diff --git "a/checkpoint-1021770/trainer_state.json" "b/checkpoint-1021770/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1021770/trainer_state.json"
@@ -0,0 +1,14322 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 29.999559594239493,
+  "eval_steps": 500,
+  "global_step": 1021770,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2330610752105713,
+      "learning_rate": 7.5e-05,
+      "loss": 47.9921,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4777156114578247,
+      "learning_rate": 0.00015,
+      "loss": 6.0359,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.119438886642456,
+      "learning_rate": 0.000225,
+      "loss": 5.8858,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3987252712249756,
+      "learning_rate": 0.0003,
+      "loss": 5.7995,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.466189980506897,
+      "learning_rate": 0.0002998529080086686,
+      "loss": 5.5973,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6185613870620728,
+      "learning_rate": 0.0002997058160173372,
+      "loss": 4.5547,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.422328472137451,
+      "learning_rate": 0.0002995587240260058,
+      "loss": 3.8361,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.422811269760132,
+      "learning_rate": 0.00029941163203467446,
+      "loss": 3.5267,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8635873794555664,
+      "learning_rate": 0.0002992645400433431,
+      "loss": 3.3318,
+      "step": 4500
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5039353370666504,
+      "learning_rate": 0.00029911744805201173,
+      "loss": 3.2049,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.018263578414917,
+      "learning_rate": 0.0002989703560606803,
+      "loss": 3.1146,
+      "step": 5500
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2091281414031982,
+      "learning_rate": 0.00029882326406934894,
+      "loss": 3.0412,
+      "step": 6000
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7145626544952393,
+      "learning_rate": 0.0002986761720780176,
+      "loss": 2.9849,
+      "step": 6500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.0529212951660156,
+      "learning_rate": 0.00029852908008668616,
+      "loss": 2.9731,
+      "step": 7000
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4529542922973633,
+      "learning_rate": 0.0002983819880953548,
+      "loss": 2.9193,
+      "step": 7500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.009725332260132,
+      "learning_rate": 0.00029823489610402343,
+      "loss": 2.8839,
+      "step": 8000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.02736759185791,
+      "learning_rate": 0.00029808780411269206,
+      "loss": 2.8483,
+      "step": 8500
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.597303867340088,
+      "learning_rate": 0.0002979407121213607,
+      "loss": 2.84,
+      "step": 9000
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.826665163040161,
+      "learning_rate": 0.0002977936201300293,
+      "loss": 2.8151,
+      "step": 9500
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.141164779663086,
+      "learning_rate": 0.0002976465281386979,
+      "loss": 2.7756,
+      "step": 10000
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0721323490142822,
+      "learning_rate": 0.00029749943614736655,
+      "loss": 2.7468,
+      "step": 10500
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0305256843566895,
+      "learning_rate": 0.0002973523441560351,
+      "loss": 2.7548,
+      "step": 11000
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7159794569015503,
+      "learning_rate": 0.00029720525216470376,
+      "loss": 2.739,
+      "step": 11500
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.015839099884033,
+      "learning_rate": 0.0002970581601733724,
+      "loss": 2.7338,
+      "step": 12000
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1496126651763916,
+      "learning_rate": 0.00029691106818204103,
+      "loss": 2.7047,
+      "step": 12500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 12.543700218200684,
+      "learning_rate": 0.00029676397619070967,
+      "loss": 2.6905,
+      "step": 13000
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.919098138809204,
+      "learning_rate": 0.00029661688419937825,
+      "loss": 2.6511,
+      "step": 13500
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.164848566055298,
+      "learning_rate": 0.0002964697922080469,
+      "loss": 2.6765,
+      "step": 14000
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.595182418823242,
+      "learning_rate": 0.0002963227002167155,
+      "loss": 2.6406,
+      "step": 14500
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.517666816711426,
+      "learning_rate": 0.0002961756082253841,
+      "loss": 2.6752,
+      "step": 15000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.788487434387207,
+      "learning_rate": 0.00029602851623405273,
+      "loss": 2.6248,
+      "step": 15500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8919869661331177,
+      "learning_rate": 0.00029588142424272137,
+      "loss": 2.6751,
+      "step": 16000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.1151628494262695,
+      "learning_rate": 0.00029573433225139,
+      "loss": 2.6024,
+      "step": 16500
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6645219326019287,
+      "learning_rate": 0.00029558724026005863,
+      "loss": 2.6017,
+      "step": 17000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.183793067932129,
+      "learning_rate": 0.0002954401482687272,
+      "loss": 2.5482,
+      "step": 17500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.3193671703338623,
+      "learning_rate": 0.00029529305627739585,
+      "loss": 2.6058,
+      "step": 18000
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.861599326133728,
+      "learning_rate": 0.0002951459642860645,
+      "loss": 2.5766,
+      "step": 18500
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.443558931350708,
+      "learning_rate": 0.00029499887229473306,
+      "loss": 2.5901,
+      "step": 19000
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1485493183135986,
+      "learning_rate": 0.0002948517803034017,
+      "loss": 2.549,
+      "step": 19500
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.802183151245117,
+      "learning_rate": 0.00029470468831207033,
+      "loss": 2.568,
+      "step": 20000
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.226308822631836,
+      "learning_rate": 0.00029455759632073897,
+      "loss": 2.5599,
+      "step": 20500
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.404243469238281,
+      "learning_rate": 0.0002944105043294076,
+      "loss": 2.5574,
+      "step": 21000
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.14946985244751,
+      "learning_rate": 0.0002942634123380762,
+      "loss": 2.5266,
+      "step": 21500
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.674615502357483,
+      "learning_rate": 0.0002941163203467448,
+      "loss": 2.5094,
+      "step": 22000
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.943082332611084,
+      "learning_rate": 0.00029396922835541345,
+      "loss": 2.4967,
+      "step": 22500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9026265144348145,
+      "learning_rate": 0.0002938221363640821,
+      "loss": 2.4974,
+      "step": 23000
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.214877128601074,
+      "learning_rate": 0.00029367504437275067,
+      "loss": 2.5329,
+      "step": 23500
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1662731170654297,
+      "learning_rate": 0.0002935279523814193,
+      "loss": 2.508,
+      "step": 24000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6833200454711914,
+      "learning_rate": 0.00029338086039008794,
+      "loss": 2.4891,
+      "step": 24500
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 10.436029434204102,
+      "learning_rate": 0.00029323376839875657,
+      "loss": 2.506,
+      "step": 25000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6364326477050781,
+      "learning_rate": 0.0002930866764074252,
+      "loss": 2.5091,
+      "step": 25500
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.408642292022705,
+      "learning_rate": 0.0002929395844160938,
+      "loss": 2.4987,
+      "step": 26000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.324516534805298,
+      "learning_rate": 0.0002927924924247624,
+      "loss": 2.5619,
+      "step": 26500
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3488011360168457,
+      "learning_rate": 0.00029264540043343106,
+      "loss": 2.4594,
+      "step": 27000
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.130359411239624,
+      "learning_rate": 0.00029249830844209964,
+      "loss": 2.4677,
+      "step": 27500
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6321048736572266,
+      "learning_rate": 0.0002923512164507683,
+      "loss": 2.4925,
+      "step": 28000
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2566709518432617,
+      "learning_rate": 0.0002922041244594369,
+      "loss": 2.4779,
+      "step": 28500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.4957294464111328,
+      "learning_rate": 0.00029205703246810554,
+      "loss": 2.4482,
+      "step": 29000
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.031252384185791,
+      "learning_rate": 0.0002919099404767742,
+      "loss": 2.4215,
+      "step": 29500
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.686828136444092,
+      "learning_rate": 0.00029176284848544276,
+      "loss": 2.4463,
+      "step": 30000
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.746907114982605,
+      "learning_rate": 0.0002916157564941114,
+      "loss": 2.4436,
+      "step": 30500
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.379333257675171,
+      "learning_rate": 0.00029146866450278,
+      "loss": 2.4244,
+      "step": 31000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.696253776550293,
+      "learning_rate": 0.0002913215725114486,
+      "loss": 2.4218,
+      "step": 31500
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.1202826499938965,
+      "learning_rate": 0.0002911744805201173,
+      "loss": 2.438,
+      "step": 32000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3910372257232666,
+      "learning_rate": 0.0002910273885287859,
+      "loss": 2.4405,
+      "step": 32500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9638476371765137,
+      "learning_rate": 0.0002908802965374545,
+      "loss": 2.4461,
+      "step": 33000
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.280348062515259,
+      "learning_rate": 0.00029073320454612315,
+      "loss": 2.4177,
+      "step": 33500
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.654728651046753,
+      "learning_rate": 0.0002905861125547917,
+      "loss": 2.414,
+      "step": 34000
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.3180739879608154,
+      "learning_rate": 0.00029043902056346036,
+      "loss": 2.367,
+      "step": 34500
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.21975040435791,
+      "learning_rate": 0.000290291928572129,
+      "loss": 2.4158,
+      "step": 35000
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7568458318710327,
+      "learning_rate": 0.0002901448365807976,
+      "loss": 2.3574,
+      "step": 35500
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.872269630432129,
+      "learning_rate": 0.00028999774458946626,
+      "loss": 2.3842,
+      "step": 36000
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.3413188457489014,
+      "learning_rate": 0.00028985065259813484,
+      "loss": 2.3313,
+      "step": 36500
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.4920777082443237,
+      "learning_rate": 0.0002897035606068035,
+      "loss": 2.4049,
+      "step": 37000
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.5813608169555664,
+      "learning_rate": 0.0002895564686154721,
+      "loss": 2.3641,
+      "step": 37500
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.110539674758911,
+      "learning_rate": 0.0002894093766241407,
+      "loss": 2.3704,
+      "step": 38000
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.397954225540161,
+      "learning_rate": 0.00028926228463280933,
+      "loss": 2.3962,
+      "step": 38500
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.290098786354065,
+      "learning_rate": 0.00028911519264147796,
+      "loss": 2.3359,
+      "step": 39000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.7063418626785278,
+      "learning_rate": 0.00028896810065014654,
+      "loss": 2.34,
+      "step": 39500
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.603582501411438,
+      "learning_rate": 0.00028882100865881523,
+      "loss": 2.3439,
+      "step": 40000
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6357001066207886,
+      "learning_rate": 0.0002886739166674838,
+      "loss": 2.3689,
+      "step": 40500
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.4969911575317383,
+      "learning_rate": 0.00028852682467615245,
+      "loss": 2.3621,
+      "step": 41000
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.7131437063217163,
+      "learning_rate": 0.0002883797326848211,
+      "loss": 2.3402,
+      "step": 41500
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.681779384613037,
+      "learning_rate": 0.00028823264069348966,
+      "loss": 2.3047,
+      "step": 42000
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.0767245292663574,
+      "learning_rate": 0.0002880855487021583,
+      "loss": 2.3529,
+      "step": 42500
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.4065330028533936,
+      "learning_rate": 0.00028793845671082693,
+      "loss": 2.3332,
+      "step": 43000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.0359294414520264,
+      "learning_rate": 0.0002877913647194955,
+      "loss": 2.3477,
+      "step": 43500
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.010524272918701,
+      "learning_rate": 0.0002876442727281642,
+      "loss": 2.3123,
+      "step": 44000
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 9.476223945617676,
+      "learning_rate": 0.0002874971807368328,
+      "loss": 2.3187,
+      "step": 44500
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.5241541862487793,
+      "learning_rate": 0.0002873500887455014,
+      "loss": 2.3473,
+      "step": 45000
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.111661434173584,
+      "learning_rate": 0.00028720299675417005,
+      "loss": 2.3122,
+      "step": 45500
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.5378273725509644,
+      "learning_rate": 0.00028705590476283863,
+      "loss": 2.3185,
+      "step": 46000
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.218921184539795,
+      "learning_rate": 0.00028690881277150727,
+      "loss": 2.3426,
+      "step": 46500
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.234276294708252,
+      "learning_rate": 0.0002867617207801759,
+      "loss": 2.2885,
+      "step": 47000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.929063081741333,
+      "learning_rate": 0.0002866146287888445,
+      "loss": 2.3047,
+      "step": 47500
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.764546513557434,
+      "learning_rate": 0.00028646753679751317,
+      "loss": 2.275,
+      "step": 48000
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.143376111984253,
+      "learning_rate": 0.00028632044480618175,
+      "loss": 2.2591,
+      "step": 48500
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.8672120571136475,
+      "learning_rate": 0.0002861733528148504,
+      "loss": 2.2462,
+      "step": 49000
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.577282667160034,
+      "learning_rate": 0.000286026260823519,
+      "loss": 2.2689,
+      "step": 49500
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.5024135112762451,
+      "learning_rate": 0.0002858791688321876,
+      "loss": 2.2886,
+      "step": 50000
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.36291241645813,
+      "learning_rate": 0.00028573207684085624,
+      "loss": 2.2102,
+      "step": 50500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.7048864364624023,
+      "learning_rate": 0.00028558498484952487,
+      "loss": 2.2734,
+      "step": 51000
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 15.023186683654785,
+      "learning_rate": 0.00028543789285819345,
+      "loss": 2.2859,
+      "step": 51500
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.5235730409622192,
+      "learning_rate": 0.00028529080086686214,
+      "loss": 2.2459,
+      "step": 52000
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.157870292663574,
+      "learning_rate": 0.0002851437088755307,
+      "loss": 2.2679,
+      "step": 52500
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 5.78060245513916,
+      "learning_rate": 0.00028499661688419936,
+      "loss": 2.2781,
+      "step": 53000
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.3039705753326416,
+      "learning_rate": 0.000284849524892868,
+      "loss": 2.2761,
+      "step": 53500
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.286205530166626,
+      "learning_rate": 0.00028470243290153657,
+      "loss": 2.2549,
+      "step": 54000
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.2150983810424805,
+      "learning_rate": 0.0002845553409102052,
+      "loss": 2.2443,
+      "step": 54500
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.0205180644989014,
+      "learning_rate": 0.00028440824891887384,
+      "loss": 2.2676,
+      "step": 55000
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.5430960655212402,
+      "learning_rate": 0.0002842611569275424,
+      "loss": 2.2403,
+      "step": 55500
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.979278802871704,
+      "learning_rate": 0.0002841140649362111,
+      "loss": 2.2409,
+      "step": 56000
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.6833264827728271,
+      "learning_rate": 0.0002839669729448797,
+      "loss": 2.2657,
+      "step": 56500
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.9541752338409424,
+      "learning_rate": 0.0002838198809535483,
+      "loss": 2.2767,
+      "step": 57000
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 5.107935905456543,
+      "learning_rate": 0.00028367278896221696,
+      "loss": 2.2774,
+      "step": 57500
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.5604417324066162,
+      "learning_rate": 0.00028352569697088554,
+      "loss": 2.2405,
+      "step": 58000
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.6418412923812866,
+      "learning_rate": 0.0002833786049795542,
+      "loss": 2.2434,
+      "step": 58500
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.5278984308242798,
+      "learning_rate": 0.0002832315129882228,
+      "loss": 2.2977,
+      "step": 59000
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.3131840229034424,
+      "learning_rate": 0.0002830844209968914,
+      "loss": 2.2236,
+      "step": 59500
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.7341011762619019,
+      "learning_rate": 0.0002829373290055601,
+      "loss": 2.2235,
+      "step": 60000
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.8039567470550537,
+      "learning_rate": 0.00028279023701422866,
+      "loss": 2.2475,
+      "step": 60500
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.142283916473389,
+      "learning_rate": 0.0002826431450228973,
+      "loss": 2.2128,
+      "step": 61000
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.2194138765335083,
+      "learning_rate": 0.00028249605303156593,
+      "loss": 2.2568,
+      "step": 61500
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.2129665613174438,
+      "learning_rate": 0.0002823489610402345,
+      "loss": 2.2282,
+      "step": 62000
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 8.789237022399902,
+      "learning_rate": 0.00028220186904890314,
+      "loss": 2.2071,
+      "step": 62500
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.4489936828613281,
+      "learning_rate": 0.0002820547770575718,
+      "loss": 2.2389,
+      "step": 63000
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.591963768005371,
+      "learning_rate": 0.0002819076850662404,
+      "loss": 2.2277,
+      "step": 63500
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.642183303833008,
+      "learning_rate": 0.00028176059307490905,
+      "loss": 2.2195,
+      "step": 64000
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.7950453758239746,
+      "learning_rate": 0.00028161350108357763,
+      "loss": 2.2179,
+      "step": 64500
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.3858351707458496,
+      "learning_rate": 0.00028146640909224626,
+      "loss": 2.2266,
+      "step": 65000
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.2545394897460938,
+      "learning_rate": 0.0002813193171009149,
+      "loss": 2.2538,
+      "step": 65500
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.3457772731781006,
+      "learning_rate": 0.0002811722251095835,
+      "loss": 2.2351,
+      "step": 66000
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.0313475131988525,
+      "learning_rate": 0.0002810251331182521,
+      "loss": 2.2574,
+      "step": 66500
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 4.175018310546875,
+      "learning_rate": 0.00028087804112692075,
+      "loss": 2.2269,
+      "step": 67000
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.963953971862793,
+      "learning_rate": 0.0002807309491355894,
+      "loss": 2.2408,
+      "step": 67500
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.8645119667053223,
+      "learning_rate": 0.000280583857144258,
+      "loss": 2.2588,
+      "step": 68000
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.8845419883728027,
+      "learning_rate": 0.0002804367651529266,
+      "loss": 2.1532,
+      "step": 68500
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 2.3963561058044434,
+      "learning_rate": 0.00028028967316159523,
+      "loss": 2.1441,
+      "step": 69000
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 3.194061040878296,
+      "learning_rate": 0.00028014258117026387,
+      "loss": 2.1184,
+      "step": 69500
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.4685026407241821,
+      "learning_rate": 0.0002799954891789325,
+      "loss": 2.1504,
+      "step": 70000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 6.733902931213379,
+      "learning_rate": 0.0002798483971876011,
+      "loss": 2.1449,
+      "step": 70500
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.5692353248596191,
+      "learning_rate": 0.0002797013051962697,
+      "loss": 2.1132,
+      "step": 71000
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.4298503398895264,
+      "learning_rate": 0.00027955421320493835,
+      "loss": 2.1557,
+      "step": 71500
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.9930598735809326,
+      "learning_rate": 0.000279407121213607,
+      "loss": 2.1509,
+      "step": 72000
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 3.2199394702911377,
+      "learning_rate": 0.0002792600292222756,
+      "loss": 2.1637,
+      "step": 72500
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 11.737617492675781,
+      "learning_rate": 0.0002791129372309442,
+      "loss": 2.1145,
+      "step": 73000
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 3.2281816005706787,
+      "learning_rate": 0.00027896584523961284,
+      "loss": 2.205,
+      "step": 73500
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.6318072080612183,
+      "learning_rate": 0.00027881875324828147,
+      "loss": 2.1355,
+      "step": 74000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.4795607328414917,
+      "learning_rate": 0.00027867166125695005,
+      "loss": 2.1594,
+      "step": 74500
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 3.246556043624878,
+      "learning_rate": 0.0002785245692656187,
+      "loss": 2.1225,
+      "step": 75000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 5.301700115203857,
+      "learning_rate": 0.0002783774772742873,
+      "loss": 2.1477,
+      "step": 75500
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.3937615156173706,
+      "learning_rate": 0.00027823038528295595,
+      "loss": 2.1377,
+      "step": 76000
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 5.8993611335754395,
+      "learning_rate": 0.0002780832932916246,
+      "loss": 2.156,
+      "step": 76500
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 3.7031867504119873,
+      "learning_rate": 0.00027793620130029317,
+      "loss": 2.1246,
+      "step": 77000
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 3.8393964767456055,
+      "learning_rate": 0.0002777891093089618,
+      "loss": 2.1503,
+      "step": 77500
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.3489915132522583,
+      "learning_rate": 0.00027764201731763044,
+      "loss": 2.1506,
+      "step": 78000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 6.942753791809082,
+      "learning_rate": 0.000277494925326299,
+      "loss": 2.1575,
+      "step": 78500
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.2154929637908936,
+      "learning_rate": 0.00027734783333496765,
+      "loss": 2.1279,
+      "step": 79000
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.8168853521347046,
+      "learning_rate": 0.0002772007413436363,
+      "loss": 2.0715,
+      "step": 79500
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.112091541290283,
+      "learning_rate": 0.0002770536493523049,
+      "loss": 2.1344,
+      "step": 80000
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.050875663757324,
+      "learning_rate": 0.00027690655736097356,
+      "loss": 2.1416,
+      "step": 80500
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.736621618270874,
+      "learning_rate": 0.00027675946536964214,
+      "loss": 2.1196,
+      "step": 81000
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.051025629043579,
+      "learning_rate": 0.0002766123733783108,
+      "loss": 2.1272,
+      "step": 81500
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 4.314809799194336,
+      "learning_rate": 0.0002764652813869794,
+      "loss": 2.1618,
+      "step": 82000
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.920485496520996,
+      "learning_rate": 0.000276318189395648,
+      "loss": 2.1354,
+      "step": 82500
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 6.310970783233643,
+      "learning_rate": 0.0002761710974043166,
+      "loss": 2.1495,
+      "step": 83000
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.929152250289917,
+      "learning_rate": 0.00027602400541298526,
+      "loss": 2.1271,
+      "step": 83500
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.9410946369171143,
+      "learning_rate": 0.0002758769134216539,
+      "loss": 2.1646,
+      "step": 84000
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.474297523498535,
+      "learning_rate": 0.00027572982143032253,
+      "loss": 2.127,
+      "step": 84500
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.6442033052444458,
+      "learning_rate": 0.0002755827294389911,
+      "loss": 2.1433,
+      "step": 85000
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.3729546070098877,
+      "learning_rate": 0.00027543563744765974,
+      "loss": 2.1121,
+      "step": 85500
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.6497186422348022,
+      "learning_rate": 0.0002752885454563284,
+      "loss": 2.1566,
+      "step": 86000
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.6162164211273193,
+      "learning_rate": 0.00027514145346499696,
+      "loss": 2.1513,
+      "step": 86500
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.7324166297912598,
+      "learning_rate": 0.0002749943614736656,
+      "loss": 2.1482,
+      "step": 87000
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.3059158325195312,
+      "learning_rate": 0.0002748472694823342,
+      "loss": 2.1346,
+      "step": 87500
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.1019114255905151,
+      "learning_rate": 0.00027470017749100286,
+      "loss": 2.123,
+      "step": 88000
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 7.39063024520874,
+      "learning_rate": 0.0002745530854996715,
+      "loss": 2.1278,
+      "step": 88500
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.9237990379333496,
+      "learning_rate": 0.0002744059935083401,
+      "loss": 2.115,
+      "step": 89000
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.4632532596588135,
+      "learning_rate": 0.0002742589015170087,
+      "loss": 2.137,
+      "step": 89500
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.9587647914886475,
+      "learning_rate": 0.00027411180952567735,
+      "loss": 2.1056,
+      "step": 90000
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 7.492849349975586,
+      "learning_rate": 0.0002739647175343459,
+      "loss": 2.1208,
+      "step": 90500
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.7255750894546509,
+      "learning_rate": 0.00027381762554301456,
+      "loss": 2.1015,
+      "step": 91000
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 8.091690063476562,
+      "learning_rate": 0.0002736705335516832,
+      "loss": 2.1073,
+      "step": 91500
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 6.672662734985352,
+      "learning_rate": 0.00027352344156035183,
+      "loss": 2.1059,
+      "step": 92000
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.948398470878601,
+      "learning_rate": 0.00027337634956902047,
+      "loss": 2.0943,
+      "step": 92500
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.4260573387145996,
+      "learning_rate": 0.00027322925757768905,
+      "loss": 2.1283,
+      "step": 93000
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.5472320318222046,
+      "learning_rate": 0.0002730821655863577,
+      "loss": 2.0882,
+      "step": 93500
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.3860925436019897,
+      "learning_rate": 0.0002729350735950263,
+      "loss": 2.1323,
+      "step": 94000
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.232808828353882,
+      "learning_rate": 0.0002727879816036949,
+      "loss": 2.1485,
+      "step": 94500
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.8404427766799927,
+      "learning_rate": 0.00027264088961236353,
+      "loss": 2.1116,
+      "step": 95000
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 2.124656915664673,
+      "learning_rate": 0.00027249379762103216,
+      "loss": 2.1076,
+      "step": 95500
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 2.2169177532196045,
+      "learning_rate": 0.0002723467056297008,
+      "loss": 2.1035,
+      "step": 96000
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.7737512588500977,
+      "learning_rate": 0.00027219961363836943,
+      "loss": 2.1369,
+      "step": 96500
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.6662325859069824,
+      "learning_rate": 0.000272052521647038,
+      "loss": 2.1131,
+      "step": 97000
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 2.6663177013397217,
+      "learning_rate": 0.00027190542965570665,
+      "loss": 2.1175,
+      "step": 97500
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.446356177330017,
+      "learning_rate": 0.0002717583376643753,
+      "loss": 2.1076,
+      "step": 98000
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.3340353965759277,
+      "learning_rate": 0.00027161124567304386,
+      "loss": 2.0877,
+      "step": 98500
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 3.389127254486084,
+      "learning_rate": 0.0002714641536817125,
+      "loss": 2.0904,
+      "step": 99000
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.044728994369507,
+      "learning_rate": 0.00027131706169038113,
+      "loss": 2.0608,
+      "step": 99500
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 17.932655334472656,
+      "learning_rate": 0.00027116996969904977,
+      "loss": 2.1136,
+      "step": 100000
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.4239097833633423,
+      "learning_rate": 0.0002710228777077184,
+      "loss": 2.0783,
+      "step": 100500
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 2.1303598880767822,
+      "learning_rate": 0.000270875785716387,
+      "loss": 2.0981,
+      "step": 101000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.2777888774871826,
+      "learning_rate": 0.0002707286937250556,
+      "loss": 2.0862,
+      "step": 101500
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 2.2035715579986572,
+      "learning_rate": 0.00027058160173372425,
+      "loss": 2.1089,
+      "step": 102000
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 2.7756547927856445,
+      "learning_rate": 0.00027043450974239283,
+      "loss": 2.0614,
+      "step": 102500
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.4222705364227295,
+      "learning_rate": 0.00027028741775106147,
+      "loss": 2.0216,
+      "step": 103000
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 1.5257524251937866,
+      "learning_rate": 0.0002701403257597301,
+      "loss": 2.0388,
+      "step": 103500
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 2.274021625518799,
+      "learning_rate": 0.00026999323376839874,
+      "loss": 2.048,
+      "step": 104000
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 5.903868198394775,
+      "learning_rate": 0.00026984614177706737,
+      "loss": 2.0008,
+      "step": 104500
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 1.561962604522705,
+      "learning_rate": 0.00026969904978573595,
+      "loss": 2.05,
+      "step": 105000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.002523422241211,
+      "learning_rate": 0.0002695519577944046,
+      "loss": 2.0217,
+      "step": 105500
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 1.9345905780792236,
+      "learning_rate": 0.0002694048658030732,
+      "loss": 2.0343,
+      "step": 106000
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.5755488872528076,
+      "learning_rate": 0.0002692577738117418,
+      "loss": 2.0133,
+      "step": 106500
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.7304731607437134,
+      "learning_rate": 0.00026911068182041044,
+      "loss": 2.0177,
+      "step": 107000
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 2.4045250415802,
+      "learning_rate": 0.00026896358982907907,
+      "loss": 2.034,
+      "step": 107500
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.5433346033096313,
+      "learning_rate": 0.0002688164978377477,
+      "loss": 2.0228,
+      "step": 108000
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 2.033250093460083,
+      "learning_rate": 0.00026866940584641634,
+      "loss": 2.0076,
+      "step": 108500
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 3.0048422813415527,
+      "learning_rate": 0.0002685223138550849,
+      "loss": 1.9876,
+      "step": 109000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.6355750560760498,
+      "learning_rate": 0.00026837522186375356,
+      "loss": 2.0259,
+      "step": 109500
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 7.954076290130615,
+      "learning_rate": 0.0002682281298724222,
+      "loss": 2.0408,
+      "step": 110000
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.3474091291427612,
+      "learning_rate": 0.0002680810378810908,
+      "loss": 2.0114,
+      "step": 110500
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 1.8665661811828613,
+      "learning_rate": 0.0002679339458897594,
+      "loss": 2.0388,
+      "step": 111000
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.452719807624817,
+      "learning_rate": 0.00026778685389842804,
+      "loss": 2.0411,
+      "step": 111500
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.9409444332122803,
+      "learning_rate": 0.0002676397619070967,
+      "loss": 2.0355,
+      "step": 112000
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.4139233827590942,
+      "learning_rate": 0.0002674926699157653,
+      "loss": 2.0306,
+      "step": 112500
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 2.548440933227539,
+      "learning_rate": 0.00026734557792443394,
+      "loss": 2.013,
+      "step": 113000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 16.565500259399414,
+      "learning_rate": 0.0002671984859331025,
+      "loss": 2.0435,
+      "step": 113500
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 2.008643865585327,
+      "learning_rate": 0.00026705139394177116,
+      "loss": 2.0094,
+      "step": 114000
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 5.655598163604736,
+      "learning_rate": 0.0002669043019504398,
+      "loss": 2.0169,
+      "step": 114500
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 13.43127727508545,
+      "learning_rate": 0.0002667572099591084,
+      "loss": 2.0273,
+      "step": 115000
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 4.050173282623291,
+      "learning_rate": 0.00026661011796777706,
+      "loss": 2.0406,
+      "step": 115500
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 2.2970187664031982,
+      "learning_rate": 0.00026646302597644564,
+      "loss": 2.0374,
+      "step": 116000
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.8749233484268188,
+      "learning_rate": 0.0002663159339851143,
+      "loss": 1.991,
+      "step": 116500
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 2.4106335639953613,
+      "learning_rate": 0.0002661688419937829,
+      "loss": 2.0425,
+      "step": 117000
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.8036812543869019,
+      "learning_rate": 0.0002660217500024515,
+      "loss": 1.9968,
+      "step": 117500
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.831998348236084,
+      "learning_rate": 0.00026587465801112013,
+      "loss": 2.0291,
+      "step": 118000
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 2.4199304580688477,
+      "learning_rate": 0.00026572756601978876,
+      "loss": 2.0186,
+      "step": 118500
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 4.254742622375488,
+      "learning_rate": 0.00026558047402845734,
+      "loss": 2.0226,
+      "step": 119000
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 12.558996200561523,
+      "learning_rate": 0.00026543338203712603,
+      "loss": 2.0368,
+      "step": 119500
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.9092944860458374,
+      "learning_rate": 0.0002652862900457946,
+      "loss": 2.02,
+      "step": 120000
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 1.9921791553497314,
+      "learning_rate": 0.00026513919805446325,
+      "loss": 2.019,
+      "step": 120500
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.1361401081085205,
+      "learning_rate": 0.0002649921060631319,
+      "loss": 1.9927,
+      "step": 121000
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.6607388257980347,
+      "learning_rate": 0.00026484501407180046,
+      "loss": 2.0722,
+      "step": 121500
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 2.246613025665283,
+      "learning_rate": 0.0002646979220804691,
+      "loss": 2.0091,
+      "step": 122000
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 5.497014045715332,
+      "learning_rate": 0.00026455083008913773,
+      "loss": 2.0182,
+      "step": 122500
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.5741316080093384,
+      "learning_rate": 0.00026440373809780637,
+      "loss": 2.0113,
+      "step": 123000
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 2.1419429779052734,
+      "learning_rate": 0.000264256646106475,
+      "loss": 2.0112,
+      "step": 123500
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 2.309093952178955,
+      "learning_rate": 0.0002641095541151436,
+      "loss": 1.9943,
+      "step": 124000
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 1.7635866403579712,
+      "learning_rate": 0.0002639624621238122,
+      "loss": 2.0081,
+      "step": 124500
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 2.3035855293273926,
+      "learning_rate": 0.00026381537013248085,
+      "loss": 1.9658,
+      "step": 125000
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 1.8141978979110718,
+      "learning_rate": 0.00026366827814114943,
+      "loss": 2.0203,
+      "step": 125500
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.8287400007247925,
+      "learning_rate": 0.00026352118614981807,
+      "loss": 1.9636,
+      "step": 126000
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.356780767440796,
+      "learning_rate": 0.0002633740941584867,
+      "loss": 2.0116,
+      "step": 126500
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 5.973343849182129,
+      "learning_rate": 0.00026322700216715534,
+      "loss": 1.9837,
+      "step": 127000
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 2.090287446975708,
+      "learning_rate": 0.00026307991017582397,
+      "loss": 2.0161,
+      "step": 127500
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.422405481338501,
+      "learning_rate": 0.00026293281818449255,
+      "loss": 2.0096,
+      "step": 128000
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 5.0919365882873535,
+      "learning_rate": 0.0002627857261931612,
+      "loss": 1.997,
+      "step": 128500
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 2.7366178035736084,
+      "learning_rate": 0.0002626386342018298,
+      "loss": 2.001,
+      "step": 129000
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 1.3529456853866577,
+      "learning_rate": 0.0002624915422104984,
+      "loss": 2.0208,
+      "step": 129500
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 2.3386831283569336,
+      "learning_rate": 0.00026234445021916704,
+      "loss": 2.0049,
+      "step": 130000
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 2.074753999710083,
+      "learning_rate": 0.00026219735822783567,
+      "loss": 1.988,
+      "step": 130500
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 4.412100791931152,
+      "learning_rate": 0.0002620502662365043,
+      "loss": 1.9851,
+      "step": 131000
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 2.1146128177642822,
+      "learning_rate": 0.00026190317424517294,
+      "loss": 2.0049,
+      "step": 131500
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 7.746240615844727,
+      "learning_rate": 0.0002617560822538415,
+      "loss": 1.9978,
+      "step": 132000
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 1.8406715393066406,
+      "learning_rate": 0.00026160899026251016,
+      "loss": 1.9677,
+      "step": 132500
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 6.467206954956055,
+      "learning_rate": 0.0002614618982711788,
+      "loss": 1.9906,
+      "step": 133000
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 2.756176471710205,
+      "learning_rate": 0.00026131480627984737,
+      "loss": 2.0179,
+      "step": 133500
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 2.753931760787964,
+      "learning_rate": 0.000261167714288516,
+      "loss": 2.0166,
+      "step": 134000
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 32.68937301635742,
+      "learning_rate": 0.00026102062229718464,
+      "loss": 2.002,
+      "step": 134500
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 2.8605151176452637,
+      "learning_rate": 0.0002608735303058533,
+      "loss": 1.9997,
+      "step": 135000
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.3387537002563477,
+      "learning_rate": 0.0002607264383145219,
+      "loss": 1.9992,
+      "step": 135500
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 2.1721653938293457,
+      "learning_rate": 0.0002605793463231905,
+      "loss": 1.9946,
+      "step": 136000
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 15.696208953857422,
+      "learning_rate": 0.0002604322543318591,
+      "loss": 1.9808,
+      "step": 136500
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 2.118614912033081,
+      "learning_rate": 0.00026028516234052776,
+      "loss": 1.9334,
+      "step": 137000
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 4.258530616760254,
+      "learning_rate": 0.00026013807034919634,
+      "loss": 1.9505,
+      "step": 137500
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.8109312057495117,
+      "learning_rate": 0.000259990978357865,
+      "loss": 1.909,
+      "step": 138000
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 1.503454327583313,
+      "learning_rate": 0.0002598438863665336,
+      "loss": 1.9283,
+      "step": 138500
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 5.558799743652344,
+      "learning_rate": 0.00025969679437520224,
+      "loss": 1.959,
+      "step": 139000
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 2.509039878845215,
+      "learning_rate": 0.0002595497023838709,
+      "loss": 1.9111,
+      "step": 139500
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.6668161153793335,
+      "learning_rate": 0.00025940261039253946,
+      "loss": 1.9116,
+      "step": 140000
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.4484660625457764,
+      "learning_rate": 0.0002592555184012081,
+      "loss": 1.9337,
+      "step": 140500
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 2.352369785308838,
+      "learning_rate": 0.00025910842640987673,
+      "loss": 1.922,
+      "step": 141000
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 17.899917602539062,
+      "learning_rate": 0.0002589613344185453,
+      "loss": 1.9144,
+      "step": 141500
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 2.010423421859741,
+      "learning_rate": 0.00025881424242721394,
+      "loss": 1.9909,
+      "step": 142000
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 2.928180694580078,
+      "learning_rate": 0.0002586671504358826,
+      "loss": 1.9295,
+      "step": 142500
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 2.5269832611083984,
+      "learning_rate": 0.0002585200584445512,
+      "loss": 1.9473,
+      "step": 143000
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 4.112135410308838,
+      "learning_rate": 0.00025837296645321985,
+      "loss": 1.9538,
+      "step": 143500
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 2.3555727005004883,
+      "learning_rate": 0.00025822587446188843,
+      "loss": 1.951,
+      "step": 144000
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 2.098503828048706,
+      "learning_rate": 0.00025807878247055706,
+      "loss": 1.9492,
+      "step": 144500
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 2.459561824798584,
+      "learning_rate": 0.0002579316904792257,
+      "loss": 1.9427,
+      "step": 145000
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 2.0252935886383057,
+      "learning_rate": 0.0002577845984878943,
+      "loss": 1.9161,
+      "step": 145500
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 1.384768009185791,
+      "learning_rate": 0.0002576375064965629,
+      "loss": 1.9325,
+      "step": 146000
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 1.6305208206176758,
+      "learning_rate": 0.00025749041450523155,
+      "loss": 1.9892,
+      "step": 146500
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 2.4807193279266357,
+      "learning_rate": 0.0002573433225139002,
+      "loss": 1.9288,
+      "step": 147000
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.798276424407959,
+      "learning_rate": 0.0002571962305225688,
+      "loss": 1.9124,
+      "step": 147500
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 3.1262764930725098,
+      "learning_rate": 0.0002570491385312374,
+      "loss": 1.9356,
+      "step": 148000
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 2.5284624099731445,
+      "learning_rate": 0.00025690204653990603,
+      "loss": 1.9519,
+      "step": 148500
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 2.51084566116333,
+      "learning_rate": 0.00025675495454857467,
+      "loss": 1.948,
+      "step": 149000
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.7908506393432617,
+      "learning_rate": 0.00025660786255724325,
+      "loss": 1.9214,
+      "step": 149500
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.8897744417190552,
+      "learning_rate": 0.0002564607705659119,
+      "loss": 1.9492,
+      "step": 150000
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 2.1504998207092285,
+      "learning_rate": 0.0002563136785745805,
+      "loss": 1.9261,
+      "step": 150500
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.7902778387069702,
+      "learning_rate": 0.00025616658658324915,
+      "loss": 1.9332,
+      "step": 151000
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.6913944482803345,
+      "learning_rate": 0.0002560194945919178,
+      "loss": 1.9391,
+      "step": 151500
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.7668604850769043,
+      "learning_rate": 0.00025587240260058637,
+      "loss": 1.961,
+      "step": 152000
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 2.929547071456909,
+      "learning_rate": 0.000255725310609255,
+      "loss": 1.9578,
+      "step": 152500
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 2.5235610008239746,
+      "learning_rate": 0.00025557821861792363,
+      "loss": 1.8975,
+      "step": 153000
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 5.4155097007751465,
+      "learning_rate": 0.0002554311266265922,
+      "loss": 1.9549,
+      "step": 153500
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 2.9047696590423584,
+      "learning_rate": 0.00025528403463526085,
+      "loss": 1.9409,
+      "step": 154000
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 3.789259910583496,
+      "learning_rate": 0.0002551369426439295,
+      "loss": 1.9267,
+      "step": 154500
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 2.7584846019744873,
+      "learning_rate": 0.0002549898506525981,
+      "loss": 1.9138,
+      "step": 155000
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 4.33261251449585,
+      "learning_rate": 0.00025484275866126675,
+      "loss": 1.9611,
+      "step": 155500
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 3.2369813919067383,
+      "learning_rate": 0.00025469566666993533,
+      "loss": 1.9354,
+      "step": 156000
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 2.6031434535980225,
+      "learning_rate": 0.00025454857467860397,
+      "loss": 1.9172,
+      "step": 156500
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 2.9214913845062256,
+      "learning_rate": 0.0002544014826872726,
+      "loss": 1.9356,
+      "step": 157000
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 2.684860944747925,
+      "learning_rate": 0.00025425439069594124,
+      "loss": 1.9494,
+      "step": 157500
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 2.98620867729187,
+      "learning_rate": 0.0002541072987046098,
+      "loss": 1.9277,
+      "step": 158000
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 1.6060062646865845,
+      "learning_rate": 0.00025396020671327845,
+      "loss": 1.9417,
+      "step": 158500
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 1.4754972457885742,
+      "learning_rate": 0.0002538131147219471,
+      "loss": 1.9691,
+      "step": 159000
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 1.5829949378967285,
+      "learning_rate": 0.0002536660227306157,
+      "loss": 1.9201,
+      "step": 159500
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 2.13678240776062,
+      "learning_rate": 0.00025351893073928436,
+      "loss": 1.9697,
+      "step": 160000
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 2.030682325363159,
+      "learning_rate": 0.00025337183874795294,
+      "loss": 1.9364,
+      "step": 160500
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 2.715879201889038,
+      "learning_rate": 0.00025322474675662157,
+      "loss": 1.9574,
+      "step": 161000
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 48.37791442871094,
+      "learning_rate": 0.0002530776547652902,
+      "loss": 1.9372,
+      "step": 161500
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 3.232931613922119,
+      "learning_rate": 0.0002529305627739588,
+      "loss": 1.9093,
+      "step": 162000
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 2.2022180557250977,
+      "learning_rate": 0.0002527834707826274,
+      "loss": 1.8969,
+      "step": 162500
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 1.3633408546447754,
+      "learning_rate": 0.00025263637879129606,
+      "loss": 1.9476,
+      "step": 163000
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.8852393627166748,
+      "learning_rate": 0.0002524892867999647,
+      "loss": 1.9467,
+      "step": 163500
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 2.2378666400909424,
+      "learning_rate": 0.0002523421948086333,
+      "loss": 1.9216,
+      "step": 164000
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 6.068374156951904,
+      "learning_rate": 0.0002521951028173019,
+      "loss": 1.9188,
+      "step": 164500
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 2.9869394302368164,
+      "learning_rate": 0.00025204801082597054,
+      "loss": 1.8981,
+      "step": 165000
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 5.120233058929443,
+      "learning_rate": 0.0002519009188346392,
+      "loss": 1.9119,
+      "step": 165500
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 24.895925521850586,
+      "learning_rate": 0.00025175382684330776,
+      "loss": 1.9263,
+      "step": 166000
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 4.085075378417969,
+      "learning_rate": 0.0002516067348519764,
+      "loss": 1.9464,
+      "step": 166500
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 2.498352527618408,
+      "learning_rate": 0.000251459642860645,
+      "loss": 1.9349,
+      "step": 167000
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 1.9715830087661743,
+      "learning_rate": 0.00025131255086931366,
+      "loss": 1.9198,
+      "step": 167500
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 6.1134161949157715,
+      "learning_rate": 0.0002511654588779823,
+      "loss": 1.9064,
+      "step": 168000
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 1.4753895998001099,
+      "learning_rate": 0.0002510183668866509,
+      "loss": 1.9277,
+      "step": 168500
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 22.2166805267334,
+      "learning_rate": 0.0002508712748953195,
+      "loss": 1.9105,
+      "step": 169000
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 9.678267478942871,
+      "learning_rate": 0.00025072418290398815,
+      "loss": 1.9361,
+      "step": 169500
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 5.086581230163574,
+      "learning_rate": 0.0002505770909126567,
+      "loss": 1.9035,
+      "step": 170000
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 3.46391224861145,
+      "learning_rate": 0.00025042999892132536,
+      "loss": 1.9096,
+      "step": 170500
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 1.564864158630371,
+      "learning_rate": 0.000250282906929994,
+      "loss": 1.8412,
+      "step": 171000
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 3.917158603668213,
+      "learning_rate": 0.00025013581493866263,
+      "loss": 1.8263,
+      "step": 171500
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 2.3305134773254395,
+      "learning_rate": 0.00024998872294733126,
+      "loss": 1.8263,
+      "step": 172000
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 2.812856435775757,
+      "learning_rate": 0.00024984163095599985,
+      "loss": 1.8238,
+      "step": 172500
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 12.236431121826172,
+      "learning_rate": 0.0002496945389646685,
+      "loss": 1.8475,
+      "step": 173000
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 1.9508650302886963,
+      "learning_rate": 0.0002495474469733371,
+      "loss": 1.8671,
+      "step": 173500
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 4.95379638671875,
+      "learning_rate": 0.0002494003549820057,
+      "loss": 1.8541,
+      "step": 174000
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 2.8690032958984375,
+      "learning_rate": 0.00024925326299067433,
+      "loss": 1.8622,
+      "step": 174500
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 2.8076915740966797,
+      "learning_rate": 0.00024910617099934296,
+      "loss": 1.8394,
+      "step": 175000
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 1.6845248937606812,
+      "learning_rate": 0.0002489590790080116,
+      "loss": 1.8572,
+      "step": 175500
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 1.74410080909729,
+      "learning_rate": 0.00024881198701668023,
+      "loss": 1.8387,
+      "step": 176000
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 2.655266761779785,
+      "learning_rate": 0.0002486648950253488,
+      "loss": 1.8776,
+      "step": 176500
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 2.5884244441986084,
+      "learning_rate": 0.00024851780303401745,
+      "loss": 1.8564,
+      "step": 177000
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 4.314496040344238,
+      "learning_rate": 0.0002483707110426861,
+      "loss": 1.8573,
+      "step": 177500
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 2.15973162651062,
+      "learning_rate": 0.00024822361905135466,
+      "loss": 1.8671,
+      "step": 178000
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 1.7802814245224,
+      "learning_rate": 0.00024807652706002335,
+      "loss": 1.8722,
+      "step": 178500
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 29.04892921447754,
+      "learning_rate": 0.00024792943506869193,
+      "loss": 1.8747,
+      "step": 179000
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 2.0370380878448486,
+      "learning_rate": 0.00024778234307736057,
+      "loss": 1.8615,
+      "step": 179500
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 2.681140661239624,
+      "learning_rate": 0.0002476352510860292,
+      "loss": 1.8603,
+      "step": 180000
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 1.609506368637085,
+      "learning_rate": 0.0002474881590946978,
+      "loss": 1.857,
+      "step": 180500
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 1.4114452600479126,
+      "learning_rate": 0.0002473410671033664,
+      "loss": 1.8849,
+      "step": 181000
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 12.606436729431152,
+      "learning_rate": 0.00024719397511203505,
+      "loss": 1.8576,
+      "step": 181500
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 1.5284700393676758,
+      "learning_rate": 0.00024704688312070363,
+      "loss": 1.8565,
+      "step": 182000
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 2.134824514389038,
+      "learning_rate": 0.0002468997911293723,
+      "loss": 1.8447,
+      "step": 182500
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 2.755667209625244,
+      "learning_rate": 0.0002467526991380409,
+      "loss": 1.8894,
+      "step": 183000
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 1.8626573085784912,
+      "learning_rate": 0.00024660560714670954,
+      "loss": 1.8788,
+      "step": 183500
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 1.2436336278915405,
+      "learning_rate": 0.00024645851515537817,
+      "loss": 1.888,
+      "step": 184000
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 1.6450908184051514,
+      "learning_rate": 0.00024631142316404675,
+      "loss": 1.8576,
+      "step": 184500
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 2.5818614959716797,
+      "learning_rate": 0.0002461643311727154,
+      "loss": 1.8634,
+      "step": 185000
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 2.4842381477355957,
+      "learning_rate": 0.000246017239181384,
+      "loss": 1.8777,
+      "step": 185500
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 5.475472927093506,
+      "learning_rate": 0.0002458701471900526,
+      "loss": 1.8537,
+      "step": 186000
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 6.852906227111816,
+      "learning_rate": 0.0002457230551987213,
+      "loss": 1.863,
+      "step": 186500
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 2.8549883365631104,
+      "learning_rate": 0.00024557596320738987,
+      "loss": 1.8643,
+      "step": 187000
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 4.323022365570068,
+      "learning_rate": 0.0002454288712160585,
+      "loss": 1.8489,
+      "step": 187500
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 1.4015731811523438,
+      "learning_rate": 0.00024528177922472714,
+      "loss": 1.879,
+      "step": 188000
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 2.5605356693267822,
+      "learning_rate": 0.0002451346872333957,
+      "loss": 1.8564,
+      "step": 188500
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 1.567606806755066,
+      "learning_rate": 0.00024498759524206436,
+      "loss": 1.8902,
+      "step": 189000
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 1.4582362174987793,
+      "learning_rate": 0.000244840503250733,
+      "loss": 1.8659,
+      "step": 189500
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 1.8250012397766113,
+      "learning_rate": 0.00024469341125940157,
+      "loss": 1.8558,
+      "step": 190000
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 6.115236759185791,
+      "learning_rate": 0.00024454631926807026,
+      "loss": 1.839,
+      "step": 190500
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 4.505608081817627,
+      "learning_rate": 0.00024439922727673884,
+      "loss": 1.846,
+      "step": 191000
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 2.1621007919311523,
+      "learning_rate": 0.0002442521352854075,
+      "loss": 1.8287,
+      "step": 191500
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 2.337688684463501,
+      "learning_rate": 0.0002441050432940761,
+      "loss": 1.88,
+      "step": 192000
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 1.921249508857727,
+      "learning_rate": 0.00024395795130274472,
+      "loss": 1.8345,
+      "step": 192500
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 1.396031379699707,
+      "learning_rate": 0.00024381085931141332,
+      "loss": 1.8527,
+      "step": 193000
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 2.6552846431732178,
+      "learning_rate": 0.00024366376732008196,
+      "loss": 1.8628,
+      "step": 193500
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 4.992696285247803,
+      "learning_rate": 0.00024351667532875057,
+      "loss": 1.8727,
+      "step": 194000
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 4.128008842468262,
+      "learning_rate": 0.00024336958333741923,
+      "loss": 1.8677,
+      "step": 194500
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 2.2797539234161377,
+      "learning_rate": 0.00024322249134608784,
+      "loss": 1.8495,
+      "step": 195000
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 1.987032175064087,
+      "learning_rate": 0.00024307539935475644,
+      "loss": 1.8555,
+      "step": 195500
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 2.1700329780578613,
+      "learning_rate": 0.00024292830736342508,
+      "loss": 1.8745,
+      "step": 196000
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 2.0741491317749023,
+      "learning_rate": 0.00024278121537209369,
+      "loss": 1.8859,
+      "step": 196500
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 3.1270885467529297,
+      "learning_rate": 0.0002426341233807623,
+      "loss": 1.8778,
+      "step": 197000
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 2.7290618419647217,
+      "learning_rate": 0.00024248703138943093,
+      "loss": 1.8418,
+      "step": 197500
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 1.5647644996643066,
+      "learning_rate": 0.00024233993939809954,
+      "loss": 1.8519,
+      "step": 198000
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 2.4231462478637695,
+      "learning_rate": 0.0002421928474067682,
+      "loss": 1.868,
+      "step": 198500
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 3.1139655113220215,
+      "learning_rate": 0.0002420457554154368,
+      "loss": 1.8255,
+      "step": 199000
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 4.595509052276611,
+      "learning_rate": 0.0002418986634241054,
+      "loss": 1.8626,
+      "step": 199500
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 7.253945827484131,
+      "learning_rate": 0.00024175157143277405,
+      "loss": 1.8714,
+      "step": 200000
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 1.5077595710754395,
+      "learning_rate": 0.00024160447944144266,
+      "loss": 1.8743,
+      "step": 200500
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 5.791823387145996,
+      "learning_rate": 0.00024145738745011126,
+      "loss": 1.8499,
+      "step": 201000
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 2.558816909790039,
+      "learning_rate": 0.0002413102954587799,
+      "loss": 1.8757,
+      "step": 201500
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 2.4598610401153564,
+      "learning_rate": 0.0002411632034674485,
+      "loss": 1.8396,
+      "step": 202000
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 1.2680063247680664,
+      "learning_rate": 0.00024101611147611717,
+      "loss": 1.9265,
+      "step": 202500
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 13.615561485290527,
+      "learning_rate": 0.00024086901948478577,
+      "loss": 1.833,
+      "step": 203000
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 1.367218017578125,
+      "learning_rate": 0.00024072192749345438,
+      "loss": 1.836,
+      "step": 203500
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 1.8230597972869873,
+      "learning_rate": 0.00024057483550212302,
+      "loss": 1.8833,
+      "step": 204000
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 3.095012903213501,
+      "learning_rate": 0.00024042774351079162,
+      "loss": 1.8106,
+      "step": 204500
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 1.357537031173706,
+      "learning_rate": 0.00024028065151946023,
+      "loss": 1.809,
+      "step": 205000
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 1.8202615976333618,
+      "learning_rate": 0.00024013355952812887,
+      "loss": 1.788,
+      "step": 205500
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 2.0572142601013184,
+      "learning_rate": 0.00023998646753679747,
+      "loss": 1.7786,
+      "step": 206000
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 2.056692123413086,
+      "learning_rate": 0.00023983937554546614,
+      "loss": 1.801,
+      "step": 206500
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.6310349702835083,
+      "learning_rate": 0.00023969228355413474,
+      "loss": 1.794,
+      "step": 207000
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 2.9542150497436523,
+      "learning_rate": 0.00023954519156280335,
+      "loss": 1.8013,
+      "step": 207500
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 2.0560832023620605,
+      "learning_rate": 0.00023939809957147199,
+      "loss": 1.809,
+      "step": 208000
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 2.991140842437744,
+      "learning_rate": 0.0002392510075801406,
+      "loss": 1.8122,
+      "step": 208500
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 4.056896686553955,
+      "learning_rate": 0.0002391039155888092,
+      "loss": 1.7895,
+      "step": 209000
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 1.4278576374053955,
+      "learning_rate": 0.00023895682359747784,
+      "loss": 1.7778,
+      "step": 209500
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 2.918614387512207,
+      "learning_rate": 0.00023880973160614644,
+      "loss": 1.7882,
+      "step": 210000
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 6.002877712249756,
+      "learning_rate": 0.0002386626396148151,
+      "loss": 1.8129,
+      "step": 210500
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 1.8039432764053345,
+      "learning_rate": 0.0002385155476234837,
+      "loss": 1.7879,
+      "step": 211000
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 2.1559925079345703,
+      "learning_rate": 0.00023836845563215232,
+      "loss": 1.825,
+      "step": 211500
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 1.6617883443832397,
+      "learning_rate": 0.00023822136364082095,
+      "loss": 1.8048,
+      "step": 212000
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 3.7312443256378174,
+      "learning_rate": 0.00023807427164948956,
+      "loss": 1.7759,
+      "step": 212500
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 4.757487773895264,
+      "learning_rate": 0.0002379271796581582,
+      "loss": 1.8357,
+      "step": 213000
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 1.4285889863967896,
+      "learning_rate": 0.0002377800876668268,
+      "loss": 1.8127,
+      "step": 213500
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 1.7722175121307373,
+      "learning_rate": 0.0002376329956754954,
+      "loss": 1.79,
+      "step": 214000
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 2.17244291305542,
+      "learning_rate": 0.00023748590368416407,
+      "loss": 1.7949,
+      "step": 214500
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 2.136143922805786,
+      "learning_rate": 0.00023733881169283268,
+      "loss": 1.8199,
+      "step": 215000
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 1.36685049533844,
+      "learning_rate": 0.00023719171970150132,
+      "loss": 1.8101,
+      "step": 215500
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 1.5481481552124023,
+      "learning_rate": 0.00023704462771016992,
+      "loss": 1.819,
+      "step": 216000
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 2.1855366230010986,
+      "learning_rate": 0.00023689753571883853,
+      "loss": 1.8162,
+      "step": 216500
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 1.5019465684890747,
+      "learning_rate": 0.00023675044372750717,
+      "loss": 1.7929,
+      "step": 217000
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 1.4323623180389404,
+      "learning_rate": 0.00023660335173617577,
+      "loss": 1.7664,
+      "step": 217500
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 3.942918539047241,
+      "learning_rate": 0.00023645625974484438,
+      "loss": 1.8165,
+      "step": 218000
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 1.4521820545196533,
+      "learning_rate": 0.00023630916775351304,
+      "loss": 1.7966,
+      "step": 218500
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 3.9033968448638916,
+      "learning_rate": 0.00023616207576218165,
+      "loss": 1.7952,
+      "step": 219000
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 2.7818336486816406,
+      "learning_rate": 0.00023601498377085029,
+      "loss": 1.7963,
+      "step": 219500
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 4.406651496887207,
+      "learning_rate": 0.0002358678917795189,
+      "loss": 1.7901,
+      "step": 220000
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 2.1108040809631348,
+      "learning_rate": 0.0002357207997881875,
+      "loss": 1.7887,
+      "step": 220500
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 6.802318096160889,
+      "learning_rate": 0.00023557370779685613,
+      "loss": 1.7811,
+      "step": 221000
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 2.7774646282196045,
+      "learning_rate": 0.00023542661580552474,
+      "loss": 1.782,
+      "step": 221500
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 1.5817331075668335,
+      "learning_rate": 0.00023527952381419335,
+      "loss": 1.7883,
+      "step": 222000
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 1.3639189004898071,
+      "learning_rate": 0.000235132431822862,
+      "loss": 1.8137,
+      "step": 222500
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 3.018841505050659,
+      "learning_rate": 0.00023498533983153062,
+      "loss": 1.8012,
+      "step": 223000
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 1.6388349533081055,
+      "learning_rate": 0.00023483824784019925,
+      "loss": 1.7776,
+      "step": 223500
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 2.2828805446624756,
+      "learning_rate": 0.00023469115584886786,
+      "loss": 1.793,
+      "step": 224000
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 1.5646781921386719,
+      "learning_rate": 0.00023454406385753647,
+      "loss": 1.8049,
+      "step": 224500
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 2.5620057582855225,
+      "learning_rate": 0.0002343969718662051,
+      "loss": 1.8082,
+      "step": 225000
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 1.1866906881332397,
+      "learning_rate": 0.0002342498798748737,
+      "loss": 1.8197,
+      "step": 225500
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 8.060154914855957,
+      "learning_rate": 0.00023410278788354232,
+      "loss": 1.7765,
+      "step": 226000
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 2.2780892848968506,
+      "learning_rate": 0.00023395569589221098,
+      "loss": 1.7903,
+      "step": 226500
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 8.265579223632812,
+      "learning_rate": 0.0002338086039008796,
+      "loss": 1.8143,
+      "step": 227000
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 1.2793028354644775,
+      "learning_rate": 0.00023366151190954822,
+      "loss": 1.7729,
+      "step": 227500
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 1.7103880643844604,
+      "learning_rate": 0.00023351441991821683,
+      "loss": 1.8019,
+      "step": 228000
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 6.143857479095459,
+      "learning_rate": 0.00023336732792688544,
+      "loss": 1.7781,
+      "step": 228500
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 3.1208572387695312,
+      "learning_rate": 0.00023322023593555407,
+      "loss": 1.791,
+      "step": 229000
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 1.774464726448059,
+      "learning_rate": 0.00023307314394422268,
+      "loss": 1.7596,
+      "step": 229500
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 1.9884730577468872,
+      "learning_rate": 0.0002329260519528913,
+      "loss": 1.7842,
+      "step": 230000
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.9491617679595947,
+      "learning_rate": 0.00023277895996155995,
+      "loss": 1.8013,
+      "step": 230500
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 1.6505547761917114,
+      "learning_rate": 0.00023263186797022856,
+      "loss": 1.7913,
+      "step": 231000
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 1.262868046760559,
+      "learning_rate": 0.0002324847759788972,
+      "loss": 1.8212,
+      "step": 231500
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 1.1688213348388672,
+      "learning_rate": 0.0002323376839875658,
+      "loss": 1.815,
+      "step": 232000
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 1.4348254203796387,
+      "learning_rate": 0.0002321905919962344,
+      "loss": 1.8335,
+      "step": 232500
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 1.977734923362732,
+      "learning_rate": 0.00023204350000490304,
+      "loss": 1.8231,
+      "step": 233000
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 12.462953567504883,
+      "learning_rate": 0.00023189640801357165,
+      "loss": 1.8086,
+      "step": 233500
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 1.9486029148101807,
+      "learning_rate": 0.00023174931602224028,
+      "loss": 1.7932,
+      "step": 234000
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 6.036319732666016,
+      "learning_rate": 0.00023160222403090892,
+      "loss": 1.7906,
+      "step": 234500
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 1.6775847673416138,
+      "learning_rate": 0.00023145513203957753,
+      "loss": 1.8026,
+      "step": 235000
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 2.1960055828094482,
+      "learning_rate": 0.00023130804004824616,
+      "loss": 1.8093,
+      "step": 235500
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 4.2877197265625,
+      "learning_rate": 0.00023116094805691477,
+      "loss": 1.8198,
+      "step": 236000
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 1.4908256530761719,
+      "learning_rate": 0.0002310138560655834,
+      "loss": 1.7882,
+      "step": 236500
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 5.148675918579102,
+      "learning_rate": 0.000230866764074252,
+      "loss": 1.786,
+      "step": 237000
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 5.5215744972229,
+      "learning_rate": 0.00023071967208292062,
+      "loss": 1.8168,
+      "step": 237500
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 21.99406623840332,
+      "learning_rate": 0.00023057258009158928,
+      "loss": 1.7993,
+      "step": 238000
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 3.8185484409332275,
+      "learning_rate": 0.0002304254881002579,
+      "loss": 1.8105,
+      "step": 238500
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 7.100237846374512,
+      "learning_rate": 0.00023027839610892652,
+      "loss": 1.7102,
+      "step": 239000
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 5.007193088531494,
+      "learning_rate": 0.00023013130411759513,
+      "loss": 1.7293,
+      "step": 239500
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 1.6522769927978516,
+      "learning_rate": 0.00022998421212626374,
+      "loss": 1.7199,
+      "step": 240000
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 1.7635401487350464,
+      "learning_rate": 0.00022983712013493237,
+      "loss": 1.7304,
+      "step": 240500
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 2.441962957382202,
+      "learning_rate": 0.00022969002814360098,
+      "loss": 1.7291,
+      "step": 241000
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 1.6155414581298828,
+      "learning_rate": 0.0002295429361522696,
+      "loss": 1.7477,
+      "step": 241500
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 1.3782049417495728,
+      "learning_rate": 0.00022939584416093825,
+      "loss": 1.7552,
+      "step": 242000
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.6964011192321777,
+      "learning_rate": 0.00022924875216960686,
+      "loss": 1.7233,
+      "step": 242500
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 1.9475128650665283,
+      "learning_rate": 0.0002291016601782755,
+      "loss": 1.7603,
+      "step": 243000
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 2.674274444580078,
+      "learning_rate": 0.0002289545681869441,
+      "loss": 1.7472,
+      "step": 243500
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 1.6163575649261475,
+      "learning_rate": 0.0002288074761956127,
+      "loss": 1.7754,
+      "step": 244000
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 1.3212227821350098,
+      "learning_rate": 0.00022866038420428134,
+      "loss": 1.7828,
+      "step": 244500
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 1.9953806400299072,
+      "learning_rate": 0.00022851329221294995,
+      "loss": 1.7289,
+      "step": 245000
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 1.5477960109710693,
+      "learning_rate": 0.00022836620022161856,
+      "loss": 1.7651,
+      "step": 245500
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 16.499107360839844,
+      "learning_rate": 0.00022821910823028722,
+      "loss": 1.7207,
+      "step": 246000
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 2.4671411514282227,
+      "learning_rate": 0.00022807201623895583,
+      "loss": 1.7438,
+      "step": 246500
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 4.774916648864746,
+      "learning_rate": 0.00022792492424762446,
+      "loss": 1.7721,
+      "step": 247000
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 2.282515287399292,
+      "learning_rate": 0.00022777783225629307,
+      "loss": 1.7145,
+      "step": 247500
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 1.9206039905548096,
+      "learning_rate": 0.00022763074026496168,
+      "loss": 1.7314,
+      "step": 248000
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 2.7325820922851562,
+      "learning_rate": 0.0002274836482736303,
+      "loss": 1.725,
+      "step": 248500
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 1.7279201745986938,
+      "learning_rate": 0.00022733655628229892,
+      "loss": 1.7095,
+      "step": 249000
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 1.8310534954071045,
+      "learning_rate": 0.00022718946429096753,
+      "loss": 1.7325,
+      "step": 249500
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 5.919315338134766,
+      "learning_rate": 0.0002270423722996362,
+      "loss": 1.7301,
+      "step": 250000
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 2.3758251667022705,
+      "learning_rate": 0.0002268952803083048,
+      "loss": 1.7529,
+      "step": 250500
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 3.524482011795044,
+      "learning_rate": 0.00022674818831697343,
+      "loss": 1.7822,
+      "step": 251000
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 2.1281306743621826,
+      "learning_rate": 0.00022660109632564204,
+      "loss": 1.7315,
+      "step": 251500
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 1.6832419633865356,
+      "learning_rate": 0.00022645400433431064,
+      "loss": 1.7553,
+      "step": 252000
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 13.24813175201416,
+      "learning_rate": 0.00022630691234297928,
+      "loss": 1.7623,
+      "step": 252500
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 1.5083931684494019,
+      "learning_rate": 0.0002261598203516479,
+      "loss": 1.7579,
+      "step": 253000
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 3.207942247390747,
+      "learning_rate": 0.00022601272836031652,
+      "loss": 1.7248,
+      "step": 253500
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 7.897548675537109,
+      "learning_rate": 0.00022586563636898516,
+      "loss": 1.7309,
+      "step": 254000
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 2.2789628505706787,
+      "learning_rate": 0.00022571854437765376,
+      "loss": 1.7444,
+      "step": 254500
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 2.129913568496704,
+      "learning_rate": 0.0002255714523863224,
+      "loss": 1.7505,
+      "step": 255000
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 2.0121564865112305,
+      "learning_rate": 0.000225424360394991,
+      "loss": 1.8085,
+      "step": 255500
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 3.321162462234497,
+      "learning_rate": 0.00022527726840365964,
+      "loss": 1.7601,
+      "step": 256000
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 8.002218246459961,
+      "learning_rate": 0.00022513017641232825,
+      "loss": 1.7719,
+      "step": 256500
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 1.610587477684021,
+      "learning_rate": 0.00022498308442099686,
+      "loss": 1.7373,
+      "step": 257000
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 5.272850513458252,
+      "learning_rate": 0.0002248359924296655,
+      "loss": 1.7481,
+      "step": 257500
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 1.519241452217102,
+      "learning_rate": 0.00022468890043833413,
+      "loss": 1.7722,
+      "step": 258000
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 3.9072492122650146,
+      "learning_rate": 0.00022454180844700276,
+      "loss": 1.7425,
+      "step": 258500
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 16.145322799682617,
+      "learning_rate": 0.00022439471645567137,
+      "loss": 1.7645,
+      "step": 259000
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 1.7309913635253906,
+      "learning_rate": 0.00022424762446433998,
+      "loss": 1.7388,
+      "step": 259500
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 1.4854837656021118,
+      "learning_rate": 0.0002241005324730086,
+      "loss": 1.7728,
+      "step": 260000
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 1.9065077304840088,
+      "learning_rate": 0.00022395344048167722,
+      "loss": 1.7306,
+      "step": 260500
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 2.6869423389434814,
+      "learning_rate": 0.00022380634849034582,
+      "loss": 1.751,
+      "step": 261000
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 2.3300232887268066,
+      "learning_rate": 0.00022365925649901446,
+      "loss": 1.7643,
+      "step": 261500
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 1.6109992265701294,
+      "learning_rate": 0.0002235121645076831,
+      "loss": 1.7848,
+      "step": 262000
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 3.2768895626068115,
+      "learning_rate": 0.00022336507251635173,
+      "loss": 1.7638,
+      "step": 262500
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 4.605926036834717,
+      "learning_rate": 0.00022321798052502034,
+      "loss": 1.7801,
+      "step": 263000
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 1.451865553855896,
+      "learning_rate": 0.00022307088853368894,
+      "loss": 1.7229,
+      "step": 263500
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 1.938750982284546,
+      "learning_rate": 0.00022292379654235758,
+      "loss": 1.7575,
+      "step": 264000
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 8.690546989440918,
+      "learning_rate": 0.00022277670455102619,
+      "loss": 1.7572,
+      "step": 264500
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 6.99992036819458,
+      "learning_rate": 0.0002226296125596948,
+      "loss": 1.7912,
+      "step": 265000
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.375183343887329,
+      "learning_rate": 0.00022248252056836343,
+      "loss": 1.7588,
+      "step": 265500
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 1.4782127141952515,
+      "learning_rate": 0.00022233542857703206,
+      "loss": 1.7929,
+      "step": 266000
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 3.282301187515259,
+      "learning_rate": 0.0002221883365857007,
+      "loss": 1.7855,
+      "step": 266500
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 2.2392537593841553,
+      "learning_rate": 0.0002220412445943693,
+      "loss": 1.7603,
+      "step": 267000
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 3.2060225009918213,
+      "learning_rate": 0.0002218941526030379,
+      "loss": 1.7416,
+      "step": 267500
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 2.6562771797180176,
+      "learning_rate": 0.00022174706061170655,
+      "loss": 1.766,
+      "step": 268000
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 3.6191391944885254,
+      "learning_rate": 0.00022159996862037516,
+      "loss": 1.7746,
+      "step": 268500
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 1.9865895509719849,
+      "learning_rate": 0.00022145287662904376,
+      "loss": 1.7841,
+      "step": 269000
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 1.3776191473007202,
+      "learning_rate": 0.0002213057846377124,
+      "loss": 1.7709,
+      "step": 269500
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 2.238267183303833,
+      "learning_rate": 0.00022115869264638103,
+      "loss": 1.7764,
+      "step": 270000
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 3.3425824642181396,
+      "learning_rate": 0.00022101160065504967,
+      "loss": 1.7901,
+      "step": 270500
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 2.183436632156372,
+      "learning_rate": 0.00022086450866371827,
+      "loss": 1.757,
+      "step": 271000
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 3.393548011779785,
+      "learning_rate": 0.00022071741667238688,
+      "loss": 1.7538,
+      "step": 271500
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 4.727015018463135,
+      "learning_rate": 0.00022057032468105552,
+      "loss": 1.7585,
+      "step": 272000
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 2.0307464599609375,
+      "learning_rate": 0.00022042323268972412,
+      "loss": 1.7659,
+      "step": 272500
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 4.098311424255371,
+      "learning_rate": 0.00022027614069839273,
+      "loss": 1.7191,
+      "step": 273000
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 2.290457248687744,
+      "learning_rate": 0.00022012904870706137,
+      "loss": 1.7206,
+      "step": 273500
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 2.1348979473114014,
+      "learning_rate": 0.00021998195671573,
+      "loss": 1.6935,
+      "step": 274000
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 2.5333049297332764,
+      "learning_rate": 0.00021983486472439864,
+      "loss": 1.6709,
+      "step": 274500
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 2.1333959102630615,
+      "learning_rate": 0.00021968777273306724,
+      "loss": 1.7318,
+      "step": 275000
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 1.6256059408187866,
+      "learning_rate": 0.00021954068074173585,
+      "loss": 1.7007,
+      "step": 275500
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 1.4450695514678955,
+      "learning_rate": 0.00021939358875040449,
+      "loss": 1.7038,
+      "step": 276000
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 4.230172634124756,
+      "learning_rate": 0.0002192464967590731,
+      "loss": 1.6738,
+      "step": 276500
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 2.046879529953003,
+      "learning_rate": 0.00021909940476774173,
+      "loss": 1.6658,
+      "step": 277000
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 2.2633895874023438,
+      "learning_rate": 0.00021895231277641034,
+      "loss": 1.6929,
+      "step": 277500
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 34.2008171081543,
+      "learning_rate": 0.00021880522078507897,
+      "loss": 1.6956,
+      "step": 278000
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 11.11546516418457,
+      "learning_rate": 0.0002186581287937476,
+      "loss": 1.7103,
+      "step": 278500
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 5.735045909881592,
+      "learning_rate": 0.0002185110368024162,
+      "loss": 1.6991,
+      "step": 279000
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 2.9397599697113037,
+      "learning_rate": 0.00021836394481108485,
+      "loss": 1.6909,
+      "step": 279500
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 1.9226768016815186,
+      "learning_rate": 0.00021821685281975345,
+      "loss": 1.6894,
+      "step": 280000
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 2.0962398052215576,
+      "learning_rate": 0.00021806976082842206,
+      "loss": 1.7094,
+      "step": 280500
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 13.345786094665527,
+      "learning_rate": 0.0002179226688370907,
+      "loss": 1.6927,
+      "step": 281000
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 3.1707746982574463,
+      "learning_rate": 0.0002177755768457593,
+      "loss": 1.7004,
+      "step": 281500
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 3.3897292613983154,
+      "learning_rate": 0.00021762848485442797,
+      "loss": 1.7094,
+      "step": 282000
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 5.791382312774658,
+      "learning_rate": 0.00021748139286309657,
+      "loss": 1.7279,
+      "step": 282500
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 2.3419294357299805,
+      "learning_rate": 0.00021733430087176518,
+      "loss": 1.7053,
+      "step": 283000
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 1.9053447246551514,
+      "learning_rate": 0.00021718720888043382,
+      "loss": 1.7225,
+      "step": 283500
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 1.8822693824768066,
+      "learning_rate": 0.00021704011688910242,
+      "loss": 1.6968,
+      "step": 284000
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 1.4814122915267944,
+      "learning_rate": 0.00021689302489777103,
+      "loss": 1.6697,
+      "step": 284500
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 8.579392433166504,
+      "learning_rate": 0.00021674593290643967,
+      "loss": 1.7206,
+      "step": 285000
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 1.9921379089355469,
+      "learning_rate": 0.00021659884091510827,
+      "loss": 1.7113,
+      "step": 285500
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 2.2425730228424072,
+      "learning_rate": 0.00021645174892377694,
+      "loss": 1.6893,
+      "step": 286000
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 4.465683460235596,
+      "learning_rate": 0.00021630465693244554,
+      "loss": 1.6997,
+      "step": 286500
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 11.377062797546387,
+      "learning_rate": 0.00021615756494111415,
+      "loss": 1.691,
+      "step": 287000
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 4.9282636642456055,
+      "learning_rate": 0.00021601047294978279,
+      "loss": 1.6722,
+      "step": 287500
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 4.008066177368164,
+      "learning_rate": 0.0002158633809584514,
+      "loss": 1.6498,
+      "step": 288000
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 1.7437248229980469,
+      "learning_rate": 0.00021571628896712,
+      "loss": 1.7336,
+      "step": 288500
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 1.627820372581482,
+      "learning_rate": 0.00021556919697578864,
+      "loss": 1.6973,
+      "step": 289000
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 5.1567254066467285,
+      "learning_rate": 0.00021542210498445724,
+      "loss": 1.6974,
+      "step": 289500
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 1.3468036651611328,
+      "learning_rate": 0.0002152750129931259,
+      "loss": 1.6815,
+      "step": 290000
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 1.6024253368377686,
+      "learning_rate": 0.0002151279210017945,
+      "loss": 1.6784,
+      "step": 290500
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 1.915425181388855,
+      "learning_rate": 0.00021498082901046312,
+      "loss": 1.7297,
+      "step": 291000
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 2.750544548034668,
+      "learning_rate": 0.00021483373701913175,
+      "loss": 1.717,
+      "step": 291500
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 8.173510551452637,
+      "learning_rate": 0.00021468664502780036,
+      "loss": 1.6971,
+      "step": 292000
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 2.2209818363189697,
+      "learning_rate": 0.00021453955303646897,
+      "loss": 1.6803,
+      "step": 292500
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 4.283031463623047,
+      "learning_rate": 0.0002143924610451376,
+      "loss": 1.7101,
+      "step": 293000
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 2.29679799079895,
+      "learning_rate": 0.0002142453690538062,
+      "loss": 1.7021,
+      "step": 293500
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 3.301649570465088,
+      "learning_rate": 0.00021409827706247487,
+      "loss": 1.727,
+      "step": 294000
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 38.071407318115234,
+      "learning_rate": 0.00021395118507114348,
+      "loss": 1.7318,
+      "step": 294500
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 1.9748557806015015,
+      "learning_rate": 0.0002138040930798121,
+      "loss": 1.7327,
+      "step": 295000
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 1.8591829538345337,
+      "learning_rate": 0.00021365700108848072,
+      "loss": 1.6944,
+      "step": 295500
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 6.686271667480469,
+      "learning_rate": 0.00021350990909714933,
+      "loss": 1.7289,
+      "step": 296000
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 2.0325803756713867,
+      "learning_rate": 0.00021336281710581794,
+      "loss": 1.7074,
+      "step": 296500
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 2.5102651119232178,
+      "learning_rate": 0.00021321572511448657,
+      "loss": 1.7061,
+      "step": 297000
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 5.305497646331787,
+      "learning_rate": 0.0002130686331231552,
+      "loss": 1.7047,
+      "step": 297500
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 2.878256320953369,
+      "learning_rate": 0.00021292154113182384,
+      "loss": 1.7109,
+      "step": 298000
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 1.7807819843292236,
+      "learning_rate": 0.00021277444914049245,
+      "loss": 1.7136,
+      "step": 298500
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 1.8971943855285645,
+      "learning_rate": 0.00021262735714916106,
+      "loss": 1.7319,
+      "step": 299000
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 2.650453805923462,
+      "learning_rate": 0.0002124802651578297,
+      "loss": 1.7436,
+      "step": 299500
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 3.341278076171875,
+      "learning_rate": 0.0002123331731664983,
+      "loss": 1.6982,
+      "step": 300000
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 4.539509296417236,
+      "learning_rate": 0.00021218608117516693,
+      "loss": 1.7149,
+      "step": 300500
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 1.6525373458862305,
+      "learning_rate": 0.00021203898918383554,
+      "loss": 1.7154,
+      "step": 301000
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 2.1805810928344727,
+      "learning_rate": 0.00021189189719250418,
+      "loss": 1.7244,
+      "step": 301500
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 1.623687744140625,
+      "learning_rate": 0.0002117448052011728,
+      "loss": 1.7109,
+      "step": 302000
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.8576191663742065,
+      "learning_rate": 0.00021159771320984142,
+      "loss": 1.71,
+      "step": 302500
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 2.7003326416015625,
+      "learning_rate": 0.00021145062121851005,
+      "loss": 1.717,
+      "step": 303000
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 1.5555850267410278,
+      "learning_rate": 0.00021130352922717866,
+      "loss": 1.6771,
+      "step": 303500
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 1.8485440015792847,
+      "learning_rate": 0.00021115643723584727,
+      "loss": 1.6907,
+      "step": 304000
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 1.9917840957641602,
+      "learning_rate": 0.0002110093452445159,
+      "loss": 1.6929,
+      "step": 304500
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 2.54939866065979,
+      "learning_rate": 0.0002108622532531845,
+      "loss": 1.7211,
+      "step": 305000
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 4.3445234298706055,
+      "learning_rate": 0.00021071516126185317,
+      "loss": 1.7038,
+      "step": 305500
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 2.9025015830993652,
+      "learning_rate": 0.00021056806927052178,
+      "loss": 1.6987,
+      "step": 306000
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 3.951500177383423,
+      "learning_rate": 0.0002104209772791904,
+      "loss": 1.7063,
+      "step": 306500
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 2.2812907695770264,
+      "learning_rate": 0.00021027388528785902,
+      "loss": 1.6754,
+      "step": 307000
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 1.313977837562561,
+      "learning_rate": 0.00021012679329652763,
+      "loss": 1.6384,
+      "step": 307500
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 1.1051950454711914,
+      "learning_rate": 0.00020997970130519624,
+      "loss": 1.6372,
+      "step": 308000
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 2.6401305198669434,
+      "learning_rate": 0.00020983260931386487,
+      "loss": 1.6535,
+      "step": 308500
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 1.477562665939331,
+      "learning_rate": 0.00020968551732253348,
+      "loss": 1.6342,
+      "step": 309000
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 1.7968777418136597,
+      "learning_rate": 0.00020953842533120214,
+      "loss": 1.6163,
+      "step": 309500
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 1.8777967691421509,
+      "learning_rate": 0.00020939133333987075,
+      "loss": 1.6466,
+      "step": 310000
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 1.970590353012085,
+      "learning_rate": 0.00020924424134853936,
+      "loss": 1.6481,
+      "step": 310500
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 11.109952926635742,
+      "learning_rate": 0.000209097149357208,
+      "loss": 1.6634,
+      "step": 311000
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 1.5415191650390625,
+      "learning_rate": 0.0002089500573658766,
+      "loss": 1.637,
+      "step": 311500
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 2.2349135875701904,
+      "learning_rate": 0.0002088029653745452,
+      "loss": 1.6373,
+      "step": 312000
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 2.3896191120147705,
+      "learning_rate": 0.00020865587338321384,
+      "loss": 1.6318,
+      "step": 312500
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 1.5920429229736328,
+      "learning_rate": 0.00020850878139188245,
+      "loss": 1.6383,
+      "step": 313000
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 1.485474944114685,
+      "learning_rate": 0.0002083616894005511,
+      "loss": 1.6392,
+      "step": 313500
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 9.920788764953613,
+      "learning_rate": 0.00020821459740921972,
+      "loss": 1.6407,
+      "step": 314000
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 5.970518589019775,
+      "learning_rate": 0.00020806750541788833,
+      "loss": 1.6389,
+      "step": 314500
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 1.5371183156967163,
+      "learning_rate": 0.00020792041342655696,
+      "loss": 1.6485,
+      "step": 315000
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 2.242612600326538,
+      "learning_rate": 0.00020777332143522557,
+      "loss": 1.6344,
+      "step": 315500
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.137301445007324,
+      "learning_rate": 0.00020762622944389418,
+      "loss": 1.6269,
+      "step": 316000
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 3.391427993774414,
+      "learning_rate": 0.0002074791374525628,
+      "loss": 1.6575,
+      "step": 316500
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 7.92528772354126,
+      "learning_rate": 0.00020733204546123142,
+      "loss": 1.6333,
+      "step": 317000
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 1.0485262870788574,
+      "learning_rate": 0.00020718495346990008,
+      "loss": 1.6574,
+      "step": 317500
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 1.6511611938476562,
+      "learning_rate": 0.0002070378614785687,
+      "loss": 1.6474,
+      "step": 318000
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 1.3929743766784668,
+      "learning_rate": 0.0002068907694872373,
+      "loss": 1.6484,
+      "step": 318500
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 1.0993067026138306,
+      "learning_rate": 0.00020674367749590593,
+      "loss": 1.6312,
+      "step": 319000
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 2.8097622394561768,
+      "learning_rate": 0.00020659658550457454,
+      "loss": 1.6438,
+      "step": 319500
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 2.274348020553589,
+      "learning_rate": 0.00020644949351324314,
+      "loss": 1.6258,
+      "step": 320000
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 1.9707189798355103,
+      "learning_rate": 0.00020630240152191178,
+      "loss": 1.6619,
+      "step": 320500
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 2.7805778980255127,
+      "learning_rate": 0.0002061553095305804,
+      "loss": 1.6537,
+      "step": 321000
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 2.856705665588379,
+      "learning_rate": 0.00020600821753924905,
+      "loss": 1.6745,
+      "step": 321500
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 1.3102498054504395,
+      "learning_rate": 0.00020586112554791766,
+      "loss": 1.6662,
+      "step": 322000
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 4.1165690422058105,
+      "learning_rate": 0.00020571403355658626,
+      "loss": 1.6531,
+      "step": 322500
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 5.66050386428833,
+      "learning_rate": 0.0002055669415652549,
+      "loss": 1.669,
+      "step": 323000
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 1.3612234592437744,
+      "learning_rate": 0.0002054198495739235,
+      "loss": 1.642,
+      "step": 323500
+    },
+    {
+      "epoch": 9.51,
+      "grad_norm": 2.5399467945098877,
+      "learning_rate": 0.00020527275758259214,
+      "loss": 1.677,
+      "step": 324000
+    },
+    {
+      "epoch": 9.53,
+      "grad_norm": 5.590959548950195,
+      "learning_rate": 0.00020512566559126075,
+      "loss": 1.6673,
+      "step": 324500
+    },
+    {
+      "epoch": 9.54,
+      "grad_norm": 1.6591867208480835,
+      "learning_rate": 0.00020497857359992936,
+      "loss": 1.6523,
+      "step": 325000
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 2.3825442790985107,
+      "learning_rate": 0.00020483148160859802,
+      "loss": 1.6472,
+      "step": 325500
+    },
+    {
+      "epoch": 9.57,
+      "grad_norm": 1.509970784187317,
+      "learning_rate": 0.00020468438961726663,
+      "loss": 1.6095,
+      "step": 326000
+    },
+    {
+      "epoch": 9.59,
+      "grad_norm": 1.0791665315628052,
+      "learning_rate": 0.00020453729762593526,
+      "loss": 1.6545,
+      "step": 326500
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 1.480979323387146,
+      "learning_rate": 0.00020439020563460387,
+      "loss": 1.6691,
+      "step": 327000
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 1.9298070669174194,
+      "learning_rate": 0.00020424311364327248,
+      "loss": 1.6561,
+      "step": 327500
+    },
+    {
+      "epoch": 9.63,
+      "grad_norm": 2.1420977115631104,
+      "learning_rate": 0.0002040960216519411,
+      "loss": 1.6332,
+      "step": 328000
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 1.6322475671768188,
+      "learning_rate": 0.00020394892966060972,
+      "loss": 1.6576,
+      "step": 328500
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 1.7701492309570312,
+      "learning_rate": 0.00020380183766927833,
+      "loss": 1.6603,
+      "step": 329000
+    },
+    {
+      "epoch": 9.67,
+      "grad_norm": 22.195663452148438,
+      "learning_rate": 0.000203654745677947,
+      "loss": 1.6605,
+      "step": 329500
+    },
+    {
+      "epoch": 9.69,
+      "grad_norm": 1.735055685043335,
+      "learning_rate": 0.0002035076536866156,
+      "loss": 1.6728,
+      "step": 330000
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 1.3548341989517212,
+      "learning_rate": 0.00020336056169528423,
+      "loss": 1.6606,
+      "step": 330500
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 2.0401058197021484,
+      "learning_rate": 0.00020321346970395284,
+      "loss": 1.6687,
+      "step": 331000
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 16.856937408447266,
+      "learning_rate": 0.00020306637771262144,
+      "loss": 1.6369,
+      "step": 331500
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 3.8989596366882324,
+      "learning_rate": 0.00020291928572129008,
+      "loss": 1.6454,
+      "step": 332000
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 1.6723474264144897,
+      "learning_rate": 0.0002027721937299587,
+      "loss": 1.6723,
+      "step": 332500
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 1.7381031513214111,
+      "learning_rate": 0.0002026251017386273,
+      "loss": 1.6672,
+      "step": 333000
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 1.7013987302780151,
+      "learning_rate": 0.00020247800974729596,
+      "loss": 1.6666,
+      "step": 333500
+    },
+    {
+      "epoch": 9.81,
+      "grad_norm": 1.6793595552444458,
+      "learning_rate": 0.00020233091775596456,
+      "loss": 1.6594,
+      "step": 334000
+    },
+    {
+      "epoch": 9.82,
+      "grad_norm": 1.2351815700531006,
+      "learning_rate": 0.0002021838257646332,
+      "loss": 1.6475,
+      "step": 334500
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 3.5721399784088135,
+      "learning_rate": 0.0002020367337733018,
+      "loss": 1.6584,
+      "step": 335000
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 1.5881208181381226,
+      "learning_rate": 0.0002018896417819704,
+      "loss": 1.6584,
+      "step": 335500
+    },
+    {
+      "epoch": 9.87,
+      "grad_norm": 2.2431886196136475,
+      "learning_rate": 0.00020174254979063905,
+      "loss": 1.6632,
+      "step": 336000
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 1.9759477376937866,
+      "learning_rate": 0.00020159545779930766,
+      "loss": 1.6098,
+      "step": 336500
+    },
+    {
+      "epoch": 9.89,
+      "grad_norm": 2.4007914066314697,
+      "learning_rate": 0.00020144836580797626,
+      "loss": 1.6553,
+      "step": 337000
+    },
+    {
+      "epoch": 9.91,
+      "grad_norm": 1.1646785736083984,
+      "learning_rate": 0.00020130127381664492,
+      "loss": 1.6917,
+      "step": 337500
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 2.7321279048919678,
+      "learning_rate": 0.00020115418182531353,
+      "loss": 1.6165,
+      "step": 338000
+    },
+    {
+      "epoch": 9.94,
+      "grad_norm": 2.1226370334625244,
+      "learning_rate": 0.00020100708983398217,
+      "loss": 1.6583,
+      "step": 338500
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 16.655624389648438,
+      "learning_rate": 0.00020085999784265077,
+      "loss": 1.6594,
+      "step": 339000
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 6.142359256744385,
+      "learning_rate": 0.00020071290585131938,
+      "loss": 1.6343,
+      "step": 339500
+    },
+    {
+      "epoch": 9.98,
+      "grad_norm": 2.1720430850982666,
+      "learning_rate": 0.00020056581385998802,
+      "loss": 1.6509,
+      "step": 340000
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 2.870986223220825,
+      "learning_rate": 0.00020041872186865662,
+      "loss": 1.6707,
+      "step": 340500
+    },
+    {
+      "epoch": 10.01,
+      "grad_norm": 1.382546067237854,
+      "learning_rate": 0.00020027162987732526,
+      "loss": 1.5759,
+      "step": 341000
+    },
+    {
+      "epoch": 10.03,
+      "grad_norm": 1.2603851556777954,
+      "learning_rate": 0.0002001245378859939,
+      "loss": 1.5937,
+      "step": 341500
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 2.455425262451172,
+      "learning_rate": 0.0001999774458946625,
+      "loss": 1.5859,
+      "step": 342000
+    },
+    {
+      "epoch": 10.06,
+      "grad_norm": 1.6993873119354248,
+      "learning_rate": 0.00019983035390333114,
+      "loss": 1.6102,
+      "step": 342500
+    },
+    {
+      "epoch": 10.07,
+      "grad_norm": 2.338402509689331,
+      "learning_rate": 0.00019968326191199974,
+      "loss": 1.6058,
+      "step": 343000
+    },
+    {
+      "epoch": 10.09,
+      "grad_norm": 1.4525296688079834,
+      "learning_rate": 0.00019953616992066838,
+      "loss": 1.5925,
+      "step": 343500
+    },
+    {
+      "epoch": 10.1,
+      "grad_norm": 1.7461038827896118,
+      "learning_rate": 0.00019938907792933699,
+      "loss": 1.5824,
+      "step": 344000
+    },
+    {
+      "epoch": 10.11,
+      "grad_norm": 20.686992645263672,
+      "learning_rate": 0.0001992419859380056,
+      "loss": 1.6173,
+      "step": 344500
+    },
+    {
+      "epoch": 10.13,
+      "grad_norm": 11.99207592010498,
+      "learning_rate": 0.00019909489394667423,
+      "loss": 1.5998,
+      "step": 345000
+    },
+    {
+      "epoch": 10.14,
+      "grad_norm": 2.926992177963257,
+      "learning_rate": 0.00019894780195534286,
+      "loss": 1.5864,
+      "step": 345500
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 3.7215046882629395,
+      "learning_rate": 0.0001988007099640115,
+      "loss": 1.5856,
+      "step": 346000
+    },
+    {
+      "epoch": 10.17,
+      "grad_norm": 1.7094500064849854,
+      "learning_rate": 0.0001986536179726801,
+      "loss": 1.5964,
+      "step": 346500
+    },
+    {
+      "epoch": 10.19,
+      "grad_norm": 2.5829222202301025,
+      "learning_rate": 0.0001985065259813487,
+      "loss": 1.5943,
+      "step": 347000
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 1.6218924522399902,
+      "learning_rate": 0.00019835943399001735,
+      "loss": 1.5808,
+      "step": 347500
+    },
+    {
+      "epoch": 10.22,
+      "grad_norm": 1.952506184577942,
+      "learning_rate": 0.00019821234199868595,
+      "loss": 1.5548,
+      "step": 348000
+    },
+    {
+      "epoch": 10.23,
+      "grad_norm": 2.1660783290863037,
+      "learning_rate": 0.00019806525000735456,
+      "loss": 1.585,
+      "step": 348500
+    },
+    {
+      "epoch": 10.25,
+      "grad_norm": 1.8295018672943115,
+      "learning_rate": 0.0001979181580160232,
+      "loss": 1.5975,
+      "step": 349000
+    },
+    {
+      "epoch": 10.26,
+      "grad_norm": 2.3375236988067627,
+      "learning_rate": 0.00019777106602469183,
+      "loss": 1.617,
+      "step": 349500
+    },
+    {
+      "epoch": 10.28,
+      "grad_norm": 1.323554515838623,
+      "learning_rate": 0.00019762397403336047,
+      "loss": 1.5915,
+      "step": 350000
+    },
+    {
+      "epoch": 10.29,
+      "grad_norm": 1.259614109992981,
+      "learning_rate": 0.00019747688204202907,
+      "loss": 1.6126,
+      "step": 350500
+    },
+    {
+      "epoch": 10.31,
+      "grad_norm": 27.243751525878906,
+      "learning_rate": 0.00019732979005069768,
+      "loss": 1.5886,
+      "step": 351000
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 1.4931347370147705,
+      "learning_rate": 0.00019718269805936632,
+      "loss": 1.6165,
+      "step": 351500
+    },
+    {
+      "epoch": 10.33,
+      "grad_norm": 1.3573870658874512,
+      "learning_rate": 0.00019703560606803492,
+      "loss": 1.5967,
+      "step": 352000
+    },
+    {
+      "epoch": 10.35,
+      "grad_norm": 2.393355369567871,
+      "learning_rate": 0.00019688851407670353,
+      "loss": 1.6033,
+      "step": 352500
+    },
+    {
+      "epoch": 10.36,
+      "grad_norm": 17.975017547607422,
+      "learning_rate": 0.00019674142208537217,
+      "loss": 1.6176,
+      "step": 353000
+    },
+    {
+      "epoch": 10.38,
+      "grad_norm": 2.195270538330078,
+      "learning_rate": 0.0001965943300940408,
+      "loss": 1.6022,
+      "step": 353500
+    },
+    {
+      "epoch": 10.39,
+      "grad_norm": 1.5013370513916016,
+      "learning_rate": 0.00019644723810270944,
+      "loss": 1.5999,
+      "step": 354000
+    },
+    {
+      "epoch": 10.41,
+      "grad_norm": 16.84026527404785,
+      "learning_rate": 0.00019630014611137804,
+      "loss": 1.6061,
+      "step": 354500
+    },
+    {
+      "epoch": 10.42,
+      "grad_norm": 2.8341875076293945,
+      "learning_rate": 0.00019615305412004665,
+      "loss": 1.5742,
+      "step": 355000
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 1.6109588146209717,
+      "learning_rate": 0.00019600596212871529,
+      "loss": 1.5941,
+      "step": 355500
+    },
+    {
+      "epoch": 10.45,
+      "grad_norm": 5.435859203338623,
+      "learning_rate": 0.0001958588701373839,
+      "loss": 1.5875,
+      "step": 356000
+    },
+    {
+      "epoch": 10.47,
+      "grad_norm": 4.453019142150879,
+      "learning_rate": 0.0001957117781460525,
+      "loss": 1.5901,
+      "step": 356500
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 2.145413875579834,
+      "learning_rate": 0.00019556468615472116,
+      "loss": 1.5946,
+      "step": 357000
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 2.531158208847046,
+      "learning_rate": 0.00019541759416338977,
+      "loss": 1.5865,
+      "step": 357500
+    },
+    {
+      "epoch": 10.51,
+      "grad_norm": 1.8745834827423096,
+      "learning_rate": 0.0001952705021720584,
+      "loss": 1.6033,
+      "step": 358000
+    },
+    {
+      "epoch": 10.53,
+      "grad_norm": 1.3398380279541016,
+      "learning_rate": 0.000195123410180727,
+      "loss": 1.6205,
+      "step": 358500
+    },
+    {
+      "epoch": 10.54,
+      "grad_norm": 6.710257053375244,
+      "learning_rate": 0.00019497631818939562,
+      "loss": 1.6171,
+      "step": 359000
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 1.5463893413543701,
+      "learning_rate": 0.00019482922619806425,
+      "loss": 1.5913,
+      "step": 359500
+    },
+    {
+      "epoch": 10.57,
+      "grad_norm": 1.32403564453125,
+      "learning_rate": 0.00019468213420673286,
+      "loss": 1.6129,
+      "step": 360000
+    },
+    {
+      "epoch": 10.58,
+      "grad_norm": 1.1276665925979614,
+      "learning_rate": 0.00019453504221540147,
+      "loss": 1.6063,
+      "step": 360500
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 3.2467503547668457,
+      "learning_rate": 0.00019438795022407013,
+      "loss": 1.6036,
+      "step": 361000
+    },
+    {
+      "epoch": 10.61,
+      "grad_norm": 2.091618537902832,
+      "learning_rate": 0.00019424085823273874,
+      "loss": 1.5919,
+      "step": 361500
+    },
+    {
+      "epoch": 10.63,
+      "grad_norm": 2.571202516555786,
+      "learning_rate": 0.00019409376624140737,
+      "loss": 1.6218,
+      "step": 362000
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 1.7798309326171875,
+      "learning_rate": 0.00019394667425007598,
+      "loss": 1.5919,
+      "step": 362500
+    },
+    {
+      "epoch": 10.66,
+      "grad_norm": 1.96729576587677,
+      "learning_rate": 0.0001937995822587446,
+      "loss": 1.6116,
+      "step": 363000
+    },
+    {
+      "epoch": 10.67,
+      "grad_norm": 3.229706048965454,
+      "learning_rate": 0.00019365249026741322,
+      "loss": 1.5993,
+      "step": 363500
+    },
+    {
+      "epoch": 10.69,
+      "grad_norm": 6.615269184112549,
+      "learning_rate": 0.00019350539827608183,
+      "loss": 1.6034,
+      "step": 364000
+    },
+    {
+      "epoch": 10.7,
+      "grad_norm": 6.847731113433838,
+      "learning_rate": 0.00019335830628475047,
+      "loss": 1.5785,
+      "step": 364500
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 1.3055931329727173,
+      "learning_rate": 0.0001932112142934191,
+      "loss": 1.6059,
+      "step": 365000
+    },
+    {
+      "epoch": 10.73,
+      "grad_norm": 5.202936172485352,
+      "learning_rate": 0.0001930641223020877,
+      "loss": 1.6145,
+      "step": 365500
+    },
+    {
+      "epoch": 10.75,
+      "grad_norm": 1.704176902770996,
+      "learning_rate": 0.00019291703031075634,
+      "loss": 1.6349,
+      "step": 366000
+    },
+    {
+      "epoch": 10.76,
+      "grad_norm": 2.4655399322509766,
+      "learning_rate": 0.00019276993831942495,
+      "loss": 1.5845,
+      "step": 366500
+    },
+    {
+      "epoch": 10.78,
+      "grad_norm": 6.769374370574951,
+      "learning_rate": 0.00019262284632809358,
+      "loss": 1.6054,
+      "step": 367000
+    },
+    {
+      "epoch": 10.79,
+      "grad_norm": 3.4355087280273438,
+      "learning_rate": 0.0001924757543367622,
+      "loss": 1.6205,
+      "step": 367500
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 1.796799659729004,
+      "learning_rate": 0.0001923286623454308,
+      "loss": 1.6138,
+      "step": 368000
+    },
+    {
+      "epoch": 10.82,
+      "grad_norm": 4.333281993865967,
+      "learning_rate": 0.00019218157035409943,
+      "loss": 1.6273,
+      "step": 368500
+    },
+    {
+      "epoch": 10.83,
+      "grad_norm": 10.447500228881836,
+      "learning_rate": 0.00019203447836276807,
+      "loss": 1.6497,
+      "step": 369000
+    },
+    {
+      "epoch": 10.85,
+      "grad_norm": 1.9053871631622314,
+      "learning_rate": 0.0001918873863714367,
+      "loss": 1.5815,
+      "step": 369500
+    },
+    {
+      "epoch": 10.86,
+      "grad_norm": 6.775355339050293,
+      "learning_rate": 0.0001917402943801053,
+      "loss": 1.5986,
+      "step": 370000
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 19.00932502746582,
+      "learning_rate": 0.00019159320238877392,
+      "loss": 1.5916,
+      "step": 370500
+    },
+    {
+      "epoch": 10.89,
+      "grad_norm": 2.1029250621795654,
+      "learning_rate": 0.00019144611039744255,
+      "loss": 1.6072,
+      "step": 371000
+    },
+    {
+      "epoch": 10.91,
+      "grad_norm": 1.3043904304504395,
+      "learning_rate": 0.00019129901840611116,
+      "loss": 1.5885,
+      "step": 371500
+    },
+    {
+      "epoch": 10.92,
+      "grad_norm": 1.8047374486923218,
+      "learning_rate": 0.00019115192641477977,
+      "loss": 1.6265,
+      "step": 372000
+    },
+    {
+      "epoch": 10.94,
+      "grad_norm": 2.591125011444092,
+      "learning_rate": 0.0001910048344234484,
+      "loss": 1.6028,
+      "step": 372500
+    },
+    {
+      "epoch": 10.95,
+      "grad_norm": 1.8858853578567505,
+      "learning_rate": 0.00019085774243211704,
+      "loss": 1.6212,
+      "step": 373000
+    },
+    {
+      "epoch": 10.97,
+      "grad_norm": 1.4778261184692383,
+      "learning_rate": 0.00019071065044078567,
+      "loss": 1.6049,
+      "step": 373500
+    },
+    {
+      "epoch": 10.98,
+      "grad_norm": 7.7040934562683105,
+      "learning_rate": 0.00019056355844945428,
+      "loss": 1.5926,
+      "step": 374000
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 1.5968307256698608,
+      "learning_rate": 0.0001904164664581229,
+      "loss": 1.6438,
+      "step": 374500
+    },
+    {
+      "epoch": 11.01,
+      "grad_norm": 1.4240366220474243,
+      "learning_rate": 0.00019026937446679152,
+      "loss": 1.5451,
+      "step": 375000
+    },
+    {
+      "epoch": 11.02,
+      "grad_norm": 2.0358214378356934,
+      "learning_rate": 0.00019012228247546013,
+      "loss": 1.5232,
+      "step": 375500
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 1.2968121767044067,
+      "learning_rate": 0.00018997519048412874,
+      "loss": 1.5417,
+      "step": 376000
+    },
+    {
+      "epoch": 11.05,
+      "grad_norm": 3.3961257934570312,
+      "learning_rate": 0.00018982809849279737,
+      "loss": 1.5513,
+      "step": 376500
+    },
+    {
+      "epoch": 11.07,
+      "grad_norm": 2.7855663299560547,
+      "learning_rate": 0.000189681006501466,
+      "loss": 1.5437,
+      "step": 377000
+    },
+    {
+      "epoch": 11.08,
+      "grad_norm": 5.572605609893799,
+      "learning_rate": 0.00018953391451013464,
+      "loss": 1.5246,
+      "step": 377500
+    },
+    {
+      "epoch": 11.1,
+      "grad_norm": 2.3603007793426514,
+      "learning_rate": 0.00018938682251880325,
+      "loss": 1.5246,
+      "step": 378000
+    },
+    {
+      "epoch": 11.11,
+      "grad_norm": 19.196584701538086,
+      "learning_rate": 0.00018923973052747186,
+      "loss": 1.5223,
+      "step": 378500
+    },
+    {
+      "epoch": 11.13,
+      "grad_norm": 2.4410743713378906,
+      "learning_rate": 0.0001890926385361405,
+      "loss": 1.5202,
+      "step": 379000
+    },
+    {
+      "epoch": 11.14,
+      "grad_norm": 1.8335517644882202,
+      "learning_rate": 0.0001889455465448091,
+      "loss": 1.5625,
+      "step": 379500
+    },
+    {
+      "epoch": 11.16,
+      "grad_norm": 2.2342159748077393,
+      "learning_rate": 0.0001887984545534777,
+      "loss": 1.5391,
+      "step": 380000
+    },
+    {
+      "epoch": 11.17,
+      "grad_norm": 5.949936389923096,
+      "learning_rate": 0.00018865136256214634,
+      "loss": 1.569,
+      "step": 380500
+    },
+    {
+      "epoch": 11.19,
+      "grad_norm": 6.047088623046875,
+      "learning_rate": 0.00018850427057081498,
+      "loss": 1.5366,
+      "step": 381000
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 16.05063819885254,
+      "learning_rate": 0.0001883571785794836,
+      "loss": 1.5343,
+      "step": 381500
+    },
+    {
+      "epoch": 11.22,
+      "grad_norm": 1.327913761138916,
+      "learning_rate": 0.00018821008658815222,
+      "loss": 1.5641,
+      "step": 382000
+    },
+    {
+      "epoch": 11.23,
+      "grad_norm": 4.132845878601074,
+      "learning_rate": 0.00018806299459682083,
+      "loss": 1.5722,
+      "step": 382500
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 1.9688639640808105,
+      "learning_rate": 0.00018791590260548946,
+      "loss": 1.5477,
+      "step": 383000
+    },
+    {
+      "epoch": 11.26,
+      "grad_norm": 1.6657782793045044,
+      "learning_rate": 0.00018776881061415807,
+      "loss": 1.5398,
+      "step": 383500
+    },
+    {
+      "epoch": 11.27,
+      "grad_norm": 2.9266579151153564,
+      "learning_rate": 0.00018762171862282668,
+      "loss": 1.556,
+      "step": 384000
+    },
+    {
+      "epoch": 11.29,
+      "grad_norm": 5.755486011505127,
+      "learning_rate": 0.0001874746266314953,
+      "loss": 1.5415,
+      "step": 384500
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 2.1186070442199707,
+      "learning_rate": 0.00018732753464016395,
+      "loss": 1.5379,
+      "step": 385000
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 3.1836068630218506,
+      "learning_rate": 0.00018718044264883258,
+      "loss": 1.5621,
+      "step": 385500
+    },
+    {
+      "epoch": 11.33,
+      "grad_norm": 4.884739875793457,
+      "learning_rate": 0.0001870333506575012,
+      "loss": 1.5791,
+      "step": 386000
+    },
+    {
+      "epoch": 11.35,
+      "grad_norm": 2.4326329231262207,
+      "learning_rate": 0.0001868862586661698,
+      "loss": 1.5647,
+      "step": 386500
+    },
+    {
+      "epoch": 11.36,
+      "grad_norm": 5.392803192138672,
+      "learning_rate": 0.00018673916667483843,
+      "loss": 1.5703,
+      "step": 387000
+    },
+    {
+      "epoch": 11.38,
+      "grad_norm": 6.940735340118408,
+      "learning_rate": 0.00018659207468350704,
+      "loss": 1.5579,
+      "step": 387500
+    },
+    {
+      "epoch": 11.39,
+      "grad_norm": 2.8021979331970215,
+      "learning_rate": 0.00018644498269217567,
+      "loss": 1.5469,
+      "step": 388000
+    },
+    {
+      "epoch": 11.41,
+      "grad_norm": 1.8717076778411865,
+      "learning_rate": 0.00018629789070084428,
+      "loss": 1.5624,
+      "step": 388500
+    },
+    {
+      "epoch": 11.42,
+      "grad_norm": 3.1023306846618652,
+      "learning_rate": 0.00018615079870951291,
+      "loss": 1.5751,
+      "step": 389000
+    },
+    {
+      "epoch": 11.44,
+      "grad_norm": 1.5794864892959595,
+      "learning_rate": 0.00018600370671818155,
+      "loss": 1.5652,
+      "step": 389500
+    },
+    {
+      "epoch": 11.45,
+      "grad_norm": 3.9795074462890625,
+      "learning_rate": 0.00018585661472685016,
+      "loss": 1.5663,
+      "step": 390000
+    },
+    {
+      "epoch": 11.47,
+      "grad_norm": 1.4186744689941406,
+      "learning_rate": 0.0001857095227355188,
+      "loss": 1.5508,
+      "step": 390500
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 3.268982172012329,
+      "learning_rate": 0.0001855624307441874,
+      "loss": 1.5666,
+      "step": 391000
+    },
+    {
+      "epoch": 11.49,
+      "grad_norm": 1.246903419494629,
+      "learning_rate": 0.000185415338752856,
+      "loss": 1.5797,
+      "step": 391500
+    },
+    {
+      "epoch": 11.51,
+      "grad_norm": 1.4437174797058105,
+      "learning_rate": 0.00018526824676152464,
+      "loss": 1.562,
+      "step": 392000
+    },
+    {
+      "epoch": 11.52,
+      "grad_norm": 1.8195850849151611,
+      "learning_rate": 0.00018512115477019325,
+      "loss": 1.5748,
+      "step": 392500
+    },
+    {
+      "epoch": 11.54,
+      "grad_norm": 1.6077349185943604,
+      "learning_rate": 0.0001849740627788619,
+      "loss": 1.5628,
+      "step": 393000
+    },
+    {
+      "epoch": 11.55,
+      "grad_norm": 1.614617943763733,
+      "learning_rate": 0.00018482697078753052,
+      "loss": 1.5992,
+      "step": 393500
+    },
+    {
+      "epoch": 11.57,
+      "grad_norm": 2.860440731048584,
+      "learning_rate": 0.00018467987879619913,
+      "loss": 1.5635,
+      "step": 394000
+    },
+    {
+      "epoch": 11.58,
+      "grad_norm": 4.223182201385498,
+      "learning_rate": 0.00018453278680486776,
+      "loss": 1.5678,
+      "step": 394500
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 2.143765687942505,
+      "learning_rate": 0.00018438569481353637,
+      "loss": 1.5714,
+      "step": 395000
+    },
+    {
+      "epoch": 11.61,
+      "grad_norm": 14.07819652557373,
+      "learning_rate": 0.00018423860282220498,
+      "loss": 1.5662,
+      "step": 395500
+    },
+    {
+      "epoch": 11.63,
+      "grad_norm": 1.6693955659866333,
+      "learning_rate": 0.0001840915108308736,
+      "loss": 1.5431,
+      "step": 396000
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 2.4783432483673096,
+      "learning_rate": 0.00018394441883954222,
+      "loss": 1.5439,
+      "step": 396500
+    },
+    {
+      "epoch": 11.66,
+      "grad_norm": 2.1813952922821045,
+      "learning_rate": 0.00018379732684821088,
+      "loss": 1.5539,
+      "step": 397000
+    },
+    {
+      "epoch": 11.67,
+      "grad_norm": 4.754596710205078,
+      "learning_rate": 0.0001836502348568795,
+      "loss": 1.5675,
+      "step": 397500
+    },
+    {
+      "epoch": 11.69,
+      "grad_norm": 1.9912656545639038,
+      "learning_rate": 0.0001835031428655481,
+      "loss": 1.5909,
+      "step": 398000
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 1.5389857292175293,
+      "learning_rate": 0.00018335605087421673,
+      "loss": 1.5843,
+      "step": 398500
+    },
+    {
+      "epoch": 11.71,
+      "grad_norm": 3.8105554580688477,
+      "learning_rate": 0.00018320895888288534,
+      "loss": 1.5804,
+      "step": 399000
+    },
+    {
+      "epoch": 11.73,
+      "grad_norm": 2.6107397079467773,
+      "learning_rate": 0.00018306186689155394,
+      "loss": 1.5685,
+      "step": 399500
+    },
+    {
+      "epoch": 11.74,
+      "grad_norm": 4.173858642578125,
+      "learning_rate": 0.00018291477490022258,
+      "loss": 1.5823,
+      "step": 400000
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 1.6713734865188599,
+      "learning_rate": 0.0001827676829088912,
+      "loss": 1.5672,
+      "step": 400500
+    },
+    {
+      "epoch": 11.77,
+      "grad_norm": 4.650341033935547,
+      "learning_rate": 0.00018262059091755985,
+      "loss": 1.5297,
+      "step": 401000
+    },
+    {
+      "epoch": 11.79,
+      "grad_norm": 1.5853548049926758,
+      "learning_rate": 0.00018247349892622846,
+      "loss": 1.5814,
+      "step": 401500
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 5.71536111831665,
+      "learning_rate": 0.00018232640693489706,
+      "loss": 1.5444,
+      "step": 402000
+    },
+    {
+      "epoch": 11.82,
+      "grad_norm": 2.5603206157684326,
+      "learning_rate": 0.0001821793149435657,
+      "loss": 1.593,
+      "step": 402500
+    },
+    {
+      "epoch": 11.83,
+      "grad_norm": 2.556755304336548,
+      "learning_rate": 0.0001820322229522343,
+      "loss": 1.5717,
+      "step": 403000
+    },
+    {
+      "epoch": 11.85,
+      "grad_norm": 2.1441986560821533,
+      "learning_rate": 0.0001818851309609029,
+      "loss": 1.5766,
+      "step": 403500
+    },
+    {
+      "epoch": 11.86,
+      "grad_norm": 1.762389898300171,
+      "learning_rate": 0.00018173803896957155,
+      "loss": 1.5857,
+      "step": 404000
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 1.7397152185440063,
+      "learning_rate": 0.00018159094697824016,
+      "loss": 1.5767,
+      "step": 404500
+    },
+    {
+      "epoch": 11.89,
+      "grad_norm": 2.2762222290039062,
+      "learning_rate": 0.00018144385498690882,
+      "loss": 1.5735,
+      "step": 405000
+    },
+    {
+      "epoch": 11.91,
+      "grad_norm": 2.9703893661499023,
+      "learning_rate": 0.00018129676299557742,
+      "loss": 1.5618,
+      "step": 405500
+    },
+    {
+      "epoch": 11.92,
+      "grad_norm": 3.4913887977600098,
+      "learning_rate": 0.00018114967100424603,
+      "loss": 1.589,
+      "step": 406000
+    },
+    {
+      "epoch": 11.93,
+      "grad_norm": 1.7856007814407349,
+      "learning_rate": 0.00018100257901291467,
+      "loss": 1.5924,
+      "step": 406500
+    },
+    {
+      "epoch": 11.95,
+      "grad_norm": 5.869511604309082,
+      "learning_rate": 0.00018085548702158327,
+      "loss": 1.5516,
+      "step": 407000
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 1.4495360851287842,
+      "learning_rate": 0.00018070839503025188,
+      "loss": 1.5721,
+      "step": 407500
+    },
+    {
+      "epoch": 11.98,
+      "grad_norm": 5.264606475830078,
+      "learning_rate": 0.00018056130303892052,
+      "loss": 1.5684,
+      "step": 408000
+    },
+    {
+      "epoch": 11.99,
+      "grad_norm": 1.1457246541976929,
+      "learning_rate": 0.00018041421104758912,
+      "loss": 1.5747,
+      "step": 408500
+    },
+    {
+      "epoch": 12.01,
+      "grad_norm": 2.9236228466033936,
+      "learning_rate": 0.00018026711905625779,
+      "loss": 1.5313,
+      "step": 409000
+    },
+    {
+      "epoch": 12.02,
+      "grad_norm": 11.91308307647705,
+      "learning_rate": 0.0001801200270649264,
+      "loss": 1.4883,
+      "step": 409500
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 1.8393372297286987,
+      "learning_rate": 0.000179972935073595,
+      "loss": 1.4828,
+      "step": 410000
+    },
+    {
+      "epoch": 12.05,
+      "grad_norm": 1.9591150283813477,
+      "learning_rate": 0.00017982584308226364,
+      "loss": 1.4832,
+      "step": 410500
+    },
+    {
+      "epoch": 12.07,
+      "grad_norm": 1.5424935817718506,
+      "learning_rate": 0.00017967875109093224,
+      "loss": 1.4863,
+      "step": 411000
+    },
+    {
+      "epoch": 12.08,
+      "grad_norm": 4.338646411895752,
+      "learning_rate": 0.00017953165909960088,
+      "loss": 1.5376,
+      "step": 411500
+    },
+    {
+      "epoch": 12.1,
+      "grad_norm": 4.14846134185791,
+      "learning_rate": 0.00017938456710826949,
+      "loss": 1.506,
+      "step": 412000
+    },
+    {
+      "epoch": 12.11,
+      "grad_norm": 6.690493106842041,
+      "learning_rate": 0.00017923747511693812,
+      "loss": 1.5015,
+      "step": 412500
+    },
+    {
+      "epoch": 12.13,
+      "grad_norm": 4.801180362701416,
+      "learning_rate": 0.00017909038312560676,
+      "loss": 1.53,
+      "step": 413000
+    },
+    {
+      "epoch": 12.14,
+      "grad_norm": 1.8997361660003662,
+      "learning_rate": 0.00017894329113427536,
+      "loss": 1.5191,
+      "step": 413500
+    },
+    {
+      "epoch": 12.16,
+      "grad_norm": 19.748411178588867,
+      "learning_rate": 0.000178796199142944,
+      "loss": 1.5213,
+      "step": 414000
+    },
+    {
+      "epoch": 12.17,
+      "grad_norm": 3.8095591068267822,
+      "learning_rate": 0.0001786491071516126,
+      "loss": 1.496,
+      "step": 414500
+    },
+    {
+      "epoch": 12.18,
+      "grad_norm": 3.752877950668335,
+      "learning_rate": 0.0001785020151602812,
+      "loss": 1.5014,
+      "step": 415000
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 2.396458625793457,
+      "learning_rate": 0.00017835492316894985,
+      "loss": 1.4683,
+      "step": 415500
+    },
+    {
+      "epoch": 12.21,
+      "grad_norm": 2.928061008453369,
+      "learning_rate": 0.00017820783117761846,
+      "loss": 1.5093,
+      "step": 416000
+    },
+    {
+      "epoch": 12.23,
+      "grad_norm": 6.80102014541626,
+      "learning_rate": 0.00017806073918628712,
+      "loss": 1.5085,
+      "step": 416500
+    },
+    {
+      "epoch": 12.24,
+      "grad_norm": 16.06849479675293,
+      "learning_rate": 0.00017791364719495572,
+      "loss": 1.5023,
+      "step": 417000
+    },
+    {
+      "epoch": 12.26,
+      "grad_norm": 9.823871612548828,
+      "learning_rate": 0.00017776655520362433,
+      "loss": 1.5677,
+      "step": 417500
+    },
+    {
+      "epoch": 12.27,
+      "grad_norm": 1.8364531993865967,
+      "learning_rate": 0.00017761946321229297,
+      "loss": 1.5146,
+      "step": 418000
+    },
+    {
+      "epoch": 12.29,
+      "grad_norm": 2.6699376106262207,
+      "learning_rate": 0.00017747237122096157,
+      "loss": 1.5106,
+      "step": 418500
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 7.037272930145264,
+      "learning_rate": 0.00017732527922963018,
+      "loss": 1.5276,
+      "step": 419000
+    },
+    {
+      "epoch": 12.32,
+      "grad_norm": 1.9254209995269775,
+      "learning_rate": 0.00017717818723829882,
+      "loss": 1.4907,
+      "step": 419500
+    },
+    {
+      "epoch": 12.33,
+      "grad_norm": 5.405977249145508,
+      "learning_rate": 0.00017703109524696742,
+      "loss": 1.5039,
+      "step": 420000
+    },
+    {
+      "epoch": 12.35,
+      "grad_norm": 6.036567211151123,
+      "learning_rate": 0.00017688400325563609,
+      "loss": 1.5258,
+      "step": 420500
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 3.601088523864746,
+      "learning_rate": 0.0001767369112643047,
+      "loss": 1.5316,
+      "step": 421000
+    },
+    {
+      "epoch": 12.38,
+      "grad_norm": 3.247906446456909,
+      "learning_rate": 0.0001765898192729733,
+      "loss": 1.5006,
+      "step": 421500
+    },
+    {
+      "epoch": 12.39,
+      "grad_norm": 2.138275384902954,
+      "learning_rate": 0.00017644272728164194,
+      "loss": 1.5125,
+      "step": 422000
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 2.024502992630005,
+      "learning_rate": 0.00017629563529031054,
+      "loss": 1.5323,
+      "step": 422500
+    },
+    {
+      "epoch": 12.42,
+      "grad_norm": 21.49098777770996,
+      "learning_rate": 0.00017614854329897915,
+      "loss": 1.5607,
+      "step": 423000
+    },
+    {
+      "epoch": 12.43,
+      "grad_norm": 3.5568199157714844,
+      "learning_rate": 0.00017600145130764779,
+      "loss": 1.4843,
+      "step": 423500
+    },
+    {
+      "epoch": 12.45,
+      "grad_norm": 1.6709766387939453,
+      "learning_rate": 0.0001758543593163164,
+      "loss": 1.5621,
+      "step": 424000
+    },
+    {
+      "epoch": 12.46,
+      "grad_norm": 2.4930756092071533,
+      "learning_rate": 0.00017570726732498505,
+      "loss": 1.5202,
+      "step": 424500
+    },
+    {
+      "epoch": 12.48,
+      "grad_norm": 1.8919436931610107,
+      "learning_rate": 0.00017556017533365366,
+      "loss": 1.4859,
+      "step": 425000
+    },
+    {
+      "epoch": 12.49,
+      "grad_norm": 1.2928180694580078,
+      "learning_rate": 0.00017541308334232227,
+      "loss": 1.5423,
+      "step": 425500
+    },
+    {
+      "epoch": 12.51,
+      "grad_norm": 3.5068464279174805,
+      "learning_rate": 0.0001752659913509909,
+      "loss": 1.5326,
+      "step": 426000
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 2.7532618045806885,
+      "learning_rate": 0.0001751188993596595,
+      "loss": 1.5276,
+      "step": 426500
+    },
+    {
+      "epoch": 12.54,
+      "grad_norm": 1.5163066387176514,
+      "learning_rate": 0.00017497180736832812,
+      "loss": 1.5306,
+      "step": 427000
+    },
+    {
+      "epoch": 12.55,
+      "grad_norm": 21.672080993652344,
+      "learning_rate": 0.00017482471537699675,
+      "loss": 1.5018,
+      "step": 427500
+    },
+    {
+      "epoch": 12.57,
+      "grad_norm": 4.169219970703125,
+      "learning_rate": 0.00017467762338566536,
+      "loss": 1.5314,
+      "step": 428000
+    },
+    {
+      "epoch": 12.58,
+      "grad_norm": 1.7433300018310547,
+      "learning_rate": 0.00017453053139433402,
+      "loss": 1.5408,
+      "step": 428500
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 1.6866792440414429,
+      "learning_rate": 0.00017438343940300263,
+      "loss": 1.5405,
+      "step": 429000
+    },
+    {
+      "epoch": 12.61,
+      "grad_norm": 12.107425689697266,
+      "learning_rate": 0.00017423634741167124,
+      "loss": 1.5375,
+      "step": 429500
+    },
+    {
+      "epoch": 12.62,
+      "grad_norm": 2.0178098678588867,
+      "learning_rate": 0.00017408925542033987,
+      "loss": 1.5137,
+      "step": 430000
+    },
+    {
+      "epoch": 12.64,
+      "grad_norm": 3.7658607959747314,
+      "learning_rate": 0.00017394216342900848,
+      "loss": 1.5138,
+      "step": 430500
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 2.0998480319976807,
+      "learning_rate": 0.00017379507143767712,
+      "loss": 1.5005,
+      "step": 431000
+    },
+    {
+      "epoch": 12.67,
+      "grad_norm": 1.7215161323547363,
+      "learning_rate": 0.00017364797944634572,
+      "loss": 1.5083,
+      "step": 431500
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 3.4021968841552734,
+      "learning_rate": 0.00017350088745501433,
+      "loss": 1.5103,
+      "step": 432000
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 5.235382080078125,
+      "learning_rate": 0.000173353795463683,
+      "loss": 1.5491,
+      "step": 432500
+    },
+    {
+      "epoch": 12.71,
+      "grad_norm": 5.455018997192383,
+      "learning_rate": 0.0001732067034723516,
+      "loss": 1.5486,
+      "step": 433000
+    },
+    {
+      "epoch": 12.73,
+      "grad_norm": 2.4046807289123535,
+      "learning_rate": 0.00017305961148102024,
+      "loss": 1.5241,
+      "step": 433500
+    },
+    {
+      "epoch": 12.74,
+      "grad_norm": 2.2005653381347656,
+      "learning_rate": 0.00017291251948968884,
+      "loss": 1.5392,
+      "step": 434000
+    },
+    {
+      "epoch": 12.76,
+      "grad_norm": 1.7510486841201782,
+      "learning_rate": 0.00017276542749835745,
+      "loss": 1.5076,
+      "step": 434500
+    },
+    {
+      "epoch": 12.77,
+      "grad_norm": 2.4069275856018066,
+      "learning_rate": 0.00017261833550702608,
+      "loss": 1.5189,
+      "step": 435000
+    },
+    {
+      "epoch": 12.79,
+      "grad_norm": 3.2867445945739746,
+      "learning_rate": 0.0001724712435156947,
+      "loss": 1.5082,
+      "step": 435500
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 1.321496605873108,
+      "learning_rate": 0.0001723241515243633,
+      "loss": 1.5364,
+      "step": 436000
+    },
+    {
+      "epoch": 12.82,
+      "grad_norm": 2.841768980026245,
+      "learning_rate": 0.00017217705953303196,
+      "loss": 1.5062,
+      "step": 436500
+    },
+    {
+      "epoch": 12.83,
+      "grad_norm": 2.678611993789673,
+      "learning_rate": 0.00017202996754170057,
+      "loss": 1.5048,
+      "step": 437000
+    },
+    {
+      "epoch": 12.85,
+      "grad_norm": 3.197082996368408,
+      "learning_rate": 0.0001718828755503692,
+      "loss": 1.5553,
+      "step": 437500
+    },
+    {
+      "epoch": 12.86,
+      "grad_norm": 2.398364543914795,
+      "learning_rate": 0.0001717357835590378,
+      "loss": 1.4866,
+      "step": 438000
+    },
+    {
+      "epoch": 12.87,
+      "grad_norm": 2.2497568130493164,
+      "learning_rate": 0.00017158869156770642,
+      "loss": 1.504,
+      "step": 438500
+    },
+    {
+      "epoch": 12.89,
+      "grad_norm": 3.9616646766662598,
+      "learning_rate": 0.00017144159957637505,
+      "loss": 1.5049,
+      "step": 439000
+    },
+    {
+      "epoch": 12.9,
+      "grad_norm": 1.5139656066894531,
+      "learning_rate": 0.00017129450758504366,
+      "loss": 1.549,
+      "step": 439500
+    },
+    {
+      "epoch": 12.92,
+      "grad_norm": 2.305665969848633,
+      "learning_rate": 0.00017114741559371227,
+      "loss": 1.5255,
+      "step": 440000
+    },
+    {
+      "epoch": 12.93,
+      "grad_norm": 2.3823301792144775,
+      "learning_rate": 0.00017100032360238093,
+      "loss": 1.4995,
+      "step": 440500
+    },
+    {
+      "epoch": 12.95,
+      "grad_norm": 1.6001240015029907,
+      "learning_rate": 0.00017085323161104954,
+      "loss": 1.5494,
+      "step": 441000
+    },
+    {
+      "epoch": 12.96,
+      "grad_norm": 2.843163013458252,
+      "learning_rate": 0.00017070613961971817,
+      "loss": 1.5394,
+      "step": 441500
+    },
+    {
+      "epoch": 12.98,
+      "grad_norm": 2.1387860774993896,
+      "learning_rate": 0.00017055904762838678,
+      "loss": 1.5456,
+      "step": 442000
+    },
+    {
+      "epoch": 12.99,
+      "grad_norm": 4.817657470703125,
+      "learning_rate": 0.0001704119556370554,
+      "loss": 1.5453,
+      "step": 442500
+    },
+    {
+      "epoch": 13.01,
+      "grad_norm": 1.376753807067871,
+      "learning_rate": 0.00017026486364572402,
+      "loss": 1.5093,
+      "step": 443000
+    },
+    {
+      "epoch": 13.02,
+      "grad_norm": 1.1947274208068848,
+      "learning_rate": 0.00017011777165439263,
+      "loss": 1.457,
+      "step": 443500
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 3.931847333908081,
+      "learning_rate": 0.00016997067966306124,
+      "loss": 1.4286,
+      "step": 444000
+    },
+    {
+      "epoch": 13.05,
+      "grad_norm": 14.636221885681152,
+      "learning_rate": 0.0001698235876717299,
+      "loss": 1.4712,
+      "step": 444500
+    },
+    {
+      "epoch": 13.07,
+      "grad_norm": 3.1448845863342285,
+      "learning_rate": 0.0001696764956803985,
+      "loss": 1.4628,
+      "step": 445000
+    },
+    {
+      "epoch": 13.08,
+      "grad_norm": 0.9527159929275513,
+      "learning_rate": 0.00016952940368906714,
+      "loss": 1.4179,
+      "step": 445500
+    },
+    {
+      "epoch": 13.09,
+      "grad_norm": 2.0560357570648193,
+      "learning_rate": 0.00016938231169773575,
+      "loss": 1.4716,
+      "step": 446000
+    },
+    {
+      "epoch": 13.11,
+      "grad_norm": 2.8239283561706543,
+      "learning_rate": 0.00016923521970640436,
+      "loss": 1.477,
+      "step": 446500
+    },
+    {
+      "epoch": 13.12,
+      "grad_norm": 3.246338367462158,
+      "learning_rate": 0.000169088127715073,
+      "loss": 1.4739,
+      "step": 447000
+    },
+    {
+      "epoch": 13.14,
+      "grad_norm": 4.943697929382324,
+      "learning_rate": 0.0001689410357237416,
+      "loss": 1.4772,
+      "step": 447500
+    },
+    {
+      "epoch": 13.15,
+      "grad_norm": 4.003148555755615,
+      "learning_rate": 0.0001687939437324102,
+      "loss": 1.4723,
+      "step": 448000
+    },
+    {
+      "epoch": 13.17,
+      "grad_norm": 1.5625710487365723,
+      "learning_rate": 0.00016864685174107887,
+      "loss": 1.4424,
+      "step": 448500
+    },
+    {
+      "epoch": 13.18,
+      "grad_norm": 2.3796448707580566,
+      "learning_rate": 0.00016849975974974748,
+      "loss": 1.512,
+      "step": 449000
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 2.314465284347534,
+      "learning_rate": 0.0001683526677584161,
+      "loss": 1.5012,
+      "step": 449500
+    },
+    {
+      "epoch": 13.21,
+      "grad_norm": 1.8954664468765259,
+      "learning_rate": 0.00016820557576708472,
+      "loss": 1.4634,
+      "step": 450000
+    },
+    {
+      "epoch": 13.23,
+      "grad_norm": 1.6504507064819336,
+      "learning_rate": 0.00016805848377575333,
+      "loss": 1.496,
+      "step": 450500
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 1.8685587644577026,
+      "learning_rate": 0.00016791139178442196,
+      "loss": 1.5054,
+      "step": 451000
+    },
+    {
+      "epoch": 13.26,
+      "grad_norm": 3.0073273181915283,
+      "learning_rate": 0.00016776429979309057,
+      "loss": 1.4714,
+      "step": 451500
+    },
+    {
+      "epoch": 13.27,
+      "grad_norm": 1.5322887897491455,
+      "learning_rate": 0.0001676172078017592,
+      "loss": 1.4683,
+      "step": 452000
+    },
+    {
+      "epoch": 13.29,
+      "grad_norm": 10.244864463806152,
+      "learning_rate": 0.00016747011581042784,
+      "loss": 1.4834,
+      "step": 452500
+    },
+    {
+      "epoch": 13.3,
+      "grad_norm": 2.7981672286987305,
+      "learning_rate": 0.00016732302381909645,
+      "loss": 1.4919,
+      "step": 453000
+    },
+    {
+      "epoch": 13.31,
+      "grad_norm": 1.3389350175857544,
+      "learning_rate": 0.00016717593182776508,
+      "loss": 1.4589,
+      "step": 453500
+    },
+    {
+      "epoch": 13.33,
+      "grad_norm": 2.1007466316223145,
+      "learning_rate": 0.0001670288398364337,
+      "loss": 1.4844,
+      "step": 454000
+    },
+    {
+      "epoch": 13.34,
+      "grad_norm": 1.9225057363510132,
+      "learning_rate": 0.00016688174784510232,
+      "loss": 1.479,
+      "step": 454500
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 6.121819972991943,
+      "learning_rate": 0.00016673465585377093,
+      "loss": 1.4713,
+      "step": 455000
+    },
+    {
+      "epoch": 13.37,
+      "grad_norm": 2.738173246383667,
+      "learning_rate": 0.00016658756386243954,
+      "loss": 1.5046,
+      "step": 455500
+    },
+    {
+      "epoch": 13.39,
+      "grad_norm": 1.3760331869125366,
+      "learning_rate": 0.00016644047187110817,
+      "loss": 1.4919,
+      "step": 456000
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 9.234387397766113,
+      "learning_rate": 0.0001662933798797768,
+      "loss": 1.4531,
+      "step": 456500
+    },
+    {
+      "epoch": 13.42,
+      "grad_norm": 1.6355879306793213,
+      "learning_rate": 0.00016614628788844544,
+      "loss": 1.442,
+      "step": 457000
+    },
+    {
+      "epoch": 13.43,
+      "grad_norm": 5.203847885131836,
+      "learning_rate": 0.00016599919589711405,
+      "loss": 1.4672,
+      "step": 457500
+    },
+    {
+      "epoch": 13.45,
+      "grad_norm": 5.590662956237793,
+      "learning_rate": 0.00016585210390578266,
+      "loss": 1.459,
+      "step": 458000
+    },
+    {
+      "epoch": 13.46,
+      "grad_norm": 1.945825457572937,
+      "learning_rate": 0.0001657050119144513,
+      "loss": 1.4632,
+      "step": 458500
+    },
+    {
+      "epoch": 13.48,
+      "grad_norm": 5.772292137145996,
+      "learning_rate": 0.0001655579199231199,
+      "loss": 1.4581,
+      "step": 459000
+    },
+    {
+      "epoch": 13.49,
+      "grad_norm": 1.2538419961929321,
+      "learning_rate": 0.0001654108279317885,
+      "loss": 1.4735,
+      "step": 459500
+    },
+    {
+      "epoch": 13.51,
+      "grad_norm": 5.774242401123047,
+      "learning_rate": 0.00016526373594045714,
+      "loss": 1.4833,
+      "step": 460000
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 1.8276828527450562,
+      "learning_rate": 0.00016511664394912578,
+      "loss": 1.5017,
+      "step": 460500
+    },
+    {
+      "epoch": 13.54,
+      "grad_norm": 2.792278528213501,
+      "learning_rate": 0.0001649695519577944,
+      "loss": 1.4677,
+      "step": 461000
+    },
+    {
+      "epoch": 13.55,
+      "grad_norm": 2.325228691101074,
+      "learning_rate": 0.00016482245996646302,
+      "loss": 1.482,
+      "step": 461500
+    },
+    {
+      "epoch": 13.56,
+      "grad_norm": 2.3256492614746094,
+      "learning_rate": 0.00016467536797513163,
+      "loss": 1.4808,
+      "step": 462000
+    },
+    {
+      "epoch": 13.58,
+      "grad_norm": 1.2064257860183716,
+      "learning_rate": 0.00016452827598380026,
+      "loss": 1.4695,
+      "step": 462500
+    },
+    {
+      "epoch": 13.59,
+      "grad_norm": 3.7189693450927734,
+      "learning_rate": 0.00016438118399246887,
+      "loss": 1.4593,
+      "step": 463000
+    },
+    {
+      "epoch": 13.61,
+      "grad_norm": 1.6200836896896362,
+      "learning_rate": 0.00016423409200113748,
+      "loss": 1.5176,
+      "step": 463500
+    },
+    {
+      "epoch": 13.62,
+      "grad_norm": 3.3271920680999756,
+      "learning_rate": 0.0001640870000098061,
+      "loss": 1.4941,
+      "step": 464000
+    },
+    {
+      "epoch": 13.64,
+      "grad_norm": 1.7106289863586426,
+      "learning_rate": 0.00016393990801847474,
+      "loss": 1.4414,
+      "step": 464500
+    },
+    {
+      "epoch": 13.65,
+      "grad_norm": 2.1748921871185303,
+      "learning_rate": 0.00016379281602714338,
+      "loss": 1.4887,
+      "step": 465000
+    },
+    {
+      "epoch": 13.67,
+      "grad_norm": 1.4012078046798706,
+      "learning_rate": 0.000163645724035812,
+      "loss": 1.4773,
+      "step": 465500
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 1.6043895483016968,
+      "learning_rate": 0.0001634986320444806,
+      "loss": 1.5182,
+      "step": 466000
+    },
+    {
+      "epoch": 13.7,
+      "grad_norm": 1.53175950050354,
+      "learning_rate": 0.00016335154005314923,
+      "loss": 1.4939,
+      "step": 466500
+    },
+    {
+      "epoch": 13.71,
+      "grad_norm": 7.83038854598999,
+      "learning_rate": 0.00016320444806181784,
+      "loss": 1.4997,
+      "step": 467000
+    },
+    {
+      "epoch": 13.73,
+      "grad_norm": 20.90247344970703,
+      "learning_rate": 0.00016305735607048644,
+      "loss": 1.4717,
+      "step": 467500
+    },
+    {
+      "epoch": 13.74,
+      "grad_norm": 1.7639943361282349,
+      "learning_rate": 0.00016291026407915508,
+      "loss": 1.4682,
+      "step": 468000
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 1.8345731496810913,
+      "learning_rate": 0.00016276317208782371,
+      "loss": 1.4941,
+      "step": 468500
+    },
+    {
+      "epoch": 13.77,
+      "grad_norm": 13.04284381866455,
+      "learning_rate": 0.00016261608009649235,
+      "loss": 1.508,
+      "step": 469000
+    },
+    {
+      "epoch": 13.78,
+      "grad_norm": 3.476304531097412,
+      "learning_rate": 0.00016246898810516096,
+      "loss": 1.4849,
+      "step": 469500
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 76.50082397460938,
+      "learning_rate": 0.00016232189611382956,
+      "loss": 1.5388,
+      "step": 470000
+    },
+    {
+      "epoch": 13.81,
+      "grad_norm": 2.03076171875,
+      "learning_rate": 0.0001621748041224982,
+      "loss": 1.4964,
+      "step": 470500
+    },
+    {
+      "epoch": 13.83,
+      "grad_norm": 5.736513137817383,
+      "learning_rate": 0.0001620277121311668,
+      "loss": 1.5211,
+      "step": 471000
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 1.7192264795303345,
+      "learning_rate": 0.0001618806201398354,
+      "loss": 1.4806,
+      "step": 471500
+    },
+    {
+      "epoch": 13.86,
+      "grad_norm": 1.4231237173080444,
+      "learning_rate": 0.00016173352814850408,
+      "loss": 1.465,
+      "step": 472000
+    },
+    {
+      "epoch": 13.87,
+      "grad_norm": 1.2601704597473145,
+      "learning_rate": 0.00016158643615717268,
+      "loss": 1.5081,
+      "step": 472500
+    },
+    {
+      "epoch": 13.89,
+      "grad_norm": 1.2420154809951782,
+      "learning_rate": 0.00016143934416584132,
+      "loss": 1.5014,
+      "step": 473000
+    },
+    {
+      "epoch": 13.9,
+      "grad_norm": 2.446563720703125,
+      "learning_rate": 0.00016129225217450993,
+      "loss": 1.4975,
+      "step": 473500
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 2.2312917709350586,
+      "learning_rate": 0.00016114516018317853,
+      "loss": 1.4827,
+      "step": 474000
+    },
+    {
+      "epoch": 13.93,
+      "grad_norm": 2.2286717891693115,
+      "learning_rate": 0.00016099806819184717,
+      "loss": 1.4932,
+      "step": 474500
+    },
+    {
+      "epoch": 13.95,
+      "grad_norm": 3.0326056480407715,
+      "learning_rate": 0.00016085097620051577,
+      "loss": 1.5015,
+      "step": 475000
+    },
+    {
+      "epoch": 13.96,
+      "grad_norm": 4.350718975067139,
+      "learning_rate": 0.0001607038842091844,
+      "loss": 1.4917,
+      "step": 475500
+    },
+    {
+      "epoch": 13.98,
+      "grad_norm": 1.2791180610656738,
+      "learning_rate": 0.00016055679221785304,
+      "loss": 1.5066,
+      "step": 476000
+    },
+    {
+      "epoch": 13.99,
+      "grad_norm": 2.8099005222320557,
+      "learning_rate": 0.00016040970022652165,
+      "loss": 1.4746,
+      "step": 476500
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 2.087524175643921,
+      "learning_rate": 0.0001602626082351903,
+      "loss": 1.4278,
+      "step": 477000
+    },
+    {
+      "epoch": 14.02,
+      "grad_norm": 5.598818778991699,
+      "learning_rate": 0.0001601155162438589,
+      "loss": 1.4266,
+      "step": 477500
+    },
+    {
+      "epoch": 14.03,
+      "grad_norm": 1.573020100593567,
+      "learning_rate": 0.00015996842425252753,
+      "loss": 1.4198,
+      "step": 478000
+    },
+    {
+      "epoch": 14.05,
+      "grad_norm": 2.8413734436035156,
+      "learning_rate": 0.00015982133226119614,
+      "loss": 1.421,
+      "step": 478500
+    },
+    {
+      "epoch": 14.06,
+      "grad_norm": 1.3834174871444702,
+      "learning_rate": 0.00015967424026986474,
+      "loss": 1.4446,
+      "step": 479000
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 1.9328341484069824,
+      "learning_rate": 0.00015952714827853338,
+      "loss": 1.4332,
+      "step": 479500
+    },
+    {
+      "epoch": 14.09,
+      "grad_norm": 4.331046104431152,
+      "learning_rate": 0.000159380056287202,
+      "loss": 1.4373,
+      "step": 480000
+    },
+    {
+      "epoch": 14.11,
+      "grad_norm": 8.744680404663086,
+      "learning_rate": 0.00015923296429587065,
+      "loss": 1.4372,
+      "step": 480500
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 1.4437713623046875,
+      "learning_rate": 0.00015908587230453926,
+      "loss": 1.467,
+      "step": 481000
+    },
+    {
+      "epoch": 14.14,
+      "grad_norm": 1.5478661060333252,
+      "learning_rate": 0.00015893878031320786,
+      "loss": 1.4209,
+      "step": 481500
+    },
+    {
+      "epoch": 14.15,
+      "grad_norm": 12.385702133178711,
+      "learning_rate": 0.0001587916883218765,
+      "loss": 1.4529,
+      "step": 482000
+    },
+    {
+      "epoch": 14.17,
+      "grad_norm": 1.585307002067566,
+      "learning_rate": 0.0001586445963305451,
+      "loss": 1.442,
+      "step": 482500
+    },
+    {
+      "epoch": 14.18,
+      "grad_norm": 2.214268207550049,
+      "learning_rate": 0.0001584975043392137,
+      "loss": 1.4349,
+      "step": 483000
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 9.036513328552246,
+      "learning_rate": 0.00015835041234788235,
+      "loss": 1.4179,
+      "step": 483500
+    },
+    {
+      "epoch": 14.21,
+      "grad_norm": 2.498725175857544,
+      "learning_rate": 0.00015820332035655098,
+      "loss": 1.4394,
+      "step": 484000
+    },
+    {
+      "epoch": 14.23,
+      "grad_norm": 7.920095920562744,
+      "learning_rate": 0.00015805622836521962,
+      "loss": 1.4494,
+      "step": 484500
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 2.6044135093688965,
+      "learning_rate": 0.00015790913637388822,
+      "loss": 1.4159,
+      "step": 485000
+    },
+    {
+      "epoch": 14.25,
+      "grad_norm": 1.906570315361023,
+      "learning_rate": 0.00015776204438255683,
+      "loss": 1.446,
+      "step": 485500
+    },
+    {
+      "epoch": 14.27,
+      "grad_norm": 1.8452608585357666,
+      "learning_rate": 0.00015761495239122547,
+      "loss": 1.3961,
+      "step": 486000
+    },
+    {
+      "epoch": 14.28,
+      "grad_norm": 1.6884217262268066,
+      "learning_rate": 0.00015746786039989407,
+      "loss": 1.4385,
+      "step": 486500
+    },
+    {
+      "epoch": 14.3,
+      "grad_norm": 1.6386513710021973,
+      "learning_rate": 0.00015732076840856268,
+      "loss": 1.3842,
+      "step": 487000
+    },
+    {
+      "epoch": 14.31,
+      "grad_norm": 5.906566619873047,
+      "learning_rate": 0.00015717367641723132,
+      "loss": 1.4241,
+      "step": 487500
+    },
+    {
+      "epoch": 14.33,
+      "grad_norm": 2.964063882827759,
+      "learning_rate": 0.00015702658442589995,
+      "loss": 1.4454,
+      "step": 488000
+    },
+    {
+      "epoch": 14.34,
+      "grad_norm": 1.53037428855896,
+      "learning_rate": 0.00015687949243456859,
+      "loss": 1.4067,
+      "step": 488500
+    },
+    {
+      "epoch": 14.36,
+      "grad_norm": 1.5273714065551758,
+      "learning_rate": 0.0001567324004432372,
+      "loss": 1.4382,
+      "step": 489000
+    },
+    {
+      "epoch": 14.37,
+      "grad_norm": 1.7097736597061157,
+      "learning_rate": 0.0001565853084519058,
+      "loss": 1.4167,
+      "step": 489500
+    },
+    {
+      "epoch": 14.39,
+      "grad_norm": 2.170961380004883,
+      "learning_rate": 0.00015643821646057444,
+      "loss": 1.4541,
+      "step": 490000
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 5.699864387512207,
+      "learning_rate": 0.00015629112446924304,
+      "loss": 1.407,
+      "step": 490500
+    },
+    {
+      "epoch": 14.42,
+      "grad_norm": 1.7430428266525269,
+      "learning_rate": 0.00015614403247791165,
+      "loss": 1.4418,
+      "step": 491000
+    },
+    {
+      "epoch": 14.43,
+      "grad_norm": 2.179090976715088,
+      "learning_rate": 0.00015599694048658029,
+      "loss": 1.4331,
+      "step": 491500
+    },
+    {
+      "epoch": 14.45,
+      "grad_norm": 1.534424901008606,
+      "learning_rate": 0.00015584984849524892,
+      "loss": 1.4536,
+      "step": 492000
+    },
+    {
+      "epoch": 14.46,
+      "grad_norm": 1.464789867401123,
+      "learning_rate": 0.00015570275650391755,
+      "loss": 1.4459,
+      "step": 492500
+    },
+    {
+      "epoch": 14.47,
+      "grad_norm": 1.9043675661087036,
+      "learning_rate": 0.00015555566451258616,
+      "loss": 1.4396,
+      "step": 493000
+    },
+    {
+      "epoch": 14.49,
+      "grad_norm": 5.7054762840271,
+      "learning_rate": 0.00015540857252125477,
+      "loss": 1.4243,
+      "step": 493500
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 4.648599624633789,
+      "learning_rate": 0.0001552614805299234,
+      "loss": 1.4607,
+      "step": 494000
+    },
+    {
+      "epoch": 14.52,
+      "grad_norm": 1.4880977869033813,
+      "learning_rate": 0.000155114388538592,
+      "loss": 1.4457,
+      "step": 494500
+    },
+    {
+      "epoch": 14.53,
+      "grad_norm": 2.5349297523498535,
+      "learning_rate": 0.00015496729654726062,
+      "loss": 1.4675,
+      "step": 495000
+    },
+    {
+      "epoch": 14.55,
+      "grad_norm": 3.5980210304260254,
+      "learning_rate": 0.00015482020455592925,
+      "loss": 1.4704,
+      "step": 495500
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 3.0215611457824707,
+      "learning_rate": 0.0001546731125645979,
+      "loss": 1.4202,
+      "step": 496000
+    },
+    {
+      "epoch": 14.58,
+      "grad_norm": 1.6257708072662354,
+      "learning_rate": 0.00015452602057326652,
+      "loss": 1.4569,
+      "step": 496500
+    },
+    {
+      "epoch": 14.59,
+      "grad_norm": 3.856419324874878,
+      "learning_rate": 0.00015437892858193513,
+      "loss": 1.4142,
+      "step": 497000
+    },
+    {
+      "epoch": 14.61,
+      "grad_norm": 4.544717788696289,
+      "learning_rate": 0.00015423183659060374,
+      "loss": 1.4437,
+      "step": 497500
+    },
+    {
+      "epoch": 14.62,
+      "grad_norm": 3.5249671936035156,
+      "learning_rate": 0.00015408474459927237,
+      "loss": 1.4432,
+      "step": 498000
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 1.8191946744918823,
+      "learning_rate": 0.00015393765260794098,
+      "loss": 1.435,
+      "step": 498500
+    },
+    {
+      "epoch": 14.65,
+      "grad_norm": 2.6009254455566406,
+      "learning_rate": 0.00015379056061660962,
+      "loss": 1.411,
+      "step": 499000
+    },
+    {
+      "epoch": 14.67,
+      "grad_norm": 2.475637674331665,
+      "learning_rate": 0.00015364346862527822,
+      "loss": 1.4568,
+      "step": 499500
+    },
+    {
+      "epoch": 14.68,
+      "grad_norm": 3.965743064880371,
+      "learning_rate": 0.00015349637663394686,
+      "loss": 1.455,
+      "step": 500000
+    },
+    {
+      "epoch": 14.69,
+      "grad_norm": 1.3109550476074219,
+      "learning_rate": 0.0001533492846426155,
+      "loss": 1.4547,
+      "step": 500500
+    },
+    {
+      "epoch": 14.71,
+      "grad_norm": 2.0032811164855957,
+      "learning_rate": 0.0001532021926512841,
+      "loss": 1.4451,
+      "step": 501000
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 1.4996896982192993,
+      "learning_rate": 0.00015305510065995274,
+      "loss": 1.4497,
+      "step": 501500
+    },
+    {
+      "epoch": 14.74,
+      "grad_norm": 2.1643595695495605,
+      "learning_rate": 0.00015290800866862134,
+      "loss": 1.4335,
+      "step": 502000
+    },
+    {
+      "epoch": 14.75,
+      "grad_norm": 1.5510461330413818,
+      "learning_rate": 0.00015276091667728995,
+      "loss": 1.44,
+      "step": 502500
+    },
+    {
+      "epoch": 14.77,
+      "grad_norm": 1.9841787815093994,
+      "learning_rate": 0.00015261382468595858,
+      "loss": 1.392,
+      "step": 503000
+    },
+    {
+      "epoch": 14.78,
+      "grad_norm": 7.774050712585449,
+      "learning_rate": 0.0001524667326946272,
+      "loss": 1.4521,
+      "step": 503500
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 1.3330835103988647,
+      "learning_rate": 0.00015231964070329585,
+      "loss": 1.4562,
+      "step": 504000
+    },
+    {
+      "epoch": 14.81,
+      "grad_norm": 2.6969919204711914,
+      "learning_rate": 0.00015217254871196446,
+      "loss": 1.4313,
+      "step": 504500
+    },
+    {
+      "epoch": 14.83,
+      "grad_norm": 1.6068675518035889,
+      "learning_rate": 0.00015202545672063307,
+      "loss": 1.4515,
+      "step": 505000
+    },
+    {
+      "epoch": 14.84,
+      "grad_norm": 3.470465660095215,
+      "learning_rate": 0.0001518783647293017,
+      "loss": 1.418,
+      "step": 505500
+    },
+    {
+      "epoch": 14.86,
+      "grad_norm": 5.018684387207031,
+      "learning_rate": 0.0001517312727379703,
+      "loss": 1.4236,
+      "step": 506000
+    },
+    {
+      "epoch": 14.87,
+      "grad_norm": 1.603996753692627,
+      "learning_rate": 0.00015158418074663892,
+      "loss": 1.4478,
+      "step": 506500
+    },
+    {
+      "epoch": 14.89,
+      "grad_norm": 2.150404930114746,
+      "learning_rate": 0.00015143708875530755,
+      "loss": 1.4523,
+      "step": 507000
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 1.252964735031128,
+      "learning_rate": 0.00015128999676397616,
+      "loss": 1.4546,
+      "step": 507500
+    },
+    {
+      "epoch": 14.92,
+      "grad_norm": 3.1075639724731445,
+      "learning_rate": 0.00015114290477264482,
+      "loss": 1.4547,
+      "step": 508000
+    },
+    {
+      "epoch": 14.93,
+      "grad_norm": 1.4597883224487305,
+      "learning_rate": 0.00015099581278131343,
+      "loss": 1.4902,
+      "step": 508500
+    },
+    {
+      "epoch": 14.94,
+      "grad_norm": 2.405596971511841,
+      "learning_rate": 0.00015084872078998204,
+      "loss": 1.4306,
+      "step": 509000
+    },
+    {
+      "epoch": 14.96,
+      "grad_norm": 1.952222228050232,
+      "learning_rate": 0.00015070162879865067,
+      "loss": 1.4209,
+      "step": 509500
+    },
+    {
+      "epoch": 14.97,
+      "grad_norm": 2.3458521366119385,
+      "learning_rate": 0.00015055453680731928,
+      "loss": 1.468,
+      "step": 510000
+    },
+    {
+      "epoch": 14.99,
+      "grad_norm": 1.6442558765411377,
+      "learning_rate": 0.0001504074448159879,
+      "loss": 1.4541,
+      "step": 510500
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 2.1197829246520996,
+      "learning_rate": 0.00015026035282465652,
+      "loss": 1.4195,
+      "step": 511000
+    },
+    {
+      "epoch": 15.02,
+      "grad_norm": 2.8808157444000244,
+      "learning_rate": 0.00015011326083332513,
+      "loss": 1.3719,
+      "step": 511500
+    },
+    {
+      "epoch": 15.03,
+      "grad_norm": 1.939942717552185,
+      "learning_rate": 0.00014996616884199377,
+      "loss": 1.3627,
+      "step": 512000
+    },
+    {
+      "epoch": 15.05,
+      "grad_norm": 4.798866271972656,
+      "learning_rate": 0.00014981907685066237,
+      "loss": 1.3839,
+      "step": 512500
+    },
+    {
+      "epoch": 15.06,
+      "grad_norm": 2.553893566131592,
+      "learning_rate": 0.000149671984859331,
+      "loss": 1.4047,
+      "step": 513000
+    },
+    {
+      "epoch": 15.08,
+      "grad_norm": 1.915277123451233,
+      "learning_rate": 0.00014952489286799964,
+      "loss": 1.3933,
+      "step": 513500
+    },
+    {
+      "epoch": 15.09,
+      "grad_norm": 1.5252777338027954,
+      "learning_rate": 0.00014937780087666825,
+      "loss": 1.3644,
+      "step": 514000
+    },
+    {
+      "epoch": 15.11,
+      "grad_norm": 1.6960341930389404,
+      "learning_rate": 0.00014923070888533686,
+      "loss": 1.3946,
+      "step": 514500
+    },
+    {
+      "epoch": 15.12,
+      "grad_norm": 9.384785652160645,
+      "learning_rate": 0.0001490836168940055,
+      "loss": 1.3706,
+      "step": 515000
+    },
+    {
+      "epoch": 15.14,
+      "grad_norm": 2.280630350112915,
+      "learning_rate": 0.00014893652490267413,
+      "loss": 1.3836,
+      "step": 515500
+    },
+    {
+      "epoch": 15.15,
+      "grad_norm": 1.8966234922409058,
+      "learning_rate": 0.00014878943291134273,
+      "loss": 1.373,
+      "step": 516000
+    },
+    {
+      "epoch": 15.16,
+      "grad_norm": 2.4201481342315674,
+      "learning_rate": 0.00014864234092001134,
+      "loss": 1.4014,
+      "step": 516500
+    },
+    {
+      "epoch": 15.18,
+      "grad_norm": 2.8996338844299316,
+      "learning_rate": 0.00014849524892867998,
+      "loss": 1.3985,
+      "step": 517000
+    },
+    {
+      "epoch": 15.19,
+      "grad_norm": 2.2007150650024414,
+      "learning_rate": 0.0001483481569373486,
+      "loss": 1.3823,
+      "step": 517500
+    },
+    {
+      "epoch": 15.21,
+      "grad_norm": 2.7956199645996094,
+      "learning_rate": 0.00014820106494601722,
+      "loss": 1.3936,
+      "step": 518000
+    },
+    {
+      "epoch": 15.22,
+      "grad_norm": 2.891369104385376,
+      "learning_rate": 0.00014805397295468585,
+      "loss": 1.3991,
+      "step": 518500
+    },
+    {
+      "epoch": 15.24,
+      "grad_norm": 7.358971118927002,
+      "learning_rate": 0.00014790688096335446,
+      "loss": 1.4115,
+      "step": 519000
+    },
+    {
+      "epoch": 15.25,
+      "grad_norm": 2.1062731742858887,
+      "learning_rate": 0.0001477597889720231,
+      "loss": 1.398,
+      "step": 519500
+    },
+    {
+      "epoch": 15.27,
+      "grad_norm": 3.914013385772705,
+      "learning_rate": 0.0001476126969806917,
+      "loss": 1.4129,
+      "step": 520000
+    },
+    {
+      "epoch": 15.28,
+      "grad_norm": 1.342411756515503,
+      "learning_rate": 0.00014746560498936034,
+      "loss": 1.4022,
+      "step": 520500
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 1.3836804628372192,
+      "learning_rate": 0.00014731851299802897,
+      "loss": 1.3849,
+      "step": 521000
+    },
+    {
+      "epoch": 15.31,
+      "grad_norm": 4.170617580413818,
+      "learning_rate": 0.00014717142100669758,
+      "loss": 1.3855,
+      "step": 521500
+    },
+    {
+      "epoch": 15.33,
+      "grad_norm": 6.060724258422852,
+      "learning_rate": 0.0001470243290153662,
+      "loss": 1.3645,
+      "step": 522000
+    },
+    {
+      "epoch": 15.34,
+      "grad_norm": 5.009141445159912,
+      "learning_rate": 0.00014687723702403482,
+      "loss": 1.4106,
+      "step": 522500
+    },
+    {
+      "epoch": 15.36,
+      "grad_norm": 4.241628646850586,
+      "learning_rate": 0.00014673014503270346,
+      "loss": 1.3983,
+      "step": 523000
+    },
+    {
+      "epoch": 15.37,
+      "grad_norm": 1.6837831735610962,
+      "learning_rate": 0.00014658305304137206,
+      "loss": 1.4023,
+      "step": 523500
+    },
+    {
+      "epoch": 15.38,
+      "grad_norm": 2.455502510070801,
+      "learning_rate": 0.00014643596105004067,
+      "loss": 1.3607,
+      "step": 524000
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 6.113306522369385,
+      "learning_rate": 0.0001462888690587093,
+      "loss": 1.4292,
+      "step": 524500
+    },
+    {
+      "epoch": 15.41,
+      "grad_norm": 1.7169700860977173,
+      "learning_rate": 0.00014614177706737794,
+      "loss": 1.364,
+      "step": 525000
+    },
+    {
+      "epoch": 15.43,
+      "grad_norm": 3.563976287841797,
+      "learning_rate": 0.00014599468507604655,
+      "loss": 1.4007,
+      "step": 525500
+    },
+    {
+      "epoch": 15.44,
+      "grad_norm": 3.6780059337615967,
+      "learning_rate": 0.00014584759308471516,
+      "loss": 1.4072,
+      "step": 526000
+    },
+    {
+      "epoch": 15.46,
+      "grad_norm": 1.8734827041625977,
+      "learning_rate": 0.0001457005010933838,
+      "loss": 1.4088,
+      "step": 526500
+    },
+    {
+      "epoch": 15.47,
+      "grad_norm": 6.644404411315918,
+      "learning_rate": 0.00014555340910205243,
+      "loss": 1.3958,
+      "step": 527000
+    },
+    {
+      "epoch": 15.49,
+      "grad_norm": 1.7657749652862549,
+      "learning_rate": 0.00014540631711072103,
+      "loss": 1.418,
+      "step": 527500
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 4.092434406280518,
+      "learning_rate": 0.00014525922511938964,
+      "loss": 1.3643,
+      "step": 528000
+    },
+    {
+      "epoch": 15.52,
+      "grad_norm": 30.432186126708984,
+      "learning_rate": 0.00014511213312805828,
+      "loss": 1.3797,
+      "step": 528500
+    },
+    {
+      "epoch": 15.53,
+      "grad_norm": 10.275367736816406,
+      "learning_rate": 0.0001449650411367269,
+      "loss": 1.3928,
+      "step": 529000
+    },
+    {
+      "epoch": 15.55,
+      "grad_norm": 20.94750213623047,
+      "learning_rate": 0.00014481794914539552,
+      "loss": 1.4131,
+      "step": 529500
+    },
+    {
+      "epoch": 15.56,
+      "grad_norm": 1.7463383674621582,
+      "learning_rate": 0.00014467085715406413,
+      "loss": 1.4173,
+      "step": 530000
+    },
+    {
+      "epoch": 15.58,
+      "grad_norm": 2.380938768386841,
+      "learning_rate": 0.00014452376516273276,
+      "loss": 1.4092,
+      "step": 530500
+    },
+    {
+      "epoch": 15.59,
+      "grad_norm": 1.8386043310165405,
+      "learning_rate": 0.0001443766731714014,
+      "loss": 1.3972,
+      "step": 531000
+    },
+    {
+      "epoch": 15.61,
+      "grad_norm": 1.3829760551452637,
+      "learning_rate": 0.00014422958118007,
+      "loss": 1.391,
+      "step": 531500
+    },
+    {
+      "epoch": 15.62,
+      "grad_norm": 2.171069383621216,
+      "learning_rate": 0.0001440824891887386,
+      "loss": 1.3993,
+      "step": 532000
+    },
+    {
+      "epoch": 15.63,
+      "grad_norm": 1.716299057006836,
+      "learning_rate": 0.00014393539719740724,
+      "loss": 1.398,
+      "step": 532500
+    },
+    {
+      "epoch": 15.65,
+      "grad_norm": 1.7643611431121826,
+      "learning_rate": 0.00014378830520607588,
+      "loss": 1.3901,
+      "step": 533000
+    },
+    {
+      "epoch": 15.66,
+      "grad_norm": 1.68152916431427,
+      "learning_rate": 0.0001436412132147445,
+      "loss": 1.3873,
+      "step": 533500
+    },
+    {
+      "epoch": 15.68,
+      "grad_norm": 2.581348419189453,
+      "learning_rate": 0.0001434941212234131,
+      "loss": 1.3713,
+      "step": 534000
+    },
+    {
+      "epoch": 15.69,
+      "grad_norm": 3.0933191776275635,
+      "learning_rate": 0.00014334702923208173,
+      "loss": 1.4283,
+      "step": 534500
+    },
+    {
+      "epoch": 15.71,
+      "grad_norm": 6.795374870300293,
+      "learning_rate": 0.00014319993724075036,
+      "loss": 1.4088,
+      "step": 535000
+    },
+    {
+      "epoch": 15.72,
+      "grad_norm": 2.514035701751709,
+      "learning_rate": 0.00014305284524941897,
+      "loss": 1.4042,
+      "step": 535500
+    },
+    {
+      "epoch": 15.74,
+      "grad_norm": 2.5651772022247314,
+      "learning_rate": 0.00014290575325808758,
+      "loss": 1.4236,
+      "step": 536000
+    },
+    {
+      "epoch": 15.75,
+      "grad_norm": 1.6859495639801025,
+      "learning_rate": 0.00014275866126675621,
+      "loss": 1.4243,
+      "step": 536500
+    },
+    {
+      "epoch": 15.77,
+      "grad_norm": 1.8449592590332031,
+      "learning_rate": 0.00014261156927542485,
+      "loss": 1.406,
+      "step": 537000
+    },
+    {
+      "epoch": 15.78,
+      "grad_norm": 2.2886695861816406,
+      "learning_rate": 0.00014246447728409346,
+      "loss": 1.4095,
+      "step": 537500
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 2.669768810272217,
+      "learning_rate": 0.00014231738529276206,
+      "loss": 1.4172,
+      "step": 538000
+    },
+    {
+      "epoch": 15.81,
+      "grad_norm": 5.082691192626953,
+      "learning_rate": 0.0001421702933014307,
+      "loss": 1.4307,
+      "step": 538500
+    },
+    {
+      "epoch": 15.83,
+      "grad_norm": 2.9871368408203125,
+      "learning_rate": 0.00014202320131009933,
+      "loss": 1.3937,
+      "step": 539000
+    },
+    {
+      "epoch": 15.84,
+      "grad_norm": 1.900804877281189,
+      "learning_rate": 0.00014187610931876794,
+      "loss": 1.4226,
+      "step": 539500
+    },
+    {
+      "epoch": 15.85,
+      "grad_norm": 1.668407678604126,
+      "learning_rate": 0.00014172901732743655,
+      "loss": 1.3807,
+      "step": 540000
+    },
+    {
+      "epoch": 15.87,
+      "grad_norm": 5.046024799346924,
+      "learning_rate": 0.00014158192533610518,
+      "loss": 1.4203,
+      "step": 540500
+    },
+    {
+      "epoch": 15.88,
+      "grad_norm": 2.8824052810668945,
+      "learning_rate": 0.00014143483334477382,
+      "loss": 1.4188,
+      "step": 541000
+    },
+    {
+      "epoch": 15.9,
+      "grad_norm": 2.688316583633423,
+      "learning_rate": 0.00014128774135344243,
+      "loss": 1.418,
+      "step": 541500
+    },
+    {
+      "epoch": 15.91,
+      "grad_norm": 2.323672294616699,
+      "learning_rate": 0.00014114064936211106,
+      "loss": 1.3825,
+      "step": 542000
+    },
+    {
+      "epoch": 15.93,
+      "grad_norm": 23.119873046875,
+      "learning_rate": 0.00014099355737077967,
+      "loss": 1.4148,
+      "step": 542500
+    },
+    {
+      "epoch": 15.94,
+      "grad_norm": 3.257922649383545,
+      "learning_rate": 0.0001408464653794483,
+      "loss": 1.4327,
+      "step": 543000
+    },
+    {
+      "epoch": 15.96,
+      "grad_norm": 1.8719940185546875,
+      "learning_rate": 0.0001406993733881169,
+      "loss": 1.3868,
+      "step": 543500
+    },
+    {
+      "epoch": 15.97,
+      "grad_norm": 2.128316640853882,
+      "learning_rate": 0.00014055228139678554,
+      "loss": 1.4102,
+      "step": 544000
+    },
+    {
+      "epoch": 15.99,
+      "grad_norm": 2.277371644973755,
+      "learning_rate": 0.00014040518940545418,
+      "loss": 1.4014,
+      "step": 544500
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 41.53245162963867,
+      "learning_rate": 0.0001402580974141228,
+      "loss": 1.4178,
+      "step": 545000
+    },
+    {
+      "epoch": 16.02,
+      "grad_norm": 1.792492151260376,
+      "learning_rate": 0.0001401110054227914,
+      "loss": 1.3353,
+      "step": 545500
+    },
+    {
+      "epoch": 16.03,
+      "grad_norm": 1.582220435142517,
+      "learning_rate": 0.00013996391343146003,
+      "loss": 1.3722,
+      "step": 546000
+    },
+    {
+      "epoch": 16.05,
+      "grad_norm": 3.7628824710845947,
+      "learning_rate": 0.00013981682144012866,
+      "loss": 1.3593,
+      "step": 546500
+    },
+    {
+      "epoch": 16.06,
+      "grad_norm": 2.833401918411255,
+      "learning_rate": 0.00013966972944879727,
+      "loss": 1.3627,
+      "step": 547000
+    },
+    {
+      "epoch": 16.07,
+      "grad_norm": 2.1202030181884766,
+      "learning_rate": 0.00013952263745746588,
+      "loss": 1.3528,
+      "step": 547500
+    },
+    {
+      "epoch": 16.09,
+      "grad_norm": 1.919555425643921,
+      "learning_rate": 0.0001393755454661345,
+      "loss": 1.3087,
+      "step": 548000
+    },
+    {
+      "epoch": 16.1,
+      "grad_norm": 2.3834056854248047,
+      "learning_rate": 0.00013922845347480315,
+      "loss": 1.3303,
+      "step": 548500
+    },
+    {
+      "epoch": 16.12,
+      "grad_norm": 2.0871472358703613,
+      "learning_rate": 0.00013908136148347176,
+      "loss": 1.3423,
+      "step": 549000
+    },
+    {
+      "epoch": 16.13,
+      "grad_norm": 3.040555953979492,
+      "learning_rate": 0.00013893426949214036,
+      "loss": 1.3402,
+      "step": 549500
+    },
+    {
+      "epoch": 16.15,
+      "grad_norm": 2.8366496562957764,
+      "learning_rate": 0.000138787177500809,
+      "loss": 1.3586,
+      "step": 550000
+    },
+    {
+      "epoch": 16.16,
+      "grad_norm": 10.008976936340332,
+      "learning_rate": 0.00013864008550947763,
+      "loss": 1.3321,
+      "step": 550500
+    },
+    {
+      "epoch": 16.18,
+      "grad_norm": 8.792502403259277,
+      "learning_rate": 0.00013849299351814624,
+      "loss": 1.3781,
+      "step": 551000
+    },
+    {
+      "epoch": 16.19,
+      "grad_norm": 8.872962951660156,
+      "learning_rate": 0.00013834590152681485,
+      "loss": 1.3687,
+      "step": 551500
+    },
+    {
+      "epoch": 16.21,
+      "grad_norm": 5.631560802459717,
+      "learning_rate": 0.00013819880953548348,
+      "loss": 1.3419,
+      "step": 552000
+    },
+    {
+      "epoch": 16.22,
+      "grad_norm": 2.04437255859375,
+      "learning_rate": 0.00013805171754415212,
+      "loss": 1.3335,
+      "step": 552500
+    },
+    {
+      "epoch": 16.24,
+      "grad_norm": 2.526149272918701,
+      "learning_rate": 0.00013790462555282072,
+      "loss": 1.3446,
+      "step": 553000
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 8.970195770263672,
+      "learning_rate": 0.00013775753356148933,
+      "loss": 1.3795,
+      "step": 553500
+    },
+    {
+      "epoch": 16.27,
+      "grad_norm": 4.344628810882568,
+      "learning_rate": 0.00013761044157015797,
+      "loss": 1.3231,
+      "step": 554000
+    },
+    {
+      "epoch": 16.28,
+      "grad_norm": 2.5846548080444336,
+      "learning_rate": 0.0001374633495788266,
+      "loss": 1.3571,
+      "step": 554500
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 20.92795181274414,
+      "learning_rate": 0.0001373162575874952,
+      "loss": 1.38,
+      "step": 555000
+    },
+    {
+      "epoch": 16.31,
+      "grad_norm": 2.3088529109954834,
+      "learning_rate": 0.00013716916559616382,
+      "loss": 1.3473,
+      "step": 555500
+    },
+    {
+      "epoch": 16.32,
+      "grad_norm": 13.217586517333984,
+      "learning_rate": 0.00013702207360483245,
+      "loss": 1.3558,
+      "step": 556000
+    },
+    {
+      "epoch": 16.34,
+      "grad_norm": 8.642449378967285,
+      "learning_rate": 0.00013687498161350109,
+      "loss": 1.382,
+      "step": 556500
+    },
+    {
+      "epoch": 16.35,
+      "grad_norm": 1.4824799299240112,
+      "learning_rate": 0.0001367278896221697,
+      "loss": 1.384,
+      "step": 557000
+    },
+    {
+      "epoch": 16.37,
+      "grad_norm": 1.741585373878479,
+      "learning_rate": 0.0001365807976308383,
+      "loss": 1.3664,
+      "step": 557500
+    },
+    {
+      "epoch": 16.38,
+      "grad_norm": 1.7038291692733765,
+      "learning_rate": 0.00013643370563950694,
+      "loss": 1.3634,
+      "step": 558000
+    },
+    {
+      "epoch": 16.4,
+      "grad_norm": 17.11383628845215,
+      "learning_rate": 0.00013628661364817557,
+      "loss": 1.3604,
+      "step": 558500
+    },
+    {
+      "epoch": 16.41,
+      "grad_norm": 19.561166763305664,
+      "learning_rate": 0.00013613952165684418,
+      "loss": 1.3731,
+      "step": 559000
+    },
+    {
+      "epoch": 16.43,
+      "grad_norm": 2.2067980766296387,
+      "learning_rate": 0.00013599242966551279,
+      "loss": 1.3472,
+      "step": 559500
+    },
+    {
+      "epoch": 16.44,
+      "grad_norm": 3.9638609886169434,
+      "learning_rate": 0.00013584533767418142,
+      "loss": 1.365,
+      "step": 560000
+    },
+    {
+      "epoch": 16.46,
+      "grad_norm": 2.2947542667388916,
+      "learning_rate": 0.00013569824568285006,
+      "loss": 1.3618,
+      "step": 560500
+    },
+    {
+      "epoch": 16.47,
+      "grad_norm": 2.3892598152160645,
+      "learning_rate": 0.00013555115369151866,
+      "loss": 1.3331,
+      "step": 561000
+    },
+    {
+      "epoch": 16.49,
+      "grad_norm": 1.9236092567443848,
+      "learning_rate": 0.00013540406170018727,
+      "loss": 1.3508,
+      "step": 561500
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 5.180337429046631,
+      "learning_rate": 0.0001352569697088559,
+      "loss": 1.3612,
+      "step": 562000
+    },
+    {
+      "epoch": 16.52,
+      "grad_norm": 8.786672592163086,
+      "learning_rate": 0.00013510987771752454,
+      "loss": 1.3776,
+      "step": 562500
+    },
+    {
+      "epoch": 16.53,
+      "grad_norm": 4.111878871917725,
+      "learning_rate": 0.00013496278572619315,
+      "loss": 1.3661,
+      "step": 563000
+    },
+    {
+      "epoch": 16.54,
+      "grad_norm": 4.706780433654785,
+      "learning_rate": 0.00013481569373486178,
+      "loss": 1.3731,
+      "step": 563500
+    },
+    {
+      "epoch": 16.56,
+      "grad_norm": 1.3788596391677856,
+      "learning_rate": 0.0001346686017435304,
+      "loss": 1.3316,
+      "step": 564000
+    },
+    {
+      "epoch": 16.57,
+      "grad_norm": 2.970449924468994,
+      "learning_rate": 0.00013452150975219902,
+      "loss": 1.3612,
+      "step": 564500
+    },
+    {
+      "epoch": 16.59,
+      "grad_norm": 2.0503463745117188,
+      "learning_rate": 0.00013437441776086763,
+      "loss": 1.3635,
+      "step": 565000
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 10.668388366699219,
+      "learning_rate": 0.00013422732576953627,
+      "loss": 1.3764,
+      "step": 565500
+    },
+    {
+      "epoch": 16.62,
+      "grad_norm": 8.57248592376709,
+      "learning_rate": 0.00013408023377820487,
+      "loss": 1.3521,
+      "step": 566000
+    },
+    {
+      "epoch": 16.63,
+      "grad_norm": 1.6269396543502808,
+      "learning_rate": 0.0001339331417868735,
+      "loss": 1.3406,
+      "step": 566500
+    },
+    {
+      "epoch": 16.65,
+      "grad_norm": 1.764863133430481,
+      "learning_rate": 0.00013378604979554212,
+      "loss": 1.3325,
+      "step": 567000
+    },
+    {
+      "epoch": 16.66,
+      "grad_norm": 1.6542813777923584,
+      "learning_rate": 0.00013363895780421075,
+      "loss": 1.3601,
+      "step": 567500
+    },
+    {
+      "epoch": 16.68,
+      "grad_norm": 1.9206827878952026,
+      "learning_rate": 0.00013349186581287936,
+      "loss": 1.3515,
+      "step": 568000
+    },
+    {
+      "epoch": 16.69,
+      "grad_norm": 2.092914581298828,
+      "learning_rate": 0.000133344773821548,
+      "loss": 1.3617,
+      "step": 568500
+    },
+    {
+      "epoch": 16.71,
+      "grad_norm": 6.570430278778076,
+      "learning_rate": 0.0001331976818302166,
+      "loss": 1.3728,
+      "step": 569000
+    },
+    {
+      "epoch": 16.72,
+      "grad_norm": 2.719400644302368,
+      "learning_rate": 0.00013305058983888524,
+      "loss": 1.3406,
+      "step": 569500
+    },
+    {
+      "epoch": 16.74,
+      "grad_norm": 9.820898056030273,
+      "learning_rate": 0.00013290349784755384,
+      "loss": 1.4036,
+      "step": 570000
+    },
+    {
+      "epoch": 16.75,
+      "grad_norm": 2.3756299018859863,
+      "learning_rate": 0.00013275640585622248,
+      "loss": 1.3735,
+      "step": 570500
+    },
+    {
+      "epoch": 16.76,
+      "grad_norm": 6.200007438659668,
+      "learning_rate": 0.00013260931386489109,
+      "loss": 1.3792,
+      "step": 571000
+    },
+    {
+      "epoch": 16.78,
+      "grad_norm": 7.884439468383789,
+      "learning_rate": 0.00013246222187355972,
+      "loss": 1.3666,
+      "step": 571500
+    },
+    {
+      "epoch": 16.79,
+      "grad_norm": 2.011915683746338,
+      "learning_rate": 0.00013231512988222833,
+      "loss": 1.3541,
+      "step": 572000
+    },
+    {
+      "epoch": 16.81,
+      "grad_norm": 6.167238712310791,
+      "learning_rate": 0.00013216803789089696,
+      "loss": 1.3781,
+      "step": 572500
+    },
+    {
+      "epoch": 16.82,
+      "grad_norm": 38.23750686645508,
+      "learning_rate": 0.00013202094589956557,
+      "loss": 1.3546,
+      "step": 573000
+    },
+    {
+      "epoch": 16.84,
+      "grad_norm": 2.6406400203704834,
+      "learning_rate": 0.0001318738539082342,
+      "loss": 1.3959,
+      "step": 573500
+    },
+    {
+      "epoch": 16.85,
+      "grad_norm": 1.713273525238037,
+      "learning_rate": 0.0001317267619169028,
+      "loss": 1.3625,
+      "step": 574000
+    },
+    {
+      "epoch": 16.87,
+      "grad_norm": 2.819561004638672,
+      "learning_rate": 0.00013157966992557145,
+      "loss": 1.354,
+      "step": 574500
+    },
+    {
+      "epoch": 16.88,
+      "grad_norm": 5.471235275268555,
+      "learning_rate": 0.00013143257793424005,
+      "loss": 1.3988,
+      "step": 575000
+    },
+    {
+      "epoch": 16.9,
+      "grad_norm": 2.4000468254089355,
+      "learning_rate": 0.0001312854859429087,
+      "loss": 1.3432,
+      "step": 575500
+    },
+    {
+      "epoch": 16.91,
+      "grad_norm": 2.053870439529419,
+      "learning_rate": 0.0001311383939515773,
+      "loss": 1.3626,
+      "step": 576000
+    },
+    {
+      "epoch": 16.93,
+      "grad_norm": 1.577664852142334,
+      "learning_rate": 0.00013099130196024593,
+      "loss": 1.3806,
+      "step": 576500
+    },
+    {
+      "epoch": 16.94,
+      "grad_norm": 16.11113166809082,
+      "learning_rate": 0.00013084420996891454,
+      "loss": 1.3787,
+      "step": 577000
+    },
+    {
+      "epoch": 16.96,
+      "grad_norm": 4.257967948913574,
+      "learning_rate": 0.00013069711797758317,
+      "loss": 1.355,
+      "step": 577500
+    },
+    {
+      "epoch": 16.97,
+      "grad_norm": 1.8505833148956299,
+      "learning_rate": 0.00013055002598625178,
+      "loss": 1.3702,
+      "step": 578000
+    },
+    {
+      "epoch": 16.98,
+      "grad_norm": 2.1093640327453613,
+      "learning_rate": 0.00013040293399492042,
+      "loss": 1.3775,
+      "step": 578500
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 1.8386383056640625,
+      "learning_rate": 0.00013025584200358902,
+      "loss": 1.3776,
+      "step": 579000
+    },
+    {
+      "epoch": 17.01,
+      "grad_norm": 7.34138298034668,
+      "learning_rate": 0.00013010875001225766,
+      "loss": 1.3215,
+      "step": 579500
+    },
+    {
+      "epoch": 17.03,
+      "grad_norm": 1.0172581672668457,
+      "learning_rate": 0.00012996165802092627,
+      "loss": 1.3147,
+      "step": 580000
+    },
+    {
+      "epoch": 17.04,
+      "grad_norm": 2.6336045265197754,
+      "learning_rate": 0.0001298145660295949,
+      "loss": 1.3365,
+      "step": 580500
+    },
+    {
+      "epoch": 17.06,
+      "grad_norm": 3.9906227588653564,
+      "learning_rate": 0.0001296674740382635,
+      "loss": 1.3245,
+      "step": 581000
+    },
+    {
+      "epoch": 17.07,
+      "grad_norm": 3.1087284088134766,
+      "learning_rate": 0.00012952038204693214,
+      "loss": 1.2864,
+      "step": 581500
+    },
+    {
+      "epoch": 17.09,
+      "grad_norm": 2.5867342948913574,
+      "learning_rate": 0.00012937329005560075,
+      "loss": 1.2947,
+      "step": 582000
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 1.2118226289749146,
+      "learning_rate": 0.00012922619806426938,
+      "loss": 1.3,
+      "step": 582500
+    },
+    {
+      "epoch": 17.12,
+      "grad_norm": 2.56510329246521,
+      "learning_rate": 0.000129079106072938,
+      "loss": 1.3147,
+      "step": 583000
+    },
+    {
+      "epoch": 17.13,
+      "grad_norm": 10.831042289733887,
+      "learning_rate": 0.00012893201408160663,
+      "loss": 1.3125,
+      "step": 583500
+    },
+    {
+      "epoch": 17.15,
+      "grad_norm": 10.36989688873291,
+      "learning_rate": 0.00012878492209027523,
+      "loss": 1.2969,
+      "step": 584000
+    },
+    {
+      "epoch": 17.16,
+      "grad_norm": 2.0238804817199707,
+      "learning_rate": 0.00012863783009894387,
+      "loss": 1.3044,
+      "step": 584500
+    },
+    {
+      "epoch": 17.18,
+      "grad_norm": 4.501575469970703,
+      "learning_rate": 0.00012849073810761248,
+      "loss": 1.3013,
+      "step": 585000
+    },
+    {
+      "epoch": 17.19,
+      "grad_norm": 4.236315727233887,
+      "learning_rate": 0.0001283436461162811,
+      "loss": 1.2991,
+      "step": 585500
+    },
+    {
+      "epoch": 17.21,
+      "grad_norm": 4.139219760894775,
+      "learning_rate": 0.00012819655412494972,
+      "loss": 1.2979,
+      "step": 586000
+    },
+    {
+      "epoch": 17.22,
+      "grad_norm": 2.1206071376800537,
+      "learning_rate": 0.00012804946213361835,
+      "loss": 1.3377,
+      "step": 586500
+    },
+    {
+      "epoch": 17.23,
+      "grad_norm": 1.7728540897369385,
+      "learning_rate": 0.000127902370142287,
+      "loss": 1.2992,
+      "step": 587000
+    },
+    {
+      "epoch": 17.25,
+      "grad_norm": 1.9550994634628296,
+      "learning_rate": 0.0001277552781509556,
+      "loss": 1.3311,
+      "step": 587500
+    },
+    {
+      "epoch": 17.26,
+      "grad_norm": 1.6412031650543213,
+      "learning_rate": 0.00012760818615962423,
+      "loss": 1.3101,
+      "step": 588000
+    },
+    {
+      "epoch": 17.28,
+      "grad_norm": 1.6908353567123413,
+      "learning_rate": 0.00012746109416829284,
+      "loss": 1.3357,
+      "step": 588500
+    },
+    {
+      "epoch": 17.29,
+      "grad_norm": 1.6260554790496826,
+      "learning_rate": 0.00012731400217696147,
+      "loss": 1.3128,
+      "step": 589000
+    },
+    {
+      "epoch": 17.31,
+      "grad_norm": 4.529758453369141,
+      "learning_rate": 0.00012716691018563008,
+      "loss": 1.3282,
+      "step": 589500
+    },
+    {
+      "epoch": 17.32,
+      "grad_norm": 1.872877836227417,
+      "learning_rate": 0.00012701981819429871,
+      "loss": 1.3271,
+      "step": 590000
+    },
+    {
+      "epoch": 17.34,
+      "grad_norm": 3.3399105072021484,
+      "learning_rate": 0.00012687272620296732,
+      "loss": 1.3104,
+      "step": 590500
+    },
+    {
+      "epoch": 17.35,
+      "grad_norm": 3.178553819656372,
+      "learning_rate": 0.00012672563421163596,
+      "loss": 1.3222,
+      "step": 591000
+    },
+    {
+      "epoch": 17.37,
+      "grad_norm": 2.042067527770996,
+      "learning_rate": 0.00012657854222030456,
+      "loss": 1.332,
+      "step": 591500
+    },
+    {
+      "epoch": 17.38,
+      "grad_norm": 2.1601064205169678,
+      "learning_rate": 0.0001264314502289732,
+      "loss": 1.3198,
+      "step": 592000
+    },
+    {
+      "epoch": 17.4,
+      "grad_norm": 1.6580477952957153,
+      "learning_rate": 0.0001262843582376418,
+      "loss": 1.3059,
+      "step": 592500
+    },
+    {
+      "epoch": 17.41,
+      "grad_norm": 3.175902843475342,
+      "learning_rate": 0.00012613726624631044,
+      "loss": 1.3262,
+      "step": 593000
+    },
+    {
+      "epoch": 17.43,
+      "grad_norm": 2.7562525272369385,
+      "learning_rate": 0.00012599017425497905,
+      "loss": 1.3389,
+      "step": 593500
+    },
+    {
+      "epoch": 17.44,
+      "grad_norm": 1.7579740285873413,
+      "learning_rate": 0.00012584308226364768,
+      "loss": 1.3308,
+      "step": 594000
+    },
+    {
+      "epoch": 17.45,
+      "grad_norm": 4.593905448913574,
+      "learning_rate": 0.0001256959902723163,
+      "loss": 1.3059,
+      "step": 594500
+    },
+    {
+      "epoch": 17.47,
+      "grad_norm": 1.438839077949524,
+      "learning_rate": 0.00012554889828098493,
+      "loss": 1.3159,
+      "step": 595000
+    },
+    {
+      "epoch": 17.48,
+      "grad_norm": 2.5471925735473633,
+      "learning_rate": 0.00012540180628965353,
+      "loss": 1.2989,
+      "step": 595500
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 1.8019795417785645,
+      "learning_rate": 0.00012525471429832217,
+      "loss": 1.3441,
+      "step": 596000
+    },
+    {
+      "epoch": 17.51,
+      "grad_norm": 2.0826618671417236,
+      "learning_rate": 0.00012510762230699078,
+      "loss": 1.3309,
+      "step": 596500
+    },
+    {
+      "epoch": 17.53,
+      "grad_norm": 1.820566177368164,
+      "learning_rate": 0.0001249605303156594,
+      "loss": 1.3224,
+      "step": 597000
+    },
+    {
+      "epoch": 17.54,
+      "grad_norm": 24.11446762084961,
+      "learning_rate": 0.00012481343832432802,
+      "loss": 1.2934,
+      "step": 597500
+    },
+    {
+      "epoch": 17.56,
+      "grad_norm": 1.5428298711776733,
+      "learning_rate": 0.00012466634633299665,
+      "loss": 1.3277,
+      "step": 598000
+    },
+    {
+      "epoch": 17.57,
+      "grad_norm": 2.171504259109497,
+      "learning_rate": 0.00012451925434166526,
+      "loss": 1.3365,
+      "step": 598500
+    },
+    {
+      "epoch": 17.59,
+      "grad_norm": 2.862025022506714,
+      "learning_rate": 0.0001243721623503339,
+      "loss": 1.3351,
+      "step": 599000
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 2.7436113357543945,
+      "learning_rate": 0.0001242250703590025,
+      "loss": 1.3203,
+      "step": 599500
+    },
+    {
+      "epoch": 17.62,
+      "grad_norm": 1.9902766942977905,
+      "learning_rate": 0.00012407797836767114,
+      "loss": 1.3324,
+      "step": 600000
+    },
+    {
+      "epoch": 17.63,
+      "grad_norm": 3.6419193744659424,
+      "learning_rate": 0.00012393088637633975,
+      "loss": 1.347,
+      "step": 600500
+    },
+    {
+      "epoch": 17.65,
+      "grad_norm": 1.8328200578689575,
+      "learning_rate": 0.00012378379438500838,
+      "loss": 1.3261,
+      "step": 601000
+    },
+    {
+      "epoch": 17.66,
+      "grad_norm": 3.2816476821899414,
+      "learning_rate": 0.000123636702393677,
+      "loss": 1.3228,
+      "step": 601500
+    },
+    {
+      "epoch": 17.67,
+      "grad_norm": 1.142104148864746,
+      "learning_rate": 0.00012348961040234562,
+      "loss": 1.3183,
+      "step": 602000
+    },
+    {
+      "epoch": 17.69,
+      "grad_norm": 2.1242527961730957,
+      "learning_rate": 0.00012334251841101423,
+      "loss": 1.3162,
+      "step": 602500
+    },
+    {
+      "epoch": 17.7,
+      "grad_norm": 9.647717475891113,
+      "learning_rate": 0.00012319542641968286,
+      "loss": 1.3473,
+      "step": 603000
+    },
+    {
+      "epoch": 17.72,
+      "grad_norm": 3.551119804382324,
+      "learning_rate": 0.00012304833442835147,
+      "loss": 1.3513,
+      "step": 603500
+    },
+    {
+      "epoch": 17.73,
+      "grad_norm": 1.6903916597366333,
+      "learning_rate": 0.0001229012424370201,
+      "loss": 1.3393,
+      "step": 604000
+    },
+    {
+      "epoch": 17.75,
+      "grad_norm": 2.54724383354187,
+      "learning_rate": 0.00012275415044568871,
+      "loss": 1.3402,
+      "step": 604500
+    },
+    {
+      "epoch": 17.76,
+      "grad_norm": 5.089727878570557,
+      "learning_rate": 0.00012260705845435735,
+      "loss": 1.303,
+      "step": 605000
+    },
+    {
+      "epoch": 17.78,
+      "grad_norm": 12.237126350402832,
+      "learning_rate": 0.00012245996646302596,
+      "loss": 1.3319,
+      "step": 605500
+    },
+    {
+      "epoch": 17.79,
+      "grad_norm": 1.576462745666504,
+      "learning_rate": 0.0001223128744716946,
+      "loss": 1.3265,
+      "step": 606000
+    },
+    {
+      "epoch": 17.81,
+      "grad_norm": 1.7792675495147705,
+      "learning_rate": 0.0001221657824803632,
+      "loss": 1.3335,
+      "step": 606500
+    },
+    {
+      "epoch": 17.82,
+      "grad_norm": 5.532106876373291,
+      "learning_rate": 0.00012201869048903183,
+      "loss": 1.3456,
+      "step": 607000
+    },
+    {
+      "epoch": 17.84,
+      "grad_norm": 3.333435297012329,
+      "learning_rate": 0.00012187159849770044,
+      "loss": 1.3332,
+      "step": 607500
+    },
+    {
+      "epoch": 17.85,
+      "grad_norm": 3.1190874576568604,
+      "learning_rate": 0.00012172450650636908,
+      "loss": 1.3368,
+      "step": 608000
+    },
+    {
+      "epoch": 17.87,
+      "grad_norm": 1.656201720237732,
+      "learning_rate": 0.0001215774145150377,
+      "loss": 1.3266,
+      "step": 608500
+    },
+    {
+      "epoch": 17.88,
+      "grad_norm": 2.088050365447998,
+      "learning_rate": 0.00012143032252370632,
+      "loss": 1.3128,
+      "step": 609000
+    },
+    {
+      "epoch": 17.9,
+      "grad_norm": 2.941950798034668,
+      "learning_rate": 0.00012128323053237493,
+      "loss": 1.3393,
+      "step": 609500
+    },
+    {
+      "epoch": 17.91,
+      "grad_norm": 1.8323218822479248,
+      "learning_rate": 0.00012113613854104356,
+      "loss": 1.3255,
+      "step": 610000
+    },
+    {
+      "epoch": 17.92,
+      "grad_norm": 1.3735178709030151,
+      "learning_rate": 0.00012098904654971218,
+      "loss": 1.3095,
+      "step": 610500
+    },
+    {
+      "epoch": 17.94,
+      "grad_norm": 2.9366865158081055,
+      "learning_rate": 0.0001208419545583808,
+      "loss": 1.3401,
+      "step": 611000
+    },
+    {
+      "epoch": 17.95,
+      "grad_norm": 3.2511463165283203,
+      "learning_rate": 0.00012069486256704941,
+      "loss": 1.3396,
+      "step": 611500
+    },
+    {
+      "epoch": 17.97,
+      "grad_norm": 2.2338857650756836,
+      "learning_rate": 0.00012054777057571804,
+      "loss": 1.3286,
+      "step": 612000
+    },
+    {
+      "epoch": 17.98,
+      "grad_norm": 1.5889378786087036,
+      "learning_rate": 0.00012040067858438667,
+      "loss": 1.3163,
+      "step": 612500
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 2.792966365814209,
+      "learning_rate": 0.00012025358659305529,
+      "loss": 1.2973,
+      "step": 613000
+    },
+    {
+      "epoch": 18.01,
+      "grad_norm": 2.239032745361328,
+      "learning_rate": 0.00012010649460172391,
+      "loss": 1.3099,
+      "step": 613500
+    },
+    {
+      "epoch": 18.03,
+      "grad_norm": 2.293813705444336,
+      "learning_rate": 0.00011995940261039253,
+      "loss": 1.2568,
+      "step": 614000
+    },
+    {
+      "epoch": 18.04,
+      "grad_norm": 2.175294876098633,
+      "learning_rate": 0.00011981231061906115,
+      "loss": 1.2661,
+      "step": 614500
+    },
+    {
+      "epoch": 18.06,
+      "grad_norm": 1.7673249244689941,
+      "learning_rate": 0.00011966521862772977,
+      "loss": 1.2747,
+      "step": 615000
+    },
+    {
+      "epoch": 18.07,
+      "grad_norm": 2.6049957275390625,
+      "learning_rate": 0.00011951812663639839,
+      "loss": 1.2851,
+      "step": 615500
+    },
+    {
+      "epoch": 18.09,
+      "grad_norm": 2.0433642864227295,
+      "learning_rate": 0.00011937103464506703,
+      "loss": 1.2846,
+      "step": 616000
+    },
+    {
+      "epoch": 18.1,
+      "grad_norm": 2.120561122894287,
+      "learning_rate": 0.00011922394265373563,
+      "loss": 1.2634,
+      "step": 616500
+    },
+    {
+      "epoch": 18.12,
+      "grad_norm": 2.6130003929138184,
+      "learning_rate": 0.00011907685066240426,
+      "loss": 1.2844,
+      "step": 617000
+    },
+    {
+      "epoch": 18.13,
+      "grad_norm": 1.8312240839004517,
+      "learning_rate": 0.00011892975867107288,
+      "loss": 1.289,
+      "step": 617500
+    },
+    {
+      "epoch": 18.14,
+      "grad_norm": 1.5402841567993164,
+      "learning_rate": 0.00011878266667974151,
+      "loss": 1.2578,
+      "step": 618000
+    },
+    {
+      "epoch": 18.16,
+      "grad_norm": 2.0804052352905273,
+      "learning_rate": 0.00011863557468841012,
+      "loss": 1.2558,
+      "step": 618500
+    },
+    {
+      "epoch": 18.17,
+      "grad_norm": 1.77811861038208,
+      "learning_rate": 0.00011848848269707874,
+      "loss": 1.2853,
+      "step": 619000
+    },
+    {
+      "epoch": 18.19,
+      "grad_norm": 6.067068099975586,
+      "learning_rate": 0.00011834139070574736,
+      "loss": 1.2941,
+      "step": 619500
+    },
+    {
+      "epoch": 18.2,
+      "grad_norm": 2.4684622287750244,
+      "learning_rate": 0.000118194298714416,
+      "loss": 1.2749,
+      "step": 620000
+    },
+    {
+      "epoch": 18.22,
+      "grad_norm": 3.7913448810577393,
+      "learning_rate": 0.0001180472067230846,
+      "loss": 1.298,
+      "step": 620500
+    },
+    {
+      "epoch": 18.23,
+      "grad_norm": 1.343802571296692,
+      "learning_rate": 0.00011790011473175322,
+      "loss": 1.2954,
+      "step": 621000
+    },
+    {
+      "epoch": 18.25,
+      "grad_norm": 1.797194480895996,
+      "learning_rate": 0.00011775302274042185,
+      "loss": 1.2718,
+      "step": 621500
+    },
+    {
+      "epoch": 18.26,
+      "grad_norm": 2.3011558055877686,
+      "learning_rate": 0.00011760593074909048,
+      "loss": 1.2966,
+      "step": 622000
+    },
+    {
+      "epoch": 18.28,
+      "grad_norm": 2.5689167976379395,
+      "learning_rate": 0.00011745883875775909,
+      "loss": 1.286,
+      "step": 622500
+    },
+    {
+      "epoch": 18.29,
+      "grad_norm": 8.428597450256348,
+      "learning_rate": 0.00011731174676642771,
+      "loss": 1.2872,
+      "step": 623000
+    },
+    {
+      "epoch": 18.31,
+      "grad_norm": 1.1936590671539307,
+      "learning_rate": 0.00011716465477509633,
+      "loss": 1.2621,
+      "step": 623500
+    },
+    {
+      "epoch": 18.32,
+      "grad_norm": 1.5052251815795898,
+      "learning_rate": 0.00011701756278376497,
+      "loss": 1.2758,
+      "step": 624000
+    },
+    {
+      "epoch": 18.34,
+      "grad_norm": 1.296823263168335,
+      "learning_rate": 0.00011687047079243357,
+      "loss": 1.266,
+      "step": 624500
+    },
+    {
+      "epoch": 18.35,
+      "grad_norm": 3.121631383895874,
+      "learning_rate": 0.0001167233788011022,
+      "loss": 1.2858,
+      "step": 625000
+    },
+    {
+      "epoch": 18.36,
+      "grad_norm": 7.022789478302002,
+      "learning_rate": 0.00011657628680977081,
+      "loss": 1.2986,
+      "step": 625500
+    },
+    {
+      "epoch": 18.38,
+      "grad_norm": 11.550426483154297,
+      "learning_rate": 0.00011642919481843945,
+      "loss": 1.2736,
+      "step": 626000
+    },
+    {
+      "epoch": 18.39,
+      "grad_norm": 2.277326822280884,
+      "learning_rate": 0.00011628210282710807,
+      "loss": 1.3052,
+      "step": 626500
+    },
+    {
+      "epoch": 18.41,
+      "grad_norm": 3.439568519592285,
+      "learning_rate": 0.00011613501083577668,
+      "loss": 1.3042,
+      "step": 627000
+    },
+    {
+      "epoch": 18.42,
+      "grad_norm": 8.868010520935059,
+      "learning_rate": 0.0001159879188444453,
+      "loss": 1.2612,
+      "step": 627500
+    },
+    {
+      "epoch": 18.44,
+      "grad_norm": 1.5216681957244873,
+      "learning_rate": 0.00011584082685311393,
+      "loss": 1.2634,
+      "step": 628000
+    },
+    {
+      "epoch": 18.45,
+      "grad_norm": 2.113112688064575,
+      "learning_rate": 0.00011569373486178256,
+      "loss": 1.2681,
+      "step": 628500
+    },
+    {
+      "epoch": 18.47,
+      "grad_norm": 2.9850034713745117,
+      "learning_rate": 0.00011554664287045116,
+      "loss": 1.2865,
+      "step": 629000
+    },
+    {
+      "epoch": 18.48,
+      "grad_norm": 2.309042453765869,
+      "learning_rate": 0.00011539955087911978,
+      "loss": 1.2674,
+      "step": 629500
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 3.1562564373016357,
+      "learning_rate": 0.00011525245888778842,
+      "loss": 1.262,
+      "step": 630000
+    },
+    {
+      "epoch": 18.51,
+      "grad_norm": 3.4192593097686768,
+      "learning_rate": 0.00011510536689645704,
+      "loss": 1.2729,
+      "step": 630500
+    },
+    {
+      "epoch": 18.53,
+      "grad_norm": 2.193237543106079,
+      "learning_rate": 0.00011495827490512565,
+      "loss": 1.2586,
+      "step": 631000
+    },
+    {
+      "epoch": 18.54,
+      "grad_norm": 8.091324806213379,
+      "learning_rate": 0.00011481118291379427,
+      "loss": 1.2649,
+      "step": 631500
+    },
+    {
+      "epoch": 18.56,
+      "grad_norm": 2.0321621894836426,
+      "learning_rate": 0.0001146640909224629,
+      "loss": 1.2792,
+      "step": 632000
+    },
+    {
+      "epoch": 18.57,
+      "grad_norm": 1.623028039932251,
+      "learning_rate": 0.00011451699893113152,
+      "loss": 1.3013,
+      "step": 632500
+    },
+    {
+      "epoch": 18.59,
+      "grad_norm": 2.7462871074676514,
+      "learning_rate": 0.00011436990693980013,
+      "loss": 1.2858,
+      "step": 633000
+    },
+    {
+      "epoch": 18.6,
+      "grad_norm": 22.743488311767578,
+      "learning_rate": 0.00011422281494846875,
+      "loss": 1.2739,
+      "step": 633500
+    },
+    {
+      "epoch": 18.61,
+      "grad_norm": 4.659852981567383,
+      "learning_rate": 0.00011407572295713739,
+      "loss": 1.2773,
+      "step": 634000
+    },
+    {
+      "epoch": 18.63,
+      "grad_norm": 1.8134876489639282,
+      "learning_rate": 0.00011392863096580601,
+      "loss": 1.274,
+      "step": 634500
+    },
+    {
+      "epoch": 18.64,
+      "grad_norm": 2.266272783279419,
+      "learning_rate": 0.00011378153897447463,
+      "loss": 1.3084,
+      "step": 635000
+    },
+    {
+      "epoch": 18.66,
+      "grad_norm": 1.4625264406204224,
+      "learning_rate": 0.00011363444698314324,
+      "loss": 1.2794,
+      "step": 635500
+    },
+    {
+      "epoch": 18.67,
+      "grad_norm": 1.57483971118927,
+      "learning_rate": 0.00011348735499181187,
+      "loss": 1.2872,
+      "step": 636000
+    },
+    {
+      "epoch": 18.69,
+      "grad_norm": 1.3935645818710327,
+      "learning_rate": 0.00011334026300048049,
+      "loss": 1.2809,
+      "step": 636500
+    },
+    {
+      "epoch": 18.7,
+      "grad_norm": 3.3207247257232666,
+      "learning_rate": 0.00011319317100914911,
+      "loss": 1.2764,
+      "step": 637000
+    },
+    {
+      "epoch": 18.72,
+      "grad_norm": 4.18394660949707,
+      "learning_rate": 0.00011304607901781772,
+      "loss": 1.2855,
+      "step": 637500
+    },
+    {
+      "epoch": 18.73,
+      "grad_norm": 1.6960937976837158,
+      "learning_rate": 0.00011289898702648636,
+      "loss": 1.2807,
+      "step": 638000
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 6.864727020263672,
+      "learning_rate": 0.00011275189503515498,
+      "loss": 1.2736,
+      "step": 638500
+    },
+    {
+      "epoch": 18.76,
+      "grad_norm": 1.7619363069534302,
+      "learning_rate": 0.0001126048030438236,
+      "loss": 1.3063,
+      "step": 639000
+    },
+    {
+      "epoch": 18.78,
+      "grad_norm": 1.6224156618118286,
+      "learning_rate": 0.0001124577110524922,
+      "loss": 1.3031,
+      "step": 639500
+    },
+    {
+      "epoch": 18.79,
+      "grad_norm": 4.564239025115967,
+      "learning_rate": 0.00011231061906116084,
+      "loss": 1.3185,
+      "step": 640000
+    },
+    {
+      "epoch": 18.81,
+      "grad_norm": 2.237443208694458,
+      "learning_rate": 0.00011216352706982946,
+      "loss": 1.2612,
+      "step": 640500
+    },
+    {
+      "epoch": 18.82,
+      "grad_norm": 1.7501612901687622,
+      "learning_rate": 0.00011201643507849808,
+      "loss": 1.278,
+      "step": 641000
+    },
+    {
+      "epoch": 18.83,
+      "grad_norm": 1.1969166994094849,
+      "learning_rate": 0.00011186934308716669,
+      "loss": 1.2979,
+      "step": 641500
+    },
+    {
+      "epoch": 18.85,
+      "grad_norm": 2.23476505279541,
+      "learning_rate": 0.00011172225109583533,
+      "loss": 1.2796,
+      "step": 642000
+    },
+    {
+      "epoch": 18.86,
+      "grad_norm": 3.1486740112304688,
+      "learning_rate": 0.00011157515910450395,
+      "loss": 1.3212,
+      "step": 642500
+    },
+    {
+      "epoch": 18.88,
+      "grad_norm": 3.5119261741638184,
+      "learning_rate": 0.00011142806711317257,
+      "loss": 1.3082,
+      "step": 643000
+    },
+    {
+      "epoch": 18.89,
+      "grad_norm": 3.0694100856781006,
+      "learning_rate": 0.00011128097512184118,
+      "loss": 1.2643,
+      "step": 643500
+    },
+    {
+      "epoch": 18.91,
+      "grad_norm": 2.235492467880249,
+      "learning_rate": 0.00011113388313050981,
+      "loss": 1.2832,
+      "step": 644000
+    },
+    {
+      "epoch": 18.92,
+      "grad_norm": 2.627898931503296,
+      "learning_rate": 0.00011098679113917843,
+      "loss": 1.2863,
+      "step": 644500
+    },
+    {
+      "epoch": 18.94,
+      "grad_norm": 1.390758991241455,
+      "learning_rate": 0.00011083969914784705,
+      "loss": 1.2949,
+      "step": 645000
+    },
+    {
+      "epoch": 18.95,
+      "grad_norm": 1.9412230253219604,
+      "learning_rate": 0.00011069260715651567,
+      "loss": 1.2765,
+      "step": 645500
+    },
+    {
+      "epoch": 18.97,
+      "grad_norm": 7.697941780090332,
+      "learning_rate": 0.0001105455151651843,
+      "loss": 1.2872,
+      "step": 646000
+    },
+    {
+      "epoch": 18.98,
+      "grad_norm": 2.0770368576049805,
+      "learning_rate": 0.00011039842317385292,
+      "loss": 1.3058,
+      "step": 646500
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 1.8311492204666138,
+      "learning_rate": 0.00011025133118252154,
+      "loss": 1.3229,
+      "step": 647000
+    },
+    {
+      "epoch": 19.01,
+      "grad_norm": 2.744004726409912,
+      "learning_rate": 0.00011010423919119017,
+      "loss": 1.253,
+      "step": 647500
+    },
+    {
+      "epoch": 19.03,
+      "grad_norm": 2.0883147716522217,
+      "learning_rate": 0.00010995714719985879,
+      "loss": 1.2348,
+      "step": 648000
+    },
+    {
+      "epoch": 19.04,
+      "grad_norm": 13.648962020874023,
+      "learning_rate": 0.0001098100552085274,
+      "loss": 1.2572,
+      "step": 648500
+    },
+    {
+      "epoch": 19.05,
+      "grad_norm": 3.89188814163208,
+      "learning_rate": 0.00010966296321719602,
+      "loss": 1.2312,
+      "step": 649000
+    },
+    {
+      "epoch": 19.07,
+      "grad_norm": 12.498522758483887,
+      "learning_rate": 0.00010951587122586466,
+      "loss": 1.2285,
+      "step": 649500
+    },
+    {
+      "epoch": 19.08,
+      "grad_norm": 1.9123058319091797,
+      "learning_rate": 0.00010936877923453328,
+      "loss": 1.2223,
+      "step": 650000
+    },
+    {
+      "epoch": 19.1,
+      "grad_norm": 1.9629552364349365,
+      "learning_rate": 0.00010922168724320188,
+      "loss": 1.2323,
+      "step": 650500
+    },
+    {
+      "epoch": 19.11,
+      "grad_norm": 16.135618209838867,
+      "learning_rate": 0.0001090745952518705,
+      "loss": 1.1976,
+      "step": 651000
+    },
+    {
+      "epoch": 19.13,
+      "grad_norm": 2.948089361190796,
+      "learning_rate": 0.00010892750326053914,
+      "loss": 1.2367,
+      "step": 651500
+    },
+    {
+      "epoch": 19.14,
+      "grad_norm": 2.4549195766448975,
+      "learning_rate": 0.00010878041126920776,
+      "loss": 1.2345,
+      "step": 652000
+    },
+    {
+      "epoch": 19.16,
+      "grad_norm": 1.7298622131347656,
+      "learning_rate": 0.00010863331927787637,
+      "loss": 1.2477,
+      "step": 652500
+    },
+    {
+      "epoch": 19.17,
+      "grad_norm": 3.237170696258545,
+      "learning_rate": 0.00010848622728654499,
+      "loss": 1.2578,
+      "step": 653000
+    },
+    {
+      "epoch": 19.19,
+      "grad_norm": 2.876091718673706,
+      "learning_rate": 0.00010833913529521362,
+      "loss": 1.2657,
+      "step": 653500
+    },
+    {
+      "epoch": 19.2,
+      "grad_norm": 2.6806657314300537,
+      "learning_rate": 0.00010819204330388225,
+      "loss": 1.2369,
+      "step": 654000
+    },
+    {
+      "epoch": 19.22,
+      "grad_norm": 1.682861328125,
+      "learning_rate": 0.00010804495131255085,
+      "loss": 1.2353,
+      "step": 654500
+    },
+    {
+      "epoch": 19.23,
+      "grad_norm": 1.420599102973938,
+      "learning_rate": 0.00010789785932121947,
+      "loss": 1.2292,
+      "step": 655000
+    },
+    {
+      "epoch": 19.25,
+      "grad_norm": 2.785423517227173,
+      "learning_rate": 0.00010775076732988811,
+      "loss": 1.2299,
+      "step": 655500
+    },
+    {
+      "epoch": 19.26,
+      "grad_norm": 3.515298843383789,
+      "learning_rate": 0.00010760367533855673,
+      "loss": 1.2343,
+      "step": 656000
+    },
+    {
+      "epoch": 19.28,
+      "grad_norm": 8.15224552154541,
+      "learning_rate": 0.00010745658334722534,
+      "loss": 1.262,
+      "step": 656500
+    },
+    {
+      "epoch": 19.29,
+      "grad_norm": 2.5358471870422363,
+      "learning_rate": 0.00010730949135589396,
+      "loss": 1.2579,
+      "step": 657000
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 2.8267860412597656,
+      "learning_rate": 0.0001071623993645626,
+      "loss": 1.2195,
+      "step": 657500
+    },
+    {
+      "epoch": 19.32,
+      "grad_norm": 2.0857648849487305,
+      "learning_rate": 0.00010701530737323122,
+      "loss": 1.2244,
+      "step": 658000
+    },
+    {
+      "epoch": 19.33,
+      "grad_norm": 2.2825379371643066,
+      "learning_rate": 0.00010686821538189984,
+      "loss": 1.2502,
+      "step": 658500
+    },
+    {
+      "epoch": 19.35,
+      "grad_norm": 2.1249475479125977,
+      "learning_rate": 0.00010672112339056844,
+      "loss": 1.2408,
+      "step": 659000
+    },
+    {
+      "epoch": 19.36,
+      "grad_norm": 1.9578863382339478,
+      "learning_rate": 0.00010657403139923708,
+      "loss": 1.2599,
+      "step": 659500
+    },
+    {
+      "epoch": 19.38,
+      "grad_norm": 1.7473647594451904,
+      "learning_rate": 0.0001064269394079057,
+      "loss": 1.2482,
+      "step": 660000
+    },
+    {
+      "epoch": 19.39,
+      "grad_norm": 1.9577364921569824,
+      "learning_rate": 0.00010627984741657432,
+      "loss": 1.2389,
+      "step": 660500
+    },
+    {
+      "epoch": 19.41,
+      "grad_norm": 6.437145233154297,
+      "learning_rate": 0.00010613275542524293,
+      "loss": 1.2327,
+      "step": 661000
+    },
+    {
+      "epoch": 19.42,
+      "grad_norm": 36.84048080444336,
+      "learning_rate": 0.00010598566343391156,
+      "loss": 1.2553,
+      "step": 661500
+    },
+    {
+      "epoch": 19.44,
+      "grad_norm": 2.4691734313964844,
+      "learning_rate": 0.00010583857144258018,
+      "loss": 1.2551,
+      "step": 662000
+    },
+    {
+      "epoch": 19.45,
+      "grad_norm": 3.6141631603240967,
+      "learning_rate": 0.0001056914794512488,
+      "loss": 1.2459,
+      "step": 662500
+    },
+    {
+      "epoch": 19.47,
+      "grad_norm": 4.1790008544921875,
+      "learning_rate": 0.00010554438745991741,
+      "loss": 1.2299,
+      "step": 663000
+    },
+    {
+      "epoch": 19.48,
+      "grad_norm": 2.3077800273895264,
+      "learning_rate": 0.00010539729546858605,
+      "loss": 1.2354,
+      "step": 663500
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 2.0375678539276123,
+      "learning_rate": 0.00010525020347725467,
+      "loss": 1.2332,
+      "step": 664000
+    },
+    {
+      "epoch": 19.51,
+      "grad_norm": 2.190852403640747,
+      "learning_rate": 0.00010510311148592329,
+      "loss": 1.2343,
+      "step": 664500
+    },
+    {
+      "epoch": 19.52,
+      "grad_norm": 1.7503656148910522,
+      "learning_rate": 0.0001049560194945919,
+      "loss": 1.2328,
+      "step": 665000
+    },
+    {
+      "epoch": 19.54,
+      "grad_norm": 2.364180326461792,
+      "learning_rate": 0.00010480892750326053,
+      "loss": 1.245,
+      "step": 665500
+    },
+    {
+      "epoch": 19.55,
+      "grad_norm": 6.837544918060303,
+      "learning_rate": 0.00010466183551192915,
+      "loss": 1.2361,
+      "step": 666000
+    },
+    {
+      "epoch": 19.57,
+      "grad_norm": 18.344188690185547,
+      "learning_rate": 0.00010451474352059777,
+      "loss": 1.2527,
+      "step": 666500
+    },
+    {
+      "epoch": 19.58,
+      "grad_norm": 4.090867519378662,
+      "learning_rate": 0.00010436765152926638,
+      "loss": 1.2573,
+      "step": 667000
+    },
+    {
+      "epoch": 19.6,
+      "grad_norm": 61.80951690673828,
+      "learning_rate": 0.00010422055953793502,
+      "loss": 1.2776,
+      "step": 667500
+    },
+    {
+      "epoch": 19.61,
+      "grad_norm": 1.9649507999420166,
+      "learning_rate": 0.00010407346754660364,
+      "loss": 1.2479,
+      "step": 668000
+    },
+    {
+      "epoch": 19.63,
+      "grad_norm": 4.030837535858154,
+      "learning_rate": 0.00010392637555527226,
+      "loss": 1.2458,
+      "step": 668500
+    },
+    {
+      "epoch": 19.64,
+      "grad_norm": 3.310805082321167,
+      "learning_rate": 0.00010377928356394088,
+      "loss": 1.2491,
+      "step": 669000
+    },
+    {
+      "epoch": 19.66,
+      "grad_norm": 6.558318138122559,
+      "learning_rate": 0.0001036321915726095,
+      "loss": 1.2562,
+      "step": 669500
+    },
+    {
+      "epoch": 19.67,
+      "grad_norm": 2.0995540618896484,
+      "learning_rate": 0.00010348509958127812,
+      "loss": 1.2529,
+      "step": 670000
+    },
+    {
+      "epoch": 19.69,
+      "grad_norm": 3.003690242767334,
+      "learning_rate": 0.00010333800758994674,
+      "loss": 1.2717,
+      "step": 670500
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 2.099637269973755,
+      "learning_rate": 0.00010319091559861536,
+      "loss": 1.2632,
+      "step": 671000
+    },
+    {
+      "epoch": 19.72,
+      "grad_norm": 35.86410140991211,
+      "learning_rate": 0.000103043823607284,
+      "loss": 1.2763,
+      "step": 671500
+    },
+    {
+      "epoch": 19.73,
+      "grad_norm": 1.7510465383529663,
+      "learning_rate": 0.0001028967316159526,
+      "loss": 1.2535,
+      "step": 672000
+    },
+    {
+      "epoch": 19.74,
+      "grad_norm": 10.661267280578613,
+      "learning_rate": 0.00010274963962462123,
+      "loss": 1.2346,
+      "step": 672500
+    },
+    {
+      "epoch": 19.76,
+      "grad_norm": 1.9645477533340454,
+      "learning_rate": 0.00010260254763328985,
+      "loss": 1.2398,
+      "step": 673000
+    },
+    {
+      "epoch": 19.77,
+      "grad_norm": 2.795703172683716,
+      "learning_rate": 0.00010245545564195848,
+      "loss": 1.2683,
+      "step": 673500
+    },
+    {
+      "epoch": 19.79,
+      "grad_norm": 3.7908451557159424,
+      "learning_rate": 0.00010230836365062709,
+      "loss": 1.2439,
+      "step": 674000
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 2.028703451156616,
+      "learning_rate": 0.00010216127165929571,
+      "loss": 1.2518,
+      "step": 674500
+    },
+    {
+      "epoch": 19.82,
+      "grad_norm": 2.059154510498047,
+      "learning_rate": 0.00010201417966796433,
+      "loss": 1.2359,
+      "step": 675000
+    },
+    {
+      "epoch": 19.83,
+      "grad_norm": 1.3817317485809326,
+      "learning_rate": 0.00010186708767663297,
+      "loss": 1.27,
+      "step": 675500
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 1.8642240762710571,
+      "learning_rate": 0.00010171999568530158,
+      "loss": 1.2604,
+      "step": 676000
+    },
+    {
+      "epoch": 19.86,
+      "grad_norm": 1.7178900241851807,
+      "learning_rate": 0.0001015729036939702,
+      "loss": 1.2509,
+      "step": 676500
+    },
+    {
+      "epoch": 19.88,
+      "grad_norm": 6.73402738571167,
+      "learning_rate": 0.00010142581170263882,
+      "loss": 1.2566,
+      "step": 677000
+    },
+    {
+      "epoch": 19.89,
+      "grad_norm": 1.7260433435440063,
+      "learning_rate": 0.00010127871971130745,
+      "loss": 1.258,
+      "step": 677500
+    },
+    {
+      "epoch": 19.91,
+      "grad_norm": 2.0348527431488037,
+      "learning_rate": 0.00010113162771997606,
+      "loss": 1.234,
+      "step": 678000
+    },
+    {
+      "epoch": 19.92,
+      "grad_norm": 3.1736955642700195,
+      "learning_rate": 0.00010098453572864468,
+      "loss": 1.241,
+      "step": 678500
+    },
+    {
+      "epoch": 19.94,
+      "grad_norm": 10.23302173614502,
+      "learning_rate": 0.0001008374437373133,
+      "loss": 1.2604,
+      "step": 679000
+    },
+    {
+      "epoch": 19.95,
+      "grad_norm": 2.974153995513916,
+      "learning_rate": 0.00010069035174598194,
+      "loss": 1.2525,
+      "step": 679500
+    },
+    {
+      "epoch": 19.97,
+      "grad_norm": 2.5226101875305176,
+      "learning_rate": 0.00010054325975465056,
+      "loss": 1.264,
+      "step": 680000
+    },
+    {
+      "epoch": 19.98,
+      "grad_norm": 2.472259521484375,
+      "learning_rate": 0.00010039616776331917,
+      "loss": 1.2299,
+      "step": 680500
+    },
+    {
+      "epoch": 19.99,
+      "grad_norm": 2.3573238849639893,
+      "learning_rate": 0.00010024907577198779,
+      "loss": 1.2657,
+      "step": 681000
+    },
+    {
+      "epoch": 20.01,
+      "grad_norm": 3.6109812259674072,
+      "learning_rate": 0.00010010198378065642,
+      "loss": 1.2247,
+      "step": 681500
+    },
+    {
+      "epoch": 20.02,
+      "grad_norm": 1.8781336545944214,
+      "learning_rate": 9.995489178932504e-05,
+      "loss": 1.2068,
+      "step": 682000
+    },
+    {
+      "epoch": 20.04,
+      "grad_norm": 1.4420850276947021,
+      "learning_rate": 9.980779979799365e-05,
+      "loss": 1.2,
+      "step": 682500
+    },
+    {
+      "epoch": 20.05,
+      "grad_norm": 3.082235097885132,
+      "learning_rate": 9.966070780666227e-05,
+      "loss": 1.2062,
+      "step": 683000
+    },
+    {
+      "epoch": 20.07,
+      "grad_norm": 1.9524058103561401,
+      "learning_rate": 9.95136158153309e-05,
+      "loss": 1.1999,
+      "step": 683500
+    },
+    {
+      "epoch": 20.08,
+      "grad_norm": 1.9696966409683228,
+      "learning_rate": 9.936652382399953e-05,
+      "loss": 1.1944,
+      "step": 684000
+    },
+    {
+      "epoch": 20.1,
+      "grad_norm": 3.851034641265869,
+      "learning_rate": 9.921943183266813e-05,
+      "loss": 1.1913,
+      "step": 684500
+    },
+    {
+      "epoch": 20.11,
+      "grad_norm": 1.7799595594406128,
+      "learning_rate": 9.907233984133676e-05,
+      "loss": 1.21,
+      "step": 685000
+    },
+    {
+      "epoch": 20.13,
+      "grad_norm": 2.3180084228515625,
+      "learning_rate": 9.892524785000539e-05,
+      "loss": 1.1741,
+      "step": 685500
+    },
+    {
+      "epoch": 20.14,
+      "grad_norm": 2.012601375579834,
+      "learning_rate": 9.877815585867401e-05,
+      "loss": 1.1956,
+      "step": 686000
+    },
+    {
+      "epoch": 20.16,
+      "grad_norm": 2.5793349742889404,
+      "learning_rate": 9.863106386734262e-05,
+      "loss": 1.2011,
+      "step": 686500
+    },
+    {
+      "epoch": 20.17,
+      "grad_norm": 1.7113804817199707,
+      "learning_rate": 9.848397187601124e-05,
+      "loss": 1.2382,
+      "step": 687000
+    },
+    {
+      "epoch": 20.19,
+      "grad_norm": 1.8239011764526367,
+      "learning_rate": 9.833687988467988e-05,
+      "loss": 1.1904,
+      "step": 687500
+    },
+    {
+      "epoch": 20.2,
+      "grad_norm": 8.440109252929688,
+      "learning_rate": 9.81897878933485e-05,
+      "loss": 1.2087,
+      "step": 688000
+    },
+    {
+      "epoch": 20.21,
+      "grad_norm": 2.4333038330078125,
+      "learning_rate": 9.80426959020171e-05,
+      "loss": 1.209,
+      "step": 688500
+    },
+    {
+      "epoch": 20.23,
+      "grad_norm": 3.217632532119751,
+      "learning_rate": 9.789560391068572e-05,
+      "loss": 1.1964,
+      "step": 689000
+    },
+    {
+      "epoch": 20.24,
+      "grad_norm": 3.0806314945220947,
+      "learning_rate": 9.774851191935436e-05,
+      "loss": 1.2227,
+      "step": 689500
+    },
+    {
+      "epoch": 20.26,
+      "grad_norm": 21.035554885864258,
+      "learning_rate": 9.760141992802298e-05,
+      "loss": 1.1899,
+      "step": 690000
+    },
+    {
+      "epoch": 20.27,
+      "grad_norm": 4.15212345123291,
+      "learning_rate": 9.74543279366916e-05,
+      "loss": 1.1919,
+      "step": 690500
+    },
+    {
+      "epoch": 20.29,
+      "grad_norm": 8.08483600616455,
+      "learning_rate": 9.730723594536021e-05,
+      "loss": 1.2083,
+      "step": 691000
+    },
+    {
+      "epoch": 20.3,
+      "grad_norm": 2.5341944694519043,
+      "learning_rate": 9.716014395402884e-05,
+      "loss": 1.181,
+      "step": 691500
+    },
+    {
+      "epoch": 20.32,
+      "grad_norm": 7.360681533813477,
+      "learning_rate": 9.701305196269747e-05,
+      "loss": 1.1975,
+      "step": 692000
+    },
+    {
+      "epoch": 20.33,
+      "grad_norm": 1.5373992919921875,
+      "learning_rate": 9.686595997136609e-05,
+      "loss": 1.191,
+      "step": 692500
+    },
+    {
+      "epoch": 20.35,
+      "grad_norm": 1.8679372072219849,
+      "learning_rate": 9.67188679800347e-05,
+      "loss": 1.213,
+      "step": 693000
+    },
+    {
+      "epoch": 20.36,
+      "grad_norm": 3.206306219100952,
+      "learning_rate": 9.657177598870333e-05,
+      "loss": 1.2051,
+      "step": 693500
+    },
+    {
+      "epoch": 20.38,
+      "grad_norm": 2.6320650577545166,
+      "learning_rate": 9.642468399737195e-05,
+      "loss": 1.1851,
+      "step": 694000
+    },
+    {
+      "epoch": 20.39,
+      "grad_norm": 3.6654086112976074,
+      "learning_rate": 9.627759200604057e-05,
+      "loss": 1.2063,
+      "step": 694500
+    },
+    {
+      "epoch": 20.41,
+      "grad_norm": 3.010348081588745,
+      "learning_rate": 9.613050001470918e-05,
+      "loss": 1.2239,
+      "step": 695000
+    },
+    {
+      "epoch": 20.42,
+      "grad_norm": 27.742935180664062,
+      "learning_rate": 9.598340802337781e-05,
+      "loss": 1.2084,
+      "step": 695500
+    },
+    {
+      "epoch": 20.43,
+      "grad_norm": 2.783250570297241,
+      "learning_rate": 9.583631603204643e-05,
+      "loss": 1.2263,
+      "step": 696000
+    },
+    {
+      "epoch": 20.45,
+      "grad_norm": 2.516063690185547,
+      "learning_rate": 9.568922404071506e-05,
+      "loss": 1.2084,
+      "step": 696500
+    },
+    {
+      "epoch": 20.46,
+      "grad_norm": 1.8266417980194092,
+      "learning_rate": 9.554213204938366e-05,
+      "loss": 1.2052,
+      "step": 697000
+    },
+    {
+      "epoch": 20.48,
+      "grad_norm": 1.2132940292358398,
+      "learning_rate": 9.53950400580523e-05,
+      "loss": 1.2132,
+      "step": 697500
+    },
+    {
+      "epoch": 20.49,
+      "grad_norm": 4.550230503082275,
+      "learning_rate": 9.524794806672092e-05,
+      "loss": 1.2263,
+      "step": 698000
+    },
+    {
+      "epoch": 20.51,
+      "grad_norm": 2.1874680519104004,
+      "learning_rate": 9.510085607538954e-05,
+      "loss": 1.1855,
+      "step": 698500
+    },
+    {
+      "epoch": 20.52,
+      "grad_norm": 1.8161512613296509,
+      "learning_rate": 9.495376408405815e-05,
+      "loss": 1.1949,
+      "step": 699000
+    },
+    {
+      "epoch": 20.54,
+      "grad_norm": 1.3943161964416504,
+      "learning_rate": 9.480667209272678e-05,
+      "loss": 1.1944,
+      "step": 699500
+    },
+    {
+      "epoch": 20.55,
+      "grad_norm": 3.2997055053710938,
+      "learning_rate": 9.46595801013954e-05,
+      "loss": 1.2213,
+      "step": 700000
+    },
+    {
+      "epoch": 20.57,
+      "grad_norm": 1.9309003353118896,
+      "learning_rate": 9.451248811006402e-05,
+      "loss": 1.2163,
+      "step": 700500
+    },
+    {
+      "epoch": 20.58,
+      "grad_norm": 2.734384775161743,
+      "learning_rate": 9.436539611873265e-05,
+      "loss": 1.1912,
+      "step": 701000
+    },
+    {
+      "epoch": 20.6,
+      "grad_norm": 2.1459152698516846,
+      "learning_rate": 9.421830412740127e-05,
+      "loss": 1.2343,
+      "step": 701500
+    },
+    {
+      "epoch": 20.61,
+      "grad_norm": 2.039071798324585,
+      "learning_rate": 9.407121213606989e-05,
+      "loss": 1.228,
+      "step": 702000
+    },
+    {
+      "epoch": 20.63,
+      "grad_norm": 3.1236155033111572,
+      "learning_rate": 9.392412014473851e-05,
+      "loss": 1.2362,
+      "step": 702500
+    },
+    {
+      "epoch": 20.64,
+      "grad_norm": 6.471434593200684,
+      "learning_rate": 9.377702815340713e-05,
+      "loss": 1.2159,
+      "step": 703000
+    },
+    {
+      "epoch": 20.66,
+      "grad_norm": 3.2641398906707764,
+      "learning_rate": 9.362993616207576e-05,
+      "loss": 1.1702,
+      "step": 703500
+    },
+    {
+      "epoch": 20.67,
+      "grad_norm": 2.0481953620910645,
+      "learning_rate": 9.348284417074437e-05,
+      "loss": 1.218,
+      "step": 704000
+    },
+    {
+      "epoch": 20.68,
+      "grad_norm": 1.6071834564208984,
+      "learning_rate": 9.3335752179413e-05,
+      "loss": 1.2422,
+      "step": 704500
+    },
+    {
+      "epoch": 20.7,
+      "grad_norm": 1.8020005226135254,
+      "learning_rate": 9.318866018808163e-05,
+      "loss": 1.2268,
+      "step": 705000
+    },
+    {
+      "epoch": 20.71,
+      "grad_norm": 4.970008373260498,
+      "learning_rate": 9.304156819675025e-05,
+      "loss": 1.2093,
+      "step": 705500
+    },
+    {
+      "epoch": 20.73,
+      "grad_norm": 2.152195453643799,
+      "learning_rate": 9.289447620541886e-05,
+      "loss": 1.2361,
+      "step": 706000
+    },
+    {
+      "epoch": 20.74,
+      "grad_norm": 5.956038475036621,
+      "learning_rate": 9.274738421408748e-05,
+      "loss": 1.1908,
+      "step": 706500
+    },
+    {
+      "epoch": 20.76,
+      "grad_norm": 2.429544687271118,
+      "learning_rate": 9.260029222275611e-05,
+      "loss": 1.1805,
+      "step": 707000
+    },
+    {
+      "epoch": 20.77,
+      "grad_norm": 2.6851959228515625,
+      "learning_rate": 9.245320023142473e-05,
+      "loss": 1.204,
+      "step": 707500
+    },
+    {
+      "epoch": 20.79,
+      "grad_norm": 2.6581480503082275,
+      "learning_rate": 9.230610824009334e-05,
+      "loss": 1.2357,
+      "step": 708000
+    },
+    {
+      "epoch": 20.8,
+      "grad_norm": 2.938990354537964,
+      "learning_rate": 9.215901624876196e-05,
+      "loss": 1.2326,
+      "step": 708500
+    },
+    {
+      "epoch": 20.82,
+      "grad_norm": 1.9101125001907349,
+      "learning_rate": 9.20119242574306e-05,
+      "loss": 1.2013,
+      "step": 709000
+    },
+    {
+      "epoch": 20.83,
+      "grad_norm": 3.4203646183013916,
+      "learning_rate": 9.186483226609922e-05,
+      "loss": 1.2021,
+      "step": 709500
+    },
+    {
+      "epoch": 20.85,
+      "grad_norm": 2.2764744758605957,
+      "learning_rate": 9.171774027476783e-05,
+      "loss": 1.223,
+      "step": 710000
+    },
+    {
+      "epoch": 20.86,
+      "grad_norm": 1.5339056253433228,
+      "learning_rate": 9.157064828343645e-05,
+      "loss": 1.2001,
+      "step": 710500
+    },
+    {
+      "epoch": 20.88,
+      "grad_norm": 1.3566075563430786,
+      "learning_rate": 9.142355629210508e-05,
+      "loss": 1.2324,
+      "step": 711000
+    },
+    {
+      "epoch": 20.89,
+      "grad_norm": 1.5568170547485352,
+      "learning_rate": 9.12764643007737e-05,
+      "loss": 1.2065,
+      "step": 711500
+    },
+    {
+      "epoch": 20.9,
+      "grad_norm": 26.298019409179688,
+      "learning_rate": 9.112937230944231e-05,
+      "loss": 1.2283,
+      "step": 712000
+    },
+    {
+      "epoch": 20.92,
+      "grad_norm": 1.2808449268341064,
+      "learning_rate": 9.098228031811093e-05,
+      "loss": 1.2531,
+      "step": 712500
+    },
+    {
+      "epoch": 20.93,
+      "grad_norm": 1.9005004167556763,
+      "learning_rate": 9.083518832677957e-05,
+      "loss": 1.2108,
+      "step": 713000
+    },
+    {
+      "epoch": 20.95,
+      "grad_norm": 6.3783721923828125,
+      "learning_rate": 9.068809633544819e-05,
+      "loss": 1.2015,
+      "step": 713500
+    },
+    {
+      "epoch": 20.96,
+      "grad_norm": 2.5035910606384277,
+      "learning_rate": 9.054100434411681e-05,
+      "loss": 1.2342,
+      "step": 714000
+    },
+    {
+      "epoch": 20.98,
+      "grad_norm": 9.022185325622559,
+      "learning_rate": 9.039391235278542e-05,
+      "loss": 1.2176,
+      "step": 714500
+    },
+    {
+      "epoch": 20.99,
+      "grad_norm": 1.442610502243042,
+      "learning_rate": 9.024682036145405e-05,
+      "loss": 1.2406,
+      "step": 715000
+    },
+    {
+      "epoch": 21.01,
+      "grad_norm": 2.419548273086548,
+      "learning_rate": 9.009972837012267e-05,
+      "loss": 1.1987,
+      "step": 715500
+    },
+    {
+      "epoch": 21.02,
+      "grad_norm": 9.644920349121094,
+      "learning_rate": 8.995263637879129e-05,
+      "loss": 1.141,
+      "step": 716000
+    },
+    {
+      "epoch": 21.04,
+      "grad_norm": 2.2750537395477295,
+      "learning_rate": 8.98055443874599e-05,
+      "loss": 1.1621,
+      "step": 716500
+    },
+    {
+      "epoch": 21.05,
+      "grad_norm": 12.630777359008789,
+      "learning_rate": 8.965845239612853e-05,
+      "loss": 1.1495,
+      "step": 717000
+    },
+    {
+      "epoch": 21.07,
+      "grad_norm": 1.7819632291793823,
+      "learning_rate": 8.951136040479716e-05,
+      "loss": 1.1687,
+      "step": 717500
+    },
+    {
+      "epoch": 21.08,
+      "grad_norm": 2.4531936645507812,
+      "learning_rate": 8.936426841346578e-05,
+      "loss": 1.1722,
+      "step": 718000
+    },
+    {
+      "epoch": 21.1,
+      "grad_norm": 4.26564884185791,
+      "learning_rate": 8.921717642213438e-05,
+      "loss": 1.1804,
+      "step": 718500
+    },
+    {
+      "epoch": 21.11,
+      "grad_norm": 3.8713653087615967,
+      "learning_rate": 8.907008443080302e-05,
+      "loss": 1.1954,
+      "step": 719000
+    },
+    {
+      "epoch": 21.12,
+      "grad_norm": 2.3605353832244873,
+      "learning_rate": 8.892299243947164e-05,
+      "loss": 1.1688,
+      "step": 719500
+    },
+    {
+      "epoch": 21.14,
+      "grad_norm": 2.322021007537842,
+      "learning_rate": 8.877590044814026e-05,
+      "loss": 1.1403,
+      "step": 720000
+    },
+    {
+      "epoch": 21.15,
+      "grad_norm": 2.5785558223724365,
+      "learning_rate": 8.862880845680887e-05,
+      "loss": 1.1812,
+      "step": 720500
+    },
+    {
+      "epoch": 21.17,
+      "grad_norm": 1.8795828819274902,
+      "learning_rate": 8.84817164654775e-05,
+      "loss": 1.1756,
+      "step": 721000
+    },
+    {
+      "epoch": 21.18,
+      "grad_norm": 1.7690235376358032,
+      "learning_rate": 8.833462447414613e-05,
+      "loss": 1.1592,
+      "step": 721500
+    },
+    {
+      "epoch": 21.2,
+      "grad_norm": 2.3973443508148193,
+      "learning_rate": 8.818753248281475e-05,
+      "loss": 1.1586,
+      "step": 722000
+    },
+    {
+      "epoch": 21.21,
+      "grad_norm": 11.784666061401367,
+      "learning_rate": 8.804044049148335e-05,
+      "loss": 1.1522,
+      "step": 722500
+    },
+    {
+      "epoch": 21.23,
+      "grad_norm": 1.8928039073944092,
+      "learning_rate": 8.789334850015199e-05,
+      "loss": 1.1739,
+      "step": 723000
+    },
+    {
+      "epoch": 21.24,
+      "grad_norm": 28.14616584777832,
+      "learning_rate": 8.774625650882061e-05,
+      "loss": 1.1844,
+      "step": 723500
+    },
+    {
+      "epoch": 21.26,
+      "grad_norm": 22.186017990112305,
+      "learning_rate": 8.759916451748923e-05,
+      "loss": 1.1501,
+      "step": 724000
+    },
+    {
+      "epoch": 21.27,
+      "grad_norm": 3.089933395385742,
+      "learning_rate": 8.745207252615785e-05,
+      "loss": 1.1687,
+      "step": 724500
+    },
+    {
+      "epoch": 21.29,
+      "grad_norm": 3.315870523452759,
+      "learning_rate": 8.730498053482647e-05,
+      "loss": 1.1695,
+      "step": 725000
+    },
+    {
+      "epoch": 21.3,
+      "grad_norm": 2.118852138519287,
+      "learning_rate": 8.71578885434951e-05,
+      "loss": 1.1831,
+      "step": 725500
+    },
+    {
+      "epoch": 21.32,
+      "grad_norm": 2.3328022956848145,
+      "learning_rate": 8.701079655216372e-05,
+      "loss": 1.1972,
+      "step": 726000
+    },
+    {
+      "epoch": 21.33,
+      "grad_norm": 2.6026008129119873,
+      "learning_rate": 8.686370456083234e-05,
+      "loss": 1.177,
+      "step": 726500
+    },
+    {
+      "epoch": 21.34,
+      "grad_norm": 1.873986840248108,
+      "learning_rate": 8.671661256950097e-05,
+      "loss": 1.1856,
+      "step": 727000
+    },
+    {
+      "epoch": 21.36,
+      "grad_norm": 2.4924209117889404,
+      "learning_rate": 8.656952057816958e-05,
+      "loss": 1.1587,
+      "step": 727500
+    },
+    {
+      "epoch": 21.37,
+      "grad_norm": 3.228402614593506,
+      "learning_rate": 8.64224285868382e-05,
+      "loss": 1.1654,
+      "step": 728000
+    },
+    {
+      "epoch": 21.39,
+      "grad_norm": 2.2008731365203857,
+      "learning_rate": 8.627533659550682e-05,
+      "loss": 1.1953,
+      "step": 728500
+    },
+    {
+      "epoch": 21.4,
+      "grad_norm": 16.22712516784668,
+      "learning_rate": 8.612824460417546e-05,
+      "loss": 1.1871,
+      "step": 729000
+    },
+    {
+      "epoch": 21.42,
+      "grad_norm": 4.7584919929504395,
+      "learning_rate": 8.598115261284406e-05,
+      "loss": 1.1865,
+      "step": 729500
+    },
+    {
+      "epoch": 21.43,
+      "grad_norm": 3.2225780487060547,
+      "learning_rate": 8.583406062151268e-05,
+      "loss": 1.1776,
+      "step": 730000
+    },
+    {
+      "epoch": 21.45,
+      "grad_norm": 5.469396591186523,
+      "learning_rate": 8.56869686301813e-05,
+      "loss": 1.1897,
+      "step": 730500
+    },
+    {
+      "epoch": 21.46,
+      "grad_norm": 1.8942010402679443,
+      "learning_rate": 8.553987663884994e-05,
+      "loss": 1.1484,
+      "step": 731000
+    },
+    {
+      "epoch": 21.48,
+      "grad_norm": 2.122103452682495,
+      "learning_rate": 8.539278464751855e-05,
+      "loss": 1.1707,
+      "step": 731500
+    },
+    {
+      "epoch": 21.49,
+      "grad_norm": 3.3333709239959717,
+      "learning_rate": 8.524569265618717e-05,
+      "loss": 1.1471,
+      "step": 732000
+    },
+    {
+      "epoch": 21.51,
+      "grad_norm": 3.002875566482544,
+      "learning_rate": 8.509860066485579e-05,
+      "loss": 1.1785,
+      "step": 732500
+    },
+    {
+      "epoch": 21.52,
+      "grad_norm": 3.3468096256256104,
+      "learning_rate": 8.495150867352442e-05,
+      "loss": 1.1705,
+      "step": 733000
+    },
+    {
+      "epoch": 21.54,
+      "grad_norm": 20.700353622436523,
+      "learning_rate": 8.480441668219303e-05,
+      "loss": 1.1792,
+      "step": 733500
+    },
+    {
+      "epoch": 21.55,
+      "grad_norm": 5.712481498718262,
+      "learning_rate": 8.465732469086165e-05,
+      "loss": 1.1633,
+      "step": 734000
+    },
+    {
+      "epoch": 21.57,
+      "grad_norm": 2.2254440784454346,
+      "learning_rate": 8.451023269953027e-05,
+      "loss": 1.1809,
+      "step": 734500
+    },
+    {
+      "epoch": 21.58,
+      "grad_norm": 5.889398574829102,
+      "learning_rate": 8.436314070819891e-05,
+      "loss": 1.1732,
+      "step": 735000
+    },
+    {
+      "epoch": 21.59,
+      "grad_norm": 6.2943572998046875,
+      "learning_rate": 8.421604871686753e-05,
+      "loss": 1.1896,
+      "step": 735500
+    },
+    {
+      "epoch": 21.61,
+      "grad_norm": 2.1915676593780518,
+      "learning_rate": 8.406895672553614e-05,
+      "loss": 1.1884,
+      "step": 736000
+    },
+    {
+      "epoch": 21.62,
+      "grad_norm": 2.950507164001465,
+      "learning_rate": 8.392186473420476e-05,
+      "loss": 1.1825,
+      "step": 736500
+    },
+    {
+      "epoch": 21.64,
+      "grad_norm": 2.338834047317505,
+      "learning_rate": 8.37747727428734e-05,
+      "loss": 1.1787,
+      "step": 737000
+    },
+    {
+      "epoch": 21.65,
+      "grad_norm": 2.806655168533325,
+      "learning_rate": 8.362768075154201e-05,
+      "loss": 1.1988,
+      "step": 737500
+    },
+    {
+      "epoch": 21.67,
+      "grad_norm": 2.509188652038574,
+      "learning_rate": 8.348058876021062e-05,
+      "loss": 1.1678,
+      "step": 738000
+    },
+    {
+      "epoch": 21.68,
+      "grad_norm": 2.3886609077453613,
+      "learning_rate": 8.333349676887924e-05,
+      "loss": 1.2138,
+      "step": 738500
+    },
+    {
+      "epoch": 21.7,
+      "grad_norm": 8.767037391662598,
+      "learning_rate": 8.318640477754788e-05,
+      "loss": 1.1682,
+      "step": 739000
+    },
+    {
+      "epoch": 21.71,
+      "grad_norm": 2.7434234619140625,
+      "learning_rate": 8.30393127862165e-05,
+      "loss": 1.1675,
+      "step": 739500
+    },
+    {
+      "epoch": 21.73,
+      "grad_norm": 3.837946891784668,
+      "learning_rate": 8.28922207948851e-05,
+      "loss": 1.1746,
+      "step": 740000
+    },
+    {
+      "epoch": 21.74,
+      "grad_norm": 2.2261147499084473,
+      "learning_rate": 8.274512880355373e-05,
+      "loss": 1.1805,
+      "step": 740500
+    },
+    {
+      "epoch": 21.76,
+      "grad_norm": 3.7343368530273438,
+      "learning_rate": 8.259803681222236e-05,
+      "loss": 1.1829,
+      "step": 741000
+    },
+    {
+      "epoch": 21.77,
+      "grad_norm": 2.696150779724121,
+      "learning_rate": 8.245094482089098e-05,
+      "loss": 1.1791,
+      "step": 741500
+    },
+    {
+      "epoch": 21.79,
+      "grad_norm": 3.374664783477783,
+      "learning_rate": 8.230385282955959e-05,
+      "loss": 1.186,
+      "step": 742000
+    },
+    {
+      "epoch": 21.8,
+      "grad_norm": 2.0599725246429443,
+      "learning_rate": 8.215676083822821e-05,
+      "loss": 1.1704,
+      "step": 742500
+    },
+    {
+      "epoch": 21.81,
+      "grad_norm": 2.960590362548828,
+      "learning_rate": 8.200966884689685e-05,
+      "loss": 1.2078,
+      "step": 743000
+    },
+    {
+      "epoch": 21.83,
+      "grad_norm": 3.514355421066284,
+      "learning_rate": 8.186257685556547e-05,
+      "loss": 1.1853,
+      "step": 743500
+    },
+    {
+      "epoch": 21.84,
+      "grad_norm": 3.7044436931610107,
+      "learning_rate": 8.171548486423408e-05,
+      "loss": 1.1775,
+      "step": 744000
+    },
+    {
+      "epoch": 21.86,
+      "grad_norm": 2.7482707500457764,
+      "learning_rate": 8.15683928729027e-05,
+      "loss": 1.1927,
+      "step": 744500
+    },
+    {
+      "epoch": 21.87,
+      "grad_norm": 3.462541103363037,
+      "learning_rate": 8.142130088157133e-05,
+      "loss": 1.155,
+      "step": 745000
+    },
+    {
+      "epoch": 21.89,
+      "grad_norm": 1.3520176410675049,
+      "learning_rate": 8.127420889023995e-05,
+      "loss": 1.1887,
+      "step": 745500
+    },
+    {
+      "epoch": 21.9,
+      "grad_norm": 3.2204971313476562,
+      "learning_rate": 8.112711689890857e-05,
+      "loss": 1.17,
+      "step": 746000
+    },
+    {
+      "epoch": 21.92,
+      "grad_norm": 2.8493075370788574,
+      "learning_rate": 8.098002490757718e-05,
+      "loss": 1.1804,
+      "step": 746500
+    },
+    {
+      "epoch": 21.93,
+      "grad_norm": 28.339750289916992,
+      "learning_rate": 8.083293291624582e-05,
+      "loss": 1.1832,
+      "step": 747000
+    },
+    {
+      "epoch": 21.95,
+      "grad_norm": 3.3736674785614014,
+      "learning_rate": 8.068584092491444e-05,
+      "loss": 1.1903,
+      "step": 747500
+    },
+    {
+      "epoch": 21.96,
+      "grad_norm": 2.166250228881836,
+      "learning_rate": 8.053874893358306e-05,
+      "loss": 1.1712,
+      "step": 748000
+    },
+    {
+      "epoch": 21.98,
+      "grad_norm": 3.713568925857544,
+      "learning_rate": 8.039165694225167e-05,
+      "loss": 1.1896,
+      "step": 748500
+    },
+    {
+      "epoch": 21.99,
+      "grad_norm": 3.077561140060425,
+      "learning_rate": 8.02445649509203e-05,
+      "loss": 1.1999,
+      "step": 749000
+    },
+    {
+      "epoch": 22.01,
+      "grad_norm": 1.1793521642684937,
+      "learning_rate": 8.009747295958892e-05,
+      "loss": 1.166,
+      "step": 749500
+    },
+    {
+      "epoch": 22.02,
+      "grad_norm": 1.4397152662277222,
+      "learning_rate": 7.995038096825754e-05,
+      "loss": 1.1327,
+      "step": 750000
+    },
+    {
+      "epoch": 22.03,
+      "grad_norm": 6.889225959777832,
+      "learning_rate": 7.980328897692615e-05,
+      "loss": 1.1533,
+      "step": 750500
+    },
+    {
+      "epoch": 22.05,
+      "grad_norm": 1.7650978565216064,
+      "learning_rate": 7.965619698559479e-05,
+      "loss": 1.1301,
+      "step": 751000
+    },
+    {
+      "epoch": 22.06,
+      "grad_norm": 1.922248125076294,
+      "learning_rate": 7.95091049942634e-05,
+      "loss": 1.1313,
+      "step": 751500
+    },
+    {
+      "epoch": 22.08,
+      "grad_norm": 9.232244491577148,
+      "learning_rate": 7.936201300293203e-05,
+      "loss": 1.1601,
+      "step": 752000
+    },
+    {
+      "epoch": 22.09,
+      "grad_norm": 2.162410259246826,
+      "learning_rate": 7.921492101160063e-05,
+      "loss": 1.1394,
+      "step": 752500
+    },
+    {
+      "epoch": 22.11,
+      "grad_norm": 2.0230438709259033,
+      "learning_rate": 7.906782902026927e-05,
+      "loss": 1.131,
+      "step": 753000
+    },
+    {
+      "epoch": 22.12,
+      "grad_norm": 3.8333919048309326,
+      "learning_rate": 7.892073702893789e-05,
+      "loss": 1.1264,
+      "step": 753500
+    },
+    {
+      "epoch": 22.14,
+      "grad_norm": 1.487121343612671,
+      "learning_rate": 7.877364503760651e-05,
+      "loss": 1.1127,
+      "step": 754000
+    },
+    {
+      "epoch": 22.15,
+      "grad_norm": 2.5997982025146484,
+      "learning_rate": 7.862655304627512e-05,
+      "loss": 1.1072,
+      "step": 754500
+    },
+    {
+      "epoch": 22.17,
+      "grad_norm": 2.5122156143188477,
+      "learning_rate": 7.847946105494375e-05,
+      "loss": 1.1091,
+      "step": 755000
+    },
+    {
+      "epoch": 22.18,
+      "grad_norm": 2.6082444190979004,
+      "learning_rate": 7.833236906361238e-05,
+      "loss": 1.142,
+      "step": 755500
+    },
+    {
+      "epoch": 22.2,
+      "grad_norm": 8.16508960723877,
+      "learning_rate": 7.8185277072281e-05,
+      "loss": 1.1247,
+      "step": 756000
+    },
+    {
+      "epoch": 22.21,
+      "grad_norm": 3.061530113220215,
+      "learning_rate": 7.803818508094962e-05,
+      "loss": 1.1526,
+      "step": 756500
+    },
+    {
+      "epoch": 22.23,
+      "grad_norm": 3.9026131629943848,
+      "learning_rate": 7.789109308961824e-05,
+      "loss": 1.152,
+      "step": 757000
+    },
+    {
+      "epoch": 22.24,
+      "grad_norm": 2.3439905643463135,
+      "learning_rate": 7.774400109828686e-05,
+      "loss": 1.1416,
+      "step": 757500
+    },
+    {
+      "epoch": 22.26,
+      "grad_norm": 0.9024741053581238,
+      "learning_rate": 7.759690910695548e-05,
+      "loss": 1.1272,
+      "step": 758000
+    },
+    {
+      "epoch": 22.27,
+      "grad_norm": 3.3680572509765625,
+      "learning_rate": 7.74498171156241e-05,
+      "loss": 1.1445,
+      "step": 758500
+    },
+    {
+      "epoch": 22.28,
+      "grad_norm": 1.695378065109253,
+      "learning_rate": 7.730272512429274e-05,
+      "loss": 1.1373,
+      "step": 759000
+    },
+    {
+      "epoch": 22.3,
+      "grad_norm": 4.116308689117432,
+      "learning_rate": 7.715563313296134e-05,
+      "loss": 1.1453,
+      "step": 759500
+    },
+    {
+      "epoch": 22.31,
+      "grad_norm": 2.014496088027954,
+      "learning_rate": 7.700854114162997e-05,
+      "loss": 1.146,
+      "step": 760000
+    },
+    {
+      "epoch": 22.33,
+      "grad_norm": 1.5276461839675903,
+      "learning_rate": 7.686144915029859e-05,
+      "loss": 1.13,
+      "step": 760500
+    },
+    {
+      "epoch": 22.34,
+      "grad_norm": 2.514599084854126,
+      "learning_rate": 7.671435715896722e-05,
+      "loss": 1.142,
+      "step": 761000
+    },
+    {
+      "epoch": 22.36,
+      "grad_norm": 2.2747581005096436,
+      "learning_rate": 7.656726516763583e-05,
+      "loss": 1.1552,
+      "step": 761500
+    },
+    {
+      "epoch": 22.37,
+      "grad_norm": 2.406036615371704,
+      "learning_rate": 7.642017317630445e-05,
+      "loss": 1.1453,
+      "step": 762000
+    },
+    {
+      "epoch": 22.39,
+      "grad_norm": 2.231992721557617,
+      "learning_rate": 7.627308118497307e-05,
+      "loss": 1.1386,
+      "step": 762500
+    },
+    {
+      "epoch": 22.4,
+      "grad_norm": 3.1717188358306885,
+      "learning_rate": 7.61259891936417e-05,
+      "loss": 1.1549,
+      "step": 763000
+    },
+    {
+      "epoch": 22.42,
+      "grad_norm": 22.27747917175293,
+      "learning_rate": 7.597889720231031e-05,
+      "loss": 1.1142,
+      "step": 763500
+    },
+    {
+      "epoch": 22.43,
+      "grad_norm": 2.7918033599853516,
+      "learning_rate": 7.583180521097893e-05,
+      "loss": 1.1108,
+      "step": 764000
+    },
+    {
+      "epoch": 22.45,
+      "grad_norm": 1.4565542936325073,
+      "learning_rate": 7.568471321964757e-05,
+      "loss": 1.1509,
+      "step": 764500
+    },
+    {
+      "epoch": 22.46,
+      "grad_norm": 3.1026453971862793,
+      "learning_rate": 7.553762122831619e-05,
+      "loss": 1.1513,
+      "step": 765000
+    },
+    {
+      "epoch": 22.48,
+      "grad_norm": 1.6490851640701294,
+      "learning_rate": 7.53905292369848e-05,
+      "loss": 1.1242,
+      "step": 765500
+    },
+    {
+      "epoch": 22.49,
+      "grad_norm": 1.7027279138565063,
+      "learning_rate": 7.524343724565342e-05,
+      "loss": 1.1561,
+      "step": 766000
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 1.2775441408157349,
+      "learning_rate": 7.509634525432205e-05,
+      "loss": 1.1168,
+      "step": 766500
+    },
+    {
+      "epoch": 22.52,
+      "grad_norm": 5.64236307144165,
+      "learning_rate": 7.494925326299067e-05,
+      "loss": 1.1465,
+      "step": 767000
+    },
+    {
+      "epoch": 22.53,
+      "grad_norm": 2.019430637359619,
+      "learning_rate": 7.480216127165928e-05,
+      "loss": 1.1162,
+      "step": 767500
+    },
+    {
+      "epoch": 22.55,
+      "grad_norm": 1.3938747644424438,
+      "learning_rate": 7.465506928032792e-05,
+      "loss": 1.1253,
+      "step": 768000
+    },
+    {
+      "epoch": 22.56,
+      "grad_norm": 2.122945785522461,
+      "learning_rate": 7.450797728899654e-05,
+      "loss": 1.1409,
+      "step": 768500
+    },
+    {
+      "epoch": 22.58,
+      "grad_norm": 1.826390266418457,
+      "learning_rate": 7.436088529766516e-05,
+      "loss": 1.1473,
+      "step": 769000
+    },
+    {
+      "epoch": 22.59,
+      "grad_norm": 1.8003076314926147,
+      "learning_rate": 7.421379330633378e-05,
+      "loss": 1.1827,
+      "step": 769500
+    },
+    {
+      "epoch": 22.61,
+      "grad_norm": 2.6314899921417236,
+      "learning_rate": 7.40667013150024e-05,
+      "loss": 1.1426,
+      "step": 770000
+    },
+    {
+      "epoch": 22.62,
+      "grad_norm": 2.7484140396118164,
+      "learning_rate": 7.391960932367102e-05,
+      "loss": 1.1603,
+      "step": 770500
+    },
+    {
+      "epoch": 22.64,
+      "grad_norm": 1.5445499420166016,
+      "learning_rate": 7.377251733233964e-05,
+      "loss": 1.1609,
+      "step": 771000
+    },
+    {
+      "epoch": 22.65,
+      "grad_norm": 2.3308262825012207,
+      "learning_rate": 7.362542534100826e-05,
+      "loss": 1.1126,
+      "step": 771500
+    },
+    {
+      "epoch": 22.67,
+      "grad_norm": 2.675467014312744,
+      "learning_rate": 7.347833334967689e-05,
+      "loss": 1.1717,
+      "step": 772000
+    },
+    {
+      "epoch": 22.68,
+      "grad_norm": 7.849360466003418,
+      "learning_rate": 7.333124135834551e-05,
+      "loss": 1.1304,
+      "step": 772500
+    },
+    {
+      "epoch": 22.7,
+      "grad_norm": 1.596022129058838,
+      "learning_rate": 7.318414936701413e-05,
+      "loss": 1.1563,
+      "step": 773000
+    },
+    {
+      "epoch": 22.71,
+      "grad_norm": 1.8678432703018188,
+      "learning_rate": 7.303705737568275e-05,
+      "loss": 1.1668,
+      "step": 773500
+    },
+    {
+      "epoch": 22.72,
+      "grad_norm": 1.3757662773132324,
+      "learning_rate": 7.288996538435137e-05,
+      "loss": 1.1484,
+      "step": 774000
+    },
+    {
+      "epoch": 22.74,
+      "grad_norm": 2.819344997406006,
+      "learning_rate": 7.274287339301999e-05,
+      "loss": 1.1213,
+      "step": 774500
+    },
+    {
+      "epoch": 22.75,
+      "grad_norm": 2.370375871658325,
+      "learning_rate": 7.259578140168861e-05,
+      "loss": 1.1654,
+      "step": 775000
+    },
+    {
+      "epoch": 22.77,
+      "grad_norm": 3.6313276290893555,
+      "learning_rate": 7.244868941035723e-05,
+      "loss": 1.13,
+      "step": 775500
+    },
+    {
+      "epoch": 22.78,
+      "grad_norm": 1.737453579902649,
+      "learning_rate": 7.230159741902585e-05,
+      "loss": 1.1283,
+      "step": 776000
+    },
+    {
+      "epoch": 22.8,
+      "grad_norm": 2.530315637588501,
+      "learning_rate": 7.215450542769448e-05,
+      "loss": 1.1413,
+      "step": 776500
+    },
+    {
+      "epoch": 22.81,
+      "grad_norm": 2.049701452255249,
+      "learning_rate": 7.20074134363631e-05,
+      "loss": 1.1863,
+      "step": 777000
+    },
+    {
+      "epoch": 22.83,
+      "grad_norm": 2.112698554992676,
+      "learning_rate": 7.186032144503172e-05,
+      "loss": 1.1217,
+      "step": 777500
+    },
+    {
+      "epoch": 22.84,
+      "grad_norm": 1.982596755027771,
+      "learning_rate": 7.171322945370034e-05,
+      "loss": 1.151,
+      "step": 778000
+    },
+    {
+      "epoch": 22.86,
+      "grad_norm": 2.8483023643493652,
+      "learning_rate": 7.156613746236896e-05,
+      "loss": 1.1371,
+      "step": 778500
+    },
+    {
+      "epoch": 22.87,
+      "grad_norm": 2.3431105613708496,
+      "learning_rate": 7.141904547103758e-05,
+      "loss": 1.1549,
+      "step": 779000
+    },
+    {
+      "epoch": 22.89,
+      "grad_norm": 2.256725549697876,
+      "learning_rate": 7.12719534797062e-05,
+      "loss": 1.1387,
+      "step": 779500
+    },
+    {
+      "epoch": 22.9,
+      "grad_norm": 3.2769088745117188,
+      "learning_rate": 7.112486148837482e-05,
+      "loss": 1.1365,
+      "step": 780000
+    },
+    {
+      "epoch": 22.92,
+      "grad_norm": 7.680245399475098,
+      "learning_rate": 7.097776949704344e-05,
+      "loss": 1.1524,
+      "step": 780500
+    },
+    {
+      "epoch": 22.93,
+      "grad_norm": 2.0272703170776367,
+      "learning_rate": 7.083067750571207e-05,
+      "loss": 1.1483,
+      "step": 781000
+    },
+    {
+      "epoch": 22.95,
+      "grad_norm": 3.6463356018066406,
+      "learning_rate": 7.068358551438069e-05,
+      "loss": 1.1188,
+      "step": 781500
+    },
+    {
+      "epoch": 22.96,
+      "grad_norm": 2.499232769012451,
+      "learning_rate": 7.053649352304931e-05,
+      "loss": 1.1434,
+      "step": 782000
+    },
+    {
+      "epoch": 22.97,
+      "grad_norm": 2.1540205478668213,
+      "learning_rate": 7.038940153171793e-05,
+      "loss": 1.1394,
+      "step": 782500
+    },
+    {
+      "epoch": 22.99,
+      "grad_norm": 2.569671154022217,
+      "learning_rate": 7.024230954038655e-05,
+      "loss": 1.1581,
+      "step": 783000
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 5.3423333168029785,
+      "learning_rate": 7.009521754905517e-05,
+      "loss": 1.1392,
+      "step": 783500
+    },
+    {
+      "epoch": 23.02,
+      "grad_norm": 1.8991897106170654,
+      "learning_rate": 6.994812555772379e-05,
+      "loss": 1.0877,
+      "step": 784000
+    },
+    {
+      "epoch": 23.03,
+      "grad_norm": 2.094095230102539,
+      "learning_rate": 6.980103356639241e-05,
+      "loss": 1.1106,
+      "step": 784500
+    },
+    {
+      "epoch": 23.05,
+      "grad_norm": 2.4271163940429688,
+      "learning_rate": 6.965394157506104e-05,
+      "loss": 1.0893,
+      "step": 785000
+    },
+    {
+      "epoch": 23.06,
+      "grad_norm": 3.5016427040100098,
+      "learning_rate": 6.950684958372966e-05,
+      "loss": 1.1119,
+      "step": 785500
+    },
+    {
+      "epoch": 23.08,
+      "grad_norm": 3.388533592224121,
+      "learning_rate": 6.935975759239828e-05,
+      "loss": 1.1135,
+      "step": 786000
+    },
+    {
+      "epoch": 23.09,
+      "grad_norm": 1.610478401184082,
+      "learning_rate": 6.92126656010669e-05,
+      "loss": 1.1064,
+      "step": 786500
+    },
+    {
+      "epoch": 23.11,
+      "grad_norm": 2.6857564449310303,
+      "learning_rate": 6.906557360973552e-05,
+      "loss": 1.105,
+      "step": 787000
+    },
+    {
+      "epoch": 23.12,
+      "grad_norm": 3.785341739654541,
+      "learning_rate": 6.891848161840414e-05,
+      "loss": 1.1109,
+      "step": 787500
+    },
+    {
+      "epoch": 23.14,
+      "grad_norm": 1.6604160070419312,
+      "learning_rate": 6.877138962707276e-05,
+      "loss": 1.1124,
+      "step": 788000
+    },
+    {
+      "epoch": 23.15,
+      "grad_norm": 1.9755994081497192,
+      "learning_rate": 6.862429763574138e-05,
+      "loss": 1.1034,
+      "step": 788500
+    },
+    {
+      "epoch": 23.17,
+      "grad_norm": 2.989527940750122,
+      "learning_rate": 6.847720564441e-05,
+      "loss": 1.1055,
+      "step": 789000
+    },
+    {
+      "epoch": 23.18,
+      "grad_norm": 2.2718911170959473,
+      "learning_rate": 6.833011365307863e-05,
+      "loss": 1.104,
+      "step": 789500
+    },
+    {
+      "epoch": 23.19,
+      "grad_norm": 4.303855895996094,
+      "learning_rate": 6.818302166174725e-05,
+      "loss": 1.1174,
+      "step": 790000
+    },
+    {
+      "epoch": 23.21,
+      "grad_norm": 2.6747548580169678,
+      "learning_rate": 6.803592967041587e-05,
+      "loss": 1.1046,
+      "step": 790500
+    },
+    {
+      "epoch": 23.22,
+      "grad_norm": 2.6597251892089844,
+      "learning_rate": 6.78888376790845e-05,
+      "loss": 1.1246,
+      "step": 791000
+    },
+    {
+      "epoch": 23.24,
+      "grad_norm": 1.6096960306167603,
+      "learning_rate": 6.774174568775311e-05,
+      "loss": 1.1103,
+      "step": 791500
+    },
+    {
+      "epoch": 23.25,
+      "grad_norm": 1.8474167585372925,
+      "learning_rate": 6.759465369642174e-05,
+      "loss": 1.1136,
+      "step": 792000
+    },
+    {
+      "epoch": 23.27,
+      "grad_norm": 6.043608665466309,
+      "learning_rate": 6.744756170509035e-05,
+      "loss": 1.1118,
+      "step": 792500
+    },
+    {
+      "epoch": 23.28,
+      "grad_norm": 10.027447700500488,
+      "learning_rate": 6.730046971375899e-05,
+      "loss": 1.115,
+      "step": 793000
+    },
+    {
+      "epoch": 23.3,
+      "grad_norm": 2.5558454990386963,
+      "learning_rate": 6.71533777224276e-05,
+      "loss": 1.0993,
+      "step": 793500
+    },
+    {
+      "epoch": 23.31,
+      "grad_norm": 2.075737476348877,
+      "learning_rate": 6.700628573109623e-05,
+      "loss": 1.1111,
+      "step": 794000
+    },
+    {
+      "epoch": 23.33,
+      "grad_norm": 2.4479551315307617,
+      "learning_rate": 6.685919373976484e-05,
+      "loss": 1.0848,
+      "step": 794500
+    },
+    {
+      "epoch": 23.34,
+      "grad_norm": 2.503129243850708,
+      "learning_rate": 6.671210174843347e-05,
+      "loss": 1.1315,
+      "step": 795000
+    },
+    {
+      "epoch": 23.36,
+      "grad_norm": 2.2025949954986572,
+      "learning_rate": 6.656500975710208e-05,
+      "loss": 1.1061,
+      "step": 795500
+    },
+    {
+      "epoch": 23.37,
+      "grad_norm": 2.5942320823669434,
+      "learning_rate": 6.641791776577071e-05,
+      "loss": 1.1244,
+      "step": 796000
+    },
+    {
+      "epoch": 23.39,
+      "grad_norm": 2.2347915172576904,
+      "learning_rate": 6.627082577443932e-05,
+      "loss": 1.0868,
+      "step": 796500
+    },
+    {
+      "epoch": 23.4,
+      "grad_norm": 2.3821372985839844,
+      "learning_rate": 6.612373378310796e-05,
+      "loss": 1.1175,
+      "step": 797000
+    },
+    {
+      "epoch": 23.41,
+      "grad_norm": 2.4445548057556152,
+      "learning_rate": 6.597664179177656e-05,
+      "loss": 1.1163,
+      "step": 797500
+    },
+    {
+      "epoch": 23.43,
+      "grad_norm": 4.338953018188477,
+      "learning_rate": 6.58295498004452e-05,
+      "loss": 1.104,
+      "step": 798000
+    },
+    {
+      "epoch": 23.44,
+      "grad_norm": 1.766503930091858,
+      "learning_rate": 6.56824578091138e-05,
+      "loss": 1.1178,
+      "step": 798500
+    },
+    {
+      "epoch": 23.46,
+      "grad_norm": 1.7907469272613525,
+      "learning_rate": 6.553536581778244e-05,
+      "loss": 1.1183,
+      "step": 799000
+    },
+    {
+      "epoch": 23.47,
+      "grad_norm": 1.722770094871521,
+      "learning_rate": 6.538827382645105e-05,
+      "loss": 1.1009,
+      "step": 799500
+    },
+    {
+      "epoch": 23.49,
+      "grad_norm": 1.7940298318862915,
+      "learning_rate": 6.524118183511968e-05,
+      "loss": 1.0938,
+      "step": 800000
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 1.5882244110107422,
+      "learning_rate": 6.50940898437883e-05,
+      "loss": 1.1024,
+      "step": 800500
+    },
+    {
+      "epoch": 23.52,
+      "grad_norm": 1.7875205278396606,
+      "learning_rate": 6.494699785245692e-05,
+      "loss": 1.092,
+      "step": 801000
+    },
+    {
+      "epoch": 23.53,
+      "grad_norm": 2.0258092880249023,
+      "learning_rate": 6.479990586112555e-05,
+      "loss": 1.1136,
+      "step": 801500
+    },
+    {
+      "epoch": 23.55,
+      "grad_norm": 1.6038285493850708,
+      "learning_rate": 6.465281386979417e-05,
+      "loss": 1.1013,
+      "step": 802000
+    },
+    {
+      "epoch": 23.56,
+      "grad_norm": 2.411367654800415,
+      "learning_rate": 6.450572187846279e-05,
+      "loss": 1.0971,
+      "step": 802500
+    },
+    {
+      "epoch": 23.58,
+      "grad_norm": 18.325363159179688,
+      "learning_rate": 6.435862988713141e-05,
+      "loss": 1.1292,
+      "step": 803000
+    },
+    {
+      "epoch": 23.59,
+      "grad_norm": 2.520692825317383,
+      "learning_rate": 6.421153789580003e-05,
+      "loss": 1.1108,
+      "step": 803500
+    },
+    {
+      "epoch": 23.61,
+      "grad_norm": 2.115523099899292,
+      "learning_rate": 6.406444590446865e-05,
+      "loss": 1.0878,
+      "step": 804000
+    },
+    {
+      "epoch": 23.62,
+      "grad_norm": 2.4668285846710205,
+      "learning_rate": 6.391735391313727e-05,
+      "loss": 1.1089,
+      "step": 804500
+    },
+    {
+      "epoch": 23.64,
+      "grad_norm": 2.2170443534851074,
+      "learning_rate": 6.37702619218059e-05,
+      "loss": 1.1189,
+      "step": 805000
+    },
+    {
+      "epoch": 23.65,
+      "grad_norm": 2.2153406143188477,
+      "learning_rate": 6.362316993047451e-05,
+      "loss": 1.1053,
+      "step": 805500
+    },
+    {
+      "epoch": 23.66,
+      "grad_norm": 2.937535285949707,
+      "learning_rate": 6.347607793914314e-05,
+      "loss": 1.108,
+      "step": 806000
+    },
+    {
+      "epoch": 23.68,
+      "grad_norm": 1.9660471677780151,
+      "learning_rate": 6.332898594781176e-05,
+      "loss": 1.1203,
+      "step": 806500
+    },
+    {
+      "epoch": 23.69,
+      "grad_norm": 3.5854222774505615,
+      "learning_rate": 6.318189395648038e-05,
+      "loss": 1.1082,
+      "step": 807000
+    },
+    {
+      "epoch": 23.71,
+      "grad_norm": 6.527500152587891,
+      "learning_rate": 6.3034801965149e-05,
+      "loss": 1.1146,
+      "step": 807500
+    },
+    {
+      "epoch": 23.72,
+      "grad_norm": 7.003795623779297,
+      "learning_rate": 6.288770997381762e-05,
+      "loss": 1.1137,
+      "step": 808000
+    },
+    {
+      "epoch": 23.74,
+      "grad_norm": 2.606776714324951,
+      "learning_rate": 6.274061798248624e-05,
+      "loss": 1.0925,
+      "step": 808500
+    },
+    {
+      "epoch": 23.75,
+      "grad_norm": 3.0098581314086914,
+      "learning_rate": 6.259352599115486e-05,
+      "loss": 1.1041,
+      "step": 809000
+    },
+    {
+      "epoch": 23.77,
+      "grad_norm": 1.7633291482925415,
+      "learning_rate": 6.244643399982348e-05,
+      "loss": 1.1257,
+      "step": 809500
+    },
+    {
+      "epoch": 23.78,
+      "grad_norm": 2.6187331676483154,
+      "learning_rate": 6.22993420084921e-05,
+      "loss": 1.0956,
+      "step": 810000
+    },
+    {
+      "epoch": 23.8,
+      "grad_norm": 1.919297456741333,
+      "learning_rate": 6.215225001716073e-05,
+      "loss": 1.0954,
+      "step": 810500
+    },
+    {
+      "epoch": 23.81,
+      "grad_norm": 3.422133445739746,
+      "learning_rate": 6.200515802582935e-05,
+      "loss": 1.0997,
+      "step": 811000
+    },
+    {
+      "epoch": 23.83,
+      "grad_norm": 1.8277616500854492,
+      "learning_rate": 6.185806603449797e-05,
+      "loss": 1.1357,
+      "step": 811500
+    },
+    {
+      "epoch": 23.84,
+      "grad_norm": 1.641455054283142,
+      "learning_rate": 6.171097404316659e-05,
+      "loss": 1.1186,
+      "step": 812000
+    },
+    {
+      "epoch": 23.86,
+      "grad_norm": 2.629831075668335,
+      "learning_rate": 6.156388205183521e-05,
+      "loss": 1.0844,
+      "step": 812500
+    },
+    {
+      "epoch": 23.87,
+      "grad_norm": 2.9981462955474854,
+      "learning_rate": 6.141679006050383e-05,
+      "loss": 1.1323,
+      "step": 813000
+    },
+    {
+      "epoch": 23.88,
+      "grad_norm": 7.224218845367432,
+      "learning_rate": 6.126969806917245e-05,
+      "loss": 1.1091,
+      "step": 813500
+    },
+    {
+      "epoch": 23.9,
+      "grad_norm": 1.445144772529602,
+      "learning_rate": 6.112260607784107e-05,
+      "loss": 1.0885,
+      "step": 814000
+    },
+    {
+      "epoch": 23.91,
+      "grad_norm": 6.933565616607666,
+      "learning_rate": 6.0975514086509695e-05,
+      "loss": 1.1264,
+      "step": 814500
+    },
+    {
+      "epoch": 23.93,
+      "grad_norm": 1.5660400390625,
+      "learning_rate": 6.082842209517832e-05,
+      "loss": 1.102,
+      "step": 815000
+    },
+    {
+      "epoch": 23.94,
+      "grad_norm": 3.0154061317443848,
+      "learning_rate": 6.068133010384694e-05,
+      "loss": 1.1043,
+      "step": 815500
+    },
+    {
+      "epoch": 23.96,
+      "grad_norm": 4.825801372528076,
+      "learning_rate": 6.0534238112515565e-05,
+      "loss": 1.1289,
+      "step": 816000
+    },
+    {
+      "epoch": 23.97,
+      "grad_norm": 2.10929012298584,
+      "learning_rate": 6.038714612118418e-05,
+      "loss": 1.112,
+      "step": 816500
+    },
+    {
+      "epoch": 23.99,
+      "grad_norm": 15.725834846496582,
+      "learning_rate": 6.024005412985281e-05,
+      "loss": 1.0857,
+      "step": 817000
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 3.745626926422119,
+      "learning_rate": 6.009296213852142e-05,
+      "loss": 1.0957,
+      "step": 817500
+    },
+    {
+      "epoch": 24.02,
+      "grad_norm": 1.5650004148483276,
+      "learning_rate": 5.994587014719005e-05,
+      "loss": 1.0553,
+      "step": 818000
+    },
+    {
+      "epoch": 24.03,
+      "grad_norm": 1.279994249343872,
+      "learning_rate": 5.9798778155858664e-05,
+      "loss": 1.0524,
+      "step": 818500
+    },
+    {
+      "epoch": 24.05,
+      "grad_norm": 1.6556503772735596,
+      "learning_rate": 5.965168616452729e-05,
+      "loss": 1.0709,
+      "step": 819000
+    },
+    {
+      "epoch": 24.06,
+      "grad_norm": 1.5529954433441162,
+      "learning_rate": 5.9504594173195906e-05,
+      "loss": 1.0952,
+      "step": 819500
+    },
+    {
+      "epoch": 24.08,
+      "grad_norm": 1.7145531177520752,
+      "learning_rate": 5.9357502181864534e-05,
+      "loss": 1.0526,
+      "step": 820000
+    },
+    {
+      "epoch": 24.09,
+      "grad_norm": 1.5692338943481445,
+      "learning_rate": 5.921041019053315e-05,
+      "loss": 1.0493,
+      "step": 820500
+    },
+    {
+      "epoch": 24.1,
+      "grad_norm": 37.72039031982422,
+      "learning_rate": 5.9063318199201776e-05,
+      "loss": 1.068,
+      "step": 821000
+    },
+    {
+      "epoch": 24.12,
+      "grad_norm": 2.0122039318084717,
+      "learning_rate": 5.891622620787039e-05,
+      "loss": 1.0775,
+      "step": 821500
+    },
+    {
+      "epoch": 24.13,
+      "grad_norm": 5.577430248260498,
+      "learning_rate": 5.876913421653902e-05,
+      "loss": 1.1098,
+      "step": 822000
+    },
+    {
+      "epoch": 24.15,
+      "grad_norm": 3.4199016094207764,
+      "learning_rate": 5.8622042225207647e-05,
+      "loss": 1.0893,
+      "step": 822500
+    },
+    {
+      "epoch": 24.16,
+      "grad_norm": 3.654630184173584,
+      "learning_rate": 5.847495023387626e-05,
+      "loss": 1.0719,
+      "step": 823000
+    },
+    {
+      "epoch": 24.18,
+      "grad_norm": 2.0985825061798096,
+      "learning_rate": 5.832785824254489e-05,
+      "loss": 1.0561,
+      "step": 823500
+    },
+    {
+      "epoch": 24.19,
+      "grad_norm": 3.724655866622925,
+      "learning_rate": 5.81807662512135e-05,
+      "loss": 1.047,
+      "step": 824000
+    },
+    {
+      "epoch": 24.21,
+      "grad_norm": 2.0398764610290527,
+      "learning_rate": 5.803367425988213e-05,
+      "loss": 1.0806,
+      "step": 824500
+    },
+    {
+      "epoch": 24.22,
+      "grad_norm": 2.9711148738861084,
+      "learning_rate": 5.7886582268550745e-05,
+      "loss": 1.104,
+      "step": 825000
+    },
+    {
+      "epoch": 24.24,
+      "grad_norm": 1.4870119094848633,
+      "learning_rate": 5.773949027721937e-05,
+      "loss": 1.0812,
+      "step": 825500
+    },
+    {
+      "epoch": 24.25,
+      "grad_norm": 2.184882640838623,
+      "learning_rate": 5.759239828588799e-05,
+      "loss": 1.071,
+      "step": 826000
+    },
+    {
+      "epoch": 24.27,
+      "grad_norm": 2.135782241821289,
+      "learning_rate": 5.7445306294556616e-05,
+      "loss": 1.081,
+      "step": 826500
+    },
+    {
+      "epoch": 24.28,
+      "grad_norm": 1.952862024307251,
+      "learning_rate": 5.729821430322523e-05,
+      "loss": 1.0688,
+      "step": 827000
+    },
+    {
+      "epoch": 24.3,
+      "grad_norm": 3.28511381149292,
+      "learning_rate": 5.715112231189386e-05,
+      "loss": 1.0719,
+      "step": 827500
+    },
+    {
+      "epoch": 24.31,
+      "grad_norm": 3.095506191253662,
+      "learning_rate": 5.700403032056247e-05,
+      "loss": 1.1089,
+      "step": 828000
+    },
+    {
+      "epoch": 24.33,
+      "grad_norm": 1.744498372077942,
+      "learning_rate": 5.68569383292311e-05,
+      "loss": 1.0666,
+      "step": 828500
+    },
+    {
+      "epoch": 24.34,
+      "grad_norm": 1.5370303392410278,
+      "learning_rate": 5.6709846337899714e-05,
+      "loss": 1.0834,
+      "step": 829000
+    },
+    {
+      "epoch": 24.35,
+      "grad_norm": 3.65459942817688,
+      "learning_rate": 5.656275434656834e-05,
+      "loss": 1.0767,
+      "step": 829500
+    },
+    {
+      "epoch": 24.37,
+      "grad_norm": 2.0333411693573,
+      "learning_rate": 5.641566235523696e-05,
+      "loss": 1.0965,
+      "step": 830000
+    },
+    {
+      "epoch": 24.38,
+      "grad_norm": 2.9158904552459717,
+      "learning_rate": 5.6268570363905585e-05,
+      "loss": 1.0726,
+      "step": 830500
+    },
+    {
+      "epoch": 24.4,
+      "grad_norm": 2.0873067378997803,
+      "learning_rate": 5.6121478372574206e-05,
+      "loss": 1.0664,
+      "step": 831000
+    },
+    {
+      "epoch": 24.41,
+      "grad_norm": 2.0477983951568604,
+      "learning_rate": 5.597438638124283e-05,
+      "loss": 1.0588,
+      "step": 831500
+    },
+    {
+      "epoch": 24.43,
+      "grad_norm": 1.9659167528152466,
+      "learning_rate": 5.582729438991145e-05,
+      "loss": 1.0986,
+      "step": 832000
+    },
+    {
+      "epoch": 24.44,
+      "grad_norm": 2.130729913711548,
+      "learning_rate": 5.568020239858007e-05,
+      "loss": 1.0713,
+      "step": 832500
+    },
+    {
+      "epoch": 24.46,
+      "grad_norm": 99.76852416992188,
+      "learning_rate": 5.553311040724869e-05,
+      "loss": 1.0803,
+      "step": 833000
+    },
+    {
+      "epoch": 24.47,
+      "grad_norm": 5.594266414642334,
+      "learning_rate": 5.538601841591731e-05,
+      "loss": 1.08,
+      "step": 833500
+    },
+    {
+      "epoch": 24.49,
+      "grad_norm": 2.3737406730651855,
+      "learning_rate": 5.523892642458593e-05,
+      "loss": 1.0931,
+      "step": 834000
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 1.6415624618530273,
+      "learning_rate": 5.5091834433254554e-05,
+      "loss": 1.0798,
+      "step": 834500
+    },
+    {
+      "epoch": 24.52,
+      "grad_norm": 7.322782516479492,
+      "learning_rate": 5.4944742441923175e-05,
+      "loss": 1.0865,
+      "step": 835000
+    },
+    {
+      "epoch": 24.53,
+      "grad_norm": 2.1509947776794434,
+      "learning_rate": 5.4797650450591796e-05,
+      "loss": 1.0782,
+      "step": 835500
+    },
+    {
+      "epoch": 24.55,
+      "grad_norm": 2.485328197479248,
+      "learning_rate": 5.465055845926042e-05,
+      "loss": 1.0585,
+      "step": 836000
+    },
+    {
+      "epoch": 24.56,
+      "grad_norm": 2.5374958515167236,
+      "learning_rate": 5.450346646792904e-05,
+      "loss": 1.0577,
+      "step": 836500
+    },
+    {
+      "epoch": 24.57,
+      "grad_norm": 2.196706533432007,
+      "learning_rate": 5.435637447659766e-05,
+      "loss": 1.0894,
+      "step": 837000
+    },
+    {
+      "epoch": 24.59,
+      "grad_norm": 2.9057724475860596,
+      "learning_rate": 5.420928248526629e-05,
+      "loss": 1.075,
+      "step": 837500
+    },
+    {
+      "epoch": 24.6,
+      "grad_norm": 2.4338793754577637,
+      "learning_rate": 5.40621904939349e-05,
+      "loss": 1.0829,
+      "step": 838000
+    },
+    {
+      "epoch": 24.62,
+      "grad_norm": 2.225640058517456,
+      "learning_rate": 5.391509850260353e-05,
+      "loss": 1.0798,
+      "step": 838500
+    },
+    {
+      "epoch": 24.63,
+      "grad_norm": 2.3546741008758545,
+      "learning_rate": 5.3768006511272144e-05,
+      "loss": 1.0887,
+      "step": 839000
+    },
+    {
+      "epoch": 24.65,
+      "grad_norm": 1.3033528327941895,
+      "learning_rate": 5.362091451994077e-05,
+      "loss": 1.0809,
+      "step": 839500
+    },
+    {
+      "epoch": 24.66,
+      "grad_norm": 2.5136497020721436,
+      "learning_rate": 5.3473822528609386e-05,
+      "loss": 1.0851,
+      "step": 840000
+    },
+    {
+      "epoch": 24.68,
+      "grad_norm": 2.0871777534484863,
+      "learning_rate": 5.3326730537278014e-05,
+      "loss": 1.0744,
+      "step": 840500
+    },
+    {
+      "epoch": 24.69,
+      "grad_norm": 5.352667808532715,
+      "learning_rate": 5.317963854594663e-05,
+      "loss": 1.102,
+      "step": 841000
+    },
+    {
+      "epoch": 24.71,
+      "grad_norm": 1.8060741424560547,
+      "learning_rate": 5.3032546554615256e-05,
+      "loss": 1.0541,
+      "step": 841500
+    },
+    {
+      "epoch": 24.72,
+      "grad_norm": 2.5160653591156006,
+      "learning_rate": 5.288545456328387e-05,
+      "loss": 1.077,
+      "step": 842000
+    },
+    {
+      "epoch": 24.74,
+      "grad_norm": 4.854517936706543,
+      "learning_rate": 5.27383625719525e-05,
+      "loss": 1.084,
+      "step": 842500
+    },
+    {
+      "epoch": 24.75,
+      "grad_norm": 7.554323196411133,
+      "learning_rate": 5.259127058062111e-05,
+      "loss": 1.0974,
+      "step": 843000
+    },
+    {
+      "epoch": 24.77,
+      "grad_norm": 1.8624228239059448,
+      "learning_rate": 5.244417858928974e-05,
+      "loss": 1.0793,
+      "step": 843500
+    },
+    {
+      "epoch": 24.78,
+      "grad_norm": 1.6777387857437134,
+      "learning_rate": 5.2297086597958355e-05,
+      "loss": 1.0746,
+      "step": 844000
+    },
+    {
+      "epoch": 24.79,
+      "grad_norm": 1.9839929342269897,
+      "learning_rate": 5.214999460662698e-05,
+      "loss": 1.0813,
+      "step": 844500
+    },
+    {
+      "epoch": 24.81,
+      "grad_norm": 1.5621155500411987,
+      "learning_rate": 5.20029026152956e-05,
+      "loss": 1.0936,
+      "step": 845000
+    },
+    {
+      "epoch": 24.82,
+      "grad_norm": 2.5677313804626465,
+      "learning_rate": 5.1855810623964225e-05,
+      "loss": 1.0845,
+      "step": 845500
+    },
+    {
+      "epoch": 24.84,
+      "grad_norm": 1.3594130277633667,
+      "learning_rate": 5.170871863263284e-05,
+      "loss": 1.0563,
+      "step": 846000
+    },
+    {
+      "epoch": 24.85,
+      "grad_norm": 1.3905824422836304,
+      "learning_rate": 5.156162664130147e-05,
+      "loss": 1.0661,
+      "step": 846500
+    },
+    {
+      "epoch": 24.87,
+      "grad_norm": 1.6563278436660767,
+      "learning_rate": 5.141453464997008e-05,
+      "loss": 1.079,
+      "step": 847000
+    },
+    {
+      "epoch": 24.88,
+      "grad_norm": 2.1937708854675293,
+      "learning_rate": 5.126744265863871e-05,
+      "loss": 1.0714,
+      "step": 847500
+    },
+    {
+      "epoch": 24.9,
+      "grad_norm": 1.874009609222412,
+      "learning_rate": 5.112035066730733e-05,
+      "loss": 1.064,
+      "step": 848000
+    },
+    {
+      "epoch": 24.91,
+      "grad_norm": 1.4228936433792114,
+      "learning_rate": 5.097325867597595e-05,
+      "loss": 1.0954,
+      "step": 848500
+    },
+    {
+      "epoch": 24.93,
+      "grad_norm": 1.7444462776184082,
+      "learning_rate": 5.082616668464457e-05,
+      "loss": 1.0921,
+      "step": 849000
+    },
+    {
+      "epoch": 24.94,
+      "grad_norm": 186.56033325195312,
+      "learning_rate": 5.0679074693313194e-05,
+      "loss": 1.0757,
+      "step": 849500
+    },
+    {
+      "epoch": 24.96,
+      "grad_norm": 4.014658451080322,
+      "learning_rate": 5.0531982701981815e-05,
+      "loss": 1.0864,
+      "step": 850000
+    },
+    {
+      "epoch": 24.97,
+      "grad_norm": 1.8258497714996338,
+      "learning_rate": 5.0384890710650436e-05,
+      "loss": 1.0691,
+      "step": 850500
+    },
+    {
+      "epoch": 24.99,
+      "grad_norm": 2.100780487060547,
+      "learning_rate": 5.023779871931906e-05,
+      "loss": 1.0722,
+      "step": 851000
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 2.250474452972412,
+      "learning_rate": 5.009070672798768e-05,
+      "loss": 1.0853,
+      "step": 851500
+    },
+    {
+      "epoch": 25.02,
+      "grad_norm": 2.6746740341186523,
+      "learning_rate": 4.99436147366563e-05,
+      "loss": 1.0277,
+      "step": 852000
+    },
+    {
+      "epoch": 25.03,
+      "grad_norm": 2.413872480392456,
+      "learning_rate": 4.979652274532492e-05,
+      "loss": 1.0804,
+      "step": 852500
+    },
+    {
+      "epoch": 25.04,
+      "grad_norm": 3.563396453857422,
+      "learning_rate": 4.964943075399354e-05,
+      "loss": 1.0638,
+      "step": 853000
+    },
+    {
+      "epoch": 25.06,
+      "grad_norm": 1.7918055057525635,
+      "learning_rate": 4.950233876266217e-05,
+      "loss": 1.0541,
+      "step": 853500
+    },
+    {
+      "epoch": 25.07,
+      "grad_norm": 2.508848190307617,
+      "learning_rate": 4.9355246771330784e-05,
+      "loss": 1.0083,
+      "step": 854000
+    },
+    {
+      "epoch": 25.09,
+      "grad_norm": 9.781774520874023,
+      "learning_rate": 4.920815477999941e-05,
+      "loss": 1.0236,
+      "step": 854500
+    },
+    {
+      "epoch": 25.1,
+      "grad_norm": 2.6363613605499268,
+      "learning_rate": 4.9061062788668026e-05,
+      "loss": 1.0356,
+      "step": 855000
+    },
+    {
+      "epoch": 25.12,
+      "grad_norm": 5.132301330566406,
+      "learning_rate": 4.8913970797336654e-05,
+      "loss": 1.0509,
+      "step": 855500
+    },
+    {
+      "epoch": 25.13,
+      "grad_norm": 2.983057737350464,
+      "learning_rate": 4.876687880600527e-05,
+      "loss": 1.0486,
+      "step": 856000
+    },
+    {
+      "epoch": 25.15,
+      "grad_norm": 3.175165891647339,
+      "learning_rate": 4.8619786814673897e-05,
+      "loss": 1.0521,
+      "step": 856500
+    },
+    {
+      "epoch": 25.16,
+      "grad_norm": 5.045122146606445,
+      "learning_rate": 4.847269482334251e-05,
+      "loss": 1.0494,
+      "step": 857000
+    },
+    {
+      "epoch": 25.18,
+      "grad_norm": 2.601104259490967,
+      "learning_rate": 4.832560283201114e-05,
+      "loss": 1.0301,
+      "step": 857500
+    },
+    {
+      "epoch": 25.19,
+      "grad_norm": 2.546464443206787,
+      "learning_rate": 4.817851084067975e-05,
+      "loss": 1.0237,
+      "step": 858000
+    },
+    {
+      "epoch": 25.21,
+      "grad_norm": 2.9643750190734863,
+      "learning_rate": 4.803141884934838e-05,
+      "loss": 1.0573,
+      "step": 858500
+    },
+    {
+      "epoch": 25.22,
+      "grad_norm": 5.178891658782959,
+      "learning_rate": 4.7884326858016995e-05,
+      "loss": 1.0588,
+      "step": 859000
+    },
+    {
+      "epoch": 25.24,
+      "grad_norm": 2.786492347717285,
+      "learning_rate": 4.773723486668562e-05,
+      "loss": 1.0537,
+      "step": 859500
+    },
+    {
+      "epoch": 25.25,
+      "grad_norm": 3.3391730785369873,
+      "learning_rate": 4.759014287535424e-05,
+      "loss": 1.0452,
+      "step": 860000
+    },
+    {
+      "epoch": 25.26,
+      "grad_norm": 2.8821425437927246,
+      "learning_rate": 4.7443050884022866e-05,
+      "loss": 1.0538,
+      "step": 860500
+    },
+    {
+      "epoch": 25.28,
+      "grad_norm": 8.122342109680176,
+      "learning_rate": 4.729595889269148e-05,
+      "loss": 1.0326,
+      "step": 861000
+    },
+    {
+      "epoch": 25.29,
+      "grad_norm": 5.042423725128174,
+      "learning_rate": 4.714886690136011e-05,
+      "loss": 1.0586,
+      "step": 861500
+    },
+    {
+      "epoch": 25.31,
+      "grad_norm": 3.068471670150757,
+      "learning_rate": 4.700177491002872e-05,
+      "loss": 1.032,
+      "step": 862000
+    },
+    {
+      "epoch": 25.32,
+      "grad_norm": 3.833307981491089,
+      "learning_rate": 4.685468291869735e-05,
+      "loss": 1.0581,
+      "step": 862500
+    },
+    {
+      "epoch": 25.34,
+      "grad_norm": 2.662344455718994,
+      "learning_rate": 4.6707590927365964e-05,
+      "loss": 1.0526,
+      "step": 863000
+    },
+    {
+      "epoch": 25.35,
+      "grad_norm": 3.626770257949829,
+      "learning_rate": 4.656049893603459e-05,
+      "loss": 1.0484,
+      "step": 863500
+    },
+    {
+      "epoch": 25.37,
+      "grad_norm": 2.1823232173919678,
+      "learning_rate": 4.6413406944703213e-05,
+      "loss": 1.0524,
+      "step": 864000
+    },
+    {
+      "epoch": 25.38,
+      "grad_norm": 2.051774740219116,
+      "learning_rate": 4.6266314953371835e-05,
+      "loss": 1.0406,
+      "step": 864500
+    },
+    {
+      "epoch": 25.4,
+      "grad_norm": 9.218499183654785,
+      "learning_rate": 4.6119222962040456e-05,
+      "loss": 1.0605,
+      "step": 865000
+    },
+    {
+      "epoch": 25.41,
+      "grad_norm": 2.4243879318237305,
+      "learning_rate": 4.597213097070908e-05,
+      "loss": 1.033,
+      "step": 865500
+    },
+    {
+      "epoch": 25.43,
+      "grad_norm": 5.0782599449157715,
+      "learning_rate": 4.58250389793777e-05,
+      "loss": 1.059,
+      "step": 866000
+    },
+    {
+      "epoch": 25.44,
+      "grad_norm": 2.4752907752990723,
+      "learning_rate": 4.567794698804632e-05,
+      "loss": 1.0348,
+      "step": 866500
+    },
+    {
+      "epoch": 25.46,
+      "grad_norm": 3.958376407623291,
+      "learning_rate": 4.553085499671494e-05,
+      "loss": 1.0677,
+      "step": 867000
+    },
+    {
+      "epoch": 25.47,
+      "grad_norm": 2.9827880859375,
+      "learning_rate": 4.538376300538356e-05,
+      "loss": 1.0371,
+      "step": 867500
+    },
+    {
+      "epoch": 25.48,
+      "grad_norm": 4.071894645690918,
+      "learning_rate": 4.523667101405218e-05,
+      "loss": 1.0578,
+      "step": 868000
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 34.7899169921875,
+      "learning_rate": 4.5089579022720804e-05,
+      "loss": 1.0903,
+      "step": 868500
+    },
+    {
+      "epoch": 25.51,
+      "grad_norm": 2.133716344833374,
+      "learning_rate": 4.4942487031389425e-05,
+      "loss": 1.0471,
+      "step": 869000
+    },
+    {
+      "epoch": 25.53,
+      "grad_norm": 2.6652045249938965,
+      "learning_rate": 4.4795395040058046e-05,
+      "loss": 1.0358,
+      "step": 869500
+    },
+    {
+      "epoch": 25.54,
+      "grad_norm": 2.98919677734375,
+      "learning_rate": 4.464830304872667e-05,
+      "loss": 1.0621,
+      "step": 870000
+    },
+    {
+      "epoch": 25.56,
+      "grad_norm": 2.7026760578155518,
+      "learning_rate": 4.4501211057395295e-05,
+      "loss": 1.0621,
+      "step": 870500
+    },
+    {
+      "epoch": 25.57,
+      "grad_norm": 3.262974977493286,
+      "learning_rate": 4.435411906606391e-05,
+      "loss": 1.0321,
+      "step": 871000
+    },
+    {
+      "epoch": 25.59,
+      "grad_norm": 1.9586899280548096,
+      "learning_rate": 4.420702707473254e-05,
+      "loss": 1.0347,
+      "step": 871500
+    },
+    {
+      "epoch": 25.6,
+      "grad_norm": 7.560853481292725,
+      "learning_rate": 4.405993508340115e-05,
+      "loss": 1.0221,
+      "step": 872000
+    },
+    {
+      "epoch": 25.62,
+      "grad_norm": 3.056516408920288,
+      "learning_rate": 4.391284309206978e-05,
+      "loss": 1.043,
+      "step": 872500
+    },
+    {
+      "epoch": 25.63,
+      "grad_norm": 2.7999415397644043,
+      "learning_rate": 4.3765751100738394e-05,
+      "loss": 1.0515,
+      "step": 873000
+    },
+    {
+      "epoch": 25.65,
+      "grad_norm": 2.5458314418792725,
+      "learning_rate": 4.361865910940702e-05,
+      "loss": 1.0599,
+      "step": 873500
+    },
+    {
+      "epoch": 25.66,
+      "grad_norm": 3.634589910507202,
+      "learning_rate": 4.3471567118075636e-05,
+      "loss": 1.0428,
+      "step": 874000
+    },
+    {
+      "epoch": 25.68,
+      "grad_norm": 2.8104031085968018,
+      "learning_rate": 4.3324475126744264e-05,
+      "loss": 1.0396,
+      "step": 874500
+    },
+    {
+      "epoch": 25.69,
+      "grad_norm": 2.319333791732788,
+      "learning_rate": 4.317738313541288e-05,
+      "loss": 1.0426,
+      "step": 875000
+    },
+    {
+      "epoch": 25.71,
+      "grad_norm": 8.262571334838867,
+      "learning_rate": 4.3030291144081506e-05,
+      "loss": 1.0616,
+      "step": 875500
+    },
+    {
+      "epoch": 25.72,
+      "grad_norm": 3.307812213897705,
+      "learning_rate": 4.288319915275012e-05,
+      "loss": 1.0706,
+      "step": 876000
+    },
+    {
+      "epoch": 25.73,
+      "grad_norm": 7.4953508377075195,
+      "learning_rate": 4.273610716141875e-05,
+      "loss": 1.0457,
+      "step": 876500
+    },
+    {
+      "epoch": 25.75,
+      "grad_norm": 3.5895800590515137,
+      "learning_rate": 4.258901517008736e-05,
+      "loss": 1.0219,
+      "step": 877000
+    },
+    {
+      "epoch": 25.76,
+      "grad_norm": 3.023827075958252,
+      "learning_rate": 4.244192317875599e-05,
+      "loss": 1.05,
+      "step": 877500
+    },
+    {
+      "epoch": 25.78,
+      "grad_norm": 2.1299614906311035,
+      "learning_rate": 4.2294831187424605e-05,
+      "loss": 1.0505,
+      "step": 878000
+    },
+    {
+      "epoch": 25.79,
+      "grad_norm": 3.042360305786133,
+      "learning_rate": 4.214773919609323e-05,
+      "loss": 1.0481,
+      "step": 878500
+    },
+    {
+      "epoch": 25.81,
+      "grad_norm": 2.3197755813598633,
+      "learning_rate": 4.200064720476185e-05,
+      "loss": 1.0181,
+      "step": 879000
+    },
+    {
+      "epoch": 25.82,
+      "grad_norm": 3.230426788330078,
+      "learning_rate": 4.1853555213430475e-05,
+      "loss": 1.0327,
+      "step": 879500
+    },
+    {
+      "epoch": 25.84,
+      "grad_norm": 7.685334205627441,
+      "learning_rate": 4.1706463222099096e-05,
+      "loss": 1.0334,
+      "step": 880000
+    },
+    {
+      "epoch": 25.85,
+      "grad_norm": 3.2607569694519043,
+      "learning_rate": 4.155937123076772e-05,
+      "loss": 1.0342,
+      "step": 880500
+    },
+    {
+      "epoch": 25.87,
+      "grad_norm": 3.470470428466797,
+      "learning_rate": 4.1412279239436345e-05,
+      "loss": 1.0508,
+      "step": 881000
+    },
+    {
+      "epoch": 25.88,
+      "grad_norm": 16.141782760620117,
+      "learning_rate": 4.126518724810496e-05,
+      "loss": 1.046,
+      "step": 881500
+    },
+    {
+      "epoch": 25.9,
+      "grad_norm": 2.900520086288452,
+      "learning_rate": 4.111809525677359e-05,
+      "loss": 1.0791,
+      "step": 882000
+    },
+    {
+      "epoch": 25.91,
+      "grad_norm": 2.5581626892089844,
+      "learning_rate": 4.09710032654422e-05,
+      "loss": 1.0386,
+      "step": 882500
+    },
+    {
+      "epoch": 25.93,
+      "grad_norm": 2.301318883895874,
+      "learning_rate": 4.082391127411083e-05,
+      "loss": 1.078,
+      "step": 883000
+    },
+    {
+      "epoch": 25.94,
+      "grad_norm": 2.5168445110321045,
+      "learning_rate": 4.0676819282779444e-05,
+      "loss": 1.0524,
+      "step": 883500
+    },
+    {
+      "epoch": 25.95,
+      "grad_norm": 3.2487659454345703,
+      "learning_rate": 4.052972729144807e-05,
+      "loss": 1.0476,
+      "step": 884000
+    },
+    {
+      "epoch": 25.97,
+      "grad_norm": 3.8711748123168945,
+      "learning_rate": 4.0382635300116686e-05,
+      "loss": 1.0349,
+      "step": 884500
+    },
+    {
+      "epoch": 25.98,
+      "grad_norm": 2.8647818565368652,
+      "learning_rate": 4.0235543308785314e-05,
+      "loss": 1.0681,
+      "step": 885000
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 2.307555675506592,
+      "learning_rate": 4.008845131745393e-05,
+      "loss": 1.0299,
+      "step": 885500
+    },
+    {
+      "epoch": 26.01,
+      "grad_norm": 5.377994537353516,
+      "learning_rate": 3.9941359326122557e-05,
+      "loss": 1.0274,
+      "step": 886000
+    },
+    {
+      "epoch": 26.03,
+      "grad_norm": 3.1154091358184814,
+      "learning_rate": 3.979426733479118e-05,
+      "loss": 1.0313,
+      "step": 886500
+    },
+    {
+      "epoch": 26.04,
+      "grad_norm": 1.3295516967773438,
+      "learning_rate": 3.96471753434598e-05,
+      "loss": 0.9962,
+      "step": 887000
+    },
+    {
+      "epoch": 26.06,
+      "grad_norm": 2.4110894203186035,
+      "learning_rate": 3.950008335212842e-05,
+      "loss": 1.0051,
+      "step": 887500
+    },
+    {
+      "epoch": 26.07,
+      "grad_norm": 1.6697067022323608,
+      "learning_rate": 3.935299136079704e-05,
+      "loss": 1.0119,
+      "step": 888000
+    },
+    {
+      "epoch": 26.09,
+      "grad_norm": 2.4017999172210693,
+      "learning_rate": 3.920589936946566e-05,
+      "loss": 1.0253,
+      "step": 888500
+    },
+    {
+      "epoch": 26.1,
+      "grad_norm": 3.1681103706359863,
+      "learning_rate": 3.905880737813428e-05,
+      "loss": 1.0213,
+      "step": 889000
+    },
+    {
+      "epoch": 26.12,
+      "grad_norm": 2.9184956550598145,
+      "learning_rate": 3.8911715386802904e-05,
+      "loss": 1.0298,
+      "step": 889500
+    },
+    {
+      "epoch": 26.13,
+      "grad_norm": 1.6290541887283325,
+      "learning_rate": 3.8764623395471526e-05,
+      "loss": 1.0094,
+      "step": 890000
+    },
+    {
+      "epoch": 26.15,
+      "grad_norm": 2.779973030090332,
+      "learning_rate": 3.861753140414015e-05,
+      "loss": 1.016,
+      "step": 890500
+    },
+    {
+      "epoch": 26.16,
+      "grad_norm": 4.393610000610352,
+      "learning_rate": 3.847043941280877e-05,
+      "loss": 1.0131,
+      "step": 891000
+    },
+    {
+      "epoch": 26.17,
+      "grad_norm": 2.503702402114868,
+      "learning_rate": 3.832334742147739e-05,
+      "loss": 1.0207,
+      "step": 891500
+    },
+    {
+      "epoch": 26.19,
+      "grad_norm": 1.3769996166229248,
+      "learning_rate": 3.817625543014601e-05,
+      "loss": 1.0305,
+      "step": 892000
+    },
+    {
+      "epoch": 26.2,
+      "grad_norm": 9.221471786499023,
+      "learning_rate": 3.802916343881463e-05,
+      "loss": 1.0258,
+      "step": 892500
+    },
+    {
+      "epoch": 26.22,
+      "grad_norm": 1.706425666809082,
+      "learning_rate": 3.788207144748326e-05,
+      "loss": 0.9808,
+      "step": 893000
+    },
+    {
+      "epoch": 26.23,
+      "grad_norm": 15.878872871398926,
+      "learning_rate": 3.7734979456151873e-05,
+      "loss": 1.0225,
+      "step": 893500
+    },
+    {
+      "epoch": 26.25,
+      "grad_norm": 9.141250610351562,
+      "learning_rate": 3.75878874648205e-05,
+      "loss": 1.0234,
+      "step": 894000
+    },
+    {
+      "epoch": 26.26,
+      "grad_norm": 1.698013424873352,
+      "learning_rate": 3.7440795473489116e-05,
+      "loss": 1.0207,
+      "step": 894500
+    },
+    {
+      "epoch": 26.28,
+      "grad_norm": 2.8760321140289307,
+      "learning_rate": 3.729370348215774e-05,
+      "loss": 1.0054,
+      "step": 895000
+    },
+    {
+      "epoch": 26.29,
+      "grad_norm": 10.192665100097656,
+      "learning_rate": 3.714661149082636e-05,
+      "loss": 1.0202,
+      "step": 895500
+    },
+    {
+      "epoch": 26.31,
+      "grad_norm": 1.8373937606811523,
+      "learning_rate": 3.699951949949498e-05,
+      "loss": 1.0399,
+      "step": 896000
+    },
+    {
+      "epoch": 26.32,
+      "grad_norm": 0.8489872217178345,
+      "learning_rate": 3.68524275081636e-05,
+      "loss": 1.0525,
+      "step": 896500
+    },
+    {
+      "epoch": 26.34,
+      "grad_norm": 1.7799832820892334,
+      "learning_rate": 3.670533551683222e-05,
+      "loss": 1.0271,
+      "step": 897000
+    },
+    {
+      "epoch": 26.35,
+      "grad_norm": 1.6424106359481812,
+      "learning_rate": 3.655824352550084e-05,
+      "loss": 1.0093,
+      "step": 897500
+    },
+    {
+      "epoch": 26.37,
+      "grad_norm": 4.4057135581970215,
+      "learning_rate": 3.6411151534169464e-05,
+      "loss": 1.0214,
+      "step": 898000
+    },
+    {
+      "epoch": 26.38,
+      "grad_norm": 1.971706509590149,
+      "learning_rate": 3.6264059542838085e-05,
+      "loss": 1.0316,
+      "step": 898500
+    },
+    {
+      "epoch": 26.39,
+      "grad_norm": 1.934280276298523,
+      "learning_rate": 3.6116967551506706e-05,
+      "loss": 1.0368,
+      "step": 899000
+    },
+    {
+      "epoch": 26.41,
+      "grad_norm": 1.6416791677474976,
+      "learning_rate": 3.596987556017533e-05,
+      "loss": 1.0123,
+      "step": 899500
+    },
+    {
+      "epoch": 26.42,
+      "grad_norm": 1.6787991523742676,
+      "learning_rate": 3.582278356884395e-05,
+      "loss": 1.0026,
+      "step": 900000
+    },
+    {
+      "epoch": 26.44,
+      "grad_norm": 2.6135306358337402,
+      "learning_rate": 3.567569157751257e-05,
+      "loss": 0.9887,
+      "step": 900500
+    },
+    {
+      "epoch": 26.45,
+      "grad_norm": 1.4282991886138916,
+      "learning_rate": 3.552859958618119e-05,
+      "loss": 1.0442,
+      "step": 901000
+    },
+    {
+      "epoch": 26.47,
+      "grad_norm": 1.8328664302825928,
+      "learning_rate": 3.538150759484981e-05,
+      "loss": 0.9983,
+      "step": 901500
+    },
+    {
+      "epoch": 26.48,
+      "grad_norm": 2.659299850463867,
+      "learning_rate": 3.523441560351843e-05,
+      "loss": 1.0397,
+      "step": 902000
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 2.29178524017334,
+      "learning_rate": 3.508732361218706e-05,
+      "loss": 1.0182,
+      "step": 902500
+    },
+    {
+      "epoch": 26.51,
+      "grad_norm": 9.058496475219727,
+      "learning_rate": 3.494023162085568e-05,
+      "loss": 1.0113,
+      "step": 903000
+    },
+    {
+      "epoch": 26.53,
+      "grad_norm": 1.5530019998550415,
+      "learning_rate": 3.47931396295243e-05,
+      "loss": 1.0243,
+      "step": 903500
+    },
+    {
+      "epoch": 26.54,
+      "grad_norm": 1.9570651054382324,
+      "learning_rate": 3.4646047638192924e-05,
+      "loss": 1.0012,
+      "step": 904000
+    },
+    {
+      "epoch": 26.56,
+      "grad_norm": 1.154820203781128,
+      "learning_rate": 3.4498955646861545e-05,
+      "loss": 1.0231,
+      "step": 904500
+    },
+    {
+      "epoch": 26.57,
+      "grad_norm": 1.5875574350357056,
+      "learning_rate": 3.4351863655530166e-05,
+      "loss": 1.0239,
+      "step": 905000
+    },
+    {
+      "epoch": 26.59,
+      "grad_norm": 2.007080554962158,
+      "learning_rate": 3.420477166419879e-05,
+      "loss": 1.0119,
+      "step": 905500
+    },
+    {
+      "epoch": 26.6,
+      "grad_norm": 1.6997588872909546,
+      "learning_rate": 3.405767967286741e-05,
+      "loss": 1.0235,
+      "step": 906000
+    },
+    {
+      "epoch": 26.62,
+      "grad_norm": 2.8754279613494873,
+      "learning_rate": 3.391058768153603e-05,
+      "loss": 1.0152,
+      "step": 906500
+    },
+    {
+      "epoch": 26.63,
+      "grad_norm": 3.1198441982269287,
+      "learning_rate": 3.376349569020465e-05,
+      "loss": 1.0351,
+      "step": 907000
+    },
+    {
+      "epoch": 26.64,
+      "grad_norm": 11.747318267822266,
+      "learning_rate": 3.361640369887327e-05,
+      "loss": 1.0166,
+      "step": 907500
+    },
+    {
+      "epoch": 26.66,
+      "grad_norm": 1.7430450916290283,
+      "learning_rate": 3.346931170754189e-05,
+      "loss": 1.0268,
+      "step": 908000
+    },
+    {
+      "epoch": 26.67,
+      "grad_norm": 1.4505376815795898,
+      "learning_rate": 3.3322219716210514e-05,
+      "loss": 1.0297,
+      "step": 908500
+    },
+    {
+      "epoch": 26.69,
+      "grad_norm": 4.441274642944336,
+      "learning_rate": 3.3175127724879135e-05,
+      "loss": 1.0145,
+      "step": 909000
+    },
+    {
+      "epoch": 26.7,
+      "grad_norm": 4.974277973175049,
+      "learning_rate": 3.302803573354776e-05,
+      "loss": 1.0163,
+      "step": 909500
+    },
+    {
+      "epoch": 26.72,
+      "grad_norm": 2.1050703525543213,
+      "learning_rate": 3.2880943742216384e-05,
+      "loss": 1.0125,
+      "step": 910000
+    },
+    {
+      "epoch": 26.73,
+      "grad_norm": 4.370170593261719,
+      "learning_rate": 3.2733851750885005e-05,
+      "loss": 1.0134,
+      "step": 910500
+    },
+    {
+      "epoch": 26.75,
+      "grad_norm": 1.755359172821045,
+      "learning_rate": 3.2586759759553626e-05,
+      "loss": 1.0174,
+      "step": 911000
+    },
+    {
+      "epoch": 26.76,
+      "grad_norm": 2.099271774291992,
+      "learning_rate": 3.243966776822225e-05,
+      "loss": 1.0142,
+      "step": 911500
+    },
+    {
+      "epoch": 26.78,
+      "grad_norm": 1.560284972190857,
+      "learning_rate": 3.229257577689087e-05,
+      "loss": 1.0393,
+      "step": 912000
+    },
+    {
+      "epoch": 26.79,
+      "grad_norm": 2.1582424640655518,
+      "learning_rate": 3.214548378555949e-05,
+      "loss": 1.016,
+      "step": 912500
+    },
+    {
+      "epoch": 26.81,
+      "grad_norm": 2.342576026916504,
+      "learning_rate": 3.199839179422811e-05,
+      "loss": 1.0262,
+      "step": 913000
+    },
+    {
+      "epoch": 26.82,
+      "grad_norm": 1.4274934530258179,
+      "learning_rate": 3.185129980289673e-05,
+      "loss": 0.9991,
+      "step": 913500
+    },
+    {
+      "epoch": 26.84,
+      "grad_norm": 3.858720064163208,
+      "learning_rate": 3.170420781156535e-05,
+      "loss": 1.0171,
+      "step": 914000
+    },
+    {
+      "epoch": 26.85,
+      "grad_norm": 2.058018922805786,
+      "learning_rate": 3.1557115820233974e-05,
+      "loss": 1.0263,
+      "step": 914500
+    },
+    {
+      "epoch": 26.86,
+      "grad_norm": 2.6626269817352295,
+      "learning_rate": 3.1410023828902595e-05,
+      "loss": 1.0257,
+      "step": 915000
+    },
+    {
+      "epoch": 26.88,
+      "grad_norm": 5.859118938446045,
+      "learning_rate": 3.1262931837571216e-05,
+      "loss": 1.023,
+      "step": 915500
+    },
+    {
+      "epoch": 26.89,
+      "grad_norm": 6.2187957763671875,
+      "learning_rate": 3.111583984623984e-05,
+      "loss": 1.0107,
+      "step": 916000
+    },
+    {
+      "epoch": 26.91,
+      "grad_norm": 2.6373274326324463,
+      "learning_rate": 3.096874785490846e-05,
+      "loss": 1.0172,
+      "step": 916500
+    },
+    {
+      "epoch": 26.92,
+      "grad_norm": 3.272045850753784,
+      "learning_rate": 3.082165586357708e-05,
+      "loss": 1.0175,
+      "step": 917000
+    },
+    {
+      "epoch": 26.94,
+      "grad_norm": 1.6807574033737183,
+      "learning_rate": 3.06745638722457e-05,
+      "loss": 1.0295,
+      "step": 917500
+    },
+    {
+      "epoch": 26.95,
+      "grad_norm": 4.664640426635742,
+      "learning_rate": 3.052747188091432e-05,
+      "loss": 1.0368,
+      "step": 918000
+    },
+    {
+      "epoch": 26.97,
+      "grad_norm": 1.5610125064849854,
+      "learning_rate": 3.0380379889582943e-05,
+      "loss": 1.0365,
+      "step": 918500
+    },
+    {
+      "epoch": 26.98,
+      "grad_norm": 2.0391860008239746,
+      "learning_rate": 3.0233287898251564e-05,
+      "loss": 1.0289,
+      "step": 919000
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 47.55400848388672,
+      "learning_rate": 3.0086195906920185e-05,
+      "loss": 1.0292,
+      "step": 919500
+    },
+    {
+      "epoch": 27.01,
+      "grad_norm": 2.6204073429107666,
+      "learning_rate": 2.9939103915588807e-05,
+      "loss": 1.0069,
+      "step": 920000
+    },
+    {
+      "epoch": 27.03,
+      "grad_norm": 9.747673034667969,
+      "learning_rate": 2.9792011924257428e-05,
+      "loss": 0.9997,
+      "step": 920500
+    },
+    {
+      "epoch": 27.04,
+      "grad_norm": 2.057465076446533,
+      "learning_rate": 2.9644919932926052e-05,
+      "loss": 0.9993,
+      "step": 921000
+    },
+    {
+      "epoch": 27.06,
+      "grad_norm": 8.264533996582031,
+      "learning_rate": 2.9497827941594673e-05,
+      "loss": 0.9862,
+      "step": 921500
+    },
+    {
+      "epoch": 27.07,
+      "grad_norm": 4.143492698669434,
+      "learning_rate": 2.9350735950263294e-05,
+      "loss": 0.9774,
+      "step": 922000
+    },
+    {
+      "epoch": 27.08,
+      "grad_norm": 2.2518653869628906,
+      "learning_rate": 2.9203643958931916e-05,
+      "loss": 1.0113,
+      "step": 922500
+    },
+    {
+      "epoch": 27.1,
+      "grad_norm": 2.745732307434082,
+      "learning_rate": 2.9056551967600537e-05,
+      "loss": 0.9734,
+      "step": 923000
+    },
+    {
+      "epoch": 27.11,
+      "grad_norm": 1.3990339040756226,
+      "learning_rate": 2.8909459976269158e-05,
+      "loss": 0.9918,
+      "step": 923500
+    },
+    {
+      "epoch": 27.13,
+      "grad_norm": 2.024524211883545,
+      "learning_rate": 2.876236798493778e-05,
+      "loss": 1.0033,
+      "step": 924000
+    },
+    {
+      "epoch": 27.14,
+      "grad_norm": 2.5025253295898438,
+      "learning_rate": 2.86152759936064e-05,
+      "loss": 1.0004,
+      "step": 924500
+    },
+    {
+      "epoch": 27.16,
+      "grad_norm": 2.7478108406066895,
+      "learning_rate": 2.846818400227502e-05,
+      "loss": 0.9901,
+      "step": 925000
+    },
+    {
+      "epoch": 27.17,
+      "grad_norm": 1.7393391132354736,
+      "learning_rate": 2.8321092010943642e-05,
+      "loss": 1.0035,
+      "step": 925500
+    },
+    {
+      "epoch": 27.19,
+      "grad_norm": 1.3920613527297974,
+      "learning_rate": 2.8174000019612263e-05,
+      "loss": 1.0194,
+      "step": 926000
+    },
+    {
+      "epoch": 27.2,
+      "grad_norm": 2.2148451805114746,
+      "learning_rate": 2.8026908028280885e-05,
+      "loss": 0.9858,
+      "step": 926500
+    },
+    {
+      "epoch": 27.22,
+      "grad_norm": 1.7391964197158813,
+      "learning_rate": 2.7879816036949506e-05,
+      "loss": 1.0069,
+      "step": 927000
+    },
+    {
+      "epoch": 27.23,
+      "grad_norm": 2.359649419784546,
+      "learning_rate": 2.7732724045618127e-05,
+      "loss": 0.9989,
+      "step": 927500
+    },
+    {
+      "epoch": 27.25,
+      "grad_norm": 2.8471827507019043,
+      "learning_rate": 2.7585632054286748e-05,
+      "loss": 0.9832,
+      "step": 928000
+    },
+    {
+      "epoch": 27.26,
+      "grad_norm": 1.788493037223816,
+      "learning_rate": 2.743854006295537e-05,
+      "loss": 1.0104,
+      "step": 928500
+    },
+    {
+      "epoch": 27.28,
+      "grad_norm": 1.7663925886154175,
+      "learning_rate": 2.7291448071623994e-05,
+      "loss": 1.0031,
+      "step": 929000
+    },
+    {
+      "epoch": 27.29,
+      "grad_norm": 2.699171304702759,
+      "learning_rate": 2.7144356080292615e-05,
+      "loss": 1.0037,
+      "step": 929500
+    },
+    {
+      "epoch": 27.31,
+      "grad_norm": 12.16844367980957,
+      "learning_rate": 2.6997264088961236e-05,
+      "loss": 1.0109,
+      "step": 930000
+    },
+    {
+      "epoch": 27.32,
+      "grad_norm": 2.0973167419433594,
+      "learning_rate": 2.6850172097629857e-05,
+      "loss": 1.0123,
+      "step": 930500
+    },
+    {
+      "epoch": 27.33,
+      "grad_norm": 1.7330328226089478,
+      "learning_rate": 2.6703080106298478e-05,
+      "loss": 1.0055,
+      "step": 931000
+    },
+    {
+      "epoch": 27.35,
+      "grad_norm": 5.502919673919678,
+      "learning_rate": 2.65559881149671e-05,
+      "loss": 1.003,
+      "step": 931500
+    },
+    {
+      "epoch": 27.36,
+      "grad_norm": 2.3308217525482178,
+      "learning_rate": 2.640889612363572e-05,
+      "loss": 1.0059,
+      "step": 932000
+    },
+    {
+      "epoch": 27.38,
+      "grad_norm": 1.6248698234558105,
+      "learning_rate": 2.626180413230434e-05,
+      "loss": 0.9849,
+      "step": 932500
+    },
+    {
+      "epoch": 27.39,
+      "grad_norm": 3.2510266304016113,
+      "learning_rate": 2.6114712140972963e-05,
+      "loss": 0.9862,
+      "step": 933000
+    },
+    {
+      "epoch": 27.41,
+      "grad_norm": 2.0555245876312256,
+      "learning_rate": 2.5967620149641584e-05,
+      "loss": 0.9844,
+      "step": 933500
+    },
+    {
+      "epoch": 27.42,
+      "grad_norm": 7.706765651702881,
+      "learning_rate": 2.5820528158310205e-05,
+      "loss": 0.9913,
+      "step": 934000
+    },
+    {
+      "epoch": 27.44,
+      "grad_norm": 1.9011601209640503,
+      "learning_rate": 2.5673436166978826e-05,
+      "loss": 1.0102,
+      "step": 934500
+    },
+    {
+      "epoch": 27.45,
+      "grad_norm": 6.408154010772705,
+      "learning_rate": 2.5526344175647447e-05,
+      "loss": 0.996,
+      "step": 935000
+    },
+    {
+      "epoch": 27.47,
+      "grad_norm": 6.746776580810547,
+      "learning_rate": 2.5379252184316068e-05,
+      "loss": 1.0273,
+      "step": 935500
+    },
+    {
+      "epoch": 27.48,
+      "grad_norm": 33.82996368408203,
+      "learning_rate": 2.523216019298469e-05,
+      "loss": 1.007,
+      "step": 936000
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 1.6131765842437744,
+      "learning_rate": 2.508506820165331e-05,
+      "loss": 0.9855,
+      "step": 936500
+    },
+    {
+      "epoch": 27.51,
+      "grad_norm": 2.5696966648101807,
+      "learning_rate": 2.493797621032193e-05,
+      "loss": 1.0153,
+      "step": 937000
+    },
+    {
+      "epoch": 27.53,
+      "grad_norm": 1.5355910062789917,
+      "learning_rate": 2.4790884218990556e-05,
+      "loss": 0.9714,
+      "step": 937500
+    },
+    {
+      "epoch": 27.54,
+      "grad_norm": 2.0435054302215576,
+      "learning_rate": 2.4643792227659177e-05,
+      "loss": 0.9997,
+      "step": 938000
+    },
+    {
+      "epoch": 27.55,
+      "grad_norm": 1.6511478424072266,
+      "learning_rate": 2.44967002363278e-05,
+      "loss": 0.9865,
+      "step": 938500
+    },
+    {
+      "epoch": 27.57,
+      "grad_norm": 1.9227514266967773,
+      "learning_rate": 2.434960824499642e-05,
+      "loss": 1.0103,
+      "step": 939000
+    },
+    {
+      "epoch": 27.58,
+      "grad_norm": 16.868221282958984,
+      "learning_rate": 2.420251625366504e-05,
+      "loss": 0.9854,
+      "step": 939500
+    },
+    {
+      "epoch": 27.6,
+      "grad_norm": 1.7769354581832886,
+      "learning_rate": 2.4055424262333662e-05,
+      "loss": 1.0176,
+      "step": 940000
+    },
+    {
+      "epoch": 27.61,
+      "grad_norm": 1.6719156503677368,
+      "learning_rate": 2.3908332271002283e-05,
+      "loss": 0.9951,
+      "step": 940500
+    },
+    {
+      "epoch": 27.63,
+      "grad_norm": 3.5106630325317383,
+      "learning_rate": 2.3761240279670904e-05,
+      "loss": 0.994,
+      "step": 941000
+    },
+    {
+      "epoch": 27.64,
+      "grad_norm": 3.102487564086914,
+      "learning_rate": 2.3614148288339525e-05,
+      "loss": 0.9802,
+      "step": 941500
+    },
+    {
+      "epoch": 27.66,
+      "grad_norm": 12.020054817199707,
+      "learning_rate": 2.3467056297008146e-05,
+      "loss": 0.9997,
+      "step": 942000
+    },
+    {
+      "epoch": 27.67,
+      "grad_norm": 2.8283021450042725,
+      "learning_rate": 2.3319964305676767e-05,
+      "loss": 0.9892,
+      "step": 942500
+    },
+    {
+      "epoch": 27.69,
+      "grad_norm": 37.08485412597656,
+      "learning_rate": 2.317287231434539e-05,
+      "loss": 0.9969,
+      "step": 943000
+    },
+    {
+      "epoch": 27.7,
+      "grad_norm": 1.4365429878234863,
+      "learning_rate": 2.302578032301401e-05,
+      "loss": 0.9748,
+      "step": 943500
+    },
+    {
+      "epoch": 27.72,
+      "grad_norm": 1.8424559831619263,
+      "learning_rate": 2.287868833168263e-05,
+      "loss": 0.9741,
+      "step": 944000
+    },
+    {
+      "epoch": 27.73,
+      "grad_norm": 2.4849884510040283,
+      "learning_rate": 2.2731596340351252e-05,
+      "loss": 0.9896,
+      "step": 944500
+    },
+    {
+      "epoch": 27.75,
+      "grad_norm": 2.0955851078033447,
+      "learning_rate": 2.2584504349019873e-05,
+      "loss": 1.012,
+      "step": 945000
+    },
+    {
+      "epoch": 27.76,
+      "grad_norm": 5.421468257904053,
+      "learning_rate": 2.2437412357688498e-05,
+      "loss": 0.9681,
+      "step": 945500
+    },
+    {
+      "epoch": 27.77,
+      "grad_norm": 3.2725253105163574,
+      "learning_rate": 2.229032036635712e-05,
+      "loss": 1.0059,
+      "step": 946000
+    },
+    {
+      "epoch": 27.79,
+      "grad_norm": 2.741119861602783,
+      "learning_rate": 2.214322837502574e-05,
+      "loss": 0.9933,
+      "step": 946500
+    },
+    {
+      "epoch": 27.8,
+      "grad_norm": 1.600335717201233,
+      "learning_rate": 2.199613638369436e-05,
+      "loss": 0.997,
+      "step": 947000
+    },
+    {
+      "epoch": 27.82,
+      "grad_norm": 3.4619009494781494,
+      "learning_rate": 2.1849044392362982e-05,
+      "loss": 0.9976,
+      "step": 947500
+    },
+    {
+      "epoch": 27.83,
+      "grad_norm": 10.748187065124512,
+      "learning_rate": 2.1701952401031603e-05,
+      "loss": 0.9727,
+      "step": 948000
+    },
+    {
+      "epoch": 27.85,
+      "grad_norm": 3.452791690826416,
+      "learning_rate": 2.1554860409700224e-05,
+      "loss": 1.001,
+      "step": 948500
+    },
+    {
+      "epoch": 27.86,
+      "grad_norm": 1.3464235067367554,
+      "learning_rate": 2.1407768418368845e-05,
+      "loss": 0.9876,
+      "step": 949000
+    },
+    {
+      "epoch": 27.88,
+      "grad_norm": 1.6416382789611816,
+      "learning_rate": 2.1260676427037467e-05,
+      "loss": 1.0062,
+      "step": 949500
+    },
+    {
+      "epoch": 27.89,
+      "grad_norm": 1.434910774230957,
+      "learning_rate": 2.1113584435706088e-05,
+      "loss": 0.9934,
+      "step": 950000
+    },
+    {
+      "epoch": 27.91,
+      "grad_norm": 5.8801045417785645,
+      "learning_rate": 2.096649244437471e-05,
+      "loss": 0.9899,
+      "step": 950500
+    },
+    {
+      "epoch": 27.92,
+      "grad_norm": 1.864310622215271,
+      "learning_rate": 2.081940045304333e-05,
+      "loss": 1.013,
+      "step": 951000
+    },
+    {
+      "epoch": 27.94,
+      "grad_norm": 2.070906639099121,
+      "learning_rate": 2.067230846171195e-05,
+      "loss": 0.9947,
+      "step": 951500
+    },
+    {
+      "epoch": 27.95,
+      "grad_norm": 2.5215396881103516,
+      "learning_rate": 2.0525216470380572e-05,
+      "loss": 0.984,
+      "step": 952000
+    },
+    {
+      "epoch": 27.97,
+      "grad_norm": 1.2144149541854858,
+      "learning_rate": 2.0378124479049193e-05,
+      "loss": 0.9993,
+      "step": 952500
+    },
+    {
+      "epoch": 27.98,
+      "grad_norm": 2.4629998207092285,
+      "learning_rate": 2.0231032487717814e-05,
+      "loss": 1.0146,
+      "step": 953000
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 4.50878381729126,
+      "learning_rate": 2.008394049638644e-05,
+      "loss": 1.0005,
+      "step": 953500
+    },
+    {
+      "epoch": 28.01,
+      "grad_norm": 2.623155355453491,
+      "learning_rate": 1.993684850505506e-05,
+      "loss": 0.9928,
+      "step": 954000
+    },
+    {
+      "epoch": 28.02,
+      "grad_norm": 2.886497735977173,
+      "learning_rate": 1.978975651372368e-05,
+      "loss": 0.9886,
+      "step": 954500
+    },
+    {
+      "epoch": 28.04,
+      "grad_norm": 2.8885116577148438,
+      "learning_rate": 1.9642664522392302e-05,
+      "loss": 0.9788,
+      "step": 955000
+    },
+    {
+      "epoch": 28.05,
+      "grad_norm": 1.946003794670105,
+      "learning_rate": 1.9495572531060923e-05,
+      "loss": 0.9647,
+      "step": 955500
+    },
+    {
+      "epoch": 28.07,
+      "grad_norm": 12.426846504211426,
+      "learning_rate": 1.9348480539729545e-05,
+      "loss": 0.9753,
+      "step": 956000
+    },
+    {
+      "epoch": 28.08,
+      "grad_norm": 1.905735969543457,
+      "learning_rate": 1.9201388548398166e-05,
+      "loss": 0.9855,
+      "step": 956500
+    },
+    {
+      "epoch": 28.1,
+      "grad_norm": 2.33889102935791,
+      "learning_rate": 1.9054296557066787e-05,
+      "loss": 0.9744,
+      "step": 957000
+    },
+    {
+      "epoch": 28.11,
+      "grad_norm": 2.3063101768493652,
+      "learning_rate": 1.8907204565735408e-05,
+      "loss": 0.9702,
+      "step": 957500
+    },
+    {
+      "epoch": 28.13,
+      "grad_norm": 2.292107582092285,
+      "learning_rate": 1.876011257440403e-05,
+      "loss": 0.9888,
+      "step": 958000
+    },
+    {
+      "epoch": 28.14,
+      "grad_norm": 2.144972324371338,
+      "learning_rate": 1.8613020583072654e-05,
+      "loss": 0.9754,
+      "step": 958500
+    },
+    {
+      "epoch": 28.16,
+      "grad_norm": 1.5302867889404297,
+      "learning_rate": 1.8465928591741275e-05,
+      "loss": 0.9816,
+      "step": 959000
+    },
+    {
+      "epoch": 28.17,
+      "grad_norm": 1.6462048292160034,
+      "learning_rate": 1.8318836600409896e-05,
+      "loss": 0.9403,
+      "step": 959500
+    },
+    {
+      "epoch": 28.19,
+      "grad_norm": 2.6783134937286377,
+      "learning_rate": 1.8171744609078517e-05,
+      "loss": 0.9762,
+      "step": 960000
+    },
+    {
+      "epoch": 28.2,
+      "grad_norm": 2.0703890323638916,
+      "learning_rate": 1.8024652617747138e-05,
+      "loss": 0.9756,
+      "step": 960500
+    },
+    {
+      "epoch": 28.22,
+      "grad_norm": 2.4419829845428467,
+      "learning_rate": 1.787756062641576e-05,
+      "loss": 0.9684,
+      "step": 961000
+    },
+    {
+      "epoch": 28.23,
+      "grad_norm": 4.775966644287109,
+      "learning_rate": 1.773046863508438e-05,
+      "loss": 0.9912,
+      "step": 961500
+    },
+    {
+      "epoch": 28.24,
+      "grad_norm": 1.0155693292617798,
+      "learning_rate": 1.7583376643753e-05,
+      "loss": 0.97,
+      "step": 962000
+    },
+    {
+      "epoch": 28.26,
+      "grad_norm": 2.105178117752075,
+      "learning_rate": 1.7436284652421623e-05,
+      "loss": 0.9699,
+      "step": 962500
+    },
+    {
+      "epoch": 28.27,
+      "grad_norm": 11.809367179870605,
+      "learning_rate": 1.7289192661090244e-05,
+      "loss": 0.9869,
+      "step": 963000
+    },
+    {
+      "epoch": 28.29,
+      "grad_norm": 2.5389130115509033,
+      "learning_rate": 1.7142100669758865e-05,
+      "loss": 0.9807,
+      "step": 963500
+    },
+    {
+      "epoch": 28.3,
+      "grad_norm": 2.3344273567199707,
+      "learning_rate": 1.6995008678427486e-05,
+      "loss": 0.9861,
+      "step": 964000
+    },
+    {
+      "epoch": 28.32,
+      "grad_norm": 2.3741703033447266,
+      "learning_rate": 1.6847916687096107e-05,
+      "loss": 0.9592,
+      "step": 964500
+    },
+    {
+      "epoch": 28.33,
+      "grad_norm": 5.04816198348999,
+      "learning_rate": 1.670082469576473e-05,
+      "loss": 0.9903,
+      "step": 965000
+    },
+    {
+      "epoch": 28.35,
+      "grad_norm": 1.2963931560516357,
+      "learning_rate": 1.6553732704433353e-05,
+      "loss": 0.9676,
+      "step": 965500
+    },
+    {
+      "epoch": 28.36,
+      "grad_norm": 3.1595191955566406,
+      "learning_rate": 1.6406640713101974e-05,
+      "loss": 0.9942,
+      "step": 966000
+    },
+    {
+      "epoch": 28.38,
+      "grad_norm": 1.5514929294586182,
+      "learning_rate": 1.6259548721770595e-05,
+      "loss": 0.9741,
+      "step": 966500
+    },
+    {
+      "epoch": 28.39,
+      "grad_norm": 1.9268239736557007,
+      "learning_rate": 1.6112456730439216e-05,
+      "loss": 0.9869,
+      "step": 967000
+    },
+    {
+      "epoch": 28.41,
+      "grad_norm": 2.7622382640838623,
+      "learning_rate": 1.5965364739107837e-05,
+      "loss": 0.9598,
+      "step": 967500
+    },
+    {
+      "epoch": 28.42,
+      "grad_norm": 1.418778657913208,
+      "learning_rate": 1.581827274777646e-05,
+      "loss": 0.9464,
+      "step": 968000
+    },
+    {
+      "epoch": 28.44,
+      "grad_norm": 2.1014456748962402,
+      "learning_rate": 1.567118075644508e-05,
+      "loss": 0.9846,
+      "step": 968500
+    },
+    {
+      "epoch": 28.45,
+      "grad_norm": 14.58630657196045,
+      "learning_rate": 1.55240887651137e-05,
+      "loss": 0.9695,
+      "step": 969000
+    },
+    {
+      "epoch": 28.46,
+      "grad_norm": 1.556394100189209,
+      "learning_rate": 1.5376996773782322e-05,
+      "loss": 0.9916,
+      "step": 969500
+    },
+    {
+      "epoch": 28.48,
+      "grad_norm": 1.8601114749908447,
+      "learning_rate": 1.5229904782450943e-05,
+      "loss": 0.9552,
+      "step": 970000
+    },
+    {
+      "epoch": 28.49,
+      "grad_norm": 2.2888152599334717,
+      "learning_rate": 1.5082812791119564e-05,
+      "loss": 0.972,
+      "step": 970500
+    },
+    {
+      "epoch": 28.51,
+      "grad_norm": 2.9334394931793213,
+      "learning_rate": 1.4935720799788187e-05,
+      "loss": 0.9986,
+      "step": 971000
+    },
+    {
+      "epoch": 28.52,
+      "grad_norm": 2.88075590133667,
+      "learning_rate": 1.4788628808456808e-05,
+      "loss": 0.9935,
+      "step": 971500
+    },
+    {
+      "epoch": 28.54,
+      "grad_norm": 4.609189510345459,
+      "learning_rate": 1.4641536817125429e-05,
+      "loss": 0.9715,
+      "step": 972000
+    },
+    {
+      "epoch": 28.55,
+      "grad_norm": 1.6899704933166504,
+      "learning_rate": 1.449444482579405e-05,
+      "loss": 0.9737,
+      "step": 972500
+    },
+    {
+      "epoch": 28.57,
+      "grad_norm": 18.26349449157715,
+      "learning_rate": 1.4347352834462671e-05,
+      "loss": 0.9728,
+      "step": 973000
+    },
+    {
+      "epoch": 28.58,
+      "grad_norm": 5.080548286437988,
+      "learning_rate": 1.4200260843131292e-05,
+      "loss": 0.9819,
+      "step": 973500
+    },
+    {
+      "epoch": 28.6,
+      "grad_norm": 3.9338316917419434,
+      "learning_rate": 1.4053168851799914e-05,
+      "loss": 0.9758,
+      "step": 974000
+    },
+    {
+      "epoch": 28.61,
+      "grad_norm": 1.4379346370697021,
+      "learning_rate": 1.3906076860468535e-05,
+      "loss": 0.9634,
+      "step": 974500
+    },
+    {
+      "epoch": 28.63,
+      "grad_norm": 2.943136215209961,
+      "learning_rate": 1.3758984869137157e-05,
+      "loss": 0.9519,
+      "step": 975000
+    },
+    {
+      "epoch": 28.64,
+      "grad_norm": 3.1957294940948486,
+      "learning_rate": 1.3611892877805779e-05,
+      "loss": 0.9727,
+      "step": 975500
+    },
+    {
+      "epoch": 28.66,
+      "grad_norm": 2.38423752784729,
+      "learning_rate": 1.34648008864744e-05,
+      "loss": 0.9787,
+      "step": 976000
+    },
+    {
+      "epoch": 28.67,
+      "grad_norm": 2.059079885482788,
+      "learning_rate": 1.331770889514302e-05,
+      "loss": 0.9851,
+      "step": 976500
+    },
+    {
+      "epoch": 28.69,
+      "grad_norm": 1.8208869695663452,
+      "learning_rate": 1.3170616903811642e-05,
+      "loss": 0.9665,
+      "step": 977000
+    },
+    {
+      "epoch": 28.7,
+      "grad_norm": 2.108597755432129,
+      "learning_rate": 1.3023524912480263e-05,
+      "loss": 0.9941,
+      "step": 977500
+    },
+    {
+      "epoch": 28.71,
+      "grad_norm": 1.5259730815887451,
+      "learning_rate": 1.2876432921148884e-05,
+      "loss": 0.98,
+      "step": 978000
+    },
+    {
+      "epoch": 28.73,
+      "grad_norm": 3.095134735107422,
+      "learning_rate": 1.2729340929817505e-05,
+      "loss": 0.9962,
+      "step": 978500
+    },
+    {
+      "epoch": 28.74,
+      "grad_norm": 1.49867582321167,
+      "learning_rate": 1.2582248938486128e-05,
+      "loss": 0.9714,
+      "step": 979000
+    },
+    {
+      "epoch": 28.76,
+      "grad_norm": 1.5933908224105835,
+      "learning_rate": 1.243515694715475e-05,
+      "loss": 0.999,
+      "step": 979500
+    },
+    {
+      "epoch": 28.77,
+      "grad_norm": 2.663355588912964,
+      "learning_rate": 1.228806495582337e-05,
+      "loss": 0.9682,
+      "step": 980000
+    },
+    {
+      "epoch": 28.79,
+      "grad_norm": 104.24024200439453,
+      "learning_rate": 1.2140972964491992e-05,
+      "loss": 0.9774,
+      "step": 980500
+    },
+    {
+      "epoch": 28.8,
+      "grad_norm": 1.7283943891525269,
+      "learning_rate": 1.1993880973160613e-05,
+      "loss": 0.9607,
+      "step": 981000
+    },
+    {
+      "epoch": 28.82,
+      "grad_norm": 2.0852441787719727,
+      "learning_rate": 1.1846788981829234e-05,
+      "loss": 1.0005,
+      "step": 981500
+    },
+    {
+      "epoch": 28.83,
+      "grad_norm": 2.0442757606506348,
+      "learning_rate": 1.1699696990497855e-05,
+      "loss": 0.9836,
+      "step": 982000
+    },
+    {
+      "epoch": 28.85,
+      "grad_norm": 1.5879892110824585,
+      "learning_rate": 1.1552604999166476e-05,
+      "loss": 0.9973,
+      "step": 982500
+    },
+    {
+      "epoch": 28.86,
+      "grad_norm": 2.7892065048217773,
+      "learning_rate": 1.14055130078351e-05,
+      "loss": 0.9615,
+      "step": 983000
+    },
+    {
+      "epoch": 28.88,
+      "grad_norm": 3.826296329498291,
+      "learning_rate": 1.1258421016503722e-05,
+      "loss": 0.9746,
+      "step": 983500
+    },
+    {
+      "epoch": 28.89,
+      "grad_norm": 7.535625457763672,
+      "learning_rate": 1.1111329025172343e-05,
+      "loss": 0.9685,
+      "step": 984000
+    },
+    {
+      "epoch": 28.91,
+      "grad_norm": 1.7329784631729126,
+      "learning_rate": 1.0964237033840964e-05,
+      "loss": 0.9745,
+      "step": 984500
+    },
+    {
+      "epoch": 28.92,
+      "grad_norm": 1.9252818822860718,
+      "learning_rate": 1.0817145042509585e-05,
+      "loss": 0.9693,
+      "step": 985000
+    },
+    {
+      "epoch": 28.93,
+      "grad_norm": 2.0630569458007812,
+      "learning_rate": 1.0670053051178206e-05,
+      "loss": 0.9634,
+      "step": 985500
+    },
+    {
+      "epoch": 28.95,
+      "grad_norm": 1.6531202793121338,
+      "learning_rate": 1.0522961059846827e-05,
+      "loss": 0.9702,
+      "step": 986000
+    },
+    {
+      "epoch": 28.96,
+      "grad_norm": 32.64999771118164,
+      "learning_rate": 1.037586906851545e-05,
+      "loss": 0.9593,
+      "step": 986500
+    },
+    {
+      "epoch": 28.98,
+      "grad_norm": 5.578742980957031,
+      "learning_rate": 1.0228777077184071e-05,
+      "loss": 0.9624,
+      "step": 987000
+    },
+    {
+      "epoch": 28.99,
+      "grad_norm": 1.9109100103378296,
+      "learning_rate": 1.0081685085852692e-05,
+      "loss": 0.9698,
+      "step": 987500
+    },
+    {
+      "epoch": 29.01,
+      "grad_norm": 3.434389352798462,
+      "learning_rate": 9.934593094521313e-06,
+      "loss": 0.9824,
+      "step": 988000
+    },
+    {
+      "epoch": 29.02,
+      "grad_norm": 3.804184913635254,
+      "learning_rate": 9.787501103189935e-06,
+      "loss": 0.9534,
+      "step": 988500
+    },
+    {
+      "epoch": 29.04,
+      "grad_norm": 2.4718267917633057,
+      "learning_rate": 9.640409111858556e-06,
+      "loss": 0.9533,
+      "step": 989000
+    },
+    {
+      "epoch": 29.05,
+      "grad_norm": 2.81648588180542,
+      "learning_rate": 9.493317120527177e-06,
+      "loss": 0.9688,
+      "step": 989500
+    },
+    {
+      "epoch": 29.07,
+      "grad_norm": 3.936281681060791,
+      "learning_rate": 9.346225129195798e-06,
+      "loss": 0.9486,
+      "step": 990000
+    },
+    {
+      "epoch": 29.08,
+      "grad_norm": 4.693751335144043,
+      "learning_rate": 9.19913313786442e-06,
+      "loss": 0.9505,
+      "step": 990500
+    },
+    {
+      "epoch": 29.1,
+      "grad_norm": 3.5278286933898926,
+      "learning_rate": 9.052041146533042e-06,
+      "loss": 0.9338,
+      "step": 991000
+    },
+    {
+      "epoch": 29.11,
+      "grad_norm": 2.9467105865478516,
+      "learning_rate": 8.904949155201663e-06,
+      "loss": 0.9602,
+      "step": 991500
+    },
+    {
+      "epoch": 29.13,
+      "grad_norm": 2.929056406021118,
+      "learning_rate": 8.757857163870284e-06,
+      "loss": 0.9469,
+      "step": 992000
+    },
+    {
+      "epoch": 29.14,
+      "grad_norm": 4.979098320007324,
+      "learning_rate": 8.610765172538905e-06,
+      "loss": 0.9364,
+      "step": 992500
+    },
+    {
+      "epoch": 29.15,
+      "grad_norm": 3.7940807342529297,
+      "learning_rate": 8.463673181207526e-06,
+      "loss": 0.9407,
+      "step": 993000
+    },
+    {
+      "epoch": 29.17,
+      "grad_norm": 2.5154898166656494,
+      "learning_rate": 8.316581189876148e-06,
+      "loss": 0.9504,
+      "step": 993500
+    },
+    {
+      "epoch": 29.18,
+      "grad_norm": 3.7596309185028076,
+      "learning_rate": 8.169489198544769e-06,
+      "loss": 0.9762,
+      "step": 994000
+    },
+    {
+      "epoch": 29.2,
+      "grad_norm": 2.13797926902771,
+      "learning_rate": 8.022397207213392e-06,
+      "loss": 0.9605,
+      "step": 994500
+    },
+    {
+      "epoch": 29.21,
+      "grad_norm": 3.1647770404815674,
+      "learning_rate": 7.875305215882013e-06,
+      "loss": 0.9611,
+      "step": 995000
+    },
+    {
+      "epoch": 29.23,
+      "grad_norm": 3.417736768722534,
+      "learning_rate": 7.728213224550634e-06,
+      "loss": 0.9611,
+      "step": 995500
+    },
+    {
+      "epoch": 29.24,
+      "grad_norm": 4.05425500869751,
+      "learning_rate": 7.581121233219255e-06,
+      "loss": 0.9733,
+      "step": 996000
+    },
+    {
+      "epoch": 29.26,
+      "grad_norm": 2.117457866668701,
+      "learning_rate": 7.434029241887876e-06,
+      "loss": 0.9494,
+      "step": 996500
+    },
+    {
+      "epoch": 29.27,
+      "grad_norm": 9.484780311584473,
+      "learning_rate": 7.286937250556497e-06,
+      "loss": 0.9842,
+      "step": 997000
+    },
+    {
+      "epoch": 29.29,
+      "grad_norm": 5.586018085479736,
+      "learning_rate": 7.139845259225119e-06,
+      "loss": 0.955,
+      "step": 997500
+    },
+    {
+      "epoch": 29.3,
+      "grad_norm": 6.439801216125488,
+      "learning_rate": 6.99275326789374e-06,
+      "loss": 0.9608,
+      "step": 998000
+    },
+    {
+      "epoch": 29.32,
+      "grad_norm": 2.890882730484009,
+      "learning_rate": 6.845661276562361e-06,
+      "loss": 0.9649,
+      "step": 998500
+    },
+    {
+      "epoch": 29.33,
+      "grad_norm": 2.0991222858428955,
+      "learning_rate": 6.6985692852309825e-06,
+      "loss": 0.9476,
+      "step": 999000
+    },
+    {
+      "epoch": 29.35,
+      "grad_norm": 2.40987229347229,
+      "learning_rate": 6.5514772938996045e-06,
+      "loss": 0.9537,
+      "step": 999500
+    },
+    {
+      "epoch": 29.36,
+      "grad_norm": 2.95318865776062,
+      "learning_rate": 6.4043853025682256e-06,
+      "loss": 0.9749,
+      "step": 1000000
+    },
+    {
+      "epoch": 29.38,
+      "grad_norm": 5.403099060058594,
+      "learning_rate": 6.257293311236847e-06,
+      "loss": 0.9496,
+      "step": 1000500
+    },
+    {
+      "epoch": 29.39,
+      "grad_norm": 3.3474481105804443,
+      "learning_rate": 6.110201319905468e-06,
+      "loss": 0.9756,
+      "step": 1001000
+    },
+    {
+      "epoch": 29.4,
+      "grad_norm": 2.7617027759552,
+      "learning_rate": 5.96310932857409e-06,
+      "loss": 0.944,
+      "step": 1001500
+    },
+    {
+      "epoch": 29.42,
+      "grad_norm": 3.7849531173706055,
+      "learning_rate": 5.816017337242711e-06,
+      "loss": 0.9616,
+      "step": 1002000
+    },
+    {
+      "epoch": 29.43,
+      "grad_norm": 2.1899170875549316,
+      "learning_rate": 5.668925345911332e-06,
+      "loss": 0.9578,
+      "step": 1002500
+    },
+    {
+      "epoch": 29.45,
+      "grad_norm": 2.423264265060425,
+      "learning_rate": 5.521833354579953e-06,
+      "loss": 0.9886,
+      "step": 1003000
+    },
+    {
+      "epoch": 29.46,
+      "grad_norm": 10.377019882202148,
+      "learning_rate": 5.374741363248575e-06,
+      "loss": 0.9559,
+      "step": 1003500
+    },
+    {
+      "epoch": 29.48,
+      "grad_norm": 3.244654655456543,
+      "learning_rate": 5.227649371917196e-06,
+      "loss": 0.9467,
+      "step": 1004000
+    },
+    {
+      "epoch": 29.49,
+      "grad_norm": 3.0262067317962646,
+      "learning_rate": 5.080557380585817e-06,
+      "loss": 0.9825,
+      "step": 1004500
+    },
+    {
+      "epoch": 29.51,
+      "grad_norm": 1.8769322633743286,
+      "learning_rate": 4.933465389254439e-06,
+      "loss": 0.9835,
+      "step": 1005000
+    },
+    {
+      "epoch": 29.52,
+      "grad_norm": 3.122030735015869,
+      "learning_rate": 4.786373397923061e-06,
+      "loss": 0.9686,
+      "step": 1005500
+    },
+    {
+      "epoch": 29.54,
+      "grad_norm": 3.3955576419830322,
+      "learning_rate": 4.639281406591682e-06,
+      "loss": 0.9549,
+      "step": 1006000
+    },
+    {
+      "epoch": 29.55,
+      "grad_norm": 11.812577247619629,
+      "learning_rate": 4.492189415260304e-06,
+      "loss": 0.944,
+      "step": 1006500
+    },
+    {
+      "epoch": 29.57,
+      "grad_norm": 3.2719273567199707,
+      "learning_rate": 4.345097423928925e-06,
+      "loss": 0.9675,
+      "step": 1007000
+    },
+    {
+      "epoch": 29.58,
+      "grad_norm": 2.0601563453674316,
+      "learning_rate": 4.198005432597546e-06,
+      "loss": 0.9816,
+      "step": 1007500
+    },
+    {
+      "epoch": 29.6,
+      "grad_norm": 2.1465187072753906,
+      "learning_rate": 4.050913441266167e-06,
+      "loss": 0.9386,
+      "step": 1008000
+    },
+    {
+      "epoch": 29.61,
+      "grad_norm": 3.070183753967285,
+      "learning_rate": 3.903821449934789e-06,
+      "loss": 0.9729,
+      "step": 1008500
+    },
+    {
+      "epoch": 29.62,
+      "grad_norm": 8.74717903137207,
+      "learning_rate": 3.7567294586034105e-06,
+      "loss": 0.9483,
+      "step": 1009000
+    },
+    {
+      "epoch": 29.64,
+      "grad_norm": 4.731165409088135,
+      "learning_rate": 3.6096374672720316e-06,
+      "loss": 0.9699,
+      "step": 1009500
+    },
+    {
+      "epoch": 29.65,
+      "grad_norm": 3.4259703159332275,
+      "learning_rate": 3.462545475940653e-06,
+      "loss": 0.9768,
+      "step": 1010000
+    },
+    {
+      "epoch": 29.67,
+      "grad_norm": 2.1280946731567383,
+      "learning_rate": 3.3154534846092743e-06,
+      "loss": 0.9732,
+      "step": 1010500
+    },
+    {
+      "epoch": 29.68,
+      "grad_norm": 2.901146411895752,
+      "learning_rate": 3.168361493277896e-06,
+      "loss": 0.968,
+      "step": 1011000
+    },
+    {
+      "epoch": 29.7,
+      "grad_norm": 3.9223461151123047,
+      "learning_rate": 3.021269501946517e-06,
+      "loss": 0.9499,
+      "step": 1011500
+    },
+    {
+      "epoch": 29.71,
+      "grad_norm": 2.9555001258850098,
+      "learning_rate": 2.8741775106151385e-06,
+      "loss": 0.9306,
+      "step": 1012000
+    },
+    {
+      "epoch": 29.73,
+      "grad_norm": 3.057680606842041,
+      "learning_rate": 2.7270855192837596e-06,
+      "loss": 0.9558,
+      "step": 1012500
+    },
+    {
+      "epoch": 29.74,
+      "grad_norm": 2.978799343109131,
+      "learning_rate": 2.579993527952381e-06,
+      "loss": 0.9701,
+      "step": 1013000
+    },
+    {
+      "epoch": 29.76,
+      "grad_norm": 2.0970911979675293,
+      "learning_rate": 2.4329015366210023e-06,
+      "loss": 0.9577,
+      "step": 1013500
+    },
+    {
+      "epoch": 29.77,
+      "grad_norm": 2.8144404888153076,
+      "learning_rate": 2.285809545289624e-06,
+      "loss": 0.9587,
+      "step": 1014000
+    },
+    {
+      "epoch": 29.79,
+      "grad_norm": 2.756420135498047,
+      "learning_rate": 2.1387175539582454e-06,
+      "loss": 0.9554,
+      "step": 1014500
+    },
+    {
+      "epoch": 29.8,
+      "grad_norm": 3.5521059036254883,
+      "learning_rate": 1.9916255626268665e-06,
+      "loss": 0.95,
+      "step": 1015000
+    },
+    {
+      "epoch": 29.82,
+      "grad_norm": 34.70965576171875,
+      "learning_rate": 1.844533571295488e-06,
+      "loss": 0.9708,
+      "step": 1015500
+    },
+    {
+      "epoch": 29.83,
+      "grad_norm": 3.2949025630950928,
+      "learning_rate": 1.6974415799641094e-06,
+      "loss": 0.9673,
+      "step": 1016000
+    },
+    {
+      "epoch": 29.84,
+      "grad_norm": 2.9649112224578857,
+      "learning_rate": 1.5503495886327307e-06,
+      "loss": 0.9758,
+      "step": 1016500
+    },
+    {
+      "epoch": 29.86,
+      "grad_norm": 2.4358630180358887,
+      "learning_rate": 1.403257597301352e-06,
+      "loss": 0.9793,
+      "step": 1017000
+    },
+    {
+      "epoch": 29.87,
+      "grad_norm": 3.3332042694091797,
+      "learning_rate": 1.2561656059699736e-06,
+      "loss": 0.9783,
+      "step": 1017500
+    },
+    {
+      "epoch": 29.89,
+      "grad_norm": 5.788940906524658,
+      "learning_rate": 1.1090736146385948e-06,
+      "loss": 0.9519,
+      "step": 1018000
+    },
+    {
+      "epoch": 29.9,
+      "grad_norm": 2.6301000118255615,
+      "learning_rate": 9.619816233072163e-07,
+      "loss": 0.9558,
+      "step": 1018500
+    },
+    {
+      "epoch": 29.92,
+      "grad_norm": 2.9536306858062744,
+      "learning_rate": 8.148896319758376e-07,
+      "loss": 0.9466,
+      "step": 1019000
+    },
+    {
+      "epoch": 29.93,
+      "grad_norm": 2.9202470779418945,
+      "learning_rate": 6.67797640644459e-07,
+      "loss": 0.9695,
+      "step": 1019500
+    },
+    {
+      "epoch": 29.95,
+      "grad_norm": 3.0824179649353027,
+      "learning_rate": 5.207056493130803e-07,
+      "loss": 0.9361,
+      "step": 1020000
+    },
+    {
+      "epoch": 29.96,
+      "grad_norm": 2.5186848640441895,
+      "learning_rate": 3.736136579817017e-07,
+      "loss": 0.9476,
+      "step": 1020500
+    },
+    {
+      "epoch": 29.98,
+      "grad_norm": 5.588615894317627,
+      "learning_rate": 2.265216666503231e-07,
+      "loss": 0.9431,
+      "step": 1021000
+    },
+    {
+      "epoch": 29.99,
+      "grad_norm": 2.025895595550537,
+      "learning_rate": 7.942967531894445e-08,
+      "loss": 0.9567,
+      "step": 1021500
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1021770,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 1.7792790799000776e+21,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}