{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 9803,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010200958890135673,
      "grad_norm": 23.863433487076062,
      "learning_rate": 3.058623619371283e-08,
      "loss": 3.0232,
      "step": 10
    },
    {
      "epoch": 0.0020401917780271347,
      "grad_norm": 32.61730608207525,
      "learning_rate": 6.457094307561598e-08,
      "loss": 2.5589,
      "step": 20
    },
    {
      "epoch": 0.003060287667040702,
      "grad_norm": 20.905749548504595,
      "learning_rate": 9.855564995751911e-08,
      "loss": 2.3109,
      "step": 30
    },
    {
      "epoch": 0.004080383556054269,
      "grad_norm": 42.04849732928689,
      "learning_rate": 1.3254035683942224e-07,
      "loss": 2.3109,
      "step": 40
    },
    {
      "epoch": 0.0051004794450678365,
      "grad_norm": 36.60798648292093,
      "learning_rate": 1.665250637213254e-07,
      "loss": 2.7944,
      "step": 50
    },
    {
      "epoch": 0.006120575334081404,
      "grad_norm": 49.921419574512164,
      "learning_rate": 2.0050977060322854e-07,
      "loss": 2.1338,
      "step": 60
    },
    {
      "epoch": 0.007140671223094971,
      "grad_norm": 20.62048890012187,
      "learning_rate": 2.3449447748513168e-07,
      "loss": 2.5731,
      "step": 70
    },
    {
      "epoch": 0.008160767112108539,
      "grad_norm": 26.274330796775203,
      "learning_rate": 2.684791843670348e-07,
      "loss": 2.8828,
      "step": 80
    },
    {
      "epoch": 0.009180863001122105,
      "grad_norm": 20.41350120571662,
      "learning_rate": 3.0246389124893796e-07,
      "loss": 1.9765,
      "step": 90
    },
    {
      "epoch": 0.010200958890135673,
      "grad_norm": 21.7806573738887,
      "learning_rate": 3.364485981308411e-07,
      "loss": 2.3008,
      "step": 100
    },
    {
      "epoch": 0.01122105477914924,
      "grad_norm": 22.347582148609074,
      "learning_rate": 3.7043330501274423e-07,
      "loss": 2.6922,
      "step": 110
    },
    {
      "epoch": 0.012241150668162807,
      "grad_norm": 22.339281756423176,
      "learning_rate": 4.044180118946474e-07,
      "loss": 2.0783,
      "step": 120
    },
    {
      "epoch": 0.013261246557176375,
      "grad_norm": 18.00882469585931,
      "learning_rate": 4.3840271877655056e-07,
      "loss": 1.9041,
      "step": 130
    },
    {
      "epoch": 0.014281342446189942,
      "grad_norm": 29.87805362484696,
      "learning_rate": 4.723874256584537e-07,
      "loss": 2.4358,
      "step": 140
    },
    {
      "epoch": 0.01530143833520351,
      "grad_norm": 23.99904284503815,
      "learning_rate": 5.063721325403568e-07,
      "loss": 1.8461,
      "step": 150
    },
    {
      "epoch": 0.016321534224217078,
      "grad_norm": 26.21234166057312,
      "learning_rate": 5.4035683942226e-07,
      "loss": 1.9981,
      "step": 160
    },
    {
      "epoch": 0.017341630113230644,
      "grad_norm": 25.119213924225146,
      "learning_rate": 5.743415463041631e-07,
      "loss": 2.2417,
      "step": 170
    },
    {
      "epoch": 0.01836172600224421,
      "grad_norm": 41.51086082925132,
      "learning_rate": 6.083262531860663e-07,
      "loss": 1.9795,
      "step": 180
    },
    {
      "epoch": 0.01938182189125778,
      "grad_norm": 15.011174011685128,
      "learning_rate": 6.423109600679693e-07,
      "loss": 2.2117,
      "step": 190
    },
    {
      "epoch": 0.020401917780271346,
      "grad_norm": 25.275011605500755,
      "learning_rate": 6.762956669498725e-07,
      "loss": 1.9294,
      "step": 200
    },
    {
      "epoch": 0.021422013669284912,
      "grad_norm": 17.075571648757318,
      "learning_rate": 7.102803738317756e-07,
      "loss": 1.8741,
      "step": 210
    },
    {
      "epoch": 0.02244210955829848,
      "grad_norm": 21.362424815092922,
      "learning_rate": 7.442650807136788e-07,
      "loss": 2.0184,
      "step": 220
    },
    {
      "epoch": 0.02346220544731205,
      "grad_norm": 17.744903305683344,
      "learning_rate": 7.782497875955819e-07,
      "loss": 1.8482,
      "step": 230
    },
    {
      "epoch": 0.024482301336325615,
      "grad_norm": 31.060221474404933,
      "learning_rate": 8.122344944774851e-07,
      "loss": 1.9851,
      "step": 240
    },
    {
      "epoch": 0.02550239722533918,
      "grad_norm": 28.398626504336214,
      "learning_rate": 8.462192013593883e-07,
      "loss": 1.966,
      "step": 250
    },
    {
      "epoch": 0.02652249311435275,
      "grad_norm": 18.723748822668217,
      "learning_rate": 8.802039082412915e-07,
      "loss": 1.9401,
      "step": 260
    },
    {
      "epoch": 0.027542589003366317,
      "grad_norm": 18.822446115487693,
      "learning_rate": 9.141886151231946e-07,
      "loss": 2.4825,
      "step": 270
    },
    {
      "epoch": 0.028562684892379883,
      "grad_norm": 19.92606022029681,
      "learning_rate": 9.481733220050977e-07,
      "loss": 1.8157,
      "step": 280
    },
    {
      "epoch": 0.02958278078139345,
      "grad_norm": 27.807529224260254,
      "learning_rate": 9.82158028887001e-07,
      "loss": 2.1421,
      "step": 290
    },
    {
      "epoch": 0.03060287667040702,
      "grad_norm": 22.101902897459876,
      "learning_rate": 9.999993842198288e-07,
      "loss": 1.7338,
      "step": 300
    },
    {
      "epoch": 0.031622972559420585,
      "grad_norm": 28.422601044401443,
      "learning_rate": 9.999940622521522e-07,
      "loss": 2.0735,
      "step": 310
    },
    {
      "epoch": 0.032643068448434155,
      "grad_norm": 20.57221723894542,
      "learning_rate": 9.999832819139521e-07,
      "loss": 2.15,
      "step": 320
    },
    {
      "epoch": 0.03366316433744772,
      "grad_norm": 27.076284498519584,
      "learning_rate": 9.999670433229166e-07,
      "loss": 1.9315,
      "step": 330
    },
    {
      "epoch": 0.03468326022646129,
      "grad_norm": 23.565505683434324,
      "learning_rate": 9.999453466563204e-07,
      "loss": 2.091,
      "step": 340
    },
    {
      "epoch": 0.03570335611547486,
      "grad_norm": 31.16939784592795,
      "learning_rate": 9.999181921510229e-07,
      "loss": 1.918,
      "step": 350
    },
    {
      "epoch": 0.03672345200448842,
      "grad_norm": 25.18768118771255,
      "learning_rate": 9.998855801034663e-07,
      "loss": 2.1623,
      "step": 360
    },
    {
      "epoch": 0.03774354789350199,
      "grad_norm": 18.937161816774555,
      "learning_rate": 9.99847510869672e-07,
      "loss": 2.0577,
      "step": 370
    },
    {
      "epoch": 0.03876364378251556,
      "grad_norm": 20.072725168709958,
      "learning_rate": 9.998039848652374e-07,
      "loss": 1.9327,
      "step": 380
    },
    {
      "epoch": 0.03978373967152912,
      "grad_norm": 23.01091948948838,
      "learning_rate": 9.997550025653299e-07,
      "loss": 1.9986,
      "step": 390
    },
    {
      "epoch": 0.04080383556054269,
      "grad_norm": 30.008833493395382,
      "learning_rate": 9.997005645046828e-07,
      "loss": 1.9093,
      "step": 400
    },
    {
      "epoch": 0.041823931449556255,
      "grad_norm": 19.456691180783,
      "learning_rate": 9.996406712775898e-07,
      "loss": 1.8422,
      "step": 410
    },
    {
      "epoch": 0.042844027338569825,
      "grad_norm": 20.002093348130398,
      "learning_rate": 9.995753235378971e-07,
      "loss": 1.7415,
      "step": 420
    },
    {
      "epoch": 0.043864123227583394,
      "grad_norm": 18.92970924547386,
      "learning_rate": 9.99504521998998e-07,
      "loss": 1.721,
      "step": 430
    },
    {
      "epoch": 0.04488421911659696,
      "grad_norm": 16.824233363035983,
      "learning_rate": 9.994282674338232e-07,
      "loss": 1.7061,
      "step": 440
    },
    {
      "epoch": 0.04590431500561053,
      "grad_norm": 26.8064289103971,
      "learning_rate": 9.993465606748341e-07,
      "loss": 2.0372,
      "step": 450
    },
    {
      "epoch": 0.0469244108946241,
      "grad_norm": 18.3467620772012,
      "learning_rate": 9.992594026140131e-07,
      "loss": 2.1637,
      "step": 460
    },
    {
      "epoch": 0.04794450678363766,
      "grad_norm": 29.787192388131537,
      "learning_rate": 9.991667942028533e-07,
      "loss": 1.9214,
      "step": 470
    },
    {
      "epoch": 0.04896460267265123,
      "grad_norm": 14.950631841799353,
      "learning_rate": 9.990687364523484e-07,
      "loss": 1.9218,
      "step": 480
    },
    {
      "epoch": 0.0499846985616648,
      "grad_norm": 15.96048957088105,
      "learning_rate": 9.989652304329827e-07,
      "loss": 1.8849,
      "step": 490
    },
    {
      "epoch": 0.05100479445067836,
      "grad_norm": 23.25559252733434,
      "learning_rate": 9.98856277274718e-07,
      "loss": 2.7678,
      "step": 500
    },
    {
      "epoch": 0.05202489033969193,
      "grad_norm": 16.981772478082267,
      "learning_rate": 9.987418781669815e-07,
      "loss": 1.7649,
      "step": 510
    },
    {
      "epoch": 0.0530449862287055,
      "grad_norm": 22.86195622557256,
      "learning_rate": 9.986220343586536e-07,
      "loss": 2.4645,
      "step": 520
    },
    {
      "epoch": 0.054065082117719064,
      "grad_norm": 20.71703992230199,
      "learning_rate": 9.984967471580538e-07,
      "loss": 2.2606,
      "step": 530
    },
    {
      "epoch": 0.055085178006732634,
      "grad_norm": 28.085771845287837,
      "learning_rate": 9.983660179329262e-07,
      "loss": 1.8932,
      "step": 540
    },
    {
      "epoch": 0.056105273895746204,
      "grad_norm": 23.920991773261413,
      "learning_rate": 9.982298481104248e-07,
      "loss": 2.3481,
      "step": 550
    },
    {
      "epoch": 0.057125369784759766,
      "grad_norm": 141.47769196105102,
      "learning_rate": 9.980882391770979e-07,
      "loss": 2.2395,
      "step": 560
    },
    {
      "epoch": 0.058145465673773336,
      "grad_norm": 28.012602335301963,
      "learning_rate": 9.97941192678872e-07,
      "loss": 2.016,
      "step": 570
    },
    {
      "epoch": 0.0591655615627869,
      "grad_norm": 20.792867426852872,
      "learning_rate": 9.977887102210346e-07,
      "loss": 2.0692,
      "step": 580
    },
    {
      "epoch": 0.06018565745180047,
      "grad_norm": 29.49661292722343,
      "learning_rate": 9.976307934682174e-07,
      "loss": 1.9136,
      "step": 590
    },
    {
      "epoch": 0.06120575334081404,
      "grad_norm": 25.056720428187944,
      "learning_rate": 9.974674441443764e-07,
      "loss": 1.8935,
      "step": 600
    },
    {
      "epoch": 0.0622258492298276,
      "grad_norm": 22.507278587608056,
      "learning_rate": 9.972986640327757e-07,
      "loss": 2.0315,
      "step": 610
    },
    {
      "epoch": 0.06324594511884117,
      "grad_norm": 21.06758645158729,
      "learning_rate": 9.971244549759657e-07,
      "loss": 1.8007,
      "step": 620
    },
    {
      "epoch": 0.06426604100785474,
      "grad_norm": 29.88602195410204,
      "learning_rate": 9.969448188757642e-07,
      "loss": 1.8421,
      "step": 630
    },
    {
      "epoch": 0.06528613689686831,
      "grad_norm": 18.58554081864542,
      "learning_rate": 9.96759757693235e-07,
      "loss": 2.0819,
      "step": 640
    },
    {
      "epoch": 0.06630623278588187,
      "grad_norm": 28.30313795460732,
      "learning_rate": 9.965692734486676e-07,
      "loss": 1.9255,
      "step": 650
    },
    {
      "epoch": 0.06732632867489544,
      "grad_norm": 23.40765460873339,
      "learning_rate": 9.963733682215536e-07,
      "loss": 1.6323,
      "step": 660
    },
    {
      "epoch": 0.068346424563909,
      "grad_norm": 42.647646700249105,
      "learning_rate": 9.96172044150565e-07,
      "loss": 1.7998,
      "step": 670
    },
    {
      "epoch": 0.06936652045292258,
      "grad_norm": 27.092471273520676,
      "learning_rate": 9.959653034335306e-07,
      "loss": 1.6738,
      "step": 680
    },
    {
      "epoch": 0.07038661634193615,
      "grad_norm": 26.61325282836434,
      "learning_rate": 9.957531483274123e-07,
      "loss": 1.9107,
      "step": 690
    },
    {
      "epoch": 0.07140671223094971,
      "grad_norm": 28.159735626915413,
      "learning_rate": 9.955355811482797e-07,
      "loss": 1.8224,
      "step": 700
    },
    {
      "epoch": 0.07242680811996327,
      "grad_norm": 20.738923221278853,
      "learning_rate": 9.953126042712858e-07,
      "loss": 1.6372,
      "step": 710
    },
    {
      "epoch": 0.07344690400897684,
      "grad_norm": 21.537289501877467,
      "learning_rate": 9.950842201306402e-07,
      "loss": 2.0209,
      "step": 720
    },
    {
      "epoch": 0.07446699989799041,
      "grad_norm": 19.18095826540949,
      "learning_rate": 9.948504312195832e-07,
      "loss": 1.8984,
      "step": 730
    },
    {
      "epoch": 0.07548709578700398,
      "grad_norm": 23.64694909729056,
      "learning_rate": 9.946112400903578e-07,
      "loss": 2.4664,
      "step": 740
    },
    {
      "epoch": 0.07650719167601755,
      "grad_norm": 20.562960110046312,
      "learning_rate": 9.943666493541826e-07,
      "loss": 1.8668,
      "step": 750
    },
    {
      "epoch": 0.07752728756503112,
      "grad_norm": 20.423437060647906,
      "learning_rate": 9.941166616812236e-07,
      "loss": 2.1648,
      "step": 760
    },
    {
      "epoch": 0.07854738345404468,
      "grad_norm": 24.117367484568646,
      "learning_rate": 9.938612798005632e-07,
      "loss": 1.9564,
      "step": 770
    },
    {
      "epoch": 0.07956747934305824,
      "grad_norm": 29.77138219841288,
      "learning_rate": 9.93600506500173e-07,
      "loss": 1.8877,
      "step": 780
    },
    {
      "epoch": 0.08058757523207181,
      "grad_norm": 19.239905828226462,
      "learning_rate": 9.933343446268808e-07,
      "loss": 1.8641,
      "step": 790
    },
    {
      "epoch": 0.08160767112108538,
      "grad_norm": 14.249507169267247,
      "learning_rate": 9.930627970863418e-07,
      "loss": 1.8001,
      "step": 800
    },
    {
      "epoch": 0.08262776701009895,
      "grad_norm": 19.579184902915916,
      "learning_rate": 9.927858668430053e-07,
      "loss": 1.8009,
      "step": 810
    },
    {
      "epoch": 0.08364786289911251,
      "grad_norm": 14.302890899225487,
      "learning_rate": 9.925035569200828e-07,
      "loss": 1.874,
      "step": 820
    },
    {
      "epoch": 0.08466795878812608,
      "grad_norm": 13.2891477784319,
      "learning_rate": 9.922158703995153e-07,
      "loss": 1.6299,
      "step": 830
    },
    {
      "epoch": 0.08568805467713965,
      "grad_norm": 20.25195601003889,
      "learning_rate": 9.919228104219394e-07,
      "loss": 2.012,
      "step": 840
    },
    {
      "epoch": 0.08670815056615322,
      "grad_norm": 23.179901692080843,
      "learning_rate": 9.916243801866531e-07,
      "loss": 1.8608,
      "step": 850
    },
    {
      "epoch": 0.08772824645516679,
      "grad_norm": 22.88003416488005,
      "learning_rate": 9.913205829515804e-07,
      "loss": 1.6534,
      "step": 860
    },
    {
      "epoch": 0.08874834234418036,
      "grad_norm": 20.87186839888629,
      "learning_rate": 9.910114220332364e-07,
      "loss": 1.8427,
      "step": 870
    },
    {
      "epoch": 0.08976843823319391,
      "grad_norm": 17.199089374639623,
      "learning_rate": 9.906969008066912e-07,
      "loss": 1.7113,
      "step": 880
    },
    {
      "epoch": 0.09078853412220748,
      "grad_norm": 16.566817453555863,
      "learning_rate": 9.903770227055322e-07,
      "loss": 2.0996,
      "step": 890
    },
    {
      "epoch": 0.09180863001122105,
      "grad_norm": 21.0245894400051,
      "learning_rate": 9.900517912218271e-07,
      "loss": 1.7114,
      "step": 900
    },
    {
      "epoch": 0.09282872590023462,
      "grad_norm": 18.818276484668242,
      "learning_rate": 9.897212099060862e-07,
      "loss": 1.8459,
      "step": 910
    },
    {
      "epoch": 0.0938488217892482,
      "grad_norm": 16.026303904081217,
      "learning_rate": 9.893852823672229e-07,
      "loss": 1.8464,
      "step": 920
    },
    {
      "epoch": 0.09486891767826176,
      "grad_norm": 27.125808362313283,
      "learning_rate": 9.890440122725146e-07,
      "loss": 1.8986,
      "step": 930
    },
    {
      "epoch": 0.09588901356727532,
      "grad_norm": 20.344732097707126,
      "learning_rate": 9.88697403347563e-07,
      "loss": 1.9212,
      "step": 940
    },
    {
      "epoch": 0.09690910945628889,
      "grad_norm": 23.88900929670636,
      "learning_rate": 9.883454593762526e-07,
      "loss": 2.0284,
      "step": 950
    },
    {
      "epoch": 0.09792920534530246,
      "grad_norm": 22.195644517653594,
      "learning_rate": 9.879881842007106e-07,
      "loss": 1.8873,
      "step": 960
    },
    {
      "epoch": 0.09894930123431603,
      "grad_norm": 16.3222708585269,
      "learning_rate": 9.876255817212636e-07,
      "loss": 1.8132,
      "step": 970
    },
    {
      "epoch": 0.0999693971233296,
      "grad_norm": 20.5034374306791,
      "learning_rate": 9.872576558963962e-07,
      "loss": 2.253,
      "step": 980
    },
    {
      "epoch": 0.10098949301234315,
      "grad_norm": 34.53933932854686,
      "learning_rate": 9.868844107427069e-07,
      "loss": 2.0532,
      "step": 990
    },
    {
      "epoch": 0.10200958890135672,
      "grad_norm": 30.345021948737934,
      "learning_rate": 9.865058503348648e-07,
      "loss": 2.2215,
      "step": 1000
    },
    {
      "epoch": 0.1030296847903703,
      "grad_norm": 119.90181962432887,
      "learning_rate": 9.861219788055648e-07,
      "loss": 2.0902,
      "step": 1010
    },
    {
      "epoch": 0.10404978067938386,
      "grad_norm": 23.05314848215449,
      "learning_rate": 9.85732800345483e-07,
      "loss": 1.8953,
      "step": 1020
    },
    {
      "epoch": 0.10506987656839743,
      "grad_norm": 14.6902353683153,
      "learning_rate": 9.853383192032302e-07,
      "loss": 2.0626,
      "step": 1030
    },
    {
      "epoch": 0.106089972457411,
      "grad_norm": 15.465866814470825,
      "learning_rate": 9.849385396853062e-07,
      "loss": 1.9264,
      "step": 1040
    },
    {
      "epoch": 0.10711006834642456,
      "grad_norm": 17.965326106654445,
      "learning_rate": 9.84533466156052e-07,
      "loss": 1.6613,
      "step": 1050
    },
    {
      "epoch": 0.10813016423543813,
      "grad_norm": 22.890470075622876,
      "learning_rate": 9.84123103037603e-07,
      "loss": 1.8367,
      "step": 1060
    },
    {
      "epoch": 0.1091502601244517,
      "grad_norm": 20.662108749162797,
      "learning_rate": 9.837074548098405e-07,
      "loss": 1.8622,
      "step": 1070
    },
    {
      "epoch": 0.11017035601346527,
      "grad_norm": 31.159665911534148,
      "learning_rate": 9.832865260103423e-07,
      "loss": 1.9955,
      "step": 1080
    },
    {
      "epoch": 0.11119045190247884,
      "grad_norm": 28.275798149125535,
      "learning_rate": 9.828603212343337e-07,
      "loss": 2.2043,
      "step": 1090
    },
    {
      "epoch": 0.11221054779149241,
      "grad_norm": 23.09932282513959,
      "learning_rate": 9.82428845134637e-07,
      "loss": 1.8363,
      "step": 1100
    },
    {
      "epoch": 0.11323064368050596,
      "grad_norm": 23.60803481043315,
      "learning_rate": 9.81992102421621e-07,
      "loss": 1.8234,
      "step": 1110
    },
    {
      "epoch": 0.11425073956951953,
      "grad_norm": 20.861825670242254,
      "learning_rate": 9.815500978631498e-07,
      "loss": 1.7806,
      "step": 1120
    },
    {
      "epoch": 0.1152708354585331,
      "grad_norm": 20.873310357797298,
      "learning_rate": 9.811028362845297e-07,
      "loss": 2.3839,
      "step": 1130
    },
    {
      "epoch": 0.11629093134754667,
      "grad_norm": 33.81197869915569,
      "learning_rate": 9.806503225684576e-07,
      "loss": 1.8077,
      "step": 1140
    },
    {
      "epoch": 0.11731102723656024,
      "grad_norm": 19.67009844103025,
      "learning_rate": 9.801925616549671e-07,
      "loss": 1.5672,
      "step": 1150
    },
    {
      "epoch": 0.1183311231255738,
      "grad_norm": 20.45219965096337,
      "learning_rate": 9.79729558541375e-07,
      "loss": 1.8723,
      "step": 1160
    },
    {
      "epoch": 0.11935121901458737,
      "grad_norm": 40.47475908930002,
      "learning_rate": 9.792613182822263e-07,
      "loss": 1.8838,
      "step": 1170
    },
    {
      "epoch": 0.12037131490360094,
      "grad_norm": 27.109561717411868,
      "learning_rate": 9.787878459892395e-07,
      "loss": 1.666,
      "step": 1180
    },
    {
      "epoch": 0.12139141079261451,
      "grad_norm": 19.39151848013294,
      "learning_rate": 9.783091468312506e-07,
      "loss": 1.7246,
      "step": 1190
    },
    {
      "epoch": 0.12241150668162808,
      "grad_norm": 133.92622685216898,
      "learning_rate": 9.77825226034156e-07,
      "loss": 2.2728,
      "step": 1200
    },
    {
      "epoch": 0.12343160257064165,
      "grad_norm": 27.651098049885245,
      "learning_rate": 9.773360888808567e-07,
      "loss": 2.1208,
      "step": 1210
    },
    {
      "epoch": 0.1244516984596552,
      "grad_norm": 17.966140169077416,
      "learning_rate": 9.768417407111995e-07,
      "loss": 1.685,
      "step": 1220
    },
    {
      "epoch": 0.12547179434866879,
      "grad_norm": 19.664438317278428,
      "learning_rate": 9.763421869219198e-07,
      "loss": 1.5492,
      "step": 1230
    },
    {
      "epoch": 0.12649189023768234,
      "grad_norm": 18.604426263284246,
      "learning_rate": 9.758374329665814e-07,
      "loss": 1.9013,
      "step": 1240
    },
    {
      "epoch": 0.1275119861266959,
      "grad_norm": 24.378982996936166,
      "learning_rate": 9.753274843555181e-07,
      "loss": 2.1396,
      "step": 1250
    },
    {
      "epoch": 0.12853208201570948,
      "grad_norm": 19.965861987041155,
      "learning_rate": 9.748123466557732e-07,
      "loss": 1.8524,
      "step": 1260
    },
    {
      "epoch": 0.12955217790472304,
      "grad_norm": 20.922374343490922,
      "learning_rate": 9.742920254910378e-07,
      "loss": 2.0343,
      "step": 1270
    },
    {
      "epoch": 0.13057227379373662,
      "grad_norm": 20.340676801514917,
      "learning_rate": 9.737665265415914e-07,
      "loss": 1.835,
      "step": 1280
    },
    {
      "epoch": 0.13159236968275018,
      "grad_norm": 26.858013112252376,
      "learning_rate": 9.732358555442375e-07,
      "loss": 1.8793,
      "step": 1290
    },
    {
      "epoch": 0.13261246557176373,
      "grad_norm": 21.728872390922554,
      "learning_rate": 9.727000182922432e-07,
      "loss": 1.8421,
      "step": 1300
    },
    {
      "epoch": 0.13363256146077732,
      "grad_norm": 99.38289549621847,
      "learning_rate": 9.72159020635274e-07,
      "loss": 2.2696,
      "step": 1310
    },
    {
      "epoch": 0.13465265734979087,
      "grad_norm": 21.663917971711452,
      "learning_rate": 9.716128684793322e-07,
      "loss": 1.5719,
      "step": 1320
    },
    {
      "epoch": 0.13567275323880446,
      "grad_norm": 38.917494149089265,
      "learning_rate": 9.710615677866894e-07,
      "loss": 1.6967,
      "step": 1330
    },
    {
      "epoch": 0.136692849127818,
      "grad_norm": 23.1844774779042,
      "learning_rate": 9.705051245758242e-07,
      "loss": 1.7595,
      "step": 1340
    },
    {
      "epoch": 0.1377129450168316,
      "grad_norm": 21.479309144716424,
      "learning_rate": 9.699435449213556e-07,
      "loss": 1.8762,
      "step": 1350
    },
    {
      "epoch": 0.13873304090584515,
      "grad_norm": 11.760203139355513,
      "learning_rate": 9.693768349539755e-07,
      "loss": 2.0624,
      "step": 1360
    },
    {
      "epoch": 0.1397531367948587,
      "grad_norm": 27.678225119649902,
      "learning_rate": 9.688050008603835e-07,
      "loss": 1.905,
      "step": 1370
    },
    {
      "epoch": 0.1407732326838723,
      "grad_norm": 20.139634211647362,
      "learning_rate": 9.682280488832188e-07,
      "loss": 1.711,
      "step": 1380
    },
    {
      "epoch": 0.14179332857288585,
      "grad_norm": 41.719184228636905,
      "learning_rate": 9.676459853209913e-07,
      "loss": 1.9367,
      "step": 1390
    },
    {
      "epoch": 0.14281342446189943,
      "grad_norm": 28.715805563284555,
      "learning_rate": 9.670588165280136e-07,
      "loss": 2.045,
      "step": 1400
    },
    {
      "epoch": 0.14383352035091299,
      "grad_norm": 12.79113783309391,
      "learning_rate": 9.664665489143314e-07,
      "loss": 2.0047,
      "step": 1410
    },
    {
      "epoch": 0.14485361623992654,
      "grad_norm": 129.09586947433615,
      "learning_rate": 9.658691889456538e-07,
      "loss": 2.6188,
      "step": 1420
    },
    {
      "epoch": 0.14587371212894013,
      "grad_norm": 17.369812023558932,
      "learning_rate": 9.652667431432822e-07,
      "loss": 1.9292,
      "step": 1430
    },
    {
      "epoch": 0.14689380801795368,
      "grad_norm": 22.978717074422672,
      "learning_rate": 9.646592180840398e-07,
      "loss": 2.0184,
      "step": 1440
    },
    {
      "epoch": 0.14791390390696726,
      "grad_norm": 21.520025549498605,
      "learning_rate": 9.640466204001986e-07,
      "loss": 2.1739,
      "step": 1450
    },
    {
      "epoch": 0.14893399979598082,
      "grad_norm": 23.41623275611073,
      "learning_rate": 9.63428956779409e-07,
      "loss": 1.6089,
      "step": 1460
    },
    {
      "epoch": 0.14995409568499438,
      "grad_norm": 26.048752921198812,
      "learning_rate": 9.628062339646242e-07,
      "loss": 1.6056,
      "step": 1470
    },
    {
      "epoch": 0.15097419157400796,
      "grad_norm": 19.239587464631775,
      "learning_rate": 9.621784587540292e-07,
      "loss": 1.6741,
      "step": 1480
    },
    {
      "epoch": 0.15199428746302152,
      "grad_norm": 12.843025287592997,
      "learning_rate": 9.615456380009643e-07,
      "loss": 1.7748,
      "step": 1490
    },
    {
      "epoch": 0.1530143833520351,
      "grad_norm": 23.187002923145943,
      "learning_rate": 9.609077786138522e-07,
      "loss": 1.5461,
      "step": 1500
    },
    {
      "epoch": 0.15403447924104866,
      "grad_norm": 26.229381992108216,
      "learning_rate": 9.602648875561211e-07,
      "loss": 1.9331,
      "step": 1510
    },
    {
      "epoch": 0.15505457513006224,
      "grad_norm": 19.911306946593058,
      "learning_rate": 9.596169718461297e-07,
      "loss": 1.7527,
      "step": 1520
    },
    {
      "epoch": 0.1560746710190758,
      "grad_norm": 19.460477292133373,
      "learning_rate": 9.5896403855709e-07,
      "loss": 1.8003,
      "step": 1530
    },
    {
      "epoch": 0.15709476690808935,
      "grad_norm": 22.754473143228722,
      "learning_rate": 9.583060948169903e-07,
      "loss": 1.944,
      "step": 1540
    },
    {
      "epoch": 0.15811486279710293,
      "grad_norm": 42.61052908995628,
      "learning_rate": 9.576431478085174e-07,
      "loss": 2.2759,
      "step": 1550
    },
    {
      "epoch": 0.1591349586861165,
      "grad_norm": 16.815636330515154,
      "learning_rate": 9.56975204768978e-07,
      "loss": 1.7898,
      "step": 1560
    },
    {
      "epoch": 0.16015505457513007,
      "grad_norm": 23.648048641863912,
      "learning_rate": 9.563022729902197e-07,
      "loss": 1.8069,
      "step": 1570
    },
    {
      "epoch": 0.16117515046414363,
      "grad_norm": 107.29861970066199,
      "learning_rate": 9.556243598185518e-07,
      "loss": 1.798,
      "step": 1580
    },
    {
      "epoch": 0.16219524635315719,
      "grad_norm": 19.293735210531185,
      "learning_rate": 9.549414726546648e-07,
      "loss": 1.6796,
      "step": 1590
    },
    {
      "epoch": 0.16321534224217077,
      "grad_norm": 19.780186438401085,
      "learning_rate": 9.542536189535497e-07,
      "loss": 1.9374,
      "step": 1600
    },
    {
      "epoch": 0.16423543813118432,
      "grad_norm": 30.85518052994413,
      "learning_rate": 9.535608062244163e-07,
      "loss": 1.7155,
      "step": 1610
    },
    {
      "epoch": 0.1652555340201979,
      "grad_norm": 20.476192194711814,
      "learning_rate": 9.528630420306114e-07,
      "loss": 1.7642,
      "step": 1620
    },
    {
      "epoch": 0.16627562990921146,
      "grad_norm": 15.873680652443559,
      "learning_rate": 9.521603339895368e-07,
      "loss": 1.8006,
      "step": 1630
    },
    {
      "epoch": 0.16729572579822502,
      "grad_norm": 33.22301261634099,
      "learning_rate": 9.51452689772565e-07,
      "loss": 1.6974,
      "step": 1640
    },
    {
      "epoch": 0.1683158216872386,
      "grad_norm": 31.11395815148964,
      "learning_rate": 9.507401171049567e-07,
      "loss": 2.1107,
      "step": 1650
    },
    {
      "epoch": 0.16933591757625216,
      "grad_norm": 16.95819401310335,
      "learning_rate": 9.500226237657754e-07,
      "loss": 1.9435,
      "step": 1660
    },
    {
      "epoch": 0.17035601346526574,
      "grad_norm": 22.035502843542794,
      "learning_rate": 9.493002175878029e-07,
      "loss": 1.825,
      "step": 1670
    },
    {
      "epoch": 0.1713761093542793,
      "grad_norm": 17.741758039501928,
      "learning_rate": 9.485729064574544e-07,
      "loss": 1.7789,
      "step": 1680
    },
    {
      "epoch": 0.17239620524329288,
      "grad_norm": 22.87942286477738,
      "learning_rate": 9.478406983146914e-07,
      "loss": 1.6026,
      "step": 1690
    },
    {
      "epoch": 0.17341630113230644,
      "grad_norm": 32.89324335246328,
      "learning_rate": 9.471036011529354e-07,
      "loss": 1.822,
      "step": 1700
    },
    {
      "epoch": 0.17443639702132,
      "grad_norm": 21.280399123568486,
      "learning_rate": 9.463616230189808e-07,
      "loss": 1.8367,
      "step": 1710
    },
    {
      "epoch": 0.17545649291033358,
      "grad_norm": 27.20313198094194,
      "learning_rate": 9.45614772012907e-07,
      "loss": 1.7817,
      "step": 1720
    },
    {
      "epoch": 0.17647658879934713,
      "grad_norm": 13.223799963182179,
      "learning_rate": 9.448630562879898e-07,
      "loss": 2.0151,
      "step": 1730
    },
    {
      "epoch": 0.17749668468836072,
      "grad_norm": 26.517164138545084,
      "learning_rate": 9.441064840506126e-07,
      "loss": 1.8553,
      "step": 1740
    },
    {
      "epoch": 0.17851678057737427,
      "grad_norm": 15.429890586382319,
      "learning_rate": 9.433450635601766e-07,
      "loss": 1.6974,
      "step": 1750
    },
    {
      "epoch": 0.17953687646638783,
      "grad_norm": 32.04664861790609,
      "learning_rate": 9.42578803129011e-07,
      "loss": 1.7994,
      "step": 1760
    },
    {
      "epoch": 0.1805569723554014,
      "grad_norm": 17.454542539607804,
      "learning_rate": 9.418077111222817e-07,
      "loss": 1.6096,
      "step": 1770
    },
    {
      "epoch": 0.18157706824441497,
      "grad_norm": 24.684742268330275,
      "learning_rate": 9.410317959579006e-07,
      "loss": 2.6363,
      "step": 1780
    },
    {
      "epoch": 0.18259716413342855,
      "grad_norm": 27.40360751651249,
      "learning_rate": 9.402510661064332e-07,
      "loss": 1.6436,
      "step": 1790
    },
    {
      "epoch": 0.1836172600224421,
      "grad_norm": 16.521139997248525,
      "learning_rate": 9.394655300910059e-07,
      "loss": 2.0406,
      "step": 1800
    },
    {
      "epoch": 0.18463735591145566,
      "grad_norm": 22.299758545479776,
      "learning_rate": 9.386751964872143e-07,
      "loss": 1.9436,
      "step": 1810
    },
    {
      "epoch": 0.18565745180046925,
      "grad_norm": 17.505216617503645,
      "learning_rate": 9.378800739230279e-07,
      "loss": 1.7628,
      "step": 1820
    },
    {
      "epoch": 0.1866775476894828,
      "grad_norm": 24.79351798839558,
      "learning_rate": 9.370801710786972e-07,
      "loss": 1.6904,
      "step": 1830
    },
    {
      "epoch": 0.1876976435784964,
      "grad_norm": 33.250977555422494,
      "learning_rate": 9.362754966866577e-07,
      "loss": 2.2315,
      "step": 1840
    },
    {
      "epoch": 0.18871773946750994,
      "grad_norm": 21.85348916357009,
      "learning_rate": 9.354660595314361e-07,
      "loss": 1.6319,
      "step": 1850
    },
    {
      "epoch": 0.18973783535652353,
      "grad_norm": 18.956166519661874,
      "learning_rate": 9.346518684495528e-07,
      "loss": 1.714,
      "step": 1860
    },
    {
      "epoch": 0.19075793124553708,
      "grad_norm": 50.86383298664566,
      "learning_rate": 9.338329323294268e-07,
      "loss": 1.5573,
      "step": 1870
    },
    {
      "epoch": 0.19177802713455064,
      "grad_norm": 22.556285003231782,
      "learning_rate": 9.330092601112777e-07,
      "loss": 1.731,
      "step": 1880
    },
    {
      "epoch": 0.19279812302356422,
      "grad_norm": 35.23700252306999,
      "learning_rate": 9.321808607870286e-07,
      "loss": 1.8926,
      "step": 1890
    },
    {
      "epoch": 0.19381821891257778,
      "grad_norm": 22.237295774699337,
      "learning_rate": 9.31347743400208e-07,
      "loss": 1.6327,
      "step": 1900
    },
    {
      "epoch": 0.19483831480159136,
      "grad_norm": 15.874129708975213,
      "learning_rate": 9.305099170458503e-07,
      "loss": 1.9731,
      "step": 1910
    },
    {
      "epoch": 0.19585841069060492,
      "grad_norm": 26.80456146418729,
      "learning_rate": 9.296673908703978e-07,
      "loss": 2.1076,
      "step": 1920
    },
    {
      "epoch": 0.19687850657961847,
      "grad_norm": 17.898429459696274,
      "learning_rate": 9.288201740715997e-07,
      "loss": 1.6725,
      "step": 1930
    },
    {
      "epoch": 0.19789860246863206,
      "grad_norm": 18.927135017032054,
      "learning_rate": 9.279682758984122e-07,
      "loss": 1.7717,
      "step": 1940
    },
    {
      "epoch": 0.1989186983576456,
      "grad_norm": 32.48437381416646,
      "learning_rate": 9.271117056508976e-07,
      "loss": 1.7586,
      "step": 1950
    },
    {
      "epoch": 0.1999387942466592,
      "grad_norm": 32.35893347047576,
      "learning_rate": 9.262504726801224e-07,
      "loss": 1.7647,
      "step": 1960
    },
    {
      "epoch": 0.20095889013567275,
      "grad_norm": 19.350050512627714,
      "learning_rate": 9.253845863880556e-07,
      "loss": 1.6479,
      "step": 1970
    },
    {
      "epoch": 0.2019789860246863,
      "grad_norm": 20.84528259331096,
      "learning_rate": 9.24514056227466e-07,
      "loss": 1.6362,
      "step": 1980
    },
    {
      "epoch": 0.2029990819136999,
      "grad_norm": 31.45220615985671,
      "learning_rate": 9.236388917018188e-07,
      "loss": 1.7597,
      "step": 1990
    },
    {
      "epoch": 0.20401917780271345,
      "grad_norm": 30.459000381470467,
      "learning_rate": 9.227591023651717e-07,
      "loss": 1.7905,
      "step": 2000
    },
    {
      "epoch": 0.20503927369172703,
      "grad_norm": 20.836662794360603,
      "learning_rate": 9.218746978220714e-07,
      "loss": 1.6241,
      "step": 2010
    },
    {
      "epoch": 0.2060593695807406,
      "grad_norm": 21.225712139479764,
      "learning_rate": 9.209856877274478e-07,
      "loss": 1.6611,
      "step": 2020
    },
    {
      "epoch": 0.20707946546975417,
      "grad_norm": 15.590415681537998,
      "learning_rate": 9.200920817865092e-07,
      "loss": 2.0811,
      "step": 2030
    },
    {
      "epoch": 0.20809956135876773,
      "grad_norm": 17.303381085531154,
      "learning_rate": 9.19193889754636e-07,
      "loss": 1.7194,
      "step": 2040
    },
    {
      "epoch": 0.20911965724778128,
      "grad_norm": 34.41317237729729,
      "learning_rate": 9.182911214372746e-07,
      "loss": 1.8071,
      "step": 2050
    },
    {
      "epoch": 0.21013975313679487,
      "grad_norm": 38.97301171398699,
      "learning_rate": 9.173837866898299e-07,
      "loss": 1.8901,
      "step": 2060
    },
    {
      "epoch": 0.21115984902580842,
      "grad_norm": 29.130870684088883,
      "learning_rate": 9.164718954175582e-07,
      "loss": 1.8154,
      "step": 2070
    },
    {
      "epoch": 0.212179944914822,
      "grad_norm": 21.870562986603385,
      "learning_rate": 9.155554575754583e-07,
      "loss": 1.6404,
      "step": 2080
    },
    {
      "epoch": 0.21320004080383556,
      "grad_norm": 21.466757393028836,
      "learning_rate": 9.146344831681638e-07,
      "loss": 1.5741,
      "step": 2090
    },
    {
      "epoch": 0.21422013669284912,
      "grad_norm": 27.26067076035061,
      "learning_rate": 9.137089822498332e-07,
      "loss": 2.0936,
      "step": 2100
    },
    {
      "epoch": 0.2152402325818627,
      "grad_norm": 21.586682054911268,
      "learning_rate": 9.127789649240403e-07,
      "loss": 2.0239,
      "step": 2110
    },
    {
      "epoch": 0.21626032847087626,
      "grad_norm": 16.430067360054906,
      "learning_rate": 9.118444413436639e-07,
      "loss": 1.7037,
      "step": 2120
    },
    {
      "epoch": 0.21728042435988984,
      "grad_norm": 64.72106203022383,
      "learning_rate": 9.109054217107772e-07,
      "loss": 1.9476,
      "step": 2130
    },
    {
      "epoch": 0.2183005202489034,
      "grad_norm": 20.398839574932314,
      "learning_rate": 9.099619162765361e-07,
      "loss": 1.7626,
      "step": 2140
    },
    {
      "epoch": 0.21932061613791695,
      "grad_norm": 23.77224290093325,
      "learning_rate": 9.090139353410675e-07,
      "loss": 2.0843,
      "step": 2150
    },
    {
      "epoch": 0.22034071202693054,
      "grad_norm": 20.065360083629308,
      "learning_rate": 9.080614892533567e-07,
      "loss": 1.5687,
      "step": 2160
    },
    {
      "epoch": 0.2213608079159441,
      "grad_norm": 31.20772528028073,
      "learning_rate": 9.071045884111341e-07,
      "loss": 1.8746,
      "step": 2170
    },
    {
      "epoch": 0.22238090380495767,
      "grad_norm": 16.06515527375252,
      "learning_rate": 9.06143243260763e-07,
      "loss": 1.9742,
      "step": 2180
    },
    {
      "epoch": 0.22340099969397123,
      "grad_norm": 20.542531329180434,
      "learning_rate": 9.051774642971237e-07,
      "loss": 1.7584,
      "step": 2190
    },
    {
      "epoch": 0.22442109558298481,
      "grad_norm": 19.633677288309446,
      "learning_rate": 9.042072620635004e-07,
      "loss": 1.6316,
      "step": 2200
    },
    {
      "epoch": 0.22544119147199837,
      "grad_norm": 25.60355283277907,
      "learning_rate": 9.032326471514655e-07,
      "loss": 1.6781,
      "step": 2210
    },
    {
      "epoch": 0.22646128736101193,
      "grad_norm": 24.004384635976272,
      "learning_rate": 9.022536302007638e-07,
      "loss": 2.2888,
      "step": 2220
    },
    {
      "epoch": 0.2274813832500255,
      "grad_norm": 23.977739827220496,
      "learning_rate": 9.012702218991969e-07,
      "loss": 1.6604,
      "step": 2230
    },
    {
      "epoch": 0.22850147913903907,
      "grad_norm": 20.698706033904543,
      "learning_rate": 9.00282432982506e-07,
      "loss": 1.9301,
      "step": 2240
    },
    {
      "epoch": 0.22952157502805265,
      "grad_norm": 36.95877531420871,
      "learning_rate": 8.99290274234255e-07,
      "loss": 1.7318,
      "step": 2250
    },
    {
      "epoch": 0.2305416709170662,
      "grad_norm": 22.292623623321983,
      "learning_rate": 8.982937564857125e-07,
      "loss": 1.6536,
      "step": 2260
    },
    {
      "epoch": 0.23156176680607976,
      "grad_norm": 23.301822503921784,
      "learning_rate": 8.972928906157342e-07,
      "loss": 1.7061,
      "step": 2270
    },
    {
      "epoch": 0.23258186269509334,
      "grad_norm": 20.562557377021463,
      "learning_rate": 8.962876875506431e-07,
      "loss": 1.8299,
      "step": 2280
    },
    {
      "epoch": 0.2336019585841069,
      "grad_norm": 15.167336457951198,
      "learning_rate": 8.952781582641111e-07,
      "loss": 1.6433,
      "step": 2290
    },
    {
      "epoch": 0.23462205447312048,
      "grad_norm": 25.72651707719158,
      "learning_rate": 8.942643137770391e-07,
      "loss": 1.9772,
      "step": 2300
    },
    {
      "epoch": 0.23564215036213404,
      "grad_norm": 31.21445137261395,
      "learning_rate": 8.932461651574364e-07,
      "loss": 1.788,
      "step": 2310
    },
    {
      "epoch": 0.2366622462511476,
      "grad_norm": 24.17577670856053,
      "learning_rate": 8.922237235202992e-07,
      "loss": 1.681,
      "step": 2320
    },
    {
      "epoch": 0.23768234214016118,
      "grad_norm": 32.9108920388422,
      "learning_rate": 8.91197000027491e-07,
      "loss": 1.61,
      "step": 2330
    },
    {
      "epoch": 0.23870243802917473,
      "grad_norm": 29.99338444946993,
      "learning_rate": 8.901660058876196e-07,
      "loss": 1.7744,
      "step": 2340
    },
    {
      "epoch": 0.23972253391818832,
      "grad_norm": 22.780125046327182,
      "learning_rate": 8.891307523559144e-07,
      "loss": 1.5905,
      "step": 2350
    },
    {
      "epoch": 0.24074262980720187,
      "grad_norm": 27.080167023035152,
      "learning_rate": 8.880912507341044e-07,
      "loss": 1.5962,
      "step": 2360
    },
    {
      "epoch": 0.24176272569621546,
      "grad_norm": 16.608145594225554,
      "learning_rate": 8.870475123702946e-07,
      "loss": 1.6814,
      "step": 2370
    },
    {
      "epoch": 0.24278282158522901,
      "grad_norm": 29.473612741064418,
      "learning_rate": 8.859995486588414e-07,
      "loss": 1.733,
      "step": 2380
    },
    {
      "epoch": 0.24380291747424257,
      "grad_norm": 18.775119422338562,
      "learning_rate": 8.849473710402293e-07,
      "loss": 2.0408,
      "step": 2390
    },
    {
      "epoch": 0.24482301336325615,
      "grad_norm": 18.849921176813794,
      "learning_rate": 8.838909910009456e-07,
      "loss": 1.7561,
      "step": 2400
    },
    {
      "epoch": 0.2458431092522697,
      "grad_norm": 19.654959532412967,
      "learning_rate": 8.82830420073354e-07,
      "loss": 1.7697,
      "step": 2410
    },
    {
      "epoch": 0.2468632051412833,
      "grad_norm": 19.292667901339268,
      "learning_rate": 8.817656698355705e-07,
      "loss": 1.738,
      "step": 2420
    },
    {
      "epoch": 0.24788330103029685,
      "grad_norm": 23.243705125204766,
      "learning_rate": 8.806967519113358e-07,
      "loss": 1.7883,
      "step": 2430
    },
    {
      "epoch": 0.2489033969193104,
      "grad_norm": 20.79797041922201,
      "learning_rate": 8.796236779698887e-07,
      "loss": 1.6558,
      "step": 2440
    },
    {
      "epoch": 0.249923492808324,
      "grad_norm": 29.647972286108892,
      "learning_rate": 8.785464597258385e-07,
      "loss": 2.162,
      "step": 2450
    },
    {
      "epoch": 0.25094358869733757,
      "grad_norm": 16.454363367319694,
      "learning_rate": 8.774651089390375e-07,
      "loss": 1.9191,
      "step": 2460
    },
    {
      "epoch": 0.2519636845863511,
      "grad_norm": 21.280469898716845,
      "learning_rate": 8.763796374144524e-07,
      "loss": 1.5506,
      "step": 2470
    },
    {
      "epoch": 0.2529837804753647,
      "grad_norm": 29.07063778229445,
      "learning_rate": 8.752900570020356e-07,
      "loss": 1.8518,
      "step": 2480
    },
    {
      "epoch": 0.25400387636437827,
      "grad_norm": 21.32064512741683,
      "learning_rate": 8.741963795965954e-07,
      "loss": 1.6652,
      "step": 2490
    },
    {
      "epoch": 0.2550239722533918,
      "grad_norm": 21.738171403387206,
      "learning_rate": 8.730986171376666e-07,
      "loss": 1.8269,
      "step": 2500
    },
    {
      "epoch": 0.2560440681424054,
      "grad_norm": 19.139779203926732,
      "learning_rate": 8.719967816093803e-07,
      "loss": 2.0439,
      "step": 2510
    },
    {
      "epoch": 0.25706416403141896,
      "grad_norm": 25.513160886893395,
      "learning_rate": 8.708908850403323e-07,
      "loss": 1.7981,
      "step": 2520
    },
    {
      "epoch": 0.25808425992043255,
      "grad_norm": 66.88513223244281,
      "learning_rate": 8.697809395034523e-07,
      "loss": 1.929,
      "step": 2530
    },
    {
      "epoch": 0.2591043558094461,
      "grad_norm": 19.942821716791418,
      "learning_rate": 8.686669571158722e-07,
      "loss": 1.872,
      "step": 2540
    },
    {
      "epoch": 0.26012445169845966,
      "grad_norm": 20.080305529697835,
      "learning_rate": 8.675489500387941e-07,
      "loss": 1.7366,
      "step": 2550
    },
    {
      "epoch": 0.26114454758747324,
      "grad_norm": 17.433667415109518,
      "learning_rate": 8.664269304773562e-07,
      "loss": 1.9505,
      "step": 2560
    },
    {
      "epoch": 0.26216464347648677,
      "grad_norm": 32.825292218495406,
      "learning_rate": 8.653009106805011e-07,
      "loss": 1.8672,
      "step": 2570
    },
    {
      "epoch": 0.26318473936550035,
      "grad_norm": 21.800144594808202,
      "learning_rate": 8.641709029408415e-07,
      "loss": 1.6437,
      "step": 2580
    },
    {
      "epoch": 0.26420483525451394,
      "grad_norm": 21.860731004203117,
      "learning_rate": 8.630369195945255e-07,
      "loss": 1.8997,
      "step": 2590
    },
    {
      "epoch": 0.26522493114352746,
      "grad_norm": 23.96511399372432,
      "learning_rate": 8.618989730211029e-07,
      "loss": 1.7695,
      "step": 2600
    },
    {
      "epoch": 0.26624502703254105,
      "grad_norm": 19.63414843717023,
      "learning_rate": 8.607570756433892e-07,
      "loss": 2.2508,
      "step": 2610
    },
    {
      "epoch": 0.26726512292155463,
      "grad_norm": 18.376314676119627,
      "learning_rate": 8.596112399273303e-07,
      "loss": 1.6267,
      "step": 2620
    },
    {
      "epoch": 0.2682852188105682,
      "grad_norm": 18.90685091254953,
      "learning_rate": 8.584614783818667e-07,
      "loss": 1.6508,
      "step": 2630
    },
    {
      "epoch": 0.26930531469958174,
      "grad_norm": 21.07560284878359,
      "learning_rate": 8.573078035587966e-07,
      "loss": 2.0577,
      "step": 2640
    },
    {
      "epoch": 0.2703254105885953,
      "grad_norm": 26.785113560169396,
      "learning_rate": 8.561502280526385e-07,
      "loss": 1.8901,
      "step": 2650
    },
    {
      "epoch": 0.2713455064776089,
      "grad_norm": 24.996504416630785,
      "learning_rate": 8.549887645004948e-07,
      "loss": 2.1266,
      "step": 2660
    },
    {
      "epoch": 0.27236560236662244,
      "grad_norm": 18.36579969820268,
      "learning_rate": 8.53823425581913e-07,
      "loss": 1.5104,
      "step": 2670
    },
    {
      "epoch": 0.273385698255636,
      "grad_norm": 19.237357823638984,
      "learning_rate": 8.526542240187471e-07,
      "loss": 1.7455,
      "step": 2680
    },
    {
      "epoch": 0.2744057941446496,
      "grad_norm": 55.978497246594245,
      "learning_rate": 8.514811725750196e-07,
      "loss": 1.6565,
      "step": 2690
    },
    {
      "epoch": 0.2754258900336632,
      "grad_norm": 26.646263900270046,
      "learning_rate": 8.503042840567811e-07,
      "loss": 1.8521,
      "step": 2700
    },
    {
      "epoch": 0.2764459859226767,
      "grad_norm": 27.324276111743366,
      "learning_rate": 8.491235713119718e-07,
      "loss": 1.8768,
      "step": 2710
    },
    {
      "epoch": 0.2774660818116903,
      "grad_norm": 22.61959710333234,
      "learning_rate": 8.479390472302796e-07,
      "loss": 1.8053,
      "step": 2720
    },
    {
      "epoch": 0.2784861777007039,
      "grad_norm": 23.802600914595835,
      "learning_rate": 8.467507247430013e-07,
      "loss": 1.5646,
      "step": 2730
    },
    {
      "epoch": 0.2795062735897174,
      "grad_norm": 33.13803576380966,
      "learning_rate": 8.455586168228994e-07,
      "loss": 1.7877,
      "step": 2740
    },
    {
      "epoch": 0.280526369478731,
      "grad_norm": 16.703965030127602,
      "learning_rate": 8.44362736484062e-07,
      "loss": 1.7068,
      "step": 2750
    },
    {
      "epoch": 0.2815464653677446,
      "grad_norm": 21.57242224959867,
      "learning_rate": 8.431630967817602e-07,
      "loss": 2.0958,
      "step": 2760
    },
    {
      "epoch": 0.2825665612567581,
      "grad_norm": 28.23530462005677,
      "learning_rate": 8.419597108123053e-07,
      "loss": 1.6524,
      "step": 2770
    },
    {
      "epoch": 0.2835866571457717,
      "grad_norm": 22.86560047831079,
      "learning_rate": 8.407525917129066e-07,
      "loss": 1.8376,
      "step": 2780
    },
    {
      "epoch": 0.2846067530347853,
      "grad_norm": 21.42467230044563,
      "learning_rate": 8.395417526615269e-07,
      "loss": 1.8883,
      "step": 2790
    },
    {
      "epoch": 0.28562684892379886,
      "grad_norm": 20.43372734978494,
      "learning_rate": 8.3832720687674e-07,
      "loss": 1.8525,
      "step": 2800
    },
    {
      "epoch": 0.2866469448128124,
      "grad_norm": 17.501981643605948,
      "learning_rate": 8.371089676175847e-07,
      "loss": 1.4422,
      "step": 2810
    },
    {
      "epoch": 0.28766704070182597,
      "grad_norm": 31.271073228017492,
      "learning_rate": 8.358870481834217e-07,
      "loss": 1.7305,
      "step": 2820
    },
    {
      "epoch": 0.28868713659083955,
      "grad_norm": 16.154637129747027,
      "learning_rate": 8.346614619137875e-07,
      "loss": 1.4987,
      "step": 2830
    },
    {
      "epoch": 0.2897072324798531,
      "grad_norm": 16.15415255242616,
      "learning_rate": 8.334322221882489e-07,
      "loss": 1.7707,
      "step": 2840
    },
    {
      "epoch": 0.29072732836886667,
      "grad_norm": 23.85481900238291,
      "learning_rate": 8.321993424262567e-07,
      "loss": 1.7265,
      "step": 2850
    },
    {
      "epoch": 0.29174742425788025,
      "grad_norm": 24.310079134510126,
      "learning_rate": 8.309628360870004e-07,
      "loss": 1.9511,
      "step": 2860
    },
    {
      "epoch": 0.29276752014689383,
      "grad_norm": 19.8007786325566,
      "learning_rate": 8.297227166692594e-07,
      "loss": 1.6476,
      "step": 2870
    },
    {
      "epoch": 0.29378761603590736,
      "grad_norm": 37.16514159273586,
      "learning_rate": 8.284789977112572e-07,
      "loss": 1.7638,
      "step": 2880
    },
    {
      "epoch": 0.29480771192492095,
      "grad_norm": 29.050076371495,
      "learning_rate": 8.272316927905127e-07,
      "loss": 1.6335,
      "step": 2890
    },
    {
      "epoch": 0.29582780781393453,
      "grad_norm": 20.60279222198699,
      "learning_rate": 8.259808155236927e-07,
      "loss": 1.6835,
      "step": 2900
    },
    {
      "epoch": 0.29684790370294806,
      "grad_norm": 32.68261500213739,
      "learning_rate": 8.247263795664623e-07,
      "loss": 2.126,
      "step": 2910
    },
    {
      "epoch": 0.29786799959196164,
      "grad_norm": 26.39389022376176,
      "learning_rate": 8.234683986133367e-07,
      "loss": 1.6937,
      "step": 2920
    },
    {
      "epoch": 0.2988880954809752,
      "grad_norm": 15.76513060111746,
      "learning_rate": 8.222068863975313e-07,
      "loss": 1.6766,
      "step": 2930
    },
    {
      "epoch": 0.29990819136998875,
      "grad_norm": 23.98934824484318,
      "learning_rate": 8.20941856690812e-07,
      "loss": 1.9212,
      "step": 2940
    },
    {
      "epoch": 0.30092828725900234,
      "grad_norm": 19.891987217024877,
      "learning_rate": 8.196733233033442e-07,
      "loss": 1.7095,
      "step": 2950
    },
    {
      "epoch": 0.3019483831480159,
      "grad_norm": 16.053266097094404,
      "learning_rate": 8.18401300083543e-07,
      "loss": 1.5561,
      "step": 2960
    },
    {
      "epoch": 0.3029684790370295,
      "grad_norm": 22.22443022563515,
      "learning_rate": 8.171258009179216e-07,
      "loss": 1.8487,
      "step": 2970
    },
    {
      "epoch": 0.30398857492604303,
      "grad_norm": 14.402269788096929,
      "learning_rate": 8.158468397309393e-07,
      "loss": 1.6095,
      "step": 2980
    },
    {
      "epoch": 0.3050086708150566,
      "grad_norm": 26.398947024578078,
      "learning_rate": 8.145644304848497e-07,
      "loss": 1.5372,
      "step": 2990
    },
    {
      "epoch": 0.3060287667040702,
      "grad_norm": 15.58920862352573,
      "learning_rate": 8.132785871795493e-07,
      "loss": 1.6421,
      "step": 3000
    },
    {
      "epoch": 0.3070488625930837,
      "grad_norm": 19.41745155967458,
      "learning_rate": 8.119893238524229e-07,
      "loss": 1.8313,
      "step": 3010
    },
    {
      "epoch": 0.3080689584820973,
      "grad_norm": 20.46644757686343,
      "learning_rate": 8.106966545781914e-07,
      "loss": 1.6337,
      "step": 3020
    },
    {
      "epoch": 0.3090890543711109,
      "grad_norm": 22.14228610597878,
      "learning_rate": 8.094005934687583e-07,
      "loss": 1.6386,
      "step": 3030
    },
    {
      "epoch": 0.3101091502601245,
      "grad_norm": 19.628647204909942,
      "learning_rate": 8.081011546730549e-07,
      "loss": 1.8201,
      "step": 3040
    },
    {
      "epoch": 0.311129246149138,
      "grad_norm": 19.604437480430438,
      "learning_rate": 8.067983523768869e-07,
      "loss": 1.6729,
      "step": 3050
    },
    {
      "epoch": 0.3121493420381516,
      "grad_norm": 22.72096591072684,
      "learning_rate": 8.054922008027783e-07,
      "loss": 1.8259,
      "step": 3060
    },
    {
      "epoch": 0.3131694379271652,
      "grad_norm": 19.775516758276876,
      "learning_rate": 8.041827142098167e-07,
      "loss": 1.7052,
      "step": 3070
    },
    {
      "epoch": 0.3141895338161787,
      "grad_norm": 21.455959346841677,
      "learning_rate": 8.028699068934981e-07,
      "loss": 1.887,
      "step": 3080
    },
    {
      "epoch": 0.3152096297051923,
      "grad_norm": 19.029050609727655,
      "learning_rate": 8.015537931855703e-07,
      "loss": 1.7535,
      "step": 3090
    },
    {
      "epoch": 0.31622972559420587,
      "grad_norm": 19.793219395568602,
      "learning_rate": 8.002343874538763e-07,
      "loss": 1.7776,
      "step": 3100
    },
    {
      "epoch": 0.3172498214832194,
      "grad_norm": 22.987842551134168,
      "learning_rate": 7.989117041021981e-07,
      "loss": 1.4253,
      "step": 3110
    },
    {
      "epoch": 0.318269917372233,
      "grad_norm": 29.202007904005438,
      "learning_rate": 7.975857575700986e-07,
      "loss": 2.0207,
      "step": 3120
    },
    {
      "epoch": 0.31929001326124656,
      "grad_norm": 23.01695871201165,
      "learning_rate": 7.962565623327646e-07,
      "loss": 1.6949,
      "step": 3130
    },
    {
      "epoch": 0.32031010915026015,
      "grad_norm": 25.456187601583725,
      "learning_rate": 7.949241329008488e-07,
      "loss": 1.756,
      "step": 3140
    },
    {
      "epoch": 0.3213302050392737,
      "grad_norm": 18.380667998512976,
      "learning_rate": 7.935884838203109e-07,
      "loss": 1.7511,
      "step": 3150
    },
    {
      "epoch": 0.32235030092828726,
      "grad_norm": 28.821988973756103,
      "learning_rate": 7.922496296722594e-07,
      "loss": 1.643,
      "step": 3160
    },
    {
      "epoch": 0.32337039681730084,
      "grad_norm": 26.283965482095564,
      "learning_rate": 7.909075850727916e-07,
      "loss": 2.3418,
      "step": 3170
    },
    {
      "epoch": 0.32439049270631437,
      "grad_norm": 18.37742978572714,
      "learning_rate": 7.895623646728349e-07,
      "loss": 1.6467,
      "step": 3180
    },
    {
      "epoch": 0.32541058859532795,
      "grad_norm": 24.717405186001496,
      "learning_rate": 7.882139831579865e-07,
      "loss": 1.8655,
      "step": 3190
    },
    {
      "epoch": 0.32643068448434154,
      "grad_norm": 17.793837488871265,
      "learning_rate": 7.868624552483531e-07,
      "loss": 1.8593,
      "step": 3200
    },
    {
      "epoch": 0.3274507803733551,
      "grad_norm": 25.835863962813505,
      "learning_rate": 7.855077956983903e-07,
      "loss": 1.575,
      "step": 3210
    },
    {
      "epoch": 0.32847087626236865,
      "grad_norm": 33.30580932489692,
      "learning_rate": 7.841500192967408e-07,
      "loss": 1.5508,
      "step": 3220
    },
    {
      "epoch": 0.32949097215138223,
      "grad_norm": 81.86677547849884,
      "learning_rate": 7.827891408660744e-07,
      "loss": 2.0876,
      "step": 3230
    },
    {
      "epoch": 0.3305110680403958,
      "grad_norm": 23.613428284016344,
      "learning_rate": 7.814251752629247e-07,
      "loss": 2.245,
      "step": 3240
    },
    {
      "epoch": 0.33153116392940934,
      "grad_norm": 23.370309369420966,
      "learning_rate": 7.800581373775279e-07,
      "loss": 1.8363,
      "step": 3250
    },
    {
      "epoch": 0.33255125981842293,
      "grad_norm": 14.002963341236308,
      "learning_rate": 7.786880421336599e-07,
      "loss": 1.6792,
      "step": 3260
    },
    {
      "epoch": 0.3335713557074365,
      "grad_norm": 16.40271464325043,
      "learning_rate": 7.773149044884731e-07,
      "loss": 1.5285,
      "step": 3270
    },
    {
      "epoch": 0.33459145159645004,
      "grad_norm": 17.096218852849372,
      "learning_rate": 7.759387394323341e-07,
      "loss": 1.7363,
      "step": 3280
    },
    {
      "epoch": 0.3356115474854636,
      "grad_norm": 19.190963996461363,
      "learning_rate": 7.745595619886581e-07,
      "loss": 1.5342,
      "step": 3290
    },
    {
      "epoch": 0.3366316433744772,
      "grad_norm": 15.964445437052554,
      "learning_rate": 7.731773872137475e-07,
      "loss": 1.5818,
      "step": 3300
    },
    {
      "epoch": 0.3376517392634908,
      "grad_norm": 18.664269231847342,
      "learning_rate": 7.717922301966253e-07,
      "loss": 1.6883,
      "step": 3310
    },
    {
      "epoch": 0.3386718351525043,
      "grad_norm": 23.1136551783112,
      "learning_rate": 7.704041060588718e-07,
      "loss": 1.6288,
      "step": 3320
    },
    {
      "epoch": 0.3396919310415179,
      "grad_norm": 13.774478800670353,
      "learning_rate": 7.690130299544583e-07,
      "loss": 1.7374,
      "step": 3330
    },
    {
      "epoch": 0.3407120269305315,
      "grad_norm": 20.996728098783063,
      "learning_rate": 7.676190170695834e-07,
      "loss": 2.0672,
      "step": 3340
    },
    {
      "epoch": 0.341732122819545,
      "grad_norm": 19.533993178325485,
      "learning_rate": 7.662220826225049e-07,
      "loss": 1.769,
      "step": 3350
    },
    {
      "epoch": 0.3427522187085586,
      "grad_norm": 17.947153092939487,
      "learning_rate": 7.648222418633757e-07,
      "loss": 1.73,
      "step": 3360
    },
    {
      "epoch": 0.3437723145975722,
      "grad_norm": 16.743189247713843,
      "learning_rate": 7.634195100740764e-07,
      "loss": 1.4,
      "step": 3370
    },
    {
      "epoch": 0.34479241048658577,
      "grad_norm": 18.95282917511057,
      "learning_rate": 7.620139025680486e-07,
      "loss": 2.1816,
      "step": 3380
    },
    {
      "epoch": 0.3458125063755993,
      "grad_norm": 21.08202053214691,
      "learning_rate": 7.606054346901276e-07,
      "loss": 1.7751,
      "step": 3390
    },
    {
      "epoch": 0.3468326022646129,
      "grad_norm": 19.611830988514413,
      "learning_rate": 7.591941218163751e-07,
      "loss": 1.6254,
      "step": 3400
    },
    {
      "epoch": 0.34785269815362646,
      "grad_norm": 25.121322524315445,
      "learning_rate": 7.577799793539115e-07,
      "loss": 1.8351,
      "step": 3410
    },
    {
      "epoch": 0.34887279404264,
      "grad_norm": 199.6082415363255,
      "learning_rate": 7.563630227407467e-07,
      "loss": 2.039,
      "step": 3420
    },
    {
      "epoch": 0.3498928899316536,
      "grad_norm": 27.740257502093968,
      "learning_rate": 7.549432674456131e-07,
      "loss": 2.0507,
      "step": 3430
    },
    {
      "epoch": 0.35091298582066716,
      "grad_norm": 16.402830963177735,
      "learning_rate": 7.535207289677958e-07,
      "loss": 1.6298,
      "step": 3440
    },
    {
      "epoch": 0.3519330817096807,
      "grad_norm": 23.305813242492984,
      "learning_rate": 7.520954228369632e-07,
      "loss": 1.6047,
      "step": 3450
    },
    {
      "epoch": 0.35295317759869427,
      "grad_norm": 22.890411145867205,
      "learning_rate": 7.506673646129982e-07,
      "loss": 1.576,
      "step": 3460
    },
    {
      "epoch": 0.35397327348770785,
      "grad_norm": 17.68813022312286,
      "learning_rate": 7.492365698858274e-07,
      "loss": 1.9312,
      "step": 3470
    },
    {
      "epoch": 0.35499336937672143,
      "grad_norm": 18.48668438865949,
      "learning_rate": 7.478030542752522e-07,
      "loss": 1.7113,
      "step": 3480
    },
    {
      "epoch": 0.35601346526573496,
      "grad_norm": 15.085586130291428,
      "learning_rate": 7.463668334307771e-07,
      "loss": 1.8197,
      "step": 3490
    },
    {
      "epoch": 0.35703356115474855,
      "grad_norm": 19.01878874363063,
      "learning_rate": 7.449279230314391e-07,
      "loss": 1.7552,
      "step": 3500
    },
    {
      "epoch": 0.35805365704376213,
      "grad_norm": 22.887155339093127,
      "learning_rate": 7.434863387856374e-07,
      "loss": 1.9644,
      "step": 3510
    },
    {
      "epoch": 0.35907375293277566,
      "grad_norm": 26.731845390006317,
      "learning_rate": 7.420420964309603e-07,
      "loss": 1.8602,
      "step": 3520
    },
    {
      "epoch": 0.36009384882178924,
      "grad_norm": 25.40096552691888,
      "learning_rate": 7.405952117340151e-07,
      "loss": 1.4797,
      "step": 3530
    },
    {
      "epoch": 0.3611139447108028,
      "grad_norm": 15.749910779782068,
      "learning_rate": 7.39145700490255e-07,
      "loss": 1.8502,
      "step": 3540
    },
    {
      "epoch": 0.3621340405998164,
      "grad_norm": 31.485003970793873,
      "learning_rate": 7.376935785238065e-07,
      "loss": 1.6267,
      "step": 3550
    },
    {
      "epoch": 0.36315413648882994,
      "grad_norm": 28.234942559465274,
      "learning_rate": 7.362388616872979e-07,
      "loss": 1.6657,
      "step": 3560
    },
    {
      "epoch": 0.3641742323778435,
      "grad_norm": 20.58217532312202,
      "learning_rate": 7.347815658616842e-07,
      "loss": 1.8823,
      "step": 3570
    },
    {
      "epoch": 0.3651943282668571,
      "grad_norm": 42.3794980034766,
      "learning_rate": 7.333217069560757e-07,
      "loss": 1.7069,
      "step": 3580
    },
    {
      "epoch": 0.36621442415587063,
      "grad_norm": 33.02400024759091,
      "learning_rate": 7.318593009075634e-07,
      "loss": 1.8266,
      "step": 3590
    },
    {
      "epoch": 0.3672345200448842,
      "grad_norm": 22.017609511527148,
      "learning_rate": 7.303943636810449e-07,
      "loss": 1.8938,
      "step": 3600
    },
    {
      "epoch": 0.3682546159338978,
      "grad_norm": 16.636294438054172,
      "learning_rate": 7.289269112690506e-07,
      "loss": 1.5259,
      "step": 3610
    },
    {
      "epoch": 0.36927471182291133,
      "grad_norm": 26.83193702313107,
      "learning_rate": 7.274569596915685e-07,
      "loss": 1.8923,
      "step": 3620
    },
    {
      "epoch": 0.3702948077119249,
      "grad_norm": 28.304457851944374,
      "learning_rate": 7.259845249958703e-07,
      "loss": 1.5357,
      "step": 3630
    },
    {
      "epoch": 0.3713149036009385,
      "grad_norm": 21.08783619574136,
      "learning_rate": 7.245096232563346e-07,
      "loss": 1.5608,
      "step": 3640
    },
    {
      "epoch": 0.3723349994899521,
      "grad_norm": 22.97453807231992,
      "learning_rate": 7.230322705742735e-07,
      "loss": 1.6042,
      "step": 3650
    },
    {
      "epoch": 0.3733550953789656,
      "grad_norm": 37.080559799061724,
      "learning_rate": 7.215524830777549e-07,
      "loss": 1.8352,
      "step": 3660
    },
    {
      "epoch": 0.3743751912679792,
      "grad_norm": 23.2654825075334,
      "learning_rate": 7.200702769214275e-07,
      "loss": 1.7155,
      "step": 3670
    },
    {
      "epoch": 0.3753952871569928,
      "grad_norm": 30.918717316143688,
      "learning_rate": 7.185856682863445e-07,
      "loss": 1.6369,
      "step": 3680
    },
    {
      "epoch": 0.3764153830460063,
      "grad_norm": 16.93120147850529,
      "learning_rate": 7.170986733797862e-07,
      "loss": 1.7429,
      "step": 3690
    },
    {
      "epoch": 0.3774354789350199,
      "grad_norm": 18.07025983667393,
      "learning_rate": 7.156093084350838e-07,
      "loss": 1.7083,
      "step": 3700
    },
    {
      "epoch": 0.37845557482403347,
      "grad_norm": 25.091070094030044,
      "learning_rate": 7.141175897114417e-07,
      "loss": 1.7783,
      "step": 3710
    },
    {
      "epoch": 0.37947567071304705,
      "grad_norm": 29.76466101433264,
      "learning_rate": 7.126235334937605e-07,
      "loss": 2.2298,
      "step": 3720
    },
    {
      "epoch": 0.3804957666020606,
      "grad_norm": 21.239736383241993,
      "learning_rate": 7.111271560924585e-07,
      "loss": 1.4927,
      "step": 3730
    },
    {
      "epoch": 0.38151586249107416,
      "grad_norm": 21.68883761882214,
      "learning_rate": 7.096284738432945e-07,
      "loss": 1.6477,
      "step": 3740
    },
    {
      "epoch": 0.38253595838008775,
      "grad_norm": 24.98493363045178,
      "learning_rate": 7.081275031071887e-07,
      "loss": 1.5521,
      "step": 3750
    },
    {
      "epoch": 0.3835560542691013,
      "grad_norm": 14.797260462502908,
      "learning_rate": 7.066242602700445e-07,
      "loss": 1.7537,
      "step": 3760
    },
    {
      "epoch": 0.38457615015811486,
      "grad_norm": 27.000361194591452,
      "learning_rate": 7.051187617425698e-07,
      "loss": 1.7058,
      "step": 3770
    },
    {
      "epoch": 0.38559624604712844,
      "grad_norm": 21.851303514335736,
      "learning_rate": 7.03611023960097e-07,
      "loss": 1.622,
      "step": 3780
    },
    {
      "epoch": 0.38661634193614197,
      "grad_norm": 16.289314156846537,
      "learning_rate": 7.021010633824047e-07,
      "loss": 1.4868,
      "step": 3790
    },
    {
      "epoch": 0.38763643782515556,
      "grad_norm": 24.06161129908453,
      "learning_rate": 7.005888964935374e-07,
      "loss": 1.8744,
      "step": 3800
    },
    {
      "epoch": 0.38865653371416914,
      "grad_norm": 23.100191144155293,
      "learning_rate": 6.990745398016252e-07,
      "loss": 1.9124,
      "step": 3810
    },
    {
      "epoch": 0.3896766296031827,
      "grad_norm": 16.846039894745555,
      "learning_rate": 6.975580098387045e-07,
      "loss": 1.6536,
      "step": 3820
    },
    {
      "epoch": 0.39069672549219625,
      "grad_norm": 22.926184852294178,
      "learning_rate": 6.960393231605368e-07,
      "loss": 1.8023,
      "step": 3830
    },
    {
      "epoch": 0.39171682138120983,
      "grad_norm": 20.174272620044174,
      "learning_rate": 6.945184963464278e-07,
      "loss": 1.6647,
      "step": 3840
    },
    {
      "epoch": 0.3927369172702234,
      "grad_norm": 23.66652766840316,
      "learning_rate": 6.92995545999048e-07,
      "loss": 1.7096,
      "step": 3850
    },
    {
      "epoch": 0.39375701315923695,
      "grad_norm": 18.91469913827273,
      "learning_rate": 6.914704887442484e-07,
      "loss": 1.5769,
      "step": 3860
    },
    {
      "epoch": 0.39477710904825053,
      "grad_norm": 23.047401360029575,
      "learning_rate": 6.899433412308826e-07,
      "loss": 1.8632,
      "step": 3870
    },
    {
      "epoch": 0.3957972049372641,
      "grad_norm": 16.907744051768756,
      "learning_rate": 6.884141201306221e-07,
      "loss": 1.6644,
      "step": 3880
    },
    {
      "epoch": 0.3968173008262777,
      "grad_norm": 21.26796913794435,
      "learning_rate": 6.868828421377761e-07,
      "loss": 1.5639,
      "step": 3890
    },
    {
      "epoch": 0.3978373967152912,
      "grad_norm": 28.244711144624997,
      "learning_rate": 6.853495239691086e-07,
      "loss": 1.9057,
      "step": 3900
    },
    {
      "epoch": 0.3988574926043048,
      "grad_norm": 20.23338404827545,
      "learning_rate": 6.838141823636554e-07,
      "loss": 1.6153,
      "step": 3910
    },
    {
      "epoch": 0.3998775884933184,
      "grad_norm": 18.621287018583235,
      "learning_rate": 6.822768340825426e-07,
      "loss": 1.9381,
      "step": 3920
    },
    {
      "epoch": 0.4008976843823319,
      "grad_norm": 20.237450088247275,
      "learning_rate": 6.807374959088023e-07,
      "loss": 1.7513,
      "step": 3930
    },
    {
      "epoch": 0.4019177802713455,
      "grad_norm": 23.274558167989564,
      "learning_rate": 6.791961846471904e-07,
      "loss": 1.7893,
      "step": 3940
    },
    {
      "epoch": 0.4029378761603591,
      "grad_norm": 25.243835376479794,
      "learning_rate": 6.776529171240027e-07,
      "loss": 1.6405,
      "step": 3950
    },
    {
      "epoch": 0.4039579720493726,
      "grad_norm": 14.768044171952967,
      "learning_rate": 6.761077101868912e-07,
      "loss": 1.8954,
      "step": 3960
    },
    {
      "epoch": 0.4049780679383862,
      "grad_norm": 24.916592983829112,
      "learning_rate": 6.745605807046804e-07,
      "loss": 1.7229,
      "step": 3970
    },
    {
      "epoch": 0.4059981638273998,
      "grad_norm": 50.48241204387922,
      "learning_rate": 6.730115455671822e-07,
      "loss": 1.5982,
      "step": 3980
    },
    {
      "epoch": 0.40701825971641337,
      "grad_norm": 23.45463937103853,
      "learning_rate": 6.714606216850134e-07,
      "loss": 1.8433,
      "step": 3990
    },
    {
      "epoch": 0.4080383556054269,
      "grad_norm": 32.081316775347574,
      "learning_rate": 6.699078259894093e-07,
      "loss": 1.5608,
      "step": 4000
    },
    {
      "epoch": 0.4090584514944405,
      "grad_norm": 19.345361357648386,
      "learning_rate": 6.683531754320397e-07,
      "loss": 1.6628,
      "step": 4010
    },
    {
      "epoch": 0.41007854738345406,
      "grad_norm": 22.71059162799784,
      "learning_rate": 6.66796686984824e-07,
      "loss": 1.4794,
      "step": 4020
    },
    {
      "epoch": 0.4110986432724676,
      "grad_norm": 24.101699215171397,
      "learning_rate": 6.652383776397448e-07,
      "loss": 1.5948,
      "step": 4030
    },
    {
      "epoch": 0.4121187391614812,
      "grad_norm": 32.44852475925317,
      "learning_rate": 6.636782644086641e-07,
      "loss": 1.5768,
      "step": 4040
    },
    {
      "epoch": 0.41313883505049476,
      "grad_norm": 24.386913203250337,
      "learning_rate": 6.621163643231362e-07,
      "loss": 1.6134,
      "step": 4050
    },
    {
      "epoch": 0.41415893093950834,
      "grad_norm": 22.01619663077069,
      "learning_rate": 6.605526944342223e-07,
      "loss": 1.5732,
      "step": 4060
    },
    {
      "epoch": 0.41517902682852187,
      "grad_norm": 23.150410653081572,
      "learning_rate": 6.589872718123044e-07,
      "loss": 1.7321,
      "step": 4070
    },
    {
      "epoch": 0.41619912271753545,
      "grad_norm": 24.652049107485926,
      "learning_rate": 6.574201135468988e-07,
      "loss": 1.8203,
      "step": 4080
    },
    {
      "epoch": 0.41721921860654904,
      "grad_norm": 18.995244838880648,
      "learning_rate": 6.558512367464698e-07,
      "loss": 2.1288,
      "step": 4090
    },
    {
      "epoch": 0.41823931449556256,
      "grad_norm": 19.38363276649229,
      "learning_rate": 6.542806585382426e-07,
      "loss": 1.9226,
      "step": 4100
    },
    {
      "epoch": 0.41925941038457615,
      "grad_norm": 18.04818975470266,
      "learning_rate": 6.52708396068016e-07,
      "loss": 1.761,
      "step": 4110
    },
    {
      "epoch": 0.42027950627358973,
      "grad_norm": 19.35487514949365,
      "learning_rate": 6.511344664999764e-07,
      "loss": 1.7642,
      "step": 4120
    },
    {
      "epoch": 0.42129960216260326,
      "grad_norm": 18.71997966396092,
      "learning_rate": 6.495588870165092e-07,
      "loss": 1.6827,
      "step": 4130
    },
    {
      "epoch": 0.42231969805161684,
      "grad_norm": 18.18412536093287,
      "learning_rate": 6.479816748180122e-07,
      "loss": 1.5543,
      "step": 4140
    },
    {
      "epoch": 0.4233397939406304,
      "grad_norm": 21.261235334184423,
      "learning_rate": 6.464028471227066e-07,
      "loss": 1.7974,
      "step": 4150
    },
    {
      "epoch": 0.424359889829644,
      "grad_norm": 21.95377850354024,
      "learning_rate": 6.448224211664506e-07,
      "loss": 1.7959,
      "step": 4160
    },
    {
      "epoch": 0.42537998571865754,
      "grad_norm": 16.734550147246225,
      "learning_rate": 6.432404142025496e-07,
      "loss": 1.5948,
      "step": 4170
    },
    {
      "epoch": 0.4264000816076711,
      "grad_norm": 19.032332250163236,
      "learning_rate": 6.416568435015695e-07,
      "loss": 1.651,
      "step": 4180
    },
    {
      "epoch": 0.4274201774966847,
      "grad_norm": 20.347305781013986,
      "learning_rate": 6.400717263511464e-07,
      "loss": 1.6155,
      "step": 4190
    },
    {
      "epoch": 0.42844027338569823,
      "grad_norm": 24.06717899990057,
      "learning_rate": 6.384850800557994e-07,
      "loss": 1.544,
      "step": 4200
    },
    {
      "epoch": 0.4294603692747118,
      "grad_norm": 22.135451020637916,
      "learning_rate": 6.368969219367413e-07,
      "loss": 1.5816,
      "step": 4210
    },
    {
      "epoch": 0.4304804651637254,
      "grad_norm": 14.741435247481006,
      "learning_rate": 6.353072693316884e-07,
      "loss": 1.8857,
      "step": 4220
    },
    {
      "epoch": 0.431500561052739,
      "grad_norm": 28.526633700552384,
      "learning_rate": 6.337161395946731e-07,
      "loss": 2.2205,
      "step": 4230
    },
    {
      "epoch": 0.4325206569417525,
      "grad_norm": 20.503606980885763,
      "learning_rate": 6.321235500958526e-07,
      "loss": 1.8561,
      "step": 4240
    },
    {
      "epoch": 0.4335407528307661,
      "grad_norm": 13.246116093524794,
      "learning_rate": 6.305295182213206e-07,
      "loss": 1.6088,
      "step": 4250
    },
    {
      "epoch": 0.4345608487197797,
      "grad_norm": 13.101878073701537,
      "learning_rate": 6.289340613729172e-07,
      "loss": 1.6656,
      "step": 4260
    },
    {
      "epoch": 0.4355809446087932,
      "grad_norm": 23.118686543209552,
      "learning_rate": 6.273371969680381e-07,
      "loss": 1.6951,
      "step": 4270
    },
    {
      "epoch": 0.4366010404978068,
      "grad_norm": 19.71217350060953,
      "learning_rate": 6.257389424394459e-07,
      "loss": 1.7617,
      "step": 4280
    },
    {
      "epoch": 0.4376211363868204,
      "grad_norm": 17.646175944610878,
      "learning_rate": 6.241393152350784e-07,
      "loss": 1.7165,
      "step": 4290
    },
    {
      "epoch": 0.4386412322758339,
      "grad_norm": 21.090842224246312,
      "learning_rate": 6.225383328178591e-07,
      "loss": 1.942,
      "step": 4300
    },
    {
      "epoch": 0.4396613281648475,
      "grad_norm": 27.138679780950476,
      "learning_rate": 6.209360126655059e-07,
      "loss": 2.1189,
      "step": 4310
    },
    {
      "epoch": 0.44068142405386107,
      "grad_norm": 20.623474854838467,
      "learning_rate": 6.193323722703407e-07,
      "loss": 1.4351,
      "step": 4320
    },
    {
      "epoch": 0.44170151994287465,
      "grad_norm": 22.803974498269323,
      "learning_rate": 6.177274291390984e-07,
      "loss": 1.5569,
      "step": 4330
    },
    {
      "epoch": 0.4427216158318882,
      "grad_norm": 17.534004780644807,
      "learning_rate": 6.161212007927354e-07,
      "loss": 1.5976,
      "step": 4340
    },
    {
      "epoch": 0.44374171172090177,
      "grad_norm": 20.883222215828617,
      "learning_rate": 6.145137047662393e-07,
      "loss": 1.6486,
      "step": 4350
    },
    {
      "epoch": 0.44476180760991535,
      "grad_norm": 21.849478591313932,
      "learning_rate": 6.129049586084359e-07,
      "loss": 1.8565,
      "step": 4360
    },
    {
      "epoch": 0.4457819034989289,
      "grad_norm": 17.231374876263576,
      "learning_rate": 6.112949798817994e-07,
      "loss": 1.6468,
      "step": 4370
    },
    {
      "epoch": 0.44680199938794246,
      "grad_norm": 24.902184291482726,
      "learning_rate": 6.09683786162259e-07,
      "loss": 2.0914,
      "step": 4380
    },
    {
      "epoch": 0.44782209527695604,
      "grad_norm": 20.63012124147058,
      "learning_rate": 6.080713950390084e-07,
      "loss": 1.6626,
      "step": 4390
    },
    {
      "epoch": 0.44884219116596963,
      "grad_norm": 21.488028861578556,
      "learning_rate": 6.064578241143129e-07,
      "loss": 1.7062,
      "step": 4400
    },
    {
      "epoch": 0.44986228705498316,
      "grad_norm": 30.417691525818796,
      "learning_rate": 6.048430910033177e-07,
      "loss": 2.3387,
      "step": 4410
    },
    {
      "epoch": 0.45088238294399674,
      "grad_norm": 21.41683407709701,
      "learning_rate": 6.032272133338553e-07,
      "loss": 1.8522,
      "step": 4420
    },
    {
      "epoch": 0.4519024788330103,
      "grad_norm": 21.129271449582312,
      "learning_rate": 6.016102087462532e-07,
      "loss": 2.1335,
      "step": 4430
    },
    {
      "epoch": 0.45292257472202385,
      "grad_norm": 21.277289106026597,
      "learning_rate": 5.999920948931415e-07,
      "loss": 1.6965,
      "step": 4440
    },
    {
      "epoch": 0.45394267061103744,
      "grad_norm": 21.34933642132836,
      "learning_rate": 5.983728894392601e-07,
      "loss": 1.6326,
      "step": 4450
    },
    {
      "epoch": 0.454962766500051,
      "grad_norm": 21.327330521187033,
      "learning_rate": 5.967526100612651e-07,
      "loss": 1.6943,
      "step": 4460
    },
    {
      "epoch": 0.45598286238906455,
      "grad_norm": 30.48997028048971,
      "learning_rate": 5.951312744475375e-07,
      "loss": 1.7056,
      "step": 4470
    },
    {
      "epoch": 0.45700295827807813,
      "grad_norm": 16.171726503014177,
      "learning_rate": 5.935089002979878e-07,
      "loss": 1.5826,
      "step": 4480
    },
    {
      "epoch": 0.4580230541670917,
      "grad_norm": 14.746897684652618,
      "learning_rate": 5.918855053238656e-07,
      "loss": 1.5686,
      "step": 4490
    },
    {
      "epoch": 0.4590431500561053,
      "grad_norm": 17.43623390701527,
      "learning_rate": 5.902611072475635e-07,
      "loss": 1.6163,
      "step": 4500
    },
    {
      "epoch": 0.4600632459451188,
      "grad_norm": 17.447298393538933,
      "learning_rate": 5.886357238024252e-07,
      "loss": 1.3748,
      "step": 4510
    },
    {
      "epoch": 0.4610833418341324,
      "grad_norm": 25.66748338323143,
      "learning_rate": 5.870093727325519e-07,
      "loss": 1.7892,
      "step": 4520
    },
    {
      "epoch": 0.462103437723146,
      "grad_norm": 28.142203473574078,
      "learning_rate": 5.853820717926074e-07,
      "loss": 1.6199,
      "step": 4530
    },
    {
      "epoch": 0.4631235336121595,
      "grad_norm": 30.26399632112453,
      "learning_rate": 5.83753838747626e-07,
      "loss": 1.6124,
      "step": 4540
    },
    {
      "epoch": 0.4641436295011731,
      "grad_norm": 14.066491879063047,
      "learning_rate": 5.821246913728173e-07,
      "loss": 1.6517,
      "step": 4550
    },
    {
      "epoch": 0.4651637253901867,
      "grad_norm": 25.28017563268791,
      "learning_rate": 5.804946474533724e-07,
      "loss": 1.7264,
      "step": 4560
    },
    {
      "epoch": 0.4661838212792003,
      "grad_norm": 13.978821741063973,
      "learning_rate": 5.788637247842701e-07,
      "loss": 2.0431,
      "step": 4570
    },
    {
      "epoch": 0.4672039171682138,
      "grad_norm": 25.844222076599042,
      "learning_rate": 5.772319411700823e-07,
      "loss": 2.078,
      "step": 4580
    },
    {
      "epoch": 0.4682240130572274,
      "grad_norm": 18.509270903687913,
      "learning_rate": 5.755993144247797e-07,
      "loss": 1.5999,
      "step": 4590
    },
    {
      "epoch": 0.46924410894624097,
      "grad_norm": 35.499934739271005,
      "learning_rate": 5.739658623715372e-07,
      "loss": 1.8381,
      "step": 4600
    },
    {
      "epoch": 0.4702642048352545,
      "grad_norm": 19.254910293375485,
      "learning_rate": 5.723316028425399e-07,
      "loss": 1.8021,
      "step": 4610
    },
    {
      "epoch": 0.4712843007242681,
      "grad_norm": 22.11391280340059,
      "learning_rate": 5.706965536787874e-07,
      "loss": 1.592,
      "step": 4620
    },
    {
      "epoch": 0.47230439661328166,
      "grad_norm": 19.50972733575566,
      "learning_rate": 5.690607327299003e-07,
      "loss": 1.5854,
      "step": 4630
    },
    {
      "epoch": 0.4733244925022952,
      "grad_norm": 23.55823443962975,
      "learning_rate": 5.674241578539244e-07,
      "loss": 1.8163,
      "step": 4640
    },
    {
      "epoch": 0.4743445883913088,
      "grad_norm": 25.88668065664756,
      "learning_rate": 5.657868469171357e-07,
      "loss": 1.7823,
      "step": 4650
    },
    {
      "epoch": 0.47536468428032236,
      "grad_norm": 19.688863771916314,
      "learning_rate": 5.641488177938462e-07,
      "loss": 1.9999,
      "step": 4660
    },
    {
      "epoch": 0.47638478016933594,
      "grad_norm": 17.148505337693656,
      "learning_rate": 5.62510088366208e-07,
      "loss": 2.0364,
      "step": 4670
    },
    {
      "epoch": 0.47740487605834947,
      "grad_norm": 38.987558216733355,
      "learning_rate": 5.608706765240182e-07,
      "loss": 1.6462,
      "step": 4680
    },
    {
      "epoch": 0.47842497194736305,
      "grad_norm": 24.8955088393481,
      "learning_rate": 5.592306001645242e-07,
      "loss": 1.5636,
      "step": 4690
    },
    {
      "epoch": 0.47944506783637664,
      "grad_norm": 20.70678152499259,
      "learning_rate": 5.575898771922271e-07,
      "loss": 1.9318,
      "step": 4700
    },
    {
      "epoch": 0.48046516372539017,
      "grad_norm": 24.132774064896505,
      "learning_rate": 5.559485255186878e-07,
      "loss": 1.8031,
      "step": 4710
    },
    {
      "epoch": 0.48148525961440375,
      "grad_norm": 26.36404049492687,
      "learning_rate": 5.543065630623301e-07,
      "loss": 1.7167,
      "step": 4720
    },
    {
      "epoch": 0.48250535550341733,
      "grad_norm": 17.824327277312417,
      "learning_rate": 5.52664007748246e-07,
      "loss": 1.6143,
      "step": 4730
    },
    {
      "epoch": 0.4835254513924309,
      "grad_norm": 30.00593227196709,
      "learning_rate": 5.510208775079991e-07,
      "loss": 1.5507,
      "step": 4740
    },
    {
      "epoch": 0.48454554728144444,
      "grad_norm": 19.211331802154554,
      "learning_rate": 5.493771902794302e-07,
      "loss": 1.6428,
      "step": 4750
    },
    {
      "epoch": 0.48556564317045803,
      "grad_norm": 26.138373900650013,
      "learning_rate": 5.4773296400646e-07,
      "loss": 1.727,
      "step": 4760
    },
    {
      "epoch": 0.4865857390594716,
      "grad_norm": 21.944997530504445,
      "learning_rate": 5.460882166388941e-07,
      "loss": 1.6417,
      "step": 4770
    },
    {
      "epoch": 0.48760583494848514,
      "grad_norm": 34.35241044153906,
      "learning_rate": 5.444429661322269e-07,
      "loss": 1.8839,
      "step": 4780
    },
    {
      "epoch": 0.4886259308374987,
      "grad_norm": 17.13438651748986,
      "learning_rate": 5.427972304474451e-07,
      "loss": 1.695,
      "step": 4790
    },
    {
      "epoch": 0.4896460267265123,
      "grad_norm": 22.457227939708634,
      "learning_rate": 5.411510275508327e-07,
      "loss": 1.6168,
      "step": 4800
    },
    {
      "epoch": 0.49066612261552583,
      "grad_norm": 21.30188707419009,
      "learning_rate": 5.39504375413774e-07,
      "loss": 1.939,
      "step": 4810
    },
    {
      "epoch": 0.4916862185045394,
      "grad_norm": 16.717440351880935,
      "learning_rate": 5.378572920125571e-07,
      "loss": 1.6379,
      "step": 4820
    },
    {
      "epoch": 0.492706314393553,
      "grad_norm": 23.609540031196865,
      "learning_rate": 5.362097953281788e-07,
      "loss": 1.6724,
      "step": 4830
    },
    {
      "epoch": 0.4937264102825666,
      "grad_norm": 27.62063749968345,
      "learning_rate": 5.34561903346147e-07,
      "loss": 1.5221,
      "step": 4840
    },
    {
      "epoch": 0.4947465061715801,
      "grad_norm": 17.995895587415315,
      "learning_rate": 5.329136340562858e-07,
      "loss": 1.6932,
      "step": 4850
    },
    {
      "epoch": 0.4957666020605937,
      "grad_norm": 20.34330553055779,
      "learning_rate": 5.312650054525375e-07,
      "loss": 1.5359,
      "step": 4860
    },
    {
      "epoch": 0.4967866979496073,
      "grad_norm": 23.70772412858853,
      "learning_rate": 5.296160355327675e-07,
      "loss": 1.4627,
      "step": 4870
    },
    {
      "epoch": 0.4978067938386208,
      "grad_norm": 21.64212788242307,
      "learning_rate": 5.279667422985675e-07,
      "loss": 1.6306,
      "step": 4880
    },
    {
      "epoch": 0.4988268897276344,
      "grad_norm": 25.564331981680812,
      "learning_rate": 5.263171437550581e-07,
      "loss": 1.6154,
      "step": 4890
    },
    {
      "epoch": 0.499846985616648,
      "grad_norm": 20.166800919476323,
      "learning_rate": 5.246672579106935e-07,
      "loss": 1.8586,
      "step": 4900
    },
    {
      "epoch": 0.5008670815056615,
      "grad_norm": 20.376782182941593,
      "learning_rate": 5.23017102777064e-07,
      "loss": 1.6741,
      "step": 4910
    },
    {
      "epoch": 0.5018871773946751,
      "grad_norm": 20.06697957717277,
      "learning_rate": 5.213666963686998e-07,
      "loss": 1.5166,
      "step": 4920
    },
    {
      "epoch": 0.5029072732836887,
      "grad_norm": 24.365408830127386,
      "learning_rate": 5.197160567028743e-07,
      "loss": 1.841,
      "step": 4930
    },
    {
      "epoch": 0.5039273691727022,
      "grad_norm": 21.62827099099707,
      "learning_rate": 5.180652017994072e-07,
      "loss": 2.0242,
      "step": 4940
    },
    {
      "epoch": 0.5049474650617158,
      "grad_norm": 31.819183548078,
      "learning_rate": 5.164141496804682e-07,
      "loss": 1.9652,
      "step": 4950
    },
    {
      "epoch": 0.5059675609507294,
      "grad_norm": 21.65926106452547,
      "learning_rate": 5.147629183703795e-07,
      "loss": 1.5916,
      "step": 4960
    },
    {
      "epoch": 0.5069876568397429,
      "grad_norm": 19.273741951209225,
      "learning_rate": 5.1311152589542e-07,
      "loss": 1.5782,
      "step": 4970
    },
    {
      "epoch": 0.5080077527287565,
      "grad_norm": 22.045492831811725,
      "learning_rate": 5.114599902836278e-07,
      "loss": 1.4014,
      "step": 4980
    },
    {
      "epoch": 0.5090278486177701,
      "grad_norm": 19.721913992432956,
      "learning_rate": 5.098083295646035e-07,
      "loss": 1.611,
      "step": 4990
    },
    {
      "epoch": 0.5100479445067836,
      "grad_norm": 16.789410640747978,
      "learning_rate": 5.081565617693139e-07,
      "loss": 1.6861,
      "step": 5000
    },
    {
      "epoch": 0.5110680403957972,
      "grad_norm": 20.42911207874605,
      "learning_rate": 5.065047049298939e-07,
      "loss": 1.8845,
      "step": 5010
    },
    {
      "epoch": 0.5120881362848108,
      "grad_norm": 17.081555701747554,
      "learning_rate": 5.048527770794515e-07,
      "loss": 2.2697,
      "step": 5020
    },
    {
      "epoch": 0.5131082321738243,
      "grad_norm": 23.509532620041945,
      "learning_rate": 5.032007962518692e-07,
      "loss": 1.8605,
      "step": 5030
    },
    {
      "epoch": 0.5141283280628379,
      "grad_norm": 17.697039730305324,
      "learning_rate": 5.015487804816081e-07,
      "loss": 1.6847,
      "step": 5040
    },
    {
      "epoch": 0.5151484239518515,
      "grad_norm": 174.30033727476336,
      "learning_rate": 4.998967478035107e-07,
      "loss": 2.098,
      "step": 5050
    },
    {
      "epoch": 0.5161685198408651,
      "grad_norm": 25.601499950987392,
      "learning_rate": 4.982447162526041e-07,
      "loss": 1.8935,
      "step": 5060
    },
    {
      "epoch": 0.5171886157298786,
      "grad_norm": 21.213269407367125,
      "learning_rate": 4.965927038639036e-07,
      "loss": 1.6402,
      "step": 5070
    },
    {
      "epoch": 0.5182087116188921,
      "grad_norm": 18.45923506218412,
      "learning_rate": 4.949407286722139e-07,
      "loss": 1.6774,
      "step": 5080
    },
    {
      "epoch": 0.5192288075079058,
      "grad_norm": 18.614115352543678,
      "learning_rate": 4.932888087119355e-07,
      "loss": 1.7381,
      "step": 5090
    },
    {
      "epoch": 0.5202489033969193,
      "grad_norm": 18.061646268407074,
      "learning_rate": 4.916369620168641e-07,
      "loss": 1.8887,
      "step": 5100
    },
    {
      "epoch": 0.5212689992859328,
      "grad_norm": 12.942053475717902,
      "learning_rate": 4.899852066199973e-07,
      "loss": 1.6345,
      "step": 5110
    },
    {
      "epoch": 0.5222890951749465,
      "grad_norm": 29.0673100840128,
      "learning_rate": 4.883335605533349e-07,
      "loss": 1.6617,
      "step": 5120
    },
    {
      "epoch": 0.52330919106396,
      "grad_norm": 22.4876956895944,
      "learning_rate": 4.86682041847683e-07,
      "loss": 1.6479,
      "step": 5130
    },
    {
      "epoch": 0.5243292869529735,
      "grad_norm": 27.56427418316557,
      "learning_rate": 4.850306685324583e-07,
      "loss": 1.4782,
      "step": 5140
    },
    {
      "epoch": 0.5253493828419872,
      "grad_norm": 30.214702756991038,
      "learning_rate": 4.833794586354895e-07,
      "loss": 1.8898,
      "step": 5150
    },
    {
      "epoch": 0.5263694787310007,
      "grad_norm": 29.528858248547927,
      "learning_rate": 4.817284301828216e-07,
      "loss": 1.6988,
      "step": 5160
    },
    {
      "epoch": 0.5273895746200142,
      "grad_norm": 22.61782038196886,
      "learning_rate": 4.800776011985185e-07,
      "loss": 1.7309,
      "step": 5170
    },
    {
      "epoch": 0.5284096705090279,
      "grad_norm": 20.321168018553745,
      "learning_rate": 4.784269897044672e-07,
      "loss": 1.6443,
      "step": 5180
    },
    {
      "epoch": 0.5294297663980414,
      "grad_norm": 18.13453460642375,
      "learning_rate": 4.7677661372017945e-07,
      "loss": 1.7276,
      "step": 5190
    },
    {
      "epoch": 0.5304498622870549,
      "grad_norm": 30.61774920943736,
      "learning_rate": 4.751264912625969e-07,
      "loss": 1.5245,
      "step": 5200
    },
    {
      "epoch": 0.5314699581760686,
      "grad_norm": 25.709511225450704,
      "learning_rate": 4.734766403458927e-07,
      "loss": 1.748,
      "step": 5210
    },
    {
      "epoch": 0.5324900540650821,
      "grad_norm": 24.91839846844101,
      "learning_rate": 4.718270789812762e-07,
      "loss": 1.6644,
      "step": 5220
    },
    {
      "epoch": 0.5335101499540957,
      "grad_norm": 23.218985303043507,
      "learning_rate": 4.701778251767952e-07,
      "loss": 1.5511,
      "step": 5230
    },
    {
      "epoch": 0.5345302458431093,
      "grad_norm": 21.696742781073834,
      "learning_rate": 4.6852889693714057e-07,
      "loss": 1.7534,
      "step": 5240
    },
    {
      "epoch": 0.5355503417321228,
      "grad_norm": 19.724895003897036,
      "learning_rate": 4.668803122634486e-07,
      "loss": 1.9025,
      "step": 5250
    },
    {
      "epoch": 0.5365704376211364,
      "grad_norm": 20.348888523778626,
      "learning_rate": 4.652320891531048e-07,
      "loss": 1.579,
      "step": 5260
    },
    {
      "epoch": 0.53759053351015,
      "grad_norm": 23.382372714632886,
      "learning_rate": 4.6358424559954794e-07,
      "loss": 1.6494,
      "step": 5270
    },
    {
      "epoch": 0.5386106293991635,
      "grad_norm": 26.861967196069617,
      "learning_rate": 4.619367995920728e-07,
      "loss": 1.946,
      "step": 5280
    },
    {
      "epoch": 0.5396307252881771,
      "grad_norm": 26.427322982423277,
      "learning_rate": 4.6028976911563467e-07,
      "loss": 1.7432,
      "step": 5290
    },
    {
      "epoch": 0.5406508211771907,
      "grad_norm": 107.83271158831703,
      "learning_rate": 4.5864317215065196e-07,
      "loss": 1.7388,
      "step": 5300
    },
    {
      "epoch": 0.5416709170662042,
      "grad_norm": 27.04638671479566,
      "learning_rate": 4.5699702667281133e-07,
      "loss": 1.77,
      "step": 5310
    },
    {
      "epoch": 0.5426910129552178,
      "grad_norm": 142.94278682776607,
      "learning_rate": 4.5535135065286937e-07,
      "loss": 1.9567,
      "step": 5320
    },
    {
      "epoch": 0.5437111088442314,
      "grad_norm": 20.56381168934149,
      "learning_rate": 4.5370616205645925e-07,
      "loss": 1.6623,
      "step": 5330
    },
    {
      "epoch": 0.5447312047332449,
      "grad_norm": 22.03257224714742,
      "learning_rate": 4.5206147884389125e-07,
      "loss": 1.7522,
      "step": 5340
    },
    {
      "epoch": 0.5457513006222585,
      "grad_norm": 15.162028740491227,
      "learning_rate": 4.5041731896995997e-07,
      "loss": 1.6784,
      "step": 5350
    },
    {
      "epoch": 0.546771396511272,
      "grad_norm": 18.873223171444806,
      "learning_rate": 4.4877370038374583e-07,
      "loss": 1.6423,
      "step": 5360
    },
    {
      "epoch": 0.5477914924002856,
      "grad_norm": 16.140138536731968,
      "learning_rate": 4.471306410284207e-07,
      "loss": 1.5781,
      "step": 5370
    },
    {
      "epoch": 0.5488115882892992,
      "grad_norm": 26.12970248802956,
      "learning_rate": 4.4548815884105077e-07,
      "loss": 1.6989,
      "step": 5380
    },
    {
      "epoch": 0.5498316841783127,
      "grad_norm": 27.887570537819006,
      "learning_rate": 4.438462717524018e-07,
      "loss": 1.5628,
      "step": 5390
    },
    {
      "epoch": 0.5508517800673264,
      "grad_norm": 19.61608213230033,
      "learning_rate": 4.422049976867429e-07,
      "loss": 1.7724,
      "step": 5400
    },
    {
      "epoch": 0.5518718759563399,
      "grad_norm": 25.580723683129282,
      "learning_rate": 4.4056435456165065e-07,
      "loss": 1.7736,
      "step": 5410
    },
    {
      "epoch": 0.5528919718453534,
      "grad_norm": 21.392685220691604,
      "learning_rate": 4.389243602878141e-07,
      "loss": 1.7573,
      "step": 5420
    },
    {
      "epoch": 0.5539120677343671,
      "grad_norm": 21.35026578711149,
      "learning_rate": 4.372850327688384e-07,
      "loss": 1.3559,
      "step": 5430
    },
    {
      "epoch": 0.5549321636233806,
      "grad_norm": 25.919502133684187,
      "learning_rate": 4.356463899010505e-07,
      "loss": 1.8172,
      "step": 5440
    },
    {
      "epoch": 0.5559522595123941,
      "grad_norm": 16.687767015511373,
      "learning_rate": 4.340084495733022e-07,
      "loss": 1.7695,
      "step": 5450
    },
    {
      "epoch": 0.5569723554014078,
      "grad_norm": 21.109701376172417,
      "learning_rate": 4.323712296667767e-07,
      "loss": 1.6166,
      "step": 5460
    },
    {
      "epoch": 0.5579924512904213,
      "grad_norm": 20.14738012754997,
      "learning_rate": 4.307347480547917e-07,
      "loss": 1.6705,
      "step": 5470
    },
    {
      "epoch": 0.5590125471794348,
      "grad_norm": 22.150409115193582,
      "learning_rate": 4.2909902260260555e-07,
      "loss": 1.5211,
      "step": 5480
    },
    {
      "epoch": 0.5600326430684485,
      "grad_norm": 24.62972426259411,
      "learning_rate": 4.274640711672212e-07,
      "loss": 1.5444,
      "step": 5490
    },
    {
      "epoch": 0.561052738957462,
      "grad_norm": 35.6972103370671,
      "learning_rate": 4.258299115971925e-07,
      "loss": 1.6748,
      "step": 5500
    },
    {
      "epoch": 0.5620728348464755,
      "grad_norm": 18.069298302313477,
      "learning_rate": 4.2419656173242766e-07,
      "loss": 1.682,
      "step": 5510
    },
    {
      "epoch": 0.5630929307354892,
      "grad_norm": 25.734531664122688,
      "learning_rate": 4.2256403940399653e-07,
      "loss": 1.4867,
      "step": 5520
    },
    {
      "epoch": 0.5641130266245027,
      "grad_norm": 18.63385933459369,
      "learning_rate": 4.209323624339338e-07,
      "loss": 1.7539,
      "step": 5530
    },
    {
      "epoch": 0.5651331225135162,
      "grad_norm": 22.094917761438904,
      "learning_rate": 4.193015486350459e-07,
      "loss": 1.6199,
      "step": 5540
    },
    {
      "epoch": 0.5661532184025299,
      "grad_norm": 25.762360669360877,
      "learning_rate": 4.1767161581071677e-07,
      "loss": 2.2628,
      "step": 5550
    },
    {
      "epoch": 0.5671733142915434,
      "grad_norm": 23.643147551297517,
      "learning_rate": 4.160425817547115e-07,
      "loss": 1.7803,
      "step": 5560
    },
    {
      "epoch": 0.568193410180557,
      "grad_norm": 29.082042983883053,
      "learning_rate": 4.1441446425098487e-07,
      "loss": 1.6527,
      "step": 5570
    },
    {
      "epoch": 0.5692135060695706,
      "grad_norm": 17.802190407054237,
      "learning_rate": 4.1278728107348484e-07,
      "loss": 1.9931,
      "step": 5580
    },
    {
      "epoch": 0.5702336019585841,
      "grad_norm": 21.07626392250645,
      "learning_rate": 4.111610499859598e-07,
      "loss": 1.8847,
      "step": 5590
    },
    {
      "epoch": 0.5712536978475977,
      "grad_norm": 28.89544155755318,
      "learning_rate": 4.095357887417642e-07,
      "loss": 1.8398,
      "step": 5600
    },
    {
      "epoch": 0.5722737937366112,
      "grad_norm": 24.758635319630983,
      "learning_rate": 4.0791151508366504e-07,
      "loss": 2.0513,
      "step": 5610
    },
    {
      "epoch": 0.5732938896256248,
      "grad_norm": 19.800800764061677,
      "learning_rate": 4.062882467436476e-07,
      "loss": 1.6694,
      "step": 5620
    },
    {
      "epoch": 0.5743139855146384,
      "grad_norm": 20.791333230303465,
      "learning_rate": 4.0466600144272266e-07,
      "loss": 1.5829,
      "step": 5630
    },
    {
      "epoch": 0.5753340814036519,
      "grad_norm": 24.16277938736952,
      "learning_rate": 4.0304479689073214e-07,
      "loss": 1.7157,
      "step": 5640
    },
    {
      "epoch": 0.5763541772926655,
      "grad_norm": 18.366614097090892,
      "learning_rate": 4.014246507861568e-07,
      "loss": 1.6298,
      "step": 5650
    },
    {
      "epoch": 0.5773742731816791,
      "grad_norm": 7.770912649648675e+22,
      "learning_rate": 3.9980558081592194e-07,
      "loss": 2.1267,
      "step": 5660
    },
    {
      "epoch": 0.5783943690706926,
      "grad_norm": 33.706031880507304,
      "learning_rate": 3.9818760465520496e-07,
      "loss": 1.5183,
      "step": 5670
    },
    {
      "epoch": 0.5794144649597062,
      "grad_norm": 29.02247102457546,
      "learning_rate": 3.9657073996724274e-07,
      "loss": 2.1971,
      "step": 5680
    },
    {
      "epoch": 0.5804345608487198,
      "grad_norm": 27.29527922334536,
      "learning_rate": 3.949550044031377e-07,
      "loss": 1.894,
      "step": 5690
    },
    {
      "epoch": 0.5814546567377333,
      "grad_norm": 36.42323044136592,
      "learning_rate": 3.933404156016662e-07,
      "loss": 1.5903,
      "step": 5700
    },
    {
      "epoch": 0.5824747526267469,
      "grad_norm": 28.391641061984362,
      "learning_rate": 3.9172699118908525e-07,
      "loss": 1.6602,
      "step": 5710
    },
    {
      "epoch": 0.5834948485157605,
      "grad_norm": 31.521224850258577,
      "learning_rate": 3.9011474877894077e-07,
      "loss": 1.8701,
      "step": 5720
    },
    {
      "epoch": 0.584514944404774,
      "grad_norm": 23.911685396539824,
      "learning_rate": 3.885037059718744e-07,
      "loss": 1.5867,
      "step": 5730
    },
    {
      "epoch": 0.5855350402937877,
      "grad_norm": 18.107758707434023,
      "learning_rate": 3.8689388035543253e-07,
      "loss": 1.5934,
      "step": 5740
    },
    {
      "epoch": 0.5865551361828012,
      "grad_norm": 27.32391152147643,
      "learning_rate": 3.852852895038727e-07,
      "loss": 1.6977,
      "step": 5750
    },
    {
      "epoch": 0.5875752320718147,
      "grad_norm": 178.08166250246055,
      "learning_rate": 3.8367795097797374e-07,
      "loss": 2.0963,
      "step": 5760
    },
    {
      "epoch": 0.5885953279608284,
      "grad_norm": 24.171129812530275,
      "learning_rate": 3.820718823248422e-07,
      "loss": 1.6089,
      "step": 5770
    },
    {
      "epoch": 0.5896154238498419,
      "grad_norm": 19.272137566629382,
      "learning_rate": 3.804671010777221e-07,
      "loss": 1.7627,
      "step": 5780
    },
    {
      "epoch": 0.5906355197388554,
      "grad_norm": 30.897451390948575,
      "learning_rate": 3.788636247558027e-07,
      "loss": 1.7348,
      "step": 5790
    },
    {
      "epoch": 0.5916556156278691,
      "grad_norm": 22.854024766457627,
      "learning_rate": 3.7726147086402736e-07,
      "loss": 1.832,
      "step": 5800
    },
    {
      "epoch": 0.5926757115168826,
      "grad_norm": 27.984278445188227,
      "learning_rate": 3.7566065689290345e-07,
      "loss": 1.5626,
      "step": 5810
    },
    {
      "epoch": 0.5936958074058961,
      "grad_norm": 22.448341099067665,
      "learning_rate": 3.740612003183097e-07,
      "loss": 1.8882,
      "step": 5820
    },
    {
      "epoch": 0.5947159032949098,
      "grad_norm": 23.859234654292095,
      "learning_rate": 3.7246311860130696e-07,
      "loss": 1.8274,
      "step": 5830
    },
    {
      "epoch": 0.5957359991839233,
      "grad_norm": 21.2647742732659,
      "learning_rate": 3.7086642918794645e-07,
      "loss": 2.0938,
      "step": 5840
    },
    {
      "epoch": 0.5967560950729368,
      "grad_norm": 19.951629393715272,
      "learning_rate": 3.6927114950908036e-07,
      "loss": 1.74,
      "step": 5850
    },
    {
      "epoch": 0.5977761909619504,
      "grad_norm": 16.903434909058284,
      "learning_rate": 3.6767729698017023e-07,
      "loss": 1.7255,
      "step": 5860
    },
    {
      "epoch": 0.598796286850964,
      "grad_norm": 32.478389015889995,
      "learning_rate": 3.6608488900109845e-07,
      "loss": 1.5527,
      "step": 5870
    },
    {
      "epoch": 0.5998163827399775,
      "grad_norm": 19.059492769007104,
      "learning_rate": 3.644939429559767e-07,
      "loss": 1.8978,
      "step": 5880
    },
    {
      "epoch": 0.6008364786289911,
      "grad_norm": 21.985800104469686,
      "learning_rate": 3.629044762129574e-07,
      "loss": 1.5734,
      "step": 5890
    },
    {
      "epoch": 0.6018565745180047,
      "grad_norm": 20.419950731931156,
      "learning_rate": 3.613165061240433e-07,
      "loss": 1.6222,
      "step": 5900
    },
    {
      "epoch": 0.6028766704070183,
      "grad_norm": 26.066394894923082,
      "learning_rate": 3.5973005002489867e-07,
      "loss": 1.6113,
      "step": 5910
    },
    {
      "epoch": 0.6038967662960318,
      "grad_norm": 19.402906718500187,
      "learning_rate": 3.581451252346593e-07,
      "loss": 1.7683,
      "step": 5920
    },
    {
      "epoch": 0.6049168621850454,
      "grad_norm": 31.758722967921386,
      "learning_rate": 3.5656174905574483e-07,
      "loss": 1.6919,
      "step": 5930
    },
    {
      "epoch": 0.605936958074059,
      "grad_norm": 29.053692055157267,
      "learning_rate": 3.5497993877366767e-07,
      "loss": 1.5923,
      "step": 5940
    },
    {
      "epoch": 0.6069570539630725,
      "grad_norm": 33.71469989534779,
      "learning_rate": 3.5339971165684624e-07,
      "loss": 1.6833,
      "step": 5950
    },
    {
      "epoch": 0.6079771498520861,
      "grad_norm": 15.445063665656633,
      "learning_rate": 3.518210849564157e-07,
      "loss": 1.6743,
      "step": 5960
    },
    {
      "epoch": 0.6089972457410997,
      "grad_norm": 25.452112902348855,
      "learning_rate": 3.5024407590603944e-07,
      "loss": 1.7873,
      "step": 5970
    },
    {
      "epoch": 0.6100173416301132,
      "grad_norm": 21.329172568132453,
      "learning_rate": 3.486687017217218e-07,
      "loss": 1.5877,
      "step": 5980
    },
    {
      "epoch": 0.6110374375191268,
      "grad_norm": 14.481212873861656,
      "learning_rate": 3.4709497960161836e-07,
      "loss": 1.6814,
      "step": 5990
    },
    {
      "epoch": 0.6120575334081404,
      "grad_norm": 16.93877214134712,
      "learning_rate": 3.4552292672585066e-07,
      "loss": 1.7529,
      "step": 6000
    },
    {
      "epoch": 0.6130776292971539,
      "grad_norm": 19.824478238855264,
      "learning_rate": 3.439525602563163e-07,
      "loss": 1.8198,
      "step": 6010
    },
    {
      "epoch": 0.6140977251861675,
      "grad_norm": 20.872815959299448,
      "learning_rate": 3.4238389733650303e-07,
      "loss": 1.4723,
      "step": 6020
    },
    {
      "epoch": 0.6151178210751811,
      "grad_norm": 23.806698092521803,
      "learning_rate": 3.40816955091301e-07,
      "loss": 1.527,
      "step": 6030
    },
    {
      "epoch": 0.6161379169641946,
      "grad_norm": 16.64697417345877,
      "learning_rate": 3.392517506268164e-07,
      "loss": 1.5573,
      "step": 6040
    },
    {
      "epoch": 0.6171580128532081,
      "grad_norm": 24.309964960248198,
      "learning_rate": 3.3768830103018363e-07,
      "loss": 2.1218,
      "step": 6050
    },
    {
      "epoch": 0.6181781087422218,
      "grad_norm": 21.89591276685123,
      "learning_rate": 3.3612662336938e-07,
      "loss": 1.7643,
      "step": 6060
    },
    {
      "epoch": 0.6191982046312353,
      "grad_norm": 47.37258830899368,
      "learning_rate": 3.345667346930384e-07,
      "loss": 1.7584,
      "step": 6070
    },
    {
      "epoch": 0.620218300520249,
      "grad_norm": 23.73359563823175,
      "learning_rate": 3.330086520302619e-07,
      "loss": 1.9482,
      "step": 6080
    },
    {
      "epoch": 0.6212383964092625,
      "grad_norm": 19.093896447171744,
      "learning_rate": 3.314523923904375e-07,
      "loss": 1.5505,
      "step": 6090
    },
    {
      "epoch": 0.622258492298276,
      "grad_norm": 21.111841066926726,
      "learning_rate": 3.2989797276305033e-07,
      "loss": 1.4888,
      "step": 6100
    },
    {
      "epoch": 0.6232785881872897,
      "grad_norm": 32.029843301786514,
      "learning_rate": 3.283454101174985e-07,
      "loss": 1.7523,
      "step": 6110
    },
    {
      "epoch": 0.6242986840763032,
      "grad_norm": 22.09504819181879,
      "learning_rate": 3.2679472140290764e-07,
      "loss": 1.7617,
      "step": 6120
    },
    {
      "epoch": 0.6253187799653167,
      "grad_norm": 30.18928187881305,
      "learning_rate": 3.252459235479461e-07,
      "loss": 1.8236,
      "step": 6130
    },
    {
      "epoch": 0.6263388758543303,
      "grad_norm": 28.49655162286157,
      "learning_rate": 3.236990334606395e-07,
      "loss": 1.3941,
      "step": 6140
    },
    {
      "epoch": 0.6273589717433439,
      "grad_norm": 24.454112148560217,
      "learning_rate": 3.2215406802818713e-07,
      "loss": 1.8038,
      "step": 6150
    },
    {
      "epoch": 0.6283790676323574,
      "grad_norm": 19.09884366111005,
      "learning_rate": 3.2061104411677655e-07,
      "loss": 1.765,
      "step": 6160
    },
    {
      "epoch": 0.629399163521371,
      "grad_norm": 13.836615066459808,
      "learning_rate": 3.1906997857140073e-07,
      "loss": 1.4457,
      "step": 6170
    },
    {
      "epoch": 0.6304192594103846,
      "grad_norm": 22.337924573464722,
      "learning_rate": 3.1753088821567233e-07,
      "loss": 1.8985,
      "step": 6180
    },
    {
      "epoch": 0.6314393552993981,
      "grad_norm": 19.055413604894852,
      "learning_rate": 3.159937898516424e-07,
      "loss": 1.6148,
      "step": 6190
    },
    {
      "epoch": 0.6324594511884117,
      "grad_norm": 28.756847537665706,
      "learning_rate": 3.144587002596147e-07,
      "loss": 1.6762,
      "step": 6200
    },
    {
      "epoch": 0.6334795470774253,
      "grad_norm": 26.83163815053663,
      "learning_rate": 3.1292563619796365e-07,
      "loss": 1.6316,
      "step": 6210
    },
    {
      "epoch": 0.6344996429664388,
      "grad_norm": 17.008271524257033,
      "learning_rate": 3.113946144029518e-07,
      "loss": 1.5391,
      "step": 6220
    },
    {
      "epoch": 0.6355197388554524,
      "grad_norm": 21.096214410946736,
      "learning_rate": 3.098656515885455e-07,
      "loss": 1.7701,
      "step": 6230
    },
    {
      "epoch": 0.636539834744466,
      "grad_norm": 14.564953501429285,
      "learning_rate": 3.083387644462346e-07,
      "loss": 1.552,
      "step": 6240
    },
    {
      "epoch": 0.6375599306334796,
      "grad_norm": 22.24409206101689,
      "learning_rate": 3.068139696448484e-07,
      "loss": 1.7022,
      "step": 6250
    },
    {
      "epoch": 0.6385800265224931,
      "grad_norm": 32.915295795583845,
      "learning_rate": 3.0529128383037467e-07,
      "loss": 1.7836,
      "step": 6260
    },
    {
      "epoch": 0.6396001224115067,
      "grad_norm": 39.79556496790107,
      "learning_rate": 3.0377072362577736e-07,
      "loss": 1.8929,
      "step": 6270
    },
    {
      "epoch": 0.6406202183005203,
      "grad_norm": 24.55610559821662,
      "learning_rate": 3.022523056308158e-07,
      "loss": 1.6828,
      "step": 6280
    },
    {
      "epoch": 0.6416403141895338,
      "grad_norm": 25.792737146577853,
      "learning_rate": 3.007360464218627e-07,
      "loss": 1.6022,
      "step": 6290
    },
    {
      "epoch": 0.6426604100785474,
      "grad_norm": 19.24518387117486,
      "learning_rate": 2.99221962551724e-07,
      "loss": 1.7734,
      "step": 6300
    },
    {
      "epoch": 0.643680505967561,
      "grad_norm": 20.08254735324983,
      "learning_rate": 2.977100705494573e-07,
      "loss": 2.1578,
      "step": 6310
    },
    {
      "epoch": 0.6447006018565745,
      "grad_norm": 22.843018993525867,
      "learning_rate": 2.962003869201923e-07,
      "loss": 1.6804,
      "step": 6320
    },
    {
      "epoch": 0.645720697745588,
      "grad_norm": 35.096697008121545,
      "learning_rate": 2.946929281449498e-07,
      "loss": 1.4639,
      "step": 6330
    },
    {
      "epoch": 0.6467407936346017,
      "grad_norm": 27.67363521650983,
      "learning_rate": 2.9318771068046233e-07,
      "loss": 1.5833,
      "step": 6340
    },
    {
      "epoch": 0.6477608895236152,
      "grad_norm": 23.721414802993387,
      "learning_rate": 2.9168475095899435e-07,
      "loss": 1.7317,
      "step": 6350
    },
    {
      "epoch": 0.6487809854126287,
      "grad_norm": 21.315376291514717,
      "learning_rate": 2.9018406538816264e-07,
      "loss": 1.7838,
      "step": 6360
    },
    {
      "epoch": 0.6498010813016424,
      "grad_norm": 28.605728502315582,
      "learning_rate": 2.88685670350758e-07,
      "loss": 1.5948,
      "step": 6370
    },
    {
      "epoch": 0.6508211771906559,
      "grad_norm": 18.745873278957642,
      "learning_rate": 2.871895822045648e-07,
      "loss": 1.6602,
      "step": 6380
    },
    {
      "epoch": 0.6518412730796694,
      "grad_norm": 18.66552888960972,
      "learning_rate": 2.8569581728218404e-07,
      "loss": 1.6812,
      "step": 6390
    },
    {
      "epoch": 0.6528613689686831,
      "grad_norm": 21.62819251055515,
      "learning_rate": 2.842043918908541e-07,
      "loss": 1.982,
      "step": 6400
    },
    {
      "epoch": 0.6538814648576966,
      "grad_norm": 18.929696669364944,
      "learning_rate": 2.827153223122734e-07,
      "loss": 1.7259,
      "step": 6410
    },
    {
      "epoch": 0.6549015607467102,
      "grad_norm": 17.14308911085822,
      "learning_rate": 2.812286248024218e-07,
      "loss": 1.6604,
      "step": 6420
    },
    {
      "epoch": 0.6559216566357238,
      "grad_norm": 19.24462556862274,
      "learning_rate": 2.7974431559138376e-07,
      "loss": 1.8042,
      "step": 6430
    },
    {
      "epoch": 0.6569417525247373,
      "grad_norm": 29.741098748731275,
      "learning_rate": 2.7826241088317093e-07,
      "loss": 1.6719,
      "step": 6440
    },
    {
      "epoch": 0.6579618484137509,
      "grad_norm": 24.48321681304582,
      "learning_rate": 2.767829268555456e-07,
      "loss": 2.1673,
      "step": 6450
    },
    {
      "epoch": 0.6589819443027645,
      "grad_norm": 20.369459137442472,
      "learning_rate": 2.7530587965984325e-07,
      "loss": 1.7385,
      "step": 6460
    },
    {
      "epoch": 0.660002040191778,
      "grad_norm": 18.410711976293207,
      "learning_rate": 2.738312854207978e-07,
      "loss": 1.5559,
      "step": 6470
    },
    {
      "epoch": 0.6610221360807916,
      "grad_norm": 18.840664497461628,
      "learning_rate": 2.7235916023636307e-07,
      "loss": 1.8145,
      "step": 6480
    },
    {
      "epoch": 0.6620422319698052,
      "grad_norm": 28.890625736858027,
      "learning_rate": 2.708895201775396e-07,
      "loss": 1.8088,
      "step": 6490
    },
    {
      "epoch": 0.6630623278588187,
      "grad_norm": 22.592351633344308,
      "learning_rate": 2.694223812881981e-07,
      "loss": 1.5063,
      "step": 6500
    },
    {
      "epoch": 0.6640824237478323,
      "grad_norm": 398.3087981607348,
      "learning_rate": 2.679577595849037e-07,
      "loss": 1.6671,
      "step": 6510
    },
    {
      "epoch": 0.6651025196368459,
      "grad_norm": 32.18163287762227,
      "learning_rate": 2.664956710567429e-07,
      "loss": 1.4634,
      "step": 6520
    },
    {
      "epoch": 0.6661226155258594,
      "grad_norm": 21.732575346646545,
      "learning_rate": 2.6503613166514616e-07,
      "loss": 1.6018,
      "step": 6530
    },
    {
      "epoch": 0.667142711414873,
      "grad_norm": 24.73462002322095,
      "learning_rate": 2.6357915734371696e-07,
      "loss": 2.0672,
      "step": 6540
    },
    {
      "epoch": 0.6681628073038866,
      "grad_norm": 19.92719688813928,
      "learning_rate": 2.6212476399805504e-07,
      "loss": 1.6276,
      "step": 6550
    },
    {
      "epoch": 0.6691829031929001,
      "grad_norm": 17.929498615882252,
      "learning_rate": 2.6067296750558505e-07,
      "loss": 1.7286,
      "step": 6560
    },
    {
      "epoch": 0.6702029990819137,
      "grad_norm": 23.373780093360992,
      "learning_rate": 2.5922378371538124e-07,
      "loss": 1.7866,
      "step": 6570
    },
    {
      "epoch": 0.6712230949709272,
      "grad_norm": 20.941368320544246,
      "learning_rate": 2.5777722844799575e-07,
      "loss": 1.6402,
      "step": 6580
    },
    {
      "epoch": 0.6722431908599409,
      "grad_norm": 21.13692779580355,
      "learning_rate": 2.563333174952851e-07,
      "loss": 1.6429,
      "step": 6590
    },
    {
      "epoch": 0.6732632867489544,
      "grad_norm": 17.614815283209243,
      "learning_rate": 2.548920666202389e-07,
      "loss": 1.6283,
      "step": 6600
    },
    {
      "epoch": 0.6742833826379679,
      "grad_norm": 25.52306466602991,
      "learning_rate": 2.5345349155680626e-07,
      "loss": 1.7585,
      "step": 6610
    },
    {
      "epoch": 0.6753034785269816,
      "grad_norm": 20.33153664867209,
      "learning_rate": 2.520176080097254e-07,
      "loss": 1.6979,
      "step": 6620
    },
    {
      "epoch": 0.6763235744159951,
      "grad_norm": 21.424245523335205,
      "learning_rate": 2.50584431654351e-07,
      "loss": 1.5832,
      "step": 6630
    },
    {
      "epoch": 0.6773436703050086,
      "grad_norm": 31.049604264185085,
      "learning_rate": 2.49153978136484e-07,
      "loss": 1.6031,
      "step": 6640
    },
    {
      "epoch": 0.6783637661940223,
      "grad_norm": 23.124537375222747,
      "learning_rate": 2.477262630722009e-07,
      "loss": 2.0283,
      "step": 6650
    },
    {
      "epoch": 0.6793838620830358,
      "grad_norm": 26.945912355507577,
      "learning_rate": 2.463013020476821e-07,
      "loss": 1.862,
      "step": 6660
    },
    {
      "epoch": 0.6804039579720493,
      "grad_norm": 22.059150575479855,
      "learning_rate": 2.44879110619043e-07,
      "loss": 1.9159,
      "step": 6670
    },
    {
      "epoch": 0.681424053861063,
      "grad_norm": 19.634252145312583,
      "learning_rate": 2.434597043121632e-07,
      "loss": 1.4844,
      "step": 6680
    },
    {
      "epoch": 0.6824441497500765,
      "grad_norm": 31.06395644669561,
      "learning_rate": 2.420430986225183e-07,
      "loss": 1.7315,
      "step": 6690
    },
    {
      "epoch": 0.68346424563909,
      "grad_norm": 11.73218227581462,
      "learning_rate": 2.40629309015009e-07,
      "loss": 1.6782,
      "step": 6700
    },
    {
      "epoch": 0.6844843415281037,
      "grad_norm": 40.474532496136526,
      "learning_rate": 2.392183509237942e-07,
      "loss": 1.6124,
      "step": 6710
    },
    {
      "epoch": 0.6855044374171172,
      "grad_norm": 17.885594049824967,
      "learning_rate": 2.3781023975212005e-07,
      "loss": 1.7385,
      "step": 6720
    },
    {
      "epoch": 0.6865245333061307,
      "grad_norm": 19.169985050789347,
      "learning_rate": 2.3640499087215488e-07,
      "loss": 1.6248,
      "step": 6730
    },
    {
      "epoch": 0.6875446291951444,
      "grad_norm": 33.60519938607105,
      "learning_rate": 2.3500261962481877e-07,
      "loss": 1.6983,
      "step": 6740
    },
    {
      "epoch": 0.6885647250841579,
      "grad_norm": 22.26210814195473,
      "learning_rate": 2.3360314131961707e-07,
      "loss": 1.4504,
      "step": 6750
    },
    {
      "epoch": 0.6895848209731715,
      "grad_norm": 14.871798879476895,
      "learning_rate": 2.3220657123447395e-07,
      "loss": 1.5498,
      "step": 6760
    },
    {
      "epoch": 0.6906049168621851,
      "grad_norm": 22.242770777064734,
      "learning_rate": 2.3081292461556384e-07,
      "loss": 1.7685,
      "step": 6770
    },
    {
      "epoch": 0.6916250127511986,
      "grad_norm": 18.26234254424413,
      "learning_rate": 2.294222166771471e-07,
      "loss": 1.4799,
      "step": 6780
    },
    {
      "epoch": 0.6926451086402122,
      "grad_norm": 29.986905037431058,
      "learning_rate": 2.2803446260140226e-07,
      "loss": 1.7332,
      "step": 6790
    },
    {
      "epoch": 0.6936652045292258,
      "grad_norm": 25.46072528103776,
      "learning_rate": 2.2664967753826125e-07,
      "loss": 1.3798,
      "step": 6800
    },
    {
      "epoch": 0.6946853004182393,
      "grad_norm": 31.653772637792102,
      "learning_rate": 2.2526787660524356e-07,
      "loss": 2.245,
      "step": 6810
    },
    {
      "epoch": 0.6957053963072529,
      "grad_norm": 19.726926151041795,
      "learning_rate": 2.2388907488729113e-07,
      "loss": 1.6074,
      "step": 6820
    },
    {
      "epoch": 0.6967254921962664,
      "grad_norm": 33.550269320637,
      "learning_rate": 2.225132874366039e-07,
      "loss": 1.7813,
      "step": 6830
    },
    {
      "epoch": 0.69774558808528,
      "grad_norm": 16.637300510478788,
      "learning_rate": 2.2114052927247601e-07,
      "loss": 1.6877,
      "step": 6840
    },
    {
      "epoch": 0.6987656839742936,
      "grad_norm": 19.08673041079158,
      "learning_rate": 2.1977081538113058e-07,
      "loss": 1.7033,
      "step": 6850
    },
    {
      "epoch": 0.6997857798633071,
      "grad_norm": 16.368079427581197,
      "learning_rate": 2.18404160715557e-07,
      "loss": 1.5335,
      "step": 6860
    },
    {
      "epoch": 0.7008058757523207,
      "grad_norm": 19.511200886748284,
      "learning_rate": 2.1704058019534737e-07,
      "loss": 1.5391,
      "step": 6870
    },
    {
      "epoch": 0.7018259716413343,
      "grad_norm": 14.599617898621338,
      "learning_rate": 2.1568008870653436e-07,
      "loss": 2.0537,
      "step": 6880
    },
    {
      "epoch": 0.7028460675303478,
      "grad_norm": 20.764490878293874,
      "learning_rate": 2.143227011014273e-07,
      "loss": 1.6096,
      "step": 6890
    },
    {
      "epoch": 0.7038661634193614,
      "grad_norm": 26.411553325067736,
      "learning_rate": 2.1296843219845128e-07,
      "loss": 1.9556,
      "step": 6900
    },
    {
      "epoch": 0.704886259308375,
      "grad_norm": 28.148319496804362,
      "learning_rate": 2.116172967819847e-07,
      "loss": 1.9273,
      "step": 6910
    },
    {
      "epoch": 0.7059063551973885,
      "grad_norm": 26.696070376751116,
      "learning_rate": 2.1026930960219807e-07,
      "loss": 1.4883,
      "step": 6920
    },
    {
      "epoch": 0.7069264510864022,
      "grad_norm": 30.65903103189678,
      "learning_rate": 2.0892448537489343e-07,
      "loss": 1.8187,
      "step": 6930
    },
    {
      "epoch": 0.7079465469754157,
      "grad_norm": 17.011352254274975,
      "learning_rate": 2.0758283878134264e-07,
      "loss": 1.5675,
      "step": 6940
    },
    {
      "epoch": 0.7089666428644292,
      "grad_norm": 25.942123506798513,
      "learning_rate": 2.0624438446812863e-07,
      "loss": 1.7619,
      "step": 6950
    },
    {
      "epoch": 0.7099867387534429,
      "grad_norm": 16.801409999903793,
      "learning_rate": 2.0490913704698332e-07,
      "loss": 1.6808,
      "step": 6960
    },
    {
      "epoch": 0.7110068346424564,
      "grad_norm": 19.621558988504034,
      "learning_rate": 2.0357711109463066e-07,
      "loss": 1.7652,
      "step": 6970
    },
    {
      "epoch": 0.7120269305314699,
      "grad_norm": 24.372785802657543,
      "learning_rate": 2.0224832115262536e-07,
      "loss": 1.7162,
      "step": 6980
    },
    {
      "epoch": 0.7130470264204836,
      "grad_norm": 26.293946623068763,
      "learning_rate": 2.009227817271958e-07,
      "loss": 1.5993,
      "step": 6990
    },
    {
      "epoch": 0.7140671223094971,
      "grad_norm": 29.869889087574588,
      "learning_rate": 1.996005072890843e-07,
      "loss": 1.7984,
      "step": 7000
    },
    {
      "epoch": 0.7150872181985106,
      "grad_norm": 19.394356480134814,
      "learning_rate": 1.982815122733899e-07,
      "loss": 1.7805,
      "step": 7010
    },
    {
      "epoch": 0.7161073140875243,
      "grad_norm": 18.784033176363376,
      "learning_rate": 1.9696581107941063e-07,
      "loss": 1.5245,
      "step": 7020
    },
    {
      "epoch": 0.7171274099765378,
      "grad_norm": 16.022826034643117,
      "learning_rate": 1.9565341807048608e-07,
      "loss": 1.5869,
      "step": 7030
    },
    {
      "epoch": 0.7181475058655513,
      "grad_norm": 16.48688567106227,
      "learning_rate": 1.943443475738414e-07,
      "loss": 1.5544,
      "step": 7040
    },
    {
      "epoch": 0.719167601754565,
      "grad_norm": 15.924426799781955,
      "learning_rate": 1.9303861388042975e-07,
      "loss": 1.8634,
      "step": 7050
    },
    {
      "epoch": 0.7201876976435785,
      "grad_norm": 23.826341797435035,
      "learning_rate": 1.9173623124477704e-07,
      "loss": 1.757,
      "step": 7060
    },
    {
      "epoch": 0.721207793532592,
      "grad_norm": 23.40658801998499,
      "learning_rate": 1.9043721388482587e-07,
      "loss": 1.6311,
      "step": 7070
    },
    {
      "epoch": 0.7222278894216057,
      "grad_norm": 22.883694603198034,
      "learning_rate": 1.891415759817812e-07,
      "loss": 1.713,
      "step": 7080
    },
    {
      "epoch": 0.7232479853106192,
      "grad_norm": 22.06160642794116,
      "learning_rate": 1.878493316799542e-07,
      "loss": 1.5854,
      "step": 7090
    },
    {
      "epoch": 0.7242680811996328,
      "grad_norm": 18.014161791765794,
      "learning_rate": 1.8656049508660893e-07,
      "loss": 1.5903,
      "step": 7100
    },
    {
      "epoch": 0.7252881770886463,
      "grad_norm": 19.7882143184373,
      "learning_rate": 1.8527508027180755e-07,
      "loss": 1.6952,
      "step": 7110
    },
    {
      "epoch": 0.7263082729776599,
      "grad_norm": 21.64014810721847,
      "learning_rate": 1.839931012682578e-07,
      "loss": 1.5395,
      "step": 7120
    },
    {
      "epoch": 0.7273283688666735,
      "grad_norm": 15.87288578826529,
      "learning_rate": 1.8271457207115825e-07,
      "loss": 1.9304,
      "step": 7130
    },
    {
      "epoch": 0.728348464755687,
      "grad_norm": 18.035395208712025,
      "learning_rate": 1.8143950663804742e-07,
      "loss": 1.7467,
      "step": 7140
    },
    {
      "epoch": 0.7293685606447006,
      "grad_norm": 19.024858797224294,
      "learning_rate": 1.8016791888864918e-07,
      "loss": 1.6865,
      "step": 7150
    },
    {
      "epoch": 0.7303886565337142,
      "grad_norm": 20.85274966136469,
      "learning_rate": 1.7889982270472263e-07,
      "loss": 2.3399,
      "step": 7160
    },
    {
      "epoch": 0.7314087524227277,
      "grad_norm": 26.789988714918955,
      "learning_rate": 1.7763523192991015e-07,
      "loss": 1.7767,
      "step": 7170
    },
    {
      "epoch": 0.7324288483117413,
      "grad_norm": 19.588372147125035,
      "learning_rate": 1.763741603695853e-07,
      "loss": 1.7922,
      "step": 7180
    },
    {
      "epoch": 0.7334489442007549,
      "grad_norm": 27.18328040760604,
      "learning_rate": 1.7511662179070385e-07,
      "loss": 1.7122,
      "step": 7190
    },
    {
      "epoch": 0.7344690400897684,
      "grad_norm": 22.957903563499766,
      "learning_rate": 1.7386262992165107e-07,
      "loss": 2.0692,
      "step": 7200
    },
    {
      "epoch": 0.735489135978782,
      "grad_norm": 39.03875591429386,
      "learning_rate": 1.7261219845209464e-07,
      "loss": 1.7537,
      "step": 7210
    },
    {
      "epoch": 0.7365092318677956,
      "grad_norm": 25.718196330076292,
      "learning_rate": 1.713653410328328e-07,
      "loss": 1.7792,
      "step": 7220
    },
    {
      "epoch": 0.7375293277568091,
      "grad_norm": 14.845271563981127,
      "learning_rate": 1.7012207127564714e-07,
      "loss": 1.9862,
      "step": 7230
    },
    {
      "epoch": 0.7385494236458227,
      "grad_norm": 21.57013646494491,
      "learning_rate": 1.6888240275315268e-07,
      "loss": 1.692,
      "step": 7240
    },
    {
      "epoch": 0.7395695195348363,
      "grad_norm": 22.296967361967983,
      "learning_rate": 1.6764634899865037e-07,
      "loss": 1.7248,
      "step": 7250
    },
    {
      "epoch": 0.7405896154238498,
      "grad_norm": 20.68636208069207,
      "learning_rate": 1.6641392350597906e-07,
      "loss": 1.7859,
      "step": 7260
    },
    {
      "epoch": 0.7416097113128635,
      "grad_norm": 26.98015098761684,
      "learning_rate": 1.6518513972936875e-07,
      "loss": 1.8914,
      "step": 7270
    },
    {
      "epoch": 0.742629807201877,
      "grad_norm": 13.65908113263917,
      "learning_rate": 1.6396001108329294e-07,
      "loss": 1.5939,
      "step": 7280
    },
    {
      "epoch": 0.7436499030908905,
      "grad_norm": 22.33428854017589,
      "learning_rate": 1.6273855094232264e-07,
      "loss": 1.5742,
      "step": 7290
    },
    {
      "epoch": 0.7446699989799042,
      "grad_norm": 26.270362979504565,
      "learning_rate": 1.615207726409802e-07,
      "loss": 1.775,
      "step": 7300
    },
    {
      "epoch": 0.7456900948689177,
      "grad_norm": 18.957041275061677,
      "learning_rate": 1.6030668947359371e-07,
      "loss": 1.8748,
      "step": 7310
    },
    {
      "epoch": 0.7467101907579312,
      "grad_norm": 25.08829151429589,
      "learning_rate": 1.590963146941525e-07,
      "loss": 1.922,
      "step": 7320
    },
    {
      "epoch": 0.7477302866469449,
      "grad_norm": 37.57298557291334,
      "learning_rate": 1.5788966151616134e-07,
      "loss": 1.6975,
      "step": 7330
    },
    {
      "epoch": 0.7487503825359584,
      "grad_norm": 26.560547206290693,
      "learning_rate": 1.566867431124969e-07,
      "loss": 1.4037,
      "step": 7340
    },
    {
      "epoch": 0.7497704784249719,
      "grad_norm": 38.308634771310565,
      "learning_rate": 1.5548757261526375e-07,
      "loss": 1.5906,
      "step": 7350
    },
    {
      "epoch": 0.7507905743139855,
      "grad_norm": 17.96410227550413,
      "learning_rate": 1.5429216311565142e-07,
      "loss": 1.9435,
      "step": 7360
    },
    {
      "epoch": 0.7518106702029991,
      "grad_norm": 23.3991789372964,
      "learning_rate": 1.5310052766379044e-07,
      "loss": 2.3403,
      "step": 7370
    },
    {
      "epoch": 0.7528307660920126,
      "grad_norm": 22.91317150966043,
      "learning_rate": 1.5191267926861156e-07,
      "loss": 1.5861,
      "step": 7380
    },
    {
      "epoch": 0.7538508619810262,
      "grad_norm": 20.599478532683882,
      "learning_rate": 1.5072863089770143e-07,
      "loss": 1.6069,
      "step": 7390
    },
    {
      "epoch": 0.7548709578700398,
      "grad_norm": 42.368331661367044,
      "learning_rate": 1.4954839547716348e-07,
      "loss": 1.729,
      "step": 7400
    },
    {
      "epoch": 0.7558910537590533,
      "grad_norm": 22.589877544707566,
      "learning_rate": 1.4837198589147486e-07,
      "loss": 1.6819,
      "step": 7410
    },
    {
      "epoch": 0.7569111496480669,
      "grad_norm": 22.871719102328232,
      "learning_rate": 1.4719941498334714e-07,
      "loss": 1.7976,
      "step": 7420
    },
    {
      "epoch": 0.7579312455370805,
      "grad_norm": 18.748971707020125,
      "learning_rate": 1.4603069555358545e-07,
      "loss": 1.6535,
      "step": 7430
    },
    {
      "epoch": 0.7589513414260941,
      "grad_norm": 28.78190784001419,
      "learning_rate": 1.4486584036094808e-07,
      "loss": 1.8561,
      "step": 7440
    },
    {
      "epoch": 0.7599714373151076,
      "grad_norm": 25.226434619051446,
      "learning_rate": 1.4370486212200899e-07,
      "loss": 1.6305,
      "step": 7450
    },
    {
      "epoch": 0.7609915332041212,
      "grad_norm": 27.92468968629957,
      "learning_rate": 1.4254777351101732e-07,
      "loss": 1.5495,
      "step": 7460
    },
    {
      "epoch": 0.7620116290931348,
      "grad_norm": 29.7295690026815,
      "learning_rate": 1.4139458715975998e-07,
      "loss": 1.9876,
      "step": 7470
    },
    {
      "epoch": 0.7630317249821483,
      "grad_norm": 26.83571576283127,
      "learning_rate": 1.402453156574232e-07,
      "loss": 1.684,
      "step": 7480
    },
    {
      "epoch": 0.7640518208711619,
      "grad_norm": 22.453312702326592,
      "learning_rate": 1.3909997155045526e-07,
      "loss": 1.6713,
      "step": 7490
    },
    {
      "epoch": 0.7650719167601755,
      "grad_norm": 36.398260519528165,
      "learning_rate": 1.379585673424296e-07,
      "loss": 1.9275,
      "step": 7500
    },
    {
      "epoch": 0.766092012649189,
      "grad_norm": 21.898480965127963,
      "learning_rate": 1.3682111549390862e-07,
      "loss": 1.6089,
      "step": 7510
    },
    {
      "epoch": 0.7671121085382026,
      "grad_norm": 29.351199703683985,
      "learning_rate": 1.356876284223069e-07,
      "loss": 1.5963,
      "step": 7520
    },
    {
      "epoch": 0.7681322044272162,
      "grad_norm": 17.20792272950376,
      "learning_rate": 1.345581185017563e-07,
      "loss": 1.5152,
      "step": 7530
    },
    {
      "epoch": 0.7691523003162297,
      "grad_norm": 18.19910721301402,
      "learning_rate": 1.3343259806297036e-07,
      "loss": 1.4488,
      "step": 7540
    },
    {
      "epoch": 0.7701723962052432,
      "grad_norm": 23.516530979837857,
      "learning_rate": 1.3231107939311042e-07,
      "loss": 1.5833,
      "step": 7550
    },
    {
      "epoch": 0.7711924920942569,
      "grad_norm": 49.23369882197886,
      "learning_rate": 1.3119357473565073e-07,
      "loss": 1.9783,
      "step": 7560
    },
    {
      "epoch": 0.7722125879832704,
      "grad_norm": 23.41167908978289,
      "learning_rate": 1.3008009629024515e-07,
      "loss": 1.817,
      "step": 7570
    },
    {
      "epoch": 0.7732326838722839,
      "grad_norm": 25.588838732555278,
      "learning_rate": 1.2897065621259372e-07,
      "loss": 1.6839,
      "step": 7580
    },
    {
      "epoch": 0.7742527797612976,
      "grad_norm": 19.412244192725158,
      "learning_rate": 1.278652666143103e-07,
      "loss": 1.7345,
      "step": 7590
    },
    {
      "epoch": 0.7752728756503111,
      "grad_norm": 28.69162997978439,
      "learning_rate": 1.2676393956279047e-07,
      "loss": 1.9476,
      "step": 7600
    },
    {
      "epoch": 0.7762929715393247,
      "grad_norm": 18.98179052685624,
      "learning_rate": 1.2566668708107885e-07,
      "loss": 2.0006,
      "step": 7610
    },
    {
      "epoch": 0.7773130674283383,
      "grad_norm": 29.430286154414116,
      "learning_rate": 1.2457352114773946e-07,
      "loss": 1.7007,
      "step": 7620
    },
    {
      "epoch": 0.7783331633173518,
      "grad_norm": 24.150404262972344,
      "learning_rate": 1.2348445369672277e-07,
      "loss": 1.6416,
      "step": 7630
    },
    {
      "epoch": 0.7793532592063654,
      "grad_norm": 20.96048945613088,
      "learning_rate": 1.2239949661723781e-07,
      "loss": 1.5704,
      "step": 7640
    },
    {
      "epoch": 0.780373355095379,
      "grad_norm": 23.113865810286573,
      "learning_rate": 1.2131866175362037e-07,
      "loss": 1.8523,
      "step": 7650
    },
    {
      "epoch": 0.7813934509843925,
      "grad_norm": 35.40691994141348,
      "learning_rate": 1.2024196090520528e-07,
      "loss": 2.0984,
      "step": 7660
    },
    {
      "epoch": 0.7824135468734061,
      "grad_norm": 32.23340175464063,
      "learning_rate": 1.1916940582619623e-07,
      "loss": 1.888,
      "step": 7670
    },
    {
      "epoch": 0.7834336427624197,
      "grad_norm": 23.265381285075236,
      "learning_rate": 1.1810100822553831e-07,
      "loss": 1.7629,
      "step": 7680
    },
    {
      "epoch": 0.7844537386514332,
      "grad_norm": 23.36424662106401,
      "learning_rate": 1.1703677976679005e-07,
      "loss": 1.4915,
      "step": 7690
    },
    {
      "epoch": 0.7854738345404468,
      "grad_norm": 38.54768696008822,
      "learning_rate": 1.1597673206799574e-07,
      "loss": 2.2836,
      "step": 7700
    },
    {
      "epoch": 0.7864939304294604,
      "grad_norm": 33.339067354592885,
      "learning_rate": 1.1492087670155937e-07,
      "loss": 1.8171,
      "step": 7710
    },
    {
      "epoch": 0.7875140263184739,
      "grad_norm": 16.642007470160333,
      "learning_rate": 1.1386922519411718e-07,
      "loss": 2.3708,
      "step": 7720
    },
    {
      "epoch": 0.7885341222074875,
      "grad_norm": 23.8154992099616,
      "learning_rate": 1.1282178902641282e-07,
      "loss": 2.074,
      "step": 7730
    },
    {
      "epoch": 0.7895542180965011,
      "grad_norm": 22.29321391651824,
      "learning_rate": 1.1177857963317122e-07,
      "loss": 1.6586,
      "step": 7740
    },
    {
      "epoch": 0.7905743139855146,
      "grad_norm": 20.849430869839942,
      "learning_rate": 1.107396084029748e-07,
      "loss": 1.5718,
      "step": 7750
    },
    {
      "epoch": 0.7915944098745282,
      "grad_norm": 26.562594063248618,
      "learning_rate": 1.0970488667813799e-07,
      "loss": 1.4991,
      "step": 7760
    },
    {
      "epoch": 0.7926145057635418,
      "grad_norm": 18.782121908759255,
      "learning_rate": 1.0867442575458402e-07,
      "loss": 1.6167,
      "step": 7770
    },
    {
      "epoch": 0.7936346016525554,
      "grad_norm": 22.935119257066393,
      "learning_rate": 1.0764823688172142e-07,
      "loss": 1.7465,
      "step": 7780
    },
    {
      "epoch": 0.7946546975415689,
      "grad_norm": 13.617513849153339,
      "learning_rate": 1.0662633126232168e-07,
      "loss": 1.6951,
      "step": 7790
    },
    {
      "epoch": 0.7956747934305824,
      "grad_norm": 16.521636961545013,
      "learning_rate": 1.0560872005239602e-07,
      "loss": 1.8369,
      "step": 7800
    },
    {
      "epoch": 0.7966948893195961,
      "grad_norm": 19.61970690445494,
      "learning_rate": 1.0459541436107488e-07,
      "loss": 1.6288,
      "step": 7810
    },
    {
      "epoch": 0.7977149852086096,
      "grad_norm": 21.481480755256573,
      "learning_rate": 1.0358642525048505e-07,
      "loss": 1.9997,
      "step": 7820
    },
    {
      "epoch": 0.7987350810976231,
      "grad_norm": 27.493809644509412,
      "learning_rate": 1.0258176373563021e-07,
      "loss": 1.4507,
      "step": 7830
    },
    {
      "epoch": 0.7997551769866368,
      "grad_norm": 21.428809296982042,
      "learning_rate": 1.0158144078427033e-07,
      "loss": 1.8233,
      "step": 7840
    },
    {
      "epoch": 0.8007752728756503,
      "grad_norm": 33.71428678156929,
      "learning_rate": 1.0058546731680162e-07,
      "loss": 1.5754,
      "step": 7850
    },
    {
      "epoch": 0.8017953687646638,
      "grad_norm": 38.77495277849617,
      "learning_rate": 9.959385420613792e-08,
      "loss": 1.6129,
      "step": 7860
    },
    {
      "epoch": 0.8028154646536775,
      "grad_norm": 19.510635912572226,
      "learning_rate": 9.860661227759087e-08,
      "loss": 2.2188,
      "step": 7870
    },
    {
      "epoch": 0.803835560542691,
      "grad_norm": 16.723040670656548,
      "learning_rate": 9.76237523087533e-08,
      "loss": 1.4559,
      "step": 7880
    },
    {
      "epoch": 0.8048556564317045,
      "grad_norm": 32.15243822669214,
      "learning_rate": 9.664528502938014e-08,
      "loss": 1.7599,
      "step": 7890
    },
    {
      "epoch": 0.8058757523207182,
      "grad_norm": 25.86596746303313,
      "learning_rate": 9.567122112127241e-08,
      "loss": 1.8331,
      "step": 7900
    },
    {
      "epoch": 0.8068958482097317,
      "grad_norm": 23.041725719423933,
      "learning_rate": 9.470157121815974e-08,
      "loss": 1.9208,
      "step": 7910
    },
    {
      "epoch": 0.8079159440987452,
      "grad_norm": 35.06406383217375,
      "learning_rate": 9.373634590558472e-08,
      "loss": 1.8437,
      "step": 7920
    },
    {
      "epoch": 0.8089360399877589,
      "grad_norm": 41.25136452735605,
      "learning_rate": 9.27755557207871e-08,
      "loss": 1.7273,
      "step": 7930
    },
    {
      "epoch": 0.8099561358767724,
      "grad_norm": 39.2089012904391,
      "learning_rate": 9.181921115258945e-08,
      "loss": 1.6692,
      "step": 7940
    },
    {
      "epoch": 0.810976231765786,
      "grad_norm": 26.782445647149878,
      "learning_rate": 9.086732264128149e-08,
      "loss": 1.6867,
      "step": 7950
    },
    {
      "epoch": 0.8119963276547996,
      "grad_norm": 20.959291278838805,
      "learning_rate": 8.991990057850713e-08,
      "loss": 1.6255,
      "step": 7960
    },
    {
      "epoch": 0.8130164235438131,
      "grad_norm": 17.835986787870524,
      "learning_rate": 8.897695530715044e-08,
      "loss": 1.5361,
      "step": 7970
    },
    {
      "epoch": 0.8140365194328267,
      "grad_norm": 28.33321886823143,
      "learning_rate": 8.803849712122291e-08,
      "loss": 1.7765,
      "step": 7980
    },
    {
      "epoch": 0.8150566153218403,
      "grad_norm": 19.032617699485062,
      "learning_rate": 8.710453626575142e-08,
      "loss": 1.6631,
      "step": 7990
    },
    {
      "epoch": 0.8160767112108538,
      "grad_norm": 27.226247437176582,
      "learning_rate": 8.617508293666581e-08,
      "loss": 1.7236,
      "step": 8000
    },
    {
      "epoch": 0.8170968070998674,
      "grad_norm": 20.522362263164414,
      "learning_rate": 8.525014728068786e-08,
      "loss": 1.5542,
      "step": 8010
    },
    {
      "epoch": 0.818116902988881,
      "grad_norm": 19.795379207693646,
      "learning_rate": 8.432973939522042e-08,
      "loss": 1.6463,
      "step": 8020
    },
    {
      "epoch": 0.8191369988778945,
      "grad_norm": 23.33391554065709,
      "learning_rate": 8.341386932823752e-08,
      "loss": 1.6336,
      "step": 8030
    },
    {
      "epoch": 0.8201570947669081,
      "grad_norm": 19.108059278422203,
      "learning_rate": 8.250254707817405e-08,
      "loss": 1.5987,
      "step": 8040
    },
    {
      "epoch": 0.8211771906559217,
      "grad_norm": 27.01218055222573,
      "learning_rate": 8.159578259381756e-08,
      "loss": 1.6374,
      "step": 8050
    },
    {
      "epoch": 0.8221972865449352,
      "grad_norm": 35.376440799029076,
      "learning_rate": 8.069358577419816e-08,
      "loss": 1.9243,
      "step": 8060
    },
    {
      "epoch": 0.8232173824339488,
      "grad_norm": 22.668709450006613,
      "learning_rate": 7.979596646848224e-08,
      "loss": 1.8702,
      "step": 8070
    },
    {
      "epoch": 0.8242374783229623,
      "grad_norm": 25.39771793053187,
      "learning_rate": 7.890293447586344e-08,
      "loss": 2.3178,
      "step": 8080
    },
    {
      "epoch": 0.8252575742119759,
      "grad_norm": 23.22588807056862,
      "learning_rate": 7.80144995454568e-08,
      "loss": 1.6089,
      "step": 8090
    },
    {
      "epoch": 0.8262776701009895,
      "grad_norm": 24.582595070781984,
      "learning_rate": 7.713067137619155e-08,
      "loss": 1.7091,
      "step": 8100
    },
    {
      "epoch": 0.827297765990003,
      "grad_norm": 16.20241718588193,
      "learning_rate": 7.625145961670515e-08,
      "loss": 1.5651,
      "step": 8110
    },
    {
      "epoch": 0.8283178618790167,
      "grad_norm": 25.486906956328838,
      "learning_rate": 7.537687386523912e-08,
      "loss": 1.6922,
      "step": 8120
    },
    {
      "epoch": 0.8293379577680302,
      "grad_norm": 22.594393753543923,
      "learning_rate": 7.450692366953271e-08,
      "loss": 1.7154,
      "step": 8130
    },
    {
      "epoch": 0.8303580536570437,
      "grad_norm": 22.873497190459474,
      "learning_rate": 7.364161852671996e-08,
      "loss": 1.6792,
      "step": 8140
    },
    {
      "epoch": 0.8313781495460574,
      "grad_norm": 20.712327174183876,
      "learning_rate": 7.278096788322512e-08,
      "loss": 1.8545,
      "step": 8150
    },
    {
      "epoch": 0.8323982454350709,
      "grad_norm": 20.598797117492094,
      "learning_rate": 7.192498113465995e-08,
      "loss": 1.6087,
      "step": 8160
    },
    {
      "epoch": 0.8334183413240844,
      "grad_norm": 35.74460986996694,
      "learning_rate": 7.107366762572104e-08,
      "loss": 1.75,
      "step": 8170
    },
    {
      "epoch": 0.8344384372130981,
      "grad_norm": 30.918337301265726,
      "learning_rate": 7.0227036650088e-08,
      "loss": 1.6747,
      "step": 8180
    },
    {
      "epoch": 0.8354585331021116,
      "grad_norm": 19.153431500017803,
      "learning_rate": 6.938509745032172e-08,
      "loss": 1.5901,
      "step": 8190
    },
    {
      "epoch": 0.8364786289911251,
      "grad_norm": 23.29877499692314,
      "learning_rate": 6.854785921776352e-08,
      "loss": 1.8583,
      "step": 8200
    },
    {
      "epoch": 0.8374987248801388,
      "grad_norm": 33.8369922024749,
      "learning_rate": 6.771533109243476e-08,
      "loss": 1.9186,
      "step": 8210
    },
    {
      "epoch": 0.8385188207691523,
      "grad_norm": 16.55193364159388,
      "learning_rate": 6.688752216293764e-08,
      "loss": 1.6276,
      "step": 8220
    },
    {
      "epoch": 0.8395389166581658,
      "grad_norm": 19.259723207160086,
      "learning_rate": 6.606444146635503e-08,
      "loss": 1.8841,
      "step": 8230
    },
    {
      "epoch": 0.8405590125471795,
      "grad_norm": 25.603281148511876,
      "learning_rate": 6.524609798815251e-08,
      "loss": 2.2438,
      "step": 8240
    },
    {
      "epoch": 0.841579108436193,
      "grad_norm": 20.841444915712064,
      "learning_rate": 6.443250066208e-08,
      "loss": 1.6694,
      "step": 8250
    },
    {
      "epoch": 0.8425992043252065,
      "grad_norm": 84.61902995038496,
      "learning_rate": 6.362365837007411e-08,
      "loss": 1.9662,
      "step": 8260
    },
    {
      "epoch": 0.8436193002142202,
      "grad_norm": 30.005187808234442,
      "learning_rate": 6.281957994216187e-08,
      "loss": 1.608,
      "step": 8270
    },
    {
      "epoch": 0.8446393961032337,
      "grad_norm": 18.089471678174192,
      "learning_rate": 6.202027415636325e-08,
      "loss": 1.8912,
      "step": 8280
    },
    {
      "epoch": 0.8456594919922473,
      "grad_norm": 20.4298166188425,
      "learning_rate": 6.122574973859645e-08,
      "loss": 1.6629,
      "step": 8290
    },
    {
      "epoch": 0.8466795878812609,
      "grad_norm": 26.030727947323555,
      "learning_rate": 6.043601536258141e-08,
      "loss": 1.8695,
      "step": 8300
    },
    {
      "epoch": 0.8476996837702744,
      "grad_norm": 16.731253061966548,
      "learning_rate": 5.965107964974652e-08,
      "loss": 1.8049,
      "step": 8310
    },
    {
      "epoch": 0.848719779659288,
      "grad_norm": 32.19178946874123,
      "learning_rate": 5.887095116913321e-08,
      "loss": 1.5545,
      "step": 8320
    },
    {
      "epoch": 0.8497398755483015,
      "grad_norm": 37.411440627682275,
      "learning_rate": 5.809563843730342e-08,
      "loss": 1.5905,
      "step": 8330
    },
    {
      "epoch": 0.8507599714373151,
      "grad_norm": 26.742774156366035,
      "learning_rate": 5.732514991824594e-08,
      "loss": 1.5766,
      "step": 8340
    },
    {
      "epoch": 0.8517800673263287,
      "grad_norm": 21.079365800931345,
      "learning_rate": 5.655949402328425e-08,
      "loss": 1.8807,
      "step": 8350
    },
    {
      "epoch": 0.8528001632153422,
      "grad_norm": 24.700393745644252,
      "learning_rate": 5.5798679110984625e-08,
      "loss": 1.9346,
      "step": 8360
    },
    {
      "epoch": 0.8538202591043558,
      "grad_norm": 22.16602828680004,
      "learning_rate": 5.504271348706541e-08,
      "loss": 1.6555,
      "step": 8370
    },
    {
      "epoch": 0.8548403549933694,
      "grad_norm": 17.97859389067489,
      "learning_rate": 5.429160540430533e-08,
      "loss": 1.6666,
      "step": 8380
    },
    {
      "epoch": 0.8558604508823829,
      "grad_norm": 18.3000648240371,
      "learning_rate": 5.354536306245422e-08,
      "loss": 1.9014,
      "step": 8390
    },
    {
      "epoch": 0.8568805467713965,
      "grad_norm": 25.96896645899262,
      "learning_rate": 5.280399460814333e-08,
      "loss": 1.7755,
      "step": 8400
    },
    {
      "epoch": 0.8579006426604101,
      "grad_norm": 22.959088643369757,
      "learning_rate": 5.206750813479599e-08,
      "loss": 1.768,
      "step": 8410
    },
    {
      "epoch": 0.8589207385494236,
      "grad_norm": 24.20527508056167,
      "learning_rate": 5.1335911682540034e-08,
      "loss": 1.8964,
      "step": 8420
    },
    {
      "epoch": 0.8599408344384372,
      "grad_norm": 17.69263103840439,
      "learning_rate": 5.060921323811923e-08,
      "loss": 1.6816,
      "step": 8430
    },
    {
      "epoch": 0.8609609303274508,
      "grad_norm": 19.23728502101208,
      "learning_rate": 4.9887420734806564e-08,
      "loss": 1.9728,
      "step": 8440
    },
    {
      "epoch": 0.8619810262164643,
      "grad_norm": 15.684029190511257,
      "learning_rate": 4.91705420523173e-08,
      "loss": 1.7444,
      "step": 8450
    },
    {
      "epoch": 0.863001122105478,
      "grad_norm": 31.200613159114106,
      "learning_rate": 4.845858501672356e-08,
      "loss": 1.5774,
      "step": 8460
    },
    {
      "epoch": 0.8640212179944915,
      "grad_norm": 39.36573898938769,
      "learning_rate": 4.7751557400368014e-08,
      "loss": 1.7016,
      "step": 8470
    },
    {
      "epoch": 0.865041313883505,
      "grad_norm": 31.200966794940115,
      "learning_rate": 4.704946692178002e-08,
      "loss": 1.8716,
      "step": 8480
    },
    {
      "epoch": 0.8660614097725187,
      "grad_norm": 18.022185177712775,
      "learning_rate": 4.635232124559013e-08,
      "loss": 1.6874,
      "step": 8490
    },
    {
      "epoch": 0.8670815056615322,
      "grad_norm": 34.54288967600377,
      "learning_rate": 4.56601279824479e-08,
      "loss": 1.8029,
      "step": 8500
    },
    {
      "epoch": 0.8681016015505457,
      "grad_norm": 29.945358371181094,
      "learning_rate": 4.4972894688937504e-08,
      "loss": 1.6582,
      "step": 8510
    },
    {
      "epoch": 0.8691216974395594,
      "grad_norm": 22.11685165201178,
      "learning_rate": 4.42906288674959e-08,
      "loss": 1.6797,
      "step": 8520
    },
    {
      "epoch": 0.8701417933285729,
      "grad_norm": 29.006836293341163,
      "learning_rate": 4.3613337966331046e-08,
      "loss": 1.7209,
      "step": 8530
    },
    {
      "epoch": 0.8711618892175864,
      "grad_norm": 15.092932386620712,
      "learning_rate": 4.2941029379339966e-08,
      "loss": 1.5863,
      "step": 8540
    },
    {
      "epoch": 0.8721819851066,
      "grad_norm": 22.010293299859292,
      "learning_rate": 4.227371044602862e-08,
      "loss": 1.5257,
      "step": 8550
    },
    {
      "epoch": 0.8732020809956136,
      "grad_norm": 21.32824593921752,
      "learning_rate": 4.1611388451431363e-08,
      "loss": 1.6375,
      "step": 8560
    },
    {
      "epoch": 0.8742221768846271,
      "grad_norm": 22.351368765383427,
      "learning_rate": 4.095407062603201e-08,
      "loss": 1.7738,
      "step": 8570
    },
    {
      "epoch": 0.8752422727736408,
      "grad_norm": 127.80618513244573,
      "learning_rate": 4.0301764145684145e-08,
      "loss": 2.2276,
      "step": 8580
    },
    {
      "epoch": 0.8762623686626543,
      "grad_norm": 15.563935367425097,
      "learning_rate": 3.9654476131533224e-08,
      "loss": 1.7511,
      "step": 8590
    },
    {
      "epoch": 0.8772824645516678,
      "grad_norm": 23.826571233036272,
      "learning_rate": 3.901221364993879e-08,
      "loss": 1.8109,
      "step": 8600
    },
    {
      "epoch": 0.8783025604406814,
      "grad_norm": 21.757144539499144,
      "learning_rate": 3.8374983712397446e-08,
      "loss": 1.6326,
      "step": 8610
    },
    {
      "epoch": 0.879322656329695,
      "grad_norm": 20.391836466293952,
      "learning_rate": 3.774279327546587e-08,
      "loss": 1.5786,
      "step": 8620
    },
    {
      "epoch": 0.8803427522187086,
      "grad_norm": 14.967953905821696,
      "learning_rate": 3.711564924068539e-08,
      "loss": 1.7269,
      "step": 8630
    },
    {
      "epoch": 0.8813628481077221,
      "grad_norm": 17.373655036007143,
      "learning_rate": 3.649355845450636e-08,
      "loss": 1.7306,
      "step": 8640
    },
    {
      "epoch": 0.8823829439967357,
      "grad_norm": 17.649447315823274,
      "learning_rate": 3.5876527708213256e-08,
      "loss": 1.608,
      "step": 8650
    },
    {
      "epoch": 0.8834030398857493,
      "grad_norm": 23.997451813665414,
      "learning_rate": 3.5264563737851196e-08,
      "loss": 1.6461,
      "step": 8660
    },
    {
      "epoch": 0.8844231357747628,
      "grad_norm": 15.579820728923217,
      "learning_rate": 3.465767322415164e-08,
      "loss": 1.4876,
      "step": 8670
    },
    {
      "epoch": 0.8854432316637764,
      "grad_norm": 26.363140475561018,
      "learning_rate": 3.405586279245981e-08,
      "loss": 1.8793,
      "step": 8680
    },
    {
      "epoch": 0.88646332755279,
      "grad_norm": 16.90307380765635,
      "learning_rate": 3.345913901266223e-08,
      "loss": 2.0268,
      "step": 8690
    },
    {
      "epoch": 0.8874834234418035,
      "grad_norm": 20.895186543223687,
      "learning_rate": 3.286750839911556e-08,
      "loss": 1.574,
      "step": 8700
    },
    {
      "epoch": 0.8885035193308171,
      "grad_norm": 18.584343988333305,
      "learning_rate": 3.22809774105744e-08,
      "loss": 1.544,
      "step": 8710
    },
    {
      "epoch": 0.8895236152198307,
      "grad_norm": 24.633105227414962,
      "learning_rate": 3.169955245012201e-08,
      "loss": 1.6392,
      "step": 8720
    },
    {
      "epoch": 0.8905437111088442,
      "grad_norm": 25.34636662643249,
      "learning_rate": 3.112323986509918e-08,
      "loss": 1.7424,
      "step": 8730
    },
    {
      "epoch": 0.8915638069978578,
      "grad_norm": 20.657323225471508,
      "learning_rate": 3.0552045947036085e-08,
      "loss": 1.6193,
      "step": 8740
    },
    {
      "epoch": 0.8925839028868714,
      "grad_norm": 27.278618821016288,
      "learning_rate": 2.9985976931582724e-08,
      "loss": 1.6316,
      "step": 8750
    },
    {
      "epoch": 0.8936039987758849,
      "grad_norm": 18.764798375928606,
      "learning_rate": 2.942503899844151e-08,
      "loss": 1.818,
      "step": 8760
    },
    {
      "epoch": 0.8946240946648985,
      "grad_norm": 19.12416141998992,
      "learning_rate": 2.8869238271299222e-08,
      "loss": 1.7018,
      "step": 8770
    },
    {
      "epoch": 0.8956441905539121,
      "grad_norm": 18.210386823268735,
      "learning_rate": 2.8318580817760563e-08,
      "loss": 1.7672,
      "step": 8780
    },
    {
      "epoch": 0.8966642864429256,
      "grad_norm": 20.657715477282977,
      "learning_rate": 2.7773072649281747e-08,
      "loss": 1.9867,
      "step": 8790
    },
    {
      "epoch": 0.8976843823319393,
      "grad_norm": 15.420457410537264,
      "learning_rate": 2.7232719721104803e-08,
      "loss": 1.5112,
      "step": 8800
    },
    {
      "epoch": 0.8987044782209528,
      "grad_norm": 29.30582102522778,
      "learning_rate": 2.6697527932192875e-08,
      "loss": 1.8693,
      "step": 8810
    },
    {
      "epoch": 0.8997245741099663,
      "grad_norm": 34.763769134622855,
      "learning_rate": 2.616750312516558e-08,
      "loss": 1.6838,
      "step": 8820
    },
    {
      "epoch": 0.90074466999898,
      "grad_norm": 27.498841907126955,
      "learning_rate": 2.564265108623509e-08,
      "loss": 1.7698,
      "step": 8830
    },
    {
      "epoch": 0.9017647658879935,
      "grad_norm": 16.226408671974106,
      "learning_rate": 2.5122977545143197e-08,
      "loss": 1.4913,
      "step": 8840
    },
    {
      "epoch": 0.902784861777007,
      "grad_norm": 17.423083222104857,
      "learning_rate": 2.4608488175098917e-08,
      "loss": 1.3155,
      "step": 8850
    },
    {
      "epoch": 0.9038049576660206,
      "grad_norm": 33.71226172329001,
      "learning_rate": 2.4099188592716093e-08,
      "loss": 1.7765,
      "step": 8860
    },
    {
      "epoch": 0.9048250535550342,
      "grad_norm": 21.39035678232929,
      "learning_rate": 2.359508435795238e-08,
      "loss": 1.7735,
      "step": 8870
    },
    {
      "epoch": 0.9058451494440477,
      "grad_norm": 46.991898862560774,
      "learning_rate": 2.3096180974048474e-08,
      "loss": 1.8635,
      "step": 8880
    },
    {
      "epoch": 0.9068652453330613,
      "grad_norm": 20.278911974730423,
      "learning_rate": 2.2602483887468217e-08,
      "loss": 1.8143,
      "step": 8890
    },
    {
      "epoch": 0.9078853412220749,
      "grad_norm": 22.256106887029187,
      "learning_rate": 2.211399848783879e-08,
      "loss": 1.8625,
      "step": 8900
    },
    {
      "epoch": 0.9089054371110884,
      "grad_norm": 35.0994813155819,
      "learning_rate": 2.163073010789229e-08,
      "loss": 1.6533,
      "step": 8910
    },
    {
      "epoch": 0.909925533000102,
      "grad_norm": 37.331167996216394,
      "learning_rate": 2.1152684023406918e-08,
      "loss": 1.705,
      "step": 8920
    },
    {
      "epoch": 0.9109456288891156,
      "grad_norm": 36.376841460429326,
      "learning_rate": 2.0679865453150038e-08,
      "loss": 1.8444,
      "step": 8930
    },
    {
      "epoch": 0.9119657247781291,
      "grad_norm": 38.802731118002455,
      "learning_rate": 2.0212279558820877e-08,
      "loss": 1.6286,
      "step": 8940
    },
    {
      "epoch": 0.9129858206671427,
      "grad_norm": 17.643644453874295,
      "learning_rate": 1.9749931444994095e-08,
      "loss": 1.6759,
      "step": 8950
    },
    {
      "epoch": 0.9140059165561563,
      "grad_norm": 26.406284822043514,
      "learning_rate": 1.9292826159064357e-08,
      "loss": 1.687,
      "step": 8960
    },
    {
      "epoch": 0.9150260124451699,
      "grad_norm": 21.035363518865424,
      "learning_rate": 1.884096869119095e-08,
      "loss": 1.8425,
      "step": 8970
    },
    {
      "epoch": 0.9160461083341834,
      "grad_norm": 20.650263900347426,
      "learning_rate": 1.8394363974243375e-08,
      "loss": 1.5209,
      "step": 8980
    },
    {
      "epoch": 0.917066204223197,
      "grad_norm": 19.235671373073703,
      "learning_rate": 1.795301688374756e-08,
      "loss": 1.542,
      "step": 8990
    },
    {
      "epoch": 0.9180863001122106,
      "grad_norm": 25.035962748311757,
      "learning_rate": 1.751693223783274e-08,
      "loss": 1.7444,
      "step": 9000
    },
    {
      "epoch": 0.9191063960012241,
      "grad_norm": 21.423592368547617,
      "learning_rate": 1.7086114797178475e-08,
      "loss": 1.596,
      "step": 9010
    },
    {
      "epoch": 0.9201264918902377,
      "grad_norm": 30.802170899218233,
      "learning_rate": 1.666056926496312e-08,
      "loss": 1.7901,
      "step": 9020
    },
    {
      "epoch": 0.9211465877792513,
      "grad_norm": 23.146789286179963,
      "learning_rate": 1.624030028681217e-08,
      "loss": 1.7938,
      "step": 9030
    },
    {
      "epoch": 0.9221666836682648,
      "grad_norm": 28.849187812310362,
      "learning_rate": 1.582531245074775e-08,
      "loss": 1.6868,
      "step": 9040
    },
    {
      "epoch": 0.9231867795572783,
      "grad_norm": 26.9045170692709,
      "learning_rate": 1.5415610287138447e-08,
      "loss": 2.0242,
      "step": 9050
    },
    {
      "epoch": 0.924206875446292,
      "grad_norm": 26.77470849592168,
      "learning_rate": 1.5011198268649772e-08,
      "loss": 1.7992,
      "step": 9060
    },
    {
      "epoch": 0.9252269713353055,
      "grad_norm": 38.29672679178296,
      "learning_rate": 1.4612080810195381e-08,
      "loss": 1.6073,
      "step": 9070
    },
    {
      "epoch": 0.926247067224319,
      "grad_norm": 19.968548383518595,
      "learning_rate": 1.4218262268889004e-08,
      "loss": 1.5158,
      "step": 9080
    },
    {
      "epoch": 0.9272671631133327,
      "grad_norm": 19.657545251084848,
      "learning_rate": 1.3829746943996745e-08,
      "loss": 1.641,
      "step": 9090
    },
    {
      "epoch": 0.9282872590023462,
      "grad_norm": 23.5048643544073,
      "learning_rate": 1.344653907689014e-08,
      "loss": 1.735,
      "step": 9100
    },
    {
      "epoch": 0.9293073548913597,
      "grad_norm": 27.1267998381856,
      "learning_rate": 1.3068642850999956e-08,
      "loss": 1.3247,
      "step": 9110
    },
    {
      "epoch": 0.9303274507803734,
      "grad_norm": 20.22101708211367,
      "learning_rate": 1.2696062391770345e-08,
      "loss": 2.3513,
      "step": 9120
    },
    {
      "epoch": 0.9313475466693869,
      "grad_norm": 18.580423295596134,
      "learning_rate": 1.232880176661405e-08,
      "loss": 1.639,
      "step": 9130
    },
    {
      "epoch": 0.9323676425584005,
      "grad_norm": 24.168158667351197,
      "learning_rate": 1.1966864984867874e-08,
      "loss": 1.8946,
      "step": 9140
    },
    {
      "epoch": 0.9333877384474141,
      "grad_norm": 16.66842625477143,
      "learning_rate": 1.1610255997748787e-08,
      "loss": 1.5375,
      "step": 9150
    },
    {
      "epoch": 0.9344078343364276,
      "grad_norm": 20.024655550399938,
      "learning_rate": 1.1258978698311051e-08,
      "loss": 1.7201,
      "step": 9160
    },
    {
      "epoch": 0.9354279302254412,
      "grad_norm": 62.69424852226504,
      "learning_rate": 1.0913036921403495e-08,
      "loss": 1.9306,
      "step": 9170
    },
    {
      "epoch": 0.9364480261144548,
      "grad_norm": 21.93361259568472,
      "learning_rate": 1.0572434443627765e-08,
      "loss": 1.6325,
      "step": 9180
    },
    {
      "epoch": 0.9374681220034683,
      "grad_norm": 21.64108449158381,
      "learning_rate": 1.0237174983297014e-08,
      "loss": 1.6123,
      "step": 9190
    },
    {
      "epoch": 0.9384882178924819,
      "grad_norm": 12.33299753596877,
      "learning_rate": 9.907262200395505e-09,
      "loss": 1.7191,
      "step": 9200
    },
    {
      "epoch": 0.9395083137814955,
      "grad_norm": 24.558758892963382,
      "learning_rate": 9.58269969653841e-09,
      "loss": 1.6432,
      "step": 9210
    },
    {
      "epoch": 0.940528409670509,
      "grad_norm": 29.10763619992857,
      "learning_rate": 9.263491014932568e-09,
      "loss": 1.6456,
      "step": 9220
    },
    {
      "epoch": 0.9415485055595226,
      "grad_norm": 20.359271009625445,
      "learning_rate": 8.949639640337846e-09,
      "loss": 1.7498,
      "step": 9230
    },
    {
      "epoch": 0.9425686014485362,
      "grad_norm": 80.07437770872143,
      "learning_rate": 8.641148999029235e-09,
      "loss": 1.6732,
      "step": 9240
    },
    {
      "epoch": 0.9435886973375497,
      "grad_norm": 16.47093403823312,
      "learning_rate": 8.338022458759086e-09,
      "loss": 1.7996,
      "step": 9250
    },
    {
      "epoch": 0.9446087932265633,
      "grad_norm": 19.053050045651066,
      "learning_rate": 8.040263328720653e-09,
      "loss": 1.7543,
      "step": 9260
    },
    {
      "epoch": 0.9456288891155769,
      "grad_norm": 17.449182067934196,
      "learning_rate": 7.747874859511894e-09,
      "loss": 1.765,
      "step": 9270
    },
    {
      "epoch": 0.9466489850045904,
      "grad_norm": 25.01504890557875,
      "learning_rate": 7.460860243099887e-09,
      "loss": 1.6661,
      "step": 9280
    },
    {
      "epoch": 0.947669080893604,
      "grad_norm": 15.925580850751395,
      "learning_rate": 7.17922261278614e-09,
      "loss": 1.501,
      "step": 9290
    },
    {
      "epoch": 0.9486891767826175,
      "grad_norm": 25.812612487055343,
      "learning_rate": 6.902965043172116e-09,
      "loss": 1.7599,
      "step": 9300
    },
    {
      "epoch": 0.9497092726716312,
      "grad_norm": 20.913017487563426,
      "learning_rate": 6.632090550126146e-09,
      "loss": 1.9602,
      "step": 9310
    },
    {
      "epoch": 0.9507293685606447,
      "grad_norm": 20.542250837845,
      "learning_rate": 6.366602090749851e-09,
      "loss": 1.5759,
      "step": 9320
    },
    {
      "epoch": 0.9517494644496582,
      "grad_norm": 26.69332932713857,
      "learning_rate": 6.106502563346605e-09,
      "loss": 1.8497,
      "step": 9330
    },
    {
      "epoch": 0.9527695603386719,
      "grad_norm": 21.015389017183082,
      "learning_rate": 5.851794807389232e-09,
      "loss": 1.9147,
      "step": 9340
    },
    {
      "epoch": 0.9537896562276854,
      "grad_norm": 26.52270852343564,
      "learning_rate": 5.602481603489307e-09,
      "loss": 1.8345,
      "step": 9350
    },
    {
      "epoch": 0.9548097521166989,
      "grad_norm": 19.389253853181625,
      "learning_rate": 5.358565673366899e-09,
      "loss": 1.9132,
      "step": 9360
    },
    {
      "epoch": 0.9558298480057126,
      "grad_norm": 20.13319347218972,
      "learning_rate": 5.1200496798207145e-09,
      "loss": 1.9792,
      "step": 9370
    },
    {
      "epoch": 0.9568499438947261,
      "grad_norm": 27.9245616395467,
      "learning_rate": 4.8869362266989456e-09,
      "loss": 1.6127,
      "step": 9380
    },
    {
      "epoch": 0.9578700397837396,
      "grad_norm": 20.99329851049813,
      "learning_rate": 4.659227858871073e-09,
      "loss": 1.6997,
      "step": 9390
    },
    {
      "epoch": 0.9588901356727533,
      "grad_norm": 25.019789484033957,
      "learning_rate": 4.436927062199836e-09,
      "loss": 1.7122,
      "step": 9400
    },
    {
      "epoch": 0.9599102315617668,
      "grad_norm": 25.327284335907226,
      "learning_rate": 4.220036263514359e-09,
      "loss": 1.7285,
      "step": 9410
    },
    {
      "epoch": 0.9609303274507803,
      "grad_norm": 23.302501234228355,
      "learning_rate": 4.008557830583348e-09,
      "loss": 1.645,
      "step": 9420
    },
    {
      "epoch": 0.961950423339794,
      "grad_norm": 19.911243106020095,
      "learning_rate": 3.802494072089601e-09,
      "loss": 1.4147,
      "step": 9430
    },
    {
      "epoch": 0.9629705192288075,
      "grad_norm": 13.940142408967743,
      "learning_rate": 3.601847237604483e-09,
      "loss": 1.6159,
      "step": 9440
    },
    {
      "epoch": 0.963990615117821,
      "grad_norm": 21.93825830657235,
      "learning_rate": 3.4066195175636027e-09,
      "loss": 1.7949,
      "step": 9450
    },
    {
      "epoch": 0.9650107110068347,
      "grad_norm": 23.199814946092065,
      "learning_rate": 3.2168130432426167e-09,
      "loss": 1.665,
      "step": 9460
    },
    {
      "epoch": 0.9660308068958482,
      "grad_norm": 22.780063112269392,
      "learning_rate": 3.0324298867343002e-09,
      "loss": 1.9056,
      "step": 9470
    },
    {
      "epoch": 0.9670509027848618,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.8534720609258436e-09,
      "loss": 1.971,
      "step": 9480
    },
    {
      "epoch": 0.9680709986738754,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.6799415194765364e-09,
      "loss": 0.0,
      "step": 9490
    },
    {
      "epoch": 0.9690910945628889,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.5118401567969517e-09,
      "loss": 0.0,
      "step": 9500
    },
    {
      "epoch": 0.9701111904519025,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.349169808027851e-09,
      "loss": 0.0,
      "step": 9510
    },
    {
      "epoch": 0.9711312863409161,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.191932249020534e-09,
      "loss": 0.0,
      "step": 9520
    },
    {
      "epoch": 0.9721513822299296,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.040129196316964e-09,
      "loss": 0.0,
      "step": 9530
    },
    {
      "epoch": 0.9731714781189432,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.8937623071314522e-09,
      "loss": 0.0,
      "step": 9540
    },
    {
      "epoch": 0.9741915740079568,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.7528331793325024e-09,
      "loss": 0.0,
      "step": 9550
    },
    {
      "epoch": 0.9752116698969703,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.617343351425049e-09,
      "loss": 0.0,
      "step": 9560
    },
    {
      "epoch": 0.9762317657859839,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.4872943025341367e-09,
      "loss": 0.0,
      "step": 9570
    },
    {
      "epoch": 0.9772518616749974,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.3626874523883225e-09,
      "loss": 0.0,
      "step": 9580
    },
    {
      "epoch": 0.978271957564011,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.2435241613045765e-09,
      "loss": 0.0,
      "step": 9590
    },
    {
      "epoch": 0.9792920534530246,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.1298057301730169e-09,
      "loss": 0.0,
      "step": 9600
    },
    {
      "epoch": 0.9803121493420381,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.0215334004431975e-09,
      "loss": 0.0,
      "step": 9610
    },
    {
      "epoch": 0.9813322452310517,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 9.187083541099538e-10,
      "loss": 0.0,
      "step": 9620
    },
    {
      "epoch": 0.9823523411200653,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 8.213317137011899e-10,
      "loss": 0.0,
      "step": 9630
    },
    {
      "epoch": 0.9833724370090788,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 7.294045422650552e-10,
      "loss": 0.0,
      "step": 9640
    },
    {
      "epoch": 0.9843925328980925,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 6.429278433585095e-10,
      "loss": 0.0,
      "step": 9650
    },
    {
      "epoch": 0.985412628787106,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 5.61902561036609e-10,
      "loss": 0.0,
      "step": 9660
    },
    {
      "epoch": 0.9864327246761195,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 4.863295798417932e-10,
      "loss": 0.0,
      "step": 9670
    },
    {
      "epoch": 0.9874528205651332,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 4.162097247945584e-10,
      "loss": 0.0,
      "step": 9680
    },
    {
      "epoch": 0.9884729164541467,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 3.515437613843542e-10,
      "loss": 0.0,
      "step": 9690
    },
    {
      "epoch": 0.9894930123431602,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.923323955610346e-10,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 0.9905131082321739,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 2.3857627372747545e-10,
      "loss": 0.0,
      "step": 9710
    },
    {
      "epoch": 0.9915332041211874,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.9027598273230194e-10,
      "loss": 0.0,
      "step": 9720
    },
    {
      "epoch": 0.9925533000102009,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.4743204986356063e-10,
      "loss": 0.0,
      "step": 9730
    },
    {
      "epoch": 0.9935733958992146,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.1004494284289067e-10,
      "loss": 0.0,
      "step": 9740
    },
    {
      "epoch": 0.9945934917882281,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 7.811506982052796e-11,
      "loss": 0.0,
      "step": 9750
    },
    {
      "epoch": 0.9956135876772416,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 5.164277937080852e-11,
      "loss": 0.0,
      "step": 9760
    },
    {
      "epoch": 0.9966336835662553,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 3.062836048822737e-11,
      "loss": 0.0,
      "step": 9770
    },
    {
      "epoch": 0.9976537794552688,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 1.5072042584551858e-11,
      "loss": 0.0,
      "step": 9780
    },
    {
      "epoch": 0.9986738753442823,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 4.973995486046156e-12,
      "loss": 0.0,
      "step": 9790
    },
    {
      "epoch": 0.999693971233296,
      "grad_norm": 1.4142135623730951,
      "learning_rate": 3.3432943175037977e-13,
      "loss": 0.0,
      "step": 9800
    }
  ],
  "logging_steps": 10,
  "max_steps": 9803,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}
