diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5634 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.402861860209136,
+  "eval_steps": 500,
+  "global_step": 40000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00550357732526142,
+      "grad_norm": 5.828019142150879,
+      "learning_rate": 6.47007042253521e-06,
+      "loss": 8.235,
+      "step": 50
+    },
+    {
+      "epoch": 0.01100715465052284,
+      "grad_norm": 15.358248710632324,
+      "learning_rate": 1.3072183098591547e-05,
+      "loss": 6.0857,
+      "step": 100
+    },
+    {
+      "epoch": 0.01651073197578426,
+      "grad_norm": 7.99273681640625,
+      "learning_rate": 1.9674295774647885e-05,
+      "loss": 4.4315,
+      "step": 150
+    },
+    {
+      "epoch": 0.02201430930104568,
+      "grad_norm": 2.8550803661346436,
+      "learning_rate": 2.6276408450704222e-05,
+      "loss": 3.681,
+      "step": 200
+    },
+    {
+      "epoch": 0.0275178866263071,
+      "grad_norm": 2.3769114017486572,
+      "learning_rate": 3.2878521126760565e-05,
+      "loss": 3.4729,
+      "step": 250
+    },
+    {
+      "epoch": 0.03302146395156852,
+      "grad_norm": 5.98811149597168,
+      "learning_rate": 3.94806338028169e-05,
+      "loss": 3.3829,
+      "step": 300
+    },
+    {
+      "epoch": 0.03852504127682994,
+      "grad_norm": 3.616163492202759,
+      "learning_rate": 4.608274647887324e-05,
+      "loss": 3.4119,
+      "step": 350
+    },
+    {
+      "epoch": 0.04402861860209136,
+      "grad_norm": 2.1746344566345215,
+      "learning_rate": 5.268485915492957e-05,
+      "loss": 3.3955,
+      "step": 400
+    },
+    {
+      "epoch": 0.04953219592735278,
+      "grad_norm": 1.3189276456832886,
+      "learning_rate": 5.928697183098591e-05,
+      "loss": 3.3177,
+      "step": 450
+    },
+    {
+      "epoch": 0.0550357732526142,
+      "grad_norm": 1.2459770441055298,
+      "learning_rate": 6.588908450704225e-05,
+      "loss": 3.3095,
+      "step": 500
+    },
+    {
+      "epoch": 0.06053935057787562,
+      "grad_norm": 1.2855902910232544,
+      "learning_rate": 7.249119718309858e-05,
+      "loss": 3.2437,
+      "step": 550
+    },
+    {
+      "epoch": 0.06604292790313704,
+      "grad_norm": 1.2871235609054565,
+      "learning_rate": 7.909330985915493e-05,
+      "loss": 3.2069,
+      "step": 600
+    },
+    {
+      "epoch": 0.07154650522839846,
+      "grad_norm": 1.0846847295761108,
+      "learning_rate": 8.569542253521127e-05,
+      "loss": 3.2512,
+      "step": 650
+    },
+    {
+      "epoch": 0.07705008255365987,
+      "grad_norm": 1.5505499839782715,
+      "learning_rate": 9.22975352112676e-05,
+      "loss": 3.1704,
+      "step": 700
+    },
+    {
+      "epoch": 0.0825536598789213,
+      "grad_norm": 1.178614616394043,
+      "learning_rate": 9.889964788732394e-05,
+      "loss": 3.1747,
+      "step": 750
+    },
+    {
+      "epoch": 0.08805723720418272,
+      "grad_norm": 1.1504727602005005,
+      "learning_rate": 0.00010550176056338028,
+      "loss": 3.1241,
+      "step": 800
+    },
+    {
+      "epoch": 0.09356081452944413,
+      "grad_norm": 1.02865469455719,
+      "learning_rate": 0.00011210387323943662,
+      "loss": 3.0481,
+      "step": 850
+    },
+    {
+      "epoch": 0.09906439185470556,
+      "grad_norm": 1.1366077661514282,
+      "learning_rate": 0.00011870598591549295,
+      "loss": 3.0201,
+      "step": 900
+    },
+    {
+      "epoch": 0.10456796917996698,
+      "grad_norm": 0.9753648042678833,
+      "learning_rate": 0.00012530809859154929,
+      "loss": 3.0145,
+      "step": 950
+    },
+    {
+      "epoch": 0.1100715465052284,
+      "grad_norm": 0.6859256625175476,
+      "learning_rate": 0.00013191021126760563,
+      "loss": 2.9671,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11557512383048982,
+      "grad_norm": 0.8368203043937683,
+      "learning_rate": 0.00013851232394366197,
+      "loss": 2.992,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12107870115575124,
+      "grad_norm": 1.1109174489974976,
+      "learning_rate": 0.0001451144366197183,
+      "loss": 2.967,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12658227848101267,
+      "grad_norm": 0.6072912812232971,
+      "learning_rate": 0.00015171654929577465,
+      "loss": 2.9501,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13208585580627408,
+      "grad_norm": 0.7659889459609985,
+      "learning_rate": 0.00015831866197183099,
+      "loss": 2.9026,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1375894331315355,
+      "grad_norm": 0.5841110348701477,
+      "learning_rate": 0.0001649207746478873,
+      "loss": 2.8972,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14309301045679693,
+      "grad_norm": 0.5997458100318909,
+      "learning_rate": 0.00017152288732394364,
+      "loss": 2.9189,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14859658778205834,
+      "grad_norm": 0.6082264184951782,
+      "learning_rate": 0.00017812499999999998,
+      "loss": 2.8877,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15410016510731975,
+      "grad_norm": 0.696685254573822,
+      "learning_rate": 0.00018472711267605632,
+      "loss": 2.8752,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15960374243258119,
+      "grad_norm": 0.6795832514762878,
+      "learning_rate": 0.00019132922535211266,
+      "loss": 2.851,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1651073197578426,
+      "grad_norm": 0.5588585734367371,
+      "learning_rate": 0.000197931338028169,
+      "loss": 2.8546,
+      "step": 1500
+    },
+    {
+      "epoch": 0.170610897083104,
+      "grad_norm": 0.5267760157585144,
+      "learning_rate": 0.00020453345070422534,
+      "loss": 2.8049,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17611447440836545,
+      "grad_norm": 0.591826856136322,
+      "learning_rate": 0.00021113556338028168,
+      "loss": 2.8071,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18161805173362686,
+      "grad_norm": 0.5463298559188843,
+      "learning_rate": 0.00021773767605633802,
+      "loss": 2.7667,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18712162905888827,
+      "grad_norm": 0.5745858550071716,
+      "learning_rate": 0.00022433978873239433,
+      "loss": 2.7968,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1926252063841497,
+      "grad_norm": 0.6239858865737915,
+      "learning_rate": 0.00023094190140845067,
+      "loss": 2.8206,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19812878370941112,
+      "grad_norm": 0.46496015787124634,
+      "learning_rate": 0.000237544014084507,
+      "loss": 2.7944,
+      "step": 1800
+    },
+    {
+      "epoch": 0.20363236103467253,
+      "grad_norm": 0.5664075016975403,
+      "learning_rate": 0.00024414612676056335,
+      "loss": 2.7837,
+      "step": 1850
+    },
+    {
+      "epoch": 0.20913593835993396,
+      "grad_norm": 0.537627100944519,
+      "learning_rate": 0.0002507482394366197,
+      "loss": 2.7179,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21463951568519538,
+      "grad_norm": 0.544585645198822,
+      "learning_rate": 0.00025735035211267603,
+      "loss": 2.7552,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2201430930104568,
+      "grad_norm": 0.5067969560623169,
+      "learning_rate": 0.0002639524647887324,
+      "loss": 2.7309,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22564667033571822,
+      "grad_norm": 0.462003618478775,
+      "learning_rate": 0.0002705545774647887,
+      "loss": 2.716,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23115024766097964,
+      "grad_norm": 0.44174736738204956,
+      "learning_rate": 0.00027715669014084505,
+      "loss": 2.7319,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23665382498624105,
+      "grad_norm": 0.5010894536972046,
+      "learning_rate": 0.0002837588028169014,
+      "loss": 2.7199,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24215740231150248,
+      "grad_norm": 0.44113022089004517,
+      "learning_rate": 0.00029036091549295773,
+      "loss": 2.7146,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2476609796367639,
+      "grad_norm": 0.4972345232963562,
+      "learning_rate": 0.0002969630281690141,
+      "loss": 2.6471,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.5614003539085388,
+      "learning_rate": 0.0002999997102213327,
+      "loss": 2.687,
+      "step": 2300
+    },
+    {
+      "epoch": 0.25866813428728674,
+      "grad_norm": 0.5223066806793213,
+      "learning_rate": 0.0002999976432184194,
+      "loss": 2.6279,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26417171161254815,
+      "grad_norm": 0.4381965100765228,
+      "learning_rate": 0.0002999935887402823,
+      "loss": 2.6889,
+      "step": 2400
+    },
+    {
+      "epoch": 0.26967528893780957,
+      "grad_norm": 0.3969985842704773,
+      "learning_rate": 0.00029998754684064345,
+      "loss": 2.6565,
+      "step": 2450
+    },
+    {
+      "epoch": 0.275178866263071,
+      "grad_norm": 0.44786953926086426,
+      "learning_rate": 0.00029997951759955823,
+      "loss": 2.6736,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28068244358833244,
+      "grad_norm": 0.3951723873615265,
+      "learning_rate": 0.0002999695011234145,
+      "loss": 2.6173,
+      "step": 2550
+    },
+    {
+      "epoch": 0.28618602091359385,
+      "grad_norm": 0.4258750081062317,
+      "learning_rate": 0.00029995749754493093,
+      "loss": 2.5992,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29168959823885526,
+      "grad_norm": 0.41892707347869873,
+      "learning_rate": 0.0002999435070231555,
+      "loss": 2.6194,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2971931755641167,
+      "grad_norm": 0.4064221978187561,
+      "learning_rate": 0.0002999275297434632,
+      "loss": 2.6265,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3026967528893781,
+      "grad_norm": 0.3920956254005432,
+      "learning_rate": 0.00029990956591755365,
+      "loss": 2.6035,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3082003302146395,
+      "grad_norm": 0.366318941116333,
+      "learning_rate": 0.0002998896157834484,
+      "loss": 2.6352,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31370390753990096,
+      "grad_norm": 0.40867865085601807,
+      "learning_rate": 0.0002998676796054875,
+      "loss": 2.5674,
+      "step": 2850
+    },
+    {
+      "epoch": 0.31920748486516237,
+      "grad_norm": 0.3685750365257263,
+      "learning_rate": 0.00029984375767432627,
+      "loss": 2.5748,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3247110621904238,
+      "grad_norm": 0.38134968280792236,
+      "learning_rate": 0.0002998178503069314,
+      "loss": 2.5703,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3302146395156852,
+      "grad_norm": 0.36195963621139526,
+      "learning_rate": 0.00029978995784657643,
+      "loss": 2.5367,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3357182168409466,
+      "grad_norm": 0.3696858584880829,
+      "learning_rate": 0.0002997600806628379,
+      "loss": 2.5693,
+      "step": 3050
+    },
+    {
+      "epoch": 0.341221794166208,
+      "grad_norm": 0.3404608964920044,
+      "learning_rate": 0.00029972821915158964,
+      "loss": 2.5454,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3467253714914695,
+      "grad_norm": 0.33220136165618896,
+      "learning_rate": 0.0002996943737349981,
+      "loss": 2.5675,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3522289488167309,
+      "grad_norm": 0.3744208812713623,
+      "learning_rate": 0.00029965854486151643,
+      "loss": 2.5453,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3577325261419923,
+      "grad_norm": 0.36920756101608276,
+      "learning_rate": 0.0002996207330058788,
+      "loss": 2.5314,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3632361034672537,
+      "grad_norm": 0.36301830410957336,
+      "learning_rate": 0.00029958093866909403,
+      "loss": 2.5361,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3687396807925151,
+      "grad_norm": 0.3135142922401428,
+      "learning_rate": 0.00029953916237843853,
+      "loss": 2.5222,
+      "step": 3350
+    },
+    {
+      "epoch": 0.37424325811777653,
+      "grad_norm": 0.3634016513824463,
+      "learning_rate": 0.00029949540468745015,
+      "loss": 2.5179,
+      "step": 3400
+    },
+    {
+      "epoch": 0.379746835443038,
+      "grad_norm": 0.370914101600647,
+      "learning_rate": 0.00029944966617592017,
+      "loss": 2.5077,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3852504127682994,
+      "grad_norm": 0.32538744807243347,
+      "learning_rate": 0.0002994019474498858,
+      "loss": 2.4873,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3907539900935608,
+      "grad_norm": 0.3077157139778137,
+      "learning_rate": 0.0002993522491416221,
+      "loss": 2.5149,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39625756741882223,
+      "grad_norm": 0.32142704725265503,
+      "learning_rate": 0.000299300571909634,
+      "loss": 2.4989,
+      "step": 3600
+    },
+    {
+      "epoch": 0.40176114474408364,
+      "grad_norm": 0.3485497832298279,
+      "learning_rate": 0.00029924691643864684,
+      "loss": 2.5114,
+      "step": 3650
+    },
+    {
+      "epoch": 0.40726472206934505,
+      "grad_norm": 0.34483280777931213,
+      "learning_rate": 0.0002991912834395981,
+      "loss": 2.4863,
+      "step": 3700
+    },
+    {
+      "epoch": 0.4127682993946065,
+      "grad_norm": 0.30801650881767273,
+      "learning_rate": 0.00029913367364962733,
+      "loss": 2.4935,
+      "step": 3750
+    },
+    {
+      "epoch": 0.41827187671986793,
+      "grad_norm": 0.3099140524864197,
+      "learning_rate": 0.00029907408783206674,
+      "loss": 2.5122,
+      "step": 3800
+    },
+    {
+      "epoch": 0.42377545404512934,
+      "grad_norm": 0.33073538541793823,
+      "learning_rate": 0.0002990125267764309,
+      "loss": 2.4613,
+      "step": 3850
+    },
+    {
+      "epoch": 0.42927903137039075,
+      "grad_norm": 0.3025052845478058,
+      "learning_rate": 0.00029894899129840653,
+      "loss": 2.4855,
+      "step": 3900
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.3200172483921051,
+      "learning_rate": 0.00029888348223984143,
+      "loss": 2.4542,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4402861860209136,
+      "grad_norm": 0.31750544905662537,
+      "learning_rate": 0.0002988160004687335,
+      "loss": 2.4714,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44578976334617504,
+      "grad_norm": 0.28732484579086304,
+      "learning_rate": 0.00029874654687921895,
+      "loss": 2.4595,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45129334067143645,
+      "grad_norm": 0.33312809467315674,
+      "learning_rate": 0.0002986751223915609,
+      "loss": 2.4797,
+      "step": 4100
+    },
+    {
+      "epoch": 0.45679691799669786,
+      "grad_norm": 0.3109307289123535,
+      "learning_rate": 0.00029860172795213695,
+      "loss": 2.4506,
+      "step": 4150
+    },
+    {
+      "epoch": 0.46230049532195927,
+      "grad_norm": 0.31127694249153137,
+      "learning_rate": 0.0002985263645334266,
+      "loss": 2.4544,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4678040726472207,
+      "grad_norm": 0.34132882952690125,
+      "learning_rate": 0.0002984490331339982,
+      "loss": 2.455,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4733076499724821,
+      "grad_norm": 0.3107958436012268,
+      "learning_rate": 0.00029836973477849634,
+      "loss": 2.4733,
+      "step": 4300
+    },
+    {
+      "epoch": 0.47881122729774356,
+      "grad_norm": 0.27321889996528625,
+      "learning_rate": 0.00029828847051762753,
+      "loss": 2.4381,
+      "step": 4350
+    },
+    {
+      "epoch": 0.48431480462300497,
+      "grad_norm": 0.3114171326160431,
+      "learning_rate": 0.0002982052414281467,
+      "loss": 2.4612,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4898183819482664,
+      "grad_norm": 0.2926501929759979,
+      "learning_rate": 0.00029812004861284294,
+      "loss": 2.4364,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4953219592735278,
+      "grad_norm": 0.2824021279811859,
+      "learning_rate": 0.00029803289320052466,
+      "loss": 2.4453,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5008255365987893,
+      "grad_norm": 0.2710123062133789,
+      "learning_rate": 0.00029794377634600485,
+      "loss": 2.4207,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.28232479095458984,
+      "learning_rate": 0.0002978526992300856,
+      "loss": 2.4332,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5118326912493121,
+      "grad_norm": 0.2661692798137665,
+      "learning_rate": 0.0002977596630595427,
+      "loss": 2.4469,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5173362685745735,
+      "grad_norm": 0.317094087600708,
+      "learning_rate": 0.0002976646690671094,
+      "loss": 2.4486,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5228398458998349,
+      "grad_norm": 0.268915593624115,
+      "learning_rate": 0.0002975677185114602,
+      "loss": 2.4447,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5283434232250963,
+      "grad_norm": 0.2714874744415283,
+      "learning_rate": 0.000297468812677194,
+      "loss": 2.4065,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5338470005503577,
+      "grad_norm": 0.2872399389743805,
+      "learning_rate": 0.0002973679528748175,
+      "loss": 2.3943,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5393505778756191,
+      "grad_norm": 0.24350598454475403,
+      "learning_rate": 0.00029726514044072736,
+      "loss": 2.3888,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5448541552008805,
+      "grad_norm": 0.2634688913822174,
+      "learning_rate": 0.00029716037673719275,
+      "loss": 2.3953,
+      "step": 4950
+    },
+    {
+      "epoch": 0.550357732526142,
+      "grad_norm": 0.2842876613140106,
+      "learning_rate": 0.0002970536631523373,
+      "loss": 2.403,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5558613098514034,
+      "grad_norm": 0.30130457878112793,
+      "learning_rate": 0.00029694500110012055,
+      "loss": 2.4393,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5613648871766649,
+      "grad_norm": 0.2594560980796814,
+      "learning_rate": 0.00029683439202031936,
+      "loss": 2.3827,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5668684645019263,
+      "grad_norm": 0.25417500734329224,
+      "learning_rate": 0.0002967218373785088,
+      "loss": 2.3789,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5723720418271877,
+      "grad_norm": 0.2533874213695526,
+      "learning_rate": 0.0002966073386660428,
+      "loss": 2.4211,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5778756191524491,
+      "grad_norm": 0.2343342900276184,
+      "learning_rate": 0.0002964908974000341,
+      "loss": 2.4332,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5833791964777105,
+      "grad_norm": 0.25453534722328186,
+      "learning_rate": 0.0002963725151233345,
+      "loss": 2.3984,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5888827738029719,
+      "grad_norm": 0.2670257091522217,
+      "learning_rate": 0.00029625219340451435,
+      "loss": 2.4126,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5943863511282333,
+      "grad_norm": 0.2728422284126282,
+      "learning_rate": 0.00029612993383784154,
+      "loss": 2.4152,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5998899284534948,
+      "grad_norm": 0.27964428067207336,
+      "learning_rate": 0.0002960057380432606,
+      "loss": 2.3675,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6053935057787562,
+      "grad_norm": 0.27354755997657776,
+      "learning_rate": 0.00029587960766637103,
+      "loss": 2.4109,
+      "step": 5500
+    },
+    {
+      "epoch": 0.6108970831040176,
+      "grad_norm": 0.26175597310066223,
+      "learning_rate": 0.000295751544378406,
+      "loss": 2.3953,
+      "step": 5550
+    },
+    {
+      "epoch": 0.616400660429279,
+      "grad_norm": 0.2590219974517822,
+      "learning_rate": 0.0002956215498762093,
+      "loss": 2.3699,
+      "step": 5600
+    },
+    {
+      "epoch": 0.6219042377545404,
+      "grad_norm": 0.2495882660150528,
+      "learning_rate": 0.0002954896258822139,
+      "loss": 2.3949,
+      "step": 5650
+    },
+    {
+      "epoch": 0.6274078150798019,
+      "grad_norm": 0.26052239537239075,
+      "learning_rate": 0.0002953557741444183,
+      "loss": 2.3642,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6329113924050633,
+      "grad_norm": 0.2627600431442261,
+      "learning_rate": 0.0002952199964363638,
+      "loss": 2.3895,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6384149697303247,
+      "grad_norm": 0.2417898327112198,
+      "learning_rate": 0.00029508229455711086,
+      "loss": 2.4087,
+      "step": 5800
+    },
+    {
+      "epoch": 0.6439185470555862,
+      "grad_norm": 0.24891215562820435,
+      "learning_rate": 0.00029494267033121525,
+      "loss": 2.3642,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6494221243808476,
+      "grad_norm": 0.2453078180551529,
+      "learning_rate": 0.0002948011256087041,
+      "loss": 2.3761,
+      "step": 5900
+    },
+    {
+      "epoch": 0.654925701706109,
+      "grad_norm": 0.24607980251312256,
+      "learning_rate": 0.0002946576622650509,
+      "loss": 2.3596,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6604292790313704,
+      "grad_norm": 0.2540590763092041,
+      "learning_rate": 0.0002945122822011513,
+      "loss": 2.3666,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6659328563566318,
+      "grad_norm": 0.25918814539909363,
+      "learning_rate": 0.00029436498734329727,
+      "loss": 2.3494,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6714364336818932,
+      "grad_norm": 0.27554330229759216,
+      "learning_rate": 0.0002942157796431521,
+      "loss": 2.3793,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6769400110071546,
+      "grad_norm": 0.22879928350448608,
+      "learning_rate": 0.00029406466107772416,
+      "loss": 2.3898,
+      "step": 6150
+    },
+    {
+      "epoch": 0.682443588332416,
+      "grad_norm": 0.2851306200027466,
+      "learning_rate": 0.00029391163364934095,
+      "loss": 2.3493,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6879471656576774,
+      "grad_norm": 0.255014568567276,
+      "learning_rate": 0.0002937566993856225,
+      "loss": 2.3959,
+      "step": 6250
+    },
+    {
+      "epoch": 0.693450742982939,
+      "grad_norm": 0.23992781341075897,
+      "learning_rate": 0.00029359986033945454,
+      "loss": 2.3588,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6989543203082004,
+      "grad_norm": 0.24902793765068054,
+      "learning_rate": 0.000293441118588961,
+      "loss": 2.3908,
+      "step": 6350
+    },
+    {
+      "epoch": 0.7044578976334618,
+      "grad_norm": 0.2294873744249344,
+      "learning_rate": 0.0002932804762374771,
+      "loss": 2.398,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7099614749587232,
+      "grad_norm": 0.23560309410095215,
+      "learning_rate": 0.00029311793541352075,
+      "loss": 2.3481,
+      "step": 6450
+    },
+    {
+      "epoch": 0.7154650522839846,
+      "grad_norm": 0.27493488788604736,
+      "learning_rate": 0.00029295349827076497,
+      "loss": 2.3248,
+      "step": 6500
+    },
+    {
+      "epoch": 0.720968629609246,
+      "grad_norm": 0.22731854021549225,
+      "learning_rate": 0.0002927871669880089,
+      "loss": 2.3444,
+      "step": 6550
+    },
+    {
+      "epoch": 0.7264722069345074,
+      "grad_norm": 0.22198539972305298,
+      "learning_rate": 0.0002926189437691492,
+      "loss": 2.3148,
+      "step": 6600
+    },
+    {
+      "epoch": 0.7319757842597688,
+      "grad_norm": 0.25049686431884766,
+      "learning_rate": 0.0002924488308431508,
+      "loss": 2.3134,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7374793615850302,
+      "grad_norm": 0.2354484349489212,
+      "learning_rate": 0.0002922768304640172,
+      "loss": 2.348,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7429829389102917,
+      "grad_norm": 0.23955915868282318,
+      "learning_rate": 0.00029210294491076094,
+      "loss": 2.3172,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7484865162355531,
+      "grad_norm": 0.2476130574941635,
+      "learning_rate": 0.000291927176487373,
+      "loss": 2.3842,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7539900935608145,
+      "grad_norm": 0.2247840166091919,
+      "learning_rate": 0.00029174952752279254,
+      "loss": 2.3302,
+      "step": 6850
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 0.22530816495418549,
+      "learning_rate": 0.000291570000370876,
+      "loss": 2.3707,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7649972482113374,
+      "grad_norm": 0.2242722362279892,
+      "learning_rate": 0.00029138859741036587,
+      "loss": 2.312,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7705008255365988,
+      "grad_norm": 0.23376357555389404,
+      "learning_rate": 0.0002912053210448592,
+      "loss": 2.3334,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7760044028618602,
+      "grad_norm": 0.23131632804870605,
+      "learning_rate": 0.0002910201737027757,
+      "loss": 2.3293,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7815079801871216,
+      "grad_norm": 0.21440596878528595,
+      "learning_rate": 0.0002908331578373256,
+      "loss": 2.3593,
+      "step": 7100
+    },
+    {
+      "epoch": 0.787011557512383,
+      "grad_norm": 0.225584477186203,
+      "learning_rate": 0.00029064427592647715,
+      "loss": 2.3464,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7925151348376445,
+      "grad_norm": 0.22819140553474426,
+      "learning_rate": 0.0002904535304729238,
+      "loss": 2.3145,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7980187121629059,
+      "grad_norm": 0.2503032982349396,
+      "learning_rate": 0.00029026092400405115,
+      "loss": 2.3376,
+      "step": 7250
+    },
+    {
+      "epoch": 0.8035222894881673,
+      "grad_norm": 0.2061757743358612,
+      "learning_rate": 0.00029006645907190295,
+      "loss": 2.3252,
+      "step": 7300
+    },
+    {
+      "epoch": 0.8090258668134287,
+      "grad_norm": 0.2099982351064682,
+      "learning_rate": 0.00028987013825314804,
+      "loss": 2.35,
+      "step": 7350
+    },
+    {
+      "epoch": 0.8145294441386901,
+      "grad_norm": 0.212050199508667,
+      "learning_rate": 0.00028967196414904573,
+      "loss": 2.3304,
+      "step": 7400
+    },
+    {
+      "epoch": 0.8200330214639515,
+      "grad_norm": 0.22283074259757996,
+      "learning_rate": 0.0002894719393854114,
+      "loss": 2.3389,
+      "step": 7450
+    },
+    {
+      "epoch": 0.825536598789213,
+      "grad_norm": 0.2060365378856659,
+      "learning_rate": 0.0002892700666125817,
+      "loss": 2.3215,
+      "step": 7500
+    },
+    {
+      "epoch": 0.8310401761144744,
+      "grad_norm": 0.22510799765586853,
+      "learning_rate": 0.0002890663485053797,
+      "loss": 2.3295,
+      "step": 7550
+    },
+    {
+      "epoch": 0.8365437534397359,
+      "grad_norm": 0.21860332787036896,
+      "learning_rate": 0.00028886078776307905,
+      "loss": 2.3286,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8420473307649973,
+      "grad_norm": 0.20771068334579468,
+      "learning_rate": 0.00028865338710936826,
+      "loss": 2.3143,
+      "step": 7650
+    },
+    {
+      "epoch": 0.8475509080902587,
+      "grad_norm": 0.23098550736904144,
+      "learning_rate": 0.0002884441492923151,
+      "loss": 2.3583,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8530544854155201,
+      "grad_norm": 0.204328715801239,
+      "learning_rate": 0.00028823307708432963,
+      "loss": 2.3425,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8585580627407815,
+      "grad_norm": 0.20610789954662323,
+      "learning_rate": 0.0002880201732821275,
+      "loss": 2.3269,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8640616400660429,
+      "grad_norm": 0.22584037482738495,
+      "learning_rate": 0.0002878054407066935,
+      "loss": 2.3061,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.22317476570606232,
+      "learning_rate": 0.0002875888822032433,
+      "loss": 2.3047,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8750687947165657,
+      "grad_norm": 0.2132745087146759,
+      "learning_rate": 0.00028737050064118645,
+      "loss": 2.2925,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8805723720418271,
+      "grad_norm": 0.21960894763469696,
+      "learning_rate": 0.00028715029891408805,
+      "loss": 2.3365,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8860759493670886,
+      "grad_norm": 0.2113264799118042,
+      "learning_rate": 0.0002869282799396305,
+      "loss": 2.2814,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8915795266923501,
+      "grad_norm": 0.19959582388401031,
+      "learning_rate": 0.00028670444665957465,
+      "loss": 2.2787,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8970831040176115,
+      "grad_norm": 0.24770890176296234,
+      "learning_rate": 0.00028647880203972115,
+      "loss": 2.3312,
+      "step": 8150
+    },
+    {
+      "epoch": 0.9025866813428729,
+      "grad_norm": 0.2229924350976944,
+      "learning_rate": 0.0002862513490698709,
+      "loss": 2.2825,
+      "step": 8200
+    },
+    {
+      "epoch": 0.9080902586681343,
+      "grad_norm": 0.23118580877780914,
+      "learning_rate": 0.0002860220907637856,
+      "loss": 2.2973,
+      "step": 8250
+    },
+    {
+      "epoch": 0.9135938359933957,
+      "grad_norm": 0.1978590488433838,
+      "learning_rate": 0.0002857910301591475,
+      "loss": 2.3106,
+      "step": 8300
+    },
+    {
+      "epoch": 0.9190974133186571,
+      "grad_norm": 0.20972274243831635,
+      "learning_rate": 0.0002855581703175198,
+      "loss": 2.3489,
+      "step": 8350
+    },
+    {
+      "epoch": 0.9246009906439185,
+      "grad_norm": 0.20369485020637512,
+      "learning_rate": 0.0002853235143243052,
+      "loss": 2.3135,
+      "step": 8400
+    },
+    {
+      "epoch": 0.93010456796918,
+      "grad_norm": 0.19085292518138885,
+      "learning_rate": 0.00028508706528870576,
+      "loss": 2.2969,
+      "step": 8450
+    },
+    {
+      "epoch": 0.9356081452944414,
+      "grad_norm": 0.21080902218818665,
+      "learning_rate": 0.0002848488263436814,
+      "loss": 2.2979,
+      "step": 8500
+    },
+    {
+      "epoch": 0.9411117226197028,
+      "grad_norm": 0.1939728856086731,
+      "learning_rate": 0.00028460880064590835,
+      "loss": 2.31,
+      "step": 8550
+    },
+    {
+      "epoch": 0.9466152999449642,
+      "grad_norm": 0.23909969627857208,
+      "learning_rate": 0.0002843669913757375,
+      "loss": 2.2585,
+      "step": 8600
+    },
+    {
+      "epoch": 0.9521188772702256,
+      "grad_norm": 0.22809971868991852,
+      "learning_rate": 0.000284123401737152,
+      "loss": 2.3058,
+      "step": 8650
+    },
+    {
+      "epoch": 0.9576224545954871,
+      "grad_norm": 0.2066497504711151,
+      "learning_rate": 0.00028387803495772513,
+      "loss": 2.2736,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9631260319207485,
+      "grad_norm": 0.1921154260635376,
+      "learning_rate": 0.00028363089428857716,
+      "loss": 2.2813,
+      "step": 8750
+    },
+    {
+      "epoch": 0.9686296092460099,
+      "grad_norm": 0.19423851370811462,
+      "learning_rate": 0.00028338198300433264,
+      "loss": 2.314,
+      "step": 8800
+    },
+    {
+      "epoch": 0.9741331865712713,
+      "grad_norm": 0.224751815199852,
+      "learning_rate": 0.0002831313044030768,
+      "loss": 2.3335,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9796367638965328,
+      "grad_norm": 0.21634523570537567,
+      "learning_rate": 0.00028287886180631183,
+      "loss": 2.3021,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9851403412217942,
+      "grad_norm": 0.19320085644721985,
+      "learning_rate": 0.00028262465855891306,
+      "loss": 2.2635,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9906439185470556,
+      "grad_norm": 0.20169596374034882,
+      "learning_rate": 0.00028236869802908453,
+      "loss": 2.3013,
+      "step": 9000
+    },
+    {
+      "epoch": 0.996147495872317,
+      "grad_norm": 0.2045382857322693,
+      "learning_rate": 0.00028211098360831407,
+      "loss": 2.2576,
+      "step": 9050
+    },
+    {
+      "epoch": 1.0016510731975785,
+      "grad_norm": 0.2171631157398224,
+      "learning_rate": 0.00028185151871132894,
+      "loss": 2.3136,
+      "step": 9100
+    },
+    {
+      "epoch": 1.00715465052284,
+      "grad_norm": 0.19541560113430023,
+      "learning_rate": 0.00028159030677605017,
+      "loss": 2.2547,
+      "step": 9150
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.19888882339000702,
+      "learning_rate": 0.0002813273512635472,
+      "loss": 2.2275,
+      "step": 9200
+    },
+    {
+      "epoch": 1.0181618051733627,
+      "grad_norm": 0.24578270316123962,
+      "learning_rate": 0.0002810626556579918,
+      "loss": 2.2775,
+      "step": 9250
+    },
+    {
+      "epoch": 1.0236653824986242,
+      "grad_norm": 0.19902436435222626,
+      "learning_rate": 0.00028079622346661217,
+      "loss": 2.2353,
+      "step": 9300
+    },
+    {
+      "epoch": 1.0291689598238856,
+      "grad_norm": 0.19234856963157654,
+      "learning_rate": 0.00028052805821964633,
+      "loss": 2.2769,
+      "step": 9350
+    },
+    {
+      "epoch": 1.034672537149147,
+      "grad_norm": 0.19912928342819214,
+      "learning_rate": 0.0002802581634702952,
+      "loss": 2.2625,
+      "step": 9400
+    },
+    {
+      "epoch": 1.0401761144744084,
+      "grad_norm": 0.1918436884880066,
+      "learning_rate": 0.00027998654279467604,
+      "loss": 2.2335,
+      "step": 9450
+    },
+    {
+      "epoch": 1.0456796917996698,
+      "grad_norm": 0.20422010123729706,
+      "learning_rate": 0.0002797131997917743,
+      "loss": 2.3022,
+      "step": 9500
+    },
+    {
+      "epoch": 1.0511832691249312,
+      "grad_norm": 0.19621697068214417,
+      "learning_rate": 0.00027943813808339657,
+      "loss": 2.2378,
+      "step": 9550
+    },
+    {
+      "epoch": 1.0566868464501926,
+      "grad_norm": 0.18403683602809906,
+      "learning_rate": 0.0002791613613141223,
+      "loss": 2.2502,
+      "step": 9600
+    },
+    {
+      "epoch": 1.062190423775454,
+      "grad_norm": 0.2140798568725586,
+      "learning_rate": 0.0002788828731512556,
+      "loss": 2.2722,
+      "step": 9650
+    },
+    {
+      "epoch": 1.0676940011007154,
+      "grad_norm": 0.19459135830402374,
+      "learning_rate": 0.0002786026772847767,
+      "loss": 2.2361,
+      "step": 9700
+    },
+    {
+      "epoch": 1.0731975784259769,
+      "grad_norm": 0.20383061468601227,
+      "learning_rate": 0.00027832077742729277,
+      "loss": 2.2471,
+      "step": 9750
+    },
+    {
+      "epoch": 1.0787011557512383,
+      "grad_norm": 0.1901775300502777,
+      "learning_rate": 0.0002780371773139891,
+      "loss": 2.2968,
+      "step": 9800
+    },
+    {
+      "epoch": 1.0842047330764997,
+      "grad_norm": 0.20100219547748566,
+      "learning_rate": 0.0002777518807025793,
+      "loss": 2.2568,
+      "step": 9850
+    },
+    {
+      "epoch": 1.089708310401761,
+      "grad_norm": 0.2127520740032196,
+      "learning_rate": 0.00027746489137325586,
+      "loss": 2.28,
+      "step": 9900
+    },
+    {
+      "epoch": 1.0952118877270225,
+      "grad_norm": 0.19059176743030548,
+      "learning_rate": 0.00027717621312863965,
+      "loss": 2.284,
+      "step": 9950
+    },
+    {
+      "epoch": 1.100715465052284,
+      "grad_norm": 0.20845113694667816,
+      "learning_rate": 0.0002768858497937299,
+      "loss": 2.2409,
+      "step": 10000
+    },
+    {
+      "epoch": 1.1062190423775453,
+      "grad_norm": 0.18969906866550446,
+      "learning_rate": 0.00027659380521585325,
+      "loss": 2.2725,
+      "step": 10050
+    },
+    {
+      "epoch": 1.1117226197028067,
+      "grad_norm": 0.19071218371391296,
+      "learning_rate": 0.0002763000832646129,
+      "loss": 2.2624,
+      "step": 10100
+    },
+    {
+      "epoch": 1.1172261970280681,
+      "grad_norm": 0.21723783016204834,
+      "learning_rate": 0.0002760046878318373,
+      "loss": 2.2453,
+      "step": 10150
+    },
+    {
+      "epoch": 1.1227297743533298,
+      "grad_norm": 0.19336852431297302,
+      "learning_rate": 0.00027570762283152874,
+      "loss": 2.2322,
+      "step": 10200
+    },
+    {
+      "epoch": 1.1282333516785912,
+      "grad_norm": 0.19500704109668732,
+      "learning_rate": 0.0002754088921998112,
+      "loss": 2.1858,
+      "step": 10250
+    },
+    {
+      "epoch": 1.1337369290038526,
+      "grad_norm": 0.19920021295547485,
+      "learning_rate": 0.0002751084998948784,
+      "loss": 2.2286,
+      "step": 10300
+    },
+    {
+      "epoch": 1.139240506329114,
+      "grad_norm": 0.18863485753536224,
+      "learning_rate": 0.00027480644989694126,
+      "loss": 2.2623,
+      "step": 10350
+    },
+    {
+      "epoch": 1.1447440836543754,
+      "grad_norm": 0.18571630120277405,
+      "learning_rate": 0.0002745027462081753,
+      "loss": 2.2616,
+      "step": 10400
+    },
+    {
+      "epoch": 1.1502476609796368,
+      "grad_norm": 0.187313050031662,
+      "learning_rate": 0.00027419739285266745,
+      "loss": 2.2608,
+      "step": 10450
+    },
+    {
+      "epoch": 1.1557512383048982,
+      "grad_norm": 0.19139795005321503,
+      "learning_rate": 0.0002738903938763628,
+      "loss": 2.2204,
+      "step": 10500
+    },
+    {
+      "epoch": 1.1612548156301596,
+      "grad_norm": 0.188013955950737,
+      "learning_rate": 0.0002735817533470109,
+      "loss": 2.2596,
+      "step": 10550
+    },
+    {
+      "epoch": 1.166758392955421,
+      "grad_norm": 0.1884177178144455,
+      "learning_rate": 0.0002732714753541122,
+      "loss": 2.2618,
+      "step": 10600
+    },
+    {
+      "epoch": 1.1722619702806825,
+      "grad_norm": 0.19187124073505402,
+      "learning_rate": 0.00027295956400886335,
+      "loss": 2.2075,
+      "step": 10650
+    },
+    {
+      "epoch": 1.1777655476059439,
+      "grad_norm": 0.19815993309020996,
+      "learning_rate": 0.0002726460234441031,
+      "loss": 2.2677,
+      "step": 10700
+    },
+    {
+      "epoch": 1.1832691249312053,
+      "grad_norm": 0.18218094110488892,
+      "learning_rate": 0.00027233085781425744,
+      "loss": 2.2697,
+      "step": 10750
+    },
+    {
+      "epoch": 1.1887727022564667,
+      "grad_norm": 0.1923886239528656,
+      "learning_rate": 0.0002720140712952845,
+      "loss": 2.2127,
+      "step": 10800
+    },
+    {
+      "epoch": 1.194276279581728,
+      "grad_norm": 0.18907909095287323,
+      "learning_rate": 0.0002716956680846193,
+      "loss": 2.1989,
+      "step": 10850
+    },
+    {
+      "epoch": 1.1997798569069895,
+      "grad_norm": 0.1902381181716919,
+      "learning_rate": 0.000271375652401118,
+      "loss": 2.2211,
+      "step": 10900
+    },
+    {
+      "epoch": 1.205283434232251,
+      "grad_norm": 0.20184637606143951,
+      "learning_rate": 0.0002710540284850023,
+      "loss": 2.2451,
+      "step": 10950
+    },
+    {
+      "epoch": 1.2107870115575123,
+      "grad_norm": 0.19903656840324402,
+      "learning_rate": 0.00027073080059780285,
+      "loss": 2.2233,
+      "step": 11000
+    },
+    {
+      "epoch": 1.2162905888827737,
+      "grad_norm": 0.19023925065994263,
+      "learning_rate": 0.000270405973022303,
+      "loss": 2.2438,
+      "step": 11050
+    },
+    {
+      "epoch": 1.2217941662080352,
+      "grad_norm": 0.17722088098526,
+      "learning_rate": 0.0002700795500624822,
+      "loss": 2.259,
+      "step": 11100
+    },
+    {
+      "epoch": 1.2272977435332966,
+      "grad_norm": 0.19715473055839539,
+      "learning_rate": 0.0002697515360434587,
+      "loss": 2.2509,
+      "step": 11150
+    },
+    {
+      "epoch": 1.232801320858558,
+      "grad_norm": 0.18065761029720306,
+      "learning_rate": 0.00026942193531143225,
+      "loss": 2.2006,
+      "step": 11200
+    },
+    {
+      "epoch": 1.2383048981838194,
+      "grad_norm": 0.18757817149162292,
+      "learning_rate": 0.00026909075223362683,
+      "loss": 2.2215,
+      "step": 11250
+    },
+    {
+      "epoch": 1.243808475509081,
+      "grad_norm": 0.18816334009170532,
+      "learning_rate": 0.00026875799119823243,
+      "loss": 2.2314,
+      "step": 11300
+    },
+    {
+      "epoch": 1.2493120528343424,
+      "grad_norm": 0.19772210717201233,
+      "learning_rate": 0.0002684236566143471,
+      "loss": 2.2256,
+      "step": 11350
+    },
+    {
+      "epoch": 1.2548156301596038,
+      "grad_norm": 0.19510945677757263,
+      "learning_rate": 0.0002680877529119184,
+      "loss": 2.2427,
+      "step": 11400
+    },
+    {
+      "epoch": 1.2603192074848653,
+      "grad_norm": 0.19960781931877136,
+      "learning_rate": 0.0002677502845416849,
+      "loss": 2.2324,
+      "step": 11450
+    },
+    {
+      "epoch": 1.2658227848101267,
+      "grad_norm": 0.1882307380437851,
+      "learning_rate": 0.0002674112559751169,
+      "loss": 2.2432,
+      "step": 11500
+    },
+    {
+      "epoch": 1.271326362135388,
+      "grad_norm": 0.18959587812423706,
+      "learning_rate": 0.00026707067170435767,
+      "loss": 2.19,
+      "step": 11550
+    },
+    {
+      "epoch": 1.2768299394606495,
+      "grad_norm": 0.18768323957920074,
+      "learning_rate": 0.0002667285362421634,
+      "loss": 2.2294,
+      "step": 11600
+    },
+    {
+      "epoch": 1.282333516785911,
+      "grad_norm": 0.20077955722808838,
+      "learning_rate": 0.00026638485412184355,
+      "loss": 2.2518,
+      "step": 11650
+    },
+    {
+      "epoch": 1.2878370941111723,
+      "grad_norm": 0.1956523358821869,
+      "learning_rate": 0.00026603962989720105,
+      "loss": 2.2662,
+      "step": 11700
+    },
+    {
+      "epoch": 1.2933406714364337,
+      "grad_norm": 0.19226029515266418,
+      "learning_rate": 0.0002656928681424718,
+      "loss": 2.2133,
+      "step": 11750
+    },
+    {
+      "epoch": 1.2988442487616951,
+      "grad_norm": 0.16987648606300354,
+      "learning_rate": 0.0002653445734522639,
+      "loss": 2.2226,
+      "step": 11800
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.17524655163288116,
+      "learning_rate": 0.00026499475044149703,
+      "loss": 2.2555,
+      "step": 11850
+    },
+    {
+      "epoch": 1.309851403412218,
+      "grad_norm": 0.19188162684440613,
+      "learning_rate": 0.00026464340374534104,
+      "loss": 2.2234,
+      "step": 11900
+    },
+    {
+      "epoch": 1.3153549807374794,
+      "grad_norm": 0.17676417529582977,
+      "learning_rate": 0.00026429053801915487,
+      "loss": 2.2188,
+      "step": 11950
+    },
+    {
+      "epoch": 1.3208585580627408,
+      "grad_norm": 0.20071938633918762,
+      "learning_rate": 0.0002639361579384245,
+      "loss": 2.234,
+      "step": 12000
+    },
+    {
+      "epoch": 1.3263621353880022,
+      "grad_norm": 0.19267459213733673,
+      "learning_rate": 0.0002635802681987012,
+      "loss": 2.235,
+      "step": 12050
+    },
+    {
+      "epoch": 1.3318657127132636,
+      "grad_norm": 0.1724405139684677,
+      "learning_rate": 0.00026322287351553944,
+      "loss": 2.2388,
+      "step": 12100
+    },
+    {
+      "epoch": 1.337369290038525,
+      "grad_norm": 0.1907806396484375,
+      "learning_rate": 0.000262863978624434,
+      "loss": 2.1937,
+      "step": 12150
+    },
+    {
+      "epoch": 1.3428728673637864,
+      "grad_norm": 0.18452796339988708,
+      "learning_rate": 0.0002625035882807578,
+      "loss": 2.2429,
+      "step": 12200
+    },
+    {
+      "epoch": 1.3483764446890478,
+      "grad_norm": 0.17565739154815674,
+      "learning_rate": 0.0002621417072596982,
+      "loss": 2.2367,
+      "step": 12250
+    },
+    {
+      "epoch": 1.3538800220143092,
+      "grad_norm": 0.1761862337589264,
+      "learning_rate": 0.00026177834035619445,
+      "loss": 2.2258,
+      "step": 12300
+    },
+    {
+      "epoch": 1.3593835993395706,
+      "grad_norm": 0.18616576492786407,
+      "learning_rate": 0.0002614134923848736,
+      "loss": 2.2207,
+      "step": 12350
+    },
+    {
+      "epoch": 1.364887176664832,
+      "grad_norm": 0.18103408813476562,
+      "learning_rate": 0.00026104716817998696,
+      "loss": 2.2346,
+      "step": 12400
+    },
+    {
+      "epoch": 1.3703907539900935,
+      "grad_norm": 0.18511448800563812,
+      "learning_rate": 0.00026067937259534595,
+      "loss": 2.2363,
+      "step": 12450
+    },
+    {
+      "epoch": 1.3758943313153549,
+      "grad_norm": 0.1737717241048813,
+      "learning_rate": 0.00026031011050425796,
+      "loss": 2.2199,
+      "step": 12500
+    },
+    {
+      "epoch": 1.3813979086406163,
+      "grad_norm": 0.18761631846427917,
+      "learning_rate": 0.00025993938679946154,
+      "loss": 2.228,
+      "step": 12550
+    },
+    {
+      "epoch": 1.3869014859658777,
+      "grad_norm": 0.18815304338932037,
+      "learning_rate": 0.0002595672063930617,
+      "loss": 2.2034,
+      "step": 12600
+    },
+    {
+      "epoch": 1.3924050632911391,
+      "grad_norm": 0.1868327558040619,
+      "learning_rate": 0.00025919357421646486,
+      "loss": 2.197,
+      "step": 12650
+    },
+    {
+      "epoch": 1.3979086406164005,
+      "grad_norm": 0.17544154822826385,
+      "learning_rate": 0.00025881849522031345,
+      "loss": 2.1956,
+      "step": 12700
+    },
+    {
+      "epoch": 1.4034122179416622,
+      "grad_norm": 0.18338458240032196,
+      "learning_rate": 0.00025844197437442033,
+      "loss": 2.2291,
+      "step": 12750
+    },
+    {
+      "epoch": 1.4089157952669236,
+      "grad_norm": 0.17127631604671478,
+      "learning_rate": 0.00025806401666770277,
+      "loss": 2.2205,
+      "step": 12800
+    },
+    {
+      "epoch": 1.414419372592185,
+      "grad_norm": 0.17331157624721527,
+      "learning_rate": 0.00025768462710811673,
+      "loss": 2.2356,
+      "step": 12850
+    },
+    {
+      "epoch": 1.4199229499174464,
+      "grad_norm": 0.16098715364933014,
+      "learning_rate": 0.00025730381072259026,
+      "loss": 2.2065,
+      "step": 12900
+    },
+    {
+      "epoch": 1.4254265272427078,
+      "grad_norm": 0.17098484933376312,
+      "learning_rate": 0.00025692157255695667,
+      "loss": 2.2323,
+      "step": 12950
+    },
+    {
+      "epoch": 1.4309301045679692,
+      "grad_norm": 0.15763573348522186,
+      "learning_rate": 0.00025653791767588823,
+      "loss": 2.2064,
+      "step": 13000
+    },
+    {
+      "epoch": 1.4364336818932306,
+      "grad_norm": 0.17587807774543762,
+      "learning_rate": 0.0002561528511628286,
+      "loss": 2.2085,
+      "step": 13050
+    },
+    {
+      "epoch": 1.441937259218492,
+      "grad_norm": 0.17770230770111084,
+      "learning_rate": 0.00025576637811992555,
+      "loss": 2.2204,
+      "step": 13100
+    },
+    {
+      "epoch": 1.4474408365437534,
+      "grad_norm": 0.18236953020095825,
+      "learning_rate": 0.0002553785036679636,
+      "loss": 2.2388,
+      "step": 13150
+    },
+    {
+      "epoch": 1.4529444138690149,
+      "grad_norm": 0.16505366563796997,
+      "learning_rate": 0.00025498923294629594,
+      "loss": 2.2345,
+      "step": 13200
+    },
+    {
+      "epoch": 1.4584479911942763,
+      "grad_norm": 0.1687227189540863,
+      "learning_rate": 0.0002545985711127763,
+      "loss": 2.1652,
+      "step": 13250
+    },
+    {
+      "epoch": 1.4639515685195377,
+      "grad_norm": 0.17849218845367432,
+      "learning_rate": 0.00025420652334369085,
+      "loss": 2.2115,
+      "step": 13300
+    },
+    {
+      "epoch": 1.469455145844799,
+      "grad_norm": 0.17708005011081696,
+      "learning_rate": 0.0002538130948336894,
+      "loss": 2.2017,
+      "step": 13350
+    },
+    {
+      "epoch": 1.4749587231700605,
+      "grad_norm": 0.17772626876831055,
+      "learning_rate": 0.0002534182907957165,
+      "loss": 2.1926,
+      "step": 13400
+    },
+    {
+      "epoch": 1.480462300495322,
+      "grad_norm": 0.16764992475509644,
+      "learning_rate": 0.00025302211646094277,
+      "loss": 2.1986,
+      "step": 13450
+    },
+    {
+      "epoch": 1.4859658778205833,
+      "grad_norm": 0.19117778539657593,
+      "learning_rate": 0.00025262457707869506,
+      "loss": 2.2118,
+      "step": 13500
+    },
+    {
+      "epoch": 1.4914694551458447,
+      "grad_norm": 0.17632248997688293,
+      "learning_rate": 0.00025222567791638743,
+      "loss": 2.1897,
+      "step": 13550
+    },
+    {
+      "epoch": 1.4969730324711064,
+      "grad_norm": 0.19731369614601135,
+      "learning_rate": 0.0002518254242594508,
+      "loss": 2.1945,
+      "step": 13600
+    },
+    {
+      "epoch": 1.5024766097963678,
+      "grad_norm": 0.16187229752540588,
+      "learning_rate": 0.0002514238214112635,
+      "loss": 2.1959,
+      "step": 13650
+    },
+    {
+      "epoch": 1.5079801871216292,
+      "grad_norm": 0.1828337460756302,
+      "learning_rate": 0.00025102087469308036,
+      "loss": 2.1994,
+      "step": 13700
+    },
+    {
+      "epoch": 1.5134837644468906,
+      "grad_norm": 0.17188578844070435,
+      "learning_rate": 0.0002506165894439628,
+      "loss": 2.2087,
+      "step": 13750
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.17914508283138275,
+      "learning_rate": 0.00025021097102070786,
+      "loss": 2.2266,
+      "step": 13800
+    },
+    {
+      "epoch": 1.5244909190974134,
+      "grad_norm": 0.17482663691043854,
+      "learning_rate": 0.0002498040247977769,
+      "loss": 2.2431,
+      "step": 13850
+    },
+    {
+      "epoch": 1.5299944964226748,
+      "grad_norm": 0.17342698574066162,
+      "learning_rate": 0.00024939575616722505,
+      "loss": 2.2045,
+      "step": 13900
+    },
+    {
+      "epoch": 1.5354980737479362,
+      "grad_norm": 0.18258632719516754,
+      "learning_rate": 0.00024898617053862904,
+      "loss": 2.1843,
+      "step": 13950
+    },
+    {
+      "epoch": 1.5410016510731976,
+      "grad_norm": 0.16960634291172028,
+      "learning_rate": 0.00024857527333901625,
+      "loss": 2.2096,
+      "step": 14000
+    },
+    {
+      "epoch": 1.546505228398459,
+      "grad_norm": 0.17008773982524872,
+      "learning_rate": 0.00024816307001279214,
+      "loss": 2.1522,
+      "step": 14050
+    },
+    {
+      "epoch": 1.5520088057237205,
+      "grad_norm": 0.16970837116241455,
+      "learning_rate": 0.00024774956602166857,
+      "loss": 2.2055,
+      "step": 14100
+    },
+    {
+      "epoch": 1.5575123830489819,
+      "grad_norm": 0.17486636340618134,
+      "learning_rate": 0.0002473347668445912,
+      "loss": 2.1906,
+      "step": 14150
+    },
+    {
+      "epoch": 1.5630159603742433,
+      "grad_norm": 0.19200065732002258,
+      "learning_rate": 0.00024691867797766684,
+      "loss": 2.1839,
+      "step": 14200
+    },
+    {
+      "epoch": 1.5685195376995047,
+      "grad_norm": 0.19441814720630646,
+      "learning_rate": 0.00024650130493409104,
+      "loss": 2.2121,
+      "step": 14250
+    },
+    {
+      "epoch": 1.574023115024766,
+      "grad_norm": 0.16507557034492493,
+      "learning_rate": 0.0002460826532440745,
+      "loss": 2.2016,
+      "step": 14300
+    },
+    {
+      "epoch": 1.5795266923500275,
+      "grad_norm": 0.16732299327850342,
+      "learning_rate": 0.00024566272845477014,
+      "loss": 2.2362,
+      "step": 14350
+    },
+    {
+      "epoch": 1.585030269675289,
+      "grad_norm": 0.18136867880821228,
+      "learning_rate": 0.00024524153613019947,
+      "loss": 2.193,
+      "step": 14400
+    },
+    {
+      "epoch": 1.5905338470005503,
+      "grad_norm": 0.16409073770046234,
+      "learning_rate": 0.0002448190818511789,
+      "loss": 2.1589,
+      "step": 14450
+    },
+    {
+      "epoch": 1.5960374243258117,
+      "grad_norm": 0.17381897568702698,
+      "learning_rate": 0.00024439537121524583,
+      "loss": 2.1834,
+      "step": 14500
+    },
+    {
+      "epoch": 1.6015410016510732,
+      "grad_norm": 0.1628628373146057,
+      "learning_rate": 0.00024397040983658436,
+      "loss": 2.172,
+      "step": 14550
+    },
+    {
+      "epoch": 1.6070445789763346,
+      "grad_norm": 0.16793246567249298,
+      "learning_rate": 0.00024354420334595105,
+      "loss": 2.176,
+      "step": 14600
+    },
+    {
+      "epoch": 1.612548156301596,
+      "grad_norm": 0.17394863069057465,
+      "learning_rate": 0.0002431167573906002,
+      "loss": 2.199,
+      "step": 14650
+    },
+    {
+      "epoch": 1.6180517336268574,
+      "grad_norm": 0.1554850935935974,
+      "learning_rate": 0.00024268807763420914,
+      "loss": 2.2042,
+      "step": 14700
+    },
+    {
+      "epoch": 1.6235553109521188,
+      "grad_norm": 0.17003560066223145,
+      "learning_rate": 0.00024225816975680312,
+      "loss": 2.1717,
+      "step": 14750
+    },
+    {
+      "epoch": 1.6290588882773802,
+      "grad_norm": 0.17207174003124237,
+      "learning_rate": 0.0002418270394546799,
+      "loss": 2.1894,
+      "step": 14800
+    },
+    {
+      "epoch": 1.6345624656026416,
+      "grad_norm": 0.1507563292980194,
+      "learning_rate": 0.00024139469244033467,
+      "loss": 2.2101,
+      "step": 14850
+    },
+    {
+      "epoch": 1.640066042927903,
+      "grad_norm": 0.17063497006893158,
+      "learning_rate": 0.0002409611344423838,
+      "loss": 2.1854,
+      "step": 14900
+    },
+    {
+      "epoch": 1.6455696202531644,
+      "grad_norm": 0.17093567550182343,
+      "learning_rate": 0.00024052637120548954,
+      "loss": 2.2028,
+      "step": 14950
+    },
+    {
+      "epoch": 1.6510731975784259,
+      "grad_norm": 0.17256326973438263,
+      "learning_rate": 0.00024009040849028341,
+      "loss": 2.2102,
+      "step": 15000
+    },
+    {
+      "epoch": 1.6565767749036873,
+      "grad_norm": 0.17648595571517944,
+      "learning_rate": 0.00023965325207329016,
+      "loss": 2.2094,
+      "step": 15050
+    },
+    {
+      "epoch": 1.6620803522289487,
+      "grad_norm": 0.17140266299247742,
+      "learning_rate": 0.0002392149077468511,
+      "loss": 2.1936,
+      "step": 15100
+    },
+    {
+      "epoch": 1.66758392955421,
+      "grad_norm": 0.16277456283569336,
+      "learning_rate": 0.00023877538131904743,
+      "loss": 2.1705,
+      "step": 15150
+    },
+    {
+      "epoch": 1.6730875068794715,
+      "grad_norm": 0.16912780702114105,
+      "learning_rate": 0.00023833467861362323,
+      "loss": 2.1834,
+      "step": 15200
+    },
+    {
+      "epoch": 1.6785910842047331,
+      "grad_norm": 0.1684638261795044,
+      "learning_rate": 0.00023789280546990824,
+      "loss": 2.2044,
+      "step": 15250
+    },
+    {
+      "epoch": 1.6840946615299945,
+      "grad_norm": 0.16793768107891083,
+      "learning_rate": 0.00023744976774274065,
+      "loss": 2.188,
+      "step": 15300
+    },
+    {
+      "epoch": 1.689598238855256,
+      "grad_norm": 0.15743038058280945,
+      "learning_rate": 0.00023700557130238944,
+      "loss": 2.183,
+      "step": 15350
+    },
+    {
+      "epoch": 1.6951018161805174,
+      "grad_norm": 0.1788504421710968,
+      "learning_rate": 0.0002365602220344765,
+      "loss": 2.1923,
+      "step": 15400
+    },
+    {
+      "epoch": 1.7006053935057788,
+      "grad_norm": 0.1614784598350525,
+      "learning_rate": 0.00023611372583989883,
+      "loss": 2.196,
+      "step": 15450
+    },
+    {
+      "epoch": 1.7061089708310402,
+      "grad_norm": 0.16596226394176483,
+      "learning_rate": 0.0002356660886347502,
+      "loss": 2.1952,
+      "step": 15500
+    },
+    {
+      "epoch": 1.7116125481563016,
+      "grad_norm": 0.18096621334552765,
+      "learning_rate": 0.00023521731635024274,
+      "loss": 2.1858,
+      "step": 15550
+    },
+    {
+      "epoch": 1.717116125481563,
+      "grad_norm": 0.1885058581829071,
+      "learning_rate": 0.0002347674149326286,
+      "loss": 2.1894,
+      "step": 15600
+    },
+    {
+      "epoch": 1.7226197028068244,
+      "grad_norm": 0.15273383259773254,
+      "learning_rate": 0.00023431639034312096,
+      "loss": 2.1896,
+      "step": 15650
+    },
+    {
+      "epoch": 1.7281232801320858,
+      "grad_norm": 0.17447490990161896,
+      "learning_rate": 0.00023386424855781495,
+      "loss": 2.1896,
+      "step": 15700
+    },
+    {
+      "epoch": 1.7336268574573472,
+      "grad_norm": 0.16734521090984344,
+      "learning_rate": 0.00023341099556760866,
+      "loss": 2.1886,
+      "step": 15750
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.17426900565624237,
+      "learning_rate": 0.0002329566373781238,
+      "loss": 2.1838,
+      "step": 15800
+    },
+    {
+      "epoch": 1.7446340121078703,
+      "grad_norm": 0.1640431135892868,
+      "learning_rate": 0.00023250118000962583,
+      "loss": 2.2026,
+      "step": 15850
+    },
+    {
+      "epoch": 1.7501375894331317,
+      "grad_norm": 0.16575609147548676,
+      "learning_rate": 0.00023204462949694447,
+      "loss": 2.1721,
+      "step": 15900
+    },
+    {
+      "epoch": 1.755641166758393,
+      "grad_norm": 0.17576967179775238,
+      "learning_rate": 0.00023158699188939371,
+      "loss": 2.1517,
+      "step": 15950
+    },
+    {
+      "epoch": 1.7611447440836545,
+      "grad_norm": 0.1661233901977539,
+      "learning_rate": 0.00023112827325069154,
+      "loss": 2.1594,
+      "step": 16000
+    },
+    {
+      "epoch": 1.766648321408916,
+      "grad_norm": 0.17433176934719086,
+      "learning_rate": 0.00023066847965887954,
+      "loss": 2.2069,
+      "step": 16050
+    },
+    {
+      "epoch": 1.7721518987341773,
+      "grad_norm": 0.164288729429245,
+      "learning_rate": 0.0002302076172062427,
+      "loss": 2.1945,
+      "step": 16100
+    },
+    {
+      "epoch": 1.7776554760594387,
+      "grad_norm": 0.16619046032428741,
+      "learning_rate": 0.00022974569199922827,
+      "loss": 2.1678,
+      "step": 16150
+    },
+    {
+      "epoch": 1.7831590533847002,
+      "grad_norm": 0.1572977751493454,
+      "learning_rate": 0.00022928271015836514,
+      "loss": 2.1834,
+      "step": 16200
+    },
+    {
+      "epoch": 1.7886626307099616,
+      "grad_norm": 0.15041351318359375,
+      "learning_rate": 0.0002288186778181826,
+      "loss": 2.1372,
+      "step": 16250
+    },
+    {
+      "epoch": 1.794166208035223,
+      "grad_norm": 0.17850619554519653,
+      "learning_rate": 0.00022835360112712913,
+      "loss": 2.2064,
+      "step": 16300
+    },
+    {
+      "epoch": 1.7996697853604844,
+      "grad_norm": 0.1657785028219223,
+      "learning_rate": 0.00022788748624749083,
+      "loss": 2.1948,
+      "step": 16350
+    },
+    {
+      "epoch": 1.8051733626857458,
+      "grad_norm": 0.16514721512794495,
+      "learning_rate": 0.00022742033935530995,
+      "loss": 2.209,
+      "step": 16400
+    },
+    {
+      "epoch": 1.8106769400110072,
+      "grad_norm": 0.1726425439119339,
+      "learning_rate": 0.00022695216664030297,
+      "loss": 2.1992,
+      "step": 16450
+    },
+    {
+      "epoch": 1.8161805173362686,
+      "grad_norm": 0.1753402054309845,
+      "learning_rate": 0.00022648297430577835,
+      "loss": 2.1929,
+      "step": 16500
+    },
+    {
+      "epoch": 1.82168409466153,
+      "grad_norm": 0.18259774148464203,
+      "learning_rate": 0.00022601276856855482,
+      "loss": 2.2083,
+      "step": 16550
+    },
+    {
+      "epoch": 1.8271876719867914,
+      "grad_norm": 0.16030113399028778,
+      "learning_rate": 0.0002255415556588786,
+      "loss": 2.1571,
+      "step": 16600
+    },
+    {
+      "epoch": 1.8326912493120529,
+      "grad_norm": 0.1431453675031662,
+      "learning_rate": 0.000225069341820341,
+      "loss": 2.149,
+      "step": 16650
+    },
+    {
+      "epoch": 1.8381948266373143,
+      "grad_norm": 0.1664692461490631,
+      "learning_rate": 0.0002245961333097956,
+      "loss": 2.1864,
+      "step": 16700
+    },
+    {
+      "epoch": 1.8436984039625757,
+      "grad_norm": 0.17130005359649658,
+      "learning_rate": 0.00022412193639727564,
+      "loss": 2.18,
+      "step": 16750
+    },
+    {
+      "epoch": 1.849201981287837,
+      "grad_norm": 0.15278129279613495,
+      "learning_rate": 0.00022364675736591058,
+      "loss": 2.1648,
+      "step": 16800
+    },
+    {
+      "epoch": 1.8547055586130985,
+      "grad_norm": 0.176371768116951,
+      "learning_rate": 0.000223170602511843,
+      "loss": 2.1582,
+      "step": 16850
+    },
+    {
+      "epoch": 1.86020913593836,
+      "grad_norm": 0.15902136266231537,
+      "learning_rate": 0.0002226934781441452,
+      "loss": 2.183,
+      "step": 16900
+    },
+    {
+      "epoch": 1.8657127132636213,
+      "grad_norm": 0.14821839332580566,
+      "learning_rate": 0.00022221539058473553,
+      "loss": 2.172,
+      "step": 16950
+    },
+    {
+      "epoch": 1.8712162905888827,
+      "grad_norm": 0.15611177682876587,
+      "learning_rate": 0.00022173634616829481,
+      "loss": 2.1715,
+      "step": 17000
+    },
+    {
+      "epoch": 1.8767198679141441,
+      "grad_norm": 0.1633993536233902,
+      "learning_rate": 0.00022125635124218207,
+      "loss": 2.1961,
+      "step": 17050
+    },
+    {
+      "epoch": 1.8822234452394055,
+      "grad_norm": 0.16087235510349274,
+      "learning_rate": 0.00022077541216635076,
+      "loss": 2.1963,
+      "step": 17100
+    },
+    {
+      "epoch": 1.887727022564667,
+      "grad_norm": 0.16733358800411224,
+      "learning_rate": 0.00022029353531326432,
+      "loss": 2.1605,
+      "step": 17150
+    },
+    {
+      "epoch": 1.8932305998899284,
+      "grad_norm": 0.16267576813697815,
+      "learning_rate": 0.00021981072706781185,
+      "loss": 2.1554,
+      "step": 17200
+    },
+    {
+      "epoch": 1.8987341772151898,
+      "grad_norm": 0.16028568148612976,
+      "learning_rate": 0.0002193269938272232,
+      "loss": 2.1793,
+      "step": 17250
+    },
+    {
+      "epoch": 1.9042377545404512,
+      "grad_norm": 0.15961337089538574,
+      "learning_rate": 0.00021884234200098473,
+      "loss": 2.1673,
+      "step": 17300
+    },
+    {
+      "epoch": 1.9097413318657126,
+      "grad_norm": 0.1637711226940155,
+      "learning_rate": 0.00021835677801075397,
+      "loss": 2.1676,
+      "step": 17350
+    },
+    {
+      "epoch": 1.915244909190974,
+      "grad_norm": 0.16490189731121063,
+      "learning_rate": 0.00021787030829027468,
+      "loss": 2.2119,
+      "step": 17400
+    },
+    {
+      "epoch": 1.9207484865162354,
+      "grad_norm": 0.17354600131511688,
+      "learning_rate": 0.00021738293928529154,
+      "loss": 2.2001,
+      "step": 17450
+    },
+    {
+      "epoch": 1.9262520638414968,
+      "grad_norm": 0.16645126044750214,
+      "learning_rate": 0.00021689467745346483,
+      "loss": 2.1915,
+      "step": 17500
+    },
+    {
+      "epoch": 1.9317556411667582,
+      "grad_norm": 0.16187366843223572,
+      "learning_rate": 0.0002164055292642849,
+      "loss": 2.1829,
+      "step": 17550
+    },
+    {
+      "epoch": 1.9372592184920197,
+      "grad_norm": 0.17859427630901337,
+      "learning_rate": 0.00021591550119898616,
+      "loss": 2.1203,
+      "step": 17600
+    },
+    {
+      "epoch": 1.9427627958172813,
+      "grad_norm": 0.15773439407348633,
+      "learning_rate": 0.00021542459975046164,
+      "loss": 2.1535,
+      "step": 17650
+    },
+    {
+      "epoch": 1.9482663731425427,
+      "grad_norm": 0.1696433126926422,
+      "learning_rate": 0.00021493283142317668,
+      "loss": 2.1271,
+      "step": 17700
+    },
+    {
+      "epoch": 1.953769950467804,
+      "grad_norm": 0.16615457832813263,
+      "learning_rate": 0.00021444020273308267,
+      "loss": 2.1688,
+      "step": 17750
+    },
+    {
+      "epoch": 1.9592735277930655,
+      "grad_norm": 0.16561925411224365,
+      "learning_rate": 0.00021394672020753102,
+      "loss": 2.1228,
+      "step": 17800
+    },
+    {
+      "epoch": 1.964777105118327,
+      "grad_norm": 0.1655699759721756,
+      "learning_rate": 0.00021345239038518638,
+      "loss": 2.1614,
+      "step": 17850
+    },
+    {
+      "epoch": 1.9702806824435883,
+      "grad_norm": 0.16106578707695007,
+      "learning_rate": 0.0002129572198159401,
+      "loss": 2.1471,
+      "step": 17900
+    },
+    {
+      "epoch": 1.9757842597688497,
+      "grad_norm": 0.1635727882385254,
+      "learning_rate": 0.00021246121506082367,
+      "loss": 2.184,
+      "step": 17950
+    },
+    {
+      "epoch": 1.9812878370941112,
+      "grad_norm": 0.15699493885040283,
+      "learning_rate": 0.0002119643826919213,
+      "loss": 2.1602,
+      "step": 18000
+    },
+    {
+      "epoch": 1.9867914144193726,
+      "grad_norm": 0.15784280002117157,
+      "learning_rate": 0.00021146672929228327,
+      "loss": 2.1781,
+      "step": 18050
+    },
+    {
+      "epoch": 1.992294991744634,
+      "grad_norm": 0.15157456696033478,
+      "learning_rate": 0.00021096826145583852,
+      "loss": 2.1532,
+      "step": 18100
+    },
+    {
+      "epoch": 1.9977985690698954,
+      "grad_norm": 0.15633012354373932,
+      "learning_rate": 0.0002104689857873074,
+      "loss": 2.1676,
+      "step": 18150
+    },
+    {
+      "epoch": 2.003302146395157,
+      "grad_norm": 0.16586118936538696,
+      "learning_rate": 0.00020996890890211386,
+      "loss": 2.1335,
+      "step": 18200
+    },
+    {
+      "epoch": 2.0088057237204184,
+      "grad_norm": 0.1573173552751541,
+      "learning_rate": 0.00020946803742629826,
+      "loss": 2.11,
+      "step": 18250
+    },
+    {
+      "epoch": 2.01430930104568,
+      "grad_norm": 0.18040034174919128,
+      "learning_rate": 0.00020896637799642915,
+      "loss": 2.1128,
+      "step": 18300
+    },
+    {
+      "epoch": 2.0198128783709413,
+      "grad_norm": 0.1664518266916275,
+      "learning_rate": 0.00020846393725951556,
+      "loss": 2.1384,
+      "step": 18350
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.17078042030334473,
+      "learning_rate": 0.0002079607218729188,
+      "loss": 2.0909,
+      "step": 18400
+    },
+    {
+      "epoch": 2.030820033021464,
+      "grad_norm": 0.15759077668190002,
+      "learning_rate": 0.00020745673850426446,
+      "loss": 2.1015,
+      "step": 18450
+    },
+    {
+      "epoch": 2.0363236103467255,
+      "grad_norm": 0.17311988770961761,
+      "learning_rate": 0.00020695199383135386,
+      "loss": 2.1436,
+      "step": 18500
+    },
+    {
+      "epoch": 2.041827187671987,
+      "grad_norm": 0.1612362265586853,
+      "learning_rate": 0.00020644649454207553,
+      "loss": 2.0868,
+      "step": 18550
+    },
+    {
+      "epoch": 2.0473307649972483,
+      "grad_norm": 0.16166241466999054,
+      "learning_rate": 0.0002059402473343169,
+      "loss": 2.0683,
+      "step": 18600
+    },
+    {
+      "epoch": 2.0528343423225097,
+      "grad_norm": 0.17205455899238586,
+      "learning_rate": 0.00020543325891587516,
+      "loss": 2.0618,
+      "step": 18650
+    },
+    {
+      "epoch": 2.058337919647771,
+      "grad_norm": 0.17534597218036652,
+      "learning_rate": 0.00020492553600436872,
+      "loss": 2.1173,
+      "step": 18700
+    },
+    {
+      "epoch": 2.0638414969730325,
+      "grad_norm": 0.15081949532032013,
+      "learning_rate": 0.00020441708532714797,
+      "loss": 2.0991,
+      "step": 18750
+    },
+    {
+      "epoch": 2.069345074298294,
+      "grad_norm": 0.15837198495864868,
+      "learning_rate": 0.00020390791362120625,
+      "loss": 2.1332,
+      "step": 18800
+    },
+    {
+      "epoch": 2.0748486516235554,
+      "grad_norm": 0.1598348468542099,
+      "learning_rate": 0.00020339802763309054,
+      "loss": 2.1164,
+      "step": 18850
+    },
+    {
+      "epoch": 2.0803522289488168,
+      "grad_norm": 0.1536133736371994,
+      "learning_rate": 0.0002028874341188122,
+      "loss": 2.1169,
+      "step": 18900
+    },
+    {
+      "epoch": 2.085855806274078,
+      "grad_norm": 0.1654282808303833,
+      "learning_rate": 0.00020237613984375711,
+      "loss": 2.1253,
+      "step": 18950
+    },
+    {
+      "epoch": 2.0913593835993396,
+      "grad_norm": 0.16015826165676117,
+      "learning_rate": 0.00020186415158259652,
+      "loss": 2.1196,
+      "step": 19000
+    },
+    {
+      "epoch": 2.096862960924601,
+      "grad_norm": 0.16205786168575287,
+      "learning_rate": 0.00020135147611919686,
+      "loss": 2.0872,
+      "step": 19050
+    },
+    {
+      "epoch": 2.1023665382498624,
+      "grad_norm": 0.16814254224300385,
+      "learning_rate": 0.00020083812024653013,
+      "loss": 2.1569,
+      "step": 19100
+    },
+    {
+      "epoch": 2.107870115575124,
+      "grad_norm": 0.15433181822299957,
+      "learning_rate": 0.00020032409076658364,
+      "loss": 2.0967,
+      "step": 19150
+    },
+    {
+      "epoch": 2.1133736929003852,
+      "grad_norm": 0.16261817514896393,
+      "learning_rate": 0.00019980939449027016,
+      "loss": 2.0938,
+      "step": 19200
+    },
+    {
+      "epoch": 2.1188772702256466,
+      "grad_norm": 0.17877838015556335,
+      "learning_rate": 0.00019929403823733748,
+      "loss": 2.1399,
+      "step": 19250
+    },
+    {
+      "epoch": 2.124380847550908,
+      "grad_norm": 0.178861603140831,
+      "learning_rate": 0.00019877802883627808,
+      "loss": 2.1362,
+      "step": 19300
+    },
+    {
+      "epoch": 2.1298844248761695,
+      "grad_norm": 0.1621016561985016,
+      "learning_rate": 0.00019826137312423873,
+      "loss": 2.1089,
+      "step": 19350
+    },
+    {
+      "epoch": 2.135388002201431,
+      "grad_norm": 0.17772269248962402,
+      "learning_rate": 0.00019774407794692986,
+      "loss": 2.1278,
+      "step": 19400
+    },
+    {
+      "epoch": 2.1408915795266923,
+      "grad_norm": 0.16339927911758423,
+      "learning_rate": 0.00019722615015853483,
+      "loss": 2.1244,
+      "step": 19450
+    },
+    {
+      "epoch": 2.1463951568519537,
+      "grad_norm": 0.16879414021968842,
+      "learning_rate": 0.0001967075966216191,
+      "loss": 2.106,
+      "step": 19500
+    },
+    {
+      "epoch": 2.151898734177215,
+      "grad_norm": 0.15002557635307312,
+      "learning_rate": 0.00019618842420703923,
+      "loss": 2.1007,
+      "step": 19550
+    },
+    {
+      "epoch": 2.1574023115024765,
+      "grad_norm": 0.17130738496780396,
+      "learning_rate": 0.0001956686397938522,
+      "loss": 2.1104,
+      "step": 19600
+    },
+    {
+      "epoch": 2.162905888827738,
+      "grad_norm": 0.16796913743019104,
+      "learning_rate": 0.00019514825026922377,
+      "loss": 2.1278,
+      "step": 19650
+    },
+    {
+      "epoch": 2.1684094661529993,
+      "grad_norm": 0.19111864268779755,
+      "learning_rate": 0.0001946272625283375,
+      "loss": 2.1402,
+      "step": 19700
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 0.17543087899684906,
+      "learning_rate": 0.00019410568347430344,
+      "loss": 2.0984,
+      "step": 19750
+    },
+    {
+      "epoch": 2.179416620803522,
+      "grad_norm": 0.15795323252677917,
+      "learning_rate": 0.00019358352001806642,
+      "loss": 2.1125,
+      "step": 19800
+    },
+    {
+      "epoch": 2.1849201981287836,
+      "grad_norm": 0.15581749379634857,
+      "learning_rate": 0.0001930607790783147,
+      "loss": 2.119,
+      "step": 19850
+    },
+    {
+      "epoch": 2.190423775454045,
+      "grad_norm": 0.17086704075336456,
+      "learning_rate": 0.00019253746758138822,
+      "loss": 2.1029,
+      "step": 19900
+    },
+    {
+      "epoch": 2.1959273527793064,
+      "grad_norm": 0.15870486199855804,
+      "learning_rate": 0.00019201359246118682,
+      "loss": 2.1133,
+      "step": 19950
+    },
+    {
+      "epoch": 2.201430930104568,
+      "grad_norm": 0.16840116679668427,
+      "learning_rate": 0.0001914891606590784,
+      "loss": 2.1196,
+      "step": 20000
+    },
+    {
+      "epoch": 2.206934507429829,
+      "grad_norm": 0.1643972396850586,
+      "learning_rate": 0.00019096417912380686,
+      "loss": 2.1173,
+      "step": 20050
+    },
+    {
+      "epoch": 2.2124380847550906,
+      "grad_norm": 0.1598280966281891,
+      "learning_rate": 0.0001904386548114001,
+      "loss": 2.1338,
+      "step": 20100
+    },
+    {
+      "epoch": 2.217941662080352,
+      "grad_norm": 0.16909152269363403,
+      "learning_rate": 0.0001899125946850779,
+      "loss": 2.0987,
+      "step": 20150
+    },
+    {
+      "epoch": 2.2234452394056134,
+      "grad_norm": 0.15597747266292572,
+      "learning_rate": 0.00018938600571515962,
+      "loss": 2.1295,
+      "step": 20200
+    },
+    {
+      "epoch": 2.228948816730875,
+      "grad_norm": 0.16759732365608215,
+      "learning_rate": 0.00018885889487897162,
+      "loss": 2.1374,
+      "step": 20250
+    },
+    {
+      "epoch": 2.2344523940561363,
+      "grad_norm": 0.15526509284973145,
+      "learning_rate": 0.00018833126916075526,
+      "loss": 2.1287,
+      "step": 20300
+    },
+    {
+      "epoch": 2.2399559713813977,
+      "grad_norm": 0.16562257707118988,
+      "learning_rate": 0.00018780313555157393,
+      "loss": 2.126,
+      "step": 20350
+    },
+    {
+      "epoch": 2.2454595487066595,
+      "grad_norm": 0.17471948266029358,
+      "learning_rate": 0.00018727450104922075,
+      "loss": 2.1175,
+      "step": 20400
+    },
+    {
+      "epoch": 2.2509631260319205,
+      "grad_norm": 0.16391079127788544,
+      "learning_rate": 0.00018674537265812548,
+      "loss": 2.0965,
+      "step": 20450
+    },
+    {
+      "epoch": 2.2564667033571824,
+      "grad_norm": 0.1693831831216812,
+      "learning_rate": 0.00018621575738926217,
+      "loss": 2.1124,
+      "step": 20500
+    },
+    {
+      "epoch": 2.2619702806824438,
+      "grad_norm": 0.16163313388824463,
+      "learning_rate": 0.00018568566226005587,
+      "loss": 2.1034,
+      "step": 20550
+    },
+    {
+      "epoch": 2.267473858007705,
+      "grad_norm": 0.1676999181509018,
+      "learning_rate": 0.00018515509429428988,
+      "loss": 2.0991,
+      "step": 20600
+    },
+    {
+      "epoch": 2.2729774353329666,
+      "grad_norm": 0.16531667113304138,
+      "learning_rate": 0.00018462406052201252,
+      "loss": 2.1183,
+      "step": 20650
+    },
+    {
+      "epoch": 2.278481012658228,
+      "grad_norm": 0.16208726167678833,
+      "learning_rate": 0.00018409256797944412,
+      "loss": 2.0981,
+      "step": 20700
+    },
+    {
+      "epoch": 2.2839845899834894,
+      "grad_norm": 0.1707569658756256,
+      "learning_rate": 0.0001835606237088837,
+      "loss": 2.0926,
+      "step": 20750
+    },
+    {
+      "epoch": 2.289488167308751,
+      "grad_norm": 0.16191871464252472,
+      "learning_rate": 0.00018302823475861582,
+      "loss": 2.0895,
+      "step": 20800
+    },
+    {
+      "epoch": 2.2949917446340122,
+      "grad_norm": 0.154753640294075,
+      "learning_rate": 0.0001824954081828168,
+      "loss": 2.0972,
+      "step": 20850
+    },
+    {
+      "epoch": 2.3004953219592736,
+      "grad_norm": 0.16868236660957336,
+      "learning_rate": 0.00018196215104146187,
+      "loss": 2.1009,
+      "step": 20900
+    },
+    {
+      "epoch": 2.305998899284535,
+      "grad_norm": 0.1602269858121872,
+      "learning_rate": 0.00018142847040023097,
+      "loss": 2.1063,
+      "step": 20950
+    },
+    {
+      "epoch": 2.3115024766097965,
+      "grad_norm": 0.17220260202884674,
+      "learning_rate": 0.00018089437333041562,
+      "loss": 2.1049,
+      "step": 21000
+    },
+    {
+      "epoch": 2.317006053935058,
+      "grad_norm": 0.14858907461166382,
+      "learning_rate": 0.0001803598669088249,
+      "loss": 2.1129,
+      "step": 21050
+    },
+    {
+      "epoch": 2.3225096312603193,
+      "grad_norm": 0.17586296796798706,
+      "learning_rate": 0.0001798249582176919,
+      "loss": 2.0961,
+      "step": 21100
+    },
+    {
+      "epoch": 2.3280132085855807,
+      "grad_norm": 0.14931316673755646,
+      "learning_rate": 0.00017928965434457983,
+      "loss": 2.1108,
+      "step": 21150
+    },
+    {
+      "epoch": 2.333516785910842,
+      "grad_norm": 0.17148137092590332,
+      "learning_rate": 0.00017875396238228798,
+      "loss": 2.1029,
+      "step": 21200
+    },
+    {
+      "epoch": 2.3390203632361035,
+      "grad_norm": 0.15888847410678864,
+      "learning_rate": 0.00017821788942875793,
+      "loss": 2.1258,
+      "step": 21250
+    },
+    {
+      "epoch": 2.344523940561365,
+      "grad_norm": 0.1749711036682129,
+      "learning_rate": 0.0001776814425869793,
+      "loss": 2.1176,
+      "step": 21300
+    },
+    {
+      "epoch": 2.3500275178866263,
+      "grad_norm": 0.16511616110801697,
+      "learning_rate": 0.0001771446289648959,
+      "loss": 2.0992,
+      "step": 21350
+    },
+    {
+      "epoch": 2.3555310952118877,
+      "grad_norm": 0.16869331896305084,
+      "learning_rate": 0.00017660745567531124,
+      "loss": 2.1274,
+      "step": 21400
+    },
+    {
+      "epoch": 2.361034672537149,
+      "grad_norm": 0.16024701297283173,
+      "learning_rate": 0.00017606992983579457,
+      "loss": 2.1379,
+      "step": 21450
+    },
+    {
+      "epoch": 2.3665382498624106,
+      "grad_norm": 0.1646272987127304,
+      "learning_rate": 0.00017553205856858637,
+      "loss": 2.1365,
+      "step": 21500
+    },
+    {
+      "epoch": 2.372041827187672,
+      "grad_norm": 0.1765459179878235,
+      "learning_rate": 0.00017499384900050404,
+      "loss": 2.1125,
+      "step": 21550
+    },
+    {
+      "epoch": 2.3775454045129334,
+      "grad_norm": 0.15617190301418304,
+      "learning_rate": 0.0001744553082628475,
+      "loss": 2.0985,
+      "step": 21600
+    },
+    {
+      "epoch": 2.383048981838195,
+      "grad_norm": 0.14687784016132355,
+      "learning_rate": 0.00017391644349130467,
+      "loss": 2.0771,
+      "step": 21650
+    },
+    {
+      "epoch": 2.388552559163456,
+      "grad_norm": 0.16840553283691406,
+      "learning_rate": 0.00017337726182585698,
+      "loss": 2.0909,
+      "step": 21700
+    },
+    {
+      "epoch": 2.3940561364887176,
+      "grad_norm": 0.16857478022575378,
+      "learning_rate": 0.00017283777041068472,
+      "loss": 2.0968,
+      "step": 21750
+    },
+    {
+      "epoch": 2.399559713813979,
+      "grad_norm": 0.1799498200416565,
+      "learning_rate": 0.00017229797639407233,
+      "loss": 2.1023,
+      "step": 21800
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 0.15700192749500275,
+      "learning_rate": 0.00017175788692831367,
+      "loss": 2.0986,
+      "step": 21850
+    },
+    {
+      "epoch": 2.410566868464502,
+      "grad_norm": 0.16778048872947693,
+      "learning_rate": 0.00017121750916961742,
+      "loss": 2.0914,
+      "step": 21900
+    },
+    {
+      "epoch": 2.4160704457897633,
+      "grad_norm": 0.15119898319244385,
+      "learning_rate": 0.0001706768502780122,
+      "loss": 2.0857,
+      "step": 21950
+    },
+    {
+      "epoch": 2.4215740231150247,
+      "grad_norm": 0.17003406584262848,
+      "learning_rate": 0.0001701359174172515,
+      "loss": 2.1366,
+      "step": 22000
+    },
+    {
+      "epoch": 2.427077600440286,
+      "grad_norm": 0.17182987928390503,
+      "learning_rate": 0.000169594717754719,
+      "loss": 2.1126,
+      "step": 22050
+    },
+    {
+      "epoch": 2.4325811777655475,
+      "grad_norm": 0.16026781499385834,
+      "learning_rate": 0.00016905325846133354,
+      "loss": 2.076,
+      "step": 22100
+    },
+    {
+      "epoch": 2.438084755090809,
+      "grad_norm": 0.16470612585544586,
+      "learning_rate": 0.00016851154671145403,
+      "loss": 2.0907,
+      "step": 22150
+    },
+    {
+      "epoch": 2.4435883324160703,
+      "grad_norm": 0.14621587097644806,
+      "learning_rate": 0.0001679695896827844,
+      "loss": 2.1073,
+      "step": 22200
+    },
+    {
+      "epoch": 2.4490919097413317,
+      "grad_norm": 0.1563323438167572,
+      "learning_rate": 0.00016742739455627862,
+      "loss": 2.0757,
+      "step": 22250
+    },
+    {
+      "epoch": 2.454595487066593,
+      "grad_norm": 0.17763760685920715,
+      "learning_rate": 0.0001668849685160455,
+      "loss": 2.0705,
+      "step": 22300
+    },
+    {
+      "epoch": 2.4600990643918546,
+      "grad_norm": 0.16651757061481476,
+      "learning_rate": 0.00016634231874925335,
+      "loss": 2.0881,
+      "step": 22350
+    },
+    {
+      "epoch": 2.465602641717116,
+      "grad_norm": 0.16752833127975464,
+      "learning_rate": 0.0001657994524460349,
+      "loss": 2.0639,
+      "step": 22400
+    },
+    {
+      "epoch": 2.4711062190423774,
+      "grad_norm": 0.14574387669563293,
+      "learning_rate": 0.000165256376799392,
+      "loss": 2.087,
+      "step": 22450
+    },
+    {
+      "epoch": 2.476609796367639,
+      "grad_norm": 0.15595164895057678,
+      "learning_rate": 0.00016471309900510043,
+      "loss": 2.0965,
+      "step": 22500
+    },
+    {
+      "epoch": 2.4821133736929,
+      "grad_norm": 0.17218752205371857,
+      "learning_rate": 0.0001641696262616142,
+      "loss": 2.0943,
+      "step": 22550
+    },
+    {
+      "epoch": 2.487616951018162,
+      "grad_norm": 0.1563566029071808,
+      "learning_rate": 0.00016362596576997057,
+      "loss": 2.0946,
+      "step": 22600
+    },
+    {
+      "epoch": 2.4931205283434235,
+      "grad_norm": 0.16190436482429504,
+      "learning_rate": 0.00016308212473369453,
+      "loss": 2.0924,
+      "step": 22650
+    },
+    {
+      "epoch": 2.498624105668685,
+      "grad_norm": 0.16890835762023926,
+      "learning_rate": 0.00016253811035870313,
+      "loss": 2.0954,
+      "step": 22700
+    },
+    {
+      "epoch": 2.5041276829939463,
+      "grad_norm": 0.16371208429336548,
+      "learning_rate": 0.00016199392985321017,
+      "loss": 2.0954,
+      "step": 22750
+    },
+    {
+      "epoch": 2.5096312603192077,
+      "grad_norm": 0.15898428857326508,
+      "learning_rate": 0.00016144959042763084,
+      "loss": 2.0766,
+      "step": 22800
+    },
+    {
+      "epoch": 2.515134837644469,
+      "grad_norm": 0.15516646206378937,
+      "learning_rate": 0.00016090509929448594,
+      "loss": 2.0891,
+      "step": 22850
+    },
+    {
+      "epoch": 2.5206384149697305,
+      "grad_norm": 0.15007217228412628,
+      "learning_rate": 0.00016036046366830635,
+      "loss": 2.078,
+      "step": 22900
+    },
+    {
+      "epoch": 2.526141992294992,
+      "grad_norm": 0.17989104986190796,
+      "learning_rate": 0.00015981569076553765,
+      "loss": 2.0888,
+      "step": 22950
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.16754071414470673,
+      "learning_rate": 0.00015927078780444403,
+      "loss": 2.0882,
+      "step": 23000
+    },
+    {
+      "epoch": 2.5371491469455147,
+      "grad_norm": 0.15897449851036072,
+      "learning_rate": 0.0001587257620050134,
+      "loss": 2.0933,
+      "step": 23050
+    },
+    {
+      "epoch": 2.542652724270776,
+      "grad_norm": 0.16594423353672028,
+      "learning_rate": 0.00015818062058886082,
+      "loss": 2.0968,
+      "step": 23100
+    },
+    {
+      "epoch": 2.5481563015960376,
+      "grad_norm": 0.1700371652841568,
+      "learning_rate": 0.00015763537077913365,
+      "loss": 2.096,
+      "step": 23150
+    },
+    {
+      "epoch": 2.553659878921299,
+      "grad_norm": 0.16013766825199127,
+      "learning_rate": 0.00015709001980041527,
+      "loss": 2.1134,
+      "step": 23200
+    },
+    {
+      "epoch": 2.5591634562465604,
+      "grad_norm": 0.1601455807685852,
+      "learning_rate": 0.00015654457487862961,
+      "loss": 2.0983,
+      "step": 23250
+    },
+    {
+      "epoch": 2.564667033571822,
+      "grad_norm": 0.16373537480831146,
+      "learning_rate": 0.00015599904324094534,
+      "loss": 2.0866,
+      "step": 23300
+    },
+    {
+      "epoch": 2.570170610897083,
+      "grad_norm": 0.15919940173625946,
+      "learning_rate": 0.0001554534321156801,
+      "loss": 2.1147,
+      "step": 23350
+    },
+    {
+      "epoch": 2.5756741882223446,
+      "grad_norm": 0.1523171216249466,
+      "learning_rate": 0.0001549077487322048,
+      "loss": 2.0887,
+      "step": 23400
+    },
+    {
+      "epoch": 2.581177765547606,
+      "grad_norm": 0.16425256431102753,
+      "learning_rate": 0.00015436200032084774,
+      "loss": 2.0918,
+      "step": 23450
+    },
+    {
+      "epoch": 2.5866813428728674,
+      "grad_norm": 0.16046775877475739,
+      "learning_rate": 0.0001538161941127988,
+      "loss": 2.0816,
+      "step": 23500
+    },
+    {
+      "epoch": 2.592184920198129,
+      "grad_norm": 0.1673419028520584,
+      "learning_rate": 0.00015327033734001375,
+      "loss": 2.1186,
+      "step": 23550
+    },
+    {
+      "epoch": 2.5976884975233903,
+      "grad_norm": 0.15573158860206604,
+      "learning_rate": 0.00015272443723511825,
+      "loss": 2.0721,
+      "step": 23600
+    },
+    {
+      "epoch": 2.6031920748486517,
+      "grad_norm": 0.16536468267440796,
+      "learning_rate": 0.00015217850103131227,
+      "loss": 2.0991,
+      "step": 23650
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 0.15281961858272552,
+      "learning_rate": 0.00015163253596227393,
+      "loss": 2.0713,
+      "step": 23700
+    },
+    {
+      "epoch": 2.6141992294991745,
+      "grad_norm": 0.1582518070936203,
+      "learning_rate": 0.0001510865492620638,
+      "loss": 2.0923,
+      "step": 23750
+    },
+    {
+      "epoch": 2.619702806824436,
+      "grad_norm": 0.1581014096736908,
+      "learning_rate": 0.0001505405481650293,
+      "loss": 2.0895,
+      "step": 23800
+    },
+    {
+      "epoch": 2.6252063841496973,
+      "grad_norm": 0.16346633434295654,
+      "learning_rate": 0.00014999453990570839,
+      "loss": 2.1011,
+      "step": 23850
+    },
+    {
+      "epoch": 2.6307099614749587,
+      "grad_norm": 0.16536889970302582,
+      "learning_rate": 0.00014944853171873387,
+      "loss": 2.113,
+      "step": 23900
+    },
+    {
+      "epoch": 2.63621353880022,
+      "grad_norm": 0.170758455991745,
+      "learning_rate": 0.00014890253083873784,
+      "loss": 2.1124,
+      "step": 23950
+    },
+    {
+      "epoch": 2.6417171161254815,
+      "grad_norm": 0.15669451653957367,
+      "learning_rate": 0.00014835654450025542,
+      "loss": 2.0894,
+      "step": 24000
+    },
+    {
+      "epoch": 2.647220693450743,
+      "grad_norm": 0.16937313973903656,
+      "learning_rate": 0.00014781057993762902,
+      "loss": 2.127,
+      "step": 24050
+    },
+    {
+      "epoch": 2.6527242707760044,
+      "grad_norm": 0.15754981338977814,
+      "learning_rate": 0.0001472646443849126,
+      "loss": 2.0959,
+      "step": 24100
+    },
+    {
+      "epoch": 2.6582278481012658,
+      "grad_norm": 0.16420376300811768,
+      "learning_rate": 0.00014671874507577574,
+      "loss": 2.0742,
+      "step": 24150
+    },
+    {
+      "epoch": 2.663731425426527,
+      "grad_norm": 0.16400571167469025,
+      "learning_rate": 0.00014617288924340775,
+      "loss": 2.0992,
+      "step": 24200
+    },
+    {
+      "epoch": 2.6692350027517886,
+      "grad_norm": 0.1658066064119339,
+      "learning_rate": 0.00014562708412042193,
+      "loss": 2.0925,
+      "step": 24250
+    },
+    {
+      "epoch": 2.67473858007705,
+      "grad_norm": 0.16872134804725647,
+      "learning_rate": 0.00014508133693875953,
+      "loss": 2.0899,
+      "step": 24300
+    },
+    {
+      "epoch": 2.6802421574023114,
+      "grad_norm": 0.17002440989017487,
+      "learning_rate": 0.00014453565492959422,
+      "loss": 2.1252,
+      "step": 24350
+    },
+    {
+      "epoch": 2.685745734727573,
+      "grad_norm": 0.16417178511619568,
+      "learning_rate": 0.00014399004532323614,
+      "loss": 2.112,
+      "step": 24400
+    },
+    {
+      "epoch": 2.6912493120528342,
+      "grad_norm": 0.172217458486557,
+      "learning_rate": 0.000143444515349036,
+      "loss": 2.0784,
+      "step": 24450
+    },
+    {
+      "epoch": 2.6967528893780957,
+      "grad_norm": 0.16687516868114471,
+      "learning_rate": 0.0001428990722352894,
+      "loss": 2.0775,
+      "step": 24500
+    },
+    {
+      "epoch": 2.702256466703357,
+      "grad_norm": 0.162861630320549,
+      "learning_rate": 0.00014235372320914108,
+      "loss": 2.0928,
+      "step": 24550
+    },
+    {
+      "epoch": 2.7077600440286185,
+      "grad_norm": 0.1640874743461609,
+      "learning_rate": 0.00014180847549648913,
+      "loss": 2.1071,
+      "step": 24600
+    },
+    {
+      "epoch": 2.71326362135388,
+      "grad_norm": 0.1654275506734848,
+      "learning_rate": 0.0001412633363218891,
+      "loss": 2.0615,
+      "step": 24650
+    },
+    {
+      "epoch": 2.7187671986791413,
+      "grad_norm": 0.14277833700180054,
+      "learning_rate": 0.00014071831290845857,
+      "loss": 2.1078,
+      "step": 24700
+    },
+    {
+      "epoch": 2.7242707760044027,
+      "grad_norm": 0.1705639362335205,
+      "learning_rate": 0.00014017341247778123,
+      "loss": 2.0858,
+      "step": 24750
+    },
+    {
+      "epoch": 2.729774353329664,
+      "grad_norm": 0.16390259563922882,
+      "learning_rate": 0.0001396286422498112,
+      "loss": 2.0508,
+      "step": 24800
+    },
+    {
+      "epoch": 2.7352779306549255,
+      "grad_norm": 0.1526353508234024,
+      "learning_rate": 0.00013908400944277738,
+      "loss": 2.0784,
+      "step": 24850
+    },
+    {
+      "epoch": 2.740781507980187,
+      "grad_norm": 0.1581793874502182,
+      "learning_rate": 0.0001385395212730879,
+      "loss": 2.0596,
+      "step": 24900
+    },
+    {
+      "epoch": 2.7462850853054483,
+      "grad_norm": 0.1692277193069458,
+      "learning_rate": 0.00013799518495523445,
+      "loss": 2.0838,
+      "step": 24950
+    },
+    {
+      "epoch": 2.7517886626307098,
+      "grad_norm": 0.1614445298910141,
+      "learning_rate": 0.0001374510077016967,
+      "loss": 2.0928,
+      "step": 25000
+    },
+    {
+      "epoch": 2.757292239955971,
+      "grad_norm": 0.15747620165348053,
+      "learning_rate": 0.00013690699672284657,
+      "loss": 2.0686,
+      "step": 25050
+    },
+    {
+      "epoch": 2.7627958172812326,
+      "grad_norm": 0.17565257847309113,
+      "learning_rate": 0.00013636315922685294,
+      "loss": 2.0971,
+      "step": 25100
+    },
+    {
+      "epoch": 2.768299394606494,
+      "grad_norm": 0.16367433965206146,
+      "learning_rate": 0.0001358195024195861,
+      "loss": 2.1052,
+      "step": 25150
+    },
+    {
+      "epoch": 2.7738029719317554,
+      "grad_norm": 0.1564227044582367,
+      "learning_rate": 0.00013527603350452206,
+      "loss": 2.0902,
+      "step": 25200
+    },
+    {
+      "epoch": 2.779306549257017,
+      "grad_norm": 0.16555672883987427,
+      "learning_rate": 0.0001347327596826473,
+      "loss": 2.0683,
+      "step": 25250
+    },
+    {
+      "epoch": 2.7848101265822782,
+      "grad_norm": 0.16134041547775269,
+      "learning_rate": 0.00013418968815236346,
+      "loss": 2.0627,
+      "step": 25300
+    },
+    {
+      "epoch": 2.7903137039075396,
+      "grad_norm": 0.17097671329975128,
+      "learning_rate": 0.0001336468261093915,
+      "loss": 2.0724,
+      "step": 25350
+    },
+    {
+      "epoch": 2.795817281232801,
+      "grad_norm": 0.1656900942325592,
+      "learning_rate": 0.00013310418074667698,
+      "loss": 2.1266,
+      "step": 25400
+    },
+    {
+      "epoch": 2.801320858558063,
+      "grad_norm": 0.1745171695947647,
+      "learning_rate": 0.00013256175925429415,
+      "loss": 2.0983,
+      "step": 25450
+    },
+    {
+      "epoch": 2.8068244358833243,
+      "grad_norm": 0.15649184584617615,
+      "learning_rate": 0.0001320195688193512,
+      "loss": 2.0672,
+      "step": 25500
+    },
+    {
+      "epoch": 2.8123280132085857,
+      "grad_norm": 0.16284936666488647,
+      "learning_rate": 0.00013147761662589473,
+      "loss": 2.0865,
+      "step": 25550
+    },
+    {
+      "epoch": 2.817831590533847,
+      "grad_norm": 0.15236727893352509,
+      "learning_rate": 0.00013093590985481456,
+      "loss": 2.0595,
+      "step": 25600
+    },
+    {
+      "epoch": 2.8233351678591085,
+      "grad_norm": 0.166220560669899,
+      "learning_rate": 0.00013039445568374872,
+      "loss": 2.0478,
+      "step": 25650
+    },
+    {
+      "epoch": 2.82883874518437,
+      "grad_norm": 0.16826435923576355,
+      "learning_rate": 0.0001298532612869883,
+      "loss": 2.0756,
+      "step": 25700
+    },
+    {
+      "epoch": 2.8343423225096314,
+      "grad_norm": 0.16447168588638306,
+      "learning_rate": 0.0001293123338353823,
+      "loss": 2.0742,
+      "step": 25750
+    },
+    {
+      "epoch": 2.8398458998348928,
+      "grad_norm": 0.157373309135437,
+      "learning_rate": 0.00012877168049624268,
+      "loss": 2.0728,
+      "step": 25800
+    },
+    {
+      "epoch": 2.845349477160154,
+      "grad_norm": 0.1642802357673645,
+      "learning_rate": 0.00012823130843324948,
+      "loss": 2.0643,
+      "step": 25850
+    },
+    {
+      "epoch": 2.8508530544854156,
+      "grad_norm": 0.15527865290641785,
+      "learning_rate": 0.00012769122480635583,
+      "loss": 2.1088,
+      "step": 25900
+    },
+    {
+      "epoch": 2.856356631810677,
+      "grad_norm": 0.175029918551445,
+      "learning_rate": 0.00012715143677169297,
+      "loss": 2.0764,
+      "step": 25950
+    },
+    {
+      "epoch": 2.8618602091359384,
+      "grad_norm": 0.17664563655853271,
+      "learning_rate": 0.0001266119514814755,
+      "loss": 2.0527,
+      "step": 26000
+    },
+    {
+      "epoch": 2.8673637864612,
+      "grad_norm": 0.16839756071567535,
+      "learning_rate": 0.00012607277608390677,
+      "loss": 2.0708,
+      "step": 26050
+    },
+    {
+      "epoch": 2.8728673637864612,
+      "grad_norm": 0.162934809923172,
+      "learning_rate": 0.00012553391772308407,
+      "loss": 2.0984,
+      "step": 26100
+    },
+    {
+      "epoch": 2.8783709411117226,
+      "grad_norm": 0.1836494356393814,
+      "learning_rate": 0.0001249953835389037,
+      "loss": 2.1118,
+      "step": 26150
+    },
+    {
+      "epoch": 2.883874518436984,
+      "grad_norm": 0.15794949233531952,
+      "learning_rate": 0.00012445718066696687,
+      "loss": 2.0541,
+      "step": 26200
+    },
+    {
+      "epoch": 2.8893780957622455,
+      "grad_norm": 0.16560156643390656,
+      "learning_rate": 0.00012391931623848476,
+      "loss": 2.1111,
+      "step": 26250
+    },
+    {
+      "epoch": 2.894881673087507,
+      "grad_norm": 0.16835126280784607,
+      "learning_rate": 0.0001233817973801842,
+      "loss": 2.0635,
+      "step": 26300
+    },
+    {
+      "epoch": 2.9003852504127683,
+      "grad_norm": 0.16151325404644012,
+      "learning_rate": 0.0001228446312142131,
+      "loss": 2.0949,
+      "step": 26350
+    },
+    {
+      "epoch": 2.9058888277380297,
+      "grad_norm": 0.16751372814178467,
+      "learning_rate": 0.00012230782485804633,
+      "loss": 2.0896,
+      "step": 26400
+    },
+    {
+      "epoch": 2.911392405063291,
+      "grad_norm": 0.18384750187397003,
+      "learning_rate": 0.00012177138542439122,
+      "loss": 2.047,
+      "step": 26450
+    },
+    {
+      "epoch": 2.9168959823885525,
+      "grad_norm": 0.15618419647216797,
+      "learning_rate": 0.00012123532002109328,
+      "loss": 2.0787,
+      "step": 26500
+    },
+    {
+      "epoch": 2.922399559713814,
+      "grad_norm": 0.15803013741970062,
+      "learning_rate": 0.00012069963575104223,
+      "loss": 2.0735,
+      "step": 26550
+    },
+    {
+      "epoch": 2.9279031370390753,
+      "grad_norm": 0.1603049784898758,
+      "learning_rate": 0.00012016433971207757,
+      "loss": 2.0802,
+      "step": 26600
+    },
+    {
+      "epoch": 2.9334067143643368,
+      "grad_norm": 0.1752861887216568,
+      "learning_rate": 0.0001196294389968949,
+      "loss": 2.0969,
+      "step": 26650
+    },
+    {
+      "epoch": 2.938910291689598,
+      "grad_norm": 0.16382624208927155,
+      "learning_rate": 0.00011909494069295175,
+      "loss": 2.0933,
+      "step": 26700
+    },
+    {
+      "epoch": 2.9444138690148596,
+      "grad_norm": 0.16802163422107697,
+      "learning_rate": 0.00011856085188237357,
+      "loss": 2.1042,
+      "step": 26750
+    },
+    {
+      "epoch": 2.949917446340121,
+      "grad_norm": 0.15907612442970276,
+      "learning_rate": 0.00011802717964186006,
+      "loss": 2.0575,
+      "step": 26800
+    },
+    {
+      "epoch": 2.9554210236653824,
+      "grad_norm": 0.16981548070907593,
+      "learning_rate": 0.00011749393104259143,
+      "loss": 2.0951,
+      "step": 26850
+    },
+    {
+      "epoch": 2.960924600990644,
+      "grad_norm": 0.15169958770275116,
+      "learning_rate": 0.0001169611131501345,
+      "loss": 2.0778,
+      "step": 26900
+    },
+    {
+      "epoch": 2.966428178315905,
+      "grad_norm": 0.1601976752281189,
+      "learning_rate": 0.00011642873302434926,
+      "loss": 2.0621,
+      "step": 26950
+    },
+    {
+      "epoch": 2.9719317556411666,
+      "grad_norm": 0.16251477599143982,
+      "learning_rate": 0.00011589679771929529,
+      "loss": 2.08,
+      "step": 27000
+    },
+    {
+      "epoch": 2.977435332966428,
+      "grad_norm": 0.15681184828281403,
+      "learning_rate": 0.00011536531428313828,
+      "loss": 2.0518,
+      "step": 27050
+    },
+    {
+      "epoch": 2.9829389102916894,
+      "grad_norm": 0.17260442674160004,
+      "learning_rate": 0.00011483428975805664,
+      "loss": 2.0899,
+      "step": 27100
+    },
+    {
+      "epoch": 2.988442487616951,
+      "grad_norm": 0.16616973280906677,
+      "learning_rate": 0.00011430373118014808,
+      "loss": 2.0787,
+      "step": 27150
+    },
+    {
+      "epoch": 2.9939460649422127,
+      "grad_norm": 0.1666904091835022,
+      "learning_rate": 0.00011377364557933658,
+      "loss": 2.0555,
+      "step": 27200
+    },
+    {
+      "epoch": 2.999449642267474,
+      "grad_norm": 0.15540987253189087,
+      "learning_rate": 0.00011324403997927922,
+      "loss": 2.0676,
+      "step": 27250
+    },
+    {
+      "epoch": 3.004953219592735,
+      "grad_norm": 0.16125087440013885,
+      "learning_rate": 0.00011271492139727294,
+      "loss": 2.0144,
+      "step": 27300
+    },
+    {
+      "epoch": 3.0104567969179965,
+      "grad_norm": 0.17550891637802124,
+      "learning_rate": 0.00011218629684416168,
+      "loss": 2.0102,
+      "step": 27350
+    },
+    {
+      "epoch": 3.015960374243258,
+      "grad_norm": 0.16927887499332428,
+      "learning_rate": 0.00011165817332424356,
+      "loss": 1.9816,
+      "step": 27400
+    },
+    {
+      "epoch": 3.0214639515685193,
+      "grad_norm": 0.17522330582141876,
+      "learning_rate": 0.00011113055783517793,
+      "loss": 2.0066,
+      "step": 27450
+    },
+    {
+      "epoch": 3.026967528893781,
+      "grad_norm": 0.160496324300766,
+      "learning_rate": 0.00011060345736789265,
+      "loss": 2.0003,
+      "step": 27500
+    },
+    {
+      "epoch": 3.0324711062190426,
+      "grad_norm": 0.17563533782958984,
+      "learning_rate": 0.00011007687890649163,
+      "loss": 2.0272,
+      "step": 27550
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.17000171542167664,
+      "learning_rate": 0.00010955082942816215,
+      "loss": 2.0447,
+      "step": 27600
+    },
+    {
+      "epoch": 3.0434782608695654,
+      "grad_norm": 0.17253416776657104,
+      "learning_rate": 0.00010902531590308236,
+      "loss": 1.9968,
+      "step": 27650
+    },
+    {
+      "epoch": 3.048981838194827,
+      "grad_norm": 0.16862879693508148,
+      "learning_rate": 0.00010850034529432919,
+      "loss": 1.9952,
+      "step": 27700
+    },
+    {
+      "epoch": 3.0544854155200882,
+      "grad_norm": 0.166275292634964,
+      "learning_rate": 0.00010797592455778562,
+      "loss": 2.0258,
+      "step": 27750
+    },
+    {
+      "epoch": 3.0599889928453496,
+      "grad_norm": 0.1734013706445694,
+      "learning_rate": 0.00010745206064204903,
+      "loss": 2.0047,
+      "step": 27800
+    },
+    {
+      "epoch": 3.065492570170611,
+      "grad_norm": 0.1760258674621582,
+      "learning_rate": 0.00010692876048833889,
+      "loss": 1.997,
+      "step": 27850
+    },
+    {
+      "epoch": 3.0709961474958725,
+      "grad_norm": 0.1616799384355545,
+      "learning_rate": 0.00010640603103040464,
+      "loss": 1.9817,
+      "step": 27900
+    },
+    {
+      "epoch": 3.076499724821134,
+      "grad_norm": 0.1775207221508026,
+      "learning_rate": 0.00010588387919443416,
+      "loss": 2.0391,
+      "step": 27950
+    },
+    {
+      "epoch": 3.0820033021463953,
+      "grad_norm": 0.17136207222938538,
+      "learning_rate": 0.00010536231189896175,
+      "loss": 1.999,
+      "step": 28000
+    },
+    {
+      "epoch": 3.0875068794716567,
+      "grad_norm": 0.1694386601448059,
+      "learning_rate": 0.00010484133605477644,
+      "loss": 2.0153,
+      "step": 28050
+    },
+    {
+      "epoch": 3.093010456796918,
+      "grad_norm": 0.17021538317203522,
+      "learning_rate": 0.00010432095856483057,
+      "loss": 2.0411,
+      "step": 28100
+    },
+    {
+      "epoch": 3.0985140341221795,
+      "grad_norm": 0.17548739910125732,
+      "learning_rate": 0.00010380118632414833,
+      "loss": 2.0416,
+      "step": 28150
+    },
+    {
+      "epoch": 3.104017611447441,
+      "grad_norm": 0.1785077303647995,
+      "learning_rate": 0.0001032820262197342,
+      "loss": 2.032,
+      "step": 28200
+    },
+    {
+      "epoch": 3.1095211887727023,
+      "grad_norm": 0.17226973176002502,
+      "learning_rate": 0.00010276348513048193,
+      "loss": 1.986,
+      "step": 28250
+    },
+    {
+      "epoch": 3.1150247660979637,
+      "grad_norm": 0.18131954967975616,
+      "learning_rate": 0.00010224556992708328,
+      "loss": 2.0149,
+      "step": 28300
+    },
+    {
+      "epoch": 3.120528343423225,
+      "grad_norm": 0.16980765759944916,
+      "learning_rate": 0.00010172828747193689,
+      "loss": 2.0309,
+      "step": 28350
+    },
+    {
+      "epoch": 3.1260319207484866,
+      "grad_norm": 0.16887415945529938,
+      "learning_rate": 0.00010121164461905759,
+      "loss": 1.9814,
+      "step": 28400
+    },
+    {
+      "epoch": 3.131535498073748,
+      "grad_norm": 0.18381917476654053,
+      "learning_rate": 0.00010069564821398536,
+      "loss": 2.0228,
+      "step": 28450
+    },
+    {
+      "epoch": 3.1370390753990094,
+      "grad_norm": 0.18095500767230988,
+      "learning_rate": 0.0001001803050936948,
+      "loss": 2.0197,
+      "step": 28500
+    },
+    {
+      "epoch": 3.142542652724271,
+      "grad_norm": 0.15869130194187164,
+      "learning_rate": 9.966562208650441e-05,
+      "loss": 2.0114,
+      "step": 28550
+    },
+    {
+      "epoch": 3.148046230049532,
+      "grad_norm": 0.17333059012889862,
+      "learning_rate": 9.915160601198616e-05,
+      "loss": 2.0025,
+      "step": 28600
+    },
+    {
+      "epoch": 3.1535498073747936,
+      "grad_norm": 0.161941796541214,
+      "learning_rate": 9.863826368087516e-05,
+      "loss": 2.0273,
+      "step": 28650
+    },
+    {
+      "epoch": 3.159053384700055,
+      "grad_norm": 0.16833285987377167,
+      "learning_rate": 9.812560189497935e-05,
+      "loss": 2.0108,
+      "step": 28700
+    },
+    {
+      "epoch": 3.1645569620253164,
+      "grad_norm": 0.15908978879451752,
+      "learning_rate": 9.76136274470895e-05,
+      "loss": 1.9852,
+      "step": 28750
+    },
+    {
+      "epoch": 3.170060539350578,
+      "grad_norm": 0.1724586933851242,
+      "learning_rate": 9.710234712088896e-05,
+      "loss": 2.0297,
+      "step": 28800
+    },
+    {
+      "epoch": 3.1755641166758393,
+      "grad_norm": 0.17113400995731354,
+      "learning_rate": 9.659176769086419e-05,
+      "loss": 1.9969,
+      "step": 28850
+    },
+    {
+      "epoch": 3.1810676940011007,
+      "grad_norm": 0.16423186659812927,
+      "learning_rate": 9.608189592221446e-05,
+      "loss": 2.0106,
+      "step": 28900
+    },
+    {
+      "epoch": 3.186571271326362,
+      "grad_norm": 0.16140978038311005,
+      "learning_rate": 9.557273857076271e-05,
+      "loss": 2.0339,
+      "step": 28950
+    },
+    {
+      "epoch": 3.1920748486516235,
+      "grad_norm": 0.1718360334634781,
+      "learning_rate": 9.506430238286566e-05,
+      "loss": 2.04,
+      "step": 29000
+    },
+    {
+      "epoch": 3.197578425976885,
+      "grad_norm": 0.164507195353508,
+      "learning_rate": 9.455659409532474e-05,
+      "loss": 2.0168,
+      "step": 29050
+    },
+    {
+      "epoch": 3.2030820033021463,
+      "grad_norm": 0.17989708483219147,
+      "learning_rate": 9.40496204352966e-05,
+      "loss": 2.0273,
+      "step": 29100
+    },
+    {
+      "epoch": 3.2085855806274077,
+      "grad_norm": 0.17319265007972717,
+      "learning_rate": 9.354338812020402e-05,
+      "loss": 1.9911,
+      "step": 29150
+    },
+    {
+      "epoch": 3.214089157952669,
+      "grad_norm": 0.17870402336120605,
+      "learning_rate": 9.303790385764694e-05,
+      "loss": 2.0103,
+      "step": 29200
+    },
+    {
+      "epoch": 3.2195927352779306,
+      "grad_norm": 0.1797982007265091,
+      "learning_rate": 9.253317434531357e-05,
+      "loss": 2.0169,
+      "step": 29250
+    },
+    {
+      "epoch": 3.225096312603192,
+      "grad_norm": 0.15193064510822296,
+      "learning_rate": 9.202920627089171e-05,
+      "loss": 2.0116,
+      "step": 29300
+    },
+    {
+      "epoch": 3.2305998899284534,
+      "grad_norm": 0.16679364442825317,
+      "learning_rate": 9.152600631197996e-05,
+      "loss": 2.0362,
+      "step": 29350
+    },
+    {
+      "epoch": 3.236103467253715,
+      "grad_norm": 0.17053711414337158,
+      "learning_rate": 9.102358113599942e-05,
+      "loss": 1.9829,
+      "step": 29400
+    },
+    {
+      "epoch": 3.241607044578976,
+      "grad_norm": 0.16923174262046814,
+      "learning_rate": 9.052193740010536e-05,
+      "loss": 2.0011,
+      "step": 29450
+    },
+    {
+      "epoch": 3.2471106219042376,
+      "grad_norm": 0.18323257565498352,
+      "learning_rate": 9.002108175109877e-05,
+      "loss": 1.9784,
+      "step": 29500
+    },
+    {
+      "epoch": 3.252614199229499,
+      "grad_norm": 0.16737329959869385,
+      "learning_rate": 8.952102082533855e-05,
+      "loss": 2.0276,
+      "step": 29550
+    },
+    {
+      "epoch": 3.2581177765547604,
+      "grad_norm": 0.17038094997406006,
+      "learning_rate": 8.902176124865348e-05,
+      "loss": 2.0315,
+      "step": 29600
+    },
+    {
+      "epoch": 3.263621353880022,
+      "grad_norm": 0.16778729856014252,
+      "learning_rate": 8.852330963625444e-05,
+      "loss": 2.0139,
+      "step": 29650
+    },
+    {
+      "epoch": 3.2691249312052832,
+      "grad_norm": 0.18061307072639465,
+      "learning_rate": 8.802567259264674e-05,
+      "loss": 2.002,
+      "step": 29700
+    },
+    {
+      "epoch": 3.2746285085305447,
+      "grad_norm": 0.17191821336746216,
+      "learning_rate": 8.752885671154253e-05,
+      "loss": 2.0218,
+      "step": 29750
+    },
+    {
+      "epoch": 3.280132085855806,
+      "grad_norm": 0.17957885563373566,
+      "learning_rate": 8.703286857577378e-05,
+      "loss": 2.0217,
+      "step": 29800
+    },
+    {
+      "epoch": 3.2856356631810675,
+      "grad_norm": 0.16803069412708282,
+      "learning_rate": 8.653771475720442e-05,
+      "loss": 2.0185,
+      "step": 29850
+    },
+    {
+      "epoch": 3.291139240506329,
+      "grad_norm": 0.1788654327392578,
+      "learning_rate": 8.604340181664395e-05,
+      "loss": 1.9973,
+      "step": 29900
+    },
+    {
+      "epoch": 3.2966428178315903,
+      "grad_norm": 0.17996376752853394,
+      "learning_rate": 8.554993630376006e-05,
+      "loss": 2.0161,
+      "step": 29950
+    },
+    {
+      "epoch": 3.3021463951568517,
+      "grad_norm": 0.18701910972595215,
+      "learning_rate": 8.50573247569921e-05,
+      "loss": 2.0156,
+      "step": 30000
+    },
+    {
+      "epoch": 3.3076499724821136,
+      "grad_norm": 0.1713022142648697,
+      "learning_rate": 8.45655737034641e-05,
+      "loss": 2.0144,
+      "step": 30050
+    },
+    {
+      "epoch": 3.313153549807375,
+      "grad_norm": 0.17666327953338623,
+      "learning_rate": 8.407468965889878e-05,
+      "loss": 2.0116,
+      "step": 30100
+    },
+    {
+      "epoch": 3.3186571271326364,
+      "grad_norm": 0.18481209874153137,
+      "learning_rate": 8.358467912753079e-05,
+      "loss": 2.0149,
+      "step": 30150
+    },
+    {
+      "epoch": 3.324160704457898,
+      "grad_norm": 0.17895914614200592,
+      "learning_rate": 8.309554860202082e-05,
+      "loss": 1.9913,
+      "step": 30200
+    },
+    {
+      "epoch": 3.329664281783159,
+      "grad_norm": 0.18427938222885132,
+      "learning_rate": 8.260730456336925e-05,
+      "loss": 1.9913,
+      "step": 30250
+    },
+    {
+      "epoch": 3.3351678591084206,
+      "grad_norm": 0.1806970238685608,
+      "learning_rate": 8.211995348083067e-05,
+      "loss": 2.0177,
+      "step": 30300
+    },
+    {
+      "epoch": 3.340671436433682,
+      "grad_norm": 0.1735043078660965,
+      "learning_rate": 8.163350181182787e-05,
+      "loss": 2.0275,
+      "step": 30350
+    },
+    {
+      "epoch": 3.3461750137589434,
+      "grad_norm": 0.17593735456466675,
+      "learning_rate": 8.11479560018664e-05,
+      "loss": 2.0525,
+      "step": 30400
+    },
+    {
+      "epoch": 3.351678591084205,
+      "grad_norm": 0.1755988448858261,
+      "learning_rate": 8.066332248444908e-05,
+      "loss": 2.0182,
+      "step": 30450
+    },
+    {
+      "epoch": 3.3571821684094663,
+      "grad_norm": 0.15805041790008545,
+      "learning_rate": 8.017960768099091e-05,
+      "loss": 1.9748,
+      "step": 30500
+    },
+    {
+      "epoch": 3.3626857457347277,
+      "grad_norm": 0.18487168848514557,
+      "learning_rate": 7.969681800073386e-05,
+      "loss": 2.021,
+      "step": 30550
+    },
+    {
+      "epoch": 3.368189323059989,
+      "grad_norm": 0.17120800912380219,
+      "learning_rate": 7.921495984066194e-05,
+      "loss": 2.0113,
+      "step": 30600
+    },
+    {
+      "epoch": 3.3736929003852505,
+      "grad_norm": 0.1721143126487732,
+      "learning_rate": 7.87340395854164e-05,
+      "loss": 1.9936,
+      "step": 30650
+    },
+    {
+      "epoch": 3.379196477710512,
+      "grad_norm": 0.17595061659812927,
+      "learning_rate": 7.825406360721139e-05,
+      "loss": 2.0044,
+      "step": 30700
+    },
+    {
+      "epoch": 3.3847000550357733,
+      "grad_norm": 0.16751642525196075,
+      "learning_rate": 7.777503826574925e-05,
+      "loss": 1.9988,
+      "step": 30750
+    },
+    {
+      "epoch": 3.3902036323610347,
+      "grad_norm": 0.17864260077476501,
+      "learning_rate": 7.729696990813634e-05,
+      "loss": 2.0189,
+      "step": 30800
+    },
+    {
+      "epoch": 3.395707209686296,
+      "grad_norm": 0.1791946142911911,
+      "learning_rate": 7.681986486879898e-05,
+      "loss": 2.0038,
+      "step": 30850
+    },
+    {
+      "epoch": 3.4012107870115575,
+      "grad_norm": 0.1704343855381012,
+      "learning_rate": 7.634372946939945e-05,
+      "loss": 1.9625,
+      "step": 30900
+    },
+    {
+      "epoch": 3.406714364336819,
+      "grad_norm": 0.17794691026210785,
+      "learning_rate": 7.586857001875235e-05,
+      "loss": 2.0017,
+      "step": 30950
+    },
+    {
+      "epoch": 3.4122179416620804,
+      "grad_norm": 0.17922665178775787,
+      "learning_rate": 7.539439281274071e-05,
+      "loss": 2.054,
+      "step": 31000
+    },
+    {
+      "epoch": 3.4177215189873418,
+      "grad_norm": 0.174763485789299,
+      "learning_rate": 7.492120413423296e-05,
+      "loss": 2.0185,
+      "step": 31050
+    },
+    {
+      "epoch": 3.423225096312603,
+      "grad_norm": 0.1687181144952774,
+      "learning_rate": 7.444901025299941e-05,
+      "loss": 2.0463,
+      "step": 31100
+    },
+    {
+      "epoch": 3.4287286736378646,
+      "grad_norm": 0.17793361842632294,
+      "learning_rate": 7.397781742562941e-05,
+      "loss": 2.0296,
+      "step": 31150
+    },
+    {
+      "epoch": 3.434232250963126,
+      "grad_norm": 0.16858656704425812,
+      "learning_rate": 7.350763189544803e-05,
+      "loss": 1.9916,
+      "step": 31200
+    },
+    {
+      "epoch": 3.4397358282883874,
+      "grad_norm": 0.16184011101722717,
+      "learning_rate": 7.303845989243384e-05,
+      "loss": 2.0314,
+      "step": 31250
+    },
+    {
+      "epoch": 3.445239405613649,
+      "grad_norm": 0.1733245998620987,
+      "learning_rate": 7.257030763313607e-05,
+      "loss": 2.0125,
+      "step": 31300
+    },
+    {
+      "epoch": 3.4507429829389102,
+      "grad_norm": 0.18471761047840118,
+      "learning_rate": 7.210318132059231e-05,
+      "loss": 1.9691,
+      "step": 31350
+    },
+    {
+      "epoch": 3.4562465602641717,
+      "grad_norm": 0.17260605096817017,
+      "learning_rate": 7.16370871442462e-05,
+      "loss": 2.0113,
+      "step": 31400
+    },
+    {
+      "epoch": 3.461750137589433,
+      "grad_norm": 0.1758509874343872,
+      "learning_rate": 7.117203127986569e-05,
+      "loss": 2.0239,
+      "step": 31450
+    },
+    {
+      "epoch": 3.4672537149146945,
+      "grad_norm": 0.18456335365772247,
+      "learning_rate": 7.070801988946098e-05,
+      "loss": 1.999,
+      "step": 31500
+    },
+    {
+      "epoch": 3.472757292239956,
+      "grad_norm": 0.17995281517505646,
+      "learning_rate": 7.024505912120293e-05,
+      "loss": 2.0141,
+      "step": 31550
+    },
+    {
+      "epoch": 3.4782608695652173,
+      "grad_norm": 0.16747882962226868,
+      "learning_rate": 6.978315510934164e-05,
+      "loss": 2.0073,
+      "step": 31600
+    },
+    {
+      "epoch": 3.4837644468904787,
+      "grad_norm": 0.1658480316400528,
+      "learning_rate": 6.932231397412522e-05,
+      "loss": 2.0079,
+      "step": 31650
+    },
+    {
+      "epoch": 3.48926802421574,
+      "grad_norm": 0.17987202107906342,
+      "learning_rate": 6.886254182171836e-05,
+      "loss": 2.0439,
+      "step": 31700
+    },
+    {
+      "epoch": 3.4947716015410015,
+      "grad_norm": 0.1672779768705368,
+      "learning_rate": 6.840384474412201e-05,
+      "loss": 1.9878,
+      "step": 31750
+    },
+    {
+      "epoch": 3.500275178866263,
+      "grad_norm": 0.1716667115688324,
+      "learning_rate": 6.794622881909207e-05,
+      "loss": 2.0248,
+      "step": 31800
+    },
+    {
+      "epoch": 3.5057787561915243,
+      "grad_norm": 0.17986662685871124,
+      "learning_rate": 6.748970011005924e-05,
+      "loss": 1.99,
+      "step": 31850
+    },
+    {
+      "epoch": 3.5112823335167858,
+      "grad_norm": 0.1808827668428421,
+      "learning_rate": 6.703426466604858e-05,
+      "loss": 1.9692,
+      "step": 31900
+    },
+    {
+      "epoch": 3.516785910842047,
+      "grad_norm": 0.17262905836105347,
+      "learning_rate": 6.65799285215993e-05,
+      "loss": 1.9636,
+      "step": 31950
+    },
+    {
+      "epoch": 3.522289488167309,
+      "grad_norm": 0.16857630014419556,
+      "learning_rate": 6.612669769668489e-05,
+      "loss": 1.9923,
+      "step": 32000
+    },
+    {
+      "epoch": 3.5277930654925704,
+      "grad_norm": 0.17498353123664856,
+      "learning_rate": 6.567457819663327e-05,
+      "loss": 2.0248,
+      "step": 32050
+    },
+    {
+      "epoch": 3.533296642817832,
+      "grad_norm": 0.16948480904102325,
+      "learning_rate": 6.522357601204719e-05,
+      "loss": 1.9966,
+      "step": 32100
+    },
+    {
+      "epoch": 3.5388002201430933,
+      "grad_norm": 0.1695474237203598,
+      "learning_rate": 6.477369711872502e-05,
+      "loss": 1.994,
+      "step": 32150
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.1761193871498108,
+      "learning_rate": 6.432494747758146e-05,
+      "loss": 1.9827,
+      "step": 32200
+    },
+    {
+      "epoch": 3.549807374793616,
+      "grad_norm": 0.17206500470638275,
+      "learning_rate": 6.387733303456858e-05,
+      "loss": 1.9944,
+      "step": 32250
+    },
+    {
+      "epoch": 3.5553109521188775,
+      "grad_norm": 0.19551701843738556,
+      "learning_rate": 6.3430859720597e-05,
+      "loss": 2.0118,
+      "step": 32300
+    },
+    {
+      "epoch": 3.560814529444139,
+      "grad_norm": 0.17827004194259644,
+      "learning_rate": 6.298553345145728e-05,
+      "loss": 1.9953,
+      "step": 32350
+    },
+    {
+      "epoch": 3.5663181067694003,
+      "grad_norm": 0.17359542846679688,
+      "learning_rate": 6.254136012774166e-05,
+      "loss": 2.0098,
+      "step": 32400
+    },
+    {
+      "epoch": 3.5718216840946617,
+      "grad_norm": 0.16225001215934753,
+      "learning_rate": 6.209834563476578e-05,
+      "loss": 2.008,
+      "step": 32450
+    },
+    {
+      "epoch": 3.577325261419923,
+      "grad_norm": 0.17888249456882477,
+      "learning_rate": 6.165649584249079e-05,
+      "loss": 1.9962,
+      "step": 32500
+    },
+    {
+      "epoch": 3.5828288387451845,
+      "grad_norm": 0.16299229860305786,
+      "learning_rate": 6.121581660544532e-05,
+      "loss": 2.0387,
+      "step": 32550
+    },
+    {
+      "epoch": 3.588332416070446,
+      "grad_norm": 0.16753186285495758,
+      "learning_rate": 6.077631376264832e-05,
+      "loss": 1.9662,
+      "step": 32600
+    },
+    {
+      "epoch": 3.5938359933957074,
+      "grad_norm": 0.18166500329971313,
+      "learning_rate": 6.033799313753129e-05,
+      "loss": 1.997,
+      "step": 32650
+    },
+    {
+      "epoch": 3.5993395707209688,
+      "grad_norm": 0.16691668331623077,
+      "learning_rate": 5.990086053786139e-05,
+      "loss": 2.0072,
+      "step": 32700
+    },
+    {
+      "epoch": 3.60484314804623,
+      "grad_norm": 0.18671706318855286,
+      "learning_rate": 5.946492175566438e-05,
+      "loss": 2.0217,
+      "step": 32750
+    },
+    {
+      "epoch": 3.6103467253714916,
+      "grad_norm": 0.16831424832344055,
+      "learning_rate": 5.9030182567147905e-05,
+      "loss": 2.0024,
+      "step": 32800
+    },
+    {
+      "epoch": 3.615850302696753,
+      "grad_norm": 0.17562079429626465,
+      "learning_rate": 5.859664873262483e-05,
+      "loss": 2.0351,
+      "step": 32850
+    },
+    {
+      "epoch": 3.6213538800220144,
+      "grad_norm": 0.18190255761146545,
+      "learning_rate": 5.816432599643719e-05,
+      "loss": 2.0117,
+      "step": 32900
+    },
+    {
+      "epoch": 3.626857457347276,
+      "grad_norm": 0.1830035150051117,
+      "learning_rate": 5.773322008687972e-05,
+      "loss": 2.012,
+      "step": 32950
+    },
+    {
+      "epoch": 3.6323610346725372,
+      "grad_norm": 0.17998354136943817,
+      "learning_rate": 5.730333671612435e-05,
+      "loss": 1.9977,
+      "step": 33000
+    },
+    {
+      "epoch": 3.6378646119977986,
+      "grad_norm": 0.1839103251695633,
+      "learning_rate": 5.687468158014421e-05,
+      "loss": 2.003,
+      "step": 33050
+    },
+    {
+      "epoch": 3.64336818932306,
+      "grad_norm": 0.18631495535373688,
+      "learning_rate": 5.6447260358638285e-05,
+      "loss": 2.0072,
+      "step": 33100
+    },
+    {
+      "epoch": 3.6488717666483215,
+      "grad_norm": 0.19609171152114868,
+      "learning_rate": 5.602107871495615e-05,
+      "loss": 1.9912,
+      "step": 33150
+    },
+    {
+      "epoch": 3.654375343973583,
+      "grad_norm": 0.16935598850250244,
+      "learning_rate": 5.5596142296022954e-05,
+      "loss": 2.0097,
+      "step": 33200
+    },
+    {
+      "epoch": 3.6598789212988443,
+      "grad_norm": 0.16973347961902618,
+      "learning_rate": 5.5172456732264435e-05,
+      "loss": 2.0151,
+      "step": 33250
+    },
+    {
+      "epoch": 3.6653824986241057,
+      "grad_norm": 0.17289844155311584,
+      "learning_rate": 5.475002763753257e-05,
+      "loss": 2.0393,
+      "step": 33300
+    },
+    {
+      "epoch": 3.670886075949367,
+      "grad_norm": 0.1864277720451355,
+      "learning_rate": 5.432886060903105e-05,
+      "loss": 2.0069,
+      "step": 33350
+    },
+    {
+      "epoch": 3.6763896532746285,
+      "grad_norm": 0.17381645739078522,
+      "learning_rate": 5.390896122724111e-05,
+      "loss": 2.0323,
+      "step": 33400
+    },
+    {
+      "epoch": 3.68189323059989,
+      "grad_norm": 0.18182097375392914,
+      "learning_rate": 5.349033505584767e-05,
+      "loss": 2.0344,
+      "step": 33450
+    },
+    {
+      "epoch": 3.6873968079251513,
+      "grad_norm": 0.17326848208904266,
+      "learning_rate": 5.3072987641665373e-05,
+      "loss": 1.9888,
+      "step": 33500
+    },
+    {
+      "epoch": 3.6929003852504128,
+      "grad_norm": 0.17132443189620972,
+      "learning_rate": 5.2656924514565515e-05,
+      "loss": 1.9741,
+      "step": 33550
+    },
+    {
+      "epoch": 3.698403962575674,
+      "grad_norm": 0.17236609756946564,
+      "learning_rate": 5.224215118740243e-05,
+      "loss": 1.9967,
+      "step": 33600
+    },
+    {
+      "epoch": 3.7039075399009356,
+      "grad_norm": 0.16628102958202362,
+      "learning_rate": 5.182867315594046e-05,
+      "loss": 2.0036,
+      "step": 33650
+    },
+    {
+      "epoch": 3.709411117226197,
+      "grad_norm": 0.17410129308700562,
+      "learning_rate": 5.141649589878134e-05,
+      "loss": 2.0173,
+      "step": 33700
+    },
+    {
+      "epoch": 3.7149146945514584,
+      "grad_norm": 0.1739521622657776,
+      "learning_rate": 5.100562487729148e-05,
+      "loss": 1.9977,
+      "step": 33750
+    },
+    {
+      "epoch": 3.72041827187672,
+      "grad_norm": 0.17988687753677368,
+      "learning_rate": 5.059606553552956e-05,
+      "loss": 1.9964,
+      "step": 33800
+    },
+    {
+      "epoch": 3.725921849201981,
+      "grad_norm": 0.17351609468460083,
+      "learning_rate": 5.018782330017448e-05,
+      "loss": 2.0078,
+      "step": 33850
+    },
+    {
+      "epoch": 3.7314254265272426,
+      "grad_norm": 0.176718607544899,
+      "learning_rate": 4.9780903580453435e-05,
+      "loss": 2.0036,
+      "step": 33900
+    },
+    {
+      "epoch": 3.736929003852504,
+      "grad_norm": 0.17504329979419708,
+      "learning_rate": 4.937531176807023e-05,
+      "loss": 1.9837,
+      "step": 33950
+    },
+    {
+      "epoch": 3.7424325811777654,
+      "grad_norm": 0.179831862449646,
+      "learning_rate": 4.897105323713374e-05,
+      "loss": 1.999,
+      "step": 34000
+    },
+    {
+      "epoch": 3.747936158503027,
+      "grad_norm": 0.17752991616725922,
+      "learning_rate": 4.8568133344086986e-05,
+      "loss": 2.0194,
+      "step": 34050
+    },
+    {
+      "epoch": 3.7534397358282883,
+      "grad_norm": 0.17888140678405762,
+      "learning_rate": 4.8166557427635774e-05,
+      "loss": 2.0108,
+      "step": 34100
+    },
+    {
+      "epoch": 3.7589433131535497,
+      "grad_norm": 0.17585836350917816,
+      "learning_rate": 4.776633080867834e-05,
+      "loss": 2.0421,
+      "step": 34150
+    },
+    {
+      "epoch": 3.764446890478811,
+      "grad_norm": 0.16860149800777435,
+      "learning_rate": 4.736745879023457e-05,
+      "loss": 2.0126,
+      "step": 34200
+    },
+    {
+      "epoch": 3.7699504678040725,
+      "grad_norm": 0.17601364850997925,
+      "learning_rate": 4.6969946657375865e-05,
+      "loss": 2.0127,
+      "step": 34250
+    },
+    {
+      "epoch": 3.775454045129334,
+      "grad_norm": 0.19177961349487305,
+      "learning_rate": 4.657379967715511e-05,
+      "loss": 2.0234,
+      "step": 34300
+    },
+    {
+      "epoch": 3.7809576224545953,
+      "grad_norm": 0.19012008607387543,
+      "learning_rate": 4.617902309853685e-05,
+      "loss": 1.976,
+      "step": 34350
+    },
+    {
+      "epoch": 3.7864611997798567,
+      "grad_norm": 0.18448469042778015,
+      "learning_rate": 4.578562215232766e-05,
+      "loss": 1.9897,
+      "step": 34400
+    },
+    {
+      "epoch": 3.791964777105118,
+      "grad_norm": 0.18167300522327423,
+      "learning_rate": 4.539360205110701e-05,
+      "loss": 2.0088,
+      "step": 34450
+    },
+    {
+      "epoch": 3.7974683544303796,
+      "grad_norm": 0.1806318610906601,
+      "learning_rate": 4.500296798915813e-05,
+      "loss": 1.9949,
+      "step": 34500
+    },
+    {
+      "epoch": 3.802971931755641,
+      "grad_norm": 0.17676065862178802,
+      "learning_rate": 4.461372514239911e-05,
+      "loss": 1.9966,
+      "step": 34550
+    },
+    {
+      "epoch": 3.8084755090809024,
+      "grad_norm": 0.16621781885623932,
+      "learning_rate": 4.422587866831446e-05,
+      "loss": 1.9952,
+      "step": 34600
+    },
+    {
+      "epoch": 3.813979086406164,
+      "grad_norm": 0.16662217676639557,
+      "learning_rate": 4.3839433705886646e-05,
+      "loss": 2.0032,
+      "step": 34650
+    },
+    {
+      "epoch": 3.819482663731425,
+      "grad_norm": 0.15986546874046326,
+      "learning_rate": 4.3454395375527966e-05,
+      "loss": 1.9974,
+      "step": 34700
+    },
+    {
+      "epoch": 3.8249862410566866,
+      "grad_norm": 0.18636077642440796,
+      "learning_rate": 4.307076877901293e-05,
+      "loss": 1.9959,
+      "step": 34750
+    },
+    {
+      "epoch": 3.830489818381948,
+      "grad_norm": 0.17939697206020355,
+      "learning_rate": 4.268855899941039e-05,
+      "loss": 1.9923,
+      "step": 34800
+    },
+    {
+      "epoch": 3.8359933957072094,
+      "grad_norm": 0.1743779331445694,
+      "learning_rate": 4.2307771101016366e-05,
+      "loss": 2.0055,
+      "step": 34850
+    },
+    {
+      "epoch": 3.841496973032471,
+      "grad_norm": 0.1767256259918213,
+      "learning_rate": 4.192841012928691e-05,
+      "loss": 1.9834,
+      "step": 34900
+    },
+    {
+      "epoch": 3.8470005503577323,
+      "grad_norm": 0.1775825172662735,
+      "learning_rate": 4.1550481110771224e-05,
+      "loss": 2.01,
+      "step": 34950
+    },
+    {
+      "epoch": 3.852504127682994,
+      "grad_norm": 0.1638861745595932,
+      "learning_rate": 4.1173989053045024e-05,
+      "loss": 2.0001,
+      "step": 35000
+    },
+    {
+      "epoch": 3.8580077050082555,
+      "grad_norm": 0.1761082410812378,
+      "learning_rate": 4.079893894464429e-05,
+      "loss": 2.0041,
+      "step": 35050
+    },
+    {
+      "epoch": 3.863511282333517,
+      "grad_norm": 0.17102594673633575,
+      "learning_rate": 4.042533575499914e-05,
+      "loss": 2.003,
+      "step": 35100
+    },
+    {
+      "epoch": 3.8690148596587783,
+      "grad_norm": 0.19267229735851288,
+      "learning_rate": 4.005318443436779e-05,
+      "loss": 1.973,
+      "step": 35150
+    },
+    {
+      "epoch": 3.8745184369840397,
+      "grad_norm": 0.16570790112018585,
+      "learning_rate": 3.968248991377137e-05,
+      "loss": 1.9834,
+      "step": 35200
+    },
+    {
+      "epoch": 3.880022014309301,
+      "grad_norm": 0.16816405951976776,
+      "learning_rate": 3.9313257104928094e-05,
+      "loss": 2.0011,
+      "step": 35250
+    },
+    {
+      "epoch": 3.8855255916345626,
+      "grad_norm": 0.16395322978496552,
+      "learning_rate": 3.8945490900188616e-05,
+      "loss": 1.9771,
+      "step": 35300
+    },
+    {
+      "epoch": 3.891029168959824,
+      "grad_norm": 0.17911794781684875,
+      "learning_rate": 3.857919617247097e-05,
+      "loss": 2.0,
+      "step": 35350
+    },
+    {
+      "epoch": 3.8965327462850854,
+      "grad_norm": 0.18344633281230927,
+      "learning_rate": 3.821437777519607e-05,
+      "loss": 1.9952,
+      "step": 35400
+    },
+    {
+      "epoch": 3.902036323610347,
+      "grad_norm": 0.1782706081867218,
+      "learning_rate": 3.785104054222334e-05,
+      "loss": 1.9952,
+      "step": 35450
+    },
+    {
+      "epoch": 3.907539900935608,
+      "grad_norm": 0.17840951681137085,
+      "learning_rate": 3.74891892877868e-05,
+      "loss": 1.9994,
+      "step": 35500
+    },
+    {
+      "epoch": 3.9130434782608696,
+      "grad_norm": 0.19014611840248108,
+      "learning_rate": 3.7128828806431046e-05,
+      "loss": 2.0023,
+      "step": 35550
+    },
+    {
+      "epoch": 3.918547055586131,
+      "grad_norm": 0.17002084851264954,
+      "learning_rate": 3.6769963872947996e-05,
+      "loss": 2.0068,
+      "step": 35600
+    },
+    {
+      "epoch": 3.9240506329113924,
+      "grad_norm": 0.16855107247829437,
+      "learning_rate": 3.641259924231345e-05,
+      "loss": 1.9847,
+      "step": 35650
+    },
+    {
+      "epoch": 3.929554210236654,
+      "grad_norm": 0.190132737159729,
+      "learning_rate": 3.605673964962414e-05,
+      "loss": 1.9955,
+      "step": 35700
+    },
+    {
+      "epoch": 3.9350577875619153,
+      "grad_norm": 0.1558249294757843,
+      "learning_rate": 3.570238981003498e-05,
+      "loss": 1.9861,
+      "step": 35750
+    },
+    {
+      "epoch": 3.9405613648871767,
+      "grad_norm": 0.18199962377548218,
+      "learning_rate": 3.53495544186966e-05,
+      "loss": 2.0136,
+      "step": 35800
+    },
+    {
+      "epoch": 3.946064942212438,
+      "grad_norm": 0.17813271284103394,
+      "learning_rate": 3.499823815069306e-05,
+      "loss": 2.0012,
+      "step": 35850
+    },
+    {
+      "epoch": 3.9515685195376995,
+      "grad_norm": 0.18256065249443054,
+      "learning_rate": 3.464844566098008e-05,
+      "loss": 1.9892,
+      "step": 35900
+    },
+    {
+      "epoch": 3.957072096862961,
+      "grad_norm": 0.1784873753786087,
+      "learning_rate": 3.4300181584323126e-05,
+      "loss": 1.9983,
+      "step": 35950
+    },
+    {
+      "epoch": 3.9625756741882223,
+      "grad_norm": 0.18530665338039398,
+      "learning_rate": 3.3953450535236226e-05,
+      "loss": 2.0357,
+      "step": 36000
+    },
+    {
+      "epoch": 3.9680792515134837,
+      "grad_norm": 0.17063435912132263,
+      "learning_rate": 3.360825710792068e-05,
+      "loss": 2.0029,
+      "step": 36050
+    },
+    {
+      "epoch": 3.973582828838745,
+      "grad_norm": 0.1718268245458603,
+      "learning_rate": 3.326460587620427e-05,
+      "loss": 1.9762,
+      "step": 36100
+    },
+    {
+      "epoch": 3.9790864061640066,
+      "grad_norm": 0.15901614725589752,
+      "learning_rate": 3.292250139348057e-05,
+      "loss": 1.9952,
+      "step": 36150
+    },
+    {
+      "epoch": 3.984589983489268,
+      "grad_norm": 0.18968409299850464,
+      "learning_rate": 3.258194819264871e-05,
+      "loss": 2.0127,
+      "step": 36200
+    },
+    {
+      "epoch": 3.9900935608145294,
+      "grad_norm": 0.18861430883407593,
+      "learning_rate": 3.2242950786053296e-05,
+      "loss": 1.9981,
+      "step": 36250
+    },
+    {
+      "epoch": 3.995597138139791,
+      "grad_norm": 0.1702040284872055,
+      "learning_rate": 3.1905513665424445e-05,
+      "loss": 2.0023,
+      "step": 36300
+    },
+    {
+      "epoch": 4.001100715465053,
+      "grad_norm": 0.18254053592681885,
+      "learning_rate": 3.156964130181859e-05,
+      "loss": 2.0038,
+      "step": 36350
+    },
+    {
+      "epoch": 4.006604292790314,
+      "grad_norm": 0.18867474794387817,
+      "learning_rate": 3.1235338145558925e-05,
+      "loss": 1.9139,
+      "step": 36400
+    },
+    {
+      "epoch": 4.0121078701155755,
+      "grad_norm": 0.1872694045305252,
+      "learning_rate": 3.090260862617671e-05,
+      "loss": 1.9484,
+      "step": 36450
+    },
+    {
+      "epoch": 4.017611447440837,
+      "grad_norm": 0.17595593631267548,
+      "learning_rate": 3.057145715235236e-05,
+      "loss": 1.9543,
+      "step": 36500
+    },
+    {
+      "epoch": 4.023115024766098,
+      "grad_norm": 0.17811767756938934,
+      "learning_rate": 3.0241888111857145e-05,
+      "loss": 1.9268,
+      "step": 36550
+    },
+    {
+      "epoch": 4.02861860209136,
+      "grad_norm": 0.17598308622837067,
+      "learning_rate": 2.9913905871495037e-05,
+      "loss": 1.9439,
+      "step": 36600
+    },
+    {
+      "epoch": 4.034122179416621,
+      "grad_norm": 0.19237573444843292,
+      "learning_rate": 2.9587514777044874e-05,
+      "loss": 1.9044,
+      "step": 36650
+    },
+    {
+      "epoch": 4.0396257567418825,
+      "grad_norm": 0.1873626410961151,
+      "learning_rate": 2.926271915320259e-05,
+      "loss": 1.943,
+      "step": 36700
+    },
+    {
+      "epoch": 4.045129334067144,
+      "grad_norm": 0.2001618891954422,
+      "learning_rate": 2.893952330352423e-05,
+      "loss": 1.9311,
+      "step": 36750
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 0.1838996410369873,
+      "learning_rate": 2.86179315103687e-05,
+      "loss": 1.9373,
+      "step": 36800
+    },
+    {
+      "epoch": 4.056136488717667,
+      "grad_norm": 0.16863055527210236,
+      "learning_rate": 2.8297948034841074e-05,
+      "loss": 1.9289,
+      "step": 36850
+    },
+    {
+      "epoch": 4.061640066042928,
+      "grad_norm": 0.19162894785404205,
+      "learning_rate": 2.797957711673619e-05,
+      "loss": 1.9293,
+      "step": 36900
+    },
+    {
+      "epoch": 4.06714364336819,
+      "grad_norm": 0.1854933500289917,
+      "learning_rate": 2.7662822974482423e-05,
+      "loss": 1.9336,
+      "step": 36950
+    },
+    {
+      "epoch": 4.072647220693451,
+      "grad_norm": 0.18568897247314453,
+      "learning_rate": 2.7347689805085733e-05,
+      "loss": 1.9431,
+      "step": 37000
+    },
+    {
+      "epoch": 4.078150798018712,
+      "grad_norm": 0.1947336494922638,
+      "learning_rate": 2.7034181784074205e-05,
+      "loss": 1.9345,
+      "step": 37050
+    },
+    {
+      "epoch": 4.083654375343974,
+      "grad_norm": 0.18774710595607758,
+      "learning_rate": 2.672230306544254e-05,
+      "loss": 1.9364,
+      "step": 37100
+    },
+    {
+      "epoch": 4.089157952669235,
+      "grad_norm": 0.1984746754169464,
+      "learning_rate": 2.6412057781597223e-05,
+      "loss": 1.954,
+      "step": 37150
+    },
+    {
+      "epoch": 4.094661529994497,
+      "grad_norm": 0.1908876746892929,
+      "learning_rate": 2.61034500433016e-05,
+      "loss": 1.9422,
+      "step": 37200
+    },
+    {
+      "epoch": 4.100165107319758,
+      "grad_norm": 0.18860432505607605,
+      "learning_rate": 2.579648393962147e-05,
+      "loss": 1.9649,
+      "step": 37250
+    },
+    {
+      "epoch": 4.105668684645019,
+      "grad_norm": 0.18678942322731018,
+      "learning_rate": 2.5491163537870922e-05,
+      "loss": 1.9544,
+      "step": 37300
+    },
+    {
+      "epoch": 4.111172261970281,
+      "grad_norm": 0.19560185074806213,
+      "learning_rate": 2.5187492883558435e-05,
+      "loss": 1.9396,
+      "step": 37350
+    },
+    {
+      "epoch": 4.116675839295542,
+      "grad_norm": 0.1825692355632782,
+      "learning_rate": 2.488547600033317e-05,
+      "loss": 1.949,
+      "step": 37400
+    },
+    {
+      "epoch": 4.122179416620804,
+      "grad_norm": 0.1951771229505539,
+      "learning_rate": 2.4585116889931873e-05,
+      "loss": 1.9482,
+      "step": 37450
+    },
+    {
+      "epoch": 4.127682993946065,
+      "grad_norm": 0.1859281212091446,
+      "learning_rate": 2.4286419532125668e-05,
+      "loss": 1.9512,
+      "step": 37500
+    },
+    {
+      "epoch": 4.1331865712713265,
+      "grad_norm": 0.19174982607364655,
+      "learning_rate": 2.3989387884667365e-05,
+      "loss": 1.9358,
+      "step": 37550
+    },
+    {
+      "epoch": 4.138690148596588,
+      "grad_norm": 0.187012180685997,
+      "learning_rate": 2.369402588323908e-05,
+      "loss": 1.9114,
+      "step": 37600
+    },
+    {
+      "epoch": 4.144193725921849,
+      "grad_norm": 0.20616699755191803,
+      "learning_rate": 2.340033744140004e-05,
+      "loss": 1.8965,
+      "step": 37650
+    },
+    {
+      "epoch": 4.149697303247111,
+      "grad_norm": 0.17636051774024963,
+      "learning_rate": 2.3108326450534725e-05,
+      "loss": 1.9549,
+      "step": 37700
+    },
+    {
+      "epoch": 4.155200880572372,
+      "grad_norm": 0.19219453632831573,
+      "learning_rate": 2.281799677980136e-05,
+      "loss": 1.8911,
+      "step": 37750
+    },
+    {
+      "epoch": 4.1607044578976335,
+      "grad_norm": 0.19742678105831146,
+      "learning_rate": 2.2529352276080513e-05,
+      "loss": 1.9562,
+      "step": 37800
+    },
+    {
+      "epoch": 4.166208035222895,
+      "grad_norm": 0.18049471080303192,
+      "learning_rate": 2.2242396763924292e-05,
+      "loss": 1.9169,
+      "step": 37850
+    },
+    {
+      "epoch": 4.171711612548156,
+      "grad_norm": 0.2108810693025589,
+      "learning_rate": 2.1957134045505588e-05,
+      "loss": 1.9363,
+      "step": 37900
+    },
+    {
+      "epoch": 4.177215189873418,
+      "grad_norm": 0.20711076259613037,
+      "learning_rate": 2.1673567900567667e-05,
+      "loss": 1.9685,
+      "step": 37950
+    },
+    {
+      "epoch": 4.182718767198679,
+      "grad_norm": 0.18670059740543365,
+      "learning_rate": 2.139170208637415e-05,
+      "loss": 1.9537,
+      "step": 38000
+    },
+    {
+      "epoch": 4.188222344523941,
+      "grad_norm": 0.17974555492401123,
+      "learning_rate": 2.1111540337659227e-05,
+      "loss": 1.9476,
+      "step": 38050
+    },
+    {
+      "epoch": 4.193725921849202,
+      "grad_norm": 0.18271256983280182,
+      "learning_rate": 2.083308636657811e-05,
+      "loss": 1.9535,
+      "step": 38100
+    },
+    {
+      "epoch": 4.199229499174463,
+      "grad_norm": 0.19260470569133759,
+      "learning_rate": 2.0556343862657855e-05,
+      "loss": 1.9477,
+      "step": 38150
+    },
+    {
+      "epoch": 4.204733076499725,
+      "grad_norm": 0.19317923486232758,
+      "learning_rate": 2.0281316492748595e-05,
+      "loss": 1.9518,
+      "step": 38200
+    },
+    {
+      "epoch": 4.210236653824986,
+      "grad_norm": 0.1854427605867386,
+      "learning_rate": 2.0008007900974738e-05,
+      "loss": 1.9245,
+      "step": 38250
+    },
+    {
+      "epoch": 4.215740231150248,
+      "grad_norm": 0.1781974732875824,
+      "learning_rate": 1.9736421708686934e-05,
+      "loss": 1.9387,
+      "step": 38300
+    },
+    {
+      "epoch": 4.221243808475509,
+      "grad_norm": 0.18503354489803314,
+      "learning_rate": 1.946656151441389e-05,
+      "loss": 1.9683,
+      "step": 38350
+    },
+    {
+      "epoch": 4.2267473858007705,
+      "grad_norm": 0.19728437066078186,
+      "learning_rate": 1.9198430893814798e-05,
+      "loss": 1.9546,
+      "step": 38400
+    },
+    {
+      "epoch": 4.232250963126032,
+      "grad_norm": 0.1957421451807022,
+      "learning_rate": 1.89320333996319e-05,
+      "loss": 1.9526,
+      "step": 38450
+    },
+    {
+      "epoch": 4.237754540451293,
+      "grad_norm": 0.18944865465164185,
+      "learning_rate": 1.866737256164349e-05,
+      "loss": 1.9467,
+      "step": 38500
+    },
+    {
+      "epoch": 4.243258117776555,
+      "grad_norm": 0.1760573536157608,
+      "learning_rate": 1.8404451886617005e-05,
+      "loss": 1.9142,
+      "step": 38550
+    },
+    {
+      "epoch": 4.248761695101816,
+      "grad_norm": 0.19618487358093262,
+      "learning_rate": 1.8143274858262702e-05,
+      "loss": 1.9491,
+      "step": 38600
+    },
+    {
+      "epoch": 4.2542652724270775,
+      "grad_norm": 0.19029423594474792,
+      "learning_rate": 1.7883844937187495e-05,
+      "loss": 1.9553,
+      "step": 38650
+    },
+    {
+      "epoch": 4.259768849752339,
+      "grad_norm": 0.19688698649406433,
+      "learning_rate": 1.762616556084891e-05,
+      "loss": 1.9698,
+      "step": 38700
+    },
+    {
+      "epoch": 4.2652724270776,
+      "grad_norm": 0.16535942256450653,
+      "learning_rate": 1.737024014350983e-05,
+      "loss": 1.9395,
+      "step": 38750
+    },
+    {
+      "epoch": 4.270776004402862,
+      "grad_norm": 0.19671253859996796,
+      "learning_rate": 1.7116072076193e-05,
+      "loss": 1.9343,
+      "step": 38800
+    },
+    {
+      "epoch": 4.276279581728123,
+      "grad_norm": 0.18672049045562744,
+      "learning_rate": 1.6863664726636278e-05,
+      "loss": 1.939,
+      "step": 38850
+    },
+    {
+      "epoch": 4.281783159053385,
+      "grad_norm": 0.19714199006557465,
+      "learning_rate": 1.6613021439247914e-05,
+      "loss": 1.9455,
+      "step": 38900
+    },
+    {
+      "epoch": 4.287286736378646,
+      "grad_norm": 0.2018548846244812,
+      "learning_rate": 1.6364145535062172e-05,
+      "loss": 1.9169,
+      "step": 38950
+    },
+    {
+      "epoch": 4.292790313703907,
+      "grad_norm": 0.19057820737361908,
+      "learning_rate": 1.611704031169555e-05,
+      "loss": 1.956,
+      "step": 39000
+    },
+    {
+      "epoch": 4.298293891029169,
+      "grad_norm": 0.18719470500946045,
+      "learning_rate": 1.5871709043302876e-05,
+      "loss": 1.9483,
+      "step": 39050
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 0.19479648768901825,
+      "learning_rate": 1.5628154980533984e-05,
+      "loss": 1.9507,
+      "step": 39100
+    },
+    {
+      "epoch": 4.309301045679692,
+      "grad_norm": 0.2034020870923996,
+      "learning_rate": 1.538638135049071e-05,
+      "loss": 1.9343,
+      "step": 39150
+    },
+    {
+      "epoch": 4.314804623004953,
+      "grad_norm": 0.18177463114261627,
+      "learning_rate": 1.5146391356684029e-05,
+      "loss": 1.9481,
+      "step": 39200
+    },
+    {
+      "epoch": 4.3203082003302145,
+      "grad_norm": 0.1850568652153015,
+      "learning_rate": 1.4908188178991714e-05,
+      "loss": 1.9324,
+      "step": 39250
+    },
+    {
+      "epoch": 4.325811777655476,
+      "grad_norm": 0.1880549043416977,
+      "learning_rate": 1.4671774973616102e-05,
+      "loss": 1.9406,
+      "step": 39300
+    },
+    {
+      "epoch": 4.331315354980737,
+      "grad_norm": 0.18556420505046844,
+      "learning_rate": 1.4437154873042279e-05,
+      "loss": 1.9516,
+      "step": 39350
+    },
+    {
+      "epoch": 4.336818932305999,
+      "grad_norm": 0.19034932553768158,
+      "learning_rate": 1.420433098599672e-05,
+      "loss": 1.9477,
+      "step": 39400
+    },
+    {
+      "epoch": 4.34232250963126,
+      "grad_norm": 0.18247225880622864,
+      "learning_rate": 1.3973306397405909e-05,
+      "loss": 1.9552,
+      "step": 39450
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "grad_norm": 0.19472351670265198,
+      "learning_rate": 1.3744084168355612e-05,
+      "loss": 1.9543,
+      "step": 39500
+    },
+    {
+      "epoch": 4.353329664281783,
+      "grad_norm": 0.18814486265182495,
+      "learning_rate": 1.3516667336050219e-05,
+      "loss": 1.956,
+      "step": 39550
+    },
+    {
+      "epoch": 4.358833241607044,
+      "grad_norm": 0.18639877438545227,
+      "learning_rate": 1.3291058913772517e-05,
+      "loss": 1.9262,
+      "step": 39600
+    },
+    {
+      "epoch": 4.364336818932306,
+      "grad_norm": 0.19503189623355865,
+      "learning_rate": 1.306726189084385e-05,
+      "loss": 1.9768,
+      "step": 39650
+    },
+    {
+      "epoch": 4.369840396257567,
+      "grad_norm": 0.1823161244392395,
+      "learning_rate": 1.2845279232584354e-05,
+      "loss": 1.9588,
+      "step": 39700
+    },
+    {
+      "epoch": 4.375343973582829,
+      "grad_norm": 0.19661200046539307,
+      "learning_rate": 1.2625113880273786e-05,
+      "loss": 1.9576,
+      "step": 39750
+    },
+    {
+      "epoch": 4.38084755090809,
+      "grad_norm": 0.17353294789791107,
+      "learning_rate": 1.2406768751112572e-05,
+      "loss": 1.9249,
+      "step": 39800
+    },
+    {
+      "epoch": 4.386351128233351,
+      "grad_norm": 0.1832091063261032,
+      "learning_rate": 1.2190246738183074e-05,
+      "loss": 1.9169,
+      "step": 39850
+    },
+    {
+      "epoch": 4.391854705558613,
+      "grad_norm": 0.18823185563087463,
+      "learning_rate": 1.197555071041122e-05,
+      "loss": 1.9368,
+      "step": 39900
+    },
+    {
+      "epoch": 4.397358282883874,
+      "grad_norm": 0.1941109150648117,
+      "learning_rate": 1.1762683512528642e-05,
+      "loss": 1.96,
+      "step": 39950
+    },
+    {
+      "epoch": 4.402861860209136,
+      "grad_norm": 0.1973351240158081,
+      "learning_rate": 1.155164796503486e-05,
+      "loss": 1.9304,
+      "step": 40000
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 45425,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.91746903834624e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}