[
  {
    "step": 100,
    "loss": 7.657252192497253,
    "perplexity": 2115.935250373028,
    "learning_rate": 0.00015150000000000019,
    "step_time": 9.368009328842163,
    "tokens_per_second": 109.30817466708918,
    "memory_mb": 756.6015625
  },
  {
    "step": 200,
    "loss": 7.487581491470337,
    "perplexity": 1785.7280665447065,
    "learning_rate": 0.0002999999999999999,
    "step_time": 10.94300365447998,
    "tokens_per_second": 93.57577063229641,
    "memory_mb": 730.99609375
  },
  {
    "step": 300,
    "loss": 7.163171410560608,
    "perplexity": 1290.9987344513497,
    "learning_rate": 0.00029794904665665113,
    "step_time": 8.949450016021729,
    "tokens_per_second": 114.42043904002891,
    "memory_mb": 733.89453125
  },
  {
    "step": 400,
    "loss": 6.677380561828613,
    "perplexity": 794.2359329078598,
    "learning_rate": 0.00029185850380610337,
    "step_time": 8.572781324386597,
    "tokens_per_second": 119.44781527169884,
    "memory_mb": 827.69140625
  },
  {
    "step": 500,
    "loss": 6.8126842975616455,
    "perplexity": 909.3083881764071,
    "learning_rate": 0.0002819134295109075,
    "step_time": 8.801953315734863,
    "tokens_per_second": 116.33781312716582,
    "memory_mb": 849.21484375
  },
  {
    "step": 600,
    "loss": 6.901162624359131,
    "perplexity": 993.4290292468936,
    "learning_rate": 0.0002684159998210713,
    "step_time": 8.660848379135132,
    "tokens_per_second": 118.23322094714423,
    "memory_mb": 734.16015625
  },
  {
    "step": 700,
    "loss": 6.6995872259140015,
    "perplexity": 812.0705542959239,
    "learning_rate": 0.0002517763273076916,
    "step_time": 9.10423469543457,
    "tokens_per_second": 112.47513209578146,
    "memory_mb": 734.64453125
  },
  {
    "step": 800,
    "loss": 6.729204297065735,
    "perplexity": 836.4814103649661,
    "learning_rate": 0.00023250000000000793,
    "step_time": 8.646829605102539,
    "tokens_per_second": 118.42490794495734,
    "memory_mb": 733.84765625
  },
  {
    "step": 900,
    "loss": 6.801577568054199,
    "perplexity": 899.2648246887062,
    "learning_rate": 0.00021117271934897237,
    "step_time": 8.66753602027893,
    "tokens_per_second": 118.14199532649263,
    "memory_mb": 839.50390625
  },
  {
    "step": 1000,
    "loss": 6.5149312019348145,
    "perplexity": 675.1475110828569,
    "learning_rate": 0.00018844250398504186,
    "step_time": 9.189276933670044,
    "tokens_per_second": 111.43423006961567,
    "memory_mb": 734.55078125
  },
  {
    "step": 1100,
    "loss": 6.551132082939148,
    "perplexity": 700.0362244642821,
    "learning_rate": 0.00016500000000000537,
    "step_time": 8.517168760299683,
    "tokens_per_second": 120.22774572379963,
    "memory_mb": 913.80078125
  },
  {
    "step": 1200,
    "loss": 6.6684452295303345,
    "perplexity": 787.170782663513,
    "learning_rate": 0.00014155749601496882,
    "step_time": 9.400139570236206,
    "tokens_per_second": 108.93455276369572,
    "memory_mb": 968.5859375
  },
  {
    "step": 1300,
    "loss": 6.406905889511108,
    "perplexity": 606.0156976890383,
    "learning_rate": 0.00011882728065103813,
    "step_time": 8.490540981292725,
    "tokens_per_second": 120.6048003603289,
    "memory_mb": 979.359375
  },
  {
    "step": 1400,
    "loss": 6.3421419858932495,
    "perplexity": 568.011682273887,
    "learning_rate": 9.750000000000261e-05,
    "step_time": 10.675879955291748,
    "tokens_per_second": 95.91715196201983,
    "memory_mb": 733.6328125
  },
  {
    "step": 1500,
    "loss": 6.335531115531921,
    "perplexity": 564.2690154518974,
    "learning_rate": 7.822367269231907e-05,
    "step_time": 8.945564270019531,
    "tokens_per_second": 114.47014062957085,
    "memory_mb": 994.3515625
  },
  {
    "step": 1600,
    "loss": 6.629261136054993,
    "perplexity": 756.9227010883292,
    "learning_rate": 6.158400017893925e-05,
    "step_time": 10.422486782073975,
    "tokens_per_second": 98.2491051714468,
    "memory_mb": 730.56640625
  },
  {
    "step": 1700,
    "loss": 6.302608251571655,
    "perplexity": 545.9941446328027,
    "learning_rate": 4.808657048910149e-05,
    "step_time": 9.476832628250122,
    "tokens_per_second": 108.05297932006208,
    "memory_mb": 745.43359375
  },
  {
    "step": 1800,
    "loss": 6.266754984855652,
    "perplexity": 526.765240241726,
    "learning_rate": 3.8141496193902704e-05,
    "step_time": 8.648972511291504,
    "tokens_per_second": 118.3955664864394,
    "memory_mb": 1009.6484375
  },
  {
    "step": 1900,
    "loss": 6.480456352233887,
    "perplexity": 652.268542568135,
    "learning_rate": 3.2050953343351995e-05,
    "step_time": 8.593879699707031,
    "tokens_per_second": 119.15456531639704,
    "memory_mb": 735.796875
  },
  {
    "step": 2000,
    "loss": 5.975515604019165,
    "perplexity": 393.6710271356782,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 8.94594669342041,
    "tokens_per_second": 114.46524723349116,
    "memory_mb": 729.84375
  },
  {
    "step": 2100,
    "loss": 6.554613709449768,
    "perplexity": 702.4777368926921,
    "learning_rate": 3.205095334333285e-05,
    "step_time": 9.199656009674072,
    "tokens_per_second": 111.3085096794047,
    "memory_mb": 752.64453125
  },
  {
    "step": 2200,
    "loss": 6.471360206604004,
    "perplexity": 646.3623155888199,
    "learning_rate": 3.814149619382671e-05,
    "step_time": 8.654725313186646,
    "tokens_per_second": 118.3168688715975,
    "memory_mb": 735.9375
  },
  {
    "step": 2300,
    "loss": 6.382450699806213,
    "perplexity": 591.3752163574818,
    "learning_rate": 4.808657048893273e-05,
    "step_time": 8.75070834159851,
    "tokens_per_second": 117.01909834340836,
    "memory_mb": 776.53125
  },
  {
    "step": 2400,
    "loss": 5.957324385643005,
    "perplexity": 386.5744152212925,
    "learning_rate": 6.158400017864459e-05,
    "step_time": 8.642782926559448,
    "tokens_per_second": 118.4803562349376,
    "memory_mb": 923.15234375
  },
  {
    "step": 2500,
    "loss": 6.218142509460449,
    "perplexity": 501.7703322170689,
    "learning_rate": 7.822367269186924e-05,
    "step_time": 9.567925691604614,
    "tokens_per_second": 107.0242425585004,
    "memory_mb": 1050.1875
  },
  {
    "step": 2600,
    "loss": 6.314482092857361,
    "perplexity": 552.515834581567,
    "learning_rate": 9.749999999937299e-05,
    "step_time": 8.212730169296265,
    "tokens_per_second": 124.68448115199003,
    "memory_mb": 738.69921875
  },
  {
    "step": 2700,
    "loss": 6.240906238555908,
    "perplexity": 513.3234937600264,
    "learning_rate": 0.00011882728065020969,
    "step_time": 8.88506555557251,
    "tokens_per_second": 115.24957172181702,
    "memory_mb": 739.125
  },
  {
    "step": 2800,
    "loss": 6.308051347732544,
    "perplexity": 548.9741461252174,
    "learning_rate": 0.0001415574960139285,
    "step_time": 9.43014669418335,
    "tokens_per_second": 108.58791842884247,
    "memory_mb": 732.40234375
  },
  {
    "step": 2900,
    "loss": 6.410398960113525,
    "perplexity": 608.136254778883,
    "learning_rate": 0.00016499999999874617,
    "step_time": 8.674461603164673,
    "tokens_per_second": 118.0476722182294,
    "memory_mb": 733.71484375
  },
  {
    "step": 3000,
    "loss": 6.160744071006775,
    "perplexity": 473.7804700263767,
    "learning_rate": 0.0001884425039835638,
    "step_time": 9.056138277053833,
    "tokens_per_second": 113.07247843096432,
    "memory_mb": 803.75390625
  },
  {
    "step": 3100,
    "loss": 6.270610332489014,
    "perplexity": 528.8000232415734,
    "learning_rate": 0.0002111727193472824,
    "step_time": 9.650378227233887,
    "tokens_per_second": 106.10983071214939,
    "memory_mb": 790.328125
  },
  {
    "step": 3200,
    "loss": 6.4413875341415405,
    "perplexity": 627.2765639068664,
    "learning_rate": 0.00023249999999811922,
    "step_time": 8.436410665512085,
    "tokens_per_second": 121.37863371043517,
    "memory_mb": 940.3671875
  },
  {
    "step": 3300,
    "loss": 6.163665413856506,
    "perplexity": 475.1665688640195,
    "learning_rate": 0.0002517763273056231,
    "step_time": 10.011188507080078,
    "tokens_per_second": 102.28555773131335,
    "memory_mb": 747.43359375
  },
  {
    "step": 3400,
    "loss": 6.067382216453552,
    "perplexity": 431.5494984476535,
    "learning_rate": 0.00026841599981884787,
    "step_time": 8.671327114105225,
    "tokens_per_second": 118.0903437876665,
    "memory_mb": 970.37109375
  },
  {
    "step": 3500,
    "loss": 6.14486300945282,
    "perplexity": 466.3157638357051,
    "learning_rate": 0.0002819134295085615,
    "step_time": 8.194751024246216,
    "tokens_per_second": 124.95803679333764,
    "memory_mb": 736.890625
  },
  {
    "step": 3600,
    "loss": 6.448354721069336,
    "perplexity": 631.6621769355031,
    "learning_rate": 0.00029185850380367053,
    "step_time": 8.76004934310913,
    "tokens_per_second": 116.89431872955184,
    "memory_mb": 781.22265625
  },
  {
    "step": 3700,
    "loss": 6.149544358253479,
    "perplexity": 468.5038682213576,
    "learning_rate": 0.00029794904665416755,
    "step_time": 11.995165824890137,
    "tokens_per_second": 85.36772354369505,
    "memory_mb": 836.3046875
  },
  {
    "step": 3800,
    "loss": 6.083824634552002,
    "perplexity": 438.70387214992155,
    "learning_rate": 0.0002999999999975028,
    "step_time": 8.194983720779419,
    "tokens_per_second": 124.95448861033346,
    "memory_mb": 766.9453125
  },
  {
    "step": 3900,
    "loss": 6.295181512832642,
    "perplexity": 541.954209109435,
    "learning_rate": 0.0002979490466541723,
    "step_time": 9.617033004760742,
    "tokens_per_second": 106.47774625428518,
    "memory_mb": 837.29296875
  },
  {
    "step": 4000,
    "loss": 5.752694249153137,
    "perplexity": 315.038309582537,
    "learning_rate": 0.0002918585038036804,
    "step_time": 10.153574228286743,
    "tokens_per_second": 100.85118569845567,
    "memory_mb": 729.46484375
  },
  {
    "step": 4100,
    "loss": 6.129071950912476,
    "perplexity": 459.00997915951336,
    "learning_rate": 0.00028191342950857645,
    "step_time": 16.923099994659424,
    "tokens_per_second": 121.01801683180422,
    "memory_mb": 866.20703125
  },
  {
    "step": 4200,
    "loss": 5.725140452384949,
    "perplexity": 306.4763075489217,
    "learning_rate": 0.0002684159998188649,
    "step_time": 16.08098530769348,
    "tokens_per_second": 127.35538033357905,
    "memory_mb": 856.5625
  },
  {
    "step": 4300,
    "loss": 5.917757749557495,
    "perplexity": 371.57760903139615,
    "learning_rate": 0.00025177632730563935,
    "step_time": 16.5029239654541,
    "tokens_per_second": 124.09922049493284,
    "memory_mb": 856.58203125
  },
  {
    "step": 4400,
    "loss": 5.828692674636841,
    "perplexity": 339.9140102647636,
    "learning_rate": 0.00023249999999813424,
    "step_time": 15.03031873703003,
    "tokens_per_second": 136.25792212605347,
    "memory_mb": 856.58203125
  },
  {
    "step": 4500,
    "loss": 5.768530249595642,
    "perplexity": 320.0669682230316,
    "learning_rate": 0.0002111727193472959,
    "step_time": 15.095525026321411,
    "tokens_per_second": 135.66934548013344,
    "memory_mb": 856.58203125
  },
  {
    "step": 4600,
    "loss": 5.824233174324036,
    "perplexity": 338.40153857021943,
    "learning_rate": 0.00018844250398357548,
    "step_time": 15.770112752914429,
    "tokens_per_second": 129.86590724416445,
    "memory_mb": 856.6796875
  },
  {
    "step": 4700,
    "loss": 5.592358708381653,
    "perplexity": 268.3678753241069,
    "learning_rate": 0.00016499999999875585,
    "step_time": 15.393234729766846,
    "tokens_per_second": 133.04546029169921,
    "memory_mb": 856.6015625
  },
  {
    "step": 4800,
    "loss": 5.626398682594299,
    "perplexity": 277.6603717835579,
    "learning_rate": 0.0001415574960139364,
    "step_time": 15.054797887802124,
    "tokens_per_second": 136.03636629750804,
    "memory_mb": 856.6796875
  },
  {
    "step": 4900,
    "loss": 5.718711733818054,
    "perplexity": 304.5123771619807,
    "learning_rate": 0.00011882728065021638,
    "step_time": 16.42769432067871,
    "tokens_per_second": 124.66752546168553,
    "memory_mb": 856.62890625
  },
  {
    "step": 5000,
    "loss": 5.441234588623047,
    "perplexity": 230.7268604519411,
    "learning_rate": 9.749999999937817e-05,
    "step_time": 15.495960474014282,
    "tokens_per_second": 132.16347598681364,
    "memory_mb": 856.71875
  },
  {
    "step": 5100,
    "loss": 5.574064016342163,
    "perplexity": 263.50280585799544,
    "learning_rate": 7.822367269187295e-05,
    "step_time": 15.050865411758423,
    "tokens_per_second": 136.07190975211358,
    "memory_mb": 856.67578125
  },
  {
    "step": 5200,
    "loss": 5.486729264259338,
    "perplexity": 241.4661419404157,
    "learning_rate": 6.158400017864684e-05,
    "step_time": 14.95394515991211,
    "tokens_per_second": 136.95382576968316,
    "memory_mb": 856.74609375
  },
  {
    "step": 5300,
    "loss": 5.643607139587402,
    "perplexity": 282.47982711304775,
    "learning_rate": 4.808657048893407e-05,
    "step_time": 14.74146294593811,
    "tokens_per_second": 138.927866759948,
    "memory_mb": 856.828125
  },
  {
    "step": 5400,
    "loss": 5.773634672164917,
    "perplexity": 321.7049020761921,
    "learning_rate": 3.8141496193827305e-05,
    "step_time": 14.770179033279419,
    "tokens_per_second": 138.6577640924697,
    "memory_mb": 856.73046875
  },
  {
    "step": 5500,
    "loss": 5.430219531059265,
    "perplexity": 228.19933676762415,
    "learning_rate": 3.2050953343333e-05,
    "step_time": 17.394481897354126,
    "tokens_per_second": 117.7384881070543,
    "memory_mb": 856.4453125
  },
  {
    "step": 5600,
    "loss": 5.905840158462524,
    "perplexity": 367.17558190769864,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 14.9873046875,
    "tokens_per_second": 136.64898677265916,
    "memory_mb": 856.71875
  },
  {
    "step": 5700,
    "loss": 5.61777663230896,
    "perplexity": 275.27666109950576,
    "learning_rate": 3.205095334333272e-05,
    "step_time": 16.493826627731323,
    "tokens_per_second": 124.16766868136386,
    "memory_mb": 856.71875
  },
  {
    "step": 5800,
    "loss": 5.823711514472961,
    "perplexity": 338.2250541104363,
    "learning_rate": 3.814149619382621e-05,
    "step_time": 15.767404556274414,
    "tokens_per_second": 129.88821290724272,
    "memory_mb": 856.7265625
  },
  {
    "step": 5900,
    "loss": 5.767146706581116,
    "perplexity": 319.6244479984378,
    "learning_rate": 4.808657048893161e-05,
    "step_time": 14.995726346969604,
    "tokens_per_second": 136.57224415900788,
    "memory_mb": 857.28125
  },
  {
    "step": 6000,
    "loss": 5.632080435752869,
    "perplexity": 279.242459738446,
    "learning_rate": 6.158400017864261e-05,
    "step_time": 15.532944679260254,
    "tokens_per_second": 131.84879250452173,
    "memory_mb": 857.25
  },
  {
    "step": 6100,
    "loss": 5.641765594482422,
    "perplexity": 281.9601064615597,
    "learning_rate": 7.822367269186639e-05,
    "step_time": 13.042147397994995,
    "tokens_per_second": 157.02935548135613,
    "memory_mb": 3038.3984375
  },
  {
    "step": 6200,
    "loss": 5.216195583343506,
    "perplexity": 184.23195401529154,
    "learning_rate": 9.749999999936928e-05,
    "step_time": 13.716249227523804,
    "tokens_per_second": 149.3119559165174,
    "memory_mb": 3027.03515625
  },
  {
    "step": 6300,
    "loss": 5.537067532539368,
    "perplexity": 253.93225847719265,
    "learning_rate": 0.00011882728065020474,
    "step_time": 13.56373405456543,
    "tokens_per_second": 150.99086960575298,
    "memory_mb": 3029.5546875
  },
  {
    "step": 6400,
    "loss": 5.462532162666321,
    "perplexity": 235.69348362718267,
    "learning_rate": 0.0001415574960139221,
    "step_time": 15.235360145568848,
    "tokens_per_second": 134.4241278468008,
    "memory_mb": 3029.3984375
  },
  {
    "step": 6500,
    "loss": 5.461255669593811,
    "perplexity": 235.39281446997174,
    "learning_rate": 0.00016499999999873858,
    "step_time": 14.625237941741943,
    "tokens_per_second": 140.03190978211683,
    "memory_mb": 3029.7421875
  },
  {
    "step": 6600,
    "loss": 5.543500542640686,
    "perplexity": 255.571072864111,
    "learning_rate": 0.00018844250398355512,
    "step_time": 13.293677568435669,
    "tokens_per_second": 154.0581971735755,
    "memory_mb": 3025.90234375
  },
  {
    "step": 6700,
    "loss": 5.326942205429077,
    "perplexity": 205.80769336539043,
    "learning_rate": 0.0002111727193472727,
    "step_time": 13.689483880996704,
    "tokens_per_second": 149.60388702768896,
    "memory_mb": 3032.14453125
  },
  {
    "step": 6800,
    "loss": 5.4400506019592285,
    "perplexity": 230.4538445816494,
    "learning_rate": 0.00023249999999810797,
    "step_time": 13.920767545700073,
    "tokens_per_second": 147.11832470994733,
    "memory_mb": 3033.22265625
  },
  {
    "step": 6900,
    "loss": 5.562488794326782,
    "perplexity": 260.47028727589134,
    "learning_rate": 0.0002517763273056109,
    "step_time": 13.11566162109375,
    "tokens_per_second": 156.1491946930247,
    "memory_mb": 3030.31640625
  },
  {
    "step": 7000,
    "loss": 5.337615728378296,
    "perplexity": 208.01615155187662,
    "learning_rate": 0.00026841599981883453,
    "step_time": 13.812754154205322,
    "tokens_per_second": 148.2687650222517,
    "memory_mb": 3030.01171875
  },
  {
    "step": 7100,
    "loss": 5.591769337654114,
    "perplexity": 268.20975375486825,
    "learning_rate": 4.0863750705554176e-05,
    "step_time": 22.385777473449707,
    "tokens_per_second": 91.48665943941405,
    "memory_mb": 778.03515625
  },
  {
    "step": 7200,
    "loss": 5.491156101226807,
    "perplexity": 242.5374426712762,
    "learning_rate": 3.860829246363548e-05,
    "step_time": 23.257835626602173,
    "tokens_per_second": 88.0563450907491,
    "memory_mb": 766.17578125
  },
  {
    "step": 7300,
    "loss": 5.590375304222107,
    "perplexity": 267.83612088021056,
    "learning_rate": 3.660737030015427e-05,
    "step_time": 22.15459370613098,
    "tokens_per_second": 92.4413251339944,
    "memory_mb": 765.515625
  },
  {
    "step": 7400,
    "loss": 5.408857345581055,
    "perplexity": 223.37619999638744,
    "learning_rate": 3.486501380605981e-05,
    "step_time": 22.82973575592041,
    "tokens_per_second": 89.70756481353028,
    "memory_mb": 762.5078125
  },
  {
    "step": 7500,
    "loss": 5.681139707565308,
    "perplexity": 293.2834969372769,
    "learning_rate": 3.338473185545381e-05,
    "step_time": 23.581167221069336,
    "tokens_per_second": 86.848966414612,
    "memory_mb": 762.328125
  },
  {
    "step": 7600,
    "loss": 5.239741921424866,
    "perplexity": 188.6214169770699,
    "learning_rate": 3.2169505539184994e-05,
    "step_time": 24.760926008224487,
    "tokens_per_second": 82.71096159003685,
    "memory_mb": 765.65625
  },
  {
    "step": 7700,
    "loss": 5.552804112434387,
    "perplexity": 257.95989121628156,
    "learning_rate": 3.122178216132881e-05,
    "step_time": 21.644299030303955,
    "tokens_per_second": 94.62075889510751,
    "memory_mb": 766.9921875
  },
  {
    "step": 7800,
    "loss": 5.3391993045806885,
    "perplexity": 208.34582193938436,
    "learning_rate": 3.054347031064272e-05,
    "step_time": 21.552024602890015,
    "tokens_per_second": 95.02587518971994,
    "memory_mb": 768.30078125
  },
  {
    "step": 7900,
    "loss": 5.07136607170105,
    "perplexity": 159.39191947680405,
    "learning_rate": 3.0135936016922528e-05,
    "step_time": 25.795836448669434,
    "tokens_per_second": 79.3926571861808,
    "memory_mb": 766.03125
  },
  {
    "step": 8000,
    "loss": 5.289915561676025,
    "perplexity": 198.32667833002583,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 20.64373207092285,
    "tokens_per_second": 99.20686787466366,
    "memory_mb": 770.73046875
  },
  {
    "step": 8100,
    "loss": 5.472690582275391,
    "perplexity": 238.09995923313426,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 20.021532773971558,
    "tokens_per_second": 102.28987076666009,
    "memory_mb": 782.06640625
  },
  {
    "step": 8200,
    "loss": 5.381044268608093,
    "perplexity": 217.24902334669338,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 19.268752336502075,
    "tokens_per_second": 106.28607209405757,
    "memory_mb": 772.1875
  },
  {
    "step": 8300,
    "loss": 5.509163498878479,
    "perplexity": 246.94447131857976,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 21.33850598335266,
    "tokens_per_second": 95.97672871745365,
    "memory_mb": 773.4140625
  },
  {
    "step": 8400,
    "loss": 5.361395001411438,
    "perplexity": 213.0219051308242,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 19.24502682685852,
    "tokens_per_second": 106.41710289235836,
    "memory_mb": 771.12890625
  },
  {
    "step": 8500,
    "loss": 5.601587295532227,
    "perplexity": 270.8559949054038,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 19.78052020072937,
    "tokens_per_second": 103.53620527757828,
    "memory_mb": 788.19140625
  },
  {
    "step": 8600,
    "loss": 5.177438259124756,
    "perplexity": 177.22821619925446,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 19.43180513381958,
    "tokens_per_second": 105.39422281646966,
    "memory_mb": 768.90625
  },
  {
    "step": 8700,
    "loss": 5.507565855979919,
    "perplexity": 246.5502572281615,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 20.27060604095459,
    "tokens_per_second": 101.03299308675011,
    "memory_mb": 770.671875
  },
  {
    "step": 8800,
    "loss": 5.28560483455658,
    "perplexity": 197.47358618400258,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 18.91090226173401,
    "tokens_per_second": 108.29731821649273,
    "memory_mb": 780.921875
  },
  {
    "step": 8900,
    "loss": 5.001006722450256,
    "perplexity": 148.56264519463585,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 18.08107566833496,
    "tokens_per_second": 113.26759743540164,
    "memory_mb": 782.58203125
  },
  {
    "step": 9000,
    "loss": 5.228554844856262,
    "perplexity": 186.5230539013194,
    "learning_rate": 2.9999999999999997e-05,
    "step_time": 19.203582286834717,
    "tokens_per_second": 106.64676878563616,
    "memory_mb": 775.125
  }
]