{
  "best_metric": 3.161168336868286,
  "best_model_checkpoint": "./cifar100_outputs/checkpoint-26565",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 26565,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018821757952192735,
      "grad_norm": 1.8490731716156006,
      "learning_rate": 0.0004998117824204781,
      "loss": 4.6141,
      "step": 10
    },
    {
      "epoch": 0.003764351590438547,
      "grad_norm": 1.622644066810608,
      "learning_rate": 0.0004996235648409562,
      "loss": 4.6699,
      "step": 20
    },
    {
      "epoch": 0.00564652738565782,
      "grad_norm": 1.3981355428695679,
      "learning_rate": 0.0004994353472614342,
      "loss": 4.6266,
      "step": 30
    },
    {
      "epoch": 0.007528703180877094,
      "grad_norm": 1.4281316995620728,
      "learning_rate": 0.0004992471296819123,
      "loss": 4.5814,
      "step": 40
    },
    {
      "epoch": 0.009410878976096368,
      "grad_norm": 1.5957081317901611,
      "learning_rate": 0.0004990589121023904,
      "loss": 4.6204,
      "step": 50
    },
    {
      "epoch": 0.01129305477131564,
      "grad_norm": 1.291913390159607,
      "learning_rate": 0.0004988706945228685,
      "loss": 4.643,
      "step": 60
    },
    {
      "epoch": 0.013175230566534914,
      "grad_norm": 0.8935217261314392,
      "learning_rate": 0.0004986824769433465,
      "loss": 4.6356,
      "step": 70
    },
    {
      "epoch": 0.015057406361754188,
      "grad_norm": 0.9641174674034119,
      "learning_rate": 0.0004984942593638246,
      "loss": 4.6652,
      "step": 80
    },
    {
      "epoch": 0.01693958215697346,
      "grad_norm": 0.9098806381225586,
      "learning_rate": 0.0004983060417843027,
      "loss": 4.61,
      "step": 90
    },
    {
      "epoch": 0.018821757952192736,
      "grad_norm": 0.8219736814498901,
      "learning_rate": 0.0004981178242047808,
      "loss": 4.5734,
      "step": 100
    },
    {
      "epoch": 0.020703933747412008,
      "grad_norm": 1.2911275625228882,
      "learning_rate": 0.0004979296066252589,
      "loss": 4.6272,
      "step": 110
    },
    {
      "epoch": 0.02258610954263128,
      "grad_norm": 0.796344518661499,
      "learning_rate": 0.0004977413890457369,
      "loss": 4.637,
      "step": 120
    },
    {
      "epoch": 0.024468285337850556,
      "grad_norm": 1.0010029077529907,
      "learning_rate": 0.000497553171466215,
      "loss": 4.6,
      "step": 130
    },
    {
      "epoch": 0.026350461133069828,
      "grad_norm": 1.2629014253616333,
      "learning_rate": 0.000497364953886693,
      "loss": 4.5761,
      "step": 140
    },
    {
      "epoch": 0.028232636928289104,
      "grad_norm": 1.2411388158798218,
      "learning_rate": 0.000497176736307171,
      "loss": 4.6005,
      "step": 150
    },
    {
      "epoch": 0.030114812723508376,
      "grad_norm": 0.893264651298523,
      "learning_rate": 0.0004969885187276491,
      "loss": 4.6009,
      "step": 160
    },
    {
      "epoch": 0.03199698851872765,
      "grad_norm": 1.3933730125427246,
      "learning_rate": 0.0004968003011481272,
      "loss": 4.5853,
      "step": 170
    },
    {
      "epoch": 0.03387916431394692,
      "grad_norm": 1.3685733079910278,
      "learning_rate": 0.0004966120835686053,
      "loss": 4.6171,
      "step": 180
    },
    {
      "epoch": 0.0357613401091662,
      "grad_norm": 0.8030378818511963,
      "learning_rate": 0.0004964238659890834,
      "loss": 4.5877,
      "step": 190
    },
    {
      "epoch": 0.03764351590438547,
      "grad_norm": 1.0896660089492798,
      "learning_rate": 0.0004962356484095614,
      "loss": 4.54,
      "step": 200
    },
    {
      "epoch": 0.039525691699604744,
      "grad_norm": 1.415959119796753,
      "learning_rate": 0.0004960474308300395,
      "loss": 4.5448,
      "step": 210
    },
    {
      "epoch": 0.041407867494824016,
      "grad_norm": 1.3994077444076538,
      "learning_rate": 0.0004958592132505176,
      "loss": 4.5989,
      "step": 220
    },
    {
      "epoch": 0.04329004329004329,
      "grad_norm": 1.2095682621002197,
      "learning_rate": 0.0004956709956709957,
      "loss": 4.5381,
      "step": 230
    },
    {
      "epoch": 0.04517221908526256,
      "grad_norm": 1.5952842235565186,
      "learning_rate": 0.0004954827780914737,
      "loss": 4.5794,
      "step": 240
    },
    {
      "epoch": 0.04705439488048184,
      "grad_norm": 1.3413859605789185,
      "learning_rate": 0.0004952945605119518,
      "loss": 4.5359,
      "step": 250
    },
    {
      "epoch": 0.04893657067570111,
      "grad_norm": 1.0710666179656982,
      "learning_rate": 0.0004951063429324299,
      "loss": 4.6534,
      "step": 260
    },
    {
      "epoch": 0.050818746470920384,
      "grad_norm": 1.269322395324707,
      "learning_rate": 0.000494918125352908,
      "loss": 4.6267,
      "step": 270
    },
    {
      "epoch": 0.052700922266139656,
      "grad_norm": 1.1679551601409912,
      "learning_rate": 0.000494729907773386,
      "loss": 4.4577,
      "step": 280
    },
    {
      "epoch": 0.05458309806135893,
      "grad_norm": 1.586521029472351,
      "learning_rate": 0.0004945416901938641,
      "loss": 4.5726,
      "step": 290
    },
    {
      "epoch": 0.05646527385657821,
      "grad_norm": 1.1866834163665771,
      "learning_rate": 0.0004943534726143422,
      "loss": 4.5967,
      "step": 300
    },
    {
      "epoch": 0.05834744965179748,
      "grad_norm": 1.2843154668807983,
      "learning_rate": 0.0004941652550348203,
      "loss": 4.557,
      "step": 310
    },
    {
      "epoch": 0.06022962544701675,
      "grad_norm": 0.9741870760917664,
      "learning_rate": 0.0004939770374552983,
      "loss": 4.5775,
      "step": 320
    },
    {
      "epoch": 0.062111801242236024,
      "grad_norm": 1.1748979091644287,
      "learning_rate": 0.0004937888198757764,
      "loss": 4.4401,
      "step": 330
    },
    {
      "epoch": 0.0639939770374553,
      "grad_norm": 1.1653861999511719,
      "learning_rate": 0.0004936006022962545,
      "loss": 4.5607,
      "step": 340
    },
    {
      "epoch": 0.06587615283267458,
      "grad_norm": 1.1719738245010376,
      "learning_rate": 0.0004934123847167326,
      "loss": 4.5217,
      "step": 350
    },
    {
      "epoch": 0.06775832862789384,
      "grad_norm": 0.9163254499435425,
      "learning_rate": 0.0004932241671372107,
      "loss": 4.6028,
      "step": 360
    },
    {
      "epoch": 0.06964050442311312,
      "grad_norm": 0.9993066191673279,
      "learning_rate": 0.0004930359495576887,
      "loss": 4.4492,
      "step": 370
    },
    {
      "epoch": 0.0715226802183324,
      "grad_norm": 1.3269693851470947,
      "learning_rate": 0.0004928477319781668,
      "loss": 4.4583,
      "step": 380
    },
    {
      "epoch": 0.07340485601355166,
      "grad_norm": 1.4873899221420288,
      "learning_rate": 0.0004926595143986449,
      "loss": 4.5468,
      "step": 390
    },
    {
      "epoch": 0.07528703180877094,
      "grad_norm": 1.2793279886245728,
      "learning_rate": 0.0004924712968191229,
      "loss": 4.5068,
      "step": 400
    },
    {
      "epoch": 0.07716920760399021,
      "grad_norm": 1.2136483192443848,
      "learning_rate": 0.0004922830792396009,
      "loss": 4.4443,
      "step": 410
    },
    {
      "epoch": 0.07905138339920949,
      "grad_norm": 1.7519340515136719,
      "learning_rate": 0.000492094861660079,
      "loss": 4.422,
      "step": 420
    },
    {
      "epoch": 0.08093355919442875,
      "grad_norm": 1.1981459856033325,
      "learning_rate": 0.0004919066440805571,
      "loss": 4.4402,
      "step": 430
    },
    {
      "epoch": 0.08281573498964803,
      "grad_norm": 1.1304644346237183,
      "learning_rate": 0.0004917184265010352,
      "loss": 4.4206,
      "step": 440
    },
    {
      "epoch": 0.08469791078486731,
      "grad_norm": 1.0927811861038208,
      "learning_rate": 0.0004915302089215133,
      "loss": 4.4864,
      "step": 450
    },
    {
      "epoch": 0.08658008658008658,
      "grad_norm": 1.6989763975143433,
      "learning_rate": 0.0004913419913419914,
      "loss": 4.4381,
      "step": 460
    },
    {
      "epoch": 0.08846226237530586,
      "grad_norm": 0.9959608316421509,
      "learning_rate": 0.0004911537737624695,
      "loss": 4.5199,
      "step": 470
    },
    {
      "epoch": 0.09034443817052512,
      "grad_norm": 1.1445202827453613,
      "learning_rate": 0.0004909655561829475,
      "loss": 4.4059,
      "step": 480
    },
    {
      "epoch": 0.0922266139657444,
      "grad_norm": 2.0817947387695312,
      "learning_rate": 0.0004907773386034255,
      "loss": 4.4654,
      "step": 490
    },
    {
      "epoch": 0.09410878976096368,
      "grad_norm": 1.2125662565231323,
      "learning_rate": 0.0004905891210239036,
      "loss": 4.4801,
      "step": 500
    },
    {
      "epoch": 0.09599096555618294,
      "grad_norm": 1.3871793746948242,
      "learning_rate": 0.0004904009034443817,
      "loss": 4.5371,
      "step": 510
    },
    {
      "epoch": 0.09787314135140222,
      "grad_norm": 1.5888006687164307,
      "learning_rate": 0.0004902126858648598,
      "loss": 4.493,
      "step": 520
    },
    {
      "epoch": 0.09975531714662149,
      "grad_norm": 1.6036688089370728,
      "learning_rate": 0.0004900244682853378,
      "loss": 4.5377,
      "step": 530
    },
    {
      "epoch": 0.10163749294184077,
      "grad_norm": 1.4229373931884766,
      "learning_rate": 0.0004898362507058159,
      "loss": 4.5266,
      "step": 540
    },
    {
      "epoch": 0.10351966873706005,
      "grad_norm": 4.853313446044922,
      "learning_rate": 0.000489648033126294,
      "loss": 4.4592,
      "step": 550
    },
    {
      "epoch": 0.10540184453227931,
      "grad_norm": 0.9515388607978821,
      "learning_rate": 0.0004894598155467721,
      "loss": 4.4654,
      "step": 560
    },
    {
      "epoch": 0.10728402032749859,
      "grad_norm": 1.1957790851593018,
      "learning_rate": 0.0004892715979672501,
      "loss": 4.4855,
      "step": 570
    },
    {
      "epoch": 0.10916619612271786,
      "grad_norm": 1.4205329418182373,
      "learning_rate": 0.0004890833803877282,
      "loss": 4.457,
      "step": 580
    },
    {
      "epoch": 0.11104837191793714,
      "grad_norm": 1.360944151878357,
      "learning_rate": 0.0004888951628082063,
      "loss": 4.508,
      "step": 590
    },
    {
      "epoch": 0.11293054771315642,
      "grad_norm": 1.8090227842330933,
      "learning_rate": 0.0004887069452286844,
      "loss": 4.4387,
      "step": 600
    },
    {
      "epoch": 0.11481272350837568,
      "grad_norm": 1.6240856647491455,
      "learning_rate": 0.0004885187276491625,
      "loss": 4.4155,
      "step": 610
    },
    {
      "epoch": 0.11669489930359496,
      "grad_norm": 1.5647474527359009,
      "learning_rate": 0.0004883305100696405,
      "loss": 4.3604,
      "step": 620
    },
    {
      "epoch": 0.11857707509881422,
      "grad_norm": 1.6423243284225464,
      "learning_rate": 0.0004881422924901186,
      "loss": 4.472,
      "step": 630
    },
    {
      "epoch": 0.1204592508940335,
      "grad_norm": 1.5964930057525635,
      "learning_rate": 0.0004879540749105967,
      "loss": 4.409,
      "step": 640
    },
    {
      "epoch": 0.12234142668925278,
      "grad_norm": 1.5382463932037354,
      "learning_rate": 0.00048776585733107476,
      "loss": 4.3099,
      "step": 650
    },
    {
      "epoch": 0.12422360248447205,
      "grad_norm": 1.5318776369094849,
      "learning_rate": 0.0004875776397515528,
      "loss": 4.4296,
      "step": 660
    },
    {
      "epoch": 0.12610577827969133,
      "grad_norm": 1.3270454406738281,
      "learning_rate": 0.00048738942217203086,
      "loss": 4.2642,
      "step": 670
    },
    {
      "epoch": 0.1279879540749106,
      "grad_norm": 1.3209527730941772,
      "learning_rate": 0.00048720120459250894,
      "loss": 4.4877,
      "step": 680
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 1.1077908277511597,
      "learning_rate": 0.000487012987012987,
      "loss": 4.4918,
      "step": 690
    },
    {
      "epoch": 0.13175230566534915,
      "grad_norm": 1.3791859149932861,
      "learning_rate": 0.0004868247694334651,
      "loss": 4.3768,
      "step": 700
    },
    {
      "epoch": 0.13363448146056842,
      "grad_norm": 1.269770622253418,
      "learning_rate": 0.00048663655185394317,
      "loss": 4.4217,
      "step": 710
    },
    {
      "epoch": 0.13551665725578768,
      "grad_norm": 1.2476558685302734,
      "learning_rate": 0.00048644833427442124,
      "loss": 4.4766,
      "step": 720
    },
    {
      "epoch": 0.13739883305100697,
      "grad_norm": 1.3043088912963867,
      "learning_rate": 0.0004862601166948993,
      "loss": 4.4827,
      "step": 730
    },
    {
      "epoch": 0.13928100884622624,
      "grad_norm": 1.1240277290344238,
      "learning_rate": 0.0004860718991153774,
      "loss": 4.4419,
      "step": 740
    },
    {
      "epoch": 0.1411631846414455,
      "grad_norm": 1.52292001247406,
      "learning_rate": 0.0004858836815358554,
      "loss": 4.4896,
      "step": 750
    },
    {
      "epoch": 0.1430453604366648,
      "grad_norm": 1.5645487308502197,
      "learning_rate": 0.0004856954639563335,
      "loss": 4.4388,
      "step": 760
    },
    {
      "epoch": 0.14492753623188406,
      "grad_norm": 1.3656686544418335,
      "learning_rate": 0.00048550724637681157,
      "loss": 4.4646,
      "step": 770
    },
    {
      "epoch": 0.14680971202710333,
      "grad_norm": 1.2550185918807983,
      "learning_rate": 0.0004853190287972897,
      "loss": 4.3995,
      "step": 780
    },
    {
      "epoch": 0.1486918878223226,
      "grad_norm": 0.9675887227058411,
      "learning_rate": 0.0004851308112177678,
      "loss": 4.4296,
      "step": 790
    },
    {
      "epoch": 0.1505740636175419,
      "grad_norm": 1.1478201150894165,
      "learning_rate": 0.00048494259363824585,
      "loss": 4.3977,
      "step": 800
    },
    {
      "epoch": 0.15245623941276115,
      "grad_norm": 1.4751043319702148,
      "learning_rate": 0.00048475437605872393,
      "loss": 4.3892,
      "step": 810
    },
    {
      "epoch": 0.15433841520798042,
      "grad_norm": 1.5579464435577393,
      "learning_rate": 0.000484566158479202,
      "loss": 4.2969,
      "step": 820
    },
    {
      "epoch": 0.1562205910031997,
      "grad_norm": 1.3829562664031982,
      "learning_rate": 0.00048437794089968003,
      "loss": 4.3862,
      "step": 830
    },
    {
      "epoch": 0.15810276679841898,
      "grad_norm": 1.6938810348510742,
      "learning_rate": 0.0004841897233201581,
      "loss": 4.3947,
      "step": 840
    },
    {
      "epoch": 0.15998494259363824,
      "grad_norm": 1.089751124382019,
      "learning_rate": 0.0004840015057406362,
      "loss": 4.4267,
      "step": 850
    },
    {
      "epoch": 0.1618671183888575,
      "grad_norm": 0.999055802822113,
      "learning_rate": 0.00048381328816111426,
      "loss": 4.504,
      "step": 860
    },
    {
      "epoch": 0.1637492941840768,
      "grad_norm": 1.2850788831710815,
      "learning_rate": 0.00048362507058159233,
      "loss": 4.4024,
      "step": 870
    },
    {
      "epoch": 0.16563146997929606,
      "grad_norm": 2.449049472808838,
      "learning_rate": 0.0004834368530020704,
      "loss": 4.3419,
      "step": 880
    },
    {
      "epoch": 0.16751364577451533,
      "grad_norm": 1.521384358406067,
      "learning_rate": 0.0004832486354225485,
      "loss": 4.4893,
      "step": 890
    },
    {
      "epoch": 0.16939582156973462,
      "grad_norm": 1.0191763639450073,
      "learning_rate": 0.00048306041784302656,
      "loss": 4.4793,
      "step": 900
    },
    {
      "epoch": 0.1712779973649539,
      "grad_norm": 1.553985834121704,
      "learning_rate": 0.00048287220026350464,
      "loss": 4.4463,
      "step": 910
    },
    {
      "epoch": 0.17316017316017315,
      "grad_norm": 1.253300428390503,
      "learning_rate": 0.00048268398268398266,
      "loss": 4.4088,
      "step": 920
    },
    {
      "epoch": 0.17504234895539245,
      "grad_norm": 1.1468899250030518,
      "learning_rate": 0.00048249576510446074,
      "loss": 4.4099,
      "step": 930
    },
    {
      "epoch": 0.1769245247506117,
      "grad_norm": 1.6279855966567993,
      "learning_rate": 0.0004823075475249388,
      "loss": 4.2465,
      "step": 940
    },
    {
      "epoch": 0.17880670054583098,
      "grad_norm": 1.0638151168823242,
      "learning_rate": 0.0004821193299454169,
      "loss": 4.2858,
      "step": 950
    },
    {
      "epoch": 0.18068887634105024,
      "grad_norm": 1.3204110860824585,
      "learning_rate": 0.00048193111236589497,
      "loss": 4.4057,
      "step": 960
    },
    {
      "epoch": 0.18257105213626953,
      "grad_norm": 1.4396264553070068,
      "learning_rate": 0.00048174289478637304,
      "loss": 4.2264,
      "step": 970
    },
    {
      "epoch": 0.1844532279314888,
      "grad_norm": 1.4043022394180298,
      "learning_rate": 0.0004815546772068511,
      "loss": 4.3651,
      "step": 980
    },
    {
      "epoch": 0.18633540372670807,
      "grad_norm": 2.881331443786621,
      "learning_rate": 0.0004813664596273292,
      "loss": 4.3227,
      "step": 990
    },
    {
      "epoch": 0.18821757952192736,
      "grad_norm": 1.3331611156463623,
      "learning_rate": 0.00048117824204780733,
      "loss": 4.4728,
      "step": 1000
    },
    {
      "epoch": 0.19009975531714662,
      "grad_norm": 1.2341046333312988,
      "learning_rate": 0.00048099002446828535,
      "loss": 4.3222,
      "step": 1010
    },
    {
      "epoch": 0.1919819311123659,
      "grad_norm": 1.335827350616455,
      "learning_rate": 0.0004808018068887634,
      "loss": 4.4133,
      "step": 1020
    },
    {
      "epoch": 0.19386410690758518,
      "grad_norm": 1.5440466403961182,
      "learning_rate": 0.0004806135893092415,
      "loss": 4.2855,
      "step": 1030
    },
    {
      "epoch": 0.19574628270280445,
      "grad_norm": 1.4489920139312744,
      "learning_rate": 0.0004804253717297196,
      "loss": 4.3718,
      "step": 1040
    },
    {
      "epoch": 0.1976284584980237,
      "grad_norm": 1.4121023416519165,
      "learning_rate": 0.00048023715415019766,
      "loss": 4.2291,
      "step": 1050
    },
    {
      "epoch": 0.19951063429324298,
      "grad_norm": 1.3732514381408691,
      "learning_rate": 0.00048004893657067573,
      "loss": 4.3611,
      "step": 1060
    },
    {
      "epoch": 0.20139281008846227,
      "grad_norm": 1.475473165512085,
      "learning_rate": 0.0004798607189911538,
      "loss": 4.6097,
      "step": 1070
    },
    {
      "epoch": 0.20327498588368154,
      "grad_norm": 1.2706341743469238,
      "learning_rate": 0.0004796725014116319,
      "loss": 4.2615,
      "step": 1080
    },
    {
      "epoch": 0.2051571616789008,
      "grad_norm": 1.344974398612976,
      "learning_rate": 0.0004794842838321099,
      "loss": 4.3608,
      "step": 1090
    },
    {
      "epoch": 0.2070393374741201,
      "grad_norm": 1.092437505722046,
      "learning_rate": 0.000479296066252588,
      "loss": 4.5277,
      "step": 1100
    },
    {
      "epoch": 0.20892151326933936,
      "grad_norm": 1.5371427536010742,
      "learning_rate": 0.00047910784867306606,
      "loss": 4.3284,
      "step": 1110
    },
    {
      "epoch": 0.21080368906455862,
      "grad_norm": 2.021251916885376,
      "learning_rate": 0.00047891963109354414,
      "loss": 4.3508,
      "step": 1120
    },
    {
      "epoch": 0.2126858648597779,
      "grad_norm": 1.4358062744140625,
      "learning_rate": 0.0004787314135140222,
      "loss": 4.3418,
      "step": 1130
    },
    {
      "epoch": 0.21456804065499718,
      "grad_norm": 1.276908040046692,
      "learning_rate": 0.0004785431959345003,
      "loss": 4.3299,
      "step": 1140
    },
    {
      "epoch": 0.21645021645021645,
      "grad_norm": 1.2175958156585693,
      "learning_rate": 0.00047835497835497837,
      "loss": 4.2586,
      "step": 1150
    },
    {
      "epoch": 0.2183323922454357,
      "grad_norm": 1.2082417011260986,
      "learning_rate": 0.00047816676077545644,
      "loss": 4.4002,
      "step": 1160
    },
    {
      "epoch": 0.220214568040655,
      "grad_norm": 1.368486762046814,
      "learning_rate": 0.0004779785431959345,
      "loss": 4.2378,
      "step": 1170
    },
    {
      "epoch": 0.22209674383587427,
      "grad_norm": 1.3902267217636108,
      "learning_rate": 0.00047779032561641254,
      "loss": 4.3557,
      "step": 1180
    },
    {
      "epoch": 0.22397891963109354,
      "grad_norm": 1.1859028339385986,
      "learning_rate": 0.0004776021080368906,
      "loss": 4.35,
      "step": 1190
    },
    {
      "epoch": 0.22586109542631283,
      "grad_norm": 1.0749539136886597,
      "learning_rate": 0.0004774138904573687,
      "loss": 4.2978,
      "step": 1200
    },
    {
      "epoch": 0.2277432712215321,
      "grad_norm": 1.339378833770752,
      "learning_rate": 0.00047722567287784677,
      "loss": 4.4142,
      "step": 1210
    },
    {
      "epoch": 0.22962544701675136,
      "grad_norm": 1.3744890689849854,
      "learning_rate": 0.0004770374552983249,
      "loss": 4.2964,
      "step": 1220
    },
    {
      "epoch": 0.23150762281197063,
      "grad_norm": 1.2274056673049927,
      "learning_rate": 0.000476849237718803,
      "loss": 4.2599,
      "step": 1230
    },
    {
      "epoch": 0.23338979860718992,
      "grad_norm": 1.4189059734344482,
      "learning_rate": 0.00047666102013928105,
      "loss": 4.1228,
      "step": 1240
    },
    {
      "epoch": 0.23527197440240918,
      "grad_norm": 1.1967359781265259,
      "learning_rate": 0.00047647280255975913,
      "loss": 4.4719,
      "step": 1250
    },
    {
      "epoch": 0.23715415019762845,
      "grad_norm": 1.1312259435653687,
      "learning_rate": 0.0004762845849802372,
      "loss": 4.29,
      "step": 1260
    },
    {
      "epoch": 0.23903632599284774,
      "grad_norm": 1.4020061492919922,
      "learning_rate": 0.00047609636740071523,
      "loss": 4.4895,
      "step": 1270
    },
    {
      "epoch": 0.240918501788067,
      "grad_norm": 1.219526767730713,
      "learning_rate": 0.0004759081498211933,
      "loss": 4.3912,
      "step": 1280
    },
    {
      "epoch": 0.24280067758328627,
      "grad_norm": 1.411958932876587,
      "learning_rate": 0.0004757199322416714,
      "loss": 4.3785,
      "step": 1290
    },
    {
      "epoch": 0.24468285337850557,
      "grad_norm": 1.2712162733078003,
      "learning_rate": 0.00047553171466214946,
      "loss": 4.4343,
      "step": 1300
    },
    {
      "epoch": 0.24656502917372483,
      "grad_norm": 1.3762128353118896,
      "learning_rate": 0.00047534349708262753,
      "loss": 4.2164,
      "step": 1310
    },
    {
      "epoch": 0.2484472049689441,
      "grad_norm": 1.3928210735321045,
      "learning_rate": 0.0004751552795031056,
      "loss": 4.4329,
      "step": 1320
    },
    {
      "epoch": 0.2503293807641634,
      "grad_norm": 6.659452438354492,
      "learning_rate": 0.0004749670619235837,
      "loss": 4.2896,
      "step": 1330
    },
    {
      "epoch": 0.25221155655938265,
      "grad_norm": 1.5140278339385986,
      "learning_rate": 0.00047477884434406176,
      "loss": 4.2381,
      "step": 1340
    },
    {
      "epoch": 0.2540937323546019,
      "grad_norm": 1.4585140943527222,
      "learning_rate": 0.00047459062676453984,
      "loss": 4.2149,
      "step": 1350
    },
    {
      "epoch": 0.2559759081498212,
      "grad_norm": 1.842198371887207,
      "learning_rate": 0.00047440240918501786,
      "loss": 4.2652,
      "step": 1360
    },
    {
      "epoch": 0.25785808394504045,
      "grad_norm": 1.442714810371399,
      "learning_rate": 0.00047421419160549594,
      "loss": 4.2783,
      "step": 1370
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 1.6346895694732666,
      "learning_rate": 0.000474025974025974,
      "loss": 4.1949,
      "step": 1380
    },
    {
      "epoch": 0.26162243553547904,
      "grad_norm": 1.3816555738449097,
      "learning_rate": 0.0004738377564464521,
      "loss": 4.3527,
      "step": 1390
    },
    {
      "epoch": 0.2635046113306983,
      "grad_norm": 1.330315113067627,
      "learning_rate": 0.00047364953886693017,
      "loss": 4.2987,
      "step": 1400
    },
    {
      "epoch": 0.26538678712591757,
      "grad_norm": 1.4008479118347168,
      "learning_rate": 0.00047346132128740824,
      "loss": 4.2562,
      "step": 1410
    },
    {
      "epoch": 0.26726896292113683,
      "grad_norm": 1.5573803186416626,
      "learning_rate": 0.0004732731037078863,
      "loss": 4.2236,
      "step": 1420
    },
    {
      "epoch": 0.2691511387163561,
      "grad_norm": 1.4606269598007202,
      "learning_rate": 0.0004730848861283644,
      "loss": 4.1763,
      "step": 1430
    },
    {
      "epoch": 0.27103331451157536,
      "grad_norm": 1.312117338180542,
      "learning_rate": 0.0004728966685488424,
      "loss": 4.3037,
      "step": 1440
    },
    {
      "epoch": 0.27291549030679463,
      "grad_norm": 1.4533745050430298,
      "learning_rate": 0.00047270845096932055,
      "loss": 4.2849,
      "step": 1450
    },
    {
      "epoch": 0.27479766610201395,
      "grad_norm": 1.4557713270187378,
      "learning_rate": 0.0004725202333897986,
      "loss": 4.2465,
      "step": 1460
    },
    {
      "epoch": 0.2766798418972332,
      "grad_norm": 1.4600456953048706,
      "learning_rate": 0.0004723320158102767,
      "loss": 4.3342,
      "step": 1470
    },
    {
      "epoch": 0.2785620176924525,
      "grad_norm": 1.6165292263031006,
      "learning_rate": 0.0004721437982307548,
      "loss": 4.1626,
      "step": 1480
    },
    {
      "epoch": 0.28044419348767174,
      "grad_norm": 1.9924274682998657,
      "learning_rate": 0.00047195558065123286,
      "loss": 4.1766,
      "step": 1490
    },
    {
      "epoch": 0.282326369282891,
      "grad_norm": 1.4779736995697021,
      "learning_rate": 0.00047176736307171093,
      "loss": 4.1712,
      "step": 1500
    },
    {
      "epoch": 0.2842085450781103,
      "grad_norm": 1.7137341499328613,
      "learning_rate": 0.000471579145492189,
      "loss": 4.3836,
      "step": 1510
    },
    {
      "epoch": 0.2860907208733296,
      "grad_norm": 1.364959955215454,
      "learning_rate": 0.0004713909279126671,
      "loss": 4.3685,
      "step": 1520
    },
    {
      "epoch": 0.28797289666854886,
      "grad_norm": 1.3166732788085938,
      "learning_rate": 0.0004712027103331451,
      "loss": 4.2714,
      "step": 1530
    },
    {
      "epoch": 0.2898550724637681,
      "grad_norm": 1.5516678094863892,
      "learning_rate": 0.0004710144927536232,
      "loss": 4.2025,
      "step": 1540
    },
    {
      "epoch": 0.2917372482589874,
      "grad_norm": 1.399585247039795,
      "learning_rate": 0.00047082627517410126,
      "loss": 4.2243,
      "step": 1550
    },
    {
      "epoch": 0.29361942405420666,
      "grad_norm": 1.48573899269104,
      "learning_rate": 0.00047063805759457934,
      "loss": 4.1842,
      "step": 1560
    },
    {
      "epoch": 0.2955015998494259,
      "grad_norm": 1.4109703302383423,
      "learning_rate": 0.0004704498400150574,
      "loss": 4.2215,
      "step": 1570
    },
    {
      "epoch": 0.2973837756446452,
      "grad_norm": 1.3801723718643188,
      "learning_rate": 0.0004702616224355355,
      "loss": 4.2273,
      "step": 1580
    },
    {
      "epoch": 0.2992659514398645,
      "grad_norm": 1.7305430173873901,
      "learning_rate": 0.00047007340485601357,
      "loss": 4.0852,
      "step": 1590
    },
    {
      "epoch": 0.3011481272350838,
      "grad_norm": 1.4525911808013916,
      "learning_rate": 0.00046988518727649164,
      "loss": 4.3404,
      "step": 1600
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 1.7163445949554443,
      "learning_rate": 0.0004696969696969697,
      "loss": 4.4006,
      "step": 1610
    },
    {
      "epoch": 0.3049124788255223,
      "grad_norm": 1.2165595293045044,
      "learning_rate": 0.00046950875211744774,
      "loss": 4.2984,
      "step": 1620
    },
    {
      "epoch": 0.30679465462074157,
      "grad_norm": 1.5305345058441162,
      "learning_rate": 0.0004693205345379258,
      "loss": 4.2391,
      "step": 1630
    },
    {
      "epoch": 0.30867683041596083,
      "grad_norm": 1.9343332052230835,
      "learning_rate": 0.0004691323169584039,
      "loss": 4.1895,
      "step": 1640
    },
    {
      "epoch": 0.3105590062111801,
      "grad_norm": 1.4054638147354126,
      "learning_rate": 0.00046894409937888197,
      "loss": 4.3532,
      "step": 1650
    },
    {
      "epoch": 0.3124411820063994,
      "grad_norm": 1.3328653573989868,
      "learning_rate": 0.0004687558817993601,
      "loss": 4.0972,
      "step": 1660
    },
    {
      "epoch": 0.3143233578016187,
      "grad_norm": 1.452494502067566,
      "learning_rate": 0.0004685676642198382,
      "loss": 4.1327,
      "step": 1670
    },
    {
      "epoch": 0.31620553359683795,
      "grad_norm": 1.5234417915344238,
      "learning_rate": 0.00046837944664031625,
      "loss": 4.2734,
      "step": 1680
    },
    {
      "epoch": 0.3180877093920572,
      "grad_norm": 1.4561686515808105,
      "learning_rate": 0.00046819122906079433,
      "loss": 4.059,
      "step": 1690
    },
    {
      "epoch": 0.3199698851872765,
      "grad_norm": 1.7445582151412964,
      "learning_rate": 0.00046800301148127235,
      "loss": 4.1765,
      "step": 1700
    },
    {
      "epoch": 0.32185206098249575,
      "grad_norm": 2.538134813308716,
      "learning_rate": 0.00046781479390175043,
      "loss": 4.2543,
      "step": 1710
    },
    {
      "epoch": 0.323734236777715,
      "grad_norm": 1.6532399654388428,
      "learning_rate": 0.0004676265763222285,
      "loss": 4.1489,
      "step": 1720
    },
    {
      "epoch": 0.32561641257293433,
      "grad_norm": 1.4668339490890503,
      "learning_rate": 0.0004674383587427066,
      "loss": 4.2505,
      "step": 1730
    },
    {
      "epoch": 0.3274985883681536,
      "grad_norm": 1.553717017173767,
      "learning_rate": 0.00046725014116318466,
      "loss": 4.2593,
      "step": 1740
    },
    {
      "epoch": 0.32938076416337286,
      "grad_norm": 1.5993154048919678,
      "learning_rate": 0.00046706192358366273,
      "loss": 4.1331,
      "step": 1750
    },
    {
      "epoch": 0.33126293995859213,
      "grad_norm": 1.5123755931854248,
      "learning_rate": 0.0004668737060041408,
      "loss": 4.3395,
      "step": 1760
    },
    {
      "epoch": 0.3331451157538114,
      "grad_norm": 1.5383719205856323,
      "learning_rate": 0.0004666854884246189,
      "loss": 4.1803,
      "step": 1770
    },
    {
      "epoch": 0.33502729154903066,
      "grad_norm": 1.5453109741210938,
      "learning_rate": 0.00046649727084509696,
      "loss": 4.3612,
      "step": 1780
    },
    {
      "epoch": 0.33690946734425,
      "grad_norm": 1.447769284248352,
      "learning_rate": 0.000466309053265575,
      "loss": 4.2859,
      "step": 1790
    },
    {
      "epoch": 0.33879164313946925,
      "grad_norm": 1.6781235933303833,
      "learning_rate": 0.00046612083568605306,
      "loss": 4.0687,
      "step": 1800
    },
    {
      "epoch": 0.3406738189346885,
      "grad_norm": 1.5212960243225098,
      "learning_rate": 0.00046593261810653114,
      "loss": 4.2642,
      "step": 1810
    },
    {
      "epoch": 0.3425559947299078,
      "grad_norm": 1.2794291973114014,
      "learning_rate": 0.0004657444005270092,
      "loss": 4.3414,
      "step": 1820
    },
    {
      "epoch": 0.34443817052512704,
      "grad_norm": 1.3889094591140747,
      "learning_rate": 0.0004655561829474873,
      "loss": 4.1447,
      "step": 1830
    },
    {
      "epoch": 0.3463203463203463,
      "grad_norm": 1.7165048122406006,
      "learning_rate": 0.00046536796536796537,
      "loss": 4.1519,
      "step": 1840
    },
    {
      "epoch": 0.34820252211556557,
      "grad_norm": 1.3793126344680786,
      "learning_rate": 0.00046517974778844344,
      "loss": 4.1838,
      "step": 1850
    },
    {
      "epoch": 0.3500846979107849,
      "grad_norm": 1.2954260110855103,
      "learning_rate": 0.0004649915302089215,
      "loss": 4.2141,
      "step": 1860
    },
    {
      "epoch": 0.35196687370600416,
      "grad_norm": 1.372908592224121,
      "learning_rate": 0.0004648033126293996,
      "loss": 4.2655,
      "step": 1870
    },
    {
      "epoch": 0.3538490495012234,
      "grad_norm": 1.400633692741394,
      "learning_rate": 0.0004646150950498776,
      "loss": 4.2115,
      "step": 1880
    },
    {
      "epoch": 0.3557312252964427,
      "grad_norm": 1.5919150114059448,
      "learning_rate": 0.00046442687747035575,
      "loss": 4.4,
      "step": 1890
    },
    {
      "epoch": 0.35761340109166195,
      "grad_norm": 1.6317856311798096,
      "learning_rate": 0.0004642386598908338,
      "loss": 4.2424,
      "step": 1900
    },
    {
      "epoch": 0.3594955768868812,
      "grad_norm": 1.5302605628967285,
      "learning_rate": 0.0004640504423113119,
      "loss": 4.3054,
      "step": 1910
    },
    {
      "epoch": 0.3613777526821005,
      "grad_norm": 1.3222270011901855,
      "learning_rate": 0.00046386222473179,
      "loss": 4.0888,
      "step": 1920
    },
    {
      "epoch": 0.3632599284773198,
      "grad_norm": 1.514804482460022,
      "learning_rate": 0.00046367400715226806,
      "loss": 4.2651,
      "step": 1930
    },
    {
      "epoch": 0.36514210427253907,
      "grad_norm": 1.5988330841064453,
      "learning_rate": 0.00046348578957274613,
      "loss": 4.2283,
      "step": 1940
    },
    {
      "epoch": 0.36702428006775834,
      "grad_norm": 1.3682210445404053,
      "learning_rate": 0.0004632975719932242,
      "loss": 4.0756,
      "step": 1950
    },
    {
      "epoch": 0.3689064558629776,
      "grad_norm": 1.4985891580581665,
      "learning_rate": 0.00046310935441370223,
      "loss": 4.1255,
      "step": 1960
    },
    {
      "epoch": 0.37078863165819687,
      "grad_norm": 1.6254980564117432,
      "learning_rate": 0.0004629211368341803,
      "loss": 4.1481,
      "step": 1970
    },
    {
      "epoch": 0.37267080745341613,
      "grad_norm": 1.5056555271148682,
      "learning_rate": 0.0004627329192546584,
      "loss": 4.1395,
      "step": 1980
    },
    {
      "epoch": 0.3745529832486354,
      "grad_norm": 1.5908918380737305,
      "learning_rate": 0.00046254470167513646,
      "loss": 4.1166,
      "step": 1990
    },
    {
      "epoch": 0.3764351590438547,
      "grad_norm": 1.4263468980789185,
      "learning_rate": 0.00046235648409561454,
      "loss": 4.1601,
      "step": 2000
    },
    {
      "epoch": 0.378317334839074,
      "grad_norm": 1.8191322088241577,
      "learning_rate": 0.0004621682665160926,
      "loss": 4.3289,
      "step": 2010
    },
    {
      "epoch": 0.38019951063429325,
      "grad_norm": 1.7897543907165527,
      "learning_rate": 0.0004619800489365707,
      "loss": 4.283,
      "step": 2020
    },
    {
      "epoch": 0.3820816864295125,
      "grad_norm": 1.5530258417129517,
      "learning_rate": 0.00046179183135704877,
      "loss": 4.2163,
      "step": 2030
    },
    {
      "epoch": 0.3839638622247318,
      "grad_norm": 1.560866355895996,
      "learning_rate": 0.00046160361377752684,
      "loss": 4.0749,
      "step": 2040
    },
    {
      "epoch": 0.38584603801995104,
      "grad_norm": 1.2918670177459717,
      "learning_rate": 0.00046141539619800486,
      "loss": 4.2007,
      "step": 2050
    },
    {
      "epoch": 0.38772821381517036,
      "grad_norm": 1.4356296062469482,
      "learning_rate": 0.00046122717861848294,
      "loss": 4.2335,
      "step": 2060
    },
    {
      "epoch": 0.38961038961038963,
      "grad_norm": 1.8680976629257202,
      "learning_rate": 0.000461038961038961,
      "loss": 4.0791,
      "step": 2070
    },
    {
      "epoch": 0.3914925654056089,
      "grad_norm": 1.8302345275878906,
      "learning_rate": 0.0004608507434594391,
      "loss": 4.1799,
      "step": 2080
    },
    {
      "epoch": 0.39337474120082816,
      "grad_norm": 1.5143201351165771,
      "learning_rate": 0.00046066252587991717,
      "loss": 4.2113,
      "step": 2090
    },
    {
      "epoch": 0.3952569169960474,
      "grad_norm": 1.3850644826889038,
      "learning_rate": 0.00046047430830039525,
      "loss": 4.2189,
      "step": 2100
    },
    {
      "epoch": 0.3971390927912667,
      "grad_norm": 1.7276452779769897,
      "learning_rate": 0.0004602860907208734,
      "loss": 4.3723,
      "step": 2110
    },
    {
      "epoch": 0.39902126858648596,
      "grad_norm": 1.5579780340194702,
      "learning_rate": 0.00046009787314135145,
      "loss": 4.2855,
      "step": 2120
    },
    {
      "epoch": 0.4009034443817053,
      "grad_norm": 1.269708275794983,
      "learning_rate": 0.00045990965556182953,
      "loss": 4.2179,
      "step": 2130
    },
    {
      "epoch": 0.40278562017692454,
      "grad_norm": 1.565470814704895,
      "learning_rate": 0.00045972143798230755,
      "loss": 4.095,
      "step": 2140
    },
    {
      "epoch": 0.4046677959721438,
      "grad_norm": 1.5279659032821655,
      "learning_rate": 0.00045953322040278563,
      "loss": 4.4537,
      "step": 2150
    },
    {
      "epoch": 0.40654997176736307,
      "grad_norm": 1.3654329776763916,
      "learning_rate": 0.0004593450028232637,
      "loss": 4.306,
      "step": 2160
    },
    {
      "epoch": 0.40843214756258234,
      "grad_norm": 1.5698778629302979,
      "learning_rate": 0.0004591567852437418,
      "loss": 4.1646,
      "step": 2170
    },
    {
      "epoch": 0.4103143233578016,
      "grad_norm": 1.4834023714065552,
      "learning_rate": 0.00045896856766421986,
      "loss": 4.1753,
      "step": 2180
    },
    {
      "epoch": 0.41219649915302087,
      "grad_norm": 1.5586764812469482,
      "learning_rate": 0.00045878035008469793,
      "loss": 4.0278,
      "step": 2190
    },
    {
      "epoch": 0.4140786749482402,
      "grad_norm": 1.7778464555740356,
      "learning_rate": 0.000458592132505176,
      "loss": 4.3505,
      "step": 2200
    },
    {
      "epoch": 0.41596085074345945,
      "grad_norm": 1.6066936254501343,
      "learning_rate": 0.0004584039149256541,
      "loss": 4.0907,
      "step": 2210
    },
    {
      "epoch": 0.4178430265386787,
      "grad_norm": 1.6600680351257324,
      "learning_rate": 0.00045821569734613216,
      "loss": 4.1453,
      "step": 2220
    },
    {
      "epoch": 0.419725202333898,
      "grad_norm": 1.248590111732483,
      "learning_rate": 0.0004580274797666102,
      "loss": 4.1283,
      "step": 2230
    },
    {
      "epoch": 0.42160737812911725,
      "grad_norm": 1.4683887958526611,
      "learning_rate": 0.00045783926218708826,
      "loss": 4.0942,
      "step": 2240
    },
    {
      "epoch": 0.4234895539243365,
      "grad_norm": 1.6800812482833862,
      "learning_rate": 0.00045765104460756634,
      "loss": 4.2249,
      "step": 2250
    },
    {
      "epoch": 0.4253717297195558,
      "grad_norm": 1.4758554697036743,
      "learning_rate": 0.0004574628270280444,
      "loss": 3.9934,
      "step": 2260
    },
    {
      "epoch": 0.4272539055147751,
      "grad_norm": 1.590091586112976,
      "learning_rate": 0.0004572746094485225,
      "loss": 4.0494,
      "step": 2270
    },
    {
      "epoch": 0.42913608130999437,
      "grad_norm": 1.5667383670806885,
      "learning_rate": 0.00045708639186900057,
      "loss": 4.1898,
      "step": 2280
    },
    {
      "epoch": 0.43101825710521363,
      "grad_norm": 1.5386614799499512,
      "learning_rate": 0.00045689817428947864,
      "loss": 4.2019,
      "step": 2290
    },
    {
      "epoch": 0.4329004329004329,
      "grad_norm": 1.4483133554458618,
      "learning_rate": 0.0004567099567099567,
      "loss": 4.2014,
      "step": 2300
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 1.5541331768035889,
      "learning_rate": 0.00045652173913043474,
      "loss": 4.0834,
      "step": 2310
    },
    {
      "epoch": 0.4366647844908714,
      "grad_norm": 1.7615141868591309,
      "learning_rate": 0.0004563335215509128,
      "loss": 4.0695,
      "step": 2320
    },
    {
      "epoch": 0.43854696028609075,
      "grad_norm": 1.6555922031402588,
      "learning_rate": 0.00045614530397139095,
      "loss": 3.9955,
      "step": 2330
    },
    {
      "epoch": 0.44042913608131,
      "grad_norm": 1.6086030006408691,
      "learning_rate": 0.000455957086391869,
      "loss": 4.2755,
      "step": 2340
    },
    {
      "epoch": 0.4423113118765293,
      "grad_norm": 1.4765291213989258,
      "learning_rate": 0.0004557688688123471,
      "loss": 4.1072,
      "step": 2350
    },
    {
      "epoch": 0.44419348767174854,
      "grad_norm": 1.5320438146591187,
      "learning_rate": 0.0004555806512328252,
      "loss": 4.2496,
      "step": 2360
    },
    {
      "epoch": 0.4460756634669678,
      "grad_norm": 1.398559808731079,
      "learning_rate": 0.00045539243365330326,
      "loss": 4.097,
      "step": 2370
    },
    {
      "epoch": 0.4479578392621871,
      "grad_norm": 1.7332468032836914,
      "learning_rate": 0.00045520421607378133,
      "loss": 4.1197,
      "step": 2380
    },
    {
      "epoch": 0.44984001505740634,
      "grad_norm": 1.7244771718978882,
      "learning_rate": 0.0004550159984942594,
      "loss": 4.2804,
      "step": 2390
    },
    {
      "epoch": 0.45172219085262566,
      "grad_norm": 1.318703532218933,
      "learning_rate": 0.00045482778091473743,
      "loss": 4.2258,
      "step": 2400
    },
    {
      "epoch": 0.4536043666478449,
      "grad_norm": 1.726775050163269,
      "learning_rate": 0.0004546395633352155,
      "loss": 4.2059,
      "step": 2410
    },
    {
      "epoch": 0.4554865424430642,
      "grad_norm": 1.4962884187698364,
      "learning_rate": 0.0004544513457556936,
      "loss": 4.0591,
      "step": 2420
    },
    {
      "epoch": 0.45736871823828346,
      "grad_norm": 1.5442137718200684,
      "learning_rate": 0.00045426312817617166,
      "loss": 4.1446,
      "step": 2430
    },
    {
      "epoch": 0.4592508940335027,
      "grad_norm": 1.510866641998291,
      "learning_rate": 0.00045407491059664974,
      "loss": 4.2509,
      "step": 2440
    },
    {
      "epoch": 0.461133069828722,
      "grad_norm": 1.864684820175171,
      "learning_rate": 0.0004538866930171278,
      "loss": 3.9895,
      "step": 2450
    },
    {
      "epoch": 0.46301524562394125,
      "grad_norm": 1.387701392173767,
      "learning_rate": 0.0004536984754376059,
      "loss": 4.1968,
      "step": 2460
    },
    {
      "epoch": 0.4648974214191606,
      "grad_norm": 2.0901427268981934,
      "learning_rate": 0.00045351025785808397,
      "loss": 4.2134,
      "step": 2470
    },
    {
      "epoch": 0.46677959721437984,
      "grad_norm": 1.6149876117706299,
      "learning_rate": 0.00045332204027856204,
      "loss": 4.0816,
      "step": 2480
    },
    {
      "epoch": 0.4686617730095991,
      "grad_norm": 1.5137728452682495,
      "learning_rate": 0.00045313382269904006,
      "loss": 4.1154,
      "step": 2490
    },
    {
      "epoch": 0.47054394880481837,
      "grad_norm": 3.7697200775146484,
      "learning_rate": 0.00045294560511951814,
      "loss": 4.3487,
      "step": 2500
    },
    {
      "epoch": 0.47242612460003763,
      "grad_norm": 1.4249904155731201,
      "learning_rate": 0.0004527573875399962,
      "loss": 4.1435,
      "step": 2510
    },
    {
      "epoch": 0.4743083003952569,
      "grad_norm": 1.586154818534851,
      "learning_rate": 0.0004525691699604743,
      "loss": 4.2354,
      "step": 2520
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 1.3356468677520752,
      "learning_rate": 0.00045238095238095237,
      "loss": 4.2267,
      "step": 2530
    },
    {
      "epoch": 0.4780726519856955,
      "grad_norm": 1.392693042755127,
      "learning_rate": 0.00045219273480143045,
      "loss": 4.2072,
      "step": 2540
    },
    {
      "epoch": 0.47995482778091475,
      "grad_norm": 1.5842061042785645,
      "learning_rate": 0.0004520045172219086,
      "loss": 4.0889,
      "step": 2550
    },
    {
      "epoch": 0.481837003576134,
      "grad_norm": 1.4757062196731567,
      "learning_rate": 0.00045181629964238665,
      "loss": 4.0046,
      "step": 2560
    },
    {
      "epoch": 0.4837191793713533,
      "grad_norm": 1.8355207443237305,
      "learning_rate": 0.0004516280820628647,
      "loss": 4.1834,
      "step": 2570
    },
    {
      "epoch": 0.48560135516657255,
      "grad_norm": 1.4048017263412476,
      "learning_rate": 0.00045143986448334275,
      "loss": 4.1623,
      "step": 2580
    },
    {
      "epoch": 0.4874835309617918,
      "grad_norm": 1.6089762449264526,
      "learning_rate": 0.00045125164690382083,
      "loss": 4.1906,
      "step": 2590
    },
    {
      "epoch": 0.48936570675701113,
      "grad_norm": 1.5948734283447266,
      "learning_rate": 0.0004510634293242989,
      "loss": 4.2207,
      "step": 2600
    },
    {
      "epoch": 0.4912478825522304,
      "grad_norm": 1.6670500040054321,
      "learning_rate": 0.000450875211744777,
      "loss": 4.289,
      "step": 2610
    },
    {
      "epoch": 0.49313005834744966,
      "grad_norm": 1.7025580406188965,
      "learning_rate": 0.00045068699416525506,
      "loss": 4.0739,
      "step": 2620
    },
    {
      "epoch": 0.4950122341426689,
      "grad_norm": 1.7070814371109009,
      "learning_rate": 0.00045049877658573313,
      "loss": 3.8954,
      "step": 2630
    },
    {
      "epoch": 0.4968944099378882,
      "grad_norm": 1.5894651412963867,
      "learning_rate": 0.0004503105590062112,
      "loss": 4.096,
      "step": 2640
    },
    {
      "epoch": 0.49877658573310746,
      "grad_norm": 1.6690545082092285,
      "learning_rate": 0.0004501223414266893,
      "loss": 4.1386,
      "step": 2650
    },
    {
      "epoch": 0.5006587615283268,
      "grad_norm": 2.1716558933258057,
      "learning_rate": 0.0004499341238471673,
      "loss": 4.2536,
      "step": 2660
    },
    {
      "epoch": 0.502540937323546,
      "grad_norm": 1.5564872026443481,
      "learning_rate": 0.0004497459062676454,
      "loss": 4.1387,
      "step": 2670
    },
    {
      "epoch": 0.5044231131187653,
      "grad_norm": 1.5847035646438599,
      "learning_rate": 0.00044955768868812346,
      "loss": 4.2487,
      "step": 2680
    },
    {
      "epoch": 0.5063052889139845,
      "grad_norm": 1.5552998781204224,
      "learning_rate": 0.00044936947110860154,
      "loss": 4.2158,
      "step": 2690
    },
    {
      "epoch": 0.5081874647092038,
      "grad_norm": 1.409266471862793,
      "learning_rate": 0.0004491812535290796,
      "loss": 4.0273,
      "step": 2700
    },
    {
      "epoch": 0.5100696405044232,
      "grad_norm": 1.5766602754592896,
      "learning_rate": 0.0004489930359495577,
      "loss": 4.1938,
      "step": 2710
    },
    {
      "epoch": 0.5119518162996424,
      "grad_norm": 1.191178321838379,
      "learning_rate": 0.00044880481837003577,
      "loss": 4.0563,
      "step": 2720
    },
    {
      "epoch": 0.5138339920948617,
      "grad_norm": 1.6059616804122925,
      "learning_rate": 0.00044861660079051384,
      "loss": 4.2029,
      "step": 2730
    },
    {
      "epoch": 0.5157161678900809,
      "grad_norm": 1.6359925270080566,
      "learning_rate": 0.0004484283832109919,
      "loss": 4.237,
      "step": 2740
    },
    {
      "epoch": 0.5175983436853002,
      "grad_norm": 1.5290448665618896,
      "learning_rate": 0.00044824016563146994,
      "loss": 4.0689,
      "step": 2750
    },
    {
      "epoch": 0.5194805194805194,
      "grad_norm": 1.4904544353485107,
      "learning_rate": 0.000448051948051948,
      "loss": 4.0665,
      "step": 2760
    },
    {
      "epoch": 0.5213626952757388,
      "grad_norm": 1.4702355861663818,
      "learning_rate": 0.00044786373047242615,
      "loss": 4.0486,
      "step": 2770
    },
    {
      "epoch": 0.5232448710709581,
      "grad_norm": 1.407709002494812,
      "learning_rate": 0.0004476755128929042,
      "loss": 4.0626,
      "step": 2780
    },
    {
      "epoch": 0.5251270468661773,
      "grad_norm": 1.6709678173065186,
      "learning_rate": 0.0004474872953133823,
      "loss": 4.1701,
      "step": 2790
    },
    {
      "epoch": 0.5270092226613966,
      "grad_norm": 1.7822628021240234,
      "learning_rate": 0.0004472990777338604,
      "loss": 4.0396,
      "step": 2800
    },
    {
      "epoch": 0.5288913984566158,
      "grad_norm": 1.638765811920166,
      "learning_rate": 0.00044711086015433846,
      "loss": 4.1889,
      "step": 2810
    },
    {
      "epoch": 0.5307735742518351,
      "grad_norm": 1.868028163909912,
      "learning_rate": 0.00044692264257481653,
      "loss": 4.0863,
      "step": 2820
    },
    {
      "epoch": 0.5326557500470543,
      "grad_norm": 1.691738247871399,
      "learning_rate": 0.0004467344249952946,
      "loss": 4.3616,
      "step": 2830
    },
    {
      "epoch": 0.5345379258422737,
      "grad_norm": 1.8202202320098877,
      "learning_rate": 0.00044654620741577263,
      "loss": 4.0839,
      "step": 2840
    },
    {
      "epoch": 0.536420101637493,
      "grad_norm": 1.5820993185043335,
      "learning_rate": 0.0004463579898362507,
      "loss": 4.0845,
      "step": 2850
    },
    {
      "epoch": 0.5383022774327122,
      "grad_norm": 1.599906325340271,
      "learning_rate": 0.0004461697722567288,
      "loss": 3.807,
      "step": 2860
    },
    {
      "epoch": 0.5401844532279315,
      "grad_norm": 1.9688726663589478,
      "learning_rate": 0.00044598155467720686,
      "loss": 3.9582,
      "step": 2870
    },
    {
      "epoch": 0.5420666290231507,
      "grad_norm": 1.6034358739852905,
      "learning_rate": 0.00044579333709768494,
      "loss": 4.1365,
      "step": 2880
    },
    {
      "epoch": 0.54394880481837,
      "grad_norm": 1.613648533821106,
      "learning_rate": 0.000445605119518163,
      "loss": 3.9885,
      "step": 2890
    },
    {
      "epoch": 0.5458309806135893,
      "grad_norm": 1.4738061428070068,
      "learning_rate": 0.0004454169019386411,
      "loss": 4.0579,
      "step": 2900
    },
    {
      "epoch": 0.5477131564088086,
      "grad_norm": 1.5942338705062866,
      "learning_rate": 0.00044522868435911917,
      "loss": 3.9049,
      "step": 2910
    },
    {
      "epoch": 0.5495953322040279,
      "grad_norm": 1.7312630414962769,
      "learning_rate": 0.0004450404667795972,
      "loss": 4.161,
      "step": 2920
    },
    {
      "epoch": 0.5514775079992471,
      "grad_norm": 1.5873860120773315,
      "learning_rate": 0.00044485224920007526,
      "loss": 3.9691,
      "step": 2930
    },
    {
      "epoch": 0.5533596837944664,
      "grad_norm": 1.842726707458496,
      "learning_rate": 0.00044466403162055334,
      "loss": 4.0464,
      "step": 2940
    },
    {
      "epoch": 0.5552418595896856,
      "grad_norm": 1.5641363859176636,
      "learning_rate": 0.0004444758140410314,
      "loss": 3.9814,
      "step": 2950
    },
    {
      "epoch": 0.557124035384905,
      "grad_norm": 1.7177900075912476,
      "learning_rate": 0.0004442875964615095,
      "loss": 4.118,
      "step": 2960
    },
    {
      "epoch": 0.5590062111801242,
      "grad_norm": 1.7312358617782593,
      "learning_rate": 0.00044409937888198757,
      "loss": 4.0702,
      "step": 2970
    },
    {
      "epoch": 0.5608883869753435,
      "grad_norm": 1.684001088142395,
      "learning_rate": 0.00044391116130246565,
      "loss": 4.0587,
      "step": 2980
    },
    {
      "epoch": 0.5627705627705628,
      "grad_norm": 1.5080180168151855,
      "learning_rate": 0.0004437229437229438,
      "loss": 4.2036,
      "step": 2990
    },
    {
      "epoch": 0.564652738565782,
      "grad_norm": 1.512826681137085,
      "learning_rate": 0.00044353472614342185,
      "loss": 4.1267,
      "step": 3000
    },
    {
      "epoch": 0.5665349143610013,
      "grad_norm": 1.5618807077407837,
      "learning_rate": 0.0004433465085638999,
      "loss": 4.0083,
      "step": 3010
    },
    {
      "epoch": 0.5684170901562206,
      "grad_norm": 1.6022028923034668,
      "learning_rate": 0.00044315829098437795,
      "loss": 3.9643,
      "step": 3020
    },
    {
      "epoch": 0.5702992659514399,
      "grad_norm": 1.5317556858062744,
      "learning_rate": 0.00044297007340485603,
      "loss": 4.2311,
      "step": 3030
    },
    {
      "epoch": 0.5721814417466592,
      "grad_norm": 1.5174047946929932,
      "learning_rate": 0.0004427818558253341,
      "loss": 4.0171,
      "step": 3040
    },
    {
      "epoch": 0.5740636175418784,
      "grad_norm": 2.2394042015075684,
      "learning_rate": 0.0004425936382458122,
      "loss": 3.998,
      "step": 3050
    },
    {
      "epoch": 0.5759457933370977,
      "grad_norm": 2.815605878829956,
      "learning_rate": 0.00044240542066629026,
      "loss": 4.0416,
      "step": 3060
    },
    {
      "epoch": 0.5778279691323169,
      "grad_norm": 1.5818989276885986,
      "learning_rate": 0.00044221720308676833,
      "loss": 4.2079,
      "step": 3070
    },
    {
      "epoch": 0.5797101449275363,
      "grad_norm": 1.8354148864746094,
      "learning_rate": 0.0004420289855072464,
      "loss": 4.0311,
      "step": 3080
    },
    {
      "epoch": 0.5815923207227555,
      "grad_norm": 1.6760532855987549,
      "learning_rate": 0.0004418407679277245,
      "loss": 4.1009,
      "step": 3090
    },
    {
      "epoch": 0.5834744965179748,
      "grad_norm": 1.7229177951812744,
      "learning_rate": 0.0004416525503482025,
      "loss": 4.198,
      "step": 3100
    },
    {
      "epoch": 0.5853566723131941,
      "grad_norm": 1.525316596031189,
      "learning_rate": 0.0004414643327686806,
      "loss": 3.8761,
      "step": 3110
    },
    {
      "epoch": 0.5872388481084133,
      "grad_norm": 1.4251718521118164,
      "learning_rate": 0.00044127611518915866,
      "loss": 3.9151,
      "step": 3120
    },
    {
      "epoch": 0.5891210239036326,
      "grad_norm": 2.061256170272827,
      "learning_rate": 0.00044108789760963674,
      "loss": 4.1635,
      "step": 3130
    },
    {
      "epoch": 0.5910031996988518,
      "grad_norm": 1.8884425163269043,
      "learning_rate": 0.0004408996800301148,
      "loss": 3.9408,
      "step": 3140
    },
    {
      "epoch": 0.5928853754940712,
      "grad_norm": 1.5586994886398315,
      "learning_rate": 0.0004407114624505929,
      "loss": 4.1696,
      "step": 3150
    },
    {
      "epoch": 0.5947675512892904,
      "grad_norm": 1.6019229888916016,
      "learning_rate": 0.00044052324487107097,
      "loss": 4.0717,
      "step": 3160
    },
    {
      "epoch": 0.5966497270845097,
      "grad_norm": 1.593672752380371,
      "learning_rate": 0.00044033502729154904,
      "loss": 4.1436,
      "step": 3170
    },
    {
      "epoch": 0.598531902879729,
      "grad_norm": 1.5448564291000366,
      "learning_rate": 0.00044014680971202707,
      "loss": 4.1807,
      "step": 3180
    },
    {
      "epoch": 0.6004140786749482,
      "grad_norm": 1.498767375946045,
      "learning_rate": 0.00043995859213250514,
      "loss": 4.0741,
      "step": 3190
    },
    {
      "epoch": 0.6022962544701675,
      "grad_norm": 1.4406611919403076,
      "learning_rate": 0.0004397703745529832,
      "loss": 3.8672,
      "step": 3200
    },
    {
      "epoch": 0.6041784302653868,
      "grad_norm": 2.3306994438171387,
      "learning_rate": 0.0004395821569734613,
      "loss": 4.0911,
      "step": 3210
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 1.641932487487793,
      "learning_rate": 0.0004393939393939394,
      "loss": 4.1758,
      "step": 3220
    },
    {
      "epoch": 0.6079427818558253,
      "grad_norm": 1.6793614625930786,
      "learning_rate": 0.0004392057218144175,
      "loss": 3.854,
      "step": 3230
    },
    {
      "epoch": 0.6098249576510446,
      "grad_norm": 1.7702709436416626,
      "learning_rate": 0.0004390175042348956,
      "loss": 4.4409,
      "step": 3240
    },
    {
      "epoch": 0.6117071334462639,
      "grad_norm": 1.4433175325393677,
      "learning_rate": 0.00043882928665537366,
      "loss": 3.8387,
      "step": 3250
    },
    {
      "epoch": 0.6135893092414831,
      "grad_norm": 1.8953951597213745,
      "learning_rate": 0.00043864106907585173,
      "loss": 4.238,
      "step": 3260
    },
    {
      "epoch": 0.6154714850367025,
      "grad_norm": 1.5574579238891602,
      "learning_rate": 0.00043845285149632975,
      "loss": 4.0235,
      "step": 3270
    },
    {
      "epoch": 0.6173536608319217,
      "grad_norm": 1.6335999965667725,
      "learning_rate": 0.00043826463391680783,
      "loss": 4.2361,
      "step": 3280
    },
    {
      "epoch": 0.619235836627141,
      "grad_norm": 1.5225999355316162,
      "learning_rate": 0.0004380764163372859,
      "loss": 3.9487,
      "step": 3290
    },
    {
      "epoch": 0.6211180124223602,
      "grad_norm": 1.7795838117599487,
      "learning_rate": 0.000437888198757764,
      "loss": 4.2099,
      "step": 3300
    },
    {
      "epoch": 0.6230001882175795,
      "grad_norm": 1.8371700048446655,
      "learning_rate": 0.00043769998117824206,
      "loss": 4.1858,
      "step": 3310
    },
    {
      "epoch": 0.6248823640127988,
      "grad_norm": 1.5186216831207275,
      "learning_rate": 0.00043751176359872014,
      "loss": 3.9859,
      "step": 3320
    },
    {
      "epoch": 0.626764539808018,
      "grad_norm": 1.853470802307129,
      "learning_rate": 0.0004373235460191982,
      "loss": 4.0308,
      "step": 3330
    },
    {
      "epoch": 0.6286467156032374,
      "grad_norm": 1.5872433185577393,
      "learning_rate": 0.0004371353284396763,
      "loss": 3.9893,
      "step": 3340
    },
    {
      "epoch": 0.6305288913984566,
      "grad_norm": 1.3795815706253052,
      "learning_rate": 0.00043694711086015437,
      "loss": 4.193,
      "step": 3350
    },
    {
      "epoch": 0.6324110671936759,
      "grad_norm": 1.5030205249786377,
      "learning_rate": 0.0004367588932806324,
      "loss": 4.0368,
      "step": 3360
    },
    {
      "epoch": 0.6342932429888951,
      "grad_norm": 1.807570219039917,
      "learning_rate": 0.00043657067570111046,
      "loss": 3.9148,
      "step": 3370
    },
    {
      "epoch": 0.6361754187841144,
      "grad_norm": 1.7171461582183838,
      "learning_rate": 0.00043638245812158854,
      "loss": 4.1182,
      "step": 3380
    },
    {
      "epoch": 0.6380575945793338,
      "grad_norm": 1.5727144479751587,
      "learning_rate": 0.0004361942405420666,
      "loss": 4.0768,
      "step": 3390
    },
    {
      "epoch": 0.639939770374553,
      "grad_norm": 1.7362148761749268,
      "learning_rate": 0.0004360060229625447,
      "loss": 3.9668,
      "step": 3400
    },
    {
      "epoch": 0.6418219461697723,
      "grad_norm": 1.6728789806365967,
      "learning_rate": 0.00043581780538302277,
      "loss": 4.0369,
      "step": 3410
    },
    {
      "epoch": 0.6437041219649915,
      "grad_norm": 1.7012757062911987,
      "learning_rate": 0.00043562958780350085,
      "loss": 4.12,
      "step": 3420
    },
    {
      "epoch": 0.6455862977602108,
      "grad_norm": 1.728248119354248,
      "learning_rate": 0.000435441370223979,
      "loss": 4.0729,
      "step": 3430
    },
    {
      "epoch": 0.64746847355543,
      "grad_norm": 1.8527989387512207,
      "learning_rate": 0.00043525315264445705,
      "loss": 4.2596,
      "step": 3440
    },
    {
      "epoch": 0.6493506493506493,
      "grad_norm": 1.7408626079559326,
      "learning_rate": 0.0004350649350649351,
      "loss": 4.1345,
      "step": 3450
    },
    {
      "epoch": 0.6512328251458687,
      "grad_norm": 1.4284261465072632,
      "learning_rate": 0.00043487671748541315,
      "loss": 4.1048,
      "step": 3460
    },
    {
      "epoch": 0.6531150009410879,
      "grad_norm": 1.3651869297027588,
      "learning_rate": 0.00043468849990589123,
      "loss": 4.0933,
      "step": 3470
    },
    {
      "epoch": 0.6549971767363072,
      "grad_norm": 1.5003199577331543,
      "learning_rate": 0.0004345002823263693,
      "loss": 4.185,
      "step": 3480
    },
    {
      "epoch": 0.6568793525315264,
      "grad_norm": 1.5484837293624878,
      "learning_rate": 0.0004343120647468474,
      "loss": 4.1184,
      "step": 3490
    },
    {
      "epoch": 0.6587615283267457,
      "grad_norm": 1.3103121519088745,
      "learning_rate": 0.00043412384716732546,
      "loss": 4.0191,
      "step": 3500
    },
    {
      "epoch": 0.6606437041219649,
      "grad_norm": 1.587458848953247,
      "learning_rate": 0.00043393562958780353,
      "loss": 4.0854,
      "step": 3510
    },
    {
      "epoch": 0.6625258799171843,
      "grad_norm": 2.2741780281066895,
      "learning_rate": 0.0004337474120082816,
      "loss": 4.0218,
      "step": 3520
    },
    {
      "epoch": 0.6644080557124036,
      "grad_norm": 1.7086808681488037,
      "learning_rate": 0.00043355919442875963,
      "loss": 3.8825,
      "step": 3530
    },
    {
      "epoch": 0.6662902315076228,
      "grad_norm": 1.551834225654602,
      "learning_rate": 0.0004333709768492377,
      "loss": 4.2044,
      "step": 3540
    },
    {
      "epoch": 0.6681724073028421,
      "grad_norm": 1.484415888786316,
      "learning_rate": 0.0004331827592697158,
      "loss": 3.9214,
      "step": 3550
    },
    {
      "epoch": 0.6700545830980613,
      "grad_norm": 1.6641745567321777,
      "learning_rate": 0.00043299454169019386,
      "loss": 4.1115,
      "step": 3560
    },
    {
      "epoch": 0.6719367588932806,
      "grad_norm": 1.9629614353179932,
      "learning_rate": 0.00043280632411067194,
      "loss": 4.0877,
      "step": 3570
    },
    {
      "epoch": 0.6738189346885,
      "grad_norm": 1.6242876052856445,
      "learning_rate": 0.00043261810653115,
      "loss": 3.9146,
      "step": 3580
    },
    {
      "epoch": 0.6757011104837192,
      "grad_norm": 1.8593428134918213,
      "learning_rate": 0.0004324298889516281,
      "loss": 4.0876,
      "step": 3590
    },
    {
      "epoch": 0.6775832862789385,
      "grad_norm": 1.5309560298919678,
      "learning_rate": 0.00043224167137210617,
      "loss": 3.9557,
      "step": 3600
    },
    {
      "epoch": 0.6794654620741577,
      "grad_norm": 1.5672924518585205,
      "learning_rate": 0.00043205345379258424,
      "loss": 4.0553,
      "step": 3610
    },
    {
      "epoch": 0.681347637869377,
      "grad_norm": 2.1588938236236572,
      "learning_rate": 0.00043186523621306227,
      "loss": 3.9586,
      "step": 3620
    },
    {
      "epoch": 0.6832298136645962,
      "grad_norm": 2.019364595413208,
      "learning_rate": 0.00043167701863354034,
      "loss": 3.9824,
      "step": 3630
    },
    {
      "epoch": 0.6851119894598156,
      "grad_norm": 2.110015869140625,
      "learning_rate": 0.0004314888010540184,
      "loss": 3.8083,
      "step": 3640
    },
    {
      "epoch": 0.6869941652550349,
      "grad_norm": 2.0237793922424316,
      "learning_rate": 0.0004313005834744965,
      "loss": 4.0203,
      "step": 3650
    },
    {
      "epoch": 0.6888763410502541,
      "grad_norm": 1.7034662961959839,
      "learning_rate": 0.0004311123658949746,
      "loss": 3.8691,
      "step": 3660
    },
    {
      "epoch": 0.6907585168454734,
      "grad_norm": 1.8328206539154053,
      "learning_rate": 0.0004309241483154527,
      "loss": 4.2907,
      "step": 3670
    },
    {
      "epoch": 0.6926406926406926,
      "grad_norm": 1.8003268241882324,
      "learning_rate": 0.0004307359307359308,
      "loss": 3.8922,
      "step": 3680
    },
    {
      "epoch": 0.6945228684359119,
      "grad_norm": 2.132638692855835,
      "learning_rate": 0.00043054771315640886,
      "loss": 4.1068,
      "step": 3690
    },
    {
      "epoch": 0.6964050442311311,
      "grad_norm": 1.8007606267929077,
      "learning_rate": 0.00043035949557688693,
      "loss": 4.0505,
      "step": 3700
    },
    {
      "epoch": 0.6982872200263505,
      "grad_norm": 1.9239156246185303,
      "learning_rate": 0.00043017127799736495,
      "loss": 3.9898,
      "step": 3710
    },
    {
      "epoch": 0.7001693958215698,
      "grad_norm": 1.4964417219161987,
      "learning_rate": 0.00042998306041784303,
      "loss": 4.0079,
      "step": 3720
    },
    {
      "epoch": 0.702051571616789,
      "grad_norm": 1.6524834632873535,
      "learning_rate": 0.0004297948428383211,
      "loss": 4.118,
      "step": 3730
    },
    {
      "epoch": 0.7039337474120083,
      "grad_norm": 1.7997394800186157,
      "learning_rate": 0.0004296066252587992,
      "loss": 4.1419,
      "step": 3740
    },
    {
      "epoch": 0.7058159232072275,
      "grad_norm": 1.7603429555892944,
      "learning_rate": 0.00042941840767927726,
      "loss": 4.0473,
      "step": 3750
    },
    {
      "epoch": 0.7076980990024468,
      "grad_norm": 1.5642491579055786,
      "learning_rate": 0.00042923019009975534,
      "loss": 4.0603,
      "step": 3760
    },
    {
      "epoch": 0.709580274797666,
      "grad_norm": 2.023898124694824,
      "learning_rate": 0.0004290419725202334,
      "loss": 3.899,
      "step": 3770
    },
    {
      "epoch": 0.7114624505928854,
      "grad_norm": 1.593841791152954,
      "learning_rate": 0.0004288537549407115,
      "loss": 3.9971,
      "step": 3780
    },
    {
      "epoch": 0.7133446263881047,
      "grad_norm": 1.7649803161621094,
      "learning_rate": 0.0004286655373611895,
      "loss": 3.8812,
      "step": 3790
    },
    {
      "epoch": 0.7152268021833239,
      "grad_norm": 1.7265058755874634,
      "learning_rate": 0.0004284773197816676,
      "loss": 4.1328,
      "step": 3800
    },
    {
      "epoch": 0.7171089779785432,
      "grad_norm": 1.5733685493469238,
      "learning_rate": 0.00042828910220214566,
      "loss": 4.0023,
      "step": 3810
    },
    {
      "epoch": 0.7189911537737624,
      "grad_norm": 1.5658947229385376,
      "learning_rate": 0.00042810088462262374,
      "loss": 4.1734,
      "step": 3820
    },
    {
      "epoch": 0.7208733295689818,
      "grad_norm": 1.6648072004318237,
      "learning_rate": 0.0004279126670431018,
      "loss": 4.1272,
      "step": 3830
    },
    {
      "epoch": 0.722755505364201,
      "grad_norm": 1.7949475049972534,
      "learning_rate": 0.0004277244494635799,
      "loss": 3.9424,
      "step": 3840
    },
    {
      "epoch": 0.7246376811594203,
      "grad_norm": 1.9354840517044067,
      "learning_rate": 0.00042753623188405797,
      "loss": 4.1731,
      "step": 3850
    },
    {
      "epoch": 0.7265198569546396,
      "grad_norm": 1.7900172472000122,
      "learning_rate": 0.00042734801430453605,
      "loss": 3.9624,
      "step": 3860
    },
    {
      "epoch": 0.7284020327498588,
      "grad_norm": 2.012629985809326,
      "learning_rate": 0.0004271597967250141,
      "loss": 4.169,
      "step": 3870
    },
    {
      "epoch": 0.7302842085450781,
      "grad_norm": 1.7707184553146362,
      "learning_rate": 0.0004269715791454922,
      "loss": 3.872,
      "step": 3880
    },
    {
      "epoch": 0.7321663843402973,
      "grad_norm": 1.6689708232879639,
      "learning_rate": 0.0004267833615659703,
      "loss": 3.9675,
      "step": 3890
    },
    {
      "epoch": 0.7340485601355167,
      "grad_norm": 1.8784040212631226,
      "learning_rate": 0.00042659514398644835,
      "loss": 4.0006,
      "step": 3900
    },
    {
      "epoch": 0.7359307359307359,
      "grad_norm": 2.014258623123169,
      "learning_rate": 0.00042640692640692643,
      "loss": 4.168,
      "step": 3910
    },
    {
      "epoch": 0.7378129117259552,
      "grad_norm": 2.2676868438720703,
      "learning_rate": 0.0004262187088274045,
      "loss": 3.9855,
      "step": 3920
    },
    {
      "epoch": 0.7396950875211745,
      "grad_norm": 1.5652729272842407,
      "learning_rate": 0.0004260304912478826,
      "loss": 4.0502,
      "step": 3930
    },
    {
      "epoch": 0.7415772633163937,
      "grad_norm": 2.0387094020843506,
      "learning_rate": 0.00042584227366836066,
      "loss": 3.8425,
      "step": 3940
    },
    {
      "epoch": 0.743459439111613,
      "grad_norm": 1.6358362436294556,
      "learning_rate": 0.00042565405608883873,
      "loss": 4.0084,
      "step": 3950
    },
    {
      "epoch": 0.7453416149068323,
      "grad_norm": 1.7088682651519775,
      "learning_rate": 0.0004254658385093168,
      "loss": 4.1777,
      "step": 3960
    },
    {
      "epoch": 0.7472237907020516,
      "grad_norm": 1.7116954326629639,
      "learning_rate": 0.00042527762092979483,
      "loss": 3.8762,
      "step": 3970
    },
    {
      "epoch": 0.7491059664972708,
      "grad_norm": 2.027803897857666,
      "learning_rate": 0.0004250894033502729,
      "loss": 3.9327,
      "step": 3980
    },
    {
      "epoch": 0.7509881422924901,
      "grad_norm": 2.3380789756774902,
      "learning_rate": 0.000424901185770751,
      "loss": 4.0325,
      "step": 3990
    },
    {
      "epoch": 0.7528703180877094,
      "grad_norm": 2.0593698024749756,
      "learning_rate": 0.00042471296819122906,
      "loss": 4.0288,
      "step": 4000
    },
    {
      "epoch": 0.7547524938829286,
      "grad_norm": 1.9557839632034302,
      "learning_rate": 0.00042452475061170714,
      "loss": 4.1358,
      "step": 4010
    },
    {
      "epoch": 0.756634669678148,
      "grad_norm": 1.571593165397644,
      "learning_rate": 0.0004243365330321852,
      "loss": 3.9639,
      "step": 4020
    },
    {
      "epoch": 0.7585168454733672,
      "grad_norm": 1.6397957801818848,
      "learning_rate": 0.0004241483154526633,
      "loss": 4.0101,
      "step": 4030
    },
    {
      "epoch": 0.7603990212685865,
      "grad_norm": 1.9671742916107178,
      "learning_rate": 0.00042396009787314137,
      "loss": 4.0802,
      "step": 4040
    },
    {
      "epoch": 0.7622811970638057,
      "grad_norm": 1.6972992420196533,
      "learning_rate": 0.00042377188029361944,
      "loss": 3.8887,
      "step": 4050
    },
    {
      "epoch": 0.764163372859025,
      "grad_norm": 1.699816107749939,
      "learning_rate": 0.00042358366271409747,
      "loss": 3.8207,
      "step": 4060
    },
    {
      "epoch": 0.7660455486542443,
      "grad_norm": 1.866045355796814,
      "learning_rate": 0.00042339544513457554,
      "loss": 3.9556,
      "step": 4070
    },
    {
      "epoch": 0.7679277244494636,
      "grad_norm": 1.6141557693481445,
      "learning_rate": 0.0004232072275550536,
      "loss": 3.8906,
      "step": 4080
    },
    {
      "epoch": 0.7698099002446829,
      "grad_norm": 2.044933319091797,
      "learning_rate": 0.0004230190099755317,
      "loss": 4.1045,
      "step": 4090
    },
    {
      "epoch": 0.7716920760399021,
      "grad_norm": 1.7318826913833618,
      "learning_rate": 0.0004228307923960098,
      "loss": 4.0916,
      "step": 4100
    },
    {
      "epoch": 0.7735742518351214,
      "grad_norm": 1.7338465452194214,
      "learning_rate": 0.0004226425748164879,
      "loss": 4.1873,
      "step": 4110
    },
    {
      "epoch": 0.7754564276303407,
      "grad_norm": 1.597584843635559,
      "learning_rate": 0.000422454357236966,
      "loss": 3.8938,
      "step": 4120
    },
    {
      "epoch": 0.7773386034255599,
      "grad_norm": 1.9002665281295776,
      "learning_rate": 0.00042226613965744406,
      "loss": 4.2353,
      "step": 4130
    },
    {
      "epoch": 0.7792207792207793,
      "grad_norm": 1.9633022546768188,
      "learning_rate": 0.0004220779220779221,
      "loss": 4.0478,
      "step": 4140
    },
    {
      "epoch": 0.7811029550159985,
      "grad_norm": 1.6098625659942627,
      "learning_rate": 0.00042188970449840015,
      "loss": 3.8488,
      "step": 4150
    },
    {
      "epoch": 0.7829851308112178,
      "grad_norm": 1.7783234119415283,
      "learning_rate": 0.00042170148691887823,
      "loss": 4.2271,
      "step": 4160
    },
    {
      "epoch": 0.784867306606437,
      "grad_norm": 1.3400135040283203,
      "learning_rate": 0.0004215132693393563,
      "loss": 3.9819,
      "step": 4170
    },
    {
      "epoch": 0.7867494824016563,
      "grad_norm": 1.805132508277893,
      "learning_rate": 0.0004213250517598344,
      "loss": 3.9726,
      "step": 4180
    },
    {
      "epoch": 0.7886316581968756,
      "grad_norm": 1.5507073402404785,
      "learning_rate": 0.00042113683418031246,
      "loss": 4.1114,
      "step": 4190
    },
    {
      "epoch": 0.7905138339920948,
      "grad_norm": 1.6533308029174805,
      "learning_rate": 0.00042094861660079054,
      "loss": 4.0381,
      "step": 4200
    },
    {
      "epoch": 0.7923960097873142,
      "grad_norm": 1.5455400943756104,
      "learning_rate": 0.0004207603990212686,
      "loss": 3.9636,
      "step": 4210
    },
    {
      "epoch": 0.7942781855825334,
      "grad_norm": 1.6970956325531006,
      "learning_rate": 0.0004205721814417467,
      "loss": 3.8815,
      "step": 4220
    },
    {
      "epoch": 0.7961603613777527,
      "grad_norm": 1.7900029420852661,
      "learning_rate": 0.0004203839638622247,
      "loss": 3.8217,
      "step": 4230
    },
    {
      "epoch": 0.7980425371729719,
      "grad_norm": 1.7710243463516235,
      "learning_rate": 0.0004201957462827028,
      "loss": 3.9771,
      "step": 4240
    },
    {
      "epoch": 0.7999247129681912,
      "grad_norm": 1.6762468814849854,
      "learning_rate": 0.00042000752870318086,
      "loss": 3.9662,
      "step": 4250
    },
    {
      "epoch": 0.8018068887634106,
      "grad_norm": 1.785457968711853,
      "learning_rate": 0.00041981931112365894,
      "loss": 4.0751,
      "step": 4260
    },
    {
      "epoch": 0.8036890645586298,
      "grad_norm": 1.6882836818695068,
      "learning_rate": 0.000419631093544137,
      "loss": 3.8162,
      "step": 4270
    },
    {
      "epoch": 0.8055712403538491,
      "grad_norm": 1.752007246017456,
      "learning_rate": 0.0004194428759646151,
      "loss": 3.9087,
      "step": 4280
    },
    {
      "epoch": 0.8074534161490683,
      "grad_norm": 1.6861228942871094,
      "learning_rate": 0.00041925465838509317,
      "loss": 4.0087,
      "step": 4290
    },
    {
      "epoch": 0.8093355919442876,
      "grad_norm": 2.41631817817688,
      "learning_rate": 0.00041906644080557125,
      "loss": 3.8155,
      "step": 4300
    },
    {
      "epoch": 0.8112177677395068,
      "grad_norm": 1.6490652561187744,
      "learning_rate": 0.0004188782232260493,
      "loss": 3.8881,
      "step": 4310
    },
    {
      "epoch": 0.8130999435347261,
      "grad_norm": 1.7375346422195435,
      "learning_rate": 0.0004186900056465274,
      "loss": 4.0091,
      "step": 4320
    },
    {
      "epoch": 0.8149821193299455,
      "grad_norm": 1.9426676034927368,
      "learning_rate": 0.0004185017880670055,
      "loss": 3.882,
      "step": 4330
    },
    {
      "epoch": 0.8168642951251647,
      "grad_norm": 1.895981788635254,
      "learning_rate": 0.00041831357048748355,
      "loss": 4.3555,
      "step": 4340
    },
    {
      "epoch": 0.818746470920384,
      "grad_norm": 1.5857805013656616,
      "learning_rate": 0.00041812535290796163,
      "loss": 3.9986,
      "step": 4350
    },
    {
      "epoch": 0.8206286467156032,
      "grad_norm": 2.0020341873168945,
      "learning_rate": 0.0004179371353284397,
      "loss": 3.8985,
      "step": 4360
    },
    {
      "epoch": 0.8225108225108225,
      "grad_norm": 1.570229172706604,
      "learning_rate": 0.0004177489177489178,
      "loss": 3.9683,
      "step": 4370
    },
    {
      "epoch": 0.8243929983060417,
      "grad_norm": 1.7034296989440918,
      "learning_rate": 0.00041756070016939586,
      "loss": 3.7937,
      "step": 4380
    },
    {
      "epoch": 0.8262751741012611,
      "grad_norm": 1.8847661018371582,
      "learning_rate": 0.00041737248258987393,
      "loss": 4.1219,
      "step": 4390
    },
    {
      "epoch": 0.8281573498964804,
      "grad_norm": 2.25219988822937,
      "learning_rate": 0.00041718426501035196,
      "loss": 4.2165,
      "step": 4400
    },
    {
      "epoch": 0.8300395256916996,
      "grad_norm": 1.679608702659607,
      "learning_rate": 0.00041699604743083003,
      "loss": 4.1739,
      "step": 4410
    },
    {
      "epoch": 0.8319217014869189,
      "grad_norm": 1.7975382804870605,
      "learning_rate": 0.0004168078298513081,
      "loss": 3.9637,
      "step": 4420
    },
    {
      "epoch": 0.8338038772821381,
      "grad_norm": 1.6868278980255127,
      "learning_rate": 0.0004166196122717862,
      "loss": 3.7738,
      "step": 4430
    },
    {
      "epoch": 0.8356860530773574,
      "grad_norm": 1.7453500032424927,
      "learning_rate": 0.00041643139469226426,
      "loss": 3.9673,
      "step": 4440
    },
    {
      "epoch": 0.8375682288725766,
      "grad_norm": 1.622382640838623,
      "learning_rate": 0.00041624317711274234,
      "loss": 4.0327,
      "step": 4450
    },
    {
      "epoch": 0.839450404667796,
      "grad_norm": 2.1576337814331055,
      "learning_rate": 0.0004160549595332204,
      "loss": 3.91,
      "step": 4460
    },
    {
      "epoch": 0.8413325804630153,
      "grad_norm": 1.755730390548706,
      "learning_rate": 0.0004158667419536985,
      "loss": 3.8582,
      "step": 4470
    },
    {
      "epoch": 0.8432147562582345,
      "grad_norm": 1.8487035036087036,
      "learning_rate": 0.00041567852437417657,
      "loss": 4.0083,
      "step": 4480
    },
    {
      "epoch": 0.8450969320534538,
      "grad_norm": 1.5082343816757202,
      "learning_rate": 0.0004154903067946546,
      "loss": 3.865,
      "step": 4490
    },
    {
      "epoch": 0.846979107848673,
      "grad_norm": 2.0205516815185547,
      "learning_rate": 0.00041530208921513267,
      "loss": 3.9817,
      "step": 4500
    },
    {
      "epoch": 0.8488612836438924,
      "grad_norm": 1.7600107192993164,
      "learning_rate": 0.00041511387163561074,
      "loss": 4.1025,
      "step": 4510
    },
    {
      "epoch": 0.8507434594391116,
      "grad_norm": 1.8236727714538574,
      "learning_rate": 0.0004149256540560888,
      "loss": 3.9428,
      "step": 4520
    },
    {
      "epoch": 0.8526256352343309,
      "grad_norm": 1.743336796760559,
      "learning_rate": 0.0004147374364765669,
      "loss": 4.0617,
      "step": 4530
    },
    {
      "epoch": 0.8545078110295502,
      "grad_norm": 1.7079042196273804,
      "learning_rate": 0.000414549218897045,
      "loss": 3.9175,
      "step": 4540
    },
    {
      "epoch": 0.8563899868247694,
      "grad_norm": 1.808450698852539,
      "learning_rate": 0.0004143610013175231,
      "loss": 3.911,
      "step": 4550
    },
    {
      "epoch": 0.8582721626199887,
      "grad_norm": 1.5372434854507446,
      "learning_rate": 0.0004141727837380012,
      "loss": 3.847,
      "step": 4560
    },
    {
      "epoch": 0.8601543384152079,
      "grad_norm": 1.904667854309082,
      "learning_rate": 0.00041398456615847926,
      "loss": 4.0016,
      "step": 4570
    },
    {
      "epoch": 0.8620365142104273,
      "grad_norm": 1.7665495872497559,
      "learning_rate": 0.0004137963485789573,
      "loss": 3.7517,
      "step": 4580
    },
    {
      "epoch": 0.8639186900056465,
      "grad_norm": 1.6588249206542969,
      "learning_rate": 0.00041360813099943535,
      "loss": 3.8672,
      "step": 4590
    },
    {
      "epoch": 0.8658008658008658,
      "grad_norm": 1.5419063568115234,
      "learning_rate": 0.00041341991341991343,
      "loss": 4.1185,
      "step": 4600
    },
    {
      "epoch": 0.8676830415960851,
      "grad_norm": 1.8266115188598633,
      "learning_rate": 0.0004132316958403915,
      "loss": 3.8609,
      "step": 4610
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 1.8634611368179321,
      "learning_rate": 0.0004130434782608696,
      "loss": 4.0257,
      "step": 4620
    },
    {
      "epoch": 0.8714473931865236,
      "grad_norm": 1.7247673273086548,
      "learning_rate": 0.00041285526068134766,
      "loss": 4.1326,
      "step": 4630
    },
    {
      "epoch": 0.8733295689817429,
      "grad_norm": 1.533218264579773,
      "learning_rate": 0.00041266704310182574,
      "loss": 3.8748,
      "step": 4640
    },
    {
      "epoch": 0.8752117447769622,
      "grad_norm": 1.7721878290176392,
      "learning_rate": 0.0004124788255223038,
      "loss": 3.991,
      "step": 4650
    },
    {
      "epoch": 0.8770939205721815,
      "grad_norm": 1.8932552337646484,
      "learning_rate": 0.00041229060794278183,
      "loss": 4.0442,
      "step": 4660
    },
    {
      "epoch": 0.8789760963674007,
      "grad_norm": 1.8900960683822632,
      "learning_rate": 0.0004121023903632599,
      "loss": 3.8255,
      "step": 4670
    },
    {
      "epoch": 0.88085827216262,
      "grad_norm": 1.672855019569397,
      "learning_rate": 0.000411914172783738,
      "loss": 3.8736,
      "step": 4680
    },
    {
      "epoch": 0.8827404479578392,
      "grad_norm": 1.7743964195251465,
      "learning_rate": 0.00041172595520421606,
      "loss": 3.9798,
      "step": 4690
    },
    {
      "epoch": 0.8846226237530586,
      "grad_norm": 1.7346552610397339,
      "learning_rate": 0.00041153773762469414,
      "loss": 3.9649,
      "step": 4700
    },
    {
      "epoch": 0.8865047995482778,
      "grad_norm": 1.8058704137802124,
      "learning_rate": 0.0004113495200451722,
      "loss": 3.8931,
      "step": 4710
    },
    {
      "epoch": 0.8883869753434971,
      "grad_norm": 1.6042428016662598,
      "learning_rate": 0.0004111613024656503,
      "loss": 3.9251,
      "step": 4720
    },
    {
      "epoch": 0.8902691511387164,
      "grad_norm": 1.6752597093582153,
      "learning_rate": 0.00041097308488612837,
      "loss": 4.0351,
      "step": 4730
    },
    {
      "epoch": 0.8921513269339356,
      "grad_norm": 1.673338532447815,
      "learning_rate": 0.00041078486730660645,
      "loss": 3.9119,
      "step": 4740
    },
    {
      "epoch": 0.8940335027291549,
      "grad_norm": 1.7413945198059082,
      "learning_rate": 0.00041059664972708447,
      "loss": 3.9578,
      "step": 4750
    },
    {
      "epoch": 0.8959156785243741,
      "grad_norm": 1.8221259117126465,
      "learning_rate": 0.00041040843214756254,
      "loss": 3.8471,
      "step": 4760
    },
    {
      "epoch": 0.8977978543195935,
      "grad_norm": 1.9183502197265625,
      "learning_rate": 0.0004102202145680407,
      "loss": 4.0963,
      "step": 4770
    },
    {
      "epoch": 0.8996800301148127,
      "grad_norm": 1.765283465385437,
      "learning_rate": 0.00041003199698851875,
      "loss": 4.006,
      "step": 4780
    },
    {
      "epoch": 0.901562205910032,
      "grad_norm": 1.787667155265808,
      "learning_rate": 0.00040984377940899683,
      "loss": 3.9434,
      "step": 4790
    },
    {
      "epoch": 0.9034443817052513,
      "grad_norm": 1.744106650352478,
      "learning_rate": 0.0004096555618294749,
      "loss": 4.0278,
      "step": 4800
    },
    {
      "epoch": 0.9053265575004705,
      "grad_norm": 1.6062647104263306,
      "learning_rate": 0.000409467344249953,
      "loss": 4.099,
      "step": 4810
    },
    {
      "epoch": 0.9072087332956899,
      "grad_norm": 1.778028964996338,
      "learning_rate": 0.00040927912667043106,
      "loss": 3.9152,
      "step": 4820
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 1.7794740200042725,
      "learning_rate": 0.00040909090909090913,
      "loss": 3.9034,
      "step": 4830
    },
    {
      "epoch": 0.9109730848861284,
      "grad_norm": 1.8644254207611084,
      "learning_rate": 0.00040890269151138716,
      "loss": 4.0704,
      "step": 4840
    },
    {
      "epoch": 0.9128552606813476,
      "grad_norm": 1.7791409492492676,
      "learning_rate": 0.00040871447393186523,
      "loss": 4.0014,
      "step": 4850
    },
    {
      "epoch": 0.9147374364765669,
      "grad_norm": 2.3520307540893555,
      "learning_rate": 0.0004085262563523433,
      "loss": 3.8349,
      "step": 4860
    },
    {
      "epoch": 0.9166196122717862,
      "grad_norm": 1.7046860456466675,
      "learning_rate": 0.0004083380387728214,
      "loss": 3.95,
      "step": 4870
    },
    {
      "epoch": 0.9185017880670054,
      "grad_norm": 2.0061821937561035,
      "learning_rate": 0.00040814982119329946,
      "loss": 3.8543,
      "step": 4880
    },
    {
      "epoch": 0.9203839638622248,
      "grad_norm": 1.903624176979065,
      "learning_rate": 0.00040796160361377754,
      "loss": 3.9311,
      "step": 4890
    },
    {
      "epoch": 0.922266139657444,
      "grad_norm": 1.834676742553711,
      "learning_rate": 0.0004077733860342556,
      "loss": 4.1316,
      "step": 4900
    },
    {
      "epoch": 0.9241483154526633,
      "grad_norm": 1.7539821863174438,
      "learning_rate": 0.0004075851684547337,
      "loss": 4.0676,
      "step": 4910
    },
    {
      "epoch": 0.9260304912478825,
      "grad_norm": 2.8327972888946533,
      "learning_rate": 0.00040739695087521177,
      "loss": 4.1533,
      "step": 4920
    },
    {
      "epoch": 0.9279126670431018,
      "grad_norm": 2.6836674213409424,
      "learning_rate": 0.0004072087332956898,
      "loss": 3.8235,
      "step": 4930
    },
    {
      "epoch": 0.9297948428383211,
      "grad_norm": 2.0225014686584473,
      "learning_rate": 0.00040702051571616787,
      "loss": 4.2813,
      "step": 4940
    },
    {
      "epoch": 0.9316770186335404,
      "grad_norm": 1.8623604774475098,
      "learning_rate": 0.00040683229813664594,
      "loss": 4.0378,
      "step": 4950
    },
    {
      "epoch": 0.9335591944287597,
      "grad_norm": 2.075603723526001,
      "learning_rate": 0.000406644080557124,
      "loss": 3.8749,
      "step": 4960
    },
    {
      "epoch": 0.9354413702239789,
      "grad_norm": 1.5741757154464722,
      "learning_rate": 0.0004064558629776021,
      "loss": 3.8205,
      "step": 4970
    },
    {
      "epoch": 0.9373235460191982,
      "grad_norm": 1.8074135780334473,
      "learning_rate": 0.0004062676453980802,
      "loss": 3.7819,
      "step": 4980
    },
    {
      "epoch": 0.9392057218144174,
      "grad_norm": 2.2221150398254395,
      "learning_rate": 0.0004060794278185583,
      "loss": 4.1042,
      "step": 4990
    },
    {
      "epoch": 0.9410878976096367,
      "grad_norm": 1.7813559770584106,
      "learning_rate": 0.0004058912102390364,
      "loss": 3.7184,
      "step": 5000
    },
    {
      "epoch": 0.9429700734048561,
      "grad_norm": 2.023097515106201,
      "learning_rate": 0.0004057029926595144,
      "loss": 3.9837,
      "step": 5010
    },
    {
      "epoch": 0.9448522492000753,
      "grad_norm": 1.661009430885315,
      "learning_rate": 0.0004055147750799925,
      "loss": 4.0824,
      "step": 5020
    },
    {
      "epoch": 0.9467344249952946,
      "grad_norm": 1.5688236951828003,
      "learning_rate": 0.00040532655750047055,
      "loss": 3.9856,
      "step": 5030
    },
    {
      "epoch": 0.9486166007905138,
      "grad_norm": 2.1541364192962646,
      "learning_rate": 0.00040513833992094863,
      "loss": 3.7916,
      "step": 5040
    },
    {
      "epoch": 0.9504987765857331,
      "grad_norm": 1.6808667182922363,
      "learning_rate": 0.0004049501223414267,
      "loss": 3.9749,
      "step": 5050
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 1.978331446647644,
      "learning_rate": 0.0004047619047619048,
      "loss": 4.1186,
      "step": 5060
    },
    {
      "epoch": 0.9542631281761716,
      "grad_norm": 1.737565279006958,
      "learning_rate": 0.00040457368718238286,
      "loss": 3.8198,
      "step": 5070
    },
    {
      "epoch": 0.956145303971391,
      "grad_norm": 1.7728204727172852,
      "learning_rate": 0.00040438546960286094,
      "loss": 3.8926,
      "step": 5080
    },
    {
      "epoch": 0.9580274797666102,
      "grad_norm": 1.7337509393692017,
      "learning_rate": 0.000404197252023339,
      "loss": 3.8826,
      "step": 5090
    },
    {
      "epoch": 0.9599096555618295,
      "grad_norm": 1.9217840433120728,
      "learning_rate": 0.00040400903444381703,
      "loss": 3.7385,
      "step": 5100
    },
    {
      "epoch": 0.9617918313570487,
      "grad_norm": 1.8276395797729492,
      "learning_rate": 0.0004038208168642951,
      "loss": 3.9753,
      "step": 5110
    },
    {
      "epoch": 0.963674007152268,
      "grad_norm": 1.7400699853897095,
      "learning_rate": 0.0004036325992847732,
      "loss": 4.0928,
      "step": 5120
    },
    {
      "epoch": 0.9655561829474872,
      "grad_norm": 1.511858582496643,
      "learning_rate": 0.00040344438170525126,
      "loss": 3.9909,
      "step": 5130
    },
    {
      "epoch": 0.9674383587427066,
      "grad_norm": 1.6028978824615479,
      "learning_rate": 0.00040325616412572934,
      "loss": 3.732,
      "step": 5140
    },
    {
      "epoch": 0.9693205345379259,
      "grad_norm": 2.316171169281006,
      "learning_rate": 0.0004030679465462074,
      "loss": 3.9026,
      "step": 5150
    },
    {
      "epoch": 0.9712027103331451,
      "grad_norm": 1.5085681676864624,
      "learning_rate": 0.0004028797289666855,
      "loss": 4.1728,
      "step": 5160
    },
    {
      "epoch": 0.9730848861283644,
      "grad_norm": 1.9052393436431885,
      "learning_rate": 0.00040269151138716357,
      "loss": 4.0342,
      "step": 5170
    },
    {
      "epoch": 0.9749670619235836,
      "grad_norm": 1.9131525754928589,
      "learning_rate": 0.00040250329380764165,
      "loss": 4.0431,
      "step": 5180
    },
    {
      "epoch": 0.9768492377188029,
      "grad_norm": 1.4925379753112793,
      "learning_rate": 0.00040231507622811967,
      "loss": 3.9273,
      "step": 5190
    },
    {
      "epoch": 0.9787314135140223,
      "grad_norm": 1.9738259315490723,
      "learning_rate": 0.00040212685864859774,
      "loss": 3.8051,
      "step": 5200
    },
    {
      "epoch": 0.9806135893092415,
      "grad_norm": 1.8526817560195923,
      "learning_rate": 0.0004019386410690759,
      "loss": 3.9126,
      "step": 5210
    },
    {
      "epoch": 0.9824957651044608,
      "grad_norm": 1.584219217300415,
      "learning_rate": 0.00040175042348955395,
      "loss": 3.9394,
      "step": 5220
    },
    {
      "epoch": 0.98437794089968,
      "grad_norm": 1.7924845218658447,
      "learning_rate": 0.00040156220591003203,
      "loss": 3.95,
      "step": 5230
    },
    {
      "epoch": 0.9862601166948993,
      "grad_norm": 1.6530725955963135,
      "learning_rate": 0.0004013739883305101,
      "loss": 3.8088,
      "step": 5240
    },
    {
      "epoch": 0.9881422924901185,
      "grad_norm": 2.2157840728759766,
      "learning_rate": 0.0004011857707509882,
      "loss": 3.9543,
      "step": 5250
    },
    {
      "epoch": 0.9900244682853379,
      "grad_norm": 1.9279638528823853,
      "learning_rate": 0.00040099755317146626,
      "loss": 3.6932,
      "step": 5260
    },
    {
      "epoch": 0.9919066440805572,
      "grad_norm": 1.7662793397903442,
      "learning_rate": 0.0004008093355919443,
      "loss": 3.8718,
      "step": 5270
    },
    {
      "epoch": 0.9937888198757764,
      "grad_norm": 1.8058240413665771,
      "learning_rate": 0.00040062111801242236,
      "loss": 4.0627,
      "step": 5280
    },
    {
      "epoch": 0.9956709956709957,
      "grad_norm": 1.6732690334320068,
      "learning_rate": 0.00040043290043290043,
      "loss": 3.9005,
      "step": 5290
    },
    {
      "epoch": 0.9975531714662149,
      "grad_norm": 1.7442728281021118,
      "learning_rate": 0.0004002446828533785,
      "loss": 3.8012,
      "step": 5300
    },
    {
      "epoch": 0.9994353472614342,
      "grad_norm": 1.855129599571228,
      "learning_rate": 0.0004000564652738566,
      "loss": 3.7207,
      "step": 5310
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.09853333333333333,
      "eval_loss": 3.863232374191284,
      "eval_runtime": 117.7167,
      "eval_samples_per_second": 63.712,
      "eval_steps_per_second": 7.968,
      "step": 5313
    },
    {
      "epoch": 1.0013175230566536,
      "grad_norm": 1.724755883216858,
      "learning_rate": 0.00039986824769433466,
      "loss": 3.9155,
      "step": 5320
    },
    {
      "epoch": 1.0031996988518728,
      "grad_norm": 1.7752708196640015,
      "learning_rate": 0.00039968003011481274,
      "loss": 3.9399,
      "step": 5330
    },
    {
      "epoch": 1.005081874647092,
      "grad_norm": 1.7419073581695557,
      "learning_rate": 0.0003994918125352908,
      "loss": 3.866,
      "step": 5340
    },
    {
      "epoch": 1.0069640504423114,
      "grad_norm": 1.829726219177246,
      "learning_rate": 0.0003993035949557689,
      "loss": 3.7939,
      "step": 5350
    },
    {
      "epoch": 1.0088462262375306,
      "grad_norm": 2.0104973316192627,
      "learning_rate": 0.0003991153773762469,
      "loss": 4.0264,
      "step": 5360
    },
    {
      "epoch": 1.0107284020327498,
      "grad_norm": 1.9856433868408203,
      "learning_rate": 0.000398927159796725,
      "loss": 3.9405,
      "step": 5370
    },
    {
      "epoch": 1.012610577827969,
      "grad_norm": 1.782480001449585,
      "learning_rate": 0.00039873894221720307,
      "loss": 3.779,
      "step": 5380
    },
    {
      "epoch": 1.0144927536231885,
      "grad_norm": 1.956235408782959,
      "learning_rate": 0.00039855072463768114,
      "loss": 3.7016,
      "step": 5390
    },
    {
      "epoch": 1.0163749294184077,
      "grad_norm": 1.7101922035217285,
      "learning_rate": 0.0003983625070581592,
      "loss": 3.974,
      "step": 5400
    },
    {
      "epoch": 1.018257105213627,
      "grad_norm": 1.6829845905303955,
      "learning_rate": 0.0003981742894786373,
      "loss": 3.9825,
      "step": 5410
    },
    {
      "epoch": 1.0201392810088463,
      "grad_norm": 1.9459203481674194,
      "learning_rate": 0.00039798607189911537,
      "loss": 3.8148,
      "step": 5420
    },
    {
      "epoch": 1.0220214568040655,
      "grad_norm": 1.8005633354187012,
      "learning_rate": 0.0003977978543195935,
      "loss": 4.0029,
      "step": 5430
    },
    {
      "epoch": 1.0239036325992847,
      "grad_norm": 1.9317513704299927,
      "learning_rate": 0.0003976096367400716,
      "loss": 3.8425,
      "step": 5440
    },
    {
      "epoch": 1.025785808394504,
      "grad_norm": 1.8140696287155151,
      "learning_rate": 0.0003974214191605496,
      "loss": 4.0525,
      "step": 5450
    },
    {
      "epoch": 1.0276679841897234,
      "grad_norm": 1.8898414373397827,
      "learning_rate": 0.0003972332015810277,
      "loss": 3.8046,
      "step": 5460
    },
    {
      "epoch": 1.0295501599849426,
      "grad_norm": 1.8269977569580078,
      "learning_rate": 0.00039704498400150575,
      "loss": 3.9079,
      "step": 5470
    },
    {
      "epoch": 1.0314323357801618,
      "grad_norm": 1.6486467123031616,
      "learning_rate": 0.00039685676642198383,
      "loss": 3.7988,
      "step": 5480
    },
    {
      "epoch": 1.0333145115753812,
      "grad_norm": 1.916007399559021,
      "learning_rate": 0.0003966685488424619,
      "loss": 3.9587,
      "step": 5490
    },
    {
      "epoch": 1.0351966873706004,
      "grad_norm": 1.9689396619796753,
      "learning_rate": 0.00039648033126294,
      "loss": 3.7973,
      "step": 5500
    },
    {
      "epoch": 1.0370788631658197,
      "grad_norm": 2.0306029319763184,
      "learning_rate": 0.00039629211368341806,
      "loss": 3.6851,
      "step": 5510
    },
    {
      "epoch": 1.0389610389610389,
      "grad_norm": 2.138746976852417,
      "learning_rate": 0.00039610389610389614,
      "loss": 3.7968,
      "step": 5520
    },
    {
      "epoch": 1.0408432147562583,
      "grad_norm": 2.021477699279785,
      "learning_rate": 0.0003959156785243742,
      "loss": 3.9782,
      "step": 5530
    },
    {
      "epoch": 1.0427253905514775,
      "grad_norm": 1.9355220794677734,
      "learning_rate": 0.00039572746094485223,
      "loss": 3.5401,
      "step": 5540
    },
    {
      "epoch": 1.0446075663466967,
      "grad_norm": 2.985809087753296,
      "learning_rate": 0.0003955392433653303,
      "loss": 3.8579,
      "step": 5550
    },
    {
      "epoch": 1.0464897421419161,
      "grad_norm": 2.070136070251465,
      "learning_rate": 0.0003953510257858084,
      "loss": 3.81,
      "step": 5560
    },
    {
      "epoch": 1.0483719179371354,
      "grad_norm": 2.088350772857666,
      "learning_rate": 0.00039516280820628646,
      "loss": 3.9072,
      "step": 5570
    },
    {
      "epoch": 1.0502540937323546,
      "grad_norm": 1.9731718301773071,
      "learning_rate": 0.00039497459062676454,
      "loss": 3.7835,
      "step": 5580
    },
    {
      "epoch": 1.0521362695275738,
      "grad_norm": 1.894665002822876,
      "learning_rate": 0.0003947863730472426,
      "loss": 3.7677,
      "step": 5590
    },
    {
      "epoch": 1.0540184453227932,
      "grad_norm": 2.145423650741577,
      "learning_rate": 0.0003945981554677207,
      "loss": 3.7991,
      "step": 5600
    },
    {
      "epoch": 1.0559006211180124,
      "grad_norm": 1.9230924844741821,
      "learning_rate": 0.00039440993788819877,
      "loss": 4.0038,
      "step": 5610
    },
    {
      "epoch": 1.0577827969132316,
      "grad_norm": 1.7206995487213135,
      "learning_rate": 0.0003942217203086768,
      "loss": 3.8487,
      "step": 5620
    },
    {
      "epoch": 1.059664972708451,
      "grad_norm": 1.8569245338439941,
      "learning_rate": 0.00039403350272915487,
      "loss": 3.8765,
      "step": 5630
    },
    {
      "epoch": 1.0615471485036703,
      "grad_norm": 2.204007148742676,
      "learning_rate": 0.00039384528514963294,
      "loss": 4.0123,
      "step": 5640
    },
    {
      "epoch": 1.0634293242988895,
      "grad_norm": 1.744128942489624,
      "learning_rate": 0.0003936570675701111,
      "loss": 4.1011,
      "step": 5650
    },
    {
      "epoch": 1.0653115000941087,
      "grad_norm": 2.6771187782287598,
      "learning_rate": 0.00039346884999058915,
      "loss": 3.8366,
      "step": 5660
    },
    {
      "epoch": 1.0671936758893281,
      "grad_norm": 2.1822588443756104,
      "learning_rate": 0.00039328063241106723,
      "loss": 3.9191,
      "step": 5670
    },
    {
      "epoch": 1.0690758516845473,
      "grad_norm": 1.7872580289840698,
      "learning_rate": 0.0003930924148315453,
      "loss": 3.7534,
      "step": 5680
    },
    {
      "epoch": 1.0709580274797665,
      "grad_norm": 1.9511586427688599,
      "learning_rate": 0.0003929041972520234,
      "loss": 3.6996,
      "step": 5690
    },
    {
      "epoch": 1.072840203274986,
      "grad_norm": 1.9602952003479004,
      "learning_rate": 0.00039271597967250146,
      "loss": 4.0202,
      "step": 5700
    },
    {
      "epoch": 1.0747223790702052,
      "grad_norm": 2.0772409439086914,
      "learning_rate": 0.0003925277620929795,
      "loss": 3.9165,
      "step": 5710
    },
    {
      "epoch": 1.0766045548654244,
      "grad_norm": 1.903748869895935,
      "learning_rate": 0.00039233954451345756,
      "loss": 3.6467,
      "step": 5720
    },
    {
      "epoch": 1.0784867306606438,
      "grad_norm": 1.9211351871490479,
      "learning_rate": 0.00039215132693393563,
      "loss": 3.8662,
      "step": 5730
    },
    {
      "epoch": 1.080368906455863,
      "grad_norm": 2.4941396713256836,
      "learning_rate": 0.0003919631093544137,
      "loss": 3.7316,
      "step": 5740
    },
    {
      "epoch": 1.0822510822510822,
      "grad_norm": 1.5563691854476929,
      "learning_rate": 0.0003917748917748918,
      "loss": 3.9315,
      "step": 5750
    },
    {
      "epoch": 1.0841332580463015,
      "grad_norm": 1.6314982175827026,
      "learning_rate": 0.00039158667419536986,
      "loss": 4.1015,
      "step": 5760
    },
    {
      "epoch": 1.0860154338415209,
      "grad_norm": 1.7421598434448242,
      "learning_rate": 0.00039139845661584794,
      "loss": 3.5859,
      "step": 5770
    },
    {
      "epoch": 1.08789760963674,
      "grad_norm": 1.7598062753677368,
      "learning_rate": 0.000391210239036326,
      "loss": 4.0092,
      "step": 5780
    },
    {
      "epoch": 1.0897797854319593,
      "grad_norm": 1.8014366626739502,
      "learning_rate": 0.0003910220214568041,
      "loss": 3.9926,
      "step": 5790
    },
    {
      "epoch": 1.0916619612271785,
      "grad_norm": 1.5961711406707764,
      "learning_rate": 0.0003908338038772821,
      "loss": 3.8729,
      "step": 5800
    },
    {
      "epoch": 1.093544137022398,
      "grad_norm": 1.7829089164733887,
      "learning_rate": 0.0003906455862977602,
      "loss": 3.7447,
      "step": 5810
    },
    {
      "epoch": 1.0954263128176172,
      "grad_norm": 2.3133766651153564,
      "learning_rate": 0.00039045736871823827,
      "loss": 3.8943,
      "step": 5820
    },
    {
      "epoch": 1.0973084886128364,
      "grad_norm": 1.9624972343444824,
      "learning_rate": 0.00039026915113871634,
      "loss": 4.1605,
      "step": 5830
    },
    {
      "epoch": 1.0991906644080558,
      "grad_norm": 1.9229626655578613,
      "learning_rate": 0.0003900809335591944,
      "loss": 3.8495,
      "step": 5840
    },
    {
      "epoch": 1.101072840203275,
      "grad_norm": 2.002578020095825,
      "learning_rate": 0.0003898927159796725,
      "loss": 3.8138,
      "step": 5850
    },
    {
      "epoch": 1.1029550159984942,
      "grad_norm": 1.9529259204864502,
      "learning_rate": 0.00038970449840015057,
      "loss": 3.9551,
      "step": 5860
    },
    {
      "epoch": 1.1048371917937136,
      "grad_norm": 1.998673677444458,
      "learning_rate": 0.0003895162808206287,
      "loss": 3.8178,
      "step": 5870
    },
    {
      "epoch": 1.1067193675889329,
      "grad_norm": 2.191293478012085,
      "learning_rate": 0.0003893280632411067,
      "loss": 3.5839,
      "step": 5880
    },
    {
      "epoch": 1.108601543384152,
      "grad_norm": 1.7941858768463135,
      "learning_rate": 0.0003891398456615848,
      "loss": 3.9038,
      "step": 5890
    },
    {
      "epoch": 1.1104837191793713,
      "grad_norm": 1.7840150594711304,
      "learning_rate": 0.0003889516280820629,
      "loss": 3.9601,
      "step": 5900
    },
    {
      "epoch": 1.1123658949745907,
      "grad_norm": 1.9555600881576538,
      "learning_rate": 0.00038876341050254095,
      "loss": 3.6967,
      "step": 5910
    },
    {
      "epoch": 1.11424807076981,
      "grad_norm": 1.902140736579895,
      "learning_rate": 0.00038857519292301903,
      "loss": 3.8388,
      "step": 5920
    },
    {
      "epoch": 1.1161302465650291,
      "grad_norm": 1.6516783237457275,
      "learning_rate": 0.0003883869753434971,
      "loss": 3.937,
      "step": 5930
    },
    {
      "epoch": 1.1180124223602483,
      "grad_norm": 1.862953543663025,
      "learning_rate": 0.0003881987577639752,
      "loss": 3.9593,
      "step": 5940
    },
    {
      "epoch": 1.1198945981554678,
      "grad_norm": 2.068586587905884,
      "learning_rate": 0.00038801054018445326,
      "loss": 3.8958,
      "step": 5950
    },
    {
      "epoch": 1.121776773950687,
      "grad_norm": 1.7183383703231812,
      "learning_rate": 0.00038782232260493134,
      "loss": 4.0018,
      "step": 5960
    },
    {
      "epoch": 1.1236589497459062,
      "grad_norm": 1.8313102722167969,
      "learning_rate": 0.00038763410502540936,
      "loss": 3.9836,
      "step": 5970
    },
    {
      "epoch": 1.1255411255411256,
      "grad_norm": 1.7724077701568604,
      "learning_rate": 0.00038744588744588743,
      "loss": 3.7711,
      "step": 5980
    },
    {
      "epoch": 1.1274233013363448,
      "grad_norm": 2.0144224166870117,
      "learning_rate": 0.0003872576698663655,
      "loss": 3.7638,
      "step": 5990
    },
    {
      "epoch": 1.129305477131564,
      "grad_norm": 2.1548473834991455,
      "learning_rate": 0.0003870694522868436,
      "loss": 3.8382,
      "step": 6000
    },
    {
      "epoch": 1.1311876529267835,
      "grad_norm": 1.721322774887085,
      "learning_rate": 0.00038688123470732166,
      "loss": 4.0194,
      "step": 6010
    },
    {
      "epoch": 1.1330698287220027,
      "grad_norm": 2.17598295211792,
      "learning_rate": 0.00038669301712779974,
      "loss": 3.7419,
      "step": 6020
    },
    {
      "epoch": 1.134952004517222,
      "grad_norm": 2.023616075515747,
      "learning_rate": 0.0003865047995482778,
      "loss": 3.6595,
      "step": 6030
    },
    {
      "epoch": 1.136834180312441,
      "grad_norm": 2.1189849376678467,
      "learning_rate": 0.0003863165819687559,
      "loss": 3.7727,
      "step": 6040
    },
    {
      "epoch": 1.1387163561076605,
      "grad_norm": 1.9913543462753296,
      "learning_rate": 0.00038612836438923397,
      "loss": 3.7929,
      "step": 6050
    },
    {
      "epoch": 1.1405985319028797,
      "grad_norm": 1.8483470678329468,
      "learning_rate": 0.000385940146809712,
      "loss": 3.7482,
      "step": 6060
    },
    {
      "epoch": 1.142480707698099,
      "grad_norm": 2.0956037044525146,
      "learning_rate": 0.00038575192923019007,
      "loss": 3.8818,
      "step": 6070
    },
    {
      "epoch": 1.1443628834933182,
      "grad_norm": 2.2748496532440186,
      "learning_rate": 0.00038556371165066814,
      "loss": 3.8901,
      "step": 6080
    },
    {
      "epoch": 1.1462450592885376,
      "grad_norm": 1.7614413499832153,
      "learning_rate": 0.0003853754940711463,
      "loss": 3.7777,
      "step": 6090
    },
    {
      "epoch": 1.1481272350837568,
      "grad_norm": 1.8067692518234253,
      "learning_rate": 0.00038518727649162435,
      "loss": 3.8457,
      "step": 6100
    },
    {
      "epoch": 1.150009410878976,
      "grad_norm": 2.009246587753296,
      "learning_rate": 0.00038499905891210243,
      "loss": 3.8384,
      "step": 6110
    },
    {
      "epoch": 1.1518915866741954,
      "grad_norm": 1.9846868515014648,
      "learning_rate": 0.0003848108413325805,
      "loss": 3.9047,
      "step": 6120
    },
    {
      "epoch": 1.1537737624694147,
      "grad_norm": 1.5468846559524536,
      "learning_rate": 0.0003846226237530586,
      "loss": 3.7324,
      "step": 6130
    },
    {
      "epoch": 1.1556559382646339,
      "grad_norm": 1.7952297925949097,
      "learning_rate": 0.00038443440617353666,
      "loss": 3.6679,
      "step": 6140
    },
    {
      "epoch": 1.1575381140598533,
      "grad_norm": 1.754523754119873,
      "learning_rate": 0.0003842461885940147,
      "loss": 3.8266,
      "step": 6150
    },
    {
      "epoch": 1.1594202898550725,
      "grad_norm": 1.77576744556427,
      "learning_rate": 0.00038405797101449276,
      "loss": 3.7629,
      "step": 6160
    },
    {
      "epoch": 1.1613024656502917,
      "grad_norm": 1.9010967016220093,
      "learning_rate": 0.00038386975343497083,
      "loss": 3.8496,
      "step": 6170
    },
    {
      "epoch": 1.163184641445511,
      "grad_norm": 1.6943968534469604,
      "learning_rate": 0.0003836815358554489,
      "loss": 3.7759,
      "step": 6180
    },
    {
      "epoch": 1.1650668172407304,
      "grad_norm": 2.6153507232666016,
      "learning_rate": 0.000383493318275927,
      "loss": 3.6857,
      "step": 6190
    },
    {
      "epoch": 1.1669489930359496,
      "grad_norm": 1.597812533378601,
      "learning_rate": 0.00038330510069640506,
      "loss": 3.8673,
      "step": 6200
    },
    {
      "epoch": 1.1688311688311688,
      "grad_norm": 1.7507632970809937,
      "learning_rate": 0.00038311688311688314,
      "loss": 3.9767,
      "step": 6210
    },
    {
      "epoch": 1.170713344626388,
      "grad_norm": 1.8110359907150269,
      "learning_rate": 0.0003829286655373612,
      "loss": 3.5311,
      "step": 6220
    },
    {
      "epoch": 1.1725955204216074,
      "grad_norm": 1.7214828729629517,
      "learning_rate": 0.00038274044795783924,
      "loss": 3.7362,
      "step": 6230
    },
    {
      "epoch": 1.1744776962168266,
      "grad_norm": 1.9230563640594482,
      "learning_rate": 0.0003825522303783173,
      "loss": 3.9467,
      "step": 6240
    },
    {
      "epoch": 1.1763598720120458,
      "grad_norm": 1.797293782234192,
      "learning_rate": 0.0003823640127987954,
      "loss": 3.9876,
      "step": 6250
    },
    {
      "epoch": 1.1782420478072653,
      "grad_norm": 1.9491268396377563,
      "learning_rate": 0.00038217579521927347,
      "loss": 3.7447,
      "step": 6260
    },
    {
      "epoch": 1.1801242236024845,
      "grad_norm": 1.9252543449401855,
      "learning_rate": 0.00038198757763975154,
      "loss": 3.9862,
      "step": 6270
    },
    {
      "epoch": 1.1820063993977037,
      "grad_norm": 2.3227150440216064,
      "learning_rate": 0.0003817993600602296,
      "loss": 3.9708,
      "step": 6280
    },
    {
      "epoch": 1.1838885751929231,
      "grad_norm": 1.932223916053772,
      "learning_rate": 0.0003816111424807077,
      "loss": 3.6916,
      "step": 6290
    },
    {
      "epoch": 1.1857707509881423,
      "grad_norm": 2.2664384841918945,
      "learning_rate": 0.00038142292490118577,
      "loss": 3.6087,
      "step": 6300
    },
    {
      "epoch": 1.1876529267833615,
      "grad_norm": 1.9968208074569702,
      "learning_rate": 0.0003812347073216639,
      "loss": 3.8954,
      "step": 6310
    },
    {
      "epoch": 1.1895351025785807,
      "grad_norm": 2.0324888229370117,
      "learning_rate": 0.0003810464897421419,
      "loss": 3.9083,
      "step": 6320
    },
    {
      "epoch": 1.1914172783738002,
      "grad_norm": 1.6557317972183228,
      "learning_rate": 0.00038085827216262,
      "loss": 3.7426,
      "step": 6330
    },
    {
      "epoch": 1.1932994541690194,
      "grad_norm": 1.7409571409225464,
      "learning_rate": 0.0003806700545830981,
      "loss": 3.7642,
      "step": 6340
    },
    {
      "epoch": 1.1951816299642386,
      "grad_norm": 1.8588122129440308,
      "learning_rate": 0.00038048183700357615,
      "loss": 3.893,
      "step": 6350
    },
    {
      "epoch": 1.1970638057594578,
      "grad_norm": 1.8750513792037964,
      "learning_rate": 0.00038029361942405423,
      "loss": 3.8427,
      "step": 6360
    },
    {
      "epoch": 1.1989459815546772,
      "grad_norm": 2.1529922485351562,
      "learning_rate": 0.0003801054018445323,
      "loss": 3.8215,
      "step": 6370
    },
    {
      "epoch": 1.2008281573498965,
      "grad_norm": 4.190012454986572,
      "learning_rate": 0.0003799171842650104,
      "loss": 3.9574,
      "step": 6380
    },
    {
      "epoch": 1.2027103331451157,
      "grad_norm": 1.9367660284042358,
      "learning_rate": 0.00037972896668548846,
      "loss": 3.8696,
      "step": 6390
    },
    {
      "epoch": 1.204592508940335,
      "grad_norm": 1.7562202215194702,
      "learning_rate": 0.00037954074910596654,
      "loss": 3.6758,
      "step": 6400
    },
    {
      "epoch": 1.2064746847355543,
      "grad_norm": 1.964674472808838,
      "learning_rate": 0.00037935253152644456,
      "loss": 3.7275,
      "step": 6410
    },
    {
      "epoch": 1.2083568605307735,
      "grad_norm": 5.957378387451172,
      "learning_rate": 0.00037916431394692263,
      "loss": 3.8423,
      "step": 6420
    },
    {
      "epoch": 1.210239036325993,
      "grad_norm": 2.2306909561157227,
      "learning_rate": 0.0003789760963674007,
      "loss": 3.8344,
      "step": 6430
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 1.6602094173431396,
      "learning_rate": 0.0003787878787878788,
      "loss": 3.7842,
      "step": 6440
    },
    {
      "epoch": 1.2140033879164314,
      "grad_norm": 1.708493947982788,
      "learning_rate": 0.00037859966120835686,
      "loss": 3.9448,
      "step": 6450
    },
    {
      "epoch": 1.2158855637116506,
      "grad_norm": 1.8723329305648804,
      "learning_rate": 0.00037841144362883494,
      "loss": 3.4864,
      "step": 6460
    },
    {
      "epoch": 1.21776773950687,
      "grad_norm": 1.9271537065505981,
      "learning_rate": 0.000378223226049313,
      "loss": 3.9397,
      "step": 6470
    },
    {
      "epoch": 1.2196499153020892,
      "grad_norm": 2.0064141750335693,
      "learning_rate": 0.0003780350084697911,
      "loss": 3.8913,
      "step": 6480
    },
    {
      "epoch": 1.2215320910973084,
      "grad_norm": 2.434368848800659,
      "learning_rate": 0.0003778467908902691,
      "loss": 3.6471,
      "step": 6490
    },
    {
      "epoch": 1.2234142668925279,
      "grad_norm": 1.9352556467056274,
      "learning_rate": 0.0003776585733107472,
      "loss": 3.8066,
      "step": 6500
    },
    {
      "epoch": 1.225296442687747,
      "grad_norm": 1.7813034057617188,
      "learning_rate": 0.00037747035573122527,
      "loss": 3.8825,
      "step": 6510
    },
    {
      "epoch": 1.2271786184829663,
      "grad_norm": 3.184835433959961,
      "learning_rate": 0.00037728213815170334,
      "loss": 3.9238,
      "step": 6520
    },
    {
      "epoch": 1.2290607942781855,
      "grad_norm": 1.8562132120132446,
      "learning_rate": 0.0003770939205721814,
      "loss": 3.7629,
      "step": 6530
    },
    {
      "epoch": 1.230942970073405,
      "grad_norm": 2.0398480892181396,
      "learning_rate": 0.00037690570299265955,
      "loss": 3.821,
      "step": 6540
    },
    {
      "epoch": 1.2328251458686241,
      "grad_norm": 2.021771192550659,
      "learning_rate": 0.00037671748541313763,
      "loss": 3.6642,
      "step": 6550
    },
    {
      "epoch": 1.2347073216638433,
      "grad_norm": 2.077044725418091,
      "learning_rate": 0.0003765292678336157,
      "loss": 3.6619,
      "step": 6560
    },
    {
      "epoch": 1.2365894974590628,
      "grad_norm": 2.127864360809326,
      "learning_rate": 0.0003763410502540938,
      "loss": 3.8543,
      "step": 6570
    },
    {
      "epoch": 1.238471673254282,
      "grad_norm": 2.043787717819214,
      "learning_rate": 0.0003761528326745718,
      "loss": 3.9452,
      "step": 6580
    },
    {
      "epoch": 1.2403538490495012,
      "grad_norm": 1.8492457866668701,
      "learning_rate": 0.0003759646150950499,
      "loss": 3.8527,
      "step": 6590
    },
    {
      "epoch": 1.2422360248447206,
      "grad_norm": 1.6727014780044556,
      "learning_rate": 0.00037577639751552796,
      "loss": 3.5721,
      "step": 6600
    },
    {
      "epoch": 1.2441182006399398,
      "grad_norm": 1.6961781978607178,
      "learning_rate": 0.00037558817993600603,
      "loss": 3.6185,
      "step": 6610
    },
    {
      "epoch": 1.246000376435159,
      "grad_norm": 2.209446430206299,
      "learning_rate": 0.0003753999623564841,
      "loss": 3.9786,
      "step": 6620
    },
    {
      "epoch": 1.2478825522303783,
      "grad_norm": 1.7225549221038818,
      "learning_rate": 0.0003752117447769622,
      "loss": 3.6741,
      "step": 6630
    },
    {
      "epoch": 1.2497647280255977,
      "grad_norm": 1.9335850477218628,
      "learning_rate": 0.00037502352719744026,
      "loss": 3.6865,
      "step": 6640
    },
    {
      "epoch": 1.251646903820817,
      "grad_norm": 2.207566738128662,
      "learning_rate": 0.00037483530961791834,
      "loss": 3.7731,
      "step": 6650
    },
    {
      "epoch": 1.253529079616036,
      "grad_norm": 2.0488553047180176,
      "learning_rate": 0.0003746470920383964,
      "loss": 3.7351,
      "step": 6660
    },
    {
      "epoch": 1.2554112554112553,
      "grad_norm": 2.094182014465332,
      "learning_rate": 0.00037445887445887444,
      "loss": 3.7246,
      "step": 6670
    },
    {
      "epoch": 1.2572934312064747,
      "grad_norm": 1.9110198020935059,
      "learning_rate": 0.0003742706568793525,
      "loss": 3.7926,
      "step": 6680
    },
    {
      "epoch": 1.259175607001694,
      "grad_norm": 1.949934959411621,
      "learning_rate": 0.0003740824392998306,
      "loss": 3.5716,
      "step": 6690
    },
    {
      "epoch": 1.2610577827969132,
      "grad_norm": 1.7895861864089966,
      "learning_rate": 0.00037389422172030867,
      "loss": 3.8369,
      "step": 6700
    },
    {
      "epoch": 1.2629399585921326,
      "grad_norm": 1.8408478498458862,
      "learning_rate": 0.00037370600414078674,
      "loss": 3.8636,
      "step": 6710
    },
    {
      "epoch": 1.2648221343873518,
      "grad_norm": 2.2939276695251465,
      "learning_rate": 0.0003735177865612648,
      "loss": 4.2916,
      "step": 6720
    },
    {
      "epoch": 1.266704310182571,
      "grad_norm": 2.4173057079315186,
      "learning_rate": 0.0003733295689817429,
      "loss": 3.7414,
      "step": 6730
    },
    {
      "epoch": 1.2685864859777904,
      "grad_norm": 3.014364719390869,
      "learning_rate": 0.00037314135140222097,
      "loss": 3.8336,
      "step": 6740
    },
    {
      "epoch": 1.2704686617730097,
      "grad_norm": 1.7591136693954468,
      "learning_rate": 0.0003729531338226991,
      "loss": 3.8213,
      "step": 6750
    },
    {
      "epoch": 1.2723508375682289,
      "grad_norm": 1.7625524997711182,
      "learning_rate": 0.0003727649162431771,
      "loss": 3.8489,
      "step": 6760
    },
    {
      "epoch": 1.274233013363448,
      "grad_norm": 1.863606333732605,
      "learning_rate": 0.0003725766986636552,
      "loss": 3.7942,
      "step": 6770
    },
    {
      "epoch": 1.2761151891586673,
      "grad_norm": 2.5267436504364014,
      "learning_rate": 0.0003723884810841333,
      "loss": 3.7174,
      "step": 6780
    },
    {
      "epoch": 1.2779973649538867,
      "grad_norm": 2.354560136795044,
      "learning_rate": 0.00037220026350461135,
      "loss": 3.7846,
      "step": 6790
    },
    {
      "epoch": 1.279879540749106,
      "grad_norm": 2.1386847496032715,
      "learning_rate": 0.00037201204592508943,
      "loss": 3.9158,
      "step": 6800
    },
    {
      "epoch": 1.2817617165443251,
      "grad_norm": 2.3380281925201416,
      "learning_rate": 0.0003718238283455675,
      "loss": 3.5337,
      "step": 6810
    },
    {
      "epoch": 1.2836438923395446,
      "grad_norm": 2.3061459064483643,
      "learning_rate": 0.0003716356107660456,
      "loss": 4.0088,
      "step": 6820
    },
    {
      "epoch": 1.2855260681347638,
      "grad_norm": 1.9884824752807617,
      "learning_rate": 0.00037144739318652366,
      "loss": 3.9162,
      "step": 6830
    },
    {
      "epoch": 1.287408243929983,
      "grad_norm": 1.793647050857544,
      "learning_rate": 0.0003712591756070017,
      "loss": 3.7617,
      "step": 6840
    },
    {
      "epoch": 1.2892904197252024,
      "grad_norm": 2.4594411849975586,
      "learning_rate": 0.00037107095802747976,
      "loss": 3.8177,
      "step": 6850
    },
    {
      "epoch": 1.2911725955204216,
      "grad_norm": 2.076636791229248,
      "learning_rate": 0.00037088274044795783,
      "loss": 4.0619,
      "step": 6860
    },
    {
      "epoch": 1.2930547713156408,
      "grad_norm": 1.7677685022354126,
      "learning_rate": 0.0003706945228684359,
      "loss": 3.8239,
      "step": 6870
    },
    {
      "epoch": 1.2949369471108603,
      "grad_norm": 1.7307604551315308,
      "learning_rate": 0.000370506305288914,
      "loss": 3.6623,
      "step": 6880
    },
    {
      "epoch": 1.2968191229060795,
      "grad_norm": 1.8652023077011108,
      "learning_rate": 0.00037031808770939206,
      "loss": 3.9653,
      "step": 6890
    },
    {
      "epoch": 1.2987012987012987,
      "grad_norm": 1.8615070581436157,
      "learning_rate": 0.00037012987012987014,
      "loss": 3.7819,
      "step": 6900
    },
    {
      "epoch": 1.300583474496518,
      "grad_norm": 2.0547096729278564,
      "learning_rate": 0.0003699416525503482,
      "loss": 3.9213,
      "step": 6910
    },
    {
      "epoch": 1.3024656502917373,
      "grad_norm": 1.9442291259765625,
      "learning_rate": 0.0003697534349708263,
      "loss": 3.8774,
      "step": 6920
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 1.550815463066101,
      "learning_rate": 0.0003695652173913043,
      "loss": 3.7824,
      "step": 6930
    },
    {
      "epoch": 1.3062300018821758,
      "grad_norm": 1.9164446592330933,
      "learning_rate": 0.0003693769998117824,
      "loss": 3.8274,
      "step": 6940
    },
    {
      "epoch": 1.308112177677395,
      "grad_norm": 2.2354953289031982,
      "learning_rate": 0.00036918878223226047,
      "loss": 3.7788,
      "step": 6950
    },
    {
      "epoch": 1.3099943534726144,
      "grad_norm": 1.9069128036499023,
      "learning_rate": 0.00036900056465273854,
      "loss": 3.5732,
      "step": 6960
    },
    {
      "epoch": 1.3118765292678336,
      "grad_norm": 2.2385337352752686,
      "learning_rate": 0.0003688123470732166,
      "loss": 3.8875,
      "step": 6970
    },
    {
      "epoch": 1.3137587050630528,
      "grad_norm": 1.7741254568099976,
      "learning_rate": 0.00036862412949369475,
      "loss": 3.7516,
      "step": 6980
    },
    {
      "epoch": 1.3156408808582722,
      "grad_norm": 1.8411750793457031,
      "learning_rate": 0.00036843591191417283,
      "loss": 3.738,
      "step": 6990
    },
    {
      "epoch": 1.3175230566534915,
      "grad_norm": 2.279778003692627,
      "learning_rate": 0.0003682476943346509,
      "loss": 3.5345,
      "step": 7000
    },
    {
      "epoch": 1.3194052324487107,
      "grad_norm": 1.9499634504318237,
      "learning_rate": 0.000368059476755129,
      "loss": 3.6431,
      "step": 7010
    },
    {
      "epoch": 1.32128740824393,
      "grad_norm": 2.2675411701202393,
      "learning_rate": 0.000367871259175607,
      "loss": 3.6623,
      "step": 7020
    },
    {
      "epoch": 1.3231695840391493,
      "grad_norm": 2.3390955924987793,
      "learning_rate": 0.0003676830415960851,
      "loss": 3.7228,
      "step": 7030
    },
    {
      "epoch": 1.3250517598343685,
      "grad_norm": 1.957190752029419,
      "learning_rate": 0.00036749482401656316,
      "loss": 3.6893,
      "step": 7040
    },
    {
      "epoch": 1.3269339356295877,
      "grad_norm": 1.7984035015106201,
      "learning_rate": 0.00036730660643704123,
      "loss": 3.6364,
      "step": 7050
    },
    {
      "epoch": 1.3288161114248072,
      "grad_norm": 3.774339199066162,
      "learning_rate": 0.0003671183888575193,
      "loss": 3.8089,
      "step": 7060
    },
    {
      "epoch": 1.3306982872200264,
      "grad_norm": 2.1288113594055176,
      "learning_rate": 0.0003669301712779974,
      "loss": 3.5106,
      "step": 7070
    },
    {
      "epoch": 1.3325804630152456,
      "grad_norm": 2.831446886062622,
      "learning_rate": 0.00036674195369847546,
      "loss": 3.7959,
      "step": 7080
    },
    {
      "epoch": 1.3344626388104648,
      "grad_norm": 2.0102405548095703,
      "learning_rate": 0.00036655373611895354,
      "loss": 3.8562,
      "step": 7090
    },
    {
      "epoch": 1.3363448146056842,
      "grad_norm": 2.1046881675720215,
      "learning_rate": 0.00036636551853943156,
      "loss": 3.8892,
      "step": 7100
    },
    {
      "epoch": 1.3382269904009034,
      "grad_norm": 1.8360822200775146,
      "learning_rate": 0.00036617730095990964,
      "loss": 3.6919,
      "step": 7110
    },
    {
      "epoch": 1.3401091661961226,
      "grad_norm": 2.020962715148926,
      "learning_rate": 0.0003659890833803877,
      "loss": 3.7452,
      "step": 7120
    },
    {
      "epoch": 1.341991341991342,
      "grad_norm": 1.8157103061676025,
      "learning_rate": 0.0003658008658008658,
      "loss": 3.4925,
      "step": 7130
    },
    {
      "epoch": 1.3438735177865613,
      "grad_norm": 2.1023714542388916,
      "learning_rate": 0.00036561264822134387,
      "loss": 3.9745,
      "step": 7140
    },
    {
      "epoch": 1.3457556935817805,
      "grad_norm": 1.8831114768981934,
      "learning_rate": 0.00036542443064182194,
      "loss": 3.8476,
      "step": 7150
    },
    {
      "epoch": 1.347637869377,
      "grad_norm": 2.2279751300811768,
      "learning_rate": 0.0003652362130623,
      "loss": 4.0432,
      "step": 7160
    },
    {
      "epoch": 1.3495200451722191,
      "grad_norm": 2.070411443710327,
      "learning_rate": 0.0003650479954827781,
      "loss": 3.4937,
      "step": 7170
    },
    {
      "epoch": 1.3514022209674383,
      "grad_norm": 2.1437418460845947,
      "learning_rate": 0.00036485977790325617,
      "loss": 3.9656,
      "step": 7180
    },
    {
      "epoch": 1.3532843967626578,
      "grad_norm": 1.7823975086212158,
      "learning_rate": 0.0003646715603237342,
      "loss": 3.8643,
      "step": 7190
    },
    {
      "epoch": 1.355166572557877,
      "grad_norm": 2.0218539237976074,
      "learning_rate": 0.0003644833427442123,
      "loss": 3.7161,
      "step": 7200
    },
    {
      "epoch": 1.3570487483530962,
      "grad_norm": 2.5077080726623535,
      "learning_rate": 0.0003642951251646904,
      "loss": 3.7599,
      "step": 7210
    },
    {
      "epoch": 1.3589309241483154,
      "grad_norm": 1.9613864421844482,
      "learning_rate": 0.0003641069075851685,
      "loss": 3.6094,
      "step": 7220
    },
    {
      "epoch": 1.3608130999435346,
      "grad_norm": 1.76811945438385,
      "learning_rate": 0.00036391869000564655,
      "loss": 3.7828,
      "step": 7230
    },
    {
      "epoch": 1.362695275738754,
      "grad_norm": 2.1831283569335938,
      "learning_rate": 0.00036373047242612463,
      "loss": 3.5293,
      "step": 7240
    },
    {
      "epoch": 1.3645774515339733,
      "grad_norm": 1.9235538244247437,
      "learning_rate": 0.0003635422548466027,
      "loss": 3.4915,
      "step": 7250
    },
    {
      "epoch": 1.3664596273291925,
      "grad_norm": 2.1489882469177246,
      "learning_rate": 0.0003633540372670808,
      "loss": 3.8052,
      "step": 7260
    },
    {
      "epoch": 1.368341803124412,
      "grad_norm": 1.9953148365020752,
      "learning_rate": 0.00036316581968755886,
      "loss": 3.8223,
      "step": 7270
    },
    {
      "epoch": 1.370223978919631,
      "grad_norm": 2.3743884563446045,
      "learning_rate": 0.0003629776021080369,
      "loss": 3.6882,
      "step": 7280
    },
    {
      "epoch": 1.3721061547148503,
      "grad_norm": 1.9586063623428345,
      "learning_rate": 0.00036278938452851496,
      "loss": 3.7565,
      "step": 7290
    },
    {
      "epoch": 1.3739883305100697,
      "grad_norm": 1.9832416772842407,
      "learning_rate": 0.00036260116694899303,
      "loss": 3.6247,
      "step": 7300
    },
    {
      "epoch": 1.375870506305289,
      "grad_norm": 1.792038917541504,
      "learning_rate": 0.0003624129493694711,
      "loss": 3.643,
      "step": 7310
    },
    {
      "epoch": 1.3777526821005082,
      "grad_norm": 2.0198590755462646,
      "learning_rate": 0.0003622247317899492,
      "loss": 3.762,
      "step": 7320
    },
    {
      "epoch": 1.3796348578957276,
      "grad_norm": 2.0416181087493896,
      "learning_rate": 0.00036203651421042726,
      "loss": 3.5613,
      "step": 7330
    },
    {
      "epoch": 1.3815170336909468,
      "grad_norm": 2.0119645595550537,
      "learning_rate": 0.00036184829663090534,
      "loss": 3.7551,
      "step": 7340
    },
    {
      "epoch": 1.383399209486166,
      "grad_norm": 1.6131541728973389,
      "learning_rate": 0.0003616600790513834,
      "loss": 3.788,
      "step": 7350
    },
    {
      "epoch": 1.3852813852813852,
      "grad_norm": 1.9785635471343994,
      "learning_rate": 0.00036147186147186144,
      "loss": 3.6012,
      "step": 7360
    },
    {
      "epoch": 1.3871635610766044,
      "grad_norm": 1.961735725402832,
      "learning_rate": 0.0003612836438923395,
      "loss": 3.52,
      "step": 7370
    },
    {
      "epoch": 1.3890457368718239,
      "grad_norm": 2.128279209136963,
      "learning_rate": 0.0003610954263128176,
      "loss": 3.8339,
      "step": 7380
    },
    {
      "epoch": 1.390927912667043,
      "grad_norm": 1.8960368633270264,
      "learning_rate": 0.00036090720873329567,
      "loss": 3.6089,
      "step": 7390
    },
    {
      "epoch": 1.3928100884622623,
      "grad_norm": 2.069545269012451,
      "learning_rate": 0.00036071899115377374,
      "loss": 3.9857,
      "step": 7400
    },
    {
      "epoch": 1.3946922642574817,
      "grad_norm": 2.1086106300354004,
      "learning_rate": 0.0003605307735742518,
      "loss": 3.9457,
      "step": 7410
    },
    {
      "epoch": 1.396574440052701,
      "grad_norm": 1.82731294631958,
      "learning_rate": 0.00036034255599472995,
      "loss": 3.8394,
      "step": 7420
    },
    {
      "epoch": 1.3984566158479201,
      "grad_norm": 1.7885284423828125,
      "learning_rate": 0.00036015433841520803,
      "loss": 3.5917,
      "step": 7430
    },
    {
      "epoch": 1.4003387916431396,
      "grad_norm": 1.8847382068634033,
      "learning_rate": 0.0003599661208356861,
      "loss": 3.8681,
      "step": 7440
    },
    {
      "epoch": 1.4022209674383588,
      "grad_norm": 1.9530550241470337,
      "learning_rate": 0.0003597779032561641,
      "loss": 3.7272,
      "step": 7450
    },
    {
      "epoch": 1.404103143233578,
      "grad_norm": 2.2070717811584473,
      "learning_rate": 0.0003595896856766422,
      "loss": 4.1098,
      "step": 7460
    },
    {
      "epoch": 1.4059853190287974,
      "grad_norm": 1.8252630233764648,
      "learning_rate": 0.0003594014680971203,
      "loss": 3.6822,
      "step": 7470
    },
    {
      "epoch": 1.4078674948240166,
      "grad_norm": 1.8794349431991577,
      "learning_rate": 0.00035921325051759836,
      "loss": 3.8877,
      "step": 7480
    },
    {
      "epoch": 1.4097496706192358,
      "grad_norm": 1.8576912879943848,
      "learning_rate": 0.00035902503293807643,
      "loss": 3.8438,
      "step": 7490
    },
    {
      "epoch": 1.411631846414455,
      "grad_norm": 2.027756690979004,
      "learning_rate": 0.0003588368153585545,
      "loss": 3.7399,
      "step": 7500
    },
    {
      "epoch": 1.4135140222096743,
      "grad_norm": 2.149160861968994,
      "learning_rate": 0.0003586485977790326,
      "loss": 3.8171,
      "step": 7510
    },
    {
      "epoch": 1.4153961980048937,
      "grad_norm": 2.319580554962158,
      "learning_rate": 0.00035846038019951066,
      "loss": 3.7717,
      "step": 7520
    },
    {
      "epoch": 1.417278373800113,
      "grad_norm": 3.2247745990753174,
      "learning_rate": 0.00035827216261998874,
      "loss": 3.7937,
      "step": 7530
    },
    {
      "epoch": 1.419160549595332,
      "grad_norm": 1.9991421699523926,
      "learning_rate": 0.00035808394504046676,
      "loss": 3.437,
      "step": 7540
    },
    {
      "epoch": 1.4210427253905515,
      "grad_norm": 2.586153984069824,
      "learning_rate": 0.00035789572746094484,
      "loss": 3.963,
      "step": 7550
    },
    {
      "epoch": 1.4229249011857708,
      "grad_norm": 2.0890746116638184,
      "learning_rate": 0.0003577075098814229,
      "loss": 3.6881,
      "step": 7560
    },
    {
      "epoch": 1.42480707698099,
      "grad_norm": 1.775581955909729,
      "learning_rate": 0.000357519292301901,
      "loss": 3.5551,
      "step": 7570
    },
    {
      "epoch": 1.4266892527762094,
      "grad_norm": 2.140167474746704,
      "learning_rate": 0.00035733107472237907,
      "loss": 3.7377,
      "step": 7580
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 1.9478838443756104,
      "learning_rate": 0.00035714285714285714,
      "loss": 3.786,
      "step": 7590
    },
    {
      "epoch": 1.4304536043666478,
      "grad_norm": 2.1335339546203613,
      "learning_rate": 0.0003569546395633352,
      "loss": 3.6391,
      "step": 7600
    },
    {
      "epoch": 1.4323357801618672,
      "grad_norm": 2.1355767250061035,
      "learning_rate": 0.0003567664219838133,
      "loss": 3.5135,
      "step": 7610
    },
    {
      "epoch": 1.4342179559570865,
      "grad_norm": 1.8865129947662354,
      "learning_rate": 0.00035657820440429137,
      "loss": 3.7835,
      "step": 7620
    },
    {
      "epoch": 1.4361001317523057,
      "grad_norm": 2.1274356842041016,
      "learning_rate": 0.0003563899868247694,
      "loss": 3.671,
      "step": 7630
    },
    {
      "epoch": 1.4379823075475249,
      "grad_norm": 2.126617670059204,
      "learning_rate": 0.0003562017692452475,
      "loss": 3.953,
      "step": 7640
    },
    {
      "epoch": 1.439864483342744,
      "grad_norm": 1.896268367767334,
      "learning_rate": 0.0003560135516657256,
      "loss": 3.7844,
      "step": 7650
    },
    {
      "epoch": 1.4417466591379635,
      "grad_norm": 2.0018656253814697,
      "learning_rate": 0.0003558253340862037,
      "loss": 3.9747,
      "step": 7660
    },
    {
      "epoch": 1.4436288349331827,
      "grad_norm": 2.149848222732544,
      "learning_rate": 0.00035563711650668175,
      "loss": 3.6292,
      "step": 7670
    },
    {
      "epoch": 1.445511010728402,
      "grad_norm": 1.9063892364501953,
      "learning_rate": 0.00035544889892715983,
      "loss": 3.9736,
      "step": 7680
    },
    {
      "epoch": 1.4473931865236214,
      "grad_norm": 2.0680341720581055,
      "learning_rate": 0.0003552606813476379,
      "loss": 3.7596,
      "step": 7690
    },
    {
      "epoch": 1.4492753623188406,
      "grad_norm": 1.9787496328353882,
      "learning_rate": 0.000355072463768116,
      "loss": 3.9205,
      "step": 7700
    },
    {
      "epoch": 1.4511575381140598,
      "grad_norm": 2.1122937202453613,
      "learning_rate": 0.000354884246188594,
      "loss": 3.8455,
      "step": 7710
    },
    {
      "epoch": 1.4530397139092792,
      "grad_norm": 2.253380298614502,
      "learning_rate": 0.0003546960286090721,
      "loss": 3.797,
      "step": 7720
    },
    {
      "epoch": 1.4549218897044984,
      "grad_norm": 1.7869551181793213,
      "learning_rate": 0.00035450781102955016,
      "loss": 3.7453,
      "step": 7730
    },
    {
      "epoch": 1.4568040654997176,
      "grad_norm": 2.2952880859375,
      "learning_rate": 0.00035431959345002823,
      "loss": 3.8178,
      "step": 7740
    },
    {
      "epoch": 1.458686241294937,
      "grad_norm": 2.147580623626709,
      "learning_rate": 0.0003541313758705063,
      "loss": 3.8506,
      "step": 7750
    },
    {
      "epoch": 1.4605684170901563,
      "grad_norm": 1.6160011291503906,
      "learning_rate": 0.0003539431582909844,
      "loss": 3.8723,
      "step": 7760
    },
    {
      "epoch": 1.4624505928853755,
      "grad_norm": 1.8009629249572754,
      "learning_rate": 0.00035375494071146246,
      "loss": 4.0286,
      "step": 7770
    },
    {
      "epoch": 1.4643327686805947,
      "grad_norm": 2.0394809246063232,
      "learning_rate": 0.00035356672313194054,
      "loss": 3.7727,
      "step": 7780
    },
    {
      "epoch": 1.466214944475814,
      "grad_norm": 1.983485221862793,
      "learning_rate": 0.0003533785055524186,
      "loss": 3.674,
      "step": 7790
    },
    {
      "epoch": 1.4680971202710333,
      "grad_norm": 1.917427659034729,
      "learning_rate": 0.00035319028797289664,
      "loss": 3.8096,
      "step": 7800
    },
    {
      "epoch": 1.4699792960662525,
      "grad_norm": 1.7408658266067505,
      "learning_rate": 0.0003530020703933747,
      "loss": 3.9625,
      "step": 7810
    },
    {
      "epoch": 1.4718614718614718,
      "grad_norm": 2.1909754276275635,
      "learning_rate": 0.0003528138528138528,
      "loss": 3.4834,
      "step": 7820
    },
    {
      "epoch": 1.4737436476566912,
      "grad_norm": 2.2905852794647217,
      "learning_rate": 0.00035262563523433087,
      "loss": 3.7214,
      "step": 7830
    },
    {
      "epoch": 1.4756258234519104,
      "grad_norm": 2.4130899906158447,
      "learning_rate": 0.00035243741765480894,
      "loss": 3.7839,
      "step": 7840
    },
    {
      "epoch": 1.4775079992471296,
      "grad_norm": 2.264302968978882,
      "learning_rate": 0.000352249200075287,
      "loss": 3.7727,
      "step": 7850
    },
    {
      "epoch": 1.479390175042349,
      "grad_norm": 2.3765761852264404,
      "learning_rate": 0.00035206098249576515,
      "loss": 3.7051,
      "step": 7860
    },
    {
      "epoch": 1.4812723508375683,
      "grad_norm": 1.75021231174469,
      "learning_rate": 0.00035187276491624323,
      "loss": 3.9133,
      "step": 7870
    },
    {
      "epoch": 1.4831545266327875,
      "grad_norm": 1.860624074935913,
      "learning_rate": 0.0003516845473367213,
      "loss": 3.8135,
      "step": 7880
    },
    {
      "epoch": 1.485036702428007,
      "grad_norm": 1.8294997215270996,
      "learning_rate": 0.0003514963297571993,
      "loss": 3.6046,
      "step": 7890
    },
    {
      "epoch": 1.486918878223226,
      "grad_norm": 1.9020800590515137,
      "learning_rate": 0.0003513081121776774,
      "loss": 3.7568,
      "step": 7900
    },
    {
      "epoch": 1.4888010540184453,
      "grad_norm": 1.8927626609802246,
      "learning_rate": 0.0003511198945981555,
      "loss": 4.121,
      "step": 7910
    },
    {
      "epoch": 1.4906832298136645,
      "grad_norm": 1.9175612926483154,
      "learning_rate": 0.00035093167701863356,
      "loss": 3.657,
      "step": 7920
    },
    {
      "epoch": 1.4925654056088837,
      "grad_norm": 2.6176860332489014,
      "learning_rate": 0.00035074345943911163,
      "loss": 3.7087,
      "step": 7930
    },
    {
      "epoch": 1.4944475814041032,
      "grad_norm": 1.8022160530090332,
      "learning_rate": 0.0003505552418595897,
      "loss": 3.9254,
      "step": 7940
    },
    {
      "epoch": 1.4963297571993224,
      "grad_norm": 1.8247177600860596,
      "learning_rate": 0.0003503670242800678,
      "loss": 3.5697,
      "step": 7950
    },
    {
      "epoch": 1.4982119329945416,
      "grad_norm": 1.8403944969177246,
      "learning_rate": 0.00035017880670054586,
      "loss": 3.591,
      "step": 7960
    },
    {
      "epoch": 1.500094108789761,
      "grad_norm": 1.9349924325942993,
      "learning_rate": 0.0003499905891210239,
      "loss": 3.6947,
      "step": 7970
    },
    {
      "epoch": 1.5019762845849802,
      "grad_norm": 2.04209303855896,
      "learning_rate": 0.00034980237154150196,
      "loss": 3.8419,
      "step": 7980
    },
    {
      "epoch": 1.5038584603801994,
      "grad_norm": 1.9403793811798096,
      "learning_rate": 0.00034961415396198004,
      "loss": 3.8686,
      "step": 7990
    },
    {
      "epoch": 1.5057406361754189,
      "grad_norm": 2.132564067840576,
      "learning_rate": 0.0003494259363824581,
      "loss": 3.7871,
      "step": 8000
    },
    {
      "epoch": 1.507622811970638,
      "grad_norm": 1.7702946662902832,
      "learning_rate": 0.0003492377188029362,
      "loss": 3.6537,
      "step": 8010
    },
    {
      "epoch": 1.5095049877658573,
      "grad_norm": 1.8843390941619873,
      "learning_rate": 0.00034904950122341427,
      "loss": 3.9158,
      "step": 8020
    },
    {
      "epoch": 1.5113871635610767,
      "grad_norm": 2.0728650093078613,
      "learning_rate": 0.00034886128364389234,
      "loss": 3.711,
      "step": 8030
    },
    {
      "epoch": 1.513269339356296,
      "grad_norm": 2.3643085956573486,
      "learning_rate": 0.0003486730660643704,
      "loss": 3.7737,
      "step": 8040
    },
    {
      "epoch": 1.5151515151515151,
      "grad_norm": 2.087571859359741,
      "learning_rate": 0.0003484848484848485,
      "loss": 3.8039,
      "step": 8050
    },
    {
      "epoch": 1.5170336909467346,
      "grad_norm": 1.801803469657898,
      "learning_rate": 0.0003482966309053265,
      "loss": 3.6293,
      "step": 8060
    },
    {
      "epoch": 1.5189158667419536,
      "grad_norm": 2.0934271812438965,
      "learning_rate": 0.0003481084133258046,
      "loss": 3.7736,
      "step": 8070
    },
    {
      "epoch": 1.520798042537173,
      "grad_norm": 2.0371158123016357,
      "learning_rate": 0.00034792019574628267,
      "loss": 3.6974,
      "step": 8080
    },
    {
      "epoch": 1.5226802183323922,
      "grad_norm": 2.2584123611450195,
      "learning_rate": 0.0003477319781667608,
      "loss": 3.8715,
      "step": 8090
    },
    {
      "epoch": 1.5245623941276114,
      "grad_norm": 1.8013052940368652,
      "learning_rate": 0.0003475437605872389,
      "loss": 3.6496,
      "step": 8100
    },
    {
      "epoch": 1.5264445699228308,
      "grad_norm": 1.840074896812439,
      "learning_rate": 0.00034735554300771695,
      "loss": 3.6919,
      "step": 8110
    },
    {
      "epoch": 1.52832674571805,
      "grad_norm": 1.870167851448059,
      "learning_rate": 0.00034716732542819503,
      "loss": 3.6268,
      "step": 8120
    },
    {
      "epoch": 1.5302089215132693,
      "grad_norm": 2.217456340789795,
      "learning_rate": 0.0003469791078486731,
      "loss": 3.7207,
      "step": 8130
    },
    {
      "epoch": 1.5320910973084887,
      "grad_norm": 2.008578300476074,
      "learning_rate": 0.0003467908902691512,
      "loss": 3.6737,
      "step": 8140
    },
    {
      "epoch": 1.533973273103708,
      "grad_norm": 1.9132500886917114,
      "learning_rate": 0.0003466026726896292,
      "loss": 3.8286,
      "step": 8150
    },
    {
      "epoch": 1.5358554488989271,
      "grad_norm": 1.8686176538467407,
      "learning_rate": 0.0003464144551101073,
      "loss": 3.6473,
      "step": 8160
    },
    {
      "epoch": 1.5377376246941465,
      "grad_norm": 2.028219699859619,
      "learning_rate": 0.00034622623753058536,
      "loss": 3.9318,
      "step": 8170
    },
    {
      "epoch": 1.5396198004893658,
      "grad_norm": 2.1390788555145264,
      "learning_rate": 0.00034603801995106343,
      "loss": 3.6349,
      "step": 8180
    },
    {
      "epoch": 1.541501976284585,
      "grad_norm": 1.924244999885559,
      "learning_rate": 0.0003458498023715415,
      "loss": 3.9472,
      "step": 8190
    },
    {
      "epoch": 1.5433841520798044,
      "grad_norm": 1.8411937952041626,
      "learning_rate": 0.0003456615847920196,
      "loss": 3.8442,
      "step": 8200
    },
    {
      "epoch": 1.5452663278750234,
      "grad_norm": 1.9796812534332275,
      "learning_rate": 0.00034547336721249766,
      "loss": 3.5435,
      "step": 8210
    },
    {
      "epoch": 1.5471485036702428,
      "grad_norm": 2.0202064514160156,
      "learning_rate": 0.00034528514963297574,
      "loss": 3.6203,
      "step": 8220
    },
    {
      "epoch": 1.549030679465462,
      "grad_norm": 2.0582363605499268,
      "learning_rate": 0.0003450969320534538,
      "loss": 3.7731,
      "step": 8230
    },
    {
      "epoch": 1.5509128552606812,
      "grad_norm": 1.8645615577697754,
      "learning_rate": 0.00034490871447393184,
      "loss": 3.9406,
      "step": 8240
    },
    {
      "epoch": 1.5527950310559007,
      "grad_norm": 2.5893502235412598,
      "learning_rate": 0.0003447204968944099,
      "loss": 3.6305,
      "step": 8250
    },
    {
      "epoch": 1.5546772068511199,
      "grad_norm": 2.3030471801757812,
      "learning_rate": 0.000344532279314888,
      "loss": 3.9195,
      "step": 8260
    },
    {
      "epoch": 1.556559382646339,
      "grad_norm": 2.0002682209014893,
      "learning_rate": 0.00034434406173536607,
      "loss": 3.8249,
      "step": 8270
    },
    {
      "epoch": 1.5584415584415585,
      "grad_norm": 1.8045579195022583,
      "learning_rate": 0.00034415584415584414,
      "loss": 3.8484,
      "step": 8280
    },
    {
      "epoch": 1.5603237342367777,
      "grad_norm": 1.8935033082962036,
      "learning_rate": 0.0003439676265763222,
      "loss": 3.6696,
      "step": 8290
    },
    {
      "epoch": 1.562205910031997,
      "grad_norm": 1.8719751834869385,
      "learning_rate": 0.00034377940899680035,
      "loss": 3.7094,
      "step": 8300
    },
    {
      "epoch": 1.5640880858272164,
      "grad_norm": 2.444491147994995,
      "learning_rate": 0.00034359119141727843,
      "loss": 4.0846,
      "step": 8310
    },
    {
      "epoch": 1.5659702616224356,
      "grad_norm": 2.4288089275360107,
      "learning_rate": 0.00034340297383775645,
      "loss": 3.7514,
      "step": 8320
    },
    {
      "epoch": 1.5678524374176548,
      "grad_norm": 2.324831008911133,
      "learning_rate": 0.0003432147562582345,
      "loss": 3.6748,
      "step": 8330
    },
    {
      "epoch": 1.5697346132128742,
      "grad_norm": 2.605628728866577,
      "learning_rate": 0.0003430265386787126,
      "loss": 3.9288,
      "step": 8340
    },
    {
      "epoch": 1.5716167890080932,
      "grad_norm": 1.9950097799301147,
      "learning_rate": 0.0003428383210991907,
      "loss": 3.5434,
      "step": 8350
    },
    {
      "epoch": 1.5734989648033126,
      "grad_norm": 2.8846535682678223,
      "learning_rate": 0.00034265010351966876,
      "loss": 3.7692,
      "step": 8360
    },
    {
      "epoch": 1.5753811405985318,
      "grad_norm": 1.8857200145721436,
      "learning_rate": 0.00034246188594014683,
      "loss": 3.5541,
      "step": 8370
    },
    {
      "epoch": 1.577263316393751,
      "grad_norm": 1.9304348230361938,
      "learning_rate": 0.0003422736683606249,
      "loss": 3.7289,
      "step": 8380
    },
    {
      "epoch": 1.5791454921889705,
      "grad_norm": 2.28515887260437,
      "learning_rate": 0.000342085450781103,
      "loss": 3.7085,
      "step": 8390
    },
    {
      "epoch": 1.5810276679841897,
      "grad_norm": 2.3015100955963135,
      "learning_rate": 0.00034189723320158106,
      "loss": 3.5947,
      "step": 8400
    },
    {
      "epoch": 1.582909843779409,
      "grad_norm": 2.521632432937622,
      "learning_rate": 0.0003417090156220591,
      "loss": 3.8023,
      "step": 8410
    },
    {
      "epoch": 1.5847920195746283,
      "grad_norm": 2.1546032428741455,
      "learning_rate": 0.00034152079804253716,
      "loss": 3.9682,
      "step": 8420
    },
    {
      "epoch": 1.5866741953698476,
      "grad_norm": 2.0820095539093018,
      "learning_rate": 0.00034133258046301524,
      "loss": 3.9084,
      "step": 8430
    },
    {
      "epoch": 1.5885563711650668,
      "grad_norm": 1.989452838897705,
      "learning_rate": 0.0003411443628834933,
      "loss": 3.793,
      "step": 8440
    },
    {
      "epoch": 1.5904385469602862,
      "grad_norm": 1.795833706855774,
      "learning_rate": 0.0003409561453039714,
      "loss": 3.7056,
      "step": 8450
    },
    {
      "epoch": 1.5923207227555054,
      "grad_norm": 2.0363857746124268,
      "learning_rate": 0.00034076792772444947,
      "loss": 3.795,
      "step": 8460
    },
    {
      "epoch": 1.5942028985507246,
      "grad_norm": 1.7800500392913818,
      "learning_rate": 0.00034057971014492754,
      "loss": 3.7061,
      "step": 8470
    },
    {
      "epoch": 1.596085074345944,
      "grad_norm": 2.08921480178833,
      "learning_rate": 0.0003403914925654056,
      "loss": 3.5798,
      "step": 8480
    },
    {
      "epoch": 1.597967250141163,
      "grad_norm": 1.9512754678726196,
      "learning_rate": 0.0003402032749858837,
      "loss": 3.8301,
      "step": 8490
    },
    {
      "epoch": 1.5998494259363825,
      "grad_norm": 2.3853600025177,
      "learning_rate": 0.0003400150574063617,
      "loss": 3.9632,
      "step": 8500
    },
    {
      "epoch": 1.601731601731602,
      "grad_norm": 2.057528018951416,
      "learning_rate": 0.0003398268398268398,
      "loss": 3.6631,
      "step": 8510
    },
    {
      "epoch": 1.6036137775268209,
      "grad_norm": 1.8551961183547974,
      "learning_rate": 0.00033963862224731787,
      "loss": 3.6559,
      "step": 8520
    },
    {
      "epoch": 1.6054959533220403,
      "grad_norm": 1.7113333940505981,
      "learning_rate": 0.000339450404667796,
      "loss": 3.5879,
      "step": 8530
    },
    {
      "epoch": 1.6073781291172595,
      "grad_norm": 1.894375205039978,
      "learning_rate": 0.0003392621870882741,
      "loss": 3.9233,
      "step": 8540
    },
    {
      "epoch": 1.6092603049124787,
      "grad_norm": 1.7292157411575317,
      "learning_rate": 0.00033907396950875215,
      "loss": 3.6998,
      "step": 8550
    },
    {
      "epoch": 1.6111424807076982,
      "grad_norm": 2.0833964347839355,
      "learning_rate": 0.00033888575192923023,
      "loss": 3.7333,
      "step": 8560
    },
    {
      "epoch": 1.6130246565029174,
      "grad_norm": 2.16780686378479,
      "learning_rate": 0.0003386975343497083,
      "loss": 3.858,
      "step": 8570
    },
    {
      "epoch": 1.6149068322981366,
      "grad_norm": 2.318399429321289,
      "learning_rate": 0.00033850931677018633,
      "loss": 3.6575,
      "step": 8580
    },
    {
      "epoch": 1.616789008093356,
      "grad_norm": 2.1125340461730957,
      "learning_rate": 0.0003383210991906644,
      "loss": 3.4829,
      "step": 8590
    },
    {
      "epoch": 1.6186711838885752,
      "grad_norm": 2.2121551036834717,
      "learning_rate": 0.0003381328816111425,
      "loss": 3.7985,
      "step": 8600
    },
    {
      "epoch": 1.6205533596837944,
      "grad_norm": 2.090364933013916,
      "learning_rate": 0.00033794466403162056,
      "loss": 3.3342,
      "step": 8610
    },
    {
      "epoch": 1.6224355354790139,
      "grad_norm": 1.9245707988739014,
      "learning_rate": 0.00033775644645209863,
      "loss": 3.7779,
      "step": 8620
    },
    {
      "epoch": 1.6243177112742329,
      "grad_norm": 1.9286541938781738,
      "learning_rate": 0.0003375682288725767,
      "loss": 3.7115,
      "step": 8630
    },
    {
      "epoch": 1.6261998870694523,
      "grad_norm": 2.4994375705718994,
      "learning_rate": 0.0003373800112930548,
      "loss": 3.9738,
      "step": 8640
    },
    {
      "epoch": 1.6280820628646717,
      "grad_norm": 2.1817786693573,
      "learning_rate": 0.00033719179371353286,
      "loss": 3.7679,
      "step": 8650
    },
    {
      "epoch": 1.6299642386598907,
      "grad_norm": 2.399630308151245,
      "learning_rate": 0.00033700357613401094,
      "loss": 3.7907,
      "step": 8660
    },
    {
      "epoch": 1.6318464144551101,
      "grad_norm": 1.7689156532287598,
      "learning_rate": 0.00033681535855448896,
      "loss": 3.7116,
      "step": 8670
    },
    {
      "epoch": 1.6337285902503293,
      "grad_norm": 2.208608627319336,
      "learning_rate": 0.00033662714097496704,
      "loss": 3.9298,
      "step": 8680
    },
    {
      "epoch": 1.6356107660455486,
      "grad_norm": 2.1811182498931885,
      "learning_rate": 0.0003364389233954451,
      "loss": 3.6754,
      "step": 8690
    },
    {
      "epoch": 1.637492941840768,
      "grad_norm": 2.085291862487793,
      "learning_rate": 0.0003362507058159232,
      "loss": 3.7016,
      "step": 8700
    },
    {
      "epoch": 1.6393751176359872,
      "grad_norm": 1.8821417093276978,
      "learning_rate": 0.00033606248823640127,
      "loss": 3.7248,
      "step": 8710
    },
    {
      "epoch": 1.6412572934312064,
      "grad_norm": 1.8898390531539917,
      "learning_rate": 0.00033587427065687934,
      "loss": 3.6938,
      "step": 8720
    },
    {
      "epoch": 1.6431394692264258,
      "grad_norm": 1.7819640636444092,
      "learning_rate": 0.0003356860530773574,
      "loss": 3.5247,
      "step": 8730
    },
    {
      "epoch": 1.645021645021645,
      "grad_norm": 1.8509273529052734,
      "learning_rate": 0.0003354978354978355,
      "loss": 3.6139,
      "step": 8740
    },
    {
      "epoch": 1.6469038208168643,
      "grad_norm": 2.061574935913086,
      "learning_rate": 0.00033530961791831363,
      "loss": 3.803,
      "step": 8750
    },
    {
      "epoch": 1.6487859966120837,
      "grad_norm": 1.8840912580490112,
      "learning_rate": 0.00033512140033879165,
      "loss": 3.5845,
      "step": 8760
    },
    {
      "epoch": 1.6506681724073027,
      "grad_norm": 2.0159287452697754,
      "learning_rate": 0.0003349331827592697,
      "loss": 3.9504,
      "step": 8770
    },
    {
      "epoch": 1.6525503482025221,
      "grad_norm": 1.6903138160705566,
      "learning_rate": 0.0003347449651797478,
      "loss": 3.6987,
      "step": 8780
    },
    {
      "epoch": 1.6544325239977415,
      "grad_norm": 2.134758234024048,
      "learning_rate": 0.0003345567476002259,
      "loss": 3.7158,
      "step": 8790
    },
    {
      "epoch": 1.6563146997929605,
      "grad_norm": 2.000469923019409,
      "learning_rate": 0.00033436853002070396,
      "loss": 3.7533,
      "step": 8800
    },
    {
      "epoch": 1.65819687558818,
      "grad_norm": 1.822241187095642,
      "learning_rate": 0.00033418031244118203,
      "loss": 3.7504,
      "step": 8810
    },
    {
      "epoch": 1.6600790513833992,
      "grad_norm": 1.8779278993606567,
      "learning_rate": 0.0003339920948616601,
      "loss": 3.6283,
      "step": 8820
    },
    {
      "epoch": 1.6619612271786184,
      "grad_norm": 1.9598212242126465,
      "learning_rate": 0.0003338038772821382,
      "loss": 3.7713,
      "step": 8830
    },
    {
      "epoch": 1.6638434029738378,
      "grad_norm": 1.7098469734191895,
      "learning_rate": 0.00033361565970261626,
      "loss": 3.5398,
      "step": 8840
    },
    {
      "epoch": 1.665725578769057,
      "grad_norm": 2.3827857971191406,
      "learning_rate": 0.0003334274421230943,
      "loss": 3.861,
      "step": 8850
    },
    {
      "epoch": 1.6676077545642762,
      "grad_norm": 1.9005036354064941,
      "learning_rate": 0.00033323922454357236,
      "loss": 3.9028,
      "step": 8860
    },
    {
      "epoch": 1.6694899303594957,
      "grad_norm": 2.504202127456665,
      "learning_rate": 0.00033305100696405044,
      "loss": 3.8158,
      "step": 8870
    },
    {
      "epoch": 1.6713721061547149,
      "grad_norm": 2.1613893508911133,
      "learning_rate": 0.0003328627893845285,
      "loss": 3.8541,
      "step": 8880
    },
    {
      "epoch": 1.673254281949934,
      "grad_norm": 2.010159492492676,
      "learning_rate": 0.0003326745718050066,
      "loss": 3.8192,
      "step": 8890
    },
    {
      "epoch": 1.6751364577451535,
      "grad_norm": 2.078258752822876,
      "learning_rate": 0.00033248635422548467,
      "loss": 3.6468,
      "step": 8900
    },
    {
      "epoch": 1.6770186335403725,
      "grad_norm": 1.9027917385101318,
      "learning_rate": 0.00033229813664596274,
      "loss": 3.918,
      "step": 8910
    },
    {
      "epoch": 1.678900809335592,
      "grad_norm": 2.3127923011779785,
      "learning_rate": 0.0003321099190664408,
      "loss": 3.5837,
      "step": 8920
    },
    {
      "epoch": 1.6807829851308114,
      "grad_norm": 1.974990963935852,
      "learning_rate": 0.00033192170148691884,
      "loss": 3.5565,
      "step": 8930
    },
    {
      "epoch": 1.6826651609260304,
      "grad_norm": 2.2739603519439697,
      "learning_rate": 0.0003317334839073969,
      "loss": 3.7445,
      "step": 8940
    },
    {
      "epoch": 1.6845473367212498,
      "grad_norm": 2.020967483520508,
      "learning_rate": 0.000331545266327875,
      "loss": 3.5698,
      "step": 8950
    },
    {
      "epoch": 1.686429512516469,
      "grad_norm": 1.9386134147644043,
      "learning_rate": 0.00033135704874835307,
      "loss": 3.5284,
      "step": 8960
    },
    {
      "epoch": 1.6883116883116882,
      "grad_norm": 1.9373177289962769,
      "learning_rate": 0.0003311688311688312,
      "loss": 3.4506,
      "step": 8970
    },
    {
      "epoch": 1.6901938641069076,
      "grad_norm": 1.9539148807525635,
      "learning_rate": 0.0003309806135893093,
      "loss": 3.8893,
      "step": 8980
    },
    {
      "epoch": 1.6920760399021268,
      "grad_norm": 2.09279203414917,
      "learning_rate": 0.00033079239600978735,
      "loss": 3.6947,
      "step": 8990
    },
    {
      "epoch": 1.693958215697346,
      "grad_norm": 1.8553531169891357,
      "learning_rate": 0.00033060417843026543,
      "loss": 3.6849,
      "step": 9000
    },
    {
      "epoch": 1.6958403914925655,
      "grad_norm": 1.9943382740020752,
      "learning_rate": 0.0003304159608507435,
      "loss": 3.862,
      "step": 9010
    },
    {
      "epoch": 1.6977225672877847,
      "grad_norm": 1.917205810546875,
      "learning_rate": 0.00033022774327122153,
      "loss": 3.6617,
      "step": 9020
    },
    {
      "epoch": 1.699604743083004,
      "grad_norm": 1.9813207387924194,
      "learning_rate": 0.0003300395256916996,
      "loss": 3.9757,
      "step": 9030
    },
    {
      "epoch": 1.7014869188782233,
      "grad_norm": 1.8657891750335693,
      "learning_rate": 0.0003298513081121777,
      "loss": 3.6165,
      "step": 9040
    },
    {
      "epoch": 1.7033690946734426,
      "grad_norm": 1.8018442392349243,
      "learning_rate": 0.00032966309053265576,
      "loss": 3.9178,
      "step": 9050
    },
    {
      "epoch": 1.7052512704686618,
      "grad_norm": 2.157297372817993,
      "learning_rate": 0.00032947487295313383,
      "loss": 3.5394,
      "step": 9060
    },
    {
      "epoch": 1.7071334462638812,
      "grad_norm": 2.088235855102539,
      "learning_rate": 0.0003292866553736119,
      "loss": 3.869,
      "step": 9070
    },
    {
      "epoch": 1.7090156220591002,
      "grad_norm": 2.250558614730835,
      "learning_rate": 0.00032909843779409,
      "loss": 3.6271,
      "step": 9080
    },
    {
      "epoch": 1.7108977978543196,
      "grad_norm": 1.9630236625671387,
      "learning_rate": 0.00032891022021456806,
      "loss": 3.6673,
      "step": 9090
    },
    {
      "epoch": 1.7127799736495388,
      "grad_norm": 2.3485467433929443,
      "learning_rate": 0.00032872200263504614,
      "loss": 3.6769,
      "step": 9100
    },
    {
      "epoch": 1.714662149444758,
      "grad_norm": 2.215182304382324,
      "learning_rate": 0.00032853378505552416,
      "loss": 3.5654,
      "step": 9110
    },
    {
      "epoch": 1.7165443252399775,
      "grad_norm": 2.0479445457458496,
      "learning_rate": 0.00032834556747600224,
      "loss": 3.763,
      "step": 9120
    },
    {
      "epoch": 1.7184265010351967,
      "grad_norm": 11.284534454345703,
      "learning_rate": 0.0003281573498964803,
      "loss": 3.5577,
      "step": 9130
    },
    {
      "epoch": 1.7203086768304159,
      "grad_norm": 2.0643694400787354,
      "learning_rate": 0.0003279691323169584,
      "loss": 3.6663,
      "step": 9140
    },
    {
      "epoch": 1.7221908526256353,
      "grad_norm": 1.885436773300171,
      "learning_rate": 0.00032778091473743647,
      "loss": 3.7123,
      "step": 9150
    },
    {
      "epoch": 1.7240730284208545,
      "grad_norm": 1.779462456703186,
      "learning_rate": 0.00032759269715791454,
      "loss": 3.7451,
      "step": 9160
    },
    {
      "epoch": 1.7259552042160737,
      "grad_norm": 1.9323447942733765,
      "learning_rate": 0.0003274044795783926,
      "loss": 3.3766,
      "step": 9170
    },
    {
      "epoch": 1.7278373800112932,
      "grad_norm": 1.9485670328140259,
      "learning_rate": 0.0003272162619988707,
      "loss": 3.5259,
      "step": 9180
    },
    {
      "epoch": 1.7297195558065124,
      "grad_norm": 2.178581476211548,
      "learning_rate": 0.0003270280444193488,
      "loss": 3.4961,
      "step": 9190
    },
    {
      "epoch": 1.7316017316017316,
      "grad_norm": 2.1433050632476807,
      "learning_rate": 0.00032683982683982685,
      "loss": 3.5747,
      "step": 9200
    },
    {
      "epoch": 1.733483907396951,
      "grad_norm": 2.328223943710327,
      "learning_rate": 0.0003266516092603049,
      "loss": 4.0973,
      "step": 9210
    },
    {
      "epoch": 1.73536608319217,
      "grad_norm": 1.82240629196167,
      "learning_rate": 0.000326463391680783,
      "loss": 3.636,
      "step": 9220
    },
    {
      "epoch": 1.7372482589873894,
      "grad_norm": 1.9444760084152222,
      "learning_rate": 0.0003262751741012611,
      "loss": 3.8834,
      "step": 9230
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 2.2533152103424072,
      "learning_rate": 0.00032608695652173916,
      "loss": 3.5912,
      "step": 9240
    },
    {
      "epoch": 1.7410126105778279,
      "grad_norm": 1.868534803390503,
      "learning_rate": 0.00032589873894221723,
      "loss": 3.7303,
      "step": 9250
    },
    {
      "epoch": 1.7428947863730473,
      "grad_norm": 2.066074848175049,
      "learning_rate": 0.0003257105213626953,
      "loss": 3.6382,
      "step": 9260
    },
    {
      "epoch": 1.7447769621682665,
      "grad_norm": 2.045961856842041,
      "learning_rate": 0.0003255223037831734,
      "loss": 3.7104,
      "step": 9270
    },
    {
      "epoch": 1.7466591379634857,
      "grad_norm": 1.985349178314209,
      "learning_rate": 0.0003253340862036514,
      "loss": 3.9342,
      "step": 9280
    },
    {
      "epoch": 1.7485413137587051,
      "grad_norm": 2.5832815170288086,
      "learning_rate": 0.0003251458686241295,
      "loss": 3.5175,
      "step": 9290
    },
    {
      "epoch": 1.7504234895539243,
      "grad_norm": 2.066349744796753,
      "learning_rate": 0.00032495765104460756,
      "loss": 3.8287,
      "step": 9300
    },
    {
      "epoch": 1.7523056653491436,
      "grad_norm": 2.0592126846313477,
      "learning_rate": 0.00032476943346508564,
      "loss": 3.5318,
      "step": 9310
    },
    {
      "epoch": 1.754187841144363,
      "grad_norm": 2.074842929840088,
      "learning_rate": 0.0003245812158855637,
      "loss": 3.6565,
      "step": 9320
    },
    {
      "epoch": 1.7560700169395822,
      "grad_norm": 2.7729220390319824,
      "learning_rate": 0.0003243929983060418,
      "loss": 3.6626,
      "step": 9330
    },
    {
      "epoch": 1.7579521927348014,
      "grad_norm": 2.1782331466674805,
      "learning_rate": 0.00032420478072651987,
      "loss": 3.6741,
      "step": 9340
    },
    {
      "epoch": 1.7598343685300208,
      "grad_norm": 2.2513933181762695,
      "learning_rate": 0.00032401656314699794,
      "loss": 3.732,
      "step": 9350
    },
    {
      "epoch": 1.7617165443252398,
      "grad_norm": 1.8551970720291138,
      "learning_rate": 0.000323828345567476,
      "loss": 3.7461,
      "step": 9360
    },
    {
      "epoch": 1.7635987201204593,
      "grad_norm": 2.0710809230804443,
      "learning_rate": 0.00032364012798795404,
      "loss": 3.5542,
      "step": 9370
    },
    {
      "epoch": 1.7654808959156785,
      "grad_norm": 2.2162725925445557,
      "learning_rate": 0.0003234519104084321,
      "loss": 3.6126,
      "step": 9380
    },
    {
      "epoch": 1.7673630717108977,
      "grad_norm": 2.5064737796783447,
      "learning_rate": 0.0003232636928289102,
      "loss": 3.5244,
      "step": 9390
    },
    {
      "epoch": 1.7692452475061171,
      "grad_norm": 2.1575369834899902,
      "learning_rate": 0.00032307547524938827,
      "loss": 3.9985,
      "step": 9400
    },
    {
      "epoch": 1.7711274233013363,
      "grad_norm": 2.06486439704895,
      "learning_rate": 0.0003228872576698664,
      "loss": 3.7231,
      "step": 9410
    },
    {
      "epoch": 1.7730095990965555,
      "grad_norm": 2.050130605697632,
      "learning_rate": 0.0003226990400903445,
      "loss": 3.5453,
      "step": 9420
    },
    {
      "epoch": 1.774891774891775,
      "grad_norm": 2.550203323364258,
      "learning_rate": 0.00032251082251082255,
      "loss": 3.7155,
      "step": 9430
    },
    {
      "epoch": 1.7767739506869942,
      "grad_norm": 2.0353000164031982,
      "learning_rate": 0.00032232260493130063,
      "loss": 3.4982,
      "step": 9440
    },
    {
      "epoch": 1.7786561264822134,
      "grad_norm": 2.8174867630004883,
      "learning_rate": 0.0003221343873517787,
      "loss": 3.7628,
      "step": 9450
    },
    {
      "epoch": 1.7805383022774328,
      "grad_norm": 1.9355627298355103,
      "learning_rate": 0.00032194616977225673,
      "loss": 3.4049,
      "step": 9460
    },
    {
      "epoch": 1.782420478072652,
      "grad_norm": 2.0293326377868652,
      "learning_rate": 0.0003217579521927348,
      "loss": 3.6546,
      "step": 9470
    },
    {
      "epoch": 1.7843026538678712,
      "grad_norm": 2.5834407806396484,
      "learning_rate": 0.0003215697346132129,
      "loss": 3.6633,
      "step": 9480
    },
    {
      "epoch": 1.7861848296630907,
      "grad_norm": 2.149320602416992,
      "learning_rate": 0.00032138151703369096,
      "loss": 3.9751,
      "step": 9490
    },
    {
      "epoch": 1.7880670054583097,
      "grad_norm": 2.027061700820923,
      "learning_rate": 0.00032119329945416903,
      "loss": 3.6954,
      "step": 9500
    },
    {
      "epoch": 1.789949181253529,
      "grad_norm": 2.0868654251098633,
      "learning_rate": 0.0003210050818746471,
      "loss": 3.8539,
      "step": 9510
    },
    {
      "epoch": 1.7918313570487485,
      "grad_norm": 1.9372637271881104,
      "learning_rate": 0.0003208168642951252,
      "loss": 3.6822,
      "step": 9520
    },
    {
      "epoch": 1.7937135328439675,
      "grad_norm": 2.4747002124786377,
      "learning_rate": 0.00032062864671560326,
      "loss": 3.6153,
      "step": 9530
    },
    {
      "epoch": 1.795595708639187,
      "grad_norm": 2.240008592605591,
      "learning_rate": 0.0003204404291360813,
      "loss": 3.5098,
      "step": 9540
    },
    {
      "epoch": 1.7974778844344061,
      "grad_norm": 1.9073539972305298,
      "learning_rate": 0.00032025221155655936,
      "loss": 3.6324,
      "step": 9550
    },
    {
      "epoch": 1.7993600602296254,
      "grad_norm": 1.914000391960144,
      "learning_rate": 0.00032006399397703744,
      "loss": 3.8111,
      "step": 9560
    },
    {
      "epoch": 1.8012422360248448,
      "grad_norm": 2.694849729537964,
      "learning_rate": 0.0003198757763975155,
      "loss": 4.0209,
      "step": 9570
    },
    {
      "epoch": 1.803124411820064,
      "grad_norm": 1.8979138135910034,
      "learning_rate": 0.0003196875588179936,
      "loss": 3.7862,
      "step": 9580
    },
    {
      "epoch": 1.8050065876152832,
      "grad_norm": 2.3798153400421143,
      "learning_rate": 0.00031949934123847167,
      "loss": 3.7394,
      "step": 9590
    },
    {
      "epoch": 1.8068887634105026,
      "grad_norm": 2.3953449726104736,
      "learning_rate": 0.00031931112365894974,
      "loss": 3.6356,
      "step": 9600
    },
    {
      "epoch": 1.8087709392057219,
      "grad_norm": 1.94780695438385,
      "learning_rate": 0.0003191229060794278,
      "loss": 3.5705,
      "step": 9610
    },
    {
      "epoch": 1.810653115000941,
      "grad_norm": 2.1115198135375977,
      "learning_rate": 0.0003189346884999059,
      "loss": 3.6439,
      "step": 9620
    },
    {
      "epoch": 1.8125352907961605,
      "grad_norm": 2.078925848007202,
      "learning_rate": 0.0003187464709203839,
      "loss": 3.6912,
      "step": 9630
    },
    {
      "epoch": 1.8144174665913795,
      "grad_norm": 1.6534162759780884,
      "learning_rate": 0.00031855825334086205,
      "loss": 3.682,
      "step": 9640
    },
    {
      "epoch": 1.816299642386599,
      "grad_norm": 2.398630142211914,
      "learning_rate": 0.0003183700357613401,
      "loss": 3.6939,
      "step": 9650
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 2.451552152633667,
      "learning_rate": 0.0003181818181818182,
      "loss": 3.5832,
      "step": 9660
    },
    {
      "epoch": 1.8200639939770373,
      "grad_norm": 1.9941877126693726,
      "learning_rate": 0.0003179936006022963,
      "loss": 3.6506,
      "step": 9670
    },
    {
      "epoch": 1.8219461697722568,
      "grad_norm": 1.82035493850708,
      "learning_rate": 0.00031780538302277436,
      "loss": 3.5554,
      "step": 9680
    },
    {
      "epoch": 1.823828345567476,
      "grad_norm": 1.8042329549789429,
      "learning_rate": 0.00031761716544325243,
      "loss": 3.5753,
      "step": 9690
    },
    {
      "epoch": 1.8257105213626952,
      "grad_norm": 2.1336681842803955,
      "learning_rate": 0.0003174289478637305,
      "loss": 3.911,
      "step": 9700
    },
    {
      "epoch": 1.8275926971579146,
      "grad_norm": 2.4392285346984863,
      "learning_rate": 0.0003172407302842086,
      "loss": 3.8885,
      "step": 9710
    },
    {
      "epoch": 1.8294748729531338,
      "grad_norm": 2.26499080657959,
      "learning_rate": 0.0003170525127046866,
      "loss": 3.4365,
      "step": 9720
    },
    {
      "epoch": 1.831357048748353,
      "grad_norm": 2.173511266708374,
      "learning_rate": 0.0003168642951251647,
      "loss": 3.5159,
      "step": 9730
    },
    {
      "epoch": 1.8332392245435725,
      "grad_norm": 1.923460841178894,
      "learning_rate": 0.00031667607754564276,
      "loss": 4.0331,
      "step": 9740
    },
    {
      "epoch": 1.8351214003387917,
      "grad_norm": 1.9043619632720947,
      "learning_rate": 0.00031648785996612084,
      "loss": 3.5768,
      "step": 9750
    },
    {
      "epoch": 1.8370035761340109,
      "grad_norm": 2.6198337078094482,
      "learning_rate": 0.0003162996423865989,
      "loss": 3.5863,
      "step": 9760
    },
    {
      "epoch": 1.8388857519292303,
      "grad_norm": 2.625149965286255,
      "learning_rate": 0.000316111424807077,
      "loss": 3.5364,
      "step": 9770
    },
    {
      "epoch": 1.8407679277244493,
      "grad_norm": 2.2209835052490234,
      "learning_rate": 0.00031592320722755507,
      "loss": 3.6741,
      "step": 9780
    },
    {
      "epoch": 1.8426501035196687,
      "grad_norm": 2.7067711353302,
      "learning_rate": 0.00031573498964803314,
      "loss": 3.5254,
      "step": 9790
    },
    {
      "epoch": 1.8445322793148882,
      "grad_norm": 2.327434539794922,
      "learning_rate": 0.00031554677206851116,
      "loss": 3.665,
      "step": 9800
    },
    {
      "epoch": 1.8464144551101072,
      "grad_norm": 2.767713785171509,
      "learning_rate": 0.00031535855448898924,
      "loss": 3.5376,
      "step": 9810
    },
    {
      "epoch": 1.8482966309053266,
      "grad_norm": 2.0066473484039307,
      "learning_rate": 0.0003151703369094673,
      "loss": 3.4858,
      "step": 9820
    },
    {
      "epoch": 1.8501788067005458,
      "grad_norm": 1.8936408758163452,
      "learning_rate": 0.0003149821193299454,
      "loss": 3.4119,
      "step": 9830
    },
    {
      "epoch": 1.852060982495765,
      "grad_norm": 2.1529366970062256,
      "learning_rate": 0.00031479390175042347,
      "loss": 3.5605,
      "step": 9840
    },
    {
      "epoch": 1.8539431582909844,
      "grad_norm": 2.163797378540039,
      "learning_rate": 0.00031460568417090155,
      "loss": 3.755,
      "step": 9850
    },
    {
      "epoch": 1.8558253340862036,
      "grad_norm": 2.1908974647521973,
      "learning_rate": 0.0003144174665913797,
      "loss": 3.4836,
      "step": 9860
    },
    {
      "epoch": 1.8577075098814229,
      "grad_norm": 2.2800731658935547,
      "learning_rate": 0.00031422924901185775,
      "loss": 3.5026,
      "step": 9870
    },
    {
      "epoch": 1.8595896856766423,
      "grad_norm": 2.2017064094543457,
      "learning_rate": 0.00031404103143233583,
      "loss": 3.3399,
      "step": 9880
    },
    {
      "epoch": 1.8614718614718615,
      "grad_norm": 2.403064489364624,
      "learning_rate": 0.00031385281385281385,
      "loss": 3.7772,
      "step": 9890
    },
    {
      "epoch": 1.8633540372670807,
      "grad_norm": 1.948598861694336,
      "learning_rate": 0.00031366459627329193,
      "loss": 3.6041,
      "step": 9900
    },
    {
      "epoch": 1.8652362130623001,
      "grad_norm": 2.0806872844696045,
      "learning_rate": 0.00031347637869377,
      "loss": 3.8162,
      "step": 9910
    },
    {
      "epoch": 1.8671183888575191,
      "grad_norm": 2.627696990966797,
      "learning_rate": 0.0003132881611142481,
      "loss": 3.6738,
      "step": 9920
    },
    {
      "epoch": 1.8690005646527386,
      "grad_norm": 2.0168612003326416,
      "learning_rate": 0.00031309994353472616,
      "loss": 3.6254,
      "step": 9930
    },
    {
      "epoch": 1.870882740447958,
      "grad_norm": 2.210606575012207,
      "learning_rate": 0.00031291172595520423,
      "loss": 3.8663,
      "step": 9940
    },
    {
      "epoch": 1.872764916243177,
      "grad_norm": 1.9852879047393799,
      "learning_rate": 0.0003127235083756823,
      "loss": 3.6297,
      "step": 9950
    },
    {
      "epoch": 1.8746470920383964,
      "grad_norm": 2.120920419692993,
      "learning_rate": 0.0003125352907961604,
      "loss": 3.9663,
      "step": 9960
    },
    {
      "epoch": 1.8765292678336156,
      "grad_norm": 2.316328763961792,
      "learning_rate": 0.00031234707321663846,
      "loss": 3.9639,
      "step": 9970
    },
    {
      "epoch": 1.8784114436288348,
      "grad_norm": 1.8062055110931396,
      "learning_rate": 0.0003121588556371165,
      "loss": 3.8023,
      "step": 9980
    },
    {
      "epoch": 1.8802936194240543,
      "grad_norm": 1.8895719051361084,
      "learning_rate": 0.00031197063805759456,
      "loss": 3.6085,
      "step": 9990
    },
    {
      "epoch": 1.8821757952192735,
      "grad_norm": 1.95108962059021,
      "learning_rate": 0.00031178242047807264,
      "loss": 3.8635,
      "step": 10000
    },
    {
      "epoch": 1.8840579710144927,
      "grad_norm": 1.824989676475525,
      "learning_rate": 0.0003115942028985507,
      "loss": 3.5738,
      "step": 10010
    },
    {
      "epoch": 1.8859401468097121,
      "grad_norm": 1.949899673461914,
      "learning_rate": 0.0003114059853190288,
      "loss": 3.4552,
      "step": 10020
    },
    {
      "epoch": 1.8878223226049313,
      "grad_norm": 2.350801944732666,
      "learning_rate": 0.00031121776773950687,
      "loss": 3.6764,
      "step": 10030
    },
    {
      "epoch": 1.8897044984001505,
      "grad_norm": 1.800048589706421,
      "learning_rate": 0.00031102955015998494,
      "loss": 3.6032,
      "step": 10040
    },
    {
      "epoch": 1.89158667419537,
      "grad_norm": 2.1716933250427246,
      "learning_rate": 0.000310841332580463,
      "loss": 3.5876,
      "step": 10050
    },
    {
      "epoch": 1.8934688499905892,
      "grad_norm": 2.26385760307312,
      "learning_rate": 0.00031065311500094104,
      "loss": 3.6223,
      "step": 10060
    },
    {
      "epoch": 1.8953510257858084,
      "grad_norm": 1.9308747053146362,
      "learning_rate": 0.0003104648974214191,
      "loss": 3.5269,
      "step": 10070
    },
    {
      "epoch": 1.8972332015810278,
      "grad_norm": 2.064584970474243,
      "learning_rate": 0.00031027667984189725,
      "loss": 3.7674,
      "step": 10080
    },
    {
      "epoch": 1.8991153773762468,
      "grad_norm": 2.29477858543396,
      "learning_rate": 0.0003100884622623753,
      "loss": 3.4748,
      "step": 10090
    },
    {
      "epoch": 1.9009975531714662,
      "grad_norm": 2.1337435245513916,
      "learning_rate": 0.0003099002446828534,
      "loss": 3.8851,
      "step": 10100
    },
    {
      "epoch": 1.9028797289666854,
      "grad_norm": 2.0595479011535645,
      "learning_rate": 0.0003097120271033315,
      "loss": 3.5353,
      "step": 10110
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 2.213545560836792,
      "learning_rate": 0.00030952380952380956,
      "loss": 3.5947,
      "step": 10120
    },
    {
      "epoch": 1.906644080557124,
      "grad_norm": 2.010716438293457,
      "learning_rate": 0.00030933559194428763,
      "loss": 3.3873,
      "step": 10130
    },
    {
      "epoch": 1.9085262563523433,
      "grad_norm": 2.145021915435791,
      "learning_rate": 0.0003091473743647657,
      "loss": 3.6004,
      "step": 10140
    },
    {
      "epoch": 1.9104084321475625,
      "grad_norm": 2.155932903289795,
      "learning_rate": 0.00030895915678524373,
      "loss": 3.8732,
      "step": 10150
    },
    {
      "epoch": 1.912290607942782,
      "grad_norm": 2.0870883464813232,
      "learning_rate": 0.0003087709392057218,
      "loss": 3.7121,
      "step": 10160
    },
    {
      "epoch": 1.9141727837380011,
      "grad_norm": 1.889845371246338,
      "learning_rate": 0.0003085827216261999,
      "loss": 3.7184,
      "step": 10170
    },
    {
      "epoch": 1.9160549595332204,
      "grad_norm": 2.0182039737701416,
      "learning_rate": 0.00030839450404667796,
      "loss": 3.7952,
      "step": 10180
    },
    {
      "epoch": 1.9179371353284398,
      "grad_norm": 1.9103964567184448,
      "learning_rate": 0.00030820628646715604,
      "loss": 3.546,
      "step": 10190
    },
    {
      "epoch": 1.919819311123659,
      "grad_norm": 1.934048056602478,
      "learning_rate": 0.0003080180688876341,
      "loss": 3.4859,
      "step": 10200
    },
    {
      "epoch": 1.9217014869188782,
      "grad_norm": 2.16024112701416,
      "learning_rate": 0.0003078298513081122,
      "loss": 3.3423,
      "step": 10210
    },
    {
      "epoch": 1.9235836627140976,
      "grad_norm": 1.946714997291565,
      "learning_rate": 0.00030764163372859027,
      "loss": 3.6281,
      "step": 10220
    },
    {
      "epoch": 1.9254658385093166,
      "grad_norm": 2.031848907470703,
      "learning_rate": 0.00030745341614906834,
      "loss": 3.6591,
      "step": 10230
    },
    {
      "epoch": 1.927348014304536,
      "grad_norm": 2.2257654666900635,
      "learning_rate": 0.00030726519856954636,
      "loss": 3.6939,
      "step": 10240
    },
    {
      "epoch": 1.9292301900997553,
      "grad_norm": 2.163264274597168,
      "learning_rate": 0.00030707698099002444,
      "loss": 3.6587,
      "step": 10250
    },
    {
      "epoch": 1.9311123658949745,
      "grad_norm": 2.368788719177246,
      "learning_rate": 0.0003068887634105025,
      "loss": 3.774,
      "step": 10260
    },
    {
      "epoch": 1.932994541690194,
      "grad_norm": 1.9467861652374268,
      "learning_rate": 0.0003067005458309806,
      "loss": 3.6306,
      "step": 10270
    },
    {
      "epoch": 1.9348767174854131,
      "grad_norm": 1.9181021451950073,
      "learning_rate": 0.00030651232825145867,
      "loss": 3.6298,
      "step": 10280
    },
    {
      "epoch": 1.9367588932806323,
      "grad_norm": 1.849297285079956,
      "learning_rate": 0.00030632411067193675,
      "loss": 3.641,
      "step": 10290
    },
    {
      "epoch": 1.9386410690758518,
      "grad_norm": 1.93975830078125,
      "learning_rate": 0.0003061358930924149,
      "loss": 3.6084,
      "step": 10300
    },
    {
      "epoch": 1.940523244871071,
      "grad_norm": 1.8382394313812256,
      "learning_rate": 0.00030594767551289295,
      "loss": 3.4179,
      "step": 10310
    },
    {
      "epoch": 1.9424054206662902,
      "grad_norm": 2.686518907546997,
      "learning_rate": 0.00030575945793337103,
      "loss": 3.7513,
      "step": 10320
    },
    {
      "epoch": 1.9442875964615096,
      "grad_norm": 2.0112316608428955,
      "learning_rate": 0.00030557124035384905,
      "loss": 3.7188,
      "step": 10330
    },
    {
      "epoch": 1.9461697722567288,
      "grad_norm": 1.9542816877365112,
      "learning_rate": 0.00030538302277432713,
      "loss": 3.5804,
      "step": 10340
    },
    {
      "epoch": 1.948051948051948,
      "grad_norm": 1.8192522525787354,
      "learning_rate": 0.0003051948051948052,
      "loss": 3.4211,
      "step": 10350
    },
    {
      "epoch": 1.9499341238471675,
      "grad_norm": 2.1841719150543213,
      "learning_rate": 0.0003050065876152833,
      "loss": 3.5976,
      "step": 10360
    },
    {
      "epoch": 1.9518162996423865,
      "grad_norm": 2.133845090866089,
      "learning_rate": 0.00030481837003576136,
      "loss": 3.5101,
      "step": 10370
    },
    {
      "epoch": 1.9536984754376059,
      "grad_norm": 2.269397497177124,
      "learning_rate": 0.00030463015245623943,
      "loss": 3.7982,
      "step": 10380
    },
    {
      "epoch": 1.955580651232825,
      "grad_norm": 2.094409704208374,
      "learning_rate": 0.0003044419348767175,
      "loss": 3.4589,
      "step": 10390
    },
    {
      "epoch": 1.9574628270280443,
      "grad_norm": 2.1520373821258545,
      "learning_rate": 0.0003042537172971956,
      "loss": 3.6693,
      "step": 10400
    },
    {
      "epoch": 1.9593450028232637,
      "grad_norm": 1.810929298400879,
      "learning_rate": 0.0003040654997176736,
      "loss": 3.8383,
      "step": 10410
    },
    {
      "epoch": 1.961227178618483,
      "grad_norm": 2.058192491531372,
      "learning_rate": 0.0003038772821381517,
      "loss": 3.5365,
      "step": 10420
    },
    {
      "epoch": 1.9631093544137022,
      "grad_norm": 2.311635732650757,
      "learning_rate": 0.00030368906455862976,
      "loss": 3.7174,
      "step": 10430
    },
    {
      "epoch": 1.9649915302089216,
      "grad_norm": 2.0313923358917236,
      "learning_rate": 0.00030350084697910784,
      "loss": 3.4804,
      "step": 10440
    },
    {
      "epoch": 1.9668737060041408,
      "grad_norm": 2.259661912918091,
      "learning_rate": 0.0003033126293995859,
      "loss": 3.8336,
      "step": 10450
    },
    {
      "epoch": 1.96875588179936,
      "grad_norm": 2.002938985824585,
      "learning_rate": 0.000303124411820064,
      "loss": 3.6662,
      "step": 10460
    },
    {
      "epoch": 1.9706380575945794,
      "grad_norm": 2.394352912902832,
      "learning_rate": 0.00030293619424054207,
      "loss": 3.6835,
      "step": 10470
    },
    {
      "epoch": 1.9725202333897986,
      "grad_norm": 2.634260892868042,
      "learning_rate": 0.00030274797666102014,
      "loss": 3.4938,
      "step": 10480
    },
    {
      "epoch": 1.9744024091850179,
      "grad_norm": 1.8893793821334839,
      "learning_rate": 0.0003025597590814982,
      "loss": 3.6379,
      "step": 10490
    },
    {
      "epoch": 1.9762845849802373,
      "grad_norm": 2.149104356765747,
      "learning_rate": 0.00030237154150197624,
      "loss": 4.076,
      "step": 10500
    },
    {
      "epoch": 1.9781667607754563,
      "grad_norm": 1.9862107038497925,
      "learning_rate": 0.0003021833239224543,
      "loss": 3.7925,
      "step": 10510
    },
    {
      "epoch": 1.9800489365706757,
      "grad_norm": 2.4172115325927734,
      "learning_rate": 0.00030199510634293245,
      "loss": 3.4149,
      "step": 10520
    },
    {
      "epoch": 1.981931112365895,
      "grad_norm": 2.012948751449585,
      "learning_rate": 0.0003018068887634105,
      "loss": 3.7194,
      "step": 10530
    },
    {
      "epoch": 1.9838132881611141,
      "grad_norm": 2.0994248390197754,
      "learning_rate": 0.0003016186711838886,
      "loss": 3.7299,
      "step": 10540
    },
    {
      "epoch": 1.9856954639563336,
      "grad_norm": 1.8276288509368896,
      "learning_rate": 0.0003014304536043667,
      "loss": 3.5345,
      "step": 10550
    },
    {
      "epoch": 1.9875776397515528,
      "grad_norm": 2.070876359939575,
      "learning_rate": 0.00030124223602484476,
      "loss": 3.8717,
      "step": 10560
    },
    {
      "epoch": 1.989459815546772,
      "grad_norm": 2.233039140701294,
      "learning_rate": 0.00030105401844532283,
      "loss": 3.4603,
      "step": 10570
    },
    {
      "epoch": 1.9913419913419914,
      "grad_norm": 2.3521862030029297,
      "learning_rate": 0.0003008658008658009,
      "loss": 3.6113,
      "step": 10580
    },
    {
      "epoch": 1.9932241671372106,
      "grad_norm": 3.8352150917053223,
      "learning_rate": 0.00030067758328627893,
      "loss": 3.8694,
      "step": 10590
    },
    {
      "epoch": 1.9951063429324298,
      "grad_norm": 2.0548229217529297,
      "learning_rate": 0.000300489365706757,
      "loss": 3.4568,
      "step": 10600
    },
    {
      "epoch": 1.9969885187276493,
      "grad_norm": 1.9041173458099365,
      "learning_rate": 0.0003003011481272351,
      "loss": 3.569,
      "step": 10610
    },
    {
      "epoch": 1.9988706945228685,
      "grad_norm": 2.2218592166900635,
      "learning_rate": 0.00030011293054771316,
      "loss": 3.5093,
      "step": 10620
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.1472,
      "eval_loss": 3.5664186477661133,
      "eval_runtime": 82.3517,
      "eval_samples_per_second": 91.073,
      "eval_steps_per_second": 11.39,
      "step": 10626
    },
    {
      "epoch": 2.0007528703180877,
      "grad_norm": 2.0325100421905518,
      "learning_rate": 0.00029992471296819124,
      "loss": 3.6207,
      "step": 10630
    },
    {
      "epoch": 2.002635046113307,
      "grad_norm": 1.9634112119674683,
      "learning_rate": 0.0002997364953886693,
      "loss": 3.8266,
      "step": 10640
    },
    {
      "epoch": 2.004517221908526,
      "grad_norm": 2.0167596340179443,
      "learning_rate": 0.0002995482778091474,
      "loss": 3.668,
      "step": 10650
    },
    {
      "epoch": 2.0063993977037455,
      "grad_norm": 1.9793752431869507,
      "learning_rate": 0.00029936006022962547,
      "loss": 3.7249,
      "step": 10660
    },
    {
      "epoch": 2.008281573498965,
      "grad_norm": 1.982495665550232,
      "learning_rate": 0.0002991718426501035,
      "loss": 3.7413,
      "step": 10670
    },
    {
      "epoch": 2.010163749294184,
      "grad_norm": 2.1777305603027344,
      "learning_rate": 0.00029898362507058156,
      "loss": 3.7726,
      "step": 10680
    },
    {
      "epoch": 2.0120459250894034,
      "grad_norm": 2.2762796878814697,
      "learning_rate": 0.00029879540749105964,
      "loss": 3.6714,
      "step": 10690
    },
    {
      "epoch": 2.013928100884623,
      "grad_norm": 1.967294692993164,
      "learning_rate": 0.0002986071899115377,
      "loss": 3.6496,
      "step": 10700
    },
    {
      "epoch": 2.015810276679842,
      "grad_norm": 2.13759708404541,
      "learning_rate": 0.0002984189723320158,
      "loss": 3.5799,
      "step": 10710
    },
    {
      "epoch": 2.0176924524750612,
      "grad_norm": 2.114488124847412,
      "learning_rate": 0.00029823075475249387,
      "loss": 3.5195,
      "step": 10720
    },
    {
      "epoch": 2.0195746282702802,
      "grad_norm": 1.7854301929473877,
      "learning_rate": 0.00029804253717297195,
      "loss": 3.5987,
      "step": 10730
    },
    {
      "epoch": 2.0214568040654997,
      "grad_norm": 2.2474656105041504,
      "learning_rate": 0.0002978543195934501,
      "loss": 3.6328,
      "step": 10740
    },
    {
      "epoch": 2.023338979860719,
      "grad_norm": 1.9125057458877563,
      "learning_rate": 0.00029766610201392815,
      "loss": 3.5598,
      "step": 10750
    },
    {
      "epoch": 2.025221155655938,
      "grad_norm": 2.0102808475494385,
      "learning_rate": 0.0002974778844344062,
      "loss": 3.5107,
      "step": 10760
    },
    {
      "epoch": 2.0271033314511575,
      "grad_norm": 2.132345676422119,
      "learning_rate": 0.00029728966685488425,
      "loss": 3.5667,
      "step": 10770
    },
    {
      "epoch": 2.028985507246377,
      "grad_norm": 3.9572272300720215,
      "learning_rate": 0.00029710144927536233,
      "loss": 3.3139,
      "step": 10780
    },
    {
      "epoch": 2.030867683041596,
      "grad_norm": 2.37929105758667,
      "learning_rate": 0.0002969132316958404,
      "loss": 3.8562,
      "step": 10790
    },
    {
      "epoch": 2.0327498588368154,
      "grad_norm": 2.9635579586029053,
      "learning_rate": 0.0002967250141163185,
      "loss": 3.5757,
      "step": 10800
    },
    {
      "epoch": 2.034632034632035,
      "grad_norm": 1.985288381576538,
      "learning_rate": 0.00029653679653679656,
      "loss": 3.5637,
      "step": 10810
    },
    {
      "epoch": 2.036514210427254,
      "grad_norm": 2.2944083213806152,
      "learning_rate": 0.00029634857895727463,
      "loss": 3.6574,
      "step": 10820
    },
    {
      "epoch": 2.038396386222473,
      "grad_norm": 2.4931163787841797,
      "learning_rate": 0.0002961603613777527,
      "loss": 3.3834,
      "step": 10830
    },
    {
      "epoch": 2.0402785620176926,
      "grad_norm": 2.0097668170928955,
      "learning_rate": 0.0002959721437982308,
      "loss": 3.4972,
      "step": 10840
    },
    {
      "epoch": 2.0421607378129116,
      "grad_norm": 2.063840627670288,
      "learning_rate": 0.0002957839262187088,
      "loss": 3.7874,
      "step": 10850
    },
    {
      "epoch": 2.044042913608131,
      "grad_norm": 2.0522704124450684,
      "learning_rate": 0.0002955957086391869,
      "loss": 3.5479,
      "step": 10860
    },
    {
      "epoch": 2.04592508940335,
      "grad_norm": 2.0130808353424072,
      "learning_rate": 0.00029540749105966496,
      "loss": 3.8158,
      "step": 10870
    },
    {
      "epoch": 2.0478072651985695,
      "grad_norm": 2.5037670135498047,
      "learning_rate": 0.00029521927348014304,
      "loss": 3.8272,
      "step": 10880
    },
    {
      "epoch": 2.049689440993789,
      "grad_norm": 2.120812177658081,
      "learning_rate": 0.0002950310559006211,
      "loss": 3.6263,
      "step": 10890
    },
    {
      "epoch": 2.051571616789008,
      "grad_norm": 2.170079469680786,
      "learning_rate": 0.0002948428383210992,
      "loss": 3.7242,
      "step": 10900
    },
    {
      "epoch": 2.0534537925842273,
      "grad_norm": 1.8562462329864502,
      "learning_rate": 0.00029465462074157727,
      "loss": 3.6286,
      "step": 10910
    },
    {
      "epoch": 2.0553359683794468,
      "grad_norm": 2.1779134273529053,
      "learning_rate": 0.00029446640316205534,
      "loss": 3.5787,
      "step": 10920
    },
    {
      "epoch": 2.0572181441746658,
      "grad_norm": 2.044370651245117,
      "learning_rate": 0.0002942781855825334,
      "loss": 3.6205,
      "step": 10930
    },
    {
      "epoch": 2.059100319969885,
      "grad_norm": 2.4833273887634277,
      "learning_rate": 0.00029408996800301144,
      "loss": 3.4728,
      "step": 10940
    },
    {
      "epoch": 2.0609824957651046,
      "grad_norm": 2.2788689136505127,
      "learning_rate": 0.0002939017504234895,
      "loss": 3.5166,
      "step": 10950
    },
    {
      "epoch": 2.0628646715603236,
      "grad_norm": 2.383826971054077,
      "learning_rate": 0.00029371353284396765,
      "loss": 3.7058,
      "step": 10960
    },
    {
      "epoch": 2.064746847355543,
      "grad_norm": 2.5065038204193115,
      "learning_rate": 0.0002935253152644457,
      "loss": 3.4283,
      "step": 10970
    },
    {
      "epoch": 2.0666290231507625,
      "grad_norm": 2.153620958328247,
      "learning_rate": 0.0002933370976849238,
      "loss": 3.3929,
      "step": 10980
    },
    {
      "epoch": 2.0685111989459815,
      "grad_norm": 2.111159563064575,
      "learning_rate": 0.0002931488801054019,
      "loss": 3.3886,
      "step": 10990
    },
    {
      "epoch": 2.070393374741201,
      "grad_norm": 2.286848545074463,
      "learning_rate": 0.00029296066252587996,
      "loss": 3.6433,
      "step": 11000
    },
    {
      "epoch": 2.0722755505364203,
      "grad_norm": 2.1899757385253906,
      "learning_rate": 0.00029277244494635803,
      "loss": 3.8215,
      "step": 11010
    },
    {
      "epoch": 2.0741577263316393,
      "grad_norm": 2.477233409881592,
      "learning_rate": 0.00029258422736683605,
      "loss": 3.7537,
      "step": 11020
    },
    {
      "epoch": 2.0760399021268587,
      "grad_norm": 2.1020686626434326,
      "learning_rate": 0.00029239600978731413,
      "loss": 3.6285,
      "step": 11030
    },
    {
      "epoch": 2.0779220779220777,
      "grad_norm": 1.9838166236877441,
      "learning_rate": 0.0002922077922077922,
      "loss": 3.8239,
      "step": 11040
    },
    {
      "epoch": 2.079804253717297,
      "grad_norm": 2.234279155731201,
      "learning_rate": 0.0002920195746282703,
      "loss": 3.5579,
      "step": 11050
    },
    {
      "epoch": 2.0816864295125166,
      "grad_norm": 2.4791712760925293,
      "learning_rate": 0.00029183135704874836,
      "loss": 3.7128,
      "step": 11060
    },
    {
      "epoch": 2.0835686053077356,
      "grad_norm": 2.0139009952545166,
      "learning_rate": 0.00029164313946922644,
      "loss": 3.497,
      "step": 11070
    },
    {
      "epoch": 2.085450781102955,
      "grad_norm": 1.918221354484558,
      "learning_rate": 0.0002914549218897045,
      "loss": 3.5534,
      "step": 11080
    },
    {
      "epoch": 2.0873329568981744,
      "grad_norm": 2.0779190063476562,
      "learning_rate": 0.0002912667043101826,
      "loss": 3.6238,
      "step": 11090
    },
    {
      "epoch": 2.0892151326933934,
      "grad_norm": 2.1959855556488037,
      "learning_rate": 0.00029107848673066067,
      "loss": 3.5225,
      "step": 11100
    },
    {
      "epoch": 2.091097308488613,
      "grad_norm": 1.7541618347167969,
      "learning_rate": 0.0002908902691511387,
      "loss": 3.7214,
      "step": 11110
    },
    {
      "epoch": 2.0929794842838323,
      "grad_norm": 1.9482086896896362,
      "learning_rate": 0.00029070205157161676,
      "loss": 3.4,
      "step": 11120
    },
    {
      "epoch": 2.0948616600790513,
      "grad_norm": 1.9692625999450684,
      "learning_rate": 0.00029051383399209484,
      "loss": 3.643,
      "step": 11130
    },
    {
      "epoch": 2.0967438358742707,
      "grad_norm": 2.198240280151367,
      "learning_rate": 0.0002903256164125729,
      "loss": 3.4483,
      "step": 11140
    },
    {
      "epoch": 2.09862601166949,
      "grad_norm": 1.9893345832824707,
      "learning_rate": 0.000290137398833051,
      "loss": 3.7261,
      "step": 11150
    },
    {
      "epoch": 2.100508187464709,
      "grad_norm": 2.090209484100342,
      "learning_rate": 0.00028994918125352907,
      "loss": 3.3484,
      "step": 11160
    },
    {
      "epoch": 2.1023903632599286,
      "grad_norm": 2.24429988861084,
      "learning_rate": 0.00028976096367400715,
      "loss": 3.604,
      "step": 11170
    },
    {
      "epoch": 2.1042725390551476,
      "grad_norm": 2.234273910522461,
      "learning_rate": 0.0002895727460944853,
      "loss": 3.4037,
      "step": 11180
    },
    {
      "epoch": 2.106154714850367,
      "grad_norm": 2.5589098930358887,
      "learning_rate": 0.00028938452851496335,
      "loss": 3.443,
      "step": 11190
    },
    {
      "epoch": 2.1080368906455864,
      "grad_norm": 2.2039730548858643,
      "learning_rate": 0.0002891963109354414,
      "loss": 3.8742,
      "step": 11200
    },
    {
      "epoch": 2.1099190664408054,
      "grad_norm": 2.143895149230957,
      "learning_rate": 0.00028900809335591945,
      "loss": 3.5561,
      "step": 11210
    },
    {
      "epoch": 2.111801242236025,
      "grad_norm": 2.077723741531372,
      "learning_rate": 0.00028881987577639753,
      "loss": 3.3852,
      "step": 11220
    },
    {
      "epoch": 2.1136834180312443,
      "grad_norm": 2.2736127376556396,
      "learning_rate": 0.0002886316581968756,
      "loss": 3.4958,
      "step": 11230
    },
    {
      "epoch": 2.1155655938264633,
      "grad_norm": 2.400620937347412,
      "learning_rate": 0.0002884434406173537,
      "loss": 3.785,
      "step": 11240
    },
    {
      "epoch": 2.1174477696216827,
      "grad_norm": 2.3708560466766357,
      "learning_rate": 0.00028825522303783176,
      "loss": 3.6247,
      "step": 11250
    },
    {
      "epoch": 2.119329945416902,
      "grad_norm": 2.456202983856201,
      "learning_rate": 0.00028806700545830983,
      "loss": 3.5149,
      "step": 11260
    },
    {
      "epoch": 2.121212121212121,
      "grad_norm": 2.273723840713501,
      "learning_rate": 0.0002878787878787879,
      "loss": 3.6566,
      "step": 11270
    },
    {
      "epoch": 2.1230942970073405,
      "grad_norm": 2.1427016258239746,
      "learning_rate": 0.00028769057029926593,
      "loss": 3.8057,
      "step": 11280
    },
    {
      "epoch": 2.12497647280256,
      "grad_norm": 2.0322492122650146,
      "learning_rate": 0.000287502352719744,
      "loss": 3.7683,
      "step": 11290
    },
    {
      "epoch": 2.126858648597779,
      "grad_norm": 2.18509578704834,
      "learning_rate": 0.0002873141351402221,
      "loss": 3.7631,
      "step": 11300
    },
    {
      "epoch": 2.1287408243929984,
      "grad_norm": 2.1193268299102783,
      "learning_rate": 0.00028712591756070016,
      "loss": 3.5023,
      "step": 11310
    },
    {
      "epoch": 2.1306230001882174,
      "grad_norm": 1.880686640739441,
      "learning_rate": 0.00028693769998117824,
      "loss": 3.6056,
      "step": 11320
    },
    {
      "epoch": 2.132505175983437,
      "grad_norm": 2.0465972423553467,
      "learning_rate": 0.0002867494824016563,
      "loss": 3.9191,
      "step": 11330
    },
    {
      "epoch": 2.1343873517786562,
      "grad_norm": 2.2074384689331055,
      "learning_rate": 0.0002865612648221344,
      "loss": 3.4342,
      "step": 11340
    },
    {
      "epoch": 2.1362695275738752,
      "grad_norm": 2.1300830841064453,
      "learning_rate": 0.00028637304724261247,
      "loss": 3.5027,
      "step": 11350
    },
    {
      "epoch": 2.1381517033690947,
      "grad_norm": 2.1655023097991943,
      "learning_rate": 0.00028618482966309054,
      "loss": 3.8552,
      "step": 11360
    },
    {
      "epoch": 2.140033879164314,
      "grad_norm": 2.5276312828063965,
      "learning_rate": 0.00028599661208356857,
      "loss": 3.5268,
      "step": 11370
    },
    {
      "epoch": 2.141916054959533,
      "grad_norm": 2.214873790740967,
      "learning_rate": 0.00028580839450404664,
      "loss": 3.3984,
      "step": 11380
    },
    {
      "epoch": 2.1437982307547525,
      "grad_norm": 2.229903221130371,
      "learning_rate": 0.0002856201769245247,
      "loss": 3.8769,
      "step": 11390
    },
    {
      "epoch": 2.145680406549972,
      "grad_norm": 2.1930768489837646,
      "learning_rate": 0.0002854319593450028,
      "loss": 3.8591,
      "step": 11400
    },
    {
      "epoch": 2.147562582345191,
      "grad_norm": 2.090045690536499,
      "learning_rate": 0.0002852437417654809,
      "loss": 3.4975,
      "step": 11410
    },
    {
      "epoch": 2.1494447581404104,
      "grad_norm": 2.0409741401672363,
      "learning_rate": 0.000285055524185959,
      "loss": 3.3442,
      "step": 11420
    },
    {
      "epoch": 2.15132693393563,
      "grad_norm": 2.0023446083068848,
      "learning_rate": 0.0002848673066064371,
      "loss": 3.7851,
      "step": 11430
    },
    {
      "epoch": 2.153209109730849,
      "grad_norm": 2.3585290908813477,
      "learning_rate": 0.00028467908902691516,
      "loss": 3.6652,
      "step": 11440
    },
    {
      "epoch": 2.155091285526068,
      "grad_norm": 1.9051262140274048,
      "learning_rate": 0.00028449087144739323,
      "loss": 3.6001,
      "step": 11450
    },
    {
      "epoch": 2.1569734613212876,
      "grad_norm": 2.1293647289276123,
      "learning_rate": 0.00028430265386787125,
      "loss": 3.8165,
      "step": 11460
    },
    {
      "epoch": 2.1588556371165066,
      "grad_norm": 1.993503451347351,
      "learning_rate": 0.00028411443628834933,
      "loss": 3.692,
      "step": 11470
    },
    {
      "epoch": 2.160737812911726,
      "grad_norm": 2.0621178150177,
      "learning_rate": 0.0002839262187088274,
      "loss": 3.7678,
      "step": 11480
    },
    {
      "epoch": 2.162619988706945,
      "grad_norm": 2.0874807834625244,
      "learning_rate": 0.0002837380011293055,
      "loss": 3.6538,
      "step": 11490
    },
    {
      "epoch": 2.1645021645021645,
      "grad_norm": 2.309434413909912,
      "learning_rate": 0.00028354978354978356,
      "loss": 3.5588,
      "step": 11500
    },
    {
      "epoch": 2.166384340297384,
      "grad_norm": 2.163496494293213,
      "learning_rate": 0.00028336156597026164,
      "loss": 3.5295,
      "step": 11510
    },
    {
      "epoch": 2.168266516092603,
      "grad_norm": 2.23266863822937,
      "learning_rate": 0.0002831733483907397,
      "loss": 3.3534,
      "step": 11520
    },
    {
      "epoch": 2.1701486918878223,
      "grad_norm": 2.2005791664123535,
      "learning_rate": 0.0002829851308112178,
      "loss": 3.4239,
      "step": 11530
    },
    {
      "epoch": 2.1720308676830418,
      "grad_norm": 2.8036704063415527,
      "learning_rate": 0.00028279691323169587,
      "loss": 3.6307,
      "step": 11540
    },
    {
      "epoch": 2.1739130434782608,
      "grad_norm": 2.1723809242248535,
      "learning_rate": 0.0002826086956521739,
      "loss": 3.5663,
      "step": 11550
    },
    {
      "epoch": 2.17579521927348,
      "grad_norm": 2.3068792819976807,
      "learning_rate": 0.00028242047807265196,
      "loss": 3.6506,
      "step": 11560
    },
    {
      "epoch": 2.1776773950686996,
      "grad_norm": 2.3664402961730957,
      "learning_rate": 0.00028223226049313004,
      "loss": 3.9696,
      "step": 11570
    },
    {
      "epoch": 2.1795595708639186,
      "grad_norm": 2.1242728233337402,
      "learning_rate": 0.0002820440429136081,
      "loss": 3.6441,
      "step": 11580
    },
    {
      "epoch": 2.181441746659138,
      "grad_norm": 1.8504153490066528,
      "learning_rate": 0.0002818558253340862,
      "loss": 3.2865,
      "step": 11590
    },
    {
      "epoch": 2.183323922454357,
      "grad_norm": 2.618985176086426,
      "learning_rate": 0.00028166760775456427,
      "loss": 3.7115,
      "step": 11600
    },
    {
      "epoch": 2.1852060982495765,
      "grad_norm": 2.038625478744507,
      "learning_rate": 0.00028147939017504235,
      "loss": 3.4573,
      "step": 11610
    },
    {
      "epoch": 2.187088274044796,
      "grad_norm": 2.480666160583496,
      "learning_rate": 0.0002812911725955205,
      "loss": 3.8504,
      "step": 11620
    },
    {
      "epoch": 2.188970449840015,
      "grad_norm": 2.2521214485168457,
      "learning_rate": 0.0002811029550159985,
      "loss": 3.614,
      "step": 11630
    },
    {
      "epoch": 2.1908526256352343,
      "grad_norm": 3.09513258934021,
      "learning_rate": 0.0002809147374364766,
      "loss": 3.8761,
      "step": 11640
    },
    {
      "epoch": 2.1927348014304537,
      "grad_norm": 2.276827335357666,
      "learning_rate": 0.00028072651985695465,
      "loss": 3.3449,
      "step": 11650
    },
    {
      "epoch": 2.1946169772256727,
      "grad_norm": 2.4314024448394775,
      "learning_rate": 0.00028053830227743273,
      "loss": 3.6035,
      "step": 11660
    },
    {
      "epoch": 2.196499153020892,
      "grad_norm": 2.1677989959716797,
      "learning_rate": 0.0002803500846979108,
      "loss": 3.7463,
      "step": 11670
    },
    {
      "epoch": 2.1983813288161116,
      "grad_norm": 2.4452247619628906,
      "learning_rate": 0.0002801618671183889,
      "loss": 3.976,
      "step": 11680
    },
    {
      "epoch": 2.2002635046113306,
      "grad_norm": 2.3938846588134766,
      "learning_rate": 0.00027997364953886696,
      "loss": 3.696,
      "step": 11690
    },
    {
      "epoch": 2.20214568040655,
      "grad_norm": 2.150054454803467,
      "learning_rate": 0.00027978543195934503,
      "loss": 3.4701,
      "step": 11700
    },
    {
      "epoch": 2.2040278562017694,
      "grad_norm": 2.1537985801696777,
      "learning_rate": 0.0002795972143798231,
      "loss": 3.63,
      "step": 11710
    },
    {
      "epoch": 2.2059100319969884,
      "grad_norm": 2.190314292907715,
      "learning_rate": 0.00027940899680030113,
      "loss": 3.4569,
      "step": 11720
    },
    {
      "epoch": 2.207792207792208,
      "grad_norm": 2.0873048305511475,
      "learning_rate": 0.0002792207792207792,
      "loss": 3.5355,
      "step": 11730
    },
    {
      "epoch": 2.2096743835874273,
      "grad_norm": 2.0455803871154785,
      "learning_rate": 0.0002790325616412573,
      "loss": 3.1505,
      "step": 11740
    },
    {
      "epoch": 2.2115565593826463,
      "grad_norm": 2.204108953475952,
      "learning_rate": 0.00027884434406173536,
      "loss": 3.523,
      "step": 11750
    },
    {
      "epoch": 2.2134387351778657,
      "grad_norm": 2.099332332611084,
      "learning_rate": 0.00027865612648221344,
      "loss": 3.6056,
      "step": 11760
    },
    {
      "epoch": 2.2153209109730847,
      "grad_norm": 2.1077349185943604,
      "learning_rate": 0.0002784679089026915,
      "loss": 3.6528,
      "step": 11770
    },
    {
      "epoch": 2.217203086768304,
      "grad_norm": 2.1245169639587402,
      "learning_rate": 0.0002782796913231696,
      "loss": 3.5215,
      "step": 11780
    },
    {
      "epoch": 2.2190852625635236,
      "grad_norm": 2.1498658657073975,
      "learning_rate": 0.00027809147374364767,
      "loss": 3.814,
      "step": 11790
    },
    {
      "epoch": 2.2209674383587426,
      "grad_norm": 2.0961034297943115,
      "learning_rate": 0.00027790325616412574,
      "loss": 3.3516,
      "step": 11800
    },
    {
      "epoch": 2.222849614153962,
      "grad_norm": 2.0248188972473145,
      "learning_rate": 0.00027771503858460377,
      "loss": 3.7715,
      "step": 11810
    },
    {
      "epoch": 2.2247317899491814,
      "grad_norm": 2.018819570541382,
      "learning_rate": 0.00027752682100508184,
      "loss": 3.645,
      "step": 11820
    },
    {
      "epoch": 2.2266139657444004,
      "grad_norm": 1.9713603258132935,
      "learning_rate": 0.0002773386034255599,
      "loss": 3.4662,
      "step": 11830
    },
    {
      "epoch": 2.22849614153962,
      "grad_norm": 2.605026960372925,
      "learning_rate": 0.000277150385846038,
      "loss": 3.5945,
      "step": 11840
    },
    {
      "epoch": 2.2303783173348393,
      "grad_norm": 1.9455453157424927,
      "learning_rate": 0.0002769621682665161,
      "loss": 3.7957,
      "step": 11850
    },
    {
      "epoch": 2.2322604931300583,
      "grad_norm": 2.095759630203247,
      "learning_rate": 0.0002767739506869942,
      "loss": 3.3663,
      "step": 11860
    },
    {
      "epoch": 2.2341426689252777,
      "grad_norm": 2.201913595199585,
      "learning_rate": 0.0002765857331074723,
      "loss": 3.3748,
      "step": 11870
    },
    {
      "epoch": 2.2360248447204967,
      "grad_norm": 2.163905143737793,
      "learning_rate": 0.00027639751552795036,
      "loss": 3.7517,
      "step": 11880
    },
    {
      "epoch": 2.237907020515716,
      "grad_norm": 2.3969902992248535,
      "learning_rate": 0.0002762092979484284,
      "loss": 3.7489,
      "step": 11890
    },
    {
      "epoch": 2.2397891963109355,
      "grad_norm": 1.9157522916793823,
      "learning_rate": 0.00027602108036890645,
      "loss": 3.4607,
      "step": 11900
    },
    {
      "epoch": 2.2416713721061545,
      "grad_norm": 2.334146499633789,
      "learning_rate": 0.00027583286278938453,
      "loss": 3.7616,
      "step": 11910
    },
    {
      "epoch": 2.243553547901374,
      "grad_norm": 2.0595529079437256,
      "learning_rate": 0.0002756446452098626,
      "loss": 3.5452,
      "step": 11920
    },
    {
      "epoch": 2.2454357236965934,
      "grad_norm": 2.5498175621032715,
      "learning_rate": 0.0002754564276303407,
      "loss": 3.3551,
      "step": 11930
    },
    {
      "epoch": 2.2473178994918124,
      "grad_norm": 2.1727492809295654,
      "learning_rate": 0.00027526821005081876,
      "loss": 3.2736,
      "step": 11940
    },
    {
      "epoch": 2.249200075287032,
      "grad_norm": 2.6587021350860596,
      "learning_rate": 0.00027507999247129684,
      "loss": 3.5636,
      "step": 11950
    },
    {
      "epoch": 2.2510822510822512,
      "grad_norm": 2.283400297164917,
      "learning_rate": 0.0002748917748917749,
      "loss": 3.5174,
      "step": 11960
    },
    {
      "epoch": 2.2529644268774702,
      "grad_norm": 2.0880675315856934,
      "learning_rate": 0.000274703557312253,
      "loss": 3.474,
      "step": 11970
    },
    {
      "epoch": 2.2548466026726897,
      "grad_norm": 2.504103660583496,
      "learning_rate": 0.000274515339732731,
      "loss": 3.5673,
      "step": 11980
    },
    {
      "epoch": 2.256728778467909,
      "grad_norm": 2.2468066215515137,
      "learning_rate": 0.0002743271221532091,
      "loss": 3.3428,
      "step": 11990
    },
    {
      "epoch": 2.258610954263128,
      "grad_norm": 2.028759717941284,
      "learning_rate": 0.00027413890457368716,
      "loss": 3.6325,
      "step": 12000
    },
    {
      "epoch": 2.2604931300583475,
      "grad_norm": 2.020503044128418,
      "learning_rate": 0.00027395068699416524,
      "loss": 3.5208,
      "step": 12010
    },
    {
      "epoch": 2.262375305853567,
      "grad_norm": 2.351266384124756,
      "learning_rate": 0.0002737624694146433,
      "loss": 3.5115,
      "step": 12020
    },
    {
      "epoch": 2.264257481648786,
      "grad_norm": 2.273869752883911,
      "learning_rate": 0.0002735742518351214,
      "loss": 3.685,
      "step": 12030
    },
    {
      "epoch": 2.2661396574440054,
      "grad_norm": 2.696059465408325,
      "learning_rate": 0.00027338603425559947,
      "loss": 3.6462,
      "step": 12040
    },
    {
      "epoch": 2.2680218332392243,
      "grad_norm": 1.843395709991455,
      "learning_rate": 0.00027319781667607755,
      "loss": 3.3474,
      "step": 12050
    },
    {
      "epoch": 2.269904009034444,
      "grad_norm": 2.270089864730835,
      "learning_rate": 0.0002730095990965556,
      "loss": 3.4742,
      "step": 12060
    },
    {
      "epoch": 2.271786184829663,
      "grad_norm": 2.3248908519744873,
      "learning_rate": 0.0002728213815170337,
      "loss": 3.4222,
      "step": 12070
    },
    {
      "epoch": 2.273668360624882,
      "grad_norm": 2.705111026763916,
      "learning_rate": 0.0002726331639375118,
      "loss": 3.5915,
      "step": 12080
    },
    {
      "epoch": 2.2755505364201016,
      "grad_norm": 2.456393003463745,
      "learning_rate": 0.00027244494635798985,
      "loss": 3.695,
      "step": 12090
    },
    {
      "epoch": 2.277432712215321,
      "grad_norm": 1.988847017288208,
      "learning_rate": 0.00027225672877846793,
      "loss": 3.54,
      "step": 12100
    },
    {
      "epoch": 2.27931488801054,
      "grad_norm": 2.099121570587158,
      "learning_rate": 0.000272068511198946,
      "loss": 3.4574,
      "step": 12110
    },
    {
      "epoch": 2.2811970638057595,
      "grad_norm": 2.102046489715576,
      "learning_rate": 0.0002718802936194241,
      "loss": 3.8274,
      "step": 12120
    },
    {
      "epoch": 2.283079239600979,
      "grad_norm": 3.1600725650787354,
      "learning_rate": 0.00027169207603990216,
      "loss": 3.6157,
      "step": 12130
    },
    {
      "epoch": 2.284961415396198,
      "grad_norm": 2.4826478958129883,
      "learning_rate": 0.00027150385846038023,
      "loss": 3.5571,
      "step": 12140
    },
    {
      "epoch": 2.2868435911914173,
      "grad_norm": 2.2563583850860596,
      "learning_rate": 0.0002713156408808583,
      "loss": 3.6172,
      "step": 12150
    },
    {
      "epoch": 2.2887257669866363,
      "grad_norm": 2.2361643314361572,
      "learning_rate": 0.00027112742330133633,
      "loss": 3.4263,
      "step": 12160
    },
    {
      "epoch": 2.2906079427818558,
      "grad_norm": 2.2822108268737793,
      "learning_rate": 0.0002709392057218144,
      "loss": 3.4694,
      "step": 12170
    },
    {
      "epoch": 2.292490118577075,
      "grad_norm": 2.0215232372283936,
      "learning_rate": 0.0002707509881422925,
      "loss": 3.6782,
      "step": 12180
    },
    {
      "epoch": 2.2943722943722946,
      "grad_norm": 2.178675651550293,
      "learning_rate": 0.00027056277056277056,
      "loss": 3.626,
      "step": 12190
    },
    {
      "epoch": 2.2962544701675136,
      "grad_norm": 2.0457639694213867,
      "learning_rate": 0.00027037455298324864,
      "loss": 3.5606,
      "step": 12200
    },
    {
      "epoch": 2.298136645962733,
      "grad_norm": 2.4033384323120117,
      "learning_rate": 0.0002701863354037267,
      "loss": 3.3515,
      "step": 12210
    },
    {
      "epoch": 2.300018821757952,
      "grad_norm": 2.2202134132385254,
      "learning_rate": 0.0002699981178242048,
      "loss": 3.5911,
      "step": 12220
    },
    {
      "epoch": 2.3019009975531715,
      "grad_norm": 2.3605339527130127,
      "learning_rate": 0.00026980990024468287,
      "loss": 3.6939,
      "step": 12230
    },
    {
      "epoch": 2.303783173348391,
      "grad_norm": 2.1413116455078125,
      "learning_rate": 0.0002696216826651609,
      "loss": 3.7362,
      "step": 12240
    },
    {
      "epoch": 2.30566534914361,
      "grad_norm": 2.47611403465271,
      "learning_rate": 0.00026943346508563897,
      "loss": 3.3767,
      "step": 12250
    },
    {
      "epoch": 2.3075475249388293,
      "grad_norm": 2.8459672927856445,
      "learning_rate": 0.00026924524750611704,
      "loss": 3.5238,
      "step": 12260
    },
    {
      "epoch": 2.3094297007340487,
      "grad_norm": 2.266383171081543,
      "learning_rate": 0.0002690570299265951,
      "loss": 3.7339,
      "step": 12270
    },
    {
      "epoch": 2.3113118765292677,
      "grad_norm": 2.063275098800659,
      "learning_rate": 0.0002688688123470732,
      "loss": 3.4752,
      "step": 12280
    },
    {
      "epoch": 2.313194052324487,
      "grad_norm": 2.5126569271087646,
      "learning_rate": 0.0002686805947675513,
      "loss": 3.596,
      "step": 12290
    },
    {
      "epoch": 2.3150762281197066,
      "grad_norm": 2.4327335357666016,
      "learning_rate": 0.0002684923771880294,
      "loss": 3.6448,
      "step": 12300
    },
    {
      "epoch": 2.3169584039149256,
      "grad_norm": 2.125715494155884,
      "learning_rate": 0.0002683041596085075,
      "loss": 3.3007,
      "step": 12310
    },
    {
      "epoch": 2.318840579710145,
      "grad_norm": 2.6347568035125732,
      "learning_rate": 0.00026811594202898556,
      "loss": 3.6187,
      "step": 12320
    },
    {
      "epoch": 2.320722755505364,
      "grad_norm": 2.244523048400879,
      "learning_rate": 0.0002679277244494636,
      "loss": 3.6203,
      "step": 12330
    },
    {
      "epoch": 2.3226049313005834,
      "grad_norm": 2.629132032394409,
      "learning_rate": 0.00026773950686994165,
      "loss": 3.5333,
      "step": 12340
    },
    {
      "epoch": 2.324487107095803,
      "grad_norm": 2.6131486892700195,
      "learning_rate": 0.00026755128929041973,
      "loss": 3.6913,
      "step": 12350
    },
    {
      "epoch": 2.326369282891022,
      "grad_norm": 2.2687394618988037,
      "learning_rate": 0.0002673630717108978,
      "loss": 3.5431,
      "step": 12360
    },
    {
      "epoch": 2.3282514586862413,
      "grad_norm": 1.9078861474990845,
      "learning_rate": 0.0002671748541313759,
      "loss": 3.721,
      "step": 12370
    },
    {
      "epoch": 2.3301336344814607,
      "grad_norm": 2.5289957523345947,
      "learning_rate": 0.00026698663655185396,
      "loss": 3.6161,
      "step": 12380
    },
    {
      "epoch": 2.3320158102766797,
      "grad_norm": 2.075119733810425,
      "learning_rate": 0.00026679841897233204,
      "loss": 3.5798,
      "step": 12390
    },
    {
      "epoch": 2.333897986071899,
      "grad_norm": 2.6997997760772705,
      "learning_rate": 0.0002666102013928101,
      "loss": 3.8208,
      "step": 12400
    },
    {
      "epoch": 2.3357801618671186,
      "grad_norm": 2.278843641281128,
      "learning_rate": 0.0002664219838132882,
      "loss": 3.7117,
      "step": 12410
    },
    {
      "epoch": 2.3376623376623376,
      "grad_norm": 2.0488789081573486,
      "learning_rate": 0.0002662337662337662,
      "loss": 3.514,
      "step": 12420
    },
    {
      "epoch": 2.339544513457557,
      "grad_norm": 2.1381616592407227,
      "learning_rate": 0.0002660455486542443,
      "loss": 3.8678,
      "step": 12430
    },
    {
      "epoch": 2.341426689252776,
      "grad_norm": 2.3508248329162598,
      "learning_rate": 0.00026585733107472236,
      "loss": 3.6683,
      "step": 12440
    },
    {
      "epoch": 2.3433088650479954,
      "grad_norm": 1.9220882654190063,
      "learning_rate": 0.00026566911349520044,
      "loss": 3.4588,
      "step": 12450
    },
    {
      "epoch": 2.345191040843215,
      "grad_norm": 2.929403066635132,
      "learning_rate": 0.0002654808959156785,
      "loss": 3.3913,
      "step": 12460
    },
    {
      "epoch": 2.3470732166384343,
      "grad_norm": 2.3121893405914307,
      "learning_rate": 0.0002652926783361566,
      "loss": 3.3751,
      "step": 12470
    },
    {
      "epoch": 2.3489553924336533,
      "grad_norm": 2.4507408142089844,
      "learning_rate": 0.00026510446075663467,
      "loss": 3.7149,
      "step": 12480
    },
    {
      "epoch": 2.3508375682288727,
      "grad_norm": 2.333505630493164,
      "learning_rate": 0.00026491624317711275,
      "loss": 3.4675,
      "step": 12490
    },
    {
      "epoch": 2.3527197440240917,
      "grad_norm": 2.398038148880005,
      "learning_rate": 0.00026472802559759077,
      "loss": 3.7468,
      "step": 12500
    },
    {
      "epoch": 2.354601919819311,
      "grad_norm": 2.136672019958496,
      "learning_rate": 0.0002645398080180689,
      "loss": 3.7055,
      "step": 12510
    },
    {
      "epoch": 2.3564840956145305,
      "grad_norm": 2.207770586013794,
      "learning_rate": 0.000264351590438547,
      "loss": 3.6393,
      "step": 12520
    },
    {
      "epoch": 2.3583662714097495,
      "grad_norm": 2.111809015274048,
      "learning_rate": 0.00026416337285902505,
      "loss": 3.4927,
      "step": 12530
    },
    {
      "epoch": 2.360248447204969,
      "grad_norm": 1.9219146966934204,
      "learning_rate": 0.00026397515527950313,
      "loss": 3.3832,
      "step": 12540
    },
    {
      "epoch": 2.3621306230001884,
      "grad_norm": 2.373422384262085,
      "learning_rate": 0.0002637869376999812,
      "loss": 3.3155,
      "step": 12550
    },
    {
      "epoch": 2.3640127987954074,
      "grad_norm": 2.4442925453186035,
      "learning_rate": 0.0002635987201204593,
      "loss": 3.6199,
      "step": 12560
    },
    {
      "epoch": 2.365894974590627,
      "grad_norm": 2.070446252822876,
      "learning_rate": 0.00026341050254093736,
      "loss": 3.5309,
      "step": 12570
    },
    {
      "epoch": 2.3677771503858462,
      "grad_norm": 2.5628628730773926,
      "learning_rate": 0.00026322228496141543,
      "loss": 3.5052,
      "step": 12580
    },
    {
      "epoch": 2.3696593261810652,
      "grad_norm": 2.644749879837036,
      "learning_rate": 0.00026303406738189346,
      "loss": 3.403,
      "step": 12590
    },
    {
      "epoch": 2.3715415019762847,
      "grad_norm": 2.175874948501587,
      "learning_rate": 0.00026284584980237153,
      "loss": 3.3338,
      "step": 12600
    },
    {
      "epoch": 2.3734236777715036,
      "grad_norm": 2.5138566493988037,
      "learning_rate": 0.0002626576322228496,
      "loss": 3.3175,
      "step": 12610
    },
    {
      "epoch": 2.375305853566723,
      "grad_norm": 2.261333465576172,
      "learning_rate": 0.0002624694146433277,
      "loss": 3.3355,
      "step": 12620
    },
    {
      "epoch": 2.3771880293619425,
      "grad_norm": 2.922394275665283,
      "learning_rate": 0.00026228119706380576,
      "loss": 3.5954,
      "step": 12630
    },
    {
      "epoch": 2.3790702051571615,
      "grad_norm": 2.592365026473999,
      "learning_rate": 0.00026209297948428384,
      "loss": 3.6931,
      "step": 12640
    },
    {
      "epoch": 2.380952380952381,
      "grad_norm": 2.305166721343994,
      "learning_rate": 0.0002619047619047619,
      "loss": 3.4529,
      "step": 12650
    },
    {
      "epoch": 2.3828345567476004,
      "grad_norm": 2.466231346130371,
      "learning_rate": 0.00026171654432524,
      "loss": 3.1573,
      "step": 12660
    },
    {
      "epoch": 2.3847167325428194,
      "grad_norm": 2.3147668838500977,
      "learning_rate": 0.00026152832674571807,
      "loss": 3.6978,
      "step": 12670
    },
    {
      "epoch": 2.386598908338039,
      "grad_norm": 2.0412979125976562,
      "learning_rate": 0.0002613401091661961,
      "loss": 3.6198,
      "step": 12680
    },
    {
      "epoch": 2.388481084133258,
      "grad_norm": 2.3344006538391113,
      "learning_rate": 0.00026115189158667417,
      "loss": 3.5463,
      "step": 12690
    },
    {
      "epoch": 2.390363259928477,
      "grad_norm": 2.2637939453125,
      "learning_rate": 0.00026096367400715224,
      "loss": 3.7714,
      "step": 12700
    },
    {
      "epoch": 2.3922454357236966,
      "grad_norm": 2.0158164501190186,
      "learning_rate": 0.0002607754564276303,
      "loss": 3.6364,
      "step": 12710
    },
    {
      "epoch": 2.3941276115189156,
      "grad_norm": 2.035423755645752,
      "learning_rate": 0.0002605872388481084,
      "loss": 3.6938,
      "step": 12720
    },
    {
      "epoch": 2.396009787314135,
      "grad_norm": 1.8480379581451416,
      "learning_rate": 0.0002603990212685865,
      "loss": 3.8096,
      "step": 12730
    },
    {
      "epoch": 2.3978919631093545,
      "grad_norm": 2.079611301422119,
      "learning_rate": 0.0002602108036890646,
      "loss": 3.4641,
      "step": 12740
    },
    {
      "epoch": 2.399774138904574,
      "grad_norm": 1.9311271905899048,
      "learning_rate": 0.0002600225861095427,
      "loss": 3.5725,
      "step": 12750
    },
    {
      "epoch": 2.401656314699793,
      "grad_norm": 2.1596946716308594,
      "learning_rate": 0.0002598343685300207,
      "loss": 3.7113,
      "step": 12760
    },
    {
      "epoch": 2.4035384904950123,
      "grad_norm": 2.4744985103607178,
      "learning_rate": 0.0002596461509504988,
      "loss": 3.5168,
      "step": 12770
    },
    {
      "epoch": 2.4054206662902313,
      "grad_norm": 2.025146245956421,
      "learning_rate": 0.00025945793337097685,
      "loss": 3.9381,
      "step": 12780
    },
    {
      "epoch": 2.4073028420854508,
      "grad_norm": 2.1392645835876465,
      "learning_rate": 0.00025926971579145493,
      "loss": 3.3267,
      "step": 12790
    },
    {
      "epoch": 2.40918501788067,
      "grad_norm": 2.1594839096069336,
      "learning_rate": 0.000259081498211933,
      "loss": 3.7087,
      "step": 12800
    },
    {
      "epoch": 2.411067193675889,
      "grad_norm": 1.840181827545166,
      "learning_rate": 0.0002588932806324111,
      "loss": 3.5356,
      "step": 12810
    },
    {
      "epoch": 2.4129493694711086,
      "grad_norm": 2.2144663333892822,
      "learning_rate": 0.00025870506305288916,
      "loss": 3.4101,
      "step": 12820
    },
    {
      "epoch": 2.414831545266328,
      "grad_norm": 2.0941808223724365,
      "learning_rate": 0.00025851684547336724,
      "loss": 3.7866,
      "step": 12830
    },
    {
      "epoch": 2.416713721061547,
      "grad_norm": 2.089059829711914,
      "learning_rate": 0.0002583286278938453,
      "loss": 3.6845,
      "step": 12840
    },
    {
      "epoch": 2.4185958968567665,
      "grad_norm": 2.912400960922241,
      "learning_rate": 0.00025814041031432334,
      "loss": 3.7385,
      "step": 12850
    },
    {
      "epoch": 2.420478072651986,
      "grad_norm": 2.2902355194091797,
      "learning_rate": 0.0002579521927348014,
      "loss": 3.7033,
      "step": 12860
    },
    {
      "epoch": 2.422360248447205,
      "grad_norm": 1.9202638864517212,
      "learning_rate": 0.0002577639751552795,
      "loss": 3.6759,
      "step": 12870
    },
    {
      "epoch": 2.4242424242424243,
      "grad_norm": 2.116422414779663,
      "learning_rate": 0.00025757575757575756,
      "loss": 3.7004,
      "step": 12880
    },
    {
      "epoch": 2.4261246000376433,
      "grad_norm": 2.2019739151000977,
      "learning_rate": 0.00025738753999623564,
      "loss": 3.4297,
      "step": 12890
    },
    {
      "epoch": 2.4280067758328627,
      "grad_norm": 2.3102312088012695,
      "learning_rate": 0.0002571993224167137,
      "loss": 3.6366,
      "step": 12900
    },
    {
      "epoch": 2.429888951628082,
      "grad_norm": 2.1590123176574707,
      "learning_rate": 0.0002570111048371918,
      "loss": 3.4755,
      "step": 12910
    },
    {
      "epoch": 2.431771127423301,
      "grad_norm": 2.459792137145996,
      "learning_rate": 0.00025682288725766987,
      "loss": 3.4308,
      "step": 12920
    },
    {
      "epoch": 2.4336533032185206,
      "grad_norm": 2.6131508350372314,
      "learning_rate": 0.00025663466967814795,
      "loss": 3.3086,
      "step": 12930
    },
    {
      "epoch": 2.43553547901374,
      "grad_norm": 2.1226155757904053,
      "learning_rate": 0.00025644645209862597,
      "loss": 3.5569,
      "step": 12940
    },
    {
      "epoch": 2.437417654808959,
      "grad_norm": 2.094432830810547,
      "learning_rate": 0.00025625823451910405,
      "loss": 3.7615,
      "step": 12950
    },
    {
      "epoch": 2.4392998306041784,
      "grad_norm": 2.2213399410247803,
      "learning_rate": 0.0002560700169395822,
      "loss": 3.42,
      "step": 12960
    },
    {
      "epoch": 2.441182006399398,
      "grad_norm": 2.1061394214630127,
      "learning_rate": 0.00025588179936006025,
      "loss": 3.6418,
      "step": 12970
    },
    {
      "epoch": 2.443064182194617,
      "grad_norm": 2.438858985900879,
      "learning_rate": 0.00025569358178053833,
      "loss": 3.3957,
      "step": 12980
    },
    {
      "epoch": 2.4449463579898363,
      "grad_norm": 2.3009002208709717,
      "learning_rate": 0.0002555053642010164,
      "loss": 3.4535,
      "step": 12990
    },
    {
      "epoch": 2.4468285337850557,
      "grad_norm": 2.1550045013427734,
      "learning_rate": 0.0002553171466214945,
      "loss": 3.5498,
      "step": 13000
    },
    {
      "epoch": 2.4487107095802747,
      "grad_norm": 2.06872820854187,
      "learning_rate": 0.00025512892904197256,
      "loss": 3.4783,
      "step": 13010
    },
    {
      "epoch": 2.450592885375494,
      "grad_norm": 1.8921680450439453,
      "learning_rate": 0.00025494071146245063,
      "loss": 3.7661,
      "step": 13020
    },
    {
      "epoch": 2.4524750611707136,
      "grad_norm": 2.2813234329223633,
      "learning_rate": 0.00025475249388292866,
      "loss": 3.5401,
      "step": 13030
    },
    {
      "epoch": 2.4543572369659326,
      "grad_norm": 2.5784573554992676,
      "learning_rate": 0.00025456427630340673,
      "loss": 3.2728,
      "step": 13040
    },
    {
      "epoch": 2.456239412761152,
      "grad_norm": 1.9490470886230469,
      "learning_rate": 0.0002543760587238848,
      "loss": 3.4593,
      "step": 13050
    },
    {
      "epoch": 2.458121588556371,
      "grad_norm": 2.069911479949951,
      "learning_rate": 0.0002541878411443629,
      "loss": 3.418,
      "step": 13060
    },
    {
      "epoch": 2.4600037643515904,
      "grad_norm": 2.3085179328918457,
      "learning_rate": 0.00025399962356484096,
      "loss": 3.3341,
      "step": 13070
    },
    {
      "epoch": 2.46188594014681,
      "grad_norm": 2.4705824851989746,
      "learning_rate": 0.00025381140598531904,
      "loss": 3.733,
      "step": 13080
    },
    {
      "epoch": 2.463768115942029,
      "grad_norm": 2.3225464820861816,
      "learning_rate": 0.0002536231884057971,
      "loss": 3.499,
      "step": 13090
    },
    {
      "epoch": 2.4656502917372483,
      "grad_norm": 2.043210506439209,
      "learning_rate": 0.0002534349708262752,
      "loss": 3.6164,
      "step": 13100
    },
    {
      "epoch": 2.4675324675324677,
      "grad_norm": 1.996595025062561,
      "learning_rate": 0.0002532467532467532,
      "loss": 3.2083,
      "step": 13110
    },
    {
      "epoch": 2.4694146433276867,
      "grad_norm": 2.048947811126709,
      "learning_rate": 0.0002530585356672313,
      "loss": 3.6867,
      "step": 13120
    },
    {
      "epoch": 2.471296819122906,
      "grad_norm": 2.1778929233551025,
      "learning_rate": 0.00025287031808770937,
      "loss": 3.3652,
      "step": 13130
    },
    {
      "epoch": 2.4731789949181255,
      "grad_norm": 2.0750463008880615,
      "learning_rate": 0.00025268210050818744,
      "loss": 3.3186,
      "step": 13140
    },
    {
      "epoch": 2.4750611707133445,
      "grad_norm": 2.405975818634033,
      "learning_rate": 0.0002524938829286655,
      "loss": 3.5292,
      "step": 13150
    },
    {
      "epoch": 2.476943346508564,
      "grad_norm": 2.224957227706909,
      "learning_rate": 0.0002523056653491436,
      "loss": 3.4451,
      "step": 13160
    },
    {
      "epoch": 2.478825522303783,
      "grad_norm": 2.276805877685547,
      "learning_rate": 0.00025211744776962167,
      "loss": 3.8172,
      "step": 13170
    },
    {
      "epoch": 2.4807076980990024,
      "grad_norm": 2.076664924621582,
      "learning_rate": 0.0002519292301900998,
      "loss": 3.4015,
      "step": 13180
    },
    {
      "epoch": 2.482589873894222,
      "grad_norm": 2.32965087890625,
      "learning_rate": 0.0002517410126105779,
      "loss": 3.4551,
      "step": 13190
    },
    {
      "epoch": 2.4844720496894412,
      "grad_norm": 2.174405336380005,
      "learning_rate": 0.0002515527950310559,
      "loss": 3.3485,
      "step": 13200
    },
    {
      "epoch": 2.4863542254846602,
      "grad_norm": 2.069518804550171,
      "learning_rate": 0.000251364577451534,
      "loss": 3.5047,
      "step": 13210
    },
    {
      "epoch": 2.4882364012798797,
      "grad_norm": 2.1788432598114014,
      "learning_rate": 0.00025117635987201205,
      "loss": 3.442,
      "step": 13220
    },
    {
      "epoch": 2.4901185770750986,
      "grad_norm": 2.014214515686035,
      "learning_rate": 0.00025098814229249013,
      "loss": 3.5642,
      "step": 13230
    },
    {
      "epoch": 2.492000752870318,
      "grad_norm": 2.407564640045166,
      "learning_rate": 0.0002507999247129682,
      "loss": 3.6167,
      "step": 13240
    },
    {
      "epoch": 2.4938829286655375,
      "grad_norm": 2.017845869064331,
      "learning_rate": 0.0002506117071334463,
      "loss": 3.7589,
      "step": 13250
    },
    {
      "epoch": 2.4957651044607565,
      "grad_norm": 2.3813204765319824,
      "learning_rate": 0.00025042348955392436,
      "loss": 3.3354,
      "step": 13260
    },
    {
      "epoch": 2.497647280255976,
      "grad_norm": 2.23264741897583,
      "learning_rate": 0.00025023527197440244,
      "loss": 3.5316,
      "step": 13270
    },
    {
      "epoch": 2.4995294560511954,
      "grad_norm": 2.2221202850341797,
      "learning_rate": 0.0002500470543948805,
      "loss": 3.5831,
      "step": 13280
    },
    {
      "epoch": 2.5014116318464144,
      "grad_norm": 2.5750515460968018,
      "learning_rate": 0.0002498588368153586,
      "loss": 3.427,
      "step": 13290
    },
    {
      "epoch": 2.503293807641634,
      "grad_norm": 2.982229471206665,
      "learning_rate": 0.00024967061923583667,
      "loss": 3.4038,
      "step": 13300
    },
    {
      "epoch": 2.505175983436853,
      "grad_norm": 3.0934581756591797,
      "learning_rate": 0.0002494824016563147,
      "loss": 3.5624,
      "step": 13310
    },
    {
      "epoch": 2.507058159232072,
      "grad_norm": 2.419921398162842,
      "learning_rate": 0.00024929418407679276,
      "loss": 3.4934,
      "step": 13320
    },
    {
      "epoch": 2.5089403350272916,
      "grad_norm": 2.056140899658203,
      "learning_rate": 0.00024910596649727084,
      "loss": 3.3226,
      "step": 13330
    },
    {
      "epoch": 2.5108225108225106,
      "grad_norm": 2.220564603805542,
      "learning_rate": 0.0002489177489177489,
      "loss": 3.5698,
      "step": 13340
    },
    {
      "epoch": 2.51270468661773,
      "grad_norm": 2.1118175983428955,
      "learning_rate": 0.000248729531338227,
      "loss": 3.4314,
      "step": 13350
    },
    {
      "epoch": 2.5145868624129495,
      "grad_norm": 2.184598684310913,
      "learning_rate": 0.00024854131375870507,
      "loss": 3.5573,
      "step": 13360
    },
    {
      "epoch": 2.516469038208169,
      "grad_norm": 2.0085346698760986,
      "learning_rate": 0.00024835309617918315,
      "loss": 3.4515,
      "step": 13370
    },
    {
      "epoch": 2.518351214003388,
      "grad_norm": 2.25004243850708,
      "learning_rate": 0.0002481648785996612,
      "loss": 3.5318,
      "step": 13380
    },
    {
      "epoch": 2.5202333897986073,
      "grad_norm": 2.047452688217163,
      "learning_rate": 0.0002479766610201393,
      "loss": 3.4558,
      "step": 13390
    },
    {
      "epoch": 2.5221155655938263,
      "grad_norm": 2.483903169631958,
      "learning_rate": 0.0002477884434406174,
      "loss": 3.6721,
      "step": 13400
    },
    {
      "epoch": 2.5239977413890458,
      "grad_norm": 2.2482049465179443,
      "learning_rate": 0.00024760022586109545,
      "loss": 3.4841,
      "step": 13410
    },
    {
      "epoch": 2.525879917184265,
      "grad_norm": 2.063401937484741,
      "learning_rate": 0.00024741200828157353,
      "loss": 3.4875,
      "step": 13420
    },
    {
      "epoch": 2.527762092979484,
      "grad_norm": 1.967253565788269,
      "learning_rate": 0.0002472237907020516,
      "loss": 3.5811,
      "step": 13430
    },
    {
      "epoch": 2.5296442687747036,
      "grad_norm": 1.9973634481430054,
      "learning_rate": 0.00024703557312252963,
      "loss": 3.4078,
      "step": 13440
    },
    {
      "epoch": 2.5315264445699226,
      "grad_norm": 2.570246458053589,
      "learning_rate": 0.0002468473555430077,
      "loss": 3.5667,
      "step": 13450
    },
    {
      "epoch": 2.533408620365142,
      "grad_norm": 2.4065957069396973,
      "learning_rate": 0.0002466591379634858,
      "loss": 3.458,
      "step": 13460
    },
    {
      "epoch": 2.5352907961603615,
      "grad_norm": 2.274855852127075,
      "learning_rate": 0.00024647092038396386,
      "loss": 3.5801,
      "step": 13470
    },
    {
      "epoch": 2.537172971955581,
      "grad_norm": 2.714076519012451,
      "learning_rate": 0.00024628270280444193,
      "loss": 3.4157,
      "step": 13480
    },
    {
      "epoch": 2.5390551477508,
      "grad_norm": 3.3540868759155273,
      "learning_rate": 0.00024609448522492,
      "loss": 3.8492,
      "step": 13490
    },
    {
      "epoch": 2.5409373235460193,
      "grad_norm": 2.6228458881378174,
      "learning_rate": 0.0002459062676453981,
      "loss": 3.4858,
      "step": 13500
    },
    {
      "epoch": 2.5428194993412383,
      "grad_norm": 2.2022087574005127,
      "learning_rate": 0.00024571805006587616,
      "loss": 3.4548,
      "step": 13510
    },
    {
      "epoch": 2.5447016751364577,
      "grad_norm": 2.2522380352020264,
      "learning_rate": 0.00024552983248635424,
      "loss": 3.5098,
      "step": 13520
    },
    {
      "epoch": 2.546583850931677,
      "grad_norm": 2.1798369884490967,
      "learning_rate": 0.0002453416149068323,
      "loss": 3.658,
      "step": 13530
    },
    {
      "epoch": 2.548466026726896,
      "grad_norm": 2.4629127979278564,
      "learning_rate": 0.0002451533973273104,
      "loss": 3.3775,
      "step": 13540
    },
    {
      "epoch": 2.5503482025221156,
      "grad_norm": 2.09586763381958,
      "learning_rate": 0.00024496517974778847,
      "loss": 3.3251,
      "step": 13550
    },
    {
      "epoch": 2.5522303783173346,
      "grad_norm": 2.1991000175476074,
      "learning_rate": 0.00024477696216826654,
      "loss": 3.4861,
      "step": 13560
    },
    {
      "epoch": 2.554112554112554,
      "grad_norm": 3.098156690597534,
      "learning_rate": 0.00024458874458874457,
      "loss": 3.3637,
      "step": 13570
    },
    {
      "epoch": 2.5559947299077734,
      "grad_norm": 2.7341153621673584,
      "learning_rate": 0.00024440052700922264,
      "loss": 3.5117,
      "step": 13580
    },
    {
      "epoch": 2.557876905702993,
      "grad_norm": 2.235184907913208,
      "learning_rate": 0.0002442123094297007,
      "loss": 3.4566,
      "step": 13590
    },
    {
      "epoch": 2.559759081498212,
      "grad_norm": 2.4505128860473633,
      "learning_rate": 0.00024402409185017882,
      "loss": 3.4255,
      "step": 13600
    },
    {
      "epoch": 2.5616412572934313,
      "grad_norm": 2.367823600769043,
      "learning_rate": 0.00024383587427065687,
      "loss": 3.7588,
      "step": 13610
    },
    {
      "epoch": 2.5635234330886503,
      "grad_norm": 2.214277982711792,
      "learning_rate": 0.00024364765669113495,
      "loss": 3.5693,
      "step": 13620
    },
    {
      "epoch": 2.5654056088838697,
      "grad_norm": 2.072028636932373,
      "learning_rate": 0.00024345943911161303,
      "loss": 3.8349,
      "step": 13630
    },
    {
      "epoch": 2.567287784679089,
      "grad_norm": 2.3604848384857178,
      "learning_rate": 0.0002432712215320911,
      "loss": 3.7526,
      "step": 13640
    },
    {
      "epoch": 2.5691699604743086,
      "grad_norm": 2.4565625190734863,
      "learning_rate": 0.00024308300395256918,
      "loss": 3.3296,
      "step": 13650
    },
    {
      "epoch": 2.5710521362695276,
      "grad_norm": 2.181539297103882,
      "learning_rate": 0.00024289478637304723,
      "loss": 3.4487,
      "step": 13660
    },
    {
      "epoch": 2.572934312064747,
      "grad_norm": 2.3154828548431396,
      "learning_rate": 0.0002427065687935253,
      "loss": 3.4625,
      "step": 13670
    },
    {
      "epoch": 2.574816487859966,
      "grad_norm": 2.2067558765411377,
      "learning_rate": 0.0002425183512140034,
      "loss": 3.6838,
      "step": 13680
    },
    {
      "epoch": 2.5766986636551854,
      "grad_norm": 2.6612298488616943,
      "learning_rate": 0.00024233013363448148,
      "loss": 3.4488,
      "step": 13690
    },
    {
      "epoch": 2.578580839450405,
      "grad_norm": 2.5054516792297363,
      "learning_rate": 0.00024214191605495953,
      "loss": 3.7849,
      "step": 13700
    },
    {
      "epoch": 2.580463015245624,
      "grad_norm": 2.342247247695923,
      "learning_rate": 0.0002419536984754376,
      "loss": 3.5564,
      "step": 13710
    },
    {
      "epoch": 2.5823451910408433,
      "grad_norm": 2.155902147293091,
      "learning_rate": 0.00024176548089591569,
      "loss": 3.5476,
      "step": 13720
    },
    {
      "epoch": 2.5842273668360622,
      "grad_norm": 2.3009862899780273,
      "learning_rate": 0.00024157726331639376,
      "loss": 3.4861,
      "step": 13730
    },
    {
      "epoch": 2.5861095426312817,
      "grad_norm": 1.888217806816101,
      "learning_rate": 0.0002413890457368718,
      "loss": 3.3693,
      "step": 13740
    },
    {
      "epoch": 2.587991718426501,
      "grad_norm": 2.2217419147491455,
      "learning_rate": 0.0002412008281573499,
      "loss": 3.828,
      "step": 13750
    },
    {
      "epoch": 2.5898738942217205,
      "grad_norm": 2.217092275619507,
      "learning_rate": 0.00024101261057782796,
      "loss": 3.3636,
      "step": 13760
    },
    {
      "epoch": 2.5917560700169395,
      "grad_norm": 2.1933679580688477,
      "learning_rate": 0.00024082439299830604,
      "loss": 3.4174,
      "step": 13770
    },
    {
      "epoch": 2.593638245812159,
      "grad_norm": 2.34127140045166,
      "learning_rate": 0.00024063617541878414,
      "loss": 3.4482,
      "step": 13780
    },
    {
      "epoch": 2.595520421607378,
      "grad_norm": 2.314049005508423,
      "learning_rate": 0.0002404479578392622,
      "loss": 3.3114,
      "step": 13790
    },
    {
      "epoch": 2.5974025974025974,
      "grad_norm": 2.2674081325531006,
      "learning_rate": 0.00024025974025974027,
      "loss": 3.637,
      "step": 13800
    },
    {
      "epoch": 2.599284773197817,
      "grad_norm": 2.1805338859558105,
      "learning_rate": 0.00024007152268021835,
      "loss": 3.6175,
      "step": 13810
    },
    {
      "epoch": 2.601166948993036,
      "grad_norm": 2.096508741378784,
      "learning_rate": 0.00023988330510069642,
      "loss": 3.8226,
      "step": 13820
    },
    {
      "epoch": 2.6030491247882552,
      "grad_norm": 2.3623275756835938,
      "learning_rate": 0.00023969508752117447,
      "loss": 3.6106,
      "step": 13830
    },
    {
      "epoch": 2.6049313005834747,
      "grad_norm": 2.232391119003296,
      "learning_rate": 0.00023950686994165255,
      "loss": 3.7322,
      "step": 13840
    },
    {
      "epoch": 2.6068134763786937,
      "grad_norm": 2.4925694465637207,
      "learning_rate": 0.00023931865236213063,
      "loss": 3.557,
      "step": 13850
    },
    {
      "epoch": 2.608695652173913,
      "grad_norm": 2.8005504608154297,
      "learning_rate": 0.0002391304347826087,
      "loss": 3.5394,
      "step": 13860
    },
    {
      "epoch": 2.6105778279691325,
      "grad_norm": 2.3423023223876953,
      "learning_rate": 0.00023894221720308678,
      "loss": 3.5762,
      "step": 13870
    },
    {
      "epoch": 2.6124600037643515,
      "grad_norm": 2.3541934490203857,
      "learning_rate": 0.00023875399962356483,
      "loss": 3.6592,
      "step": 13880
    },
    {
      "epoch": 2.614342179559571,
      "grad_norm": 2.138944387435913,
      "learning_rate": 0.0002385657820440429,
      "loss": 3.3244,
      "step": 13890
    },
    {
      "epoch": 2.61622435535479,
      "grad_norm": 2.374363899230957,
      "learning_rate": 0.000238377564464521,
      "loss": 3.4745,
      "step": 13900
    },
    {
      "epoch": 2.6181065311500094,
      "grad_norm": 2.2343404293060303,
      "learning_rate": 0.00023818934688499908,
      "loss": 3.4339,
      "step": 13910
    },
    {
      "epoch": 2.619988706945229,
      "grad_norm": 5.174120903015137,
      "learning_rate": 0.00023800112930547713,
      "loss": 3.4124,
      "step": 13920
    },
    {
      "epoch": 2.621870882740448,
      "grad_norm": 2.39028000831604,
      "learning_rate": 0.0002378129117259552,
      "loss": 3.404,
      "step": 13930
    },
    {
      "epoch": 2.623753058535667,
      "grad_norm": 2.3732285499572754,
      "learning_rate": 0.00023762469414643329,
      "loss": 3.6385,
      "step": 13940
    },
    {
      "epoch": 2.6256352343308866,
      "grad_norm": 2.5302624702453613,
      "learning_rate": 0.00023743647656691136,
      "loss": 3.4885,
      "step": 13950
    },
    {
      "epoch": 2.6275174101261056,
      "grad_norm": 2.292924642562866,
      "learning_rate": 0.0002372482589873894,
      "loss": 3.4066,
      "step": 13960
    },
    {
      "epoch": 2.629399585921325,
      "grad_norm": 2.267543315887451,
      "learning_rate": 0.0002370600414078675,
      "loss": 3.1002,
      "step": 13970
    },
    {
      "epoch": 2.6312817617165445,
      "grad_norm": 2.347487688064575,
      "learning_rate": 0.00023687182382834556,
      "loss": 3.5494,
      "step": 13980
    },
    {
      "epoch": 2.6331639375117635,
      "grad_norm": 2.2723233699798584,
      "learning_rate": 0.00023668360624882364,
      "loss": 3.5419,
      "step": 13990
    },
    {
      "epoch": 2.635046113306983,
      "grad_norm": 2.240119457244873,
      "learning_rate": 0.00023649538866930172,
      "loss": 3.5517,
      "step": 14000
    },
    {
      "epoch": 2.636928289102202,
      "grad_norm": 2.802652359008789,
      "learning_rate": 0.0002363071710897798,
      "loss": 3.4292,
      "step": 14010
    },
    {
      "epoch": 2.6388104648974213,
      "grad_norm": 2.53572678565979,
      "learning_rate": 0.00023611895351025787,
      "loss": 3.1934,
      "step": 14020
    },
    {
      "epoch": 2.6406926406926408,
      "grad_norm": 2.339966297149658,
      "learning_rate": 0.00023593073593073595,
      "loss": 3.5413,
      "step": 14030
    },
    {
      "epoch": 2.64257481648786,
      "grad_norm": 2.430595636367798,
      "learning_rate": 0.00023574251835121402,
      "loss": 3.3905,
      "step": 14040
    },
    {
      "epoch": 2.644456992283079,
      "grad_norm": 2.0948545932769775,
      "learning_rate": 0.00023555430077169207,
      "loss": 3.34,
      "step": 14050
    },
    {
      "epoch": 2.6463391680782986,
      "grad_norm": 2.448134422302246,
      "learning_rate": 0.00023536608319217015,
      "loss": 3.5613,
      "step": 14060
    },
    {
      "epoch": 2.6482213438735176,
      "grad_norm": 2.4141626358032227,
      "learning_rate": 0.00023517786561264823,
      "loss": 3.6671,
      "step": 14070
    },
    {
      "epoch": 2.650103519668737,
      "grad_norm": 2.3373255729675293,
      "learning_rate": 0.0002349896480331263,
      "loss": 3.8727,
      "step": 14080
    },
    {
      "epoch": 2.6519856954639565,
      "grad_norm": 2.106760025024414,
      "learning_rate": 0.00023480143045360435,
      "loss": 3.5622,
      "step": 14090
    },
    {
      "epoch": 2.6538678712591754,
      "grad_norm": 2.769371271133423,
      "learning_rate": 0.00023461321287408243,
      "loss": 3.4578,
      "step": 14100
    },
    {
      "epoch": 2.655750047054395,
      "grad_norm": 2.594217300415039,
      "learning_rate": 0.0002344249952945605,
      "loss": 3.3999,
      "step": 14110
    },
    {
      "epoch": 2.6576322228496143,
      "grad_norm": 2.42844820022583,
      "learning_rate": 0.0002342367777150386,
      "loss": 3.326,
      "step": 14120
    },
    {
      "epoch": 2.6595143986448333,
      "grad_norm": 2.072188138961792,
      "learning_rate": 0.00023404856013551668,
      "loss": 3.5705,
      "step": 14130
    },
    {
      "epoch": 2.6613965744400527,
      "grad_norm": 2.64915132522583,
      "learning_rate": 0.00023386034255599473,
      "loss": 3.5551,
      "step": 14140
    },
    {
      "epoch": 2.663278750235272,
      "grad_norm": 1.9408377408981323,
      "learning_rate": 0.0002336721249764728,
      "loss": 3.402,
      "step": 14150
    },
    {
      "epoch": 2.665160926030491,
      "grad_norm": 2.603893756866455,
      "learning_rate": 0.00023348390739695089,
      "loss": 3.5726,
      "step": 14160
    },
    {
      "epoch": 2.6670431018257106,
      "grad_norm": 2.6545052528381348,
      "learning_rate": 0.00023329568981742896,
      "loss": 3.5343,
      "step": 14170
    },
    {
      "epoch": 2.6689252776209296,
      "grad_norm": 2.0449416637420654,
      "learning_rate": 0.000233107472237907,
      "loss": 3.6172,
      "step": 14180
    },
    {
      "epoch": 2.670807453416149,
      "grad_norm": 2.6366517543792725,
      "learning_rate": 0.0002329192546583851,
      "loss": 3.4519,
      "step": 14190
    },
    {
      "epoch": 2.6726896292113684,
      "grad_norm": 2.118492841720581,
      "learning_rate": 0.00023273103707886316,
      "loss": 3.4333,
      "step": 14200
    },
    {
      "epoch": 2.674571805006588,
      "grad_norm": 2.4209890365600586,
      "learning_rate": 0.00023254281949934124,
      "loss": 3.5463,
      "step": 14210
    },
    {
      "epoch": 2.676453980801807,
      "grad_norm": 2.1032493114471436,
      "learning_rate": 0.0002323546019198193,
      "loss": 3.6745,
      "step": 14220
    },
    {
      "epoch": 2.6783361565970263,
      "grad_norm": 2.586094379425049,
      "learning_rate": 0.0002321663843402974,
      "loss": 3.475,
      "step": 14230
    },
    {
      "epoch": 2.6802183323922453,
      "grad_norm": 1.9233129024505615,
      "learning_rate": 0.00023197816676077547,
      "loss": 3.4387,
      "step": 14240
    },
    {
      "epoch": 2.6821005081874647,
      "grad_norm": 2.343977689743042,
      "learning_rate": 0.00023178994918125355,
      "loss": 3.7495,
      "step": 14250
    },
    {
      "epoch": 2.683982683982684,
      "grad_norm": 1.9606075286865234,
      "learning_rate": 0.00023160173160173162,
      "loss": 3.8449,
      "step": 14260
    },
    {
      "epoch": 2.685864859777903,
      "grad_norm": 2.037147045135498,
      "learning_rate": 0.00023141351402220967,
      "loss": 3.6515,
      "step": 14270
    },
    {
      "epoch": 2.6877470355731226,
      "grad_norm": 2.135225296020508,
      "learning_rate": 0.00023122529644268775,
      "loss": 3.4237,
      "step": 14280
    },
    {
      "epoch": 2.6896292113683415,
      "grad_norm": 2.5087924003601074,
      "learning_rate": 0.00023103707886316583,
      "loss": 3.654,
      "step": 14290
    },
    {
      "epoch": 2.691511387163561,
      "grad_norm": 2.4269285202026367,
      "learning_rate": 0.0002308488612836439,
      "loss": 3.333,
      "step": 14300
    },
    {
      "epoch": 2.6933935629587804,
      "grad_norm": 2.0803372859954834,
      "learning_rate": 0.00023066064370412195,
      "loss": 3.4516,
      "step": 14310
    },
    {
      "epoch": 2.695275738754,
      "grad_norm": 2.165856122970581,
      "learning_rate": 0.00023047242612460003,
      "loss": 3.6354,
      "step": 14320
    },
    {
      "epoch": 2.697157914549219,
      "grad_norm": 2.180093288421631,
      "learning_rate": 0.0002302842085450781,
      "loss": 3.4879,
      "step": 14330
    },
    {
      "epoch": 2.6990400903444383,
      "grad_norm": 2.306670904159546,
      "learning_rate": 0.0002300959909655562,
      "loss": 3.5027,
      "step": 14340
    },
    {
      "epoch": 2.7009222661396572,
      "grad_norm": 1.9162133932113647,
      "learning_rate": 0.00022990777338603426,
      "loss": 3.6179,
      "step": 14350
    },
    {
      "epoch": 2.7028044419348767,
      "grad_norm": 2.0616095066070557,
      "learning_rate": 0.00022971955580651233,
      "loss": 3.5577,
      "step": 14360
    },
    {
      "epoch": 2.704686617730096,
      "grad_norm": 2.197929859161377,
      "learning_rate": 0.0002295313382269904,
      "loss": 3.2005,
      "step": 14370
    },
    {
      "epoch": 2.7065687935253155,
      "grad_norm": 2.8560714721679688,
      "learning_rate": 0.00022934312064746849,
      "loss": 3.3947,
      "step": 14380
    },
    {
      "epoch": 2.7084509693205345,
      "grad_norm": 2.397538185119629,
      "learning_rate": 0.00022915490306794656,
      "loss": 3.4252,
      "step": 14390
    },
    {
      "epoch": 2.710333145115754,
      "grad_norm": 2.24406099319458,
      "learning_rate": 0.0002289666854884246,
      "loss": 3.3707,
      "step": 14400
    },
    {
      "epoch": 2.712215320910973,
      "grad_norm": 2.7690091133117676,
      "learning_rate": 0.0002287784679089027,
      "loss": 3.5741,
      "step": 14410
    },
    {
      "epoch": 2.7140974967061924,
      "grad_norm": 2.4531538486480713,
      "learning_rate": 0.00022859025032938076,
      "loss": 3.2768,
      "step": 14420
    },
    {
      "epoch": 2.715979672501412,
      "grad_norm": 2.0439727306365967,
      "learning_rate": 0.00022840203274985884,
      "loss": 3.2689,
      "step": 14430
    },
    {
      "epoch": 2.717861848296631,
      "grad_norm": 2.421132802963257,
      "learning_rate": 0.0002282138151703369,
      "loss": 3.2724,
      "step": 14440
    },
    {
      "epoch": 2.7197440240918502,
      "grad_norm": 2.507495164871216,
      "learning_rate": 0.000228025597590815,
      "loss": 3.2866,
      "step": 14450
    },
    {
      "epoch": 2.721626199887069,
      "grad_norm": 2.518876075744629,
      "learning_rate": 0.00022783738001129307,
      "loss": 3.4307,
      "step": 14460
    },
    {
      "epoch": 2.7235083756822887,
      "grad_norm": 2.889340400695801,
      "learning_rate": 0.00022764916243177115,
      "loss": 3.3949,
      "step": 14470
    },
    {
      "epoch": 2.725390551477508,
      "grad_norm": 3.0600709915161133,
      "learning_rate": 0.00022746094485224922,
      "loss": 3.5561,
      "step": 14480
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 2.439391851425171,
      "learning_rate": 0.00022727272727272727,
      "loss": 3.5498,
      "step": 14490
    },
    {
      "epoch": 2.7291549030679465,
      "grad_norm": 2.550637722015381,
      "learning_rate": 0.00022708450969320535,
      "loss": 3.344,
      "step": 14500
    },
    {
      "epoch": 2.731037078863166,
      "grad_norm": 2.25809645652771,
      "learning_rate": 0.00022689629211368343,
      "loss": 3.5153,
      "step": 14510
    },
    {
      "epoch": 2.732919254658385,
      "grad_norm": 2.413231611251831,
      "learning_rate": 0.0002267080745341615,
      "loss": 3.5793,
      "step": 14520
    },
    {
      "epoch": 2.7348014304536044,
      "grad_norm": 2.406266450881958,
      "learning_rate": 0.00022651985695463955,
      "loss": 3.3934,
      "step": 14530
    },
    {
      "epoch": 2.736683606248824,
      "grad_norm": 2.2091832160949707,
      "learning_rate": 0.00022633163937511763,
      "loss": 3.3904,
      "step": 14540
    },
    {
      "epoch": 2.7385657820440428,
      "grad_norm": 2.170200824737549,
      "learning_rate": 0.0002261434217955957,
      "loss": 3.6256,
      "step": 14550
    },
    {
      "epoch": 2.740447957839262,
      "grad_norm": 2.615487575531006,
      "learning_rate": 0.0002259552042160738,
      "loss": 3.6725,
      "step": 14560
    },
    {
      "epoch": 2.742330133634481,
      "grad_norm": 2.3170371055603027,
      "learning_rate": 0.00022576698663655186,
      "loss": 3.4314,
      "step": 14570
    },
    {
      "epoch": 2.7442123094297006,
      "grad_norm": 2.150153398513794,
      "learning_rate": 0.00022557876905702993,
      "loss": 3.4444,
      "step": 14580
    },
    {
      "epoch": 2.74609448522492,
      "grad_norm": 2.0498740673065186,
      "learning_rate": 0.000225390551477508,
      "loss": 3.578,
      "step": 14590
    },
    {
      "epoch": 2.7479766610201395,
      "grad_norm": 2.061810255050659,
      "learning_rate": 0.00022520233389798609,
      "loss": 3.4493,
      "step": 14600
    },
    {
      "epoch": 2.7498588368153585,
      "grad_norm": 2.162282705307007,
      "learning_rate": 0.00022501411631846416,
      "loss": 3.4586,
      "step": 14610
    },
    {
      "epoch": 2.751741012610578,
      "grad_norm": 2.4143502712249756,
      "learning_rate": 0.0002248258987389422,
      "loss": 3.603,
      "step": 14620
    },
    {
      "epoch": 2.753623188405797,
      "grad_norm": 2.4602837562561035,
      "learning_rate": 0.0002246376811594203,
      "loss": 3.4643,
      "step": 14630
    },
    {
      "epoch": 2.7555053642010163,
      "grad_norm": 2.0327515602111816,
      "learning_rate": 0.00022444946357989836,
      "loss": 3.3329,
      "step": 14640
    },
    {
      "epoch": 2.7573875399962358,
      "grad_norm": 3.8538384437561035,
      "learning_rate": 0.00022426124600037644,
      "loss": 3.4552,
      "step": 14650
    },
    {
      "epoch": 2.759269715791455,
      "grad_norm": 2.4830667972564697,
      "learning_rate": 0.0002240730284208545,
      "loss": 3.5332,
      "step": 14660
    },
    {
      "epoch": 2.761151891586674,
      "grad_norm": 2.0901987552642822,
      "learning_rate": 0.0002238848108413326,
      "loss": 3.5226,
      "step": 14670
    },
    {
      "epoch": 2.7630340673818936,
      "grad_norm": 3.789461374282837,
      "learning_rate": 0.00022369659326181067,
      "loss": 3.5308,
      "step": 14680
    },
    {
      "epoch": 2.7649162431771126,
      "grad_norm": 2.136140823364258,
      "learning_rate": 0.00022350837568228875,
      "loss": 3.6822,
      "step": 14690
    },
    {
      "epoch": 2.766798418972332,
      "grad_norm": 2.2833197116851807,
      "learning_rate": 0.0002233201581027668,
      "loss": 3.6,
      "step": 14700
    },
    {
      "epoch": 2.7686805947675515,
      "grad_norm": 2.147174596786499,
      "learning_rate": 0.00022313194052324487,
      "loss": 3.4034,
      "step": 14710
    },
    {
      "epoch": 2.7705627705627704,
      "grad_norm": 2.3008553981781006,
      "learning_rate": 0.00022294372294372295,
      "loss": 3.462,
      "step": 14720
    },
    {
      "epoch": 2.77244494635799,
      "grad_norm": 2.269583225250244,
      "learning_rate": 0.00022275550536420103,
      "loss": 3.3978,
      "step": 14730
    },
    {
      "epoch": 2.774327122153209,
      "grad_norm": 2.5759148597717285,
      "learning_rate": 0.0002225672877846791,
      "loss": 3.4445,
      "step": 14740
    },
    {
      "epoch": 2.7762092979484283,
      "grad_norm": 2.253596067428589,
      "learning_rate": 0.00022237907020515715,
      "loss": 3.4524,
      "step": 14750
    },
    {
      "epoch": 2.7780914737436477,
      "grad_norm": 2.3863766193389893,
      "learning_rate": 0.00022219085262563523,
      "loss": 3.5521,
      "step": 14760
    },
    {
      "epoch": 2.779973649538867,
      "grad_norm": 2.2985453605651855,
      "learning_rate": 0.0002220026350461133,
      "loss": 3.5765,
      "step": 14770
    },
    {
      "epoch": 2.781855825334086,
      "grad_norm": 3.2895071506500244,
      "learning_rate": 0.0002218144174665914,
      "loss": 3.4979,
      "step": 14780
    },
    {
      "epoch": 2.7837380011293056,
      "grad_norm": 2.392838478088379,
      "learning_rate": 0.00022162619988706946,
      "loss": 3.4905,
      "step": 14790
    },
    {
      "epoch": 2.7856201769245246,
      "grad_norm": 2.6074109077453613,
      "learning_rate": 0.00022143798230754753,
      "loss": 3.6035,
      "step": 14800
    },
    {
      "epoch": 2.787502352719744,
      "grad_norm": 2.218205690383911,
      "learning_rate": 0.0002212497647280256,
      "loss": 3.5234,
      "step": 14810
    },
    {
      "epoch": 2.7893845285149634,
      "grad_norm": 2.9633259773254395,
      "learning_rate": 0.00022106154714850369,
      "loss": 3.6222,
      "step": 14820
    },
    {
      "epoch": 2.7912667043101824,
      "grad_norm": 2.1877028942108154,
      "learning_rate": 0.00022087332956898174,
      "loss": 3.722,
      "step": 14830
    },
    {
      "epoch": 2.793148880105402,
      "grad_norm": 2.6830685138702393,
      "learning_rate": 0.0002206851119894598,
      "loss": 3.4448,
      "step": 14840
    },
    {
      "epoch": 2.795031055900621,
      "grad_norm": 2.181081771850586,
      "learning_rate": 0.0002204968944099379,
      "loss": 3.5549,
      "step": 14850
    },
    {
      "epoch": 2.7969132316958403,
      "grad_norm": 2.0416409969329834,
      "learning_rate": 0.00022030867683041596,
      "loss": 3.6628,
      "step": 14860
    },
    {
      "epoch": 2.7987954074910597,
      "grad_norm": 2.270547866821289,
      "learning_rate": 0.00022012045925089404,
      "loss": 3.378,
      "step": 14870
    },
    {
      "epoch": 2.800677583286279,
      "grad_norm": 3.774535655975342,
      "learning_rate": 0.0002199322416713721,
      "loss": 3.4061,
      "step": 14880
    },
    {
      "epoch": 2.802559759081498,
      "grad_norm": 2.869417428970337,
      "learning_rate": 0.0002197440240918502,
      "loss": 3.5096,
      "step": 14890
    },
    {
      "epoch": 2.8044419348767176,
      "grad_norm": 2.835237503051758,
      "learning_rate": 0.00021955580651232827,
      "loss": 3.3399,
      "step": 14900
    },
    {
      "epoch": 2.8063241106719365,
      "grad_norm": 2.811671495437622,
      "learning_rate": 0.00021936758893280635,
      "loss": 3.4577,
      "step": 14910
    },
    {
      "epoch": 2.808206286467156,
      "grad_norm": 2.3493731021881104,
      "learning_rate": 0.0002191793713532844,
      "loss": 3.5547,
      "step": 14920
    },
    {
      "epoch": 2.8100884622623754,
      "grad_norm": 2.611755132675171,
      "learning_rate": 0.00021899115377376247,
      "loss": 3.4613,
      "step": 14930
    },
    {
      "epoch": 2.811970638057595,
      "grad_norm": 2.213094711303711,
      "learning_rate": 0.00021880293619424055,
      "loss": 3.2579,
      "step": 14940
    },
    {
      "epoch": 2.813852813852814,
      "grad_norm": 3.3088784217834473,
      "learning_rate": 0.00021861471861471863,
      "loss": 3.4697,
      "step": 14950
    },
    {
      "epoch": 2.8157349896480333,
      "grad_norm": 2.3907647132873535,
      "learning_rate": 0.00021842650103519667,
      "loss": 3.5001,
      "step": 14960
    },
    {
      "epoch": 2.8176171654432522,
      "grad_norm": 2.6922473907470703,
      "learning_rate": 0.00021823828345567475,
      "loss": 3.2857,
      "step": 14970
    },
    {
      "epoch": 2.8194993412384717,
      "grad_norm": 2.418233633041382,
      "learning_rate": 0.00021805006587615283,
      "loss": 3.4984,
      "step": 14980
    },
    {
      "epoch": 2.821381517033691,
      "grad_norm": 2.222984790802002,
      "learning_rate": 0.0002178618482966309,
      "loss": 3.4617,
      "step": 14990
    },
    {
      "epoch": 2.82326369282891,
      "grad_norm": 2.219573974609375,
      "learning_rate": 0.000217673630717109,
      "loss": 3.5526,
      "step": 15000
    },
    {
      "epoch": 2.8251458686241295,
      "grad_norm": 2.521733522415161,
      "learning_rate": 0.00021748541313758706,
      "loss": 3.5159,
      "step": 15010
    },
    {
      "epoch": 2.8270280444193485,
      "grad_norm": 1.963568925857544,
      "learning_rate": 0.00021729719555806513,
      "loss": 3.5813,
      "step": 15020
    },
    {
      "epoch": 2.828910220214568,
      "grad_norm": 2.4848878383636475,
      "learning_rate": 0.0002171089779785432,
      "loss": 3.4577,
      "step": 15030
    },
    {
      "epoch": 2.8307923960097874,
      "grad_norm": 3.544420003890991,
      "learning_rate": 0.00021692076039902129,
      "loss": 3.4027,
      "step": 15040
    },
    {
      "epoch": 2.832674571805007,
      "grad_norm": 2.314347743988037,
      "learning_rate": 0.00021673254281949934,
      "loss": 3.5866,
      "step": 15050
    },
    {
      "epoch": 2.834556747600226,
      "grad_norm": 2.8646903038024902,
      "learning_rate": 0.0002165443252399774,
      "loss": 3.3809,
      "step": 15060
    },
    {
      "epoch": 2.8364389233954452,
      "grad_norm": 2.6082592010498047,
      "learning_rate": 0.0002163561076604555,
      "loss": 3.5331,
      "step": 15070
    },
    {
      "epoch": 2.838321099190664,
      "grad_norm": 2.5774693489074707,
      "learning_rate": 0.00021616789008093356,
      "loss": 3.484,
      "step": 15080
    },
    {
      "epoch": 2.8402032749858837,
      "grad_norm": 2.395798921585083,
      "learning_rate": 0.00021597967250141161,
      "loss": 3.4579,
      "step": 15090
    },
    {
      "epoch": 2.842085450781103,
      "grad_norm": 2.2384536266326904,
      "learning_rate": 0.0002157914549218897,
      "loss": 3.6846,
      "step": 15100
    },
    {
      "epoch": 2.843967626576322,
      "grad_norm": 1.9982672929763794,
      "learning_rate": 0.0002156032373423678,
      "loss": 3.3189,
      "step": 15110
    },
    {
      "epoch": 2.8458498023715415,
      "grad_norm": 2.47128963470459,
      "learning_rate": 0.00021541501976284587,
      "loss": 3.5565,
      "step": 15120
    },
    {
      "epoch": 2.847731978166761,
      "grad_norm": 2.4589641094207764,
      "learning_rate": 0.00021522680218332395,
      "loss": 3.4213,
      "step": 15130
    },
    {
      "epoch": 2.84961415396198,
      "grad_norm": 2.188739538192749,
      "learning_rate": 0.000215038584603802,
      "loss": 3.624,
      "step": 15140
    },
    {
      "epoch": 2.8514963297571994,
      "grad_norm": 2.4727349281311035,
      "learning_rate": 0.00021485036702428007,
      "loss": 3.5586,
      "step": 15150
    },
    {
      "epoch": 2.853378505552419,
      "grad_norm": 2.101638078689575,
      "learning_rate": 0.00021466214944475815,
      "loss": 3.2808,
      "step": 15160
    },
    {
      "epoch": 2.8552606813476378,
      "grad_norm": 2.9845190048217773,
      "learning_rate": 0.00021447393186523623,
      "loss": 3.5844,
      "step": 15170
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 2.1420416831970215,
      "learning_rate": 0.00021428571428571427,
      "loss": 3.1567,
      "step": 15180
    },
    {
      "epoch": 2.859025032938076,
      "grad_norm": 2.4418587684631348,
      "learning_rate": 0.00021409749670619235,
      "loss": 3.3212,
      "step": 15190
    },
    {
      "epoch": 2.8609072087332956,
      "grad_norm": 3.390158176422119,
      "learning_rate": 0.00021390927912667043,
      "loss": 3.3359,
      "step": 15200
    },
    {
      "epoch": 2.862789384528515,
      "grad_norm": 2.4932174682617188,
      "learning_rate": 0.0002137210615471485,
      "loss": 3.5236,
      "step": 15210
    },
    {
      "epoch": 2.8646715603237345,
      "grad_norm": 2.0627095699310303,
      "learning_rate": 0.0002135328439676266,
      "loss": 3.3669,
      "step": 15220
    },
    {
      "epoch": 2.8665537361189535,
      "grad_norm": 3.4032983779907227,
      "learning_rate": 0.00021334462638810466,
      "loss": 3.4306,
      "step": 15230
    },
    {
      "epoch": 2.868435911914173,
      "grad_norm": 2.5139946937561035,
      "learning_rate": 0.00021315640880858273,
      "loss": 3.4435,
      "step": 15240
    },
    {
      "epoch": 2.870318087709392,
      "grad_norm": 2.587301254272461,
      "learning_rate": 0.0002129681912290608,
      "loss": 3.4574,
      "step": 15250
    },
    {
      "epoch": 2.8722002635046113,
      "grad_norm": 2.299530267715454,
      "learning_rate": 0.00021277997364953889,
      "loss": 3.5394,
      "step": 15260
    },
    {
      "epoch": 2.8740824392998308,
      "grad_norm": 1.9752168655395508,
      "learning_rate": 0.00021259175607001694,
      "loss": 3.2919,
      "step": 15270
    },
    {
      "epoch": 2.8759646150950497,
      "grad_norm": 2.2354319095611572,
      "learning_rate": 0.000212403538490495,
      "loss": 3.6547,
      "step": 15280
    },
    {
      "epoch": 2.877846790890269,
      "grad_norm": 2.2502505779266357,
      "learning_rate": 0.0002122153209109731,
      "loss": 3.2796,
      "step": 15290
    },
    {
      "epoch": 2.879728966685488,
      "grad_norm": 2.381803274154663,
      "learning_rate": 0.00021202710333145116,
      "loss": 3.5734,
      "step": 15300
    },
    {
      "epoch": 2.8816111424807076,
      "grad_norm": 2.6856446266174316,
      "learning_rate": 0.00021183888575192921,
      "loss": 3.5075,
      "step": 15310
    },
    {
      "epoch": 2.883493318275927,
      "grad_norm": 2.4393763542175293,
      "learning_rate": 0.0002116506681724073,
      "loss": 3.5841,
      "step": 15320
    },
    {
      "epoch": 2.8853754940711465,
      "grad_norm": 2.8853402137756348,
      "learning_rate": 0.00021146245059288537,
      "loss": 3.5743,
      "step": 15330
    },
    {
      "epoch": 2.8872576698663655,
      "grad_norm": 2.005941152572632,
      "learning_rate": 0.00021127423301336347,
      "loss": 3.4861,
      "step": 15340
    },
    {
      "epoch": 2.889139845661585,
      "grad_norm": 2.736389398574829,
      "learning_rate": 0.00021108601543384155,
      "loss": 3.3339,
      "step": 15350
    },
    {
      "epoch": 2.891022021456804,
      "grad_norm": 2.5647521018981934,
      "learning_rate": 0.0002108977978543196,
      "loss": 3.4417,
      "step": 15360
    },
    {
      "epoch": 2.8929041972520233,
      "grad_norm": 2.37971568107605,
      "learning_rate": 0.00021070958027479767,
      "loss": 3.3668,
      "step": 15370
    },
    {
      "epoch": 2.8947863730472427,
      "grad_norm": 2.441014289855957,
      "learning_rate": 0.00021052136269527575,
      "loss": 3.5414,
      "step": 15380
    },
    {
      "epoch": 2.8966685488424617,
      "grad_norm": 3.1848180294036865,
      "learning_rate": 0.00021033314511575383,
      "loss": 3.6509,
      "step": 15390
    },
    {
      "epoch": 2.898550724637681,
      "grad_norm": 2.5912351608276367,
      "learning_rate": 0.00021014492753623187,
      "loss": 3.2655,
      "step": 15400
    },
    {
      "epoch": 2.9004329004329006,
      "grad_norm": 3.1243958473205566,
      "learning_rate": 0.00020995670995670995,
      "loss": 3.6256,
      "step": 15410
    },
    {
      "epoch": 2.9023150762281196,
      "grad_norm": 2.3207552433013916,
      "learning_rate": 0.00020976849237718803,
      "loss": 3.1971,
      "step": 15420
    },
    {
      "epoch": 2.904197252023339,
      "grad_norm": 2.6017887592315674,
      "learning_rate": 0.0002095802747976661,
      "loss": 3.61,
      "step": 15430
    },
    {
      "epoch": 2.9060794278185584,
      "grad_norm": 2.248102903366089,
      "learning_rate": 0.00020939205721814415,
      "loss": 3.5052,
      "step": 15440
    },
    {
      "epoch": 2.9079616036137774,
      "grad_norm": 2.675691604614258,
      "learning_rate": 0.00020920383963862226,
      "loss": 3.5116,
      "step": 15450
    },
    {
      "epoch": 2.909843779408997,
      "grad_norm": 2.4814398288726807,
      "learning_rate": 0.00020901562205910033,
      "loss": 3.2238,
      "step": 15460
    },
    {
      "epoch": 2.911725955204216,
      "grad_norm": 2.3238048553466797,
      "learning_rate": 0.0002088274044795784,
      "loss": 3.3525,
      "step": 15470
    },
    {
      "epoch": 2.9136081309994353,
      "grad_norm": 2.466846466064453,
      "learning_rate": 0.00020863918690005649,
      "loss": 3.6302,
      "step": 15480
    },
    {
      "epoch": 2.9154903067946547,
      "grad_norm": 2.3691577911376953,
      "learning_rate": 0.00020845096932053454,
      "loss": 3.1262,
      "step": 15490
    },
    {
      "epoch": 2.917372482589874,
      "grad_norm": 2.283759593963623,
      "learning_rate": 0.0002082627517410126,
      "loss": 3.2809,
      "step": 15500
    },
    {
      "epoch": 2.919254658385093,
      "grad_norm": 2.8975417613983154,
      "learning_rate": 0.0002080745341614907,
      "loss": 3.6239,
      "step": 15510
    },
    {
      "epoch": 2.9211368341803126,
      "grad_norm": 2.6288318634033203,
      "learning_rate": 0.00020788631658196876,
      "loss": 3.6008,
      "step": 15520
    },
    {
      "epoch": 2.9230190099755315,
      "grad_norm": 2.5137789249420166,
      "learning_rate": 0.00020769809900244681,
      "loss": 3.4699,
      "step": 15530
    },
    {
      "epoch": 2.924901185770751,
      "grad_norm": 2.2313966751098633,
      "learning_rate": 0.0002075098814229249,
      "loss": 3.4451,
      "step": 15540
    },
    {
      "epoch": 2.9267833615659704,
      "grad_norm": 2.7938575744628906,
      "learning_rate": 0.00020732166384340297,
      "loss": 3.6125,
      "step": 15550
    },
    {
      "epoch": 2.9286655373611894,
      "grad_norm": 2.416111946105957,
      "learning_rate": 0.00020713344626388107,
      "loss": 3.5303,
      "step": 15560
    },
    {
      "epoch": 2.930547713156409,
      "grad_norm": 3.317490816116333,
      "learning_rate": 0.00020694522868435912,
      "loss": 3.3833,
      "step": 15570
    },
    {
      "epoch": 2.932429888951628,
      "grad_norm": 2.4095804691314697,
      "learning_rate": 0.0002067570111048372,
      "loss": 3.5556,
      "step": 15580
    },
    {
      "epoch": 2.9343120647468472,
      "grad_norm": 2.984218120574951,
      "learning_rate": 0.00020656879352531527,
      "loss": 3.2939,
      "step": 15590
    },
    {
      "epoch": 2.9361942405420667,
      "grad_norm": 3.039593458175659,
      "learning_rate": 0.00020638057594579335,
      "loss": 3.792,
      "step": 15600
    },
    {
      "epoch": 2.938076416337286,
      "grad_norm": 2.3210456371307373,
      "learning_rate": 0.00020619235836627143,
      "loss": 3.4311,
      "step": 15610
    },
    {
      "epoch": 2.939958592132505,
      "grad_norm": 3.5044655799865723,
      "learning_rate": 0.00020600414078674947,
      "loss": 3.6649,
      "step": 15620
    },
    {
      "epoch": 2.9418407679277245,
      "grad_norm": 2.563467264175415,
      "learning_rate": 0.00020581592320722755,
      "loss": 3.7411,
      "step": 15630
    },
    {
      "epoch": 2.9437229437229435,
      "grad_norm": 2.5478553771972656,
      "learning_rate": 0.00020562770562770563,
      "loss": 3.6339,
      "step": 15640
    },
    {
      "epoch": 2.945605119518163,
      "grad_norm": 2.3911826610565186,
      "learning_rate": 0.0002054394880481837,
      "loss": 3.588,
      "step": 15650
    },
    {
      "epoch": 2.9474872953133824,
      "grad_norm": 2.204178810119629,
      "learning_rate": 0.00020525127046866175,
      "loss": 3.2706,
      "step": 15660
    },
    {
      "epoch": 2.949369471108602,
      "grad_norm": 2.1773793697357178,
      "learning_rate": 0.00020506305288913986,
      "loss": 3.5213,
      "step": 15670
    },
    {
      "epoch": 2.951251646903821,
      "grad_norm": 2.37015962600708,
      "learning_rate": 0.00020487483530961793,
      "loss": 3.5603,
      "step": 15680
    },
    {
      "epoch": 2.9531338226990402,
      "grad_norm": 2.6659293174743652,
      "learning_rate": 0.000204686617730096,
      "loss": 3.2855,
      "step": 15690
    },
    {
      "epoch": 2.955015998494259,
      "grad_norm": 2.7114241123199463,
      "learning_rate": 0.00020449840015057406,
      "loss": 3.2266,
      "step": 15700
    },
    {
      "epoch": 2.9568981742894787,
      "grad_norm": 2.4781360626220703,
      "learning_rate": 0.00020431018257105214,
      "loss": 3.3296,
      "step": 15710
    },
    {
      "epoch": 2.958780350084698,
      "grad_norm": 2.8543200492858887,
      "learning_rate": 0.0002041219649915302,
      "loss": 3.384,
      "step": 15720
    },
    {
      "epoch": 2.960662525879917,
      "grad_norm": 2.0750515460968018,
      "learning_rate": 0.0002039337474120083,
      "loss": 3.3709,
      "step": 15730
    },
    {
      "epoch": 2.9625447016751365,
      "grad_norm": 2.2768306732177734,
      "learning_rate": 0.00020374552983248636,
      "loss": 3.5925,
      "step": 15740
    },
    {
      "epoch": 2.9644268774703555,
      "grad_norm": 3.9188880920410156,
      "learning_rate": 0.00020355731225296441,
      "loss": 3.3453,
      "step": 15750
    },
    {
      "epoch": 2.966309053265575,
      "grad_norm": 2.0652997493743896,
      "learning_rate": 0.0002033690946734425,
      "loss": 3.4702,
      "step": 15760
    },
    {
      "epoch": 2.9681912290607944,
      "grad_norm": 2.6457831859588623,
      "learning_rate": 0.00020318087709392057,
      "loss": 3.6202,
      "step": 15770
    },
    {
      "epoch": 2.970073404856014,
      "grad_norm": 2.8359549045562744,
      "learning_rate": 0.00020299265951439867,
      "loss": 3.4095,
      "step": 15780
    },
    {
      "epoch": 2.9719555806512328,
      "grad_norm": 2.460312843322754,
      "learning_rate": 0.00020280444193487672,
      "loss": 3.2642,
      "step": 15790
    },
    {
      "epoch": 2.973837756446452,
      "grad_norm": 2.7726497650146484,
      "learning_rate": 0.0002026162243553548,
      "loss": 3.2432,
      "step": 15800
    },
    {
      "epoch": 2.975719932241671,
      "grad_norm": 2.8087449073791504,
      "learning_rate": 0.00020242800677583287,
      "loss": 3.9875,
      "step": 15810
    },
    {
      "epoch": 2.9776021080368906,
      "grad_norm": 2.3682594299316406,
      "learning_rate": 0.00020223978919631095,
      "loss": 3.2547,
      "step": 15820
    },
    {
      "epoch": 2.97948428383211,
      "grad_norm": 2.968564748764038,
      "learning_rate": 0.00020205157161678903,
      "loss": 3.3891,
      "step": 15830
    },
    {
      "epoch": 2.981366459627329,
      "grad_norm": 2.396745204925537,
      "learning_rate": 0.00020186335403726707,
      "loss": 3.4696,
      "step": 15840
    },
    {
      "epoch": 2.9832486354225485,
      "grad_norm": 3.5899250507354736,
      "learning_rate": 0.00020167513645774515,
      "loss": 3.6297,
      "step": 15850
    },
    {
      "epoch": 2.9851308112177675,
      "grad_norm": 2.914106607437134,
      "learning_rate": 0.00020148691887822323,
      "loss": 3.2882,
      "step": 15860
    },
    {
      "epoch": 2.987012987012987,
      "grad_norm": 2.3831381797790527,
      "learning_rate": 0.0002012987012987013,
      "loss": 3.6653,
      "step": 15870
    },
    {
      "epoch": 2.9888951628082063,
      "grad_norm": 2.077132225036621,
      "learning_rate": 0.00020111048371917935,
      "loss": 3.3957,
      "step": 15880
    },
    {
      "epoch": 2.9907773386034258,
      "grad_norm": 2.5786349773406982,
      "learning_rate": 0.00020092226613965746,
      "loss": 3.5659,
      "step": 15890
    },
    {
      "epoch": 2.9926595143986447,
      "grad_norm": 2.3395864963531494,
      "learning_rate": 0.00020073404856013553,
      "loss": 3.2228,
      "step": 15900
    },
    {
      "epoch": 2.994541690193864,
      "grad_norm": 2.300260066986084,
      "learning_rate": 0.0002005458309806136,
      "loss": 3.336,
      "step": 15910
    },
    {
      "epoch": 2.996423865989083,
      "grad_norm": 3.540156364440918,
      "learning_rate": 0.00020035761340109166,
      "loss": 3.3551,
      "step": 15920
    },
    {
      "epoch": 2.9983060417843026,
      "grad_norm": 2.3956615924835205,
      "learning_rate": 0.00020016939582156974,
      "loss": 3.3675,
      "step": 15930
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.166,
      "eval_loss": 3.4389488697052,
      "eval_runtime": 87.0069,
      "eval_samples_per_second": 86.2,
      "eval_steps_per_second": 10.781,
      "step": 15939
    },
    {
      "epoch": 3.000188217579522,
      "grad_norm": 2.3464903831481934,
      "learning_rate": 0.0001999811782420478,
      "loss": 3.5502,
      "step": 15940
    },
    {
      "epoch": 3.002070393374741,
      "grad_norm": 2.4275286197662354,
      "learning_rate": 0.0001997929606625259,
      "loss": 3.3322,
      "step": 15950
    },
    {
      "epoch": 3.0039525691699605,
      "grad_norm": 2.139082908630371,
      "learning_rate": 0.00019960474308300396,
      "loss": 3.2053,
      "step": 15960
    },
    {
      "epoch": 3.00583474496518,
      "grad_norm": 2.784451961517334,
      "learning_rate": 0.00019941652550348201,
      "loss": 3.5752,
      "step": 15970
    },
    {
      "epoch": 3.007716920760399,
      "grad_norm": 2.6467702388763428,
      "learning_rate": 0.0001992283079239601,
      "loss": 3.346,
      "step": 15980
    },
    {
      "epoch": 3.0095990965556183,
      "grad_norm": 2.888674736022949,
      "learning_rate": 0.00019904009034443817,
      "loss": 3.4121,
      "step": 15990
    },
    {
      "epoch": 3.0114812723508377,
      "grad_norm": 2.2215163707733154,
      "learning_rate": 0.00019885187276491627,
      "loss": 3.5202,
      "step": 16000
    },
    {
      "epoch": 3.0133634481460567,
      "grad_norm": 3.2592546939849854,
      "learning_rate": 0.00019866365518539432,
      "loss": 3.1009,
      "step": 16010
    },
    {
      "epoch": 3.015245623941276,
      "grad_norm": 2.3010451793670654,
      "learning_rate": 0.0001984754376058724,
      "loss": 3.3903,
      "step": 16020
    },
    {
      "epoch": 3.0171277997364956,
      "grad_norm": 2.2252073287963867,
      "learning_rate": 0.00019828722002635047,
      "loss": 3.3718,
      "step": 16030
    },
    {
      "epoch": 3.0190099755317146,
      "grad_norm": 2.8217551708221436,
      "learning_rate": 0.00019809900244682855,
      "loss": 3.2882,
      "step": 16040
    },
    {
      "epoch": 3.020892151326934,
      "grad_norm": 2.717163324356079,
      "learning_rate": 0.0001979107848673066,
      "loss": 3.1833,
      "step": 16050
    },
    {
      "epoch": 3.022774327122153,
      "grad_norm": 2.6875417232513428,
      "learning_rate": 0.00019772256728778467,
      "loss": 3.6372,
      "step": 16060
    },
    {
      "epoch": 3.0246565029173724,
      "grad_norm": 2.6113834381103516,
      "learning_rate": 0.00019753434970826275,
      "loss": 3.2069,
      "step": 16070
    },
    {
      "epoch": 3.026538678712592,
      "grad_norm": 3.7651712894439697,
      "learning_rate": 0.00019734613212874083,
      "loss": 3.5204,
      "step": 16080
    },
    {
      "epoch": 3.028420854507811,
      "grad_norm": 3.171018600463867,
      "learning_rate": 0.0001971579145492189,
      "loss": 3.4848,
      "step": 16090
    },
    {
      "epoch": 3.0303030303030303,
      "grad_norm": 2.6116936206817627,
      "learning_rate": 0.00019696969696969695,
      "loss": 3.525,
      "step": 16100
    },
    {
      "epoch": 3.0321852060982497,
      "grad_norm": 2.1383917331695557,
      "learning_rate": 0.00019678147939017506,
      "loss": 3.1989,
      "step": 16110
    },
    {
      "epoch": 3.0340673818934687,
      "grad_norm": 2.373936891555786,
      "learning_rate": 0.00019659326181065313,
      "loss": 3.3162,
      "step": 16120
    },
    {
      "epoch": 3.035949557688688,
      "grad_norm": 1.9612059593200684,
      "learning_rate": 0.0001964050442311312,
      "loss": 3.58,
      "step": 16130
    },
    {
      "epoch": 3.0378317334839076,
      "grad_norm": 2.311664342880249,
      "learning_rate": 0.00019621682665160926,
      "loss": 3.5013,
      "step": 16140
    },
    {
      "epoch": 3.0397139092791265,
      "grad_norm": 2.8051741123199463,
      "learning_rate": 0.00019602860907208734,
      "loss": 3.3944,
      "step": 16150
    },
    {
      "epoch": 3.041596085074346,
      "grad_norm": 2.781449794769287,
      "learning_rate": 0.0001958403914925654,
      "loss": 3.2989,
      "step": 16160
    },
    {
      "epoch": 3.0434782608695654,
      "grad_norm": 2.8339180946350098,
      "learning_rate": 0.0001956521739130435,
      "loss": 3.513,
      "step": 16170
    },
    {
      "epoch": 3.0453604366647844,
      "grad_norm": 2.3824877738952637,
      "learning_rate": 0.00019546395633352154,
      "loss": 3.4308,
      "step": 16180
    },
    {
      "epoch": 3.047242612460004,
      "grad_norm": 2.3761489391326904,
      "learning_rate": 0.00019527573875399961,
      "loss": 3.2124,
      "step": 16190
    },
    {
      "epoch": 3.049124788255223,
      "grad_norm": 2.7145802974700928,
      "learning_rate": 0.0001950875211744777,
      "loss": 3.2744,
      "step": 16200
    },
    {
      "epoch": 3.0510069640504422,
      "grad_norm": 3.7028284072875977,
      "learning_rate": 0.00019489930359495577,
      "loss": 3.3786,
      "step": 16210
    },
    {
      "epoch": 3.0528891398456617,
      "grad_norm": 2.8431806564331055,
      "learning_rate": 0.00019471108601543387,
      "loss": 3.5511,
      "step": 16220
    },
    {
      "epoch": 3.0547713156408807,
      "grad_norm": 2.4497873783111572,
      "learning_rate": 0.00019452286843591192,
      "loss": 3.4525,
      "step": 16230
    },
    {
      "epoch": 3.0566534914361,
      "grad_norm": 3.198949098587036,
      "learning_rate": 0.00019433465085639,
      "loss": 3.2419,
      "step": 16240
    },
    {
      "epoch": 3.0585356672313195,
      "grad_norm": 2.400282859802246,
      "learning_rate": 0.00019414643327686807,
      "loss": 3.2808,
      "step": 16250
    },
    {
      "epoch": 3.0604178430265385,
      "grad_norm": 3.089738607406616,
      "learning_rate": 0.00019395821569734615,
      "loss": 3.5481,
      "step": 16260
    },
    {
      "epoch": 3.062300018821758,
      "grad_norm": 2.3858842849731445,
      "learning_rate": 0.0001937699981178242,
      "loss": 3.4233,
      "step": 16270
    },
    {
      "epoch": 3.0641821946169774,
      "grad_norm": 2.711099863052368,
      "learning_rate": 0.00019358178053830227,
      "loss": 3.7983,
      "step": 16280
    },
    {
      "epoch": 3.0660643704121964,
      "grad_norm": 2.758885622024536,
      "learning_rate": 0.00019339356295878035,
      "loss": 3.5862,
      "step": 16290
    },
    {
      "epoch": 3.067946546207416,
      "grad_norm": 2.564115285873413,
      "learning_rate": 0.00019320534537925843,
      "loss": 3.4485,
      "step": 16300
    },
    {
      "epoch": 3.0698287220026352,
      "grad_norm": 2.551232099533081,
      "learning_rate": 0.00019301712779973648,
      "loss": 3.2021,
      "step": 16310
    },
    {
      "epoch": 3.0717108977978542,
      "grad_norm": 2.466330051422119,
      "learning_rate": 0.00019282891022021455,
      "loss": 3.5043,
      "step": 16320
    },
    {
      "epoch": 3.0735930735930737,
      "grad_norm": 2.2110447883605957,
      "learning_rate": 0.00019264069264069266,
      "loss": 3.5946,
      "step": 16330
    },
    {
      "epoch": 3.075475249388293,
      "grad_norm": 2.7100703716278076,
      "learning_rate": 0.00019245247506117073,
      "loss": 3.4292,
      "step": 16340
    },
    {
      "epoch": 3.077357425183512,
      "grad_norm": 2.8344309329986572,
      "learning_rate": 0.0001922642574816488,
      "loss": 3.4147,
      "step": 16350
    },
    {
      "epoch": 3.0792396009787315,
      "grad_norm": 2.4452314376831055,
      "learning_rate": 0.00019207603990212686,
      "loss": 3.2555,
      "step": 16360
    },
    {
      "epoch": 3.0811217767739505,
      "grad_norm": 2.6006791591644287,
      "learning_rate": 0.00019188782232260494,
      "loss": 3.4488,
      "step": 16370
    },
    {
      "epoch": 3.08300395256917,
      "grad_norm": 2.4451346397399902,
      "learning_rate": 0.000191699604743083,
      "loss": 3.3258,
      "step": 16380
    },
    {
      "epoch": 3.0848861283643894,
      "grad_norm": 2.379737615585327,
      "learning_rate": 0.0001915113871635611,
      "loss": 3.6737,
      "step": 16390
    },
    {
      "epoch": 3.0867683041596083,
      "grad_norm": 2.5672545433044434,
      "learning_rate": 0.00019132316958403914,
      "loss": 3.4565,
      "step": 16400
    },
    {
      "epoch": 3.0886504799548278,
      "grad_norm": 2.4402852058410645,
      "learning_rate": 0.00019113495200451721,
      "loss": 3.6682,
      "step": 16410
    },
    {
      "epoch": 3.090532655750047,
      "grad_norm": 3.9088499546051025,
      "learning_rate": 0.0001909467344249953,
      "loss": 3.2161,
      "step": 16420
    },
    {
      "epoch": 3.092414831545266,
      "grad_norm": 2.7074153423309326,
      "learning_rate": 0.00019075851684547337,
      "loss": 3.1454,
      "step": 16430
    },
    {
      "epoch": 3.0942970073404856,
      "grad_norm": 2.2738146781921387,
      "learning_rate": 0.00019057029926595144,
      "loss": 3.196,
      "step": 16440
    },
    {
      "epoch": 3.096179183135705,
      "grad_norm": 2.12372088432312,
      "learning_rate": 0.00019038208168642952,
      "loss": 3.172,
      "step": 16450
    },
    {
      "epoch": 3.098061358930924,
      "grad_norm": 2.799208641052246,
      "learning_rate": 0.0001901938641069076,
      "loss": 3.3201,
      "step": 16460
    },
    {
      "epoch": 3.0999435347261435,
      "grad_norm": 2.7520501613616943,
      "learning_rate": 0.00019000564652738567,
      "loss": 3.4104,
      "step": 16470
    },
    {
      "epoch": 3.1018257105213625,
      "grad_norm": 2.407956838607788,
      "learning_rate": 0.00018981742894786375,
      "loss": 3.2854,
      "step": 16480
    },
    {
      "epoch": 3.103707886316582,
      "grad_norm": 2.214393377304077,
      "learning_rate": 0.0001896292113683418,
      "loss": 3.1286,
      "step": 16490
    },
    {
      "epoch": 3.1055900621118013,
      "grad_norm": 2.534752130508423,
      "learning_rate": 0.00018944099378881987,
      "loss": 3.4013,
      "step": 16500
    },
    {
      "epoch": 3.1074722379070203,
      "grad_norm": 2.607753038406372,
      "learning_rate": 0.00018925277620929795,
      "loss": 3.5454,
      "step": 16510
    },
    {
      "epoch": 3.1093544137022397,
      "grad_norm": 2.6956064701080322,
      "learning_rate": 0.00018906455862977603,
      "loss": 3.354,
      "step": 16520
    },
    {
      "epoch": 3.111236589497459,
      "grad_norm": 2.5441951751708984,
      "learning_rate": 0.00018887634105025408,
      "loss": 3.5804,
      "step": 16530
    },
    {
      "epoch": 3.113118765292678,
      "grad_norm": 2.281205177307129,
      "learning_rate": 0.00018868812347073215,
      "loss": 3.3036,
      "step": 16540
    },
    {
      "epoch": 3.1150009410878976,
      "grad_norm": 2.4613826274871826,
      "learning_rate": 0.00018849990589121026,
      "loss": 3.4355,
      "step": 16550
    },
    {
      "epoch": 3.116883116883117,
      "grad_norm": 2.2320241928100586,
      "learning_rate": 0.00018831168831168833,
      "loss": 3.5351,
      "step": 16560
    },
    {
      "epoch": 3.118765292678336,
      "grad_norm": 2.5939552783966064,
      "learning_rate": 0.0001881234707321664,
      "loss": 3.3774,
      "step": 16570
    },
    {
      "epoch": 3.1206474684735555,
      "grad_norm": 2.816632032394409,
      "learning_rate": 0.00018793525315264446,
      "loss": 3.5098,
      "step": 16580
    },
    {
      "epoch": 3.122529644268775,
      "grad_norm": 2.7928173542022705,
      "learning_rate": 0.00018774703557312254,
      "loss": 3.0776,
      "step": 16590
    },
    {
      "epoch": 3.124411820063994,
      "grad_norm": 1.9923349618911743,
      "learning_rate": 0.0001875588179936006,
      "loss": 3.6454,
      "step": 16600
    },
    {
      "epoch": 3.1262939958592133,
      "grad_norm": 2.296696901321411,
      "learning_rate": 0.0001873706004140787,
      "loss": 3.527,
      "step": 16610
    },
    {
      "epoch": 3.1281761716544327,
      "grad_norm": 2.5350286960601807,
      "learning_rate": 0.00018718238283455674,
      "loss": 3.5192,
      "step": 16620
    },
    {
      "epoch": 3.1300583474496517,
      "grad_norm": 2.438202381134033,
      "learning_rate": 0.00018699416525503481,
      "loss": 3.3552,
      "step": 16630
    },
    {
      "epoch": 3.131940523244871,
      "grad_norm": 2.678471565246582,
      "learning_rate": 0.0001868059476755129,
      "loss": 3.5464,
      "step": 16640
    },
    {
      "epoch": 3.13382269904009,
      "grad_norm": 2.5259385108947754,
      "learning_rate": 0.00018661773009599097,
      "loss": 3.308,
      "step": 16650
    },
    {
      "epoch": 3.1357048748353096,
      "grad_norm": 2.369147539138794,
      "learning_rate": 0.00018642951251646904,
      "loss": 3.3693,
      "step": 16660
    },
    {
      "epoch": 3.137587050630529,
      "grad_norm": 2.7190983295440674,
      "learning_rate": 0.00018624129493694712,
      "loss": 3.3369,
      "step": 16670
    },
    {
      "epoch": 3.139469226425748,
      "grad_norm": 3.0043201446533203,
      "learning_rate": 0.0001860530773574252,
      "loss": 3.4178,
      "step": 16680
    },
    {
      "epoch": 3.1413514022209674,
      "grad_norm": 2.472367525100708,
      "learning_rate": 0.00018586485977790327,
      "loss": 3.9501,
      "step": 16690
    },
    {
      "epoch": 3.143233578016187,
      "grad_norm": 2.0951249599456787,
      "learning_rate": 0.00018567664219838135,
      "loss": 3.4487,
      "step": 16700
    },
    {
      "epoch": 3.145115753811406,
      "grad_norm": 2.3645644187927246,
      "learning_rate": 0.0001854884246188594,
      "loss": 3.5567,
      "step": 16710
    },
    {
      "epoch": 3.1469979296066253,
      "grad_norm": 2.6005163192749023,
      "learning_rate": 0.00018530020703933747,
      "loss": 3.1231,
      "step": 16720
    },
    {
      "epoch": 3.1488801054018447,
      "grad_norm": 2.6122794151306152,
      "learning_rate": 0.00018511198945981555,
      "loss": 2.9662,
      "step": 16730
    },
    {
      "epoch": 3.1507622811970637,
      "grad_norm": 2.8639791011810303,
      "learning_rate": 0.00018492377188029363,
      "loss": 3.5101,
      "step": 16740
    },
    {
      "epoch": 3.152644456992283,
      "grad_norm": 2.615168571472168,
      "learning_rate": 0.00018473555430077168,
      "loss": 3.3929,
      "step": 16750
    },
    {
      "epoch": 3.1545266327875026,
      "grad_norm": 2.6724283695220947,
      "learning_rate": 0.00018454733672124975,
      "loss": 3.3804,
      "step": 16760
    },
    {
      "epoch": 3.1564088085827215,
      "grad_norm": 2.6021275520324707,
      "learning_rate": 0.00018435911914172786,
      "loss": 3.5446,
      "step": 16770
    },
    {
      "epoch": 3.158290984377941,
      "grad_norm": 2.5110085010528564,
      "learning_rate": 0.00018417090156220593,
      "loss": 3.372,
      "step": 16780
    },
    {
      "epoch": 3.16017316017316,
      "grad_norm": 2.4576356410980225,
      "learning_rate": 0.00018398268398268398,
      "loss": 3.3233,
      "step": 16790
    },
    {
      "epoch": 3.1620553359683794,
      "grad_norm": 2.512538433074951,
      "learning_rate": 0.00018379446640316206,
      "loss": 3.5851,
      "step": 16800
    },
    {
      "epoch": 3.163937511763599,
      "grad_norm": 2.883690595626831,
      "learning_rate": 0.00018360624882364014,
      "loss": 3.7707,
      "step": 16810
    },
    {
      "epoch": 3.165819687558818,
      "grad_norm": 2.1599953174591064,
      "learning_rate": 0.0001834180312441182,
      "loss": 3.273,
      "step": 16820
    },
    {
      "epoch": 3.1677018633540373,
      "grad_norm": 2.552882432937622,
      "learning_rate": 0.0001832298136645963,
      "loss": 3.1311,
      "step": 16830
    },
    {
      "epoch": 3.1695840391492567,
      "grad_norm": 3.2530431747436523,
      "learning_rate": 0.00018304159608507434,
      "loss": 3.5942,
      "step": 16840
    },
    {
      "epoch": 3.1714662149444757,
      "grad_norm": 1.7661073207855225,
      "learning_rate": 0.00018285337850555241,
      "loss": 3.3458,
      "step": 16850
    },
    {
      "epoch": 3.173348390739695,
      "grad_norm": 2.1382436752319336,
      "learning_rate": 0.0001826651609260305,
      "loss": 2.9106,
      "step": 16860
    },
    {
      "epoch": 3.1752305665349145,
      "grad_norm": 2.80940580368042,
      "learning_rate": 0.00018247694334650857,
      "loss": 3.3798,
      "step": 16870
    },
    {
      "epoch": 3.1771127423301335,
      "grad_norm": 2.5640830993652344,
      "learning_rate": 0.00018228872576698662,
      "loss": 3.1892,
      "step": 16880
    },
    {
      "epoch": 3.178994918125353,
      "grad_norm": 2.5059783458709717,
      "learning_rate": 0.00018210050818746472,
      "loss": 3.5314,
      "step": 16890
    },
    {
      "epoch": 3.1808770939205724,
      "grad_norm": 2.6395926475524902,
      "learning_rate": 0.0001819122906079428,
      "loss": 3.5268,
      "step": 16900
    },
    {
      "epoch": 3.1827592697157914,
      "grad_norm": 2.771196126937866,
      "learning_rate": 0.00018172407302842087,
      "loss": 3.3903,
      "step": 16910
    },
    {
      "epoch": 3.184641445511011,
      "grad_norm": 2.2118568420410156,
      "learning_rate": 0.00018153585544889892,
      "loss": 3.5801,
      "step": 16920
    },
    {
      "epoch": 3.18652362130623,
      "grad_norm": 3.436131238937378,
      "learning_rate": 0.000181347637869377,
      "loss": 3.2285,
      "step": 16930
    },
    {
      "epoch": 3.1884057971014492,
      "grad_norm": 2.3162379264831543,
      "learning_rate": 0.00018115942028985507,
      "loss": 3.5493,
      "step": 16940
    },
    {
      "epoch": 3.1902879728966687,
      "grad_norm": 2.5042316913604736,
      "learning_rate": 0.00018097120271033315,
      "loss": 3.8026,
      "step": 16950
    },
    {
      "epoch": 3.1921701486918876,
      "grad_norm": 2.483301877975464,
      "learning_rate": 0.00018078298513081123,
      "loss": 3.2815,
      "step": 16960
    },
    {
      "epoch": 3.194052324487107,
      "grad_norm": 2.8055741786956787,
      "learning_rate": 0.00018059476755128928,
      "loss": 3.1948,
      "step": 16970
    },
    {
      "epoch": 3.1959345002823265,
      "grad_norm": 3.0562124252319336,
      "learning_rate": 0.00018040654997176735,
      "loss": 3.2669,
      "step": 16980
    },
    {
      "epoch": 3.1978166760775455,
      "grad_norm": 2.5893020629882812,
      "learning_rate": 0.00018021833239224543,
      "loss": 3.4459,
      "step": 16990
    },
    {
      "epoch": 3.199698851872765,
      "grad_norm": 3.113239049911499,
      "learning_rate": 0.00018003011481272353,
      "loss": 3.3262,
      "step": 17000
    },
    {
      "epoch": 3.2015810276679844,
      "grad_norm": 2.3117594718933105,
      "learning_rate": 0.00017984189723320158,
      "loss": 3.1799,
      "step": 17010
    },
    {
      "epoch": 3.2034632034632033,
      "grad_norm": 2.710742712020874,
      "learning_rate": 0.00017965367965367966,
      "loss": 3.4581,
      "step": 17020
    },
    {
      "epoch": 3.2053453792584228,
      "grad_norm": 2.3555855751037598,
      "learning_rate": 0.00017946546207415774,
      "loss": 3.1941,
      "step": 17030
    },
    {
      "epoch": 3.207227555053642,
      "grad_norm": 2.6905875205993652,
      "learning_rate": 0.0001792772444946358,
      "loss": 3.4441,
      "step": 17040
    },
    {
      "epoch": 3.209109730848861,
      "grad_norm": 2.401326894760132,
      "learning_rate": 0.00017908902691511386,
      "loss": 3.2281,
      "step": 17050
    },
    {
      "epoch": 3.2109919066440806,
      "grad_norm": 2.5542731285095215,
      "learning_rate": 0.00017890080933559194,
      "loss": 3.5194,
      "step": 17060
    },
    {
      "epoch": 3.2128740824393,
      "grad_norm": 2.9633452892303467,
      "learning_rate": 0.00017871259175607001,
      "loss": 3.4064,
      "step": 17070
    },
    {
      "epoch": 3.214756258234519,
      "grad_norm": 2.62150502204895,
      "learning_rate": 0.0001785243741765481,
      "loss": 3.4583,
      "step": 17080
    },
    {
      "epoch": 3.2166384340297385,
      "grad_norm": 2.9389212131500244,
      "learning_rate": 0.00017833615659702617,
      "loss": 3.4426,
      "step": 17090
    },
    {
      "epoch": 3.2185206098249575,
      "grad_norm": 2.7606170177459717,
      "learning_rate": 0.00017814793901750422,
      "loss": 3.3341,
      "step": 17100
    },
    {
      "epoch": 3.220402785620177,
      "grad_norm": 3.0567126274108887,
      "learning_rate": 0.00017795972143798232,
      "loss": 3.1219,
      "step": 17110
    },
    {
      "epoch": 3.2222849614153963,
      "grad_norm": 2.618769645690918,
      "learning_rate": 0.0001777715038584604,
      "loss": 3.3479,
      "step": 17120
    },
    {
      "epoch": 3.2241671372106153,
      "grad_norm": 3.627694606781006,
      "learning_rate": 0.00017758328627893847,
      "loss": 3.4016,
      "step": 17130
    },
    {
      "epoch": 3.2260493130058348,
      "grad_norm": 2.895933151245117,
      "learning_rate": 0.00017739506869941652,
      "loss": 3.4261,
      "step": 17140
    },
    {
      "epoch": 3.227931488801054,
      "grad_norm": 2.9938414096832275,
      "learning_rate": 0.0001772068511198946,
      "loss": 3.2197,
      "step": 17150
    },
    {
      "epoch": 3.229813664596273,
      "grad_norm": 2.576021194458008,
      "learning_rate": 0.00017701863354037267,
      "loss": 3.6523,
      "step": 17160
    },
    {
      "epoch": 3.2316958403914926,
      "grad_norm": 2.8756418228149414,
      "learning_rate": 0.00017683041596085075,
      "loss": 3.3199,
      "step": 17170
    },
    {
      "epoch": 3.233578016186712,
      "grad_norm": 4.0849289894104,
      "learning_rate": 0.00017664219838132883,
      "loss": 3.7093,
      "step": 17180
    },
    {
      "epoch": 3.235460191981931,
      "grad_norm": 2.724503993988037,
      "learning_rate": 0.00017645398080180688,
      "loss": 3.3134,
      "step": 17190
    },
    {
      "epoch": 3.2373423677771505,
      "grad_norm": 3.1503915786743164,
      "learning_rate": 0.00017626576322228495,
      "loss": 3.5615,
      "step": 17200
    },
    {
      "epoch": 3.2392245435723694,
      "grad_norm": 2.5145812034606934,
      "learning_rate": 0.00017607754564276303,
      "loss": 3.3714,
      "step": 17210
    },
    {
      "epoch": 3.241106719367589,
      "grad_norm": 2.7142868041992188,
      "learning_rate": 0.00017588932806324113,
      "loss": 3.6091,
      "step": 17220
    },
    {
      "epoch": 3.2429888951628083,
      "grad_norm": 3.1583940982818604,
      "learning_rate": 0.00017570111048371918,
      "loss": 3.3126,
      "step": 17230
    },
    {
      "epoch": 3.2448710709580273,
      "grad_norm": 2.589491128921509,
      "learning_rate": 0.00017551289290419726,
      "loss": 3.4296,
      "step": 17240
    },
    {
      "epoch": 3.2467532467532467,
      "grad_norm": 2.6531646251678467,
      "learning_rate": 0.00017532467532467534,
      "loss": 3.2807,
      "step": 17250
    },
    {
      "epoch": 3.248635422548466,
      "grad_norm": 2.35854434967041,
      "learning_rate": 0.0001751364577451534,
      "loss": 3.1465,
      "step": 17260
    },
    {
      "epoch": 3.250517598343685,
      "grad_norm": 2.510300636291504,
      "learning_rate": 0.00017494824016563146,
      "loss": 3.6213,
      "step": 17270
    },
    {
      "epoch": 3.2523997741389046,
      "grad_norm": 2.660430669784546,
      "learning_rate": 0.00017476002258610954,
      "loss": 3.7127,
      "step": 17280
    },
    {
      "epoch": 3.254281949934124,
      "grad_norm": 2.52522349357605,
      "learning_rate": 0.00017457180500658761,
      "loss": 3.3203,
      "step": 17290
    },
    {
      "epoch": 3.256164125729343,
      "grad_norm": 2.3675825595855713,
      "learning_rate": 0.0001743835874270657,
      "loss": 3.6297,
      "step": 17300
    },
    {
      "epoch": 3.2580463015245624,
      "grad_norm": 2.6578762531280518,
      "learning_rate": 0.00017419536984754377,
      "loss": 3.3509,
      "step": 17310
    },
    {
      "epoch": 3.259928477319782,
      "grad_norm": 2.3674263954162598,
      "learning_rate": 0.00017400715226802182,
      "loss": 3.1971,
      "step": 17320
    },
    {
      "epoch": 3.261810653115001,
      "grad_norm": 2.461933135986328,
      "learning_rate": 0.00017381893468849992,
      "loss": 3.2012,
      "step": 17330
    },
    {
      "epoch": 3.2636928289102203,
      "grad_norm": 3.2475147247314453,
      "learning_rate": 0.000173630717108978,
      "loss": 3.4897,
      "step": 17340
    },
    {
      "epoch": 3.2655750047054397,
      "grad_norm": 2.4707467555999756,
      "learning_rate": 0.00017344249952945607,
      "loss": 3.4132,
      "step": 17350
    },
    {
      "epoch": 3.2674571805006587,
      "grad_norm": 2.574896812438965,
      "learning_rate": 0.00017325428194993412,
      "loss": 3.4592,
      "step": 17360
    },
    {
      "epoch": 3.269339356295878,
      "grad_norm": 2.799384117126465,
      "learning_rate": 0.0001730660643704122,
      "loss": 3.4034,
      "step": 17370
    },
    {
      "epoch": 3.271221532091097,
      "grad_norm": 3.077925205230713,
      "learning_rate": 0.00017287784679089027,
      "loss": 3.4267,
      "step": 17380
    },
    {
      "epoch": 3.2731037078863165,
      "grad_norm": 2.2687697410583496,
      "learning_rate": 0.00017268962921136835,
      "loss": 3.4686,
      "step": 17390
    },
    {
      "epoch": 3.274985883681536,
      "grad_norm": 2.218032121658325,
      "learning_rate": 0.0001725014116318464,
      "loss": 3.1861,
      "step": 17400
    },
    {
      "epoch": 3.276868059476755,
      "grad_norm": 2.9106338024139404,
      "learning_rate": 0.00017231319405232448,
      "loss": 3.5033,
      "step": 17410
    },
    {
      "epoch": 3.2787502352719744,
      "grad_norm": 3.4965925216674805,
      "learning_rate": 0.00017212497647280255,
      "loss": 3.5055,
      "step": 17420
    },
    {
      "epoch": 3.280632411067194,
      "grad_norm": 2.3185698986053467,
      "learning_rate": 0.00017193675889328063,
      "loss": 3.4843,
      "step": 17430
    },
    {
      "epoch": 3.282514586862413,
      "grad_norm": 2.887524366378784,
      "learning_rate": 0.00017174854131375873,
      "loss": 3.5924,
      "step": 17440
    },
    {
      "epoch": 3.2843967626576323,
      "grad_norm": 3.108182668685913,
      "learning_rate": 0.00017156032373423678,
      "loss": 3.3346,
      "step": 17450
    },
    {
      "epoch": 3.2862789384528517,
      "grad_norm": 3.007628917694092,
      "learning_rate": 0.00017137210615471486,
      "loss": 3.2649,
      "step": 17460
    },
    {
      "epoch": 3.2881611142480707,
      "grad_norm": 2.5218400955200195,
      "learning_rate": 0.00017118388857519294,
      "loss": 3.2575,
      "step": 17470
    },
    {
      "epoch": 3.29004329004329,
      "grad_norm": 4.477243900299072,
      "learning_rate": 0.000170995670995671,
      "loss": 3.343,
      "step": 17480
    },
    {
      "epoch": 3.291925465838509,
      "grad_norm": 4.656927108764648,
      "learning_rate": 0.00017080745341614906,
      "loss": 3.32,
      "step": 17490
    },
    {
      "epoch": 3.2938076416337285,
      "grad_norm": 3.0149948596954346,
      "learning_rate": 0.00017061923583662714,
      "loss": 3.6426,
      "step": 17500
    },
    {
      "epoch": 3.295689817428948,
      "grad_norm": 2.364891529083252,
      "learning_rate": 0.00017043101825710521,
      "loss": 3.4555,
      "step": 17510
    },
    {
      "epoch": 3.2975719932241674,
      "grad_norm": 2.841169595718384,
      "learning_rate": 0.0001702428006775833,
      "loss": 2.9926,
      "step": 17520
    },
    {
      "epoch": 3.2994541690193864,
      "grad_norm": 2.4284143447875977,
      "learning_rate": 0.00017005458309806134,
      "loss": 3.4719,
      "step": 17530
    },
    {
      "epoch": 3.301336344814606,
      "grad_norm": 2.8778367042541504,
      "learning_rate": 0.00016986636551853942,
      "loss": 3.2167,
      "step": 17540
    },
    {
      "epoch": 3.303218520609825,
      "grad_norm": 3.5048258304595947,
      "learning_rate": 0.00016967814793901752,
      "loss": 3.2399,
      "step": 17550
    },
    {
      "epoch": 3.3051006964050442,
      "grad_norm": 4.433194160461426,
      "learning_rate": 0.0001694899303594956,
      "loss": 3.3544,
      "step": 17560
    },
    {
      "epoch": 3.3069828722002637,
      "grad_norm": 3.8717234134674072,
      "learning_rate": 0.00016930171277997367,
      "loss": 3.3017,
      "step": 17570
    },
    {
      "epoch": 3.3088650479954826,
      "grad_norm": 2.5483531951904297,
      "learning_rate": 0.00016911349520045172,
      "loss": 3.4301,
      "step": 17580
    },
    {
      "epoch": 3.310747223790702,
      "grad_norm": 2.6057353019714355,
      "learning_rate": 0.0001689252776209298,
      "loss": 3.9394,
      "step": 17590
    },
    {
      "epoch": 3.3126293995859215,
      "grad_norm": 3.1346988677978516,
      "learning_rate": 0.00016873706004140787,
      "loss": 3.4337,
      "step": 17600
    },
    {
      "epoch": 3.3145115753811405,
      "grad_norm": 2.919283151626587,
      "learning_rate": 0.00016854884246188595,
      "loss": 3.105,
      "step": 17610
    },
    {
      "epoch": 3.31639375117636,
      "grad_norm": 2.9236936569213867,
      "learning_rate": 0.000168360624882364,
      "loss": 3.6559,
      "step": 17620
    },
    {
      "epoch": 3.3182759269715794,
      "grad_norm": 2.968472719192505,
      "learning_rate": 0.00016817240730284208,
      "loss": 3.5618,
      "step": 17630
    },
    {
      "epoch": 3.3201581027667983,
      "grad_norm": 3.1104965209960938,
      "learning_rate": 0.00016798418972332015,
      "loss": 3.2912,
      "step": 17640
    },
    {
      "epoch": 3.322040278562018,
      "grad_norm": 3.617663860321045,
      "learning_rate": 0.00016779597214379823,
      "loss": 3.116,
      "step": 17650
    },
    {
      "epoch": 3.3239224543572368,
      "grad_norm": 2.541982650756836,
      "learning_rate": 0.0001676077545642763,
      "loss": 3.6969,
      "step": 17660
    },
    {
      "epoch": 3.325804630152456,
      "grad_norm": 2.6385083198547363,
      "learning_rate": 0.00016741953698475438,
      "loss": 3.2644,
      "step": 17670
    },
    {
      "epoch": 3.3276868059476756,
      "grad_norm": 2.365657329559326,
      "learning_rate": 0.00016723131940523246,
      "loss": 3.3977,
      "step": 17680
    },
    {
      "epoch": 3.3295689817428946,
      "grad_norm": 3.0871853828430176,
      "learning_rate": 0.00016704310182571054,
      "loss": 3.4599,
      "step": 17690
    },
    {
      "epoch": 3.331451157538114,
      "grad_norm": 2.563349962234497,
      "learning_rate": 0.0001668548842461886,
      "loss": 3.3422,
      "step": 17700
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 3.314103126525879,
      "learning_rate": 0.00016666666666666666,
      "loss": 3.5439,
      "step": 17710
    },
    {
      "epoch": 3.3352155091285525,
      "grad_norm": 2.4972023963928223,
      "learning_rate": 0.00016647844908714474,
      "loss": 3.2783,
      "step": 17720
    },
    {
      "epoch": 3.337097684923772,
      "grad_norm": 3.040225028991699,
      "learning_rate": 0.00016629023150762281,
      "loss": 3.3601,
      "step": 17730
    },
    {
      "epoch": 3.3389798607189913,
      "grad_norm": 2.6701977252960205,
      "learning_rate": 0.0001661020139281009,
      "loss": 3.6541,
      "step": 17740
    },
    {
      "epoch": 3.3408620365142103,
      "grad_norm": 2.771026372909546,
      "learning_rate": 0.00016591379634857894,
      "loss": 3.6395,
      "step": 17750
    },
    {
      "epoch": 3.3427442123094298,
      "grad_norm": 2.5545403957366943,
      "learning_rate": 0.00016572557876905702,
      "loss": 3.4282,
      "step": 17760
    },
    {
      "epoch": 3.3446263881046487,
      "grad_norm": 3.2837064266204834,
      "learning_rate": 0.00016553736118953512,
      "loss": 3.3282,
      "step": 17770
    },
    {
      "epoch": 3.346508563899868,
      "grad_norm": 2.680157423019409,
      "learning_rate": 0.0001653491436100132,
      "loss": 3.5378,
      "step": 17780
    },
    {
      "epoch": 3.3483907396950876,
      "grad_norm": 2.6319072246551514,
      "learning_rate": 0.00016516092603049125,
      "loss": 3.3281,
      "step": 17790
    },
    {
      "epoch": 3.350272915490307,
      "grad_norm": 2.3553097248077393,
      "learning_rate": 0.00016497270845096932,
      "loss": 3.3224,
      "step": 17800
    },
    {
      "epoch": 3.352155091285526,
      "grad_norm": 2.488459348678589,
      "learning_rate": 0.0001647844908714474,
      "loss": 3.2201,
      "step": 17810
    },
    {
      "epoch": 3.3540372670807455,
      "grad_norm": 2.488111972808838,
      "learning_rate": 0.00016459627329192547,
      "loss": 3.3382,
      "step": 17820
    },
    {
      "epoch": 3.3559194428759644,
      "grad_norm": 3.3139944076538086,
      "learning_rate": 0.00016440805571240355,
      "loss": 3.5223,
      "step": 17830
    },
    {
      "epoch": 3.357801618671184,
      "grad_norm": 2.8468685150146484,
      "learning_rate": 0.0001642198381328816,
      "loss": 3.5038,
      "step": 17840
    },
    {
      "epoch": 3.3596837944664033,
      "grad_norm": 2.1689863204956055,
      "learning_rate": 0.00016403162055335968,
      "loss": 3.2975,
      "step": 17850
    },
    {
      "epoch": 3.3615659702616223,
      "grad_norm": 2.7081663608551025,
      "learning_rate": 0.00016384340297383775,
      "loss": 3.3812,
      "step": 17860
    },
    {
      "epoch": 3.3634481460568417,
      "grad_norm": 3.3801076412200928,
      "learning_rate": 0.00016365518539431583,
      "loss": 3.347,
      "step": 17870
    },
    {
      "epoch": 3.365330321852061,
      "grad_norm": 2.969923973083496,
      "learning_rate": 0.0001634669678147939,
      "loss": 3.4156,
      "step": 17880
    },
    {
      "epoch": 3.36721249764728,
      "grad_norm": 2.6103897094726562,
      "learning_rate": 0.00016327875023527198,
      "loss": 3.1598,
      "step": 17890
    },
    {
      "epoch": 3.3690946734424996,
      "grad_norm": 2.7593328952789307,
      "learning_rate": 0.00016309053265575006,
      "loss": 3.1087,
      "step": 17900
    },
    {
      "epoch": 3.370976849237719,
      "grad_norm": 2.5040123462677,
      "learning_rate": 0.00016290231507622814,
      "loss": 3.3374,
      "step": 17910
    },
    {
      "epoch": 3.372859025032938,
      "grad_norm": 2.686152935028076,
      "learning_rate": 0.0001627140974967062,
      "loss": 3.6755,
      "step": 17920
    },
    {
      "epoch": 3.3747412008281574,
      "grad_norm": 2.570594549179077,
      "learning_rate": 0.00016252587991718426,
      "loss": 3.0671,
      "step": 17930
    },
    {
      "epoch": 3.3766233766233764,
      "grad_norm": 3.6530938148498535,
      "learning_rate": 0.00016233766233766234,
      "loss": 3.4958,
      "step": 17940
    },
    {
      "epoch": 3.378505552418596,
      "grad_norm": 2.7650959491729736,
      "learning_rate": 0.00016214944475814041,
      "loss": 3.2266,
      "step": 17950
    },
    {
      "epoch": 3.3803877282138153,
      "grad_norm": 2.8386785984039307,
      "learning_rate": 0.0001619612271786185,
      "loss": 3.513,
      "step": 17960
    },
    {
      "epoch": 3.3822699040090343,
      "grad_norm": 2.6067397594451904,
      "learning_rate": 0.00016177300959909654,
      "loss": 3.2123,
      "step": 17970
    },
    {
      "epoch": 3.3841520798042537,
      "grad_norm": 2.2532970905303955,
      "learning_rate": 0.00016158479201957462,
      "loss": 3.475,
      "step": 17980
    },
    {
      "epoch": 3.386034255599473,
      "grad_norm": 2.9290244579315186,
      "learning_rate": 0.00016139657444005272,
      "loss": 3.4137,
      "step": 17990
    },
    {
      "epoch": 3.387916431394692,
      "grad_norm": 2.3853273391723633,
      "learning_rate": 0.0001612083568605308,
      "loss": 3.3931,
      "step": 18000
    },
    {
      "epoch": 3.3897986071899116,
      "grad_norm": 2.483642339706421,
      "learning_rate": 0.00016102013928100885,
      "loss": 3.3691,
      "step": 18010
    },
    {
      "epoch": 3.391680782985131,
      "grad_norm": 3.018707752227783,
      "learning_rate": 0.00016083192170148692,
      "loss": 3.5776,
      "step": 18020
    },
    {
      "epoch": 3.39356295878035,
      "grad_norm": 3.4228579998016357,
      "learning_rate": 0.000160643704121965,
      "loss": 3.3006,
      "step": 18030
    },
    {
      "epoch": 3.3954451345755694,
      "grad_norm": 2.5835790634155273,
      "learning_rate": 0.00016045548654244307,
      "loss": 3.4533,
      "step": 18040
    },
    {
      "epoch": 3.3973273103707884,
      "grad_norm": 3.668458938598633,
      "learning_rate": 0.00016026726896292115,
      "loss": 3.4412,
      "step": 18050
    },
    {
      "epoch": 3.399209486166008,
      "grad_norm": 2.2307446002960205,
      "learning_rate": 0.0001600790513833992,
      "loss": 3.1922,
      "step": 18060
    },
    {
      "epoch": 3.4010916619612273,
      "grad_norm": 2.579155206680298,
      "learning_rate": 0.00015989083380387728,
      "loss": 3.6517,
      "step": 18070
    },
    {
      "epoch": 3.4029738377564467,
      "grad_norm": 2.7033772468566895,
      "learning_rate": 0.00015970261622435535,
      "loss": 3.4283,
      "step": 18080
    },
    {
      "epoch": 3.4048560135516657,
      "grad_norm": 2.3598473072052,
      "learning_rate": 0.00015951439864483343,
      "loss": 3.4329,
      "step": 18090
    },
    {
      "epoch": 3.406738189346885,
      "grad_norm": 2.421225070953369,
      "learning_rate": 0.0001593261810653115,
      "loss": 2.9126,
      "step": 18100
    },
    {
      "epoch": 3.408620365142104,
      "grad_norm": 2.460188388824463,
      "learning_rate": 0.00015913796348578958,
      "loss": 3.1849,
      "step": 18110
    },
    {
      "epoch": 3.4105025409373235,
      "grad_norm": 2.6091690063476562,
      "learning_rate": 0.00015894974590626766,
      "loss": 3.4228,
      "step": 18120
    },
    {
      "epoch": 3.412384716732543,
      "grad_norm": 2.700566053390503,
      "learning_rate": 0.00015876152832674574,
      "loss": 3.4774,
      "step": 18130
    },
    {
      "epoch": 3.414266892527762,
      "grad_norm": 2.6323583126068115,
      "learning_rate": 0.00015857331074722378,
      "loss": 3.3954,
      "step": 18140
    },
    {
      "epoch": 3.4161490683229814,
      "grad_norm": 2.687390089035034,
      "learning_rate": 0.00015838509316770186,
      "loss": 3.2594,
      "step": 18150
    },
    {
      "epoch": 3.418031244118201,
      "grad_norm": 2.705817222595215,
      "learning_rate": 0.00015819687558817994,
      "loss": 3.3878,
      "step": 18160
    },
    {
      "epoch": 3.41991341991342,
      "grad_norm": 2.4590365886688232,
      "learning_rate": 0.00015800865800865801,
      "loss": 3.0333,
      "step": 18170
    },
    {
      "epoch": 3.4217955957086392,
      "grad_norm": 2.336115598678589,
      "learning_rate": 0.0001578204404291361,
      "loss": 3.1843,
      "step": 18180
    },
    {
      "epoch": 3.4236777715038587,
      "grad_norm": 2.8003556728363037,
      "learning_rate": 0.00015763222284961414,
      "loss": 3.1254,
      "step": 18190
    },
    {
      "epoch": 3.4255599472990776,
      "grad_norm": 3.6783621311187744,
      "learning_rate": 0.00015744400527009222,
      "loss": 3.4193,
      "step": 18200
    },
    {
      "epoch": 3.427442123094297,
      "grad_norm": 2.280116558074951,
      "learning_rate": 0.00015725578769057032,
      "loss": 3.2221,
      "step": 18210
    },
    {
      "epoch": 3.429324298889516,
      "grad_norm": 2.6370630264282227,
      "learning_rate": 0.0001570675701110484,
      "loss": 3.4528,
      "step": 18220
    },
    {
      "epoch": 3.4312064746847355,
      "grad_norm": 2.820404291152954,
      "learning_rate": 0.00015687935253152645,
      "loss": 3.4164,
      "step": 18230
    },
    {
      "epoch": 3.433088650479955,
      "grad_norm": 2.970364809036255,
      "learning_rate": 0.00015669113495200452,
      "loss": 3.4036,
      "step": 18240
    },
    {
      "epoch": 3.434970826275174,
      "grad_norm": 2.6260478496551514,
      "learning_rate": 0.0001565029173724826,
      "loss": 3.572,
      "step": 18250
    },
    {
      "epoch": 3.4368530020703933,
      "grad_norm": 2.628375768661499,
      "learning_rate": 0.00015631469979296067,
      "loss": 3.1255,
      "step": 18260
    },
    {
      "epoch": 3.438735177865613,
      "grad_norm": 3.0799124240875244,
      "learning_rate": 0.00015612648221343872,
      "loss": 3.2365,
      "step": 18270
    },
    {
      "epoch": 3.4406173536608318,
      "grad_norm": 2.8040168285369873,
      "learning_rate": 0.0001559382646339168,
      "loss": 3.3586,
      "step": 18280
    },
    {
      "epoch": 3.442499529456051,
      "grad_norm": 2.400254249572754,
      "learning_rate": 0.00015575004705439488,
      "loss": 3.1926,
      "step": 18290
    },
    {
      "epoch": 3.4443817052512706,
      "grad_norm": 2.8902792930603027,
      "learning_rate": 0.00015556182947487295,
      "loss": 3.0558,
      "step": 18300
    },
    {
      "epoch": 3.4462638810464896,
      "grad_norm": 2.6344738006591797,
      "learning_rate": 0.00015537361189535103,
      "loss": 3.3546,
      "step": 18310
    },
    {
      "epoch": 3.448146056841709,
      "grad_norm": 2.634481906890869,
      "learning_rate": 0.0001551853943158291,
      "loss": 3.3618,
      "step": 18320
    },
    {
      "epoch": 3.4500282326369285,
      "grad_norm": 2.5868468284606934,
      "learning_rate": 0.00015499717673630718,
      "loss": 3.3979,
      "step": 18330
    },
    {
      "epoch": 3.4519104084321475,
      "grad_norm": 2.580573320388794,
      "learning_rate": 0.00015480895915678526,
      "loss": 3.344,
      "step": 18340
    },
    {
      "epoch": 3.453792584227367,
      "grad_norm": 2.240208148956299,
      "learning_rate": 0.00015462074157726334,
      "loss": 3.406,
      "step": 18350
    },
    {
      "epoch": 3.4556747600225863,
      "grad_norm": 3.5148346424102783,
      "learning_rate": 0.00015443252399774138,
      "loss": 3.0173,
      "step": 18360
    },
    {
      "epoch": 3.4575569358178053,
      "grad_norm": 2.5353798866271973,
      "learning_rate": 0.00015424430641821946,
      "loss": 3.4003,
      "step": 18370
    },
    {
      "epoch": 3.4594391116130248,
      "grad_norm": 2.6224095821380615,
      "learning_rate": 0.00015405608883869754,
      "loss": 3.3535,
      "step": 18380
    },
    {
      "epoch": 3.4613212874082437,
      "grad_norm": 2.489593029022217,
      "learning_rate": 0.00015386787125917561,
      "loss": 3.4504,
      "step": 18390
    },
    {
      "epoch": 3.463203463203463,
      "grad_norm": 2.6428322792053223,
      "learning_rate": 0.00015367965367965366,
      "loss": 3.4557,
      "step": 18400
    },
    {
      "epoch": 3.4650856389986826,
      "grad_norm": 2.464837074279785,
      "learning_rate": 0.00015349143610013174,
      "loss": 3.0501,
      "step": 18410
    },
    {
      "epoch": 3.4669678147939016,
      "grad_norm": 3.0857815742492676,
      "learning_rate": 0.00015330321852060982,
      "loss": 3.3766,
      "step": 18420
    },
    {
      "epoch": 3.468849990589121,
      "grad_norm": 2.774864912033081,
      "learning_rate": 0.00015311500094108792,
      "loss": 3.2938,
      "step": 18430
    },
    {
      "epoch": 3.4707321663843405,
      "grad_norm": 2.6012887954711914,
      "learning_rate": 0.000152926783361566,
      "loss": 3.3509,
      "step": 18440
    },
    {
      "epoch": 3.4726143421795594,
      "grad_norm": 3.4910905361175537,
      "learning_rate": 0.00015273856578204405,
      "loss": 3.1945,
      "step": 18450
    },
    {
      "epoch": 3.474496517974779,
      "grad_norm": 3.091975450515747,
      "learning_rate": 0.00015255034820252212,
      "loss": 3.2423,
      "step": 18460
    },
    {
      "epoch": 3.4763786937699983,
      "grad_norm": 2.299147129058838,
      "learning_rate": 0.0001523621306230002,
      "loss": 3.1601,
      "step": 18470
    },
    {
      "epoch": 3.4782608695652173,
      "grad_norm": 2.402646064758301,
      "learning_rate": 0.00015217391304347827,
      "loss": 3.1178,
      "step": 18480
    },
    {
      "epoch": 3.4801430453604367,
      "grad_norm": 3.045456647872925,
      "learning_rate": 0.00015198569546395632,
      "loss": 3.527,
      "step": 18490
    },
    {
      "epoch": 3.4820252211556557,
      "grad_norm": 2.992879867553711,
      "learning_rate": 0.0001517974778844344,
      "loss": 3.4033,
      "step": 18500
    },
    {
      "epoch": 3.483907396950875,
      "grad_norm": 2.254138469696045,
      "learning_rate": 0.00015160926030491248,
      "loss": 2.9717,
      "step": 18510
    },
    {
      "epoch": 3.4857895727460946,
      "grad_norm": 2.8303282260894775,
      "learning_rate": 0.00015142104272539055,
      "loss": 3.3559,
      "step": 18520
    },
    {
      "epoch": 3.4876717485413136,
      "grad_norm": 2.7462894916534424,
      "learning_rate": 0.00015123282514586863,
      "loss": 3.3438,
      "step": 18530
    },
    {
      "epoch": 3.489553924336533,
      "grad_norm": 3.280153274536133,
      "learning_rate": 0.00015104460756634668,
      "loss": 3.4967,
      "step": 18540
    },
    {
      "epoch": 3.4914361001317524,
      "grad_norm": 3.010605573654175,
      "learning_rate": 0.00015085638998682478,
      "loss": 3.3245,
      "step": 18550
    },
    {
      "epoch": 3.4933182759269714,
      "grad_norm": 2.8119089603424072,
      "learning_rate": 0.00015066817240730286,
      "loss": 3.0419,
      "step": 18560
    },
    {
      "epoch": 3.495200451722191,
      "grad_norm": 3.2307541370391846,
      "learning_rate": 0.00015047995482778094,
      "loss": 3.4478,
      "step": 18570
    },
    {
      "epoch": 3.4970826275174103,
      "grad_norm": 2.8820271492004395,
      "learning_rate": 0.00015029173724825898,
      "loss": 3.5247,
      "step": 18580
    },
    {
      "epoch": 3.4989648033126293,
      "grad_norm": 2.232989549636841,
      "learning_rate": 0.00015010351966873706,
      "loss": 3.0745,
      "step": 18590
    },
    {
      "epoch": 3.5008469791078487,
      "grad_norm": 3.2824294567108154,
      "learning_rate": 0.00014991530208921514,
      "loss": 3.4003,
      "step": 18600
    },
    {
      "epoch": 3.5027291549030677,
      "grad_norm": 2.7704029083251953,
      "learning_rate": 0.00014972708450969321,
      "loss": 3.2815,
      "step": 18610
    },
    {
      "epoch": 3.504611330698287,
      "grad_norm": 3.6349306106567383,
      "learning_rate": 0.00014953886693017126,
      "loss": 3.6239,
      "step": 18620
    },
    {
      "epoch": 3.5064935064935066,
      "grad_norm": 3.6622002124786377,
      "learning_rate": 0.00014935064935064934,
      "loss": 3.4647,
      "step": 18630
    },
    {
      "epoch": 3.508375682288726,
      "grad_norm": 2.65909481048584,
      "learning_rate": 0.00014916243177112742,
      "loss": 3.0913,
      "step": 18640
    },
    {
      "epoch": 3.510257858083945,
      "grad_norm": 2.877439022064209,
      "learning_rate": 0.0001489742141916055,
      "loss": 2.859,
      "step": 18650
    },
    {
      "epoch": 3.5121400338791644,
      "grad_norm": 2.6584038734436035,
      "learning_rate": 0.0001487859966120836,
      "loss": 3.2673,
      "step": 18660
    },
    {
      "epoch": 3.5140222096743834,
      "grad_norm": 3.749934196472168,
      "learning_rate": 0.00014859777903256165,
      "loss": 3.7862,
      "step": 18670
    },
    {
      "epoch": 3.515904385469603,
      "grad_norm": 2.909475088119507,
      "learning_rate": 0.00014840956145303972,
      "loss": 3.4305,
      "step": 18680
    },
    {
      "epoch": 3.5177865612648223,
      "grad_norm": 3.0480315685272217,
      "learning_rate": 0.0001482213438735178,
      "loss": 3.2649,
      "step": 18690
    },
    {
      "epoch": 3.5196687370600412,
      "grad_norm": 3.11883282661438,
      "learning_rate": 0.00014803312629399587,
      "loss": 3.3649,
      "step": 18700
    },
    {
      "epoch": 3.5215509128552607,
      "grad_norm": 2.5214035511016846,
      "learning_rate": 0.00014784490871447392,
      "loss": 3.6397,
      "step": 18710
    },
    {
      "epoch": 3.52343308865048,
      "grad_norm": 3.1389122009277344,
      "learning_rate": 0.000147656691134952,
      "loss": 3.5458,
      "step": 18720
    },
    {
      "epoch": 3.525315264445699,
      "grad_norm": 2.5999817848205566,
      "learning_rate": 0.00014746847355543008,
      "loss": 3.372,
      "step": 18730
    },
    {
      "epoch": 3.5271974402409185,
      "grad_norm": 2.3595402240753174,
      "learning_rate": 0.00014728025597590815,
      "loss": 3.1727,
      "step": 18740
    },
    {
      "epoch": 3.529079616036138,
      "grad_norm": 2.473238468170166,
      "learning_rate": 0.0001470920383963862,
      "loss": 3.4563,
      "step": 18750
    },
    {
      "epoch": 3.530961791831357,
      "grad_norm": 5.704708576202393,
      "learning_rate": 0.00014690382081686428,
      "loss": 3.2771,
      "step": 18760
    },
    {
      "epoch": 3.5328439676265764,
      "grad_norm": 2.5150506496429443,
      "learning_rate": 0.00014671560323734238,
      "loss": 3.0659,
      "step": 18770
    },
    {
      "epoch": 3.5347261434217954,
      "grad_norm": 2.901177406311035,
      "learning_rate": 0.00014652738565782046,
      "loss": 3.1914,
      "step": 18780
    },
    {
      "epoch": 3.536608319217015,
      "grad_norm": 2.633591413497925,
      "learning_rate": 0.00014633916807829854,
      "loss": 3.5465,
      "step": 18790
    },
    {
      "epoch": 3.5384904950122342,
      "grad_norm": 2.6768107414245605,
      "learning_rate": 0.00014615095049877658,
      "loss": 3.2693,
      "step": 18800
    },
    {
      "epoch": 3.5403726708074537,
      "grad_norm": 2.265665054321289,
      "learning_rate": 0.00014596273291925466,
      "loss": 3.3685,
      "step": 18810
    },
    {
      "epoch": 3.5422548466026726,
      "grad_norm": 2.9285104274749756,
      "learning_rate": 0.00014577451533973274,
      "loss": 3.2335,
      "step": 18820
    },
    {
      "epoch": 3.544137022397892,
      "grad_norm": 3.480781078338623,
      "learning_rate": 0.00014558629776021081,
      "loss": 3.2812,
      "step": 18830
    },
    {
      "epoch": 3.546019198193111,
      "grad_norm": 2.353689432144165,
      "learning_rate": 0.00014539808018068886,
      "loss": 3.3355,
      "step": 18840
    },
    {
      "epoch": 3.5479013739883305,
      "grad_norm": 3.253490447998047,
      "learning_rate": 0.00014520986260116694,
      "loss": 2.9411,
      "step": 18850
    },
    {
      "epoch": 3.54978354978355,
      "grad_norm": 2.585083484649658,
      "learning_rate": 0.00014502164502164502,
      "loss": 3.3769,
      "step": 18860
    },
    {
      "epoch": 3.551665725578769,
      "grad_norm": 2.4804880619049072,
      "learning_rate": 0.0001448334274421231,
      "loss": 3.3273,
      "step": 18870
    },
    {
      "epoch": 3.5535479013739883,
      "grad_norm": 3.0086333751678467,
      "learning_rate": 0.00014464520986260117,
      "loss": 3.364,
      "step": 18880
    },
    {
      "epoch": 3.5554300771692073,
      "grad_norm": 2.616567373275757,
      "learning_rate": 0.00014445699228307925,
      "loss": 3.5107,
      "step": 18890
    },
    {
      "epoch": 3.5573122529644268,
      "grad_norm": 2.6643967628479004,
      "learning_rate": 0.00014426877470355732,
      "loss": 3.2795,
      "step": 18900
    },
    {
      "epoch": 3.559194428759646,
      "grad_norm": 3.104766607284546,
      "learning_rate": 0.0001440805571240354,
      "loss": 3.1807,
      "step": 18910
    },
    {
      "epoch": 3.5610766045548656,
      "grad_norm": 2.417785406112671,
      "learning_rate": 0.00014389233954451347,
      "loss": 3.0671,
      "step": 18920
    },
    {
      "epoch": 3.5629587803500846,
      "grad_norm": 3.15303897857666,
      "learning_rate": 0.00014370412196499152,
      "loss": 3.1559,
      "step": 18930
    },
    {
      "epoch": 3.564840956145304,
      "grad_norm": 2.950138807296753,
      "learning_rate": 0.0001435159043854696,
      "loss": 3.2226,
      "step": 18940
    },
    {
      "epoch": 3.566723131940523,
      "grad_norm": 3.2707948684692383,
      "learning_rate": 0.00014332768680594768,
      "loss": 3.4891,
      "step": 18950
    },
    {
      "epoch": 3.5686053077357425,
      "grad_norm": 2.545849084854126,
      "learning_rate": 0.00014313946922642575,
      "loss": 3.3594,
      "step": 18960
    },
    {
      "epoch": 3.570487483530962,
      "grad_norm": 3.009657621383667,
      "learning_rate": 0.0001429512516469038,
      "loss": 3.6016,
      "step": 18970
    },
    {
      "epoch": 3.5723696593261813,
      "grad_norm": 2.9132399559020996,
      "learning_rate": 0.00014276303406738188,
      "loss": 3.4437,
      "step": 18980
    },
    {
      "epoch": 3.5742518351214003,
      "grad_norm": 2.985321044921875,
      "learning_rate": 0.00014257481648785998,
      "loss": 2.7064,
      "step": 18990
    },
    {
      "epoch": 3.5761340109166198,
      "grad_norm": 2.5817177295684814,
      "learning_rate": 0.00014238659890833806,
      "loss": 3.5903,
      "step": 19000
    },
    {
      "epoch": 3.5780161867118387,
      "grad_norm": 2.643543004989624,
      "learning_rate": 0.0001421983813288161,
      "loss": 3.3309,
      "step": 19010
    },
    {
      "epoch": 3.579898362507058,
      "grad_norm": 2.5157105922698975,
      "learning_rate": 0.00014201016374929418,
      "loss": 3.7219,
      "step": 19020
    },
    {
      "epoch": 3.5817805383022776,
      "grad_norm": 2.6616153717041016,
      "learning_rate": 0.00014182194616977226,
      "loss": 3.2806,
      "step": 19030
    },
    {
      "epoch": 3.5836627140974966,
      "grad_norm": 2.7307190895080566,
      "learning_rate": 0.00014163372859025034,
      "loss": 3.3157,
      "step": 19040
    },
    {
      "epoch": 3.585544889892716,
      "grad_norm": 2.92560076713562,
      "learning_rate": 0.00014144551101072841,
      "loss": 3.0866,
      "step": 19050
    },
    {
      "epoch": 3.587427065687935,
      "grad_norm": 3.155165433883667,
      "learning_rate": 0.00014125729343120646,
      "loss": 3.4034,
      "step": 19060
    },
    {
      "epoch": 3.5893092414831544,
      "grad_norm": 2.795790195465088,
      "learning_rate": 0.00014106907585168454,
      "loss": 3.2257,
      "step": 19070
    },
    {
      "epoch": 3.591191417278374,
      "grad_norm": 2.648715019226074,
      "learning_rate": 0.00014088085827216262,
      "loss": 3.3564,
      "step": 19080
    },
    {
      "epoch": 3.5930735930735933,
      "grad_norm": 2.7506496906280518,
      "learning_rate": 0.0001406926406926407,
      "loss": 3.197,
      "step": 19090
    },
    {
      "epoch": 3.5949557688688123,
      "grad_norm": 2.9696731567382812,
      "learning_rate": 0.00014050442311311877,
      "loss": 3.3379,
      "step": 19100
    },
    {
      "epoch": 3.5968379446640317,
      "grad_norm": 2.6705081462860107,
      "learning_rate": 0.00014031620553359685,
      "loss": 3.588,
      "step": 19110
    },
    {
      "epoch": 3.5987201204592507,
      "grad_norm": 3.0333800315856934,
      "learning_rate": 0.00014012798795407492,
      "loss": 3.3548,
      "step": 19120
    },
    {
      "epoch": 3.60060229625447,
      "grad_norm": 2.8488924503326416,
      "learning_rate": 0.000139939770374553,
      "loss": 2.8974,
      "step": 19130
    },
    {
      "epoch": 3.6024844720496896,
      "grad_norm": 2.7044355869293213,
      "learning_rate": 0.00013975155279503105,
      "loss": 3.3713,
      "step": 19140
    },
    {
      "epoch": 3.6043666478449086,
      "grad_norm": 2.565526247024536,
      "learning_rate": 0.00013956333521550912,
      "loss": 3.2275,
      "step": 19150
    },
    {
      "epoch": 3.606248823640128,
      "grad_norm": 3.3287127017974854,
      "learning_rate": 0.0001393751176359872,
      "loss": 3.2934,
      "step": 19160
    },
    {
      "epoch": 3.608130999435347,
      "grad_norm": 3.1183736324310303,
      "learning_rate": 0.00013918690005646528,
      "loss": 3.375,
      "step": 19170
    },
    {
      "epoch": 3.6100131752305664,
      "grad_norm": 2.3812272548675537,
      "learning_rate": 0.00013899868247694335,
      "loss": 2.9646,
      "step": 19180
    },
    {
      "epoch": 3.611895351025786,
      "grad_norm": 2.601849317550659,
      "learning_rate": 0.0001388104648974214,
      "loss": 3.4268,
      "step": 19190
    },
    {
      "epoch": 3.6137775268210053,
      "grad_norm": 2.694033622741699,
      "learning_rate": 0.00013862224731789948,
      "loss": 3.7199,
      "step": 19200
    },
    {
      "epoch": 3.6156597026162243,
      "grad_norm": 2.751224994659424,
      "learning_rate": 0.00013843402973837758,
      "loss": 3.5829,
      "step": 19210
    },
    {
      "epoch": 3.6175418784114437,
      "grad_norm": 2.70204496383667,
      "learning_rate": 0.00013824581215885566,
      "loss": 3.3441,
      "step": 19220
    },
    {
      "epoch": 3.6194240542066627,
      "grad_norm": 3.35942006111145,
      "learning_rate": 0.0001380575945793337,
      "loss": 3.0396,
      "step": 19230
    },
    {
      "epoch": 3.621306230001882,
      "grad_norm": 2.7449188232421875,
      "learning_rate": 0.00013786937699981178,
      "loss": 3.4542,
      "step": 19240
    },
    {
      "epoch": 3.6231884057971016,
      "grad_norm": 2.499721050262451,
      "learning_rate": 0.00013768115942028986,
      "loss": 3.2617,
      "step": 19250
    },
    {
      "epoch": 3.625070581592321,
      "grad_norm": 3.022339344024658,
      "learning_rate": 0.00013749294184076794,
      "loss": 3.5071,
      "step": 19260
    },
    {
      "epoch": 3.62695275738754,
      "grad_norm": 3.4480507373809814,
      "learning_rate": 0.00013730472426124601,
      "loss": 3.3696,
      "step": 19270
    },
    {
      "epoch": 3.6288349331827594,
      "grad_norm": 2.719674587249756,
      "learning_rate": 0.00013711650668172406,
      "loss": 3.2398,
      "step": 19280
    },
    {
      "epoch": 3.6307171089779784,
      "grad_norm": 2.420034646987915,
      "learning_rate": 0.00013692828910220214,
      "loss": 3.399,
      "step": 19290
    },
    {
      "epoch": 3.632599284773198,
      "grad_norm": 2.120476007461548,
      "learning_rate": 0.00013674007152268022,
      "loss": 3.0942,
      "step": 19300
    },
    {
      "epoch": 3.6344814605684173,
      "grad_norm": 2.2568111419677734,
      "learning_rate": 0.0001365518539431583,
      "loss": 3.5591,
      "step": 19310
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 2.630584478378296,
      "learning_rate": 0.00013636363636363637,
      "loss": 3.3086,
      "step": 19320
    },
    {
      "epoch": 3.6382458121588557,
      "grad_norm": 2.691880941390991,
      "learning_rate": 0.00013617541878411445,
      "loss": 3.3616,
      "step": 19330
    },
    {
      "epoch": 3.6401279879540747,
      "grad_norm": 3.3408854007720947,
      "learning_rate": 0.00013598720120459252,
      "loss": 3.2374,
      "step": 19340
    },
    {
      "epoch": 3.642010163749294,
      "grad_norm": 2.621610641479492,
      "learning_rate": 0.0001357989836250706,
      "loss": 3.3335,
      "step": 19350
    },
    {
      "epoch": 3.6438923395445135,
      "grad_norm": 2.847215414047241,
      "learning_rate": 0.00013561076604554865,
      "loss": 3.3562,
      "step": 19360
    },
    {
      "epoch": 3.645774515339733,
      "grad_norm": 2.9248063564300537,
      "learning_rate": 0.00013542254846602672,
      "loss": 3.4876,
      "step": 19370
    },
    {
      "epoch": 3.647656691134952,
      "grad_norm": 2.526078701019287,
      "learning_rate": 0.0001352343308865048,
      "loss": 3.4567,
      "step": 19380
    },
    {
      "epoch": 3.6495388669301714,
      "grad_norm": 2.450305938720703,
      "learning_rate": 0.00013504611330698288,
      "loss": 3.4384,
      "step": 19390
    },
    {
      "epoch": 3.6514210427253904,
      "grad_norm": 2.7489380836486816,
      "learning_rate": 0.00013485789572746095,
      "loss": 3.598,
      "step": 19400
    },
    {
      "epoch": 3.65330321852061,
      "grad_norm": 2.759094476699829,
      "learning_rate": 0.000134669678147939,
      "loss": 3.567,
      "step": 19410
    },
    {
      "epoch": 3.6551853943158292,
      "grad_norm": 3.249265670776367,
      "learning_rate": 0.00013448146056841708,
      "loss": 3.227,
      "step": 19420
    },
    {
      "epoch": 3.657067570111048,
      "grad_norm": 3.172544240951538,
      "learning_rate": 0.00013429324298889518,
      "loss": 3.1833,
      "step": 19430
    },
    {
      "epoch": 3.6589497459062676,
      "grad_norm": 2.491276502609253,
      "learning_rate": 0.00013410502540937326,
      "loss": 3.4064,
      "step": 19440
    },
    {
      "epoch": 3.660831921701487,
      "grad_norm": 3.467625379562378,
      "learning_rate": 0.0001339168078298513,
      "loss": 3.5775,
      "step": 19450
    },
    {
      "epoch": 3.662714097496706,
      "grad_norm": 2.929133653640747,
      "learning_rate": 0.00013372859025032938,
      "loss": 3.3226,
      "step": 19460
    },
    {
      "epoch": 3.6645962732919255,
      "grad_norm": 2.4177849292755127,
      "learning_rate": 0.00013354037267080746,
      "loss": 3.4322,
      "step": 19470
    },
    {
      "epoch": 3.666478449087145,
      "grad_norm": 2.494344711303711,
      "learning_rate": 0.00013335215509128554,
      "loss": 3.2928,
      "step": 19480
    },
    {
      "epoch": 3.668360624882364,
      "grad_norm": 3.252093553543091,
      "learning_rate": 0.0001331639375117636,
      "loss": 3.2858,
      "step": 19490
    },
    {
      "epoch": 3.6702428006775834,
      "grad_norm": 2.993792772293091,
      "learning_rate": 0.00013297571993224166,
      "loss": 3.1031,
      "step": 19500
    },
    {
      "epoch": 3.6721249764728023,
      "grad_norm": 3.514948606491089,
      "learning_rate": 0.00013278750235271974,
      "loss": 3.2385,
      "step": 19510
    },
    {
      "epoch": 3.6740071522680218,
      "grad_norm": 3.2754671573638916,
      "learning_rate": 0.00013259928477319782,
      "loss": 3.3253,
      "step": 19520
    },
    {
      "epoch": 3.675889328063241,
      "grad_norm": 3.0171761512756348,
      "learning_rate": 0.0001324110671936759,
      "loss": 3.1884,
      "step": 19530
    },
    {
      "epoch": 3.6777715038584606,
      "grad_norm": 3.297513246536255,
      "learning_rate": 0.00013222284961415397,
      "loss": 3.2366,
      "step": 19540
    },
    {
      "epoch": 3.6796536796536796,
      "grad_norm": 2.680819272994995,
      "learning_rate": 0.00013203463203463205,
      "loss": 3.5181,
      "step": 19550
    },
    {
      "epoch": 3.681535855448899,
      "grad_norm": 2.5602333545684814,
      "learning_rate": 0.00013184641445511012,
      "loss": 3.2382,
      "step": 19560
    },
    {
      "epoch": 3.683418031244118,
      "grad_norm": 2.667515993118286,
      "learning_rate": 0.0001316581968755882,
      "loss": 3.2604,
      "step": 19570
    },
    {
      "epoch": 3.6853002070393375,
      "grad_norm": 3.419037103652954,
      "learning_rate": 0.00013146997929606625,
      "loss": 3.1958,
      "step": 19580
    },
    {
      "epoch": 3.687182382834557,
      "grad_norm": 2.5114715099334717,
      "learning_rate": 0.00013128176171654432,
      "loss": 3.2694,
      "step": 19590
    },
    {
      "epoch": 3.689064558629776,
      "grad_norm": 2.8820993900299072,
      "learning_rate": 0.0001310935441370224,
      "loss": 3.2628,
      "step": 19600
    },
    {
      "epoch": 3.6909467344249953,
      "grad_norm": 2.631727695465088,
      "learning_rate": 0.00013090532655750048,
      "loss": 3.3984,
      "step": 19610
    },
    {
      "epoch": 3.6928289102202143,
      "grad_norm": 3.3341405391693115,
      "learning_rate": 0.00013071710897797853,
      "loss": 3.3687,
      "step": 19620
    },
    {
      "epoch": 3.6947110860154337,
      "grad_norm": 3.213141918182373,
      "learning_rate": 0.0001305288913984566,
      "loss": 3.6688,
      "step": 19630
    },
    {
      "epoch": 3.696593261810653,
      "grad_norm": 2.9482338428497314,
      "learning_rate": 0.00013034067381893468,
      "loss": 3.2922,
      "step": 19640
    },
    {
      "epoch": 3.6984754376058726,
      "grad_norm": 2.2478370666503906,
      "learning_rate": 0.00013015245623941278,
      "loss": 3.1485,
      "step": 19650
    },
    {
      "epoch": 3.7003576134010916,
      "grad_norm": 2.829216480255127,
      "learning_rate": 0.00012996423865989086,
      "loss": 3.1318,
      "step": 19660
    },
    {
      "epoch": 3.702239789196311,
      "grad_norm": 2.8538355827331543,
      "learning_rate": 0.0001297760210803689,
      "loss": 3.3705,
      "step": 19670
    },
    {
      "epoch": 3.70412196499153,
      "grad_norm": 2.4689252376556396,
      "learning_rate": 0.00012958780350084698,
      "loss": 3.2183,
      "step": 19680
    },
    {
      "epoch": 3.7060041407867494,
      "grad_norm": 2.2744150161743164,
      "learning_rate": 0.00012939958592132506,
      "loss": 3.3251,
      "step": 19690
    },
    {
      "epoch": 3.707886316581969,
      "grad_norm": 4.31705904006958,
      "learning_rate": 0.00012921136834180314,
      "loss": 3.4272,
      "step": 19700
    },
    {
      "epoch": 3.709768492377188,
      "grad_norm": 2.2697396278381348,
      "learning_rate": 0.0001290231507622812,
      "loss": 3.3771,
      "step": 19710
    },
    {
      "epoch": 3.7116506681724073,
      "grad_norm": 2.675645351409912,
      "learning_rate": 0.00012883493318275926,
      "loss": 3.5572,
      "step": 19720
    },
    {
      "epoch": 3.7135328439676267,
      "grad_norm": 3.0459091663360596,
      "learning_rate": 0.00012864671560323734,
      "loss": 3.2407,
      "step": 19730
    },
    {
      "epoch": 3.7154150197628457,
      "grad_norm": 2.8712270259857178,
      "learning_rate": 0.00012845849802371542,
      "loss": 3.3708,
      "step": 19740
    },
    {
      "epoch": 3.717297195558065,
      "grad_norm": 2.0616369247436523,
      "learning_rate": 0.00012827028044419347,
      "loss": 3.0517,
      "step": 19750
    },
    {
      "epoch": 3.7191793713532846,
      "grad_norm": 2.9897196292877197,
      "learning_rate": 0.00012808206286467157,
      "loss": 3.506,
      "step": 19760
    },
    {
      "epoch": 3.7210615471485036,
      "grad_norm": 2.4094443321228027,
      "learning_rate": 0.00012789384528514965,
      "loss": 3.2432,
      "step": 19770
    },
    {
      "epoch": 3.722943722943723,
      "grad_norm": 2.9298617839813232,
      "learning_rate": 0.00012770562770562772,
      "loss": 3.4935,
      "step": 19780
    },
    {
      "epoch": 3.724825898738942,
      "grad_norm": 2.423041820526123,
      "learning_rate": 0.0001275174101261058,
      "loss": 3.2528,
      "step": 19790
    },
    {
      "epoch": 3.7267080745341614,
      "grad_norm": 2.4703598022460938,
      "learning_rate": 0.00012732919254658385,
      "loss": 3.7537,
      "step": 19800
    },
    {
      "epoch": 3.728590250329381,
      "grad_norm": 2.2207388877868652,
      "learning_rate": 0.00012714097496706192,
      "loss": 3.1028,
      "step": 19810
    },
    {
      "epoch": 3.7304724261246003,
      "grad_norm": 2.2429516315460205,
      "learning_rate": 0.00012695275738754,
      "loss": 3.133,
      "step": 19820
    },
    {
      "epoch": 3.7323546019198193,
      "grad_norm": 2.499889850616455,
      "learning_rate": 0.00012676453980801808,
      "loss": 3.444,
      "step": 19830
    },
    {
      "epoch": 3.7342367777150387,
      "grad_norm": 3.0017619132995605,
      "learning_rate": 0.00012657632222849613,
      "loss": 3.4312,
      "step": 19840
    },
    {
      "epoch": 3.7361189535102577,
      "grad_norm": 2.957941770553589,
      "learning_rate": 0.0001263881046489742,
      "loss": 3.1642,
      "step": 19850
    },
    {
      "epoch": 3.738001129305477,
      "grad_norm": 3.130078077316284,
      "learning_rate": 0.00012619988706945228,
      "loss": 3.1626,
      "step": 19860
    },
    {
      "epoch": 3.7398833051006966,
      "grad_norm": 2.6402502059936523,
      "learning_rate": 0.00012601166948993038,
      "loss": 3.3958,
      "step": 19870
    },
    {
      "epoch": 3.7417654808959155,
      "grad_norm": 2.338499069213867,
      "learning_rate": 0.00012582345191040846,
      "loss": 3.4364,
      "step": 19880
    },
    {
      "epoch": 3.743647656691135,
      "grad_norm": 2.4969937801361084,
      "learning_rate": 0.0001256352343308865,
      "loss": 3.5577,
      "step": 19890
    },
    {
      "epoch": 3.745529832486354,
      "grad_norm": 2.380075454711914,
      "learning_rate": 0.00012544701675136458,
      "loss": 3.244,
      "step": 19900
    },
    {
      "epoch": 3.7474120082815734,
      "grad_norm": 2.291037082672119,
      "learning_rate": 0.00012525879917184266,
      "loss": 3.0756,
      "step": 19910
    },
    {
      "epoch": 3.749294184076793,
      "grad_norm": 2.3379275798797607,
      "learning_rate": 0.00012507058159232074,
      "loss": 3.1479,
      "step": 19920
    },
    {
      "epoch": 3.7511763598720123,
      "grad_norm": 2.849090337753296,
      "learning_rate": 0.00012488236401279881,
      "loss": 3.3691,
      "step": 19930
    },
    {
      "epoch": 3.7530585356672312,
      "grad_norm": 2.6632776260375977,
      "learning_rate": 0.00012469414643327686,
      "loss": 3.309,
      "step": 19940
    },
    {
      "epoch": 3.7549407114624507,
      "grad_norm": 1.9748684167861938,
      "learning_rate": 0.00012450592885375494,
      "loss": 3.2076,
      "step": 19950
    },
    {
      "epoch": 3.7568228872576697,
      "grad_norm": 2.6560304164886475,
      "learning_rate": 0.00012431771127423302,
      "loss": 3.3641,
      "step": 19960
    },
    {
      "epoch": 3.758705063052889,
      "grad_norm": 3.091458797454834,
      "learning_rate": 0.0001241294936947111,
      "loss": 3.7965,
      "step": 19970
    },
    {
      "epoch": 3.7605872388481085,
      "grad_norm": 2.292750835418701,
      "learning_rate": 0.00012394127611518917,
      "loss": 3.0958,
      "step": 19980
    },
    {
      "epoch": 3.762469414643328,
      "grad_norm": 2.81014347076416,
      "learning_rate": 0.00012375305853566725,
      "loss": 3.4827,
      "step": 19990
    },
    {
      "epoch": 3.764351590438547,
      "grad_norm": 2.6002442836761475,
      "learning_rate": 0.0001235648409561453,
      "loss": 2.8435,
      "step": 20000
    },
    {
      "epoch": 3.7662337662337664,
      "grad_norm": 2.743107318878174,
      "learning_rate": 0.00012337662337662337,
      "loss": 3.2989,
      "step": 20010
    },
    {
      "epoch": 3.7681159420289854,
      "grad_norm": 3.365095615386963,
      "learning_rate": 0.00012318840579710145,
      "loss": 3.516,
      "step": 20020
    },
    {
      "epoch": 3.769998117824205,
      "grad_norm": 2.544158458709717,
      "learning_rate": 0.00012300018821757952,
      "loss": 3.2268,
      "step": 20030
    },
    {
      "epoch": 3.7718802936194242,
      "grad_norm": 3.235170364379883,
      "learning_rate": 0.0001228119706380576,
      "loss": 3.2683,
      "step": 20040
    },
    {
      "epoch": 3.773762469414643,
      "grad_norm": 3.5878753662109375,
      "learning_rate": 0.00012262375305853568,
      "loss": 3.4592,
      "step": 20050
    },
    {
      "epoch": 3.7756446452098626,
      "grad_norm": 2.7629101276397705,
      "learning_rate": 0.00012243553547901375,
      "loss": 3.1159,
      "step": 20060
    },
    {
      "epoch": 3.7775268210050816,
      "grad_norm": 2.5514304637908936,
      "learning_rate": 0.0001222473178994918,
      "loss": 3.1758,
      "step": 20070
    },
    {
      "epoch": 3.779408996800301,
      "grad_norm": 2.6327264308929443,
      "learning_rate": 0.00012205910031996989,
      "loss": 3.5325,
      "step": 20080
    },
    {
      "epoch": 3.7812911725955205,
      "grad_norm": 2.35088849067688,
      "learning_rate": 0.00012187088274044796,
      "loss": 3.2252,
      "step": 20090
    },
    {
      "epoch": 3.78317334839074,
      "grad_norm": 3.5405220985412598,
      "learning_rate": 0.00012168266516092603,
      "loss": 3.3319,
      "step": 20100
    },
    {
      "epoch": 3.785055524185959,
      "grad_norm": 3.110680103302002,
      "learning_rate": 0.0001214944475814041,
      "loss": 3.3915,
      "step": 20110
    },
    {
      "epoch": 3.7869376999811784,
      "grad_norm": 2.190539598464966,
      "learning_rate": 0.00012130623000188218,
      "loss": 3.2086,
      "step": 20120
    },
    {
      "epoch": 3.7888198757763973,
      "grad_norm": 2.484727621078491,
      "learning_rate": 0.00012111801242236026,
      "loss": 3.5845,
      "step": 20130
    },
    {
      "epoch": 3.7907020515716168,
      "grad_norm": 2.2589473724365234,
      "learning_rate": 0.00012092979484283832,
      "loss": 3.2769,
      "step": 20140
    },
    {
      "epoch": 3.792584227366836,
      "grad_norm": 2.6829514503479004,
      "learning_rate": 0.0001207415772633164,
      "loss": 3.2124,
      "step": 20150
    },
    {
      "epoch": 3.794466403162055,
      "grad_norm": 2.3834097385406494,
      "learning_rate": 0.00012055335968379446,
      "loss": 3.3095,
      "step": 20160
    },
    {
      "epoch": 3.7963485789572746,
      "grad_norm": 2.6875038146972656,
      "learning_rate": 0.00012036514210427254,
      "loss": 3.329,
      "step": 20170
    },
    {
      "epoch": 3.7982307547524936,
      "grad_norm": 3.3290762901306152,
      "learning_rate": 0.00012017692452475062,
      "loss": 2.977,
      "step": 20180
    },
    {
      "epoch": 3.800112930547713,
      "grad_norm": 2.4440839290618896,
      "learning_rate": 0.00011998870694522869,
      "loss": 3.6461,
      "step": 20190
    },
    {
      "epoch": 3.8019951063429325,
      "grad_norm": 2.430534601211548,
      "learning_rate": 0.00011980048936570676,
      "loss": 3.1679,
      "step": 20200
    },
    {
      "epoch": 3.803877282138152,
      "grad_norm": 2.636639356613159,
      "learning_rate": 0.00011961227178618483,
      "loss": 3.361,
      "step": 20210
    },
    {
      "epoch": 3.805759457933371,
      "grad_norm": 2.4769604206085205,
      "learning_rate": 0.0001194240542066629,
      "loss": 3.3063,
      "step": 20220
    },
    {
      "epoch": 3.8076416337285903,
      "grad_norm": 2.9437851905822754,
      "learning_rate": 0.00011923583662714098,
      "loss": 3.138,
      "step": 20230
    },
    {
      "epoch": 3.8095238095238093,
      "grad_norm": 3.349299192428589,
      "learning_rate": 0.00011904761904761905,
      "loss": 3.107,
      "step": 20240
    },
    {
      "epoch": 3.8114059853190287,
      "grad_norm": 2.5073299407958984,
      "learning_rate": 0.00011885940146809712,
      "loss": 3.5264,
      "step": 20250
    },
    {
      "epoch": 3.813288161114248,
      "grad_norm": 3.159978151321411,
      "learning_rate": 0.0001186711838885752,
      "loss": 3.3368,
      "step": 20260
    },
    {
      "epoch": 3.8151703369094676,
      "grad_norm": 2.660733461380005,
      "learning_rate": 0.00011848296630905326,
      "loss": 3.0852,
      "step": 20270
    },
    {
      "epoch": 3.8170525127046866,
      "grad_norm": 2.7388837337493896,
      "learning_rate": 0.00011829474872953134,
      "loss": 3.2356,
      "step": 20280
    },
    {
      "epoch": 3.818934688499906,
      "grad_norm": 3.001278877258301,
      "learning_rate": 0.00011810653115000942,
      "loss": 3.3562,
      "step": 20290
    },
    {
      "epoch": 3.820816864295125,
      "grad_norm": 3.217205047607422,
      "learning_rate": 0.00011791831357048749,
      "loss": 3.3602,
      "step": 20300
    },
    {
      "epoch": 3.8226990400903444,
      "grad_norm": 2.8541181087493896,
      "learning_rate": 0.00011773009599096556,
      "loss": 3.3973,
      "step": 20310
    },
    {
      "epoch": 3.824581215885564,
      "grad_norm": 2.6750447750091553,
      "learning_rate": 0.00011754187841144363,
      "loss": 2.9959,
      "step": 20320
    },
    {
      "epoch": 3.826463391680783,
      "grad_norm": 8.156944274902344,
      "learning_rate": 0.0001173536608319217,
      "loss": 3.3854,
      "step": 20330
    },
    {
      "epoch": 3.8283455674760023,
      "grad_norm": 2.9745311737060547,
      "learning_rate": 0.00011716544325239978,
      "loss": 3.0903,
      "step": 20340
    },
    {
      "epoch": 3.8302277432712213,
      "grad_norm": 2.6606485843658447,
      "learning_rate": 0.00011697722567287785,
      "loss": 3.3105,
      "step": 20350
    },
    {
      "epoch": 3.8321099190664407,
      "grad_norm": 2.661266803741455,
      "learning_rate": 0.00011678900809335592,
      "loss": 3.2123,
      "step": 20360
    },
    {
      "epoch": 3.83399209486166,
      "grad_norm": 2.7658510208129883,
      "learning_rate": 0.000116600790513834,
      "loss": 3.4492,
      "step": 20370
    },
    {
      "epoch": 3.8358742706568796,
      "grad_norm": 4.285975456237793,
      "learning_rate": 0.00011641257293431206,
      "loss": 3.338,
      "step": 20380
    },
    {
      "epoch": 3.8377564464520986,
      "grad_norm": 3.6666557788848877,
      "learning_rate": 0.00011622435535479014,
      "loss": 3.388,
      "step": 20390
    },
    {
      "epoch": 3.839638622247318,
      "grad_norm": 2.724447250366211,
      "learning_rate": 0.00011603613777526822,
      "loss": 3.0808,
      "step": 20400
    },
    {
      "epoch": 3.841520798042537,
      "grad_norm": 2.9280121326446533,
      "learning_rate": 0.00011584792019574629,
      "loss": 3.1565,
      "step": 20410
    },
    {
      "epoch": 3.8434029738377564,
      "grad_norm": 3.070014238357544,
      "learning_rate": 0.00011565970261622436,
      "loss": 3.2554,
      "step": 20420
    },
    {
      "epoch": 3.845285149632976,
      "grad_norm": 2.623056650161743,
      "learning_rate": 0.00011547148503670243,
      "loss": 3.8541,
      "step": 20430
    },
    {
      "epoch": 3.847167325428195,
      "grad_norm": 2.4293594360351562,
      "learning_rate": 0.0001152832674571805,
      "loss": 3.303,
      "step": 20440
    },
    {
      "epoch": 3.8490495012234143,
      "grad_norm": 2.4312355518341064,
      "learning_rate": 0.00011509504987765858,
      "loss": 3.2644,
      "step": 20450
    },
    {
      "epoch": 3.8509316770186337,
      "grad_norm": 3.47027325630188,
      "learning_rate": 0.00011490683229813665,
      "loss": 3.3996,
      "step": 20460
    },
    {
      "epoch": 3.8528138528138527,
      "grad_norm": 2.698883533477783,
      "learning_rate": 0.00011471861471861472,
      "loss": 3.2559,
      "step": 20470
    },
    {
      "epoch": 3.854696028609072,
      "grad_norm": 2.323240280151367,
      "learning_rate": 0.00011453039713909279,
      "loss": 3.4054,
      "step": 20480
    },
    {
      "epoch": 3.8565782044042916,
      "grad_norm": 2.6935620307922363,
      "learning_rate": 0.00011434217955957086,
      "loss": 3.2728,
      "step": 20490
    },
    {
      "epoch": 3.8584603801995105,
      "grad_norm": 3.0033633708953857,
      "learning_rate": 0.00011415396198004894,
      "loss": 3.1861,
      "step": 20500
    },
    {
      "epoch": 3.86034255599473,
      "grad_norm": 2.8364367485046387,
      "learning_rate": 0.00011396574440052702,
      "loss": 3.3365,
      "step": 20510
    },
    {
      "epoch": 3.862224731789949,
      "grad_norm": 3.5239908695220947,
      "learning_rate": 0.00011377752682100509,
      "loss": 3.4504,
      "step": 20520
    },
    {
      "epoch": 3.8641069075851684,
      "grad_norm": 2.6138646602630615,
      "learning_rate": 0.00011358930924148316,
      "loss": 3.3374,
      "step": 20530
    },
    {
      "epoch": 3.865989083380388,
      "grad_norm": 2.864724636077881,
      "learning_rate": 0.00011340109166196123,
      "loss": 3.4036,
      "step": 20540
    },
    {
      "epoch": 3.8678712591756073,
      "grad_norm": 3.0685441493988037,
      "learning_rate": 0.0001132128740824393,
      "loss": 3.3029,
      "step": 20550
    },
    {
      "epoch": 3.8697534349708262,
      "grad_norm": 2.7818939685821533,
      "learning_rate": 0.00011302465650291738,
      "loss": 3.388,
      "step": 20560
    },
    {
      "epoch": 3.8716356107660457,
      "grad_norm": 3.0230255126953125,
      "learning_rate": 0.00011283643892339545,
      "loss": 3.1544,
      "step": 20570
    },
    {
      "epoch": 3.8735177865612647,
      "grad_norm": 2.7873122692108154,
      "learning_rate": 0.00011264822134387352,
      "loss": 2.8861,
      "step": 20580
    },
    {
      "epoch": 3.875399962356484,
      "grad_norm": 2.592653274536133,
      "learning_rate": 0.00011246000376435159,
      "loss": 3.415,
      "step": 20590
    },
    {
      "epoch": 3.8772821381517035,
      "grad_norm": 2.8197875022888184,
      "learning_rate": 0.00011227178618482966,
      "loss": 3.1919,
      "step": 20600
    },
    {
      "epoch": 3.8791643139469225,
      "grad_norm": 2.8188552856445312,
      "learning_rate": 0.00011208356860530773,
      "loss": 3.2859,
      "step": 20610
    },
    {
      "epoch": 3.881046489742142,
      "grad_norm": 3.598754644393921,
      "learning_rate": 0.00011189535102578582,
      "loss": 3.322,
      "step": 20620
    },
    {
      "epoch": 3.882928665537361,
      "grad_norm": 2.8028104305267334,
      "learning_rate": 0.00011170713344626389,
      "loss": 3.3449,
      "step": 20630
    },
    {
      "epoch": 3.8848108413325804,
      "grad_norm": 3.098341226577759,
      "learning_rate": 0.00011151891586674196,
      "loss": 3.3943,
      "step": 20640
    },
    {
      "epoch": 3.8866930171278,
      "grad_norm": 2.6026523113250732,
      "learning_rate": 0.00011133069828722003,
      "loss": 3.168,
      "step": 20650
    },
    {
      "epoch": 3.8885751929230192,
      "grad_norm": 2.6130552291870117,
      "learning_rate": 0.0001111424807076981,
      "loss": 3.3081,
      "step": 20660
    },
    {
      "epoch": 3.890457368718238,
      "grad_norm": 2.8330435752868652,
      "learning_rate": 0.00011095426312817617,
      "loss": 3.2736,
      "step": 20670
    },
    {
      "epoch": 3.8923395445134576,
      "grad_norm": 2.8563363552093506,
      "learning_rate": 0.00011076604554865425,
      "loss": 3.2376,
      "step": 20680
    },
    {
      "epoch": 3.8942217203086766,
      "grad_norm": 2.8903121948242188,
      "learning_rate": 0.00011057782796913232,
      "loss": 3.0781,
      "step": 20690
    },
    {
      "epoch": 3.896103896103896,
      "grad_norm": 3.307441234588623,
      "learning_rate": 0.00011038961038961039,
      "loss": 3.0561,
      "step": 20700
    },
    {
      "epoch": 3.8979860718991155,
      "grad_norm": 2.68328595161438,
      "learning_rate": 0.00011020139281008846,
      "loss": 3.358,
      "step": 20710
    },
    {
      "epoch": 3.8998682476943345,
      "grad_norm": 2.7960171699523926,
      "learning_rate": 0.00011001317523056653,
      "loss": 3.4266,
      "step": 20720
    },
    {
      "epoch": 3.901750423489554,
      "grad_norm": 2.629624128341675,
      "learning_rate": 0.00010982495765104462,
      "loss": 3.1278,
      "step": 20730
    },
    {
      "epoch": 3.9036325992847734,
      "grad_norm": 3.141634225845337,
      "learning_rate": 0.00010963674007152269,
      "loss": 3.5527,
      "step": 20740
    },
    {
      "epoch": 3.9055147750799923,
      "grad_norm": 2.913501739501953,
      "learning_rate": 0.00010944852249200076,
      "loss": 3.4974,
      "step": 20750
    },
    {
      "epoch": 3.9073969508752118,
      "grad_norm": 2.5849661827087402,
      "learning_rate": 0.00010926030491247883,
      "loss": 3.1552,
      "step": 20760
    },
    {
      "epoch": 3.909279126670431,
      "grad_norm": 2.997143030166626,
      "learning_rate": 0.0001090720873329569,
      "loss": 3.302,
      "step": 20770
    },
    {
      "epoch": 3.91116130246565,
      "grad_norm": 2.5189380645751953,
      "learning_rate": 0.00010888386975343497,
      "loss": 3.3858,
      "step": 20780
    },
    {
      "epoch": 3.9130434782608696,
      "grad_norm": 3.202120780944824,
      "learning_rate": 0.00010869565217391305,
      "loss": 3.8227,
      "step": 20790
    },
    {
      "epoch": 3.9149256540560886,
      "grad_norm": 2.4213297367095947,
      "learning_rate": 0.00010850743459439112,
      "loss": 3.4291,
      "step": 20800
    },
    {
      "epoch": 3.916807829851308,
      "grad_norm": 2.5922842025756836,
      "learning_rate": 0.00010831921701486919,
      "loss": 3.1879,
      "step": 20810
    },
    {
      "epoch": 3.9186900056465275,
      "grad_norm": 2.7009775638580322,
      "learning_rate": 0.00010813099943534726,
      "loss": 3.1444,
      "step": 20820
    },
    {
      "epoch": 3.920572181441747,
      "grad_norm": 2.252577066421509,
      "learning_rate": 0.00010794278185582533,
      "loss": 3.2912,
      "step": 20830
    },
    {
      "epoch": 3.922454357236966,
      "grad_norm": 3.549348831176758,
      "learning_rate": 0.00010775456427630342,
      "loss": 3.2258,
      "step": 20840
    },
    {
      "epoch": 3.9243365330321853,
      "grad_norm": 3.6950178146362305,
      "learning_rate": 0.00010756634669678148,
      "loss": 3.1597,
      "step": 20850
    },
    {
      "epoch": 3.9262187088274043,
      "grad_norm": 2.4231810569763184,
      "learning_rate": 0.00010737812911725956,
      "loss": 2.8874,
      "step": 20860
    },
    {
      "epoch": 3.9281008846226237,
      "grad_norm": 2.3907105922698975,
      "learning_rate": 0.00010718991153773763,
      "loss": 3.15,
      "step": 20870
    },
    {
      "epoch": 3.929983060417843,
      "grad_norm": 2.8133437633514404,
      "learning_rate": 0.0001070016939582157,
      "loss": 3.5011,
      "step": 20880
    },
    {
      "epoch": 3.931865236213062,
      "grad_norm": 2.9195053577423096,
      "learning_rate": 0.00010681347637869377,
      "loss": 3.3655,
      "step": 20890
    },
    {
      "epoch": 3.9337474120082816,
      "grad_norm": 3.61535906791687,
      "learning_rate": 0.00010662525879917185,
      "loss": 3.2766,
      "step": 20900
    },
    {
      "epoch": 3.9356295878035006,
      "grad_norm": 3.8378422260284424,
      "learning_rate": 0.00010643704121964992,
      "loss": 3.515,
      "step": 20910
    },
    {
      "epoch": 3.93751176359872,
      "grad_norm": 2.60741925239563,
      "learning_rate": 0.00010624882364012799,
      "loss": 3.3687,
      "step": 20920
    },
    {
      "epoch": 3.9393939393939394,
      "grad_norm": 2.7538509368896484,
      "learning_rate": 0.00010606060606060606,
      "loss": 3.2729,
      "step": 20930
    },
    {
      "epoch": 3.941276115189159,
      "grad_norm": 2.943190813064575,
      "learning_rate": 0.00010587238848108413,
      "loss": 3.3263,
      "step": 20940
    },
    {
      "epoch": 3.943158290984378,
      "grad_norm": 2.7204387187957764,
      "learning_rate": 0.00010568417090156222,
      "loss": 2.9551,
      "step": 20950
    },
    {
      "epoch": 3.9450404667795973,
      "grad_norm": 2.7319447994232178,
      "learning_rate": 0.00010549595332204028,
      "loss": 3.5181,
      "step": 20960
    },
    {
      "epoch": 3.9469226425748163,
      "grad_norm": 3.1706159114837646,
      "learning_rate": 0.00010530773574251836,
      "loss": 3.3221,
      "step": 20970
    },
    {
      "epoch": 3.9488048183700357,
      "grad_norm": 2.647839307785034,
      "learning_rate": 0.00010511951816299642,
      "loss": 3.5333,
      "step": 20980
    },
    {
      "epoch": 3.950686994165255,
      "grad_norm": 3.027841329574585,
      "learning_rate": 0.0001049313005834745,
      "loss": 3.2927,
      "step": 20990
    },
    {
      "epoch": 3.9525691699604746,
      "grad_norm": 2.752403974533081,
      "learning_rate": 0.00010474308300395257,
      "loss": 3.2126,
      "step": 21000
    },
    {
      "epoch": 3.9544513457556936,
      "grad_norm": 2.5150058269500732,
      "learning_rate": 0.00010455486542443065,
      "loss": 3.5071,
      "step": 21010
    },
    {
      "epoch": 3.956333521550913,
      "grad_norm": 3.4964075088500977,
      "learning_rate": 0.00010436664784490872,
      "loss": 3.2479,
      "step": 21020
    },
    {
      "epoch": 3.958215697346132,
      "grad_norm": 2.55086612701416,
      "learning_rate": 0.00010417843026538679,
      "loss": 3.3188,
      "step": 21030
    },
    {
      "epoch": 3.9600978731413514,
      "grad_norm": 3.518401622772217,
      "learning_rate": 0.00010399021268586486,
      "loss": 3.1695,
      "step": 21040
    },
    {
      "epoch": 3.961980048936571,
      "grad_norm": 2.718132972717285,
      "learning_rate": 0.00010380199510634293,
      "loss": 3.3029,
      "step": 21050
    },
    {
      "epoch": 3.96386222473179,
      "grad_norm": 3.0271036624908447,
      "learning_rate": 0.00010361377752682102,
      "loss": 3.0488,
      "step": 21060
    },
    {
      "epoch": 3.9657444005270093,
      "grad_norm": 2.508157968521118,
      "learning_rate": 0.00010342555994729908,
      "loss": 3.2239,
      "step": 21070
    },
    {
      "epoch": 3.9676265763222283,
      "grad_norm": 3.0764923095703125,
      "learning_rate": 0.00010323734236777716,
      "loss": 3.2407,
      "step": 21080
    },
    {
      "epoch": 3.9695087521174477,
      "grad_norm": 2.797504425048828,
      "learning_rate": 0.00010304912478825522,
      "loss": 3.5661,
      "step": 21090
    },
    {
      "epoch": 3.971390927912667,
      "grad_norm": 3.5259289741516113,
      "learning_rate": 0.0001028609072087333,
      "loss": 3.7717,
      "step": 21100
    },
    {
      "epoch": 3.9732731037078866,
      "grad_norm": 2.6272242069244385,
      "learning_rate": 0.00010267268962921137,
      "loss": 3.4713,
      "step": 21110
    },
    {
      "epoch": 3.9751552795031055,
      "grad_norm": 2.6657400131225586,
      "learning_rate": 0.00010248447204968945,
      "loss": 3.1154,
      "step": 21120
    },
    {
      "epoch": 3.977037455298325,
      "grad_norm": 2.7080202102661133,
      "learning_rate": 0.00010229625447016752,
      "loss": 3.552,
      "step": 21130
    },
    {
      "epoch": 3.978919631093544,
      "grad_norm": 2.928029775619507,
      "learning_rate": 0.00010210803689064559,
      "loss": 3.0428,
      "step": 21140
    },
    {
      "epoch": 3.9808018068887634,
      "grad_norm": 3.023933172225952,
      "learning_rate": 0.00010191981931112366,
      "loss": 3.2027,
      "step": 21150
    },
    {
      "epoch": 3.982683982683983,
      "grad_norm": 2.8178048133850098,
      "learning_rate": 0.00010173160173160173,
      "loss": 3.3582,
      "step": 21160
    },
    {
      "epoch": 3.984566158479202,
      "grad_norm": 2.425347089767456,
      "learning_rate": 0.00010154338415207982,
      "loss": 3.1945,
      "step": 21170
    },
    {
      "epoch": 3.9864483342744212,
      "grad_norm": 4.833703994750977,
      "learning_rate": 0.00010135516657255788,
      "loss": 3.3424,
      "step": 21180
    },
    {
      "epoch": 3.9883305100696402,
      "grad_norm": 3.0012435913085938,
      "learning_rate": 0.00010116694899303596,
      "loss": 3.3914,
      "step": 21190
    },
    {
      "epoch": 3.9902126858648597,
      "grad_norm": 2.9188785552978516,
      "learning_rate": 0.00010097873141351402,
      "loss": 3.1256,
      "step": 21200
    },
    {
      "epoch": 3.992094861660079,
      "grad_norm": 2.435695171356201,
      "learning_rate": 0.0001007905138339921,
      "loss": 3.1203,
      "step": 21210
    },
    {
      "epoch": 3.9939770374552985,
      "grad_norm": 3.414642810821533,
      "learning_rate": 0.00010060229625447016,
      "loss": 3.371,
      "step": 21220
    },
    {
      "epoch": 3.9958592132505175,
      "grad_norm": 2.751316785812378,
      "learning_rate": 0.00010041407867494825,
      "loss": 3.0688,
      "step": 21230
    },
    {
      "epoch": 3.997741389045737,
      "grad_norm": 2.6124749183654785,
      "learning_rate": 0.00010022586109542632,
      "loss": 3.2969,
      "step": 21240
    },
    {
      "epoch": 3.999623564840956,
      "grad_norm": 2.908557653427124,
      "learning_rate": 0.00010003764351590439,
      "loss": 2.9505,
      "step": 21250
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.20933333333333334,
      "eval_loss": 3.2325947284698486,
      "eval_runtime": 94.8421,
      "eval_samples_per_second": 79.079,
      "eval_steps_per_second": 9.89,
      "step": 21252
    },
    {
      "epoch": 4.001505740636175,
      "grad_norm": 2.891474962234497,
      "learning_rate": 9.984942593638246e-05,
      "loss": 3.1685,
      "step": 21260
    },
    {
      "epoch": 4.003387916431395,
      "grad_norm": 2.610499858856201,
      "learning_rate": 9.966120835686053e-05,
      "loss": 2.977,
      "step": 21270
    },
    {
      "epoch": 4.005270092226614,
      "grad_norm": 2.853829860687256,
      "learning_rate": 9.947299077733862e-05,
      "loss": 3.1707,
      "step": 21280
    },
    {
      "epoch": 4.007152268021834,
      "grad_norm": 3.7910237312316895,
      "learning_rate": 9.928477319781668e-05,
      "loss": 3.3487,
      "step": 21290
    },
    {
      "epoch": 4.009034443817052,
      "grad_norm": 3.1136045455932617,
      "learning_rate": 9.909655561829476e-05,
      "loss": 3.354,
      "step": 21300
    },
    {
      "epoch": 4.010916619612272,
      "grad_norm": 2.9077839851379395,
      "learning_rate": 9.890833803877282e-05,
      "loss": 3.208,
      "step": 21310
    },
    {
      "epoch": 4.012798795407491,
      "grad_norm": 2.4560554027557373,
      "learning_rate": 9.87201204592509e-05,
      "loss": 3.4798,
      "step": 21320
    },
    {
      "epoch": 4.0146809712027105,
      "grad_norm": 3.8288331031799316,
      "learning_rate": 9.853190287972896e-05,
      "loss": 3.3021,
      "step": 21330
    },
    {
      "epoch": 4.01656314699793,
      "grad_norm": 2.5452582836151123,
      "learning_rate": 9.834368530020705e-05,
      "loss": 3.2399,
      "step": 21340
    },
    {
      "epoch": 4.0184453227931485,
      "grad_norm": 3.3501482009887695,
      "learning_rate": 9.815546772068511e-05,
      "loss": 3.1027,
      "step": 21350
    },
    {
      "epoch": 4.020327498588368,
      "grad_norm": 2.5182175636291504,
      "learning_rate": 9.796725014116319e-05,
      "loss": 3.075,
      "step": 21360
    },
    {
      "epoch": 4.022209674383587,
      "grad_norm": 2.5452823638916016,
      "learning_rate": 9.777903256164126e-05,
      "loss": 3.3111,
      "step": 21370
    },
    {
      "epoch": 4.024091850178807,
      "grad_norm": 2.716010570526123,
      "learning_rate": 9.759081498211933e-05,
      "loss": 3.1905,
      "step": 21380
    },
    {
      "epoch": 4.025974025974026,
      "grad_norm": 2.6728432178497314,
      "learning_rate": 9.740259740259742e-05,
      "loss": 3.4892,
      "step": 21390
    },
    {
      "epoch": 4.027856201769246,
      "grad_norm": 3.552994728088379,
      "learning_rate": 9.721437982307548e-05,
      "loss": 3.2987,
      "step": 21400
    },
    {
      "epoch": 4.029738377564464,
      "grad_norm": 2.918489933013916,
      "learning_rate": 9.702616224355356e-05,
      "loss": 3.4667,
      "step": 21410
    },
    {
      "epoch": 4.031620553359684,
      "grad_norm": 3.082736015319824,
      "learning_rate": 9.683794466403162e-05,
      "loss": 3.2245,
      "step": 21420
    },
    {
      "epoch": 4.033502729154903,
      "grad_norm": 2.7363715171813965,
      "learning_rate": 9.66497270845097e-05,
      "loss": 3.3247,
      "step": 21430
    },
    {
      "epoch": 4.0353849049501225,
      "grad_norm": 2.707773447036743,
      "learning_rate": 9.646150950498776e-05,
      "loss": 3.3738,
      "step": 21440
    },
    {
      "epoch": 4.037267080745342,
      "grad_norm": 2.463040828704834,
      "learning_rate": 9.627329192546585e-05,
      "loss": 3.1729,
      "step": 21450
    },
    {
      "epoch": 4.0391492565405605,
      "grad_norm": 3.271688938140869,
      "learning_rate": 9.608507434594391e-05,
      "loss": 3.1545,
      "step": 21460
    },
    {
      "epoch": 4.04103143233578,
      "grad_norm": 2.990217685699463,
      "learning_rate": 9.589685676642199e-05,
      "loss": 2.9567,
      "step": 21470
    },
    {
      "epoch": 4.042913608130999,
      "grad_norm": 2.7719483375549316,
      "learning_rate": 9.570863918690006e-05,
      "loss": 3.0559,
      "step": 21480
    },
    {
      "epoch": 4.044795783926219,
      "grad_norm": 2.990938425064087,
      "learning_rate": 9.552042160737813e-05,
      "loss": 3.3036,
      "step": 21490
    },
    {
      "epoch": 4.046677959721438,
      "grad_norm": 2.75420880317688,
      "learning_rate": 9.53322040278562e-05,
      "loss": 3.2986,
      "step": 21500
    },
    {
      "epoch": 4.048560135516658,
      "grad_norm": 3.4968814849853516,
      "learning_rate": 9.514398644833428e-05,
      "loss": 3.2688,
      "step": 21510
    },
    {
      "epoch": 4.050442311311876,
      "grad_norm": 3.2569799423217773,
      "learning_rate": 9.495576886881236e-05,
      "loss": 3.388,
      "step": 21520
    },
    {
      "epoch": 4.052324487107096,
      "grad_norm": 3.2045726776123047,
      "learning_rate": 9.476755128929042e-05,
      "loss": 3.0191,
      "step": 21530
    },
    {
      "epoch": 4.054206662902315,
      "grad_norm": 2.2563700675964355,
      "learning_rate": 9.45793337097685e-05,
      "loss": 3.3321,
      "step": 21540
    },
    {
      "epoch": 4.0560888386975344,
      "grad_norm": 2.6916544437408447,
      "learning_rate": 9.439111613024656e-05,
      "loss": 3.5008,
      "step": 21550
    },
    {
      "epoch": 4.057971014492754,
      "grad_norm": 3.3092386722564697,
      "learning_rate": 9.420289855072465e-05,
      "loss": 3.1675,
      "step": 21560
    },
    {
      "epoch": 4.059853190287973,
      "grad_norm": 2.2328038215637207,
      "learning_rate": 9.401468097120271e-05,
      "loss": 3.2129,
      "step": 21570
    },
    {
      "epoch": 4.061735366083192,
      "grad_norm": 2.782254695892334,
      "learning_rate": 9.382646339168079e-05,
      "loss": 3.1073,
      "step": 21580
    },
    {
      "epoch": 4.063617541878411,
      "grad_norm": 2.7193851470947266,
      "learning_rate": 9.363824581215885e-05,
      "loss": 3.1614,
      "step": 21590
    },
    {
      "epoch": 4.065499717673631,
      "grad_norm": 2.849473714828491,
      "learning_rate": 9.345002823263693e-05,
      "loss": 2.7911,
      "step": 21600
    },
    {
      "epoch": 4.06738189346885,
      "grad_norm": 3.0582633018493652,
      "learning_rate": 9.3261810653115e-05,
      "loss": 3.5445,
      "step": 21610
    },
    {
      "epoch": 4.06926406926407,
      "grad_norm": 2.8459339141845703,
      "learning_rate": 9.307359307359308e-05,
      "loss": 3.1633,
      "step": 21620
    },
    {
      "epoch": 4.071146245059288,
      "grad_norm": 2.8063242435455322,
      "learning_rate": 9.288537549407116e-05,
      "loss": 3.3386,
      "step": 21630
    },
    {
      "epoch": 4.073028420854508,
      "grad_norm": 3.0272722244262695,
      "learning_rate": 9.269715791454922e-05,
      "loss": 3.1299,
      "step": 21640
    },
    {
      "epoch": 4.074910596649727,
      "grad_norm": 2.7872672080993652,
      "learning_rate": 9.25089403350273e-05,
      "loss": 3.2995,
      "step": 21650
    },
    {
      "epoch": 4.076792772444946,
      "grad_norm": 3.4354023933410645,
      "learning_rate": 9.232072275550536e-05,
      "loss": 3.205,
      "step": 21660
    },
    {
      "epoch": 4.078674948240166,
      "grad_norm": 2.8815746307373047,
      "learning_rate": 9.213250517598345e-05,
      "loss": 3.1771,
      "step": 21670
    },
    {
      "epoch": 4.080557124035385,
      "grad_norm": 3.5228142738342285,
      "learning_rate": 9.194428759646151e-05,
      "loss": 3.0383,
      "step": 21680
    },
    {
      "epoch": 4.082439299830604,
      "grad_norm": 3.062067985534668,
      "learning_rate": 9.175607001693959e-05,
      "loss": 3.3322,
      "step": 21690
    },
    {
      "epoch": 4.084321475625823,
      "grad_norm": 4.2620015144348145,
      "learning_rate": 9.156785243741765e-05,
      "loss": 3.1793,
      "step": 21700
    },
    {
      "epoch": 4.086203651421043,
      "grad_norm": 4.1625285148620605,
      "learning_rate": 9.137963485789573e-05,
      "loss": 3.3753,
      "step": 21710
    },
    {
      "epoch": 4.088085827216262,
      "grad_norm": 3.0724854469299316,
      "learning_rate": 9.11914172783738e-05,
      "loss": 3.183,
      "step": 21720
    },
    {
      "epoch": 4.089968003011482,
      "grad_norm": 4.273207664489746,
      "learning_rate": 9.100319969885188e-05,
      "loss": 3.0482,
      "step": 21730
    },
    {
      "epoch": 4.0918501788067,
      "grad_norm": 2.5469064712524414,
      "learning_rate": 9.081498211932996e-05,
      "loss": 3.234,
      "step": 21740
    },
    {
      "epoch": 4.0937323546019195,
      "grad_norm": 2.6905412673950195,
      "learning_rate": 9.062676453980802e-05,
      "loss": 3.2977,
      "step": 21750
    },
    {
      "epoch": 4.095614530397139,
      "grad_norm": 3.4132399559020996,
      "learning_rate": 9.04385469602861e-05,
      "loss": 3.3364,
      "step": 21760
    },
    {
      "epoch": 4.097496706192358,
      "grad_norm": 3.024416208267212,
      "learning_rate": 9.025032938076416e-05,
      "loss": 3.4897,
      "step": 21770
    },
    {
      "epoch": 4.099378881987578,
      "grad_norm": 2.8688106536865234,
      "learning_rate": 9.006211180124225e-05,
      "loss": 3.2479,
      "step": 21780
    },
    {
      "epoch": 4.101261057782797,
      "grad_norm": 4.9075398445129395,
      "learning_rate": 8.987389422172031e-05,
      "loss": 2.818,
      "step": 21790
    },
    {
      "epoch": 4.103143233578016,
      "grad_norm": 2.8078277111053467,
      "learning_rate": 8.968567664219839e-05,
      "loss": 3.1297,
      "step": 21800
    },
    {
      "epoch": 4.105025409373235,
      "grad_norm": 3.4160094261169434,
      "learning_rate": 8.949745906267645e-05,
      "loss": 3.3984,
      "step": 21810
    },
    {
      "epoch": 4.106907585168455,
      "grad_norm": 3.5791547298431396,
      "learning_rate": 8.930924148315453e-05,
      "loss": 2.9892,
      "step": 21820
    },
    {
      "epoch": 4.108789760963674,
      "grad_norm": 2.7223494052886963,
      "learning_rate": 8.912102390363259e-05,
      "loss": 3.1671,
      "step": 21830
    },
    {
      "epoch": 4.1106719367588935,
      "grad_norm": 3.5885062217712402,
      "learning_rate": 8.893280632411068e-05,
      "loss": 3.1588,
      "step": 21840
    },
    {
      "epoch": 4.112554112554113,
      "grad_norm": 3.027226686477661,
      "learning_rate": 8.874458874458876e-05,
      "loss": 3.1116,
      "step": 21850
    },
    {
      "epoch": 4.1144362883493315,
      "grad_norm": 2.6472368240356445,
      "learning_rate": 8.855637116506682e-05,
      "loss": 3.3602,
      "step": 21860
    },
    {
      "epoch": 4.116318464144551,
      "grad_norm": 3.1893270015716553,
      "learning_rate": 8.83681535855449e-05,
      "loss": 3.0538,
      "step": 21870
    },
    {
      "epoch": 4.11820063993977,
      "grad_norm": 3.10514760017395,
      "learning_rate": 8.817993600602296e-05,
      "loss": 3.4478,
      "step": 21880
    },
    {
      "epoch": 4.12008281573499,
      "grad_norm": 2.870718002319336,
      "learning_rate": 8.799171842650105e-05,
      "loss": 3.3483,
      "step": 21890
    },
    {
      "epoch": 4.121964991530209,
      "grad_norm": 2.7556862831115723,
      "learning_rate": 8.780350084697911e-05,
      "loss": 3.0957,
      "step": 21900
    },
    {
      "epoch": 4.123847167325428,
      "grad_norm": 3.0777435302734375,
      "learning_rate": 8.761528326745719e-05,
      "loss": 3.0024,
      "step": 21910
    },
    {
      "epoch": 4.125729343120647,
      "grad_norm": 4.2900567054748535,
      "learning_rate": 8.742706568793525e-05,
      "loss": 3.3981,
      "step": 21920
    },
    {
      "epoch": 4.127611518915867,
      "grad_norm": 3.218428373336792,
      "learning_rate": 8.723884810841333e-05,
      "loss": 3.1353,
      "step": 21930
    },
    {
      "epoch": 4.129493694711086,
      "grad_norm": 3.2349698543548584,
      "learning_rate": 8.705063052889139e-05,
      "loss": 3.1707,
      "step": 21940
    },
    {
      "epoch": 4.1313758705063055,
      "grad_norm": 3.2318453788757324,
      "learning_rate": 8.686241294936948e-05,
      "loss": 3.1009,
      "step": 21950
    },
    {
      "epoch": 4.133258046301525,
      "grad_norm": 3.21386456489563,
      "learning_rate": 8.667419536984754e-05,
      "loss": 3.1957,
      "step": 21960
    },
    {
      "epoch": 4.1351402220967435,
      "grad_norm": 2.730018138885498,
      "learning_rate": 8.648597779032562e-05,
      "loss": 3.1705,
      "step": 21970
    },
    {
      "epoch": 4.137022397891963,
      "grad_norm": 3.171736001968384,
      "learning_rate": 8.62977602108037e-05,
      "loss": 3.7192,
      "step": 21980
    },
    {
      "epoch": 4.138904573687182,
      "grad_norm": 3.063369035720825,
      "learning_rate": 8.610954263128176e-05,
      "loss": 3.3647,
      "step": 21990
    },
    {
      "epoch": 4.140786749482402,
      "grad_norm": 3.849029779434204,
      "learning_rate": 8.592132505175985e-05,
      "loss": 3.2821,
      "step": 22000
    },
    {
      "epoch": 4.142668925277621,
      "grad_norm": 4.603402614593506,
      "learning_rate": 8.573310747223791e-05,
      "loss": 3.2007,
      "step": 22010
    },
    {
      "epoch": 4.144551101072841,
      "grad_norm": 2.3862357139587402,
      "learning_rate": 8.554488989271599e-05,
      "loss": 3.3233,
      "step": 22020
    },
    {
      "epoch": 4.146433276868059,
      "grad_norm": 3.104445457458496,
      "learning_rate": 8.535667231319405e-05,
      "loss": 2.9834,
      "step": 22030
    },
    {
      "epoch": 4.148315452663279,
      "grad_norm": 3.098273515701294,
      "learning_rate": 8.516845473367213e-05,
      "loss": 3.2513,
      "step": 22040
    },
    {
      "epoch": 4.150197628458498,
      "grad_norm": 3.133579730987549,
      "learning_rate": 8.498023715415019e-05,
      "loss": 3.2901,
      "step": 22050
    },
    {
      "epoch": 4.1520798042537175,
      "grad_norm": 2.9663634300231934,
      "learning_rate": 8.479201957462828e-05,
      "loss": 3.4831,
      "step": 22060
    },
    {
      "epoch": 4.153961980048937,
      "grad_norm": 2.842804193496704,
      "learning_rate": 8.460380199510634e-05,
      "loss": 3.1227,
      "step": 22070
    },
    {
      "epoch": 4.1558441558441555,
      "grad_norm": 3.0301401615142822,
      "learning_rate": 8.441558441558442e-05,
      "loss": 3.1276,
      "step": 22080
    },
    {
      "epoch": 4.157726331639375,
      "grad_norm": 3.2375950813293457,
      "learning_rate": 8.42273668360625e-05,
      "loss": 3.3902,
      "step": 22090
    },
    {
      "epoch": 4.159608507434594,
      "grad_norm": 2.83935809135437,
      "learning_rate": 8.403914925654056e-05,
      "loss": 3.5444,
      "step": 22100
    },
    {
      "epoch": 4.161490683229814,
      "grad_norm": 2.878446578979492,
      "learning_rate": 8.385093167701865e-05,
      "loss": 3.2114,
      "step": 22110
    },
    {
      "epoch": 4.163372859025033,
      "grad_norm": 2.6201987266540527,
      "learning_rate": 8.366271409749671e-05,
      "loss": 3.2031,
      "step": 22120
    },
    {
      "epoch": 4.165255034820253,
      "grad_norm": 3.6080262660980225,
      "learning_rate": 8.347449651797479e-05,
      "loss": 3.3242,
      "step": 22130
    },
    {
      "epoch": 4.167137210615471,
      "grad_norm": 4.1051411628723145,
      "learning_rate": 8.328627893845285e-05,
      "loss": 3.4341,
      "step": 22140
    },
    {
      "epoch": 4.169019386410691,
      "grad_norm": 4.344814300537109,
      "learning_rate": 8.309806135893093e-05,
      "loss": 3.2216,
      "step": 22150
    },
    {
      "epoch": 4.17090156220591,
      "grad_norm": 3.445652723312378,
      "learning_rate": 8.290984377940899e-05,
      "loss": 2.9959,
      "step": 22160
    },
    {
      "epoch": 4.1727837380011294,
      "grad_norm": 3.2097978591918945,
      "learning_rate": 8.272162619988708e-05,
      "loss": 3.2192,
      "step": 22170
    },
    {
      "epoch": 4.174665913796349,
      "grad_norm": 3.208712339401245,
      "learning_rate": 8.253340862036514e-05,
      "loss": 3.1598,
      "step": 22180
    },
    {
      "epoch": 4.176548089591567,
      "grad_norm": 3.730483055114746,
      "learning_rate": 8.234519104084322e-05,
      "loss": 3.4319,
      "step": 22190
    },
    {
      "epoch": 4.178430265386787,
      "grad_norm": 3.2604448795318604,
      "learning_rate": 8.215697346132128e-05,
      "loss": 3.0354,
      "step": 22200
    },
    {
      "epoch": 4.180312441182006,
      "grad_norm": 2.9703927040100098,
      "learning_rate": 8.196875588179936e-05,
      "loss": 3.4206,
      "step": 22210
    },
    {
      "epoch": 4.182194616977226,
      "grad_norm": 3.236666440963745,
      "learning_rate": 8.178053830227745e-05,
      "loss": 3.3407,
      "step": 22220
    },
    {
      "epoch": 4.184076792772445,
      "grad_norm": 3.3799564838409424,
      "learning_rate": 8.159232072275551e-05,
      "loss": 2.9908,
      "step": 22230
    },
    {
      "epoch": 4.185958968567665,
      "grad_norm": 3.3381948471069336,
      "learning_rate": 8.140410314323359e-05,
      "loss": 3.1457,
      "step": 22240
    },
    {
      "epoch": 4.187841144362883,
      "grad_norm": 2.3282175064086914,
      "learning_rate": 8.121588556371165e-05,
      "loss": 3.0359,
      "step": 22250
    },
    {
      "epoch": 4.189723320158103,
      "grad_norm": 2.6722915172576904,
      "learning_rate": 8.102766798418973e-05,
      "loss": 3.6484,
      "step": 22260
    },
    {
      "epoch": 4.191605495953322,
      "grad_norm": 3.148141622543335,
      "learning_rate": 8.083945040466779e-05,
      "loss": 3.3126,
      "step": 22270
    },
    {
      "epoch": 4.193487671748541,
      "grad_norm": 2.9252452850341797,
      "learning_rate": 8.065123282514588e-05,
      "loss": 3.1325,
      "step": 22280
    },
    {
      "epoch": 4.195369847543761,
      "grad_norm": 3.091193437576294,
      "learning_rate": 8.046301524562394e-05,
      "loss": 3.4042,
      "step": 22290
    },
    {
      "epoch": 4.19725202333898,
      "grad_norm": 3.0783419609069824,
      "learning_rate": 8.027479766610202e-05,
      "loss": 2.6655,
      "step": 22300
    },
    {
      "epoch": 4.199134199134199,
      "grad_norm": 3.132167100906372,
      "learning_rate": 8.008658008658008e-05,
      "loss": 3.463,
      "step": 22310
    },
    {
      "epoch": 4.201016374929418,
      "grad_norm": 3.468315601348877,
      "learning_rate": 7.989836250705816e-05,
      "loss": 3.6242,
      "step": 22320
    },
    {
      "epoch": 4.202898550724638,
      "grad_norm": 3.142845392227173,
      "learning_rate": 7.971014492753622e-05,
      "loss": 3.2898,
      "step": 22330
    },
    {
      "epoch": 4.204780726519857,
      "grad_norm": 2.87841534614563,
      "learning_rate": 7.952192734801431e-05,
      "loss": 3.0523,
      "step": 22340
    },
    {
      "epoch": 4.206662902315077,
      "grad_norm": 2.8895421028137207,
      "learning_rate": 7.933370976849239e-05,
      "loss": 3.2179,
      "step": 22350
    },
    {
      "epoch": 4.208545078110295,
      "grad_norm": 3.0601487159729004,
      "learning_rate": 7.914549218897045e-05,
      "loss": 3.3932,
      "step": 22360
    },
    {
      "epoch": 4.2104272539055145,
      "grad_norm": 2.776848793029785,
      "learning_rate": 7.895727460944853e-05,
      "loss": 3.21,
      "step": 22370
    },
    {
      "epoch": 4.212309429700734,
      "grad_norm": 3.218050718307495,
      "learning_rate": 7.876905702992659e-05,
      "loss": 3.3461,
      "step": 22380
    },
    {
      "epoch": 4.214191605495953,
      "grad_norm": 2.931539535522461,
      "learning_rate": 7.858083945040468e-05,
      "loss": 3.0586,
      "step": 22390
    },
    {
      "epoch": 4.216073781291173,
      "grad_norm": 3.7356948852539062,
      "learning_rate": 7.839262187088274e-05,
      "loss": 3.2584,
      "step": 22400
    },
    {
      "epoch": 4.217955957086392,
      "grad_norm": 2.919062852859497,
      "learning_rate": 7.820440429136082e-05,
      "loss": 3.0511,
      "step": 22410
    },
    {
      "epoch": 4.219838132881611,
      "grad_norm": 3.2050087451934814,
      "learning_rate": 7.801618671183888e-05,
      "loss": 3.26,
      "step": 22420
    },
    {
      "epoch": 4.22172030867683,
      "grad_norm": 3.362515926361084,
      "learning_rate": 7.782796913231696e-05,
      "loss": 3.176,
      "step": 22430
    },
    {
      "epoch": 4.22360248447205,
      "grad_norm": 2.7202816009521484,
      "learning_rate": 7.763975155279502e-05,
      "loss": 3.3341,
      "step": 22440
    },
    {
      "epoch": 4.225484660267269,
      "grad_norm": 2.699904203414917,
      "learning_rate": 7.745153397327311e-05,
      "loss": 3.4156,
      "step": 22450
    },
    {
      "epoch": 4.2273668360624885,
      "grad_norm": 3.7537598609924316,
      "learning_rate": 7.726331639375119e-05,
      "loss": 3.239,
      "step": 22460
    },
    {
      "epoch": 4.229249011857707,
      "grad_norm": 2.781452178955078,
      "learning_rate": 7.707509881422925e-05,
      "loss": 3.1356,
      "step": 22470
    },
    {
      "epoch": 4.2311311876529265,
      "grad_norm": 4.503787994384766,
      "learning_rate": 7.688688123470733e-05,
      "loss": 3.3394,
      "step": 22480
    },
    {
      "epoch": 4.233013363448146,
      "grad_norm": 2.8764593601226807,
      "learning_rate": 7.669866365518539e-05,
      "loss": 3.3024,
      "step": 22490
    },
    {
      "epoch": 4.234895539243365,
      "grad_norm": 2.42519211769104,
      "learning_rate": 7.651044607566348e-05,
      "loss": 3.2218,
      "step": 22500
    },
    {
      "epoch": 4.236777715038585,
      "grad_norm": 3.315824508666992,
      "learning_rate": 7.632222849614154e-05,
      "loss": 2.7431,
      "step": 22510
    },
    {
      "epoch": 4.238659890833804,
      "grad_norm": 2.6028430461883545,
      "learning_rate": 7.613401091661962e-05,
      "loss": 3.4724,
      "step": 22520
    },
    {
      "epoch": 4.240542066629023,
      "grad_norm": 3.285188674926758,
      "learning_rate": 7.594579333709768e-05,
      "loss": 3.3218,
      "step": 22530
    },
    {
      "epoch": 4.242424242424242,
      "grad_norm": 3.028482675552368,
      "learning_rate": 7.575757575757576e-05,
      "loss": 3.4616,
      "step": 22540
    },
    {
      "epoch": 4.244306418219462,
      "grad_norm": 3.097353219985962,
      "learning_rate": 7.556935817805382e-05,
      "loss": 3.359,
      "step": 22550
    },
    {
      "epoch": 4.246188594014681,
      "grad_norm": 3.4922540187835693,
      "learning_rate": 7.538114059853191e-05,
      "loss": 3.0898,
      "step": 22560
    },
    {
      "epoch": 4.2480707698099005,
      "grad_norm": 2.766566753387451,
      "learning_rate": 7.519292301900997e-05,
      "loss": 3.1267,
      "step": 22570
    },
    {
      "epoch": 4.24995294560512,
      "grad_norm": 4.96455717086792,
      "learning_rate": 7.500470543948805e-05,
      "loss": 3.3342,
      "step": 22580
    },
    {
      "epoch": 4.2518351214003385,
      "grad_norm": 4.132570743560791,
      "learning_rate": 7.481648785996613e-05,
      "loss": 3.0486,
      "step": 22590
    },
    {
      "epoch": 4.253717297195558,
      "grad_norm": 2.6853671073913574,
      "learning_rate": 7.462827028044419e-05,
      "loss": 3.0899,
      "step": 22600
    },
    {
      "epoch": 4.255599472990777,
      "grad_norm": 2.830772638320923,
      "learning_rate": 7.444005270092228e-05,
      "loss": 3.3514,
      "step": 22610
    },
    {
      "epoch": 4.257481648785997,
      "grad_norm": 3.1268768310546875,
      "learning_rate": 7.425183512140034e-05,
      "loss": 3.2823,
      "step": 22620
    },
    {
      "epoch": 4.259363824581216,
      "grad_norm": 3.380945920944214,
      "learning_rate": 7.406361754187842e-05,
      "loss": 3.5613,
      "step": 22630
    },
    {
      "epoch": 4.261246000376435,
      "grad_norm": 2.9997637271881104,
      "learning_rate": 7.387539996235648e-05,
      "loss": 3.1986,
      "step": 22640
    },
    {
      "epoch": 4.263128176171654,
      "grad_norm": 3.073002338409424,
      "learning_rate": 7.368718238283456e-05,
      "loss": 3.122,
      "step": 22650
    },
    {
      "epoch": 4.265010351966874,
      "grad_norm": 2.674787759780884,
      "learning_rate": 7.349896480331262e-05,
      "loss": 3.1517,
      "step": 22660
    },
    {
      "epoch": 4.266892527762093,
      "grad_norm": 2.988823652267456,
      "learning_rate": 7.331074722379071e-05,
      "loss": 3.3729,
      "step": 22670
    },
    {
      "epoch": 4.2687747035573125,
      "grad_norm": 2.2312140464782715,
      "learning_rate": 7.312252964426877e-05,
      "loss": 3.2023,
      "step": 22680
    },
    {
      "epoch": 4.270656879352532,
      "grad_norm": 2.4428207874298096,
      "learning_rate": 7.293431206474685e-05,
      "loss": 3.0415,
      "step": 22690
    },
    {
      "epoch": 4.2725390551477505,
      "grad_norm": 2.431781530380249,
      "learning_rate": 7.274609448522493e-05,
      "loss": 3.0563,
      "step": 22700
    },
    {
      "epoch": 4.27442123094297,
      "grad_norm": 2.9765193462371826,
      "learning_rate": 7.255787690570299e-05,
      "loss": 3.0556,
      "step": 22710
    },
    {
      "epoch": 4.276303406738189,
      "grad_norm": 2.8088014125823975,
      "learning_rate": 7.236965932618108e-05,
      "loss": 3.443,
      "step": 22720
    },
    {
      "epoch": 4.278185582533409,
      "grad_norm": 2.609876871109009,
      "learning_rate": 7.218144174665914e-05,
      "loss": 3.1875,
      "step": 22730
    },
    {
      "epoch": 4.280067758328628,
      "grad_norm": 3.226731300354004,
      "learning_rate": 7.199322416713722e-05,
      "loss": 3.415,
      "step": 22740
    },
    {
      "epoch": 4.281949934123848,
      "grad_norm": 2.7580599784851074,
      "learning_rate": 7.180500658761528e-05,
      "loss": 3.058,
      "step": 22750
    },
    {
      "epoch": 4.283832109919066,
      "grad_norm": 2.7163636684417725,
      "learning_rate": 7.161678900809336e-05,
      "loss": 3.0862,
      "step": 22760
    },
    {
      "epoch": 4.285714285714286,
      "grad_norm": 2.734267473220825,
      "learning_rate": 7.142857142857142e-05,
      "loss": 3.2912,
      "step": 22770
    },
    {
      "epoch": 4.287596461509505,
      "grad_norm": 3.4720916748046875,
      "learning_rate": 7.124035384904951e-05,
      "loss": 3.1295,
      "step": 22780
    },
    {
      "epoch": 4.2894786373047245,
      "grad_norm": 2.542668104171753,
      "learning_rate": 7.105213626952757e-05,
      "loss": 3.3263,
      "step": 22790
    },
    {
      "epoch": 4.291360813099944,
      "grad_norm": 2.874943494796753,
      "learning_rate": 7.086391869000565e-05,
      "loss": 3.111,
      "step": 22800
    },
    {
      "epoch": 4.293242988895162,
      "grad_norm": 3.1843760013580322,
      "learning_rate": 7.067570111048371e-05,
      "loss": 3.1787,
      "step": 22810
    },
    {
      "epoch": 4.295125164690382,
      "grad_norm": 2.4640989303588867,
      "learning_rate": 7.048748353096179e-05,
      "loss": 3.0852,
      "step": 22820
    },
    {
      "epoch": 4.297007340485601,
      "grad_norm": 2.776172161102295,
      "learning_rate": 7.029926595143988e-05,
      "loss": 2.6763,
      "step": 22830
    },
    {
      "epoch": 4.298889516280821,
      "grad_norm": 3.5256826877593994,
      "learning_rate": 7.011104837191794e-05,
      "loss": 2.8748,
      "step": 22840
    },
    {
      "epoch": 4.30077169207604,
      "grad_norm": 3.026275396347046,
      "learning_rate": 6.992283079239602e-05,
      "loss": 3.1198,
      "step": 22850
    },
    {
      "epoch": 4.30265386787126,
      "grad_norm": 3.187401533126831,
      "learning_rate": 6.973461321287408e-05,
      "loss": 3.2494,
      "step": 22860
    },
    {
      "epoch": 4.304536043666478,
      "grad_norm": 2.384568452835083,
      "learning_rate": 6.954639563335216e-05,
      "loss": 3.1702,
      "step": 22870
    },
    {
      "epoch": 4.306418219461698,
      "grad_norm": 3.4012176990509033,
      "learning_rate": 6.935817805383022e-05,
      "loss": 3.0068,
      "step": 22880
    },
    {
      "epoch": 4.308300395256917,
      "grad_norm": 3.3398947715759277,
      "learning_rate": 6.916996047430831e-05,
      "loss": 3.1857,
      "step": 22890
    },
    {
      "epoch": 4.310182571052136,
      "grad_norm": 2.42055606842041,
      "learning_rate": 6.898174289478637e-05,
      "loss": 3.0783,
      "step": 22900
    },
    {
      "epoch": 4.312064746847356,
      "grad_norm": 3.519979476928711,
      "learning_rate": 6.879352531526445e-05,
      "loss": 3.2145,
      "step": 22910
    },
    {
      "epoch": 4.313946922642575,
      "grad_norm": 3.9459290504455566,
      "learning_rate": 6.860530773574251e-05,
      "loss": 3.5089,
      "step": 22920
    },
    {
      "epoch": 4.315829098437794,
      "grad_norm": 2.7116644382476807,
      "learning_rate": 6.841709015622059e-05,
      "loss": 3.2327,
      "step": 22930
    },
    {
      "epoch": 4.317711274233013,
      "grad_norm": 3.385389566421509,
      "learning_rate": 6.822887257669865e-05,
      "loss": 3.3809,
      "step": 22940
    },
    {
      "epoch": 4.319593450028233,
      "grad_norm": 3.4308905601501465,
      "learning_rate": 6.804065499717674e-05,
      "loss": 3.067,
      "step": 22950
    },
    {
      "epoch": 4.321475625823452,
      "grad_norm": 2.6280534267425537,
      "learning_rate": 6.785243741765482e-05,
      "loss": 3.3243,
      "step": 22960
    },
    {
      "epoch": 4.323357801618672,
      "grad_norm": 2.3873131275177,
      "learning_rate": 6.766421983813288e-05,
      "loss": 2.8105,
      "step": 22970
    },
    {
      "epoch": 4.32523997741389,
      "grad_norm": 2.873793601989746,
      "learning_rate": 6.747600225861096e-05,
      "loss": 3.563,
      "step": 22980
    },
    {
      "epoch": 4.3271221532091095,
      "grad_norm": 3.03075909614563,
      "learning_rate": 6.728778467908902e-05,
      "loss": 3.2434,
      "step": 22990
    },
    {
      "epoch": 4.329004329004329,
      "grad_norm": 3.1492440700531006,
      "learning_rate": 6.709956709956711e-05,
      "loss": 3.1969,
      "step": 23000
    },
    {
      "epoch": 4.330886504799548,
      "grad_norm": 3.263035774230957,
      "learning_rate": 6.691134952004517e-05,
      "loss": 3.1421,
      "step": 23010
    },
    {
      "epoch": 4.332768680594768,
      "grad_norm": 3.79087495803833,
      "learning_rate": 6.672313194052325e-05,
      "loss": 3.2209,
      "step": 23020
    },
    {
      "epoch": 4.334650856389986,
      "grad_norm": 3.310776710510254,
      "learning_rate": 6.653491436100131e-05,
      "loss": 3.321,
      "step": 23030
    },
    {
      "epoch": 4.336533032185206,
      "grad_norm": 3.0996947288513184,
      "learning_rate": 6.634669678147939e-05,
      "loss": 3.0465,
      "step": 23040
    },
    {
      "epoch": 4.338415207980425,
      "grad_norm": 3.124666452407837,
      "learning_rate": 6.615847920195745e-05,
      "loss": 3.2509,
      "step": 23050
    },
    {
      "epoch": 4.340297383775645,
      "grad_norm": 2.6973137855529785,
      "learning_rate": 6.597026162243554e-05,
      "loss": 3.1465,
      "step": 23060
    },
    {
      "epoch": 4.342179559570864,
      "grad_norm": 2.4782652854919434,
      "learning_rate": 6.578204404291362e-05,
      "loss": 3.3048,
      "step": 23070
    },
    {
      "epoch": 4.3440617353660835,
      "grad_norm": 3.4553232192993164,
      "learning_rate": 6.559382646339168e-05,
      "loss": 3.2484,
      "step": 23080
    },
    {
      "epoch": 4.345943911161302,
      "grad_norm": 2.5292747020721436,
      "learning_rate": 6.540560888386976e-05,
      "loss": 3.2178,
      "step": 23090
    },
    {
      "epoch": 4.3478260869565215,
      "grad_norm": 2.555924892425537,
      "learning_rate": 6.521739130434782e-05,
      "loss": 3.1644,
      "step": 23100
    },
    {
      "epoch": 4.349708262751741,
      "grad_norm": 2.9614477157592773,
      "learning_rate": 6.502917372482591e-05,
      "loss": 3.2833,
      "step": 23110
    },
    {
      "epoch": 4.35159043854696,
      "grad_norm": 2.6555614471435547,
      "learning_rate": 6.484095614530397e-05,
      "loss": 3.3832,
      "step": 23120
    },
    {
      "epoch": 4.35347261434218,
      "grad_norm": 2.8622424602508545,
      "learning_rate": 6.465273856578205e-05,
      "loss": 3.1168,
      "step": 23130
    },
    {
      "epoch": 4.355354790137399,
      "grad_norm": 4.439013481140137,
      "learning_rate": 6.446452098626011e-05,
      "loss": 3.0878,
      "step": 23140
    },
    {
      "epoch": 4.357236965932618,
      "grad_norm": 4.426267147064209,
      "learning_rate": 6.427630340673819e-05,
      "loss": 3.3674,
      "step": 23150
    },
    {
      "epoch": 4.359119141727837,
      "grad_norm": 3.5063087940216064,
      "learning_rate": 6.408808582721625e-05,
      "loss": 3.321,
      "step": 23160
    },
    {
      "epoch": 4.361001317523057,
      "grad_norm": 3.535888433456421,
      "learning_rate": 6.389986824769434e-05,
      "loss": 3.2064,
      "step": 23170
    },
    {
      "epoch": 4.362883493318276,
      "grad_norm": 2.365108013153076,
      "learning_rate": 6.37116506681724e-05,
      "loss": 3.0767,
      "step": 23180
    },
    {
      "epoch": 4.3647656691134955,
      "grad_norm": 2.9245288372039795,
      "learning_rate": 6.352343308865048e-05,
      "loss": 3.1926,
      "step": 23190
    },
    {
      "epoch": 4.366647844908714,
      "grad_norm": 3.0852577686309814,
      "learning_rate": 6.333521550912856e-05,
      "loss": 3.4855,
      "step": 23200
    },
    {
      "epoch": 4.3685300207039335,
      "grad_norm": 3.742438316345215,
      "learning_rate": 6.314699792960662e-05,
      "loss": 3.4457,
      "step": 23210
    },
    {
      "epoch": 4.370412196499153,
      "grad_norm": 2.9332962036132812,
      "learning_rate": 6.295878035008471e-05,
      "loss": 3.053,
      "step": 23220
    },
    {
      "epoch": 4.372294372294372,
      "grad_norm": 2.669461250305176,
      "learning_rate": 6.277056277056277e-05,
      "loss": 3.2083,
      "step": 23230
    },
    {
      "epoch": 4.374176548089592,
      "grad_norm": 2.6461141109466553,
      "learning_rate": 6.258234519104085e-05,
      "loss": 3.1934,
      "step": 23240
    },
    {
      "epoch": 4.376058723884811,
      "grad_norm": 4.071197986602783,
      "learning_rate": 6.239412761151891e-05,
      "loss": 3.2333,
      "step": 23250
    },
    {
      "epoch": 4.37794089968003,
      "grad_norm": 3.3596227169036865,
      "learning_rate": 6.220591003199699e-05,
      "loss": 3.3347,
      "step": 23260
    },
    {
      "epoch": 4.379823075475249,
      "grad_norm": 2.816563129425049,
      "learning_rate": 6.201769245247507e-05,
      "loss": 3.2306,
      "step": 23270
    },
    {
      "epoch": 4.381705251270469,
      "grad_norm": 2.9401190280914307,
      "learning_rate": 6.182947487295314e-05,
      "loss": 3.4218,
      "step": 23280
    },
    {
      "epoch": 4.383587427065688,
      "grad_norm": 2.8984081745147705,
      "learning_rate": 6.16412572934312e-05,
      "loss": 3.3952,
      "step": 23290
    },
    {
      "epoch": 4.3854696028609075,
      "grad_norm": 2.7690205574035645,
      "learning_rate": 6.145303971390928e-05,
      "loss": 3.0585,
      "step": 23300
    },
    {
      "epoch": 4.387351778656127,
      "grad_norm": 2.664947271347046,
      "learning_rate": 6.126482213438736e-05,
      "loss": 3.1813,
      "step": 23310
    },
    {
      "epoch": 4.3892339544513455,
      "grad_norm": 3.077943801879883,
      "learning_rate": 6.107660455486542e-05,
      "loss": 3.1159,
      "step": 23320
    },
    {
      "epoch": 4.391116130246565,
      "grad_norm": 3.109022378921509,
      "learning_rate": 6.08883869753435e-05,
      "loss": 3.1827,
      "step": 23330
    },
    {
      "epoch": 4.392998306041784,
      "grad_norm": 2.6905243396759033,
      "learning_rate": 6.0700169395821566e-05,
      "loss": 2.7874,
      "step": 23340
    },
    {
      "epoch": 4.394880481837004,
      "grad_norm": 2.7580599784851074,
      "learning_rate": 6.051195181629964e-05,
      "loss": 3.0988,
      "step": 23350
    },
    {
      "epoch": 4.396762657632223,
      "grad_norm": 2.6568217277526855,
      "learning_rate": 6.032373423677771e-05,
      "loss": 3.0959,
      "step": 23360
    },
    {
      "epoch": 4.398644833427442,
      "grad_norm": 3.559309720993042,
      "learning_rate": 6.013551665725579e-05,
      "loss": 3.3385,
      "step": 23370
    },
    {
      "epoch": 4.400527009222661,
      "grad_norm": 2.7499680519104004,
      "learning_rate": 5.9947299077733865e-05,
      "loss": 3.2577,
      "step": 23380
    },
    {
      "epoch": 4.402409185017881,
      "grad_norm": 3.9470832347869873,
      "learning_rate": 5.9759081498211935e-05,
      "loss": 3.2105,
      "step": 23390
    },
    {
      "epoch": 4.4042913608131,
      "grad_norm": 2.9097394943237305,
      "learning_rate": 5.957086391869001e-05,
      "loss": 3.0114,
      "step": 23400
    },
    {
      "epoch": 4.4061735366083195,
      "grad_norm": 3.433927536010742,
      "learning_rate": 5.938264633916808e-05,
      "loss": 2.8237,
      "step": 23410
    },
    {
      "epoch": 4.408055712403539,
      "grad_norm": 3.0882952213287354,
      "learning_rate": 5.919442875964615e-05,
      "loss": 3.2526,
      "step": 23420
    },
    {
      "epoch": 4.409937888198757,
      "grad_norm": 4.191737651824951,
      "learning_rate": 5.900621118012423e-05,
      "loss": 3.1569,
      "step": 23430
    },
    {
      "epoch": 4.411820063993977,
      "grad_norm": 3.4990599155426025,
      "learning_rate": 5.88179936006023e-05,
      "loss": 3.2241,
      "step": 23440
    },
    {
      "epoch": 4.413702239789196,
      "grad_norm": 3.3133044242858887,
      "learning_rate": 5.8629776021080366e-05,
      "loss": 3.0933,
      "step": 23450
    },
    {
      "epoch": 4.415584415584416,
      "grad_norm": 3.126660108566284,
      "learning_rate": 5.844155844155844e-05,
      "loss": 3.3786,
      "step": 23460
    },
    {
      "epoch": 4.417466591379635,
      "grad_norm": 2.7241156101226807,
      "learning_rate": 5.825334086203651e-05,
      "loss": 3.2118,
      "step": 23470
    },
    {
      "epoch": 4.419348767174855,
      "grad_norm": 2.985119581222534,
      "learning_rate": 5.806512328251459e-05,
      "loss": 3.2246,
      "step": 23480
    },
    {
      "epoch": 4.421230942970073,
      "grad_norm": 2.953031301498413,
      "learning_rate": 5.787690570299266e-05,
      "loss": 3.2377,
      "step": 23490
    },
    {
      "epoch": 4.423113118765293,
      "grad_norm": 2.686232089996338,
      "learning_rate": 5.7688688123470735e-05,
      "loss": 3.2919,
      "step": 23500
    },
    {
      "epoch": 4.424995294560512,
      "grad_norm": 2.8897335529327393,
      "learning_rate": 5.750047054394881e-05,
      "loss": 3.4231,
      "step": 23510
    },
    {
      "epoch": 4.426877470355731,
      "grad_norm": 2.7605607509613037,
      "learning_rate": 5.731225296442688e-05,
      "loss": 3.3684,
      "step": 23520
    },
    {
      "epoch": 4.428759646150951,
      "grad_norm": 3.12101674079895,
      "learning_rate": 5.712403538490495e-05,
      "loss": 3.1358,
      "step": 23530
    },
    {
      "epoch": 4.430641821946169,
      "grad_norm": 3.256334066390991,
      "learning_rate": 5.693581780538303e-05,
      "loss": 3.0332,
      "step": 23540
    },
    {
      "epoch": 4.432523997741389,
      "grad_norm": 4.081264019012451,
      "learning_rate": 5.67476002258611e-05,
      "loss": 3.0865,
      "step": 23550
    },
    {
      "epoch": 4.434406173536608,
      "grad_norm": 3.149183511734009,
      "learning_rate": 5.6559382646339166e-05,
      "loss": 3.2729,
      "step": 23560
    },
    {
      "epoch": 4.436288349331828,
      "grad_norm": 7.6489105224609375,
      "learning_rate": 5.637116506681724e-05,
      "loss": 3.2411,
      "step": 23570
    },
    {
      "epoch": 4.438170525127047,
      "grad_norm": 2.61975359916687,
      "learning_rate": 5.618294748729531e-05,
      "loss": 3.3824,
      "step": 23580
    },
    {
      "epoch": 4.440052700922266,
      "grad_norm": 2.765589475631714,
      "learning_rate": 5.599472990777338e-05,
      "loss": 3.3724,
      "step": 23590
    },
    {
      "epoch": 4.441934876717485,
      "grad_norm": 3.076176881790161,
      "learning_rate": 5.580651232825146e-05,
      "loss": 3.197,
      "step": 23600
    },
    {
      "epoch": 4.4438170525127045,
      "grad_norm": 3.716524600982666,
      "learning_rate": 5.561829474872953e-05,
      "loss": 3.4739,
      "step": 23610
    },
    {
      "epoch": 4.445699228307924,
      "grad_norm": 3.2613918781280518,
      "learning_rate": 5.5430077169207605e-05,
      "loss": 3.2186,
      "step": 23620
    },
    {
      "epoch": 4.447581404103143,
      "grad_norm": 3.2381677627563477,
      "learning_rate": 5.524185958968568e-05,
      "loss": 3.1227,
      "step": 23630
    },
    {
      "epoch": 4.449463579898363,
      "grad_norm": 3.682084083557129,
      "learning_rate": 5.505364201016375e-05,
      "loss": 2.9952,
      "step": 23640
    },
    {
      "epoch": 4.451345755693581,
      "grad_norm": 3.6691043376922607,
      "learning_rate": 5.486542443064183e-05,
      "loss": 3.2097,
      "step": 23650
    },
    {
      "epoch": 4.453227931488801,
      "grad_norm": 2.951138973236084,
      "learning_rate": 5.46772068511199e-05,
      "loss": 3.339,
      "step": 23660
    },
    {
      "epoch": 4.45511010728402,
      "grad_norm": 2.816178321838379,
      "learning_rate": 5.4488989271597966e-05,
      "loss": 3.0707,
      "step": 23670
    },
    {
      "epoch": 4.45699228307924,
      "grad_norm": 2.796182870864868,
      "learning_rate": 5.430077169207604e-05,
      "loss": 3.2437,
      "step": 23680
    },
    {
      "epoch": 4.458874458874459,
      "grad_norm": 2.747709274291992,
      "learning_rate": 5.411255411255411e-05,
      "loss": 2.9044,
      "step": 23690
    },
    {
      "epoch": 4.4607566346696785,
      "grad_norm": 7.65395975112915,
      "learning_rate": 5.392433653303218e-05,
      "loss": 2.8958,
      "step": 23700
    },
    {
      "epoch": 4.462638810464897,
      "grad_norm": 2.6658778190612793,
      "learning_rate": 5.373611895351026e-05,
      "loss": 3.4358,
      "step": 23710
    },
    {
      "epoch": 4.4645209862601165,
      "grad_norm": 3.1606075763702393,
      "learning_rate": 5.354790137398833e-05,
      "loss": 3.2838,
      "step": 23720
    },
    {
      "epoch": 4.466403162055336,
      "grad_norm": 3.0979714393615723,
      "learning_rate": 5.3359683794466405e-05,
      "loss": 3.4613,
      "step": 23730
    },
    {
      "epoch": 4.468285337850555,
      "grad_norm": 3.3036980628967285,
      "learning_rate": 5.3171466214944474e-05,
      "loss": 3.0426,
      "step": 23740
    },
    {
      "epoch": 4.470167513645775,
      "grad_norm": 2.8867993354797363,
      "learning_rate": 5.298324863542255e-05,
      "loss": 2.9376,
      "step": 23750
    },
    {
      "epoch": 4.472049689440993,
      "grad_norm": 3.0002734661102295,
      "learning_rate": 5.279503105590063e-05,
      "loss": 3.0216,
      "step": 23760
    },
    {
      "epoch": 4.473931865236213,
      "grad_norm": 3.7616660594940186,
      "learning_rate": 5.26068134763787e-05,
      "loss": 3.338,
      "step": 23770
    },
    {
      "epoch": 4.475814041031432,
      "grad_norm": 2.7143394947052,
      "learning_rate": 5.2418595896856766e-05,
      "loss": 3.2587,
      "step": 23780
    },
    {
      "epoch": 4.477696216826652,
      "grad_norm": 3.7299182415008545,
      "learning_rate": 5.223037831733484e-05,
      "loss": 2.9254,
      "step": 23790
    },
    {
      "epoch": 4.479578392621871,
      "grad_norm": 2.715489149093628,
      "learning_rate": 5.204216073781291e-05,
      "loss": 3.1811,
      "step": 23800
    },
    {
      "epoch": 4.4814605684170905,
      "grad_norm": 3.3261804580688477,
      "learning_rate": 5.185394315829098e-05,
      "loss": 3.3265,
      "step": 23810
    },
    {
      "epoch": 4.483342744212309,
      "grad_norm": 2.696645736694336,
      "learning_rate": 5.166572557876906e-05,
      "loss": 3.1978,
      "step": 23820
    },
    {
      "epoch": 4.4852249200075285,
      "grad_norm": 2.7663300037384033,
      "learning_rate": 5.147750799924713e-05,
      "loss": 2.9375,
      "step": 23830
    },
    {
      "epoch": 4.487107095802748,
      "grad_norm": 3.4459638595581055,
      "learning_rate": 5.1289290419725205e-05,
      "loss": 3.4786,
      "step": 23840
    },
    {
      "epoch": 4.488989271597967,
      "grad_norm": 2.642287492752075,
      "learning_rate": 5.1101072840203274e-05,
      "loss": 3.0888,
      "step": 23850
    },
    {
      "epoch": 4.490871447393187,
      "grad_norm": 2.4454216957092285,
      "learning_rate": 5.0912855260681344e-05,
      "loss": 3.0148,
      "step": 23860
    },
    {
      "epoch": 4.492753623188406,
      "grad_norm": 2.8533213138580322,
      "learning_rate": 5.072463768115943e-05,
      "loss": 3.3603,
      "step": 23870
    },
    {
      "epoch": 4.494635798983625,
      "grad_norm": 2.8486480712890625,
      "learning_rate": 5.05364201016375e-05,
      "loss": 3.0267,
      "step": 23880
    },
    {
      "epoch": 4.496517974778844,
      "grad_norm": 3.5169730186462402,
      "learning_rate": 5.0348202522115566e-05,
      "loss": 2.9836,
      "step": 23890
    },
    {
      "epoch": 4.498400150574064,
      "grad_norm": 4.43604040145874,
      "learning_rate": 5.015998494259364e-05,
      "loss": 3.2031,
      "step": 23900
    },
    {
      "epoch": 4.500282326369283,
      "grad_norm": 2.625504493713379,
      "learning_rate": 4.997176736307171e-05,
      "loss": 3.1256,
      "step": 23910
    },
    {
      "epoch": 4.5021645021645025,
      "grad_norm": 2.9266598224639893,
      "learning_rate": 4.978354978354978e-05,
      "loss": 3.1863,
      "step": 23920
    },
    {
      "epoch": 4.504046677959721,
      "grad_norm": 4.946619987487793,
      "learning_rate": 4.959533220402786e-05,
      "loss": 3.4948,
      "step": 23930
    },
    {
      "epoch": 4.5059288537549405,
      "grad_norm": 3.0961945056915283,
      "learning_rate": 4.940711462450593e-05,
      "loss": 3.4399,
      "step": 23940
    },
    {
      "epoch": 4.50781102955016,
      "grad_norm": 2.682321548461914,
      "learning_rate": 4.9218897044984e-05,
      "loss": 3.3231,
      "step": 23950
    },
    {
      "epoch": 4.509693205345379,
      "grad_norm": 3.179645299911499,
      "learning_rate": 4.9030679465462074e-05,
      "loss": 3.2811,
      "step": 23960
    },
    {
      "epoch": 4.511575381140599,
      "grad_norm": 2.6911354064941406,
      "learning_rate": 4.8842461885940144e-05,
      "loss": 3.2388,
      "step": 23970
    },
    {
      "epoch": 4.513457556935818,
      "grad_norm": 3.0853166580200195,
      "learning_rate": 4.865424430641822e-05,
      "loss": 3.6197,
      "step": 23980
    },
    {
      "epoch": 4.515339732731037,
      "grad_norm": 2.753788471221924,
      "learning_rate": 4.84660267268963e-05,
      "loss": 3.111,
      "step": 23990
    },
    {
      "epoch": 4.517221908526256,
      "grad_norm": 2.9684507846832275,
      "learning_rate": 4.8277809147374366e-05,
      "loss": 3.0002,
      "step": 24000
    },
    {
      "epoch": 4.519104084321476,
      "grad_norm": 3.3698248863220215,
      "learning_rate": 4.808959156785244e-05,
      "loss": 3.3243,
      "step": 24010
    },
    {
      "epoch": 4.520986260116695,
      "grad_norm": 2.9527013301849365,
      "learning_rate": 4.790137398833051e-05,
      "loss": 3.1075,
      "step": 24020
    },
    {
      "epoch": 4.5228684359119145,
      "grad_norm": 3.412144899368286,
      "learning_rate": 4.771315640880858e-05,
      "loss": 3.189,
      "step": 24030
    },
    {
      "epoch": 4.524750611707134,
      "grad_norm": 2.9642202854156494,
      "learning_rate": 4.752493882928666e-05,
      "loss": 3.205,
      "step": 24040
    },
    {
      "epoch": 4.526632787502352,
      "grad_norm": 3.5863893032073975,
      "learning_rate": 4.733672124976473e-05,
      "loss": 3.2214,
      "step": 24050
    },
    {
      "epoch": 4.528514963297572,
      "grad_norm": 2.6433331966400146,
      "learning_rate": 4.71485036702428e-05,
      "loss": 3.347,
      "step": 24060
    },
    {
      "epoch": 4.530397139092791,
      "grad_norm": 2.9724504947662354,
      "learning_rate": 4.6960286090720874e-05,
      "loss": 3.021,
      "step": 24070
    },
    {
      "epoch": 4.532279314888011,
      "grad_norm": 3.0654139518737793,
      "learning_rate": 4.6772068511198944e-05,
      "loss": 3.4734,
      "step": 24080
    },
    {
      "epoch": 4.53416149068323,
      "grad_norm": 2.883718729019165,
      "learning_rate": 4.658385093167702e-05,
      "loss": 3.0313,
      "step": 24090
    },
    {
      "epoch": 4.536043666478449,
      "grad_norm": 3.460390090942383,
      "learning_rate": 4.639563335215509e-05,
      "loss": 3.4964,
      "step": 24100
    },
    {
      "epoch": 4.537925842273668,
      "grad_norm": 2.498680830001831,
      "learning_rate": 4.620741577263316e-05,
      "loss": 2.9641,
      "step": 24110
    },
    {
      "epoch": 4.539808018068888,
      "grad_norm": 4.274486064910889,
      "learning_rate": 4.601919819311124e-05,
      "loss": 3.1713,
      "step": 24120
    },
    {
      "epoch": 4.541690193864107,
      "grad_norm": 3.244508743286133,
      "learning_rate": 4.583098061358931e-05,
      "loss": 3.2632,
      "step": 24130
    },
    {
      "epoch": 4.543572369659326,
      "grad_norm": 2.921966791152954,
      "learning_rate": 4.564276303406738e-05,
      "loss": 3.0531,
      "step": 24140
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 3.98917555809021,
      "learning_rate": 4.545454545454546e-05,
      "loss": 3.3034,
      "step": 24150
    },
    {
      "epoch": 4.547336721249764,
      "grad_norm": 3.7164669036865234,
      "learning_rate": 4.526632787502353e-05,
      "loss": 3.162,
      "step": 24160
    },
    {
      "epoch": 4.549218897044984,
      "grad_norm": 3.392164945602417,
      "learning_rate": 4.50781102955016e-05,
      "loss": 3.0602,
      "step": 24170
    },
    {
      "epoch": 4.551101072840203,
      "grad_norm": 3.1076338291168213,
      "learning_rate": 4.4889892715979674e-05,
      "loss": 3.3768,
      "step": 24180
    },
    {
      "epoch": 4.552983248635423,
      "grad_norm": 2.799534797668457,
      "learning_rate": 4.4701675136457744e-05,
      "loss": 2.9775,
      "step": 24190
    },
    {
      "epoch": 4.554865424430642,
      "grad_norm": 2.9202513694763184,
      "learning_rate": 4.451345755693582e-05,
      "loss": 3.2604,
      "step": 24200
    },
    {
      "epoch": 4.556747600225862,
      "grad_norm": 4.149763107299805,
      "learning_rate": 4.432523997741389e-05,
      "loss": 3.0419,
      "step": 24210
    },
    {
      "epoch": 4.55862977602108,
      "grad_norm": 2.8485655784606934,
      "learning_rate": 4.413702239789196e-05,
      "loss": 3.0743,
      "step": 24220
    },
    {
      "epoch": 4.5605119518162995,
      "grad_norm": 2.8361072540283203,
      "learning_rate": 4.3948804818370036e-05,
      "loss": 3.3415,
      "step": 24230
    },
    {
      "epoch": 4.562394127611519,
      "grad_norm": 3.362912178039551,
      "learning_rate": 4.376058723884811e-05,
      "loss": 3.1703,
      "step": 24240
    },
    {
      "epoch": 4.564276303406738,
      "grad_norm": 2.6403450965881348,
      "learning_rate": 4.357236965932618e-05,
      "loss": 2.8701,
      "step": 24250
    },
    {
      "epoch": 4.566158479201958,
      "grad_norm": 2.5263137817382812,
      "learning_rate": 4.338415207980426e-05,
      "loss": 2.8825,
      "step": 24260
    },
    {
      "epoch": 4.568040654997176,
      "grad_norm": 2.7215659618377686,
      "learning_rate": 4.319593450028233e-05,
      "loss": 2.9142,
      "step": 24270
    },
    {
      "epoch": 4.569922830792396,
      "grad_norm": 3.132044553756714,
      "learning_rate": 4.30077169207604e-05,
      "loss": 3.4087,
      "step": 24280
    },
    {
      "epoch": 4.571805006587615,
      "grad_norm": 3.3651986122131348,
      "learning_rate": 4.2819499341238474e-05,
      "loss": 3.3137,
      "step": 24290
    },
    {
      "epoch": 4.573687182382835,
      "grad_norm": 3.7178475856781006,
      "learning_rate": 4.2631281761716544e-05,
      "loss": 2.882,
      "step": 24300
    },
    {
      "epoch": 4.575569358178054,
      "grad_norm": 5.011250972747803,
      "learning_rate": 4.244306418219462e-05,
      "loss": 3.4289,
      "step": 24310
    },
    {
      "epoch": 4.577451533973273,
      "grad_norm": 3.3479485511779785,
      "learning_rate": 4.225484660267269e-05,
      "loss": 3.28,
      "step": 24320
    },
    {
      "epoch": 4.579333709768492,
      "grad_norm": 2.715578079223633,
      "learning_rate": 4.206662902315076e-05,
      "loss": 2.9993,
      "step": 24330
    },
    {
      "epoch": 4.5812158855637115,
      "grad_norm": 3.1181604862213135,
      "learning_rate": 4.1878411443628836e-05,
      "loss": 2.9622,
      "step": 24340
    },
    {
      "epoch": 4.583098061358931,
      "grad_norm": 4.305251121520996,
      "learning_rate": 4.1690193864106906e-05,
      "loss": 3.034,
      "step": 24350
    },
    {
      "epoch": 4.58498023715415,
      "grad_norm": 3.093553304672241,
      "learning_rate": 4.150197628458498e-05,
      "loss": 3.6412,
      "step": 24360
    },
    {
      "epoch": 4.58686241294937,
      "grad_norm": 4.383254051208496,
      "learning_rate": 4.131375870506306e-05,
      "loss": 3.6157,
      "step": 24370
    },
    {
      "epoch": 4.588744588744589,
      "grad_norm": 3.1755826473236084,
      "learning_rate": 4.112554112554113e-05,
      "loss": 3.4622,
      "step": 24380
    },
    {
      "epoch": 4.590626764539808,
      "grad_norm": 2.9450700283050537,
      "learning_rate": 4.09373235460192e-05,
      "loss": 3.3494,
      "step": 24390
    },
    {
      "epoch": 4.592508940335027,
      "grad_norm": 3.380802869796753,
      "learning_rate": 4.0749105966497274e-05,
      "loss": 3.2481,
      "step": 24400
    },
    {
      "epoch": 4.594391116130247,
      "grad_norm": 2.74828839302063,
      "learning_rate": 4.0560888386975344e-05,
      "loss": 3.2937,
      "step": 24410
    },
    {
      "epoch": 4.596273291925466,
      "grad_norm": 2.6550493240356445,
      "learning_rate": 4.0372670807453414e-05,
      "loss": 2.9292,
      "step": 24420
    },
    {
      "epoch": 4.5981554677206855,
      "grad_norm": 2.6913087368011475,
      "learning_rate": 4.018445322793149e-05,
      "loss": 3.3403,
      "step": 24430
    },
    {
      "epoch": 4.600037643515904,
      "grad_norm": 3.7004897594451904,
      "learning_rate": 3.999623564840956e-05,
      "loss": 3.1304,
      "step": 24440
    },
    {
      "epoch": 4.6019198193111235,
      "grad_norm": 3.044382095336914,
      "learning_rate": 3.9808018068887636e-05,
      "loss": 2.8982,
      "step": 24450
    },
    {
      "epoch": 4.603801995106343,
      "grad_norm": 2.8291525840759277,
      "learning_rate": 3.9619800489365706e-05,
      "loss": 3.1667,
      "step": 24460
    },
    {
      "epoch": 4.605684170901562,
      "grad_norm": 2.786212205886841,
      "learning_rate": 3.9431582909843775e-05,
      "loss": 3.0917,
      "step": 24470
    },
    {
      "epoch": 4.607566346696782,
      "grad_norm": 2.716911554336548,
      "learning_rate": 3.924336533032186e-05,
      "loss": 3.5009,
      "step": 24480
    },
    {
      "epoch": 4.609448522492,
      "grad_norm": 2.9290614128112793,
      "learning_rate": 3.905514775079993e-05,
      "loss": 3.2019,
      "step": 24490
    },
    {
      "epoch": 4.61133069828722,
      "grad_norm": 2.7946932315826416,
      "learning_rate": 3.8866930171278e-05,
      "loss": 3.3843,
      "step": 24500
    },
    {
      "epoch": 4.613212874082439,
      "grad_norm": 3.1367475986480713,
      "learning_rate": 3.8678712591756074e-05,
      "loss": 3.1811,
      "step": 24510
    },
    {
      "epoch": 4.615095049877659,
      "grad_norm": 2.5652503967285156,
      "learning_rate": 3.8490495012234144e-05,
      "loss": 3.0921,
      "step": 24520
    },
    {
      "epoch": 4.616977225672878,
      "grad_norm": 2.903272867202759,
      "learning_rate": 3.8302277432712214e-05,
      "loss": 3.1698,
      "step": 24530
    },
    {
      "epoch": 4.6188594014680975,
      "grad_norm": 3.0952274799346924,
      "learning_rate": 3.811405985319029e-05,
      "loss": 2.991,
      "step": 24540
    },
    {
      "epoch": 4.620741577263316,
      "grad_norm": 2.6256113052368164,
      "learning_rate": 3.792584227366836e-05,
      "loss": 3.0366,
      "step": 24550
    },
    {
      "epoch": 4.6226237530585355,
      "grad_norm": 2.872602939605713,
      "learning_rate": 3.7737624694146436e-05,
      "loss": 3.065,
      "step": 24560
    },
    {
      "epoch": 4.624505928853755,
      "grad_norm": 3.981386184692383,
      "learning_rate": 3.7549407114624506e-05,
      "loss": 3.0991,
      "step": 24570
    },
    {
      "epoch": 4.626388104648974,
      "grad_norm": 6.0364251136779785,
      "learning_rate": 3.7361189535102575e-05,
      "loss": 3.1146,
      "step": 24580
    },
    {
      "epoch": 4.628270280444194,
      "grad_norm": 3.1531736850738525,
      "learning_rate": 3.717297195558065e-05,
      "loss": 3.3461,
      "step": 24590
    },
    {
      "epoch": 4.630152456239413,
      "grad_norm": 3.7676281929016113,
      "learning_rate": 3.698475437605872e-05,
      "loss": 3.0683,
      "step": 24600
    },
    {
      "epoch": 4.632034632034632,
      "grad_norm": 2.6802709102630615,
      "learning_rate": 3.67965367965368e-05,
      "loss": 3.2123,
      "step": 24610
    },
    {
      "epoch": 4.633916807829851,
      "grad_norm": 3.7825965881347656,
      "learning_rate": 3.6608319217014874e-05,
      "loss": 3.3615,
      "step": 24620
    },
    {
      "epoch": 4.635798983625071,
      "grad_norm": 2.9915659427642822,
      "learning_rate": 3.6420101637492944e-05,
      "loss": 3.3229,
      "step": 24630
    },
    {
      "epoch": 4.63768115942029,
      "grad_norm": 3.3726351261138916,
      "learning_rate": 3.6231884057971014e-05,
      "loss": 3.3137,
      "step": 24640
    },
    {
      "epoch": 4.6395633352155095,
      "grad_norm": 2.611403465270996,
      "learning_rate": 3.604366647844909e-05,
      "loss": 3.2665,
      "step": 24650
    },
    {
      "epoch": 4.641445511010728,
      "grad_norm": 4.3528523445129395,
      "learning_rate": 3.585544889892716e-05,
      "loss": 3.1482,
      "step": 24660
    },
    {
      "epoch": 4.643327686805947,
      "grad_norm": 2.827160596847534,
      "learning_rate": 3.5667231319405236e-05,
      "loss": 3.5753,
      "step": 24670
    },
    {
      "epoch": 4.645209862601167,
      "grad_norm": 2.940286159515381,
      "learning_rate": 3.5479013739883306e-05,
      "loss": 3.2837,
      "step": 24680
    },
    {
      "epoch": 4.647092038396386,
      "grad_norm": 2.595956325531006,
      "learning_rate": 3.5290796160361375e-05,
      "loss": 3.1817,
      "step": 24690
    },
    {
      "epoch": 4.648974214191606,
      "grad_norm": 3.200382947921753,
      "learning_rate": 3.510257858083945e-05,
      "loss": 3.4257,
      "step": 24700
    },
    {
      "epoch": 4.650856389986825,
      "grad_norm": 2.6763854026794434,
      "learning_rate": 3.491436100131752e-05,
      "loss": 2.9655,
      "step": 24710
    },
    {
      "epoch": 4.652738565782044,
      "grad_norm": 3.2725934982299805,
      "learning_rate": 3.472614342179559e-05,
      "loss": 3.2608,
      "step": 24720
    },
    {
      "epoch": 4.654620741577263,
      "grad_norm": 2.985708236694336,
      "learning_rate": 3.4537925842273674e-05,
      "loss": 3.1168,
      "step": 24730
    },
    {
      "epoch": 4.656502917372483,
      "grad_norm": 3.0978457927703857,
      "learning_rate": 3.4349708262751744e-05,
      "loss": 3.0778,
      "step": 24740
    },
    {
      "epoch": 4.658385093167702,
      "grad_norm": 8.759063720703125,
      "learning_rate": 3.4161490683229814e-05,
      "loss": 3.246,
      "step": 24750
    },
    {
      "epoch": 4.660267268962921,
      "grad_norm": 3.0920560359954834,
      "learning_rate": 3.397327310370789e-05,
      "loss": 3.0907,
      "step": 24760
    },
    {
      "epoch": 4.662149444758141,
      "grad_norm": 2.807457685470581,
      "learning_rate": 3.378505552418596e-05,
      "loss": 3.2512,
      "step": 24770
    },
    {
      "epoch": 4.664031620553359,
      "grad_norm": 3.2192564010620117,
      "learning_rate": 3.359683794466403e-05,
      "loss": 3.4434,
      "step": 24780
    },
    {
      "epoch": 4.665913796348579,
      "grad_norm": 3.1085925102233887,
      "learning_rate": 3.3408620365142106e-05,
      "loss": 3.375,
      "step": 24790
    },
    {
      "epoch": 4.667795972143798,
      "grad_norm": 3.6151530742645264,
      "learning_rate": 3.3220402785620175e-05,
      "loss": 3.2606,
      "step": 24800
    },
    {
      "epoch": 4.669678147939018,
      "grad_norm": 2.7845237255096436,
      "learning_rate": 3.303218520609825e-05,
      "loss": 3.1624,
      "step": 24810
    },
    {
      "epoch": 4.671560323734237,
      "grad_norm": 2.3765504360198975,
      "learning_rate": 3.284396762657632e-05,
      "loss": 3.4659,
      "step": 24820
    },
    {
      "epoch": 4.673442499529456,
      "grad_norm": 3.301711082458496,
      "learning_rate": 3.265575004705439e-05,
      "loss": 3.5859,
      "step": 24830
    },
    {
      "epoch": 4.675324675324675,
      "grad_norm": 2.6286189556121826,
      "learning_rate": 3.246753246753247e-05,
      "loss": 3.0791,
      "step": 24840
    },
    {
      "epoch": 4.6772068511198945,
      "grad_norm": 2.879110813140869,
      "learning_rate": 3.2279314888010544e-05,
      "loss": 3.1536,
      "step": 24850
    },
    {
      "epoch": 4.679089026915114,
      "grad_norm": 3.2365987300872803,
      "learning_rate": 3.2091097308488614e-05,
      "loss": 3.1614,
      "step": 24860
    },
    {
      "epoch": 4.680971202710333,
      "grad_norm": 3.932792901992798,
      "learning_rate": 3.190287972896669e-05,
      "loss": 3.0712,
      "step": 24870
    },
    {
      "epoch": 4.682853378505552,
      "grad_norm": 3.3613767623901367,
      "learning_rate": 3.171466214944476e-05,
      "loss": 3.4802,
      "step": 24880
    },
    {
      "epoch": 4.684735554300771,
      "grad_norm": 2.611473798751831,
      "learning_rate": 3.152644456992283e-05,
      "loss": 3.2437,
      "step": 24890
    },
    {
      "epoch": 4.686617730095991,
      "grad_norm": 3.1896893978118896,
      "learning_rate": 3.1338226990400906e-05,
      "loss": 3.2623,
      "step": 24900
    },
    {
      "epoch": 4.68849990589121,
      "grad_norm": 2.306182861328125,
      "learning_rate": 3.1150009410878975e-05,
      "loss": 2.9952,
      "step": 24910
    },
    {
      "epoch": 4.69038208168643,
      "grad_norm": 2.9280362129211426,
      "learning_rate": 3.096179183135705e-05,
      "loss": 3.3888,
      "step": 24920
    },
    {
      "epoch": 4.692264257481649,
      "grad_norm": 3.746704578399658,
      "learning_rate": 3.077357425183512e-05,
      "loss": 2.9443,
      "step": 24930
    },
    {
      "epoch": 4.6941464332768685,
      "grad_norm": 3.5688564777374268,
      "learning_rate": 3.05853566723132e-05,
      "loss": 3.258,
      "step": 24940
    },
    {
      "epoch": 4.696028609072087,
      "grad_norm": 4.903964042663574,
      "learning_rate": 3.0397139092791268e-05,
      "loss": 3.4195,
      "step": 24950
    },
    {
      "epoch": 4.6979107848673065,
      "grad_norm": 3.729482412338257,
      "learning_rate": 3.020892151326934e-05,
      "loss": 3.3093,
      "step": 24960
    },
    {
      "epoch": 4.699792960662526,
      "grad_norm": 3.215060234069824,
      "learning_rate": 3.0020703933747414e-05,
      "loss": 3.475,
      "step": 24970
    },
    {
      "epoch": 4.701675136457745,
      "grad_norm": 2.280484914779663,
      "learning_rate": 2.9832486354225483e-05,
      "loss": 3.1306,
      "step": 24980
    },
    {
      "epoch": 4.703557312252965,
      "grad_norm": 3.377382278442383,
      "learning_rate": 2.9644268774703556e-05,
      "loss": 3.2126,
      "step": 24990
    },
    {
      "epoch": 4.705439488048183,
      "grad_norm": 3.27429461479187,
      "learning_rate": 2.945605119518163e-05,
      "loss": 3.281,
      "step": 25000
    },
    {
      "epoch": 4.707321663843403,
      "grad_norm": 2.8424630165100098,
      "learning_rate": 2.9267833615659706e-05,
      "loss": 3.2675,
      "step": 25010
    },
    {
      "epoch": 4.709203839638622,
      "grad_norm": 3.1172735691070557,
      "learning_rate": 2.9079616036137775e-05,
      "loss": 3.4589,
      "step": 25020
    },
    {
      "epoch": 4.711086015433842,
      "grad_norm": 2.8301455974578857,
      "learning_rate": 2.889139845661585e-05,
      "loss": 3.3261,
      "step": 25030
    },
    {
      "epoch": 4.712968191229061,
      "grad_norm": 2.538022041320801,
      "learning_rate": 2.870318087709392e-05,
      "loss": 3.0696,
      "step": 25040
    },
    {
      "epoch": 4.71485036702428,
      "grad_norm": 2.5753090381622314,
      "learning_rate": 2.851496329757199e-05,
      "loss": 2.9831,
      "step": 25050
    },
    {
      "epoch": 4.716732542819499,
      "grad_norm": 3.2386507987976074,
      "learning_rate": 2.8326745718050064e-05,
      "loss": 3.04,
      "step": 25060
    },
    {
      "epoch": 4.7186147186147185,
      "grad_norm": 2.7614169120788574,
      "learning_rate": 2.813852813852814e-05,
      "loss": 3.292,
      "step": 25070
    },
    {
      "epoch": 4.720496894409938,
      "grad_norm": 3.875685214996338,
      "learning_rate": 2.7950310559006214e-05,
      "loss": 3.3236,
      "step": 25080
    },
    {
      "epoch": 4.722379070205157,
      "grad_norm": 3.0858395099639893,
      "learning_rate": 2.7762092979484283e-05,
      "loss": 3.2861,
      "step": 25090
    },
    {
      "epoch": 4.724261246000377,
      "grad_norm": 4.012641906738281,
      "learning_rate": 2.7573875399962356e-05,
      "loss": 3.3092,
      "step": 25100
    },
    {
      "epoch": 4.726143421795595,
      "grad_norm": 2.8968312740325928,
      "learning_rate": 2.738565782044043e-05,
      "loss": 3.0026,
      "step": 25110
    },
    {
      "epoch": 4.728025597590815,
      "grad_norm": 4.2459001541137695,
      "learning_rate": 2.7197440240918502e-05,
      "loss": 2.9701,
      "step": 25120
    },
    {
      "epoch": 4.729907773386034,
      "grad_norm": 3.4018383026123047,
      "learning_rate": 2.7009222661396575e-05,
      "loss": 3.1652,
      "step": 25130
    },
    {
      "epoch": 4.731789949181254,
      "grad_norm": 3.798372268676758,
      "learning_rate": 2.682100508187465e-05,
      "loss": 3.2211,
      "step": 25140
    },
    {
      "epoch": 4.733672124976473,
      "grad_norm": 3.093381643295288,
      "learning_rate": 2.663278750235272e-05,
      "loss": 3.2467,
      "step": 25150
    },
    {
      "epoch": 4.7355543007716925,
      "grad_norm": 2.5847713947296143,
      "learning_rate": 2.644456992283079e-05,
      "loss": 2.995,
      "step": 25160
    },
    {
      "epoch": 4.737436476566911,
      "grad_norm": 3.9761364459991455,
      "learning_rate": 2.6256352343308864e-05,
      "loss": 2.8856,
      "step": 25170
    },
    {
      "epoch": 4.7393186523621305,
      "grad_norm": 2.505748987197876,
      "learning_rate": 2.6068134763786937e-05,
      "loss": 3.0557,
      "step": 25180
    },
    {
      "epoch": 4.74120082815735,
      "grad_norm": 3.296593189239502,
      "learning_rate": 2.5879917184265014e-05,
      "loss": 3.0296,
      "step": 25190
    },
    {
      "epoch": 4.743083003952569,
      "grad_norm": 3.624117136001587,
      "learning_rate": 2.5691699604743083e-05,
      "loss": 2.8997,
      "step": 25200
    },
    {
      "epoch": 4.744965179747789,
      "grad_norm": 3.9669125080108643,
      "learning_rate": 2.5503482025221156e-05,
      "loss": 2.946,
      "step": 25210
    },
    {
      "epoch": 4.746847355543007,
      "grad_norm": 2.9287898540496826,
      "learning_rate": 2.531526444569923e-05,
      "loss": 3.2531,
      "step": 25220
    },
    {
      "epoch": 4.748729531338227,
      "grad_norm": 3.1249217987060547,
      "learning_rate": 2.5127046866177302e-05,
      "loss": 3.2925,
      "step": 25230
    },
    {
      "epoch": 4.750611707133446,
      "grad_norm": 3.173673152923584,
      "learning_rate": 2.4938829286655372e-05,
      "loss": 3.2142,
      "step": 25240
    },
    {
      "epoch": 4.752493882928666,
      "grad_norm": 2.8763136863708496,
      "learning_rate": 2.475061170713345e-05,
      "loss": 3.1028,
      "step": 25250
    },
    {
      "epoch": 4.754376058723885,
      "grad_norm": 4.05729866027832,
      "learning_rate": 2.456239412761152e-05,
      "loss": 2.7233,
      "step": 25260
    },
    {
      "epoch": 4.7562582345191045,
      "grad_norm": 3.805598497390747,
      "learning_rate": 2.437417654808959e-05,
      "loss": 3.1988,
      "step": 25270
    },
    {
      "epoch": 4.758140410314323,
      "grad_norm": 3.6921346187591553,
      "learning_rate": 2.4185958968567664e-05,
      "loss": 2.8693,
      "step": 25280
    },
    {
      "epoch": 4.760022586109542,
      "grad_norm": 3.240833044052124,
      "learning_rate": 2.3997741389045737e-05,
      "loss": 3.0211,
      "step": 25290
    },
    {
      "epoch": 4.761904761904762,
      "grad_norm": 3.152642250061035,
      "learning_rate": 2.380952380952381e-05,
      "loss": 3.0866,
      "step": 25300
    },
    {
      "epoch": 4.763786937699981,
      "grad_norm": 3.251298427581787,
      "learning_rate": 2.3621306230001883e-05,
      "loss": 3.1911,
      "step": 25310
    },
    {
      "epoch": 4.765669113495201,
      "grad_norm": 2.7727744579315186,
      "learning_rate": 2.3433088650479956e-05,
      "loss": 3.2782,
      "step": 25320
    },
    {
      "epoch": 4.76755128929042,
      "grad_norm": 2.8768370151519775,
      "learning_rate": 2.324487107095803e-05,
      "loss": 2.9626,
      "step": 25330
    },
    {
      "epoch": 4.769433465085639,
      "grad_norm": 3.1602566242218018,
      "learning_rate": 2.30566534914361e-05,
      "loss": 3.0035,
      "step": 25340
    },
    {
      "epoch": 4.771315640880858,
      "grad_norm": 4.934927940368652,
      "learning_rate": 2.2868435911914172e-05,
      "loss": 2.9412,
      "step": 25350
    },
    {
      "epoch": 4.773197816676078,
      "grad_norm": 3.632086992263794,
      "learning_rate": 2.2680218332392245e-05,
      "loss": 3.4639,
      "step": 25360
    },
    {
      "epoch": 4.775079992471297,
      "grad_norm": 3.2546494007110596,
      "learning_rate": 2.249200075287032e-05,
      "loss": 2.9484,
      "step": 25370
    },
    {
      "epoch": 4.776962168266516,
      "grad_norm": 4.427943229675293,
      "learning_rate": 2.230378317334839e-05,
      "loss": 3.2541,
      "step": 25380
    },
    {
      "epoch": 4.778844344061735,
      "grad_norm": 2.785226583480835,
      "learning_rate": 2.2115565593826464e-05,
      "loss": 2.9992,
      "step": 25390
    },
    {
      "epoch": 4.780726519856954,
      "grad_norm": 2.93110990524292,
      "learning_rate": 2.1927348014304537e-05,
      "loss": 3.2238,
      "step": 25400
    },
    {
      "epoch": 4.782608695652174,
      "grad_norm": 2.6001505851745605,
      "learning_rate": 2.173913043478261e-05,
      "loss": 3.1899,
      "step": 25410
    },
    {
      "epoch": 4.784490871447393,
      "grad_norm": 2.8628385066986084,
      "learning_rate": 2.155091285526068e-05,
      "loss": 3.0132,
      "step": 25420
    },
    {
      "epoch": 4.786373047242613,
      "grad_norm": 2.6850054264068604,
      "learning_rate": 2.1362695275738753e-05,
      "loss": 3.4278,
      "step": 25430
    },
    {
      "epoch": 4.788255223037831,
      "grad_norm": 2.7586615085601807,
      "learning_rate": 2.117447769621683e-05,
      "loss": 3.428,
      "step": 25440
    },
    {
      "epoch": 4.790137398833051,
      "grad_norm": 3.3150601387023926,
      "learning_rate": 2.09862601166949e-05,
      "loss": 3.1502,
      "step": 25450
    },
    {
      "epoch": 4.79201957462827,
      "grad_norm": 3.676438093185425,
      "learning_rate": 2.0798042537172972e-05,
      "loss": 3.3348,
      "step": 25460
    },
    {
      "epoch": 4.7939017504234895,
      "grad_norm": 3.2945897579193115,
      "learning_rate": 2.0609824957651045e-05,
      "loss": 3.1048,
      "step": 25470
    },
    {
      "epoch": 4.795783926218709,
      "grad_norm": 3.1217949390411377,
      "learning_rate": 2.0421607378129118e-05,
      "loss": 3.0479,
      "step": 25480
    },
    {
      "epoch": 4.797666102013928,
      "grad_norm": 2.6365010738372803,
      "learning_rate": 2.0233389798607188e-05,
      "loss": 2.9137,
      "step": 25490
    },
    {
      "epoch": 4.799548277809148,
      "grad_norm": 3.463871955871582,
      "learning_rate": 2.0045172219085264e-05,
      "loss": 3.129,
      "step": 25500
    },
    {
      "epoch": 4.801430453604366,
      "grad_norm": 3.1459686756134033,
      "learning_rate": 1.9856954639563337e-05,
      "loss": 3.3096,
      "step": 25510
    },
    {
      "epoch": 4.803312629399586,
      "grad_norm": 2.775165557861328,
      "learning_rate": 1.9668737060041407e-05,
      "loss": 3.2217,
      "step": 25520
    },
    {
      "epoch": 4.805194805194805,
      "grad_norm": 3.2702062129974365,
      "learning_rate": 1.948051948051948e-05,
      "loss": 2.9853,
      "step": 25530
    },
    {
      "epoch": 4.807076980990025,
      "grad_norm": 2.980114698410034,
      "learning_rate": 1.9292301900997553e-05,
      "loss": 2.9909,
      "step": 25540
    },
    {
      "epoch": 4.808959156785244,
      "grad_norm": 3.2365455627441406,
      "learning_rate": 1.9104084321475626e-05,
      "loss": 2.9872,
      "step": 25550
    },
    {
      "epoch": 4.810841332580463,
      "grad_norm": 2.9393603801727295,
      "learning_rate": 1.89158667419537e-05,
      "loss": 3.4253,
      "step": 25560
    },
    {
      "epoch": 4.812723508375682,
      "grad_norm": 2.5274322032928467,
      "learning_rate": 1.8727649162431772e-05,
      "loss": 2.8669,
      "step": 25570
    },
    {
      "epoch": 4.8146056841709015,
      "grad_norm": 3.172722816467285,
      "learning_rate": 1.8539431582909845e-05,
      "loss": 2.9687,
      "step": 25580
    },
    {
      "epoch": 4.816487859966121,
      "grad_norm": 3.130019426345825,
      "learning_rate": 1.8351214003387918e-05,
      "loss": 3.0413,
      "step": 25590
    },
    {
      "epoch": 4.81837003576134,
      "grad_norm": 3.0461792945861816,
      "learning_rate": 1.8162996423865988e-05,
      "loss": 3.2432,
      "step": 25600
    },
    {
      "epoch": 4.820252211556559,
      "grad_norm": 2.9114632606506348,
      "learning_rate": 1.797477884434406e-05,
      "loss": 3.1182,
      "step": 25610
    },
    {
      "epoch": 4.822134387351778,
      "grad_norm": 3.019514799118042,
      "learning_rate": 1.7786561264822137e-05,
      "loss": 2.9704,
      "step": 25620
    },
    {
      "epoch": 4.824016563146998,
      "grad_norm": 3.043949604034424,
      "learning_rate": 1.7598343685300207e-05,
      "loss": 3.5735,
      "step": 25630
    },
    {
      "epoch": 4.825898738942217,
      "grad_norm": 2.620448589324951,
      "learning_rate": 1.741012610577828e-05,
      "loss": 2.9745,
      "step": 25640
    },
    {
      "epoch": 4.827780914737437,
      "grad_norm": 2.6471874713897705,
      "learning_rate": 1.7221908526256353e-05,
      "loss": 2.6752,
      "step": 25650
    },
    {
      "epoch": 4.829663090532656,
      "grad_norm": 2.99849271774292,
      "learning_rate": 1.7033690946734426e-05,
      "loss": 3.3014,
      "step": 25660
    },
    {
      "epoch": 4.8315452663278755,
      "grad_norm": 3.3653836250305176,
      "learning_rate": 1.6845473367212496e-05,
      "loss": 2.8202,
      "step": 25670
    },
    {
      "epoch": 4.833427442123094,
      "grad_norm": 3.4769065380096436,
      "learning_rate": 1.6657255787690572e-05,
      "loss": 3.2475,
      "step": 25680
    },
    {
      "epoch": 4.8353096179183135,
      "grad_norm": 3.774672508239746,
      "learning_rate": 1.6469038208168645e-05,
      "loss": 3.2216,
      "step": 25690
    },
    {
      "epoch": 4.837191793713533,
      "grad_norm": 4.837364196777344,
      "learning_rate": 1.6280820628646715e-05,
      "loss": 3.3699,
      "step": 25700
    },
    {
      "epoch": 4.839073969508752,
      "grad_norm": 2.9457051753997803,
      "learning_rate": 1.6092603049124788e-05,
      "loss": 2.9287,
      "step": 25710
    },
    {
      "epoch": 4.840956145303972,
      "grad_norm": 2.990029811859131,
      "learning_rate": 1.590438546960286e-05,
      "loss": 3.0793,
      "step": 25720
    },
    {
      "epoch": 4.84283832109919,
      "grad_norm": 3.9842264652252197,
      "learning_rate": 1.5716167890080934e-05,
      "loss": 3.043,
      "step": 25730
    },
    {
      "epoch": 4.84472049689441,
      "grad_norm": 4.764008045196533,
      "learning_rate": 1.5527950310559007e-05,
      "loss": 2.9053,
      "step": 25740
    },
    {
      "epoch": 4.846602672689629,
      "grad_norm": 3.5892012119293213,
      "learning_rate": 1.533973273103708e-05,
      "loss": 3.1736,
      "step": 25750
    },
    {
      "epoch": 4.848484848484849,
      "grad_norm": 3.524770498275757,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 3.1188,
      "step": 25760
    },
    {
      "epoch": 4.850367024280068,
      "grad_norm": 3.6661603450775146,
      "learning_rate": 1.4963297571993224e-05,
      "loss": 3.3539,
      "step": 25770
    },
    {
      "epoch": 4.852249200075287,
      "grad_norm": 2.802971124649048,
      "learning_rate": 1.4775079992471297e-05,
      "loss": 3.0607,
      "step": 25780
    },
    {
      "epoch": 4.854131375870506,
      "grad_norm": 2.968571662902832,
      "learning_rate": 1.458686241294937e-05,
      "loss": 3.0637,
      "step": 25790
    },
    {
      "epoch": 4.8560135516657255,
      "grad_norm": 4.406318187713623,
      "learning_rate": 1.4398644833427442e-05,
      "loss": 3.3069,
      "step": 25800
    },
    {
      "epoch": 4.857895727460945,
      "grad_norm": 2.874358654022217,
      "learning_rate": 1.4210427253905515e-05,
      "loss": 3.045,
      "step": 25810
    },
    {
      "epoch": 4.859777903256164,
      "grad_norm": 3.047579526901245,
      "learning_rate": 1.4022209674383588e-05,
      "loss": 2.9849,
      "step": 25820
    },
    {
      "epoch": 4.861660079051384,
      "grad_norm": 2.9793665409088135,
      "learning_rate": 1.383399209486166e-05,
      "loss": 3.2543,
      "step": 25830
    },
    {
      "epoch": 4.863542254846602,
      "grad_norm": 3.1359031200408936,
      "learning_rate": 1.3645774515339732e-05,
      "loss": 3.0561,
      "step": 25840
    },
    {
      "epoch": 4.865424430641822,
      "grad_norm": 3.4481136798858643,
      "learning_rate": 1.3457556935817807e-05,
      "loss": 3.231,
      "step": 25850
    },
    {
      "epoch": 4.867306606437041,
      "grad_norm": 2.9049251079559326,
      "learning_rate": 1.3269339356295878e-05,
      "loss": 3.3585,
      "step": 25860
    },
    {
      "epoch": 4.869188782232261,
      "grad_norm": 3.000098943710327,
      "learning_rate": 1.3081121776773951e-05,
      "loss": 3.2807,
      "step": 25870
    },
    {
      "epoch": 4.87107095802748,
      "grad_norm": 3.4274730682373047,
      "learning_rate": 1.2892904197252024e-05,
      "loss": 3.1417,
      "step": 25880
    },
    {
      "epoch": 4.8729531338226995,
      "grad_norm": 3.498537063598633,
      "learning_rate": 1.2704686617730096e-05,
      "loss": 3.2978,
      "step": 25890
    },
    {
      "epoch": 4.874835309617918,
      "grad_norm": 2.6304216384887695,
      "learning_rate": 1.2516469038208169e-05,
      "loss": 3.3949,
      "step": 25900
    },
    {
      "epoch": 4.876717485413137,
      "grad_norm": 3.301097869873047,
      "learning_rate": 1.2328251458686242e-05,
      "loss": 2.9152,
      "step": 25910
    },
    {
      "epoch": 4.878599661208357,
      "grad_norm": 4.412207126617432,
      "learning_rate": 1.2140033879164315e-05,
      "loss": 3.3158,
      "step": 25920
    },
    {
      "epoch": 4.880481837003576,
      "grad_norm": 3.0507736206054688,
      "learning_rate": 1.1951816299642386e-05,
      "loss": 3.2019,
      "step": 25930
    },
    {
      "epoch": 4.882364012798796,
      "grad_norm": 3.0898468494415283,
      "learning_rate": 1.176359872012046e-05,
      "loss": 3.1667,
      "step": 25940
    },
    {
      "epoch": 4.884246188594014,
      "grad_norm": 2.6355340480804443,
      "learning_rate": 1.1575381140598532e-05,
      "loss": 3.1116,
      "step": 25950
    },
    {
      "epoch": 4.886128364389234,
      "grad_norm": 3.2803468704223633,
      "learning_rate": 1.1387163561076605e-05,
      "loss": 3.0631,
      "step": 25960
    },
    {
      "epoch": 4.888010540184453,
      "grad_norm": 3.071220636367798,
      "learning_rate": 1.1198945981554678e-05,
      "loss": 3.0295,
      "step": 25970
    },
    {
      "epoch": 4.889892715979673,
      "grad_norm": 3.0252721309661865,
      "learning_rate": 1.101072840203275e-05,
      "loss": 3.0742,
      "step": 25980
    },
    {
      "epoch": 4.891774891774892,
      "grad_norm": 3.041609764099121,
      "learning_rate": 1.0822510822510823e-05,
      "loss": 3.1043,
      "step": 25990
    },
    {
      "epoch": 4.893657067570111,
      "grad_norm": 3.819404125213623,
      "learning_rate": 1.0634293242988896e-05,
      "loss": 3.275,
      "step": 26000
    },
    {
      "epoch": 4.89553924336533,
      "grad_norm": 2.8616020679473877,
      "learning_rate": 1.0446075663466969e-05,
      "loss": 2.8112,
      "step": 26010
    },
    {
      "epoch": 4.897421419160549,
      "grad_norm": 3.4349212646484375,
      "learning_rate": 1.025785808394504e-05,
      "loss": 3.3595,
      "step": 26020
    },
    {
      "epoch": 4.899303594955769,
      "grad_norm": 3.0160281658172607,
      "learning_rate": 1.0069640504423115e-05,
      "loss": 3.1097,
      "step": 26030
    },
    {
      "epoch": 4.901185770750988,
      "grad_norm": 3.218116521835327,
      "learning_rate": 9.881422924901186e-06,
      "loss": 2.9423,
      "step": 26040
    },
    {
      "epoch": 4.903067946546208,
      "grad_norm": 3.245553970336914,
      "learning_rate": 9.693205345379259e-06,
      "loss": 3.151,
      "step": 26050
    },
    {
      "epoch": 4.904950122341427,
      "grad_norm": 3.4128801822662354,
      "learning_rate": 9.504987765857332e-06,
      "loss": 3.1474,
      "step": 26060
    },
    {
      "epoch": 4.906832298136646,
      "grad_norm": 3.061704397201538,
      "learning_rate": 9.316770186335403e-06,
      "loss": 3.1783,
      "step": 26070
    },
    {
      "epoch": 4.908714473931865,
      "grad_norm": 3.863232374191284,
      "learning_rate": 9.128552606813476e-06,
      "loss": 3.4587,
      "step": 26080
    },
    {
      "epoch": 4.9105966497270845,
      "grad_norm": 3.27815580368042,
      "learning_rate": 8.94033502729155e-06,
      "loss": 3.2556,
      "step": 26090
    },
    {
      "epoch": 4.912478825522304,
      "grad_norm": 3.329728364944458,
      "learning_rate": 8.752117447769623e-06,
      "loss": 3.0733,
      "step": 26100
    },
    {
      "epoch": 4.914361001317523,
      "grad_norm": 4.6205058097839355,
      "learning_rate": 8.563899868247694e-06,
      "loss": 3.2327,
      "step": 26110
    },
    {
      "epoch": 4.916243177112742,
      "grad_norm": 3.4472804069519043,
      "learning_rate": 8.375682288725767e-06,
      "loss": 3.2555,
      "step": 26120
    },
    {
      "epoch": 4.918125352907961,
      "grad_norm": 3.038895845413208,
      "learning_rate": 8.18746470920384e-06,
      "loss": 3.4315,
      "step": 26130
    },
    {
      "epoch": 4.920007528703181,
      "grad_norm": 2.9684219360351562,
      "learning_rate": 7.999247129681913e-06,
      "loss": 3.3593,
      "step": 26140
    },
    {
      "epoch": 4.9218897044984,
      "grad_norm": 5.256326675415039,
      "learning_rate": 7.811029550159986e-06,
      "loss": 3.2857,
      "step": 26150
    },
    {
      "epoch": 4.92377188029362,
      "grad_norm": 2.682718515396118,
      "learning_rate": 7.622811970638058e-06,
      "loss": 3.2377,
      "step": 26160
    },
    {
      "epoch": 4.925654056088838,
      "grad_norm": 3.5212459564208984,
      "learning_rate": 7.43459439111613e-06,
      "loss": 3.3286,
      "step": 26170
    },
    {
      "epoch": 4.927536231884058,
      "grad_norm": 3.4938385486602783,
      "learning_rate": 7.246376811594203e-06,
      "loss": 3.1179,
      "step": 26180
    },
    {
      "epoch": 4.929418407679277,
      "grad_norm": 3.7432050704956055,
      "learning_rate": 7.0581592320722764e-06,
      "loss": 3.3646,
      "step": 26190
    },
    {
      "epoch": 4.9313005834744965,
      "grad_norm": 4.2980475425720215,
      "learning_rate": 6.869941652550348e-06,
      "loss": 3.1579,
      "step": 26200
    },
    {
      "epoch": 4.933182759269716,
      "grad_norm": 3.2238101959228516,
      "learning_rate": 6.681724073028421e-06,
      "loss": 3.4587,
      "step": 26210
    },
    {
      "epoch": 4.935064935064935,
      "grad_norm": 3.5493414402008057,
      "learning_rate": 6.493506493506494e-06,
      "loss": 3.1526,
      "step": 26220
    },
    {
      "epoch": 4.936947110860155,
      "grad_norm": 2.7393136024475098,
      "learning_rate": 6.305288913984566e-06,
      "loss": 3.2802,
      "step": 26230
    },
    {
      "epoch": 4.938829286655373,
      "grad_norm": 4.225271701812744,
      "learning_rate": 6.117071334462639e-06,
      "loss": 3.3651,
      "step": 26240
    },
    {
      "epoch": 4.940711462450593,
      "grad_norm": 3.390509605407715,
      "learning_rate": 5.928853754940711e-06,
      "loss": 3.1389,
      "step": 26250
    },
    {
      "epoch": 4.942593638245812,
      "grad_norm": 3.0110936164855957,
      "learning_rate": 5.740636175418784e-06,
      "loss": 3.01,
      "step": 26260
    },
    {
      "epoch": 4.944475814041032,
      "grad_norm": 3.559267997741699,
      "learning_rate": 5.552418595896857e-06,
      "loss": 3.1301,
      "step": 26270
    },
    {
      "epoch": 4.946357989836251,
      "grad_norm": 2.8467063903808594,
      "learning_rate": 5.3642010163749295e-06,
      "loss": 3.0362,
      "step": 26280
    },
    {
      "epoch": 4.94824016563147,
      "grad_norm": 2.625588893890381,
      "learning_rate": 5.175983436853002e-06,
      "loss": 3.0604,
      "step": 26290
    },
    {
      "epoch": 4.950122341426689,
      "grad_norm": 2.751812696456909,
      "learning_rate": 4.987765857331075e-06,
      "loss": 2.7247,
      "step": 26300
    },
    {
      "epoch": 4.9520045172219085,
      "grad_norm": 3.6566710472106934,
      "learning_rate": 4.799548277809147e-06,
      "loss": 3.0319,
      "step": 26310
    },
    {
      "epoch": 4.953886693017128,
      "grad_norm": 4.1501288414001465,
      "learning_rate": 4.61133069828722e-06,
      "loss": 3.2094,
      "step": 26320
    },
    {
      "epoch": 4.955768868812347,
      "grad_norm": 2.843015670776367,
      "learning_rate": 4.423113118765293e-06,
      "loss": 3.1845,
      "step": 26330
    },
    {
      "epoch": 4.957651044607566,
      "grad_norm": 3.4206199645996094,
      "learning_rate": 4.234895539243365e-06,
      "loss": 3.0502,
      "step": 26340
    },
    {
      "epoch": 4.959533220402785,
      "grad_norm": 3.2094857692718506,
      "learning_rate": 4.046677959721438e-06,
      "loss": 3.3604,
      "step": 26350
    },
    {
      "epoch": 4.961415396198005,
      "grad_norm": 3.3807573318481445,
      "learning_rate": 3.85846038019951e-06,
      "loss": 3.3002,
      "step": 26360
    },
    {
      "epoch": 4.963297571993224,
      "grad_norm": 2.6988401412963867,
      "learning_rate": 3.6702428006775834e-06,
      "loss": 3.0346,
      "step": 26370
    },
    {
      "epoch": 4.965179747788444,
      "grad_norm": 2.562386989593506,
      "learning_rate": 3.482025221155656e-06,
      "loss": 3.1302,
      "step": 26380
    },
    {
      "epoch": 4.967061923583663,
      "grad_norm": 2.9049251079559326,
      "learning_rate": 3.2938076416337287e-06,
      "loss": 3.1375,
      "step": 26390
    },
    {
      "epoch": 4.9689440993788825,
      "grad_norm": 5.675838947296143,
      "learning_rate": 3.1055900621118013e-06,
      "loss": 3.312,
      "step": 26400
    },
    {
      "epoch": 4.970826275174101,
      "grad_norm": 2.4691736698150635,
      "learning_rate": 2.917372482589874e-06,
      "loss": 3.1683,
      "step": 26410
    },
    {
      "epoch": 4.9727084509693205,
      "grad_norm": 3.464921236038208,
      "learning_rate": 2.7291549030679465e-06,
      "loss": 3.2794,
      "step": 26420
    },
    {
      "epoch": 4.97459062676454,
      "grad_norm": 3.111931562423706,
      "learning_rate": 2.540937323546019e-06,
      "loss": 3.2953,
      "step": 26430
    },
    {
      "epoch": 4.976472802559759,
      "grad_norm": 3.5703999996185303,
      "learning_rate": 2.352719744024092e-06,
      "loss": 2.9351,
      "step": 26440
    },
    {
      "epoch": 4.978354978354979,
      "grad_norm": 2.725428342819214,
      "learning_rate": 2.1645021645021648e-06,
      "loss": 2.988,
      "step": 26450
    },
    {
      "epoch": 4.980237154150197,
      "grad_norm": 2.8327372074127197,
      "learning_rate": 1.976284584980237e-06,
      "loss": 2.7986,
      "step": 26460
    },
    {
      "epoch": 4.982119329945417,
      "grad_norm": 3.3417296409606934,
      "learning_rate": 1.7880670054583098e-06,
      "loss": 3.096,
      "step": 26470
    },
    {
      "epoch": 4.984001505740636,
      "grad_norm": 2.8281989097595215,
      "learning_rate": 1.5998494259363826e-06,
      "loss": 3.0627,
      "step": 26480
    },
    {
      "epoch": 4.985883681535856,
      "grad_norm": 2.6077282428741455,
      "learning_rate": 1.411631846414455e-06,
      "loss": 3.1875,
      "step": 26490
    },
    {
      "epoch": 4.987765857331075,
      "grad_norm": 3.3847365379333496,
      "learning_rate": 1.2234142668925278e-06,
      "loss": 3.311,
      "step": 26500
    },
    {
      "epoch": 4.989648033126294,
      "grad_norm": 4.011746883392334,
      "learning_rate": 1.0351966873706004e-06,
      "loss": 2.8797,
      "step": 26510
    },
    {
      "epoch": 4.991530208921513,
      "grad_norm": 3.3813626766204834,
      "learning_rate": 8.469791078486731e-07,
      "loss": 3.1457,
      "step": 26520
    },
    {
      "epoch": 4.993412384716732,
      "grad_norm": 2.967564344406128,
      "learning_rate": 6.587615283267458e-07,
      "loss": 3.2162,
      "step": 26530
    },
    {
      "epoch": 4.995294560511952,
      "grad_norm": 2.9905147552490234,
      "learning_rate": 4.7054394880481837e-07,
      "loss": 3.188,
      "step": 26540
    },
    {
      "epoch": 4.997176736307171,
      "grad_norm": 3.274362802505493,
      "learning_rate": 2.8232636928289103e-07,
      "loss": 2.7407,
      "step": 26550
    },
    {
      "epoch": 4.999058912102391,
      "grad_norm": 3.104670524597168,
      "learning_rate": 9.410878976096367e-08,
      "loss": 3.1158,
      "step": 26560
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.22226666666666667,
      "eval_loss": 3.161168336868286,
      "eval_runtime": 116.485,
      "eval_samples_per_second": 64.386,
      "eval_steps_per_second": 8.053,
      "step": 26565
    },
    {
      "epoch": 5.0,
      "step": 26565,
      "total_flos": 1.64815115092992e+19,
      "train_loss": 3.597675289453456,
      "train_runtime": 6995.4319,
      "train_samples_per_second": 30.377,
      "train_steps_per_second": 3.797
    }
  ],
  "logging_steps": 10,
  "max_steps": 26565,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 1.64815115092992e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}