{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6658956296218413,
  "eval_steps": 500,
  "global_step": 190000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00021028283040689727,
      "grad_norm": 2.9733738899230957,
      "learning_rate": 4.99907125083237e-05,
      "loss": 9.4089,
      "step": 60
    },
    {
      "epoch": 0.00042056566081379455,
      "grad_norm": 2.3245558738708496,
      "learning_rate": 4.9980198366803356e-05,
      "loss": 3.6104,
      "step": 120
    },
    {
      "epoch": 0.0006308484912206919,
      "grad_norm": 2.0401523113250732,
      "learning_rate": 4.996968422528301e-05,
      "loss": 3.3119,
      "step": 180
    },
    {
      "epoch": 0.0008411313216275891,
      "grad_norm": 1.8371763229370117,
      "learning_rate": 4.995917008376266e-05,
      "loss": 3.2129,
      "step": 240
    },
    {
      "epoch": 0.0010514141520344864,
      "grad_norm": 1.6748260259628296,
      "learning_rate": 4.994865594224232e-05,
      "loss": 3.1313,
      "step": 300
    },
    {
      "epoch": 0.0012616969824413838,
      "grad_norm": 1.8345626592636108,
      "learning_rate": 4.993814180072197e-05,
      "loss": 3.1001,
      "step": 360
    },
    {
      "epoch": 0.0014719798128482809,
      "grad_norm": 1.54319429397583,
      "learning_rate": 4.992762765920163e-05,
      "loss": 3.0693,
      "step": 420
    },
    {
      "epoch": 0.0016822626432551782,
      "grad_norm": 1.4675054550170898,
      "learning_rate": 4.991711351768129e-05,
      "loss": 3.0448,
      "step": 480
    },
    {
      "epoch": 0.0018925454736620755,
      "grad_norm": 1.5000808238983154,
      "learning_rate": 4.990659937616094e-05,
      "loss": 3.0369,
      "step": 540
    },
    {
      "epoch": 0.002102828304068973,
      "grad_norm": 1.3771166801452637,
      "learning_rate": 4.989608523464059e-05,
      "loss": 3.021,
      "step": 600
    },
    {
      "epoch": 0.00231311113447587,
      "grad_norm": 1.3988029956817627,
      "learning_rate": 4.988557109312025e-05,
      "loss": 3.0092,
      "step": 660
    },
    {
      "epoch": 0.0025233939648827675,
      "grad_norm": 1.3383244276046753,
      "learning_rate": 4.98750569515999e-05,
      "loss": 2.9906,
      "step": 720
    },
    {
      "epoch": 0.0027336767952896644,
      "grad_norm": 1.3521662950515747,
      "learning_rate": 4.986454281007956e-05,
      "loss": 2.9699,
      "step": 780
    },
    {
      "epoch": 0.0029439596256965617,
      "grad_norm": 1.287575125694275,
      "learning_rate": 4.985402866855921e-05,
      "loss": 2.9622,
      "step": 840
    },
    {
      "epoch": 0.003154242456103459,
      "grad_norm": 1.2298351526260376,
      "learning_rate": 4.984351452703887e-05,
      "loss": 2.9578,
      "step": 900
    },
    {
      "epoch": 0.0033645252865103564,
      "grad_norm": 1.3640484809875488,
      "learning_rate": 4.9833000385518524e-05,
      "loss": 2.9551,
      "step": 960
    },
    {
      "epoch": 0.0035748081169172537,
      "grad_norm": 1.273837685585022,
      "learning_rate": 4.9822486243998175e-05,
      "loss": 2.9444,
      "step": 1020
    },
    {
      "epoch": 0.003785090947324151,
      "grad_norm": 1.2357532978057861,
      "learning_rate": 4.9811972102477834e-05,
      "loss": 2.9326,
      "step": 1080
    },
    {
      "epoch": 0.003995373777731048,
      "grad_norm": 1.211605429649353,
      "learning_rate": 4.980145796095749e-05,
      "loss": 2.923,
      "step": 1140
    },
    {
      "epoch": 0.004205656608137946,
      "grad_norm": 1.1577990055084229,
      "learning_rate": 4.9790943819437145e-05,
      "loss": 2.9242,
      "step": 1200
    },
    {
      "epoch": 0.004415939438544843,
      "grad_norm": 1.2026646137237549,
      "learning_rate": 4.97804296779168e-05,
      "loss": 2.9175,
      "step": 1260
    },
    {
      "epoch": 0.00462622226895174,
      "grad_norm": 1.2141482830047607,
      "learning_rate": 4.9769915536396455e-05,
      "loss": 2.9319,
      "step": 1320
    },
    {
      "epoch": 0.004836505099358637,
      "grad_norm": 1.1758204698562622,
      "learning_rate": 4.975940139487611e-05,
      "loss": 2.9047,
      "step": 1380
    },
    {
      "epoch": 0.005046787929765535,
      "grad_norm": 1.2033660411834717,
      "learning_rate": 4.9748887253355766e-05,
      "loss": 2.9102,
      "step": 1440
    },
    {
      "epoch": 0.005257070760172432,
      "grad_norm": 1.2934941053390503,
      "learning_rate": 4.9738373111835424e-05,
      "loss": 2.9022,
      "step": 1500
    },
    {
      "epoch": 0.005467353590579329,
      "grad_norm": 1.1089513301849365,
      "learning_rate": 4.9727858970315076e-05,
      "loss": 2.8968,
      "step": 1560
    },
    {
      "epoch": 0.005677636420986227,
      "grad_norm": 1.081939697265625,
      "learning_rate": 4.9717344828794735e-05,
      "loss": 2.8894,
      "step": 1620
    },
    {
      "epoch": 0.0058879192513931235,
      "grad_norm": 1.0430926084518433,
      "learning_rate": 4.970683068727439e-05,
      "loss": 2.8958,
      "step": 1680
    },
    {
      "epoch": 0.006098202081800021,
      "grad_norm": 1.118507742881775,
      "learning_rate": 4.969631654575404e-05,
      "loss": 2.8837,
      "step": 1740
    },
    {
      "epoch": 0.006308484912206918,
      "grad_norm": 1.163057804107666,
      "learning_rate": 4.96858024042337e-05,
      "loss": 2.8754,
      "step": 1800
    },
    {
      "epoch": 0.006518767742613816,
      "grad_norm": 1.0965882539749146,
      "learning_rate": 4.9675288262713356e-05,
      "loss": 2.8799,
      "step": 1860
    },
    {
      "epoch": 0.006729050573020713,
      "grad_norm": 1.0341143608093262,
      "learning_rate": 4.966477412119301e-05,
      "loss": 2.8794,
      "step": 1920
    },
    {
      "epoch": 0.0069393334034276105,
      "grad_norm": 0.9922598004341125,
      "learning_rate": 4.965425997967266e-05,
      "loss": 2.8769,
      "step": 1980
    },
    {
      "epoch": 0.0071496162338345074,
      "grad_norm": 1.0550235509872437,
      "learning_rate": 4.964374583815232e-05,
      "loss": 2.8557,
      "step": 2040
    },
    {
      "epoch": 0.007359899064241404,
      "grad_norm": 1.0422568321228027,
      "learning_rate": 4.963323169663197e-05,
      "loss": 2.8654,
      "step": 2100
    },
    {
      "epoch": 0.007570181894648302,
      "grad_norm": 1.0454896688461304,
      "learning_rate": 4.962271755511162e-05,
      "loss": 2.8542,
      "step": 2160
    },
    {
      "epoch": 0.007780464725055199,
      "grad_norm": 1.0227279663085938,
      "learning_rate": 4.961220341359129e-05,
      "loss": 2.8552,
      "step": 2220
    },
    {
      "epoch": 0.007990747555462096,
      "grad_norm": 1.0329540967941284,
      "learning_rate": 4.960168927207094e-05,
      "loss": 2.8527,
      "step": 2280
    },
    {
      "epoch": 0.008201030385868995,
      "grad_norm": 1.0209529399871826,
      "learning_rate": 4.959117513055059e-05,
      "loss": 2.8433,
      "step": 2340
    },
    {
      "epoch": 0.008411313216275891,
      "grad_norm": 1.0325701236724854,
      "learning_rate": 4.958066098903025e-05,
      "loss": 2.842,
      "step": 2400
    },
    {
      "epoch": 0.008621596046682788,
      "grad_norm": 0.9857275485992432,
      "learning_rate": 4.95701468475099e-05,
      "loss": 2.8324,
      "step": 2460
    },
    {
      "epoch": 0.008831878877089685,
      "grad_norm": 1.027936339378357,
      "learning_rate": 4.9559632705989554e-05,
      "loss": 2.8347,
      "step": 2520
    },
    {
      "epoch": 0.009042161707496582,
      "grad_norm": 1.045196771621704,
      "learning_rate": 4.954911856446921e-05,
      "loss": 2.8279,
      "step": 2580
    },
    {
      "epoch": 0.00925244453790348,
      "grad_norm": 1.0517592430114746,
      "learning_rate": 4.953860442294887e-05,
      "loss": 2.82,
      "step": 2640
    },
    {
      "epoch": 0.009462727368310378,
      "grad_norm": 1.0048534870147705,
      "learning_rate": 4.952809028142852e-05,
      "loss": 2.814,
      "step": 2700
    },
    {
      "epoch": 0.009673010198717275,
      "grad_norm": 1.0240980386734009,
      "learning_rate": 4.9517576139908175e-05,
      "loss": 2.7985,
      "step": 2760
    },
    {
      "epoch": 0.009883293029124171,
      "grad_norm": 1.019655704498291,
      "learning_rate": 4.9507061998387834e-05,
      "loss": 2.7932,
      "step": 2820
    },
    {
      "epoch": 0.01009357585953107,
      "grad_norm": 1.0587692260742188,
      "learning_rate": 4.9496547856867486e-05,
      "loss": 2.7943,
      "step": 2880
    },
    {
      "epoch": 0.010303858689937967,
      "grad_norm": 1.0488303899765015,
      "learning_rate": 4.9486033715347144e-05,
      "loss": 2.7792,
      "step": 2940
    },
    {
      "epoch": 0.010514141520344864,
      "grad_norm": 1.0349452495574951,
      "learning_rate": 4.94755195738268e-05,
      "loss": 2.7717,
      "step": 3000
    },
    {
      "epoch": 0.01072442435075176,
      "grad_norm": 1.0631664991378784,
      "learning_rate": 4.9465005432306455e-05,
      "loss": 2.7696,
      "step": 3060
    },
    {
      "epoch": 0.010934707181158658,
      "grad_norm": 1.026247501373291,
      "learning_rate": 4.945449129078611e-05,
      "loss": 2.7586,
      "step": 3120
    },
    {
      "epoch": 0.011144990011565556,
      "grad_norm": 1.049814224243164,
      "learning_rate": 4.9443977149265766e-05,
      "loss": 2.746,
      "step": 3180
    },
    {
      "epoch": 0.011355272841972453,
      "grad_norm": 1.0707862377166748,
      "learning_rate": 4.943346300774542e-05,
      "loss": 2.7169,
      "step": 3240
    },
    {
      "epoch": 0.01156555567237935,
      "grad_norm": 1.092517614364624,
      "learning_rate": 4.9422948866225076e-05,
      "loss": 2.719,
      "step": 3300
    },
    {
      "epoch": 0.011775838502786247,
      "grad_norm": 0.9983085989952087,
      "learning_rate": 4.9412434724704735e-05,
      "loss": 2.7114,
      "step": 3360
    },
    {
      "epoch": 0.011986121333193146,
      "grad_norm": 1.0133140087127686,
      "learning_rate": 4.940192058318439e-05,
      "loss": 2.6965,
      "step": 3420
    },
    {
      "epoch": 0.012196404163600042,
      "grad_norm": 1.1006369590759277,
      "learning_rate": 4.939140644166404e-05,
      "loss": 2.7053,
      "step": 3480
    },
    {
      "epoch": 0.01240668699400694,
      "grad_norm": 1.0474873781204224,
      "learning_rate": 4.93808923001437e-05,
      "loss": 2.6907,
      "step": 3540
    },
    {
      "epoch": 0.012616969824413836,
      "grad_norm": 1.033618450164795,
      "learning_rate": 4.937037815862335e-05,
      "loss": 2.6892,
      "step": 3600
    },
    {
      "epoch": 0.012827252654820733,
      "grad_norm": 1.0913538932800293,
      "learning_rate": 4.935986401710301e-05,
      "loss": 2.6782,
      "step": 3660
    },
    {
      "epoch": 0.013037535485227632,
      "grad_norm": 1.0223588943481445,
      "learning_rate": 4.934934987558266e-05,
      "loss": 2.6607,
      "step": 3720
    },
    {
      "epoch": 0.013247818315634529,
      "grad_norm": 1.0294113159179688,
      "learning_rate": 4.933883573406232e-05,
      "loss": 2.66,
      "step": 3780
    },
    {
      "epoch": 0.013458101146041426,
      "grad_norm": 0.9733858704566956,
      "learning_rate": 4.932832159254197e-05,
      "loss": 2.6583,
      "step": 3840
    },
    {
      "epoch": 0.013668383976448322,
      "grad_norm": 1.1031690835952759,
      "learning_rate": 4.931780745102162e-05,
      "loss": 2.6736,
      "step": 3900
    },
    {
      "epoch": 0.013878666806855221,
      "grad_norm": 1.0241620540618896,
      "learning_rate": 4.930729330950128e-05,
      "loss": 2.652,
      "step": 3960
    },
    {
      "epoch": 0.014088949637262118,
      "grad_norm": 1.017519235610962,
      "learning_rate": 4.929677916798094e-05,
      "loss": 2.6346,
      "step": 4020
    },
    {
      "epoch": 0.014299232467669015,
      "grad_norm": 1.0698292255401611,
      "learning_rate": 4.928626502646059e-05,
      "loss": 2.6272,
      "step": 4080
    },
    {
      "epoch": 0.014509515298075912,
      "grad_norm": 1.0114396810531616,
      "learning_rate": 4.927575088494025e-05,
      "loss": 2.6419,
      "step": 4140
    },
    {
      "epoch": 0.014719798128482809,
      "grad_norm": 1.0519788265228271,
      "learning_rate": 4.92652367434199e-05,
      "loss": 2.6147,
      "step": 4200
    },
    {
      "epoch": 0.014930080958889707,
      "grad_norm": 0.952059268951416,
      "learning_rate": 4.9254722601899554e-05,
      "loss": 2.6179,
      "step": 4260
    },
    {
      "epoch": 0.015140363789296604,
      "grad_norm": 0.9879328012466431,
      "learning_rate": 4.924420846037921e-05,
      "loss": 2.6115,
      "step": 4320
    },
    {
      "epoch": 0.015350646619703501,
      "grad_norm": 1.000038504600525,
      "learning_rate": 4.923369431885887e-05,
      "loss": 2.6155,
      "step": 4380
    },
    {
      "epoch": 0.015560929450110398,
      "grad_norm": 1.020724892616272,
      "learning_rate": 4.922318017733852e-05,
      "loss": 2.5966,
      "step": 4440
    },
    {
      "epoch": 0.015771212280517297,
      "grad_norm": 1.0393227338790894,
      "learning_rate": 4.9212666035818175e-05,
      "loss": 2.6104,
      "step": 4500
    },
    {
      "epoch": 0.015981495110924192,
      "grad_norm": 0.9846431612968445,
      "learning_rate": 4.9202151894297834e-05,
      "loss": 2.5907,
      "step": 4560
    },
    {
      "epoch": 0.01619177794133109,
      "grad_norm": 1.0070515871047974,
      "learning_rate": 4.9191637752777485e-05,
      "loss": 2.5962,
      "step": 4620
    },
    {
      "epoch": 0.01640206077173799,
      "grad_norm": 0.9761502742767334,
      "learning_rate": 4.918112361125714e-05,
      "loss": 2.6027,
      "step": 4680
    },
    {
      "epoch": 0.016612343602144884,
      "grad_norm": 0.9863872528076172,
      "learning_rate": 4.91706094697368e-05,
      "loss": 2.5745,
      "step": 4740
    },
    {
      "epoch": 0.016822626432551783,
      "grad_norm": 0.9633762240409851,
      "learning_rate": 4.9160095328216455e-05,
      "loss": 2.5901,
      "step": 4800
    },
    {
      "epoch": 0.017032909262958678,
      "grad_norm": 1.013360619544983,
      "learning_rate": 4.9149581186696107e-05,
      "loss": 2.57,
      "step": 4860
    },
    {
      "epoch": 0.017243192093365577,
      "grad_norm": 0.9540980458259583,
      "learning_rate": 4.9139067045175765e-05,
      "loss": 2.5718,
      "step": 4920
    },
    {
      "epoch": 0.017453474923772475,
      "grad_norm": 0.937667965888977,
      "learning_rate": 4.912855290365542e-05,
      "loss": 2.5899,
      "step": 4980
    },
    {
      "epoch": 0.01766375775417937,
      "grad_norm": 0.9429223537445068,
      "learning_rate": 4.911803876213507e-05,
      "loss": 2.591,
      "step": 5040
    },
    {
      "epoch": 0.01787404058458627,
      "grad_norm": 0.9466859102249146,
      "learning_rate": 4.910752462061473e-05,
      "loss": 2.5675,
      "step": 5100
    },
    {
      "epoch": 0.018084323414993164,
      "grad_norm": 0.9759580492973328,
      "learning_rate": 4.9097010479094386e-05,
      "loss": 2.5778,
      "step": 5160
    },
    {
      "epoch": 0.018294606245400063,
      "grad_norm": 0.974449634552002,
      "learning_rate": 4.908649633757404e-05,
      "loss": 2.5657,
      "step": 5220
    },
    {
      "epoch": 0.01850488907580696,
      "grad_norm": 0.9969838857650757,
      "learning_rate": 4.90759821960537e-05,
      "loss": 2.5451,
      "step": 5280
    },
    {
      "epoch": 0.018715171906213857,
      "grad_norm": 0.8910425901412964,
      "learning_rate": 4.906546805453335e-05,
      "loss": 2.5686,
      "step": 5340
    },
    {
      "epoch": 0.018925454736620755,
      "grad_norm": 0.964389979839325,
      "learning_rate": 4.905495391301301e-05,
      "loss": 2.5562,
      "step": 5400
    },
    {
      "epoch": 0.019135737567027654,
      "grad_norm": 0.9532023072242737,
      "learning_rate": 4.904443977149266e-05,
      "loss": 2.5487,
      "step": 5460
    },
    {
      "epoch": 0.01934602039743455,
      "grad_norm": 0.8950223326683044,
      "learning_rate": 4.903392562997232e-05,
      "loss": 2.5552,
      "step": 5520
    },
    {
      "epoch": 0.019556303227841448,
      "grad_norm": 0.97803795337677,
      "learning_rate": 4.902341148845197e-05,
      "loss": 2.5639,
      "step": 5580
    },
    {
      "epoch": 0.019766586058248343,
      "grad_norm": 0.9063622951507568,
      "learning_rate": 4.901289734693162e-05,
      "loss": 2.5484,
      "step": 5640
    },
    {
      "epoch": 0.01997686888865524,
      "grad_norm": 1.014898419380188,
      "learning_rate": 4.900238320541128e-05,
      "loss": 2.5646,
      "step": 5700
    },
    {
      "epoch": 0.02018715171906214,
      "grad_norm": 0.938632607460022,
      "learning_rate": 4.899186906389094e-05,
      "loss": 2.5457,
      "step": 5760
    },
    {
      "epoch": 0.020397434549469035,
      "grad_norm": 0.9069220423698425,
      "learning_rate": 4.898135492237059e-05,
      "loss": 2.5633,
      "step": 5820
    },
    {
      "epoch": 0.020607717379875934,
      "grad_norm": 0.8817732334136963,
      "learning_rate": 4.897084078085025e-05,
      "loss": 2.5471,
      "step": 5880
    },
    {
      "epoch": 0.02081800021028283,
      "grad_norm": 1.003666639328003,
      "learning_rate": 4.89603266393299e-05,
      "loss": 2.5588,
      "step": 5940
    },
    {
      "epoch": 0.021028283040689728,
      "grad_norm": 0.9436509609222412,
      "learning_rate": 4.8949812497809553e-05,
      "loss": 2.536,
      "step": 6000
    },
    {
      "epoch": 0.021238565871096626,
      "grad_norm": 0.9367396235466003,
      "learning_rate": 4.893929835628921e-05,
      "loss": 2.5427,
      "step": 6060
    },
    {
      "epoch": 0.02144884870150352,
      "grad_norm": 0.8808217644691467,
      "learning_rate": 4.892878421476887e-05,
      "loss": 2.5447,
      "step": 6120
    },
    {
      "epoch": 0.02165913153191042,
      "grad_norm": 0.9140152335166931,
      "learning_rate": 4.891827007324852e-05,
      "loss": 2.5298,
      "step": 6180
    },
    {
      "epoch": 0.021869414362317315,
      "grad_norm": 0.9449228644371033,
      "learning_rate": 4.8907755931728175e-05,
      "loss": 2.5421,
      "step": 6240
    },
    {
      "epoch": 0.022079697192724214,
      "grad_norm": 0.9297705292701721,
      "learning_rate": 4.889724179020783e-05,
      "loss": 2.5394,
      "step": 6300
    },
    {
      "epoch": 0.022289980023131113,
      "grad_norm": 0.8961461186408997,
      "learning_rate": 4.8886727648687485e-05,
      "loss": 2.5296,
      "step": 6360
    },
    {
      "epoch": 0.022500262853538008,
      "grad_norm": 0.9122869372367859,
      "learning_rate": 4.887621350716714e-05,
      "loss": 2.5292,
      "step": 6420
    },
    {
      "epoch": 0.022710545683944906,
      "grad_norm": 0.8803234100341797,
      "learning_rate": 4.88656993656468e-05,
      "loss": 2.5429,
      "step": 6480
    },
    {
      "epoch": 0.0229208285143518,
      "grad_norm": 0.9121540784835815,
      "learning_rate": 4.8855185224126454e-05,
      "loss": 2.5127,
      "step": 6540
    },
    {
      "epoch": 0.0231311113447587,
      "grad_norm": 0.9358422756195068,
      "learning_rate": 4.8844671082606106e-05,
      "loss": 2.535,
      "step": 6600
    },
    {
      "epoch": 0.0233413941751656,
      "grad_norm": 0.8811025023460388,
      "learning_rate": 4.8834156941085765e-05,
      "loss": 2.5332,
      "step": 6660
    },
    {
      "epoch": 0.023551677005572494,
      "grad_norm": 0.9080128073692322,
      "learning_rate": 4.882364279956542e-05,
      "loss": 2.5034,
      "step": 6720
    },
    {
      "epoch": 0.023761959835979393,
      "grad_norm": 0.8822489380836487,
      "learning_rate": 4.881312865804507e-05,
      "loss": 2.5028,
      "step": 6780
    },
    {
      "epoch": 0.02397224266638629,
      "grad_norm": 0.9486945271492004,
      "learning_rate": 4.880261451652473e-05,
      "loss": 2.519,
      "step": 6840
    },
    {
      "epoch": 0.024182525496793186,
      "grad_norm": 0.8917908072471619,
      "learning_rate": 4.8792100375004386e-05,
      "loss": 2.5171,
      "step": 6900
    },
    {
      "epoch": 0.024392808327200085,
      "grad_norm": 0.9071120023727417,
      "learning_rate": 4.878158623348404e-05,
      "loss": 2.5382,
      "step": 6960
    },
    {
      "epoch": 0.02460309115760698,
      "grad_norm": 0.8386140465736389,
      "learning_rate": 4.877107209196369e-05,
      "loss": 2.5238,
      "step": 7020
    },
    {
      "epoch": 0.02481337398801388,
      "grad_norm": 0.857929527759552,
      "learning_rate": 4.876055795044335e-05,
      "loss": 2.5112,
      "step": 7080
    },
    {
      "epoch": 0.025023656818420777,
      "grad_norm": 0.9252939820289612,
      "learning_rate": 4.8750043808923e-05,
      "loss": 2.5164,
      "step": 7140
    },
    {
      "epoch": 0.025233939648827673,
      "grad_norm": 0.8985560536384583,
      "learning_rate": 4.873952966740266e-05,
      "loss": 2.5206,
      "step": 7200
    },
    {
      "epoch": 0.02544422247923457,
      "grad_norm": 0.8806078433990479,
      "learning_rate": 4.872901552588232e-05,
      "loss": 2.5276,
      "step": 7260
    },
    {
      "epoch": 0.025654505309641466,
      "grad_norm": 0.8562397956848145,
      "learning_rate": 4.871850138436197e-05,
      "loss": 2.5144,
      "step": 7320
    },
    {
      "epoch": 0.025864788140048365,
      "grad_norm": 0.8808496594429016,
      "learning_rate": 4.870798724284162e-05,
      "loss": 2.5116,
      "step": 7380
    },
    {
      "epoch": 0.026075070970455264,
      "grad_norm": 0.8252414464950562,
      "learning_rate": 4.869747310132128e-05,
      "loss": 2.5146,
      "step": 7440
    },
    {
      "epoch": 0.02628535380086216,
      "grad_norm": 0.9716061949729919,
      "learning_rate": 4.868695895980093e-05,
      "loss": 2.5224,
      "step": 7500
    },
    {
      "epoch": 0.026495636631269057,
      "grad_norm": 0.9421200752258301,
      "learning_rate": 4.867644481828059e-05,
      "loss": 2.5087,
      "step": 7560
    },
    {
      "epoch": 0.026705919461675953,
      "grad_norm": 0.8816734552383423,
      "learning_rate": 4.866593067676025e-05,
      "loss": 2.5218,
      "step": 7620
    },
    {
      "epoch": 0.02691620229208285,
      "grad_norm": 0.8930853605270386,
      "learning_rate": 4.86554165352399e-05,
      "loss": 2.5142,
      "step": 7680
    },
    {
      "epoch": 0.02712648512248975,
      "grad_norm": 0.8809117078781128,
      "learning_rate": 4.864490239371955e-05,
      "loss": 2.5103,
      "step": 7740
    },
    {
      "epoch": 0.027336767952896645,
      "grad_norm": 0.8628544807434082,
      "learning_rate": 4.863438825219921e-05,
      "loss": 2.5115,
      "step": 7800
    },
    {
      "epoch": 0.027547050783303544,
      "grad_norm": 0.8847160339355469,
      "learning_rate": 4.8623874110678864e-05,
      "loss": 2.5095,
      "step": 7860
    },
    {
      "epoch": 0.027757333613710442,
      "grad_norm": 0.888922393321991,
      "learning_rate": 4.861335996915852e-05,
      "loss": 2.5135,
      "step": 7920
    },
    {
      "epoch": 0.027967616444117337,
      "grad_norm": 0.8904682993888855,
      "learning_rate": 4.8602845827638174e-05,
      "loss": 2.4958,
      "step": 7980
    },
    {
      "epoch": 0.028177899274524236,
      "grad_norm": 1.176128625869751,
      "learning_rate": 4.859233168611783e-05,
      "loss": 2.5173,
      "step": 8040
    },
    {
      "epoch": 0.02838818210493113,
      "grad_norm": 0.833441972732544,
      "learning_rate": 4.8581817544597485e-05,
      "loss": 2.4879,
      "step": 8100
    },
    {
      "epoch": 0.02859846493533803,
      "grad_norm": 0.8262945413589478,
      "learning_rate": 4.857130340307714e-05,
      "loss": 2.4856,
      "step": 8160
    },
    {
      "epoch": 0.02880874776574493,
      "grad_norm": 0.9492835402488708,
      "learning_rate": 4.8560789261556795e-05,
      "loss": 2.5002,
      "step": 8220
    },
    {
      "epoch": 0.029019030596151824,
      "grad_norm": 0.8316044211387634,
      "learning_rate": 4.8550275120036454e-05,
      "loss": 2.4947,
      "step": 8280
    },
    {
      "epoch": 0.029229313426558722,
      "grad_norm": 0.828999400138855,
      "learning_rate": 4.8539760978516106e-05,
      "loss": 2.4924,
      "step": 8340
    },
    {
      "epoch": 0.029439596256965617,
      "grad_norm": 0.8634353876113892,
      "learning_rate": 4.8529246836995765e-05,
      "loss": 2.5055,
      "step": 8400
    },
    {
      "epoch": 0.029649879087372516,
      "grad_norm": 0.8393489122390747,
      "learning_rate": 4.8518732695475417e-05,
      "loss": 2.4965,
      "step": 8460
    },
    {
      "epoch": 0.029860161917779415,
      "grad_norm": 0.824374794960022,
      "learning_rate": 4.850821855395507e-05,
      "loss": 2.4955,
      "step": 8520
    },
    {
      "epoch": 0.03007044474818631,
      "grad_norm": 0.8524509072303772,
      "learning_rate": 4.849770441243473e-05,
      "loss": 2.4894,
      "step": 8580
    },
    {
      "epoch": 0.03028072757859321,
      "grad_norm": 0.8297523856163025,
      "learning_rate": 4.8487190270914386e-05,
      "loss": 2.4959,
      "step": 8640
    },
    {
      "epoch": 0.030491010409000104,
      "grad_norm": 0.8216577768325806,
      "learning_rate": 4.847667612939404e-05,
      "loss": 2.4985,
      "step": 8700
    },
    {
      "epoch": 0.030701293239407002,
      "grad_norm": 0.7975873351097107,
      "learning_rate": 4.846616198787369e-05,
      "loss": 2.4817,
      "step": 8760
    },
    {
      "epoch": 0.0309115760698139,
      "grad_norm": 0.8472198843955994,
      "learning_rate": 4.845564784635335e-05,
      "loss": 2.4909,
      "step": 8820
    },
    {
      "epoch": 0.031121858900220796,
      "grad_norm": 0.837291955947876,
      "learning_rate": 4.8445133704833e-05,
      "loss": 2.4946,
      "step": 8880
    },
    {
      "epoch": 0.03133214173062769,
      "grad_norm": 0.8038502931594849,
      "learning_rate": 4.843461956331265e-05,
      "loss": 2.4971,
      "step": 8940
    },
    {
      "epoch": 0.03154242456103459,
      "grad_norm": 0.7669224739074707,
      "learning_rate": 4.842410542179232e-05,
      "loss": 2.4905,
      "step": 9000
    },
    {
      "epoch": 0.03175270739144149,
      "grad_norm": 0.8333732485771179,
      "learning_rate": 4.841359128027197e-05,
      "loss": 2.4949,
      "step": 9060
    },
    {
      "epoch": 0.031962990221848384,
      "grad_norm": 0.9027146697044373,
      "learning_rate": 4.840307713875162e-05,
      "loss": 2.4948,
      "step": 9120
    },
    {
      "epoch": 0.032173273052255286,
      "grad_norm": 0.8094586133956909,
      "learning_rate": 4.839256299723128e-05,
      "loss": 2.4733,
      "step": 9180
    },
    {
      "epoch": 0.03238355588266218,
      "grad_norm": 0.8115667104721069,
      "learning_rate": 4.838204885571093e-05,
      "loss": 2.4736,
      "step": 9240
    },
    {
      "epoch": 0.032593838713069076,
      "grad_norm": 0.8178938031196594,
      "learning_rate": 4.8371534714190584e-05,
      "loss": 2.4747,
      "step": 9300
    },
    {
      "epoch": 0.03280412154347598,
      "grad_norm": 0.8295975923538208,
      "learning_rate": 4.836102057267025e-05,
      "loss": 2.4927,
      "step": 9360
    },
    {
      "epoch": 0.03301440437388287,
      "grad_norm": 0.813774824142456,
      "learning_rate": 4.83505064311499e-05,
      "loss": 2.4935,
      "step": 9420
    },
    {
      "epoch": 0.03322468720428977,
      "grad_norm": 0.8150737881660461,
      "learning_rate": 4.833999228962955e-05,
      "loss": 2.4735,
      "step": 9480
    },
    {
      "epoch": 0.033434970034696664,
      "grad_norm": 0.8141120672225952,
      "learning_rate": 4.832947814810921e-05,
      "loss": 2.4798,
      "step": 9540
    },
    {
      "epoch": 0.033645252865103566,
      "grad_norm": 0.8410353064537048,
      "learning_rate": 4.8318964006588864e-05,
      "loss": 2.4706,
      "step": 9600
    },
    {
      "epoch": 0.03385553569551046,
      "grad_norm": 0.7936863303184509,
      "learning_rate": 4.8308449865068515e-05,
      "loss": 2.5007,
      "step": 9660
    },
    {
      "epoch": 0.034065818525917356,
      "grad_norm": 0.8532397150993347,
      "learning_rate": 4.8297935723548174e-05,
      "loss": 2.4819,
      "step": 9720
    },
    {
      "epoch": 0.03427610135632426,
      "grad_norm": 0.7777956128120422,
      "learning_rate": 4.828742158202783e-05,
      "loss": 2.465,
      "step": 9780
    },
    {
      "epoch": 0.03448638418673115,
      "grad_norm": 0.8209832310676575,
      "learning_rate": 4.8276907440507485e-05,
      "loss": 2.4618,
      "step": 9840
    },
    {
      "epoch": 0.03469666701713805,
      "grad_norm": 0.782342255115509,
      "learning_rate": 4.8266393298987137e-05,
      "loss": 2.4844,
      "step": 9900
    },
    {
      "epoch": 0.03490694984754495,
      "grad_norm": 0.8156614303588867,
      "learning_rate": 4.8255879157466795e-05,
      "loss": 2.4842,
      "step": 9960
    },
    {
      "epoch": 0.035117232677951846,
      "grad_norm": 0.8090491890907288,
      "learning_rate": 4.824536501594645e-05,
      "loss": 2.4631,
      "step": 10020
    },
    {
      "epoch": 0.03532751550835874,
      "grad_norm": 0.7655121684074402,
      "learning_rate": 4.8234850874426106e-05,
      "loss": 2.4701,
      "step": 10080
    },
    {
      "epoch": 0.03553779833876564,
      "grad_norm": 0.8753265738487244,
      "learning_rate": 4.8224336732905764e-05,
      "loss": 2.4683,
      "step": 10140
    },
    {
      "epoch": 0.03574808116917254,
      "grad_norm": 0.7905852794647217,
      "learning_rate": 4.8213822591385416e-05,
      "loss": 2.4685,
      "step": 10200
    },
    {
      "epoch": 0.03595836399957943,
      "grad_norm": 0.7805492281913757,
      "learning_rate": 4.820330844986507e-05,
      "loss": 2.4762,
      "step": 10260
    },
    {
      "epoch": 0.03616864682998633,
      "grad_norm": 0.8051442503929138,
      "learning_rate": 4.819279430834473e-05,
      "loss": 2.4542,
      "step": 10320
    },
    {
      "epoch": 0.03637892966039323,
      "grad_norm": 0.7971628308296204,
      "learning_rate": 4.818228016682438e-05,
      "loss": 2.4665,
      "step": 10380
    },
    {
      "epoch": 0.036589212490800126,
      "grad_norm": 0.8155962228775024,
      "learning_rate": 4.817176602530404e-05,
      "loss": 2.4912,
      "step": 10440
    },
    {
      "epoch": 0.03679949532120702,
      "grad_norm": 0.791641116142273,
      "learning_rate": 4.816125188378369e-05,
      "loss": 2.4663,
      "step": 10500
    },
    {
      "epoch": 0.03700977815161392,
      "grad_norm": 0.8577168583869934,
      "learning_rate": 4.815073774226335e-05,
      "loss": 2.4579,
      "step": 10560
    },
    {
      "epoch": 0.03722006098202082,
      "grad_norm": 0.7734136581420898,
      "learning_rate": 4.8140223600743e-05,
      "loss": 2.4546,
      "step": 10620
    },
    {
      "epoch": 0.03743034381242771,
      "grad_norm": 0.8331671357154846,
      "learning_rate": 4.812970945922265e-05,
      "loss": 2.4737,
      "step": 10680
    },
    {
      "epoch": 0.037640626642834615,
      "grad_norm": 0.791526734828949,
      "learning_rate": 4.811919531770231e-05,
      "loss": 2.4678,
      "step": 10740
    },
    {
      "epoch": 0.03785090947324151,
      "grad_norm": 0.7544797658920288,
      "learning_rate": 4.810868117618197e-05,
      "loss": 2.4595,
      "step": 10800
    },
    {
      "epoch": 0.038061192303648406,
      "grad_norm": 0.8361566066741943,
      "learning_rate": 4.809816703466162e-05,
      "loss": 2.4601,
      "step": 10860
    },
    {
      "epoch": 0.03827147513405531,
      "grad_norm": 0.7700713872909546,
      "learning_rate": 4.808765289314128e-05,
      "loss": 2.4651,
      "step": 10920
    },
    {
      "epoch": 0.0384817579644622,
      "grad_norm": 0.7866743803024292,
      "learning_rate": 4.807713875162093e-05,
      "loss": 2.4652,
      "step": 10980
    },
    {
      "epoch": 0.0386920407948691,
      "grad_norm": 0.7954710721969604,
      "learning_rate": 4.8066624610100583e-05,
      "loss": 2.4655,
      "step": 11040
    },
    {
      "epoch": 0.03890232362527599,
      "grad_norm": 0.8627223372459412,
      "learning_rate": 4.805611046858024e-05,
      "loss": 2.4638,
      "step": 11100
    },
    {
      "epoch": 0.039112606455682895,
      "grad_norm": 0.76466304063797,
      "learning_rate": 4.80455963270599e-05,
      "loss": 2.4467,
      "step": 11160
    },
    {
      "epoch": 0.03932288928608979,
      "grad_norm": 0.8471527099609375,
      "learning_rate": 4.803508218553955e-05,
      "loss": 2.4494,
      "step": 11220
    },
    {
      "epoch": 0.039533172116496686,
      "grad_norm": 0.7519867420196533,
      "learning_rate": 4.802456804401921e-05,
      "loss": 2.4583,
      "step": 11280
    },
    {
      "epoch": 0.03974345494690359,
      "grad_norm": 0.8605870008468628,
      "learning_rate": 4.801405390249886e-05,
      "loss": 2.4565,
      "step": 11340
    },
    {
      "epoch": 0.03995373777731048,
      "grad_norm": 0.7822230458259583,
      "learning_rate": 4.8003539760978515e-05,
      "loss": 2.4636,
      "step": 11400
    },
    {
      "epoch": 0.04016402060771738,
      "grad_norm": 0.7491495609283447,
      "learning_rate": 4.7993025619458174e-05,
      "loss": 2.4636,
      "step": 11460
    },
    {
      "epoch": 0.04037430343812428,
      "grad_norm": 0.81394362449646,
      "learning_rate": 4.798251147793783e-05,
      "loss": 2.4613,
      "step": 11520
    },
    {
      "epoch": 0.040584586268531175,
      "grad_norm": 0.8277611136436462,
      "learning_rate": 4.7971997336417484e-05,
      "loss": 2.445,
      "step": 11580
    },
    {
      "epoch": 0.04079486909893807,
      "grad_norm": 0.7311505675315857,
      "learning_rate": 4.7961483194897136e-05,
      "loss": 2.4439,
      "step": 11640
    },
    {
      "epoch": 0.041005151929344966,
      "grad_norm": 0.7802140712738037,
      "learning_rate": 4.7950969053376795e-05,
      "loss": 2.4629,
      "step": 11700
    },
    {
      "epoch": 0.04121543475975187,
      "grad_norm": 0.7538201212882996,
      "learning_rate": 4.794045491185645e-05,
      "loss": 2.4535,
      "step": 11760
    },
    {
      "epoch": 0.04142571759015876,
      "grad_norm": 0.7967063784599304,
      "learning_rate": 4.7929940770336105e-05,
      "loss": 2.4552,
      "step": 11820
    },
    {
      "epoch": 0.04163600042056566,
      "grad_norm": 0.819740891456604,
      "learning_rate": 4.7919426628815764e-05,
      "loss": 2.4601,
      "step": 11880
    },
    {
      "epoch": 0.04184628325097256,
      "grad_norm": 0.8193987607955933,
      "learning_rate": 4.7908912487295416e-05,
      "loss": 2.4774,
      "step": 11940
    },
    {
      "epoch": 0.042056566081379455,
      "grad_norm": 0.83516526222229,
      "learning_rate": 4.789839834577507e-05,
      "loss": 2.4546,
      "step": 12000
    },
    {
      "epoch": 0.04226684891178635,
      "grad_norm": 0.8305575847625732,
      "learning_rate": 4.7887884204254727e-05,
      "loss": 2.4476,
      "step": 12060
    },
    {
      "epoch": 0.04247713174219325,
      "grad_norm": 0.8162139654159546,
      "learning_rate": 4.787737006273438e-05,
      "loss": 2.4527,
      "step": 12120
    },
    {
      "epoch": 0.04268741457260015,
      "grad_norm": 0.8285430073738098,
      "learning_rate": 4.786685592121404e-05,
      "loss": 2.4485,
      "step": 12180
    },
    {
      "epoch": 0.04289769740300704,
      "grad_norm": 0.9468696713447571,
      "learning_rate": 4.785634177969369e-05,
      "loss": 2.4469,
      "step": 12240
    },
    {
      "epoch": 0.043107980233413945,
      "grad_norm": 0.7479498386383057,
      "learning_rate": 4.784582763817335e-05,
      "loss": 2.432,
      "step": 12300
    },
    {
      "epoch": 0.04331826306382084,
      "grad_norm": 0.7362015843391418,
      "learning_rate": 4.7835313496653e-05,
      "loss": 2.4626,
      "step": 12360
    },
    {
      "epoch": 0.043528545894227735,
      "grad_norm": 0.8287525177001953,
      "learning_rate": 4.782479935513265e-05,
      "loss": 2.456,
      "step": 12420
    },
    {
      "epoch": 0.04373882872463463,
      "grad_norm": 0.7658047080039978,
      "learning_rate": 4.781428521361231e-05,
      "loss": 2.4587,
      "step": 12480
    },
    {
      "epoch": 0.04394911155504153,
      "grad_norm": 0.7921282052993774,
      "learning_rate": 4.780377107209197e-05,
      "loss": 2.4617,
      "step": 12540
    },
    {
      "epoch": 0.04415939438544843,
      "grad_norm": 0.7566713690757751,
      "learning_rate": 4.779325693057162e-05,
      "loss": 2.4299,
      "step": 12600
    },
    {
      "epoch": 0.04436967721585532,
      "grad_norm": 0.7757437825202942,
      "learning_rate": 4.778274278905128e-05,
      "loss": 2.4452,
      "step": 12660
    },
    {
      "epoch": 0.044579960046262225,
      "grad_norm": 0.7873514294624329,
      "learning_rate": 4.777222864753093e-05,
      "loss": 2.4432,
      "step": 12720
    },
    {
      "epoch": 0.04479024287666912,
      "grad_norm": 0.7305336594581604,
      "learning_rate": 4.776171450601058e-05,
      "loss": 2.4468,
      "step": 12780
    },
    {
      "epoch": 0.045000525707076015,
      "grad_norm": 0.7629138231277466,
      "learning_rate": 4.775120036449024e-05,
      "loss": 2.4392,
      "step": 12840
    },
    {
      "epoch": 0.04521080853748292,
      "grad_norm": 0.7269495129585266,
      "learning_rate": 4.77406862229699e-05,
      "loss": 2.4201,
      "step": 12900
    },
    {
      "epoch": 0.04542109136788981,
      "grad_norm": 0.8234984278678894,
      "learning_rate": 4.773017208144955e-05,
      "loss": 2.4379,
      "step": 12960
    },
    {
      "epoch": 0.04563137419829671,
      "grad_norm": 0.7635879516601562,
      "learning_rate": 4.7719657939929204e-05,
      "loss": 2.4549,
      "step": 13020
    },
    {
      "epoch": 0.0458416570287036,
      "grad_norm": 0.8231070637702942,
      "learning_rate": 4.770914379840886e-05,
      "loss": 2.434,
      "step": 13080
    },
    {
      "epoch": 0.046051939859110505,
      "grad_norm": 0.7481459975242615,
      "learning_rate": 4.7698629656888515e-05,
      "loss": 2.4546,
      "step": 13140
    },
    {
      "epoch": 0.0462622226895174,
      "grad_norm": 0.7934229373931885,
      "learning_rate": 4.7688115515368174e-05,
      "loss": 2.4424,
      "step": 13200
    },
    {
      "epoch": 0.046472505519924295,
      "grad_norm": 0.7335939407348633,
      "learning_rate": 4.767760137384783e-05,
      "loss": 2.4549,
      "step": 13260
    },
    {
      "epoch": 0.0466827883503312,
      "grad_norm": 0.8233382105827332,
      "learning_rate": 4.7667087232327484e-05,
      "loss": 2.4531,
      "step": 13320
    },
    {
      "epoch": 0.04689307118073809,
      "grad_norm": 0.8387680053710938,
      "learning_rate": 4.7656573090807136e-05,
      "loss": 2.4555,
      "step": 13380
    },
    {
      "epoch": 0.04710335401114499,
      "grad_norm": 0.7505112886428833,
      "learning_rate": 4.7646058949286795e-05,
      "loss": 2.4447,
      "step": 13440
    },
    {
      "epoch": 0.04731363684155189,
      "grad_norm": 0.7546599507331848,
      "learning_rate": 4.7635544807766447e-05,
      "loss": 2.449,
      "step": 13500
    },
    {
      "epoch": 0.047523919671958785,
      "grad_norm": 0.7598650455474854,
      "learning_rate": 4.76250306662461e-05,
      "loss": 2.4502,
      "step": 13560
    },
    {
      "epoch": 0.04773420250236568,
      "grad_norm": 0.7494813203811646,
      "learning_rate": 4.7614516524725764e-05,
      "loss": 2.4391,
      "step": 13620
    },
    {
      "epoch": 0.04794448533277258,
      "grad_norm": 0.7718535661697388,
      "learning_rate": 4.7604002383205416e-05,
      "loss": 2.431,
      "step": 13680
    },
    {
      "epoch": 0.04815476816317948,
      "grad_norm": 0.6967277526855469,
      "learning_rate": 4.759348824168507e-05,
      "loss": 2.4261,
      "step": 13740
    },
    {
      "epoch": 0.04836505099358637,
      "grad_norm": 0.7197991013526917,
      "learning_rate": 4.7582974100164726e-05,
      "loss": 2.4463,
      "step": 13800
    },
    {
      "epoch": 0.04857533382399327,
      "grad_norm": 0.7507856488227844,
      "learning_rate": 4.757245995864438e-05,
      "loss": 2.4337,
      "step": 13860
    },
    {
      "epoch": 0.04878561665440017,
      "grad_norm": 0.7825746536254883,
      "learning_rate": 4.756194581712403e-05,
      "loss": 2.4446,
      "step": 13920
    },
    {
      "epoch": 0.048995899484807065,
      "grad_norm": 0.7825769782066345,
      "learning_rate": 4.755143167560369e-05,
      "loss": 2.441,
      "step": 13980
    },
    {
      "epoch": 0.04920618231521396,
      "grad_norm": 0.7822974324226379,
      "learning_rate": 4.754091753408335e-05,
      "loss": 2.4236,
      "step": 14040
    },
    {
      "epoch": 0.04941646514562086,
      "grad_norm": 0.7625136375427246,
      "learning_rate": 4.7530403392563e-05,
      "loss": 2.4392,
      "step": 14100
    },
    {
      "epoch": 0.04962674797602776,
      "grad_norm": 0.7651515007019043,
      "learning_rate": 4.751988925104265e-05,
      "loss": 2.4476,
      "step": 14160
    },
    {
      "epoch": 0.04983703080643465,
      "grad_norm": 0.6852614879608154,
      "learning_rate": 4.750937510952231e-05,
      "loss": 2.4496,
      "step": 14220
    },
    {
      "epoch": 0.050047313636841555,
      "grad_norm": 0.7274338006973267,
      "learning_rate": 4.749886096800196e-05,
      "loss": 2.4484,
      "step": 14280
    },
    {
      "epoch": 0.05025759646724845,
      "grad_norm": 0.7408413887023926,
      "learning_rate": 4.748834682648162e-05,
      "loss": 2.4413,
      "step": 14340
    },
    {
      "epoch": 0.050467879297655345,
      "grad_norm": 0.7533018589019775,
      "learning_rate": 4.747783268496128e-05,
      "loss": 2.4458,
      "step": 14400
    },
    {
      "epoch": 0.05067816212806225,
      "grad_norm": 0.7672397494316101,
      "learning_rate": 4.746731854344093e-05,
      "loss": 2.4379,
      "step": 14460
    },
    {
      "epoch": 0.05088844495846914,
      "grad_norm": 0.705220103263855,
      "learning_rate": 4.745680440192058e-05,
      "loss": 2.4361,
      "step": 14520
    },
    {
      "epoch": 0.05109872778887604,
      "grad_norm": 0.7222245335578918,
      "learning_rate": 4.744629026040024e-05,
      "loss": 2.4512,
      "step": 14580
    },
    {
      "epoch": 0.05130901061928293,
      "grad_norm": 0.7020009756088257,
      "learning_rate": 4.7435776118879893e-05,
      "loss": 2.4419,
      "step": 14640
    },
    {
      "epoch": 0.051519293449689835,
      "grad_norm": 0.7522382140159607,
      "learning_rate": 4.742526197735955e-05,
      "loss": 2.4284,
      "step": 14700
    },
    {
      "epoch": 0.05172957628009673,
      "grad_norm": 0.7863227128982544,
      "learning_rate": 4.7414923071531216e-05,
      "loss": 2.433,
      "step": 14760
    },
    {
      "epoch": 0.051939859110503625,
      "grad_norm": 0.7541500329971313,
      "learning_rate": 4.740440893001087e-05,
      "loss": 2.4364,
      "step": 14820
    },
    {
      "epoch": 0.05215014194091053,
      "grad_norm": 0.7301027178764343,
      "learning_rate": 4.739389478849052e-05,
      "loss": 2.4339,
      "step": 14880
    },
    {
      "epoch": 0.05236042477131742,
      "grad_norm": 0.7505426406860352,
      "learning_rate": 4.738338064697018e-05,
      "loss": 2.4269,
      "step": 14940
    },
    {
      "epoch": 0.05257070760172432,
      "grad_norm": 0.7464006543159485,
      "learning_rate": 4.737286650544983e-05,
      "loss": 2.4242,
      "step": 15000
    },
    {
      "epoch": 0.05278099043213122,
      "grad_norm": 0.7655802965164185,
      "learning_rate": 4.736235236392948e-05,
      "loss": 2.4396,
      "step": 15060
    },
    {
      "epoch": 0.052991273262538115,
      "grad_norm": 0.7519718408584595,
      "learning_rate": 4.735183822240915e-05,
      "loss": 2.4231,
      "step": 15120
    },
    {
      "epoch": 0.05320155609294501,
      "grad_norm": 0.8537678718566895,
      "learning_rate": 4.73413240808888e-05,
      "loss": 2.4434,
      "step": 15180
    },
    {
      "epoch": 0.053411838923351905,
      "grad_norm": 0.7247916460037231,
      "learning_rate": 4.733080993936845e-05,
      "loss": 2.4366,
      "step": 15240
    },
    {
      "epoch": 0.05362212175375881,
      "grad_norm": 0.7293427586555481,
      "learning_rate": 4.732029579784811e-05,
      "loss": 2.4335,
      "step": 15300
    },
    {
      "epoch": 0.0538324045841657,
      "grad_norm": 0.7361655831336975,
      "learning_rate": 4.730978165632776e-05,
      "loss": 2.4303,
      "step": 15360
    },
    {
      "epoch": 0.0540426874145726,
      "grad_norm": 0.7800238132476807,
      "learning_rate": 4.7299267514807414e-05,
      "loss": 2.4291,
      "step": 15420
    },
    {
      "epoch": 0.0542529702449795,
      "grad_norm": 0.7535472512245178,
      "learning_rate": 4.728875337328707e-05,
      "loss": 2.4263,
      "step": 15480
    },
    {
      "epoch": 0.054463253075386395,
      "grad_norm": 0.7635684609413147,
      "learning_rate": 4.727823923176673e-05,
      "loss": 2.4356,
      "step": 15540
    },
    {
      "epoch": 0.05467353590579329,
      "grad_norm": 0.7387227416038513,
      "learning_rate": 4.726772509024638e-05,
      "loss": 2.4258,
      "step": 15600
    },
    {
      "epoch": 0.05488381873620019,
      "grad_norm": 0.7634516358375549,
      "learning_rate": 4.7257210948726035e-05,
      "loss": 2.4369,
      "step": 15660
    },
    {
      "epoch": 0.05509410156660709,
      "grad_norm": 0.7039297223091125,
      "learning_rate": 4.7246696807205694e-05,
      "loss": 2.4319,
      "step": 15720
    },
    {
      "epoch": 0.05530438439701398,
      "grad_norm": 0.7102476954460144,
      "learning_rate": 4.7236182665685346e-05,
      "loss": 2.4395,
      "step": 15780
    },
    {
      "epoch": 0.055514667227420884,
      "grad_norm": 0.7064028382301331,
      "learning_rate": 4.7225668524165004e-05,
      "loss": 2.4184,
      "step": 15840
    },
    {
      "epoch": 0.05572495005782778,
      "grad_norm": 0.7453619241714478,
      "learning_rate": 4.721515438264466e-05,
      "loss": 2.4262,
      "step": 15900
    },
    {
      "epoch": 0.055935232888234675,
      "grad_norm": 0.7171881198883057,
      "learning_rate": 4.7204640241124315e-05,
      "loss": 2.4381,
      "step": 15960
    },
    {
      "epoch": 0.05614551571864157,
      "grad_norm": 0.6928536891937256,
      "learning_rate": 4.719412609960397e-05,
      "loss": 2.4174,
      "step": 16020
    },
    {
      "epoch": 0.05635579854904847,
      "grad_norm": 0.7249700427055359,
      "learning_rate": 4.7183611958083625e-05,
      "loss": 2.4245,
      "step": 16080
    },
    {
      "epoch": 0.05656608137945537,
      "grad_norm": 0.7394696474075317,
      "learning_rate": 4.717309781656328e-05,
      "loss": 2.4238,
      "step": 16140
    },
    {
      "epoch": 0.05677636420986226,
      "grad_norm": 0.7761276364326477,
      "learning_rate": 4.7162583675042936e-05,
      "loss": 2.4203,
      "step": 16200
    },
    {
      "epoch": 0.056986647040269164,
      "grad_norm": 0.8082882761955261,
      "learning_rate": 4.7152069533522595e-05,
      "loss": 2.4302,
      "step": 16260
    },
    {
      "epoch": 0.05719692987067606,
      "grad_norm": 0.7704823613166809,
      "learning_rate": 4.7141555392002247e-05,
      "loss": 2.4378,
      "step": 16320
    },
    {
      "epoch": 0.057407212701082955,
      "grad_norm": 0.7104964256286621,
      "learning_rate": 4.71310412504819e-05,
      "loss": 2.4272,
      "step": 16380
    },
    {
      "epoch": 0.05761749553148986,
      "grad_norm": 0.8135157823562622,
      "learning_rate": 4.712070234465356e-05,
      "loss": 2.4248,
      "step": 16440
    },
    {
      "epoch": 0.05782777836189675,
      "grad_norm": 0.7562858462333679,
      "learning_rate": 4.7110188203133214e-05,
      "loss": 2.4342,
      "step": 16500
    },
    {
      "epoch": 0.05803806119230365,
      "grad_norm": 0.7549763917922974,
      "learning_rate": 4.709967406161287e-05,
      "loss": 2.4231,
      "step": 16560
    },
    {
      "epoch": 0.05824834402271054,
      "grad_norm": 0.7127776741981506,
      "learning_rate": 4.708915992009253e-05,
      "loss": 2.4152,
      "step": 16620
    },
    {
      "epoch": 0.058458626853117444,
      "grad_norm": 0.7201217412948608,
      "learning_rate": 4.7078645778572183e-05,
      "loss": 2.4197,
      "step": 16680
    },
    {
      "epoch": 0.05866890968352434,
      "grad_norm": 0.7064502835273743,
      "learning_rate": 4.7068131637051835e-05,
      "loss": 2.4165,
      "step": 16740
    },
    {
      "epoch": 0.058879192513931235,
      "grad_norm": 0.7064214944839478,
      "learning_rate": 4.7057617495531494e-05,
      "loss": 2.4084,
      "step": 16800
    },
    {
      "epoch": 0.05908947534433814,
      "grad_norm": 0.6829055547714233,
      "learning_rate": 4.7047103354011146e-05,
      "loss": 2.4121,
      "step": 16860
    },
    {
      "epoch": 0.05929975817474503,
      "grad_norm": 0.7540745735168457,
      "learning_rate": 4.70365892124908e-05,
      "loss": 2.4226,
      "step": 16920
    },
    {
      "epoch": 0.05951004100515193,
      "grad_norm": 0.7283081412315369,
      "learning_rate": 4.702607507097046e-05,
      "loss": 2.4147,
      "step": 16980
    },
    {
      "epoch": 0.05972032383555883,
      "grad_norm": 0.7194704413414001,
      "learning_rate": 4.7015560929450115e-05,
      "loss": 2.4159,
      "step": 17040
    },
    {
      "epoch": 0.059930606665965724,
      "grad_norm": 0.7521179914474487,
      "learning_rate": 4.700504678792977e-05,
      "loss": 2.433,
      "step": 17100
    },
    {
      "epoch": 0.06014088949637262,
      "grad_norm": 0.7045489549636841,
      "learning_rate": 4.6994532646409426e-05,
      "loss": 2.4229,
      "step": 17160
    },
    {
      "epoch": 0.06035117232677952,
      "grad_norm": 0.7060059309005737,
      "learning_rate": 4.698401850488908e-05,
      "loss": 2.4234,
      "step": 17220
    },
    {
      "epoch": 0.06056145515718642,
      "grad_norm": 0.6854456663131714,
      "learning_rate": 4.697350436336873e-05,
      "loss": 2.4157,
      "step": 17280
    },
    {
      "epoch": 0.06077173798759331,
      "grad_norm": 0.7461974620819092,
      "learning_rate": 4.696299022184839e-05,
      "loss": 2.4253,
      "step": 17340
    },
    {
      "epoch": 0.06098202081800021,
      "grad_norm": 0.7423505783081055,
      "learning_rate": 4.695247608032805e-05,
      "loss": 2.411,
      "step": 17400
    },
    {
      "epoch": 0.06119230364840711,
      "grad_norm": 0.7172077298164368,
      "learning_rate": 4.69419619388077e-05,
      "loss": 2.4272,
      "step": 17460
    },
    {
      "epoch": 0.061402586478814004,
      "grad_norm": 0.7238907217979431,
      "learning_rate": 4.693144779728735e-05,
      "loss": 2.4395,
      "step": 17520
    },
    {
      "epoch": 0.0616128693092209,
      "grad_norm": 0.7185387015342712,
      "learning_rate": 4.692093365576701e-05,
      "loss": 2.424,
      "step": 17580
    },
    {
      "epoch": 0.0618231521396278,
      "grad_norm": 0.6714202165603638,
      "learning_rate": 4.691041951424666e-05,
      "loss": 2.396,
      "step": 17640
    },
    {
      "epoch": 0.0620334349700347,
      "grad_norm": 0.7396364212036133,
      "learning_rate": 4.689990537272632e-05,
      "loss": 2.4254,
      "step": 17700
    },
    {
      "epoch": 0.06224371780044159,
      "grad_norm": 0.6906604170799255,
      "learning_rate": 4.688939123120598e-05,
      "loss": 2.4044,
      "step": 17760
    },
    {
      "epoch": 0.062454000630848494,
      "grad_norm": 0.6801305413246155,
      "learning_rate": 4.687887708968563e-05,
      "loss": 2.4268,
      "step": 17820
    },
    {
      "epoch": 0.06266428346125538,
      "grad_norm": 0.7206404805183411,
      "learning_rate": 4.686836294816528e-05,
      "loss": 2.4268,
      "step": 17880
    },
    {
      "epoch": 0.06287456629166228,
      "grad_norm": 0.6780009865760803,
      "learning_rate": 4.685784880664494e-05,
      "loss": 2.4236,
      "step": 17940
    },
    {
      "epoch": 0.06308484912206919,
      "grad_norm": 0.7205548882484436,
      "learning_rate": 4.684733466512459e-05,
      "loss": 2.4329,
      "step": 18000
    },
    {
      "epoch": 0.06329513195247607,
      "grad_norm": 0.7508593797683716,
      "learning_rate": 4.683682052360425e-05,
      "loss": 2.4133,
      "step": 18060
    },
    {
      "epoch": 0.06350541478288298,
      "grad_norm": 0.6937260031700134,
      "learning_rate": 4.6826306382083903e-05,
      "loss": 2.4156,
      "step": 18120
    },
    {
      "epoch": 0.06371569761328988,
      "grad_norm": 0.7223955392837524,
      "learning_rate": 4.681579224056356e-05,
      "loss": 2.4226,
      "step": 18180
    },
    {
      "epoch": 0.06392598044369677,
      "grad_norm": 0.7173060774803162,
      "learning_rate": 4.6805278099043214e-05,
      "loss": 2.4309,
      "step": 18240
    },
    {
      "epoch": 0.06413626327410367,
      "grad_norm": 0.7185004949569702,
      "learning_rate": 4.6794763957522866e-05,
      "loss": 2.4121,
      "step": 18300
    },
    {
      "epoch": 0.06434654610451057,
      "grad_norm": 0.7266471982002258,
      "learning_rate": 4.6784249816002525e-05,
      "loss": 2.4137,
      "step": 18360
    },
    {
      "epoch": 0.06455682893491746,
      "grad_norm": 0.6976909637451172,
      "learning_rate": 4.677373567448218e-05,
      "loss": 2.4061,
      "step": 18420
    },
    {
      "epoch": 0.06476711176532436,
      "grad_norm": 0.6547685861587524,
      "learning_rate": 4.6763221532961835e-05,
      "loss": 2.4076,
      "step": 18480
    },
    {
      "epoch": 0.06497739459573126,
      "grad_norm": 0.7449997067451477,
      "learning_rate": 4.6752707391441494e-05,
      "loss": 2.4172,
      "step": 18540
    },
    {
      "epoch": 0.06518767742613815,
      "grad_norm": 0.7313099503517151,
      "learning_rate": 4.6742193249921146e-05,
      "loss": 2.4074,
      "step": 18600
    },
    {
      "epoch": 0.06539796025654505,
      "grad_norm": 0.6708427667617798,
      "learning_rate": 4.67316791084008e-05,
      "loss": 2.3998,
      "step": 18660
    },
    {
      "epoch": 0.06560824308695196,
      "grad_norm": 0.7288368940353394,
      "learning_rate": 4.6721164966880456e-05,
      "loss": 2.4041,
      "step": 18720
    },
    {
      "epoch": 0.06581852591735884,
      "grad_norm": 0.6853367686271667,
      "learning_rate": 4.6710650825360115e-05,
      "loss": 2.3981,
      "step": 18780
    },
    {
      "epoch": 0.06602880874776575,
      "grad_norm": 0.7069407105445862,
      "learning_rate": 4.670013668383977e-05,
      "loss": 2.4008,
      "step": 18840
    },
    {
      "epoch": 0.06623909157817265,
      "grad_norm": 0.7321661114692688,
      "learning_rate": 4.6689622542319425e-05,
      "loss": 2.3965,
      "step": 18900
    },
    {
      "epoch": 0.06644937440857954,
      "grad_norm": 0.6575349569320679,
      "learning_rate": 4.667910840079908e-05,
      "loss": 2.4077,
      "step": 18960
    },
    {
      "epoch": 0.06665965723898644,
      "grad_norm": 0.6651709675788879,
      "learning_rate": 4.666859425927873e-05,
      "loss": 2.4041,
      "step": 19020
    },
    {
      "epoch": 0.06686994006939333,
      "grad_norm": 0.7022081613540649,
      "learning_rate": 4.665808011775839e-05,
      "loss": 2.4001,
      "step": 19080
    },
    {
      "epoch": 0.06708022289980023,
      "grad_norm": 0.67169189453125,
      "learning_rate": 4.6647565976238047e-05,
      "loss": 2.4178,
      "step": 19140
    },
    {
      "epoch": 0.06729050573020713,
      "grad_norm": 0.6954366564750671,
      "learning_rate": 4.66370518347177e-05,
      "loss": 2.4271,
      "step": 19200
    },
    {
      "epoch": 0.06750078856061402,
      "grad_norm": 0.6861891150474548,
      "learning_rate": 4.662653769319735e-05,
      "loss": 2.4049,
      "step": 19260
    },
    {
      "epoch": 0.06771107139102092,
      "grad_norm": 0.7052733898162842,
      "learning_rate": 4.661602355167701e-05,
      "loss": 2.4011,
      "step": 19320
    },
    {
      "epoch": 0.06792135422142782,
      "grad_norm": 0.6501637697219849,
      "learning_rate": 4.660550941015666e-05,
      "loss": 2.4239,
      "step": 19380
    },
    {
      "epoch": 0.06813163705183471,
      "grad_norm": 0.6948041319847107,
      "learning_rate": 4.659499526863631e-05,
      "loss": 2.4054,
      "step": 19440
    },
    {
      "epoch": 0.06834191988224161,
      "grad_norm": 0.7107264995574951,
      "learning_rate": 4.658448112711598e-05,
      "loss": 2.4261,
      "step": 19500
    },
    {
      "epoch": 0.06855220271264852,
      "grad_norm": 0.7571120262145996,
      "learning_rate": 4.657396698559563e-05,
      "loss": 2.4051,
      "step": 19560
    },
    {
      "epoch": 0.0687624855430554,
      "grad_norm": 0.7028728127479553,
      "learning_rate": 4.656345284407528e-05,
      "loss": 2.4063,
      "step": 19620
    },
    {
      "epoch": 0.0689727683734623,
      "grad_norm": 0.699620246887207,
      "learning_rate": 4.655293870255494e-05,
      "loss": 2.4099,
      "step": 19680
    },
    {
      "epoch": 0.06918305120386921,
      "grad_norm": 0.6919541358947754,
      "learning_rate": 4.654242456103459e-05,
      "loss": 2.4114,
      "step": 19740
    },
    {
      "epoch": 0.0693933340342761,
      "grad_norm": 0.696038544178009,
      "learning_rate": 4.6531910419514244e-05,
      "loss": 2.4016,
      "step": 19800
    },
    {
      "epoch": 0.069603616864683,
      "grad_norm": 0.6753870248794556,
      "learning_rate": 4.65213962779939e-05,
      "loss": 2.3975,
      "step": 19860
    },
    {
      "epoch": 0.0698138996950899,
      "grad_norm": 0.719841718673706,
      "learning_rate": 4.651088213647356e-05,
      "loss": 2.4118,
      "step": 19920
    },
    {
      "epoch": 0.07002418252549679,
      "grad_norm": 0.6914800405502319,
      "learning_rate": 4.6500367994953214e-05,
      "loss": 2.4107,
      "step": 19980
    },
    {
      "epoch": 0.07023446535590369,
      "grad_norm": 0.6995266675949097,
      "learning_rate": 4.6489853853432866e-05,
      "loss": 2.4027,
      "step": 20040
    },
    {
      "epoch": 0.0704447481863106,
      "grad_norm": 0.7216110229492188,
      "learning_rate": 4.6479339711912524e-05,
      "loss": 2.4194,
      "step": 20100
    },
    {
      "epoch": 0.07065503101671748,
      "grad_norm": 0.6979327201843262,
      "learning_rate": 4.6468825570392176e-05,
      "loss": 2.3914,
      "step": 20160
    },
    {
      "epoch": 0.07086531384712438,
      "grad_norm": 0.6791154742240906,
      "learning_rate": 4.6458311428871835e-05,
      "loss": 2.4015,
      "step": 20220
    },
    {
      "epoch": 0.07107559667753129,
      "grad_norm": 0.6936553120613098,
      "learning_rate": 4.6447797287351493e-05,
      "loss": 2.4082,
      "step": 20280
    },
    {
      "epoch": 0.07128587950793817,
      "grad_norm": 0.6697074770927429,
      "learning_rate": 4.6437283145831145e-05,
      "loss": 2.408,
      "step": 20340
    },
    {
      "epoch": 0.07149616233834508,
      "grad_norm": 0.681769609451294,
      "learning_rate": 4.64267690043108e-05,
      "loss": 2.3901,
      "step": 20400
    },
    {
      "epoch": 0.07170644516875198,
      "grad_norm": 0.6847583055496216,
      "learning_rate": 4.6416254862790456e-05,
      "loss": 2.3975,
      "step": 20460
    },
    {
      "epoch": 0.07191672799915887,
      "grad_norm": 0.6366531848907471,
      "learning_rate": 4.640574072127011e-05,
      "loss": 2.4028,
      "step": 20520
    },
    {
      "epoch": 0.07212701082956577,
      "grad_norm": 0.6561765074729919,
      "learning_rate": 4.6395226579749766e-05,
      "loss": 2.4038,
      "step": 20580
    },
    {
      "epoch": 0.07233729365997266,
      "grad_norm": 0.7223414182662964,
      "learning_rate": 4.6384712438229425e-05,
      "loss": 2.4159,
      "step": 20640
    },
    {
      "epoch": 0.07254757649037956,
      "grad_norm": 0.7201683521270752,
      "learning_rate": 4.637419829670908e-05,
      "loss": 2.4054,
      "step": 20700
    },
    {
      "epoch": 0.07275785932078646,
      "grad_norm": 0.661088764667511,
      "learning_rate": 4.636368415518873e-05,
      "loss": 2.398,
      "step": 20760
    },
    {
      "epoch": 0.07296814215119335,
      "grad_norm": 0.6699717044830322,
      "learning_rate": 4.635317001366839e-05,
      "loss": 2.4058,
      "step": 20820
    },
    {
      "epoch": 0.07317842498160025,
      "grad_norm": 0.680035412311554,
      "learning_rate": 4.634265587214804e-05,
      "loss": 2.412,
      "step": 20880
    },
    {
      "epoch": 0.07338870781200715,
      "grad_norm": 0.6644302606582642,
      "learning_rate": 4.63321417306277e-05,
      "loss": 2.4015,
      "step": 20940
    },
    {
      "epoch": 0.07359899064241404,
      "grad_norm": 0.6539902091026306,
      "learning_rate": 4.632162758910735e-05,
      "loss": 2.3749,
      "step": 21000
    },
    {
      "epoch": 0.07380927347282094,
      "grad_norm": 0.6808243989944458,
      "learning_rate": 4.631111344758701e-05,
      "loss": 2.3918,
      "step": 21060
    },
    {
      "epoch": 0.07401955630322785,
      "grad_norm": 0.7031143307685852,
      "learning_rate": 4.630059930606666e-05,
      "loss": 2.4004,
      "step": 21120
    },
    {
      "epoch": 0.07422983913363473,
      "grad_norm": 0.6713913083076477,
      "learning_rate": 4.629008516454631e-05,
      "loss": 2.4042,
      "step": 21180
    },
    {
      "epoch": 0.07444012196404164,
      "grad_norm": 0.7224510312080383,
      "learning_rate": 4.627957102302598e-05,
      "loss": 2.4053,
      "step": 21240
    },
    {
      "epoch": 0.07465040479444854,
      "grad_norm": 0.7205802798271179,
      "learning_rate": 4.626905688150563e-05,
      "loss": 2.4049,
      "step": 21300
    },
    {
      "epoch": 0.07486068762485543,
      "grad_norm": 0.6812241077423096,
      "learning_rate": 4.625854273998528e-05,
      "loss": 2.3989,
      "step": 21360
    },
    {
      "epoch": 0.07507097045526233,
      "grad_norm": 0.699661374092102,
      "learning_rate": 4.624802859846494e-05,
      "loss": 2.388,
      "step": 21420
    },
    {
      "epoch": 0.07528125328566923,
      "grad_norm": 0.6710296869277954,
      "learning_rate": 4.623751445694459e-05,
      "loss": 2.3954,
      "step": 21480
    },
    {
      "epoch": 0.07549153611607612,
      "grad_norm": 0.6275144219398499,
      "learning_rate": 4.6227000315424244e-05,
      "loss": 2.3954,
      "step": 21540
    },
    {
      "epoch": 0.07570181894648302,
      "grad_norm": 0.7540141344070435,
      "learning_rate": 4.62164861739039e-05,
      "loss": 2.4054,
      "step": 21600
    },
    {
      "epoch": 0.07591210177688992,
      "grad_norm": Infinity,
      "learning_rate": 4.620614726807556e-05,
      "loss": 2.3994,
      "step": 21660
    },
    {
      "epoch": 0.07612238460729681,
      "grad_norm": 0.6901912093162537,
      "learning_rate": 4.619563312655522e-05,
      "loss": 2.3948,
      "step": 21720
    },
    {
      "epoch": 0.07633266743770371,
      "grad_norm": 0.6476166844367981,
      "learning_rate": 4.618511898503488e-05,
      "loss": 2.3979,
      "step": 21780
    },
    {
      "epoch": 0.07654295026811062,
      "grad_norm": 0.7146350741386414,
      "learning_rate": 4.617460484351453e-05,
      "loss": 2.3922,
      "step": 21840
    },
    {
      "epoch": 0.0767532330985175,
      "grad_norm": 0.6991199851036072,
      "learning_rate": 4.616409070199418e-05,
      "loss": 2.3935,
      "step": 21900
    },
    {
      "epoch": 0.0769635159289244,
      "grad_norm": 0.6704103946685791,
      "learning_rate": 4.615357656047384e-05,
      "loss": 2.3735,
      "step": 21960
    },
    {
      "epoch": 0.0771737987593313,
      "grad_norm": 0.6334531903266907,
      "learning_rate": 4.614306241895349e-05,
      "loss": 2.3881,
      "step": 22020
    },
    {
      "epoch": 0.0773840815897382,
      "grad_norm": 0.642454981803894,
      "learning_rate": 4.613254827743315e-05,
      "loss": 2.4056,
      "step": 22080
    },
    {
      "epoch": 0.0775943644201451,
      "grad_norm": 0.6881276965141296,
      "learning_rate": 4.612203413591281e-05,
      "loss": 2.398,
      "step": 22140
    },
    {
      "epoch": 0.07780464725055199,
      "grad_norm": 0.6868687272071838,
      "learning_rate": 4.611151999439246e-05,
      "loss": 2.392,
      "step": 22200
    },
    {
      "epoch": 0.07801493008095889,
      "grad_norm": 0.6720156073570251,
      "learning_rate": 4.610100585287211e-05,
      "loss": 2.4043,
      "step": 22260
    },
    {
      "epoch": 0.07822521291136579,
      "grad_norm": 0.679771900177002,
      "learning_rate": 4.609049171135177e-05,
      "loss": 2.3936,
      "step": 22320
    },
    {
      "epoch": 0.07843549574177268,
      "grad_norm": 0.7134901285171509,
      "learning_rate": 4.607997756983142e-05,
      "loss": 2.3959,
      "step": 22380
    },
    {
      "epoch": 0.07864577857217958,
      "grad_norm": 0.7048931121826172,
      "learning_rate": 4.606946342831108e-05,
      "loss": 2.3805,
      "step": 22440
    },
    {
      "epoch": 0.07885606140258648,
      "grad_norm": 0.7281864285469055,
      "learning_rate": 4.6058949286790734e-05,
      "loss": 2.3867,
      "step": 22500
    },
    {
      "epoch": 0.07906634423299337,
      "grad_norm": 0.6903355121612549,
      "learning_rate": 4.604843514527039e-05,
      "loss": 2.3922,
      "step": 22560
    },
    {
      "epoch": 0.07927662706340027,
      "grad_norm": 0.6869490146636963,
      "learning_rate": 4.6037921003750044e-05,
      "loss": 2.3956,
      "step": 22620
    },
    {
      "epoch": 0.07948690989380718,
      "grad_norm": 0.6450217962265015,
      "learning_rate": 4.6027406862229696e-05,
      "loss": 2.399,
      "step": 22680
    },
    {
      "epoch": 0.07969719272421406,
      "grad_norm": 0.6367782354354858,
      "learning_rate": 4.6016892720709355e-05,
      "loss": 2.3827,
      "step": 22740
    },
    {
      "epoch": 0.07990747555462097,
      "grad_norm": 0.6485390067100525,
      "learning_rate": 4.6006378579189014e-05,
      "loss": 2.3938,
      "step": 22800
    },
    {
      "epoch": 0.08011775838502787,
      "grad_norm": 0.7188836932182312,
      "learning_rate": 4.5995864437668666e-05,
      "loss": 2.3812,
      "step": 22860
    },
    {
      "epoch": 0.08032804121543476,
      "grad_norm": 0.67445969581604,
      "learning_rate": 4.5985350296148324e-05,
      "loss": 2.3891,
      "step": 22920
    },
    {
      "epoch": 0.08053832404584166,
      "grad_norm": 0.7052184343338013,
      "learning_rate": 4.5974836154627976e-05,
      "loss": 2.3974,
      "step": 22980
    },
    {
      "epoch": 0.08074860687624856,
      "grad_norm": 0.6907039284706116,
      "learning_rate": 4.596432201310763e-05,
      "loss": 2.386,
      "step": 23040
    },
    {
      "epoch": 0.08095888970665545,
      "grad_norm": 0.7139337062835693,
      "learning_rate": 4.595380787158729e-05,
      "loss": 2.3803,
      "step": 23100
    },
    {
      "epoch": 0.08116917253706235,
      "grad_norm": 0.684907853603363,
      "learning_rate": 4.5943293730066945e-05,
      "loss": 2.3946,
      "step": 23160
    },
    {
      "epoch": 0.08137945536746925,
      "grad_norm": 0.6158885955810547,
      "learning_rate": 4.59327795885466e-05,
      "loss": 2.4151,
      "step": 23220
    },
    {
      "epoch": 0.08158973819787614,
      "grad_norm": 0.701888382434845,
      "learning_rate": 4.592226544702625e-05,
      "loss": 2.3844,
      "step": 23280
    },
    {
      "epoch": 0.08180002102828304,
      "grad_norm": 0.668491542339325,
      "learning_rate": 4.591175130550591e-05,
      "loss": 2.3991,
      "step": 23340
    },
    {
      "epoch": 0.08201030385868993,
      "grad_norm": 0.7079835534095764,
      "learning_rate": 4.590123716398556e-05,
      "loss": 2.386,
      "step": 23400
    },
    {
      "epoch": 0.08222058668909683,
      "grad_norm": 0.725784957408905,
      "learning_rate": 4.589072302246522e-05,
      "loss": 2.3815,
      "step": 23460
    },
    {
      "epoch": 0.08243086951950374,
      "grad_norm": 0.6603190302848816,
      "learning_rate": 4.588020888094488e-05,
      "loss": 2.3864,
      "step": 23520
    },
    {
      "epoch": 0.08264115234991062,
      "grad_norm": 0.7041152715682983,
      "learning_rate": 4.586969473942453e-05,
      "loss": 2.3903,
      "step": 23580
    },
    {
      "epoch": 0.08285143518031753,
      "grad_norm": 0.6685366034507751,
      "learning_rate": 4.585918059790418e-05,
      "loss": 2.3856,
      "step": 23640
    },
    {
      "epoch": 0.08306171801072443,
      "grad_norm": 0.6484189033508301,
      "learning_rate": 4.584866645638384e-05,
      "loss": 2.3873,
      "step": 23700
    },
    {
      "epoch": 0.08327200084113132,
      "grad_norm": 0.6959061026573181,
      "learning_rate": 4.58383275505555e-05,
      "loss": 2.3866,
      "step": 23760
    },
    {
      "epoch": 0.08348228367153822,
      "grad_norm": 0.6646124720573425,
      "learning_rate": 4.5827813409035155e-05,
      "loss": 2.3863,
      "step": 23820
    },
    {
      "epoch": 0.08369256650194512,
      "grad_norm": 0.6963408589363098,
      "learning_rate": 4.581729926751481e-05,
      "loss": 2.3821,
      "step": 23880
    },
    {
      "epoch": 0.08390284933235201,
      "grad_norm": 0.667786717414856,
      "learning_rate": 4.5806785125994466e-05,
      "loss": 2.393,
      "step": 23940
    },
    {
      "epoch": 0.08411313216275891,
      "grad_norm": 0.6659138202667236,
      "learning_rate": 4.579627098447412e-05,
      "loss": 2.3862,
      "step": 24000
    },
    {
      "epoch": 0.08432341499316581,
      "grad_norm": 0.7164613008499146,
      "learning_rate": 4.5785756842953776e-05,
      "loss": 2.3981,
      "step": 24060
    },
    {
      "epoch": 0.0845336978235727,
      "grad_norm": 0.6524563431739807,
      "learning_rate": 4.577524270143343e-05,
      "loss": 2.395,
      "step": 24120
    },
    {
      "epoch": 0.0847439806539796,
      "grad_norm": 1.5108562707901,
      "learning_rate": 4.576472855991309e-05,
      "loss": 2.3708,
      "step": 24180
    },
    {
      "epoch": 0.0849542634843865,
      "grad_norm": 0.641019344329834,
      "learning_rate": 4.575421441839274e-05,
      "loss": 2.3911,
      "step": 24240
    },
    {
      "epoch": 0.0851645463147934,
      "grad_norm": 0.7017983198165894,
      "learning_rate": 4.57437002768724e-05,
      "loss": 2.3829,
      "step": 24300
    },
    {
      "epoch": 0.0853748291452003,
      "grad_norm": 0.6863337755203247,
      "learning_rate": 4.573318613535205e-05,
      "loss": 2.3787,
      "step": 24360
    },
    {
      "epoch": 0.0855851119756072,
      "grad_norm": 0.623172402381897,
      "learning_rate": 4.572267199383171e-05,
      "loss": 2.3866,
      "step": 24420
    },
    {
      "epoch": 0.08579539480601409,
      "grad_norm": 0.6611483693122864,
      "learning_rate": 4.571215785231136e-05,
      "loss": 2.3928,
      "step": 24480
    },
    {
      "epoch": 0.08600567763642099,
      "grad_norm": 0.6451998353004456,
      "learning_rate": 4.570164371079101e-05,
      "loss": 2.3727,
      "step": 24540
    },
    {
      "epoch": 0.08621596046682789,
      "grad_norm": 0.6573619246482849,
      "learning_rate": 4.569112956927067e-05,
      "loss": 2.3647,
      "step": 24600
    },
    {
      "epoch": 0.08642624329723478,
      "grad_norm": 0.67535001039505,
      "learning_rate": 4.568061542775033e-05,
      "loss": 2.3882,
      "step": 24660
    },
    {
      "epoch": 0.08663652612764168,
      "grad_norm": 0.6458430290222168,
      "learning_rate": 4.567010128622998e-05,
      "loss": 2.393,
      "step": 24720
    },
    {
      "epoch": 0.08684680895804857,
      "grad_norm": 0.6228342652320862,
      "learning_rate": 4.565958714470964e-05,
      "loss": 2.3856,
      "step": 24780
    },
    {
      "epoch": 0.08705709178845547,
      "grad_norm": 0.6460654735565186,
      "learning_rate": 4.564907300318929e-05,
      "loss": 2.3851,
      "step": 24840
    },
    {
      "epoch": 0.08726737461886237,
      "grad_norm": 0.694816529750824,
      "learning_rate": 4.5638558861668944e-05,
      "loss": 2.3789,
      "step": 24900
    },
    {
      "epoch": 0.08747765744926926,
      "grad_norm": 0.7187425494194031,
      "learning_rate": 4.56280447201486e-05,
      "loss": 2.375,
      "step": 24960
    },
    {
      "epoch": 0.08768794027967616,
      "grad_norm": 0.6361470222473145,
      "learning_rate": 4.561753057862826e-05,
      "loss": 2.3805,
      "step": 25020
    },
    {
      "epoch": 0.08789822311008307,
      "grad_norm": 0.6624230146408081,
      "learning_rate": 4.560701643710791e-05,
      "loss": 2.385,
      "step": 25080
    },
    {
      "epoch": 0.08810850594048995,
      "grad_norm": 0.6224386096000671,
      "learning_rate": 4.5596502295587565e-05,
      "loss": 2.3751,
      "step": 25140
    },
    {
      "epoch": 0.08831878877089686,
      "grad_norm": 0.6870352029800415,
      "learning_rate": 4.558598815406722e-05,
      "loss": 2.3736,
      "step": 25200
    },
    {
      "epoch": 0.08852907160130376,
      "grad_norm": 0.6708427667617798,
      "learning_rate": 4.5575474012546875e-05,
      "loss": 2.405,
      "step": 25260
    },
    {
      "epoch": 0.08873935443171065,
      "grad_norm": 0.6544594168663025,
      "learning_rate": 4.556495987102653e-05,
      "loss": 2.3976,
      "step": 25320
    },
    {
      "epoch": 0.08894963726211755,
      "grad_norm": 0.6882736086845398,
      "learning_rate": 4.555444572950619e-05,
      "loss": 2.393,
      "step": 25380
    },
    {
      "epoch": 0.08915992009252445,
      "grad_norm": 0.6833152770996094,
      "learning_rate": 4.5543931587985844e-05,
      "loss": 2.3912,
      "step": 25440
    },
    {
      "epoch": 0.08937020292293134,
      "grad_norm": 0.6555745005607605,
      "learning_rate": 4.5533417446465496e-05,
      "loss": 2.386,
      "step": 25500
    },
    {
      "epoch": 0.08958048575333824,
      "grad_norm": 0.6967830657958984,
      "learning_rate": 4.5522903304945155e-05,
      "loss": 2.3719,
      "step": 25560
    },
    {
      "epoch": 0.08979076858374514,
      "grad_norm": 0.6678874492645264,
      "learning_rate": 4.551238916342481e-05,
      "loss": 2.3911,
      "step": 25620
    },
    {
      "epoch": 0.09000105141415203,
      "grad_norm": 0.6675252914428711,
      "learning_rate": 4.550187502190446e-05,
      "loss": 2.3802,
      "step": 25680
    },
    {
      "epoch": 0.09021133424455893,
      "grad_norm": 0.6637398600578308,
      "learning_rate": 4.549136088038412e-05,
      "loss": 2.3885,
      "step": 25740
    },
    {
      "epoch": 0.09042161707496583,
      "grad_norm": 0.6243078112602234,
      "learning_rate": 4.5480846738863776e-05,
      "loss": 2.3837,
      "step": 25800
    },
    {
      "epoch": 0.09063189990537272,
      "grad_norm": 0.7010598182678223,
      "learning_rate": 4.547033259734343e-05,
      "loss": 2.3577,
      "step": 25860
    },
    {
      "epoch": 0.09084218273577963,
      "grad_norm": 0.6444143652915955,
      "learning_rate": 4.545981845582308e-05,
      "loss": 2.3873,
      "step": 25920
    },
    {
      "epoch": 0.09105246556618653,
      "grad_norm": 0.6662720441818237,
      "learning_rate": 4.544930431430274e-05,
      "loss": 2.378,
      "step": 25980
    },
    {
      "epoch": 0.09126274839659342,
      "grad_norm": 0.7123090624809265,
      "learning_rate": 4.54387901727824e-05,
      "loss": 2.3748,
      "step": 26040
    },
    {
      "epoch": 0.09147303122700032,
      "grad_norm": 0.6969013214111328,
      "learning_rate": 4.542827603126205e-05,
      "loss": 2.3761,
      "step": 26100
    },
    {
      "epoch": 0.0916833140574072,
      "grad_norm": 0.6556299328804016,
      "learning_rate": 4.541776188974171e-05,
      "loss": 2.3884,
      "step": 26160
    },
    {
      "epoch": 0.09189359688781411,
      "grad_norm": 0.6420555710792542,
      "learning_rate": 4.5407422983913365e-05,
      "loss": 2.3777,
      "step": 26220
    },
    {
      "epoch": 0.09210387971822101,
      "grad_norm": 0.6758638024330139,
      "learning_rate": 4.5396908842393024e-05,
      "loss": 2.3934,
      "step": 26280
    },
    {
      "epoch": 0.0923141625486279,
      "grad_norm": 0.6616193652153015,
      "learning_rate": 4.5386394700872676e-05,
      "loss": 2.3729,
      "step": 26340
    },
    {
      "epoch": 0.0925244453790348,
      "grad_norm": 0.7162423729896545,
      "learning_rate": 4.537588055935233e-05,
      "loss": 2.3803,
      "step": 26400
    },
    {
      "epoch": 0.0927347282094417,
      "grad_norm": 0.6722175478935242,
      "learning_rate": 4.5365366417831986e-05,
      "loss": 2.3835,
      "step": 26460
    },
    {
      "epoch": 0.09294501103984859,
      "grad_norm": 0.6524060368537903,
      "learning_rate": 4.5354852276311645e-05,
      "loss": 2.3756,
      "step": 26520
    },
    {
      "epoch": 0.09315529387025549,
      "grad_norm": 0.6283015608787537,
      "learning_rate": 4.53443381347913e-05,
      "loss": 2.3789,
      "step": 26580
    },
    {
      "epoch": 0.0933655767006624,
      "grad_norm": 0.6441583037376404,
      "learning_rate": 4.533382399327095e-05,
      "loss": 2.3858,
      "step": 26640
    },
    {
      "epoch": 0.09357585953106928,
      "grad_norm": 0.6654236316680908,
      "learning_rate": 4.532330985175061e-05,
      "loss": 2.3774,
      "step": 26700
    },
    {
      "epoch": 0.09378614236147619,
      "grad_norm": 0.6412244439125061,
      "learning_rate": 4.531279571023026e-05,
      "loss": 2.3693,
      "step": 26760
    },
    {
      "epoch": 0.09399642519188309,
      "grad_norm": 0.7081546783447266,
      "learning_rate": 4.530228156870992e-05,
      "loss": 2.3834,
      "step": 26820
    },
    {
      "epoch": 0.09420670802228998,
      "grad_norm": 0.6857917904853821,
      "learning_rate": 4.5291767427189576e-05,
      "loss": 2.3694,
      "step": 26880
    },
    {
      "epoch": 0.09441699085269688,
      "grad_norm": 0.6636059880256653,
      "learning_rate": 4.528125328566923e-05,
      "loss": 2.375,
      "step": 26940
    },
    {
      "epoch": 0.09462727368310378,
      "grad_norm": 0.6544620990753174,
      "learning_rate": 4.527073914414888e-05,
      "loss": 2.3687,
      "step": 27000
    },
    {
      "epoch": 0.09483755651351067,
      "grad_norm": 0.6736893653869629,
      "learning_rate": 4.526022500262854e-05,
      "loss": 2.3776,
      "step": 27060
    },
    {
      "epoch": 0.09504783934391757,
      "grad_norm": 0.6209065318107605,
      "learning_rate": 4.524971086110819e-05,
      "loss": 2.3871,
      "step": 27120
    },
    {
      "epoch": 0.09525812217432447,
      "grad_norm": 0.6604148149490356,
      "learning_rate": 4.523919671958784e-05,
      "loss": 2.3553,
      "step": 27180
    },
    {
      "epoch": 0.09546840500473136,
      "grad_norm": 0.6733397245407104,
      "learning_rate": 4.522868257806751e-05,
      "loss": 2.3763,
      "step": 27240
    },
    {
      "epoch": 0.09567868783513826,
      "grad_norm": 0.6688505411148071,
      "learning_rate": 4.521816843654716e-05,
      "loss": 2.3845,
      "step": 27300
    },
    {
      "epoch": 0.09588897066554516,
      "grad_norm": 0.6299710273742676,
      "learning_rate": 4.520765429502681e-05,
      "loss": 2.3823,
      "step": 27360
    },
    {
      "epoch": 0.09609925349595205,
      "grad_norm": 0.6818243861198425,
      "learning_rate": 4.519714015350647e-05,
      "loss": 2.3822,
      "step": 27420
    },
    {
      "epoch": 0.09630953632635895,
      "grad_norm": 0.671530544757843,
      "learning_rate": 4.518662601198612e-05,
      "loss": 2.3907,
      "step": 27480
    },
    {
      "epoch": 0.09651981915676586,
      "grad_norm": 0.7086201310157776,
      "learning_rate": 4.5176111870465774e-05,
      "loss": 2.3716,
      "step": 27540
    },
    {
      "epoch": 0.09673010198717275,
      "grad_norm": 0.6694490909576416,
      "learning_rate": 4.516559772894543e-05,
      "loss": 2.3833,
      "step": 27600
    },
    {
      "epoch": 0.09694038481757965,
      "grad_norm": 0.630606472492218,
      "learning_rate": 4.51552588231171e-05,
      "loss": 2.3933,
      "step": 27660
    },
    {
      "epoch": 0.09715066764798654,
      "grad_norm": 0.6745237708091736,
      "learning_rate": 4.514474468159675e-05,
      "loss": 2.3681,
      "step": 27720
    },
    {
      "epoch": 0.09736095047839344,
      "grad_norm": 0.6740299463272095,
      "learning_rate": 4.513423054007641e-05,
      "loss": 2.3852,
      "step": 27780
    },
    {
      "epoch": 0.09757123330880034,
      "grad_norm": 0.6464400887489319,
      "learning_rate": 4.512371639855606e-05,
      "loss": 2.3856,
      "step": 27840
    },
    {
      "epoch": 0.09778151613920723,
      "grad_norm": 0.6964374780654907,
      "learning_rate": 4.511320225703571e-05,
      "loss": 2.3825,
      "step": 27900
    },
    {
      "epoch": 0.09799179896961413,
      "grad_norm": 0.6555659770965576,
      "learning_rate": 4.510268811551537e-05,
      "loss": 2.3811,
      "step": 27960
    },
    {
      "epoch": 0.09820208180002103,
      "grad_norm": 0.6736533045768738,
      "learning_rate": 4.509217397399503e-05,
      "loss": 2.3686,
      "step": 28020
    },
    {
      "epoch": 0.09841236463042792,
      "grad_norm": 0.710355818271637,
      "learning_rate": 4.508165983247468e-05,
      "loss": 2.3703,
      "step": 28080
    },
    {
      "epoch": 0.09862264746083482,
      "grad_norm": 0.6771557331085205,
      "learning_rate": 4.507114569095434e-05,
      "loss": 2.3663,
      "step": 28140
    },
    {
      "epoch": 0.09883293029124172,
      "grad_norm": 0.6405073404312134,
      "learning_rate": 4.506063154943399e-05,
      "loss": 2.3757,
      "step": 28200
    },
    {
      "epoch": 0.09904321312164861,
      "grad_norm": 0.6605597734451294,
      "learning_rate": 4.505011740791364e-05,
      "loss": 2.3752,
      "step": 28260
    },
    {
      "epoch": 0.09925349595205551,
      "grad_norm": 0.6761056184768677,
      "learning_rate": 4.50396032663933e-05,
      "loss": 2.3628,
      "step": 28320
    },
    {
      "epoch": 0.09946377878246242,
      "grad_norm": 0.6279546022415161,
      "learning_rate": 4.502908912487296e-05,
      "loss": 2.3638,
      "step": 28380
    },
    {
      "epoch": 0.0996740616128693,
      "grad_norm": 0.7104046940803528,
      "learning_rate": 4.501857498335261e-05,
      "loss": 2.3686,
      "step": 28440
    },
    {
      "epoch": 0.09988434444327621,
      "grad_norm": 0.625745415687561,
      "learning_rate": 4.5008060841832264e-05,
      "loss": 2.3646,
      "step": 28500
    },
    {
      "epoch": 0.10009462727368311,
      "grad_norm": 0.6523934602737427,
      "learning_rate": 4.499754670031192e-05,
      "loss": 2.3637,
      "step": 28560
    },
    {
      "epoch": 0.10030491010409,
      "grad_norm": 0.678256630897522,
      "learning_rate": 4.4987032558791575e-05,
      "loss": 2.3852,
      "step": 28620
    },
    {
      "epoch": 0.1005151929344969,
      "grad_norm": 0.6635131239891052,
      "learning_rate": 4.4976518417271227e-05,
      "loss": 2.3798,
      "step": 28680
    },
    {
      "epoch": 0.1007254757649038,
      "grad_norm": 0.728843629360199,
      "learning_rate": 4.496600427575089e-05,
      "loss": 2.3787,
      "step": 28740
    },
    {
      "epoch": 0.10093575859531069,
      "grad_norm": 0.6350687146186829,
      "learning_rate": 4.4955490134230544e-05,
      "loss": 2.3804,
      "step": 28800
    },
    {
      "epoch": 0.10114604142571759,
      "grad_norm": 0.732272207736969,
      "learning_rate": 4.4944975992710196e-05,
      "loss": 2.3734,
      "step": 28860
    },
    {
      "epoch": 0.1013563242561245,
      "grad_norm": 0.6215546727180481,
      "learning_rate": 4.4934461851189854e-05,
      "loss": 2.3583,
      "step": 28920
    },
    {
      "epoch": 0.10156660708653138,
      "grad_norm": 0.600857138633728,
      "learning_rate": 4.4923947709669506e-05,
      "loss": 2.3719,
      "step": 28980
    },
    {
      "epoch": 0.10177688991693828,
      "grad_norm": 0.6272040009498596,
      "learning_rate": 4.491343356814916e-05,
      "loss": 2.3675,
      "step": 29040
    },
    {
      "epoch": 0.10198717274734517,
      "grad_norm": 0.6435662508010864,
      "learning_rate": 4.490291942662882e-05,
      "loss": 2.3619,
      "step": 29100
    },
    {
      "epoch": 0.10219745557775207,
      "grad_norm": 0.6466601490974426,
      "learning_rate": 4.4892405285108476e-05,
      "loss": 2.3576,
      "step": 29160
    },
    {
      "epoch": 0.10240773840815898,
      "grad_norm": 0.6503751873970032,
      "learning_rate": 4.488189114358813e-05,
      "loss": 2.3659,
      "step": 29220
    },
    {
      "epoch": 0.10261802123856587,
      "grad_norm": 0.6413028240203857,
      "learning_rate": 4.487137700206778e-05,
      "loss": 2.3686,
      "step": 29280
    },
    {
      "epoch": 0.10282830406897277,
      "grad_norm": 0.6793463826179504,
      "learning_rate": 4.486086286054744e-05,
      "loss": 2.357,
      "step": 29340
    },
    {
      "epoch": 0.10303858689937967,
      "grad_norm": 0.6493575572967529,
      "learning_rate": 4.485034871902709e-05,
      "loss": 2.3692,
      "step": 29400
    },
    {
      "epoch": 0.10324886972978656,
      "grad_norm": 0.6862606406211853,
      "learning_rate": 4.483983457750675e-05,
      "loss": 2.3631,
      "step": 29460
    },
    {
      "epoch": 0.10345915256019346,
      "grad_norm": 0.688431441783905,
      "learning_rate": 4.482932043598641e-05,
      "loss": 2.3779,
      "step": 29520
    },
    {
      "epoch": 0.10366943539060036,
      "grad_norm": 0.6671889424324036,
      "learning_rate": 4.481880629446606e-05,
      "loss": 2.372,
      "step": 29580
    },
    {
      "epoch": 0.10387971822100725,
      "grad_norm": 0.6443668603897095,
      "learning_rate": 4.480829215294571e-05,
      "loss": 2.3927,
      "step": 29640
    },
    {
      "epoch": 0.10409000105141415,
      "grad_norm": 0.6309412717819214,
      "learning_rate": 4.479777801142537e-05,
      "loss": 2.3609,
      "step": 29700
    },
    {
      "epoch": 0.10430028388182105,
      "grad_norm": 0.6563585996627808,
      "learning_rate": 4.478726386990502e-05,
      "loss": 2.3551,
      "step": 29760
    },
    {
      "epoch": 0.10451056671222794,
      "grad_norm": 0.6455681920051575,
      "learning_rate": 4.477674972838468e-05,
      "loss": 2.3806,
      "step": 29820
    },
    {
      "epoch": 0.10472084954263484,
      "grad_norm": 0.6657047867774963,
      "learning_rate": 4.476623558686434e-05,
      "loss": 2.3618,
      "step": 29880
    },
    {
      "epoch": 0.10493113237304175,
      "grad_norm": 0.6885735392570496,
      "learning_rate": 4.475572144534399e-05,
      "loss": 2.3622,
      "step": 29940
    },
    {
      "epoch": 0.10514141520344863,
      "grad_norm": 0.6443381905555725,
      "learning_rate": 4.474520730382364e-05,
      "loss": 2.3561,
      "step": 30000
    },
    {
      "epoch": 0.10535169803385554,
      "grad_norm": 0.6550012826919556,
      "learning_rate": 4.47346931623033e-05,
      "loss": 2.3585,
      "step": 30060
    },
    {
      "epoch": 0.10556198086426244,
      "grad_norm": 0.651603102684021,
      "learning_rate": 4.472417902078295e-05,
      "loss": 2.3669,
      "step": 30120
    },
    {
      "epoch": 0.10577226369466933,
      "grad_norm": 0.676773726940155,
      "learning_rate": 4.471366487926261e-05,
      "loss": 2.3702,
      "step": 30180
    },
    {
      "epoch": 0.10598254652507623,
      "grad_norm": 0.6347634196281433,
      "learning_rate": 4.4703150737742264e-05,
      "loss": 2.371,
      "step": 30240
    },
    {
      "epoch": 0.10619282935548313,
      "grad_norm": 0.67087721824646,
      "learning_rate": 4.469263659622192e-05,
      "loss": 2.3546,
      "step": 30300
    },
    {
      "epoch": 0.10640311218589002,
      "grad_norm": 0.6727715730667114,
      "learning_rate": 4.4682122454701574e-05,
      "loss": 2.3523,
      "step": 30360
    },
    {
      "epoch": 0.10661339501629692,
      "grad_norm": 0.6418828368186951,
      "learning_rate": 4.4671608313181226e-05,
      "loss": 2.3702,
      "step": 30420
    },
    {
      "epoch": 0.10682367784670381,
      "grad_norm": 0.6562620997428894,
      "learning_rate": 4.4661094171660885e-05,
      "loss": 2.3854,
      "step": 30480
    },
    {
      "epoch": 0.10703396067711071,
      "grad_norm": 0.6004977822303772,
      "learning_rate": 4.4650580030140544e-05,
      "loss": 2.3542,
      "step": 30540
    },
    {
      "epoch": 0.10724424350751761,
      "grad_norm": 0.6666843891143799,
      "learning_rate": 4.4640065888620195e-05,
      "loss": 2.3609,
      "step": 30600
    },
    {
      "epoch": 0.1074545263379245,
      "grad_norm": 0.5965372323989868,
      "learning_rate": 4.4629551747099854e-05,
      "loss": 2.3746,
      "step": 30660
    },
    {
      "epoch": 0.1076648091683314,
      "grad_norm": 0.6549592018127441,
      "learning_rate": 4.4619037605579506e-05,
      "loss": 2.3612,
      "step": 30720
    },
    {
      "epoch": 0.1078750919987383,
      "grad_norm": 0.679852306842804,
      "learning_rate": 4.460852346405916e-05,
      "loss": 2.3665,
      "step": 30780
    },
    {
      "epoch": 0.1080853748291452,
      "grad_norm": 0.6266508102416992,
      "learning_rate": 4.4598009322538817e-05,
      "loss": 2.3547,
      "step": 30840
    },
    {
      "epoch": 0.1082956576595521,
      "grad_norm": 0.6630394458770752,
      "learning_rate": 4.4587495181018475e-05,
      "loss": 2.3633,
      "step": 30900
    },
    {
      "epoch": 0.108505940489959,
      "grad_norm": 0.645500898361206,
      "learning_rate": 4.457715627519013e-05,
      "loss": 2.3528,
      "step": 30960
    },
    {
      "epoch": 0.10871622332036589,
      "grad_norm": 0.6527357697486877,
      "learning_rate": 4.456664213366979e-05,
      "loss": 2.3743,
      "step": 31020
    },
    {
      "epoch": 0.10892650615077279,
      "grad_norm": 0.7146820425987244,
      "learning_rate": 4.455612799214944e-05,
      "loss": 2.3685,
      "step": 31080
    },
    {
      "epoch": 0.10913678898117969,
      "grad_norm": 0.6792079210281372,
      "learning_rate": 4.4545613850629095e-05,
      "loss": 2.3618,
      "step": 31140
    },
    {
      "epoch": 0.10934707181158658,
      "grad_norm": 0.6431834697723389,
      "learning_rate": 4.4535099709108754e-05,
      "loss": 2.3716,
      "step": 31200
    },
    {
      "epoch": 0.10955735464199348,
      "grad_norm": 0.7037909030914307,
      "learning_rate": 4.4524585567588405e-05,
      "loss": 2.3665,
      "step": 31260
    },
    {
      "epoch": 0.10976763747240038,
      "grad_norm": 0.637527585029602,
      "learning_rate": 4.4514071426068064e-05,
      "loss": 2.3625,
      "step": 31320
    },
    {
      "epoch": 0.10997792030280727,
      "grad_norm": 0.6294242739677429,
      "learning_rate": 4.450355728454772e-05,
      "loss": 2.3697,
      "step": 31380
    },
    {
      "epoch": 0.11018820313321417,
      "grad_norm": 0.6632541418075562,
      "learning_rate": 4.4493043143027375e-05,
      "loss": 2.3689,
      "step": 31440
    },
    {
      "epoch": 0.11039848596362108,
      "grad_norm": 0.6736090183258057,
      "learning_rate": 4.4482529001507027e-05,
      "loss": 2.3597,
      "step": 31500
    },
    {
      "epoch": 0.11060876879402796,
      "grad_norm": 0.6545177698135376,
      "learning_rate": 4.4472014859986685e-05,
      "loss": 2.3501,
      "step": 31560
    },
    {
      "epoch": 0.11081905162443487,
      "grad_norm": 0.6885281801223755,
      "learning_rate": 4.446150071846634e-05,
      "loss": 2.3684,
      "step": 31620
    },
    {
      "epoch": 0.11102933445484177,
      "grad_norm": 0.6284212470054626,
      "learning_rate": 4.4450986576945996e-05,
      "loss": 2.3692,
      "step": 31680
    },
    {
      "epoch": 0.11123961728524866,
      "grad_norm": 0.6197623014450073,
      "learning_rate": 4.444047243542565e-05,
      "loss": 2.3573,
      "step": 31740
    },
    {
      "epoch": 0.11144990011565556,
      "grad_norm": 0.6184737086296082,
      "learning_rate": 4.4429958293905306e-05,
      "loss": 2.3704,
      "step": 31800
    },
    {
      "epoch": 0.11166018294606245,
      "grad_norm": 0.6636645793914795,
      "learning_rate": 4.441944415238496e-05,
      "loss": 2.3807,
      "step": 31860
    },
    {
      "epoch": 0.11187046577646935,
      "grad_norm": 0.6567378640174866,
      "learning_rate": 4.440893001086461e-05,
      "loss": 2.3758,
      "step": 31920
    },
    {
      "epoch": 0.11208074860687625,
      "grad_norm": 0.6212893724441528,
      "learning_rate": 4.439841586934427e-05,
      "loss": 2.3483,
      "step": 31980
    },
    {
      "epoch": 0.11229103143728314,
      "grad_norm": 0.6712754964828491,
      "learning_rate": 4.438790172782393e-05,
      "loss": 2.3579,
      "step": 32040
    },
    {
      "epoch": 0.11250131426769004,
      "grad_norm": 0.6450668573379517,
      "learning_rate": 4.437738758630358e-05,
      "loss": 2.3723,
      "step": 32100
    },
    {
      "epoch": 0.11271159709809694,
      "grad_norm": 0.6592389345169067,
      "learning_rate": 4.436687344478324e-05,
      "loss": 2.3569,
      "step": 32160
    },
    {
      "epoch": 0.11292187992850383,
      "grad_norm": 0.6862047910690308,
      "learning_rate": 4.435635930326289e-05,
      "loss": 2.3567,
      "step": 32220
    },
    {
      "epoch": 0.11313216275891073,
      "grad_norm": 0.6675071716308594,
      "learning_rate": 4.434584516174254e-05,
      "loss": 2.3642,
      "step": 32280
    },
    {
      "epoch": 0.11334244558931764,
      "grad_norm": 0.6118093132972717,
      "learning_rate": 4.43353310202222e-05,
      "loss": 2.3598,
      "step": 32340
    },
    {
      "epoch": 0.11355272841972452,
      "grad_norm": 0.6482064127922058,
      "learning_rate": 4.432481687870186e-05,
      "loss": 2.3656,
      "step": 32400
    },
    {
      "epoch": 0.11376301125013143,
      "grad_norm": 0.6413975954055786,
      "learning_rate": 4.431430273718151e-05,
      "loss": 2.3498,
      "step": 32460
    },
    {
      "epoch": 0.11397329408053833,
      "grad_norm": 0.596386194229126,
      "learning_rate": 4.430378859566117e-05,
      "loss": 2.3548,
      "step": 32520
    },
    {
      "epoch": 0.11418357691094522,
      "grad_norm": 0.618043839931488,
      "learning_rate": 4.429327445414082e-05,
      "loss": 2.3578,
      "step": 32580
    },
    {
      "epoch": 0.11439385974135212,
      "grad_norm": 0.629632294178009,
      "learning_rate": 4.4282760312620473e-05,
      "loss": 2.3597,
      "step": 32640
    },
    {
      "epoch": 0.11460414257175902,
      "grad_norm": 0.6268094182014465,
      "learning_rate": 4.427224617110013e-05,
      "loss": 2.3605,
      "step": 32700
    },
    {
      "epoch": 0.11481442540216591,
      "grad_norm": 0.608088493347168,
      "learning_rate": 4.426173202957979e-05,
      "loss": 2.3664,
      "step": 32760
    },
    {
      "epoch": 0.11502470823257281,
      "grad_norm": 0.7458576560020447,
      "learning_rate": 4.425121788805944e-05,
      "loss": 2.3507,
      "step": 32820
    },
    {
      "epoch": 0.11523499106297971,
      "grad_norm": 0.6672834753990173,
      "learning_rate": 4.4240703746539095e-05,
      "loss": 2.3508,
      "step": 32880
    },
    {
      "epoch": 0.1154452738933866,
      "grad_norm": 0.5789446830749512,
      "learning_rate": 4.423018960501875e-05,
      "loss": 2.36,
      "step": 32940
    },
    {
      "epoch": 0.1156555567237935,
      "grad_norm": 0.694132924079895,
      "learning_rate": 4.4219675463498405e-05,
      "loss": 2.3433,
      "step": 33000
    },
    {
      "epoch": 0.1158658395542004,
      "grad_norm": 0.6312335729598999,
      "learning_rate": 4.4209161321978064e-05,
      "loss": 2.3366,
      "step": 33060
    },
    {
      "epoch": 0.1160761223846073,
      "grad_norm": 0.6660616993904114,
      "learning_rate": 4.419864718045772e-05,
      "loss": 2.3512,
      "step": 33120
    },
    {
      "epoch": 0.1162864052150142,
      "grad_norm": 0.6671555638313293,
      "learning_rate": 4.4188133038937374e-05,
      "loss": 2.3515,
      "step": 33180
    },
    {
      "epoch": 0.11649668804542108,
      "grad_norm": 0.645706057548523,
      "learning_rate": 4.4177618897417026e-05,
      "loss": 2.3594,
      "step": 33240
    },
    {
      "epoch": 0.11670697087582799,
      "grad_norm": 0.6335362792015076,
      "learning_rate": 4.4167104755896685e-05,
      "loss": 2.3508,
      "step": 33300
    },
    {
      "epoch": 0.11691725370623489,
      "grad_norm": 0.6129640340805054,
      "learning_rate": 4.415659061437634e-05,
      "loss": 2.3498,
      "step": 33360
    },
    {
      "epoch": 0.11712753653664178,
      "grad_norm": 0.6517692804336548,
      "learning_rate": 4.4146076472855995e-05,
      "loss": 2.3586,
      "step": 33420
    },
    {
      "epoch": 0.11733781936704868,
      "grad_norm": 0.6196044683456421,
      "learning_rate": 4.413556233133565e-05,
      "loss": 2.3515,
      "step": 33480
    },
    {
      "epoch": 0.11754810219745558,
      "grad_norm": 0.634680986404419,
      "learning_rate": 4.4125048189815306e-05,
      "loss": 2.3583,
      "step": 33540
    },
    {
      "epoch": 0.11775838502786247,
      "grad_norm": 0.6536049246788025,
      "learning_rate": 4.411453404829496e-05,
      "loss": 2.3498,
      "step": 33600
    },
    {
      "epoch": 0.11796866785826937,
      "grad_norm": 0.6144905090332031,
      "learning_rate": 4.410401990677461e-05,
      "loss": 2.3533,
      "step": 33660
    },
    {
      "epoch": 0.11817895068867627,
      "grad_norm": 0.6263406872749329,
      "learning_rate": 4.409350576525427e-05,
      "loss": 2.3425,
      "step": 33720
    },
    {
      "epoch": 0.11838923351908316,
      "grad_norm": 0.5995636582374573,
      "learning_rate": 4.408299162373393e-05,
      "loss": 2.355,
      "step": 33780
    },
    {
      "epoch": 0.11859951634949006,
      "grad_norm": 0.6604700684547424,
      "learning_rate": 4.407247748221358e-05,
      "loss": 2.3319,
      "step": 33840
    },
    {
      "epoch": 0.11880979917989697,
      "grad_norm": 0.683630645275116,
      "learning_rate": 4.406196334069324e-05,
      "loss": 2.3547,
      "step": 33900
    },
    {
      "epoch": 0.11902008201030385,
      "grad_norm": 0.6503610610961914,
      "learning_rate": 4.405144919917289e-05,
      "loss": 2.3574,
      "step": 33960
    },
    {
      "epoch": 0.11923036484071076,
      "grad_norm": 0.659476101398468,
      "learning_rate": 4.404093505765254e-05,
      "loss": 2.3298,
      "step": 34020
    },
    {
      "epoch": 0.11944064767111766,
      "grad_norm": 0.636908233165741,
      "learning_rate": 4.40304209161322e-05,
      "loss": 2.3475,
      "step": 34080
    },
    {
      "epoch": 0.11965093050152455,
      "grad_norm": 0.6105729937553406,
      "learning_rate": 4.401990677461186e-05,
      "loss": 2.3422,
      "step": 34140
    },
    {
      "epoch": 0.11986121333193145,
      "grad_norm": 0.6372390389442444,
      "learning_rate": 4.400939263309151e-05,
      "loss": 2.3513,
      "step": 34200
    },
    {
      "epoch": 0.12007149616233835,
      "grad_norm": 0.6719101071357727,
      "learning_rate": 4.399887849157117e-05,
      "loss": 2.3412,
      "step": 34260
    },
    {
      "epoch": 0.12028177899274524,
      "grad_norm": 0.6186497807502747,
      "learning_rate": 4.398836435005082e-05,
      "loss": 2.3487,
      "step": 34320
    },
    {
      "epoch": 0.12049206182315214,
      "grad_norm": 0.6582601070404053,
      "learning_rate": 4.397785020853047e-05,
      "loss": 2.3543,
      "step": 34380
    },
    {
      "epoch": 0.12070234465355904,
      "grad_norm": 0.5645360946655273,
      "learning_rate": 4.396733606701013e-05,
      "loss": 2.3708,
      "step": 34440
    },
    {
      "epoch": 0.12091262748396593,
      "grad_norm": 0.6398698687553406,
      "learning_rate": 4.395682192548979e-05,
      "loss": 2.3539,
      "step": 34500
    },
    {
      "epoch": 0.12112291031437283,
      "grad_norm": 0.6094596982002258,
      "learning_rate": 4.394630778396944e-05,
      "loss": 2.3537,
      "step": 34560
    },
    {
      "epoch": 0.12133319314477972,
      "grad_norm": 0.6569386720657349,
      "learning_rate": 4.3935793642449094e-05,
      "loss": 2.3436,
      "step": 34620
    },
    {
      "epoch": 0.12154347597518662,
      "grad_norm": 0.6402146816253662,
      "learning_rate": 4.392527950092875e-05,
      "loss": 2.3567,
      "step": 34680
    },
    {
      "epoch": 0.12175375880559353,
      "grad_norm": 0.7214573621749878,
      "learning_rate": 4.3914765359408405e-05,
      "loss": 2.357,
      "step": 34740
    },
    {
      "epoch": 0.12196404163600041,
      "grad_norm": 0.6399426460266113,
      "learning_rate": 4.390425121788806e-05,
      "loss": 2.3413,
      "step": 34800
    },
    {
      "epoch": 0.12217432446640732,
      "grad_norm": 0.6547138094902039,
      "learning_rate": 4.389373707636772e-05,
      "loss": 2.3567,
      "step": 34860
    },
    {
      "epoch": 0.12238460729681422,
      "grad_norm": 0.6688662171363831,
      "learning_rate": 4.3883222934847374e-05,
      "loss": 2.3543,
      "step": 34920
    },
    {
      "epoch": 0.1225948901272211,
      "grad_norm": 0.6135913729667664,
      "learning_rate": 4.3872708793327026e-05,
      "loss": 2.3516,
      "step": 34980
    },
    {
      "epoch": 0.12280517295762801,
      "grad_norm": 0.6633687615394592,
      "learning_rate": 4.3862194651806685e-05,
      "loss": 2.363,
      "step": 35040
    },
    {
      "epoch": 0.12301545578803491,
      "grad_norm": 0.6762205958366394,
      "learning_rate": 4.3851680510286337e-05,
      "loss": 2.3551,
      "step": 35100
    },
    {
      "epoch": 0.1232257386184418,
      "grad_norm": 0.6045565009117126,
      "learning_rate": 4.384116636876599e-05,
      "loss": 2.3614,
      "step": 35160
    },
    {
      "epoch": 0.1234360214488487,
      "grad_norm": 0.6369898915290833,
      "learning_rate": 4.383065222724565e-05,
      "loss": 2.3384,
      "step": 35220
    },
    {
      "epoch": 0.1236463042792556,
      "grad_norm": 0.607323169708252,
      "learning_rate": 4.3820138085725306e-05,
      "loss": 2.3433,
      "step": 35280
    },
    {
      "epoch": 0.12385658710966249,
      "grad_norm": 0.6833062767982483,
      "learning_rate": 4.380962394420496e-05,
      "loss": 2.351,
      "step": 35340
    },
    {
      "epoch": 0.1240668699400694,
      "grad_norm": 0.6462218761444092,
      "learning_rate": 4.379910980268461e-05,
      "loss": 2.3512,
      "step": 35400
    },
    {
      "epoch": 0.1242771527704763,
      "grad_norm": 0.628410279750824,
      "learning_rate": 4.378859566116427e-05,
      "loss": 2.3415,
      "step": 35460
    },
    {
      "epoch": 0.12448743560088318,
      "grad_norm": 0.6561073064804077,
      "learning_rate": 4.377808151964392e-05,
      "loss": 2.3513,
      "step": 35520
    },
    {
      "epoch": 0.12469771843129009,
      "grad_norm": 0.605626106262207,
      "learning_rate": 4.376756737812358e-05,
      "loss": 2.3387,
      "step": 35580
    },
    {
      "epoch": 0.12490800126169699,
      "grad_norm": 0.6336895823478699,
      "learning_rate": 4.375705323660324e-05,
      "loss": 2.3564,
      "step": 35640
    },
    {
      "epoch": 0.12511828409210388,
      "grad_norm": 0.6178215742111206,
      "learning_rate": 4.374653909508289e-05,
      "loss": 2.3507,
      "step": 35700
    },
    {
      "epoch": 0.12532856692251076,
      "grad_norm": 0.5754424333572388,
      "learning_rate": 4.373602495356254e-05,
      "loss": 2.3547,
      "step": 35760
    },
    {
      "epoch": 0.12553884975291768,
      "grad_norm": 0.6096973419189453,
      "learning_rate": 4.37255108120422e-05,
      "loss": 2.3464,
      "step": 35820
    },
    {
      "epoch": 0.12574913258332457,
      "grad_norm": 0.6017065048217773,
      "learning_rate": 4.371499667052185e-05,
      "loss": 2.3488,
      "step": 35880
    },
    {
      "epoch": 0.12595941541373146,
      "grad_norm": 0.6396207809448242,
      "learning_rate": 4.370448252900151e-05,
      "loss": 2.334,
      "step": 35940
    },
    {
      "epoch": 0.12616969824413837,
      "grad_norm": 0.6168996691703796,
      "learning_rate": 4.369396838748116e-05,
      "loss": 2.3496,
      "step": 36000
    },
    {
      "epoch": 0.12637998107454526,
      "grad_norm": 0.62862229347229,
      "learning_rate": 4.368345424596082e-05,
      "loss": 2.3338,
      "step": 36060
    },
    {
      "epoch": 0.12659026390495215,
      "grad_norm": 0.6999679207801819,
      "learning_rate": 4.367294010444047e-05,
      "loss": 2.3498,
      "step": 36120
    },
    {
      "epoch": 0.12680054673535907,
      "grad_norm": 0.6003287434577942,
      "learning_rate": 4.366242596292013e-05,
      "loss": 2.3495,
      "step": 36180
    },
    {
      "epoch": 0.12701082956576595,
      "grad_norm": 0.6159112453460693,
      "learning_rate": 4.3651911821399783e-05,
      "loss": 2.3434,
      "step": 36240
    },
    {
      "epoch": 0.12722111239617284,
      "grad_norm": 0.6516057252883911,
      "learning_rate": 4.364139767987944e-05,
      "loss": 2.3495,
      "step": 36300
    },
    {
      "epoch": 0.12743139522657976,
      "grad_norm": 0.6326673626899719,
      "learning_rate": 4.3630883538359094e-05,
      "loss": 2.3433,
      "step": 36360
    },
    {
      "epoch": 0.12764167805698665,
      "grad_norm": 0.6475709676742554,
      "learning_rate": 4.362036939683875e-05,
      "loss": 2.3693,
      "step": 36420
    },
    {
      "epoch": 0.12785196088739353,
      "grad_norm": 0.6244445443153381,
      "learning_rate": 4.3609855255318405e-05,
      "loss": 2.3536,
      "step": 36480
    },
    {
      "epoch": 0.12806224371780045,
      "grad_norm": 0.6203553080558777,
      "learning_rate": 4.3599341113798056e-05,
      "loss": 2.3531,
      "step": 36540
    },
    {
      "epoch": 0.12827252654820734,
      "grad_norm": 0.6234232187271118,
      "learning_rate": 4.3588826972277715e-05,
      "loss": 2.3542,
      "step": 36600
    },
    {
      "epoch": 0.12848280937861423,
      "grad_norm": 0.6129841208457947,
      "learning_rate": 4.3578312830757374e-05,
      "loss": 2.344,
      "step": 36660
    },
    {
      "epoch": 0.12869309220902114,
      "grad_norm": 0.6914127469062805,
      "learning_rate": 4.3567798689237026e-05,
      "loss": 2.343,
      "step": 36720
    },
    {
      "epoch": 0.12890337503942803,
      "grad_norm": 0.6331791281700134,
      "learning_rate": 4.3557284547716684e-05,
      "loss": 2.3464,
      "step": 36780
    },
    {
      "epoch": 0.12911365786983492,
      "grad_norm": 0.6370502710342407,
      "learning_rate": 4.3546770406196336e-05,
      "loss": 2.3419,
      "step": 36840
    },
    {
      "epoch": 0.12932394070024184,
      "grad_norm": 0.6535687446594238,
      "learning_rate": 4.353625626467599e-05,
      "loss": 2.3459,
      "step": 36900
    },
    {
      "epoch": 0.12953422353064872,
      "grad_norm": 0.6368798613548279,
      "learning_rate": 4.352591735884765e-05,
      "loss": 2.3495,
      "step": 36960
    },
    {
      "epoch": 0.1297445063610556,
      "grad_norm": 0.6685027480125427,
      "learning_rate": 4.351557845301931e-05,
      "loss": 2.3321,
      "step": 37020
    },
    {
      "epoch": 0.12995478919146253,
      "grad_norm": 0.6186925768852234,
      "learning_rate": 4.350506431149897e-05,
      "loss": 2.3328,
      "step": 37080
    },
    {
      "epoch": 0.13016507202186942,
      "grad_norm": 0.6169596314430237,
      "learning_rate": 4.349455016997862e-05,
      "loss": 2.3439,
      "step": 37140
    },
    {
      "epoch": 0.1303753548522763,
      "grad_norm": 0.6129602789878845,
      "learning_rate": 4.348403602845828e-05,
      "loss": 2.344,
      "step": 37200
    },
    {
      "epoch": 0.13058563768268322,
      "grad_norm": 0.6711073517799377,
      "learning_rate": 4.347352188693794e-05,
      "loss": 2.349,
      "step": 37260
    },
    {
      "epoch": 0.1307959205130901,
      "grad_norm": 0.6636022925376892,
      "learning_rate": 4.346300774541759e-05,
      "loss": 2.3578,
      "step": 37320
    },
    {
      "epoch": 0.131006203343497,
      "grad_norm": 0.6326186060905457,
      "learning_rate": 4.345249360389724e-05,
      "loss": 2.3404,
      "step": 37380
    },
    {
      "epoch": 0.1312164861739039,
      "grad_norm": 0.6219196915626526,
      "learning_rate": 4.34419794623769e-05,
      "loss": 2.3423,
      "step": 37440
    },
    {
      "epoch": 0.1314267690043108,
      "grad_norm": 0.6334632635116577,
      "learning_rate": 4.343146532085655e-05,
      "loss": 2.3453,
      "step": 37500
    },
    {
      "epoch": 0.1316370518347177,
      "grad_norm": 0.6453350782394409,
      "learning_rate": 4.342095117933621e-05,
      "loss": 2.3581,
      "step": 37560
    },
    {
      "epoch": 0.1318473346651246,
      "grad_norm": 0.637695848941803,
      "learning_rate": 4.341043703781586e-05,
      "loss": 2.3506,
      "step": 37620
    },
    {
      "epoch": 0.1320576174955315,
      "grad_norm": 0.6321239471435547,
      "learning_rate": 4.339992289629552e-05,
      "loss": 2.3492,
      "step": 37680
    },
    {
      "epoch": 0.13226790032593838,
      "grad_norm": 0.5901476740837097,
      "learning_rate": 4.338940875477517e-05,
      "loss": 2.343,
      "step": 37740
    },
    {
      "epoch": 0.1324781831563453,
      "grad_norm": 0.6054787635803223,
      "learning_rate": 4.337889461325483e-05,
      "loss": 2.3409,
      "step": 37800
    },
    {
      "epoch": 0.13268846598675219,
      "grad_norm": 0.6033499240875244,
      "learning_rate": 4.336838047173449e-05,
      "loss": 2.3468,
      "step": 37860
    },
    {
      "epoch": 0.13289874881715907,
      "grad_norm": 0.6457664370536804,
      "learning_rate": 4.335786633021414e-05,
      "loss": 2.3436,
      "step": 37920
    },
    {
      "epoch": 0.133109031647566,
      "grad_norm": 0.6080107688903809,
      "learning_rate": 4.3347352188693794e-05,
      "loss": 2.3483,
      "step": 37980
    },
    {
      "epoch": 0.13331931447797288,
      "grad_norm": 0.6590386629104614,
      "learning_rate": 4.333683804717345e-05,
      "loss": 2.3528,
      "step": 38040
    },
    {
      "epoch": 0.13352959730837977,
      "grad_norm": 0.6028419137001038,
      "learning_rate": 4.3326323905653104e-05,
      "loss": 2.3359,
      "step": 38100
    },
    {
      "epoch": 0.13373988013878665,
      "grad_norm": 0.7010486721992493,
      "learning_rate": 4.3315809764132756e-05,
      "loss": 2.3474,
      "step": 38160
    },
    {
      "epoch": 0.13395016296919357,
      "grad_norm": 0.6186957955360413,
      "learning_rate": 4.330529562261242e-05,
      "loss": 2.3457,
      "step": 38220
    },
    {
      "epoch": 0.13416044579960046,
      "grad_norm": 0.6543868184089661,
      "learning_rate": 4.3294781481092073e-05,
      "loss": 2.3572,
      "step": 38280
    },
    {
      "epoch": 0.13437072863000735,
      "grad_norm": 0.613800048828125,
      "learning_rate": 4.3284267339571725e-05,
      "loss": 2.3444,
      "step": 38340
    },
    {
      "epoch": 0.13458101146041426,
      "grad_norm": 0.6139937043190002,
      "learning_rate": 4.3273753198051384e-05,
      "loss": 2.3632,
      "step": 38400
    },
    {
      "epoch": 0.13479129429082115,
      "grad_norm": 0.6320657730102539,
      "learning_rate": 4.3263239056531036e-05,
      "loss": 2.3523,
      "step": 38460
    },
    {
      "epoch": 0.13500157712122804,
      "grad_norm": 0.6350893974304199,
      "learning_rate": 4.325272491501069e-05,
      "loss": 2.3477,
      "step": 38520
    },
    {
      "epoch": 0.13521185995163496,
      "grad_norm": 0.6405870914459229,
      "learning_rate": 4.3242210773490346e-05,
      "loss": 2.3401,
      "step": 38580
    },
    {
      "epoch": 0.13542214278204184,
      "grad_norm": 0.6332811117172241,
      "learning_rate": 4.3231696631970005e-05,
      "loss": 2.3313,
      "step": 38640
    },
    {
      "epoch": 0.13563242561244873,
      "grad_norm": 0.6080389022827148,
      "learning_rate": 4.322118249044966e-05,
      "loss": 2.3336,
      "step": 38700
    },
    {
      "epoch": 0.13584270844285565,
      "grad_norm": 0.5950392484664917,
      "learning_rate": 4.321066834892931e-05,
      "loss": 2.3321,
      "step": 38760
    },
    {
      "epoch": 0.13605299127326254,
      "grad_norm": 0.5927405953407288,
      "learning_rate": 4.320015420740897e-05,
      "loss": 2.3487,
      "step": 38820
    },
    {
      "epoch": 0.13626327410366942,
      "grad_norm": 0.6823748350143433,
      "learning_rate": 4.318964006588862e-05,
      "loss": 2.3424,
      "step": 38880
    },
    {
      "epoch": 0.13647355693407634,
      "grad_norm": 0.6291486024856567,
      "learning_rate": 4.317912592436828e-05,
      "loss": 2.348,
      "step": 38940
    },
    {
      "epoch": 0.13668383976448323,
      "grad_norm": 0.608342170715332,
      "learning_rate": 4.316861178284794e-05,
      "loss": 2.3439,
      "step": 39000
    },
    {
      "epoch": 0.13689412259489012,
      "grad_norm": 0.6415091156959534,
      "learning_rate": 4.315809764132759e-05,
      "loss": 2.3409,
      "step": 39060
    },
    {
      "epoch": 0.13710440542529703,
      "grad_norm": 0.6027123928070068,
      "learning_rate": 4.314758349980724e-05,
      "loss": 2.3452,
      "step": 39120
    },
    {
      "epoch": 0.13731468825570392,
      "grad_norm": 0.6144362688064575,
      "learning_rate": 4.31370693582869e-05,
      "loss": 2.3268,
      "step": 39180
    },
    {
      "epoch": 0.1375249710861108,
      "grad_norm": 0.616381824016571,
      "learning_rate": 4.3126730452458556e-05,
      "loss": 2.3301,
      "step": 39240
    },
    {
      "epoch": 0.13773525391651772,
      "grad_norm": 0.5856786370277405,
      "learning_rate": 4.3116216310938215e-05,
      "loss": 2.3289,
      "step": 39300
    },
    {
      "epoch": 0.1379455367469246,
      "grad_norm": 0.612820565700531,
      "learning_rate": 4.310570216941787e-05,
      "loss": 2.3392,
      "step": 39360
    },
    {
      "epoch": 0.1381558195773315,
      "grad_norm": 0.653420090675354,
      "learning_rate": 4.3095188027897526e-05,
      "loss": 2.3422,
      "step": 39420
    },
    {
      "epoch": 0.13836610240773842,
      "grad_norm": 0.6993768215179443,
      "learning_rate": 4.308467388637718e-05,
      "loss": 2.3276,
      "step": 39480
    },
    {
      "epoch": 0.1385763852381453,
      "grad_norm": 0.620498538017273,
      "learning_rate": 4.3074159744856836e-05,
      "loss": 2.327,
      "step": 39540
    },
    {
      "epoch": 0.1387866680685522,
      "grad_norm": 0.6352885365486145,
      "learning_rate": 4.306364560333649e-05,
      "loss": 2.3349,
      "step": 39600
    },
    {
      "epoch": 0.1389969508989591,
      "grad_norm": 0.6150739789009094,
      "learning_rate": 4.305313146181614e-05,
      "loss": 2.34,
      "step": 39660
    },
    {
      "epoch": 0.139207233729366,
      "grad_norm": 0.6017259359359741,
      "learning_rate": 4.30426173202958e-05,
      "loss": 2.3376,
      "step": 39720
    },
    {
      "epoch": 0.13941751655977289,
      "grad_norm": 0.6711851358413696,
      "learning_rate": 4.303210317877546e-05,
      "loss": 2.351,
      "step": 39780
    },
    {
      "epoch": 0.1396277993901798,
      "grad_norm": 0.7069224715232849,
      "learning_rate": 4.302158903725511e-05,
      "loss": 2.3402,
      "step": 39840
    },
    {
      "epoch": 0.1398380822205867,
      "grad_norm": 0.5861261487007141,
      "learning_rate": 4.301107489573477e-05,
      "loss": 2.3418,
      "step": 39900
    },
    {
      "epoch": 0.14004836505099358,
      "grad_norm": 0.6228580474853516,
      "learning_rate": 4.300056075421442e-05,
      "loss": 2.3228,
      "step": 39960
    },
    {
      "epoch": 0.1402586478814005,
      "grad_norm": 0.6415411233901978,
      "learning_rate": 4.299004661269407e-05,
      "loss": 2.3498,
      "step": 40020
    },
    {
      "epoch": 0.14046893071180738,
      "grad_norm": 0.6277108192443848,
      "learning_rate": 4.297953247117373e-05,
      "loss": 2.348,
      "step": 40080
    },
    {
      "epoch": 0.14067921354221427,
      "grad_norm": 0.6571968793869019,
      "learning_rate": 4.296901832965339e-05,
      "loss": 2.3394,
      "step": 40140
    },
    {
      "epoch": 0.1408894963726212,
      "grad_norm": 0.6432043313980103,
      "learning_rate": 4.295850418813304e-05,
      "loss": 2.3297,
      "step": 40200
    },
    {
      "epoch": 0.14109977920302808,
      "grad_norm": 0.6432401537895203,
      "learning_rate": 4.294799004661269e-05,
      "loss": 2.3301,
      "step": 40260
    },
    {
      "epoch": 0.14131006203343496,
      "grad_norm": 0.6234648823738098,
      "learning_rate": 4.293747590509235e-05,
      "loss": 2.3284,
      "step": 40320
    },
    {
      "epoch": 0.14152034486384188,
      "grad_norm": 0.6374291181564331,
      "learning_rate": 4.2926961763572e-05,
      "loss": 2.3368,
      "step": 40380
    },
    {
      "epoch": 0.14173062769424877,
      "grad_norm": 0.6450021862983704,
      "learning_rate": 4.291644762205166e-05,
      "loss": 2.3456,
      "step": 40440
    },
    {
      "epoch": 0.14194091052465566,
      "grad_norm": 0.5875654220581055,
      "learning_rate": 4.290593348053132e-05,
      "loss": 2.3504,
      "step": 40500
    },
    {
      "epoch": 0.14215119335506257,
      "grad_norm": 0.5929096937179565,
      "learning_rate": 4.289541933901097e-05,
      "loss": 2.3415,
      "step": 40560
    },
    {
      "epoch": 0.14236147618546946,
      "grad_norm": 0.6588292121887207,
      "learning_rate": 4.2884905197490624e-05,
      "loss": 2.3225,
      "step": 40620
    },
    {
      "epoch": 0.14257175901587635,
      "grad_norm": 0.6448934078216553,
      "learning_rate": 4.287439105597028e-05,
      "loss": 2.3272,
      "step": 40680
    },
    {
      "epoch": 0.14278204184628326,
      "grad_norm": 0.6730201840400696,
      "learning_rate": 4.2863876914449935e-05,
      "loss": 2.3358,
      "step": 40740
    },
    {
      "epoch": 0.14299232467669015,
      "grad_norm": 0.6035958528518677,
      "learning_rate": 4.2853362772929594e-05,
      "loss": 2.32,
      "step": 40800
    },
    {
      "epoch": 0.14320260750709704,
      "grad_norm": 0.5733596086502075,
      "learning_rate": 4.284284863140925e-05,
      "loss": 2.3386,
      "step": 40860
    },
    {
      "epoch": 0.14341289033750396,
      "grad_norm": 0.6223961710929871,
      "learning_rate": 4.2832334489888904e-05,
      "loss": 2.339,
      "step": 40920
    },
    {
      "epoch": 0.14362317316791084,
      "grad_norm": 0.6346945762634277,
      "learning_rate": 4.2821820348368556e-05,
      "loss": 2.3336,
      "step": 40980
    },
    {
      "epoch": 0.14383345599831773,
      "grad_norm": 0.6121115684509277,
      "learning_rate": 4.2811306206848215e-05,
      "loss": 2.3331,
      "step": 41040
    },
    {
      "epoch": 0.14404373882872462,
      "grad_norm": 0.6739269495010376,
      "learning_rate": 4.280079206532787e-05,
      "loss": 2.3314,
      "step": 41100
    },
    {
      "epoch": 0.14425402165913154,
      "grad_norm": 0.6606531143188477,
      "learning_rate": 4.2790277923807525e-05,
      "loss": 2.3389,
      "step": 41160
    },
    {
      "epoch": 0.14446430448953843,
      "grad_norm": 0.6564496159553528,
      "learning_rate": 4.277976378228718e-05,
      "loss": 2.3303,
      "step": 41220
    },
    {
      "epoch": 0.1446745873199453,
      "grad_norm": 0.6618686318397522,
      "learning_rate": 4.2769249640766836e-05,
      "loss": 2.3352,
      "step": 41280
    },
    {
      "epoch": 0.14488487015035223,
      "grad_norm": 0.6552087068557739,
      "learning_rate": 4.275873549924649e-05,
      "loss": 2.3315,
      "step": 41340
    },
    {
      "epoch": 0.14509515298075912,
      "grad_norm": 0.6370447874069214,
      "learning_rate": 4.274822135772614e-05,
      "loss": 2.3395,
      "step": 41400
    },
    {
      "epoch": 0.145305435811166,
      "grad_norm": 0.6118485927581787,
      "learning_rate": 4.27377072162058e-05,
      "loss": 2.3397,
      "step": 41460
    },
    {
      "epoch": 0.14551571864157292,
      "grad_norm": 0.6253863573074341,
      "learning_rate": 4.272719307468546e-05,
      "loss": 2.3379,
      "step": 41520
    },
    {
      "epoch": 0.1457260014719798,
      "grad_norm": 0.6244592666625977,
      "learning_rate": 4.271667893316511e-05,
      "loss": 2.3415,
      "step": 41580
    },
    {
      "epoch": 0.1459362843023867,
      "grad_norm": 0.6361585259437561,
      "learning_rate": 4.270616479164477e-05,
      "loss": 2.3334,
      "step": 41640
    },
    {
      "epoch": 0.14614656713279361,
      "grad_norm": 0.6284440159797668,
      "learning_rate": 4.2695825885816425e-05,
      "loss": 2.3311,
      "step": 41700
    },
    {
      "epoch": 0.1463568499632005,
      "grad_norm": 0.6048732995986938,
      "learning_rate": 4.2685311744296083e-05,
      "loss": 2.3239,
      "step": 41760
    },
    {
      "epoch": 0.1465671327936074,
      "grad_norm": 0.6068998575210571,
      "learning_rate": 4.2674797602775735e-05,
      "loss": 2.3411,
      "step": 41820
    },
    {
      "epoch": 0.1467774156240143,
      "grad_norm": 0.6173762083053589,
      "learning_rate": 4.266428346125539e-05,
      "loss": 2.3328,
      "step": 41880
    },
    {
      "epoch": 0.1469876984544212,
      "grad_norm": 0.6197360157966614,
      "learning_rate": 4.2653769319735046e-05,
      "loss": 2.329,
      "step": 41940
    },
    {
      "epoch": 0.14719798128482808,
      "grad_norm": 0.6429170966148376,
      "learning_rate": 4.2643255178214705e-05,
      "loss": 2.3444,
      "step": 42000
    },
    {
      "epoch": 0.147408264115235,
      "grad_norm": 0.6106884479522705,
      "learning_rate": 4.2632741036694356e-05,
      "loss": 2.3446,
      "step": 42060
    },
    {
      "epoch": 0.1476185469456419,
      "grad_norm": 0.6438376903533936,
      "learning_rate": 4.262222689517401e-05,
      "loss": 2.3502,
      "step": 42120
    },
    {
      "epoch": 0.14782882977604878,
      "grad_norm": 0.6259479522705078,
      "learning_rate": 4.261171275365367e-05,
      "loss": 2.3399,
      "step": 42180
    },
    {
      "epoch": 0.1480391126064557,
      "grad_norm": 0.5971313714981079,
      "learning_rate": 4.260119861213332e-05,
      "loss": 2.325,
      "step": 42240
    },
    {
      "epoch": 0.14824939543686258,
      "grad_norm": 0.6188226342201233,
      "learning_rate": 4.259068447061297e-05,
      "loss": 2.3217,
      "step": 42300
    },
    {
      "epoch": 0.14845967826726947,
      "grad_norm": 0.6155485510826111,
      "learning_rate": 4.2580170329092636e-05,
      "loss": 2.336,
      "step": 42360
    },
    {
      "epoch": 0.14866996109767638,
      "grad_norm": 0.5590870976448059,
      "learning_rate": 4.256965618757229e-05,
      "loss": 2.3412,
      "step": 42420
    },
    {
      "epoch": 0.14888024392808327,
      "grad_norm": 0.686099112033844,
      "learning_rate": 4.255914204605194e-05,
      "loss": 2.3134,
      "step": 42480
    },
    {
      "epoch": 0.14909052675849016,
      "grad_norm": 0.6008979082107544,
      "learning_rate": 4.25486279045316e-05,
      "loss": 2.3503,
      "step": 42540
    },
    {
      "epoch": 0.14930080958889708,
      "grad_norm": 0.6094452738761902,
      "learning_rate": 4.253811376301125e-05,
      "loss": 2.3473,
      "step": 42600
    },
    {
      "epoch": 0.14951109241930396,
      "grad_norm": 0.6370200514793396,
      "learning_rate": 4.252759962149091e-05,
      "loss": 2.3404,
      "step": 42660
    },
    {
      "epoch": 0.14972137524971085,
      "grad_norm": 0.653773844242096,
      "learning_rate": 4.251708547997056e-05,
      "loss": 2.3368,
      "step": 42720
    },
    {
      "epoch": 0.14993165808011777,
      "grad_norm": 0.6160172820091248,
      "learning_rate": 4.250657133845022e-05,
      "loss": 2.3275,
      "step": 42780
    },
    {
      "epoch": 0.15014194091052466,
      "grad_norm": 0.5883677005767822,
      "learning_rate": 4.249605719692987e-05,
      "loss": 2.324,
      "step": 42840
    },
    {
      "epoch": 0.15035222374093155,
      "grad_norm": 0.6456954479217529,
      "learning_rate": 4.2485543055409524e-05,
      "loss": 2.3437,
      "step": 42900
    },
    {
      "epoch": 0.15056250657133846,
      "grad_norm": 0.6041179299354553,
      "learning_rate": 4.247502891388918e-05,
      "loss": 2.3269,
      "step": 42960
    },
    {
      "epoch": 0.15077278940174535,
      "grad_norm": 0.5934765338897705,
      "learning_rate": 4.246451477236884e-05,
      "loss": 2.3242,
      "step": 43020
    },
    {
      "epoch": 0.15098307223215224,
      "grad_norm": 0.6217716336250305,
      "learning_rate": 4.245400063084849e-05,
      "loss": 2.3276,
      "step": 43080
    },
    {
      "epoch": 0.15119335506255915,
      "grad_norm": 0.58519446849823,
      "learning_rate": 4.244348648932815e-05,
      "loss": 2.3267,
      "step": 43140
    },
    {
      "epoch": 0.15140363789296604,
      "grad_norm": 0.6344335675239563,
      "learning_rate": 4.24329723478078e-05,
      "loss": 2.3307,
      "step": 43200
    },
    {
      "epoch": 0.15161392072337293,
      "grad_norm": 0.6420978903770447,
      "learning_rate": 4.2422458206287455e-05,
      "loss": 2.332,
      "step": 43260
    },
    {
      "epoch": 0.15182420355377985,
      "grad_norm": 0.6562826037406921,
      "learning_rate": 4.2411944064767114e-05,
      "loss": 2.3135,
      "step": 43320
    },
    {
      "epoch": 0.15203448638418673,
      "grad_norm": 0.6492423415184021,
      "learning_rate": 4.240142992324677e-05,
      "loss": 2.3133,
      "step": 43380
    },
    {
      "epoch": 0.15224476921459362,
      "grad_norm": 0.6478837132453918,
      "learning_rate": 4.2390915781726424e-05,
      "loss": 2.3247,
      "step": 43440
    },
    {
      "epoch": 0.15245505204500054,
      "grad_norm": 0.5964265465736389,
      "learning_rate": 4.238040164020608e-05,
      "loss": 2.3222,
      "step": 43500
    },
    {
      "epoch": 0.15266533487540743,
      "grad_norm": 0.6163307428359985,
      "learning_rate": 4.2369887498685735e-05,
      "loss": 2.3225,
      "step": 43560
    },
    {
      "epoch": 0.15287561770581432,
      "grad_norm": 0.6812646389007568,
      "learning_rate": 4.235937335716539e-05,
      "loss": 2.3266,
      "step": 43620
    },
    {
      "epoch": 0.15308590053622123,
      "grad_norm": 0.6045469045639038,
      "learning_rate": 4.2348859215645046e-05,
      "loss": 2.3195,
      "step": 43680
    },
    {
      "epoch": 0.15329618336662812,
      "grad_norm": 0.6138345003128052,
      "learning_rate": 4.2338345074124704e-05,
      "loss": 2.3475,
      "step": 43740
    },
    {
      "epoch": 0.153506466197035,
      "grad_norm": 0.6312504410743713,
      "learning_rate": 4.2327830932604356e-05,
      "loss": 2.3244,
      "step": 43800
    },
    {
      "epoch": 0.1537167490274419,
      "grad_norm": 0.5887283086776733,
      "learning_rate": 4.231731679108401e-05,
      "loss": 2.3253,
      "step": 43860
    },
    {
      "epoch": 0.1539270318578488,
      "grad_norm": 0.644916296005249,
      "learning_rate": 4.230680264956367e-05,
      "loss": 2.3534,
      "step": 43920
    },
    {
      "epoch": 0.1541373146882557,
      "grad_norm": 0.6598151922225952,
      "learning_rate": 4.229628850804332e-05,
      "loss": 2.3247,
      "step": 43980
    },
    {
      "epoch": 0.1543475975186626,
      "grad_norm": 0.6058512330055237,
      "learning_rate": 4.228577436652297e-05,
      "loss": 2.3278,
      "step": 44040
    },
    {
      "epoch": 0.1545578803490695,
      "grad_norm": 0.7185304760932922,
      "learning_rate": 4.2275260225002636e-05,
      "loss": 2.3231,
      "step": 44100
    },
    {
      "epoch": 0.1547681631794764,
      "grad_norm": 0.6237541437149048,
      "learning_rate": 4.226474608348229e-05,
      "loss": 2.3131,
      "step": 44160
    },
    {
      "epoch": 0.15497844600988328,
      "grad_norm": 0.6727291941642761,
      "learning_rate": 4.225423194196194e-05,
      "loss": 2.3176,
      "step": 44220
    },
    {
      "epoch": 0.1551887288402902,
      "grad_norm": 0.6596560478210449,
      "learning_rate": 4.22437178004416e-05,
      "loss": 2.324,
      "step": 44280
    },
    {
      "epoch": 0.15539901167069708,
      "grad_norm": 0.6178886890411377,
      "learning_rate": 4.223320365892125e-05,
      "loss": 2.3304,
      "step": 44340
    },
    {
      "epoch": 0.15560929450110397,
      "grad_norm": 0.6122422814369202,
      "learning_rate": 4.2222864753092914e-05,
      "loss": 2.3411,
      "step": 44400
    },
    {
      "epoch": 0.1558195773315109,
      "grad_norm": 0.6111167669296265,
      "learning_rate": 4.2212350611572566e-05,
      "loss": 2.327,
      "step": 44460
    },
    {
      "epoch": 0.15602986016191778,
      "grad_norm": 0.6797990798950195,
      "learning_rate": 4.220183647005222e-05,
      "loss": 2.3295,
      "step": 44520
    },
    {
      "epoch": 0.15624014299232467,
      "grad_norm": 0.5918498039245605,
      "learning_rate": 4.219132232853188e-05,
      "loss": 2.3297,
      "step": 44580
    },
    {
      "epoch": 0.15645042582273158,
      "grad_norm": 0.6331505179405212,
      "learning_rate": 4.2180808187011535e-05,
      "loss": 2.3268,
      "step": 44640
    },
    {
      "epoch": 0.15666070865313847,
      "grad_norm": 0.5968581438064575,
      "learning_rate": 4.217029404549119e-05,
      "loss": 2.3499,
      "step": 44700
    },
    {
      "epoch": 0.15687099148354536,
      "grad_norm": 0.7291754484176636,
      "learning_rate": 4.215977990397084e-05,
      "loss": 2.3382,
      "step": 44760
    },
    {
      "epoch": 0.15708127431395227,
      "grad_norm": 0.62226802110672,
      "learning_rate": 4.21492657624505e-05,
      "loss": 2.3379,
      "step": 44820
    },
    {
      "epoch": 0.15729155714435916,
      "grad_norm": 0.6174754500389099,
      "learning_rate": 4.2138751620930156e-05,
      "loss": 2.3331,
      "step": 44880
    },
    {
      "epoch": 0.15750183997476605,
      "grad_norm": 0.6238417625427246,
      "learning_rate": 4.212823747940981e-05,
      "loss": 2.3259,
      "step": 44940
    },
    {
      "epoch": 0.15771212280517297,
      "grad_norm": 0.6212115287780762,
      "learning_rate": 4.211772333788947e-05,
      "loss": 2.3301,
      "step": 45000
    },
    {
      "epoch": 0.15792240563557985,
      "grad_norm": 0.6135154962539673,
      "learning_rate": 4.210720919636912e-05,
      "loss": 2.3235,
      "step": 45060
    },
    {
      "epoch": 0.15813268846598674,
      "grad_norm": 0.6578890085220337,
      "learning_rate": 4.209669505484877e-05,
      "loss": 2.3273,
      "step": 45120
    },
    {
      "epoch": 0.15834297129639366,
      "grad_norm": 0.6077913045883179,
      "learning_rate": 4.208618091332843e-05,
      "loss": 2.3394,
      "step": 45180
    },
    {
      "epoch": 0.15855325412680055,
      "grad_norm": 0.6092454791069031,
      "learning_rate": 4.207566677180809e-05,
      "loss": 2.332,
      "step": 45240
    },
    {
      "epoch": 0.15876353695720744,
      "grad_norm": 0.6163653135299683,
      "learning_rate": 4.206515263028774e-05,
      "loss": 2.3237,
      "step": 45300
    },
    {
      "epoch": 0.15897381978761435,
      "grad_norm": 0.5905401706695557,
      "learning_rate": 4.205463848876739e-05,
      "loss": 2.3199,
      "step": 45360
    },
    {
      "epoch": 0.15918410261802124,
      "grad_norm": 0.6775098443031311,
      "learning_rate": 4.204412434724705e-05,
      "loss": 2.323,
      "step": 45420
    },
    {
      "epoch": 0.15939438544842813,
      "grad_norm": 0.5804296135902405,
      "learning_rate": 4.20336102057267e-05,
      "loss": 2.3197,
      "step": 45480
    },
    {
      "epoch": 0.15960466827883504,
      "grad_norm": 0.5796772241592407,
      "learning_rate": 4.2023096064206354e-05,
      "loss": 2.3177,
      "step": 45540
    },
    {
      "epoch": 0.15981495110924193,
      "grad_norm": 0.6753705739974976,
      "learning_rate": 4.201258192268602e-05,
      "loss": 2.3467,
      "step": 45600
    },
    {
      "epoch": 0.16002523393964882,
      "grad_norm": 0.580149233341217,
      "learning_rate": 4.200206778116567e-05,
      "loss": 2.3185,
      "step": 45660
    },
    {
      "epoch": 0.16023551677005574,
      "grad_norm": 0.5798102617263794,
      "learning_rate": 4.1991553639645324e-05,
      "loss": 2.3195,
      "step": 45720
    },
    {
      "epoch": 0.16044579960046262,
      "grad_norm": 0.6332999467849731,
      "learning_rate": 4.198103949812498e-05,
      "loss": 2.3237,
      "step": 45780
    },
    {
      "epoch": 0.1606560824308695,
      "grad_norm": 0.5849390029907227,
      "learning_rate": 4.1970525356604634e-05,
      "loss": 2.3137,
      "step": 45840
    },
    {
      "epoch": 0.16086636526127643,
      "grad_norm": 0.5851735472679138,
      "learning_rate": 4.1960011215084286e-05,
      "loss": 2.3095,
      "step": 45900
    },
    {
      "epoch": 0.16107664809168332,
      "grad_norm": 0.6217033267021179,
      "learning_rate": 4.194949707356395e-05,
      "loss": 2.343,
      "step": 45960
    },
    {
      "epoch": 0.1612869309220902,
      "grad_norm": 0.6043165922164917,
      "learning_rate": 4.19389829320436e-05,
      "loss": 2.3256,
      "step": 46020
    },
    {
      "epoch": 0.16149721375249712,
      "grad_norm": 0.608424186706543,
      "learning_rate": 4.1928468790523255e-05,
      "loss": 2.3435,
      "step": 46080
    },
    {
      "epoch": 0.161707496582904,
      "grad_norm": 0.6008337736129761,
      "learning_rate": 4.1917954649002914e-05,
      "loss": 2.3228,
      "step": 46140
    },
    {
      "epoch": 0.1619177794133109,
      "grad_norm": 0.5890358090400696,
      "learning_rate": 4.1907440507482566e-05,
      "loss": 2.3246,
      "step": 46200
    },
    {
      "epoch": 0.1621280622437178,
      "grad_norm": 0.5866249799728394,
      "learning_rate": 4.189692636596222e-05,
      "loss": 2.3048,
      "step": 46260
    },
    {
      "epoch": 0.1623383450741247,
      "grad_norm": 0.6539163589477539,
      "learning_rate": 4.1886412224441876e-05,
      "loss": 2.3154,
      "step": 46320
    },
    {
      "epoch": 0.1625486279045316,
      "grad_norm": 0.6630322933197021,
      "learning_rate": 4.1875898082921535e-05,
      "loss": 2.332,
      "step": 46380
    },
    {
      "epoch": 0.1627589107349385,
      "grad_norm": 0.6159778237342834,
      "learning_rate": 4.186538394140119e-05,
      "loss": 2.3144,
      "step": 46440
    },
    {
      "epoch": 0.1629691935653454,
      "grad_norm": 0.6124069690704346,
      "learning_rate": 4.185504503557285e-05,
      "loss": 2.3312,
      "step": 46500
    },
    {
      "epoch": 0.16317947639575228,
      "grad_norm": 0.5926678776741028,
      "learning_rate": 4.18445308940525e-05,
      "loss": 2.3286,
      "step": 46560
    },
    {
      "epoch": 0.16338975922615917,
      "grad_norm": 0.6444182395935059,
      "learning_rate": 4.1834016752532155e-05,
      "loss": 2.3426,
      "step": 46620
    },
    {
      "epoch": 0.1636000420565661,
      "grad_norm": 0.5816701650619507,
      "learning_rate": 4.182350261101181e-05,
      "loss": 2.3237,
      "step": 46680
    },
    {
      "epoch": 0.16381032488697297,
      "grad_norm": 0.6558911800384521,
      "learning_rate": 4.1812988469491465e-05,
      "loss": 2.3219,
      "step": 46740
    },
    {
      "epoch": 0.16402060771737986,
      "grad_norm": 0.5754097700119019,
      "learning_rate": 4.1802474327971124e-05,
      "loss": 2.3159,
      "step": 46800
    },
    {
      "epoch": 0.16423089054778678,
      "grad_norm": 0.6297584176063538,
      "learning_rate": 4.179196018645078e-05,
      "loss": 2.3223,
      "step": 46860
    },
    {
      "epoch": 0.16444117337819367,
      "grad_norm": 0.6096706390380859,
      "learning_rate": 4.1781446044930434e-05,
      "loss": 2.324,
      "step": 46920
    },
    {
      "epoch": 0.16465145620860056,
      "grad_norm": 0.5916000008583069,
      "learning_rate": 4.1770931903410086e-05,
      "loss": 2.3375,
      "step": 46980
    },
    {
      "epoch": 0.16486173903900747,
      "grad_norm": 0.5777989029884338,
      "learning_rate": 4.1760417761889745e-05,
      "loss": 2.3196,
      "step": 47040
    },
    {
      "epoch": 0.16507202186941436,
      "grad_norm": 0.6046702861785889,
      "learning_rate": 4.17499036203694e-05,
      "loss": 2.329,
      "step": 47100
    },
    {
      "epoch": 0.16528230469982125,
      "grad_norm": 0.5970363020896912,
      "learning_rate": 4.1739389478849056e-05,
      "loss": 2.3165,
      "step": 47160
    },
    {
      "epoch": 0.16549258753022816,
      "grad_norm": 0.5796226263046265,
      "learning_rate": 4.172887533732871e-05,
      "loss": 2.3139,
      "step": 47220
    },
    {
      "epoch": 0.16570287036063505,
      "grad_norm": 0.6245923042297363,
      "learning_rate": 4.1718361195808366e-05,
      "loss": 2.3177,
      "step": 47280
    },
    {
      "epoch": 0.16591315319104194,
      "grad_norm": 0.6365753412246704,
      "learning_rate": 4.170784705428802e-05,
      "loss": 2.3105,
      "step": 47340
    },
    {
      "epoch": 0.16612343602144886,
      "grad_norm": 0.6340218186378479,
      "learning_rate": 4.169733291276767e-05,
      "loss": 2.3289,
      "step": 47400
    },
    {
      "epoch": 0.16633371885185574,
      "grad_norm": 0.6164852976799011,
      "learning_rate": 4.1686818771247335e-05,
      "loss": 2.3347,
      "step": 47460
    },
    {
      "epoch": 0.16654400168226263,
      "grad_norm": 0.6079014539718628,
      "learning_rate": 4.167630462972699e-05,
      "loss": 2.3058,
      "step": 47520
    },
    {
      "epoch": 0.16675428451266955,
      "grad_norm": 0.6427564024925232,
      "learning_rate": 4.166579048820664e-05,
      "loss": 2.2973,
      "step": 47580
    },
    {
      "epoch": 0.16696456734307644,
      "grad_norm": 0.6605886220932007,
      "learning_rate": 4.16552763466863e-05,
      "loss": 2.3225,
      "step": 47640
    },
    {
      "epoch": 0.16717485017348332,
      "grad_norm": 0.6974831223487854,
      "learning_rate": 4.164476220516595e-05,
      "loss": 2.3205,
      "step": 47700
    },
    {
      "epoch": 0.16738513300389024,
      "grad_norm": 0.6362879872322083,
      "learning_rate": 4.16342480636456e-05,
      "loss": 2.3157,
      "step": 47760
    },
    {
      "epoch": 0.16759541583429713,
      "grad_norm": 0.5881654024124146,
      "learning_rate": 4.162373392212526e-05,
      "loss": 2.3122,
      "step": 47820
    },
    {
      "epoch": 0.16780569866470402,
      "grad_norm": 0.6076338291168213,
      "learning_rate": 4.161321978060492e-05,
      "loss": 2.3195,
      "step": 47880
    },
    {
      "epoch": 0.16801598149511093,
      "grad_norm": 0.6087121367454529,
      "learning_rate": 4.160270563908457e-05,
      "loss": 2.3136,
      "step": 47940
    },
    {
      "epoch": 0.16822626432551782,
      "grad_norm": 0.5977998971939087,
      "learning_rate": 4.159219149756422e-05,
      "loss": 2.3206,
      "step": 48000
    },
    {
      "epoch": 0.1684365471559247,
      "grad_norm": 0.638401985168457,
      "learning_rate": 4.158167735604388e-05,
      "loss": 2.3223,
      "step": 48060
    },
    {
      "epoch": 0.16864682998633163,
      "grad_norm": 0.5896310806274414,
      "learning_rate": 4.157116321452353e-05,
      "loss": 2.3258,
      "step": 48120
    },
    {
      "epoch": 0.1688571128167385,
      "grad_norm": 0.6041130423545837,
      "learning_rate": 4.156064907300319e-05,
      "loss": 2.3068,
      "step": 48180
    },
    {
      "epoch": 0.1690673956471454,
      "grad_norm": 0.6105507016181946,
      "learning_rate": 4.155013493148285e-05,
      "loss": 2.3208,
      "step": 48240
    },
    {
      "epoch": 0.16927767847755232,
      "grad_norm": 0.6303457021713257,
      "learning_rate": 4.15396207899625e-05,
      "loss": 2.3028,
      "step": 48300
    },
    {
      "epoch": 0.1694879613079592,
      "grad_norm": 0.6402522325515747,
      "learning_rate": 4.1529106648442154e-05,
      "loss": 2.3328,
      "step": 48360
    },
    {
      "epoch": 0.1696982441383661,
      "grad_norm": 0.622160017490387,
      "learning_rate": 4.151859250692181e-05,
      "loss": 2.314,
      "step": 48420
    },
    {
      "epoch": 0.169908526968773,
      "grad_norm": 0.5710818767547607,
      "learning_rate": 4.1508078365401465e-05,
      "loss": 2.3165,
      "step": 48480
    },
    {
      "epoch": 0.1701188097991799,
      "grad_norm": 0.6494461297988892,
      "learning_rate": 4.1497564223881124e-05,
      "loss": 2.3201,
      "step": 48540
    },
    {
      "epoch": 0.1703290926295868,
      "grad_norm": 0.586834728717804,
      "learning_rate": 4.1487050082360775e-05,
      "loss": 2.3251,
      "step": 48600
    },
    {
      "epoch": 0.1705393754599937,
      "grad_norm": 0.6685080528259277,
      "learning_rate": 4.1476535940840434e-05,
      "loss": 2.3327,
      "step": 48660
    },
    {
      "epoch": 0.1707496582904006,
      "grad_norm": 0.5974341034889221,
      "learning_rate": 4.1466021799320086e-05,
      "loss": 2.3246,
      "step": 48720
    },
    {
      "epoch": 0.17095994112080748,
      "grad_norm": 0.5994948744773865,
      "learning_rate": 4.1455507657799745e-05,
      "loss": 2.3205,
      "step": 48780
    },
    {
      "epoch": 0.1711702239512144,
      "grad_norm": 0.6069355607032776,
      "learning_rate": 4.1444993516279397e-05,
      "loss": 2.3273,
      "step": 48840
    },
    {
      "epoch": 0.17138050678162128,
      "grad_norm": 0.6142827272415161,
      "learning_rate": 4.1434479374759055e-05,
      "loss": 2.3084,
      "step": 48900
    },
    {
      "epoch": 0.17159078961202817,
      "grad_norm": 0.6005485653877258,
      "learning_rate": 4.142396523323871e-05,
      "loss": 2.3108,
      "step": 48960
    },
    {
      "epoch": 0.1718010724424351,
      "grad_norm": 0.6493135690689087,
      "learning_rate": 4.1413451091718366e-05,
      "loss": 2.3054,
      "step": 49020
    },
    {
      "epoch": 0.17201135527284198,
      "grad_norm": 0.628297746181488,
      "learning_rate": 4.140293695019802e-05,
      "loss": 2.3217,
      "step": 49080
    },
    {
      "epoch": 0.17222163810324886,
      "grad_norm": 0.5488650798797607,
      "learning_rate": 4.139242280867767e-05,
      "loss": 2.3233,
      "step": 49140
    },
    {
      "epoch": 0.17243192093365578,
      "grad_norm": 0.6183356642723083,
      "learning_rate": 4.138190866715733e-05,
      "loss": 2.3258,
      "step": 49200
    },
    {
      "epoch": 0.17264220376406267,
      "grad_norm": 0.6453129649162292,
      "learning_rate": 4.1371569761328985e-05,
      "loss": 2.3094,
      "step": 49260
    },
    {
      "epoch": 0.17285248659446956,
      "grad_norm": 0.638854444026947,
      "learning_rate": 4.1361055619808644e-05,
      "loss": 2.3394,
      "step": 49320
    },
    {
      "epoch": 0.17306276942487647,
      "grad_norm": 0.6387530565261841,
      "learning_rate": 4.13505414782883e-05,
      "loss": 2.3375,
      "step": 49380
    },
    {
      "epoch": 0.17327305225528336,
      "grad_norm": 0.6037068963050842,
      "learning_rate": 4.1340027336767955e-05,
      "loss": 2.3175,
      "step": 49440
    },
    {
      "epoch": 0.17348333508569025,
      "grad_norm": 0.6161598563194275,
      "learning_rate": 4.132951319524761e-05,
      "loss": 2.3104,
      "step": 49500
    },
    {
      "epoch": 0.17369361791609714,
      "grad_norm": 0.5830373167991638,
      "learning_rate": 4.1318999053727265e-05,
      "loss": 2.3251,
      "step": 49560
    },
    {
      "epoch": 0.17390390074650405,
      "grad_norm": 0.6282650828361511,
      "learning_rate": 4.130848491220692e-05,
      "loss": 2.3182,
      "step": 49620
    },
    {
      "epoch": 0.17411418357691094,
      "grad_norm": 0.6157172918319702,
      "learning_rate": 4.1297970770686576e-05,
      "loss": 2.323,
      "step": 49680
    },
    {
      "epoch": 0.17432446640731783,
      "grad_norm": 0.5994360446929932,
      "learning_rate": 4.1287456629166234e-05,
      "loss": 2.3281,
      "step": 49740
    },
    {
      "epoch": 0.17453474923772475,
      "grad_norm": 0.6262275576591492,
      "learning_rate": 4.1276942487645886e-05,
      "loss": 2.3086,
      "step": 49800
    },
    {
      "epoch": 0.17474503206813163,
      "grad_norm": 0.672465443611145,
      "learning_rate": 4.126642834612554e-05,
      "loss": 2.2974,
      "step": 49860
    },
    {
      "epoch": 0.17495531489853852,
      "grad_norm": 0.586807370185852,
      "learning_rate": 4.12559142046052e-05,
      "loss": 2.3197,
      "step": 49920
    },
    {
      "epoch": 0.17516559772894544,
      "grad_norm": 0.6669567823410034,
      "learning_rate": 4.124540006308485e-05,
      "loss": 2.3177,
      "step": 49980
    },
    {
      "epoch": 0.17537588055935233,
      "grad_norm": 0.5670884847640991,
      "learning_rate": 4.123488592156451e-05,
      "loss": 2.3208,
      "step": 50040
    },
    {
      "epoch": 0.17558616338975921,
      "grad_norm": 0.5929347276687622,
      "learning_rate": 4.1224371780044166e-05,
      "loss": 2.324,
      "step": 50100
    },
    {
      "epoch": 0.17579644622016613,
      "grad_norm": 0.5971206426620483,
      "learning_rate": 4.121385763852382e-05,
      "loss": 2.3158,
      "step": 50160
    },
    {
      "epoch": 0.17600672905057302,
      "grad_norm": 0.5975629687309265,
      "learning_rate": 4.120334349700347e-05,
      "loss": 2.3187,
      "step": 50220
    },
    {
      "epoch": 0.1762170118809799,
      "grad_norm": 0.5721208453178406,
      "learning_rate": 4.119282935548313e-05,
      "loss": 2.3269,
      "step": 50280
    },
    {
      "epoch": 0.17642729471138682,
      "grad_norm": 0.593453586101532,
      "learning_rate": 4.118231521396278e-05,
      "loss": 2.3168,
      "step": 50340
    },
    {
      "epoch": 0.1766375775417937,
      "grad_norm": 0.6251164078712463,
      "learning_rate": 4.117180107244244e-05,
      "loss": 2.334,
      "step": 50400
    },
    {
      "epoch": 0.1768478603722006,
      "grad_norm": 0.616355836391449,
      "learning_rate": 4.116128693092209e-05,
      "loss": 2.3135,
      "step": 50460
    },
    {
      "epoch": 0.17705814320260752,
      "grad_norm": 0.6078146696090698,
      "learning_rate": 4.115077278940175e-05,
      "loss": 2.3236,
      "step": 50520
    },
    {
      "epoch": 0.1772684260330144,
      "grad_norm": 0.5824019908905029,
      "learning_rate": 4.11402586478814e-05,
      "loss": 2.3015,
      "step": 50580
    },
    {
      "epoch": 0.1774787088634213,
      "grad_norm": 0.6472707986831665,
      "learning_rate": 4.1129744506361053e-05,
      "loss": 2.2933,
      "step": 50640
    },
    {
      "epoch": 0.1776889916938282,
      "grad_norm": 0.599815845489502,
      "learning_rate": 4.111923036484071e-05,
      "loss": 2.3065,
      "step": 50700
    },
    {
      "epoch": 0.1778992745242351,
      "grad_norm": 0.591781735420227,
      "learning_rate": 4.110871622332037e-05,
      "loss": 2.3104,
      "step": 50760
    },
    {
      "epoch": 0.17810955735464198,
      "grad_norm": 0.5977340340614319,
      "learning_rate": 4.109820208180002e-05,
      "loss": 2.3213,
      "step": 50820
    },
    {
      "epoch": 0.1783198401850489,
      "grad_norm": 0.6299940347671509,
      "learning_rate": 4.108768794027968e-05,
      "loss": 2.3093,
      "step": 50880
    },
    {
      "epoch": 0.1785301230154558,
      "grad_norm": 0.6001994013786316,
      "learning_rate": 4.107717379875933e-05,
      "loss": 2.314,
      "step": 50940
    },
    {
      "epoch": 0.17874040584586268,
      "grad_norm": 0.6062608361244202,
      "learning_rate": 4.1066659657238985e-05,
      "loss": 2.3098,
      "step": 51000
    },
    {
      "epoch": 0.1789506886762696,
      "grad_norm": 0.6256763935089111,
      "learning_rate": 4.1056145515718644e-05,
      "loss": 2.3324,
      "step": 51060
    },
    {
      "epoch": 0.17916097150667648,
      "grad_norm": 0.5938847064971924,
      "learning_rate": 4.10456313741983e-05,
      "loss": 2.3,
      "step": 51120
    },
    {
      "epoch": 0.17937125433708337,
      "grad_norm": 0.5945448875427246,
      "learning_rate": 4.1035117232677954e-05,
      "loss": 2.3263,
      "step": 51180
    },
    {
      "epoch": 0.17958153716749028,
      "grad_norm": 0.5838569402694702,
      "learning_rate": 4.1024603091157606e-05,
      "loss": 2.3236,
      "step": 51240
    },
    {
      "epoch": 0.17979181999789717,
      "grad_norm": 0.6150086522102356,
      "learning_rate": 4.1014088949637265e-05,
      "loss": 2.3205,
      "step": 51300
    },
    {
      "epoch": 0.18000210282830406,
      "grad_norm": 0.6472041606903076,
      "learning_rate": 4.100357480811692e-05,
      "loss": 2.3125,
      "step": 51360
    },
    {
      "epoch": 0.18021238565871098,
      "grad_norm": 0.6316036581993103,
      "learning_rate": 4.0993060666596575e-05,
      "loss": 2.3159,
      "step": 51420
    },
    {
      "epoch": 0.18042266848911787,
      "grad_norm": 0.5957314968109131,
      "learning_rate": 4.0982546525076234e-05,
      "loss": 2.3227,
      "step": 51480
    },
    {
      "epoch": 0.18063295131952475,
      "grad_norm": 0.5779405236244202,
      "learning_rate": 4.0972032383555886e-05,
      "loss": 2.314,
      "step": 51540
    },
    {
      "epoch": 0.18084323414993167,
      "grad_norm": 0.5753687024116516,
      "learning_rate": 4.096151824203554e-05,
      "loss": 2.3037,
      "step": 51600
    },
    {
      "epoch": 0.18105351698033856,
      "grad_norm": 0.6188888549804688,
      "learning_rate": 4.0951004100515197e-05,
      "loss": 2.3152,
      "step": 51660
    },
    {
      "epoch": 0.18126379981074545,
      "grad_norm": 0.5780879259109497,
      "learning_rate": 4.094048995899485e-05,
      "loss": 2.3216,
      "step": 51720
    },
    {
      "epoch": 0.18147408264115236,
      "grad_norm": 0.6299008727073669,
      "learning_rate": 4.09299758174745e-05,
      "loss": 2.3047,
      "step": 51780
    },
    {
      "epoch": 0.18168436547155925,
      "grad_norm": 0.575741708278656,
      "learning_rate": 4.0919461675954166e-05,
      "loss": 2.305,
      "step": 51840
    },
    {
      "epoch": 0.18189464830196614,
      "grad_norm": 0.6241801977157593,
      "learning_rate": 4.090894753443382e-05,
      "loss": 2.2994,
      "step": 51900
    },
    {
      "epoch": 0.18210493113237305,
      "grad_norm": 0.5695961713790894,
      "learning_rate": 4.089843339291347e-05,
      "loss": 2.3188,
      "step": 51960
    },
    {
      "epoch": 0.18231521396277994,
      "grad_norm": 0.6108229160308838,
      "learning_rate": 4.088791925139313e-05,
      "loss": 2.3052,
      "step": 52020
    },
    {
      "epoch": 0.18252549679318683,
      "grad_norm": 0.6097207069396973,
      "learning_rate": 4.087740510987278e-05,
      "loss": 2.3067,
      "step": 52080
    },
    {
      "epoch": 0.18273577962359375,
      "grad_norm": 0.5715197324752808,
      "learning_rate": 4.086689096835243e-05,
      "loss": 2.307,
      "step": 52140
    },
    {
      "epoch": 0.18294606245400064,
      "grad_norm": 0.5958813428878784,
      "learning_rate": 4.085637682683209e-05,
      "loss": 2.3099,
      "step": 52200
    },
    {
      "epoch": 0.18315634528440752,
      "grad_norm": 0.6055279970169067,
      "learning_rate": 4.084586268531175e-05,
      "loss": 2.3367,
      "step": 52260
    },
    {
      "epoch": 0.1833666281148144,
      "grad_norm": 0.6448001861572266,
      "learning_rate": 4.08353485437914e-05,
      "loss": 2.3191,
      "step": 52320
    },
    {
      "epoch": 0.18357691094522133,
      "grad_norm": 0.5830652713775635,
      "learning_rate": 4.082483440227105e-05,
      "loss": 2.3125,
      "step": 52380
    },
    {
      "epoch": 0.18378719377562822,
      "grad_norm": 0.645237922668457,
      "learning_rate": 4.081432026075071e-05,
      "loss": 2.3147,
      "step": 52440
    },
    {
      "epoch": 0.1839974766060351,
      "grad_norm": 0.6656805276870728,
      "learning_rate": 4.0803806119230364e-05,
      "loss": 2.2889,
      "step": 52500
    },
    {
      "epoch": 0.18420775943644202,
      "grad_norm": 0.6012744307518005,
      "learning_rate": 4.079329197771002e-05,
      "loss": 2.3095,
      "step": 52560
    },
    {
      "epoch": 0.1844180422668489,
      "grad_norm": 0.6312623023986816,
      "learning_rate": 4.0782953071881686e-05,
      "loss": 2.3014,
      "step": 52620
    },
    {
      "epoch": 0.1846283250972558,
      "grad_norm": 0.6151764392852783,
      "learning_rate": 4.077243893036134e-05,
      "loss": 2.3117,
      "step": 52680
    },
    {
      "epoch": 0.1848386079276627,
      "grad_norm": 0.6274310350418091,
      "learning_rate": 4.0761924788841e-05,
      "loss": 2.3017,
      "step": 52740
    },
    {
      "epoch": 0.1850488907580696,
      "grad_norm": 0.6389456987380981,
      "learning_rate": 4.075141064732065e-05,
      "loss": 2.2992,
      "step": 52800
    },
    {
      "epoch": 0.1852591735884765,
      "grad_norm": 0.6331002116203308,
      "learning_rate": 4.07408965058003e-05,
      "loss": 2.3032,
      "step": 52860
    },
    {
      "epoch": 0.1854694564188834,
      "grad_norm": 0.5969839692115784,
      "learning_rate": 4.073038236427996e-05,
      "loss": 2.3023,
      "step": 52920
    },
    {
      "epoch": 0.1856797392492903,
      "grad_norm": 0.6058233976364136,
      "learning_rate": 4.071986822275962e-05,
      "loss": 2.3079,
      "step": 52980
    },
    {
      "epoch": 0.18589002207969718,
      "grad_norm": 0.600603461265564,
      "learning_rate": 4.070935408123927e-05,
      "loss": 2.3255,
      "step": 53040
    },
    {
      "epoch": 0.1861003049101041,
      "grad_norm": 0.5847122073173523,
      "learning_rate": 4.069883993971892e-05,
      "loss": 2.2967,
      "step": 53100
    },
    {
      "epoch": 0.18631058774051099,
      "grad_norm": 0.6589469313621521,
      "learning_rate": 4.068832579819858e-05,
      "loss": 2.3096,
      "step": 53160
    },
    {
      "epoch": 0.18652087057091787,
      "grad_norm": 0.5959056615829468,
      "learning_rate": 4.067781165667823e-05,
      "loss": 2.3039,
      "step": 53220
    },
    {
      "epoch": 0.1867311534013248,
      "grad_norm": 0.6054456830024719,
      "learning_rate": 4.0667297515157884e-05,
      "loss": 2.3166,
      "step": 53280
    },
    {
      "epoch": 0.18694143623173168,
      "grad_norm": 0.670484721660614,
      "learning_rate": 4.065678337363755e-05,
      "loss": 2.3216,
      "step": 53340
    },
    {
      "epoch": 0.18715171906213857,
      "grad_norm": 0.6493387222290039,
      "learning_rate": 4.06462692321172e-05,
      "loss": 2.3181,
      "step": 53400
    },
    {
      "epoch": 0.18736200189254548,
      "grad_norm": 0.5590341687202454,
      "learning_rate": 4.0635755090596853e-05,
      "loss": 2.3218,
      "step": 53460
    },
    {
      "epoch": 0.18757228472295237,
      "grad_norm": 0.5755470395088196,
      "learning_rate": 4.062524094907651e-05,
      "loss": 2.3121,
      "step": 53520
    },
    {
      "epoch": 0.18778256755335926,
      "grad_norm": 0.5840538144111633,
      "learning_rate": 4.0614726807556164e-05,
      "loss": 2.2996,
      "step": 53580
    },
    {
      "epoch": 0.18799285038376617,
      "grad_norm": 0.5699300765991211,
      "learning_rate": 4.0604212666035816e-05,
      "loss": 2.3193,
      "step": 53640
    },
    {
      "epoch": 0.18820313321417306,
      "grad_norm": 0.6303991675376892,
      "learning_rate": 4.0593698524515475e-05,
      "loss": 2.3158,
      "step": 53700
    },
    {
      "epoch": 0.18841341604457995,
      "grad_norm": 0.5751924514770508,
      "learning_rate": 4.058318438299513e-05,
      "loss": 2.2999,
      "step": 53760
    },
    {
      "epoch": 0.18862369887498687,
      "grad_norm": 0.5767543911933899,
      "learning_rate": 4.0572670241474785e-05,
      "loss": 2.3222,
      "step": 53820
    },
    {
      "epoch": 0.18883398170539376,
      "grad_norm": 0.6244462728500366,
      "learning_rate": 4.056215609995444e-05,
      "loss": 2.3067,
      "step": 53880
    },
    {
      "epoch": 0.18904426453580064,
      "grad_norm": 0.6316798329353333,
      "learning_rate": 4.05518171941261e-05,
      "loss": 2.3029,
      "step": 53940
    },
    {
      "epoch": 0.18925454736620756,
      "grad_norm": 0.6083167791366577,
      "learning_rate": 4.054130305260575e-05,
      "loss": 2.3053,
      "step": 54000
    },
    {
      "epoch": 0.18946483019661445,
      "grad_norm": 0.6451045870780945,
      "learning_rate": 4.053078891108541e-05,
      "loss": 2.3087,
      "step": 54060
    },
    {
      "epoch": 0.18967511302702134,
      "grad_norm": 0.558189868927002,
      "learning_rate": 4.052027476956506e-05,
      "loss": 2.309,
      "step": 54120
    },
    {
      "epoch": 0.18988539585742825,
      "grad_norm": 0.5685776472091675,
      "learning_rate": 4.050976062804472e-05,
      "loss": 2.2987,
      "step": 54180
    },
    {
      "epoch": 0.19009567868783514,
      "grad_norm": 0.5583335757255554,
      "learning_rate": 4.049924648652438e-05,
      "loss": 2.2946,
      "step": 54240
    },
    {
      "epoch": 0.19030596151824203,
      "grad_norm": 0.6071650981903076,
      "learning_rate": 4.048873234500403e-05,
      "loss": 2.3021,
      "step": 54300
    },
    {
      "epoch": 0.19051624434864894,
      "grad_norm": 0.5754538178443909,
      "learning_rate": 4.0478218203483684e-05,
      "loss": 2.3088,
      "step": 54360
    },
    {
      "epoch": 0.19072652717905583,
      "grad_norm": 0.5936737060546875,
      "learning_rate": 4.046770406196334e-05,
      "loss": 2.3125,
      "step": 54420
    },
    {
      "epoch": 0.19093681000946272,
      "grad_norm": 0.609310507774353,
      "learning_rate": 4.0457189920443e-05,
      "loss": 2.3399,
      "step": 54480
    },
    {
      "epoch": 0.19114709283986964,
      "grad_norm": 0.60650235414505,
      "learning_rate": 4.0446675778922654e-05,
      "loss": 2.3021,
      "step": 54540
    },
    {
      "epoch": 0.19135737567027652,
      "grad_norm": 0.5556688904762268,
      "learning_rate": 4.0436161637402306e-05,
      "loss": 2.3061,
      "step": 54600
    },
    {
      "epoch": 0.1915676585006834,
      "grad_norm": 0.668526291847229,
      "learning_rate": 4.0425647495881964e-05,
      "loss": 2.3005,
      "step": 54660
    },
    {
      "epoch": 0.19177794133109033,
      "grad_norm": 0.5854858160018921,
      "learning_rate": 4.0415133354361616e-05,
      "loss": 2.3084,
      "step": 54720
    },
    {
      "epoch": 0.19198822416149722,
      "grad_norm": 0.5923261642456055,
      "learning_rate": 4.0404619212841275e-05,
      "loss": 2.3094,
      "step": 54780
    },
    {
      "epoch": 0.1921985069919041,
      "grad_norm": 0.5517012476921082,
      "learning_rate": 4.0394105071320934e-05,
      "loss": 2.2968,
      "step": 54840
    },
    {
      "epoch": 0.19240878982231102,
      "grad_norm": 0.5993329286575317,
      "learning_rate": 4.0383590929800585e-05,
      "loss": 2.3174,
      "step": 54900
    },
    {
      "epoch": 0.1926190726527179,
      "grad_norm": 0.5918632745742798,
      "learning_rate": 4.037307678828024e-05,
      "loss": 2.3063,
      "step": 54960
    },
    {
      "epoch": 0.1928293554831248,
      "grad_norm": 0.5919117331504822,
      "learning_rate": 4.0362562646759896e-05,
      "loss": 2.2953,
      "step": 55020
    },
    {
      "epoch": 0.19303963831353171,
      "grad_norm": 0.6217793226242065,
      "learning_rate": 4.035204850523955e-05,
      "loss": 2.3082,
      "step": 55080
    },
    {
      "epoch": 0.1932499211439386,
      "grad_norm": 0.6075329780578613,
      "learning_rate": 4.03415343637192e-05,
      "loss": 2.3098,
      "step": 55140
    },
    {
      "epoch": 0.1934602039743455,
      "grad_norm": 0.5899872183799744,
      "learning_rate": 4.0331020222198865e-05,
      "loss": 2.312,
      "step": 55200
    },
    {
      "epoch": 0.19367048680475238,
      "grad_norm": 0.5910099744796753,
      "learning_rate": 4.032050608067852e-05,
      "loss": 2.3145,
      "step": 55260
    },
    {
      "epoch": 0.1938807696351593,
      "grad_norm": 0.6759595274925232,
      "learning_rate": 4.030999193915817e-05,
      "loss": 2.3064,
      "step": 55320
    },
    {
      "epoch": 0.19409105246556618,
      "grad_norm": 0.5908769369125366,
      "learning_rate": 4.029947779763783e-05,
      "loss": 2.3159,
      "step": 55380
    },
    {
      "epoch": 0.19430133529597307,
      "grad_norm": 0.5796085596084595,
      "learning_rate": 4.028896365611748e-05,
      "loss": 2.2936,
      "step": 55440
    },
    {
      "epoch": 0.19451161812638,
      "grad_norm": 0.623555064201355,
      "learning_rate": 4.027844951459713e-05,
      "loss": 2.3031,
      "step": 55500
    },
    {
      "epoch": 0.19472190095678688,
      "grad_norm": 0.610206663608551,
      "learning_rate": 4.026793537307679e-05,
      "loss": 2.3181,
      "step": 55560
    },
    {
      "epoch": 0.19493218378719376,
      "grad_norm": 0.5862197875976562,
      "learning_rate": 4.025742123155645e-05,
      "loss": 2.2862,
      "step": 55620
    },
    {
      "epoch": 0.19514246661760068,
      "grad_norm": 0.5683301091194153,
      "learning_rate": 4.02469070900361e-05,
      "loss": 2.3057,
      "step": 55680
    },
    {
      "epoch": 0.19535274944800757,
      "grad_norm": 0.5699648261070251,
      "learning_rate": 4.023639294851575e-05,
      "loss": 2.311,
      "step": 55740
    },
    {
      "epoch": 0.19556303227841446,
      "grad_norm": 0.5977410674095154,
      "learning_rate": 4.022587880699541e-05,
      "loss": 2.3056,
      "step": 55800
    },
    {
      "epoch": 0.19577331510882137,
      "grad_norm": 0.5807138681411743,
      "learning_rate": 4.021536466547506e-05,
      "loss": 2.3131,
      "step": 55860
    },
    {
      "epoch": 0.19598359793922826,
      "grad_norm": 0.5496934056282043,
      "learning_rate": 4.020485052395472e-05,
      "loss": 2.3164,
      "step": 55920
    },
    {
      "epoch": 0.19619388076963515,
      "grad_norm": 0.6061669588088989,
      "learning_rate": 4.019433638243438e-05,
      "loss": 2.3044,
      "step": 55980
    },
    {
      "epoch": 0.19640416360004206,
      "grad_norm": 0.5988995432853699,
      "learning_rate": 4.018382224091403e-05,
      "loss": 2.2983,
      "step": 56040
    },
    {
      "epoch": 0.19661444643044895,
      "grad_norm": 0.6225556135177612,
      "learning_rate": 4.0173308099393684e-05,
      "loss": 2.293,
      "step": 56100
    },
    {
      "epoch": 0.19682472926085584,
      "grad_norm": 0.5846222639083862,
      "learning_rate": 4.016279395787334e-05,
      "loss": 2.2929,
      "step": 56160
    },
    {
      "epoch": 0.19703501209126276,
      "grad_norm": 0.5997016429901123,
      "learning_rate": 4.0152279816352995e-05,
      "loss": 2.3001,
      "step": 56220
    },
    {
      "epoch": 0.19724529492166964,
      "grad_norm": 0.5757994055747986,
      "learning_rate": 4.0141765674832653e-05,
      "loss": 2.3129,
      "step": 56280
    },
    {
      "epoch": 0.19745557775207653,
      "grad_norm": 0.5928000211715698,
      "learning_rate": 4.0131251533312305e-05,
      "loss": 2.3106,
      "step": 56340
    },
    {
      "epoch": 0.19766586058248345,
      "grad_norm": 0.6106668710708618,
      "learning_rate": 4.0120737391791964e-05,
      "loss": 2.3057,
      "step": 56400
    },
    {
      "epoch": 0.19787614341289034,
      "grad_norm": 0.5892142057418823,
      "learning_rate": 4.0110223250271616e-05,
      "loss": 2.2884,
      "step": 56460
    },
    {
      "epoch": 0.19808642624329723,
      "grad_norm": 0.6026231646537781,
      "learning_rate": 4.009970910875127e-05,
      "loss": 2.31,
      "step": 56520
    },
    {
      "epoch": 0.19829670907370414,
      "grad_norm": 0.6028878092765808,
      "learning_rate": 4.0089194967230926e-05,
      "loss": 2.2865,
      "step": 56580
    },
    {
      "epoch": 0.19850699190411103,
      "grad_norm": 0.5761397480964661,
      "learning_rate": 4.0078680825710585e-05,
      "loss": 2.3016,
      "step": 56640
    },
    {
      "epoch": 0.19871727473451792,
      "grad_norm": 0.6288372278213501,
      "learning_rate": 4.006816668419024e-05,
      "loss": 2.3027,
      "step": 56700
    },
    {
      "epoch": 0.19892755756492483,
      "grad_norm": 0.5907147526741028,
      "learning_rate": 4.0057652542669896e-05,
      "loss": 2.3144,
      "step": 56760
    },
    {
      "epoch": 0.19913784039533172,
      "grad_norm": 0.5984941720962524,
      "learning_rate": 4.004713840114955e-05,
      "loss": 2.3066,
      "step": 56820
    },
    {
      "epoch": 0.1993481232257386,
      "grad_norm": 0.5911430716514587,
      "learning_rate": 4.00366242596292e-05,
      "loss": 2.2991,
      "step": 56880
    },
    {
      "epoch": 0.19955840605614553,
      "grad_norm": 0.6023363471031189,
      "learning_rate": 4.002611011810886e-05,
      "loss": 2.3113,
      "step": 56940
    },
    {
      "epoch": 0.19976868888655241,
      "grad_norm": 0.5403707027435303,
      "learning_rate": 4.001559597658852e-05,
      "loss": 2.313,
      "step": 57000
    },
    {
      "epoch": 0.1999789717169593,
      "grad_norm": 0.5953503847122192,
      "learning_rate": 4.000508183506817e-05,
      "loss": 2.3128,
      "step": 57060
    },
    {
      "epoch": 0.20018925454736622,
      "grad_norm": 0.5829181671142578,
      "learning_rate": 3.999456769354783e-05,
      "loss": 2.3079,
      "step": 57120
    },
    {
      "epoch": 0.2003995373777731,
      "grad_norm": 0.6059377789497375,
      "learning_rate": 3.998405355202748e-05,
      "loss": 2.3212,
      "step": 57180
    },
    {
      "epoch": 0.20060982020818,
      "grad_norm": 0.6386002898216248,
      "learning_rate": 3.997353941050713e-05,
      "loss": 2.318,
      "step": 57240
    },
    {
      "epoch": 0.2008201030385869,
      "grad_norm": 0.5913571119308472,
      "learning_rate": 3.996302526898679e-05,
      "loss": 2.3003,
      "step": 57300
    },
    {
      "epoch": 0.2010303858689938,
      "grad_norm": 0.571596086025238,
      "learning_rate": 3.995251112746645e-05,
      "loss": 2.3026,
      "step": 57360
    },
    {
      "epoch": 0.2012406686994007,
      "grad_norm": 0.6805840730667114,
      "learning_rate": 3.99419969859461e-05,
      "loss": 2.3095,
      "step": 57420
    },
    {
      "epoch": 0.2014509515298076,
      "grad_norm": 0.601517379283905,
      "learning_rate": 3.993148284442575e-05,
      "loss": 2.2977,
      "step": 57480
    },
    {
      "epoch": 0.2016612343602145,
      "grad_norm": 0.604802131652832,
      "learning_rate": 3.992096870290541e-05,
      "loss": 2.2887,
      "step": 57540
    },
    {
      "epoch": 0.20187151719062138,
      "grad_norm": 0.6532741189002991,
      "learning_rate": 3.991045456138506e-05,
      "loss": 2.2981,
      "step": 57600
    },
    {
      "epoch": 0.2020818000210283,
      "grad_norm": 0.5688276290893555,
      "learning_rate": 3.9899940419864715e-05,
      "loss": 2.3068,
      "step": 57660
    },
    {
      "epoch": 0.20229208285143518,
      "grad_norm": 0.6363274455070496,
      "learning_rate": 3.988942627834438e-05,
      "loss": 2.305,
      "step": 57720
    },
    {
      "epoch": 0.20250236568184207,
      "grad_norm": 0.5459544062614441,
      "learning_rate": 3.987891213682403e-05,
      "loss": 2.3052,
      "step": 57780
    },
    {
      "epoch": 0.202712648512249,
      "grad_norm": 0.6446651816368103,
      "learning_rate": 3.9868397995303684e-05,
      "loss": 2.3085,
      "step": 57840
    },
    {
      "epoch": 0.20292293134265588,
      "grad_norm": 0.6298580169677734,
      "learning_rate": 3.985788385378334e-05,
      "loss": 2.3129,
      "step": 57900
    },
    {
      "epoch": 0.20313321417306276,
      "grad_norm": 0.6669092774391174,
      "learning_rate": 3.9847369712262995e-05,
      "loss": 2.2961,
      "step": 57960
    },
    {
      "epoch": 0.20334349700346965,
      "grad_norm": 0.5807875990867615,
      "learning_rate": 3.983703080643466e-05,
      "loss": 2.3062,
      "step": 58020
    },
    {
      "epoch": 0.20355377983387657,
      "grad_norm": 0.6042791604995728,
      "learning_rate": 3.982651666491431e-05,
      "loss": 2.3205,
      "step": 58080
    },
    {
      "epoch": 0.20376406266428346,
      "grad_norm": 0.581607460975647,
      "learning_rate": 3.981600252339397e-05,
      "loss": 2.3042,
      "step": 58140
    },
    {
      "epoch": 0.20397434549469035,
      "grad_norm": 0.5742453336715698,
      "learning_rate": 3.980548838187362e-05,
      "loss": 2.3127,
      "step": 58200
    },
    {
      "epoch": 0.20418462832509726,
      "grad_norm": 0.5976558923721313,
      "learning_rate": 3.979497424035328e-05,
      "loss": 2.3015,
      "step": 58260
    },
    {
      "epoch": 0.20439491115550415,
      "grad_norm": 0.6173827052116394,
      "learning_rate": 3.978446009883293e-05,
      "loss": 2.2928,
      "step": 58320
    },
    {
      "epoch": 0.20460519398591104,
      "grad_norm": 0.5937358736991882,
      "learning_rate": 3.977394595731258e-05,
      "loss": 2.3008,
      "step": 58380
    },
    {
      "epoch": 0.20481547681631795,
      "grad_norm": 0.6002318859100342,
      "learning_rate": 3.976343181579224e-05,
      "loss": 2.3133,
      "step": 58440
    },
    {
      "epoch": 0.20502575964672484,
      "grad_norm": 0.6029914021492004,
      "learning_rate": 3.97529176742719e-05,
      "loss": 2.2874,
      "step": 58500
    },
    {
      "epoch": 0.20523604247713173,
      "grad_norm": 0.6260485649108887,
      "learning_rate": 3.974240353275155e-05,
      "loss": 2.2919,
      "step": 58560
    },
    {
      "epoch": 0.20544632530753865,
      "grad_norm": 0.5648069977760315,
      "learning_rate": 3.973188939123121e-05,
      "loss": 2.3182,
      "step": 58620
    },
    {
      "epoch": 0.20565660813794553,
      "grad_norm": 0.6081042289733887,
      "learning_rate": 3.972137524971086e-05,
      "loss": 2.2971,
      "step": 58680
    },
    {
      "epoch": 0.20586689096835242,
      "grad_norm": 0.5963698029518127,
      "learning_rate": 3.9710861108190515e-05,
      "loss": 2.311,
      "step": 58740
    },
    {
      "epoch": 0.20607717379875934,
      "grad_norm": 0.6485723257064819,
      "learning_rate": 3.9700346966670174e-05,
      "loss": 2.3095,
      "step": 58800
    },
    {
      "epoch": 0.20628745662916623,
      "grad_norm": 0.613882839679718,
      "learning_rate": 3.968983282514983e-05,
      "loss": 2.3224,
      "step": 58860
    },
    {
      "epoch": 0.20649773945957312,
      "grad_norm": 0.5611439347267151,
      "learning_rate": 3.9679318683629484e-05,
      "loss": 2.3028,
      "step": 58920
    },
    {
      "epoch": 0.20670802228998003,
      "grad_norm": 0.5237680673599243,
      "learning_rate": 3.9668804542109136e-05,
      "loss": 2.2831,
      "step": 58980
    },
    {
      "epoch": 0.20691830512038692,
      "grad_norm": 0.5781338214874268,
      "learning_rate": 3.9658290400588795e-05,
      "loss": 2.2963,
      "step": 59040
    },
    {
      "epoch": 0.2071285879507938,
      "grad_norm": 0.5803892016410828,
      "learning_rate": 3.964777625906845e-05,
      "loss": 2.3121,
      "step": 59100
    },
    {
      "epoch": 0.20733887078120072,
      "grad_norm": 0.5665882229804993,
      "learning_rate": 3.96372621175481e-05,
      "loss": 2.3089,
      "step": 59160
    },
    {
      "epoch": 0.2075491536116076,
      "grad_norm": 0.5831686854362488,
      "learning_rate": 3.9626747976027764e-05,
      "loss": 2.2884,
      "step": 59220
    },
    {
      "epoch": 0.2077594364420145,
      "grad_norm": 0.6503099203109741,
      "learning_rate": 3.9616233834507416e-05,
      "loss": 2.2925,
      "step": 59280
    },
    {
      "epoch": 0.20796971927242142,
      "grad_norm": 0.5764198303222656,
      "learning_rate": 3.960571969298707e-05,
      "loss": 2.2878,
      "step": 59340
    },
    {
      "epoch": 0.2081800021028283,
      "grad_norm": 0.5917394161224365,
      "learning_rate": 3.9595205551466726e-05,
      "loss": 2.3005,
      "step": 59400
    },
    {
      "epoch": 0.2083902849332352,
      "grad_norm": 0.6093656420707703,
      "learning_rate": 3.958469140994638e-05,
      "loss": 2.3097,
      "step": 59460
    },
    {
      "epoch": 0.2086005677636421,
      "grad_norm": 0.6051207184791565,
      "learning_rate": 3.957417726842603e-05,
      "loss": 2.3081,
      "step": 59520
    },
    {
      "epoch": 0.208810850594049,
      "grad_norm": 0.6025633811950684,
      "learning_rate": 3.9563663126905696e-05,
      "loss": 2.3096,
      "step": 59580
    },
    {
      "epoch": 0.20902113342445588,
      "grad_norm": 0.581791341304779,
      "learning_rate": 3.955314898538535e-05,
      "loss": 2.2989,
      "step": 59640
    },
    {
      "epoch": 0.2092314162548628,
      "grad_norm": 0.6260595917701721,
      "learning_rate": 3.9542634843865e-05,
      "loss": 2.3093,
      "step": 59700
    },
    {
      "epoch": 0.2094416990852697,
      "grad_norm": 0.5583322644233704,
      "learning_rate": 3.953212070234466e-05,
      "loss": 2.2863,
      "step": 59760
    },
    {
      "epoch": 0.20965198191567658,
      "grad_norm": 0.590339183807373,
      "learning_rate": 3.952160656082431e-05,
      "loss": 2.2983,
      "step": 59820
    },
    {
      "epoch": 0.2098622647460835,
      "grad_norm": 0.5544622540473938,
      "learning_rate": 3.951109241930396e-05,
      "loss": 2.2918,
      "step": 59880
    },
    {
      "epoch": 0.21007254757649038,
      "grad_norm": 0.5708463788032532,
      "learning_rate": 3.950057827778362e-05,
      "loss": 2.292,
      "step": 59940
    },
    {
      "epoch": 0.21028283040689727,
      "grad_norm": 0.617348313331604,
      "learning_rate": 3.949006413626328e-05,
      "loss": 2.308,
      "step": 60000
    },
    {
      "epoch": 0.21049311323730419,
      "grad_norm": 0.5817039608955383,
      "learning_rate": 3.947954999474293e-05,
      "loss": 2.3057,
      "step": 60060
    },
    {
      "epoch": 0.21070339606771107,
      "grad_norm": 0.5671586990356445,
      "learning_rate": 3.9469211088914595e-05,
      "loss": 2.3052,
      "step": 60120
    },
    {
      "epoch": 0.21091367889811796,
      "grad_norm": 0.5275577902793884,
      "learning_rate": 3.945869694739425e-05,
      "loss": 2.3053,
      "step": 60180
    },
    {
      "epoch": 0.21112396172852488,
      "grad_norm": 0.5590606927871704,
      "learning_rate": 3.94481828058739e-05,
      "loss": 2.2958,
      "step": 60240
    },
    {
      "epoch": 0.21133424455893177,
      "grad_norm": 0.5458212494850159,
      "learning_rate": 3.943766866435356e-05,
      "loss": 2.2908,
      "step": 60300
    },
    {
      "epoch": 0.21154452738933865,
      "grad_norm": 0.6047812700271606,
      "learning_rate": 3.9427154522833216e-05,
      "loss": 2.2973,
      "step": 60360
    },
    {
      "epoch": 0.21175481021974557,
      "grad_norm": 0.5924034714698792,
      "learning_rate": 3.941664038131287e-05,
      "loss": 2.2874,
      "step": 60420
    },
    {
      "epoch": 0.21196509305015246,
      "grad_norm": 0.6144199371337891,
      "learning_rate": 3.940612623979253e-05,
      "loss": 2.3011,
      "step": 60480
    },
    {
      "epoch": 0.21217537588055935,
      "grad_norm": 0.5948887467384338,
      "learning_rate": 3.939561209827218e-05,
      "loss": 2.2952,
      "step": 60540
    },
    {
      "epoch": 0.21238565871096626,
      "grad_norm": 0.5655609369277954,
      "learning_rate": 3.9385273192443836e-05,
      "loss": 2.2903,
      "step": 60600
    },
    {
      "epoch": 0.21259594154137315,
      "grad_norm": 0.5578904747962952,
      "learning_rate": 3.9374759050923494e-05,
      "loss": 2.2985,
      "step": 60660
    },
    {
      "epoch": 0.21280622437178004,
      "grad_norm": 0.5913036465644836,
      "learning_rate": 3.9364244909403146e-05,
      "loss": 2.2939,
      "step": 60720
    },
    {
      "epoch": 0.21301650720218693,
      "grad_norm": 0.5955153107643127,
      "learning_rate": 3.93537307678828e-05,
      "loss": 2.2823,
      "step": 60780
    },
    {
      "epoch": 0.21322679003259384,
      "grad_norm": 0.5599097013473511,
      "learning_rate": 3.9343216626362464e-05,
      "loss": 2.293,
      "step": 60840
    },
    {
      "epoch": 0.21343707286300073,
      "grad_norm": 0.5825154185295105,
      "learning_rate": 3.9332702484842116e-05,
      "loss": 2.3027,
      "step": 60900
    },
    {
      "epoch": 0.21364735569340762,
      "grad_norm": 0.6176333427429199,
      "learning_rate": 3.932218834332177e-05,
      "loss": 2.2903,
      "step": 60960
    },
    {
      "epoch": 0.21385763852381454,
      "grad_norm": 0.6115411520004272,
      "learning_rate": 3.9311674201801426e-05,
      "loss": 2.2938,
      "step": 61020
    },
    {
      "epoch": 0.21406792135422142,
      "grad_norm": 0.6089323163032532,
      "learning_rate": 3.930116006028108e-05,
      "loss": 2.2794,
      "step": 61080
    },
    {
      "epoch": 0.2142782041846283,
      "grad_norm": 0.5954441428184509,
      "learning_rate": 3.929064591876073e-05,
      "loss": 2.2942,
      "step": 61140
    },
    {
      "epoch": 0.21448848701503523,
      "grad_norm": 0.5842456221580505,
      "learning_rate": 3.9280131777240395e-05,
      "loss": 2.3082,
      "step": 61200
    },
    {
      "epoch": 0.21469876984544212,
      "grad_norm": 0.5549436211585999,
      "learning_rate": 3.926961763572005e-05,
      "loss": 2.2954,
      "step": 61260
    },
    {
      "epoch": 0.214909052675849,
      "grad_norm": 0.6503235101699829,
      "learning_rate": 3.92591034941997e-05,
      "loss": 2.2934,
      "step": 61320
    },
    {
      "epoch": 0.21511933550625592,
      "grad_norm": 0.5893281102180481,
      "learning_rate": 3.924858935267936e-05,
      "loss": 2.3044,
      "step": 61380
    },
    {
      "epoch": 0.2153296183366628,
      "grad_norm": 0.6054825186729431,
      "learning_rate": 3.923807521115901e-05,
      "loss": 2.3034,
      "step": 61440
    },
    {
      "epoch": 0.2155399011670697,
      "grad_norm": 0.7815125584602356,
      "learning_rate": 3.922756106963867e-05,
      "loss": 2.3021,
      "step": 61500
    },
    {
      "epoch": 0.2157501839974766,
      "grad_norm": 0.6160147190093994,
      "learning_rate": 3.921704692811832e-05,
      "loss": 2.2933,
      "step": 61560
    },
    {
      "epoch": 0.2159604668278835,
      "grad_norm": 0.6206346154212952,
      "learning_rate": 3.920653278659798e-05,
      "loss": 2.2961,
      "step": 61620
    },
    {
      "epoch": 0.2161707496582904,
      "grad_norm": 0.5747255682945251,
      "learning_rate": 3.919601864507763e-05,
      "loss": 2.2893,
      "step": 61680
    },
    {
      "epoch": 0.2163810324886973,
      "grad_norm": 0.5848297476768494,
      "learning_rate": 3.918550450355728e-05,
      "loss": 2.2864,
      "step": 61740
    },
    {
      "epoch": 0.2165913153191042,
      "grad_norm": 0.6018233895301819,
      "learning_rate": 3.917499036203694e-05,
      "loss": 2.2894,
      "step": 61800
    },
    {
      "epoch": 0.21680159814951108,
      "grad_norm": 0.5732107162475586,
      "learning_rate": 3.91644762205166e-05,
      "loss": 2.2931,
      "step": 61860
    },
    {
      "epoch": 0.217011880979918,
      "grad_norm": 0.6022079586982727,
      "learning_rate": 3.915396207899625e-05,
      "loss": 2.2948,
      "step": 61920
    },
    {
      "epoch": 0.2172221638103249,
      "grad_norm": 0.596991777420044,
      "learning_rate": 3.914344793747591e-05,
      "loss": 2.2946,
      "step": 61980
    },
    {
      "epoch": 0.21743244664073177,
      "grad_norm": 0.5973013043403625,
      "learning_rate": 3.913293379595556e-05,
      "loss": 2.2917,
      "step": 62040
    },
    {
      "epoch": 0.2176427294711387,
      "grad_norm": 0.6209733486175537,
      "learning_rate": 3.9122419654435214e-05,
      "loss": 2.2975,
      "step": 62100
    },
    {
      "epoch": 0.21785301230154558,
      "grad_norm": 0.5817129015922546,
      "learning_rate": 3.911190551291487e-05,
      "loss": 2.311,
      "step": 62160
    },
    {
      "epoch": 0.21806329513195247,
      "grad_norm": 0.6193563342094421,
      "learning_rate": 3.910139137139453e-05,
      "loss": 2.294,
      "step": 62220
    },
    {
      "epoch": 0.21827357796235938,
      "grad_norm": 0.5551738142967224,
      "learning_rate": 3.9090877229874184e-05,
      "loss": 2.2894,
      "step": 62280
    },
    {
      "epoch": 0.21848386079276627,
      "grad_norm": 0.6071308255195618,
      "learning_rate": 3.9080363088353835e-05,
      "loss": 2.299,
      "step": 62340
    },
    {
      "epoch": 0.21869414362317316,
      "grad_norm": 0.5778204798698425,
      "learning_rate": 3.9069848946833494e-05,
      "loss": 2.2934,
      "step": 62400
    },
    {
      "epoch": 0.21890442645358008,
      "grad_norm": 0.5507710576057434,
      "learning_rate": 3.9059334805313146e-05,
      "loss": 2.3035,
      "step": 62460
    },
    {
      "epoch": 0.21911470928398696,
      "grad_norm": 0.6067638993263245,
      "learning_rate": 3.90488206637928e-05,
      "loss": 2.3043,
      "step": 62520
    },
    {
      "epoch": 0.21932499211439385,
      "grad_norm": 0.5624009966850281,
      "learning_rate": 3.9038306522272463e-05,
      "loss": 2.28,
      "step": 62580
    },
    {
      "epoch": 0.21953527494480077,
      "grad_norm": 0.6132268905639648,
      "learning_rate": 3.9027792380752115e-05,
      "loss": 2.2813,
      "step": 62640
    },
    {
      "epoch": 0.21974555777520766,
      "grad_norm": 0.5769537091255188,
      "learning_rate": 3.901727823923177e-05,
      "loss": 2.2959,
      "step": 62700
    },
    {
      "epoch": 0.21995584060561454,
      "grad_norm": 0.5875435471534729,
      "learning_rate": 3.9006764097711426e-05,
      "loss": 2.3173,
      "step": 62760
    },
    {
      "epoch": 0.22016612343602146,
      "grad_norm": 0.6600687503814697,
      "learning_rate": 3.899624995619108e-05,
      "loss": 2.279,
      "step": 62820
    },
    {
      "epoch": 0.22037640626642835,
      "grad_norm": 0.5903270244598389,
      "learning_rate": 3.898573581467073e-05,
      "loss": 2.2965,
      "step": 62880
    },
    {
      "epoch": 0.22058668909683524,
      "grad_norm": 0.6876611113548279,
      "learning_rate": 3.897522167315039e-05,
      "loss": 2.3025,
      "step": 62940
    },
    {
      "epoch": 0.22079697192724215,
      "grad_norm": 0.5623339414596558,
      "learning_rate": 3.896470753163005e-05,
      "loss": 2.3011,
      "step": 63000
    },
    {
      "epoch": 0.22100725475764904,
      "grad_norm": 0.627585232257843,
      "learning_rate": 3.89541933901097e-05,
      "loss": 2.2827,
      "step": 63060
    },
    {
      "epoch": 0.22121753758805593,
      "grad_norm": 0.5718169808387756,
      "learning_rate": 3.894367924858936e-05,
      "loss": 2.281,
      "step": 63120
    },
    {
      "epoch": 0.22142782041846285,
      "grad_norm": 0.608727216720581,
      "learning_rate": 3.893316510706901e-05,
      "loss": 2.3046,
      "step": 63180
    },
    {
      "epoch": 0.22163810324886973,
      "grad_norm": 0.5940541625022888,
      "learning_rate": 3.892265096554866e-05,
      "loss": 2.2994,
      "step": 63240
    },
    {
      "epoch": 0.22184838607927662,
      "grad_norm": 0.6359009742736816,
      "learning_rate": 3.891213682402832e-05,
      "loss": 2.2937,
      "step": 63300
    },
    {
      "epoch": 0.22205866890968354,
      "grad_norm": 0.6548600792884827,
      "learning_rate": 3.890162268250798e-05,
      "loss": 2.2911,
      "step": 63360
    },
    {
      "epoch": 0.22226895174009043,
      "grad_norm": 0.6042421460151672,
      "learning_rate": 3.889110854098763e-05,
      "loss": 2.2918,
      "step": 63420
    },
    {
      "epoch": 0.22247923457049731,
      "grad_norm": 0.597143292427063,
      "learning_rate": 3.888059439946728e-05,
      "loss": 2.3005,
      "step": 63480
    },
    {
      "epoch": 0.22268951740090423,
      "grad_norm": 0.624727189540863,
      "learning_rate": 3.887008025794694e-05,
      "loss": 2.2891,
      "step": 63540
    },
    {
      "epoch": 0.22289980023131112,
      "grad_norm": 0.6133010387420654,
      "learning_rate": 3.885956611642659e-05,
      "loss": 2.2943,
      "step": 63600
    },
    {
      "epoch": 0.223110083061718,
      "grad_norm": 0.6171077489852905,
      "learning_rate": 3.884905197490625e-05,
      "loss": 2.2894,
      "step": 63660
    },
    {
      "epoch": 0.2233203658921249,
      "grad_norm": 0.6293420791625977,
      "learning_rate": 3.883853783338591e-05,
      "loss": 2.2883,
      "step": 63720
    },
    {
      "epoch": 0.2235306487225318,
      "grad_norm": 0.6117975115776062,
      "learning_rate": 3.882802369186556e-05,
      "loss": 2.2944,
      "step": 63780
    },
    {
      "epoch": 0.2237409315529387,
      "grad_norm": 0.5496307015419006,
      "learning_rate": 3.8817509550345214e-05,
      "loss": 2.2912,
      "step": 63840
    },
    {
      "epoch": 0.2239512143833456,
      "grad_norm": 0.7082515954971313,
      "learning_rate": 3.880699540882487e-05,
      "loss": 2.2942,
      "step": 63900
    },
    {
      "epoch": 0.2241614972137525,
      "grad_norm": 0.5779778957366943,
      "learning_rate": 3.8796481267304525e-05,
      "loss": 2.29,
      "step": 63960
    },
    {
      "epoch": 0.2243717800441594,
      "grad_norm": 0.5756336450576782,
      "learning_rate": 3.878596712578418e-05,
      "loss": 2.2815,
      "step": 64020
    },
    {
      "epoch": 0.22458206287456628,
      "grad_norm": 0.5937342643737793,
      "learning_rate": 3.8775452984263835e-05,
      "loss": 2.2985,
      "step": 64080
    },
    {
      "epoch": 0.2247923457049732,
      "grad_norm": 0.5614263415336609,
      "learning_rate": 3.8764938842743494e-05,
      "loss": 2.2983,
      "step": 64140
    },
    {
      "epoch": 0.22500262853538008,
      "grad_norm": 0.5805337429046631,
      "learning_rate": 3.8754424701223146e-05,
      "loss": 2.295,
      "step": 64200
    },
    {
      "epoch": 0.22521291136578697,
      "grad_norm": 0.6046919822692871,
      "learning_rate": 3.87439105597028e-05,
      "loss": 2.3119,
      "step": 64260
    },
    {
      "epoch": 0.2254231941961939,
      "grad_norm": 0.5776734948158264,
      "learning_rate": 3.8733396418182456e-05,
      "loss": 2.296,
      "step": 64320
    },
    {
      "epoch": 0.22563347702660078,
      "grad_norm": 0.5678257942199707,
      "learning_rate": 3.8722882276662115e-05,
      "loss": 2.2921,
      "step": 64380
    },
    {
      "epoch": 0.22584375985700766,
      "grad_norm": 0.5991799831390381,
      "learning_rate": 3.871236813514177e-05,
      "loss": 2.3095,
      "step": 64440
    },
    {
      "epoch": 0.22605404268741458,
      "grad_norm": 0.6048367619514465,
      "learning_rate": 3.8701853993621426e-05,
      "loss": 2.3009,
      "step": 64500
    },
    {
      "epoch": 0.22626432551782147,
      "grad_norm": 0.5879917740821838,
      "learning_rate": 3.869133985210108e-05,
      "loss": 2.2973,
      "step": 64560
    },
    {
      "epoch": 0.22647460834822836,
      "grad_norm": 0.601833701133728,
      "learning_rate": 3.868082571058073e-05,
      "loss": 2.2898,
      "step": 64620
    },
    {
      "epoch": 0.22668489117863527,
      "grad_norm": 0.6042653918266296,
      "learning_rate": 3.867031156906039e-05,
      "loss": 2.2839,
      "step": 64680
    },
    {
      "epoch": 0.22689517400904216,
      "grad_norm": 0.5772842764854431,
      "learning_rate": 3.8659972663232045e-05,
      "loss": 2.2911,
      "step": 64740
    },
    {
      "epoch": 0.22710545683944905,
      "grad_norm": 0.5636753439903259,
      "learning_rate": 3.8649458521711704e-05,
      "loss": 2.2955,
      "step": 64800
    },
    {
      "epoch": 0.22731573966985597,
      "grad_norm": 0.5744955539703369,
      "learning_rate": 3.863894438019136e-05,
      "loss": 2.2872,
      "step": 64860
    },
    {
      "epoch": 0.22752602250026285,
      "grad_norm": 0.575678288936615,
      "learning_rate": 3.8628430238671014e-05,
      "loss": 2.2935,
      "step": 64920
    },
    {
      "epoch": 0.22773630533066974,
      "grad_norm": 0.6447136402130127,
      "learning_rate": 3.8617916097150666e-05,
      "loss": 2.2983,
      "step": 64980
    },
    {
      "epoch": 0.22794658816107666,
      "grad_norm": 0.5995298624038696,
      "learning_rate": 3.8607401955630325e-05,
      "loss": 2.2867,
      "step": 65040
    },
    {
      "epoch": 0.22815687099148355,
      "grad_norm": 0.609114944934845,
      "learning_rate": 3.859688781410998e-05,
      "loss": 2.294,
      "step": 65100
    },
    {
      "epoch": 0.22836715382189043,
      "grad_norm": 0.615568220615387,
      "learning_rate": 3.8586373672589635e-05,
      "loss": 2.3041,
      "step": 65160
    },
    {
      "epoch": 0.22857743665229735,
      "grad_norm": 0.6183356046676636,
      "learning_rate": 3.8575859531069294e-05,
      "loss": 2.2961,
      "step": 65220
    },
    {
      "epoch": 0.22878771948270424,
      "grad_norm": 0.589805006980896,
      "learning_rate": 3.8565345389548946e-05,
      "loss": 2.2756,
      "step": 65280
    },
    {
      "epoch": 0.22899800231311113,
      "grad_norm": 0.6104287505149841,
      "learning_rate": 3.85548312480286e-05,
      "loss": 2.2903,
      "step": 65340
    },
    {
      "epoch": 0.22920828514351804,
      "grad_norm": 0.6121728420257568,
      "learning_rate": 3.854431710650826e-05,
      "loss": 2.288,
      "step": 65400
    },
    {
      "epoch": 0.22941856797392493,
      "grad_norm": 0.5765408277511597,
      "learning_rate": 3.853380296498791e-05,
      "loss": 2.2867,
      "step": 65460
    },
    {
      "epoch": 0.22962885080433182,
      "grad_norm": 0.6292064785957336,
      "learning_rate": 3.852328882346757e-05,
      "loss": 2.2856,
      "step": 65520
    },
    {
      "epoch": 0.22983913363473873,
      "grad_norm": 0.5729442834854126,
      "learning_rate": 3.851277468194722e-05,
      "loss": 2.2967,
      "step": 65580
    },
    {
      "epoch": 0.23004941646514562,
      "grad_norm": 0.5671252608299255,
      "learning_rate": 3.850226054042688e-05,
      "loss": 2.2795,
      "step": 65640
    },
    {
      "epoch": 0.2302596992955525,
      "grad_norm": 0.6085238456726074,
      "learning_rate": 3.849174639890653e-05,
      "loss": 2.2816,
      "step": 65700
    },
    {
      "epoch": 0.23046998212595943,
      "grad_norm": 0.5640025734901428,
      "learning_rate": 3.848123225738619e-05,
      "loss": 2.282,
      "step": 65760
    },
    {
      "epoch": 0.23068026495636632,
      "grad_norm": 0.5420010685920715,
      "learning_rate": 3.847071811586584e-05,
      "loss": 2.2999,
      "step": 65820
    },
    {
      "epoch": 0.2308905477867732,
      "grad_norm": 0.5791419148445129,
      "learning_rate": 3.84602039743455e-05,
      "loss": 2.288,
      "step": 65880
    },
    {
      "epoch": 0.23110083061718012,
      "grad_norm": 0.679742693901062,
      "learning_rate": 3.844968983282515e-05,
      "loss": 2.2958,
      "step": 65940
    },
    {
      "epoch": 0.231311113447587,
      "grad_norm": 0.6004040241241455,
      "learning_rate": 3.843917569130481e-05,
      "loss": 2.3013,
      "step": 66000
    },
    {
      "epoch": 0.2315213962779939,
      "grad_norm": 0.6382125616073608,
      "learning_rate": 3.842866154978446e-05,
      "loss": 2.2935,
      "step": 66060
    },
    {
      "epoch": 0.2317316791084008,
      "grad_norm": 0.5905836820602417,
      "learning_rate": 3.841814740826411e-05,
      "loss": 2.2975,
      "step": 66120
    },
    {
      "epoch": 0.2319419619388077,
      "grad_norm": 0.598331868648529,
      "learning_rate": 3.840763326674377e-05,
      "loss": 2.2866,
      "step": 66180
    },
    {
      "epoch": 0.2321522447692146,
      "grad_norm": 0.5576470494270325,
      "learning_rate": 3.839711912522343e-05,
      "loss": 2.2794,
      "step": 66240
    },
    {
      "epoch": 0.2323625275996215,
      "grad_norm": 0.6076599955558777,
      "learning_rate": 3.838660498370308e-05,
      "loss": 2.2855,
      "step": 66300
    },
    {
      "epoch": 0.2325728104300284,
      "grad_norm": 0.5614954829216003,
      "learning_rate": 3.837609084218274e-05,
      "loss": 2.2823,
      "step": 66360
    },
    {
      "epoch": 0.23278309326043528,
      "grad_norm": 0.5770753622055054,
      "learning_rate": 3.836557670066239e-05,
      "loss": 2.3076,
      "step": 66420
    },
    {
      "epoch": 0.23299337609084217,
      "grad_norm": 0.558489978313446,
      "learning_rate": 3.8355062559142045e-05,
      "loss": 2.2918,
      "step": 66480
    },
    {
      "epoch": 0.23320365892124909,
      "grad_norm": 0.6074410080909729,
      "learning_rate": 3.8344548417621704e-05,
      "loss": 2.2886,
      "step": 66540
    },
    {
      "epoch": 0.23341394175165597,
      "grad_norm": 0.5700733065605164,
      "learning_rate": 3.833403427610136e-05,
      "loss": 2.2817,
      "step": 66600
    },
    {
      "epoch": 0.23362422458206286,
      "grad_norm": 0.5876428484916687,
      "learning_rate": 3.8323520134581014e-05,
      "loss": 2.2925,
      "step": 66660
    },
    {
      "epoch": 0.23383450741246978,
      "grad_norm": 0.5912514925003052,
      "learning_rate": 3.8313005993060666e-05,
      "loss": 2.2865,
      "step": 66720
    },
    {
      "epoch": 0.23404479024287667,
      "grad_norm": 0.5970344543457031,
      "learning_rate": 3.830266708723233e-05,
      "loss": 2.3007,
      "step": 66780
    },
    {
      "epoch": 0.23425507307328355,
      "grad_norm": 0.615527331829071,
      "learning_rate": 3.829215294571198e-05,
      "loss": 2.2843,
      "step": 66840
    },
    {
      "epoch": 0.23446535590369047,
      "grad_norm": 0.6198700070381165,
      "learning_rate": 3.828163880419164e-05,
      "loss": 2.2807,
      "step": 66900
    },
    {
      "epoch": 0.23467563873409736,
      "grad_norm": 0.5815827250480652,
      "learning_rate": 3.827112466267129e-05,
      "loss": 2.2841,
      "step": 66960
    },
    {
      "epoch": 0.23488592156450425,
      "grad_norm": 0.5822238922119141,
      "learning_rate": 3.826061052115095e-05,
      "loss": 2.2937,
      "step": 67020
    },
    {
      "epoch": 0.23509620439491116,
      "grad_norm": 0.700834333896637,
      "learning_rate": 3.825009637963061e-05,
      "loss": 2.2888,
      "step": 67080
    },
    {
      "epoch": 0.23530648722531805,
      "grad_norm": 0.6379315853118896,
      "learning_rate": 3.823958223811026e-05,
      "loss": 2.2833,
      "step": 67140
    },
    {
      "epoch": 0.23551677005572494,
      "grad_norm": 0.566112220287323,
      "learning_rate": 3.8229068096589913e-05,
      "loss": 2.2859,
      "step": 67200
    },
    {
      "epoch": 0.23572705288613185,
      "grad_norm": 0.5224472284317017,
      "learning_rate": 3.821855395506957e-05,
      "loss": 2.2963,
      "step": 67260
    },
    {
      "epoch": 0.23593733571653874,
      "grad_norm": 0.6125534772872925,
      "learning_rate": 3.8208039813549224e-05,
      "loss": 2.2776,
      "step": 67320
    },
    {
      "epoch": 0.23614761854694563,
      "grad_norm": 0.5589282512664795,
      "learning_rate": 3.819752567202888e-05,
      "loss": 2.2687,
      "step": 67380
    },
    {
      "epoch": 0.23635790137735255,
      "grad_norm": 0.6050557494163513,
      "learning_rate": 3.8187011530508535e-05,
      "loss": 2.2958,
      "step": 67440
    },
    {
      "epoch": 0.23656818420775944,
      "grad_norm": 0.5693143010139465,
      "learning_rate": 3.817649738898819e-05,
      "loss": 2.281,
      "step": 67500
    },
    {
      "epoch": 0.23677846703816632,
      "grad_norm": 0.6809577941894531,
      "learning_rate": 3.8165983247467845e-05,
      "loss": 2.3013,
      "step": 67560
    },
    {
      "epoch": 0.23698874986857324,
      "grad_norm": 0.5879551768302917,
      "learning_rate": 3.81554691059475e-05,
      "loss": 2.2664,
      "step": 67620
    },
    {
      "epoch": 0.23719903269898013,
      "grad_norm": 0.5711112022399902,
      "learning_rate": 3.8144954964427156e-05,
      "loss": 2.2878,
      "step": 67680
    },
    {
      "epoch": 0.23740931552938702,
      "grad_norm": 0.5781254172325134,
      "learning_rate": 3.8134440822906814e-05,
      "loss": 2.2977,
      "step": 67740
    },
    {
      "epoch": 0.23761959835979393,
      "grad_norm": 0.6531150341033936,
      "learning_rate": 3.8123926681386466e-05,
      "loss": 2.2855,
      "step": 67800
    },
    {
      "epoch": 0.23782988119020082,
      "grad_norm": 0.5889065265655518,
      "learning_rate": 3.8113412539866125e-05,
      "loss": 2.3085,
      "step": 67860
    },
    {
      "epoch": 0.2380401640206077,
      "grad_norm": 0.5385099649429321,
      "learning_rate": 3.810289839834578e-05,
      "loss": 2.2844,
      "step": 67920
    },
    {
      "epoch": 0.23825044685101462,
      "grad_norm": 0.5548147559165955,
      "learning_rate": 3.809238425682543e-05,
      "loss": 2.2847,
      "step": 67980
    },
    {
      "epoch": 0.2384607296814215,
      "grad_norm": 0.6021646857261658,
      "learning_rate": 3.808187011530509e-05,
      "loss": 2.2768,
      "step": 68040
    },
    {
      "epoch": 0.2386710125118284,
      "grad_norm": 0.592646598815918,
      "learning_rate": 3.8071355973784746e-05,
      "loss": 2.2908,
      "step": 68100
    },
    {
      "epoch": 0.23888129534223532,
      "grad_norm": 0.6466455459594727,
      "learning_rate": 3.80608418322644e-05,
      "loss": 2.2919,
      "step": 68160
    },
    {
      "epoch": 0.2390915781726422,
      "grad_norm": 0.5685292482376099,
      "learning_rate": 3.805032769074405e-05,
      "loss": 2.2784,
      "step": 68220
    },
    {
      "epoch": 0.2393018610030491,
      "grad_norm": 0.564929187297821,
      "learning_rate": 3.803981354922371e-05,
      "loss": 2.2802,
      "step": 68280
    },
    {
      "epoch": 0.239512143833456,
      "grad_norm": 0.609919011592865,
      "learning_rate": 3.802929940770336e-05,
      "loss": 2.2781,
      "step": 68340
    },
    {
      "epoch": 0.2397224266638629,
      "grad_norm": 0.5926073789596558,
      "learning_rate": 3.801878526618302e-05,
      "loss": 2.2773,
      "step": 68400
    },
    {
      "epoch": 0.23993270949426979,
      "grad_norm": 0.5856402516365051,
      "learning_rate": 3.800827112466268e-05,
      "loss": 2.286,
      "step": 68460
    },
    {
      "epoch": 0.2401429923246767,
      "grad_norm": 0.5757060050964355,
      "learning_rate": 3.799775698314233e-05,
      "loss": 2.2801,
      "step": 68520
    },
    {
      "epoch": 0.2403532751550836,
      "grad_norm": 0.626157820224762,
      "learning_rate": 3.798724284162198e-05,
      "loss": 2.2915,
      "step": 68580
    },
    {
      "epoch": 0.24056355798549048,
      "grad_norm": 0.5402255654335022,
      "learning_rate": 3.797672870010164e-05,
      "loss": 2.2782,
      "step": 68640
    },
    {
      "epoch": 0.2407738408158974,
      "grad_norm": 0.5578147768974304,
      "learning_rate": 3.796621455858129e-05,
      "loss": 2.2887,
      "step": 68700
    },
    {
      "epoch": 0.24098412364630428,
      "grad_norm": 0.6032596826553345,
      "learning_rate": 3.7955700417060944e-05,
      "loss": 2.2862,
      "step": 68760
    },
    {
      "epoch": 0.24119440647671117,
      "grad_norm": 0.5816863775253296,
      "learning_rate": 3.794518627554061e-05,
      "loss": 2.269,
      "step": 68820
    },
    {
      "epoch": 0.2414046893071181,
      "grad_norm": 0.5900723338127136,
      "learning_rate": 3.793467213402026e-05,
      "loss": 2.2907,
      "step": 68880
    },
    {
      "epoch": 0.24161497213752497,
      "grad_norm": 0.5583692789077759,
      "learning_rate": 3.792433322819192e-05,
      "loss": 2.29,
      "step": 68940
    },
    {
      "epoch": 0.24182525496793186,
      "grad_norm": 0.5381126999855042,
      "learning_rate": 3.791381908667158e-05,
      "loss": 2.2981,
      "step": 69000
    },
    {
      "epoch": 0.24203553779833878,
      "grad_norm": 0.5909293293952942,
      "learning_rate": 3.790330494515123e-05,
      "loss": 2.2989,
      "step": 69060
    },
    {
      "epoch": 0.24224582062874567,
      "grad_norm": 0.6078698635101318,
      "learning_rate": 3.789279080363089e-05,
      "loss": 2.296,
      "step": 69120
    },
    {
      "epoch": 0.24245610345915256,
      "grad_norm": 0.5708657503128052,
      "learning_rate": 3.788227666211054e-05,
      "loss": 2.2714,
      "step": 69180
    },
    {
      "epoch": 0.24266638628955944,
      "grad_norm": 0.5961434245109558,
      "learning_rate": 3.78717625205902e-05,
      "loss": 2.2737,
      "step": 69240
    },
    {
      "epoch": 0.24287666911996636,
      "grad_norm": 0.5856456160545349,
      "learning_rate": 3.786124837906985e-05,
      "loss": 2.2953,
      "step": 69300
    },
    {
      "epoch": 0.24308695195037325,
      "grad_norm": 0.5971105098724365,
      "learning_rate": 3.785073423754951e-05,
      "loss": 2.2791,
      "step": 69360
    },
    {
      "epoch": 0.24329723478078014,
      "grad_norm": 0.5927890539169312,
      "learning_rate": 3.784022009602916e-05,
      "loss": 2.2845,
      "step": 69420
    },
    {
      "epoch": 0.24350751761118705,
      "grad_norm": 0.5798436403274536,
      "learning_rate": 3.782970595450881e-05,
      "loss": 2.2845,
      "step": 69480
    },
    {
      "epoch": 0.24371780044159394,
      "grad_norm": 0.574324905872345,
      "learning_rate": 3.781919181298847e-05,
      "loss": 2.2817,
      "step": 69540
    },
    {
      "epoch": 0.24392808327200083,
      "grad_norm": 0.5704073905944824,
      "learning_rate": 3.780867767146813e-05,
      "loss": 2.2779,
      "step": 69600
    },
    {
      "epoch": 0.24413836610240774,
      "grad_norm": 0.5989373922348022,
      "learning_rate": 3.779816352994778e-05,
      "loss": 2.299,
      "step": 69660
    },
    {
      "epoch": 0.24434864893281463,
      "grad_norm": 0.5521675944328308,
      "learning_rate": 3.778764938842744e-05,
      "loss": 2.2701,
      "step": 69720
    },
    {
      "epoch": 0.24455893176322152,
      "grad_norm": 0.6123749017715454,
      "learning_rate": 3.777713524690709e-05,
      "loss": 2.2782,
      "step": 69780
    },
    {
      "epoch": 0.24476921459362844,
      "grad_norm": 0.5558038353919983,
      "learning_rate": 3.7766621105386744e-05,
      "loss": 2.2697,
      "step": 69840
    },
    {
      "epoch": 0.24497949742403533,
      "grad_norm": 0.5577847957611084,
      "learning_rate": 3.77561069638664e-05,
      "loss": 2.2813,
      "step": 69900
    },
    {
      "epoch": 0.2451897802544422,
      "grad_norm": 0.5782567262649536,
      "learning_rate": 3.774559282234606e-05,
      "loss": 2.2702,
      "step": 69960
    },
    {
      "epoch": 0.24540006308484913,
      "grad_norm": 0.6026614904403687,
      "learning_rate": 3.7735078680825713e-05,
      "loss": 2.2734,
      "step": 70020
    },
    {
      "epoch": 0.24561034591525602,
      "grad_norm": 0.5850802659988403,
      "learning_rate": 3.7724564539305365e-05,
      "loss": 2.2959,
      "step": 70080
    },
    {
      "epoch": 0.2458206287456629,
      "grad_norm": 0.6082926988601685,
      "learning_rate": 3.7714050397785024e-05,
      "loss": 2.2909,
      "step": 70140
    },
    {
      "epoch": 0.24603091157606982,
      "grad_norm": 0.6162857413291931,
      "learning_rate": 3.7703536256264676e-05,
      "loss": 2.293,
      "step": 70200
    },
    {
      "epoch": 0.2462411944064767,
      "grad_norm": 0.5475439429283142,
      "learning_rate": 3.769302211474433e-05,
      "loss": 2.2745,
      "step": 70260
    },
    {
      "epoch": 0.2464514772368836,
      "grad_norm": 0.6155129671096802,
      "learning_rate": 3.768250797322399e-05,
      "loss": 2.2821,
      "step": 70320
    },
    {
      "epoch": 0.24666176006729051,
      "grad_norm": 0.5935275554656982,
      "learning_rate": 3.7671993831703645e-05,
      "loss": 2.2783,
      "step": 70380
    },
    {
      "epoch": 0.2468720428976974,
      "grad_norm": 0.5896009802818298,
      "learning_rate": 3.76614796901833e-05,
      "loss": 2.2785,
      "step": 70440
    },
    {
      "epoch": 0.2470823257281043,
      "grad_norm": 0.6749953627586365,
      "learning_rate": 3.7650965548662956e-05,
      "loss": 2.2916,
      "step": 70500
    },
    {
      "epoch": 0.2472926085585112,
      "grad_norm": 0.6071747541427612,
      "learning_rate": 3.764045140714261e-05,
      "loss": 2.2915,
      "step": 70560
    },
    {
      "epoch": 0.2475028913889181,
      "grad_norm": 0.5885292291641235,
      "learning_rate": 3.762993726562226e-05,
      "loss": 2.2668,
      "step": 70620
    },
    {
      "epoch": 0.24771317421932498,
      "grad_norm": 0.5685102343559265,
      "learning_rate": 3.761942312410192e-05,
      "loss": 2.2716,
      "step": 70680
    },
    {
      "epoch": 0.2479234570497319,
      "grad_norm": 0.6011393666267395,
      "learning_rate": 3.760890898258158e-05,
      "loss": 2.2737,
      "step": 70740
    },
    {
      "epoch": 0.2481337398801388,
      "grad_norm": 0.5806678533554077,
      "learning_rate": 3.759839484106123e-05,
      "loss": 2.2944,
      "step": 70800
    },
    {
      "epoch": 0.24834402271054568,
      "grad_norm": 0.6233952641487122,
      "learning_rate": 3.758788069954088e-05,
      "loss": 2.2818,
      "step": 70860
    },
    {
      "epoch": 0.2485543055409526,
      "grad_norm": 0.6003621816635132,
      "learning_rate": 3.757736655802054e-05,
      "loss": 2.2701,
      "step": 70920
    },
    {
      "epoch": 0.24876458837135948,
      "grad_norm": 0.5553110241889954,
      "learning_rate": 3.7567027652192196e-05,
      "loss": 2.2701,
      "step": 70980
    },
    {
      "epoch": 0.24897487120176637,
      "grad_norm": 0.5705835223197937,
      "learning_rate": 3.7556513510671855e-05,
      "loss": 2.2765,
      "step": 71040
    },
    {
      "epoch": 0.24918515403217328,
      "grad_norm": 0.5779468417167664,
      "learning_rate": 3.7545999369151514e-05,
      "loss": 2.2657,
      "step": 71100
    },
    {
      "epoch": 0.24939543686258017,
      "grad_norm": 0.6075929403305054,
      "learning_rate": 3.7535485227631166e-05,
      "loss": 2.2864,
      "step": 71160
    },
    {
      "epoch": 0.24960571969298706,
      "grad_norm": 0.5834728479385376,
      "learning_rate": 3.7524971086110824e-05,
      "loss": 2.2949,
      "step": 71220
    },
    {
      "epoch": 0.24981600252339398,
      "grad_norm": 0.5804185271263123,
      "learning_rate": 3.7514456944590476e-05,
      "loss": 2.2847,
      "step": 71280
    },
    {
      "epoch": 0.25002628535380084,
      "grad_norm": 0.595839262008667,
      "learning_rate": 3.750394280307013e-05,
      "loss": 2.292,
      "step": 71340
    },
    {
      "epoch": 0.25023656818420775,
      "grad_norm": 0.6016972661018372,
      "learning_rate": 3.749342866154979e-05,
      "loss": 2.2834,
      "step": 71400
    },
    {
      "epoch": 0.25044685101461467,
      "grad_norm": 0.6161638498306274,
      "learning_rate": 3.7482914520029445e-05,
      "loss": 2.2901,
      "step": 71460
    },
    {
      "epoch": 0.25065713384502153,
      "grad_norm": 0.5625993609428406,
      "learning_rate": 3.74724003785091e-05,
      "loss": 2.2812,
      "step": 71520
    },
    {
      "epoch": 0.25086741667542845,
      "grad_norm": 0.6411647796630859,
      "learning_rate": 3.746188623698875e-05,
      "loss": 2.2818,
      "step": 71580
    },
    {
      "epoch": 0.25107769950583536,
      "grad_norm": 0.5416374802589417,
      "learning_rate": 3.745137209546841e-05,
      "loss": 2.2785,
      "step": 71640
    },
    {
      "epoch": 0.2512879823362422,
      "grad_norm": 0.6235836744308472,
      "learning_rate": 3.744085795394806e-05,
      "loss": 2.2747,
      "step": 71700
    },
    {
      "epoch": 0.25149826516664914,
      "grad_norm": 0.60355144739151,
      "learning_rate": 3.743034381242771e-05,
      "loss": 2.2697,
      "step": 71760
    },
    {
      "epoch": 0.25170854799705605,
      "grad_norm": 0.5651815533638,
      "learning_rate": 3.741982967090738e-05,
      "loss": 2.2981,
      "step": 71820
    },
    {
      "epoch": 0.2519188308274629,
      "grad_norm": 0.552401602268219,
      "learning_rate": 3.740931552938703e-05,
      "loss": 2.2984,
      "step": 71880
    },
    {
      "epoch": 0.25212911365786983,
      "grad_norm": 0.5791208148002625,
      "learning_rate": 3.739880138786668e-05,
      "loss": 2.2848,
      "step": 71940
    },
    {
      "epoch": 0.25233939648827675,
      "grad_norm": 0.5935322046279907,
      "learning_rate": 3.738828724634634e-05,
      "loss": 2.3134,
      "step": 72000
    },
    {
      "epoch": 0.2525496793186836,
      "grad_norm": 0.5474521517753601,
      "learning_rate": 3.737777310482599e-05,
      "loss": 2.2772,
      "step": 72060
    },
    {
      "epoch": 0.2527599621490905,
      "grad_norm": 0.5810613036155701,
      "learning_rate": 3.736725896330564e-05,
      "loss": 2.288,
      "step": 72120
    },
    {
      "epoch": 0.25297024497949744,
      "grad_norm": 0.5361061096191406,
      "learning_rate": 3.735674482178531e-05,
      "loss": 2.2766,
      "step": 72180
    },
    {
      "epoch": 0.2531805278099043,
      "grad_norm": 0.5803009867668152,
      "learning_rate": 3.734640591595696e-05,
      "loss": 2.2782,
      "step": 72240
    },
    {
      "epoch": 0.2533908106403112,
      "grad_norm": 0.5926686525344849,
      "learning_rate": 3.733589177443662e-05,
      "loss": 2.2903,
      "step": 72300
    },
    {
      "epoch": 0.25360109347071813,
      "grad_norm": 0.5496491193771362,
      "learning_rate": 3.7325377632916276e-05,
      "loss": 2.2587,
      "step": 72360
    },
    {
      "epoch": 0.253811376301125,
      "grad_norm": 0.6137828826904297,
      "learning_rate": 3.731486349139593e-05,
      "loss": 2.2849,
      "step": 72420
    },
    {
      "epoch": 0.2540216591315319,
      "grad_norm": 0.5940037965774536,
      "learning_rate": 3.730434934987558e-05,
      "loss": 2.2847,
      "step": 72480
    },
    {
      "epoch": 0.2542319419619388,
      "grad_norm": 0.5658112168312073,
      "learning_rate": 3.729383520835524e-05,
      "loss": 2.2715,
      "step": 72540
    },
    {
      "epoch": 0.2544422247923457,
      "grad_norm": 0.6902830004692078,
      "learning_rate": 3.728332106683489e-05,
      "loss": 2.2756,
      "step": 72600
    },
    {
      "epoch": 0.2546525076227526,
      "grad_norm": 0.6103239059448242,
      "learning_rate": 3.727280692531455e-05,
      "loss": 2.288,
      "step": 72660
    },
    {
      "epoch": 0.2548627904531595,
      "grad_norm": 0.5635161995887756,
      "learning_rate": 3.726229278379421e-05,
      "loss": 2.2926,
      "step": 72720
    },
    {
      "epoch": 0.2550730732835664,
      "grad_norm": 0.6250701546669006,
      "learning_rate": 3.725177864227386e-05,
      "loss": 2.267,
      "step": 72780
    },
    {
      "epoch": 0.2552833561139733,
      "grad_norm": 0.6795318722724915,
      "learning_rate": 3.724126450075351e-05,
      "loss": 2.2794,
      "step": 72840
    },
    {
      "epoch": 0.2554936389443802,
      "grad_norm": 0.5607568025588989,
      "learning_rate": 3.723075035923317e-05,
      "loss": 2.2719,
      "step": 72900
    },
    {
      "epoch": 0.25570392177478707,
      "grad_norm": 0.5337287187576294,
      "learning_rate": 3.722023621771282e-05,
      "loss": 2.2686,
      "step": 72960
    },
    {
      "epoch": 0.255914204605194,
      "grad_norm": 0.6685793995857239,
      "learning_rate": 3.720972207619248e-05,
      "loss": 2.2956,
      "step": 73020
    },
    {
      "epoch": 0.2561244874356009,
      "grad_norm": 0.6170177459716797,
      "learning_rate": 3.719920793467214e-05,
      "loss": 2.2859,
      "step": 73080
    },
    {
      "epoch": 0.25633477026600776,
      "grad_norm": 0.5965674519538879,
      "learning_rate": 3.718869379315179e-05,
      "loss": 2.284,
      "step": 73140
    },
    {
      "epoch": 0.2565450530964147,
      "grad_norm": 0.5725794434547424,
      "learning_rate": 3.7178179651631444e-05,
      "loss": 2.2957,
      "step": 73200
    },
    {
      "epoch": 0.2567553359268216,
      "grad_norm": 0.6222959756851196,
      "learning_rate": 3.71676655101111e-05,
      "loss": 2.2802,
      "step": 73260
    },
    {
      "epoch": 0.25696561875722845,
      "grad_norm": 0.5478679537773132,
      "learning_rate": 3.715715136859076e-05,
      "loss": 2.2772,
      "step": 73320
    },
    {
      "epoch": 0.25717590158763537,
      "grad_norm": 0.5699739456176758,
      "learning_rate": 3.714663722707041e-05,
      "loss": 2.2794,
      "step": 73380
    },
    {
      "epoch": 0.2573861844180423,
      "grad_norm": 0.5548140406608582,
      "learning_rate": 3.7136123085550065e-05,
      "loss": 2.2788,
      "step": 73440
    },
    {
      "epoch": 0.25759646724844915,
      "grad_norm": 0.5604926943778992,
      "learning_rate": 3.7125608944029723e-05,
      "loss": 2.285,
      "step": 73500
    },
    {
      "epoch": 0.25780675007885606,
      "grad_norm": 0.6234787106513977,
      "learning_rate": 3.7115094802509375e-05,
      "loss": 2.2875,
      "step": 73560
    },
    {
      "epoch": 0.258017032909263,
      "grad_norm": 0.5991029143333435,
      "learning_rate": 3.710458066098903e-05,
      "loss": 2.2752,
      "step": 73620
    },
    {
      "epoch": 0.25822731573966984,
      "grad_norm": 0.628025472164154,
      "learning_rate": 3.709406651946869e-05,
      "loss": 2.2855,
      "step": 73680
    },
    {
      "epoch": 0.25843759857007675,
      "grad_norm": 0.5676313638687134,
      "learning_rate": 3.7083552377948345e-05,
      "loss": 2.284,
      "step": 73740
    },
    {
      "epoch": 0.25864788140048367,
      "grad_norm": 0.5893223285675049,
      "learning_rate": 3.7073038236427996e-05,
      "loss": 2.2686,
      "step": 73800
    },
    {
      "epoch": 0.25885816423089053,
      "grad_norm": 0.5592080354690552,
      "learning_rate": 3.7062524094907655e-05,
      "loss": 2.2821,
      "step": 73860
    },
    {
      "epoch": 0.25906844706129745,
      "grad_norm": 0.578217625617981,
      "learning_rate": 3.705200995338731e-05,
      "loss": 2.2657,
      "step": 73920
    },
    {
      "epoch": 0.25927872989170436,
      "grad_norm": 0.5730153918266296,
      "learning_rate": 3.704149581186696e-05,
      "loss": 2.2831,
      "step": 73980
    },
    {
      "epoch": 0.2594890127221112,
      "grad_norm": 0.5621018409729004,
      "learning_rate": 3.703098167034662e-05,
      "loss": 2.2782,
      "step": 74040
    },
    {
      "epoch": 0.25969929555251814,
      "grad_norm": 0.5932819843292236,
      "learning_rate": 3.7020467528826276e-05,
      "loss": 2.2905,
      "step": 74100
    },
    {
      "epoch": 0.25990957838292505,
      "grad_norm": 0.6077775359153748,
      "learning_rate": 3.700995338730593e-05,
      "loss": 2.2759,
      "step": 74160
    },
    {
      "epoch": 0.2601198612133319,
      "grad_norm": 0.6019622087478638,
      "learning_rate": 3.699943924578558e-05,
      "loss": 2.2719,
      "step": 74220
    },
    {
      "epoch": 0.26033014404373883,
      "grad_norm": 0.5670755505561829,
      "learning_rate": 3.698892510426524e-05,
      "loss": 2.2659,
      "step": 74280
    },
    {
      "epoch": 0.26054042687414575,
      "grad_norm": 0.6045207977294922,
      "learning_rate": 3.697841096274489e-05,
      "loss": 2.2675,
      "step": 74340
    },
    {
      "epoch": 0.2607507097045526,
      "grad_norm": 0.5820584893226624,
      "learning_rate": 3.696789682122455e-05,
      "loss": 2.2751,
      "step": 74400
    },
    {
      "epoch": 0.2609609925349595,
      "grad_norm": 0.6069955825805664,
      "learning_rate": 3.695738267970421e-05,
      "loss": 2.286,
      "step": 74460
    },
    {
      "epoch": 0.26117127536536644,
      "grad_norm": 0.5582893490791321,
      "learning_rate": 3.694686853818386e-05,
      "loss": 2.2748,
      "step": 74520
    },
    {
      "epoch": 0.2613815581957733,
      "grad_norm": 0.7658640146255493,
      "learning_rate": 3.693635439666351e-05,
      "loss": 2.2724,
      "step": 74580
    },
    {
      "epoch": 0.2615918410261802,
      "grad_norm": 0.5628880262374878,
      "learning_rate": 3.692584025514317e-05,
      "loss": 2.2802,
      "step": 74640
    },
    {
      "epoch": 0.26180212385658713,
      "grad_norm": 0.5615341067314148,
      "learning_rate": 3.691532611362282e-05,
      "loss": 2.2764,
      "step": 74700
    },
    {
      "epoch": 0.262012406686994,
      "grad_norm": 0.5396032333374023,
      "learning_rate": 3.690481197210248e-05,
      "loss": 2.2799,
      "step": 74760
    },
    {
      "epoch": 0.2622226895174009,
      "grad_norm": 0.6700000166893005,
      "learning_rate": 3.689429783058214e-05,
      "loss": 2.2818,
      "step": 74820
    },
    {
      "epoch": 0.2624329723478078,
      "grad_norm": 0.5692336559295654,
      "learning_rate": 3.688378368906179e-05,
      "loss": 2.2745,
      "step": 74880
    },
    {
      "epoch": 0.2626432551782147,
      "grad_norm": 0.5630716681480408,
      "learning_rate": 3.687326954754144e-05,
      "loss": 2.2765,
      "step": 74940
    },
    {
      "epoch": 0.2628535380086216,
      "grad_norm": 0.5756769180297852,
      "learning_rate": 3.68627554060211e-05,
      "loss": 2.2772,
      "step": 75000
    },
    {
      "epoch": 0.2630638208390285,
      "grad_norm": 0.5566816329956055,
      "learning_rate": 3.6852241264500754e-05,
      "loss": 2.2685,
      "step": 75060
    },
    {
      "epoch": 0.2632741036694354,
      "grad_norm": 0.6051713228225708,
      "learning_rate": 3.684172712298041e-05,
      "loss": 2.2603,
      "step": 75120
    },
    {
      "epoch": 0.2634843864998423,
      "grad_norm": 0.6107643842697144,
      "learning_rate": 3.6831212981460064e-05,
      "loss": 2.2592,
      "step": 75180
    },
    {
      "epoch": 0.2636946693302492,
      "grad_norm": 0.5542864203453064,
      "learning_rate": 3.682069883993972e-05,
      "loss": 2.273,
      "step": 75240
    },
    {
      "epoch": 0.26390495216065607,
      "grad_norm": 0.5695670247077942,
      "learning_rate": 3.6810184698419375e-05,
      "loss": 2.2682,
      "step": 75300
    },
    {
      "epoch": 0.264115234991063,
      "grad_norm": 0.5360891222953796,
      "learning_rate": 3.679967055689903e-05,
      "loss": 2.2794,
      "step": 75360
    },
    {
      "epoch": 0.2643255178214699,
      "grad_norm": 0.5911653637886047,
      "learning_rate": 3.6789156415378686e-05,
      "loss": 2.2841,
      "step": 75420
    },
    {
      "epoch": 0.26453580065187676,
      "grad_norm": 0.5800861716270447,
      "learning_rate": 3.6778642273858344e-05,
      "loss": 2.2822,
      "step": 75480
    },
    {
      "epoch": 0.2647460834822837,
      "grad_norm": 0.5655189156532288,
      "learning_rate": 3.6768128132337996e-05,
      "loss": 2.2963,
      "step": 75540
    },
    {
      "epoch": 0.2649563663126906,
      "grad_norm": 0.5841854810714722,
      "learning_rate": 3.6757613990817655e-05,
      "loss": 2.2742,
      "step": 75600
    },
    {
      "epoch": 0.26516664914309745,
      "grad_norm": 0.5612586736679077,
      "learning_rate": 3.674709984929731e-05,
      "loss": 2.2644,
      "step": 75660
    },
    {
      "epoch": 0.26537693197350437,
      "grad_norm": 0.5804318785667419,
      "learning_rate": 3.673658570777696e-05,
      "loss": 2.2799,
      "step": 75720
    },
    {
      "epoch": 0.2655872148039113,
      "grad_norm": 0.6640058159828186,
      "learning_rate": 3.672607156625662e-05,
      "loss": 2.2752,
      "step": 75780
    },
    {
      "epoch": 0.26579749763431815,
      "grad_norm": 0.5463032126426697,
      "learning_rate": 3.6715557424736276e-05,
      "loss": 2.2671,
      "step": 75840
    },
    {
      "epoch": 0.26600778046472506,
      "grad_norm": 0.5558159351348877,
      "learning_rate": 3.670504328321593e-05,
      "loss": 2.2825,
      "step": 75900
    },
    {
      "epoch": 0.266218063295132,
      "grad_norm": 0.6093794703483582,
      "learning_rate": 3.669470437738759e-05,
      "loss": 2.2823,
      "step": 75960
    },
    {
      "epoch": 0.26642834612553884,
      "grad_norm": 0.6040659546852112,
      "learning_rate": 3.6684190235867244e-05,
      "loss": 2.2763,
      "step": 76020
    },
    {
      "epoch": 0.26663862895594576,
      "grad_norm": 0.5455958247184753,
      "learning_rate": 3.6673676094346896e-05,
      "loss": 2.2805,
      "step": 76080
    },
    {
      "epoch": 0.26684891178635267,
      "grad_norm": 0.5874274969100952,
      "learning_rate": 3.6663161952826554e-05,
      "loss": 2.2779,
      "step": 76140
    },
    {
      "epoch": 0.26705919461675953,
      "grad_norm": 0.5681596398353577,
      "learning_rate": 3.6652647811306206e-05,
      "loss": 2.2686,
      "step": 76200
    },
    {
      "epoch": 0.26726947744716645,
      "grad_norm": 0.5700655579566956,
      "learning_rate": 3.6642133669785865e-05,
      "loss": 2.277,
      "step": 76260
    },
    {
      "epoch": 0.2674797602775733,
      "grad_norm": 0.5863887071609497,
      "learning_rate": 3.6631619528265523e-05,
      "loss": 2.2858,
      "step": 76320
    },
    {
      "epoch": 0.2676900431079802,
      "grad_norm": 0.5644739270210266,
      "learning_rate": 3.6621105386745175e-05,
      "loss": 2.2728,
      "step": 76380
    },
    {
      "epoch": 0.26790032593838714,
      "grad_norm": 0.6302114129066467,
      "learning_rate": 3.661059124522483e-05,
      "loss": 2.2855,
      "step": 76440
    },
    {
      "epoch": 0.268110608768794,
      "grad_norm": 0.5792679786682129,
      "learning_rate": 3.6600077103704486e-05,
      "loss": 2.2767,
      "step": 76500
    },
    {
      "epoch": 0.2683208915992009,
      "grad_norm": 0.585135817527771,
      "learning_rate": 3.658956296218414e-05,
      "loss": 2.2628,
      "step": 76560
    },
    {
      "epoch": 0.26853117442960783,
      "grad_norm": 0.6189137697219849,
      "learning_rate": 3.6579048820663796e-05,
      "loss": 2.2751,
      "step": 76620
    },
    {
      "epoch": 0.2687414572600147,
      "grad_norm": 0.5395883321762085,
      "learning_rate": 3.656853467914345e-05,
      "loss": 2.2943,
      "step": 76680
    },
    {
      "epoch": 0.2689517400904216,
      "grad_norm": 0.5798149704933167,
      "learning_rate": 3.655802053762311e-05,
      "loss": 2.2818,
      "step": 76740
    },
    {
      "epoch": 0.2691620229208285,
      "grad_norm": 0.5785930752754211,
      "learning_rate": 3.654750639610276e-05,
      "loss": 2.277,
      "step": 76800
    },
    {
      "epoch": 0.2693723057512354,
      "grad_norm": 0.5721600651741028,
      "learning_rate": 3.653699225458241e-05,
      "loss": 2.2691,
      "step": 76860
    },
    {
      "epoch": 0.2695825885816423,
      "grad_norm": 0.5851025581359863,
      "learning_rate": 3.652647811306207e-05,
      "loss": 2.2701,
      "step": 76920
    },
    {
      "epoch": 0.2697928714120492,
      "grad_norm": 0.6369785070419312,
      "learning_rate": 3.651596397154173e-05,
      "loss": 2.2694,
      "step": 76980
    },
    {
      "epoch": 0.2700031542424561,
      "grad_norm": 0.5723124146461487,
      "learning_rate": 3.650544983002138e-05,
      "loss": 2.277,
      "step": 77040
    },
    {
      "epoch": 0.270213437072863,
      "grad_norm": 0.537835955619812,
      "learning_rate": 3.649493568850104e-05,
      "loss": 2.2772,
      "step": 77100
    },
    {
      "epoch": 0.2704237199032699,
      "grad_norm": 0.574788510799408,
      "learning_rate": 3.648442154698069e-05,
      "loss": 2.2613,
      "step": 77160
    },
    {
      "epoch": 0.27063400273367677,
      "grad_norm": 0.5730419158935547,
      "learning_rate": 3.647390740546034e-05,
      "loss": 2.2719,
      "step": 77220
    },
    {
      "epoch": 0.2708442855640837,
      "grad_norm": 0.5613835453987122,
      "learning_rate": 3.646339326394e-05,
      "loss": 2.2883,
      "step": 77280
    },
    {
      "epoch": 0.2710545683944906,
      "grad_norm": 0.5720829963684082,
      "learning_rate": 3.645287912241966e-05,
      "loss": 2.2798,
      "step": 77340
    },
    {
      "epoch": 0.27126485122489746,
      "grad_norm": 0.5370036959648132,
      "learning_rate": 3.644236498089931e-05,
      "loss": 2.248,
      "step": 77400
    },
    {
      "epoch": 0.2714751340553044,
      "grad_norm": 0.6213563084602356,
      "learning_rate": 3.643185083937897e-05,
      "loss": 2.269,
      "step": 77460
    },
    {
      "epoch": 0.2716854168857113,
      "grad_norm": 0.5907732844352722,
      "learning_rate": 3.642133669785862e-05,
      "loss": 2.2824,
      "step": 77520
    },
    {
      "epoch": 0.27189569971611816,
      "grad_norm": 0.5852854251861572,
      "learning_rate": 3.6410822556338274e-05,
      "loss": 2.2667,
      "step": 77580
    },
    {
      "epoch": 0.27210598254652507,
      "grad_norm": 0.6155591607093811,
      "learning_rate": 3.640030841481793e-05,
      "loss": 2.2729,
      "step": 77640
    },
    {
      "epoch": 0.272316265376932,
      "grad_norm": 0.6252135038375854,
      "learning_rate": 3.638979427329759e-05,
      "loss": 2.2559,
      "step": 77700
    },
    {
      "epoch": 0.27252654820733885,
      "grad_norm": 0.6540042161941528,
      "learning_rate": 3.637928013177724e-05,
      "loss": 2.2865,
      "step": 77760
    },
    {
      "epoch": 0.27273683103774576,
      "grad_norm": 0.5478835701942444,
      "learning_rate": 3.6368765990256895e-05,
      "loss": 2.2831,
      "step": 77820
    },
    {
      "epoch": 0.2729471138681527,
      "grad_norm": 0.5724520683288574,
      "learning_rate": 3.6358251848736554e-05,
      "loss": 2.2792,
      "step": 77880
    },
    {
      "epoch": 0.27315739669855954,
      "grad_norm": 0.5985337495803833,
      "learning_rate": 3.6347737707216206e-05,
      "loss": 2.2833,
      "step": 77940
    },
    {
      "epoch": 0.27336767952896646,
      "grad_norm": 0.589272677898407,
      "learning_rate": 3.633722356569586e-05,
      "loss": 2.2795,
      "step": 78000
    },
    {
      "epoch": 0.2735779623593734,
      "grad_norm": 0.5734955072402954,
      "learning_rate": 3.632670942417552e-05,
      "loss": 2.2781,
      "step": 78060
    },
    {
      "epoch": 0.27378824518978023,
      "grad_norm": 0.5893633961677551,
      "learning_rate": 3.6316195282655175e-05,
      "loss": 2.2795,
      "step": 78120
    },
    {
      "epoch": 0.27399852802018715,
      "grad_norm": 0.6065170168876648,
      "learning_rate": 3.630568114113483e-05,
      "loss": 2.2781,
      "step": 78180
    },
    {
      "epoch": 0.27420881085059406,
      "grad_norm": 0.6370767951011658,
      "learning_rate": 3.6295166999614486e-05,
      "loss": 2.2815,
      "step": 78240
    },
    {
      "epoch": 0.2744190936810009,
      "grad_norm": 0.6043940782546997,
      "learning_rate": 3.628465285809414e-05,
      "loss": 2.2856,
      "step": 78300
    },
    {
      "epoch": 0.27462937651140784,
      "grad_norm": 0.5856173634529114,
      "learning_rate": 3.627413871657379e-05,
      "loss": 2.2742,
      "step": 78360
    },
    {
      "epoch": 0.27483965934181476,
      "grad_norm": 0.5545368194580078,
      "learning_rate": 3.626362457505345e-05,
      "loss": 2.2841,
      "step": 78420
    },
    {
      "epoch": 0.2750499421722216,
      "grad_norm": 0.5876592993736267,
      "learning_rate": 3.625328566922511e-05,
      "loss": 2.2842,
      "step": 78480
    },
    {
      "epoch": 0.27526022500262853,
      "grad_norm": 0.5744016766548157,
      "learning_rate": 3.6242771527704764e-05,
      "loss": 2.2797,
      "step": 78540
    },
    {
      "epoch": 0.27547050783303545,
      "grad_norm": 0.6173490285873413,
      "learning_rate": 3.623225738618442e-05,
      "loss": 2.2653,
      "step": 78600
    },
    {
      "epoch": 0.2756807906634423,
      "grad_norm": 0.5842175483703613,
      "learning_rate": 3.6221743244664074e-05,
      "loss": 2.2698,
      "step": 78660
    },
    {
      "epoch": 0.2758910734938492,
      "grad_norm": 0.6094750761985779,
      "learning_rate": 3.6211229103143726e-05,
      "loss": 2.281,
      "step": 78720
    },
    {
      "epoch": 0.27610135632425614,
      "grad_norm": 0.5415861010551453,
      "learning_rate": 3.6200714961623385e-05,
      "loss": 2.2665,
      "step": 78780
    },
    {
      "epoch": 0.276311639154663,
      "grad_norm": 0.58269864320755,
      "learning_rate": 3.6190200820103044e-05,
      "loss": 2.2607,
      "step": 78840
    },
    {
      "epoch": 0.2765219219850699,
      "grad_norm": 0.568885087966919,
      "learning_rate": 3.6179686678582696e-05,
      "loss": 2.2534,
      "step": 78900
    },
    {
      "epoch": 0.27673220481547683,
      "grad_norm": 0.5955713987350464,
      "learning_rate": 3.6169172537062354e-05,
      "loss": 2.2671,
      "step": 78960
    },
    {
      "epoch": 0.2769424876458837,
      "grad_norm": 0.5937778949737549,
      "learning_rate": 3.6158658395542006e-05,
      "loss": 2.2728,
      "step": 79020
    },
    {
      "epoch": 0.2771527704762906,
      "grad_norm": 0.5668724179267883,
      "learning_rate": 3.614814425402166e-05,
      "loss": 2.2773,
      "step": 79080
    },
    {
      "epoch": 0.2773630533066975,
      "grad_norm": 0.5798020362854004,
      "learning_rate": 3.613763011250132e-05,
      "loss": 2.2796,
      "step": 79140
    },
    {
      "epoch": 0.2775733361371044,
      "grad_norm": 0.5528576374053955,
      "learning_rate": 3.6127115970980975e-05,
      "loss": 2.2819,
      "step": 79200
    },
    {
      "epoch": 0.2777836189675113,
      "grad_norm": 0.8906698226928711,
      "learning_rate": 3.611660182946063e-05,
      "loss": 2.2767,
      "step": 79260
    },
    {
      "epoch": 0.2779939017979182,
      "grad_norm": 0.5634118318557739,
      "learning_rate": 3.610608768794028e-05,
      "loss": 2.291,
      "step": 79320
    },
    {
      "epoch": 0.2782041846283251,
      "grad_norm": 0.615436851978302,
      "learning_rate": 3.609557354641994e-05,
      "loss": 2.2881,
      "step": 79380
    },
    {
      "epoch": 0.278414467458732,
      "grad_norm": 0.6048182249069214,
      "learning_rate": 3.608505940489959e-05,
      "loss": 2.2766,
      "step": 79440
    },
    {
      "epoch": 0.2786247502891389,
      "grad_norm": 0.5954188704490662,
      "learning_rate": 3.607454526337924e-05,
      "loss": 2.2794,
      "step": 79500
    },
    {
      "epoch": 0.27883503311954577,
      "grad_norm": 0.6068901419639587,
      "learning_rate": 3.606403112185891e-05,
      "loss": 2.2886,
      "step": 79560
    },
    {
      "epoch": 0.2790453159499527,
      "grad_norm": 0.5282168388366699,
      "learning_rate": 3.605351698033856e-05,
      "loss": 2.281,
      "step": 79620
    },
    {
      "epoch": 0.2792555987803596,
      "grad_norm": 0.5673637986183167,
      "learning_rate": 3.604300283881821e-05,
      "loss": 2.2773,
      "step": 79680
    },
    {
      "epoch": 0.27946588161076646,
      "grad_norm": 0.5759354829788208,
      "learning_rate": 3.603248869729787e-05,
      "loss": 2.2663,
      "step": 79740
    },
    {
      "epoch": 0.2796761644411734,
      "grad_norm": 0.580322265625,
      "learning_rate": 3.602197455577752e-05,
      "loss": 2.2748,
      "step": 79800
    },
    {
      "epoch": 0.2798864472715803,
      "grad_norm": 0.5620171427726746,
      "learning_rate": 3.601146041425717e-05,
      "loss": 2.2677,
      "step": 79860
    },
    {
      "epoch": 0.28009673010198716,
      "grad_norm": 0.5752658247947693,
      "learning_rate": 3.600094627273683e-05,
      "loss": 2.277,
      "step": 79920
    },
    {
      "epoch": 0.2803070129323941,
      "grad_norm": 0.6061094403266907,
      "learning_rate": 3.599043213121649e-05,
      "loss": 2.2687,
      "step": 79980
    },
    {
      "epoch": 0.280517295762801,
      "grad_norm": 0.5554862022399902,
      "learning_rate": 3.597991798969614e-05,
      "loss": 2.2693,
      "step": 80040
    },
    {
      "epoch": 0.28072757859320785,
      "grad_norm": 0.5555817484855652,
      "learning_rate": 3.59694038481758e-05,
      "loss": 2.2764,
      "step": 80100
    },
    {
      "epoch": 0.28093786142361477,
      "grad_norm": 0.5457671880722046,
      "learning_rate": 3.595888970665545e-05,
      "loss": 2.2667,
      "step": 80160
    },
    {
      "epoch": 0.2811481442540217,
      "grad_norm": 0.5533132553100586,
      "learning_rate": 3.5948375565135105e-05,
      "loss": 2.2753,
      "step": 80220
    },
    {
      "epoch": 0.28135842708442854,
      "grad_norm": 0.6642926931381226,
      "learning_rate": 3.5937861423614764e-05,
      "loss": 2.2584,
      "step": 80280
    },
    {
      "epoch": 0.28156870991483546,
      "grad_norm": 0.5732621550559998,
      "learning_rate": 3.592734728209442e-05,
      "loss": 2.2761,
      "step": 80340
    },
    {
      "epoch": 0.2817789927452424,
      "grad_norm": 0.5810340046882629,
      "learning_rate": 3.5916833140574074e-05,
      "loss": 2.2622,
      "step": 80400
    },
    {
      "epoch": 0.28198927557564923,
      "grad_norm": 0.5583139061927795,
      "learning_rate": 3.5906318999053726e-05,
      "loss": 2.2803,
      "step": 80460
    },
    {
      "epoch": 0.28219955840605615,
      "grad_norm": 0.5932782888412476,
      "learning_rate": 3.5895804857533385e-05,
      "loss": 2.2638,
      "step": 80520
    },
    {
      "epoch": 0.28240984123646307,
      "grad_norm": 0.606884241104126,
      "learning_rate": 3.5885290716013037e-05,
      "loss": 2.2761,
      "step": 80580
    },
    {
      "epoch": 0.2826201240668699,
      "grad_norm": 0.6091174483299255,
      "learning_rate": 3.5874776574492695e-05,
      "loss": 2.2745,
      "step": 80640
    },
    {
      "epoch": 0.28283040689727684,
      "grad_norm": 0.568898618221283,
      "learning_rate": 3.5864262432972354e-05,
      "loss": 2.2727,
      "step": 80700
    },
    {
      "epoch": 0.28304068972768376,
      "grad_norm": 0.5533196926116943,
      "learning_rate": 3.5853748291452006e-05,
      "loss": 2.2476,
      "step": 80760
    },
    {
      "epoch": 0.2832509725580906,
      "grad_norm": 0.5467692017555237,
      "learning_rate": 3.584323414993166e-05,
      "loss": 2.2744,
      "step": 80820
    },
    {
      "epoch": 0.28346125538849754,
      "grad_norm": 0.5872514843940735,
      "learning_rate": 3.5832720008411316e-05,
      "loss": 2.2623,
      "step": 80880
    },
    {
      "epoch": 0.28367153821890445,
      "grad_norm": 0.6108036637306213,
      "learning_rate": 3.582220586689097e-05,
      "loss": 2.2543,
      "step": 80940
    },
    {
      "epoch": 0.2838818210493113,
      "grad_norm": 0.5775202512741089,
      "learning_rate": 3.581169172537063e-05,
      "loss": 2.2559,
      "step": 81000
    },
    {
      "epoch": 0.2840921038797182,
      "grad_norm": 0.5871195793151855,
      "learning_rate": 3.580117758385028e-05,
      "loss": 2.2459,
      "step": 81060
    },
    {
      "epoch": 0.28430238671012514,
      "grad_norm": 0.5602725744247437,
      "learning_rate": 3.579066344232994e-05,
      "loss": 2.2584,
      "step": 81120
    },
    {
      "epoch": 0.284512669540532,
      "grad_norm": 0.5468248128890991,
      "learning_rate": 3.578014930080959e-05,
      "loss": 2.2736,
      "step": 81180
    },
    {
      "epoch": 0.2847229523709389,
      "grad_norm": 0.5603057146072388,
      "learning_rate": 3.576963515928924e-05,
      "loss": 2.2622,
      "step": 81240
    },
    {
      "epoch": 0.28493323520134584,
      "grad_norm": 0.5584391355514526,
      "learning_rate": 3.57591210177689e-05,
      "loss": 2.2554,
      "step": 81300
    },
    {
      "epoch": 0.2851435180317527,
      "grad_norm": 0.6015111804008484,
      "learning_rate": 3.574860687624856e-05,
      "loss": 2.2682,
      "step": 81360
    },
    {
      "epoch": 0.2853538008621596,
      "grad_norm": 0.5918145775794983,
      "learning_rate": 3.573809273472821e-05,
      "loss": 2.272,
      "step": 81420
    },
    {
      "epoch": 0.28556408369256653,
      "grad_norm": 0.6344346404075623,
      "learning_rate": 3.572757859320787e-05,
      "loss": 2.2662,
      "step": 81480
    },
    {
      "epoch": 0.2857743665229734,
      "grad_norm": 0.5758938193321228,
      "learning_rate": 3.571706445168752e-05,
      "loss": 2.2616,
      "step": 81540
    },
    {
      "epoch": 0.2859846493533803,
      "grad_norm": 0.6150460839271545,
      "learning_rate": 3.570655031016717e-05,
      "loss": 2.2674,
      "step": 81600
    },
    {
      "epoch": 0.2861949321837872,
      "grad_norm": 0.6692140102386475,
      "learning_rate": 3.569603616864683e-05,
      "loss": 2.2647,
      "step": 81660
    },
    {
      "epoch": 0.2864052150141941,
      "grad_norm": 0.5545239448547363,
      "learning_rate": 3.568552202712649e-05,
      "loss": 2.2596,
      "step": 81720
    },
    {
      "epoch": 0.286615497844601,
      "grad_norm": 0.6395848393440247,
      "learning_rate": 3.567500788560614e-05,
      "loss": 2.2774,
      "step": 81780
    },
    {
      "epoch": 0.2868257806750079,
      "grad_norm": 0.5730613470077515,
      "learning_rate": 3.5664493744085794e-05,
      "loss": 2.2655,
      "step": 81840
    },
    {
      "epoch": 0.2870360635054148,
      "grad_norm": 0.5833149552345276,
      "learning_rate": 3.565397960256545e-05,
      "loss": 2.2705,
      "step": 81900
    },
    {
      "epoch": 0.2872463463358217,
      "grad_norm": 0.5819433331489563,
      "learning_rate": 3.5643465461045105e-05,
      "loss": 2.2533,
      "step": 81960
    },
    {
      "epoch": 0.28745662916622855,
      "grad_norm": 0.5378984808921814,
      "learning_rate": 3.563295131952476e-05,
      "loss": 2.2749,
      "step": 82020
    },
    {
      "epoch": 0.28766691199663547,
      "grad_norm": 0.643327534198761,
      "learning_rate": 3.562243717800442e-05,
      "loss": 2.2694,
      "step": 82080
    },
    {
      "epoch": 0.2878771948270424,
      "grad_norm": 0.5695509910583496,
      "learning_rate": 3.5611923036484074e-05,
      "loss": 2.2715,
      "step": 82140
    },
    {
      "epoch": 0.28808747765744924,
      "grad_norm": 0.5622131824493408,
      "learning_rate": 3.5601408894963726e-05,
      "loss": 2.2797,
      "step": 82200
    },
    {
      "epoch": 0.28829776048785616,
      "grad_norm": 0.58394855260849,
      "learning_rate": 3.5590894753443384e-05,
      "loss": 2.264,
      "step": 82260
    },
    {
      "epoch": 0.2885080433182631,
      "grad_norm": 0.5865631699562073,
      "learning_rate": 3.5580380611923036e-05,
      "loss": 2.2757,
      "step": 82320
    },
    {
      "epoch": 0.28871832614866993,
      "grad_norm": 0.5562753677368164,
      "learning_rate": 3.556986647040269e-05,
      "loss": 2.2632,
      "step": 82380
    },
    {
      "epoch": 0.28892860897907685,
      "grad_norm": 0.5615153312683105,
      "learning_rate": 3.5559352328882354e-05,
      "loss": 2.2613,
      "step": 82440
    },
    {
      "epoch": 0.28913889180948377,
      "grad_norm": 0.5816681385040283,
      "learning_rate": 3.5548838187362006e-05,
      "loss": 2.2741,
      "step": 82500
    },
    {
      "epoch": 0.2893491746398906,
      "grad_norm": 0.6041331887245178,
      "learning_rate": 3.553832404584166e-05,
      "loss": 2.2675,
      "step": 82560
    },
    {
      "epoch": 0.28955945747029754,
      "grad_norm": 0.5498755574226379,
      "learning_rate": 3.552798514001332e-05,
      "loss": 2.2622,
      "step": 82620
    },
    {
      "epoch": 0.28976974030070446,
      "grad_norm": 0.5720410943031311,
      "learning_rate": 3.551747099849297e-05,
      "loss": 2.2657,
      "step": 82680
    },
    {
      "epoch": 0.2899800231311113,
      "grad_norm": 0.5457740426063538,
      "learning_rate": 3.550713209266464e-05,
      "loss": 2.2648,
      "step": 82740
    },
    {
      "epoch": 0.29019030596151824,
      "grad_norm": 0.6136612892150879,
      "learning_rate": 3.549661795114429e-05,
      "loss": 2.2572,
      "step": 82800
    },
    {
      "epoch": 0.29040058879192515,
      "grad_norm": 0.5554794073104858,
      "learning_rate": 3.548610380962394e-05,
      "loss": 2.262,
      "step": 82860
    },
    {
      "epoch": 0.290610871622332,
      "grad_norm": 0.5288531184196472,
      "learning_rate": 3.5475589668103606e-05,
      "loss": 2.2772,
      "step": 82920
    },
    {
      "epoch": 0.29082115445273893,
      "grad_norm": 0.5184205174446106,
      "learning_rate": 3.546507552658326e-05,
      "loss": 2.2646,
      "step": 82980
    },
    {
      "epoch": 0.29103143728314584,
      "grad_norm": 0.5476679801940918,
      "learning_rate": 3.545456138506291e-05,
      "loss": 2.283,
      "step": 83040
    },
    {
      "epoch": 0.2912417201135527,
      "grad_norm": 0.602214515209198,
      "learning_rate": 3.544404724354257e-05,
      "loss": 2.253,
      "step": 83100
    },
    {
      "epoch": 0.2914520029439596,
      "grad_norm": 0.5712595582008362,
      "learning_rate": 3.543353310202222e-05,
      "loss": 2.2706,
      "step": 83160
    },
    {
      "epoch": 0.29166228577436654,
      "grad_norm": 0.6233317852020264,
      "learning_rate": 3.542301896050187e-05,
      "loss": 2.2605,
      "step": 83220
    },
    {
      "epoch": 0.2918725686047734,
      "grad_norm": 0.5277979969978333,
      "learning_rate": 3.541250481898153e-05,
      "loss": 2.2423,
      "step": 83280
    },
    {
      "epoch": 0.2920828514351803,
      "grad_norm": 0.564591109752655,
      "learning_rate": 3.540199067746119e-05,
      "loss": 2.2628,
      "step": 83340
    },
    {
      "epoch": 0.29229313426558723,
      "grad_norm": 0.5598502159118652,
      "learning_rate": 3.539147653594084e-05,
      "loss": 2.2807,
      "step": 83400
    },
    {
      "epoch": 0.2925034170959941,
      "grad_norm": 0.5494130849838257,
      "learning_rate": 3.5380962394420494e-05,
      "loss": 2.2698,
      "step": 83460
    },
    {
      "epoch": 0.292713699926401,
      "grad_norm": 0.5903727412223816,
      "learning_rate": 3.537044825290015e-05,
      "loss": 2.2692,
      "step": 83520
    },
    {
      "epoch": 0.2929239827568079,
      "grad_norm": 0.5893160104751587,
      "learning_rate": 3.5359934111379804e-05,
      "loss": 2.2593,
      "step": 83580
    },
    {
      "epoch": 0.2931342655872148,
      "grad_norm": 0.5820626616477966,
      "learning_rate": 3.534941996985946e-05,
      "loss": 2.2657,
      "step": 83640
    },
    {
      "epoch": 0.2933445484176217,
      "grad_norm": 0.6435462236404419,
      "learning_rate": 3.533890582833912e-05,
      "loss": 2.2636,
      "step": 83700
    },
    {
      "epoch": 0.2935548312480286,
      "grad_norm": 0.5747917890548706,
      "learning_rate": 3.5328391686818774e-05,
      "loss": 2.2617,
      "step": 83760
    },
    {
      "epoch": 0.2937651140784355,
      "grad_norm": 0.5375548005104065,
      "learning_rate": 3.5317877545298425e-05,
      "loss": 2.2723,
      "step": 83820
    },
    {
      "epoch": 0.2939753969088424,
      "grad_norm": 0.5366535186767578,
      "learning_rate": 3.5307363403778084e-05,
      "loss": 2.2755,
      "step": 83880
    },
    {
      "epoch": 0.2941856797392493,
      "grad_norm": 0.641517698764801,
      "learning_rate": 3.5296849262257736e-05,
      "loss": 2.271,
      "step": 83940
    },
    {
      "epoch": 0.29439596256965617,
      "grad_norm": 0.5670305490493774,
      "learning_rate": 3.5286335120737395e-05,
      "loss": 2.2707,
      "step": 84000
    },
    {
      "epoch": 0.2946062454000631,
      "grad_norm": 0.5636521577835083,
      "learning_rate": 3.527582097921705e-05,
      "loss": 2.2622,
      "step": 84060
    },
    {
      "epoch": 0.29481652823047,
      "grad_norm": 0.5666058659553528,
      "learning_rate": 3.5265306837696705e-05,
      "loss": 2.256,
      "step": 84120
    },
    {
      "epoch": 0.29502681106087686,
      "grad_norm": 0.621285617351532,
      "learning_rate": 3.525479269617636e-05,
      "loss": 2.2726,
      "step": 84180
    },
    {
      "epoch": 0.2952370938912838,
      "grad_norm": 0.5541791319847107,
      "learning_rate": 3.5244278554656016e-05,
      "loss": 2.2598,
      "step": 84240
    },
    {
      "epoch": 0.2954473767216907,
      "grad_norm": 0.5623100996017456,
      "learning_rate": 3.523376441313567e-05,
      "loss": 2.2616,
      "step": 84300
    },
    {
      "epoch": 0.29565765955209755,
      "grad_norm": 0.617722749710083,
      "learning_rate": 3.5223250271615326e-05,
      "loss": 2.2695,
      "step": 84360
    },
    {
      "epoch": 0.29586794238250447,
      "grad_norm": 0.5718526840209961,
      "learning_rate": 3.521273613009498e-05,
      "loss": 2.2756,
      "step": 84420
    },
    {
      "epoch": 0.2960782252129114,
      "grad_norm": 0.5954556465148926,
      "learning_rate": 3.520222198857464e-05,
      "loss": 2.2605,
      "step": 84480
    },
    {
      "epoch": 0.29628850804331824,
      "grad_norm": 0.5449030995368958,
      "learning_rate": 3.519170784705429e-05,
      "loss": 2.2795,
      "step": 84540
    },
    {
      "epoch": 0.29649879087372516,
      "grad_norm": 0.6317994594573975,
      "learning_rate": 3.518119370553394e-05,
      "loss": 2.2734,
      "step": 84600
    },
    {
      "epoch": 0.2967090737041321,
      "grad_norm": 0.5306956768035889,
      "learning_rate": 3.51706795640136e-05,
      "loss": 2.2734,
      "step": 84660
    },
    {
      "epoch": 0.29691935653453894,
      "grad_norm": 0.5752288103103638,
      "learning_rate": 3.516016542249326e-05,
      "loss": 2.2721,
      "step": 84720
    },
    {
      "epoch": 0.29712963936494585,
      "grad_norm": 0.5798787474632263,
      "learning_rate": 3.514965128097291e-05,
      "loss": 2.263,
      "step": 84780
    },
    {
      "epoch": 0.29733992219535277,
      "grad_norm": 0.6193447113037109,
      "learning_rate": 3.513913713945257e-05,
      "loss": 2.2682,
      "step": 84840
    },
    {
      "epoch": 0.29755020502575963,
      "grad_norm": 0.606647789478302,
      "learning_rate": 3.512862299793222e-05,
      "loss": 2.2723,
      "step": 84900
    },
    {
      "epoch": 0.29776048785616654,
      "grad_norm": 0.5479643940925598,
      "learning_rate": 3.5118284092103884e-05,
      "loss": 2.2642,
      "step": 84960
    },
    {
      "epoch": 0.29797077068657346,
      "grad_norm": 0.5397103428840637,
      "learning_rate": 3.5107769950583536e-05,
      "loss": 2.2639,
      "step": 85020
    },
    {
      "epoch": 0.2981810535169803,
      "grad_norm": 0.5763701796531677,
      "learning_rate": 3.509725580906319e-05,
      "loss": 2.282,
      "step": 85080
    },
    {
      "epoch": 0.29839133634738724,
      "grad_norm": 0.5760320425033569,
      "learning_rate": 3.508674166754285e-05,
      "loss": 2.2701,
      "step": 85140
    },
    {
      "epoch": 0.29860161917779415,
      "grad_norm": 0.5381782054901123,
      "learning_rate": 3.5076227526022505e-05,
      "loss": 2.2619,
      "step": 85200
    },
    {
      "epoch": 0.298811902008201,
      "grad_norm": 0.5738365054130554,
      "learning_rate": 3.506571338450216e-05,
      "loss": 2.2596,
      "step": 85260
    },
    {
      "epoch": 0.29902218483860793,
      "grad_norm": 0.5890007019042969,
      "learning_rate": 3.505519924298181e-05,
      "loss": 2.2801,
      "step": 85320
    },
    {
      "epoch": 0.29923246766901485,
      "grad_norm": 0.5713682174682617,
      "learning_rate": 3.504468510146147e-05,
      "loss": 2.2637,
      "step": 85380
    },
    {
      "epoch": 0.2994427504994217,
      "grad_norm": 0.5621296167373657,
      "learning_rate": 3.503417095994112e-05,
      "loss": 2.2639,
      "step": 85440
    },
    {
      "epoch": 0.2996530333298286,
      "grad_norm": 0.5415980815887451,
      "learning_rate": 3.502365681842078e-05,
      "loss": 2.2729,
      "step": 85500
    },
    {
      "epoch": 0.29986331616023554,
      "grad_norm": 0.6131845116615295,
      "learning_rate": 3.501314267690044e-05,
      "loss": 2.2626,
      "step": 85560
    },
    {
      "epoch": 0.3000735989906424,
      "grad_norm": 0.5747252702713013,
      "learning_rate": 3.500262853538009e-05,
      "loss": 2.2589,
      "step": 85620
    },
    {
      "epoch": 0.3002838818210493,
      "grad_norm": 0.5767868161201477,
      "learning_rate": 3.499211439385974e-05,
      "loss": 2.2655,
      "step": 85680
    },
    {
      "epoch": 0.30049416465145623,
      "grad_norm": 0.563033938407898,
      "learning_rate": 3.49816002523394e-05,
      "loss": 2.2744,
      "step": 85740
    },
    {
      "epoch": 0.3007044474818631,
      "grad_norm": 0.503409743309021,
      "learning_rate": 3.497108611081905e-05,
      "loss": 2.2667,
      "step": 85800
    },
    {
      "epoch": 0.30091473031227,
      "grad_norm": 0.5511951446533203,
      "learning_rate": 3.496057196929871e-05,
      "loss": 2.2694,
      "step": 85860
    },
    {
      "epoch": 0.3011250131426769,
      "grad_norm": 0.5811784267425537,
      "learning_rate": 3.495005782777836e-05,
      "loss": 2.2629,
      "step": 85920
    },
    {
      "epoch": 0.3013352959730838,
      "grad_norm": 0.5567927956581116,
      "learning_rate": 3.493954368625802e-05,
      "loss": 2.2606,
      "step": 85980
    },
    {
      "epoch": 0.3015455788034907,
      "grad_norm": 0.567691445350647,
      "learning_rate": 3.492902954473767e-05,
      "loss": 2.263,
      "step": 86040
    },
    {
      "epoch": 0.3017558616338976,
      "grad_norm": 0.5785837769508362,
      "learning_rate": 3.4918515403217325e-05,
      "loss": 2.2551,
      "step": 86100
    },
    {
      "epoch": 0.3019661444643045,
      "grad_norm": 0.6118089556694031,
      "learning_rate": 3.490800126169698e-05,
      "loss": 2.2685,
      "step": 86160
    },
    {
      "epoch": 0.3021764272947114,
      "grad_norm": 0.5302484631538391,
      "learning_rate": 3.489748712017664e-05,
      "loss": 2.2555,
      "step": 86220
    },
    {
      "epoch": 0.3023867101251183,
      "grad_norm": 0.6238006949424744,
      "learning_rate": 3.4886972978656294e-05,
      "loss": 2.2602,
      "step": 86280
    },
    {
      "epoch": 0.30259699295552517,
      "grad_norm": 0.5831438302993774,
      "learning_rate": 3.487645883713595e-05,
      "loss": 2.2522,
      "step": 86340
    },
    {
      "epoch": 0.3028072757859321,
      "grad_norm": 0.5737531781196594,
      "learning_rate": 3.4865944695615604e-05,
      "loss": 2.2572,
      "step": 86400
    },
    {
      "epoch": 0.303017558616339,
      "grad_norm": 0.6034284234046936,
      "learning_rate": 3.4855430554095256e-05,
      "loss": 2.2708,
      "step": 86460
    },
    {
      "epoch": 0.30322784144674586,
      "grad_norm": 0.6044401526451111,
      "learning_rate": 3.4844916412574915e-05,
      "loss": 2.2659,
      "step": 86520
    },
    {
      "epoch": 0.3034381242771528,
      "grad_norm": 0.5900658965110779,
      "learning_rate": 3.4834402271054574e-05,
      "loss": 2.2673,
      "step": 86580
    },
    {
      "epoch": 0.3036484071075597,
      "grad_norm": 0.5390185117721558,
      "learning_rate": 3.4823888129534225e-05,
      "loss": 2.2505,
      "step": 86640
    },
    {
      "epoch": 0.30385868993796655,
      "grad_norm": 0.5748520493507385,
      "learning_rate": 3.4813373988013884e-05,
      "loss": 2.2634,
      "step": 86700
    },
    {
      "epoch": 0.30406897276837347,
      "grad_norm": 0.5243687033653259,
      "learning_rate": 3.4802859846493536e-05,
      "loss": 2.2798,
      "step": 86760
    },
    {
      "epoch": 0.3042792555987804,
      "grad_norm": 0.5927646160125732,
      "learning_rate": 3.479234570497319e-05,
      "loss": 2.2792,
      "step": 86820
    },
    {
      "epoch": 0.30448953842918725,
      "grad_norm": 0.5824649333953857,
      "learning_rate": 3.4781831563452847e-05,
      "loss": 2.2633,
      "step": 86880
    },
    {
      "epoch": 0.30469982125959416,
      "grad_norm": 0.5818813443183899,
      "learning_rate": 3.4771317421932505e-05,
      "loss": 2.271,
      "step": 86940
    },
    {
      "epoch": 0.3049101040900011,
      "grad_norm": 0.5972533822059631,
      "learning_rate": 3.476080328041216e-05,
      "loss": 2.282,
      "step": 87000
    },
    {
      "epoch": 0.30512038692040794,
      "grad_norm": 0.5627744793891907,
      "learning_rate": 3.475028913889181e-05,
      "loss": 2.2621,
      "step": 87060
    },
    {
      "epoch": 0.30533066975081485,
      "grad_norm": 0.5838892459869385,
      "learning_rate": 3.473977499737147e-05,
      "loss": 2.2751,
      "step": 87120
    },
    {
      "epoch": 0.30554095258122177,
      "grad_norm": 0.5534815788269043,
      "learning_rate": 3.472926085585112e-05,
      "loss": 2.2741,
      "step": 87180
    },
    {
      "epoch": 0.30575123541162863,
      "grad_norm": 0.5433311462402344,
      "learning_rate": 3.471874671433077e-05,
      "loss": 2.2716,
      "step": 87240
    },
    {
      "epoch": 0.30596151824203555,
      "grad_norm": 0.5968735218048096,
      "learning_rate": 3.470823257281044e-05,
      "loss": 2.2503,
      "step": 87300
    },
    {
      "epoch": 0.30617180107244246,
      "grad_norm": 0.5322306752204895,
      "learning_rate": 3.469771843129009e-05,
      "loss": 2.2795,
      "step": 87360
    },
    {
      "epoch": 0.3063820839028493,
      "grad_norm": 0.6331864595413208,
      "learning_rate": 3.468720428976974e-05,
      "loss": 2.2611,
      "step": 87420
    },
    {
      "epoch": 0.30659236673325624,
      "grad_norm": 0.5504761338233948,
      "learning_rate": 3.46766901482494e-05,
      "loss": 2.2672,
      "step": 87480
    },
    {
      "epoch": 0.30680264956366315,
      "grad_norm": 0.6416230201721191,
      "learning_rate": 3.466617600672905e-05,
      "loss": 2.2727,
      "step": 87540
    },
    {
      "epoch": 0.30701293239407,
      "grad_norm": 0.639982283115387,
      "learning_rate": 3.46556618652087e-05,
      "loss": 2.2618,
      "step": 87600
    },
    {
      "epoch": 0.30722321522447693,
      "grad_norm": 0.6075161099433899,
      "learning_rate": 3.464514772368836e-05,
      "loss": 2.251,
      "step": 87660
    },
    {
      "epoch": 0.3074334980548838,
      "grad_norm": 0.5911880135536194,
      "learning_rate": 3.463463358216802e-05,
      "loss": 2.2673,
      "step": 87720
    },
    {
      "epoch": 0.3076437808852907,
      "grad_norm": 0.5482556223869324,
      "learning_rate": 3.462411944064767e-05,
      "loss": 2.2676,
      "step": 87780
    },
    {
      "epoch": 0.3078540637156976,
      "grad_norm": 0.5883685946464539,
      "learning_rate": 3.4613605299127324e-05,
      "loss": 2.2726,
      "step": 87840
    },
    {
      "epoch": 0.3080643465461045,
      "grad_norm": 0.5492207407951355,
      "learning_rate": 3.460309115760698e-05,
      "loss": 2.2536,
      "step": 87900
    },
    {
      "epoch": 0.3082746293765114,
      "grad_norm": 0.5360828042030334,
      "learning_rate": 3.4592577016086635e-05,
      "loss": 2.2705,
      "step": 87960
    },
    {
      "epoch": 0.3084849122069183,
      "grad_norm": 0.5943337082862854,
      "learning_rate": 3.4582062874566293e-05,
      "loss": 2.2732,
      "step": 88020
    },
    {
      "epoch": 0.3086951950373252,
      "grad_norm": 0.5922872424125671,
      "learning_rate": 3.457154873304595e-05,
      "loss": 2.2563,
      "step": 88080
    },
    {
      "epoch": 0.3089054778677321,
      "grad_norm": 0.5607509613037109,
      "learning_rate": 3.456120982721761e-05,
      "loss": 2.2701,
      "step": 88140
    },
    {
      "epoch": 0.309115760698139,
      "grad_norm": 0.6036087870597839,
      "learning_rate": 3.455069568569727e-05,
      "loss": 2.2575,
      "step": 88200
    },
    {
      "epoch": 0.30932604352854587,
      "grad_norm": 0.5599225163459778,
      "learning_rate": 3.454018154417692e-05,
      "loss": 2.2664,
      "step": 88260
    },
    {
      "epoch": 0.3095363263589528,
      "grad_norm": 0.5826290249824524,
      "learning_rate": 3.452966740265657e-05,
      "loss": 2.2645,
      "step": 88320
    },
    {
      "epoch": 0.3097466091893597,
      "grad_norm": 0.5541812777519226,
      "learning_rate": 3.451915326113623e-05,
      "loss": 2.256,
      "step": 88380
    },
    {
      "epoch": 0.30995689201976656,
      "grad_norm": 0.5463770627975464,
      "learning_rate": 3.450863911961589e-05,
      "loss": 2.2494,
      "step": 88440
    },
    {
      "epoch": 0.3101671748501735,
      "grad_norm": 0.554961085319519,
      "learning_rate": 3.449812497809554e-05,
      "loss": 2.2609,
      "step": 88500
    },
    {
      "epoch": 0.3103774576805804,
      "grad_norm": 0.5289806127548218,
      "learning_rate": 3.448761083657519e-05,
      "loss": 2.2712,
      "step": 88560
    },
    {
      "epoch": 0.31058774051098725,
      "grad_norm": 0.5908351540565491,
      "learning_rate": 3.447709669505485e-05,
      "loss": 2.2589,
      "step": 88620
    },
    {
      "epoch": 0.31079802334139417,
      "grad_norm": 0.5531824231147766,
      "learning_rate": 3.4466582553534503e-05,
      "loss": 2.2567,
      "step": 88680
    },
    {
      "epoch": 0.3110083061718011,
      "grad_norm": 0.6163316369056702,
      "learning_rate": 3.4456068412014155e-05,
      "loss": 2.2585,
      "step": 88740
    },
    {
      "epoch": 0.31121858900220795,
      "grad_norm": 0.527815580368042,
      "learning_rate": 3.444555427049382e-05,
      "loss": 2.27,
      "step": 88800
    },
    {
      "epoch": 0.31142887183261486,
      "grad_norm": 0.5704740285873413,
      "learning_rate": 3.443504012897347e-05,
      "loss": 2.2537,
      "step": 88860
    },
    {
      "epoch": 0.3116391546630218,
      "grad_norm": 0.5804399847984314,
      "learning_rate": 3.4424525987453125e-05,
      "loss": 2.2601,
      "step": 88920
    },
    {
      "epoch": 0.31184943749342864,
      "grad_norm": 0.5516076683998108,
      "learning_rate": 3.441401184593278e-05,
      "loss": 2.2565,
      "step": 88980
    },
    {
      "epoch": 0.31205972032383555,
      "grad_norm": 0.5606161952018738,
      "learning_rate": 3.4403497704412435e-05,
      "loss": 2.2563,
      "step": 89040
    },
    {
      "epoch": 0.31227000315424247,
      "grad_norm": 0.5703347325325012,
      "learning_rate": 3.439298356289209e-05,
      "loss": 2.2505,
      "step": 89100
    },
    {
      "epoch": 0.31248028598464933,
      "grad_norm": 0.5731121301651001,
      "learning_rate": 3.438246942137175e-05,
      "loss": 2.2589,
      "step": 89160
    },
    {
      "epoch": 0.31269056881505625,
      "grad_norm": 0.5524592995643616,
      "learning_rate": 3.4371955279851404e-05,
      "loss": 2.2535,
      "step": 89220
    },
    {
      "epoch": 0.31290085164546316,
      "grad_norm": 0.5376356244087219,
      "learning_rate": 3.4361441138331056e-05,
      "loss": 2.2718,
      "step": 89280
    },
    {
      "epoch": 0.31311113447587,
      "grad_norm": 0.5426408052444458,
      "learning_rate": 3.4350926996810715e-05,
      "loss": 2.2534,
      "step": 89340
    },
    {
      "epoch": 0.31332141730627694,
      "grad_norm": 0.5779808759689331,
      "learning_rate": 3.434041285529037e-05,
      "loss": 2.2609,
      "step": 89400
    },
    {
      "epoch": 0.31353170013668386,
      "grad_norm": 0.5719578862190247,
      "learning_rate": 3.432989871377002e-05,
      "loss": 2.2715,
      "step": 89460
    },
    {
      "epoch": 0.3137419829670907,
      "grad_norm": 0.5462282299995422,
      "learning_rate": 3.431938457224968e-05,
      "loss": 2.2761,
      "step": 89520
    },
    {
      "epoch": 0.31395226579749763,
      "grad_norm": 0.5807678699493408,
      "learning_rate": 3.4308870430729336e-05,
      "loss": 2.2539,
      "step": 89580
    },
    {
      "epoch": 0.31416254862790455,
      "grad_norm": 0.5386305451393127,
      "learning_rate": 3.429835628920899e-05,
      "loss": 2.2604,
      "step": 89640
    },
    {
      "epoch": 0.3143728314583114,
      "grad_norm": 0.5434944033622742,
      "learning_rate": 3.428784214768864e-05,
      "loss": 2.2533,
      "step": 89700
    },
    {
      "epoch": 0.3145831142887183,
      "grad_norm": 0.5666850209236145,
      "learning_rate": 3.42773280061683e-05,
      "loss": 2.2525,
      "step": 89760
    },
    {
      "epoch": 0.31479339711912524,
      "grad_norm": 0.5859403014183044,
      "learning_rate": 3.426681386464795e-05,
      "loss": 2.2715,
      "step": 89820
    },
    {
      "epoch": 0.3150036799495321,
      "grad_norm": 0.6053626537322998,
      "learning_rate": 3.425629972312761e-05,
      "loss": 2.2643,
      "step": 89880
    },
    {
      "epoch": 0.315213962779939,
      "grad_norm": 0.6361802816390991,
      "learning_rate": 3.424578558160727e-05,
      "loss": 2.268,
      "step": 89940
    },
    {
      "epoch": 0.31542424561034593,
      "grad_norm": 0.6449795961380005,
      "learning_rate": 3.423527144008692e-05,
      "loss": 2.2574,
      "step": 90000
    },
    {
      "epoch": 0.3156345284407528,
      "grad_norm": 0.57269287109375,
      "learning_rate": 3.422475729856657e-05,
      "loss": 2.2697,
      "step": 90060
    },
    {
      "epoch": 0.3158448112711597,
      "grad_norm": 0.5491210222244263,
      "learning_rate": 3.421424315704623e-05,
      "loss": 2.2598,
      "step": 90120
    },
    {
      "epoch": 0.3160550941015666,
      "grad_norm": 0.5665245652198792,
      "learning_rate": 3.420372901552588e-05,
      "loss": 2.2456,
      "step": 90180
    },
    {
      "epoch": 0.3162653769319735,
      "grad_norm": 0.5336186289787292,
      "learning_rate": 3.419321487400554e-05,
      "loss": 2.2659,
      "step": 90240
    },
    {
      "epoch": 0.3164756597623804,
      "grad_norm": 0.539472222328186,
      "learning_rate": 3.418270073248519e-05,
      "loss": 2.2555,
      "step": 90300
    },
    {
      "epoch": 0.3166859425927873,
      "grad_norm": 0.6023342609405518,
      "learning_rate": 3.417218659096485e-05,
      "loss": 2.2755,
      "step": 90360
    },
    {
      "epoch": 0.3168962254231942,
      "grad_norm": 0.5498049855232239,
      "learning_rate": 3.41616724494445e-05,
      "loss": 2.2514,
      "step": 90420
    },
    {
      "epoch": 0.3171065082536011,
      "grad_norm": 0.5622275471687317,
      "learning_rate": 3.4151158307924155e-05,
      "loss": 2.2569,
      "step": 90480
    },
    {
      "epoch": 0.317316791084008,
      "grad_norm": 0.5502556562423706,
      "learning_rate": 3.4140644166403814e-05,
      "loss": 2.2784,
      "step": 90540
    },
    {
      "epoch": 0.31752707391441487,
      "grad_norm": 0.5545588731765747,
      "learning_rate": 3.413013002488347e-05,
      "loss": 2.2701,
      "step": 90600
    },
    {
      "epoch": 0.3177373567448218,
      "grad_norm": 0.5292103886604309,
      "learning_rate": 3.4119615883363124e-05,
      "loss": 2.268,
      "step": 90660
    },
    {
      "epoch": 0.3179476395752287,
      "grad_norm": 0.6121399402618408,
      "learning_rate": 3.410910174184278e-05,
      "loss": 2.2552,
      "step": 90720
    },
    {
      "epoch": 0.31815792240563556,
      "grad_norm": 0.6200298070907593,
      "learning_rate": 3.4098587600322435e-05,
      "loss": 2.2734,
      "step": 90780
    },
    {
      "epoch": 0.3183682052360425,
      "grad_norm": 0.5391733646392822,
      "learning_rate": 3.408807345880209e-05,
      "loss": 2.2668,
      "step": 90840
    },
    {
      "epoch": 0.3185784880664494,
      "grad_norm": 0.5934666395187378,
      "learning_rate": 3.4077559317281745e-05,
      "loss": 2.2687,
      "step": 90900
    },
    {
      "epoch": 0.31878877089685626,
      "grad_norm": 0.5021581053733826,
      "learning_rate": 3.4067045175761404e-05,
      "loss": 2.2582,
      "step": 90960
    },
    {
      "epoch": 0.31899905372726317,
      "grad_norm": 0.7432253956794739,
      "learning_rate": 3.4056531034241056e-05,
      "loss": 2.2719,
      "step": 91020
    },
    {
      "epoch": 0.3192093365576701,
      "grad_norm": 0.5567831993103027,
      "learning_rate": 3.4046016892720715e-05,
      "loss": 2.2607,
      "step": 91080
    },
    {
      "epoch": 0.31941961938807695,
      "grad_norm": 0.5400500893592834,
      "learning_rate": 3.4035502751200366e-05,
      "loss": 2.2592,
      "step": 91140
    },
    {
      "epoch": 0.31962990221848386,
      "grad_norm": 0.5561904907226562,
      "learning_rate": 3.402498860968002e-05,
      "loss": 2.2624,
      "step": 91200
    },
    {
      "epoch": 0.3198401850488908,
      "grad_norm": 0.6148996949195862,
      "learning_rate": 3.401447446815968e-05,
      "loss": 2.2486,
      "step": 91260
    },
    {
      "epoch": 0.32005046787929764,
      "grad_norm": 0.5738281607627869,
      "learning_rate": 3.4003960326639336e-05,
      "loss": 2.2401,
      "step": 91320
    },
    {
      "epoch": 0.32026075070970456,
      "grad_norm": 0.5140102505683899,
      "learning_rate": 3.399344618511899e-05,
      "loss": 2.2603,
      "step": 91380
    },
    {
      "epoch": 0.32047103354011147,
      "grad_norm": 0.6118866205215454,
      "learning_rate": 3.398293204359864e-05,
      "loss": 2.2582,
      "step": 91440
    },
    {
      "epoch": 0.32068131637051833,
      "grad_norm": 0.5432943105697632,
      "learning_rate": 3.39724179020783e-05,
      "loss": 2.2531,
      "step": 91500
    },
    {
      "epoch": 0.32089159920092525,
      "grad_norm": 0.5328633785247803,
      "learning_rate": 3.396190376055795e-05,
      "loss": 2.2455,
      "step": 91560
    },
    {
      "epoch": 0.32110188203133216,
      "grad_norm": 0.5897437930107117,
      "learning_rate": 3.39513896190376e-05,
      "loss": 2.2739,
      "step": 91620
    },
    {
      "epoch": 0.321312164861739,
      "grad_norm": 0.5548875331878662,
      "learning_rate": 3.394087547751727e-05,
      "loss": 2.2699,
      "step": 91680
    },
    {
      "epoch": 0.32152244769214594,
      "grad_norm": 0.5671025514602661,
      "learning_rate": 3.393036133599692e-05,
      "loss": 2.2655,
      "step": 91740
    },
    {
      "epoch": 0.32173273052255286,
      "grad_norm": 0.5613976716995239,
      "learning_rate": 3.391984719447657e-05,
      "loss": 2.265,
      "step": 91800
    },
    {
      "epoch": 0.3219430133529597,
      "grad_norm": 0.5736813545227051,
      "learning_rate": 3.390933305295623e-05,
      "loss": 2.2543,
      "step": 91860
    },
    {
      "epoch": 0.32215329618336663,
      "grad_norm": 0.5219939947128296,
      "learning_rate": 3.389881891143588e-05,
      "loss": 2.2588,
      "step": 91920
    },
    {
      "epoch": 0.32236357901377355,
      "grad_norm": 0.5467992424964905,
      "learning_rate": 3.3888304769915534e-05,
      "loss": 2.2552,
      "step": 91980
    },
    {
      "epoch": 0.3225738618441804,
      "grad_norm": 0.572704017162323,
      "learning_rate": 3.387779062839519e-05,
      "loss": 2.2651,
      "step": 92040
    },
    {
      "epoch": 0.3227841446745873,
      "grad_norm": 0.5442659258842468,
      "learning_rate": 3.386727648687485e-05,
      "loss": 2.2542,
      "step": 92100
    },
    {
      "epoch": 0.32299442750499424,
      "grad_norm": 0.6716965436935425,
      "learning_rate": 3.38567623453545e-05,
      "loss": 2.2431,
      "step": 92160
    },
    {
      "epoch": 0.3232047103354011,
      "grad_norm": 0.5147510170936584,
      "learning_rate": 3.3846248203834155e-05,
      "loss": 2.2577,
      "step": 92220
    },
    {
      "epoch": 0.323414993165808,
      "grad_norm": 0.523707389831543,
      "learning_rate": 3.3835734062313813e-05,
      "loss": 2.2588,
      "step": 92280
    },
    {
      "epoch": 0.32362527599621493,
      "grad_norm": 0.5979717373847961,
      "learning_rate": 3.3825219920793465e-05,
      "loss": 2.2678,
      "step": 92340
    },
    {
      "epoch": 0.3238355588266218,
      "grad_norm": 0.5593013763427734,
      "learning_rate": 3.381488101496513e-05,
      "loss": 2.2686,
      "step": 92400
    },
    {
      "epoch": 0.3240458416570287,
      "grad_norm": 0.5766727924346924,
      "learning_rate": 3.380436687344479e-05,
      "loss": 2.2619,
      "step": 92460
    },
    {
      "epoch": 0.3242561244874356,
      "grad_norm": 0.5491504669189453,
      "learning_rate": 3.379385273192444e-05,
      "loss": 2.2583,
      "step": 92520
    },
    {
      "epoch": 0.3244664073178425,
      "grad_norm": 0.570665717124939,
      "learning_rate": 3.37833385904041e-05,
      "loss": 2.2549,
      "step": 92580
    },
    {
      "epoch": 0.3246766901482494,
      "grad_norm": 0.7060606479644775,
      "learning_rate": 3.377282444888375e-05,
      "loss": 2.2543,
      "step": 92640
    },
    {
      "epoch": 0.3248869729786563,
      "grad_norm": 0.5789251327514648,
      "learning_rate": 3.37623103073634e-05,
      "loss": 2.2744,
      "step": 92700
    },
    {
      "epoch": 0.3250972558090632,
      "grad_norm": 0.6093466281890869,
      "learning_rate": 3.375179616584306e-05,
      "loss": 2.2537,
      "step": 92760
    },
    {
      "epoch": 0.3253075386394701,
      "grad_norm": 0.6546159982681274,
      "learning_rate": 3.374128202432272e-05,
      "loss": 2.2615,
      "step": 92820
    },
    {
      "epoch": 0.325517821469877,
      "grad_norm": 0.5653020739555359,
      "learning_rate": 3.373076788280237e-05,
      "loss": 2.2522,
      "step": 92880
    },
    {
      "epoch": 0.32572810430028387,
      "grad_norm": 0.5708128213882446,
      "learning_rate": 3.372025374128202e-05,
      "loss": 2.2478,
      "step": 92940
    },
    {
      "epoch": 0.3259383871306908,
      "grad_norm": 0.5512128472328186,
      "learning_rate": 3.370973959976168e-05,
      "loss": 2.2609,
      "step": 93000
    },
    {
      "epoch": 0.3261486699610977,
      "grad_norm": 0.5367293953895569,
      "learning_rate": 3.3699225458241334e-05,
      "loss": 2.2584,
      "step": 93060
    },
    {
      "epoch": 0.32635895279150456,
      "grad_norm": 0.5934156775474548,
      "learning_rate": 3.3688711316720986e-05,
      "loss": 2.2551,
      "step": 93120
    },
    {
      "epoch": 0.3265692356219115,
      "grad_norm": 0.5676745772361755,
      "learning_rate": 3.367819717520065e-05,
      "loss": 2.2589,
      "step": 93180
    },
    {
      "epoch": 0.32677951845231834,
      "grad_norm": 0.5735729932785034,
      "learning_rate": 3.36676830336803e-05,
      "loss": 2.2476,
      "step": 93240
    },
    {
      "epoch": 0.32698980128272526,
      "grad_norm": 0.5471116304397583,
      "learning_rate": 3.3657168892159955e-05,
      "loss": 2.2497,
      "step": 93300
    },
    {
      "epoch": 0.3272000841131322,
      "grad_norm": 0.5850885510444641,
      "learning_rate": 3.3646654750639614e-05,
      "loss": 2.2593,
      "step": 93360
    },
    {
      "epoch": 0.32741036694353903,
      "grad_norm": 0.5297808647155762,
      "learning_rate": 3.3636140609119266e-05,
      "loss": 2.2455,
      "step": 93420
    },
    {
      "epoch": 0.32762064977394595,
      "grad_norm": 0.5714202523231506,
      "learning_rate": 3.362562646759892e-05,
      "loss": 2.2594,
      "step": 93480
    },
    {
      "epoch": 0.32783093260435286,
      "grad_norm": 0.6715126633644104,
      "learning_rate": 3.3615112326078576e-05,
      "loss": 2.2555,
      "step": 93540
    },
    {
      "epoch": 0.3280412154347597,
      "grad_norm": 0.5289738178253174,
      "learning_rate": 3.3604598184558235e-05,
      "loss": 2.2483,
      "step": 93600
    },
    {
      "epoch": 0.32825149826516664,
      "grad_norm": 0.5812007784843445,
      "learning_rate": 3.359408404303789e-05,
      "loss": 2.2677,
      "step": 93660
    },
    {
      "epoch": 0.32846178109557356,
      "grad_norm": 0.5916142463684082,
      "learning_rate": 3.3583569901517545e-05,
      "loss": 2.2611,
      "step": 93720
    },
    {
      "epoch": 0.3286720639259804,
      "grad_norm": 0.575390100479126,
      "learning_rate": 3.35730557599972e-05,
      "loss": 2.2667,
      "step": 93780
    },
    {
      "epoch": 0.32888234675638733,
      "grad_norm": 0.559893012046814,
      "learning_rate": 3.356254161847685e-05,
      "loss": 2.2486,
      "step": 93840
    },
    {
      "epoch": 0.32909262958679425,
      "grad_norm": 0.5065498948097229,
      "learning_rate": 3.355202747695651e-05,
      "loss": 2.2562,
      "step": 93900
    },
    {
      "epoch": 0.3293029124172011,
      "grad_norm": 0.6010339856147766,
      "learning_rate": 3.3541513335436166e-05,
      "loss": 2.2599,
      "step": 93960
    },
    {
      "epoch": 0.329513195247608,
      "grad_norm": 0.5235238671302795,
      "learning_rate": 3.353099919391582e-05,
      "loss": 2.2639,
      "step": 94020
    },
    {
      "epoch": 0.32972347807801494,
      "grad_norm": 0.48841866850852966,
      "learning_rate": 3.352048505239547e-05,
      "loss": 2.2623,
      "step": 94080
    },
    {
      "epoch": 0.3299337609084218,
      "grad_norm": 0.5267378687858582,
      "learning_rate": 3.350997091087513e-05,
      "loss": 2.2622,
      "step": 94140
    },
    {
      "epoch": 0.3301440437388287,
      "grad_norm": 0.6126842498779297,
      "learning_rate": 3.349945676935478e-05,
      "loss": 2.266,
      "step": 94200
    },
    {
      "epoch": 0.33035432656923563,
      "grad_norm": 0.5331740379333496,
      "learning_rate": 3.348894262783444e-05,
      "loss": 2.2528,
      "step": 94260
    },
    {
      "epoch": 0.3305646093996425,
      "grad_norm": 0.5459609031677246,
      "learning_rate": 3.34784284863141e-05,
      "loss": 2.2501,
      "step": 94320
    },
    {
      "epoch": 0.3307748922300494,
      "grad_norm": 0.5753666162490845,
      "learning_rate": 3.346791434479375e-05,
      "loss": 2.241,
      "step": 94380
    },
    {
      "epoch": 0.3309851750604563,
      "grad_norm": 0.5683814287185669,
      "learning_rate": 3.34574002032734e-05,
      "loss": 2.2516,
      "step": 94440
    },
    {
      "epoch": 0.3311954578908632,
      "grad_norm": 0.5722650289535522,
      "learning_rate": 3.3447061297445066e-05,
      "loss": 2.2594,
      "step": 94500
    },
    {
      "epoch": 0.3314057407212701,
      "grad_norm": 0.5353830456733704,
      "learning_rate": 3.343654715592472e-05,
      "loss": 2.2654,
      "step": 94560
    },
    {
      "epoch": 0.331616023551677,
      "grad_norm": 0.5557458400726318,
      "learning_rate": 3.3426033014404376e-05,
      "loss": 2.2348,
      "step": 94620
    },
    {
      "epoch": 0.3318263063820839,
      "grad_norm": 0.5775477886199951,
      "learning_rate": 3.3415518872884035e-05,
      "loss": 2.2588,
      "step": 94680
    },
    {
      "epoch": 0.3320365892124908,
      "grad_norm": 0.6219456195831299,
      "learning_rate": 3.340500473136369e-05,
      "loss": 2.2453,
      "step": 94740
    },
    {
      "epoch": 0.3322468720428977,
      "grad_norm": 0.5657910108566284,
      "learning_rate": 3.339449058984334e-05,
      "loss": 2.2507,
      "step": 94800
    },
    {
      "epoch": 0.3324571548733046,
      "grad_norm": 0.5465739369392395,
      "learning_rate": 3.3383976448323e-05,
      "loss": 2.2523,
      "step": 94860
    },
    {
      "epoch": 0.3326674377037115,
      "grad_norm": 0.5367757081985474,
      "learning_rate": 3.337346230680265e-05,
      "loss": 2.2658,
      "step": 94920
    },
    {
      "epoch": 0.3328777205341184,
      "grad_norm": 0.5744025707244873,
      "learning_rate": 3.33629481652823e-05,
      "loss": 2.2767,
      "step": 94980
    },
    {
      "epoch": 0.33308800336452526,
      "grad_norm": 0.5318028330802917,
      "learning_rate": 3.335243402376197e-05,
      "loss": 2.2766,
      "step": 95040
    },
    {
      "epoch": 0.3332982861949322,
      "grad_norm": 0.5595847368240356,
      "learning_rate": 3.334191988224162e-05,
      "loss": 2.2378,
      "step": 95100
    },
    {
      "epoch": 0.3335085690253391,
      "grad_norm": 0.5904885530471802,
      "learning_rate": 3.333140574072127e-05,
      "loss": 2.2638,
      "step": 95160
    },
    {
      "epoch": 0.33371885185574596,
      "grad_norm": 0.5781721472740173,
      "learning_rate": 3.332089159920093e-05,
      "loss": 2.2609,
      "step": 95220
    },
    {
      "epoch": 0.3339291346861529,
      "grad_norm": 0.5953025221824646,
      "learning_rate": 3.331037745768058e-05,
      "loss": 2.2427,
      "step": 95280
    },
    {
      "epoch": 0.3341394175165598,
      "grad_norm": 0.5774268507957458,
      "learning_rate": 3.329986331616023e-05,
      "loss": 2.2708,
      "step": 95340
    },
    {
      "epoch": 0.33434970034696665,
      "grad_norm": 0.6212142705917358,
      "learning_rate": 3.328934917463989e-05,
      "loss": 2.2473,
      "step": 95400
    },
    {
      "epoch": 0.33455998317737357,
      "grad_norm": 0.6178370714187622,
      "learning_rate": 3.327883503311955e-05,
      "loss": 2.2389,
      "step": 95460
    },
    {
      "epoch": 0.3347702660077805,
      "grad_norm": 0.630793035030365,
      "learning_rate": 3.32683208915992e-05,
      "loss": 2.2598,
      "step": 95520
    },
    {
      "epoch": 0.33498054883818734,
      "grad_norm": 0.6330223083496094,
      "learning_rate": 3.3257806750078854e-05,
      "loss": 2.2453,
      "step": 95580
    },
    {
      "epoch": 0.33519083166859426,
      "grad_norm": 0.6235046982765198,
      "learning_rate": 3.324729260855851e-05,
      "loss": 2.2458,
      "step": 95640
    },
    {
      "epoch": 0.3354011144990012,
      "grad_norm": 0.5611142516136169,
      "learning_rate": 3.3236778467038165e-05,
      "loss": 2.2608,
      "step": 95700
    },
    {
      "epoch": 0.33561139732940803,
      "grad_norm": 0.5523935556411743,
      "learning_rate": 3.322626432551782e-05,
      "loss": 2.2592,
      "step": 95760
    },
    {
      "epoch": 0.33582168015981495,
      "grad_norm": 0.5253689289093018,
      "learning_rate": 3.321575018399748e-05,
      "loss": 2.2541,
      "step": 95820
    },
    {
      "epoch": 0.33603196299022187,
      "grad_norm": 0.578079104423523,
      "learning_rate": 3.3205236042477134e-05,
      "loss": 2.2629,
      "step": 95880
    },
    {
      "epoch": 0.3362422458206287,
      "grad_norm": 0.5487265586853027,
      "learning_rate": 3.3194721900956786e-05,
      "loss": 2.2482,
      "step": 95940
    },
    {
      "epoch": 0.33645252865103564,
      "grad_norm": 0.5192381739616394,
      "learning_rate": 3.3184207759436444e-05,
      "loss": 2.2556,
      "step": 96000
    },
    {
      "epoch": 0.33666281148144256,
      "grad_norm": 0.6124681234359741,
      "learning_rate": 3.3173693617916096e-05,
      "loss": 2.2545,
      "step": 96060
    },
    {
      "epoch": 0.3368730943118494,
      "grad_norm": 0.5543643832206726,
      "learning_rate": 3.3163179476395755e-05,
      "loss": 2.2566,
      "step": 96120
    },
    {
      "epoch": 0.33708337714225634,
      "grad_norm": 0.5779359936714172,
      "learning_rate": 3.315266533487541e-05,
      "loss": 2.2612,
      "step": 96180
    },
    {
      "epoch": 0.33729365997266325,
      "grad_norm": 0.5384775400161743,
      "learning_rate": 3.3142151193355066e-05,
      "loss": 2.258,
      "step": 96240
    },
    {
      "epoch": 0.3375039428030701,
      "grad_norm": 0.6284984946250916,
      "learning_rate": 3.313163705183472e-05,
      "loss": 2.2617,
      "step": 96300
    },
    {
      "epoch": 0.337714225633477,
      "grad_norm": 0.5733190774917603,
      "learning_rate": 3.312112291031437e-05,
      "loss": 2.2411,
      "step": 96360
    },
    {
      "epoch": 0.33792450846388394,
      "grad_norm": 0.5528522729873657,
      "learning_rate": 3.311060876879403e-05,
      "loss": 2.2512,
      "step": 96420
    },
    {
      "epoch": 0.3381347912942908,
      "grad_norm": 0.6423466205596924,
      "learning_rate": 3.310009462727369e-05,
      "loss": 2.2564,
      "step": 96480
    },
    {
      "epoch": 0.3383450741246977,
      "grad_norm": 0.5352001786231995,
      "learning_rate": 3.308975572144535e-05,
      "loss": 2.2525,
      "step": 96540
    },
    {
      "epoch": 0.33855535695510464,
      "grad_norm": 0.5893850922584534,
      "learning_rate": 3.3079241579925e-05,
      "loss": 2.2474,
      "step": 96600
    },
    {
      "epoch": 0.3387656397855115,
      "grad_norm": 0.5544783473014832,
      "learning_rate": 3.3068727438404654e-05,
      "loss": 2.2595,
      "step": 96660
    },
    {
      "epoch": 0.3389759226159184,
      "grad_norm": 0.5462645292282104,
      "learning_rate": 3.305821329688431e-05,
      "loss": 2.2581,
      "step": 96720
    },
    {
      "epoch": 0.33918620544632533,
      "grad_norm": 0.5372522473335266,
      "learning_rate": 3.3047699155363965e-05,
      "loss": 2.2553,
      "step": 96780
    },
    {
      "epoch": 0.3393964882767322,
      "grad_norm": 0.5661049485206604,
      "learning_rate": 3.303718501384362e-05,
      "loss": 2.2475,
      "step": 96840
    },
    {
      "epoch": 0.3396067711071391,
      "grad_norm": 0.5360636711120605,
      "learning_rate": 3.3026670872323276e-05,
      "loss": 2.2632,
      "step": 96900
    },
    {
      "epoch": 0.339817053937546,
      "grad_norm": 0.5690281987190247,
      "learning_rate": 3.3016156730802934e-05,
      "loss": 2.2427,
      "step": 96960
    },
    {
      "epoch": 0.3400273367679529,
      "grad_norm": 0.5777586102485657,
      "learning_rate": 3.3005642589282586e-05,
      "loss": 2.2635,
      "step": 97020
    },
    {
      "epoch": 0.3402376195983598,
      "grad_norm": 0.570685625076294,
      "learning_rate": 3.299512844776224e-05,
      "loss": 2.2486,
      "step": 97080
    },
    {
      "epoch": 0.3404479024287667,
      "grad_norm": 0.5590705275535583,
      "learning_rate": 3.29846143062419e-05,
      "loss": 2.259,
      "step": 97140
    },
    {
      "epoch": 0.3406581852591736,
      "grad_norm": 0.5950242877006531,
      "learning_rate": 3.297410016472155e-05,
      "loss": 2.2606,
      "step": 97200
    },
    {
      "epoch": 0.3408684680895805,
      "grad_norm": 0.5764995813369751,
      "learning_rate": 3.296358602320121e-05,
      "loss": 2.2652,
      "step": 97260
    },
    {
      "epoch": 0.3410787509199874,
      "grad_norm": 0.5674739480018616,
      "learning_rate": 3.2953071881680866e-05,
      "loss": 2.2421,
      "step": 97320
    },
    {
      "epoch": 0.34128903375039427,
      "grad_norm": 0.5347117781639099,
      "learning_rate": 3.294255774016052e-05,
      "loss": 2.2452,
      "step": 97380
    },
    {
      "epoch": 0.3414993165808012,
      "grad_norm": 0.5449021458625793,
      "learning_rate": 3.293204359864017e-05,
      "loss": 2.2509,
      "step": 97440
    },
    {
      "epoch": 0.3417095994112081,
      "grad_norm": 0.5273362994194031,
      "learning_rate": 3.292152945711983e-05,
      "loss": 2.2523,
      "step": 97500
    },
    {
      "epoch": 0.34191988224161496,
      "grad_norm": 0.5723702907562256,
      "learning_rate": 3.291101531559948e-05,
      "loss": 2.248,
      "step": 97560
    },
    {
      "epoch": 0.3421301650720219,
      "grad_norm": 0.5732927918434143,
      "learning_rate": 3.290050117407914e-05,
      "loss": 2.2491,
      "step": 97620
    },
    {
      "epoch": 0.3423404479024288,
      "grad_norm": 0.5697137117385864,
      "learning_rate": 3.28899870325588e-05,
      "loss": 2.2542,
      "step": 97680
    },
    {
      "epoch": 0.34255073073283565,
      "grad_norm": 0.5517762899398804,
      "learning_rate": 3.287947289103845e-05,
      "loss": 2.258,
      "step": 97740
    },
    {
      "epoch": 0.34276101356324257,
      "grad_norm": 0.5736147165298462,
      "learning_rate": 3.28689587495181e-05,
      "loss": 2.2695,
      "step": 97800
    },
    {
      "epoch": 0.3429712963936495,
      "grad_norm": 0.5735359191894531,
      "learning_rate": 3.285844460799776e-05,
      "loss": 2.2411,
      "step": 97860
    },
    {
      "epoch": 0.34318157922405634,
      "grad_norm": 0.5607940554618835,
      "learning_rate": 3.284793046647741e-05,
      "loss": 2.2488,
      "step": 97920
    },
    {
      "epoch": 0.34339186205446326,
      "grad_norm": 0.5465531349182129,
      "learning_rate": 3.283741632495707e-05,
      "loss": 2.2673,
      "step": 97980
    },
    {
      "epoch": 0.3436021448848702,
      "grad_norm": 0.6117886900901794,
      "learning_rate": 3.282690218343672e-05,
      "loss": 2.2608,
      "step": 98040
    },
    {
      "epoch": 0.34381242771527704,
      "grad_norm": 0.5727432370185852,
      "learning_rate": 3.281638804191638e-05,
      "loss": 2.2667,
      "step": 98100
    },
    {
      "epoch": 0.34402271054568395,
      "grad_norm": 0.5974268913269043,
      "learning_rate": 3.280587390039603e-05,
      "loss": 2.2436,
      "step": 98160
    },
    {
      "epoch": 0.34423299337609087,
      "grad_norm": 0.5546492338180542,
      "learning_rate": 3.2795359758875685e-05,
      "loss": 2.2442,
      "step": 98220
    },
    {
      "epoch": 0.34444327620649773,
      "grad_norm": 0.5560635328292847,
      "learning_rate": 3.2784845617355344e-05,
      "loss": 2.2408,
      "step": 98280
    },
    {
      "epoch": 0.34465355903690464,
      "grad_norm": 0.570353090763092,
      "learning_rate": 3.2774331475835e-05,
      "loss": 2.2486,
      "step": 98340
    },
    {
      "epoch": 0.34486384186731156,
      "grad_norm": 0.6234951615333557,
      "learning_rate": 3.2763817334314654e-05,
      "loss": 2.2398,
      "step": 98400
    },
    {
      "epoch": 0.3450741246977184,
      "grad_norm": 0.5658458471298218,
      "learning_rate": 3.275330319279431e-05,
      "loss": 2.2492,
      "step": 98460
    },
    {
      "epoch": 0.34528440752812534,
      "grad_norm": 0.6331504583358765,
      "learning_rate": 3.2742789051273965e-05,
      "loss": 2.2568,
      "step": 98520
    },
    {
      "epoch": 0.34549469035853225,
      "grad_norm": 0.5121544003486633,
      "learning_rate": 3.273245014544563e-05,
      "loss": 2.2432,
      "step": 98580
    },
    {
      "epoch": 0.3457049731889391,
      "grad_norm": 0.5852991342544556,
      "learning_rate": 3.272193600392528e-05,
      "loss": 2.2823,
      "step": 98640
    },
    {
      "epoch": 0.34591525601934603,
      "grad_norm": 0.5442538261413574,
      "learning_rate": 3.271142186240493e-05,
      "loss": 2.2429,
      "step": 98700
    },
    {
      "epoch": 0.34612553884975295,
      "grad_norm": 0.5579994320869446,
      "learning_rate": 3.270090772088459e-05,
      "loss": 2.235,
      "step": 98760
    },
    {
      "epoch": 0.3463358216801598,
      "grad_norm": 0.5348595976829529,
      "learning_rate": 3.269039357936425e-05,
      "loss": 2.247,
      "step": 98820
    },
    {
      "epoch": 0.3465461045105667,
      "grad_norm": 0.5732651352882385,
      "learning_rate": 3.26798794378439e-05,
      "loss": 2.249,
      "step": 98880
    },
    {
      "epoch": 0.3467563873409736,
      "grad_norm": 0.5671512484550476,
      "learning_rate": 3.2669365296323553e-05,
      "loss": 2.2515,
      "step": 98940
    },
    {
      "epoch": 0.3469666701713805,
      "grad_norm": 0.5615571737289429,
      "learning_rate": 3.265885115480321e-05,
      "loss": 2.2373,
      "step": 99000
    },
    {
      "epoch": 0.3471769530017874,
      "grad_norm": 0.5443949699401855,
      "learning_rate": 3.2648337013282864e-05,
      "loss": 2.2428,
      "step": 99060
    },
    {
      "epoch": 0.3473872358321943,
      "grad_norm": 0.5558658838272095,
      "learning_rate": 3.263782287176252e-05,
      "loss": 2.2556,
      "step": 99120
    },
    {
      "epoch": 0.3475975186626012,
      "grad_norm": 0.7935243248939514,
      "learning_rate": 3.262730873024218e-05,
      "loss": 2.2481,
      "step": 99180
    },
    {
      "epoch": 0.3478078014930081,
      "grad_norm": 0.5348489880561829,
      "learning_rate": 3.261679458872183e-05,
      "loss": 2.2458,
      "step": 99240
    },
    {
      "epoch": 0.34801808432341497,
      "grad_norm": 0.5577119588851929,
      "learning_rate": 3.2606280447201485e-05,
      "loss": 2.2317,
      "step": 99300
    },
    {
      "epoch": 0.3482283671538219,
      "grad_norm": 0.6064680218696594,
      "learning_rate": 3.2595766305681144e-05,
      "loss": 2.2508,
      "step": 99360
    },
    {
      "epoch": 0.3484386499842288,
      "grad_norm": 0.534321129322052,
      "learning_rate": 3.2585252164160796e-05,
      "loss": 2.2448,
      "step": 99420
    },
    {
      "epoch": 0.34864893281463566,
      "grad_norm": 0.6524277925491333,
      "learning_rate": 3.2574738022640454e-05,
      "loss": 2.2563,
      "step": 99480
    },
    {
      "epoch": 0.3488592156450426,
      "grad_norm": 0.5712284445762634,
      "learning_rate": 3.2564223881120106e-05,
      "loss": 2.2582,
      "step": 99540
    },
    {
      "epoch": 0.3490694984754495,
      "grad_norm": 0.6521320343017578,
      "learning_rate": 3.2553709739599765e-05,
      "loss": 2.2551,
      "step": 99600
    },
    {
      "epoch": 0.34927978130585635,
      "grad_norm": 0.5247908234596252,
      "learning_rate": 3.254319559807942e-05,
      "loss": 2.2573,
      "step": 99660
    },
    {
      "epoch": 0.34949006413626327,
      "grad_norm": 0.5874356627464294,
      "learning_rate": 3.253268145655907e-05,
      "loss": 2.2516,
      "step": 99720
    },
    {
      "epoch": 0.3497003469666702,
      "grad_norm": 0.5620294213294983,
      "learning_rate": 3.252216731503873e-05,
      "loss": 2.2459,
      "step": 99780
    },
    {
      "epoch": 0.34991062979707704,
      "grad_norm": 0.5107635259628296,
      "learning_rate": 3.2511653173518386e-05,
      "loss": 2.2529,
      "step": 99840
    },
    {
      "epoch": 0.35012091262748396,
      "grad_norm": 0.5404499173164368,
      "learning_rate": 3.250113903199804e-05,
      "loss": 2.254,
      "step": 99900
    },
    {
      "epoch": 0.3503311954578909,
      "grad_norm": 0.6408905982971191,
      "learning_rate": 3.24906248904777e-05,
      "loss": 2.2524,
      "step": 99960
    },
    {
      "epoch": 0.35054147828829774,
      "grad_norm": 0.5619672536849976,
      "learning_rate": 3.248011074895735e-05,
      "loss": 2.262,
      "step": 100020
    },
    {
      "epoch": 0.35075176111870465,
      "grad_norm": 0.5574461817741394,
      "learning_rate": 3.2469596607437e-05,
      "loss": 2.2523,
      "step": 100080
    },
    {
      "epoch": 0.35096204394911157,
      "grad_norm": 0.5604238510131836,
      "learning_rate": 3.245908246591666e-05,
      "loss": 2.25,
      "step": 100140
    },
    {
      "epoch": 0.35117232677951843,
      "grad_norm": 0.5675678849220276,
      "learning_rate": 3.244856832439632e-05,
      "loss": 2.2511,
      "step": 100200
    },
    {
      "epoch": 0.35138260960992534,
      "grad_norm": 0.5655011534690857,
      "learning_rate": 3.243805418287597e-05,
      "loss": 2.2651,
      "step": 100260
    },
    {
      "epoch": 0.35159289244033226,
      "grad_norm": 0.5608187317848206,
      "learning_rate": 3.2427715277047634e-05,
      "loss": 2.2427,
      "step": 100320
    },
    {
      "epoch": 0.3518031752707391,
      "grad_norm": 0.597961962223053,
      "learning_rate": 3.2417201135527285e-05,
      "loss": 2.2632,
      "step": 100380
    },
    {
      "epoch": 0.35201345810114604,
      "grad_norm": 0.6023119688034058,
      "learning_rate": 3.240668699400694e-05,
      "loss": 2.2564,
      "step": 100440
    },
    {
      "epoch": 0.35222374093155295,
      "grad_norm": 0.5368738174438477,
      "learning_rate": 3.2396172852486596e-05,
      "loss": 2.2621,
      "step": 100500
    },
    {
      "epoch": 0.3524340237619598,
      "grad_norm": 0.5581424832344055,
      "learning_rate": 3.238565871096625e-05,
      "loss": 2.2489,
      "step": 100560
    },
    {
      "epoch": 0.35264430659236673,
      "grad_norm": 0.5462180376052856,
      "learning_rate": 3.2375144569445907e-05,
      "loss": 2.2416,
      "step": 100620
    },
    {
      "epoch": 0.35285458942277365,
      "grad_norm": 0.5432653427124023,
      "learning_rate": 3.2364630427925565e-05,
      "loss": 2.2484,
      "step": 100680
    },
    {
      "epoch": 0.3530648722531805,
      "grad_norm": 0.7162202596664429,
      "learning_rate": 3.235411628640522e-05,
      "loss": 2.2335,
      "step": 100740
    },
    {
      "epoch": 0.3532751550835874,
      "grad_norm": 0.5798888802528381,
      "learning_rate": 3.234360214488487e-05,
      "loss": 2.2471,
      "step": 100800
    },
    {
      "epoch": 0.35348543791399434,
      "grad_norm": 0.56444251537323,
      "learning_rate": 3.233308800336453e-05,
      "loss": 2.2548,
      "step": 100860
    },
    {
      "epoch": 0.3536957207444012,
      "grad_norm": 0.5319738388061523,
      "learning_rate": 3.232257386184418e-05,
      "loss": 2.2546,
      "step": 100920
    },
    {
      "epoch": 0.3539060035748081,
      "grad_norm": 0.5734124779701233,
      "learning_rate": 3.231205972032384e-05,
      "loss": 2.2415,
      "step": 100980
    },
    {
      "epoch": 0.35411628640521503,
      "grad_norm": 0.5583657026290894,
      "learning_rate": 3.23015455788035e-05,
      "loss": 2.2401,
      "step": 101040
    },
    {
      "epoch": 0.3543265692356219,
      "grad_norm": 0.5711774826049805,
      "learning_rate": 3.229103143728315e-05,
      "loss": 2.2468,
      "step": 101100
    },
    {
      "epoch": 0.3545368520660288,
      "grad_norm": 0.5239578485488892,
      "learning_rate": 3.22805172957628e-05,
      "loss": 2.2571,
      "step": 101160
    },
    {
      "epoch": 0.3547471348964357,
      "grad_norm": 0.56839919090271,
      "learning_rate": 3.227000315424246e-05,
      "loss": 2.2466,
      "step": 101220
    },
    {
      "epoch": 0.3549574177268426,
      "grad_norm": 0.585908830165863,
      "learning_rate": 3.225948901272211e-05,
      "loss": 2.2559,
      "step": 101280
    },
    {
      "epoch": 0.3551677005572495,
      "grad_norm": 0.628909707069397,
      "learning_rate": 3.224897487120177e-05,
      "loss": 2.2551,
      "step": 101340
    },
    {
      "epoch": 0.3553779833876564,
      "grad_norm": 0.6182568073272705,
      "learning_rate": 3.223846072968142e-05,
      "loss": 2.2379,
      "step": 101400
    },
    {
      "epoch": 0.3555882662180633,
      "grad_norm": 0.5343201160430908,
      "learning_rate": 3.222794658816108e-05,
      "loss": 2.2346,
      "step": 101460
    },
    {
      "epoch": 0.3557985490484702,
      "grad_norm": 0.5548133850097656,
      "learning_rate": 3.221743244664073e-05,
      "loss": 2.2461,
      "step": 101520
    },
    {
      "epoch": 0.3560088318788771,
      "grad_norm": 0.6163520812988281,
      "learning_rate": 3.2206918305120384e-05,
      "loss": 2.2595,
      "step": 101580
    },
    {
      "epoch": 0.35621911470928397,
      "grad_norm": 0.5295227766036987,
      "learning_rate": 3.219640416360004e-05,
      "loss": 2.2445,
      "step": 101640
    },
    {
      "epoch": 0.3564293975396909,
      "grad_norm": 0.5932655930519104,
      "learning_rate": 3.21858900220797e-05,
      "loss": 2.2535,
      "step": 101700
    },
    {
      "epoch": 0.3566396803700978,
      "grad_norm": 0.5422408580780029,
      "learning_rate": 3.2175375880559353e-05,
      "loss": 2.2535,
      "step": 101760
    },
    {
      "epoch": 0.35684996320050466,
      "grad_norm": 0.5753630995750427,
      "learning_rate": 3.216486173903901e-05,
      "loss": 2.2419,
      "step": 101820
    },
    {
      "epoch": 0.3570602460309116,
      "grad_norm": 0.5593083500862122,
      "learning_rate": 3.2154347597518664e-05,
      "loss": 2.2441,
      "step": 101880
    },
    {
      "epoch": 0.3572705288613185,
      "grad_norm": 0.6243857145309448,
      "learning_rate": 3.2143833455998316e-05,
      "loss": 2.2362,
      "step": 101940
    },
    {
      "epoch": 0.35748081169172535,
      "grad_norm": 0.5592348575592041,
      "learning_rate": 3.2133319314477975e-05,
      "loss": 2.2408,
      "step": 102000
    },
    {
      "epoch": 0.35769109452213227,
      "grad_norm": 0.5877608060836792,
      "learning_rate": 3.212280517295763e-05,
      "loss": 2.2413,
      "step": 102060
    },
    {
      "epoch": 0.3579013773525392,
      "grad_norm": 0.5965809226036072,
      "learning_rate": 3.2112291031437285e-05,
      "loss": 2.2449,
      "step": 102120
    },
    {
      "epoch": 0.35811166018294605,
      "grad_norm": 0.5869651436805725,
      "learning_rate": 3.210177688991694e-05,
      "loss": 2.2437,
      "step": 102180
    },
    {
      "epoch": 0.35832194301335296,
      "grad_norm": 0.6029491424560547,
      "learning_rate": 3.2091262748396596e-05,
      "loss": 2.2413,
      "step": 102240
    },
    {
      "epoch": 0.3585322258437599,
      "grad_norm": 0.628241777420044,
      "learning_rate": 3.208074860687625e-05,
      "loss": 2.251,
      "step": 102300
    },
    {
      "epoch": 0.35874250867416674,
      "grad_norm": 0.5701780319213867,
      "learning_rate": 3.20702344653559e-05,
      "loss": 2.2615,
      "step": 102360
    },
    {
      "epoch": 0.35895279150457365,
      "grad_norm": 0.5451123714447021,
      "learning_rate": 3.2059720323835565e-05,
      "loss": 2.2572,
      "step": 102420
    },
    {
      "epoch": 0.35916307433498057,
      "grad_norm": 0.5642544627189636,
      "learning_rate": 3.204920618231522e-05,
      "loss": 2.2575,
      "step": 102480
    },
    {
      "epoch": 0.35937335716538743,
      "grad_norm": 0.6085901260375977,
      "learning_rate": 3.203869204079487e-05,
      "loss": 2.2534,
      "step": 102540
    },
    {
      "epoch": 0.35958363999579435,
      "grad_norm": Infinity,
      "learning_rate": 3.202835313496653e-05,
      "loss": 2.2543,
      "step": 102600
    },
    {
      "epoch": 0.35979392282620126,
      "grad_norm": 0.6252976059913635,
      "learning_rate": 3.2017838993446185e-05,
      "loss": 2.2496,
      "step": 102660
    },
    {
      "epoch": 0.3600042056566081,
      "grad_norm": 0.6806179285049438,
      "learning_rate": 3.200732485192584e-05,
      "loss": 2.255,
      "step": 102720
    },
    {
      "epoch": 0.36021448848701504,
      "grad_norm": 0.5911528468132019,
      "learning_rate": 3.1996810710405495e-05,
      "loss": 2.247,
      "step": 102780
    },
    {
      "epoch": 0.36042477131742195,
      "grad_norm": 0.5509805083274841,
      "learning_rate": 3.1986296568885154e-05,
      "loss": 2.2584,
      "step": 102840
    },
    {
      "epoch": 0.3606350541478288,
      "grad_norm": 0.5932069420814514,
      "learning_rate": 3.1975782427364806e-05,
      "loss": 2.2533,
      "step": 102900
    },
    {
      "epoch": 0.36084533697823573,
      "grad_norm": 0.5719885230064392,
      "learning_rate": 3.1965268285844464e-05,
      "loss": 2.2415,
      "step": 102960
    },
    {
      "epoch": 0.36105561980864265,
      "grad_norm": 0.5080400705337524,
      "learning_rate": 3.1954754144324116e-05,
      "loss": 2.257,
      "step": 103020
    },
    {
      "epoch": 0.3612659026390495,
      "grad_norm": 0.5492014288902283,
      "learning_rate": 3.194424000280377e-05,
      "loss": 2.2447,
      "step": 103080
    },
    {
      "epoch": 0.3614761854694564,
      "grad_norm": 0.5836805105209351,
      "learning_rate": 3.193372586128343e-05,
      "loss": 2.2559,
      "step": 103140
    },
    {
      "epoch": 0.36168646829986334,
      "grad_norm": 0.563239574432373,
      "learning_rate": 3.1923211719763085e-05,
      "loss": 2.2493,
      "step": 103200
    },
    {
      "epoch": 0.3618967511302702,
      "grad_norm": 0.5607362389564514,
      "learning_rate": 3.191269757824274e-05,
      "loss": 2.2417,
      "step": 103260
    },
    {
      "epoch": 0.3621070339606771,
      "grad_norm": 0.5219749808311462,
      "learning_rate": 3.1902183436722396e-05,
      "loss": 2.2553,
      "step": 103320
    },
    {
      "epoch": 0.36231731679108403,
      "grad_norm": 0.5371667742729187,
      "learning_rate": 3.189166929520205e-05,
      "loss": 2.2527,
      "step": 103380
    },
    {
      "epoch": 0.3625275996214909,
      "grad_norm": 0.6101118326187134,
      "learning_rate": 3.18811551536817e-05,
      "loss": 2.2503,
      "step": 103440
    },
    {
      "epoch": 0.3627378824518978,
      "grad_norm": 0.578412652015686,
      "learning_rate": 3.187064101216136e-05,
      "loss": 2.2537,
      "step": 103500
    },
    {
      "epoch": 0.3629481652823047,
      "grad_norm": 0.5483526587486267,
      "learning_rate": 3.186012687064102e-05,
      "loss": 2.2547,
      "step": 103560
    },
    {
      "epoch": 0.3631584481127116,
      "grad_norm": 0.5780754685401917,
      "learning_rate": 3.184961272912067e-05,
      "loss": 2.2482,
      "step": 103620
    },
    {
      "epoch": 0.3633687309431185,
      "grad_norm": 0.5565093159675598,
      "learning_rate": 3.183909858760033e-05,
      "loss": 2.2445,
      "step": 103680
    },
    {
      "epoch": 0.3635790137735254,
      "grad_norm": 0.5821536183357239,
      "learning_rate": 3.182858444607998e-05,
      "loss": 2.2456,
      "step": 103740
    },
    {
      "epoch": 0.3637892966039323,
      "grad_norm": 0.5497275590896606,
      "learning_rate": 3.181807030455963e-05,
      "loss": 2.2416,
      "step": 103800
    },
    {
      "epoch": 0.3639995794343392,
      "grad_norm": 0.5313774347305298,
      "learning_rate": 3.180755616303929e-05,
      "loss": 2.2555,
      "step": 103860
    },
    {
      "epoch": 0.3642098622647461,
      "grad_norm": 0.5351090431213379,
      "learning_rate": 3.179704202151895e-05,
      "loss": 2.2355,
      "step": 103920
    },
    {
      "epoch": 0.36442014509515297,
      "grad_norm": 0.5801105499267578,
      "learning_rate": 3.17865278799986e-05,
      "loss": 2.2415,
      "step": 103980
    },
    {
      "epoch": 0.3646304279255599,
      "grad_norm": 0.5337709784507751,
      "learning_rate": 3.177601373847825e-05,
      "loss": 2.2519,
      "step": 104040
    },
    {
      "epoch": 0.3648407107559668,
      "grad_norm": 0.6514520049095154,
      "learning_rate": 3.176549959695791e-05,
      "loss": 2.2415,
      "step": 104100
    },
    {
      "epoch": 0.36505099358637366,
      "grad_norm": 0.5500940680503845,
      "learning_rate": 3.175498545543756e-05,
      "loss": 2.2488,
      "step": 104160
    },
    {
      "epoch": 0.3652612764167806,
      "grad_norm": 0.6053885221481323,
      "learning_rate": 3.1744471313917215e-05,
      "loss": 2.2521,
      "step": 104220
    },
    {
      "epoch": 0.3654715592471875,
      "grad_norm": 0.5517741441726685,
      "learning_rate": 3.173395717239688e-05,
      "loss": 2.2446,
      "step": 104280
    },
    {
      "epoch": 0.36568184207759435,
      "grad_norm": 0.5213547348976135,
      "learning_rate": 3.172344303087653e-05,
      "loss": 2.2482,
      "step": 104340
    },
    {
      "epoch": 0.36589212490800127,
      "grad_norm": 0.5915991067886353,
      "learning_rate": 3.1712928889356184e-05,
      "loss": 2.2425,
      "step": 104400
    },
    {
      "epoch": 0.3661024077384082,
      "grad_norm": 0.5863001346588135,
      "learning_rate": 3.170241474783584e-05,
      "loss": 2.2535,
      "step": 104460
    },
    {
      "epoch": 0.36631269056881505,
      "grad_norm": 0.5261284708976746,
      "learning_rate": 3.1691900606315495e-05,
      "loss": 2.2545,
      "step": 104520
    },
    {
      "epoch": 0.36652297339922196,
      "grad_norm": 0.5854162573814392,
      "learning_rate": 3.168138646479515e-05,
      "loss": 2.2545,
      "step": 104580
    },
    {
      "epoch": 0.3667332562296288,
      "grad_norm": 0.5391114354133606,
      "learning_rate": 3.1670872323274805e-05,
      "loss": 2.2462,
      "step": 104640
    },
    {
      "epoch": 0.36694353906003574,
      "grad_norm": 0.551304280757904,
      "learning_rate": 3.1660358181754464e-05,
      "loss": 2.2438,
      "step": 104700
    },
    {
      "epoch": 0.36715382189044266,
      "grad_norm": 0.5472719073295593,
      "learning_rate": 3.1649844040234116e-05,
      "loss": 2.2427,
      "step": 104760
    },
    {
      "epoch": 0.3673641047208495,
      "grad_norm": 0.5311259031295776,
      "learning_rate": 3.163932989871377e-05,
      "loss": 2.2521,
      "step": 104820
    },
    {
      "epoch": 0.36757438755125643,
      "grad_norm": 0.6197320818901062,
      "learning_rate": 3.1628815757193427e-05,
      "loss": 2.2467,
      "step": 104880
    },
    {
      "epoch": 0.36778467038166335,
      "grad_norm": 0.5505039095878601,
      "learning_rate": 3.161830161567308e-05,
      "loss": 2.2456,
      "step": 104940
    },
    {
      "epoch": 0.3679949532120702,
      "grad_norm": 0.5435019135475159,
      "learning_rate": 3.160778747415274e-05,
      "loss": 2.2502,
      "step": 105000
    },
    {
      "epoch": 0.3682052360424771,
      "grad_norm": 0.5461676716804504,
      "learning_rate": 3.1597273332632396e-05,
      "loss": 2.2565,
      "step": 105060
    },
    {
      "epoch": 0.36841551887288404,
      "grad_norm": 0.5456289649009705,
      "learning_rate": 3.158693442680405e-05,
      "loss": 2.2574,
      "step": 105120
    },
    {
      "epoch": 0.3686258017032909,
      "grad_norm": 0.4910428524017334,
      "learning_rate": 3.157642028528371e-05,
      "loss": 2.2321,
      "step": 105180
    },
    {
      "epoch": 0.3688360845336978,
      "grad_norm": 0.5281582474708557,
      "learning_rate": 3.1565906143763363e-05,
      "loss": 2.237,
      "step": 105240
    },
    {
      "epoch": 0.36904636736410473,
      "grad_norm": 0.5398959517478943,
      "learning_rate": 3.1555392002243015e-05,
      "loss": 2.2487,
      "step": 105300
    },
    {
      "epoch": 0.3692566501945116,
      "grad_norm": 0.5760674476623535,
      "learning_rate": 3.1544877860722674e-05,
      "loss": 2.2384,
      "step": 105360
    },
    {
      "epoch": 0.3694669330249185,
      "grad_norm": 0.5500927567481995,
      "learning_rate": 3.153436371920233e-05,
      "loss": 2.2577,
      "step": 105420
    },
    {
      "epoch": 0.3696772158553254,
      "grad_norm": 0.5500715970993042,
      "learning_rate": 3.1523849577681985e-05,
      "loss": 2.2433,
      "step": 105480
    },
    {
      "epoch": 0.3698874986857323,
      "grad_norm": 0.5827727317810059,
      "learning_rate": 3.1513335436161636e-05,
      "loss": 2.2465,
      "step": 105540
    },
    {
      "epoch": 0.3700977815161392,
      "grad_norm": 0.5601902604103088,
      "learning_rate": 3.1502821294641295e-05,
      "loss": 2.25,
      "step": 105600
    },
    {
      "epoch": 0.3703080643465461,
      "grad_norm": 0.561234712600708,
      "learning_rate": 3.149230715312095e-05,
      "loss": 2.2434,
      "step": 105660
    },
    {
      "epoch": 0.370518347176953,
      "grad_norm": 0.5326609015464783,
      "learning_rate": 3.14817930116006e-05,
      "loss": 2.2481,
      "step": 105720
    },
    {
      "epoch": 0.3707286300073599,
      "grad_norm": 0.6148693561553955,
      "learning_rate": 3.1471278870080264e-05,
      "loss": 2.2336,
      "step": 105780
    },
    {
      "epoch": 0.3709389128377668,
      "grad_norm": 0.5935849547386169,
      "learning_rate": 3.1460764728559916e-05,
      "loss": 2.2555,
      "step": 105840
    },
    {
      "epoch": 0.37114919566817367,
      "grad_norm": 0.5288058519363403,
      "learning_rate": 3.145025058703957e-05,
      "loss": 2.2413,
      "step": 105900
    },
    {
      "epoch": 0.3713594784985806,
      "grad_norm": 0.5751053690910339,
      "learning_rate": 3.143973644551923e-05,
      "loss": 2.2566,
      "step": 105960
    },
    {
      "epoch": 0.3715697613289875,
      "grad_norm": 0.528931200504303,
      "learning_rate": 3.142922230399888e-05,
      "loss": 2.236,
      "step": 106020
    },
    {
      "epoch": 0.37178004415939436,
      "grad_norm": 0.5181067585945129,
      "learning_rate": 3.141870816247853e-05,
      "loss": 2.2493,
      "step": 106080
    },
    {
      "epoch": 0.3719903269898013,
      "grad_norm": 0.5679474472999573,
      "learning_rate": 3.1408194020958196e-05,
      "loss": 2.2428,
      "step": 106140
    },
    {
      "epoch": 0.3722006098202082,
      "grad_norm": 0.5477309226989746,
      "learning_rate": 3.139767987943785e-05,
      "loss": 2.2476,
      "step": 106200
    },
    {
      "epoch": 0.37241089265061506,
      "grad_norm": 0.531114399433136,
      "learning_rate": 3.13871657379175e-05,
      "loss": 2.2521,
      "step": 106260
    },
    {
      "epoch": 0.37262117548102197,
      "grad_norm": 0.6383332014083862,
      "learning_rate": 3.137665159639716e-05,
      "loss": 2.2458,
      "step": 106320
    },
    {
      "epoch": 0.3728314583114289,
      "grad_norm": 0.6344099640846252,
      "learning_rate": 3.136613745487681e-05,
      "loss": 2.2645,
      "step": 106380
    },
    {
      "epoch": 0.37304174114183575,
      "grad_norm": 0.5700390934944153,
      "learning_rate": 3.135562331335646e-05,
      "loss": 2.2407,
      "step": 106440
    },
    {
      "epoch": 0.37325202397224266,
      "grad_norm": 0.5498290061950684,
      "learning_rate": 3.134510917183612e-05,
      "loss": 2.2249,
      "step": 106500
    },
    {
      "epoch": 0.3734623068026496,
      "grad_norm": 0.560438871383667,
      "learning_rate": 3.133459503031578e-05,
      "loss": 2.2556,
      "step": 106560
    },
    {
      "epoch": 0.37367258963305644,
      "grad_norm": 0.5165132284164429,
      "learning_rate": 3.132408088879543e-05,
      "loss": 2.231,
      "step": 106620
    },
    {
      "epoch": 0.37388287246346336,
      "grad_norm": 0.7113255262374878,
      "learning_rate": 3.131356674727508e-05,
      "loss": 2.2551,
      "step": 106680
    },
    {
      "epoch": 0.37409315529387027,
      "grad_norm": 0.5605031847953796,
      "learning_rate": 3.130305260575474e-05,
      "loss": 2.2449,
      "step": 106740
    },
    {
      "epoch": 0.37430343812427713,
      "grad_norm": 0.49553272128105164,
      "learning_rate": 3.1292538464234394e-05,
      "loss": 2.2307,
      "step": 106800
    },
    {
      "epoch": 0.37451372095468405,
      "grad_norm": 0.558988094329834,
      "learning_rate": 3.128202432271405e-05,
      "loss": 2.2406,
      "step": 106860
    },
    {
      "epoch": 0.37472400378509096,
      "grad_norm": 0.5793077349662781,
      "learning_rate": 3.127151018119371e-05,
      "loss": 2.2598,
      "step": 106920
    },
    {
      "epoch": 0.3749342866154978,
      "grad_norm": 0.6070676445960999,
      "learning_rate": 3.126099603967336e-05,
      "loss": 2.246,
      "step": 106980
    },
    {
      "epoch": 0.37514456944590474,
      "grad_norm": 0.5537981390953064,
      "learning_rate": 3.1250481898153015e-05,
      "loss": 2.2416,
      "step": 107040
    },
    {
      "epoch": 0.37535485227631166,
      "grad_norm": 0.6169135570526123,
      "learning_rate": 3.1239967756632674e-05,
      "loss": 2.2434,
      "step": 107100
    },
    {
      "epoch": 0.3755651351067185,
      "grad_norm": 0.546450138092041,
      "learning_rate": 3.1229453615112326e-05,
      "loss": 2.2498,
      "step": 107160
    },
    {
      "epoch": 0.37577541793712543,
      "grad_norm": 0.5148147940635681,
      "learning_rate": 3.1218939473591984e-05,
      "loss": 2.2508,
      "step": 107220
    },
    {
      "epoch": 0.37598570076753235,
      "grad_norm": 0.5672466158866882,
      "learning_rate": 3.1208425332071636e-05,
      "loss": 2.2454,
      "step": 107280
    },
    {
      "epoch": 0.3761959835979392,
      "grad_norm": 0.5429059267044067,
      "learning_rate": 3.1197911190551295e-05,
      "loss": 2.2466,
      "step": 107340
    },
    {
      "epoch": 0.3764062664283461,
      "grad_norm": 0.6130840182304382,
      "learning_rate": 3.118739704903095e-05,
      "loss": 2.2551,
      "step": 107400
    },
    {
      "epoch": 0.37661654925875304,
      "grad_norm": 0.5849588513374329,
      "learning_rate": 3.11768829075106e-05,
      "loss": 2.2551,
      "step": 107460
    },
    {
      "epoch": 0.3768268320891599,
      "grad_norm": 0.6018226146697998,
      "learning_rate": 3.116636876599026e-05,
      "loss": 2.2457,
      "step": 107520
    },
    {
      "epoch": 0.3770371149195668,
      "grad_norm": 0.5993386507034302,
      "learning_rate": 3.1155854624469916e-05,
      "loss": 2.2304,
      "step": 107580
    },
    {
      "epoch": 0.37724739774997373,
      "grad_norm": 0.6036729216575623,
      "learning_rate": 3.114534048294957e-05,
      "loss": 2.2313,
      "step": 107640
    },
    {
      "epoch": 0.3774576805803806,
      "grad_norm": 0.5473097562789917,
      "learning_rate": 3.1134826341429227e-05,
      "loss": 2.2337,
      "step": 107700
    },
    {
      "epoch": 0.3776679634107875,
      "grad_norm": 0.526314914226532,
      "learning_rate": 3.112431219990888e-05,
      "loss": 2.2359,
      "step": 107760
    },
    {
      "epoch": 0.3778782462411944,
      "grad_norm": 0.5568339824676514,
      "learning_rate": 3.111379805838853e-05,
      "loss": 2.232,
      "step": 107820
    },
    {
      "epoch": 0.3780885290716013,
      "grad_norm": 0.5934057235717773,
      "learning_rate": 3.110328391686819e-05,
      "loss": 2.249,
      "step": 107880
    },
    {
      "epoch": 0.3782988119020082,
      "grad_norm": 0.5619835257530212,
      "learning_rate": 3.109276977534785e-05,
      "loss": 2.2361,
      "step": 107940
    },
    {
      "epoch": 0.3785090947324151,
      "grad_norm": 0.5756205320358276,
      "learning_rate": 3.10822556338275e-05,
      "loss": 2.238,
      "step": 108000
    },
    {
      "epoch": 0.378719377562822,
      "grad_norm": 0.5448141694068909,
      "learning_rate": 3.107174149230715e-05,
      "loss": 2.2498,
      "step": 108060
    },
    {
      "epoch": 0.3789296603932289,
      "grad_norm": 0.5649576783180237,
      "learning_rate": 3.106122735078681e-05,
      "loss": 2.2352,
      "step": 108120
    },
    {
      "epoch": 0.3791399432236358,
      "grad_norm": 0.5113899111747742,
      "learning_rate": 3.105071320926646e-05,
      "loss": 2.2526,
      "step": 108180
    },
    {
      "epoch": 0.37935022605404267,
      "grad_norm": 0.5608682036399841,
      "learning_rate": 3.104019906774612e-05,
      "loss": 2.2474,
      "step": 108240
    },
    {
      "epoch": 0.3795605088844496,
      "grad_norm": 0.5401106476783752,
      "learning_rate": 3.102968492622578e-05,
      "loss": 2.2552,
      "step": 108300
    },
    {
      "epoch": 0.3797707917148565,
      "grad_norm": 0.5243397951126099,
      "learning_rate": 3.101917078470543e-05,
      "loss": 2.2306,
      "step": 108360
    },
    {
      "epoch": 0.37998107454526336,
      "grad_norm": 0.5430238246917725,
      "learning_rate": 3.100865664318508e-05,
      "loss": 2.2566,
      "step": 108420
    },
    {
      "epoch": 0.3801913573756703,
      "grad_norm": 0.5654410719871521,
      "learning_rate": 3.099814250166474e-05,
      "loss": 2.2548,
      "step": 108480
    },
    {
      "epoch": 0.3804016402060772,
      "grad_norm": 0.6027359366416931,
      "learning_rate": 3.0987628360144394e-05,
      "loss": 2.2466,
      "step": 108540
    },
    {
      "epoch": 0.38061192303648406,
      "grad_norm": 0.547913670539856,
      "learning_rate": 3.0977114218624046e-05,
      "loss": 2.2357,
      "step": 108600
    },
    {
      "epoch": 0.380822205866891,
      "grad_norm": 0.5652959942817688,
      "learning_rate": 3.096660007710371e-05,
      "loss": 2.2411,
      "step": 108660
    },
    {
      "epoch": 0.3810324886972979,
      "grad_norm": 0.5216713547706604,
      "learning_rate": 3.095608593558336e-05,
      "loss": 2.2482,
      "step": 108720
    },
    {
      "epoch": 0.38124277152770475,
      "grad_norm": 0.5479360222816467,
      "learning_rate": 3.0945571794063015e-05,
      "loss": 2.2338,
      "step": 108780
    },
    {
      "epoch": 0.38145305435811167,
      "grad_norm": 0.5628937482833862,
      "learning_rate": 3.0935057652542673e-05,
      "loss": 2.2531,
      "step": 108840
    },
    {
      "epoch": 0.3816633371885186,
      "grad_norm": 0.5069226622581482,
      "learning_rate": 3.0924543511022325e-05,
      "loss": 2.2533,
      "step": 108900
    },
    {
      "epoch": 0.38187362001892544,
      "grad_norm": 0.5676142573356628,
      "learning_rate": 3.091402936950198e-05,
      "loss": 2.2329,
      "step": 108960
    },
    {
      "epoch": 0.38208390284933236,
      "grad_norm": 0.50908362865448,
      "learning_rate": 3.0903515227981636e-05,
      "loss": 2.2376,
      "step": 109020
    },
    {
      "epoch": 0.3822941856797393,
      "grad_norm": 0.5596504211425781,
      "learning_rate": 3.0893001086461295e-05,
      "loss": 2.2539,
      "step": 109080
    },
    {
      "epoch": 0.38250446851014613,
      "grad_norm": 0.5801405906677246,
      "learning_rate": 3.0882486944940946e-05,
      "loss": 2.2478,
      "step": 109140
    },
    {
      "epoch": 0.38271475134055305,
      "grad_norm": 0.5595638751983643,
      "learning_rate": 3.08719728034206e-05,
      "loss": 2.247,
      "step": 109200
    },
    {
      "epoch": 0.38292503417095997,
      "grad_norm": 0.5531991124153137,
      "learning_rate": 3.086163389759226e-05,
      "loss": 2.2325,
      "step": 109260
    },
    {
      "epoch": 0.3831353170013668,
      "grad_norm": 0.5818877816200256,
      "learning_rate": 3.0851119756071914e-05,
      "loss": 2.2396,
      "step": 109320
    },
    {
      "epoch": 0.38334559983177374,
      "grad_norm": 0.550900399684906,
      "learning_rate": 3.084060561455157e-05,
      "loss": 2.2317,
      "step": 109380
    },
    {
      "epoch": 0.38355588266218066,
      "grad_norm": 0.5770589709281921,
      "learning_rate": 3.083009147303123e-05,
      "loss": 2.2427,
      "step": 109440
    },
    {
      "epoch": 0.3837661654925875,
      "grad_norm": 0.5315117239952087,
      "learning_rate": 3.081957733151088e-05,
      "loss": 2.2436,
      "step": 109500
    },
    {
      "epoch": 0.38397644832299443,
      "grad_norm": 0.5745555758476257,
      "learning_rate": 3.080906318999054e-05,
      "loss": 2.2297,
      "step": 109560
    },
    {
      "epoch": 0.38418673115340135,
      "grad_norm": 0.5811091661453247,
      "learning_rate": 3.0798549048470194e-05,
      "loss": 2.2418,
      "step": 109620
    },
    {
      "epoch": 0.3843970139838082,
      "grad_norm": 0.6055805683135986,
      "learning_rate": 3.0788034906949846e-05,
      "loss": 2.2344,
      "step": 109680
    },
    {
      "epoch": 0.3846072968142151,
      "grad_norm": 0.5680029392242432,
      "learning_rate": 3.0777520765429504e-05,
      "loss": 2.2571,
      "step": 109740
    },
    {
      "epoch": 0.38481757964462204,
      "grad_norm": 0.5134311318397522,
      "learning_rate": 3.076700662390916e-05,
      "loss": 2.2575,
      "step": 109800
    },
    {
      "epoch": 0.3850278624750289,
      "grad_norm": 0.5660803914070129,
      "learning_rate": 3.0756492482388815e-05,
      "loss": 2.2517,
      "step": 109860
    },
    {
      "epoch": 0.3852381453054358,
      "grad_norm": 0.5265384912490845,
      "learning_rate": 3.074597834086847e-05,
      "loss": 2.2545,
      "step": 109920
    },
    {
      "epoch": 0.38544842813584274,
      "grad_norm": 0.4810716211795807,
      "learning_rate": 3.0735464199348126e-05,
      "loss": 2.2417,
      "step": 109980
    },
    {
      "epoch": 0.3856587109662496,
      "grad_norm": 0.516758143901825,
      "learning_rate": 3.072495005782778e-05,
      "loss": 2.2409,
      "step": 110040
    },
    {
      "epoch": 0.3858689937966565,
      "grad_norm": 0.5345315337181091,
      "learning_rate": 3.071443591630743e-05,
      "loss": 2.2502,
      "step": 110100
    },
    {
      "epoch": 0.38607927662706343,
      "grad_norm": 0.5337819457054138,
      "learning_rate": 3.0703921774787095e-05,
      "loss": 2.2227,
      "step": 110160
    },
    {
      "epoch": 0.3862895594574703,
      "grad_norm": 0.5085687637329102,
      "learning_rate": 3.069340763326675e-05,
      "loss": 2.247,
      "step": 110220
    },
    {
      "epoch": 0.3864998422878772,
      "grad_norm": 0.5997688174247742,
      "learning_rate": 3.06828934917464e-05,
      "loss": 2.2545,
      "step": 110280
    },
    {
      "epoch": 0.38671012511828406,
      "grad_norm": 0.6165827512741089,
      "learning_rate": 3.067237935022606e-05,
      "loss": 2.2427,
      "step": 110340
    },
    {
      "epoch": 0.386920407948691,
      "grad_norm": 0.6079306602478027,
      "learning_rate": 3.066186520870571e-05,
      "loss": 2.2245,
      "step": 110400
    },
    {
      "epoch": 0.3871306907790979,
      "grad_norm": 0.5653731226921082,
      "learning_rate": 3.065135106718536e-05,
      "loss": 2.2416,
      "step": 110460
    },
    {
      "epoch": 0.38734097360950476,
      "grad_norm": 0.5574362277984619,
      "learning_rate": 3.064083692566502e-05,
      "loss": 2.2417,
      "step": 110520
    },
    {
      "epoch": 0.3875512564399117,
      "grad_norm": 0.5841783285140991,
      "learning_rate": 3.063032278414468e-05,
      "loss": 2.2382,
      "step": 110580
    },
    {
      "epoch": 0.3877615392703186,
      "grad_norm": 0.5657100677490234,
      "learning_rate": 3.061980864262433e-05,
      "loss": 2.2302,
      "step": 110640
    },
    {
      "epoch": 0.38797182210072545,
      "grad_norm": 0.5238633155822754,
      "learning_rate": 3.060929450110398e-05,
      "loss": 2.2378,
      "step": 110700
    },
    {
      "epoch": 0.38818210493113237,
      "grad_norm": 0.5261026620864868,
      "learning_rate": 3.059878035958364e-05,
      "loss": 2.239,
      "step": 110760
    },
    {
      "epoch": 0.3883923877615393,
      "grad_norm": 0.5436447858810425,
      "learning_rate": 3.058826621806329e-05,
      "loss": 2.2478,
      "step": 110820
    },
    {
      "epoch": 0.38860267059194614,
      "grad_norm": 0.5481789112091064,
      "learning_rate": 3.057775207654295e-05,
      "loss": 2.2473,
      "step": 110880
    },
    {
      "epoch": 0.38881295342235306,
      "grad_norm": 0.6277444958686829,
      "learning_rate": 3.056723793502261e-05,
      "loss": 2.2423,
      "step": 110940
    },
    {
      "epoch": 0.38902323625276,
      "grad_norm": 0.5612151622772217,
      "learning_rate": 3.055672379350226e-05,
      "loss": 2.2426,
      "step": 111000
    },
    {
      "epoch": 0.38923351908316683,
      "grad_norm": 0.563457727432251,
      "learning_rate": 3.0546209651981914e-05,
      "loss": 2.2374,
      "step": 111060
    },
    {
      "epoch": 0.38944380191357375,
      "grad_norm": 0.5553929805755615,
      "learning_rate": 3.053569551046157e-05,
      "loss": 2.2368,
      "step": 111120
    },
    {
      "epoch": 0.38965408474398067,
      "grad_norm": 0.6097853183746338,
      "learning_rate": 3.0525181368941224e-05,
      "loss": 2.2409,
      "step": 111180
    },
    {
      "epoch": 0.3898643675743875,
      "grad_norm": 0.5294462442398071,
      "learning_rate": 3.0514667227420883e-05,
      "loss": 2.2392,
      "step": 111240
    },
    {
      "epoch": 0.39007465040479444,
      "grad_norm": 0.5460028648376465,
      "learning_rate": 3.0504328321592547e-05,
      "loss": 2.2325,
      "step": 111300
    },
    {
      "epoch": 0.39028493323520136,
      "grad_norm": 0.5618406534194946,
      "learning_rate": 3.04938141800722e-05,
      "loss": 2.2472,
      "step": 111360
    },
    {
      "epoch": 0.3904952160656082,
      "grad_norm": 0.5654280185699463,
      "learning_rate": 3.0483300038551854e-05,
      "loss": 2.242,
      "step": 111420
    },
    {
      "epoch": 0.39070549889601514,
      "grad_norm": 0.5807438492774963,
      "learning_rate": 3.047278589703151e-05,
      "loss": 2.2386,
      "step": 111480
    },
    {
      "epoch": 0.39091578172642205,
      "grad_norm": 0.6127345561981201,
      "learning_rate": 3.046227175551116e-05,
      "loss": 2.2501,
      "step": 111540
    },
    {
      "epoch": 0.3911260645568289,
      "grad_norm": 0.5517969727516174,
      "learning_rate": 3.0451757613990817e-05,
      "loss": 2.2367,
      "step": 111600
    },
    {
      "epoch": 0.39133634738723583,
      "grad_norm": 0.5994294881820679,
      "learning_rate": 3.0441243472470475e-05,
      "loss": 2.2497,
      "step": 111660
    },
    {
      "epoch": 0.39154663021764274,
      "grad_norm": 0.5764245986938477,
      "learning_rate": 3.043072933095013e-05,
      "loss": 2.2423,
      "step": 111720
    },
    {
      "epoch": 0.3917569130480496,
      "grad_norm": 0.5494112968444824,
      "learning_rate": 3.0420215189429786e-05,
      "loss": 2.2354,
      "step": 111780
    },
    {
      "epoch": 0.3919671958784565,
      "grad_norm": 0.561445415019989,
      "learning_rate": 3.0409701047909438e-05,
      "loss": 2.24,
      "step": 111840
    },
    {
      "epoch": 0.39217747870886344,
      "grad_norm": 0.5689985156059265,
      "learning_rate": 3.0399186906389093e-05,
      "loss": 2.2409,
      "step": 111900
    },
    {
      "epoch": 0.3923877615392703,
      "grad_norm": 0.5241578817367554,
      "learning_rate": 3.038867276486875e-05,
      "loss": 2.2314,
      "step": 111960
    },
    {
      "epoch": 0.3925980443696772,
      "grad_norm": 0.5331136584281921,
      "learning_rate": 3.0378158623348407e-05,
      "loss": 2.2364,
      "step": 112020
    },
    {
      "epoch": 0.39280832720008413,
      "grad_norm": 0.5528892278671265,
      "learning_rate": 3.0367644481828062e-05,
      "loss": 2.2523,
      "step": 112080
    },
    {
      "epoch": 0.393018610030491,
      "grad_norm": 0.5552469491958618,
      "learning_rate": 3.0357130340307714e-05,
      "loss": 2.2384,
      "step": 112140
    },
    {
      "epoch": 0.3932288928608979,
      "grad_norm": 0.5582413673400879,
      "learning_rate": 3.034661619878737e-05,
      "loss": 2.2439,
      "step": 112200
    },
    {
      "epoch": 0.3934391756913048,
      "grad_norm": 0.4989684820175171,
      "learning_rate": 3.0336102057267025e-05,
      "loss": 2.2515,
      "step": 112260
    },
    {
      "epoch": 0.3936494585217117,
      "grad_norm": 0.574006974697113,
      "learning_rate": 3.0325587915746677e-05,
      "loss": 2.239,
      "step": 112320
    },
    {
      "epoch": 0.3938597413521186,
      "grad_norm": 0.5915408730506897,
      "learning_rate": 3.031507377422634e-05,
      "loss": 2.2411,
      "step": 112380
    },
    {
      "epoch": 0.3940700241825255,
      "grad_norm": 0.5183863043785095,
      "learning_rate": 3.030455963270599e-05,
      "loss": 2.2469,
      "step": 112440
    },
    {
      "epoch": 0.3942803070129324,
      "grad_norm": 0.5208684802055359,
      "learning_rate": 3.0294045491185646e-05,
      "loss": 2.2343,
      "step": 112500
    },
    {
      "epoch": 0.3944905898433393,
      "grad_norm": 0.6256120800971985,
      "learning_rate": 3.02835313496653e-05,
      "loss": 2.2467,
      "step": 112560
    },
    {
      "epoch": 0.3947008726737462,
      "grad_norm": 0.5246924161911011,
      "learning_rate": 3.0273017208144956e-05,
      "loss": 2.2413,
      "step": 112620
    },
    {
      "epoch": 0.39491115550415307,
      "grad_norm": 0.5094478130340576,
      "learning_rate": 3.0262503066624608e-05,
      "loss": 2.2345,
      "step": 112680
    },
    {
      "epoch": 0.39512143833456,
      "grad_norm": 0.5629433393478394,
      "learning_rate": 3.025198892510427e-05,
      "loss": 2.2398,
      "step": 112740
    },
    {
      "epoch": 0.3953317211649669,
      "grad_norm": 0.541553795337677,
      "learning_rate": 3.0241474783583922e-05,
      "loss": 2.2266,
      "step": 112800
    },
    {
      "epoch": 0.39554200399537376,
      "grad_norm": 0.5277448892593384,
      "learning_rate": 3.0230960642063578e-05,
      "loss": 2.2339,
      "step": 112860
    },
    {
      "epoch": 0.3957522868257807,
      "grad_norm": 0.6435659527778625,
      "learning_rate": 3.0220446500543233e-05,
      "loss": 2.2537,
      "step": 112920
    },
    {
      "epoch": 0.3959625696561876,
      "grad_norm": 0.5504876375198364,
      "learning_rate": 3.0209932359022885e-05,
      "loss": 2.2507,
      "step": 112980
    },
    {
      "epoch": 0.39617285248659445,
      "grad_norm": 0.5389472842216492,
      "learning_rate": 3.019941821750254e-05,
      "loss": 2.233,
      "step": 113040
    },
    {
      "epoch": 0.39638313531700137,
      "grad_norm": 0.5711050629615784,
      "learning_rate": 3.01889040759822e-05,
      "loss": 2.2264,
      "step": 113100
    },
    {
      "epoch": 0.3965934181474083,
      "grad_norm": 0.5223866105079651,
      "learning_rate": 3.0178389934461854e-05,
      "loss": 2.2274,
      "step": 113160
    },
    {
      "epoch": 0.39680370097781514,
      "grad_norm": 0.5716981291770935,
      "learning_rate": 3.016787579294151e-05,
      "loss": 2.2541,
      "step": 113220
    },
    {
      "epoch": 0.39701398380822206,
      "grad_norm": 0.6482357978820801,
      "learning_rate": 3.015736165142116e-05,
      "loss": 2.2481,
      "step": 113280
    },
    {
      "epoch": 0.397224266638629,
      "grad_norm": 0.5465265512466431,
      "learning_rate": 3.0146847509900816e-05,
      "loss": 2.2432,
      "step": 113340
    },
    {
      "epoch": 0.39743454946903584,
      "grad_norm": 0.5352566838264465,
      "learning_rate": 3.0136508604072477e-05,
      "loss": 2.2309,
      "step": 113400
    },
    {
      "epoch": 0.39764483229944275,
      "grad_norm": 0.5502161979675293,
      "learning_rate": 3.0125994462552132e-05,
      "loss": 2.2392,
      "step": 113460
    },
    {
      "epoch": 0.39785511512984967,
      "grad_norm": 0.536171555519104,
      "learning_rate": 3.011548032103179e-05,
      "loss": 2.2289,
      "step": 113520
    },
    {
      "epoch": 0.39806539796025653,
      "grad_norm": 0.6111114621162415,
      "learning_rate": 3.0104966179511446e-05,
      "loss": 2.2382,
      "step": 113580
    },
    {
      "epoch": 0.39827568079066344,
      "grad_norm": 0.610998272895813,
      "learning_rate": 3.00944520379911e-05,
      "loss": 2.2357,
      "step": 113640
    },
    {
      "epoch": 0.39848596362107036,
      "grad_norm": 0.5478922128677368,
      "learning_rate": 3.0083937896470753e-05,
      "loss": 2.2417,
      "step": 113700
    },
    {
      "epoch": 0.3986962464514772,
      "grad_norm": 0.515874445438385,
      "learning_rate": 3.007342375495041e-05,
      "loss": 2.2375,
      "step": 113760
    },
    {
      "epoch": 0.39890652928188414,
      "grad_norm": 0.5432803630828857,
      "learning_rate": 3.0062909613430064e-05,
      "loss": 2.2242,
      "step": 113820
    },
    {
      "epoch": 0.39911681211229105,
      "grad_norm": 0.6252090334892273,
      "learning_rate": 3.0052395471909723e-05,
      "loss": 2.2431,
      "step": 113880
    },
    {
      "epoch": 0.3993270949426979,
      "grad_norm": 0.5408844351768494,
      "learning_rate": 3.0041881330389378e-05,
      "loss": 2.2413,
      "step": 113940
    },
    {
      "epoch": 0.39953737777310483,
      "grad_norm": 0.5749211311340332,
      "learning_rate": 3.003136718886903e-05,
      "loss": 2.223,
      "step": 114000
    },
    {
      "epoch": 0.39974766060351175,
      "grad_norm": 0.568860650062561,
      "learning_rate": 3.0020853047348685e-05,
      "loss": 2.255,
      "step": 114060
    },
    {
      "epoch": 0.3999579434339186,
      "grad_norm": 0.5700159072875977,
      "learning_rate": 3.001033890582834e-05,
      "loss": 2.2504,
      "step": 114120
    },
    {
      "epoch": 0.4001682262643255,
      "grad_norm": 0.5364213585853577,
      "learning_rate": 2.9999824764307992e-05,
      "loss": 2.2327,
      "step": 114180
    },
    {
      "epoch": 0.40037850909473244,
      "grad_norm": 0.6105502247810364,
      "learning_rate": 2.9989310622787654e-05,
      "loss": 2.2264,
      "step": 114240
    },
    {
      "epoch": 0.4005887919251393,
      "grad_norm": 0.5763128399848938,
      "learning_rate": 2.9978796481267306e-05,
      "loss": 2.2373,
      "step": 114300
    },
    {
      "epoch": 0.4007990747555462,
      "grad_norm": 0.5186144113540649,
      "learning_rate": 2.996828233974696e-05,
      "loss": 2.2397,
      "step": 114360
    },
    {
      "epoch": 0.40100935758595313,
      "grad_norm": 0.5978050827980042,
      "learning_rate": 2.9957768198226617e-05,
      "loss": 2.2451,
      "step": 114420
    },
    {
      "epoch": 0.40121964041636,
      "grad_norm": 0.533627450466156,
      "learning_rate": 2.994725405670627e-05,
      "loss": 2.2421,
      "step": 114480
    },
    {
      "epoch": 0.4014299232467669,
      "grad_norm": 0.5692892670631409,
      "learning_rate": 2.9936739915185924e-05,
      "loss": 2.2384,
      "step": 114540
    },
    {
      "epoch": 0.4016402060771738,
      "grad_norm": 0.5549047589302063,
      "learning_rate": 2.9926225773665582e-05,
      "loss": 2.2379,
      "step": 114600
    },
    {
      "epoch": 0.4018504889075807,
      "grad_norm": 0.6541472673416138,
      "learning_rate": 2.9915711632145238e-05,
      "loss": 2.2458,
      "step": 114660
    },
    {
      "epoch": 0.4020607717379876,
      "grad_norm": 0.5438767671585083,
      "learning_rate": 2.9905197490624893e-05,
      "loss": 2.2562,
      "step": 114720
    },
    {
      "epoch": 0.4022710545683945,
      "grad_norm": 0.5015159249305725,
      "learning_rate": 2.9894683349104545e-05,
      "loss": 2.2245,
      "step": 114780
    },
    {
      "epoch": 0.4024813373988014,
      "grad_norm": 0.5856261849403381,
      "learning_rate": 2.98841692075842e-05,
      "loss": 2.241,
      "step": 114840
    },
    {
      "epoch": 0.4026916202292083,
      "grad_norm": 0.5348838567733765,
      "learning_rate": 2.9873655066063855e-05,
      "loss": 2.2259,
      "step": 114900
    },
    {
      "epoch": 0.4029019030596152,
      "grad_norm": 0.5396873950958252,
      "learning_rate": 2.9863140924543514e-05,
      "loss": 2.2413,
      "step": 114960
    },
    {
      "epoch": 0.40311218589002207,
      "grad_norm": 0.6850928068161011,
      "learning_rate": 2.985262678302317e-05,
      "loss": 2.2454,
      "step": 115020
    },
    {
      "epoch": 0.403322468720429,
      "grad_norm": 0.5155946016311646,
      "learning_rate": 2.984211264150282e-05,
      "loss": 2.2516,
      "step": 115080
    },
    {
      "epoch": 0.4035327515508359,
      "grad_norm": 0.5300692915916443,
      "learning_rate": 2.9831598499982477e-05,
      "loss": 2.242,
      "step": 115140
    },
    {
      "epoch": 0.40374303438124276,
      "grad_norm": 0.5764443874359131,
      "learning_rate": 2.9821084358462132e-05,
      "loss": 2.2258,
      "step": 115200
    },
    {
      "epoch": 0.4039533172116497,
      "grad_norm": 0.5413774251937866,
      "learning_rate": 2.9810570216941787e-05,
      "loss": 2.2469,
      "step": 115260
    },
    {
      "epoch": 0.4041636000420566,
      "grad_norm": 0.5148273706436157,
      "learning_rate": 2.9800056075421446e-05,
      "loss": 2.2292,
      "step": 115320
    },
    {
      "epoch": 0.40437388287246345,
      "grad_norm": 0.5263270735740662,
      "learning_rate": 2.97895419339011e-05,
      "loss": 2.2364,
      "step": 115380
    },
    {
      "epoch": 0.40458416570287037,
      "grad_norm": 0.5516897439956665,
      "learning_rate": 2.977920302807276e-05,
      "loss": 2.2345,
      "step": 115440
    },
    {
      "epoch": 0.4047944485332773,
      "grad_norm": 0.5485020279884338,
      "learning_rate": 2.9768688886552414e-05,
      "loss": 2.2291,
      "step": 115500
    },
    {
      "epoch": 0.40500473136368415,
      "grad_norm": 0.5054183602333069,
      "learning_rate": 2.975817474503207e-05,
      "loss": 2.2505,
      "step": 115560
    },
    {
      "epoch": 0.40521501419409106,
      "grad_norm": 0.5678335428237915,
      "learning_rate": 2.9747660603511724e-05,
      "loss": 2.2476,
      "step": 115620
    },
    {
      "epoch": 0.405425297024498,
      "grad_norm": 0.5526294112205505,
      "learning_rate": 2.9737146461991376e-05,
      "loss": 2.244,
      "step": 115680
    },
    {
      "epoch": 0.40563557985490484,
      "grad_norm": 0.5054203271865845,
      "learning_rate": 2.9726632320471038e-05,
      "loss": 2.2392,
      "step": 115740
    },
    {
      "epoch": 0.40584586268531175,
      "grad_norm": 0.5515086054801941,
      "learning_rate": 2.971611817895069e-05,
      "loss": 2.2366,
      "step": 115800
    },
    {
      "epoch": 0.4060561455157186,
      "grad_norm": 0.5307000875473022,
      "learning_rate": 2.9705604037430345e-05,
      "loss": 2.2323,
      "step": 115860
    },
    {
      "epoch": 0.40626642834612553,
      "grad_norm": 0.5576381087303162,
      "learning_rate": 2.969508989591e-05,
      "loss": 2.2366,
      "step": 115920
    },
    {
      "epoch": 0.40647671117653245,
      "grad_norm": 0.5790101885795593,
      "learning_rate": 2.9684575754389656e-05,
      "loss": 2.2342,
      "step": 115980
    },
    {
      "epoch": 0.4066869940069393,
      "grad_norm": 0.5370448231697083,
      "learning_rate": 2.9674061612869308e-05,
      "loss": 2.2266,
      "step": 116040
    },
    {
      "epoch": 0.4068972768373462,
      "grad_norm": 0.5727744102478027,
      "learning_rate": 2.966354747134897e-05,
      "loss": 2.2409,
      "step": 116100
    },
    {
      "epoch": 0.40710755966775314,
      "grad_norm": 0.5130985379219055,
      "learning_rate": 2.965303332982862e-05,
      "loss": 2.2479,
      "step": 116160
    },
    {
      "epoch": 0.40731784249816,
      "grad_norm": 0.5241026878356934,
      "learning_rate": 2.9642519188308277e-05,
      "loss": 2.2545,
      "step": 116220
    },
    {
      "epoch": 0.4075281253285669,
      "grad_norm": 0.49381840229034424,
      "learning_rate": 2.9632005046787932e-05,
      "loss": 2.2361,
      "step": 116280
    },
    {
      "epoch": 0.40773840815897383,
      "grad_norm": 0.5293487310409546,
      "learning_rate": 2.9621490905267584e-05,
      "loss": 2.2476,
      "step": 116340
    },
    {
      "epoch": 0.4079486909893807,
      "grad_norm": 0.5284563302993774,
      "learning_rate": 2.961097676374724e-05,
      "loss": 2.2249,
      "step": 116400
    },
    {
      "epoch": 0.4081589738197876,
      "grad_norm": 0.545966625213623,
      "learning_rate": 2.9600462622226898e-05,
      "loss": 2.2227,
      "step": 116460
    },
    {
      "epoch": 0.4083692566501945,
      "grad_norm": 0.574855625629425,
      "learning_rate": 2.9589948480706553e-05,
      "loss": 2.2341,
      "step": 116520
    },
    {
      "epoch": 0.4085795394806014,
      "grad_norm": 0.5358082056045532,
      "learning_rate": 2.957943433918621e-05,
      "loss": 2.2397,
      "step": 116580
    },
    {
      "epoch": 0.4087898223110083,
      "grad_norm": 0.5895236134529114,
      "learning_rate": 2.956892019766586e-05,
      "loss": 2.2342,
      "step": 116640
    },
    {
      "epoch": 0.4090001051414152,
      "grad_norm": 0.5704084634780884,
      "learning_rate": 2.9558406056145516e-05,
      "loss": 2.2417,
      "step": 116700
    },
    {
      "epoch": 0.4092103879718221,
      "grad_norm": 0.58211350440979,
      "learning_rate": 2.954789191462517e-05,
      "loss": 2.2371,
      "step": 116760
    },
    {
      "epoch": 0.409420670802229,
      "grad_norm": 0.532575786113739,
      "learning_rate": 2.953737777310483e-05,
      "loss": 2.239,
      "step": 116820
    },
    {
      "epoch": 0.4096309536326359,
      "grad_norm": 0.5325069427490234,
      "learning_rate": 2.9526863631584485e-05,
      "loss": 2.2369,
      "step": 116880
    },
    {
      "epoch": 0.40984123646304277,
      "grad_norm": 0.5889403223991394,
      "learning_rate": 2.9516349490064137e-05,
      "loss": 2.2481,
      "step": 116940
    },
    {
      "epoch": 0.4100515192934497,
      "grad_norm": 0.6322012543678284,
      "learning_rate": 2.9505835348543792e-05,
      "loss": 2.2413,
      "step": 117000
    },
    {
      "epoch": 0.4102618021238566,
      "grad_norm": 0.5821205377578735,
      "learning_rate": 2.9495321207023447e-05,
      "loss": 2.2249,
      "step": 117060
    },
    {
      "epoch": 0.41047208495426346,
      "grad_norm": 0.6500152349472046,
      "learning_rate": 2.94848070655031e-05,
      "loss": 2.2425,
      "step": 117120
    },
    {
      "epoch": 0.4106823677846704,
      "grad_norm": 0.5491619110107422,
      "learning_rate": 2.947429292398276e-05,
      "loss": 2.2478,
      "step": 117180
    },
    {
      "epoch": 0.4108926506150773,
      "grad_norm": 0.5800027251243591,
      "learning_rate": 2.9463778782462413e-05,
      "loss": 2.2458,
      "step": 117240
    },
    {
      "epoch": 0.41110293344548415,
      "grad_norm": 0.5748924612998962,
      "learning_rate": 2.945326464094207e-05,
      "loss": 2.2259,
      "step": 117300
    },
    {
      "epoch": 0.41131321627589107,
      "grad_norm": 0.5880306363105774,
      "learning_rate": 2.9442750499421724e-05,
      "loss": 2.2461,
      "step": 117360
    },
    {
      "epoch": 0.411523499106298,
      "grad_norm": 0.5313839912414551,
      "learning_rate": 2.9432236357901376e-05,
      "loss": 2.2411,
      "step": 117420
    },
    {
      "epoch": 0.41173378193670485,
      "grad_norm": 0.5506525635719299,
      "learning_rate": 2.942172221638103e-05,
      "loss": 2.2414,
      "step": 117480
    },
    {
      "epoch": 0.41194406476711176,
      "grad_norm": 0.5372014045715332,
      "learning_rate": 2.941138331055269e-05,
      "loss": 2.2413,
      "step": 117540
    },
    {
      "epoch": 0.4121543475975187,
      "grad_norm": 0.5816333889961243,
      "learning_rate": 2.9400869169032354e-05,
      "loss": 2.2294,
      "step": 117600
    },
    {
      "epoch": 0.41236463042792554,
      "grad_norm": 0.5727115273475647,
      "learning_rate": 2.9390355027512005e-05,
      "loss": 2.221,
      "step": 117660
    },
    {
      "epoch": 0.41257491325833245,
      "grad_norm": 0.5720790028572083,
      "learning_rate": 2.938001612168367e-05,
      "loss": 2.242,
      "step": 117720
    },
    {
      "epoch": 0.41278519608873937,
      "grad_norm": 0.5449382066726685,
      "learning_rate": 2.936950198016332e-05,
      "loss": 2.2247,
      "step": 117780
    },
    {
      "epoch": 0.41299547891914623,
      "grad_norm": 0.5297885537147522,
      "learning_rate": 2.9358987838642977e-05,
      "loss": 2.233,
      "step": 117840
    },
    {
      "epoch": 0.41320576174955315,
      "grad_norm": 0.5624762177467346,
      "learning_rate": 2.9348473697122632e-05,
      "loss": 2.2335,
      "step": 117900
    },
    {
      "epoch": 0.41341604457996006,
      "grad_norm": 0.5435742735862732,
      "learning_rate": 2.9337959555602284e-05,
      "loss": 2.2378,
      "step": 117960
    },
    {
      "epoch": 0.4136263274103669,
      "grad_norm": 0.5343254208564758,
      "learning_rate": 2.9327445414081946e-05,
      "loss": 2.2461,
      "step": 118020
    },
    {
      "epoch": 0.41383661024077384,
      "grad_norm": 0.5433058738708496,
      "learning_rate": 2.9316931272561598e-05,
      "loss": 2.2361,
      "step": 118080
    },
    {
      "epoch": 0.41404689307118075,
      "grad_norm": 0.5210029482841492,
      "learning_rate": 2.9306417131041253e-05,
      "loss": 2.2497,
      "step": 118140
    },
    {
      "epoch": 0.4142571759015876,
      "grad_norm": 0.5953100919723511,
      "learning_rate": 2.9295902989520908e-05,
      "loss": 2.2403,
      "step": 118200
    },
    {
      "epoch": 0.41446745873199453,
      "grad_norm": 0.523826003074646,
      "learning_rate": 2.928538884800056e-05,
      "loss": 2.2405,
      "step": 118260
    },
    {
      "epoch": 0.41467774156240145,
      "grad_norm": 0.5504089593887329,
      "learning_rate": 2.9274874706480215e-05,
      "loss": 2.2415,
      "step": 118320
    },
    {
      "epoch": 0.4148880243928083,
      "grad_norm": 0.5553892850875854,
      "learning_rate": 2.9264360564959874e-05,
      "loss": 2.2383,
      "step": 118380
    },
    {
      "epoch": 0.4150983072232152,
      "grad_norm": 0.5315834879875183,
      "learning_rate": 2.925384642343953e-05,
      "loss": 2.2172,
      "step": 118440
    },
    {
      "epoch": 0.41530859005362214,
      "grad_norm": 0.5641074776649475,
      "learning_rate": 2.9243332281919185e-05,
      "loss": 2.2516,
      "step": 118500
    },
    {
      "epoch": 0.415518872884029,
      "grad_norm": 0.513152539730072,
      "learning_rate": 2.9232818140398837e-05,
      "loss": 2.2416,
      "step": 118560
    },
    {
      "epoch": 0.4157291557144359,
      "grad_norm": 0.5532053112983704,
      "learning_rate": 2.9222303998878492e-05,
      "loss": 2.2424,
      "step": 118620
    },
    {
      "epoch": 0.41593943854484283,
      "grad_norm": 0.529827892780304,
      "learning_rate": 2.9211789857358147e-05,
      "loss": 2.2246,
      "step": 118680
    },
    {
      "epoch": 0.4161497213752497,
      "grad_norm": 0.5687451958656311,
      "learning_rate": 2.9201275715837806e-05,
      "loss": 2.2251,
      "step": 118740
    },
    {
      "epoch": 0.4163600042056566,
      "grad_norm": 0.5417942404747009,
      "learning_rate": 2.919076157431746e-05,
      "loss": 2.22,
      "step": 118800
    },
    {
      "epoch": 0.4165702870360635,
      "grad_norm": 0.5505846738815308,
      "learning_rate": 2.9180247432797113e-05,
      "loss": 2.2247,
      "step": 118860
    },
    {
      "epoch": 0.4167805698664704,
      "grad_norm": 0.5387886762619019,
      "learning_rate": 2.9169733291276768e-05,
      "loss": 2.2375,
      "step": 118920
    },
    {
      "epoch": 0.4169908526968773,
      "grad_norm": 0.5633653402328491,
      "learning_rate": 2.9159219149756423e-05,
      "loss": 2.2265,
      "step": 118980
    },
    {
      "epoch": 0.4172011355272842,
      "grad_norm": 0.581901490688324,
      "learning_rate": 2.9148705008236075e-05,
      "loss": 2.2427,
      "step": 119040
    },
    {
      "epoch": 0.4174114183576911,
      "grad_norm": 0.5256428122520447,
      "learning_rate": 2.9138190866715737e-05,
      "loss": 2.2371,
      "step": 119100
    },
    {
      "epoch": 0.417621701188098,
      "grad_norm": 0.516146719455719,
      "learning_rate": 2.912767672519539e-05,
      "loss": 2.2485,
      "step": 119160
    },
    {
      "epoch": 0.4178319840185049,
      "grad_norm": 0.5308356285095215,
      "learning_rate": 2.9117162583675045e-05,
      "loss": 2.2391,
      "step": 119220
    },
    {
      "epoch": 0.41804226684891177,
      "grad_norm": 0.5721306800842285,
      "learning_rate": 2.91066484421547e-05,
      "loss": 2.2283,
      "step": 119280
    },
    {
      "epoch": 0.4182525496793187,
      "grad_norm": 0.5151183605194092,
      "learning_rate": 2.9096134300634352e-05,
      "loss": 2.2278,
      "step": 119340
    },
    {
      "epoch": 0.4184628325097256,
      "grad_norm": 0.539119303226471,
      "learning_rate": 2.9085620159114007e-05,
      "loss": 2.2244,
      "step": 119400
    },
    {
      "epoch": 0.41867311534013246,
      "grad_norm": 0.5463353395462036,
      "learning_rate": 2.9075106017593666e-05,
      "loss": 2.2349,
      "step": 119460
    },
    {
      "epoch": 0.4188833981705394,
      "grad_norm": 0.6453050971031189,
      "learning_rate": 2.906459187607332e-05,
      "loss": 2.2309,
      "step": 119520
    },
    {
      "epoch": 0.4190936810009463,
      "grad_norm": 0.5602206587791443,
      "learning_rate": 2.9054077734552976e-05,
      "loss": 2.2323,
      "step": 119580
    },
    {
      "epoch": 0.41930396383135315,
      "grad_norm": 0.5998344421386719,
      "learning_rate": 2.904356359303263e-05,
      "loss": 2.2313,
      "step": 119640
    },
    {
      "epoch": 0.41951424666176007,
      "grad_norm": 0.5444914698600769,
      "learning_rate": 2.9033049451512283e-05,
      "loss": 2.2332,
      "step": 119700
    },
    {
      "epoch": 0.419724529492167,
      "grad_norm": 0.5113641023635864,
      "learning_rate": 2.902253530999194e-05,
      "loss": 2.2325,
      "step": 119760
    },
    {
      "epoch": 0.41993481232257385,
      "grad_norm": 0.5486562848091125,
      "learning_rate": 2.9012021168471597e-05,
      "loss": 2.2463,
      "step": 119820
    },
    {
      "epoch": 0.42014509515298076,
      "grad_norm": 0.5450092554092407,
      "learning_rate": 2.9001507026951253e-05,
      "loss": 2.2495,
      "step": 119880
    },
    {
      "epoch": 0.4203553779833877,
      "grad_norm": 0.6255912184715271,
      "learning_rate": 2.8990992885430908e-05,
      "loss": 2.2348,
      "step": 119940
    },
    {
      "epoch": 0.42056566081379454,
      "grad_norm": 0.5884232521057129,
      "learning_rate": 2.898047874391056e-05,
      "loss": 2.2301,
      "step": 120000
    },
    {
      "epoch": 0.42077594364420146,
      "grad_norm": 0.5232421159744263,
      "learning_rate": 2.8969964602390215e-05,
      "loss": 2.2267,
      "step": 120060
    },
    {
      "epoch": 0.42098622647460837,
      "grad_norm": 0.5445395112037659,
      "learning_rate": 2.895945046086987e-05,
      "loss": 2.2177,
      "step": 120120
    },
    {
      "epoch": 0.42119650930501523,
      "grad_norm": 0.5900552868843079,
      "learning_rate": 2.894893631934953e-05,
      "loss": 2.2466,
      "step": 120180
    },
    {
      "epoch": 0.42140679213542215,
      "grad_norm": 0.547518253326416,
      "learning_rate": 2.8938422177829184e-05,
      "loss": 2.2245,
      "step": 120240
    },
    {
      "epoch": 0.42161707496582906,
      "grad_norm": 0.6022685766220093,
      "learning_rate": 2.8927908036308836e-05,
      "loss": 2.2354,
      "step": 120300
    },
    {
      "epoch": 0.4218273577962359,
      "grad_norm": 0.5935965776443481,
      "learning_rate": 2.891739389478849e-05,
      "loss": 2.2352,
      "step": 120360
    },
    {
      "epoch": 0.42203764062664284,
      "grad_norm": 0.5695683360099792,
      "learning_rate": 2.8906879753268147e-05,
      "loss": 2.2308,
      "step": 120420
    },
    {
      "epoch": 0.42224792345704976,
      "grad_norm": 0.5888732075691223,
      "learning_rate": 2.88963656117478e-05,
      "loss": 2.2399,
      "step": 120480
    },
    {
      "epoch": 0.4224582062874566,
      "grad_norm": 0.5700323581695557,
      "learning_rate": 2.888585147022746e-05,
      "loss": 2.2396,
      "step": 120540
    },
    {
      "epoch": 0.42266848911786353,
      "grad_norm": 0.5631459951400757,
      "learning_rate": 2.8875337328707113e-05,
      "loss": 2.232,
      "step": 120600
    },
    {
      "epoch": 0.42287877194827045,
      "grad_norm": 0.5992614030838013,
      "learning_rate": 2.8864823187186768e-05,
      "loss": 2.2455,
      "step": 120660
    },
    {
      "epoch": 0.4230890547786773,
      "grad_norm": 0.531584620475769,
      "learning_rate": 2.8854309045666423e-05,
      "loss": 2.2286,
      "step": 120720
    },
    {
      "epoch": 0.4232993376090842,
      "grad_norm": 0.5879722237586975,
      "learning_rate": 2.8843794904146075e-05,
      "loss": 2.229,
      "step": 120780
    },
    {
      "epoch": 0.42350962043949114,
      "grad_norm": 0.5823795795440674,
      "learning_rate": 2.883328076262573e-05,
      "loss": 2.2367,
      "step": 120840
    },
    {
      "epoch": 0.423719903269898,
      "grad_norm": 0.5151508450508118,
      "learning_rate": 2.882276662110539e-05,
      "loss": 2.2285,
      "step": 120900
    },
    {
      "epoch": 0.4239301861003049,
      "grad_norm": 0.5895393490791321,
      "learning_rate": 2.8812252479585044e-05,
      "loss": 2.2495,
      "step": 120960
    },
    {
      "epoch": 0.42414046893071183,
      "grad_norm": 0.5240002274513245,
      "learning_rate": 2.88017383380647e-05,
      "loss": 2.2464,
      "step": 121020
    },
    {
      "epoch": 0.4243507517611187,
      "grad_norm": 0.560350239276886,
      "learning_rate": 2.879122419654435e-05,
      "loss": 2.2292,
      "step": 121080
    },
    {
      "epoch": 0.4245610345915256,
      "grad_norm": 0.5635443329811096,
      "learning_rate": 2.8780710055024007e-05,
      "loss": 2.2249,
      "step": 121140
    },
    {
      "epoch": 0.4247713174219325,
      "grad_norm": 0.525646448135376,
      "learning_rate": 2.8770195913503662e-05,
      "loss": 2.2328,
      "step": 121200
    },
    {
      "epoch": 0.4249816002523394,
      "grad_norm": 0.5428566336631775,
      "learning_rate": 2.875968177198332e-05,
      "loss": 2.2304,
      "step": 121260
    },
    {
      "epoch": 0.4251918830827463,
      "grad_norm": 0.5613175630569458,
      "learning_rate": 2.8749167630462976e-05,
      "loss": 2.2184,
      "step": 121320
    },
    {
      "epoch": 0.4254021659131532,
      "grad_norm": 0.544946551322937,
      "learning_rate": 2.8738653488942628e-05,
      "loss": 2.2286,
      "step": 121380
    },
    {
      "epoch": 0.4256124487435601,
      "grad_norm": 0.6552244424819946,
      "learning_rate": 2.8728139347422283e-05,
      "loss": 2.2402,
      "step": 121440
    },
    {
      "epoch": 0.425822731573967,
      "grad_norm": 0.5511813163757324,
      "learning_rate": 2.871762520590194e-05,
      "loss": 2.226,
      "step": 121500
    },
    {
      "epoch": 0.42603301440437386,
      "grad_norm": 0.5645471811294556,
      "learning_rate": 2.8707111064381594e-05,
      "loss": 2.2363,
      "step": 121560
    },
    {
      "epoch": 0.42624329723478077,
      "grad_norm": 0.5755658149719238,
      "learning_rate": 2.8696596922861252e-05,
      "loss": 2.2314,
      "step": 121620
    },
    {
      "epoch": 0.4264535800651877,
      "grad_norm": 0.5792255997657776,
      "learning_rate": 2.8686082781340908e-05,
      "loss": 2.2355,
      "step": 121680
    },
    {
      "epoch": 0.42666386289559455,
      "grad_norm": 0.5449994802474976,
      "learning_rate": 2.8675743875512568e-05,
      "loss": 2.2387,
      "step": 121740
    },
    {
      "epoch": 0.42687414572600146,
      "grad_norm": 0.5818429589271545,
      "learning_rate": 2.866522973399222e-05,
      "loss": 2.2386,
      "step": 121800
    },
    {
      "epoch": 0.4270844285564084,
      "grad_norm": 0.5970384478569031,
      "learning_rate": 2.8654715592471875e-05,
      "loss": 2.2309,
      "step": 121860
    },
    {
      "epoch": 0.42729471138681524,
      "grad_norm": 0.6303051114082336,
      "learning_rate": 2.864420145095153e-05,
      "loss": 2.2248,
      "step": 121920
    },
    {
      "epoch": 0.42750499421722216,
      "grad_norm": 0.5991727113723755,
      "learning_rate": 2.8633687309431183e-05,
      "loss": 2.2394,
      "step": 121980
    },
    {
      "epoch": 0.42771527704762907,
      "grad_norm": 0.5650383830070496,
      "learning_rate": 2.8623173167910845e-05,
      "loss": 2.2439,
      "step": 122040
    },
    {
      "epoch": 0.42792555987803593,
      "grad_norm": 0.547942578792572,
      "learning_rate": 2.8612659026390496e-05,
      "loss": 2.225,
      "step": 122100
    },
    {
      "epoch": 0.42813584270844285,
      "grad_norm": 0.5970585942268372,
      "learning_rate": 2.8602144884870152e-05,
      "loss": 2.2238,
      "step": 122160
    },
    {
      "epoch": 0.42834612553884976,
      "grad_norm": 0.540296733379364,
      "learning_rate": 2.8591630743349807e-05,
      "loss": 2.2304,
      "step": 122220
    },
    {
      "epoch": 0.4285564083692566,
      "grad_norm": 0.5640031695365906,
      "learning_rate": 2.8581116601829462e-05,
      "loss": 2.2392,
      "step": 122280
    },
    {
      "epoch": 0.42876669119966354,
      "grad_norm": 0.5584647059440613,
      "learning_rate": 2.8570602460309114e-05,
      "loss": 2.2253,
      "step": 122340
    },
    {
      "epoch": 0.42897697403007046,
      "grad_norm": 0.5100136995315552,
      "learning_rate": 2.8560088318788776e-05,
      "loss": 2.2381,
      "step": 122400
    },
    {
      "epoch": 0.4291872568604773,
      "grad_norm": 0.5283793807029724,
      "learning_rate": 2.8549574177268428e-05,
      "loss": 2.225,
      "step": 122460
    },
    {
      "epoch": 0.42939753969088423,
      "grad_norm": 0.5295258164405823,
      "learning_rate": 2.8539060035748083e-05,
      "loss": 2.23,
      "step": 122520
    },
    {
      "epoch": 0.42960782252129115,
      "grad_norm": 0.5125938057899475,
      "learning_rate": 2.852854589422774e-05,
      "loss": 2.2247,
      "step": 122580
    },
    {
      "epoch": 0.429818105351698,
      "grad_norm": 0.6334744095802307,
      "learning_rate": 2.851803175270739e-05,
      "loss": 2.233,
      "step": 122640
    },
    {
      "epoch": 0.4300283881821049,
      "grad_norm": 0.5219427943229675,
      "learning_rate": 2.8507517611187046e-05,
      "loss": 2.2167,
      "step": 122700
    },
    {
      "epoch": 0.43023867101251184,
      "grad_norm": 0.5502570867538452,
      "learning_rate": 2.8497003469666705e-05,
      "loss": 2.2263,
      "step": 122760
    },
    {
      "epoch": 0.4304489538429187,
      "grad_norm": 0.5381381511688232,
      "learning_rate": 2.848648932814636e-05,
      "loss": 2.2359,
      "step": 122820
    },
    {
      "epoch": 0.4306592366733256,
      "grad_norm": 0.5731381773948669,
      "learning_rate": 2.8475975186626015e-05,
      "loss": 2.2329,
      "step": 122880
    },
    {
      "epoch": 0.43086951950373253,
      "grad_norm": 0.5249903202056885,
      "learning_rate": 2.8465461045105667e-05,
      "loss": 2.2451,
      "step": 122940
    },
    {
      "epoch": 0.4310798023341394,
      "grad_norm": 0.517389178276062,
      "learning_rate": 2.8454946903585322e-05,
      "loss": 2.2341,
      "step": 123000
    },
    {
      "epoch": 0.4312900851645463,
      "grad_norm": 0.5333336591720581,
      "learning_rate": 2.8444432762064978e-05,
      "loss": 2.2407,
      "step": 123060
    },
    {
      "epoch": 0.4315003679949532,
      "grad_norm": 0.527782678604126,
      "learning_rate": 2.8433918620544636e-05,
      "loss": 2.2298,
      "step": 123120
    },
    {
      "epoch": 0.4317106508253601,
      "grad_norm": 0.5925357341766357,
      "learning_rate": 2.842340447902429e-05,
      "loss": 2.2361,
      "step": 123180
    },
    {
      "epoch": 0.431920933655767,
      "grad_norm": 0.5482273697853088,
      "learning_rate": 2.8412890337503943e-05,
      "loss": 2.2371,
      "step": 123240
    },
    {
      "epoch": 0.4321312164861739,
      "grad_norm": 0.5719531774520874,
      "learning_rate": 2.84023761959836e-05,
      "loss": 2.229,
      "step": 123300
    },
    {
      "epoch": 0.4323414993165808,
      "grad_norm": 0.6021667122840881,
      "learning_rate": 2.8391862054463254e-05,
      "loss": 2.2275,
      "step": 123360
    },
    {
      "epoch": 0.4325517821469877,
      "grad_norm": 0.5164570212364197,
      "learning_rate": 2.8381347912942906e-05,
      "loss": 2.2231,
      "step": 123420
    },
    {
      "epoch": 0.4327620649773946,
      "grad_norm": 0.6197656989097595,
      "learning_rate": 2.8370833771422568e-05,
      "loss": 2.2398,
      "step": 123480
    },
    {
      "epoch": 0.43297234780780147,
      "grad_norm": 0.5924232006072998,
      "learning_rate": 2.836031962990222e-05,
      "loss": 2.2379,
      "step": 123540
    },
    {
      "epoch": 0.4331826306382084,
      "grad_norm": 0.57561856508255,
      "learning_rate": 2.8349805488381875e-05,
      "loss": 2.244,
      "step": 123600
    },
    {
      "epoch": 0.4333929134686153,
      "grad_norm": 0.5562826991081238,
      "learning_rate": 2.833929134686153e-05,
      "loss": 2.2281,
      "step": 123660
    },
    {
      "epoch": 0.43360319629902216,
      "grad_norm": 0.5709443688392639,
      "learning_rate": 2.8328777205341182e-05,
      "loss": 2.2216,
      "step": 123720
    },
    {
      "epoch": 0.4338134791294291,
      "grad_norm": 0.6017040014266968,
      "learning_rate": 2.8318263063820838e-05,
      "loss": 2.2352,
      "step": 123780
    },
    {
      "epoch": 0.434023761959836,
      "grad_norm": 0.5744826793670654,
      "learning_rate": 2.8307924157992498e-05,
      "loss": 2.2266,
      "step": 123840
    },
    {
      "epoch": 0.43423404479024286,
      "grad_norm": 0.6745397448539734,
      "learning_rate": 2.829741001647216e-05,
      "loss": 2.2247,
      "step": 123900
    },
    {
      "epoch": 0.4344443276206498,
      "grad_norm": 0.5829909443855286,
      "learning_rate": 2.8286895874951812e-05,
      "loss": 2.2301,
      "step": 123960
    },
    {
      "epoch": 0.4346546104510567,
      "grad_norm": 0.5703389644622803,
      "learning_rate": 2.8276381733431467e-05,
      "loss": 2.2126,
      "step": 124020
    },
    {
      "epoch": 0.43486489328146355,
      "grad_norm": 0.5177561640739441,
      "learning_rate": 2.8265867591911123e-05,
      "loss": 2.2376,
      "step": 124080
    },
    {
      "epoch": 0.43507517611187047,
      "grad_norm": 0.5599654912948608,
      "learning_rate": 2.8255353450390774e-05,
      "loss": 2.2116,
      "step": 124140
    },
    {
      "epoch": 0.4352854589422774,
      "grad_norm": 0.5665854215621948,
      "learning_rate": 2.824483930887043e-05,
      "loss": 2.2437,
      "step": 124200
    },
    {
      "epoch": 0.43549574177268424,
      "grad_norm": 0.548435628414154,
      "learning_rate": 2.823432516735009e-05,
      "loss": 2.2466,
      "step": 124260
    },
    {
      "epoch": 0.43570602460309116,
      "grad_norm": 0.5328386425971985,
      "learning_rate": 2.8223811025829744e-05,
      "loss": 2.2386,
      "step": 124320
    },
    {
      "epoch": 0.4359163074334981,
      "grad_norm": 0.5652334094047546,
      "learning_rate": 2.82132968843094e-05,
      "loss": 2.2331,
      "step": 124380
    },
    {
      "epoch": 0.43612659026390493,
      "grad_norm": 0.5891857147216797,
      "learning_rate": 2.820278274278905e-05,
      "loss": 2.2185,
      "step": 124440
    },
    {
      "epoch": 0.43633687309431185,
      "grad_norm": 0.5438734889030457,
      "learning_rate": 2.8192268601268706e-05,
      "loss": 2.2226,
      "step": 124500
    },
    {
      "epoch": 0.43654715592471877,
      "grad_norm": 0.5453084111213684,
      "learning_rate": 2.818175445974836e-05,
      "loss": 2.2288,
      "step": 124560
    },
    {
      "epoch": 0.4367574387551256,
      "grad_norm": 0.5740223526954651,
      "learning_rate": 2.817124031822802e-05,
      "loss": 2.2327,
      "step": 124620
    },
    {
      "epoch": 0.43696772158553254,
      "grad_norm": 0.5052223801612854,
      "learning_rate": 2.8160726176707675e-05,
      "loss": 2.2366,
      "step": 124680
    },
    {
      "epoch": 0.43717800441593946,
      "grad_norm": 0.5633878707885742,
      "learning_rate": 2.8150212035187327e-05,
      "loss": 2.2296,
      "step": 124740
    },
    {
      "epoch": 0.4373882872463463,
      "grad_norm": 0.5381173491477966,
      "learning_rate": 2.8139697893666983e-05,
      "loss": 2.2313,
      "step": 124800
    },
    {
      "epoch": 0.43759857007675323,
      "grad_norm": 0.5788121223449707,
      "learning_rate": 2.8129183752146638e-05,
      "loss": 2.2267,
      "step": 124860
    },
    {
      "epoch": 0.43780885290716015,
      "grad_norm": 0.5180872678756714,
      "learning_rate": 2.8118669610626293e-05,
      "loss": 2.2213,
      "step": 124920
    },
    {
      "epoch": 0.438019135737567,
      "grad_norm": 0.5612934827804565,
      "learning_rate": 2.8108155469105952e-05,
      "loss": 2.2218,
      "step": 124980
    },
    {
      "epoch": 0.4382294185679739,
      "grad_norm": 0.7243056893348694,
      "learning_rate": 2.8097641327585607e-05,
      "loss": 2.2406,
      "step": 125040
    },
    {
      "epoch": 0.43843970139838084,
      "grad_norm": 0.6163976788520813,
      "learning_rate": 2.808712718606526e-05,
      "loss": 2.2208,
      "step": 125100
    },
    {
      "epoch": 0.4386499842287877,
      "grad_norm": 0.57260662317276,
      "learning_rate": 2.8076613044544914e-05,
      "loss": 2.2152,
      "step": 125160
    },
    {
      "epoch": 0.4388602670591946,
      "grad_norm": 0.5954247713088989,
      "learning_rate": 2.806609890302457e-05,
      "loss": 2.2249,
      "step": 125220
    },
    {
      "epoch": 0.43907054988960154,
      "grad_norm": 0.48894116282463074,
      "learning_rate": 2.805558476150422e-05,
      "loss": 2.2506,
      "step": 125280
    },
    {
      "epoch": 0.4392808327200084,
      "grad_norm": 0.5188723802566528,
      "learning_rate": 2.8045070619983883e-05,
      "loss": 2.2224,
      "step": 125340
    },
    {
      "epoch": 0.4394911155504153,
      "grad_norm": 0.5828328728675842,
      "learning_rate": 2.8034556478463535e-05,
      "loss": 2.2445,
      "step": 125400
    },
    {
      "epoch": 0.43970139838082223,
      "grad_norm": 0.521661102771759,
      "learning_rate": 2.802404233694319e-05,
      "loss": 2.2302,
      "step": 125460
    },
    {
      "epoch": 0.4399116812112291,
      "grad_norm": 0.5403749942779541,
      "learning_rate": 2.8013528195422846e-05,
      "loss": 2.2135,
      "step": 125520
    },
    {
      "epoch": 0.440121964041636,
      "grad_norm": 0.534483015537262,
      "learning_rate": 2.8003014053902498e-05,
      "loss": 2.212,
      "step": 125580
    },
    {
      "epoch": 0.4403322468720429,
      "grad_norm": 0.6184211373329163,
      "learning_rate": 2.7992499912382153e-05,
      "loss": 2.2396,
      "step": 125640
    },
    {
      "epoch": 0.4405425297024498,
      "grad_norm": 0.5149890780448914,
      "learning_rate": 2.7981985770861812e-05,
      "loss": 2.2368,
      "step": 125700
    },
    {
      "epoch": 0.4407528125328567,
      "grad_norm": 0.5126898884773254,
      "learning_rate": 2.7971471629341467e-05,
      "loss": 2.2277,
      "step": 125760
    },
    {
      "epoch": 0.4409630953632636,
      "grad_norm": 0.5691506862640381,
      "learning_rate": 2.7960957487821122e-05,
      "loss": 2.2398,
      "step": 125820
    },
    {
      "epoch": 0.4411733781936705,
      "grad_norm": 0.5742629170417786,
      "learning_rate": 2.7950443346300774e-05,
      "loss": 2.2313,
      "step": 125880
    },
    {
      "epoch": 0.4413836610240774,
      "grad_norm": 0.5491039156913757,
      "learning_rate": 2.793992920478043e-05,
      "loss": 2.2295,
      "step": 125940
    },
    {
      "epoch": 0.4415939438544843,
      "grad_norm": 0.5712140798568726,
      "learning_rate": 2.792959029895209e-05,
      "loss": 2.2311,
      "step": 126000
    },
    {
      "epoch": 0.44180422668489117,
      "grad_norm": 0.5529006123542786,
      "learning_rate": 2.7919076157431745e-05,
      "loss": 2.2285,
      "step": 126060
    },
    {
      "epoch": 0.4420145095152981,
      "grad_norm": 0.5630046129226685,
      "learning_rate": 2.7908562015911404e-05,
      "loss": 2.2245,
      "step": 126120
    },
    {
      "epoch": 0.442224792345705,
      "grad_norm": 0.6187325119972229,
      "learning_rate": 2.789804787439106e-05,
      "loss": 2.2175,
      "step": 126180
    },
    {
      "epoch": 0.44243507517611186,
      "grad_norm": 0.508385419845581,
      "learning_rate": 2.7887533732870714e-05,
      "loss": 2.2245,
      "step": 126240
    },
    {
      "epoch": 0.4426453580065188,
      "grad_norm": 0.5613511800765991,
      "learning_rate": 2.7877019591350366e-05,
      "loss": 2.2226,
      "step": 126300
    },
    {
      "epoch": 0.4428556408369257,
      "grad_norm": 0.5883780121803284,
      "learning_rate": 2.786650544983002e-05,
      "loss": 2.2158,
      "step": 126360
    },
    {
      "epoch": 0.44306592366733255,
      "grad_norm": 0.6000725626945496,
      "learning_rate": 2.7855991308309677e-05,
      "loss": 2.2262,
      "step": 126420
    },
    {
      "epoch": 0.44327620649773947,
      "grad_norm": 0.5127400755882263,
      "learning_rate": 2.7845477166789336e-05,
      "loss": 2.2276,
      "step": 126480
    },
    {
      "epoch": 0.4434864893281464,
      "grad_norm": 0.5251140594482422,
      "learning_rate": 2.783496302526899e-05,
      "loss": 2.2197,
      "step": 126540
    },
    {
      "epoch": 0.44369677215855324,
      "grad_norm": 0.5944713950157166,
      "learning_rate": 2.7824448883748643e-05,
      "loss": 2.2275,
      "step": 126600
    },
    {
      "epoch": 0.44390705498896016,
      "grad_norm": 0.5731028318405151,
      "learning_rate": 2.7813934742228298e-05,
      "loss": 2.235,
      "step": 126660
    },
    {
      "epoch": 0.4441173378193671,
      "grad_norm": 0.5536123514175415,
      "learning_rate": 2.7803420600707953e-05,
      "loss": 2.2396,
      "step": 126720
    },
    {
      "epoch": 0.44432762064977394,
      "grad_norm": 0.5599493384361267,
      "learning_rate": 2.7792906459187605e-05,
      "loss": 2.226,
      "step": 126780
    },
    {
      "epoch": 0.44453790348018085,
      "grad_norm": 0.548050045967102,
      "learning_rate": 2.7782392317667267e-05,
      "loss": 2.2504,
      "step": 126840
    },
    {
      "epoch": 0.44474818631058777,
      "grad_norm": 0.5116492509841919,
      "learning_rate": 2.777187817614692e-05,
      "loss": 2.2346,
      "step": 126900
    },
    {
      "epoch": 0.44495846914099463,
      "grad_norm": 0.5121557712554932,
      "learning_rate": 2.7761364034626574e-05,
      "loss": 2.2365,
      "step": 126960
    },
    {
      "epoch": 0.44516875197140154,
      "grad_norm": 0.5654640793800354,
      "learning_rate": 2.775084989310623e-05,
      "loss": 2.2114,
      "step": 127020
    },
    {
      "epoch": 0.44537903480180846,
      "grad_norm": 0.5340287089347839,
      "learning_rate": 2.774033575158588e-05,
      "loss": 2.2296,
      "step": 127080
    },
    {
      "epoch": 0.4455893176322153,
      "grad_norm": 0.5115010738372803,
      "learning_rate": 2.7729821610065537e-05,
      "loss": 2.2265,
      "step": 127140
    },
    {
      "epoch": 0.44579960046262224,
      "grad_norm": 0.5538489818572998,
      "learning_rate": 2.7719307468545196e-05,
      "loss": 2.2066,
      "step": 127200
    },
    {
      "epoch": 0.4460098832930291,
      "grad_norm": 0.5297344923019409,
      "learning_rate": 2.770879332702485e-05,
      "loss": 2.218,
      "step": 127260
    },
    {
      "epoch": 0.446220166123436,
      "grad_norm": 0.5513839721679688,
      "learning_rate": 2.7698279185504506e-05,
      "loss": 2.2197,
      "step": 127320
    },
    {
      "epoch": 0.44643044895384293,
      "grad_norm": 0.5162503719329834,
      "learning_rate": 2.7687765043984158e-05,
      "loss": 2.2308,
      "step": 127380
    },
    {
      "epoch": 0.4466407317842498,
      "grad_norm": 0.6151419878005981,
      "learning_rate": 2.7677250902463813e-05,
      "loss": 2.2455,
      "step": 127440
    },
    {
      "epoch": 0.4468510146146567,
      "grad_norm": 0.6089375615119934,
      "learning_rate": 2.766673676094347e-05,
      "loss": 2.2147,
      "step": 127500
    },
    {
      "epoch": 0.4470612974450636,
      "grad_norm": 0.5278417468070984,
      "learning_rate": 2.7656222619423127e-05,
      "loss": 2.2258,
      "step": 127560
    },
    {
      "epoch": 0.4472715802754705,
      "grad_norm": 0.5466154217720032,
      "learning_rate": 2.7645708477902783e-05,
      "loss": 2.2321,
      "step": 127620
    },
    {
      "epoch": 0.4474818631058774,
      "grad_norm": 0.5713569521903992,
      "learning_rate": 2.7635194336382438e-05,
      "loss": 2.2191,
      "step": 127680
    },
    {
      "epoch": 0.4476921459362843,
      "grad_norm": 0.560494601726532,
      "learning_rate": 2.762468019486209e-05,
      "loss": 2.2269,
      "step": 127740
    },
    {
      "epoch": 0.4479024287666912,
      "grad_norm": 0.48673978447914124,
      "learning_rate": 2.7614166053341745e-05,
      "loss": 2.2386,
      "step": 127800
    },
    {
      "epoch": 0.4481127115970981,
      "grad_norm": 0.5395427346229553,
      "learning_rate": 2.76036519118214e-05,
      "loss": 2.2298,
      "step": 127860
    },
    {
      "epoch": 0.448322994427505,
      "grad_norm": 0.5101057887077332,
      "learning_rate": 2.759313777030106e-05,
      "loss": 2.2421,
      "step": 127920
    },
    {
      "epoch": 0.44853327725791187,
      "grad_norm": 0.5033383369445801,
      "learning_rate": 2.758279886447272e-05,
      "loss": 2.2284,
      "step": 127980
    },
    {
      "epoch": 0.4487435600883188,
      "grad_norm": 0.5280095934867859,
      "learning_rate": 2.7572284722952375e-05,
      "loss": 2.2458,
      "step": 128040
    },
    {
      "epoch": 0.4489538429187257,
      "grad_norm": 0.5307741761207581,
      "learning_rate": 2.7561770581432027e-05,
      "loss": 2.2245,
      "step": 128100
    },
    {
      "epoch": 0.44916412574913256,
      "grad_norm": 0.5375804305076599,
      "learning_rate": 2.7551256439911682e-05,
      "loss": 2.2116,
      "step": 128160
    },
    {
      "epoch": 0.4493744085795395,
      "grad_norm": 0.5175926685333252,
      "learning_rate": 2.7540742298391337e-05,
      "loss": 2.2212,
      "step": 128220
    },
    {
      "epoch": 0.4495846914099464,
      "grad_norm": 0.5948784947395325,
      "learning_rate": 2.753022815687099e-05,
      "loss": 2.2265,
      "step": 128280
    },
    {
      "epoch": 0.44979497424035325,
      "grad_norm": 0.5095413327217102,
      "learning_rate": 2.751971401535065e-05,
      "loss": 2.227,
      "step": 128340
    },
    {
      "epoch": 0.45000525707076017,
      "grad_norm": 0.5439273118972778,
      "learning_rate": 2.7509199873830303e-05,
      "loss": 2.2472,
      "step": 128400
    },
    {
      "epoch": 0.4502155399011671,
      "grad_norm": 0.5273983478546143,
      "learning_rate": 2.749868573230996e-05,
      "loss": 2.2369,
      "step": 128460
    },
    {
      "epoch": 0.45042582273157394,
      "grad_norm": 0.5817850828170776,
      "learning_rate": 2.7488171590789614e-05,
      "loss": 2.229,
      "step": 128520
    },
    {
      "epoch": 0.45063610556198086,
      "grad_norm": 0.5278428792953491,
      "learning_rate": 2.747765744926927e-05,
      "loss": 2.2221,
      "step": 128580
    },
    {
      "epoch": 0.4508463883923878,
      "grad_norm": 0.529971718788147,
      "learning_rate": 2.746714330774892e-05,
      "loss": 2.2413,
      "step": 128640
    },
    {
      "epoch": 0.45105667122279464,
      "grad_norm": 0.5319401621818542,
      "learning_rate": 2.7456629166228583e-05,
      "loss": 2.2259,
      "step": 128700
    },
    {
      "epoch": 0.45126695405320155,
      "grad_norm": 0.5638734698295593,
      "learning_rate": 2.7446115024708235e-05,
      "loss": 2.2072,
      "step": 128760
    },
    {
      "epoch": 0.45147723688360847,
      "grad_norm": 0.567353367805481,
      "learning_rate": 2.743560088318789e-05,
      "loss": 2.206,
      "step": 128820
    },
    {
      "epoch": 0.45168751971401533,
      "grad_norm": 0.5370661616325378,
      "learning_rate": 2.7425086741667545e-05,
      "loss": 2.2205,
      "step": 128880
    },
    {
      "epoch": 0.45189780254442224,
      "grad_norm": 0.5796499848365784,
      "learning_rate": 2.7414572600147197e-05,
      "loss": 2.2259,
      "step": 128940
    },
    {
      "epoch": 0.45210808537482916,
      "grad_norm": 0.5625753402709961,
      "learning_rate": 2.7404058458626852e-05,
      "loss": 2.2237,
      "step": 129000
    },
    {
      "epoch": 0.452318368205236,
      "grad_norm": 0.5433998703956604,
      "learning_rate": 2.739354431710651e-05,
      "loss": 2.2162,
      "step": 129060
    },
    {
      "epoch": 0.45252865103564294,
      "grad_norm": 0.5345425009727478,
      "learning_rate": 2.7383030175586166e-05,
      "loss": 2.2191,
      "step": 129120
    },
    {
      "epoch": 0.45273893386604985,
      "grad_norm": 0.492722749710083,
      "learning_rate": 2.737251603406582e-05,
      "loss": 2.2256,
      "step": 129180
    },
    {
      "epoch": 0.4529492166964567,
      "grad_norm": 0.5437445640563965,
      "learning_rate": 2.7362001892545474e-05,
      "loss": 2.2252,
      "step": 129240
    },
    {
      "epoch": 0.45315949952686363,
      "grad_norm": 0.5385192036628723,
      "learning_rate": 2.735148775102513e-05,
      "loss": 2.2337,
      "step": 129300
    },
    {
      "epoch": 0.45336978235727055,
      "grad_norm": 0.5326671004295349,
      "learning_rate": 2.7340973609504784e-05,
      "loss": 2.2292,
      "step": 129360
    },
    {
      "epoch": 0.4535800651876774,
      "grad_norm": 0.582104504108429,
      "learning_rate": 2.7330459467984443e-05,
      "loss": 2.225,
      "step": 129420
    },
    {
      "epoch": 0.4537903480180843,
      "grad_norm": 0.5414701700210571,
      "learning_rate": 2.7319945326464098e-05,
      "loss": 2.2438,
      "step": 129480
    },
    {
      "epoch": 0.45400063084849124,
      "grad_norm": 0.5443952679634094,
      "learning_rate": 2.730943118494375e-05,
      "loss": 2.2414,
      "step": 129540
    },
    {
      "epoch": 0.4542109136788981,
      "grad_norm": 0.49736344814300537,
      "learning_rate": 2.7298917043423405e-05,
      "loss": 2.2184,
      "step": 129600
    },
    {
      "epoch": 0.454421196509305,
      "grad_norm": 0.5797542929649353,
      "learning_rate": 2.728840290190306e-05,
      "loss": 2.224,
      "step": 129660
    },
    {
      "epoch": 0.45463147933971193,
      "grad_norm": 0.5145884156227112,
      "learning_rate": 2.7277888760382712e-05,
      "loss": 2.2343,
      "step": 129720
    },
    {
      "epoch": 0.4548417621701188,
      "grad_norm": 0.5669538378715515,
      "learning_rate": 2.7267374618862374e-05,
      "loss": 2.2406,
      "step": 129780
    },
    {
      "epoch": 0.4550520450005257,
      "grad_norm": 0.531952977180481,
      "learning_rate": 2.7256860477342026e-05,
      "loss": 2.2407,
      "step": 129840
    },
    {
      "epoch": 0.4552623278309326,
      "grad_norm": 0.5253896117210388,
      "learning_rate": 2.724634633582168e-05,
      "loss": 2.2334,
      "step": 129900
    },
    {
      "epoch": 0.4554726106613395,
      "grad_norm": 0.5913567543029785,
      "learning_rate": 2.7235832194301337e-05,
      "loss": 2.2157,
      "step": 129960
    },
    {
      "epoch": 0.4556828934917464,
      "grad_norm": 0.5832261443138123,
      "learning_rate": 2.722531805278099e-05,
      "loss": 2.2168,
      "step": 130020
    },
    {
      "epoch": 0.4558931763221533,
      "grad_norm": 0.528374195098877,
      "learning_rate": 2.7214803911260644e-05,
      "loss": 2.2204,
      "step": 130080
    },
    {
      "epoch": 0.4561034591525602,
      "grad_norm": 0.5559998154640198,
      "learning_rate": 2.7204465005432305e-05,
      "loss": 2.2149,
      "step": 130140
    },
    {
      "epoch": 0.4563137419829671,
      "grad_norm": 0.5101924538612366,
      "learning_rate": 2.7193950863911967e-05,
      "loss": 2.2408,
      "step": 130200
    },
    {
      "epoch": 0.456524024813374,
      "grad_norm": 0.5665587186813354,
      "learning_rate": 2.718343672239162e-05,
      "loss": 2.2305,
      "step": 130260
    },
    {
      "epoch": 0.45673430764378087,
      "grad_norm": 0.601974606513977,
      "learning_rate": 2.7172922580871274e-05,
      "loss": 2.2291,
      "step": 130320
    },
    {
      "epoch": 0.4569445904741878,
      "grad_norm": 0.4964762032032013,
      "learning_rate": 2.716240843935093e-05,
      "loss": 2.2179,
      "step": 130380
    },
    {
      "epoch": 0.4571548733045947,
      "grad_norm": 0.5069715976715088,
      "learning_rate": 2.715189429783058e-05,
      "loss": 2.2339,
      "step": 130440
    },
    {
      "epoch": 0.45736515613500156,
      "grad_norm": 0.560584545135498,
      "learning_rate": 2.7141380156310236e-05,
      "loss": 2.2287,
      "step": 130500
    },
    {
      "epoch": 0.4575754389654085,
      "grad_norm": 0.5845340490341187,
      "learning_rate": 2.7130866014789895e-05,
      "loss": 2.2249,
      "step": 130560
    },
    {
      "epoch": 0.4577857217958154,
      "grad_norm": 0.5387807488441467,
      "learning_rate": 2.712035187326955e-05,
      "loss": 2.226,
      "step": 130620
    },
    {
      "epoch": 0.45799600462622225,
      "grad_norm": 0.5556718111038208,
      "learning_rate": 2.7109837731749206e-05,
      "loss": 2.2328,
      "step": 130680
    },
    {
      "epoch": 0.45820628745662917,
      "grad_norm": 0.5116903185844421,
      "learning_rate": 2.7099323590228857e-05,
      "loss": 2.228,
      "step": 130740
    },
    {
      "epoch": 0.4584165702870361,
      "grad_norm": 0.5489668846130371,
      "learning_rate": 2.7088809448708513e-05,
      "loss": 2.2253,
      "step": 130800
    },
    {
      "epoch": 0.45862685311744295,
      "grad_norm": 0.49988293647766113,
      "learning_rate": 2.7078295307188168e-05,
      "loss": 2.2123,
      "step": 130860
    },
    {
      "epoch": 0.45883713594784986,
      "grad_norm": 0.5553171634674072,
      "learning_rate": 2.706795640135983e-05,
      "loss": 2.2295,
      "step": 130920
    },
    {
      "epoch": 0.4590474187782568,
      "grad_norm": 0.5072939991950989,
      "learning_rate": 2.7057442259839484e-05,
      "loss": 2.2345,
      "step": 130980
    },
    {
      "epoch": 0.45925770160866364,
      "grad_norm": 0.5555667877197266,
      "learning_rate": 2.7046928118319142e-05,
      "loss": 2.2448,
      "step": 131040
    },
    {
      "epoch": 0.45946798443907055,
      "grad_norm": 0.5642482042312622,
      "learning_rate": 2.7036413976798798e-05,
      "loss": 2.2164,
      "step": 131100
    },
    {
      "epoch": 0.45967826726947747,
      "grad_norm": 0.651848316192627,
      "learning_rate": 2.702589983527845e-05,
      "loss": 2.2258,
      "step": 131160
    },
    {
      "epoch": 0.45988855009988433,
      "grad_norm": 0.6740097403526306,
      "learning_rate": 2.7015385693758105e-05,
      "loss": 2.2263,
      "step": 131220
    },
    {
      "epoch": 0.46009883293029125,
      "grad_norm": 0.5537322759628296,
      "learning_rate": 2.700487155223776e-05,
      "loss": 2.2224,
      "step": 131280
    },
    {
      "epoch": 0.46030911576069816,
      "grad_norm": 0.5565202236175537,
      "learning_rate": 2.6994357410717412e-05,
      "loss": 2.2227,
      "step": 131340
    },
    {
      "epoch": 0.460519398591105,
      "grad_norm": 0.5631937384605408,
      "learning_rate": 2.6983843269197074e-05,
      "loss": 2.2206,
      "step": 131400
    },
    {
      "epoch": 0.46072968142151194,
      "grad_norm": 0.5715699195861816,
      "learning_rate": 2.6973329127676726e-05,
      "loss": 2.2145,
      "step": 131460
    },
    {
      "epoch": 0.46093996425191885,
      "grad_norm": 0.5436623096466064,
      "learning_rate": 2.696281498615638e-05,
      "loss": 2.2266,
      "step": 131520
    },
    {
      "epoch": 0.4611502470823257,
      "grad_norm": 0.5260350108146667,
      "learning_rate": 2.6952300844636037e-05,
      "loss": 2.2117,
      "step": 131580
    },
    {
      "epoch": 0.46136052991273263,
      "grad_norm": 0.49663183093070984,
      "learning_rate": 2.694178670311569e-05,
      "loss": 2.227,
      "step": 131640
    },
    {
      "epoch": 0.46157081274313955,
      "grad_norm": 0.5875915288925171,
      "learning_rate": 2.6931272561595344e-05,
      "loss": 2.2093,
      "step": 131700
    },
    {
      "epoch": 0.4617810955735464,
      "grad_norm": 0.5310595631599426,
      "learning_rate": 2.6920758420075002e-05,
      "loss": 2.227,
      "step": 131760
    },
    {
      "epoch": 0.4619913784039533,
      "grad_norm": 0.5796633958816528,
      "learning_rate": 2.6910244278554658e-05,
      "loss": 2.2256,
      "step": 131820
    },
    {
      "epoch": 0.46220166123436024,
      "grad_norm": 0.5720958113670349,
      "learning_rate": 2.6899730137034313e-05,
      "loss": 2.222,
      "step": 131880
    },
    {
      "epoch": 0.4624119440647671,
      "grad_norm": 0.5652265548706055,
      "learning_rate": 2.6889215995513968e-05,
      "loss": 2.2221,
      "step": 131940
    },
    {
      "epoch": 0.462622226895174,
      "grad_norm": 0.5959270000457764,
      "learning_rate": 2.687870185399362e-05,
      "loss": 2.2209,
      "step": 132000
    },
    {
      "epoch": 0.46283250972558093,
      "grad_norm": 0.573942244052887,
      "learning_rate": 2.6868187712473282e-05,
      "loss": 2.2271,
      "step": 132060
    },
    {
      "epoch": 0.4630427925559878,
      "grad_norm": 0.5455266237258911,
      "learning_rate": 2.6857673570952934e-05,
      "loss": 2.2122,
      "step": 132120
    },
    {
      "epoch": 0.4632530753863947,
      "grad_norm": 0.5603702664375305,
      "learning_rate": 2.684715942943259e-05,
      "loss": 2.2281,
      "step": 132180
    },
    {
      "epoch": 0.4634633582168016,
      "grad_norm": 0.548740565776825,
      "learning_rate": 2.6836645287912245e-05,
      "loss": 2.2262,
      "step": 132240
    },
    {
      "epoch": 0.4636736410472085,
      "grad_norm": 0.5343613624572754,
      "learning_rate": 2.6826131146391897e-05,
      "loss": 2.2154,
      "step": 132300
    },
    {
      "epoch": 0.4638839238776154,
      "grad_norm": 0.5475893020629883,
      "learning_rate": 2.6815617004871552e-05,
      "loss": 2.2279,
      "step": 132360
    },
    {
      "epoch": 0.4640942067080223,
      "grad_norm": 0.522315502166748,
      "learning_rate": 2.680510286335121e-05,
      "loss": 2.2222,
      "step": 132420
    },
    {
      "epoch": 0.4643044895384292,
      "grad_norm": 0.5225598812103271,
      "learning_rate": 2.6794588721830866e-05,
      "loss": 2.2285,
      "step": 132480
    },
    {
      "epoch": 0.4645147723688361,
      "grad_norm": 0.5249638557434082,
      "learning_rate": 2.678407458031052e-05,
      "loss": 2.2349,
      "step": 132540
    },
    {
      "epoch": 0.464725055199243,
      "grad_norm": 0.5643758773803711,
      "learning_rate": 2.6773560438790173e-05,
      "loss": 2.2221,
      "step": 132600
    },
    {
      "epoch": 0.46493533802964987,
      "grad_norm": 0.4918893873691559,
      "learning_rate": 2.6763046297269828e-05,
      "loss": 2.2187,
      "step": 132660
    },
    {
      "epoch": 0.4651456208600568,
      "grad_norm": 0.5675564408302307,
      "learning_rate": 2.6752532155749483e-05,
      "loss": 2.2358,
      "step": 132720
    },
    {
      "epoch": 0.46535590369046365,
      "grad_norm": 0.5786041021347046,
      "learning_rate": 2.6742018014229142e-05,
      "loss": 2.2305,
      "step": 132780
    },
    {
      "epoch": 0.46556618652087056,
      "grad_norm": 0.6326382756233215,
      "learning_rate": 2.6731503872708797e-05,
      "loss": 2.2225,
      "step": 132840
    },
    {
      "epoch": 0.4657764693512775,
      "grad_norm": 0.5252036452293396,
      "learning_rate": 2.672098973118845e-05,
      "loss": 2.2237,
      "step": 132900
    },
    {
      "epoch": 0.46598675218168434,
      "grad_norm": 0.5439803004264832,
      "learning_rate": 2.6710475589668105e-05,
      "loss": 2.2233,
      "step": 132960
    },
    {
      "epoch": 0.46619703501209125,
      "grad_norm": 0.5557428598403931,
      "learning_rate": 2.669996144814776e-05,
      "loss": 2.2139,
      "step": 133020
    },
    {
      "epoch": 0.46640731784249817,
      "grad_norm": 0.5631593465805054,
      "learning_rate": 2.6689447306627412e-05,
      "loss": 2.2353,
      "step": 133080
    },
    {
      "epoch": 0.46661760067290503,
      "grad_norm": 0.5134449005126953,
      "learning_rate": 2.6679108400799076e-05,
      "loss": 2.2281,
      "step": 133140
    },
    {
      "epoch": 0.46682788350331195,
      "grad_norm": 0.554543137550354,
      "learning_rate": 2.6668594259278728e-05,
      "loss": 2.2387,
      "step": 133200
    },
    {
      "epoch": 0.46703816633371886,
      "grad_norm": 0.5604194402694702,
      "learning_rate": 2.665808011775839e-05,
      "loss": 2.2123,
      "step": 133260
    },
    {
      "epoch": 0.4672484491641257,
      "grad_norm": 0.5490676760673523,
      "learning_rate": 2.664756597623804e-05,
      "loss": 2.2075,
      "step": 133320
    },
    {
      "epoch": 0.46745873199453264,
      "grad_norm": 0.5634739995002747,
      "learning_rate": 2.6637051834717697e-05,
      "loss": 2.2261,
      "step": 133380
    },
    {
      "epoch": 0.46766901482493956,
      "grad_norm": 0.5604760646820068,
      "learning_rate": 2.6626537693197352e-05,
      "loss": 2.2322,
      "step": 133440
    },
    {
      "epoch": 0.4678792976553464,
      "grad_norm": 0.5978127121925354,
      "learning_rate": 2.6616023551677004e-05,
      "loss": 2.237,
      "step": 133500
    },
    {
      "epoch": 0.46808958048575333,
      "grad_norm": 0.6151342988014221,
      "learning_rate": 2.660550941015666e-05,
      "loss": 2.2133,
      "step": 133560
    },
    {
      "epoch": 0.46829986331616025,
      "grad_norm": 0.597916305065155,
      "learning_rate": 2.6594995268636318e-05,
      "loss": 2.2218,
      "step": 133620
    },
    {
      "epoch": 0.4685101461465671,
      "grad_norm": 0.4812368154525757,
      "learning_rate": 2.6584481127115973e-05,
      "loss": 2.2089,
      "step": 133680
    },
    {
      "epoch": 0.468720428976974,
      "grad_norm": 0.5443069338798523,
      "learning_rate": 2.657396698559563e-05,
      "loss": 2.2308,
      "step": 133740
    },
    {
      "epoch": 0.46893071180738094,
      "grad_norm": 0.5664808750152588,
      "learning_rate": 2.656345284407528e-05,
      "loss": 2.2107,
      "step": 133800
    },
    {
      "epoch": 0.4691409946377878,
      "grad_norm": 0.5802755355834961,
      "learning_rate": 2.6552938702554936e-05,
      "loss": 2.2307,
      "step": 133860
    },
    {
      "epoch": 0.4693512774681947,
      "grad_norm": 0.5749623775482178,
      "learning_rate": 2.654242456103459e-05,
      "loss": 2.2241,
      "step": 133920
    },
    {
      "epoch": 0.46956156029860163,
      "grad_norm": 0.5218759775161743,
      "learning_rate": 2.653191041951425e-05,
      "loss": 2.2253,
      "step": 133980
    },
    {
      "epoch": 0.4697718431290085,
      "grad_norm": 0.5462409257888794,
      "learning_rate": 2.6521396277993905e-05,
      "loss": 2.2294,
      "step": 134040
    },
    {
      "epoch": 0.4699821259594154,
      "grad_norm": 0.5934569835662842,
      "learning_rate": 2.6510882136473557e-05,
      "loss": 2.2173,
      "step": 134100
    },
    {
      "epoch": 0.4701924087898223,
      "grad_norm": 0.5049318671226501,
      "learning_rate": 2.6500367994953212e-05,
      "loss": 2.2172,
      "step": 134160
    },
    {
      "epoch": 0.4704026916202292,
      "grad_norm": 0.5666798949241638,
      "learning_rate": 2.6489853853432867e-05,
      "loss": 2.2277,
      "step": 134220
    },
    {
      "epoch": 0.4706129744506361,
      "grad_norm": 0.5260395407676697,
      "learning_rate": 2.6479339711912526e-05,
      "loss": 2.2152,
      "step": 134280
    },
    {
      "epoch": 0.470823257281043,
      "grad_norm": 0.5611526966094971,
      "learning_rate": 2.646882557039218e-05,
      "loss": 2.2153,
      "step": 134340
    },
    {
      "epoch": 0.4710335401114499,
      "grad_norm": 0.5322962999343872,
      "learning_rate": 2.6458311428871833e-05,
      "loss": 2.2136,
      "step": 134400
    },
    {
      "epoch": 0.4712438229418568,
      "grad_norm": 0.5388842821121216,
      "learning_rate": 2.644779728735149e-05,
      "loss": 2.2204,
      "step": 134460
    },
    {
      "epoch": 0.4714541057722637,
      "grad_norm": 0.5794078707695007,
      "learning_rate": 2.6437283145831144e-05,
      "loss": 2.2324,
      "step": 134520
    },
    {
      "epoch": 0.47166438860267057,
      "grad_norm": 0.5857319235801697,
      "learning_rate": 2.64267690043108e-05,
      "loss": 2.2398,
      "step": 134580
    },
    {
      "epoch": 0.4718746714330775,
      "grad_norm": 0.5116488933563232,
      "learning_rate": 2.6416254862790458e-05,
      "loss": 2.2248,
      "step": 134640
    },
    {
      "epoch": 0.4720849542634844,
      "grad_norm": 0.5278642177581787,
      "learning_rate": 2.6405740721270113e-05,
      "loss": 2.2372,
      "step": 134700
    },
    {
      "epoch": 0.47229523709389126,
      "grad_norm": 0.5648896098136902,
      "learning_rate": 2.6395226579749765e-05,
      "loss": 2.22,
      "step": 134760
    },
    {
      "epoch": 0.4725055199242982,
      "grad_norm": 0.5393842458724976,
      "learning_rate": 2.638471243822942e-05,
      "loss": 2.2364,
      "step": 134820
    },
    {
      "epoch": 0.4727158027547051,
      "grad_norm": 0.558422863483429,
      "learning_rate": 2.6374198296709075e-05,
      "loss": 2.2243,
      "step": 134880
    },
    {
      "epoch": 0.47292608558511195,
      "grad_norm": 0.6029451489448547,
      "learning_rate": 2.6363684155188727e-05,
      "loss": 2.2318,
      "step": 134940
    },
    {
      "epoch": 0.47313636841551887,
      "grad_norm": 0.7082358002662659,
      "learning_rate": 2.635317001366839e-05,
      "loss": 2.2211,
      "step": 135000
    },
    {
      "epoch": 0.4733466512459258,
      "grad_norm": 0.551494300365448,
      "learning_rate": 2.634265587214804e-05,
      "loss": 2.2234,
      "step": 135060
    },
    {
      "epoch": 0.47355693407633265,
      "grad_norm": 0.4907706081867218,
      "learning_rate": 2.6332141730627697e-05,
      "loss": 2.2184,
      "step": 135120
    },
    {
      "epoch": 0.47376721690673956,
      "grad_norm": 0.5930429100990295,
      "learning_rate": 2.6321627589107352e-05,
      "loss": 2.2248,
      "step": 135180
    },
    {
      "epoch": 0.4739774997371465,
      "grad_norm": 0.6308527588844299,
      "learning_rate": 2.6311113447587004e-05,
      "loss": 2.2286,
      "step": 135240
    },
    {
      "epoch": 0.47418778256755334,
      "grad_norm": 0.5455154180526733,
      "learning_rate": 2.630059930606666e-05,
      "loss": 2.2261,
      "step": 135300
    },
    {
      "epoch": 0.47439806539796026,
      "grad_norm": 0.5316567420959473,
      "learning_rate": 2.6290085164546318e-05,
      "loss": 2.2183,
      "step": 135360
    },
    {
      "epoch": 0.47460834822836717,
      "grad_norm": 0.5488483309745789,
      "learning_rate": 2.6279571023025973e-05,
      "loss": 2.2377,
      "step": 135420
    },
    {
      "epoch": 0.47481863105877403,
      "grad_norm": 0.5488642454147339,
      "learning_rate": 2.6269056881505628e-05,
      "loss": 2.213,
      "step": 135480
    },
    {
      "epoch": 0.47502891388918095,
      "grad_norm": 0.5235707759857178,
      "learning_rate": 2.625854273998528e-05,
      "loss": 2.2269,
      "step": 135540
    },
    {
      "epoch": 0.47523919671958786,
      "grad_norm": 0.5823555588722229,
      "learning_rate": 2.6248028598464935e-05,
      "loss": 2.2304,
      "step": 135600
    },
    {
      "epoch": 0.4754494795499947,
      "grad_norm": 0.5386966466903687,
      "learning_rate": 2.623751445694459e-05,
      "loss": 2.2283,
      "step": 135660
    },
    {
      "epoch": 0.47565976238040164,
      "grad_norm": 0.5936428904533386,
      "learning_rate": 2.622700031542425e-05,
      "loss": 2.2251,
      "step": 135720
    },
    {
      "epoch": 0.47587004521080856,
      "grad_norm": 0.5601207613945007,
      "learning_rate": 2.6216486173903905e-05,
      "loss": 2.235,
      "step": 135780
    },
    {
      "epoch": 0.4760803280412154,
      "grad_norm": 0.4968276023864746,
      "learning_rate": 2.6205972032383557e-05,
      "loss": 2.2109,
      "step": 135840
    },
    {
      "epoch": 0.47629061087162233,
      "grad_norm": 0.5630781650543213,
      "learning_rate": 2.6195457890863212e-05,
      "loss": 2.2279,
      "step": 135900
    },
    {
      "epoch": 0.47650089370202925,
      "grad_norm": 0.552478551864624,
      "learning_rate": 2.6184943749342867e-05,
      "loss": 2.2225,
      "step": 135960
    },
    {
      "epoch": 0.4767111765324361,
      "grad_norm": 0.5814388394355774,
      "learning_rate": 2.617442960782252e-05,
      "loss": 2.2248,
      "step": 136020
    },
    {
      "epoch": 0.476921459362843,
      "grad_norm": 0.5658822059631348,
      "learning_rate": 2.616391546630218e-05,
      "loss": 2.2234,
      "step": 136080
    },
    {
      "epoch": 0.47713174219324994,
      "grad_norm": 0.5445215106010437,
      "learning_rate": 2.6153401324781833e-05,
      "loss": 2.2142,
      "step": 136140
    },
    {
      "epoch": 0.4773420250236568,
      "grad_norm": 0.6274155378341675,
      "learning_rate": 2.6142887183261488e-05,
      "loss": 2.2332,
      "step": 136200
    },
    {
      "epoch": 0.4775523078540637,
      "grad_norm": 0.527101457118988,
      "learning_rate": 2.6132373041741143e-05,
      "loss": 2.2325,
      "step": 136260
    },
    {
      "epoch": 0.47776259068447063,
      "grad_norm": 0.5380088090896606,
      "learning_rate": 2.6121858900220795e-05,
      "loss": 2.2114,
      "step": 136320
    },
    {
      "epoch": 0.4779728735148775,
      "grad_norm": 0.5434974431991577,
      "learning_rate": 2.611134475870045e-05,
      "loss": 2.2182,
      "step": 136380
    },
    {
      "epoch": 0.4781831563452844,
      "grad_norm": 0.5429152846336365,
      "learning_rate": 2.610083061718011e-05,
      "loss": 2.2297,
      "step": 136440
    },
    {
      "epoch": 0.4783934391756913,
      "grad_norm": 0.5782210826873779,
      "learning_rate": 2.6090316475659765e-05,
      "loss": 2.2281,
      "step": 136500
    },
    {
      "epoch": 0.4786037220060982,
      "grad_norm": 0.5653987526893616,
      "learning_rate": 2.607980233413942e-05,
      "loss": 2.2234,
      "step": 136560
    },
    {
      "epoch": 0.4788140048365051,
      "grad_norm": 0.5530957579612732,
      "learning_rate": 2.6069288192619075e-05,
      "loss": 2.208,
      "step": 136620
    },
    {
      "epoch": 0.479024287666912,
      "grad_norm": 0.5284966230392456,
      "learning_rate": 2.6058774051098727e-05,
      "loss": 2.2094,
      "step": 136680
    },
    {
      "epoch": 0.4792345704973189,
      "grad_norm": 0.5516782999038696,
      "learning_rate": 2.6048259909578382e-05,
      "loss": 2.2212,
      "step": 136740
    },
    {
      "epoch": 0.4794448533277258,
      "grad_norm": 0.498627245426178,
      "learning_rate": 2.603774576805804e-05,
      "loss": 2.222,
      "step": 136800
    },
    {
      "epoch": 0.4796551361581327,
      "grad_norm": 0.5247315168380737,
      "learning_rate": 2.6027231626537696e-05,
      "loss": 2.2061,
      "step": 136860
    },
    {
      "epoch": 0.47986541898853957,
      "grad_norm": 0.6753693222999573,
      "learning_rate": 2.601671748501735e-05,
      "loss": 2.2286,
      "step": 136920
    },
    {
      "epoch": 0.4800757018189465,
      "grad_norm": 0.5473688244819641,
      "learning_rate": 2.6006203343497003e-05,
      "loss": 2.2231,
      "step": 136980
    },
    {
      "epoch": 0.4802859846493534,
      "grad_norm": 0.5400590300559998,
      "learning_rate": 2.599568920197666e-05,
      "loss": 2.2172,
      "step": 137040
    },
    {
      "epoch": 0.48049626747976026,
      "grad_norm": 0.5710771679878235,
      "learning_rate": 2.5985175060456314e-05,
      "loss": 2.2218,
      "step": 137100
    },
    {
      "epoch": 0.4807065503101672,
      "grad_norm": 0.5183876752853394,
      "learning_rate": 2.5974660918935973e-05,
      "loss": 2.226,
      "step": 137160
    },
    {
      "epoch": 0.4809168331405741,
      "grad_norm": 0.5513122081756592,
      "learning_rate": 2.5964322013107633e-05,
      "loss": 2.2181,
      "step": 137220
    },
    {
      "epoch": 0.48112711597098096,
      "grad_norm": 0.4864659905433655,
      "learning_rate": 2.595380787158729e-05,
      "loss": 2.2214,
      "step": 137280
    },
    {
      "epoch": 0.4813373988013879,
      "grad_norm": 0.5089411735534668,
      "learning_rate": 2.5943293730066944e-05,
      "loss": 2.2298,
      "step": 137340
    },
    {
      "epoch": 0.4815476816317948,
      "grad_norm": 0.6495891809463501,
      "learning_rate": 2.5932779588546596e-05,
      "loss": 2.2157,
      "step": 137400
    },
    {
      "epoch": 0.48175796446220165,
      "grad_norm": 0.5468581914901733,
      "learning_rate": 2.592226544702625e-05,
      "loss": 2.2176,
      "step": 137460
    },
    {
      "epoch": 0.48196824729260856,
      "grad_norm": 0.49662095308303833,
      "learning_rate": 2.5911751305505906e-05,
      "loss": 2.2285,
      "step": 137520
    },
    {
      "epoch": 0.4821785301230155,
      "grad_norm": 0.5509756207466125,
      "learning_rate": 2.5901237163985565e-05,
      "loss": 2.2188,
      "step": 137580
    },
    {
      "epoch": 0.48238881295342234,
      "grad_norm": 0.4889494776725769,
      "learning_rate": 2.589072302246522e-05,
      "loss": 2.2104,
      "step": 137640
    },
    {
      "epoch": 0.48259909578382926,
      "grad_norm": 0.5467410087585449,
      "learning_rate": 2.5880208880944872e-05,
      "loss": 2.2241,
      "step": 137700
    },
    {
      "epoch": 0.4828093786142362,
      "grad_norm": 0.5617779493331909,
      "learning_rate": 2.5869694739424527e-05,
      "loss": 2.2173,
      "step": 137760
    },
    {
      "epoch": 0.48301966144464303,
      "grad_norm": 0.5608574748039246,
      "learning_rate": 2.5859180597904183e-05,
      "loss": 2.225,
      "step": 137820
    },
    {
      "epoch": 0.48322994427504995,
      "grad_norm": 0.5132883787155151,
      "learning_rate": 2.5848666456383834e-05,
      "loss": 2.2,
      "step": 137880
    },
    {
      "epoch": 0.48344022710545687,
      "grad_norm": 0.5200146436691284,
      "learning_rate": 2.5838152314863497e-05,
      "loss": 2.2331,
      "step": 137940
    },
    {
      "epoch": 0.4836505099358637,
      "grad_norm": 0.5218937993049622,
      "learning_rate": 2.582763817334315e-05,
      "loss": 2.2115,
      "step": 138000
    },
    {
      "epoch": 0.48386079276627064,
      "grad_norm": 0.5255389213562012,
      "learning_rate": 2.5817124031822804e-05,
      "loss": 2.2251,
      "step": 138060
    },
    {
      "epoch": 0.48407107559667756,
      "grad_norm": 0.5643746852874756,
      "learning_rate": 2.580660989030246e-05,
      "loss": 2.2332,
      "step": 138120
    },
    {
      "epoch": 0.4842813584270844,
      "grad_norm": 0.6352805495262146,
      "learning_rate": 2.579609574878211e-05,
      "loss": 2.2218,
      "step": 138180
    },
    {
      "epoch": 0.48449164125749133,
      "grad_norm": 0.559818685054779,
      "learning_rate": 2.5785756842953775e-05,
      "loss": 2.2327,
      "step": 138240
    },
    {
      "epoch": 0.48470192408789825,
      "grad_norm": 0.5820249319076538,
      "learning_rate": 2.5775242701433427e-05,
      "loss": 2.2132,
      "step": 138300
    },
    {
      "epoch": 0.4849122069183051,
      "grad_norm": 0.551038384437561,
      "learning_rate": 2.5764728559913082e-05,
      "loss": 2.2305,
      "step": 138360
    },
    {
      "epoch": 0.485122489748712,
      "grad_norm": 0.512161374092102,
      "learning_rate": 2.575421441839274e-05,
      "loss": 2.2239,
      "step": 138420
    },
    {
      "epoch": 0.4853327725791189,
      "grad_norm": 0.5664234161376953,
      "learning_rate": 2.5743700276872396e-05,
      "loss": 2.2149,
      "step": 138480
    },
    {
      "epoch": 0.4855430554095258,
      "grad_norm": 0.525309681892395,
      "learning_rate": 2.573318613535205e-05,
      "loss": 2.2167,
      "step": 138540
    },
    {
      "epoch": 0.4857533382399327,
      "grad_norm": 0.5458988547325134,
      "learning_rate": 2.5722671993831703e-05,
      "loss": 2.2209,
      "step": 138600
    },
    {
      "epoch": 0.4859636210703396,
      "grad_norm": 0.5710783004760742,
      "learning_rate": 2.571215785231136e-05,
      "loss": 2.2239,
      "step": 138660
    },
    {
      "epoch": 0.4861739039007465,
      "grad_norm": 0.537351131439209,
      "learning_rate": 2.5701643710791014e-05,
      "loss": 2.2072,
      "step": 138720
    },
    {
      "epoch": 0.4863841867311534,
      "grad_norm": 0.5294988751411438,
      "learning_rate": 2.5691129569270672e-05,
      "loss": 2.2131,
      "step": 138780
    },
    {
      "epoch": 0.48659446956156027,
      "grad_norm": 0.4968686103820801,
      "learning_rate": 2.5680615427750328e-05,
      "loss": 2.2291,
      "step": 138840
    },
    {
      "epoch": 0.4868047523919672,
      "grad_norm": 0.5579273700714111,
      "learning_rate": 2.567010128622998e-05,
      "loss": 2.2172,
      "step": 138900
    },
    {
      "epoch": 0.4870150352223741,
      "grad_norm": 0.5695374608039856,
      "learning_rate": 2.5659587144709635e-05,
      "loss": 2.2108,
      "step": 138960
    },
    {
      "epoch": 0.48722531805278096,
      "grad_norm": 0.5216060280799866,
      "learning_rate": 2.564907300318929e-05,
      "loss": 2.2162,
      "step": 139020
    },
    {
      "epoch": 0.4874356008831879,
      "grad_norm": 0.491592139005661,
      "learning_rate": 2.563855886166895e-05,
      "loss": 2.2196,
      "step": 139080
    },
    {
      "epoch": 0.4876458837135948,
      "grad_norm": 0.6150323152542114,
      "learning_rate": 2.5628044720148604e-05,
      "loss": 2.205,
      "step": 139140
    },
    {
      "epoch": 0.48785616654400166,
      "grad_norm": 0.5466335415840149,
      "learning_rate": 2.5617705814320265e-05,
      "loss": 2.2297,
      "step": 139200
    },
    {
      "epoch": 0.4880664493744086,
      "grad_norm": 0.6138894557952881,
      "learning_rate": 2.560719167279992e-05,
      "loss": 2.2275,
      "step": 139260
    },
    {
      "epoch": 0.4882767322048155,
      "grad_norm": 0.5531201362609863,
      "learning_rate": 2.559667753127957e-05,
      "loss": 2.2157,
      "step": 139320
    },
    {
      "epoch": 0.48848701503522235,
      "grad_norm": 0.5969268083572388,
      "learning_rate": 2.5586163389759227e-05,
      "loss": 2.2197,
      "step": 139380
    },
    {
      "epoch": 0.48869729786562927,
      "grad_norm": 0.5635274648666382,
      "learning_rate": 2.5575649248238882e-05,
      "loss": 2.2253,
      "step": 139440
    },
    {
      "epoch": 0.4889075806960362,
      "grad_norm": 0.6807565689086914,
      "learning_rate": 2.5565135106718534e-05,
      "loss": 2.2206,
      "step": 139500
    },
    {
      "epoch": 0.48911786352644304,
      "grad_norm": 0.5284523963928223,
      "learning_rate": 2.5554620965198196e-05,
      "loss": 2.2179,
      "step": 139560
    },
    {
      "epoch": 0.48932814635684996,
      "grad_norm": 0.614186704158783,
      "learning_rate": 2.5544106823677848e-05,
      "loss": 2.2238,
      "step": 139620
    },
    {
      "epoch": 0.4895384291872569,
      "grad_norm": 0.48403292894363403,
      "learning_rate": 2.5533592682157503e-05,
      "loss": 2.2155,
      "step": 139680
    },
    {
      "epoch": 0.48974871201766373,
      "grad_norm": 0.536378026008606,
      "learning_rate": 2.552307854063716e-05,
      "loss": 2.2065,
      "step": 139740
    },
    {
      "epoch": 0.48995899484807065,
      "grad_norm": 0.5436242818832397,
      "learning_rate": 2.551256439911681e-05,
      "loss": 2.2155,
      "step": 139800
    },
    {
      "epoch": 0.49016927767847757,
      "grad_norm": 0.5471640825271606,
      "learning_rate": 2.5502050257596466e-05,
      "loss": 2.2185,
      "step": 139860
    },
    {
      "epoch": 0.4903795605088844,
      "grad_norm": 0.5677446722984314,
      "learning_rate": 2.5491536116076124e-05,
      "loss": 2.2149,
      "step": 139920
    },
    {
      "epoch": 0.49058984333929134,
      "grad_norm": 0.5640824437141418,
      "learning_rate": 2.548102197455578e-05,
      "loss": 2.219,
      "step": 139980
    },
    {
      "epoch": 0.49080012616969826,
      "grad_norm": 0.54488605260849,
      "learning_rate": 2.5470507833035435e-05,
      "loss": 2.2243,
      "step": 140040
    },
    {
      "epoch": 0.4910104090001051,
      "grad_norm": 0.587646484375,
      "learning_rate": 2.5459993691515087e-05,
      "loss": 2.2201,
      "step": 140100
    },
    {
      "epoch": 0.49122069183051204,
      "grad_norm": 0.5216631293296814,
      "learning_rate": 2.5449479549994742e-05,
      "loss": 2.2111,
      "step": 140160
    },
    {
      "epoch": 0.49143097466091895,
      "grad_norm": 0.5511109828948975,
      "learning_rate": 2.5438965408474397e-05,
      "loss": 2.2034,
      "step": 140220
    },
    {
      "epoch": 0.4916412574913258,
      "grad_norm": 0.5907667875289917,
      "learning_rate": 2.5428451266954056e-05,
      "loss": 2.2101,
      "step": 140280
    },
    {
      "epoch": 0.4918515403217327,
      "grad_norm": 0.6119207739830017,
      "learning_rate": 2.541793712543371e-05,
      "loss": 2.2313,
      "step": 140340
    },
    {
      "epoch": 0.49206182315213964,
      "grad_norm": 0.5808233618736267,
      "learning_rate": 2.5407422983913363e-05,
      "loss": 2.2082,
      "step": 140400
    },
    {
      "epoch": 0.4922721059825465,
      "grad_norm": 0.5466307401657104,
      "learning_rate": 2.539690884239302e-05,
      "loss": 2.2109,
      "step": 140460
    },
    {
      "epoch": 0.4924823888129534,
      "grad_norm": 0.5470625758171082,
      "learning_rate": 2.5386394700872674e-05,
      "loss": 2.2139,
      "step": 140520
    },
    {
      "epoch": 0.49269267164336034,
      "grad_norm": 0.5294373035430908,
      "learning_rate": 2.5375880559352326e-05,
      "loss": 2.2275,
      "step": 140580
    },
    {
      "epoch": 0.4929029544737672,
      "grad_norm": 0.5608066320419312,
      "learning_rate": 2.5365366417831988e-05,
      "loss": 2.2222,
      "step": 140640
    },
    {
      "epoch": 0.4931132373041741,
      "grad_norm": 0.5671436190605164,
      "learning_rate": 2.535485227631164e-05,
      "loss": 2.2208,
      "step": 140700
    },
    {
      "epoch": 0.49332352013458103,
      "grad_norm": 0.5754434466362,
      "learning_rate": 2.5344338134791295e-05,
      "loss": 2.2126,
      "step": 140760
    },
    {
      "epoch": 0.4935338029649879,
      "grad_norm": 0.5340360403060913,
      "learning_rate": 2.533382399327095e-05,
      "loss": 2.2138,
      "step": 140820
    },
    {
      "epoch": 0.4937440857953948,
      "grad_norm": 0.5738496780395508,
      "learning_rate": 2.5323309851750606e-05,
      "loss": 2.2305,
      "step": 140880
    },
    {
      "epoch": 0.4939543686258017,
      "grad_norm": 0.49853968620300293,
      "learning_rate": 2.5312795710230257e-05,
      "loss": 2.1923,
      "step": 140940
    },
    {
      "epoch": 0.4941646514562086,
      "grad_norm": 0.5699918866157532,
      "learning_rate": 2.530228156870992e-05,
      "loss": 2.2172,
      "step": 141000
    },
    {
      "epoch": 0.4943749342866155,
      "grad_norm": 0.6300356984138489,
      "learning_rate": 2.529176742718957e-05,
      "loss": 2.217,
      "step": 141060
    },
    {
      "epoch": 0.4945852171170224,
      "grad_norm": 0.5313805937767029,
      "learning_rate": 2.5281253285669227e-05,
      "loss": 2.2302,
      "step": 141120
    },
    {
      "epoch": 0.4947954999474293,
      "grad_norm": 0.6249646544456482,
      "learning_rate": 2.5270739144148882e-05,
      "loss": 2.2078,
      "step": 141180
    },
    {
      "epoch": 0.4950057827778362,
      "grad_norm": 0.5782462954521179,
      "learning_rate": 2.5260225002628534e-05,
      "loss": 2.2239,
      "step": 141240
    },
    {
      "epoch": 0.4952160656082431,
      "grad_norm": 0.5095561146736145,
      "learning_rate": 2.524971086110819e-05,
      "loss": 2.2215,
      "step": 141300
    },
    {
      "epoch": 0.49542634843864997,
      "grad_norm": 0.5634046196937561,
      "learning_rate": 2.5239196719587848e-05,
      "loss": 2.2081,
      "step": 141360
    },
    {
      "epoch": 0.4956366312690569,
      "grad_norm": 0.5690566301345825,
      "learning_rate": 2.5228682578067503e-05,
      "loss": 2.2248,
      "step": 141420
    },
    {
      "epoch": 0.4958469140994638,
      "grad_norm": 0.5184747576713562,
      "learning_rate": 2.521816843654716e-05,
      "loss": 2.2148,
      "step": 141480
    },
    {
      "epoch": 0.49605719692987066,
      "grad_norm": 0.5188051462173462,
      "learning_rate": 2.520765429502681e-05,
      "loss": 2.2166,
      "step": 141540
    },
    {
      "epoch": 0.4962674797602776,
      "grad_norm": 0.5549147725105286,
      "learning_rate": 2.5197140153506466e-05,
      "loss": 2.2322,
      "step": 141600
    },
    {
      "epoch": 0.4964777625906845,
      "grad_norm": 0.5586085319519043,
      "learning_rate": 2.5186626011986124e-05,
      "loss": 2.2367,
      "step": 141660
    },
    {
      "epoch": 0.49668804542109135,
      "grad_norm": 0.5334410667419434,
      "learning_rate": 2.517611187046578e-05,
      "loss": 2.2272,
      "step": 141720
    },
    {
      "epoch": 0.49689832825149827,
      "grad_norm": 0.560259222984314,
      "learning_rate": 2.5165597728945435e-05,
      "loss": 2.2138,
      "step": 141780
    },
    {
      "epoch": 0.4971086110819052,
      "grad_norm": 0.5895169377326965,
      "learning_rate": 2.5155083587425087e-05,
      "loss": 2.2213,
      "step": 141840
    },
    {
      "epoch": 0.49731889391231204,
      "grad_norm": 0.5676740407943726,
      "learning_rate": 2.5144569445904742e-05,
      "loss": 2.2199,
      "step": 141900
    },
    {
      "epoch": 0.49752917674271896,
      "grad_norm": 0.507685661315918,
      "learning_rate": 2.5134055304384397e-05,
      "loss": 2.2129,
      "step": 141960
    },
    {
      "epoch": 0.4977394595731259,
      "grad_norm": 0.5849261283874512,
      "learning_rate": 2.5123541162864056e-05,
      "loss": 2.2321,
      "step": 142020
    },
    {
      "epoch": 0.49794974240353274,
      "grad_norm": 0.5333006978034973,
      "learning_rate": 2.511302702134371e-05,
      "loss": 2.2164,
      "step": 142080
    },
    {
      "epoch": 0.49816002523393965,
      "grad_norm": 0.5627948045730591,
      "learning_rate": 2.5102512879823363e-05,
      "loss": 2.2245,
      "step": 142140
    },
    {
      "epoch": 0.49837030806434657,
      "grad_norm": 0.5709524154663086,
      "learning_rate": 2.509199873830302e-05,
      "loss": 2.2225,
      "step": 142200
    },
    {
      "epoch": 0.49858059089475343,
      "grad_norm": 0.5809540748596191,
      "learning_rate": 2.5081484596782674e-05,
      "loss": 2.2103,
      "step": 142260
    },
    {
      "epoch": 0.49879087372516034,
      "grad_norm": 0.5410124659538269,
      "learning_rate": 2.5070970455262326e-05,
      "loss": 2.2285,
      "step": 142320
    },
    {
      "epoch": 0.49900115655556726,
      "grad_norm": 0.5636032819747925,
      "learning_rate": 2.5060456313741988e-05,
      "loss": 2.2171,
      "step": 142380
    },
    {
      "epoch": 0.4992114393859741,
      "grad_norm": 0.5565786361694336,
      "learning_rate": 2.504994217222164e-05,
      "loss": 2.2195,
      "step": 142440
    },
    {
      "epoch": 0.49942172221638104,
      "grad_norm": 0.5169489979743958,
      "learning_rate": 2.5039428030701295e-05,
      "loss": 2.232,
      "step": 142500
    },
    {
      "epoch": 0.49963200504678795,
      "grad_norm": 0.5828045010566711,
      "learning_rate": 2.502891388918095e-05,
      "loss": 2.226,
      "step": 142560
    },
    {
      "epoch": 0.4998422878771948,
      "grad_norm": 0.5285956859588623,
      "learning_rate": 2.5018399747660602e-05,
      "loss": 2.2235,
      "step": 142620
    },
    {
      "epoch": 0.5000525707076017,
      "grad_norm": 0.5771108865737915,
      "learning_rate": 2.5007885606140257e-05,
      "loss": 2.2077,
      "step": 142680
    },
    {
      "epoch": 0.5002628535380086,
      "grad_norm": 0.5183426737785339,
      "learning_rate": 2.4997371464619916e-05,
      "loss": 2.2273,
      "step": 142740
    },
    {
      "epoch": 0.5004731363684155,
      "grad_norm": 0.5981959700584412,
      "learning_rate": 2.4986857323099568e-05,
      "loss": 2.2246,
      "step": 142800
    },
    {
      "epoch": 0.5006834191988224,
      "grad_norm": 0.5162586569786072,
      "learning_rate": 2.4976343181579226e-05,
      "loss": 2.215,
      "step": 142860
    },
    {
      "epoch": 0.5008937020292293,
      "grad_norm": 0.5488058924674988,
      "learning_rate": 2.4965829040058882e-05,
      "loss": 2.2118,
      "step": 142920
    },
    {
      "epoch": 0.5011039848596363,
      "grad_norm": 0.5368149280548096,
      "learning_rate": 2.4955314898538534e-05,
      "loss": 2.216,
      "step": 142980
    },
    {
      "epoch": 0.5013142676900431,
      "grad_norm": 0.5487305521965027,
      "learning_rate": 2.4944800757018192e-05,
      "loss": 2.211,
      "step": 143040
    },
    {
      "epoch": 0.50152455052045,
      "grad_norm": 0.5786258578300476,
      "learning_rate": 2.4934286615497844e-05,
      "loss": 2.2345,
      "step": 143100
    },
    {
      "epoch": 0.5017348333508569,
      "grad_norm": 0.549926221370697,
      "learning_rate": 2.49237724739775e-05,
      "loss": 2.2161,
      "step": 143160
    },
    {
      "epoch": 0.5019451161812638,
      "grad_norm": 0.5187004208564758,
      "learning_rate": 2.4913258332457158e-05,
      "loss": 2.233,
      "step": 143220
    },
    {
      "epoch": 0.5021553990116707,
      "grad_norm": 0.5308973789215088,
      "learning_rate": 2.490274419093681e-05,
      "loss": 2.2098,
      "step": 143280
    },
    {
      "epoch": 0.5023656818420776,
      "grad_norm": 0.5926359295845032,
      "learning_rate": 2.4892230049416465e-05,
      "loss": 2.2082,
      "step": 143340
    },
    {
      "epoch": 0.5025759646724844,
      "grad_norm": 0.5576338768005371,
      "learning_rate": 2.488171590789612e-05,
      "loss": 2.2275,
      "step": 143400
    },
    {
      "epoch": 0.5027862475028914,
      "grad_norm": 0.5443142652511597,
      "learning_rate": 2.4871201766375776e-05,
      "loss": 2.2097,
      "step": 143460
    },
    {
      "epoch": 0.5029965303332983,
      "grad_norm": 0.5324608683586121,
      "learning_rate": 2.486068762485543e-05,
      "loss": 2.2276,
      "step": 143520
    },
    {
      "epoch": 0.5032068131637052,
      "grad_norm": 0.6219227313995361,
      "learning_rate": 2.4850173483335086e-05,
      "loss": 2.2128,
      "step": 143580
    },
    {
      "epoch": 0.5034170959941121,
      "grad_norm": 0.5511086583137512,
      "learning_rate": 2.483965934181474e-05,
      "loss": 2.2231,
      "step": 143640
    },
    {
      "epoch": 0.503627378824519,
      "grad_norm": 0.5457072257995605,
      "learning_rate": 2.4829145200294397e-05,
      "loss": 2.2117,
      "step": 143700
    },
    {
      "epoch": 0.5038376616549258,
      "grad_norm": 0.5231776237487793,
      "learning_rate": 2.4818631058774052e-05,
      "loss": 2.2244,
      "step": 143760
    },
    {
      "epoch": 0.5040479444853327,
      "grad_norm": 0.5896267890930176,
      "learning_rate": 2.4808116917253708e-05,
      "loss": 2.2073,
      "step": 143820
    },
    {
      "epoch": 0.5042582273157397,
      "grad_norm": 0.5525411367416382,
      "learning_rate": 2.4797602775733363e-05,
      "loss": 2.2325,
      "step": 143880
    },
    {
      "epoch": 0.5044685101461466,
      "grad_norm": 0.5126237869262695,
      "learning_rate": 2.4787088634213018e-05,
      "loss": 2.2142,
      "step": 143940
    },
    {
      "epoch": 0.5046787929765535,
      "grad_norm": 0.5332605242729187,
      "learning_rate": 2.4776574492692673e-05,
      "loss": 2.2041,
      "step": 144000
    },
    {
      "epoch": 0.5048890758069604,
      "grad_norm": 0.5159221291542053,
      "learning_rate": 2.4766060351172325e-05,
      "loss": 2.2258,
      "step": 144060
    },
    {
      "epoch": 0.5050993586373672,
      "grad_norm": 0.546165406703949,
      "learning_rate": 2.4755546209651984e-05,
      "loss": 2.2118,
      "step": 144120
    },
    {
      "epoch": 0.5053096414677741,
      "grad_norm": 0.6250487565994263,
      "learning_rate": 2.474503206813164e-05,
      "loss": 2.2144,
      "step": 144180
    },
    {
      "epoch": 0.505519924298181,
      "grad_norm": 0.5558093190193176,
      "learning_rate": 2.4734517926611294e-05,
      "loss": 2.2113,
      "step": 144240
    },
    {
      "epoch": 0.505730207128588,
      "grad_norm": 0.5311017036437988,
      "learning_rate": 2.472400378509095e-05,
      "loss": 2.2064,
      "step": 144300
    },
    {
      "epoch": 0.5059404899589949,
      "grad_norm": 0.5303267240524292,
      "learning_rate": 2.47134896435706e-05,
      "loss": 2.2178,
      "step": 144360
    },
    {
      "epoch": 0.5061507727894018,
      "grad_norm": 0.5508550405502319,
      "learning_rate": 2.470297550205026e-05,
      "loss": 2.2266,
      "step": 144420
    },
    {
      "epoch": 0.5063610556198086,
      "grad_norm": 0.5501142144203186,
      "learning_rate": 2.4692461360529916e-05,
      "loss": 2.2181,
      "step": 144480
    },
    {
      "epoch": 0.5065713384502155,
      "grad_norm": 0.5321212410926819,
      "learning_rate": 2.4681947219009567e-05,
      "loss": 2.2072,
      "step": 144540
    },
    {
      "epoch": 0.5067816212806224,
      "grad_norm": 0.598643958568573,
      "learning_rate": 2.4671433077489226e-05,
      "loss": 2.2166,
      "step": 144600
    },
    {
      "epoch": 0.5069919041110293,
      "grad_norm": 0.5446153879165649,
      "learning_rate": 2.4660918935968878e-05,
      "loss": 2.2168,
      "step": 144660
    },
    {
      "epoch": 0.5072021869414363,
      "grad_norm": 0.5070498585700989,
      "learning_rate": 2.4650404794448533e-05,
      "loss": 2.2242,
      "step": 144720
    },
    {
      "epoch": 0.5074124697718432,
      "grad_norm": 0.5821142792701721,
      "learning_rate": 2.4639890652928192e-05,
      "loss": 2.2081,
      "step": 144780
    },
    {
      "epoch": 0.50762275260225,
      "grad_norm": 0.6096354126930237,
      "learning_rate": 2.4629376511407844e-05,
      "loss": 2.2235,
      "step": 144840
    },
    {
      "epoch": 0.5078330354326569,
      "grad_norm": 0.5673126578330994,
      "learning_rate": 2.46188623698875e-05,
      "loss": 2.2095,
      "step": 144900
    },
    {
      "epoch": 0.5080433182630638,
      "grad_norm": 0.6011648774147034,
      "learning_rate": 2.4608348228367158e-05,
      "loss": 2.2172,
      "step": 144960
    },
    {
      "epoch": 0.5082536010934707,
      "grad_norm": 0.5500099062919617,
      "learning_rate": 2.459783408684681e-05,
      "loss": 2.2266,
      "step": 145020
    },
    {
      "epoch": 0.5084638839238776,
      "grad_norm": 0.6012193560600281,
      "learning_rate": 2.4587319945326465e-05,
      "loss": 2.2244,
      "step": 145080
    },
    {
      "epoch": 0.5086741667542846,
      "grad_norm": 0.5006005764007568,
      "learning_rate": 2.457680580380612e-05,
      "loss": 2.2187,
      "step": 145140
    },
    {
      "epoch": 0.5088844495846914,
      "grad_norm": 0.5464049577713013,
      "learning_rate": 2.456646689797778e-05,
      "loss": 2.2264,
      "step": 145200
    },
    {
      "epoch": 0.5090947324150983,
      "grad_norm": 0.5614712238311768,
      "learning_rate": 2.4555952756457436e-05,
      "loss": 2.223,
      "step": 145260
    },
    {
      "epoch": 0.5093050152455052,
      "grad_norm": 0.5789541602134705,
      "learning_rate": 2.454543861493709e-05,
      "loss": 2.1983,
      "step": 145320
    },
    {
      "epoch": 0.5095152980759121,
      "grad_norm": 0.5474076271057129,
      "learning_rate": 2.4534924473416747e-05,
      "loss": 2.2146,
      "step": 145380
    },
    {
      "epoch": 0.509725580906319,
      "grad_norm": 0.49787166714668274,
      "learning_rate": 2.4524410331896402e-05,
      "loss": 2.2116,
      "step": 145440
    },
    {
      "epoch": 0.509935863736726,
      "grad_norm": 0.5263041853904724,
      "learning_rate": 2.4513896190376057e-05,
      "loss": 2.2207,
      "step": 145500
    },
    {
      "epoch": 0.5101461465671328,
      "grad_norm": 0.5450867414474487,
      "learning_rate": 2.4503382048855712e-05,
      "loss": 2.2196,
      "step": 145560
    },
    {
      "epoch": 0.5103564293975397,
      "grad_norm": 0.5154622793197632,
      "learning_rate": 2.4492867907335368e-05,
      "loss": 2.2239,
      "step": 145620
    },
    {
      "epoch": 0.5105667122279466,
      "grad_norm": 0.5420225858688354,
      "learning_rate": 2.4482353765815023e-05,
      "loss": 2.2192,
      "step": 145680
    },
    {
      "epoch": 0.5107769950583535,
      "grad_norm": 0.544823944568634,
      "learning_rate": 2.4471839624294675e-05,
      "loss": 2.2148,
      "step": 145740
    },
    {
      "epoch": 0.5109872778887604,
      "grad_norm": 0.5374436378479004,
      "learning_rate": 2.4461325482774334e-05,
      "loss": 2.2034,
      "step": 145800
    },
    {
      "epoch": 0.5111975607191673,
      "grad_norm": 0.5344963073730469,
      "learning_rate": 2.445081134125399e-05,
      "loss": 2.2237,
      "step": 145860
    },
    {
      "epoch": 0.5114078435495741,
      "grad_norm": 0.5483312010765076,
      "learning_rate": 2.444029719973364e-05,
      "loss": 2.2139,
      "step": 145920
    },
    {
      "epoch": 0.511618126379981,
      "grad_norm": 0.5935961604118347,
      "learning_rate": 2.44297830582133e-05,
      "loss": 2.2039,
      "step": 145980
    },
    {
      "epoch": 0.511828409210388,
      "grad_norm": 0.4969542920589447,
      "learning_rate": 2.441926891669295e-05,
      "loss": 2.2228,
      "step": 146040
    },
    {
      "epoch": 0.5120386920407949,
      "grad_norm": 0.5557306408882141,
      "learning_rate": 2.4408754775172607e-05,
      "loss": 2.2329,
      "step": 146100
    },
    {
      "epoch": 0.5122489748712018,
      "grad_norm": 0.6203109622001648,
      "learning_rate": 2.4398240633652265e-05,
      "loss": 2.2258,
      "step": 146160
    },
    {
      "epoch": 0.5124592577016087,
      "grad_norm": 0.5021337270736694,
      "learning_rate": 2.4387726492131917e-05,
      "loss": 2.2227,
      "step": 146220
    },
    {
      "epoch": 0.5126695405320155,
      "grad_norm": 0.5517904162406921,
      "learning_rate": 2.4377212350611572e-05,
      "loss": 2.213,
      "step": 146280
    },
    {
      "epoch": 0.5128798233624224,
      "grad_norm": 0.5719086527824402,
      "learning_rate": 2.4366698209091228e-05,
      "loss": 2.2296,
      "step": 146340
    },
    {
      "epoch": 0.5130901061928294,
      "grad_norm": 0.6046878695487976,
      "learning_rate": 2.4356184067570883e-05,
      "loss": 2.2074,
      "step": 146400
    },
    {
      "epoch": 0.5133003890232363,
      "grad_norm": 0.5404784679412842,
      "learning_rate": 2.4345669926050538e-05,
      "loss": 2.231,
      "step": 146460
    },
    {
      "epoch": 0.5135106718536432,
      "grad_norm": 0.49592384696006775,
      "learning_rate": 2.4335155784530194e-05,
      "loss": 2.2109,
      "step": 146520
    },
    {
      "epoch": 0.5137209546840501,
      "grad_norm": 0.5282294750213623,
      "learning_rate": 2.432464164300985e-05,
      "loss": 2.2199,
      "step": 146580
    },
    {
      "epoch": 0.5139312375144569,
      "grad_norm": 0.5357757806777954,
      "learning_rate": 2.4314127501489508e-05,
      "loss": 2.2123,
      "step": 146640
    },
    {
      "epoch": 0.5141415203448638,
      "grad_norm": 0.5113288760185242,
      "learning_rate": 2.430361335996916e-05,
      "loss": 2.2078,
      "step": 146700
    },
    {
      "epoch": 0.5143518031752707,
      "grad_norm": 0.5509752631187439,
      "learning_rate": 2.4293099218448815e-05,
      "loss": 2.2105,
      "step": 146760
    },
    {
      "epoch": 0.5145620860056777,
      "grad_norm": 0.5632410049438477,
      "learning_rate": 2.428258507692847e-05,
      "loss": 2.2158,
      "step": 146820
    },
    {
      "epoch": 0.5147723688360846,
      "grad_norm": 0.525120198726654,
      "learning_rate": 2.4272070935408125e-05,
      "loss": 2.2228,
      "step": 146880
    },
    {
      "epoch": 0.5149826516664914,
      "grad_norm": 0.5465018153190613,
      "learning_rate": 2.426155679388778e-05,
      "loss": 2.201,
      "step": 146940
    },
    {
      "epoch": 0.5151929344968983,
      "grad_norm": 0.5666688084602356,
      "learning_rate": 2.4251042652367436e-05,
      "loss": 2.199,
      "step": 147000
    },
    {
      "epoch": 0.5154032173273052,
      "grad_norm": 0.5029938220977783,
      "learning_rate": 2.424052851084709e-05,
      "loss": 2.21,
      "step": 147060
    },
    {
      "epoch": 0.5156135001577121,
      "grad_norm": 0.5832599401473999,
      "learning_rate": 2.4230014369326746e-05,
      "loss": 2.2302,
      "step": 147120
    },
    {
      "epoch": 0.515823782988119,
      "grad_norm": 0.5170128345489502,
      "learning_rate": 2.4219675463498407e-05,
      "loss": 2.2179,
      "step": 147180
    },
    {
      "epoch": 0.516034065818526,
      "grad_norm": 0.6044071316719055,
      "learning_rate": 2.4209161321978062e-05,
      "loss": 2.2212,
      "step": 147240
    },
    {
      "epoch": 0.5162443486489328,
      "grad_norm": 0.5323086977005005,
      "learning_rate": 2.4198647180457717e-05,
      "loss": 2.2144,
      "step": 147300
    },
    {
      "epoch": 0.5164546314793397,
      "grad_norm": 0.5192455649375916,
      "learning_rate": 2.4188133038937373e-05,
      "loss": 2.2064,
      "step": 147360
    },
    {
      "epoch": 0.5166649143097466,
      "grad_norm": 0.5478724241256714,
      "learning_rate": 2.4177618897417025e-05,
      "loss": 2.2047,
      "step": 147420
    },
    {
      "epoch": 0.5168751971401535,
      "grad_norm": 0.5389701724052429,
      "learning_rate": 2.4167104755896683e-05,
      "loss": 2.2104,
      "step": 147480
    },
    {
      "epoch": 0.5170854799705604,
      "grad_norm": 0.5311550498008728,
      "learning_rate": 2.415659061437634e-05,
      "loss": 2.214,
      "step": 147540
    },
    {
      "epoch": 0.5172957628009673,
      "grad_norm": 0.6084071397781372,
      "learning_rate": 2.414607647285599e-05,
      "loss": 2.2091,
      "step": 147600
    },
    {
      "epoch": 0.5175060456313741,
      "grad_norm": 0.5623484253883362,
      "learning_rate": 2.413556233133565e-05,
      "loss": 2.2095,
      "step": 147660
    },
    {
      "epoch": 0.5177163284617811,
      "grad_norm": 0.509027898311615,
      "learning_rate": 2.41250481898153e-05,
      "loss": 2.2212,
      "step": 147720
    },
    {
      "epoch": 0.517926611292188,
      "grad_norm": 0.5712621808052063,
      "learning_rate": 2.4114534048294956e-05,
      "loss": 2.2106,
      "step": 147780
    },
    {
      "epoch": 0.5181368941225949,
      "grad_norm": 0.5070745348930359,
      "learning_rate": 2.4104019906774615e-05,
      "loss": 2.2146,
      "step": 147840
    },
    {
      "epoch": 0.5183471769530018,
      "grad_norm": 0.5508371591567993,
      "learning_rate": 2.4093505765254267e-05,
      "loss": 2.2136,
      "step": 147900
    },
    {
      "epoch": 0.5185574597834087,
      "grad_norm": 0.5198752284049988,
      "learning_rate": 2.4082991623733922e-05,
      "loss": 2.2137,
      "step": 147960
    },
    {
      "epoch": 0.5187677426138155,
      "grad_norm": 0.555348813533783,
      "learning_rate": 2.4072477482213577e-05,
      "loss": 2.2097,
      "step": 148020
    },
    {
      "epoch": 0.5189780254442224,
      "grad_norm": 0.5508757829666138,
      "learning_rate": 2.4061963340693233e-05,
      "loss": 2.2097,
      "step": 148080
    },
    {
      "epoch": 0.5191883082746294,
      "grad_norm": 0.5607093572616577,
      "learning_rate": 2.4051449199172888e-05,
      "loss": 2.2048,
      "step": 148140
    },
    {
      "epoch": 0.5193985911050363,
      "grad_norm": 0.5992648601531982,
      "learning_rate": 2.4040935057652543e-05,
      "loss": 2.2078,
      "step": 148200
    },
    {
      "epoch": 0.5196088739354432,
      "grad_norm": 0.5487757921218872,
      "learning_rate": 2.40304209161322e-05,
      "loss": 2.2259,
      "step": 148260
    },
    {
      "epoch": 0.5198191567658501,
      "grad_norm": 0.5373034477233887,
      "learning_rate": 2.4019906774611854e-05,
      "loss": 2.219,
      "step": 148320
    },
    {
      "epoch": 0.5200294395962569,
      "grad_norm": 0.502025306224823,
      "learning_rate": 2.400939263309151e-05,
      "loss": 2.2192,
      "step": 148380
    },
    {
      "epoch": 0.5202397224266638,
      "grad_norm": 0.5354958176612854,
      "learning_rate": 2.3998878491571164e-05,
      "loss": 2.2023,
      "step": 148440
    },
    {
      "epoch": 0.5204500052570707,
      "grad_norm": 0.5596054196357727,
      "learning_rate": 2.398836435005082e-05,
      "loss": 2.2085,
      "step": 148500
    },
    {
      "epoch": 0.5206602880874777,
      "grad_norm": 0.5417528748512268,
      "learning_rate": 2.3977850208530475e-05,
      "loss": 2.2219,
      "step": 148560
    },
    {
      "epoch": 0.5208705709178846,
      "grad_norm": 0.636630117893219,
      "learning_rate": 2.396733606701013e-05,
      "loss": 2.2194,
      "step": 148620
    },
    {
      "epoch": 0.5210808537482915,
      "grad_norm": 0.4849955439567566,
      "learning_rate": 2.3956821925489782e-05,
      "loss": 2.2215,
      "step": 148680
    },
    {
      "epoch": 0.5212911365786983,
      "grad_norm": 0.5673374533653259,
      "learning_rate": 2.394630778396944e-05,
      "loss": 2.2132,
      "step": 148740
    },
    {
      "epoch": 0.5215014194091052,
      "grad_norm": 0.563437283039093,
      "learning_rate": 2.3935793642449096e-05,
      "loss": 2.1998,
      "step": 148800
    },
    {
      "epoch": 0.5217117022395121,
      "grad_norm": 0.5512784123420715,
      "learning_rate": 2.392527950092875e-05,
      "loss": 2.212,
      "step": 148860
    },
    {
      "epoch": 0.521921985069919,
      "grad_norm": 0.5081817507743835,
      "learning_rate": 2.3914765359408407e-05,
      "loss": 2.2078,
      "step": 148920
    },
    {
      "epoch": 0.522132267900326,
      "grad_norm": 0.5619904398918152,
      "learning_rate": 2.390425121788806e-05,
      "loss": 2.2137,
      "step": 148980
    },
    {
      "epoch": 0.5223425507307329,
      "grad_norm": 0.5273348093032837,
      "learning_rate": 2.3893737076367717e-05,
      "loss": 2.2204,
      "step": 149040
    },
    {
      "epoch": 0.5225528335611397,
      "grad_norm": 0.5501142740249634,
      "learning_rate": 2.3883222934847372e-05,
      "loss": 2.2188,
      "step": 149100
    },
    {
      "epoch": 0.5227631163915466,
      "grad_norm": 0.5600405335426331,
      "learning_rate": 2.3872708793327024e-05,
      "loss": 2.2064,
      "step": 149160
    },
    {
      "epoch": 0.5229733992219535,
      "grad_norm": 0.5650786757469177,
      "learning_rate": 2.3862194651806683e-05,
      "loss": 2.2123,
      "step": 149220
    },
    {
      "epoch": 0.5231836820523604,
      "grad_norm": 0.49991318583488464,
      "learning_rate": 2.3851680510286338e-05,
      "loss": 2.2215,
      "step": 149280
    },
    {
      "epoch": 0.5233939648827673,
      "grad_norm": 0.5382364392280579,
      "learning_rate": 2.384116636876599e-05,
      "loss": 2.2115,
      "step": 149340
    },
    {
      "epoch": 0.5236042477131743,
      "grad_norm": 0.5215093493461609,
      "learning_rate": 2.383065222724565e-05,
      "loss": 2.2176,
      "step": 149400
    },
    {
      "epoch": 0.5238145305435811,
      "grad_norm": 0.5873035192489624,
      "learning_rate": 2.38201380857253e-05,
      "loss": 2.2033,
      "step": 149460
    },
    {
      "epoch": 0.524024813373988,
      "grad_norm": 0.5551034808158875,
      "learning_rate": 2.3809623944204956e-05,
      "loss": 2.2174,
      "step": 149520
    },
    {
      "epoch": 0.5242350962043949,
      "grad_norm": 0.5522902607917786,
      "learning_rate": 2.3799109802684615e-05,
      "loss": 2.202,
      "step": 149580
    },
    {
      "epoch": 0.5244453790348018,
      "grad_norm": 0.6057192087173462,
      "learning_rate": 2.3788595661164267e-05,
      "loss": 2.2006,
      "step": 149640
    },
    {
      "epoch": 0.5246556618652087,
      "grad_norm": 0.5531818270683289,
      "learning_rate": 2.3778081519643922e-05,
      "loss": 2.2174,
      "step": 149700
    },
    {
      "epoch": 0.5248659446956156,
      "grad_norm": 0.5004293322563171,
      "learning_rate": 2.3767567378123577e-05,
      "loss": 2.2027,
      "step": 149760
    },
    {
      "epoch": 0.5250762275260225,
      "grad_norm": 0.5355766415596008,
      "learning_rate": 2.3757053236603232e-05,
      "loss": 2.2172,
      "step": 149820
    },
    {
      "epoch": 0.5252865103564294,
      "grad_norm": 0.5332327485084534,
      "learning_rate": 2.3746539095082888e-05,
      "loss": 2.2174,
      "step": 149880
    },
    {
      "epoch": 0.5254967931868363,
      "grad_norm": 0.5604259371757507,
      "learning_rate": 2.3736024953562543e-05,
      "loss": 2.2165,
      "step": 149940
    },
    {
      "epoch": 0.5257070760172432,
      "grad_norm": 0.5731934905052185,
      "learning_rate": 2.3725510812042198e-05,
      "loss": 2.2177,
      "step": 150000
    },
    {
      "epoch": 0.5259173588476501,
      "grad_norm": 0.5286061763763428,
      "learning_rate": 2.3714996670521854e-05,
      "loss": 2.1979,
      "step": 150060
    },
    {
      "epoch": 0.526127641678057,
      "grad_norm": 0.5667920112609863,
      "learning_rate": 2.370448252900151e-05,
      "loss": 2.2066,
      "step": 150120
    },
    {
      "epoch": 0.5263379245084638,
      "grad_norm": 0.5224534869194031,
      "learning_rate": 2.3693968387481164e-05,
      "loss": 2.199,
      "step": 150180
    },
    {
      "epoch": 0.5265482073388708,
      "grad_norm": 0.5209370851516724,
      "learning_rate": 2.368345424596082e-05,
      "loss": 2.2104,
      "step": 150240
    },
    {
      "epoch": 0.5267584901692777,
      "grad_norm": 0.4856961965560913,
      "learning_rate": 2.3672940104440475e-05,
      "loss": 2.2186,
      "step": 150300
    },
    {
      "epoch": 0.5269687729996846,
      "grad_norm": 0.5362430810928345,
      "learning_rate": 2.366242596292013e-05,
      "loss": 2.2201,
      "step": 150360
    },
    {
      "epoch": 0.5271790558300915,
      "grad_norm": 0.5632970929145813,
      "learning_rate": 2.3651911821399782e-05,
      "loss": 2.2002,
      "step": 150420
    },
    {
      "epoch": 0.5273893386604984,
      "grad_norm": 0.5169422626495361,
      "learning_rate": 2.364139767987944e-05,
      "loss": 2.2105,
      "step": 150480
    },
    {
      "epoch": 0.5275996214909052,
      "grad_norm": 0.5175157785415649,
      "learning_rate": 2.3630883538359096e-05,
      "loss": 2.2193,
      "step": 150540
    },
    {
      "epoch": 0.5278099043213121,
      "grad_norm": 0.5866751074790955,
      "learning_rate": 2.3620544632530756e-05,
      "loss": 2.2127,
      "step": 150600
    },
    {
      "epoch": 0.5280201871517191,
      "grad_norm": 0.5538196563720703,
      "learning_rate": 2.3610030491010408e-05,
      "loss": 2.2139,
      "step": 150660
    },
    {
      "epoch": 0.528230469982126,
      "grad_norm": 0.5771428942680359,
      "learning_rate": 2.3599516349490063e-05,
      "loss": 2.2089,
      "step": 150720
    },
    {
      "epoch": 0.5284407528125329,
      "grad_norm": 0.5791342258453369,
      "learning_rate": 2.3589002207969722e-05,
      "loss": 2.2157,
      "step": 150780
    },
    {
      "epoch": 0.5286510356429398,
      "grad_norm": 0.5100274085998535,
      "learning_rate": 2.3578488066449374e-05,
      "loss": 2.2075,
      "step": 150840
    },
    {
      "epoch": 0.5288613184733466,
      "grad_norm": 0.5530140399932861,
      "learning_rate": 2.356797392492903e-05,
      "loss": 2.2152,
      "step": 150900
    },
    {
      "epoch": 0.5290716013037535,
      "grad_norm": 0.590545654296875,
      "learning_rate": 2.3557459783408688e-05,
      "loss": 2.2085,
      "step": 150960
    },
    {
      "epoch": 0.5292818841341604,
      "grad_norm": 0.5271270275115967,
      "learning_rate": 2.354694564188834e-05,
      "loss": 2.2225,
      "step": 151020
    },
    {
      "epoch": 0.5294921669645674,
      "grad_norm": 0.5703055262565613,
      "learning_rate": 2.3536431500367995e-05,
      "loss": 2.228,
      "step": 151080
    },
    {
      "epoch": 0.5297024497949743,
      "grad_norm": 0.5086321830749512,
      "learning_rate": 2.352591735884765e-05,
      "loss": 2.2178,
      "step": 151140
    },
    {
      "epoch": 0.5299127326253812,
      "grad_norm": 0.5059917569160461,
      "learning_rate": 2.3515403217327306e-05,
      "loss": 2.2016,
      "step": 151200
    },
    {
      "epoch": 0.530123015455788,
      "grad_norm": 0.5583618879318237,
      "learning_rate": 2.3504889075806964e-05,
      "loss": 2.229,
      "step": 151260
    },
    {
      "epoch": 0.5303332982861949,
      "grad_norm": 0.5437110662460327,
      "learning_rate": 2.3494374934286616e-05,
      "loss": 2.1988,
      "step": 151320
    },
    {
      "epoch": 0.5305435811166018,
      "grad_norm": 0.5776737332344055,
      "learning_rate": 2.348386079276627e-05,
      "loss": 2.2158,
      "step": 151380
    },
    {
      "epoch": 0.5307538639470087,
      "grad_norm": 0.5363079309463501,
      "learning_rate": 2.3473346651245927e-05,
      "loss": 2.2242,
      "step": 151440
    },
    {
      "epoch": 0.5309641467774157,
      "grad_norm": 0.5857604742050171,
      "learning_rate": 2.3462832509725582e-05,
      "loss": 2.214,
      "step": 151500
    },
    {
      "epoch": 0.5311744296078226,
      "grad_norm": 0.5153557062149048,
      "learning_rate": 2.3452318368205237e-05,
      "loss": 2.2025,
      "step": 151560
    },
    {
      "epoch": 0.5313847124382294,
      "grad_norm": 0.6134850978851318,
      "learning_rate": 2.3441804226684893e-05,
      "loss": 2.2091,
      "step": 151620
    },
    {
      "epoch": 0.5315949952686363,
      "grad_norm": 0.5441007018089294,
      "learning_rate": 2.3431290085164548e-05,
      "loss": 2.2208,
      "step": 151680
    },
    {
      "epoch": 0.5318052780990432,
      "grad_norm": 0.565891444683075,
      "learning_rate": 2.3420775943644203e-05,
      "loss": 2.2169,
      "step": 151740
    },
    {
      "epoch": 0.5320155609294501,
      "grad_norm": 0.5420317053794861,
      "learning_rate": 2.341026180212386e-05,
      "loss": 2.2207,
      "step": 151800
    },
    {
      "epoch": 0.532225843759857,
      "grad_norm": 0.5484797954559326,
      "learning_rate": 2.3399747660603514e-05,
      "loss": 2.2072,
      "step": 151860
    },
    {
      "epoch": 0.532436126590264,
      "grad_norm": 0.559556245803833,
      "learning_rate": 2.338923351908317e-05,
      "loss": 2.2114,
      "step": 151920
    },
    {
      "epoch": 0.5326464094206708,
      "grad_norm": 0.5181391835212708,
      "learning_rate": 2.3378719377562824e-05,
      "loss": 2.2165,
      "step": 151980
    },
    {
      "epoch": 0.5328566922510777,
      "grad_norm": 0.5498536825180054,
      "learning_rate": 2.336820523604248e-05,
      "loss": 2.2114,
      "step": 152040
    },
    {
      "epoch": 0.5330669750814846,
      "grad_norm": 0.559701144695282,
      "learning_rate": 2.335769109452213e-05,
      "loss": 2.2195,
      "step": 152100
    },
    {
      "epoch": 0.5332772579118915,
      "grad_norm": 0.5087351202964783,
      "learning_rate": 2.334717695300179e-05,
      "loss": 2.2114,
      "step": 152160
    },
    {
      "epoch": 0.5334875407422984,
      "grad_norm": 0.590671718120575,
      "learning_rate": 2.3336662811481445e-05,
      "loss": 2.2161,
      "step": 152220
    },
    {
      "epoch": 0.5336978235727053,
      "grad_norm": 0.5296396017074585,
      "learning_rate": 2.3326148669961097e-05,
      "loss": 2.2173,
      "step": 152280
    },
    {
      "epoch": 0.5339081064031121,
      "grad_norm": 0.4890645742416382,
      "learning_rate": 2.3315634528440756e-05,
      "loss": 2.2094,
      "step": 152340
    },
    {
      "epoch": 0.5341183892335191,
      "grad_norm": 0.5311561822891235,
      "learning_rate": 2.3305120386920408e-05,
      "loss": 2.2167,
      "step": 152400
    },
    {
      "epoch": 0.534328672063926,
      "grad_norm": 0.5763415694236755,
      "learning_rate": 2.3294606245400063e-05,
      "loss": 2.2274,
      "step": 152460
    },
    {
      "epoch": 0.5345389548943329,
      "grad_norm": 0.5071849822998047,
      "learning_rate": 2.3284092103879722e-05,
      "loss": 2.1984,
      "step": 152520
    },
    {
      "epoch": 0.5347492377247398,
      "grad_norm": 0.5014827847480774,
      "learning_rate": 2.3273577962359374e-05,
      "loss": 2.228,
      "step": 152580
    },
    {
      "epoch": 0.5349595205551466,
      "grad_norm": 0.562168538570404,
      "learning_rate": 2.326306382083903e-05,
      "loss": 2.1957,
      "step": 152640
    },
    {
      "epoch": 0.5351698033855535,
      "grad_norm": 0.5440489649772644,
      "learning_rate": 2.3252549679318684e-05,
      "loss": 2.2289,
      "step": 152700
    },
    {
      "epoch": 0.5353800862159604,
      "grad_norm": 0.5407817959785461,
      "learning_rate": 2.324203553779834e-05,
      "loss": 2.2165,
      "step": 152760
    },
    {
      "epoch": 0.5355903690463674,
      "grad_norm": 0.5259541869163513,
      "learning_rate": 2.3231521396277995e-05,
      "loss": 2.2268,
      "step": 152820
    },
    {
      "epoch": 0.5358006518767743,
      "grad_norm": 0.5506186485290527,
      "learning_rate": 2.322100725475765e-05,
      "loss": 2.2008,
      "step": 152880
    },
    {
      "epoch": 0.5360109347071812,
      "grad_norm": 0.5958437323570251,
      "learning_rate": 2.3210493113237305e-05,
      "loss": 2.2186,
      "step": 152940
    },
    {
      "epoch": 0.536221217537588,
      "grad_norm": 0.5881892442703247,
      "learning_rate": 2.319997897171696e-05,
      "loss": 2.224,
      "step": 153000
    },
    {
      "epoch": 0.5364315003679949,
      "grad_norm": 0.5412174463272095,
      "learning_rate": 2.3189464830196616e-05,
      "loss": 2.2077,
      "step": 153060
    },
    {
      "epoch": 0.5366417831984018,
      "grad_norm": 0.5946020483970642,
      "learning_rate": 2.317895068867627e-05,
      "loss": 2.2145,
      "step": 153120
    },
    {
      "epoch": 0.5368520660288087,
      "grad_norm": 0.5380927920341492,
      "learning_rate": 2.3168436547155927e-05,
      "loss": 2.2029,
      "step": 153180
    },
    {
      "epoch": 0.5370623488592157,
      "grad_norm": 0.5236719250679016,
      "learning_rate": 2.3157922405635582e-05,
      "loss": 2.22,
      "step": 153240
    },
    {
      "epoch": 0.5372726316896226,
      "grad_norm": 0.6787263751029968,
      "learning_rate": 2.3147408264115237e-05,
      "loss": 2.2274,
      "step": 153300
    },
    {
      "epoch": 0.5374829145200294,
      "grad_norm": 0.5436735153198242,
      "learning_rate": 2.3137069358286898e-05,
      "loss": 2.2076,
      "step": 153360
    },
    {
      "epoch": 0.5376931973504363,
      "grad_norm": 0.5597900748252869,
      "learning_rate": 2.3126555216766553e-05,
      "loss": 2.2129,
      "step": 153420
    },
    {
      "epoch": 0.5379034801808432,
      "grad_norm": 0.5178171992301941,
      "learning_rate": 2.3116041075246205e-05,
      "loss": 2.215,
      "step": 153480
    },
    {
      "epoch": 0.5381137630112501,
      "grad_norm": 0.5543037056922913,
      "learning_rate": 2.3105526933725863e-05,
      "loss": 2.2151,
      "step": 153540
    },
    {
      "epoch": 0.538324045841657,
      "grad_norm": 0.6140149235725403,
      "learning_rate": 2.3095012792205515e-05,
      "loss": 2.1995,
      "step": 153600
    },
    {
      "epoch": 0.538534328672064,
      "grad_norm": 0.5255257487297058,
      "learning_rate": 2.3084498650685174e-05,
      "loss": 2.225,
      "step": 153660
    },
    {
      "epoch": 0.5387446115024708,
      "grad_norm": 0.5231916904449463,
      "learning_rate": 2.307398450916483e-05,
      "loss": 2.2208,
      "step": 153720
    },
    {
      "epoch": 0.5389548943328777,
      "grad_norm": 0.5627821087837219,
      "learning_rate": 2.306347036764448e-05,
      "loss": 2.2165,
      "step": 153780
    },
    {
      "epoch": 0.5391651771632846,
      "grad_norm": 0.5242416262626648,
      "learning_rate": 2.305295622612414e-05,
      "loss": 2.1959,
      "step": 153840
    },
    {
      "epoch": 0.5393754599936915,
      "grad_norm": 0.5642799735069275,
      "learning_rate": 2.3042442084603795e-05,
      "loss": 2.2203,
      "step": 153900
    },
    {
      "epoch": 0.5395857428240984,
      "grad_norm": 0.5174145698547363,
      "learning_rate": 2.3031927943083447e-05,
      "loss": 2.2069,
      "step": 153960
    },
    {
      "epoch": 0.5397960256545054,
      "grad_norm": 0.5788621306419373,
      "learning_rate": 2.3021413801563106e-05,
      "loss": 2.2111,
      "step": 154020
    },
    {
      "epoch": 0.5400063084849122,
      "grad_norm": 0.5217883586883545,
      "learning_rate": 2.3010899660042758e-05,
      "loss": 2.2185,
      "step": 154080
    },
    {
      "epoch": 0.5402165913153191,
      "grad_norm": 0.5840219855308533,
      "learning_rate": 2.3000385518522413e-05,
      "loss": 2.2113,
      "step": 154140
    },
    {
      "epoch": 0.540426874145726,
      "grad_norm": 0.5109094977378845,
      "learning_rate": 2.298987137700207e-05,
      "loss": 2.2095,
      "step": 154200
    },
    {
      "epoch": 0.5406371569761329,
      "grad_norm": 0.5547183156013489,
      "learning_rate": 2.2979357235481723e-05,
      "loss": 2.1991,
      "step": 154260
    },
    {
      "epoch": 0.5408474398065398,
      "grad_norm": 0.5175747275352478,
      "learning_rate": 2.296884309396138e-05,
      "loss": 2.2171,
      "step": 154320
    },
    {
      "epoch": 0.5410577226369467,
      "grad_norm": 0.5823888778686523,
      "learning_rate": 2.2958328952441034e-05,
      "loss": 2.2003,
      "step": 154380
    },
    {
      "epoch": 0.5412680054673535,
      "grad_norm": 0.5631999969482422,
      "learning_rate": 2.294781481092069e-05,
      "loss": 2.2135,
      "step": 154440
    },
    {
      "epoch": 0.5414782882977605,
      "grad_norm": 0.7047464847564697,
      "learning_rate": 2.2937300669400345e-05,
      "loss": 2.2136,
      "step": 154500
    },
    {
      "epoch": 0.5416885711281674,
      "grad_norm": 0.5780332088470459,
      "learning_rate": 2.292678652788e-05,
      "loss": 2.2104,
      "step": 154560
    },
    {
      "epoch": 0.5418988539585743,
      "grad_norm": 0.5036734938621521,
      "learning_rate": 2.2916272386359655e-05,
      "loss": 2.2023,
      "step": 154620
    },
    {
      "epoch": 0.5421091367889812,
      "grad_norm": 0.5215955972671509,
      "learning_rate": 2.290575824483931e-05,
      "loss": 2.196,
      "step": 154680
    },
    {
      "epoch": 0.5423194196193881,
      "grad_norm": 0.5139001607894897,
      "learning_rate": 2.2895244103318966e-05,
      "loss": 2.2188,
      "step": 154740
    },
    {
      "epoch": 0.5425297024497949,
      "grad_norm": 0.5407582521438599,
      "learning_rate": 2.288472996179862e-05,
      "loss": 2.2033,
      "step": 154800
    },
    {
      "epoch": 0.5427399852802018,
      "grad_norm": 0.6017780900001526,
      "learning_rate": 2.2874215820278276e-05,
      "loss": 2.2156,
      "step": 154860
    },
    {
      "epoch": 0.5429502681106088,
      "grad_norm": 0.6027617454528809,
      "learning_rate": 2.286370167875793e-05,
      "loss": 2.2104,
      "step": 154920
    },
    {
      "epoch": 0.5431605509410157,
      "grad_norm": 0.5565388798713684,
      "learning_rate": 2.2853187537237587e-05,
      "loss": 2.2037,
      "step": 154980
    },
    {
      "epoch": 0.5433708337714226,
      "grad_norm": 0.5749856233596802,
      "learning_rate": 2.284267339571724e-05,
      "loss": 2.2239,
      "step": 155040
    },
    {
      "epoch": 0.5435811166018295,
      "grad_norm": 0.5343650579452515,
      "learning_rate": 2.2832159254196897e-05,
      "loss": 2.1944,
      "step": 155100
    },
    {
      "epoch": 0.5437913994322363,
      "grad_norm": 0.5500574707984924,
      "learning_rate": 2.2821645112676553e-05,
      "loss": 2.2129,
      "step": 155160
    },
    {
      "epoch": 0.5440016822626432,
      "grad_norm": 0.5807549953460693,
      "learning_rate": 2.2811130971156205e-05,
      "loss": 2.2116,
      "step": 155220
    },
    {
      "epoch": 0.5442119650930501,
      "grad_norm": 0.5436511635780334,
      "learning_rate": 2.2800616829635863e-05,
      "loss": 2.2123,
      "step": 155280
    },
    {
      "epoch": 0.5444222479234571,
      "grad_norm": 0.5555745363235474,
      "learning_rate": 2.2790102688115515e-05,
      "loss": 2.2056,
      "step": 155340
    },
    {
      "epoch": 0.544632530753864,
      "grad_norm": 0.5600705742835999,
      "learning_rate": 2.277958854659517e-05,
      "loss": 2.2064,
      "step": 155400
    },
    {
      "epoch": 0.5448428135842709,
      "grad_norm": 0.5020048022270203,
      "learning_rate": 2.276907440507483e-05,
      "loss": 2.2199,
      "step": 155460
    },
    {
      "epoch": 0.5450530964146777,
      "grad_norm": 0.5523339509963989,
      "learning_rate": 2.275856026355448e-05,
      "loss": 2.2202,
      "step": 155520
    },
    {
      "epoch": 0.5452633792450846,
      "grad_norm": 0.6014343500137329,
      "learning_rate": 2.2748046122034136e-05,
      "loss": 2.2138,
      "step": 155580
    },
    {
      "epoch": 0.5454736620754915,
      "grad_norm": 0.5882730484008789,
      "learning_rate": 2.273753198051379e-05,
      "loss": 2.2182,
      "step": 155640
    },
    {
      "epoch": 0.5456839449058984,
      "grad_norm": 0.5294257998466492,
      "learning_rate": 2.2727017838993447e-05,
      "loss": 2.2024,
      "step": 155700
    },
    {
      "epoch": 0.5458942277363054,
      "grad_norm": 0.4940982460975647,
      "learning_rate": 2.2716503697473102e-05,
      "loss": 2.2176,
      "step": 155760
    },
    {
      "epoch": 0.5461045105667123,
      "grad_norm": 0.6698052883148193,
      "learning_rate": 2.2705989555952757e-05,
      "loss": 2.2183,
      "step": 155820
    },
    {
      "epoch": 0.5463147933971191,
      "grad_norm": 0.5630810260772705,
      "learning_rate": 2.2695475414432413e-05,
      "loss": 2.2259,
      "step": 155880
    },
    {
      "epoch": 0.546525076227526,
      "grad_norm": 0.6864802241325378,
      "learning_rate": 2.2684961272912068e-05,
      "loss": 2.2025,
      "step": 155940
    },
    {
      "epoch": 0.5467353590579329,
      "grad_norm": 0.5602729320526123,
      "learning_rate": 2.2674447131391723e-05,
      "loss": 2.2127,
      "step": 156000
    },
    {
      "epoch": 0.5469456418883398,
      "grad_norm": 0.5600816607475281,
      "learning_rate": 2.266393298987138e-05,
      "loss": 2.2099,
      "step": 156060
    },
    {
      "epoch": 0.5471559247187467,
      "grad_norm": 0.5645787715911865,
      "learning_rate": 2.265359408404304e-05,
      "loss": 2.2127,
      "step": 156120
    },
    {
      "epoch": 0.5473662075491537,
      "grad_norm": 0.59485924243927,
      "learning_rate": 2.2643079942522694e-05,
      "loss": 2.2066,
      "step": 156180
    },
    {
      "epoch": 0.5475764903795605,
      "grad_norm": 0.5252460837364197,
      "learning_rate": 2.263256580100235e-05,
      "loss": 2.2002,
      "step": 156240
    },
    {
      "epoch": 0.5477867732099674,
      "grad_norm": 0.5992864966392517,
      "learning_rate": 2.2622051659482005e-05,
      "loss": 2.218,
      "step": 156300
    },
    {
      "epoch": 0.5479970560403743,
      "grad_norm": 0.5336859226226807,
      "learning_rate": 2.261153751796166e-05,
      "loss": 2.2217,
      "step": 156360
    },
    {
      "epoch": 0.5482073388707812,
      "grad_norm": 0.5347955226898193,
      "learning_rate": 2.2601023376441315e-05,
      "loss": 2.2083,
      "step": 156420
    },
    {
      "epoch": 0.5484176217011881,
      "grad_norm": 0.5723451972007751,
      "learning_rate": 2.259050923492097e-05,
      "loss": 2.2145,
      "step": 156480
    },
    {
      "epoch": 0.548627904531595,
      "grad_norm": 0.5852446556091309,
      "learning_rate": 2.2579995093400626e-05,
      "loss": 2.2054,
      "step": 156540
    },
    {
      "epoch": 0.5488381873620019,
      "grad_norm": 0.5354224443435669,
      "learning_rate": 2.256948095188028e-05,
      "loss": 2.199,
      "step": 156600
    },
    {
      "epoch": 0.5490484701924088,
      "grad_norm": 0.5295231938362122,
      "learning_rate": 2.2558966810359937e-05,
      "loss": 2.2171,
      "step": 156660
    },
    {
      "epoch": 0.5492587530228157,
      "grad_norm": 0.5262629985809326,
      "learning_rate": 2.254845266883959e-05,
      "loss": 2.2272,
      "step": 156720
    },
    {
      "epoch": 0.5494690358532226,
      "grad_norm": 0.5583018660545349,
      "learning_rate": 2.2537938527319247e-05,
      "loss": 2.2085,
      "step": 156780
    },
    {
      "epoch": 0.5496793186836295,
      "grad_norm": 0.6145007610321045,
      "learning_rate": 2.2527424385798902e-05,
      "loss": 2.1992,
      "step": 156840
    },
    {
      "epoch": 0.5498896015140364,
      "grad_norm": 0.4931606650352478,
      "learning_rate": 2.2516910244278554e-05,
      "loss": 2.2027,
      "step": 156900
    },
    {
      "epoch": 0.5500998843444432,
      "grad_norm": 0.568114697933197,
      "learning_rate": 2.2506396102758213e-05,
      "loss": 2.2207,
      "step": 156960
    },
    {
      "epoch": 0.5503101671748502,
      "grad_norm": 0.5709725022315979,
      "learning_rate": 2.2495881961237865e-05,
      "loss": 2.2085,
      "step": 157020
    },
    {
      "epoch": 0.5505204500052571,
      "grad_norm": 0.5881896615028381,
      "learning_rate": 2.248536781971752e-05,
      "loss": 2.2058,
      "step": 157080
    },
    {
      "epoch": 0.550730732835664,
      "grad_norm": 0.5648079514503479,
      "learning_rate": 2.247485367819718e-05,
      "loss": 2.2209,
      "step": 157140
    },
    {
      "epoch": 0.5509410156660709,
      "grad_norm": 0.6545466184616089,
      "learning_rate": 2.246433953667683e-05,
      "loss": 2.2135,
      "step": 157200
    },
    {
      "epoch": 0.5511512984964778,
      "grad_norm": 0.5051990151405334,
      "learning_rate": 2.2453825395156486e-05,
      "loss": 2.1934,
      "step": 157260
    },
    {
      "epoch": 0.5513615813268846,
      "grad_norm": 0.5131949186325073,
      "learning_rate": 2.244331125363614e-05,
      "loss": 2.2062,
      "step": 157320
    },
    {
      "epoch": 0.5515718641572915,
      "grad_norm": 0.5909935832023621,
      "learning_rate": 2.2432797112115796e-05,
      "loss": 2.2123,
      "step": 157380
    },
    {
      "epoch": 0.5517821469876985,
      "grad_norm": 0.5474815368652344,
      "learning_rate": 2.2422282970595452e-05,
      "loss": 2.201,
      "step": 157440
    },
    {
      "epoch": 0.5519924298181054,
      "grad_norm": 0.5805018544197083,
      "learning_rate": 2.2411768829075107e-05,
      "loss": 2.2214,
      "step": 157500
    },
    {
      "epoch": 0.5522027126485123,
      "grad_norm": 0.6290664672851562,
      "learning_rate": 2.2401254687554762e-05,
      "loss": 2.2116,
      "step": 157560
    },
    {
      "epoch": 0.5524129954789192,
      "grad_norm": 0.5544005036354065,
      "learning_rate": 2.2390740546034418e-05,
      "loss": 2.2184,
      "step": 157620
    },
    {
      "epoch": 0.552623278309326,
      "grad_norm": 0.591930627822876,
      "learning_rate": 2.2380226404514073e-05,
      "loss": 2.2026,
      "step": 157680
    },
    {
      "epoch": 0.5528335611397329,
      "grad_norm": 0.5133166909217834,
      "learning_rate": 2.2369712262993728e-05,
      "loss": 2.2067,
      "step": 157740
    },
    {
      "epoch": 0.5530438439701398,
      "grad_norm": 0.5512820482254028,
      "learning_rate": 2.2359198121473383e-05,
      "loss": 2.2148,
      "step": 157800
    },
    {
      "epoch": 0.5532541268005468,
      "grad_norm": 0.5863820314407349,
      "learning_rate": 2.234868397995304e-05,
      "loss": 2.2151,
      "step": 157860
    },
    {
      "epoch": 0.5534644096309537,
      "grad_norm": 0.510724663734436,
      "learning_rate": 2.2338169838432694e-05,
      "loss": 2.2123,
      "step": 157920
    },
    {
      "epoch": 0.5536746924613606,
      "grad_norm": 0.522894024848938,
      "learning_rate": 2.2327655696912346e-05,
      "loss": 2.2078,
      "step": 157980
    },
    {
      "epoch": 0.5538849752917674,
      "grad_norm": 0.5410124659538269,
      "learning_rate": 2.2317141555392005e-05,
      "loss": 2.2038,
      "step": 158040
    },
    {
      "epoch": 0.5540952581221743,
      "grad_norm": 0.5146405696868896,
      "learning_rate": 2.230662741387166e-05,
      "loss": 2.1995,
      "step": 158100
    },
    {
      "epoch": 0.5543055409525812,
      "grad_norm": 0.5585925579071045,
      "learning_rate": 2.2296113272351312e-05,
      "loss": 2.219,
      "step": 158160
    },
    {
      "epoch": 0.5545158237829881,
      "grad_norm": 0.530822217464447,
      "learning_rate": 2.228559913083097e-05,
      "loss": 2.2088,
      "step": 158220
    },
    {
      "epoch": 0.554726106613395,
      "grad_norm": 0.5129376649856567,
      "learning_rate": 2.2275084989310622e-05,
      "loss": 2.2021,
      "step": 158280
    },
    {
      "epoch": 0.5549363894438019,
      "grad_norm": 0.5571216344833374,
      "learning_rate": 2.2264570847790278e-05,
      "loss": 2.2075,
      "step": 158340
    },
    {
      "epoch": 0.5551466722742088,
      "grad_norm": 0.5772547125816345,
      "learning_rate": 2.2254231941961938e-05,
      "loss": 2.2175,
      "step": 158400
    },
    {
      "epoch": 0.5553569551046157,
      "grad_norm": 0.508144199848175,
      "learning_rate": 2.2243717800441597e-05,
      "loss": 2.2162,
      "step": 158460
    },
    {
      "epoch": 0.5555672379350226,
      "grad_norm": 0.6046881079673767,
      "learning_rate": 2.2233203658921252e-05,
      "loss": 2.1968,
      "step": 158520
    },
    {
      "epoch": 0.5557775207654295,
      "grad_norm": 0.530588686466217,
      "learning_rate": 2.2222689517400904e-05,
      "loss": 2.2059,
      "step": 158580
    },
    {
      "epoch": 0.5559878035958364,
      "grad_norm": 0.5466212034225464,
      "learning_rate": 2.2212175375880563e-05,
      "loss": 2.2118,
      "step": 158640
    },
    {
      "epoch": 0.5561980864262432,
      "grad_norm": 0.5684359073638916,
      "learning_rate": 2.2201661234360214e-05,
      "loss": 2.2001,
      "step": 158700
    },
    {
      "epoch": 0.5564083692566502,
      "grad_norm": 0.5453093647956848,
      "learning_rate": 2.219114709283987e-05,
      "loss": 2.1946,
      "step": 158760
    },
    {
      "epoch": 0.5566186520870571,
      "grad_norm": 0.5468719601631165,
      "learning_rate": 2.218063295131953e-05,
      "loss": 2.2095,
      "step": 158820
    },
    {
      "epoch": 0.556828934917464,
      "grad_norm": 0.5114817023277283,
      "learning_rate": 2.217011880979918e-05,
      "loss": 2.204,
      "step": 158880
    },
    {
      "epoch": 0.5570392177478709,
      "grad_norm": 0.5197003483772278,
      "learning_rate": 2.2159604668278836e-05,
      "loss": 2.204,
      "step": 158940
    },
    {
      "epoch": 0.5572495005782778,
      "grad_norm": 0.5789101719856262,
      "learning_rate": 2.214909052675849e-05,
      "loss": 2.2304,
      "step": 159000
    },
    {
      "epoch": 0.5574597834086846,
      "grad_norm": 0.5665631890296936,
      "learning_rate": 2.2138576385238146e-05,
      "loss": 2.2046,
      "step": 159060
    },
    {
      "epoch": 0.5576700662390915,
      "grad_norm": 0.5702418684959412,
      "learning_rate": 2.21280622437178e-05,
      "loss": 2.2051,
      "step": 159120
    },
    {
      "epoch": 0.5578803490694985,
      "grad_norm": 0.5206588506698608,
      "learning_rate": 2.2117548102197457e-05,
      "loss": 2.1996,
      "step": 159180
    },
    {
      "epoch": 0.5580906318999054,
      "grad_norm": 0.5394989848136902,
      "learning_rate": 2.2107033960677112e-05,
      "loss": 2.1917,
      "step": 159240
    },
    {
      "epoch": 0.5583009147303123,
      "grad_norm": 0.5701467394828796,
      "learning_rate": 2.2096519819156767e-05,
      "loss": 2.195,
      "step": 159300
    },
    {
      "epoch": 0.5585111975607192,
      "grad_norm": 0.5451265573501587,
      "learning_rate": 2.2086005677636423e-05,
      "loss": 2.2014,
      "step": 159360
    },
    {
      "epoch": 0.558721480391126,
      "grad_norm": 0.5584667325019836,
      "learning_rate": 2.2075491536116078e-05,
      "loss": 2.2138,
      "step": 159420
    },
    {
      "epoch": 0.5589317632215329,
      "grad_norm": 0.5295857787132263,
      "learning_rate": 2.2064977394595733e-05,
      "loss": 2.2191,
      "step": 159480
    },
    {
      "epoch": 0.5591420460519398,
      "grad_norm": 0.678843080997467,
      "learning_rate": 2.205446325307539e-05,
      "loss": 2.2222,
      "step": 159540
    },
    {
      "epoch": 0.5593523288823468,
      "grad_norm": 0.5450401306152344,
      "learning_rate": 2.2043949111555044e-05,
      "loss": 2.2098,
      "step": 159600
    },
    {
      "epoch": 0.5595626117127537,
      "grad_norm": 0.5541589856147766,
      "learning_rate": 2.2033434970034696e-05,
      "loss": 2.2096,
      "step": 159660
    },
    {
      "epoch": 0.5597728945431606,
      "grad_norm": 0.5362442135810852,
      "learning_rate": 2.2022920828514354e-05,
      "loss": 2.1889,
      "step": 159720
    },
    {
      "epoch": 0.5599831773735674,
      "grad_norm": 0.5870858430862427,
      "learning_rate": 2.201240668699401e-05,
      "loss": 2.2146,
      "step": 159780
    },
    {
      "epoch": 0.5601934602039743,
      "grad_norm": 0.5245428085327148,
      "learning_rate": 2.200189254547366e-05,
      "loss": 2.2165,
      "step": 159840
    },
    {
      "epoch": 0.5604037430343812,
      "grad_norm": 0.5324500203132629,
      "learning_rate": 2.199137840395332e-05,
      "loss": 2.2096,
      "step": 159900
    },
    {
      "epoch": 0.5606140258647881,
      "grad_norm": 0.5151851773262024,
      "learning_rate": 2.1980864262432972e-05,
      "loss": 2.2044,
      "step": 159960
    },
    {
      "epoch": 0.5608243086951951,
      "grad_norm": 0.532479465007782,
      "learning_rate": 2.1970350120912627e-05,
      "loss": 2.2112,
      "step": 160020
    },
    {
      "epoch": 0.561034591525602,
      "grad_norm": 0.5969590544700623,
      "learning_rate": 2.1959835979392286e-05,
      "loss": 2.2185,
      "step": 160080
    },
    {
      "epoch": 0.5612448743560088,
      "grad_norm": 0.5450465083122253,
      "learning_rate": 2.1949321837871938e-05,
      "loss": 2.204,
      "step": 160140
    },
    {
      "epoch": 0.5614551571864157,
      "grad_norm": 0.5737215280532837,
      "learning_rate": 2.1938807696351593e-05,
      "loss": 2.2075,
      "step": 160200
    },
    {
      "epoch": 0.5616654400168226,
      "grad_norm": 0.5520704984664917,
      "learning_rate": 2.1928293554831252e-05,
      "loss": 2.1986,
      "step": 160260
    },
    {
      "epoch": 0.5618757228472295,
      "grad_norm": 0.5581754446029663,
      "learning_rate": 2.1917779413310904e-05,
      "loss": 2.2046,
      "step": 160320
    },
    {
      "epoch": 0.5620860056776364,
      "grad_norm": 0.556607186794281,
      "learning_rate": 2.190726527179056e-05,
      "loss": 2.2018,
      "step": 160380
    },
    {
      "epoch": 0.5622962885080434,
      "grad_norm": 0.5431088209152222,
      "learning_rate": 2.1896751130270214e-05,
      "loss": 2.2187,
      "step": 160440
    },
    {
      "epoch": 0.5625065713384502,
      "grad_norm": 0.5602589249610901,
      "learning_rate": 2.188623698874987e-05,
      "loss": 2.2032,
      "step": 160500
    },
    {
      "epoch": 0.5627168541688571,
      "grad_norm": 0.5283489227294922,
      "learning_rate": 2.1875722847229525e-05,
      "loss": 2.1863,
      "step": 160560
    },
    {
      "epoch": 0.562927136999264,
      "grad_norm": 0.6582157015800476,
      "learning_rate": 2.186520870570918e-05,
      "loss": 2.1946,
      "step": 160620
    },
    {
      "epoch": 0.5631374198296709,
      "grad_norm": 0.5504642128944397,
      "learning_rate": 2.1854694564188835e-05,
      "loss": 2.2088,
      "step": 160680
    },
    {
      "epoch": 0.5633477026600778,
      "grad_norm": 0.5475605726242065,
      "learning_rate": 2.184418042266849e-05,
      "loss": 2.2134,
      "step": 160740
    },
    {
      "epoch": 0.5635579854904847,
      "grad_norm": 0.5584597587585449,
      "learning_rate": 2.1833666281148146e-05,
      "loss": 2.2034,
      "step": 160800
    },
    {
      "epoch": 0.5637682683208916,
      "grad_norm": 0.6042028069496155,
      "learning_rate": 2.18231521396278e-05,
      "loss": 2.2178,
      "step": 160860
    },
    {
      "epoch": 0.5639785511512985,
      "grad_norm": 0.5251367092132568,
      "learning_rate": 2.1812637998107453e-05,
      "loss": 2.1982,
      "step": 160920
    },
    {
      "epoch": 0.5641888339817054,
      "grad_norm": 0.518165111541748,
      "learning_rate": 2.1802123856587112e-05,
      "loss": 2.2248,
      "step": 160980
    },
    {
      "epoch": 0.5643991168121123,
      "grad_norm": 0.5064350366592407,
      "learning_rate": 2.1791609715066767e-05,
      "loss": 2.2202,
      "step": 161040
    },
    {
      "epoch": 0.5646093996425192,
      "grad_norm": 0.5454937219619751,
      "learning_rate": 2.178109557354642e-05,
      "loss": 2.2062,
      "step": 161100
    },
    {
      "epoch": 0.5648196824729261,
      "grad_norm": 0.592663049697876,
      "learning_rate": 2.1770581432026078e-05,
      "loss": 2.2157,
      "step": 161160
    },
    {
      "epoch": 0.5650299653033329,
      "grad_norm": 0.5274229049682617,
      "learning_rate": 2.1760067290505733e-05,
      "loss": 2.2065,
      "step": 161220
    },
    {
      "epoch": 0.5652402481337399,
      "grad_norm": 0.5737342834472656,
      "learning_rate": 2.1749553148985385e-05,
      "loss": 2.1895,
      "step": 161280
    },
    {
      "epoch": 0.5654505309641468,
      "grad_norm": 0.5299097895622253,
      "learning_rate": 2.1739039007465043e-05,
      "loss": 2.2206,
      "step": 161340
    },
    {
      "epoch": 0.5656608137945537,
      "grad_norm": 0.547879159450531,
      "learning_rate": 2.1728524865944695e-05,
      "loss": 2.2083,
      "step": 161400
    },
    {
      "epoch": 0.5658710966249606,
      "grad_norm": 0.548787534236908,
      "learning_rate": 2.171801072442435e-05,
      "loss": 2.2084,
      "step": 161460
    },
    {
      "epoch": 0.5660813794553675,
      "grad_norm": 0.5963943600654602,
      "learning_rate": 2.170749658290401e-05,
      "loss": 2.2096,
      "step": 161520
    },
    {
      "epoch": 0.5662916622857743,
      "grad_norm": 0.5302314162254333,
      "learning_rate": 2.169698244138366e-05,
      "loss": 2.2034,
      "step": 161580
    },
    {
      "epoch": 0.5665019451161812,
      "grad_norm": 0.531383216381073,
      "learning_rate": 2.1686468299863316e-05,
      "loss": 2.1979,
      "step": 161640
    },
    {
      "epoch": 0.5667122279465882,
      "grad_norm": 0.5811280012130737,
      "learning_rate": 2.1675954158342972e-05,
      "loss": 2.2139,
      "step": 161700
    },
    {
      "epoch": 0.5669225107769951,
      "grad_norm": 0.5571404695510864,
      "learning_rate": 2.1665440016822627e-05,
      "loss": 2.21,
      "step": 161760
    },
    {
      "epoch": 0.567132793607402,
      "grad_norm": 0.5254374742507935,
      "learning_rate": 2.1654925875302282e-05,
      "loss": 2.2137,
      "step": 161820
    },
    {
      "epoch": 0.5673430764378089,
      "grad_norm": 0.5230144262313843,
      "learning_rate": 2.1644411733781938e-05,
      "loss": 2.227,
      "step": 161880
    },
    {
      "epoch": 0.5675533592682157,
      "grad_norm": 0.524534285068512,
      "learning_rate": 2.1633897592261593e-05,
      "loss": 2.2066,
      "step": 161940
    },
    {
      "epoch": 0.5677636420986226,
      "grad_norm": 0.5795220136642456,
      "learning_rate": 2.1623383450741248e-05,
      "loss": 2.21,
      "step": 162000
    },
    {
      "epoch": 0.5679739249290295,
      "grad_norm": 0.5567732453346252,
      "learning_rate": 2.1612869309220903e-05,
      "loss": 2.2213,
      "step": 162060
    },
    {
      "epoch": 0.5681842077594365,
      "grad_norm": 0.528534471988678,
      "learning_rate": 2.160235516770056e-05,
      "loss": 2.196,
      "step": 162120
    },
    {
      "epoch": 0.5683944905898434,
      "grad_norm": 0.5799056887626648,
      "learning_rate": 2.1591841026180214e-05,
      "loss": 2.2136,
      "step": 162180
    },
    {
      "epoch": 0.5686047734202503,
      "grad_norm": 0.5151017308235168,
      "learning_rate": 2.158132688465987e-05,
      "loss": 2.2089,
      "step": 162240
    },
    {
      "epoch": 0.5688150562506571,
      "grad_norm": 0.6659327745437622,
      "learning_rate": 2.1570812743139524e-05,
      "loss": 2.2154,
      "step": 162300
    },
    {
      "epoch": 0.569025339081064,
      "grad_norm": 0.5421218276023865,
      "learning_rate": 2.1560473837311185e-05,
      "loss": 2.2131,
      "step": 162360
    },
    {
      "epoch": 0.5692356219114709,
      "grad_norm": 0.5917002558708191,
      "learning_rate": 2.154995969579084e-05,
      "loss": 2.2086,
      "step": 162420
    },
    {
      "epoch": 0.5694459047418778,
      "grad_norm": 0.5525832772254944,
      "learning_rate": 2.1539445554270496e-05,
      "loss": 2.2043,
      "step": 162480
    },
    {
      "epoch": 0.5696561875722848,
      "grad_norm": 0.5710530281066895,
      "learning_rate": 2.152893141275015e-05,
      "loss": 2.2084,
      "step": 162540
    },
    {
      "epoch": 0.5698664704026917,
      "grad_norm": 0.5232380032539368,
      "learning_rate": 2.1518417271229803e-05,
      "loss": 2.2021,
      "step": 162600
    },
    {
      "epoch": 0.5700767532330985,
      "grad_norm": 0.5594772696495056,
      "learning_rate": 2.150790312970946e-05,
      "loss": 2.2137,
      "step": 162660
    },
    {
      "epoch": 0.5702870360635054,
      "grad_norm": 0.4988769292831421,
      "learning_rate": 2.1497388988189117e-05,
      "loss": 2.1901,
      "step": 162720
    },
    {
      "epoch": 0.5704973188939123,
      "grad_norm": 0.5252644419670105,
      "learning_rate": 2.148687484666877e-05,
      "loss": 2.2049,
      "step": 162780
    },
    {
      "epoch": 0.5707076017243192,
      "grad_norm": 0.5093538165092468,
      "learning_rate": 2.1476360705148427e-05,
      "loss": 2.1893,
      "step": 162840
    },
    {
      "epoch": 0.5709178845547261,
      "grad_norm": 0.6334115266799927,
      "learning_rate": 2.1465846563628083e-05,
      "loss": 2.2228,
      "step": 162900
    },
    {
      "epoch": 0.5711281673851331,
      "grad_norm": 0.5815603733062744,
      "learning_rate": 2.1455332422107734e-05,
      "loss": 2.2282,
      "step": 162960
    },
    {
      "epoch": 0.5713384502155399,
      "grad_norm": 0.5428621172904968,
      "learning_rate": 2.1444818280587393e-05,
      "loss": 2.195,
      "step": 163020
    },
    {
      "epoch": 0.5715487330459468,
      "grad_norm": 0.5853855013847351,
      "learning_rate": 2.1434304139067045e-05,
      "loss": 2.2075,
      "step": 163080
    },
    {
      "epoch": 0.5717590158763537,
      "grad_norm": 0.5330763459205627,
      "learning_rate": 2.14237899975467e-05,
      "loss": 2.2133,
      "step": 163140
    },
    {
      "epoch": 0.5719692987067606,
      "grad_norm": 0.6086885333061218,
      "learning_rate": 2.141327585602636e-05,
      "loss": 2.2198,
      "step": 163200
    },
    {
      "epoch": 0.5721795815371675,
      "grad_norm": 0.5251525044441223,
      "learning_rate": 2.140276171450601e-05,
      "loss": 2.2149,
      "step": 163260
    },
    {
      "epoch": 0.5723898643675744,
      "grad_norm": 0.5404868721961975,
      "learning_rate": 2.1392247572985666e-05,
      "loss": 2.1866,
      "step": 163320
    },
    {
      "epoch": 0.5726001471979812,
      "grad_norm": 0.5327269434928894,
      "learning_rate": 2.138173343146532e-05,
      "loss": 2.2018,
      "step": 163380
    },
    {
      "epoch": 0.5728104300283882,
      "grad_norm": 0.5136888027191162,
      "learning_rate": 2.1371219289944977e-05,
      "loss": 2.2092,
      "step": 163440
    },
    {
      "epoch": 0.5730207128587951,
      "grad_norm": 0.5119737386703491,
      "learning_rate": 2.1360705148424632e-05,
      "loss": 2.1987,
      "step": 163500
    },
    {
      "epoch": 0.573230995689202,
      "grad_norm": 0.519402027130127,
      "learning_rate": 2.1350191006904287e-05,
      "loss": 2.2073,
      "step": 163560
    },
    {
      "epoch": 0.5734412785196089,
      "grad_norm": 0.5646945238113403,
      "learning_rate": 2.1339676865383943e-05,
      "loss": 2.2054,
      "step": 163620
    },
    {
      "epoch": 0.5736515613500158,
      "grad_norm": 0.5275649428367615,
      "learning_rate": 2.1329162723863598e-05,
      "loss": 2.2091,
      "step": 163680
    },
    {
      "epoch": 0.5738618441804226,
      "grad_norm": 0.5596910715103149,
      "learning_rate": 2.1318648582343253e-05,
      "loss": 2.2094,
      "step": 163740
    },
    {
      "epoch": 0.5740721270108295,
      "grad_norm": 0.5533624887466431,
      "learning_rate": 2.130813444082291e-05,
      "loss": 2.1974,
      "step": 163800
    },
    {
      "epoch": 0.5742824098412365,
      "grad_norm": 0.5841308236122131,
      "learning_rate": 2.1297620299302564e-05,
      "loss": 2.2073,
      "step": 163860
    },
    {
      "epoch": 0.5744926926716434,
      "grad_norm": 0.5480549931526184,
      "learning_rate": 2.128710615778222e-05,
      "loss": 2.2091,
      "step": 163920
    },
    {
      "epoch": 0.5747029755020503,
      "grad_norm": 0.5628250241279602,
      "learning_rate": 2.1276592016261874e-05,
      "loss": 2.1969,
      "step": 163980
    },
    {
      "epoch": 0.5749132583324571,
      "grad_norm": 0.5552694797515869,
      "learning_rate": 2.1266077874741526e-05,
      "loss": 2.2012,
      "step": 164040
    },
    {
      "epoch": 0.575123541162864,
      "grad_norm": 0.5316144824028015,
      "learning_rate": 2.1255563733221185e-05,
      "loss": 2.2024,
      "step": 164100
    },
    {
      "epoch": 0.5753338239932709,
      "grad_norm": 0.5356711745262146,
      "learning_rate": 2.124504959170084e-05,
      "loss": 2.21,
      "step": 164160
    },
    {
      "epoch": 0.5755441068236778,
      "grad_norm": 0.5729750394821167,
      "learning_rate": 2.1234535450180492e-05,
      "loss": 2.2165,
      "step": 164220
    },
    {
      "epoch": 0.5757543896540848,
      "grad_norm": 0.5203530788421631,
      "learning_rate": 2.122402130866015e-05,
      "loss": 2.2064,
      "step": 164280
    },
    {
      "epoch": 0.5759646724844917,
      "grad_norm": 0.5358246564865112,
      "learning_rate": 2.1213507167139802e-05,
      "loss": 2.1968,
      "step": 164340
    },
    {
      "epoch": 0.5761749553148985,
      "grad_norm": 0.5929320454597473,
      "learning_rate": 2.1203168261311466e-05,
      "loss": 2.2166,
      "step": 164400
    },
    {
      "epoch": 0.5763852381453054,
      "grad_norm": 0.5435928106307983,
      "learning_rate": 2.1192654119791118e-05,
      "loss": 2.1965,
      "step": 164460
    },
    {
      "epoch": 0.5765955209757123,
      "grad_norm": 0.6071345210075378,
      "learning_rate": 2.1182139978270777e-05,
      "loss": 2.1928,
      "step": 164520
    },
    {
      "epoch": 0.5768058038061192,
      "grad_norm": 0.5283774137496948,
      "learning_rate": 2.1171625836750432e-05,
      "loss": 2.1927,
      "step": 164580
    },
    {
      "epoch": 0.5770160866365261,
      "grad_norm": 0.5249363780021667,
      "learning_rate": 2.1161111695230084e-05,
      "loss": 2.1854,
      "step": 164640
    },
    {
      "epoch": 0.5772263694669331,
      "grad_norm": 0.5568847060203552,
      "learning_rate": 2.1150597553709743e-05,
      "loss": 2.1961,
      "step": 164700
    },
    {
      "epoch": 0.5774366522973399,
      "grad_norm": 0.545061469078064,
      "learning_rate": 2.1140083412189395e-05,
      "loss": 2.2026,
      "step": 164760
    },
    {
      "epoch": 0.5776469351277468,
      "grad_norm": 0.5466181635856628,
      "learning_rate": 2.112956927066905e-05,
      "loss": 2.201,
      "step": 164820
    },
    {
      "epoch": 0.5778572179581537,
      "grad_norm": 0.5336632132530212,
      "learning_rate": 2.111905512914871e-05,
      "loss": 2.2067,
      "step": 164880
    },
    {
      "epoch": 0.5780675007885606,
      "grad_norm": 0.5472707152366638,
      "learning_rate": 2.110854098762836e-05,
      "loss": 2.196,
      "step": 164940
    },
    {
      "epoch": 0.5782777836189675,
      "grad_norm": 0.5615684390068054,
      "learning_rate": 2.1098026846108016e-05,
      "loss": 2.1958,
      "step": 165000
    },
    {
      "epoch": 0.5784880664493744,
      "grad_norm": 0.5983463525772095,
      "learning_rate": 2.108751270458767e-05,
      "loss": 2.204,
      "step": 165060
    },
    {
      "epoch": 0.5786983492797813,
      "grad_norm": 0.5347958207130432,
      "learning_rate": 2.1076998563067326e-05,
      "loss": 2.1982,
      "step": 165120
    },
    {
      "epoch": 0.5789086321101882,
      "grad_norm": 0.5867051482200623,
      "learning_rate": 2.106648442154698e-05,
      "loss": 2.1985,
      "step": 165180
    },
    {
      "epoch": 0.5791189149405951,
      "grad_norm": 0.5597366094589233,
      "learning_rate": 2.1055970280026637e-05,
      "loss": 2.2171,
      "step": 165240
    },
    {
      "epoch": 0.579329197771002,
      "grad_norm": 0.5548207759857178,
      "learning_rate": 2.1045456138506292e-05,
      "loss": 2.2144,
      "step": 165300
    },
    {
      "epoch": 0.5795394806014089,
      "grad_norm": 0.5555901527404785,
      "learning_rate": 2.1034941996985947e-05,
      "loss": 2.2007,
      "step": 165360
    },
    {
      "epoch": 0.5797497634318158,
      "grad_norm": 0.5185126662254333,
      "learning_rate": 2.1024427855465603e-05,
      "loss": 2.2192,
      "step": 165420
    },
    {
      "epoch": 0.5799600462622226,
      "grad_norm": 0.5712375044822693,
      "learning_rate": 2.1013913713945258e-05,
      "loss": 2.2049,
      "step": 165480
    },
    {
      "epoch": 0.5801703290926296,
      "grad_norm": 0.5157219171524048,
      "learning_rate": 2.1003399572424913e-05,
      "loss": 2.2082,
      "step": 165540
    },
    {
      "epoch": 0.5803806119230365,
      "grad_norm": 0.5245518684387207,
      "learning_rate": 2.099288543090457e-05,
      "loss": 2.2327,
      "step": 165600
    },
    {
      "epoch": 0.5805908947534434,
      "grad_norm": 0.5682709813117981,
      "learning_rate": 2.0982371289384224e-05,
      "loss": 2.2124,
      "step": 165660
    },
    {
      "epoch": 0.5808011775838503,
      "grad_norm": 0.5247425436973572,
      "learning_rate": 2.0971857147863876e-05,
      "loss": 2.1882,
      "step": 165720
    },
    {
      "epoch": 0.5810114604142572,
      "grad_norm": 0.5044289231300354,
      "learning_rate": 2.0961343006343534e-05,
      "loss": 2.1964,
      "step": 165780
    },
    {
      "epoch": 0.581221743244664,
      "grad_norm": 0.5318524837493896,
      "learning_rate": 2.095082886482319e-05,
      "loss": 2.2039,
      "step": 165840
    },
    {
      "epoch": 0.5814320260750709,
      "grad_norm": 0.5390702486038208,
      "learning_rate": 2.094031472330284e-05,
      "loss": 2.2166,
      "step": 165900
    },
    {
      "epoch": 0.5816423089054779,
      "grad_norm": 0.5584826469421387,
      "learning_rate": 2.09298005817825e-05,
      "loss": 2.204,
      "step": 165960
    },
    {
      "epoch": 0.5818525917358848,
      "grad_norm": 0.6024618148803711,
      "learning_rate": 2.0919286440262152e-05,
      "loss": 2.1936,
      "step": 166020
    },
    {
      "epoch": 0.5820628745662917,
      "grad_norm": 0.5568735003471375,
      "learning_rate": 2.0908772298741807e-05,
      "loss": 2.1944,
      "step": 166080
    },
    {
      "epoch": 0.5822731573966986,
      "grad_norm": 0.4855721890926361,
      "learning_rate": 2.0898258157221466e-05,
      "loss": 2.1942,
      "step": 166140
    },
    {
      "epoch": 0.5824834402271054,
      "grad_norm": 0.5432448387145996,
      "learning_rate": 2.0887744015701118e-05,
      "loss": 2.217,
      "step": 166200
    },
    {
      "epoch": 0.5826937230575123,
      "grad_norm": 0.5233843922615051,
      "learning_rate": 2.0877229874180773e-05,
      "loss": 2.178,
      "step": 166260
    },
    {
      "epoch": 0.5829040058879192,
      "grad_norm": 0.5257920026779175,
      "learning_rate": 2.086671573266043e-05,
      "loss": 2.2078,
      "step": 166320
    },
    {
      "epoch": 0.5831142887183262,
      "grad_norm": 0.4958641529083252,
      "learning_rate": 2.0856201591140084e-05,
      "loss": 2.2105,
      "step": 166380
    },
    {
      "epoch": 0.5833245715487331,
      "grad_norm": 0.5342472195625305,
      "learning_rate": 2.084568744961974e-05,
      "loss": 2.2074,
      "step": 166440
    },
    {
      "epoch": 0.58353485437914,
      "grad_norm": 0.5222705006599426,
      "learning_rate": 2.08353485437914e-05,
      "loss": 2.2021,
      "step": 166500
    },
    {
      "epoch": 0.5837451372095468,
      "grad_norm": 0.588127076625824,
      "learning_rate": 2.082483440227106e-05,
      "loss": 2.2006,
      "step": 166560
    },
    {
      "epoch": 0.5839554200399537,
      "grad_norm": 0.5478534698486328,
      "learning_rate": 2.081432026075071e-05,
      "loss": 2.1941,
      "step": 166620
    },
    {
      "epoch": 0.5841657028703606,
      "grad_norm": 0.6026087403297424,
      "learning_rate": 2.0803806119230365e-05,
      "loss": 2.1984,
      "step": 166680
    },
    {
      "epoch": 0.5843759857007675,
      "grad_norm": 0.5317620038986206,
      "learning_rate": 2.079329197771002e-05,
      "loss": 2.2086,
      "step": 166740
    },
    {
      "epoch": 0.5845862685311745,
      "grad_norm": 0.5124843120574951,
      "learning_rate": 2.0782777836189676e-05,
      "loss": 2.2168,
      "step": 166800
    },
    {
      "epoch": 0.5847965513615814,
      "grad_norm": 0.5259480476379395,
      "learning_rate": 2.077226369466933e-05,
      "loss": 2.2097,
      "step": 166860
    },
    {
      "epoch": 0.5850068341919882,
      "grad_norm": 0.5472617149353027,
      "learning_rate": 2.0761749553148987e-05,
      "loss": 2.204,
      "step": 166920
    },
    {
      "epoch": 0.5852171170223951,
      "grad_norm": 0.540069043636322,
      "learning_rate": 2.0751235411628642e-05,
      "loss": 2.2079,
      "step": 166980
    },
    {
      "epoch": 0.585427399852802,
      "grad_norm": 0.5131522417068481,
      "learning_rate": 2.0740721270108297e-05,
      "loss": 2.2178,
      "step": 167040
    },
    {
      "epoch": 0.5856376826832089,
      "grad_norm": 0.5550267100334167,
      "learning_rate": 2.0730207128587952e-05,
      "loss": 2.2018,
      "step": 167100
    },
    {
      "epoch": 0.5858479655136158,
      "grad_norm": 0.5923956036567688,
      "learning_rate": 2.0719692987067608e-05,
      "loss": 2.1929,
      "step": 167160
    },
    {
      "epoch": 0.5860582483440228,
      "grad_norm": 0.541883111000061,
      "learning_rate": 2.070917884554726e-05,
      "loss": 2.1886,
      "step": 167220
    },
    {
      "epoch": 0.5862685311744296,
      "grad_norm": 0.5703868865966797,
      "learning_rate": 2.0698664704026918e-05,
      "loss": 2.2095,
      "step": 167280
    },
    {
      "epoch": 0.5864788140048365,
      "grad_norm": 0.5646779537200928,
      "learning_rate": 2.0688150562506574e-05,
      "loss": 2.2195,
      "step": 167340
    },
    {
      "epoch": 0.5866890968352434,
      "grad_norm": 0.5657762289047241,
      "learning_rate": 2.0677636420986225e-05,
      "loss": 2.2057,
      "step": 167400
    },
    {
      "epoch": 0.5868993796656503,
      "grad_norm": 0.5300728678703308,
      "learning_rate": 2.0667122279465884e-05,
      "loss": 2.2111,
      "step": 167460
    },
    {
      "epoch": 0.5871096624960572,
      "grad_norm": 0.5432790517807007,
      "learning_rate": 2.065660813794554e-05,
      "loss": 2.199,
      "step": 167520
    },
    {
      "epoch": 0.5873199453264641,
      "grad_norm": 0.5578957796096802,
      "learning_rate": 2.064609399642519e-05,
      "loss": 2.1944,
      "step": 167580
    },
    {
      "epoch": 0.587530228156871,
      "grad_norm": 0.6584115624427795,
      "learning_rate": 2.063557985490485e-05,
      "loss": 2.2217,
      "step": 167640
    },
    {
      "epoch": 0.5877405109872779,
      "grad_norm": 0.5259327292442322,
      "learning_rate": 2.0625065713384502e-05,
      "loss": 2.2109,
      "step": 167700
    },
    {
      "epoch": 0.5879507938176848,
      "grad_norm": 0.5304680466651917,
      "learning_rate": 2.0614551571864157e-05,
      "loss": 2.1984,
      "step": 167760
    },
    {
      "epoch": 0.5881610766480917,
      "grad_norm": 0.515880823135376,
      "learning_rate": 2.0604037430343816e-05,
      "loss": 2.2042,
      "step": 167820
    },
    {
      "epoch": 0.5883713594784986,
      "grad_norm": 0.6544479131698608,
      "learning_rate": 2.0593523288823468e-05,
      "loss": 2.2056,
      "step": 167880
    },
    {
      "epoch": 0.5885816423089055,
      "grad_norm": 0.5079035758972168,
      "learning_rate": 2.0583009147303123e-05,
      "loss": 2.2061,
      "step": 167940
    },
    {
      "epoch": 0.5887919251393123,
      "grad_norm": 0.5269632935523987,
      "learning_rate": 2.0572495005782778e-05,
      "loss": 2.2078,
      "step": 168000
    },
    {
      "epoch": 0.5890022079697192,
      "grad_norm": 0.6275792121887207,
      "learning_rate": 2.0561980864262434e-05,
      "loss": 2.2012,
      "step": 168060
    },
    {
      "epoch": 0.5892124908001262,
      "grad_norm": 0.5413856506347656,
      "learning_rate": 2.055146672274209e-05,
      "loss": 2.1943,
      "step": 168120
    },
    {
      "epoch": 0.5894227736305331,
      "grad_norm": 0.5271443724632263,
      "learning_rate": 2.0540952581221744e-05,
      "loss": 2.1954,
      "step": 168180
    },
    {
      "epoch": 0.58963305646094,
      "grad_norm": 0.5837568044662476,
      "learning_rate": 2.0530613675393408e-05,
      "loss": 2.2011,
      "step": 168240
    },
    {
      "epoch": 0.5898433392913469,
      "grad_norm": 0.5604709386825562,
      "learning_rate": 2.052009953387306e-05,
      "loss": 2.1948,
      "step": 168300
    },
    {
      "epoch": 0.5900536221217537,
      "grad_norm": 0.5001662969589233,
      "learning_rate": 2.0509585392352715e-05,
      "loss": 2.2032,
      "step": 168360
    },
    {
      "epoch": 0.5902639049521606,
      "grad_norm": 0.5290040969848633,
      "learning_rate": 2.049907125083237e-05,
      "loss": 2.2068,
      "step": 168420
    },
    {
      "epoch": 0.5904741877825676,
      "grad_norm": 0.5256708264350891,
      "learning_rate": 2.0488557109312026e-05,
      "loss": 2.2006,
      "step": 168480
    },
    {
      "epoch": 0.5906844706129745,
      "grad_norm": 0.5151348114013672,
      "learning_rate": 2.047804296779168e-05,
      "loss": 2.18,
      "step": 168540
    },
    {
      "epoch": 0.5908947534433814,
      "grad_norm": 0.555904746055603,
      "learning_rate": 2.0467528826271336e-05,
      "loss": 2.2069,
      "step": 168600
    },
    {
      "epoch": 0.5911050362737883,
      "grad_norm": 0.5817997455596924,
      "learning_rate": 2.045701468475099e-05,
      "loss": 2.1841,
      "step": 168660
    },
    {
      "epoch": 0.5913153191041951,
      "grad_norm": 0.6211709976196289,
      "learning_rate": 2.0446500543230647e-05,
      "loss": 2.1905,
      "step": 168720
    },
    {
      "epoch": 0.591525601934602,
      "grad_norm": 0.5097838640213013,
      "learning_rate": 2.0435986401710302e-05,
      "loss": 2.2004,
      "step": 168780
    },
    {
      "epoch": 0.5917358847650089,
      "grad_norm": 0.5386205315589905,
      "learning_rate": 2.0425472260189957e-05,
      "loss": 2.1964,
      "step": 168840
    },
    {
      "epoch": 0.5919461675954159,
      "grad_norm": 0.5313547849655151,
      "learning_rate": 2.041495811866961e-05,
      "loss": 2.1926,
      "step": 168900
    },
    {
      "epoch": 0.5921564504258228,
      "grad_norm": 0.5706102848052979,
      "learning_rate": 2.0404443977149268e-05,
      "loss": 2.2198,
      "step": 168960
    },
    {
      "epoch": 0.5923667332562297,
      "grad_norm": 0.5210338234901428,
      "learning_rate": 2.0393929835628923e-05,
      "loss": 2.2126,
      "step": 169020
    },
    {
      "epoch": 0.5925770160866365,
      "grad_norm": 0.6056246161460876,
      "learning_rate": 2.0383415694108575e-05,
      "loss": 2.185,
      "step": 169080
    },
    {
      "epoch": 0.5927872989170434,
      "grad_norm": 0.5859523415565491,
      "learning_rate": 2.0372901552588234e-05,
      "loss": 2.2102,
      "step": 169140
    },
    {
      "epoch": 0.5929975817474503,
      "grad_norm": 0.5275964736938477,
      "learning_rate": 2.036238741106789e-05,
      "loss": 2.209,
      "step": 169200
    },
    {
      "epoch": 0.5932078645778572,
      "grad_norm": 0.5471135973930359,
      "learning_rate": 2.035187326954754e-05,
      "loss": 2.1826,
      "step": 169260
    },
    {
      "epoch": 0.5934181474082642,
      "grad_norm": 0.5576337575912476,
      "learning_rate": 2.03413591280272e-05,
      "loss": 2.1961,
      "step": 169320
    },
    {
      "epoch": 0.5936284302386711,
      "grad_norm": 0.5606521368026733,
      "learning_rate": 2.033084498650685e-05,
      "loss": 2.2022,
      "step": 169380
    },
    {
      "epoch": 0.5938387130690779,
      "grad_norm": 0.5469135642051697,
      "learning_rate": 2.0320330844986507e-05,
      "loss": 2.2126,
      "step": 169440
    },
    {
      "epoch": 0.5940489958994848,
      "grad_norm": 0.5704513788223267,
      "learning_rate": 2.0309816703466165e-05,
      "loss": 2.1926,
      "step": 169500
    },
    {
      "epoch": 0.5942592787298917,
      "grad_norm": 0.5261359214782715,
      "learning_rate": 2.0299302561945817e-05,
      "loss": 2.2098,
      "step": 169560
    },
    {
      "epoch": 0.5944695615602986,
      "grad_norm": 0.5451574921607971,
      "learning_rate": 2.0288788420425473e-05,
      "loss": 2.1908,
      "step": 169620
    },
    {
      "epoch": 0.5946798443907055,
      "grad_norm": 0.599793553352356,
      "learning_rate": 2.0278274278905128e-05,
      "loss": 2.2151,
      "step": 169680
    },
    {
      "epoch": 0.5948901272211123,
      "grad_norm": 0.5122420191764832,
      "learning_rate": 2.0267760137384783e-05,
      "loss": 2.1889,
      "step": 169740
    },
    {
      "epoch": 0.5951004100515193,
      "grad_norm": 0.49233028292655945,
      "learning_rate": 2.025724599586444e-05,
      "loss": 2.1872,
      "step": 169800
    },
    {
      "epoch": 0.5953106928819262,
      "grad_norm": 0.553433358669281,
      "learning_rate": 2.0246731854344094e-05,
      "loss": 2.2133,
      "step": 169860
    },
    {
      "epoch": 0.5955209757123331,
      "grad_norm": 0.6067487001419067,
      "learning_rate": 2.023621771282375e-05,
      "loss": 2.2073,
      "step": 169920
    },
    {
      "epoch": 0.59573125854274,
      "grad_norm": 0.5803714394569397,
      "learning_rate": 2.0225703571303404e-05,
      "loss": 2.2029,
      "step": 169980
    },
    {
      "epoch": 0.5959415413731469,
      "grad_norm": 0.5488730669021606,
      "learning_rate": 2.021518942978306e-05,
      "loss": 2.2062,
      "step": 170040
    },
    {
      "epoch": 0.5961518242035537,
      "grad_norm": 0.5398715734481812,
      "learning_rate": 2.0204675288262715e-05,
      "loss": 2.2067,
      "step": 170100
    },
    {
      "epoch": 0.5963621070339606,
      "grad_norm": 0.5281250476837158,
      "learning_rate": 2.019416114674237e-05,
      "loss": 2.192,
      "step": 170160
    },
    {
      "epoch": 0.5965723898643676,
      "grad_norm": 0.53722083568573,
      "learning_rate": 2.0183647005222025e-05,
      "loss": 2.2,
      "step": 170220
    },
    {
      "epoch": 0.5967826726947745,
      "grad_norm": 0.5472487807273865,
      "learning_rate": 2.017313286370168e-05,
      "loss": 2.1978,
      "step": 170280
    },
    {
      "epoch": 0.5969929555251814,
      "grad_norm": 0.6086287498474121,
      "learning_rate": 2.0162618722181333e-05,
      "loss": 2.1849,
      "step": 170340
    },
    {
      "epoch": 0.5972032383555883,
      "grad_norm": 0.655676543712616,
      "learning_rate": 2.015210458066099e-05,
      "loss": 2.2063,
      "step": 170400
    },
    {
      "epoch": 0.5974135211859951,
      "grad_norm": 0.5415231585502625,
      "learning_rate": 2.0141590439140647e-05,
      "loss": 2.185,
      "step": 170460
    },
    {
      "epoch": 0.597623804016402,
      "grad_norm": 0.5675277709960938,
      "learning_rate": 2.01310762976203e-05,
      "loss": 2.2077,
      "step": 170520
    },
    {
      "epoch": 0.5978340868468089,
      "grad_norm": 0.5531576871871948,
      "learning_rate": 2.0120562156099957e-05,
      "loss": 2.1948,
      "step": 170580
    },
    {
      "epoch": 0.5980443696772159,
      "grad_norm": 0.5377048850059509,
      "learning_rate": 2.011004801457961e-05,
      "loss": 2.1867,
      "step": 170640
    },
    {
      "epoch": 0.5982546525076228,
      "grad_norm": 0.47934389114379883,
      "learning_rate": 2.0099533873059264e-05,
      "loss": 2.2068,
      "step": 170700
    },
    {
      "epoch": 0.5984649353380297,
      "grad_norm": 0.576343297958374,
      "learning_rate": 2.0089019731538923e-05,
      "loss": 2.1905,
      "step": 170760
    },
    {
      "epoch": 0.5986752181684365,
      "grad_norm": 0.5875000953674316,
      "learning_rate": 2.0078505590018575e-05,
      "loss": 2.2133,
      "step": 170820
    },
    {
      "epoch": 0.5988855009988434,
      "grad_norm": 0.5581299066543579,
      "learning_rate": 2.006799144849823e-05,
      "loss": 2.2149,
      "step": 170880
    },
    {
      "epoch": 0.5990957838292503,
      "grad_norm": 0.5211154818534851,
      "learning_rate": 2.0057477306977885e-05,
      "loss": 2.2056,
      "step": 170940
    },
    {
      "epoch": 0.5993060666596572,
      "grad_norm": 0.6828522682189941,
      "learning_rate": 2.004696316545754e-05,
      "loss": 2.2017,
      "step": 171000
    },
    {
      "epoch": 0.5995163494900642,
      "grad_norm": 0.6186325550079346,
      "learning_rate": 2.0036449023937196e-05,
      "loss": 2.2047,
      "step": 171060
    },
    {
      "epoch": 0.5997266323204711,
      "grad_norm": 0.5010774731636047,
      "learning_rate": 2.002593488241685e-05,
      "loss": 2.2054,
      "step": 171120
    },
    {
      "epoch": 0.5999369151508779,
      "grad_norm": 0.5373931527137756,
      "learning_rate": 2.0015420740896507e-05,
      "loss": 2.2007,
      "step": 171180
    },
    {
      "epoch": 0.6001471979812848,
      "grad_norm": 0.6149768829345703,
      "learning_rate": 2.0004906599376162e-05,
      "loss": 2.1917,
      "step": 171240
    },
    {
      "epoch": 0.6003574808116917,
      "grad_norm": 0.5230481028556824,
      "learning_rate": 1.9994392457855817e-05,
      "loss": 2.1969,
      "step": 171300
    },
    {
      "epoch": 0.6005677636420986,
      "grad_norm": 0.6458109021186829,
      "learning_rate": 1.9983878316335472e-05,
      "loss": 2.1982,
      "step": 171360
    },
    {
      "epoch": 0.6007780464725055,
      "grad_norm": 0.5456601977348328,
      "learning_rate": 1.9973364174815128e-05,
      "loss": 2.1954,
      "step": 171420
    },
    {
      "epoch": 0.6009883293029125,
      "grad_norm": 0.534690260887146,
      "learning_rate": 1.9962850033294783e-05,
      "loss": 2.2124,
      "step": 171480
    },
    {
      "epoch": 0.6011986121333193,
      "grad_norm": 0.5408187508583069,
      "learning_rate": 1.9952335891774438e-05,
      "loss": 2.2105,
      "step": 171540
    },
    {
      "epoch": 0.6014088949637262,
      "grad_norm": 0.5347569584846497,
      "learning_rate": 1.994182175025409e-05,
      "loss": 2.2046,
      "step": 171600
    },
    {
      "epoch": 0.6016191777941331,
      "grad_norm": 0.5162372589111328,
      "learning_rate": 1.993130760873375e-05,
      "loss": 2.1956,
      "step": 171660
    },
    {
      "epoch": 0.60182946062454,
      "grad_norm": 0.570464551448822,
      "learning_rate": 1.9920793467213404e-05,
      "loss": 2.1895,
      "step": 171720
    },
    {
      "epoch": 0.6020397434549469,
      "grad_norm": 0.5477625727653503,
      "learning_rate": 1.9910279325693056e-05,
      "loss": 2.1948,
      "step": 171780
    },
    {
      "epoch": 0.6022500262853538,
      "grad_norm": 0.5018092393875122,
      "learning_rate": 1.9899765184172715e-05,
      "loss": 2.2074,
      "step": 171840
    },
    {
      "epoch": 0.6024603091157607,
      "grad_norm": 0.4974982738494873,
      "learning_rate": 1.9889251042652367e-05,
      "loss": 2.2148,
      "step": 171900
    },
    {
      "epoch": 0.6026705919461676,
      "grad_norm": 0.5513793230056763,
      "learning_rate": 1.9878736901132022e-05,
      "loss": 2.2062,
      "step": 171960
    },
    {
      "epoch": 0.6028808747765745,
      "grad_norm": 0.5236059427261353,
      "learning_rate": 1.986822275961168e-05,
      "loss": 2.1883,
      "step": 172020
    },
    {
      "epoch": 0.6030911576069814,
      "grad_norm": 0.5051701664924622,
      "learning_rate": 1.9857708618091332e-05,
      "loss": 2.2044,
      "step": 172080
    },
    {
      "epoch": 0.6033014404373883,
      "grad_norm": 0.5532500147819519,
      "learning_rate": 1.9847194476570988e-05,
      "loss": 2.1961,
      "step": 172140
    },
    {
      "epoch": 0.6035117232677952,
      "grad_norm": 0.5269227623939514,
      "learning_rate": 1.9836855570742648e-05,
      "loss": 2.1847,
      "step": 172200
    },
    {
      "epoch": 0.603722006098202,
      "grad_norm": 0.5466260313987732,
      "learning_rate": 1.9826341429222307e-05,
      "loss": 2.2018,
      "step": 172260
    },
    {
      "epoch": 0.603932288928609,
      "grad_norm": 0.5222917795181274,
      "learning_rate": 1.981582728770196e-05,
      "loss": 2.2044,
      "step": 172320
    },
    {
      "epoch": 0.6041425717590159,
      "grad_norm": 0.5396358370780945,
      "learning_rate": 1.9805313146181614e-05,
      "loss": 2.1967,
      "step": 172380
    },
    {
      "epoch": 0.6043528545894228,
      "grad_norm": 0.5147253274917603,
      "learning_rate": 1.9794799004661273e-05,
      "loss": 2.198,
      "step": 172440
    },
    {
      "epoch": 0.6045631374198297,
      "grad_norm": 0.5578526258468628,
      "learning_rate": 1.9784284863140925e-05,
      "loss": 2.2013,
      "step": 172500
    },
    {
      "epoch": 0.6047734202502366,
      "grad_norm": 0.5797805786132812,
      "learning_rate": 1.977377072162058e-05,
      "loss": 2.1967,
      "step": 172560
    },
    {
      "epoch": 0.6049837030806434,
      "grad_norm": 0.6064589023590088,
      "learning_rate": 1.9763256580100235e-05,
      "loss": 2.2014,
      "step": 172620
    },
    {
      "epoch": 0.6051939859110503,
      "grad_norm": 0.5353837609291077,
      "learning_rate": 1.975274243857989e-05,
      "loss": 2.2,
      "step": 172680
    },
    {
      "epoch": 0.6054042687414573,
      "grad_norm": 0.5772452354431152,
      "learning_rate": 1.9742228297059546e-05,
      "loss": 2.1983,
      "step": 172740
    },
    {
      "epoch": 0.6056145515718642,
      "grad_norm": 0.5389087200164795,
      "learning_rate": 1.97317141555392e-05,
      "loss": 2.1982,
      "step": 172800
    },
    {
      "epoch": 0.6058248344022711,
      "grad_norm": 0.5934854745864868,
      "learning_rate": 1.9721200014018856e-05,
      "loss": 2.206,
      "step": 172860
    },
    {
      "epoch": 0.606035117232678,
      "grad_norm": 0.5362783670425415,
      "learning_rate": 1.971068587249851e-05,
      "loss": 2.1964,
      "step": 172920
    },
    {
      "epoch": 0.6062454000630848,
      "grad_norm": 0.5497432947158813,
      "learning_rate": 1.9700171730978167e-05,
      "loss": 2.2167,
      "step": 172980
    },
    {
      "epoch": 0.6064556828934917,
      "grad_norm": 0.5316357016563416,
      "learning_rate": 1.9689657589457822e-05,
      "loss": 2.1987,
      "step": 173040
    },
    {
      "epoch": 0.6066659657238986,
      "grad_norm": 0.5472717881202698,
      "learning_rate": 1.9679143447937477e-05,
      "loss": 2.2006,
      "step": 173100
    },
    {
      "epoch": 0.6068762485543056,
      "grad_norm": 0.564258337020874,
      "learning_rate": 1.9668629306417133e-05,
      "loss": 2.1874,
      "step": 173160
    },
    {
      "epoch": 0.6070865313847125,
      "grad_norm": 0.5468765497207642,
      "learning_rate": 1.9658115164896788e-05,
      "loss": 2.2048,
      "step": 173220
    },
    {
      "epoch": 0.6072968142151194,
      "grad_norm": 0.5391786694526672,
      "learning_rate": 1.964760102337644e-05,
      "loss": 2.2042,
      "step": 173280
    },
    {
      "epoch": 0.6075070970455262,
      "grad_norm": 0.5502793788909912,
      "learning_rate": 1.96370868818561e-05,
      "loss": 2.2096,
      "step": 173340
    },
    {
      "epoch": 0.6077173798759331,
      "grad_norm": 0.49307966232299805,
      "learning_rate": 1.9626572740335754e-05,
      "loss": 2.2055,
      "step": 173400
    },
    {
      "epoch": 0.60792766270634,
      "grad_norm": 0.5799115896224976,
      "learning_rate": 1.9616058598815406e-05,
      "loss": 2.1992,
      "step": 173460
    },
    {
      "epoch": 0.6081379455367469,
      "grad_norm": 0.554755449295044,
      "learning_rate": 1.9605544457295064e-05,
      "loss": 2.1998,
      "step": 173520
    },
    {
      "epoch": 0.6083482283671539,
      "grad_norm": 0.5916181802749634,
      "learning_rate": 1.9595030315774716e-05,
      "loss": 2.2041,
      "step": 173580
    },
    {
      "epoch": 0.6085585111975608,
      "grad_norm": 0.5109410285949707,
      "learning_rate": 1.958451617425437e-05,
      "loss": 2.1912,
      "step": 173640
    },
    {
      "epoch": 0.6087687940279676,
      "grad_norm": 0.6058209538459778,
      "learning_rate": 1.957400203273403e-05,
      "loss": 2.2032,
      "step": 173700
    },
    {
      "epoch": 0.6089790768583745,
      "grad_norm": 0.5700769424438477,
      "learning_rate": 1.9563487891213682e-05,
      "loss": 2.2047,
      "step": 173760
    },
    {
      "epoch": 0.6091893596887814,
      "grad_norm": 0.5366501212120056,
      "learning_rate": 1.9552973749693337e-05,
      "loss": 2.2086,
      "step": 173820
    },
    {
      "epoch": 0.6093996425191883,
      "grad_norm": 0.530959963798523,
      "learning_rate": 1.9542459608172996e-05,
      "loss": 2.194,
      "step": 173880
    },
    {
      "epoch": 0.6096099253495952,
      "grad_norm": 0.5777100324630737,
      "learning_rate": 1.9531945466652648e-05,
      "loss": 2.187,
      "step": 173940
    },
    {
      "epoch": 0.6098202081800022,
      "grad_norm": 0.5903493762016296,
      "learning_rate": 1.9521431325132303e-05,
      "loss": 2.1917,
      "step": 174000
    },
    {
      "epoch": 0.610030491010409,
      "grad_norm": 0.5649183988571167,
      "learning_rate": 1.951091718361196e-05,
      "loss": 2.2008,
      "step": 174060
    },
    {
      "epoch": 0.6102407738408159,
      "grad_norm": 0.5999853610992432,
      "learning_rate": 1.9500578277783622e-05,
      "loss": 2.1982,
      "step": 174120
    },
    {
      "epoch": 0.6104510566712228,
      "grad_norm": 0.5888971090316772,
      "learning_rate": 1.9490064136263274e-05,
      "loss": 2.1858,
      "step": 174180
    },
    {
      "epoch": 0.6106613395016297,
      "grad_norm": 0.5282642841339111,
      "learning_rate": 1.947954999474293e-05,
      "loss": 2.2078,
      "step": 174240
    },
    {
      "epoch": 0.6108716223320366,
      "grad_norm": 0.5580457448959351,
      "learning_rate": 1.9469035853222585e-05,
      "loss": 2.2056,
      "step": 174300
    },
    {
      "epoch": 0.6110819051624435,
      "grad_norm": 0.5449835658073425,
      "learning_rate": 1.945852171170224e-05,
      "loss": 2.1955,
      "step": 174360
    },
    {
      "epoch": 0.6112921879928503,
      "grad_norm": 0.575751543045044,
      "learning_rate": 1.9448007570181895e-05,
      "loss": 2.2093,
      "step": 174420
    },
    {
      "epoch": 0.6115024708232573,
      "grad_norm": 0.5661749243736267,
      "learning_rate": 1.943749342866155e-05,
      "loss": 2.1966,
      "step": 174480
    },
    {
      "epoch": 0.6117127536536642,
      "grad_norm": 0.5168628692626953,
      "learning_rate": 1.9426979287141206e-05,
      "loss": 2.1901,
      "step": 174540
    },
    {
      "epoch": 0.6119230364840711,
      "grad_norm": 0.5125284194946289,
      "learning_rate": 1.941646514562086e-05,
      "loss": 2.1967,
      "step": 174600
    },
    {
      "epoch": 0.612133319314478,
      "grad_norm": 0.5725422501564026,
      "learning_rate": 1.9405951004100516e-05,
      "loss": 2.1878,
      "step": 174660
    },
    {
      "epoch": 0.6123436021448849,
      "grad_norm": 0.5970425009727478,
      "learning_rate": 1.9395436862580172e-05,
      "loss": 2.2048,
      "step": 174720
    },
    {
      "epoch": 0.6125538849752917,
      "grad_norm": 0.6201969385147095,
      "learning_rate": 1.9384922721059827e-05,
      "loss": 2.2,
      "step": 174780
    },
    {
      "epoch": 0.6127641678056986,
      "grad_norm": 0.5165169835090637,
      "learning_rate": 1.9374408579539482e-05,
      "loss": 2.2012,
      "step": 174840
    },
    {
      "epoch": 0.6129744506361056,
      "grad_norm": 0.5577533841133118,
      "learning_rate": 1.9363894438019138e-05,
      "loss": 2.1943,
      "step": 174900
    },
    {
      "epoch": 0.6131847334665125,
      "grad_norm": 0.5277051329612732,
      "learning_rate": 1.935338029649879e-05,
      "loss": 2.2078,
      "step": 174960
    },
    {
      "epoch": 0.6133950162969194,
      "grad_norm": 0.5218917727470398,
      "learning_rate": 1.9342866154978448e-05,
      "loss": 2.2117,
      "step": 175020
    },
    {
      "epoch": 0.6136052991273263,
      "grad_norm": 0.5269528031349182,
      "learning_rate": 1.9332352013458103e-05,
      "loss": 2.1972,
      "step": 175080
    },
    {
      "epoch": 0.6138155819577331,
      "grad_norm": 0.6118609309196472,
      "learning_rate": 1.9321837871937755e-05,
      "loss": 2.2075,
      "step": 175140
    },
    {
      "epoch": 0.61402586478814,
      "grad_norm": 0.545382559299469,
      "learning_rate": 1.9311323730417414e-05,
      "loss": 2.1901,
      "step": 175200
    },
    {
      "epoch": 0.614236147618547,
      "grad_norm": 0.6513542532920837,
      "learning_rate": 1.9300809588897066e-05,
      "loss": 2.1867,
      "step": 175260
    },
    {
      "epoch": 0.6144464304489539,
      "grad_norm": 0.5605107545852661,
      "learning_rate": 1.929029544737672e-05,
      "loss": 2.1884,
      "step": 175320
    },
    {
      "epoch": 0.6146567132793608,
      "grad_norm": 0.6436792612075806,
      "learning_rate": 1.927978130585638e-05,
      "loss": 2.1898,
      "step": 175380
    },
    {
      "epoch": 0.6148669961097676,
      "grad_norm": 0.5533410310745239,
      "learning_rate": 1.9269267164336032e-05,
      "loss": 2.2032,
      "step": 175440
    },
    {
      "epoch": 0.6150772789401745,
      "grad_norm": 0.501750111579895,
      "learning_rate": 1.9258753022815687e-05,
      "loss": 2.2081,
      "step": 175500
    },
    {
      "epoch": 0.6152875617705814,
      "grad_norm": 0.498298317193985,
      "learning_rate": 1.9248238881295346e-05,
      "loss": 2.2056,
      "step": 175560
    },
    {
      "epoch": 0.6154978446009883,
      "grad_norm": 0.5598739981651306,
      "learning_rate": 1.9237724739774998e-05,
      "loss": 2.179,
      "step": 175620
    },
    {
      "epoch": 0.6157081274313952,
      "grad_norm": 0.5532486438751221,
      "learning_rate": 1.9227210598254653e-05,
      "loss": 2.1799,
      "step": 175680
    },
    {
      "epoch": 0.6159184102618022,
      "grad_norm": 0.5259189009666443,
      "learning_rate": 1.9216696456734308e-05,
      "loss": 2.189,
      "step": 175740
    },
    {
      "epoch": 0.616128693092209,
      "grad_norm": 0.5915912389755249,
      "learning_rate": 1.9206182315213963e-05,
      "loss": 2.1899,
      "step": 175800
    },
    {
      "epoch": 0.6163389759226159,
      "grad_norm": 0.511455774307251,
      "learning_rate": 1.919566817369362e-05,
      "loss": 2.2011,
      "step": 175860
    },
    {
      "epoch": 0.6165492587530228,
      "grad_norm": 0.5347529649734497,
      "learning_rate": 1.9185154032173274e-05,
      "loss": 2.2029,
      "step": 175920
    },
    {
      "epoch": 0.6167595415834297,
      "grad_norm": 0.5264248251914978,
      "learning_rate": 1.917463989065293e-05,
      "loss": 2.1933,
      "step": 175980
    },
    {
      "epoch": 0.6169698244138366,
      "grad_norm": 0.5930898189544678,
      "learning_rate": 1.9164125749132585e-05,
      "loss": 2.1939,
      "step": 176040
    },
    {
      "epoch": 0.6171801072442435,
      "grad_norm": 0.6320661902427673,
      "learning_rate": 1.915361160761224e-05,
      "loss": 2.1923,
      "step": 176100
    },
    {
      "epoch": 0.6173903900746504,
      "grad_norm": 0.5503607988357544,
      "learning_rate": 1.9143097466091895e-05,
      "loss": 2.1956,
      "step": 176160
    },
    {
      "epoch": 0.6176006729050573,
      "grad_norm": 0.5331404209136963,
      "learning_rate": 1.9132583324571547e-05,
      "loss": 2.2246,
      "step": 176220
    },
    {
      "epoch": 0.6178109557354642,
      "grad_norm": 0.5129519701004028,
      "learning_rate": 1.9122069183051206e-05,
      "loss": 2.1986,
      "step": 176280
    },
    {
      "epoch": 0.6180212385658711,
      "grad_norm": 0.5717986226081848,
      "learning_rate": 1.911155504153086e-05,
      "loss": 2.1992,
      "step": 176340
    },
    {
      "epoch": 0.618231521396278,
      "grad_norm": 0.5437119603157043,
      "learning_rate": 1.9101040900010513e-05,
      "loss": 2.1926,
      "step": 176400
    },
    {
      "epoch": 0.6184418042266849,
      "grad_norm": 0.5349199175834656,
      "learning_rate": 1.909052675849017e-05,
      "loss": 2.2062,
      "step": 176460
    },
    {
      "epoch": 0.6186520870570917,
      "grad_norm": 0.5290020704269409,
      "learning_rate": 1.9080012616969827e-05,
      "loss": 2.2002,
      "step": 176520
    },
    {
      "epoch": 0.6188623698874987,
      "grad_norm": 0.5506412386894226,
      "learning_rate": 1.906949847544948e-05,
      "loss": 2.1958,
      "step": 176580
    },
    {
      "epoch": 0.6190726527179056,
      "grad_norm": 0.5792592167854309,
      "learning_rate": 1.9058984333929137e-05,
      "loss": 2.2002,
      "step": 176640
    },
    {
      "epoch": 0.6192829355483125,
      "grad_norm": 0.5670069456100464,
      "learning_rate": 1.904847019240879e-05,
      "loss": 2.1975,
      "step": 176700
    },
    {
      "epoch": 0.6194932183787194,
      "grad_norm": 0.6118037700653076,
      "learning_rate": 1.9037956050888445e-05,
      "loss": 2.186,
      "step": 176760
    },
    {
      "epoch": 0.6197035012091263,
      "grad_norm": 0.5524449348449707,
      "learning_rate": 1.9027441909368103e-05,
      "loss": 2.197,
      "step": 176820
    },
    {
      "epoch": 0.6199137840395331,
      "grad_norm": 0.5148321986198425,
      "learning_rate": 1.9016927767847755e-05,
      "loss": 2.1969,
      "step": 176880
    },
    {
      "epoch": 0.62012406686994,
      "grad_norm": 0.5635314583778381,
      "learning_rate": 1.900641362632741e-05,
      "loss": 2.2055,
      "step": 176940
    },
    {
      "epoch": 0.620334349700347,
      "grad_norm": 0.5319798588752747,
      "learning_rate": 1.8995899484807066e-05,
      "loss": 2.1948,
      "step": 177000
    },
    {
      "epoch": 0.6205446325307539,
      "grad_norm": 0.6794459819793701,
      "learning_rate": 1.898538534328672e-05,
      "loss": 2.2103,
      "step": 177060
    },
    {
      "epoch": 0.6207549153611608,
      "grad_norm": 0.6157546639442444,
      "learning_rate": 1.8974871201766376e-05,
      "loss": 2.2019,
      "step": 177120
    },
    {
      "epoch": 0.6209651981915677,
      "grad_norm": 0.5367477536201477,
      "learning_rate": 1.896435706024603e-05,
      "loss": 2.1928,
      "step": 177180
    },
    {
      "epoch": 0.6211754810219745,
      "grad_norm": 0.5531861782073975,
      "learning_rate": 1.8953842918725687e-05,
      "loss": 2.2016,
      "step": 177240
    },
    {
      "epoch": 0.6213857638523814,
      "grad_norm": 0.5841910243034363,
      "learning_rate": 1.8943328777205342e-05,
      "loss": 2.1906,
      "step": 177300
    },
    {
      "epoch": 0.6215960466827883,
      "grad_norm": 0.5331421494483948,
      "learning_rate": 1.8932814635684997e-05,
      "loss": 2.2002,
      "step": 177360
    },
    {
      "epoch": 0.6218063295131953,
      "grad_norm": 0.5444777011871338,
      "learning_rate": 1.8922300494164653e-05,
      "loss": 2.1937,
      "step": 177420
    },
    {
      "epoch": 0.6220166123436022,
      "grad_norm": 0.5303032994270325,
      "learning_rate": 1.8911786352644308e-05,
      "loss": 2.2036,
      "step": 177480
    },
    {
      "epoch": 0.6222268951740091,
      "grad_norm": 0.5160943865776062,
      "learning_rate": 1.890144744681597e-05,
      "loss": 2.1968,
      "step": 177540
    },
    {
      "epoch": 0.6224371780044159,
      "grad_norm": 0.6058926582336426,
      "learning_rate": 1.8890933305295624e-05,
      "loss": 2.1976,
      "step": 177600
    },
    {
      "epoch": 0.6226474608348228,
      "grad_norm": 0.6021593809127808,
      "learning_rate": 1.888041916377528e-05,
      "loss": 2.1935,
      "step": 177660
    },
    {
      "epoch": 0.6228577436652297,
      "grad_norm": 0.5436554551124573,
      "learning_rate": 1.8869905022254934e-05,
      "loss": 2.2202,
      "step": 177720
    },
    {
      "epoch": 0.6230680264956366,
      "grad_norm": 0.527804970741272,
      "learning_rate": 1.885939088073459e-05,
      "loss": 2.1995,
      "step": 177780
    },
    {
      "epoch": 0.6232783093260436,
      "grad_norm": 0.6060287952423096,
      "learning_rate": 1.8848876739214245e-05,
      "loss": 2.2054,
      "step": 177840
    },
    {
      "epoch": 0.6234885921564505,
      "grad_norm": 0.5040077567100525,
      "learning_rate": 1.8838362597693897e-05,
      "loss": 2.1946,
      "step": 177900
    },
    {
      "epoch": 0.6236988749868573,
      "grad_norm": 0.5401580929756165,
      "learning_rate": 1.8827848456173555e-05,
      "loss": 2.1775,
      "step": 177960
    },
    {
      "epoch": 0.6239091578172642,
      "grad_norm": 0.6010558009147644,
      "learning_rate": 1.881733431465321e-05,
      "loss": 2.1757,
      "step": 178020
    },
    {
      "epoch": 0.6241194406476711,
      "grad_norm": 0.6008610725402832,
      "learning_rate": 1.8806820173132863e-05,
      "loss": 2.1986,
      "step": 178080
    },
    {
      "epoch": 0.624329723478078,
      "grad_norm": 0.6357063055038452,
      "learning_rate": 1.879630603161252e-05,
      "loss": 2.1998,
      "step": 178140
    },
    {
      "epoch": 0.6245400063084849,
      "grad_norm": 0.5567299127578735,
      "learning_rate": 1.8785791890092176e-05,
      "loss": 2.1938,
      "step": 178200
    },
    {
      "epoch": 0.6247502891388919,
      "grad_norm": 0.5733098983764648,
      "learning_rate": 1.877527774857183e-05,
      "loss": 2.2024,
      "step": 178260
    },
    {
      "epoch": 0.6249605719692987,
      "grad_norm": 0.6415252685546875,
      "learning_rate": 1.8764763607051487e-05,
      "loss": 2.2091,
      "step": 178320
    },
    {
      "epoch": 0.6251708547997056,
      "grad_norm": 0.5701323747634888,
      "learning_rate": 1.875424946553114e-05,
      "loss": 2.2194,
      "step": 178380
    },
    {
      "epoch": 0.6253811376301125,
      "grad_norm": 0.5979422330856323,
      "learning_rate": 1.8743735324010794e-05,
      "loss": 2.2182,
      "step": 178440
    },
    {
      "epoch": 0.6255914204605194,
      "grad_norm": 0.5337743759155273,
      "learning_rate": 1.8733221182490453e-05,
      "loss": 2.2097,
      "step": 178500
    },
    {
      "epoch": 0.6258017032909263,
      "grad_norm": 0.5219767689704895,
      "learning_rate": 1.8722707040970105e-05,
      "loss": 2.2057,
      "step": 178560
    },
    {
      "epoch": 0.6260119861213332,
      "grad_norm": 0.5585417747497559,
      "learning_rate": 1.871219289944976e-05,
      "loss": 2.1904,
      "step": 178620
    },
    {
      "epoch": 0.62622226895174,
      "grad_norm": 0.5472621321678162,
      "learning_rate": 1.8701678757929415e-05,
      "loss": 2.1978,
      "step": 178680
    },
    {
      "epoch": 0.626432551782147,
      "grad_norm": 0.5493637919425964,
      "learning_rate": 1.869116461640907e-05,
      "loss": 2.2048,
      "step": 178740
    },
    {
      "epoch": 0.6266428346125539,
      "grad_norm": 0.5349371433258057,
      "learning_rate": 1.8680650474888726e-05,
      "loss": 2.1892,
      "step": 178800
    },
    {
      "epoch": 0.6268531174429608,
      "grad_norm": 0.5108512043952942,
      "learning_rate": 1.867013633336838e-05,
      "loss": 2.2086,
      "step": 178860
    },
    {
      "epoch": 0.6270634002733677,
      "grad_norm": 0.5267765522003174,
      "learning_rate": 1.8659622191848036e-05,
      "loss": 2.1787,
      "step": 178920
    },
    {
      "epoch": 0.6272736831037746,
      "grad_norm": 0.5868079662322998,
      "learning_rate": 1.8649108050327692e-05,
      "loss": 2.2027,
      "step": 178980
    },
    {
      "epoch": 0.6274839659341814,
      "grad_norm": 0.5836610794067383,
      "learning_rate": 1.8638593908807347e-05,
      "loss": 2.1919,
      "step": 179040
    },
    {
      "epoch": 0.6276942487645883,
      "grad_norm": 0.5579856038093567,
      "learning_rate": 1.8628079767287002e-05,
      "loss": 2.1972,
      "step": 179100
    },
    {
      "epoch": 0.6279045315949953,
      "grad_norm": 0.5386043190956116,
      "learning_rate": 1.8617565625766658e-05,
      "loss": 2.1811,
      "step": 179160
    },
    {
      "epoch": 0.6281148144254022,
      "grad_norm": 0.5331193208694458,
      "learning_rate": 1.8607051484246313e-05,
      "loss": 2.189,
      "step": 179220
    },
    {
      "epoch": 0.6283250972558091,
      "grad_norm": 0.546983540058136,
      "learning_rate": 1.8596537342725968e-05,
      "loss": 2.1948,
      "step": 179280
    },
    {
      "epoch": 0.628535380086216,
      "grad_norm": 0.531303882598877,
      "learning_rate": 1.858602320120562e-05,
      "loss": 2.1898,
      "step": 179340
    },
    {
      "epoch": 0.6287456629166228,
      "grad_norm": 0.5148618221282959,
      "learning_rate": 1.857550905968528e-05,
      "loss": 2.187,
      "step": 179400
    },
    {
      "epoch": 0.6289559457470297,
      "grad_norm": 0.5330983996391296,
      "learning_rate": 1.8564994918164934e-05,
      "loss": 2.1847,
      "step": 179460
    },
    {
      "epoch": 0.6291662285774366,
      "grad_norm": 0.5286121368408203,
      "learning_rate": 1.8554480776644586e-05,
      "loss": 2.2227,
      "step": 179520
    },
    {
      "epoch": 0.6293765114078436,
      "grad_norm": 0.5923421382904053,
      "learning_rate": 1.8543966635124245e-05,
      "loss": 2.1934,
      "step": 179580
    },
    {
      "epoch": 0.6295867942382505,
      "grad_norm": 0.5364999175071716,
      "learning_rate": 1.8533452493603896e-05,
      "loss": 2.176,
      "step": 179640
    },
    {
      "epoch": 0.6297970770686574,
      "grad_norm": 0.5409441590309143,
      "learning_rate": 1.852293835208355e-05,
      "loss": 2.2059,
      "step": 179700
    },
    {
      "epoch": 0.6300073598990642,
      "grad_norm": 0.5342493653297424,
      "learning_rate": 1.851242421056321e-05,
      "loss": 2.1979,
      "step": 179760
    },
    {
      "epoch": 0.6302176427294711,
      "grad_norm": 0.5905653834342957,
      "learning_rate": 1.8501910069042862e-05,
      "loss": 2.1909,
      "step": 179820
    },
    {
      "epoch": 0.630427925559878,
      "grad_norm": 0.5535312294960022,
      "learning_rate": 1.8491395927522518e-05,
      "loss": 2.1869,
      "step": 179880
    },
    {
      "epoch": 0.630638208390285,
      "grad_norm": 0.5361486077308655,
      "learning_rate": 1.8480881786002173e-05,
      "loss": 2.1917,
      "step": 179940
    },
    {
      "epoch": 0.6308484912206919,
      "grad_norm": 0.5434253215789795,
      "learning_rate": 1.8470367644481828e-05,
      "loss": 2.1744,
      "step": 180000
    },
    {
      "epoch": 0.6310587740510988,
      "grad_norm": 0.6362529993057251,
      "learning_rate": 1.8459853502961483e-05,
      "loss": 2.1978,
      "step": 180060
    },
    {
      "epoch": 0.6312690568815056,
      "grad_norm": 0.5024562478065491,
      "learning_rate": 1.844933936144114e-05,
      "loss": 2.1903,
      "step": 180120
    },
    {
      "epoch": 0.6314793397119125,
      "grad_norm": 0.5112056732177734,
      "learning_rate": 1.8438825219920794e-05,
      "loss": 2.2001,
      "step": 180180
    },
    {
      "epoch": 0.6316896225423194,
      "grad_norm": 0.5255155563354492,
      "learning_rate": 1.842831107840045e-05,
      "loss": 2.1856,
      "step": 180240
    },
    {
      "epoch": 0.6318999053727263,
      "grad_norm": 0.5178373456001282,
      "learning_rate": 1.8417796936880104e-05,
      "loss": 2.1976,
      "step": 180300
    },
    {
      "epoch": 0.6321101882031332,
      "grad_norm": 0.5733485221862793,
      "learning_rate": 1.840728279535976e-05,
      "loss": 2.1893,
      "step": 180360
    },
    {
      "epoch": 0.6323204710335402,
      "grad_norm": 0.6175151467323303,
      "learning_rate": 1.8396768653839415e-05,
      "loss": 2.1902,
      "step": 180420
    },
    {
      "epoch": 0.632530753863947,
      "grad_norm": 0.5645368099212646,
      "learning_rate": 1.838625451231907e-05,
      "loss": 2.1836,
      "step": 180480
    },
    {
      "epoch": 0.6327410366943539,
      "grad_norm": 0.5591124296188354,
      "learning_rate": 1.8375740370798726e-05,
      "loss": 2.1984,
      "step": 180540
    },
    {
      "epoch": 0.6329513195247608,
      "grad_norm": 0.5352583527565002,
      "learning_rate": 1.836522622927838e-05,
      "loss": 2.1962,
      "step": 180600
    },
    {
      "epoch": 0.6331616023551677,
      "grad_norm": 0.6056286096572876,
      "learning_rate": 1.8354712087758036e-05,
      "loss": 2.1884,
      "step": 180660
    },
    {
      "epoch": 0.6333718851855746,
      "grad_norm": 0.7630537748336792,
      "learning_rate": 1.834419794623769e-05,
      "loss": 2.1988,
      "step": 180720
    },
    {
      "epoch": 0.6335821680159816,
      "grad_norm": 0.5268065333366394,
      "learning_rate": 1.8333683804717347e-05,
      "loss": 2.1808,
      "step": 180780
    },
    {
      "epoch": 0.6337924508463884,
      "grad_norm": 0.5918155312538147,
      "learning_rate": 1.8323169663197002e-05,
      "loss": 2.1979,
      "step": 180840
    },
    {
      "epoch": 0.6340027336767953,
      "grad_norm": 0.5612367987632751,
      "learning_rate": 1.8312655521676654e-05,
      "loss": 2.2091,
      "step": 180900
    },
    {
      "epoch": 0.6342130165072022,
      "grad_norm": 0.5907817482948303,
      "learning_rate": 1.8302141380156313e-05,
      "loss": 2.2071,
      "step": 180960
    },
    {
      "epoch": 0.6344232993376091,
      "grad_norm": 0.5619683265686035,
      "learning_rate": 1.8291627238635968e-05,
      "loss": 2.1921,
      "step": 181020
    },
    {
      "epoch": 0.634633582168016,
      "grad_norm": 0.6120419502258301,
      "learning_rate": 1.828111309711562e-05,
      "loss": 2.1918,
      "step": 181080
    },
    {
      "epoch": 0.6348438649984228,
      "grad_norm": 0.5742617249488831,
      "learning_rate": 1.827059895559528e-05,
      "loss": 2.2084,
      "step": 181140
    },
    {
      "epoch": 0.6350541478288297,
      "grad_norm": 0.4941420257091522,
      "learning_rate": 1.8260084814074934e-05,
      "loss": 2.1804,
      "step": 181200
    },
    {
      "epoch": 0.6352644306592367,
      "grad_norm": 0.5339905619621277,
      "learning_rate": 1.8249570672554586e-05,
      "loss": 2.194,
      "step": 181260
    },
    {
      "epoch": 0.6354747134896436,
      "grad_norm": 0.5762035846710205,
      "learning_rate": 1.8239056531034244e-05,
      "loss": 2.1876,
      "step": 181320
    },
    {
      "epoch": 0.6356849963200505,
      "grad_norm": 0.55418860912323,
      "learning_rate": 1.8228542389513896e-05,
      "loss": 2.2229,
      "step": 181380
    },
    {
      "epoch": 0.6358952791504574,
      "grad_norm": 0.6099485158920288,
      "learning_rate": 1.821802824799355e-05,
      "loss": 2.1944,
      "step": 181440
    },
    {
      "epoch": 0.6361055619808642,
      "grad_norm": 0.5443301200866699,
      "learning_rate": 1.820751410647321e-05,
      "loss": 2.1773,
      "step": 181500
    },
    {
      "epoch": 0.6363158448112711,
      "grad_norm": 0.5454010367393494,
      "learning_rate": 1.8196999964952862e-05,
      "loss": 2.2003,
      "step": 181560
    },
    {
      "epoch": 0.636526127641678,
      "grad_norm": 0.5376167297363281,
      "learning_rate": 1.8186661059124522e-05,
      "loss": 2.183,
      "step": 181620
    },
    {
      "epoch": 0.636736410472085,
      "grad_norm": 0.6612685918807983,
      "learning_rate": 1.8176146917604178e-05,
      "loss": 2.2047,
      "step": 181680
    },
    {
      "epoch": 0.6369466933024919,
      "grad_norm": 0.5321860313415527,
      "learning_rate": 1.8165632776083833e-05,
      "loss": 2.2129,
      "step": 181740
    },
    {
      "epoch": 0.6371569761328988,
      "grad_norm": 0.5190796852111816,
      "learning_rate": 1.815511863456349e-05,
      "loss": 2.1888,
      "step": 181800
    },
    {
      "epoch": 0.6373672589633056,
      "grad_norm": 0.5587844252586365,
      "learning_rate": 1.8144604493043144e-05,
      "loss": 2.206,
      "step": 181860
    },
    {
      "epoch": 0.6375775417937125,
      "grad_norm": 0.5742557048797607,
      "learning_rate": 1.81340903515228e-05,
      "loss": 2.1899,
      "step": 181920
    },
    {
      "epoch": 0.6377878246241194,
      "grad_norm": 0.5507059097290039,
      "learning_rate": 1.8123576210002454e-05,
      "loss": 2.2189,
      "step": 181980
    },
    {
      "epoch": 0.6379981074545263,
      "grad_norm": 0.5660874247550964,
      "learning_rate": 1.811306206848211e-05,
      "loss": 2.1985,
      "step": 182040
    },
    {
      "epoch": 0.6382083902849333,
      "grad_norm": 0.5507509112358093,
      "learning_rate": 1.8102547926961765e-05,
      "loss": 2.191,
      "step": 182100
    },
    {
      "epoch": 0.6384186731153402,
      "grad_norm": 0.6052454710006714,
      "learning_rate": 1.809203378544142e-05,
      "loss": 2.1788,
      "step": 182160
    },
    {
      "epoch": 0.638628955945747,
      "grad_norm": 0.5226848125457764,
      "learning_rate": 1.8081519643921075e-05,
      "loss": 2.1897,
      "step": 182220
    },
    {
      "epoch": 0.6388392387761539,
      "grad_norm": 0.5561570525169373,
      "learning_rate": 1.8071005502400727e-05,
      "loss": 2.1932,
      "step": 182280
    },
    {
      "epoch": 0.6390495216065608,
      "grad_norm": 0.5580832362174988,
      "learning_rate": 1.8060491360880386e-05,
      "loss": 2.1983,
      "step": 182340
    },
    {
      "epoch": 0.6392598044369677,
      "grad_norm": 0.5689629316329956,
      "learning_rate": 1.804997721936004e-05,
      "loss": 2.1837,
      "step": 182400
    },
    {
      "epoch": 0.6394700872673746,
      "grad_norm": 0.5467937588691711,
      "learning_rate": 1.8039463077839693e-05,
      "loss": 2.2017,
      "step": 182460
    },
    {
      "epoch": 0.6396803700977816,
      "grad_norm": 0.5341092348098755,
      "learning_rate": 1.802894893631935e-05,
      "loss": 2.1936,
      "step": 182520
    },
    {
      "epoch": 0.6398906529281884,
      "grad_norm": 0.5286155939102173,
      "learning_rate": 1.8018434794799004e-05,
      "loss": 2.2146,
      "step": 182580
    },
    {
      "epoch": 0.6401009357585953,
      "grad_norm": 0.5773108601570129,
      "learning_rate": 1.800792065327866e-05,
      "loss": 2.2021,
      "step": 182640
    },
    {
      "epoch": 0.6403112185890022,
      "grad_norm": 0.5345368385314941,
      "learning_rate": 1.7997406511758318e-05,
      "loss": 2.1836,
      "step": 182700
    },
    {
      "epoch": 0.6405215014194091,
      "grad_norm": 0.5411863923072815,
      "learning_rate": 1.798689237023797e-05,
      "loss": 2.1921,
      "step": 182760
    },
    {
      "epoch": 0.640731784249816,
      "grad_norm": 0.5300861597061157,
      "learning_rate": 1.7976378228717628e-05,
      "loss": 2.206,
      "step": 182820
    },
    {
      "epoch": 0.6409420670802229,
      "grad_norm": 0.5324180722236633,
      "learning_rate": 1.796586408719728e-05,
      "loss": 2.1792,
      "step": 182880
    },
    {
      "epoch": 0.6411523499106297,
      "grad_norm": 0.6221733689308167,
      "learning_rate": 1.7955349945676935e-05,
      "loss": 2.2037,
      "step": 182940
    },
    {
      "epoch": 0.6413626327410367,
      "grad_norm": 0.5904287695884705,
      "learning_rate": 1.7944835804156594e-05,
      "loss": 2.1877,
      "step": 183000
    },
    {
      "epoch": 0.6415729155714436,
      "grad_norm": 0.610864520072937,
      "learning_rate": 1.7934321662636246e-05,
      "loss": 2.1882,
      "step": 183060
    },
    {
      "epoch": 0.6417831984018505,
      "grad_norm": 0.6096898913383484,
      "learning_rate": 1.79238075211159e-05,
      "loss": 2.1964,
      "step": 183120
    },
    {
      "epoch": 0.6419934812322574,
      "grad_norm": 0.5498456358909607,
      "learning_rate": 1.791329337959556e-05,
      "loss": 2.1888,
      "step": 183180
    },
    {
      "epoch": 0.6422037640626643,
      "grad_norm": 0.5559455156326294,
      "learning_rate": 1.790277923807521e-05,
      "loss": 2.192,
      "step": 183240
    },
    {
      "epoch": 0.6424140468930711,
      "grad_norm": 0.5066351294517517,
      "learning_rate": 1.7892265096554867e-05,
      "loss": 2.1838,
      "step": 183300
    },
    {
      "epoch": 0.642624329723478,
      "grad_norm": 0.5600133538246155,
      "learning_rate": 1.7881750955034522e-05,
      "loss": 2.1902,
      "step": 183360
    },
    {
      "epoch": 0.642834612553885,
      "grad_norm": 0.6039444804191589,
      "learning_rate": 1.7871236813514177e-05,
      "loss": 2.1976,
      "step": 183420
    },
    {
      "epoch": 0.6430448953842919,
      "grad_norm": 0.524412214756012,
      "learning_rate": 1.7860722671993833e-05,
      "loss": 2.1931,
      "step": 183480
    },
    {
      "epoch": 0.6432551782146988,
      "grad_norm": 0.5259726047515869,
      "learning_rate": 1.7850208530473488e-05,
      "loss": 2.2079,
      "step": 183540
    },
    {
      "epoch": 0.6434654610451057,
      "grad_norm": 0.5919510126113892,
      "learning_rate": 1.7839694388953143e-05,
      "loss": 2.1983,
      "step": 183600
    },
    {
      "epoch": 0.6436757438755125,
      "grad_norm": 0.5589147806167603,
      "learning_rate": 1.78291802474328e-05,
      "loss": 2.2145,
      "step": 183660
    },
    {
      "epoch": 0.6438860267059194,
      "grad_norm": 0.581453800201416,
      "learning_rate": 1.781884134160446e-05,
      "loss": 2.1932,
      "step": 183720
    },
    {
      "epoch": 0.6440963095363264,
      "grad_norm": 0.5136606693267822,
      "learning_rate": 1.7808327200084114e-05,
      "loss": 2.1914,
      "step": 183780
    },
    {
      "epoch": 0.6443065923667333,
      "grad_norm": 0.5781887173652649,
      "learning_rate": 1.779781305856377e-05,
      "loss": 2.2011,
      "step": 183840
    },
    {
      "epoch": 0.6445168751971402,
      "grad_norm": 0.5550873875617981,
      "learning_rate": 1.7787298917043425e-05,
      "loss": 2.1915,
      "step": 183900
    },
    {
      "epoch": 0.6447271580275471,
      "grad_norm": 0.5426739454269409,
      "learning_rate": 1.7776784775523077e-05,
      "loss": 2.2125,
      "step": 183960
    },
    {
      "epoch": 0.6449374408579539,
      "grad_norm": 0.5990216732025146,
      "learning_rate": 1.7766270634002736e-05,
      "loss": 2.1797,
      "step": 184020
    },
    {
      "epoch": 0.6451477236883608,
      "grad_norm": 0.5692834258079529,
      "learning_rate": 1.775575649248239e-05,
      "loss": 2.1976,
      "step": 184080
    },
    {
      "epoch": 0.6453580065187677,
      "grad_norm": 0.5651482343673706,
      "learning_rate": 1.7745242350962043e-05,
      "loss": 2.1983,
      "step": 184140
    },
    {
      "epoch": 0.6455682893491747,
      "grad_norm": 0.5360584259033203,
      "learning_rate": 1.77347282094417e-05,
      "loss": 2.2047,
      "step": 184200
    },
    {
      "epoch": 0.6457785721795816,
      "grad_norm": 0.5235934257507324,
      "learning_rate": 1.7724214067921353e-05,
      "loss": 2.1807,
      "step": 184260
    },
    {
      "epoch": 0.6459888550099885,
      "grad_norm": 0.5651029348373413,
      "learning_rate": 1.771369992640101e-05,
      "loss": 2.2096,
      "step": 184320
    },
    {
      "epoch": 0.6461991378403953,
      "grad_norm": 0.5518803000450134,
      "learning_rate": 1.7703185784880667e-05,
      "loss": 2.2053,
      "step": 184380
    },
    {
      "epoch": 0.6464094206708022,
      "grad_norm": 0.5300690531730652,
      "learning_rate": 1.769267164336032e-05,
      "loss": 2.1847,
      "step": 184440
    },
    {
      "epoch": 0.6466197035012091,
      "grad_norm": 0.5407575368881226,
      "learning_rate": 1.7682157501839974e-05,
      "loss": 2.1976,
      "step": 184500
    },
    {
      "epoch": 0.646829986331616,
      "grad_norm": 0.6042479276657104,
      "learning_rate": 1.767164336031963e-05,
      "loss": 2.1695,
      "step": 184560
    },
    {
      "epoch": 0.647040269162023,
      "grad_norm": 0.5453103184700012,
      "learning_rate": 1.7661129218799285e-05,
      "loss": 2.1789,
      "step": 184620
    },
    {
      "epoch": 0.6472505519924299,
      "grad_norm": 0.5787133574485779,
      "learning_rate": 1.765061507727894e-05,
      "loss": 2.1866,
      "step": 184680
    },
    {
      "epoch": 0.6474608348228367,
      "grad_norm": 0.5406185984611511,
      "learning_rate": 1.7640100935758596e-05,
      "loss": 2.1825,
      "step": 184740
    },
    {
      "epoch": 0.6476711176532436,
      "grad_norm": 0.5635714530944824,
      "learning_rate": 1.762958679423825e-05,
      "loss": 2.1914,
      "step": 184800
    },
    {
      "epoch": 0.6478814004836505,
      "grad_norm": 0.5155267715454102,
      "learning_rate": 1.7619072652717906e-05,
      "loss": 2.19,
      "step": 184860
    },
    {
      "epoch": 0.6480916833140574,
      "grad_norm": 0.5049726366996765,
      "learning_rate": 1.760855851119756e-05,
      "loss": 2.1987,
      "step": 184920
    },
    {
      "epoch": 0.6483019661444643,
      "grad_norm": 0.522000253200531,
      "learning_rate": 1.7598044369677217e-05,
      "loss": 2.1838,
      "step": 184980
    },
    {
      "epoch": 0.6485122489748713,
      "grad_norm": 0.566197395324707,
      "learning_rate": 1.7587530228156872e-05,
      "loss": 2.1912,
      "step": 185040
    },
    {
      "epoch": 0.6487225318052781,
      "grad_norm": 0.6304585337638855,
      "learning_rate": 1.7577016086636527e-05,
      "loss": 2.1952,
      "step": 185100
    },
    {
      "epoch": 0.648932814635685,
      "grad_norm": 0.5605672001838684,
      "learning_rate": 1.7566501945116182e-05,
      "loss": 2.2132,
      "step": 185160
    },
    {
      "epoch": 0.6491430974660919,
      "grad_norm": 0.5913508534431458,
      "learning_rate": 1.7555987803595838e-05,
      "loss": 2.1916,
      "step": 185220
    },
    {
      "epoch": 0.6493533802964988,
      "grad_norm": 0.5016937851905823,
      "learning_rate": 1.7545473662075493e-05,
      "loss": 2.1779,
      "step": 185280
    },
    {
      "epoch": 0.6495636631269057,
      "grad_norm": 0.5301057696342468,
      "learning_rate": 1.7534959520555148e-05,
      "loss": 2.1948,
      "step": 185340
    },
    {
      "epoch": 0.6497739459573126,
      "grad_norm": 0.5172202587127686,
      "learning_rate": 1.7524445379034804e-05,
      "loss": 2.1957,
      "step": 185400
    },
    {
      "epoch": 0.6499842287877194,
      "grad_norm": 0.5441404581069946,
      "learning_rate": 1.751393123751446e-05,
      "loss": 2.2083,
      "step": 185460
    },
    {
      "epoch": 0.6501945116181264,
      "grad_norm": 0.5819316506385803,
      "learning_rate": 1.750341709599411e-05,
      "loss": 2.1794,
      "step": 185520
    },
    {
      "epoch": 0.6504047944485333,
      "grad_norm": 0.5801051259040833,
      "learning_rate": 1.749290295447377e-05,
      "loss": 2.1856,
      "step": 185580
    },
    {
      "epoch": 0.6506150772789402,
      "grad_norm": 0.547092854976654,
      "learning_rate": 1.7482388812953425e-05,
      "loss": 2.1984,
      "step": 185640
    },
    {
      "epoch": 0.6508253601093471,
      "grad_norm": 0.5455633997917175,
      "learning_rate": 1.7472049907125085e-05,
      "loss": 2.1923,
      "step": 185700
    },
    {
      "epoch": 0.651035642939754,
      "grad_norm": 0.5476002097129822,
      "learning_rate": 1.746153576560474e-05,
      "loss": 2.1953,
      "step": 185760
    },
    {
      "epoch": 0.6512459257701608,
      "grad_norm": 0.5624036192893982,
      "learning_rate": 1.74511968597764e-05,
      "loss": 2.198,
      "step": 185820
    },
    {
      "epoch": 0.6514562086005677,
      "grad_norm": 0.5351853370666504,
      "learning_rate": 1.7440682718256053e-05,
      "loss": 2.1904,
      "step": 185880
    },
    {
      "epoch": 0.6516664914309747,
      "grad_norm": 0.5587154626846313,
      "learning_rate": 1.743016857673571e-05,
      "loss": 2.1892,
      "step": 185940
    },
    {
      "epoch": 0.6518767742613816,
      "grad_norm": 0.5613061189651489,
      "learning_rate": 1.7419654435215367e-05,
      "loss": 2.1974,
      "step": 186000
    },
    {
      "epoch": 0.6520870570917885,
      "grad_norm": 0.5189383029937744,
      "learning_rate": 1.740914029369502e-05,
      "loss": 2.1953,
      "step": 186060
    },
    {
      "epoch": 0.6522973399221954,
      "grad_norm": 0.5712246298789978,
      "learning_rate": 1.7398626152174677e-05,
      "loss": 2.2143,
      "step": 186120
    },
    {
      "epoch": 0.6525076227526022,
      "grad_norm": 0.5922529697418213,
      "learning_rate": 1.7388112010654333e-05,
      "loss": 2.2,
      "step": 186180
    },
    {
      "epoch": 0.6527179055830091,
      "grad_norm": 0.5634300708770752,
      "learning_rate": 1.7377597869133985e-05,
      "loss": 2.1873,
      "step": 186240
    },
    {
      "epoch": 0.652928188413416,
      "grad_norm": 0.5620796084403992,
      "learning_rate": 1.7367083727613643e-05,
      "loss": 2.1842,
      "step": 186300
    },
    {
      "epoch": 0.653138471243823,
      "grad_norm": 0.49786993861198425,
      "learning_rate": 1.7356569586093295e-05,
      "loss": 2.1809,
      "step": 186360
    },
    {
      "epoch": 0.6533487540742299,
      "grad_norm": 0.5227112174034119,
      "learning_rate": 1.734605544457295e-05,
      "loss": 2.1901,
      "step": 186420
    },
    {
      "epoch": 0.6535590369046367,
      "grad_norm": 0.604454517364502,
      "learning_rate": 1.733554130305261e-05,
      "loss": 2.1968,
      "step": 186480
    },
    {
      "epoch": 0.6537693197350436,
      "grad_norm": 0.5914393663406372,
      "learning_rate": 1.732502716153226e-05,
      "loss": 2.1877,
      "step": 186540
    },
    {
      "epoch": 0.6539796025654505,
      "grad_norm": 0.5598288178443909,
      "learning_rate": 1.7314513020011916e-05,
      "loss": 2.1957,
      "step": 186600
    },
    {
      "epoch": 0.6541898853958574,
      "grad_norm": 0.517230749130249,
      "learning_rate": 1.730399887849157e-05,
      "loss": 2.1898,
      "step": 186660
    },
    {
      "epoch": 0.6544001682262643,
      "grad_norm": 0.6065497398376465,
      "learning_rate": 1.7293484736971227e-05,
      "loss": 2.2012,
      "step": 186720
    },
    {
      "epoch": 0.6546104510566713,
      "grad_norm": 0.5286918878555298,
      "learning_rate": 1.7282970595450882e-05,
      "loss": 2.1883,
      "step": 186780
    },
    {
      "epoch": 0.6548207338870781,
      "grad_norm": 0.5354787111282349,
      "learning_rate": 1.7272456453930537e-05,
      "loss": 2.1809,
      "step": 186840
    },
    {
      "epoch": 0.655031016717485,
      "grad_norm": 0.5585691928863525,
      "learning_rate": 1.7261942312410193e-05,
      "loss": 2.1927,
      "step": 186900
    },
    {
      "epoch": 0.6552412995478919,
      "grad_norm": 0.58662348985672,
      "learning_rate": 1.7251428170889848e-05,
      "loss": 2.2075,
      "step": 186960
    },
    {
      "epoch": 0.6554515823782988,
      "grad_norm": 0.5285871624946594,
      "learning_rate": 1.7240914029369503e-05,
      "loss": 2.1944,
      "step": 187020
    },
    {
      "epoch": 0.6556618652087057,
      "grad_norm": 0.5373408794403076,
      "learning_rate": 1.723039988784916e-05,
      "loss": 2.2032,
      "step": 187080
    },
    {
      "epoch": 0.6558721480391126,
      "grad_norm": 0.5267037153244019,
      "learning_rate": 1.7219885746328814e-05,
      "loss": 2.1964,
      "step": 187140
    },
    {
      "epoch": 0.6560824308695195,
      "grad_norm": 0.5143727660179138,
      "learning_rate": 1.720937160480847e-05,
      "loss": 2.2158,
      "step": 187200
    },
    {
      "epoch": 0.6562927136999264,
      "grad_norm": 0.6170992255210876,
      "learning_rate": 1.7198857463288124e-05,
      "loss": 2.1874,
      "step": 187260
    },
    {
      "epoch": 0.6565029965303333,
      "grad_norm": 0.5920122265815735,
      "learning_rate": 1.7188343321767776e-05,
      "loss": 2.1934,
      "step": 187320
    },
    {
      "epoch": 0.6567132793607402,
      "grad_norm": 0.5491796135902405,
      "learning_rate": 1.7177829180247435e-05,
      "loss": 2.1803,
      "step": 187380
    },
    {
      "epoch": 0.6569235621911471,
      "grad_norm": 0.5345472097396851,
      "learning_rate": 1.716731503872709e-05,
      "loss": 2.1783,
      "step": 187440
    },
    {
      "epoch": 0.657133845021554,
      "grad_norm": 0.5320336222648621,
      "learning_rate": 1.7156800897206742e-05,
      "loss": 2.18,
      "step": 187500
    },
    {
      "epoch": 0.6573441278519608,
      "grad_norm": 0.54201740026474,
      "learning_rate": 1.71462867556864e-05,
      "loss": 2.1915,
      "step": 187560
    },
    {
      "epoch": 0.6575544106823678,
      "grad_norm": 0.5722365975379944,
      "learning_rate": 1.7135772614166053e-05,
      "loss": 2.184,
      "step": 187620
    },
    {
      "epoch": 0.6577646935127747,
      "grad_norm": 0.5633969902992249,
      "learning_rate": 1.7125258472645708e-05,
      "loss": 2.1872,
      "step": 187680
    },
    {
      "epoch": 0.6579749763431816,
      "grad_norm": 0.606042742729187,
      "learning_rate": 1.7114744331125367e-05,
      "loss": 2.178,
      "step": 187740
    },
    {
      "epoch": 0.6581852591735885,
      "grad_norm": 0.520921528339386,
      "learning_rate": 1.710423018960502e-05,
      "loss": 2.1992,
      "step": 187800
    },
    {
      "epoch": 0.6583955420039954,
      "grad_norm": 0.5178558230400085,
      "learning_rate": 1.7093716048084674e-05,
      "loss": 2.1974,
      "step": 187860
    },
    {
      "epoch": 0.6586058248344022,
      "grad_norm": 0.5311471223831177,
      "learning_rate": 1.708320190656433e-05,
      "loss": 2.1847,
      "step": 187920
    },
    {
      "epoch": 0.6588161076648091,
      "grad_norm": 0.5017029643058777,
      "learning_rate": 1.7072687765043984e-05,
      "loss": 2.1965,
      "step": 187980
    },
    {
      "epoch": 0.659026390495216,
      "grad_norm": 0.6602824330329895,
      "learning_rate": 1.706217362352364e-05,
      "loss": 2.1865,
      "step": 188040
    },
    {
      "epoch": 0.659236673325623,
      "grad_norm": 0.6342161297798157,
      "learning_rate": 1.7051659482003295e-05,
      "loss": 2.1896,
      "step": 188100
    },
    {
      "epoch": 0.6594469561560299,
      "grad_norm": 0.5871784090995789,
      "learning_rate": 1.704114534048295e-05,
      "loss": 2.1825,
      "step": 188160
    },
    {
      "epoch": 0.6596572389864368,
      "grad_norm": 0.5610417723655701,
      "learning_rate": 1.7030631198962605e-05,
      "loss": 2.1948,
      "step": 188220
    },
    {
      "epoch": 0.6598675218168436,
      "grad_norm": 0.54507976770401,
      "learning_rate": 1.702011705744226e-05,
      "loss": 2.187,
      "step": 188280
    },
    {
      "epoch": 0.6600778046472505,
      "grad_norm": 0.5555378198623657,
      "learning_rate": 1.7009602915921916e-05,
      "loss": 2.1982,
      "step": 188340
    },
    {
      "epoch": 0.6602880874776574,
      "grad_norm": 0.5375819802284241,
      "learning_rate": 1.699908877440157e-05,
      "loss": 2.1917,
      "step": 188400
    },
    {
      "epoch": 0.6604983703080644,
      "grad_norm": 0.48358219861984253,
      "learning_rate": 1.6988574632881227e-05,
      "loss": 2.1878,
      "step": 188460
    },
    {
      "epoch": 0.6607086531384713,
      "grad_norm": 0.59239262342453,
      "learning_rate": 1.6978060491360882e-05,
      "loss": 2.1837,
      "step": 188520
    },
    {
      "epoch": 0.6609189359688782,
      "grad_norm": 0.5474352836608887,
      "learning_rate": 1.6967546349840534e-05,
      "loss": 2.2024,
      "step": 188580
    },
    {
      "epoch": 0.661129218799285,
      "grad_norm": 0.5559254884719849,
      "learning_rate": 1.6957032208320192e-05,
      "loss": 2.1843,
      "step": 188640
    },
    {
      "epoch": 0.6613395016296919,
      "grad_norm": 0.6271349191665649,
      "learning_rate": 1.6946518066799848e-05,
      "loss": 2.1855,
      "step": 188700
    },
    {
      "epoch": 0.6615497844600988,
      "grad_norm": 0.5410918593406677,
      "learning_rate": 1.69360039252795e-05,
      "loss": 2.1845,
      "step": 188760
    },
    {
      "epoch": 0.6617600672905057,
      "grad_norm": 0.5364717245101929,
      "learning_rate": 1.6925489783759158e-05,
      "loss": 2.1826,
      "step": 188820
    },
    {
      "epoch": 0.6619703501209127,
      "grad_norm": 0.5298910737037659,
      "learning_rate": 1.691497564223881e-05,
      "loss": 2.1894,
      "step": 188880
    },
    {
      "epoch": 0.6621806329513196,
      "grad_norm": 0.5474604964256287,
      "learning_rate": 1.6904461500718465e-05,
      "loss": 2.1728,
      "step": 188940
    },
    {
      "epoch": 0.6623909157817264,
      "grad_norm": 0.5495792627334595,
      "learning_rate": 1.6893947359198124e-05,
      "loss": 2.1935,
      "step": 189000
    },
    {
      "epoch": 0.6626011986121333,
      "grad_norm": 0.5411957502365112,
      "learning_rate": 1.6883433217677776e-05,
      "loss": 2.1797,
      "step": 189060
    },
    {
      "epoch": 0.6628114814425402,
      "grad_norm": 0.5660207271575928,
      "learning_rate": 1.687291907615743e-05,
      "loss": 2.1799,
      "step": 189120
    },
    {
      "epoch": 0.6630217642729471,
      "grad_norm": 0.5661222338676453,
      "learning_rate": 1.686240493463709e-05,
      "loss": 2.1854,
      "step": 189180
    },
    {
      "epoch": 0.663232047103354,
      "grad_norm": 0.6380460262298584,
      "learning_rate": 1.6851890793116742e-05,
      "loss": 2.1977,
      "step": 189240
    },
    {
      "epoch": 0.663442329933761,
      "grad_norm": 0.5798426866531372,
      "learning_rate": 1.6841376651596397e-05,
      "loss": 2.1977,
      "step": 189300
    },
    {
      "epoch": 0.6636526127641678,
      "grad_norm": 0.588367760181427,
      "learning_rate": 1.6830862510076052e-05,
      "loss": 2.1875,
      "step": 189360
    },
    {
      "epoch": 0.6638628955945747,
      "grad_norm": 0.5828709602355957,
      "learning_rate": 1.6820348368555708e-05,
      "loss": 2.1876,
      "step": 189420
    },
    {
      "epoch": 0.6640731784249816,
      "grad_norm": 0.6115942001342773,
      "learning_rate": 1.6809834227035363e-05,
      "loss": 2.201,
      "step": 189480
    },
    {
      "epoch": 0.6642834612553885,
      "grad_norm": 0.6153404116630554,
      "learning_rate": 1.6799320085515018e-05,
      "loss": 2.1897,
      "step": 189540
    },
    {
      "epoch": 0.6644937440857954,
      "grad_norm": 0.6499913930892944,
      "learning_rate": 1.6788805943994673e-05,
      "loss": 2.1946,
      "step": 189600
    },
    {
      "epoch": 0.6647040269162023,
      "grad_norm": 0.5375574827194214,
      "learning_rate": 1.677829180247433e-05,
      "loss": 2.1858,
      "step": 189660
    },
    {
      "epoch": 0.6649143097466091,
      "grad_norm": 0.5913978815078735,
      "learning_rate": 1.6767777660953984e-05,
      "loss": 2.1969,
      "step": 189720
    },
    {
      "epoch": 0.6651245925770161,
      "grad_norm": 0.531616747379303,
      "learning_rate": 1.675726351943364e-05,
      "loss": 2.1766,
      "step": 189780
    },
    {
      "epoch": 0.665334875407423,
      "grad_norm": 0.5702480673789978,
      "learning_rate": 1.6746749377913295e-05,
      "loss": 2.1935,
      "step": 189840
    },
    {
      "epoch": 0.6655451582378299,
      "grad_norm": 0.5366515517234802,
      "learning_rate": 1.6736410472084955e-05,
      "loss": 2.1953,
      "step": 189900
    },
    {
      "epoch": 0.6657554410682368,
      "grad_norm": 0.571426272392273,
      "learning_rate": 1.672589633056461e-05,
      "loss": 2.2015,
      "step": 189960
    }
  ],
  "logging_steps": 60,
  "max_steps": 285330,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.692720539303936e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}