diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10462 @@
+{
+  "best_global_step": 148440,
+  "best_metric": 0.7365977168083191,
+  "best_model_checkpoint": "./output_optimized/checkpoint-148440",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 148440,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0033683643222850983,
+      "grad_norm": 1.9248961210250854,
+      "learning_rate": 2.000808407437348e-06,
+      "loss": 5.0547,
+      "step": 100
+    },
+    {
+      "epoch": 0.006736728644570197,
+      "grad_norm": 1.0100284814834595,
+      "learning_rate": 4.021827000808407e-06,
+      "loss": 4.9226,
+      "step": 200
+    },
+    {
+      "epoch": 0.010105092966855295,
+      "grad_norm": 0.8452675938606262,
+      "learning_rate": 6.042845594179466e-06,
+      "loss": 4.8032,
+      "step": 300
+    },
+    {
+      "epoch": 0.013473457289140393,
+      "grad_norm": 0.878520131111145,
+      "learning_rate": 8.063864187550524e-06,
+      "loss": 4.7193,
+      "step": 400
+    },
+    {
+      "epoch": 0.016841821611425493,
+      "grad_norm": 1.0584090948104858,
+      "learning_rate": 1.0084882780921583e-05,
+      "loss": 4.5769,
+      "step": 500
+    },
+    {
+      "epoch": 0.02021018593371059,
+      "grad_norm": 1.1221024990081787,
+      "learning_rate": 1.2105901374292643e-05,
+      "loss": 4.3638,
+      "step": 600
+    },
+    {
+      "epoch": 0.02357855025599569,
+      "grad_norm": 1.0861926078796387,
+      "learning_rate": 1.41269199676637e-05,
+      "loss": 4.155,
+      "step": 700
+    },
+    {
+      "epoch": 0.026946914578280787,
+      "grad_norm": 0.9905880093574524,
+      "learning_rate": 1.6147938561034762e-05,
+      "loss": 3.9701,
+      "step": 800
+    },
+    {
+      "epoch": 0.030315278900565887,
+      "grad_norm": 0.9011399149894714,
+      "learning_rate": 1.816895715440582e-05,
+      "loss": 3.7974,
+      "step": 900
+    },
+    {
+      "epoch": 0.03368364322285099,
+      "grad_norm": 0.8480072617530823,
+      "learning_rate": 2.0189975747776877e-05,
+      "loss": 3.6336,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03705200754513608,
+      "grad_norm": 0.7879598736763,
+      "learning_rate": 2.2210994341147935e-05,
+      "loss": 3.4894,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04042037186742118,
+      "grad_norm": 0.7154058814048767,
+      "learning_rate": 2.4232012934518997e-05,
+      "loss": 3.3559,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04378873618970628,
+      "grad_norm": 0.6783929467201233,
+      "learning_rate": 2.6253031527890058e-05,
+      "loss": 3.235,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04715710051199138,
+      "grad_norm": 0.6816830635070801,
+      "learning_rate": 2.8274050121261112e-05,
+      "loss": 3.1325,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05052546483427647,
+      "grad_norm": 0.6334635615348816,
+      "learning_rate": 3.029506871463217e-05,
+      "loss": 3.0338,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05389382915656157,
+      "grad_norm": 0.6427187323570251,
+      "learning_rate": 3.2316087308003235e-05,
+      "loss": 2.9431,
+      "step": 1600
+    },
+    {
+      "epoch": 0.057262193478846674,
+      "grad_norm": 1.2221870422363281,
+      "learning_rate": 3.433710590137429e-05,
+      "loss": 2.8522,
+      "step": 1700
+    },
+    {
+      "epoch": 0.060630557801131774,
+      "grad_norm": 1.4814372062683105,
+      "learning_rate": 3.635812449474535e-05,
+      "loss": 2.7769,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06399892212341687,
+      "grad_norm": 1.1986867189407349,
+      "learning_rate": 3.837914308811641e-05,
+      "loss": 2.7072,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06736728644570197,
+      "grad_norm": 1.0901182889938354,
+      "learning_rate": 4.0400161681487466e-05,
+      "loss": 2.6491,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07073565076798706,
+      "grad_norm": 1.8428642749786377,
+      "learning_rate": 4.2421180274858524e-05,
+      "loss": 2.5944,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07410401509027216,
+      "grad_norm": 1.2476333379745483,
+      "learning_rate": 4.444219886822959e-05,
+      "loss": 2.5473,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07747237941255726,
+      "grad_norm": 1.3376599550247192,
+      "learning_rate": 4.6463217461600646e-05,
+      "loss": 2.5115,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08084074373484236,
+      "grad_norm": 2.028588056564331,
+      "learning_rate": 4.84842360549717e-05,
+      "loss": 2.4653,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08420910805712746,
+      "grad_norm": 1.5413949489593506,
+      "learning_rate": 5.0505254648342755e-05,
+      "loss": 2.4203,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08757747237941256,
+      "grad_norm": 1.3644788265228271,
+      "learning_rate": 5.252627324171382e-05,
+      "loss": 2.3903,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09094583670169766,
+      "grad_norm": 1.130286693572998,
+      "learning_rate": 5.454729183508488e-05,
+      "loss": 2.3527,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09431420102398276,
+      "grad_norm": 1.6935234069824219,
+      "learning_rate": 5.6568310428455935e-05,
+      "loss": 2.3229,
+      "step": 2800
+    },
+    {
+      "epoch": 0.09768256534626785,
+      "grad_norm": 1.9905304908752441,
+      "learning_rate": 5.8589329021827e-05,
+      "loss": 2.2926,
+      "step": 2900
+    },
+    {
+      "epoch": 0.10105092966855295,
+      "grad_norm": 1.4488565921783447,
+      "learning_rate": 6.061034761519806e-05,
+      "loss": 2.2595,
+      "step": 3000
+    },
+    {
+      "epoch": 0.10441929399083805,
+      "grad_norm": 1.6210366487503052,
+      "learning_rate": 6.263136620856912e-05,
+      "loss": 2.2317,
+      "step": 3100
+    },
+    {
+      "epoch": 0.10778765831312315,
+      "grad_norm": 1.6805219650268555,
+      "learning_rate": 6.465238480194017e-05,
+      "loss": 2.1987,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11115602263540825,
+      "grad_norm": 1.6502385139465332,
+      "learning_rate": 6.667340339531123e-05,
+      "loss": 2.181,
+      "step": 3300
+    },
+    {
+      "epoch": 0.11452438695769335,
+      "grad_norm": 1.831292986869812,
+      "learning_rate": 6.869442198868228e-05,
+      "loss": 2.1532,
+      "step": 3400
+    },
+    {
+      "epoch": 0.11789275127997845,
+      "grad_norm": 1.9188601970672607,
+      "learning_rate": 7.071544058205335e-05,
+      "loss": 2.1127,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12126111560226355,
+      "grad_norm": 1.6280624866485596,
+      "learning_rate": 7.273645917542441e-05,
+      "loss": 2.0921,
+      "step": 3600
+    },
+    {
+      "epoch": 0.12462947992454863,
+      "grad_norm": 1.5468984842300415,
+      "learning_rate": 7.475747776879546e-05,
+      "loss": 2.0626,
+      "step": 3700
+    },
+    {
+      "epoch": 0.12799784424683375,
+      "grad_norm": 1.7183716297149658,
+      "learning_rate": 7.677849636216651e-05,
+      "loss": 2.0302,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13136620856911885,
+      "grad_norm": 1.8704299926757812,
+      "learning_rate": 7.879951495553758e-05,
+      "loss": 1.9946,
+      "step": 3900
+    },
+    {
+      "epoch": 0.13473457289140395,
+      "grad_norm": 1.719117522239685,
+      "learning_rate": 8.082053354890864e-05,
+      "loss": 1.955,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13810293721368902,
+      "grad_norm": 1.8330260515213013,
+      "learning_rate": 8.28415521422797e-05,
+      "loss": 1.9137,
+      "step": 4100
+    },
+    {
+      "epoch": 0.14147130153597412,
+      "grad_norm": 2.341217517852783,
+      "learning_rate": 8.486257073565076e-05,
+      "loss": 1.8765,
+      "step": 4200
+    },
+    {
+      "epoch": 0.14483966585825922,
+      "grad_norm": 1.792738914489746,
+      "learning_rate": 8.688358932902182e-05,
+      "loss": 1.837,
+      "step": 4300
+    },
+    {
+      "epoch": 0.14820803018054432,
+      "grad_norm": 2.237147092819214,
+      "learning_rate": 8.890460792239287e-05,
+      "loss": 1.7995,
+      "step": 4400
+    },
+    {
+      "epoch": 0.15157639450282942,
+      "grad_norm": 1.8367396593093872,
+      "learning_rate": 9.092562651576394e-05,
+      "loss": 1.7706,
+      "step": 4500
+    },
+    {
+      "epoch": 0.15494475882511452,
+      "grad_norm": 2.0587222576141357,
+      "learning_rate": 9.2946645109135e-05,
+      "loss": 1.7431,
+      "step": 4600
+    },
+    {
+      "epoch": 0.15831312314739962,
+      "grad_norm": 1.8988635540008545,
+      "learning_rate": 9.496766370250605e-05,
+      "loss": 1.7179,
+      "step": 4700
+    },
+    {
+      "epoch": 0.16168148746968472,
+      "grad_norm": 2.192547559738159,
+      "learning_rate": 9.698868229587712e-05,
+      "loss": 1.6932,
+      "step": 4800
+    },
+    {
+      "epoch": 0.16504985179196982,
+      "grad_norm": 1.6913732290267944,
+      "learning_rate": 9.900970088924818e-05,
+      "loss": 1.6705,
+      "step": 4900
+    },
+    {
+      "epoch": 0.16841821611425492,
+      "grad_norm": 1.705934762954712,
+      "learning_rate": 0.00010103071948261923,
+      "loss": 1.6509,
+      "step": 5000
+    },
+    {
+      "epoch": 0.17178658043654002,
+      "grad_norm": 1.9764398336410522,
+      "learning_rate": 0.0001030517380759903,
+      "loss": 1.6212,
+      "step": 5100
+    },
+    {
+      "epoch": 0.17515494475882512,
+      "grad_norm": 2.109279155731201,
+      "learning_rate": 0.00010507275666936134,
+      "loss": 1.6087,
+      "step": 5200
+    },
+    {
+      "epoch": 0.17852330908111022,
+      "grad_norm": 2.036076784133911,
+      "learning_rate": 0.0001070937752627324,
+      "loss": 1.5878,
+      "step": 5300
+    },
+    {
+      "epoch": 0.18189167340339532,
+      "grad_norm": 2.1286652088165283,
+      "learning_rate": 0.00010911479385610347,
+      "loss": 1.5641,
+      "step": 5400
+    },
+    {
+      "epoch": 0.18526003772568042,
+      "grad_norm": 2.04007625579834,
+      "learning_rate": 0.00011113581244947452,
+      "loss": 1.5462,
+      "step": 5500
+    },
+    {
+      "epoch": 0.18862840204796552,
+      "grad_norm": 1.9469410181045532,
+      "learning_rate": 0.00011315683104284558,
+      "loss": 1.528,
+      "step": 5600
+    },
+    {
+      "epoch": 0.1919967663702506,
+      "grad_norm": 1.9078123569488525,
+      "learning_rate": 0.00011517784963621665,
+      "loss": 1.5101,
+      "step": 5700
+    },
+    {
+      "epoch": 0.1953651306925357,
+      "grad_norm": 2.1443777084350586,
+      "learning_rate": 0.0001171988682295877,
+      "loss": 1.4885,
+      "step": 5800
+    },
+    {
+      "epoch": 0.1987334950148208,
+      "grad_norm": 1.8993617296218872,
+      "learning_rate": 0.00011921988682295876,
+      "loss": 1.478,
+      "step": 5900
+    },
+    {
+      "epoch": 0.2021018593371059,
+      "grad_norm": 1.7812656164169312,
+      "learning_rate": 0.00012124090541632983,
+      "loss": 1.4621,
+      "step": 6000
+    },
+    {
+      "epoch": 0.205470223659391,
+      "grad_norm": 1.7858940362930298,
+      "learning_rate": 0.0001232619240097009,
+      "loss": 1.4494,
+      "step": 6100
+    },
+    {
+      "epoch": 0.2088385879816761,
+      "grad_norm": 1.9336419105529785,
+      "learning_rate": 0.00012528294260307193,
+      "loss": 1.4393,
+      "step": 6200
+    },
+    {
+      "epoch": 0.2122069523039612,
+      "grad_norm": 1.85440993309021,
+      "learning_rate": 0.000127303961196443,
+      "loss": 1.43,
+      "step": 6300
+    },
+    {
+      "epoch": 0.2155753166262463,
+      "grad_norm": 1.583737850189209,
+      "learning_rate": 0.00012932497978981406,
+      "loss": 1.4198,
+      "step": 6400
+    },
+    {
+      "epoch": 0.2189436809485314,
+      "grad_norm": 2.076510429382324,
+      "learning_rate": 0.00013134599838318512,
+      "loss": 1.4034,
+      "step": 6500
+    },
+    {
+      "epoch": 0.2223120452708165,
+      "grad_norm": 1.6790341138839722,
+      "learning_rate": 0.0001333670169765562,
+      "loss": 1.3937,
+      "step": 6600
+    },
+    {
+      "epoch": 0.2256804095931016,
+      "grad_norm": 1.9147748947143555,
+      "learning_rate": 0.00013538803556992725,
+      "loss": 1.3818,
+      "step": 6700
+    },
+    {
+      "epoch": 0.2290487739153867,
+      "grad_norm": 1.6485368013381958,
+      "learning_rate": 0.0001374090541632983,
+      "loss": 1.3758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.2324171382376718,
+      "grad_norm": 2.087151527404785,
+      "learning_rate": 0.00013943007275666935,
+      "loss": 1.3696,
+      "step": 6900
+    },
+    {
+      "epoch": 0.2357855025599569,
+      "grad_norm": 1.7565312385559082,
+      "learning_rate": 0.0001414510913500404,
+      "loss": 1.3579,
+      "step": 7000
+    },
+    {
+      "epoch": 0.239153866882242,
+      "grad_norm": 1.83383047580719,
+      "learning_rate": 0.00014347210994341146,
+      "loss": 1.3532,
+      "step": 7100
+    },
+    {
+      "epoch": 0.2425222312045271,
+      "grad_norm": 1.8977510929107666,
+      "learning_rate": 0.00014549312853678252,
+      "loss": 1.3465,
+      "step": 7200
+    },
+    {
+      "epoch": 0.24589059552681217,
+      "grad_norm": 2.1144802570343018,
+      "learning_rate": 0.00014751414713015358,
+      "loss": 1.3389,
+      "step": 7300
+    },
+    {
+      "epoch": 0.24925895984909727,
+      "grad_norm": 1.6824164390563965,
+      "learning_rate": 0.00014953516572352465,
+      "loss": 1.3283,
+      "step": 7400
+    },
+    {
+      "epoch": 0.2526273241713824,
+      "grad_norm": 1.8628549575805664,
+      "learning_rate": 0.00015155618431689571,
+      "loss": 1.3225,
+      "step": 7500
+    },
+    {
+      "epoch": 0.2559956884936675,
+      "grad_norm": 1.651308298110962,
+      "learning_rate": 0.00015357720291026675,
+      "loss": 1.3153,
+      "step": 7600
+    },
+    {
+      "epoch": 0.2593640528159526,
+      "grad_norm": 1.6041486263275146,
+      "learning_rate": 0.00015559822150363782,
+      "loss": 1.3056,
+      "step": 7700
+    },
+    {
+      "epoch": 0.2627324171382377,
+      "grad_norm": 1.9386959075927734,
+      "learning_rate": 0.00015761924009700888,
+      "loss": 1.3046,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2661007814605228,
+      "grad_norm": 1.8345019817352295,
+      "learning_rate": 0.00015964025869037995,
+      "loss": 1.2884,
+      "step": 7900
+    },
+    {
+      "epoch": 0.2694691457828079,
+      "grad_norm": 1.4670854806900024,
+      "learning_rate": 0.000161661277283751,
+      "loss": 1.286,
+      "step": 8000
+    },
+    {
+      "epoch": 0.27283751010509294,
+      "grad_norm": 1.8858684301376343,
+      "learning_rate": 0.00016368229587712205,
+      "loss": 1.2774,
+      "step": 8100
+    },
+    {
+      "epoch": 0.27620587442737804,
+      "grad_norm": 1.8875221014022827,
+      "learning_rate": 0.0001657033144704931,
+      "loss": 1.2736,
+      "step": 8200
+    },
+    {
+      "epoch": 0.27957423874966314,
+      "grad_norm": 1.7752630710601807,
+      "learning_rate": 0.00016772433306386418,
+      "loss": 1.2746,
+      "step": 8300
+    },
+    {
+      "epoch": 0.28294260307194824,
+      "grad_norm": 1.5532513856887817,
+      "learning_rate": 0.00016974535165723524,
+      "loss": 1.267,
+      "step": 8400
+    },
+    {
+      "epoch": 0.28631096739423334,
+      "grad_norm": 1.6465749740600586,
+      "learning_rate": 0.0001717663702506063,
+      "loss": 1.2599,
+      "step": 8500
+    },
+    {
+      "epoch": 0.28967933171651844,
+      "grad_norm": 1.5771738290786743,
+      "learning_rate": 0.00017378738884397737,
+      "loss": 1.2557,
+      "step": 8600
+    },
+    {
+      "epoch": 0.29304769603880354,
+      "grad_norm": 1.4705991744995117,
+      "learning_rate": 0.0001758084074373484,
+      "loss": 1.252,
+      "step": 8700
+    },
+    {
+      "epoch": 0.29641606036108864,
+      "grad_norm": 1.489914059638977,
+      "learning_rate": 0.00017782942603071947,
+      "loss": 1.2425,
+      "step": 8800
+    },
+    {
+      "epoch": 0.29978442468337374,
+      "grad_norm": 1.5901821851730347,
+      "learning_rate": 0.00017985044462409054,
+      "loss": 1.2341,
+      "step": 8900
+    },
+    {
+      "epoch": 0.30315278900565884,
+      "grad_norm": 1.5143710374832153,
+      "learning_rate": 0.0001818714632174616,
+      "loss": 1.2309,
+      "step": 9000
+    },
+    {
+      "epoch": 0.30652115332794394,
+      "grad_norm": 1.5409547090530396,
+      "learning_rate": 0.00018389248181083267,
+      "loss": 1.2258,
+      "step": 9100
+    },
+    {
+      "epoch": 0.30988951765022904,
+      "grad_norm": 1.4094816446304321,
+      "learning_rate": 0.00018591350040420368,
+      "loss": 1.2197,
+      "step": 9200
+    },
+    {
+      "epoch": 0.31325788197251414,
+      "grad_norm": 1.6748660802841187,
+      "learning_rate": 0.00018793451899757474,
+      "loss": 1.2176,
+      "step": 9300
+    },
+    {
+      "epoch": 0.31662624629479924,
+      "grad_norm": 1.587318778038025,
+      "learning_rate": 0.0001899555375909458,
+      "loss": 1.2091,
+      "step": 9400
+    },
+    {
+      "epoch": 0.31999461061708434,
+      "grad_norm": 1.5422818660736084,
+      "learning_rate": 0.00019197655618431687,
+      "loss": 1.2079,
+      "step": 9500
+    },
+    {
+      "epoch": 0.32336297493936944,
+      "grad_norm": 1.30134117603302,
+      "learning_rate": 0.00019399757477768793,
+      "loss": 1.2017,
+      "step": 9600
+    },
+    {
+      "epoch": 0.32673133926165454,
+      "grad_norm": 1.368249773979187,
+      "learning_rate": 0.00019601859337105897,
+      "loss": 1.2026,
+      "step": 9700
+    },
+    {
+      "epoch": 0.33009970358393964,
+      "grad_norm": 1.5968406200408936,
+      "learning_rate": 0.00019803961196443004,
+      "loss": 1.1902,
+      "step": 9800
+    },
+    {
+      "epoch": 0.33346806790622474,
+      "grad_norm": 1.435455083847046,
+      "learning_rate": 0.0002000606305578011,
+      "loss": 1.1894,
+      "step": 9900
+    },
+    {
+      "epoch": 0.33683643222850984,
+      "grad_norm": 1.4132752418518066,
+      "learning_rate": 0.00020208164915117217,
+      "loss": 1.1841,
+      "step": 10000
+    },
+    {
+      "epoch": 0.34020479655079494,
+      "grad_norm": 1.5038225650787354,
+      "learning_rate": 0.00020410266774454323,
+      "loss": 1.1859,
+      "step": 10100
+    },
+    {
+      "epoch": 0.34357316087308004,
+      "grad_norm": 1.4201886653900146,
+      "learning_rate": 0.0002061236863379143,
+      "loss": 1.1743,
+      "step": 10200
+    },
+    {
+      "epoch": 0.34694152519536514,
+      "grad_norm": 1.3544988632202148,
+      "learning_rate": 0.00020814470493128533,
+      "loss": 1.1724,
+      "step": 10300
+    },
+    {
+      "epoch": 0.35030988951765024,
+      "grad_norm": 1.6340460777282715,
+      "learning_rate": 0.0002101657235246564,
+      "loss": 1.1679,
+      "step": 10400
+    },
+    {
+      "epoch": 0.35367825383993534,
+      "grad_norm": 1.3137534856796265,
+      "learning_rate": 0.00021218674211802746,
+      "loss": 1.1602,
+      "step": 10500
+    },
+    {
+      "epoch": 0.35704661816222044,
+      "grad_norm": 1.3838586807250977,
+      "learning_rate": 0.00021420776071139853,
+      "loss": 1.1602,
+      "step": 10600
+    },
+    {
+      "epoch": 0.36041498248450554,
+      "grad_norm": 1.46292245388031,
+      "learning_rate": 0.0002162287793047696,
+      "loss": 1.154,
+      "step": 10700
+    },
+    {
+      "epoch": 0.36378334680679064,
+      "grad_norm": 1.2189207077026367,
+      "learning_rate": 0.00021824979789814066,
+      "loss": 1.1518,
+      "step": 10800
+    },
+    {
+      "epoch": 0.36715171112907574,
+      "grad_norm": 1.3567001819610596,
+      "learning_rate": 0.0002202708164915117,
+      "loss": 1.1437,
+      "step": 10900
+    },
+    {
+      "epoch": 0.37052007545136084,
+      "grad_norm": 1.3379132747650146,
+      "learning_rate": 0.00022229183508488276,
+      "loss": 1.1418,
+      "step": 11000
+    },
+    {
+      "epoch": 0.37388843977364594,
+      "grad_norm": 1.2323216199874878,
+      "learning_rate": 0.00022431285367825382,
+      "loss": 1.1382,
+      "step": 11100
+    },
+    {
+      "epoch": 0.37725680409593104,
+      "grad_norm": 1.9002209901809692,
+      "learning_rate": 0.0002263338722716249,
+      "loss": 1.1323,
+      "step": 11200
+    },
+    {
+      "epoch": 0.3806251684182161,
+      "grad_norm": 1.3124207258224487,
+      "learning_rate": 0.00022835489086499595,
+      "loss": 1.1336,
+      "step": 11300
+    },
+    {
+      "epoch": 0.3839935327405012,
+      "grad_norm": 1.3445236682891846,
+      "learning_rate": 0.000230375909458367,
+      "loss": 1.1259,
+      "step": 11400
+    },
+    {
+      "epoch": 0.3873618970627863,
+      "grad_norm": 1.2994790077209473,
+      "learning_rate": 0.00023239692805173805,
+      "loss": 1.1298,
+      "step": 11500
+    },
+    {
+      "epoch": 0.3907302613850714,
+      "grad_norm": 1.2289458513259888,
+      "learning_rate": 0.00023441794664510912,
+      "loss": 1.1226,
+      "step": 11600
+    },
+    {
+      "epoch": 0.3940986257073565,
+      "grad_norm": 1.1772109270095825,
+      "learning_rate": 0.00023643896523848018,
+      "loss": 1.1215,
+      "step": 11700
+    },
+    {
+      "epoch": 0.3974669900296416,
+      "grad_norm": 1.3592746257781982,
+      "learning_rate": 0.00023845998383185125,
+      "loss": 1.1179,
+      "step": 11800
+    },
+    {
+      "epoch": 0.4008353543519267,
+      "grad_norm": 1.1571407318115234,
+      "learning_rate": 0.0002404810024252223,
+      "loss": 1.1124,
+      "step": 11900
+    },
+    {
+      "epoch": 0.4042037186742118,
+      "grad_norm": 1.3592592477798462,
+      "learning_rate": 0.00024250202101859335,
+      "loss": 1.1083,
+      "step": 12000
+    },
+    {
+      "epoch": 0.4075720829964969,
+      "grad_norm": 1.2856664657592773,
+      "learning_rate": 0.00024452303961196444,
+      "loss": 1.1083,
+      "step": 12100
+    },
+    {
+      "epoch": 0.410940447318782,
+      "grad_norm": 1.2512900829315186,
+      "learning_rate": 0.0002465440582053355,
+      "loss": 1.0984,
+      "step": 12200
+    },
+    {
+      "epoch": 0.4143088116410671,
+      "grad_norm": 1.239823818206787,
+      "learning_rate": 0.0002485650767987065,
+      "loss": 1.0984,
+      "step": 12300
+    },
+    {
+      "epoch": 0.4176771759633522,
+      "grad_norm": 1.4879858493804932,
+      "learning_rate": 0.0002505860953920776,
+      "loss": 1.0962,
+      "step": 12400
+    },
+    {
+      "epoch": 0.4210455402856373,
+      "grad_norm": 1.1708803176879883,
+      "learning_rate": 0.00025260711398544865,
+      "loss": 1.0914,
+      "step": 12500
+    },
+    {
+      "epoch": 0.4244139046079224,
+      "grad_norm": 1.092463731765747,
+      "learning_rate": 0.0002546281325788197,
+      "loss": 1.089,
+      "step": 12600
+    },
+    {
+      "epoch": 0.4277822689302075,
+      "grad_norm": 1.2004188299179077,
+      "learning_rate": 0.0002566491511721908,
+      "loss": 1.0884,
+      "step": 12700
+    },
+    {
+      "epoch": 0.4311506332524926,
+      "grad_norm": 1.1339149475097656,
+      "learning_rate": 0.0002586701697655618,
+      "loss": 1.0857,
+      "step": 12800
+    },
+    {
+      "epoch": 0.4345189975747777,
+      "grad_norm": 1.3253908157348633,
+      "learning_rate": 0.00026069118835893285,
+      "loss": 1.0821,
+      "step": 12900
+    },
+    {
+      "epoch": 0.4378873618970628,
+      "grad_norm": 1.2655895948410034,
+      "learning_rate": 0.00026271220695230394,
+      "loss": 1.0775,
+      "step": 13000
+    },
+    {
+      "epoch": 0.4412557262193479,
+      "grad_norm": 1.4968757629394531,
+      "learning_rate": 0.000264733225545675,
+      "loss": 1.0764,
+      "step": 13100
+    },
+    {
+      "epoch": 0.444624090541633,
+      "grad_norm": 1.200173020362854,
+      "learning_rate": 0.00026675424413904607,
+      "loss": 1.0768,
+      "step": 13200
+    },
+    {
+      "epoch": 0.4479924548639181,
+      "grad_norm": 1.3085741996765137,
+      "learning_rate": 0.0002687752627324171,
+      "loss": 1.0739,
+      "step": 13300
+    },
+    {
+      "epoch": 0.4513608191862032,
+      "grad_norm": 1.181569218635559,
+      "learning_rate": 0.00027079628132578815,
+      "loss": 1.0685,
+      "step": 13400
+    },
+    {
+      "epoch": 0.4547291835084883,
+      "grad_norm": 1.2471662759780884,
+      "learning_rate": 0.00027281729991915924,
+      "loss": 1.0668,
+      "step": 13500
+    },
+    {
+      "epoch": 0.4580975478307734,
+      "grad_norm": 1.0714460611343384,
+      "learning_rate": 0.0002748383185125303,
+      "loss": 1.0687,
+      "step": 13600
+    },
+    {
+      "epoch": 0.4614659121530585,
+      "grad_norm": 1.2705806493759155,
+      "learning_rate": 0.00027685933710590137,
+      "loss": 1.0602,
+      "step": 13700
+    },
+    {
+      "epoch": 0.4648342764753436,
+      "grad_norm": 1.199216365814209,
+      "learning_rate": 0.0002788803556992724,
+      "loss": 1.0572,
+      "step": 13800
+    },
+    {
+      "epoch": 0.4682026407976287,
+      "grad_norm": 1.1781370639801025,
+      "learning_rate": 0.00028090137429264344,
+      "loss": 1.056,
+      "step": 13900
+    },
+    {
+      "epoch": 0.4715710051199138,
+      "grad_norm": 1.1787018775939941,
+      "learning_rate": 0.00028292239288601453,
+      "loss": 1.0582,
+      "step": 14000
+    },
+    {
+      "epoch": 0.4749393694421989,
+      "grad_norm": 1.0408787727355957,
+      "learning_rate": 0.00028494341147938557,
+      "loss": 1.0509,
+      "step": 14100
+    },
+    {
+      "epoch": 0.478307733764484,
+      "grad_norm": 1.2054550647735596,
+      "learning_rate": 0.00028696443007275666,
+      "loss": 1.0496,
+      "step": 14200
+    },
+    {
+      "epoch": 0.4816760980867691,
+      "grad_norm": 1.059328317642212,
+      "learning_rate": 0.0002889854486661277,
+      "loss": 1.0484,
+      "step": 14300
+    },
+    {
+      "epoch": 0.4850444624090542,
+      "grad_norm": 1.0218919515609741,
+      "learning_rate": 0.0002910064672594988,
+      "loss": 1.0443,
+      "step": 14400
+    },
+    {
+      "epoch": 0.4884128267313393,
+      "grad_norm": 1.569550633430481,
+      "learning_rate": 0.00029302748585286983,
+      "loss": 1.0439,
+      "step": 14500
+    },
+    {
+      "epoch": 0.49178119105362433,
+      "grad_norm": 1.2660326957702637,
+      "learning_rate": 0.00029504850444624087,
+      "loss": 1.0412,
+      "step": 14600
+    },
+    {
+      "epoch": 0.49514955537590943,
+      "grad_norm": 0.9795782566070557,
+      "learning_rate": 0.00029706952303961196,
+      "loss": 1.0362,
+      "step": 14700
+    },
+    {
+      "epoch": 0.49851791969819453,
+      "grad_norm": 1.0511739253997803,
+      "learning_rate": 0.000299090541632983,
+      "loss": 1.0379,
+      "step": 14800
+    },
+    {
+      "epoch": 0.5018862840204796,
+      "grad_norm": 1.077668309211731,
+      "learning_rate": 0.00029987649330818285,
+      "loss": 1.034,
+      "step": 14900
+    },
+    {
+      "epoch": 0.5052546483427648,
+      "grad_norm": 0.9709302186965942,
+      "learning_rate": 0.0002996519356866972,
+      "loss": 1.0297,
+      "step": 15000
+    },
+    {
+      "epoch": 0.5086230126650498,
+      "grad_norm": 1.038855791091919,
+      "learning_rate": 0.0002994273780652115,
+      "loss": 1.0295,
+      "step": 15100
+    },
+    {
+      "epoch": 0.511991376987335,
+      "grad_norm": 1.1095309257507324,
+      "learning_rate": 0.00029920282044372586,
+      "loss": 1.0241,
+      "step": 15200
+    },
+    {
+      "epoch": 0.51535974130962,
+      "grad_norm": 1.0058341026306152,
+      "learning_rate": 0.00029897826282224014,
+      "loss": 1.0233,
+      "step": 15300
+    },
+    {
+      "epoch": 0.5187281056319052,
+      "grad_norm": 1.1029912233352661,
+      "learning_rate": 0.00029875370520075447,
+      "loss": 1.0253,
+      "step": 15400
+    },
+    {
+      "epoch": 0.5220964699541902,
+      "grad_norm": 1.1280447244644165,
+      "learning_rate": 0.0002985291475792688,
+      "loss": 1.02,
+      "step": 15500
+    },
+    {
+      "epoch": 0.5254648342764754,
+      "grad_norm": 1.0095000267028809,
+      "learning_rate": 0.00029830458995778314,
+      "loss": 1.0185,
+      "step": 15600
+    },
+    {
+      "epoch": 0.5288331985987604,
+      "grad_norm": 1.071540117263794,
+      "learning_rate": 0.0002980800323362975,
+      "loss": 1.0133,
+      "step": 15700
+    },
+    {
+      "epoch": 0.5322015629210456,
+      "grad_norm": 0.9709872007369995,
+      "learning_rate": 0.0002978554747148118,
+      "loss": 1.0189,
+      "step": 15800
+    },
+    {
+      "epoch": 0.5355699272433306,
+      "grad_norm": 1.1805214881896973,
+      "learning_rate": 0.0002976309170933261,
+      "loss": 1.0145,
+      "step": 15900
+    },
+    {
+      "epoch": 0.5389382915656158,
+      "grad_norm": 1.1302651166915894,
+      "learning_rate": 0.0002974063594718405,
+      "loss": 1.0088,
+      "step": 16000
+    },
+    {
+      "epoch": 0.5423066558879008,
+      "grad_norm": 1.280207872390747,
+      "learning_rate": 0.00029718180185035476,
+      "loss": 1.0126,
+      "step": 16100
+    },
+    {
+      "epoch": 0.5456750202101859,
+      "grad_norm": 1.024566888809204,
+      "learning_rate": 0.0002969572442288691,
+      "loss": 1.006,
+      "step": 16200
+    },
+    {
+      "epoch": 0.549043384532471,
+      "grad_norm": 1.192209243774414,
+      "learning_rate": 0.00029673268660738343,
+      "loss": 1.0076,
+      "step": 16300
+    },
+    {
+      "epoch": 0.5524117488547561,
+      "grad_norm": 0.9792165756225586,
+      "learning_rate": 0.0002965081289858977,
+      "loss": 1.005,
+      "step": 16400
+    },
+    {
+      "epoch": 0.5557801131770412,
+      "grad_norm": 1.6060813665390015,
+      "learning_rate": 0.0002962835713644121,
+      "loss": 0.9999,
+      "step": 16500
+    },
+    {
+      "epoch": 0.5591484774993263,
+      "grad_norm": 0.9630849957466125,
+      "learning_rate": 0.0002960590137429264,
+      "loss": 1.0024,
+      "step": 16600
+    },
+    {
+      "epoch": 0.5625168418216114,
+      "grad_norm": 0.9396387934684753,
+      "learning_rate": 0.0002958344561214407,
+      "loss": 1.0004,
+      "step": 16700
+    },
+    {
+      "epoch": 0.5658852061438965,
+      "grad_norm": 0.9031047821044922,
+      "learning_rate": 0.00029560989849995506,
+      "loss": 0.9984,
+      "step": 16800
+    },
+    {
+      "epoch": 0.5692535704661816,
+      "grad_norm": 1.0625028610229492,
+      "learning_rate": 0.0002953853408784694,
+      "loss": 0.9932,
+      "step": 16900
+    },
+    {
+      "epoch": 0.5726219347884667,
+      "grad_norm": 0.9275569319725037,
+      "learning_rate": 0.0002951607832569837,
+      "loss": 0.9925,
+      "step": 17000
+    },
+    {
+      "epoch": 0.5759902991107518,
+      "grad_norm": 0.8847247362136841,
+      "learning_rate": 0.00029493622563549806,
+      "loss": 0.9909,
+      "step": 17100
+    },
+    {
+      "epoch": 0.5793586634330369,
+      "grad_norm": 0.9581294059753418,
+      "learning_rate": 0.0002947116680140124,
+      "loss": 0.9923,
+      "step": 17200
+    },
+    {
+      "epoch": 0.582727027755322,
+      "grad_norm": 1.003164529800415,
+      "learning_rate": 0.0002944871103925267,
+      "loss": 0.9914,
+      "step": 17300
+    },
+    {
+      "epoch": 0.5860953920776071,
+      "grad_norm": 1.010026454925537,
+      "learning_rate": 0.000294262552771041,
+      "loss": 0.9854,
+      "step": 17400
+    },
+    {
+      "epoch": 0.5894637563998922,
+      "grad_norm": 0.9266247153282166,
+      "learning_rate": 0.00029403799514955535,
+      "loss": 0.9879,
+      "step": 17500
+    },
+    {
+      "epoch": 0.5928321207221773,
+      "grad_norm": 0.9909249544143677,
+      "learning_rate": 0.0002938134375280697,
+      "loss": 0.9876,
+      "step": 17600
+    },
+    {
+      "epoch": 0.5962004850444624,
+      "grad_norm": 0.8353651165962219,
+      "learning_rate": 0.000293588879906584,
+      "loss": 0.9846,
+      "step": 17700
+    },
+    {
+      "epoch": 0.5995688493667475,
+      "grad_norm": 0.9142294526100159,
+      "learning_rate": 0.00029336432228509835,
+      "loss": 0.9877,
+      "step": 17800
+    },
+    {
+      "epoch": 0.6029372136890326,
+      "grad_norm": 0.9139926433563232,
+      "learning_rate": 0.00029313976466361263,
+      "loss": 0.9831,
+      "step": 17900
+    },
+    {
+      "epoch": 0.6063055780113177,
+      "grad_norm": 0.8871977925300598,
+      "learning_rate": 0.000292915207042127,
+      "loss": 0.9817,
+      "step": 18000
+    },
+    {
+      "epoch": 0.6096739423336028,
+      "grad_norm": 0.9932221174240112,
+      "learning_rate": 0.0002926906494206413,
+      "loss": 0.979,
+      "step": 18100
+    },
+    {
+      "epoch": 0.6130423066558879,
+      "grad_norm": 0.9240766167640686,
+      "learning_rate": 0.00029246609179915564,
+      "loss": 0.9774,
+      "step": 18200
+    },
+    {
+      "epoch": 0.616410670978173,
+      "grad_norm": 0.932101845741272,
+      "learning_rate": 0.00029224153417767,
+      "loss": 0.9797,
+      "step": 18300
+    },
+    {
+      "epoch": 0.6197790353004581,
+      "grad_norm": 0.9871794581413269,
+      "learning_rate": 0.00029201697655618426,
+      "loss": 0.9809,
+      "step": 18400
+    },
+    {
+      "epoch": 0.6231473996227432,
+      "grad_norm": 0.8771729469299316,
+      "learning_rate": 0.00029179241893469865,
+      "loss": 0.9761,
+      "step": 18500
+    },
+    {
+      "epoch": 0.6265157639450283,
+      "grad_norm": 1.103968858718872,
+      "learning_rate": 0.0002915678613132129,
+      "loss": 0.9779,
+      "step": 18600
+    },
+    {
+      "epoch": 0.6298841282673134,
+      "grad_norm": 1.0848268270492554,
+      "learning_rate": 0.00029134330369172726,
+      "loss": 0.9722,
+      "step": 18700
+    },
+    {
+      "epoch": 0.6332524925895985,
+      "grad_norm": 0.8907010555267334,
+      "learning_rate": 0.0002911187460702416,
+      "loss": 0.9709,
+      "step": 18800
+    },
+    {
+      "epoch": 0.6366208569118836,
+      "grad_norm": 0.9081377983093262,
+      "learning_rate": 0.00029089418844875593,
+      "loss": 0.9681,
+      "step": 18900
+    },
+    {
+      "epoch": 0.6399892212341687,
+      "grad_norm": 0.873905599117279,
+      "learning_rate": 0.00029066963082727027,
+      "loss": 0.9735,
+      "step": 19000
+    },
+    {
+      "epoch": 0.6433575855564538,
+      "grad_norm": 0.9111950397491455,
+      "learning_rate": 0.0002904450732057846,
+      "loss": 0.97,
+      "step": 19100
+    },
+    {
+      "epoch": 0.6467259498787389,
+      "grad_norm": 0.9769060611724854,
+      "learning_rate": 0.0002902205155842989,
+      "loss": 0.9688,
+      "step": 19200
+    },
+    {
+      "epoch": 0.650094314201024,
+      "grad_norm": 1.0822559595108032,
+      "learning_rate": 0.0002899959579628132,
+      "loss": 0.968,
+      "step": 19300
+    },
+    {
+      "epoch": 0.6534626785233091,
+      "grad_norm": 0.8573871850967407,
+      "learning_rate": 0.00028977140034132755,
+      "loss": 0.967,
+      "step": 19400
+    },
+    {
+      "epoch": 0.6568310428455941,
+      "grad_norm": 0.989267885684967,
+      "learning_rate": 0.0002895468427198419,
+      "loss": 0.9652,
+      "step": 19500
+    },
+    {
+      "epoch": 0.6601994071678793,
+      "grad_norm": 0.9819543361663818,
+      "learning_rate": 0.0002893222850983562,
+      "loss": 0.9597,
+      "step": 19600
+    },
+    {
+      "epoch": 0.6635677714901643,
+      "grad_norm": 0.9204864501953125,
+      "learning_rate": 0.00028909772747687056,
+      "loss": 0.9641,
+      "step": 19700
+    },
+    {
+      "epoch": 0.6669361358124495,
+      "grad_norm": 1.1161561012268066,
+      "learning_rate": 0.0002888731698553849,
+      "loss": 0.9619,
+      "step": 19800
+    },
+    {
+      "epoch": 0.6703045001347345,
+      "grad_norm": 0.8925914764404297,
+      "learning_rate": 0.0002886486122338992,
+      "loss": 0.9599,
+      "step": 19900
+    },
+    {
+      "epoch": 0.6736728644570197,
+      "grad_norm": 0.9228368401527405,
+      "learning_rate": 0.0002884240546124135,
+      "loss": 0.9603,
+      "step": 20000
+    },
+    {
+      "epoch": 0.6770412287793047,
+      "grad_norm": 0.8357170224189758,
+      "learning_rate": 0.00028819949699092785,
+      "loss": 0.9562,
+      "step": 20100
+    },
+    {
+      "epoch": 0.6804095931015899,
+      "grad_norm": 0.9358044266700745,
+      "learning_rate": 0.0002879749393694422,
+      "loss": 0.963,
+      "step": 20200
+    },
+    {
+      "epoch": 0.6837779574238749,
+      "grad_norm": 0.83344966173172,
+      "learning_rate": 0.0002877503817479565,
+      "loss": 0.9555,
+      "step": 20300
+    },
+    {
+      "epoch": 0.6871463217461601,
+      "grad_norm": 0.8138599395751953,
+      "learning_rate": 0.0002875258241264708,
+      "loss": 0.9572,
+      "step": 20400
+    },
+    {
+      "epoch": 0.6905146860684451,
+      "grad_norm": 0.886358380317688,
+      "learning_rate": 0.0002873012665049852,
+      "loss": 0.9525,
+      "step": 20500
+    },
+    {
+      "epoch": 0.6938830503907303,
+      "grad_norm": 0.9100881814956665,
+      "learning_rate": 0.00028707670888349947,
+      "loss": 0.9545,
+      "step": 20600
+    },
+    {
+      "epoch": 0.6972514147130153,
+      "grad_norm": 0.8902551531791687,
+      "learning_rate": 0.0002868521512620138,
+      "loss": 0.9542,
+      "step": 20700
+    },
+    {
+      "epoch": 0.7006197790353005,
+      "grad_norm": 1.24783456325531,
+      "learning_rate": 0.00028662759364052814,
+      "loss": 0.9536,
+      "step": 20800
+    },
+    {
+      "epoch": 0.7039881433575855,
+      "grad_norm": 0.831633448600769,
+      "learning_rate": 0.0002864030360190425,
+      "loss": 0.9526,
+      "step": 20900
+    },
+    {
+      "epoch": 0.7073565076798707,
+      "grad_norm": 0.8300578594207764,
+      "learning_rate": 0.0002861784783975568,
+      "loss": 0.9478,
+      "step": 21000
+    },
+    {
+      "epoch": 0.7107248720021557,
+      "grad_norm": 0.9245336055755615,
+      "learning_rate": 0.0002859539207760711,
+      "loss": 0.9531,
+      "step": 21100
+    },
+    {
+      "epoch": 0.7140932363244409,
+      "grad_norm": 0.8717476725578308,
+      "learning_rate": 0.0002857293631545854,
+      "loss": 0.9487,
+      "step": 21200
+    },
+    {
+      "epoch": 0.7174616006467259,
+      "grad_norm": 0.8705008625984192,
+      "learning_rate": 0.00028550480553309976,
+      "loss": 0.9486,
+      "step": 21300
+    },
+    {
+      "epoch": 0.7208299649690111,
+      "grad_norm": 0.8324209451675415,
+      "learning_rate": 0.0002852802479116141,
+      "loss": 0.9506,
+      "step": 21400
+    },
+    {
+      "epoch": 0.7241983292912961,
+      "grad_norm": 0.9107707738876343,
+      "learning_rate": 0.00028505569029012843,
+      "loss": 0.9465,
+      "step": 21500
+    },
+    {
+      "epoch": 0.7275666936135813,
+      "grad_norm": 0.8204140067100525,
+      "learning_rate": 0.00028483113266864277,
+      "loss": 0.9462,
+      "step": 21600
+    },
+    {
+      "epoch": 0.7309350579358663,
+      "grad_norm": 0.8158605098724365,
+      "learning_rate": 0.00028460657504715705,
+      "loss": 0.9483,
+      "step": 21700
+    },
+    {
+      "epoch": 0.7343034222581515,
+      "grad_norm": 0.8872929811477661,
+      "learning_rate": 0.00028438201742567144,
+      "loss": 0.9461,
+      "step": 21800
+    },
+    {
+      "epoch": 0.7376717865804365,
+      "grad_norm": 0.8722573518753052,
+      "learning_rate": 0.0002841574598041857,
+      "loss": 0.943,
+      "step": 21900
+    },
+    {
+      "epoch": 0.7410401509027217,
+      "grad_norm": 0.8522630929946899,
+      "learning_rate": 0.00028393290218270005,
+      "loss": 0.9449,
+      "step": 22000
+    },
+    {
+      "epoch": 0.7444085152250067,
+      "grad_norm": 0.850837767124176,
+      "learning_rate": 0.0002837083445612144,
+      "loss": 0.9441,
+      "step": 22100
+    },
+    {
+      "epoch": 0.7477768795472919,
+      "grad_norm": 0.7917930483818054,
+      "learning_rate": 0.0002834837869397287,
+      "loss": 0.9442,
+      "step": 22200
+    },
+    {
+      "epoch": 0.7511452438695769,
+      "grad_norm": 0.9168843030929565,
+      "learning_rate": 0.00028325922931824306,
+      "loss": 0.9427,
+      "step": 22300
+    },
+    {
+      "epoch": 0.7545136081918621,
+      "grad_norm": 0.9117637276649475,
+      "learning_rate": 0.00028303467169675734,
+      "loss": 0.9436,
+      "step": 22400
+    },
+    {
+      "epoch": 0.7578819725141471,
+      "grad_norm": 0.8345003724098206,
+      "learning_rate": 0.0002828101140752717,
+      "loss": 0.9394,
+      "step": 22500
+    },
+    {
+      "epoch": 0.7612503368364322,
+      "grad_norm": 0.8290796875953674,
+      "learning_rate": 0.000282585556453786,
+      "loss": 0.938,
+      "step": 22600
+    },
+    {
+      "epoch": 0.7646187011587173,
+      "grad_norm": 0.7990386486053467,
+      "learning_rate": 0.00028236099883230034,
+      "loss": 0.9393,
+      "step": 22700
+    },
+    {
+      "epoch": 0.7679870654810024,
+      "grad_norm": 0.7871207594871521,
+      "learning_rate": 0.0002821364412108147,
+      "loss": 0.9376,
+      "step": 22800
+    },
+    {
+      "epoch": 0.7713554298032875,
+      "grad_norm": 0.7800641059875488,
+      "learning_rate": 0.000281911883589329,
+      "loss": 0.9366,
+      "step": 22900
+    },
+    {
+      "epoch": 0.7747237941255726,
+      "grad_norm": 0.8521484136581421,
+      "learning_rate": 0.00028168732596784335,
+      "loss": 0.9377,
+      "step": 23000
+    },
+    {
+      "epoch": 0.7780921584478577,
+      "grad_norm": 0.8726188540458679,
+      "learning_rate": 0.00028146276834635763,
+      "loss": 0.9344,
+      "step": 23100
+    },
+    {
+      "epoch": 0.7814605227701428,
+      "grad_norm": 0.9499660730361938,
+      "learning_rate": 0.00028123821072487197,
+      "loss": 0.9357,
+      "step": 23200
+    },
+    {
+      "epoch": 0.7848288870924279,
+      "grad_norm": 0.8431583642959595,
+      "learning_rate": 0.0002810136531033863,
+      "loss": 0.9373,
+      "step": 23300
+    },
+    {
+      "epoch": 0.788197251414713,
+      "grad_norm": 0.8140375018119812,
+      "learning_rate": 0.00028078909548190064,
+      "loss": 0.9352,
+      "step": 23400
+    },
+    {
+      "epoch": 0.7915656157369981,
+      "grad_norm": 0.8821849226951599,
+      "learning_rate": 0.00028056453786041497,
+      "loss": 0.9346,
+      "step": 23500
+    },
+    {
+      "epoch": 0.7949339800592832,
+      "grad_norm": 1.0091123580932617,
+      "learning_rate": 0.0002803399802389293,
+      "loss": 0.9328,
+      "step": 23600
+    },
+    {
+      "epoch": 0.7983023443815683,
+      "grad_norm": 1.0306652784347534,
+      "learning_rate": 0.0002801154226174436,
+      "loss": 0.9332,
+      "step": 23700
+    },
+    {
+      "epoch": 0.8016707087038534,
+      "grad_norm": 0.8444597721099854,
+      "learning_rate": 0.000279890864995958,
+      "loss": 0.9372,
+      "step": 23800
+    },
+    {
+      "epoch": 0.8050390730261385,
+      "grad_norm": 0.8706274628639221,
+      "learning_rate": 0.00027966630737447226,
+      "loss": 0.9351,
+      "step": 23900
+    },
+    {
+      "epoch": 0.8084074373484236,
+      "grad_norm": 0.8325883150100708,
+      "learning_rate": 0.0002794417497529866,
+      "loss": 0.9296,
+      "step": 24000
+    },
+    {
+      "epoch": 0.8117758016707087,
+      "grad_norm": 0.7604151964187622,
+      "learning_rate": 0.00027921719213150093,
+      "loss": 0.9292,
+      "step": 24100
+    },
+    {
+      "epoch": 0.8151441659929938,
+      "grad_norm": 0.916677713394165,
+      "learning_rate": 0.0002789926345100152,
+      "loss": 0.9274,
+      "step": 24200
+    },
+    {
+      "epoch": 0.8185125303152789,
+      "grad_norm": 0.8420447111129761,
+      "learning_rate": 0.0002787680768885296,
+      "loss": 0.9293,
+      "step": 24300
+    },
+    {
+      "epoch": 0.821880894637564,
+      "grad_norm": 0.8805976510047913,
+      "learning_rate": 0.0002785435192670439,
+      "loss": 0.9243,
+      "step": 24400
+    },
+    {
+      "epoch": 0.8252492589598491,
+      "grad_norm": 0.8013381361961365,
+      "learning_rate": 0.0002783189616455582,
+      "loss": 0.9296,
+      "step": 24500
+    },
+    {
+      "epoch": 0.8286176232821342,
+      "grad_norm": 0.8348533511161804,
+      "learning_rate": 0.00027809440402407255,
+      "loss": 0.9256,
+      "step": 24600
+    },
+    {
+      "epoch": 0.8319859876044193,
+      "grad_norm": 0.7982239723205566,
+      "learning_rate": 0.0002778698464025869,
+      "loss": 0.9273,
+      "step": 24700
+    },
+    {
+      "epoch": 0.8353543519267044,
+      "grad_norm": 0.9021079540252686,
+      "learning_rate": 0.0002776452887811012,
+      "loss": 0.9238,
+      "step": 24800
+    },
+    {
+      "epoch": 0.8387227162489895,
+      "grad_norm": 0.7782942652702332,
+      "learning_rate": 0.0002774207311596155,
+      "loss": 0.9248,
+      "step": 24900
+    },
+    {
+      "epoch": 0.8420910805712746,
+      "grad_norm": 0.8153879046440125,
+      "learning_rate": 0.00027719617353812984,
+      "loss": 0.9272,
+      "step": 25000
+    },
+    {
+      "epoch": 0.8454594448935597,
+      "grad_norm": 0.9402956366539001,
+      "learning_rate": 0.00027697161591664417,
+      "loss": 0.9243,
+      "step": 25100
+    },
+    {
+      "epoch": 0.8488278092158448,
+      "grad_norm": 0.7734837532043457,
+      "learning_rate": 0.0002767470582951585,
+      "loss": 0.9262,
+      "step": 25200
+    },
+    {
+      "epoch": 0.8521961735381299,
+      "grad_norm": 0.7865928411483765,
+      "learning_rate": 0.00027652250067367284,
+      "loss": 0.9247,
+      "step": 25300
+    },
+    {
+      "epoch": 0.855564537860415,
+      "grad_norm": 0.8157060742378235,
+      "learning_rate": 0.0002762979430521872,
+      "loss": 0.9235,
+      "step": 25400
+    },
+    {
+      "epoch": 0.8589329021827001,
+      "grad_norm": 0.9254295825958252,
+      "learning_rate": 0.0002760733854307015,
+      "loss": 0.9201,
+      "step": 25500
+    },
+    {
+      "epoch": 0.8623012665049852,
+      "grad_norm": 0.7986881136894226,
+      "learning_rate": 0.00027584882780921585,
+      "loss": 0.9215,
+      "step": 25600
+    },
+    {
+      "epoch": 0.8656696308272703,
+      "grad_norm": 1.0841712951660156,
+      "learning_rate": 0.00027562427018773013,
+      "loss": 0.9234,
+      "step": 25700
+    },
+    {
+      "epoch": 0.8690379951495554,
+      "grad_norm": 0.8564463257789612,
+      "learning_rate": 0.00027539971256624446,
+      "loss": 0.9238,
+      "step": 25800
+    },
+    {
+      "epoch": 0.8724063594718404,
+      "grad_norm": 0.7546072602272034,
+      "learning_rate": 0.0002751751549447588,
+      "loss": 0.9195,
+      "step": 25900
+    },
+    {
+      "epoch": 0.8757747237941256,
+      "grad_norm": 1.1822130680084229,
+      "learning_rate": 0.00027495059732327313,
+      "loss": 0.9204,
+      "step": 26000
+    },
+    {
+      "epoch": 0.8791430881164106,
+      "grad_norm": 0.7953358292579651,
+      "learning_rate": 0.00027472603970178747,
+      "loss": 0.9234,
+      "step": 26100
+    },
+    {
+      "epoch": 0.8825114524386958,
+      "grad_norm": 0.7510402798652649,
+      "learning_rate": 0.00027450148208030175,
+      "loss": 0.9193,
+      "step": 26200
+    },
+    {
+      "epoch": 0.8858798167609808,
+      "grad_norm": 1.1970155239105225,
+      "learning_rate": 0.00027427692445881614,
+      "loss": 0.9165,
+      "step": 26300
+    },
+    {
+      "epoch": 0.889248181083266,
+      "grad_norm": 1.0077903270721436,
+      "learning_rate": 0.0002740523668373304,
+      "loss": 0.9136,
+      "step": 26400
+    },
+    {
+      "epoch": 0.892616545405551,
+      "grad_norm": 0.7909373044967651,
+      "learning_rate": 0.00027382780921584476,
+      "loss": 0.9162,
+      "step": 26500
+    },
+    {
+      "epoch": 0.8959849097278362,
+      "grad_norm": 0.7484354376792908,
+      "learning_rate": 0.0002736032515943591,
+      "loss": 0.9151,
+      "step": 26600
+    },
+    {
+      "epoch": 0.8993532740501212,
+      "grad_norm": 0.7733712792396545,
+      "learning_rate": 0.0002733786939728734,
+      "loss": 0.9171,
+      "step": 26700
+    },
+    {
+      "epoch": 0.9027216383724064,
+      "grad_norm": 1.0612813234329224,
+      "learning_rate": 0.00027315413635138776,
+      "loss": 0.9202,
+      "step": 26800
+    },
+    {
+      "epoch": 0.9060900026946914,
+      "grad_norm": 0.8826086521148682,
+      "learning_rate": 0.00027292957872990204,
+      "loss": 0.9184,
+      "step": 26900
+    },
+    {
+      "epoch": 0.9094583670169766,
+      "grad_norm": 0.7911275029182434,
+      "learning_rate": 0.0002727050211084164,
+      "loss": 0.9144,
+      "step": 27000
+    },
+    {
+      "epoch": 0.9128267313392616,
+      "grad_norm": 0.7702584862709045,
+      "learning_rate": 0.0002724804634869307,
+      "loss": 0.9086,
+      "step": 27100
+    },
+    {
+      "epoch": 0.9161950956615468,
+      "grad_norm": 0.7692966461181641,
+      "learning_rate": 0.00027225590586544505,
+      "loss": 0.9127,
+      "step": 27200
+    },
+    {
+      "epoch": 0.9195634599838318,
+      "grad_norm": 0.7272462844848633,
+      "learning_rate": 0.0002720313482439594,
+      "loss": 0.9129,
+      "step": 27300
+    },
+    {
+      "epoch": 0.922931824306117,
+      "grad_norm": 0.7643866539001465,
+      "learning_rate": 0.0002718067906224737,
+      "loss": 0.9127,
+      "step": 27400
+    },
+    {
+      "epoch": 0.926300188628402,
+      "grad_norm": 0.8528723120689392,
+      "learning_rate": 0.000271582233000988,
+      "loss": 0.9118,
+      "step": 27500
+    },
+    {
+      "epoch": 0.9296685529506872,
+      "grad_norm": 0.9294866323471069,
+      "learning_rate": 0.0002713576753795024,
+      "loss": 0.913,
+      "step": 27600
+    },
+    {
+      "epoch": 0.9330369172729722,
+      "grad_norm": 0.7819476127624512,
+      "learning_rate": 0.00027113311775801667,
+      "loss": 0.9094,
+      "step": 27700
+    },
+    {
+      "epoch": 0.9364052815952574,
+      "grad_norm": 0.9195557236671448,
+      "learning_rate": 0.000270908560136531,
+      "loss": 0.9102,
+      "step": 27800
+    },
+    {
+      "epoch": 0.9397736459175424,
+      "grad_norm": 0.8723948001861572,
+      "learning_rate": 0.00027068400251504534,
+      "loss": 0.9099,
+      "step": 27900
+    },
+    {
+      "epoch": 0.9431420102398276,
+      "grad_norm": 0.7643413543701172,
+      "learning_rate": 0.0002704594448935597,
+      "loss": 0.9101,
+      "step": 28000
+    },
+    {
+      "epoch": 0.9465103745621126,
+      "grad_norm": 0.8065772652626038,
+      "learning_rate": 0.000270234887272074,
+      "loss": 0.9099,
+      "step": 28100
+    },
+    {
+      "epoch": 0.9498787388843978,
+      "grad_norm": 0.7567630410194397,
+      "learning_rate": 0.0002700103296505883,
+      "loss": 0.9092,
+      "step": 28200
+    },
+    {
+      "epoch": 0.9532471032066828,
+      "grad_norm": 0.8164196014404297,
+      "learning_rate": 0.0002697857720291026,
+      "loss": 0.9115,
+      "step": 28300
+    },
+    {
+      "epoch": 0.956615467528968,
+      "grad_norm": 0.7787773013114929,
+      "learning_rate": 0.00026956121440761696,
+      "loss": 0.9062,
+      "step": 28400
+    },
+    {
+      "epoch": 0.959983831851253,
+      "grad_norm": 0.724734902381897,
+      "learning_rate": 0.0002693366567861313,
+      "loss": 0.9078,
+      "step": 28500
+    },
+    {
+      "epoch": 0.9633521961735382,
+      "grad_norm": 0.813438355922699,
+      "learning_rate": 0.00026911209916464563,
+      "loss": 0.9075,
+      "step": 28600
+    },
+    {
+      "epoch": 0.9667205604958232,
+      "grad_norm": 0.9264329671859741,
+      "learning_rate": 0.00026888754154315997,
+      "loss": 0.9052,
+      "step": 28700
+    },
+    {
+      "epoch": 0.9700889248181084,
+      "grad_norm": 0.8521473407745361,
+      "learning_rate": 0.0002686629839216743,
+      "loss": 0.9034,
+      "step": 28800
+    },
+    {
+      "epoch": 0.9734572891403934,
+      "grad_norm": 0.7979128360748291,
+      "learning_rate": 0.0002684384263001886,
+      "loss": 0.9052,
+      "step": 28900
+    },
+    {
+      "epoch": 0.9768256534626786,
+      "grad_norm": 0.7806681394577026,
+      "learning_rate": 0.0002682138686787029,
+      "loss": 0.9039,
+      "step": 29000
+    },
+    {
+      "epoch": 0.9801940177849636,
+      "grad_norm": 0.7266373038291931,
+      "learning_rate": 0.00026798931105721725,
+      "loss": 0.9072,
+      "step": 29100
+    },
+    {
+      "epoch": 0.9835623821072487,
+      "grad_norm": 0.8608720302581787,
+      "learning_rate": 0.0002677647534357316,
+      "loss": 0.9062,
+      "step": 29200
+    },
+    {
+      "epoch": 0.9869307464295338,
+      "grad_norm": 0.7421156764030457,
+      "learning_rate": 0.0002675401958142459,
+      "loss": 0.9059,
+      "step": 29300
+    },
+    {
+      "epoch": 0.9902991107518189,
+      "grad_norm": 0.7319119572639465,
+      "learning_rate": 0.00026731563819276026,
+      "loss": 0.9074,
+      "step": 29400
+    },
+    {
+      "epoch": 0.993667475074104,
+      "grad_norm": 1.1928077936172485,
+      "learning_rate": 0.00026709108057127454,
+      "loss": 0.9024,
+      "step": 29500
+    },
+    {
+      "epoch": 0.9970358393963891,
+      "grad_norm": 0.7490862011909485,
+      "learning_rate": 0.00026686652294978893,
+      "loss": 0.904,
+      "step": 29600
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.840101957321167,
+      "eval_runtime": 1.8082,
+      "eval_samples_per_second": 2765.256,
+      "eval_steps_per_second": 43.691,
+      "step": 29688
+    },
+    {
+      "epoch": 1.0004042037186742,
+      "grad_norm": 0.7771435379981995,
+      "learning_rate": 0.0002666419653283032,
+      "loss": 0.9014,
+      "step": 29700
+    },
+    {
+      "epoch": 1.0037725680409593,
+      "grad_norm": 0.7980936765670776,
+      "learning_rate": 0.00026641740770681755,
+      "loss": 0.9045,
+      "step": 29800
+    },
+    {
+      "epoch": 1.0071409323632443,
+      "grad_norm": 0.7477487921714783,
+      "learning_rate": 0.0002661928500853319,
+      "loss": 0.904,
+      "step": 29900
+    },
+    {
+      "epoch": 1.0105092966855296,
+      "grad_norm": 0.7866721153259277,
+      "learning_rate": 0.00026596829246384616,
+      "loss": 0.905,
+      "step": 30000
+    },
+    {
+      "epoch": 1.0138776610078146,
+      "grad_norm": 0.8210276365280151,
+      "learning_rate": 0.00026574373484236055,
+      "loss": 0.9023,
+      "step": 30100
+    },
+    {
+      "epoch": 1.0172460253300997,
+      "grad_norm": 0.683873176574707,
+      "learning_rate": 0.00026551917722087483,
+      "loss": 0.899,
+      "step": 30200
+    },
+    {
+      "epoch": 1.0206143896523847,
+      "grad_norm": 0.772179901599884,
+      "learning_rate": 0.00026529461959938917,
+      "loss": 0.8972,
+      "step": 30300
+    },
+    {
+      "epoch": 1.02398275397467,
+      "grad_norm": 0.9590583443641663,
+      "learning_rate": 0.0002650700619779035,
+      "loss": 0.9048,
+      "step": 30400
+    },
+    {
+      "epoch": 1.027351118296955,
+      "grad_norm": 0.8030016422271729,
+      "learning_rate": 0.00026484550435641784,
+      "loss": 0.9007,
+      "step": 30500
+    },
+    {
+      "epoch": 1.03071948261924,
+      "grad_norm": 0.7494839429855347,
+      "learning_rate": 0.0002646209467349322,
+      "loss": 0.9002,
+      "step": 30600
+    },
+    {
+      "epoch": 1.0340878469415251,
+      "grad_norm": 0.7339987754821777,
+      "learning_rate": 0.00026439638911344645,
+      "loss": 0.8995,
+      "step": 30700
+    },
+    {
+      "epoch": 1.0374562112638104,
+      "grad_norm": 0.7874321937561035,
+      "learning_rate": 0.0002641718314919608,
+      "loss": 0.8982,
+      "step": 30800
+    },
+    {
+      "epoch": 1.0408245755860954,
+      "grad_norm": 0.6752808690071106,
+      "learning_rate": 0.0002639472738704751,
+      "loss": 0.8986,
+      "step": 30900
+    },
+    {
+      "epoch": 1.0441929399083805,
+      "grad_norm": 0.758162796497345,
+      "learning_rate": 0.00026372271624898946,
+      "loss": 0.8963,
+      "step": 31000
+    },
+    {
+      "epoch": 1.0475613042306655,
+      "grad_norm": 1.2287030220031738,
+      "learning_rate": 0.0002634981586275038,
+      "loss": 0.8999,
+      "step": 31100
+    },
+    {
+      "epoch": 1.0509296685529508,
+      "grad_norm": 0.7714762687683105,
+      "learning_rate": 0.00026327360100601813,
+      "loss": 0.8952,
+      "step": 31200
+    },
+    {
+      "epoch": 1.0542980328752358,
+      "grad_norm": 0.7721061706542969,
+      "learning_rate": 0.00026304904338453247,
+      "loss": 0.895,
+      "step": 31300
+    },
+    {
+      "epoch": 1.0576663971975209,
+      "grad_norm": 0.7775672078132629,
+      "learning_rate": 0.0002628244857630468,
+      "loss": 0.8973,
+      "step": 31400
+    },
+    {
+      "epoch": 1.061034761519806,
+      "grad_norm": 0.8213030695915222,
+      "learning_rate": 0.0002625999281415611,
+      "loss": 0.8958,
+      "step": 31500
+    },
+    {
+      "epoch": 1.0644031258420912,
+      "grad_norm": 1.3196107149124146,
+      "learning_rate": 0.0002623753705200754,
+      "loss": 0.8953,
+      "step": 31600
+    },
+    {
+      "epoch": 1.0677714901643762,
+      "grad_norm": 0.7078897953033447,
+      "learning_rate": 0.00026215081289858975,
+      "loss": 0.8981,
+      "step": 31700
+    },
+    {
+      "epoch": 1.0711398544866613,
+      "grad_norm": 0.7529389262199402,
+      "learning_rate": 0.0002619262552771041,
+      "loss": 0.8963,
+      "step": 31800
+    },
+    {
+      "epoch": 1.0745082188089463,
+      "grad_norm": 0.7418708801269531,
+      "learning_rate": 0.0002617016976556184,
+      "loss": 0.8931,
+      "step": 31900
+    },
+    {
+      "epoch": 1.0778765831312316,
+      "grad_norm": 0.6806597113609314,
+      "learning_rate": 0.0002614771400341327,
+      "loss": 0.8938,
+      "step": 32000
+    },
+    {
+      "epoch": 1.0812449474535166,
+      "grad_norm": 0.8359425067901611,
+      "learning_rate": 0.0002612525824126471,
+      "loss": 0.8909,
+      "step": 32100
+    },
+    {
+      "epoch": 1.0846133117758017,
+      "grad_norm": 0.7026517391204834,
+      "learning_rate": 0.0002610280247911614,
+      "loss": 0.8932,
+      "step": 32200
+    },
+    {
+      "epoch": 1.0879816760980867,
+      "grad_norm": 0.9777092933654785,
+      "learning_rate": 0.0002608034671696757,
+      "loss": 0.8936,
+      "step": 32300
+    },
+    {
+      "epoch": 1.0913500404203718,
+      "grad_norm": 0.7617833018302917,
+      "learning_rate": 0.00026057890954819004,
+      "loss": 0.8957,
+      "step": 32400
+    },
+    {
+      "epoch": 1.094718404742657,
+      "grad_norm": 0.7257563471794128,
+      "learning_rate": 0.0002603543519267044,
+      "loss": 0.8947,
+      "step": 32500
+    },
+    {
+      "epoch": 1.098086769064942,
+      "grad_norm": 0.6984378695487976,
+      "learning_rate": 0.0002601297943052187,
+      "loss": 0.8933,
+      "step": 32600
+    },
+    {
+      "epoch": 1.1014551333872271,
+      "grad_norm": 0.7216737270355225,
+      "learning_rate": 0.000259905236683733,
+      "loss": 0.8912,
+      "step": 32700
+    },
+    {
+      "epoch": 1.1048234977095122,
+      "grad_norm": 0.8986912965774536,
+      "learning_rate": 0.00025968067906224733,
+      "loss": 0.8924,
+      "step": 32800
+    },
+    {
+      "epoch": 1.1081918620317974,
+      "grad_norm": 0.7360677719116211,
+      "learning_rate": 0.00025945612144076167,
+      "loss": 0.8912,
+      "step": 32900
+    },
+    {
+      "epoch": 1.1115602263540825,
+      "grad_norm": 0.7221621870994568,
+      "learning_rate": 0.000259231563819276,
+      "loss": 0.8903,
+      "step": 33000
+    },
+    {
+      "epoch": 1.1149285906763675,
+      "grad_norm": 0.8666985034942627,
+      "learning_rate": 0.00025900700619779034,
+      "loss": 0.8869,
+      "step": 33100
+    },
+    {
+      "epoch": 1.1182969549986526,
+      "grad_norm": 0.7224547266960144,
+      "learning_rate": 0.00025878244857630467,
+      "loss": 0.8909,
+      "step": 33200
+    },
+    {
+      "epoch": 1.1216653193209378,
+      "grad_norm": 1.036876916885376,
+      "learning_rate": 0.000258557890954819,
+      "loss": 0.8931,
+      "step": 33300
+    },
+    {
+      "epoch": 1.1250336836432229,
+      "grad_norm": 0.795993447303772,
+      "learning_rate": 0.00025833333333333334,
+      "loss": 0.8853,
+      "step": 33400
+    },
+    {
+      "epoch": 1.128402047965508,
+      "grad_norm": 0.7044444680213928,
+      "learning_rate": 0.0002581087757118476,
+      "loss": 0.8913,
+      "step": 33500
+    },
+    {
+      "epoch": 1.131770412287793,
+      "grad_norm": 0.7511852979660034,
+      "learning_rate": 0.00025788421809036196,
+      "loss": 0.8886,
+      "step": 33600
+    },
+    {
+      "epoch": 1.1351387766100782,
+      "grad_norm": 0.8065422177314758,
+      "learning_rate": 0.0002576596604688763,
+      "loss": 0.8882,
+      "step": 33700
+    },
+    {
+      "epoch": 1.1385071409323633,
+      "grad_norm": 0.7372477650642395,
+      "learning_rate": 0.00025743510284739063,
+      "loss": 0.8886,
+      "step": 33800
+    },
+    {
+      "epoch": 1.1418755052546483,
+      "grad_norm": 0.7889522314071655,
+      "learning_rate": 0.00025721054522590496,
+      "loss": 0.8886,
+      "step": 33900
+    },
+    {
+      "epoch": 1.1452438695769334,
+      "grad_norm": 0.6768883466720581,
+      "learning_rate": 0.00025698598760441925,
+      "loss": 0.8895,
+      "step": 34000
+    },
+    {
+      "epoch": 1.1486122338992186,
+      "grad_norm": 0.8599404096603394,
+      "learning_rate": 0.00025676142998293363,
+      "loss": 0.8904,
+      "step": 34100
+    },
+    {
+      "epoch": 1.1519805982215037,
+      "grad_norm": 0.716754138469696,
+      "learning_rate": 0.0002565368723614479,
+      "loss": 0.8898,
+      "step": 34200
+    },
+    {
+      "epoch": 1.1553489625437887,
+      "grad_norm": 0.7366636991500854,
+      "learning_rate": 0.00025631231473996225,
+      "loss": 0.8872,
+      "step": 34300
+    },
+    {
+      "epoch": 1.1587173268660738,
+      "grad_norm": 0.7212886810302734,
+      "learning_rate": 0.0002560877571184766,
+      "loss": 0.8897,
+      "step": 34400
+    },
+    {
+      "epoch": 1.162085691188359,
+      "grad_norm": 0.768371045589447,
+      "learning_rate": 0.00025586319949699087,
+      "loss": 0.8871,
+      "step": 34500
+    },
+    {
+      "epoch": 1.165454055510644,
+      "grad_norm": 0.7179331183433533,
+      "learning_rate": 0.00025563864187550526,
+      "loss": 0.8864,
+      "step": 34600
+    },
+    {
+      "epoch": 1.1688224198329291,
+      "grad_norm": 0.7503130435943604,
+      "learning_rate": 0.00025541408425401954,
+      "loss": 0.8857,
+      "step": 34700
+    },
+    {
+      "epoch": 1.1721907841552142,
+      "grad_norm": 0.7497594952583313,
+      "learning_rate": 0.00025518952663253387,
+      "loss": 0.8858,
+      "step": 34800
+    },
+    {
+      "epoch": 1.1755591484774994,
+      "grad_norm": 0.6965382695198059,
+      "learning_rate": 0.0002549649690110482,
+      "loss": 0.8867,
+      "step": 34900
+    },
+    {
+      "epoch": 1.1789275127997845,
+      "grad_norm": 0.7309035658836365,
+      "learning_rate": 0.00025474041138956254,
+      "loss": 0.8865,
+      "step": 35000
+    },
+    {
+      "epoch": 1.1822958771220695,
+      "grad_norm": 0.692935049533844,
+      "learning_rate": 0.0002545158537680769,
+      "loss": 0.8839,
+      "step": 35100
+    },
+    {
+      "epoch": 1.1856642414443546,
+      "grad_norm": 0.7206865549087524,
+      "learning_rate": 0.0002542912961465912,
+      "loss": 0.8833,
+      "step": 35200
+    },
+    {
+      "epoch": 1.1890326057666396,
+      "grad_norm": 0.718701958656311,
+      "learning_rate": 0.0002540667385251055,
+      "loss": 0.8831,
+      "step": 35300
+    },
+    {
+      "epoch": 1.1924009700889249,
+      "grad_norm": 0.797205924987793,
+      "learning_rate": 0.0002538421809036199,
+      "loss": 0.8851,
+      "step": 35400
+    },
+    {
+      "epoch": 1.19576933441121,
+      "grad_norm": 0.6977315545082092,
+      "learning_rate": 0.00025361762328213416,
+      "loss": 0.8834,
+      "step": 35500
+    },
+    {
+      "epoch": 1.199137698733495,
+      "grad_norm": 0.768139660358429,
+      "learning_rate": 0.0002533930656606485,
+      "loss": 0.8822,
+      "step": 35600
+    },
+    {
+      "epoch": 1.2025060630557802,
+      "grad_norm": 0.7986385226249695,
+      "learning_rate": 0.00025316850803916283,
+      "loss": 0.8816,
+      "step": 35700
+    },
+    {
+      "epoch": 1.2058744273780653,
+      "grad_norm": 0.7320616841316223,
+      "learning_rate": 0.00025294395041767717,
+      "loss": 0.8827,
+      "step": 35800
+    },
+    {
+      "epoch": 1.2092427917003503,
+      "grad_norm": 0.7750969529151917,
+      "learning_rate": 0.0002527193927961915,
+      "loss": 0.8851,
+      "step": 35900
+    },
+    {
+      "epoch": 1.2126111560226354,
+      "grad_norm": 0.7603546380996704,
+      "learning_rate": 0.0002524948351747058,
+      "loss": 0.8834,
+      "step": 36000
+    },
+    {
+      "epoch": 1.2159795203449204,
+      "grad_norm": 0.6906490325927734,
+      "learning_rate": 0.0002522702775532201,
+      "loss": 0.8835,
+      "step": 36100
+    },
+    {
+      "epoch": 1.2193478846672057,
+      "grad_norm": 0.7611385583877563,
+      "learning_rate": 0.00025204571993173446,
+      "loss": 0.8832,
+      "step": 36200
+    },
+    {
+      "epoch": 1.2227162489894907,
+      "grad_norm": 0.732187807559967,
+      "learning_rate": 0.0002518211623102488,
+      "loss": 0.8813,
+      "step": 36300
+    },
+    {
+      "epoch": 1.2260846133117758,
+      "grad_norm": 0.7710540890693665,
+      "learning_rate": 0.0002515966046887631,
+      "loss": 0.8835,
+      "step": 36400
+    },
+    {
+      "epoch": 1.2294529776340608,
+      "grad_norm": 0.7388759255409241,
+      "learning_rate": 0.0002513720470672774,
+      "loss": 0.8846,
+      "step": 36500
+    },
+    {
+      "epoch": 1.232821341956346,
+      "grad_norm": 0.7000882029533386,
+      "learning_rate": 0.0002511474894457918,
+      "loss": 0.8818,
+      "step": 36600
+    },
+    {
+      "epoch": 1.2361897062786311,
+      "grad_norm": 0.7058309316635132,
+      "learning_rate": 0.0002509229318243061,
+      "loss": 0.8827,
+      "step": 36700
+    },
+    {
+      "epoch": 1.2395580706009162,
+      "grad_norm": 0.80439692735672,
+      "learning_rate": 0.0002506983742028204,
+      "loss": 0.8793,
+      "step": 36800
+    },
+    {
+      "epoch": 1.2429264349232012,
+      "grad_norm": 0.7056805491447449,
+      "learning_rate": 0.00025047381658133475,
+      "loss": 0.8763,
+      "step": 36900
+    },
+    {
+      "epoch": 1.2462947992454865,
+      "grad_norm": 0.7477532625198364,
+      "learning_rate": 0.0002502492589598491,
+      "loss": 0.8803,
+      "step": 37000
+    },
+    {
+      "epoch": 1.2496631635677715,
+      "grad_norm": 0.8339030742645264,
+      "learning_rate": 0.0002500247013383634,
+      "loss": 0.8782,
+      "step": 37100
+    },
+    {
+      "epoch": 1.2530315278900566,
+      "grad_norm": 0.7052040100097656,
+      "learning_rate": 0.00024980014371687775,
+      "loss": 0.8834,
+      "step": 37200
+    },
+    {
+      "epoch": 1.2563998922123416,
+      "grad_norm": 0.7932031750679016,
+      "learning_rate": 0.00024957558609539204,
+      "loss": 0.882,
+      "step": 37300
+    },
+    {
+      "epoch": 1.2597682565346267,
+      "grad_norm": 0.7311996817588806,
+      "learning_rate": 0.00024935102847390637,
+      "loss": 0.8802,
+      "step": 37400
+    },
+    {
+      "epoch": 1.263136620856912,
+      "grad_norm": 0.7274471521377563,
+      "learning_rate": 0.0002491264708524207,
+      "loss": 0.8801,
+      "step": 37500
+    },
+    {
+      "epoch": 1.266504985179197,
+      "grad_norm": 0.6959684491157532,
+      "learning_rate": 0.00024890191323093504,
+      "loss": 0.8742,
+      "step": 37600
+    },
+    {
+      "epoch": 1.269873349501482,
+      "grad_norm": 0.7355061173439026,
+      "learning_rate": 0.0002486773556094494,
+      "loss": 0.8763,
+      "step": 37700
+    },
+    {
+      "epoch": 1.2732417138237673,
+      "grad_norm": 0.6900234818458557,
+      "learning_rate": 0.00024845279798796366,
+      "loss": 0.879,
+      "step": 37800
+    },
+    {
+      "epoch": 1.2766100781460523,
+      "grad_norm": 0.7259871363639832,
+      "learning_rate": 0.00024822824036647805,
+      "loss": 0.8769,
+      "step": 37900
+    },
+    {
+      "epoch": 1.2799784424683374,
+      "grad_norm": 0.7086994051933289,
+      "learning_rate": 0.00024800368274499233,
+      "loss": 0.877,
+      "step": 38000
+    },
+    {
+      "epoch": 1.2833468067906224,
+      "grad_norm": 0.7197995185852051,
+      "learning_rate": 0.00024777912512350666,
+      "loss": 0.8757,
+      "step": 38100
+    },
+    {
+      "epoch": 1.2867151711129075,
+      "grad_norm": 0.7302994728088379,
+      "learning_rate": 0.000247554567502021,
+      "loss": 0.8801,
+      "step": 38200
+    },
+    {
+      "epoch": 1.2900835354351927,
+      "grad_norm": 0.6908255815505981,
+      "learning_rate": 0.00024733000988053533,
+      "loss": 0.8766,
+      "step": 38300
+    },
+    {
+      "epoch": 1.2934518997574778,
+      "grad_norm": 0.8338357210159302,
+      "learning_rate": 0.00024710545225904967,
+      "loss": 0.874,
+      "step": 38400
+    },
+    {
+      "epoch": 1.2968202640797628,
+      "grad_norm": 0.7342631220817566,
+      "learning_rate": 0.00024688089463756395,
+      "loss": 0.8786,
+      "step": 38500
+    },
+    {
+      "epoch": 1.300188628402048,
+      "grad_norm": 0.7488550543785095,
+      "learning_rate": 0.0002466563370160783,
+      "loss": 0.8813,
+      "step": 38600
+    },
+    {
+      "epoch": 1.3035569927243331,
+      "grad_norm": 0.7316462993621826,
+      "learning_rate": 0.0002464317793945926,
+      "loss": 0.8755,
+      "step": 38700
+    },
+    {
+      "epoch": 1.3069253570466182,
+      "grad_norm": 0.7482060790061951,
+      "learning_rate": 0.00024620722177310695,
+      "loss": 0.8737,
+      "step": 38800
+    },
+    {
+      "epoch": 1.3102937213689032,
+      "grad_norm": 0.7243971228599548,
+      "learning_rate": 0.0002459826641516213,
+      "loss": 0.8759,
+      "step": 38900
+    },
+    {
+      "epoch": 1.3136620856911883,
+      "grad_norm": 0.8776415586471558,
+      "learning_rate": 0.0002457581065301356,
+      "loss": 0.8757,
+      "step": 39000
+    },
+    {
+      "epoch": 1.3170304500134735,
+      "grad_norm": 0.6730260848999023,
+      "learning_rate": 0.00024553354890864996,
+      "loss": 0.873,
+      "step": 39100
+    },
+    {
+      "epoch": 1.3203988143357586,
+      "grad_norm": 0.6937080025672913,
+      "learning_rate": 0.0002453089912871643,
+      "loss": 0.8737,
+      "step": 39200
+    },
+    {
+      "epoch": 1.3237671786580436,
+      "grad_norm": 0.7004138827323914,
+      "learning_rate": 0.0002450844336656786,
+      "loss": 0.8771,
+      "step": 39300
+    },
+    {
+      "epoch": 1.3271355429803289,
+      "grad_norm": 0.6719589829444885,
+      "learning_rate": 0.0002448598760441929,
+      "loss": 0.8758,
+      "step": 39400
+    },
+    {
+      "epoch": 1.330503907302614,
+      "grad_norm": 0.738394021987915,
+      "learning_rate": 0.00024463531842270725,
+      "loss": 0.8743,
+      "step": 39500
+    },
+    {
+      "epoch": 1.333872271624899,
+      "grad_norm": 0.7103344202041626,
+      "learning_rate": 0.0002444107608012216,
+      "loss": 0.8712,
+      "step": 39600
+    },
+    {
+      "epoch": 1.337240635947184,
+      "grad_norm": 0.69880610704422,
+      "learning_rate": 0.0002441862031797359,
+      "loss": 0.8737,
+      "step": 39700
+    },
+    {
+      "epoch": 1.340609000269469,
+      "grad_norm": 0.8922190070152283,
+      "learning_rate": 0.00024396164555825023,
+      "loss": 0.8718,
+      "step": 39800
+    },
+    {
+      "epoch": 1.3439773645917543,
+      "grad_norm": 0.7306973338127136,
+      "learning_rate": 0.00024373708793676456,
+      "loss": 0.8712,
+      "step": 39900
+    },
+    {
+      "epoch": 1.3473457289140394,
+      "grad_norm": 0.7311689257621765,
+      "learning_rate": 0.00024351253031527887,
+      "loss": 0.875,
+      "step": 40000
+    },
+    {
+      "epoch": 1.3507140932363244,
+      "grad_norm": 0.7093273997306824,
+      "learning_rate": 0.00024328797269379323,
+      "loss": 0.8711,
+      "step": 40100
+    },
+    {
+      "epoch": 1.3540824575586097,
+      "grad_norm": 0.6579886078834534,
+      "learning_rate": 0.00024306341507230754,
+      "loss": 0.8773,
+      "step": 40200
+    },
+    {
+      "epoch": 1.3574508218808947,
+      "grad_norm": 0.6832658648490906,
+      "learning_rate": 0.00024283885745082185,
+      "loss": 0.8727,
+      "step": 40300
+    },
+    {
+      "epoch": 1.3608191862031798,
+      "grad_norm": 0.7372367978096008,
+      "learning_rate": 0.00024261429982933618,
+      "loss": 0.8718,
+      "step": 40400
+    },
+    {
+      "epoch": 1.3641875505254648,
+      "grad_norm": 0.691271185874939,
+      "learning_rate": 0.0002423897422078505,
+      "loss": 0.8716,
+      "step": 40500
+    },
+    {
+      "epoch": 1.3675559148477499,
+      "grad_norm": 0.7311553359031677,
+      "learning_rate": 0.00024216518458636485,
+      "loss": 0.8711,
+      "step": 40600
+    },
+    {
+      "epoch": 1.3709242791700351,
+      "grad_norm": 0.7508808970451355,
+      "learning_rate": 0.00024194062696487916,
+      "loss": 0.8708,
+      "step": 40700
+    },
+    {
+      "epoch": 1.3742926434923202,
+      "grad_norm": 0.707360029220581,
+      "learning_rate": 0.0002417160693433935,
+      "loss": 0.8727,
+      "step": 40800
+    },
+    {
+      "epoch": 1.3776610078146052,
+      "grad_norm": 0.6759727001190186,
+      "learning_rate": 0.0002414915117219078,
+      "loss": 0.8681,
+      "step": 40900
+    },
+    {
+      "epoch": 1.3810293721368903,
+      "grad_norm": 0.6768555045127869,
+      "learning_rate": 0.00024126695410042217,
+      "loss": 0.8709,
+      "step": 41000
+    },
+    {
+      "epoch": 1.3843977364591753,
+      "grad_norm": 0.7042515873908997,
+      "learning_rate": 0.00024104239647893647,
+      "loss": 0.8711,
+      "step": 41100
+    },
+    {
+      "epoch": 1.3877661007814606,
+      "grad_norm": 0.7361947894096375,
+      "learning_rate": 0.00024081783885745078,
+      "loss": 0.87,
+      "step": 41200
+    },
+    {
+      "epoch": 1.3911344651037456,
+      "grad_norm": 0.7217181921005249,
+      "learning_rate": 0.00024059328123596512,
+      "loss": 0.8673,
+      "step": 41300
+    },
+    {
+      "epoch": 1.3945028294260307,
+      "grad_norm": 0.6747815608978271,
+      "learning_rate": 0.00024036872361447945,
+      "loss": 0.8686,
+      "step": 41400
+    },
+    {
+      "epoch": 1.397871193748316,
+      "grad_norm": 0.8375660181045532,
+      "learning_rate": 0.0002401441659929938,
+      "loss": 0.8711,
+      "step": 41500
+    },
+    {
+      "epoch": 1.401239558070601,
+      "grad_norm": 0.7188234925270081,
+      "learning_rate": 0.0002399196083715081,
+      "loss": 0.8687,
+      "step": 41600
+    },
+    {
+      "epoch": 1.404607922392886,
+      "grad_norm": 0.7048287987709045,
+      "learning_rate": 0.00023969505075002243,
+      "loss": 0.8663,
+      "step": 41700
+    },
+    {
+      "epoch": 1.407976286715171,
+      "grad_norm": 0.703464925289154,
+      "learning_rate": 0.00023947049312853677,
+      "loss": 0.8698,
+      "step": 41800
+    },
+    {
+      "epoch": 1.411344651037456,
+      "grad_norm": 0.725428581237793,
+      "learning_rate": 0.0002392459355070511,
+      "loss": 0.8704,
+      "step": 41900
+    },
+    {
+      "epoch": 1.4147130153597414,
+      "grad_norm": 0.7408603429794312,
+      "learning_rate": 0.0002390213778855654,
+      "loss": 0.868,
+      "step": 42000
+    },
+    {
+      "epoch": 1.4180813796820264,
+      "grad_norm": 0.718657910823822,
+      "learning_rate": 0.00023879682026407974,
+      "loss": 0.8684,
+      "step": 42100
+    },
+    {
+      "epoch": 1.4214497440043115,
+      "grad_norm": 0.6897197961807251,
+      "learning_rate": 0.00023857226264259408,
+      "loss": 0.8638,
+      "step": 42200
+    },
+    {
+      "epoch": 1.4248181083265967,
+      "grad_norm": 0.7002888917922974,
+      "learning_rate": 0.0002383477050211084,
+      "loss": 0.8694,
+      "step": 42300
+    },
+    {
+      "epoch": 1.4281864726488818,
+      "grad_norm": 0.6644707322120667,
+      "learning_rate": 0.00023812314739962272,
+      "loss": 0.8684,
+      "step": 42400
+    },
+    {
+      "epoch": 1.4315548369711668,
+      "grad_norm": 0.7583789229393005,
+      "learning_rate": 0.00023789858977813703,
+      "loss": 0.8671,
+      "step": 42500
+    },
+    {
+      "epoch": 1.4349232012934519,
+      "grad_norm": 1.1469764709472656,
+      "learning_rate": 0.0002376740321566514,
+      "loss": 0.8659,
+      "step": 42600
+    },
+    {
+      "epoch": 1.438291565615737,
+      "grad_norm": 0.7378877997398376,
+      "learning_rate": 0.0002374494745351657,
+      "loss": 0.8668,
+      "step": 42700
+    },
+    {
+      "epoch": 1.4416599299380222,
+      "grad_norm": 0.6720541715621948,
+      "learning_rate": 0.00023722491691368004,
+      "loss": 0.867,
+      "step": 42800
+    },
+    {
+      "epoch": 1.4450282942603072,
+      "grad_norm": 0.6979005336761475,
+      "learning_rate": 0.00023700035929219435,
+      "loss": 0.865,
+      "step": 42900
+    },
+    {
+      "epoch": 1.4483966585825923,
+      "grad_norm": 0.6991161108016968,
+      "learning_rate": 0.0002367758016707087,
+      "loss": 0.8646,
+      "step": 43000
+    },
+    {
+      "epoch": 1.4517650229048775,
+      "grad_norm": 0.7050434350967407,
+      "learning_rate": 0.00023655124404922302,
+      "loss": 0.8663,
+      "step": 43100
+    },
+    {
+      "epoch": 1.4551333872271626,
+      "grad_norm": 0.733540415763855,
+      "learning_rate": 0.00023632668642773732,
+      "loss": 0.869,
+      "step": 43200
+    },
+    {
+      "epoch": 1.4585017515494476,
+      "grad_norm": 0.6488509774208069,
+      "learning_rate": 0.00023610212880625166,
+      "loss": 0.8645,
+      "step": 43300
+    },
+    {
+      "epoch": 1.4618701158717327,
+      "grad_norm": 0.760858416557312,
+      "learning_rate": 0.00023587757118476597,
+      "loss": 0.8671,
+      "step": 43400
+    },
+    {
+      "epoch": 1.4652384801940177,
+      "grad_norm": 0.6800232529640198,
+      "learning_rate": 0.00023565301356328033,
+      "loss": 0.8647,
+      "step": 43500
+    },
+    {
+      "epoch": 1.468606844516303,
+      "grad_norm": 0.6814864873886108,
+      "learning_rate": 0.00023542845594179464,
+      "loss": 0.8679,
+      "step": 43600
+    },
+    {
+      "epoch": 1.471975208838588,
+      "grad_norm": 0.6680698394775391,
+      "learning_rate": 0.00023520389832030897,
+      "loss": 0.8638,
+      "step": 43700
+    },
+    {
+      "epoch": 1.475343573160873,
+      "grad_norm": 0.7103099226951599,
+      "learning_rate": 0.00023497934069882328,
+      "loss": 0.8625,
+      "step": 43800
+    },
+    {
+      "epoch": 1.478711937483158,
+      "grad_norm": 0.6946255564689636,
+      "learning_rate": 0.00023475478307733764,
+      "loss": 0.8636,
+      "step": 43900
+    },
+    {
+      "epoch": 1.4820803018054431,
+      "grad_norm": 0.7505577802658081,
+      "learning_rate": 0.00023453022545585195,
+      "loss": 0.8622,
+      "step": 44000
+    },
+    {
+      "epoch": 1.4854486661277284,
+      "grad_norm": 0.623587965965271,
+      "learning_rate": 0.00023430566783436626,
+      "loss": 0.8617,
+      "step": 44100
+    },
+    {
+      "epoch": 1.4888170304500135,
+      "grad_norm": 0.7254552841186523,
+      "learning_rate": 0.0002340811102128806,
+      "loss": 0.8631,
+      "step": 44200
+    },
+    {
+      "epoch": 1.4921853947722985,
+      "grad_norm": 0.7500590682029724,
+      "learning_rate": 0.00023385655259139493,
+      "loss": 0.8665,
+      "step": 44300
+    },
+    {
+      "epoch": 1.4955537590945838,
+      "grad_norm": 0.6879542469978333,
+      "learning_rate": 0.00023363199496990926,
+      "loss": 0.8652,
+      "step": 44400
+    },
+    {
+      "epoch": 1.4989221234168688,
+      "grad_norm": 1.074501872062683,
+      "learning_rate": 0.00023340743734842357,
+      "loss": 0.8635,
+      "step": 44500
+    },
+    {
+      "epoch": 1.5022904877391539,
+      "grad_norm": 0.6877649426460266,
+      "learning_rate": 0.0002331828797269379,
+      "loss": 0.8639,
+      "step": 44600
+    },
+    {
+      "epoch": 1.5056588520614391,
+      "grad_norm": 0.6956282258033752,
+      "learning_rate": 0.00023295832210545224,
+      "loss": 0.8625,
+      "step": 44700
+    },
+    {
+      "epoch": 1.509027216383724,
+      "grad_norm": 0.6834612488746643,
+      "learning_rate": 0.00023273376448396658,
+      "loss": 0.863,
+      "step": 44800
+    },
+    {
+      "epoch": 1.5123955807060092,
+      "grad_norm": 0.6679215431213379,
+      "learning_rate": 0.00023250920686248089,
+      "loss": 0.8604,
+      "step": 44900
+    },
+    {
+      "epoch": 1.5157639450282943,
+      "grad_norm": 0.7261891961097717,
+      "learning_rate": 0.00023228464924099525,
+      "loss": 0.8614,
+      "step": 45000
+    },
+    {
+      "epoch": 1.5191323093505793,
+      "grad_norm": 0.6932804584503174,
+      "learning_rate": 0.00023206009161950956,
+      "loss": 0.8629,
+      "step": 45100
+    },
+    {
+      "epoch": 1.5225006736728646,
+      "grad_norm": 0.9139769673347473,
+      "learning_rate": 0.00023183553399802386,
+      "loss": 0.8592,
+      "step": 45200
+    },
+    {
+      "epoch": 1.5258690379951494,
+      "grad_norm": 0.7025532126426697,
+      "learning_rate": 0.0002316109763765382,
+      "loss": 0.8623,
+      "step": 45300
+    },
+    {
+      "epoch": 1.5292374023174347,
+      "grad_norm": 0.6872456073760986,
+      "learning_rate": 0.0002313864187550525,
+      "loss": 0.8634,
+      "step": 45400
+    },
+    {
+      "epoch": 1.5326057666397197,
+      "grad_norm": 0.6915197968482971,
+      "learning_rate": 0.00023116186113356687,
+      "loss": 0.8604,
+      "step": 45500
+    },
+    {
+      "epoch": 1.5359741309620047,
+      "grad_norm": 0.6892699599266052,
+      "learning_rate": 0.00023093730351208118,
+      "loss": 0.8591,
+      "step": 45600
+    },
+    {
+      "epoch": 1.53934249528429,
+      "grad_norm": 0.7258453369140625,
+      "learning_rate": 0.00023071274589059551,
+      "loss": 0.8611,
+      "step": 45700
+    },
+    {
+      "epoch": 1.542710859606575,
+      "grad_norm": 0.6410971879959106,
+      "learning_rate": 0.00023048818826910982,
+      "loss": 0.8585,
+      "step": 45800
+    },
+    {
+      "epoch": 1.54607922392886,
+      "grad_norm": 0.7481923699378967,
+      "learning_rate": 0.00023026363064762418,
+      "loss": 0.8612,
+      "step": 45900
+    },
+    {
+      "epoch": 1.5494475882511454,
+      "grad_norm": 0.6729730367660522,
+      "learning_rate": 0.0002300390730261385,
+      "loss": 0.8594,
+      "step": 46000
+    },
+    {
+      "epoch": 1.5528159525734302,
+      "grad_norm": 0.6492398977279663,
+      "learning_rate": 0.0002298145154046528,
+      "loss": 0.8596,
+      "step": 46100
+    },
+    {
+      "epoch": 1.5561843168957155,
+      "grad_norm": 0.7804999947547913,
+      "learning_rate": 0.00022958995778316714,
+      "loss": 0.8588,
+      "step": 46200
+    },
+    {
+      "epoch": 1.5595526812180005,
+      "grad_norm": 0.692258358001709,
+      "learning_rate": 0.00022936540016168144,
+      "loss": 0.8644,
+      "step": 46300
+    },
+    {
+      "epoch": 1.5629210455402855,
+      "grad_norm": 0.8399534821510315,
+      "learning_rate": 0.0002291408425401958,
+      "loss": 0.8563,
+      "step": 46400
+    },
+    {
+      "epoch": 1.5662894098625708,
+      "grad_norm": 0.7541986107826233,
+      "learning_rate": 0.00022891628491871011,
+      "loss": 0.8597,
+      "step": 46500
+    },
+    {
+      "epoch": 1.5696577741848559,
+      "grad_norm": 0.7269881367683411,
+      "learning_rate": 0.00022869172729722445,
+      "loss": 0.8613,
+      "step": 46600
+    },
+    {
+      "epoch": 1.573026138507141,
+      "grad_norm": 0.6875913143157959,
+      "learning_rate": 0.00022846716967573876,
+      "loss": 0.8569,
+      "step": 46700
+    },
+    {
+      "epoch": 1.5763945028294262,
+      "grad_norm": 0.6820959448814392,
+      "learning_rate": 0.00022824261205425312,
+      "loss": 0.8578,
+      "step": 46800
+    },
+    {
+      "epoch": 1.579762867151711,
+      "grad_norm": 0.63446044921875,
+      "learning_rate": 0.00022801805443276743,
+      "loss": 0.8583,
+      "step": 46900
+    },
+    {
+      "epoch": 1.5831312314739963,
+      "grad_norm": 0.7082163691520691,
+      "learning_rate": 0.00022779349681128174,
+      "loss": 0.8591,
+      "step": 47000
+    },
+    {
+      "epoch": 1.5864995957962813,
+      "grad_norm": 0.6886746883392334,
+      "learning_rate": 0.0002275689391897961,
+      "loss": 0.8569,
+      "step": 47100
+    },
+    {
+      "epoch": 1.5898679601185663,
+      "grad_norm": 0.6877068281173706,
+      "learning_rate": 0.0002273443815683104,
+      "loss": 0.8563,
+      "step": 47200
+    },
+    {
+      "epoch": 1.5932363244408516,
+      "grad_norm": 0.7213451266288757,
+      "learning_rate": 0.00022711982394682474,
+      "loss": 0.8576,
+      "step": 47300
+    },
+    {
+      "epoch": 1.5966046887631367,
+      "grad_norm": 0.6784006357192993,
+      "learning_rate": 0.00022689526632533905,
+      "loss": 0.8561,
+      "step": 47400
+    },
+    {
+      "epoch": 1.5999730530854217,
+      "grad_norm": 0.7651084661483765,
+      "learning_rate": 0.0002266707087038534,
+      "loss": 0.8533,
+      "step": 47500
+    },
+    {
+      "epoch": 1.603341417407707,
+      "grad_norm": 0.6762063503265381,
+      "learning_rate": 0.00022644615108236772,
+      "loss": 0.8593,
+      "step": 47600
+    },
+    {
+      "epoch": 1.6067097817299918,
+      "grad_norm": 0.6897014379501343,
+      "learning_rate": 0.00022622159346088205,
+      "loss": 0.8584,
+      "step": 47700
+    },
+    {
+      "epoch": 1.610078146052277,
+      "grad_norm": 0.642955482006073,
+      "learning_rate": 0.00022599703583939636,
+      "loss": 0.8536,
+      "step": 47800
+    },
+    {
+      "epoch": 1.613446510374562,
+      "grad_norm": 0.676459014415741,
+      "learning_rate": 0.00022577247821791067,
+      "loss": 0.8561,
+      "step": 47900
+    },
+    {
+      "epoch": 1.6168148746968471,
+      "grad_norm": 0.7143009901046753,
+      "learning_rate": 0.00022554792059642503,
+      "loss": 0.8593,
+      "step": 48000
+    },
+    {
+      "epoch": 1.6201832390191324,
+      "grad_norm": 0.7367132306098938,
+      "learning_rate": 0.00022532336297493934,
+      "loss": 0.8547,
+      "step": 48100
+    },
+    {
+      "epoch": 1.6235516033414175,
+      "grad_norm": 0.6949586868286133,
+      "learning_rate": 0.00022509880535345368,
+      "loss": 0.8577,
+      "step": 48200
+    },
+    {
+      "epoch": 1.6269199676637025,
+      "grad_norm": 0.672558605670929,
+      "learning_rate": 0.00022487424773196798,
+      "loss": 0.8526,
+      "step": 48300
+    },
+    {
+      "epoch": 1.6302883319859878,
+      "grad_norm": 0.6996020078659058,
+      "learning_rate": 0.00022464969011048235,
+      "loss": 0.8527,
+      "step": 48400
+    },
+    {
+      "epoch": 1.6336566963082726,
+      "grad_norm": 0.7007562518119812,
+      "learning_rate": 0.00022442513248899666,
+      "loss": 0.8536,
+      "step": 48500
+    },
+    {
+      "epoch": 1.6370250606305579,
+      "grad_norm": 0.7103307247161865,
+      "learning_rate": 0.000224200574867511,
+      "loss": 0.8557,
+      "step": 48600
+    },
+    {
+      "epoch": 1.640393424952843,
+      "grad_norm": 0.7483230829238892,
+      "learning_rate": 0.0002239760172460253,
+      "loss": 0.857,
+      "step": 48700
+    },
+    {
+      "epoch": 1.643761789275128,
+      "grad_norm": 0.7092981338500977,
+      "learning_rate": 0.00022375145962453966,
+      "loss": 0.8589,
+      "step": 48800
+    },
+    {
+      "epoch": 1.6471301535974132,
+      "grad_norm": 0.7270293235778809,
+      "learning_rate": 0.00022352690200305397,
+      "loss": 0.8553,
+      "step": 48900
+    },
+    {
+      "epoch": 1.650498517919698,
+      "grad_norm": 0.6916648149490356,
+      "learning_rate": 0.00022330234438156828,
+      "loss": 0.8566,
+      "step": 49000
+    },
+    {
+      "epoch": 1.6538668822419833,
+      "grad_norm": 0.6687049269676208,
+      "learning_rate": 0.0002230777867600826,
+      "loss": 0.8532,
+      "step": 49100
+    },
+    {
+      "epoch": 1.6572352465642683,
+      "grad_norm": 0.689947783946991,
+      "learning_rate": 0.00022285322913859695,
+      "loss": 0.8556,
+      "step": 49200
+    },
+    {
+      "epoch": 1.6606036108865534,
+      "grad_norm": 0.7345608472824097,
+      "learning_rate": 0.00022262867151711128,
+      "loss": 0.8526,
+      "step": 49300
+    },
+    {
+      "epoch": 1.6639719752088387,
+      "grad_norm": 0.6420150399208069,
+      "learning_rate": 0.0002224041138956256,
+      "loss": 0.8584,
+      "step": 49400
+    },
+    {
+      "epoch": 1.6673403395311237,
+      "grad_norm": 0.7315524816513062,
+      "learning_rate": 0.00022217955627413993,
+      "loss": 0.8516,
+      "step": 49500
+    },
+    {
+      "epoch": 1.6707087038534087,
+      "grad_norm": 0.674518883228302,
+      "learning_rate": 0.00022195499865265426,
+      "loss": 0.8564,
+      "step": 49600
+    },
+    {
+      "epoch": 1.674077068175694,
+      "grad_norm": 0.6845901608467102,
+      "learning_rate": 0.0002217304410311686,
+      "loss": 0.8493,
+      "step": 49700
+    },
+    {
+      "epoch": 1.6774454324979788,
+      "grad_norm": 0.6932186484336853,
+      "learning_rate": 0.0002215058834096829,
+      "loss": 0.8516,
+      "step": 49800
+    },
+    {
+      "epoch": 1.680813796820264,
+      "grad_norm": 0.6544727087020874,
+      "learning_rate": 0.0002212813257881972,
+      "loss": 0.8519,
+      "step": 49900
+    },
+    {
+      "epoch": 1.6841821611425492,
+      "grad_norm": 0.7253223657608032,
+      "learning_rate": 0.00022105676816671157,
+      "loss": 0.8529,
+      "step": 50000
+    },
+    {
+      "epoch": 1.6875505254648342,
+      "grad_norm": 0.6576797962188721,
+      "learning_rate": 0.00022083221054522588,
+      "loss": 0.8528,
+      "step": 50100
+    },
+    {
+      "epoch": 1.6909188897871195,
+      "grad_norm": 0.715744137763977,
+      "learning_rate": 0.00022060765292374022,
+      "loss": 0.8537,
+      "step": 50200
+    },
+    {
+      "epoch": 1.6942872541094045,
+      "grad_norm": 0.6994728446006775,
+      "learning_rate": 0.00022038309530225453,
+      "loss": 0.852,
+      "step": 50300
+    },
+    {
+      "epoch": 1.6976556184316896,
+      "grad_norm": 0.6934739947319031,
+      "learning_rate": 0.0002201585376807689,
+      "loss": 0.8509,
+      "step": 50400
+    },
+    {
+      "epoch": 1.7010239827539748,
+      "grad_norm": 0.7031779885292053,
+      "learning_rate": 0.0002199339800592832,
+      "loss": 0.853,
+      "step": 50500
+    },
+    {
+      "epoch": 1.7043923470762596,
+      "grad_norm": 0.7014051079750061,
+      "learning_rate": 0.00021970942243779753,
+      "loss": 0.8514,
+      "step": 50600
+    },
+    {
+      "epoch": 1.707760711398545,
+      "grad_norm": 0.8519027233123779,
+      "learning_rate": 0.00021948486481631184,
+      "loss": 0.8525,
+      "step": 50700
+    },
+    {
+      "epoch": 1.71112907572083,
+      "grad_norm": 0.6523889899253845,
+      "learning_rate": 0.00021926030719482615,
+      "loss": 0.8514,
+      "step": 50800
+    },
+    {
+      "epoch": 1.714497440043115,
+      "grad_norm": 0.7346422076225281,
+      "learning_rate": 0.0002190357495733405,
+      "loss": 0.856,
+      "step": 50900
+    },
+    {
+      "epoch": 1.7178658043654003,
+      "grad_norm": 0.7833214402198792,
+      "learning_rate": 0.00021881119195185482,
+      "loss": 0.8518,
+      "step": 51000
+    },
+    {
+      "epoch": 1.7212341686876853,
+      "grad_norm": 0.599700391292572,
+      "learning_rate": 0.00021858663433036915,
+      "loss": 0.8539,
+      "step": 51100
+    },
+    {
+      "epoch": 1.7246025330099704,
+      "grad_norm": 0.7568506002426147,
+      "learning_rate": 0.00021836207670888346,
+      "loss": 0.8504,
+      "step": 51200
+    },
+    {
+      "epoch": 1.7279708973322556,
+      "grad_norm": 0.71254563331604,
+      "learning_rate": 0.00021813751908739782,
+      "loss": 0.849,
+      "step": 51300
+    },
+    {
+      "epoch": 1.7313392616545404,
+      "grad_norm": 0.6965556144714355,
+      "learning_rate": 0.00021791296146591213,
+      "loss": 0.8506,
+      "step": 51400
+    },
+    {
+      "epoch": 1.7347076259768257,
+      "grad_norm": 0.6838847994804382,
+      "learning_rate": 0.00021768840384442647,
+      "loss": 0.8487,
+      "step": 51500
+    },
+    {
+      "epoch": 1.7380759902991108,
+      "grad_norm": 0.6968240737915039,
+      "learning_rate": 0.00021746384622294077,
+      "loss": 0.8498,
+      "step": 51600
+    },
+    {
+      "epoch": 1.7414443546213958,
+      "grad_norm": 0.6424885988235474,
+      "learning_rate": 0.00021723928860145514,
+      "loss": 0.8495,
+      "step": 51700
+    },
+    {
+      "epoch": 1.744812718943681,
+      "grad_norm": 0.6591705083847046,
+      "learning_rate": 0.00021701473097996945,
+      "loss": 0.8474,
+      "step": 51800
+    },
+    {
+      "epoch": 1.7481810832659659,
+      "grad_norm": 0.9693089723587036,
+      "learning_rate": 0.00021679017335848375,
+      "loss": 0.8468,
+      "step": 51900
+    },
+    {
+      "epoch": 1.7515494475882512,
+      "grad_norm": 0.6744303107261658,
+      "learning_rate": 0.0002165656157369981,
+      "loss": 0.8488,
+      "step": 52000
+    },
+    {
+      "epoch": 1.7549178119105362,
+      "grad_norm": 0.7250896096229553,
+      "learning_rate": 0.00021634105811551242,
+      "loss": 0.8501,
+      "step": 52100
+    },
+    {
+      "epoch": 1.7582861762328212,
+      "grad_norm": 0.6125505566596985,
+      "learning_rate": 0.00021611650049402676,
+      "loss": 0.8481,
+      "step": 52200
+    },
+    {
+      "epoch": 1.7616545405551065,
+      "grad_norm": 0.6853183507919312,
+      "learning_rate": 0.00021589194287254107,
+      "loss": 0.8496,
+      "step": 52300
+    },
+    {
+      "epoch": 1.7650229048773916,
+      "grad_norm": 0.6613684892654419,
+      "learning_rate": 0.0002156673852510554,
+      "loss": 0.8496,
+      "step": 52400
+    },
+    {
+      "epoch": 1.7683912691996766,
+      "grad_norm": 0.6338353157043457,
+      "learning_rate": 0.00021544282762956974,
+      "loss": 0.8473,
+      "step": 52500
+    },
+    {
+      "epoch": 1.7717596335219619,
+      "grad_norm": 0.7311720848083496,
+      "learning_rate": 0.00021521827000808407,
+      "loss": 0.8494,
+      "step": 52600
+    },
+    {
+      "epoch": 1.7751279978442467,
+      "grad_norm": 0.8986194133758545,
+      "learning_rate": 0.00021499371238659838,
+      "loss": 0.847,
+      "step": 52700
+    },
+    {
+      "epoch": 1.778496362166532,
+      "grad_norm": 0.7145596146583557,
+      "learning_rate": 0.0002147691547651127,
+      "loss": 0.8513,
+      "step": 52800
+    },
+    {
+      "epoch": 1.781864726488817,
+      "grad_norm": 0.673674464225769,
+      "learning_rate": 0.00021454459714362705,
+      "loss": 0.8496,
+      "step": 52900
+    },
+    {
+      "epoch": 1.785233090811102,
+      "grad_norm": 0.6568505167961121,
+      "learning_rate": 0.00021432003952214136,
+      "loss": 0.8477,
+      "step": 53000
+    },
+    {
+      "epoch": 1.7886014551333873,
+      "grad_norm": 0.8661649227142334,
+      "learning_rate": 0.0002140954819006557,
+      "loss": 0.8437,
+      "step": 53100
+    },
+    {
+      "epoch": 1.7919698194556724,
+      "grad_norm": 0.636997401714325,
+      "learning_rate": 0.00021387092427917,
+      "loss": 0.8498,
+      "step": 53200
+    },
+    {
+      "epoch": 1.7953381837779574,
+      "grad_norm": 0.6258701682090759,
+      "learning_rate": 0.00021364636665768436,
+      "loss": 0.846,
+      "step": 53300
+    },
+    {
+      "epoch": 1.7987065481002427,
+      "grad_norm": 0.7167558073997498,
+      "learning_rate": 0.00021342180903619867,
+      "loss": 0.8496,
+      "step": 53400
+    },
+    {
+      "epoch": 1.8020749124225275,
+      "grad_norm": 0.725511372089386,
+      "learning_rate": 0.000213197251414713,
+      "loss": 0.8503,
+      "step": 53500
+    },
+    {
+      "epoch": 1.8054432767448128,
+      "grad_norm": 0.7038520574569702,
+      "learning_rate": 0.00021297269379322732,
+      "loss": 0.8486,
+      "step": 53600
+    },
+    {
+      "epoch": 1.8088116410670978,
+      "grad_norm": 0.6768267154693604,
+      "learning_rate": 0.00021274813617174162,
+      "loss": 0.8475,
+      "step": 53700
+    },
+    {
+      "epoch": 1.8121800053893828,
+      "grad_norm": 0.6625707745552063,
+      "learning_rate": 0.000212523578550256,
+      "loss": 0.8466,
+      "step": 53800
+    },
+    {
+      "epoch": 1.815548369711668,
+      "grad_norm": 0.6589378118515015,
+      "learning_rate": 0.0002122990209287703,
+      "loss": 0.8449,
+      "step": 53900
+    },
+    {
+      "epoch": 1.8189167340339532,
+      "grad_norm": 0.8694218993186951,
+      "learning_rate": 0.00021207446330728463,
+      "loss": 0.8464,
+      "step": 54000
+    },
+    {
+      "epoch": 1.8222850983562382,
+      "grad_norm": 0.6879692673683167,
+      "learning_rate": 0.00021184990568579894,
+      "loss": 0.8464,
+      "step": 54100
+    },
+    {
+      "epoch": 1.8256534626785235,
+      "grad_norm": 0.7406269907951355,
+      "learning_rate": 0.0002116253480643133,
+      "loss": 0.8451,
+      "step": 54200
+    },
+    {
+      "epoch": 1.8290218270008083,
+      "grad_norm": 0.645255982875824,
+      "learning_rate": 0.0002114007904428276,
+      "loss": 0.8459,
+      "step": 54300
+    },
+    {
+      "epoch": 1.8323901913230936,
+      "grad_norm": 0.6275530457496643,
+      "learning_rate": 0.00021117623282134194,
+      "loss": 0.848,
+      "step": 54400
+    },
+    {
+      "epoch": 1.8357585556453786,
+      "grad_norm": 0.673546314239502,
+      "learning_rate": 0.00021095167519985625,
+      "loss": 0.8421,
+      "step": 54500
+    },
+    {
+      "epoch": 1.8391269199676636,
+      "grad_norm": 0.6627029180526733,
+      "learning_rate": 0.00021072711757837061,
+      "loss": 0.8443,
+      "step": 54600
+    },
+    {
+      "epoch": 1.842495284289949,
+      "grad_norm": 0.6747744083404541,
+      "learning_rate": 0.00021050255995688492,
+      "loss": 0.8439,
+      "step": 54700
+    },
+    {
+      "epoch": 1.845863648612234,
+      "grad_norm": 0.6658967733383179,
+      "learning_rate": 0.00021027800233539923,
+      "loss": 0.8435,
+      "step": 54800
+    },
+    {
+      "epoch": 1.849232012934519,
+      "grad_norm": 0.6818722486495972,
+      "learning_rate": 0.00021005344471391357,
+      "loss": 0.8433,
+      "step": 54900
+    },
+    {
+      "epoch": 1.852600377256804,
+      "grad_norm": 0.6817068457603455,
+      "learning_rate": 0.0002098288870924279,
+      "loss": 0.8449,
+      "step": 55000
+    },
+    {
+      "epoch": 1.855968741579089,
+      "grad_norm": 0.6656964421272278,
+      "learning_rate": 0.00020960432947094224,
+      "loss": 0.8448,
+      "step": 55100
+    },
+    {
+      "epoch": 1.8593371059013744,
+      "grad_norm": 0.7307213544845581,
+      "learning_rate": 0.00020937977184945654,
+      "loss": 0.8439,
+      "step": 55200
+    },
+    {
+      "epoch": 1.8627054702236594,
+      "grad_norm": 0.7471979856491089,
+      "learning_rate": 0.00020915521422797088,
+      "loss": 0.8461,
+      "step": 55300
+    },
+    {
+      "epoch": 1.8660738345459444,
+      "grad_norm": 0.6660134792327881,
+      "learning_rate": 0.00020893065660648521,
+      "loss": 0.849,
+      "step": 55400
+    },
+    {
+      "epoch": 1.8694421988682297,
+      "grad_norm": 0.6332405209541321,
+      "learning_rate": 0.00020870609898499955,
+      "loss": 0.8417,
+      "step": 55500
+    },
+    {
+      "epoch": 1.8728105631905145,
+      "grad_norm": 0.6369178295135498,
+      "learning_rate": 0.00020848154136351386,
+      "loss": 0.844,
+      "step": 55600
+    },
+    {
+      "epoch": 1.8761789275127998,
+      "grad_norm": 0.6878752112388611,
+      "learning_rate": 0.00020825698374202817,
+      "loss": 0.8435,
+      "step": 55700
+    },
+    {
+      "epoch": 1.8795472918350848,
+      "grad_norm": 0.7077382206916809,
+      "learning_rate": 0.00020803242612054253,
+      "loss": 0.8448,
+      "step": 55800
+    },
+    {
+      "epoch": 1.8829156561573699,
+      "grad_norm": 0.6542516350746155,
+      "learning_rate": 0.00020780786849905684,
+      "loss": 0.8441,
+      "step": 55900
+    },
+    {
+      "epoch": 1.8862840204796552,
+      "grad_norm": 0.6779966950416565,
+      "learning_rate": 0.00020758331087757117,
+      "loss": 0.8452,
+      "step": 56000
+    },
+    {
+      "epoch": 1.8896523848019402,
+      "grad_norm": 0.7874273657798767,
+      "learning_rate": 0.00020735875325608548,
+      "loss": 0.8408,
+      "step": 56100
+    },
+    {
+      "epoch": 1.8930207491242252,
+      "grad_norm": 0.7043356895446777,
+      "learning_rate": 0.00020713419563459984,
+      "loss": 0.8459,
+      "step": 56200
+    },
+    {
+      "epoch": 1.8963891134465105,
+      "grad_norm": 0.6597324013710022,
+      "learning_rate": 0.00020690963801311415,
+      "loss": 0.8418,
+      "step": 56300
+    },
+    {
+      "epoch": 1.8997574777687953,
+      "grad_norm": 0.6724010705947876,
+      "learning_rate": 0.00020668508039162848,
+      "loss": 0.8459,
+      "step": 56400
+    },
+    {
+      "epoch": 1.9031258420910806,
+      "grad_norm": 0.7157804369926453,
+      "learning_rate": 0.0002064605227701428,
+      "loss": 0.8459,
+      "step": 56500
+    },
+    {
+      "epoch": 1.9064942064133656,
+      "grad_norm": 0.730675995349884,
+      "learning_rate": 0.0002062359651486571,
+      "loss": 0.8477,
+      "step": 56600
+    },
+    {
+      "epoch": 1.9098625707356507,
+      "grad_norm": 0.6512274146080017,
+      "learning_rate": 0.00020601140752717146,
+      "loss": 0.8424,
+      "step": 56700
+    },
+    {
+      "epoch": 1.913230935057936,
+      "grad_norm": 0.679719090461731,
+      "learning_rate": 0.00020578684990568577,
+      "loss": 0.8465,
+      "step": 56800
+    },
+    {
+      "epoch": 1.916599299380221,
+      "grad_norm": 0.7341735363006592,
+      "learning_rate": 0.0002055622922842001,
+      "loss": 0.8465,
+      "step": 56900
+    },
+    {
+      "epoch": 1.919967663702506,
+      "grad_norm": 0.6398690342903137,
+      "learning_rate": 0.00020533773466271441,
+      "loss": 0.8417,
+      "step": 57000
+    },
+    {
+      "epoch": 1.9233360280247913,
+      "grad_norm": 0.8629583716392517,
+      "learning_rate": 0.00020511317704122878,
+      "loss": 0.8438,
+      "step": 57100
+    },
+    {
+      "epoch": 1.9267043923470761,
+      "grad_norm": 0.6348667740821838,
+      "learning_rate": 0.00020488861941974308,
+      "loss": 0.8411,
+      "step": 57200
+    },
+    {
+      "epoch": 1.9300727566693614,
+      "grad_norm": 1.0562591552734375,
+      "learning_rate": 0.00020466406179825742,
+      "loss": 0.8401,
+      "step": 57300
+    },
+    {
+      "epoch": 1.9334411209916464,
+      "grad_norm": 0.7126754522323608,
+      "learning_rate": 0.00020443950417677173,
+      "loss": 0.8469,
+      "step": 57400
+    },
+    {
+      "epoch": 1.9368094853139315,
+      "grad_norm": 0.624739408493042,
+      "learning_rate": 0.00020421494655528606,
+      "loss": 0.8422,
+      "step": 57500
+    },
+    {
+      "epoch": 1.9401778496362168,
+      "grad_norm": 0.6490176916122437,
+      "learning_rate": 0.0002039903889338004,
+      "loss": 0.841,
+      "step": 57600
+    },
+    {
+      "epoch": 1.9435462139585018,
+      "grad_norm": 0.6586236357688904,
+      "learning_rate": 0.0002037658313123147,
+      "loss": 0.8412,
+      "step": 57700
+    },
+    {
+      "epoch": 1.9469145782807868,
+      "grad_norm": 0.66822749376297,
+      "learning_rate": 0.00020354127369082904,
+      "loss": 0.8418,
+      "step": 57800
+    },
+    {
+      "epoch": 1.9502829426030721,
+      "grad_norm": 0.6851320266723633,
+      "learning_rate": 0.00020331671606934338,
+      "loss": 0.8414,
+      "step": 57900
+    },
+    {
+      "epoch": 1.953651306925357,
+      "grad_norm": 0.6610788702964783,
+      "learning_rate": 0.0002030921584478577,
+      "loss": 0.8409,
+      "step": 58000
+    },
+    {
+      "epoch": 1.9570196712476422,
+      "grad_norm": 0.6205683946609497,
+      "learning_rate": 0.00020286760082637202,
+      "loss": 0.8384,
+      "step": 58100
+    },
+    {
+      "epoch": 1.9603880355699272,
+      "grad_norm": 0.6619114875793457,
+      "learning_rate": 0.00020264304320488636,
+      "loss": 0.8404,
+      "step": 58200
+    },
+    {
+      "epoch": 1.9637563998922123,
+      "grad_norm": 0.6718551516532898,
+      "learning_rate": 0.0002024184855834007,
+      "loss": 0.8428,
+      "step": 58300
+    },
+    {
+      "epoch": 1.9671247642144976,
+      "grad_norm": 0.696954607963562,
+      "learning_rate": 0.00020219392796191503,
+      "loss": 0.8377,
+      "step": 58400
+    },
+    {
+      "epoch": 1.9704931285367824,
+      "grad_norm": 0.7444251775741577,
+      "learning_rate": 0.00020196937034042933,
+      "loss": 0.8399,
+      "step": 58500
+    },
+    {
+      "epoch": 1.9738614928590676,
+      "grad_norm": 1.1385860443115234,
+      "learning_rate": 0.00020174481271894364,
+      "loss": 0.8424,
+      "step": 58600
+    },
+    {
+      "epoch": 1.9772298571813527,
+      "grad_norm": 0.6890325546264648,
+      "learning_rate": 0.000201520255097458,
+      "loss": 0.8441,
+      "step": 58700
+    },
+    {
+      "epoch": 1.9805982215036377,
+      "grad_norm": 0.6715498566627502,
+      "learning_rate": 0.0002012956974759723,
+      "loss": 0.8426,
+      "step": 58800
+    },
+    {
+      "epoch": 1.983966585825923,
+      "grad_norm": 0.683191180229187,
+      "learning_rate": 0.00020107113985448665,
+      "loss": 0.8408,
+      "step": 58900
+    },
+    {
+      "epoch": 1.987334950148208,
+      "grad_norm": 0.7549853324890137,
+      "learning_rate": 0.00020084658223300096,
+      "loss": 0.8406,
+      "step": 59000
+    },
+    {
+      "epoch": 1.990703314470493,
+      "grad_norm": 0.6465044617652893,
+      "learning_rate": 0.00020062202461151532,
+      "loss": 0.8398,
+      "step": 59100
+    },
+    {
+      "epoch": 1.9940716787927784,
+      "grad_norm": 0.6415732502937317,
+      "learning_rate": 0.00020039746699002963,
+      "loss": 0.8385,
+      "step": 59200
+    },
+    {
+      "epoch": 1.9974400431150632,
+      "grad_norm": 0.6739877462387085,
+      "learning_rate": 0.00020017290936854396,
+      "loss": 0.8405,
+      "step": 59300
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.7839618921279907,
+      "eval_runtime": 1.804,
+      "eval_samples_per_second": 2771.644,
+      "eval_steps_per_second": 43.792,
+      "step": 59376
+    },
+    {
+      "epoch": 2.0008084074373484,
+      "grad_norm": 0.7029405832290649,
+      "learning_rate": 0.00019994835174705827,
+      "loss": 0.8394,
+      "step": 59400
+    },
+    {
+      "epoch": 2.0041767717596337,
+      "grad_norm": 1.100113034248352,
+      "learning_rate": 0.00019972379412557258,
+      "loss": 0.8415,
+      "step": 59500
+    },
+    {
+      "epoch": 2.0075451360819185,
+      "grad_norm": 0.6430971026420593,
+      "learning_rate": 0.00019949923650408694,
+      "loss": 0.8368,
+      "step": 59600
+    },
+    {
+      "epoch": 2.010913500404204,
+      "grad_norm": 0.6618381142616272,
+      "learning_rate": 0.00019927467888260125,
+      "loss": 0.8374,
+      "step": 59700
+    },
+    {
+      "epoch": 2.0142818647264886,
+      "grad_norm": 0.7243971228599548,
+      "learning_rate": 0.00019905012126111558,
+      "loss": 0.84,
+      "step": 59800
+    },
+    {
+      "epoch": 2.017650229048774,
+      "grad_norm": 0.6579388380050659,
+      "learning_rate": 0.0001988255636396299,
+      "loss": 0.8395,
+      "step": 59900
+    },
+    {
+      "epoch": 2.021018593371059,
+      "grad_norm": 0.6798832416534424,
+      "learning_rate": 0.00019860100601814425,
+      "loss": 0.8378,
+      "step": 60000
+    },
+    {
+      "epoch": 2.024386957693344,
+      "grad_norm": 0.7237014174461365,
+      "learning_rate": 0.00019837644839665856,
+      "loss": 0.8403,
+      "step": 60100
+    },
+    {
+      "epoch": 2.0277553220156292,
+      "grad_norm": 0.6715816259384155,
+      "learning_rate": 0.0001981518907751729,
+      "loss": 0.8412,
+      "step": 60200
+    },
+    {
+      "epoch": 2.0311236863379145,
+      "grad_norm": 0.6749352812767029,
+      "learning_rate": 0.0001979273331536872,
+      "loss": 0.8387,
+      "step": 60300
+    },
+    {
+      "epoch": 2.0344920506601993,
+      "grad_norm": 0.6498907804489136,
+      "learning_rate": 0.00019770277553220154,
+      "loss": 0.8365,
+      "step": 60400
+    },
+    {
+      "epoch": 2.0378604149824846,
+      "grad_norm": 0.6706451177597046,
+      "learning_rate": 0.00019747821791071588,
+      "loss": 0.8353,
+      "step": 60500
+    },
+    {
+      "epoch": 2.0412287793047694,
+      "grad_norm": 0.6508938670158386,
+      "learning_rate": 0.00019725366028923018,
+      "loss": 0.8386,
+      "step": 60600
+    },
+    {
+      "epoch": 2.0445971436270547,
+      "grad_norm": 0.6738646626472473,
+      "learning_rate": 0.00019702910266774452,
+      "loss": 0.8357,
+      "step": 60700
+    },
+    {
+      "epoch": 2.04796550794934,
+      "grad_norm": 0.6249734163284302,
+      "learning_rate": 0.00019680454504625885,
+      "loss": 0.8381,
+      "step": 60800
+    },
+    {
+      "epoch": 2.051333872271625,
+      "grad_norm": 0.72087162733078,
+      "learning_rate": 0.0001965799874247732,
+      "loss": 0.8377,
+      "step": 60900
+    },
+    {
+      "epoch": 2.05470223659391,
+      "grad_norm": 0.7066697478294373,
+      "learning_rate": 0.0001963554298032875,
+      "loss": 0.8397,
+      "step": 61000
+    },
+    {
+      "epoch": 2.0580706009161953,
+      "grad_norm": 0.7159758806228638,
+      "learning_rate": 0.00019613087218180186,
+      "loss": 0.8382,
+      "step": 61100
+    },
+    {
+      "epoch": 2.06143896523848,
+      "grad_norm": 0.7034109830856323,
+      "learning_rate": 0.00019590631456031617,
+      "loss": 0.8386,
+      "step": 61200
+    },
+    {
+      "epoch": 2.0648073295607654,
+      "grad_norm": 0.648295521736145,
+      "learning_rate": 0.0001956817569388305,
+      "loss": 0.8353,
+      "step": 61300
+    },
+    {
+      "epoch": 2.0681756938830502,
+      "grad_norm": 0.909184992313385,
+      "learning_rate": 0.0001954571993173448,
+      "loss": 0.8366,
+      "step": 61400
+    },
+    {
+      "epoch": 2.0715440582053355,
+      "grad_norm": 0.6607633233070374,
+      "learning_rate": 0.00019523264169585912,
+      "loss": 0.8389,
+      "step": 61500
+    },
+    {
+      "epoch": 2.0749124225276208,
+      "grad_norm": 0.6451642513275146,
+      "learning_rate": 0.00019500808407437348,
+      "loss": 0.8379,
+      "step": 61600
+    },
+    {
+      "epoch": 2.0782807868499056,
+      "grad_norm": 0.6556397080421448,
+      "learning_rate": 0.0001947835264528878,
+      "loss": 0.837,
+      "step": 61700
+    },
+    {
+      "epoch": 2.081649151172191,
+      "grad_norm": 0.6268017292022705,
+      "learning_rate": 0.00019455896883140212,
+      "loss": 0.8379,
+      "step": 61800
+    },
+    {
+      "epoch": 2.0850175154944757,
+      "grad_norm": 0.6620351672172546,
+      "learning_rate": 0.00019433441120991643,
+      "loss": 0.8377,
+      "step": 61900
+    },
+    {
+      "epoch": 2.088385879816761,
+      "grad_norm": 0.6334593296051025,
+      "learning_rate": 0.0001941098535884308,
+      "loss": 0.8367,
+      "step": 62000
+    },
+    {
+      "epoch": 2.091754244139046,
+      "grad_norm": 0.6498568058013916,
+      "learning_rate": 0.0001938852959669451,
+      "loss": 0.836,
+      "step": 62100
+    },
+    {
+      "epoch": 2.095122608461331,
+      "grad_norm": 0.6664881110191345,
+      "learning_rate": 0.00019366073834545944,
+      "loss": 0.8366,
+      "step": 62200
+    },
+    {
+      "epoch": 2.0984909727836163,
+      "grad_norm": 0.6434110999107361,
+      "learning_rate": 0.00019343618072397375,
+      "loss": 0.8339,
+      "step": 62300
+    },
+    {
+      "epoch": 2.1018593371059016,
+      "grad_norm": 0.6958553194999695,
+      "learning_rate": 0.00019321162310248805,
+      "loss": 0.8363,
+      "step": 62400
+    },
+    {
+      "epoch": 2.1052277014281864,
+      "grad_norm": 0.7187645435333252,
+      "learning_rate": 0.00019298706548100242,
+      "loss": 0.8337,
+      "step": 62500
+    },
+    {
+      "epoch": 2.1085960657504716,
+      "grad_norm": 0.6984855532646179,
+      "learning_rate": 0.00019276250785951672,
+      "loss": 0.8363,
+      "step": 62600
+    },
+    {
+      "epoch": 2.1119644300727565,
+      "grad_norm": 0.7342332601547241,
+      "learning_rate": 0.00019253795023803106,
+      "loss": 0.8385,
+      "step": 62700
+    },
+    {
+      "epoch": 2.1153327943950417,
+      "grad_norm": 0.7250052690505981,
+      "learning_rate": 0.00019231339261654537,
+      "loss": 0.8333,
+      "step": 62800
+    },
+    {
+      "epoch": 2.118701158717327,
+      "grad_norm": 0.6984183192253113,
+      "learning_rate": 0.00019208883499505973,
+      "loss": 0.8343,
+      "step": 62900
+    },
+    {
+      "epoch": 2.122069523039612,
+      "grad_norm": 1.0108722448349,
+      "learning_rate": 0.00019186427737357404,
+      "loss": 0.836,
+      "step": 63000
+    },
+    {
+      "epoch": 2.125437887361897,
+      "grad_norm": 0.6725150942802429,
+      "learning_rate": 0.00019163971975208837,
+      "loss": 0.8372,
+      "step": 63100
+    },
+    {
+      "epoch": 2.1288062516841824,
+      "grad_norm": 0.6805692911148071,
+      "learning_rate": 0.0001914151621306027,
+      "loss": 0.8311,
+      "step": 63200
+    },
+    {
+      "epoch": 2.132174616006467,
+      "grad_norm": 0.9061579704284668,
+      "learning_rate": 0.00019119060450911702,
+      "loss": 0.8343,
+      "step": 63300
+    },
+    {
+      "epoch": 2.1355429803287524,
+      "grad_norm": 0.6339781880378723,
+      "learning_rate": 0.00019096604688763135,
+      "loss": 0.8405,
+      "step": 63400
+    },
+    {
+      "epoch": 2.1389113446510373,
+      "grad_norm": 0.6159859895706177,
+      "learning_rate": 0.00019074148926614566,
+      "loss": 0.8352,
+      "step": 63500
+    },
+    {
+      "epoch": 2.1422797089733225,
+      "grad_norm": 0.6404457092285156,
+      "learning_rate": 0.00019051693164466002,
+      "loss": 0.835,
+      "step": 63600
+    },
+    {
+      "epoch": 2.145648073295608,
+      "grad_norm": 0.619216799736023,
+      "learning_rate": 0.00019029237402317433,
+      "loss": 0.8323,
+      "step": 63700
+    },
+    {
+      "epoch": 2.1490164376178926,
+      "grad_norm": 0.6672142744064331,
+      "learning_rate": 0.00019006781640168867,
+      "loss": 0.8318,
+      "step": 63800
+    },
+    {
+      "epoch": 2.152384801940178,
+      "grad_norm": 0.6713272929191589,
+      "learning_rate": 0.00018984325878020297,
+      "loss": 0.835,
+      "step": 63900
+    },
+    {
+      "epoch": 2.155753166262463,
+      "grad_norm": 0.6657119989395142,
+      "learning_rate": 0.00018961870115871734,
+      "loss": 0.8312,
+      "step": 64000
+    },
+    {
+      "epoch": 2.159121530584748,
+      "grad_norm": 0.7010686993598938,
+      "learning_rate": 0.00018939414353723164,
+      "loss": 0.8334,
+      "step": 64100
+    },
+    {
+      "epoch": 2.1624898949070332,
+      "grad_norm": 0.67507404088974,
+      "learning_rate": 0.00018916958591574598,
+      "loss": 0.8337,
+      "step": 64200
+    },
+    {
+      "epoch": 2.165858259229318,
+      "grad_norm": 0.7008448243141174,
+      "learning_rate": 0.0001889450282942603,
+      "loss": 0.8334,
+      "step": 64300
+    },
+    {
+      "epoch": 2.1692266235516033,
+      "grad_norm": 0.634978175163269,
+      "learning_rate": 0.0001887204706727746,
+      "loss": 0.8372,
+      "step": 64400
+    },
+    {
+      "epoch": 2.1725949878738886,
+      "grad_norm": 0.7248919010162354,
+      "learning_rate": 0.00018849591305128896,
+      "loss": 0.8355,
+      "step": 64500
+    },
+    {
+      "epoch": 2.1759633521961734,
+      "grad_norm": 0.6367628574371338,
+      "learning_rate": 0.00018827135542980327,
+      "loss": 0.8328,
+      "step": 64600
+    },
+    {
+      "epoch": 2.1793317165184587,
+      "grad_norm": 0.712471604347229,
+      "learning_rate": 0.0001880467978083176,
+      "loss": 0.8296,
+      "step": 64700
+    },
+    {
+      "epoch": 2.1827000808407435,
+      "grad_norm": 0.671159565448761,
+      "learning_rate": 0.0001878222401868319,
+      "loss": 0.8339,
+      "step": 64800
+    },
+    {
+      "epoch": 2.186068445163029,
+      "grad_norm": 0.6872281432151794,
+      "learning_rate": 0.00018759768256534627,
+      "loss": 0.8313,
+      "step": 64900
+    },
+    {
+      "epoch": 2.189436809485314,
+      "grad_norm": 0.7017882466316223,
+      "learning_rate": 0.00018737312494386058,
+      "loss": 0.8327,
+      "step": 65000
+    },
+    {
+      "epoch": 2.192805173807599,
+      "grad_norm": 0.7563288807868958,
+      "learning_rate": 0.00018714856732237491,
+      "loss": 0.8342,
+      "step": 65100
+    },
+    {
+      "epoch": 2.196173538129884,
+      "grad_norm": 0.6648709177970886,
+      "learning_rate": 0.00018692400970088922,
+      "loss": 0.8318,
+      "step": 65200
+    },
+    {
+      "epoch": 2.1995419024521694,
+      "grad_norm": 0.6673408150672913,
+      "learning_rate": 0.00018669945207940356,
+      "loss": 0.8328,
+      "step": 65300
+    },
+    {
+      "epoch": 2.2029102667744542,
+      "grad_norm": 0.8705912232398987,
+      "learning_rate": 0.0001864748944579179,
+      "loss": 0.831,
+      "step": 65400
+    },
+    {
+      "epoch": 2.2062786310967395,
+      "grad_norm": 0.6253674030303955,
+      "learning_rate": 0.0001862503368364322,
+      "loss": 0.8322,
+      "step": 65500
+    },
+    {
+      "epoch": 2.2096469954190243,
+      "grad_norm": 0.7175179719924927,
+      "learning_rate": 0.00018602577921494654,
+      "loss": 0.8316,
+      "step": 65600
+    },
+    {
+      "epoch": 2.2130153597413096,
+      "grad_norm": 0.6952610611915588,
+      "learning_rate": 0.00018580122159346087,
+      "loss": 0.831,
+      "step": 65700
+    },
+    {
+      "epoch": 2.216383724063595,
+      "grad_norm": 0.6487058401107788,
+      "learning_rate": 0.0001855766639719752,
+      "loss": 0.8323,
+      "step": 65800
+    },
+    {
+      "epoch": 2.2197520883858797,
+      "grad_norm": 0.7087188959121704,
+      "learning_rate": 0.00018535210635048951,
+      "loss": 0.8337,
+      "step": 65900
+    },
+    {
+      "epoch": 2.223120452708165,
+      "grad_norm": 0.689513087272644,
+      "learning_rate": 0.00018512754872900385,
+      "loss": 0.834,
+      "step": 66000
+    },
+    {
+      "epoch": 2.22648881703045,
+      "grad_norm": 1.3590748310089111,
+      "learning_rate": 0.00018490299110751818,
+      "loss": 0.8328,
+      "step": 66100
+    },
+    {
+      "epoch": 2.229857181352735,
+      "grad_norm": 0.640274167060852,
+      "learning_rate": 0.0001846784334860325,
+      "loss": 0.8333,
+      "step": 66200
+    },
+    {
+      "epoch": 2.2332255456750203,
+      "grad_norm": 0.794620156288147,
+      "learning_rate": 0.00018445387586454683,
+      "loss": 0.8343,
+      "step": 66300
+    },
+    {
+      "epoch": 2.236593909997305,
+      "grad_norm": 0.6511592268943787,
+      "learning_rate": 0.00018422931824306114,
+      "loss": 0.8328,
+      "step": 66400
+    },
+    {
+      "epoch": 2.2399622743195904,
+      "grad_norm": 0.6505751609802246,
+      "learning_rate": 0.0001840047606215755,
+      "loss": 0.8302,
+      "step": 66500
+    },
+    {
+      "epoch": 2.2433306386418757,
+      "grad_norm": 0.7967960834503174,
+      "learning_rate": 0.0001837802030000898,
+      "loss": 0.83,
+      "step": 66600
+    },
+    {
+      "epoch": 2.2466990029641605,
+      "grad_norm": 0.708914577960968,
+      "learning_rate": 0.00018355564537860414,
+      "loss": 0.8305,
+      "step": 66700
+    },
+    {
+      "epoch": 2.2500673672864457,
+      "grad_norm": 0.6149790287017822,
+      "learning_rate": 0.00018333108775711845,
+      "loss": 0.8334,
+      "step": 66800
+    },
+    {
+      "epoch": 2.2534357316087306,
+      "grad_norm": 0.6834396719932556,
+      "learning_rate": 0.0001831065301356328,
+      "loss": 0.8314,
+      "step": 66900
+    },
+    {
+      "epoch": 2.256804095931016,
+      "grad_norm": 0.7162107229232788,
+      "learning_rate": 0.00018288197251414712,
+      "loss": 0.8327,
+      "step": 67000
+    },
+    {
+      "epoch": 2.260172460253301,
+      "grad_norm": 0.6464916467666626,
+      "learning_rate": 0.00018265741489266143,
+      "loss": 0.8311,
+      "step": 67100
+    },
+    {
+      "epoch": 2.263540824575586,
+      "grad_norm": 0.680050253868103,
+      "learning_rate": 0.00018243285727117576,
+      "loss": 0.8293,
+      "step": 67200
+    },
+    {
+      "epoch": 2.266909188897871,
+      "grad_norm": 0.7750843167304993,
+      "learning_rate": 0.00018220829964969007,
+      "loss": 0.8286,
+      "step": 67300
+    },
+    {
+      "epoch": 2.2702775532201565,
+      "grad_norm": 0.6611768007278442,
+      "learning_rate": 0.00018198374202820443,
+      "loss": 0.8311,
+      "step": 67400
+    },
+    {
+      "epoch": 2.2736459175424413,
+      "grad_norm": 0.7073565125465393,
+      "learning_rate": 0.00018175918440671874,
+      "loss": 0.829,
+      "step": 67500
+    },
+    {
+      "epoch": 2.2770142818647265,
+      "grad_norm": 0.6587579846382141,
+      "learning_rate": 0.00018153462678523308,
+      "loss": 0.8301,
+      "step": 67600
+    },
+    {
+      "epoch": 2.280382646187012,
+      "grad_norm": 0.6691922545433044,
+      "learning_rate": 0.00018131006916374739,
+      "loss": 0.8291,
+      "step": 67700
+    },
+    {
+      "epoch": 2.2837510105092966,
+      "grad_norm": 0.6645218133926392,
+      "learning_rate": 0.00018108551154226175,
+      "loss": 0.8294,
+      "step": 67800
+    },
+    {
+      "epoch": 2.287119374831582,
+      "grad_norm": 0.6749672293663025,
+      "learning_rate": 0.00018086095392077606,
+      "loss": 0.8324,
+      "step": 67900
+    },
+    {
+      "epoch": 2.2904877391538667,
+      "grad_norm": 0.6658663153648376,
+      "learning_rate": 0.0001806363962992904,
+      "loss": 0.8292,
+      "step": 68000
+    },
+    {
+      "epoch": 2.293856103476152,
+      "grad_norm": 0.7315119504928589,
+      "learning_rate": 0.0001804118386778047,
+      "loss": 0.8289,
+      "step": 68100
+    },
+    {
+      "epoch": 2.2972244677984373,
+      "grad_norm": 0.6700948476791382,
+      "learning_rate": 0.00018018728105631903,
+      "loss": 0.8266,
+      "step": 68200
+    },
+    {
+      "epoch": 2.300592832120722,
+      "grad_norm": 0.6325820088386536,
+      "learning_rate": 0.00017996272343483337,
+      "loss": 0.8312,
+      "step": 68300
+    },
+    {
+      "epoch": 2.3039611964430073,
+      "grad_norm": 0.6814998984336853,
+      "learning_rate": 0.00017973816581334768,
+      "loss": 0.8288,
+      "step": 68400
+    },
+    {
+      "epoch": 2.307329560765292,
+      "grad_norm": 0.6832602024078369,
+      "learning_rate": 0.000179513608191862,
+      "loss": 0.8278,
+      "step": 68500
+    },
+    {
+      "epoch": 2.3106979250875774,
+      "grad_norm": 0.7399811148643494,
+      "learning_rate": 0.00017928905057037635,
+      "loss": 0.8313,
+      "step": 68600
+    },
+    {
+      "epoch": 2.3140662894098627,
+      "grad_norm": 0.6988467574119568,
+      "learning_rate": 0.00017906449294889068,
+      "loss": 0.8298,
+      "step": 68700
+    },
+    {
+      "epoch": 2.3174346537321475,
+      "grad_norm": 0.6190904974937439,
+      "learning_rate": 0.000178839935327405,
+      "loss": 0.8307,
+      "step": 68800
+    },
+    {
+      "epoch": 2.320803018054433,
+      "grad_norm": 0.6427486538887024,
+      "learning_rate": 0.00017861537770591933,
+      "loss": 0.8309,
+      "step": 68900
+    },
+    {
+      "epoch": 2.324171382376718,
+      "grad_norm": 0.661953330039978,
+      "learning_rate": 0.00017839082008443366,
+      "loss": 0.8291,
+      "step": 69000
+    },
+    {
+      "epoch": 2.327539746699003,
+      "grad_norm": 0.6629980206489563,
+      "learning_rate": 0.00017816626246294797,
+      "loss": 0.8268,
+      "step": 69100
+    },
+    {
+      "epoch": 2.330908111021288,
+      "grad_norm": 0.8455718755722046,
+      "learning_rate": 0.0001779417048414623,
+      "loss": 0.8281,
+      "step": 69200
+    },
+    {
+      "epoch": 2.3342764753435734,
+      "grad_norm": 0.6195480227470398,
+      "learning_rate": 0.0001777171472199766,
+      "loss": 0.8271,
+      "step": 69300
+    },
+    {
+      "epoch": 2.3376448396658582,
+      "grad_norm": 0.6320804357528687,
+      "learning_rate": 0.00017749258959849098,
+      "loss": 0.8272,
+      "step": 69400
+    },
+    {
+      "epoch": 2.3410132039881435,
+      "grad_norm": 0.6865086555480957,
+      "learning_rate": 0.00017726803197700528,
+      "loss": 0.83,
+      "step": 69500
+    },
+    {
+      "epoch": 2.3443815683104283,
+      "grad_norm": 0.6443759799003601,
+      "learning_rate": 0.00017704347435551962,
+      "loss": 0.828,
+      "step": 69600
+    },
+    {
+      "epoch": 2.3477499326327136,
+      "grad_norm": 0.6651887893676758,
+      "learning_rate": 0.00017681891673403393,
+      "loss": 0.8309,
+      "step": 69700
+    },
+    {
+      "epoch": 2.351118296954999,
+      "grad_norm": 0.6643931269645691,
+      "learning_rate": 0.0001765943591125483,
+      "loss": 0.8285,
+      "step": 69800
+    },
+    {
+      "epoch": 2.3544866612772837,
+      "grad_norm": 0.6697126030921936,
+      "learning_rate": 0.0001763698014910626,
+      "loss": 0.8264,
+      "step": 69900
+    },
+    {
+      "epoch": 2.357855025599569,
+      "grad_norm": 0.6785570979118347,
+      "learning_rate": 0.0001761452438695769,
+      "loss": 0.8255,
+      "step": 70000
+    },
+    {
+      "epoch": 2.3612233899218538,
+      "grad_norm": 0.6412234306335449,
+      "learning_rate": 0.00017592068624809124,
+      "loss": 0.8271,
+      "step": 70100
+    },
+    {
+      "epoch": 2.364591754244139,
+      "grad_norm": 0.7083507180213928,
+      "learning_rate": 0.00017569612862660555,
+      "loss": 0.83,
+      "step": 70200
+    },
+    {
+      "epoch": 2.3679601185664243,
+      "grad_norm": 0.6519679427146912,
+      "learning_rate": 0.0001754715710051199,
+      "loss": 0.8249,
+      "step": 70300
+    },
+    {
+      "epoch": 2.371328482888709,
+      "grad_norm": 0.7082831263542175,
+      "learning_rate": 0.00017524701338363422,
+      "loss": 0.8287,
+      "step": 70400
+    },
+    {
+      "epoch": 2.3746968472109944,
+      "grad_norm": 0.6893306374549866,
+      "learning_rate": 0.00017502245576214855,
+      "loss": 0.8241,
+      "step": 70500
+    },
+    {
+      "epoch": 2.378065211533279,
+      "grad_norm": 0.6624453663825989,
+      "learning_rate": 0.00017479789814066286,
+      "loss": 0.8281,
+      "step": 70600
+    },
+    {
+      "epoch": 2.3814335758555645,
+      "grad_norm": 0.7071661353111267,
+      "learning_rate": 0.00017457334051917722,
+      "loss": 0.8257,
+      "step": 70700
+    },
+    {
+      "epoch": 2.3848019401778497,
+      "grad_norm": 0.6531967520713806,
+      "learning_rate": 0.00017434878289769153,
+      "loss": 0.8253,
+      "step": 70800
+    },
+    {
+      "epoch": 2.3881703045001346,
+      "grad_norm": 0.6394172310829163,
+      "learning_rate": 0.00017412422527620587,
+      "loss": 0.8291,
+      "step": 70900
+    },
+    {
+      "epoch": 2.39153866882242,
+      "grad_norm": 0.7370265126228333,
+      "learning_rate": 0.00017389966765472018,
+      "loss": 0.8282,
+      "step": 71000
+    },
+    {
+      "epoch": 2.394907033144705,
+      "grad_norm": 0.6256112456321716,
+      "learning_rate": 0.0001736751100332345,
+      "loss": 0.8267,
+      "step": 71100
+    },
+    {
+      "epoch": 2.39827539746699,
+      "grad_norm": 0.6319020390510559,
+      "learning_rate": 0.00017345055241174885,
+      "loss": 0.8271,
+      "step": 71200
+    },
+    {
+      "epoch": 2.401643761789275,
+      "grad_norm": 0.638664186000824,
+      "learning_rate": 0.00017322599479026315,
+      "loss": 0.8259,
+      "step": 71300
+    },
+    {
+      "epoch": 2.4050121261115605,
+      "grad_norm": 0.6793828010559082,
+      "learning_rate": 0.0001730014371687775,
+      "loss": 0.8265,
+      "step": 71400
+    },
+    {
+      "epoch": 2.4083804904338453,
+      "grad_norm": 0.7026681303977966,
+      "learning_rate": 0.00017277687954729182,
+      "loss": 0.8296,
+      "step": 71500
+    },
+    {
+      "epoch": 2.4117488547561305,
+      "grad_norm": 0.6481872200965881,
+      "learning_rate": 0.00017255232192580616,
+      "loss": 0.826,
+      "step": 71600
+    },
+    {
+      "epoch": 2.4151172190784154,
+      "grad_norm": 0.68873530626297,
+      "learning_rate": 0.00017232776430432047,
+      "loss": 0.8283,
+      "step": 71700
+    },
+    {
+      "epoch": 2.4184855834007006,
+      "grad_norm": 0.6869419813156128,
+      "learning_rate": 0.0001721032066828348,
+      "loss": 0.8251,
+      "step": 71800
+    },
+    {
+      "epoch": 2.421853947722986,
+      "grad_norm": 0.6462306380271912,
+      "learning_rate": 0.00017187864906134914,
+      "loss": 0.8265,
+      "step": 71900
+    },
+    {
+      "epoch": 2.4252223120452707,
+      "grad_norm": 0.6632818579673767,
+      "learning_rate": 0.00017165409143986345,
+      "loss": 0.8287,
+      "step": 72000
+    },
+    {
+      "epoch": 2.428590676367556,
+      "grad_norm": 0.7588053345680237,
+      "learning_rate": 0.00017142953381837778,
+      "loss": 0.8278,
+      "step": 72100
+    },
+    {
+      "epoch": 2.431959040689841,
+      "grad_norm": 0.6804139614105225,
+      "learning_rate": 0.0001712049761968921,
+      "loss": 0.8266,
+      "step": 72200
+    },
+    {
+      "epoch": 2.435327405012126,
+      "grad_norm": 0.8674142956733704,
+      "learning_rate": 0.00017098041857540645,
+      "loss": 0.8242,
+      "step": 72300
+    },
+    {
+      "epoch": 2.4386957693344113,
+      "grad_norm": 1.0071535110473633,
+      "learning_rate": 0.00017075586095392076,
+      "loss": 0.8294,
+      "step": 72400
+    },
+    {
+      "epoch": 2.442064133656696,
+      "grad_norm": 0.6555808782577515,
+      "learning_rate": 0.0001705313033324351,
+      "loss": 0.8256,
+      "step": 72500
+    },
+    {
+      "epoch": 2.4454324979789814,
+      "grad_norm": 0.6730812788009644,
+      "learning_rate": 0.0001703067457109494,
+      "loss": 0.8254,
+      "step": 72600
+    },
+    {
+      "epoch": 2.4488008623012663,
+      "grad_norm": 0.6546606421470642,
+      "learning_rate": 0.00017008218808946377,
+      "loss": 0.8257,
+      "step": 72700
+    },
+    {
+      "epoch": 2.4521692266235515,
+      "grad_norm": 0.6847641468048096,
+      "learning_rate": 0.00016985763046797807,
+      "loss": 0.824,
+      "step": 72800
+    },
+    {
+      "epoch": 2.455537590945837,
+      "grad_norm": 0.6383925080299377,
+      "learning_rate": 0.00016963307284649238,
+      "loss": 0.8238,
+      "step": 72900
+    },
+    {
+      "epoch": 2.4589059552681216,
+      "grad_norm": 0.6601071357727051,
+      "learning_rate": 0.00016940851522500672,
+      "loss": 0.8215,
+      "step": 73000
+    },
+    {
+      "epoch": 2.462274319590407,
+      "grad_norm": 0.6254110336303711,
+      "learning_rate": 0.00016918395760352102,
+      "loss": 0.8254,
+      "step": 73100
+    },
+    {
+      "epoch": 2.465642683912692,
+      "grad_norm": 0.6367729902267456,
+      "learning_rate": 0.0001689593999820354,
+      "loss": 0.8201,
+      "step": 73200
+    },
+    {
+      "epoch": 2.469011048234977,
+      "grad_norm": 0.6448660492897034,
+      "learning_rate": 0.0001687348423605497,
+      "loss": 0.8237,
+      "step": 73300
+    },
+    {
+      "epoch": 2.4723794125572622,
+      "grad_norm": 0.6661122441291809,
+      "learning_rate": 0.00016851028473906403,
+      "loss": 0.8257,
+      "step": 73400
+    },
+    {
+      "epoch": 2.4757477768795475,
+      "grad_norm": 0.619644820690155,
+      "learning_rate": 0.00016828572711757834,
+      "loss": 0.8238,
+      "step": 73500
+    },
+    {
+      "epoch": 2.4791161412018323,
+      "grad_norm": 0.7607592344284058,
+      "learning_rate": 0.0001680611694960927,
+      "loss": 0.8225,
+      "step": 73600
+    },
+    {
+      "epoch": 2.4824845055241176,
+      "grad_norm": 0.6733140349388123,
+      "learning_rate": 0.000167836611874607,
+      "loss": 0.8212,
+      "step": 73700
+    },
+    {
+      "epoch": 2.4858528698464024,
+      "grad_norm": 0.9612058401107788,
+      "learning_rate": 0.00016761205425312134,
+      "loss": 0.8214,
+      "step": 73800
+    },
+    {
+      "epoch": 2.4892212341686877,
+      "grad_norm": 0.618076503276825,
+      "learning_rate": 0.00016738749663163565,
+      "loss": 0.8231,
+      "step": 73900
+    },
+    {
+      "epoch": 2.492589598490973,
+      "grad_norm": 0.6606545448303223,
+      "learning_rate": 0.00016716293901015,
+      "loss": 0.8235,
+      "step": 74000
+    },
+    {
+      "epoch": 2.4959579628132578,
+      "grad_norm": 0.6311343312263489,
+      "learning_rate": 0.00016693838138866432,
+      "loss": 0.8227,
+      "step": 74100
+    },
+    {
+      "epoch": 2.499326327135543,
+      "grad_norm": 0.713347315788269,
+      "learning_rate": 0.00016671382376717863,
+      "loss": 0.825,
+      "step": 74200
+    },
+    {
+      "epoch": 2.502694691457828,
+      "grad_norm": 0.658001720905304,
+      "learning_rate": 0.00016648926614569297,
+      "loss": 0.8227,
+      "step": 74300
+    },
+    {
+      "epoch": 2.506063055780113,
+      "grad_norm": 0.7130460739135742,
+      "learning_rate": 0.0001662647085242073,
+      "loss": 0.823,
+      "step": 74400
+    },
+    {
+      "epoch": 2.5094314201023984,
+      "grad_norm": 0.7092128992080688,
+      "learning_rate": 0.00016604015090272164,
+      "loss": 0.8271,
+      "step": 74500
+    },
+    {
+      "epoch": 2.512799784424683,
+      "grad_norm": 0.6681484580039978,
+      "learning_rate": 0.00016581559328123594,
+      "loss": 0.8208,
+      "step": 74600
+    },
+    {
+      "epoch": 2.5161681487469685,
+      "grad_norm": 0.649299144744873,
+      "learning_rate": 0.00016559103565975028,
+      "loss": 0.8229,
+      "step": 74700
+    },
+    {
+      "epoch": 2.5195365130692533,
+      "grad_norm": 0.6099591255187988,
+      "learning_rate": 0.00016536647803826461,
+      "loss": 0.8221,
+      "step": 74800
+    },
+    {
+      "epoch": 2.5229048773915386,
+      "grad_norm": 0.655128538608551,
+      "learning_rate": 0.00016514192041677892,
+      "loss": 0.8216,
+      "step": 74900
+    },
+    {
+      "epoch": 2.526273241713824,
+      "grad_norm": 2.5822091102600098,
+      "learning_rate": 0.00016491736279529326,
+      "loss": 0.8213,
+      "step": 75000
+    },
+    {
+      "epoch": 2.529641606036109,
+      "grad_norm": 0.6650177240371704,
+      "learning_rate": 0.00016469280517380757,
+      "loss": 0.8279,
+      "step": 75100
+    },
+    {
+      "epoch": 2.533009970358394,
+      "grad_norm": 0.9380492568016052,
+      "learning_rate": 0.00016446824755232193,
+      "loss": 0.8245,
+      "step": 75200
+    },
+    {
+      "epoch": 2.536378334680679,
+      "grad_norm": 0.6475105881690979,
+      "learning_rate": 0.00016424368993083624,
+      "loss": 0.8233,
+      "step": 75300
+    },
+    {
+      "epoch": 2.539746699002964,
+      "grad_norm": 0.6452984809875488,
+      "learning_rate": 0.00016401913230935057,
+      "loss": 0.8218,
+      "step": 75400
+    },
+    {
+      "epoch": 2.5431150633252493,
+      "grad_norm": 1.478945255279541,
+      "learning_rate": 0.00016379457468786488,
+      "loss": 0.825,
+      "step": 75500
+    },
+    {
+      "epoch": 2.5464834276475345,
+      "grad_norm": 0.6364376544952393,
+      "learning_rate": 0.00016357001706637924,
+      "loss": 0.8215,
+      "step": 75600
+    },
+    {
+      "epoch": 2.5498517919698194,
+      "grad_norm": 0.5982120633125305,
+      "learning_rate": 0.00016334545944489355,
+      "loss": 0.8244,
+      "step": 75700
+    },
+    {
+      "epoch": 2.5532201562921046,
+      "grad_norm": 0.641855001449585,
+      "learning_rate": 0.00016312090182340786,
+      "loss": 0.8238,
+      "step": 75800
+    },
+    {
+      "epoch": 2.5565885206143895,
+      "grad_norm": 0.6564063429832458,
+      "learning_rate": 0.0001628963442019222,
+      "loss": 0.8228,
+      "step": 75900
+    },
+    {
+      "epoch": 2.5599568849366747,
+      "grad_norm": 0.6515690684318542,
+      "learning_rate": 0.0001626717865804365,
+      "loss": 0.825,
+      "step": 76000
+    },
+    {
+      "epoch": 2.56332524925896,
+      "grad_norm": 0.7063090205192566,
+      "learning_rate": 0.00016244722895895086,
+      "loss": 0.8232,
+      "step": 76100
+    },
+    {
+      "epoch": 2.566693613581245,
+      "grad_norm": 1.018594741821289,
+      "learning_rate": 0.00016222267133746517,
+      "loss": 0.8219,
+      "step": 76200
+    },
+    {
+      "epoch": 2.57006197790353,
+      "grad_norm": 0.7534065842628479,
+      "learning_rate": 0.0001619981137159795,
+      "loss": 0.8208,
+      "step": 76300
+    },
+    {
+      "epoch": 2.573430342225815,
+      "grad_norm": 0.6062216758728027,
+      "learning_rate": 0.00016177355609449382,
+      "loss": 0.8224,
+      "step": 76400
+    },
+    {
+      "epoch": 2.5767987065481,
+      "grad_norm": 0.6766044497489929,
+      "learning_rate": 0.00016154899847300818,
+      "loss": 0.823,
+      "step": 76500
+    },
+    {
+      "epoch": 2.5801670708703854,
+      "grad_norm": 0.668302595615387,
+      "learning_rate": 0.00016132444085152249,
+      "loss": 0.8211,
+      "step": 76600
+    },
+    {
+      "epoch": 2.5835354351926707,
+      "grad_norm": 0.8911457061767578,
+      "learning_rate": 0.0001610998832300368,
+      "loss": 0.8212,
+      "step": 76700
+    },
+    {
+      "epoch": 2.5869037995149555,
+      "grad_norm": 0.718221127986908,
+      "learning_rate": 0.00016087532560855113,
+      "loss": 0.8228,
+      "step": 76800
+    },
+    {
+      "epoch": 2.590272163837241,
+      "grad_norm": 0.7011673450469971,
+      "learning_rate": 0.00016065076798706546,
+      "loss": 0.8216,
+      "step": 76900
+    },
+    {
+      "epoch": 2.5936405281595256,
+      "grad_norm": 0.6552968621253967,
+      "learning_rate": 0.0001604262103655798,
+      "loss": 0.8216,
+      "step": 77000
+    },
+    {
+      "epoch": 2.597008892481811,
+      "grad_norm": 0.6576195955276489,
+      "learning_rate": 0.0001602016527440941,
+      "loss": 0.8197,
+      "step": 77100
+    },
+    {
+      "epoch": 2.600377256804096,
+      "grad_norm": 0.6853031516075134,
+      "learning_rate": 0.00015997709512260847,
+      "loss": 0.8186,
+      "step": 77200
+    },
+    {
+      "epoch": 2.603745621126381,
+      "grad_norm": 0.631230890750885,
+      "learning_rate": 0.00015975253750112278,
+      "loss": 0.8212,
+      "step": 77300
+    },
+    {
+      "epoch": 2.6071139854486662,
+      "grad_norm": 0.6358488202095032,
+      "learning_rate": 0.0001595279798796371,
+      "loss": 0.8221,
+      "step": 77400
+    },
+    {
+      "epoch": 2.610482349770951,
+      "grad_norm": 1.2744354009628296,
+      "learning_rate": 0.00015930342225815142,
+      "loss": 0.8214,
+      "step": 77500
+    },
+    {
+      "epoch": 2.6138507140932363,
+      "grad_norm": 0.9848027229309082,
+      "learning_rate": 0.00015907886463666578,
+      "loss": 0.8238,
+      "step": 77600
+    },
+    {
+      "epoch": 2.6172190784155216,
+      "grad_norm": 0.7024006843566895,
+      "learning_rate": 0.0001588543070151801,
+      "loss": 0.8203,
+      "step": 77700
+    },
+    {
+      "epoch": 2.6205874427378064,
+      "grad_norm": 0.628036379814148,
+      "learning_rate": 0.0001586297493936944,
+      "loss": 0.8214,
+      "step": 77800
+    },
+    {
+      "epoch": 2.6239558070600917,
+      "grad_norm": 1.1920697689056396,
+      "learning_rate": 0.00015840519177220873,
+      "loss": 0.8203,
+      "step": 77900
+    },
+    {
+      "epoch": 2.6273241713823765,
+      "grad_norm": 0.6503965258598328,
+      "learning_rate": 0.00015818063415072304,
+      "loss": 0.8213,
+      "step": 78000
+    },
+    {
+      "epoch": 2.6306925357046618,
+      "grad_norm": 0.6561105847358704,
+      "learning_rate": 0.0001579560765292374,
+      "loss": 0.8215,
+      "step": 78100
+    },
+    {
+      "epoch": 2.634060900026947,
+      "grad_norm": 0.9598469734191895,
+      "learning_rate": 0.0001577315189077517,
+      "loss": 0.8189,
+      "step": 78200
+    },
+    {
+      "epoch": 2.637429264349232,
+      "grad_norm": 0.6700890064239502,
+      "learning_rate": 0.00015750696128626605,
+      "loss": 0.8217,
+      "step": 78300
+    },
+    {
+      "epoch": 2.640797628671517,
+      "grad_norm": 0.6581085920333862,
+      "learning_rate": 0.00015728240366478036,
+      "loss": 0.8207,
+      "step": 78400
+    },
+    {
+      "epoch": 2.644165992993802,
+      "grad_norm": 0.6442667841911316,
+      "learning_rate": 0.00015705784604329472,
+      "loss": 0.8198,
+      "step": 78500
+    },
+    {
+      "epoch": 2.647534357316087,
+      "grad_norm": 1.1659986972808838,
+      "learning_rate": 0.00015683328842180903,
+      "loss": 0.8232,
+      "step": 78600
+    },
+    {
+      "epoch": 2.6509027216383725,
+      "grad_norm": 0.6638743877410889,
+      "learning_rate": 0.00015660873080032333,
+      "loss": 0.8184,
+      "step": 78700
+    },
+    {
+      "epoch": 2.6542710859606578,
+      "grad_norm": 0.649681806564331,
+      "learning_rate": 0.00015638417317883767,
+      "loss": 0.8195,
+      "step": 78800
+    },
+    {
+      "epoch": 2.6576394502829426,
+      "grad_norm": 1.0518876314163208,
+      "learning_rate": 0.00015615961555735198,
+      "loss": 0.8222,
+      "step": 78900
+    },
+    {
+      "epoch": 2.661007814605228,
+      "grad_norm": 0.6661698222160339,
+      "learning_rate": 0.00015593505793586634,
+      "loss": 0.8192,
+      "step": 79000
+    },
+    {
+      "epoch": 2.6643761789275127,
+      "grad_norm": 0.6558882594108582,
+      "learning_rate": 0.00015571050031438065,
+      "loss": 0.8151,
+      "step": 79100
+    },
+    {
+      "epoch": 2.667744543249798,
+      "grad_norm": 0.621672511100769,
+      "learning_rate": 0.00015548594269289498,
+      "loss": 0.8185,
+      "step": 79200
+    },
+    {
+      "epoch": 2.671112907572083,
+      "grad_norm": 0.6606272459030151,
+      "learning_rate": 0.00015526138507140932,
+      "loss": 0.8178,
+      "step": 79300
+    },
+    {
+      "epoch": 2.674481271894368,
+      "grad_norm": 0.685043215751648,
+      "learning_rate": 0.00015503682744992365,
+      "loss": 0.8199,
+      "step": 79400
+    },
+    {
+      "epoch": 2.6778496362166533,
+      "grad_norm": 0.6863681077957153,
+      "learning_rate": 0.00015481226982843796,
+      "loss": 0.8209,
+      "step": 79500
+    },
+    {
+      "epoch": 2.681218000538938,
+      "grad_norm": 0.6480740308761597,
+      "learning_rate": 0.00015458771220695227,
+      "loss": 0.8209,
+      "step": 79600
+    },
+    {
+      "epoch": 2.6845863648612234,
+      "grad_norm": 0.6727776527404785,
+      "learning_rate": 0.00015436315458546663,
+      "loss": 0.8207,
+      "step": 79700
+    },
+    {
+      "epoch": 2.6879547291835086,
+      "grad_norm": 0.7330679893493652,
+      "learning_rate": 0.00015413859696398094,
+      "loss": 0.8217,
+      "step": 79800
+    },
+    {
+      "epoch": 2.6913230935057935,
+      "grad_norm": 0.7128227949142456,
+      "learning_rate": 0.00015391403934249528,
+      "loss": 0.8229,
+      "step": 79900
+    },
+    {
+      "epoch": 2.6946914578280787,
+      "grad_norm": 0.6492688059806824,
+      "learning_rate": 0.00015368948172100958,
+      "loss": 0.8162,
+      "step": 80000
+    },
+    {
+      "epoch": 2.6980598221503636,
+      "grad_norm": 0.8806473016738892,
+      "learning_rate": 0.00015346492409952395,
+      "loss": 0.8227,
+      "step": 80100
+    },
+    {
+      "epoch": 2.701428186472649,
+      "grad_norm": 0.8799885511398315,
+      "learning_rate": 0.00015324036647803825,
+      "loss": 0.8165,
+      "step": 80200
+    },
+    {
+      "epoch": 2.704796550794934,
+      "grad_norm": 0.6979735493659973,
+      "learning_rate": 0.0001530158088565526,
+      "loss": 0.8198,
+      "step": 80300
+    },
+    {
+      "epoch": 2.7081649151172194,
+      "grad_norm": 0.6624419689178467,
+      "learning_rate": 0.0001527912512350669,
+      "loss": 0.8194,
+      "step": 80400
+    },
+    {
+      "epoch": 2.711533279439504,
+      "grad_norm": 0.7164821624755859,
+      "learning_rate": 0.00015256669361358126,
+      "loss": 0.8193,
+      "step": 80500
+    },
+    {
+      "epoch": 2.7149016437617894,
+      "grad_norm": 0.6688589453697205,
+      "learning_rate": 0.00015234213599209557,
+      "loss": 0.8156,
+      "step": 80600
+    },
+    {
+      "epoch": 2.7182700080840743,
+      "grad_norm": 0.6661184430122375,
+      "learning_rate": 0.00015211757837060988,
+      "loss": 0.8199,
+      "step": 80700
+    },
+    {
+      "epoch": 2.7216383724063595,
+      "grad_norm": 0.6526447534561157,
+      "learning_rate": 0.0001518930207491242,
+      "loss": 0.8187,
+      "step": 80800
+    },
+    {
+      "epoch": 2.725006736728645,
+      "grad_norm": 0.6510984897613525,
+      "learning_rate": 0.00015166846312763852,
+      "loss": 0.8189,
+      "step": 80900
+    },
+    {
+      "epoch": 2.7283751010509296,
+      "grad_norm": 0.6921165585517883,
+      "learning_rate": 0.00015144390550615288,
+      "loss": 0.8182,
+      "step": 81000
+    },
+    {
+      "epoch": 2.731743465373215,
+      "grad_norm": 0.7041354179382324,
+      "learning_rate": 0.0001512193478846672,
+      "loss": 0.8208,
+      "step": 81100
+    },
+    {
+      "epoch": 2.7351118296954997,
+      "grad_norm": 0.639445424079895,
+      "learning_rate": 0.00015099479026318152,
+      "loss": 0.8202,
+      "step": 81200
+    },
+    {
+      "epoch": 2.738480194017785,
+      "grad_norm": 0.7262235879898071,
+      "learning_rate": 0.00015077023264169583,
+      "loss": 0.8184,
+      "step": 81300
+    },
+    {
+      "epoch": 2.7418485583400702,
+      "grad_norm": 0.6470584273338318,
+      "learning_rate": 0.0001505456750202102,
+      "loss": 0.8199,
+      "step": 81400
+    },
+    {
+      "epoch": 2.745216922662355,
+      "grad_norm": 0.6758275628089905,
+      "learning_rate": 0.0001503211173987245,
+      "loss": 0.817,
+      "step": 81500
+    },
+    {
+      "epoch": 2.7485852869846403,
+      "grad_norm": 0.6550074815750122,
+      "learning_rate": 0.0001500965597772388,
+      "loss": 0.82,
+      "step": 81600
+    },
+    {
+      "epoch": 2.751953651306925,
+      "grad_norm": 0.6312419176101685,
+      "learning_rate": 0.00014987200215575315,
+      "loss": 0.8209,
+      "step": 81700
+    },
+    {
+      "epoch": 2.7553220156292104,
+      "grad_norm": 0.6456059813499451,
+      "learning_rate": 0.00014964744453426748,
+      "loss": 0.8173,
+      "step": 81800
+    },
+    {
+      "epoch": 2.7586903799514957,
+      "grad_norm": 0.6251012086868286,
+      "learning_rate": 0.00014942288691278182,
+      "loss": 0.8147,
+      "step": 81900
+    },
+    {
+      "epoch": 2.7620587442737805,
+      "grad_norm": 0.6424401998519897,
+      "learning_rate": 0.00014919832929129613,
+      "loss": 0.8173,
+      "step": 82000
+    },
+    {
+      "epoch": 2.7654271085960658,
+      "grad_norm": 0.7199423313140869,
+      "learning_rate": 0.00014897377166981046,
+      "loss": 0.8163,
+      "step": 82100
+    },
+    {
+      "epoch": 2.7687954729183506,
+      "grad_norm": 0.6573197841644287,
+      "learning_rate": 0.0001487492140483248,
+      "loss": 0.8145,
+      "step": 82200
+    },
+    {
+      "epoch": 2.772163837240636,
+      "grad_norm": 0.6857665777206421,
+      "learning_rate": 0.0001485246564268391,
+      "loss": 0.8188,
+      "step": 82300
+    },
+    {
+      "epoch": 2.775532201562921,
+      "grad_norm": 0.6398062705993652,
+      "learning_rate": 0.00014830009880535344,
+      "loss": 0.8173,
+      "step": 82400
+    },
+    {
+      "epoch": 2.7789005658852064,
+      "grad_norm": 0.6743867993354797,
+      "learning_rate": 0.00014807554118386777,
+      "loss": 0.8176,
+      "step": 82500
+    },
+    {
+      "epoch": 2.782268930207491,
+      "grad_norm": 0.6104719042778015,
+      "learning_rate": 0.0001478509835623821,
+      "loss": 0.8171,
+      "step": 82600
+    },
+    {
+      "epoch": 2.7856372945297765,
+      "grad_norm": 0.6518858075141907,
+      "learning_rate": 0.00014762642594089642,
+      "loss": 0.8203,
+      "step": 82700
+    },
+    {
+      "epoch": 2.7890056588520613,
+      "grad_norm": 0.7392122745513916,
+      "learning_rate": 0.00014740186831941075,
+      "loss": 0.8171,
+      "step": 82800
+    },
+    {
+      "epoch": 2.7923740231743466,
+      "grad_norm": 0.6652575135231018,
+      "learning_rate": 0.0001471773106979251,
+      "loss": 0.8155,
+      "step": 82900
+    },
+    {
+      "epoch": 2.795742387496632,
+      "grad_norm": 0.698665201663971,
+      "learning_rate": 0.0001469527530764394,
+      "loss": 0.8202,
+      "step": 83000
+    },
+    {
+      "epoch": 2.7991107518189167,
+      "grad_norm": 0.6012236475944519,
+      "learning_rate": 0.00014672819545495373,
+      "loss": 0.817,
+      "step": 83100
+    },
+    {
+      "epoch": 2.802479116141202,
+      "grad_norm": 0.6791641116142273,
+      "learning_rate": 0.00014650363783346804,
+      "loss": 0.8185,
+      "step": 83200
+    },
+    {
+      "epoch": 2.8058474804634868,
+      "grad_norm": 0.694733738899231,
+      "learning_rate": 0.00014627908021198237,
+      "loss": 0.8161,
+      "step": 83300
+    },
+    {
+      "epoch": 2.809215844785772,
+      "grad_norm": 1.395378589630127,
+      "learning_rate": 0.0001460545225904967,
+      "loss": 0.8136,
+      "step": 83400
+    },
+    {
+      "epoch": 2.8125842091080573,
+      "grad_norm": 1.1096270084381104,
+      "learning_rate": 0.00014582996496901104,
+      "loss": 0.8187,
+      "step": 83500
+    },
+    {
+      "epoch": 2.815952573430342,
+      "grad_norm": 0.6426212787628174,
+      "learning_rate": 0.00014560540734752535,
+      "loss": 0.817,
+      "step": 83600
+    },
+    {
+      "epoch": 2.8193209377526274,
+      "grad_norm": 0.6201661229133606,
+      "learning_rate": 0.0001453808497260397,
+      "loss": 0.8166,
+      "step": 83700
+    },
+    {
+      "epoch": 2.822689302074912,
+      "grad_norm": 0.6646463871002197,
+      "learning_rate": 0.00014515629210455402,
+      "loss": 0.82,
+      "step": 83800
+    },
+    {
+      "epoch": 2.8260576663971975,
+      "grad_norm": 0.6990267038345337,
+      "learning_rate": 0.00014493173448306833,
+      "loss": 0.8169,
+      "step": 83900
+    },
+    {
+      "epoch": 2.8294260307194827,
+      "grad_norm": 0.671810507774353,
+      "learning_rate": 0.00014470717686158267,
+      "loss": 0.8167,
+      "step": 84000
+    },
+    {
+      "epoch": 2.8327943950417676,
+      "grad_norm": 0.686876118183136,
+      "learning_rate": 0.000144482619240097,
+      "loss": 0.8137,
+      "step": 84100
+    },
+    {
+      "epoch": 2.836162759364053,
+      "grad_norm": 0.6434644460678101,
+      "learning_rate": 0.0001442580616186113,
+      "loss": 0.8157,
+      "step": 84200
+    },
+    {
+      "epoch": 2.8395311236863376,
+      "grad_norm": 0.652999758720398,
+      "learning_rate": 0.00014403350399712564,
+      "loss": 0.8121,
+      "step": 84300
+    },
+    {
+      "epoch": 2.842899488008623,
+      "grad_norm": 0.6468531489372253,
+      "learning_rate": 0.00014380894637563998,
+      "loss": 0.8175,
+      "step": 84400
+    },
+    {
+      "epoch": 2.846267852330908,
+      "grad_norm": 0.6018803119659424,
+      "learning_rate": 0.00014358438875415432,
+      "loss": 0.816,
+      "step": 84500
+    },
+    {
+      "epoch": 2.8496362166531934,
+      "grad_norm": 0.6775005459785461,
+      "learning_rate": 0.00014335983113266862,
+      "loss": 0.8128,
+      "step": 84600
+    },
+    {
+      "epoch": 2.8530045809754783,
+      "grad_norm": 0.664910078048706,
+      "learning_rate": 0.00014313527351118296,
+      "loss": 0.8172,
+      "step": 84700
+    },
+    {
+      "epoch": 2.8563729452977635,
+      "grad_norm": 0.6959900259971619,
+      "learning_rate": 0.0001429107158896973,
+      "loss": 0.8135,
+      "step": 84800
+    },
+    {
+      "epoch": 2.8597413096200484,
+      "grad_norm": 0.6213033199310303,
+      "learning_rate": 0.0001426861582682116,
+      "loss": 0.8153,
+      "step": 84900
+    },
+    {
+      "epoch": 2.8631096739423336,
+      "grad_norm": 0.6437749266624451,
+      "learning_rate": 0.00014246160064672594,
+      "loss": 0.8114,
+      "step": 85000
+    },
+    {
+      "epoch": 2.866478038264619,
+      "grad_norm": 0.6763966083526611,
+      "learning_rate": 0.00014223704302524027,
+      "loss": 0.8138,
+      "step": 85100
+    },
+    {
+      "epoch": 2.8698464025869037,
+      "grad_norm": 1.10175621509552,
+      "learning_rate": 0.00014201248540375458,
+      "loss": 0.8153,
+      "step": 85200
+    },
+    {
+      "epoch": 2.873214766909189,
+      "grad_norm": 0.6517946124076843,
+      "learning_rate": 0.00014178792778226892,
+      "loss": 0.8147,
+      "step": 85300
+    },
+    {
+      "epoch": 2.876583131231474,
+      "grad_norm": 0.7234548926353455,
+      "learning_rate": 0.00014156337016078325,
+      "loss": 0.8182,
+      "step": 85400
+    },
+    {
+      "epoch": 2.879951495553759,
+      "grad_norm": 0.6792501211166382,
+      "learning_rate": 0.00014133881253929759,
+      "loss": 0.8154,
+      "step": 85500
+    },
+    {
+      "epoch": 2.8833198598760443,
+      "grad_norm": 0.6222261786460876,
+      "learning_rate": 0.0001411142549178119,
+      "loss": 0.816,
+      "step": 85600
+    },
+    {
+      "epoch": 2.886688224198329,
+      "grad_norm": 0.6346508264541626,
+      "learning_rate": 0.00014088969729632623,
+      "loss": 0.8159,
+      "step": 85700
+    },
+    {
+      "epoch": 2.8900565885206144,
+      "grad_norm": 0.7334688305854797,
+      "learning_rate": 0.00014066513967484054,
+      "loss": 0.8166,
+      "step": 85800
+    },
+    {
+      "epoch": 2.8934249528428992,
+      "grad_norm": 0.6864719986915588,
+      "learning_rate": 0.00014044058205335487,
+      "loss": 0.819,
+      "step": 85900
+    },
+    {
+      "epoch": 2.8967933171651845,
+      "grad_norm": 0.5927285552024841,
+      "learning_rate": 0.0001402160244318692,
+      "loss": 0.8148,
+      "step": 86000
+    },
+    {
+      "epoch": 2.90016168148747,
+      "grad_norm": 0.660213828086853,
+      "learning_rate": 0.00013999146681038354,
+      "loss": 0.815,
+      "step": 86100
+    },
+    {
+      "epoch": 2.903530045809755,
+      "grad_norm": 0.6887788772583008,
+      "learning_rate": 0.00013976690918889785,
+      "loss": 0.8143,
+      "step": 86200
+    },
+    {
+      "epoch": 2.90689841013204,
+      "grad_norm": 1.1385151147842407,
+      "learning_rate": 0.00013954235156741219,
+      "loss": 0.8139,
+      "step": 86300
+    },
+    {
+      "epoch": 2.910266774454325,
+      "grad_norm": 0.6464802026748657,
+      "learning_rate": 0.00013931779394592652,
+      "loss": 0.8149,
+      "step": 86400
+    },
+    {
+      "epoch": 2.91363513877661,
+      "grad_norm": 0.6867853999137878,
+      "learning_rate": 0.00013909323632444086,
+      "loss": 0.8139,
+      "step": 86500
+    },
+    {
+      "epoch": 2.9170035030988952,
+      "grad_norm": 0.8868036866188049,
+      "learning_rate": 0.00013886867870295516,
+      "loss": 0.8139,
+      "step": 86600
+    },
+    {
+      "epoch": 2.9203718674211805,
+      "grad_norm": 0.6516538262367249,
+      "learning_rate": 0.0001386441210814695,
+      "loss": 0.8153,
+      "step": 86700
+    },
+    {
+      "epoch": 2.9237402317434653,
+      "grad_norm": 0.6790093779563904,
+      "learning_rate": 0.0001384195634599838,
+      "loss": 0.8133,
+      "step": 86800
+    },
+    {
+      "epoch": 2.9271085960657506,
+      "grad_norm": 0.6333130598068237,
+      "learning_rate": 0.00013819500583849814,
+      "loss": 0.8117,
+      "step": 86900
+    },
+    {
+      "epoch": 2.9304769603880354,
+      "grad_norm": 0.7102107405662537,
+      "learning_rate": 0.00013797044821701248,
+      "loss": 0.8128,
+      "step": 87000
+    },
+    {
+      "epoch": 2.9338453247103207,
+      "grad_norm": 0.7193422913551331,
+      "learning_rate": 0.00013774589059552679,
+      "loss": 0.8121,
+      "step": 87100
+    },
+    {
+      "epoch": 2.937213689032606,
+      "grad_norm": 0.7304584980010986,
+      "learning_rate": 0.00013752133297404112,
+      "loss": 0.8114,
+      "step": 87200
+    },
+    {
+      "epoch": 2.9405820533548908,
+      "grad_norm": 0.8924300074577332,
+      "learning_rate": 0.00013729677535255546,
+      "loss": 0.8137,
+      "step": 87300
+    },
+    {
+      "epoch": 2.943950417677176,
+      "grad_norm": 0.6813507080078125,
+      "learning_rate": 0.0001370722177310698,
+      "loss": 0.8109,
+      "step": 87400
+    },
+    {
+      "epoch": 2.947318781999461,
+      "grad_norm": 0.6427081823348999,
+      "learning_rate": 0.0001368476601095841,
+      "loss": 0.8125,
+      "step": 87500
+    },
+    {
+      "epoch": 2.950687146321746,
+      "grad_norm": 0.6571387052536011,
+      "learning_rate": 0.00013662310248809843,
+      "loss": 0.8132,
+      "step": 87600
+    },
+    {
+      "epoch": 2.9540555106440314,
+      "grad_norm": 0.7705689072608948,
+      "learning_rate": 0.00013639854486661277,
+      "loss": 0.8129,
+      "step": 87700
+    },
+    {
+      "epoch": 2.957423874966316,
+      "grad_norm": 0.7075904011726379,
+      "learning_rate": 0.00013617398724512708,
+      "loss": 0.8143,
+      "step": 87800
+    },
+    {
+      "epoch": 2.9607922392886015,
+      "grad_norm": 0.6926144957542419,
+      "learning_rate": 0.0001359494296236414,
+      "loss": 0.8115,
+      "step": 87900
+    },
+    {
+      "epoch": 2.9641606036108863,
+      "grad_norm": 0.7183883190155029,
+      "learning_rate": 0.00013572487200215575,
+      "loss": 0.8122,
+      "step": 88000
+    },
+    {
+      "epoch": 2.9675289679331716,
+      "grad_norm": 0.6963924169540405,
+      "learning_rate": 0.00013550031438067006,
+      "loss": 0.8151,
+      "step": 88100
+    },
+    {
+      "epoch": 2.970897332255457,
+      "grad_norm": 0.6360912919044495,
+      "learning_rate": 0.0001352757567591844,
+      "loss": 0.8129,
+      "step": 88200
+    },
+    {
+      "epoch": 2.974265696577742,
+      "grad_norm": 0.7141982316970825,
+      "learning_rate": 0.00013505119913769873,
+      "loss": 0.8142,
+      "step": 88300
+    },
+    {
+      "epoch": 2.977634060900027,
+      "grad_norm": 0.6647577285766602,
+      "learning_rate": 0.00013482664151621306,
+      "loss": 0.811,
+      "step": 88400
+    },
+    {
+      "epoch": 2.981002425222312,
+      "grad_norm": 0.6686524152755737,
+      "learning_rate": 0.00013460208389472737,
+      "loss": 0.8149,
+      "step": 88500
+    },
+    {
+      "epoch": 2.984370789544597,
+      "grad_norm": 0.66321861743927,
+      "learning_rate": 0.0001343775262732417,
+      "loss": 0.8103,
+      "step": 88600
+    },
+    {
+      "epoch": 2.9877391538668823,
+      "grad_norm": 0.622572660446167,
+      "learning_rate": 0.00013415296865175601,
+      "loss": 0.8144,
+      "step": 88700
+    },
+    {
+      "epoch": 2.9911075181891675,
+      "grad_norm": 0.667248547077179,
+      "learning_rate": 0.00013392841103027035,
+      "loss": 0.8108,
+      "step": 88800
+    },
+    {
+      "epoch": 2.9944758825114524,
+      "grad_norm": 0.6621103882789612,
+      "learning_rate": 0.00013370385340878468,
+      "loss": 0.8106,
+      "step": 88900
+    },
+    {
+      "epoch": 2.9978442468337376,
+      "grad_norm": 0.6303636431694031,
+      "learning_rate": 0.00013347929578729902,
+      "loss": 0.8104,
+      "step": 89000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.760505199432373,
+      "eval_runtime": 1.8138,
+      "eval_samples_per_second": 2756.665,
+      "eval_steps_per_second": 43.555,
+      "step": 89064
+    },
+    {
+      "epoch": 3.0012126111560224,
+      "grad_norm": 0.6279102563858032,
+      "learning_rate": 0.00013325473816581333,
+      "loss": 0.813,
+      "step": 89100
+    },
+    {
+      "epoch": 3.0045809754783077,
+      "grad_norm": 0.6355727910995483,
+      "learning_rate": 0.00013303018054432766,
+      "loss": 0.811,
+      "step": 89200
+    },
+    {
+      "epoch": 3.007949339800593,
+      "grad_norm": 0.6710761189460754,
+      "learning_rate": 0.000132805622922842,
+      "loss": 0.8147,
+      "step": 89300
+    },
+    {
+      "epoch": 3.011317704122878,
+      "grad_norm": 0.6306372284889221,
+      "learning_rate": 0.00013258106530135633,
+      "loss": 0.8151,
+      "step": 89400
+    },
+    {
+      "epoch": 3.014686068445163,
+      "grad_norm": 0.6803897023200989,
+      "learning_rate": 0.00013235650767987064,
+      "loss": 0.8086,
+      "step": 89500
+    },
+    {
+      "epoch": 3.0180544327674483,
+      "grad_norm": 0.6367260217666626,
+      "learning_rate": 0.00013213195005838498,
+      "loss": 0.8119,
+      "step": 89600
+    },
+    {
+      "epoch": 3.021422797089733,
+      "grad_norm": 0.664561927318573,
+      "learning_rate": 0.00013190739243689928,
+      "loss": 0.8094,
+      "step": 89700
+    },
+    {
+      "epoch": 3.0247911614120184,
+      "grad_norm": 0.6986654996871948,
+      "learning_rate": 0.00013168283481541362,
+      "loss": 0.8116,
+      "step": 89800
+    },
+    {
+      "epoch": 3.0281595257343032,
+      "grad_norm": 0.6583049893379211,
+      "learning_rate": 0.00013145827719392795,
+      "loss": 0.8148,
+      "step": 89900
+    },
+    {
+      "epoch": 3.0315278900565885,
+      "grad_norm": 0.6716769337654114,
+      "learning_rate": 0.00013123371957244226,
+      "loss": 0.8137,
+      "step": 90000
+    },
+    {
+      "epoch": 3.034896254378874,
+      "grad_norm": 0.7089846134185791,
+      "learning_rate": 0.0001310091619509566,
+      "loss": 0.809,
+      "step": 90100
+    },
+    {
+      "epoch": 3.0382646187011586,
+      "grad_norm": 0.6873953938484192,
+      "learning_rate": 0.00013078460432947093,
+      "loss": 0.8113,
+      "step": 90200
+    },
+    {
+      "epoch": 3.041632983023444,
+      "grad_norm": 0.6583080887794495,
+      "learning_rate": 0.00013056004670798527,
+      "loss": 0.8067,
+      "step": 90300
+    },
+    {
+      "epoch": 3.045001347345729,
+      "grad_norm": 0.6893338561058044,
+      "learning_rate": 0.00013033548908649958,
+      "loss": 0.8126,
+      "step": 90400
+    },
+    {
+      "epoch": 3.048369711668014,
+      "grad_norm": 0.9407336115837097,
+      "learning_rate": 0.0001301109314650139,
+      "loss": 0.8077,
+      "step": 90500
+    },
+    {
+      "epoch": 3.0517380759902992,
+      "grad_norm": 1.1255403757095337,
+      "learning_rate": 0.00012988637384352822,
+      "loss": 0.8104,
+      "step": 90600
+    },
+    {
+      "epoch": 3.055106440312584,
+      "grad_norm": 0.6687456369400024,
+      "learning_rate": 0.00012966181622204255,
+      "loss": 0.8122,
+      "step": 90700
+    },
+    {
+      "epoch": 3.0584748046348693,
+      "grad_norm": 0.6100497245788574,
+      "learning_rate": 0.0001294372586005569,
+      "loss": 0.8098,
+      "step": 90800
+    },
+    {
+      "epoch": 3.0618431689571546,
+      "grad_norm": 0.6621761918067932,
+      "learning_rate": 0.00012921270097907123,
+      "loss": 0.8118,
+      "step": 90900
+    },
+    {
+      "epoch": 3.0652115332794394,
+      "grad_norm": 0.8076705932617188,
+      "learning_rate": 0.00012898814335758553,
+      "loss": 0.811,
+      "step": 91000
+    },
+    {
+      "epoch": 3.0685798976017247,
+      "grad_norm": 0.6705955266952515,
+      "learning_rate": 0.00012876358573609987,
+      "loss": 0.8116,
+      "step": 91100
+    },
+    {
+      "epoch": 3.0719482619240095,
+      "grad_norm": 0.6365945339202881,
+      "learning_rate": 0.0001285390281146142,
+      "loss": 0.8074,
+      "step": 91200
+    },
+    {
+      "epoch": 3.0753166262462948,
+      "grad_norm": 0.9314165115356445,
+      "learning_rate": 0.00012831447049312854,
+      "loss": 0.8092,
+      "step": 91300
+    },
+    {
+      "epoch": 3.07868499056858,
+      "grad_norm": 0.6733311414718628,
+      "learning_rate": 0.00012808991287164285,
+      "loss": 0.8102,
+      "step": 91400
+    },
+    {
+      "epoch": 3.082053354890865,
+      "grad_norm": 0.6910605430603027,
+      "learning_rate": 0.00012786535525015718,
+      "loss": 0.8065,
+      "step": 91500
+    },
+    {
+      "epoch": 3.08542171921315,
+      "grad_norm": 0.7043003439903259,
+      "learning_rate": 0.0001276407976286715,
+      "loss": 0.8073,
+      "step": 91600
+    },
+    {
+      "epoch": 3.0887900835354354,
+      "grad_norm": 0.6764921545982361,
+      "learning_rate": 0.00012741624000718583,
+      "loss": 0.8089,
+      "step": 91700
+    },
+    {
+      "epoch": 3.09215844785772,
+      "grad_norm": 0.6997144818305969,
+      "learning_rate": 0.00012719168238570016,
+      "loss": 0.8088,
+      "step": 91800
+    },
+    {
+      "epoch": 3.0955268121800055,
+      "grad_norm": 0.6801837086677551,
+      "learning_rate": 0.0001269671247642145,
+      "loss": 0.8094,
+      "step": 91900
+    },
+    {
+      "epoch": 3.0988951765022903,
+      "grad_norm": 0.6339113116264343,
+      "learning_rate": 0.0001267425671427288,
+      "loss": 0.8109,
+      "step": 92000
+    },
+    {
+      "epoch": 3.1022635408245756,
+      "grad_norm": 0.6691506505012512,
+      "learning_rate": 0.00012651800952124314,
+      "loss": 0.8135,
+      "step": 92100
+    },
+    {
+      "epoch": 3.105631905146861,
+      "grad_norm": 0.6617900133132935,
+      "learning_rate": 0.00012629345189975747,
+      "loss": 0.8106,
+      "step": 92200
+    },
+    {
+      "epoch": 3.1090002694691456,
+      "grad_norm": 0.6737276911735535,
+      "learning_rate": 0.0001260688942782718,
+      "loss": 0.8128,
+      "step": 92300
+    },
+    {
+      "epoch": 3.112368633791431,
+      "grad_norm": 0.670802652835846,
+      "learning_rate": 0.00012584433665678612,
+      "loss": 0.8106,
+      "step": 92400
+    },
+    {
+      "epoch": 3.115736998113716,
+      "grad_norm": 0.626956045627594,
+      "learning_rate": 0.00012561977903530045,
+      "loss": 0.8114,
+      "step": 92500
+    },
+    {
+      "epoch": 3.119105362436001,
+      "grad_norm": 0.6243528127670288,
+      "learning_rate": 0.00012539522141381476,
+      "loss": 0.8115,
+      "step": 92600
+    },
+    {
+      "epoch": 3.1224737267582863,
+      "grad_norm": 0.6828027367591858,
+      "learning_rate": 0.0001251706637923291,
+      "loss": 0.807,
+      "step": 92700
+    },
+    {
+      "epoch": 3.125842091080571,
+      "grad_norm": 0.704557478427887,
+      "learning_rate": 0.00012494610617084343,
+      "loss": 0.8085,
+      "step": 92800
+    },
+    {
+      "epoch": 3.1292104554028564,
+      "grad_norm": 0.6188080310821533,
+      "learning_rate": 0.00012472154854935774,
+      "loss": 0.8079,
+      "step": 92900
+    },
+    {
+      "epoch": 3.1325788197251416,
+      "grad_norm": 0.6640317440032959,
+      "learning_rate": 0.00012449699092787207,
+      "loss": 0.808,
+      "step": 93000
+    },
+    {
+      "epoch": 3.1359471840474265,
+      "grad_norm": 0.6489530801773071,
+      "learning_rate": 0.0001242724333063864,
+      "loss": 0.8089,
+      "step": 93100
+    },
+    {
+      "epoch": 3.1393155483697117,
+      "grad_norm": 0.6527587175369263,
+      "learning_rate": 0.00012404787568490074,
+      "loss": 0.809,
+      "step": 93200
+    },
+    {
+      "epoch": 3.1426839126919965,
+      "grad_norm": 0.6802580952644348,
+      "learning_rate": 0.00012382331806341508,
+      "loss": 0.8099,
+      "step": 93300
+    },
+    {
+      "epoch": 3.146052277014282,
+      "grad_norm": 0.6751042604446411,
+      "learning_rate": 0.0001235987604419294,
+      "loss": 0.8103,
+      "step": 93400
+    },
+    {
+      "epoch": 3.149420641336567,
+      "grad_norm": 0.723804235458374,
+      "learning_rate": 0.0001233742028204437,
+      "loss": 0.8043,
+      "step": 93500
+    },
+    {
+      "epoch": 3.152789005658852,
+      "grad_norm": 0.768860399723053,
+      "learning_rate": 0.00012314964519895803,
+      "loss": 0.8105,
+      "step": 93600
+    },
+    {
+      "epoch": 3.156157369981137,
+      "grad_norm": 0.6464242339134216,
+      "learning_rate": 0.00012292508757747237,
+      "loss": 0.808,
+      "step": 93700
+    },
+    {
+      "epoch": 3.1595257343034224,
+      "grad_norm": 0.7125059962272644,
+      "learning_rate": 0.0001227005299559867,
+      "loss": 0.8103,
+      "step": 93800
+    },
+    {
+      "epoch": 3.1628940986257073,
+      "grad_norm": 0.6584749221801758,
+      "learning_rate": 0.000122475972334501,
+      "loss": 0.81,
+      "step": 93900
+    },
+    {
+      "epoch": 3.1662624629479925,
+      "grad_norm": 0.6689501404762268,
+      "learning_rate": 0.00012225141471301534,
+      "loss": 0.8086,
+      "step": 94000
+    },
+    {
+      "epoch": 3.169630827270278,
+      "grad_norm": 0.6383669376373291,
+      "learning_rate": 0.00012202685709152968,
+      "loss": 0.8097,
+      "step": 94100
+    },
+    {
+      "epoch": 3.1729991915925626,
+      "grad_norm": 0.6786794066429138,
+      "learning_rate": 0.000121802299470044,
+      "loss": 0.809,
+      "step": 94200
+    },
+    {
+      "epoch": 3.176367555914848,
+      "grad_norm": 0.6704023480415344,
+      "learning_rate": 0.00012157774184855834,
+      "loss": 0.8086,
+      "step": 94300
+    },
+    {
+      "epoch": 3.1797359202371327,
+      "grad_norm": 0.652862012386322,
+      "learning_rate": 0.00012135318422707266,
+      "loss": 0.808,
+      "step": 94400
+    },
+    {
+      "epoch": 3.183104284559418,
+      "grad_norm": 0.8095204830169678,
+      "learning_rate": 0.00012112862660558698,
+      "loss": 0.8066,
+      "step": 94500
+    },
+    {
+      "epoch": 3.1864726488817032,
+      "grad_norm": 0.6581931710243225,
+      "learning_rate": 0.0001209040689841013,
+      "loss": 0.8061,
+      "step": 94600
+    },
+    {
+      "epoch": 3.189841013203988,
+      "grad_norm": 0.6642458438873291,
+      "learning_rate": 0.00012067951136261564,
+      "loss": 0.8079,
+      "step": 94700
+    },
+    {
+      "epoch": 3.1932093775262733,
+      "grad_norm": 0.6264484524726868,
+      "learning_rate": 0.00012045495374112996,
+      "loss": 0.8108,
+      "step": 94800
+    },
+    {
+      "epoch": 3.196577741848558,
+      "grad_norm": 0.6631668210029602,
+      "learning_rate": 0.0001202303961196443,
+      "loss": 0.8076,
+      "step": 94900
+    },
+    {
+      "epoch": 3.1999461061708434,
+      "grad_norm": 0.636448085308075,
+      "learning_rate": 0.00012000583849815862,
+      "loss": 0.805,
+      "step": 95000
+    },
+    {
+      "epoch": 3.2033144704931287,
+      "grad_norm": 0.6331253051757812,
+      "learning_rate": 0.00011978128087667295,
+      "loss": 0.8089,
+      "step": 95100
+    },
+    {
+      "epoch": 3.2066828348154135,
+      "grad_norm": 0.6623615026473999,
+      "learning_rate": 0.00011955672325518727,
+      "loss": 0.8089,
+      "step": 95200
+    },
+    {
+      "epoch": 3.2100511991376988,
+      "grad_norm": 0.671399712562561,
+      "learning_rate": 0.00011933216563370161,
+      "loss": 0.8054,
+      "step": 95300
+    },
+    {
+      "epoch": 3.213419563459984,
+      "grad_norm": 0.6822311282157898,
+      "learning_rate": 0.00011910760801221592,
+      "loss": 0.8057,
+      "step": 95400
+    },
+    {
+      "epoch": 3.216787927782269,
+      "grad_norm": 0.6769167184829712,
+      "learning_rate": 0.00011888305039073025,
+      "loss": 0.8065,
+      "step": 95500
+    },
+    {
+      "epoch": 3.220156292104554,
+      "grad_norm": 0.6246688961982727,
+      "learning_rate": 0.00011865849276924457,
+      "loss": 0.8063,
+      "step": 95600
+    },
+    {
+      "epoch": 3.223524656426839,
+      "grad_norm": 0.6401100754737854,
+      "learning_rate": 0.00011843393514775891,
+      "loss": 0.807,
+      "step": 95700
+    },
+    {
+      "epoch": 3.226893020749124,
+      "grad_norm": 0.6910848021507263,
+      "learning_rate": 0.00011820937752627323,
+      "loss": 0.8068,
+      "step": 95800
+    },
+    {
+      "epoch": 3.2302613850714095,
+      "grad_norm": 0.7022745013237,
+      "learning_rate": 0.00011798481990478756,
+      "loss": 0.8077,
+      "step": 95900
+    },
+    {
+      "epoch": 3.2336297493936943,
+      "grad_norm": 0.7098489999771118,
+      "learning_rate": 0.00011776026228330189,
+      "loss": 0.8079,
+      "step": 96000
+    },
+    {
+      "epoch": 3.2369981137159796,
+      "grad_norm": 1.6580332517623901,
+      "learning_rate": 0.00011753570466181622,
+      "loss": 0.8052,
+      "step": 96100
+    },
+    {
+      "epoch": 3.240366478038265,
+      "grad_norm": 0.6385944485664368,
+      "learning_rate": 0.00011731114704033054,
+      "loss": 0.8097,
+      "step": 96200
+    },
+    {
+      "epoch": 3.2437348423605497,
+      "grad_norm": 0.6737959384918213,
+      "learning_rate": 0.00011708658941884488,
+      "loss": 0.8075,
+      "step": 96300
+    },
+    {
+      "epoch": 3.247103206682835,
+      "grad_norm": 0.6414308547973633,
+      "learning_rate": 0.00011686203179735919,
+      "loss": 0.8074,
+      "step": 96400
+    },
+    {
+      "epoch": 3.2504715710051197,
+      "grad_norm": 0.6727792024612427,
+      "learning_rate": 0.00011663747417587351,
+      "loss": 0.8082,
+      "step": 96500
+    },
+    {
+      "epoch": 3.253839935327405,
+      "grad_norm": 0.6280369162559509,
+      "learning_rate": 0.00011641291655438784,
+      "loss": 0.8098,
+      "step": 96600
+    },
+    {
+      "epoch": 3.2572082996496903,
+      "grad_norm": 0.6988112330436707,
+      "learning_rate": 0.00011618835893290216,
+      "loss": 0.8079,
+      "step": 96700
+    },
+    {
+      "epoch": 3.260576663971975,
+      "grad_norm": 0.6650980710983276,
+      "learning_rate": 0.0001159638013114165,
+      "loss": 0.8084,
+      "step": 96800
+    },
+    {
+      "epoch": 3.2639450282942604,
+      "grad_norm": 0.6886364817619324,
+      "learning_rate": 0.00011573924368993082,
+      "loss": 0.8074,
+      "step": 96900
+    },
+    {
+      "epoch": 3.267313392616545,
+      "grad_norm": 0.6584846377372742,
+      "learning_rate": 0.00011551468606844516,
+      "loss": 0.8111,
+      "step": 97000
+    },
+    {
+      "epoch": 3.2706817569388305,
+      "grad_norm": 0.9834907650947571,
+      "learning_rate": 0.00011529012844695948,
+      "loss": 0.8088,
+      "step": 97100
+    },
+    {
+      "epoch": 3.2740501212611157,
+      "grad_norm": 0.6571055054664612,
+      "learning_rate": 0.00011506557082547381,
+      "loss": 0.8043,
+      "step": 97200
+    },
+    {
+      "epoch": 3.2774184855834005,
+      "grad_norm": 0.6906171441078186,
+      "learning_rate": 0.00011484101320398814,
+      "loss": 0.8081,
+      "step": 97300
+    },
+    {
+      "epoch": 3.280786849905686,
+      "grad_norm": 0.7476776838302612,
+      "learning_rate": 0.00011461645558250246,
+      "loss": 0.8064,
+      "step": 97400
+    },
+    {
+      "epoch": 3.284155214227971,
+      "grad_norm": 0.6860908269882202,
+      "learning_rate": 0.00011439189796101678,
+      "loss": 0.8073,
+      "step": 97500
+    },
+    {
+      "epoch": 3.287523578550256,
+      "grad_norm": 0.6590797901153564,
+      "learning_rate": 0.00011416734033953111,
+      "loss": 0.807,
+      "step": 97600
+    },
+    {
+      "epoch": 3.290891942872541,
+      "grad_norm": 0.7175418138504028,
+      "learning_rate": 0.00011394278271804544,
+      "loss": 0.8073,
+      "step": 97700
+    },
+    {
+      "epoch": 3.2942603071948264,
+      "grad_norm": 0.6721409559249878,
+      "learning_rate": 0.00011371822509655977,
+      "loss": 0.806,
+      "step": 97800
+    },
+    {
+      "epoch": 3.2976286715171113,
+      "grad_norm": 0.6370182633399963,
+      "learning_rate": 0.00011349366747507409,
+      "loss": 0.8094,
+      "step": 97900
+    },
+    {
+      "epoch": 3.3009970358393965,
+      "grad_norm": 0.6653867959976196,
+      "learning_rate": 0.00011326910985358843,
+      "loss": 0.8021,
+      "step": 98000
+    },
+    {
+      "epoch": 3.3043654001616813,
+      "grad_norm": 0.635477602481842,
+      "learning_rate": 0.00011304455223210275,
+      "loss": 0.8074,
+      "step": 98100
+    },
+    {
+      "epoch": 3.3077337644839666,
+      "grad_norm": 0.7132477760314941,
+      "learning_rate": 0.00011281999461061708,
+      "loss": 0.8048,
+      "step": 98200
+    },
+    {
+      "epoch": 3.311102128806252,
+      "grad_norm": 0.661605954170227,
+      "learning_rate": 0.00011259543698913139,
+      "loss": 0.8065,
+      "step": 98300
+    },
+    {
+      "epoch": 3.3144704931285367,
+      "grad_norm": 0.7334872484207153,
+      "learning_rate": 0.00011237087936764573,
+      "loss": 0.8056,
+      "step": 98400
+    },
+    {
+      "epoch": 3.317838857450822,
+      "grad_norm": 0.7113956212997437,
+      "learning_rate": 0.00011214632174616005,
+      "loss": 0.8068,
+      "step": 98500
+    },
+    {
+      "epoch": 3.321207221773107,
+      "grad_norm": 0.6793413758277893,
+      "learning_rate": 0.00011192176412467438,
+      "loss": 0.8102,
+      "step": 98600
+    },
+    {
+      "epoch": 3.324575586095392,
+      "grad_norm": 0.6595569849014282,
+      "learning_rate": 0.0001116972065031887,
+      "loss": 0.8045,
+      "step": 98700
+    },
+    {
+      "epoch": 3.3279439504176773,
+      "grad_norm": 0.6264058351516724,
+      "learning_rate": 0.00011147264888170304,
+      "loss": 0.8059,
+      "step": 98800
+    },
+    {
+      "epoch": 3.331312314739962,
+      "grad_norm": 0.7037299275398254,
+      "learning_rate": 0.00011124809126021736,
+      "loss": 0.8044,
+      "step": 98900
+    },
+    {
+      "epoch": 3.3346806790622474,
+      "grad_norm": 0.6255789995193481,
+      "learning_rate": 0.0001110235336387317,
+      "loss": 0.8081,
+      "step": 99000
+    },
+    {
+      "epoch": 3.3380490433845322,
+      "grad_norm": 0.6675742864608765,
+      "learning_rate": 0.00011079897601724602,
+      "loss": 0.8028,
+      "step": 99100
+    },
+    {
+      "epoch": 3.3414174077068175,
+      "grad_norm": 0.6799077391624451,
+      "learning_rate": 0.00011057441839576035,
+      "loss": 0.804,
+      "step": 99200
+    },
+    {
+      "epoch": 3.3447857720291028,
+      "grad_norm": 0.6606206893920898,
+      "learning_rate": 0.00011034986077427466,
+      "loss": 0.8064,
+      "step": 99300
+    },
+    {
+      "epoch": 3.3481541363513876,
+      "grad_norm": 0.7276676893234253,
+      "learning_rate": 0.00011012530315278898,
+      "loss": 0.8074,
+      "step": 99400
+    },
+    {
+      "epoch": 3.351522500673673,
+      "grad_norm": 0.6872825026512146,
+      "learning_rate": 0.00010990074553130332,
+      "loss": 0.8056,
+      "step": 99500
+    },
+    {
+      "epoch": 3.354890864995958,
+      "grad_norm": 0.6640327572822571,
+      "learning_rate": 0.00010967618790981764,
+      "loss": 0.8078,
+      "step": 99600
+    },
+    {
+      "epoch": 3.358259229318243,
+      "grad_norm": 0.9947742819786072,
+      "learning_rate": 0.00010945163028833198,
+      "loss": 0.8018,
+      "step": 99700
+    },
+    {
+      "epoch": 3.361627593640528,
+      "grad_norm": 0.6360524296760559,
+      "learning_rate": 0.0001092270726668463,
+      "loss": 0.807,
+      "step": 99800
+    },
+    {
+      "epoch": 3.3649959579628135,
+      "grad_norm": 0.7029403448104858,
+      "learning_rate": 0.00010900251504536063,
+      "loss": 0.8059,
+      "step": 99900
+    },
+    {
+      "epoch": 3.3683643222850983,
+      "grad_norm": 0.7347244024276733,
+      "learning_rate": 0.00010877795742387497,
+      "loss": 0.8069,
+      "step": 100000
+    },
+    {
+      "epoch": 3.3717326866073836,
+      "grad_norm": 0.6576654314994812,
+      "learning_rate": 0.00010855339980238929,
+      "loss": 0.8035,
+      "step": 100100
+    },
+    {
+      "epoch": 3.3751010509296684,
+      "grad_norm": 0.6341889500617981,
+      "learning_rate": 0.0001083288421809036,
+      "loss": 0.8065,
+      "step": 100200
+    },
+    {
+      "epoch": 3.3784694152519537,
+      "grad_norm": 0.6809447407722473,
+      "learning_rate": 0.00010810428455941793,
+      "loss": 0.8071,
+      "step": 100300
+    },
+    {
+      "epoch": 3.381837779574239,
+      "grad_norm": 0.6789582967758179,
+      "learning_rate": 0.00010787972693793226,
+      "loss": 0.8081,
+      "step": 100400
+    },
+    {
+      "epoch": 3.3852061438965237,
+      "grad_norm": 0.6672530174255371,
+      "learning_rate": 0.00010765516931644659,
+      "loss": 0.8079,
+      "step": 100500
+    },
+    {
+      "epoch": 3.388574508218809,
+      "grad_norm": 0.6484895348548889,
+      "learning_rate": 0.00010743061169496091,
+      "loss": 0.8053,
+      "step": 100600
+    },
+    {
+      "epoch": 3.391942872541094,
+      "grad_norm": 0.6628451943397522,
+      "learning_rate": 0.00010720605407347525,
+      "loss": 0.8041,
+      "step": 100700
+    },
+    {
+      "epoch": 3.395311236863379,
+      "grad_norm": 0.6461237668991089,
+      "learning_rate": 0.00010698149645198957,
+      "loss": 0.804,
+      "step": 100800
+    },
+    {
+      "epoch": 3.3986796011856644,
+      "grad_norm": 0.6682327389717102,
+      "learning_rate": 0.0001067569388305039,
+      "loss": 0.8049,
+      "step": 100900
+    },
+    {
+      "epoch": 3.402047965507949,
+      "grad_norm": 0.6611519455909729,
+      "learning_rate": 0.00010653238120901823,
+      "loss": 0.8004,
+      "step": 101000
+    },
+    {
+      "epoch": 3.4054163298302345,
+      "grad_norm": 0.6624406576156616,
+      "learning_rate": 0.00010630782358753256,
+      "loss": 0.8039,
+      "step": 101100
+    },
+    {
+      "epoch": 3.4087846941525197,
+      "grad_norm": 0.6566054224967957,
+      "learning_rate": 0.00010608326596604687,
+      "loss": 0.8058,
+      "step": 101200
+    },
+    {
+      "epoch": 3.4121530584748045,
+      "grad_norm": 0.6751037240028381,
+      "learning_rate": 0.0001058587083445612,
+      "loss": 0.8051,
+      "step": 101300
+    },
+    {
+      "epoch": 3.41552142279709,
+      "grad_norm": 0.6572412252426147,
+      "learning_rate": 0.00010563415072307553,
+      "loss": 0.8048,
+      "step": 101400
+    },
+    {
+      "epoch": 3.418889787119375,
+      "grad_norm": 0.7792493104934692,
+      "learning_rate": 0.00010540959310158986,
+      "loss": 0.8044,
+      "step": 101500
+    },
+    {
+      "epoch": 3.42225815144166,
+      "grad_norm": 0.6584118604660034,
+      "learning_rate": 0.00010518503548010418,
+      "loss": 0.8026,
+      "step": 101600
+    },
+    {
+      "epoch": 3.425626515763945,
+      "grad_norm": 0.6414441466331482,
+      "learning_rate": 0.00010496047785861852,
+      "loss": 0.8068,
+      "step": 101700
+    },
+    {
+      "epoch": 3.42899488008623,
+      "grad_norm": 0.6849080920219421,
+      "learning_rate": 0.00010473592023713284,
+      "loss": 0.8052,
+      "step": 101800
+    },
+    {
+      "epoch": 3.4323632444085153,
+      "grad_norm": 0.6778447031974792,
+      "learning_rate": 0.00010451136261564717,
+      "loss": 0.8069,
+      "step": 101900
+    },
+    {
+      "epoch": 3.4357316087308005,
+      "grad_norm": 0.6514096260070801,
+      "learning_rate": 0.0001042868049941615,
+      "loss": 0.8042,
+      "step": 102000
+    },
+    {
+      "epoch": 3.4390999730530853,
+      "grad_norm": 0.630409300327301,
+      "learning_rate": 0.00010406224737267583,
+      "loss": 0.8029,
+      "step": 102100
+    },
+    {
+      "epoch": 3.4424683373753706,
+      "grad_norm": 0.709867000579834,
+      "learning_rate": 0.00010383768975119014,
+      "loss": 0.8026,
+      "step": 102200
+    },
+    {
+      "epoch": 3.4458367016976554,
+      "grad_norm": 0.6706274151802063,
+      "learning_rate": 0.00010361313212970447,
+      "loss": 0.8069,
+      "step": 102300
+    },
+    {
+      "epoch": 3.4492050660199407,
+      "grad_norm": 0.6600052118301392,
+      "learning_rate": 0.0001033885745082188,
+      "loss": 0.8028,
+      "step": 102400
+    },
+    {
+      "epoch": 3.452573430342226,
+      "grad_norm": 0.6819061040878296,
+      "learning_rate": 0.00010316401688673313,
+      "loss": 0.8063,
+      "step": 102500
+    },
+    {
+      "epoch": 3.455941794664511,
+      "grad_norm": 0.6513516902923584,
+      "learning_rate": 0.00010293945926524745,
+      "loss": 0.8008,
+      "step": 102600
+    },
+    {
+      "epoch": 3.459310158986796,
+      "grad_norm": 0.623355507850647,
+      "learning_rate": 0.00010271490164376179,
+      "loss": 0.8035,
+      "step": 102700
+    },
+    {
+      "epoch": 3.462678523309081,
+      "grad_norm": 0.6720972657203674,
+      "learning_rate": 0.00010249034402227611,
+      "loss": 0.8057,
+      "step": 102800
+    },
+    {
+      "epoch": 3.466046887631366,
+      "grad_norm": 0.6593225002288818,
+      "learning_rate": 0.00010226578640079045,
+      "loss": 0.8026,
+      "step": 102900
+    },
+    {
+      "epoch": 3.4694152519536514,
+      "grad_norm": 0.6316161751747131,
+      "learning_rate": 0.00010204122877930477,
+      "loss": 0.8019,
+      "step": 103000
+    },
+    {
+      "epoch": 3.4727836162759362,
+      "grad_norm": 0.7166170477867126,
+      "learning_rate": 0.00010181667115781907,
+      "loss": 0.8046,
+      "step": 103100
+    },
+    {
+      "epoch": 3.4761519805982215,
+      "grad_norm": 0.6723181009292603,
+      "learning_rate": 0.00010159211353633341,
+      "loss": 0.8038,
+      "step": 103200
+    },
+    {
+      "epoch": 3.4795203449205068,
+      "grad_norm": 0.6460021734237671,
+      "learning_rate": 0.00010136755591484773,
+      "loss": 0.8049,
+      "step": 103300
+    },
+    {
+      "epoch": 3.4828887092427916,
+      "grad_norm": 0.7485737800598145,
+      "learning_rate": 0.00010114299829336207,
+      "loss": 0.805,
+      "step": 103400
+    },
+    {
+      "epoch": 3.486257073565077,
+      "grad_norm": 0.67794269323349,
+      "learning_rate": 0.00010091844067187639,
+      "loss": 0.8026,
+      "step": 103500
+    },
+    {
+      "epoch": 3.489625437887362,
+      "grad_norm": 0.6387248635292053,
+      "learning_rate": 0.00010069388305039072,
+      "loss": 0.8043,
+      "step": 103600
+    },
+    {
+      "epoch": 3.492993802209647,
+      "grad_norm": 0.662192702293396,
+      "learning_rate": 0.00010046932542890505,
+      "loss": 0.8031,
+      "step": 103700
+    },
+    {
+      "epoch": 3.496362166531932,
+      "grad_norm": 0.7354533076286316,
+      "learning_rate": 0.00010024476780741938,
+      "loss": 0.8051,
+      "step": 103800
+    },
+    {
+      "epoch": 3.499730530854217,
+      "grad_norm": 0.7070392370223999,
+      "learning_rate": 0.0001000202101859337,
+      "loss": 0.8036,
+      "step": 103900
+    },
+    {
+      "epoch": 3.5030988951765023,
+      "grad_norm": 0.7134096622467041,
+      "learning_rate": 9.979565256444804e-05,
+      "loss": 0.8034,
+      "step": 104000
+    },
+    {
+      "epoch": 3.5064672594987876,
+      "grad_norm": 0.6784800291061401,
+      "learning_rate": 9.957109494296235e-05,
+      "loss": 0.8029,
+      "step": 104100
+    },
+    {
+      "epoch": 3.5098356238210724,
+      "grad_norm": 0.6493478417396545,
+      "learning_rate": 9.934653732147668e-05,
+      "loss": 0.8022,
+      "step": 104200
+    },
+    {
+      "epoch": 3.5132039881433577,
+      "grad_norm": 0.6267081499099731,
+      "learning_rate": 9.9121979699991e-05,
+      "loss": 0.8027,
+      "step": 104300
+    },
+    {
+      "epoch": 3.5165723524656425,
+      "grad_norm": 0.7198253273963928,
+      "learning_rate": 9.889742207850534e-05,
+      "loss": 0.8021,
+      "step": 104400
+    },
+    {
+      "epoch": 3.5199407167879277,
+      "grad_norm": 0.7415684461593628,
+      "learning_rate": 9.867286445701966e-05,
+      "loss": 0.8049,
+      "step": 104500
+    },
+    {
+      "epoch": 3.523309081110213,
+      "grad_norm": 0.6261735558509827,
+      "learning_rate": 9.8448306835534e-05,
+      "loss": 0.8018,
+      "step": 104600
+    },
+    {
+      "epoch": 3.526677445432498,
+      "grad_norm": 0.6972131133079529,
+      "learning_rate": 9.822374921404832e-05,
+      "loss": 0.8046,
+      "step": 104700
+    },
+    {
+      "epoch": 3.530045809754783,
+      "grad_norm": 0.657211184501648,
+      "learning_rate": 9.799919159256265e-05,
+      "loss": 0.8049,
+      "step": 104800
+    },
+    {
+      "epoch": 3.533414174077068,
+      "grad_norm": 0.6501233577728271,
+      "learning_rate": 9.777463397107697e-05,
+      "loss": 0.8004,
+      "step": 104900
+    },
+    {
+      "epoch": 3.536782538399353,
+      "grad_norm": 0.7275915741920471,
+      "learning_rate": 9.75500763495913e-05,
+      "loss": 0.804,
+      "step": 105000
+    },
+    {
+      "epoch": 3.5401509027216385,
+      "grad_norm": 0.6765680313110352,
+      "learning_rate": 9.732551872810562e-05,
+      "loss": 0.8044,
+      "step": 105100
+    },
+    {
+      "epoch": 3.5435192670439237,
+      "grad_norm": 0.7218645811080933,
+      "learning_rate": 9.710096110661995e-05,
+      "loss": 0.8034,
+      "step": 105200
+    },
+    {
+      "epoch": 3.5468876313662085,
+      "grad_norm": 0.713777482509613,
+      "learning_rate": 9.687640348513427e-05,
+      "loss": 0.8034,
+      "step": 105300
+    },
+    {
+      "epoch": 3.550255995688494,
+      "grad_norm": 0.6668462157249451,
+      "learning_rate": 9.665184586364861e-05,
+      "loss": 0.8029,
+      "step": 105400
+    },
+    {
+      "epoch": 3.5536243600107786,
+      "grad_norm": 0.7213618755340576,
+      "learning_rate": 9.642728824216293e-05,
+      "loss": 0.802,
+      "step": 105500
+    },
+    {
+      "epoch": 3.556992724333064,
+      "grad_norm": 0.6867049932479858,
+      "learning_rate": 9.620273062067726e-05,
+      "loss": 0.8027,
+      "step": 105600
+    },
+    {
+      "epoch": 3.560361088655349,
+      "grad_norm": 0.6288276314735413,
+      "learning_rate": 9.597817299919159e-05,
+      "loss": 0.8017,
+      "step": 105700
+    },
+    {
+      "epoch": 3.563729452977634,
+      "grad_norm": 0.6391323208808899,
+      "learning_rate": 9.575361537770592e-05,
+      "loss": 0.8013,
+      "step": 105800
+    },
+    {
+      "epoch": 3.5670978172999193,
+      "grad_norm": 0.6860449314117432,
+      "learning_rate": 9.552905775622024e-05,
+      "loss": 0.7998,
+      "step": 105900
+    },
+    {
+      "epoch": 3.570466181622204,
+      "grad_norm": 0.7907363772392273,
+      "learning_rate": 9.530450013473455e-05,
+      "loss": 0.8067,
+      "step": 106000
+    },
+    {
+      "epoch": 3.5738345459444893,
+      "grad_norm": 0.6776504516601562,
+      "learning_rate": 9.507994251324889e-05,
+      "loss": 0.8017,
+      "step": 106100
+    },
+    {
+      "epoch": 3.5772029102667746,
+      "grad_norm": 0.7182029485702515,
+      "learning_rate": 9.485538489176321e-05,
+      "loss": 0.7997,
+      "step": 106200
+    },
+    {
+      "epoch": 3.5805712745890594,
+      "grad_norm": 0.6875755786895752,
+      "learning_rate": 9.463082727027754e-05,
+      "loss": 0.8043,
+      "step": 106300
+    },
+    {
+      "epoch": 3.5839396389113447,
+      "grad_norm": 0.6848233938217163,
+      "learning_rate": 9.440626964879187e-05,
+      "loss": 0.8025,
+      "step": 106400
+    },
+    {
+      "epoch": 3.5873080032336295,
+      "grad_norm": 0.6334069967269897,
+      "learning_rate": 9.41817120273062e-05,
+      "loss": 0.8032,
+      "step": 106500
+    },
+    {
+      "epoch": 3.590676367555915,
+      "grad_norm": 0.6613095998764038,
+      "learning_rate": 9.395715440582052e-05,
+      "loss": 0.8014,
+      "step": 106600
+    },
+    {
+      "epoch": 3.5940447318782,
+      "grad_norm": 0.6903994083404541,
+      "learning_rate": 9.373259678433486e-05,
+      "loss": 0.8043,
+      "step": 106700
+    },
+    {
+      "epoch": 3.597413096200485,
+      "grad_norm": 0.6296311616897583,
+      "learning_rate": 9.350803916284918e-05,
+      "loss": 0.8003,
+      "step": 106800
+    },
+    {
+      "epoch": 3.60078146052277,
+      "grad_norm": 0.7521107196807861,
+      "learning_rate": 9.328348154136351e-05,
+      "loss": 0.7999,
+      "step": 106900
+    },
+    {
+      "epoch": 3.604149824845055,
+      "grad_norm": 0.6513036489486694,
+      "learning_rate": 9.305892391987782e-05,
+      "loss": 0.8039,
+      "step": 107000
+    },
+    {
+      "epoch": 3.6075181891673402,
+      "grad_norm": 0.6684443950653076,
+      "learning_rate": 9.283436629839216e-05,
+      "loss": 0.8027,
+      "step": 107100
+    },
+    {
+      "epoch": 3.6108865534896255,
+      "grad_norm": 0.6627715826034546,
+      "learning_rate": 9.260980867690648e-05,
+      "loss": 0.8025,
+      "step": 107200
+    },
+    {
+      "epoch": 3.6142549178119108,
+      "grad_norm": 0.7347738742828369,
+      "learning_rate": 9.238525105542081e-05,
+      "loss": 0.8036,
+      "step": 107300
+    },
+    {
+      "epoch": 3.6176232821341956,
+      "grad_norm": 0.7007977366447449,
+      "learning_rate": 9.216069343393514e-05,
+      "loss": 0.8035,
+      "step": 107400
+    },
+    {
+      "epoch": 3.620991646456481,
+      "grad_norm": 0.6845256686210632,
+      "learning_rate": 9.193613581244947e-05,
+      "loss": 0.8032,
+      "step": 107500
+    },
+    {
+      "epoch": 3.6243600107787657,
+      "grad_norm": 0.7089165449142456,
+      "learning_rate": 9.171157819096379e-05,
+      "loss": 0.8046,
+      "step": 107600
+    },
+    {
+      "epoch": 3.627728375101051,
+      "grad_norm": 0.7869235873222351,
+      "learning_rate": 9.148702056947813e-05,
+      "loss": 0.8019,
+      "step": 107700
+    },
+    {
+      "epoch": 3.631096739423336,
+      "grad_norm": 0.6592691540718079,
+      "learning_rate": 9.126246294799245e-05,
+      "loss": 0.8012,
+      "step": 107800
+    },
+    {
+      "epoch": 3.634465103745621,
+      "grad_norm": 0.6670995354652405,
+      "learning_rate": 9.103790532650677e-05,
+      "loss": 0.8028,
+      "step": 107900
+    },
+    {
+      "epoch": 3.6378334680679063,
+      "grad_norm": 0.7031666040420532,
+      "learning_rate": 9.081334770502109e-05,
+      "loss": 0.8017,
+      "step": 108000
+    },
+    {
+      "epoch": 3.641201832390191,
+      "grad_norm": 0.7147188782691956,
+      "learning_rate": 9.058879008353543e-05,
+      "loss": 0.7982,
+      "step": 108100
+    },
+    {
+      "epoch": 3.6445701967124764,
+      "grad_norm": 0.7295541763305664,
+      "learning_rate": 9.036423246204975e-05,
+      "loss": 0.8005,
+      "step": 108200
+    },
+    {
+      "epoch": 3.6479385610347617,
+      "grad_norm": 0.6590719819068909,
+      "learning_rate": 9.013967484056408e-05,
+      "loss": 0.8037,
+      "step": 108300
+    },
+    {
+      "epoch": 3.6513069253570465,
+      "grad_norm": 0.6898632049560547,
+      "learning_rate": 8.99151172190784e-05,
+      "loss": 0.8002,
+      "step": 108400
+    },
+    {
+      "epoch": 3.6546752896793318,
+      "grad_norm": 0.661747395992279,
+      "learning_rate": 8.969055959759274e-05,
+      "loss": 0.8035,
+      "step": 108500
+    },
+    {
+      "epoch": 3.6580436540016166,
+      "grad_norm": 0.7051920294761658,
+      "learning_rate": 8.946600197610706e-05,
+      "loss": 0.801,
+      "step": 108600
+    },
+    {
+      "epoch": 3.661412018323902,
+      "grad_norm": 0.6946534514427185,
+      "learning_rate": 8.92414443546214e-05,
+      "loss": 0.803,
+      "step": 108700
+    },
+    {
+      "epoch": 3.664780382646187,
+      "grad_norm": 0.649159848690033,
+      "learning_rate": 8.901688673313572e-05,
+      "loss": 0.8003,
+      "step": 108800
+    },
+    {
+      "epoch": 3.6681487469684724,
+      "grad_norm": 0.6959517002105713,
+      "learning_rate": 8.879232911165003e-05,
+      "loss": 0.8025,
+      "step": 108900
+    },
+    {
+      "epoch": 3.671517111290757,
+      "grad_norm": 0.7102181315422058,
+      "learning_rate": 8.856777149016436e-05,
+      "loss": 0.8003,
+      "step": 109000
+    },
+    {
+      "epoch": 3.6748854756130425,
+      "grad_norm": 0.6565383076667786,
+      "learning_rate": 8.83432138686787e-05,
+      "loss": 0.7994,
+      "step": 109100
+    },
+    {
+      "epoch": 3.6782538399353273,
+      "grad_norm": 0.6567991375923157,
+      "learning_rate": 8.811865624719302e-05,
+      "loss": 0.802,
+      "step": 109200
+    },
+    {
+      "epoch": 3.6816222042576126,
+      "grad_norm": 0.6707866787910461,
+      "learning_rate": 8.789409862570736e-05,
+      "loss": 0.7997,
+      "step": 109300
+    },
+    {
+      "epoch": 3.684990568579898,
+      "grad_norm": 0.6689081192016602,
+      "learning_rate": 8.766954100422168e-05,
+      "loss": 0.8005,
+      "step": 109400
+    },
+    {
+      "epoch": 3.6883589329021826,
+      "grad_norm": 0.6506887078285217,
+      "learning_rate": 8.744498338273601e-05,
+      "loss": 0.8017,
+      "step": 109500
+    },
+    {
+      "epoch": 3.691727297224468,
+      "grad_norm": 0.6316550970077515,
+      "learning_rate": 8.722042576125033e-05,
+      "loss": 0.8001,
+      "step": 109600
+    },
+    {
+      "epoch": 3.6950956615467527,
+      "grad_norm": 0.6964483261108398,
+      "learning_rate": 8.699586813976467e-05,
+      "loss": 0.8005,
+      "step": 109700
+    },
+    {
+      "epoch": 3.698464025869038,
+      "grad_norm": 0.6844159364700317,
+      "learning_rate": 8.677131051827898e-05,
+      "loss": 0.7971,
+      "step": 109800
+    },
+    {
+      "epoch": 3.7018323901913233,
+      "grad_norm": 0.5916749835014343,
+      "learning_rate": 8.65467528967933e-05,
+      "loss": 0.8008,
+      "step": 109900
+    },
+    {
+      "epoch": 3.705200754513608,
+      "grad_norm": 0.7180382609367371,
+      "learning_rate": 8.632219527530763e-05,
+      "loss": 0.7967,
+      "step": 110000
+    },
+    {
+      "epoch": 3.7085691188358934,
+      "grad_norm": 0.6756430864334106,
+      "learning_rate": 8.609763765382196e-05,
+      "loss": 0.8002,
+      "step": 110100
+    },
+    {
+      "epoch": 3.711937483158178,
+      "grad_norm": 0.6347695589065552,
+      "learning_rate": 8.587308003233629e-05,
+      "loss": 0.801,
+      "step": 110200
+    },
+    {
+      "epoch": 3.7153058474804634,
+      "grad_norm": 0.712821364402771,
+      "learning_rate": 8.564852241085061e-05,
+      "loss": 0.8,
+      "step": 110300
+    },
+    {
+      "epoch": 3.7186742118027487,
+      "grad_norm": 0.7329118251800537,
+      "learning_rate": 8.542396478936495e-05,
+      "loss": 0.7991,
+      "step": 110400
+    },
+    {
+      "epoch": 3.7220425761250335,
+      "grad_norm": 0.6399374008178711,
+      "learning_rate": 8.519940716787927e-05,
+      "loss": 0.802,
+      "step": 110500
+    },
+    {
+      "epoch": 3.725410940447319,
+      "grad_norm": 0.7076860666275024,
+      "learning_rate": 8.49748495463936e-05,
+      "loss": 0.8004,
+      "step": 110600
+    },
+    {
+      "epoch": 3.7287793047696036,
+      "grad_norm": 0.6736636161804199,
+      "learning_rate": 8.475029192490793e-05,
+      "loss": 0.7985,
+      "step": 110700
+    },
+    {
+      "epoch": 3.732147669091889,
+      "grad_norm": 0.7174369692802429,
+      "learning_rate": 8.452573430342225e-05,
+      "loss": 0.7997,
+      "step": 110800
+    },
+    {
+      "epoch": 3.735516033414174,
+      "grad_norm": 0.6668689250946045,
+      "learning_rate": 8.430117668193657e-05,
+      "loss": 0.7984,
+      "step": 110900
+    },
+    {
+      "epoch": 3.7388843977364594,
+      "grad_norm": 0.6991139054298401,
+      "learning_rate": 8.40766190604509e-05,
+      "loss": 0.8011,
+      "step": 111000
+    },
+    {
+      "epoch": 3.7422527620587442,
+      "grad_norm": 0.6559997797012329,
+      "learning_rate": 8.385206143896523e-05,
+      "loss": 0.8002,
+      "step": 111100
+    },
+    {
+      "epoch": 3.7456211263810295,
+      "grad_norm": 0.6859176754951477,
+      "learning_rate": 8.362750381747956e-05,
+      "loss": 0.799,
+      "step": 111200
+    },
+    {
+      "epoch": 3.7489894907033143,
+      "grad_norm": 0.7021101117134094,
+      "learning_rate": 8.340294619599388e-05,
+      "loss": 0.796,
+      "step": 111300
+    },
+    {
+      "epoch": 3.7523578550255996,
+      "grad_norm": 0.6929513216018677,
+      "learning_rate": 8.317838857450822e-05,
+      "loss": 0.7977,
+      "step": 111400
+    },
+    {
+      "epoch": 3.755726219347885,
+      "grad_norm": 0.6644559502601624,
+      "learning_rate": 8.295383095302254e-05,
+      "loss": 0.7994,
+      "step": 111500
+    },
+    {
+      "epoch": 3.7590945836701697,
+      "grad_norm": 0.6280823349952698,
+      "learning_rate": 8.272927333153687e-05,
+      "loss": 0.7974,
+      "step": 111600
+    },
+    {
+      "epoch": 3.762462947992455,
+      "grad_norm": 0.699496328830719,
+      "learning_rate": 8.25047157100512e-05,
+      "loss": 0.7995,
+      "step": 111700
+    },
+    {
+      "epoch": 3.7658313123147398,
+      "grad_norm": 0.7624558806419373,
+      "learning_rate": 8.228015808856552e-05,
+      "loss": 0.8004,
+      "step": 111800
+    },
+    {
+      "epoch": 3.769199676637025,
+      "grad_norm": 0.7288152575492859,
+      "learning_rate": 8.205560046707984e-05,
+      "loss": 0.8029,
+      "step": 111900
+    },
+    {
+      "epoch": 3.7725680409593103,
+      "grad_norm": 0.6420552730560303,
+      "learning_rate": 8.183104284559417e-05,
+      "loss": 0.8015,
+      "step": 112000
+    },
+    {
+      "epoch": 3.775936405281595,
+      "grad_norm": 0.6827369332313538,
+      "learning_rate": 8.16064852241085e-05,
+      "loss": 0.7997,
+      "step": 112100
+    },
+    {
+      "epoch": 3.7793047696038804,
+      "grad_norm": 0.7149909138679504,
+      "learning_rate": 8.138192760262283e-05,
+      "loss": 0.8008,
+      "step": 112200
+    },
+    {
+      "epoch": 3.782673133926165,
+      "grad_norm": 0.7044945955276489,
+      "learning_rate": 8.115736998113715e-05,
+      "loss": 0.7974,
+      "step": 112300
+    },
+    {
+      "epoch": 3.7860414982484505,
+      "grad_norm": 0.7744246125221252,
+      "learning_rate": 8.093281235965149e-05,
+      "loss": 0.7971,
+      "step": 112400
+    },
+    {
+      "epoch": 3.7894098625707358,
+      "grad_norm": 0.6370006203651428,
+      "learning_rate": 8.070825473816581e-05,
+      "loss": 0.7987,
+      "step": 112500
+    },
+    {
+      "epoch": 3.792778226893021,
+      "grad_norm": 0.6973426342010498,
+      "learning_rate": 8.048369711668015e-05,
+      "loss": 0.7999,
+      "step": 112600
+    },
+    {
+      "epoch": 3.796146591215306,
+      "grad_norm": 0.7414847612380981,
+      "learning_rate": 8.025913949519445e-05,
+      "loss": 0.7979,
+      "step": 112700
+    },
+    {
+      "epoch": 3.799514955537591,
+      "grad_norm": 0.6913410425186157,
+      "learning_rate": 8.003458187370878e-05,
+      "loss": 0.796,
+      "step": 112800
+    },
+    {
+      "epoch": 3.802883319859876,
+      "grad_norm": 0.6863036155700684,
+      "learning_rate": 7.981002425222311e-05,
+      "loss": 0.7975,
+      "step": 112900
+    },
+    {
+      "epoch": 3.806251684182161,
+      "grad_norm": 0.804077684879303,
+      "learning_rate": 7.958546663073743e-05,
+      "loss": 0.7968,
+      "step": 113000
+    },
+    {
+      "epoch": 3.8096200485044465,
+      "grad_norm": 1.013036847114563,
+      "learning_rate": 7.936090900925177e-05,
+      "loss": 0.7999,
+      "step": 113100
+    },
+    {
+      "epoch": 3.8129884128267313,
+      "grad_norm": 0.7045647501945496,
+      "learning_rate": 7.913635138776609e-05,
+      "loss": 0.7991,
+      "step": 113200
+    },
+    {
+      "epoch": 3.8163567771490166,
+      "grad_norm": 0.6505812406539917,
+      "learning_rate": 7.891179376628042e-05,
+      "loss": 0.8006,
+      "step": 113300
+    },
+    {
+      "epoch": 3.8197251414713014,
+      "grad_norm": 0.6592190861701965,
+      "learning_rate": 7.868723614479475e-05,
+      "loss": 0.7991,
+      "step": 113400
+    },
+    {
+      "epoch": 3.8230935057935866,
+      "grad_norm": 0.6840342879295349,
+      "learning_rate": 7.846267852330908e-05,
+      "loss": 0.798,
+      "step": 113500
+    },
+    {
+      "epoch": 3.826461870115872,
+      "grad_norm": 0.6523902416229248,
+      "learning_rate": 7.82381209018234e-05,
+      "loss": 0.7982,
+      "step": 113600
+    },
+    {
+      "epoch": 3.8298302344381567,
+      "grad_norm": 0.6831576228141785,
+      "learning_rate": 7.801356328033772e-05,
+      "loss": 0.7988,
+      "step": 113700
+    },
+    {
+      "epoch": 3.833198598760442,
+      "grad_norm": 0.6503862738609314,
+      "learning_rate": 7.778900565885205e-05,
+      "loss": 0.7995,
+      "step": 113800
+    },
+    {
+      "epoch": 3.836566963082727,
+      "grad_norm": 0.6413772702217102,
+      "learning_rate": 7.756444803736638e-05,
+      "loss": 0.7992,
+      "step": 113900
+    },
+    {
+      "epoch": 3.839935327405012,
+      "grad_norm": 0.708017885684967,
+      "learning_rate": 7.73398904158807e-05,
+      "loss": 0.7995,
+      "step": 114000
+    },
+    {
+      "epoch": 3.8433036917272974,
+      "grad_norm": 0.686399519443512,
+      "learning_rate": 7.711533279439504e-05,
+      "loss": 0.7974,
+      "step": 114100
+    },
+    {
+      "epoch": 3.846672056049582,
+      "grad_norm": 0.6614728569984436,
+      "learning_rate": 7.689077517290936e-05,
+      "loss": 0.7972,
+      "step": 114200
+    },
+    {
+      "epoch": 3.8500404203718674,
+      "grad_norm": 1.1526070833206177,
+      "learning_rate": 7.66662175514237e-05,
+      "loss": 0.795,
+      "step": 114300
+    },
+    {
+      "epoch": 3.8534087846941523,
+      "grad_norm": 0.6846965551376343,
+      "learning_rate": 7.644165992993802e-05,
+      "loss": 0.799,
+      "step": 114400
+    },
+    {
+      "epoch": 3.8567771490164375,
+      "grad_norm": 0.699941098690033,
+      "learning_rate": 7.621710230845235e-05,
+      "loss": 0.796,
+      "step": 114500
+    },
+    {
+      "epoch": 3.860145513338723,
+      "grad_norm": 0.7409800291061401,
+      "learning_rate": 7.599254468696666e-05,
+      "loss": 0.7975,
+      "step": 114600
+    },
+    {
+      "epoch": 3.863513877661008,
+      "grad_norm": 0.7050840854644775,
+      "learning_rate": 7.5767987065481e-05,
+      "loss": 0.7965,
+      "step": 114700
+    },
+    {
+      "epoch": 3.866882241983293,
+      "grad_norm": 0.6288536787033081,
+      "learning_rate": 7.554342944399532e-05,
+      "loss": 0.7987,
+      "step": 114800
+    },
+    {
+      "epoch": 3.870250606305578,
+      "grad_norm": 0.69952791929245,
+      "learning_rate": 7.531887182250965e-05,
+      "loss": 0.7983,
+      "step": 114900
+    },
+    {
+      "epoch": 3.873618970627863,
+      "grad_norm": 0.7967835664749146,
+      "learning_rate": 7.509431420102397e-05,
+      "loss": 0.7992,
+      "step": 115000
+    },
+    {
+      "epoch": 3.8769873349501482,
+      "grad_norm": 0.6878296732902527,
+      "learning_rate": 7.486975657953831e-05,
+      "loss": 0.7953,
+      "step": 115100
+    },
+    {
+      "epoch": 3.8803556992724335,
+      "grad_norm": 0.750624418258667,
+      "learning_rate": 7.464519895805263e-05,
+      "loss": 0.7972,
+      "step": 115200
+    },
+    {
+      "epoch": 3.8837240635947183,
+      "grad_norm": 0.6286699771881104,
+      "learning_rate": 7.442064133656697e-05,
+      "loss": 0.7979,
+      "step": 115300
+    },
+    {
+      "epoch": 3.8870924279170036,
+      "grad_norm": 0.7046734094619751,
+      "learning_rate": 7.419608371508129e-05,
+      "loss": 0.7991,
+      "step": 115400
+    },
+    {
+      "epoch": 3.8904607922392884,
+      "grad_norm": 0.6798041462898254,
+      "learning_rate": 7.397152609359561e-05,
+      "loss": 0.8019,
+      "step": 115500
+    },
+    {
+      "epoch": 3.8938291565615737,
+      "grad_norm": 0.6814720034599304,
+      "learning_rate": 7.374696847210994e-05,
+      "loss": 0.7992,
+      "step": 115600
+    },
+    {
+      "epoch": 3.897197520883859,
+      "grad_norm": 0.6247894167900085,
+      "learning_rate": 7.352241085062427e-05,
+      "loss": 0.7959,
+      "step": 115700
+    },
+    {
+      "epoch": 3.900565885206144,
+      "grad_norm": 0.6836827397346497,
+      "learning_rate": 7.329785322913859e-05,
+      "loss": 0.7971,
+      "step": 115800
+    },
+    {
+      "epoch": 3.903934249528429,
+      "grad_norm": 0.6945433616638184,
+      "learning_rate": 7.307329560765291e-05,
+      "loss": 0.7947,
+      "step": 115900
+    },
+    {
+      "epoch": 3.907302613850714,
+      "grad_norm": 0.7244613766670227,
+      "learning_rate": 7.284873798616724e-05,
+      "loss": 0.7966,
+      "step": 116000
+    },
+    {
+      "epoch": 3.910670978172999,
+      "grad_norm": 0.6739228367805481,
+      "learning_rate": 7.262418036468158e-05,
+      "loss": 0.7987,
+      "step": 116100
+    },
+    {
+      "epoch": 3.9140393424952844,
+      "grad_norm": 0.7027273774147034,
+      "learning_rate": 7.23996227431959e-05,
+      "loss": 0.7966,
+      "step": 116200
+    },
+    {
+      "epoch": 3.9174077068175692,
+      "grad_norm": 0.7074873447418213,
+      "learning_rate": 7.217506512171022e-05,
+      "loss": 0.7953,
+      "step": 116300
+    },
+    {
+      "epoch": 3.9207760711398545,
+      "grad_norm": 0.6511521935462952,
+      "learning_rate": 7.195050750022454e-05,
+      "loss": 0.8015,
+      "step": 116400
+    },
+    {
+      "epoch": 3.9241444354621393,
+      "grad_norm": 0.7246668934822083,
+      "learning_rate": 7.172594987873888e-05,
+      "loss": 0.7963,
+      "step": 116500
+    },
+    {
+      "epoch": 3.9275127997844246,
+      "grad_norm": 0.6860557198524475,
+      "learning_rate": 7.15013922572532e-05,
+      "loss": 0.7973,
+      "step": 116600
+    },
+    {
+      "epoch": 3.93088116410671,
+      "grad_norm": 0.7153899669647217,
+      "learning_rate": 7.127683463576754e-05,
+      "loss": 0.7959,
+      "step": 116700
+    },
+    {
+      "epoch": 3.934249528428995,
+      "grad_norm": 0.6767284274101257,
+      "learning_rate": 7.105227701428186e-05,
+      "loss": 0.7971,
+      "step": 116800
+    },
+    {
+      "epoch": 3.93761789275128,
+      "grad_norm": 0.6651219129562378,
+      "learning_rate": 7.082771939279618e-05,
+      "loss": 0.7959,
+      "step": 116900
+    },
+    {
+      "epoch": 3.940986257073565,
+      "grad_norm": 0.6394347548484802,
+      "learning_rate": 7.060316177131051e-05,
+      "loss": 0.7988,
+      "step": 117000
+    },
+    {
+      "epoch": 3.94435462139585,
+      "grad_norm": 0.6922232508659363,
+      "learning_rate": 7.037860414982484e-05,
+      "loss": 0.7961,
+      "step": 117100
+    },
+    {
+      "epoch": 3.9477229857181353,
+      "grad_norm": 0.6803662776947021,
+      "learning_rate": 7.015404652833917e-05,
+      "loss": 0.7982,
+      "step": 117200
+    },
+    {
+      "epoch": 3.9510913500404206,
+      "grad_norm": 0.7751142978668213,
+      "learning_rate": 6.992948890685349e-05,
+      "loss": 0.7968,
+      "step": 117300
+    },
+    {
+      "epoch": 3.9544597143627054,
+      "grad_norm": 0.6750118136405945,
+      "learning_rate": 6.970493128536781e-05,
+      "loss": 0.7955,
+      "step": 117400
+    },
+    {
+      "epoch": 3.9578280786849906,
+      "grad_norm": 0.6720499992370605,
+      "learning_rate": 6.948037366388215e-05,
+      "loss": 0.7975,
+      "step": 117500
+    },
+    {
+      "epoch": 3.9611964430072755,
+      "grad_norm": 0.7234087586402893,
+      "learning_rate": 6.925581604239647e-05,
+      "loss": 0.7954,
+      "step": 117600
+    },
+    {
+      "epoch": 3.9645648073295607,
+      "grad_norm": 0.6893304586410522,
+      "learning_rate": 6.903125842091079e-05,
+      "loss": 0.7958,
+      "step": 117700
+    },
+    {
+      "epoch": 3.967933171651846,
+      "grad_norm": 0.6690725684165955,
+      "learning_rate": 6.880670079942513e-05,
+      "loss": 0.794,
+      "step": 117800
+    },
+    {
+      "epoch": 3.971301535974131,
+      "grad_norm": 0.6747570633888245,
+      "learning_rate": 6.858214317793945e-05,
+      "loss": 0.7962,
+      "step": 117900
+    },
+    {
+      "epoch": 3.974669900296416,
+      "grad_norm": 0.7051836848258972,
+      "learning_rate": 6.835758555645378e-05,
+      "loss": 0.7947,
+      "step": 118000
+    },
+    {
+      "epoch": 3.978038264618701,
+      "grad_norm": 0.6995989084243774,
+      "learning_rate": 6.81330279349681e-05,
+      "loss": 0.795,
+      "step": 118100
+    },
+    {
+      "epoch": 3.981406628940986,
+      "grad_norm": 0.6704210638999939,
+      "learning_rate": 6.790847031348243e-05,
+      "loss": 0.7944,
+      "step": 118200
+    },
+    {
+      "epoch": 3.9847749932632714,
+      "grad_norm": 0.6781893372535706,
+      "learning_rate": 6.768391269199676e-05,
+      "loss": 0.7954,
+      "step": 118300
+    },
+    {
+      "epoch": 3.9881433575855567,
+      "grad_norm": 0.6674184203147888,
+      "learning_rate": 6.745935507051109e-05,
+      "loss": 0.7984,
+      "step": 118400
+    },
+    {
+      "epoch": 3.9915117219078415,
+      "grad_norm": 0.6545355319976807,
+      "learning_rate": 6.723479744902542e-05,
+      "loss": 0.7946,
+      "step": 118500
+    },
+    {
+      "epoch": 3.994880086230127,
+      "grad_norm": 0.6851162910461426,
+      "learning_rate": 6.701023982753974e-05,
+      "loss": 0.7954,
+      "step": 118600
+    },
+    {
+      "epoch": 3.9982484505524116,
+      "grad_norm": 0.6962630152702332,
+      "learning_rate": 6.678568220605406e-05,
+      "loss": 0.7949,
+      "step": 118700
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.7465363144874573,
+      "eval_runtime": 1.8135,
+      "eval_samples_per_second": 2757.131,
+      "eval_steps_per_second": 43.563,
+      "step": 118752
+    },
+    {
+      "epoch": 4.001616814874697,
+      "grad_norm": 0.7494738101959229,
+      "learning_rate": 6.65611245845684e-05,
+      "loss": 0.7985,
+      "step": 118800
+    },
+    {
+      "epoch": 4.004985179196982,
+      "grad_norm": 0.6624446511268616,
+      "learning_rate": 6.633656696308272e-05,
+      "loss": 0.7954,
+      "step": 118900
+    },
+    {
+      "epoch": 4.008353543519267,
+      "grad_norm": 0.69657963514328,
+      "learning_rate": 6.611200934159706e-05,
+      "loss": 0.7957,
+      "step": 119000
+    },
+    {
+      "epoch": 4.011721907841552,
+      "grad_norm": 0.706929087638855,
+      "learning_rate": 6.588745172011138e-05,
+      "loss": 0.7969,
+      "step": 119100
+    },
+    {
+      "epoch": 4.015090272163837,
+      "grad_norm": 0.7379885911941528,
+      "learning_rate": 6.56628940986257e-05,
+      "loss": 0.7968,
+      "step": 119200
+    },
+    {
+      "epoch": 4.018458636486122,
+      "grad_norm": 0.6987789869308472,
+      "learning_rate": 6.543833647714002e-05,
+      "loss": 0.7959,
+      "step": 119300
+    },
+    {
+      "epoch": 4.021827000808408,
+      "grad_norm": 0.6569750308990479,
+      "learning_rate": 6.521377885565436e-05,
+      "loss": 0.7917,
+      "step": 119400
+    },
+    {
+      "epoch": 4.025195365130693,
+      "grad_norm": 0.7094253301620483,
+      "learning_rate": 6.498922123416869e-05,
+      "loss": 0.7937,
+      "step": 119500
+    },
+    {
+      "epoch": 4.028563729452977,
+      "grad_norm": 0.7103100419044495,
+      "learning_rate": 6.476466361268301e-05,
+      "loss": 0.7988,
+      "step": 119600
+    },
+    {
+      "epoch": 4.0319320937752625,
+      "grad_norm": 0.7065451145172119,
+      "learning_rate": 6.454010599119733e-05,
+      "loss": 0.7905,
+      "step": 119700
+    },
+    {
+      "epoch": 4.035300458097548,
+      "grad_norm": 0.690153181552887,
+      "learning_rate": 6.431554836971166e-05,
+      "loss": 0.7948,
+      "step": 119800
+    },
+    {
+      "epoch": 4.038668822419833,
+      "grad_norm": 0.7387445569038391,
+      "learning_rate": 6.409099074822599e-05,
+      "loss": 0.7944,
+      "step": 119900
+    },
+    {
+      "epoch": 4.042037186742118,
+      "grad_norm": 0.6826400756835938,
+      "learning_rate": 6.386643312674031e-05,
+      "loss": 0.7935,
+      "step": 120000
+    },
+    {
+      "epoch": 4.045405551064403,
+      "grad_norm": 0.6468490362167358,
+      "learning_rate": 6.364187550525463e-05,
+      "loss": 0.796,
+      "step": 120100
+    },
+    {
+      "epoch": 4.048773915386688,
+      "grad_norm": 0.6864830851554871,
+      "learning_rate": 6.341731788376897e-05,
+      "loss": 0.7948,
+      "step": 120200
+    },
+    {
+      "epoch": 4.052142279708973,
+      "grad_norm": 0.7382389903068542,
+      "learning_rate": 6.319276026228329e-05,
+      "loss": 0.7943,
+      "step": 120300
+    },
+    {
+      "epoch": 4.0555106440312585,
+      "grad_norm": 0.7259221076965332,
+      "learning_rate": 6.296820264079763e-05,
+      "loss": 0.7975,
+      "step": 120400
+    },
+    {
+      "epoch": 4.058879008353544,
+      "grad_norm": 0.7078028917312622,
+      "learning_rate": 6.274364501931195e-05,
+      "loss": 0.7969,
+      "step": 120500
+    },
+    {
+      "epoch": 4.062247372675829,
+      "grad_norm": 0.6797070503234863,
+      "learning_rate": 6.251908739782627e-05,
+      "loss": 0.7946,
+      "step": 120600
+    },
+    {
+      "epoch": 4.065615736998113,
+      "grad_norm": 0.7093040347099304,
+      "learning_rate": 6.22945297763406e-05,
+      "loss": 0.7931,
+      "step": 120700
+    },
+    {
+      "epoch": 4.068984101320399,
+      "grad_norm": 0.6741282939910889,
+      "learning_rate": 6.206997215485493e-05,
+      "loss": 0.7975,
+      "step": 120800
+    },
+    {
+      "epoch": 4.072352465642684,
+      "grad_norm": 0.6645729541778564,
+      "learning_rate": 6.184541453336926e-05,
+      "loss": 0.7959,
+      "step": 120900
+    },
+    {
+      "epoch": 4.075720829964969,
+      "grad_norm": 0.6510828137397766,
+      "learning_rate": 6.162085691188358e-05,
+      "loss": 0.7945,
+      "step": 121000
+    },
+    {
+      "epoch": 4.0790891942872545,
+      "grad_norm": 0.7214908599853516,
+      "learning_rate": 6.13962992903979e-05,
+      "loss": 0.7934,
+      "step": 121100
+    },
+    {
+      "epoch": 4.082457558609539,
+      "grad_norm": 0.7020803093910217,
+      "learning_rate": 6.117174166891224e-05,
+      "loss": 0.7952,
+      "step": 121200
+    },
+    {
+      "epoch": 4.085825922931824,
+      "grad_norm": 0.7014907002449036,
+      "learning_rate": 6.094718404742656e-05,
+      "loss": 0.7932,
+      "step": 121300
+    },
+    {
+      "epoch": 4.089194287254109,
+      "grad_norm": 0.6577868461608887,
+      "learning_rate": 6.072262642594089e-05,
+      "loss": 0.7904,
+      "step": 121400
+    },
+    {
+      "epoch": 4.092562651576395,
+      "grad_norm": 0.6450611352920532,
+      "learning_rate": 6.049806880445522e-05,
+      "loss": 0.7917,
+      "step": 121500
+    },
+    {
+      "epoch": 4.09593101589868,
+      "grad_norm": 0.7224249243736267,
+      "learning_rate": 6.027351118296954e-05,
+      "loss": 0.793,
+      "step": 121600
+    },
+    {
+      "epoch": 4.099299380220964,
+      "grad_norm": 0.6740891337394714,
+      "learning_rate": 6.004895356148387e-05,
+      "loss": 0.7929,
+      "step": 121700
+    },
+    {
+      "epoch": 4.10266774454325,
+      "grad_norm": 0.7219831347465515,
+      "learning_rate": 5.98243959399982e-05,
+      "loss": 0.7921,
+      "step": 121800
+    },
+    {
+      "epoch": 4.106036108865535,
+      "grad_norm": 0.7117908596992493,
+      "learning_rate": 5.9599838318512525e-05,
+      "loss": 0.7936,
+      "step": 121900
+    },
+    {
+      "epoch": 4.10940447318782,
+      "grad_norm": 0.8350685238838196,
+      "learning_rate": 5.9375280697026854e-05,
+      "loss": 0.7924,
+      "step": 122000
+    },
+    {
+      "epoch": 4.112772837510105,
+      "grad_norm": 0.6902172565460205,
+      "learning_rate": 5.9150723075541175e-05,
+      "loss": 0.7931,
+      "step": 122100
+    },
+    {
+      "epoch": 4.116141201832391,
+      "grad_norm": 0.6757128238677979,
+      "learning_rate": 5.8926165454055504e-05,
+      "loss": 0.7954,
+      "step": 122200
+    },
+    {
+      "epoch": 4.119509566154675,
+      "grad_norm": 0.6877591013908386,
+      "learning_rate": 5.870160783256983e-05,
+      "loss": 0.7916,
+      "step": 122300
+    },
+    {
+      "epoch": 4.12287793047696,
+      "grad_norm": 0.677144467830658,
+      "learning_rate": 5.847705021108416e-05,
+      "loss": 0.7951,
+      "step": 122400
+    },
+    {
+      "epoch": 4.1262462947992455,
+      "grad_norm": 0.6870157122612,
+      "learning_rate": 5.825249258959848e-05,
+      "loss": 0.7931,
+      "step": 122500
+    },
+    {
+      "epoch": 4.129614659121531,
+      "grad_norm": 0.7329301834106445,
+      "learning_rate": 5.802793496811281e-05,
+      "loss": 0.7947,
+      "step": 122600
+    },
+    {
+      "epoch": 4.132983023443816,
+      "grad_norm": 0.7153013348579407,
+      "learning_rate": 5.780337734662714e-05,
+      "loss": 0.7927,
+      "step": 122700
+    },
+    {
+      "epoch": 4.1363513877661005,
+      "grad_norm": 0.6681983470916748,
+      "learning_rate": 5.757881972514147e-05,
+      "loss": 0.7966,
+      "step": 122800
+    },
+    {
+      "epoch": 4.139719752088386,
+      "grad_norm": 0.6760982275009155,
+      "learning_rate": 5.7354262103655796e-05,
+      "loss": 0.7932,
+      "step": 122900
+    },
+    {
+      "epoch": 4.143088116410671,
+      "grad_norm": 0.6279290914535522,
+      "learning_rate": 5.712970448217012e-05,
+      "loss": 0.7936,
+      "step": 123000
+    },
+    {
+      "epoch": 4.146456480732956,
+      "grad_norm": 0.6763176918029785,
+      "learning_rate": 5.6905146860684446e-05,
+      "loss": 0.7966,
+      "step": 123100
+    },
+    {
+      "epoch": 4.1498248450552415,
+      "grad_norm": 0.6843128204345703,
+      "learning_rate": 5.6680589239198774e-05,
+      "loss": 0.7933,
+      "step": 123200
+    },
+    {
+      "epoch": 4.153193209377526,
+      "grad_norm": 0.6466397643089294,
+      "learning_rate": 5.64560316177131e-05,
+      "loss": 0.7955,
+      "step": 123300
+    },
+    {
+      "epoch": 4.156561573699811,
+      "grad_norm": 0.6983882188796997,
+      "learning_rate": 5.623147399622743e-05,
+      "loss": 0.7951,
+      "step": 123400
+    },
+    {
+      "epoch": 4.159929938022096,
+      "grad_norm": 0.6982767581939697,
+      "learning_rate": 5.600691637474175e-05,
+      "loss": 0.7932,
+      "step": 123500
+    },
+    {
+      "epoch": 4.163298302344382,
+      "grad_norm": 0.6864616870880127,
+      "learning_rate": 5.578235875325608e-05,
+      "loss": 0.7926,
+      "step": 123600
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 0.7273553609848022,
+      "learning_rate": 5.555780113177041e-05,
+      "loss": 0.7935,
+      "step": 123700
+    },
+    {
+      "epoch": 4.170035030988951,
+      "grad_norm": 0.6874902248382568,
+      "learning_rate": 5.533324351028474e-05,
+      "loss": 0.7913,
+      "step": 123800
+    },
+    {
+      "epoch": 4.173403395311237,
+      "grad_norm": 0.7229161262512207,
+      "learning_rate": 5.5108685888799067e-05,
+      "loss": 0.7887,
+      "step": 123900
+    },
+    {
+      "epoch": 4.176771759633522,
+      "grad_norm": 0.7190577387809753,
+      "learning_rate": 5.488412826731338e-05,
+      "loss": 0.7908,
+      "step": 124000
+    },
+    {
+      "epoch": 4.180140123955807,
+      "grad_norm": 0.6708100438117981,
+      "learning_rate": 5.4659570645827717e-05,
+      "loss": 0.7937,
+      "step": 124100
+    },
+    {
+      "epoch": 4.183508488278092,
+      "grad_norm": 0.8221725821495056,
+      "learning_rate": 5.4435013024342045e-05,
+      "loss": 0.7923,
+      "step": 124200
+    },
+    {
+      "epoch": 4.186876852600378,
+      "grad_norm": 0.7301611304283142,
+      "learning_rate": 5.4210455402856373e-05,
+      "loss": 0.7946,
+      "step": 124300
+    },
+    {
+      "epoch": 4.190245216922662,
+      "grad_norm": 0.7508817315101624,
+      "learning_rate": 5.39858977813707e-05,
+      "loss": 0.7921,
+      "step": 124400
+    },
+    {
+      "epoch": 4.193613581244947,
+      "grad_norm": 0.7159464359283447,
+      "learning_rate": 5.376134015988502e-05,
+      "loss": 0.7916,
+      "step": 124500
+    },
+    {
+      "epoch": 4.196981945567233,
+      "grad_norm": 0.6572859883308411,
+      "learning_rate": 5.3536782538399345e-05,
+      "loss": 0.7932,
+      "step": 124600
+    },
+    {
+      "epoch": 4.200350309889518,
+      "grad_norm": 0.8845502138137817,
+      "learning_rate": 5.3312224916913674e-05,
+      "loss": 0.7914,
+      "step": 124700
+    },
+    {
+      "epoch": 4.203718674211803,
+      "grad_norm": 0.7034029364585876,
+      "learning_rate": 5.3087667295428e-05,
+      "loss": 0.7925,
+      "step": 124800
+    },
+    {
+      "epoch": 4.2070870385340875,
+      "grad_norm": 0.66123366355896,
+      "learning_rate": 5.2863109673942324e-05,
+      "loss": 0.7956,
+      "step": 124900
+    },
+    {
+      "epoch": 4.210455402856373,
+      "grad_norm": 0.7376932501792908,
+      "learning_rate": 5.263855205245665e-05,
+      "loss": 0.7891,
+      "step": 125000
+    },
+    {
+      "epoch": 4.213823767178658,
+      "grad_norm": 0.7154867649078369,
+      "learning_rate": 5.241399443097098e-05,
+      "loss": 0.7921,
+      "step": 125100
+    },
+    {
+      "epoch": 4.217192131500943,
+      "grad_norm": 0.6891977190971375,
+      "learning_rate": 5.218943680948531e-05,
+      "loss": 0.7938,
+      "step": 125200
+    },
+    {
+      "epoch": 4.220560495823229,
+      "grad_norm": 0.713810384273529,
+      "learning_rate": 5.196487918799964e-05,
+      "loss": 0.7926,
+      "step": 125300
+    },
+    {
+      "epoch": 4.223928860145513,
+      "grad_norm": 0.7049047350883484,
+      "learning_rate": 5.174032156651396e-05,
+      "loss": 0.7955,
+      "step": 125400
+    },
+    {
+      "epoch": 4.227297224467798,
+      "grad_norm": 0.7068188786506653,
+      "learning_rate": 5.151576394502829e-05,
+      "loss": 0.7966,
+      "step": 125500
+    },
+    {
+      "epoch": 4.2306655887900835,
+      "grad_norm": 0.7107961177825928,
+      "learning_rate": 5.1291206323542616e-05,
+      "loss": 0.7959,
+      "step": 125600
+    },
+    {
+      "epoch": 4.234033953112369,
+      "grad_norm": 0.7235453128814697,
+      "learning_rate": 5.1066648702056944e-05,
+      "loss": 0.7904,
+      "step": 125700
+    },
+    {
+      "epoch": 4.237402317434654,
+      "grad_norm": 0.7281498908996582,
+      "learning_rate": 5.084209108057127e-05,
+      "loss": 0.7916,
+      "step": 125800
+    },
+    {
+      "epoch": 4.240770681756938,
+      "grad_norm": 0.726445198059082,
+      "learning_rate": 5.0617533459085594e-05,
+      "loss": 0.7905,
+      "step": 125900
+    },
+    {
+      "epoch": 4.244139046079224,
+      "grad_norm": 0.7289313673973083,
+      "learning_rate": 5.039297583759992e-05,
+      "loss": 0.796,
+      "step": 126000
+    },
+    {
+      "epoch": 4.247507410401509,
+      "grad_norm": 0.6762425899505615,
+      "learning_rate": 5.016841821611425e-05,
+      "loss": 0.7898,
+      "step": 126100
+    },
+    {
+      "epoch": 4.250875774723794,
+      "grad_norm": 0.7091333866119385,
+      "learning_rate": 4.994386059462858e-05,
+      "loss": 0.7946,
+      "step": 126200
+    },
+    {
+      "epoch": 4.2542441390460795,
+      "grad_norm": 0.6724031567573547,
+      "learning_rate": 4.971930297314291e-05,
+      "loss": 0.7931,
+      "step": 126300
+    },
+    {
+      "epoch": 4.257612503368365,
+      "grad_norm": 0.7157333493232727,
+      "learning_rate": 4.949474535165723e-05,
+      "loss": 0.7925,
+      "step": 126400
+    },
+    {
+      "epoch": 4.260980867690649,
+      "grad_norm": 0.7710690498352051,
+      "learning_rate": 4.927018773017156e-05,
+      "loss": 0.795,
+      "step": 126500
+    },
+    {
+      "epoch": 4.264349232012934,
+      "grad_norm": 0.6310470104217529,
+      "learning_rate": 4.9045630108685886e-05,
+      "loss": 0.7931,
+      "step": 126600
+    },
+    {
+      "epoch": 4.26771759633522,
+      "grad_norm": 0.6831243634223938,
+      "learning_rate": 4.8821072487200215e-05,
+      "loss": 0.7926,
+      "step": 126700
+    },
+    {
+      "epoch": 4.271085960657505,
+      "grad_norm": 0.675401508808136,
+      "learning_rate": 4.859651486571454e-05,
+      "loss": 0.7902,
+      "step": 126800
+    },
+    {
+      "epoch": 4.27445432497979,
+      "grad_norm": 0.715557873249054,
+      "learning_rate": 4.8371957244228865e-05,
+      "loss": 0.7932,
+      "step": 126900
+    },
+    {
+      "epoch": 4.2778226893020745,
+      "grad_norm": 0.6567670702934265,
+      "learning_rate": 4.814739962274319e-05,
+      "loss": 0.7925,
+      "step": 127000
+    },
+    {
+      "epoch": 4.28119105362436,
+      "grad_norm": 0.6738788485527039,
+      "learning_rate": 4.792284200125752e-05,
+      "loss": 0.7932,
+      "step": 127100
+    },
+    {
+      "epoch": 4.284559417946645,
+      "grad_norm": 0.6749903559684753,
+      "learning_rate": 4.769828437977185e-05,
+      "loss": 0.792,
+      "step": 127200
+    },
+    {
+      "epoch": 4.28792778226893,
+      "grad_norm": 0.6749402284622192,
+      "learning_rate": 4.7473726758286165e-05,
+      "loss": 0.7924,
+      "step": 127300
+    },
+    {
+      "epoch": 4.291296146591216,
+      "grad_norm": 0.7213864326477051,
+      "learning_rate": 4.724916913680049e-05,
+      "loss": 0.7923,
+      "step": 127400
+    },
+    {
+      "epoch": 4.2946645109135,
+      "grad_norm": 0.6684932112693787,
+      "learning_rate": 4.702461151531482e-05,
+      "loss": 0.7906,
+      "step": 127500
+    },
+    {
+      "epoch": 4.298032875235785,
+      "grad_norm": 0.676459014415741,
+      "learning_rate": 4.680005389382916e-05,
+      "loss": 0.791,
+      "step": 127600
+    },
+    {
+      "epoch": 4.3014012395580705,
+      "grad_norm": 0.6472882032394409,
+      "learning_rate": 4.6575496272343485e-05,
+      "loss": 0.7893,
+      "step": 127700
+    },
+    {
+      "epoch": 4.304769603880356,
+      "grad_norm": 0.6770561337471008,
+      "learning_rate": 4.63509386508578e-05,
+      "loss": 0.7907,
+      "step": 127800
+    },
+    {
+      "epoch": 4.308137968202641,
+      "grad_norm": 0.6275637149810791,
+      "learning_rate": 4.612638102937213e-05,
+      "loss": 0.7913,
+      "step": 127900
+    },
+    {
+      "epoch": 4.311506332524926,
+      "grad_norm": 0.6540498733520508,
+      "learning_rate": 4.590182340788646e-05,
+      "loss": 0.793,
+      "step": 128000
+    },
+    {
+      "epoch": 4.314874696847211,
+      "grad_norm": 0.7203840017318726,
+      "learning_rate": 4.5677265786400785e-05,
+      "loss": 0.7913,
+      "step": 128100
+    },
+    {
+      "epoch": 4.318243061169496,
+      "grad_norm": 0.6884647607803345,
+      "learning_rate": 4.5452708164915114e-05,
+      "loss": 0.7928,
+      "step": 128200
+    },
+    {
+      "epoch": 4.321611425491781,
+      "grad_norm": 0.8688404560089111,
+      "learning_rate": 4.5228150543429435e-05,
+      "loss": 0.791,
+      "step": 128300
+    },
+    {
+      "epoch": 4.3249797898140665,
+      "grad_norm": 0.6714907884597778,
+      "learning_rate": 4.5003592921943764e-05,
+      "loss": 0.7909,
+      "step": 128400
+    },
+    {
+      "epoch": 4.328348154136352,
+      "grad_norm": 0.6698452830314636,
+      "learning_rate": 4.477903530045809e-05,
+      "loss": 0.7934,
+      "step": 128500
+    },
+    {
+      "epoch": 4.331716518458636,
+      "grad_norm": 0.6693661212921143,
+      "learning_rate": 4.455447767897242e-05,
+      "loss": 0.7917,
+      "step": 128600
+    },
+    {
+      "epoch": 4.335084882780921,
+      "grad_norm": 0.6978726983070374,
+      "learning_rate": 4.432992005748675e-05,
+      "loss": 0.7903,
+      "step": 128700
+    },
+    {
+      "epoch": 4.338453247103207,
+      "grad_norm": 0.7052512168884277,
+      "learning_rate": 4.410536243600107e-05,
+      "loss": 0.7909,
+      "step": 128800
+    },
+    {
+      "epoch": 4.341821611425492,
+      "grad_norm": 0.6648239493370056,
+      "learning_rate": 4.38808048145154e-05,
+      "loss": 0.793,
+      "step": 128900
+    },
+    {
+      "epoch": 4.345189975747777,
+      "grad_norm": 0.6616061329841614,
+      "learning_rate": 4.365624719302973e-05,
+      "loss": 0.7915,
+      "step": 129000
+    },
+    {
+      "epoch": 4.348558340070062,
+      "grad_norm": 0.7084757089614868,
+      "learning_rate": 4.3431689571544056e-05,
+      "loss": 0.7918,
+      "step": 129100
+    },
+    {
+      "epoch": 4.351926704392347,
+      "grad_norm": 0.6875206828117371,
+      "learning_rate": 4.3207131950058384e-05,
+      "loss": 0.7893,
+      "step": 129200
+    },
+    {
+      "epoch": 4.355295068714632,
+      "grad_norm": 0.6854655742645264,
+      "learning_rate": 4.2982574328572706e-05,
+      "loss": 0.7859,
+      "step": 129300
+    },
+    {
+      "epoch": 4.358663433036917,
+      "grad_norm": 0.7505910992622375,
+      "learning_rate": 4.2758016707087034e-05,
+      "loss": 0.7934,
+      "step": 129400
+    },
+    {
+      "epoch": 4.362031797359203,
+      "grad_norm": 0.6831994652748108,
+      "learning_rate": 4.253345908560136e-05,
+      "loss": 0.7941,
+      "step": 129500
+    },
+    {
+      "epoch": 4.365400161681487,
+      "grad_norm": 0.7224695086479187,
+      "learning_rate": 4.230890146411569e-05,
+      "loss": 0.7907,
+      "step": 129600
+    },
+    {
+      "epoch": 4.368768526003772,
+      "grad_norm": 0.7102277874946594,
+      "learning_rate": 4.208434384263001e-05,
+      "loss": 0.7926,
+      "step": 129700
+    },
+    {
+      "epoch": 4.372136890326058,
+      "grad_norm": 0.6760639548301697,
+      "learning_rate": 4.185978622114434e-05,
+      "loss": 0.7949,
+      "step": 129800
+    },
+    {
+      "epoch": 4.375505254648343,
+      "grad_norm": 0.6891269087791443,
+      "learning_rate": 4.163522859965867e-05,
+      "loss": 0.7901,
+      "step": 129900
+    },
+    {
+      "epoch": 4.378873618970628,
+      "grad_norm": 0.6708014607429504,
+      "learning_rate": 4.1410670978173e-05,
+      "loss": 0.7894,
+      "step": 130000
+    },
+    {
+      "epoch": 4.382241983292913,
+      "grad_norm": 0.7219834327697754,
+      "learning_rate": 4.1186113356687327e-05,
+      "loss": 0.7917,
+      "step": 130100
+    },
+    {
+      "epoch": 4.385610347615198,
+      "grad_norm": 0.6797767281532288,
+      "learning_rate": 4.096155573520165e-05,
+      "loss": 0.79,
+      "step": 130200
+    },
+    {
+      "epoch": 4.388978711937483,
+      "grad_norm": 1.1183526515960693,
+      "learning_rate": 4.073699811371598e-05,
+      "loss": 0.7903,
+      "step": 130300
+    },
+    {
+      "epoch": 4.392347076259768,
+      "grad_norm": 0.6638786792755127,
+      "learning_rate": 4.0512440492230305e-05,
+      "loss": 0.7929,
+      "step": 130400
+    },
+    {
+      "epoch": 4.3957154405820535,
+      "grad_norm": 0.6631746888160706,
+      "learning_rate": 4.0287882870744633e-05,
+      "loss": 0.7895,
+      "step": 130500
+    },
+    {
+      "epoch": 4.399083804904339,
+      "grad_norm": 0.6432344317436218,
+      "learning_rate": 4.006332524925896e-05,
+      "loss": 0.793,
+      "step": 130600
+    },
+    {
+      "epoch": 4.402452169226623,
+      "grad_norm": 0.6888932585716248,
+      "learning_rate": 3.983876762777328e-05,
+      "loss": 0.7863,
+      "step": 130700
+    },
+    {
+      "epoch": 4.4058205335489085,
+      "grad_norm": 0.6781389713287354,
+      "learning_rate": 3.9614210006287605e-05,
+      "loss": 0.7894,
+      "step": 130800
+    },
+    {
+      "epoch": 4.409188897871194,
+      "grad_norm": 0.6817391514778137,
+      "learning_rate": 3.9389652384801934e-05,
+      "loss": 0.7915,
+      "step": 130900
+    },
+    {
+      "epoch": 4.412557262193479,
+      "grad_norm": 0.7040595412254333,
+      "learning_rate": 3.916509476331627e-05,
+      "loss": 0.7898,
+      "step": 131000
+    },
+    {
+      "epoch": 4.415925626515764,
+      "grad_norm": 0.6829173564910889,
+      "learning_rate": 3.89405371418306e-05,
+      "loss": 0.7911,
+      "step": 131100
+    },
+    {
+      "epoch": 4.419293990838049,
+      "grad_norm": 0.7113239765167236,
+      "learning_rate": 3.871597952034491e-05,
+      "loss": 0.7871,
+      "step": 131200
+    },
+    {
+      "epoch": 4.422662355160334,
+      "grad_norm": 0.6862262487411499,
+      "learning_rate": 3.849142189885924e-05,
+      "loss": 0.7915,
+      "step": 131300
+    },
+    {
+      "epoch": 4.426030719482619,
+      "grad_norm": 0.6750577092170715,
+      "learning_rate": 3.826686427737357e-05,
+      "loss": 0.7925,
+      "step": 131400
+    },
+    {
+      "epoch": 4.429399083804904,
+      "grad_norm": 0.7142935395240784,
+      "learning_rate": 3.80423066558879e-05,
+      "loss": 0.7906,
+      "step": 131500
+    },
+    {
+      "epoch": 4.43276744812719,
+      "grad_norm": 0.7272329330444336,
+      "learning_rate": 3.7817749034402226e-05,
+      "loss": 0.7898,
+      "step": 131600
+    },
+    {
+      "epoch": 4.436135812449475,
+      "grad_norm": 0.7627023458480835,
+      "learning_rate": 3.759319141291655e-05,
+      "loss": 0.7907,
+      "step": 131700
+    },
+    {
+      "epoch": 4.439504176771759,
+      "grad_norm": 0.6552876830101013,
+      "learning_rate": 3.7368633791430876e-05,
+      "loss": 0.7901,
+      "step": 131800
+    },
+    {
+      "epoch": 4.442872541094045,
+      "grad_norm": 0.7480065822601318,
+      "learning_rate": 3.7144076169945204e-05,
+      "loss": 0.7917,
+      "step": 131900
+    },
+    {
+      "epoch": 4.44624090541633,
+      "grad_norm": 0.7101287841796875,
+      "learning_rate": 3.691951854845953e-05,
+      "loss": 0.7909,
+      "step": 132000
+    },
+    {
+      "epoch": 4.449609269738615,
+      "grad_norm": 0.7063937187194824,
+      "learning_rate": 3.669496092697386e-05,
+      "loss": 0.7897,
+      "step": 132100
+    },
+    {
+      "epoch": 4.4529776340609,
+      "grad_norm": 0.7050164341926575,
+      "learning_rate": 3.647040330548819e-05,
+      "loss": 0.7924,
+      "step": 132200
+    },
+    {
+      "epoch": 4.456345998383185,
+      "grad_norm": 0.7028830051422119,
+      "learning_rate": 3.624584568400251e-05,
+      "loss": 0.7894,
+      "step": 132300
+    },
+    {
+      "epoch": 4.45971436270547,
+      "grad_norm": 0.7394464015960693,
+      "learning_rate": 3.602128806251684e-05,
+      "loss": 0.7916,
+      "step": 132400
+    },
+    {
+      "epoch": 4.463082727027755,
+      "grad_norm": 0.6921053528785706,
+      "learning_rate": 3.579673044103116e-05,
+      "loss": 0.7888,
+      "step": 132500
+    },
+    {
+      "epoch": 4.466451091350041,
+      "grad_norm": 0.6709455847740173,
+      "learning_rate": 3.557217281954549e-05,
+      "loss": 0.7915,
+      "step": 132600
+    },
+    {
+      "epoch": 4.469819455672326,
+      "grad_norm": 0.7549142241477966,
+      "learning_rate": 3.534761519805982e-05,
+      "loss": 0.7925,
+      "step": 132700
+    },
+    {
+      "epoch": 4.47318781999461,
+      "grad_norm": 0.716698944568634,
+      "learning_rate": 3.5123057576574146e-05,
+      "loss": 0.7893,
+      "step": 132800
+    },
+    {
+      "epoch": 4.4765561843168955,
+      "grad_norm": 0.6615895628929138,
+      "learning_rate": 3.4898499955088475e-05,
+      "loss": 0.7864,
+      "step": 132900
+    },
+    {
+      "epoch": 4.479924548639181,
+      "grad_norm": 0.8180909156799316,
+      "learning_rate": 3.4673942333602796e-05,
+      "loss": 0.7902,
+      "step": 133000
+    },
+    {
+      "epoch": 4.483292912961466,
+      "grad_norm": 0.7339180111885071,
+      "learning_rate": 3.4449384712117125e-05,
+      "loss": 0.789,
+      "step": 133100
+    },
+    {
+      "epoch": 4.486661277283751,
+      "grad_norm": 0.921172559261322,
+      "learning_rate": 3.422482709063145e-05,
+      "loss": 0.7876,
+      "step": 133200
+    },
+    {
+      "epoch": 4.490029641606036,
+      "grad_norm": 0.7154456973075867,
+      "learning_rate": 3.400026946914578e-05,
+      "loss": 0.7911,
+      "step": 133300
+    },
+    {
+      "epoch": 4.493398005928321,
+      "grad_norm": 0.9519571661949158,
+      "learning_rate": 3.377571184766011e-05,
+      "loss": 0.7895,
+      "step": 133400
+    },
+    {
+      "epoch": 4.496766370250606,
+      "grad_norm": 0.7096812725067139,
+      "learning_rate": 3.355115422617443e-05,
+      "loss": 0.7892,
+      "step": 133500
+    },
+    {
+      "epoch": 4.5001347345728915,
+      "grad_norm": 0.6969826817512512,
+      "learning_rate": 3.332659660468876e-05,
+      "loss": 0.7913,
+      "step": 133600
+    },
+    {
+      "epoch": 4.503503098895177,
+      "grad_norm": 0.6900309920310974,
+      "learning_rate": 3.310203898320309e-05,
+      "loss": 0.7909,
+      "step": 133700
+    },
+    {
+      "epoch": 4.506871463217461,
+      "grad_norm": 0.7131757736206055,
+      "learning_rate": 3.287748136171742e-05,
+      "loss": 0.7932,
+      "step": 133800
+    },
+    {
+      "epoch": 4.510239827539746,
+      "grad_norm": 0.6708106398582458,
+      "learning_rate": 3.265292374023174e-05,
+      "loss": 0.7889,
+      "step": 133900
+    },
+    {
+      "epoch": 4.513608191862032,
+      "grad_norm": 0.7623175382614136,
+      "learning_rate": 3.242836611874607e-05,
+      "loss": 0.7906,
+      "step": 134000
+    },
+    {
+      "epoch": 4.516976556184317,
+      "grad_norm": 0.7173711657524109,
+      "learning_rate": 3.2203808497260395e-05,
+      "loss": 0.7869,
+      "step": 134100
+    },
+    {
+      "epoch": 4.520344920506602,
+      "grad_norm": 0.7341943979263306,
+      "learning_rate": 3.197925087577472e-05,
+      "loss": 0.7897,
+      "step": 134200
+    },
+    {
+      "epoch": 4.5237132848288875,
+      "grad_norm": 0.7385048866271973,
+      "learning_rate": 3.1754693254289045e-05,
+      "loss": 0.7884,
+      "step": 134300
+    },
+    {
+      "epoch": 4.527081649151172,
+      "grad_norm": 0.7369481921195984,
+      "learning_rate": 3.1530135632803374e-05,
+      "loss": 0.7922,
+      "step": 134400
+    },
+    {
+      "epoch": 4.530450013473457,
+      "grad_norm": 0.7261357307434082,
+      "learning_rate": 3.13055780113177e-05,
+      "loss": 0.7896,
+      "step": 134500
+    },
+    {
+      "epoch": 4.533818377795742,
+      "grad_norm": 0.7368954420089722,
+      "learning_rate": 3.108102038983203e-05,
+      "loss": 0.7877,
+      "step": 134600
+    },
+    {
+      "epoch": 4.537186742118028,
+      "grad_norm": 0.6391497850418091,
+      "learning_rate": 3.085646276834635e-05,
+      "loss": 0.7883,
+      "step": 134700
+    },
+    {
+      "epoch": 4.540555106440313,
+      "grad_norm": 0.6790878176689148,
+      "learning_rate": 3.063190514686068e-05,
+      "loss": 0.7878,
+      "step": 134800
+    },
+    {
+      "epoch": 4.543923470762598,
+      "grad_norm": 0.6800740361213684,
+      "learning_rate": 3.0407347525375006e-05,
+      "loss": 0.7879,
+      "step": 134900
+    },
+    {
+      "epoch": 4.5472918350848825,
+      "grad_norm": 0.6817995309829712,
+      "learning_rate": 3.0182789903889334e-05,
+      "loss": 0.7893,
+      "step": 135000
+    },
+    {
+      "epoch": 4.550660199407168,
+      "grad_norm": 0.706876814365387,
+      "learning_rate": 2.995823228240366e-05,
+      "loss": 0.7877,
+      "step": 135100
+    },
+    {
+      "epoch": 4.554028563729453,
+      "grad_norm": 0.6842048168182373,
+      "learning_rate": 2.9733674660917988e-05,
+      "loss": 0.7887,
+      "step": 135200
+    },
+    {
+      "epoch": 4.557396928051738,
+      "grad_norm": 0.8431084752082825,
+      "learning_rate": 2.9509117039432316e-05,
+      "loss": 0.7895,
+      "step": 135300
+    },
+    {
+      "epoch": 4.560765292374024,
+      "grad_norm": 0.6965360045433044,
+      "learning_rate": 2.928455941794664e-05,
+      "loss": 0.7857,
+      "step": 135400
+    },
+    {
+      "epoch": 4.564133656696308,
+      "grad_norm": 0.675323486328125,
+      "learning_rate": 2.906000179646097e-05,
+      "loss": 0.7917,
+      "step": 135500
+    },
+    {
+      "epoch": 4.567502021018593,
+      "grad_norm": 0.6971179842948914,
+      "learning_rate": 2.8835444174975295e-05,
+      "loss": 0.791,
+      "step": 135600
+    },
+    {
+      "epoch": 4.5708703853408785,
+      "grad_norm": 0.7096483111381531,
+      "learning_rate": 2.8610886553489623e-05,
+      "loss": 0.7886,
+      "step": 135700
+    },
+    {
+      "epoch": 4.574238749663164,
+      "grad_norm": 0.6643834710121155,
+      "learning_rate": 2.838632893200395e-05,
+      "loss": 0.7903,
+      "step": 135800
+    },
+    {
+      "epoch": 4.577607113985449,
+      "grad_norm": 0.6781139373779297,
+      "learning_rate": 2.8161771310518276e-05,
+      "loss": 0.7861,
+      "step": 135900
+    },
+    {
+      "epoch": 4.580975478307733,
+      "grad_norm": 0.69442218542099,
+      "learning_rate": 2.7937213689032605e-05,
+      "loss": 0.7884,
+      "step": 136000
+    },
+    {
+      "epoch": 4.584343842630019,
+      "grad_norm": 0.6849011778831482,
+      "learning_rate": 2.771265606754693e-05,
+      "loss": 0.7892,
+      "step": 136100
+    },
+    {
+      "epoch": 4.587712206952304,
+      "grad_norm": 0.7133069634437561,
+      "learning_rate": 2.7488098446061258e-05,
+      "loss": 0.7904,
+      "step": 136200
+    },
+    {
+      "epoch": 4.591080571274589,
+      "grad_norm": 0.7414596080780029,
+      "learning_rate": 2.7263540824575583e-05,
+      "loss": 0.7933,
+      "step": 136300
+    },
+    {
+      "epoch": 4.5944489355968745,
+      "grad_norm": 0.7594481706619263,
+      "learning_rate": 2.703898320308991e-05,
+      "loss": 0.7901,
+      "step": 136400
+    },
+    {
+      "epoch": 4.597817299919159,
+      "grad_norm": 0.7436200976371765,
+      "learning_rate": 2.681442558160424e-05,
+      "loss": 0.7863,
+      "step": 136500
+    },
+    {
+      "epoch": 4.601185664241444,
+      "grad_norm": 0.7052866220474243,
+      "learning_rate": 2.6589867960118562e-05,
+      "loss": 0.7899,
+      "step": 136600
+    },
+    {
+      "epoch": 4.604554028563729,
+      "grad_norm": 0.720867395401001,
+      "learning_rate": 2.636531033863289e-05,
+      "loss": 0.786,
+      "step": 136700
+    },
+    {
+      "epoch": 4.607922392886015,
+      "grad_norm": 0.7140068411827087,
+      "learning_rate": 2.6140752717147215e-05,
+      "loss": 0.7886,
+      "step": 136800
+    },
+    {
+      "epoch": 4.6112907572083,
+      "grad_norm": 0.6707795858383179,
+      "learning_rate": 2.5916195095661544e-05,
+      "loss": 0.792,
+      "step": 136900
+    },
+    {
+      "epoch": 4.614659121530584,
+      "grad_norm": 0.6954900622367859,
+      "learning_rate": 2.5691637474175872e-05,
+      "loss": 0.785,
+      "step": 137000
+    },
+    {
+      "epoch": 4.61802748585287,
+      "grad_norm": 0.695578396320343,
+      "learning_rate": 2.5467079852690197e-05,
+      "loss": 0.7897,
+      "step": 137100
+    },
+    {
+      "epoch": 4.621395850175155,
+      "grad_norm": 0.8920716047286987,
+      "learning_rate": 2.5242522231204525e-05,
+      "loss": 0.7912,
+      "step": 137200
+    },
+    {
+      "epoch": 4.62476421449744,
+      "grad_norm": 0.6703862547874451,
+      "learning_rate": 2.501796460971885e-05,
+      "loss": 0.79,
+      "step": 137300
+    },
+    {
+      "epoch": 4.628132578819725,
+      "grad_norm": 0.7049610018730164,
+      "learning_rate": 2.479340698823318e-05,
+      "loss": 0.7876,
+      "step": 137400
+    },
+    {
+      "epoch": 4.63150094314201,
+      "grad_norm": 0.7894124388694763,
+      "learning_rate": 2.4568849366747504e-05,
+      "loss": 0.7897,
+      "step": 137500
+    },
+    {
+      "epoch": 4.634869307464295,
+      "grad_norm": 2.398905038833618,
+      "learning_rate": 2.4344291745261832e-05,
+      "loss": 0.7909,
+      "step": 137600
+    },
+    {
+      "epoch": 4.63823767178658,
+      "grad_norm": 0.6975995302200317,
+      "learning_rate": 2.411973412377616e-05,
+      "loss": 0.789,
+      "step": 137700
+    },
+    {
+      "epoch": 4.641606036108866,
+      "grad_norm": 0.717632532119751,
+      "learning_rate": 2.3895176502290486e-05,
+      "loss": 0.7883,
+      "step": 137800
+    },
+    {
+      "epoch": 4.644974400431151,
+      "grad_norm": 0.7279338240623474,
+      "learning_rate": 2.3670618880804814e-05,
+      "loss": 0.787,
+      "step": 137900
+    },
+    {
+      "epoch": 4.648342764753436,
+      "grad_norm": 0.6684018969535828,
+      "learning_rate": 2.344606125931914e-05,
+      "loss": 0.7875,
+      "step": 138000
+    },
+    {
+      "epoch": 4.6517111290757205,
+      "grad_norm": 0.7201947569847107,
+      "learning_rate": 2.3221503637833468e-05,
+      "loss": 0.791,
+      "step": 138100
+    },
+    {
+      "epoch": 4.655079493398006,
+      "grad_norm": 0.71978360414505,
+      "learning_rate": 2.2996946016347796e-05,
+      "loss": 0.7871,
+      "step": 138200
+    },
+    {
+      "epoch": 4.658447857720291,
+      "grad_norm": 0.6821112632751465,
+      "learning_rate": 2.2772388394862118e-05,
+      "loss": 0.7888,
+      "step": 138300
+    },
+    {
+      "epoch": 4.661816222042576,
+      "grad_norm": 0.7123835682868958,
+      "learning_rate": 2.2547830773376446e-05,
+      "loss": 0.7879,
+      "step": 138400
+    },
+    {
+      "epoch": 4.6651845863648616,
+      "grad_norm": 0.6627094745635986,
+      "learning_rate": 2.232327315189077e-05,
+      "loss": 0.7901,
+      "step": 138500
+    },
+    {
+      "epoch": 4.668552950687147,
+      "grad_norm": 0.6200422644615173,
+      "learning_rate": 2.20987155304051e-05,
+      "loss": 0.791,
+      "step": 138600
+    },
+    {
+      "epoch": 4.671921315009431,
+      "grad_norm": 0.7372182011604309,
+      "learning_rate": 2.1874157908919425e-05,
+      "loss": 0.7846,
+      "step": 138700
+    },
+    {
+      "epoch": 4.6752896793317165,
+      "grad_norm": 0.7178613543510437,
+      "learning_rate": 2.1649600287433753e-05,
+      "loss": 0.7903,
+      "step": 138800
+    },
+    {
+      "epoch": 4.678658043654002,
+      "grad_norm": 0.7218438386917114,
+      "learning_rate": 2.142504266594808e-05,
+      "loss": 0.7881,
+      "step": 138900
+    },
+    {
+      "epoch": 4.682026407976287,
+      "grad_norm": 0.7254891395568848,
+      "learning_rate": 2.1200485044462406e-05,
+      "loss": 0.7882,
+      "step": 139000
+    },
+    {
+      "epoch": 4.685394772298572,
+      "grad_norm": 0.6931572556495667,
+      "learning_rate": 2.0975927422976735e-05,
+      "loss": 0.7878,
+      "step": 139100
+    },
+    {
+      "epoch": 4.688763136620857,
+      "grad_norm": 0.6938092708587646,
+      "learning_rate": 2.075136980149106e-05,
+      "loss": 0.7859,
+      "step": 139200
+    },
+    {
+      "epoch": 4.692131500943142,
+      "grad_norm": 0.7220520973205566,
+      "learning_rate": 2.0526812180005388e-05,
+      "loss": 0.7874,
+      "step": 139300
+    },
+    {
+      "epoch": 4.695499865265427,
+      "grad_norm": 0.7422808408737183,
+      "learning_rate": 2.0302254558519717e-05,
+      "loss": 0.7861,
+      "step": 139400
+    },
+    {
+      "epoch": 4.698868229587712,
+      "grad_norm": 0.6828125715255737,
+      "learning_rate": 2.007769693703404e-05,
+      "loss": 0.7864,
+      "step": 139500
+    },
+    {
+      "epoch": 4.702236593909998,
+      "grad_norm": 0.7483955025672913,
+      "learning_rate": 1.985313931554837e-05,
+      "loss": 0.7847,
+      "step": 139600
+    },
+    {
+      "epoch": 4.705604958232282,
+      "grad_norm": 0.7267476916313171,
+      "learning_rate": 1.9628581694062695e-05,
+      "loss": 0.788,
+      "step": 139700
+    },
+    {
+      "epoch": 4.708973322554567,
+      "grad_norm": 0.6834740042686462,
+      "learning_rate": 1.9404024072577024e-05,
+      "loss": 0.7845,
+      "step": 139800
+    },
+    {
+      "epoch": 4.712341686876853,
+      "grad_norm": 0.7835246324539185,
+      "learning_rate": 1.9179466451091345e-05,
+      "loss": 0.7856,
+      "step": 139900
+    },
+    {
+      "epoch": 4.715710051199138,
+      "grad_norm": 0.6620598435401917,
+      "learning_rate": 1.8954908829605674e-05,
+      "loss": 0.7915,
+      "step": 140000
+    },
+    {
+      "epoch": 4.719078415521423,
+      "grad_norm": 0.7632565498352051,
+      "learning_rate": 1.8730351208120002e-05,
+      "loss": 0.7878,
+      "step": 140100
+    },
+    {
+      "epoch": 4.7224467798437075,
+      "grad_norm": 0.6834028363227844,
+      "learning_rate": 1.8505793586634327e-05,
+      "loss": 0.7852,
+      "step": 140200
+    },
+    {
+      "epoch": 4.725815144165993,
+      "grad_norm": 0.8396201729774475,
+      "learning_rate": 1.8281235965148655e-05,
+      "loss": 0.7893,
+      "step": 140300
+    },
+    {
+      "epoch": 4.729183508488278,
+      "grad_norm": 0.7341758608818054,
+      "learning_rate": 1.8056678343662984e-05,
+      "loss": 0.7892,
+      "step": 140400
+    },
+    {
+      "epoch": 4.732551872810563,
+      "grad_norm": 0.74070143699646,
+      "learning_rate": 1.783212072217731e-05,
+      "loss": 0.7864,
+      "step": 140500
+    },
+    {
+      "epoch": 4.735920237132849,
+      "grad_norm": 0.7226638793945312,
+      "learning_rate": 1.7607563100691637e-05,
+      "loss": 0.7861,
+      "step": 140600
+    },
+    {
+      "epoch": 4.739288601455133,
+      "grad_norm": 0.6856757402420044,
+      "learning_rate": 1.7383005479205962e-05,
+      "loss": 0.7878,
+      "step": 140700
+    },
+    {
+      "epoch": 4.742656965777418,
+      "grad_norm": 0.695364773273468,
+      "learning_rate": 1.7158447857720287e-05,
+      "loss": 0.788,
+      "step": 140800
+    },
+    {
+      "epoch": 4.7460253300997035,
+      "grad_norm": 0.7357635498046875,
+      "learning_rate": 1.6933890236234616e-05,
+      "loss": 0.7853,
+      "step": 140900
+    },
+    {
+      "epoch": 4.749393694421989,
+      "grad_norm": 0.6697304248809814,
+      "learning_rate": 1.6709332614748944e-05,
+      "loss": 0.7868,
+      "step": 141000
+    },
+    {
+      "epoch": 4.752762058744274,
+      "grad_norm": 0.719422459602356,
+      "learning_rate": 1.648477499326327e-05,
+      "loss": 0.7887,
+      "step": 141100
+    },
+    {
+      "epoch": 4.756130423066558,
+      "grad_norm": 0.650025486946106,
+      "learning_rate": 1.6260217371777598e-05,
+      "loss": 0.7901,
+      "step": 141200
+    },
+    {
+      "epoch": 4.759498787388844,
+      "grad_norm": 0.6734464168548584,
+      "learning_rate": 1.6035659750291923e-05,
+      "loss": 0.7885,
+      "step": 141300
+    },
+    {
+      "epoch": 4.762867151711129,
+      "grad_norm": 0.6939591765403748,
+      "learning_rate": 1.581110212880625e-05,
+      "loss": 0.7896,
+      "step": 141400
+    },
+    {
+      "epoch": 4.766235516033414,
+      "grad_norm": 0.6683725118637085,
+      "learning_rate": 1.558654450732058e-05,
+      "loss": 0.7909,
+      "step": 141500
+    },
+    {
+      "epoch": 4.7696038803556995,
+      "grad_norm": 0.7492696642875671,
+      "learning_rate": 1.5361986885834905e-05,
+      "loss": 0.7856,
+      "step": 141600
+    },
+    {
+      "epoch": 4.772972244677985,
+      "grad_norm": 0.6370770931243896,
+      "learning_rate": 1.5137429264349231e-05,
+      "loss": 0.7858,
+      "step": 141700
+    },
+    {
+      "epoch": 4.776340609000269,
+      "grad_norm": 0.7032959461212158,
+      "learning_rate": 1.4912871642863558e-05,
+      "loss": 0.7833,
+      "step": 141800
+    },
+    {
+      "epoch": 4.779708973322554,
+      "grad_norm": 0.6765040159225464,
+      "learning_rate": 1.4688314021377885e-05,
+      "loss": 0.7861,
+      "step": 141900
+    },
+    {
+      "epoch": 4.78307733764484,
+      "grad_norm": 0.6935933232307434,
+      "learning_rate": 1.446375639989221e-05,
+      "loss": 0.7834,
+      "step": 142000
+    },
+    {
+      "epoch": 4.786445701967125,
+      "grad_norm": 0.716052770614624,
+      "learning_rate": 1.4239198778406538e-05,
+      "loss": 0.7857,
+      "step": 142100
+    },
+    {
+      "epoch": 4.78981406628941,
+      "grad_norm": 0.7378547787666321,
+      "learning_rate": 1.4014641156920865e-05,
+      "loss": 0.7849,
+      "step": 142200
+    },
+    {
+      "epoch": 4.793182430611695,
+      "grad_norm": 0.7271299958229065,
+      "learning_rate": 1.3790083535435192e-05,
+      "loss": 0.7877,
+      "step": 142300
+    },
+    {
+      "epoch": 4.79655079493398,
+      "grad_norm": 0.6609006524085999,
+      "learning_rate": 1.3565525913949518e-05,
+      "loss": 0.7857,
+      "step": 142400
+    },
+    {
+      "epoch": 4.799919159256265,
+      "grad_norm": 0.6987965703010559,
+      "learning_rate": 1.3340968292463845e-05,
+      "loss": 0.7848,
+      "step": 142500
+    },
+    {
+      "epoch": 4.80328752357855,
+      "grad_norm": 0.6643743515014648,
+      "learning_rate": 1.3116410670978172e-05,
+      "loss": 0.7872,
+      "step": 142600
+    },
+    {
+      "epoch": 4.806655887900836,
+      "grad_norm": 0.642590343952179,
+      "learning_rate": 1.28918530494925e-05,
+      "loss": 0.7858,
+      "step": 142700
+    },
+    {
+      "epoch": 4.810024252223121,
+      "grad_norm": 0.6913971304893494,
+      "learning_rate": 1.2667295428006827e-05,
+      "loss": 0.7862,
+      "step": 142800
+    },
+    {
+      "epoch": 4.813392616545405,
+      "grad_norm": 0.7790234684944153,
+      "learning_rate": 1.2442737806521152e-05,
+      "loss": 0.7869,
+      "step": 142900
+    },
+    {
+      "epoch": 4.8167609808676906,
+      "grad_norm": 0.6860402226448059,
+      "learning_rate": 1.2218180185035479e-05,
+      "loss": 0.7872,
+      "step": 143000
+    },
+    {
+      "epoch": 4.820129345189976,
+      "grad_norm": 0.73018479347229,
+      "learning_rate": 1.1993622563549805e-05,
+      "loss": 0.785,
+      "step": 143100
+    },
+    {
+      "epoch": 4.823497709512261,
+      "grad_norm": 0.7606022357940674,
+      "learning_rate": 1.1769064942064132e-05,
+      "loss": 0.7835,
+      "step": 143200
+    },
+    {
+      "epoch": 4.826866073834546,
+      "grad_norm": 0.7062585949897766,
+      "learning_rate": 1.154450732057846e-05,
+      "loss": 0.7863,
+      "step": 143300
+    },
+    {
+      "epoch": 4.830234438156831,
+      "grad_norm": 0.7451142072677612,
+      "learning_rate": 1.1319949699092787e-05,
+      "loss": 0.7853,
+      "step": 143400
+    },
+    {
+      "epoch": 4.833602802479116,
+      "grad_norm": 0.6708253622055054,
+      "learning_rate": 1.1095392077607114e-05,
+      "loss": 0.7872,
+      "step": 143500
+    },
+    {
+      "epoch": 4.836971166801401,
+      "grad_norm": 0.6949145197868347,
+      "learning_rate": 1.0870834456121439e-05,
+      "loss": 0.7877,
+      "step": 143600
+    },
+    {
+      "epoch": 4.8403395311236865,
+      "grad_norm": 0.7620200514793396,
+      "learning_rate": 1.0646276834635766e-05,
+      "loss": 0.784,
+      "step": 143700
+    },
+    {
+      "epoch": 4.843707895445972,
+      "grad_norm": 0.6912148594856262,
+      "learning_rate": 1.0421719213150092e-05,
+      "loss": 0.7838,
+      "step": 143800
+    },
+    {
+      "epoch": 4.847076259768256,
+      "grad_norm": 0.6649619936943054,
+      "learning_rate": 1.019716159166442e-05,
+      "loss": 0.7847,
+      "step": 143900
+    },
+    {
+      "epoch": 4.8504446240905414,
+      "grad_norm": 0.6840892434120178,
+      "learning_rate": 9.972603970178747e-06,
+      "loss": 0.7842,
+      "step": 144000
+    },
+    {
+      "epoch": 4.853812988412827,
+      "grad_norm": 0.6763936877250671,
+      "learning_rate": 9.748046348693074e-06,
+      "loss": 0.7884,
+      "step": 144100
+    },
+    {
+      "epoch": 4.857181352735112,
+      "grad_norm": 0.7060201168060303,
+      "learning_rate": 9.523488727207401e-06,
+      "loss": 0.7869,
+      "step": 144200
+    },
+    {
+      "epoch": 4.860549717057397,
+      "grad_norm": 0.6832892298698425,
+      "learning_rate": 9.298931105721728e-06,
+      "loss": 0.7846,
+      "step": 144300
+    },
+    {
+      "epoch": 4.863918081379682,
+      "grad_norm": 0.6947805285453796,
+      "learning_rate": 9.074373484236054e-06,
+      "loss": 0.7876,
+      "step": 144400
+    },
+    {
+      "epoch": 4.867286445701967,
+      "grad_norm": 0.6749753355979919,
+      "learning_rate": 8.849815862750381e-06,
+      "loss": 0.7848,
+      "step": 144500
+    },
+    {
+      "epoch": 4.870654810024252,
+      "grad_norm": 0.7636469602584839,
+      "learning_rate": 8.625258241264708e-06,
+      "loss": 0.7846,
+      "step": 144600
+    },
+    {
+      "epoch": 4.874023174346537,
+      "grad_norm": 0.6952735781669617,
+      "learning_rate": 8.400700619779035e-06,
+      "loss": 0.7889,
+      "step": 144700
+    },
+    {
+      "epoch": 4.877391538668823,
+      "grad_norm": 0.6861172318458557,
+      "learning_rate": 8.176142998293361e-06,
+      "loss": 0.787,
+      "step": 144800
+    },
+    {
+      "epoch": 4.880759902991107,
+      "grad_norm": 0.7696015238761902,
+      "learning_rate": 7.951585376807688e-06,
+      "loss": 0.7858,
+      "step": 144900
+    },
+    {
+      "epoch": 4.884128267313392,
+      "grad_norm": 0.7360905408859253,
+      "learning_rate": 7.727027755322015e-06,
+      "loss": 0.7875,
+      "step": 145000
+    },
+    {
+      "epoch": 4.887496631635678,
+      "grad_norm": 0.7402172088623047,
+      "learning_rate": 7.502470133836341e-06,
+      "loss": 0.7901,
+      "step": 145100
+    },
+    {
+      "epoch": 4.890864995957963,
+      "grad_norm": 0.7550304532051086,
+      "learning_rate": 7.277912512350668e-06,
+      "loss": 0.7881,
+      "step": 145200
+    },
+    {
+      "epoch": 4.894233360280248,
+      "grad_norm": 0.7490784525871277,
+      "learning_rate": 7.053354890864996e-06,
+      "loss": 0.7866,
+      "step": 145300
+    },
+    {
+      "epoch": 4.8976017246025325,
+      "grad_norm": 0.675995945930481,
+      "learning_rate": 6.828797269379322e-06,
+      "loss": 0.7868,
+      "step": 145400
+    },
+    {
+      "epoch": 4.900970088924818,
+      "grad_norm": 0.6862391829490662,
+      "learning_rate": 6.604239647893648e-06,
+      "loss": 0.7851,
+      "step": 145500
+    },
+    {
+      "epoch": 4.904338453247103,
+      "grad_norm": 0.6939913630485535,
+      "learning_rate": 6.379682026407976e-06,
+      "loss": 0.7871,
+      "step": 145600
+    },
+    {
+      "epoch": 4.907706817569388,
+      "grad_norm": 0.741671085357666,
+      "learning_rate": 6.155124404922303e-06,
+      "loss": 0.7829,
+      "step": 145700
+    },
+    {
+      "epoch": 4.911075181891674,
+      "grad_norm": 0.6528514623641968,
+      "learning_rate": 5.930566783436629e-06,
+      "loss": 0.7869,
+      "step": 145800
+    },
+    {
+      "epoch": 4.914443546213959,
+      "grad_norm": 0.6991773843765259,
+      "learning_rate": 5.706009161950956e-06,
+      "loss": 0.7847,
+      "step": 145900
+    },
+    {
+      "epoch": 4.917811910536243,
+      "grad_norm": 0.6840393543243408,
+      "learning_rate": 5.481451540465283e-06,
+      "loss": 0.7874,
+      "step": 146000
+    },
+    {
+      "epoch": 4.9211802748585285,
+      "grad_norm": 0.6707117557525635,
+      "learning_rate": 5.2568939189796095e-06,
+      "loss": 0.7874,
+      "step": 146100
+    },
+    {
+      "epoch": 4.924548639180814,
+      "grad_norm": 0.7077836990356445,
+      "learning_rate": 5.032336297493937e-06,
+      "loss": 0.7858,
+      "step": 146200
+    },
+    {
+      "epoch": 4.927917003503099,
+      "grad_norm": 0.6887540817260742,
+      "learning_rate": 4.807778676008263e-06,
+      "loss": 0.7834,
+      "step": 146300
+    },
+    {
+      "epoch": 4.931285367825384,
+      "grad_norm": 0.710809051990509,
+      "learning_rate": 4.5832210545225905e-06,
+      "loss": 0.7855,
+      "step": 146400
+    },
+    {
+      "epoch": 4.93465373214767,
+      "grad_norm": 0.6486634016036987,
+      "learning_rate": 4.358663433036916e-06,
+      "loss": 0.7833,
+      "step": 146500
+    },
+    {
+      "epoch": 4.938022096469954,
+      "grad_norm": 0.7380653619766235,
+      "learning_rate": 4.134105811551244e-06,
+      "loss": 0.7825,
+      "step": 146600
+    },
+    {
+      "epoch": 4.941390460792239,
+      "grad_norm": 0.672187328338623,
+      "learning_rate": 3.909548190065571e-06,
+      "loss": 0.7853,
+      "step": 146700
+    },
+    {
+      "epoch": 4.9447588251145245,
+      "grad_norm": 0.7624046206474304,
+      "learning_rate": 3.684990568579897e-06,
+      "loss": 0.787,
+      "step": 146800
+    },
+    {
+      "epoch": 4.94812718943681,
+      "grad_norm": 0.7582620978355408,
+      "learning_rate": 3.460432947094224e-06,
+      "loss": 0.7852,
+      "step": 146900
+    },
+    {
+      "epoch": 4.951495553759095,
+      "grad_norm": 0.7094969153404236,
+      "learning_rate": 3.235875325608551e-06,
+      "loss": 0.7845,
+      "step": 147000
+    },
+    {
+      "epoch": 4.954863918081379,
+      "grad_norm": 0.7196788787841797,
+      "learning_rate": 3.0113177041228775e-06,
+      "loss": 0.7844,
+      "step": 147100
+    },
+    {
+      "epoch": 4.958232282403665,
+      "grad_norm": 0.6927585005760193,
+      "learning_rate": 2.7867600826372042e-06,
+      "loss": 0.7839,
+      "step": 147200
+    },
+    {
+      "epoch": 4.96160064672595,
+      "grad_norm": 0.7253730893135071,
+      "learning_rate": 2.5622024611515314e-06,
+      "loss": 0.7838,
+      "step": 147300
+    },
+    {
+      "epoch": 4.964969011048235,
+      "grad_norm": 0.6914359331130981,
+      "learning_rate": 2.337644839665858e-06,
+      "loss": 0.7853,
+      "step": 147400
+    },
+    {
+      "epoch": 4.9683373753705204,
+      "grad_norm": 0.7005605101585388,
+      "learning_rate": 2.113087218180185e-06,
+      "loss": 0.7852,
+      "step": 147500
+    },
+    {
+      "epoch": 4.971705739692805,
+      "grad_norm": 0.6463395357131958,
+      "learning_rate": 1.8885295966945117e-06,
+      "loss": 0.7868,
+      "step": 147600
+    },
+    {
+      "epoch": 4.97507410401509,
+      "grad_norm": 0.6650646328926086,
+      "learning_rate": 1.6639719752088385e-06,
+      "loss": 0.7874,
+      "step": 147700
+    },
+    {
+      "epoch": 4.978442468337375,
+      "grad_norm": 0.7527862191200256,
+      "learning_rate": 1.4394143537231652e-06,
+      "loss": 0.7866,
+      "step": 147800
+    },
+    {
+      "epoch": 4.981810832659661,
+      "grad_norm": 0.7495971322059631,
+      "learning_rate": 1.2148567322374921e-06,
+      "loss": 0.786,
+      "step": 147900
+    },
+    {
+      "epoch": 4.985179196981946,
+      "grad_norm": 0.6791830062866211,
+      "learning_rate": 9.902991107518188e-07,
+      "loss": 0.7863,
+      "step": 148000
+    },
+    {
+      "epoch": 4.98854756130423,
+      "grad_norm": 0.6754565834999084,
+      "learning_rate": 7.657414892661456e-07,
+      "loss": 0.7832,
+      "step": 148100
+    },
+    {
+      "epoch": 4.9919159256265155,
+      "grad_norm": 0.6369759440422058,
+      "learning_rate": 5.411838677804724e-07,
+      "loss": 0.7807,
+      "step": 148200
+    },
+    {
+      "epoch": 4.995284289948801,
+      "grad_norm": 0.7656301259994507,
+      "learning_rate": 3.166262462947992e-07,
+      "loss": 0.7841,
+      "step": 148300
+    },
+    {
+      "epoch": 4.998652654271086,
+      "grad_norm": 0.6411105394363403,
+      "learning_rate": 9.206862480912601e-08,
+      "loss": 0.787,
+      "step": 148400
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.7365977168083191,
+      "eval_runtime": 1.819,
+      "eval_samples_per_second": 2748.792,
+      "eval_steps_per_second": 43.431,
+      "step": 148440
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 148440,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8701968384000000.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}