate": 7.166169016040551e-07,
      "loss": 3.3084,
      "step": 215500
    },
    {
      "epoch": 2.999791681133255,
      "grad_norm": 1.916053056716919,
      "learning_rate": 2.222067911950559e-08,
      "loss": 3.2611,
      "step": 216000
    }
  ],
  "logging_steps": 500,
  "max_steps": 216015,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.3301724526811136e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}