diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.995311767463667,
+  "eval_steps": 500,
+  "global_step": 5330,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.009376465072667605,
+      "grad_norm": 0.5424641966819763,
+      "learning_rate": 4.99998914337006e-05,
+      "loss": 0.8677,
+      "step": 5
+    },
+    {
+      "epoch": 0.01875293014533521,
+      "grad_norm": 0.5437079071998596,
+      "learning_rate": 4.999956573574533e-05,
+      "loss": 0.8147,
+      "step": 10
+    },
+    {
+      "epoch": 0.02812939521800281,
+      "grad_norm": 0.487265944480896,
+      "learning_rate": 4.9999022908962976e-05,
+      "loss": 0.7416,
+      "step": 15
+    },
+    {
+      "epoch": 0.03750586029067042,
+      "grad_norm": 0.5504423975944519,
+      "learning_rate": 4.999826295806815e-05,
+      "loss": 0.7843,
+      "step": 20
+    },
+    {
+      "epoch": 0.04688232536333802,
+      "grad_norm": 0.6139317154884338,
+      "learning_rate": 4.999728588966127e-05,
+      "loss": 0.7553,
+      "step": 25
+    },
+    {
+      "epoch": 0.05625879043600562,
+      "grad_norm": 0.6714254021644592,
+      "learning_rate": 4.999609171222846e-05,
+      "loss": 0.7018,
+      "step": 30
+    },
+    {
+      "epoch": 0.06563525550867323,
+      "grad_norm": 0.6426434516906738,
+      "learning_rate": 4.9994680436141516e-05,
+      "loss": 0.7141,
+      "step": 35
+    },
+    {
+      "epoch": 0.07501172058134084,
+      "grad_norm": 0.612634003162384,
+      "learning_rate": 4.99930520736578e-05,
+      "loss": 0.7257,
+      "step": 40
+    },
+    {
+      "epoch": 0.08438818565400844,
+      "grad_norm": 0.6088927984237671,
+      "learning_rate": 4.999120663892013e-05,
+      "loss": 0.7492,
+      "step": 45
+    },
+    {
+      "epoch": 0.09376465072667604,
+      "grad_norm": 0.5448371171951294,
+      "learning_rate": 4.998914414795668e-05,
+      "loss": 0.6998,
+      "step": 50
+    },
+    {
+      "epoch": 0.10314111579934365,
+      "grad_norm": 0.609154462814331,
+      "learning_rate": 4.9986864618680795e-05,
+      "loss": 0.7014,
+      "step": 55
+    },
+    {
+      "epoch": 0.11251758087201125,
+      "grad_norm": 0.8684519529342651,
+      "learning_rate": 4.99843680708909e-05,
+      "loss": 0.6824,
+      "step": 60
+    },
+    {
+      "epoch": 0.12189404594467886,
+      "grad_norm": 0.5986927151679993,
+      "learning_rate": 4.998165452627025e-05,
+      "loss": 0.6789,
+      "step": 65
+    },
+    {
+      "epoch": 0.13127051101734646,
+      "grad_norm": 0.6483319401741028,
+      "learning_rate": 4.997872400838682e-05,
+      "loss": 0.6439,
+      "step": 70
+    },
+    {
+      "epoch": 0.14064697609001406,
+      "grad_norm": 0.7569819092750549,
+      "learning_rate": 4.9975576542693044e-05,
+      "loss": 0.6833,
+      "step": 75
+    },
+    {
+      "epoch": 0.15002344116268168,
+      "grad_norm": 0.7457460165023804,
+      "learning_rate": 4.997221215652562e-05,
+      "loss": 0.6264,
+      "step": 80
+    },
+    {
+      "epoch": 0.15939990623534928,
+      "grad_norm": 0.7782825827598572,
+      "learning_rate": 4.996863087910526e-05,
+      "loss": 0.651,
+      "step": 85
+    },
+    {
+      "epoch": 0.16877637130801687,
+      "grad_norm": 0.8220712542533875,
+      "learning_rate": 4.9964832741536444e-05,
+      "loss": 0.6575,
+      "step": 90
+    },
+    {
+      "epoch": 0.1781528363806845,
+      "grad_norm": 0.7562811374664307,
+      "learning_rate": 4.996081777680716e-05,
+      "loss": 0.6336,
+      "step": 95
+    },
+    {
+      "epoch": 0.1875293014533521,
+      "grad_norm": 0.7624123096466064,
+      "learning_rate": 4.9956586019788584e-05,
+      "loss": 0.6131,
+      "step": 100
+    },
+    {
+      "epoch": 0.19690576652601968,
+      "grad_norm": 0.8074551820755005,
+      "learning_rate": 4.995213750723484e-05,
+      "loss": 0.662,
+      "step": 105
+    },
+    {
+      "epoch": 0.2062822315986873,
+      "grad_norm": 0.7907121181488037,
+      "learning_rate": 4.9947472277782584e-05,
+      "loss": 0.6428,
+      "step": 110
+    },
+    {
+      "epoch": 0.2156586966713549,
+      "grad_norm": 0.8095118999481201,
+      "learning_rate": 4.994259037195076e-05,
+      "loss": 0.6605,
+      "step": 115
+    },
+    {
+      "epoch": 0.2250351617440225,
+      "grad_norm": 0.8278319239616394,
+      "learning_rate": 4.993749183214021e-05,
+      "loss": 0.6526,
+      "step": 120
+    },
+    {
+      "epoch": 0.23441162681669012,
+      "grad_norm": 0.7899359464645386,
+      "learning_rate": 4.993217670263328e-05,
+      "loss": 0.6534,
+      "step": 125
+    },
+    {
+      "epoch": 0.2437880918893577,
+      "grad_norm": 0.8587161302566528,
+      "learning_rate": 4.992664502959351e-05,
+      "loss": 0.6343,
+      "step": 130
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.8511154651641846,
+      "learning_rate": 4.992089686106516e-05,
+      "loss": 0.6664,
+      "step": 135
+    },
+    {
+      "epoch": 0.26254102203469293,
+      "grad_norm": 0.8315441608428955,
+      "learning_rate": 4.991493224697281e-05,
+      "loss": 0.611,
+      "step": 140
+    },
+    {
+      "epoch": 0.2719174871073605,
+      "grad_norm": 0.87080317735672,
+      "learning_rate": 4.990875123912096e-05,
+      "loss": 0.6259,
+      "step": 145
+    },
+    {
+      "epoch": 0.2812939521800281,
+      "grad_norm": 0.8811519145965576,
+      "learning_rate": 4.990235389119352e-05,
+      "loss": 0.6398,
+      "step": 150
+    },
+    {
+      "epoch": 0.2906704172526957,
+      "grad_norm": 0.8086720108985901,
+      "learning_rate": 4.989574025875342e-05,
+      "loss": 0.6155,
+      "step": 155
+    },
+    {
+      "epoch": 0.30004688232536336,
+      "grad_norm": 0.8305373191833496,
+      "learning_rate": 4.9888910399242065e-05,
+      "loss": 0.6121,
+      "step": 160
+    },
+    {
+      "epoch": 0.30942334739803096,
+      "grad_norm": 0.8462594747543335,
+      "learning_rate": 4.988186437197885e-05,
+      "loss": 0.6582,
+      "step": 165
+    },
+    {
+      "epoch": 0.31879981247069855,
+      "grad_norm": 0.9462889432907104,
+      "learning_rate": 4.987460223816067e-05,
+      "loss": 0.6403,
+      "step": 170
+    },
+    {
+      "epoch": 0.32817627754336615,
+      "grad_norm": 0.7495555281639099,
+      "learning_rate": 4.986712406086137e-05,
+      "loss": 0.6339,
+      "step": 175
+    },
+    {
+      "epoch": 0.33755274261603374,
+      "grad_norm": 0.9177277684211731,
+      "learning_rate": 4.985942990503119e-05,
+      "loss": 0.6206,
+      "step": 180
+    },
+    {
+      "epoch": 0.34692920768870134,
+      "grad_norm": 0.9031923413276672,
+      "learning_rate": 4.985151983749621e-05,
+      "loss": 0.6357,
+      "step": 185
+    },
+    {
+      "epoch": 0.356305672761369,
+      "grad_norm": 0.8324641585350037,
+      "learning_rate": 4.984339392695777e-05,
+      "loss": 0.5994,
+      "step": 190
+    },
+    {
+      "epoch": 0.3656821378340366,
+      "grad_norm": 0.9211899042129517,
+      "learning_rate": 4.9835052243991874e-05,
+      "loss": 0.6596,
+      "step": 195
+    },
+    {
+      "epoch": 0.3750586029067042,
+      "grad_norm": 0.8922693729400635,
+      "learning_rate": 4.9826494861048576e-05,
+      "loss": 0.6399,
+      "step": 200
+    },
+    {
+      "epoch": 0.38443506797937177,
+      "grad_norm": 0.9406248331069946,
+      "learning_rate": 4.981772185245135e-05,
+      "loss": 0.6146,
+      "step": 205
+    },
+    {
+      "epoch": 0.39381153305203936,
+      "grad_norm": 0.9842830896377563,
+      "learning_rate": 4.980873329439644e-05,
+      "loss": 0.6333,
+      "step": 210
+    },
+    {
+      "epoch": 0.40318799812470696,
+      "grad_norm": 1.0240492820739746,
+      "learning_rate": 4.979952926495219e-05,
+      "loss": 0.6117,
+      "step": 215
+    },
+    {
+      "epoch": 0.4125644631973746,
+      "grad_norm": 0.9410423636436462,
+      "learning_rate": 4.979010984405842e-05,
+      "loss": 0.599,
+      "step": 220
+    },
+    {
+      "epoch": 0.4219409282700422,
+      "grad_norm": 0.8562084436416626,
+      "learning_rate": 4.978047511352565e-05,
+      "loss": 0.5964,
+      "step": 225
+    },
+    {
+      "epoch": 0.4313173933427098,
+      "grad_norm": 1.0595966577529907,
+      "learning_rate": 4.9770625157034436e-05,
+      "loss": 0.5823,
+      "step": 230
+    },
+    {
+      "epoch": 0.4406938584153774,
+      "grad_norm": 0.8417986631393433,
+      "learning_rate": 4.976056006013465e-05,
+      "loss": 0.6277,
+      "step": 235
+    },
+    {
+      "epoch": 0.450070323488045,
+      "grad_norm": 1.0082447528839111,
+      "learning_rate": 4.975027991024473e-05,
+      "loss": 0.6065,
+      "step": 240
+    },
+    {
+      "epoch": 0.45944678856071264,
+      "grad_norm": 0.8976945281028748,
+      "learning_rate": 4.973978479665088e-05,
+      "loss": 0.6246,
+      "step": 245
+    },
+    {
+      "epoch": 0.46882325363338023,
+      "grad_norm": 0.9026245474815369,
+      "learning_rate": 4.972907481050637e-05,
+      "loss": 0.5986,
+      "step": 250
+    },
+    {
+      "epoch": 0.4781997187060478,
+      "grad_norm": 0.9327179193496704,
+      "learning_rate": 4.971815004483068e-05,
+      "loss": 0.6198,
+      "step": 255
+    },
+    {
+      "epoch": 0.4875761837787154,
+      "grad_norm": 0.9296254515647888,
+      "learning_rate": 4.970701059450872e-05,
+      "loss": 0.5762,
+      "step": 260
+    },
+    {
+      "epoch": 0.496952648851383,
+      "grad_norm": 1.0290290117263794,
+      "learning_rate": 4.969565655628999e-05,
+      "loss": 0.5778,
+      "step": 265
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.9617416262626648,
+      "learning_rate": 4.968408802878778e-05,
+      "loss": 0.6195,
+      "step": 270
+    },
+    {
+      "epoch": 0.5157055789967182,
+      "grad_norm": 0.84481281042099,
+      "learning_rate": 4.9672305112478266e-05,
+      "loss": 0.575,
+      "step": 275
+    },
+    {
+      "epoch": 0.5250820440693859,
+      "grad_norm": 0.9828088879585266,
+      "learning_rate": 4.9660307909699645e-05,
+      "loss": 0.6162,
+      "step": 280
+    },
+    {
+      "epoch": 0.5344585091420534,
+      "grad_norm": 0.8879114985466003,
+      "learning_rate": 4.9648096524651285e-05,
+      "loss": 0.5918,
+      "step": 285
+    },
+    {
+      "epoch": 0.543834974214721,
+      "grad_norm": 0.9867956042289734,
+      "learning_rate": 4.963567106339276e-05,
+      "loss": 0.574,
+      "step": 290
+    },
+    {
+      "epoch": 0.5532114392873887,
+      "grad_norm": 0.9731470346450806,
+      "learning_rate": 4.9623031633842995e-05,
+      "loss": 0.6111,
+      "step": 295
+    },
+    {
+      "epoch": 0.5625879043600562,
+      "grad_norm": 0.9366615414619446,
+      "learning_rate": 4.961017834577927e-05,
+      "loss": 0.6199,
+      "step": 300
+    },
+    {
+      "epoch": 0.5719643694327239,
+      "grad_norm": 0.8803014159202576,
+      "learning_rate": 4.9597111310836294e-05,
+      "loss": 0.5866,
+      "step": 305
+    },
+    {
+      "epoch": 0.5813408345053914,
+      "grad_norm": 0.8761358857154846,
+      "learning_rate": 4.958383064250525e-05,
+      "loss": 0.5915,
+      "step": 310
+    },
+    {
+      "epoch": 0.5907172995780591,
+      "grad_norm": 0.9882873296737671,
+      "learning_rate": 4.957033645613276e-05,
+      "loss": 0.5865,
+      "step": 315
+    },
+    {
+      "epoch": 0.6000937646507267,
+      "grad_norm": 0.9283161163330078,
+      "learning_rate": 4.955662886891995e-05,
+      "loss": 0.6528,
+      "step": 320
+    },
+    {
+      "epoch": 0.6094702297233943,
+      "grad_norm": 0.9495806097984314,
+      "learning_rate": 4.954270799992138e-05,
+      "loss": 0.5969,
+      "step": 325
+    },
+    {
+      "epoch": 0.6188466947960619,
+      "grad_norm": 0.947560727596283,
+      "learning_rate": 4.952857397004401e-05,
+      "loss": 0.5814,
+      "step": 330
+    },
+    {
+      "epoch": 0.6282231598687295,
+      "grad_norm": 1.0182502269744873,
+      "learning_rate": 4.951422690204622e-05,
+      "loss": 0.6264,
+      "step": 335
+    },
+    {
+      "epoch": 0.6375996249413971,
+      "grad_norm": 0.9604515433311462,
+      "learning_rate": 4.949966692053663e-05,
+      "loss": 0.5847,
+      "step": 340
+    },
+    {
+      "epoch": 0.6469760900140648,
+      "grad_norm": 0.9579267501831055,
+      "learning_rate": 4.948489415197311e-05,
+      "loss": 0.6255,
+      "step": 345
+    },
+    {
+      "epoch": 0.6563525550867323,
+      "grad_norm": 1.0120905637741089,
+      "learning_rate": 4.946990872466164e-05,
+      "loss": 0.6465,
+      "step": 350
+    },
+    {
+      "epoch": 0.6657290201593999,
+      "grad_norm": 1.0143414735794067,
+      "learning_rate": 4.9454710768755224e-05,
+      "loss": 0.559,
+      "step": 355
+    },
+    {
+      "epoch": 0.6751054852320675,
+      "grad_norm": 0.9492619633674622,
+      "learning_rate": 4.943930041625272e-05,
+      "loss": 0.6202,
+      "step": 360
+    },
+    {
+      "epoch": 0.6844819503047351,
+      "grad_norm": 0.9203887581825256,
+      "learning_rate": 4.942367780099773e-05,
+      "loss": 0.5514,
+      "step": 365
+    },
+    {
+      "epoch": 0.6938584153774027,
+      "grad_norm": 1.0443195104599,
+      "learning_rate": 4.940784305867741e-05,
+      "loss": 0.5831,
+      "step": 370
+    },
+    {
+      "epoch": 0.7032348804500703,
+      "grad_norm": 1.0183184146881104,
+      "learning_rate": 4.939179632682131e-05,
+      "loss": 0.602,
+      "step": 375
+    },
+    {
+      "epoch": 0.712611345522738,
+      "grad_norm": 0.9733455181121826,
+      "learning_rate": 4.937553774480018e-05,
+      "loss": 0.5955,
+      "step": 380
+    },
+    {
+      "epoch": 0.7219878105954055,
+      "grad_norm": 1.0156100988388062,
+      "learning_rate": 4.9359067453824745e-05,
+      "loss": 0.561,
+      "step": 385
+    },
+    {
+      "epoch": 0.7313642756680732,
+      "grad_norm": 1.0459623336791992,
+      "learning_rate": 4.934238559694448e-05,
+      "loss": 0.5894,
+      "step": 390
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 1.0285617113113403,
+      "learning_rate": 4.932549231904638e-05,
+      "loss": 0.5954,
+      "step": 395
+    },
+    {
+      "epoch": 0.7501172058134083,
+      "grad_norm": 1.0028026103973389,
+      "learning_rate": 4.9308387766853725e-05,
+      "loss": 0.6117,
+      "step": 400
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 1.0255147218704224,
+      "learning_rate": 4.9291072088924714e-05,
+      "loss": 0.5808,
+      "step": 405
+    },
+    {
+      "epoch": 0.7688701359587435,
+      "grad_norm": 0.9642876982688904,
+      "learning_rate": 4.92735454356513e-05,
+      "loss": 0.5858,
+      "step": 410
+    },
+    {
+      "epoch": 0.7782466010314112,
+      "grad_norm": 0.9397988319396973,
+      "learning_rate": 4.925580795925778e-05,
+      "loss": 0.5747,
+      "step": 415
+    },
+    {
+      "epoch": 0.7876230661040787,
+      "grad_norm": 1.084703803062439,
+      "learning_rate": 4.9237859813799535e-05,
+      "loss": 0.6057,
+      "step": 420
+    },
+    {
+      "epoch": 0.7969995311767464,
+      "grad_norm": 1.1365820169448853,
+      "learning_rate": 4.9219701155161666e-05,
+      "loss": 0.5915,
+      "step": 425
+    },
+    {
+      "epoch": 0.8063759962494139,
+      "grad_norm": 1.0651617050170898,
+      "learning_rate": 4.9201332141057623e-05,
+      "loss": 0.5727,
+      "step": 430
+    },
+    {
+      "epoch": 0.8157524613220816,
+      "grad_norm": 1.0743099451065063,
+      "learning_rate": 4.91827529310279e-05,
+      "loss": 0.5274,
+      "step": 435
+    },
+    {
+      "epoch": 0.8251289263947492,
+      "grad_norm": 0.9993802905082703,
+      "learning_rate": 4.9163963686438575e-05,
+      "loss": 0.5488,
+      "step": 440
+    },
+    {
+      "epoch": 0.8345053914674168,
+      "grad_norm": 1.00540292263031,
+      "learning_rate": 4.914496457047995e-05,
+      "loss": 0.5947,
+      "step": 445
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "grad_norm": 1.0636786222457886,
+      "learning_rate": 4.912575574816511e-05,
+      "loss": 0.5809,
+      "step": 450
+    },
+    {
+      "epoch": 0.853258321612752,
+      "grad_norm": 0.9834513068199158,
+      "learning_rate": 4.9106337386328524e-05,
+      "loss": 0.5694,
+      "step": 455
+    },
+    {
+      "epoch": 0.8626347866854196,
+      "grad_norm": 1.037613034248352,
+      "learning_rate": 4.908670965362457e-05,
+      "loss": 0.611,
+      "step": 460
+    },
+    {
+      "epoch": 0.8720112517580872,
+      "grad_norm": 1.004014492034912,
+      "learning_rate": 4.906687272052608e-05,
+      "loss": 0.5837,
+      "step": 465
+    },
+    {
+      "epoch": 0.8813877168307548,
+      "grad_norm": 0.9947504997253418,
+      "learning_rate": 4.9046826759322825e-05,
+      "loss": 0.6068,
+      "step": 470
+    },
+    {
+      "epoch": 0.8907641819034224,
+      "grad_norm": 1.0303345918655396,
+      "learning_rate": 4.902657194412009e-05,
+      "loss": 0.6247,
+      "step": 475
+    },
+    {
+      "epoch": 0.90014064697609,
+      "grad_norm": 1.0110838413238525,
+      "learning_rate": 4.9006108450837095e-05,
+      "loss": 0.5827,
+      "step": 480
+    },
+    {
+      "epoch": 0.9095171120487576,
+      "grad_norm": 1.014485239982605,
+      "learning_rate": 4.89854364572055e-05,
+      "loss": 0.5582,
+      "step": 485
+    },
+    {
+      "epoch": 0.9188935771214253,
+      "grad_norm": 0.980571448802948,
+      "learning_rate": 4.8964556142767845e-05,
+      "loss": 0.5915,
+      "step": 490
+    },
+    {
+      "epoch": 0.9282700421940928,
+      "grad_norm": 1.0147678852081299,
+      "learning_rate": 4.894346768887602e-05,
+      "loss": 0.6365,
+      "step": 495
+    },
+    {
+      "epoch": 0.9376465072667605,
+      "grad_norm": 1.1158618927001953,
+      "learning_rate": 4.892217127868965e-05,
+      "loss": 0.5784,
+      "step": 500
+    },
+    {
+      "epoch": 0.947022972339428,
+      "grad_norm": 0.9598096609115601,
+      "learning_rate": 4.890066709717454e-05,
+      "loss": 0.5929,
+      "step": 505
+    },
+    {
+      "epoch": 0.9563994374120957,
+      "grad_norm": 0.9693619012832642,
+      "learning_rate": 4.8878955331101026e-05,
+      "loss": 0.626,
+      "step": 510
+    },
+    {
+      "epoch": 0.9657759024847632,
+      "grad_norm": 0.8924005031585693,
+      "learning_rate": 4.885703616904241e-05,
+      "loss": 0.5658,
+      "step": 515
+    },
+    {
+      "epoch": 0.9751523675574308,
+      "grad_norm": 1.0127456188201904,
+      "learning_rate": 4.8834909801373264e-05,
+      "loss": 0.5631,
+      "step": 520
+    },
+    {
+      "epoch": 0.9845288326300985,
+      "grad_norm": 1.0072908401489258,
+      "learning_rate": 4.881257642026783e-05,
+      "loss": 0.5816,
+      "step": 525
+    },
+    {
+      "epoch": 0.993905297702766,
+      "grad_norm": 0.9498401880264282,
+      "learning_rate": 4.879003621969831e-05,
+      "loss": 0.581,
+      "step": 530
+    },
+    {
+      "epoch": 1.0032817627754336,
+      "grad_norm": 1.1577348709106445,
+      "learning_rate": 4.876728939543318e-05,
+      "loss": 0.5467,
+      "step": 535
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 1.048673391342163,
+      "learning_rate": 4.874433614503554e-05,
+      "loss": 0.538,
+      "step": 540
+    },
+    {
+      "epoch": 1.0220346929207689,
+      "grad_norm": 1.0235296487808228,
+      "learning_rate": 4.872117666786134e-05,
+      "loss": 0.5274,
+      "step": 545
+    },
+    {
+      "epoch": 1.0314111579934364,
+      "grad_norm": 0.9855211973190308,
+      "learning_rate": 4.869781116505768e-05,
+      "loss": 0.5977,
+      "step": 550
+    },
+    {
+      "epoch": 1.0407876230661042,
+      "grad_norm": 1.0761393308639526,
+      "learning_rate": 4.8674239839561055e-05,
+      "loss": 0.5568,
+      "step": 555
+    },
+    {
+      "epoch": 1.0501640881387717,
+      "grad_norm": 0.9937365651130676,
+      "learning_rate": 4.8650462896095597e-05,
+      "loss": 0.5378,
+      "step": 560
+    },
+    {
+      "epoch": 1.0595405532114393,
+      "grad_norm": 1.1459683179855347,
+      "learning_rate": 4.862648054117127e-05,
+      "loss": 0.5836,
+      "step": 565
+    },
+    {
+      "epoch": 1.0689170182841068,
+      "grad_norm": 1.0522550344467163,
+      "learning_rate": 4.860229298308213e-05,
+      "loss": 0.5818,
+      "step": 570
+    },
+    {
+      "epoch": 1.0782934833567746,
+      "grad_norm": 1.1253257989883423,
+      "learning_rate": 4.8577900431904466e-05,
+      "loss": 0.5647,
+      "step": 575
+    },
+    {
+      "epoch": 1.087669948429442,
+      "grad_norm": 1.0557312965393066,
+      "learning_rate": 4.8553303099495e-05,
+      "loss": 0.5702,
+      "step": 580
+    },
+    {
+      "epoch": 1.0970464135021096,
+      "grad_norm": 1.1547330617904663,
+      "learning_rate": 4.852850119948904e-05,
+      "loss": 0.5631,
+      "step": 585
+    },
+    {
+      "epoch": 1.1064228785747774,
+      "grad_norm": 0.9754932522773743,
+      "learning_rate": 4.8503494947298634e-05,
+      "loss": 0.5508,
+      "step": 590
+    },
+    {
+      "epoch": 1.115799343647445,
+      "grad_norm": 1.0107804536819458,
+      "learning_rate": 4.847828456011066e-05,
+      "loss": 0.5659,
+      "step": 595
+    },
+    {
+      "epoch": 1.1251758087201125,
+      "grad_norm": 1.0393505096435547,
+      "learning_rate": 4.845287025688503e-05,
+      "loss": 0.5501,
+      "step": 600
+    },
+    {
+      "epoch": 1.1345522737927802,
+      "grad_norm": 1.1557810306549072,
+      "learning_rate": 4.842725225835266e-05,
+      "loss": 0.5629,
+      "step": 605
+    },
+    {
+      "epoch": 1.1439287388654478,
+      "grad_norm": 1.0539976358413696,
+      "learning_rate": 4.8401430787013666e-05,
+      "loss": 0.5359,
+      "step": 610
+    },
+    {
+      "epoch": 1.1533052039381153,
+      "grad_norm": 1.1011552810668945,
+      "learning_rate": 4.837540606713538e-05,
+      "loss": 0.5294,
+      "step": 615
+    },
+    {
+      "epoch": 1.1626816690107828,
+      "grad_norm": 1.1274709701538086,
+      "learning_rate": 4.8349178324750387e-05,
+      "loss": 0.5286,
+      "step": 620
+    },
+    {
+      "epoch": 1.1720581340834506,
+      "grad_norm": 1.0979019403457642,
+      "learning_rate": 4.832274778765462e-05,
+      "loss": 0.5511,
+      "step": 625
+    },
+    {
+      "epoch": 1.1814345991561181,
+      "grad_norm": 1.095531702041626,
+      "learning_rate": 4.8296114685405324e-05,
+      "loss": 0.5687,
+      "step": 630
+    },
+    {
+      "epoch": 1.1908110642287857,
+      "grad_norm": 1.0479497909545898,
+      "learning_rate": 4.826927924931908e-05,
+      "loss": 0.5589,
+      "step": 635
+    },
+    {
+      "epoch": 1.2001875293014534,
+      "grad_norm": 1.013268232345581,
+      "learning_rate": 4.824224171246981e-05,
+      "loss": 0.5294,
+      "step": 640
+    },
+    {
+      "epoch": 1.209563994374121,
+      "grad_norm": 1.1136912107467651,
+      "learning_rate": 4.821500230968674e-05,
+      "loss": 0.5316,
+      "step": 645
+    },
+    {
+      "epoch": 1.2189404594467885,
+      "grad_norm": 1.1556483507156372,
+      "learning_rate": 4.8187561277552374e-05,
+      "loss": 0.5839,
+      "step": 650
+    },
+    {
+      "epoch": 1.228316924519456,
+      "grad_norm": 1.102048397064209,
+      "learning_rate": 4.8159918854400394e-05,
+      "loss": 0.5622,
+      "step": 655
+    },
+    {
+      "epoch": 1.2376933895921238,
+      "grad_norm": 1.1714756488800049,
+      "learning_rate": 4.813207528031366e-05,
+      "loss": 0.5794,
+      "step": 660
+    },
+    {
+      "epoch": 1.2470698546647914,
+      "grad_norm": 1.3068993091583252,
+      "learning_rate": 4.810403079712208e-05,
+      "loss": 0.567,
+      "step": 665
+    },
+    {
+      "epoch": 1.256446319737459,
+      "grad_norm": 1.248699426651001,
+      "learning_rate": 4.807578564840051e-05,
+      "loss": 0.6111,
+      "step": 670
+    },
+    {
+      "epoch": 1.2658227848101267,
+      "grad_norm": 1.0111894607543945,
+      "learning_rate": 4.804734007946665e-05,
+      "loss": 0.604,
+      "step": 675
+    },
+    {
+      "epoch": 1.2751992498827942,
+      "grad_norm": 1.0763133764266968,
+      "learning_rate": 4.801869433737891e-05,
+      "loss": 0.5297,
+      "step": 680
+    },
+    {
+      "epoch": 1.2845757149554617,
+      "grad_norm": 1.1137092113494873,
+      "learning_rate": 4.798984867093428e-05,
+      "loss": 0.5701,
+      "step": 685
+    },
+    {
+      "epoch": 1.2939521800281293,
+      "grad_norm": 1.072448492050171,
+      "learning_rate": 4.796080333066613e-05,
+      "loss": 0.5849,
+      "step": 690
+    },
+    {
+      "epoch": 1.303328645100797,
+      "grad_norm": 1.1339977979660034,
+      "learning_rate": 4.7931558568842064e-05,
+      "loss": 0.5795,
+      "step": 695
+    },
+    {
+      "epoch": 1.3127051101734646,
+      "grad_norm": 1.0227925777435303,
+      "learning_rate": 4.790211463946174e-05,
+      "loss": 0.5386,
+      "step": 700
+    },
+    {
+      "epoch": 1.3220815752461323,
+      "grad_norm": 0.8955875039100647,
+      "learning_rate": 4.7872471798254624e-05,
+      "loss": 0.5619,
+      "step": 705
+    },
+    {
+      "epoch": 1.3314580403187999,
+      "grad_norm": 1.033470630645752,
+      "learning_rate": 4.784263030267781e-05,
+      "loss": 0.5533,
+      "step": 710
+    },
+    {
+      "epoch": 1.3408345053914674,
+      "grad_norm": 1.0284202098846436,
+      "learning_rate": 4.781259041191375e-05,
+      "loss": 0.5125,
+      "step": 715
+    },
+    {
+      "epoch": 1.350210970464135,
+      "grad_norm": 1.1263688802719116,
+      "learning_rate": 4.7782352386868035e-05,
+      "loss": 0.5577,
+      "step": 720
+    },
+    {
+      "epoch": 1.3595874355368025,
+      "grad_norm": 1.1555782556533813,
+      "learning_rate": 4.7751916490167094e-05,
+      "loss": 0.5614,
+      "step": 725
+    },
+    {
+      "epoch": 1.3689639006094703,
+      "grad_norm": 1.003699541091919,
+      "learning_rate": 4.7721282986155945e-05,
+      "loss": 0.5648,
+      "step": 730
+    },
+    {
+      "epoch": 1.3783403656821378,
+      "grad_norm": 1.0394999980926514,
+      "learning_rate": 4.769045214089588e-05,
+      "loss": 0.5572,
+      "step": 735
+    },
+    {
+      "epoch": 1.3877168307548056,
+      "grad_norm": 1.209919810295105,
+      "learning_rate": 4.7659424222162165e-05,
+      "loss": 0.5295,
+      "step": 740
+    },
+    {
+      "epoch": 1.397093295827473,
+      "grad_norm": 1.1991006135940552,
+      "learning_rate": 4.76281994994417e-05,
+      "loss": 0.577,
+      "step": 745
+    },
+    {
+      "epoch": 1.4064697609001406,
+      "grad_norm": 1.029775857925415,
+      "learning_rate": 4.7596778243930694e-05,
+      "loss": 0.5495,
+      "step": 750
+    },
+    {
+      "epoch": 1.4158462259728082,
+      "grad_norm": 1.0505032539367676,
+      "learning_rate": 4.7565160728532307e-05,
+      "loss": 0.5662,
+      "step": 755
+    },
+    {
+      "epoch": 1.4252226910454757,
+      "grad_norm": 0.9723071455955505,
+      "learning_rate": 4.7533347227854265e-05,
+      "loss": 0.591,
+      "step": 760
+    },
+    {
+      "epoch": 1.4345991561181435,
+      "grad_norm": 1.046174168586731,
+      "learning_rate": 4.750133801820649e-05,
+      "loss": 0.572,
+      "step": 765
+    },
+    {
+      "epoch": 1.443975621190811,
+      "grad_norm": 1.1850031614303589,
+      "learning_rate": 4.7469133377598695e-05,
+      "loss": 0.5458,
+      "step": 770
+    },
+    {
+      "epoch": 1.4533520862634788,
+      "grad_norm": 1.0828883647918701,
+      "learning_rate": 4.743673358573799e-05,
+      "loss": 0.5458,
+      "step": 775
+    },
+    {
+      "epoch": 1.4627285513361463,
+      "grad_norm": 1.0885288715362549,
+      "learning_rate": 4.740413892402639e-05,
+      "loss": 0.5633,
+      "step": 780
+    },
+    {
+      "epoch": 1.4721050164088139,
+      "grad_norm": 1.0618774890899658,
+      "learning_rate": 4.7371349675558465e-05,
+      "loss": 0.5552,
+      "step": 785
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 1.0389914512634277,
+      "learning_rate": 4.7338366125118775e-05,
+      "loss": 0.6061,
+      "step": 790
+    },
+    {
+      "epoch": 1.4908579465541492,
+      "grad_norm": 1.114457368850708,
+      "learning_rate": 4.73051885591795e-05,
+      "loss": 0.5525,
+      "step": 795
+    },
+    {
+      "epoch": 1.5002344116268167,
+      "grad_norm": 1.0990691184997559,
+      "learning_rate": 4.727181726589789e-05,
+      "loss": 0.6171,
+      "step": 800
+    },
+    {
+      "epoch": 1.5096108766994845,
+      "grad_norm": 1.0763331651687622,
+      "learning_rate": 4.7238252535113756e-05,
+      "loss": 0.5442,
+      "step": 805
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 1.1251211166381836,
+      "learning_rate": 4.7204494658346996e-05,
+      "loss": 0.5293,
+      "step": 810
+    },
+    {
+      "epoch": 1.5283638068448195,
+      "grad_norm": 0.9672238230705261,
+      "learning_rate": 4.717054392879503e-05,
+      "loss": 0.5362,
+      "step": 815
+    },
+    {
+      "epoch": 1.537740271917487,
+      "grad_norm": 1.0139844417572021,
+      "learning_rate": 4.713640064133025e-05,
+      "loss": 0.5805,
+      "step": 820
+    },
+    {
+      "epoch": 1.5471167369901546,
+      "grad_norm": 1.0186749696731567,
+      "learning_rate": 4.7102065092497504e-05,
+      "loss": 0.554,
+      "step": 825
+    },
+    {
+      "epoch": 1.5564932020628222,
+      "grad_norm": 1.1397581100463867,
+      "learning_rate": 4.706753758051145e-05,
+      "loss": 0.568,
+      "step": 830
+    },
+    {
+      "epoch": 1.56586966713549,
+      "grad_norm": 0.9560020565986633,
+      "learning_rate": 4.7032818405254054e-05,
+      "loss": 0.5074,
+      "step": 835
+    },
+    {
+      "epoch": 1.5752461322081577,
+      "grad_norm": 1.1008929014205933,
+      "learning_rate": 4.699790786827188e-05,
+      "loss": 0.5672,
+      "step": 840
+    },
+    {
+      "epoch": 1.5846225972808252,
+      "grad_norm": 1.0296518802642822,
+      "learning_rate": 4.6962806272773564e-05,
+      "loss": 0.5534,
+      "step": 845
+    },
+    {
+      "epoch": 1.5939990623534928,
+      "grad_norm": 1.013056755065918,
+      "learning_rate": 4.6927513923627124e-05,
+      "loss": 0.5753,
+      "step": 850
+    },
+    {
+      "epoch": 1.6033755274261603,
+      "grad_norm": 1.1257952451705933,
+      "learning_rate": 4.689203112735735e-05,
+      "loss": 0.5789,
+      "step": 855
+    },
+    {
+      "epoch": 1.6127519924988278,
+      "grad_norm": 1.0546296834945679,
+      "learning_rate": 4.68563581921431e-05,
+      "loss": 0.5512,
+      "step": 860
+    },
+    {
+      "epoch": 1.6221284575714956,
+      "grad_norm": 1.0948747396469116,
+      "learning_rate": 4.682049542781468e-05,
+      "loss": 0.5319,
+      "step": 865
+    },
+    {
+      "epoch": 1.6315049226441631,
+      "grad_norm": 1.2167490720748901,
+      "learning_rate": 4.6784443145851074e-05,
+      "loss": 0.573,
+      "step": 870
+    },
+    {
+      "epoch": 1.640881387716831,
+      "grad_norm": 1.0839264392852783,
+      "learning_rate": 4.6748201659377335e-05,
+      "loss": 0.5372,
+      "step": 875
+    },
+    {
+      "epoch": 1.6502578527894984,
+      "grad_norm": 1.1405868530273438,
+      "learning_rate": 4.671177128316176e-05,
+      "loss": 0.5389,
+      "step": 880
+    },
+    {
+      "epoch": 1.659634317862166,
+      "grad_norm": 1.003694772720337,
+      "learning_rate": 4.6675152333613256e-05,
+      "loss": 0.5828,
+      "step": 885
+    },
+    {
+      "epoch": 1.6690107829348335,
+      "grad_norm": 0.9864010214805603,
+      "learning_rate": 4.663834512877853e-05,
+      "loss": 0.5193,
+      "step": 890
+    },
+    {
+      "epoch": 1.678387248007501,
+      "grad_norm": 1.081709861755371,
+      "learning_rate": 4.6601349988339345e-05,
+      "loss": 0.5626,
+      "step": 895
+    },
+    {
+      "epoch": 1.6877637130801688,
+      "grad_norm": 1.120210886001587,
+      "learning_rate": 4.6564167233609736e-05,
+      "loss": 0.598,
+      "step": 900
+    },
+    {
+      "epoch": 1.6971401781528364,
+      "grad_norm": 1.0161548852920532,
+      "learning_rate": 4.6526797187533225e-05,
+      "loss": 0.5661,
+      "step": 905
+    },
+    {
+      "epoch": 1.7065166432255041,
+      "grad_norm": 0.9947847723960876,
+      "learning_rate": 4.648924017468003e-05,
+      "loss": 0.5996,
+      "step": 910
+    },
+    {
+      "epoch": 1.7158931082981717,
+      "grad_norm": 1.0159549713134766,
+      "learning_rate": 4.645149652124422e-05,
+      "loss": 0.5589,
+      "step": 915
+    },
+    {
+      "epoch": 1.7252695733708392,
+      "grad_norm": 1.0072029829025269,
+      "learning_rate": 4.6413566555040896e-05,
+      "loss": 0.6435,
+      "step": 920
+    },
+    {
+      "epoch": 1.7346460384435067,
+      "grad_norm": 1.1263002157211304,
+      "learning_rate": 4.6375450605503345e-05,
+      "loss": 0.5335,
+      "step": 925
+    },
+    {
+      "epoch": 1.7440225035161743,
+      "grad_norm": 1.0950555801391602,
+      "learning_rate": 4.633714900368018e-05,
+      "loss": 0.5713,
+      "step": 930
+    },
+    {
+      "epoch": 1.753398968588842,
+      "grad_norm": 1.0297006368637085,
+      "learning_rate": 4.6298662082232446e-05,
+      "loss": 0.5544,
+      "step": 935
+    },
+    {
+      "epoch": 1.7627754336615096,
+      "grad_norm": 1.1370675563812256,
+      "learning_rate": 4.625999017543075e-05,
+      "loss": 0.6001,
+      "step": 940
+    },
+    {
+      "epoch": 1.7721518987341773,
+      "grad_norm": 1.1817889213562012,
+      "learning_rate": 4.622113361915237e-05,
+      "loss": 0.5393,
+      "step": 945
+    },
+    {
+      "epoch": 1.7815283638068449,
+      "grad_norm": 1.1695425510406494,
+      "learning_rate": 4.618209275087829e-05,
+      "loss": 0.5749,
+      "step": 950
+    },
+    {
+      "epoch": 1.7909048288795124,
+      "grad_norm": 1.1453884840011597,
+      "learning_rate": 4.614286790969034e-05,
+      "loss": 0.5495,
+      "step": 955
+    },
+    {
+      "epoch": 1.80028129395218,
+      "grad_norm": 1.067964792251587,
+      "learning_rate": 4.610345943626817e-05,
+      "loss": 0.6157,
+      "step": 960
+    },
+    {
+      "epoch": 1.8096577590248475,
+      "grad_norm": 1.0637770891189575,
+      "learning_rate": 4.606386767288636e-05,
+      "loss": 0.5457,
+      "step": 965
+    },
+    {
+      "epoch": 1.8190342240975153,
+      "grad_norm": 1.0843828916549683,
+      "learning_rate": 4.602409296341141e-05,
+      "loss": 0.5589,
+      "step": 970
+    },
+    {
+      "epoch": 1.8284106891701828,
+      "grad_norm": 1.0681754350662231,
+      "learning_rate": 4.598413565329875e-05,
+      "loss": 0.5396,
+      "step": 975
+    },
+    {
+      "epoch": 1.8377871542428506,
+      "grad_norm": 0.9996591806411743,
+      "learning_rate": 4.5943996089589775e-05,
+      "loss": 0.5272,
+      "step": 980
+    },
+    {
+      "epoch": 1.847163619315518,
+      "grad_norm": 1.180212140083313,
+      "learning_rate": 4.590367462090879e-05,
+      "loss": 0.5361,
+      "step": 985
+    },
+    {
+      "epoch": 1.8565400843881856,
+      "grad_norm": 0.9685352444648743,
+      "learning_rate": 4.586317159746001e-05,
+      "loss": 0.5275,
+      "step": 990
+    },
+    {
+      "epoch": 1.8659165494608532,
+      "grad_norm": 1.0599303245544434,
+      "learning_rate": 4.5822487371024495e-05,
+      "loss": 0.5352,
+      "step": 995
+    },
+    {
+      "epoch": 1.8752930145335207,
+      "grad_norm": 1.0124273300170898,
+      "learning_rate": 4.5781622294957136e-05,
+      "loss": 0.5327,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8846694796061885,
+      "grad_norm": 1.1367263793945312,
+      "learning_rate": 4.5740576724183525e-05,
+      "loss": 0.5183,
+      "step": 1005
+    },
+    {
+      "epoch": 1.8940459446788562,
+      "grad_norm": 1.050596833229065,
+      "learning_rate": 4.569935101519692e-05,
+      "loss": 0.5301,
+      "step": 1010
+    },
+    {
+      "epoch": 1.9034224097515238,
+      "grad_norm": 0.9859219789505005,
+      "learning_rate": 4.565794552605514e-05,
+      "loss": 0.5641,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9127988748241913,
+      "grad_norm": 1.123216986656189,
+      "learning_rate": 4.561636061637745e-05,
+      "loss": 0.5391,
+      "step": 1020
+    },
+    {
+      "epoch": 1.9221753398968588,
+      "grad_norm": 0.9957221150398254,
+      "learning_rate": 4.557459664734141e-05,
+      "loss": 0.5367,
+      "step": 1025
+    },
+    {
+      "epoch": 1.9315518049695264,
+      "grad_norm": 1.1165965795516968,
+      "learning_rate": 4.553265398167981e-05,
+      "loss": 0.5418,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9409282700421941,
+      "grad_norm": 0.9105197191238403,
+      "learning_rate": 4.549053298367742e-05,
+      "loss": 0.5466,
+      "step": 1035
+    },
+    {
+      "epoch": 1.9503047351148617,
+      "grad_norm": 1.0891969203948975,
+      "learning_rate": 4.5448234019167945e-05,
+      "loss": 0.5408,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9596812001875294,
+      "grad_norm": 1.0737484693527222,
+      "learning_rate": 4.540575745553072e-05,
+      "loss": 0.579,
+      "step": 1045
+    },
+    {
+      "epoch": 1.969057665260197,
+      "grad_norm": 1.1235281229019165,
+      "learning_rate": 4.536310366168763e-05,
+      "loss": 0.5185,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9784341303328645,
+      "grad_norm": 1.1056690216064453,
+      "learning_rate": 4.532027300809983e-05,
+      "loss": 0.523,
+      "step": 1055
+    },
+    {
+      "epoch": 1.987810595405532,
+      "grad_norm": 1.0960348844528198,
+      "learning_rate": 4.5277265866764565e-05,
+      "loss": 0.5206,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9971870604781996,
+      "grad_norm": 1.0557410717010498,
+      "learning_rate": 4.5234082611211926e-05,
+      "loss": 0.5457,
+      "step": 1065
+    },
+    {
+      "epoch": 2.006563525550867,
+      "grad_norm": 1.0027008056640625,
+      "learning_rate": 4.519072361650163e-05,
+      "loss": 0.4943,
+      "step": 1070
+    },
+    {
+      "epoch": 2.015939990623535,
+      "grad_norm": 0.993651270866394,
+      "learning_rate": 4.51471892592197e-05,
+      "loss": 0.5032,
+      "step": 1075
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 1.10842764377594,
+      "learning_rate": 4.5103479917475286e-05,
+      "loss": 0.5602,
+      "step": 1080
+    },
+    {
+      "epoch": 2.03469292076887,
+      "grad_norm": 0.9827935099601746,
+      "learning_rate": 4.505959597089729e-05,
+      "loss": 0.5449,
+      "step": 1085
+    },
+    {
+      "epoch": 2.0440693858415377,
+      "grad_norm": 1.0879554748535156,
+      "learning_rate": 4.501553780063113e-05,
+      "loss": 0.5494,
+      "step": 1090
+    },
+    {
+      "epoch": 2.0534458509142053,
+      "grad_norm": 1.146403431892395,
+      "learning_rate": 4.4971305789335415e-05,
+      "loss": 0.5545,
+      "step": 1095
+    },
+    {
+      "epoch": 2.062822315986873,
+      "grad_norm": 1.1221157312393188,
+      "learning_rate": 4.4926900321178595e-05,
+      "loss": 0.5208,
+      "step": 1100
+    },
+    {
+      "epoch": 2.0721987810595404,
+      "grad_norm": 1.009045958518982,
+      "learning_rate": 4.488232178183567e-05,
+      "loss": 0.4887,
+      "step": 1105
+    },
+    {
+      "epoch": 2.0815752461322083,
+      "grad_norm": 1.0460408926010132,
+      "learning_rate": 4.483757055848479e-05,
+      "loss": 0.513,
+      "step": 1110
+    },
+    {
+      "epoch": 2.090951711204876,
+      "grad_norm": 1.0761908292770386,
+      "learning_rate": 4.479264703980394e-05,
+      "loss": 0.5304,
+      "step": 1115
+    },
+    {
+      "epoch": 2.1003281762775434,
+      "grad_norm": 1.0238643884658813,
+      "learning_rate": 4.4747551615967534e-05,
+      "loss": 0.538,
+      "step": 1120
+    },
+    {
+      "epoch": 2.109704641350211,
+      "grad_norm": 1.1239683628082275,
+      "learning_rate": 4.470228467864304e-05,
+      "loss": 0.5371,
+      "step": 1125
+    },
+    {
+      "epoch": 2.1190811064228785,
+      "grad_norm": 1.1968084573745728,
+      "learning_rate": 4.4656846620987557e-05,
+      "loss": 0.5467,
+      "step": 1130
+    },
+    {
+      "epoch": 2.128457571495546,
+      "grad_norm": 1.1302961111068726,
+      "learning_rate": 4.461123783764444e-05,
+      "loss": 0.5449,
+      "step": 1135
+    },
+    {
+      "epoch": 2.1378340365682136,
+      "grad_norm": 1.1595929861068726,
+      "learning_rate": 4.4565458724739825e-05,
+      "loss": 0.6024,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1472105016408816,
+      "grad_norm": 1.2119770050048828,
+      "learning_rate": 4.4519509679879226e-05,
+      "loss": 0.5108,
+      "step": 1145
+    },
+    {
+      "epoch": 2.156586966713549,
+      "grad_norm": 1.1276174783706665,
+      "learning_rate": 4.447339110214405e-05,
+      "loss": 0.5024,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1659634317862166,
+      "grad_norm": 1.0845879316329956,
+      "learning_rate": 4.4427103392088185e-05,
+      "loss": 0.5101,
+      "step": 1155
+    },
+    {
+      "epoch": 2.175339896858884,
+      "grad_norm": 1.1071850061416626,
+      "learning_rate": 4.438064695173446e-05,
+      "loss": 0.527,
+      "step": 1160
+    },
+    {
+      "epoch": 2.1847163619315517,
+      "grad_norm": 1.1622778177261353,
+      "learning_rate": 4.433402218457116e-05,
+      "loss": 0.5446,
+      "step": 1165
+    },
+    {
+      "epoch": 2.1940928270042193,
+      "grad_norm": 1.0629466772079468,
+      "learning_rate": 4.428722949554857e-05,
+      "loss": 0.5749,
+      "step": 1170
+    },
+    {
+      "epoch": 2.2034692920768872,
+      "grad_norm": 1.2850778102874756,
+      "learning_rate": 4.424026929107543e-05,
+      "loss": 0.5384,
+      "step": 1175
+    },
+    {
+      "epoch": 2.212845757149555,
+      "grad_norm": 1.1147748231887817,
+      "learning_rate": 4.419314197901537e-05,
+      "loss": 0.5335,
+      "step": 1180
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 1.001845121383667,
+      "learning_rate": 4.4145847968683435e-05,
+      "loss": 0.5318,
+      "step": 1185
+    },
+    {
+      "epoch": 2.23159868729489,
+      "grad_norm": 1.266741394996643,
+      "learning_rate": 4.4098387670842466e-05,
+      "loss": 0.5133,
+      "step": 1190
+    },
+    {
+      "epoch": 2.2409751523675574,
+      "grad_norm": 1.1320396661758423,
+      "learning_rate": 4.405076149769959e-05,
+      "loss": 0.5075,
+      "step": 1195
+    },
+    {
+      "epoch": 2.250351617440225,
+      "grad_norm": 1.1551005840301514,
+      "learning_rate": 4.400296986290258e-05,
+      "loss": 0.5445,
+      "step": 1200
+    },
+    {
+      "epoch": 2.2597280825128925,
+      "grad_norm": 1.1459821462631226,
+      "learning_rate": 4.395501318153632e-05,
+      "loss": 0.5349,
+      "step": 1205
+    },
+    {
+      "epoch": 2.2691045475855605,
+      "grad_norm": 1.0856013298034668,
+      "learning_rate": 4.390689187011917e-05,
+      "loss": 0.569,
+      "step": 1210
+    },
+    {
+      "epoch": 2.278481012658228,
+      "grad_norm": 1.1259822845458984,
+      "learning_rate": 4.385860634659934e-05,
+      "loss": 0.5112,
+      "step": 1215
+    },
+    {
+      "epoch": 2.2878574777308955,
+      "grad_norm": 1.196035385131836,
+      "learning_rate": 4.3810157030351276e-05,
+      "loss": 0.5034,
+      "step": 1220
+    },
+    {
+      "epoch": 2.297233942803563,
+      "grad_norm": 1.0859311819076538,
+      "learning_rate": 4.3761544342172015e-05,
+      "loss": 0.5288,
+      "step": 1225
+    },
+    {
+      "epoch": 2.3066104078762306,
+      "grad_norm": 1.1294561624526978,
+      "learning_rate": 4.371276870427753e-05,
+      "loss": 0.5044,
+      "step": 1230
+    },
+    {
+      "epoch": 2.315986872948898,
+      "grad_norm": 1.164184808731079,
+      "learning_rate": 4.366383054029906e-05,
+      "loss": 0.5372,
+      "step": 1235
+    },
+    {
+      "epoch": 2.3253633380215657,
+      "grad_norm": 1.2330281734466553,
+      "learning_rate": 4.3614730275279457e-05,
+      "loss": 0.5404,
+      "step": 1240
+    },
+    {
+      "epoch": 2.3347398030942337,
+      "grad_norm": 1.1626160144805908,
+      "learning_rate": 4.3565468335669413e-05,
+      "loss": 0.5335,
+      "step": 1245
+    },
+    {
+      "epoch": 2.344116268166901,
+      "grad_norm": 1.0611025094985962,
+      "learning_rate": 4.351604514932387e-05,
+      "loss": 0.5499,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3534927332395688,
+      "grad_norm": 1.1000903844833374,
+      "learning_rate": 4.346646114549822e-05,
+      "loss": 0.5526,
+      "step": 1255
+    },
+    {
+      "epoch": 2.3628691983122363,
+      "grad_norm": 1.0685826539993286,
+      "learning_rate": 4.341671675484459e-05,
+      "loss": 0.5169,
+      "step": 1260
+    },
+    {
+      "epoch": 2.372245663384904,
+      "grad_norm": 1.0964394807815552,
+      "learning_rate": 4.336681240940815e-05,
+      "loss": 0.5258,
+      "step": 1265
+    },
+    {
+      "epoch": 2.3816221284575714,
+      "grad_norm": 1.1604883670806885,
+      "learning_rate": 4.331674854262331e-05,
+      "loss": 0.5842,
+      "step": 1270
+    },
+    {
+      "epoch": 2.390998593530239,
+      "grad_norm": 1.0219110250473022,
+      "learning_rate": 4.326652558930996e-05,
+      "loss": 0.5026,
+      "step": 1275
+    },
+    {
+      "epoch": 2.400375058602907,
+      "grad_norm": 1.116335391998291,
+      "learning_rate": 4.321614398566972e-05,
+      "loss": 0.5736,
+      "step": 1280
+    },
+    {
+      "epoch": 2.4097515236755744,
+      "grad_norm": 1.157214641571045,
+      "learning_rate": 4.316560416928213e-05,
+      "loss": 0.5505,
+      "step": 1285
+    },
+    {
+      "epoch": 2.419127988748242,
+      "grad_norm": 1.011484980583191,
+      "learning_rate": 4.3114906579100853e-05,
+      "loss": 0.5362,
+      "step": 1290
+    },
+    {
+      "epoch": 2.4285044538209095,
+      "grad_norm": 1.1310287714004517,
+      "learning_rate": 4.306405165544988e-05,
+      "loss": 0.4937,
+      "step": 1295
+    },
+    {
+      "epoch": 2.437880918893577,
+      "grad_norm": 1.1262575387954712,
+      "learning_rate": 4.301303984001967e-05,
+      "loss": 0.5365,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4472573839662446,
+      "grad_norm": 1.09992253780365,
+      "learning_rate": 4.296187157586336e-05,
+      "loss": 0.5906,
+      "step": 1305
+    },
+    {
+      "epoch": 2.456633849038912,
+      "grad_norm": 1.2821314334869385,
+      "learning_rate": 4.291054730739286e-05,
+      "loss": 0.552,
+      "step": 1310
+    },
+    {
+      "epoch": 2.46601031411158,
+      "grad_norm": 1.0789141654968262,
+      "learning_rate": 4.2859067480375045e-05,
+      "loss": 0.5239,
+      "step": 1315
+    },
+    {
+      "epoch": 2.4753867791842477,
+      "grad_norm": 1.1485294103622437,
+      "learning_rate": 4.2807432541927865e-05,
+      "loss": 0.5117,
+      "step": 1320
+    },
+    {
+      "epoch": 2.484763244256915,
+      "grad_norm": 1.1945081949234009,
+      "learning_rate": 4.275564294051646e-05,
+      "loss": 0.5448,
+      "step": 1325
+    },
+    {
+      "epoch": 2.4941397093295827,
+      "grad_norm": 1.2146923542022705,
+      "learning_rate": 4.2703699125949245e-05,
+      "loss": 0.5245,
+      "step": 1330
+    },
+    {
+      "epoch": 2.5035161744022503,
+      "grad_norm": 1.0163376331329346,
+      "learning_rate": 4.265160154937404e-05,
+      "loss": 0.5185,
+      "step": 1335
+    },
+    {
+      "epoch": 2.512892639474918,
+      "grad_norm": 1.2290781736373901,
+      "learning_rate": 4.259935066327415e-05,
+      "loss": 0.5493,
+      "step": 1340
+    },
+    {
+      "epoch": 2.5222691045475853,
+      "grad_norm": 1.1765304803848267,
+      "learning_rate": 4.254694692146439e-05,
+      "loss": 0.5021,
+      "step": 1345
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 1.1869114637374878,
+      "learning_rate": 4.2494390779087187e-05,
+      "loss": 0.5017,
+      "step": 1350
+    },
+    {
+      "epoch": 2.541022034692921,
+      "grad_norm": 1.1636326313018799,
+      "learning_rate": 4.2441682692608603e-05,
+      "loss": 0.5441,
+      "step": 1355
+    },
+    {
+      "epoch": 2.5503984997655884,
+      "grad_norm": 1.0827562808990479,
+      "learning_rate": 4.238882311981441e-05,
+      "loss": 0.5431,
+      "step": 1360
+    },
+    {
+      "epoch": 2.559774964838256,
+      "grad_norm": 1.1022346019744873,
+      "learning_rate": 4.233581251980604e-05,
+      "loss": 0.501,
+      "step": 1365
+    },
+    {
+      "epoch": 2.5691514299109235,
+      "grad_norm": 1.1429698467254639,
+      "learning_rate": 4.228265135299669e-05,
+      "loss": 0.5005,
+      "step": 1370
+    },
+    {
+      "epoch": 2.578527894983591,
+      "grad_norm": 1.0954620838165283,
+      "learning_rate": 4.222934008110725e-05,
+      "loss": 0.5459,
+      "step": 1375
+    },
+    {
+      "epoch": 2.5879043600562586,
+      "grad_norm": 1.0992670059204102,
+      "learning_rate": 4.2175879167162304e-05,
+      "loss": 0.5321,
+      "step": 1380
+    },
+    {
+      "epoch": 2.5972808251289266,
+      "grad_norm": 1.1005558967590332,
+      "learning_rate": 4.212226907548614e-05,
+      "loss": 0.5123,
+      "step": 1385
+    },
+    {
+      "epoch": 2.606657290201594,
+      "grad_norm": 1.1369819641113281,
+      "learning_rate": 4.206851027169871e-05,
+      "loss": 0.4994,
+      "step": 1390
+    },
+    {
+      "epoch": 2.6160337552742616,
+      "grad_norm": 1.0780795812606812,
+      "learning_rate": 4.2014603222711576e-05,
+      "loss": 0.5599,
+      "step": 1395
+    },
+    {
+      "epoch": 2.625410220346929,
+      "grad_norm": 1.1751298904418945,
+      "learning_rate": 4.196054839672382e-05,
+      "loss": 0.5245,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6347866854195967,
+      "grad_norm": 1.1803909540176392,
+      "learning_rate": 4.1906346263218044e-05,
+      "loss": 0.5195,
+      "step": 1405
+    },
+    {
+      "epoch": 2.6441631504922647,
+      "grad_norm": 1.1183876991271973,
+      "learning_rate": 4.1851997292956255e-05,
+      "loss": 0.5026,
+      "step": 1410
+    },
+    {
+      "epoch": 2.653539615564932,
+      "grad_norm": 1.1361398696899414,
+      "learning_rate": 4.179750195797578e-05,
+      "loss": 0.5048,
+      "step": 1415
+    },
+    {
+      "epoch": 2.6629160806375998,
+      "grad_norm": 1.2964445352554321,
+      "learning_rate": 4.174286073158516e-05,
+      "loss": 0.5008,
+      "step": 1420
+    },
+    {
+      "epoch": 2.6722925457102673,
+      "grad_norm": 1.1892164945602417,
+      "learning_rate": 4.1688074088360065e-05,
+      "loss": 0.5291,
+      "step": 1425
+    },
+    {
+      "epoch": 2.681669010782935,
+      "grad_norm": 1.292514681816101,
+      "learning_rate": 4.163314250413913e-05,
+      "loss": 0.5285,
+      "step": 1430
+    },
+    {
+      "epoch": 2.6910454758556024,
+      "grad_norm": 1.1883822679519653,
+      "learning_rate": 4.157806645601988e-05,
+      "loss": 0.5208,
+      "step": 1435
+    },
+    {
+      "epoch": 2.70042194092827,
+      "grad_norm": 1.0695146322250366,
+      "learning_rate": 4.152284642235452e-05,
+      "loss": 0.52,
+      "step": 1440
+    },
+    {
+      "epoch": 2.709798406000938,
+      "grad_norm": 1.0255844593048096,
+      "learning_rate": 4.1467482882745835e-05,
+      "loss": 0.5586,
+      "step": 1445
+    },
+    {
+      "epoch": 2.719174871073605,
+      "grad_norm": 1.1997429132461548,
+      "learning_rate": 4.141197631804298e-05,
+      "loss": 0.533,
+      "step": 1450
+    },
+    {
+      "epoch": 2.728551336146273,
+      "grad_norm": 1.1059077978134155,
+      "learning_rate": 4.1356327210337345e-05,
+      "loss": 0.5146,
+      "step": 1455
+    },
+    {
+      "epoch": 2.7379278012189405,
+      "grad_norm": 1.206176519393921,
+      "learning_rate": 4.1300536042958354e-05,
+      "loss": 0.5406,
+      "step": 1460
+    },
+    {
+      "epoch": 2.747304266291608,
+      "grad_norm": 0.9949771165847778,
+      "learning_rate": 4.1244603300469254e-05,
+      "loss": 0.5477,
+      "step": 1465
+    },
+    {
+      "epoch": 2.7566807313642756,
+      "grad_norm": 1.0117969512939453,
+      "learning_rate": 4.118852946866291e-05,
+      "loss": 0.5082,
+      "step": 1470
+    },
+    {
+      "epoch": 2.766057196436943,
+      "grad_norm": 1.1483266353607178,
+      "learning_rate": 4.113231503455758e-05,
+      "loss": 0.5116,
+      "step": 1475
+    },
+    {
+      "epoch": 2.775433661509611,
+      "grad_norm": 1.1810563802719116,
+      "learning_rate": 4.107596048639274e-05,
+      "loss": 0.5229,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7848101265822782,
+      "grad_norm": 1.2007412910461426,
+      "learning_rate": 4.1019466313624746e-05,
+      "loss": 0.5392,
+      "step": 1485
+    },
+    {
+      "epoch": 2.794186591654946,
+      "grad_norm": 1.1348439455032349,
+      "learning_rate": 4.0962833006922675e-05,
+      "loss": 0.535,
+      "step": 1490
+    },
+    {
+      "epoch": 2.8035630567276137,
+      "grad_norm": 1.1027777194976807,
+      "learning_rate": 4.0906061058163995e-05,
+      "loss": 0.5483,
+      "step": 1495
+    },
+    {
+      "epoch": 2.8129395218002813,
+      "grad_norm": 1.1508432626724243,
+      "learning_rate": 4.0849150960430356e-05,
+      "loss": 0.5065,
+      "step": 1500
+    },
+    {
+      "epoch": 2.822315986872949,
+      "grad_norm": 1.0971708297729492,
+      "learning_rate": 4.079210320800324e-05,
+      "loss": 0.4962,
+      "step": 1505
+    },
+    {
+      "epoch": 2.8316924519456164,
+      "grad_norm": 1.1189675331115723,
+      "learning_rate": 4.0734918296359716e-05,
+      "loss": 0.5153,
+      "step": 1510
+    },
+    {
+      "epoch": 2.8410689170182843,
+      "grad_norm": 1.0645800828933716,
+      "learning_rate": 4.0677596722168135e-05,
+      "loss": 0.5478,
+      "step": 1515
+    },
+    {
+      "epoch": 2.8504453820909514,
+      "grad_norm": 1.1716704368591309,
+      "learning_rate": 4.0620138983283785e-05,
+      "loss": 0.5172,
+      "step": 1520
+    },
+    {
+      "epoch": 2.8598218471636194,
+      "grad_norm": 1.1528739929199219,
+      "learning_rate": 4.0562545578744585e-05,
+      "loss": 0.5502,
+      "step": 1525
+    },
+    {
+      "epoch": 2.869198312236287,
+      "grad_norm": 1.069067358970642,
+      "learning_rate": 4.050481700876677e-05,
+      "loss": 0.5052,
+      "step": 1530
+    },
+    {
+      "epoch": 2.8785747773089545,
+      "grad_norm": 1.031403660774231,
+      "learning_rate": 4.044695377474051e-05,
+      "loss": 0.5309,
+      "step": 1535
+    },
+    {
+      "epoch": 2.887951242381622,
+      "grad_norm": 1.043045163154602,
+      "learning_rate": 4.038895637922559e-05,
+      "loss": 0.4996,
+      "step": 1540
+    },
+    {
+      "epoch": 2.8973277074542896,
+      "grad_norm": 1.1269640922546387,
+      "learning_rate": 4.033082532594701e-05,
+      "loss": 0.569,
+      "step": 1545
+    },
+    {
+      "epoch": 2.9067041725269576,
+      "grad_norm": 1.156980037689209,
+      "learning_rate": 4.027256111979063e-05,
+      "loss": 0.5168,
+      "step": 1550
+    },
+    {
+      "epoch": 2.916080637599625,
+      "grad_norm": 1.197581171989441,
+      "learning_rate": 4.021416426679881e-05,
+      "loss": 0.5172,
+      "step": 1555
+    },
+    {
+      "epoch": 2.9254571026722926,
+      "grad_norm": 1.2223148345947266,
+      "learning_rate": 4.015563527416595e-05,
+      "loss": 0.5251,
+      "step": 1560
+    },
+    {
+      "epoch": 2.93483356774496,
+      "grad_norm": 1.0682929754257202,
+      "learning_rate": 4.0096974650234154e-05,
+      "loss": 0.5499,
+      "step": 1565
+    },
+    {
+      "epoch": 2.9442100328176277,
+      "grad_norm": 1.0679572820663452,
+      "learning_rate": 4.003818290448876e-05,
+      "loss": 0.5518,
+      "step": 1570
+    },
+    {
+      "epoch": 2.9535864978902953,
+      "grad_norm": 1.2242721319198608,
+      "learning_rate": 3.9979260547553955e-05,
+      "loss": 0.5367,
+      "step": 1575
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 1.1870614290237427,
+      "learning_rate": 3.992020809118832e-05,
+      "loss": 0.5245,
+      "step": 1580
+    },
+    {
+      "epoch": 2.972339428035631,
+      "grad_norm": 1.1091721057891846,
+      "learning_rate": 3.986102604828038e-05,
+      "loss": 0.4939,
+      "step": 1585
+    },
+    {
+      "epoch": 2.9817158931082983,
+      "grad_norm": 1.1817296743392944,
+      "learning_rate": 3.980171493284418e-05,
+      "loss": 0.5723,
+      "step": 1590
+    },
+    {
+      "epoch": 2.991092358180966,
+      "grad_norm": 1.1225743293762207,
+      "learning_rate": 3.974227526001477e-05,
+      "loss": 0.551,
+      "step": 1595
+    },
+    {
+      "epoch": 3.0004688232536334,
+      "grad_norm": 1.5355123281478882,
+      "learning_rate": 3.9682707546043785e-05,
+      "loss": 0.5302,
+      "step": 1600
+    },
+    {
+      "epoch": 3.009845288326301,
+      "grad_norm": 1.2028840780258179,
+      "learning_rate": 3.9623012308294923e-05,
+      "loss": 0.5514,
+      "step": 1605
+    },
+    {
+      "epoch": 3.0192217533989685,
+      "grad_norm": 1.2248326539993286,
+      "learning_rate": 3.9563190065239474e-05,
+      "loss": 0.5004,
+      "step": 1610
+    },
+    {
+      "epoch": 3.028598218471636,
+      "grad_norm": 1.1607325077056885,
+      "learning_rate": 3.950324133645179e-05,
+      "loss": 0.5292,
+      "step": 1615
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 1.1387099027633667,
+      "learning_rate": 3.9443166642604814e-05,
+      "loss": 0.5275,
+      "step": 1620
+    },
+    {
+      "epoch": 3.0473511486169715,
+      "grad_norm": 0.9971510767936707,
+      "learning_rate": 3.938296650546552e-05,
+      "loss": 0.5223,
+      "step": 1625
+    },
+    {
+      "epoch": 3.056727613689639,
+      "grad_norm": 1.1528209447860718,
+      "learning_rate": 3.932264144789038e-05,
+      "loss": 0.4827,
+      "step": 1630
+    },
+    {
+      "epoch": 3.0661040787623066,
+      "grad_norm": 1.2538106441497803,
+      "learning_rate": 3.9262191993820884e-05,
+      "loss": 0.4863,
+      "step": 1635
+    },
+    {
+      "epoch": 3.075480543834974,
+      "grad_norm": 1.024010419845581,
+      "learning_rate": 3.920161866827889e-05,
+      "loss": 0.5363,
+      "step": 1640
+    },
+    {
+      "epoch": 3.0848570089076417,
+      "grad_norm": 1.1442774534225464,
+      "learning_rate": 3.914092199736217e-05,
+      "loss": 0.4711,
+      "step": 1645
+    },
+    {
+      "epoch": 3.0942334739803092,
+      "grad_norm": 1.2986931800842285,
+      "learning_rate": 3.908010250823972e-05,
+      "loss": 0.5237,
+      "step": 1650
+    },
+    {
+      "epoch": 3.103609939052977,
+      "grad_norm": 1.2176250219345093,
+      "learning_rate": 3.901916072914732e-05,
+      "loss": 0.5261,
+      "step": 1655
+    },
+    {
+      "epoch": 3.1129864041256448,
+      "grad_norm": 1.2821696996688843,
+      "learning_rate": 3.895809718938283e-05,
+      "loss": 0.4975,
+      "step": 1660
+    },
+    {
+      "epoch": 3.1223628691983123,
+      "grad_norm": 1.1697841882705688,
+      "learning_rate": 3.889691241930166e-05,
+      "loss": 0.5053,
+      "step": 1665
+    },
+    {
+      "epoch": 3.13173933427098,
+      "grad_norm": 1.1689724922180176,
+      "learning_rate": 3.883560695031213e-05,
+      "loss": 0.4911,
+      "step": 1670
+    },
+    {
+      "epoch": 3.1411157993436474,
+      "grad_norm": 1.1652592420578003,
+      "learning_rate": 3.8774181314870885e-05,
+      "loss": 0.4567,
+      "step": 1675
+    },
+    {
+      "epoch": 3.150492264416315,
+      "grad_norm": 1.212613821029663,
+      "learning_rate": 3.871263604647822e-05,
+      "loss": 0.5374,
+      "step": 1680
+    },
+    {
+      "epoch": 3.1598687294889825,
+      "grad_norm": 1.272538661956787,
+      "learning_rate": 3.865097167967352e-05,
+      "loss": 0.4974,
+      "step": 1685
+    },
+    {
+      "epoch": 3.1692451945616504,
+      "grad_norm": 1.1123679876327515,
+      "learning_rate": 3.858918875003053e-05,
+      "loss": 0.5184,
+      "step": 1690
+    },
+    {
+      "epoch": 3.178621659634318,
+      "grad_norm": 1.2083110809326172,
+      "learning_rate": 3.8527287794152786e-05,
+      "loss": 0.4811,
+      "step": 1695
+    },
+    {
+      "epoch": 3.1879981247069855,
+      "grad_norm": 1.1418672800064087,
+      "learning_rate": 3.846526934966891e-05,
+      "loss": 0.5137,
+      "step": 1700
+    },
+    {
+      "epoch": 3.197374589779653,
+      "grad_norm": 1.2072086334228516,
+      "learning_rate": 3.840313395522793e-05,
+      "loss": 0.463,
+      "step": 1705
+    },
+    {
+      "epoch": 3.2067510548523206,
+      "grad_norm": 1.3163800239562988,
+      "learning_rate": 3.834088215049464e-05,
+      "loss": 0.5014,
+      "step": 1710
+    },
+    {
+      "epoch": 3.216127519924988,
+      "grad_norm": 1.1487891674041748,
+      "learning_rate": 3.827851447614489e-05,
+      "loss": 0.536,
+      "step": 1715
+    },
+    {
+      "epoch": 3.2255039849976557,
+      "grad_norm": 1.149970293045044,
+      "learning_rate": 3.821603147386088e-05,
+      "loss": 0.5505,
+      "step": 1720
+    },
+    {
+      "epoch": 3.2348804500703237,
+      "grad_norm": 1.3420171737670898,
+      "learning_rate": 3.815343368632648e-05,
+      "loss": 0.5114,
+      "step": 1725
+    },
+    {
+      "epoch": 3.244256915142991,
+      "grad_norm": 1.2148125171661377,
+      "learning_rate": 3.80907216572225e-05,
+      "loss": 0.4733,
+      "step": 1730
+    },
+    {
+      "epoch": 3.2536333802156587,
+      "grad_norm": 1.2376741170883179,
+      "learning_rate": 3.802789593122198e-05,
+      "loss": 0.4584,
+      "step": 1735
+    },
+    {
+      "epoch": 3.2630098452883263,
+      "grad_norm": 1.0415185689926147,
+      "learning_rate": 3.796495705398544e-05,
+      "loss": 0.5423,
+      "step": 1740
+    },
+    {
+      "epoch": 3.272386310360994,
+      "grad_norm": 1.2215274572372437,
+      "learning_rate": 3.790190557215615e-05,
+      "loss": 0.5127,
+      "step": 1745
+    },
+    {
+      "epoch": 3.2817627754336613,
+      "grad_norm": 1.1470167636871338,
+      "learning_rate": 3.783874203335542e-05,
+      "loss": 0.4913,
+      "step": 1750
+    },
+    {
+      "epoch": 3.291139240506329,
+      "grad_norm": 1.1880097389221191,
+      "learning_rate": 3.777546698617776e-05,
+      "loss": 0.5296,
+      "step": 1755
+    },
+    {
+      "epoch": 3.300515705578997,
+      "grad_norm": 1.154988169670105,
+      "learning_rate": 3.77120809801862e-05,
+      "loss": 0.4661,
+      "step": 1760
+    },
+    {
+      "epoch": 3.3098921706516644,
+      "grad_norm": 1.2969125509262085,
+      "learning_rate": 3.764858456590746e-05,
+      "loss": 0.4864,
+      "step": 1765
+    },
+    {
+      "epoch": 3.319268635724332,
+      "grad_norm": 1.122007966041565,
+      "learning_rate": 3.758497829482721e-05,
+      "loss": 0.5037,
+      "step": 1770
+    },
+    {
+      "epoch": 3.3286451007969995,
+      "grad_norm": 1.2779842615127563,
+      "learning_rate": 3.752126271938524e-05,
+      "loss": 0.5081,
+      "step": 1775
+    },
+    {
+      "epoch": 3.338021565869667,
+      "grad_norm": 1.1773457527160645,
+      "learning_rate": 3.7457438392970686e-05,
+      "loss": 0.5379,
+      "step": 1780
+    },
+    {
+      "epoch": 3.3473980309423346,
+      "grad_norm": 1.1523633003234863,
+      "learning_rate": 3.7393505869917224e-05,
+      "loss": 0.4975,
+      "step": 1785
+    },
+    {
+      "epoch": 3.356774496015002,
+      "grad_norm": 1.164303183555603,
+      "learning_rate": 3.732946570549825e-05,
+      "loss": 0.5219,
+      "step": 1790
+    },
+    {
+      "epoch": 3.36615096108767,
+      "grad_norm": 1.123823881149292,
+      "learning_rate": 3.7265318455922057e-05,
+      "loss": 0.5423,
+      "step": 1795
+    },
+    {
+      "epoch": 3.3755274261603376,
+      "grad_norm": 1.2550562620162964,
+      "learning_rate": 3.720106467832701e-05,
+      "loss": 0.5189,
+      "step": 1800
+    },
+    {
+      "epoch": 3.384903891233005,
+      "grad_norm": 1.1335469484329224,
+      "learning_rate": 3.7136704930776686e-05,
+      "loss": 0.5173,
+      "step": 1805
+    },
+    {
+      "epoch": 3.3942803563056727,
+      "grad_norm": 1.2015936374664307,
+      "learning_rate": 3.707223977225507e-05,
+      "loss": 0.5167,
+      "step": 1810
+    },
+    {
+      "epoch": 3.4036568213783402,
+      "grad_norm": 1.1734511852264404,
+      "learning_rate": 3.7007669762661645e-05,
+      "loss": 0.5309,
+      "step": 1815
+    },
+    {
+      "epoch": 3.413033286451008,
+      "grad_norm": 1.1762455701828003,
+      "learning_rate": 3.694299546280657e-05,
+      "loss": 0.5469,
+      "step": 1820
+    },
+    {
+      "epoch": 3.4224097515236753,
+      "grad_norm": 1.1089357137680054,
+      "learning_rate": 3.6878217434405816e-05,
+      "loss": 0.4952,
+      "step": 1825
+    },
+    {
+      "epoch": 3.4317862165963433,
+      "grad_norm": 1.1444052457809448,
+      "learning_rate": 3.681333624007623e-05,
+      "loss": 0.4927,
+      "step": 1830
+    },
+    {
+      "epoch": 3.441162681669011,
+      "grad_norm": 1.0770821571350098,
+      "learning_rate": 3.674835244333071e-05,
+      "loss": 0.5248,
+      "step": 1835
+    },
+    {
+      "epoch": 3.4505391467416784,
+      "grad_norm": 1.1776275634765625,
+      "learning_rate": 3.6683266608573286e-05,
+      "loss": 0.4653,
+      "step": 1840
+    },
+    {
+      "epoch": 3.459915611814346,
+      "grad_norm": 1.1338536739349365,
+      "learning_rate": 3.6618079301094216e-05,
+      "loss": 0.4843,
+      "step": 1845
+    },
+    {
+      "epoch": 3.4692920768870135,
+      "grad_norm": 1.0945082902908325,
+      "learning_rate": 3.6552791087065075e-05,
+      "loss": 0.5148,
+      "step": 1850
+    },
+    {
+      "epoch": 3.4786685419596814,
+      "grad_norm": 1.1242570877075195,
+      "learning_rate": 3.648740253353385e-05,
+      "loss": 0.4963,
+      "step": 1855
+    },
+    {
+      "epoch": 3.488045007032349,
+      "grad_norm": 1.1743115186691284,
+      "learning_rate": 3.642191420842e-05,
+      "loss": 0.4976,
+      "step": 1860
+    },
+    {
+      "epoch": 3.4974214721050165,
+      "grad_norm": 1.1451455354690552,
+      "learning_rate": 3.635632668050954e-05,
+      "loss": 0.4649,
+      "step": 1865
+    },
+    {
+      "epoch": 3.506797937177684,
+      "grad_norm": 1.2638212442398071,
+      "learning_rate": 3.6290640519450074e-05,
+      "loss": 0.512,
+      "step": 1870
+    },
+    {
+      "epoch": 3.5161744022503516,
+      "grad_norm": 1.1948256492614746,
+      "learning_rate": 3.622485629574589e-05,
+      "loss": 0.5213,
+      "step": 1875
+    },
+    {
+      "epoch": 3.525550867323019,
+      "grad_norm": 1.14603590965271,
+      "learning_rate": 3.6158974580752954e-05,
+      "loss": 0.4847,
+      "step": 1880
+    },
+    {
+      "epoch": 3.5349273323956867,
+      "grad_norm": 1.1715166568756104,
+      "learning_rate": 3.6092995946673994e-05,
+      "loss": 0.5436,
+      "step": 1885
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 1.2392544746398926,
+      "learning_rate": 3.60269209665535e-05,
+      "loss": 0.5019,
+      "step": 1890
+    },
+    {
+      "epoch": 3.5536802625410218,
+      "grad_norm": 1.2725554704666138,
+      "learning_rate": 3.596075021427275e-05,
+      "loss": 0.5178,
+      "step": 1895
+    },
+    {
+      "epoch": 3.5630567276136897,
+      "grad_norm": 1.1060048341751099,
+      "learning_rate": 3.589448426454486e-05,
+      "loss": 0.5126,
+      "step": 1900
+    },
+    {
+      "epoch": 3.5724331926863573,
+      "grad_norm": 1.23256516456604,
+      "learning_rate": 3.582812369290972e-05,
+      "loss": 0.5216,
+      "step": 1905
+    },
+    {
+      "epoch": 3.581809657759025,
+      "grad_norm": 1.1210943460464478,
+      "learning_rate": 3.5761669075729084e-05,
+      "loss": 0.5305,
+      "step": 1910
+    },
+    {
+      "epoch": 3.5911861228316924,
+      "grad_norm": 1.0895915031433105,
+      "learning_rate": 3.569512099018149e-05,
+      "loss": 0.5053,
+      "step": 1915
+    },
+    {
+      "epoch": 3.60056258790436,
+      "grad_norm": 1.150411605834961,
+      "learning_rate": 3.562848001425729e-05,
+      "loss": 0.508,
+      "step": 1920
+    },
+    {
+      "epoch": 3.609939052977028,
+      "grad_norm": 1.1784210205078125,
+      "learning_rate": 3.556174672675362e-05,
+      "loss": 0.4804,
+      "step": 1925
+    },
+    {
+      "epoch": 3.6193155180496954,
+      "grad_norm": 1.1341195106506348,
+      "learning_rate": 3.549492170726937e-05,
+      "loss": 0.5413,
+      "step": 1930
+    },
+    {
+      "epoch": 3.628691983122363,
+      "grad_norm": 1.1819273233413696,
+      "learning_rate": 3.5428005536200134e-05,
+      "loss": 0.5451,
+      "step": 1935
+    },
+    {
+      "epoch": 3.6380684481950305,
+      "grad_norm": 1.2183939218521118,
+      "learning_rate": 3.53609987947332e-05,
+      "loss": 0.5362,
+      "step": 1940
+    },
+    {
+      "epoch": 3.647444913267698,
+      "grad_norm": 1.2285958528518677,
+      "learning_rate": 3.5293902064842496e-05,
+      "loss": 0.5377,
+      "step": 1945
+    },
+    {
+      "epoch": 3.6568213783403656,
+      "grad_norm": 1.158765435218811,
+      "learning_rate": 3.5226715929283506e-05,
+      "loss": 0.5314,
+      "step": 1950
+    },
+    {
+      "epoch": 3.666197843413033,
+      "grad_norm": 1.2746514081954956,
+      "learning_rate": 3.515944097158823e-05,
+      "loss": 0.5168,
+      "step": 1955
+    },
+    {
+      "epoch": 3.675574308485701,
+      "grad_norm": 1.2066044807434082,
+      "learning_rate": 3.509207777606013e-05,
+      "loss": 0.527,
+      "step": 1960
+    },
+    {
+      "epoch": 3.6849507735583686,
+      "grad_norm": 1.3743610382080078,
+      "learning_rate": 3.502462692776905e-05,
+      "loss": 0.4951,
+      "step": 1965
+    },
+    {
+      "epoch": 3.694327238631036,
+      "grad_norm": 1.14555823802948,
+      "learning_rate": 3.49570890125461e-05,
+      "loss": 0.5925,
+      "step": 1970
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 1.2276118993759155,
+      "learning_rate": 3.488946461697858e-05,
+      "loss": 0.5096,
+      "step": 1975
+    },
+    {
+      "epoch": 3.7130801687763713,
+      "grad_norm": 1.0634855031967163,
+      "learning_rate": 3.482175432840495e-05,
+      "loss": 0.5127,
+      "step": 1980
+    },
+    {
+      "epoch": 3.722456633849039,
+      "grad_norm": 1.0928593873977661,
+      "learning_rate": 3.4753958734909636e-05,
+      "loss": 0.5134,
+      "step": 1985
+    },
+    {
+      "epoch": 3.7318330989217063,
+      "grad_norm": 1.102906346321106,
+      "learning_rate": 3.468607842531797e-05,
+      "loss": 0.5144,
+      "step": 1990
+    },
+    {
+      "epoch": 3.7412095639943743,
+      "grad_norm": 1.2499363422393799,
+      "learning_rate": 3.461811398919108e-05,
+      "loss": 0.5211,
+      "step": 1995
+    },
+    {
+      "epoch": 3.750586029067042,
+      "grad_norm": 1.1789065599441528,
+      "learning_rate": 3.455006601682075e-05,
+      "loss": 0.5263,
+      "step": 2000
+    },
+    {
+      "epoch": 3.7599624941397094,
+      "grad_norm": 1.1923739910125732,
+      "learning_rate": 3.44819350992243e-05,
+      "loss": 0.5419,
+      "step": 2005
+    },
+    {
+      "epoch": 3.769338959212377,
+      "grad_norm": 1.1823238134384155,
+      "learning_rate": 3.441372182813946e-05,
+      "loss": 0.4625,
+      "step": 2010
+    },
+    {
+      "epoch": 3.7787154242850445,
+      "grad_norm": 1.1649651527404785,
+      "learning_rate": 3.434542679601922e-05,
+      "loss": 0.4627,
+      "step": 2015
+    },
+    {
+      "epoch": 3.788091889357712,
+      "grad_norm": 1.072548747062683,
+      "learning_rate": 3.427705059602671e-05,
+      "loss": 0.5012,
+      "step": 2020
+    },
+    {
+      "epoch": 3.7974683544303796,
+      "grad_norm": 1.261210322380066,
+      "learning_rate": 3.420859382202997e-05,
+      "loss": 0.5192,
+      "step": 2025
+    },
+    {
+      "epoch": 3.8068448195030475,
+      "grad_norm": 1.3214399814605713,
+      "learning_rate": 3.414005706859693e-05,
+      "loss": 0.5196,
+      "step": 2030
+    },
+    {
+      "epoch": 3.816221284575715,
+      "grad_norm": 1.2286287546157837,
+      "learning_rate": 3.407144093099011e-05,
+      "loss": 0.5301,
+      "step": 2035
+    },
+    {
+      "epoch": 3.8255977496483826,
+      "grad_norm": 1.370466947555542,
+      "learning_rate": 3.400274600516152e-05,
+      "loss": 0.494,
+      "step": 2040
+    },
+    {
+      "epoch": 3.83497421472105,
+      "grad_norm": 1.1601370573043823,
+      "learning_rate": 3.393397288774747e-05,
+      "loss": 0.5295,
+      "step": 2045
+    },
+    {
+      "epoch": 3.8443506797937177,
+      "grad_norm": 1.202243685722351,
+      "learning_rate": 3.386512217606339e-05,
+      "loss": 0.474,
+      "step": 2050
+    },
+    {
+      "epoch": 3.8537271448663852,
+      "grad_norm": 1.3419731855392456,
+      "learning_rate": 3.379619446809865e-05,
+      "loss": 0.536,
+      "step": 2055
+    },
+    {
+      "epoch": 3.8631036099390528,
+      "grad_norm": 1.1033244132995605,
+      "learning_rate": 3.372719036251132e-05,
+      "loss": 0.5191,
+      "step": 2060
+    },
+    {
+      "epoch": 3.8724800750117208,
+      "grad_norm": 1.19711172580719,
+      "learning_rate": 3.365811045862305e-05,
+      "loss": 0.4945,
+      "step": 2065
+    },
+    {
+      "epoch": 3.8818565400843883,
+      "grad_norm": 1.1615053415298462,
+      "learning_rate": 3.3588955356413795e-05,
+      "loss": 0.5127,
+      "step": 2070
+    },
+    {
+      "epoch": 3.891233005157056,
+      "grad_norm": 1.1314269304275513,
+      "learning_rate": 3.351972565651664e-05,
+      "loss": 0.558,
+      "step": 2075
+    },
+    {
+      "epoch": 3.9006094702297234,
+      "grad_norm": 1.2228426933288574,
+      "learning_rate": 3.3450421960212566e-05,
+      "loss": 0.5032,
+      "step": 2080
+    },
+    {
+      "epoch": 3.909985935302391,
+      "grad_norm": 1.273769497871399,
+      "learning_rate": 3.338104486942526e-05,
+      "loss": 0.5007,
+      "step": 2085
+    },
+    {
+      "epoch": 3.9193624003750585,
+      "grad_norm": 1.2002629041671753,
+      "learning_rate": 3.3311594986715814e-05,
+      "loss": 0.4874,
+      "step": 2090
+    },
+    {
+      "epoch": 3.928738865447726,
+      "grad_norm": 1.285919427871704,
+      "learning_rate": 3.324207291527758e-05,
+      "loss": 0.53,
+      "step": 2095
+    },
+    {
+      "epoch": 3.938115330520394,
+      "grad_norm": 1.1449804306030273,
+      "learning_rate": 3.317247925893089e-05,
+      "loss": 0.5267,
+      "step": 2100
+    },
+    {
+      "epoch": 3.9474917955930615,
+      "grad_norm": 1.2187373638153076,
+      "learning_rate": 3.31028146221178e-05,
+      "loss": 0.498,
+      "step": 2105
+    },
+    {
+      "epoch": 3.956868260665729,
+      "grad_norm": 1.281288981437683,
+      "learning_rate": 3.3033079609896834e-05,
+      "loss": 0.5103,
+      "step": 2110
+    },
+    {
+      "epoch": 3.9662447257383966,
+      "grad_norm": 1.1786112785339355,
+      "learning_rate": 3.2963274827937794e-05,
+      "loss": 0.4943,
+      "step": 2115
+    },
+    {
+      "epoch": 3.975621190811064,
+      "grad_norm": 1.1950743198394775,
+      "learning_rate": 3.289340088251642e-05,
+      "loss": 0.5308,
+      "step": 2120
+    },
+    {
+      "epoch": 3.9849976558837317,
+      "grad_norm": 1.2143981456756592,
+      "learning_rate": 3.282345838050916e-05,
+      "loss": 0.5103,
+      "step": 2125
+    },
+    {
+      "epoch": 3.994374120956399,
+      "grad_norm": 1.1443520784378052,
+      "learning_rate": 3.275344792938791e-05,
+      "loss": 0.5076,
+      "step": 2130
+    },
+    {
+      "epoch": 4.003750586029067,
+      "grad_norm": 1.1350758075714111,
+      "learning_rate": 3.268337013721472e-05,
+      "loss": 0.4976,
+      "step": 2135
+    },
+    {
+      "epoch": 4.013127051101734,
+      "grad_norm": 1.1434862613677979,
+      "learning_rate": 3.2613225612636525e-05,
+      "loss": 0.5194,
+      "step": 2140
+    },
+    {
+      "epoch": 4.022503516174402,
+      "grad_norm": 1.2293566465377808,
+      "learning_rate": 3.2543014964879816e-05,
+      "loss": 0.5,
+      "step": 2145
+    },
+    {
+      "epoch": 4.03187998124707,
+      "grad_norm": 1.263427495956421,
+      "learning_rate": 3.247273880374542e-05,
+      "loss": 0.513,
+      "step": 2150
+    },
+    {
+      "epoch": 4.041256446319737,
+      "grad_norm": 1.2102352380752563,
+      "learning_rate": 3.240239773960316e-05,
+      "loss": 0.484,
+      "step": 2155
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 1.2546935081481934,
+      "learning_rate": 3.2331992383386566e-05,
+      "loss": 0.4915,
+      "step": 2160
+    },
+    {
+      "epoch": 4.060009376465072,
+      "grad_norm": 1.1183348894119263,
+      "learning_rate": 3.226152334658754e-05,
+      "loss": 0.5173,
+      "step": 2165
+    },
+    {
+      "epoch": 4.06938584153774,
+      "grad_norm": 1.2030504941940308,
+      "learning_rate": 3.21909912412511e-05,
+      "loss": 0.5144,
+      "step": 2170
+    },
+    {
+      "epoch": 4.0787623066104075,
+      "grad_norm": 1.1170116662979126,
+      "learning_rate": 3.2120396679970004e-05,
+      "loss": 0.4967,
+      "step": 2175
+    },
+    {
+      "epoch": 4.0881387716830755,
+      "grad_norm": 1.1743519306182861,
+      "learning_rate": 3.2049740275879493e-05,
+      "loss": 0.4679,
+      "step": 2180
+    },
+    {
+      "epoch": 4.0975152367557435,
+      "grad_norm": 1.2604570388793945,
+      "learning_rate": 3.197902264265189e-05,
+      "loss": 0.4765,
+      "step": 2185
+    },
+    {
+      "epoch": 4.106891701828411,
+      "grad_norm": 1.3032381534576416,
+      "learning_rate": 3.190824439449137e-05,
+      "loss": 0.532,
+      "step": 2190
+    },
+    {
+      "epoch": 4.1162681669010786,
+      "grad_norm": 1.1729470491409302,
+      "learning_rate": 3.1837406146128495e-05,
+      "loss": 0.4973,
+      "step": 2195
+    },
+    {
+      "epoch": 4.125644631973746,
+      "grad_norm": 1.1268198490142822,
+      "learning_rate": 3.176650851281499e-05,
+      "loss": 0.468,
+      "step": 2200
+    },
+    {
+      "epoch": 4.135021097046414,
+      "grad_norm": 1.2962627410888672,
+      "learning_rate": 3.1695552110318375e-05,
+      "loss": 0.4801,
+      "step": 2205
+    },
+    {
+      "epoch": 4.144397562119081,
+      "grad_norm": 1.2923821210861206,
+      "learning_rate": 3.162453755491655e-05,
+      "loss": 0.4751,
+      "step": 2210
+    },
+    {
+      "epoch": 4.153774027191749,
+      "grad_norm": 1.1911282539367676,
+      "learning_rate": 3.155346546339251e-05,
+      "loss": 0.4857,
+      "step": 2215
+    },
+    {
+      "epoch": 4.163150492264417,
+      "grad_norm": 1.2961620092391968,
+      "learning_rate": 3.1482336453028986e-05,
+      "loss": 0.485,
+      "step": 2220
+    },
+    {
+      "epoch": 4.172526957337084,
+      "grad_norm": 1.1935573816299438,
+      "learning_rate": 3.141115114160305e-05,
+      "loss": 0.4817,
+      "step": 2225
+    },
+    {
+      "epoch": 4.181903422409752,
+      "grad_norm": 1.259291172027588,
+      "learning_rate": 3.133991014738076e-05,
+      "loss": 0.5237,
+      "step": 2230
+    },
+    {
+      "epoch": 4.191279887482419,
+      "grad_norm": 1.1934033632278442,
+      "learning_rate": 3.126861408911182e-05,
+      "loss": 0.5139,
+      "step": 2235
+    },
+    {
+      "epoch": 4.200656352555087,
+      "grad_norm": 1.196466088294983,
+      "learning_rate": 3.1197263586024155e-05,
+      "loss": 0.5199,
+      "step": 2240
+    },
+    {
+      "epoch": 4.210032817627754,
+      "grad_norm": 1.0982701778411865,
+      "learning_rate": 3.112585925781858e-05,
+      "loss": 0.4637,
+      "step": 2245
+    },
+    {
+      "epoch": 4.219409282700422,
+      "grad_norm": 1.259890079498291,
+      "learning_rate": 3.105440172466337e-05,
+      "loss": 0.5248,
+      "step": 2250
+    },
+    {
+      "epoch": 4.22878574777309,
+      "grad_norm": 1.1860095262527466,
+      "learning_rate": 3.098289160718895e-05,
+      "loss": 0.4885,
+      "step": 2255
+    },
+    {
+      "epoch": 4.238162212845757,
+      "grad_norm": 1.1539103984832764,
+      "learning_rate": 3.09113295264824e-05,
+      "loss": 0.4893,
+      "step": 2260
+    },
+    {
+      "epoch": 4.247538677918425,
+      "grad_norm": 1.2394139766693115,
+      "learning_rate": 3.083971610408215e-05,
+      "loss": 0.4819,
+      "step": 2265
+    },
+    {
+      "epoch": 4.256915142991092,
+      "grad_norm": 1.1855919361114502,
+      "learning_rate": 3.076805196197255e-05,
+      "loss": 0.5539,
+      "step": 2270
+    },
+    {
+      "epoch": 4.26629160806376,
+      "grad_norm": 1.2501834630966187,
+      "learning_rate": 3.0696337722578444e-05,
+      "loss": 0.4927,
+      "step": 2275
+    },
+    {
+      "epoch": 4.275668073136427,
+      "grad_norm": 1.1885732412338257,
+      "learning_rate": 3.0624574008759805e-05,
+      "loss": 0.4914,
+      "step": 2280
+    },
+    {
+      "epoch": 4.285044538209095,
+      "grad_norm": 1.2143566608428955,
+      "learning_rate": 3.05527614438063e-05,
+      "loss": 0.5062,
+      "step": 2285
+    },
+    {
+      "epoch": 4.294421003281763,
+      "grad_norm": 1.1372582912445068,
+      "learning_rate": 3.0480900651431876e-05,
+      "loss": 0.5349,
+      "step": 2290
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 1.2711280584335327,
+      "learning_rate": 3.0408992255769375e-05,
+      "loss": 0.4706,
+      "step": 2295
+    },
+    {
+      "epoch": 4.313173933427098,
+      "grad_norm": 1.214854121208191,
+      "learning_rate": 3.0337036881365045e-05,
+      "loss": 0.4896,
+      "step": 2300
+    },
+    {
+      "epoch": 4.322550398499765,
+      "grad_norm": 1.1747026443481445,
+      "learning_rate": 3.026503515317321e-05,
+      "loss": 0.4542,
+      "step": 2305
+    },
+    {
+      "epoch": 4.331926863572433,
+      "grad_norm": 1.382872223854065,
+      "learning_rate": 3.0192987696550746e-05,
+      "loss": 0.489,
+      "step": 2310
+    },
+    {
+      "epoch": 4.3413033286451,
+      "grad_norm": 1.1877027750015259,
+      "learning_rate": 3.0120895137251716e-05,
+      "loss": 0.5012,
+      "step": 2315
+    },
+    {
+      "epoch": 4.350679793717768,
+      "grad_norm": 1.1503034830093384,
+      "learning_rate": 3.0048758101421914e-05,
+      "loss": 0.4673,
+      "step": 2320
+    },
+    {
+      "epoch": 4.360056258790436,
+      "grad_norm": 1.1902778148651123,
+      "learning_rate": 2.9976577215593427e-05,
+      "loss": 0.4686,
+      "step": 2325
+    },
+    {
+      "epoch": 4.369432723863103,
+      "grad_norm": 1.1509406566619873,
+      "learning_rate": 2.9904353106679178e-05,
+      "loss": 0.4863,
+      "step": 2330
+    },
+    {
+      "epoch": 4.378809188935771,
+      "grad_norm": 1.2297817468643188,
+      "learning_rate": 2.983208640196751e-05,
+      "loss": 0.4852,
+      "step": 2335
+    },
+    {
+      "epoch": 4.3881856540084385,
+      "grad_norm": 1.140858769416809,
+      "learning_rate": 2.975977772911671e-05,
+      "loss": 0.4491,
+      "step": 2340
+    },
+    {
+      "epoch": 4.3975621190811065,
+      "grad_norm": 1.1405256986618042,
+      "learning_rate": 2.9687427716149603e-05,
+      "loss": 0.4961,
+      "step": 2345
+    },
+    {
+      "epoch": 4.4069385841537745,
+      "grad_norm": 1.1603128910064697,
+      "learning_rate": 2.9615036991448015e-05,
+      "loss": 0.4803,
+      "step": 2350
+    },
+    {
+      "epoch": 4.416315049226442,
+      "grad_norm": 1.2632277011871338,
+      "learning_rate": 2.9542606183747407e-05,
+      "loss": 0.5293,
+      "step": 2355
+    },
+    {
+      "epoch": 4.42569151429911,
+      "grad_norm": 1.3100402355194092,
+      "learning_rate": 2.947013592213137e-05,
+      "loss": 0.4822,
+      "step": 2360
+    },
+    {
+      "epoch": 4.435067979371777,
+      "grad_norm": 1.2090730667114258,
+      "learning_rate": 2.9397626836026133e-05,
+      "loss": 0.5287,
+      "step": 2365
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 1.2461273670196533,
+      "learning_rate": 2.9325079555195163e-05,
+      "loss": 0.4607,
+      "step": 2370
+    },
+    {
+      "epoch": 4.453820909517112,
+      "grad_norm": 1.2689801454544067,
+      "learning_rate": 2.9252494709733635e-05,
+      "loss": 0.4919,
+      "step": 2375
+    },
+    {
+      "epoch": 4.46319737458978,
+      "grad_norm": 1.1873716115951538,
+      "learning_rate": 2.9179872930063e-05,
+      "loss": 0.4835,
+      "step": 2380
+    },
+    {
+      "epoch": 4.472573839662447,
+      "grad_norm": 1.1069796085357666,
+      "learning_rate": 2.9107214846925486e-05,
+      "loss": 0.493,
+      "step": 2385
+    },
+    {
+      "epoch": 4.481950304735115,
+      "grad_norm": 1.200538158416748,
+      "learning_rate": 2.9034521091378635e-05,
+      "loss": 0.4829,
+      "step": 2390
+    },
+    {
+      "epoch": 4.491326769807783,
+      "grad_norm": 1.1566411256790161,
+      "learning_rate": 2.8961792294789796e-05,
+      "loss": 0.5174,
+      "step": 2395
+    },
+    {
+      "epoch": 4.50070323488045,
+      "grad_norm": 1.2605206966400146,
+      "learning_rate": 2.8889029088830686e-05,
+      "loss": 0.4787,
+      "step": 2400
+    },
+    {
+      "epoch": 4.510079699953118,
+      "grad_norm": 1.1897013187408447,
+      "learning_rate": 2.8816232105471863e-05,
+      "loss": 0.4817,
+      "step": 2405
+    },
+    {
+      "epoch": 4.519456165025785,
+      "grad_norm": 1.3637303113937378,
+      "learning_rate": 2.8743401976977257e-05,
+      "loss": 0.5553,
+      "step": 2410
+    },
+    {
+      "epoch": 4.528832630098453,
+      "grad_norm": 1.2789379358291626,
+      "learning_rate": 2.867053933589866e-05,
+      "loss": 0.4798,
+      "step": 2415
+    },
+    {
+      "epoch": 4.538209095171121,
+      "grad_norm": 1.3142014741897583,
+      "learning_rate": 2.8597644815070263e-05,
+      "loss": 0.4799,
+      "step": 2420
+    },
+    {
+      "epoch": 4.547585560243788,
+      "grad_norm": 1.2164462804794312,
+      "learning_rate": 2.8524719047603133e-05,
+      "loss": 0.511,
+      "step": 2425
+    },
+    {
+      "epoch": 4.556962025316456,
+      "grad_norm": 1.3221874237060547,
+      "learning_rate": 2.845176266687974e-05,
+      "loss": 0.5036,
+      "step": 2430
+    },
+    {
+      "epoch": 4.566338490389123,
+      "grad_norm": 1.257280707359314,
+      "learning_rate": 2.83787763065484e-05,
+      "loss": 0.5445,
+      "step": 2435
+    },
+    {
+      "epoch": 4.575714955461791,
+      "grad_norm": 1.1325232982635498,
+      "learning_rate": 2.8305760600517862e-05,
+      "loss": 0.482,
+      "step": 2440
+    },
+    {
+      "epoch": 4.585091420534458,
+      "grad_norm": 1.201951026916504,
+      "learning_rate": 2.8232716182951712e-05,
+      "loss": 0.5115,
+      "step": 2445
+    },
+    {
+      "epoch": 4.594467885607126,
+      "grad_norm": 1.2108012437820435,
+      "learning_rate": 2.815964368826292e-05,
+      "loss": 0.5376,
+      "step": 2450
+    },
+    {
+      "epoch": 4.603844350679793,
+      "grad_norm": 1.1058579683303833,
+      "learning_rate": 2.808654375110832e-05,
+      "loss": 0.4949,
+      "step": 2455
+    },
+    {
+      "epoch": 4.613220815752461,
+      "grad_norm": 1.3397616147994995,
+      "learning_rate": 2.8013417006383076e-05,
+      "loss": 0.4716,
+      "step": 2460
+    },
+    {
+      "epoch": 4.622597280825129,
+      "grad_norm": 1.204738974571228,
+      "learning_rate": 2.7940264089215208e-05,
+      "loss": 0.4845,
+      "step": 2465
+    },
+    {
+      "epoch": 4.631973745897796,
+      "grad_norm": 1.0744824409484863,
+      "learning_rate": 2.7867085634960016e-05,
+      "loss": 0.5383,
+      "step": 2470
+    },
+    {
+      "epoch": 4.641350210970464,
+      "grad_norm": 1.1593892574310303,
+      "learning_rate": 2.779388227919463e-05,
+      "loss": 0.4851,
+      "step": 2475
+    },
+    {
+      "epoch": 4.650726676043131,
+      "grad_norm": 1.2528319358825684,
+      "learning_rate": 2.772065465771244e-05,
+      "loss": 0.4903,
+      "step": 2480
+    },
+    {
+      "epoch": 4.660103141115799,
+      "grad_norm": 1.327526330947876,
+      "learning_rate": 2.7647403406517607e-05,
+      "loss": 0.5113,
+      "step": 2485
+    },
+    {
+      "epoch": 4.669479606188467,
+      "grad_norm": 1.246399164199829,
+      "learning_rate": 2.7574129161819495e-05,
+      "loss": 0.5093,
+      "step": 2490
+    },
+    {
+      "epoch": 4.6788560712611345,
+      "grad_norm": 1.1293491125106812,
+      "learning_rate": 2.7500832560027207e-05,
+      "loss": 0.4943,
+      "step": 2495
+    },
+    {
+      "epoch": 4.688232536333802,
+      "grad_norm": 1.3458096981048584,
+      "learning_rate": 2.7427514237744e-05,
+      "loss": 0.5064,
+      "step": 2500
+    },
+    {
+      "epoch": 4.6976090014064695,
+      "grad_norm": 1.2340059280395508,
+      "learning_rate": 2.7354174831761792e-05,
+      "loss": 0.4965,
+      "step": 2505
+    },
+    {
+      "epoch": 4.7069854664791375,
+      "grad_norm": 1.1967331171035767,
+      "learning_rate": 2.7280814979055612e-05,
+      "loss": 0.5535,
+      "step": 2510
+    },
+    {
+      "epoch": 4.716361931551805,
+      "grad_norm": 1.1821253299713135,
+      "learning_rate": 2.7207435316778087e-05,
+      "loss": 0.4779,
+      "step": 2515
+    },
+    {
+      "epoch": 4.725738396624473,
+      "grad_norm": 1.1687015295028687,
+      "learning_rate": 2.713403648225388e-05,
+      "loss": 0.513,
+      "step": 2520
+    },
+    {
+      "epoch": 4.73511486169714,
+      "grad_norm": 1.223876953125,
+      "learning_rate": 2.7060619112974173e-05,
+      "loss": 0.4993,
+      "step": 2525
+    },
+    {
+      "epoch": 4.744491326769808,
+      "grad_norm": 1.3277308940887451,
+      "learning_rate": 2.698718384659114e-05,
+      "loss": 0.4977,
+      "step": 2530
+    },
+    {
+      "epoch": 4.753867791842476,
+      "grad_norm": 1.2935707569122314,
+      "learning_rate": 2.69137313209124e-05,
+      "loss": 0.4844,
+      "step": 2535
+    },
+    {
+      "epoch": 4.763244256915143,
+      "grad_norm": 1.2303117513656616,
+      "learning_rate": 2.684026217389544e-05,
+      "loss": 0.4829,
+      "step": 2540
+    },
+    {
+      "epoch": 4.772620721987811,
+      "grad_norm": 1.1118587255477905,
+      "learning_rate": 2.6766777043642166e-05,
+      "loss": 0.5146,
+      "step": 2545
+    },
+    {
+      "epoch": 4.781997187060478,
+      "grad_norm": 1.2319899797439575,
+      "learning_rate": 2.6693276568393245e-05,
+      "loss": 0.4873,
+      "step": 2550
+    },
+    {
+      "epoch": 4.791373652133146,
+      "grad_norm": 1.2883573770523071,
+      "learning_rate": 2.6619761386522647e-05,
+      "loss": 0.5248,
+      "step": 2555
+    },
+    {
+      "epoch": 4.800750117205814,
+      "grad_norm": 1.2751914262771606,
+      "learning_rate": 2.6546232136532083e-05,
+      "loss": 0.5243,
+      "step": 2560
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 1.2558832168579102,
+      "learning_rate": 2.647268945704543e-05,
+      "loss": 0.494,
+      "step": 2565
+    },
+    {
+      "epoch": 4.819503047351149,
+      "grad_norm": 1.210188627243042,
+      "learning_rate": 2.639913398680322e-05,
+      "loss": 0.473,
+      "step": 2570
+    },
+    {
+      "epoch": 4.828879512423816,
+      "grad_norm": 1.2350958585739136,
+      "learning_rate": 2.632556636465706e-05,
+      "loss": 0.5097,
+      "step": 2575
+    },
+    {
+      "epoch": 4.838255977496484,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 2.6251987229564123e-05,
+      "loss": 0.5153,
+      "step": 2580
+    },
+    {
+      "epoch": 4.847632442569151,
+      "grad_norm": 1.173352837562561,
+      "learning_rate": 2.6178397220581547e-05,
+      "loss": 0.4543,
+      "step": 2585
+    },
+    {
+      "epoch": 4.857008907641819,
+      "grad_norm": 1.298449158668518,
+      "learning_rate": 2.610479697686093e-05,
+      "loss": 0.4527,
+      "step": 2590
+    },
+    {
+      "epoch": 4.866385372714487,
+      "grad_norm": 1.2860742807388306,
+      "learning_rate": 2.6031187137642766e-05,
+      "loss": 0.4778,
+      "step": 2595
+    },
+    {
+      "epoch": 4.875761837787154,
+      "grad_norm": 1.269826054573059,
+      "learning_rate": 2.595756834225089e-05,
+      "loss": 0.5012,
+      "step": 2600
+    },
+    {
+      "epoch": 4.885138302859822,
+      "grad_norm": 1.2211941480636597,
+      "learning_rate": 2.5883941230086895e-05,
+      "loss": 0.5295,
+      "step": 2605
+    },
+    {
+      "epoch": 4.894514767932489,
+      "grad_norm": 1.2985845804214478,
+      "learning_rate": 2.5810306440624644e-05,
+      "loss": 0.4894,
+      "step": 2610
+    },
+    {
+      "epoch": 4.903891233005157,
+      "grad_norm": 1.2949026823043823,
+      "learning_rate": 2.5736664613404664e-05,
+      "loss": 0.4943,
+      "step": 2615
+    },
+    {
+      "epoch": 4.913267698077824,
+      "grad_norm": 1.2144726514816284,
+      "learning_rate": 2.566301638802861e-05,
+      "loss": 0.4552,
+      "step": 2620
+    },
+    {
+      "epoch": 4.922644163150492,
+      "grad_norm": 1.252120018005371,
+      "learning_rate": 2.558936240415369e-05,
+      "loss": 0.4625,
+      "step": 2625
+    },
+    {
+      "epoch": 4.93202062822316,
+      "grad_norm": 1.2007100582122803,
+      "learning_rate": 2.551570330148716e-05,
+      "loss": 0.502,
+      "step": 2630
+    },
+    {
+      "epoch": 4.941397093295827,
+      "grad_norm": 1.197721004486084,
+      "learning_rate": 2.5442039719780702e-05,
+      "loss": 0.5138,
+      "step": 2635
+    },
+    {
+      "epoch": 4.950773558368495,
+      "grad_norm": 1.1850709915161133,
+      "learning_rate": 2.5368372298824922e-05,
+      "loss": 0.4996,
+      "step": 2640
+    },
+    {
+      "epoch": 4.960150023441162,
+      "grad_norm": 1.2153841257095337,
+      "learning_rate": 2.529470167844376e-05,
+      "loss": 0.4681,
+      "step": 2645
+    },
+    {
+      "epoch": 4.96952648851383,
+      "grad_norm": 1.1117486953735352,
+      "learning_rate": 2.5221028498488947e-05,
+      "loss": 0.4663,
+      "step": 2650
+    },
+    {
+      "epoch": 4.978902953586498,
+      "grad_norm": 1.2405316829681396,
+      "learning_rate": 2.514735339883444e-05,
+      "loss": 0.5197,
+      "step": 2655
+    },
+    {
+      "epoch": 4.9882794186591655,
+      "grad_norm": 1.233486294746399,
+      "learning_rate": 2.507367701937087e-05,
+      "loss": 0.4955,
+      "step": 2660
+    },
+    {
+      "epoch": 4.9976558837318334,
+      "grad_norm": 1.2090688943862915,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5063,
+      "step": 2665
+    },
+    {
+      "epoch": 5.0070323488045005,
+      "grad_norm": 1.1730905771255493,
+      "learning_rate": 2.492632298062913e-05,
+      "loss": 0.4801,
+      "step": 2670
+    },
+    {
+      "epoch": 5.0164088138771685,
+      "grad_norm": 1.3483178615570068,
+      "learning_rate": 2.4852646601165567e-05,
+      "loss": 0.469,
+      "step": 2675
+    },
+    {
+      "epoch": 5.025785278949836,
+      "grad_norm": 1.2244436740875244,
+      "learning_rate": 2.4778971501511063e-05,
+      "loss": 0.4697,
+      "step": 2680
+    },
+    {
+      "epoch": 5.035161744022504,
+      "grad_norm": 1.3525062799453735,
+      "learning_rate": 2.4705298321556248e-05,
+      "loss": 0.4674,
+      "step": 2685
+    },
+    {
+      "epoch": 5.044538209095171,
+      "grad_norm": 1.188866376876831,
+      "learning_rate": 2.4631627701175084e-05,
+      "loss": 0.4627,
+      "step": 2690
+    },
+    {
+      "epoch": 5.053914674167839,
+      "grad_norm": 1.3100556135177612,
+      "learning_rate": 2.4557960280219297e-05,
+      "loss": 0.4592,
+      "step": 2695
+    },
+    {
+      "epoch": 5.063291139240507,
+      "grad_norm": 1.2546091079711914,
+      "learning_rate": 2.448429669851285e-05,
+      "loss": 0.4903,
+      "step": 2700
+    },
+    {
+      "epoch": 5.072667604313174,
+      "grad_norm": 1.3357013463974,
+      "learning_rate": 2.4410637595846317e-05,
+      "loss": 0.5137,
+      "step": 2705
+    },
+    {
+      "epoch": 5.082044069385842,
+      "grad_norm": 1.1900633573532104,
+      "learning_rate": 2.43369836119714e-05,
+      "loss": 0.4808,
+      "step": 2710
+    },
+    {
+      "epoch": 5.091420534458509,
+      "grad_norm": 1.2639002799987793,
+      "learning_rate": 2.426333538659534e-05,
+      "loss": 0.4783,
+      "step": 2715
+    },
+    {
+      "epoch": 5.100796999531177,
+      "grad_norm": 1.219259262084961,
+      "learning_rate": 2.4189693559375365e-05,
+      "loss": 0.4733,
+      "step": 2720
+    },
+    {
+      "epoch": 5.110173464603844,
+      "grad_norm": 1.367392659187317,
+      "learning_rate": 2.4116058769913108e-05,
+      "loss": 0.468,
+      "step": 2725
+    },
+    {
+      "epoch": 5.119549929676512,
+      "grad_norm": 1.185295820236206,
+      "learning_rate": 2.4042431657749117e-05,
+      "loss": 0.453,
+      "step": 2730
+    },
+    {
+      "epoch": 5.12892639474918,
+      "grad_norm": 1.2299050092697144,
+      "learning_rate": 2.3968812862357233e-05,
+      "loss": 0.4812,
+      "step": 2735
+    },
+    {
+      "epoch": 5.138302859821847,
+      "grad_norm": 1.297749400138855,
+      "learning_rate": 2.3895203023139073e-05,
+      "loss": 0.4931,
+      "step": 2740
+    },
+    {
+      "epoch": 5.147679324894515,
+      "grad_norm": 1.3923060894012451,
+      "learning_rate": 2.3821602779418462e-05,
+      "loss": 0.5565,
+      "step": 2745
+    },
+    {
+      "epoch": 5.157055789967182,
+      "grad_norm": 1.205780267715454,
+      "learning_rate": 2.3748012770435883e-05,
+      "loss": 0.4738,
+      "step": 2750
+    },
+    {
+      "epoch": 5.16643225503985,
+      "grad_norm": 1.2132831811904907,
+      "learning_rate": 2.3674433635342942e-05,
+      "loss": 0.4735,
+      "step": 2755
+    },
+    {
+      "epoch": 5.175808720112517,
+      "grad_norm": 1.2221248149871826,
+      "learning_rate": 2.3600866013196787e-05,
+      "loss": 0.498,
+      "step": 2760
+    },
+    {
+      "epoch": 5.185185185185185,
+      "grad_norm": 1.336977243423462,
+      "learning_rate": 2.3527310542954575e-05,
+      "loss": 0.458,
+      "step": 2765
+    },
+    {
+      "epoch": 5.194561650257853,
+      "grad_norm": 1.2728416919708252,
+      "learning_rate": 2.3453767863467923e-05,
+      "loss": 0.5213,
+      "step": 2770
+    },
+    {
+      "epoch": 5.20393811533052,
+      "grad_norm": 1.2689812183380127,
+      "learning_rate": 2.3380238613477355e-05,
+      "loss": 0.5083,
+      "step": 2775
+    },
+    {
+      "epoch": 5.213314580403188,
+      "grad_norm": 1.3632757663726807,
+      "learning_rate": 2.3306723431606758e-05,
+      "loss": 0.4538,
+      "step": 2780
+    },
+    {
+      "epoch": 5.222691045475855,
+      "grad_norm": 1.2548248767852783,
+      "learning_rate": 2.3233222956357846e-05,
+      "loss": 0.4885,
+      "step": 2785
+    },
+    {
+      "epoch": 5.232067510548523,
+      "grad_norm": 1.2769088745117188,
+      "learning_rate": 2.3159737826104565e-05,
+      "loss": 0.481,
+      "step": 2790
+    },
+    {
+      "epoch": 5.241443975621191,
+      "grad_norm": 1.1989548206329346,
+      "learning_rate": 2.3086268679087607e-05,
+      "loss": 0.4823,
+      "step": 2795
+    },
+    {
+      "epoch": 5.250820440693858,
+      "grad_norm": 1.2793382406234741,
+      "learning_rate": 2.3012816153408863e-05,
+      "loss": 0.4658,
+      "step": 2800
+    },
+    {
+      "epoch": 5.260196905766526,
+      "grad_norm": 1.2584115266799927,
+      "learning_rate": 2.293938088702584e-05,
+      "loss": 0.4903,
+      "step": 2805
+    },
+    {
+      "epoch": 5.269573370839193,
+      "grad_norm": 1.2669495344161987,
+      "learning_rate": 2.286596351774613e-05,
+      "loss": 0.4763,
+      "step": 2810
+    },
+    {
+      "epoch": 5.278949835911861,
+      "grad_norm": 1.3669885396957397,
+      "learning_rate": 2.2792564683221916e-05,
+      "loss": 0.525,
+      "step": 2815
+    },
+    {
+      "epoch": 5.2883263009845285,
+      "grad_norm": 1.139404535293579,
+      "learning_rate": 2.271918502094439e-05,
+      "loss": 0.4847,
+      "step": 2820
+    },
+    {
+      "epoch": 5.2977027660571965,
+      "grad_norm": 1.2239021062850952,
+      "learning_rate": 2.2645825168238214e-05,
+      "loss": 0.4677,
+      "step": 2825
+    },
+    {
+      "epoch": 5.307079231129864,
+      "grad_norm": 1.437768816947937,
+      "learning_rate": 2.2572485762256005e-05,
+      "loss": 0.4657,
+      "step": 2830
+    },
+    {
+      "epoch": 5.3164556962025316,
+      "grad_norm": 1.2063370943069458,
+      "learning_rate": 2.2499167439972795e-05,
+      "loss": 0.4795,
+      "step": 2835
+    },
+    {
+      "epoch": 5.3258321612751995,
+      "grad_norm": 1.3254681825637817,
+      "learning_rate": 2.2425870838180507e-05,
+      "loss": 0.4799,
+      "step": 2840
+    },
+    {
+      "epoch": 5.335208626347867,
+      "grad_norm": 1.3319010734558105,
+      "learning_rate": 2.2352596593482405e-05,
+      "loss": 0.4756,
+      "step": 2845
+    },
+    {
+      "epoch": 5.344585091420535,
+      "grad_norm": 1.3138788938522339,
+      "learning_rate": 2.2279345342287567e-05,
+      "loss": 0.4777,
+      "step": 2850
+    },
+    {
+      "epoch": 5.353961556493202,
+      "grad_norm": 1.1732454299926758,
+      "learning_rate": 2.2206117720805377e-05,
+      "loss": 0.4928,
+      "step": 2855
+    },
+    {
+      "epoch": 5.36333802156587,
+      "grad_norm": 1.300582766532898,
+      "learning_rate": 2.2132914365039993e-05,
+      "loss": 0.506,
+      "step": 2860
+    },
+    {
+      "epoch": 5.372714486638538,
+      "grad_norm": 1.1914080381393433,
+      "learning_rate": 2.2059735910784808e-05,
+      "loss": 0.4698,
+      "step": 2865
+    },
+    {
+      "epoch": 5.382090951711205,
+      "grad_norm": 1.2935923337936401,
+      "learning_rate": 2.1986582993616926e-05,
+      "loss": 0.4904,
+      "step": 2870
+    },
+    {
+      "epoch": 5.391467416783873,
+      "grad_norm": 1.3232049942016602,
+      "learning_rate": 2.1913456248891686e-05,
+      "loss": 0.497,
+      "step": 2875
+    },
+    {
+      "epoch": 5.40084388185654,
+      "grad_norm": 1.2711055278778076,
+      "learning_rate": 2.1840356311737084e-05,
+      "loss": 0.4529,
+      "step": 2880
+    },
+    {
+      "epoch": 5.410220346929208,
+      "grad_norm": 1.3390828371047974,
+      "learning_rate": 2.176728381704829e-05,
+      "loss": 0.5213,
+      "step": 2885
+    },
+    {
+      "epoch": 5.419596812001875,
+      "grad_norm": 1.4285824298858643,
+      "learning_rate": 2.169423939948215e-05,
+      "loss": 0.5365,
+      "step": 2890
+    },
+    {
+      "epoch": 5.428973277074543,
+      "grad_norm": 1.3551216125488281,
+      "learning_rate": 2.162122369345161e-05,
+      "loss": 0.4519,
+      "step": 2895
+    },
+    {
+      "epoch": 5.438349742147211,
+      "grad_norm": 1.3975772857666016,
+      "learning_rate": 2.154823733312027e-05,
+      "loss": 0.4738,
+      "step": 2900
+    },
+    {
+      "epoch": 5.447726207219878,
+      "grad_norm": 1.330074429512024,
+      "learning_rate": 2.1475280952396866e-05,
+      "loss": 0.453,
+      "step": 2905
+    },
+    {
+      "epoch": 5.457102672292546,
+      "grad_norm": 1.3424428701400757,
+      "learning_rate": 2.140235518492975e-05,
+      "loss": 0.4801,
+      "step": 2910
+    },
+    {
+      "epoch": 5.466479137365213,
+      "grad_norm": 1.3250641822814941,
+      "learning_rate": 2.1329460664101346e-05,
+      "loss": 0.4819,
+      "step": 2915
+    },
+    {
+      "epoch": 5.475855602437881,
+      "grad_norm": 1.2409799098968506,
+      "learning_rate": 2.125659802302275e-05,
+      "loss": 0.4713,
+      "step": 2920
+    },
+    {
+      "epoch": 5.485232067510548,
+      "grad_norm": 1.2620609998703003,
+      "learning_rate": 2.1183767894528136e-05,
+      "loss": 0.5022,
+      "step": 2925
+    },
+    {
+      "epoch": 5.494608532583216,
+      "grad_norm": 1.2690008878707886,
+      "learning_rate": 2.1110970911169316e-05,
+      "loss": 0.436,
+      "step": 2930
+    },
+    {
+      "epoch": 5.503984997655884,
+      "grad_norm": 1.26837956905365,
+      "learning_rate": 2.1038207705210206e-05,
+      "loss": 0.4927,
+      "step": 2935
+    },
+    {
+      "epoch": 5.513361462728551,
+      "grad_norm": 1.2761749029159546,
+      "learning_rate": 2.096547890862137e-05,
+      "loss": 0.4926,
+      "step": 2940
+    },
+    {
+      "epoch": 5.522737927801219,
+      "grad_norm": 1.2363426685333252,
+      "learning_rate": 2.0892785153074517e-05,
+      "loss": 0.5321,
+      "step": 2945
+    },
+    {
+      "epoch": 5.532114392873886,
+      "grad_norm": 1.2713450193405151,
+      "learning_rate": 2.0820127069937008e-05,
+      "loss": 0.4828,
+      "step": 2950
+    },
+    {
+      "epoch": 5.541490857946554,
+      "grad_norm": 1.2500605583190918,
+      "learning_rate": 2.074750529026637e-05,
+      "loss": 0.4816,
+      "step": 2955
+    },
+    {
+      "epoch": 5.550867323019221,
+      "grad_norm": 1.3841161727905273,
+      "learning_rate": 2.0674920444804847e-05,
+      "loss": 0.492,
+      "step": 2960
+    },
+    {
+      "epoch": 5.560243788091889,
+      "grad_norm": 1.3208250999450684,
+      "learning_rate": 2.060237316397387e-05,
+      "loss": 0.4834,
+      "step": 2965
+    },
+    {
+      "epoch": 5.569620253164557,
+      "grad_norm": 1.097853660583496,
+      "learning_rate": 2.0529864077868643e-05,
+      "loss": 0.4657,
+      "step": 2970
+    },
+    {
+      "epoch": 5.578996718237224,
+      "grad_norm": 1.1769808530807495,
+      "learning_rate": 2.0457393816252596e-05,
+      "loss": 0.492,
+      "step": 2975
+    },
+    {
+      "epoch": 5.588373183309892,
+      "grad_norm": 1.2151196002960205,
+      "learning_rate": 2.0384963008551995e-05,
+      "loss": 0.5049,
+      "step": 2980
+    },
+    {
+      "epoch": 5.5977496483825595,
+      "grad_norm": 1.2380439043045044,
+      "learning_rate": 2.0312572283850403e-05,
+      "loss": 0.4933,
+      "step": 2985
+    },
+    {
+      "epoch": 5.6071261134552275,
+      "grad_norm": 1.3202080726623535,
+      "learning_rate": 2.0240222270883288e-05,
+      "loss": 0.5056,
+      "step": 2990
+    },
+    {
+      "epoch": 5.616502578527895,
+      "grad_norm": 1.2747589349746704,
+      "learning_rate": 2.0167913598032507e-05,
+      "loss": 0.5095,
+      "step": 2995
+    },
+    {
+      "epoch": 5.625879043600563,
+      "grad_norm": 1.3257839679718018,
+      "learning_rate": 2.0095646893320828e-05,
+      "loss": 0.4526,
+      "step": 3000
+    },
+    {
+      "epoch": 5.6352555086732306,
+      "grad_norm": 1.2410848140716553,
+      "learning_rate": 2.0023422784406576e-05,
+      "loss": 0.517,
+      "step": 3005
+    },
+    {
+      "epoch": 5.644631973745898,
+      "grad_norm": 1.2765707969665527,
+      "learning_rate": 1.9951241898578085e-05,
+      "loss": 0.4845,
+      "step": 3010
+    },
+    {
+      "epoch": 5.654008438818566,
+      "grad_norm": 1.2771388292312622,
+      "learning_rate": 1.9879104862748287e-05,
+      "loss": 0.481,
+      "step": 3015
+    },
+    {
+      "epoch": 5.663384903891233,
+      "grad_norm": 1.1813123226165771,
+      "learning_rate": 1.980701230344926e-05,
+      "loss": 0.4779,
+      "step": 3020
+    },
+    {
+      "epoch": 5.672761368963901,
+      "grad_norm": 1.1853107213974,
+      "learning_rate": 1.9734964846826794e-05,
+      "loss": 0.4695,
+      "step": 3025
+    },
+    {
+      "epoch": 5.682137834036569,
+      "grad_norm": 1.3022593259811401,
+      "learning_rate": 1.9662963118634954e-05,
+      "loss": 0.4825,
+      "step": 3030
+    },
+    {
+      "epoch": 5.691514299109236,
+      "grad_norm": 1.24039626121521,
+      "learning_rate": 1.9591007744230637e-05,
+      "loss": 0.447,
+      "step": 3035
+    },
+    {
+      "epoch": 5.700890764181904,
+      "grad_norm": 1.204289197921753,
+      "learning_rate": 1.9519099348568127e-05,
+      "loss": 0.5317,
+      "step": 3040
+    },
+    {
+      "epoch": 5.710267229254571,
+      "grad_norm": 1.3570585250854492,
+      "learning_rate": 1.944723855619371e-05,
+      "loss": 0.4577,
+      "step": 3045
+    },
+    {
+      "epoch": 5.719643694327239,
+      "grad_norm": 1.3854038715362549,
+      "learning_rate": 1.93754259912402e-05,
+      "loss": 0.4658,
+      "step": 3050
+    },
+    {
+      "epoch": 5.729020159399906,
+      "grad_norm": 1.3811026811599731,
+      "learning_rate": 1.9303662277421568e-05,
+      "loss": 0.4575,
+      "step": 3055
+    },
+    {
+      "epoch": 5.738396624472574,
+      "grad_norm": 1.3897615671157837,
+      "learning_rate": 1.9231948038027462e-05,
+      "loss": 0.5074,
+      "step": 3060
+    },
+    {
+      "epoch": 5.747773089545241,
+      "grad_norm": 1.433873176574707,
+      "learning_rate": 1.9160283895917855e-05,
+      "loss": 0.5143,
+      "step": 3065
+    },
+    {
+      "epoch": 5.757149554617909,
+      "grad_norm": 1.3998658657073975,
+      "learning_rate": 1.9088670473517605e-05,
+      "loss": 0.4909,
+      "step": 3070
+    },
+    {
+      "epoch": 5.766526019690577,
+      "grad_norm": 1.3678226470947266,
+      "learning_rate": 1.9017108392811065e-05,
+      "loss": 0.4833,
+      "step": 3075
+    },
+    {
+      "epoch": 5.775902484763244,
+      "grad_norm": 1.3105368614196777,
+      "learning_rate": 1.8945598275336633e-05,
+      "loss": 0.4889,
+      "step": 3080
+    },
+    {
+      "epoch": 5.785278949835912,
+      "grad_norm": 1.2549711465835571,
+      "learning_rate": 1.8874140742181424e-05,
+      "loss": 0.4632,
+      "step": 3085
+    },
+    {
+      "epoch": 5.794655414908579,
+      "grad_norm": 1.233004093170166,
+      "learning_rate": 1.8802736413975844e-05,
+      "loss": 0.4631,
+      "step": 3090
+    },
+    {
+      "epoch": 5.804031879981247,
+      "grad_norm": 1.2922793626785278,
+      "learning_rate": 1.8731385910888177e-05,
+      "loss": 0.4995,
+      "step": 3095
+    },
+    {
+      "epoch": 5.813408345053915,
+      "grad_norm": 1.2658621072769165,
+      "learning_rate": 1.866008985261924e-05,
+      "loss": 0.4892,
+      "step": 3100
+    },
+    {
+      "epoch": 5.822784810126582,
+      "grad_norm": 1.3431642055511475,
+      "learning_rate": 1.8588848858396957e-05,
+      "loss": 0.5213,
+      "step": 3105
+    },
+    {
+      "epoch": 5.83216127519925,
+      "grad_norm": 1.3486206531524658,
+      "learning_rate": 1.8517663546971013e-05,
+      "loss": 0.4488,
+      "step": 3110
+    },
+    {
+      "epoch": 5.841537740271917,
+      "grad_norm": 1.5124202966690063,
+      "learning_rate": 1.844653453660749e-05,
+      "loss": 0.485,
+      "step": 3115
+    },
+    {
+      "epoch": 5.850914205344585,
+      "grad_norm": 1.310178518295288,
+      "learning_rate": 1.8375462445083464e-05,
+      "loss": 0.4959,
+      "step": 3120
+    },
+    {
+      "epoch": 5.860290670417252,
+      "grad_norm": 1.4783662557601929,
+      "learning_rate": 1.830444788968163e-05,
+      "loss": 0.4824,
+      "step": 3125
+    },
+    {
+      "epoch": 5.86966713548992,
+      "grad_norm": 1.2521761655807495,
+      "learning_rate": 1.8233491487185006e-05,
+      "loss": 0.4672,
+      "step": 3130
+    },
+    {
+      "epoch": 5.8790436005625875,
+      "grad_norm": 1.256402850151062,
+      "learning_rate": 1.8162593853871514e-05,
+      "loss": 0.4611,
+      "step": 3135
+    },
+    {
+      "epoch": 5.888420065635255,
+      "grad_norm": 1.3934487104415894,
+      "learning_rate": 1.8091755605508643e-05,
+      "loss": 0.4362,
+      "step": 3140
+    },
+    {
+      "epoch": 5.897796530707923,
+      "grad_norm": 1.394781470298767,
+      "learning_rate": 1.8020977357348106e-05,
+      "loss": 0.4675,
+      "step": 3145
+    },
+    {
+      "epoch": 5.9071729957805905,
+      "grad_norm": 1.2938883304595947,
+      "learning_rate": 1.7950259724120512e-05,
+      "loss": 0.4997,
+      "step": 3150
+    },
+    {
+      "epoch": 5.9165494608532585,
+      "grad_norm": 1.2647290229797363,
+      "learning_rate": 1.7879603320029992e-05,
+      "loss": 0.4983,
+      "step": 3155
+    },
+    {
+      "epoch": 5.925925925925926,
+      "grad_norm": 1.269392967224121,
+      "learning_rate": 1.7809008758748913e-05,
+      "loss": 0.4957,
+      "step": 3160
+    },
+    {
+      "epoch": 5.935302390998594,
+      "grad_norm": 1.3376566171646118,
+      "learning_rate": 1.7738476653412468e-05,
+      "loss": 0.4618,
+      "step": 3165
+    },
+    {
+      "epoch": 5.944678856071262,
+      "grad_norm": 1.2967661619186401,
+      "learning_rate": 1.766800761661344e-05,
+      "loss": 0.5211,
+      "step": 3170
+    },
+    {
+      "epoch": 5.954055321143929,
+      "grad_norm": 1.3205426931381226,
+      "learning_rate": 1.759760226039684e-05,
+      "loss": 0.5053,
+      "step": 3175
+    },
+    {
+      "epoch": 5.963431786216597,
+      "grad_norm": 1.259722352027893,
+      "learning_rate": 1.752726119625459e-05,
+      "loss": 0.4874,
+      "step": 3180
+    },
+    {
+      "epoch": 5.972808251289264,
+      "grad_norm": 1.304119348526001,
+      "learning_rate": 1.7456985035120193e-05,
+      "loss": 0.4907,
+      "step": 3185
+    },
+    {
+      "epoch": 5.982184716361932,
+      "grad_norm": 1.2843273878097534,
+      "learning_rate": 1.7386774387363484e-05,
+      "loss": 0.5289,
+      "step": 3190
+    },
+    {
+      "epoch": 5.991561181434599,
+      "grad_norm": 1.2098652124404907,
+      "learning_rate": 1.731662986278528e-05,
+      "loss": 0.4907,
+      "step": 3195
+    },
+    {
+      "epoch": 6.000937646507267,
+      "grad_norm": 1.4787734746932983,
+      "learning_rate": 1.724655207061209e-05,
+      "loss": 0.4716,
+      "step": 3200
+    },
+    {
+      "epoch": 6.010314111579935,
+      "grad_norm": 1.2539989948272705,
+      "learning_rate": 1.7176541619490847e-05,
+      "loss": 0.4719,
+      "step": 3205
+    },
+    {
+      "epoch": 6.019690576652602,
+      "grad_norm": 1.2070831060409546,
+      "learning_rate": 1.710659911748359e-05,
+      "loss": 0.4488,
+      "step": 3210
+    },
+    {
+      "epoch": 6.02906704172527,
+      "grad_norm": 1.1978720426559448,
+      "learning_rate": 1.703672517206221e-05,
+      "loss": 0.4868,
+      "step": 3215
+    },
+    {
+      "epoch": 6.038443506797937,
+      "grad_norm": 1.1691241264343262,
+      "learning_rate": 1.696692039010317e-05,
+      "loss": 0.5243,
+      "step": 3220
+    },
+    {
+      "epoch": 6.047819971870605,
+      "grad_norm": 1.2439446449279785,
+      "learning_rate": 1.6897185377882215e-05,
+      "loss": 0.4689,
+      "step": 3225
+    },
+    {
+      "epoch": 6.057196436943272,
+      "grad_norm": 1.3329085111618042,
+      "learning_rate": 1.6827520741069118e-05,
+      "loss": 0.4991,
+      "step": 3230
+    },
+    {
+      "epoch": 6.06657290201594,
+      "grad_norm": 1.346922516822815,
+      "learning_rate": 1.6757927084722426e-05,
+      "loss": 0.4646,
+      "step": 3235
+    },
+    {
+      "epoch": 6.075949367088608,
+      "grad_norm": 1.3174694776535034,
+      "learning_rate": 1.6688405013284192e-05,
+      "loss": 0.4256,
+      "step": 3240
+    },
+    {
+      "epoch": 6.085325832161275,
+      "grad_norm": 1.360258936882019,
+      "learning_rate": 1.6618955130574755e-05,
+      "loss": 0.4901,
+      "step": 3245
+    },
+    {
+      "epoch": 6.094702297233943,
+      "grad_norm": 1.352583646774292,
+      "learning_rate": 1.6549578039787436e-05,
+      "loss": 0.5018,
+      "step": 3250
+    },
+    {
+      "epoch": 6.10407876230661,
+      "grad_norm": 1.3360068798065186,
+      "learning_rate": 1.6480274343483364e-05,
+      "loss": 0.4922,
+      "step": 3255
+    },
+    {
+      "epoch": 6.113455227379278,
+      "grad_norm": 1.3001676797866821,
+      "learning_rate": 1.6411044643586204e-05,
+      "loss": 0.4572,
+      "step": 3260
+    },
+    {
+      "epoch": 6.122831692451945,
+      "grad_norm": 1.2709695100784302,
+      "learning_rate": 1.6341889541376958e-05,
+      "loss": 0.5083,
+      "step": 3265
+    },
+    {
+      "epoch": 6.132208157524613,
+      "grad_norm": 1.4063165187835693,
+      "learning_rate": 1.627280963748869e-05,
+      "loss": 0.4781,
+      "step": 3270
+    },
+    {
+      "epoch": 6.141584622597281,
+      "grad_norm": 1.299294114112854,
+      "learning_rate": 1.6203805531901355e-05,
+      "loss": 0.4825,
+      "step": 3275
+    },
+    {
+      "epoch": 6.150961087669948,
+      "grad_norm": 1.3920867443084717,
+      "learning_rate": 1.613487782393661e-05,
+      "loss": 0.4693,
+      "step": 3280
+    },
+    {
+      "epoch": 6.160337552742616,
+      "grad_norm": 1.3572577238082886,
+      "learning_rate": 1.6066027112252542e-05,
+      "loss": 0.4759,
+      "step": 3285
+    },
+    {
+      "epoch": 6.169714017815283,
+      "grad_norm": 1.3963364362716675,
+      "learning_rate": 1.5997253994838484e-05,
+      "loss": 0.4976,
+      "step": 3290
+    },
+    {
+      "epoch": 6.179090482887951,
+      "grad_norm": 1.2881590127944946,
+      "learning_rate": 1.5928559069009895e-05,
+      "loss": 0.4738,
+      "step": 3295
+    },
+    {
+      "epoch": 6.1884669479606185,
+      "grad_norm": 1.4318541288375854,
+      "learning_rate": 1.5859942931403072e-05,
+      "loss": 0.4884,
+      "step": 3300
+    },
+    {
+      "epoch": 6.1978434130332865,
+      "grad_norm": 1.3195756673812866,
+      "learning_rate": 1.579140617797003e-05,
+      "loss": 0.4881,
+      "step": 3305
+    },
+    {
+      "epoch": 6.207219878105954,
+      "grad_norm": 1.3620742559432983,
+      "learning_rate": 1.5722949403973308e-05,
+      "loss": 0.4649,
+      "step": 3310
+    },
+    {
+      "epoch": 6.2165963431786215,
+      "grad_norm": 1.2172105312347412,
+      "learning_rate": 1.5654573203980784e-05,
+      "loss": 0.4996,
+      "step": 3315
+    },
+    {
+      "epoch": 6.2259728082512895,
+      "grad_norm": 1.1952767372131348,
+      "learning_rate": 1.5586278171860546e-05,
+      "loss": 0.4695,
+      "step": 3320
+    },
+    {
+      "epoch": 6.235349273323957,
+      "grad_norm": 1.3092693090438843,
+      "learning_rate": 1.55180649007757e-05,
+      "loss": 0.4603,
+      "step": 3325
+    },
+    {
+      "epoch": 6.244725738396625,
+      "grad_norm": 1.4018226861953735,
+      "learning_rate": 1.5449933983179256e-05,
+      "loss": 0.4632,
+      "step": 3330
+    },
+    {
+      "epoch": 6.254102203469292,
+      "grad_norm": 1.454005479812622,
+      "learning_rate": 1.5381886010808927e-05,
+      "loss": 0.5189,
+      "step": 3335
+    },
+    {
+      "epoch": 6.26347866854196,
+      "grad_norm": 1.30441415309906,
+      "learning_rate": 1.5313921574682032e-05,
+      "loss": 0.4877,
+      "step": 3340
+    },
+    {
+      "epoch": 6.272855133614628,
+      "grad_norm": 1.3858163356781006,
+      "learning_rate": 1.5246041265090363e-05,
+      "loss": 0.4778,
+      "step": 3345
+    },
+    {
+      "epoch": 6.282231598687295,
+      "grad_norm": 1.3887982368469238,
+      "learning_rate": 1.517824567159506e-05,
+      "loss": 0.4678,
+      "step": 3350
+    },
+    {
+      "epoch": 6.291608063759963,
+      "grad_norm": 1.3607546091079712,
+      "learning_rate": 1.5110535383021424e-05,
+      "loss": 0.4589,
+      "step": 3355
+    },
+    {
+      "epoch": 6.30098452883263,
+      "grad_norm": 1.2462177276611328,
+      "learning_rate": 1.5042910987453909e-05,
+      "loss": 0.4662,
+      "step": 3360
+    },
+    {
+      "epoch": 6.310360993905298,
+      "grad_norm": 1.3966500759124756,
+      "learning_rate": 1.497537307223095e-05,
+      "loss": 0.5021,
+      "step": 3365
+    },
+    {
+      "epoch": 6.319737458977965,
+      "grad_norm": 1.3240716457366943,
+      "learning_rate": 1.4907922223939874e-05,
+      "loss": 0.4999,
+      "step": 3370
+    },
+    {
+      "epoch": 6.329113924050633,
+      "grad_norm": 1.3514400720596313,
+      "learning_rate": 1.4840559028411776e-05,
+      "loss": 0.4635,
+      "step": 3375
+    },
+    {
+      "epoch": 6.338490389123301,
+      "grad_norm": 1.3836942911148071,
+      "learning_rate": 1.4773284070716503e-05,
+      "loss": 0.4582,
+      "step": 3380
+    },
+    {
+      "epoch": 6.347866854195968,
+      "grad_norm": 1.2241392135620117,
+      "learning_rate": 1.470609793515751e-05,
+      "loss": 0.4462,
+      "step": 3385
+    },
+    {
+      "epoch": 6.357243319268636,
+      "grad_norm": 1.2426246404647827,
+      "learning_rate": 1.4639001205266803e-05,
+      "loss": 0.4785,
+      "step": 3390
+    },
+    {
+      "epoch": 6.366619784341303,
+      "grad_norm": 1.2843676805496216,
+      "learning_rate": 1.4571994463799876e-05,
+      "loss": 0.5108,
+      "step": 3395
+    },
+    {
+      "epoch": 6.375996249413971,
+      "grad_norm": 1.2921011447906494,
+      "learning_rate": 1.4505078292730632e-05,
+      "loss": 0.4786,
+      "step": 3400
+    },
+    {
+      "epoch": 6.385372714486638,
+      "grad_norm": 1.3774598836898804,
+      "learning_rate": 1.4438253273246378e-05,
+      "loss": 0.4896,
+      "step": 3405
+    },
+    {
+      "epoch": 6.394749179559306,
+      "grad_norm": 1.3065452575683594,
+      "learning_rate": 1.4371519985742715e-05,
+      "loss": 0.4571,
+      "step": 3410
+    },
+    {
+      "epoch": 6.404125644631974,
+      "grad_norm": 1.3350876569747925,
+      "learning_rate": 1.4304879009818516e-05,
+      "loss": 0.4605,
+      "step": 3415
+    },
+    {
+      "epoch": 6.413502109704641,
+      "grad_norm": 1.355252742767334,
+      "learning_rate": 1.4238330924270927e-05,
+      "loss": 0.4748,
+      "step": 3420
+    },
+    {
+      "epoch": 6.422878574777309,
+      "grad_norm": 1.3216321468353271,
+      "learning_rate": 1.4171876307090284e-05,
+      "loss": 0.4698,
+      "step": 3425
+    },
+    {
+      "epoch": 6.432255039849976,
+      "grad_norm": 1.210968255996704,
+      "learning_rate": 1.4105515735455149e-05,
+      "loss": 0.4958,
+      "step": 3430
+    },
+    {
+      "epoch": 6.441631504922644,
+      "grad_norm": 1.4339016675949097,
+      "learning_rate": 1.4039249785727249e-05,
+      "loss": 0.465,
+      "step": 3435
+    },
+    {
+      "epoch": 6.451007969995311,
+      "grad_norm": 1.317304253578186,
+      "learning_rate": 1.3973079033446501e-05,
+      "loss": 0.475,
+      "step": 3440
+    },
+    {
+      "epoch": 6.460384435067979,
+      "grad_norm": 1.2484238147735596,
+      "learning_rate": 1.3907004053326006e-05,
+      "loss": 0.4765,
+      "step": 3445
+    },
+    {
+      "epoch": 6.469760900140647,
+      "grad_norm": 1.2497808933258057,
+      "learning_rate": 1.3841025419247045e-05,
+      "loss": 0.4909,
+      "step": 3450
+    },
+    {
+      "epoch": 6.479137365213314,
+      "grad_norm": 1.2647196054458618,
+      "learning_rate": 1.3775143704254123e-05,
+      "loss": 0.4789,
+      "step": 3455
+    },
+    {
+      "epoch": 6.488513830285982,
+      "grad_norm": 1.5578006505966187,
+      "learning_rate": 1.3709359480549932e-05,
+      "loss": 0.4795,
+      "step": 3460
+    },
+    {
+      "epoch": 6.4978902953586495,
+      "grad_norm": 1.169047236442566,
+      "learning_rate": 1.364367331949047e-05,
+      "loss": 0.439,
+      "step": 3465
+    },
+    {
+      "epoch": 6.5072667604313175,
+      "grad_norm": 1.4554437398910522,
+      "learning_rate": 1.3578085791580008e-05,
+      "loss": 0.469,
+      "step": 3470
+    },
+    {
+      "epoch": 6.5166432255039854,
+      "grad_norm": 1.3108760118484497,
+      "learning_rate": 1.3512597466466157e-05,
+      "loss": 0.4596,
+      "step": 3475
+    },
+    {
+      "epoch": 6.5260196905766525,
+      "grad_norm": 1.3502819538116455,
+      "learning_rate": 1.3447208912934927e-05,
+      "loss": 0.4533,
+      "step": 3480
+    },
+    {
+      "epoch": 6.5353961556493205,
+      "grad_norm": 1.3211690187454224,
+      "learning_rate": 1.3381920698905787e-05,
+      "loss": 0.5055,
+      "step": 3485
+    },
+    {
+      "epoch": 6.544772620721988,
+      "grad_norm": 1.5334270000457764,
+      "learning_rate": 1.3316733391426716e-05,
+      "loss": 0.4792,
+      "step": 3490
+    },
+    {
+      "epoch": 6.554149085794656,
+      "grad_norm": 1.3144179582595825,
+      "learning_rate": 1.32516475566693e-05,
+      "loss": 0.4855,
+      "step": 3495
+    },
+    {
+      "epoch": 6.563525550867323,
+      "grad_norm": 1.3895609378814697,
+      "learning_rate": 1.3186663759923782e-05,
+      "loss": 0.4679,
+      "step": 3500
+    },
+    {
+      "epoch": 6.572902015939991,
+      "grad_norm": 1.2866381406784058,
+      "learning_rate": 1.3121782565594185e-05,
+      "loss": 0.4823,
+      "step": 3505
+    },
+    {
+      "epoch": 6.582278481012658,
+      "grad_norm": 1.4270572662353516,
+      "learning_rate": 1.3057004537193423e-05,
+      "loss": 0.46,
+      "step": 3510
+    },
+    {
+      "epoch": 6.591654946085326,
+      "grad_norm": 1.2864598035812378,
+      "learning_rate": 1.2992330237338366e-05,
+      "loss": 0.479,
+      "step": 3515
+    },
+    {
+      "epoch": 6.601031411157994,
+      "grad_norm": 1.2378607988357544,
+      "learning_rate": 1.2927760227744943e-05,
+      "loss": 0.4779,
+      "step": 3520
+    },
+    {
+      "epoch": 6.610407876230661,
+      "grad_norm": 1.569650411605835,
+      "learning_rate": 1.2863295069223321e-05,
+      "loss": 0.4695,
+      "step": 3525
+    },
+    {
+      "epoch": 6.619784341303329,
+      "grad_norm": 1.2559000253677368,
+      "learning_rate": 1.2798935321673e-05,
+      "loss": 0.5207,
+      "step": 3530
+    },
+    {
+      "epoch": 6.629160806375996,
+      "grad_norm": 1.191889762878418,
+      "learning_rate": 1.2734681544077951e-05,
+      "loss": 0.4483,
+      "step": 3535
+    },
+    {
+      "epoch": 6.638537271448664,
+      "grad_norm": 1.3526393175125122,
+      "learning_rate": 1.2670534294501756e-05,
+      "loss": 0.4859,
+      "step": 3540
+    },
+    {
+      "epoch": 6.647913736521332,
+      "grad_norm": 1.417015790939331,
+      "learning_rate": 1.2606494130082786e-05,
+      "loss": 0.4775,
+      "step": 3545
+    },
+    {
+      "epoch": 6.657290201593999,
+      "grad_norm": 1.2867164611816406,
+      "learning_rate": 1.2542561607029322e-05,
+      "loss": 0.459,
+      "step": 3550
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 1.322531819343567,
+      "learning_rate": 1.2478737280614764e-05,
+      "loss": 0.4835,
+      "step": 3555
+    },
+    {
+      "epoch": 6.676043131739334,
+      "grad_norm": 1.273796558380127,
+      "learning_rate": 1.2415021705172799e-05,
+      "loss": 0.4854,
+      "step": 3560
+    },
+    {
+      "epoch": 6.685419596812002,
+      "grad_norm": 1.235628366470337,
+      "learning_rate": 1.2351415434092545e-05,
+      "loss": 0.5066,
+      "step": 3565
+    },
+    {
+      "epoch": 6.694796061884669,
+      "grad_norm": 1.3556861877441406,
+      "learning_rate": 1.2287919019813807e-05,
+      "loss": 0.4483,
+      "step": 3570
+    },
+    {
+      "epoch": 6.704172526957337,
+      "grad_norm": 1.3843202590942383,
+      "learning_rate": 1.2224533013822238e-05,
+      "loss": 0.4945,
+      "step": 3575
+    },
+    {
+      "epoch": 6.713548992030004,
+      "grad_norm": 1.4850592613220215,
+      "learning_rate": 1.2161257966644588e-05,
+      "loss": 0.4709,
+      "step": 3580
+    },
+    {
+      "epoch": 6.722925457102672,
+      "grad_norm": 1.38179349899292,
+      "learning_rate": 1.2098094427843854e-05,
+      "loss": 0.4652,
+      "step": 3585
+    },
+    {
+      "epoch": 6.73230192217534,
+      "grad_norm": 1.4235470294952393,
+      "learning_rate": 1.2035042946014572e-05,
+      "loss": 0.4783,
+      "step": 3590
+    },
+    {
+      "epoch": 6.741678387248007,
+      "grad_norm": 1.3751304149627686,
+      "learning_rate": 1.1972104068778028e-05,
+      "loss": 0.4574,
+      "step": 3595
+    },
+    {
+      "epoch": 6.751054852320675,
+      "grad_norm": 1.156980276107788,
+      "learning_rate": 1.1909278342777513e-05,
+      "loss": 0.4643,
+      "step": 3600
+    },
+    {
+      "epoch": 6.760431317393342,
+      "grad_norm": 1.2472528219223022,
+      "learning_rate": 1.1846566313673524e-05,
+      "loss": 0.5495,
+      "step": 3605
+    },
+    {
+      "epoch": 6.76980778246601,
+      "grad_norm": 1.321594476699829,
+      "learning_rate": 1.1783968526139121e-05,
+      "loss": 0.4563,
+      "step": 3610
+    },
+    {
+      "epoch": 6.779184247538678,
+      "grad_norm": 1.2460840940475464,
+      "learning_rate": 1.172148552385511e-05,
+      "loss": 0.5021,
+      "step": 3615
+    },
+    {
+      "epoch": 6.788560712611345,
+      "grad_norm": 1.3775686025619507,
+      "learning_rate": 1.1659117849505367e-05,
+      "loss": 0.4628,
+      "step": 3620
+    },
+    {
+      "epoch": 6.797937177684013,
+      "grad_norm": 1.3026732206344604,
+      "learning_rate": 1.1596866044772076e-05,
+      "loss": 0.4432,
+      "step": 3625
+    },
+    {
+      "epoch": 6.8073136427566805,
+      "grad_norm": 1.2298964262008667,
+      "learning_rate": 1.1534730650331096e-05,
+      "loss": 0.5062,
+      "step": 3630
+    },
+    {
+      "epoch": 6.8166901078293485,
+      "grad_norm": 1.306643009185791,
+      "learning_rate": 1.1472712205847216e-05,
+      "loss": 0.4682,
+      "step": 3635
+    },
+    {
+      "epoch": 6.826066572902016,
+      "grad_norm": 1.3525081872940063,
+      "learning_rate": 1.1410811249969475e-05,
+      "loss": 0.4834,
+      "step": 3640
+    },
+    {
+      "epoch": 6.8354430379746836,
+      "grad_norm": 1.29501211643219,
+      "learning_rate": 1.1349028320326487e-05,
+      "loss": 0.4335,
+      "step": 3645
+    },
+    {
+      "epoch": 6.844819503047351,
+      "grad_norm": 1.3382041454315186,
+      "learning_rate": 1.1287363953521779e-05,
+      "loss": 0.473,
+      "step": 3650
+    },
+    {
+      "epoch": 6.854195968120019,
+      "grad_norm": 1.502648949623108,
+      "learning_rate": 1.122581868512912e-05,
+      "loss": 0.4659,
+      "step": 3655
+    },
+    {
+      "epoch": 6.863572433192687,
+      "grad_norm": 1.387737512588501,
+      "learning_rate": 1.1164393049687868e-05,
+      "loss": 0.4398,
+      "step": 3660
+    },
+    {
+      "epoch": 6.872948898265354,
+      "grad_norm": 1.3075759410858154,
+      "learning_rate": 1.1103087580698351e-05,
+      "loss": 0.4731,
+      "step": 3665
+    },
+    {
+      "epoch": 6.882325363338022,
+      "grad_norm": 1.4493883848190308,
+      "learning_rate": 1.104190281061718e-05,
+      "loss": 0.4692,
+      "step": 3670
+    },
+    {
+      "epoch": 6.891701828410689,
+      "grad_norm": 1.3822391033172607,
+      "learning_rate": 1.0980839270852683e-05,
+      "loss": 0.4649,
+      "step": 3675
+    },
+    {
+      "epoch": 6.901078293483357,
+      "grad_norm": 1.3482156991958618,
+      "learning_rate": 1.0919897491760279e-05,
+      "loss": 0.4574,
+      "step": 3680
+    },
+    {
+      "epoch": 6.910454758556025,
+      "grad_norm": 1.3633962869644165,
+      "learning_rate": 1.0859078002637842e-05,
+      "loss": 0.4478,
+      "step": 3685
+    },
+    {
+      "epoch": 6.919831223628692,
+      "grad_norm": 1.3784875869750977,
+      "learning_rate": 1.0798381331721109e-05,
+      "loss": 0.5066,
+      "step": 3690
+    },
+    {
+      "epoch": 6.92920768870136,
+      "grad_norm": 1.1772392988204956,
+      "learning_rate": 1.0737808006179118e-05,
+      "loss": 0.4928,
+      "step": 3695
+    },
+    {
+      "epoch": 6.938584153774027,
+      "grad_norm": 1.3571972846984863,
+      "learning_rate": 1.0677358552109618e-05,
+      "loss": 0.4555,
+      "step": 3700
+    },
+    {
+      "epoch": 6.947960618846695,
+      "grad_norm": 1.3081351518630981,
+      "learning_rate": 1.0617033494534486e-05,
+      "loss": 0.4612,
+      "step": 3705
+    },
+    {
+      "epoch": 6.957337083919363,
+      "grad_norm": 1.3350365161895752,
+      "learning_rate": 1.0556833357395188e-05,
+      "loss": 0.4623,
+      "step": 3710
+    },
+    {
+      "epoch": 6.96671354899203,
+      "grad_norm": 1.2777769565582275,
+      "learning_rate": 1.0496758663548209e-05,
+      "loss": 0.4773,
+      "step": 3715
+    },
+    {
+      "epoch": 6.976090014064698,
+      "grad_norm": 1.2693568468093872,
+      "learning_rate": 1.0436809934760527e-05,
+      "loss": 0.5141,
+      "step": 3720
+    },
+    {
+      "epoch": 6.985466479137365,
+      "grad_norm": 1.286055088043213,
+      "learning_rate": 1.0376987691705084e-05,
+      "loss": 0.4389,
+      "step": 3725
+    },
+    {
+      "epoch": 6.994842944210033,
+      "grad_norm": 1.3331400156021118,
+      "learning_rate": 1.031729245395622e-05,
+      "loss": 0.4674,
+      "step": 3730
+    },
+    {
+      "epoch": 7.0042194092827,
+      "grad_norm": 1.291988730430603,
+      "learning_rate": 1.0257724739985236e-05,
+      "loss": 0.4593,
+      "step": 3735
+    },
+    {
+      "epoch": 7.013595874355368,
+      "grad_norm": 1.2212333679199219,
+      "learning_rate": 1.0198285067155827e-05,
+      "loss": 0.4549,
+      "step": 3740
+    },
+    {
+      "epoch": 7.022972339428035,
+      "grad_norm": 1.3065071105957031,
+      "learning_rate": 1.0138973951719613e-05,
+      "loss": 0.4889,
+      "step": 3745
+    },
+    {
+      "epoch": 7.032348804500703,
+      "grad_norm": 1.3612645864486694,
+      "learning_rate": 1.0079791908811683e-05,
+      "loss": 0.4611,
+      "step": 3750
+    },
+    {
+      "epoch": 7.041725269573371,
+      "grad_norm": 1.3851178884506226,
+      "learning_rate": 1.002073945244605e-05,
+      "loss": 0.4634,
+      "step": 3755
+    },
+    {
+      "epoch": 7.051101734646038,
+      "grad_norm": 1.2975175380706787,
+      "learning_rate": 9.961817095511242e-06,
+      "loss": 0.4778,
+      "step": 3760
+    },
+    {
+      "epoch": 7.060478199718706,
+      "grad_norm": 1.5221518278121948,
+      "learning_rate": 9.903025349765848e-06,
+      "loss": 0.4878,
+      "step": 3765
+    },
+    {
+      "epoch": 7.069854664791373,
+      "grad_norm": 1.3909027576446533,
+      "learning_rate": 9.844364725834057e-06,
+      "loss": 0.4548,
+      "step": 3770
+    },
+    {
+      "epoch": 7.079231129864041,
+      "grad_norm": 1.3161048889160156,
+      "learning_rate": 9.78583573320119e-06,
+      "loss": 0.4826,
+      "step": 3775
+    },
+    {
+      "epoch": 7.0886075949367084,
+      "grad_norm": 1.2947895526885986,
+      "learning_rate": 9.727438880209366e-06,
+      "loss": 0.4825,
+      "step": 3780
+    },
+    {
+      "epoch": 7.097984060009376,
+      "grad_norm": 1.4308338165283203,
+      "learning_rate": 9.669174674052991e-06,
+      "loss": 0.5366,
+      "step": 3785
+    },
+    {
+      "epoch": 7.107360525082044,
+      "grad_norm": 1.3448426723480225,
+      "learning_rate": 9.611043620774419e-06,
+      "loss": 0.4573,
+      "step": 3790
+    },
+    {
+      "epoch": 7.1167369901547115,
+      "grad_norm": 1.419171690940857,
+      "learning_rate": 9.553046225259495e-06,
+      "loss": 0.4612,
+      "step": 3795
+    },
+    {
+      "epoch": 7.1261134552273795,
+      "grad_norm": 1.443497657775879,
+      "learning_rate": 9.495182991233236e-06,
+      "loss": 0.4467,
+      "step": 3800
+    },
+    {
+      "epoch": 7.135489920300047,
+      "grad_norm": 1.2052894830703735,
+      "learning_rate": 9.43745442125542e-06,
+      "loss": 0.5063,
+      "step": 3805
+    },
+    {
+      "epoch": 7.144866385372715,
+      "grad_norm": 1.330528974533081,
+      "learning_rate": 9.379861016716224e-06,
+      "loss": 0.454,
+      "step": 3810
+    },
+    {
+      "epoch": 7.154242850445382,
+      "grad_norm": 1.3853237628936768,
+      "learning_rate": 9.322403277831871e-06,
+      "loss": 0.4322,
+      "step": 3815
+    },
+    {
+      "epoch": 7.16361931551805,
+      "grad_norm": 1.3971184492111206,
+      "learning_rate": 9.265081703640285e-06,
+      "loss": 0.4567,
+      "step": 3820
+    },
+    {
+      "epoch": 7.172995780590718,
+      "grad_norm": 1.2893038988113403,
+      "learning_rate": 9.207896791996764e-06,
+      "loss": 0.4753,
+      "step": 3825
+    },
+    {
+      "epoch": 7.182372245663385,
+      "grad_norm": 1.5246171951293945,
+      "learning_rate": 9.150849039569655e-06,
+      "loss": 0.4622,
+      "step": 3830
+    },
+    {
+      "epoch": 7.191748710736053,
+      "grad_norm": 1.383436679840088,
+      "learning_rate": 9.09393894183601e-06,
+      "loss": 0.5124,
+      "step": 3835
+    },
+    {
+      "epoch": 7.20112517580872,
+      "grad_norm": 1.3397338390350342,
+      "learning_rate": 9.037166993077337e-06,
+      "loss": 0.5021,
+      "step": 3840
+    },
+    {
+      "epoch": 7.210501640881388,
+      "grad_norm": 1.5347236394882202,
+      "learning_rate": 8.980533686375261e-06,
+      "loss": 0.4429,
+      "step": 3845
+    },
+    {
+      "epoch": 7.219878105954056,
+      "grad_norm": 1.397948145866394,
+      "learning_rate": 8.92403951360726e-06,
+      "loss": 0.4885,
+      "step": 3850
+    },
+    {
+      "epoch": 7.229254571026723,
+      "grad_norm": 1.363653540611267,
+      "learning_rate": 8.867684965442425e-06,
+      "loss": 0.5183,
+      "step": 3855
+    },
+    {
+      "epoch": 7.238631036099391,
+      "grad_norm": 1.3129911422729492,
+      "learning_rate": 8.811470531337102e-06,
+      "loss": 0.4651,
+      "step": 3860
+    },
+    {
+      "epoch": 7.248007501172058,
+      "grad_norm": 1.320569634437561,
+      "learning_rate": 8.755396699530752e-06,
+      "loss": 0.5209,
+      "step": 3865
+    },
+    {
+      "epoch": 7.257383966244726,
+      "grad_norm": 1.318885326385498,
+      "learning_rate": 8.699463957041649e-06,
+      "loss": 0.4355,
+      "step": 3870
+    },
+    {
+      "epoch": 7.266760431317393,
+      "grad_norm": 1.3796628713607788,
+      "learning_rate": 8.643672789662665e-06,
+      "loss": 0.5222,
+      "step": 3875
+    },
+    {
+      "epoch": 7.276136896390061,
+      "grad_norm": 1.3845789432525635,
+      "learning_rate": 8.588023681957028e-06,
+      "loss": 0.4599,
+      "step": 3880
+    },
+    {
+      "epoch": 7.285513361462728,
+      "grad_norm": 1.3544453382492065,
+      "learning_rate": 8.532517117254171e-06,
+      "loss": 0.4725,
+      "step": 3885
+    },
+    {
+      "epoch": 7.294889826535396,
+      "grad_norm": 1.3245503902435303,
+      "learning_rate": 8.477153577645481e-06,
+      "loss": 0.4482,
+      "step": 3890
+    },
+    {
+      "epoch": 7.304266291608064,
+      "grad_norm": 1.3104580640792847,
+      "learning_rate": 8.421933543980126e-06,
+      "loss": 0.4436,
+      "step": 3895
+    },
+    {
+      "epoch": 7.313642756680731,
+      "grad_norm": 1.5002598762512207,
+      "learning_rate": 8.36685749586087e-06,
+      "loss": 0.454,
+      "step": 3900
+    },
+    {
+      "epoch": 7.323019221753399,
+      "grad_norm": 1.5575108528137207,
+      "learning_rate": 8.311925911639942e-06,
+      "loss": 0.4618,
+      "step": 3905
+    },
+    {
+      "epoch": 7.332395686826066,
+      "grad_norm": 1.364050269126892,
+      "learning_rate": 8.257139268414844e-06,
+      "loss": 0.4821,
+      "step": 3910
+    },
+    {
+      "epoch": 7.341772151898734,
+      "grad_norm": 1.3934012651443481,
+      "learning_rate": 8.202498042024226e-06,
+      "loss": 0.4742,
+      "step": 3915
+    },
+    {
+      "epoch": 7.351148616971402,
+      "grad_norm": 1.3544234037399292,
+      "learning_rate": 8.14800270704375e-06,
+      "loss": 0.4353,
+      "step": 3920
+    },
+    {
+      "epoch": 7.360525082044069,
+      "grad_norm": 1.3235310316085815,
+      "learning_rate": 8.093653736781958e-06,
+      "loss": 0.4559,
+      "step": 3925
+    },
+    {
+      "epoch": 7.369901547116737,
+      "grad_norm": 1.3521572351455688,
+      "learning_rate": 8.039451603276185e-06,
+      "loss": 0.4618,
+      "step": 3930
+    },
+    {
+      "epoch": 7.379278012189404,
+      "grad_norm": 1.3934191465377808,
+      "learning_rate": 7.985396777288436e-06,
+      "loss": 0.5106,
+      "step": 3935
+    },
+    {
+      "epoch": 7.388654477262072,
+      "grad_norm": 1.4107383489608765,
+      "learning_rate": 7.931489728301292e-06,
+      "loss": 0.4424,
+      "step": 3940
+    },
+    {
+      "epoch": 7.3980309423347395,
+      "grad_norm": 1.3612831830978394,
+      "learning_rate": 7.877730924513866e-06,
+      "loss": 0.4505,
+      "step": 3945
+    },
+    {
+      "epoch": 7.407407407407407,
+      "grad_norm": 1.4322446584701538,
+      "learning_rate": 7.8241208328377e-06,
+      "loss": 0.493,
+      "step": 3950
+    },
+    {
+      "epoch": 7.416783872480075,
+      "grad_norm": 1.3748643398284912,
+      "learning_rate": 7.770659918892754e-06,
+      "loss": 0.4625,
+      "step": 3955
+    },
+    {
+      "epoch": 7.4261603375527425,
+      "grad_norm": 1.2667847871780396,
+      "learning_rate": 7.71734864700331e-06,
+      "loss": 0.4802,
+      "step": 3960
+    },
+    {
+      "epoch": 7.4355368026254105,
+      "grad_norm": 1.2717978954315186,
+      "learning_rate": 7.66418748019396e-06,
+      "loss": 0.4531,
+      "step": 3965
+    },
+    {
+      "epoch": 7.444913267698078,
+      "grad_norm": 1.2851226329803467,
+      "learning_rate": 7.611176880185597e-06,
+      "loss": 0.4634,
+      "step": 3970
+    },
+    {
+      "epoch": 7.454289732770746,
+      "grad_norm": 1.4703879356384277,
+      "learning_rate": 7.558317307391399e-06,
+      "loss": 0.4735,
+      "step": 3975
+    },
+    {
+      "epoch": 7.463666197843413,
+      "grad_norm": 1.3519517183303833,
+      "learning_rate": 7.505609220912821e-06,
+      "loss": 0.4823,
+      "step": 3980
+    },
+    {
+      "epoch": 7.473042662916081,
+      "grad_norm": 1.2879433631896973,
+      "learning_rate": 7.453053078535613e-06,
+      "loss": 0.4431,
+      "step": 3985
+    },
+    {
+      "epoch": 7.482419127988749,
+      "grad_norm": 1.2699778079986572,
+      "learning_rate": 7.4006493367258515e-06,
+      "loss": 0.4913,
+      "step": 3990
+    },
+    {
+      "epoch": 7.491795593061416,
+      "grad_norm": 1.4511828422546387,
+      "learning_rate": 7.348398450625956e-06,
+      "loss": 0.5012,
+      "step": 3995
+    },
+    {
+      "epoch": 7.501172058134084,
+      "grad_norm": 1.3058310747146606,
+      "learning_rate": 7.2963008740507656e-06,
+      "loss": 0.4804,
+      "step": 4000
+    },
+    {
+      "epoch": 7.510548523206751,
+      "grad_norm": 1.341951847076416,
+      "learning_rate": 7.244357059483551e-06,
+      "loss": 0.4285,
+      "step": 4005
+    },
+    {
+      "epoch": 7.519924988279419,
+      "grad_norm": 1.3997609615325928,
+      "learning_rate": 7.192567458072138e-06,
+      "loss": 0.471,
+      "step": 4010
+    },
+    {
+      "epoch": 7.529301453352086,
+      "grad_norm": 1.3489426374435425,
+      "learning_rate": 7.14093251962496e-06,
+      "loss": 0.4749,
+      "step": 4015
+    },
+    {
+      "epoch": 7.538677918424754,
+      "grad_norm": 1.3656972646713257,
+      "learning_rate": 7.089452692607146e-06,
+      "loss": 0.4834,
+      "step": 4020
+    },
+    {
+      "epoch": 7.548054383497422,
+      "grad_norm": 1.3216897249221802,
+      "learning_rate": 7.038128424136644e-06,
+      "loss": 0.4667,
+      "step": 4025
+    },
+    {
+      "epoch": 7.557430848570089,
+      "grad_norm": 1.290844440460205,
+      "learning_rate": 6.986960159980327e-06,
+      "loss": 0.4688,
+      "step": 4030
+    },
+    {
+      "epoch": 7.566807313642757,
+      "grad_norm": 1.351767897605896,
+      "learning_rate": 6.935948344550119e-06,
+      "loss": 0.4733,
+      "step": 4035
+    },
+    {
+      "epoch": 7.576183778715424,
+      "grad_norm": 1.3938384056091309,
+      "learning_rate": 6.885093420899152e-06,
+      "loss": 0.4715,
+      "step": 4040
+    },
+    {
+      "epoch": 7.585560243788092,
+      "grad_norm": 1.3741334676742554,
+      "learning_rate": 6.8343958307178795e-06,
+      "loss": 0.4597,
+      "step": 4045
+    },
+    {
+      "epoch": 7.594936708860759,
+      "grad_norm": 1.3019955158233643,
+      "learning_rate": 6.783856014330281e-06,
+      "loss": 0.4745,
+      "step": 4050
+    },
+    {
+      "epoch": 7.604313173933427,
+      "grad_norm": 1.342897891998291,
+      "learning_rate": 6.7334744106900414e-06,
+      "loss": 0.461,
+      "step": 4055
+    },
+    {
+      "epoch": 7.613689639006095,
+      "grad_norm": 1.525770902633667,
+      "learning_rate": 6.68325145737669e-06,
+      "loss": 0.4434,
+      "step": 4060
+    },
+    {
+      "epoch": 7.623066104078762,
+      "grad_norm": 1.3355739116668701,
+      "learning_rate": 6.633187590591855e-06,
+      "loss": 0.4923,
+      "step": 4065
+    },
+    {
+      "epoch": 7.63244256915143,
+      "grad_norm": 1.3165456056594849,
+      "learning_rate": 6.583283245155414e-06,
+      "loss": 0.4516,
+      "step": 4070
+    },
+    {
+      "epoch": 7.641819034224097,
+      "grad_norm": 1.3211554288864136,
+      "learning_rate": 6.5335388545017915e-06,
+      "loss": 0.4835,
+      "step": 4075
+    },
+    {
+      "epoch": 7.651195499296765,
+      "grad_norm": 1.3713293075561523,
+      "learning_rate": 6.483954850676133e-06,
+      "loss": 0.4544,
+      "step": 4080
+    },
+    {
+      "epoch": 7.660571964369433,
+      "grad_norm": 1.3454351425170898,
+      "learning_rate": 6.434531664330587e-06,
+      "loss": 0.4759,
+      "step": 4085
+    },
+    {
+      "epoch": 7.6699484294421,
+      "grad_norm": 1.3137215375900269,
+      "learning_rate": 6.385269724720547e-06,
+      "loss": 0.4414,
+      "step": 4090
+    },
+    {
+      "epoch": 7.679324894514768,
+      "grad_norm": 1.197231650352478,
+      "learning_rate": 6.336169459700933e-06,
+      "loss": 0.4473,
+      "step": 4095
+    },
+    {
+      "epoch": 7.688701359587435,
+      "grad_norm": 1.4338058233261108,
+      "learning_rate": 6.28723129572247e-06,
+      "loss": 0.5081,
+      "step": 4100
+    },
+    {
+      "epoch": 7.698077824660103,
+      "grad_norm": 1.359397530555725,
+      "learning_rate": 6.238455657827999e-06,
+      "loss": 0.4727,
+      "step": 4105
+    },
+    {
+      "epoch": 7.7074542897327705,
+      "grad_norm": 1.3532447814941406,
+      "learning_rate": 6.189842969648737e-06,
+      "loss": 0.4345,
+      "step": 4110
+    },
+    {
+      "epoch": 7.7168307548054385,
+      "grad_norm": 1.3469390869140625,
+      "learning_rate": 6.14139365340067e-06,
+      "loss": 0.4512,
+      "step": 4115
+    },
+    {
+      "epoch": 7.7262072198781055,
+      "grad_norm": 1.3108094930648804,
+      "learning_rate": 6.0931081298808316e-06,
+      "loss": 0.4419,
+      "step": 4120
+    },
+    {
+      "epoch": 7.7355836849507735,
+      "grad_norm": 1.4088788032531738,
+      "learning_rate": 6.044986818463683e-06,
+      "loss": 0.4436,
+      "step": 4125
+    },
+    {
+      "epoch": 7.7449601500234415,
+      "grad_norm": 1.2557556629180908,
+      "learning_rate": 5.997030137097426e-06,
+      "loss": 0.4834,
+      "step": 4130
+    },
+    {
+      "epoch": 7.754336615096109,
+      "grad_norm": 1.2912348508834839,
+      "learning_rate": 5.9492385023004196e-06,
+      "loss": 0.4311,
+      "step": 4135
+    },
+    {
+      "epoch": 7.763713080168777,
+      "grad_norm": 1.3706984519958496,
+      "learning_rate": 5.901612329157535e-06,
+      "loss": 0.5012,
+      "step": 4140
+    },
+    {
+      "epoch": 7.773089545241444,
+      "grad_norm": 1.3478624820709229,
+      "learning_rate": 5.854152031316576e-06,
+      "loss": 0.4896,
+      "step": 4145
+    },
+    {
+      "epoch": 7.782466010314112,
+      "grad_norm": 1.4457701444625854,
+      "learning_rate": 5.806858020984629e-06,
+      "loss": 0.4757,
+      "step": 4150
+    },
+    {
+      "epoch": 7.79184247538678,
+      "grad_norm": 1.3146065473556519,
+      "learning_rate": 5.759730708924571e-06,
+      "loss": 0.4583,
+      "step": 4155
+    },
+    {
+      "epoch": 7.801218940459447,
+      "grad_norm": 1.3273216485977173,
+      "learning_rate": 5.712770504451426e-06,
+      "loss": 0.4696,
+      "step": 4160
+    },
+    {
+      "epoch": 7.810595405532115,
+      "grad_norm": 1.4631688594818115,
+      "learning_rate": 5.665977815428841e-06,
+      "loss": 0.4897,
+      "step": 4165
+    },
+    {
+      "epoch": 7.819971870604782,
+      "grad_norm": 1.3307130336761475,
+      "learning_rate": 5.619353048265552e-06,
+      "loss": 0.449,
+      "step": 4170
+    },
+    {
+      "epoch": 7.82934833567745,
+      "grad_norm": 1.4473830461502075,
+      "learning_rate": 5.572896607911818e-06,
+      "loss": 0.4666,
+      "step": 4175
+    },
+    {
+      "epoch": 7.838724800750117,
+      "grad_norm": 1.2786786556243896,
+      "learning_rate": 5.526608897855953e-06,
+      "loss": 0.4782,
+      "step": 4180
+    },
+    {
+      "epoch": 7.848101265822785,
+      "grad_norm": 1.2860808372497559,
+      "learning_rate": 5.480490320120784e-06,
+      "loss": 0.5104,
+      "step": 4185
+    },
+    {
+      "epoch": 7.857477730895452,
+      "grad_norm": 1.3066420555114746,
+      "learning_rate": 5.434541275260182e-06,
+      "loss": 0.4603,
+      "step": 4190
+    },
+    {
+      "epoch": 7.86685419596812,
+      "grad_norm": 1.4550244808197021,
+      "learning_rate": 5.388762162355565e-06,
+      "loss": 0.4539,
+      "step": 4195
+    },
+    {
+      "epoch": 7.876230661040788,
+      "grad_norm": 1.2913579940795898,
+      "learning_rate": 5.343153379012444e-06,
+      "loss": 0.5024,
+      "step": 4200
+    },
+    {
+      "epoch": 7.885607126113455,
+      "grad_norm": 1.4126182794570923,
+      "learning_rate": 5.2977153213569635e-06,
+      "loss": 0.4666,
+      "step": 4205
+    },
+    {
+      "epoch": 7.894983591186123,
+      "grad_norm": 1.3481818437576294,
+      "learning_rate": 5.252448384032471e-06,
+      "loss": 0.4396,
+      "step": 4210
+    },
+    {
+      "epoch": 7.90436005625879,
+      "grad_norm": 1.3086531162261963,
+      "learning_rate": 5.207352960196069e-06,
+      "loss": 0.467,
+      "step": 4215
+    },
+    {
+      "epoch": 7.913736521331458,
+      "grad_norm": 1.3286422491073608,
+      "learning_rate": 5.162429441515221e-06,
+      "loss": 0.4677,
+      "step": 4220
+    },
+    {
+      "epoch": 7.923112986404126,
+      "grad_norm": 1.292620062828064,
+      "learning_rate": 5.117678218164338e-06,
+      "loss": 0.5007,
+      "step": 4225
+    },
+    {
+      "epoch": 7.932489451476793,
+      "grad_norm": 1.3954863548278809,
+      "learning_rate": 5.073099678821413e-06,
+      "loss": 0.4522,
+      "step": 4230
+    },
+    {
+      "epoch": 7.941865916549461,
+      "grad_norm": 1.3693162202835083,
+      "learning_rate": 5.028694210664592e-06,
+      "loss": 0.475,
+      "step": 4235
+    },
+    {
+      "epoch": 7.951242381622128,
+      "grad_norm": 1.4868087768554688,
+      "learning_rate": 4.984462199368872e-06,
+      "loss": 0.4216,
+      "step": 4240
+    },
+    {
+      "epoch": 7.960618846694796,
+      "grad_norm": 1.2957206964492798,
+      "learning_rate": 4.940404029102713e-06,
+      "loss": 0.4799,
+      "step": 4245
+    },
+    {
+      "epoch": 7.969995311767463,
+      "grad_norm": 1.325158953666687,
+      "learning_rate": 4.8965200825247245e-06,
+      "loss": 0.4964,
+      "step": 4250
+    },
+    {
+      "epoch": 7.979371776840131,
+      "grad_norm": 1.2667571306228638,
+      "learning_rate": 4.8528107407803e-06,
+      "loss": 0.4637,
+      "step": 4255
+    },
+    {
+      "epoch": 7.988748241912798,
+      "grad_norm": 3.300675630569458,
+      "learning_rate": 4.809276383498376e-06,
+      "loss": 0.4787,
+      "step": 4260
+    },
+    {
+      "epoch": 7.998124706985466,
+      "grad_norm": 1.408666729927063,
+      "learning_rate": 4.765917388788071e-06,
+      "loss": 0.5012,
+      "step": 4265
+    },
+    {
+      "epoch": 8.007501172058134,
+      "grad_norm": 1.2651296854019165,
+      "learning_rate": 4.722734133235438e-06,
+      "loss": 0.4882,
+      "step": 4270
+    },
+    {
+      "epoch": 8.016877637130802,
+      "grad_norm": 1.4006842374801636,
+      "learning_rate": 4.679726991900177e-06,
+      "loss": 0.4834,
+      "step": 4275
+    },
+    {
+      "epoch": 8.026254102203469,
+      "grad_norm": 1.3617703914642334,
+      "learning_rate": 4.636896338312374e-06,
+      "loss": 0.4774,
+      "step": 4280
+    },
+    {
+      "epoch": 8.035630567276137,
+      "grad_norm": 1.3175445795059204,
+      "learning_rate": 4.594242544469282e-06,
+      "loss": 0.4667,
+      "step": 4285
+    },
+    {
+      "epoch": 8.045007032348805,
+      "grad_norm": 1.3159610033035278,
+      "learning_rate": 4.551765980832059e-06,
+      "loss": 0.4927,
+      "step": 4290
+    },
+    {
+      "epoch": 8.054383497421473,
+      "grad_norm": 1.3321839570999146,
+      "learning_rate": 4.509467016322577e-06,
+      "loss": 0.4911,
+      "step": 4295
+    },
+    {
+      "epoch": 8.06375996249414,
+      "grad_norm": 1.460774302482605,
+      "learning_rate": 4.467346018320198e-06,
+      "loss": 0.4546,
+      "step": 4300
+    },
+    {
+      "epoch": 8.073136427566807,
+      "grad_norm": 1.5414506196975708,
+      "learning_rate": 4.425403352658591e-06,
+      "loss": 0.4451,
+      "step": 4305
+    },
+    {
+      "epoch": 8.082512892639475,
+      "grad_norm": 1.357359528541565,
+      "learning_rate": 4.383639383622557e-06,
+      "loss": 0.4917,
+      "step": 4310
+    },
+    {
+      "epoch": 8.091889357712143,
+      "grad_norm": 1.3124933242797852,
+      "learning_rate": 4.342054473944865e-06,
+      "loss": 0.4624,
+      "step": 4315
+    },
+    {
+      "epoch": 8.10126582278481,
+      "grad_norm": 1.4539119005203247,
+      "learning_rate": 4.300648984803085e-06,
+      "loss": 0.468,
+      "step": 4320
+    },
+    {
+      "epoch": 8.110642287857477,
+      "grad_norm": 1.4366650581359863,
+      "learning_rate": 4.259423275816476e-06,
+      "loss": 0.4399,
+      "step": 4325
+    },
+    {
+      "epoch": 8.120018752930145,
+      "grad_norm": 1.4772125482559204,
+      "learning_rate": 4.218377705042867e-06,
+      "loss": 0.4713,
+      "step": 4330
+    },
+    {
+      "epoch": 8.129395218002813,
+      "grad_norm": 1.4118038415908813,
+      "learning_rate": 4.177512628975508e-06,
+      "loss": 0.4594,
+      "step": 4335
+    },
+    {
+      "epoch": 8.13877168307548,
+      "grad_norm": 1.411610722541809,
+      "learning_rate": 4.1368284025399965e-06,
+      "loss": 0.4343,
+      "step": 4340
+    },
+    {
+      "epoch": 8.148148148148149,
+      "grad_norm": 1.1540132761001587,
+      "learning_rate": 4.096325379091215e-06,
+      "loss": 0.4792,
+      "step": 4345
+    },
+    {
+      "epoch": 8.157524613220815,
+      "grad_norm": 1.4441187381744385,
+      "learning_rate": 4.0560039104102305e-06,
+      "loss": 0.479,
+      "step": 4350
+    },
+    {
+      "epoch": 8.166901078293483,
+      "grad_norm": 1.4413646459579468,
+      "learning_rate": 4.015864346701251e-06,
+      "loss": 0.449,
+      "step": 4355
+    },
+    {
+      "epoch": 8.176277543366151,
+      "grad_norm": 1.3881680965423584,
+      "learning_rate": 3.975907036588594e-06,
+      "loss": 0.4678,
+      "step": 4360
+    },
+    {
+      "epoch": 8.185654008438819,
+      "grad_norm": 1.2951891422271729,
+      "learning_rate": 3.93613232711364e-06,
+      "loss": 0.5058,
+      "step": 4365
+    },
+    {
+      "epoch": 8.195030473511487,
+      "grad_norm": 1.299561619758606,
+      "learning_rate": 3.8965405637318294e-06,
+      "loss": 0.4468,
+      "step": 4370
+    },
+    {
+      "epoch": 8.204406938584153,
+      "grad_norm": 1.3333089351654053,
+      "learning_rate": 3.857132090309662e-06,
+      "loss": 0.5227,
+      "step": 4375
+    },
+    {
+      "epoch": 8.213783403656821,
+      "grad_norm": 1.4621747732162476,
+      "learning_rate": 3.817907249121713e-06,
+      "loss": 0.4476,
+      "step": 4380
+    },
+    {
+      "epoch": 8.22315986872949,
+      "grad_norm": 1.398207187652588,
+      "learning_rate": 3.77886638084764e-06,
+      "loss": 0.4478,
+      "step": 4385
+    },
+    {
+      "epoch": 8.232536333802157,
+      "grad_norm": 1.302186369895935,
+      "learning_rate": 3.7400098245692572e-06,
+      "loss": 0.4615,
+      "step": 4390
+    },
+    {
+      "epoch": 8.241912798874825,
+      "grad_norm": 1.3900314569473267,
+      "learning_rate": 3.7013379177675557e-06,
+      "loss": 0.4739,
+      "step": 4395
+    },
+    {
+      "epoch": 8.251289263947491,
+      "grad_norm": 1.3765474557876587,
+      "learning_rate": 3.662850996319825e-06,
+      "loss": 0.46,
+      "step": 4400
+    },
+    {
+      "epoch": 8.26066572902016,
+      "grad_norm": 1.336715579032898,
+      "learning_rate": 3.624549394496654e-06,
+      "loss": 0.462,
+      "step": 4405
+    },
+    {
+      "epoch": 8.270042194092827,
+      "grad_norm": 1.4132641553878784,
+      "learning_rate": 3.586433444959103e-06,
+      "loss": 0.4619,
+      "step": 4410
+    },
+    {
+      "epoch": 8.279418659165495,
+      "grad_norm": 1.4156947135925293,
+      "learning_rate": 3.5485034787557797e-06,
+      "loss": 0.4997,
+      "step": 4415
+    },
+    {
+      "epoch": 8.288795124238161,
+      "grad_norm": 1.3596758842468262,
+      "learning_rate": 3.5107598253199758e-06,
+      "loss": 0.4313,
+      "step": 4420
+    },
+    {
+      "epoch": 8.29817158931083,
+      "grad_norm": 1.3741769790649414,
+      "learning_rate": 3.473202812466775e-06,
+      "loss": 0.4627,
+      "step": 4425
+    },
+    {
+      "epoch": 8.307548054383497,
+      "grad_norm": 1.4133784770965576,
+      "learning_rate": 3.4358327663902677e-06,
+      "loss": 0.4752,
+      "step": 4430
+    },
+    {
+      "epoch": 8.316924519456165,
+      "grad_norm": 1.456164836883545,
+      "learning_rate": 3.3986500116606563e-06,
+      "loss": 0.448,
+      "step": 4435
+    },
+    {
+      "epoch": 8.326300984528833,
+      "grad_norm": 1.2815409898757935,
+      "learning_rate": 3.3616548712214756e-06,
+      "loss": 0.4445,
+      "step": 4440
+    },
+    {
+      "epoch": 8.3356774496015,
+      "grad_norm": 1.2286583185195923,
+      "learning_rate": 3.324847666386749e-06,
+      "loss": 0.5237,
+      "step": 4445
+    },
+    {
+      "epoch": 8.345053914674168,
+      "grad_norm": 1.30026113986969,
+      "learning_rate": 3.288228716838246e-06,
+      "loss": 0.4536,
+      "step": 4450
+    },
+    {
+      "epoch": 8.354430379746836,
+      "grad_norm": 1.3132426738739014,
+      "learning_rate": 3.251798340622675e-06,
+      "loss": 0.4902,
+      "step": 4455
+    },
+    {
+      "epoch": 8.363806844819504,
+      "grad_norm": 1.2999507188796997,
+      "learning_rate": 3.2155568541489268e-06,
+      "loss": 0.4647,
+      "step": 4460
+    },
+    {
+      "epoch": 8.37318330989217,
+      "grad_norm": 1.3368269205093384,
+      "learning_rate": 3.179504572185324e-06,
+      "loss": 0.4788,
+      "step": 4465
+    },
+    {
+      "epoch": 8.382559774964838,
+      "grad_norm": 1.3750991821289062,
+      "learning_rate": 3.143641807856898e-06,
+      "loss": 0.4537,
+      "step": 4470
+    },
+    {
+      "epoch": 8.391936240037506,
+      "grad_norm": 1.4691765308380127,
+      "learning_rate": 3.1079688726426527e-06,
+      "loss": 0.4691,
+      "step": 4475
+    },
+    {
+      "epoch": 8.401312705110174,
+      "grad_norm": 1.3552100658416748,
+      "learning_rate": 3.0724860763728767e-06,
+      "loss": 0.4293,
+      "step": 4480
+    },
+    {
+      "epoch": 8.410689170182842,
+      "grad_norm": 1.262357473373413,
+      "learning_rate": 3.037193727226445e-06,
+      "loss": 0.4705,
+      "step": 4485
+    },
+    {
+      "epoch": 8.420065635255508,
+      "grad_norm": 1.440129280090332,
+      "learning_rate": 3.0020921317281264e-06,
+      "loss": 0.4945,
+      "step": 4490
+    },
+    {
+      "epoch": 8.429442100328176,
+      "grad_norm": 1.467227578163147,
+      "learning_rate": 2.9671815947459464e-06,
+      "loss": 0.4426,
+      "step": 4495
+    },
+    {
+      "epoch": 8.438818565400844,
+      "grad_norm": 1.374812364578247,
+      "learning_rate": 2.9324624194885436e-06,
+      "loss": 0.4601,
+      "step": 4500
+    },
+    {
+      "epoch": 8.448195030473512,
+      "grad_norm": 1.269289255142212,
+      "learning_rate": 2.897934907502503e-06,
+      "loss": 0.4699,
+      "step": 4505
+    },
+    {
+      "epoch": 8.45757149554618,
+      "grad_norm": 1.4129917621612549,
+      "learning_rate": 2.8635993586697553e-06,
+      "loss": 0.4467,
+      "step": 4510
+    },
+    {
+      "epoch": 8.466947960618846,
+      "grad_norm": 1.451233148574829,
+      "learning_rate": 2.829456071204978e-06,
+      "loss": 0.4313,
+      "step": 4515
+    },
+    {
+      "epoch": 8.476324425691514,
+      "grad_norm": 1.4040919542312622,
+      "learning_rate": 2.795505341653007e-06,
+      "loss": 0.4342,
+      "step": 4520
+    },
+    {
+      "epoch": 8.485700890764182,
+      "grad_norm": 1.4583274126052856,
+      "learning_rate": 2.761747464886244e-06,
+      "loss": 0.5055,
+      "step": 4525
+    },
+    {
+      "epoch": 8.49507735583685,
+      "grad_norm": 1.350627064704895,
+      "learning_rate": 2.728182734102111e-06,
+      "loss": 0.4588,
+      "step": 4530
+    },
+    {
+      "epoch": 8.504453820909518,
+      "grad_norm": 1.3828774690628052,
+      "learning_rate": 2.6948114408204967e-06,
+      "loss": 0.4596,
+      "step": 4535
+    },
+    {
+      "epoch": 8.513830285982184,
+      "grad_norm": 1.5524414777755737,
+      "learning_rate": 2.6616338748812255e-06,
+      "loss": 0.4124,
+      "step": 4540
+    },
+    {
+      "epoch": 8.523206751054852,
+      "grad_norm": 1.3723361492156982,
+      "learning_rate": 2.628650324441548e-06,
+      "loss": 0.4617,
+      "step": 4545
+    },
+    {
+      "epoch": 8.53258321612752,
+      "grad_norm": 1.2759183645248413,
+      "learning_rate": 2.595861075973613e-06,
+      "loss": 0.4806,
+      "step": 4550
+    },
+    {
+      "epoch": 8.541959681200188,
+      "grad_norm": 1.4044135808944702,
+      "learning_rate": 2.5632664142620165e-06,
+      "loss": 0.443,
+      "step": 4555
+    },
+    {
+      "epoch": 8.551336146272854,
+      "grad_norm": 1.3814681768417358,
+      "learning_rate": 2.530866622401304e-06,
+      "loss": 0.4374,
+      "step": 4560
+    },
+    {
+      "epoch": 8.560712611345522,
+      "grad_norm": 1.384630560874939,
+      "learning_rate": 2.4986619817935152e-06,
+      "loss": 0.4929,
+      "step": 4565
+    },
+    {
+      "epoch": 8.57008907641819,
+      "grad_norm": 1.4086894989013672,
+      "learning_rate": 2.4666527721457416e-06,
+      "loss": 0.4436,
+      "step": 4570
+    },
+    {
+      "epoch": 8.579465541490858,
+      "grad_norm": 1.3705902099609375,
+      "learning_rate": 2.4348392714676984e-06,
+      "loss": 0.4618,
+      "step": 4575
+    },
+    {
+      "epoch": 8.588842006563526,
+      "grad_norm": 1.3637340068817139,
+      "learning_rate": 2.40322175606931e-06,
+      "loss": 0.445,
+      "step": 4580
+    },
+    {
+      "epoch": 8.598218471636192,
+      "grad_norm": 1.3644124269485474,
+      "learning_rate": 2.371800500558305e-06,
+      "loss": 0.4817,
+      "step": 4585
+    },
+    {
+      "epoch": 8.60759493670886,
+      "grad_norm": 1.3477566242218018,
+      "learning_rate": 2.3405757778378445e-06,
+      "loss": 0.4967,
+      "step": 4590
+    },
+    {
+      "epoch": 8.616971401781528,
+      "grad_norm": 1.400787115097046,
+      "learning_rate": 2.3095478591041276e-06,
+      "loss": 0.494,
+      "step": 4595
+    },
+    {
+      "epoch": 8.626347866854196,
+      "grad_norm": 1.5081714391708374,
+      "learning_rate": 2.278717013844059e-06,
+      "loss": 0.4938,
+      "step": 4600
+    },
+    {
+      "epoch": 8.635724331926864,
+      "grad_norm": 1.3515801429748535,
+      "learning_rate": 2.2480835098329073e-06,
+      "loss": 0.5367,
+      "step": 4605
+    },
+    {
+      "epoch": 8.64510079699953,
+      "grad_norm": 1.3089500665664673,
+      "learning_rate": 2.2176476131319707e-06,
+      "loss": 0.4752,
+      "step": 4610
+    },
+    {
+      "epoch": 8.654477262072199,
+      "grad_norm": 1.2583814859390259,
+      "learning_rate": 2.1874095880862505e-06,
+      "loss": 0.4566,
+      "step": 4615
+    },
+    {
+      "epoch": 8.663853727144867,
+      "grad_norm": 1.4750242233276367,
+      "learning_rate": 2.1573696973221922e-06,
+      "loss": 0.5024,
+      "step": 4620
+    },
+    {
+      "epoch": 8.673230192217535,
+      "grad_norm": 1.3479623794555664,
+      "learning_rate": 2.127528201745377e-06,
+      "loss": 0.4544,
+      "step": 4625
+    },
+    {
+      "epoch": 8.6826066572902,
+      "grad_norm": 1.2190600633621216,
+      "learning_rate": 2.0978853605382624e-06,
+      "loss": 0.4804,
+      "step": 4630
+    },
+    {
+      "epoch": 8.691983122362869,
+      "grad_norm": 1.2463667392730713,
+      "learning_rate": 2.0684414311579357e-06,
+      "loss": 0.4394,
+      "step": 4635
+    },
+    {
+      "epoch": 8.701359587435537,
+      "grad_norm": 1.4247466325759888,
+      "learning_rate": 2.0391966693338733e-06,
+      "loss": 0.4428,
+      "step": 4640
+    },
+    {
+      "epoch": 8.710736052508205,
+      "grad_norm": 1.4727615118026733,
+      "learning_rate": 2.010151329065721e-06,
+      "loss": 0.4277,
+      "step": 4645
+    },
+    {
+      "epoch": 8.720112517580873,
+      "grad_norm": 1.2548913955688477,
+      "learning_rate": 1.9813056626210886e-06,
+      "loss": 0.4202,
+      "step": 4650
+    },
+    {
+      "epoch": 8.729488982653539,
+      "grad_norm": 1.4516143798828125,
+      "learning_rate": 1.9526599205333536e-06,
+      "loss": 0.4236,
+      "step": 4655
+    },
+    {
+      "epoch": 8.738865447726207,
+      "grad_norm": 1.397210955619812,
+      "learning_rate": 1.9242143515994933e-06,
+      "loss": 0.5411,
+      "step": 4660
+    },
+    {
+      "epoch": 8.748241912798875,
+      "grad_norm": 1.376376986503601,
+      "learning_rate": 1.895969202877923e-06,
+      "loss": 0.4761,
+      "step": 4665
+    },
+    {
+      "epoch": 8.757618377871543,
+      "grad_norm": 1.3811997175216675,
+      "learning_rate": 1.8679247196863425e-06,
+      "loss": 0.4762,
+      "step": 4670
+    },
+    {
+      "epoch": 8.76699484294421,
+      "grad_norm": 1.288142204284668,
+      "learning_rate": 1.8400811455996092e-06,
+      "loss": 0.4814,
+      "step": 4675
+    },
+    {
+      "epoch": 8.776371308016877,
+      "grad_norm": 1.3083816766738892,
+      "learning_rate": 1.8124387224476347e-06,
+      "loss": 0.4529,
+      "step": 4680
+    },
+    {
+      "epoch": 8.785747773089545,
+      "grad_norm": 1.2878098487854004,
+      "learning_rate": 1.7849976903132588e-06,
+      "loss": 0.4391,
+      "step": 4685
+    },
+    {
+      "epoch": 8.795124238162213,
+      "grad_norm": 1.303001880645752,
+      "learning_rate": 1.757758287530195e-06,
+      "loss": 0.4894,
+      "step": 4690
+    },
+    {
+      "epoch": 8.804500703234881,
+      "grad_norm": 1.3268485069274902,
+      "learning_rate": 1.7307207506809282e-06,
+      "loss": 0.4825,
+      "step": 4695
+    },
+    {
+      "epoch": 8.813877168307549,
+      "grad_norm": 1.3460689783096313,
+      "learning_rate": 1.7038853145946804e-06,
+      "loss": 0.4902,
+      "step": 4700
+    },
+    {
+      "epoch": 8.823253633380215,
+      "grad_norm": 1.3170233964920044,
+      "learning_rate": 1.6772522123453816e-06,
+      "loss": 0.4537,
+      "step": 4705
+    },
+    {
+      "epoch": 8.832630098452883,
+      "grad_norm": 1.4489531517028809,
+      "learning_rate": 1.6508216752496141e-06,
+      "loss": 0.4396,
+      "step": 4710
+    },
+    {
+      "epoch": 8.842006563525551,
+      "grad_norm": 1.458016037940979,
+      "learning_rate": 1.624593932864632e-06,
+      "loss": 0.4696,
+      "step": 4715
+    },
+    {
+      "epoch": 8.85138302859822,
+      "grad_norm": 1.4190688133239746,
+      "learning_rate": 1.5985692129863395e-06,
+      "loss": 0.447,
+      "step": 4720
+    },
+    {
+      "epoch": 8.860759493670885,
+      "grad_norm": 1.3865045309066772,
+      "learning_rate": 1.5727477416473456e-06,
+      "loss": 0.4521,
+      "step": 4725
+    },
+    {
+      "epoch": 8.870135958743553,
+      "grad_norm": 1.3155893087387085,
+      "learning_rate": 1.547129743114978e-06,
+      "loss": 0.4531,
+      "step": 4730
+    },
+    {
+      "epoch": 8.879512423816221,
+      "grad_norm": 1.5441020727157593,
+      "learning_rate": 1.5217154398893373e-06,
+      "loss": 0.4411,
+      "step": 4735
+    },
+    {
+      "epoch": 8.88888888888889,
+      "grad_norm": 1.5488996505737305,
+      "learning_rate": 1.496505052701372e-06,
+      "loss": 0.4536,
+      "step": 4740
+    },
+    {
+      "epoch": 8.898265353961557,
+      "grad_norm": 1.4361801147460938,
+      "learning_rate": 1.471498800510962e-06,
+      "loss": 0.4556,
+      "step": 4745
+    },
+    {
+      "epoch": 8.907641819034223,
+      "grad_norm": 1.4031137228012085,
+      "learning_rate": 1.4466969005050013e-06,
+      "loss": 0.471,
+      "step": 4750
+    },
+    {
+      "epoch": 8.917018284106891,
+      "grad_norm": 1.3877636194229126,
+      "learning_rate": 1.4220995680955417e-06,
+      "loss": 0.4636,
+      "step": 4755
+    },
+    {
+      "epoch": 8.92639474917956,
+      "grad_norm": 1.4993648529052734,
+      "learning_rate": 1.3977070169178763e-06,
+      "loss": 0.4869,
+      "step": 4760
+    },
+    {
+      "epoch": 8.935771214252227,
+      "grad_norm": 1.4594330787658691,
+      "learning_rate": 1.373519458828737e-06,
+      "loss": 0.4603,
+      "step": 4765
+    },
+    {
+      "epoch": 8.945147679324894,
+      "grad_norm": 1.3764735460281372,
+      "learning_rate": 1.349537103904408e-06,
+      "loss": 0.4551,
+      "step": 4770
+    },
+    {
+      "epoch": 8.954524144397562,
+      "grad_norm": 1.3424842357635498,
+      "learning_rate": 1.3257601604389464e-06,
+      "loss": 0.4682,
+      "step": 4775
+    },
+    {
+      "epoch": 8.96390060947023,
+      "grad_norm": 1.338976502418518,
+      "learning_rate": 1.3021888349423222e-06,
+      "loss": 0.4869,
+      "step": 4780
+    },
+    {
+      "epoch": 8.973277074542898,
+      "grad_norm": 1.3037632703781128,
+      "learning_rate": 1.2788233321386621e-06,
+      "loss": 0.4708,
+      "step": 4785
+    },
+    {
+      "epoch": 8.982653539615566,
+      "grad_norm": 1.2927602529525757,
+      "learning_rate": 1.2556638549644644e-06,
+      "loss": 0.4763,
+      "step": 4790
+    },
+    {
+      "epoch": 8.992030004688232,
+      "grad_norm": 1.396018385887146,
+      "learning_rate": 1.2327106045668279e-06,
+      "loss": 0.4321,
+      "step": 4795
+    },
+    {
+      "epoch": 9.0014064697609,
+      "grad_norm": 1.4099509716033936,
+      "learning_rate": 1.2099637803016983e-06,
+      "loss": 0.4568,
+      "step": 4800
+    },
+    {
+      "epoch": 9.010782934833568,
+      "grad_norm": 1.358614444732666,
+      "learning_rate": 1.187423579732172e-06,
+      "loss": 0.4319,
+      "step": 4805
+    },
+    {
+      "epoch": 9.020159399906236,
+      "grad_norm": 1.388286828994751,
+      "learning_rate": 1.1650901986267365e-06,
+      "loss": 0.4318,
+      "step": 4810
+    },
+    {
+      "epoch": 9.029535864978904,
+      "grad_norm": 1.297178030014038,
+      "learning_rate": 1.142963830957594e-06,
+      "loss": 0.4779,
+      "step": 4815
+    },
+    {
+      "epoch": 9.03891233005157,
+      "grad_norm": 1.4789464473724365,
+      "learning_rate": 1.1210446688989768e-06,
+      "loss": 0.4774,
+      "step": 4820
+    },
+    {
+      "epoch": 9.048288795124238,
+      "grad_norm": 1.3909670114517212,
+      "learning_rate": 1.0993329028254644e-06,
+      "loss": 0.4502,
+      "step": 4825
+    },
+    {
+      "epoch": 9.057665260196906,
+      "grad_norm": 1.2993029356002808,
+      "learning_rate": 1.0778287213103478e-06,
+      "loss": 0.4914,
+      "step": 4830
+    },
+    {
+      "epoch": 9.067041725269574,
+      "grad_norm": 1.3750618696212769,
+      "learning_rate": 1.0565323111239783e-06,
+      "loss": 0.5026,
+      "step": 4835
+    },
+    {
+      "epoch": 9.076418190342242,
+      "grad_norm": 1.5192205905914307,
+      "learning_rate": 1.0354438572321546e-06,
+      "loss": 0.4502,
+      "step": 4840
+    },
+    {
+      "epoch": 9.085794655414908,
+      "grad_norm": 1.319626808166504,
+      "learning_rate": 1.0145635427945028e-06,
+      "loss": 0.4581,
+      "step": 4845
+    },
+    {
+      "epoch": 9.095171120487576,
+      "grad_norm": 1.2741262912750244,
+      "learning_rate": 9.938915491629063e-07,
+      "loss": 0.4844,
+      "step": 4850
+    },
+    {
+      "epoch": 9.104547585560244,
+      "grad_norm": 1.4752476215362549,
+      "learning_rate": 9.734280558799102e-07,
+      "loss": 0.4665,
+      "step": 4855
+    },
+    {
+      "epoch": 9.113924050632912,
+      "grad_norm": 1.3631614446640015,
+      "learning_rate": 9.531732406771771e-07,
+      "loss": 0.4676,
+      "step": 4860
+    },
+    {
+      "epoch": 9.123300515705578,
+      "grad_norm": 1.3976688385009766,
+      "learning_rate": 9.331272794739276e-07,
+      "loss": 0.4515,
+      "step": 4865
+    },
+    {
+      "epoch": 9.132676980778246,
+      "grad_norm": 1.3497931957244873,
+      "learning_rate": 9.132903463754256e-07,
+      "loss": 0.4952,
+      "step": 4870
+    },
+    {
+      "epoch": 9.142053445850914,
+      "grad_norm": 1.49704909324646,
+      "learning_rate": 8.936626136714754e-07,
+      "loss": 0.4212,
+      "step": 4875
+    },
+    {
+      "epoch": 9.151429910923582,
+      "grad_norm": 1.2458924055099487,
+      "learning_rate": 8.742442518348965e-07,
+      "loss": 0.4894,
+      "step": 4880
+    },
+    {
+      "epoch": 9.16080637599625,
+      "grad_norm": 1.3910996913909912,
+      "learning_rate": 8.550354295200596e-07,
+      "loss": 0.4994,
+      "step": 4885
+    },
+    {
+      "epoch": 9.170182841068916,
+      "grad_norm": 1.3585845232009888,
+      "learning_rate": 8.360363135614307e-07,
+      "loss": 0.4901,
+      "step": 4890
+    },
+    {
+      "epoch": 9.179559306141584,
+      "grad_norm": 1.2690738439559937,
+      "learning_rate": 8.172470689721018e-07,
+      "loss": 0.439,
+      "step": 4895
+    },
+    {
+      "epoch": 9.188935771214252,
+      "grad_norm": 1.4131815433502197,
+      "learning_rate": 7.986678589423758e-07,
+      "loss": 0.4916,
+      "step": 4900
+    },
+    {
+      "epoch": 9.19831223628692,
+      "grad_norm": 1.5426223278045654,
+      "learning_rate": 7.802988448383398e-07,
+      "loss": 0.4551,
+      "step": 4905
+    },
+    {
+      "epoch": 9.207688701359588,
+      "grad_norm": 1.3847144842147827,
+      "learning_rate": 7.621401862004634e-07,
+      "loss": 0.4374,
+      "step": 4910
+    },
+    {
+      "epoch": 9.217065166432254,
+      "grad_norm": 1.2885124683380127,
+      "learning_rate": 7.44192040742217e-07,
+      "loss": 0.4394,
+      "step": 4915
+    },
+    {
+      "epoch": 9.226441631504922,
+      "grad_norm": 1.4092646837234497,
+      "learning_rate": 7.264545643486997e-07,
+      "loss": 0.4776,
+      "step": 4920
+    },
+    {
+      "epoch": 9.23581809657759,
+      "grad_norm": 1.365407109260559,
+      "learning_rate": 7.089279110752856e-07,
+      "loss": 0.4765,
+      "step": 4925
+    },
+    {
+      "epoch": 9.245194561650258,
+      "grad_norm": 1.4221059083938599,
+      "learning_rate": 6.916122331462799e-07,
+      "loss": 0.4483,
+      "step": 4930
+    },
+    {
+      "epoch": 9.254571026722925,
+      "grad_norm": 1.3975940942764282,
+      "learning_rate": 6.74507680953615e-07,
+      "loss": 0.4666,
+      "step": 4935
+    },
+    {
+      "epoch": 9.263947491795593,
+      "grad_norm": 1.2930744886398315,
+      "learning_rate": 6.576144030555259e-07,
+      "loss": 0.4642,
+      "step": 4940
+    },
+    {
+      "epoch": 9.27332395686826,
+      "grad_norm": 1.360459566116333,
+      "learning_rate": 6.409325461752602e-07,
+      "loss": 0.4778,
+      "step": 4945
+    },
+    {
+      "epoch": 9.282700421940929,
+      "grad_norm": 1.3444581031799316,
+      "learning_rate": 6.244622551998203e-07,
+      "loss": 0.5415,
+      "step": 4950
+    },
+    {
+      "epoch": 9.292076887013597,
+      "grad_norm": 1.3890564441680908,
+      "learning_rate": 6.082036731786895e-07,
+      "loss": 0.4408,
+      "step": 4955
+    },
+    {
+      "epoch": 9.301453352086263,
+      "grad_norm": 1.3739839792251587,
+      "learning_rate": 5.921569413225913e-07,
+      "loss": 0.4813,
+      "step": 4960
+    },
+    {
+      "epoch": 9.31082981715893,
+      "grad_norm": 1.3291054964065552,
+      "learning_rate": 5.763221990022743e-07,
+      "loss": 0.4678,
+      "step": 4965
+    },
+    {
+      "epoch": 9.320206282231599,
+      "grad_norm": 1.2954983711242676,
+      "learning_rate": 5.606995837472817e-07,
+      "loss": 0.4045,
+      "step": 4970
+    },
+    {
+      "epoch": 9.329582747304267,
+      "grad_norm": 1.3745497465133667,
+      "learning_rate": 5.452892312447777e-07,
+      "loss": 0.4684,
+      "step": 4975
+    },
+    {
+      "epoch": 9.338959212376935,
+      "grad_norm": 1.3169890642166138,
+      "learning_rate": 5.300912753383625e-07,
+      "loss": 0.5042,
+      "step": 4980
+    },
+    {
+      "epoch": 9.348335677449601,
+      "grad_norm": 1.461635708808899,
+      "learning_rate": 5.151058480269006e-07,
+      "loss": 0.4415,
+      "step": 4985
+    },
+    {
+      "epoch": 9.357712142522269,
+      "grad_norm": 1.3662902116775513,
+      "learning_rate": 5.003330794633776e-07,
+      "loss": 0.5308,
+      "step": 4990
+    },
+    {
+      "epoch": 9.367088607594937,
+      "grad_norm": 1.3519859313964844,
+      "learning_rate": 4.857730979537873e-07,
+      "loss": 0.4722,
+      "step": 4995
+    },
+    {
+      "epoch": 9.376465072667605,
+      "grad_norm": 1.4192826747894287,
+      "learning_rate": 4.714260299559875e-07,
+      "loss": 0.4407,
+      "step": 5000
+    },
+    {
+      "epoch": 9.385841537740271,
+      "grad_norm": 1.2914866209030151,
+      "learning_rate": 4.5729200007862683e-07,
+      "loss": 0.4575,
+      "step": 5005
+    },
+    {
+      "epoch": 9.395218002812939,
+      "grad_norm": 1.4089744091033936,
+      "learning_rate": 4.4337113108005314e-07,
+      "loss": 0.4909,
+      "step": 5010
+    },
+    {
+      "epoch": 9.404594467885607,
+      "grad_norm": 1.3321195840835571,
+      "learning_rate": 4.296635438672425e-07,
+      "loss": 0.518,
+      "step": 5015
+    },
+    {
+      "epoch": 9.413970932958275,
+      "grad_norm": 1.284547209739685,
+      "learning_rate": 4.161693574947556e-07,
+      "loss": 0.4587,
+      "step": 5020
+    },
+    {
+      "epoch": 9.423347398030943,
+      "grad_norm": 1.3978044986724854,
+      "learning_rate": 4.0288868916370514e-07,
+      "loss": 0.4571,
+      "step": 5025
+    },
+    {
+      "epoch": 9.43272386310361,
+      "grad_norm": 1.3437304496765137,
+      "learning_rate": 3.8982165422073445e-07,
+      "loss": 0.447,
+      "step": 5030
+    },
+    {
+      "epoch": 9.442100328176277,
+      "grad_norm": 1.3242155313491821,
+      "learning_rate": 3.7696836615700727e-07,
+      "loss": 0.4907,
+      "step": 5035
+    },
+    {
+      "epoch": 9.451476793248945,
+      "grad_norm": 1.3593541383743286,
+      "learning_rate": 3.6432893660723886e-07,
+      "loss": 0.4578,
+      "step": 5040
+    },
+    {
+      "epoch": 9.460853258321613,
+      "grad_norm": 1.454704999923706,
+      "learning_rate": 3.519034753487166e-07,
+      "loss": 0.5003,
+      "step": 5045
+    },
+    {
+      "epoch": 9.470229723394281,
+      "grad_norm": 1.438981056213379,
+      "learning_rate": 3.396920903003559e-07,
+      "loss": 0.472,
+      "step": 5050
+    },
+    {
+      "epoch": 9.479606188466947,
+      "grad_norm": 1.2579084634780884,
+      "learning_rate": 3.276948875217373e-07,
+      "loss": 0.4534,
+      "step": 5055
+    },
+    {
+      "epoch": 9.488982653539615,
+      "grad_norm": 1.3231197595596313,
+      "learning_rate": 3.1591197121222107e-07,
+      "loss": 0.4678,
+      "step": 5060
+    },
+    {
+      "epoch": 9.498359118612283,
+      "grad_norm": 1.4480105638504028,
+      "learning_rate": 3.043434437100118e-07,
+      "loss": 0.4331,
+      "step": 5065
+    },
+    {
+      "epoch": 9.507735583684951,
+      "grad_norm": 1.2928885221481323,
+      "learning_rate": 2.9298940549128964e-07,
+      "loss": 0.4439,
+      "step": 5070
+    },
+    {
+      "epoch": 9.517112048757618,
+      "grad_norm": 1.4796031713485718,
+      "learning_rate": 2.818499551693221e-07,
+      "loss": 0.4496,
+      "step": 5075
+    },
+    {
+      "epoch": 9.526488513830285,
+      "grad_norm": 1.4499142169952393,
+      "learning_rate": 2.7092518949362875e-07,
+      "loss": 0.453,
+      "step": 5080
+    },
+    {
+      "epoch": 9.535864978902953,
+      "grad_norm": 1.4941402673721313,
+      "learning_rate": 2.6021520334911786e-07,
+      "loss": 0.5147,
+      "step": 5085
+    },
+    {
+      "epoch": 9.545241443975621,
+      "grad_norm": 1.256858468055725,
+      "learning_rate": 2.4972008975527593e-07,
+      "loss": 0.4778,
+      "step": 5090
+    },
+    {
+      "epoch": 9.55461790904829,
+      "grad_norm": 1.2863494157791138,
+      "learning_rate": 2.3943993986534905e-07,
+      "loss": 0.4641,
+      "step": 5095
+    },
+    {
+      "epoch": 9.563994374120956,
+      "grad_norm": 1.4388316869735718,
+      "learning_rate": 2.2937484296556566e-07,
+      "loss": 0.4418,
+      "step": 5100
+    },
+    {
+      "epoch": 9.573370839193624,
+      "grad_norm": 1.2195985317230225,
+      "learning_rate": 2.1952488647435654e-07,
+      "loss": 0.5068,
+      "step": 5105
+    },
+    {
+      "epoch": 9.582747304266292,
+      "grad_norm": 1.4117023944854736,
+      "learning_rate": 2.0989015594158058e-07,
+      "loss": 0.42,
+      "step": 5110
+    },
+    {
+      "epoch": 9.59212376933896,
+      "grad_norm": 1.4613057374954224,
+      "learning_rate": 2.0047073504780579e-07,
+      "loss": 0.4307,
+      "step": 5115
+    },
+    {
+      "epoch": 9.601500234411628,
+      "grad_norm": 1.3232667446136475,
+      "learning_rate": 1.9126670560356553e-07,
+      "loss": 0.4639,
+      "step": 5120
+    },
+    {
+      "epoch": 9.610876699484294,
+      "grad_norm": 1.3385889530181885,
+      "learning_rate": 1.8227814754865068e-07,
+      "loss": 0.4664,
+      "step": 5125
+    },
+    {
+      "epoch": 9.620253164556962,
+      "grad_norm": 1.4078065156936646,
+      "learning_rate": 1.735051389514214e-07,
+      "loss": 0.4853,
+      "step": 5130
+    },
+    {
+      "epoch": 9.62962962962963,
+      "grad_norm": 1.393649697303772,
+      "learning_rate": 1.6494775600812417e-07,
+      "loss": 0.4494,
+      "step": 5135
+    },
+    {
+      "epoch": 9.639006094702298,
+      "grad_norm": 1.3628828525543213,
+      "learning_rate": 1.5660607304223141e-07,
+      "loss": 0.5302,
+      "step": 5140
+    },
+    {
+      "epoch": 9.648382559774966,
+      "grad_norm": 1.385543942451477,
+      "learning_rate": 1.4848016250378904e-07,
+      "loss": 0.4483,
+      "step": 5145
+    },
+    {
+      "epoch": 9.657759024847632,
+      "grad_norm": 1.5173555612564087,
+      "learning_rate": 1.4057009496881158e-07,
+      "loss": 0.48,
+      "step": 5150
+    },
+    {
+      "epoch": 9.6671354899203,
+      "grad_norm": 1.2662690877914429,
+      "learning_rate": 1.3287593913862972e-07,
+      "loss": 0.4586,
+      "step": 5155
+    },
+    {
+      "epoch": 9.676511954992968,
+      "grad_norm": 1.4856374263763428,
+      "learning_rate": 1.2539776183932982e-07,
+      "loss": 0.4381,
+      "step": 5160
+    },
+    {
+      "epoch": 9.685888420065636,
+      "grad_norm": 1.289402961730957,
+      "learning_rate": 1.1813562802115708e-07,
+      "loss": 0.4319,
+      "step": 5165
+    },
+    {
+      "epoch": 9.695264885138302,
+      "grad_norm": 1.3352842330932617,
+      "learning_rate": 1.1108960075794372e-07,
+      "loss": 0.4772,
+      "step": 5170
+    },
+    {
+      "epoch": 9.70464135021097,
+      "grad_norm": 1.4623725414276123,
+      "learning_rate": 1.0425974124658455e-07,
+      "loss": 0.5125,
+      "step": 5175
+    },
+    {
+      "epoch": 9.714017815283638,
+      "grad_norm": 1.2473069429397583,
+      "learning_rate": 9.764610880648451e-08,
+      "loss": 0.4311,
+      "step": 5180
+    },
+    {
+      "epoch": 9.723394280356306,
+      "grad_norm": 1.4001269340515137,
+      "learning_rate": 9.124876087904799e-08,
+      "loss": 0.5005,
+      "step": 5185
+    },
+    {
+      "epoch": 9.732770745428974,
+      "grad_norm": 1.2903003692626953,
+      "learning_rate": 8.506775302719039e-08,
+      "loss": 0.4248,
+      "step": 5190
+    },
+    {
+      "epoch": 9.74214721050164,
+      "grad_norm": 1.4147828817367554,
+      "learning_rate": 7.910313893484123e-08,
+      "loss": 0.4287,
+      "step": 5195
+    },
+    {
+      "epoch": 9.751523675574308,
+      "grad_norm": 1.3648351430892944,
+      "learning_rate": 7.335497040648898e-08,
+      "loss": 0.4535,
+      "step": 5200
+    },
+    {
+      "epoch": 9.760900140646976,
+      "grad_norm": 1.4243569374084473,
+      "learning_rate": 6.78232973667231e-08,
+      "loss": 0.4731,
+      "step": 5205
+    },
+    {
+      "epoch": 9.770276605719644,
+      "grad_norm": 1.604968786239624,
+      "learning_rate": 6.250816785980385e-08,
+      "loss": 0.473,
+      "step": 5210
+    },
+    {
+      "epoch": 9.77965307079231,
+      "grad_norm": 1.3988664150238037,
+      "learning_rate": 5.7409628049245877e-08,
+      "loss": 0.4412,
+      "step": 5215
+    },
+    {
+      "epoch": 9.789029535864978,
+      "grad_norm": 1.3431826829910278,
+      "learning_rate": 5.2527722217421416e-08,
+      "loss": 0.4863,
+      "step": 5220
+    },
+    {
+      "epoch": 9.798406000937646,
+      "grad_norm": 1.5021179914474487,
+      "learning_rate": 4.786249276516608e-08,
+      "loss": 0.4684,
+      "step": 5225
+    },
+    {
+      "epoch": 9.807782466010314,
+      "grad_norm": 1.2139686346054077,
+      "learning_rate": 4.3413980211412516e-08,
+      "loss": 0.432,
+      "step": 5230
+    },
+    {
+      "epoch": 9.817158931082982,
+      "grad_norm": 1.4108506441116333,
+      "learning_rate": 3.918222319284348e-08,
+      "loss": 0.4606,
+      "step": 5235
+    },
+    {
+      "epoch": 9.826535396155649,
+      "grad_norm": 1.3711954355239868,
+      "learning_rate": 3.516725846355873e-08,
+      "loss": 0.4296,
+      "step": 5240
+    },
+    {
+      "epoch": 9.835911861228317,
+      "grad_norm": 1.3619983196258545,
+      "learning_rate": 3.136912089474753e-08,
+      "loss": 0.4437,
+      "step": 5245
+    },
+    {
+      "epoch": 9.845288326300984,
+      "grad_norm": 1.3800973892211914,
+      "learning_rate": 2.7787843474386123e-08,
+      "loss": 0.4616,
+      "step": 5250
+    },
+    {
+      "epoch": 9.854664791373652,
+      "grad_norm": 1.390395998954773,
+      "learning_rate": 2.4423457306957388e-08,
+      "loss": 0.4561,
+      "step": 5255
+    },
+    {
+      "epoch": 9.86404125644632,
+      "grad_norm": 1.2912839651107788,
+      "learning_rate": 2.127599161318161e-08,
+      "loss": 0.4436,
+      "step": 5260
+    },
+    {
+      "epoch": 9.873417721518987,
+      "grad_norm": 1.3427574634552002,
+      "learning_rate": 1.834547372975004e-08,
+      "loss": 0.4756,
+      "step": 5265
+    },
+    {
+      "epoch": 9.882794186591655,
+      "grad_norm": 1.3428921699523926,
+      "learning_rate": 1.5631929109102828e-08,
+      "loss": 0.4342,
+      "step": 5270
+    },
+    {
+      "epoch": 9.892170651664323,
+      "grad_norm": 1.280137538909912,
+      "learning_rate": 1.3135381319204221e-08,
+      "loss": 0.5266,
+      "step": 5275
+    },
+    {
+      "epoch": 9.90154711673699,
+      "grad_norm": 1.453433632850647,
+      "learning_rate": 1.0855852043323289e-08,
+      "loss": 0.4431,
+      "step": 5280
+    },
+    {
+      "epoch": 9.910923581809659,
+      "grad_norm": 1.3490616083145142,
+      "learning_rate": 8.793361079870167e-09,
+      "loss": 0.4554,
+      "step": 5285
+    },
+    {
+      "epoch": 9.920300046882325,
+      "grad_norm": 1.2622390985488892,
+      "learning_rate": 6.947926342204536e-09,
+      "loss": 0.4767,
+      "step": 5290
+    },
+    {
+      "epoch": 9.929676511954993,
+      "grad_norm": 1.3238799571990967,
+      "learning_rate": 5.31956385848853e-09,
+      "loss": 0.4756,
+      "step": 5295
+    },
+    {
+      "epoch": 9.93905297702766,
+      "grad_norm": 1.2758700847625732,
+      "learning_rate": 3.908287771542396e-09,
+      "loss": 0.4476,
+      "step": 5300
+    },
+    {
+      "epoch": 9.948429442100329,
+      "grad_norm": 1.4579707384109497,
+      "learning_rate": 2.7141103387307022e-09,
+      "loss": 0.469,
+      "step": 5305
+    },
+    {
+      "epoch": 9.957805907172995,
+      "grad_norm": 1.3778728246688843,
+      "learning_rate": 1.737041931845762e-09,
+      "loss": 0.4784,
+      "step": 5310
+    },
+    {
+      "epoch": 9.967182372245663,
+      "grad_norm": 1.6131629943847656,
+      "learning_rate": 9.770910370243692e-10,
+      "loss": 0.4766,
+      "step": 5315
+    },
+    {
+      "epoch": 9.976558837318331,
+      "grad_norm": 1.383500576019287,
+      "learning_rate": 4.3426425467008035e-10,
+      "loss": 0.4395,
+      "step": 5320
+    },
+    {
+      "epoch": 9.985935302390999,
+      "grad_norm": 1.423774242401123,
+      "learning_rate": 1.0856629940048101e-10,
+      "loss": 0.4473,
+      "step": 5325
+    },
+    {
+      "epoch": 9.995311767463667,
+      "grad_norm": 1.3990870714187622,
+      "learning_rate": 0.0,
+      "loss": 0.4661,
+      "step": 5330
+    },
+    {
+      "epoch": 9.995311767463667,
+      "step": 5330,
+      "total_flos": 6.079064529052828e+18,
+      "train_loss": 0.5077758979171123,
+      "train_runtime": 117216.291,
+      "train_samples_per_second": 0.728,
+      "train_steps_per_second": 0.045
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 5330,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 100,
+  "total_flos": 6.079064529052828e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}