{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.000162162025201,
  "eval_steps": 3084,
  "global_step": 12336,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016216202520099224,
      "grad_norm": 1.1505268812179565,
      "learning_rate": 6.091370558375634e-06,
      "loss": 3.27203106880188,
      "step": 2
    },
    {
      "epoch": 0.0003243240504019845,
      "grad_norm": 1.1518614292144775,
      "learning_rate": 1.82741116751269e-05,
      "loss": 3.2960915565490723,
      "step": 4
    },
    {
      "epoch": 0.0004864860756029767,
      "grad_norm": 1.1301591396331787,
      "learning_rate": 3.0456852791878168e-05,
      "loss": 3.3159894943237305,
      "step": 6
    },
    {
      "epoch": 0.000648648100803969,
      "grad_norm": 1.154915690422058,
      "learning_rate": 4.263959390862944e-05,
      "loss": 3.3586864471435547,
      "step": 8
    },
    {
      "epoch": 0.0008108101260049612,
      "grad_norm": 1.088661789894104,
      "learning_rate": 5.4822335025380706e-05,
      "loss": 3.29746675491333,
      "step": 10
    },
    {
      "epoch": 0.0009729721512059534,
      "grad_norm": 1.0727648735046387,
      "learning_rate": 6.700507614213197e-05,
      "loss": 3.255526065826416,
      "step": 12
    },
    {
      "epoch": 0.0011351341764069456,
      "grad_norm": 1.0574971437454224,
      "learning_rate": 7.918781725888325e-05,
      "loss": 3.2961537837982178,
      "step": 14
    },
    {
      "epoch": 0.001297296201607938,
      "grad_norm": 1.0183849334716797,
      "learning_rate": 9.137055837563452e-05,
      "loss": 3.2811856269836426,
      "step": 16
    },
    {
      "epoch": 0.00145945822680893,
      "grad_norm": 1.0246390104293823,
      "learning_rate": 0.00010355329949238577,
      "loss": 3.341029167175293,
      "step": 18
    },
    {
      "epoch": 0.0016216202520099223,
      "grad_norm": 1.0190030336380005,
      "learning_rate": 0.00011573604060913704,
      "loss": 3.3245861530303955,
      "step": 20
    },
    {
      "epoch": 0.0017837822772109144,
      "grad_norm": 0.9651017785072327,
      "learning_rate": 0.00012791878172588832,
      "loss": 3.353879451751709,
      "step": 22
    },
    {
      "epoch": 0.0019459443024119067,
      "grad_norm": 0.9217729568481445,
      "learning_rate": 0.00014010152284263958,
      "loss": 3.2668230533599854,
      "step": 24
    },
    {
      "epoch": 0.002108106327612899,
      "grad_norm": 0.8605936765670776,
      "learning_rate": 0.00015228426395939084,
      "loss": 3.2448906898498535,
      "step": 26
    },
    {
      "epoch": 0.002270268352813891,
      "grad_norm": 0.8758831024169922,
      "learning_rate": 0.0001644670050761421,
      "loss": 3.3152410984039307,
      "step": 28
    },
    {
      "epoch": 0.0024324303780148835,
      "grad_norm": 0.8324900269508362,
      "learning_rate": 0.00017664974619289339,
      "loss": 3.2823314666748047,
      "step": 30
    },
    {
      "epoch": 0.002594592403215876,
      "grad_norm": 0.7941120266914368,
      "learning_rate": 0.00018883248730964465,
      "loss": 3.1972389221191406,
      "step": 32
    },
    {
      "epoch": 0.002756754428416868,
      "grad_norm": 0.7831990718841553,
      "learning_rate": 0.00020101522842639593,
      "loss": 3.2079737186431885,
      "step": 34
    },
    {
      "epoch": 0.00291891645361786,
      "grad_norm": 0.7255515456199646,
      "learning_rate": 0.0002131979695431472,
      "loss": 3.1902036666870117,
      "step": 36
    },
    {
      "epoch": 0.0030810784788188523,
      "grad_norm": 0.6976752281188965,
      "learning_rate": 0.00022538071065989846,
      "loss": 3.2157955169677734,
      "step": 38
    },
    {
      "epoch": 0.0032432405040198446,
      "grad_norm": 0.6698542833328247,
      "learning_rate": 0.00023756345177664974,
      "loss": 3.19138240814209,
      "step": 40
    },
    {
      "epoch": 0.003405402529220837,
      "grad_norm": 0.6469593644142151,
      "learning_rate": 0.00024974619289340103,
      "loss": 3.1880784034729004,
      "step": 42
    },
    {
      "epoch": 0.003567564554421829,
      "grad_norm": 0.628739058971405,
      "learning_rate": 0.00026192893401015224,
      "loss": 3.2353568077087402,
      "step": 44
    },
    {
      "epoch": 0.003729726579622821,
      "grad_norm": 0.5827344059944153,
      "learning_rate": 0.0002741116751269035,
      "loss": 3.196615695953369,
      "step": 46
    },
    {
      "epoch": 0.0038918886048238135,
      "grad_norm": 0.559890866279602,
      "learning_rate": 0.0002862944162436548,
      "loss": 3.195606231689453,
      "step": 48
    },
    {
      "epoch": 0.004054050630024806,
      "grad_norm": 0.5410118103027344,
      "learning_rate": 0.0002984771573604061,
      "loss": 3.1604042053222656,
      "step": 50
    },
    {
      "epoch": 0.004216212655225798,
      "grad_norm": 0.521488606929779,
      "learning_rate": 0.00031065989847715733,
      "loss": 3.1378462314605713,
      "step": 52
    },
    {
      "epoch": 0.00437837468042679,
      "grad_norm": 0.4823688864707947,
      "learning_rate": 0.0003228426395939086,
      "loss": 3.137131690979004,
      "step": 54
    },
    {
      "epoch": 0.004540536705627782,
      "grad_norm": 0.4367760121822357,
      "learning_rate": 0.00033502538071065985,
      "loss": 3.1108250617980957,
      "step": 56
    },
    {
      "epoch": 0.004702698730828775,
      "grad_norm": 0.42890316247940063,
      "learning_rate": 0.0003472081218274111,
      "loss": 3.0908260345458984,
      "step": 58
    },
    {
      "epoch": 0.004864860756029767,
      "grad_norm": 0.4048822522163391,
      "learning_rate": 0.00035939086294416243,
      "loss": 3.14139461517334,
      "step": 60
    },
    {
      "epoch": 0.005027022781230759,
      "grad_norm": 0.3952430784702301,
      "learning_rate": 0.00037157360406091364,
      "loss": 3.189903736114502,
      "step": 62
    },
    {
      "epoch": 0.005189184806431752,
      "grad_norm": 0.37357133626937866,
      "learning_rate": 0.00038375634517766495,
      "loss": 3.1876959800720215,
      "step": 64
    },
    {
      "epoch": 0.0053513468316327435,
      "grad_norm": 0.3550533652305603,
      "learning_rate": 0.0003959390862944162,
      "loss": 3.1674551963806152,
      "step": 66
    },
    {
      "epoch": 0.005513508856833736,
      "grad_norm": 0.3177494406700134,
      "learning_rate": 0.0004081218274111675,
      "loss": 3.101983070373535,
      "step": 68
    },
    {
      "epoch": 0.005675670882034728,
      "grad_norm": 0.3120107650756836,
      "learning_rate": 0.00042030456852791873,
      "loss": 3.1081576347351074,
      "step": 70
    },
    {
      "epoch": 0.00583783290723572,
      "grad_norm": 0.28324973583221436,
      "learning_rate": 0.00043248730964467005,
      "loss": 3.113741397857666,
      "step": 72
    },
    {
      "epoch": 0.005999994932436713,
      "grad_norm": 0.26636359095573425,
      "learning_rate": 0.0004446700507614213,
      "loss": 3.105936288833618,
      "step": 74
    },
    {
      "epoch": 0.006162156957637705,
      "grad_norm": 0.27505505084991455,
      "learning_rate": 0.0004568527918781725,
      "loss": 3.202652931213379,
      "step": 76
    },
    {
      "epoch": 0.006324318982838697,
      "grad_norm": 0.24881111085414886,
      "learning_rate": 0.00046903553299492383,
      "loss": 3.1451287269592285,
      "step": 78
    },
    {
      "epoch": 0.006486481008039689,
      "grad_norm": 0.2395073026418686,
      "learning_rate": 0.00048121827411167504,
      "loss": 3.1096341609954834,
      "step": 80
    },
    {
      "epoch": 0.006648643033240681,
      "grad_norm": 0.2236504703760147,
      "learning_rate": 0.0004934010152284263,
      "loss": 3.150496244430542,
      "step": 82
    },
    {
      "epoch": 0.006810805058441674,
      "grad_norm": 0.20683768391609192,
      "learning_rate": 0.0005055837563451776,
      "loss": 3.054837703704834,
      "step": 84
    },
    {
      "epoch": 0.006972967083642666,
      "grad_norm": 0.2191886156797409,
      "learning_rate": 0.0005177664974619289,
      "loss": 3.0978028774261475,
      "step": 86
    },
    {
      "epoch": 0.007135129108843658,
      "grad_norm": 0.20551079511642456,
      "learning_rate": 0.0005299492385786802,
      "loss": 3.1202709674835205,
      "step": 88
    },
    {
      "epoch": 0.00729729113404465,
      "grad_norm": 0.2187616527080536,
      "learning_rate": 0.0005421319796954315,
      "loss": 3.087888717651367,
      "step": 90
    },
    {
      "epoch": 0.007459453159245642,
      "grad_norm": 0.22148892283439636,
      "learning_rate": 0.0005543147208121827,
      "loss": 3.033322334289551,
      "step": 92
    },
    {
      "epoch": 0.007621615184446635,
      "grad_norm": 0.19334140419960022,
      "learning_rate": 0.000566497461928934,
      "loss": 3.1089961528778076,
      "step": 94
    },
    {
      "epoch": 0.007783777209647627,
      "grad_norm": 0.2052568793296814,
      "learning_rate": 0.0005786802030456852,
      "loss": 3.137206554412842,
      "step": 96
    },
    {
      "epoch": 0.007945939234848619,
      "grad_norm": 0.1829306185245514,
      "learning_rate": 0.0005908629441624364,
      "loss": 3.0947084426879883,
      "step": 98
    },
    {
      "epoch": 0.008108101260049612,
      "grad_norm": 0.1768820881843567,
      "learning_rate": 0.0006030456852791877,
      "loss": 3.082709312438965,
      "step": 100
    },
    {
      "epoch": 0.008270263285250604,
      "grad_norm": 0.1676890105009079,
      "learning_rate": 0.0006152284263959391,
      "loss": 3.1184029579162598,
      "step": 102
    },
    {
      "epoch": 0.008432425310451595,
      "grad_norm": 0.16701531410217285,
      "learning_rate": 0.0006274111675126903,
      "loss": 3.075084686279297,
      "step": 104
    },
    {
      "epoch": 0.008594587335652588,
      "grad_norm": 0.15712954103946686,
      "learning_rate": 0.0006395939086294416,
      "loss": 3.082538366317749,
      "step": 106
    },
    {
      "epoch": 0.00875674936085358,
      "grad_norm": 0.1508082002401352,
      "learning_rate": 0.0006517766497461928,
      "loss": 3.0721402168273926,
      "step": 108
    },
    {
      "epoch": 0.008918911386054572,
      "grad_norm": 0.14864599704742432,
      "learning_rate": 0.0006639593908629441,
      "loss": 3.0853652954101562,
      "step": 110
    },
    {
      "epoch": 0.009081073411255565,
      "grad_norm": 0.13963720202445984,
      "learning_rate": 0.0006761421319796954,
      "loss": 3.061211347579956,
      "step": 112
    },
    {
      "epoch": 0.009243235436456557,
      "grad_norm": 0.12978024780750275,
      "learning_rate": 0.0006883248730964467,
      "loss": 3.0585741996765137,
      "step": 114
    },
    {
      "epoch": 0.00940539746165755,
      "grad_norm": 0.1366254985332489,
      "learning_rate": 0.0007005076142131978,
      "loss": 3.1098732948303223,
      "step": 116
    },
    {
      "epoch": 0.009567559486858541,
      "grad_norm": 0.1287386566400528,
      "learning_rate": 0.0007126903553299491,
      "loss": 3.0545196533203125,
      "step": 118
    },
    {
      "epoch": 0.009729721512059534,
      "grad_norm": 0.11731719225645065,
      "learning_rate": 0.0007248730964467005,
      "loss": 3.0955309867858887,
      "step": 120
    },
    {
      "epoch": 0.009891883537260527,
      "grad_norm": 0.12735258042812347,
      "learning_rate": 0.0007370558375634517,
      "loss": 3.0806546211242676,
      "step": 122
    },
    {
      "epoch": 0.010054045562461518,
      "grad_norm": 0.11629543453454971,
      "learning_rate": 0.000749238578680203,
      "loss": 3.116793632507324,
      "step": 124
    },
    {
      "epoch": 0.01021620758766251,
      "grad_norm": 0.10294725745916367,
      "learning_rate": 0.0007614213197969543,
      "loss": 3.0566506385803223,
      "step": 126
    },
    {
      "epoch": 0.010378369612863503,
      "grad_norm": 0.10898813605308533,
      "learning_rate": 0.0007736040609137056,
      "loss": 3.0814855098724365,
      "step": 128
    },
    {
      "epoch": 0.010540531638064494,
      "grad_norm": 0.11384740471839905,
      "learning_rate": 0.0007857868020304567,
      "loss": 3.04201078414917,
      "step": 130
    },
    {
      "epoch": 0.010702693663265487,
      "grad_norm": 0.0997062399983406,
      "learning_rate": 0.000797969543147208,
      "loss": 3.0501389503479004,
      "step": 132
    },
    {
      "epoch": 0.01086485568846648,
      "grad_norm": 0.09038319438695908,
      "learning_rate": 0.0008101522842639593,
      "loss": 3.042616605758667,
      "step": 134
    },
    {
      "epoch": 0.011027017713667472,
      "grad_norm": 0.10113442689180374,
      "learning_rate": 0.0008223350253807105,
      "loss": 3.0287296772003174,
      "step": 136
    },
    {
      "epoch": 0.011189179738868463,
      "grad_norm": 0.08830210566520691,
      "learning_rate": 0.0008345177664974619,
      "loss": 3.1338868141174316,
      "step": 138
    },
    {
      "epoch": 0.011351341764069456,
      "grad_norm": 0.09539781510829926,
      "learning_rate": 0.0008467005076142132,
      "loss": 3.0429229736328125,
      "step": 140
    },
    {
      "epoch": 0.011513503789270449,
      "grad_norm": 0.08422872424125671,
      "learning_rate": 0.0008588832487309644,
      "loss": 3.0130550861358643,
      "step": 142
    },
    {
      "epoch": 0.01167566581447144,
      "grad_norm": 0.08192985504865646,
      "learning_rate": 0.0008710659898477156,
      "loss": 3.0422134399414062,
      "step": 144
    },
    {
      "epoch": 0.011837827839672433,
      "grad_norm": 0.08447181433439255,
      "learning_rate": 0.0008832487309644669,
      "loss": 3.0073728561401367,
      "step": 146
    },
    {
      "epoch": 0.011999989864873425,
      "grad_norm": 0.07773733884096146,
      "learning_rate": 0.0008954314720812182,
      "loss": 3.0086498260498047,
      "step": 148
    },
    {
      "epoch": 0.012162151890074416,
      "grad_norm": 0.0762648731470108,
      "learning_rate": 0.0009076142131979695,
      "loss": 3.0244011878967285,
      "step": 150
    },
    {
      "epoch": 0.01232431391527541,
      "grad_norm": 0.08988801389932632,
      "learning_rate": 0.0009197969543147207,
      "loss": 3.0724058151245117,
      "step": 152
    },
    {
      "epoch": 0.012486475940476402,
      "grad_norm": 0.07769739627838135,
      "learning_rate": 0.000931979695431472,
      "loss": 3.0304055213928223,
      "step": 154
    },
    {
      "epoch": 0.012648637965677395,
      "grad_norm": 0.07071869820356369,
      "learning_rate": 0.0009441624365482233,
      "loss": 3.048305034637451,
      "step": 156
    },
    {
      "epoch": 0.012810799990878386,
      "grad_norm": 0.07106027752161026,
      "learning_rate": 0.0009563451776649745,
      "loss": 3.0263147354125977,
      "step": 158
    },
    {
      "epoch": 0.012972962016079378,
      "grad_norm": 0.07345829159021378,
      "learning_rate": 0.0009685279187817258,
      "loss": 3.0326833724975586,
      "step": 160
    },
    {
      "epoch": 0.013135124041280371,
      "grad_norm": 0.069247767329216,
      "learning_rate": 0.000980710659898477,
      "loss": 3.0317368507385254,
      "step": 162
    },
    {
      "epoch": 0.013297286066481362,
      "grad_norm": 0.07600503414869308,
      "learning_rate": 0.0009928934010152283,
      "loss": 3.053802013397217,
      "step": 164
    },
    {
      "epoch": 0.013459448091682355,
      "grad_norm": 0.06923208385705948,
      "learning_rate": 0.0010050761421319795,
      "loss": 3.017022132873535,
      "step": 166
    },
    {
      "epoch": 0.013621610116883348,
      "grad_norm": 0.06457912176847458,
      "learning_rate": 0.001017258883248731,
      "loss": 3.0685906410217285,
      "step": 168
    },
    {
      "epoch": 0.013783772142084339,
      "grad_norm": 0.07060800492763519,
      "learning_rate": 0.0010294416243654821,
      "loss": 3.0297493934631348,
      "step": 170
    },
    {
      "epoch": 0.013945934167285332,
      "grad_norm": 0.07214295119047165,
      "learning_rate": 0.0010416243654822336,
      "loss": 3.041106939315796,
      "step": 172
    },
    {
      "epoch": 0.014108096192486324,
      "grad_norm": 0.0979209765791893,
      "learning_rate": 0.0010538071065989846,
      "loss": 3.0834975242614746,
      "step": 174
    },
    {
      "epoch": 0.014270258217687315,
      "grad_norm": 0.07776852697134018,
      "learning_rate": 0.001065989847715736,
      "loss": 2.985560655593872,
      "step": 176
    },
    {
      "epoch": 0.014432420242888308,
      "grad_norm": 0.09235746413469315,
      "learning_rate": 0.0010781725888324872,
      "loss": 3.0245742797851562,
      "step": 178
    },
    {
      "epoch": 0.0145945822680893,
      "grad_norm": 0.13174434006214142,
      "learning_rate": 0.0010903553299492384,
      "loss": 2.9688825607299805,
      "step": 180
    },
    {
      "epoch": 0.014756744293290294,
      "grad_norm": 0.06018657982349396,
      "learning_rate": 0.0011025380710659898,
      "loss": 3.0319437980651855,
      "step": 182
    },
    {
      "epoch": 0.014918906318491285,
      "grad_norm": 0.12183063477277756,
      "learning_rate": 0.001114720812182741,
      "loss": 2.996735095977783,
      "step": 184
    },
    {
      "epoch": 0.015081068343692277,
      "grad_norm": 0.061322491616010666,
      "learning_rate": 0.0011269035532994924,
      "loss": 3.0492444038391113,
      "step": 186
    },
    {
      "epoch": 0.01524323036889327,
      "grad_norm": 0.07974984496831894,
      "learning_rate": 0.0011390862944162434,
      "loss": 2.9929585456848145,
      "step": 188
    },
    {
      "epoch": 0.015405392394094261,
      "grad_norm": 0.0526207834482193,
      "learning_rate": 0.0011512690355329949,
      "loss": 3.0313217639923096,
      "step": 190
    },
    {
      "epoch": 0.015567554419295254,
      "grad_norm": 0.0681115984916687,
      "learning_rate": 0.001163451776649746,
      "loss": 3.0273892879486084,
      "step": 192
    },
    {
      "epoch": 0.015729716444496247,
      "grad_norm": 0.057305335998535156,
      "learning_rate": 0.0011756345177664975,
      "loss": 2.958958625793457,
      "step": 194
    },
    {
      "epoch": 0.015891878469697238,
      "grad_norm": 0.09620480239391327,
      "learning_rate": 0.0011878172588832487,
      "loss": 2.9589405059814453,
      "step": 196
    },
    {
      "epoch": 0.016054040494898232,
      "grad_norm": 0.05802655592560768,
      "learning_rate": 0.0012,
      "loss": 2.965613842010498,
      "step": 198
    },
    {
      "epoch": 0.016216202520099223,
      "grad_norm": 0.05243109166622162,
      "learning_rate": 0.0012,
      "loss": 3.0005226135253906,
      "step": 200
    },
    {
      "epoch": 0.016378364545300214,
      "grad_norm": 0.04796193167567253,
      "learning_rate": 0.0012,
      "loss": 2.941563606262207,
      "step": 202
    },
    {
      "epoch": 0.01654052657050121,
      "grad_norm": 0.05740799382328987,
      "learning_rate": 0.0012,
      "loss": 2.9851419925689697,
      "step": 204
    },
    {
      "epoch": 0.0167026885957022,
      "grad_norm": 0.04726474732160568,
      "learning_rate": 0.0012,
      "loss": 2.9557199478149414,
      "step": 206
    },
    {
      "epoch": 0.01686485062090319,
      "grad_norm": 0.05039890110492706,
      "learning_rate": 0.0012,
      "loss": 2.968472480773926,
      "step": 208
    },
    {
      "epoch": 0.017027012646104185,
      "grad_norm": 0.05482833459973335,
      "learning_rate": 0.0012,
      "loss": 2.9228992462158203,
      "step": 210
    },
    {
      "epoch": 0.017189174671305176,
      "grad_norm": 0.043529778718948364,
      "learning_rate": 0.0012,
      "loss": 2.972501754760742,
      "step": 212
    },
    {
      "epoch": 0.017351336696506167,
      "grad_norm": 0.04418805241584778,
      "learning_rate": 0.0012,
      "loss": 3.0476248264312744,
      "step": 214
    },
    {
      "epoch": 0.01751349872170716,
      "grad_norm": 0.045405033975839615,
      "learning_rate": 0.0012,
      "loss": 3.006955146789551,
      "step": 216
    },
    {
      "epoch": 0.017675660746908153,
      "grad_norm": 0.041875723749399185,
      "learning_rate": 0.0012,
      "loss": 2.963393211364746,
      "step": 218
    },
    {
      "epoch": 0.017837822772109144,
      "grad_norm": 0.04650777205824852,
      "learning_rate": 0.0012,
      "loss": 2.9626660346984863,
      "step": 220
    },
    {
      "epoch": 0.017999984797310138,
      "grad_norm": 0.047178201377391815,
      "learning_rate": 0.0012,
      "loss": 3.0212175846099854,
      "step": 222
    },
    {
      "epoch": 0.01816214682251113,
      "grad_norm": 0.06972131878137589,
      "learning_rate": 0.0012,
      "loss": 2.94985294342041,
      "step": 224
    },
    {
      "epoch": 0.01832430884771212,
      "grad_norm": 0.04383312910795212,
      "learning_rate": 0.0012,
      "loss": 2.9764933586120605,
      "step": 226
    },
    {
      "epoch": 0.018486470872913115,
      "grad_norm": 0.05111587420105934,
      "learning_rate": 0.0012,
      "loss": 2.959449529647827,
      "step": 228
    },
    {
      "epoch": 0.018648632898114106,
      "grad_norm": 0.042525358498096466,
      "learning_rate": 0.0012,
      "loss": 2.9851438999176025,
      "step": 230
    },
    {
      "epoch": 0.0188107949233151,
      "grad_norm": 0.047227662056684494,
      "learning_rate": 0.0012,
      "loss": 2.972661018371582,
      "step": 232
    },
    {
      "epoch": 0.01897295694851609,
      "grad_norm": 0.03954648971557617,
      "learning_rate": 0.0012,
      "loss": 2.9962611198425293,
      "step": 234
    },
    {
      "epoch": 0.019135118973717082,
      "grad_norm": 0.04237870126962662,
      "learning_rate": 0.0012,
      "loss": 2.9288787841796875,
      "step": 236
    },
    {
      "epoch": 0.019297280998918077,
      "grad_norm": 0.038656897842884064,
      "learning_rate": 0.0012,
      "loss": 2.9308741092681885,
      "step": 238
    },
    {
      "epoch": 0.019459443024119068,
      "grad_norm": 0.041273996233940125,
      "learning_rate": 0.0012,
      "loss": 2.993577480316162,
      "step": 240
    },
    {
      "epoch": 0.01962160504932006,
      "grad_norm": 0.06833764165639877,
      "learning_rate": 0.0012,
      "loss": 2.8978216648101807,
      "step": 242
    },
    {
      "epoch": 0.019783767074521053,
      "grad_norm": 0.055236730724573135,
      "learning_rate": 0.0012,
      "loss": 3.0273079872131348,
      "step": 244
    },
    {
      "epoch": 0.019945929099722044,
      "grad_norm": 0.0737483948469162,
      "learning_rate": 0.0012,
      "loss": 2.9471025466918945,
      "step": 246
    },
    {
      "epoch": 0.020108091124923035,
      "grad_norm": 0.04837452620267868,
      "learning_rate": 0.0012,
      "loss": 2.9191503524780273,
      "step": 248
    },
    {
      "epoch": 0.02027025315012403,
      "grad_norm": 0.03750497102737427,
      "learning_rate": 0.0012,
      "loss": 3.0062742233276367,
      "step": 250
    },
    {
      "epoch": 0.02043241517532502,
      "grad_norm": 0.10225784033536911,
      "learning_rate": 0.0012,
      "loss": 2.914576292037964,
      "step": 252
    },
    {
      "epoch": 0.020594577200526012,
      "grad_norm": 0.05738373473286629,
      "learning_rate": 0.0012,
      "loss": 2.934493064880371,
      "step": 254
    },
    {
      "epoch": 0.020756739225727006,
      "grad_norm": 0.08224239200353622,
      "learning_rate": 0.0012,
      "loss": 2.9525957107543945,
      "step": 256
    },
    {
      "epoch": 0.020918901250927997,
      "grad_norm": 0.076959989964962,
      "learning_rate": 0.0012,
      "loss": 2.9638874530792236,
      "step": 258
    },
    {
      "epoch": 0.02108106327612899,
      "grad_norm": 0.04292575269937515,
      "learning_rate": 0.0012,
      "loss": 2.9107131958007812,
      "step": 260
    },
    {
      "epoch": 0.021243225301329983,
      "grad_norm": 0.10289739817380905,
      "learning_rate": 0.0012,
      "loss": 2.961691379547119,
      "step": 262
    },
    {
      "epoch": 0.021405387326530974,
      "grad_norm": 0.03500393033027649,
      "learning_rate": 0.0012,
      "loss": 2.91906476020813,
      "step": 264
    },
    {
      "epoch": 0.021567549351731965,
      "grad_norm": 0.11181014031171799,
      "learning_rate": 0.0012,
      "loss": 2.8747916221618652,
      "step": 266
    },
    {
      "epoch": 0.02172971137693296,
      "grad_norm": 0.04349331557750702,
      "learning_rate": 0.0012,
      "loss": 2.9167447090148926,
      "step": 268
    },
    {
      "epoch": 0.02189187340213395,
      "grad_norm": 0.08939475566148758,
      "learning_rate": 0.0012,
      "loss": 2.9420483112335205,
      "step": 270
    },
    {
      "epoch": 0.022054035427334945,
      "grad_norm": 0.06359287351369858,
      "learning_rate": 0.0012,
      "loss": 2.987565517425537,
      "step": 272
    },
    {
      "epoch": 0.022216197452535936,
      "grad_norm": 0.0515572652220726,
      "learning_rate": 0.0012,
      "loss": 2.9143576622009277,
      "step": 274
    },
    {
      "epoch": 0.022378359477736927,
      "grad_norm": 0.06436209380626678,
      "learning_rate": 0.0012,
      "loss": 2.8995940685272217,
      "step": 276
    },
    {
      "epoch": 0.02254052150293792,
      "grad_norm": 0.0340852253139019,
      "learning_rate": 0.0012,
      "loss": 2.9294300079345703,
      "step": 278
    },
    {
      "epoch": 0.022702683528138912,
      "grad_norm": 0.04865012317895889,
      "learning_rate": 0.0012,
      "loss": 2.916050434112549,
      "step": 280
    },
    {
      "epoch": 0.022864845553339903,
      "grad_norm": 0.04238403961062431,
      "learning_rate": 0.0012,
      "loss": 2.935431957244873,
      "step": 282
    },
    {
      "epoch": 0.023027007578540898,
      "grad_norm": 0.044609420001506805,
      "learning_rate": 0.0012,
      "loss": 2.8778584003448486,
      "step": 284
    },
    {
      "epoch": 0.02318916960374189,
      "grad_norm": 0.038976237177848816,
      "learning_rate": 0.0012,
      "loss": 2.8866772651672363,
      "step": 286
    },
    {
      "epoch": 0.02335133162894288,
      "grad_norm": 0.036394260823726654,
      "learning_rate": 0.0012,
      "loss": 2.86003041267395,
      "step": 288
    },
    {
      "epoch": 0.023513493654143874,
      "grad_norm": 0.060845550149679184,
      "learning_rate": 0.0012,
      "loss": 2.8118181228637695,
      "step": 290
    },
    {
      "epoch": 0.023675655679344865,
      "grad_norm": 0.041568875312805176,
      "learning_rate": 0.0012,
      "loss": 2.870222806930542,
      "step": 292
    },
    {
      "epoch": 0.023837817704545856,
      "grad_norm": 0.040509313344955444,
      "learning_rate": 0.0012,
      "loss": 2.9193146228790283,
      "step": 294
    },
    {
      "epoch": 0.02399997972974685,
      "grad_norm": 0.05546777695417404,
      "learning_rate": 0.0012,
      "loss": 2.9777228832244873,
      "step": 296
    },
    {
      "epoch": 0.024162141754947842,
      "grad_norm": 0.03491999953985214,
      "learning_rate": 0.0012,
      "loss": 2.943509817123413,
      "step": 298
    },
    {
      "epoch": 0.024324303780148833,
      "grad_norm": 0.05588215962052345,
      "learning_rate": 0.0012,
      "loss": 2.9129638671875,
      "step": 300
    },
    {
      "epoch": 0.024486465805349827,
      "grad_norm": 0.05662793293595314,
      "learning_rate": 0.0012,
      "loss": 2.9758787155151367,
      "step": 302
    },
    {
      "epoch": 0.02464862783055082,
      "grad_norm": 0.033741045743227005,
      "learning_rate": 0.0012,
      "loss": 2.884683609008789,
      "step": 304
    },
    {
      "epoch": 0.02481078985575181,
      "grad_norm": 0.04745311662554741,
      "learning_rate": 0.0012,
      "loss": 2.914041042327881,
      "step": 306
    },
    {
      "epoch": 0.024972951880952804,
      "grad_norm": 0.0395323820412159,
      "learning_rate": 0.0012,
      "loss": 2.9073486328125,
      "step": 308
    },
    {
      "epoch": 0.025135113906153795,
      "grad_norm": 0.035278309136629105,
      "learning_rate": 0.0012,
      "loss": 2.9476728439331055,
      "step": 310
    },
    {
      "epoch": 0.02529727593135479,
      "grad_norm": 0.05067719146609306,
      "learning_rate": 0.0012,
      "loss": 2.923830509185791,
      "step": 312
    },
    {
      "epoch": 0.02545943795655578,
      "grad_norm": 0.04607498645782471,
      "learning_rate": 0.0012,
      "loss": 2.8913328647613525,
      "step": 314
    },
    {
      "epoch": 0.02562159998175677,
      "grad_norm": 0.03795121982693672,
      "learning_rate": 0.0012,
      "loss": 2.9100489616394043,
      "step": 316
    },
    {
      "epoch": 0.025783762006957766,
      "grad_norm": 0.0442856065928936,
      "learning_rate": 0.0012,
      "loss": 2.9114983081817627,
      "step": 318
    },
    {
      "epoch": 0.025945924032158757,
      "grad_norm": 0.042419251054525375,
      "learning_rate": 0.0012,
      "loss": 2.8812994956970215,
      "step": 320
    },
    {
      "epoch": 0.026108086057359748,
      "grad_norm": 0.03309924900531769,
      "learning_rate": 0.0012,
      "loss": 2.8561344146728516,
      "step": 322
    },
    {
      "epoch": 0.026270248082560742,
      "grad_norm": 0.06521229445934296,
      "learning_rate": 0.0012,
      "loss": 2.902982234954834,
      "step": 324
    },
    {
      "epoch": 0.026432410107761733,
      "grad_norm": 0.04802558198571205,
      "learning_rate": 0.0012,
      "loss": 2.863579273223877,
      "step": 326
    },
    {
      "epoch": 0.026594572132962725,
      "grad_norm": 0.04056398570537567,
      "learning_rate": 0.0012,
      "loss": 2.9300172328948975,
      "step": 328
    },
    {
      "epoch": 0.02675673415816372,
      "grad_norm": 0.03733952343463898,
      "learning_rate": 0.0012,
      "loss": 2.882927417755127,
      "step": 330
    },
    {
      "epoch": 0.02691889618336471,
      "grad_norm": 0.03937390446662903,
      "learning_rate": 0.0012,
      "loss": 2.9494688510894775,
      "step": 332
    },
    {
      "epoch": 0.0270810582085657,
      "grad_norm": 0.03592396154999733,
      "learning_rate": 0.0012,
      "loss": 2.9109911918640137,
      "step": 334
    },
    {
      "epoch": 0.027243220233766696,
      "grad_norm": 0.03635367378592491,
      "learning_rate": 0.0012,
      "loss": 2.9188942909240723,
      "step": 336
    },
    {
      "epoch": 0.027405382258967687,
      "grad_norm": 0.03285514935851097,
      "learning_rate": 0.0012,
      "loss": 2.907867431640625,
      "step": 338
    },
    {
      "epoch": 0.027567544284168678,
      "grad_norm": 0.03414156287908554,
      "learning_rate": 0.0012,
      "loss": 2.897500514984131,
      "step": 340
    },
    {
      "epoch": 0.027729706309369672,
      "grad_norm": 0.03253110125660896,
      "learning_rate": 0.0012,
      "loss": 2.8711211681365967,
      "step": 342
    },
    {
      "epoch": 0.027891868334570663,
      "grad_norm": 0.03527747094631195,
      "learning_rate": 0.0012,
      "loss": 2.907806873321533,
      "step": 344
    },
    {
      "epoch": 0.028054030359771654,
      "grad_norm": 0.04623907059431076,
      "learning_rate": 0.0012,
      "loss": 2.9123048782348633,
      "step": 346
    },
    {
      "epoch": 0.02821619238497265,
      "grad_norm": 0.03669530153274536,
      "learning_rate": 0.0012,
      "loss": 2.875178337097168,
      "step": 348
    },
    {
      "epoch": 0.02837835441017364,
      "grad_norm": 0.03393073007464409,
      "learning_rate": 0.0012,
      "loss": 2.8421082496643066,
      "step": 350
    },
    {
      "epoch": 0.02854051643537463,
      "grad_norm": 0.03223411738872528,
      "learning_rate": 0.0012,
      "loss": 2.868607521057129,
      "step": 352
    },
    {
      "epoch": 0.028702678460575625,
      "grad_norm": 0.032954320311546326,
      "learning_rate": 0.0012,
      "loss": 2.8840231895446777,
      "step": 354
    },
    {
      "epoch": 0.028864840485776616,
      "grad_norm": 0.03337768092751503,
      "learning_rate": 0.0012,
      "loss": 2.880872964859009,
      "step": 356
    },
    {
      "epoch": 0.02902700251097761,
      "grad_norm": 0.032412368804216385,
      "learning_rate": 0.0012,
      "loss": 2.8338088989257812,
      "step": 358
    },
    {
      "epoch": 0.0291891645361786,
      "grad_norm": 0.033116020262241364,
      "learning_rate": 0.0012,
      "loss": 2.8295068740844727,
      "step": 360
    },
    {
      "epoch": 0.029351326561379593,
      "grad_norm": 0.03446929529309273,
      "learning_rate": 0.0012,
      "loss": 2.9334521293640137,
      "step": 362
    },
    {
      "epoch": 0.029513488586580587,
      "grad_norm": 0.037121064960956573,
      "learning_rate": 0.0012,
      "loss": 2.8303370475769043,
      "step": 364
    },
    {
      "epoch": 0.029675650611781578,
      "grad_norm": 0.03445614501833916,
      "learning_rate": 0.0012,
      "loss": 2.8712241649627686,
      "step": 366
    },
    {
      "epoch": 0.02983781263698257,
      "grad_norm": 0.03678618744015694,
      "learning_rate": 0.0012,
      "loss": 2.8344922065734863,
      "step": 368
    },
    {
      "epoch": 0.029999974662183564,
      "grad_norm": 0.04280025511980057,
      "learning_rate": 0.0012,
      "loss": 2.8747785091400146,
      "step": 370
    },
    {
      "epoch": 0.030162136687384555,
      "grad_norm": 0.03331344574689865,
      "learning_rate": 0.0012,
      "loss": 2.863950252532959,
      "step": 372
    },
    {
      "epoch": 0.030324298712585546,
      "grad_norm": 0.03843531385064125,
      "learning_rate": 0.0012,
      "loss": 2.871696949005127,
      "step": 374
    },
    {
      "epoch": 0.03048646073778654,
      "grad_norm": 0.038714345544576645,
      "learning_rate": 0.0012,
      "loss": 2.8897619247436523,
      "step": 376
    },
    {
      "epoch": 0.03064862276298753,
      "grad_norm": 0.034631162881851196,
      "learning_rate": 0.0012,
      "loss": 2.83707332611084,
      "step": 378
    },
    {
      "epoch": 0.030810784788188522,
      "grad_norm": 0.032213106751441956,
      "learning_rate": 0.0012,
      "loss": 2.8370254039764404,
      "step": 380
    },
    {
      "epoch": 0.030972946813389517,
      "grad_norm": 0.03538420796394348,
      "learning_rate": 0.0012,
      "loss": 2.92585825920105,
      "step": 382
    },
    {
      "epoch": 0.031135108838590508,
      "grad_norm": 0.033038973808288574,
      "learning_rate": 0.0012,
      "loss": 2.9070730209350586,
      "step": 384
    },
    {
      "epoch": 0.0312972708637915,
      "grad_norm": 0.04414618760347366,
      "learning_rate": 0.0012,
      "loss": 2.8657431602478027,
      "step": 386
    },
    {
      "epoch": 0.03145943288899249,
      "grad_norm": 0.0418984480202198,
      "learning_rate": 0.0012,
      "loss": 2.880760669708252,
      "step": 388
    },
    {
      "epoch": 0.03162159491419349,
      "grad_norm": 0.030810100957751274,
      "learning_rate": 0.0012,
      "loss": 2.90207839012146,
      "step": 390
    },
    {
      "epoch": 0.031783756939394475,
      "grad_norm": 0.03251317888498306,
      "learning_rate": 0.0012,
      "loss": 2.8063254356384277,
      "step": 392
    },
    {
      "epoch": 0.03194591896459547,
      "grad_norm": 0.041920531541109085,
      "learning_rate": 0.0012,
      "loss": 2.895156145095825,
      "step": 394
    },
    {
      "epoch": 0.032108080989796464,
      "grad_norm": 0.04542321711778641,
      "learning_rate": 0.0012,
      "loss": 2.846189498901367,
      "step": 396
    },
    {
      "epoch": 0.03227024301499745,
      "grad_norm": 0.028101852163672447,
      "learning_rate": 0.0012,
      "loss": 2.843764305114746,
      "step": 398
    },
    {
      "epoch": 0.032432405040198446,
      "grad_norm": 0.06268654763698578,
      "learning_rate": 0.0012,
      "loss": 2.8165717124938965,
      "step": 400
    },
    {
      "epoch": 0.03259456706539944,
      "grad_norm": 0.03920038789510727,
      "learning_rate": 0.0012,
      "loss": 2.889784097671509,
      "step": 402
    },
    {
      "epoch": 0.03275672909060043,
      "grad_norm": 0.029086723923683167,
      "learning_rate": 0.0012,
      "loss": 2.8110880851745605,
      "step": 404
    },
    {
      "epoch": 0.03291889111580142,
      "grad_norm": 0.032011114060878754,
      "learning_rate": 0.0012,
      "loss": 2.888796329498291,
      "step": 406
    },
    {
      "epoch": 0.03308105314100242,
      "grad_norm": 0.039376482367515564,
      "learning_rate": 0.0012,
      "loss": 2.8050618171691895,
      "step": 408
    },
    {
      "epoch": 0.033243215166203405,
      "grad_norm": 0.068088598549366,
      "learning_rate": 0.0012,
      "loss": 2.90848970413208,
      "step": 410
    },
    {
      "epoch": 0.0334053771914044,
      "grad_norm": 0.04729756340384483,
      "learning_rate": 0.0012,
      "loss": 2.8389358520507812,
      "step": 412
    },
    {
      "epoch": 0.033567539216605394,
      "grad_norm": 0.027843791991472244,
      "learning_rate": 0.0012,
      "loss": 2.8776798248291016,
      "step": 414
    },
    {
      "epoch": 0.03372970124180638,
      "grad_norm": 0.03880459815263748,
      "learning_rate": 0.0012,
      "loss": 2.8874850273132324,
      "step": 416
    },
    {
      "epoch": 0.033891863267007376,
      "grad_norm": 0.03327984735369682,
      "learning_rate": 0.0012,
      "loss": 2.8472132682800293,
      "step": 418
    },
    {
      "epoch": 0.03405402529220837,
      "grad_norm": 0.028868863359093666,
      "learning_rate": 0.0012,
      "loss": 2.847853899002075,
      "step": 420
    },
    {
      "epoch": 0.03421618731740936,
      "grad_norm": 0.025681525468826294,
      "learning_rate": 0.0012,
      "loss": 2.8830881118774414,
      "step": 422
    },
    {
      "epoch": 0.03437834934261035,
      "grad_norm": 0.05296497419476509,
      "learning_rate": 0.0012,
      "loss": 2.94162917137146,
      "step": 424
    },
    {
      "epoch": 0.03454051136781135,
      "grad_norm": 0.03395327180624008,
      "learning_rate": 0.0012,
      "loss": 2.8577234745025635,
      "step": 426
    },
    {
      "epoch": 0.034702673393012334,
      "grad_norm": 0.0339728482067585,
      "learning_rate": 0.0012,
      "loss": 2.9462122917175293,
      "step": 428
    },
    {
      "epoch": 0.03486483541821333,
      "grad_norm": 0.07809537649154663,
      "learning_rate": 0.0012,
      "loss": 2.853940963745117,
      "step": 430
    },
    {
      "epoch": 0.03502699744341432,
      "grad_norm": 0.0599173866212368,
      "learning_rate": 0.0012,
      "loss": 2.7668843269348145,
      "step": 432
    },
    {
      "epoch": 0.03518915946861531,
      "grad_norm": 0.031115280464291573,
      "learning_rate": 0.0012,
      "loss": 2.8470771312713623,
      "step": 434
    },
    {
      "epoch": 0.035351321493816305,
      "grad_norm": 0.033972762525081635,
      "learning_rate": 0.0012,
      "loss": 2.8085694313049316,
      "step": 436
    },
    {
      "epoch": 0.0355134835190173,
      "grad_norm": 0.051337484270334244,
      "learning_rate": 0.0012,
      "loss": 2.8032588958740234,
      "step": 438
    },
    {
      "epoch": 0.03567564554421829,
      "grad_norm": 0.02822447009384632,
      "learning_rate": 0.0012,
      "loss": 2.83536958694458,
      "step": 440
    },
    {
      "epoch": 0.03583780756941928,
      "grad_norm": 0.0636354312300682,
      "learning_rate": 0.0012,
      "loss": 2.807875633239746,
      "step": 442
    },
    {
      "epoch": 0.035999969594620276,
      "grad_norm": 0.04920722544193268,
      "learning_rate": 0.0012,
      "loss": 2.82302188873291,
      "step": 444
    },
    {
      "epoch": 0.036162131619821264,
      "grad_norm": 0.040734563022851944,
      "learning_rate": 0.0012,
      "loss": 2.8399364948272705,
      "step": 446
    },
    {
      "epoch": 0.03632429364502226,
      "grad_norm": 0.057042092084884644,
      "learning_rate": 0.0012,
      "loss": 2.83658766746521,
      "step": 448
    },
    {
      "epoch": 0.03648645567022325,
      "grad_norm": 0.03276890888810158,
      "learning_rate": 0.0012,
      "loss": 2.8834710121154785,
      "step": 450
    },
    {
      "epoch": 0.03664861769542424,
      "grad_norm": 0.05333075672388077,
      "learning_rate": 0.0012,
      "loss": 2.8143153190612793,
      "step": 452
    },
    {
      "epoch": 0.036810779720625235,
      "grad_norm": 0.039645060896873474,
      "learning_rate": 0.0012,
      "loss": 2.8261075019836426,
      "step": 454
    },
    {
      "epoch": 0.03697294174582623,
      "grad_norm": 0.04189448431134224,
      "learning_rate": 0.0012,
      "loss": 2.861607551574707,
      "step": 456
    },
    {
      "epoch": 0.037135103771027224,
      "grad_norm": 0.05914913862943649,
      "learning_rate": 0.0012,
      "loss": 2.8012685775756836,
      "step": 458
    },
    {
      "epoch": 0.03729726579622821,
      "grad_norm": 0.04268115013837814,
      "learning_rate": 0.0012,
      "loss": 2.8496251106262207,
      "step": 460
    },
    {
      "epoch": 0.037459427821429206,
      "grad_norm": 0.05694008991122246,
      "learning_rate": 0.0012,
      "loss": 2.8754420280456543,
      "step": 462
    },
    {
      "epoch": 0.0376215898466302,
      "grad_norm": 0.054185617715120316,
      "learning_rate": 0.0012,
      "loss": 2.8582987785339355,
      "step": 464
    },
    {
      "epoch": 0.03778375187183119,
      "grad_norm": 0.028196515515446663,
      "learning_rate": 0.0012,
      "loss": 2.8305468559265137,
      "step": 466
    },
    {
      "epoch": 0.03794591389703218,
      "grad_norm": 0.024743732064962387,
      "learning_rate": 0.0012,
      "loss": 2.8433823585510254,
      "step": 468
    },
    {
      "epoch": 0.03810807592223318,
      "grad_norm": 0.03771902620792389,
      "learning_rate": 0.0012,
      "loss": 2.8622894287109375,
      "step": 470
    },
    {
      "epoch": 0.038270237947434164,
      "grad_norm": 0.048127640038728714,
      "learning_rate": 0.0012,
      "loss": 2.844299793243408,
      "step": 472
    },
    {
      "epoch": 0.03843239997263516,
      "grad_norm": 0.025167779996991158,
      "learning_rate": 0.0012,
      "loss": 2.8418996334075928,
      "step": 474
    },
    {
      "epoch": 0.03859456199783615,
      "grad_norm": 0.057061817497015,
      "learning_rate": 0.0012,
      "loss": 2.85254168510437,
      "step": 476
    },
    {
      "epoch": 0.03875672402303714,
      "grad_norm": 0.043872248381376266,
      "learning_rate": 0.0012,
      "loss": 2.8328447341918945,
      "step": 478
    },
    {
      "epoch": 0.038918886048238135,
      "grad_norm": 0.026863591745495796,
      "learning_rate": 0.0012,
      "loss": 2.8883285522460938,
      "step": 480
    },
    {
      "epoch": 0.03908104807343913,
      "grad_norm": 0.05471372231841087,
      "learning_rate": 0.0012,
      "loss": 2.891606330871582,
      "step": 482
    },
    {
      "epoch": 0.03924321009864012,
      "grad_norm": 0.03392742574214935,
      "learning_rate": 0.0012,
      "loss": 2.8919410705566406,
      "step": 484
    },
    {
      "epoch": 0.03940537212384111,
      "grad_norm": 0.0282833781093359,
      "learning_rate": 0.0012,
      "loss": 2.8019509315490723,
      "step": 486
    },
    {
      "epoch": 0.039567534149042106,
      "grad_norm": 0.05026375874876976,
      "learning_rate": 0.0012,
      "loss": 2.8088438510894775,
      "step": 488
    },
    {
      "epoch": 0.039729696174243094,
      "grad_norm": 0.03285565599799156,
      "learning_rate": 0.0012,
      "loss": 2.858989715576172,
      "step": 490
    },
    {
      "epoch": 0.03989185819944409,
      "grad_norm": 0.027263140305876732,
      "learning_rate": 0.0012,
      "loss": 2.828458309173584,
      "step": 492
    },
    {
      "epoch": 0.04005402022464508,
      "grad_norm": 0.04072710871696472,
      "learning_rate": 0.0012,
      "loss": 2.798424243927002,
      "step": 494
    },
    {
      "epoch": 0.04021618224984607,
      "grad_norm": 0.026240479201078415,
      "learning_rate": 0.0012,
      "loss": 2.7871651649475098,
      "step": 496
    },
    {
      "epoch": 0.040378344275047065,
      "grad_norm": 0.033734604716300964,
      "learning_rate": 0.0012,
      "loss": 2.9278483390808105,
      "step": 498
    },
    {
      "epoch": 0.04054050630024806,
      "grad_norm": 0.03572112321853638,
      "learning_rate": 0.0012,
      "loss": 2.8660635948181152,
      "step": 500
    },
    {
      "epoch": 0.04070266832544905,
      "grad_norm": 0.02979310043156147,
      "learning_rate": 0.0012,
      "loss": 2.829638957977295,
      "step": 502
    },
    {
      "epoch": 0.04086483035065004,
      "grad_norm": 0.051394663751125336,
      "learning_rate": 0.0012,
      "loss": 2.8415937423706055,
      "step": 504
    },
    {
      "epoch": 0.041026992375851036,
      "grad_norm": 0.04895510897040367,
      "learning_rate": 0.0012,
      "loss": 2.829357862472534,
      "step": 506
    },
    {
      "epoch": 0.041189154401052024,
      "grad_norm": 0.030360236763954163,
      "learning_rate": 0.0012,
      "loss": 2.8804614543914795,
      "step": 508
    },
    {
      "epoch": 0.04135131642625302,
      "grad_norm": 0.03648420423269272,
      "learning_rate": 0.0012,
      "loss": 2.8491716384887695,
      "step": 510
    },
    {
      "epoch": 0.04151347845145401,
      "grad_norm": 0.06702860444784164,
      "learning_rate": 0.0012,
      "loss": 2.7563395500183105,
      "step": 512
    },
    {
      "epoch": 0.041675640476655,
      "grad_norm": 0.023179104551672935,
      "learning_rate": 0.0012,
      "loss": 2.81121563911438,
      "step": 514
    },
    {
      "epoch": 0.041837802501855995,
      "grad_norm": 0.056094612926244736,
      "learning_rate": 0.0012,
      "loss": 2.8118040561676025,
      "step": 516
    },
    {
      "epoch": 0.04199996452705699,
      "grad_norm": 0.04948415979743004,
      "learning_rate": 0.0012,
      "loss": 2.8427138328552246,
      "step": 518
    },
    {
      "epoch": 0.04216212655225798,
      "grad_norm": 0.022890165448188782,
      "learning_rate": 0.0012,
      "loss": 2.839265823364258,
      "step": 520
    },
    {
      "epoch": 0.04232428857745897,
      "grad_norm": 0.06509353220462799,
      "learning_rate": 0.0012,
      "loss": 2.871161937713623,
      "step": 522
    },
    {
      "epoch": 0.042486450602659966,
      "grad_norm": 0.06935494393110275,
      "learning_rate": 0.0012,
      "loss": 2.8360791206359863,
      "step": 524
    },
    {
      "epoch": 0.04264861262786095,
      "grad_norm": 0.022866254672408104,
      "learning_rate": 0.0012,
      "loss": 2.7958312034606934,
      "step": 526
    },
    {
      "epoch": 0.04281077465306195,
      "grad_norm": 0.08423201739788055,
      "learning_rate": 0.0012,
      "loss": 2.826429843902588,
      "step": 528
    },
    {
      "epoch": 0.04297293667826294,
      "grad_norm": 0.04014052823185921,
      "learning_rate": 0.0012,
      "loss": 2.8846688270568848,
      "step": 530
    },
    {
      "epoch": 0.04313509870346393,
      "grad_norm": 0.02544465847313404,
      "learning_rate": 0.0012,
      "loss": 2.786289691925049,
      "step": 532
    },
    {
      "epoch": 0.043297260728664924,
      "grad_norm": 0.02443767711520195,
      "learning_rate": 0.0012,
      "loss": 2.7855491638183594,
      "step": 534
    },
    {
      "epoch": 0.04345942275386592,
      "grad_norm": 0.02681577205657959,
      "learning_rate": 0.0012,
      "loss": 2.7973501682281494,
      "step": 536
    },
    {
      "epoch": 0.043621584779066906,
      "grad_norm": 0.041308000683784485,
      "learning_rate": 0.0012,
      "loss": 2.843874931335449,
      "step": 538
    },
    {
      "epoch": 0.0437837468042679,
      "grad_norm": 0.06161566823720932,
      "learning_rate": 0.0012,
      "loss": 2.831456184387207,
      "step": 540
    },
    {
      "epoch": 0.043945908829468895,
      "grad_norm": 0.023288529366254807,
      "learning_rate": 0.0012,
      "loss": 2.8346807956695557,
      "step": 542
    },
    {
      "epoch": 0.04410807085466989,
      "grad_norm": 0.045438386499881744,
      "learning_rate": 0.0012,
      "loss": 2.817714214324951,
      "step": 544
    },
    {
      "epoch": 0.04427023287987088,
      "grad_norm": 0.04769458994269371,
      "learning_rate": 0.0012,
      "loss": 2.8768701553344727,
      "step": 546
    },
    {
      "epoch": 0.04443239490507187,
      "grad_norm": 0.04741879180073738,
      "learning_rate": 0.0012,
      "loss": 2.842287540435791,
      "step": 548
    },
    {
      "epoch": 0.044594556930272866,
      "grad_norm": 0.027781469747424126,
      "learning_rate": 0.0012,
      "loss": 2.8438291549682617,
      "step": 550
    },
    {
      "epoch": 0.044756718955473854,
      "grad_norm": 0.06673149764537811,
      "learning_rate": 0.0012,
      "loss": 2.8073034286499023,
      "step": 552
    },
    {
      "epoch": 0.04491888098067485,
      "grad_norm": 0.036107905209064484,
      "learning_rate": 0.0012,
      "loss": 2.8380887508392334,
      "step": 554
    },
    {
      "epoch": 0.04508104300587584,
      "grad_norm": 0.04221151024103165,
      "learning_rate": 0.0012,
      "loss": 2.866011142730713,
      "step": 556
    },
    {
      "epoch": 0.04524320503107683,
      "grad_norm": 0.08147833496332169,
      "learning_rate": 0.0012,
      "loss": 2.7935068607330322,
      "step": 558
    },
    {
      "epoch": 0.045405367056277825,
      "grad_norm": 0.036353740841150284,
      "learning_rate": 0.0012,
      "loss": 2.810122489929199,
      "step": 560
    },
    {
      "epoch": 0.04556752908147882,
      "grad_norm": 0.03876449912786484,
      "learning_rate": 0.0012,
      "loss": 2.7434215545654297,
      "step": 562
    },
    {
      "epoch": 0.04572969110667981,
      "grad_norm": 0.0318748876452446,
      "learning_rate": 0.0012,
      "loss": 2.7557473182678223,
      "step": 564
    },
    {
      "epoch": 0.0458918531318808,
      "grad_norm": 0.046737801283597946,
      "learning_rate": 0.0012,
      "loss": 2.8102121353149414,
      "step": 566
    },
    {
      "epoch": 0.046054015157081796,
      "grad_norm": 0.038842953741550446,
      "learning_rate": 0.0012,
      "loss": 2.8063628673553467,
      "step": 568
    },
    {
      "epoch": 0.04621617718228278,
      "grad_norm": 0.03262815251946449,
      "learning_rate": 0.0012,
      "loss": 2.904482126235962,
      "step": 570
    },
    {
      "epoch": 0.04637833920748378,
      "grad_norm": 0.02620743401348591,
      "learning_rate": 0.0012,
      "loss": 2.795499801635742,
      "step": 572
    },
    {
      "epoch": 0.04654050123268477,
      "grad_norm": 0.027253350242972374,
      "learning_rate": 0.0012,
      "loss": 2.792022228240967,
      "step": 574
    },
    {
      "epoch": 0.04670266325788576,
      "grad_norm": 0.03230276331305504,
      "learning_rate": 0.0012,
      "loss": 2.815505027770996,
      "step": 576
    },
    {
      "epoch": 0.046864825283086754,
      "grad_norm": 0.026045402511954308,
      "learning_rate": 0.0012,
      "loss": 2.7369742393493652,
      "step": 578
    },
    {
      "epoch": 0.04702698730828775,
      "grad_norm": 0.03883018344640732,
      "learning_rate": 0.0012,
      "loss": 2.845569133758545,
      "step": 580
    },
    {
      "epoch": 0.047189149333488736,
      "grad_norm": 0.025563260540366173,
      "learning_rate": 0.0012,
      "loss": 2.786860942840576,
      "step": 582
    },
    {
      "epoch": 0.04735131135868973,
      "grad_norm": 0.03573295846581459,
      "learning_rate": 0.0012,
      "loss": 2.8056833744049072,
      "step": 584
    },
    {
      "epoch": 0.047513473383890725,
      "grad_norm": 0.05407102033495903,
      "learning_rate": 0.0012,
      "loss": 2.7710492610931396,
      "step": 586
    },
    {
      "epoch": 0.04767563540909171,
      "grad_norm": 0.06135999783873558,
      "learning_rate": 0.0012,
      "loss": 2.7939109802246094,
      "step": 588
    },
    {
      "epoch": 0.04783779743429271,
      "grad_norm": 0.04313323274254799,
      "learning_rate": 0.0012,
      "loss": 2.7943201065063477,
      "step": 590
    },
    {
      "epoch": 0.0479999594594937,
      "grad_norm": 0.02449094131588936,
      "learning_rate": 0.0012,
      "loss": 2.803926944732666,
      "step": 592
    },
    {
      "epoch": 0.04816212148469469,
      "grad_norm": 0.04876024276018143,
      "learning_rate": 0.0012,
      "loss": 2.835420608520508,
      "step": 594
    },
    {
      "epoch": 0.048324283509895684,
      "grad_norm": 0.031132498756051064,
      "learning_rate": 0.0012,
      "loss": 2.834179401397705,
      "step": 596
    },
    {
      "epoch": 0.04848644553509668,
      "grad_norm": 0.028030376881361008,
      "learning_rate": 0.0012,
      "loss": 2.7764053344726562,
      "step": 598
    },
    {
      "epoch": 0.048648607560297666,
      "grad_norm": 0.05269145965576172,
      "learning_rate": 0.0012,
      "loss": 2.797335147857666,
      "step": 600
    },
    {
      "epoch": 0.04881076958549866,
      "grad_norm": 0.0514092892408371,
      "learning_rate": 0.0012,
      "loss": 2.850727081298828,
      "step": 602
    },
    {
      "epoch": 0.048972931610699655,
      "grad_norm": 0.025368347764015198,
      "learning_rate": 0.0012,
      "loss": 2.8369102478027344,
      "step": 604
    },
    {
      "epoch": 0.04913509363590064,
      "grad_norm": 0.039563871920108795,
      "learning_rate": 0.0012,
      "loss": 2.8211445808410645,
      "step": 606
    },
    {
      "epoch": 0.04929725566110164,
      "grad_norm": 0.033700402826070786,
      "learning_rate": 0.0012,
      "loss": 2.828765869140625,
      "step": 608
    },
    {
      "epoch": 0.04945941768630263,
      "grad_norm": 0.028599323704838753,
      "learning_rate": 0.0012,
      "loss": 2.820280075073242,
      "step": 610
    },
    {
      "epoch": 0.04962157971150362,
      "grad_norm": 0.036730241030454636,
      "learning_rate": 0.0012,
      "loss": 2.842223644256592,
      "step": 612
    },
    {
      "epoch": 0.04978374173670461,
      "grad_norm": 0.03669724985957146,
      "learning_rate": 0.0012,
      "loss": 2.805368423461914,
      "step": 614
    },
    {
      "epoch": 0.04994590376190561,
      "grad_norm": 0.031159549951553345,
      "learning_rate": 0.0012,
      "loss": 2.8155274391174316,
      "step": 616
    },
    {
      "epoch": 0.050108065787106595,
      "grad_norm": 0.029385393485426903,
      "learning_rate": 0.0012,
      "loss": 2.8164424896240234,
      "step": 618
    },
    {
      "epoch": 0.05027022781230759,
      "grad_norm": 0.027909869328141212,
      "learning_rate": 0.0012,
      "loss": 2.8224427700042725,
      "step": 620
    },
    {
      "epoch": 0.050432389837508584,
      "grad_norm": 0.02664153464138508,
      "learning_rate": 0.0012,
      "loss": 2.8481903076171875,
      "step": 622
    },
    {
      "epoch": 0.05059455186270958,
      "grad_norm": 0.026661543175578117,
      "learning_rate": 0.0012,
      "loss": 2.8390402793884277,
      "step": 624
    },
    {
      "epoch": 0.050756713887910566,
      "grad_norm": 0.04138673469424248,
      "learning_rate": 0.0012,
      "loss": 2.8232765197753906,
      "step": 626
    },
    {
      "epoch": 0.05091887591311156,
      "grad_norm": 0.0460287407040596,
      "learning_rate": 0.0012,
      "loss": 2.811755418777466,
      "step": 628
    },
    {
      "epoch": 0.051081037938312555,
      "grad_norm": 0.02654406800866127,
      "learning_rate": 0.0012,
      "loss": 2.845198154449463,
      "step": 630
    },
    {
      "epoch": 0.05124319996351354,
      "grad_norm": 0.025368520990014076,
      "learning_rate": 0.0012,
      "loss": 2.824252128601074,
      "step": 632
    },
    {
      "epoch": 0.05140536198871454,
      "grad_norm": 0.03204388916492462,
      "learning_rate": 0.0012,
      "loss": 2.7024435997009277,
      "step": 634
    },
    {
      "epoch": 0.05156752401391553,
      "grad_norm": 0.02571740187704563,
      "learning_rate": 0.0012,
      "loss": 2.8911068439483643,
      "step": 636
    },
    {
      "epoch": 0.05172968603911652,
      "grad_norm": 0.02564145438373089,
      "learning_rate": 0.0012,
      "loss": 2.7574634552001953,
      "step": 638
    },
    {
      "epoch": 0.051891848064317514,
      "grad_norm": 0.027438810095191002,
      "learning_rate": 0.0012,
      "loss": 2.805967092514038,
      "step": 640
    },
    {
      "epoch": 0.05205401008951851,
      "grad_norm": 0.024335384368896484,
      "learning_rate": 0.0012,
      "loss": 2.7979812622070312,
      "step": 642
    },
    {
      "epoch": 0.052216172114719496,
      "grad_norm": 0.06589711457490921,
      "learning_rate": 0.0012,
      "loss": 2.8238401412963867,
      "step": 644
    },
    {
      "epoch": 0.05237833413992049,
      "grad_norm": 0.06956661492586136,
      "learning_rate": 0.0012,
      "loss": 2.8247427940368652,
      "step": 646
    },
    {
      "epoch": 0.052540496165121485,
      "grad_norm": 0.04586252197623253,
      "learning_rate": 0.0012,
      "loss": 2.788926601409912,
      "step": 648
    },
    {
      "epoch": 0.05270265819032247,
      "grad_norm": 0.026384446769952774,
      "learning_rate": 0.0012,
      "loss": 2.798828601837158,
      "step": 650
    },
    {
      "epoch": 0.05286482021552347,
      "grad_norm": 0.037902530282735825,
      "learning_rate": 0.0012,
      "loss": 2.764737606048584,
      "step": 652
    },
    {
      "epoch": 0.05302698224072446,
      "grad_norm": 0.0633036196231842,
      "learning_rate": 0.0012,
      "loss": 2.8301806449890137,
      "step": 654
    },
    {
      "epoch": 0.05318914426592545,
      "grad_norm": 0.024945799261331558,
      "learning_rate": 0.0012,
      "loss": 2.7808139324188232,
      "step": 656
    },
    {
      "epoch": 0.053351306291126444,
      "grad_norm": 0.03666963055729866,
      "learning_rate": 0.0012,
      "loss": 2.807668685913086,
      "step": 658
    },
    {
      "epoch": 0.05351346831632744,
      "grad_norm": 0.053004130721092224,
      "learning_rate": 0.0012,
      "loss": 2.8035409450531006,
      "step": 660
    },
    {
      "epoch": 0.053675630341528426,
      "grad_norm": 0.024705028161406517,
      "learning_rate": 0.0012,
      "loss": 2.8032069206237793,
      "step": 662
    },
    {
      "epoch": 0.05383779236672942,
      "grad_norm": 0.025177186354994774,
      "learning_rate": 0.0012,
      "loss": 2.8052453994750977,
      "step": 664
    },
    {
      "epoch": 0.053999954391930415,
      "grad_norm": 0.034903284162282944,
      "learning_rate": 0.0012,
      "loss": 2.7878708839416504,
      "step": 666
    },
    {
      "epoch": 0.0541621164171314,
      "grad_norm": 0.056320901960134506,
      "learning_rate": 0.0012,
      "loss": 2.776088237762451,
      "step": 668
    },
    {
      "epoch": 0.0543242784423324,
      "grad_norm": 0.043577469885349274,
      "learning_rate": 0.0012,
      "loss": 2.776820182800293,
      "step": 670
    },
    {
      "epoch": 0.05448644046753339,
      "grad_norm": 0.025632433593273163,
      "learning_rate": 0.0012,
      "loss": 2.830261707305908,
      "step": 672
    },
    {
      "epoch": 0.05464860249273438,
      "grad_norm": 0.03919898346066475,
      "learning_rate": 0.0012,
      "loss": 2.8092567920684814,
      "step": 674
    },
    {
      "epoch": 0.05481076451793537,
      "grad_norm": 0.04060964286327362,
      "learning_rate": 0.0012,
      "loss": 2.835178852081299,
      "step": 676
    },
    {
      "epoch": 0.05497292654313637,
      "grad_norm": 0.027036098763346672,
      "learning_rate": 0.0012,
      "loss": 2.8325159549713135,
      "step": 678
    },
    {
      "epoch": 0.055135088568337355,
      "grad_norm": 0.023708799853920937,
      "learning_rate": 0.0012,
      "loss": 2.8138012886047363,
      "step": 680
    },
    {
      "epoch": 0.05529725059353835,
      "grad_norm": 0.02368602715432644,
      "learning_rate": 0.0012,
      "loss": 2.767200469970703,
      "step": 682
    },
    {
      "epoch": 0.055459412618739344,
      "grad_norm": 0.031109118834137917,
      "learning_rate": 0.0012,
      "loss": 2.7789130210876465,
      "step": 684
    },
    {
      "epoch": 0.05562157464394033,
      "grad_norm": 0.038877882063388824,
      "learning_rate": 0.0012,
      "loss": 2.806872844696045,
      "step": 686
    },
    {
      "epoch": 0.055783736669141326,
      "grad_norm": 0.038816746324300766,
      "learning_rate": 0.0012,
      "loss": 2.8016700744628906,
      "step": 688
    },
    {
      "epoch": 0.05594589869434232,
      "grad_norm": 0.024513792246580124,
      "learning_rate": 0.0012,
      "loss": 2.836843490600586,
      "step": 690
    },
    {
      "epoch": 0.05610806071954331,
      "grad_norm": 0.0438697375357151,
      "learning_rate": 0.0012,
      "loss": 2.792125701904297,
      "step": 692
    },
    {
      "epoch": 0.0562702227447443,
      "grad_norm": 0.0349336676299572,
      "learning_rate": 0.0012,
      "loss": 2.734417676925659,
      "step": 694
    },
    {
      "epoch": 0.0564323847699453,
      "grad_norm": 0.026638654991984367,
      "learning_rate": 0.0012,
      "loss": 2.8219213485717773,
      "step": 696
    },
    {
      "epoch": 0.056594546795146285,
      "grad_norm": 0.027787193655967712,
      "learning_rate": 0.0012,
      "loss": 2.7754476070404053,
      "step": 698
    },
    {
      "epoch": 0.05675670882034728,
      "grad_norm": 0.028173036873340607,
      "learning_rate": 0.0012,
      "loss": 2.7949600219726562,
      "step": 700
    },
    {
      "epoch": 0.056918870845548274,
      "grad_norm": 0.029065780341625214,
      "learning_rate": 0.0012,
      "loss": 2.7767863273620605,
      "step": 702
    },
    {
      "epoch": 0.05708103287074926,
      "grad_norm": 0.02568637952208519,
      "learning_rate": 0.0012,
      "loss": 2.7806527614593506,
      "step": 704
    },
    {
      "epoch": 0.057243194895950256,
      "grad_norm": 0.03151519596576691,
      "learning_rate": 0.0012,
      "loss": 2.826061964035034,
      "step": 706
    },
    {
      "epoch": 0.05740535692115125,
      "grad_norm": 0.03912341222167015,
      "learning_rate": 0.0012,
      "loss": 2.7657876014709473,
      "step": 708
    },
    {
      "epoch": 0.057567518946352245,
      "grad_norm": 0.07272133976221085,
      "learning_rate": 0.0012,
      "loss": 2.8463995456695557,
      "step": 710
    },
    {
      "epoch": 0.05772968097155323,
      "grad_norm": 0.06251130998134613,
      "learning_rate": 0.0012,
      "loss": 2.7903594970703125,
      "step": 712
    },
    {
      "epoch": 0.05789184299675423,
      "grad_norm": 0.029281314462423325,
      "learning_rate": 0.0012,
      "loss": 2.8069164752960205,
      "step": 714
    },
    {
      "epoch": 0.05805400502195522,
      "grad_norm": 0.03900211676955223,
      "learning_rate": 0.0012,
      "loss": 2.786116600036621,
      "step": 716
    },
    {
      "epoch": 0.05821616704715621,
      "grad_norm": 0.06058591231703758,
      "learning_rate": 0.0012,
      "loss": 2.868616819381714,
      "step": 718
    },
    {
      "epoch": 0.0583783290723572,
      "grad_norm": 0.025215644389390945,
      "learning_rate": 0.0012,
      "loss": 2.802328109741211,
      "step": 720
    },
    {
      "epoch": 0.0585404910975582,
      "grad_norm": 0.0588529109954834,
      "learning_rate": 0.0012,
      "loss": 2.801304817199707,
      "step": 722
    },
    {
      "epoch": 0.058702653122759185,
      "grad_norm": 0.06197066605091095,
      "learning_rate": 0.0012,
      "loss": 2.8366098403930664,
      "step": 724
    },
    {
      "epoch": 0.05886481514796018,
      "grad_norm": 0.04359520971775055,
      "learning_rate": 0.0012,
      "loss": 2.7818775177001953,
      "step": 726
    },
    {
      "epoch": 0.059026977173161174,
      "grad_norm": 0.023816874250769615,
      "learning_rate": 0.0012,
      "loss": 2.721273899078369,
      "step": 728
    },
    {
      "epoch": 0.05918913919836216,
      "grad_norm": 0.06182929128408432,
      "learning_rate": 0.0012,
      "loss": 2.7765350341796875,
      "step": 730
    },
    {
      "epoch": 0.059351301223563156,
      "grad_norm": 0.06214679777622223,
      "learning_rate": 0.0012,
      "loss": 2.829639434814453,
      "step": 732
    },
    {
      "epoch": 0.05951346324876415,
      "grad_norm": 0.02848348207771778,
      "learning_rate": 0.0012,
      "loss": 2.7520251274108887,
      "step": 734
    },
    {
      "epoch": 0.05967562527396514,
      "grad_norm": 0.061107609421014786,
      "learning_rate": 0.0012,
      "loss": 2.803643226623535,
      "step": 736
    },
    {
      "epoch": 0.05983778729916613,
      "grad_norm": 0.029128488153219223,
      "learning_rate": 0.0012,
      "loss": 2.7634835243225098,
      "step": 738
    },
    {
      "epoch": 0.05999994932436713,
      "grad_norm": 0.027682267129421234,
      "learning_rate": 0.0012,
      "loss": 2.815765857696533,
      "step": 740
    },
    {
      "epoch": 0.060162111349568115,
      "grad_norm": 0.06570383906364441,
      "learning_rate": 0.0012,
      "loss": 2.74450945854187,
      "step": 742
    },
    {
      "epoch": 0.06032427337476911,
      "grad_norm": 0.035338908433914185,
      "learning_rate": 0.0012,
      "loss": 2.8067989349365234,
      "step": 744
    },
    {
      "epoch": 0.060486435399970104,
      "grad_norm": 0.030907299369573593,
      "learning_rate": 0.0012,
      "loss": 2.7756733894348145,
      "step": 746
    },
    {
      "epoch": 0.06064859742517109,
      "grad_norm": 0.05596936494112015,
      "learning_rate": 0.0012,
      "loss": 2.8504397869110107,
      "step": 748
    },
    {
      "epoch": 0.060810759450372086,
      "grad_norm": 0.03347066417336464,
      "learning_rate": 0.0012,
      "loss": 2.772042751312256,
      "step": 750
    },
    {
      "epoch": 0.06097292147557308,
      "grad_norm": 0.02728777565062046,
      "learning_rate": 0.0012,
      "loss": 2.8392181396484375,
      "step": 752
    },
    {
      "epoch": 0.06113508350077407,
      "grad_norm": 0.027127817273139954,
      "learning_rate": 0.0012,
      "loss": 2.8267297744750977,
      "step": 754
    },
    {
      "epoch": 0.06129724552597506,
      "grad_norm": 0.03181108087301254,
      "learning_rate": 0.0012,
      "loss": 2.7723388671875,
      "step": 756
    },
    {
      "epoch": 0.06145940755117606,
      "grad_norm": 0.032515935599803925,
      "learning_rate": 0.0012,
      "loss": 2.8127593994140625,
      "step": 758
    },
    {
      "epoch": 0.061621569576377044,
      "grad_norm": 0.022973768413066864,
      "learning_rate": 0.0012,
      "loss": 2.7951812744140625,
      "step": 760
    },
    {
      "epoch": 0.06178373160157804,
      "grad_norm": 0.04086555168032646,
      "learning_rate": 0.0012,
      "loss": 2.797055244445801,
      "step": 762
    },
    {
      "epoch": 0.06194589362677903,
      "grad_norm": 0.04892946034669876,
      "learning_rate": 0.0012,
      "loss": 2.810792922973633,
      "step": 764
    },
    {
      "epoch": 0.06210805565198002,
      "grad_norm": 0.04004444181919098,
      "learning_rate": 0.0012,
      "loss": 2.823395252227783,
      "step": 766
    },
    {
      "epoch": 0.062270217677181015,
      "grad_norm": 0.029567470774054527,
      "learning_rate": 0.0012,
      "loss": 2.783790349960327,
      "step": 768
    },
    {
      "epoch": 0.06243237970238201,
      "grad_norm": 0.03947019577026367,
      "learning_rate": 0.0012,
      "loss": 2.8382463455200195,
      "step": 770
    },
    {
      "epoch": 0.062594541727583,
      "grad_norm": 0.028545424342155457,
      "learning_rate": 0.0012,
      "loss": 2.8311233520507812,
      "step": 772
    },
    {
      "epoch": 0.062756703752784,
      "grad_norm": 0.025397077202796936,
      "learning_rate": 0.0012,
      "loss": 2.762181282043457,
      "step": 774
    },
    {
      "epoch": 0.06291886577798499,
      "grad_norm": 0.024300845339894295,
      "learning_rate": 0.0012,
      "loss": 2.7773141860961914,
      "step": 776
    },
    {
      "epoch": 0.06308102780318597,
      "grad_norm": 0.0369003564119339,
      "learning_rate": 0.0012,
      "loss": 2.819627285003662,
      "step": 778
    },
    {
      "epoch": 0.06324318982838698,
      "grad_norm": 0.024226414039731026,
      "learning_rate": 0.0012,
      "loss": 2.8142220973968506,
      "step": 780
    },
    {
      "epoch": 0.06340535185358796,
      "grad_norm": 0.024554291740059853,
      "learning_rate": 0.0012,
      "loss": 2.758183479309082,
      "step": 782
    },
    {
      "epoch": 0.06356751387878895,
      "grad_norm": 0.03885577246546745,
      "learning_rate": 0.0012,
      "loss": 2.66410493850708,
      "step": 784
    },
    {
      "epoch": 0.06372967590398995,
      "grad_norm": 0.036815788596868515,
      "learning_rate": 0.0012,
      "loss": 2.753741979598999,
      "step": 786
    },
    {
      "epoch": 0.06389183792919094,
      "grad_norm": 0.04101785644888878,
      "learning_rate": 0.0012,
      "loss": 2.824448347091675,
      "step": 788
    },
    {
      "epoch": 0.06405399995439193,
      "grad_norm": 0.029839634895324707,
      "learning_rate": 0.0012,
      "loss": 2.8299484252929688,
      "step": 790
    },
    {
      "epoch": 0.06421616197959293,
      "grad_norm": 0.025884294882416725,
      "learning_rate": 0.0012,
      "loss": 2.81380558013916,
      "step": 792
    },
    {
      "epoch": 0.06437832400479392,
      "grad_norm": 0.03394615650177002,
      "learning_rate": 0.0012,
      "loss": 2.78205943107605,
      "step": 794
    },
    {
      "epoch": 0.0645404860299949,
      "grad_norm": 0.03133104741573334,
      "learning_rate": 0.0012,
      "loss": 2.7857553958892822,
      "step": 796
    },
    {
      "epoch": 0.0647026480551959,
      "grad_norm": 0.03087289072573185,
      "learning_rate": 0.0012,
      "loss": 2.8368561267852783,
      "step": 798
    },
    {
      "epoch": 0.06486481008039689,
      "grad_norm": 0.0237034372985363,
      "learning_rate": 0.0012,
      "loss": 2.806675910949707,
      "step": 800
    },
    {
      "epoch": 0.06502697210559788,
      "grad_norm": 0.026956629008054733,
      "learning_rate": 0.0012,
      "loss": 2.787815570831299,
      "step": 802
    },
    {
      "epoch": 0.06518913413079888,
      "grad_norm": 0.02599433995783329,
      "learning_rate": 0.0012,
      "loss": 2.813039779663086,
      "step": 804
    },
    {
      "epoch": 0.06535129615599987,
      "grad_norm": 0.024729805067181587,
      "learning_rate": 0.0012,
      "loss": 2.817988634109497,
      "step": 806
    },
    {
      "epoch": 0.06551345818120086,
      "grad_norm": 0.03431496396660805,
      "learning_rate": 0.0012,
      "loss": 2.7925562858581543,
      "step": 808
    },
    {
      "epoch": 0.06567562020640186,
      "grad_norm": 0.025360431522130966,
      "learning_rate": 0.0012,
      "loss": 2.779433012008667,
      "step": 810
    },
    {
      "epoch": 0.06583778223160285,
      "grad_norm": 0.024445965886116028,
      "learning_rate": 0.0012,
      "loss": 2.8073272705078125,
      "step": 812
    },
    {
      "epoch": 0.06599994425680383,
      "grad_norm": 0.034767016768455505,
      "learning_rate": 0.0012,
      "loss": 2.810288906097412,
      "step": 814
    },
    {
      "epoch": 0.06616210628200483,
      "grad_norm": 0.045350708067417145,
      "learning_rate": 0.0012,
      "loss": 2.797250747680664,
      "step": 816
    },
    {
      "epoch": 0.06632426830720582,
      "grad_norm": 0.053747858852148056,
      "learning_rate": 0.0012,
      "loss": 2.750704288482666,
      "step": 818
    },
    {
      "epoch": 0.06648643033240681,
      "grad_norm": 0.08504091203212738,
      "learning_rate": 0.0012,
      "loss": 2.772592067718506,
      "step": 820
    },
    {
      "epoch": 0.06664859235760781,
      "grad_norm": 0.06701505929231644,
      "learning_rate": 0.0012,
      "loss": 2.767937421798706,
      "step": 822
    },
    {
      "epoch": 0.0668107543828088,
      "grad_norm": 0.039016690105199814,
      "learning_rate": 0.0012,
      "loss": 2.7261157035827637,
      "step": 824
    },
    {
      "epoch": 0.06697291640800979,
      "grad_norm": 0.033956386148929596,
      "learning_rate": 0.0012,
      "loss": 2.8389339447021484,
      "step": 826
    },
    {
      "epoch": 0.06713507843321079,
      "grad_norm": 0.032032426446676254,
      "learning_rate": 0.0012,
      "loss": 2.7438859939575195,
      "step": 828
    },
    {
      "epoch": 0.06729724045841178,
      "grad_norm": 0.04052675515413284,
      "learning_rate": 0.0012,
      "loss": 2.785806179046631,
      "step": 830
    },
    {
      "epoch": 0.06745940248361276,
      "grad_norm": 0.02818346582353115,
      "learning_rate": 0.0012,
      "loss": 2.7568745613098145,
      "step": 832
    },
    {
      "epoch": 0.06762156450881376,
      "grad_norm": 0.026192141696810722,
      "learning_rate": 0.0012,
      "loss": 2.8302643299102783,
      "step": 834
    },
    {
      "epoch": 0.06778372653401475,
      "grad_norm": 0.02827424183487892,
      "learning_rate": 0.0012,
      "loss": 2.779984951019287,
      "step": 836
    },
    {
      "epoch": 0.06794588855921574,
      "grad_norm": 0.03487604111433029,
      "learning_rate": 0.0012,
      "loss": 2.702725410461426,
      "step": 838
    },
    {
      "epoch": 0.06810805058441674,
      "grad_norm": 0.045235466212034225,
      "learning_rate": 0.0012,
      "loss": 2.832655906677246,
      "step": 840
    },
    {
      "epoch": 0.06827021260961773,
      "grad_norm": 0.02925250679254532,
      "learning_rate": 0.0012,
      "loss": 2.8279879093170166,
      "step": 842
    },
    {
      "epoch": 0.06843237463481872,
      "grad_norm": 0.02353719249367714,
      "learning_rate": 0.0012,
      "loss": 2.769491195678711,
      "step": 844
    },
    {
      "epoch": 0.06859453666001972,
      "grad_norm": 0.02411200851202011,
      "learning_rate": 0.0012,
      "loss": 2.826003074645996,
      "step": 846
    },
    {
      "epoch": 0.0687566986852207,
      "grad_norm": 0.035606566816568375,
      "learning_rate": 0.0012,
      "loss": 2.829493522644043,
      "step": 848
    },
    {
      "epoch": 0.06891886071042169,
      "grad_norm": 0.03537715971469879,
      "learning_rate": 0.0012,
      "loss": 2.7795121669769287,
      "step": 850
    },
    {
      "epoch": 0.0690810227356227,
      "grad_norm": 0.052140604704618454,
      "learning_rate": 0.0012,
      "loss": 2.7831873893737793,
      "step": 852
    },
    {
      "epoch": 0.06924318476082368,
      "grad_norm": 0.02908746525645256,
      "learning_rate": 0.0012,
      "loss": 2.7725038528442383,
      "step": 854
    },
    {
      "epoch": 0.06940534678602467,
      "grad_norm": 0.025618096813559532,
      "learning_rate": 0.0012,
      "loss": 2.805943012237549,
      "step": 856
    },
    {
      "epoch": 0.06956750881122567,
      "grad_norm": 0.03063592128455639,
      "learning_rate": 0.0012,
      "loss": 2.79844331741333,
      "step": 858
    },
    {
      "epoch": 0.06972967083642666,
      "grad_norm": 0.02835642360150814,
      "learning_rate": 0.0012,
      "loss": 2.779445171356201,
      "step": 860
    },
    {
      "epoch": 0.06989183286162765,
      "grad_norm": 0.06037187948822975,
      "learning_rate": 0.0012,
      "loss": 2.7961087226867676,
      "step": 862
    },
    {
      "epoch": 0.07005399488682865,
      "grad_norm": 0.03179816156625748,
      "learning_rate": 0.0012,
      "loss": 2.833096504211426,
      "step": 864
    },
    {
      "epoch": 0.07021615691202963,
      "grad_norm": 0.04850596934556961,
      "learning_rate": 0.0012,
      "loss": 2.845719814300537,
      "step": 866
    },
    {
      "epoch": 0.07037831893723062,
      "grad_norm": 0.02338014915585518,
      "learning_rate": 0.0012,
      "loss": 2.845247268676758,
      "step": 868
    },
    {
      "epoch": 0.07054048096243162,
      "grad_norm": 0.02516208402812481,
      "learning_rate": 0.0012,
      "loss": 2.7131333351135254,
      "step": 870
    },
    {
      "epoch": 0.07070264298763261,
      "grad_norm": 0.02463376522064209,
      "learning_rate": 0.0012,
      "loss": 2.714649200439453,
      "step": 872
    },
    {
      "epoch": 0.0708648050128336,
      "grad_norm": 0.05112173780798912,
      "learning_rate": 0.0012,
      "loss": 2.816636085510254,
      "step": 874
    },
    {
      "epoch": 0.0710269670380346,
      "grad_norm": 0.025543179363012314,
      "learning_rate": 0.0012,
      "loss": 2.7541613578796387,
      "step": 876
    },
    {
      "epoch": 0.07118912906323559,
      "grad_norm": 0.028953127562999725,
      "learning_rate": 0.0012,
      "loss": 2.796741008758545,
      "step": 878
    },
    {
      "epoch": 0.07135129108843657,
      "grad_norm": 0.022424345836043358,
      "learning_rate": 0.0012,
      "loss": 2.8408074378967285,
      "step": 880
    },
    {
      "epoch": 0.07151345311363758,
      "grad_norm": 0.025431664660573006,
      "learning_rate": 0.0012,
      "loss": 2.804001808166504,
      "step": 882
    },
    {
      "epoch": 0.07167561513883856,
      "grad_norm": 0.02760249376296997,
      "learning_rate": 0.0012,
      "loss": 2.7420363426208496,
      "step": 884
    },
    {
      "epoch": 0.07183777716403955,
      "grad_norm": 0.05799056962132454,
      "learning_rate": 0.0012,
      "loss": 2.771162986755371,
      "step": 886
    },
    {
      "epoch": 0.07199993918924055,
      "grad_norm": 0.04749896004796028,
      "learning_rate": 0.0012,
      "loss": 2.738199472427368,
      "step": 888
    },
    {
      "epoch": 0.07216210121444154,
      "grad_norm": 0.029499683529138565,
      "learning_rate": 0.0012,
      "loss": 2.85471773147583,
      "step": 890
    },
    {
      "epoch": 0.07232426323964253,
      "grad_norm": 0.030919194221496582,
      "learning_rate": 0.0012,
      "loss": 2.8237147331237793,
      "step": 892
    },
    {
      "epoch": 0.07248642526484353,
      "grad_norm": 0.0450918935239315,
      "learning_rate": 0.0012,
      "loss": 2.733566999435425,
      "step": 894
    },
    {
      "epoch": 0.07264858729004452,
      "grad_norm": 0.06414259225130081,
      "learning_rate": 0.0012,
      "loss": 2.7656538486480713,
      "step": 896
    },
    {
      "epoch": 0.0728107493152455,
      "grad_norm": 0.03156546875834465,
      "learning_rate": 0.0012,
      "loss": 2.7540946006774902,
      "step": 898
    },
    {
      "epoch": 0.0729729113404465,
      "grad_norm": 0.02481449581682682,
      "learning_rate": 0.0012,
      "loss": 2.8112637996673584,
      "step": 900
    },
    {
      "epoch": 0.0731350733656475,
      "grad_norm": 0.0348629355430603,
      "learning_rate": 0.0012,
      "loss": 2.765500545501709,
      "step": 902
    },
    {
      "epoch": 0.07329723539084848,
      "grad_norm": 0.05459243431687355,
      "learning_rate": 0.0012,
      "loss": 2.751729965209961,
      "step": 904
    },
    {
      "epoch": 0.07345939741604948,
      "grad_norm": 0.02861364744603634,
      "learning_rate": 0.0012,
      "loss": 2.76962947845459,
      "step": 906
    },
    {
      "epoch": 0.07362155944125047,
      "grad_norm": 0.032960131764411926,
      "learning_rate": 0.0012,
      "loss": 2.7886404991149902,
      "step": 908
    },
    {
      "epoch": 0.07378372146645146,
      "grad_norm": 0.024978509172797203,
      "learning_rate": 0.0012,
      "loss": 2.784071683883667,
      "step": 910
    },
    {
      "epoch": 0.07394588349165246,
      "grad_norm": 0.02670152671635151,
      "learning_rate": 0.0012,
      "loss": 2.7669010162353516,
      "step": 912
    },
    {
      "epoch": 0.07410804551685345,
      "grad_norm": 0.028396405279636383,
      "learning_rate": 0.0012,
      "loss": 2.7421586513519287,
      "step": 914
    },
    {
      "epoch": 0.07427020754205445,
      "grad_norm": 0.024763578549027443,
      "learning_rate": 0.0012,
      "loss": 2.843698501586914,
      "step": 916
    },
    {
      "epoch": 0.07443236956725544,
      "grad_norm": 0.02642650716006756,
      "learning_rate": 0.0012,
      "loss": 2.7192554473876953,
      "step": 918
    },
    {
      "epoch": 0.07459453159245642,
      "grad_norm": 0.02783787250518799,
      "learning_rate": 0.0012,
      "loss": 2.7670631408691406,
      "step": 920
    },
    {
      "epoch": 0.07475669361765742,
      "grad_norm": 0.031894829124212265,
      "learning_rate": 0.0012,
      "loss": 2.820521354675293,
      "step": 922
    },
    {
      "epoch": 0.07491885564285841,
      "grad_norm": 0.03239092603325844,
      "learning_rate": 0.0012,
      "loss": 2.8430583477020264,
      "step": 924
    },
    {
      "epoch": 0.0750810176680594,
      "grad_norm": 0.02872905693948269,
      "learning_rate": 0.0012,
      "loss": 2.815396308898926,
      "step": 926
    },
    {
      "epoch": 0.0752431796932604,
      "grad_norm": 0.027452627196907997,
      "learning_rate": 0.0012,
      "loss": 2.7703070640563965,
      "step": 928
    },
    {
      "epoch": 0.07540534171846139,
      "grad_norm": 0.02415386587381363,
      "learning_rate": 0.0012,
      "loss": 2.745851993560791,
      "step": 930
    },
    {
      "epoch": 0.07556750374366238,
      "grad_norm": 0.029828744009137154,
      "learning_rate": 0.0012,
      "loss": 2.786896228790283,
      "step": 932
    },
    {
      "epoch": 0.07572966576886338,
      "grad_norm": 0.021135564893484116,
      "learning_rate": 0.0012,
      "loss": 2.807582378387451,
      "step": 934
    },
    {
      "epoch": 0.07589182779406436,
      "grad_norm": 0.025239141657948494,
      "learning_rate": 0.0012,
      "loss": 2.7566704750061035,
      "step": 936
    },
    {
      "epoch": 0.07605398981926535,
      "grad_norm": 0.025347452610731125,
      "learning_rate": 0.0012,
      "loss": 2.8352484703063965,
      "step": 938
    },
    {
      "epoch": 0.07621615184446635,
      "grad_norm": 0.024229686707258224,
      "learning_rate": 0.0012,
      "loss": 2.7926220893859863,
      "step": 940
    },
    {
      "epoch": 0.07637831386966734,
      "grad_norm": 0.027682164683938026,
      "learning_rate": 0.0012,
      "loss": 2.7090587615966797,
      "step": 942
    },
    {
      "epoch": 0.07654047589486833,
      "grad_norm": 0.04599779471755028,
      "learning_rate": 0.0012,
      "loss": 2.760251522064209,
      "step": 944
    },
    {
      "epoch": 0.07670263792006933,
      "grad_norm": 0.02718338370323181,
      "learning_rate": 0.0012,
      "loss": 2.822281837463379,
      "step": 946
    },
    {
      "epoch": 0.07686479994527032,
      "grad_norm": 0.0723159983754158,
      "learning_rate": 0.0012,
      "loss": 2.720212459564209,
      "step": 948
    },
    {
      "epoch": 0.0770269619704713,
      "grad_norm": 0.05825444683432579,
      "learning_rate": 0.0012,
      "loss": 2.7600760459899902,
      "step": 950
    },
    {
      "epoch": 0.0771891239956723,
      "grad_norm": 0.030322087928652763,
      "learning_rate": 0.0012,
      "loss": 2.7999167442321777,
      "step": 952
    },
    {
      "epoch": 0.0773512860208733,
      "grad_norm": 0.045720212161540985,
      "learning_rate": 0.0012,
      "loss": 2.7893378734588623,
      "step": 954
    },
    {
      "epoch": 0.07751344804607428,
      "grad_norm": 0.04409421980381012,
      "learning_rate": 0.0012,
      "loss": 2.806281089782715,
      "step": 956
    },
    {
      "epoch": 0.07767561007127528,
      "grad_norm": 0.0432383231818676,
      "learning_rate": 0.0012,
      "loss": 2.7087950706481934,
      "step": 958
    },
    {
      "epoch": 0.07783777209647627,
      "grad_norm": 0.026679597795009613,
      "learning_rate": 0.0012,
      "loss": 2.816490888595581,
      "step": 960
    },
    {
      "epoch": 0.07799993412167726,
      "grad_norm": 0.03117910586297512,
      "learning_rate": 0.0012,
      "loss": 2.706005573272705,
      "step": 962
    },
    {
      "epoch": 0.07816209614687826,
      "grad_norm": 0.026260944083333015,
      "learning_rate": 0.0012,
      "loss": 2.77134370803833,
      "step": 964
    },
    {
      "epoch": 0.07832425817207925,
      "grad_norm": 0.036938201636075974,
      "learning_rate": 0.0012,
      "loss": 2.7950849533081055,
      "step": 966
    },
    {
      "epoch": 0.07848642019728024,
      "grad_norm": 0.03832279145717621,
      "learning_rate": 0.0012,
      "loss": 2.8194103240966797,
      "step": 968
    },
    {
      "epoch": 0.07864858222248124,
      "grad_norm": 0.02591184712946415,
      "learning_rate": 0.0012,
      "loss": 2.7933173179626465,
      "step": 970
    },
    {
      "epoch": 0.07881074424768222,
      "grad_norm": 0.025888357311487198,
      "learning_rate": 0.0012,
      "loss": 2.77181339263916,
      "step": 972
    },
    {
      "epoch": 0.07897290627288321,
      "grad_norm": 0.05541665107011795,
      "learning_rate": 0.0012,
      "loss": 2.781693458557129,
      "step": 974
    },
    {
      "epoch": 0.07913506829808421,
      "grad_norm": 0.028384117409586906,
      "learning_rate": 0.0012,
      "loss": 2.8059463500976562,
      "step": 976
    },
    {
      "epoch": 0.0792972303232852,
      "grad_norm": 0.023060407489538193,
      "learning_rate": 0.0012,
      "loss": 2.7344486713409424,
      "step": 978
    },
    {
      "epoch": 0.07945939234848619,
      "grad_norm": 0.024311941117048264,
      "learning_rate": 0.0012,
      "loss": 2.743359088897705,
      "step": 980
    },
    {
      "epoch": 0.07962155437368719,
      "grad_norm": 0.022074192762374878,
      "learning_rate": 0.0012,
      "loss": 2.7640795707702637,
      "step": 982
    },
    {
      "epoch": 0.07978371639888818,
      "grad_norm": 0.024267742410302162,
      "learning_rate": 0.0012,
      "loss": 2.840012550354004,
      "step": 984
    },
    {
      "epoch": 0.07994587842408916,
      "grad_norm": 0.06090960279107094,
      "learning_rate": 0.0012,
      "loss": 2.7791833877563477,
      "step": 986
    },
    {
      "epoch": 0.08010804044929017,
      "grad_norm": 0.06649359315633774,
      "learning_rate": 0.0012,
      "loss": 2.727996349334717,
      "step": 988
    },
    {
      "epoch": 0.08027020247449115,
      "grad_norm": 0.05936787649989128,
      "learning_rate": 0.0012,
      "loss": 2.7204489707946777,
      "step": 990
    },
    {
      "epoch": 0.08043236449969214,
      "grad_norm": 0.024678971618413925,
      "learning_rate": 0.0012,
      "loss": 2.7954349517822266,
      "step": 992
    },
    {
      "epoch": 0.08059452652489314,
      "grad_norm": 0.04641995579004288,
      "learning_rate": 0.0012,
      "loss": 2.722850799560547,
      "step": 994
    },
    {
      "epoch": 0.08075668855009413,
      "grad_norm": 0.03037053346633911,
      "learning_rate": 0.0012,
      "loss": 2.7645506858825684,
      "step": 996
    },
    {
      "epoch": 0.08091885057529512,
      "grad_norm": 0.025215134024620056,
      "learning_rate": 0.0012,
      "loss": 2.7619595527648926,
      "step": 998
    },
    {
      "epoch": 0.08108101260049612,
      "grad_norm": 0.02511732093989849,
      "learning_rate": 0.0012,
      "loss": 2.7923355102539062,
      "step": 1000
    },
    {
      "epoch": 0.0812431746256971,
      "grad_norm": 0.048819512128829956,
      "learning_rate": 0.0012,
      "loss": 2.808535099029541,
      "step": 1002
    },
    {
      "epoch": 0.0814053366508981,
      "grad_norm": 0.052475277334451675,
      "learning_rate": 0.0012,
      "loss": 2.688096046447754,
      "step": 1004
    },
    {
      "epoch": 0.0815674986760991,
      "grad_norm": 0.02680508978664875,
      "learning_rate": 0.0012,
      "loss": 2.72548770904541,
      "step": 1006
    },
    {
      "epoch": 0.08172966070130008,
      "grad_norm": 0.031177466735243797,
      "learning_rate": 0.0012,
      "loss": 2.7714505195617676,
      "step": 1008
    },
    {
      "epoch": 0.08189182272650107,
      "grad_norm": 0.029796253889799118,
      "learning_rate": 0.0012,
      "loss": 2.7348508834838867,
      "step": 1010
    },
    {
      "epoch": 0.08205398475170207,
      "grad_norm": 0.022941160947084427,
      "learning_rate": 0.0012,
      "loss": 2.7542660236358643,
      "step": 1012
    },
    {
      "epoch": 0.08221614677690306,
      "grad_norm": 0.025124235078692436,
      "learning_rate": 0.0012,
      "loss": 2.710036277770996,
      "step": 1014
    },
    {
      "epoch": 0.08237830880210405,
      "grad_norm": 0.02273031696677208,
      "learning_rate": 0.0012,
      "loss": 2.747563362121582,
      "step": 1016
    },
    {
      "epoch": 0.08254047082730505,
      "grad_norm": 0.02489432692527771,
      "learning_rate": 0.0012,
      "loss": 2.796889305114746,
      "step": 1018
    },
    {
      "epoch": 0.08270263285250604,
      "grad_norm": 0.023969141766428947,
      "learning_rate": 0.0012,
      "loss": 2.7140040397644043,
      "step": 1020
    },
    {
      "epoch": 0.08286479487770702,
      "grad_norm": 0.02375999093055725,
      "learning_rate": 0.0012,
      "loss": 2.7370851039886475,
      "step": 1022
    },
    {
      "epoch": 0.08302695690290803,
      "grad_norm": 0.021311169490218163,
      "learning_rate": 0.0012,
      "loss": 2.775613307952881,
      "step": 1024
    },
    {
      "epoch": 0.08318911892810901,
      "grad_norm": 0.023362906649708748,
      "learning_rate": 0.0012,
      "loss": 2.7674880027770996,
      "step": 1026
    },
    {
      "epoch": 0.08335128095331,
      "grad_norm": 0.04492713510990143,
      "learning_rate": 0.0012,
      "loss": 2.6898679733276367,
      "step": 1028
    },
    {
      "epoch": 0.083513442978511,
      "grad_norm": 0.06262896955013275,
      "learning_rate": 0.0012,
      "loss": 2.790565013885498,
      "step": 1030
    },
    {
      "epoch": 0.08367560500371199,
      "grad_norm": 0.0645773857831955,
      "learning_rate": 0.0012,
      "loss": 2.799091339111328,
      "step": 1032
    },
    {
      "epoch": 0.08383776702891298,
      "grad_norm": 0.032336339354515076,
      "learning_rate": 0.0012,
      "loss": 2.686673641204834,
      "step": 1034
    },
    {
      "epoch": 0.08399992905411398,
      "grad_norm": 0.022203318774700165,
      "learning_rate": 0.0012,
      "loss": 2.73785662651062,
      "step": 1036
    },
    {
      "epoch": 0.08416209107931497,
      "grad_norm": 0.043484438210725784,
      "learning_rate": 0.0012,
      "loss": 2.7503318786621094,
      "step": 1038
    },
    {
      "epoch": 0.08432425310451595,
      "grad_norm": 0.03414354845881462,
      "learning_rate": 0.0012,
      "loss": 2.712611675262451,
      "step": 1040
    },
    {
      "epoch": 0.08448641512971695,
      "grad_norm": 0.024671901017427444,
      "learning_rate": 0.0012,
      "loss": 2.749001979827881,
      "step": 1042
    },
    {
      "epoch": 0.08464857715491794,
      "grad_norm": 0.0292985662817955,
      "learning_rate": 0.0012,
      "loss": 2.854358196258545,
      "step": 1044
    },
    {
      "epoch": 0.08481073918011893,
      "grad_norm": 0.03245013952255249,
      "learning_rate": 0.0012,
      "loss": 2.811643600463867,
      "step": 1046
    },
    {
      "epoch": 0.08497290120531993,
      "grad_norm": 0.028530344367027283,
      "learning_rate": 0.0012,
      "loss": 2.74368953704834,
      "step": 1048
    },
    {
      "epoch": 0.08513506323052092,
      "grad_norm": 0.030086925253272057,
      "learning_rate": 0.0012,
      "loss": 2.745392322540283,
      "step": 1050
    },
    {
      "epoch": 0.0852972252557219,
      "grad_norm": 0.027866575866937637,
      "learning_rate": 0.0012,
      "loss": 2.790403366088867,
      "step": 1052
    },
    {
      "epoch": 0.08545938728092291,
      "grad_norm": 0.041287779808044434,
      "learning_rate": 0.0012,
      "loss": 2.7675869464874268,
      "step": 1054
    },
    {
      "epoch": 0.0856215493061239,
      "grad_norm": 0.02807474322617054,
      "learning_rate": 0.0012,
      "loss": 2.752105236053467,
      "step": 1056
    },
    {
      "epoch": 0.08578371133132488,
      "grad_norm": 0.025875110179185867,
      "learning_rate": 0.0012,
      "loss": 2.7977962493896484,
      "step": 1058
    },
    {
      "epoch": 0.08594587335652588,
      "grad_norm": 0.02795221097767353,
      "learning_rate": 0.0012,
      "loss": 2.731785297393799,
      "step": 1060
    },
    {
      "epoch": 0.08610803538172687,
      "grad_norm": 0.02454310841858387,
      "learning_rate": 0.0012,
      "loss": 2.7293553352355957,
      "step": 1062
    },
    {
      "epoch": 0.08627019740692786,
      "grad_norm": 0.024422883987426758,
      "learning_rate": 0.0012,
      "loss": 2.751359462738037,
      "step": 1064
    },
    {
      "epoch": 0.08643235943212886,
      "grad_norm": 0.02392309531569481,
      "learning_rate": 0.0012,
      "loss": 2.7016305923461914,
      "step": 1066
    },
    {
      "epoch": 0.08659452145732985,
      "grad_norm": 0.04153026267886162,
      "learning_rate": 0.0012,
      "loss": 2.7328882217407227,
      "step": 1068
    },
    {
      "epoch": 0.08675668348253084,
      "grad_norm": 0.03895338997244835,
      "learning_rate": 0.0012,
      "loss": 2.78132963180542,
      "step": 1070
    },
    {
      "epoch": 0.08691884550773184,
      "grad_norm": 0.030635757371783257,
      "learning_rate": 0.0012,
      "loss": 2.7614450454711914,
      "step": 1072
    },
    {
      "epoch": 0.08708100753293282,
      "grad_norm": 0.02579147182404995,
      "learning_rate": 0.0012,
      "loss": 2.6937599182128906,
      "step": 1074
    },
    {
      "epoch": 0.08724316955813381,
      "grad_norm": 0.029640695080161095,
      "learning_rate": 0.0012,
      "loss": 2.779022216796875,
      "step": 1076
    },
    {
      "epoch": 0.08740533158333481,
      "grad_norm": 0.0364222377538681,
      "learning_rate": 0.0012,
      "loss": 2.734147787094116,
      "step": 1078
    },
    {
      "epoch": 0.0875674936085358,
      "grad_norm": 0.03568345308303833,
      "learning_rate": 0.0012,
      "loss": 2.807281970977783,
      "step": 1080
    },
    {
      "epoch": 0.0877296556337368,
      "grad_norm": 0.028410300612449646,
      "learning_rate": 0.0012,
      "loss": 2.801797389984131,
      "step": 1082
    },
    {
      "epoch": 0.08789181765893779,
      "grad_norm": 0.02672436833381653,
      "learning_rate": 0.0012,
      "loss": 2.7180299758911133,
      "step": 1084
    },
    {
      "epoch": 0.08805397968413878,
      "grad_norm": 0.03266593813896179,
      "learning_rate": 0.0012,
      "loss": 2.7161946296691895,
      "step": 1086
    },
    {
      "epoch": 0.08821614170933978,
      "grad_norm": 0.04000142589211464,
      "learning_rate": 0.0012,
      "loss": 2.744049310684204,
      "step": 1088
    },
    {
      "epoch": 0.08837830373454077,
      "grad_norm": 0.08053399622440338,
      "learning_rate": 0.0012,
      "loss": 2.7895689010620117,
      "step": 1090
    },
    {
      "epoch": 0.08854046575974175,
      "grad_norm": 0.03901616111397743,
      "learning_rate": 0.0012,
      "loss": 2.7513537406921387,
      "step": 1092
    },
    {
      "epoch": 0.08870262778494276,
      "grad_norm": 0.02465078793466091,
      "learning_rate": 0.0012,
      "loss": 2.7239041328430176,
      "step": 1094
    },
    {
      "epoch": 0.08886478981014374,
      "grad_norm": 0.04712425544857979,
      "learning_rate": 0.0012,
      "loss": 2.779444694519043,
      "step": 1096
    },
    {
      "epoch": 0.08902695183534473,
      "grad_norm": 0.049210790544748306,
      "learning_rate": 0.0012,
      "loss": 2.7917919158935547,
      "step": 1098
    },
    {
      "epoch": 0.08918911386054573,
      "grad_norm": 0.03542398661375046,
      "learning_rate": 0.0012,
      "loss": 2.754694938659668,
      "step": 1100
    },
    {
      "epoch": 0.08935127588574672,
      "grad_norm": 0.025969313457608223,
      "learning_rate": 0.0012,
      "loss": 2.7661666870117188,
      "step": 1102
    },
    {
      "epoch": 0.08951343791094771,
      "grad_norm": 0.0233487319201231,
      "learning_rate": 0.0012,
      "loss": 2.7222137451171875,
      "step": 1104
    },
    {
      "epoch": 0.08967559993614871,
      "grad_norm": 0.026491370052099228,
      "learning_rate": 0.0012,
      "loss": 2.7061703205108643,
      "step": 1106
    },
    {
      "epoch": 0.0898377619613497,
      "grad_norm": 0.040290359407663345,
      "learning_rate": 0.0012,
      "loss": 2.800100326538086,
      "step": 1108
    },
    {
      "epoch": 0.08999992398655068,
      "grad_norm": 0.04862072691321373,
      "learning_rate": 0.0012,
      "loss": 2.748018264770508,
      "step": 1110
    },
    {
      "epoch": 0.09016208601175169,
      "grad_norm": 0.028823379427194595,
      "learning_rate": 0.0012,
      "loss": 2.7595653533935547,
      "step": 1112
    },
    {
      "epoch": 0.09032424803695267,
      "grad_norm": 0.025598380714654922,
      "learning_rate": 0.0012,
      "loss": 2.821131467819214,
      "step": 1114
    },
    {
      "epoch": 0.09048641006215366,
      "grad_norm": 0.022280829027295113,
      "learning_rate": 0.0012,
      "loss": 2.784642219543457,
      "step": 1116
    },
    {
      "epoch": 0.09064857208735466,
      "grad_norm": 0.038682203739881516,
      "learning_rate": 0.0012,
      "loss": 2.7989282608032227,
      "step": 1118
    },
    {
      "epoch": 0.09081073411255565,
      "grad_norm": 0.025830470025539398,
      "learning_rate": 0.0012,
      "loss": 2.7720770835876465,
      "step": 1120
    },
    {
      "epoch": 0.09097289613775664,
      "grad_norm": 0.027381742373108864,
      "learning_rate": 0.0012,
      "loss": 2.786921501159668,
      "step": 1122
    },
    {
      "epoch": 0.09113505816295764,
      "grad_norm": 0.03642972558736801,
      "learning_rate": 0.0012,
      "loss": 2.754455804824829,
      "step": 1124
    },
    {
      "epoch": 0.09129722018815863,
      "grad_norm": 0.04399402439594269,
      "learning_rate": 0.0012,
      "loss": 2.8094966411590576,
      "step": 1126
    },
    {
      "epoch": 0.09145938221335961,
      "grad_norm": 0.06144993007183075,
      "learning_rate": 0.0012,
      "loss": 2.776930332183838,
      "step": 1128
    },
    {
      "epoch": 0.09162154423856061,
      "grad_norm": 0.04460948705673218,
      "learning_rate": 0.0012,
      "loss": 2.739543914794922,
      "step": 1130
    },
    {
      "epoch": 0.0917837062637616,
      "grad_norm": 0.03262336552143097,
      "learning_rate": 0.0012,
      "loss": 2.765697479248047,
      "step": 1132
    },
    {
      "epoch": 0.09194586828896259,
      "grad_norm": 0.02276168018579483,
      "learning_rate": 0.0012,
      "loss": 2.7543463706970215,
      "step": 1134
    },
    {
      "epoch": 0.09210803031416359,
      "grad_norm": 0.07815255224704742,
      "learning_rate": 0.0012,
      "loss": 2.774268627166748,
      "step": 1136
    },
    {
      "epoch": 0.09227019233936458,
      "grad_norm": 0.051665641367435455,
      "learning_rate": 0.0012,
      "loss": 2.698397159576416,
      "step": 1138
    },
    {
      "epoch": 0.09243235436456557,
      "grad_norm": 0.026452740654349327,
      "learning_rate": 0.0012,
      "loss": 2.7602243423461914,
      "step": 1140
    },
    {
      "epoch": 0.09259451638976657,
      "grad_norm": 0.05439820513129234,
      "learning_rate": 0.0012,
      "loss": 2.697218418121338,
      "step": 1142
    },
    {
      "epoch": 0.09275667841496756,
      "grad_norm": 0.030556757003068924,
      "learning_rate": 0.0012,
      "loss": 2.770596504211426,
      "step": 1144
    },
    {
      "epoch": 0.09291884044016854,
      "grad_norm": 0.025409957394003868,
      "learning_rate": 0.0012,
      "loss": 2.7264060974121094,
      "step": 1146
    },
    {
      "epoch": 0.09308100246536954,
      "grad_norm": 0.02610226906836033,
      "learning_rate": 0.0012,
      "loss": 2.7310783863067627,
      "step": 1148
    },
    {
      "epoch": 0.09324316449057053,
      "grad_norm": 0.03229987248778343,
      "learning_rate": 0.0012,
      "loss": 2.774404764175415,
      "step": 1150
    },
    {
      "epoch": 0.09340532651577152,
      "grad_norm": 0.030262308195233345,
      "learning_rate": 0.0012,
      "loss": 2.817232131958008,
      "step": 1152
    },
    {
      "epoch": 0.09356748854097252,
      "grad_norm": 0.04802423343062401,
      "learning_rate": 0.0012,
      "loss": 2.7392044067382812,
      "step": 1154
    },
    {
      "epoch": 0.09372965056617351,
      "grad_norm": 0.04878299683332443,
      "learning_rate": 0.0012,
      "loss": 2.767995834350586,
      "step": 1156
    },
    {
      "epoch": 0.0938918125913745,
      "grad_norm": 0.030496280640363693,
      "learning_rate": 0.0012,
      "loss": 2.7353549003601074,
      "step": 1158
    },
    {
      "epoch": 0.0940539746165755,
      "grad_norm": 0.03691063076257706,
      "learning_rate": 0.0012,
      "loss": 2.7707319259643555,
      "step": 1160
    },
    {
      "epoch": 0.09421613664177649,
      "grad_norm": 0.03194038197398186,
      "learning_rate": 0.0012,
      "loss": 2.795442819595337,
      "step": 1162
    },
    {
      "epoch": 0.09437829866697747,
      "grad_norm": 0.026768658310174942,
      "learning_rate": 0.0012,
      "loss": 2.7839550971984863,
      "step": 1164
    },
    {
      "epoch": 0.09454046069217847,
      "grad_norm": 0.028010308742523193,
      "learning_rate": 0.0012,
      "loss": 2.73832106590271,
      "step": 1166
    },
    {
      "epoch": 0.09470262271737946,
      "grad_norm": 0.027960728853940964,
      "learning_rate": 0.0012,
      "loss": 2.7796201705932617,
      "step": 1168
    },
    {
      "epoch": 0.09486478474258045,
      "grad_norm": 0.04547376558184624,
      "learning_rate": 0.0012,
      "loss": 2.762030601501465,
      "step": 1170
    },
    {
      "epoch": 0.09502694676778145,
      "grad_norm": 0.02435302734375,
      "learning_rate": 0.0012,
      "loss": 2.6880838871002197,
      "step": 1172
    },
    {
      "epoch": 0.09518910879298244,
      "grad_norm": 0.0351199172437191,
      "learning_rate": 0.0012,
      "loss": 2.780952215194702,
      "step": 1174
    },
    {
      "epoch": 0.09535127081818343,
      "grad_norm": 0.05042312666773796,
      "learning_rate": 0.0012,
      "loss": 2.7097830772399902,
      "step": 1176
    },
    {
      "epoch": 0.09551343284338443,
      "grad_norm": 0.02355182357132435,
      "learning_rate": 0.0012,
      "loss": 2.7913482189178467,
      "step": 1178
    },
    {
      "epoch": 0.09567559486858541,
      "grad_norm": 0.04151438921689987,
      "learning_rate": 0.0012,
      "loss": 2.752685546875,
      "step": 1180
    },
    {
      "epoch": 0.0958377568937864,
      "grad_norm": 0.06660644710063934,
      "learning_rate": 0.0012,
      "loss": 2.755188465118408,
      "step": 1182
    },
    {
      "epoch": 0.0959999189189874,
      "grad_norm": 0.04496704787015915,
      "learning_rate": 0.0012,
      "loss": 2.752829074859619,
      "step": 1184
    },
    {
      "epoch": 0.09616208094418839,
      "grad_norm": 0.02510944940149784,
      "learning_rate": 0.0012,
      "loss": 2.7343034744262695,
      "step": 1186
    },
    {
      "epoch": 0.09632424296938938,
      "grad_norm": 0.056764595210552216,
      "learning_rate": 0.0012,
      "loss": 2.7603721618652344,
      "step": 1188
    },
    {
      "epoch": 0.09648640499459038,
      "grad_norm": 0.0532778762280941,
      "learning_rate": 0.0012,
      "loss": 2.850139617919922,
      "step": 1190
    },
    {
      "epoch": 0.09664856701979137,
      "grad_norm": 0.02853897400200367,
      "learning_rate": 0.0012,
      "loss": 2.7569329738616943,
      "step": 1192
    },
    {
      "epoch": 0.09681072904499236,
      "grad_norm": 0.04359349235892296,
      "learning_rate": 0.0012,
      "loss": 2.6882362365722656,
      "step": 1194
    },
    {
      "epoch": 0.09697289107019336,
      "grad_norm": 0.0385415181517601,
      "learning_rate": 0.0012,
      "loss": 2.695469379425049,
      "step": 1196
    },
    {
      "epoch": 0.09713505309539434,
      "grad_norm": 0.02929791994392872,
      "learning_rate": 0.0012,
      "loss": 2.759072780609131,
      "step": 1198
    },
    {
      "epoch": 0.09729721512059533,
      "grad_norm": 0.02092457003891468,
      "learning_rate": 0.0012,
      "loss": 2.7560324668884277,
      "step": 1200
    },
    {
      "epoch": 0.09745937714579633,
      "grad_norm": 0.03039129078388214,
      "learning_rate": 0.0012,
      "loss": 2.7911787033081055,
      "step": 1202
    },
    {
      "epoch": 0.09762153917099732,
      "grad_norm": 0.03591299429535866,
      "learning_rate": 0.0012,
      "loss": 2.7280561923980713,
      "step": 1204
    },
    {
      "epoch": 0.09778370119619831,
      "grad_norm": 0.029090706259012222,
      "learning_rate": 0.0012,
      "loss": 2.7574658393859863,
      "step": 1206
    },
    {
      "epoch": 0.09794586322139931,
      "grad_norm": 0.02829216793179512,
      "learning_rate": 0.0012,
      "loss": 2.7861075401306152,
      "step": 1208
    },
    {
      "epoch": 0.0981080252466003,
      "grad_norm": 0.027130667120218277,
      "learning_rate": 0.0012,
      "loss": 2.7064902782440186,
      "step": 1210
    },
    {
      "epoch": 0.09827018727180128,
      "grad_norm": 0.022945718839764595,
      "learning_rate": 0.0012,
      "loss": 2.809319019317627,
      "step": 1212
    },
    {
      "epoch": 0.09843234929700229,
      "grad_norm": 0.03205443173646927,
      "learning_rate": 0.0012,
      "loss": 2.730339527130127,
      "step": 1214
    },
    {
      "epoch": 0.09859451132220327,
      "grad_norm": 0.04260207712650299,
      "learning_rate": 0.0012,
      "loss": 2.7816386222839355,
      "step": 1216
    },
    {
      "epoch": 0.09875667334740426,
      "grad_norm": 0.05965240299701691,
      "learning_rate": 0.0012,
      "loss": 2.7062463760375977,
      "step": 1218
    },
    {
      "epoch": 0.09891883537260526,
      "grad_norm": 0.02470826357603073,
      "learning_rate": 0.0012,
      "loss": 2.7463603019714355,
      "step": 1220
    },
    {
      "epoch": 0.09908099739780625,
      "grad_norm": 0.04084325209259987,
      "learning_rate": 0.0012,
      "loss": 2.746112823486328,
      "step": 1222
    },
    {
      "epoch": 0.09924315942300724,
      "grad_norm": 0.06011278182268143,
      "learning_rate": 0.0012,
      "loss": 2.6953470706939697,
      "step": 1224
    },
    {
      "epoch": 0.09940532144820824,
      "grad_norm": 0.02530486509203911,
      "learning_rate": 0.0012,
      "loss": 2.766115665435791,
      "step": 1226
    },
    {
      "epoch": 0.09956748347340923,
      "grad_norm": 0.024790074676275253,
      "learning_rate": 0.0012,
      "loss": 2.783571481704712,
      "step": 1228
    },
    {
      "epoch": 0.09972964549861021,
      "grad_norm": 0.037671882659196854,
      "learning_rate": 0.0012,
      "loss": 2.687596321105957,
      "step": 1230
    },
    {
      "epoch": 0.09989180752381122,
      "grad_norm": 0.040392909198999405,
      "learning_rate": 0.0012,
      "loss": 2.7882590293884277,
      "step": 1232
    },
    {
      "epoch": 0.1000539695490122,
      "grad_norm": 0.029194243252277374,
      "learning_rate": 0.0012,
      "loss": 2.713937759399414,
      "step": 1234
    },
    {
      "epoch": 0.10021613157421319,
      "grad_norm": 0.025602800771594048,
      "learning_rate": 0.0012,
      "loss": 2.7743921279907227,
      "step": 1236
    },
    {
      "epoch": 0.10037829359941419,
      "grad_norm": 0.032877031713724136,
      "learning_rate": 0.0012,
      "loss": 2.7449562549591064,
      "step": 1238
    },
    {
      "epoch": 0.10054045562461518,
      "grad_norm": 0.05113174393773079,
      "learning_rate": 0.0012,
      "loss": 2.7856178283691406,
      "step": 1240
    },
    {
      "epoch": 0.10070261764981617,
      "grad_norm": 0.030856836587190628,
      "learning_rate": 0.0012,
      "loss": 2.8084206581115723,
      "step": 1242
    },
    {
      "epoch": 0.10086477967501717,
      "grad_norm": 0.024027377367019653,
      "learning_rate": 0.0012,
      "loss": 2.8544321060180664,
      "step": 1244
    },
    {
      "epoch": 0.10102694170021816,
      "grad_norm": 0.024614250287413597,
      "learning_rate": 0.0012,
      "loss": 2.6910505294799805,
      "step": 1246
    },
    {
      "epoch": 0.10118910372541916,
      "grad_norm": 0.027601411566138268,
      "learning_rate": 0.0012,
      "loss": 2.7672157287597656,
      "step": 1248
    },
    {
      "epoch": 0.10135126575062015,
      "grad_norm": 0.04442698135972023,
      "learning_rate": 0.0012,
      "loss": 2.7823472023010254,
      "step": 1250
    },
    {
      "epoch": 0.10151342777582113,
      "grad_norm": 0.023598814383149147,
      "learning_rate": 0.0012,
      "loss": 2.739293098449707,
      "step": 1252
    },
    {
      "epoch": 0.10167558980102213,
      "grad_norm": 0.022507747635245323,
      "learning_rate": 0.0012,
      "loss": 2.683602809906006,
      "step": 1254
    },
    {
      "epoch": 0.10183775182622312,
      "grad_norm": 0.02415647730231285,
      "learning_rate": 0.0012,
      "loss": 2.7098212242126465,
      "step": 1256
    },
    {
      "epoch": 0.10199991385142411,
      "grad_norm": 0.02264566719532013,
      "learning_rate": 0.0012,
      "loss": 2.7709813117980957,
      "step": 1258
    },
    {
      "epoch": 0.10216207587662511,
      "grad_norm": 0.02702777460217476,
      "learning_rate": 0.0012,
      "loss": 2.786830425262451,
      "step": 1260
    },
    {
      "epoch": 0.1023242379018261,
      "grad_norm": 0.023167895153164864,
      "learning_rate": 0.0012,
      "loss": 2.7364678382873535,
      "step": 1262
    },
    {
      "epoch": 0.10248639992702709,
      "grad_norm": 0.02530748024582863,
      "learning_rate": 0.0012,
      "loss": 2.7198081016540527,
      "step": 1264
    },
    {
      "epoch": 0.10264856195222809,
      "grad_norm": 0.030470840632915497,
      "learning_rate": 0.0012,
      "loss": 2.756988525390625,
      "step": 1266
    },
    {
      "epoch": 0.10281072397742907,
      "grad_norm": 0.03320765495300293,
      "learning_rate": 0.0012,
      "loss": 2.7991833686828613,
      "step": 1268
    },
    {
      "epoch": 0.10297288600263006,
      "grad_norm": 0.03422698751091957,
      "learning_rate": 0.0012,
      "loss": 2.7741098403930664,
      "step": 1270
    },
    {
      "epoch": 0.10313504802783106,
      "grad_norm": 0.038381561636924744,
      "learning_rate": 0.0012,
      "loss": 2.731234073638916,
      "step": 1272
    },
    {
      "epoch": 0.10329721005303205,
      "grad_norm": 0.036338284611701965,
      "learning_rate": 0.0012,
      "loss": 2.672527313232422,
      "step": 1274
    },
    {
      "epoch": 0.10345937207823304,
      "grad_norm": 0.023698270320892334,
      "learning_rate": 0.0012,
      "loss": 2.7187418937683105,
      "step": 1276
    },
    {
      "epoch": 0.10362153410343404,
      "grad_norm": 0.03188573941588402,
      "learning_rate": 0.0012,
      "loss": 2.7931437492370605,
      "step": 1278
    },
    {
      "epoch": 0.10378369612863503,
      "grad_norm": 0.02925541065633297,
      "learning_rate": 0.0012,
      "loss": 2.7273130416870117,
      "step": 1280
    },
    {
      "epoch": 0.10394585815383602,
      "grad_norm": 0.02538011595606804,
      "learning_rate": 0.0012,
      "loss": 2.7764580249786377,
      "step": 1282
    },
    {
      "epoch": 0.10410802017903702,
      "grad_norm": 0.027568891644477844,
      "learning_rate": 0.0012,
      "loss": 2.8202147483825684,
      "step": 1284
    },
    {
      "epoch": 0.104270182204238,
      "grad_norm": 0.022250477224588394,
      "learning_rate": 0.0012,
      "loss": 2.8209822177886963,
      "step": 1286
    },
    {
      "epoch": 0.10443234422943899,
      "grad_norm": 0.02799615077674389,
      "learning_rate": 0.0012,
      "loss": 2.8171486854553223,
      "step": 1288
    },
    {
      "epoch": 0.10459450625464,
      "grad_norm": 0.03224198892712593,
      "learning_rate": 0.0012,
      "loss": 2.7370476722717285,
      "step": 1290
    },
    {
      "epoch": 0.10475666827984098,
      "grad_norm": 0.028362056240439415,
      "learning_rate": 0.0012,
      "loss": 2.753636360168457,
      "step": 1292
    },
    {
      "epoch": 0.10491883030504197,
      "grad_norm": 0.0318898968398571,
      "learning_rate": 0.0012,
      "loss": 2.7783284187316895,
      "step": 1294
    },
    {
      "epoch": 0.10508099233024297,
      "grad_norm": 0.0541328564286232,
      "learning_rate": 0.0012,
      "loss": 2.7659811973571777,
      "step": 1296
    },
    {
      "epoch": 0.10524315435544396,
      "grad_norm": 0.032322514802217484,
      "learning_rate": 0.0012,
      "loss": 2.7411041259765625,
      "step": 1298
    },
    {
      "epoch": 0.10540531638064495,
      "grad_norm": 0.024282462894916534,
      "learning_rate": 0.0012,
      "loss": 2.7637813091278076,
      "step": 1300
    },
    {
      "epoch": 0.10556747840584595,
      "grad_norm": 0.022128138691186905,
      "learning_rate": 0.0012,
      "loss": 2.7288966178894043,
      "step": 1302
    },
    {
      "epoch": 0.10572964043104693,
      "grad_norm": 0.04440990462899208,
      "learning_rate": 0.0012,
      "loss": 2.738029956817627,
      "step": 1304
    },
    {
      "epoch": 0.10589180245624792,
      "grad_norm": 0.021835120394825935,
      "learning_rate": 0.0012,
      "loss": 2.83172607421875,
      "step": 1306
    },
    {
      "epoch": 0.10605396448144892,
      "grad_norm": 0.023942964151501656,
      "learning_rate": 0.0012,
      "loss": 2.736145257949829,
      "step": 1308
    },
    {
      "epoch": 0.10621612650664991,
      "grad_norm": 0.023241357877850533,
      "learning_rate": 0.0012,
      "loss": 2.7344002723693848,
      "step": 1310
    },
    {
      "epoch": 0.1063782885318509,
      "grad_norm": 0.03730713203549385,
      "learning_rate": 0.0012,
      "loss": 2.7215051651000977,
      "step": 1312
    },
    {
      "epoch": 0.1065404505570519,
      "grad_norm": 0.02678019367158413,
      "learning_rate": 0.0012,
      "loss": 2.7300448417663574,
      "step": 1314
    },
    {
      "epoch": 0.10670261258225289,
      "grad_norm": 0.02418391965329647,
      "learning_rate": 0.0012,
      "loss": 2.792816162109375,
      "step": 1316
    },
    {
      "epoch": 0.10686477460745387,
      "grad_norm": 0.02268834412097931,
      "learning_rate": 0.0012,
      "loss": 2.756596565246582,
      "step": 1318
    },
    {
      "epoch": 0.10702693663265488,
      "grad_norm": 0.04041912406682968,
      "learning_rate": 0.0012,
      "loss": 2.6576547622680664,
      "step": 1320
    },
    {
      "epoch": 0.10718909865785586,
      "grad_norm": 0.031375959515571594,
      "learning_rate": 0.0012,
      "loss": 2.7804160118103027,
      "step": 1322
    },
    {
      "epoch": 0.10735126068305685,
      "grad_norm": 0.03847081959247589,
      "learning_rate": 0.0012,
      "loss": 2.697819709777832,
      "step": 1324
    },
    {
      "epoch": 0.10751342270825785,
      "grad_norm": 0.0689343586564064,
      "learning_rate": 0.0012,
      "loss": 2.7994790077209473,
      "step": 1326
    },
    {
      "epoch": 0.10767558473345884,
      "grad_norm": 0.0803372859954834,
      "learning_rate": 0.0012,
      "loss": 2.748582363128662,
      "step": 1328
    },
    {
      "epoch": 0.10783774675865983,
      "grad_norm": 0.047531530261039734,
      "learning_rate": 0.0012,
      "loss": 2.770766496658325,
      "step": 1330
    },
    {
      "epoch": 0.10799990878386083,
      "grad_norm": 0.046813998371362686,
      "learning_rate": 0.0012,
      "loss": 2.7530710697174072,
      "step": 1332
    },
    {
      "epoch": 0.10816207080906182,
      "grad_norm": 0.08986669033765793,
      "learning_rate": 0.0012,
      "loss": 2.8073997497558594,
      "step": 1334
    },
    {
      "epoch": 0.1083242328342628,
      "grad_norm": 0.06169890612363815,
      "learning_rate": 0.0012,
      "loss": 2.725494384765625,
      "step": 1336
    },
    {
      "epoch": 0.1084863948594638,
      "grad_norm": 0.03196362778544426,
      "learning_rate": 0.0012,
      "loss": 2.741255283355713,
      "step": 1338
    },
    {
      "epoch": 0.1086485568846648,
      "grad_norm": 0.07789518684148788,
      "learning_rate": 0.0012,
      "loss": 2.762739658355713,
      "step": 1340
    },
    {
      "epoch": 0.10881071890986578,
      "grad_norm": 0.060307521373033524,
      "learning_rate": 0.0012,
      "loss": 2.788882255554199,
      "step": 1342
    },
    {
      "epoch": 0.10897288093506678,
      "grad_norm": 0.027100738137960434,
      "learning_rate": 0.0012,
      "loss": 2.7174854278564453,
      "step": 1344
    },
    {
      "epoch": 0.10913504296026777,
      "grad_norm": 0.040277149528265,
      "learning_rate": 0.0012,
      "loss": 2.7594690322875977,
      "step": 1346
    },
    {
      "epoch": 0.10929720498546876,
      "grad_norm": 0.024659577757120132,
      "learning_rate": 0.0012,
      "loss": 2.7152037620544434,
      "step": 1348
    },
    {
      "epoch": 0.10945936701066976,
      "grad_norm": 0.02328571304678917,
      "learning_rate": 0.0012,
      "loss": 2.7199926376342773,
      "step": 1350
    },
    {
      "epoch": 0.10962152903587075,
      "grad_norm": 0.03263581544160843,
      "learning_rate": 0.0012,
      "loss": 2.80489182472229,
      "step": 1352
    },
    {
      "epoch": 0.10978369106107173,
      "grad_norm": 0.03987548500299454,
      "learning_rate": 0.0012,
      "loss": 2.761406660079956,
      "step": 1354
    },
    {
      "epoch": 0.10994585308627274,
      "grad_norm": 0.03113541007041931,
      "learning_rate": 0.0012,
      "loss": 2.692812442779541,
      "step": 1356
    },
    {
      "epoch": 0.11010801511147372,
      "grad_norm": 0.043459292501211166,
      "learning_rate": 0.0012,
      "loss": 2.7689356803894043,
      "step": 1358
    },
    {
      "epoch": 0.11027017713667471,
      "grad_norm": 0.025332557037472725,
      "learning_rate": 0.0012,
      "loss": 2.775092840194702,
      "step": 1360
    },
    {
      "epoch": 0.11043233916187571,
      "grad_norm": 0.03201674297451973,
      "learning_rate": 0.0012,
      "loss": 2.7174272537231445,
      "step": 1362
    },
    {
      "epoch": 0.1105945011870767,
      "grad_norm": 0.031124718487262726,
      "learning_rate": 0.0012,
      "loss": 2.772355556488037,
      "step": 1364
    },
    {
      "epoch": 0.11075666321227769,
      "grad_norm": 0.05714285001158714,
      "learning_rate": 0.0012,
      "loss": 2.7570972442626953,
      "step": 1366
    },
    {
      "epoch": 0.11091882523747869,
      "grad_norm": 0.04502076283097267,
      "learning_rate": 0.0012,
      "loss": 2.70524263381958,
      "step": 1368
    },
    {
      "epoch": 0.11108098726267968,
      "grad_norm": 0.028944728896021843,
      "learning_rate": 0.0012,
      "loss": 2.7358150482177734,
      "step": 1370
    },
    {
      "epoch": 0.11124314928788066,
      "grad_norm": 0.06490097939968109,
      "learning_rate": 0.0012,
      "loss": 2.674391746520996,
      "step": 1372
    },
    {
      "epoch": 0.11140531131308166,
      "grad_norm": 0.047356486320495605,
      "learning_rate": 0.0012,
      "loss": 2.6668035984039307,
      "step": 1374
    },
    {
      "epoch": 0.11156747333828265,
      "grad_norm": 0.0337027944624424,
      "learning_rate": 0.0012,
      "loss": 2.7850239276885986,
      "step": 1376
    },
    {
      "epoch": 0.11172963536348364,
      "grad_norm": 0.025771580636501312,
      "learning_rate": 0.0012,
      "loss": 2.741304397583008,
      "step": 1378
    },
    {
      "epoch": 0.11189179738868464,
      "grad_norm": 0.04813729599118233,
      "learning_rate": 0.0012,
      "loss": 2.753368377685547,
      "step": 1380
    },
    {
      "epoch": 0.11205395941388563,
      "grad_norm": 0.023975279182195663,
      "learning_rate": 0.0012,
      "loss": 2.8363144397735596,
      "step": 1382
    },
    {
      "epoch": 0.11221612143908662,
      "grad_norm": 0.025591908022761345,
      "learning_rate": 0.0012,
      "loss": 2.7627170085906982,
      "step": 1384
    },
    {
      "epoch": 0.11237828346428762,
      "grad_norm": 0.024834509938955307,
      "learning_rate": 0.0012,
      "loss": 2.723206043243408,
      "step": 1386
    },
    {
      "epoch": 0.1125404454894886,
      "grad_norm": 0.02998102642595768,
      "learning_rate": 0.0012,
      "loss": 2.771827220916748,
      "step": 1388
    },
    {
      "epoch": 0.11270260751468959,
      "grad_norm": 0.03927347809076309,
      "learning_rate": 0.0012,
      "loss": 2.730198860168457,
      "step": 1390
    },
    {
      "epoch": 0.1128647695398906,
      "grad_norm": 0.021769296377897263,
      "learning_rate": 0.0012,
      "loss": 2.7649168968200684,
      "step": 1392
    },
    {
      "epoch": 0.11302693156509158,
      "grad_norm": 0.025123098865151405,
      "learning_rate": 0.0012,
      "loss": 2.7908968925476074,
      "step": 1394
    },
    {
      "epoch": 0.11318909359029257,
      "grad_norm": 0.025065166875720024,
      "learning_rate": 0.0012,
      "loss": 2.765408515930176,
      "step": 1396
    },
    {
      "epoch": 0.11335125561549357,
      "grad_norm": 0.02583008073270321,
      "learning_rate": 0.0012,
      "loss": 2.7143218517303467,
      "step": 1398
    },
    {
      "epoch": 0.11351341764069456,
      "grad_norm": 0.028857912868261337,
      "learning_rate": 0.0012,
      "loss": 2.7834458351135254,
      "step": 1400
    },
    {
      "epoch": 0.11367557966589555,
      "grad_norm": 0.034672535955905914,
      "learning_rate": 0.0012,
      "loss": 2.8016281127929688,
      "step": 1402
    },
    {
      "epoch": 0.11383774169109655,
      "grad_norm": 0.031050221994519234,
      "learning_rate": 0.0012,
      "loss": 2.74448823928833,
      "step": 1404
    },
    {
      "epoch": 0.11399990371629753,
      "grad_norm": 0.030553316697478294,
      "learning_rate": 0.0012,
      "loss": 2.7857394218444824,
      "step": 1406
    },
    {
      "epoch": 0.11416206574149852,
      "grad_norm": 0.02364938333630562,
      "learning_rate": 0.0012,
      "loss": 2.7744295597076416,
      "step": 1408
    },
    {
      "epoch": 0.11432422776669952,
      "grad_norm": 0.024436116218566895,
      "learning_rate": 0.0012,
      "loss": 2.781743049621582,
      "step": 1410
    },
    {
      "epoch": 0.11448638979190051,
      "grad_norm": 0.02348971553146839,
      "learning_rate": 0.0012,
      "loss": 2.805253744125366,
      "step": 1412
    },
    {
      "epoch": 0.1146485518171015,
      "grad_norm": 0.05129342898726463,
      "learning_rate": 0.0012,
      "loss": 2.7088065147399902,
      "step": 1414
    },
    {
      "epoch": 0.1148107138423025,
      "grad_norm": 0.05223388224840164,
      "learning_rate": 0.0012,
      "loss": 2.743961811065674,
      "step": 1416
    },
    {
      "epoch": 0.11497287586750349,
      "grad_norm": 0.05396222695708275,
      "learning_rate": 0.0012,
      "loss": 2.7854340076446533,
      "step": 1418
    },
    {
      "epoch": 0.11513503789270449,
      "grad_norm": 0.05315963178873062,
      "learning_rate": 0.0012,
      "loss": 2.7788805961608887,
      "step": 1420
    },
    {
      "epoch": 0.11529719991790548,
      "grad_norm": 0.028025532141327858,
      "learning_rate": 0.0012,
      "loss": 2.788364887237549,
      "step": 1422
    },
    {
      "epoch": 0.11545936194310646,
      "grad_norm": 0.06860300898551941,
      "learning_rate": 0.0012,
      "loss": 2.7135491371154785,
      "step": 1424
    },
    {
      "epoch": 0.11562152396830747,
      "grad_norm": 0.05767658352851868,
      "learning_rate": 0.0012,
      "loss": 2.719625949859619,
      "step": 1426
    },
    {
      "epoch": 0.11578368599350845,
      "grad_norm": 0.027722377330064774,
      "learning_rate": 0.0012,
      "loss": 2.7531771659851074,
      "step": 1428
    },
    {
      "epoch": 0.11594584801870944,
      "grad_norm": 0.08026549220085144,
      "learning_rate": 0.0012,
      "loss": 2.7163596153259277,
      "step": 1430
    },
    {
      "epoch": 0.11610801004391044,
      "grad_norm": 0.030180811882019043,
      "learning_rate": 0.0012,
      "loss": 2.7063424587249756,
      "step": 1432
    },
    {
      "epoch": 0.11627017206911143,
      "grad_norm": 0.032185789197683334,
      "learning_rate": 0.0012,
      "loss": 2.712311029434204,
      "step": 1434
    },
    {
      "epoch": 0.11643233409431242,
      "grad_norm": 0.06179207190871239,
      "learning_rate": 0.0012,
      "loss": 2.8045907020568848,
      "step": 1436
    },
    {
      "epoch": 0.11659449611951342,
      "grad_norm": 0.04108645021915436,
      "learning_rate": 0.0012,
      "loss": 2.683917999267578,
      "step": 1438
    },
    {
      "epoch": 0.1167566581447144,
      "grad_norm": 0.03703099116683006,
      "learning_rate": 0.0012,
      "loss": 2.756218194961548,
      "step": 1440
    },
    {
      "epoch": 0.1169188201699154,
      "grad_norm": 0.03026178479194641,
      "learning_rate": 0.0012,
      "loss": 2.751915454864502,
      "step": 1442
    },
    {
      "epoch": 0.1170809821951164,
      "grad_norm": 0.04384521022439003,
      "learning_rate": 0.0012,
      "loss": 2.7216262817382812,
      "step": 1444
    },
    {
      "epoch": 0.11724314422031738,
      "grad_norm": 0.04245149716734886,
      "learning_rate": 0.0012,
      "loss": 2.7403202056884766,
      "step": 1446
    },
    {
      "epoch": 0.11740530624551837,
      "grad_norm": 0.03110373020172119,
      "learning_rate": 0.0012,
      "loss": 2.689162254333496,
      "step": 1448
    },
    {
      "epoch": 0.11756746827071937,
      "grad_norm": 0.02902667224407196,
      "learning_rate": 0.0012,
      "loss": 2.769796848297119,
      "step": 1450
    },
    {
      "epoch": 0.11772963029592036,
      "grad_norm": 0.031385935842990875,
      "learning_rate": 0.0012,
      "loss": 2.7748255729675293,
      "step": 1452
    },
    {
      "epoch": 0.11789179232112135,
      "grad_norm": 0.05582403019070625,
      "learning_rate": 0.0012,
      "loss": 2.7414960861206055,
      "step": 1454
    },
    {
      "epoch": 0.11805395434632235,
      "grad_norm": 0.033284008502960205,
      "learning_rate": 0.0012,
      "loss": 2.6856753826141357,
      "step": 1456
    },
    {
      "epoch": 0.11821611637152334,
      "grad_norm": 0.027904706075787544,
      "learning_rate": 0.0012,
      "loss": 2.7751550674438477,
      "step": 1458
    },
    {
      "epoch": 0.11837827839672432,
      "grad_norm": 0.023098643869161606,
      "learning_rate": 0.0012,
      "loss": 2.735558032989502,
      "step": 1460
    },
    {
      "epoch": 0.11854044042192532,
      "grad_norm": 0.045249342918395996,
      "learning_rate": 0.0012,
      "loss": 2.775961399078369,
      "step": 1462
    },
    {
      "epoch": 0.11870260244712631,
      "grad_norm": 0.05580924451351166,
      "learning_rate": 0.0012,
      "loss": 2.6839632987976074,
      "step": 1464
    },
    {
      "epoch": 0.1188647644723273,
      "grad_norm": 0.05070424824953079,
      "learning_rate": 0.0012,
      "loss": 2.6689765453338623,
      "step": 1466
    },
    {
      "epoch": 0.1190269264975283,
      "grad_norm": 0.029027247801423073,
      "learning_rate": 0.0012,
      "loss": 2.7070388793945312,
      "step": 1468
    },
    {
      "epoch": 0.11918908852272929,
      "grad_norm": 0.03474850952625275,
      "learning_rate": 0.0012,
      "loss": 2.753340244293213,
      "step": 1470
    },
    {
      "epoch": 0.11935125054793028,
      "grad_norm": 0.07401382178068161,
      "learning_rate": 0.0012,
      "loss": 2.744272232055664,
      "step": 1472
    },
    {
      "epoch": 0.11951341257313128,
      "grad_norm": 0.051949936896562576,
      "learning_rate": 0.0012,
      "loss": 2.7471656799316406,
      "step": 1474
    },
    {
      "epoch": 0.11967557459833227,
      "grad_norm": 0.040167827159166336,
      "learning_rate": 0.0012,
      "loss": 2.726132392883301,
      "step": 1476
    },
    {
      "epoch": 0.11983773662353325,
      "grad_norm": 0.04855886474251747,
      "learning_rate": 0.0012,
      "loss": 2.7329037189483643,
      "step": 1478
    },
    {
      "epoch": 0.11999989864873425,
      "grad_norm": 0.03901258483529091,
      "learning_rate": 0.0012,
      "loss": 2.700920581817627,
      "step": 1480
    },
    {
      "epoch": 0.12016206067393524,
      "grad_norm": 0.026086660102009773,
      "learning_rate": 0.0012,
      "loss": 2.7010412216186523,
      "step": 1482
    },
    {
      "epoch": 0.12032422269913623,
      "grad_norm": 0.025722723454236984,
      "learning_rate": 0.0012,
      "loss": 2.7871813774108887,
      "step": 1484
    },
    {
      "epoch": 0.12048638472433723,
      "grad_norm": 0.036070317029953,
      "learning_rate": 0.0012,
      "loss": 2.7261314392089844,
      "step": 1486
    },
    {
      "epoch": 0.12064854674953822,
      "grad_norm": 0.03768303245306015,
      "learning_rate": 0.0012,
      "loss": 2.6897776126861572,
      "step": 1488
    },
    {
      "epoch": 0.1208107087747392,
      "grad_norm": 0.02736426517367363,
      "learning_rate": 0.0012,
      "loss": 2.8370108604431152,
      "step": 1490
    },
    {
      "epoch": 0.12097287079994021,
      "grad_norm": 0.024671869352459908,
      "learning_rate": 0.0012,
      "loss": 2.7796788215637207,
      "step": 1492
    },
    {
      "epoch": 0.1211350328251412,
      "grad_norm": 0.03255629912018776,
      "learning_rate": 0.0012,
      "loss": 2.7805304527282715,
      "step": 1494
    },
    {
      "epoch": 0.12129719485034218,
      "grad_norm": 0.023984601721167564,
      "learning_rate": 0.0012,
      "loss": 2.666665554046631,
      "step": 1496
    },
    {
      "epoch": 0.12145935687554318,
      "grad_norm": 0.022576548159122467,
      "learning_rate": 0.0012,
      "loss": 2.7616045475006104,
      "step": 1498
    },
    {
      "epoch": 0.12162151890074417,
      "grad_norm": 0.02151445858180523,
      "learning_rate": 0.0012,
      "loss": 2.7687418460845947,
      "step": 1500
    },
    {
      "epoch": 0.12178368092594516,
      "grad_norm": 0.07631471008062363,
      "learning_rate": 0.0012,
      "loss": 2.7569355964660645,
      "step": 1502
    },
    {
      "epoch": 0.12194584295114616,
      "grad_norm": 0.06072380393743515,
      "learning_rate": 0.0012,
      "loss": 2.761544704437256,
      "step": 1504
    },
    {
      "epoch": 0.12210800497634715,
      "grad_norm": 0.027703238651156425,
      "learning_rate": 0.0012,
      "loss": 2.7307305335998535,
      "step": 1506
    },
    {
      "epoch": 0.12227016700154814,
      "grad_norm": 0.05135719105601311,
      "learning_rate": 0.0012,
      "loss": 2.7685141563415527,
      "step": 1508
    },
    {
      "epoch": 0.12243232902674914,
      "grad_norm": 0.054383061826229095,
      "learning_rate": 0.0012,
      "loss": 2.7436463832855225,
      "step": 1510
    },
    {
      "epoch": 0.12259449105195012,
      "grad_norm": 0.031467001885175705,
      "learning_rate": 0.0012,
      "loss": 2.8094046115875244,
      "step": 1512
    },
    {
      "epoch": 0.12275665307715111,
      "grad_norm": 0.028357679024338722,
      "learning_rate": 0.0012,
      "loss": 2.6915926933288574,
      "step": 1514
    },
    {
      "epoch": 0.12291881510235211,
      "grad_norm": 0.04917743057012558,
      "learning_rate": 0.0012,
      "loss": 2.726445198059082,
      "step": 1516
    },
    {
      "epoch": 0.1230809771275531,
      "grad_norm": 0.027511343359947205,
      "learning_rate": 0.0012,
      "loss": 2.740560531616211,
      "step": 1518
    },
    {
      "epoch": 0.12324313915275409,
      "grad_norm": 0.03349863365292549,
      "learning_rate": 0.0012,
      "loss": 2.793884754180908,
      "step": 1520
    },
    {
      "epoch": 0.12340530117795509,
      "grad_norm": 0.03031662106513977,
      "learning_rate": 0.0012,
      "loss": 2.7407069206237793,
      "step": 1522
    },
    {
      "epoch": 0.12356746320315608,
      "grad_norm": 0.02800528146326542,
      "learning_rate": 0.0012,
      "loss": 2.792966604232788,
      "step": 1524
    },
    {
      "epoch": 0.12372962522835707,
      "grad_norm": 0.024668985977768898,
      "learning_rate": 0.0012,
      "loss": 2.7444100379943848,
      "step": 1526
    },
    {
      "epoch": 0.12389178725355807,
      "grad_norm": 0.0364600345492363,
      "learning_rate": 0.0012,
      "loss": 2.733487129211426,
      "step": 1528
    },
    {
      "epoch": 0.12405394927875905,
      "grad_norm": 0.02268456481397152,
      "learning_rate": 0.0012,
      "loss": 2.774501323699951,
      "step": 1530
    },
    {
      "epoch": 0.12421611130396004,
      "grad_norm": 0.0248273815959692,
      "learning_rate": 0.0012,
      "loss": 2.658921003341675,
      "step": 1532
    },
    {
      "epoch": 0.12437827332916104,
      "grad_norm": 0.02220734767615795,
      "learning_rate": 0.0012,
      "loss": 2.820063591003418,
      "step": 1534
    },
    {
      "epoch": 0.12454043535436203,
      "grad_norm": 0.024972498416900635,
      "learning_rate": 0.0012,
      "loss": 2.735910415649414,
      "step": 1536
    },
    {
      "epoch": 0.12470259737956302,
      "grad_norm": 0.03247813135385513,
      "learning_rate": 0.0012,
      "loss": 2.7745277881622314,
      "step": 1538
    },
    {
      "epoch": 0.12486475940476402,
      "grad_norm": 0.03741436451673508,
      "learning_rate": 0.0012,
      "loss": 2.7830405235290527,
      "step": 1540
    },
    {
      "epoch": 0.12502692142996502,
      "grad_norm": 0.02500065602362156,
      "learning_rate": 0.0012,
      "loss": 2.7310128211975098,
      "step": 1542
    },
    {
      "epoch": 0.125189083455166,
      "grad_norm": 0.024196993559598923,
      "learning_rate": 0.0012,
      "loss": 2.7297792434692383,
      "step": 1544
    },
    {
      "epoch": 0.125351245480367,
      "grad_norm": 0.022958973422646523,
      "learning_rate": 0.0012,
      "loss": 2.696354389190674,
      "step": 1546
    },
    {
      "epoch": 0.125513407505568,
      "grad_norm": 0.034133732318878174,
      "learning_rate": 0.0012,
      "loss": 2.7335805892944336,
      "step": 1548
    },
    {
      "epoch": 0.12567556953076897,
      "grad_norm": 0.07697222381830215,
      "learning_rate": 0.0012,
      "loss": 2.747321128845215,
      "step": 1550
    },
    {
      "epoch": 0.12583773155596997,
      "grad_norm": 0.05702817812561989,
      "learning_rate": 0.0012,
      "loss": 2.6886138916015625,
      "step": 1552
    },
    {
      "epoch": 0.12599989358117097,
      "grad_norm": 0.025946229696273804,
      "learning_rate": 0.0012,
      "loss": 2.7143356800079346,
      "step": 1554
    },
    {
      "epoch": 0.12616205560637195,
      "grad_norm": 0.025382546707987785,
      "learning_rate": 0.0012,
      "loss": 2.7276151180267334,
      "step": 1556
    },
    {
      "epoch": 0.12632421763157295,
      "grad_norm": 0.0493818037211895,
      "learning_rate": 0.0012,
      "loss": 2.700432062149048,
      "step": 1558
    },
    {
      "epoch": 0.12648637965677395,
      "grad_norm": 0.02371765859425068,
      "learning_rate": 0.0012,
      "loss": 2.7282190322875977,
      "step": 1560
    },
    {
      "epoch": 0.12664854168197492,
      "grad_norm": 0.03169611096382141,
      "learning_rate": 0.0012,
      "loss": 2.713120222091675,
      "step": 1562
    },
    {
      "epoch": 0.12681070370717593,
      "grad_norm": 0.04718024656176567,
      "learning_rate": 0.0012,
      "loss": 2.791527509689331,
      "step": 1564
    },
    {
      "epoch": 0.12697286573237693,
      "grad_norm": 0.053549375385046005,
      "learning_rate": 0.0012,
      "loss": 2.782393455505371,
      "step": 1566
    },
    {
      "epoch": 0.1271350277575779,
      "grad_norm": 0.024598341435194016,
      "learning_rate": 0.0012,
      "loss": 2.7684240341186523,
      "step": 1568
    },
    {
      "epoch": 0.1272971897827789,
      "grad_norm": 0.023727748543024063,
      "learning_rate": 0.0012,
      "loss": 2.765033721923828,
      "step": 1570
    },
    {
      "epoch": 0.1274593518079799,
      "grad_norm": 0.03716399520635605,
      "learning_rate": 0.0012,
      "loss": 2.7537903785705566,
      "step": 1572
    },
    {
      "epoch": 0.12762151383318088,
      "grad_norm": 0.033523909747600555,
      "learning_rate": 0.0012,
      "loss": 2.7535414695739746,
      "step": 1574
    },
    {
      "epoch": 0.12778367585838188,
      "grad_norm": 0.032600536942481995,
      "learning_rate": 0.0012,
      "loss": 2.775965929031372,
      "step": 1576
    },
    {
      "epoch": 0.12794583788358288,
      "grad_norm": 0.0277189202606678,
      "learning_rate": 0.0012,
      "loss": 2.7484540939331055,
      "step": 1578
    },
    {
      "epoch": 0.12810799990878385,
      "grad_norm": 0.02335728518664837,
      "learning_rate": 0.0012,
      "loss": 2.7322487831115723,
      "step": 1580
    },
    {
      "epoch": 0.12827016193398486,
      "grad_norm": 0.03295409306883812,
      "learning_rate": 0.0012,
      "loss": 2.773493766784668,
      "step": 1582
    },
    {
      "epoch": 0.12843232395918586,
      "grad_norm": 0.03093794360756874,
      "learning_rate": 0.0012,
      "loss": 2.7198705673217773,
      "step": 1584
    },
    {
      "epoch": 0.12859448598438683,
      "grad_norm": 0.023317910730838776,
      "learning_rate": 0.0012,
      "loss": 2.7236814498901367,
      "step": 1586
    },
    {
      "epoch": 0.12875664800958783,
      "grad_norm": 0.026997849345207214,
      "learning_rate": 0.0012,
      "loss": 2.7131423950195312,
      "step": 1588
    },
    {
      "epoch": 0.12891881003478883,
      "grad_norm": 0.04018936678767204,
      "learning_rate": 0.0012,
      "loss": 2.702033519744873,
      "step": 1590
    },
    {
      "epoch": 0.1290809720599898,
      "grad_norm": 0.027253182604908943,
      "learning_rate": 0.0012,
      "loss": 2.7418556213378906,
      "step": 1592
    },
    {
      "epoch": 0.1292431340851908,
      "grad_norm": 0.04193807765841484,
      "learning_rate": 0.0012,
      "loss": 2.7319347858428955,
      "step": 1594
    },
    {
      "epoch": 0.1294052961103918,
      "grad_norm": 0.05556900426745415,
      "learning_rate": 0.0012,
      "loss": 2.7937002182006836,
      "step": 1596
    },
    {
      "epoch": 0.12956745813559278,
      "grad_norm": 0.033654216676950455,
      "learning_rate": 0.0012,
      "loss": 2.6745352745056152,
      "step": 1598
    },
    {
      "epoch": 0.12972962016079378,
      "grad_norm": 0.05059616640210152,
      "learning_rate": 0.0012,
      "loss": 2.7382652759552,
      "step": 1600
    },
    {
      "epoch": 0.1298917821859948,
      "grad_norm": 0.021628092974424362,
      "learning_rate": 0.0012,
      "loss": 2.746356964111328,
      "step": 1602
    },
    {
      "epoch": 0.13005394421119576,
      "grad_norm": 0.02944726124405861,
      "learning_rate": 0.0012,
      "loss": 2.76786208152771,
      "step": 1604
    },
    {
      "epoch": 0.13021610623639676,
      "grad_norm": 0.03934677317738533,
      "learning_rate": 0.0012,
      "loss": 2.6974940299987793,
      "step": 1606
    },
    {
      "epoch": 0.13037826826159776,
      "grad_norm": 0.05215495079755783,
      "learning_rate": 0.0012,
      "loss": 2.766866683959961,
      "step": 1608
    },
    {
      "epoch": 0.13054043028679874,
      "grad_norm": 0.0765022411942482,
      "learning_rate": 0.0012,
      "loss": 2.74731183052063,
      "step": 1610
    },
    {
      "epoch": 0.13070259231199974,
      "grad_norm": 0.036953043192625046,
      "learning_rate": 0.0012,
      "loss": 2.7989957332611084,
      "step": 1612
    },
    {
      "epoch": 0.13086475433720074,
      "grad_norm": 0.04233867675065994,
      "learning_rate": 0.0012,
      "loss": 2.7232255935668945,
      "step": 1614
    },
    {
      "epoch": 0.1310269163624017,
      "grad_norm": 0.05487754940986633,
      "learning_rate": 0.0012,
      "loss": 2.7671656608581543,
      "step": 1616
    },
    {
      "epoch": 0.13118907838760271,
      "grad_norm": 0.04921456798911095,
      "learning_rate": 0.0012,
      "loss": 2.7051339149475098,
      "step": 1618
    },
    {
      "epoch": 0.13135124041280372,
      "grad_norm": 0.05579725652933121,
      "learning_rate": 0.0012,
      "loss": 2.742370128631592,
      "step": 1620
    },
    {
      "epoch": 0.1315134024380047,
      "grad_norm": 0.07181261479854584,
      "learning_rate": 0.0012,
      "loss": 2.735299825668335,
      "step": 1622
    },
    {
      "epoch": 0.1316755644632057,
      "grad_norm": 0.023596692830324173,
      "learning_rate": 0.0012,
      "loss": 2.7718255519866943,
      "step": 1624
    },
    {
      "epoch": 0.1318377264884067,
      "grad_norm": 0.03511602431535721,
      "learning_rate": 0.0012,
      "loss": 2.6739869117736816,
      "step": 1626
    },
    {
      "epoch": 0.13199988851360767,
      "grad_norm": 0.03642137348651886,
      "learning_rate": 0.0012,
      "loss": 2.752081871032715,
      "step": 1628
    },
    {
      "epoch": 0.13216205053880867,
      "grad_norm": 0.02859467640519142,
      "learning_rate": 0.0012,
      "loss": 2.7186946868896484,
      "step": 1630
    },
    {
      "epoch": 0.13232421256400967,
      "grad_norm": 0.045843567699193954,
      "learning_rate": 0.0012,
      "loss": 2.7606260776519775,
      "step": 1632
    },
    {
      "epoch": 0.13248637458921064,
      "grad_norm": 0.03693564608693123,
      "learning_rate": 0.0012,
      "loss": 2.7284626960754395,
      "step": 1634
    },
    {
      "epoch": 0.13264853661441164,
      "grad_norm": 0.03791423887014389,
      "learning_rate": 0.0012,
      "loss": 2.8201940059661865,
      "step": 1636
    },
    {
      "epoch": 0.13281069863961265,
      "grad_norm": 0.039293739944696426,
      "learning_rate": 0.0012,
      "loss": 2.7198944091796875,
      "step": 1638
    },
    {
      "epoch": 0.13297286066481362,
      "grad_norm": 0.03036259301006794,
      "learning_rate": 0.0012,
      "loss": 2.788905620574951,
      "step": 1640
    },
    {
      "epoch": 0.13313502269001462,
      "grad_norm": 0.03425733745098114,
      "learning_rate": 0.0012,
      "loss": 2.771714687347412,
      "step": 1642
    },
    {
      "epoch": 0.13329718471521562,
      "grad_norm": 0.023477928712964058,
      "learning_rate": 0.0012,
      "loss": 2.750119209289551,
      "step": 1644
    },
    {
      "epoch": 0.1334593467404166,
      "grad_norm": 0.03076671063899994,
      "learning_rate": 0.0012,
      "loss": 2.767557144165039,
      "step": 1646
    },
    {
      "epoch": 0.1336215087656176,
      "grad_norm": 0.03047414869070053,
      "learning_rate": 0.0012,
      "loss": 2.7467246055603027,
      "step": 1648
    },
    {
      "epoch": 0.1337836707908186,
      "grad_norm": 0.03596234321594238,
      "learning_rate": 0.0012,
      "loss": 2.6628830432891846,
      "step": 1650
    },
    {
      "epoch": 0.13394583281601957,
      "grad_norm": 0.029109004884958267,
      "learning_rate": 0.0012,
      "loss": 2.6838743686676025,
      "step": 1652
    },
    {
      "epoch": 0.13410799484122057,
      "grad_norm": 0.0274164117872715,
      "learning_rate": 0.0012,
      "loss": 2.711520195007324,
      "step": 1654
    },
    {
      "epoch": 0.13427015686642157,
      "grad_norm": 0.044820040464401245,
      "learning_rate": 0.0012,
      "loss": 2.738852024078369,
      "step": 1656
    },
    {
      "epoch": 0.13443231889162255,
      "grad_norm": 0.025214694440364838,
      "learning_rate": 0.0012,
      "loss": 2.715020179748535,
      "step": 1658
    },
    {
      "epoch": 0.13459448091682355,
      "grad_norm": 0.0463993065059185,
      "learning_rate": 0.0012,
      "loss": 2.65008282661438,
      "step": 1660
    },
    {
      "epoch": 0.13475664294202455,
      "grad_norm": 0.06422653794288635,
      "learning_rate": 0.0012,
      "loss": 2.678666114807129,
      "step": 1662
    },
    {
      "epoch": 0.13491880496722553,
      "grad_norm": 0.0455496720969677,
      "learning_rate": 0.0012,
      "loss": 2.7705585956573486,
      "step": 1664
    },
    {
      "epoch": 0.13508096699242653,
      "grad_norm": 0.02874034084379673,
      "learning_rate": 0.0012,
      "loss": 2.7285633087158203,
      "step": 1666
    },
    {
      "epoch": 0.13524312901762753,
      "grad_norm": 0.021711448207497597,
      "learning_rate": 0.0012,
      "loss": 2.7460708618164062,
      "step": 1668
    },
    {
      "epoch": 0.1354052910428285,
      "grad_norm": 0.038542356342077255,
      "learning_rate": 0.0012,
      "loss": 2.7167763710021973,
      "step": 1670
    },
    {
      "epoch": 0.1355674530680295,
      "grad_norm": 0.025961726903915405,
      "learning_rate": 0.0012,
      "loss": 2.6904358863830566,
      "step": 1672
    },
    {
      "epoch": 0.1357296150932305,
      "grad_norm": 0.04002934321761131,
      "learning_rate": 0.0012,
      "loss": 2.6990389823913574,
      "step": 1674
    },
    {
      "epoch": 0.13589177711843148,
      "grad_norm": 0.06684131920337677,
      "learning_rate": 0.0012,
      "loss": 2.7391371726989746,
      "step": 1676
    },
    {
      "epoch": 0.13605393914363248,
      "grad_norm": 0.028230158612132072,
      "learning_rate": 0.0012,
      "loss": 2.6983964443206787,
      "step": 1678
    },
    {
      "epoch": 0.13621610116883348,
      "grad_norm": 0.03164998069405556,
      "learning_rate": 0.0012,
      "loss": 2.753408432006836,
      "step": 1680
    },
    {
      "epoch": 0.13637826319403445,
      "grad_norm": 0.02242370881140232,
      "learning_rate": 0.0012,
      "loss": 2.6380679607391357,
      "step": 1682
    },
    {
      "epoch": 0.13654042521923546,
      "grad_norm": 0.03942084684967995,
      "learning_rate": 0.0012,
      "loss": 2.712604522705078,
      "step": 1684
    },
    {
      "epoch": 0.13670258724443646,
      "grad_norm": 0.04492565616965294,
      "learning_rate": 0.0012,
      "loss": 2.802072048187256,
      "step": 1686
    },
    {
      "epoch": 0.13686474926963743,
      "grad_norm": 0.03955453634262085,
      "learning_rate": 0.0012,
      "loss": 2.709958076477051,
      "step": 1688
    },
    {
      "epoch": 0.13702691129483843,
      "grad_norm": 0.04045148193836212,
      "learning_rate": 0.0012,
      "loss": 2.767653465270996,
      "step": 1690
    },
    {
      "epoch": 0.13718907332003943,
      "grad_norm": 0.035839661955833435,
      "learning_rate": 0.0012,
      "loss": 2.759517192840576,
      "step": 1692
    },
    {
      "epoch": 0.1373512353452404,
      "grad_norm": 0.028093751519918442,
      "learning_rate": 0.0012,
      "loss": 2.731966495513916,
      "step": 1694
    },
    {
      "epoch": 0.1375133973704414,
      "grad_norm": 0.03287062421441078,
      "learning_rate": 0.0012,
      "loss": 2.709146022796631,
      "step": 1696
    },
    {
      "epoch": 0.1376755593956424,
      "grad_norm": 0.026375506073236465,
      "learning_rate": 0.0012,
      "loss": 2.727025032043457,
      "step": 1698
    },
    {
      "epoch": 0.13783772142084338,
      "grad_norm": 0.046890366822481155,
      "learning_rate": 0.0012,
      "loss": 2.7058522701263428,
      "step": 1700
    },
    {
      "epoch": 0.13799988344604439,
      "grad_norm": 0.03419497609138489,
      "learning_rate": 0.0012,
      "loss": 2.6797499656677246,
      "step": 1702
    },
    {
      "epoch": 0.1381620454712454,
      "grad_norm": 0.055954769253730774,
      "learning_rate": 0.0012,
      "loss": 2.7431774139404297,
      "step": 1704
    },
    {
      "epoch": 0.13832420749644636,
      "grad_norm": 0.0467267781496048,
      "learning_rate": 0.0012,
      "loss": 2.7600245475769043,
      "step": 1706
    },
    {
      "epoch": 0.13848636952164736,
      "grad_norm": 0.024708492681384087,
      "learning_rate": 0.0012,
      "loss": 2.6903624534606934,
      "step": 1708
    },
    {
      "epoch": 0.13864853154684836,
      "grad_norm": 0.032527364790439606,
      "learning_rate": 0.0012,
      "loss": 2.731632709503174,
      "step": 1710
    },
    {
      "epoch": 0.13881069357204934,
      "grad_norm": 0.025494959205389023,
      "learning_rate": 0.0012,
      "loss": 2.7771735191345215,
      "step": 1712
    },
    {
      "epoch": 0.13897285559725034,
      "grad_norm": 0.03230839967727661,
      "learning_rate": 0.0012,
      "loss": 2.7555370330810547,
      "step": 1714
    },
    {
      "epoch": 0.13913501762245134,
      "grad_norm": 0.035262394696474075,
      "learning_rate": 0.0012,
      "loss": 2.747642755508423,
      "step": 1716
    },
    {
      "epoch": 0.1392971796476523,
      "grad_norm": 0.0394895039498806,
      "learning_rate": 0.0012,
      "loss": 2.6813158988952637,
      "step": 1718
    },
    {
      "epoch": 0.13945934167285332,
      "grad_norm": 0.02594420313835144,
      "learning_rate": 0.0012,
      "loss": 2.7465977668762207,
      "step": 1720
    },
    {
      "epoch": 0.13962150369805432,
      "grad_norm": 0.033606551587581635,
      "learning_rate": 0.0012,
      "loss": 2.752027750015259,
      "step": 1722
    },
    {
      "epoch": 0.1397836657232553,
      "grad_norm": 0.02462681382894516,
      "learning_rate": 0.0012,
      "loss": 2.7558751106262207,
      "step": 1724
    },
    {
      "epoch": 0.1399458277484563,
      "grad_norm": 0.034549836069345474,
      "learning_rate": 0.0012,
      "loss": 2.7276694774627686,
      "step": 1726
    },
    {
      "epoch": 0.1401079897736573,
      "grad_norm": 0.030665112659335136,
      "learning_rate": 0.0012,
      "loss": 2.719968318939209,
      "step": 1728
    },
    {
      "epoch": 0.14027015179885827,
      "grad_norm": 0.03800811618566513,
      "learning_rate": 0.0012,
      "loss": 2.747292995452881,
      "step": 1730
    },
    {
      "epoch": 0.14043231382405927,
      "grad_norm": 0.0379587821662426,
      "learning_rate": 0.0012,
      "loss": 2.7321505546569824,
      "step": 1732
    },
    {
      "epoch": 0.14059447584926027,
      "grad_norm": 0.05312243476510048,
      "learning_rate": 0.0012,
      "loss": 2.6631054878234863,
      "step": 1734
    },
    {
      "epoch": 0.14075663787446124,
      "grad_norm": 0.06865030527114868,
      "learning_rate": 0.0012,
      "loss": 2.7574496269226074,
      "step": 1736
    },
    {
      "epoch": 0.14091879989966224,
      "grad_norm": 0.030548300594091415,
      "learning_rate": 0.0012,
      "loss": 2.807978391647339,
      "step": 1738
    },
    {
      "epoch": 0.14108096192486325,
      "grad_norm": 0.06683232635259628,
      "learning_rate": 0.0012,
      "loss": 2.6463308334350586,
      "step": 1740
    },
    {
      "epoch": 0.14124312395006422,
      "grad_norm": 0.05188342183828354,
      "learning_rate": 0.0012,
      "loss": 2.7388877868652344,
      "step": 1742
    },
    {
      "epoch": 0.14140528597526522,
      "grad_norm": 0.025564778596162796,
      "learning_rate": 0.0012,
      "loss": 2.7470004558563232,
      "step": 1744
    },
    {
      "epoch": 0.14156744800046622,
      "grad_norm": 0.06131884828209877,
      "learning_rate": 0.0012,
      "loss": 2.748053550720215,
      "step": 1746
    },
    {
      "epoch": 0.1417296100256672,
      "grad_norm": 0.06519073992967606,
      "learning_rate": 0.0012,
      "loss": 2.719069242477417,
      "step": 1748
    },
    {
      "epoch": 0.1418917720508682,
      "grad_norm": 0.03174464777112007,
      "learning_rate": 0.0012,
      "loss": 2.75553035736084,
      "step": 1750
    },
    {
      "epoch": 0.1420539340760692,
      "grad_norm": 0.04713595658540726,
      "learning_rate": 0.0012,
      "loss": 2.750699996948242,
      "step": 1752
    },
    {
      "epoch": 0.14221609610127017,
      "grad_norm": 0.05602327734231949,
      "learning_rate": 0.0012,
      "loss": 2.6722350120544434,
      "step": 1754
    },
    {
      "epoch": 0.14237825812647117,
      "grad_norm": 0.024334868416190147,
      "learning_rate": 0.0012,
      "loss": 2.7350223064422607,
      "step": 1756
    },
    {
      "epoch": 0.14254042015167218,
      "grad_norm": 0.052320439368486404,
      "learning_rate": 0.0012,
      "loss": 2.785900592803955,
      "step": 1758
    },
    {
      "epoch": 0.14270258217687315,
      "grad_norm": 0.0256134495139122,
      "learning_rate": 0.0012,
      "loss": 2.7505505084991455,
      "step": 1760
    },
    {
      "epoch": 0.14286474420207415,
      "grad_norm": 0.057191669940948486,
      "learning_rate": 0.0012,
      "loss": 2.676105499267578,
      "step": 1762
    },
    {
      "epoch": 0.14302690622727515,
      "grad_norm": 0.055449023842811584,
      "learning_rate": 0.0012,
      "loss": 2.737976551055908,
      "step": 1764
    },
    {
      "epoch": 0.14318906825247613,
      "grad_norm": 0.024520184844732285,
      "learning_rate": 0.0012,
      "loss": 2.726931095123291,
      "step": 1766
    },
    {
      "epoch": 0.14335123027767713,
      "grad_norm": 0.060250744223594666,
      "learning_rate": 0.0012,
      "loss": 2.802435874938965,
      "step": 1768
    },
    {
      "epoch": 0.14351339230287813,
      "grad_norm": 0.06962408870458603,
      "learning_rate": 0.0012,
      "loss": 2.7352097034454346,
      "step": 1770
    },
    {
      "epoch": 0.1436755543280791,
      "grad_norm": 0.026441358029842377,
      "learning_rate": 0.0012,
      "loss": 2.7960071563720703,
      "step": 1772
    },
    {
      "epoch": 0.1438377163532801,
      "grad_norm": 0.041460659354925156,
      "learning_rate": 0.0012,
      "loss": 2.7807655334472656,
      "step": 1774
    },
    {
      "epoch": 0.1439998783784811,
      "grad_norm": 0.061341721564531326,
      "learning_rate": 0.0012,
      "loss": 2.731433629989624,
      "step": 1776
    },
    {
      "epoch": 0.14416204040368208,
      "grad_norm": 0.025960300117731094,
      "learning_rate": 0.0012,
      "loss": 2.727919101715088,
      "step": 1778
    },
    {
      "epoch": 0.14432420242888308,
      "grad_norm": 0.05536384880542755,
      "learning_rate": 0.0012,
      "loss": 2.667982578277588,
      "step": 1780
    },
    {
      "epoch": 0.14448636445408408,
      "grad_norm": 0.04591980203986168,
      "learning_rate": 0.0012,
      "loss": 2.715188503265381,
      "step": 1782
    },
    {
      "epoch": 0.14464852647928506,
      "grad_norm": 0.033970821648836136,
      "learning_rate": 0.0012,
      "loss": 2.7242040634155273,
      "step": 1784
    },
    {
      "epoch": 0.14481068850448606,
      "grad_norm": 0.040980249643325806,
      "learning_rate": 0.0012,
      "loss": 2.69766902923584,
      "step": 1786
    },
    {
      "epoch": 0.14497285052968706,
      "grad_norm": 0.024872055277228355,
      "learning_rate": 0.0012,
      "loss": 2.7051262855529785,
      "step": 1788
    },
    {
      "epoch": 0.14513501255488803,
      "grad_norm": 0.04481930658221245,
      "learning_rate": 0.0012,
      "loss": 2.7436976432800293,
      "step": 1790
    },
    {
      "epoch": 0.14529717458008903,
      "grad_norm": 0.07727087289094925,
      "learning_rate": 0.0012,
      "loss": 2.758662223815918,
      "step": 1792
    },
    {
      "epoch": 0.14545933660529003,
      "grad_norm": 0.022576095536351204,
      "learning_rate": 0.0012,
      "loss": 2.7240264415740967,
      "step": 1794
    },
    {
      "epoch": 0.145621498630491,
      "grad_norm": 0.06359665840864182,
      "learning_rate": 0.0012,
      "loss": 2.695845127105713,
      "step": 1796
    },
    {
      "epoch": 0.145783660655692,
      "grad_norm": 0.04279274865984917,
      "learning_rate": 0.0012,
      "loss": 2.6913747787475586,
      "step": 1798
    },
    {
      "epoch": 0.145945822680893,
      "grad_norm": 0.025194628164172173,
      "learning_rate": 0.0012,
      "loss": 2.7337284088134766,
      "step": 1800
    },
    {
      "epoch": 0.14610798470609399,
      "grad_norm": 0.04457344859838486,
      "learning_rate": 0.0012,
      "loss": 2.7907252311706543,
      "step": 1802
    },
    {
      "epoch": 0.146270146731295,
      "grad_norm": 0.046919114887714386,
      "learning_rate": 0.0012,
      "loss": 2.7343263626098633,
      "step": 1804
    },
    {
      "epoch": 0.146432308756496,
      "grad_norm": 0.034811653196811676,
      "learning_rate": 0.0012,
      "loss": 2.7448320388793945,
      "step": 1806
    },
    {
      "epoch": 0.14659447078169696,
      "grad_norm": 0.05553847551345825,
      "learning_rate": 0.0012,
      "loss": 2.742560386657715,
      "step": 1808
    },
    {
      "epoch": 0.14675663280689796,
      "grad_norm": 0.0407397523522377,
      "learning_rate": 0.0012,
      "loss": 2.7569963932037354,
      "step": 1810
    },
    {
      "epoch": 0.14691879483209896,
      "grad_norm": 0.028414903208613396,
      "learning_rate": 0.0012,
      "loss": 2.769904851913452,
      "step": 1812
    },
    {
      "epoch": 0.14708095685729994,
      "grad_norm": 0.06152932345867157,
      "learning_rate": 0.0012,
      "loss": 2.694079875946045,
      "step": 1814
    },
    {
      "epoch": 0.14724311888250094,
      "grad_norm": 0.04396386072039604,
      "learning_rate": 0.0012,
      "loss": 2.69907283782959,
      "step": 1816
    },
    {
      "epoch": 0.14740528090770194,
      "grad_norm": 0.025318380445241928,
      "learning_rate": 0.0012,
      "loss": 2.7327170372009277,
      "step": 1818
    },
    {
      "epoch": 0.14756744293290291,
      "grad_norm": 0.04080936312675476,
      "learning_rate": 0.0012,
      "loss": 2.669325351715088,
      "step": 1820
    },
    {
      "epoch": 0.14772960495810392,
      "grad_norm": 0.038354046642780304,
      "learning_rate": 0.0012,
      "loss": 2.7635750770568848,
      "step": 1822
    },
    {
      "epoch": 0.14789176698330492,
      "grad_norm": 0.04054105281829834,
      "learning_rate": 0.0012,
      "loss": 2.667959690093994,
      "step": 1824
    },
    {
      "epoch": 0.1480539290085059,
      "grad_norm": 0.025114266201853752,
      "learning_rate": 0.0012,
      "loss": 2.7170238494873047,
      "step": 1826
    },
    {
      "epoch": 0.1482160910337069,
      "grad_norm": 0.025008268654346466,
      "learning_rate": 0.0012,
      "loss": 2.789632797241211,
      "step": 1828
    },
    {
      "epoch": 0.1483782530589079,
      "grad_norm": 0.03146105632185936,
      "learning_rate": 0.0012,
      "loss": 2.7106401920318604,
      "step": 1830
    },
    {
      "epoch": 0.1485404150841089,
      "grad_norm": 0.023103540763258934,
      "learning_rate": 0.0012,
      "loss": 2.7555148601531982,
      "step": 1832
    },
    {
      "epoch": 0.14870257710930987,
      "grad_norm": 0.02667209692299366,
      "learning_rate": 0.0012,
      "loss": 2.6982879638671875,
      "step": 1834
    },
    {
      "epoch": 0.14886473913451087,
      "grad_norm": 0.025596842169761658,
      "learning_rate": 0.0012,
      "loss": 2.758913516998291,
      "step": 1836
    },
    {
      "epoch": 0.14902690115971187,
      "grad_norm": 0.04801538214087486,
      "learning_rate": 0.0012,
      "loss": 2.715991973876953,
      "step": 1838
    },
    {
      "epoch": 0.14918906318491285,
      "grad_norm": 0.04243363440036774,
      "learning_rate": 0.0012,
      "loss": 2.7390527725219727,
      "step": 1840
    },
    {
      "epoch": 0.14935122521011385,
      "grad_norm": 0.02761574648320675,
      "learning_rate": 0.0012,
      "loss": 2.78446626663208,
      "step": 1842
    },
    {
      "epoch": 0.14951338723531485,
      "grad_norm": 0.057846758514642715,
      "learning_rate": 0.0012,
      "loss": 2.7774806022644043,
      "step": 1844
    },
    {
      "epoch": 0.14967554926051582,
      "grad_norm": 0.045362673699855804,
      "learning_rate": 0.0012,
      "loss": 2.7005488872528076,
      "step": 1846
    },
    {
      "epoch": 0.14983771128571682,
      "grad_norm": 0.05218088626861572,
      "learning_rate": 0.0012,
      "loss": 2.7068238258361816,
      "step": 1848
    },
    {
      "epoch": 0.14999987331091783,
      "grad_norm": 0.02770119719207287,
      "learning_rate": 0.0012,
      "loss": 2.761991024017334,
      "step": 1850
    },
    {
      "epoch": 0.1501620353361188,
      "grad_norm": 0.060202326625585556,
      "learning_rate": 0.0012,
      "loss": 2.725707530975342,
      "step": 1852
    },
    {
      "epoch": 0.1503241973613198,
      "grad_norm": 0.07335687428712845,
      "learning_rate": 0.0012,
      "loss": 2.730203628540039,
      "step": 1854
    },
    {
      "epoch": 0.1504863593865208,
      "grad_norm": 0.02891639620065689,
      "learning_rate": 0.0012,
      "loss": 2.708801507949829,
      "step": 1856
    },
    {
      "epoch": 0.15064852141172178,
      "grad_norm": 0.03552323952317238,
      "learning_rate": 0.0012,
      "loss": 2.72413969039917,
      "step": 1858
    },
    {
      "epoch": 0.15081068343692278,
      "grad_norm": 0.03860405460000038,
      "learning_rate": 0.0012,
      "loss": 2.7099509239196777,
      "step": 1860
    },
    {
      "epoch": 0.15097284546212378,
      "grad_norm": 0.024803688749670982,
      "learning_rate": 0.0012,
      "loss": 2.735090970993042,
      "step": 1862
    },
    {
      "epoch": 0.15113500748732475,
      "grad_norm": 0.02554907277226448,
      "learning_rate": 0.0012,
      "loss": 2.7920470237731934,
      "step": 1864
    },
    {
      "epoch": 0.15129716951252575,
      "grad_norm": 0.08711431175470352,
      "learning_rate": 0.0012,
      "loss": 2.753896951675415,
      "step": 1866
    },
    {
      "epoch": 0.15145933153772675,
      "grad_norm": 0.07359342277050018,
      "learning_rate": 0.0012,
      "loss": 2.6743831634521484,
      "step": 1868
    },
    {
      "epoch": 0.15162149356292773,
      "grad_norm": 0.02407222054898739,
      "learning_rate": 0.0012,
      "loss": 2.646705389022827,
      "step": 1870
    },
    {
      "epoch": 0.15178365558812873,
      "grad_norm": 0.048123009502887726,
      "learning_rate": 0.0012,
      "loss": 2.70823335647583,
      "step": 1872
    },
    {
      "epoch": 0.15194581761332973,
      "grad_norm": 0.05961233377456665,
      "learning_rate": 0.0012,
      "loss": 2.718547821044922,
      "step": 1874
    },
    {
      "epoch": 0.1521079796385307,
      "grad_norm": 0.03018476441502571,
      "learning_rate": 0.0012,
      "loss": 2.7308273315429688,
      "step": 1876
    },
    {
      "epoch": 0.1522701416637317,
      "grad_norm": 0.02668023109436035,
      "learning_rate": 0.0012,
      "loss": 2.7008306980133057,
      "step": 1878
    },
    {
      "epoch": 0.1524323036889327,
      "grad_norm": 0.03342357277870178,
      "learning_rate": 0.0012,
      "loss": 2.7694311141967773,
      "step": 1880
    },
    {
      "epoch": 0.15259446571413368,
      "grad_norm": 0.03750356659293175,
      "learning_rate": 0.0012,
      "loss": 2.8274755477905273,
      "step": 1882
    },
    {
      "epoch": 0.15275662773933468,
      "grad_norm": 0.030065178871154785,
      "learning_rate": 0.0012,
      "loss": 2.727346897125244,
      "step": 1884
    },
    {
      "epoch": 0.15291878976453568,
      "grad_norm": 0.026032399386167526,
      "learning_rate": 0.0012,
      "loss": 2.7526559829711914,
      "step": 1886
    },
    {
      "epoch": 0.15308095178973666,
      "grad_norm": 0.034453097730875015,
      "learning_rate": 0.0012,
      "loss": 2.709530830383301,
      "step": 1888
    },
    {
      "epoch": 0.15324311381493766,
      "grad_norm": 0.030391251668334007,
      "learning_rate": 0.0012,
      "loss": 2.7780683040618896,
      "step": 1890
    },
    {
      "epoch": 0.15340527584013866,
      "grad_norm": 0.02900552563369274,
      "learning_rate": 0.0012,
      "loss": 2.6898112297058105,
      "step": 1892
    },
    {
      "epoch": 0.15356743786533963,
      "grad_norm": 0.023332148790359497,
      "learning_rate": 0.0012,
      "loss": 2.7618255615234375,
      "step": 1894
    },
    {
      "epoch": 0.15372959989054064,
      "grad_norm": 0.02196205034852028,
      "learning_rate": 0.0012,
      "loss": 2.7152585983276367,
      "step": 1896
    },
    {
      "epoch": 0.15389176191574164,
      "grad_norm": 0.027053551748394966,
      "learning_rate": 0.0012,
      "loss": 2.743807554244995,
      "step": 1898
    },
    {
      "epoch": 0.1540539239409426,
      "grad_norm": 0.02627457305788994,
      "learning_rate": 0.0012,
      "loss": 2.7119359970092773,
      "step": 1900
    },
    {
      "epoch": 0.1542160859661436,
      "grad_norm": 0.024235567077994347,
      "learning_rate": 0.0012,
      "loss": 2.6912646293640137,
      "step": 1902
    },
    {
      "epoch": 0.1543782479913446,
      "grad_norm": 0.023947298526763916,
      "learning_rate": 0.0012,
      "loss": 2.7236108779907227,
      "step": 1904
    },
    {
      "epoch": 0.1545404100165456,
      "grad_norm": 0.027155540883541107,
      "learning_rate": 0.0012,
      "loss": 2.7701573371887207,
      "step": 1906
    },
    {
      "epoch": 0.1547025720417466,
      "grad_norm": 0.026496533304452896,
      "learning_rate": 0.0012,
      "loss": 2.727247714996338,
      "step": 1908
    },
    {
      "epoch": 0.1548647340669476,
      "grad_norm": 0.028977518901228905,
      "learning_rate": 0.0012,
      "loss": 2.6781859397888184,
      "step": 1910
    },
    {
      "epoch": 0.15502689609214856,
      "grad_norm": 0.025533655658364296,
      "learning_rate": 0.0012,
      "loss": 2.7434582710266113,
      "step": 1912
    },
    {
      "epoch": 0.15518905811734957,
      "grad_norm": 0.021744512021541595,
      "learning_rate": 0.0012,
      "loss": 2.728518486022949,
      "step": 1914
    },
    {
      "epoch": 0.15535122014255057,
      "grad_norm": 0.036230288445949554,
      "learning_rate": 0.0012,
      "loss": 2.7146177291870117,
      "step": 1916
    },
    {
      "epoch": 0.15551338216775154,
      "grad_norm": 0.05425127595663071,
      "learning_rate": 0.0012,
      "loss": 2.6996593475341797,
      "step": 1918
    },
    {
      "epoch": 0.15567554419295254,
      "grad_norm": 0.04537057504057884,
      "learning_rate": 0.0012,
      "loss": 2.6985349655151367,
      "step": 1920
    },
    {
      "epoch": 0.15583770621815354,
      "grad_norm": 0.02379826456308365,
      "learning_rate": 0.0012,
      "loss": 2.682090997695923,
      "step": 1922
    },
    {
      "epoch": 0.15599986824335452,
      "grad_norm": 0.03879925608634949,
      "learning_rate": 0.0012,
      "loss": 2.7456653118133545,
      "step": 1924
    },
    {
      "epoch": 0.15616203026855552,
      "grad_norm": 0.045712318271398544,
      "learning_rate": 0.0012,
      "loss": 2.7793121337890625,
      "step": 1926
    },
    {
      "epoch": 0.15632419229375652,
      "grad_norm": 0.04605862498283386,
      "learning_rate": 0.0012,
      "loss": 2.731492042541504,
      "step": 1928
    },
    {
      "epoch": 0.1564863543189575,
      "grad_norm": 0.025156021118164062,
      "learning_rate": 0.0012,
      "loss": 2.7236340045928955,
      "step": 1930
    },
    {
      "epoch": 0.1566485163441585,
      "grad_norm": 0.02487189881503582,
      "learning_rate": 0.0012,
      "loss": 2.713637590408325,
      "step": 1932
    },
    {
      "epoch": 0.1568106783693595,
      "grad_norm": 0.0237958412617445,
      "learning_rate": 0.0012,
      "loss": 2.754927158355713,
      "step": 1934
    },
    {
      "epoch": 0.15697284039456047,
      "grad_norm": 0.05953480675816536,
      "learning_rate": 0.0012,
      "loss": 2.6644704341888428,
      "step": 1936
    },
    {
      "epoch": 0.15713500241976147,
      "grad_norm": 0.0345805361866951,
      "learning_rate": 0.0012,
      "loss": 2.7283883094787598,
      "step": 1938
    },
    {
      "epoch": 0.15729716444496247,
      "grad_norm": 0.0328529067337513,
      "learning_rate": 0.0012,
      "loss": 2.693020820617676,
      "step": 1940
    },
    {
      "epoch": 0.15745932647016345,
      "grad_norm": 0.024471575394272804,
      "learning_rate": 0.0012,
      "loss": 2.759561777114868,
      "step": 1942
    },
    {
      "epoch": 0.15762148849536445,
      "grad_norm": 0.033353377133607864,
      "learning_rate": 0.0012,
      "loss": 2.698369264602661,
      "step": 1944
    },
    {
      "epoch": 0.15778365052056545,
      "grad_norm": 0.022027406841516495,
      "learning_rate": 0.0012,
      "loss": 2.7359421253204346,
      "step": 1946
    },
    {
      "epoch": 0.15794581254576642,
      "grad_norm": 0.02367263473570347,
      "learning_rate": 0.0012,
      "loss": 2.693741798400879,
      "step": 1948
    },
    {
      "epoch": 0.15810797457096742,
      "grad_norm": 0.023186802864074707,
      "learning_rate": 0.0012,
      "loss": 2.8029491901397705,
      "step": 1950
    },
    {
      "epoch": 0.15827013659616843,
      "grad_norm": 0.03148287162184715,
      "learning_rate": 0.0012,
      "loss": 2.741569995880127,
      "step": 1952
    },
    {
      "epoch": 0.1584322986213694,
      "grad_norm": 0.06713547557592392,
      "learning_rate": 0.0012,
      "loss": 2.714311361312866,
      "step": 1954
    },
    {
      "epoch": 0.1585944606465704,
      "grad_norm": 0.02234683930873871,
      "learning_rate": 0.0012,
      "loss": 2.7321972846984863,
      "step": 1956
    },
    {
      "epoch": 0.1587566226717714,
      "grad_norm": 0.03921957314014435,
      "learning_rate": 0.0012,
      "loss": 2.7466416358947754,
      "step": 1958
    },
    {
      "epoch": 0.15891878469697238,
      "grad_norm": 0.0532679446041584,
      "learning_rate": 0.0012,
      "loss": 2.666926622390747,
      "step": 1960
    },
    {
      "epoch": 0.15908094672217338,
      "grad_norm": 0.039678145200014114,
      "learning_rate": 0.0012,
      "loss": 2.6859984397888184,
      "step": 1962
    },
    {
      "epoch": 0.15924310874737438,
      "grad_norm": 0.03096655011177063,
      "learning_rate": 0.0012,
      "loss": 2.769064426422119,
      "step": 1964
    },
    {
      "epoch": 0.15940527077257535,
      "grad_norm": 0.03644552826881409,
      "learning_rate": 0.0012,
      "loss": 2.729377031326294,
      "step": 1966
    },
    {
      "epoch": 0.15956743279777635,
      "grad_norm": 0.023593273013830185,
      "learning_rate": 0.0012,
      "loss": 2.738727331161499,
      "step": 1968
    },
    {
      "epoch": 0.15972959482297736,
      "grad_norm": 0.05963050201535225,
      "learning_rate": 0.0012,
      "loss": 2.7985901832580566,
      "step": 1970
    },
    {
      "epoch": 0.15989175684817833,
      "grad_norm": 0.03883720934391022,
      "learning_rate": 0.0012,
      "loss": 2.7169904708862305,
      "step": 1972
    },
    {
      "epoch": 0.16005391887337933,
      "grad_norm": 0.037231069058179855,
      "learning_rate": 0.0012,
      "loss": 2.702781915664673,
      "step": 1974
    },
    {
      "epoch": 0.16021608089858033,
      "grad_norm": 0.030991990119218826,
      "learning_rate": 0.0012,
      "loss": 2.728119373321533,
      "step": 1976
    },
    {
      "epoch": 0.1603782429237813,
      "grad_norm": 0.02976652793586254,
      "learning_rate": 0.0012,
      "loss": 2.7577507495880127,
      "step": 1978
    },
    {
      "epoch": 0.1605404049489823,
      "grad_norm": 0.02410745620727539,
      "learning_rate": 0.0012,
      "loss": 2.7429795265197754,
      "step": 1980
    },
    {
      "epoch": 0.1607025669741833,
      "grad_norm": 0.02967197820544243,
      "learning_rate": 0.0012,
      "loss": 2.7828619480133057,
      "step": 1982
    },
    {
      "epoch": 0.16086472899938428,
      "grad_norm": 0.04056461527943611,
      "learning_rate": 0.0012,
      "loss": 2.7042291164398193,
      "step": 1984
    },
    {
      "epoch": 0.16102689102458528,
      "grad_norm": 0.02482522465288639,
      "learning_rate": 0.0012,
      "loss": 2.7404823303222656,
      "step": 1986
    },
    {
      "epoch": 0.16118905304978629,
      "grad_norm": 0.030895352363586426,
      "learning_rate": 0.0012,
      "loss": 2.664069175720215,
      "step": 1988
    },
    {
      "epoch": 0.16135121507498726,
      "grad_norm": 0.025476375594735146,
      "learning_rate": 0.0012,
      "loss": 2.791372776031494,
      "step": 1990
    },
    {
      "epoch": 0.16151337710018826,
      "grad_norm": 0.03843531385064125,
      "learning_rate": 0.0012,
      "loss": 2.689337730407715,
      "step": 1992
    },
    {
      "epoch": 0.16167553912538926,
      "grad_norm": 0.038494840264320374,
      "learning_rate": 0.0012,
      "loss": 2.7282543182373047,
      "step": 1994
    },
    {
      "epoch": 0.16183770115059024,
      "grad_norm": 0.057491742074489594,
      "learning_rate": 0.0012,
      "loss": 2.6860828399658203,
      "step": 1996
    },
    {
      "epoch": 0.16199986317579124,
      "grad_norm": 0.024999914690852165,
      "learning_rate": 0.0012,
      "loss": 2.6361024379730225,
      "step": 1998
    },
    {
      "epoch": 0.16216202520099224,
      "grad_norm": 0.02645510621368885,
      "learning_rate": 0.0012,
      "loss": 2.7091779708862305,
      "step": 2000
    },
    {
      "epoch": 0.1623241872261932,
      "grad_norm": 0.04748566076159477,
      "learning_rate": 0.0012,
      "loss": 2.7209010124206543,
      "step": 2002
    },
    {
      "epoch": 0.1624863492513942,
      "grad_norm": 0.05596374720335007,
      "learning_rate": 0.0012,
      "loss": 2.6391029357910156,
      "step": 2004
    },
    {
      "epoch": 0.16264851127659521,
      "grad_norm": 0.062242839485406876,
      "learning_rate": 0.0012,
      "loss": 2.746389865875244,
      "step": 2006
    },
    {
      "epoch": 0.1628106733017962,
      "grad_norm": 0.02265295572578907,
      "learning_rate": 0.0012,
      "loss": 2.696781635284424,
      "step": 2008
    },
    {
      "epoch": 0.1629728353269972,
      "grad_norm": 0.0512719564139843,
      "learning_rate": 0.0012,
      "loss": 2.7442054748535156,
      "step": 2010
    },
    {
      "epoch": 0.1631349973521982,
      "grad_norm": 0.0576956681907177,
      "learning_rate": 0.0012,
      "loss": 2.7668466567993164,
      "step": 2012
    },
    {
      "epoch": 0.16329715937739916,
      "grad_norm": 0.03548339381814003,
      "learning_rate": 0.0012,
      "loss": 2.708108425140381,
      "step": 2014
    },
    {
      "epoch": 0.16345932140260017,
      "grad_norm": 0.0265990998595953,
      "learning_rate": 0.0012,
      "loss": 2.710696220397949,
      "step": 2016
    },
    {
      "epoch": 0.16362148342780117,
      "grad_norm": 0.028373170644044876,
      "learning_rate": 0.0012,
      "loss": 2.7058658599853516,
      "step": 2018
    },
    {
      "epoch": 0.16378364545300214,
      "grad_norm": 0.039339061826467514,
      "learning_rate": 0.0012,
      "loss": 2.6969332695007324,
      "step": 2020
    },
    {
      "epoch": 0.16394580747820314,
      "grad_norm": 0.02266733907163143,
      "learning_rate": 0.0012,
      "loss": 2.6208858489990234,
      "step": 2022
    },
    {
      "epoch": 0.16410796950340414,
      "grad_norm": 0.03304461017251015,
      "learning_rate": 0.0012,
      "loss": 2.6749136447906494,
      "step": 2024
    },
    {
      "epoch": 0.16427013152860512,
      "grad_norm": 0.026189737021923065,
      "learning_rate": 0.0012,
      "loss": 2.761638641357422,
      "step": 2026
    },
    {
      "epoch": 0.16443229355380612,
      "grad_norm": 0.03636905923485756,
      "learning_rate": 0.0012,
      "loss": 2.6809797286987305,
      "step": 2028
    },
    {
      "epoch": 0.16459445557900712,
      "grad_norm": 0.032276228070259094,
      "learning_rate": 0.0012,
      "loss": 2.7107865810394287,
      "step": 2030
    },
    {
      "epoch": 0.1647566176042081,
      "grad_norm": 0.03629191219806671,
      "learning_rate": 0.0012,
      "loss": 2.738797187805176,
      "step": 2032
    },
    {
      "epoch": 0.1649187796294091,
      "grad_norm": 0.04197670891880989,
      "learning_rate": 0.0012,
      "loss": 2.702542781829834,
      "step": 2034
    },
    {
      "epoch": 0.1650809416546101,
      "grad_norm": 0.040572937577962875,
      "learning_rate": 0.0012,
      "loss": 2.7755279541015625,
      "step": 2036
    },
    {
      "epoch": 0.16524310367981107,
      "grad_norm": 0.024032549932599068,
      "learning_rate": 0.0012,
      "loss": 2.722550630569458,
      "step": 2038
    },
    {
      "epoch": 0.16540526570501207,
      "grad_norm": 0.02315390296280384,
      "learning_rate": 0.0012,
      "loss": 2.759287118911743,
      "step": 2040
    },
    {
      "epoch": 0.16556742773021307,
      "grad_norm": 0.04772220551967621,
      "learning_rate": 0.0012,
      "loss": 2.6942226886749268,
      "step": 2042
    },
    {
      "epoch": 0.16572958975541405,
      "grad_norm": 0.03675943613052368,
      "learning_rate": 0.0012,
      "loss": 2.642526149749756,
      "step": 2044
    },
    {
      "epoch": 0.16589175178061505,
      "grad_norm": 0.03647027164697647,
      "learning_rate": 0.0012,
      "loss": 2.7855634689331055,
      "step": 2046
    },
    {
      "epoch": 0.16605391380581605,
      "grad_norm": 0.04033167287707329,
      "learning_rate": 0.0012,
      "loss": 2.7240357398986816,
      "step": 2048
    },
    {
      "epoch": 0.16621607583101702,
      "grad_norm": 0.037436168640851974,
      "learning_rate": 0.0012,
      "loss": 2.745492458343506,
      "step": 2050
    },
    {
      "epoch": 0.16637823785621803,
      "grad_norm": 0.02238456904888153,
      "learning_rate": 0.0012,
      "loss": 2.748833179473877,
      "step": 2052
    },
    {
      "epoch": 0.16654039988141903,
      "grad_norm": 0.03202953562140465,
      "learning_rate": 0.0012,
      "loss": 2.7119431495666504,
      "step": 2054
    },
    {
      "epoch": 0.16670256190662,
      "grad_norm": 0.0576016828417778,
      "learning_rate": 0.0012,
      "loss": 2.763415813446045,
      "step": 2056
    },
    {
      "epoch": 0.166864723931821,
      "grad_norm": 0.025856146588921547,
      "learning_rate": 0.0012,
      "loss": 2.647017002105713,
      "step": 2058
    },
    {
      "epoch": 0.167026885957022,
      "grad_norm": 0.02948053553700447,
      "learning_rate": 0.0012,
      "loss": 2.6793999671936035,
      "step": 2060
    },
    {
      "epoch": 0.16718904798222298,
      "grad_norm": 0.041251812130212784,
      "learning_rate": 0.0012,
      "loss": 2.6705493927001953,
      "step": 2062
    },
    {
      "epoch": 0.16735121000742398,
      "grad_norm": 0.025495806708931923,
      "learning_rate": 0.0012,
      "loss": 2.712280750274658,
      "step": 2064
    },
    {
      "epoch": 0.16751337203262498,
      "grad_norm": 0.03599660098552704,
      "learning_rate": 0.0012,
      "loss": 2.7650623321533203,
      "step": 2066
    },
    {
      "epoch": 0.16767553405782595,
      "grad_norm": 0.02406834252178669,
      "learning_rate": 0.0012,
      "loss": 2.759732246398926,
      "step": 2068
    },
    {
      "epoch": 0.16783769608302695,
      "grad_norm": 0.02784694731235504,
      "learning_rate": 0.0012,
      "loss": 2.6601028442382812,
      "step": 2070
    },
    {
      "epoch": 0.16799985810822796,
      "grad_norm": 0.03020234778523445,
      "learning_rate": 0.0012,
      "loss": 2.743163585662842,
      "step": 2072
    },
    {
      "epoch": 0.16816202013342893,
      "grad_norm": 0.029128724709153175,
      "learning_rate": 0.0012,
      "loss": 2.720031261444092,
      "step": 2074
    },
    {
      "epoch": 0.16832418215862993,
      "grad_norm": 0.035132914781570435,
      "learning_rate": 0.0012,
      "loss": 2.709052085876465,
      "step": 2076
    },
    {
      "epoch": 0.16848634418383093,
      "grad_norm": 0.026463491842150688,
      "learning_rate": 0.0012,
      "loss": 2.67083477973938,
      "step": 2078
    },
    {
      "epoch": 0.1686485062090319,
      "grad_norm": 0.03936498984694481,
      "learning_rate": 0.0012,
      "loss": 2.6700778007507324,
      "step": 2080
    },
    {
      "epoch": 0.1688106682342329,
      "grad_norm": 0.029072808101773262,
      "learning_rate": 0.0012,
      "loss": 2.7123730182647705,
      "step": 2082
    },
    {
      "epoch": 0.1689728302594339,
      "grad_norm": 0.024036433547735214,
      "learning_rate": 0.0012,
      "loss": 2.7239789962768555,
      "step": 2084
    },
    {
      "epoch": 0.16913499228463488,
      "grad_norm": 0.032044995576143265,
      "learning_rate": 0.0012,
      "loss": 2.717738151550293,
      "step": 2086
    },
    {
      "epoch": 0.16929715430983588,
      "grad_norm": 0.05296298488974571,
      "learning_rate": 0.0012,
      "loss": 2.730583667755127,
      "step": 2088
    },
    {
      "epoch": 0.16945931633503689,
      "grad_norm": 0.059763211756944656,
      "learning_rate": 0.0012,
      "loss": 2.716948986053467,
      "step": 2090
    },
    {
      "epoch": 0.16962147836023786,
      "grad_norm": 0.0606759712100029,
      "learning_rate": 0.0012,
      "loss": 2.6313626766204834,
      "step": 2092
    },
    {
      "epoch": 0.16978364038543886,
      "grad_norm": 0.026697689667344093,
      "learning_rate": 0.0012,
      "loss": 2.73433780670166,
      "step": 2094
    },
    {
      "epoch": 0.16994580241063986,
      "grad_norm": 0.038482312113046646,
      "learning_rate": 0.0012,
      "loss": 2.783747434616089,
      "step": 2096
    },
    {
      "epoch": 0.17010796443584084,
      "grad_norm": 0.059857115149497986,
      "learning_rate": 0.0012,
      "loss": 2.737879991531372,
      "step": 2098
    },
    {
      "epoch": 0.17027012646104184,
      "grad_norm": 0.07122787833213806,
      "learning_rate": 0.0012,
      "loss": 2.736626386642456,
      "step": 2100
    },
    {
      "epoch": 0.17043228848624284,
      "grad_norm": 0.024399712681770325,
      "learning_rate": 0.0012,
      "loss": 2.7318148612976074,
      "step": 2102
    },
    {
      "epoch": 0.1705944505114438,
      "grad_norm": 0.05522390082478523,
      "learning_rate": 0.0012,
      "loss": 2.6516165733337402,
      "step": 2104
    },
    {
      "epoch": 0.17075661253664481,
      "grad_norm": 0.04821356385946274,
      "learning_rate": 0.0012,
      "loss": 2.649806499481201,
      "step": 2106
    },
    {
      "epoch": 0.17091877456184582,
      "grad_norm": 0.028881505131721497,
      "learning_rate": 0.0012,
      "loss": 2.752748727798462,
      "step": 2108
    },
    {
      "epoch": 0.1710809365870468,
      "grad_norm": 0.026502329856157303,
      "learning_rate": 0.0012,
      "loss": 2.7115094661712646,
      "step": 2110
    },
    {
      "epoch": 0.1712430986122478,
      "grad_norm": 0.048590585589408875,
      "learning_rate": 0.0012,
      "loss": 2.7530710697174072,
      "step": 2112
    },
    {
      "epoch": 0.1714052606374488,
      "grad_norm": 0.04069884493947029,
      "learning_rate": 0.0012,
      "loss": 2.8182663917541504,
      "step": 2114
    },
    {
      "epoch": 0.17156742266264977,
      "grad_norm": 0.023906687274575233,
      "learning_rate": 0.0012,
      "loss": 2.734062671661377,
      "step": 2116
    },
    {
      "epoch": 0.17172958468785077,
      "grad_norm": 0.039869919419288635,
      "learning_rate": 0.0012,
      "loss": 2.6775341033935547,
      "step": 2118
    },
    {
      "epoch": 0.17189174671305177,
      "grad_norm": 0.04953397437930107,
      "learning_rate": 0.0012,
      "loss": 2.7568230628967285,
      "step": 2120
    },
    {
      "epoch": 0.17205390873825274,
      "grad_norm": 0.05024636164307594,
      "learning_rate": 0.0012,
      "loss": 2.691988706588745,
      "step": 2122
    },
    {
      "epoch": 0.17221607076345374,
      "grad_norm": 0.025018706917762756,
      "learning_rate": 0.0012,
      "loss": 2.711871385574341,
      "step": 2124
    },
    {
      "epoch": 0.17237823278865474,
      "grad_norm": 0.03926042094826698,
      "learning_rate": 0.0012,
      "loss": 2.727396011352539,
      "step": 2126
    },
    {
      "epoch": 0.17254039481385572,
      "grad_norm": 0.042397137731313705,
      "learning_rate": 0.0012,
      "loss": 2.7021872997283936,
      "step": 2128
    },
    {
      "epoch": 0.17270255683905672,
      "grad_norm": 0.0229585450142622,
      "learning_rate": 0.0012,
      "loss": 2.732343912124634,
      "step": 2130
    },
    {
      "epoch": 0.17286471886425772,
      "grad_norm": 0.022223403677344322,
      "learning_rate": 0.0012,
      "loss": 2.7199912071228027,
      "step": 2132
    },
    {
      "epoch": 0.1730268808894587,
      "grad_norm": 0.04514741897583008,
      "learning_rate": 0.0012,
      "loss": 2.6833784580230713,
      "step": 2134
    },
    {
      "epoch": 0.1731890429146597,
      "grad_norm": 0.04263390228152275,
      "learning_rate": 0.0012,
      "loss": 2.6208548545837402,
      "step": 2136
    },
    {
      "epoch": 0.1733512049398607,
      "grad_norm": 0.036768969148397446,
      "learning_rate": 0.0012,
      "loss": 2.7830677032470703,
      "step": 2138
    },
    {
      "epoch": 0.17351336696506167,
      "grad_norm": 0.022001391276717186,
      "learning_rate": 0.0012,
      "loss": 2.7148728370666504,
      "step": 2140
    },
    {
      "epoch": 0.17367552899026267,
      "grad_norm": 0.022117769345641136,
      "learning_rate": 0.0012,
      "loss": 2.766353130340576,
      "step": 2142
    },
    {
      "epoch": 0.17383769101546367,
      "grad_norm": 0.04036613926291466,
      "learning_rate": 0.0012,
      "loss": 2.7151436805725098,
      "step": 2144
    },
    {
      "epoch": 0.17399985304066465,
      "grad_norm": 0.05033177509903908,
      "learning_rate": 0.0012,
      "loss": 2.786353588104248,
      "step": 2146
    },
    {
      "epoch": 0.17416201506586565,
      "grad_norm": 0.024325354024767876,
      "learning_rate": 0.0012,
      "loss": 2.711942672729492,
      "step": 2148
    },
    {
      "epoch": 0.17432417709106665,
      "grad_norm": 0.023068256676197052,
      "learning_rate": 0.0012,
      "loss": 2.7278010845184326,
      "step": 2150
    },
    {
      "epoch": 0.17448633911626762,
      "grad_norm": 0.051669903099536896,
      "learning_rate": 0.0012,
      "loss": 2.75547456741333,
      "step": 2152
    },
    {
      "epoch": 0.17464850114146863,
      "grad_norm": 0.04827481135725975,
      "learning_rate": 0.0012,
      "loss": 2.737154483795166,
      "step": 2154
    },
    {
      "epoch": 0.17481066316666963,
      "grad_norm": 0.033639878034591675,
      "learning_rate": 0.0012,
      "loss": 2.7454090118408203,
      "step": 2156
    },
    {
      "epoch": 0.1749728251918706,
      "grad_norm": 0.02334778942167759,
      "learning_rate": 0.0012,
      "loss": 2.7312731742858887,
      "step": 2158
    },
    {
      "epoch": 0.1751349872170716,
      "grad_norm": 0.046179041266441345,
      "learning_rate": 0.0012,
      "loss": 2.6605706214904785,
      "step": 2160
    },
    {
      "epoch": 0.1752971492422726,
      "grad_norm": 0.05803694948554039,
      "learning_rate": 0.0012,
      "loss": 2.6837141513824463,
      "step": 2162
    },
    {
      "epoch": 0.1754593112674736,
      "grad_norm": 0.03290850669145584,
      "learning_rate": 0.0012,
      "loss": 2.695559024810791,
      "step": 2164
    },
    {
      "epoch": 0.17562147329267458,
      "grad_norm": 0.023084884509444237,
      "learning_rate": 0.0012,
      "loss": 2.710212230682373,
      "step": 2166
    },
    {
      "epoch": 0.17578363531787558,
      "grad_norm": 0.03274243324995041,
      "learning_rate": 0.0012,
      "loss": 2.759770393371582,
      "step": 2168
    },
    {
      "epoch": 0.17594579734307658,
      "grad_norm": 0.03585462272167206,
      "learning_rate": 0.0012,
      "loss": 2.6785941123962402,
      "step": 2170
    },
    {
      "epoch": 0.17610795936827756,
      "grad_norm": 0.030521783977746964,
      "learning_rate": 0.0012,
      "loss": 2.7462780475616455,
      "step": 2172
    },
    {
      "epoch": 0.17627012139347856,
      "grad_norm": 0.03140690177679062,
      "learning_rate": 0.0012,
      "loss": 2.6939616203308105,
      "step": 2174
    },
    {
      "epoch": 0.17643228341867956,
      "grad_norm": 0.02499580755829811,
      "learning_rate": 0.0012,
      "loss": 2.672567844390869,
      "step": 2176
    },
    {
      "epoch": 0.17659444544388053,
      "grad_norm": 0.03590945154428482,
      "learning_rate": 0.0012,
      "loss": 2.735933303833008,
      "step": 2178
    },
    {
      "epoch": 0.17675660746908153,
      "grad_norm": 0.02630893513560295,
      "learning_rate": 0.0012,
      "loss": 2.717409610748291,
      "step": 2180
    },
    {
      "epoch": 0.17691876949428254,
      "grad_norm": 0.022761056199669838,
      "learning_rate": 0.0012,
      "loss": 2.663046360015869,
      "step": 2182
    },
    {
      "epoch": 0.1770809315194835,
      "grad_norm": 0.03033497929573059,
      "learning_rate": 0.0012,
      "loss": 2.74444317817688,
      "step": 2184
    },
    {
      "epoch": 0.1772430935446845,
      "grad_norm": 0.02844209410250187,
      "learning_rate": 0.0012,
      "loss": 2.709834098815918,
      "step": 2186
    },
    {
      "epoch": 0.1774052555698855,
      "grad_norm": 0.03220690041780472,
      "learning_rate": 0.0012,
      "loss": 2.768150806427002,
      "step": 2188
    },
    {
      "epoch": 0.17756741759508649,
      "grad_norm": 0.062319155782461166,
      "learning_rate": 0.0012,
      "loss": 2.727530002593994,
      "step": 2190
    },
    {
      "epoch": 0.1777295796202875,
      "grad_norm": 0.029833899810910225,
      "learning_rate": 0.0012,
      "loss": 2.718817710876465,
      "step": 2192
    },
    {
      "epoch": 0.1778917416454885,
      "grad_norm": 0.057182881981134415,
      "learning_rate": 0.0012,
      "loss": 2.699248790740967,
      "step": 2194
    },
    {
      "epoch": 0.17805390367068946,
      "grad_norm": 0.07224905490875244,
      "learning_rate": 0.0012,
      "loss": 2.748046875,
      "step": 2196
    },
    {
      "epoch": 0.17821606569589046,
      "grad_norm": 0.04215170443058014,
      "learning_rate": 0.0012,
      "loss": 2.7129669189453125,
      "step": 2198
    },
    {
      "epoch": 0.17837822772109146,
      "grad_norm": 0.05318208411335945,
      "learning_rate": 0.0012,
      "loss": 2.707000255584717,
      "step": 2200
    },
    {
      "epoch": 0.17854038974629244,
      "grad_norm": 0.08464711904525757,
      "learning_rate": 0.0012,
      "loss": 2.713369369506836,
      "step": 2202
    },
    {
      "epoch": 0.17870255177149344,
      "grad_norm": 0.025962693616747856,
      "learning_rate": 0.0012,
      "loss": 2.6886444091796875,
      "step": 2204
    },
    {
      "epoch": 0.17886471379669444,
      "grad_norm": 0.038629014045000076,
      "learning_rate": 0.0012,
      "loss": 2.7479348182678223,
      "step": 2206
    },
    {
      "epoch": 0.17902687582189541,
      "grad_norm": 0.03370707854628563,
      "learning_rate": 0.0012,
      "loss": 2.7481179237365723,
      "step": 2208
    },
    {
      "epoch": 0.17918903784709642,
      "grad_norm": 0.02458900399506092,
      "learning_rate": 0.0012,
      "loss": 2.768352508544922,
      "step": 2210
    },
    {
      "epoch": 0.17935119987229742,
      "grad_norm": 0.023991646245121956,
      "learning_rate": 0.0012,
      "loss": 2.7292237281799316,
      "step": 2212
    },
    {
      "epoch": 0.1795133618974984,
      "grad_norm": 0.024124465882778168,
      "learning_rate": 0.0012,
      "loss": 2.789297103881836,
      "step": 2214
    },
    {
      "epoch": 0.1796755239226994,
      "grad_norm": 0.038784921169281006,
      "learning_rate": 0.0012,
      "loss": 2.6633992195129395,
      "step": 2216
    },
    {
      "epoch": 0.1798376859479004,
      "grad_norm": 0.04855378344655037,
      "learning_rate": 0.0012,
      "loss": 2.6800336837768555,
      "step": 2218
    },
    {
      "epoch": 0.17999984797310137,
      "grad_norm": 0.04416748881340027,
      "learning_rate": 0.0012,
      "loss": 2.6927490234375,
      "step": 2220
    },
    {
      "epoch": 0.18016200999830237,
      "grad_norm": 0.039457015693187714,
      "learning_rate": 0.0012,
      "loss": 2.748074531555176,
      "step": 2222
    },
    {
      "epoch": 0.18032417202350337,
      "grad_norm": 0.04165374115109444,
      "learning_rate": 0.0012,
      "loss": 2.7640695571899414,
      "step": 2224
    },
    {
      "epoch": 0.18048633404870434,
      "grad_norm": 0.03284842520952225,
      "learning_rate": 0.0012,
      "loss": 2.64888858795166,
      "step": 2226
    },
    {
      "epoch": 0.18064849607390535,
      "grad_norm": 0.02412067912518978,
      "learning_rate": 0.0012,
      "loss": 2.7390336990356445,
      "step": 2228
    },
    {
      "epoch": 0.18081065809910635,
      "grad_norm": 0.04261936619877815,
      "learning_rate": 0.0012,
      "loss": 2.6985726356506348,
      "step": 2230
    },
    {
      "epoch": 0.18097282012430732,
      "grad_norm": 0.029773211106657982,
      "learning_rate": 0.0012,
      "loss": 2.761223316192627,
      "step": 2232
    },
    {
      "epoch": 0.18113498214950832,
      "grad_norm": 0.021433643996715546,
      "learning_rate": 0.0012,
      "loss": 2.6578540802001953,
      "step": 2234
    },
    {
      "epoch": 0.18129714417470932,
      "grad_norm": 0.0245012566447258,
      "learning_rate": 0.0012,
      "loss": 2.7483417987823486,
      "step": 2236
    },
    {
      "epoch": 0.1814593061999103,
      "grad_norm": 0.024525830522179604,
      "learning_rate": 0.0012,
      "loss": 2.7421674728393555,
      "step": 2238
    },
    {
      "epoch": 0.1816214682251113,
      "grad_norm": 0.031120343133807182,
      "learning_rate": 0.0012,
      "loss": 2.731092691421509,
      "step": 2240
    },
    {
      "epoch": 0.1817836302503123,
      "grad_norm": 0.040459394454956055,
      "learning_rate": 0.0012,
      "loss": 2.6953673362731934,
      "step": 2242
    },
    {
      "epoch": 0.18194579227551327,
      "grad_norm": 0.06581681966781616,
      "learning_rate": 0.0012,
      "loss": 2.786656618118286,
      "step": 2244
    },
    {
      "epoch": 0.18210795430071428,
      "grad_norm": 0.03320524841547012,
      "learning_rate": 0.0012,
      "loss": 2.7394042015075684,
      "step": 2246
    },
    {
      "epoch": 0.18227011632591528,
      "grad_norm": 0.03370809182524681,
      "learning_rate": 0.0012,
      "loss": 2.7714805603027344,
      "step": 2248
    },
    {
      "epoch": 0.18243227835111625,
      "grad_norm": 0.03489774465560913,
      "learning_rate": 0.0012,
      "loss": 2.681706666946411,
      "step": 2250
    },
    {
      "epoch": 0.18259444037631725,
      "grad_norm": 0.027355749160051346,
      "learning_rate": 0.0012,
      "loss": 2.725304365158081,
      "step": 2252
    },
    {
      "epoch": 0.18275660240151825,
      "grad_norm": 0.04110114648938179,
      "learning_rate": 0.0012,
      "loss": 2.691678524017334,
      "step": 2254
    },
    {
      "epoch": 0.18291876442671923,
      "grad_norm": 0.028653385117650032,
      "learning_rate": 0.0012,
      "loss": 2.7302169799804688,
      "step": 2256
    },
    {
      "epoch": 0.18308092645192023,
      "grad_norm": 0.02331748977303505,
      "learning_rate": 0.0012,
      "loss": 2.68292236328125,
      "step": 2258
    },
    {
      "epoch": 0.18324308847712123,
      "grad_norm": 0.055397603660821915,
      "learning_rate": 0.0012,
      "loss": 2.7372360229492188,
      "step": 2260
    },
    {
      "epoch": 0.1834052505023222,
      "grad_norm": 0.034301064908504486,
      "learning_rate": 0.0012,
      "loss": 2.729640007019043,
      "step": 2262
    },
    {
      "epoch": 0.1835674125275232,
      "grad_norm": 0.027365053072571754,
      "learning_rate": 0.0012,
      "loss": 2.7173986434936523,
      "step": 2264
    },
    {
      "epoch": 0.1837295745527242,
      "grad_norm": 0.0495884083211422,
      "learning_rate": 0.0012,
      "loss": 2.638090133666992,
      "step": 2266
    },
    {
      "epoch": 0.18389173657792518,
      "grad_norm": 0.06081481650471687,
      "learning_rate": 0.0012,
      "loss": 2.6738102436065674,
      "step": 2268
    },
    {
      "epoch": 0.18405389860312618,
      "grad_norm": 0.03772469609975815,
      "learning_rate": 0.0012,
      "loss": 2.7381978034973145,
      "step": 2270
    },
    {
      "epoch": 0.18421606062832718,
      "grad_norm": 0.024908430874347687,
      "learning_rate": 0.0012,
      "loss": 2.710465431213379,
      "step": 2272
    },
    {
      "epoch": 0.18437822265352816,
      "grad_norm": 0.027680441737174988,
      "learning_rate": 0.0012,
      "loss": 2.7008137702941895,
      "step": 2274
    },
    {
      "epoch": 0.18454038467872916,
      "grad_norm": 0.051100779324769974,
      "learning_rate": 0.0012,
      "loss": 2.7465033531188965,
      "step": 2276
    },
    {
      "epoch": 0.18470254670393016,
      "grad_norm": 0.02480226941406727,
      "learning_rate": 0.0012,
      "loss": 2.6622676849365234,
      "step": 2278
    },
    {
      "epoch": 0.18486470872913113,
      "grad_norm": 0.030865605920553207,
      "learning_rate": 0.0012,
      "loss": 2.7685892581939697,
      "step": 2280
    },
    {
      "epoch": 0.18502687075433213,
      "grad_norm": 0.022489452734589577,
      "learning_rate": 0.0012,
      "loss": 2.7630198001861572,
      "step": 2282
    },
    {
      "epoch": 0.18518903277953314,
      "grad_norm": 0.0214128065854311,
      "learning_rate": 0.0012,
      "loss": 2.7405927181243896,
      "step": 2284
    },
    {
      "epoch": 0.1853511948047341,
      "grad_norm": 0.023799337446689606,
      "learning_rate": 0.0012,
      "loss": 2.70550274848938,
      "step": 2286
    },
    {
      "epoch": 0.1855133568299351,
      "grad_norm": 0.03071829304099083,
      "learning_rate": 0.0012,
      "loss": 2.726137161254883,
      "step": 2288
    },
    {
      "epoch": 0.1856755188551361,
      "grad_norm": 0.026335610076785088,
      "learning_rate": 0.0012,
      "loss": 2.7295632362365723,
      "step": 2290
    },
    {
      "epoch": 0.18583768088033709,
      "grad_norm": 0.022120025008916855,
      "learning_rate": 0.0012,
      "loss": 2.708770751953125,
      "step": 2292
    },
    {
      "epoch": 0.1859998429055381,
      "grad_norm": 0.03553251922130585,
      "learning_rate": 0.0012,
      "loss": 2.692206859588623,
      "step": 2294
    },
    {
      "epoch": 0.1861620049307391,
      "grad_norm": 0.04490605741739273,
      "learning_rate": 0.0012,
      "loss": 2.7292141914367676,
      "step": 2296
    },
    {
      "epoch": 0.18632416695594006,
      "grad_norm": 0.05324702709913254,
      "learning_rate": 0.0012,
      "loss": 2.7091283798217773,
      "step": 2298
    },
    {
      "epoch": 0.18648632898114106,
      "grad_norm": 0.02801034040749073,
      "learning_rate": 0.0012,
      "loss": 2.7173850536346436,
      "step": 2300
    },
    {
      "epoch": 0.18664849100634207,
      "grad_norm": 0.041019752621650696,
      "learning_rate": 0.0012,
      "loss": 2.6840691566467285,
      "step": 2302
    },
    {
      "epoch": 0.18681065303154304,
      "grad_norm": 0.04209590703248978,
      "learning_rate": 0.0012,
      "loss": 2.766345739364624,
      "step": 2304
    },
    {
      "epoch": 0.18697281505674404,
      "grad_norm": 0.04437068849802017,
      "learning_rate": 0.0012,
      "loss": 2.7219290733337402,
      "step": 2306
    },
    {
      "epoch": 0.18713497708194504,
      "grad_norm": 0.024695688858628273,
      "learning_rate": 0.0012,
      "loss": 2.6887941360473633,
      "step": 2308
    },
    {
      "epoch": 0.18729713910714602,
      "grad_norm": 0.038547880947589874,
      "learning_rate": 0.0012,
      "loss": 2.6336350440979004,
      "step": 2310
    },
    {
      "epoch": 0.18745930113234702,
      "grad_norm": 0.06798165291547775,
      "learning_rate": 0.0012,
      "loss": 2.7017767429351807,
      "step": 2312
    },
    {
      "epoch": 0.18762146315754802,
      "grad_norm": 0.04238548129796982,
      "learning_rate": 0.0012,
      "loss": 2.7292282581329346,
      "step": 2314
    },
    {
      "epoch": 0.187783625182749,
      "grad_norm": 0.03862670436501503,
      "learning_rate": 0.0012,
      "loss": 2.720640182495117,
      "step": 2316
    },
    {
      "epoch": 0.18794578720795,
      "grad_norm": 0.038994766771793365,
      "learning_rate": 0.0012,
      "loss": 2.6756162643432617,
      "step": 2318
    },
    {
      "epoch": 0.188107949233151,
      "grad_norm": 0.03237957879900932,
      "learning_rate": 0.0012,
      "loss": 2.748821258544922,
      "step": 2320
    },
    {
      "epoch": 0.18827011125835197,
      "grad_norm": 0.02992452308535576,
      "learning_rate": 0.0012,
      "loss": 2.724195957183838,
      "step": 2322
    },
    {
      "epoch": 0.18843227328355297,
      "grad_norm": 0.02556726522743702,
      "learning_rate": 0.0012,
      "loss": 2.687242269515991,
      "step": 2324
    },
    {
      "epoch": 0.18859443530875397,
      "grad_norm": 0.0561026893556118,
      "learning_rate": 0.0012,
      "loss": 2.7235071659088135,
      "step": 2326
    },
    {
      "epoch": 0.18875659733395495,
      "grad_norm": 0.058455098420381546,
      "learning_rate": 0.0012,
      "loss": 2.754371404647827,
      "step": 2328
    },
    {
      "epoch": 0.18891875935915595,
      "grad_norm": 0.02912844531238079,
      "learning_rate": 0.0012,
      "loss": 2.6267857551574707,
      "step": 2330
    },
    {
      "epoch": 0.18908092138435695,
      "grad_norm": 0.02365407906472683,
      "learning_rate": 0.0012,
      "loss": 2.6957311630249023,
      "step": 2332
    },
    {
      "epoch": 0.18924308340955792,
      "grad_norm": 0.026833439245820045,
      "learning_rate": 0.0012,
      "loss": 2.7038841247558594,
      "step": 2334
    },
    {
      "epoch": 0.18940524543475892,
      "grad_norm": 0.023421816527843475,
      "learning_rate": 0.0012,
      "loss": 2.776146411895752,
      "step": 2336
    },
    {
      "epoch": 0.18956740745995992,
      "grad_norm": 0.037446215748786926,
      "learning_rate": 0.0012,
      "loss": 2.7220301628112793,
      "step": 2338
    },
    {
      "epoch": 0.1897295694851609,
      "grad_norm": 0.05852378159761429,
      "learning_rate": 0.0012,
      "loss": 2.6833693981170654,
      "step": 2340
    },
    {
      "epoch": 0.1898917315103619,
      "grad_norm": 0.04491305351257324,
      "learning_rate": 0.0012,
      "loss": 2.687333583831787,
      "step": 2342
    },
    {
      "epoch": 0.1900538935355629,
      "grad_norm": 0.02369772456586361,
      "learning_rate": 0.0012,
      "loss": 2.7633719444274902,
      "step": 2344
    },
    {
      "epoch": 0.19021605556076387,
      "grad_norm": 0.07714489847421646,
      "learning_rate": 0.0012,
      "loss": 2.6891937255859375,
      "step": 2346
    },
    {
      "epoch": 0.19037821758596488,
      "grad_norm": 0.0445769838988781,
      "learning_rate": 0.0012,
      "loss": 2.674459457397461,
      "step": 2348
    },
    {
      "epoch": 0.19054037961116588,
      "grad_norm": 0.033622387796640396,
      "learning_rate": 0.0012,
      "loss": 2.683220624923706,
      "step": 2350
    },
    {
      "epoch": 0.19070254163636685,
      "grad_norm": 0.06959934532642365,
      "learning_rate": 0.0012,
      "loss": 2.6818289756774902,
      "step": 2352
    },
    {
      "epoch": 0.19086470366156785,
      "grad_norm": 0.061595212668180466,
      "learning_rate": 0.0012,
      "loss": 2.7803821563720703,
      "step": 2354
    },
    {
      "epoch": 0.19102686568676885,
      "grad_norm": 0.02779592014849186,
      "learning_rate": 0.0012,
      "loss": 2.7055504322052,
      "step": 2356
    },
    {
      "epoch": 0.19118902771196983,
      "grad_norm": 0.04706176742911339,
      "learning_rate": 0.0012,
      "loss": 2.738804817199707,
      "step": 2358
    },
    {
      "epoch": 0.19135118973717083,
      "grad_norm": 0.05392534285783768,
      "learning_rate": 0.0012,
      "loss": 2.690220832824707,
      "step": 2360
    },
    {
      "epoch": 0.19151335176237183,
      "grad_norm": 0.022804994136095047,
      "learning_rate": 0.0012,
      "loss": 2.7363224029541016,
      "step": 2362
    },
    {
      "epoch": 0.1916755137875728,
      "grad_norm": 0.040942393243312836,
      "learning_rate": 0.0012,
      "loss": 2.708456516265869,
      "step": 2364
    },
    {
      "epoch": 0.1918376758127738,
      "grad_norm": 0.023316146805882454,
      "learning_rate": 0.0012,
      "loss": 2.6905875205993652,
      "step": 2366
    },
    {
      "epoch": 0.1919998378379748,
      "grad_norm": 0.021588560193777084,
      "learning_rate": 0.0012,
      "loss": 2.7608938217163086,
      "step": 2368
    },
    {
      "epoch": 0.19216199986317578,
      "grad_norm": 0.02619745023548603,
      "learning_rate": 0.0012,
      "loss": 2.686023235321045,
      "step": 2370
    },
    {
      "epoch": 0.19232416188837678,
      "grad_norm": 0.0225941501557827,
      "learning_rate": 0.0012,
      "loss": 2.7111093997955322,
      "step": 2372
    },
    {
      "epoch": 0.19248632391357778,
      "grad_norm": 0.02340855449438095,
      "learning_rate": 0.0012,
      "loss": 2.721397876739502,
      "step": 2374
    },
    {
      "epoch": 0.19264848593877876,
      "grad_norm": 0.03818884864449501,
      "learning_rate": 0.0012,
      "loss": 2.7324421405792236,
      "step": 2376
    },
    {
      "epoch": 0.19281064796397976,
      "grad_norm": 0.023415112867951393,
      "learning_rate": 0.0012,
      "loss": 2.7831122875213623,
      "step": 2378
    },
    {
      "epoch": 0.19297280998918076,
      "grad_norm": 0.03956650197505951,
      "learning_rate": 0.0012,
      "loss": 2.706190824508667,
      "step": 2380
    },
    {
      "epoch": 0.19313497201438173,
      "grad_norm": 0.022519247606396675,
      "learning_rate": 0.0012,
      "loss": 2.79661226272583,
      "step": 2382
    },
    {
      "epoch": 0.19329713403958274,
      "grad_norm": 0.028169704601168633,
      "learning_rate": 0.0012,
      "loss": 2.70712947845459,
      "step": 2384
    },
    {
      "epoch": 0.19345929606478374,
      "grad_norm": 0.028596369549632072,
      "learning_rate": 0.0012,
      "loss": 2.6207518577575684,
      "step": 2386
    },
    {
      "epoch": 0.1936214580899847,
      "grad_norm": 0.05694541335105896,
      "learning_rate": 0.0012,
      "loss": 2.725215196609497,
      "step": 2388
    },
    {
      "epoch": 0.1937836201151857,
      "grad_norm": 0.04210190102458,
      "learning_rate": 0.0012,
      "loss": 2.682623863220215,
      "step": 2390
    },
    {
      "epoch": 0.1939457821403867,
      "grad_norm": 0.0259614996612072,
      "learning_rate": 0.0012,
      "loss": 2.708599805831909,
      "step": 2392
    },
    {
      "epoch": 0.1941079441655877,
      "grad_norm": 0.029756246134638786,
      "learning_rate": 0.0012,
      "loss": 2.7137880325317383,
      "step": 2394
    },
    {
      "epoch": 0.1942701061907887,
      "grad_norm": 0.027375297620892525,
      "learning_rate": 0.0012,
      "loss": 2.7028276920318604,
      "step": 2396
    },
    {
      "epoch": 0.1944322682159897,
      "grad_norm": 0.027150513604283333,
      "learning_rate": 0.0012,
      "loss": 2.79002046585083,
      "step": 2398
    },
    {
      "epoch": 0.19459443024119066,
      "grad_norm": 0.02847253531217575,
      "learning_rate": 0.0012,
      "loss": 2.710066795349121,
      "step": 2400
    },
    {
      "epoch": 0.19475659226639166,
      "grad_norm": 0.03619396686553955,
      "learning_rate": 0.0012,
      "loss": 2.6948206424713135,
      "step": 2402
    },
    {
      "epoch": 0.19491875429159267,
      "grad_norm": 0.022893154993653297,
      "learning_rate": 0.0012,
      "loss": 2.7043795585632324,
      "step": 2404
    },
    {
      "epoch": 0.19508091631679364,
      "grad_norm": 0.0416647270321846,
      "learning_rate": 0.0012,
      "loss": 2.737370491027832,
      "step": 2406
    },
    {
      "epoch": 0.19524307834199464,
      "grad_norm": 0.025717176496982574,
      "learning_rate": 0.0012,
      "loss": 2.7711312770843506,
      "step": 2408
    },
    {
      "epoch": 0.19540524036719564,
      "grad_norm": 0.03783318027853966,
      "learning_rate": 0.0012,
      "loss": 2.621974468231201,
      "step": 2410
    },
    {
      "epoch": 0.19556740239239662,
      "grad_norm": 0.02389744110405445,
      "learning_rate": 0.0012,
      "loss": 2.6908316612243652,
      "step": 2412
    },
    {
      "epoch": 0.19572956441759762,
      "grad_norm": 0.03455303609371185,
      "learning_rate": 0.0012,
      "loss": 2.737456798553467,
      "step": 2414
    },
    {
      "epoch": 0.19589172644279862,
      "grad_norm": 0.025888636708259583,
      "learning_rate": 0.0012,
      "loss": 2.7588393688201904,
      "step": 2416
    },
    {
      "epoch": 0.1960538884679996,
      "grad_norm": 0.029050888493657112,
      "learning_rate": 0.0012,
      "loss": 2.708723545074463,
      "step": 2418
    },
    {
      "epoch": 0.1962160504932006,
      "grad_norm": 0.03700469806790352,
      "learning_rate": 0.0012,
      "loss": 2.6878514289855957,
      "step": 2420
    },
    {
      "epoch": 0.1963782125184016,
      "grad_norm": 0.03047638013958931,
      "learning_rate": 0.0012,
      "loss": 2.696800708770752,
      "step": 2422
    },
    {
      "epoch": 0.19654037454360257,
      "grad_norm": 0.028901295736432076,
      "learning_rate": 0.0012,
      "loss": 2.7178854942321777,
      "step": 2424
    },
    {
      "epoch": 0.19670253656880357,
      "grad_norm": 0.03014433942735195,
      "learning_rate": 0.0012,
      "loss": 2.7439560890197754,
      "step": 2426
    },
    {
      "epoch": 0.19686469859400457,
      "grad_norm": 0.02426535077393055,
      "learning_rate": 0.0012,
      "loss": 2.761549472808838,
      "step": 2428
    },
    {
      "epoch": 0.19702686061920555,
      "grad_norm": 0.03823578357696533,
      "learning_rate": 0.0012,
      "loss": 2.6896286010742188,
      "step": 2430
    },
    {
      "epoch": 0.19718902264440655,
      "grad_norm": 0.05330615118145943,
      "learning_rate": 0.0012,
      "loss": 2.7278168201446533,
      "step": 2432
    },
    {
      "epoch": 0.19735118466960755,
      "grad_norm": 0.0746062621474266,
      "learning_rate": 0.0012,
      "loss": 2.7256150245666504,
      "step": 2434
    },
    {
      "epoch": 0.19751334669480852,
      "grad_norm": 0.06512253731489182,
      "learning_rate": 0.0012,
      "loss": 2.724959135055542,
      "step": 2436
    },
    {
      "epoch": 0.19767550872000952,
      "grad_norm": 0.02460281364619732,
      "learning_rate": 0.0012,
      "loss": 2.7640507221221924,
      "step": 2438
    },
    {
      "epoch": 0.19783767074521053,
      "grad_norm": 0.03540654480457306,
      "learning_rate": 0.0012,
      "loss": 2.680037021636963,
      "step": 2440
    },
    {
      "epoch": 0.1979998327704115,
      "grad_norm": 0.039146460592746735,
      "learning_rate": 0.0012,
      "loss": 2.6704583168029785,
      "step": 2442
    },
    {
      "epoch": 0.1981619947956125,
      "grad_norm": 0.0587041862308979,
      "learning_rate": 0.0012,
      "loss": 2.712947368621826,
      "step": 2444
    },
    {
      "epoch": 0.1983241568208135,
      "grad_norm": 0.033042680472135544,
      "learning_rate": 0.0012,
      "loss": 2.7127060890197754,
      "step": 2446
    },
    {
      "epoch": 0.19848631884601448,
      "grad_norm": 0.025407694280147552,
      "learning_rate": 0.0012,
      "loss": 2.688307285308838,
      "step": 2448
    },
    {
      "epoch": 0.19864848087121548,
      "grad_norm": 0.024788381531834602,
      "learning_rate": 0.0012,
      "loss": 2.7294816970825195,
      "step": 2450
    },
    {
      "epoch": 0.19881064289641648,
      "grad_norm": 0.02223445475101471,
      "learning_rate": 0.0012,
      "loss": 2.74395489692688,
      "step": 2452
    },
    {
      "epoch": 0.19897280492161745,
      "grad_norm": 0.02615446411073208,
      "learning_rate": 0.0012,
      "loss": 2.7185354232788086,
      "step": 2454
    },
    {
      "epoch": 0.19913496694681845,
      "grad_norm": 0.02585953287780285,
      "learning_rate": 0.0012,
      "loss": 2.7531375885009766,
      "step": 2456
    },
    {
      "epoch": 0.19929712897201945,
      "grad_norm": 0.046702634543180466,
      "learning_rate": 0.0012,
      "loss": 2.607008934020996,
      "step": 2458
    },
    {
      "epoch": 0.19945929099722043,
      "grad_norm": 0.031049257144331932,
      "learning_rate": 0.0012,
      "loss": 2.7247965335845947,
      "step": 2460
    },
    {
      "epoch": 0.19962145302242143,
      "grad_norm": 0.023124296218156815,
      "learning_rate": 0.0012,
      "loss": 2.673142433166504,
      "step": 2462
    },
    {
      "epoch": 0.19978361504762243,
      "grad_norm": 0.023175545036792755,
      "learning_rate": 0.0012,
      "loss": 2.6904091835021973,
      "step": 2464
    },
    {
      "epoch": 0.1999457770728234,
      "grad_norm": 0.05057899281382561,
      "learning_rate": 0.0012,
      "loss": 2.6929643154144287,
      "step": 2466
    },
    {
      "epoch": 0.2001079390980244,
      "grad_norm": 0.04757823422551155,
      "learning_rate": 0.0012,
      "loss": 2.6855320930480957,
      "step": 2468
    },
    {
      "epoch": 0.2002701011232254,
      "grad_norm": 0.08671563863754272,
      "learning_rate": 0.0012,
      "loss": 2.7248125076293945,
      "step": 2470
    },
    {
      "epoch": 0.20043226314842638,
      "grad_norm": 0.028200550004839897,
      "learning_rate": 0.0012,
      "loss": 2.6771535873413086,
      "step": 2472
    },
    {
      "epoch": 0.20059442517362738,
      "grad_norm": 0.030317336320877075,
      "learning_rate": 0.0012,
      "loss": 2.6856815814971924,
      "step": 2474
    },
    {
      "epoch": 0.20075658719882838,
      "grad_norm": 0.0387442521750927,
      "learning_rate": 0.0012,
      "loss": 2.682368755340576,
      "step": 2476
    },
    {
      "epoch": 0.20091874922402936,
      "grad_norm": 0.025360295549035072,
      "learning_rate": 0.0012,
      "loss": 2.715118646621704,
      "step": 2478
    },
    {
      "epoch": 0.20108091124923036,
      "grad_norm": 0.024383211508393288,
      "learning_rate": 0.0012,
      "loss": 2.743002414703369,
      "step": 2480
    },
    {
      "epoch": 0.20124307327443136,
      "grad_norm": 0.025550469756126404,
      "learning_rate": 0.0012,
      "loss": 2.7249679565429688,
      "step": 2482
    },
    {
      "epoch": 0.20140523529963233,
      "grad_norm": 0.027135765179991722,
      "learning_rate": 0.0012,
      "loss": 2.722769260406494,
      "step": 2484
    },
    {
      "epoch": 0.20156739732483334,
      "grad_norm": 0.023784229531884193,
      "learning_rate": 0.0012,
      "loss": 2.7006125450134277,
      "step": 2486
    },
    {
      "epoch": 0.20172955935003434,
      "grad_norm": 0.02704133465886116,
      "learning_rate": 0.0012,
      "loss": 2.6948506832122803,
      "step": 2488
    },
    {
      "epoch": 0.2018917213752353,
      "grad_norm": 0.024582015350461006,
      "learning_rate": 0.0012,
      "loss": 2.716273307800293,
      "step": 2490
    },
    {
      "epoch": 0.2020538834004363,
      "grad_norm": 0.05677264183759689,
      "learning_rate": 0.0012,
      "loss": 2.781299352645874,
      "step": 2492
    },
    {
      "epoch": 0.20221604542563731,
      "grad_norm": 0.04668521136045456,
      "learning_rate": 0.0012,
      "loss": 2.754725694656372,
      "step": 2494
    },
    {
      "epoch": 0.20237820745083832,
      "grad_norm": 0.028287675231695175,
      "learning_rate": 0.0012,
      "loss": 2.630161762237549,
      "step": 2496
    },
    {
      "epoch": 0.2025403694760393,
      "grad_norm": 0.0231868214905262,
      "learning_rate": 0.0012,
      "loss": 2.695427894592285,
      "step": 2498
    },
    {
      "epoch": 0.2027025315012403,
      "grad_norm": 0.03224097192287445,
      "learning_rate": 0.0012,
      "loss": 2.6387555599212646,
      "step": 2500
    },
    {
      "epoch": 0.2028646935264413,
      "grad_norm": 0.05752271041274071,
      "learning_rate": 0.0012,
      "loss": 2.663142442703247,
      "step": 2502
    },
    {
      "epoch": 0.20302685555164227,
      "grad_norm": 0.04486864060163498,
      "learning_rate": 0.0012,
      "loss": 2.7042388916015625,
      "step": 2504
    },
    {
      "epoch": 0.20318901757684327,
      "grad_norm": 0.04243026301264763,
      "learning_rate": 0.0012,
      "loss": 2.734429359436035,
      "step": 2506
    },
    {
      "epoch": 0.20335117960204427,
      "grad_norm": 0.036406684666872025,
      "learning_rate": 0.0012,
      "loss": 2.6473779678344727,
      "step": 2508
    },
    {
      "epoch": 0.20351334162724524,
      "grad_norm": 0.03085904009640217,
      "learning_rate": 0.0012,
      "loss": 2.736022472381592,
      "step": 2510
    },
    {
      "epoch": 0.20367550365244624,
      "grad_norm": 0.031782638281583786,
      "learning_rate": 0.0012,
      "loss": 2.753170967102051,
      "step": 2512
    },
    {
      "epoch": 0.20383766567764725,
      "grad_norm": 0.062040507793426514,
      "learning_rate": 0.0012,
      "loss": 2.7441039085388184,
      "step": 2514
    },
    {
      "epoch": 0.20399982770284822,
      "grad_norm": 0.03257039189338684,
      "learning_rate": 0.0012,
      "loss": 2.7329843044281006,
      "step": 2516
    },
    {
      "epoch": 0.20416198972804922,
      "grad_norm": 0.023291543126106262,
      "learning_rate": 0.0012,
      "loss": 2.737569808959961,
      "step": 2518
    },
    {
      "epoch": 0.20432415175325022,
      "grad_norm": 0.029739869758486748,
      "learning_rate": 0.0012,
      "loss": 2.695953369140625,
      "step": 2520
    },
    {
      "epoch": 0.2044863137784512,
      "grad_norm": 0.04321277141571045,
      "learning_rate": 0.0012,
      "loss": 2.6584630012512207,
      "step": 2522
    },
    {
      "epoch": 0.2046484758036522,
      "grad_norm": 0.04372958466410637,
      "learning_rate": 0.0012,
      "loss": 2.714160442352295,
      "step": 2524
    },
    {
      "epoch": 0.2048106378288532,
      "grad_norm": 0.03152601420879364,
      "learning_rate": 0.0012,
      "loss": 2.6290335655212402,
      "step": 2526
    },
    {
      "epoch": 0.20497279985405417,
      "grad_norm": 0.02332163229584694,
      "learning_rate": 0.0012,
      "loss": 2.721635341644287,
      "step": 2528
    },
    {
      "epoch": 0.20513496187925517,
      "grad_norm": 0.030069833621382713,
      "learning_rate": 0.0012,
      "loss": 2.7168922424316406,
      "step": 2530
    },
    {
      "epoch": 0.20529712390445617,
      "grad_norm": 0.0342598482966423,
      "learning_rate": 0.0012,
      "loss": 2.653258800506592,
      "step": 2532
    },
    {
      "epoch": 0.20545928592965715,
      "grad_norm": 0.027472175657749176,
      "learning_rate": 0.0012,
      "loss": 2.6650328636169434,
      "step": 2534
    },
    {
      "epoch": 0.20562144795485815,
      "grad_norm": 0.039559200406074524,
      "learning_rate": 0.0012,
      "loss": 2.7032063007354736,
      "step": 2536
    },
    {
      "epoch": 0.20578360998005915,
      "grad_norm": 0.031412724405527115,
      "learning_rate": 0.0012,
      "loss": 2.735264301300049,
      "step": 2538
    },
    {
      "epoch": 0.20594577200526012,
      "grad_norm": 0.02298039011657238,
      "learning_rate": 0.0012,
      "loss": 2.6440324783325195,
      "step": 2540
    },
    {
      "epoch": 0.20610793403046113,
      "grad_norm": 0.025743596255779266,
      "learning_rate": 0.0012,
      "loss": 2.6853222846984863,
      "step": 2542
    },
    {
      "epoch": 0.20627009605566213,
      "grad_norm": 0.04747503995895386,
      "learning_rate": 0.0012,
      "loss": 2.768951892852783,
      "step": 2544
    },
    {
      "epoch": 0.2064322580808631,
      "grad_norm": 0.03576238080859184,
      "learning_rate": 0.0012,
      "loss": 2.7182693481445312,
      "step": 2546
    },
    {
      "epoch": 0.2065944201060641,
      "grad_norm": 0.04720182344317436,
      "learning_rate": 0.0012,
      "loss": 2.721086263656616,
      "step": 2548
    },
    {
      "epoch": 0.2067565821312651,
      "grad_norm": 0.04899141564965248,
      "learning_rate": 0.0012,
      "loss": 2.7513370513916016,
      "step": 2550
    },
    {
      "epoch": 0.20691874415646608,
      "grad_norm": 0.025994611904025078,
      "learning_rate": 0.0012,
      "loss": 2.7479939460754395,
      "step": 2552
    },
    {
      "epoch": 0.20708090618166708,
      "grad_norm": 0.043324049562215805,
      "learning_rate": 0.0012,
      "loss": 2.6732876300811768,
      "step": 2554
    },
    {
      "epoch": 0.20724306820686808,
      "grad_norm": 0.07680746912956238,
      "learning_rate": 0.0012,
      "loss": 2.7565455436706543,
      "step": 2556
    },
    {
      "epoch": 0.20740523023206905,
      "grad_norm": 0.06900375336408615,
      "learning_rate": 0.0012,
      "loss": 2.700136184692383,
      "step": 2558
    },
    {
      "epoch": 0.20756739225727006,
      "grad_norm": 0.02754143998026848,
      "learning_rate": 0.0012,
      "loss": 2.6541402339935303,
      "step": 2560
    },
    {
      "epoch": 0.20772955428247106,
      "grad_norm": 0.06890005618333817,
      "learning_rate": 0.0012,
      "loss": 2.65523099899292,
      "step": 2562
    },
    {
      "epoch": 0.20789171630767203,
      "grad_norm": 0.06096918508410454,
      "learning_rate": 0.0012,
      "loss": 2.746666431427002,
      "step": 2564
    },
    {
      "epoch": 0.20805387833287303,
      "grad_norm": 0.028222884982824326,
      "learning_rate": 0.0012,
      "loss": 2.70015287399292,
      "step": 2566
    },
    {
      "epoch": 0.20821604035807403,
      "grad_norm": 0.05295107513666153,
      "learning_rate": 0.0012,
      "loss": 2.815093755722046,
      "step": 2568
    },
    {
      "epoch": 0.208378202383275,
      "grad_norm": 0.025558985769748688,
      "learning_rate": 0.0012,
      "loss": 2.7026915550231934,
      "step": 2570
    },
    {
      "epoch": 0.208540364408476,
      "grad_norm": 0.03220168501138687,
      "learning_rate": 0.0012,
      "loss": 2.7710213661193848,
      "step": 2572
    },
    {
      "epoch": 0.208702526433677,
      "grad_norm": 0.030459081754088402,
      "learning_rate": 0.0012,
      "loss": 2.7555150985717773,
      "step": 2574
    },
    {
      "epoch": 0.20886468845887798,
      "grad_norm": 0.029777774587273598,
      "learning_rate": 0.0012,
      "loss": 2.740530014038086,
      "step": 2576
    },
    {
      "epoch": 0.20902685048407899,
      "grad_norm": 0.03175332024693489,
      "learning_rate": 0.0012,
      "loss": 2.6905908584594727,
      "step": 2578
    },
    {
      "epoch": 0.20918901250928,
      "grad_norm": 0.030048001557588577,
      "learning_rate": 0.0012,
      "loss": 2.695611000061035,
      "step": 2580
    },
    {
      "epoch": 0.20935117453448096,
      "grad_norm": 0.034684717655181885,
      "learning_rate": 0.0012,
      "loss": 2.696147918701172,
      "step": 2582
    },
    {
      "epoch": 0.20951333655968196,
      "grad_norm": 0.034506820142269135,
      "learning_rate": 0.0012,
      "loss": 2.716726779937744,
      "step": 2584
    },
    {
      "epoch": 0.20967549858488296,
      "grad_norm": 0.023879745975136757,
      "learning_rate": 0.0012,
      "loss": 2.714205026626587,
      "step": 2586
    },
    {
      "epoch": 0.20983766061008394,
      "grad_norm": 0.022438710555434227,
      "learning_rate": 0.0012,
      "loss": 2.8331093788146973,
      "step": 2588
    },
    {
      "epoch": 0.20999982263528494,
      "grad_norm": 0.044555261731147766,
      "learning_rate": 0.0012,
      "loss": 2.683816909790039,
      "step": 2590
    },
    {
      "epoch": 0.21016198466048594,
      "grad_norm": 0.06211457774043083,
      "learning_rate": 0.0012,
      "loss": 2.683915853500366,
      "step": 2592
    },
    {
      "epoch": 0.2103241466856869,
      "grad_norm": 0.022729013115167618,
      "learning_rate": 0.0012,
      "loss": 2.664198637008667,
      "step": 2594
    },
    {
      "epoch": 0.21048630871088791,
      "grad_norm": 0.03142024576663971,
      "learning_rate": 0.0012,
      "loss": 2.6953113079071045,
      "step": 2596
    },
    {
      "epoch": 0.21064847073608892,
      "grad_norm": 0.028472406789660454,
      "learning_rate": 0.0012,
      "loss": 2.7266957759857178,
      "step": 2598
    },
    {
      "epoch": 0.2108106327612899,
      "grad_norm": 0.030940108001232147,
      "learning_rate": 0.0012,
      "loss": 2.670102119445801,
      "step": 2600
    },
    {
      "epoch": 0.2109727947864909,
      "grad_norm": 0.029938235878944397,
      "learning_rate": 0.0012,
      "loss": 2.679610252380371,
      "step": 2602
    },
    {
      "epoch": 0.2111349568116919,
      "grad_norm": 0.024099105969071388,
      "learning_rate": 0.0012,
      "loss": 2.666499137878418,
      "step": 2604
    },
    {
      "epoch": 0.21129711883689287,
      "grad_norm": 0.03489365428686142,
      "learning_rate": 0.0012,
      "loss": 2.673786163330078,
      "step": 2606
    },
    {
      "epoch": 0.21145928086209387,
      "grad_norm": 0.022393032908439636,
      "learning_rate": 0.0012,
      "loss": 2.754635810852051,
      "step": 2608
    },
    {
      "epoch": 0.21162144288729487,
      "grad_norm": 0.024563265964388847,
      "learning_rate": 0.0012,
      "loss": 2.6340718269348145,
      "step": 2610
    },
    {
      "epoch": 0.21178360491249584,
      "grad_norm": 0.029067538678646088,
      "learning_rate": 0.0012,
      "loss": 2.7172279357910156,
      "step": 2612
    },
    {
      "epoch": 0.21194576693769684,
      "grad_norm": 0.06105156987905502,
      "learning_rate": 0.0012,
      "loss": 2.7226691246032715,
      "step": 2614
    },
    {
      "epoch": 0.21210792896289785,
      "grad_norm": 0.05807894840836525,
      "learning_rate": 0.0012,
      "loss": 2.690627098083496,
      "step": 2616
    },
    {
      "epoch": 0.21227009098809882,
      "grad_norm": 0.025796793401241302,
      "learning_rate": 0.0012,
      "loss": 2.7728779315948486,
      "step": 2618
    },
    {
      "epoch": 0.21243225301329982,
      "grad_norm": 0.04237738251686096,
      "learning_rate": 0.0012,
      "loss": 2.6732311248779297,
      "step": 2620
    },
    {
      "epoch": 0.21259441503850082,
      "grad_norm": 0.049213558435440063,
      "learning_rate": 0.0012,
      "loss": 2.6890478134155273,
      "step": 2622
    },
    {
      "epoch": 0.2127565770637018,
      "grad_norm": 0.05468490347266197,
      "learning_rate": 0.0012,
      "loss": 2.7291975021362305,
      "step": 2624
    },
    {
      "epoch": 0.2129187390889028,
      "grad_norm": 0.023097500205039978,
      "learning_rate": 0.0012,
      "loss": 2.686221122741699,
      "step": 2626
    },
    {
      "epoch": 0.2130809011141038,
      "grad_norm": 0.027266204357147217,
      "learning_rate": 0.0012,
      "loss": 2.658346176147461,
      "step": 2628
    },
    {
      "epoch": 0.21324306313930477,
      "grad_norm": 0.04112926498055458,
      "learning_rate": 0.0012,
      "loss": 2.7069075107574463,
      "step": 2630
    },
    {
      "epoch": 0.21340522516450577,
      "grad_norm": 0.03029409423470497,
      "learning_rate": 0.0012,
      "loss": 2.711606740951538,
      "step": 2632
    },
    {
      "epoch": 0.21356738718970678,
      "grad_norm": 0.03548325225710869,
      "learning_rate": 0.0012,
      "loss": 2.755829334259033,
      "step": 2634
    },
    {
      "epoch": 0.21372954921490775,
      "grad_norm": 0.02522600255906582,
      "learning_rate": 0.0012,
      "loss": 2.734246253967285,
      "step": 2636
    },
    {
      "epoch": 0.21389171124010875,
      "grad_norm": 0.025410674512386322,
      "learning_rate": 0.0012,
      "loss": 2.718306541442871,
      "step": 2638
    },
    {
      "epoch": 0.21405387326530975,
      "grad_norm": 0.04756923392415047,
      "learning_rate": 0.0012,
      "loss": 2.6377596855163574,
      "step": 2640
    },
    {
      "epoch": 0.21421603529051073,
      "grad_norm": 0.06601852923631668,
      "learning_rate": 0.0012,
      "loss": 2.711369752883911,
      "step": 2642
    },
    {
      "epoch": 0.21437819731571173,
      "grad_norm": 0.037763796746730804,
      "learning_rate": 0.0012,
      "loss": 2.6998214721679688,
      "step": 2644
    },
    {
      "epoch": 0.21454035934091273,
      "grad_norm": 0.026870662346482277,
      "learning_rate": 0.0012,
      "loss": 2.713142156600952,
      "step": 2646
    },
    {
      "epoch": 0.2147025213661137,
      "grad_norm": 0.0419919490814209,
      "learning_rate": 0.0012,
      "loss": 2.7699320316314697,
      "step": 2648
    },
    {
      "epoch": 0.2148646833913147,
      "grad_norm": 0.0669235959649086,
      "learning_rate": 0.0012,
      "loss": 2.758902072906494,
      "step": 2650
    },
    {
      "epoch": 0.2150268454165157,
      "grad_norm": 0.023727647960186005,
      "learning_rate": 0.0012,
      "loss": 2.666785955429077,
      "step": 2652
    },
    {
      "epoch": 0.21518900744171668,
      "grad_norm": 0.03400159999728203,
      "learning_rate": 0.0012,
      "loss": 2.632807731628418,
      "step": 2654
    },
    {
      "epoch": 0.21535116946691768,
      "grad_norm": 0.04031413048505783,
      "learning_rate": 0.0012,
      "loss": 2.6994757652282715,
      "step": 2656
    },
    {
      "epoch": 0.21551333149211868,
      "grad_norm": 0.03916064277291298,
      "learning_rate": 0.0012,
      "loss": 2.7688050270080566,
      "step": 2658
    },
    {
      "epoch": 0.21567549351731966,
      "grad_norm": 0.02631252631545067,
      "learning_rate": 0.0012,
      "loss": 2.708001136779785,
      "step": 2660
    },
    {
      "epoch": 0.21583765554252066,
      "grad_norm": 0.030986202880740166,
      "learning_rate": 0.0012,
      "loss": 2.6988372802734375,
      "step": 2662
    },
    {
      "epoch": 0.21599981756772166,
      "grad_norm": 0.02321590483188629,
      "learning_rate": 0.0012,
      "loss": 2.7433319091796875,
      "step": 2664
    },
    {
      "epoch": 0.21616197959292263,
      "grad_norm": 0.026086371392011642,
      "learning_rate": 0.0012,
      "loss": 2.6715688705444336,
      "step": 2666
    },
    {
      "epoch": 0.21632414161812363,
      "grad_norm": 0.04720166325569153,
      "learning_rate": 0.0012,
      "loss": 2.715719223022461,
      "step": 2668
    },
    {
      "epoch": 0.21648630364332463,
      "grad_norm": 0.05205851420760155,
      "learning_rate": 0.0012,
      "loss": 2.675518035888672,
      "step": 2670
    },
    {
      "epoch": 0.2166484656685256,
      "grad_norm": 0.02583330310881138,
      "learning_rate": 0.0012,
      "loss": 2.7759571075439453,
      "step": 2672
    },
    {
      "epoch": 0.2168106276937266,
      "grad_norm": 0.02949129417538643,
      "learning_rate": 0.0012,
      "loss": 2.7249627113342285,
      "step": 2674
    },
    {
      "epoch": 0.2169727897189276,
      "grad_norm": 0.04117889702320099,
      "learning_rate": 0.0012,
      "loss": 2.6567580699920654,
      "step": 2676
    },
    {
      "epoch": 0.21713495174412858,
      "grad_norm": 0.04847737029194832,
      "learning_rate": 0.0012,
      "loss": 2.665503740310669,
      "step": 2678
    },
    {
      "epoch": 0.2172971137693296,
      "grad_norm": 0.042472898960113525,
      "learning_rate": 0.0012,
      "loss": 2.683225154876709,
      "step": 2680
    },
    {
      "epoch": 0.2174592757945306,
      "grad_norm": 0.023585597053170204,
      "learning_rate": 0.0012,
      "loss": 2.6844992637634277,
      "step": 2682
    },
    {
      "epoch": 0.21762143781973156,
      "grad_norm": 0.02943081222474575,
      "learning_rate": 0.0012,
      "loss": 2.678370952606201,
      "step": 2684
    },
    {
      "epoch": 0.21778359984493256,
      "grad_norm": 0.034338150173425674,
      "learning_rate": 0.0012,
      "loss": 2.6282172203063965,
      "step": 2686
    },
    {
      "epoch": 0.21794576187013356,
      "grad_norm": 0.04840384051203728,
      "learning_rate": 0.0012,
      "loss": 2.7766611576080322,
      "step": 2688
    },
    {
      "epoch": 0.21810792389533454,
      "grad_norm": 0.026890872046351433,
      "learning_rate": 0.0012,
      "loss": 2.7719829082489014,
      "step": 2690
    },
    {
      "epoch": 0.21827008592053554,
      "grad_norm": 0.02696947008371353,
      "learning_rate": 0.0012,
      "loss": 2.703124523162842,
      "step": 2692
    },
    {
      "epoch": 0.21843224794573654,
      "grad_norm": 0.023366134613752365,
      "learning_rate": 0.0012,
      "loss": 2.709270477294922,
      "step": 2694
    },
    {
      "epoch": 0.21859440997093751,
      "grad_norm": 0.024939343333244324,
      "learning_rate": 0.0012,
      "loss": 2.686697483062744,
      "step": 2696
    },
    {
      "epoch": 0.21875657199613852,
      "grad_norm": 0.023228449746966362,
      "learning_rate": 0.0012,
      "loss": 2.7200522422790527,
      "step": 2698
    },
    {
      "epoch": 0.21891873402133952,
      "grad_norm": 0.044245243072509766,
      "learning_rate": 0.0012,
      "loss": 2.6799399852752686,
      "step": 2700
    },
    {
      "epoch": 0.2190808960465405,
      "grad_norm": 0.061103105545043945,
      "learning_rate": 0.0012,
      "loss": 2.645995855331421,
      "step": 2702
    },
    {
      "epoch": 0.2192430580717415,
      "grad_norm": 0.04153219237923622,
      "learning_rate": 0.0012,
      "loss": 2.652069091796875,
      "step": 2704
    },
    {
      "epoch": 0.2194052200969425,
      "grad_norm": 0.046631090342998505,
      "learning_rate": 0.0012,
      "loss": 2.7270891666412354,
      "step": 2706
    },
    {
      "epoch": 0.21956738212214347,
      "grad_norm": 0.026760362088680267,
      "learning_rate": 0.0012,
      "loss": 2.6457133293151855,
      "step": 2708
    },
    {
      "epoch": 0.21972954414734447,
      "grad_norm": 0.024516526609659195,
      "learning_rate": 0.0012,
      "loss": 2.683292865753174,
      "step": 2710
    },
    {
      "epoch": 0.21989170617254547,
      "grad_norm": 0.03695599362254143,
      "learning_rate": 0.0012,
      "loss": 2.7303619384765625,
      "step": 2712
    },
    {
      "epoch": 0.22005386819774644,
      "grad_norm": 0.022896502166986465,
      "learning_rate": 0.0012,
      "loss": 2.7315621376037598,
      "step": 2714
    },
    {
      "epoch": 0.22021603022294745,
      "grad_norm": 0.02406260184943676,
      "learning_rate": 0.0012,
      "loss": 2.7103586196899414,
      "step": 2716
    },
    {
      "epoch": 0.22037819224814845,
      "grad_norm": 0.036936014890670776,
      "learning_rate": 0.0012,
      "loss": 2.6617655754089355,
      "step": 2718
    },
    {
      "epoch": 0.22054035427334942,
      "grad_norm": 0.02719937637448311,
      "learning_rate": 0.0012,
      "loss": 2.749484062194824,
      "step": 2720
    },
    {
      "epoch": 0.22070251629855042,
      "grad_norm": 0.024181397631764412,
      "learning_rate": 0.0012,
      "loss": 2.772202253341675,
      "step": 2722
    },
    {
      "epoch": 0.22086467832375142,
      "grad_norm": 0.03249494731426239,
      "learning_rate": 0.0012,
      "loss": 2.7056093215942383,
      "step": 2724
    },
    {
      "epoch": 0.2210268403489524,
      "grad_norm": 0.045897986739873886,
      "learning_rate": 0.0012,
      "loss": 2.715111017227173,
      "step": 2726
    },
    {
      "epoch": 0.2211890023741534,
      "grad_norm": 0.03741440176963806,
      "learning_rate": 0.0012,
      "loss": 2.6848230361938477,
      "step": 2728
    },
    {
      "epoch": 0.2213511643993544,
      "grad_norm": 0.02519441582262516,
      "learning_rate": 0.0012,
      "loss": 2.7314388751983643,
      "step": 2730
    },
    {
      "epoch": 0.22151332642455537,
      "grad_norm": 0.03042658604681492,
      "learning_rate": 0.0012,
      "loss": 2.740631103515625,
      "step": 2732
    },
    {
      "epoch": 0.22167548844975637,
      "grad_norm": 0.034313466399908066,
      "learning_rate": 0.0012,
      "loss": 2.7267117500305176,
      "step": 2734
    },
    {
      "epoch": 0.22183765047495738,
      "grad_norm": 0.04769130051136017,
      "learning_rate": 0.0012,
      "loss": 2.6804144382476807,
      "step": 2736
    },
    {
      "epoch": 0.22199981250015835,
      "grad_norm": 0.03859303519129753,
      "learning_rate": 0.0012,
      "loss": 2.731602907180786,
      "step": 2738
    },
    {
      "epoch": 0.22216197452535935,
      "grad_norm": 0.028201395645737648,
      "learning_rate": 0.0012,
      "loss": 2.664102554321289,
      "step": 2740
    },
    {
      "epoch": 0.22232413655056035,
      "grad_norm": 0.026347897946834564,
      "learning_rate": 0.0012,
      "loss": 2.8045904636383057,
      "step": 2742
    },
    {
      "epoch": 0.22248629857576133,
      "grad_norm": 0.02340748906135559,
      "learning_rate": 0.0012,
      "loss": 2.649343967437744,
      "step": 2744
    },
    {
      "epoch": 0.22264846060096233,
      "grad_norm": 0.02434348315000534,
      "learning_rate": 0.0012,
      "loss": 2.7041306495666504,
      "step": 2746
    },
    {
      "epoch": 0.22281062262616333,
      "grad_norm": 0.03234028443694115,
      "learning_rate": 0.0012,
      "loss": 2.698368549346924,
      "step": 2748
    },
    {
      "epoch": 0.2229727846513643,
      "grad_norm": 0.026559555903077126,
      "learning_rate": 0.0012,
      "loss": 2.7612149715423584,
      "step": 2750
    },
    {
      "epoch": 0.2231349466765653,
      "grad_norm": 0.07333895564079285,
      "learning_rate": 0.0012,
      "loss": 2.670485019683838,
      "step": 2752
    },
    {
      "epoch": 0.2232971087017663,
      "grad_norm": 0.050742167979478836,
      "learning_rate": 0.0012,
      "loss": 2.712620735168457,
      "step": 2754
    },
    {
      "epoch": 0.22345927072696728,
      "grad_norm": 0.023564158007502556,
      "learning_rate": 0.0012,
      "loss": 2.7691030502319336,
      "step": 2756
    },
    {
      "epoch": 0.22362143275216828,
      "grad_norm": 0.037301499396562576,
      "learning_rate": 0.0012,
      "loss": 2.8056366443634033,
      "step": 2758
    },
    {
      "epoch": 0.22378359477736928,
      "grad_norm": 0.04158635810017586,
      "learning_rate": 0.0012,
      "loss": 2.7453298568725586,
      "step": 2760
    },
    {
      "epoch": 0.22394575680257026,
      "grad_norm": 0.02942144125699997,
      "learning_rate": 0.0012,
      "loss": 2.6738457679748535,
      "step": 2762
    },
    {
      "epoch": 0.22410791882777126,
      "grad_norm": 0.02320464700460434,
      "learning_rate": 0.0012,
      "loss": 2.676687479019165,
      "step": 2764
    },
    {
      "epoch": 0.22427008085297226,
      "grad_norm": 0.029597975313663483,
      "learning_rate": 0.0012,
      "loss": 2.7044920921325684,
      "step": 2766
    },
    {
      "epoch": 0.22443224287817323,
      "grad_norm": 0.039712924510240555,
      "learning_rate": 0.0012,
      "loss": 2.7505764961242676,
      "step": 2768
    },
    {
      "epoch": 0.22459440490337423,
      "grad_norm": 0.029591361060738564,
      "learning_rate": 0.0012,
      "loss": 2.7050366401672363,
      "step": 2770
    },
    {
      "epoch": 0.22475656692857524,
      "grad_norm": 0.030782552435994148,
      "learning_rate": 0.0012,
      "loss": 2.6971611976623535,
      "step": 2772
    },
    {
      "epoch": 0.2249187289537762,
      "grad_norm": 0.07113740593194962,
      "learning_rate": 0.0012,
      "loss": 2.734320640563965,
      "step": 2774
    },
    {
      "epoch": 0.2250808909789772,
      "grad_norm": 0.03909816965460777,
      "learning_rate": 0.0012,
      "loss": 2.649656295776367,
      "step": 2776
    },
    {
      "epoch": 0.2252430530041782,
      "grad_norm": 0.057593535631895065,
      "learning_rate": 0.0012,
      "loss": 2.7272284030914307,
      "step": 2778
    },
    {
      "epoch": 0.22540521502937919,
      "grad_norm": 0.07467789202928543,
      "learning_rate": 0.0012,
      "loss": 2.702073574066162,
      "step": 2780
    },
    {
      "epoch": 0.2255673770545802,
      "grad_norm": 0.030561212450265884,
      "learning_rate": 0.0012,
      "loss": 2.7654082775115967,
      "step": 2782
    },
    {
      "epoch": 0.2257295390797812,
      "grad_norm": 0.07736421376466751,
      "learning_rate": 0.0012,
      "loss": 2.667335271835327,
      "step": 2784
    },
    {
      "epoch": 0.22589170110498216,
      "grad_norm": 0.0521356575191021,
      "learning_rate": 0.0012,
      "loss": 2.774446725845337,
      "step": 2786
    },
    {
      "epoch": 0.22605386313018316,
      "grad_norm": 0.051966696977615356,
      "learning_rate": 0.0012,
      "loss": 2.6638216972351074,
      "step": 2788
    },
    {
      "epoch": 0.22621602515538417,
      "grad_norm": 0.05819011479616165,
      "learning_rate": 0.0012,
      "loss": 2.7929792404174805,
      "step": 2790
    },
    {
      "epoch": 0.22637818718058514,
      "grad_norm": 0.02633701078593731,
      "learning_rate": 0.0012,
      "loss": 2.7190840244293213,
      "step": 2792
    },
    {
      "epoch": 0.22654034920578614,
      "grad_norm": 0.02742697484791279,
      "learning_rate": 0.0012,
      "loss": 2.675930976867676,
      "step": 2794
    },
    {
      "epoch": 0.22670251123098714,
      "grad_norm": 0.032797083258628845,
      "learning_rate": 0.0012,
      "loss": 2.6925878524780273,
      "step": 2796
    },
    {
      "epoch": 0.22686467325618812,
      "grad_norm": 0.04347538203001022,
      "learning_rate": 0.0012,
      "loss": 2.6957077980041504,
      "step": 2798
    },
    {
      "epoch": 0.22702683528138912,
      "grad_norm": 0.04053313285112381,
      "learning_rate": 0.0012,
      "loss": 2.6885266304016113,
      "step": 2800
    },
    {
      "epoch": 0.22718899730659012,
      "grad_norm": 0.02685297094285488,
      "learning_rate": 0.0012,
      "loss": 2.7097301483154297,
      "step": 2802
    },
    {
      "epoch": 0.2273511593317911,
      "grad_norm": 0.032442327588796616,
      "learning_rate": 0.0012,
      "loss": 2.7477712631225586,
      "step": 2804
    },
    {
      "epoch": 0.2275133213569921,
      "grad_norm": 0.024996532127261162,
      "learning_rate": 0.0012,
      "loss": 2.7522315979003906,
      "step": 2806
    },
    {
      "epoch": 0.2276754833821931,
      "grad_norm": 0.027762869372963905,
      "learning_rate": 0.0012,
      "loss": 2.7089598178863525,
      "step": 2808
    },
    {
      "epoch": 0.22783764540739407,
      "grad_norm": 0.06107150390744209,
      "learning_rate": 0.0012,
      "loss": 2.734276294708252,
      "step": 2810
    },
    {
      "epoch": 0.22799980743259507,
      "grad_norm": 0.025749702006578445,
      "learning_rate": 0.0012,
      "loss": 2.704011917114258,
      "step": 2812
    },
    {
      "epoch": 0.22816196945779607,
      "grad_norm": 0.02388303168118,
      "learning_rate": 0.0012,
      "loss": 2.6579625606536865,
      "step": 2814
    },
    {
      "epoch": 0.22832413148299704,
      "grad_norm": 0.029134662821888924,
      "learning_rate": 0.0012,
      "loss": 2.682537317276001,
      "step": 2816
    },
    {
      "epoch": 0.22848629350819805,
      "grad_norm": 0.04318234696984291,
      "learning_rate": 0.0012,
      "loss": 2.746854782104492,
      "step": 2818
    },
    {
      "epoch": 0.22864845553339905,
      "grad_norm": 0.027025466784834862,
      "learning_rate": 0.0012,
      "loss": 2.700883626937866,
      "step": 2820
    },
    {
      "epoch": 0.22881061755860002,
      "grad_norm": 0.031755946576595306,
      "learning_rate": 0.0012,
      "loss": 2.705829620361328,
      "step": 2822
    },
    {
      "epoch": 0.22897277958380102,
      "grad_norm": 0.02333097532391548,
      "learning_rate": 0.0012,
      "loss": 2.675741195678711,
      "step": 2824
    },
    {
      "epoch": 0.22913494160900202,
      "grad_norm": 0.033646389842033386,
      "learning_rate": 0.0012,
      "loss": 2.7212445735931396,
      "step": 2826
    },
    {
      "epoch": 0.229297103634203,
      "grad_norm": 0.03737325221300125,
      "learning_rate": 0.0012,
      "loss": 2.6549830436706543,
      "step": 2828
    },
    {
      "epoch": 0.229459265659404,
      "grad_norm": 0.02655629627406597,
      "learning_rate": 0.0012,
      "loss": 2.6546378135681152,
      "step": 2830
    },
    {
      "epoch": 0.229621427684605,
      "grad_norm": 0.04245191812515259,
      "learning_rate": 0.0012,
      "loss": 2.65451979637146,
      "step": 2832
    },
    {
      "epoch": 0.229783589709806,
      "grad_norm": 0.06096874177455902,
      "learning_rate": 0.0012,
      "loss": 2.702150821685791,
      "step": 2834
    },
    {
      "epoch": 0.22994575173500698,
      "grad_norm": 0.041071608662605286,
      "learning_rate": 0.0012,
      "loss": 2.6965019702911377,
      "step": 2836
    },
    {
      "epoch": 0.23010791376020798,
      "grad_norm": 0.02452712506055832,
      "learning_rate": 0.0012,
      "loss": 2.714177370071411,
      "step": 2838
    },
    {
      "epoch": 0.23027007578540898,
      "grad_norm": 0.03769677132368088,
      "learning_rate": 0.0012,
      "loss": 2.6800379753112793,
      "step": 2840
    },
    {
      "epoch": 0.23043223781060995,
      "grad_norm": 0.028943553566932678,
      "learning_rate": 0.0012,
      "loss": 2.658557176589966,
      "step": 2842
    },
    {
      "epoch": 0.23059439983581095,
      "grad_norm": 0.0291745588183403,
      "learning_rate": 0.0012,
      "loss": 2.6527557373046875,
      "step": 2844
    },
    {
      "epoch": 0.23075656186101196,
      "grad_norm": 0.052874624729156494,
      "learning_rate": 0.0012,
      "loss": 2.7515225410461426,
      "step": 2846
    },
    {
      "epoch": 0.23091872388621293,
      "grad_norm": 0.06321548670530319,
      "learning_rate": 0.0012,
      "loss": 2.7276034355163574,
      "step": 2848
    },
    {
      "epoch": 0.23108088591141393,
      "grad_norm": 0.034337639808654785,
      "learning_rate": 0.0012,
      "loss": 2.7227704524993896,
      "step": 2850
    },
    {
      "epoch": 0.23124304793661493,
      "grad_norm": 0.048881176859140396,
      "learning_rate": 0.0012,
      "loss": 2.7442402839660645,
      "step": 2852
    },
    {
      "epoch": 0.2314052099618159,
      "grad_norm": 0.053290579468011856,
      "learning_rate": 0.0012,
      "loss": 2.7214441299438477,
      "step": 2854
    },
    {
      "epoch": 0.2315673719870169,
      "grad_norm": 0.028741000220179558,
      "learning_rate": 0.0012,
      "loss": 2.685905933380127,
      "step": 2856
    },
    {
      "epoch": 0.2317295340122179,
      "grad_norm": 0.02242858335375786,
      "learning_rate": 0.0012,
      "loss": 2.656764030456543,
      "step": 2858
    },
    {
      "epoch": 0.23189169603741888,
      "grad_norm": 0.02841844968497753,
      "learning_rate": 0.0012,
      "loss": 2.658510208129883,
      "step": 2860
    },
    {
      "epoch": 0.23205385806261988,
      "grad_norm": 0.04125404730439186,
      "learning_rate": 0.0012,
      "loss": 2.732283353805542,
      "step": 2862
    },
    {
      "epoch": 0.23221602008782088,
      "grad_norm": 0.021460941061377525,
      "learning_rate": 0.0012,
      "loss": 2.716339588165283,
      "step": 2864
    },
    {
      "epoch": 0.23237818211302186,
      "grad_norm": 0.026140952482819557,
      "learning_rate": 0.0012,
      "loss": 2.6974706649780273,
      "step": 2866
    },
    {
      "epoch": 0.23254034413822286,
      "grad_norm": 0.036853499710559845,
      "learning_rate": 0.0012,
      "loss": 2.5887396335601807,
      "step": 2868
    },
    {
      "epoch": 0.23270250616342386,
      "grad_norm": 0.025372516363859177,
      "learning_rate": 0.0012,
      "loss": 2.708413600921631,
      "step": 2870
    },
    {
      "epoch": 0.23286466818862483,
      "grad_norm": 0.03013041615486145,
      "learning_rate": 0.0012,
      "loss": 2.712085247039795,
      "step": 2872
    },
    {
      "epoch": 0.23302683021382584,
      "grad_norm": 0.0403306782245636,
      "learning_rate": 0.0012,
      "loss": 2.7305564880371094,
      "step": 2874
    },
    {
      "epoch": 0.23318899223902684,
      "grad_norm": 0.03557124733924866,
      "learning_rate": 0.0012,
      "loss": 2.736374855041504,
      "step": 2876
    },
    {
      "epoch": 0.2333511542642278,
      "grad_norm": 0.05232984200119972,
      "learning_rate": 0.0012,
      "loss": 2.645061731338501,
      "step": 2878
    },
    {
      "epoch": 0.2335133162894288,
      "grad_norm": 0.030214188620448112,
      "learning_rate": 0.0012,
      "loss": 2.6806068420410156,
      "step": 2880
    },
    {
      "epoch": 0.23367547831462981,
      "grad_norm": 0.03074566088616848,
      "learning_rate": 0.0012,
      "loss": 2.6533548831939697,
      "step": 2882
    },
    {
      "epoch": 0.2338376403398308,
      "grad_norm": 0.05163377523422241,
      "learning_rate": 0.0012,
      "loss": 2.7234747409820557,
      "step": 2884
    },
    {
      "epoch": 0.2339998023650318,
      "grad_norm": 0.034109048545360565,
      "learning_rate": 0.0012,
      "loss": 2.6832995414733887,
      "step": 2886
    },
    {
      "epoch": 0.2341619643902328,
      "grad_norm": 0.029000919312238693,
      "learning_rate": 0.0012,
      "loss": 2.7550172805786133,
      "step": 2888
    },
    {
      "epoch": 0.23432412641543376,
      "grad_norm": 0.06878913938999176,
      "learning_rate": 0.0012,
      "loss": 2.7185745239257812,
      "step": 2890
    },
    {
      "epoch": 0.23448628844063477,
      "grad_norm": 0.08221475780010223,
      "learning_rate": 0.0012,
      "loss": 2.7055697441101074,
      "step": 2892
    },
    {
      "epoch": 0.23464845046583577,
      "grad_norm": 0.04346026852726936,
      "learning_rate": 0.0012,
      "loss": 2.6785905361175537,
      "step": 2894
    },
    {
      "epoch": 0.23481061249103674,
      "grad_norm": 0.0351329930126667,
      "learning_rate": 0.0012,
      "loss": 2.7213144302368164,
      "step": 2896
    },
    {
      "epoch": 0.23497277451623774,
      "grad_norm": 0.03127681836485863,
      "learning_rate": 0.0012,
      "loss": 2.7356700897216797,
      "step": 2898
    },
    {
      "epoch": 0.23513493654143874,
      "grad_norm": 0.026036668568849564,
      "learning_rate": 0.0012,
      "loss": 2.7393388748168945,
      "step": 2900
    },
    {
      "epoch": 0.23529709856663972,
      "grad_norm": 0.03232694789767265,
      "learning_rate": 0.0012,
      "loss": 2.753933906555176,
      "step": 2902
    },
    {
      "epoch": 0.23545926059184072,
      "grad_norm": 0.042771752923727036,
      "learning_rate": 0.0012,
      "loss": 2.722442865371704,
      "step": 2904
    },
    {
      "epoch": 0.23562142261704172,
      "grad_norm": 0.038743019104003906,
      "learning_rate": 0.0012,
      "loss": 2.691138982772827,
      "step": 2906
    },
    {
      "epoch": 0.2357835846422427,
      "grad_norm": 0.028451189398765564,
      "learning_rate": 0.0012,
      "loss": 2.7278988361358643,
      "step": 2908
    },
    {
      "epoch": 0.2359457466674437,
      "grad_norm": 0.057042196393013,
      "learning_rate": 0.0012,
      "loss": 2.7573413848876953,
      "step": 2910
    },
    {
      "epoch": 0.2361079086926447,
      "grad_norm": 0.04705628752708435,
      "learning_rate": 0.0012,
      "loss": 2.655301570892334,
      "step": 2912
    },
    {
      "epoch": 0.23627007071784567,
      "grad_norm": 0.023507090285420418,
      "learning_rate": 0.0012,
      "loss": 2.6205923557281494,
      "step": 2914
    },
    {
      "epoch": 0.23643223274304667,
      "grad_norm": 0.04417256638407707,
      "learning_rate": 0.0012,
      "loss": 2.7014803886413574,
      "step": 2916
    },
    {
      "epoch": 0.23659439476824767,
      "grad_norm": 0.045058973133563995,
      "learning_rate": 0.0012,
      "loss": 2.6965816020965576,
      "step": 2918
    },
    {
      "epoch": 0.23675655679344865,
      "grad_norm": 0.045096512883901596,
      "learning_rate": 0.0012,
      "loss": 2.7122762203216553,
      "step": 2920
    },
    {
      "epoch": 0.23691871881864965,
      "grad_norm": 0.02462535724043846,
      "learning_rate": 0.0012,
      "loss": 2.6927058696746826,
      "step": 2922
    },
    {
      "epoch": 0.23708088084385065,
      "grad_norm": 0.028241291642189026,
      "learning_rate": 0.0012,
      "loss": 2.7468831539154053,
      "step": 2924
    },
    {
      "epoch": 0.23724304286905162,
      "grad_norm": 0.02855856530368328,
      "learning_rate": 0.0012,
      "loss": 2.7453536987304688,
      "step": 2926
    },
    {
      "epoch": 0.23740520489425262,
      "grad_norm": 0.03652961552143097,
      "learning_rate": 0.0012,
      "loss": 2.682443618774414,
      "step": 2928
    },
    {
      "epoch": 0.23756736691945363,
      "grad_norm": 0.03904666751623154,
      "learning_rate": 0.0012,
      "loss": 2.743823528289795,
      "step": 2930
    },
    {
      "epoch": 0.2377295289446546,
      "grad_norm": 0.026538116857409477,
      "learning_rate": 0.0012,
      "loss": 2.66038179397583,
      "step": 2932
    },
    {
      "epoch": 0.2378916909698556,
      "grad_norm": 0.02222716622054577,
      "learning_rate": 0.0012,
      "loss": 2.719388008117676,
      "step": 2934
    },
    {
      "epoch": 0.2380538529950566,
      "grad_norm": 0.02930467762053013,
      "learning_rate": 0.0012,
      "loss": 2.7048630714416504,
      "step": 2936
    },
    {
      "epoch": 0.23821601502025758,
      "grad_norm": 0.029494009912014008,
      "learning_rate": 0.0012,
      "loss": 2.6893296241760254,
      "step": 2938
    },
    {
      "epoch": 0.23837817704545858,
      "grad_norm": 0.04731793701648712,
      "learning_rate": 0.0012,
      "loss": 2.745288848876953,
      "step": 2940
    },
    {
      "epoch": 0.23854033907065958,
      "grad_norm": 0.04855873063206673,
      "learning_rate": 0.0012,
      "loss": 2.708726406097412,
      "step": 2942
    },
    {
      "epoch": 0.23870250109586055,
      "grad_norm": 0.04547962546348572,
      "learning_rate": 0.0012,
      "loss": 2.6840901374816895,
      "step": 2944
    },
    {
      "epoch": 0.23886466312106155,
      "grad_norm": 0.04309828579425812,
      "learning_rate": 0.0012,
      "loss": 2.7357711791992188,
      "step": 2946
    },
    {
      "epoch": 0.23902682514626256,
      "grad_norm": 0.022218814119696617,
      "learning_rate": 0.0012,
      "loss": 2.6624534130096436,
      "step": 2948
    },
    {
      "epoch": 0.23918898717146353,
      "grad_norm": 0.03915882483124733,
      "learning_rate": 0.0012,
      "loss": 2.648439884185791,
      "step": 2950
    },
    {
      "epoch": 0.23935114919666453,
      "grad_norm": 0.04326235502958298,
      "learning_rate": 0.0012,
      "loss": 2.7180233001708984,
      "step": 2952
    },
    {
      "epoch": 0.23951331122186553,
      "grad_norm": 0.024075401946902275,
      "learning_rate": 0.0012,
      "loss": 2.730076789855957,
      "step": 2954
    },
    {
      "epoch": 0.2396754732470665,
      "grad_norm": 0.024948474019765854,
      "learning_rate": 0.0012,
      "loss": 2.740007162094116,
      "step": 2956
    },
    {
      "epoch": 0.2398376352722675,
      "grad_norm": 0.05033247917890549,
      "learning_rate": 0.0012,
      "loss": 2.7145447731018066,
      "step": 2958
    },
    {
      "epoch": 0.2399997972974685,
      "grad_norm": 0.0345492921769619,
      "learning_rate": 0.0012,
      "loss": 2.67803692817688,
      "step": 2960
    },
    {
      "epoch": 0.24016195932266948,
      "grad_norm": 0.025833234190940857,
      "learning_rate": 0.0012,
      "loss": 2.6812551021575928,
      "step": 2962
    },
    {
      "epoch": 0.24032412134787048,
      "grad_norm": 0.028289256617426872,
      "learning_rate": 0.0012,
      "loss": 2.6760103702545166,
      "step": 2964
    },
    {
      "epoch": 0.24048628337307149,
      "grad_norm": 0.029800422489643097,
      "learning_rate": 0.0012,
      "loss": 2.69877290725708,
      "step": 2966
    },
    {
      "epoch": 0.24064844539827246,
      "grad_norm": 0.039615448564291,
      "learning_rate": 0.0012,
      "loss": 2.692929744720459,
      "step": 2968
    },
    {
      "epoch": 0.24081060742347346,
      "grad_norm": 0.024519197642803192,
      "learning_rate": 0.0012,
      "loss": 2.7033047676086426,
      "step": 2970
    },
    {
      "epoch": 0.24097276944867446,
      "grad_norm": 0.036123231053352356,
      "learning_rate": 0.0012,
      "loss": 2.682861328125,
      "step": 2972
    },
    {
      "epoch": 0.24113493147387544,
      "grad_norm": 0.03002551570534706,
      "learning_rate": 0.0012,
      "loss": 2.7515974044799805,
      "step": 2974
    },
    {
      "epoch": 0.24129709349907644,
      "grad_norm": 0.02414497546851635,
      "learning_rate": 0.0012,
      "loss": 2.7447280883789062,
      "step": 2976
    },
    {
      "epoch": 0.24145925552427744,
      "grad_norm": 0.030363792553544044,
      "learning_rate": 0.0012,
      "loss": 2.652357578277588,
      "step": 2978
    },
    {
      "epoch": 0.2416214175494784,
      "grad_norm": 0.02320041134953499,
      "learning_rate": 0.0012,
      "loss": 2.7556262016296387,
      "step": 2980
    },
    {
      "epoch": 0.2417835795746794,
      "grad_norm": 0.023286588490009308,
      "learning_rate": 0.0012,
      "loss": 2.7384612560272217,
      "step": 2982
    },
    {
      "epoch": 0.24194574159988042,
      "grad_norm": 0.03206389769911766,
      "learning_rate": 0.0012,
      "loss": 2.7049694061279297,
      "step": 2984
    },
    {
      "epoch": 0.2421079036250814,
      "grad_norm": 0.02349797450006008,
      "learning_rate": 0.0012,
      "loss": 2.7601380348205566,
      "step": 2986
    },
    {
      "epoch": 0.2422700656502824,
      "grad_norm": 0.03043835237622261,
      "learning_rate": 0.0012,
      "loss": 2.6987645626068115,
      "step": 2988
    },
    {
      "epoch": 0.2424322276754834,
      "grad_norm": 0.05747264623641968,
      "learning_rate": 0.0012,
      "loss": 2.744269371032715,
      "step": 2990
    },
    {
      "epoch": 0.24259438970068437,
      "grad_norm": 0.05755302682518959,
      "learning_rate": 0.0012,
      "loss": 2.6756484508514404,
      "step": 2992
    },
    {
      "epoch": 0.24275655172588537,
      "grad_norm": 0.02599668689072132,
      "learning_rate": 0.0012,
      "loss": 2.6890993118286133,
      "step": 2994
    },
    {
      "epoch": 0.24291871375108637,
      "grad_norm": 0.04187505319714546,
      "learning_rate": 0.0012,
      "loss": 2.6810641288757324,
      "step": 2996
    },
    {
      "epoch": 0.24308087577628734,
      "grad_norm": 0.04365087300539017,
      "learning_rate": 0.0012,
      "loss": 2.7097220420837402,
      "step": 2998
    },
    {
      "epoch": 0.24324303780148834,
      "grad_norm": 0.04581552743911743,
      "learning_rate": 0.0012,
      "loss": 2.7067666053771973,
      "step": 3000
    },
    {
      "epoch": 0.24340519982668934,
      "grad_norm": 0.02331043779850006,
      "learning_rate": 0.0012,
      "loss": 2.677687644958496,
      "step": 3002
    },
    {
      "epoch": 0.24356736185189032,
      "grad_norm": 0.05717064440250397,
      "learning_rate": 0.0012,
      "loss": 2.6726109981536865,
      "step": 3004
    },
    {
      "epoch": 0.24372952387709132,
      "grad_norm": 0.052816957235336304,
      "learning_rate": 0.0012,
      "loss": 2.728684425354004,
      "step": 3006
    },
    {
      "epoch": 0.24389168590229232,
      "grad_norm": 0.027332931756973267,
      "learning_rate": 0.0012,
      "loss": 2.735991954803467,
      "step": 3008
    },
    {
      "epoch": 0.2440538479274933,
      "grad_norm": 0.03058428317308426,
      "learning_rate": 0.0012,
      "loss": 2.7501816749572754,
      "step": 3010
    },
    {
      "epoch": 0.2442160099526943,
      "grad_norm": 0.04475148022174835,
      "learning_rate": 0.0012,
      "loss": 2.6527295112609863,
      "step": 3012
    },
    {
      "epoch": 0.2443781719778953,
      "grad_norm": 0.04596131294965744,
      "learning_rate": 0.0012,
      "loss": 2.6639606952667236,
      "step": 3014
    },
    {
      "epoch": 0.24454033400309627,
      "grad_norm": 0.037373535335063934,
      "learning_rate": 0.0012,
      "loss": 2.6902084350585938,
      "step": 3016
    },
    {
      "epoch": 0.24470249602829727,
      "grad_norm": 0.022757459431886673,
      "learning_rate": 0.0012,
      "loss": 2.721843719482422,
      "step": 3018
    },
    {
      "epoch": 0.24486465805349827,
      "grad_norm": 0.02472841367125511,
      "learning_rate": 0.0012,
      "loss": 2.7419254779815674,
      "step": 3020
    },
    {
      "epoch": 0.24502682007869925,
      "grad_norm": 0.03143644332885742,
      "learning_rate": 0.0012,
      "loss": 2.7031636238098145,
      "step": 3022
    },
    {
      "epoch": 0.24518898210390025,
      "grad_norm": 0.02282712049782276,
      "learning_rate": 0.0012,
      "loss": 2.729095458984375,
      "step": 3024
    },
    {
      "epoch": 0.24535114412910125,
      "grad_norm": 0.026938559487462044,
      "learning_rate": 0.0012,
      "loss": 2.691408395767212,
      "step": 3026
    },
    {
      "epoch": 0.24551330615430222,
      "grad_norm": 0.0256523247808218,
      "learning_rate": 0.0012,
      "loss": 2.6700899600982666,
      "step": 3028
    },
    {
      "epoch": 0.24567546817950323,
      "grad_norm": 0.02838793396949768,
      "learning_rate": 0.0012,
      "loss": 2.7482333183288574,
      "step": 3030
    },
    {
      "epoch": 0.24583763020470423,
      "grad_norm": 0.03450194001197815,
      "learning_rate": 0.0012,
      "loss": 2.7157211303710938,
      "step": 3032
    },
    {
      "epoch": 0.2459997922299052,
      "grad_norm": 0.029580967500805855,
      "learning_rate": 0.0012,
      "loss": 2.714209794998169,
      "step": 3034
    },
    {
      "epoch": 0.2461619542551062,
      "grad_norm": 0.028379635885357857,
      "learning_rate": 0.0012,
      "loss": 2.782045841217041,
      "step": 3036
    },
    {
      "epoch": 0.2463241162803072,
      "grad_norm": 0.023457098752260208,
      "learning_rate": 0.0012,
      "loss": 2.739910125732422,
      "step": 3038
    },
    {
      "epoch": 0.24648627830550818,
      "grad_norm": 0.02652040682733059,
      "learning_rate": 0.0012,
      "loss": 2.677304983139038,
      "step": 3040
    },
    {
      "epoch": 0.24664844033070918,
      "grad_norm": 0.025741707533597946,
      "learning_rate": 0.0012,
      "loss": 2.6716113090515137,
      "step": 3042
    },
    {
      "epoch": 0.24681060235591018,
      "grad_norm": 0.027116701006889343,
      "learning_rate": 0.0012,
      "loss": 2.7316160202026367,
      "step": 3044
    },
    {
      "epoch": 0.24697276438111115,
      "grad_norm": 0.038706161081790924,
      "learning_rate": 0.0012,
      "loss": 2.716090679168701,
      "step": 3046
    },
    {
      "epoch": 0.24713492640631216,
      "grad_norm": 0.025895502418279648,
      "learning_rate": 0.0012,
      "loss": 2.6427206993103027,
      "step": 3048
    },
    {
      "epoch": 0.24729708843151316,
      "grad_norm": 0.02891554869711399,
      "learning_rate": 0.0012,
      "loss": 2.7056121826171875,
      "step": 3050
    },
    {
      "epoch": 0.24745925045671413,
      "grad_norm": 0.046539802104234695,
      "learning_rate": 0.0012,
      "loss": 2.7071194648742676,
      "step": 3052
    },
    {
      "epoch": 0.24762141248191513,
      "grad_norm": 0.0480872318148613,
      "learning_rate": 0.0012,
      "loss": 2.7272846698760986,
      "step": 3054
    },
    {
      "epoch": 0.24778357450711613,
      "grad_norm": 0.028197817504405975,
      "learning_rate": 0.0012,
      "loss": 2.698120594024658,
      "step": 3056
    },
    {
      "epoch": 0.2479457365323171,
      "grad_norm": 0.038864556699991226,
      "learning_rate": 0.0012,
      "loss": 2.713909149169922,
      "step": 3058
    },
    {
      "epoch": 0.2481078985575181,
      "grad_norm": 0.06061799079179764,
      "learning_rate": 0.0012,
      "loss": 2.7462077140808105,
      "step": 3060
    },
    {
      "epoch": 0.2482700605827191,
      "grad_norm": 0.04336366802453995,
      "learning_rate": 0.0012,
      "loss": 2.6750240325927734,
      "step": 3062
    },
    {
      "epoch": 0.24843222260792008,
      "grad_norm": 0.05114663392305374,
      "learning_rate": 0.0012,
      "loss": 2.7012345790863037,
      "step": 3064
    },
    {
      "epoch": 0.24859438463312108,
      "grad_norm": 0.07936888188123703,
      "learning_rate": 0.0012,
      "loss": 2.6978936195373535,
      "step": 3066
    },
    {
      "epoch": 0.2487565466583221,
      "grad_norm": 0.05205497145652771,
      "learning_rate": 0.0012,
      "loss": 2.7630085945129395,
      "step": 3068
    },
    {
      "epoch": 0.24891870868352306,
      "grad_norm": 0.02691664732992649,
      "learning_rate": 0.0012,
      "loss": 2.68579363822937,
      "step": 3070
    },
    {
      "epoch": 0.24908087070872406,
      "grad_norm": 0.05447423458099365,
      "learning_rate": 0.0012,
      "loss": 2.6574480533599854,
      "step": 3072
    },
    {
      "epoch": 0.24924303273392506,
      "grad_norm": 0.05942544341087341,
      "learning_rate": 0.0012,
      "loss": 2.7274510860443115,
      "step": 3074
    },
    {
      "epoch": 0.24940519475912604,
      "grad_norm": 0.03165377676486969,
      "learning_rate": 0.0012,
      "loss": 2.705920696258545,
      "step": 3076
    },
    {
      "epoch": 0.24956735678432704,
      "grad_norm": 0.04917531460523605,
      "learning_rate": 0.0012,
      "loss": 2.665682554244995,
      "step": 3078
    },
    {
      "epoch": 0.24972951880952804,
      "grad_norm": 0.05379106476902962,
      "learning_rate": 0.0012,
      "loss": 2.7204461097717285,
      "step": 3080
    },
    {
      "epoch": 0.249891680834729,
      "grad_norm": 0.032674819231033325,
      "learning_rate": 0.0012,
      "loss": 2.7539446353912354,
      "step": 3082
    },
    {
      "epoch": 0.25005384285993004,
      "grad_norm": 0.0495276115834713,
      "learning_rate": 0.0012,
      "loss": 2.710740089416504,
      "step": 3084
    },
    {
      "epoch": 0.25005384285993004,
      "eval_loss": 2.706743001937866,
      "eval_runtime": 1148.8388,
      "eval_samples_per_second": 27.761,
      "eval_steps_per_second": 6.941,
      "step": 3084
    },
    {
      "epoch": 0.250216004885131,
      "grad_norm": 0.025464095175266266,
      "learning_rate": 0.0012,
      "loss": 2.7786099910736084,
      "step": 3086
    },
    {
      "epoch": 0.250378166910332,
      "grad_norm": 0.06016185134649277,
      "learning_rate": 0.0012,
      "loss": 2.766223430633545,
      "step": 3088
    },
    {
      "epoch": 0.250540328935533,
      "grad_norm": 0.046437818557024,
      "learning_rate": 0.0012,
      "loss": 2.697214126586914,
      "step": 3090
    },
    {
      "epoch": 0.250702490960734,
      "grad_norm": 0.026641886681318283,
      "learning_rate": 0.0012,
      "loss": 2.7508461475372314,
      "step": 3092
    },
    {
      "epoch": 0.25086465298593497,
      "grad_norm": 0.06901624798774719,
      "learning_rate": 0.0012,
      "loss": 2.656351327896118,
      "step": 3094
    },
    {
      "epoch": 0.251026815011136,
      "grad_norm": 0.04452091455459595,
      "learning_rate": 0.0012,
      "loss": 2.732724189758301,
      "step": 3096
    },
    {
      "epoch": 0.25118897703633697,
      "grad_norm": 0.030806398019194603,
      "learning_rate": 0.0012,
      "loss": 2.698793411254883,
      "step": 3098
    },
    {
      "epoch": 0.25135113906153794,
      "grad_norm": 0.05095498263835907,
      "learning_rate": 0.0012,
      "loss": 2.6625728607177734,
      "step": 3100
    },
    {
      "epoch": 0.25151330108673897,
      "grad_norm": 0.04286812245845795,
      "learning_rate": 0.0012,
      "loss": 2.684832811355591,
      "step": 3102
    },
    {
      "epoch": 0.25167546311193995,
      "grad_norm": 0.046234190464019775,
      "learning_rate": 0.0012,
      "loss": 2.7156176567077637,
      "step": 3104
    },
    {
      "epoch": 0.2518376251371409,
      "grad_norm": 0.03626799210906029,
      "learning_rate": 0.0012,
      "loss": 2.7174618244171143,
      "step": 3106
    },
    {
      "epoch": 0.25199978716234195,
      "grad_norm": 0.026352932676672935,
      "learning_rate": 0.0012,
      "loss": 2.740522623062134,
      "step": 3108
    },
    {
      "epoch": 0.2521619491875429,
      "grad_norm": 0.02382308803498745,
      "learning_rate": 0.0012,
      "loss": 2.7239813804626465,
      "step": 3110
    },
    {
      "epoch": 0.2523241112127439,
      "grad_norm": 0.028859488666057587,
      "learning_rate": 0.0012,
      "loss": 2.7407522201538086,
      "step": 3112
    },
    {
      "epoch": 0.2524862732379449,
      "grad_norm": 0.031002240255475044,
      "learning_rate": 0.0012,
      "loss": 2.673027992248535,
      "step": 3114
    },
    {
      "epoch": 0.2526484352631459,
      "grad_norm": 0.023334836587309837,
      "learning_rate": 0.0012,
      "loss": 2.7239437103271484,
      "step": 3116
    },
    {
      "epoch": 0.25281059728834687,
      "grad_norm": 0.049146637320518494,
      "learning_rate": 0.0012,
      "loss": 2.738558292388916,
      "step": 3118
    },
    {
      "epoch": 0.2529727593135479,
      "grad_norm": 0.03505962714552879,
      "learning_rate": 0.0012,
      "loss": 2.7162933349609375,
      "step": 3120
    },
    {
      "epoch": 0.2531349213387489,
      "grad_norm": 0.02460108883678913,
      "learning_rate": 0.0012,
      "loss": 2.6966609954833984,
      "step": 3122
    },
    {
      "epoch": 0.25329708336394985,
      "grad_norm": 0.030199700966477394,
      "learning_rate": 0.0012,
      "loss": 2.7258946895599365,
      "step": 3124
    },
    {
      "epoch": 0.2534592453891509,
      "grad_norm": 0.06267234683036804,
      "learning_rate": 0.0012,
      "loss": 2.709623098373413,
      "step": 3126
    },
    {
      "epoch": 0.25362140741435185,
      "grad_norm": 0.03916580229997635,
      "learning_rate": 0.0012,
      "loss": 2.7193312644958496,
      "step": 3128
    },
    {
      "epoch": 0.2537835694395528,
      "grad_norm": 0.03466412052512169,
      "learning_rate": 0.0012,
      "loss": 2.7282071113586426,
      "step": 3130
    },
    {
      "epoch": 0.25394573146475385,
      "grad_norm": 0.03088618628680706,
      "learning_rate": 0.0012,
      "loss": 2.64432430267334,
      "step": 3132
    },
    {
      "epoch": 0.25410789348995483,
      "grad_norm": 0.031057704240083694,
      "learning_rate": 0.0012,
      "loss": 2.694491386413574,
      "step": 3134
    },
    {
      "epoch": 0.2542700555151558,
      "grad_norm": 0.0358721949160099,
      "learning_rate": 0.0012,
      "loss": 2.7441422939300537,
      "step": 3136
    },
    {
      "epoch": 0.25443221754035683,
      "grad_norm": 0.04232353717088699,
      "learning_rate": 0.0012,
      "loss": 2.66253399848938,
      "step": 3138
    },
    {
      "epoch": 0.2545943795655578,
      "grad_norm": 0.05062701925635338,
      "learning_rate": 0.0012,
      "loss": 2.6405766010284424,
      "step": 3140
    },
    {
      "epoch": 0.2547565415907588,
      "grad_norm": 0.057788584381341934,
      "learning_rate": 0.0012,
      "loss": 2.7195348739624023,
      "step": 3142
    },
    {
      "epoch": 0.2549187036159598,
      "grad_norm": 0.036095183342695236,
      "learning_rate": 0.0012,
      "loss": 2.70542049407959,
      "step": 3144
    },
    {
      "epoch": 0.2550808656411608,
      "grad_norm": 0.02916887030005455,
      "learning_rate": 0.0012,
      "loss": 2.676398754119873,
      "step": 3146
    },
    {
      "epoch": 0.25524302766636175,
      "grad_norm": 0.024923890829086304,
      "learning_rate": 0.0012,
      "loss": 2.6730246543884277,
      "step": 3148
    },
    {
      "epoch": 0.2554051896915628,
      "grad_norm": 0.028262589126825333,
      "learning_rate": 0.0012,
      "loss": 2.7004265785217285,
      "step": 3150
    },
    {
      "epoch": 0.25556735171676376,
      "grad_norm": 0.032034751027822495,
      "learning_rate": 0.0012,
      "loss": 2.6558680534362793,
      "step": 3152
    },
    {
      "epoch": 0.25572951374196473,
      "grad_norm": 0.03755755349993706,
      "learning_rate": 0.0012,
      "loss": 2.720576286315918,
      "step": 3154
    },
    {
      "epoch": 0.25589167576716576,
      "grad_norm": 0.05924734100699425,
      "learning_rate": 0.0012,
      "loss": 2.749281406402588,
      "step": 3156
    },
    {
      "epoch": 0.25605383779236673,
      "grad_norm": 0.055141013115644455,
      "learning_rate": 0.0012,
      "loss": 2.684856414794922,
      "step": 3158
    },
    {
      "epoch": 0.2562159998175677,
      "grad_norm": 0.03853808715939522,
      "learning_rate": 0.0012,
      "loss": 2.6368765830993652,
      "step": 3160
    },
    {
      "epoch": 0.25637816184276874,
      "grad_norm": 0.025613242760300636,
      "learning_rate": 0.0012,
      "loss": 2.6722846031188965,
      "step": 3162
    },
    {
      "epoch": 0.2565403238679697,
      "grad_norm": 0.03880007937550545,
      "learning_rate": 0.0012,
      "loss": 2.715503215789795,
      "step": 3164
    },
    {
      "epoch": 0.2567024858931707,
      "grad_norm": 0.043405696749687195,
      "learning_rate": 0.0012,
      "loss": 2.6442573070526123,
      "step": 3166
    },
    {
      "epoch": 0.2568646479183717,
      "grad_norm": 0.05081886798143387,
      "learning_rate": 0.0012,
      "loss": 2.715001344680786,
      "step": 3168
    },
    {
      "epoch": 0.2570268099435727,
      "grad_norm": 0.04044853150844574,
      "learning_rate": 0.0012,
      "loss": 2.7986626625061035,
      "step": 3170
    },
    {
      "epoch": 0.25718897196877366,
      "grad_norm": 0.03718748688697815,
      "learning_rate": 0.0012,
      "loss": 2.6646499633789062,
      "step": 3172
    },
    {
      "epoch": 0.2573511339939747,
      "grad_norm": 0.027414999902248383,
      "learning_rate": 0.0012,
      "loss": 2.656191349029541,
      "step": 3174
    },
    {
      "epoch": 0.25751329601917566,
      "grad_norm": 0.025602731853723526,
      "learning_rate": 0.0012,
      "loss": 2.7239646911621094,
      "step": 3176
    },
    {
      "epoch": 0.25767545804437664,
      "grad_norm": 0.02682488039135933,
      "learning_rate": 0.0012,
      "loss": 2.6843442916870117,
      "step": 3178
    },
    {
      "epoch": 0.25783762006957767,
      "grad_norm": 0.04244659096002579,
      "learning_rate": 0.0012,
      "loss": 2.6453537940979004,
      "step": 3180
    },
    {
      "epoch": 0.25799978209477864,
      "grad_norm": 0.04514368250966072,
      "learning_rate": 0.0012,
      "loss": 2.651777744293213,
      "step": 3182
    },
    {
      "epoch": 0.2581619441199796,
      "grad_norm": 0.02918044477701187,
      "learning_rate": 0.0012,
      "loss": 2.7033815383911133,
      "step": 3184
    },
    {
      "epoch": 0.25832410614518064,
      "grad_norm": 0.026646940037608147,
      "learning_rate": 0.0012,
      "loss": 2.693155288696289,
      "step": 3186
    },
    {
      "epoch": 0.2584862681703816,
      "grad_norm": 0.03185809403657913,
      "learning_rate": 0.0012,
      "loss": 2.7529475688934326,
      "step": 3188
    },
    {
      "epoch": 0.2586484301955826,
      "grad_norm": 0.07881644368171692,
      "learning_rate": 0.0012,
      "loss": 2.7207608222961426,
      "step": 3190
    },
    {
      "epoch": 0.2588105922207836,
      "grad_norm": 0.03693808615207672,
      "learning_rate": 0.0012,
      "loss": 2.7092556953430176,
      "step": 3192
    },
    {
      "epoch": 0.2589727542459846,
      "grad_norm": 0.03018597513437271,
      "learning_rate": 0.0012,
      "loss": 2.7472147941589355,
      "step": 3194
    },
    {
      "epoch": 0.25913491627118557,
      "grad_norm": 0.05551399663090706,
      "learning_rate": 0.0012,
      "loss": 2.7042384147644043,
      "step": 3196
    },
    {
      "epoch": 0.2592970782963866,
      "grad_norm": 0.036580298095941544,
      "learning_rate": 0.0012,
      "loss": 2.679096221923828,
      "step": 3198
    },
    {
      "epoch": 0.25945924032158757,
      "grad_norm": 0.03176465258002281,
      "learning_rate": 0.0012,
      "loss": 2.688896656036377,
      "step": 3200
    },
    {
      "epoch": 0.25962140234678854,
      "grad_norm": 0.042499661445617676,
      "learning_rate": 0.0012,
      "loss": 2.6345958709716797,
      "step": 3202
    },
    {
      "epoch": 0.2597835643719896,
      "grad_norm": 0.042917732149362564,
      "learning_rate": 0.0012,
      "loss": 2.7509822845458984,
      "step": 3204
    },
    {
      "epoch": 0.25994572639719055,
      "grad_norm": 0.026120005175471306,
      "learning_rate": 0.0012,
      "loss": 2.7259302139282227,
      "step": 3206
    },
    {
      "epoch": 0.2601078884223915,
      "grad_norm": 0.05197150632739067,
      "learning_rate": 0.0012,
      "loss": 2.7258353233337402,
      "step": 3208
    },
    {
      "epoch": 0.26027005044759255,
      "grad_norm": 0.046105578541755676,
      "learning_rate": 0.0012,
      "loss": 2.685565710067749,
      "step": 3210
    },
    {
      "epoch": 0.2604322124727935,
      "grad_norm": 0.022977130487561226,
      "learning_rate": 0.0012,
      "loss": 2.6859092712402344,
      "step": 3212
    },
    {
      "epoch": 0.2605943744979945,
      "grad_norm": 0.029574308544397354,
      "learning_rate": 0.0012,
      "loss": 2.6732797622680664,
      "step": 3214
    },
    {
      "epoch": 0.2607565365231955,
      "grad_norm": 0.050850093364715576,
      "learning_rate": 0.0012,
      "loss": 2.74126935005188,
      "step": 3216
    },
    {
      "epoch": 0.2609186985483965,
      "grad_norm": 0.04028540104627609,
      "learning_rate": 0.0012,
      "loss": 2.736311912536621,
      "step": 3218
    },
    {
      "epoch": 0.2610808605735975,
      "grad_norm": 0.02311103604733944,
      "learning_rate": 0.0012,
      "loss": 2.7284226417541504,
      "step": 3220
    },
    {
      "epoch": 0.2612430225987985,
      "grad_norm": 0.060158953070640564,
      "learning_rate": 0.0012,
      "loss": 2.741988182067871,
      "step": 3222
    },
    {
      "epoch": 0.2614051846239995,
      "grad_norm": 0.04696779325604439,
      "learning_rate": 0.0012,
      "loss": 2.7558510303497314,
      "step": 3224
    },
    {
      "epoch": 0.26156734664920045,
      "grad_norm": 0.028465600684285164,
      "learning_rate": 0.0012,
      "loss": 2.7546544075012207,
      "step": 3226
    },
    {
      "epoch": 0.2617295086744015,
      "grad_norm": 0.06247063726186752,
      "learning_rate": 0.0012,
      "loss": 2.6845479011535645,
      "step": 3228
    },
    {
      "epoch": 0.26189167069960245,
      "grad_norm": 0.07436327636241913,
      "learning_rate": 0.0012,
      "loss": 2.7005743980407715,
      "step": 3230
    },
    {
      "epoch": 0.2620538327248034,
      "grad_norm": 0.024377668276429176,
      "learning_rate": 0.0012,
      "loss": 2.6666245460510254,
      "step": 3232
    },
    {
      "epoch": 0.26221599475000446,
      "grad_norm": 0.0289163701236248,
      "learning_rate": 0.0012,
      "loss": 2.754148244857788,
      "step": 3234
    },
    {
      "epoch": 0.26237815677520543,
      "grad_norm": 0.037568964064121246,
      "learning_rate": 0.0012,
      "loss": 2.6955699920654297,
      "step": 3236
    },
    {
      "epoch": 0.2625403188004064,
      "grad_norm": 0.024897411465644836,
      "learning_rate": 0.0012,
      "loss": 2.6906280517578125,
      "step": 3238
    },
    {
      "epoch": 0.26270248082560743,
      "grad_norm": 0.03809221461415291,
      "learning_rate": 0.0012,
      "loss": 2.7076287269592285,
      "step": 3240
    },
    {
      "epoch": 0.2628646428508084,
      "grad_norm": 0.02983851172029972,
      "learning_rate": 0.0012,
      "loss": 2.74932861328125,
      "step": 3242
    },
    {
      "epoch": 0.2630268048760094,
      "grad_norm": 0.06424614042043686,
      "learning_rate": 0.0012,
      "loss": 2.707307815551758,
      "step": 3244
    },
    {
      "epoch": 0.2631889669012104,
      "grad_norm": 0.09227762371301651,
      "learning_rate": 0.0012,
      "loss": 2.788041114807129,
      "step": 3246
    },
    {
      "epoch": 0.2633511289264114,
      "grad_norm": 0.04488295689225197,
      "learning_rate": 0.0012,
      "loss": 2.679727077484131,
      "step": 3248
    },
    {
      "epoch": 0.26351329095161236,
      "grad_norm": 0.050375133752822876,
      "learning_rate": 0.0012,
      "loss": 2.6348352432250977,
      "step": 3250
    },
    {
      "epoch": 0.2636754529768134,
      "grad_norm": 0.027697710320353508,
      "learning_rate": 0.0012,
      "loss": 2.73537015914917,
      "step": 3252
    },
    {
      "epoch": 0.26383761500201436,
      "grad_norm": 0.030587103217840195,
      "learning_rate": 0.0012,
      "loss": 2.715822458267212,
      "step": 3254
    },
    {
      "epoch": 0.26399977702721533,
      "grad_norm": 0.04313896968960762,
      "learning_rate": 0.0012,
      "loss": 2.646972417831421,
      "step": 3256
    },
    {
      "epoch": 0.26416193905241636,
      "grad_norm": 0.025649752467870712,
      "learning_rate": 0.0012,
      "loss": 2.678760051727295,
      "step": 3258
    },
    {
      "epoch": 0.26432410107761733,
      "grad_norm": 0.056272607296705246,
      "learning_rate": 0.0012,
      "loss": 2.6650452613830566,
      "step": 3260
    },
    {
      "epoch": 0.2644862631028183,
      "grad_norm": 0.047381024807691574,
      "learning_rate": 0.0012,
      "loss": 2.6748623847961426,
      "step": 3262
    },
    {
      "epoch": 0.26464842512801934,
      "grad_norm": 0.02208693139255047,
      "learning_rate": 0.0012,
      "loss": 2.6903815269470215,
      "step": 3264
    },
    {
      "epoch": 0.2648105871532203,
      "grad_norm": 0.037391241639852524,
      "learning_rate": 0.0012,
      "loss": 2.677198886871338,
      "step": 3266
    },
    {
      "epoch": 0.2649727491784213,
      "grad_norm": 0.023066626861691475,
      "learning_rate": 0.0012,
      "loss": 2.657888650894165,
      "step": 3268
    },
    {
      "epoch": 0.2651349112036223,
      "grad_norm": 0.04518509656190872,
      "learning_rate": 0.0012,
      "loss": 2.6381120681762695,
      "step": 3270
    },
    {
      "epoch": 0.2652970732288233,
      "grad_norm": 0.04959309473633766,
      "learning_rate": 0.0012,
      "loss": 2.682558536529541,
      "step": 3272
    },
    {
      "epoch": 0.26545923525402426,
      "grad_norm": 0.03266984224319458,
      "learning_rate": 0.0012,
      "loss": 2.683133602142334,
      "step": 3274
    },
    {
      "epoch": 0.2656213972792253,
      "grad_norm": 0.02810799516737461,
      "learning_rate": 0.0012,
      "loss": 2.7147676944732666,
      "step": 3276
    },
    {
      "epoch": 0.26578355930442626,
      "grad_norm": 0.03871092572808266,
      "learning_rate": 0.0012,
      "loss": 2.6928534507751465,
      "step": 3278
    },
    {
      "epoch": 0.26594572132962724,
      "grad_norm": 0.038465358316898346,
      "learning_rate": 0.0012,
      "loss": 2.703782558441162,
      "step": 3280
    },
    {
      "epoch": 0.26610788335482827,
      "grad_norm": 0.02419455349445343,
      "learning_rate": 0.0012,
      "loss": 2.744680404663086,
      "step": 3282
    },
    {
      "epoch": 0.26627004538002924,
      "grad_norm": 0.05309522897005081,
      "learning_rate": 0.0012,
      "loss": 2.662966728210449,
      "step": 3284
    },
    {
      "epoch": 0.2664322074052302,
      "grad_norm": 0.04303261637687683,
      "learning_rate": 0.0012,
      "loss": 2.6391220092773438,
      "step": 3286
    },
    {
      "epoch": 0.26659436943043124,
      "grad_norm": 0.034409262239933014,
      "learning_rate": 0.0012,
      "loss": 2.672701358795166,
      "step": 3288
    },
    {
      "epoch": 0.2667565314556322,
      "grad_norm": 0.04416009038686752,
      "learning_rate": 0.0012,
      "loss": 2.6793527603149414,
      "step": 3290
    },
    {
      "epoch": 0.2669186934808332,
      "grad_norm": 0.042341962456703186,
      "learning_rate": 0.0012,
      "loss": 2.7108216285705566,
      "step": 3292
    },
    {
      "epoch": 0.2670808555060342,
      "grad_norm": 0.02624794840812683,
      "learning_rate": 0.0012,
      "loss": 2.742063045501709,
      "step": 3294
    },
    {
      "epoch": 0.2672430175312352,
      "grad_norm": 0.02475673146545887,
      "learning_rate": 0.0012,
      "loss": 2.660421371459961,
      "step": 3296
    },
    {
      "epoch": 0.26740517955643617,
      "grad_norm": 0.034188251942396164,
      "learning_rate": 0.0012,
      "loss": 2.6731176376342773,
      "step": 3298
    },
    {
      "epoch": 0.2675673415816372,
      "grad_norm": 0.02954721264541149,
      "learning_rate": 0.0012,
      "loss": 2.7065696716308594,
      "step": 3300
    },
    {
      "epoch": 0.26772950360683817,
      "grad_norm": 0.02907450869679451,
      "learning_rate": 0.0012,
      "loss": 2.624898910522461,
      "step": 3302
    },
    {
      "epoch": 0.26789166563203914,
      "grad_norm": 0.025875717401504517,
      "learning_rate": 0.0012,
      "loss": 2.6593265533447266,
      "step": 3304
    },
    {
      "epoch": 0.2680538276572402,
      "grad_norm": 0.025464605540037155,
      "learning_rate": 0.0012,
      "loss": 2.710283041000366,
      "step": 3306
    },
    {
      "epoch": 0.26821598968244115,
      "grad_norm": 0.03594018518924713,
      "learning_rate": 0.0012,
      "loss": 2.7009902000427246,
      "step": 3308
    },
    {
      "epoch": 0.2683781517076421,
      "grad_norm": 0.0438288189470768,
      "learning_rate": 0.0012,
      "loss": 2.6992993354797363,
      "step": 3310
    },
    {
      "epoch": 0.26854031373284315,
      "grad_norm": 0.04333999380469322,
      "learning_rate": 0.0012,
      "loss": 2.6685636043548584,
      "step": 3312
    },
    {
      "epoch": 0.2687024757580441,
      "grad_norm": 0.044407930225133896,
      "learning_rate": 0.0012,
      "loss": 2.656789541244507,
      "step": 3314
    },
    {
      "epoch": 0.2688646377832451,
      "grad_norm": 0.02645237185060978,
      "learning_rate": 0.0012,
      "loss": 2.7108044624328613,
      "step": 3316
    },
    {
      "epoch": 0.2690267998084461,
      "grad_norm": 0.02657177858054638,
      "learning_rate": 0.0012,
      "loss": 2.738434314727783,
      "step": 3318
    },
    {
      "epoch": 0.2691889618336471,
      "grad_norm": 0.02532898634672165,
      "learning_rate": 0.0012,
      "loss": 2.6656277179718018,
      "step": 3320
    },
    {
      "epoch": 0.2693511238588481,
      "grad_norm": 0.023588353767991066,
      "learning_rate": 0.0012,
      "loss": 2.691995859146118,
      "step": 3322
    },
    {
      "epoch": 0.2695132858840491,
      "grad_norm": 0.04213135689496994,
      "learning_rate": 0.0012,
      "loss": 2.6219871044158936,
      "step": 3324
    },
    {
      "epoch": 0.2696754479092501,
      "grad_norm": 0.022214137017726898,
      "learning_rate": 0.0012,
      "loss": 2.6215779781341553,
      "step": 3326
    },
    {
      "epoch": 0.26983760993445105,
      "grad_norm": 0.026171518489718437,
      "learning_rate": 0.0012,
      "loss": 2.7192671298980713,
      "step": 3328
    },
    {
      "epoch": 0.2699997719596521,
      "grad_norm": 0.0247233584523201,
      "learning_rate": 0.0012,
      "loss": 2.717039108276367,
      "step": 3330
    },
    {
      "epoch": 0.27016193398485305,
      "grad_norm": 0.022396760061383247,
      "learning_rate": 0.0012,
      "loss": 2.7138988971710205,
      "step": 3332
    },
    {
      "epoch": 0.270324096010054,
      "grad_norm": 0.02898336388170719,
      "learning_rate": 0.0012,
      "loss": 2.6109986305236816,
      "step": 3334
    },
    {
      "epoch": 0.27048625803525506,
      "grad_norm": 0.024767817929387093,
      "learning_rate": 0.0012,
      "loss": 2.6808571815490723,
      "step": 3336
    },
    {
      "epoch": 0.27064842006045603,
      "grad_norm": 0.051403384655714035,
      "learning_rate": 0.0012,
      "loss": 2.6509509086608887,
      "step": 3338
    },
    {
      "epoch": 0.270810582085657,
      "grad_norm": 0.034167684614658356,
      "learning_rate": 0.0012,
      "loss": 2.6740212440490723,
      "step": 3340
    },
    {
      "epoch": 0.27097274411085803,
      "grad_norm": 0.03301997482776642,
      "learning_rate": 0.0012,
      "loss": 2.5694403648376465,
      "step": 3342
    },
    {
      "epoch": 0.271134906136059,
      "grad_norm": 0.03220667690038681,
      "learning_rate": 0.0012,
      "loss": 2.6218926906585693,
      "step": 3344
    },
    {
      "epoch": 0.27129706816126,
      "grad_norm": 0.041436076164245605,
      "learning_rate": 0.0012,
      "loss": 2.681213617324829,
      "step": 3346
    },
    {
      "epoch": 0.271459230186461,
      "grad_norm": 0.037761591374874115,
      "learning_rate": 0.0012,
      "loss": 2.748034954071045,
      "step": 3348
    },
    {
      "epoch": 0.271621392211662,
      "grad_norm": 0.04704133793711662,
      "learning_rate": 0.0012,
      "loss": 2.684917449951172,
      "step": 3350
    },
    {
      "epoch": 0.27178355423686296,
      "grad_norm": 0.0377856008708477,
      "learning_rate": 0.0012,
      "loss": 2.689357280731201,
      "step": 3352
    },
    {
      "epoch": 0.271945716262064,
      "grad_norm": 0.027899116277694702,
      "learning_rate": 0.0012,
      "loss": 2.7213926315307617,
      "step": 3354
    },
    {
      "epoch": 0.27210787828726496,
      "grad_norm": 0.02712581865489483,
      "learning_rate": 0.0012,
      "loss": 2.633711338043213,
      "step": 3356
    },
    {
      "epoch": 0.27227004031246593,
      "grad_norm": 0.026077115908265114,
      "learning_rate": 0.0012,
      "loss": 2.671825885772705,
      "step": 3358
    },
    {
      "epoch": 0.27243220233766696,
      "grad_norm": 0.05857793241739273,
      "learning_rate": 0.0012,
      "loss": 2.7283287048339844,
      "step": 3360
    },
    {
      "epoch": 0.27259436436286794,
      "grad_norm": 0.05957653746008873,
      "learning_rate": 0.0012,
      "loss": 2.688400983810425,
      "step": 3362
    },
    {
      "epoch": 0.2727565263880689,
      "grad_norm": 0.03173523023724556,
      "learning_rate": 0.0012,
      "loss": 2.6775784492492676,
      "step": 3364
    },
    {
      "epoch": 0.27291868841326994,
      "grad_norm": 0.025698021054267883,
      "learning_rate": 0.0012,
      "loss": 2.701725721359253,
      "step": 3366
    },
    {
      "epoch": 0.2730808504384709,
      "grad_norm": 0.062258146703243256,
      "learning_rate": 0.0012,
      "loss": 2.675168514251709,
      "step": 3368
    },
    {
      "epoch": 0.2732430124636719,
      "grad_norm": 0.06586690992116928,
      "learning_rate": 0.0012,
      "loss": 2.709415912628174,
      "step": 3370
    },
    {
      "epoch": 0.2734051744888729,
      "grad_norm": 0.027446705847978592,
      "learning_rate": 0.0012,
      "loss": 2.700045108795166,
      "step": 3372
    },
    {
      "epoch": 0.2735673365140739,
      "grad_norm": 0.03567645326256752,
      "learning_rate": 0.0012,
      "loss": 2.6949167251586914,
      "step": 3374
    },
    {
      "epoch": 0.27372949853927486,
      "grad_norm": 0.06902799755334854,
      "learning_rate": 0.0012,
      "loss": 2.722456455230713,
      "step": 3376
    },
    {
      "epoch": 0.2738916605644759,
      "grad_norm": 0.023962106555700302,
      "learning_rate": 0.0012,
      "loss": 2.6852378845214844,
      "step": 3378
    },
    {
      "epoch": 0.27405382258967687,
      "grad_norm": 0.051534950733184814,
      "learning_rate": 0.0012,
      "loss": 2.7253613471984863,
      "step": 3380
    },
    {
      "epoch": 0.27421598461487784,
      "grad_norm": 0.056445445865392685,
      "learning_rate": 0.0012,
      "loss": 2.6954450607299805,
      "step": 3382
    },
    {
      "epoch": 0.27437814664007887,
      "grad_norm": 0.02535751834511757,
      "learning_rate": 0.0012,
      "loss": 2.746809959411621,
      "step": 3384
    },
    {
      "epoch": 0.27454030866527984,
      "grad_norm": 0.056104157119989395,
      "learning_rate": 0.0012,
      "loss": 2.7321667671203613,
      "step": 3386
    },
    {
      "epoch": 0.2747024706904808,
      "grad_norm": 0.056694578379392624,
      "learning_rate": 0.0012,
      "loss": 2.7035298347473145,
      "step": 3388
    },
    {
      "epoch": 0.27486463271568184,
      "grad_norm": 0.02361370250582695,
      "learning_rate": 0.0012,
      "loss": 2.685255289077759,
      "step": 3390
    },
    {
      "epoch": 0.2750267947408828,
      "grad_norm": 0.038406383246183395,
      "learning_rate": 0.0012,
      "loss": 2.750577211380005,
      "step": 3392
    },
    {
      "epoch": 0.2751889567660838,
      "grad_norm": 0.026955027133226395,
      "learning_rate": 0.0012,
      "loss": 2.699561595916748,
      "step": 3394
    },
    {
      "epoch": 0.2753511187912848,
      "grad_norm": 0.02667987160384655,
      "learning_rate": 0.0012,
      "loss": 2.6750807762145996,
      "step": 3396
    },
    {
      "epoch": 0.2755132808164858,
      "grad_norm": 0.02508557215332985,
      "learning_rate": 0.0012,
      "loss": 2.711872100830078,
      "step": 3398
    },
    {
      "epoch": 0.27567544284168677,
      "grad_norm": 0.029491793364286423,
      "learning_rate": 0.0012,
      "loss": 2.6747124195098877,
      "step": 3400
    },
    {
      "epoch": 0.2758376048668878,
      "grad_norm": 0.03525857627391815,
      "learning_rate": 0.0012,
      "loss": 2.7172508239746094,
      "step": 3402
    },
    {
      "epoch": 0.27599976689208877,
      "grad_norm": 0.025942260399460793,
      "learning_rate": 0.0012,
      "loss": 2.7003962993621826,
      "step": 3404
    },
    {
      "epoch": 0.27616192891728975,
      "grad_norm": 0.025403669103980064,
      "learning_rate": 0.0012,
      "loss": 2.6858837604522705,
      "step": 3406
    },
    {
      "epoch": 0.2763240909424908,
      "grad_norm": 0.031971994787454605,
      "learning_rate": 0.0012,
      "loss": 2.6971631050109863,
      "step": 3408
    },
    {
      "epoch": 0.27648625296769175,
      "grad_norm": 0.07256284356117249,
      "learning_rate": 0.0012,
      "loss": 2.6326119899749756,
      "step": 3410
    },
    {
      "epoch": 0.2766484149928927,
      "grad_norm": 0.06786856800317764,
      "learning_rate": 0.0012,
      "loss": 2.7071354389190674,
      "step": 3412
    },
    {
      "epoch": 0.27681057701809375,
      "grad_norm": 0.027229972183704376,
      "learning_rate": 0.0012,
      "loss": 2.733405113220215,
      "step": 3414
    },
    {
      "epoch": 0.2769727390432947,
      "grad_norm": 0.08579825609922409,
      "learning_rate": 0.0012,
      "loss": 2.7172346115112305,
      "step": 3416
    },
    {
      "epoch": 0.2771349010684957,
      "grad_norm": 0.034744758158922195,
      "learning_rate": 0.0012,
      "loss": 2.6621968746185303,
      "step": 3418
    },
    {
      "epoch": 0.2772970630936967,
      "grad_norm": 0.042785074561834335,
      "learning_rate": 0.0012,
      "loss": 2.7360119819641113,
      "step": 3420
    },
    {
      "epoch": 0.2774592251188977,
      "grad_norm": 0.044172901660203934,
      "learning_rate": 0.0012,
      "loss": 2.7165112495422363,
      "step": 3422
    },
    {
      "epoch": 0.2776213871440987,
      "grad_norm": 0.02289487048983574,
      "learning_rate": 0.0012,
      "loss": 2.59725284576416,
      "step": 3424
    },
    {
      "epoch": 0.2777835491692997,
      "grad_norm": 0.05455096811056137,
      "learning_rate": 0.0012,
      "loss": 2.711531162261963,
      "step": 3426
    },
    {
      "epoch": 0.2779457111945007,
      "grad_norm": 0.04947530850768089,
      "learning_rate": 0.0012,
      "loss": 2.6868748664855957,
      "step": 3428
    },
    {
      "epoch": 0.27810787321970165,
      "grad_norm": 0.03327358886599541,
      "learning_rate": 0.0012,
      "loss": 2.716512441635132,
      "step": 3430
    },
    {
      "epoch": 0.2782700352449027,
      "grad_norm": 0.029675984755158424,
      "learning_rate": 0.0012,
      "loss": 2.6575822830200195,
      "step": 3432
    },
    {
      "epoch": 0.27843219727010365,
      "grad_norm": 0.0428876094520092,
      "learning_rate": 0.0012,
      "loss": 2.6518349647521973,
      "step": 3434
    },
    {
      "epoch": 0.2785943592953046,
      "grad_norm": 0.0239908117800951,
      "learning_rate": 0.0012,
      "loss": 2.725343704223633,
      "step": 3436
    },
    {
      "epoch": 0.27875652132050566,
      "grad_norm": 0.0331391915678978,
      "learning_rate": 0.0012,
      "loss": 2.734609603881836,
      "step": 3438
    },
    {
      "epoch": 0.27891868334570663,
      "grad_norm": 0.03869687393307686,
      "learning_rate": 0.0012,
      "loss": 2.6671547889709473,
      "step": 3440
    },
    {
      "epoch": 0.2790808453709076,
      "grad_norm": 0.02352280355989933,
      "learning_rate": 0.0012,
      "loss": 2.629085063934326,
      "step": 3442
    },
    {
      "epoch": 0.27924300739610863,
      "grad_norm": 0.023099998012185097,
      "learning_rate": 0.0012,
      "loss": 2.66153621673584,
      "step": 3444
    },
    {
      "epoch": 0.2794051694213096,
      "grad_norm": 0.03449149429798126,
      "learning_rate": 0.0012,
      "loss": 2.6385626792907715,
      "step": 3446
    },
    {
      "epoch": 0.2795673314465106,
      "grad_norm": 0.024186749011278152,
      "learning_rate": 0.0012,
      "loss": 2.7156903743743896,
      "step": 3448
    },
    {
      "epoch": 0.2797294934717116,
      "grad_norm": 0.025279713794589043,
      "learning_rate": 0.0012,
      "loss": 2.7237765789031982,
      "step": 3450
    },
    {
      "epoch": 0.2798916554969126,
      "grad_norm": 0.02943737618625164,
      "learning_rate": 0.0012,
      "loss": 2.730128765106201,
      "step": 3452
    },
    {
      "epoch": 0.28005381752211356,
      "grad_norm": 0.037672434002161026,
      "learning_rate": 0.0012,
      "loss": 2.7390153408050537,
      "step": 3454
    },
    {
      "epoch": 0.2802159795473146,
      "grad_norm": 0.029889510944485664,
      "learning_rate": 0.0012,
      "loss": 2.6797053813934326,
      "step": 3456
    },
    {
      "epoch": 0.28037814157251556,
      "grad_norm": 0.030534891411662102,
      "learning_rate": 0.0012,
      "loss": 2.6973929405212402,
      "step": 3458
    },
    {
      "epoch": 0.28054030359771653,
      "grad_norm": 0.026959313079714775,
      "learning_rate": 0.0012,
      "loss": 2.7582669258117676,
      "step": 3460
    },
    {
      "epoch": 0.28070246562291756,
      "grad_norm": 0.02806617133319378,
      "learning_rate": 0.0012,
      "loss": 2.6738619804382324,
      "step": 3462
    },
    {
      "epoch": 0.28086462764811854,
      "grad_norm": 0.040059201419353485,
      "learning_rate": 0.0012,
      "loss": 2.6807899475097656,
      "step": 3464
    },
    {
      "epoch": 0.2810267896733195,
      "grad_norm": 0.03827083110809326,
      "learning_rate": 0.0012,
      "loss": 2.6423797607421875,
      "step": 3466
    },
    {
      "epoch": 0.28118895169852054,
      "grad_norm": 0.02492240071296692,
      "learning_rate": 0.0012,
      "loss": 2.6731886863708496,
      "step": 3468
    },
    {
      "epoch": 0.2813511137237215,
      "grad_norm": 0.02867594175040722,
      "learning_rate": 0.0012,
      "loss": 2.6982192993164062,
      "step": 3470
    },
    {
      "epoch": 0.2815132757489225,
      "grad_norm": 0.03879070654511452,
      "learning_rate": 0.0012,
      "loss": 2.6698200702667236,
      "step": 3472
    },
    {
      "epoch": 0.2816754377741235,
      "grad_norm": 0.025196261703968048,
      "learning_rate": 0.0012,
      "loss": 2.711275100708008,
      "step": 3474
    },
    {
      "epoch": 0.2818375997993245,
      "grad_norm": 0.024772468954324722,
      "learning_rate": 0.0012,
      "loss": 2.694671630859375,
      "step": 3476
    },
    {
      "epoch": 0.28199976182452546,
      "grad_norm": 0.03070407547056675,
      "learning_rate": 0.0012,
      "loss": 2.704007387161255,
      "step": 3478
    },
    {
      "epoch": 0.2821619238497265,
      "grad_norm": 0.030590400099754333,
      "learning_rate": 0.0012,
      "loss": 2.7575936317443848,
      "step": 3480
    },
    {
      "epoch": 0.28232408587492747,
      "grad_norm": 0.042900193482637405,
      "learning_rate": 0.0012,
      "loss": 2.7417349815368652,
      "step": 3482
    },
    {
      "epoch": 0.28248624790012844,
      "grad_norm": 0.032352592796087265,
      "learning_rate": 0.0012,
      "loss": 2.6817786693573,
      "step": 3484
    },
    {
      "epoch": 0.28264840992532947,
      "grad_norm": 0.025442739948630333,
      "learning_rate": 0.0012,
      "loss": 2.716799736022949,
      "step": 3486
    },
    {
      "epoch": 0.28281057195053044,
      "grad_norm": 0.026958301663398743,
      "learning_rate": 0.0012,
      "loss": 2.6722917556762695,
      "step": 3488
    },
    {
      "epoch": 0.2829727339757314,
      "grad_norm": 0.023728664964437485,
      "learning_rate": 0.0012,
      "loss": 2.656172275543213,
      "step": 3490
    },
    {
      "epoch": 0.28313489600093245,
      "grad_norm": 0.04292122274637222,
      "learning_rate": 0.0012,
      "loss": 2.7169852256774902,
      "step": 3492
    },
    {
      "epoch": 0.2832970580261334,
      "grad_norm": 0.051968954503536224,
      "learning_rate": 0.0012,
      "loss": 2.712042808532715,
      "step": 3494
    },
    {
      "epoch": 0.2834592200513344,
      "grad_norm": 0.033374663442373276,
      "learning_rate": 0.0012,
      "loss": 2.6976680755615234,
      "step": 3496
    },
    {
      "epoch": 0.2836213820765354,
      "grad_norm": 0.02469482645392418,
      "learning_rate": 0.0012,
      "loss": 2.6709794998168945,
      "step": 3498
    },
    {
      "epoch": 0.2837835441017364,
      "grad_norm": 0.02986256033182144,
      "learning_rate": 0.0012,
      "loss": 2.676215648651123,
      "step": 3500
    },
    {
      "epoch": 0.28394570612693737,
      "grad_norm": 0.05596422404050827,
      "learning_rate": 0.0012,
      "loss": 2.699978828430176,
      "step": 3502
    },
    {
      "epoch": 0.2841078681521384,
      "grad_norm": 0.04223903268575668,
      "learning_rate": 0.0012,
      "loss": 2.7485060691833496,
      "step": 3504
    },
    {
      "epoch": 0.2842700301773394,
      "grad_norm": 0.0314704030752182,
      "learning_rate": 0.0012,
      "loss": 2.7206785678863525,
      "step": 3506
    },
    {
      "epoch": 0.28443219220254035,
      "grad_norm": 0.02730206400156021,
      "learning_rate": 0.0012,
      "loss": 2.73266339302063,
      "step": 3508
    },
    {
      "epoch": 0.2845943542277414,
      "grad_norm": 0.0674768015742302,
      "learning_rate": 0.0012,
      "loss": 2.69364070892334,
      "step": 3510
    },
    {
      "epoch": 0.28475651625294235,
      "grad_norm": 0.09402772039175034,
      "learning_rate": 0.0012,
      "loss": 2.6782639026641846,
      "step": 3512
    },
    {
      "epoch": 0.2849186782781433,
      "grad_norm": 0.06807664036750793,
      "learning_rate": 0.0012,
      "loss": 2.749244451522827,
      "step": 3514
    },
    {
      "epoch": 0.28508084030334435,
      "grad_norm": 0.03156169131398201,
      "learning_rate": 0.0012,
      "loss": 2.710909128189087,
      "step": 3516
    },
    {
      "epoch": 0.2852430023285453,
      "grad_norm": 0.06480380892753601,
      "learning_rate": 0.0012,
      "loss": 2.651754856109619,
      "step": 3518
    },
    {
      "epoch": 0.2854051643537463,
      "grad_norm": 0.046458303928375244,
      "learning_rate": 0.0012,
      "loss": 2.707864284515381,
      "step": 3520
    },
    {
      "epoch": 0.28556732637894733,
      "grad_norm": 0.027149055153131485,
      "learning_rate": 0.0012,
      "loss": 2.7567739486694336,
      "step": 3522
    },
    {
      "epoch": 0.2857294884041483,
      "grad_norm": 0.0373716875910759,
      "learning_rate": 0.0012,
      "loss": 2.755173683166504,
      "step": 3524
    },
    {
      "epoch": 0.2858916504293493,
      "grad_norm": 0.0296830665320158,
      "learning_rate": 0.0012,
      "loss": 2.722036838531494,
      "step": 3526
    },
    {
      "epoch": 0.2860538124545503,
      "grad_norm": 0.03533012419939041,
      "learning_rate": 0.0012,
      "loss": 2.667806625366211,
      "step": 3528
    },
    {
      "epoch": 0.2862159744797513,
      "grad_norm": 0.02245168387889862,
      "learning_rate": 0.0012,
      "loss": 2.6608524322509766,
      "step": 3530
    },
    {
      "epoch": 0.28637813650495225,
      "grad_norm": 0.05533822625875473,
      "learning_rate": 0.0012,
      "loss": 2.7685041427612305,
      "step": 3532
    },
    {
      "epoch": 0.2865402985301533,
      "grad_norm": 0.07195763289928436,
      "learning_rate": 0.0012,
      "loss": 2.6916115283966064,
      "step": 3534
    },
    {
      "epoch": 0.28670246055535425,
      "grad_norm": 0.03082442283630371,
      "learning_rate": 0.0012,
      "loss": 2.7094998359680176,
      "step": 3536
    },
    {
      "epoch": 0.28686462258055523,
      "grad_norm": 0.05350085720419884,
      "learning_rate": 0.0012,
      "loss": 2.672337055206299,
      "step": 3538
    },
    {
      "epoch": 0.28702678460575626,
      "grad_norm": 0.05598202347755432,
      "learning_rate": 0.0012,
      "loss": 2.699902057647705,
      "step": 3540
    },
    {
      "epoch": 0.28718894663095723,
      "grad_norm": 0.023680400103330612,
      "learning_rate": 0.0012,
      "loss": 2.6753921508789062,
      "step": 3542
    },
    {
      "epoch": 0.2873511086561582,
      "grad_norm": 0.033278997987508774,
      "learning_rate": 0.0012,
      "loss": 2.6810150146484375,
      "step": 3544
    },
    {
      "epoch": 0.28751327068135923,
      "grad_norm": 0.041635919362306595,
      "learning_rate": 0.0012,
      "loss": 2.7025699615478516,
      "step": 3546
    },
    {
      "epoch": 0.2876754327065602,
      "grad_norm": 0.024568982422351837,
      "learning_rate": 0.0012,
      "loss": 2.6890640258789062,
      "step": 3548
    },
    {
      "epoch": 0.2878375947317612,
      "grad_norm": 0.03303627669811249,
      "learning_rate": 0.0012,
      "loss": 2.679427146911621,
      "step": 3550
    },
    {
      "epoch": 0.2879997567569622,
      "grad_norm": 0.0395730696618557,
      "learning_rate": 0.0012,
      "loss": 2.6837871074676514,
      "step": 3552
    },
    {
      "epoch": 0.2881619187821632,
      "grad_norm": 0.029711754992604256,
      "learning_rate": 0.0012,
      "loss": 2.707287311553955,
      "step": 3554
    },
    {
      "epoch": 0.28832408080736416,
      "grad_norm": 0.02617994323372841,
      "learning_rate": 0.0012,
      "loss": 2.6686501502990723,
      "step": 3556
    },
    {
      "epoch": 0.2884862428325652,
      "grad_norm": 0.03996168449521065,
      "learning_rate": 0.0012,
      "loss": 2.656383514404297,
      "step": 3558
    },
    {
      "epoch": 0.28864840485776616,
      "grad_norm": 0.03033224679529667,
      "learning_rate": 0.0012,
      "loss": 2.7056589126586914,
      "step": 3560
    },
    {
      "epoch": 0.28881056688296713,
      "grad_norm": 0.03605812415480614,
      "learning_rate": 0.0012,
      "loss": 2.7384822368621826,
      "step": 3562
    },
    {
      "epoch": 0.28897272890816816,
      "grad_norm": 0.027953658252954483,
      "learning_rate": 0.0012,
      "loss": 2.631700038909912,
      "step": 3564
    },
    {
      "epoch": 0.28913489093336914,
      "grad_norm": 0.02535293996334076,
      "learning_rate": 0.0012,
      "loss": 2.703561782836914,
      "step": 3566
    },
    {
      "epoch": 0.2892970529585701,
      "grad_norm": 0.03304172307252884,
      "learning_rate": 0.0012,
      "loss": 2.692748785018921,
      "step": 3568
    },
    {
      "epoch": 0.28945921498377114,
      "grad_norm": 0.026046443730592728,
      "learning_rate": 0.0012,
      "loss": 2.74316143989563,
      "step": 3570
    },
    {
      "epoch": 0.2896213770089721,
      "grad_norm": 0.051511187106370926,
      "learning_rate": 0.0012,
      "loss": 2.6958556175231934,
      "step": 3572
    },
    {
      "epoch": 0.2897835390341731,
      "grad_norm": 0.04437864199280739,
      "learning_rate": 0.0012,
      "loss": 2.652636766433716,
      "step": 3574
    },
    {
      "epoch": 0.2899457010593741,
      "grad_norm": 0.038056522607803345,
      "learning_rate": 0.0012,
      "loss": 2.719395637512207,
      "step": 3576
    },
    {
      "epoch": 0.2901078630845751,
      "grad_norm": 0.03011668473482132,
      "learning_rate": 0.0012,
      "loss": 2.7007455825805664,
      "step": 3578
    },
    {
      "epoch": 0.29027002510977606,
      "grad_norm": 0.033636778593063354,
      "learning_rate": 0.0012,
      "loss": 2.6402759552001953,
      "step": 3580
    },
    {
      "epoch": 0.2904321871349771,
      "grad_norm": 0.04526745527982712,
      "learning_rate": 0.0012,
      "loss": 2.804429769515991,
      "step": 3582
    },
    {
      "epoch": 0.29059434916017807,
      "grad_norm": 0.0353056900203228,
      "learning_rate": 0.0012,
      "loss": 2.6783668994903564,
      "step": 3584
    },
    {
      "epoch": 0.29075651118537904,
      "grad_norm": 0.023121891543269157,
      "learning_rate": 0.0012,
      "loss": 2.7022287845611572,
      "step": 3586
    },
    {
      "epoch": 0.29091867321058007,
      "grad_norm": 0.022688275203108788,
      "learning_rate": 0.0012,
      "loss": 2.736365795135498,
      "step": 3588
    },
    {
      "epoch": 0.29108083523578104,
      "grad_norm": 0.042628079652786255,
      "learning_rate": 0.0012,
      "loss": 2.6732373237609863,
      "step": 3590
    },
    {
      "epoch": 0.291242997260982,
      "grad_norm": 0.047835830599069595,
      "learning_rate": 0.0012,
      "loss": 2.7087533473968506,
      "step": 3592
    },
    {
      "epoch": 0.29140515928618305,
      "grad_norm": 0.03671550005674362,
      "learning_rate": 0.0012,
      "loss": 2.69978666305542,
      "step": 3594
    },
    {
      "epoch": 0.291567321311384,
      "grad_norm": 0.03001483343541622,
      "learning_rate": 0.0012,
      "loss": 2.6498661041259766,
      "step": 3596
    },
    {
      "epoch": 0.291729483336585,
      "grad_norm": 0.02340870536863804,
      "learning_rate": 0.0012,
      "loss": 2.638415813446045,
      "step": 3598
    },
    {
      "epoch": 0.291891645361786,
      "grad_norm": 0.028180744498968124,
      "learning_rate": 0.0012,
      "loss": 2.696946620941162,
      "step": 3600
    },
    {
      "epoch": 0.292053807386987,
      "grad_norm": 0.05225491523742676,
      "learning_rate": 0.0012,
      "loss": 2.730280876159668,
      "step": 3602
    },
    {
      "epoch": 0.29221596941218797,
      "grad_norm": 0.05067569389939308,
      "learning_rate": 0.0012,
      "loss": 2.679872989654541,
      "step": 3604
    },
    {
      "epoch": 0.292378131437389,
      "grad_norm": 0.03838733956217766,
      "learning_rate": 0.0012,
      "loss": 2.699526309967041,
      "step": 3606
    },
    {
      "epoch": 0.29254029346259,
      "grad_norm": 0.027493610978126526,
      "learning_rate": 0.0012,
      "loss": 2.7031474113464355,
      "step": 3608
    },
    {
      "epoch": 0.29270245548779095,
      "grad_norm": 0.0230643842369318,
      "learning_rate": 0.0012,
      "loss": 2.7085962295532227,
      "step": 3610
    },
    {
      "epoch": 0.292864617512992,
      "grad_norm": 0.02337685413658619,
      "learning_rate": 0.0012,
      "loss": 2.6566317081451416,
      "step": 3612
    },
    {
      "epoch": 0.29302677953819295,
      "grad_norm": 0.03310178592801094,
      "learning_rate": 0.0012,
      "loss": 2.697751045227051,
      "step": 3614
    },
    {
      "epoch": 0.2931889415633939,
      "grad_norm": 0.045669883489608765,
      "learning_rate": 0.0012,
      "loss": 2.621551513671875,
      "step": 3616
    },
    {
      "epoch": 0.29335110358859495,
      "grad_norm": 0.02368960529565811,
      "learning_rate": 0.0012,
      "loss": 2.6756081581115723,
      "step": 3618
    },
    {
      "epoch": 0.2935132656137959,
      "grad_norm": 0.031625665724277496,
      "learning_rate": 0.0012,
      "loss": 2.698126792907715,
      "step": 3620
    },
    {
      "epoch": 0.2936754276389969,
      "grad_norm": 0.024485928937792778,
      "learning_rate": 0.0012,
      "loss": 2.724820137023926,
      "step": 3622
    },
    {
      "epoch": 0.29383758966419793,
      "grad_norm": 0.032417185604572296,
      "learning_rate": 0.0012,
      "loss": 2.726928234100342,
      "step": 3624
    },
    {
      "epoch": 0.2939997516893989,
      "grad_norm": 0.0331973060965538,
      "learning_rate": 0.0012,
      "loss": 2.671325922012329,
      "step": 3626
    },
    {
      "epoch": 0.2941619137145999,
      "grad_norm": 0.02339404821395874,
      "learning_rate": 0.0012,
      "loss": 2.69635009765625,
      "step": 3628
    },
    {
      "epoch": 0.2943240757398009,
      "grad_norm": 0.028472797945141792,
      "learning_rate": 0.0012,
      "loss": 2.678435802459717,
      "step": 3630
    },
    {
      "epoch": 0.2944862377650019,
      "grad_norm": 0.03564121946692467,
      "learning_rate": 0.0012,
      "loss": 2.6359901428222656,
      "step": 3632
    },
    {
      "epoch": 0.29464839979020285,
      "grad_norm": 0.023458661511540413,
      "learning_rate": 0.0012,
      "loss": 2.684422254562378,
      "step": 3634
    },
    {
      "epoch": 0.2948105618154039,
      "grad_norm": 0.02832651138305664,
      "learning_rate": 0.0012,
      "loss": 2.7294576168060303,
      "step": 3636
    },
    {
      "epoch": 0.29497272384060486,
      "grad_norm": 0.02532975561916828,
      "learning_rate": 0.0012,
      "loss": 2.6911275386810303,
      "step": 3638
    },
    {
      "epoch": 0.29513488586580583,
      "grad_norm": 0.022844597697257996,
      "learning_rate": 0.0012,
      "loss": 2.694549083709717,
      "step": 3640
    },
    {
      "epoch": 0.29529704789100686,
      "grad_norm": 0.03353709727525711,
      "learning_rate": 0.0012,
      "loss": 2.7189459800720215,
      "step": 3642
    },
    {
      "epoch": 0.29545920991620783,
      "grad_norm": 0.025217637419700623,
      "learning_rate": 0.0012,
      "loss": 2.7028591632843018,
      "step": 3644
    },
    {
      "epoch": 0.2956213719414088,
      "grad_norm": 0.02925613336265087,
      "learning_rate": 0.0012,
      "loss": 2.680400848388672,
      "step": 3646
    },
    {
      "epoch": 0.29578353396660984,
      "grad_norm": 0.03541376441717148,
      "learning_rate": 0.0012,
      "loss": 2.7020506858825684,
      "step": 3648
    },
    {
      "epoch": 0.2959456959918108,
      "grad_norm": 0.05031898617744446,
      "learning_rate": 0.0012,
      "loss": 2.6616806983947754,
      "step": 3650
    },
    {
      "epoch": 0.2961078580170118,
      "grad_norm": 0.04350168630480766,
      "learning_rate": 0.0012,
      "loss": 2.674945831298828,
      "step": 3652
    },
    {
      "epoch": 0.2962700200422128,
      "grad_norm": 0.04427933320403099,
      "learning_rate": 0.0012,
      "loss": 2.6760106086730957,
      "step": 3654
    },
    {
      "epoch": 0.2964321820674138,
      "grad_norm": 0.02261785976588726,
      "learning_rate": 0.0012,
      "loss": 2.741924285888672,
      "step": 3656
    },
    {
      "epoch": 0.29659434409261476,
      "grad_norm": 0.04006347432732582,
      "learning_rate": 0.0012,
      "loss": 2.647575616836548,
      "step": 3658
    },
    {
      "epoch": 0.2967565061178158,
      "grad_norm": 0.05429626628756523,
      "learning_rate": 0.0012,
      "loss": 2.6915526390075684,
      "step": 3660
    },
    {
      "epoch": 0.29691866814301676,
      "grad_norm": 0.043431445956230164,
      "learning_rate": 0.0012,
      "loss": 2.695077419281006,
      "step": 3662
    },
    {
      "epoch": 0.2970808301682178,
      "grad_norm": 0.022876404225826263,
      "learning_rate": 0.0012,
      "loss": 2.693634510040283,
      "step": 3664
    },
    {
      "epoch": 0.29724299219341876,
      "grad_norm": 0.026048757135868073,
      "learning_rate": 0.0012,
      "loss": 2.7669689655303955,
      "step": 3666
    },
    {
      "epoch": 0.29740515421861974,
      "grad_norm": 0.02601180411875248,
      "learning_rate": 0.0012,
      "loss": 2.7198028564453125,
      "step": 3668
    },
    {
      "epoch": 0.29756731624382077,
      "grad_norm": 0.02709449641406536,
      "learning_rate": 0.0012,
      "loss": 2.666476249694824,
      "step": 3670
    },
    {
      "epoch": 0.29772947826902174,
      "grad_norm": 0.039310626685619354,
      "learning_rate": 0.0012,
      "loss": 2.669680118560791,
      "step": 3672
    },
    {
      "epoch": 0.2978916402942227,
      "grad_norm": 0.03509191423654556,
      "learning_rate": 0.0012,
      "loss": 2.7227492332458496,
      "step": 3674
    },
    {
      "epoch": 0.29805380231942374,
      "grad_norm": 0.025562595576047897,
      "learning_rate": 0.0012,
      "loss": 2.6683568954467773,
      "step": 3676
    },
    {
      "epoch": 0.2982159643446247,
      "grad_norm": 0.025002365931868553,
      "learning_rate": 0.0012,
      "loss": 2.6933846473693848,
      "step": 3678
    },
    {
      "epoch": 0.2983781263698257,
      "grad_norm": 0.026637600734829903,
      "learning_rate": 0.0012,
      "loss": 2.743687629699707,
      "step": 3680
    },
    {
      "epoch": 0.2985402883950267,
      "grad_norm": 0.02405521087348461,
      "learning_rate": 0.0012,
      "loss": 2.7156505584716797,
      "step": 3682
    },
    {
      "epoch": 0.2987024504202277,
      "grad_norm": 0.05293811112642288,
      "learning_rate": 0.0012,
      "loss": 2.721112012863159,
      "step": 3684
    },
    {
      "epoch": 0.29886461244542867,
      "grad_norm": 0.04536886885762215,
      "learning_rate": 0.0012,
      "loss": 2.7647643089294434,
      "step": 3686
    },
    {
      "epoch": 0.2990267744706297,
      "grad_norm": 0.050010085105895996,
      "learning_rate": 0.0012,
      "loss": 2.6548736095428467,
      "step": 3688
    },
    {
      "epoch": 0.29918893649583067,
      "grad_norm": 0.030322179198265076,
      "learning_rate": 0.0012,
      "loss": 2.734814405441284,
      "step": 3690
    },
    {
      "epoch": 0.29935109852103164,
      "grad_norm": 0.03388204053044319,
      "learning_rate": 0.0012,
      "loss": 2.701995372772217,
      "step": 3692
    },
    {
      "epoch": 0.2995132605462327,
      "grad_norm": 0.028199832886457443,
      "learning_rate": 0.0012,
      "loss": 2.799410820007324,
      "step": 3694
    },
    {
      "epoch": 0.29967542257143365,
      "grad_norm": 0.02324892207980156,
      "learning_rate": 0.0012,
      "loss": 2.713097333908081,
      "step": 3696
    },
    {
      "epoch": 0.2998375845966346,
      "grad_norm": 0.0591513067483902,
      "learning_rate": 0.0012,
      "loss": 2.757615327835083,
      "step": 3698
    },
    {
      "epoch": 0.29999974662183565,
      "grad_norm": 0.05210267752408981,
      "learning_rate": 0.0012,
      "loss": 2.688347339630127,
      "step": 3700
    },
    {
      "epoch": 0.3001619086470366,
      "grad_norm": 0.03342953696846962,
      "learning_rate": 0.0012,
      "loss": 2.6842477321624756,
      "step": 3702
    },
    {
      "epoch": 0.3003240706722376,
      "grad_norm": 0.024021189659833908,
      "learning_rate": 0.0012,
      "loss": 2.7201550006866455,
      "step": 3704
    },
    {
      "epoch": 0.3004862326974386,
      "grad_norm": 0.02403446100652218,
      "learning_rate": 0.0012,
      "loss": 2.7346882820129395,
      "step": 3706
    },
    {
      "epoch": 0.3006483947226396,
      "grad_norm": 0.024291357025504112,
      "learning_rate": 0.0012,
      "loss": 2.717869281768799,
      "step": 3708
    },
    {
      "epoch": 0.3008105567478406,
      "grad_norm": 0.023977352306246758,
      "learning_rate": 0.0012,
      "loss": 2.767469882965088,
      "step": 3710
    },
    {
      "epoch": 0.3009727187730416,
      "grad_norm": 0.04197358340024948,
      "learning_rate": 0.0012,
      "loss": 2.6975953578948975,
      "step": 3712
    },
    {
      "epoch": 0.3011348807982426,
      "grad_norm": 0.05655201897025108,
      "learning_rate": 0.0012,
      "loss": 2.733441114425659,
      "step": 3714
    },
    {
      "epoch": 0.30129704282344355,
      "grad_norm": 0.02409788779914379,
      "learning_rate": 0.0012,
      "loss": 2.7251994609832764,
      "step": 3716
    },
    {
      "epoch": 0.3014592048486446,
      "grad_norm": 0.0362386554479599,
      "learning_rate": 0.0012,
      "loss": 2.709096670150757,
      "step": 3718
    },
    {
      "epoch": 0.30162136687384555,
      "grad_norm": 0.039796024560928345,
      "learning_rate": 0.0012,
      "loss": 2.7431087493896484,
      "step": 3720
    },
    {
      "epoch": 0.3017835288990465,
      "grad_norm": 0.05555464327335358,
      "learning_rate": 0.0012,
      "loss": 2.683638572692871,
      "step": 3722
    },
    {
      "epoch": 0.30194569092424756,
      "grad_norm": 0.04605889320373535,
      "learning_rate": 0.0012,
      "loss": 2.62292742729187,
      "step": 3724
    },
    {
      "epoch": 0.30210785294944853,
      "grad_norm": 0.023300940170884132,
      "learning_rate": 0.0012,
      "loss": 2.689199686050415,
      "step": 3726
    },
    {
      "epoch": 0.3022700149746495,
      "grad_norm": 0.029349539428949356,
      "learning_rate": 0.0012,
      "loss": 2.7296624183654785,
      "step": 3728
    },
    {
      "epoch": 0.30243217699985053,
      "grad_norm": 0.02549344301223755,
      "learning_rate": 0.0012,
      "loss": 2.6400132179260254,
      "step": 3730
    },
    {
      "epoch": 0.3025943390250515,
      "grad_norm": 0.024829773232340813,
      "learning_rate": 0.0012,
      "loss": 2.717426300048828,
      "step": 3732
    },
    {
      "epoch": 0.3027565010502525,
      "grad_norm": 0.04791424423456192,
      "learning_rate": 0.0012,
      "loss": 2.7165684700012207,
      "step": 3734
    },
    {
      "epoch": 0.3029186630754535,
      "grad_norm": 0.046665944159030914,
      "learning_rate": 0.0012,
      "loss": 2.686127185821533,
      "step": 3736
    },
    {
      "epoch": 0.3030808251006545,
      "grad_norm": 0.02407858707010746,
      "learning_rate": 0.0012,
      "loss": 2.6641502380371094,
      "step": 3738
    },
    {
      "epoch": 0.30324298712585546,
      "grad_norm": 0.035174526274204254,
      "learning_rate": 0.0012,
      "loss": 2.78633975982666,
      "step": 3740
    },
    {
      "epoch": 0.3034051491510565,
      "grad_norm": 0.02868524380028248,
      "learning_rate": 0.0012,
      "loss": 2.6539199352264404,
      "step": 3742
    },
    {
      "epoch": 0.30356731117625746,
      "grad_norm": 0.02770049497485161,
      "learning_rate": 0.0012,
      "loss": 2.7306642532348633,
      "step": 3744
    },
    {
      "epoch": 0.30372947320145843,
      "grad_norm": 0.024599019438028336,
      "learning_rate": 0.0012,
      "loss": 2.6750845909118652,
      "step": 3746
    },
    {
      "epoch": 0.30389163522665946,
      "grad_norm": 0.0351044163107872,
      "learning_rate": 0.0012,
      "loss": 2.7023439407348633,
      "step": 3748
    },
    {
      "epoch": 0.30405379725186044,
      "grad_norm": 0.03264794498682022,
      "learning_rate": 0.0012,
      "loss": 2.7034811973571777,
      "step": 3750
    },
    {
      "epoch": 0.3042159592770614,
      "grad_norm": 0.026273315772414207,
      "learning_rate": 0.0012,
      "loss": 2.692228317260742,
      "step": 3752
    },
    {
      "epoch": 0.30437812130226244,
      "grad_norm": 0.025718839839100838,
      "learning_rate": 0.0012,
      "loss": 2.7083544731140137,
      "step": 3754
    },
    {
      "epoch": 0.3045402833274634,
      "grad_norm": 0.042023856192827225,
      "learning_rate": 0.0012,
      "loss": 2.7299652099609375,
      "step": 3756
    },
    {
      "epoch": 0.3047024453526644,
      "grad_norm": 0.025624427944421768,
      "learning_rate": 0.0012,
      "loss": 2.7034225463867188,
      "step": 3758
    },
    {
      "epoch": 0.3048646073778654,
      "grad_norm": 0.02557194232940674,
      "learning_rate": 0.0012,
      "loss": 2.7125327587127686,
      "step": 3760
    },
    {
      "epoch": 0.3050267694030664,
      "grad_norm": 0.028254183009266853,
      "learning_rate": 0.0012,
      "loss": 2.583076000213623,
      "step": 3762
    },
    {
      "epoch": 0.30518893142826736,
      "grad_norm": 0.02969275414943695,
      "learning_rate": 0.0012,
      "loss": 2.674924373626709,
      "step": 3764
    },
    {
      "epoch": 0.3053510934534684,
      "grad_norm": 0.023198354989290237,
      "learning_rate": 0.0012,
      "loss": 2.6776323318481445,
      "step": 3766
    },
    {
      "epoch": 0.30551325547866937,
      "grad_norm": 0.028331594541668892,
      "learning_rate": 0.0012,
      "loss": 2.7228565216064453,
      "step": 3768
    },
    {
      "epoch": 0.30567541750387034,
      "grad_norm": 0.029825031757354736,
      "learning_rate": 0.0012,
      "loss": 2.733900547027588,
      "step": 3770
    },
    {
      "epoch": 0.30583757952907137,
      "grad_norm": 0.03939323127269745,
      "learning_rate": 0.0012,
      "loss": 2.7257180213928223,
      "step": 3772
    },
    {
      "epoch": 0.30599974155427234,
      "grad_norm": 0.0222818274050951,
      "learning_rate": 0.0012,
      "loss": 2.7190232276916504,
      "step": 3774
    },
    {
      "epoch": 0.3061619035794733,
      "grad_norm": 0.027049144729971886,
      "learning_rate": 0.0012,
      "loss": 2.673417568206787,
      "step": 3776
    },
    {
      "epoch": 0.30632406560467434,
      "grad_norm": 0.06094063073396683,
      "learning_rate": 0.0012,
      "loss": 2.706291437149048,
      "step": 3778
    },
    {
      "epoch": 0.3064862276298753,
      "grad_norm": 0.03126060962677002,
      "learning_rate": 0.0012,
      "loss": 2.7614994049072266,
      "step": 3780
    },
    {
      "epoch": 0.3066483896550763,
      "grad_norm": 0.04172632843255997,
      "learning_rate": 0.0012,
      "loss": 2.6919021606445312,
      "step": 3782
    },
    {
      "epoch": 0.3068105516802773,
      "grad_norm": 0.022776680067181587,
      "learning_rate": 0.0012,
      "loss": 2.7068276405334473,
      "step": 3784
    },
    {
      "epoch": 0.3069727137054783,
      "grad_norm": 0.024387851357460022,
      "learning_rate": 0.0012,
      "loss": 2.709970474243164,
      "step": 3786
    },
    {
      "epoch": 0.30713487573067927,
      "grad_norm": 0.025880644097924232,
      "learning_rate": 0.0012,
      "loss": 2.695392608642578,
      "step": 3788
    },
    {
      "epoch": 0.3072970377558803,
      "grad_norm": 0.028253119438886642,
      "learning_rate": 0.0012,
      "loss": 2.6640372276306152,
      "step": 3790
    },
    {
      "epoch": 0.30745919978108127,
      "grad_norm": 0.03030557557940483,
      "learning_rate": 0.0012,
      "loss": 2.693894624710083,
      "step": 3792
    },
    {
      "epoch": 0.30762136180628225,
      "grad_norm": 0.042788103222846985,
      "learning_rate": 0.0012,
      "loss": 2.752267837524414,
      "step": 3794
    },
    {
      "epoch": 0.3077835238314833,
      "grad_norm": 0.02604059875011444,
      "learning_rate": 0.0012,
      "loss": 2.6545286178588867,
      "step": 3796
    },
    {
      "epoch": 0.30794568585668425,
      "grad_norm": 0.02681553177535534,
      "learning_rate": 0.0012,
      "loss": 2.6789164543151855,
      "step": 3798
    },
    {
      "epoch": 0.3081078478818852,
      "grad_norm": 0.04500815272331238,
      "learning_rate": 0.0012,
      "loss": 2.716567039489746,
      "step": 3800
    },
    {
      "epoch": 0.30827000990708625,
      "grad_norm": 0.039189212024211884,
      "learning_rate": 0.0012,
      "loss": 2.670332431793213,
      "step": 3802
    },
    {
      "epoch": 0.3084321719322872,
      "grad_norm": 0.024372821673750877,
      "learning_rate": 0.0012,
      "loss": 2.6647143363952637,
      "step": 3804
    },
    {
      "epoch": 0.3085943339574882,
      "grad_norm": 0.022618554532527924,
      "learning_rate": 0.0012,
      "loss": 2.649052619934082,
      "step": 3806
    },
    {
      "epoch": 0.3087564959826892,
      "grad_norm": 0.03050205484032631,
      "learning_rate": 0.0012,
      "loss": 2.746875286102295,
      "step": 3808
    },
    {
      "epoch": 0.3089186580078902,
      "grad_norm": 0.02441958338022232,
      "learning_rate": 0.0012,
      "loss": 2.644151210784912,
      "step": 3810
    },
    {
      "epoch": 0.3090808200330912,
      "grad_norm": 0.023843029513955116,
      "learning_rate": 0.0012,
      "loss": 2.716470241546631,
      "step": 3812
    },
    {
      "epoch": 0.3092429820582922,
      "grad_norm": 0.053568627685308456,
      "learning_rate": 0.0012,
      "loss": 2.730515718460083,
      "step": 3814
    },
    {
      "epoch": 0.3094051440834932,
      "grad_norm": 0.04697383567690849,
      "learning_rate": 0.0012,
      "loss": 2.683626174926758,
      "step": 3816
    },
    {
      "epoch": 0.30956730610869415,
      "grad_norm": 0.06416673213243484,
      "learning_rate": 0.0012,
      "loss": 2.691037178039551,
      "step": 3818
    },
    {
      "epoch": 0.3097294681338952,
      "grad_norm": 0.04794636368751526,
      "learning_rate": 0.0012,
      "loss": 2.7614150047302246,
      "step": 3820
    },
    {
      "epoch": 0.30989163015909615,
      "grad_norm": 0.049511414021253586,
      "learning_rate": 0.0012,
      "loss": 2.6627089977264404,
      "step": 3822
    },
    {
      "epoch": 0.31005379218429713,
      "grad_norm": 0.02292368933558464,
      "learning_rate": 0.0012,
      "loss": 2.669041156768799,
      "step": 3824
    },
    {
      "epoch": 0.31021595420949816,
      "grad_norm": 0.024858007207512856,
      "learning_rate": 0.0012,
      "loss": 2.751755714416504,
      "step": 3826
    },
    {
      "epoch": 0.31037811623469913,
      "grad_norm": 0.02785998396575451,
      "learning_rate": 0.0012,
      "loss": 2.736415386199951,
      "step": 3828
    },
    {
      "epoch": 0.3105402782599001,
      "grad_norm": 0.029795218259096146,
      "learning_rate": 0.0012,
      "loss": 2.749938488006592,
      "step": 3830
    },
    {
      "epoch": 0.31070244028510113,
      "grad_norm": 0.045968540012836456,
      "learning_rate": 0.0012,
      "loss": 2.6863725185394287,
      "step": 3832
    },
    {
      "epoch": 0.3108646023103021,
      "grad_norm": 0.022855881601572037,
      "learning_rate": 0.0012,
      "loss": 2.743227958679199,
      "step": 3834
    },
    {
      "epoch": 0.3110267643355031,
      "grad_norm": 0.027005968615412712,
      "learning_rate": 0.0012,
      "loss": 2.7069528102874756,
      "step": 3836
    },
    {
      "epoch": 0.3111889263607041,
      "grad_norm": 0.029173657298088074,
      "learning_rate": 0.0012,
      "loss": 2.6854825019836426,
      "step": 3838
    },
    {
      "epoch": 0.3113510883859051,
      "grad_norm": 0.025054017081856728,
      "learning_rate": 0.0012,
      "loss": 2.6591310501098633,
      "step": 3840
    },
    {
      "epoch": 0.31151325041110606,
      "grad_norm": 0.05583212152123451,
      "learning_rate": 0.0012,
      "loss": 2.753845691680908,
      "step": 3842
    },
    {
      "epoch": 0.3116754124363071,
      "grad_norm": 0.05816887319087982,
      "learning_rate": 0.0012,
      "loss": 2.7082576751708984,
      "step": 3844
    },
    {
      "epoch": 0.31183757446150806,
      "grad_norm": 0.02601740136742592,
      "learning_rate": 0.0012,
      "loss": 2.6716644763946533,
      "step": 3846
    },
    {
      "epoch": 0.31199973648670903,
      "grad_norm": 0.04756230488419533,
      "learning_rate": 0.0012,
      "loss": 2.669003963470459,
      "step": 3848
    },
    {
      "epoch": 0.31216189851191006,
      "grad_norm": 0.07218702882528305,
      "learning_rate": 0.0012,
      "loss": 2.679992437362671,
      "step": 3850
    },
    {
      "epoch": 0.31232406053711104,
      "grad_norm": 0.053871918469667435,
      "learning_rate": 0.0012,
      "loss": 2.698293447494507,
      "step": 3852
    },
    {
      "epoch": 0.312486222562312,
      "grad_norm": 0.030948275700211525,
      "learning_rate": 0.0012,
      "loss": 2.7368717193603516,
      "step": 3854
    },
    {
      "epoch": 0.31264838458751304,
      "grad_norm": 0.05761713162064552,
      "learning_rate": 0.0012,
      "loss": 2.651543140411377,
      "step": 3856
    },
    {
      "epoch": 0.312810546612714,
      "grad_norm": 0.054613836109638214,
      "learning_rate": 0.0012,
      "loss": 2.723102569580078,
      "step": 3858
    },
    {
      "epoch": 0.312972708637915,
      "grad_norm": 0.032494816929101944,
      "learning_rate": 0.0012,
      "loss": 2.6783580780029297,
      "step": 3860
    },
    {
      "epoch": 0.313134870663116,
      "grad_norm": 0.04250982403755188,
      "learning_rate": 0.0012,
      "loss": 2.7360970973968506,
      "step": 3862
    },
    {
      "epoch": 0.313297032688317,
      "grad_norm": 0.03387052193284035,
      "learning_rate": 0.0012,
      "loss": 2.6936049461364746,
      "step": 3864
    },
    {
      "epoch": 0.31345919471351796,
      "grad_norm": 0.0386977419257164,
      "learning_rate": 0.0012,
      "loss": 2.702322006225586,
      "step": 3866
    },
    {
      "epoch": 0.313621356738719,
      "grad_norm": 0.0661354809999466,
      "learning_rate": 0.0012,
      "loss": 2.660881280899048,
      "step": 3868
    },
    {
      "epoch": 0.31378351876391997,
      "grad_norm": 0.044697828590869904,
      "learning_rate": 0.0012,
      "loss": 2.748222827911377,
      "step": 3870
    },
    {
      "epoch": 0.31394568078912094,
      "grad_norm": 0.035948850214481354,
      "learning_rate": 0.0012,
      "loss": 2.7098724842071533,
      "step": 3872
    },
    {
      "epoch": 0.31410784281432197,
      "grad_norm": 0.06279655545949936,
      "learning_rate": 0.0012,
      "loss": 2.690635919570923,
      "step": 3874
    },
    {
      "epoch": 0.31427000483952294,
      "grad_norm": 0.0443095862865448,
      "learning_rate": 0.0012,
      "loss": 2.7090022563934326,
      "step": 3876
    },
    {
      "epoch": 0.3144321668647239,
      "grad_norm": 0.021667925640940666,
      "learning_rate": 0.0012,
      "loss": 2.7256593704223633,
      "step": 3878
    },
    {
      "epoch": 0.31459432888992495,
      "grad_norm": 0.04149143025279045,
      "learning_rate": 0.0012,
      "loss": 2.7211639881134033,
      "step": 3880
    },
    {
      "epoch": 0.3147564909151259,
      "grad_norm": 0.03237299993634224,
      "learning_rate": 0.0012,
      "loss": 2.685361623764038,
      "step": 3882
    },
    {
      "epoch": 0.3149186529403269,
      "grad_norm": 0.032661259174346924,
      "learning_rate": 0.0012,
      "loss": 2.678969383239746,
      "step": 3884
    },
    {
      "epoch": 0.3150808149655279,
      "grad_norm": 0.04987683892250061,
      "learning_rate": 0.0012,
      "loss": 2.7165088653564453,
      "step": 3886
    },
    {
      "epoch": 0.3152429769907289,
      "grad_norm": 0.02749650739133358,
      "learning_rate": 0.0012,
      "loss": 2.7451157569885254,
      "step": 3888
    },
    {
      "epoch": 0.31540513901592987,
      "grad_norm": 0.03411014378070831,
      "learning_rate": 0.0012,
      "loss": 2.6786155700683594,
      "step": 3890
    },
    {
      "epoch": 0.3155673010411309,
      "grad_norm": 0.06097760796546936,
      "learning_rate": 0.0012,
      "loss": 2.7631354331970215,
      "step": 3892
    },
    {
      "epoch": 0.3157294630663319,
      "grad_norm": 0.05750631168484688,
      "learning_rate": 0.0012,
      "loss": 2.7177300453186035,
      "step": 3894
    },
    {
      "epoch": 0.31589162509153285,
      "grad_norm": 0.02418399229645729,
      "learning_rate": 0.0012,
      "loss": 2.7000651359558105,
      "step": 3896
    },
    {
      "epoch": 0.3160537871167339,
      "grad_norm": 0.06676550954580307,
      "learning_rate": 0.0012,
      "loss": 2.690202236175537,
      "step": 3898
    },
    {
      "epoch": 0.31621594914193485,
      "grad_norm": 0.023713983595371246,
      "learning_rate": 0.0012,
      "loss": 2.6911728382110596,
      "step": 3900
    },
    {
      "epoch": 0.3163781111671358,
      "grad_norm": 0.03482888638973236,
      "learning_rate": 0.0012,
      "loss": 2.7403368949890137,
      "step": 3902
    },
    {
      "epoch": 0.31654027319233685,
      "grad_norm": 0.07559531927108765,
      "learning_rate": 0.0012,
      "loss": 2.667371988296509,
      "step": 3904
    },
    {
      "epoch": 0.3167024352175378,
      "grad_norm": 0.04674387723207474,
      "learning_rate": 0.0012,
      "loss": 2.7321043014526367,
      "step": 3906
    },
    {
      "epoch": 0.3168645972427388,
      "grad_norm": 0.03008096106350422,
      "learning_rate": 0.0012,
      "loss": 2.7375149726867676,
      "step": 3908
    },
    {
      "epoch": 0.31702675926793983,
      "grad_norm": 0.03420838341116905,
      "learning_rate": 0.0012,
      "loss": 2.7357869148254395,
      "step": 3910
    },
    {
      "epoch": 0.3171889212931408,
      "grad_norm": 0.031565070152282715,
      "learning_rate": 0.0012,
      "loss": 2.726573944091797,
      "step": 3912
    },
    {
      "epoch": 0.3173510833183418,
      "grad_norm": 0.03924984857439995,
      "learning_rate": 0.0012,
      "loss": 2.756632089614868,
      "step": 3914
    },
    {
      "epoch": 0.3175132453435428,
      "grad_norm": 0.024420689791440964,
      "learning_rate": 0.0012,
      "loss": 2.6168975830078125,
      "step": 3916
    },
    {
      "epoch": 0.3176754073687438,
      "grad_norm": 0.02620626799762249,
      "learning_rate": 0.0012,
      "loss": 2.6857752799987793,
      "step": 3918
    },
    {
      "epoch": 0.31783756939394475,
      "grad_norm": 0.028034795075654984,
      "learning_rate": 0.0012,
      "loss": 2.7361645698547363,
      "step": 3920
    },
    {
      "epoch": 0.3179997314191458,
      "grad_norm": 0.02264219895005226,
      "learning_rate": 0.0012,
      "loss": 2.687436580657959,
      "step": 3922
    },
    {
      "epoch": 0.31816189344434676,
      "grad_norm": 0.03352418169379234,
      "learning_rate": 0.0012,
      "loss": 2.659391403198242,
      "step": 3924
    },
    {
      "epoch": 0.31832405546954773,
      "grad_norm": 0.02360348217189312,
      "learning_rate": 0.0012,
      "loss": 2.6756582260131836,
      "step": 3926
    },
    {
      "epoch": 0.31848621749474876,
      "grad_norm": 0.0293061975389719,
      "learning_rate": 0.0012,
      "loss": 2.645812749862671,
      "step": 3928
    },
    {
      "epoch": 0.31864837951994973,
      "grad_norm": 0.03540433943271637,
      "learning_rate": 0.0012,
      "loss": 2.7024598121643066,
      "step": 3930
    },
    {
      "epoch": 0.3188105415451507,
      "grad_norm": 0.024899544194340706,
      "learning_rate": 0.0012,
      "loss": 2.716977834701538,
      "step": 3932
    },
    {
      "epoch": 0.31897270357035173,
      "grad_norm": 0.026198582723736763,
      "learning_rate": 0.0012,
      "loss": 2.7317276000976562,
      "step": 3934
    },
    {
      "epoch": 0.3191348655955527,
      "grad_norm": 0.025045795366168022,
      "learning_rate": 0.0012,
      "loss": 2.705200672149658,
      "step": 3936
    },
    {
      "epoch": 0.3192970276207537,
      "grad_norm": 0.02377483993768692,
      "learning_rate": 0.0012,
      "loss": 2.65891170501709,
      "step": 3938
    },
    {
      "epoch": 0.3194591896459547,
      "grad_norm": 0.03066643886268139,
      "learning_rate": 0.0012,
      "loss": 2.5848464965820312,
      "step": 3940
    },
    {
      "epoch": 0.3196213516711557,
      "grad_norm": 0.02842777594923973,
      "learning_rate": 0.0012,
      "loss": 2.6733100414276123,
      "step": 3942
    },
    {
      "epoch": 0.31978351369635666,
      "grad_norm": 0.0376637801527977,
      "learning_rate": 0.0012,
      "loss": 2.7369213104248047,
      "step": 3944
    },
    {
      "epoch": 0.3199456757215577,
      "grad_norm": 0.039570312947034836,
      "learning_rate": 0.0012,
      "loss": 2.659731864929199,
      "step": 3946
    },
    {
      "epoch": 0.32010783774675866,
      "grad_norm": 0.02870190516114235,
      "learning_rate": 0.0012,
      "loss": 2.692413330078125,
      "step": 3948
    },
    {
      "epoch": 0.32026999977195963,
      "grad_norm": 0.025957081466913223,
      "learning_rate": 0.0012,
      "loss": 2.7083284854888916,
      "step": 3950
    },
    {
      "epoch": 0.32043216179716066,
      "grad_norm": 0.0276778731495142,
      "learning_rate": 0.0012,
      "loss": 2.6810684204101562,
      "step": 3952
    },
    {
      "epoch": 0.32059432382236164,
      "grad_norm": 0.029137808829545975,
      "learning_rate": 0.0012,
      "loss": 2.681272268295288,
      "step": 3954
    },
    {
      "epoch": 0.3207564858475626,
      "grad_norm": 0.026820801198482513,
      "learning_rate": 0.0012,
      "loss": 2.687767505645752,
      "step": 3956
    },
    {
      "epoch": 0.32091864787276364,
      "grad_norm": 0.04154570773243904,
      "learning_rate": 0.0012,
      "loss": 2.7384495735168457,
      "step": 3958
    },
    {
      "epoch": 0.3210808098979646,
      "grad_norm": 0.02963838167488575,
      "learning_rate": 0.0012,
      "loss": 2.6386983394622803,
      "step": 3960
    },
    {
      "epoch": 0.3212429719231656,
      "grad_norm": 0.0485372431576252,
      "learning_rate": 0.0012,
      "loss": 2.694655418395996,
      "step": 3962
    },
    {
      "epoch": 0.3214051339483666,
      "grad_norm": 0.027584413066506386,
      "learning_rate": 0.0012,
      "loss": 2.6704959869384766,
      "step": 3964
    },
    {
      "epoch": 0.3215672959735676,
      "grad_norm": 0.02528453804552555,
      "learning_rate": 0.0012,
      "loss": 2.651270866394043,
      "step": 3966
    },
    {
      "epoch": 0.32172945799876856,
      "grad_norm": 0.024877000600099564,
      "learning_rate": 0.0012,
      "loss": 2.677248954772949,
      "step": 3968
    },
    {
      "epoch": 0.3218916200239696,
      "grad_norm": 0.03933575004339218,
      "learning_rate": 0.0012,
      "loss": 2.742457866668701,
      "step": 3970
    },
    {
      "epoch": 0.32205378204917057,
      "grad_norm": 0.027232497930526733,
      "learning_rate": 0.0012,
      "loss": 2.6652469635009766,
      "step": 3972
    },
    {
      "epoch": 0.32221594407437154,
      "grad_norm": 0.028630372136831284,
      "learning_rate": 0.0012,
      "loss": 2.6839804649353027,
      "step": 3974
    },
    {
      "epoch": 0.32237810609957257,
      "grad_norm": 0.029072068631649017,
      "learning_rate": 0.0012,
      "loss": 2.662029266357422,
      "step": 3976
    },
    {
      "epoch": 0.32254026812477354,
      "grad_norm": 0.026241011917591095,
      "learning_rate": 0.0012,
      "loss": 2.7335855960845947,
      "step": 3978
    },
    {
      "epoch": 0.3227024301499745,
      "grad_norm": 0.037648092955350876,
      "learning_rate": 0.0012,
      "loss": 2.722362995147705,
      "step": 3980
    },
    {
      "epoch": 0.32286459217517555,
      "grad_norm": 0.029527999460697174,
      "learning_rate": 0.0012,
      "loss": 2.705349922180176,
      "step": 3982
    },
    {
      "epoch": 0.3230267542003765,
      "grad_norm": 0.03933212161064148,
      "learning_rate": 0.0012,
      "loss": 2.6994524002075195,
      "step": 3984
    },
    {
      "epoch": 0.3231889162255775,
      "grad_norm": 0.0385088287293911,
      "learning_rate": 0.0012,
      "loss": 2.7076468467712402,
      "step": 3986
    },
    {
      "epoch": 0.3233510782507785,
      "grad_norm": 0.0263025164604187,
      "learning_rate": 0.0012,
      "loss": 2.6752357482910156,
      "step": 3988
    },
    {
      "epoch": 0.3235132402759795,
      "grad_norm": 0.04427499696612358,
      "learning_rate": 0.0012,
      "loss": 2.6727871894836426,
      "step": 3990
    },
    {
      "epoch": 0.32367540230118047,
      "grad_norm": 0.026519354432821274,
      "learning_rate": 0.0012,
      "loss": 2.626192331314087,
      "step": 3992
    },
    {
      "epoch": 0.3238375643263815,
      "grad_norm": 0.02737540379166603,
      "learning_rate": 0.0012,
      "loss": 2.6988868713378906,
      "step": 3994
    },
    {
      "epoch": 0.3239997263515825,
      "grad_norm": 0.022153133526444435,
      "learning_rate": 0.0012,
      "loss": 2.6514816284179688,
      "step": 3996
    },
    {
      "epoch": 0.32416188837678345,
      "grad_norm": 0.023441309109330177,
      "learning_rate": 0.0012,
      "loss": 2.6688244342803955,
      "step": 3998
    },
    {
      "epoch": 0.3243240504019845,
      "grad_norm": 0.024457532912492752,
      "learning_rate": 0.0012,
      "loss": 2.6981468200683594,
      "step": 4000
    },
    {
      "epoch": 0.32448621242718545,
      "grad_norm": 0.0254792720079422,
      "learning_rate": 0.0012,
      "loss": 2.7103841304779053,
      "step": 4002
    },
    {
      "epoch": 0.3246483744523864,
      "grad_norm": 0.039224252104759216,
      "learning_rate": 0.0012,
      "loss": 2.6745429039001465,
      "step": 4004
    },
    {
      "epoch": 0.32481053647758745,
      "grad_norm": 0.03335333243012428,
      "learning_rate": 0.0012,
      "loss": 2.5991339683532715,
      "step": 4006
    },
    {
      "epoch": 0.3249726985027884,
      "grad_norm": 0.025135602802038193,
      "learning_rate": 0.0012,
      "loss": 2.673530101776123,
      "step": 4008
    },
    {
      "epoch": 0.3251348605279894,
      "grad_norm": 0.028662320226430893,
      "learning_rate": 0.0012,
      "loss": 2.6993770599365234,
      "step": 4010
    },
    {
      "epoch": 0.32529702255319043,
      "grad_norm": 0.04759817197918892,
      "learning_rate": 0.0012,
      "loss": 2.708803653717041,
      "step": 4012
    },
    {
      "epoch": 0.3254591845783914,
      "grad_norm": 0.04556750878691673,
      "learning_rate": 0.0012,
      "loss": 2.689265489578247,
      "step": 4014
    },
    {
      "epoch": 0.3256213466035924,
      "grad_norm": 0.028755977749824524,
      "learning_rate": 0.0012,
      "loss": 2.67560076713562,
      "step": 4016
    },
    {
      "epoch": 0.3257835086287934,
      "grad_norm": 0.022281857207417488,
      "learning_rate": 0.0012,
      "loss": 2.717399835586548,
      "step": 4018
    },
    {
      "epoch": 0.3259456706539944,
      "grad_norm": 0.029895232990384102,
      "learning_rate": 0.0012,
      "loss": 2.6738834381103516,
      "step": 4020
    },
    {
      "epoch": 0.32610783267919535,
      "grad_norm": 0.02988537773489952,
      "learning_rate": 0.0012,
      "loss": 2.6490983963012695,
      "step": 4022
    },
    {
      "epoch": 0.3262699947043964,
      "grad_norm": 0.03310006856918335,
      "learning_rate": 0.0012,
      "loss": 2.762295722961426,
      "step": 4024
    },
    {
      "epoch": 0.32643215672959736,
      "grad_norm": 0.05105002969503403,
      "learning_rate": 0.0012,
      "loss": 2.725933790206909,
      "step": 4026
    },
    {
      "epoch": 0.32659431875479833,
      "grad_norm": 0.061814773827791214,
      "learning_rate": 0.0012,
      "loss": 2.7230491638183594,
      "step": 4028
    },
    {
      "epoch": 0.32675648077999936,
      "grad_norm": 0.03495914861559868,
      "learning_rate": 0.0012,
      "loss": 2.675342082977295,
      "step": 4030
    },
    {
      "epoch": 0.32691864280520033,
      "grad_norm": 0.030521871522068977,
      "learning_rate": 0.0012,
      "loss": 2.762193202972412,
      "step": 4032
    },
    {
      "epoch": 0.3270808048304013,
      "grad_norm": 0.02494516223669052,
      "learning_rate": 0.0012,
      "loss": 2.6463069915771484,
      "step": 4034
    },
    {
      "epoch": 0.32724296685560234,
      "grad_norm": 0.0594274178147316,
      "learning_rate": 0.0012,
      "loss": 2.667158603668213,
      "step": 4036
    },
    {
      "epoch": 0.3274051288808033,
      "grad_norm": 0.024581247940659523,
      "learning_rate": 0.0012,
      "loss": 2.690526008605957,
      "step": 4038
    },
    {
      "epoch": 0.3275672909060043,
      "grad_norm": 0.03437405452132225,
      "learning_rate": 0.0012,
      "loss": 2.704155683517456,
      "step": 4040
    },
    {
      "epoch": 0.3277294529312053,
      "grad_norm": 0.025397779420018196,
      "learning_rate": 0.0012,
      "loss": 2.683350086212158,
      "step": 4042
    },
    {
      "epoch": 0.3278916149564063,
      "grad_norm": 0.02932305447757244,
      "learning_rate": 0.0012,
      "loss": 2.6526036262512207,
      "step": 4044
    },
    {
      "epoch": 0.32805377698160726,
      "grad_norm": 0.026818087324500084,
      "learning_rate": 0.0012,
      "loss": 2.8019943237304688,
      "step": 4046
    },
    {
      "epoch": 0.3282159390068083,
      "grad_norm": 0.03312470018863678,
      "learning_rate": 0.0012,
      "loss": 2.668884754180908,
      "step": 4048
    },
    {
      "epoch": 0.32837810103200926,
      "grad_norm": 0.03557547554373741,
      "learning_rate": 0.0012,
      "loss": 2.720975399017334,
      "step": 4050
    },
    {
      "epoch": 0.32854026305721024,
      "grad_norm": 0.044299062341451645,
      "learning_rate": 0.0012,
      "loss": 2.700623035430908,
      "step": 4052
    },
    {
      "epoch": 0.32870242508241126,
      "grad_norm": 0.06346964091062546,
      "learning_rate": 0.0012,
      "loss": 2.7567641735076904,
      "step": 4054
    },
    {
      "epoch": 0.32886458710761224,
      "grad_norm": 0.026318617165088654,
      "learning_rate": 0.0012,
      "loss": 2.713080406188965,
      "step": 4056
    },
    {
      "epoch": 0.3290267491328132,
      "grad_norm": 0.030698137357831,
      "learning_rate": 0.0012,
      "loss": 2.6577508449554443,
      "step": 4058
    },
    {
      "epoch": 0.32918891115801424,
      "grad_norm": 0.026593206450343132,
      "learning_rate": 0.0012,
      "loss": 2.6737022399902344,
      "step": 4060
    },
    {
      "epoch": 0.3293510731832152,
      "grad_norm": 0.03199136629700661,
      "learning_rate": 0.0012,
      "loss": 2.6899139881134033,
      "step": 4062
    },
    {
      "epoch": 0.3295132352084162,
      "grad_norm": 0.024911193177103996,
      "learning_rate": 0.0012,
      "loss": 2.6872286796569824,
      "step": 4064
    },
    {
      "epoch": 0.3296753972336172,
      "grad_norm": 0.03435820713639259,
      "learning_rate": 0.0012,
      "loss": 2.6864147186279297,
      "step": 4066
    },
    {
      "epoch": 0.3298375592588182,
      "grad_norm": 0.028654159978032112,
      "learning_rate": 0.0012,
      "loss": 2.673910617828369,
      "step": 4068
    },
    {
      "epoch": 0.32999972128401917,
      "grad_norm": 0.02483632229268551,
      "learning_rate": 0.0012,
      "loss": 2.749208927154541,
      "step": 4070
    },
    {
      "epoch": 0.3301618833092202,
      "grad_norm": 0.03720923513174057,
      "learning_rate": 0.0012,
      "loss": 2.7306466102600098,
      "step": 4072
    },
    {
      "epoch": 0.33032404533442117,
      "grad_norm": 0.04665082320570946,
      "learning_rate": 0.0012,
      "loss": 2.640836238861084,
      "step": 4074
    },
    {
      "epoch": 0.33048620735962214,
      "grad_norm": 0.031417116522789,
      "learning_rate": 0.0012,
      "loss": 2.69173526763916,
      "step": 4076
    },
    {
      "epoch": 0.33064836938482317,
      "grad_norm": 0.030122363939881325,
      "learning_rate": 0.0012,
      "loss": 2.6855297088623047,
      "step": 4078
    },
    {
      "epoch": 0.33081053141002414,
      "grad_norm": 0.04317516088485718,
      "learning_rate": 0.0012,
      "loss": 2.6892542839050293,
      "step": 4080
    },
    {
      "epoch": 0.3309726934352251,
      "grad_norm": 0.03332914784550667,
      "learning_rate": 0.0012,
      "loss": 2.6664929389953613,
      "step": 4082
    },
    {
      "epoch": 0.33113485546042615,
      "grad_norm": 0.04580341652035713,
      "learning_rate": 0.0012,
      "loss": 2.7104411125183105,
      "step": 4084
    },
    {
      "epoch": 0.3312970174856271,
      "grad_norm": 0.06663374602794647,
      "learning_rate": 0.0012,
      "loss": 2.771939277648926,
      "step": 4086
    },
    {
      "epoch": 0.3314591795108281,
      "grad_norm": 0.046179886907339096,
      "learning_rate": 0.0012,
      "loss": 2.672429084777832,
      "step": 4088
    },
    {
      "epoch": 0.3316213415360291,
      "grad_norm": 0.025597969070076942,
      "learning_rate": 0.0012,
      "loss": 2.653226375579834,
      "step": 4090
    },
    {
      "epoch": 0.3317835035612301,
      "grad_norm": 0.07005797326564789,
      "learning_rate": 0.0012,
      "loss": 2.7293760776519775,
      "step": 4092
    },
    {
      "epoch": 0.33194566558643107,
      "grad_norm": 0.03085990995168686,
      "learning_rate": 0.0012,
      "loss": 2.6748294830322266,
      "step": 4094
    },
    {
      "epoch": 0.3321078276116321,
      "grad_norm": 0.029840674251317978,
      "learning_rate": 0.0012,
      "loss": 2.6957039833068848,
      "step": 4096
    },
    {
      "epoch": 0.3322699896368331,
      "grad_norm": 0.02744124084711075,
      "learning_rate": 0.0012,
      "loss": 2.6816699504852295,
      "step": 4098
    },
    {
      "epoch": 0.33243215166203405,
      "grad_norm": 0.0348665751516819,
      "learning_rate": 0.0012,
      "loss": 2.7123632431030273,
      "step": 4100
    },
    {
      "epoch": 0.3325943136872351,
      "grad_norm": 0.026478439569473267,
      "learning_rate": 0.0012,
      "loss": 2.6721513271331787,
      "step": 4102
    },
    {
      "epoch": 0.33275647571243605,
      "grad_norm": 0.02796625718474388,
      "learning_rate": 0.0012,
      "loss": 2.625600814819336,
      "step": 4104
    },
    {
      "epoch": 0.332918637737637,
      "grad_norm": 0.024616587907075882,
      "learning_rate": 0.0012,
      "loss": 2.6933746337890625,
      "step": 4106
    },
    {
      "epoch": 0.33308079976283805,
      "grad_norm": 0.023638945072889328,
      "learning_rate": 0.0012,
      "loss": 2.6719560623168945,
      "step": 4108
    },
    {
      "epoch": 0.333242961788039,
      "grad_norm": 0.053053561598062515,
      "learning_rate": 0.0012,
      "loss": 2.7124111652374268,
      "step": 4110
    },
    {
      "epoch": 0.33340512381324,
      "grad_norm": 0.04956590011715889,
      "learning_rate": 0.0012,
      "loss": 2.619550943374634,
      "step": 4112
    },
    {
      "epoch": 0.33356728583844103,
      "grad_norm": 0.02516121044754982,
      "learning_rate": 0.0012,
      "loss": 2.7118892669677734,
      "step": 4114
    },
    {
      "epoch": 0.333729447863642,
      "grad_norm": 0.02304116077721119,
      "learning_rate": 0.0012,
      "loss": 2.690077781677246,
      "step": 4116
    },
    {
      "epoch": 0.333891609888843,
      "grad_norm": 0.024999845772981644,
      "learning_rate": 0.0012,
      "loss": 2.679293394088745,
      "step": 4118
    },
    {
      "epoch": 0.334053771914044,
      "grad_norm": 0.043608538806438446,
      "learning_rate": 0.0012,
      "loss": 2.722790241241455,
      "step": 4120
    },
    {
      "epoch": 0.334215933939245,
      "grad_norm": 0.04557528346776962,
      "learning_rate": 0.0012,
      "loss": 2.750230312347412,
      "step": 4122
    },
    {
      "epoch": 0.33437809596444595,
      "grad_norm": 0.023795122280716896,
      "learning_rate": 0.0012,
      "loss": 2.690037727355957,
      "step": 4124
    },
    {
      "epoch": 0.334540257989647,
      "grad_norm": 0.03397229313850403,
      "learning_rate": 0.0012,
      "loss": 2.6693105697631836,
      "step": 4126
    },
    {
      "epoch": 0.33470242001484796,
      "grad_norm": 0.039219360798597336,
      "learning_rate": 0.0012,
      "loss": 2.699524402618408,
      "step": 4128
    },
    {
      "epoch": 0.33486458204004893,
      "grad_norm": 0.023970555514097214,
      "learning_rate": 0.0012,
      "loss": 2.6712048053741455,
      "step": 4130
    },
    {
      "epoch": 0.33502674406524996,
      "grad_norm": 0.024630066007375717,
      "learning_rate": 0.0012,
      "loss": 2.7105536460876465,
      "step": 4132
    },
    {
      "epoch": 0.33518890609045093,
      "grad_norm": 0.061389680951833725,
      "learning_rate": 0.0012,
      "loss": 2.6748528480529785,
      "step": 4134
    },
    {
      "epoch": 0.3353510681156519,
      "grad_norm": 0.05554387718439102,
      "learning_rate": 0.0012,
      "loss": 2.6943485736846924,
      "step": 4136
    },
    {
      "epoch": 0.33551323014085294,
      "grad_norm": 0.036920350044965744,
      "learning_rate": 0.0012,
      "loss": 2.767704963684082,
      "step": 4138
    },
    {
      "epoch": 0.3356753921660539,
      "grad_norm": 0.04806669056415558,
      "learning_rate": 0.0012,
      "loss": 2.693190574645996,
      "step": 4140
    },
    {
      "epoch": 0.3358375541912549,
      "grad_norm": 0.060605064034461975,
      "learning_rate": 0.0012,
      "loss": 2.679232358932495,
      "step": 4142
    },
    {
      "epoch": 0.3359997162164559,
      "grad_norm": 0.023867804557085037,
      "learning_rate": 0.0012,
      "loss": 2.634706497192383,
      "step": 4144
    },
    {
      "epoch": 0.3361618782416569,
      "grad_norm": 0.05015748366713524,
      "learning_rate": 0.0012,
      "loss": 2.7431280612945557,
      "step": 4146
    },
    {
      "epoch": 0.33632404026685786,
      "grad_norm": 0.05829876288771629,
      "learning_rate": 0.0012,
      "loss": 2.6246330738067627,
      "step": 4148
    },
    {
      "epoch": 0.3364862022920589,
      "grad_norm": 0.02756594866514206,
      "learning_rate": 0.0012,
      "loss": 2.7407774925231934,
      "step": 4150
    },
    {
      "epoch": 0.33664836431725986,
      "grad_norm": 0.04624181613326073,
      "learning_rate": 0.0012,
      "loss": 2.6223182678222656,
      "step": 4152
    },
    {
      "epoch": 0.33681052634246084,
      "grad_norm": 0.047243840992450714,
      "learning_rate": 0.0012,
      "loss": 2.738889455795288,
      "step": 4154
    },
    {
      "epoch": 0.33697268836766187,
      "grad_norm": 0.02329108864068985,
      "learning_rate": 0.0012,
      "loss": 2.6883139610290527,
      "step": 4156
    },
    {
      "epoch": 0.33713485039286284,
      "grad_norm": 0.022362997755408287,
      "learning_rate": 0.0012,
      "loss": 2.732016086578369,
      "step": 4158
    },
    {
      "epoch": 0.3372970124180638,
      "grad_norm": 0.026382233947515488,
      "learning_rate": 0.0012,
      "loss": 2.669762134552002,
      "step": 4160
    },
    {
      "epoch": 0.33745917444326484,
      "grad_norm": 0.029124580323696136,
      "learning_rate": 0.0012,
      "loss": 2.7155306339263916,
      "step": 4162
    },
    {
      "epoch": 0.3376213364684658,
      "grad_norm": 0.0252310149371624,
      "learning_rate": 0.0012,
      "loss": 2.6671950817108154,
      "step": 4164
    },
    {
      "epoch": 0.3377834984936668,
      "grad_norm": 0.03614935651421547,
      "learning_rate": 0.0012,
      "loss": 2.729607343673706,
      "step": 4166
    },
    {
      "epoch": 0.3379456605188678,
      "grad_norm": 0.042566969990730286,
      "learning_rate": 0.0012,
      "loss": 2.605884075164795,
      "step": 4168
    },
    {
      "epoch": 0.3381078225440688,
      "grad_norm": 0.023970408365130424,
      "learning_rate": 0.0012,
      "loss": 2.7337660789489746,
      "step": 4170
    },
    {
      "epoch": 0.33826998456926977,
      "grad_norm": 0.0316929817199707,
      "learning_rate": 0.0012,
      "loss": 2.6651487350463867,
      "step": 4172
    },
    {
      "epoch": 0.3384321465944708,
      "grad_norm": 0.023114634677767754,
      "learning_rate": 0.0012,
      "loss": 2.7078981399536133,
      "step": 4174
    },
    {
      "epoch": 0.33859430861967177,
      "grad_norm": 0.02513173222541809,
      "learning_rate": 0.0012,
      "loss": 2.741229295730591,
      "step": 4176
    },
    {
      "epoch": 0.33875647064487274,
      "grad_norm": 0.023904602974653244,
      "learning_rate": 0.0012,
      "loss": 2.691443920135498,
      "step": 4178
    },
    {
      "epoch": 0.33891863267007377,
      "grad_norm": 0.02708498202264309,
      "learning_rate": 0.0012,
      "loss": 2.650110960006714,
      "step": 4180
    },
    {
      "epoch": 0.33908079469527475,
      "grad_norm": 0.034339189529418945,
      "learning_rate": 0.0012,
      "loss": 2.66355299949646,
      "step": 4182
    },
    {
      "epoch": 0.3392429567204757,
      "grad_norm": 0.025008419528603554,
      "learning_rate": 0.0012,
      "loss": 2.64084529876709,
      "step": 4184
    },
    {
      "epoch": 0.33940511874567675,
      "grad_norm": 0.046196579933166504,
      "learning_rate": 0.0012,
      "loss": 2.6392979621887207,
      "step": 4186
    },
    {
      "epoch": 0.3395672807708777,
      "grad_norm": 0.037288058549165726,
      "learning_rate": 0.0012,
      "loss": 2.7648067474365234,
      "step": 4188
    },
    {
      "epoch": 0.3397294427960787,
      "grad_norm": 0.027782684192061424,
      "learning_rate": 0.0012,
      "loss": 2.6749942302703857,
      "step": 4190
    },
    {
      "epoch": 0.3398916048212797,
      "grad_norm": 0.024659613147377968,
      "learning_rate": 0.0012,
      "loss": 2.7293343544006348,
      "step": 4192
    },
    {
      "epoch": 0.3400537668464807,
      "grad_norm": 0.02927209436893463,
      "learning_rate": 0.0012,
      "loss": 2.7238411903381348,
      "step": 4194
    },
    {
      "epoch": 0.34021592887168167,
      "grad_norm": 0.05914444848895073,
      "learning_rate": 0.0012,
      "loss": 2.722947120666504,
      "step": 4196
    },
    {
      "epoch": 0.3403780908968827,
      "grad_norm": 0.07142481952905655,
      "learning_rate": 0.0012,
      "loss": 2.7263216972351074,
      "step": 4198
    },
    {
      "epoch": 0.3405402529220837,
      "grad_norm": 0.029212357476353645,
      "learning_rate": 0.0012,
      "loss": 2.7035999298095703,
      "step": 4200
    },
    {
      "epoch": 0.34070241494728465,
      "grad_norm": 0.04042855650186539,
      "learning_rate": 0.0012,
      "loss": 2.6815640926361084,
      "step": 4202
    },
    {
      "epoch": 0.3408645769724857,
      "grad_norm": 0.04151708632707596,
      "learning_rate": 0.0012,
      "loss": 2.681485652923584,
      "step": 4204
    },
    {
      "epoch": 0.34102673899768665,
      "grad_norm": 0.02847793884575367,
      "learning_rate": 0.0012,
      "loss": 2.6891016960144043,
      "step": 4206
    },
    {
      "epoch": 0.3411889010228876,
      "grad_norm": 0.03479878231883049,
      "learning_rate": 0.0012,
      "loss": 2.6998019218444824,
      "step": 4208
    },
    {
      "epoch": 0.34135106304808865,
      "grad_norm": 0.03736049681901932,
      "learning_rate": 0.0012,
      "loss": 2.714029312133789,
      "step": 4210
    },
    {
      "epoch": 0.34151322507328963,
      "grad_norm": 0.0282556414604187,
      "learning_rate": 0.0012,
      "loss": 2.631664276123047,
      "step": 4212
    },
    {
      "epoch": 0.3416753870984906,
      "grad_norm": 0.021514033898711205,
      "learning_rate": 0.0012,
      "loss": 2.6926121711730957,
      "step": 4214
    },
    {
      "epoch": 0.34183754912369163,
      "grad_norm": 0.05193188413977623,
      "learning_rate": 0.0012,
      "loss": 2.745917558670044,
      "step": 4216
    },
    {
      "epoch": 0.3419997111488926,
      "grad_norm": 0.057567715644836426,
      "learning_rate": 0.0012,
      "loss": 2.6794798374176025,
      "step": 4218
    },
    {
      "epoch": 0.3421618731740936,
      "grad_norm": 0.038360487669706345,
      "learning_rate": 0.0012,
      "loss": 2.6495189666748047,
      "step": 4220
    },
    {
      "epoch": 0.3423240351992946,
      "grad_norm": 0.023221692070364952,
      "learning_rate": 0.0012,
      "loss": 2.6438448429107666,
      "step": 4222
    },
    {
      "epoch": 0.3424861972244956,
      "grad_norm": 0.04477328062057495,
      "learning_rate": 0.0012,
      "loss": 2.684333562850952,
      "step": 4224
    },
    {
      "epoch": 0.34264835924969655,
      "grad_norm": 0.05162075534462929,
      "learning_rate": 0.0012,
      "loss": 2.68581485748291,
      "step": 4226
    },
    {
      "epoch": 0.3428105212748976,
      "grad_norm": 0.025699902325868607,
      "learning_rate": 0.0012,
      "loss": 2.657613754272461,
      "step": 4228
    },
    {
      "epoch": 0.34297268330009856,
      "grad_norm": 0.024010343477129936,
      "learning_rate": 0.0012,
      "loss": 2.6472411155700684,
      "step": 4230
    },
    {
      "epoch": 0.34313484532529953,
      "grad_norm": 0.024835210293531418,
      "learning_rate": 0.0012,
      "loss": 2.6761531829833984,
      "step": 4232
    },
    {
      "epoch": 0.34329700735050056,
      "grad_norm": 0.02711544744670391,
      "learning_rate": 0.0012,
      "loss": 2.699814796447754,
      "step": 4234
    },
    {
      "epoch": 0.34345916937570153,
      "grad_norm": 0.03726671636104584,
      "learning_rate": 0.0012,
      "loss": 2.675323963165283,
      "step": 4236
    },
    {
      "epoch": 0.3436213314009025,
      "grad_norm": 0.03976675495505333,
      "learning_rate": 0.0012,
      "loss": 2.6971335411071777,
      "step": 4238
    },
    {
      "epoch": 0.34378349342610354,
      "grad_norm": 0.047716621309518814,
      "learning_rate": 0.0012,
      "loss": 2.724395275115967,
      "step": 4240
    },
    {
      "epoch": 0.3439456554513045,
      "grad_norm": 0.022694962099194527,
      "learning_rate": 0.0012,
      "loss": 2.664010524749756,
      "step": 4242
    },
    {
      "epoch": 0.3441078174765055,
      "grad_norm": 0.025400560349225998,
      "learning_rate": 0.0012,
      "loss": 2.6662960052490234,
      "step": 4244
    },
    {
      "epoch": 0.3442699795017065,
      "grad_norm": 0.040109433233737946,
      "learning_rate": 0.0012,
      "loss": 2.6672170162200928,
      "step": 4246
    },
    {
      "epoch": 0.3444321415269075,
      "grad_norm": 0.058041468262672424,
      "learning_rate": 0.0012,
      "loss": 2.7085416316986084,
      "step": 4248
    },
    {
      "epoch": 0.34459430355210846,
      "grad_norm": 0.027318213135004044,
      "learning_rate": 0.0012,
      "loss": 2.7064037322998047,
      "step": 4250
    },
    {
      "epoch": 0.3447564655773095,
      "grad_norm": 0.023730270564556122,
      "learning_rate": 0.0012,
      "loss": 2.690824508666992,
      "step": 4252
    },
    {
      "epoch": 0.34491862760251046,
      "grad_norm": 0.042072877287864685,
      "learning_rate": 0.0012,
      "loss": 2.692138671875,
      "step": 4254
    },
    {
      "epoch": 0.34508078962771144,
      "grad_norm": 0.03319491073489189,
      "learning_rate": 0.0012,
      "loss": 2.707387685775757,
      "step": 4256
    },
    {
      "epoch": 0.34524295165291247,
      "grad_norm": 0.027884718030691147,
      "learning_rate": 0.0012,
      "loss": 2.688188076019287,
      "step": 4258
    },
    {
      "epoch": 0.34540511367811344,
      "grad_norm": 0.02545115165412426,
      "learning_rate": 0.0012,
      "loss": 2.684865951538086,
      "step": 4260
    },
    {
      "epoch": 0.3455672757033144,
      "grad_norm": 0.026645712554454803,
      "learning_rate": 0.0012,
      "loss": 2.718709707260132,
      "step": 4262
    },
    {
      "epoch": 0.34572943772851544,
      "grad_norm": 0.045483749359846115,
      "learning_rate": 0.0012,
      "loss": 2.6586787700653076,
      "step": 4264
    },
    {
      "epoch": 0.3458915997537164,
      "grad_norm": 0.02479964680969715,
      "learning_rate": 0.0012,
      "loss": 2.7374744415283203,
      "step": 4266
    },
    {
      "epoch": 0.3460537617789174,
      "grad_norm": 0.04527059197425842,
      "learning_rate": 0.0012,
      "loss": 2.7299396991729736,
      "step": 4268
    },
    {
      "epoch": 0.3462159238041184,
      "grad_norm": 0.03935789316892624,
      "learning_rate": 0.0012,
      "loss": 2.7046399116516113,
      "step": 4270
    },
    {
      "epoch": 0.3463780858293194,
      "grad_norm": 0.04302796721458435,
      "learning_rate": 0.0012,
      "loss": 2.687070846557617,
      "step": 4272
    },
    {
      "epoch": 0.34654024785452037,
      "grad_norm": 0.02867903932929039,
      "learning_rate": 0.0012,
      "loss": 2.678657293319702,
      "step": 4274
    },
    {
      "epoch": 0.3467024098797214,
      "grad_norm": 0.03739440441131592,
      "learning_rate": 0.0012,
      "loss": 2.6800832748413086,
      "step": 4276
    },
    {
      "epoch": 0.34686457190492237,
      "grad_norm": 0.04378298670053482,
      "learning_rate": 0.0012,
      "loss": 2.7151520252227783,
      "step": 4278
    },
    {
      "epoch": 0.34702673393012334,
      "grad_norm": 0.024784604087471962,
      "learning_rate": 0.0012,
      "loss": 2.696993589401245,
      "step": 4280
    },
    {
      "epoch": 0.3471888959553244,
      "grad_norm": 0.029819011688232422,
      "learning_rate": 0.0012,
      "loss": 2.7214319705963135,
      "step": 4282
    },
    {
      "epoch": 0.34735105798052535,
      "grad_norm": 0.051211897283792496,
      "learning_rate": 0.0012,
      "loss": 2.6860201358795166,
      "step": 4284
    },
    {
      "epoch": 0.3475132200057263,
      "grad_norm": 0.024730898439884186,
      "learning_rate": 0.0012,
      "loss": 2.7572529315948486,
      "step": 4286
    },
    {
      "epoch": 0.34767538203092735,
      "grad_norm": 0.026327118277549744,
      "learning_rate": 0.0012,
      "loss": 2.773470401763916,
      "step": 4288
    },
    {
      "epoch": 0.3478375440561283,
      "grad_norm": 0.050847917795181274,
      "learning_rate": 0.0012,
      "loss": 2.639288902282715,
      "step": 4290
    },
    {
      "epoch": 0.3479997060813293,
      "grad_norm": 0.06527058035135269,
      "learning_rate": 0.0012,
      "loss": 2.679884433746338,
      "step": 4292
    },
    {
      "epoch": 0.3481618681065303,
      "grad_norm": 0.03230326622724533,
      "learning_rate": 0.0012,
      "loss": 2.747467279434204,
      "step": 4294
    },
    {
      "epoch": 0.3483240301317313,
      "grad_norm": 0.024561133235692978,
      "learning_rate": 0.0012,
      "loss": 2.6808362007141113,
      "step": 4296
    },
    {
      "epoch": 0.3484861921569323,
      "grad_norm": 0.037208300083875656,
      "learning_rate": 0.0012,
      "loss": 2.725965976715088,
      "step": 4298
    },
    {
      "epoch": 0.3486483541821333,
      "grad_norm": 0.024827878922224045,
      "learning_rate": 0.0012,
      "loss": 2.657229423522949,
      "step": 4300
    },
    {
      "epoch": 0.3488105162073343,
      "grad_norm": 0.046633101999759674,
      "learning_rate": 0.0012,
      "loss": 2.706916093826294,
      "step": 4302
    },
    {
      "epoch": 0.34897267823253525,
      "grad_norm": 0.042865511029958725,
      "learning_rate": 0.0012,
      "loss": 2.6783056259155273,
      "step": 4304
    },
    {
      "epoch": 0.3491348402577363,
      "grad_norm": 0.06200164929032326,
      "learning_rate": 0.0012,
      "loss": 2.693697452545166,
      "step": 4306
    },
    {
      "epoch": 0.34929700228293725,
      "grad_norm": 0.04811514541506767,
      "learning_rate": 0.0012,
      "loss": 2.6558051109313965,
      "step": 4308
    },
    {
      "epoch": 0.3494591643081382,
      "grad_norm": 0.03019232116639614,
      "learning_rate": 0.0012,
      "loss": 2.712017059326172,
      "step": 4310
    },
    {
      "epoch": 0.34962132633333926,
      "grad_norm": 0.029916448518633842,
      "learning_rate": 0.0012,
      "loss": 2.7340564727783203,
      "step": 4312
    },
    {
      "epoch": 0.34978348835854023,
      "grad_norm": 0.03361848369240761,
      "learning_rate": 0.0012,
      "loss": 2.6992597579956055,
      "step": 4314
    },
    {
      "epoch": 0.3499456503837412,
      "grad_norm": 0.036177705973386765,
      "learning_rate": 0.0012,
      "loss": 2.691565752029419,
      "step": 4316
    },
    {
      "epoch": 0.35010781240894223,
      "grad_norm": 0.05140072479844093,
      "learning_rate": 0.0012,
      "loss": 2.6429333686828613,
      "step": 4318
    },
    {
      "epoch": 0.3502699744341432,
      "grad_norm": 0.03286268934607506,
      "learning_rate": 0.0012,
      "loss": 2.654623031616211,
      "step": 4320
    },
    {
      "epoch": 0.3504321364593442,
      "grad_norm": 0.02895548939704895,
      "learning_rate": 0.0012,
      "loss": 2.7015419006347656,
      "step": 4322
    },
    {
      "epoch": 0.3505942984845452,
      "grad_norm": 0.028689613565802574,
      "learning_rate": 0.0012,
      "loss": 2.688793182373047,
      "step": 4324
    },
    {
      "epoch": 0.3507564605097462,
      "grad_norm": 0.037071552127599716,
      "learning_rate": 0.0012,
      "loss": 2.7169408798217773,
      "step": 4326
    },
    {
      "epoch": 0.3509186225349472,
      "grad_norm": 0.035978931933641434,
      "learning_rate": 0.0012,
      "loss": 2.646965503692627,
      "step": 4328
    },
    {
      "epoch": 0.3510807845601482,
      "grad_norm": 0.02505083754658699,
      "learning_rate": 0.0012,
      "loss": 2.7461886405944824,
      "step": 4330
    },
    {
      "epoch": 0.35124294658534916,
      "grad_norm": 0.027629496529698372,
      "learning_rate": 0.0012,
      "loss": 2.664991617202759,
      "step": 4332
    },
    {
      "epoch": 0.3514051086105502,
      "grad_norm": 0.024265408515930176,
      "learning_rate": 0.0012,
      "loss": 2.6350643634796143,
      "step": 4334
    },
    {
      "epoch": 0.35156727063575116,
      "grad_norm": 0.027043383568525314,
      "learning_rate": 0.0012,
      "loss": 2.678248405456543,
      "step": 4336
    },
    {
      "epoch": 0.35172943266095213,
      "grad_norm": 0.024526115506887436,
      "learning_rate": 0.0012,
      "loss": 2.6953368186950684,
      "step": 4338
    },
    {
      "epoch": 0.35189159468615316,
      "grad_norm": 0.022610612213611603,
      "learning_rate": 0.0012,
      "loss": 2.6262025833129883,
      "step": 4340
    },
    {
      "epoch": 0.35205375671135414,
      "grad_norm": 0.03755807504057884,
      "learning_rate": 0.0012,
      "loss": 2.7120742797851562,
      "step": 4342
    },
    {
      "epoch": 0.3522159187365551,
      "grad_norm": 0.030840376392006874,
      "learning_rate": 0.0012,
      "loss": 2.728550672531128,
      "step": 4344
    },
    {
      "epoch": 0.35237808076175614,
      "grad_norm": 0.024688435718417168,
      "learning_rate": 0.0012,
      "loss": 2.7120895385742188,
      "step": 4346
    },
    {
      "epoch": 0.3525402427869571,
      "grad_norm": 0.023864291608333588,
      "learning_rate": 0.0012,
      "loss": 2.698718309402466,
      "step": 4348
    },
    {
      "epoch": 0.3527024048121581,
      "grad_norm": 0.040959496051073074,
      "learning_rate": 0.0012,
      "loss": 2.6618828773498535,
      "step": 4350
    },
    {
      "epoch": 0.3528645668373591,
      "grad_norm": 0.03855977579951286,
      "learning_rate": 0.0012,
      "loss": 2.6304330825805664,
      "step": 4352
    },
    {
      "epoch": 0.3530267288625601,
      "grad_norm": 0.02747306600213051,
      "learning_rate": 0.0012,
      "loss": 2.6451797485351562,
      "step": 4354
    },
    {
      "epoch": 0.35318889088776106,
      "grad_norm": 0.08699292689561844,
      "learning_rate": 0.0012,
      "loss": 2.652311325073242,
      "step": 4356
    },
    {
      "epoch": 0.3533510529129621,
      "grad_norm": 0.07981257885694504,
      "learning_rate": 0.0012,
      "loss": 2.6644253730773926,
      "step": 4358
    },
    {
      "epoch": 0.35351321493816307,
      "grad_norm": 0.029331035912036896,
      "learning_rate": 0.0012,
      "loss": 2.691559314727783,
      "step": 4360
    },
    {
      "epoch": 0.35367537696336404,
      "grad_norm": 0.05227319523692131,
      "learning_rate": 0.0012,
      "loss": 2.7466065883636475,
      "step": 4362
    },
    {
      "epoch": 0.35383753898856507,
      "grad_norm": 0.08191391080617905,
      "learning_rate": 0.0012,
      "loss": 2.6692662239074707,
      "step": 4364
    },
    {
      "epoch": 0.35399970101376604,
      "grad_norm": 0.028714574873447418,
      "learning_rate": 0.0012,
      "loss": 2.61026668548584,
      "step": 4366
    },
    {
      "epoch": 0.354161863038967,
      "grad_norm": 0.046111684292554855,
      "learning_rate": 0.0012,
      "loss": 2.6982061862945557,
      "step": 4368
    },
    {
      "epoch": 0.35432402506416805,
      "grad_norm": 0.024998530745506287,
      "learning_rate": 0.0012,
      "loss": 2.7012178897857666,
      "step": 4370
    },
    {
      "epoch": 0.354486187089369,
      "grad_norm": 0.03833502158522606,
      "learning_rate": 0.0012,
      "loss": 2.584350109100342,
      "step": 4372
    },
    {
      "epoch": 0.35464834911457,
      "grad_norm": 0.023251397535204887,
      "learning_rate": 0.0012,
      "loss": 2.6964058876037598,
      "step": 4374
    },
    {
      "epoch": 0.354810511139771,
      "grad_norm": 0.025545651093125343,
      "learning_rate": 0.0012,
      "loss": 2.707651138305664,
      "step": 4376
    },
    {
      "epoch": 0.354972673164972,
      "grad_norm": 0.023591298609972,
      "learning_rate": 0.0012,
      "loss": 2.6596031188964844,
      "step": 4378
    },
    {
      "epoch": 0.35513483519017297,
      "grad_norm": 0.023737505078315735,
      "learning_rate": 0.0012,
      "loss": 2.628347158432007,
      "step": 4380
    },
    {
      "epoch": 0.355296997215374,
      "grad_norm": 0.040206484496593475,
      "learning_rate": 0.0012,
      "loss": 2.6505675315856934,
      "step": 4382
    },
    {
      "epoch": 0.355459159240575,
      "grad_norm": 0.03015020117163658,
      "learning_rate": 0.0012,
      "loss": 2.679780960083008,
      "step": 4384
    },
    {
      "epoch": 0.35562132126577595,
      "grad_norm": 0.023069262504577637,
      "learning_rate": 0.0012,
      "loss": 2.719302177429199,
      "step": 4386
    },
    {
      "epoch": 0.355783483290977,
      "grad_norm": 0.03473250940442085,
      "learning_rate": 0.0012,
      "loss": 2.7033281326293945,
      "step": 4388
    },
    {
      "epoch": 0.35594564531617795,
      "grad_norm": 0.023007391020655632,
      "learning_rate": 0.0012,
      "loss": 2.6826181411743164,
      "step": 4390
    },
    {
      "epoch": 0.3561078073413789,
      "grad_norm": 0.0256130900233984,
      "learning_rate": 0.0012,
      "loss": 2.6811411380767822,
      "step": 4392
    },
    {
      "epoch": 0.35626996936657995,
      "grad_norm": 0.024335592985153198,
      "learning_rate": 0.0012,
      "loss": 2.683067798614502,
      "step": 4394
    },
    {
      "epoch": 0.3564321313917809,
      "grad_norm": 0.028140198439359665,
      "learning_rate": 0.0012,
      "loss": 2.7155399322509766,
      "step": 4396
    },
    {
      "epoch": 0.3565942934169819,
      "grad_norm": 0.03286842629313469,
      "learning_rate": 0.0012,
      "loss": 2.7253050804138184,
      "step": 4398
    },
    {
      "epoch": 0.35675645544218293,
      "grad_norm": 0.021635690703988075,
      "learning_rate": 0.0012,
      "loss": 2.7849442958831787,
      "step": 4400
    },
    {
      "epoch": 0.3569186174673839,
      "grad_norm": 0.024096405133605003,
      "learning_rate": 0.0012,
      "loss": 2.6658618450164795,
      "step": 4402
    },
    {
      "epoch": 0.3570807794925849,
      "grad_norm": 0.031883083283901215,
      "learning_rate": 0.0012,
      "loss": 2.67940616607666,
      "step": 4404
    },
    {
      "epoch": 0.3572429415177859,
      "grad_norm": 0.03173130378127098,
      "learning_rate": 0.0012,
      "loss": 2.623486042022705,
      "step": 4406
    },
    {
      "epoch": 0.3574051035429869,
      "grad_norm": 0.044244199991226196,
      "learning_rate": 0.0012,
      "loss": 2.698143720626831,
      "step": 4408
    },
    {
      "epoch": 0.35756726556818785,
      "grad_norm": 0.09148424118757248,
      "learning_rate": 0.0012,
      "loss": 2.7134299278259277,
      "step": 4410
    },
    {
      "epoch": 0.3577294275933889,
      "grad_norm": 0.049412913620471954,
      "learning_rate": 0.0012,
      "loss": 2.6523938179016113,
      "step": 4412
    },
    {
      "epoch": 0.35789158961858986,
      "grad_norm": 0.029846694320440292,
      "learning_rate": 0.0012,
      "loss": 2.7490317821502686,
      "step": 4414
    },
    {
      "epoch": 0.35805375164379083,
      "grad_norm": 0.025617342442274094,
      "learning_rate": 0.0012,
      "loss": 2.658965587615967,
      "step": 4416
    },
    {
      "epoch": 0.35821591366899186,
      "grad_norm": 0.04093112051486969,
      "learning_rate": 0.0012,
      "loss": 2.6945743560791016,
      "step": 4418
    },
    {
      "epoch": 0.35837807569419283,
      "grad_norm": 0.03096291609108448,
      "learning_rate": 0.0012,
      "loss": 2.6959786415100098,
      "step": 4420
    },
    {
      "epoch": 0.3585402377193938,
      "grad_norm": 0.028606008738279343,
      "learning_rate": 0.0012,
      "loss": 2.712280750274658,
      "step": 4422
    },
    {
      "epoch": 0.35870239974459484,
      "grad_norm": 0.022742221131920815,
      "learning_rate": 0.0012,
      "loss": 2.70170259475708,
      "step": 4424
    },
    {
      "epoch": 0.3588645617697958,
      "grad_norm": 0.02699866145849228,
      "learning_rate": 0.0012,
      "loss": 2.6940507888793945,
      "step": 4426
    },
    {
      "epoch": 0.3590267237949968,
      "grad_norm": 0.02711099572479725,
      "learning_rate": 0.0012,
      "loss": 2.7589492797851562,
      "step": 4428
    },
    {
      "epoch": 0.3591888858201978,
      "grad_norm": 0.02675597555935383,
      "learning_rate": 0.0012,
      "loss": 2.656113624572754,
      "step": 4430
    },
    {
      "epoch": 0.3593510478453988,
      "grad_norm": 0.023084670305252075,
      "learning_rate": 0.0012,
      "loss": 2.675797939300537,
      "step": 4432
    },
    {
      "epoch": 0.35951320987059976,
      "grad_norm": 0.03565461188554764,
      "learning_rate": 0.0012,
      "loss": 2.697751522064209,
      "step": 4434
    },
    {
      "epoch": 0.3596753718958008,
      "grad_norm": 0.027736585587263107,
      "learning_rate": 0.0012,
      "loss": 2.7096450328826904,
      "step": 4436
    },
    {
      "epoch": 0.35983753392100176,
      "grad_norm": 0.03657485917210579,
      "learning_rate": 0.0012,
      "loss": 2.725951671600342,
      "step": 4438
    },
    {
      "epoch": 0.35999969594620274,
      "grad_norm": 0.05346885696053505,
      "learning_rate": 0.0012,
      "loss": 2.728531837463379,
      "step": 4440
    },
    {
      "epoch": 0.36016185797140376,
      "grad_norm": 0.035348135977983475,
      "learning_rate": 0.0012,
      "loss": 2.694122314453125,
      "step": 4442
    },
    {
      "epoch": 0.36032401999660474,
      "grad_norm": 0.029181595891714096,
      "learning_rate": 0.0012,
      "loss": 2.6346354484558105,
      "step": 4444
    },
    {
      "epoch": 0.3604861820218057,
      "grad_norm": 0.03417433053255081,
      "learning_rate": 0.0012,
      "loss": 2.6375160217285156,
      "step": 4446
    },
    {
      "epoch": 0.36064834404700674,
      "grad_norm": 0.027788976207375526,
      "learning_rate": 0.0012,
      "loss": 2.724367141723633,
      "step": 4448
    },
    {
      "epoch": 0.3608105060722077,
      "grad_norm": 0.025220340117812157,
      "learning_rate": 0.0012,
      "loss": 2.691659450531006,
      "step": 4450
    },
    {
      "epoch": 0.3609726680974087,
      "grad_norm": 0.040768176317214966,
      "learning_rate": 0.0012,
      "loss": 2.6693618297576904,
      "step": 4452
    },
    {
      "epoch": 0.3611348301226097,
      "grad_norm": 0.03948557376861572,
      "learning_rate": 0.0012,
      "loss": 2.683051824569702,
      "step": 4454
    },
    {
      "epoch": 0.3612969921478107,
      "grad_norm": 0.025758983567357063,
      "learning_rate": 0.0012,
      "loss": 2.660193920135498,
      "step": 4456
    },
    {
      "epoch": 0.36145915417301167,
      "grad_norm": 0.026208091527223587,
      "learning_rate": 0.0012,
      "loss": 2.711869478225708,
      "step": 4458
    },
    {
      "epoch": 0.3616213161982127,
      "grad_norm": 0.06176064908504486,
      "learning_rate": 0.0012,
      "loss": 2.6660399436950684,
      "step": 4460
    },
    {
      "epoch": 0.36178347822341367,
      "grad_norm": 0.06676062941551208,
      "learning_rate": 0.0012,
      "loss": 2.722703218460083,
      "step": 4462
    },
    {
      "epoch": 0.36194564024861464,
      "grad_norm": 0.024685006588697433,
      "learning_rate": 0.0012,
      "loss": 2.715381622314453,
      "step": 4464
    },
    {
      "epoch": 0.36210780227381567,
      "grad_norm": 0.044701479375362396,
      "learning_rate": 0.0012,
      "loss": 2.6253039836883545,
      "step": 4466
    },
    {
      "epoch": 0.36226996429901664,
      "grad_norm": 0.049150437116622925,
      "learning_rate": 0.0012,
      "loss": 2.6730129718780518,
      "step": 4468
    },
    {
      "epoch": 0.3624321263242176,
      "grad_norm": 0.026057148352265358,
      "learning_rate": 0.0012,
      "loss": 2.6863131523132324,
      "step": 4470
    },
    {
      "epoch": 0.36259428834941865,
      "grad_norm": 0.08086767792701721,
      "learning_rate": 0.0012,
      "loss": 2.681279182434082,
      "step": 4472
    },
    {
      "epoch": 0.3627564503746196,
      "grad_norm": 0.06988751888275146,
      "learning_rate": 0.0012,
      "loss": 2.6587491035461426,
      "step": 4474
    },
    {
      "epoch": 0.3629186123998206,
      "grad_norm": 0.03207610547542572,
      "learning_rate": 0.0012,
      "loss": 2.628239393234253,
      "step": 4476
    },
    {
      "epoch": 0.3630807744250216,
      "grad_norm": 0.03217802196741104,
      "learning_rate": 0.0012,
      "loss": 2.659209728240967,
      "step": 4478
    },
    {
      "epoch": 0.3632429364502226,
      "grad_norm": 0.03574107587337494,
      "learning_rate": 0.0012,
      "loss": 2.7510533332824707,
      "step": 4480
    },
    {
      "epoch": 0.36340509847542357,
      "grad_norm": 0.029928019270300865,
      "learning_rate": 0.0012,
      "loss": 2.6568708419799805,
      "step": 4482
    },
    {
      "epoch": 0.3635672605006246,
      "grad_norm": 0.0393172912299633,
      "learning_rate": 0.0012,
      "loss": 2.691864490509033,
      "step": 4484
    },
    {
      "epoch": 0.3637294225258256,
      "grad_norm": 0.03063596971333027,
      "learning_rate": 0.0012,
      "loss": 2.670974016189575,
      "step": 4486
    },
    {
      "epoch": 0.36389158455102655,
      "grad_norm": 0.025318477302789688,
      "learning_rate": 0.0012,
      "loss": 2.6790928840637207,
      "step": 4488
    },
    {
      "epoch": 0.3640537465762276,
      "grad_norm": 0.023679519072175026,
      "learning_rate": 0.0012,
      "loss": 2.6692776679992676,
      "step": 4490
    },
    {
      "epoch": 0.36421590860142855,
      "grad_norm": 0.028859233483672142,
      "learning_rate": 0.0012,
      "loss": 2.718724489212036,
      "step": 4492
    },
    {
      "epoch": 0.3643780706266295,
      "grad_norm": 0.0274050273001194,
      "learning_rate": 0.0012,
      "loss": 2.7971599102020264,
      "step": 4494
    },
    {
      "epoch": 0.36454023265183055,
      "grad_norm": 0.023900479078292847,
      "learning_rate": 0.0012,
      "loss": 2.7337470054626465,
      "step": 4496
    },
    {
      "epoch": 0.3647023946770315,
      "grad_norm": 0.030630310997366905,
      "learning_rate": 0.0012,
      "loss": 2.6705150604248047,
      "step": 4498
    },
    {
      "epoch": 0.3648645567022325,
      "grad_norm": 0.02544301189482212,
      "learning_rate": 0.0012,
      "loss": 2.7017459869384766,
      "step": 4500
    },
    {
      "epoch": 0.36502671872743353,
      "grad_norm": 0.047425106167793274,
      "learning_rate": 0.0012,
      "loss": 2.702084541320801,
      "step": 4502
    },
    {
      "epoch": 0.3651888807526345,
      "grad_norm": 0.028940215706825256,
      "learning_rate": 0.0012,
      "loss": 2.7161030769348145,
      "step": 4504
    },
    {
      "epoch": 0.3653510427778355,
      "grad_norm": 0.0575803741812706,
      "learning_rate": 0.0012,
      "loss": 2.7230703830718994,
      "step": 4506
    },
    {
      "epoch": 0.3655132048030365,
      "grad_norm": 0.06745222955942154,
      "learning_rate": 0.0012,
      "loss": 2.695676803588867,
      "step": 4508
    },
    {
      "epoch": 0.3656753668282375,
      "grad_norm": 0.031069288030266762,
      "learning_rate": 0.0012,
      "loss": 2.6616177558898926,
      "step": 4510
    },
    {
      "epoch": 0.36583752885343845,
      "grad_norm": 0.0387544147670269,
      "learning_rate": 0.0012,
      "loss": 2.64145565032959,
      "step": 4512
    },
    {
      "epoch": 0.3659996908786395,
      "grad_norm": 0.07576430588960648,
      "learning_rate": 0.0012,
      "loss": 2.7004101276397705,
      "step": 4514
    },
    {
      "epoch": 0.36616185290384046,
      "grad_norm": 0.05134518817067146,
      "learning_rate": 0.0012,
      "loss": 2.6687960624694824,
      "step": 4516
    },
    {
      "epoch": 0.36632401492904143,
      "grad_norm": 0.0423412024974823,
      "learning_rate": 0.0012,
      "loss": 2.7123022079467773,
      "step": 4518
    },
    {
      "epoch": 0.36648617695424246,
      "grad_norm": 0.06811556965112686,
      "learning_rate": 0.0012,
      "loss": 2.693458080291748,
      "step": 4520
    },
    {
      "epoch": 0.36664833897944343,
      "grad_norm": 0.03561345860362053,
      "learning_rate": 0.0012,
      "loss": 2.6519601345062256,
      "step": 4522
    },
    {
      "epoch": 0.3668105010046444,
      "grad_norm": 0.031456444412469864,
      "learning_rate": 0.0012,
      "loss": 2.6885595321655273,
      "step": 4524
    },
    {
      "epoch": 0.36697266302984544,
      "grad_norm": 0.04968511313199997,
      "learning_rate": 0.0012,
      "loss": 2.712066411972046,
      "step": 4526
    },
    {
      "epoch": 0.3671348250550464,
      "grad_norm": 0.043511711061000824,
      "learning_rate": 0.0012,
      "loss": 2.6727168560028076,
      "step": 4528
    },
    {
      "epoch": 0.3672969870802474,
      "grad_norm": 0.02436407282948494,
      "learning_rate": 0.0012,
      "loss": 2.6458563804626465,
      "step": 4530
    },
    {
      "epoch": 0.3674591491054484,
      "grad_norm": 0.03145044669508934,
      "learning_rate": 0.0012,
      "loss": 2.704129219055176,
      "step": 4532
    },
    {
      "epoch": 0.3676213111306494,
      "grad_norm": 0.025346804410219193,
      "learning_rate": 0.0012,
      "loss": 2.7245419025421143,
      "step": 4534
    },
    {
      "epoch": 0.36778347315585036,
      "grad_norm": 0.02237383835017681,
      "learning_rate": 0.0012,
      "loss": 2.6700053215026855,
      "step": 4536
    },
    {
      "epoch": 0.3679456351810514,
      "grad_norm": 0.03321722522377968,
      "learning_rate": 0.0012,
      "loss": 2.652122735977173,
      "step": 4538
    },
    {
      "epoch": 0.36810779720625236,
      "grad_norm": 0.032311923801898956,
      "learning_rate": 0.0012,
      "loss": 2.6814827919006348,
      "step": 4540
    },
    {
      "epoch": 0.36826995923145334,
      "grad_norm": 0.029192110523581505,
      "learning_rate": 0.0012,
      "loss": 2.713193416595459,
      "step": 4542
    },
    {
      "epoch": 0.36843212125665437,
      "grad_norm": 0.02546260878443718,
      "learning_rate": 0.0012,
      "loss": 2.6321005821228027,
      "step": 4544
    },
    {
      "epoch": 0.36859428328185534,
      "grad_norm": 0.04169449955224991,
      "learning_rate": 0.0012,
      "loss": 2.684314727783203,
      "step": 4546
    },
    {
      "epoch": 0.3687564453070563,
      "grad_norm": 0.045515548437833786,
      "learning_rate": 0.0012,
      "loss": 2.7139711380004883,
      "step": 4548
    },
    {
      "epoch": 0.36891860733225734,
      "grad_norm": 0.03672289848327637,
      "learning_rate": 0.0012,
      "loss": 2.6975502967834473,
      "step": 4550
    },
    {
      "epoch": 0.3690807693574583,
      "grad_norm": 0.029205111786723137,
      "learning_rate": 0.0012,
      "loss": 2.697786808013916,
      "step": 4552
    },
    {
      "epoch": 0.3692429313826593,
      "grad_norm": 0.031219391152262688,
      "learning_rate": 0.0012,
      "loss": 2.6448559761047363,
      "step": 4554
    },
    {
      "epoch": 0.3694050934078603,
      "grad_norm": 0.026072908192873,
      "learning_rate": 0.0012,
      "loss": 2.7066707611083984,
      "step": 4556
    },
    {
      "epoch": 0.3695672554330613,
      "grad_norm": 0.02505200169980526,
      "learning_rate": 0.0012,
      "loss": 2.6739532947540283,
      "step": 4558
    },
    {
      "epoch": 0.36972941745826227,
      "grad_norm": 0.04063449054956436,
      "learning_rate": 0.0012,
      "loss": 2.646573305130005,
      "step": 4560
    },
    {
      "epoch": 0.3698915794834633,
      "grad_norm": 0.024704020470380783,
      "learning_rate": 0.0012,
      "loss": 2.717033624649048,
      "step": 4562
    },
    {
      "epoch": 0.37005374150866427,
      "grad_norm": 0.044270895421504974,
      "learning_rate": 0.0012,
      "loss": 2.690143346786499,
      "step": 4564
    },
    {
      "epoch": 0.37021590353386524,
      "grad_norm": 0.08869249373674393,
      "learning_rate": 0.0012,
      "loss": 2.708207368850708,
      "step": 4566
    },
    {
      "epoch": 0.37037806555906627,
      "grad_norm": 0.06175697594881058,
      "learning_rate": 0.0012,
      "loss": 2.5895230770111084,
      "step": 4568
    },
    {
      "epoch": 0.37054022758426725,
      "grad_norm": 0.02667888067662716,
      "learning_rate": 0.0012,
      "loss": 2.6703805923461914,
      "step": 4570
    },
    {
      "epoch": 0.3707023896094682,
      "grad_norm": 0.08872358500957489,
      "learning_rate": 0.0012,
      "loss": 2.6468639373779297,
      "step": 4572
    },
    {
      "epoch": 0.37086455163466925,
      "grad_norm": 0.057345110923051834,
      "learning_rate": 0.0012,
      "loss": 2.65407657623291,
      "step": 4574
    },
    {
      "epoch": 0.3710267136598702,
      "grad_norm": 0.04000714048743248,
      "learning_rate": 0.0012,
      "loss": 2.649468183517456,
      "step": 4576
    },
    {
      "epoch": 0.3711888756850712,
      "grad_norm": 0.04441901296377182,
      "learning_rate": 0.0012,
      "loss": 2.694729804992676,
      "step": 4578
    },
    {
      "epoch": 0.3713510377102722,
      "grad_norm": 0.03191729262471199,
      "learning_rate": 0.0012,
      "loss": 2.65506649017334,
      "step": 4580
    },
    {
      "epoch": 0.3715131997354732,
      "grad_norm": 0.0273039098829031,
      "learning_rate": 0.0012,
      "loss": 2.7483115196228027,
      "step": 4582
    },
    {
      "epoch": 0.37167536176067417,
      "grad_norm": 0.023960918188095093,
      "learning_rate": 0.0012,
      "loss": 2.68587064743042,
      "step": 4584
    },
    {
      "epoch": 0.3718375237858752,
      "grad_norm": 0.032551851123571396,
      "learning_rate": 0.0012,
      "loss": 2.687370777130127,
      "step": 4586
    },
    {
      "epoch": 0.3719996858110762,
      "grad_norm": 0.041604794561862946,
      "learning_rate": 0.0012,
      "loss": 2.755986213684082,
      "step": 4588
    },
    {
      "epoch": 0.37216184783627715,
      "grad_norm": 0.02348725125193596,
      "learning_rate": 0.0012,
      "loss": 2.6972270011901855,
      "step": 4590
    },
    {
      "epoch": 0.3723240098614782,
      "grad_norm": 0.045764755457639694,
      "learning_rate": 0.0012,
      "loss": 2.694267988204956,
      "step": 4592
    },
    {
      "epoch": 0.37248617188667915,
      "grad_norm": 0.04365480691194534,
      "learning_rate": 0.0012,
      "loss": 2.698216438293457,
      "step": 4594
    },
    {
      "epoch": 0.3726483339118801,
      "grad_norm": 0.03464466705918312,
      "learning_rate": 0.0012,
      "loss": 2.6429882049560547,
      "step": 4596
    },
    {
      "epoch": 0.37281049593708115,
      "grad_norm": 0.024724625051021576,
      "learning_rate": 0.0012,
      "loss": 2.7161526679992676,
      "step": 4598
    },
    {
      "epoch": 0.37297265796228213,
      "grad_norm": 0.02679443173110485,
      "learning_rate": 0.0012,
      "loss": 2.700075626373291,
      "step": 4600
    },
    {
      "epoch": 0.3731348199874831,
      "grad_norm": 0.023325223475694656,
      "learning_rate": 0.0012,
      "loss": 2.5981621742248535,
      "step": 4602
    },
    {
      "epoch": 0.37329698201268413,
      "grad_norm": 0.035331983119249344,
      "learning_rate": 0.0012,
      "loss": 2.7206997871398926,
      "step": 4604
    },
    {
      "epoch": 0.3734591440378851,
      "grad_norm": 0.03297495096921921,
      "learning_rate": 0.0012,
      "loss": 2.709559440612793,
      "step": 4606
    },
    {
      "epoch": 0.3736213060630861,
      "grad_norm": 0.0385405533015728,
      "learning_rate": 0.0012,
      "loss": 2.6969194412231445,
      "step": 4608
    },
    {
      "epoch": 0.3737834680882871,
      "grad_norm": 0.024726038798689842,
      "learning_rate": 0.0012,
      "loss": 2.6825265884399414,
      "step": 4610
    },
    {
      "epoch": 0.3739456301134881,
      "grad_norm": 0.06499360501766205,
      "learning_rate": 0.0012,
      "loss": 2.700178861618042,
      "step": 4612
    },
    {
      "epoch": 0.37410779213868905,
      "grad_norm": 0.047421909868717194,
      "learning_rate": 0.0012,
      "loss": 2.6439785957336426,
      "step": 4614
    },
    {
      "epoch": 0.3742699541638901,
      "grad_norm": 0.03432715684175491,
      "learning_rate": 0.0012,
      "loss": 2.6648035049438477,
      "step": 4616
    },
    {
      "epoch": 0.37443211618909106,
      "grad_norm": 0.03599085658788681,
      "learning_rate": 0.0012,
      "loss": 2.6497087478637695,
      "step": 4618
    },
    {
      "epoch": 0.37459427821429203,
      "grad_norm": 0.04576596990227699,
      "learning_rate": 0.0012,
      "loss": 2.690424919128418,
      "step": 4620
    },
    {
      "epoch": 0.37475644023949306,
      "grad_norm": 0.02372639626264572,
      "learning_rate": 0.0012,
      "loss": 2.630099296569824,
      "step": 4622
    },
    {
      "epoch": 0.37491860226469403,
      "grad_norm": 0.029867662116885185,
      "learning_rate": 0.0012,
      "loss": 2.669156551361084,
      "step": 4624
    },
    {
      "epoch": 0.375080764289895,
      "grad_norm": 0.02405640110373497,
      "learning_rate": 0.0012,
      "loss": 2.700439929962158,
      "step": 4626
    },
    {
      "epoch": 0.37524292631509604,
      "grad_norm": 0.02645825408399105,
      "learning_rate": 0.0012,
      "loss": 2.6621017456054688,
      "step": 4628
    },
    {
      "epoch": 0.375405088340297,
      "grad_norm": 0.02374751679599285,
      "learning_rate": 0.0012,
      "loss": 2.7279601097106934,
      "step": 4630
    },
    {
      "epoch": 0.375567250365498,
      "grad_norm": 0.04245816543698311,
      "learning_rate": 0.0012,
      "loss": 2.6605491638183594,
      "step": 4632
    },
    {
      "epoch": 0.375729412390699,
      "grad_norm": 0.03016858920454979,
      "learning_rate": 0.0012,
      "loss": 2.679145336151123,
      "step": 4634
    },
    {
      "epoch": 0.3758915744159,
      "grad_norm": 0.027132602408528328,
      "learning_rate": 0.0012,
      "loss": 2.683394193649292,
      "step": 4636
    },
    {
      "epoch": 0.37605373644110096,
      "grad_norm": 0.02566659450531006,
      "learning_rate": 0.0012,
      "loss": 2.657562255859375,
      "step": 4638
    },
    {
      "epoch": 0.376215898466302,
      "grad_norm": 0.03762071952223778,
      "learning_rate": 0.0012,
      "loss": 2.63862943649292,
      "step": 4640
    },
    {
      "epoch": 0.37637806049150296,
      "grad_norm": 0.029794320464134216,
      "learning_rate": 0.0012,
      "loss": 2.6761178970336914,
      "step": 4642
    },
    {
      "epoch": 0.37654022251670394,
      "grad_norm": 0.03493839502334595,
      "learning_rate": 0.0012,
      "loss": 2.687554359436035,
      "step": 4644
    },
    {
      "epoch": 0.37670238454190497,
      "grad_norm": 0.027367370203137398,
      "learning_rate": 0.0012,
      "loss": 2.654033660888672,
      "step": 4646
    },
    {
      "epoch": 0.37686454656710594,
      "grad_norm": 0.029897576197981834,
      "learning_rate": 0.0012,
      "loss": 2.704519748687744,
      "step": 4648
    },
    {
      "epoch": 0.3770267085923069,
      "grad_norm": 0.026547731831669807,
      "learning_rate": 0.0012,
      "loss": 2.742990493774414,
      "step": 4650
    },
    {
      "epoch": 0.37718887061750794,
      "grad_norm": 0.022149937227368355,
      "learning_rate": 0.0012,
      "loss": 2.6650924682617188,
      "step": 4652
    },
    {
      "epoch": 0.3773510326427089,
      "grad_norm": 0.026423625648021698,
      "learning_rate": 0.0012,
      "loss": 2.6989474296569824,
      "step": 4654
    },
    {
      "epoch": 0.3775131946679099,
      "grad_norm": 0.028326449915766716,
      "learning_rate": 0.0012,
      "loss": 2.711559534072876,
      "step": 4656
    },
    {
      "epoch": 0.3776753566931109,
      "grad_norm": 0.023982763290405273,
      "learning_rate": 0.0012,
      "loss": 2.669172525405884,
      "step": 4658
    },
    {
      "epoch": 0.3778375187183119,
      "grad_norm": 0.03270784020423889,
      "learning_rate": 0.0012,
      "loss": 2.7109408378601074,
      "step": 4660
    },
    {
      "epoch": 0.37799968074351287,
      "grad_norm": 0.030566267669200897,
      "learning_rate": 0.0012,
      "loss": 2.6296489238739014,
      "step": 4662
    },
    {
      "epoch": 0.3781618427687139,
      "grad_norm": 0.03945379704236984,
      "learning_rate": 0.0012,
      "loss": 2.695953130722046,
      "step": 4664
    },
    {
      "epoch": 0.37832400479391487,
      "grad_norm": 0.05583828315138817,
      "learning_rate": 0.0012,
      "loss": 2.688363552093506,
      "step": 4666
    },
    {
      "epoch": 0.37848616681911584,
      "grad_norm": 0.032572995871305466,
      "learning_rate": 0.0012,
      "loss": 2.6983447074890137,
      "step": 4668
    },
    {
      "epoch": 0.3786483288443169,
      "grad_norm": 0.025633079931139946,
      "learning_rate": 0.0012,
      "loss": 2.700899362564087,
      "step": 4670
    },
    {
      "epoch": 0.37881049086951785,
      "grad_norm": 0.043904900550842285,
      "learning_rate": 0.0012,
      "loss": 2.5942676067352295,
      "step": 4672
    },
    {
      "epoch": 0.3789726528947188,
      "grad_norm": 0.060619134455919266,
      "learning_rate": 0.0012,
      "loss": 2.7295479774475098,
      "step": 4674
    },
    {
      "epoch": 0.37913481491991985,
      "grad_norm": 0.034675464034080505,
      "learning_rate": 0.0012,
      "loss": 2.6169679164886475,
      "step": 4676
    },
    {
      "epoch": 0.3792969769451208,
      "grad_norm": 0.02484562061727047,
      "learning_rate": 0.0012,
      "loss": 2.626415252685547,
      "step": 4678
    },
    {
      "epoch": 0.3794591389703218,
      "grad_norm": 0.04893616959452629,
      "learning_rate": 0.0012,
      "loss": 2.6935949325561523,
      "step": 4680
    },
    {
      "epoch": 0.3796213009955228,
      "grad_norm": 0.03700084611773491,
      "learning_rate": 0.0012,
      "loss": 2.6995623111724854,
      "step": 4682
    },
    {
      "epoch": 0.3797834630207238,
      "grad_norm": 0.028696324676275253,
      "learning_rate": 0.0012,
      "loss": 2.676297664642334,
      "step": 4684
    },
    {
      "epoch": 0.3799456250459248,
      "grad_norm": 0.024192508310079575,
      "learning_rate": 0.0012,
      "loss": 2.7539544105529785,
      "step": 4686
    },
    {
      "epoch": 0.3801077870711258,
      "grad_norm": 0.026399172842502594,
      "learning_rate": 0.0012,
      "loss": 2.656883716583252,
      "step": 4688
    },
    {
      "epoch": 0.3802699490963268,
      "grad_norm": 0.029416276142001152,
      "learning_rate": 0.0012,
      "loss": 2.7231404781341553,
      "step": 4690
    },
    {
      "epoch": 0.38043211112152775,
      "grad_norm": 0.03652145341038704,
      "learning_rate": 0.0012,
      "loss": 2.670234203338623,
      "step": 4692
    },
    {
      "epoch": 0.3805942731467288,
      "grad_norm": 0.025926213711500168,
      "learning_rate": 0.0012,
      "loss": 2.691707134246826,
      "step": 4694
    },
    {
      "epoch": 0.38075643517192975,
      "grad_norm": 0.029094751924276352,
      "learning_rate": 0.0012,
      "loss": 2.7128257751464844,
      "step": 4696
    },
    {
      "epoch": 0.3809185971971307,
      "grad_norm": 0.029374312609434128,
      "learning_rate": 0.0012,
      "loss": 2.661712169647217,
      "step": 4698
    },
    {
      "epoch": 0.38108075922233176,
      "grad_norm": 0.02437419444322586,
      "learning_rate": 0.0012,
      "loss": 2.7123942375183105,
      "step": 4700
    },
    {
      "epoch": 0.38124292124753273,
      "grad_norm": 0.024104056879878044,
      "learning_rate": 0.0012,
      "loss": 2.6808552742004395,
      "step": 4702
    },
    {
      "epoch": 0.3814050832727337,
      "grad_norm": 0.022894909605383873,
      "learning_rate": 0.0012,
      "loss": 2.7183926105499268,
      "step": 4704
    },
    {
      "epoch": 0.38156724529793473,
      "grad_norm": 0.023922527208924294,
      "learning_rate": 0.0012,
      "loss": 2.7599937915802,
      "step": 4706
    },
    {
      "epoch": 0.3817294073231357,
      "grad_norm": 0.023969585075974464,
      "learning_rate": 0.0012,
      "loss": 2.7213940620422363,
      "step": 4708
    },
    {
      "epoch": 0.3818915693483367,
      "grad_norm": 0.027068687602877617,
      "learning_rate": 0.0012,
      "loss": 2.695631504058838,
      "step": 4710
    },
    {
      "epoch": 0.3820537313735377,
      "grad_norm": 0.021675704047083855,
      "learning_rate": 0.0012,
      "loss": 2.648353099822998,
      "step": 4712
    },
    {
      "epoch": 0.3822158933987387,
      "grad_norm": 0.02634415216743946,
      "learning_rate": 0.0012,
      "loss": 2.7302136421203613,
      "step": 4714
    },
    {
      "epoch": 0.38237805542393966,
      "grad_norm": 0.026178322732448578,
      "learning_rate": 0.0012,
      "loss": 2.706188678741455,
      "step": 4716
    },
    {
      "epoch": 0.3825402174491407,
      "grad_norm": 0.031239794567227364,
      "learning_rate": 0.0012,
      "loss": 2.697861671447754,
      "step": 4718
    },
    {
      "epoch": 0.38270237947434166,
      "grad_norm": 0.045124445110559464,
      "learning_rate": 0.0012,
      "loss": 2.6749939918518066,
      "step": 4720
    },
    {
      "epoch": 0.38286454149954263,
      "grad_norm": 0.028740687295794487,
      "learning_rate": 0.0012,
      "loss": 2.7250022888183594,
      "step": 4722
    },
    {
      "epoch": 0.38302670352474366,
      "grad_norm": 0.026818126440048218,
      "learning_rate": 0.0012,
      "loss": 2.692445755004883,
      "step": 4724
    },
    {
      "epoch": 0.38318886554994464,
      "grad_norm": 0.06118964031338692,
      "learning_rate": 0.0012,
      "loss": 2.7575368881225586,
      "step": 4726
    },
    {
      "epoch": 0.3833510275751456,
      "grad_norm": 0.032254938036203384,
      "learning_rate": 0.0012,
      "loss": 2.697134494781494,
      "step": 4728
    },
    {
      "epoch": 0.38351318960034664,
      "grad_norm": 0.026288948953151703,
      "learning_rate": 0.0012,
      "loss": 2.6851139068603516,
      "step": 4730
    },
    {
      "epoch": 0.3836753516255476,
      "grad_norm": 0.023170961067080498,
      "learning_rate": 0.0012,
      "loss": 2.7026963233947754,
      "step": 4732
    },
    {
      "epoch": 0.3838375136507486,
      "grad_norm": 0.02400396205484867,
      "learning_rate": 0.0012,
      "loss": 2.568598985671997,
      "step": 4734
    },
    {
      "epoch": 0.3839996756759496,
      "grad_norm": 0.030943144112825394,
      "learning_rate": 0.0012,
      "loss": 2.6740589141845703,
      "step": 4736
    },
    {
      "epoch": 0.3841618377011506,
      "grad_norm": 0.022596728056669235,
      "learning_rate": 0.0012,
      "loss": 2.6736297607421875,
      "step": 4738
    },
    {
      "epoch": 0.38432399972635156,
      "grad_norm": 0.025242827832698822,
      "learning_rate": 0.0012,
      "loss": 2.6817920207977295,
      "step": 4740
    },
    {
      "epoch": 0.3844861617515526,
      "grad_norm": 0.03902869299054146,
      "learning_rate": 0.0012,
      "loss": 2.7014379501342773,
      "step": 4742
    },
    {
      "epoch": 0.38464832377675356,
      "grad_norm": 0.06279482692480087,
      "learning_rate": 0.0012,
      "loss": 2.7414021492004395,
      "step": 4744
    },
    {
      "epoch": 0.38481048580195454,
      "grad_norm": 0.06419175863265991,
      "learning_rate": 0.0012,
      "loss": 2.6934714317321777,
      "step": 4746
    },
    {
      "epoch": 0.38497264782715557,
      "grad_norm": 0.0659211203455925,
      "learning_rate": 0.0012,
      "loss": 2.5822200775146484,
      "step": 4748
    },
    {
      "epoch": 0.38513480985235654,
      "grad_norm": 0.03956206142902374,
      "learning_rate": 0.0012,
      "loss": 2.6533873081207275,
      "step": 4750
    },
    {
      "epoch": 0.3852969718775575,
      "grad_norm": 0.03093840554356575,
      "learning_rate": 0.0012,
      "loss": 2.7221012115478516,
      "step": 4752
    },
    {
      "epoch": 0.38545913390275854,
      "grad_norm": 0.025348160415887833,
      "learning_rate": 0.0012,
      "loss": 2.7263855934143066,
      "step": 4754
    },
    {
      "epoch": 0.3856212959279595,
      "grad_norm": 0.026257522404193878,
      "learning_rate": 0.0012,
      "loss": 2.6994268894195557,
      "step": 4756
    },
    {
      "epoch": 0.3857834579531605,
      "grad_norm": 0.04074221849441528,
      "learning_rate": 0.0012,
      "loss": 2.7086358070373535,
      "step": 4758
    },
    {
      "epoch": 0.3859456199783615,
      "grad_norm": 0.030814865604043007,
      "learning_rate": 0.0012,
      "loss": 2.6680731773376465,
      "step": 4760
    },
    {
      "epoch": 0.3861077820035625,
      "grad_norm": 0.029269875958561897,
      "learning_rate": 0.0012,
      "loss": 2.6461987495422363,
      "step": 4762
    },
    {
      "epoch": 0.38626994402876347,
      "grad_norm": 0.04032265022397041,
      "learning_rate": 0.0012,
      "loss": 2.7402873039245605,
      "step": 4764
    },
    {
      "epoch": 0.3864321060539645,
      "grad_norm": 0.025329194962978363,
      "learning_rate": 0.0012,
      "loss": 2.693962812423706,
      "step": 4766
    },
    {
      "epoch": 0.38659426807916547,
      "grad_norm": 0.026710128411650658,
      "learning_rate": 0.0012,
      "loss": 2.7577619552612305,
      "step": 4768
    },
    {
      "epoch": 0.38675643010436644,
      "grad_norm": 0.024439699947834015,
      "learning_rate": 0.0012,
      "loss": 2.636456251144409,
      "step": 4770
    },
    {
      "epoch": 0.3869185921295675,
      "grad_norm": 0.02657175436615944,
      "learning_rate": 0.0012,
      "loss": 2.6804471015930176,
      "step": 4772
    },
    {
      "epoch": 0.38708075415476845,
      "grad_norm": 0.04019076004624367,
      "learning_rate": 0.0012,
      "loss": 2.6767001152038574,
      "step": 4774
    },
    {
      "epoch": 0.3872429161799694,
      "grad_norm": 0.06517177075147629,
      "learning_rate": 0.0012,
      "loss": 2.7317538261413574,
      "step": 4776
    },
    {
      "epoch": 0.38740507820517045,
      "grad_norm": 0.059285033494234085,
      "learning_rate": 0.0012,
      "loss": 2.6766104698181152,
      "step": 4778
    },
    {
      "epoch": 0.3875672402303714,
      "grad_norm": 0.028735466301441193,
      "learning_rate": 0.0012,
      "loss": 2.6938815116882324,
      "step": 4780
    },
    {
      "epoch": 0.3877294022555724,
      "grad_norm": 0.06623665243387222,
      "learning_rate": 0.0012,
      "loss": 2.6881771087646484,
      "step": 4782
    },
    {
      "epoch": 0.3878915642807734,
      "grad_norm": 0.09839429706335068,
      "learning_rate": 0.0012,
      "loss": 2.7030816078186035,
      "step": 4784
    },
    {
      "epoch": 0.3880537263059744,
      "grad_norm": 0.029471363872289658,
      "learning_rate": 0.0012,
      "loss": 2.7320122718811035,
      "step": 4786
    },
    {
      "epoch": 0.3882158883311754,
      "grad_norm": 0.050292909145355225,
      "learning_rate": 0.0012,
      "loss": 2.6884899139404297,
      "step": 4788
    },
    {
      "epoch": 0.3883780503563764,
      "grad_norm": 0.030686477199196815,
      "learning_rate": 0.0012,
      "loss": 2.621795654296875,
      "step": 4790
    },
    {
      "epoch": 0.3885402123815774,
      "grad_norm": 0.028859930112957954,
      "learning_rate": 0.0012,
      "loss": 2.6578989028930664,
      "step": 4792
    },
    {
      "epoch": 0.38870237440677835,
      "grad_norm": 0.02566026896238327,
      "learning_rate": 0.0012,
      "loss": 2.6747448444366455,
      "step": 4794
    },
    {
      "epoch": 0.3888645364319794,
      "grad_norm": 0.02642199955880642,
      "learning_rate": 0.0012,
      "loss": 2.6766581535339355,
      "step": 4796
    },
    {
      "epoch": 0.38902669845718035,
      "grad_norm": 0.03442374989390373,
      "learning_rate": 0.0012,
      "loss": 2.6568875312805176,
      "step": 4798
    },
    {
      "epoch": 0.3891888604823813,
      "grad_norm": 0.028053859248757362,
      "learning_rate": 0.0012,
      "loss": 2.7319746017456055,
      "step": 4800
    },
    {
      "epoch": 0.38935102250758236,
      "grad_norm": 0.0348837673664093,
      "learning_rate": 0.0012,
      "loss": 2.7233898639678955,
      "step": 4802
    },
    {
      "epoch": 0.38951318453278333,
      "grad_norm": 0.03327113017439842,
      "learning_rate": 0.0012,
      "loss": 2.7111244201660156,
      "step": 4804
    },
    {
      "epoch": 0.3896753465579843,
      "grad_norm": 0.031236723065376282,
      "learning_rate": 0.0012,
      "loss": 2.6887941360473633,
      "step": 4806
    },
    {
      "epoch": 0.38983750858318533,
      "grad_norm": 0.05817505717277527,
      "learning_rate": 0.0012,
      "loss": 2.678469181060791,
      "step": 4808
    },
    {
      "epoch": 0.3899996706083863,
      "grad_norm": 0.06143788993358612,
      "learning_rate": 0.0012,
      "loss": 2.717592239379883,
      "step": 4810
    },
    {
      "epoch": 0.3901618326335873,
      "grad_norm": 0.04866853728890419,
      "learning_rate": 0.0012,
      "loss": 2.6681387424468994,
      "step": 4812
    },
    {
      "epoch": 0.3903239946587883,
      "grad_norm": 0.02786880172789097,
      "learning_rate": 0.0012,
      "loss": 2.6828293800354004,
      "step": 4814
    },
    {
      "epoch": 0.3904861566839893,
      "grad_norm": 0.02740390971302986,
      "learning_rate": 0.0012,
      "loss": 2.7359910011291504,
      "step": 4816
    },
    {
      "epoch": 0.39064831870919026,
      "grad_norm": 0.060222078114748,
      "learning_rate": 0.0012,
      "loss": 2.6842262744903564,
      "step": 4818
    },
    {
      "epoch": 0.3908104807343913,
      "grad_norm": 0.03311151638627052,
      "learning_rate": 0.0012,
      "loss": 2.710219383239746,
      "step": 4820
    },
    {
      "epoch": 0.39097264275959226,
      "grad_norm": 0.024432463571429253,
      "learning_rate": 0.0012,
      "loss": 2.6565632820129395,
      "step": 4822
    },
    {
      "epoch": 0.39113480478479323,
      "grad_norm": 0.029658230021595955,
      "learning_rate": 0.0012,
      "loss": 2.6660385131835938,
      "step": 4824
    },
    {
      "epoch": 0.39129696680999426,
      "grad_norm": 0.04234076291322708,
      "learning_rate": 0.0012,
      "loss": 2.7322139739990234,
      "step": 4826
    },
    {
      "epoch": 0.39145912883519524,
      "grad_norm": 0.050408005714416504,
      "learning_rate": 0.0012,
      "loss": 2.6492152214050293,
      "step": 4828
    },
    {
      "epoch": 0.3916212908603962,
      "grad_norm": 0.052850209176540375,
      "learning_rate": 0.0012,
      "loss": 2.7012414932250977,
      "step": 4830
    },
    {
      "epoch": 0.39178345288559724,
      "grad_norm": 0.02429366670548916,
      "learning_rate": 0.0012,
      "loss": 2.667654514312744,
      "step": 4832
    },
    {
      "epoch": 0.3919456149107982,
      "grad_norm": 0.07471387088298798,
      "learning_rate": 0.0012,
      "loss": 2.7122607231140137,
      "step": 4834
    },
    {
      "epoch": 0.3921077769359992,
      "grad_norm": 0.05565783381462097,
      "learning_rate": 0.0012,
      "loss": 2.7030820846557617,
      "step": 4836
    },
    {
      "epoch": 0.3922699389612002,
      "grad_norm": 0.0339232012629509,
      "learning_rate": 0.0012,
      "loss": 2.7678165435791016,
      "step": 4838
    },
    {
      "epoch": 0.3924321009864012,
      "grad_norm": 0.026339827105402946,
      "learning_rate": 0.0012,
      "loss": 2.6990795135498047,
      "step": 4840
    },
    {
      "epoch": 0.39259426301160216,
      "grad_norm": 0.04239979386329651,
      "learning_rate": 0.0012,
      "loss": 2.626628875732422,
      "step": 4842
    },
    {
      "epoch": 0.3927564250368032,
      "grad_norm": 0.06370248645544052,
      "learning_rate": 0.0012,
      "loss": 2.6793746948242188,
      "step": 4844
    },
    {
      "epoch": 0.39291858706200417,
      "grad_norm": 0.023445947095751762,
      "learning_rate": 0.0012,
      "loss": 2.662670612335205,
      "step": 4846
    },
    {
      "epoch": 0.39308074908720514,
      "grad_norm": 0.08457820862531662,
      "learning_rate": 0.0012,
      "loss": 2.7349600791931152,
      "step": 4848
    },
    {
      "epoch": 0.39324291111240617,
      "grad_norm": 0.0435062050819397,
      "learning_rate": 0.0012,
      "loss": 2.677957057952881,
      "step": 4850
    },
    {
      "epoch": 0.39340507313760714,
      "grad_norm": 0.02940850518643856,
      "learning_rate": 0.0012,
      "loss": 2.681241035461426,
      "step": 4852
    },
    {
      "epoch": 0.3935672351628081,
      "grad_norm": 0.04980946332216263,
      "learning_rate": 0.0012,
      "loss": 2.74363112449646,
      "step": 4854
    },
    {
      "epoch": 0.39372939718800914,
      "grad_norm": 0.029815109446644783,
      "learning_rate": 0.0012,
      "loss": 2.6510820388793945,
      "step": 4856
    },
    {
      "epoch": 0.3938915592132101,
      "grad_norm": 0.035432808101177216,
      "learning_rate": 0.0012,
      "loss": 2.629812240600586,
      "step": 4858
    },
    {
      "epoch": 0.3940537212384111,
      "grad_norm": 0.06938643008470535,
      "learning_rate": 0.0012,
      "loss": 2.7834548950195312,
      "step": 4860
    },
    {
      "epoch": 0.3942158832636121,
      "grad_norm": 0.02250070311129093,
      "learning_rate": 0.0012,
      "loss": 2.6933507919311523,
      "step": 4862
    },
    {
      "epoch": 0.3943780452888131,
      "grad_norm": 0.027389992028474808,
      "learning_rate": 0.0012,
      "loss": 2.7134971618652344,
      "step": 4864
    },
    {
      "epoch": 0.39454020731401407,
      "grad_norm": 0.027697879821062088,
      "learning_rate": 0.0012,
      "loss": 2.7377378940582275,
      "step": 4866
    },
    {
      "epoch": 0.3947023693392151,
      "grad_norm": 0.024633187800645828,
      "learning_rate": 0.0012,
      "loss": 2.7059438228607178,
      "step": 4868
    },
    {
      "epoch": 0.39486453136441607,
      "grad_norm": 0.030216852203011513,
      "learning_rate": 0.0012,
      "loss": 2.6237101554870605,
      "step": 4870
    },
    {
      "epoch": 0.39502669338961705,
      "grad_norm": 0.028726696968078613,
      "learning_rate": 0.0012,
      "loss": 2.7095041275024414,
      "step": 4872
    },
    {
      "epoch": 0.3951888554148181,
      "grad_norm": 0.022040778771042824,
      "learning_rate": 0.0012,
      "loss": 2.647516965866089,
      "step": 4874
    },
    {
      "epoch": 0.39535101744001905,
      "grad_norm": 0.0284586139023304,
      "learning_rate": 0.0012,
      "loss": 2.703775405883789,
      "step": 4876
    },
    {
      "epoch": 0.39551317946522,
      "grad_norm": 0.027038833126425743,
      "learning_rate": 0.0012,
      "loss": 2.6686325073242188,
      "step": 4878
    },
    {
      "epoch": 0.39567534149042105,
      "grad_norm": 0.0253834780305624,
      "learning_rate": 0.0012,
      "loss": 2.7275314331054688,
      "step": 4880
    },
    {
      "epoch": 0.395837503515622,
      "grad_norm": 0.024842046201229095,
      "learning_rate": 0.0012,
      "loss": 2.6941256523132324,
      "step": 4882
    },
    {
      "epoch": 0.395999665540823,
      "grad_norm": 0.03504138067364693,
      "learning_rate": 0.0012,
      "loss": 2.6996688842773438,
      "step": 4884
    },
    {
      "epoch": 0.396161827566024,
      "grad_norm": 0.03294457495212555,
      "learning_rate": 0.0012,
      "loss": 2.6790571212768555,
      "step": 4886
    },
    {
      "epoch": 0.396323989591225,
      "grad_norm": 0.03451908007264137,
      "learning_rate": 0.0012,
      "loss": 2.7840452194213867,
      "step": 4888
    },
    {
      "epoch": 0.396486151616426,
      "grad_norm": 0.04203465208411217,
      "learning_rate": 0.0012,
      "loss": 2.632627010345459,
      "step": 4890
    },
    {
      "epoch": 0.396648313641627,
      "grad_norm": 0.027664843946695328,
      "learning_rate": 0.0012,
      "loss": 2.660050392150879,
      "step": 4892
    },
    {
      "epoch": 0.396810475666828,
      "grad_norm": 0.02793406881392002,
      "learning_rate": 0.0012,
      "loss": 2.7134525775909424,
      "step": 4894
    },
    {
      "epoch": 0.39697263769202895,
      "grad_norm": 0.0284196138381958,
      "learning_rate": 0.0012,
      "loss": 2.6856374740600586,
      "step": 4896
    },
    {
      "epoch": 0.39713479971723,
      "grad_norm": 0.06478801369667053,
      "learning_rate": 0.0012,
      "loss": 2.6476664543151855,
      "step": 4898
    },
    {
      "epoch": 0.39729696174243095,
      "grad_norm": 0.0937022864818573,
      "learning_rate": 0.0012,
      "loss": 2.6896982192993164,
      "step": 4900
    },
    {
      "epoch": 0.3974591237676319,
      "grad_norm": 0.03466242924332619,
      "learning_rate": 0.0012,
      "loss": 2.7195658683776855,
      "step": 4902
    },
    {
      "epoch": 0.39762128579283296,
      "grad_norm": 0.05331851541996002,
      "learning_rate": 0.0012,
      "loss": 2.6661338806152344,
      "step": 4904
    },
    {
      "epoch": 0.39778344781803393,
      "grad_norm": 0.05468218773603439,
      "learning_rate": 0.0012,
      "loss": 2.7233071327209473,
      "step": 4906
    },
    {
      "epoch": 0.3979456098432349,
      "grad_norm": 0.02738098055124283,
      "learning_rate": 0.0012,
      "loss": 2.70479416847229,
      "step": 4908
    },
    {
      "epoch": 0.39810777186843593,
      "grad_norm": 0.06460115313529968,
      "learning_rate": 0.0012,
      "loss": 2.7323808670043945,
      "step": 4910
    },
    {
      "epoch": 0.3982699338936369,
      "grad_norm": 0.08126769959926605,
      "learning_rate": 0.0012,
      "loss": 2.6886954307556152,
      "step": 4912
    },
    {
      "epoch": 0.3984320959188379,
      "grad_norm": 0.029987599700689316,
      "learning_rate": 0.0012,
      "loss": 2.686911106109619,
      "step": 4914
    },
    {
      "epoch": 0.3985942579440389,
      "grad_norm": 0.04924807697534561,
      "learning_rate": 0.0012,
      "loss": 2.6559762954711914,
      "step": 4916
    },
    {
      "epoch": 0.3987564199692399,
      "grad_norm": 0.04682996869087219,
      "learning_rate": 0.0012,
      "loss": 2.661510467529297,
      "step": 4918
    },
    {
      "epoch": 0.39891858199444086,
      "grad_norm": 0.029171863570809364,
      "learning_rate": 0.0012,
      "loss": 2.6518242359161377,
      "step": 4920
    },
    {
      "epoch": 0.3990807440196419,
      "grad_norm": 0.05303361639380455,
      "learning_rate": 0.0012,
      "loss": 2.712749481201172,
      "step": 4922
    },
    {
      "epoch": 0.39924290604484286,
      "grad_norm": 0.03771597519516945,
      "learning_rate": 0.0012,
      "loss": 2.6913907527923584,
      "step": 4924
    },
    {
      "epoch": 0.39940506807004383,
      "grad_norm": 0.024527551606297493,
      "learning_rate": 0.0012,
      "loss": 2.6802148818969727,
      "step": 4926
    },
    {
      "epoch": 0.39956723009524486,
      "grad_norm": 0.028203696012496948,
      "learning_rate": 0.0012,
      "loss": 2.744990825653076,
      "step": 4928
    },
    {
      "epoch": 0.39972939212044584,
      "grad_norm": 0.026227660477161407,
      "learning_rate": 0.0012,
      "loss": 2.636396884918213,
      "step": 4930
    },
    {
      "epoch": 0.3998915541456468,
      "grad_norm": 0.04083673283457756,
      "learning_rate": 0.0012,
      "loss": 2.728837490081787,
      "step": 4932
    },
    {
      "epoch": 0.40005371617084784,
      "grad_norm": 0.05343370512127876,
      "learning_rate": 0.0012,
      "loss": 2.7540879249572754,
      "step": 4934
    },
    {
      "epoch": 0.4002158781960488,
      "grad_norm": 0.026652498170733452,
      "learning_rate": 0.0012,
      "loss": 2.6529524326324463,
      "step": 4936
    },
    {
      "epoch": 0.4003780402212498,
      "grad_norm": 0.024849163368344307,
      "learning_rate": 0.0012,
      "loss": 2.6908326148986816,
      "step": 4938
    },
    {
      "epoch": 0.4005402022464508,
      "grad_norm": 0.042730920016765594,
      "learning_rate": 0.0012,
      "loss": 2.712940216064453,
      "step": 4940
    },
    {
      "epoch": 0.4007023642716518,
      "grad_norm": 0.031303469091653824,
      "learning_rate": 0.0012,
      "loss": 2.6579949855804443,
      "step": 4942
    },
    {
      "epoch": 0.40086452629685276,
      "grad_norm": 0.027662931010127068,
      "learning_rate": 0.0012,
      "loss": 2.69244122505188,
      "step": 4944
    },
    {
      "epoch": 0.4010266883220538,
      "grad_norm": 0.021837834268808365,
      "learning_rate": 0.0012,
      "loss": 2.650083065032959,
      "step": 4946
    },
    {
      "epoch": 0.40118885034725477,
      "grad_norm": 0.03221727907657623,
      "learning_rate": 0.0012,
      "loss": 2.7854793071746826,
      "step": 4948
    },
    {
      "epoch": 0.40135101237245574,
      "grad_norm": 0.028153786435723305,
      "learning_rate": 0.0012,
      "loss": 2.6946654319763184,
      "step": 4950
    },
    {
      "epoch": 0.40151317439765677,
      "grad_norm": 0.04554562643170357,
      "learning_rate": 0.0012,
      "loss": 2.759256362915039,
      "step": 4952
    },
    {
      "epoch": 0.40167533642285774,
      "grad_norm": 0.056818172335624695,
      "learning_rate": 0.0012,
      "loss": 2.6881823539733887,
      "step": 4954
    },
    {
      "epoch": 0.4018374984480587,
      "grad_norm": 0.04493682086467743,
      "learning_rate": 0.0012,
      "loss": 2.7216930389404297,
      "step": 4956
    },
    {
      "epoch": 0.40199966047325975,
      "grad_norm": 0.02384767308831215,
      "learning_rate": 0.0012,
      "loss": 2.7467339038848877,
      "step": 4958
    },
    {
      "epoch": 0.4021618224984607,
      "grad_norm": 0.03561066463589668,
      "learning_rate": 0.0012,
      "loss": 2.6498584747314453,
      "step": 4960
    },
    {
      "epoch": 0.4023239845236617,
      "grad_norm": 0.05906302109360695,
      "learning_rate": 0.0012,
      "loss": 2.7047982215881348,
      "step": 4962
    },
    {
      "epoch": 0.4024861465488627,
      "grad_norm": 0.052777063101530075,
      "learning_rate": 0.0012,
      "loss": 2.690382719039917,
      "step": 4964
    },
    {
      "epoch": 0.4026483085740637,
      "grad_norm": 0.03417985513806343,
      "learning_rate": 0.0012,
      "loss": 2.749615430831909,
      "step": 4966
    },
    {
      "epoch": 0.40281047059926467,
      "grad_norm": 0.029731586575508118,
      "learning_rate": 0.0012,
      "loss": 2.6733198165893555,
      "step": 4968
    },
    {
      "epoch": 0.4029726326244657,
      "grad_norm": 0.04140664264559746,
      "learning_rate": 0.0012,
      "loss": 2.6530792713165283,
      "step": 4970
    },
    {
      "epoch": 0.4031347946496667,
      "grad_norm": 0.031136559322476387,
      "learning_rate": 0.0012,
      "loss": 2.660606861114502,
      "step": 4972
    },
    {
      "epoch": 0.40329695667486765,
      "grad_norm": 0.030022965744137764,
      "learning_rate": 0.0012,
      "loss": 2.7066571712493896,
      "step": 4974
    },
    {
      "epoch": 0.4034591187000687,
      "grad_norm": 0.03913067281246185,
      "learning_rate": 0.0012,
      "loss": 2.6651439666748047,
      "step": 4976
    },
    {
      "epoch": 0.40362128072526965,
      "grad_norm": 0.0263929795473814,
      "learning_rate": 0.0012,
      "loss": 2.721503257751465,
      "step": 4978
    },
    {
      "epoch": 0.4037834427504706,
      "grad_norm": 0.038239795714616776,
      "learning_rate": 0.0012,
      "loss": 2.630924701690674,
      "step": 4980
    },
    {
      "epoch": 0.40394560477567165,
      "grad_norm": 0.027734186500310898,
      "learning_rate": 0.0012,
      "loss": 2.6345300674438477,
      "step": 4982
    },
    {
      "epoch": 0.4041077668008726,
      "grad_norm": 0.023503858596086502,
      "learning_rate": 0.0012,
      "loss": 2.708258867263794,
      "step": 4984
    },
    {
      "epoch": 0.4042699288260736,
      "grad_norm": 0.02752828039228916,
      "learning_rate": 0.0012,
      "loss": 2.7197279930114746,
      "step": 4986
    },
    {
      "epoch": 0.40443209085127463,
      "grad_norm": 0.025008223950862885,
      "learning_rate": 0.0012,
      "loss": 2.7142581939697266,
      "step": 4988
    },
    {
      "epoch": 0.4045942528764756,
      "grad_norm": 0.0231292936950922,
      "learning_rate": 0.0012,
      "loss": 2.697725296020508,
      "step": 4990
    },
    {
      "epoch": 0.40475641490167663,
      "grad_norm": 0.026422733440995216,
      "learning_rate": 0.0012,
      "loss": 2.7059361934661865,
      "step": 4992
    },
    {
      "epoch": 0.4049185769268776,
      "grad_norm": 0.037755757570266724,
      "learning_rate": 0.0012,
      "loss": 2.654449462890625,
      "step": 4994
    },
    {
      "epoch": 0.4050807389520786,
      "grad_norm": 0.02538072131574154,
      "learning_rate": 0.0012,
      "loss": 2.6661839485168457,
      "step": 4996
    },
    {
      "epoch": 0.4052429009772796,
      "grad_norm": 0.029634103178977966,
      "learning_rate": 0.0012,
      "loss": 2.6535112857818604,
      "step": 4998
    },
    {
      "epoch": 0.4054050630024806,
      "grad_norm": 0.02927539125084877,
      "learning_rate": 0.0012,
      "loss": 2.6083879470825195,
      "step": 5000
    },
    {
      "epoch": 0.40556722502768155,
      "grad_norm": 0.02696283534169197,
      "learning_rate": 0.0012,
      "loss": 2.6826062202453613,
      "step": 5002
    },
    {
      "epoch": 0.4057293870528826,
      "grad_norm": 0.02558475360274315,
      "learning_rate": 0.0012,
      "loss": 2.6644062995910645,
      "step": 5004
    },
    {
      "epoch": 0.40589154907808356,
      "grad_norm": 0.023946870118379593,
      "learning_rate": 0.0012,
      "loss": 2.6465840339660645,
      "step": 5006
    },
    {
      "epoch": 0.40605371110328453,
      "grad_norm": 0.030150577425956726,
      "learning_rate": 0.0012,
      "loss": 2.669609546661377,
      "step": 5008
    },
    {
      "epoch": 0.40621587312848556,
      "grad_norm": 0.024974733591079712,
      "learning_rate": 0.0012,
      "loss": 2.662815570831299,
      "step": 5010
    },
    {
      "epoch": 0.40637803515368653,
      "grad_norm": 0.030544716864824295,
      "learning_rate": 0.0012,
      "loss": 2.62497615814209,
      "step": 5012
    },
    {
      "epoch": 0.4065401971788875,
      "grad_norm": 0.052532147616147995,
      "learning_rate": 0.0012,
      "loss": 2.663715362548828,
      "step": 5014
    },
    {
      "epoch": 0.40670235920408854,
      "grad_norm": 0.025858808308839798,
      "learning_rate": 0.0012,
      "loss": 2.6379880905151367,
      "step": 5016
    },
    {
      "epoch": 0.4068645212292895,
      "grad_norm": 0.0324845016002655,
      "learning_rate": 0.0012,
      "loss": 2.6783180236816406,
      "step": 5018
    },
    {
      "epoch": 0.4070266832544905,
      "grad_norm": 0.023177005350589752,
      "learning_rate": 0.0012,
      "loss": 2.696388006210327,
      "step": 5020
    },
    {
      "epoch": 0.4071888452796915,
      "grad_norm": 0.03338293731212616,
      "learning_rate": 0.0012,
      "loss": 2.730208396911621,
      "step": 5022
    },
    {
      "epoch": 0.4073510073048925,
      "grad_norm": 0.036363743245601654,
      "learning_rate": 0.0012,
      "loss": 2.663347005844116,
      "step": 5024
    },
    {
      "epoch": 0.40751316933009346,
      "grad_norm": 0.024128053337335587,
      "learning_rate": 0.0012,
      "loss": 2.6825222969055176,
      "step": 5026
    },
    {
      "epoch": 0.4076753313552945,
      "grad_norm": 0.025597376748919487,
      "learning_rate": 0.0012,
      "loss": 2.674476385116577,
      "step": 5028
    },
    {
      "epoch": 0.40783749338049546,
      "grad_norm": 0.028868986293673515,
      "learning_rate": 0.0012,
      "loss": 2.652596950531006,
      "step": 5030
    },
    {
      "epoch": 0.40799965540569644,
      "grad_norm": 0.0227936003357172,
      "learning_rate": 0.0012,
      "loss": 2.678126335144043,
      "step": 5032
    },
    {
      "epoch": 0.40816181743089747,
      "grad_norm": 0.025252150371670723,
      "learning_rate": 0.0012,
      "loss": 2.663339138031006,
      "step": 5034
    },
    {
      "epoch": 0.40832397945609844,
      "grad_norm": 0.024325162172317505,
      "learning_rate": 0.0012,
      "loss": 2.674516439437866,
      "step": 5036
    },
    {
      "epoch": 0.4084861414812994,
      "grad_norm": 0.04415741562843323,
      "learning_rate": 0.0012,
      "loss": 2.662438154220581,
      "step": 5038
    },
    {
      "epoch": 0.40864830350650044,
      "grad_norm": 0.03085862472653389,
      "learning_rate": 0.0012,
      "loss": 2.709048271179199,
      "step": 5040
    },
    {
      "epoch": 0.4088104655317014,
      "grad_norm": 0.05122816935181618,
      "learning_rate": 0.0012,
      "loss": 2.665116310119629,
      "step": 5042
    },
    {
      "epoch": 0.4089726275569024,
      "grad_norm": 0.061004411429166794,
      "learning_rate": 0.0012,
      "loss": 2.693666934967041,
      "step": 5044
    },
    {
      "epoch": 0.4091347895821034,
      "grad_norm": 0.05239752680063248,
      "learning_rate": 0.0012,
      "loss": 2.7042322158813477,
      "step": 5046
    },
    {
      "epoch": 0.4092969516073044,
      "grad_norm": 0.031819432973861694,
      "learning_rate": 0.0012,
      "loss": 2.635470390319824,
      "step": 5048
    },
    {
      "epoch": 0.40945911363250537,
      "grad_norm": 0.025034843012690544,
      "learning_rate": 0.0012,
      "loss": 2.6527981758117676,
      "step": 5050
    },
    {
      "epoch": 0.4096212756577064,
      "grad_norm": 0.023329811170697212,
      "learning_rate": 0.0012,
      "loss": 2.6206185817718506,
      "step": 5052
    },
    {
      "epoch": 0.40978343768290737,
      "grad_norm": 0.03393389284610748,
      "learning_rate": 0.0012,
      "loss": 2.7017276287078857,
      "step": 5054
    },
    {
      "epoch": 0.40994559970810834,
      "grad_norm": 0.03046322613954544,
      "learning_rate": 0.0012,
      "loss": 2.715012550354004,
      "step": 5056
    },
    {
      "epoch": 0.4101077617333094,
      "grad_norm": 0.02861681580543518,
      "learning_rate": 0.0012,
      "loss": 2.668421983718872,
      "step": 5058
    },
    {
      "epoch": 0.41026992375851035,
      "grad_norm": 0.02346843108534813,
      "learning_rate": 0.0012,
      "loss": 2.6403636932373047,
      "step": 5060
    },
    {
      "epoch": 0.4104320857837113,
      "grad_norm": 0.03218868747353554,
      "learning_rate": 0.0012,
      "loss": 2.6405842304229736,
      "step": 5062
    },
    {
      "epoch": 0.41059424780891235,
      "grad_norm": 0.024687817320227623,
      "learning_rate": 0.0012,
      "loss": 2.7141454219818115,
      "step": 5064
    },
    {
      "epoch": 0.4107564098341133,
      "grad_norm": 0.03668462857604027,
      "learning_rate": 0.0012,
      "loss": 2.6332974433898926,
      "step": 5066
    },
    {
      "epoch": 0.4109185718593143,
      "grad_norm": 0.025837523862719536,
      "learning_rate": 0.0012,
      "loss": 2.7045741081237793,
      "step": 5068
    },
    {
      "epoch": 0.4110807338845153,
      "grad_norm": 0.02249220386147499,
      "learning_rate": 0.0012,
      "loss": 2.6233716011047363,
      "step": 5070
    },
    {
      "epoch": 0.4112428959097163,
      "grad_norm": 0.027651866897940636,
      "learning_rate": 0.0012,
      "loss": 2.674363136291504,
      "step": 5072
    },
    {
      "epoch": 0.4114050579349173,
      "grad_norm": 0.029498793184757233,
      "learning_rate": 0.0012,
      "loss": 2.676908016204834,
      "step": 5074
    },
    {
      "epoch": 0.4115672199601183,
      "grad_norm": 0.024567387998104095,
      "learning_rate": 0.0012,
      "loss": 2.689384937286377,
      "step": 5076
    },
    {
      "epoch": 0.4117293819853193,
      "grad_norm": 0.023552488535642624,
      "learning_rate": 0.0012,
      "loss": 2.7453784942626953,
      "step": 5078
    },
    {
      "epoch": 0.41189154401052025,
      "grad_norm": 0.025059517472982407,
      "learning_rate": 0.0012,
      "loss": 2.6700053215026855,
      "step": 5080
    },
    {
      "epoch": 0.4120537060357213,
      "grad_norm": 0.047040704637765884,
      "learning_rate": 0.0012,
      "loss": 2.7025182247161865,
      "step": 5082
    },
    {
      "epoch": 0.41221586806092225,
      "grad_norm": 0.04679681360721588,
      "learning_rate": 0.0012,
      "loss": 2.704392433166504,
      "step": 5084
    },
    {
      "epoch": 0.4123780300861232,
      "grad_norm": 0.03793664649128914,
      "learning_rate": 0.0012,
      "loss": 2.681948661804199,
      "step": 5086
    },
    {
      "epoch": 0.41254019211132426,
      "grad_norm": 0.06304685771465302,
      "learning_rate": 0.0012,
      "loss": 2.7237324714660645,
      "step": 5088
    },
    {
      "epoch": 0.41270235413652523,
      "grad_norm": 0.08211499452590942,
      "learning_rate": 0.0012,
      "loss": 2.7018089294433594,
      "step": 5090
    },
    {
      "epoch": 0.4128645161617262,
      "grad_norm": 0.0558927059173584,
      "learning_rate": 0.0012,
      "loss": 2.6593127250671387,
      "step": 5092
    },
    {
      "epoch": 0.41302667818692723,
      "grad_norm": 0.026076665148139,
      "learning_rate": 0.0012,
      "loss": 2.774883270263672,
      "step": 5094
    },
    {
      "epoch": 0.4131888402121282,
      "grad_norm": 0.04328735172748566,
      "learning_rate": 0.0012,
      "loss": 2.6768124103546143,
      "step": 5096
    },
    {
      "epoch": 0.4133510022373292,
      "grad_norm": 0.03237491101026535,
      "learning_rate": 0.0012,
      "loss": 2.709946632385254,
      "step": 5098
    },
    {
      "epoch": 0.4135131642625302,
      "grad_norm": 0.03518221899867058,
      "learning_rate": 0.0012,
      "loss": 2.6136627197265625,
      "step": 5100
    },
    {
      "epoch": 0.4136753262877312,
      "grad_norm": 0.030188310891389847,
      "learning_rate": 0.0012,
      "loss": 2.667217493057251,
      "step": 5102
    },
    {
      "epoch": 0.41383748831293216,
      "grad_norm": 0.024258747696876526,
      "learning_rate": 0.0012,
      "loss": 2.6965270042419434,
      "step": 5104
    },
    {
      "epoch": 0.4139996503381332,
      "grad_norm": 0.05078992247581482,
      "learning_rate": 0.0012,
      "loss": 2.632145404815674,
      "step": 5106
    },
    {
      "epoch": 0.41416181236333416,
      "grad_norm": 0.024537203833460808,
      "learning_rate": 0.0012,
      "loss": 2.6592652797698975,
      "step": 5108
    },
    {
      "epoch": 0.41432397438853513,
      "grad_norm": 0.039840925484895706,
      "learning_rate": 0.0012,
      "loss": 2.6029677391052246,
      "step": 5110
    },
    {
      "epoch": 0.41448613641373616,
      "grad_norm": 0.049152206629514694,
      "learning_rate": 0.0012,
      "loss": 2.6972904205322266,
      "step": 5112
    },
    {
      "epoch": 0.41464829843893714,
      "grad_norm": 0.03062584064900875,
      "learning_rate": 0.0012,
      "loss": 2.635758876800537,
      "step": 5114
    },
    {
      "epoch": 0.4148104604641381,
      "grad_norm": 0.022821135818958282,
      "learning_rate": 0.0012,
      "loss": 2.7102789878845215,
      "step": 5116
    },
    {
      "epoch": 0.41497262248933914,
      "grad_norm": 0.024616554379463196,
      "learning_rate": 0.0012,
      "loss": 2.7596418857574463,
      "step": 5118
    },
    {
      "epoch": 0.4151347845145401,
      "grad_norm": 0.027005400508642197,
      "learning_rate": 0.0012,
      "loss": 2.709672451019287,
      "step": 5120
    },
    {
      "epoch": 0.4152969465397411,
      "grad_norm": 0.027019087225198746,
      "learning_rate": 0.0012,
      "loss": 2.638044834136963,
      "step": 5122
    },
    {
      "epoch": 0.4154591085649421,
      "grad_norm": 0.03742021694779396,
      "learning_rate": 0.0012,
      "loss": 2.677262783050537,
      "step": 5124
    },
    {
      "epoch": 0.4156212705901431,
      "grad_norm": 0.025293750688433647,
      "learning_rate": 0.0012,
      "loss": 2.7324142456054688,
      "step": 5126
    },
    {
      "epoch": 0.41578343261534406,
      "grad_norm": 0.02723565697669983,
      "learning_rate": 0.0012,
      "loss": 2.6362156867980957,
      "step": 5128
    },
    {
      "epoch": 0.4159455946405451,
      "grad_norm": 0.030869856476783752,
      "learning_rate": 0.0012,
      "loss": 2.6209025382995605,
      "step": 5130
    },
    {
      "epoch": 0.41610775666574606,
      "grad_norm": 0.06726556271314621,
      "learning_rate": 0.0012,
      "loss": 2.698805332183838,
      "step": 5132
    },
    {
      "epoch": 0.41626991869094704,
      "grad_norm": 0.09057871997356415,
      "learning_rate": 0.0012,
      "loss": 2.5888454914093018,
      "step": 5134
    },
    {
      "epoch": 0.41643208071614807,
      "grad_norm": 0.055097684264183044,
      "learning_rate": 0.0012,
      "loss": 2.6716384887695312,
      "step": 5136
    },
    {
      "epoch": 0.41659424274134904,
      "grad_norm": 0.02638394385576248,
      "learning_rate": 0.0012,
      "loss": 2.7141571044921875,
      "step": 5138
    },
    {
      "epoch": 0.41675640476655,
      "grad_norm": 0.038269829005002975,
      "learning_rate": 0.0012,
      "loss": 2.7252819538116455,
      "step": 5140
    },
    {
      "epoch": 0.41691856679175104,
      "grad_norm": 0.041987884789705276,
      "learning_rate": 0.0012,
      "loss": 2.6953136920928955,
      "step": 5142
    },
    {
      "epoch": 0.417080728816952,
      "grad_norm": 0.02769935131072998,
      "learning_rate": 0.0012,
      "loss": 2.733095645904541,
      "step": 5144
    },
    {
      "epoch": 0.417242890842153,
      "grad_norm": 0.04458322003483772,
      "learning_rate": 0.0012,
      "loss": 2.6698713302612305,
      "step": 5146
    },
    {
      "epoch": 0.417405052867354,
      "grad_norm": 0.06172165274620056,
      "learning_rate": 0.0012,
      "loss": 2.648332118988037,
      "step": 5148
    },
    {
      "epoch": 0.417567214892555,
      "grad_norm": 0.021582866087555885,
      "learning_rate": 0.0012,
      "loss": 2.6087727546691895,
      "step": 5150
    },
    {
      "epoch": 0.41772937691775597,
      "grad_norm": 0.028040191158652306,
      "learning_rate": 0.0012,
      "loss": 2.6744136810302734,
      "step": 5152
    },
    {
      "epoch": 0.417891538942957,
      "grad_norm": 0.026544824242591858,
      "learning_rate": 0.0012,
      "loss": 2.762768268585205,
      "step": 5154
    },
    {
      "epoch": 0.41805370096815797,
      "grad_norm": 0.03298662230372429,
      "learning_rate": 0.0012,
      "loss": 2.745115041732788,
      "step": 5156
    },
    {
      "epoch": 0.41821586299335894,
      "grad_norm": 0.04569371044635773,
      "learning_rate": 0.0012,
      "loss": 2.710834503173828,
      "step": 5158
    },
    {
      "epoch": 0.41837802501856,
      "grad_norm": 0.043509289622306824,
      "learning_rate": 0.0012,
      "loss": 2.6398138999938965,
      "step": 5160
    },
    {
      "epoch": 0.41854018704376095,
      "grad_norm": 0.02612403593957424,
      "learning_rate": 0.0012,
      "loss": 2.729748249053955,
      "step": 5162
    },
    {
      "epoch": 0.4187023490689619,
      "grad_norm": 0.029606910422444344,
      "learning_rate": 0.0012,
      "loss": 2.7026896476745605,
      "step": 5164
    },
    {
      "epoch": 0.41886451109416295,
      "grad_norm": 0.03589525446295738,
      "learning_rate": 0.0012,
      "loss": 2.7268166542053223,
      "step": 5166
    },
    {
      "epoch": 0.4190266731193639,
      "grad_norm": 0.02981496788561344,
      "learning_rate": 0.0012,
      "loss": 2.7091784477233887,
      "step": 5168
    },
    {
      "epoch": 0.4191888351445649,
      "grad_norm": 0.03534706309437752,
      "learning_rate": 0.0012,
      "loss": 2.734772205352783,
      "step": 5170
    },
    {
      "epoch": 0.4193509971697659,
      "grad_norm": 0.036266375333070755,
      "learning_rate": 0.0012,
      "loss": 2.719088554382324,
      "step": 5172
    },
    {
      "epoch": 0.4195131591949669,
      "grad_norm": 0.02895570918917656,
      "learning_rate": 0.0012,
      "loss": 2.6810569763183594,
      "step": 5174
    },
    {
      "epoch": 0.4196753212201679,
      "grad_norm": 0.044203806668519974,
      "learning_rate": 0.0012,
      "loss": 2.7059454917907715,
      "step": 5176
    },
    {
      "epoch": 0.4198374832453689,
      "grad_norm": 0.03616291284561157,
      "learning_rate": 0.0012,
      "loss": 2.6833367347717285,
      "step": 5178
    },
    {
      "epoch": 0.4199996452705699,
      "grad_norm": 0.023214535787701607,
      "learning_rate": 0.0012,
      "loss": 2.7206506729125977,
      "step": 5180
    },
    {
      "epoch": 0.42016180729577085,
      "grad_norm": 0.024271732196211815,
      "learning_rate": 0.0012,
      "loss": 2.5936670303344727,
      "step": 5182
    },
    {
      "epoch": 0.4203239693209719,
      "grad_norm": 0.04127153009176254,
      "learning_rate": 0.0012,
      "loss": 2.6864705085754395,
      "step": 5184
    },
    {
      "epoch": 0.42048613134617285,
      "grad_norm": 0.05822586640715599,
      "learning_rate": 0.0012,
      "loss": 2.6861226558685303,
      "step": 5186
    },
    {
      "epoch": 0.4206482933713738,
      "grad_norm": 0.06088555231690407,
      "learning_rate": 0.0012,
      "loss": 2.6153860092163086,
      "step": 5188
    },
    {
      "epoch": 0.42081045539657486,
      "grad_norm": 0.03198784217238426,
      "learning_rate": 0.0012,
      "loss": 2.6595563888549805,
      "step": 5190
    },
    {
      "epoch": 0.42097261742177583,
      "grad_norm": 0.05589326471090317,
      "learning_rate": 0.0012,
      "loss": 2.6727147102355957,
      "step": 5192
    },
    {
      "epoch": 0.4211347794469768,
      "grad_norm": 0.061539143323898315,
      "learning_rate": 0.0012,
      "loss": 2.675150156021118,
      "step": 5194
    },
    {
      "epoch": 0.42129694147217783,
      "grad_norm": 0.030880054458975792,
      "learning_rate": 0.0012,
      "loss": 2.675149440765381,
      "step": 5196
    },
    {
      "epoch": 0.4214591034973788,
      "grad_norm": 0.08204270154237747,
      "learning_rate": 0.0012,
      "loss": 2.663980007171631,
      "step": 5198
    },
    {
      "epoch": 0.4216212655225798,
      "grad_norm": 0.05189688876271248,
      "learning_rate": 0.0012,
      "loss": 2.717583179473877,
      "step": 5200
    },
    {
      "epoch": 0.4217834275477808,
      "grad_norm": 0.04515138268470764,
      "learning_rate": 0.0012,
      "loss": 2.671226978302002,
      "step": 5202
    },
    {
      "epoch": 0.4219455895729818,
      "grad_norm": 0.06118212267756462,
      "learning_rate": 0.0012,
      "loss": 2.6820359230041504,
      "step": 5204
    },
    {
      "epoch": 0.42210775159818276,
      "grad_norm": 0.026727894321084023,
      "learning_rate": 0.0012,
      "loss": 2.672628402709961,
      "step": 5206
    },
    {
      "epoch": 0.4222699136233838,
      "grad_norm": 0.036944154649972916,
      "learning_rate": 0.0012,
      "loss": 2.679034948348999,
      "step": 5208
    },
    {
      "epoch": 0.42243207564858476,
      "grad_norm": 0.0314788743853569,
      "learning_rate": 0.0012,
      "loss": 2.693382501602173,
      "step": 5210
    },
    {
      "epoch": 0.42259423767378573,
      "grad_norm": 0.04316367208957672,
      "learning_rate": 0.0012,
      "loss": 2.6800003051757812,
      "step": 5212
    },
    {
      "epoch": 0.42275639969898676,
      "grad_norm": 0.04382234439253807,
      "learning_rate": 0.0012,
      "loss": 2.683487892150879,
      "step": 5214
    },
    {
      "epoch": 0.42291856172418774,
      "grad_norm": 0.025264481082558632,
      "learning_rate": 0.0012,
      "loss": 2.6635239124298096,
      "step": 5216
    },
    {
      "epoch": 0.4230807237493887,
      "grad_norm": 0.06489188224077225,
      "learning_rate": 0.0012,
      "loss": 2.668395519256592,
      "step": 5218
    },
    {
      "epoch": 0.42324288577458974,
      "grad_norm": 0.02690781280398369,
      "learning_rate": 0.0012,
      "loss": 2.678006649017334,
      "step": 5220
    },
    {
      "epoch": 0.4234050477997907,
      "grad_norm": 0.027262957766652107,
      "learning_rate": 0.0012,
      "loss": 2.6892004013061523,
      "step": 5222
    },
    {
      "epoch": 0.4235672098249917,
      "grad_norm": 0.023830510675907135,
      "learning_rate": 0.0012,
      "loss": 2.6663382053375244,
      "step": 5224
    },
    {
      "epoch": 0.4237293718501927,
      "grad_norm": 0.03814254701137543,
      "learning_rate": 0.0012,
      "loss": 2.6576883792877197,
      "step": 5226
    },
    {
      "epoch": 0.4238915338753937,
      "grad_norm": 0.04448002949357033,
      "learning_rate": 0.0012,
      "loss": 2.7076525688171387,
      "step": 5228
    },
    {
      "epoch": 0.42405369590059466,
      "grad_norm": 0.0246022529900074,
      "learning_rate": 0.0012,
      "loss": 2.6460533142089844,
      "step": 5230
    },
    {
      "epoch": 0.4242158579257957,
      "grad_norm": 0.05088943615555763,
      "learning_rate": 0.0012,
      "loss": 2.702594757080078,
      "step": 5232
    },
    {
      "epoch": 0.42437801995099667,
      "grad_norm": 0.04603502154350281,
      "learning_rate": 0.0012,
      "loss": 2.619196891784668,
      "step": 5234
    },
    {
      "epoch": 0.42454018197619764,
      "grad_norm": 0.025565192103385925,
      "learning_rate": 0.0012,
      "loss": 2.712834119796753,
      "step": 5236
    },
    {
      "epoch": 0.42470234400139867,
      "grad_norm": 0.04507641866803169,
      "learning_rate": 0.0012,
      "loss": 2.671867847442627,
      "step": 5238
    },
    {
      "epoch": 0.42486450602659964,
      "grad_norm": 0.04814305156469345,
      "learning_rate": 0.0012,
      "loss": 2.711933135986328,
      "step": 5240
    },
    {
      "epoch": 0.4250266680518006,
      "grad_norm": 0.02451285347342491,
      "learning_rate": 0.0012,
      "loss": 2.708604574203491,
      "step": 5242
    },
    {
      "epoch": 0.42518883007700164,
      "grad_norm": 0.032122742384672165,
      "learning_rate": 0.0012,
      "loss": 2.7103610038757324,
      "step": 5244
    },
    {
      "epoch": 0.4253509921022026,
      "grad_norm": 0.04579661786556244,
      "learning_rate": 0.0012,
      "loss": 2.6660664081573486,
      "step": 5246
    },
    {
      "epoch": 0.4255131541274036,
      "grad_norm": 0.024825669825077057,
      "learning_rate": 0.0012,
      "loss": 2.7073516845703125,
      "step": 5248
    },
    {
      "epoch": 0.4256753161526046,
      "grad_norm": 0.036970220506191254,
      "learning_rate": 0.0012,
      "loss": 2.688936233520508,
      "step": 5250
    },
    {
      "epoch": 0.4258374781778056,
      "grad_norm": 0.04525221139192581,
      "learning_rate": 0.0012,
      "loss": 2.631633996963501,
      "step": 5252
    },
    {
      "epoch": 0.42599964020300657,
      "grad_norm": 0.022361960262060165,
      "learning_rate": 0.0012,
      "loss": 2.656352996826172,
      "step": 5254
    },
    {
      "epoch": 0.4261618022282076,
      "grad_norm": 0.026972714811563492,
      "learning_rate": 0.0012,
      "loss": 2.6432173252105713,
      "step": 5256
    },
    {
      "epoch": 0.42632396425340857,
      "grad_norm": 0.040360577404499054,
      "learning_rate": 0.0012,
      "loss": 2.63301420211792,
      "step": 5258
    },
    {
      "epoch": 0.42648612627860955,
      "grad_norm": 0.040385857224464417,
      "learning_rate": 0.0012,
      "loss": 2.658219337463379,
      "step": 5260
    },
    {
      "epoch": 0.4266482883038106,
      "grad_norm": 0.02855714038014412,
      "learning_rate": 0.0012,
      "loss": 2.6997179985046387,
      "step": 5262
    },
    {
      "epoch": 0.42681045032901155,
      "grad_norm": 0.02334594540297985,
      "learning_rate": 0.0012,
      "loss": 2.635927677154541,
      "step": 5264
    },
    {
      "epoch": 0.4269726123542125,
      "grad_norm": 0.02368699386715889,
      "learning_rate": 0.0012,
      "loss": 2.6997671127319336,
      "step": 5266
    },
    {
      "epoch": 0.42713477437941355,
      "grad_norm": 0.02539813704788685,
      "learning_rate": 0.0012,
      "loss": 2.680825710296631,
      "step": 5268
    },
    {
      "epoch": 0.4272969364046145,
      "grad_norm": 0.02668221853673458,
      "learning_rate": 0.0012,
      "loss": 2.7064428329467773,
      "step": 5270
    },
    {
      "epoch": 0.4274590984298155,
      "grad_norm": 0.03889841213822365,
      "learning_rate": 0.0012,
      "loss": 2.6545464992523193,
      "step": 5272
    },
    {
      "epoch": 0.4276212604550165,
      "grad_norm": 0.03566752001643181,
      "learning_rate": 0.0012,
      "loss": 2.665985584259033,
      "step": 5274
    },
    {
      "epoch": 0.4277834224802175,
      "grad_norm": 0.04262900725007057,
      "learning_rate": 0.0012,
      "loss": 2.659977436065674,
      "step": 5276
    },
    {
      "epoch": 0.4279455845054185,
      "grad_norm": 0.027572572231292725,
      "learning_rate": 0.0012,
      "loss": 2.671746253967285,
      "step": 5278
    },
    {
      "epoch": 0.4281077465306195,
      "grad_norm": 0.025689711794257164,
      "learning_rate": 0.0012,
      "loss": 2.636260986328125,
      "step": 5280
    },
    {
      "epoch": 0.4282699085558205,
      "grad_norm": 0.055362433195114136,
      "learning_rate": 0.0012,
      "loss": 2.7179877758026123,
      "step": 5282
    },
    {
      "epoch": 0.42843207058102145,
      "grad_norm": 0.06932824850082397,
      "learning_rate": 0.0012,
      "loss": 2.6375033855438232,
      "step": 5284
    },
    {
      "epoch": 0.4285942326062225,
      "grad_norm": 0.0445658415555954,
      "learning_rate": 0.0012,
      "loss": 2.6615309715270996,
      "step": 5286
    },
    {
      "epoch": 0.42875639463142345,
      "grad_norm": 0.024823995307087898,
      "learning_rate": 0.0012,
      "loss": 2.6886091232299805,
      "step": 5288
    },
    {
      "epoch": 0.42891855665662443,
      "grad_norm": 0.048461880534887314,
      "learning_rate": 0.0012,
      "loss": 2.655289649963379,
      "step": 5290
    },
    {
      "epoch": 0.42908071868182546,
      "grad_norm": 0.042357929050922394,
      "learning_rate": 0.0012,
      "loss": 2.6640067100524902,
      "step": 5292
    },
    {
      "epoch": 0.42924288070702643,
      "grad_norm": 0.04519741237163544,
      "learning_rate": 0.0012,
      "loss": 2.782918930053711,
      "step": 5294
    },
    {
      "epoch": 0.4294050427322274,
      "grad_norm": 0.05684025213122368,
      "learning_rate": 0.0012,
      "loss": 2.758002758026123,
      "step": 5296
    },
    {
      "epoch": 0.42956720475742843,
      "grad_norm": 0.02411201037466526,
      "learning_rate": 0.0012,
      "loss": 2.709298610687256,
      "step": 5298
    },
    {
      "epoch": 0.4297293667826294,
      "grad_norm": 0.051108453422784805,
      "learning_rate": 0.0012,
      "loss": 2.690524101257324,
      "step": 5300
    },
    {
      "epoch": 0.4298915288078304,
      "grad_norm": 0.048993002623319626,
      "learning_rate": 0.0012,
      "loss": 2.6670470237731934,
      "step": 5302
    },
    {
      "epoch": 0.4300536908330314,
      "grad_norm": 0.023434529080986977,
      "learning_rate": 0.0012,
      "loss": 2.6343774795532227,
      "step": 5304
    },
    {
      "epoch": 0.4302158528582324,
      "grad_norm": 0.02653324604034424,
      "learning_rate": 0.0012,
      "loss": 2.6501288414001465,
      "step": 5306
    },
    {
      "epoch": 0.43037801488343336,
      "grad_norm": 0.04657597467303276,
      "learning_rate": 0.0012,
      "loss": 2.722669839859009,
      "step": 5308
    },
    {
      "epoch": 0.4305401769086344,
      "grad_norm": 0.04663746803998947,
      "learning_rate": 0.0012,
      "loss": 2.6871559619903564,
      "step": 5310
    },
    {
      "epoch": 0.43070233893383536,
      "grad_norm": 0.02693290077149868,
      "learning_rate": 0.0012,
      "loss": 2.723902702331543,
      "step": 5312
    },
    {
      "epoch": 0.43086450095903633,
      "grad_norm": 0.03412333130836487,
      "learning_rate": 0.0012,
      "loss": 2.696002721786499,
      "step": 5314
    },
    {
      "epoch": 0.43102666298423736,
      "grad_norm": 0.04384445771574974,
      "learning_rate": 0.0012,
      "loss": 2.7252120971679688,
      "step": 5316
    },
    {
      "epoch": 0.43118882500943834,
      "grad_norm": 0.03547601401805878,
      "learning_rate": 0.0012,
      "loss": 2.64021635055542,
      "step": 5318
    },
    {
      "epoch": 0.4313509870346393,
      "grad_norm": 0.03614396974444389,
      "learning_rate": 0.0012,
      "loss": 2.7306950092315674,
      "step": 5320
    },
    {
      "epoch": 0.43151314905984034,
      "grad_norm": 0.030565623193979263,
      "learning_rate": 0.0012,
      "loss": 2.6753129959106445,
      "step": 5322
    },
    {
      "epoch": 0.4316753110850413,
      "grad_norm": 0.0559113547205925,
      "learning_rate": 0.0012,
      "loss": 2.7002768516540527,
      "step": 5324
    },
    {
      "epoch": 0.4318374731102423,
      "grad_norm": 0.030994962900877,
      "learning_rate": 0.0012,
      "loss": 2.7094576358795166,
      "step": 5326
    },
    {
      "epoch": 0.4319996351354433,
      "grad_norm": 0.02719217725098133,
      "learning_rate": 0.0012,
      "loss": 2.637791156768799,
      "step": 5328
    },
    {
      "epoch": 0.4321617971606443,
      "grad_norm": 0.03754093125462532,
      "learning_rate": 0.0012,
      "loss": 2.6550536155700684,
      "step": 5330
    },
    {
      "epoch": 0.43232395918584526,
      "grad_norm": 0.034262776374816895,
      "learning_rate": 0.0012,
      "loss": 2.702155351638794,
      "step": 5332
    },
    {
      "epoch": 0.4324861212110463,
      "grad_norm": 0.04307897016406059,
      "learning_rate": 0.0012,
      "loss": 2.7151553630828857,
      "step": 5334
    },
    {
      "epoch": 0.43264828323624727,
      "grad_norm": 0.06675084680318832,
      "learning_rate": 0.0012,
      "loss": 2.653306484222412,
      "step": 5336
    },
    {
      "epoch": 0.43281044526144824,
      "grad_norm": 0.03555292636156082,
      "learning_rate": 0.0012,
      "loss": 2.6558408737182617,
      "step": 5338
    },
    {
      "epoch": 0.43297260728664927,
      "grad_norm": 0.026836063712835312,
      "learning_rate": 0.0012,
      "loss": 2.670980930328369,
      "step": 5340
    },
    {
      "epoch": 0.43313476931185024,
      "grad_norm": 0.0550982803106308,
      "learning_rate": 0.0012,
      "loss": 2.7024571895599365,
      "step": 5342
    },
    {
      "epoch": 0.4332969313370512,
      "grad_norm": 0.06855601072311401,
      "learning_rate": 0.0012,
      "loss": 2.674290180206299,
      "step": 5344
    },
    {
      "epoch": 0.43345909336225225,
      "grad_norm": 0.0652850940823555,
      "learning_rate": 0.0012,
      "loss": 2.691406011581421,
      "step": 5346
    },
    {
      "epoch": 0.4336212553874532,
      "grad_norm": 0.025265570729970932,
      "learning_rate": 0.0012,
      "loss": 2.655158519744873,
      "step": 5348
    },
    {
      "epoch": 0.4337834174126542,
      "grad_norm": 0.03860032185912132,
      "learning_rate": 0.0012,
      "loss": 2.67584228515625,
      "step": 5350
    },
    {
      "epoch": 0.4339455794378552,
      "grad_norm": 0.0287553109228611,
      "learning_rate": 0.0012,
      "loss": 2.7081458568573,
      "step": 5352
    },
    {
      "epoch": 0.4341077414630562,
      "grad_norm": 0.024969106540083885,
      "learning_rate": 0.0012,
      "loss": 2.6430416107177734,
      "step": 5354
    },
    {
      "epoch": 0.43426990348825717,
      "grad_norm": 0.03392801433801651,
      "learning_rate": 0.0012,
      "loss": 2.657205581665039,
      "step": 5356
    },
    {
      "epoch": 0.4344320655134582,
      "grad_norm": 0.026384558528661728,
      "learning_rate": 0.0012,
      "loss": 2.70654296875,
      "step": 5358
    },
    {
      "epoch": 0.4345942275386592,
      "grad_norm": 0.028680957853794098,
      "learning_rate": 0.0012,
      "loss": 2.659419536590576,
      "step": 5360
    },
    {
      "epoch": 0.43475638956386015,
      "grad_norm": 0.032140735536813736,
      "learning_rate": 0.0012,
      "loss": 2.6858365535736084,
      "step": 5362
    },
    {
      "epoch": 0.4349185515890612,
      "grad_norm": 0.026000525802373886,
      "learning_rate": 0.0012,
      "loss": 2.65645694732666,
      "step": 5364
    },
    {
      "epoch": 0.43508071361426215,
      "grad_norm": 0.03518463671207428,
      "learning_rate": 0.0012,
      "loss": 2.681845188140869,
      "step": 5366
    },
    {
      "epoch": 0.4352428756394631,
      "grad_norm": 0.03324217349290848,
      "learning_rate": 0.0012,
      "loss": 2.6911964416503906,
      "step": 5368
    },
    {
      "epoch": 0.43540503766466415,
      "grad_norm": 0.03307127580046654,
      "learning_rate": 0.0012,
      "loss": 2.735297679901123,
      "step": 5370
    },
    {
      "epoch": 0.4355671996898651,
      "grad_norm": 0.03662750869989395,
      "learning_rate": 0.0012,
      "loss": 2.6404459476470947,
      "step": 5372
    },
    {
      "epoch": 0.4357293617150661,
      "grad_norm": 0.027124179527163506,
      "learning_rate": 0.0012,
      "loss": 2.6598925590515137,
      "step": 5374
    },
    {
      "epoch": 0.43589152374026713,
      "grad_norm": 0.02608444169163704,
      "learning_rate": 0.0012,
      "loss": 2.6959991455078125,
      "step": 5376
    },
    {
      "epoch": 0.4360536857654681,
      "grad_norm": 0.0243439432233572,
      "learning_rate": 0.0012,
      "loss": 2.662440061569214,
      "step": 5378
    },
    {
      "epoch": 0.4362158477906691,
      "grad_norm": 0.025256779044866562,
      "learning_rate": 0.0012,
      "loss": 2.683216094970703,
      "step": 5380
    },
    {
      "epoch": 0.4363780098158701,
      "grad_norm": 0.024775506928563118,
      "learning_rate": 0.0012,
      "loss": 2.730325222015381,
      "step": 5382
    },
    {
      "epoch": 0.4365401718410711,
      "grad_norm": 0.02898758463561535,
      "learning_rate": 0.0012,
      "loss": 2.7461118698120117,
      "step": 5384
    },
    {
      "epoch": 0.43670233386627205,
      "grad_norm": 0.026340514421463013,
      "learning_rate": 0.0012,
      "loss": 2.6570301055908203,
      "step": 5386
    },
    {
      "epoch": 0.4368644958914731,
      "grad_norm": 0.023090709000825882,
      "learning_rate": 0.0012,
      "loss": 2.6203463077545166,
      "step": 5388
    },
    {
      "epoch": 0.43702665791667406,
      "grad_norm": 0.03677474334836006,
      "learning_rate": 0.0012,
      "loss": 2.7228128910064697,
      "step": 5390
    },
    {
      "epoch": 0.43718881994187503,
      "grad_norm": 0.05920642241835594,
      "learning_rate": 0.0012,
      "loss": 2.6157307624816895,
      "step": 5392
    },
    {
      "epoch": 0.43735098196707606,
      "grad_norm": 0.042061060667037964,
      "learning_rate": 0.0012,
      "loss": 2.6332273483276367,
      "step": 5394
    },
    {
      "epoch": 0.43751314399227703,
      "grad_norm": 0.029538096860051155,
      "learning_rate": 0.0012,
      "loss": 2.680403232574463,
      "step": 5396
    },
    {
      "epoch": 0.437675306017478,
      "grad_norm": 0.027255989611148834,
      "learning_rate": 0.0012,
      "loss": 2.653233051300049,
      "step": 5398
    },
    {
      "epoch": 0.43783746804267903,
      "grad_norm": 0.03097512386739254,
      "learning_rate": 0.0012,
      "loss": 2.6767406463623047,
      "step": 5400
    },
    {
      "epoch": 0.43799963006788,
      "grad_norm": 0.024701515212655067,
      "learning_rate": 0.0012,
      "loss": 2.7603015899658203,
      "step": 5402
    },
    {
      "epoch": 0.438161792093081,
      "grad_norm": 0.025286519899964333,
      "learning_rate": 0.0012,
      "loss": 2.6992721557617188,
      "step": 5404
    },
    {
      "epoch": 0.438323954118282,
      "grad_norm": 0.024016989395022392,
      "learning_rate": 0.0012,
      "loss": 2.688765287399292,
      "step": 5406
    },
    {
      "epoch": 0.438486116143483,
      "grad_norm": 0.0339885950088501,
      "learning_rate": 0.0012,
      "loss": 2.6963624954223633,
      "step": 5408
    },
    {
      "epoch": 0.43864827816868396,
      "grad_norm": 0.03226419538259506,
      "learning_rate": 0.0012,
      "loss": 2.6404170989990234,
      "step": 5410
    },
    {
      "epoch": 0.438810440193885,
      "grad_norm": 0.026962948963046074,
      "learning_rate": 0.0012,
      "loss": 2.6757354736328125,
      "step": 5412
    },
    {
      "epoch": 0.43897260221908596,
      "grad_norm": 0.034063950181007385,
      "learning_rate": 0.0012,
      "loss": 2.7281460762023926,
      "step": 5414
    },
    {
      "epoch": 0.43913476424428693,
      "grad_norm": 0.042819418013095856,
      "learning_rate": 0.0012,
      "loss": 2.6699607372283936,
      "step": 5416
    },
    {
      "epoch": 0.43929692626948796,
      "grad_norm": 0.05860035493969917,
      "learning_rate": 0.0012,
      "loss": 2.650681734085083,
      "step": 5418
    },
    {
      "epoch": 0.43945908829468894,
      "grad_norm": 0.08235103636980057,
      "learning_rate": 0.0012,
      "loss": 2.697479009628296,
      "step": 5420
    },
    {
      "epoch": 0.4396212503198899,
      "grad_norm": 0.06509173661470413,
      "learning_rate": 0.0012,
      "loss": 2.669159412384033,
      "step": 5422
    },
    {
      "epoch": 0.43978341234509094,
      "grad_norm": 0.025796763598918915,
      "learning_rate": 0.0012,
      "loss": 2.684377670288086,
      "step": 5424
    },
    {
      "epoch": 0.4399455743702919,
      "grad_norm": 0.029121654108166695,
      "learning_rate": 0.0012,
      "loss": 2.6725525856018066,
      "step": 5426
    },
    {
      "epoch": 0.4401077363954929,
      "grad_norm": 0.025207405909895897,
      "learning_rate": 0.0012,
      "loss": 2.730060338973999,
      "step": 5428
    },
    {
      "epoch": 0.4402698984206939,
      "grad_norm": 0.02303890511393547,
      "learning_rate": 0.0012,
      "loss": 2.6581597328186035,
      "step": 5430
    },
    {
      "epoch": 0.4404320604458949,
      "grad_norm": 0.045297421514987946,
      "learning_rate": 0.0012,
      "loss": 2.6718454360961914,
      "step": 5432
    },
    {
      "epoch": 0.44059422247109586,
      "grad_norm": 0.05424115061759949,
      "learning_rate": 0.0012,
      "loss": 2.7281014919281006,
      "step": 5434
    },
    {
      "epoch": 0.4407563844962969,
      "grad_norm": 0.035065773874521255,
      "learning_rate": 0.0012,
      "loss": 2.6775341033935547,
      "step": 5436
    },
    {
      "epoch": 0.44091854652149787,
      "grad_norm": 0.039268460124731064,
      "learning_rate": 0.0012,
      "loss": 2.7495059967041016,
      "step": 5438
    },
    {
      "epoch": 0.44108070854669884,
      "grad_norm": 0.0465148501098156,
      "learning_rate": 0.0012,
      "loss": 2.703550100326538,
      "step": 5440
    },
    {
      "epoch": 0.44124287057189987,
      "grad_norm": 0.06358474493026733,
      "learning_rate": 0.0012,
      "loss": 2.7276017665863037,
      "step": 5442
    },
    {
      "epoch": 0.44140503259710084,
      "grad_norm": 0.028677338734269142,
      "learning_rate": 0.0012,
      "loss": 2.6929924488067627,
      "step": 5444
    },
    {
      "epoch": 0.4415671946223018,
      "grad_norm": 0.038367945700883865,
      "learning_rate": 0.0012,
      "loss": 2.6803994178771973,
      "step": 5446
    },
    {
      "epoch": 0.44172935664750285,
      "grad_norm": 0.028578484430909157,
      "learning_rate": 0.0012,
      "loss": 2.7011172771453857,
      "step": 5448
    },
    {
      "epoch": 0.4418915186727038,
      "grad_norm": 0.031120723113417625,
      "learning_rate": 0.0012,
      "loss": 2.6997036933898926,
      "step": 5450
    },
    {
      "epoch": 0.4420536806979048,
      "grad_norm": 0.026935650035738945,
      "learning_rate": 0.0012,
      "loss": 2.6954925060272217,
      "step": 5452
    },
    {
      "epoch": 0.4422158427231058,
      "grad_norm": 0.029925161972641945,
      "learning_rate": 0.0012,
      "loss": 2.714973211288452,
      "step": 5454
    },
    {
      "epoch": 0.4423780047483068,
      "grad_norm": 0.026228351518511772,
      "learning_rate": 0.0012,
      "loss": 2.6433796882629395,
      "step": 5456
    },
    {
      "epoch": 0.44254016677350777,
      "grad_norm": 0.034024838358163834,
      "learning_rate": 0.0012,
      "loss": 2.6882455348968506,
      "step": 5458
    },
    {
      "epoch": 0.4427023287987088,
      "grad_norm": 0.04583149775862694,
      "learning_rate": 0.0012,
      "loss": 2.646425485610962,
      "step": 5460
    },
    {
      "epoch": 0.4428644908239098,
      "grad_norm": 0.04614220932126045,
      "learning_rate": 0.0012,
      "loss": 2.6709139347076416,
      "step": 5462
    },
    {
      "epoch": 0.44302665284911075,
      "grad_norm": 0.03840501606464386,
      "learning_rate": 0.0012,
      "loss": 2.6864418983459473,
      "step": 5464
    },
    {
      "epoch": 0.4431888148743118,
      "grad_norm": 0.026528485119342804,
      "learning_rate": 0.0012,
      "loss": 2.6094882488250732,
      "step": 5466
    },
    {
      "epoch": 0.44335097689951275,
      "grad_norm": 0.043532464653253555,
      "learning_rate": 0.0012,
      "loss": 2.6851158142089844,
      "step": 5468
    },
    {
      "epoch": 0.4435131389247137,
      "grad_norm": 0.04594384506344795,
      "learning_rate": 0.0012,
      "loss": 2.6832118034362793,
      "step": 5470
    },
    {
      "epoch": 0.44367530094991475,
      "grad_norm": 0.029128646478056908,
      "learning_rate": 0.0012,
      "loss": 2.6625401973724365,
      "step": 5472
    },
    {
      "epoch": 0.4438374629751157,
      "grad_norm": 0.03349275887012482,
      "learning_rate": 0.0012,
      "loss": 2.7575721740722656,
      "step": 5474
    },
    {
      "epoch": 0.4439996250003167,
      "grad_norm": 0.07266229391098022,
      "learning_rate": 0.0012,
      "loss": 2.7360615730285645,
      "step": 5476
    },
    {
      "epoch": 0.44416178702551773,
      "grad_norm": 0.03390222787857056,
      "learning_rate": 0.0012,
      "loss": 2.7095093727111816,
      "step": 5478
    },
    {
      "epoch": 0.4443239490507187,
      "grad_norm": 0.03309578448534012,
      "learning_rate": 0.0012,
      "loss": 2.7068347930908203,
      "step": 5480
    },
    {
      "epoch": 0.4444861110759197,
      "grad_norm": 0.04940641298890114,
      "learning_rate": 0.0012,
      "loss": 2.7203497886657715,
      "step": 5482
    },
    {
      "epoch": 0.4446482731011207,
      "grad_norm": 0.02693997137248516,
      "learning_rate": 0.0012,
      "loss": 2.6431243419647217,
      "step": 5484
    },
    {
      "epoch": 0.4448104351263217,
      "grad_norm": 0.026093972846865654,
      "learning_rate": 0.0012,
      "loss": 2.7028751373291016,
      "step": 5486
    },
    {
      "epoch": 0.44497259715152265,
      "grad_norm": 0.04258375242352486,
      "learning_rate": 0.0012,
      "loss": 2.656160354614258,
      "step": 5488
    },
    {
      "epoch": 0.4451347591767237,
      "grad_norm": 0.06392578035593033,
      "learning_rate": 0.0012,
      "loss": 2.6514241695404053,
      "step": 5490
    },
    {
      "epoch": 0.44529692120192466,
      "grad_norm": 0.04473789036273956,
      "learning_rate": 0.0012,
      "loss": 2.6616897583007812,
      "step": 5492
    },
    {
      "epoch": 0.44545908322712563,
      "grad_norm": 0.027086658403277397,
      "learning_rate": 0.0012,
      "loss": 2.7040228843688965,
      "step": 5494
    },
    {
      "epoch": 0.44562124525232666,
      "grad_norm": 0.025424262508749962,
      "learning_rate": 0.0012,
      "loss": 2.7073307037353516,
      "step": 5496
    },
    {
      "epoch": 0.44578340727752763,
      "grad_norm": 0.036781735718250275,
      "learning_rate": 0.0012,
      "loss": 2.67449688911438,
      "step": 5498
    },
    {
      "epoch": 0.4459455693027286,
      "grad_norm": 0.023965176194906235,
      "learning_rate": 0.0012,
      "loss": 2.696019172668457,
      "step": 5500
    },
    {
      "epoch": 0.44610773132792964,
      "grad_norm": 0.036264508962631226,
      "learning_rate": 0.0012,
      "loss": 2.6555209159851074,
      "step": 5502
    },
    {
      "epoch": 0.4462698933531306,
      "grad_norm": 0.026078635826706886,
      "learning_rate": 0.0012,
      "loss": 2.637861967086792,
      "step": 5504
    },
    {
      "epoch": 0.4464320553783316,
      "grad_norm": 0.02267296239733696,
      "learning_rate": 0.0012,
      "loss": 2.7130303382873535,
      "step": 5506
    },
    {
      "epoch": 0.4465942174035326,
      "grad_norm": 0.05749819055199623,
      "learning_rate": 0.0012,
      "loss": 2.6640172004699707,
      "step": 5508
    },
    {
      "epoch": 0.4467563794287336,
      "grad_norm": 0.09578739106655121,
      "learning_rate": 0.0012,
      "loss": 2.635089159011841,
      "step": 5510
    },
    {
      "epoch": 0.44691854145393456,
      "grad_norm": 0.06754839420318604,
      "learning_rate": 0.0012,
      "loss": 2.6750717163085938,
      "step": 5512
    },
    {
      "epoch": 0.4470807034791356,
      "grad_norm": 0.04597495123744011,
      "learning_rate": 0.0012,
      "loss": 2.700826406478882,
      "step": 5514
    },
    {
      "epoch": 0.44724286550433656,
      "grad_norm": 0.07169143855571747,
      "learning_rate": 0.0012,
      "loss": 2.676513433456421,
      "step": 5516
    },
    {
      "epoch": 0.44740502752953754,
      "grad_norm": 0.041663993149995804,
      "learning_rate": 0.0012,
      "loss": 2.6527676582336426,
      "step": 5518
    },
    {
      "epoch": 0.44756718955473856,
      "grad_norm": 0.03456132858991623,
      "learning_rate": 0.0012,
      "loss": 2.693882942199707,
      "step": 5520
    },
    {
      "epoch": 0.44772935157993954,
      "grad_norm": 0.05466552451252937,
      "learning_rate": 0.0012,
      "loss": 2.73097562789917,
      "step": 5522
    },
    {
      "epoch": 0.4478915136051405,
      "grad_norm": 0.02931159734725952,
      "learning_rate": 0.0012,
      "loss": 2.693565607070923,
      "step": 5524
    },
    {
      "epoch": 0.44805367563034154,
      "grad_norm": 0.0584050677716732,
      "learning_rate": 0.0012,
      "loss": 2.6819868087768555,
      "step": 5526
    },
    {
      "epoch": 0.4482158376555425,
      "grad_norm": 0.040505293756723404,
      "learning_rate": 0.0012,
      "loss": 2.6534440517425537,
      "step": 5528
    },
    {
      "epoch": 0.4483779996807435,
      "grad_norm": 0.028095152229070663,
      "learning_rate": 0.0012,
      "loss": 2.7096152305603027,
      "step": 5530
    },
    {
      "epoch": 0.4485401617059445,
      "grad_norm": 0.03089808113873005,
      "learning_rate": 0.0012,
      "loss": 2.6692748069763184,
      "step": 5532
    },
    {
      "epoch": 0.4487023237311455,
      "grad_norm": 0.023754416033625603,
      "learning_rate": 0.0012,
      "loss": 2.7333555221557617,
      "step": 5534
    },
    {
      "epoch": 0.44886448575634647,
      "grad_norm": 0.028884587809443474,
      "learning_rate": 0.0012,
      "loss": 2.739346981048584,
      "step": 5536
    },
    {
      "epoch": 0.4490266477815475,
      "grad_norm": 0.028333494439721107,
      "learning_rate": 0.0012,
      "loss": 2.6923484802246094,
      "step": 5538
    },
    {
      "epoch": 0.44918880980674847,
      "grad_norm": 0.02191239595413208,
      "learning_rate": 0.0012,
      "loss": 2.635219097137451,
      "step": 5540
    },
    {
      "epoch": 0.44935097183194944,
      "grad_norm": 0.03335493803024292,
      "learning_rate": 0.0012,
      "loss": 2.6706838607788086,
      "step": 5542
    },
    {
      "epoch": 0.44951313385715047,
      "grad_norm": 0.034538041800260544,
      "learning_rate": 0.0012,
      "loss": 2.606062650680542,
      "step": 5544
    },
    {
      "epoch": 0.44967529588235144,
      "grad_norm": 0.03267938271164894,
      "learning_rate": 0.0012,
      "loss": 2.7150325775146484,
      "step": 5546
    },
    {
      "epoch": 0.4498374579075524,
      "grad_norm": 0.035395413637161255,
      "learning_rate": 0.0012,
      "loss": 2.6706995964050293,
      "step": 5548
    },
    {
      "epoch": 0.44999961993275345,
      "grad_norm": 0.02287847176194191,
      "learning_rate": 0.0012,
      "loss": 2.6976139545440674,
      "step": 5550
    },
    {
      "epoch": 0.4501617819579544,
      "grad_norm": 0.03589823842048645,
      "learning_rate": 0.0012,
      "loss": 2.577960252761841,
      "step": 5552
    },
    {
      "epoch": 0.4503239439831554,
      "grad_norm": 0.025265155360102654,
      "learning_rate": 0.0012,
      "loss": 2.6720874309539795,
      "step": 5554
    },
    {
      "epoch": 0.4504861060083564,
      "grad_norm": 0.043598324060440063,
      "learning_rate": 0.0012,
      "loss": 2.6561765670776367,
      "step": 5556
    },
    {
      "epoch": 0.4506482680335574,
      "grad_norm": 0.05952271446585655,
      "learning_rate": 0.0012,
      "loss": 2.7153759002685547,
      "step": 5558
    },
    {
      "epoch": 0.45081043005875837,
      "grad_norm": 0.05035829171538353,
      "learning_rate": 0.0012,
      "loss": 2.699455738067627,
      "step": 5560
    },
    {
      "epoch": 0.4509725920839594,
      "grad_norm": 0.026375344023108482,
      "learning_rate": 0.0012,
      "loss": 2.688333034515381,
      "step": 5562
    },
    {
      "epoch": 0.4511347541091604,
      "grad_norm": 0.05252157151699066,
      "learning_rate": 0.0012,
      "loss": 2.643498420715332,
      "step": 5564
    },
    {
      "epoch": 0.45129691613436135,
      "grad_norm": 0.054000768810510635,
      "learning_rate": 0.0012,
      "loss": 2.6700568199157715,
      "step": 5566
    },
    {
      "epoch": 0.4514590781595624,
      "grad_norm": 0.02301282435655594,
      "learning_rate": 0.0012,
      "loss": 2.6697421073913574,
      "step": 5568
    },
    {
      "epoch": 0.45162124018476335,
      "grad_norm": 0.050671376287937164,
      "learning_rate": 0.0012,
      "loss": 2.7389750480651855,
      "step": 5570
    },
    {
      "epoch": 0.4517834022099643,
      "grad_norm": 0.04837118089199066,
      "learning_rate": 0.0012,
      "loss": 2.7269370555877686,
      "step": 5572
    },
    {
      "epoch": 0.45194556423516535,
      "grad_norm": 0.025417566299438477,
      "learning_rate": 0.0012,
      "loss": 2.68469500541687,
      "step": 5574
    },
    {
      "epoch": 0.4521077262603663,
      "grad_norm": 0.07300276309251785,
      "learning_rate": 0.0012,
      "loss": 2.6658191680908203,
      "step": 5576
    },
    {
      "epoch": 0.4522698882855673,
      "grad_norm": 0.07474657148122787,
      "learning_rate": 0.0012,
      "loss": 2.6280953884124756,
      "step": 5578
    },
    {
      "epoch": 0.45243205031076833,
      "grad_norm": 0.0360095277428627,
      "learning_rate": 0.0012,
      "loss": 2.678866386413574,
      "step": 5580
    },
    {
      "epoch": 0.4525942123359693,
      "grad_norm": 0.05592654272913933,
      "learning_rate": 0.0012,
      "loss": 2.6895554065704346,
      "step": 5582
    },
    {
      "epoch": 0.4527563743611703,
      "grad_norm": 0.07131792604923248,
      "learning_rate": 0.0012,
      "loss": 2.7049636840820312,
      "step": 5584
    },
    {
      "epoch": 0.4529185363863713,
      "grad_norm": 0.024890733882784843,
      "learning_rate": 0.0012,
      "loss": 2.7239298820495605,
      "step": 5586
    },
    {
      "epoch": 0.4530806984115723,
      "grad_norm": 0.038064125925302505,
      "learning_rate": 0.0012,
      "loss": 2.6072287559509277,
      "step": 5588
    },
    {
      "epoch": 0.45324286043677325,
      "grad_norm": 0.02355383336544037,
      "learning_rate": 0.0012,
      "loss": 2.780606746673584,
      "step": 5590
    },
    {
      "epoch": 0.4534050224619743,
      "grad_norm": 0.03415745496749878,
      "learning_rate": 0.0012,
      "loss": 2.6815149784088135,
      "step": 5592
    },
    {
      "epoch": 0.45356718448717526,
      "grad_norm": 0.06483452022075653,
      "learning_rate": 0.0012,
      "loss": 2.7119553089141846,
      "step": 5594
    },
    {
      "epoch": 0.45372934651237623,
      "grad_norm": 0.042394816875457764,
      "learning_rate": 0.0012,
      "loss": 2.6592161655426025,
      "step": 5596
    },
    {
      "epoch": 0.45389150853757726,
      "grad_norm": 0.025126444175839424,
      "learning_rate": 0.0012,
      "loss": 2.738304615020752,
      "step": 5598
    },
    {
      "epoch": 0.45405367056277823,
      "grad_norm": 0.02425636537373066,
      "learning_rate": 0.0012,
      "loss": 2.699376344680786,
      "step": 5600
    },
    {
      "epoch": 0.4542158325879792,
      "grad_norm": 0.02925361506640911,
      "learning_rate": 0.0012,
      "loss": 2.744877338409424,
      "step": 5602
    },
    {
      "epoch": 0.45437799461318024,
      "grad_norm": 0.0338042676448822,
      "learning_rate": 0.0012,
      "loss": 2.6871933937072754,
      "step": 5604
    },
    {
      "epoch": 0.4545401566383812,
      "grad_norm": 0.0501515232026577,
      "learning_rate": 0.0012,
      "loss": 2.6725382804870605,
      "step": 5606
    },
    {
      "epoch": 0.4547023186635822,
      "grad_norm": 0.028245041146874428,
      "learning_rate": 0.0012,
      "loss": 2.6850125789642334,
      "step": 5608
    },
    {
      "epoch": 0.4548644806887832,
      "grad_norm": 0.025794511660933495,
      "learning_rate": 0.0012,
      "loss": 2.740814685821533,
      "step": 5610
    },
    {
      "epoch": 0.4550266427139842,
      "grad_norm": 0.0333353690803051,
      "learning_rate": 0.0012,
      "loss": 2.7670416831970215,
      "step": 5612
    },
    {
      "epoch": 0.45518880473918516,
      "grad_norm": 0.028320396319031715,
      "learning_rate": 0.0012,
      "loss": 2.6588759422302246,
      "step": 5614
    },
    {
      "epoch": 0.4553509667643862,
      "grad_norm": 0.030774610117077827,
      "learning_rate": 0.0012,
      "loss": 2.657806396484375,
      "step": 5616
    },
    {
      "epoch": 0.45551312878958716,
      "grad_norm": 0.027605639770627022,
      "learning_rate": 0.0012,
      "loss": 2.646699905395508,
      "step": 5618
    },
    {
      "epoch": 0.45567529081478814,
      "grad_norm": 0.024521905928850174,
      "learning_rate": 0.0012,
      "loss": 2.697990894317627,
      "step": 5620
    },
    {
      "epoch": 0.45583745283998917,
      "grad_norm": 0.04273606464266777,
      "learning_rate": 0.0012,
      "loss": 2.6767468452453613,
      "step": 5622
    },
    {
      "epoch": 0.45599961486519014,
      "grad_norm": 0.059797417372465134,
      "learning_rate": 0.0012,
      "loss": 2.6675539016723633,
      "step": 5624
    },
    {
      "epoch": 0.4561617768903911,
      "grad_norm": 0.04139013588428497,
      "learning_rate": 0.0012,
      "loss": 2.7016284465789795,
      "step": 5626
    },
    {
      "epoch": 0.45632393891559214,
      "grad_norm": 0.0306281428784132,
      "learning_rate": 0.0012,
      "loss": 2.7611212730407715,
      "step": 5628
    },
    {
      "epoch": 0.4564861009407931,
      "grad_norm": 0.028105609118938446,
      "learning_rate": 0.0012,
      "loss": 2.724008798599243,
      "step": 5630
    },
    {
      "epoch": 0.4566482629659941,
      "grad_norm": 0.02320462092757225,
      "learning_rate": 0.0012,
      "loss": 2.594954013824463,
      "step": 5632
    },
    {
      "epoch": 0.4568104249911951,
      "grad_norm": 0.03882677108049393,
      "learning_rate": 0.0012,
      "loss": 2.581223249435425,
      "step": 5634
    },
    {
      "epoch": 0.4569725870163961,
      "grad_norm": 0.030186204239726067,
      "learning_rate": 0.0012,
      "loss": 2.62916898727417,
      "step": 5636
    },
    {
      "epoch": 0.45713474904159707,
      "grad_norm": 0.022850284352898598,
      "learning_rate": 0.0012,
      "loss": 2.6772475242614746,
      "step": 5638
    },
    {
      "epoch": 0.4572969110667981,
      "grad_norm": 0.028421791270375252,
      "learning_rate": 0.0012,
      "loss": 2.636748790740967,
      "step": 5640
    },
    {
      "epoch": 0.45745907309199907,
      "grad_norm": 0.028631621971726418,
      "learning_rate": 0.0012,
      "loss": 2.7103562355041504,
      "step": 5642
    },
    {
      "epoch": 0.45762123511720004,
      "grad_norm": 0.039457160979509354,
      "learning_rate": 0.0012,
      "loss": 2.7292139530181885,
      "step": 5644
    },
    {
      "epoch": 0.45778339714240107,
      "grad_norm": 0.06133134663105011,
      "learning_rate": 0.0012,
      "loss": 2.7350945472717285,
      "step": 5646
    },
    {
      "epoch": 0.45794555916760205,
      "grad_norm": 0.05562873184680939,
      "learning_rate": 0.0012,
      "loss": 2.716440439224243,
      "step": 5648
    },
    {
      "epoch": 0.458107721192803,
      "grad_norm": 0.02593534253537655,
      "learning_rate": 0.0012,
      "loss": 2.6423768997192383,
      "step": 5650
    },
    {
      "epoch": 0.45826988321800405,
      "grad_norm": 0.026568807661533356,
      "learning_rate": 0.0012,
      "loss": 2.701559543609619,
      "step": 5652
    },
    {
      "epoch": 0.458432045243205,
      "grad_norm": 0.029310202226042747,
      "learning_rate": 0.0012,
      "loss": 2.698080539703369,
      "step": 5654
    },
    {
      "epoch": 0.458594207268406,
      "grad_norm": 0.03853340819478035,
      "learning_rate": 0.0012,
      "loss": 2.7108235359191895,
      "step": 5656
    },
    {
      "epoch": 0.458756369293607,
      "grad_norm": 0.031082624569535255,
      "learning_rate": 0.0012,
      "loss": 2.716254711151123,
      "step": 5658
    },
    {
      "epoch": 0.458918531318808,
      "grad_norm": 0.032566145062446594,
      "learning_rate": 0.0012,
      "loss": 2.7320103645324707,
      "step": 5660
    },
    {
      "epoch": 0.459080693344009,
      "grad_norm": 0.029270723462104797,
      "learning_rate": 0.0012,
      "loss": 2.6772828102111816,
      "step": 5662
    },
    {
      "epoch": 0.45924285536921,
      "grad_norm": 0.024534763768315315,
      "learning_rate": 0.0012,
      "loss": 2.6903562545776367,
      "step": 5664
    },
    {
      "epoch": 0.459405017394411,
      "grad_norm": 0.04746159911155701,
      "learning_rate": 0.0012,
      "loss": 2.624497890472412,
      "step": 5666
    },
    {
      "epoch": 0.459567179419612,
      "grad_norm": 0.04537773132324219,
      "learning_rate": 0.0012,
      "loss": 2.6708984375,
      "step": 5668
    },
    {
      "epoch": 0.459729341444813,
      "grad_norm": 0.06859485059976578,
      "learning_rate": 0.0012,
      "loss": 2.691673517227173,
      "step": 5670
    },
    {
      "epoch": 0.45989150347001395,
      "grad_norm": 0.05161122605204582,
      "learning_rate": 0.0012,
      "loss": 2.7177953720092773,
      "step": 5672
    },
    {
      "epoch": 0.460053665495215,
      "grad_norm": 0.024916818365454674,
      "learning_rate": 0.0012,
      "loss": 2.7223048210144043,
      "step": 5674
    },
    {
      "epoch": 0.46021582752041595,
      "grad_norm": 0.03285098448395729,
      "learning_rate": 0.0012,
      "loss": 2.678412914276123,
      "step": 5676
    },
    {
      "epoch": 0.46037798954561693,
      "grad_norm": 0.029567645862698555,
      "learning_rate": 0.0012,
      "loss": 2.698580741882324,
      "step": 5678
    },
    {
      "epoch": 0.46054015157081796,
      "grad_norm": 0.041454024612903595,
      "learning_rate": 0.0012,
      "loss": 2.6842808723449707,
      "step": 5680
    },
    {
      "epoch": 0.46070231359601893,
      "grad_norm": 0.054270777851343155,
      "learning_rate": 0.0012,
      "loss": 2.6746997833251953,
      "step": 5682
    },
    {
      "epoch": 0.4608644756212199,
      "grad_norm": 0.04238879308104515,
      "learning_rate": 0.0012,
      "loss": 2.742664337158203,
      "step": 5684
    },
    {
      "epoch": 0.46102663764642093,
      "grad_norm": 0.022707058116793633,
      "learning_rate": 0.0012,
      "loss": 2.657574415206909,
      "step": 5686
    },
    {
      "epoch": 0.4611887996716219,
      "grad_norm": 0.02929896116256714,
      "learning_rate": 0.0012,
      "loss": 2.685140609741211,
      "step": 5688
    },
    {
      "epoch": 0.4613509616968229,
      "grad_norm": 0.02322245016694069,
      "learning_rate": 0.0012,
      "loss": 2.6025328636169434,
      "step": 5690
    },
    {
      "epoch": 0.4615131237220239,
      "grad_norm": 0.035856690257787704,
      "learning_rate": 0.0012,
      "loss": 2.6315178871154785,
      "step": 5692
    },
    {
      "epoch": 0.4616752857472249,
      "grad_norm": 0.05686206743121147,
      "learning_rate": 0.0012,
      "loss": 2.625947952270508,
      "step": 5694
    },
    {
      "epoch": 0.46183744777242586,
      "grad_norm": 0.03299645707011223,
      "learning_rate": 0.0012,
      "loss": 2.6236844062805176,
      "step": 5696
    },
    {
      "epoch": 0.4619996097976269,
      "grad_norm": 0.023212185129523277,
      "learning_rate": 0.0012,
      "loss": 2.6900811195373535,
      "step": 5698
    },
    {
      "epoch": 0.46216177182282786,
      "grad_norm": 0.05735580250620842,
      "learning_rate": 0.0012,
      "loss": 2.6580731868743896,
      "step": 5700
    },
    {
      "epoch": 0.46232393384802883,
      "grad_norm": 0.06567972153425217,
      "learning_rate": 0.0012,
      "loss": 2.636695384979248,
      "step": 5702
    },
    {
      "epoch": 0.46248609587322986,
      "grad_norm": 0.02406308427453041,
      "learning_rate": 0.0012,
      "loss": 2.654853343963623,
      "step": 5704
    },
    {
      "epoch": 0.46264825789843084,
      "grad_norm": 0.034666724503040314,
      "learning_rate": 0.0012,
      "loss": 2.678280830383301,
      "step": 5706
    },
    {
      "epoch": 0.4628104199236318,
      "grad_norm": 0.0522906519472599,
      "learning_rate": 0.0012,
      "loss": 2.6353092193603516,
      "step": 5708
    },
    {
      "epoch": 0.46297258194883284,
      "grad_norm": 0.029242148622870445,
      "learning_rate": 0.0012,
      "loss": 2.696457862854004,
      "step": 5710
    },
    {
      "epoch": 0.4631347439740338,
      "grad_norm": 0.04029950499534607,
      "learning_rate": 0.0012,
      "loss": 2.7304930686950684,
      "step": 5712
    },
    {
      "epoch": 0.4632969059992348,
      "grad_norm": 0.05431315302848816,
      "learning_rate": 0.0012,
      "loss": 2.6797666549682617,
      "step": 5714
    },
    {
      "epoch": 0.4634590680244358,
      "grad_norm": 0.06274858862161636,
      "learning_rate": 0.0012,
      "loss": 2.652545213699341,
      "step": 5716
    },
    {
      "epoch": 0.4636212300496368,
      "grad_norm": 0.026094472035765648,
      "learning_rate": 0.0012,
      "loss": 2.6246724128723145,
      "step": 5718
    },
    {
      "epoch": 0.46378339207483776,
      "grad_norm": 0.03028496541082859,
      "learning_rate": 0.0012,
      "loss": 2.7127318382263184,
      "step": 5720
    },
    {
      "epoch": 0.4639455541000388,
      "grad_norm": 0.04429222643375397,
      "learning_rate": 0.0012,
      "loss": 2.7151999473571777,
      "step": 5722
    },
    {
      "epoch": 0.46410771612523977,
      "grad_norm": 0.02583923004567623,
      "learning_rate": 0.0012,
      "loss": 2.6312255859375,
      "step": 5724
    },
    {
      "epoch": 0.46426987815044074,
      "grad_norm": 0.04864686354994774,
      "learning_rate": 0.0012,
      "loss": 2.6873092651367188,
      "step": 5726
    },
    {
      "epoch": 0.46443204017564177,
      "grad_norm": 0.051121268421411514,
      "learning_rate": 0.0012,
      "loss": 2.666459560394287,
      "step": 5728
    },
    {
      "epoch": 0.46459420220084274,
      "grad_norm": 0.0431971438229084,
      "learning_rate": 0.0012,
      "loss": 2.7089619636535645,
      "step": 5730
    },
    {
      "epoch": 0.4647563642260437,
      "grad_norm": 0.04601959139108658,
      "learning_rate": 0.0012,
      "loss": 2.657979965209961,
      "step": 5732
    },
    {
      "epoch": 0.46491852625124475,
      "grad_norm": 0.08271805942058563,
      "learning_rate": 0.0012,
      "loss": 2.6342320442199707,
      "step": 5734
    },
    {
      "epoch": 0.4650806882764457,
      "grad_norm": 0.04332241415977478,
      "learning_rate": 0.0012,
      "loss": 2.6872398853302,
      "step": 5736
    },
    {
      "epoch": 0.4652428503016467,
      "grad_norm": 0.02825390361249447,
      "learning_rate": 0.0012,
      "loss": 2.669734239578247,
      "step": 5738
    },
    {
      "epoch": 0.4654050123268477,
      "grad_norm": 0.04291360452771187,
      "learning_rate": 0.0012,
      "loss": 2.6966867446899414,
      "step": 5740
    },
    {
      "epoch": 0.4655671743520487,
      "grad_norm": 0.04576437547802925,
      "learning_rate": 0.0012,
      "loss": 2.66791033744812,
      "step": 5742
    },
    {
      "epoch": 0.46572933637724967,
      "grad_norm": 0.026296311989426613,
      "learning_rate": 0.0012,
      "loss": 2.7744760513305664,
      "step": 5744
    },
    {
      "epoch": 0.4658914984024507,
      "grad_norm": 0.03179056942462921,
      "learning_rate": 0.0012,
      "loss": 2.669276714324951,
      "step": 5746
    },
    {
      "epoch": 0.4660536604276517,
      "grad_norm": 0.04280898720026016,
      "learning_rate": 0.0012,
      "loss": 2.6758170127868652,
      "step": 5748
    },
    {
      "epoch": 0.46621582245285265,
      "grad_norm": 0.02484147436916828,
      "learning_rate": 0.0012,
      "loss": 2.6386971473693848,
      "step": 5750
    },
    {
      "epoch": 0.4663779844780537,
      "grad_norm": 0.042410071939229965,
      "learning_rate": 0.0012,
      "loss": 2.73990535736084,
      "step": 5752
    },
    {
      "epoch": 0.46654014650325465,
      "grad_norm": 0.030618196353316307,
      "learning_rate": 0.0012,
      "loss": 2.6841719150543213,
      "step": 5754
    },
    {
      "epoch": 0.4667023085284556,
      "grad_norm": 0.026073934510350227,
      "learning_rate": 0.0012,
      "loss": 2.7035155296325684,
      "step": 5756
    },
    {
      "epoch": 0.46686447055365665,
      "grad_norm": 0.02378384955227375,
      "learning_rate": 0.0012,
      "loss": 2.6992576122283936,
      "step": 5758
    },
    {
      "epoch": 0.4670266325788576,
      "grad_norm": 0.02681259997189045,
      "learning_rate": 0.0012,
      "loss": 2.6766016483306885,
      "step": 5760
    },
    {
      "epoch": 0.4671887946040586,
      "grad_norm": 0.023988544940948486,
      "learning_rate": 0.0012,
      "loss": 2.7006330490112305,
      "step": 5762
    },
    {
      "epoch": 0.46735095662925963,
      "grad_norm": 0.03368532285094261,
      "learning_rate": 0.0012,
      "loss": 2.6404364109039307,
      "step": 5764
    },
    {
      "epoch": 0.4675131186544606,
      "grad_norm": 0.02584677003324032,
      "learning_rate": 0.0012,
      "loss": 2.6703648567199707,
      "step": 5766
    },
    {
      "epoch": 0.4676752806796616,
      "grad_norm": 0.024559741839766502,
      "learning_rate": 0.0012,
      "loss": 2.6696598529815674,
      "step": 5768
    },
    {
      "epoch": 0.4678374427048626,
      "grad_norm": 0.024273786693811417,
      "learning_rate": 0.0012,
      "loss": 2.686516761779785,
      "step": 5770
    },
    {
      "epoch": 0.4679996047300636,
      "grad_norm": 0.03266880661249161,
      "learning_rate": 0.0012,
      "loss": 2.6600341796875,
      "step": 5772
    },
    {
      "epoch": 0.46816176675526455,
      "grad_norm": 0.039835501462221146,
      "learning_rate": 0.0012,
      "loss": 2.735896110534668,
      "step": 5774
    },
    {
      "epoch": 0.4683239287804656,
      "grad_norm": 0.054444797337055206,
      "learning_rate": 0.0012,
      "loss": 2.707552909851074,
      "step": 5776
    },
    {
      "epoch": 0.46848609080566656,
      "grad_norm": 0.05091795697808266,
      "learning_rate": 0.0012,
      "loss": 2.5955209732055664,
      "step": 5778
    },
    {
      "epoch": 0.46864825283086753,
      "grad_norm": 0.027814971283078194,
      "learning_rate": 0.0012,
      "loss": 2.7621095180511475,
      "step": 5780
    },
    {
      "epoch": 0.46881041485606856,
      "grad_norm": 0.03703116253018379,
      "learning_rate": 0.0012,
      "loss": 2.6652278900146484,
      "step": 5782
    },
    {
      "epoch": 0.46897257688126953,
      "grad_norm": 0.04196619987487793,
      "learning_rate": 0.0012,
      "loss": 2.738744020462036,
      "step": 5784
    },
    {
      "epoch": 0.4691347389064705,
      "grad_norm": 0.03367151692509651,
      "learning_rate": 0.0012,
      "loss": 2.7056326866149902,
      "step": 5786
    },
    {
      "epoch": 0.46929690093167153,
      "grad_norm": 0.027067258954048157,
      "learning_rate": 0.0012,
      "loss": 2.666717767715454,
      "step": 5788
    },
    {
      "epoch": 0.4694590629568725,
      "grad_norm": 0.023748790845274925,
      "learning_rate": 0.0012,
      "loss": 2.7040388584136963,
      "step": 5790
    },
    {
      "epoch": 0.4696212249820735,
      "grad_norm": 0.04272811859846115,
      "learning_rate": 0.0012,
      "loss": 2.645411252975464,
      "step": 5792
    },
    {
      "epoch": 0.4697833870072745,
      "grad_norm": 0.027488240972161293,
      "learning_rate": 0.0012,
      "loss": 2.6226282119750977,
      "step": 5794
    },
    {
      "epoch": 0.4699455490324755,
      "grad_norm": 0.03904043510556221,
      "learning_rate": 0.0012,
      "loss": 2.7135396003723145,
      "step": 5796
    },
    {
      "epoch": 0.47010771105767646,
      "grad_norm": 0.03985447809100151,
      "learning_rate": 0.0012,
      "loss": 2.683408737182617,
      "step": 5798
    },
    {
      "epoch": 0.4702698730828775,
      "grad_norm": 0.040667880326509476,
      "learning_rate": 0.0012,
      "loss": 2.6941957473754883,
      "step": 5800
    },
    {
      "epoch": 0.47043203510807846,
      "grad_norm": 0.058085497468709946,
      "learning_rate": 0.0012,
      "loss": 2.6454596519470215,
      "step": 5802
    },
    {
      "epoch": 0.47059419713327943,
      "grad_norm": 0.023693662136793137,
      "learning_rate": 0.0012,
      "loss": 2.6839051246643066,
      "step": 5804
    },
    {
      "epoch": 0.47075635915848046,
      "grad_norm": 0.02827884443104267,
      "learning_rate": 0.0012,
      "loss": 2.706333637237549,
      "step": 5806
    },
    {
      "epoch": 0.47091852118368144,
      "grad_norm": 0.03511302173137665,
      "learning_rate": 0.0012,
      "loss": 2.7134504318237305,
      "step": 5808
    },
    {
      "epoch": 0.4710806832088824,
      "grad_norm": 0.02861478179693222,
      "learning_rate": 0.0012,
      "loss": 2.7538437843322754,
      "step": 5810
    },
    {
      "epoch": 0.47124284523408344,
      "grad_norm": 0.02566535584628582,
      "learning_rate": 0.0012,
      "loss": 2.7213358879089355,
      "step": 5812
    },
    {
      "epoch": 0.4714050072592844,
      "grad_norm": 0.037011951208114624,
      "learning_rate": 0.0012,
      "loss": 2.7367630004882812,
      "step": 5814
    },
    {
      "epoch": 0.4715671692844854,
      "grad_norm": 0.079286128282547,
      "learning_rate": 0.0012,
      "loss": 2.7420952320098877,
      "step": 5816
    },
    {
      "epoch": 0.4717293313096864,
      "grad_norm": 0.06242755427956581,
      "learning_rate": 0.0012,
      "loss": 2.6956067085266113,
      "step": 5818
    },
    {
      "epoch": 0.4718914933348874,
      "grad_norm": 0.027378451079130173,
      "learning_rate": 0.0012,
      "loss": 2.6736855506896973,
      "step": 5820
    },
    {
      "epoch": 0.47205365536008836,
      "grad_norm": 0.07388366013765335,
      "learning_rate": 0.0012,
      "loss": 2.677349090576172,
      "step": 5822
    },
    {
      "epoch": 0.4722158173852894,
      "grad_norm": 0.07330229878425598,
      "learning_rate": 0.0012,
      "loss": 2.688077926635742,
      "step": 5824
    },
    {
      "epoch": 0.47237797941049037,
      "grad_norm": 0.036280449479818344,
      "learning_rate": 0.0012,
      "loss": 2.6635384559631348,
      "step": 5826
    },
    {
      "epoch": 0.47254014143569134,
      "grad_norm": 0.03193242847919464,
      "learning_rate": 0.0012,
      "loss": 2.7179670333862305,
      "step": 5828
    },
    {
      "epoch": 0.47270230346089237,
      "grad_norm": 0.022939864546060562,
      "learning_rate": 0.0012,
      "loss": 2.670544385910034,
      "step": 5830
    },
    {
      "epoch": 0.47286446548609334,
      "grad_norm": 0.040037281811237335,
      "learning_rate": 0.0012,
      "loss": 2.6842594146728516,
      "step": 5832
    },
    {
      "epoch": 0.4730266275112943,
      "grad_norm": 0.050177063792943954,
      "learning_rate": 0.0012,
      "loss": 2.675791025161743,
      "step": 5834
    },
    {
      "epoch": 0.47318878953649535,
      "grad_norm": 0.02331562712788582,
      "learning_rate": 0.0012,
      "loss": 2.67891788482666,
      "step": 5836
    },
    {
      "epoch": 0.4733509515616963,
      "grad_norm": 0.02561144158244133,
      "learning_rate": 0.0012,
      "loss": 2.6635026931762695,
      "step": 5838
    },
    {
      "epoch": 0.4735131135868973,
      "grad_norm": 0.033037394285202026,
      "learning_rate": 0.0012,
      "loss": 2.673772096633911,
      "step": 5840
    },
    {
      "epoch": 0.4736752756120983,
      "grad_norm": 0.024271158501505852,
      "learning_rate": 0.0012,
      "loss": 2.655855178833008,
      "step": 5842
    },
    {
      "epoch": 0.4738374376372993,
      "grad_norm": 0.03896170109510422,
      "learning_rate": 0.0012,
      "loss": 2.7533273696899414,
      "step": 5844
    },
    {
      "epoch": 0.47399959966250027,
      "grad_norm": 0.06166965514421463,
      "learning_rate": 0.0012,
      "loss": 2.668181896209717,
      "step": 5846
    },
    {
      "epoch": 0.4741617616877013,
      "grad_norm": 0.037218254059553146,
      "learning_rate": 0.0012,
      "loss": 2.6754446029663086,
      "step": 5848
    },
    {
      "epoch": 0.4743239237129023,
      "grad_norm": 0.03510380536317825,
      "learning_rate": 0.0012,
      "loss": 2.6795475482940674,
      "step": 5850
    },
    {
      "epoch": 0.47448608573810325,
      "grad_norm": 0.03291625529527664,
      "learning_rate": 0.0012,
      "loss": 2.639270544052124,
      "step": 5852
    },
    {
      "epoch": 0.4746482477633043,
      "grad_norm": 0.02879556640982628,
      "learning_rate": 0.0012,
      "loss": 2.6273393630981445,
      "step": 5854
    },
    {
      "epoch": 0.47481040978850525,
      "grad_norm": 0.025490332394838333,
      "learning_rate": 0.0012,
      "loss": 2.6400537490844727,
      "step": 5856
    },
    {
      "epoch": 0.4749725718137062,
      "grad_norm": 0.02585497498512268,
      "learning_rate": 0.0012,
      "loss": 2.681746482849121,
      "step": 5858
    },
    {
      "epoch": 0.47513473383890725,
      "grad_norm": 0.023141633719205856,
      "learning_rate": 0.0012,
      "loss": 2.6567482948303223,
      "step": 5860
    },
    {
      "epoch": 0.4752968958641082,
      "grad_norm": 0.02885153703391552,
      "learning_rate": 0.0012,
      "loss": 2.7437744140625,
      "step": 5862
    },
    {
      "epoch": 0.4754590578893092,
      "grad_norm": 0.02738010883331299,
      "learning_rate": 0.0012,
      "loss": 2.700010299682617,
      "step": 5864
    },
    {
      "epoch": 0.47562121991451023,
      "grad_norm": 0.02634126879274845,
      "learning_rate": 0.0012,
      "loss": 2.680152654647827,
      "step": 5866
    },
    {
      "epoch": 0.4757833819397112,
      "grad_norm": 0.024230176582932472,
      "learning_rate": 0.0012,
      "loss": 2.7748541831970215,
      "step": 5868
    },
    {
      "epoch": 0.4759455439649122,
      "grad_norm": 0.029528237879276276,
      "learning_rate": 0.0012,
      "loss": 2.7234864234924316,
      "step": 5870
    },
    {
      "epoch": 0.4761077059901132,
      "grad_norm": 0.031596094369888306,
      "learning_rate": 0.0012,
      "loss": 2.6414999961853027,
      "step": 5872
    },
    {
      "epoch": 0.4762698680153142,
      "grad_norm": 0.02782192826271057,
      "learning_rate": 0.0012,
      "loss": 2.749053955078125,
      "step": 5874
    },
    {
      "epoch": 0.47643203004051515,
      "grad_norm": 0.04070867598056793,
      "learning_rate": 0.0012,
      "loss": 2.6632041931152344,
      "step": 5876
    },
    {
      "epoch": 0.4765941920657162,
      "grad_norm": 0.03565070033073425,
      "learning_rate": 0.0012,
      "loss": 2.584764003753662,
      "step": 5878
    },
    {
      "epoch": 0.47675635409091716,
      "grad_norm": 0.05752735212445259,
      "learning_rate": 0.0012,
      "loss": 2.6628055572509766,
      "step": 5880
    },
    {
      "epoch": 0.47691851611611813,
      "grad_norm": 0.03514270484447479,
      "learning_rate": 0.0012,
      "loss": 2.7216877937316895,
      "step": 5882
    },
    {
      "epoch": 0.47708067814131916,
      "grad_norm": 0.025480516254901886,
      "learning_rate": 0.0012,
      "loss": 2.7210118770599365,
      "step": 5884
    },
    {
      "epoch": 0.47724284016652013,
      "grad_norm": 0.029046455398201942,
      "learning_rate": 0.0012,
      "loss": 2.71038818359375,
      "step": 5886
    },
    {
      "epoch": 0.4774050021917211,
      "grad_norm": 0.025437746196985245,
      "learning_rate": 0.0012,
      "loss": 2.6246180534362793,
      "step": 5888
    },
    {
      "epoch": 0.47756716421692214,
      "grad_norm": 0.023144718259572983,
      "learning_rate": 0.0012,
      "loss": 2.7484660148620605,
      "step": 5890
    },
    {
      "epoch": 0.4777293262421231,
      "grad_norm": 0.053510088473558426,
      "learning_rate": 0.0012,
      "loss": 2.6055402755737305,
      "step": 5892
    },
    {
      "epoch": 0.4778914882673241,
      "grad_norm": 0.026827383786439896,
      "learning_rate": 0.0012,
      "loss": 2.735499858856201,
      "step": 5894
    },
    {
      "epoch": 0.4780536502925251,
      "grad_norm": 0.04576931893825531,
      "learning_rate": 0.0012,
      "loss": 2.6686625480651855,
      "step": 5896
    },
    {
      "epoch": 0.4782158123177261,
      "grad_norm": 0.048829708248376846,
      "learning_rate": 0.0012,
      "loss": 2.669370651245117,
      "step": 5898
    },
    {
      "epoch": 0.47837797434292706,
      "grad_norm": 0.03552674874663353,
      "learning_rate": 0.0012,
      "loss": 2.6920433044433594,
      "step": 5900
    },
    {
      "epoch": 0.4785401363681281,
      "grad_norm": 0.031351104378700256,
      "learning_rate": 0.0012,
      "loss": 2.6258435249328613,
      "step": 5902
    },
    {
      "epoch": 0.47870229839332906,
      "grad_norm": 0.026548316702246666,
      "learning_rate": 0.0012,
      "loss": 2.705029010772705,
      "step": 5904
    },
    {
      "epoch": 0.47886446041853004,
      "grad_norm": 0.03427843749523163,
      "learning_rate": 0.0012,
      "loss": 2.648995876312256,
      "step": 5906
    },
    {
      "epoch": 0.47902662244373106,
      "grad_norm": 0.02670031040906906,
      "learning_rate": 0.0012,
      "loss": 2.656022071838379,
      "step": 5908
    },
    {
      "epoch": 0.47918878446893204,
      "grad_norm": 0.0248801801353693,
      "learning_rate": 0.0012,
      "loss": 2.6600513458251953,
      "step": 5910
    },
    {
      "epoch": 0.479350946494133,
      "grad_norm": 0.036762744188308716,
      "learning_rate": 0.0012,
      "loss": 2.712005615234375,
      "step": 5912
    },
    {
      "epoch": 0.47951310851933404,
      "grad_norm": 0.030970091000199318,
      "learning_rate": 0.0012,
      "loss": 2.6876072883605957,
      "step": 5914
    },
    {
      "epoch": 0.479675270544535,
      "grad_norm": 0.0461006760597229,
      "learning_rate": 0.0012,
      "loss": 2.689802646636963,
      "step": 5916
    },
    {
      "epoch": 0.479837432569736,
      "grad_norm": 0.0336257703602314,
      "learning_rate": 0.0012,
      "loss": 2.6882195472717285,
      "step": 5918
    },
    {
      "epoch": 0.479999594594937,
      "grad_norm": 0.02419021725654602,
      "learning_rate": 0.0012,
      "loss": 2.6764109134674072,
      "step": 5920
    },
    {
      "epoch": 0.480161756620138,
      "grad_norm": 0.02529451623558998,
      "learning_rate": 0.0012,
      "loss": 2.6584837436676025,
      "step": 5922
    },
    {
      "epoch": 0.48032391864533897,
      "grad_norm": 0.038035936653614044,
      "learning_rate": 0.0012,
      "loss": 2.6533679962158203,
      "step": 5924
    },
    {
      "epoch": 0.48048608067054,
      "grad_norm": 0.026328561827540398,
      "learning_rate": 0.0012,
      "loss": 2.635396957397461,
      "step": 5926
    },
    {
      "epoch": 0.48064824269574097,
      "grad_norm": 0.024859318509697914,
      "learning_rate": 0.0012,
      "loss": 2.710692882537842,
      "step": 5928
    },
    {
      "epoch": 0.48081040472094194,
      "grad_norm": 0.026818403974175453,
      "learning_rate": 0.0012,
      "loss": 2.663167953491211,
      "step": 5930
    },
    {
      "epoch": 0.48097256674614297,
      "grad_norm": 0.02877137064933777,
      "learning_rate": 0.0012,
      "loss": 2.6820108890533447,
      "step": 5932
    },
    {
      "epoch": 0.48113472877134394,
      "grad_norm": 0.0576925165951252,
      "learning_rate": 0.0012,
      "loss": 2.6496365070343018,
      "step": 5934
    },
    {
      "epoch": 0.4812968907965449,
      "grad_norm": 0.05977887660264969,
      "learning_rate": 0.0012,
      "loss": 2.6011266708374023,
      "step": 5936
    },
    {
      "epoch": 0.48145905282174595,
      "grad_norm": 0.023821759968996048,
      "learning_rate": 0.0012,
      "loss": 2.6622250080108643,
      "step": 5938
    },
    {
      "epoch": 0.4816212148469469,
      "grad_norm": 0.02863987535238266,
      "learning_rate": 0.0012,
      "loss": 2.633828639984131,
      "step": 5940
    },
    {
      "epoch": 0.4817833768721479,
      "grad_norm": 0.0632578581571579,
      "learning_rate": 0.0012,
      "loss": 2.703906297683716,
      "step": 5942
    },
    {
      "epoch": 0.4819455388973489,
      "grad_norm": 0.0437769815325737,
      "learning_rate": 0.0012,
      "loss": 2.6617953777313232,
      "step": 5944
    },
    {
      "epoch": 0.4821077009225499,
      "grad_norm": 0.045486826449632645,
      "learning_rate": 0.0012,
      "loss": 2.662353038787842,
      "step": 5946
    },
    {
      "epoch": 0.48226986294775087,
      "grad_norm": 0.026900717988610268,
      "learning_rate": 0.0012,
      "loss": 2.6639084815979004,
      "step": 5948
    },
    {
      "epoch": 0.4824320249729519,
      "grad_norm": 0.05871735140681267,
      "learning_rate": 0.0012,
      "loss": 2.6787850856781006,
      "step": 5950
    },
    {
      "epoch": 0.4825941869981529,
      "grad_norm": 0.06429410725831985,
      "learning_rate": 0.0012,
      "loss": 2.6145682334899902,
      "step": 5952
    },
    {
      "epoch": 0.48275634902335385,
      "grad_norm": 0.035739824175834656,
      "learning_rate": 0.0012,
      "loss": 2.692279577255249,
      "step": 5954
    },
    {
      "epoch": 0.4829185110485549,
      "grad_norm": 0.02588878944516182,
      "learning_rate": 0.0012,
      "loss": 2.67110538482666,
      "step": 5956
    },
    {
      "epoch": 0.48308067307375585,
      "grad_norm": 0.025643879547715187,
      "learning_rate": 0.0012,
      "loss": 2.6444408893585205,
      "step": 5958
    },
    {
      "epoch": 0.4832428350989568,
      "grad_norm": 0.02911308780312538,
      "learning_rate": 0.0012,
      "loss": 2.684387445449829,
      "step": 5960
    },
    {
      "epoch": 0.48340499712415785,
      "grad_norm": 0.027196412906050682,
      "learning_rate": 0.0012,
      "loss": 2.6893773078918457,
      "step": 5962
    },
    {
      "epoch": 0.4835671591493588,
      "grad_norm": 0.027496758848428726,
      "learning_rate": 0.0012,
      "loss": 2.669646978378296,
      "step": 5964
    },
    {
      "epoch": 0.4837293211745598,
      "grad_norm": 0.02937772311270237,
      "learning_rate": 0.0012,
      "loss": 2.662504196166992,
      "step": 5966
    },
    {
      "epoch": 0.48389148319976083,
      "grad_norm": 0.025098437443375587,
      "learning_rate": 0.0012,
      "loss": 2.663727045059204,
      "step": 5968
    },
    {
      "epoch": 0.4840536452249618,
      "grad_norm": 0.024463756009936333,
      "learning_rate": 0.0012,
      "loss": 2.6821939945220947,
      "step": 5970
    },
    {
      "epoch": 0.4842158072501628,
      "grad_norm": 0.02944275364279747,
      "learning_rate": 0.0012,
      "loss": 2.677825450897217,
      "step": 5972
    },
    {
      "epoch": 0.4843779692753638,
      "grad_norm": 0.02797560766339302,
      "learning_rate": 0.0012,
      "loss": 2.62399959564209,
      "step": 5974
    },
    {
      "epoch": 0.4845401313005648,
      "grad_norm": 0.02799796313047409,
      "learning_rate": 0.0012,
      "loss": 2.7118048667907715,
      "step": 5976
    },
    {
      "epoch": 0.48470229332576575,
      "grad_norm": 0.05142607539892197,
      "learning_rate": 0.0012,
      "loss": 2.684844493865967,
      "step": 5978
    },
    {
      "epoch": 0.4848644553509668,
      "grad_norm": 0.02341652661561966,
      "learning_rate": 0.0012,
      "loss": 2.7180423736572266,
      "step": 5980
    },
    {
      "epoch": 0.48502661737616776,
      "grad_norm": 0.04287418723106384,
      "learning_rate": 0.0012,
      "loss": 2.663483142852783,
      "step": 5982
    },
    {
      "epoch": 0.48518877940136873,
      "grad_norm": 0.04996604844927788,
      "learning_rate": 0.0012,
      "loss": 2.723170280456543,
      "step": 5984
    },
    {
      "epoch": 0.48535094142656976,
      "grad_norm": 0.02299046330153942,
      "learning_rate": 0.0012,
      "loss": 2.7229628562927246,
      "step": 5986
    },
    {
      "epoch": 0.48551310345177073,
      "grad_norm": 0.04292365536093712,
      "learning_rate": 0.0012,
      "loss": 2.682373523712158,
      "step": 5988
    },
    {
      "epoch": 0.4856752654769717,
      "grad_norm": 0.051302794367074966,
      "learning_rate": 0.0012,
      "loss": 2.670732021331787,
      "step": 5990
    },
    {
      "epoch": 0.48583742750217274,
      "grad_norm": 0.03082759492099285,
      "learning_rate": 0.0012,
      "loss": 2.652235984802246,
      "step": 5992
    },
    {
      "epoch": 0.4859995895273737,
      "grad_norm": 0.02278500236570835,
      "learning_rate": 0.0012,
      "loss": 2.7359237670898438,
      "step": 5994
    },
    {
      "epoch": 0.4861617515525747,
      "grad_norm": 0.038349248468875885,
      "learning_rate": 0.0012,
      "loss": 2.6650524139404297,
      "step": 5996
    },
    {
      "epoch": 0.4863239135777757,
      "grad_norm": 0.044459447264671326,
      "learning_rate": 0.0012,
      "loss": 2.6529476642608643,
      "step": 5998
    },
    {
      "epoch": 0.4864860756029767,
      "grad_norm": 0.03146527335047722,
      "learning_rate": 0.0012,
      "loss": 2.708113193511963,
      "step": 6000
    },
    {
      "epoch": 0.48664823762817766,
      "grad_norm": 0.02455829083919525,
      "learning_rate": 0.0012,
      "loss": 2.645026206970215,
      "step": 6002
    },
    {
      "epoch": 0.4868103996533787,
      "grad_norm": 0.031781721860170364,
      "learning_rate": 0.0012,
      "loss": 2.671874761581421,
      "step": 6004
    },
    {
      "epoch": 0.48697256167857966,
      "grad_norm": 0.027796445414423943,
      "learning_rate": 0.0012,
      "loss": 2.698092460632324,
      "step": 6006
    },
    {
      "epoch": 0.48713472370378064,
      "grad_norm": 0.03427404910326004,
      "learning_rate": 0.0012,
      "loss": 2.6810426712036133,
      "step": 6008
    },
    {
      "epoch": 0.48729688572898167,
      "grad_norm": 0.023512350395321846,
      "learning_rate": 0.0012,
      "loss": 2.6786386966705322,
      "step": 6010
    },
    {
      "epoch": 0.48745904775418264,
      "grad_norm": 0.028170693665742874,
      "learning_rate": 0.0012,
      "loss": 2.678025960922241,
      "step": 6012
    },
    {
      "epoch": 0.4876212097793836,
      "grad_norm": 0.023106073960661888,
      "learning_rate": 0.0012,
      "loss": 2.6011977195739746,
      "step": 6014
    },
    {
      "epoch": 0.48778337180458464,
      "grad_norm": 0.025407500565052032,
      "learning_rate": 0.0012,
      "loss": 2.6823337078094482,
      "step": 6016
    },
    {
      "epoch": 0.4879455338297856,
      "grad_norm": 0.03086146153509617,
      "learning_rate": 0.0012,
      "loss": 2.711669921875,
      "step": 6018
    },
    {
      "epoch": 0.4881076958549866,
      "grad_norm": 0.022738659754395485,
      "learning_rate": 0.0012,
      "loss": 2.7347543239593506,
      "step": 6020
    },
    {
      "epoch": 0.4882698578801876,
      "grad_norm": 0.024666056036949158,
      "learning_rate": 0.0012,
      "loss": 2.688998222351074,
      "step": 6022
    },
    {
      "epoch": 0.4884320199053886,
      "grad_norm": 0.02965303137898445,
      "learning_rate": 0.0012,
      "loss": 2.64898681640625,
      "step": 6024
    },
    {
      "epoch": 0.48859418193058957,
      "grad_norm": 0.0252519678324461,
      "learning_rate": 0.0012,
      "loss": 2.7092480659484863,
      "step": 6026
    },
    {
      "epoch": 0.4887563439557906,
      "grad_norm": 0.023217303678393364,
      "learning_rate": 0.0012,
      "loss": 2.6940064430236816,
      "step": 6028
    },
    {
      "epoch": 0.48891850598099157,
      "grad_norm": 0.02598014660179615,
      "learning_rate": 0.0012,
      "loss": 2.6875810623168945,
      "step": 6030
    },
    {
      "epoch": 0.48908066800619254,
      "grad_norm": 0.03128565847873688,
      "learning_rate": 0.0012,
      "loss": 2.647637128829956,
      "step": 6032
    },
    {
      "epoch": 0.48924283003139357,
      "grad_norm": 0.0643620565533638,
      "learning_rate": 0.0012,
      "loss": 2.708709716796875,
      "step": 6034
    },
    {
      "epoch": 0.48940499205659455,
      "grad_norm": 0.049066923558712006,
      "learning_rate": 0.0012,
      "loss": 2.664194107055664,
      "step": 6036
    },
    {
      "epoch": 0.4895671540817955,
      "grad_norm": 0.025760110467672348,
      "learning_rate": 0.0012,
      "loss": 2.6228747367858887,
      "step": 6038
    },
    {
      "epoch": 0.48972931610699655,
      "grad_norm": 0.06746158748865128,
      "learning_rate": 0.0012,
      "loss": 2.6632025241851807,
      "step": 6040
    },
    {
      "epoch": 0.4898914781321975,
      "grad_norm": 0.05256195366382599,
      "learning_rate": 0.0012,
      "loss": 2.7019639015197754,
      "step": 6042
    },
    {
      "epoch": 0.4900536401573985,
      "grad_norm": 0.049446720629930496,
      "learning_rate": 0.0012,
      "loss": 2.6999101638793945,
      "step": 6044
    },
    {
      "epoch": 0.4902158021825995,
      "grad_norm": 0.02518298476934433,
      "learning_rate": 0.0012,
      "loss": 2.6663076877593994,
      "step": 6046
    },
    {
      "epoch": 0.4903779642078005,
      "grad_norm": 0.03597588837146759,
      "learning_rate": 0.0012,
      "loss": 2.6641812324523926,
      "step": 6048
    },
    {
      "epoch": 0.49054012623300147,
      "grad_norm": 0.027303606271743774,
      "learning_rate": 0.0012,
      "loss": 2.7390975952148438,
      "step": 6050
    },
    {
      "epoch": 0.4907022882582025,
      "grad_norm": 0.05683049559593201,
      "learning_rate": 0.0012,
      "loss": 2.730191230773926,
      "step": 6052
    },
    {
      "epoch": 0.4908644502834035,
      "grad_norm": 0.03343624249100685,
      "learning_rate": 0.0012,
      "loss": 2.7301836013793945,
      "step": 6054
    },
    {
      "epoch": 0.49102661230860445,
      "grad_norm": 0.03026890754699707,
      "learning_rate": 0.0012,
      "loss": 2.6802730560302734,
      "step": 6056
    },
    {
      "epoch": 0.4911887743338055,
      "grad_norm": 0.02257544919848442,
      "learning_rate": 0.0012,
      "loss": 2.748900890350342,
      "step": 6058
    },
    {
      "epoch": 0.49135093635900645,
      "grad_norm": 0.025826817378401756,
      "learning_rate": 0.0012,
      "loss": 2.681641101837158,
      "step": 6060
    },
    {
      "epoch": 0.4915130983842074,
      "grad_norm": 0.027955664321780205,
      "learning_rate": 0.0012,
      "loss": 2.680473804473877,
      "step": 6062
    },
    {
      "epoch": 0.49167526040940845,
      "grad_norm": 0.02344954013824463,
      "learning_rate": 0.0012,
      "loss": 2.670592784881592,
      "step": 6064
    },
    {
      "epoch": 0.49183742243460943,
      "grad_norm": 0.047516874969005585,
      "learning_rate": 0.0012,
      "loss": 2.641305923461914,
      "step": 6066
    },
    {
      "epoch": 0.4919995844598104,
      "grad_norm": 0.029564907774329185,
      "learning_rate": 0.0012,
      "loss": 2.6052870750427246,
      "step": 6068
    },
    {
      "epoch": 0.49216174648501143,
      "grad_norm": 0.04884330555796623,
      "learning_rate": 0.0012,
      "loss": 2.6415085792541504,
      "step": 6070
    },
    {
      "epoch": 0.4923239085102124,
      "grad_norm": 0.03921743482351303,
      "learning_rate": 0.0012,
      "loss": 2.697779655456543,
      "step": 6072
    },
    {
      "epoch": 0.4924860705354134,
      "grad_norm": 0.0322207435965538,
      "learning_rate": 0.0012,
      "loss": 2.6614809036254883,
      "step": 6074
    },
    {
      "epoch": 0.4926482325606144,
      "grad_norm": 0.030659010633826256,
      "learning_rate": 0.0012,
      "loss": 2.697369337081909,
      "step": 6076
    },
    {
      "epoch": 0.4928103945858154,
      "grad_norm": 0.047842368483543396,
      "learning_rate": 0.0012,
      "loss": 2.694990634918213,
      "step": 6078
    },
    {
      "epoch": 0.49297255661101635,
      "grad_norm": 0.051729779690504074,
      "learning_rate": 0.0012,
      "loss": 2.7229793071746826,
      "step": 6080
    },
    {
      "epoch": 0.4931347186362174,
      "grad_norm": 0.03766797110438347,
      "learning_rate": 0.0012,
      "loss": 2.668210029602051,
      "step": 6082
    },
    {
      "epoch": 0.49329688066141836,
      "grad_norm": 0.03139709308743477,
      "learning_rate": 0.0012,
      "loss": 2.6442699432373047,
      "step": 6084
    },
    {
      "epoch": 0.49345904268661933,
      "grad_norm": 0.02382260002195835,
      "learning_rate": 0.0012,
      "loss": 2.640990734100342,
      "step": 6086
    },
    {
      "epoch": 0.49362120471182036,
      "grad_norm": 0.025537574663758278,
      "learning_rate": 0.0012,
      "loss": 2.732921838760376,
      "step": 6088
    },
    {
      "epoch": 0.49378336673702133,
      "grad_norm": 0.028772681951522827,
      "learning_rate": 0.0012,
      "loss": 2.6409854888916016,
      "step": 6090
    },
    {
      "epoch": 0.4939455287622223,
      "grad_norm": 0.041198913007974625,
      "learning_rate": 0.0012,
      "loss": 2.6840627193450928,
      "step": 6092
    },
    {
      "epoch": 0.49410769078742334,
      "grad_norm": 0.031509559601545334,
      "learning_rate": 0.0012,
      "loss": 2.6681222915649414,
      "step": 6094
    },
    {
      "epoch": 0.4942698528126243,
      "grad_norm": 0.02834041230380535,
      "learning_rate": 0.0012,
      "loss": 2.6061177253723145,
      "step": 6096
    },
    {
      "epoch": 0.4944320148378253,
      "grad_norm": 0.03751141577959061,
      "learning_rate": 0.0012,
      "loss": 2.666855812072754,
      "step": 6098
    },
    {
      "epoch": 0.4945941768630263,
      "grad_norm": 0.03388716280460358,
      "learning_rate": 0.0012,
      "loss": 2.6935715675354004,
      "step": 6100
    },
    {
      "epoch": 0.4947563388882273,
      "grad_norm": 0.024672072380781174,
      "learning_rate": 0.0012,
      "loss": 2.7191085815429688,
      "step": 6102
    },
    {
      "epoch": 0.49491850091342826,
      "grad_norm": 0.02390669845044613,
      "learning_rate": 0.0012,
      "loss": 2.723609447479248,
      "step": 6104
    },
    {
      "epoch": 0.4950806629386293,
      "grad_norm": 0.022579537704586983,
      "learning_rate": 0.0012,
      "loss": 2.715139865875244,
      "step": 6106
    },
    {
      "epoch": 0.49524282496383026,
      "grad_norm": 0.026829702779650688,
      "learning_rate": 0.0012,
      "loss": 2.660804271697998,
      "step": 6108
    },
    {
      "epoch": 0.49540498698903124,
      "grad_norm": 0.027626540511846542,
      "learning_rate": 0.0012,
      "loss": 2.6624460220336914,
      "step": 6110
    },
    {
      "epoch": 0.49556714901423227,
      "grad_norm": 0.03578836843371391,
      "learning_rate": 0.0012,
      "loss": 2.730079174041748,
      "step": 6112
    },
    {
      "epoch": 0.49572931103943324,
      "grad_norm": 0.025920266285538673,
      "learning_rate": 0.0012,
      "loss": 2.6794843673706055,
      "step": 6114
    },
    {
      "epoch": 0.4958914730646342,
      "grad_norm": 0.02396787703037262,
      "learning_rate": 0.0012,
      "loss": 2.7309181690216064,
      "step": 6116
    },
    {
      "epoch": 0.49605363508983524,
      "grad_norm": 0.030222227796912193,
      "learning_rate": 0.0012,
      "loss": 2.6056103706359863,
      "step": 6118
    },
    {
      "epoch": 0.4962157971150362,
      "grad_norm": 0.039993803948163986,
      "learning_rate": 0.0012,
      "loss": 2.6345343589782715,
      "step": 6120
    },
    {
      "epoch": 0.4963779591402372,
      "grad_norm": 0.025621065869927406,
      "learning_rate": 0.0012,
      "loss": 2.6205897331237793,
      "step": 6122
    },
    {
      "epoch": 0.4965401211654382,
      "grad_norm": 0.028645295649766922,
      "learning_rate": 0.0012,
      "loss": 2.6063690185546875,
      "step": 6124
    },
    {
      "epoch": 0.4967022831906392,
      "grad_norm": 0.029201645404100418,
      "learning_rate": 0.0012,
      "loss": 2.710296630859375,
      "step": 6126
    },
    {
      "epoch": 0.49686444521584017,
      "grad_norm": 0.025022905319929123,
      "learning_rate": 0.0012,
      "loss": 2.7110657691955566,
      "step": 6128
    },
    {
      "epoch": 0.4970266072410412,
      "grad_norm": 0.025708569213747978,
      "learning_rate": 0.0012,
      "loss": 2.6787362098693848,
      "step": 6130
    },
    {
      "epoch": 0.49718876926624217,
      "grad_norm": 0.023427218198776245,
      "learning_rate": 0.0012,
      "loss": 2.661881446838379,
      "step": 6132
    },
    {
      "epoch": 0.49735093129144314,
      "grad_norm": 0.023662328720092773,
      "learning_rate": 0.0012,
      "loss": 2.790097713470459,
      "step": 6134
    },
    {
      "epoch": 0.4975130933166442,
      "grad_norm": 0.03595462441444397,
      "learning_rate": 0.0012,
      "loss": 2.7554538249969482,
      "step": 6136
    },
    {
      "epoch": 0.49767525534184515,
      "grad_norm": 0.023304414004087448,
      "learning_rate": 0.0012,
      "loss": 2.6714348793029785,
      "step": 6138
    },
    {
      "epoch": 0.4978374173670461,
      "grad_norm": 0.029635649174451828,
      "learning_rate": 0.0012,
      "loss": 2.7070295810699463,
      "step": 6140
    },
    {
      "epoch": 0.49799957939224715,
      "grad_norm": 0.029072744771838188,
      "learning_rate": 0.0012,
      "loss": 2.615445852279663,
      "step": 6142
    },
    {
      "epoch": 0.4981617414174481,
      "grad_norm": 0.046227242797613144,
      "learning_rate": 0.0012,
      "loss": 2.68272066116333,
      "step": 6144
    },
    {
      "epoch": 0.4983239034426491,
      "grad_norm": 0.047080669552087784,
      "learning_rate": 0.0012,
      "loss": 2.6824049949645996,
      "step": 6146
    },
    {
      "epoch": 0.4984860654678501,
      "grad_norm": 0.060287151485681534,
      "learning_rate": 0.0012,
      "loss": 2.733553886413574,
      "step": 6148
    },
    {
      "epoch": 0.4986482274930511,
      "grad_norm": 0.06356231123209,
      "learning_rate": 0.0012,
      "loss": 2.6920599937438965,
      "step": 6150
    },
    {
      "epoch": 0.4988103895182521,
      "grad_norm": 0.02289752848446369,
      "learning_rate": 0.0012,
      "loss": 2.7169408798217773,
      "step": 6152
    },
    {
      "epoch": 0.4989725515434531,
      "grad_norm": 0.0360487625002861,
      "learning_rate": 0.0012,
      "loss": 2.672801971435547,
      "step": 6154
    },
    {
      "epoch": 0.4991347135686541,
      "grad_norm": 0.04210498556494713,
      "learning_rate": 0.0012,
      "loss": 2.6930861473083496,
      "step": 6156
    },
    {
      "epoch": 0.49929687559385505,
      "grad_norm": 0.06245649233460426,
      "learning_rate": 0.0012,
      "loss": 2.677263021469116,
      "step": 6158
    },
    {
      "epoch": 0.4994590376190561,
      "grad_norm": 0.0506514273583889,
      "learning_rate": 0.0012,
      "loss": 2.655043601989746,
      "step": 6160
    },
    {
      "epoch": 0.49962119964425705,
      "grad_norm": 0.030834568664431572,
      "learning_rate": 0.0012,
      "loss": 2.716580390930176,
      "step": 6162
    },
    {
      "epoch": 0.499783361669458,
      "grad_norm": 0.02501816116273403,
      "learning_rate": 0.0012,
      "loss": 2.6684346199035645,
      "step": 6164
    },
    {
      "epoch": 0.49994552369465906,
      "grad_norm": 0.03315926715731621,
      "learning_rate": 0.0012,
      "loss": 2.6918582916259766,
      "step": 6166
    },
    {
      "epoch": 0.5001076857198601,
      "grad_norm": 0.023402139544487,
      "learning_rate": 0.0012,
      "loss": 2.7085680961608887,
      "step": 6168
    },
    {
      "epoch": 0.5001076857198601,
      "eval_loss": 2.679360866546631,
      "eval_runtime": 1148.1423,
      "eval_samples_per_second": 27.778,
      "eval_steps_per_second": 6.945,
      "step": 6168
    },
    {
      "epoch": 0.500269847745061,
      "grad_norm": 0.028830645605921745,
      "learning_rate": 0.0012,
      "loss": 2.665560007095337,
      "step": 6170
    },
    {
      "epoch": 0.500432009770262,
      "grad_norm": 0.0335165299475193,
      "learning_rate": 0.0012,
      "loss": 2.6592893600463867,
      "step": 6172
    },
    {
      "epoch": 0.5005941717954631,
      "grad_norm": 0.04323822632431984,
      "learning_rate": 0.0012,
      "loss": 2.644383192062378,
      "step": 6174
    },
    {
      "epoch": 0.500756333820664,
      "grad_norm": 0.058639660477638245,
      "learning_rate": 0.0012,
      "loss": 2.7279722690582275,
      "step": 6176
    },
    {
      "epoch": 0.500918495845865,
      "grad_norm": 0.04016340523958206,
      "learning_rate": 0.0012,
      "loss": 2.6707282066345215,
      "step": 6178
    },
    {
      "epoch": 0.501080657871066,
      "grad_norm": 0.023978501558303833,
      "learning_rate": 0.0012,
      "loss": 2.745539665222168,
      "step": 6180
    },
    {
      "epoch": 0.501242819896267,
      "grad_norm": 0.0320555604994297,
      "learning_rate": 0.0012,
      "loss": 2.7328438758850098,
      "step": 6182
    },
    {
      "epoch": 0.501404981921468,
      "grad_norm": 0.02562594786286354,
      "learning_rate": 0.0012,
      "loss": 2.669320583343506,
      "step": 6184
    },
    {
      "epoch": 0.501567143946669,
      "grad_norm": 0.03184087947010994,
      "learning_rate": 0.0012,
      "loss": 2.652068614959717,
      "step": 6186
    },
    {
      "epoch": 0.5017293059718699,
      "grad_norm": 0.03669057786464691,
      "learning_rate": 0.0012,
      "loss": 2.6465959548950195,
      "step": 6188
    },
    {
      "epoch": 0.501891467997071,
      "grad_norm": 0.03280002623796463,
      "learning_rate": 0.0012,
      "loss": 2.6693081855773926,
      "step": 6190
    },
    {
      "epoch": 0.502053630022272,
      "grad_norm": 0.036630451679229736,
      "learning_rate": 0.0012,
      "loss": 2.6992335319519043,
      "step": 6192
    },
    {
      "epoch": 0.5022157920474729,
      "grad_norm": 0.03281830623745918,
      "learning_rate": 0.0012,
      "loss": 2.643293857574463,
      "step": 6194
    },
    {
      "epoch": 0.5023779540726739,
      "grad_norm": 0.02536400966346264,
      "learning_rate": 0.0012,
      "loss": 2.6972804069519043,
      "step": 6196
    },
    {
      "epoch": 0.502540116097875,
      "grad_norm": 0.024268893525004387,
      "learning_rate": 0.0012,
      "loss": 2.6405346393585205,
      "step": 6198
    },
    {
      "epoch": 0.5027022781230759,
      "grad_norm": 0.026247762143611908,
      "learning_rate": 0.0012,
      "loss": 2.673598289489746,
      "step": 6200
    },
    {
      "epoch": 0.5028644401482769,
      "grad_norm": 0.034116633236408234,
      "learning_rate": 0.0012,
      "loss": 2.6469016075134277,
      "step": 6202
    },
    {
      "epoch": 0.5030266021734779,
      "grad_norm": 0.025891099125146866,
      "learning_rate": 0.0012,
      "loss": 2.550724983215332,
      "step": 6204
    },
    {
      "epoch": 0.5031887641986789,
      "grad_norm": 0.03096785582602024,
      "learning_rate": 0.0012,
      "loss": 2.7132673263549805,
      "step": 6206
    },
    {
      "epoch": 0.5033509262238799,
      "grad_norm": 0.03838050737977028,
      "learning_rate": 0.0012,
      "loss": 2.6731653213500977,
      "step": 6208
    },
    {
      "epoch": 0.5035130882490809,
      "grad_norm": 0.04073137417435646,
      "learning_rate": 0.0012,
      "loss": 2.734673500061035,
      "step": 6210
    },
    {
      "epoch": 0.5036752502742818,
      "grad_norm": 0.026230866089463234,
      "learning_rate": 0.0012,
      "loss": 2.7295103073120117,
      "step": 6212
    },
    {
      "epoch": 0.5038374122994829,
      "grad_norm": 0.025852276012301445,
      "learning_rate": 0.0012,
      "loss": 2.6720705032348633,
      "step": 6214
    },
    {
      "epoch": 0.5039995743246839,
      "grad_norm": 0.030499035492539406,
      "learning_rate": 0.0012,
      "loss": 2.640767812728882,
      "step": 6216
    },
    {
      "epoch": 0.5041617363498848,
      "grad_norm": 0.03170297294855118,
      "learning_rate": 0.0012,
      "loss": 2.6736843585968018,
      "step": 6218
    },
    {
      "epoch": 0.5043238983750858,
      "grad_norm": 0.027209291234612465,
      "learning_rate": 0.0012,
      "loss": 2.7383816242218018,
      "step": 6220
    },
    {
      "epoch": 0.5044860604002869,
      "grad_norm": 0.02823229320347309,
      "learning_rate": 0.0012,
      "loss": 2.6210145950317383,
      "step": 6222
    },
    {
      "epoch": 0.5046482224254878,
      "grad_norm": 0.02621215209364891,
      "learning_rate": 0.0012,
      "loss": 2.7078778743743896,
      "step": 6224
    },
    {
      "epoch": 0.5048103844506888,
      "grad_norm": 0.03027057833969593,
      "learning_rate": 0.0012,
      "loss": 2.67930269241333,
      "step": 6226
    },
    {
      "epoch": 0.5049725464758898,
      "grad_norm": 0.02978723868727684,
      "learning_rate": 0.0012,
      "loss": 2.65316104888916,
      "step": 6228
    },
    {
      "epoch": 0.5051347085010908,
      "grad_norm": 0.032530784606933594,
      "learning_rate": 0.0012,
      "loss": 2.613713264465332,
      "step": 6230
    },
    {
      "epoch": 0.5052968705262918,
      "grad_norm": 0.027755647897720337,
      "learning_rate": 0.0012,
      "loss": 2.6933841705322266,
      "step": 6232
    },
    {
      "epoch": 0.5054590325514928,
      "grad_norm": 0.02696635015308857,
      "learning_rate": 0.0012,
      "loss": 2.6970160007476807,
      "step": 6234
    },
    {
      "epoch": 0.5056211945766937,
      "grad_norm": 0.036420512944459915,
      "learning_rate": 0.0012,
      "loss": 2.602081060409546,
      "step": 6236
    },
    {
      "epoch": 0.5057833566018948,
      "grad_norm": 0.034890200942754745,
      "learning_rate": 0.0012,
      "loss": 2.6921186447143555,
      "step": 6238
    },
    {
      "epoch": 0.5059455186270958,
      "grad_norm": 0.024809502065181732,
      "learning_rate": 0.0012,
      "loss": 2.644493341445923,
      "step": 6240
    },
    {
      "epoch": 0.5061076806522967,
      "grad_norm": 0.02447095885872841,
      "learning_rate": 0.0012,
      "loss": 2.6298325061798096,
      "step": 6242
    },
    {
      "epoch": 0.5062698426774978,
      "grad_norm": 0.026572786271572113,
      "learning_rate": 0.0012,
      "loss": 2.6454949378967285,
      "step": 6244
    },
    {
      "epoch": 0.5064320047026988,
      "grad_norm": 0.02960502915084362,
      "learning_rate": 0.0012,
      "loss": 2.7013401985168457,
      "step": 6246
    },
    {
      "epoch": 0.5065941667278997,
      "grad_norm": 0.030792195349931717,
      "learning_rate": 0.0012,
      "loss": 2.664243698120117,
      "step": 6248
    },
    {
      "epoch": 0.5067563287531007,
      "grad_norm": 0.048746779561042786,
      "learning_rate": 0.0012,
      "loss": 2.6015548706054688,
      "step": 6250
    },
    {
      "epoch": 0.5069184907783018,
      "grad_norm": 0.0543992705643177,
      "learning_rate": 0.0012,
      "loss": 2.672394275665283,
      "step": 6252
    },
    {
      "epoch": 0.5070806528035027,
      "grad_norm": 0.031485140323638916,
      "learning_rate": 0.0012,
      "loss": 2.6809558868408203,
      "step": 6254
    },
    {
      "epoch": 0.5072428148287037,
      "grad_norm": 0.042567178606987,
      "learning_rate": 0.0012,
      "loss": 2.6668295860290527,
      "step": 6256
    },
    {
      "epoch": 0.5074049768539047,
      "grad_norm": 0.038864366710186005,
      "learning_rate": 0.0012,
      "loss": 2.6641461849212646,
      "step": 6258
    },
    {
      "epoch": 0.5075671388791057,
      "grad_norm": 0.02627614140510559,
      "learning_rate": 0.0012,
      "loss": 2.647794723510742,
      "step": 6260
    },
    {
      "epoch": 0.5077293009043067,
      "grad_norm": 0.04943471401929855,
      "learning_rate": 0.0012,
      "loss": 2.7079310417175293,
      "step": 6262
    },
    {
      "epoch": 0.5078914629295077,
      "grad_norm": 0.034324049949645996,
      "learning_rate": 0.0012,
      "loss": 2.7034337520599365,
      "step": 6264
    },
    {
      "epoch": 0.5080536249547086,
      "grad_norm": 0.025621144101023674,
      "learning_rate": 0.0012,
      "loss": 2.710996389389038,
      "step": 6266
    },
    {
      "epoch": 0.5082157869799097,
      "grad_norm": 0.03060862608253956,
      "learning_rate": 0.0012,
      "loss": 2.6636762619018555,
      "step": 6268
    },
    {
      "epoch": 0.5083779490051107,
      "grad_norm": 0.0250102449208498,
      "learning_rate": 0.0012,
      "loss": 2.652021884918213,
      "step": 6270
    },
    {
      "epoch": 0.5085401110303116,
      "grad_norm": 0.022338591516017914,
      "learning_rate": 0.0012,
      "loss": 2.7217514514923096,
      "step": 6272
    },
    {
      "epoch": 0.5087022730555126,
      "grad_norm": 0.049872979521751404,
      "learning_rate": 0.0012,
      "loss": 2.645437240600586,
      "step": 6274
    },
    {
      "epoch": 0.5088644350807137,
      "grad_norm": 0.06748571246862411,
      "learning_rate": 0.0012,
      "loss": 2.735492706298828,
      "step": 6276
    },
    {
      "epoch": 0.5090265971059146,
      "grad_norm": 0.06056679040193558,
      "learning_rate": 0.0012,
      "loss": 2.676872730255127,
      "step": 6278
    },
    {
      "epoch": 0.5091887591311156,
      "grad_norm": 0.04095802456140518,
      "learning_rate": 0.0012,
      "loss": 2.6890525817871094,
      "step": 6280
    },
    {
      "epoch": 0.5093509211563166,
      "grad_norm": 0.03925388678908348,
      "learning_rate": 0.0012,
      "loss": 2.702153205871582,
      "step": 6282
    },
    {
      "epoch": 0.5095130831815176,
      "grad_norm": 0.04758167639374733,
      "learning_rate": 0.0012,
      "loss": 2.6536598205566406,
      "step": 6284
    },
    {
      "epoch": 0.5096752452067186,
      "grad_norm": 0.07110653817653656,
      "learning_rate": 0.0012,
      "loss": 2.6437509059906006,
      "step": 6286
    },
    {
      "epoch": 0.5098374072319196,
      "grad_norm": 0.02730695530772209,
      "learning_rate": 0.0012,
      "loss": 2.684358596801758,
      "step": 6288
    },
    {
      "epoch": 0.5099995692571205,
      "grad_norm": 0.03446701169013977,
      "learning_rate": 0.0012,
      "loss": 2.6124067306518555,
      "step": 6290
    },
    {
      "epoch": 0.5101617312823216,
      "grad_norm": 0.02625180035829544,
      "learning_rate": 0.0012,
      "loss": 2.6626477241516113,
      "step": 6292
    },
    {
      "epoch": 0.5103238933075226,
      "grad_norm": 0.026305941864848137,
      "learning_rate": 0.0012,
      "loss": 2.7111101150512695,
      "step": 6294
    },
    {
      "epoch": 0.5104860553327235,
      "grad_norm": 0.05625757575035095,
      "learning_rate": 0.0012,
      "loss": 2.682149648666382,
      "step": 6296
    },
    {
      "epoch": 0.5106482173579245,
      "grad_norm": 0.035012297332286835,
      "learning_rate": 0.0012,
      "loss": 2.6808953285217285,
      "step": 6298
    },
    {
      "epoch": 0.5108103793831256,
      "grad_norm": 0.026915762573480606,
      "learning_rate": 0.0012,
      "loss": 2.6579480171203613,
      "step": 6300
    },
    {
      "epoch": 0.5109725414083265,
      "grad_norm": 0.035912640392780304,
      "learning_rate": 0.0012,
      "loss": 2.731208562850952,
      "step": 6302
    },
    {
      "epoch": 0.5111347034335275,
      "grad_norm": 0.03544101491570473,
      "learning_rate": 0.0012,
      "loss": 2.658627986907959,
      "step": 6304
    },
    {
      "epoch": 0.5112968654587285,
      "grad_norm": 0.044915955513715744,
      "learning_rate": 0.0012,
      "loss": 2.739630699157715,
      "step": 6306
    },
    {
      "epoch": 0.5114590274839295,
      "grad_norm": 0.03974173963069916,
      "learning_rate": 0.0012,
      "loss": 2.710655927658081,
      "step": 6308
    },
    {
      "epoch": 0.5116211895091305,
      "grad_norm": 0.025581002235412598,
      "learning_rate": 0.0012,
      "loss": 2.6344571113586426,
      "step": 6310
    },
    {
      "epoch": 0.5117833515343315,
      "grad_norm": 0.02764456905424595,
      "learning_rate": 0.0012,
      "loss": 2.693081855773926,
      "step": 6312
    },
    {
      "epoch": 0.5119455135595324,
      "grad_norm": 0.032293204218149185,
      "learning_rate": 0.0012,
      "loss": 2.7279272079467773,
      "step": 6314
    },
    {
      "epoch": 0.5121076755847335,
      "grad_norm": 0.024946654215455055,
      "learning_rate": 0.0012,
      "loss": 2.758237838745117,
      "step": 6316
    },
    {
      "epoch": 0.5122698376099345,
      "grad_norm": 0.028068093582987785,
      "learning_rate": 0.0012,
      "loss": 2.607734203338623,
      "step": 6318
    },
    {
      "epoch": 0.5124319996351354,
      "grad_norm": 0.026754116639494896,
      "learning_rate": 0.0012,
      "loss": 2.7021799087524414,
      "step": 6320
    },
    {
      "epoch": 0.5125941616603364,
      "grad_norm": 0.02977832220494747,
      "learning_rate": 0.0012,
      "loss": 2.71040678024292,
      "step": 6322
    },
    {
      "epoch": 0.5127563236855375,
      "grad_norm": 0.03177084028720856,
      "learning_rate": 0.0012,
      "loss": 2.715879440307617,
      "step": 6324
    },
    {
      "epoch": 0.5129184857107384,
      "grad_norm": 0.030616145581007004,
      "learning_rate": 0.0012,
      "loss": 2.6905980110168457,
      "step": 6326
    },
    {
      "epoch": 0.5130806477359394,
      "grad_norm": 0.046938568353652954,
      "learning_rate": 0.0012,
      "loss": 2.683260917663574,
      "step": 6328
    },
    {
      "epoch": 0.5132428097611405,
      "grad_norm": 0.03395875170826912,
      "learning_rate": 0.0012,
      "loss": 2.6931419372558594,
      "step": 6330
    },
    {
      "epoch": 0.5134049717863414,
      "grad_norm": 0.024785427376627922,
      "learning_rate": 0.0012,
      "loss": 2.6921639442443848,
      "step": 6332
    },
    {
      "epoch": 0.5135671338115424,
      "grad_norm": 0.029886187985539436,
      "learning_rate": 0.0012,
      "loss": 2.759793281555176,
      "step": 6334
    },
    {
      "epoch": 0.5137292958367434,
      "grad_norm": 0.02518186718225479,
      "learning_rate": 0.0012,
      "loss": 2.669214963912964,
      "step": 6336
    },
    {
      "epoch": 0.5138914578619443,
      "grad_norm": 0.026663057506084442,
      "learning_rate": 0.0012,
      "loss": 2.6637775897979736,
      "step": 6338
    },
    {
      "epoch": 0.5140536198871454,
      "grad_norm": 0.05204489454627037,
      "learning_rate": 0.0012,
      "loss": 2.699587345123291,
      "step": 6340
    },
    {
      "epoch": 0.5142157819123464,
      "grad_norm": 0.030575107783079147,
      "learning_rate": 0.0012,
      "loss": 2.675292491912842,
      "step": 6342
    },
    {
      "epoch": 0.5143779439375473,
      "grad_norm": 0.023289784789085388,
      "learning_rate": 0.0012,
      "loss": 2.6641221046447754,
      "step": 6344
    },
    {
      "epoch": 0.5145401059627484,
      "grad_norm": 0.06270144134759903,
      "learning_rate": 0.0012,
      "loss": 2.7271363735198975,
      "step": 6346
    },
    {
      "epoch": 0.5147022679879494,
      "grad_norm": 0.0333215594291687,
      "learning_rate": 0.0012,
      "loss": 2.6701974868774414,
      "step": 6348
    },
    {
      "epoch": 0.5148644300131503,
      "grad_norm": 0.02347845770418644,
      "learning_rate": 0.0012,
      "loss": 2.728470802307129,
      "step": 6350
    },
    {
      "epoch": 0.5150265920383513,
      "grad_norm": 0.04488786309957504,
      "learning_rate": 0.0012,
      "loss": 2.638420581817627,
      "step": 6352
    },
    {
      "epoch": 0.5151887540635524,
      "grad_norm": 0.03931264951825142,
      "learning_rate": 0.0012,
      "loss": 2.703871250152588,
      "step": 6354
    },
    {
      "epoch": 0.5153509160887533,
      "grad_norm": 0.07253532856702805,
      "learning_rate": 0.0012,
      "loss": 2.6691946983337402,
      "step": 6356
    },
    {
      "epoch": 0.5155130781139543,
      "grad_norm": 0.035515010356903076,
      "learning_rate": 0.0012,
      "loss": 2.6805202960968018,
      "step": 6358
    },
    {
      "epoch": 0.5156752401391553,
      "grad_norm": 0.030542753636837006,
      "learning_rate": 0.0012,
      "loss": 2.7143428325653076,
      "step": 6360
    },
    {
      "epoch": 0.5158374021643563,
      "grad_norm": 0.06120137870311737,
      "learning_rate": 0.0012,
      "loss": 2.6817550659179688,
      "step": 6362
    },
    {
      "epoch": 0.5159995641895573,
      "grad_norm": 0.027137406170368195,
      "learning_rate": 0.0012,
      "loss": 2.6771175861358643,
      "step": 6364
    },
    {
      "epoch": 0.5161617262147583,
      "grad_norm": 0.07061584293842316,
      "learning_rate": 0.0012,
      "loss": 2.6681201457977295,
      "step": 6366
    },
    {
      "epoch": 0.5163238882399592,
      "grad_norm": 0.052292194217443466,
      "learning_rate": 0.0012,
      "loss": 2.6683340072631836,
      "step": 6368
    },
    {
      "epoch": 0.5164860502651603,
      "grad_norm": 0.02591688185930252,
      "learning_rate": 0.0012,
      "loss": 2.7140910625457764,
      "step": 6370
    },
    {
      "epoch": 0.5166482122903613,
      "grad_norm": 0.03319443762302399,
      "learning_rate": 0.0012,
      "loss": 2.700399160385132,
      "step": 6372
    },
    {
      "epoch": 0.5168103743155622,
      "grad_norm": 0.035803016275167465,
      "learning_rate": 0.0012,
      "loss": 2.7139625549316406,
      "step": 6374
    },
    {
      "epoch": 0.5169725363407632,
      "grad_norm": 0.02621559612452984,
      "learning_rate": 0.0012,
      "loss": 2.691054344177246,
      "step": 6376
    },
    {
      "epoch": 0.5171346983659643,
      "grad_norm": 0.02877230942249298,
      "learning_rate": 0.0012,
      "loss": 2.6278059482574463,
      "step": 6378
    },
    {
      "epoch": 0.5172968603911652,
      "grad_norm": 0.024591414257884026,
      "learning_rate": 0.0012,
      "loss": 2.633239269256592,
      "step": 6380
    },
    {
      "epoch": 0.5174590224163662,
      "grad_norm": 0.027224641293287277,
      "learning_rate": 0.0012,
      "loss": 2.7243199348449707,
      "step": 6382
    },
    {
      "epoch": 0.5176211844415672,
      "grad_norm": 0.024947937577962875,
      "learning_rate": 0.0012,
      "loss": 2.6855971813201904,
      "step": 6384
    },
    {
      "epoch": 0.5177833464667682,
      "grad_norm": 0.026515290141105652,
      "learning_rate": 0.0012,
      "loss": 2.645705223083496,
      "step": 6386
    },
    {
      "epoch": 0.5179455084919692,
      "grad_norm": 0.03269050270318985,
      "learning_rate": 0.0012,
      "loss": 2.6747994422912598,
      "step": 6388
    },
    {
      "epoch": 0.5181076705171702,
      "grad_norm": 0.03528539463877678,
      "learning_rate": 0.0012,
      "loss": 2.6573984622955322,
      "step": 6390
    },
    {
      "epoch": 0.5182698325423711,
      "grad_norm": 0.06017623096704483,
      "learning_rate": 0.0012,
      "loss": 2.646686553955078,
      "step": 6392
    },
    {
      "epoch": 0.5184319945675722,
      "grad_norm": 0.06045915186405182,
      "learning_rate": 0.0012,
      "loss": 2.680044412612915,
      "step": 6394
    },
    {
      "epoch": 0.5185941565927732,
      "grad_norm": 0.03145146369934082,
      "learning_rate": 0.0012,
      "loss": 2.7478737831115723,
      "step": 6396
    },
    {
      "epoch": 0.5187563186179741,
      "grad_norm": 0.03559846058487892,
      "learning_rate": 0.0012,
      "loss": 2.663881301879883,
      "step": 6398
    },
    {
      "epoch": 0.5189184806431751,
      "grad_norm": 0.046408772468566895,
      "learning_rate": 0.0012,
      "loss": 2.6903226375579834,
      "step": 6400
    },
    {
      "epoch": 0.5190806426683762,
      "grad_norm": 0.03140931576490402,
      "learning_rate": 0.0012,
      "loss": 2.7023158073425293,
      "step": 6402
    },
    {
      "epoch": 0.5192428046935771,
      "grad_norm": 0.03374788910150528,
      "learning_rate": 0.0012,
      "loss": 2.619065999984741,
      "step": 6404
    },
    {
      "epoch": 0.5194049667187781,
      "grad_norm": 0.028461840003728867,
      "learning_rate": 0.0012,
      "loss": 2.655031681060791,
      "step": 6406
    },
    {
      "epoch": 0.5195671287439791,
      "grad_norm": 0.0228200051933527,
      "learning_rate": 0.0012,
      "loss": 2.667118549346924,
      "step": 6408
    },
    {
      "epoch": 0.5197292907691801,
      "grad_norm": 0.025269832462072372,
      "learning_rate": 0.0012,
      "loss": 2.5838756561279297,
      "step": 6410
    },
    {
      "epoch": 0.5198914527943811,
      "grad_norm": 0.024373164400458336,
      "learning_rate": 0.0012,
      "loss": 2.638892889022827,
      "step": 6412
    },
    {
      "epoch": 0.5200536148195821,
      "grad_norm": 0.03681135177612305,
      "learning_rate": 0.0012,
      "loss": 2.68037748336792,
      "step": 6414
    },
    {
      "epoch": 0.520215776844783,
      "grad_norm": 0.04033764824271202,
      "learning_rate": 0.0012,
      "loss": 2.6752593517303467,
      "step": 6416
    },
    {
      "epoch": 0.5203779388699841,
      "grad_norm": 0.03563384711742401,
      "learning_rate": 0.0012,
      "loss": 2.6812186241149902,
      "step": 6418
    },
    {
      "epoch": 0.5205401008951851,
      "grad_norm": 0.032375358045101166,
      "learning_rate": 0.0012,
      "loss": 2.7048721313476562,
      "step": 6420
    },
    {
      "epoch": 0.520702262920386,
      "grad_norm": 0.03820287063717842,
      "learning_rate": 0.0012,
      "loss": 2.6927475929260254,
      "step": 6422
    },
    {
      "epoch": 0.520864424945587,
      "grad_norm": 0.03493191674351692,
      "learning_rate": 0.0012,
      "loss": 2.6979990005493164,
      "step": 6424
    },
    {
      "epoch": 0.5210265869707881,
      "grad_norm": 0.02511579357087612,
      "learning_rate": 0.0012,
      "loss": 2.72802996635437,
      "step": 6426
    },
    {
      "epoch": 0.521188748995989,
      "grad_norm": 0.025242606177926064,
      "learning_rate": 0.0012,
      "loss": 2.683788299560547,
      "step": 6428
    },
    {
      "epoch": 0.52135091102119,
      "grad_norm": 0.04259972274303436,
      "learning_rate": 0.0012,
      "loss": 2.6289753913879395,
      "step": 6430
    },
    {
      "epoch": 0.521513073046391,
      "grad_norm": 0.02373078092932701,
      "learning_rate": 0.0012,
      "loss": 2.720810651779175,
      "step": 6432
    },
    {
      "epoch": 0.521675235071592,
      "grad_norm": 0.02208353392779827,
      "learning_rate": 0.0012,
      "loss": 2.66469144821167,
      "step": 6434
    },
    {
      "epoch": 0.521837397096793,
      "grad_norm": 0.03071313165128231,
      "learning_rate": 0.0012,
      "loss": 2.6002769470214844,
      "step": 6436
    },
    {
      "epoch": 0.521999559121994,
      "grad_norm": 0.031330838799476624,
      "learning_rate": 0.0012,
      "loss": 2.6847214698791504,
      "step": 6438
    },
    {
      "epoch": 0.522161721147195,
      "grad_norm": 0.03760603815317154,
      "learning_rate": 0.0012,
      "loss": 2.65450382232666,
      "step": 6440
    },
    {
      "epoch": 0.522323883172396,
      "grad_norm": 0.044229455292224884,
      "learning_rate": 0.0012,
      "loss": 2.6553187370300293,
      "step": 6442
    },
    {
      "epoch": 0.522486045197597,
      "grad_norm": 0.045238394290208817,
      "learning_rate": 0.0012,
      "loss": 2.6932871341705322,
      "step": 6444
    },
    {
      "epoch": 0.5226482072227979,
      "grad_norm": 0.036355286836624146,
      "learning_rate": 0.0012,
      "loss": 2.5951380729675293,
      "step": 6446
    },
    {
      "epoch": 0.522810369247999,
      "grad_norm": 0.035010263323783875,
      "learning_rate": 0.0012,
      "loss": 2.701305627822876,
      "step": 6448
    },
    {
      "epoch": 0.5229725312732,
      "grad_norm": 0.053713828325271606,
      "learning_rate": 0.0012,
      "loss": 2.731527805328369,
      "step": 6450
    },
    {
      "epoch": 0.5231346932984009,
      "grad_norm": 0.054887767881155014,
      "learning_rate": 0.0012,
      "loss": 2.677133083343506,
      "step": 6452
    },
    {
      "epoch": 0.5232968553236019,
      "grad_norm": 0.028446607291698456,
      "learning_rate": 0.0012,
      "loss": 2.718367099761963,
      "step": 6454
    },
    {
      "epoch": 0.523459017348803,
      "grad_norm": 0.030353480949997902,
      "learning_rate": 0.0012,
      "loss": 2.6715734004974365,
      "step": 6456
    },
    {
      "epoch": 0.5236211793740039,
      "grad_norm": 0.043931744992733,
      "learning_rate": 0.0012,
      "loss": 2.6327147483825684,
      "step": 6458
    },
    {
      "epoch": 0.5237833413992049,
      "grad_norm": 0.025933295488357544,
      "learning_rate": 0.0012,
      "loss": 2.7211670875549316,
      "step": 6460
    },
    {
      "epoch": 0.5239455034244059,
      "grad_norm": 0.04234271124005318,
      "learning_rate": 0.0012,
      "loss": 2.6406760215759277,
      "step": 6462
    },
    {
      "epoch": 0.5241076654496069,
      "grad_norm": 0.033267952501773834,
      "learning_rate": 0.0012,
      "loss": 2.652498245239258,
      "step": 6464
    },
    {
      "epoch": 0.5242698274748079,
      "grad_norm": 0.03222408518195152,
      "learning_rate": 0.0012,
      "loss": 2.7018401622772217,
      "step": 6466
    },
    {
      "epoch": 0.5244319895000089,
      "grad_norm": 0.023161552846431732,
      "learning_rate": 0.0012,
      "loss": 2.6772806644439697,
      "step": 6468
    },
    {
      "epoch": 0.5245941515252098,
      "grad_norm": 0.042098719626665115,
      "learning_rate": 0.0012,
      "loss": 2.6685094833374023,
      "step": 6470
    },
    {
      "epoch": 0.5247563135504109,
      "grad_norm": 0.03632482886314392,
      "learning_rate": 0.0012,
      "loss": 2.6711645126342773,
      "step": 6472
    },
    {
      "epoch": 0.5249184755756119,
      "grad_norm": 0.05496395379304886,
      "learning_rate": 0.0012,
      "loss": 2.7051355838775635,
      "step": 6474
    },
    {
      "epoch": 0.5250806376008128,
      "grad_norm": 0.02303532138466835,
      "learning_rate": 0.0012,
      "loss": 2.7121071815490723,
      "step": 6476
    },
    {
      "epoch": 0.5252427996260138,
      "grad_norm": 0.037338122725486755,
      "learning_rate": 0.0012,
      "loss": 2.707040786743164,
      "step": 6478
    },
    {
      "epoch": 0.5254049616512149,
      "grad_norm": 0.04403949901461601,
      "learning_rate": 0.0012,
      "loss": 2.615412712097168,
      "step": 6480
    },
    {
      "epoch": 0.5255671236764158,
      "grad_norm": 0.027736200019717216,
      "learning_rate": 0.0012,
      "loss": 2.697514057159424,
      "step": 6482
    },
    {
      "epoch": 0.5257292857016168,
      "grad_norm": 0.030689364299178123,
      "learning_rate": 0.0012,
      "loss": 2.75594425201416,
      "step": 6484
    },
    {
      "epoch": 0.5258914477268178,
      "grad_norm": 0.03657183423638344,
      "learning_rate": 0.0012,
      "loss": 2.6930031776428223,
      "step": 6486
    },
    {
      "epoch": 0.5260536097520188,
      "grad_norm": 0.027363870292901993,
      "learning_rate": 0.0012,
      "loss": 2.7172417640686035,
      "step": 6488
    },
    {
      "epoch": 0.5262157717772198,
      "grad_norm": 0.028079699724912643,
      "learning_rate": 0.0012,
      "loss": 2.701550006866455,
      "step": 6490
    },
    {
      "epoch": 0.5263779338024208,
      "grad_norm": 0.030828874558210373,
      "learning_rate": 0.0012,
      "loss": 2.6729376316070557,
      "step": 6492
    },
    {
      "epoch": 0.5265400958276217,
      "grad_norm": 0.031320009380578995,
      "learning_rate": 0.0012,
      "loss": 2.712378978729248,
      "step": 6494
    },
    {
      "epoch": 0.5267022578528228,
      "grad_norm": 0.027659857645630836,
      "learning_rate": 0.0012,
      "loss": 2.6788220405578613,
      "step": 6496
    },
    {
      "epoch": 0.5268644198780238,
      "grad_norm": 0.02677592821419239,
      "learning_rate": 0.0012,
      "loss": 2.6522865295410156,
      "step": 6498
    },
    {
      "epoch": 0.5270265819032247,
      "grad_norm": 0.03222321718931198,
      "learning_rate": 0.0012,
      "loss": 2.705526828765869,
      "step": 6500
    },
    {
      "epoch": 0.5271887439284257,
      "grad_norm": 0.023547803983092308,
      "learning_rate": 0.0012,
      "loss": 2.617645263671875,
      "step": 6502
    },
    {
      "epoch": 0.5273509059536268,
      "grad_norm": 0.027882764115929604,
      "learning_rate": 0.0012,
      "loss": 2.65853214263916,
      "step": 6504
    },
    {
      "epoch": 0.5275130679788277,
      "grad_norm": 0.03077998198568821,
      "learning_rate": 0.0012,
      "loss": 2.7207770347595215,
      "step": 6506
    },
    {
      "epoch": 0.5276752300040287,
      "grad_norm": 0.030225267633795738,
      "learning_rate": 0.0012,
      "loss": 2.7579612731933594,
      "step": 6508
    },
    {
      "epoch": 0.5278373920292297,
      "grad_norm": 0.03749452531337738,
      "learning_rate": 0.0012,
      "loss": 2.637831211090088,
      "step": 6510
    },
    {
      "epoch": 0.5279995540544307,
      "grad_norm": 0.03546426445245743,
      "learning_rate": 0.0012,
      "loss": 2.6585817337036133,
      "step": 6512
    },
    {
      "epoch": 0.5281617160796317,
      "grad_norm": 0.04172097146511078,
      "learning_rate": 0.0012,
      "loss": 2.7414748668670654,
      "step": 6514
    },
    {
      "epoch": 0.5283238781048327,
      "grad_norm": 0.0754442811012268,
      "learning_rate": 0.0012,
      "loss": 2.6963534355163574,
      "step": 6516
    },
    {
      "epoch": 0.5284860401300336,
      "grad_norm": 0.06908786296844482,
      "learning_rate": 0.0012,
      "loss": 2.7036821842193604,
      "step": 6518
    },
    {
      "epoch": 0.5286482021552347,
      "grad_norm": 0.022437147796154022,
      "learning_rate": 0.0012,
      "loss": 2.6793198585510254,
      "step": 6520
    },
    {
      "epoch": 0.5288103641804357,
      "grad_norm": 0.05043584108352661,
      "learning_rate": 0.0012,
      "loss": 2.663694143295288,
      "step": 6522
    },
    {
      "epoch": 0.5289725262056366,
      "grad_norm": 0.05804309621453285,
      "learning_rate": 0.0012,
      "loss": 2.665865421295166,
      "step": 6524
    },
    {
      "epoch": 0.5291346882308376,
      "grad_norm": 0.026971381157636642,
      "learning_rate": 0.0012,
      "loss": 2.7207345962524414,
      "step": 6526
    },
    {
      "epoch": 0.5292968502560387,
      "grad_norm": 0.029066814109683037,
      "learning_rate": 0.0012,
      "loss": 2.727616786956787,
      "step": 6528
    },
    {
      "epoch": 0.5294590122812396,
      "grad_norm": 0.035985711961984634,
      "learning_rate": 0.0012,
      "loss": 2.664004325866699,
      "step": 6530
    },
    {
      "epoch": 0.5296211743064406,
      "grad_norm": 0.031937915831804276,
      "learning_rate": 0.0012,
      "loss": 2.7084405422210693,
      "step": 6532
    },
    {
      "epoch": 0.5297833363316417,
      "grad_norm": 0.02610844001173973,
      "learning_rate": 0.0012,
      "loss": 2.6809425354003906,
      "step": 6534
    },
    {
      "epoch": 0.5299454983568426,
      "grad_norm": 0.023270340636372566,
      "learning_rate": 0.0012,
      "loss": 2.6684470176696777,
      "step": 6536
    },
    {
      "epoch": 0.5301076603820436,
      "grad_norm": 0.023628223687410355,
      "learning_rate": 0.0012,
      "loss": 2.7664237022399902,
      "step": 6538
    },
    {
      "epoch": 0.5302698224072446,
      "grad_norm": 0.026790114119648933,
      "learning_rate": 0.0012,
      "loss": 2.750603437423706,
      "step": 6540
    },
    {
      "epoch": 0.5304319844324455,
      "grad_norm": 0.026052897796034813,
      "learning_rate": 0.0012,
      "loss": 2.6613030433654785,
      "step": 6542
    },
    {
      "epoch": 0.5305941464576466,
      "grad_norm": 0.02519894950091839,
      "learning_rate": 0.0012,
      "loss": 2.671581745147705,
      "step": 6544
    },
    {
      "epoch": 0.5307563084828476,
      "grad_norm": 0.026686394587159157,
      "learning_rate": 0.0012,
      "loss": 2.6713640689849854,
      "step": 6546
    },
    {
      "epoch": 0.5309184705080485,
      "grad_norm": 0.04174739494919777,
      "learning_rate": 0.0012,
      "loss": 2.675323486328125,
      "step": 6548
    },
    {
      "epoch": 0.5310806325332496,
      "grad_norm": 0.05928557366132736,
      "learning_rate": 0.0012,
      "loss": 2.727539539337158,
      "step": 6550
    },
    {
      "epoch": 0.5312427945584506,
      "grad_norm": 0.07249127328395844,
      "learning_rate": 0.0012,
      "loss": 2.7163376808166504,
      "step": 6552
    },
    {
      "epoch": 0.5314049565836515,
      "grad_norm": 0.03195861726999283,
      "learning_rate": 0.0012,
      "loss": 2.716005325317383,
      "step": 6554
    },
    {
      "epoch": 0.5315671186088525,
      "grad_norm": 0.025067083537578583,
      "learning_rate": 0.0012,
      "loss": 2.6165075302124023,
      "step": 6556
    },
    {
      "epoch": 0.5317292806340536,
      "grad_norm": 0.031217684969305992,
      "learning_rate": 0.0012,
      "loss": 2.6965370178222656,
      "step": 6558
    },
    {
      "epoch": 0.5318914426592545,
      "grad_norm": 0.04564634710550308,
      "learning_rate": 0.0012,
      "loss": 2.7097179889678955,
      "step": 6560
    },
    {
      "epoch": 0.5320536046844555,
      "grad_norm": 0.048129186034202576,
      "learning_rate": 0.0012,
      "loss": 2.6989471912384033,
      "step": 6562
    },
    {
      "epoch": 0.5322157667096565,
      "grad_norm": 0.025948932394385338,
      "learning_rate": 0.0012,
      "loss": 2.7177999019622803,
      "step": 6564
    },
    {
      "epoch": 0.5323779287348575,
      "grad_norm": 0.054429929703474045,
      "learning_rate": 0.0012,
      "loss": 2.666252613067627,
      "step": 6566
    },
    {
      "epoch": 0.5325400907600585,
      "grad_norm": 0.0300915464758873,
      "learning_rate": 0.0012,
      "loss": 2.6902599334716797,
      "step": 6568
    },
    {
      "epoch": 0.5327022527852595,
      "grad_norm": 0.02570764347910881,
      "learning_rate": 0.0012,
      "loss": 2.6404905319213867,
      "step": 6570
    },
    {
      "epoch": 0.5328644148104604,
      "grad_norm": 0.02564556524157524,
      "learning_rate": 0.0012,
      "loss": 2.6571359634399414,
      "step": 6572
    },
    {
      "epoch": 0.5330265768356615,
      "grad_norm": 0.028720976784825325,
      "learning_rate": 0.0012,
      "loss": 2.617806911468506,
      "step": 6574
    },
    {
      "epoch": 0.5331887388608625,
      "grad_norm": 0.03342447057366371,
      "learning_rate": 0.0012,
      "loss": 2.6563873291015625,
      "step": 6576
    },
    {
      "epoch": 0.5333509008860634,
      "grad_norm": 0.02755781076848507,
      "learning_rate": 0.0012,
      "loss": 2.5783379077911377,
      "step": 6578
    },
    {
      "epoch": 0.5335130629112644,
      "grad_norm": 0.04113197699189186,
      "learning_rate": 0.0012,
      "loss": 2.6471123695373535,
      "step": 6580
    },
    {
      "epoch": 0.5336752249364655,
      "grad_norm": 0.036233823746442795,
      "learning_rate": 0.0012,
      "loss": 2.7160861492156982,
      "step": 6582
    },
    {
      "epoch": 0.5338373869616664,
      "grad_norm": 0.034589942544698715,
      "learning_rate": 0.0012,
      "loss": 2.5786080360412598,
      "step": 6584
    },
    {
      "epoch": 0.5339995489868674,
      "grad_norm": 0.02460213378071785,
      "learning_rate": 0.0012,
      "loss": 2.6382646560668945,
      "step": 6586
    },
    {
      "epoch": 0.5341617110120684,
      "grad_norm": 0.02452712319791317,
      "learning_rate": 0.0012,
      "loss": 2.7011802196502686,
      "step": 6588
    },
    {
      "epoch": 0.5343238730372694,
      "grad_norm": 0.03312335163354874,
      "learning_rate": 0.0012,
      "loss": 2.6738595962524414,
      "step": 6590
    },
    {
      "epoch": 0.5344860350624704,
      "grad_norm": 0.04001075029373169,
      "learning_rate": 0.0012,
      "loss": 2.631290912628174,
      "step": 6592
    },
    {
      "epoch": 0.5346481970876714,
      "grad_norm": 0.02692515403032303,
      "learning_rate": 0.0012,
      "loss": 2.645805597305298,
      "step": 6594
    },
    {
      "epoch": 0.5348103591128723,
      "grad_norm": 0.045811235904693604,
      "learning_rate": 0.0012,
      "loss": 2.6728014945983887,
      "step": 6596
    },
    {
      "epoch": 0.5349725211380734,
      "grad_norm": 0.032172802835702896,
      "learning_rate": 0.0012,
      "loss": 2.71293306350708,
      "step": 6598
    },
    {
      "epoch": 0.5351346831632744,
      "grad_norm": 0.04373571649193764,
      "learning_rate": 0.0012,
      "loss": 2.6507012844085693,
      "step": 6600
    },
    {
      "epoch": 0.5352968451884753,
      "grad_norm": 0.0312141552567482,
      "learning_rate": 0.0012,
      "loss": 2.6440539360046387,
      "step": 6602
    },
    {
      "epoch": 0.5354590072136763,
      "grad_norm": 0.02721269801259041,
      "learning_rate": 0.0012,
      "loss": 2.5975193977355957,
      "step": 6604
    },
    {
      "epoch": 0.5356211692388774,
      "grad_norm": 0.024984784424304962,
      "learning_rate": 0.0012,
      "loss": 2.6386008262634277,
      "step": 6606
    },
    {
      "epoch": 0.5357833312640783,
      "grad_norm": 0.02529660426080227,
      "learning_rate": 0.0012,
      "loss": 2.6865527629852295,
      "step": 6608
    },
    {
      "epoch": 0.5359454932892793,
      "grad_norm": 0.048532746732234955,
      "learning_rate": 0.0012,
      "loss": 2.704451084136963,
      "step": 6610
    },
    {
      "epoch": 0.5361076553144803,
      "grad_norm": 0.05920787900686264,
      "learning_rate": 0.0012,
      "loss": 2.6975784301757812,
      "step": 6612
    },
    {
      "epoch": 0.5362698173396813,
      "grad_norm": 0.04597599059343338,
      "learning_rate": 0.0012,
      "loss": 2.6974384784698486,
      "step": 6614
    },
    {
      "epoch": 0.5364319793648823,
      "grad_norm": 0.023146124556660652,
      "learning_rate": 0.0012,
      "loss": 2.7220940589904785,
      "step": 6616
    },
    {
      "epoch": 0.5365941413900833,
      "grad_norm": 0.03330466151237488,
      "learning_rate": 0.0012,
      "loss": 2.7100820541381836,
      "step": 6618
    },
    {
      "epoch": 0.5367563034152842,
      "grad_norm": 0.04005567729473114,
      "learning_rate": 0.0012,
      "loss": 2.660245418548584,
      "step": 6620
    },
    {
      "epoch": 0.5369184654404853,
      "grad_norm": 0.05448184162378311,
      "learning_rate": 0.0012,
      "loss": 2.6268796920776367,
      "step": 6622
    },
    {
      "epoch": 0.5370806274656863,
      "grad_norm": 0.05734500661492348,
      "learning_rate": 0.0012,
      "loss": 2.6563568115234375,
      "step": 6624
    },
    {
      "epoch": 0.5372427894908872,
      "grad_norm": 0.03784773126244545,
      "learning_rate": 0.0012,
      "loss": 2.639258623123169,
      "step": 6626
    },
    {
      "epoch": 0.5374049515160882,
      "grad_norm": 0.02542293258011341,
      "learning_rate": 0.0012,
      "loss": 2.76218843460083,
      "step": 6628
    },
    {
      "epoch": 0.5375671135412893,
      "grad_norm": 0.022792872041463852,
      "learning_rate": 0.0012,
      "loss": 2.66951322555542,
      "step": 6630
    },
    {
      "epoch": 0.5377292755664902,
      "grad_norm": 0.029493851587176323,
      "learning_rate": 0.0012,
      "loss": 2.721268653869629,
      "step": 6632
    },
    {
      "epoch": 0.5378914375916912,
      "grad_norm": 0.02990178018808365,
      "learning_rate": 0.0012,
      "loss": 2.6435317993164062,
      "step": 6634
    },
    {
      "epoch": 0.5380535996168923,
      "grad_norm": 0.023480340838432312,
      "learning_rate": 0.0012,
      "loss": 2.6785728931427,
      "step": 6636
    },
    {
      "epoch": 0.5382157616420932,
      "grad_norm": 0.0372522808611393,
      "learning_rate": 0.0012,
      "loss": 2.769015312194824,
      "step": 6638
    },
    {
      "epoch": 0.5383779236672942,
      "grad_norm": 0.03725928068161011,
      "learning_rate": 0.0012,
      "loss": 2.730328321456909,
      "step": 6640
    },
    {
      "epoch": 0.5385400856924952,
      "grad_norm": 0.044370874762535095,
      "learning_rate": 0.0012,
      "loss": 2.720224618911743,
      "step": 6642
    },
    {
      "epoch": 0.5387022477176961,
      "grad_norm": 0.02684248425066471,
      "learning_rate": 0.0012,
      "loss": 2.63295316696167,
      "step": 6644
    },
    {
      "epoch": 0.5388644097428972,
      "grad_norm": 0.0524991974234581,
      "learning_rate": 0.0012,
      "loss": 2.7402536869049072,
      "step": 6646
    },
    {
      "epoch": 0.5390265717680982,
      "grad_norm": 0.07407867908477783,
      "learning_rate": 0.0012,
      "loss": 2.653918743133545,
      "step": 6648
    },
    {
      "epoch": 0.5391887337932991,
      "grad_norm": 0.027015632018446922,
      "learning_rate": 0.0012,
      "loss": 2.7281765937805176,
      "step": 6650
    },
    {
      "epoch": 0.5393508958185002,
      "grad_norm": 0.034959014505147934,
      "learning_rate": 0.0012,
      "loss": 2.7002222537994385,
      "step": 6652
    },
    {
      "epoch": 0.5395130578437012,
      "grad_norm": 0.028898967429995537,
      "learning_rate": 0.0012,
      "loss": 2.7198843955993652,
      "step": 6654
    },
    {
      "epoch": 0.5396752198689021,
      "grad_norm": 0.02412573993206024,
      "learning_rate": 0.0012,
      "loss": 2.712085723876953,
      "step": 6656
    },
    {
      "epoch": 0.5398373818941031,
      "grad_norm": 0.03465362265706062,
      "learning_rate": 0.0012,
      "loss": 2.6665046215057373,
      "step": 6658
    },
    {
      "epoch": 0.5399995439193042,
      "grad_norm": 0.04145431891083717,
      "learning_rate": 0.0012,
      "loss": 2.6555728912353516,
      "step": 6660
    },
    {
      "epoch": 0.5401617059445051,
      "grad_norm": 0.0277504064142704,
      "learning_rate": 0.0012,
      "loss": 2.707655906677246,
      "step": 6662
    },
    {
      "epoch": 0.5403238679697061,
      "grad_norm": 0.03016875497996807,
      "learning_rate": 0.0012,
      "loss": 2.7197961807250977,
      "step": 6664
    },
    {
      "epoch": 0.5404860299949071,
      "grad_norm": 0.041542306542396545,
      "learning_rate": 0.0012,
      "loss": 2.6515965461730957,
      "step": 6666
    },
    {
      "epoch": 0.540648192020108,
      "grad_norm": 0.04931865260004997,
      "learning_rate": 0.0012,
      "loss": 2.6662096977233887,
      "step": 6668
    },
    {
      "epoch": 0.5408103540453091,
      "grad_norm": 0.028387319296598434,
      "learning_rate": 0.0012,
      "loss": 2.7323555946350098,
      "step": 6670
    },
    {
      "epoch": 0.5409725160705101,
      "grad_norm": 0.02449892833828926,
      "learning_rate": 0.0012,
      "loss": 2.6938095092773438,
      "step": 6672
    },
    {
      "epoch": 0.541134678095711,
      "grad_norm": 0.03043942153453827,
      "learning_rate": 0.0012,
      "loss": 2.708310604095459,
      "step": 6674
    },
    {
      "epoch": 0.5412968401209121,
      "grad_norm": 0.02529538981616497,
      "learning_rate": 0.0012,
      "loss": 2.6401805877685547,
      "step": 6676
    },
    {
      "epoch": 0.5414590021461131,
      "grad_norm": 0.03307458013296127,
      "learning_rate": 0.0012,
      "loss": 2.6979455947875977,
      "step": 6678
    },
    {
      "epoch": 0.541621164171314,
      "grad_norm": 0.03823784366250038,
      "learning_rate": 0.0012,
      "loss": 2.675240993499756,
      "step": 6680
    },
    {
      "epoch": 0.541783326196515,
      "grad_norm": 0.04137999564409256,
      "learning_rate": 0.0012,
      "loss": 2.7310924530029297,
      "step": 6682
    },
    {
      "epoch": 0.5419454882217161,
      "grad_norm": 0.046890124678611755,
      "learning_rate": 0.0012,
      "loss": 2.6885266304016113,
      "step": 6684
    },
    {
      "epoch": 0.542107650246917,
      "grad_norm": 0.052456021308898926,
      "learning_rate": 0.0012,
      "loss": 2.6341729164123535,
      "step": 6686
    },
    {
      "epoch": 0.542269812272118,
      "grad_norm": 0.03154517710208893,
      "learning_rate": 0.0012,
      "loss": 2.6172125339508057,
      "step": 6688
    },
    {
      "epoch": 0.542431974297319,
      "grad_norm": 0.03423500433564186,
      "learning_rate": 0.0012,
      "loss": 2.6908631324768066,
      "step": 6690
    },
    {
      "epoch": 0.54259413632252,
      "grad_norm": 0.03783315420150757,
      "learning_rate": 0.0012,
      "loss": 2.660407066345215,
      "step": 6692
    },
    {
      "epoch": 0.542756298347721,
      "grad_norm": 0.039785485714673996,
      "learning_rate": 0.0012,
      "loss": 2.695058822631836,
      "step": 6694
    },
    {
      "epoch": 0.542918460372922,
      "grad_norm": 0.027753517031669617,
      "learning_rate": 0.0012,
      "loss": 2.6808505058288574,
      "step": 6696
    },
    {
      "epoch": 0.5430806223981229,
      "grad_norm": 0.02359328232705593,
      "learning_rate": 0.0012,
      "loss": 2.677489757537842,
      "step": 6698
    },
    {
      "epoch": 0.543242784423324,
      "grad_norm": 0.031941600143909454,
      "learning_rate": 0.0012,
      "loss": 2.71182918548584,
      "step": 6700
    },
    {
      "epoch": 0.543404946448525,
      "grad_norm": 0.05925151705741882,
      "learning_rate": 0.0012,
      "loss": 2.6797032356262207,
      "step": 6702
    },
    {
      "epoch": 0.5435671084737259,
      "grad_norm": 0.0427766814827919,
      "learning_rate": 0.0012,
      "loss": 2.600956916809082,
      "step": 6704
    },
    {
      "epoch": 0.5437292704989269,
      "grad_norm": 0.023161055520176888,
      "learning_rate": 0.0012,
      "loss": 2.6366677284240723,
      "step": 6706
    },
    {
      "epoch": 0.543891432524128,
      "grad_norm": 0.0613589882850647,
      "learning_rate": 0.0012,
      "loss": 2.635741710662842,
      "step": 6708
    },
    {
      "epoch": 0.5440535945493289,
      "grad_norm": 0.048382215201854706,
      "learning_rate": 0.0012,
      "loss": 2.656585216522217,
      "step": 6710
    },
    {
      "epoch": 0.5442157565745299,
      "grad_norm": 0.04819461703300476,
      "learning_rate": 0.0012,
      "loss": 2.6602582931518555,
      "step": 6712
    },
    {
      "epoch": 0.544377918599731,
      "grad_norm": 0.05128031596541405,
      "learning_rate": 0.0012,
      "loss": 2.733170986175537,
      "step": 6714
    },
    {
      "epoch": 0.5445400806249319,
      "grad_norm": 0.02327827364206314,
      "learning_rate": 0.0012,
      "loss": 2.5677990913391113,
      "step": 6716
    },
    {
      "epoch": 0.5447022426501329,
      "grad_norm": 0.024229587987065315,
      "learning_rate": 0.0012,
      "loss": 2.596475124359131,
      "step": 6718
    },
    {
      "epoch": 0.5448644046753339,
      "grad_norm": 0.037496283650398254,
      "learning_rate": 0.0012,
      "loss": 2.737755298614502,
      "step": 6720
    },
    {
      "epoch": 0.5450265667005348,
      "grad_norm": 0.024564141407608986,
      "learning_rate": 0.0012,
      "loss": 2.7069807052612305,
      "step": 6722
    },
    {
      "epoch": 0.5451887287257359,
      "grad_norm": 0.03274565935134888,
      "learning_rate": 0.0012,
      "loss": 2.726253032684326,
      "step": 6724
    },
    {
      "epoch": 0.5453508907509369,
      "grad_norm": 0.032798513770103455,
      "learning_rate": 0.0012,
      "loss": 2.632831573486328,
      "step": 6726
    },
    {
      "epoch": 0.5455130527761378,
      "grad_norm": 0.029673367738723755,
      "learning_rate": 0.0012,
      "loss": 2.6589441299438477,
      "step": 6728
    },
    {
      "epoch": 0.5456752148013388,
      "grad_norm": 0.024454884231090546,
      "learning_rate": 0.0012,
      "loss": 2.7102723121643066,
      "step": 6730
    },
    {
      "epoch": 0.5458373768265399,
      "grad_norm": 0.02450478821992874,
      "learning_rate": 0.0012,
      "loss": 2.7089409828186035,
      "step": 6732
    },
    {
      "epoch": 0.5459995388517408,
      "grad_norm": 0.030588237568736076,
      "learning_rate": 0.0012,
      "loss": 2.711021900177002,
      "step": 6734
    },
    {
      "epoch": 0.5461617008769418,
      "grad_norm": 0.03316376730799675,
      "learning_rate": 0.0012,
      "loss": 2.6686205863952637,
      "step": 6736
    },
    {
      "epoch": 0.5463238629021429,
      "grad_norm": 0.03379342332482338,
      "learning_rate": 0.0012,
      "loss": 2.6358108520507812,
      "step": 6738
    },
    {
      "epoch": 0.5464860249273438,
      "grad_norm": 0.02633415162563324,
      "learning_rate": 0.0012,
      "loss": 2.71540904045105,
      "step": 6740
    },
    {
      "epoch": 0.5466481869525448,
      "grad_norm": 0.02352391555905342,
      "learning_rate": 0.0012,
      "loss": 2.720874071121216,
      "step": 6742
    },
    {
      "epoch": 0.5468103489777458,
      "grad_norm": 0.0274822860956192,
      "learning_rate": 0.0012,
      "loss": 2.6690497398376465,
      "step": 6744
    },
    {
      "epoch": 0.5469725110029467,
      "grad_norm": 0.023614127188920975,
      "learning_rate": 0.0012,
      "loss": 2.7112934589385986,
      "step": 6746
    },
    {
      "epoch": 0.5471346730281478,
      "grad_norm": 0.03455526754260063,
      "learning_rate": 0.0012,
      "loss": 2.6280059814453125,
      "step": 6748
    },
    {
      "epoch": 0.5472968350533488,
      "grad_norm": 0.05398969352245331,
      "learning_rate": 0.0012,
      "loss": 2.7023978233337402,
      "step": 6750
    },
    {
      "epoch": 0.5474589970785497,
      "grad_norm": 0.034019578248262405,
      "learning_rate": 0.0012,
      "loss": 2.7008635997772217,
      "step": 6752
    },
    {
      "epoch": 0.5476211591037508,
      "grad_norm": 0.035736437886953354,
      "learning_rate": 0.0012,
      "loss": 2.626615285873413,
      "step": 6754
    },
    {
      "epoch": 0.5477833211289518,
      "grad_norm": 0.03363034874200821,
      "learning_rate": 0.0012,
      "loss": 2.692126750946045,
      "step": 6756
    },
    {
      "epoch": 0.5479454831541527,
      "grad_norm": 0.02615177258849144,
      "learning_rate": 0.0012,
      "loss": 2.742604970932007,
      "step": 6758
    },
    {
      "epoch": 0.5481076451793537,
      "grad_norm": 0.03136158362030983,
      "learning_rate": 0.0012,
      "loss": 2.716859817504883,
      "step": 6760
    },
    {
      "epoch": 0.5482698072045548,
      "grad_norm": 0.024623511359095573,
      "learning_rate": 0.0012,
      "loss": 2.6771106719970703,
      "step": 6762
    },
    {
      "epoch": 0.5484319692297557,
      "grad_norm": 0.026630401611328125,
      "learning_rate": 0.0012,
      "loss": 2.756042957305908,
      "step": 6764
    },
    {
      "epoch": 0.5485941312549567,
      "grad_norm": 0.031137557700276375,
      "learning_rate": 0.0012,
      "loss": 2.688462495803833,
      "step": 6766
    },
    {
      "epoch": 0.5487562932801577,
      "grad_norm": 0.028574500232934952,
      "learning_rate": 0.0012,
      "loss": 2.7438619136810303,
      "step": 6768
    },
    {
      "epoch": 0.5489184553053587,
      "grad_norm": 0.0474340058863163,
      "learning_rate": 0.0012,
      "loss": 2.7152929306030273,
      "step": 6770
    },
    {
      "epoch": 0.5490806173305597,
      "grad_norm": 0.04815001040697098,
      "learning_rate": 0.0012,
      "loss": 2.689779043197632,
      "step": 6772
    },
    {
      "epoch": 0.5492427793557607,
      "grad_norm": 0.043586939573287964,
      "learning_rate": 0.0012,
      "loss": 2.7085771560668945,
      "step": 6774
    },
    {
      "epoch": 0.5494049413809616,
      "grad_norm": 0.027396516874432564,
      "learning_rate": 0.0012,
      "loss": 2.660918951034546,
      "step": 6776
    },
    {
      "epoch": 0.5495671034061627,
      "grad_norm": 0.05722186714410782,
      "learning_rate": 0.0012,
      "loss": 2.669950485229492,
      "step": 6778
    },
    {
      "epoch": 0.5497292654313637,
      "grad_norm": 0.11116544902324677,
      "learning_rate": 0.0012,
      "loss": 2.6589508056640625,
      "step": 6780
    },
    {
      "epoch": 0.5498914274565646,
      "grad_norm": 0.046535808593034744,
      "learning_rate": 0.0012,
      "loss": 2.7277212142944336,
      "step": 6782
    },
    {
      "epoch": 0.5500535894817656,
      "grad_norm": 0.0338413305580616,
      "learning_rate": 0.0012,
      "loss": 2.5748329162597656,
      "step": 6784
    },
    {
      "epoch": 0.5502157515069667,
      "grad_norm": 0.026388274505734444,
      "learning_rate": 0.0012,
      "loss": 2.653675079345703,
      "step": 6786
    },
    {
      "epoch": 0.5503779135321676,
      "grad_norm": 0.028265537694096565,
      "learning_rate": 0.0012,
      "loss": 2.7237565517425537,
      "step": 6788
    },
    {
      "epoch": 0.5505400755573686,
      "grad_norm": 0.05536464601755142,
      "learning_rate": 0.0012,
      "loss": 2.618159532546997,
      "step": 6790
    },
    {
      "epoch": 0.5507022375825696,
      "grad_norm": 0.06360327452421188,
      "learning_rate": 0.0012,
      "loss": 2.656525135040283,
      "step": 6792
    },
    {
      "epoch": 0.5508643996077706,
      "grad_norm": 0.05073021352291107,
      "learning_rate": 0.0012,
      "loss": 2.6811280250549316,
      "step": 6794
    },
    {
      "epoch": 0.5510265616329716,
      "grad_norm": 0.026065269485116005,
      "learning_rate": 0.0012,
      "loss": 2.7774083614349365,
      "step": 6796
    },
    {
      "epoch": 0.5511887236581726,
      "grad_norm": 0.024943837895989418,
      "learning_rate": 0.0012,
      "loss": 2.6833391189575195,
      "step": 6798
    },
    {
      "epoch": 0.5513508856833735,
      "grad_norm": 0.034065183252096176,
      "learning_rate": 0.0012,
      "loss": 2.698575496673584,
      "step": 6800
    },
    {
      "epoch": 0.5515130477085746,
      "grad_norm": 0.030705202370882034,
      "learning_rate": 0.0012,
      "loss": 2.6830692291259766,
      "step": 6802
    },
    {
      "epoch": 0.5516752097337756,
      "grad_norm": 0.025996524840593338,
      "learning_rate": 0.0012,
      "loss": 2.711272716522217,
      "step": 6804
    },
    {
      "epoch": 0.5518373717589765,
      "grad_norm": 0.030957072973251343,
      "learning_rate": 0.0012,
      "loss": 2.680187463760376,
      "step": 6806
    },
    {
      "epoch": 0.5519995337841775,
      "grad_norm": 0.024432897567749023,
      "learning_rate": 0.0012,
      "loss": 2.6867289543151855,
      "step": 6808
    },
    {
      "epoch": 0.5521616958093786,
      "grad_norm": 0.02852768450975418,
      "learning_rate": 0.0012,
      "loss": 2.6838369369506836,
      "step": 6810
    },
    {
      "epoch": 0.5523238578345795,
      "grad_norm": 0.03301677480340004,
      "learning_rate": 0.0012,
      "loss": 2.7100729942321777,
      "step": 6812
    },
    {
      "epoch": 0.5524860198597805,
      "grad_norm": 0.03904037922620773,
      "learning_rate": 0.0012,
      "loss": 2.6362552642822266,
      "step": 6814
    },
    {
      "epoch": 0.5526481818849815,
      "grad_norm": 0.04175678640604019,
      "learning_rate": 0.0012,
      "loss": 2.6825509071350098,
      "step": 6816
    },
    {
      "epoch": 0.5528103439101825,
      "grad_norm": 0.05805923417210579,
      "learning_rate": 0.0012,
      "loss": 2.623218059539795,
      "step": 6818
    },
    {
      "epoch": 0.5529725059353835,
      "grad_norm": 0.033090636134147644,
      "learning_rate": 0.0012,
      "loss": 2.673963785171509,
      "step": 6820
    },
    {
      "epoch": 0.5531346679605845,
      "grad_norm": 0.025990737602114677,
      "learning_rate": 0.0012,
      "loss": 2.674398183822632,
      "step": 6822
    },
    {
      "epoch": 0.5532968299857854,
      "grad_norm": 0.03755463659763336,
      "learning_rate": 0.0012,
      "loss": 2.6238484382629395,
      "step": 6824
    },
    {
      "epoch": 0.5534589920109865,
      "grad_norm": 0.04393076151609421,
      "learning_rate": 0.0012,
      "loss": 2.721930503845215,
      "step": 6826
    },
    {
      "epoch": 0.5536211540361875,
      "grad_norm": 0.04427732899785042,
      "learning_rate": 0.0012,
      "loss": 2.6821532249450684,
      "step": 6828
    },
    {
      "epoch": 0.5537833160613884,
      "grad_norm": 0.033188361674547195,
      "learning_rate": 0.0012,
      "loss": 2.5905985832214355,
      "step": 6830
    },
    {
      "epoch": 0.5539454780865894,
      "grad_norm": 0.025083420798182487,
      "learning_rate": 0.0012,
      "loss": 2.6549317836761475,
      "step": 6832
    },
    {
      "epoch": 0.5541076401117905,
      "grad_norm": 0.02696244791150093,
      "learning_rate": 0.0012,
      "loss": 2.7423038482666016,
      "step": 6834
    },
    {
      "epoch": 0.5542698021369914,
      "grad_norm": 0.02724801003932953,
      "learning_rate": 0.0012,
      "loss": 2.6596384048461914,
      "step": 6836
    },
    {
      "epoch": 0.5544319641621924,
      "grad_norm": 0.0267502311617136,
      "learning_rate": 0.0012,
      "loss": 2.621659278869629,
      "step": 6838
    },
    {
      "epoch": 0.5545941261873935,
      "grad_norm": 0.030303724110126495,
      "learning_rate": 0.0012,
      "loss": 2.711533546447754,
      "step": 6840
    },
    {
      "epoch": 0.5547562882125944,
      "grad_norm": 0.024753805249929428,
      "learning_rate": 0.0012,
      "loss": 2.6480374336242676,
      "step": 6842
    },
    {
      "epoch": 0.5549184502377954,
      "grad_norm": 0.03044600784778595,
      "learning_rate": 0.0012,
      "loss": 2.6552011966705322,
      "step": 6844
    },
    {
      "epoch": 0.5550806122629964,
      "grad_norm": 0.025154661387205124,
      "learning_rate": 0.0012,
      "loss": 2.656614065170288,
      "step": 6846
    },
    {
      "epoch": 0.5552427742881973,
      "grad_norm": 0.023171881213784218,
      "learning_rate": 0.0012,
      "loss": 2.6952292919158936,
      "step": 6848
    },
    {
      "epoch": 0.5554049363133984,
      "grad_norm": 0.028429273515939713,
      "learning_rate": 0.0012,
      "loss": 2.6959986686706543,
      "step": 6850
    },
    {
      "epoch": 0.5555670983385994,
      "grad_norm": 0.0237446166574955,
      "learning_rate": 0.0012,
      "loss": 2.6315500736236572,
      "step": 6852
    },
    {
      "epoch": 0.5557292603638003,
      "grad_norm": 0.03246494010090828,
      "learning_rate": 0.0012,
      "loss": 2.63254451751709,
      "step": 6854
    },
    {
      "epoch": 0.5558914223890014,
      "grad_norm": 0.030606959015130997,
      "learning_rate": 0.0012,
      "loss": 2.6673216819763184,
      "step": 6856
    },
    {
      "epoch": 0.5560535844142024,
      "grad_norm": 0.03155069053173065,
      "learning_rate": 0.0012,
      "loss": 2.6401586532592773,
      "step": 6858
    },
    {
      "epoch": 0.5562157464394033,
      "grad_norm": 0.04356325417757034,
      "learning_rate": 0.0012,
      "loss": 2.6911463737487793,
      "step": 6860
    },
    {
      "epoch": 0.5563779084646043,
      "grad_norm": 0.05521580949425697,
      "learning_rate": 0.0012,
      "loss": 2.6726064682006836,
      "step": 6862
    },
    {
      "epoch": 0.5565400704898054,
      "grad_norm": 0.03726234287023544,
      "learning_rate": 0.0012,
      "loss": 2.6907947063446045,
      "step": 6864
    },
    {
      "epoch": 0.5567022325150063,
      "grad_norm": 0.027013877406716347,
      "learning_rate": 0.0012,
      "loss": 2.681715965270996,
      "step": 6866
    },
    {
      "epoch": 0.5568643945402073,
      "grad_norm": 0.03716660663485527,
      "learning_rate": 0.0012,
      "loss": 2.703312397003174,
      "step": 6868
    },
    {
      "epoch": 0.5570265565654083,
      "grad_norm": 0.027382327243685722,
      "learning_rate": 0.0012,
      "loss": 2.744025468826294,
      "step": 6870
    },
    {
      "epoch": 0.5571887185906093,
      "grad_norm": 0.0233429204672575,
      "learning_rate": 0.0012,
      "loss": 2.6740975379943848,
      "step": 6872
    },
    {
      "epoch": 0.5573508806158103,
      "grad_norm": 0.02807898446917534,
      "learning_rate": 0.0012,
      "loss": 2.6345551013946533,
      "step": 6874
    },
    {
      "epoch": 0.5575130426410113,
      "grad_norm": 0.03247606009244919,
      "learning_rate": 0.0012,
      "loss": 2.6967954635620117,
      "step": 6876
    },
    {
      "epoch": 0.5576752046662122,
      "grad_norm": 0.04745175689458847,
      "learning_rate": 0.0012,
      "loss": 2.676121711730957,
      "step": 6878
    },
    {
      "epoch": 0.5578373666914133,
      "grad_norm": 0.06728771328926086,
      "learning_rate": 0.0012,
      "loss": 2.7573189735412598,
      "step": 6880
    },
    {
      "epoch": 0.5579995287166143,
      "grad_norm": 0.05127999931573868,
      "learning_rate": 0.0012,
      "loss": 2.6812427043914795,
      "step": 6882
    },
    {
      "epoch": 0.5581616907418152,
      "grad_norm": 0.029560888186097145,
      "learning_rate": 0.0012,
      "loss": 2.6298601627349854,
      "step": 6884
    },
    {
      "epoch": 0.5583238527670162,
      "grad_norm": 0.027253853157162666,
      "learning_rate": 0.0012,
      "loss": 2.650726795196533,
      "step": 6886
    },
    {
      "epoch": 0.5584860147922173,
      "grad_norm": 0.026052048429846764,
      "learning_rate": 0.0012,
      "loss": 2.635193109512329,
      "step": 6888
    },
    {
      "epoch": 0.5586481768174182,
      "grad_norm": 0.07230863720178604,
      "learning_rate": 0.0012,
      "loss": 2.73876690864563,
      "step": 6890
    },
    {
      "epoch": 0.5588103388426192,
      "grad_norm": 0.0687812939286232,
      "learning_rate": 0.0012,
      "loss": 2.6750593185424805,
      "step": 6892
    },
    {
      "epoch": 0.5589725008678202,
      "grad_norm": 0.05378777161240578,
      "learning_rate": 0.0012,
      "loss": 2.6446030139923096,
      "step": 6894
    },
    {
      "epoch": 0.5591346628930212,
      "grad_norm": 0.028558332473039627,
      "learning_rate": 0.0012,
      "loss": 2.6664657592773438,
      "step": 6896
    },
    {
      "epoch": 0.5592968249182222,
      "grad_norm": 0.06346588581800461,
      "learning_rate": 0.0012,
      "loss": 2.656446933746338,
      "step": 6898
    },
    {
      "epoch": 0.5594589869434232,
      "grad_norm": 0.03806954249739647,
      "learning_rate": 0.0012,
      "loss": 2.7254385948181152,
      "step": 6900
    },
    {
      "epoch": 0.5596211489686241,
      "grad_norm": 0.035886943340301514,
      "learning_rate": 0.0012,
      "loss": 2.6873998641967773,
      "step": 6902
    },
    {
      "epoch": 0.5597833109938252,
      "grad_norm": 0.024612918496131897,
      "learning_rate": 0.0012,
      "loss": 2.6166207790374756,
      "step": 6904
    },
    {
      "epoch": 0.5599454730190262,
      "grad_norm": 0.026923397555947304,
      "learning_rate": 0.0012,
      "loss": 2.661602258682251,
      "step": 6906
    },
    {
      "epoch": 0.5601076350442271,
      "grad_norm": 0.035930193960666656,
      "learning_rate": 0.0012,
      "loss": 2.651637077331543,
      "step": 6908
    },
    {
      "epoch": 0.5602697970694281,
      "grad_norm": 0.03921961039304733,
      "learning_rate": 0.0012,
      "loss": 2.6747524738311768,
      "step": 6910
    },
    {
      "epoch": 0.5604319590946292,
      "grad_norm": 0.06754613667726517,
      "learning_rate": 0.0012,
      "loss": 2.7362070083618164,
      "step": 6912
    },
    {
      "epoch": 0.5605941211198301,
      "grad_norm": 0.07598903775215149,
      "learning_rate": 0.0012,
      "loss": 2.733760356903076,
      "step": 6914
    },
    {
      "epoch": 0.5607562831450311,
      "grad_norm": 0.03937598690390587,
      "learning_rate": 0.0012,
      "loss": 2.590930461883545,
      "step": 6916
    },
    {
      "epoch": 0.5609184451702321,
      "grad_norm": 0.029124200344085693,
      "learning_rate": 0.0012,
      "loss": 2.682680606842041,
      "step": 6918
    },
    {
      "epoch": 0.5610806071954331,
      "grad_norm": 0.06347868591547012,
      "learning_rate": 0.0012,
      "loss": 2.686478614807129,
      "step": 6920
    },
    {
      "epoch": 0.5612427692206341,
      "grad_norm": 0.030547283589839935,
      "learning_rate": 0.0012,
      "loss": 2.636763572692871,
      "step": 6922
    },
    {
      "epoch": 0.5614049312458351,
      "grad_norm": 0.04276842623949051,
      "learning_rate": 0.0012,
      "loss": 2.6195850372314453,
      "step": 6924
    },
    {
      "epoch": 0.561567093271036,
      "grad_norm": 0.07153602689504623,
      "learning_rate": 0.0012,
      "loss": 2.688568592071533,
      "step": 6926
    },
    {
      "epoch": 0.5617292552962371,
      "grad_norm": 0.037177689373493195,
      "learning_rate": 0.0012,
      "loss": 2.701345205307007,
      "step": 6928
    },
    {
      "epoch": 0.5618914173214381,
      "grad_norm": 0.04054165259003639,
      "learning_rate": 0.0012,
      "loss": 2.727489471435547,
      "step": 6930
    },
    {
      "epoch": 0.562053579346639,
      "grad_norm": 0.030180765315890312,
      "learning_rate": 0.0012,
      "loss": 2.7163405418395996,
      "step": 6932
    },
    {
      "epoch": 0.56221574137184,
      "grad_norm": 0.025027496740221977,
      "learning_rate": 0.0012,
      "loss": 2.71852707862854,
      "step": 6934
    },
    {
      "epoch": 0.5623779033970411,
      "grad_norm": 0.033478155732154846,
      "learning_rate": 0.0012,
      "loss": 2.695958137512207,
      "step": 6936
    },
    {
      "epoch": 0.562540065422242,
      "grad_norm": 0.033528883010149,
      "learning_rate": 0.0012,
      "loss": 2.667928695678711,
      "step": 6938
    },
    {
      "epoch": 0.562702227447443,
      "grad_norm": 0.051014311611652374,
      "learning_rate": 0.0012,
      "loss": 2.6779417991638184,
      "step": 6940
    },
    {
      "epoch": 0.5628643894726441,
      "grad_norm": 0.024048764258623123,
      "learning_rate": 0.0012,
      "loss": 2.6246447563171387,
      "step": 6942
    },
    {
      "epoch": 0.563026551497845,
      "grad_norm": 0.037668485194444656,
      "learning_rate": 0.0012,
      "loss": 2.6290245056152344,
      "step": 6944
    },
    {
      "epoch": 0.563188713523046,
      "grad_norm": 0.024379877373576164,
      "learning_rate": 0.0012,
      "loss": 2.7113733291625977,
      "step": 6946
    },
    {
      "epoch": 0.563350875548247,
      "grad_norm": 0.02698666602373123,
      "learning_rate": 0.0012,
      "loss": 2.727133274078369,
      "step": 6948
    },
    {
      "epoch": 0.563513037573448,
      "grad_norm": 0.02832374908030033,
      "learning_rate": 0.0012,
      "loss": 2.6087708473205566,
      "step": 6950
    },
    {
      "epoch": 0.563675199598649,
      "grad_norm": 0.03263133019208908,
      "learning_rate": 0.0012,
      "loss": 2.638288974761963,
      "step": 6952
    },
    {
      "epoch": 0.56383736162385,
      "grad_norm": 0.02814212255179882,
      "learning_rate": 0.0012,
      "loss": 2.663238525390625,
      "step": 6954
    },
    {
      "epoch": 0.5639995236490509,
      "grad_norm": 0.027297256514430046,
      "learning_rate": 0.0012,
      "loss": 2.674130916595459,
      "step": 6956
    },
    {
      "epoch": 0.564161685674252,
      "grad_norm": 0.025512468069791794,
      "learning_rate": 0.0012,
      "loss": 2.713367223739624,
      "step": 6958
    },
    {
      "epoch": 0.564323847699453,
      "grad_norm": 0.033194687217473984,
      "learning_rate": 0.0012,
      "loss": 2.6713027954101562,
      "step": 6960
    },
    {
      "epoch": 0.5644860097246539,
      "grad_norm": 0.030654996633529663,
      "learning_rate": 0.0012,
      "loss": 2.611663341522217,
      "step": 6962
    },
    {
      "epoch": 0.5646481717498549,
      "grad_norm": 0.0363130085170269,
      "learning_rate": 0.0012,
      "loss": 2.7502572536468506,
      "step": 6964
    },
    {
      "epoch": 0.564810333775056,
      "grad_norm": 0.036641400307416916,
      "learning_rate": 0.0012,
      "loss": 2.643526554107666,
      "step": 6966
    },
    {
      "epoch": 0.5649724958002569,
      "grad_norm": 0.02761419303715229,
      "learning_rate": 0.0012,
      "loss": 2.6107139587402344,
      "step": 6968
    },
    {
      "epoch": 0.5651346578254579,
      "grad_norm": 0.024724502116441727,
      "learning_rate": 0.0012,
      "loss": 2.67476749420166,
      "step": 6970
    },
    {
      "epoch": 0.5652968198506589,
      "grad_norm": 0.03289889171719551,
      "learning_rate": 0.0012,
      "loss": 2.7274162769317627,
      "step": 6972
    },
    {
      "epoch": 0.5654589818758599,
      "grad_norm": 0.05219056457281113,
      "learning_rate": 0.0012,
      "loss": 2.6688485145568848,
      "step": 6974
    },
    {
      "epoch": 0.5656211439010609,
      "grad_norm": 0.034907273948192596,
      "learning_rate": 0.0012,
      "loss": 2.6441667079925537,
      "step": 6976
    },
    {
      "epoch": 0.5657833059262619,
      "grad_norm": 0.028755253180861473,
      "learning_rate": 0.0012,
      "loss": 2.644514560699463,
      "step": 6978
    },
    {
      "epoch": 0.5659454679514628,
      "grad_norm": 0.03451044112443924,
      "learning_rate": 0.0012,
      "loss": 2.6715526580810547,
      "step": 6980
    },
    {
      "epoch": 0.5661076299766639,
      "grad_norm": 0.02653173729777336,
      "learning_rate": 0.0012,
      "loss": 2.6731796264648438,
      "step": 6982
    },
    {
      "epoch": 0.5662697920018649,
      "grad_norm": 0.029187368229031563,
      "learning_rate": 0.0012,
      "loss": 2.6954002380371094,
      "step": 6984
    },
    {
      "epoch": 0.5664319540270658,
      "grad_norm": 0.04212227836251259,
      "learning_rate": 0.0012,
      "loss": 2.7223901748657227,
      "step": 6986
    },
    {
      "epoch": 0.5665941160522668,
      "grad_norm": 0.04310845956206322,
      "learning_rate": 0.0012,
      "loss": 2.677572727203369,
      "step": 6988
    },
    {
      "epoch": 0.5667562780774679,
      "grad_norm": 0.060332197695970535,
      "learning_rate": 0.0012,
      "loss": 2.6567254066467285,
      "step": 6990
    },
    {
      "epoch": 0.5669184401026688,
      "grad_norm": 0.050108954310417175,
      "learning_rate": 0.0012,
      "loss": 2.6128199100494385,
      "step": 6992
    },
    {
      "epoch": 0.5670806021278698,
      "grad_norm": 0.03328549489378929,
      "learning_rate": 0.0012,
      "loss": 2.6290125846862793,
      "step": 6994
    },
    {
      "epoch": 0.5672427641530708,
      "grad_norm": 0.03157627210021019,
      "learning_rate": 0.0012,
      "loss": 2.672501564025879,
      "step": 6996
    },
    {
      "epoch": 0.5674049261782718,
      "grad_norm": 0.047680191695690155,
      "learning_rate": 0.0012,
      "loss": 2.6352734565734863,
      "step": 6998
    },
    {
      "epoch": 0.5675670882034728,
      "grad_norm": 0.0680319219827652,
      "learning_rate": 0.0012,
      "loss": 2.625818967819214,
      "step": 7000
    },
    {
      "epoch": 0.5677292502286738,
      "grad_norm": 0.026072947308421135,
      "learning_rate": 0.0012,
      "loss": 2.6897730827331543,
      "step": 7002
    },
    {
      "epoch": 0.5678914122538747,
      "grad_norm": 0.03587883710861206,
      "learning_rate": 0.0012,
      "loss": 2.6061272621154785,
      "step": 7004
    },
    {
      "epoch": 0.5680535742790758,
      "grad_norm": 0.0542241632938385,
      "learning_rate": 0.0012,
      "loss": 2.6478593349456787,
      "step": 7006
    },
    {
      "epoch": 0.5682157363042768,
      "grad_norm": 0.03605169802904129,
      "learning_rate": 0.0012,
      "loss": 2.6953225135803223,
      "step": 7008
    },
    {
      "epoch": 0.5683778983294777,
      "grad_norm": 0.02807118184864521,
      "learning_rate": 0.0012,
      "loss": 2.641144275665283,
      "step": 7010
    },
    {
      "epoch": 0.5685400603546787,
      "grad_norm": 0.040648866444826126,
      "learning_rate": 0.0012,
      "loss": 2.6172919273376465,
      "step": 7012
    },
    {
      "epoch": 0.5687022223798798,
      "grad_norm": 0.0381409153342247,
      "learning_rate": 0.0012,
      "loss": 2.689079761505127,
      "step": 7014
    },
    {
      "epoch": 0.5688643844050807,
      "grad_norm": 0.04196237400174141,
      "learning_rate": 0.0012,
      "loss": 2.650205612182617,
      "step": 7016
    },
    {
      "epoch": 0.5690265464302817,
      "grad_norm": 0.026821551844477654,
      "learning_rate": 0.0012,
      "loss": 2.6356120109558105,
      "step": 7018
    },
    {
      "epoch": 0.5691887084554828,
      "grad_norm": 0.030319128185510635,
      "learning_rate": 0.0012,
      "loss": 2.6321048736572266,
      "step": 7020
    },
    {
      "epoch": 0.5693508704806837,
      "grad_norm": 0.034652967005968094,
      "learning_rate": 0.0012,
      "loss": 2.6810755729675293,
      "step": 7022
    },
    {
      "epoch": 0.5695130325058847,
      "grad_norm": 0.038105450570583344,
      "learning_rate": 0.0012,
      "loss": 2.665055274963379,
      "step": 7024
    },
    {
      "epoch": 0.5696751945310857,
      "grad_norm": 0.029435433447360992,
      "learning_rate": 0.0012,
      "loss": 2.6777234077453613,
      "step": 7026
    },
    {
      "epoch": 0.5698373565562866,
      "grad_norm": 0.04601333662867546,
      "learning_rate": 0.0012,
      "loss": 2.6322522163391113,
      "step": 7028
    },
    {
      "epoch": 0.5699995185814877,
      "grad_norm": 0.025336122140288353,
      "learning_rate": 0.0012,
      "loss": 2.691044807434082,
      "step": 7030
    },
    {
      "epoch": 0.5701616806066887,
      "grad_norm": 0.02309499680995941,
      "learning_rate": 0.0012,
      "loss": 2.738874673843384,
      "step": 7032
    },
    {
      "epoch": 0.5703238426318896,
      "grad_norm": 0.027356503531336784,
      "learning_rate": 0.0012,
      "loss": 2.662252902984619,
      "step": 7034
    },
    {
      "epoch": 0.5704860046570907,
      "grad_norm": 0.044128429144620895,
      "learning_rate": 0.0012,
      "loss": 2.7471513748168945,
      "step": 7036
    },
    {
      "epoch": 0.5706481666822917,
      "grad_norm": 0.02417231723666191,
      "learning_rate": 0.0012,
      "loss": 2.6572041511535645,
      "step": 7038
    },
    {
      "epoch": 0.5708103287074926,
      "grad_norm": 0.0362798273563385,
      "learning_rate": 0.0012,
      "loss": 2.633111000061035,
      "step": 7040
    },
    {
      "epoch": 0.5709724907326936,
      "grad_norm": 0.033929795026779175,
      "learning_rate": 0.0012,
      "loss": 2.670991897583008,
      "step": 7042
    },
    {
      "epoch": 0.5711346527578947,
      "grad_norm": 0.023443862795829773,
      "learning_rate": 0.0012,
      "loss": 2.645406723022461,
      "step": 7044
    },
    {
      "epoch": 0.5712968147830956,
      "grad_norm": 0.02210919000208378,
      "learning_rate": 0.0012,
      "loss": 2.623778820037842,
      "step": 7046
    },
    {
      "epoch": 0.5714589768082966,
      "grad_norm": 0.025057654827833176,
      "learning_rate": 0.0012,
      "loss": 2.6248531341552734,
      "step": 7048
    },
    {
      "epoch": 0.5716211388334976,
      "grad_norm": 0.024335067719221115,
      "learning_rate": 0.0012,
      "loss": 2.6175990104675293,
      "step": 7050
    },
    {
      "epoch": 0.5717833008586986,
      "grad_norm": 0.04282648488879204,
      "learning_rate": 0.0012,
      "loss": 2.6656365394592285,
      "step": 7052
    },
    {
      "epoch": 0.5719454628838996,
      "grad_norm": 0.026622870936989784,
      "learning_rate": 0.0012,
      "loss": 2.7207164764404297,
      "step": 7054
    },
    {
      "epoch": 0.5721076249091006,
      "grad_norm": 0.03063627891242504,
      "learning_rate": 0.0012,
      "loss": 2.7051706314086914,
      "step": 7056
    },
    {
      "epoch": 0.5722697869343015,
      "grad_norm": 0.024039408192038536,
      "learning_rate": 0.0012,
      "loss": 2.685793876647949,
      "step": 7058
    },
    {
      "epoch": 0.5724319489595026,
      "grad_norm": 0.02467469312250614,
      "learning_rate": 0.0012,
      "loss": 2.7113232612609863,
      "step": 7060
    },
    {
      "epoch": 0.5725941109847036,
      "grad_norm": 0.025653846561908722,
      "learning_rate": 0.0012,
      "loss": 2.6420464515686035,
      "step": 7062
    },
    {
      "epoch": 0.5727562730099045,
      "grad_norm": 0.025794735178351402,
      "learning_rate": 0.0012,
      "loss": 2.6486802101135254,
      "step": 7064
    },
    {
      "epoch": 0.5729184350351055,
      "grad_norm": 0.03153698518872261,
      "learning_rate": 0.0012,
      "loss": 2.670048236846924,
      "step": 7066
    },
    {
      "epoch": 0.5730805970603066,
      "grad_norm": 0.033449914306402206,
      "learning_rate": 0.0012,
      "loss": 2.6049094200134277,
      "step": 7068
    },
    {
      "epoch": 0.5732427590855075,
      "grad_norm": 0.02541041001677513,
      "learning_rate": 0.0012,
      "loss": 2.652434825897217,
      "step": 7070
    },
    {
      "epoch": 0.5734049211107085,
      "grad_norm": 0.047255922108888626,
      "learning_rate": 0.0012,
      "loss": 2.7108922004699707,
      "step": 7072
    },
    {
      "epoch": 0.5735670831359095,
      "grad_norm": 0.04405702278017998,
      "learning_rate": 0.0012,
      "loss": 2.661588430404663,
      "step": 7074
    },
    {
      "epoch": 0.5737292451611105,
      "grad_norm": 0.049770060926675797,
      "learning_rate": 0.0012,
      "loss": 2.7161502838134766,
      "step": 7076
    },
    {
      "epoch": 0.5738914071863115,
      "grad_norm": 0.030669918283820152,
      "learning_rate": 0.0012,
      "loss": 2.714554786682129,
      "step": 7078
    },
    {
      "epoch": 0.5740535692115125,
      "grad_norm": 0.023693962022662163,
      "learning_rate": 0.0012,
      "loss": 2.6460390090942383,
      "step": 7080
    },
    {
      "epoch": 0.5742157312367134,
      "grad_norm": 0.031368687748909,
      "learning_rate": 0.0012,
      "loss": 2.707367420196533,
      "step": 7082
    },
    {
      "epoch": 0.5743778932619145,
      "grad_norm": 0.028368841856718063,
      "learning_rate": 0.0012,
      "loss": 2.650106906890869,
      "step": 7084
    },
    {
      "epoch": 0.5745400552871155,
      "grad_norm": 0.03423615172505379,
      "learning_rate": 0.0012,
      "loss": 2.680901050567627,
      "step": 7086
    },
    {
      "epoch": 0.5747022173123164,
      "grad_norm": 0.02364112064242363,
      "learning_rate": 0.0012,
      "loss": 2.66662335395813,
      "step": 7088
    },
    {
      "epoch": 0.5748643793375174,
      "grad_norm": 0.03329897299408913,
      "learning_rate": 0.0012,
      "loss": 2.622053623199463,
      "step": 7090
    },
    {
      "epoch": 0.5750265413627185,
      "grad_norm": 0.03333480283617973,
      "learning_rate": 0.0012,
      "loss": 2.638665199279785,
      "step": 7092
    },
    {
      "epoch": 0.5751887033879194,
      "grad_norm": 0.032098911702632904,
      "learning_rate": 0.0012,
      "loss": 2.6247401237487793,
      "step": 7094
    },
    {
      "epoch": 0.5753508654131204,
      "grad_norm": 0.024505330249667168,
      "learning_rate": 0.0012,
      "loss": 2.7576279640197754,
      "step": 7096
    },
    {
      "epoch": 0.5755130274383214,
      "grad_norm": 0.023451585322618484,
      "learning_rate": 0.0012,
      "loss": 2.6306164264678955,
      "step": 7098
    },
    {
      "epoch": 0.5756751894635224,
      "grad_norm": 0.03327390179038048,
      "learning_rate": 0.0012,
      "loss": 2.7178139686584473,
      "step": 7100
    },
    {
      "epoch": 0.5758373514887234,
      "grad_norm": 0.04389270767569542,
      "learning_rate": 0.0012,
      "loss": 2.6542158126831055,
      "step": 7102
    },
    {
      "epoch": 0.5759995135139244,
      "grad_norm": 0.038066938519477844,
      "learning_rate": 0.0012,
      "loss": 2.6733169555664062,
      "step": 7104
    },
    {
      "epoch": 0.5761616755391253,
      "grad_norm": 0.029216811060905457,
      "learning_rate": 0.0012,
      "loss": 2.622323513031006,
      "step": 7106
    },
    {
      "epoch": 0.5763238375643264,
      "grad_norm": 0.040130864828825,
      "learning_rate": 0.0012,
      "loss": 2.6267170906066895,
      "step": 7108
    },
    {
      "epoch": 0.5764859995895274,
      "grad_norm": 0.04130801931023598,
      "learning_rate": 0.0012,
      "loss": 2.7519664764404297,
      "step": 7110
    },
    {
      "epoch": 0.5766481616147283,
      "grad_norm": 0.022545846179127693,
      "learning_rate": 0.0012,
      "loss": 2.6731739044189453,
      "step": 7112
    },
    {
      "epoch": 0.5768103236399293,
      "grad_norm": 0.030208270996809006,
      "learning_rate": 0.0012,
      "loss": 2.598997116088867,
      "step": 7114
    },
    {
      "epoch": 0.5769724856651304,
      "grad_norm": 0.06557008624076843,
      "learning_rate": 0.0012,
      "loss": 2.6462976932525635,
      "step": 7116
    },
    {
      "epoch": 0.5771346476903313,
      "grad_norm": 0.03403101861476898,
      "learning_rate": 0.0012,
      "loss": 2.6953234672546387,
      "step": 7118
    },
    {
      "epoch": 0.5772968097155323,
      "grad_norm": 0.02320307306945324,
      "learning_rate": 0.0012,
      "loss": 2.659618854522705,
      "step": 7120
    },
    {
      "epoch": 0.5774589717407334,
      "grad_norm": 0.0258095171302557,
      "learning_rate": 0.0012,
      "loss": 2.6393256187438965,
      "step": 7122
    },
    {
      "epoch": 0.5776211337659343,
      "grad_norm": 0.0738554447889328,
      "learning_rate": 0.0012,
      "loss": 2.6876862049102783,
      "step": 7124
    },
    {
      "epoch": 0.5777832957911353,
      "grad_norm": 0.045812323689460754,
      "learning_rate": 0.0012,
      "loss": 2.681698799133301,
      "step": 7126
    },
    {
      "epoch": 0.5779454578163363,
      "grad_norm": 0.030366359278559685,
      "learning_rate": 0.0012,
      "loss": 2.6523964405059814,
      "step": 7128
    },
    {
      "epoch": 0.5781076198415372,
      "grad_norm": 0.02698003686964512,
      "learning_rate": 0.0012,
      "loss": 2.705235481262207,
      "step": 7130
    },
    {
      "epoch": 0.5782697818667383,
      "grad_norm": 0.03251587972044945,
      "learning_rate": 0.0012,
      "loss": 2.656818389892578,
      "step": 7132
    },
    {
      "epoch": 0.5784319438919393,
      "grad_norm": 0.053022339940071106,
      "learning_rate": 0.0012,
      "loss": 2.6166837215423584,
      "step": 7134
    },
    {
      "epoch": 0.5785941059171402,
      "grad_norm": 0.031745027750730515,
      "learning_rate": 0.0012,
      "loss": 2.6377739906311035,
      "step": 7136
    },
    {
      "epoch": 0.5787562679423413,
      "grad_norm": 0.024853384122252464,
      "learning_rate": 0.0012,
      "loss": 2.716770648956299,
      "step": 7138
    },
    {
      "epoch": 0.5789184299675423,
      "grad_norm": 0.048989493399858475,
      "learning_rate": 0.0012,
      "loss": 2.6814732551574707,
      "step": 7140
    },
    {
      "epoch": 0.5790805919927432,
      "grad_norm": 0.04945606738328934,
      "learning_rate": 0.0012,
      "loss": 2.7338104248046875,
      "step": 7142
    },
    {
      "epoch": 0.5792427540179442,
      "grad_norm": 0.034124597907066345,
      "learning_rate": 0.0012,
      "loss": 2.6097426414489746,
      "step": 7144
    },
    {
      "epoch": 0.5794049160431453,
      "grad_norm": 0.0331517718732357,
      "learning_rate": 0.0012,
      "loss": 2.6306326389312744,
      "step": 7146
    },
    {
      "epoch": 0.5795670780683462,
      "grad_norm": 0.05869806185364723,
      "learning_rate": 0.0012,
      "loss": 2.7117228507995605,
      "step": 7148
    },
    {
      "epoch": 0.5797292400935472,
      "grad_norm": 0.028890741989016533,
      "learning_rate": 0.0012,
      "loss": 2.7152037620544434,
      "step": 7150
    },
    {
      "epoch": 0.5798914021187482,
      "grad_norm": 0.024719303473830223,
      "learning_rate": 0.0012,
      "loss": 2.70426082611084,
      "step": 7152
    },
    {
      "epoch": 0.5800535641439492,
      "grad_norm": 0.024375688284635544,
      "learning_rate": 0.0012,
      "loss": 2.6847121715545654,
      "step": 7154
    },
    {
      "epoch": 0.5802157261691502,
      "grad_norm": 0.029922276735305786,
      "learning_rate": 0.0012,
      "loss": 2.7463669776916504,
      "step": 7156
    },
    {
      "epoch": 0.5803778881943512,
      "grad_norm": 0.030218038707971573,
      "learning_rate": 0.0012,
      "loss": 2.732914447784424,
      "step": 7158
    },
    {
      "epoch": 0.5805400502195521,
      "grad_norm": 0.024258580058813095,
      "learning_rate": 0.0012,
      "loss": 2.701516628265381,
      "step": 7160
    },
    {
      "epoch": 0.5807022122447532,
      "grad_norm": 0.02613108791410923,
      "learning_rate": 0.0012,
      "loss": 2.70843505859375,
      "step": 7162
    },
    {
      "epoch": 0.5808643742699542,
      "grad_norm": 0.044709477573633194,
      "learning_rate": 0.0012,
      "loss": 2.672023296356201,
      "step": 7164
    },
    {
      "epoch": 0.5810265362951551,
      "grad_norm": 0.026955751702189445,
      "learning_rate": 0.0012,
      "loss": 2.7081313133239746,
      "step": 7166
    },
    {
      "epoch": 0.5811886983203561,
      "grad_norm": 0.026240263134241104,
      "learning_rate": 0.0012,
      "loss": 2.682748317718506,
      "step": 7168
    },
    {
      "epoch": 0.5813508603455572,
      "grad_norm": 0.02526525780558586,
      "learning_rate": 0.0012,
      "loss": 2.663424015045166,
      "step": 7170
    },
    {
      "epoch": 0.5815130223707581,
      "grad_norm": 0.04698105528950691,
      "learning_rate": 0.0012,
      "loss": 2.634249687194824,
      "step": 7172
    },
    {
      "epoch": 0.5816751843959591,
      "grad_norm": 0.031020507216453552,
      "learning_rate": 0.0012,
      "loss": 2.7196311950683594,
      "step": 7174
    },
    {
      "epoch": 0.5818373464211601,
      "grad_norm": 0.02417561039328575,
      "learning_rate": 0.0012,
      "loss": 2.656266212463379,
      "step": 7176
    },
    {
      "epoch": 0.5819995084463611,
      "grad_norm": 0.03056269884109497,
      "learning_rate": 0.0012,
      "loss": 2.677823543548584,
      "step": 7178
    },
    {
      "epoch": 0.5821616704715621,
      "grad_norm": 0.03556473180651665,
      "learning_rate": 0.0012,
      "loss": 2.7249581813812256,
      "step": 7180
    },
    {
      "epoch": 0.5823238324967631,
      "grad_norm": 0.04725753515958786,
      "learning_rate": 0.0012,
      "loss": 2.6998679637908936,
      "step": 7182
    },
    {
      "epoch": 0.582485994521964,
      "grad_norm": 0.06748384982347488,
      "learning_rate": 0.0012,
      "loss": 2.617743968963623,
      "step": 7184
    },
    {
      "epoch": 0.5826481565471651,
      "grad_norm": 0.05327565222978592,
      "learning_rate": 0.0012,
      "loss": 2.6675305366516113,
      "step": 7186
    },
    {
      "epoch": 0.5828103185723661,
      "grad_norm": 0.029688309878110886,
      "learning_rate": 0.0012,
      "loss": 2.6540465354919434,
      "step": 7188
    },
    {
      "epoch": 0.582972480597567,
      "grad_norm": 0.05110746622085571,
      "learning_rate": 0.0012,
      "loss": 2.704094648361206,
      "step": 7190
    },
    {
      "epoch": 0.583134642622768,
      "grad_norm": 0.03280246630311012,
      "learning_rate": 0.0012,
      "loss": 2.6689553260803223,
      "step": 7192
    },
    {
      "epoch": 0.5832968046479691,
      "grad_norm": 0.025744248181581497,
      "learning_rate": 0.0012,
      "loss": 2.676692485809326,
      "step": 7194
    },
    {
      "epoch": 0.58345896667317,
      "grad_norm": 0.02802746370434761,
      "learning_rate": 0.0012,
      "loss": 2.66709566116333,
      "step": 7196
    },
    {
      "epoch": 0.583621128698371,
      "grad_norm": 0.023788506165146828,
      "learning_rate": 0.0012,
      "loss": 2.6178059577941895,
      "step": 7198
    },
    {
      "epoch": 0.583783290723572,
      "grad_norm": 0.02170496992766857,
      "learning_rate": 0.0012,
      "loss": 2.627042293548584,
      "step": 7200
    },
    {
      "epoch": 0.583945452748773,
      "grad_norm": 0.04365713894367218,
      "learning_rate": 0.0012,
      "loss": 2.653581142425537,
      "step": 7202
    },
    {
      "epoch": 0.584107614773974,
      "grad_norm": 0.025216195732355118,
      "learning_rate": 0.0012,
      "loss": 2.712671995162964,
      "step": 7204
    },
    {
      "epoch": 0.584269776799175,
      "grad_norm": 0.03052983246743679,
      "learning_rate": 0.0012,
      "loss": 2.6476216316223145,
      "step": 7206
    },
    {
      "epoch": 0.5844319388243759,
      "grad_norm": 0.021455632522702217,
      "learning_rate": 0.0012,
      "loss": 2.618819236755371,
      "step": 7208
    },
    {
      "epoch": 0.584594100849577,
      "grad_norm": 0.03130802884697914,
      "learning_rate": 0.0012,
      "loss": 2.651093006134033,
      "step": 7210
    },
    {
      "epoch": 0.584756262874778,
      "grad_norm": 0.02535160817205906,
      "learning_rate": 0.0012,
      "loss": 2.6715502738952637,
      "step": 7212
    },
    {
      "epoch": 0.5849184248999789,
      "grad_norm": 0.024597572162747383,
      "learning_rate": 0.0012,
      "loss": 2.7157459259033203,
      "step": 7214
    },
    {
      "epoch": 0.58508058692518,
      "grad_norm": 0.059112295508384705,
      "learning_rate": 0.0012,
      "loss": 2.641176700592041,
      "step": 7216
    },
    {
      "epoch": 0.585242748950381,
      "grad_norm": 0.050385959446430206,
      "learning_rate": 0.0012,
      "loss": 2.674006462097168,
      "step": 7218
    },
    {
      "epoch": 0.5854049109755819,
      "grad_norm": 0.031018376350402832,
      "learning_rate": 0.0012,
      "loss": 2.679870367050171,
      "step": 7220
    },
    {
      "epoch": 0.5855670730007829,
      "grad_norm": 0.02531994692981243,
      "learning_rate": 0.0012,
      "loss": 2.7366011142730713,
      "step": 7222
    },
    {
      "epoch": 0.585729235025984,
      "grad_norm": 0.03422960638999939,
      "learning_rate": 0.0012,
      "loss": 2.6209166049957275,
      "step": 7224
    },
    {
      "epoch": 0.5858913970511849,
      "grad_norm": 0.02267528884112835,
      "learning_rate": 0.0012,
      "loss": 2.683635711669922,
      "step": 7226
    },
    {
      "epoch": 0.5860535590763859,
      "grad_norm": 0.02561207115650177,
      "learning_rate": 0.0012,
      "loss": 2.6997690200805664,
      "step": 7228
    },
    {
      "epoch": 0.5862157211015869,
      "grad_norm": 0.02389494515955448,
      "learning_rate": 0.0012,
      "loss": 2.6786019802093506,
      "step": 7230
    },
    {
      "epoch": 0.5863778831267878,
      "grad_norm": 0.024922117590904236,
      "learning_rate": 0.0012,
      "loss": 2.674345016479492,
      "step": 7232
    },
    {
      "epoch": 0.5865400451519889,
      "grad_norm": 0.027077853679656982,
      "learning_rate": 0.0012,
      "loss": 2.688817024230957,
      "step": 7234
    },
    {
      "epoch": 0.5867022071771899,
      "grad_norm": 0.024360770359635353,
      "learning_rate": 0.0012,
      "loss": 2.689929246902466,
      "step": 7236
    },
    {
      "epoch": 0.5868643692023908,
      "grad_norm": 0.027452750131487846,
      "learning_rate": 0.0012,
      "loss": 2.623702049255371,
      "step": 7238
    },
    {
      "epoch": 0.5870265312275919,
      "grad_norm": 0.02554905228316784,
      "learning_rate": 0.0012,
      "loss": 2.6271753311157227,
      "step": 7240
    },
    {
      "epoch": 0.5871886932527929,
      "grad_norm": 0.061307914555072784,
      "learning_rate": 0.0012,
      "loss": 2.6853654384613037,
      "step": 7242
    },
    {
      "epoch": 0.5873508552779938,
      "grad_norm": 0.07868704944849014,
      "learning_rate": 0.0012,
      "loss": 2.652308464050293,
      "step": 7244
    },
    {
      "epoch": 0.5875130173031948,
      "grad_norm": 0.04762626066803932,
      "learning_rate": 0.0012,
      "loss": 2.6198625564575195,
      "step": 7246
    },
    {
      "epoch": 0.5876751793283959,
      "grad_norm": 0.02437223494052887,
      "learning_rate": 0.0012,
      "loss": 2.6607112884521484,
      "step": 7248
    },
    {
      "epoch": 0.5878373413535968,
      "grad_norm": 0.06150480732321739,
      "learning_rate": 0.0012,
      "loss": 2.651585578918457,
      "step": 7250
    },
    {
      "epoch": 0.5879995033787978,
      "grad_norm": 0.04376852512359619,
      "learning_rate": 0.0012,
      "loss": 2.7107887268066406,
      "step": 7252
    },
    {
      "epoch": 0.5881616654039988,
      "grad_norm": 0.027734970673918724,
      "learning_rate": 0.0012,
      "loss": 2.68180251121521,
      "step": 7254
    },
    {
      "epoch": 0.5883238274291998,
      "grad_norm": 0.053271204233169556,
      "learning_rate": 0.0012,
      "loss": 2.654632329940796,
      "step": 7256
    },
    {
      "epoch": 0.5884859894544008,
      "grad_norm": 0.07733266055583954,
      "learning_rate": 0.0012,
      "loss": 2.6907668113708496,
      "step": 7258
    },
    {
      "epoch": 0.5886481514796018,
      "grad_norm": 0.033175550401210785,
      "learning_rate": 0.0012,
      "loss": 2.672717571258545,
      "step": 7260
    },
    {
      "epoch": 0.5888103135048027,
      "grad_norm": 0.04872220754623413,
      "learning_rate": 0.0012,
      "loss": 2.666192054748535,
      "step": 7262
    },
    {
      "epoch": 0.5889724755300038,
      "grad_norm": 0.03841711953282356,
      "learning_rate": 0.0012,
      "loss": 2.69374942779541,
      "step": 7264
    },
    {
      "epoch": 0.5891346375552048,
      "grad_norm": 0.04571523889899254,
      "learning_rate": 0.0012,
      "loss": 2.628969192504883,
      "step": 7266
    },
    {
      "epoch": 0.5892967995804057,
      "grad_norm": 0.06161152571439743,
      "learning_rate": 0.0012,
      "loss": 2.6895766258239746,
      "step": 7268
    },
    {
      "epoch": 0.5894589616056067,
      "grad_norm": 0.027166886255145073,
      "learning_rate": 0.0012,
      "loss": 2.6316349506378174,
      "step": 7270
    },
    {
      "epoch": 0.5896211236308078,
      "grad_norm": 0.03699856624007225,
      "learning_rate": 0.0012,
      "loss": 2.7086517810821533,
      "step": 7272
    },
    {
      "epoch": 0.5897832856560087,
      "grad_norm": 0.04996020346879959,
      "learning_rate": 0.0012,
      "loss": 2.6568315029144287,
      "step": 7274
    },
    {
      "epoch": 0.5899454476812097,
      "grad_norm": 0.05045962706208229,
      "learning_rate": 0.0012,
      "loss": 2.6784768104553223,
      "step": 7276
    },
    {
      "epoch": 0.5901076097064107,
      "grad_norm": 0.02276870794594288,
      "learning_rate": 0.0012,
      "loss": 2.6743826866149902,
      "step": 7278
    },
    {
      "epoch": 0.5902697717316117,
      "grad_norm": 0.03607509285211563,
      "learning_rate": 0.0012,
      "loss": 2.6473824977874756,
      "step": 7280
    },
    {
      "epoch": 0.5904319337568127,
      "grad_norm": 0.024761322885751724,
      "learning_rate": 0.0012,
      "loss": 2.7198028564453125,
      "step": 7282
    },
    {
      "epoch": 0.5905940957820137,
      "grad_norm": 0.04541534557938576,
      "learning_rate": 0.0012,
      "loss": 2.6554737091064453,
      "step": 7284
    },
    {
      "epoch": 0.5907562578072146,
      "grad_norm": 0.04355943202972412,
      "learning_rate": 0.0012,
      "loss": 2.7436695098876953,
      "step": 7286
    },
    {
      "epoch": 0.5909184198324157,
      "grad_norm": 0.027587998658418655,
      "learning_rate": 0.0012,
      "loss": 2.59865665435791,
      "step": 7288
    },
    {
      "epoch": 0.5910805818576167,
      "grad_norm": 0.024217478930950165,
      "learning_rate": 0.0012,
      "loss": 2.67372465133667,
      "step": 7290
    },
    {
      "epoch": 0.5912427438828176,
      "grad_norm": 0.033502910286188126,
      "learning_rate": 0.0012,
      "loss": 2.6923954486846924,
      "step": 7292
    },
    {
      "epoch": 0.5914049059080186,
      "grad_norm": 0.03498483821749687,
      "learning_rate": 0.0012,
      "loss": 2.6447296142578125,
      "step": 7294
    },
    {
      "epoch": 0.5915670679332197,
      "grad_norm": 0.05157732218503952,
      "learning_rate": 0.0012,
      "loss": 2.6620640754699707,
      "step": 7296
    },
    {
      "epoch": 0.5917292299584206,
      "grad_norm": 0.05896627530455589,
      "learning_rate": 0.0012,
      "loss": 2.672266960144043,
      "step": 7298
    },
    {
      "epoch": 0.5918913919836216,
      "grad_norm": 0.028435299172997475,
      "learning_rate": 0.0012,
      "loss": 2.5645978450775146,
      "step": 7300
    },
    {
      "epoch": 0.5920535540088226,
      "grad_norm": 0.031980000436306,
      "learning_rate": 0.0012,
      "loss": 2.718092203140259,
      "step": 7302
    },
    {
      "epoch": 0.5922157160340236,
      "grad_norm": 0.04036010429263115,
      "learning_rate": 0.0012,
      "loss": 2.5811350345611572,
      "step": 7304
    },
    {
      "epoch": 0.5923778780592246,
      "grad_norm": 0.03833182528614998,
      "learning_rate": 0.0012,
      "loss": 2.6736297607421875,
      "step": 7306
    },
    {
      "epoch": 0.5925400400844256,
      "grad_norm": 0.024270856752991676,
      "learning_rate": 0.0012,
      "loss": 2.7293753623962402,
      "step": 7308
    },
    {
      "epoch": 0.5927022021096265,
      "grad_norm": 0.039729874581098557,
      "learning_rate": 0.0012,
      "loss": 2.684875011444092,
      "step": 7310
    },
    {
      "epoch": 0.5928643641348276,
      "grad_norm": 0.04064951092004776,
      "learning_rate": 0.0012,
      "loss": 2.6544625759124756,
      "step": 7312
    },
    {
      "epoch": 0.5930265261600286,
      "grad_norm": 0.03264759108424187,
      "learning_rate": 0.0012,
      "loss": 2.6695690155029297,
      "step": 7314
    },
    {
      "epoch": 0.5931886881852295,
      "grad_norm": 0.02773749642074108,
      "learning_rate": 0.0012,
      "loss": 2.683619976043701,
      "step": 7316
    },
    {
      "epoch": 0.5933508502104305,
      "grad_norm": 0.0254158116877079,
      "learning_rate": 0.0012,
      "loss": 2.710350513458252,
      "step": 7318
    },
    {
      "epoch": 0.5935130122356316,
      "grad_norm": 0.025839870795607567,
      "learning_rate": 0.0012,
      "loss": 2.664492130279541,
      "step": 7320
    },
    {
      "epoch": 0.5936751742608326,
      "grad_norm": 0.029366230592131615,
      "learning_rate": 0.0012,
      "loss": 2.6888136863708496,
      "step": 7322
    },
    {
      "epoch": 0.5938373362860335,
      "grad_norm": 0.04462215676903725,
      "learning_rate": 0.0012,
      "loss": 2.6930880546569824,
      "step": 7324
    },
    {
      "epoch": 0.5939994983112346,
      "grad_norm": 0.04077832028269768,
      "learning_rate": 0.0012,
      "loss": 2.6187803745269775,
      "step": 7326
    },
    {
      "epoch": 0.5941616603364356,
      "grad_norm": 0.04982251301407814,
      "learning_rate": 0.0012,
      "loss": 2.690774440765381,
      "step": 7328
    },
    {
      "epoch": 0.5943238223616365,
      "grad_norm": 0.04044247046113014,
      "learning_rate": 0.0012,
      "loss": 2.6377370357513428,
      "step": 7330
    },
    {
      "epoch": 0.5944859843868375,
      "grad_norm": 0.023902785032987595,
      "learning_rate": 0.0012,
      "loss": 2.6931588649749756,
      "step": 7332
    },
    {
      "epoch": 0.5946481464120386,
      "grad_norm": 0.027176758274435997,
      "learning_rate": 0.0012,
      "loss": 2.688897132873535,
      "step": 7334
    },
    {
      "epoch": 0.5948103084372395,
      "grad_norm": 0.02904612198472023,
      "learning_rate": 0.0012,
      "loss": 2.7117550373077393,
      "step": 7336
    },
    {
      "epoch": 0.5949724704624405,
      "grad_norm": 0.06109444797039032,
      "learning_rate": 0.0012,
      "loss": 2.6537466049194336,
      "step": 7338
    },
    {
      "epoch": 0.5951346324876415,
      "grad_norm": 0.05861647427082062,
      "learning_rate": 0.0012,
      "loss": 2.6942334175109863,
      "step": 7340
    },
    {
      "epoch": 0.5952967945128425,
      "grad_norm": 0.027524417266249657,
      "learning_rate": 0.0012,
      "loss": 2.6211071014404297,
      "step": 7342
    },
    {
      "epoch": 0.5954589565380435,
      "grad_norm": 0.02447892166674137,
      "learning_rate": 0.0012,
      "loss": 2.672010660171509,
      "step": 7344
    },
    {
      "epoch": 0.5956211185632445,
      "grad_norm": 0.05020981654524803,
      "learning_rate": 0.0012,
      "loss": 2.6350789070129395,
      "step": 7346
    },
    {
      "epoch": 0.5957832805884454,
      "grad_norm": 0.026602154597640038,
      "learning_rate": 0.0012,
      "loss": 2.658402919769287,
      "step": 7348
    },
    {
      "epoch": 0.5959454426136465,
      "grad_norm": 0.026792190968990326,
      "learning_rate": 0.0012,
      "loss": 2.643232822418213,
      "step": 7350
    },
    {
      "epoch": 0.5961076046388475,
      "grad_norm": 0.02716570347547531,
      "learning_rate": 0.0012,
      "loss": 2.662014961242676,
      "step": 7352
    },
    {
      "epoch": 0.5962697666640484,
      "grad_norm": 0.039728082716464996,
      "learning_rate": 0.0012,
      "loss": 2.626377582550049,
      "step": 7354
    },
    {
      "epoch": 0.5964319286892494,
      "grad_norm": 0.04630686342716217,
      "learning_rate": 0.0012,
      "loss": 2.668184280395508,
      "step": 7356
    },
    {
      "epoch": 0.5965940907144505,
      "grad_norm": 0.06244850903749466,
      "learning_rate": 0.0012,
      "loss": 2.633678913116455,
      "step": 7358
    },
    {
      "epoch": 0.5967562527396514,
      "grad_norm": 0.031039124354720116,
      "learning_rate": 0.0012,
      "loss": 2.700267791748047,
      "step": 7360
    },
    {
      "epoch": 0.5969184147648524,
      "grad_norm": 0.05244478955864906,
      "learning_rate": 0.0012,
      "loss": 2.7081410884857178,
      "step": 7362
    },
    {
      "epoch": 0.5970805767900534,
      "grad_norm": 0.057638879865407944,
      "learning_rate": 0.0012,
      "loss": 2.6516916751861572,
      "step": 7364
    },
    {
      "epoch": 0.5972427388152544,
      "grad_norm": 0.035455696284770966,
      "learning_rate": 0.0012,
      "loss": 2.7036726474761963,
      "step": 7366
    },
    {
      "epoch": 0.5974049008404554,
      "grad_norm": 0.029959596693515778,
      "learning_rate": 0.0012,
      "loss": 2.634906768798828,
      "step": 7368
    },
    {
      "epoch": 0.5975670628656564,
      "grad_norm": 0.07227431237697601,
      "learning_rate": 0.0012,
      "loss": 2.60658597946167,
      "step": 7370
    },
    {
      "epoch": 0.5977292248908573,
      "grad_norm": 0.023981191217899323,
      "learning_rate": 0.0012,
      "loss": 2.7054576873779297,
      "step": 7372
    },
    {
      "epoch": 0.5978913869160584,
      "grad_norm": 0.03552450239658356,
      "learning_rate": 0.0012,
      "loss": 2.686148166656494,
      "step": 7374
    },
    {
      "epoch": 0.5980535489412594,
      "grad_norm": 0.05105069652199745,
      "learning_rate": 0.0012,
      "loss": 2.754770517349243,
      "step": 7376
    },
    {
      "epoch": 0.5982157109664603,
      "grad_norm": 0.06257119029760361,
      "learning_rate": 0.0012,
      "loss": 2.7093870639801025,
      "step": 7378
    },
    {
      "epoch": 0.5983778729916613,
      "grad_norm": 0.029903022572398186,
      "learning_rate": 0.0012,
      "loss": 2.5836517810821533,
      "step": 7380
    },
    {
      "epoch": 0.5985400350168624,
      "grad_norm": 0.02975860983133316,
      "learning_rate": 0.0012,
      "loss": 2.736400842666626,
      "step": 7382
    },
    {
      "epoch": 0.5987021970420633,
      "grad_norm": 0.03971276804804802,
      "learning_rate": 0.0012,
      "loss": 2.616830348968506,
      "step": 7384
    },
    {
      "epoch": 0.5988643590672643,
      "grad_norm": 0.02965393476188183,
      "learning_rate": 0.0012,
      "loss": 2.6381616592407227,
      "step": 7386
    },
    {
      "epoch": 0.5990265210924653,
      "grad_norm": 0.03216595575213432,
      "learning_rate": 0.0012,
      "loss": 2.6207664012908936,
      "step": 7388
    },
    {
      "epoch": 0.5991886831176663,
      "grad_norm": 0.02466467209160328,
      "learning_rate": 0.0012,
      "loss": 2.727292537689209,
      "step": 7390
    },
    {
      "epoch": 0.5993508451428673,
      "grad_norm": 0.024088643491268158,
      "learning_rate": 0.0012,
      "loss": 2.6984317302703857,
      "step": 7392
    },
    {
      "epoch": 0.5995130071680683,
      "grad_norm": 0.02417336218059063,
      "learning_rate": 0.0012,
      "loss": 2.667461395263672,
      "step": 7394
    },
    {
      "epoch": 0.5996751691932692,
      "grad_norm": 0.04045163840055466,
      "learning_rate": 0.0012,
      "loss": 2.7011070251464844,
      "step": 7396
    },
    {
      "epoch": 0.5998373312184703,
      "grad_norm": 0.030740704387426376,
      "learning_rate": 0.0012,
      "loss": 2.7017710208892822,
      "step": 7398
    },
    {
      "epoch": 0.5999994932436713,
      "grad_norm": 0.026988454163074493,
      "learning_rate": 0.0012,
      "loss": 2.693019151687622,
      "step": 7400
    },
    {
      "epoch": 0.6001616552688722,
      "grad_norm": 0.027043882757425308,
      "learning_rate": 0.0012,
      "loss": 2.663689136505127,
      "step": 7402
    },
    {
      "epoch": 0.6003238172940732,
      "grad_norm": 0.02644982375204563,
      "learning_rate": 0.0012,
      "loss": 2.703644275665283,
      "step": 7404
    },
    {
      "epoch": 0.6004859793192743,
      "grad_norm": 0.026407748460769653,
      "learning_rate": 0.0012,
      "loss": 2.694833278656006,
      "step": 7406
    },
    {
      "epoch": 0.6006481413444752,
      "grad_norm": 0.03307900205254555,
      "learning_rate": 0.0012,
      "loss": 2.6493635177612305,
      "step": 7408
    },
    {
      "epoch": 0.6008103033696762,
      "grad_norm": 0.07540225982666016,
      "learning_rate": 0.0012,
      "loss": 2.6249899864196777,
      "step": 7410
    },
    {
      "epoch": 0.6009724653948773,
      "grad_norm": 0.06192155182361603,
      "learning_rate": 0.0012,
      "loss": 2.6196441650390625,
      "step": 7412
    },
    {
      "epoch": 0.6011346274200782,
      "grad_norm": 0.03384125232696533,
      "learning_rate": 0.0012,
      "loss": 2.6534790992736816,
      "step": 7414
    },
    {
      "epoch": 0.6012967894452792,
      "grad_norm": 0.03594216704368591,
      "learning_rate": 0.0012,
      "loss": 2.682218074798584,
      "step": 7416
    },
    {
      "epoch": 0.6014589514704802,
      "grad_norm": 0.05883806198835373,
      "learning_rate": 0.0012,
      "loss": 2.6681160926818848,
      "step": 7418
    },
    {
      "epoch": 0.6016211134956811,
      "grad_norm": 0.036412760615348816,
      "learning_rate": 0.0012,
      "loss": 2.629136562347412,
      "step": 7420
    },
    {
      "epoch": 0.6017832755208822,
      "grad_norm": 0.026067664846777916,
      "learning_rate": 0.0012,
      "loss": 2.642256736755371,
      "step": 7422
    },
    {
      "epoch": 0.6019454375460832,
      "grad_norm": 0.028370670974254608,
      "learning_rate": 0.0012,
      "loss": 2.631575584411621,
      "step": 7424
    },
    {
      "epoch": 0.6021075995712841,
      "grad_norm": 0.02732965536415577,
      "learning_rate": 0.0012,
      "loss": 2.6911091804504395,
      "step": 7426
    },
    {
      "epoch": 0.6022697615964852,
      "grad_norm": 0.02589712105691433,
      "learning_rate": 0.0012,
      "loss": 2.651817798614502,
      "step": 7428
    },
    {
      "epoch": 0.6024319236216862,
      "grad_norm": 0.028014758601784706,
      "learning_rate": 0.0012,
      "loss": 2.7430872917175293,
      "step": 7430
    },
    {
      "epoch": 0.6025940856468871,
      "grad_norm": 0.02978094294667244,
      "learning_rate": 0.0012,
      "loss": 2.614408016204834,
      "step": 7432
    },
    {
      "epoch": 0.6027562476720881,
      "grad_norm": 0.02261500433087349,
      "learning_rate": 0.0012,
      "loss": 2.6620335578918457,
      "step": 7434
    },
    {
      "epoch": 0.6029184096972892,
      "grad_norm": 0.04191069304943085,
      "learning_rate": 0.0012,
      "loss": 2.660133123397827,
      "step": 7436
    },
    {
      "epoch": 0.6030805717224901,
      "grad_norm": 0.06501773744821548,
      "learning_rate": 0.0012,
      "loss": 2.621868133544922,
      "step": 7438
    },
    {
      "epoch": 0.6032427337476911,
      "grad_norm": 0.04073552414774895,
      "learning_rate": 0.0012,
      "loss": 2.6021289825439453,
      "step": 7440
    },
    {
      "epoch": 0.6034048957728921,
      "grad_norm": 0.034508053213357925,
      "learning_rate": 0.0012,
      "loss": 2.5962018966674805,
      "step": 7442
    },
    {
      "epoch": 0.603567057798093,
      "grad_norm": 0.040614787489175797,
      "learning_rate": 0.0012,
      "loss": 2.715564250946045,
      "step": 7444
    },
    {
      "epoch": 0.6037292198232941,
      "grad_norm": 0.04112826660275459,
      "learning_rate": 0.0012,
      "loss": 2.7097010612487793,
      "step": 7446
    },
    {
      "epoch": 0.6038913818484951,
      "grad_norm": 0.02365444228053093,
      "learning_rate": 0.0012,
      "loss": 2.7151546478271484,
      "step": 7448
    },
    {
      "epoch": 0.604053543873696,
      "grad_norm": 0.026429342105984688,
      "learning_rate": 0.0012,
      "loss": 2.7360777854919434,
      "step": 7450
    },
    {
      "epoch": 0.6042157058988971,
      "grad_norm": 0.03351885825395584,
      "learning_rate": 0.0012,
      "loss": 2.7190966606140137,
      "step": 7452
    },
    {
      "epoch": 0.6043778679240981,
      "grad_norm": 0.04416413977742195,
      "learning_rate": 0.0012,
      "loss": 2.7010700702667236,
      "step": 7454
    },
    {
      "epoch": 0.604540029949299,
      "grad_norm": 0.0437028631567955,
      "learning_rate": 0.0012,
      "loss": 2.6415629386901855,
      "step": 7456
    },
    {
      "epoch": 0.6047021919745,
      "grad_norm": 0.03386016562581062,
      "learning_rate": 0.0012,
      "loss": 2.6253113746643066,
      "step": 7458
    },
    {
      "epoch": 0.6048643539997011,
      "grad_norm": 0.03438775613903999,
      "learning_rate": 0.0012,
      "loss": 2.684262275695801,
      "step": 7460
    },
    {
      "epoch": 0.605026516024902,
      "grad_norm": 0.03919399157166481,
      "learning_rate": 0.0012,
      "loss": 2.680710792541504,
      "step": 7462
    },
    {
      "epoch": 0.605188678050103,
      "grad_norm": 0.02937966212630272,
      "learning_rate": 0.0012,
      "loss": 2.62863826751709,
      "step": 7464
    },
    {
      "epoch": 0.605350840075304,
      "grad_norm": 0.0234469473361969,
      "learning_rate": 0.0012,
      "loss": 2.681255340576172,
      "step": 7466
    },
    {
      "epoch": 0.605513002100505,
      "grad_norm": 0.03081965260207653,
      "learning_rate": 0.0012,
      "loss": 2.6170053482055664,
      "step": 7468
    },
    {
      "epoch": 0.605675164125706,
      "grad_norm": 0.02969764731824398,
      "learning_rate": 0.0012,
      "loss": 2.6705358028411865,
      "step": 7470
    },
    {
      "epoch": 0.605837326150907,
      "grad_norm": 0.025745363906025887,
      "learning_rate": 0.0012,
      "loss": 2.6436893939971924,
      "step": 7472
    },
    {
      "epoch": 0.6059994881761079,
      "grad_norm": 0.024698331952095032,
      "learning_rate": 0.0012,
      "loss": 2.678773880004883,
      "step": 7474
    },
    {
      "epoch": 0.606161650201309,
      "grad_norm": 0.025299591943621635,
      "learning_rate": 0.0012,
      "loss": 2.661860704421997,
      "step": 7476
    },
    {
      "epoch": 0.60632381222651,
      "grad_norm": 0.031097179278731346,
      "learning_rate": 0.0012,
      "loss": 2.637317657470703,
      "step": 7478
    },
    {
      "epoch": 0.6064859742517109,
      "grad_norm": 0.041172485798597336,
      "learning_rate": 0.0012,
      "loss": 2.6795411109924316,
      "step": 7480
    },
    {
      "epoch": 0.6066481362769119,
      "grad_norm": 0.02889208123087883,
      "learning_rate": 0.0012,
      "loss": 2.670262336730957,
      "step": 7482
    },
    {
      "epoch": 0.606810298302113,
      "grad_norm": 0.02629091776907444,
      "learning_rate": 0.0012,
      "loss": 2.6003477573394775,
      "step": 7484
    },
    {
      "epoch": 0.6069724603273139,
      "grad_norm": 0.03312195464968681,
      "learning_rate": 0.0012,
      "loss": 2.636974811553955,
      "step": 7486
    },
    {
      "epoch": 0.6071346223525149,
      "grad_norm": 0.03490237891674042,
      "learning_rate": 0.0012,
      "loss": 2.698397636413574,
      "step": 7488
    },
    {
      "epoch": 0.607296784377716,
      "grad_norm": 0.026202430948615074,
      "learning_rate": 0.0012,
      "loss": 2.692392349243164,
      "step": 7490
    },
    {
      "epoch": 0.6074589464029169,
      "grad_norm": 0.030183231458067894,
      "learning_rate": 0.0012,
      "loss": 2.6695516109466553,
      "step": 7492
    },
    {
      "epoch": 0.6076211084281179,
      "grad_norm": 0.06922779232263565,
      "learning_rate": 0.0012,
      "loss": 2.6556222438812256,
      "step": 7494
    },
    {
      "epoch": 0.6077832704533189,
      "grad_norm": 0.07168378680944443,
      "learning_rate": 0.0012,
      "loss": 2.658874750137329,
      "step": 7496
    },
    {
      "epoch": 0.6079454324785198,
      "grad_norm": 0.03630850836634636,
      "learning_rate": 0.0012,
      "loss": 2.600267171859741,
      "step": 7498
    },
    {
      "epoch": 0.6081075945037209,
      "grad_norm": 0.04770057648420334,
      "learning_rate": 0.0012,
      "loss": 2.661954402923584,
      "step": 7500
    },
    {
      "epoch": 0.6082697565289219,
      "grad_norm": 0.06880483776330948,
      "learning_rate": 0.0012,
      "loss": 2.6805005073547363,
      "step": 7502
    },
    {
      "epoch": 0.6084319185541228,
      "grad_norm": 0.02565770410001278,
      "learning_rate": 0.0012,
      "loss": 2.628715991973877,
      "step": 7504
    },
    {
      "epoch": 0.6085940805793238,
      "grad_norm": 0.044690992683172226,
      "learning_rate": 0.0012,
      "loss": 2.6669178009033203,
      "step": 7506
    },
    {
      "epoch": 0.6087562426045249,
      "grad_norm": 0.03871871158480644,
      "learning_rate": 0.0012,
      "loss": 2.6584627628326416,
      "step": 7508
    },
    {
      "epoch": 0.6089184046297258,
      "grad_norm": 0.04060292989015579,
      "learning_rate": 0.0012,
      "loss": 2.6283764839172363,
      "step": 7510
    },
    {
      "epoch": 0.6090805666549268,
      "grad_norm": 0.053236402571201324,
      "learning_rate": 0.0012,
      "loss": 2.6693098545074463,
      "step": 7512
    },
    {
      "epoch": 0.6092427286801279,
      "grad_norm": 0.037442512810230255,
      "learning_rate": 0.0012,
      "loss": 2.6087136268615723,
      "step": 7514
    },
    {
      "epoch": 0.6094048907053288,
      "grad_norm": 0.025856249034404755,
      "learning_rate": 0.0012,
      "loss": 2.6575474739074707,
      "step": 7516
    },
    {
      "epoch": 0.6095670527305298,
      "grad_norm": 0.03905525431036949,
      "learning_rate": 0.0012,
      "loss": 2.689236640930176,
      "step": 7518
    },
    {
      "epoch": 0.6097292147557308,
      "grad_norm": 0.030110551044344902,
      "learning_rate": 0.0012,
      "loss": 2.70747709274292,
      "step": 7520
    },
    {
      "epoch": 0.6098913767809317,
      "grad_norm": 0.02453673630952835,
      "learning_rate": 0.0012,
      "loss": 2.6340436935424805,
      "step": 7522
    },
    {
      "epoch": 0.6100535388061328,
      "grad_norm": 0.04741143062710762,
      "learning_rate": 0.0012,
      "loss": 2.5898334980010986,
      "step": 7524
    },
    {
      "epoch": 0.6102157008313338,
      "grad_norm": 0.029748737812042236,
      "learning_rate": 0.0012,
      "loss": 2.6448354721069336,
      "step": 7526
    },
    {
      "epoch": 0.6103778628565347,
      "grad_norm": 0.06667052954435349,
      "learning_rate": 0.0012,
      "loss": 2.7061538696289062,
      "step": 7528
    },
    {
      "epoch": 0.6105400248817358,
      "grad_norm": 0.035178206861019135,
      "learning_rate": 0.0012,
      "loss": 2.624929428100586,
      "step": 7530
    },
    {
      "epoch": 0.6107021869069368,
      "grad_norm": 0.025145133957266808,
      "learning_rate": 0.0012,
      "loss": 2.700529098510742,
      "step": 7532
    },
    {
      "epoch": 0.6108643489321377,
      "grad_norm": 0.05587422475218773,
      "learning_rate": 0.0012,
      "loss": 2.7359347343444824,
      "step": 7534
    },
    {
      "epoch": 0.6110265109573387,
      "grad_norm": 0.05961275473237038,
      "learning_rate": 0.0012,
      "loss": 2.62239933013916,
      "step": 7536
    },
    {
      "epoch": 0.6111886729825398,
      "grad_norm": 0.025349846109747887,
      "learning_rate": 0.0012,
      "loss": 2.627552032470703,
      "step": 7538
    },
    {
      "epoch": 0.6113508350077407,
      "grad_norm": 0.05931536480784416,
      "learning_rate": 0.0012,
      "loss": 2.6659467220306396,
      "step": 7540
    },
    {
      "epoch": 0.6115129970329417,
      "grad_norm": 0.05746973678469658,
      "learning_rate": 0.0012,
      "loss": 2.668863296508789,
      "step": 7542
    },
    {
      "epoch": 0.6116751590581427,
      "grad_norm": 0.0416809506714344,
      "learning_rate": 0.0012,
      "loss": 2.713031768798828,
      "step": 7544
    },
    {
      "epoch": 0.6118373210833437,
      "grad_norm": 0.04156620427966118,
      "learning_rate": 0.0012,
      "loss": 2.5731143951416016,
      "step": 7546
    },
    {
      "epoch": 0.6119994831085447,
      "grad_norm": 0.05237229913473129,
      "learning_rate": 0.0012,
      "loss": 2.723766803741455,
      "step": 7548
    },
    {
      "epoch": 0.6121616451337457,
      "grad_norm": 0.029249761253595352,
      "learning_rate": 0.0012,
      "loss": 2.6871533393859863,
      "step": 7550
    },
    {
      "epoch": 0.6123238071589466,
      "grad_norm": 0.03494785726070404,
      "learning_rate": 0.0012,
      "loss": 2.7176575660705566,
      "step": 7552
    },
    {
      "epoch": 0.6124859691841477,
      "grad_norm": 0.03381054848432541,
      "learning_rate": 0.0012,
      "loss": 2.6444735527038574,
      "step": 7554
    },
    {
      "epoch": 0.6126481312093487,
      "grad_norm": 0.04141880199313164,
      "learning_rate": 0.0012,
      "loss": 2.6377816200256348,
      "step": 7556
    },
    {
      "epoch": 0.6128102932345496,
      "grad_norm": 0.06089909002184868,
      "learning_rate": 0.0012,
      "loss": 2.650219440460205,
      "step": 7558
    },
    {
      "epoch": 0.6129724552597506,
      "grad_norm": 0.025298580527305603,
      "learning_rate": 0.0012,
      "loss": 2.6815171241760254,
      "step": 7560
    },
    {
      "epoch": 0.6131346172849517,
      "grad_norm": 0.046990204602479935,
      "learning_rate": 0.0012,
      "loss": 2.6070027351379395,
      "step": 7562
    },
    {
      "epoch": 0.6132967793101526,
      "grad_norm": 0.043039847165346146,
      "learning_rate": 0.0012,
      "loss": 2.649355888366699,
      "step": 7564
    },
    {
      "epoch": 0.6134589413353536,
      "grad_norm": 0.03228619322180748,
      "learning_rate": 0.0012,
      "loss": 2.6563568115234375,
      "step": 7566
    },
    {
      "epoch": 0.6136211033605546,
      "grad_norm": 0.07343880087137222,
      "learning_rate": 0.0012,
      "loss": 2.6473259925842285,
      "step": 7568
    },
    {
      "epoch": 0.6137832653857556,
      "grad_norm": 0.033858247101306915,
      "learning_rate": 0.0012,
      "loss": 2.5872697830200195,
      "step": 7570
    },
    {
      "epoch": 0.6139454274109566,
      "grad_norm": 0.024823574349284172,
      "learning_rate": 0.0012,
      "loss": 2.677908420562744,
      "step": 7572
    },
    {
      "epoch": 0.6141075894361576,
      "grad_norm": 0.03478936478495598,
      "learning_rate": 0.0012,
      "loss": 2.654458522796631,
      "step": 7574
    },
    {
      "epoch": 0.6142697514613585,
      "grad_norm": 0.048874903470277786,
      "learning_rate": 0.0012,
      "loss": 2.7312097549438477,
      "step": 7576
    },
    {
      "epoch": 0.6144319134865596,
      "grad_norm": 0.042779941111803055,
      "learning_rate": 0.0012,
      "loss": 2.651949405670166,
      "step": 7578
    },
    {
      "epoch": 0.6145940755117606,
      "grad_norm": 0.029054567217826843,
      "learning_rate": 0.0012,
      "loss": 2.6671652793884277,
      "step": 7580
    },
    {
      "epoch": 0.6147562375369615,
      "grad_norm": 0.03201567381620407,
      "learning_rate": 0.0012,
      "loss": 2.744020938873291,
      "step": 7582
    },
    {
      "epoch": 0.6149183995621625,
      "grad_norm": 0.028705798089504242,
      "learning_rate": 0.0012,
      "loss": 2.6697332859039307,
      "step": 7584
    },
    {
      "epoch": 0.6150805615873636,
      "grad_norm": 0.03609717637300491,
      "learning_rate": 0.0012,
      "loss": 2.695713520050049,
      "step": 7586
    },
    {
      "epoch": 0.6152427236125645,
      "grad_norm": 0.028048187494277954,
      "learning_rate": 0.0012,
      "loss": 2.6480183601379395,
      "step": 7588
    },
    {
      "epoch": 0.6154048856377655,
      "grad_norm": 0.026618199422955513,
      "learning_rate": 0.0012,
      "loss": 2.6912477016448975,
      "step": 7590
    },
    {
      "epoch": 0.6155670476629665,
      "grad_norm": 0.05040533468127251,
      "learning_rate": 0.0012,
      "loss": 2.67608642578125,
      "step": 7592
    },
    {
      "epoch": 0.6157292096881675,
      "grad_norm": 0.04510386660695076,
      "learning_rate": 0.0012,
      "loss": 2.665748119354248,
      "step": 7594
    },
    {
      "epoch": 0.6158913717133685,
      "grad_norm": 0.028724195435643196,
      "learning_rate": 0.0012,
      "loss": 2.6662282943725586,
      "step": 7596
    },
    {
      "epoch": 0.6160535337385695,
      "grad_norm": 0.025700286030769348,
      "learning_rate": 0.0012,
      "loss": 2.660219192504883,
      "step": 7598
    },
    {
      "epoch": 0.6162156957637704,
      "grad_norm": 0.02998029999434948,
      "learning_rate": 0.0012,
      "loss": 2.578090190887451,
      "step": 7600
    },
    {
      "epoch": 0.6163778577889715,
      "grad_norm": 0.03009391389787197,
      "learning_rate": 0.0012,
      "loss": 2.604916572570801,
      "step": 7602
    },
    {
      "epoch": 0.6165400198141725,
      "grad_norm": 0.025565704330801964,
      "learning_rate": 0.0012,
      "loss": 2.693929672241211,
      "step": 7604
    },
    {
      "epoch": 0.6167021818393734,
      "grad_norm": 0.026402028277516365,
      "learning_rate": 0.0012,
      "loss": 2.6257686614990234,
      "step": 7606
    },
    {
      "epoch": 0.6168643438645744,
      "grad_norm": 0.042027249932289124,
      "learning_rate": 0.0012,
      "loss": 2.615142345428467,
      "step": 7608
    },
    {
      "epoch": 0.6170265058897755,
      "grad_norm": 0.024231353774666786,
      "learning_rate": 0.0012,
      "loss": 2.6631383895874023,
      "step": 7610
    },
    {
      "epoch": 0.6171886679149764,
      "grad_norm": 0.04036235064268112,
      "learning_rate": 0.0012,
      "loss": 2.6365203857421875,
      "step": 7612
    },
    {
      "epoch": 0.6173508299401774,
      "grad_norm": 0.049406472593545914,
      "learning_rate": 0.0012,
      "loss": 2.6917808055877686,
      "step": 7614
    },
    {
      "epoch": 0.6175129919653785,
      "grad_norm": 0.05757250264286995,
      "learning_rate": 0.0012,
      "loss": 2.7212963104248047,
      "step": 7616
    },
    {
      "epoch": 0.6176751539905794,
      "grad_norm": 0.04073324054479599,
      "learning_rate": 0.0012,
      "loss": 2.730930805206299,
      "step": 7618
    },
    {
      "epoch": 0.6178373160157804,
      "grad_norm": 0.02463964931666851,
      "learning_rate": 0.0012,
      "loss": 2.6627426147460938,
      "step": 7620
    },
    {
      "epoch": 0.6179994780409814,
      "grad_norm": 0.06722128391265869,
      "learning_rate": 0.0012,
      "loss": 2.648538112640381,
      "step": 7622
    },
    {
      "epoch": 0.6181616400661823,
      "grad_norm": 0.0711342990398407,
      "learning_rate": 0.0012,
      "loss": 2.655177116394043,
      "step": 7624
    },
    {
      "epoch": 0.6183238020913834,
      "grad_norm": 0.029694750905036926,
      "learning_rate": 0.0012,
      "loss": 2.668825626373291,
      "step": 7626
    },
    {
      "epoch": 0.6184859641165844,
      "grad_norm": 0.034533217549324036,
      "learning_rate": 0.0012,
      "loss": 2.7320146560668945,
      "step": 7628
    },
    {
      "epoch": 0.6186481261417853,
      "grad_norm": 0.04802347347140312,
      "learning_rate": 0.0012,
      "loss": 2.645716667175293,
      "step": 7630
    },
    {
      "epoch": 0.6188102881669864,
      "grad_norm": 0.0837225615978241,
      "learning_rate": 0.0012,
      "loss": 2.619846820831299,
      "step": 7632
    },
    {
      "epoch": 0.6189724501921874,
      "grad_norm": 0.04412268474698067,
      "learning_rate": 0.0012,
      "loss": 2.6346445083618164,
      "step": 7634
    },
    {
      "epoch": 0.6191346122173883,
      "grad_norm": 0.028287675231695175,
      "learning_rate": 0.0012,
      "loss": 2.713472366333008,
      "step": 7636
    },
    {
      "epoch": 0.6192967742425893,
      "grad_norm": 0.035305194556713104,
      "learning_rate": 0.0012,
      "loss": 2.702327251434326,
      "step": 7638
    },
    {
      "epoch": 0.6194589362677904,
      "grad_norm": 0.03282155841588974,
      "learning_rate": 0.0012,
      "loss": 2.6280877590179443,
      "step": 7640
    },
    {
      "epoch": 0.6196210982929913,
      "grad_norm": 0.026242639869451523,
      "learning_rate": 0.0012,
      "loss": 2.646244764328003,
      "step": 7642
    },
    {
      "epoch": 0.6197832603181923,
      "grad_norm": 0.031517863273620605,
      "learning_rate": 0.0012,
      "loss": 2.73921275138855,
      "step": 7644
    },
    {
      "epoch": 0.6199454223433933,
      "grad_norm": 0.0328710712492466,
      "learning_rate": 0.0012,
      "loss": 2.6893749237060547,
      "step": 7646
    },
    {
      "epoch": 0.6201075843685943,
      "grad_norm": 0.02743251621723175,
      "learning_rate": 0.0012,
      "loss": 2.6485228538513184,
      "step": 7648
    },
    {
      "epoch": 0.6202697463937953,
      "grad_norm": 0.045533549040555954,
      "learning_rate": 0.0012,
      "loss": 2.6609010696411133,
      "step": 7650
    },
    {
      "epoch": 0.6204319084189963,
      "grad_norm": 0.05240781977772713,
      "learning_rate": 0.0012,
      "loss": 2.7439231872558594,
      "step": 7652
    },
    {
      "epoch": 0.6205940704441972,
      "grad_norm": 0.0299599077552557,
      "learning_rate": 0.0012,
      "loss": 2.662640333175659,
      "step": 7654
    },
    {
      "epoch": 0.6207562324693983,
      "grad_norm": 0.02379630133509636,
      "learning_rate": 0.0012,
      "loss": 2.745495319366455,
      "step": 7656
    },
    {
      "epoch": 0.6209183944945993,
      "grad_norm": 0.040373045951128006,
      "learning_rate": 0.0012,
      "loss": 2.664515733718872,
      "step": 7658
    },
    {
      "epoch": 0.6210805565198002,
      "grad_norm": 0.0489702932536602,
      "learning_rate": 0.0012,
      "loss": 2.69944429397583,
      "step": 7660
    },
    {
      "epoch": 0.6212427185450012,
      "grad_norm": 0.03250596672296524,
      "learning_rate": 0.0012,
      "loss": 2.7115776538848877,
      "step": 7662
    },
    {
      "epoch": 0.6214048805702023,
      "grad_norm": 0.037134360522031784,
      "learning_rate": 0.0012,
      "loss": 2.7136502265930176,
      "step": 7664
    },
    {
      "epoch": 0.6215670425954032,
      "grad_norm": 0.04723097383975983,
      "learning_rate": 0.0012,
      "loss": 2.6603312492370605,
      "step": 7666
    },
    {
      "epoch": 0.6217292046206042,
      "grad_norm": 0.04208579659461975,
      "learning_rate": 0.0012,
      "loss": 2.7040810585021973,
      "step": 7668
    },
    {
      "epoch": 0.6218913666458052,
      "grad_norm": 0.027709675952792168,
      "learning_rate": 0.0012,
      "loss": 2.675142765045166,
      "step": 7670
    },
    {
      "epoch": 0.6220535286710062,
      "grad_norm": 0.027134334668517113,
      "learning_rate": 0.0012,
      "loss": 2.6581039428710938,
      "step": 7672
    },
    {
      "epoch": 0.6222156906962072,
      "grad_norm": 0.028401115909218788,
      "learning_rate": 0.0012,
      "loss": 2.6830625534057617,
      "step": 7674
    },
    {
      "epoch": 0.6223778527214082,
      "grad_norm": 0.024822333827614784,
      "learning_rate": 0.0012,
      "loss": 2.6317920684814453,
      "step": 7676
    },
    {
      "epoch": 0.6225400147466091,
      "grad_norm": 0.025808988139033318,
      "learning_rate": 0.0012,
      "loss": 2.6947481632232666,
      "step": 7678
    },
    {
      "epoch": 0.6227021767718102,
      "grad_norm": 0.04262221232056618,
      "learning_rate": 0.0012,
      "loss": 2.6120858192443848,
      "step": 7680
    },
    {
      "epoch": 0.6228643387970112,
      "grad_norm": 0.05925511196255684,
      "learning_rate": 0.0012,
      "loss": 2.6905341148376465,
      "step": 7682
    },
    {
      "epoch": 0.6230265008222121,
      "grad_norm": 0.04595748707652092,
      "learning_rate": 0.0012,
      "loss": 2.69940447807312,
      "step": 7684
    },
    {
      "epoch": 0.6231886628474131,
      "grad_norm": 0.024853842332959175,
      "learning_rate": 0.0012,
      "loss": 2.671476364135742,
      "step": 7686
    },
    {
      "epoch": 0.6233508248726142,
      "grad_norm": 0.02791244350373745,
      "learning_rate": 0.0012,
      "loss": 2.6561405658721924,
      "step": 7688
    },
    {
      "epoch": 0.6235129868978151,
      "grad_norm": 0.039697807282209396,
      "learning_rate": 0.0012,
      "loss": 2.707806348800659,
      "step": 7690
    },
    {
      "epoch": 0.6236751489230161,
      "grad_norm": 0.04279274120926857,
      "learning_rate": 0.0012,
      "loss": 2.6667466163635254,
      "step": 7692
    },
    {
      "epoch": 0.6238373109482171,
      "grad_norm": 0.03613406419754028,
      "learning_rate": 0.0012,
      "loss": 2.719198226928711,
      "step": 7694
    },
    {
      "epoch": 0.6239994729734181,
      "grad_norm": 0.040611058473587036,
      "learning_rate": 0.0012,
      "loss": 2.646146059036255,
      "step": 7696
    },
    {
      "epoch": 0.6241616349986191,
      "grad_norm": 0.03546816483139992,
      "learning_rate": 0.0012,
      "loss": 2.7146904468536377,
      "step": 7698
    },
    {
      "epoch": 0.6243237970238201,
      "grad_norm": 0.02941977046430111,
      "learning_rate": 0.0012,
      "loss": 2.686943769454956,
      "step": 7700
    },
    {
      "epoch": 0.624485959049021,
      "grad_norm": 0.02362479269504547,
      "learning_rate": 0.0012,
      "loss": 2.7175827026367188,
      "step": 7702
    },
    {
      "epoch": 0.6246481210742221,
      "grad_norm": 0.0302107036113739,
      "learning_rate": 0.0012,
      "loss": 2.771603584289551,
      "step": 7704
    },
    {
      "epoch": 0.6248102830994231,
      "grad_norm": 0.025453848764300346,
      "learning_rate": 0.0012,
      "loss": 2.737955093383789,
      "step": 7706
    },
    {
      "epoch": 0.624972445124624,
      "grad_norm": 0.029418861493468285,
      "learning_rate": 0.0012,
      "loss": 2.6444716453552246,
      "step": 7708
    },
    {
      "epoch": 0.625134607149825,
      "grad_norm": 0.023260358721017838,
      "learning_rate": 0.0012,
      "loss": 2.605001926422119,
      "step": 7710
    },
    {
      "epoch": 0.6252967691750261,
      "grad_norm": 0.027020908892154694,
      "learning_rate": 0.0012,
      "loss": 2.6665914058685303,
      "step": 7712
    },
    {
      "epoch": 0.625458931200227,
      "grad_norm": 0.024948514997959137,
      "learning_rate": 0.0012,
      "loss": 2.716789722442627,
      "step": 7714
    },
    {
      "epoch": 0.625621093225428,
      "grad_norm": 0.03216419368982315,
      "learning_rate": 0.0012,
      "loss": 2.7589032649993896,
      "step": 7716
    },
    {
      "epoch": 0.6257832552506291,
      "grad_norm": 0.024186275899410248,
      "learning_rate": 0.0012,
      "loss": 2.691920518875122,
      "step": 7718
    },
    {
      "epoch": 0.62594541727583,
      "grad_norm": 0.025984864681959152,
      "learning_rate": 0.0012,
      "loss": 2.779952049255371,
      "step": 7720
    },
    {
      "epoch": 0.626107579301031,
      "grad_norm": 0.04169992357492447,
      "learning_rate": 0.0012,
      "loss": 2.678285598754883,
      "step": 7722
    },
    {
      "epoch": 0.626269741326232,
      "grad_norm": 0.05631319433450699,
      "learning_rate": 0.0012,
      "loss": 2.5774221420288086,
      "step": 7724
    },
    {
      "epoch": 0.626431903351433,
      "grad_norm": 0.022865507751703262,
      "learning_rate": 0.0012,
      "loss": 2.6893043518066406,
      "step": 7726
    },
    {
      "epoch": 0.626594065376634,
      "grad_norm": 0.032767221331596375,
      "learning_rate": 0.0012,
      "loss": 2.7019553184509277,
      "step": 7728
    },
    {
      "epoch": 0.626756227401835,
      "grad_norm": 0.040474168956279755,
      "learning_rate": 0.0012,
      "loss": 2.7295572757720947,
      "step": 7730
    },
    {
      "epoch": 0.6269183894270359,
      "grad_norm": 0.043459899723529816,
      "learning_rate": 0.0012,
      "loss": 2.6793041229248047,
      "step": 7732
    },
    {
      "epoch": 0.627080551452237,
      "grad_norm": 0.06483909487724304,
      "learning_rate": 0.0012,
      "loss": 2.704470634460449,
      "step": 7734
    },
    {
      "epoch": 0.627242713477438,
      "grad_norm": 0.036590542644262314,
      "learning_rate": 0.0012,
      "loss": 2.627566337585449,
      "step": 7736
    },
    {
      "epoch": 0.6274048755026389,
      "grad_norm": 0.052700940519571304,
      "learning_rate": 0.0012,
      "loss": 2.6984519958496094,
      "step": 7738
    },
    {
      "epoch": 0.6275670375278399,
      "grad_norm": 0.04620101675391197,
      "learning_rate": 0.0012,
      "loss": 2.6526575088500977,
      "step": 7740
    },
    {
      "epoch": 0.627729199553041,
      "grad_norm": 0.05038208141922951,
      "learning_rate": 0.0012,
      "loss": 2.7047533988952637,
      "step": 7742
    },
    {
      "epoch": 0.6278913615782419,
      "grad_norm": 0.04213647171854973,
      "learning_rate": 0.0012,
      "loss": 2.6716341972351074,
      "step": 7744
    },
    {
      "epoch": 0.6280535236034429,
      "grad_norm": 0.025445841252803802,
      "learning_rate": 0.0012,
      "loss": 2.646305799484253,
      "step": 7746
    },
    {
      "epoch": 0.6282156856286439,
      "grad_norm": 0.028074700385332108,
      "learning_rate": 0.0012,
      "loss": 2.6259326934814453,
      "step": 7748
    },
    {
      "epoch": 0.6283778476538449,
      "grad_norm": 0.042534299194812775,
      "learning_rate": 0.0012,
      "loss": 2.645380973815918,
      "step": 7750
    },
    {
      "epoch": 0.6285400096790459,
      "grad_norm": 0.033822160214185715,
      "learning_rate": 0.0012,
      "loss": 2.714808464050293,
      "step": 7752
    },
    {
      "epoch": 0.6287021717042469,
      "grad_norm": 0.03320557251572609,
      "learning_rate": 0.0012,
      "loss": 2.668668746948242,
      "step": 7754
    },
    {
      "epoch": 0.6288643337294478,
      "grad_norm": 0.03324955329298973,
      "learning_rate": 0.0012,
      "loss": 2.68684720993042,
      "step": 7756
    },
    {
      "epoch": 0.6290264957546489,
      "grad_norm": 0.02373148687183857,
      "learning_rate": 0.0012,
      "loss": 2.6352810859680176,
      "step": 7758
    },
    {
      "epoch": 0.6291886577798499,
      "grad_norm": 0.027509452775120735,
      "learning_rate": 0.0012,
      "loss": 2.639005661010742,
      "step": 7760
    },
    {
      "epoch": 0.6293508198050508,
      "grad_norm": 0.024674473330378532,
      "learning_rate": 0.0012,
      "loss": 2.637528896331787,
      "step": 7762
    },
    {
      "epoch": 0.6295129818302518,
      "grad_norm": 0.028433866798877716,
      "learning_rate": 0.0012,
      "loss": 2.676894426345825,
      "step": 7764
    },
    {
      "epoch": 0.6296751438554529,
      "grad_norm": 0.024299124255776405,
      "learning_rate": 0.0012,
      "loss": 2.6516475677490234,
      "step": 7766
    },
    {
      "epoch": 0.6298373058806538,
      "grad_norm": 0.04310794919729233,
      "learning_rate": 0.0012,
      "loss": 2.70501971244812,
      "step": 7768
    },
    {
      "epoch": 0.6299994679058548,
      "grad_norm": 0.03142084181308746,
      "learning_rate": 0.0012,
      "loss": 2.661600112915039,
      "step": 7770
    },
    {
      "epoch": 0.6301616299310558,
      "grad_norm": 0.03380948677659035,
      "learning_rate": 0.0012,
      "loss": 2.6937313079833984,
      "step": 7772
    },
    {
      "epoch": 0.6303237919562568,
      "grad_norm": 0.03631936386227608,
      "learning_rate": 0.0012,
      "loss": 2.6648483276367188,
      "step": 7774
    },
    {
      "epoch": 0.6304859539814578,
      "grad_norm": 0.027466367930173874,
      "learning_rate": 0.0012,
      "loss": 2.6861653327941895,
      "step": 7776
    },
    {
      "epoch": 0.6306481160066588,
      "grad_norm": 0.04524189978837967,
      "learning_rate": 0.0012,
      "loss": 2.7415616512298584,
      "step": 7778
    },
    {
      "epoch": 0.6308102780318597,
      "grad_norm": 0.040024254471063614,
      "learning_rate": 0.0012,
      "loss": 2.666381359100342,
      "step": 7780
    },
    {
      "epoch": 0.6309724400570608,
      "grad_norm": 0.04405586048960686,
      "learning_rate": 0.0012,
      "loss": 2.6007699966430664,
      "step": 7782
    },
    {
      "epoch": 0.6311346020822618,
      "grad_norm": 0.045017343014478683,
      "learning_rate": 0.0012,
      "loss": 2.6793861389160156,
      "step": 7784
    },
    {
      "epoch": 0.6312967641074627,
      "grad_norm": 0.028595687821507454,
      "learning_rate": 0.0012,
      "loss": 2.692370891571045,
      "step": 7786
    },
    {
      "epoch": 0.6314589261326637,
      "grad_norm": 0.05150289088487625,
      "learning_rate": 0.0012,
      "loss": 2.6904706954956055,
      "step": 7788
    },
    {
      "epoch": 0.6316210881578648,
      "grad_norm": 0.0592479333281517,
      "learning_rate": 0.0012,
      "loss": 2.594191789627075,
      "step": 7790
    },
    {
      "epoch": 0.6317832501830657,
      "grad_norm": 0.05473474785685539,
      "learning_rate": 0.0012,
      "loss": 2.6687355041503906,
      "step": 7792
    },
    {
      "epoch": 0.6319454122082667,
      "grad_norm": 0.026114389300346375,
      "learning_rate": 0.0012,
      "loss": 2.6644256114959717,
      "step": 7794
    },
    {
      "epoch": 0.6321075742334678,
      "grad_norm": 0.024659402668476105,
      "learning_rate": 0.0012,
      "loss": 2.6728224754333496,
      "step": 7796
    },
    {
      "epoch": 0.6322697362586687,
      "grad_norm": 0.024026887491345406,
      "learning_rate": 0.0012,
      "loss": 2.6955788135528564,
      "step": 7798
    },
    {
      "epoch": 0.6324318982838697,
      "grad_norm": 0.028630884364247322,
      "learning_rate": 0.0012,
      "loss": 2.6537768840789795,
      "step": 7800
    },
    {
      "epoch": 0.6325940603090707,
      "grad_norm": 0.023907126858830452,
      "learning_rate": 0.0012,
      "loss": 2.6654887199401855,
      "step": 7802
    },
    {
      "epoch": 0.6327562223342716,
      "grad_norm": 0.025802256539463997,
      "learning_rate": 0.0012,
      "loss": 2.662257194519043,
      "step": 7804
    },
    {
      "epoch": 0.6329183843594727,
      "grad_norm": 0.04391759634017944,
      "learning_rate": 0.0012,
      "loss": 2.677668571472168,
      "step": 7806
    },
    {
      "epoch": 0.6330805463846737,
      "grad_norm": 0.036363184452056885,
      "learning_rate": 0.0012,
      "loss": 2.717437505722046,
      "step": 7808
    },
    {
      "epoch": 0.6332427084098746,
      "grad_norm": 0.06255286186933517,
      "learning_rate": 0.0012,
      "loss": 2.6601548194885254,
      "step": 7810
    },
    {
      "epoch": 0.6334048704350757,
      "grad_norm": 0.06387349218130112,
      "learning_rate": 0.0012,
      "loss": 2.65065860748291,
      "step": 7812
    },
    {
      "epoch": 0.6335670324602767,
      "grad_norm": 0.04958900064229965,
      "learning_rate": 0.0012,
      "loss": 2.6600561141967773,
      "step": 7814
    },
    {
      "epoch": 0.6337291944854776,
      "grad_norm": 0.024784788489341736,
      "learning_rate": 0.0012,
      "loss": 2.6197633743286133,
      "step": 7816
    },
    {
      "epoch": 0.6338913565106786,
      "grad_norm": 0.029968777671456337,
      "learning_rate": 0.0012,
      "loss": 2.700713872909546,
      "step": 7818
    },
    {
      "epoch": 0.6340535185358797,
      "grad_norm": 0.025230051949620247,
      "learning_rate": 0.0012,
      "loss": 2.650078296661377,
      "step": 7820
    },
    {
      "epoch": 0.6342156805610806,
      "grad_norm": 0.02557825669646263,
      "learning_rate": 0.0012,
      "loss": 2.6151015758514404,
      "step": 7822
    },
    {
      "epoch": 0.6343778425862816,
      "grad_norm": 0.031053451821208,
      "learning_rate": 0.0012,
      "loss": 2.718113899230957,
      "step": 7824
    },
    {
      "epoch": 0.6345400046114826,
      "grad_norm": 0.03452207148075104,
      "learning_rate": 0.0012,
      "loss": 2.6014628410339355,
      "step": 7826
    },
    {
      "epoch": 0.6347021666366836,
      "grad_norm": 0.026917049661278725,
      "learning_rate": 0.0012,
      "loss": 2.6505918502807617,
      "step": 7828
    },
    {
      "epoch": 0.6348643286618846,
      "grad_norm": 0.04788810759782791,
      "learning_rate": 0.0012,
      "loss": 2.6606969833374023,
      "step": 7830
    },
    {
      "epoch": 0.6350264906870856,
      "grad_norm": 0.06249278038740158,
      "learning_rate": 0.0012,
      "loss": 2.660310745239258,
      "step": 7832
    },
    {
      "epoch": 0.6351886527122865,
      "grad_norm": 0.028865410014986992,
      "learning_rate": 0.0012,
      "loss": 2.7238540649414062,
      "step": 7834
    },
    {
      "epoch": 0.6353508147374876,
      "grad_norm": 0.032712750136852264,
      "learning_rate": 0.0012,
      "loss": 2.6504507064819336,
      "step": 7836
    },
    {
      "epoch": 0.6355129767626886,
      "grad_norm": 0.04368513822555542,
      "learning_rate": 0.0012,
      "loss": 2.644920587539673,
      "step": 7838
    },
    {
      "epoch": 0.6356751387878895,
      "grad_norm": 0.030948510393500328,
      "learning_rate": 0.0012,
      "loss": 2.7352614402770996,
      "step": 7840
    },
    {
      "epoch": 0.6358373008130905,
      "grad_norm": 0.04912412539124489,
      "learning_rate": 0.0012,
      "loss": 2.6583361625671387,
      "step": 7842
    },
    {
      "epoch": 0.6359994628382916,
      "grad_norm": 0.02523570880293846,
      "learning_rate": 0.0012,
      "loss": 2.608673095703125,
      "step": 7844
    },
    {
      "epoch": 0.6361616248634925,
      "grad_norm": 0.04276378080248833,
      "learning_rate": 0.0012,
      "loss": 2.6717543601989746,
      "step": 7846
    },
    {
      "epoch": 0.6363237868886935,
      "grad_norm": 0.04431458190083504,
      "learning_rate": 0.0012,
      "loss": 2.681196689605713,
      "step": 7848
    },
    {
      "epoch": 0.6364859489138945,
      "grad_norm": 0.025992123410105705,
      "learning_rate": 0.0012,
      "loss": 2.649388074874878,
      "step": 7850
    },
    {
      "epoch": 0.6366481109390955,
      "grad_norm": 0.036005664616823196,
      "learning_rate": 0.0012,
      "loss": 2.7012455463409424,
      "step": 7852
    },
    {
      "epoch": 0.6368102729642965,
      "grad_norm": 0.03172362968325615,
      "learning_rate": 0.0012,
      "loss": 2.6584882736206055,
      "step": 7854
    },
    {
      "epoch": 0.6369724349894975,
      "grad_norm": 0.03090614266693592,
      "learning_rate": 0.0012,
      "loss": 2.647672176361084,
      "step": 7856
    },
    {
      "epoch": 0.6371345970146984,
      "grad_norm": 0.02423260547220707,
      "learning_rate": 0.0012,
      "loss": 2.6649279594421387,
      "step": 7858
    },
    {
      "epoch": 0.6372967590398995,
      "grad_norm": 0.02768014557659626,
      "learning_rate": 0.0012,
      "loss": 2.6870079040527344,
      "step": 7860
    },
    {
      "epoch": 0.6374589210651005,
      "grad_norm": 0.0451471172273159,
      "learning_rate": 0.0012,
      "loss": 2.7072134017944336,
      "step": 7862
    },
    {
      "epoch": 0.6376210830903014,
      "grad_norm": 0.03840259090065956,
      "learning_rate": 0.0012,
      "loss": 2.7363364696502686,
      "step": 7864
    },
    {
      "epoch": 0.6377832451155024,
      "grad_norm": 0.030282452702522278,
      "learning_rate": 0.0012,
      "loss": 2.674351930618286,
      "step": 7866
    },
    {
      "epoch": 0.6379454071407035,
      "grad_norm": 0.044061433523893356,
      "learning_rate": 0.0012,
      "loss": 2.6705141067504883,
      "step": 7868
    },
    {
      "epoch": 0.6381075691659044,
      "grad_norm": 0.03016877919435501,
      "learning_rate": 0.0012,
      "loss": 2.618594169616699,
      "step": 7870
    },
    {
      "epoch": 0.6382697311911054,
      "grad_norm": 0.041007790714502335,
      "learning_rate": 0.0012,
      "loss": 2.6951613426208496,
      "step": 7872
    },
    {
      "epoch": 0.6384318932163064,
      "grad_norm": 0.038392744958400726,
      "learning_rate": 0.0012,
      "loss": 2.7187724113464355,
      "step": 7874
    },
    {
      "epoch": 0.6385940552415074,
      "grad_norm": 0.031296536326408386,
      "learning_rate": 0.0012,
      "loss": 2.6747422218322754,
      "step": 7876
    },
    {
      "epoch": 0.6387562172667084,
      "grad_norm": 0.02266651764512062,
      "learning_rate": 0.0012,
      "loss": 2.6563949584960938,
      "step": 7878
    },
    {
      "epoch": 0.6389183792919094,
      "grad_norm": 0.04354286566376686,
      "learning_rate": 0.0012,
      "loss": 2.691863536834717,
      "step": 7880
    },
    {
      "epoch": 0.6390805413171103,
      "grad_norm": 0.040460988879203796,
      "learning_rate": 0.0012,
      "loss": 2.603435754776001,
      "step": 7882
    },
    {
      "epoch": 0.6392427033423114,
      "grad_norm": 0.024061091244220734,
      "learning_rate": 0.0012,
      "loss": 2.642179489135742,
      "step": 7884
    },
    {
      "epoch": 0.6394048653675124,
      "grad_norm": 0.05153767019510269,
      "learning_rate": 0.0012,
      "loss": 2.664945125579834,
      "step": 7886
    },
    {
      "epoch": 0.6395670273927133,
      "grad_norm": 0.06063957139849663,
      "learning_rate": 0.0012,
      "loss": 2.7585840225219727,
      "step": 7888
    },
    {
      "epoch": 0.6397291894179143,
      "grad_norm": 0.02887694351375103,
      "learning_rate": 0.0012,
      "loss": 2.7300467491149902,
      "step": 7890
    },
    {
      "epoch": 0.6398913514431154,
      "grad_norm": 0.035527754575014114,
      "learning_rate": 0.0012,
      "loss": 2.687703847885132,
      "step": 7892
    },
    {
      "epoch": 0.6400535134683163,
      "grad_norm": 0.04448281228542328,
      "learning_rate": 0.0012,
      "loss": 2.6936733722686768,
      "step": 7894
    },
    {
      "epoch": 0.6402156754935173,
      "grad_norm": 0.05912408232688904,
      "learning_rate": 0.0012,
      "loss": 2.6595940589904785,
      "step": 7896
    },
    {
      "epoch": 0.6403778375187184,
      "grad_norm": 0.023125065490603447,
      "learning_rate": 0.0012,
      "loss": 2.630281925201416,
      "step": 7898
    },
    {
      "epoch": 0.6405399995439193,
      "grad_norm": 0.036279018968343735,
      "learning_rate": 0.0012,
      "loss": 2.6882705688476562,
      "step": 7900
    },
    {
      "epoch": 0.6407021615691203,
      "grad_norm": 0.0526435524225235,
      "learning_rate": 0.0012,
      "loss": 2.7074155807495117,
      "step": 7902
    },
    {
      "epoch": 0.6408643235943213,
      "grad_norm": 0.026883559301495552,
      "learning_rate": 0.0012,
      "loss": 2.6144771575927734,
      "step": 7904
    },
    {
      "epoch": 0.6410264856195222,
      "grad_norm": 0.02703884243965149,
      "learning_rate": 0.0012,
      "loss": 2.6142311096191406,
      "step": 7906
    },
    {
      "epoch": 0.6411886476447233,
      "grad_norm": 0.047394443303346634,
      "learning_rate": 0.0012,
      "loss": 2.5982258319854736,
      "step": 7908
    },
    {
      "epoch": 0.6413508096699243,
      "grad_norm": 0.02674596756696701,
      "learning_rate": 0.0012,
      "loss": 2.685194492340088,
      "step": 7910
    },
    {
      "epoch": 0.6415129716951252,
      "grad_norm": 0.030475720763206482,
      "learning_rate": 0.0012,
      "loss": 2.602053165435791,
      "step": 7912
    },
    {
      "epoch": 0.6416751337203263,
      "grad_norm": 0.05910048633813858,
      "learning_rate": 0.0012,
      "loss": 2.7472972869873047,
      "step": 7914
    },
    {
      "epoch": 0.6418372957455273,
      "grad_norm": 0.07643469423055649,
      "learning_rate": 0.0012,
      "loss": 2.695553779602051,
      "step": 7916
    },
    {
      "epoch": 0.6419994577707282,
      "grad_norm": 0.029595905914902687,
      "learning_rate": 0.0012,
      "loss": 2.6894850730895996,
      "step": 7918
    },
    {
      "epoch": 0.6421616197959292,
      "grad_norm": 0.11036810278892517,
      "learning_rate": 0.0012,
      "loss": 2.675507068634033,
      "step": 7920
    },
    {
      "epoch": 0.6423237818211303,
      "grad_norm": 0.06474505364894867,
      "learning_rate": 0.0012,
      "loss": 2.6747777462005615,
      "step": 7922
    },
    {
      "epoch": 0.6424859438463312,
      "grad_norm": 0.02979455143213272,
      "learning_rate": 0.0012,
      "loss": 2.6890952587127686,
      "step": 7924
    },
    {
      "epoch": 0.6426481058715322,
      "grad_norm": 0.05747007578611374,
      "learning_rate": 0.0012,
      "loss": 2.6548595428466797,
      "step": 7926
    },
    {
      "epoch": 0.6428102678967332,
      "grad_norm": 0.030022162944078445,
      "learning_rate": 0.0012,
      "loss": 2.6289374828338623,
      "step": 7928
    },
    {
      "epoch": 0.6429724299219342,
      "grad_norm": 0.03554845228791237,
      "learning_rate": 0.0012,
      "loss": 2.7036404609680176,
      "step": 7930
    },
    {
      "epoch": 0.6431345919471352,
      "grad_norm": 0.04711224138736725,
      "learning_rate": 0.0012,
      "loss": 2.6939916610717773,
      "step": 7932
    },
    {
      "epoch": 0.6432967539723362,
      "grad_norm": 0.029162436723709106,
      "learning_rate": 0.0012,
      "loss": 2.591928482055664,
      "step": 7934
    },
    {
      "epoch": 0.6434589159975371,
      "grad_norm": 0.024285079911351204,
      "learning_rate": 0.0012,
      "loss": 2.6414332389831543,
      "step": 7936
    },
    {
      "epoch": 0.6436210780227382,
      "grad_norm": 0.029341671615839005,
      "learning_rate": 0.0012,
      "loss": 2.6873550415039062,
      "step": 7938
    },
    {
      "epoch": 0.6437832400479392,
      "grad_norm": 0.025009753182530403,
      "learning_rate": 0.0012,
      "loss": 2.695401191711426,
      "step": 7940
    },
    {
      "epoch": 0.6439454020731401,
      "grad_norm": 0.025177473202347755,
      "learning_rate": 0.0012,
      "loss": 2.578293800354004,
      "step": 7942
    },
    {
      "epoch": 0.6441075640983411,
      "grad_norm": 0.02372266724705696,
      "learning_rate": 0.0012,
      "loss": 2.6093204021453857,
      "step": 7944
    },
    {
      "epoch": 0.6442697261235422,
      "grad_norm": 0.026333291083574295,
      "learning_rate": 0.0012,
      "loss": 2.6679491996765137,
      "step": 7946
    },
    {
      "epoch": 0.6444318881487431,
      "grad_norm": 0.02576783485710621,
      "learning_rate": 0.0012,
      "loss": 2.653761386871338,
      "step": 7948
    },
    {
      "epoch": 0.6445940501739441,
      "grad_norm": 0.029459545388817787,
      "learning_rate": 0.0012,
      "loss": 2.6252663135528564,
      "step": 7950
    },
    {
      "epoch": 0.6447562121991451,
      "grad_norm": 0.03919835016131401,
      "learning_rate": 0.0012,
      "loss": 2.6777186393737793,
      "step": 7952
    },
    {
      "epoch": 0.6449183742243461,
      "grad_norm": 0.028542116284370422,
      "learning_rate": 0.0012,
      "loss": 2.698887348175049,
      "step": 7954
    },
    {
      "epoch": 0.6450805362495471,
      "grad_norm": 0.041377466171979904,
      "learning_rate": 0.0012,
      "loss": 2.6740856170654297,
      "step": 7956
    },
    {
      "epoch": 0.6452426982747481,
      "grad_norm": 0.026738854125142097,
      "learning_rate": 0.0012,
      "loss": 2.7587521076202393,
      "step": 7958
    },
    {
      "epoch": 0.645404860299949,
      "grad_norm": 0.044661130756139755,
      "learning_rate": 0.0012,
      "loss": 2.5866708755493164,
      "step": 7960
    },
    {
      "epoch": 0.6455670223251501,
      "grad_norm": 0.057468123733997345,
      "learning_rate": 0.0012,
      "loss": 2.644503355026245,
      "step": 7962
    },
    {
      "epoch": 0.6457291843503511,
      "grad_norm": 0.041436463594436646,
      "learning_rate": 0.0012,
      "loss": 2.7061166763305664,
      "step": 7964
    },
    {
      "epoch": 0.645891346375552,
      "grad_norm": 0.029554156586527824,
      "learning_rate": 0.0012,
      "loss": 2.651020050048828,
      "step": 7966
    },
    {
      "epoch": 0.646053508400753,
      "grad_norm": 0.035855285823345184,
      "learning_rate": 0.0012,
      "loss": 2.6907262802124023,
      "step": 7968
    },
    {
      "epoch": 0.6462156704259541,
      "grad_norm": 0.0524868480861187,
      "learning_rate": 0.0012,
      "loss": 2.6283164024353027,
      "step": 7970
    },
    {
      "epoch": 0.646377832451155,
      "grad_norm": 0.03427949920296669,
      "learning_rate": 0.0012,
      "loss": 2.699143409729004,
      "step": 7972
    },
    {
      "epoch": 0.646539994476356,
      "grad_norm": 0.03707023337483406,
      "learning_rate": 0.0012,
      "loss": 2.666109085083008,
      "step": 7974
    },
    {
      "epoch": 0.646702156501557,
      "grad_norm": 0.055571068078279495,
      "learning_rate": 0.0012,
      "loss": 2.6474833488464355,
      "step": 7976
    },
    {
      "epoch": 0.646864318526758,
      "grad_norm": 0.04922313988208771,
      "learning_rate": 0.0012,
      "loss": 2.727583646774292,
      "step": 7978
    },
    {
      "epoch": 0.647026480551959,
      "grad_norm": 0.0329282246530056,
      "learning_rate": 0.0012,
      "loss": 2.6789255142211914,
      "step": 7980
    },
    {
      "epoch": 0.64718864257716,
      "grad_norm": 0.027015915140509605,
      "learning_rate": 0.0012,
      "loss": 2.6279473304748535,
      "step": 7982
    },
    {
      "epoch": 0.6473508046023609,
      "grad_norm": 0.04910627752542496,
      "learning_rate": 0.0012,
      "loss": 2.6609115600585938,
      "step": 7984
    },
    {
      "epoch": 0.647512966627562,
      "grad_norm": 0.03130998834967613,
      "learning_rate": 0.0012,
      "loss": 2.648806095123291,
      "step": 7986
    },
    {
      "epoch": 0.647675128652763,
      "grad_norm": 0.04197905585169792,
      "learning_rate": 0.0012,
      "loss": 2.6663691997528076,
      "step": 7988
    },
    {
      "epoch": 0.6478372906779639,
      "grad_norm": 0.06529869884252548,
      "learning_rate": 0.0012,
      "loss": 2.6902403831481934,
      "step": 7990
    },
    {
      "epoch": 0.647999452703165,
      "grad_norm": 0.06489566713571548,
      "learning_rate": 0.0012,
      "loss": 2.7005324363708496,
      "step": 7992
    },
    {
      "epoch": 0.648161614728366,
      "grad_norm": 0.025371139869093895,
      "learning_rate": 0.0012,
      "loss": 2.6083240509033203,
      "step": 7994
    },
    {
      "epoch": 0.6483237767535669,
      "grad_norm": 0.024228442460298538,
      "learning_rate": 0.0012,
      "loss": 2.716801643371582,
      "step": 7996
    },
    {
      "epoch": 0.6484859387787679,
      "grad_norm": 0.04115039482712746,
      "learning_rate": 0.0012,
      "loss": 2.685211658477783,
      "step": 7998
    },
    {
      "epoch": 0.648648100803969,
      "grad_norm": 0.022875415161252022,
      "learning_rate": 0.0012,
      "loss": 2.6821866035461426,
      "step": 8000
    },
    {
      "epoch": 0.6488102628291699,
      "grad_norm": 0.027056654915213585,
      "learning_rate": 0.0012,
      "loss": 2.6660842895507812,
      "step": 8002
    },
    {
      "epoch": 0.6489724248543709,
      "grad_norm": 0.027666974812746048,
      "learning_rate": 0.0012,
      "loss": 2.6945347785949707,
      "step": 8004
    },
    {
      "epoch": 0.6491345868795719,
      "grad_norm": 0.026555482298135757,
      "learning_rate": 0.0012,
      "loss": 2.6195244789123535,
      "step": 8006
    },
    {
      "epoch": 0.6492967489047728,
      "grad_norm": 0.03173981234431267,
      "learning_rate": 0.0012,
      "loss": 2.635850667953491,
      "step": 8008
    },
    {
      "epoch": 0.6494589109299739,
      "grad_norm": 0.03213329613208771,
      "learning_rate": 0.0012,
      "loss": 2.669126033782959,
      "step": 8010
    },
    {
      "epoch": 0.6496210729551749,
      "grad_norm": 0.04665551707148552,
      "learning_rate": 0.0012,
      "loss": 2.604041576385498,
      "step": 8012
    },
    {
      "epoch": 0.6497832349803758,
      "grad_norm": 0.054284725338220596,
      "learning_rate": 0.0012,
      "loss": 2.680166721343994,
      "step": 8014
    },
    {
      "epoch": 0.6499453970055769,
      "grad_norm": 0.024625806137919426,
      "learning_rate": 0.0012,
      "loss": 2.6328489780426025,
      "step": 8016
    },
    {
      "epoch": 0.6501075590307779,
      "grad_norm": 0.02649722807109356,
      "learning_rate": 0.0012,
      "loss": 2.6716322898864746,
      "step": 8018
    },
    {
      "epoch": 0.6502697210559788,
      "grad_norm": 0.02605276368558407,
      "learning_rate": 0.0012,
      "loss": 2.6852030754089355,
      "step": 8020
    },
    {
      "epoch": 0.6504318830811798,
      "grad_norm": 0.027412403374910355,
      "learning_rate": 0.0012,
      "loss": 2.6554617881774902,
      "step": 8022
    },
    {
      "epoch": 0.6505940451063809,
      "grad_norm": 0.029976224526762962,
      "learning_rate": 0.0012,
      "loss": 2.6566004753112793,
      "step": 8024
    },
    {
      "epoch": 0.6507562071315818,
      "grad_norm": 0.03373529016971588,
      "learning_rate": 0.0012,
      "loss": 2.710371971130371,
      "step": 8026
    },
    {
      "epoch": 0.6509183691567828,
      "grad_norm": 0.04039378836750984,
      "learning_rate": 0.0012,
      "loss": 2.5786774158477783,
      "step": 8028
    },
    {
      "epoch": 0.6510805311819838,
      "grad_norm": 0.040415141731500626,
      "learning_rate": 0.0012,
      "loss": 2.645345687866211,
      "step": 8030
    },
    {
      "epoch": 0.6512426932071848,
      "grad_norm": 0.046487417072057724,
      "learning_rate": 0.0012,
      "loss": 2.724811553955078,
      "step": 8032
    },
    {
      "epoch": 0.6514048552323858,
      "grad_norm": 0.0441647432744503,
      "learning_rate": 0.0012,
      "loss": 2.65350079536438,
      "step": 8034
    },
    {
      "epoch": 0.6515670172575868,
      "grad_norm": 0.024929963052272797,
      "learning_rate": 0.0012,
      "loss": 2.722721576690674,
      "step": 8036
    },
    {
      "epoch": 0.6517291792827877,
      "grad_norm": 0.028224021196365356,
      "learning_rate": 0.0012,
      "loss": 2.6208319664001465,
      "step": 8038
    },
    {
      "epoch": 0.6518913413079888,
      "grad_norm": 0.02227029763162136,
      "learning_rate": 0.0012,
      "loss": 2.684049606323242,
      "step": 8040
    },
    {
      "epoch": 0.6520535033331898,
      "grad_norm": 0.03510322421789169,
      "learning_rate": 0.0012,
      "loss": 2.6578104496002197,
      "step": 8042
    },
    {
      "epoch": 0.6522156653583907,
      "grad_norm": 0.02445952035486698,
      "learning_rate": 0.0012,
      "loss": 2.6271743774414062,
      "step": 8044
    },
    {
      "epoch": 0.6523778273835917,
      "grad_norm": 0.02600482478737831,
      "learning_rate": 0.0012,
      "loss": 2.625241756439209,
      "step": 8046
    },
    {
      "epoch": 0.6525399894087928,
      "grad_norm": 0.023394733667373657,
      "learning_rate": 0.0012,
      "loss": 2.6769683361053467,
      "step": 8048
    },
    {
      "epoch": 0.6527021514339937,
      "grad_norm": 0.025055130943655968,
      "learning_rate": 0.0012,
      "loss": 2.677837371826172,
      "step": 8050
    },
    {
      "epoch": 0.6528643134591947,
      "grad_norm": 0.024587206542491913,
      "learning_rate": 0.0012,
      "loss": 2.6379518508911133,
      "step": 8052
    },
    {
      "epoch": 0.6530264754843957,
      "grad_norm": 0.02477140538394451,
      "learning_rate": 0.0012,
      "loss": 2.7078359127044678,
      "step": 8054
    },
    {
      "epoch": 0.6531886375095967,
      "grad_norm": 0.03324050456285477,
      "learning_rate": 0.0012,
      "loss": 2.677288055419922,
      "step": 8056
    },
    {
      "epoch": 0.6533507995347977,
      "grad_norm": 0.03515487536787987,
      "learning_rate": 0.0012,
      "loss": 2.639411449432373,
      "step": 8058
    },
    {
      "epoch": 0.6535129615599987,
      "grad_norm": 0.03560230880975723,
      "learning_rate": 0.0012,
      "loss": 2.6215157508850098,
      "step": 8060
    },
    {
      "epoch": 0.6536751235851996,
      "grad_norm": 0.026229573413729668,
      "learning_rate": 0.0012,
      "loss": 2.693690776824951,
      "step": 8062
    },
    {
      "epoch": 0.6538372856104007,
      "grad_norm": 0.02595142088830471,
      "learning_rate": 0.0012,
      "loss": 2.675210952758789,
      "step": 8064
    },
    {
      "epoch": 0.6539994476356017,
      "grad_norm": 0.039194341748952866,
      "learning_rate": 0.0012,
      "loss": 2.713968276977539,
      "step": 8066
    },
    {
      "epoch": 0.6541616096608026,
      "grad_norm": 0.025508275255560875,
      "learning_rate": 0.0012,
      "loss": 2.7003846168518066,
      "step": 8068
    },
    {
      "epoch": 0.6543237716860036,
      "grad_norm": 0.024688048288226128,
      "learning_rate": 0.0012,
      "loss": 2.6209824085235596,
      "step": 8070
    },
    {
      "epoch": 0.6544859337112047,
      "grad_norm": 0.022515777498483658,
      "learning_rate": 0.0012,
      "loss": 2.681490898132324,
      "step": 8072
    },
    {
      "epoch": 0.6546480957364056,
      "grad_norm": 0.055974770337343216,
      "learning_rate": 0.0012,
      "loss": 2.6361117362976074,
      "step": 8074
    },
    {
      "epoch": 0.6548102577616066,
      "grad_norm": 0.0899575799703598,
      "learning_rate": 0.0012,
      "loss": 2.7113895416259766,
      "step": 8076
    },
    {
      "epoch": 0.6549724197868076,
      "grad_norm": 0.10126933455467224,
      "learning_rate": 0.0012,
      "loss": 2.7550132274627686,
      "step": 8078
    },
    {
      "epoch": 0.6551345818120086,
      "grad_norm": 0.04052496701478958,
      "learning_rate": 0.0012,
      "loss": 2.7175817489624023,
      "step": 8080
    },
    {
      "epoch": 0.6552967438372096,
      "grad_norm": 0.03333786129951477,
      "learning_rate": 0.0012,
      "loss": 2.619941234588623,
      "step": 8082
    },
    {
      "epoch": 0.6554589058624106,
      "grad_norm": 0.03230750933289528,
      "learning_rate": 0.0012,
      "loss": 2.64487886428833,
      "step": 8084
    },
    {
      "epoch": 0.6556210678876115,
      "grad_norm": 0.03350953385233879,
      "learning_rate": 0.0012,
      "loss": 2.6684701442718506,
      "step": 8086
    },
    {
      "epoch": 0.6557832299128126,
      "grad_norm": 0.024862876161932945,
      "learning_rate": 0.0012,
      "loss": 2.7242465019226074,
      "step": 8088
    },
    {
      "epoch": 0.6559453919380136,
      "grad_norm": 0.025659993290901184,
      "learning_rate": 0.0012,
      "loss": 2.6600711345672607,
      "step": 8090
    },
    {
      "epoch": 0.6561075539632145,
      "grad_norm": 0.043750714510679245,
      "learning_rate": 0.0012,
      "loss": 2.6883938312530518,
      "step": 8092
    },
    {
      "epoch": 0.6562697159884155,
      "grad_norm": 0.025937369093298912,
      "learning_rate": 0.0012,
      "loss": 2.6330299377441406,
      "step": 8094
    },
    {
      "epoch": 0.6564318780136166,
      "grad_norm": 0.026330826804041862,
      "learning_rate": 0.0012,
      "loss": 2.627136707305908,
      "step": 8096
    },
    {
      "epoch": 0.6565940400388175,
      "grad_norm": 0.025433652102947235,
      "learning_rate": 0.0012,
      "loss": 2.557776689529419,
      "step": 8098
    },
    {
      "epoch": 0.6567562020640185,
      "grad_norm": 0.02387945167720318,
      "learning_rate": 0.0012,
      "loss": 2.7015271186828613,
      "step": 8100
    },
    {
      "epoch": 0.6569183640892196,
      "grad_norm": 0.024347038939595222,
      "learning_rate": 0.0012,
      "loss": 2.6345324516296387,
      "step": 8102
    },
    {
      "epoch": 0.6570805261144205,
      "grad_norm": 0.039674848318099976,
      "learning_rate": 0.0012,
      "loss": 2.647897243499756,
      "step": 8104
    },
    {
      "epoch": 0.6572426881396215,
      "grad_norm": 0.03335815295577049,
      "learning_rate": 0.0012,
      "loss": 2.647706985473633,
      "step": 8106
    },
    {
      "epoch": 0.6574048501648225,
      "grad_norm": 0.02926114946603775,
      "learning_rate": 0.0012,
      "loss": 2.680840015411377,
      "step": 8108
    },
    {
      "epoch": 0.6575670121900234,
      "grad_norm": 0.024534588679671288,
      "learning_rate": 0.0012,
      "loss": 2.619514226913452,
      "step": 8110
    },
    {
      "epoch": 0.6577291742152245,
      "grad_norm": 0.03136385977268219,
      "learning_rate": 0.0012,
      "loss": 2.6950275897979736,
      "step": 8112
    },
    {
      "epoch": 0.6578913362404255,
      "grad_norm": 0.026157846674323082,
      "learning_rate": 0.0012,
      "loss": 2.6186904907226562,
      "step": 8114
    },
    {
      "epoch": 0.6580534982656264,
      "grad_norm": 0.025654837489128113,
      "learning_rate": 0.0012,
      "loss": 2.6348915100097656,
      "step": 8116
    },
    {
      "epoch": 0.6582156602908275,
      "grad_norm": 0.036802928894758224,
      "learning_rate": 0.0012,
      "loss": 2.7089388370513916,
      "step": 8118
    },
    {
      "epoch": 0.6583778223160285,
      "grad_norm": 0.027866747230291367,
      "learning_rate": 0.0012,
      "loss": 2.6619882583618164,
      "step": 8120
    },
    {
      "epoch": 0.6585399843412294,
      "grad_norm": 0.03373255953192711,
      "learning_rate": 0.0012,
      "loss": 2.6835904121398926,
      "step": 8122
    },
    {
      "epoch": 0.6587021463664304,
      "grad_norm": 0.03141677752137184,
      "learning_rate": 0.0012,
      "loss": 2.6964802742004395,
      "step": 8124
    },
    {
      "epoch": 0.6588643083916315,
      "grad_norm": 0.026937466114759445,
      "learning_rate": 0.0012,
      "loss": 2.634063243865967,
      "step": 8126
    },
    {
      "epoch": 0.6590264704168324,
      "grad_norm": 0.03336050361394882,
      "learning_rate": 0.0012,
      "loss": 2.697477340698242,
      "step": 8128
    },
    {
      "epoch": 0.6591886324420334,
      "grad_norm": 0.02558707445859909,
      "learning_rate": 0.0012,
      "loss": 2.6614766120910645,
      "step": 8130
    },
    {
      "epoch": 0.6593507944672344,
      "grad_norm": 0.046782415360212326,
      "learning_rate": 0.0012,
      "loss": 2.7101378440856934,
      "step": 8132
    },
    {
      "epoch": 0.6595129564924354,
      "grad_norm": 0.03828199580311775,
      "learning_rate": 0.0012,
      "loss": 2.6409473419189453,
      "step": 8134
    },
    {
      "epoch": 0.6596751185176364,
      "grad_norm": 0.035570431500673294,
      "learning_rate": 0.0012,
      "loss": 2.6497433185577393,
      "step": 8136
    },
    {
      "epoch": 0.6598372805428374,
      "grad_norm": 0.02913297526538372,
      "learning_rate": 0.0012,
      "loss": 2.729975700378418,
      "step": 8138
    },
    {
      "epoch": 0.6599994425680383,
      "grad_norm": 0.028301255777478218,
      "learning_rate": 0.0012,
      "loss": 2.650369644165039,
      "step": 8140
    },
    {
      "epoch": 0.6601616045932394,
      "grad_norm": 0.046083323657512665,
      "learning_rate": 0.0012,
      "loss": 2.6569931507110596,
      "step": 8142
    },
    {
      "epoch": 0.6603237666184404,
      "grad_norm": 0.03451770171523094,
      "learning_rate": 0.0012,
      "loss": 2.6982812881469727,
      "step": 8144
    },
    {
      "epoch": 0.6604859286436413,
      "grad_norm": 0.0382692813873291,
      "learning_rate": 0.0012,
      "loss": 2.637091875076294,
      "step": 8146
    },
    {
      "epoch": 0.6606480906688423,
      "grad_norm": 0.057290226221084595,
      "learning_rate": 0.0012,
      "loss": 2.6160812377929688,
      "step": 8148
    },
    {
      "epoch": 0.6608102526940434,
      "grad_norm": 0.050387781113386154,
      "learning_rate": 0.0012,
      "loss": 2.7057671546936035,
      "step": 8150
    },
    {
      "epoch": 0.6609724147192443,
      "grad_norm": 0.025329770520329475,
      "learning_rate": 0.0012,
      "loss": 2.729529857635498,
      "step": 8152
    },
    {
      "epoch": 0.6611345767444453,
      "grad_norm": 0.034457240253686905,
      "learning_rate": 0.0012,
      "loss": 2.7318739891052246,
      "step": 8154
    },
    {
      "epoch": 0.6612967387696463,
      "grad_norm": 0.025771385058760643,
      "learning_rate": 0.0012,
      "loss": 2.634303569793701,
      "step": 8156
    },
    {
      "epoch": 0.6614589007948473,
      "grad_norm": 0.02418314293026924,
      "learning_rate": 0.0012,
      "loss": 2.6748759746551514,
      "step": 8158
    },
    {
      "epoch": 0.6616210628200483,
      "grad_norm": 0.026810284703969955,
      "learning_rate": 0.0012,
      "loss": 2.629711151123047,
      "step": 8160
    },
    {
      "epoch": 0.6617832248452493,
      "grad_norm": 0.025066787376999855,
      "learning_rate": 0.0012,
      "loss": 2.589428186416626,
      "step": 8162
    },
    {
      "epoch": 0.6619453868704502,
      "grad_norm": 0.02437845803797245,
      "learning_rate": 0.0012,
      "loss": 2.7084007263183594,
      "step": 8164
    },
    {
      "epoch": 0.6621075488956513,
      "grad_norm": 0.026299241930246353,
      "learning_rate": 0.0012,
      "loss": 2.6652517318725586,
      "step": 8166
    },
    {
      "epoch": 0.6622697109208523,
      "grad_norm": 0.02863742597401142,
      "learning_rate": 0.0012,
      "loss": 2.641650676727295,
      "step": 8168
    },
    {
      "epoch": 0.6624318729460532,
      "grad_norm": 0.029923418536782265,
      "learning_rate": 0.0012,
      "loss": 2.656867504119873,
      "step": 8170
    },
    {
      "epoch": 0.6625940349712542,
      "grad_norm": 0.026221947744488716,
      "learning_rate": 0.0012,
      "loss": 2.6678900718688965,
      "step": 8172
    },
    {
      "epoch": 0.6627561969964553,
      "grad_norm": 0.052301328629255295,
      "learning_rate": 0.0012,
      "loss": 2.6518971920013428,
      "step": 8174
    },
    {
      "epoch": 0.6629183590216562,
      "grad_norm": 0.04030179604887962,
      "learning_rate": 0.0012,
      "loss": 2.5893988609313965,
      "step": 8176
    },
    {
      "epoch": 0.6630805210468572,
      "grad_norm": 0.029391063377261162,
      "learning_rate": 0.0012,
      "loss": 2.6833720207214355,
      "step": 8178
    },
    {
      "epoch": 0.6632426830720582,
      "grad_norm": 0.02989056520164013,
      "learning_rate": 0.0012,
      "loss": 2.7014856338500977,
      "step": 8180
    },
    {
      "epoch": 0.6634048450972592,
      "grad_norm": 0.045719657093286514,
      "learning_rate": 0.0012,
      "loss": 2.729691505432129,
      "step": 8182
    },
    {
      "epoch": 0.6635670071224602,
      "grad_norm": 0.06007017940282822,
      "learning_rate": 0.0012,
      "loss": 2.7117810249328613,
      "step": 8184
    },
    {
      "epoch": 0.6637291691476612,
      "grad_norm": 0.05731342360377312,
      "learning_rate": 0.0012,
      "loss": 2.6173439025878906,
      "step": 8186
    },
    {
      "epoch": 0.6638913311728621,
      "grad_norm": 0.043483201414346695,
      "learning_rate": 0.0012,
      "loss": 2.723672389984131,
      "step": 8188
    },
    {
      "epoch": 0.6640534931980632,
      "grad_norm": 0.02765815332531929,
      "learning_rate": 0.0012,
      "loss": 2.636183261871338,
      "step": 8190
    },
    {
      "epoch": 0.6642156552232642,
      "grad_norm": 0.03945249319076538,
      "learning_rate": 0.0012,
      "loss": 2.6756038665771484,
      "step": 8192
    },
    {
      "epoch": 0.6643778172484651,
      "grad_norm": 0.040668997913599014,
      "learning_rate": 0.0012,
      "loss": 2.6732869148254395,
      "step": 8194
    },
    {
      "epoch": 0.6645399792736661,
      "grad_norm": 0.027934985235333443,
      "learning_rate": 0.0012,
      "loss": 2.687490940093994,
      "step": 8196
    },
    {
      "epoch": 0.6647021412988672,
      "grad_norm": 0.035481955856084824,
      "learning_rate": 0.0012,
      "loss": 2.7317309379577637,
      "step": 8198
    },
    {
      "epoch": 0.6648643033240681,
      "grad_norm": 0.03527505323290825,
      "learning_rate": 0.0012,
      "loss": 2.630800724029541,
      "step": 8200
    },
    {
      "epoch": 0.6650264653492691,
      "grad_norm": 0.0485035665333271,
      "learning_rate": 0.0012,
      "loss": 2.661041736602783,
      "step": 8202
    },
    {
      "epoch": 0.6651886273744702,
      "grad_norm": 0.027026424184441566,
      "learning_rate": 0.0012,
      "loss": 2.6317520141601562,
      "step": 8204
    },
    {
      "epoch": 0.6653507893996711,
      "grad_norm": 0.03265989571809769,
      "learning_rate": 0.0012,
      "loss": 2.7085347175598145,
      "step": 8206
    },
    {
      "epoch": 0.6655129514248721,
      "grad_norm": 0.032105453312397,
      "learning_rate": 0.0012,
      "loss": 2.667010545730591,
      "step": 8208
    },
    {
      "epoch": 0.6656751134500731,
      "grad_norm": 0.04197046533226967,
      "learning_rate": 0.0012,
      "loss": 2.658613681793213,
      "step": 8210
    },
    {
      "epoch": 0.665837275475274,
      "grad_norm": 0.038663629442453384,
      "learning_rate": 0.0012,
      "loss": 2.670064926147461,
      "step": 8212
    },
    {
      "epoch": 0.6659994375004751,
      "grad_norm": 0.02236200124025345,
      "learning_rate": 0.0012,
      "loss": 2.656104564666748,
      "step": 8214
    },
    {
      "epoch": 0.6661615995256761,
      "grad_norm": 0.04644607752561569,
      "learning_rate": 0.0012,
      "loss": 2.6695733070373535,
      "step": 8216
    },
    {
      "epoch": 0.666323761550877,
      "grad_norm": 0.030448701232671738,
      "learning_rate": 0.0012,
      "loss": 2.717395782470703,
      "step": 8218
    },
    {
      "epoch": 0.666485923576078,
      "grad_norm": 0.030394207686185837,
      "learning_rate": 0.0012,
      "loss": 2.706129789352417,
      "step": 8220
    },
    {
      "epoch": 0.6666480856012791,
      "grad_norm": 0.027133464813232422,
      "learning_rate": 0.0012,
      "loss": 2.7265233993530273,
      "step": 8222
    },
    {
      "epoch": 0.66681024762648,
      "grad_norm": 0.036136385053396225,
      "learning_rate": 0.0012,
      "loss": 2.664473056793213,
      "step": 8224
    },
    {
      "epoch": 0.666972409651681,
      "grad_norm": 0.023256227374076843,
      "learning_rate": 0.0012,
      "loss": 2.671337604522705,
      "step": 8226
    },
    {
      "epoch": 0.6671345716768821,
      "grad_norm": 0.024262448772788048,
      "learning_rate": 0.0012,
      "loss": 2.6266236305236816,
      "step": 8228
    },
    {
      "epoch": 0.667296733702083,
      "grad_norm": 0.044287558645009995,
      "learning_rate": 0.0012,
      "loss": 2.6802353858947754,
      "step": 8230
    },
    {
      "epoch": 0.667458895727284,
      "grad_norm": 0.02350819855928421,
      "learning_rate": 0.0012,
      "loss": 2.617231845855713,
      "step": 8232
    },
    {
      "epoch": 0.667621057752485,
      "grad_norm": 0.03448973596096039,
      "learning_rate": 0.0012,
      "loss": 2.5875649452209473,
      "step": 8234
    },
    {
      "epoch": 0.667783219777686,
      "grad_norm": 0.042574986815452576,
      "learning_rate": 0.0012,
      "loss": 2.672496795654297,
      "step": 8236
    },
    {
      "epoch": 0.667945381802887,
      "grad_norm": 0.07011502236127853,
      "learning_rate": 0.0012,
      "loss": 2.697450876235962,
      "step": 8238
    },
    {
      "epoch": 0.668107543828088,
      "grad_norm": 0.04186716675758362,
      "learning_rate": 0.0012,
      "loss": 2.7081680297851562,
      "step": 8240
    },
    {
      "epoch": 0.6682697058532889,
      "grad_norm": 0.028594372794032097,
      "learning_rate": 0.0012,
      "loss": 2.577232837677002,
      "step": 8242
    },
    {
      "epoch": 0.66843186787849,
      "grad_norm": 0.028401410207152367,
      "learning_rate": 0.0012,
      "loss": 2.5981531143188477,
      "step": 8244
    },
    {
      "epoch": 0.668594029903691,
      "grad_norm": 0.027566460892558098,
      "learning_rate": 0.0012,
      "loss": 2.6209635734558105,
      "step": 8246
    },
    {
      "epoch": 0.6687561919288919,
      "grad_norm": 0.05358128622174263,
      "learning_rate": 0.0012,
      "loss": 2.627187728881836,
      "step": 8248
    },
    {
      "epoch": 0.6689183539540929,
      "grad_norm": 0.05333159863948822,
      "learning_rate": 0.0012,
      "loss": 2.6437172889709473,
      "step": 8250
    },
    {
      "epoch": 0.669080515979294,
      "grad_norm": 0.06454062461853027,
      "learning_rate": 0.0012,
      "loss": 2.6040287017822266,
      "step": 8252
    },
    {
      "epoch": 0.6692426780044949,
      "grad_norm": 0.04828261956572533,
      "learning_rate": 0.0012,
      "loss": 2.673731803894043,
      "step": 8254
    },
    {
      "epoch": 0.6694048400296959,
      "grad_norm": 0.030800700187683105,
      "learning_rate": 0.0012,
      "loss": 2.741844654083252,
      "step": 8256
    },
    {
      "epoch": 0.6695670020548969,
      "grad_norm": 0.038950707763433456,
      "learning_rate": 0.0012,
      "loss": 2.6918411254882812,
      "step": 8258
    },
    {
      "epoch": 0.6697291640800979,
      "grad_norm": 0.02772998996078968,
      "learning_rate": 0.0012,
      "loss": 2.6783039569854736,
      "step": 8260
    },
    {
      "epoch": 0.6698913261052989,
      "grad_norm": 0.027087803930044174,
      "learning_rate": 0.0012,
      "loss": 2.682694435119629,
      "step": 8262
    },
    {
      "epoch": 0.6700534881304999,
      "grad_norm": 0.032706648111343384,
      "learning_rate": 0.0012,
      "loss": 2.586167812347412,
      "step": 8264
    },
    {
      "epoch": 0.6702156501557008,
      "grad_norm": 0.02739170379936695,
      "learning_rate": 0.0012,
      "loss": 2.6828837394714355,
      "step": 8266
    },
    {
      "epoch": 0.6703778121809019,
      "grad_norm": 0.023583652451634407,
      "learning_rate": 0.0012,
      "loss": 2.623688220977783,
      "step": 8268
    },
    {
      "epoch": 0.6705399742061029,
      "grad_norm": 0.042497023940086365,
      "learning_rate": 0.0012,
      "loss": 2.705758810043335,
      "step": 8270
    },
    {
      "epoch": 0.6707021362313038,
      "grad_norm": 0.028066733852028847,
      "learning_rate": 0.0012,
      "loss": 2.6836516857147217,
      "step": 8272
    },
    {
      "epoch": 0.6708642982565048,
      "grad_norm": 0.027835380285978317,
      "learning_rate": 0.0012,
      "loss": 2.6059346199035645,
      "step": 8274
    },
    {
      "epoch": 0.6710264602817059,
      "grad_norm": 0.03143010661005974,
      "learning_rate": 0.0012,
      "loss": 2.7227017879486084,
      "step": 8276
    },
    {
      "epoch": 0.6711886223069068,
      "grad_norm": 0.029432078823447227,
      "learning_rate": 0.0012,
      "loss": 2.675346851348877,
      "step": 8278
    },
    {
      "epoch": 0.6713507843321078,
      "grad_norm": 0.032509367913007736,
      "learning_rate": 0.0012,
      "loss": 2.7014245986938477,
      "step": 8280
    },
    {
      "epoch": 0.6715129463573088,
      "grad_norm": 0.050104133784770966,
      "learning_rate": 0.0012,
      "loss": 2.6357080936431885,
      "step": 8282
    },
    {
      "epoch": 0.6716751083825098,
      "grad_norm": 0.02737520821392536,
      "learning_rate": 0.0012,
      "loss": 2.6799111366271973,
      "step": 8284
    },
    {
      "epoch": 0.6718372704077108,
      "grad_norm": 0.04080406576395035,
      "learning_rate": 0.0012,
      "loss": 2.6702628135681152,
      "step": 8286
    },
    {
      "epoch": 0.6719994324329118,
      "grad_norm": 0.033019229769706726,
      "learning_rate": 0.0012,
      "loss": 2.674022674560547,
      "step": 8288
    },
    {
      "epoch": 0.6721615944581127,
      "grad_norm": 0.02693108096718788,
      "learning_rate": 0.0012,
      "loss": 2.6391232013702393,
      "step": 8290
    },
    {
      "epoch": 0.6723237564833138,
      "grad_norm": 0.024769717827439308,
      "learning_rate": 0.0012,
      "loss": 2.7082457542419434,
      "step": 8292
    },
    {
      "epoch": 0.6724859185085148,
      "grad_norm": 0.034896548837423325,
      "learning_rate": 0.0012,
      "loss": 2.6461713314056396,
      "step": 8294
    },
    {
      "epoch": 0.6726480805337157,
      "grad_norm": 0.030885199084877968,
      "learning_rate": 0.0012,
      "loss": 2.762599468231201,
      "step": 8296
    },
    {
      "epoch": 0.6728102425589167,
      "grad_norm": 0.032396577298641205,
      "learning_rate": 0.0012,
      "loss": 2.722081184387207,
      "step": 8298
    },
    {
      "epoch": 0.6729724045841178,
      "grad_norm": 0.038598719984292984,
      "learning_rate": 0.0012,
      "loss": 2.676454544067383,
      "step": 8300
    },
    {
      "epoch": 0.6731345666093187,
      "grad_norm": 0.025920873507857323,
      "learning_rate": 0.0012,
      "loss": 2.6364846229553223,
      "step": 8302
    },
    {
      "epoch": 0.6732967286345197,
      "grad_norm": 0.025356458500027657,
      "learning_rate": 0.0012,
      "loss": 2.664726734161377,
      "step": 8304
    },
    {
      "epoch": 0.6734588906597208,
      "grad_norm": 0.040351491421461105,
      "learning_rate": 0.0012,
      "loss": 2.591484546661377,
      "step": 8306
    },
    {
      "epoch": 0.6736210526849217,
      "grad_norm": 0.04764646664261818,
      "learning_rate": 0.0012,
      "loss": 2.6647157669067383,
      "step": 8308
    },
    {
      "epoch": 0.6737832147101227,
      "grad_norm": 0.02637697383761406,
      "learning_rate": 0.0012,
      "loss": 2.6663169860839844,
      "step": 8310
    },
    {
      "epoch": 0.6739453767353237,
      "grad_norm": 0.03410610929131508,
      "learning_rate": 0.0012,
      "loss": 2.6315479278564453,
      "step": 8312
    },
    {
      "epoch": 0.6741075387605246,
      "grad_norm": 0.03135222569108009,
      "learning_rate": 0.0012,
      "loss": 2.606955051422119,
      "step": 8314
    },
    {
      "epoch": 0.6742697007857257,
      "grad_norm": 0.02779833786189556,
      "learning_rate": 0.0012,
      "loss": 2.6850695610046387,
      "step": 8316
    },
    {
      "epoch": 0.6744318628109267,
      "grad_norm": 0.042068738490343094,
      "learning_rate": 0.0012,
      "loss": 2.586811065673828,
      "step": 8318
    },
    {
      "epoch": 0.6745940248361276,
      "grad_norm": 0.03166736662387848,
      "learning_rate": 0.0012,
      "loss": 2.6395397186279297,
      "step": 8320
    },
    {
      "epoch": 0.6747561868613287,
      "grad_norm": 0.028107954189181328,
      "learning_rate": 0.0012,
      "loss": 2.703287124633789,
      "step": 8322
    },
    {
      "epoch": 0.6749183488865297,
      "grad_norm": 0.025819437578320503,
      "learning_rate": 0.0012,
      "loss": 2.5824215412139893,
      "step": 8324
    },
    {
      "epoch": 0.6750805109117306,
      "grad_norm": 0.02621254324913025,
      "learning_rate": 0.0012,
      "loss": 2.704874038696289,
      "step": 8326
    },
    {
      "epoch": 0.6752426729369316,
      "grad_norm": 0.025448027998209,
      "learning_rate": 0.0012,
      "loss": 2.718169689178467,
      "step": 8328
    },
    {
      "epoch": 0.6754048349621327,
      "grad_norm": 0.03518032655119896,
      "learning_rate": 0.0012,
      "loss": 2.6771068572998047,
      "step": 8330
    },
    {
      "epoch": 0.6755669969873336,
      "grad_norm": 0.03067641332745552,
      "learning_rate": 0.0012,
      "loss": 2.662783622741699,
      "step": 8332
    },
    {
      "epoch": 0.6757291590125346,
      "grad_norm": 0.04848621040582657,
      "learning_rate": 0.0012,
      "loss": 2.6963515281677246,
      "step": 8334
    },
    {
      "epoch": 0.6758913210377356,
      "grad_norm": 0.07933943718671799,
      "learning_rate": 0.0012,
      "loss": 2.6476974487304688,
      "step": 8336
    },
    {
      "epoch": 0.6760534830629366,
      "grad_norm": 0.04042487591505051,
      "learning_rate": 0.0012,
      "loss": 2.6661124229431152,
      "step": 8338
    },
    {
      "epoch": 0.6762156450881376,
      "grad_norm": 0.024482540786266327,
      "learning_rate": 0.0012,
      "loss": 2.666339635848999,
      "step": 8340
    },
    {
      "epoch": 0.6763778071133386,
      "grad_norm": 0.05608786642551422,
      "learning_rate": 0.0012,
      "loss": 2.725184202194214,
      "step": 8342
    },
    {
      "epoch": 0.6765399691385395,
      "grad_norm": 0.04155753552913666,
      "learning_rate": 0.0012,
      "loss": 2.660576343536377,
      "step": 8344
    },
    {
      "epoch": 0.6767021311637406,
      "grad_norm": 0.023376071825623512,
      "learning_rate": 0.0012,
      "loss": 2.6521501541137695,
      "step": 8346
    },
    {
      "epoch": 0.6768642931889416,
      "grad_norm": 0.033249713480472565,
      "learning_rate": 0.0012,
      "loss": 2.7213244438171387,
      "step": 8348
    },
    {
      "epoch": 0.6770264552141425,
      "grad_norm": 0.03502403944730759,
      "learning_rate": 0.0012,
      "loss": 2.6450843811035156,
      "step": 8350
    },
    {
      "epoch": 0.6771886172393435,
      "grad_norm": 0.02857770025730133,
      "learning_rate": 0.0012,
      "loss": 2.668182373046875,
      "step": 8352
    },
    {
      "epoch": 0.6773507792645446,
      "grad_norm": 0.033826276659965515,
      "learning_rate": 0.0012,
      "loss": 2.709182024002075,
      "step": 8354
    },
    {
      "epoch": 0.6775129412897455,
      "grad_norm": 0.04745963588356972,
      "learning_rate": 0.0012,
      "loss": 2.609622001647949,
      "step": 8356
    },
    {
      "epoch": 0.6776751033149465,
      "grad_norm": 0.0895543172955513,
      "learning_rate": 0.0012,
      "loss": 2.626157283782959,
      "step": 8358
    },
    {
      "epoch": 0.6778372653401475,
      "grad_norm": 0.031232712790369987,
      "learning_rate": 0.0012,
      "loss": 2.6415200233459473,
      "step": 8360
    },
    {
      "epoch": 0.6779994273653485,
      "grad_norm": 0.03867412731051445,
      "learning_rate": 0.0012,
      "loss": 2.6815919876098633,
      "step": 8362
    },
    {
      "epoch": 0.6781615893905495,
      "grad_norm": 0.059500813484191895,
      "learning_rate": 0.0012,
      "loss": 2.687343120574951,
      "step": 8364
    },
    {
      "epoch": 0.6783237514157505,
      "grad_norm": 0.03587467223405838,
      "learning_rate": 0.0012,
      "loss": 2.7168760299682617,
      "step": 8366
    },
    {
      "epoch": 0.6784859134409514,
      "grad_norm": 0.02733183652162552,
      "learning_rate": 0.0012,
      "loss": 2.7560601234436035,
      "step": 8368
    },
    {
      "epoch": 0.6786480754661525,
      "grad_norm": 0.04250570759177208,
      "learning_rate": 0.0012,
      "loss": 2.751984119415283,
      "step": 8370
    },
    {
      "epoch": 0.6788102374913535,
      "grad_norm": 0.044479720294475555,
      "learning_rate": 0.0012,
      "loss": 2.647921562194824,
      "step": 8372
    },
    {
      "epoch": 0.6789723995165544,
      "grad_norm": 0.05337509512901306,
      "learning_rate": 0.0012,
      "loss": 2.732278823852539,
      "step": 8374
    },
    {
      "epoch": 0.6791345615417554,
      "grad_norm": 0.049300871789455414,
      "learning_rate": 0.0012,
      "loss": 2.6410470008850098,
      "step": 8376
    },
    {
      "epoch": 0.6792967235669565,
      "grad_norm": 0.028192266821861267,
      "learning_rate": 0.0012,
      "loss": 2.614175796508789,
      "step": 8378
    },
    {
      "epoch": 0.6794588855921574,
      "grad_norm": 0.05595553293824196,
      "learning_rate": 0.0012,
      "loss": 2.7480885982513428,
      "step": 8380
    },
    {
      "epoch": 0.6796210476173584,
      "grad_norm": 0.03127378970384598,
      "learning_rate": 0.0012,
      "loss": 2.648238182067871,
      "step": 8382
    },
    {
      "epoch": 0.6797832096425594,
      "grad_norm": 0.027469750493764877,
      "learning_rate": 0.0012,
      "loss": 2.5693016052246094,
      "step": 8384
    },
    {
      "epoch": 0.6799453716677604,
      "grad_norm": 0.04272852838039398,
      "learning_rate": 0.0012,
      "loss": 2.6508162021636963,
      "step": 8386
    },
    {
      "epoch": 0.6801075336929614,
      "grad_norm": 0.02721875160932541,
      "learning_rate": 0.0012,
      "loss": 2.639375686645508,
      "step": 8388
    },
    {
      "epoch": 0.6802696957181624,
      "grad_norm": 0.02338951826095581,
      "learning_rate": 0.0012,
      "loss": 2.609992027282715,
      "step": 8390
    },
    {
      "epoch": 0.6804318577433633,
      "grad_norm": 0.031841278076171875,
      "learning_rate": 0.0012,
      "loss": 2.6936941146850586,
      "step": 8392
    },
    {
      "epoch": 0.6805940197685644,
      "grad_norm": 0.02510780096054077,
      "learning_rate": 0.0012,
      "loss": 2.6206107139587402,
      "step": 8394
    },
    {
      "epoch": 0.6807561817937654,
      "grad_norm": 0.03176417574286461,
      "learning_rate": 0.0012,
      "loss": 2.687666893005371,
      "step": 8396
    },
    {
      "epoch": 0.6809183438189663,
      "grad_norm": 0.061602864414453506,
      "learning_rate": 0.0012,
      "loss": 2.684544563293457,
      "step": 8398
    },
    {
      "epoch": 0.6810805058441673,
      "grad_norm": 0.03101024404168129,
      "learning_rate": 0.0012,
      "loss": 2.688723564147949,
      "step": 8400
    },
    {
      "epoch": 0.6812426678693684,
      "grad_norm": 0.029766719788312912,
      "learning_rate": 0.0012,
      "loss": 2.6490976810455322,
      "step": 8402
    },
    {
      "epoch": 0.6814048298945693,
      "grad_norm": 0.026392262428998947,
      "learning_rate": 0.0012,
      "loss": 2.619870901107788,
      "step": 8404
    },
    {
      "epoch": 0.6815669919197703,
      "grad_norm": 0.029441671445965767,
      "learning_rate": 0.0012,
      "loss": 2.734269142150879,
      "step": 8406
    },
    {
      "epoch": 0.6817291539449714,
      "grad_norm": 0.023721037432551384,
      "learning_rate": 0.0012,
      "loss": 2.6988062858581543,
      "step": 8408
    },
    {
      "epoch": 0.6818913159701723,
      "grad_norm": 0.030188629403710365,
      "learning_rate": 0.0012,
      "loss": 2.638692855834961,
      "step": 8410
    },
    {
      "epoch": 0.6820534779953733,
      "grad_norm": 0.028510214760899544,
      "learning_rate": 0.0012,
      "loss": 2.6507701873779297,
      "step": 8412
    },
    {
      "epoch": 0.6822156400205743,
      "grad_norm": 0.0360572412610054,
      "learning_rate": 0.0012,
      "loss": 2.657837152481079,
      "step": 8414
    },
    {
      "epoch": 0.6823778020457753,
      "grad_norm": 0.03679678216576576,
      "learning_rate": 0.0012,
      "loss": 2.6854324340820312,
      "step": 8416
    },
    {
      "epoch": 0.6825399640709763,
      "grad_norm": 0.027476152405142784,
      "learning_rate": 0.0012,
      "loss": 2.667778491973877,
      "step": 8418
    },
    {
      "epoch": 0.6827021260961773,
      "grad_norm": 0.028030846267938614,
      "learning_rate": 0.0012,
      "loss": 2.6541552543640137,
      "step": 8420
    },
    {
      "epoch": 0.6828642881213782,
      "grad_norm": 0.025480078533291817,
      "learning_rate": 0.0012,
      "loss": 2.652846336364746,
      "step": 8422
    },
    {
      "epoch": 0.6830264501465793,
      "grad_norm": 0.045619457960128784,
      "learning_rate": 0.0012,
      "loss": 2.693748950958252,
      "step": 8424
    },
    {
      "epoch": 0.6831886121717803,
      "grad_norm": 0.050953466445207596,
      "learning_rate": 0.0012,
      "loss": 2.654221773147583,
      "step": 8426
    },
    {
      "epoch": 0.6833507741969812,
      "grad_norm": 0.03284648060798645,
      "learning_rate": 0.0012,
      "loss": 2.637650966644287,
      "step": 8428
    },
    {
      "epoch": 0.6835129362221822,
      "grad_norm": 0.028096958994865417,
      "learning_rate": 0.0012,
      "loss": 2.6904702186584473,
      "step": 8430
    },
    {
      "epoch": 0.6836750982473833,
      "grad_norm": 0.0309634730219841,
      "learning_rate": 0.0012,
      "loss": 2.633997917175293,
      "step": 8432
    },
    {
      "epoch": 0.6838372602725842,
      "grad_norm": 0.035346921533346176,
      "learning_rate": 0.0012,
      "loss": 2.6479480266571045,
      "step": 8434
    },
    {
      "epoch": 0.6839994222977852,
      "grad_norm": 0.024192973971366882,
      "learning_rate": 0.0012,
      "loss": 2.69758939743042,
      "step": 8436
    },
    {
      "epoch": 0.6841615843229862,
      "grad_norm": 0.0266108438372612,
      "learning_rate": 0.0012,
      "loss": 2.6568784713745117,
      "step": 8438
    },
    {
      "epoch": 0.6843237463481872,
      "grad_norm": 0.022661281749606133,
      "learning_rate": 0.0012,
      "loss": 2.6878437995910645,
      "step": 8440
    },
    {
      "epoch": 0.6844859083733882,
      "grad_norm": 0.044145870953798294,
      "learning_rate": 0.0012,
      "loss": 2.5807065963745117,
      "step": 8442
    },
    {
      "epoch": 0.6846480703985892,
      "grad_norm": 0.03600097820162773,
      "learning_rate": 0.0012,
      "loss": 2.6535749435424805,
      "step": 8444
    },
    {
      "epoch": 0.6848102324237901,
      "grad_norm": 0.02368786372244358,
      "learning_rate": 0.0012,
      "loss": 2.6561267375946045,
      "step": 8446
    },
    {
      "epoch": 0.6849723944489912,
      "grad_norm": 0.027306770905852318,
      "learning_rate": 0.0012,
      "loss": 2.676327705383301,
      "step": 8448
    },
    {
      "epoch": 0.6851345564741922,
      "grad_norm": 0.025378450751304626,
      "learning_rate": 0.0012,
      "loss": 2.6475448608398438,
      "step": 8450
    },
    {
      "epoch": 0.6852967184993931,
      "grad_norm": 0.027133917436003685,
      "learning_rate": 0.0012,
      "loss": 2.687474250793457,
      "step": 8452
    },
    {
      "epoch": 0.6854588805245941,
      "grad_norm": 0.032738957554101944,
      "learning_rate": 0.0012,
      "loss": 2.595427989959717,
      "step": 8454
    },
    {
      "epoch": 0.6856210425497952,
      "grad_norm": 0.034587666392326355,
      "learning_rate": 0.0012,
      "loss": 2.6702687740325928,
      "step": 8456
    },
    {
      "epoch": 0.6857832045749961,
      "grad_norm": 0.03410341590642929,
      "learning_rate": 0.0012,
      "loss": 2.61977481842041,
      "step": 8458
    },
    {
      "epoch": 0.6859453666001971,
      "grad_norm": 0.02630285732448101,
      "learning_rate": 0.0012,
      "loss": 2.5960922241210938,
      "step": 8460
    },
    {
      "epoch": 0.6861075286253981,
      "grad_norm": 0.026247384026646614,
      "learning_rate": 0.0012,
      "loss": 2.6697816848754883,
      "step": 8462
    },
    {
      "epoch": 0.6862696906505991,
      "grad_norm": 0.04655957967042923,
      "learning_rate": 0.0012,
      "loss": 2.690925121307373,
      "step": 8464
    },
    {
      "epoch": 0.6864318526758001,
      "grad_norm": 0.06016412004828453,
      "learning_rate": 0.0012,
      "loss": 2.6612303256988525,
      "step": 8466
    },
    {
      "epoch": 0.6865940147010011,
      "grad_norm": 0.04557114467024803,
      "learning_rate": 0.0012,
      "loss": 2.7226593494415283,
      "step": 8468
    },
    {
      "epoch": 0.686756176726202,
      "grad_norm": 0.026219157502055168,
      "learning_rate": 0.0012,
      "loss": 2.6943037509918213,
      "step": 8470
    },
    {
      "epoch": 0.6869183387514031,
      "grad_norm": 0.04230811819434166,
      "learning_rate": 0.0012,
      "loss": 2.683415412902832,
      "step": 8472
    },
    {
      "epoch": 0.6870805007766041,
      "grad_norm": 0.05602823197841644,
      "learning_rate": 0.0012,
      "loss": 2.7148585319519043,
      "step": 8474
    },
    {
      "epoch": 0.687242662801805,
      "grad_norm": 0.08466006070375443,
      "learning_rate": 0.0012,
      "loss": 2.6756863594055176,
      "step": 8476
    },
    {
      "epoch": 0.687404824827006,
      "grad_norm": 0.028271446004509926,
      "learning_rate": 0.0012,
      "loss": 2.666656494140625,
      "step": 8478
    },
    {
      "epoch": 0.6875669868522071,
      "grad_norm": 0.023937629535794258,
      "learning_rate": 0.0012,
      "loss": 2.641880989074707,
      "step": 8480
    },
    {
      "epoch": 0.687729148877408,
      "grad_norm": 0.03814785182476044,
      "learning_rate": 0.0012,
      "loss": 2.7281992435455322,
      "step": 8482
    },
    {
      "epoch": 0.687891310902609,
      "grad_norm": 0.03122173622250557,
      "learning_rate": 0.0012,
      "loss": 2.672543525695801,
      "step": 8484
    },
    {
      "epoch": 0.68805347292781,
      "grad_norm": 0.035691842436790466,
      "learning_rate": 0.0012,
      "loss": 2.6550750732421875,
      "step": 8486
    },
    {
      "epoch": 0.688215634953011,
      "grad_norm": 0.05853237211704254,
      "learning_rate": 0.0012,
      "loss": 2.697315216064453,
      "step": 8488
    },
    {
      "epoch": 0.688377796978212,
      "grad_norm": 0.031405285000801086,
      "learning_rate": 0.0012,
      "loss": 2.6354291439056396,
      "step": 8490
    },
    {
      "epoch": 0.688539959003413,
      "grad_norm": 0.03629547357559204,
      "learning_rate": 0.0012,
      "loss": 2.6952853202819824,
      "step": 8492
    },
    {
      "epoch": 0.688702121028614,
      "grad_norm": 0.04336605221033096,
      "learning_rate": 0.0012,
      "loss": 2.642889976501465,
      "step": 8494
    },
    {
      "epoch": 0.688864283053815,
      "grad_norm": 0.03998495265841484,
      "learning_rate": 0.0012,
      "loss": 2.636319160461426,
      "step": 8496
    },
    {
      "epoch": 0.689026445079016,
      "grad_norm": 0.031240206211805344,
      "learning_rate": 0.0012,
      "loss": 2.6297683715820312,
      "step": 8498
    },
    {
      "epoch": 0.6891886071042169,
      "grad_norm": 0.08126174658536911,
      "learning_rate": 0.0012,
      "loss": 2.633967399597168,
      "step": 8500
    },
    {
      "epoch": 0.689350769129418,
      "grad_norm": 0.07611432671546936,
      "learning_rate": 0.0012,
      "loss": 2.683037281036377,
      "step": 8502
    },
    {
      "epoch": 0.689512931154619,
      "grad_norm": 0.028107257559895515,
      "learning_rate": 0.0012,
      "loss": 2.7002415657043457,
      "step": 8504
    },
    {
      "epoch": 0.6896750931798199,
      "grad_norm": 0.0537600964307785,
      "learning_rate": 0.0012,
      "loss": 2.6840944290161133,
      "step": 8506
    },
    {
      "epoch": 0.6898372552050209,
      "grad_norm": 0.06788382679224014,
      "learning_rate": 0.0012,
      "loss": 2.7010574340820312,
      "step": 8508
    },
    {
      "epoch": 0.689999417230222,
      "grad_norm": 0.0246721338480711,
      "learning_rate": 0.0012,
      "loss": 2.625377655029297,
      "step": 8510
    },
    {
      "epoch": 0.6901615792554229,
      "grad_norm": 0.0683896616101265,
      "learning_rate": 0.0012,
      "loss": 2.7005653381347656,
      "step": 8512
    },
    {
      "epoch": 0.6903237412806239,
      "grad_norm": 0.03329528495669365,
      "learning_rate": 0.0012,
      "loss": 2.63729190826416,
      "step": 8514
    },
    {
      "epoch": 0.6904859033058249,
      "grad_norm": 0.044985901564359665,
      "learning_rate": 0.0012,
      "loss": 2.692992687225342,
      "step": 8516
    },
    {
      "epoch": 0.6906480653310259,
      "grad_norm": 0.05154905840754509,
      "learning_rate": 0.0012,
      "loss": 2.7204484939575195,
      "step": 8518
    },
    {
      "epoch": 0.6908102273562269,
      "grad_norm": 0.02527666836977005,
      "learning_rate": 0.0012,
      "loss": 2.7272303104400635,
      "step": 8520
    },
    {
      "epoch": 0.6909723893814279,
      "grad_norm": 0.0691557377576828,
      "learning_rate": 0.0012,
      "loss": 2.6161739826202393,
      "step": 8522
    },
    {
      "epoch": 0.6911345514066288,
      "grad_norm": 0.038511354476213455,
      "learning_rate": 0.0012,
      "loss": 2.668657064437866,
      "step": 8524
    },
    {
      "epoch": 0.6912967134318299,
      "grad_norm": 0.02717726118862629,
      "learning_rate": 0.0012,
      "loss": 2.70289945602417,
      "step": 8526
    },
    {
      "epoch": 0.6914588754570309,
      "grad_norm": 0.04280782490968704,
      "learning_rate": 0.0012,
      "loss": 2.693016767501831,
      "step": 8528
    },
    {
      "epoch": 0.6916210374822318,
      "grad_norm": 0.027911372482776642,
      "learning_rate": 0.0012,
      "loss": 2.6557769775390625,
      "step": 8530
    },
    {
      "epoch": 0.6917831995074328,
      "grad_norm": 0.027752060443162918,
      "learning_rate": 0.0012,
      "loss": 2.6207988262176514,
      "step": 8532
    },
    {
      "epoch": 0.6919453615326339,
      "grad_norm": 0.04785566404461861,
      "learning_rate": 0.0012,
      "loss": 2.636080741882324,
      "step": 8534
    },
    {
      "epoch": 0.6921075235578348,
      "grad_norm": 0.06546515226364136,
      "learning_rate": 0.0012,
      "loss": 2.689380645751953,
      "step": 8536
    },
    {
      "epoch": 0.6922696855830358,
      "grad_norm": 0.034295499324798584,
      "learning_rate": 0.0012,
      "loss": 2.7401418685913086,
      "step": 8538
    },
    {
      "epoch": 0.6924318476082368,
      "grad_norm": 0.041001662611961365,
      "learning_rate": 0.0012,
      "loss": 2.6581597328186035,
      "step": 8540
    },
    {
      "epoch": 0.6925940096334378,
      "grad_norm": 0.05949440598487854,
      "learning_rate": 0.0012,
      "loss": 2.655618667602539,
      "step": 8542
    },
    {
      "epoch": 0.6927561716586388,
      "grad_norm": 0.028999263420701027,
      "learning_rate": 0.0012,
      "loss": 2.7180628776550293,
      "step": 8544
    },
    {
      "epoch": 0.6929183336838398,
      "grad_norm": 0.02992015890777111,
      "learning_rate": 0.0012,
      "loss": 2.665790557861328,
      "step": 8546
    },
    {
      "epoch": 0.6930804957090407,
      "grad_norm": 0.03048398531973362,
      "learning_rate": 0.0012,
      "loss": 2.683086633682251,
      "step": 8548
    },
    {
      "epoch": 0.6932426577342418,
      "grad_norm": 0.02529023215174675,
      "learning_rate": 0.0012,
      "loss": 2.6539220809936523,
      "step": 8550
    },
    {
      "epoch": 0.6934048197594428,
      "grad_norm": 0.028478646650910378,
      "learning_rate": 0.0012,
      "loss": 2.6665828227996826,
      "step": 8552
    },
    {
      "epoch": 0.6935669817846437,
      "grad_norm": 0.02760060876607895,
      "learning_rate": 0.0012,
      "loss": 2.629783868789673,
      "step": 8554
    },
    {
      "epoch": 0.6937291438098447,
      "grad_norm": 0.031182287260890007,
      "learning_rate": 0.0012,
      "loss": 2.6407713890075684,
      "step": 8556
    },
    {
      "epoch": 0.6938913058350458,
      "grad_norm": 0.03083610162138939,
      "learning_rate": 0.0012,
      "loss": 2.6316328048706055,
      "step": 8558
    },
    {
      "epoch": 0.6940534678602467,
      "grad_norm": 0.025884415954351425,
      "learning_rate": 0.0012,
      "loss": 2.6744978427886963,
      "step": 8560
    },
    {
      "epoch": 0.6942156298854477,
      "grad_norm": 0.025235792621970177,
      "learning_rate": 0.0012,
      "loss": 2.71779727935791,
      "step": 8562
    },
    {
      "epoch": 0.6943777919106487,
      "grad_norm": 0.023856516927480698,
      "learning_rate": 0.0012,
      "loss": 2.732503890991211,
      "step": 8564
    },
    {
      "epoch": 0.6945399539358497,
      "grad_norm": 0.036000095307826996,
      "learning_rate": 0.0012,
      "loss": 2.689201831817627,
      "step": 8566
    },
    {
      "epoch": 0.6947021159610507,
      "grad_norm": 0.029448585584759712,
      "learning_rate": 0.0012,
      "loss": 2.6442220211029053,
      "step": 8568
    },
    {
      "epoch": 0.6948642779862517,
      "grad_norm": 0.04132663458585739,
      "learning_rate": 0.0012,
      "loss": 2.6916251182556152,
      "step": 8570
    },
    {
      "epoch": 0.6950264400114526,
      "grad_norm": 0.027083249762654305,
      "learning_rate": 0.0012,
      "loss": 2.7002267837524414,
      "step": 8572
    },
    {
      "epoch": 0.6951886020366537,
      "grad_norm": 0.029331088066101074,
      "learning_rate": 0.0012,
      "loss": 2.7153563499450684,
      "step": 8574
    },
    {
      "epoch": 0.6953507640618547,
      "grad_norm": 0.023203805088996887,
      "learning_rate": 0.0012,
      "loss": 2.7307865619659424,
      "step": 8576
    },
    {
      "epoch": 0.6955129260870556,
      "grad_norm": 0.03641580417752266,
      "learning_rate": 0.0012,
      "loss": 2.671090602874756,
      "step": 8578
    },
    {
      "epoch": 0.6956750881122566,
      "grad_norm": 0.03487534448504448,
      "learning_rate": 0.0012,
      "loss": 2.626248836517334,
      "step": 8580
    },
    {
      "epoch": 0.6958372501374577,
      "grad_norm": 0.027494877576828003,
      "learning_rate": 0.0012,
      "loss": 2.749505043029785,
      "step": 8582
    },
    {
      "epoch": 0.6959994121626586,
      "grad_norm": 0.02450578473508358,
      "learning_rate": 0.0012,
      "loss": 2.680988311767578,
      "step": 8584
    },
    {
      "epoch": 0.6961615741878596,
      "grad_norm": 0.0297468863427639,
      "learning_rate": 0.0012,
      "loss": 2.665536403656006,
      "step": 8586
    },
    {
      "epoch": 0.6963237362130607,
      "grad_norm": 0.03979470580816269,
      "learning_rate": 0.0012,
      "loss": 2.656566619873047,
      "step": 8588
    },
    {
      "epoch": 0.6964858982382616,
      "grad_norm": 0.02503410167992115,
      "learning_rate": 0.0012,
      "loss": 2.733855962753296,
      "step": 8590
    },
    {
      "epoch": 0.6966480602634626,
      "grad_norm": 0.028731439262628555,
      "learning_rate": 0.0012,
      "loss": 2.744328022003174,
      "step": 8592
    },
    {
      "epoch": 0.6968102222886636,
      "grad_norm": 0.024757742881774902,
      "learning_rate": 0.0012,
      "loss": 2.6032216548919678,
      "step": 8594
    },
    {
      "epoch": 0.6969723843138645,
      "grad_norm": 0.04004983231425285,
      "learning_rate": 0.0012,
      "loss": 2.641725540161133,
      "step": 8596
    },
    {
      "epoch": 0.6971345463390656,
      "grad_norm": 0.032018955796957016,
      "learning_rate": 0.0012,
      "loss": 2.638659954071045,
      "step": 8598
    },
    {
      "epoch": 0.6972967083642666,
      "grad_norm": 0.03440537676215172,
      "learning_rate": 0.0012,
      "loss": 2.653308391571045,
      "step": 8600
    },
    {
      "epoch": 0.6974588703894675,
      "grad_norm": 0.04231015220284462,
      "learning_rate": 0.0012,
      "loss": 2.700446128845215,
      "step": 8602
    },
    {
      "epoch": 0.6976210324146686,
      "grad_norm": 0.0559907965362072,
      "learning_rate": 0.0012,
      "loss": 2.6293230056762695,
      "step": 8604
    },
    {
      "epoch": 0.6977831944398696,
      "grad_norm": 0.044112011790275574,
      "learning_rate": 0.0012,
      "loss": 2.7104406356811523,
      "step": 8606
    },
    {
      "epoch": 0.6979453564650705,
      "grad_norm": 0.03368498384952545,
      "learning_rate": 0.0012,
      "loss": 2.7509875297546387,
      "step": 8608
    },
    {
      "epoch": 0.6981075184902715,
      "grad_norm": 0.04544554278254509,
      "learning_rate": 0.0012,
      "loss": 2.655486822128296,
      "step": 8610
    },
    {
      "epoch": 0.6982696805154726,
      "grad_norm": 0.028834184631705284,
      "learning_rate": 0.0012,
      "loss": 2.7182626724243164,
      "step": 8612
    },
    {
      "epoch": 0.6984318425406735,
      "grad_norm": 0.027375007048249245,
      "learning_rate": 0.0012,
      "loss": 2.6922035217285156,
      "step": 8614
    },
    {
      "epoch": 0.6985940045658745,
      "grad_norm": 0.0241688285022974,
      "learning_rate": 0.0012,
      "loss": 2.6466567516326904,
      "step": 8616
    },
    {
      "epoch": 0.6987561665910755,
      "grad_norm": 0.036647483706474304,
      "learning_rate": 0.0012,
      "loss": 2.6766762733459473,
      "step": 8618
    },
    {
      "epoch": 0.6989183286162765,
      "grad_norm": 0.03546033054590225,
      "learning_rate": 0.0012,
      "loss": 2.640211582183838,
      "step": 8620
    },
    {
      "epoch": 0.6990804906414775,
      "grad_norm": 0.03331276774406433,
      "learning_rate": 0.0012,
      "loss": 2.6770853996276855,
      "step": 8622
    },
    {
      "epoch": 0.6992426526666785,
      "grad_norm": 0.025953415781259537,
      "learning_rate": 0.0012,
      "loss": 2.6946654319763184,
      "step": 8624
    },
    {
      "epoch": 0.6994048146918794,
      "grad_norm": 0.035762984305620193,
      "learning_rate": 0.0012,
      "loss": 2.6450839042663574,
      "step": 8626
    },
    {
      "epoch": 0.6995669767170805,
      "grad_norm": 0.04406505823135376,
      "learning_rate": 0.0012,
      "loss": 2.7027571201324463,
      "step": 8628
    },
    {
      "epoch": 0.6997291387422815,
      "grad_norm": 0.04153851792216301,
      "learning_rate": 0.0012,
      "loss": 2.6152923107147217,
      "step": 8630
    },
    {
      "epoch": 0.6998913007674824,
      "grad_norm": 0.02696523629128933,
      "learning_rate": 0.0012,
      "loss": 2.576767921447754,
      "step": 8632
    },
    {
      "epoch": 0.7000534627926834,
      "grad_norm": 0.05947086215019226,
      "learning_rate": 0.0012,
      "loss": 2.6885104179382324,
      "step": 8634
    },
    {
      "epoch": 0.7002156248178845,
      "grad_norm": 0.08442918211221695,
      "learning_rate": 0.0012,
      "loss": 2.6674845218658447,
      "step": 8636
    },
    {
      "epoch": 0.7003777868430854,
      "grad_norm": 0.023591477423906326,
      "learning_rate": 0.0012,
      "loss": 2.654860496520996,
      "step": 8638
    },
    {
      "epoch": 0.7005399488682864,
      "grad_norm": 0.06138006970286369,
      "learning_rate": 0.0012,
      "loss": 2.6902337074279785,
      "step": 8640
    },
    {
      "epoch": 0.7007021108934874,
      "grad_norm": 0.06022512540221214,
      "learning_rate": 0.0012,
      "loss": 2.6362552642822266,
      "step": 8642
    },
    {
      "epoch": 0.7008642729186884,
      "grad_norm": 0.028080588206648827,
      "learning_rate": 0.0012,
      "loss": 2.705230474472046,
      "step": 8644
    },
    {
      "epoch": 0.7010264349438894,
      "grad_norm": 0.05317092314362526,
      "learning_rate": 0.0012,
      "loss": 2.6654810905456543,
      "step": 8646
    },
    {
      "epoch": 0.7011885969690904,
      "grad_norm": 0.04560936987400055,
      "learning_rate": 0.0012,
      "loss": 2.6646978855133057,
      "step": 8648
    },
    {
      "epoch": 0.7013507589942913,
      "grad_norm": 0.029701534658670425,
      "learning_rate": 0.0012,
      "loss": 2.6622676849365234,
      "step": 8650
    },
    {
      "epoch": 0.7015129210194924,
      "grad_norm": 0.03643471375107765,
      "learning_rate": 0.0012,
      "loss": 2.6696600914001465,
      "step": 8652
    },
    {
      "epoch": 0.7016750830446934,
      "grad_norm": 0.03002760000526905,
      "learning_rate": 0.0012,
      "loss": 2.694554090499878,
      "step": 8654
    },
    {
      "epoch": 0.7018372450698944,
      "grad_norm": 0.04686393216252327,
      "learning_rate": 0.0012,
      "loss": 2.7041921615600586,
      "step": 8656
    },
    {
      "epoch": 0.7019994070950953,
      "grad_norm": 0.03129863366484642,
      "learning_rate": 0.0012,
      "loss": 2.6476500034332275,
      "step": 8658
    },
    {
      "epoch": 0.7021615691202964,
      "grad_norm": 0.029012667015194893,
      "learning_rate": 0.0012,
      "loss": 2.65971040725708,
      "step": 8660
    },
    {
      "epoch": 0.7023237311454974,
      "grad_norm": 0.03651824966073036,
      "learning_rate": 0.0012,
      "loss": 2.6084179878234863,
      "step": 8662
    },
    {
      "epoch": 0.7024858931706983,
      "grad_norm": 0.03287771716713905,
      "learning_rate": 0.0012,
      "loss": 2.662571907043457,
      "step": 8664
    },
    {
      "epoch": 0.7026480551958993,
      "grad_norm": 0.025160402059555054,
      "learning_rate": 0.0012,
      "loss": 2.7155609130859375,
      "step": 8666
    },
    {
      "epoch": 0.7028102172211004,
      "grad_norm": 0.028307685628533363,
      "learning_rate": 0.0012,
      "loss": 2.6635165214538574,
      "step": 8668
    },
    {
      "epoch": 0.7029723792463013,
      "grad_norm": 0.036019910126924515,
      "learning_rate": 0.0012,
      "loss": 2.6713953018188477,
      "step": 8670
    },
    {
      "epoch": 0.7031345412715023,
      "grad_norm": 0.03194982931017876,
      "learning_rate": 0.0012,
      "loss": 2.6899189949035645,
      "step": 8672
    },
    {
      "epoch": 0.7032967032967034,
      "grad_norm": 0.04383343085646629,
      "learning_rate": 0.0012,
      "loss": 2.673558235168457,
      "step": 8674
    },
    {
      "epoch": 0.7034588653219043,
      "grad_norm": 0.033808931708335876,
      "learning_rate": 0.0012,
      "loss": 2.702840566635132,
      "step": 8676
    },
    {
      "epoch": 0.7036210273471053,
      "grad_norm": 0.055316925048828125,
      "learning_rate": 0.0012,
      "loss": 2.6778688430786133,
      "step": 8678
    },
    {
      "epoch": 0.7037831893723063,
      "grad_norm": 0.04853253439068794,
      "learning_rate": 0.0012,
      "loss": 2.644131660461426,
      "step": 8680
    },
    {
      "epoch": 0.7039453513975072,
      "grad_norm": 0.027682246640324593,
      "learning_rate": 0.0012,
      "loss": 2.649571180343628,
      "step": 8682
    },
    {
      "epoch": 0.7041075134227083,
      "grad_norm": 0.02585749328136444,
      "learning_rate": 0.0012,
      "loss": 2.62388014793396,
      "step": 8684
    },
    {
      "epoch": 0.7042696754479093,
      "grad_norm": 0.029642188921570778,
      "learning_rate": 0.0012,
      "loss": 2.6023077964782715,
      "step": 8686
    },
    {
      "epoch": 0.7044318374731102,
      "grad_norm": 0.04123442992568016,
      "learning_rate": 0.0012,
      "loss": 2.603579521179199,
      "step": 8688
    },
    {
      "epoch": 0.7045939994983113,
      "grad_norm": 0.04627152904868126,
      "learning_rate": 0.0012,
      "loss": 2.68182373046875,
      "step": 8690
    },
    {
      "epoch": 0.7047561615235123,
      "grad_norm": 0.04478532448410988,
      "learning_rate": 0.0012,
      "loss": 2.725367546081543,
      "step": 8692
    },
    {
      "epoch": 0.7049183235487132,
      "grad_norm": 0.03249760717153549,
      "learning_rate": 0.0012,
      "loss": 2.660846471786499,
      "step": 8694
    },
    {
      "epoch": 0.7050804855739142,
      "grad_norm": 0.023941827937960625,
      "learning_rate": 0.0012,
      "loss": 2.6652450561523438,
      "step": 8696
    },
    {
      "epoch": 0.7052426475991153,
      "grad_norm": 0.038955654948949814,
      "learning_rate": 0.0012,
      "loss": 2.7403156757354736,
      "step": 8698
    },
    {
      "epoch": 0.7054048096243162,
      "grad_norm": 0.02382783405482769,
      "learning_rate": 0.0012,
      "loss": 2.6578266620635986,
      "step": 8700
    },
    {
      "epoch": 0.7055669716495172,
      "grad_norm": 0.03803052380681038,
      "learning_rate": 0.0012,
      "loss": 2.703348159790039,
      "step": 8702
    },
    {
      "epoch": 0.7057291336747182,
      "grad_norm": 0.03514324873685837,
      "learning_rate": 0.0012,
      "loss": 2.6547677516937256,
      "step": 8704
    },
    {
      "epoch": 0.7058912956999192,
      "grad_norm": 0.05462735891342163,
      "learning_rate": 0.0012,
      "loss": 2.6237149238586426,
      "step": 8706
    },
    {
      "epoch": 0.7060534577251202,
      "grad_norm": 0.026031019166111946,
      "learning_rate": 0.0012,
      "loss": 2.677155017852783,
      "step": 8708
    },
    {
      "epoch": 0.7062156197503212,
      "grad_norm": 0.037388402968645096,
      "learning_rate": 0.0012,
      "loss": 2.677119255065918,
      "step": 8710
    },
    {
      "epoch": 0.7063777817755221,
      "grad_norm": 0.04375799000263214,
      "learning_rate": 0.0012,
      "loss": 2.650489568710327,
      "step": 8712
    },
    {
      "epoch": 0.7065399438007232,
      "grad_norm": 0.0444779098033905,
      "learning_rate": 0.0012,
      "loss": 2.6550345420837402,
      "step": 8714
    },
    {
      "epoch": 0.7067021058259242,
      "grad_norm": 0.05702374130487442,
      "learning_rate": 0.0012,
      "loss": 2.7060928344726562,
      "step": 8716
    },
    {
      "epoch": 0.7068642678511251,
      "grad_norm": 0.05732167884707451,
      "learning_rate": 0.0012,
      "loss": 2.752884864807129,
      "step": 8718
    },
    {
      "epoch": 0.7070264298763261,
      "grad_norm": 0.024413149803876877,
      "learning_rate": 0.0012,
      "loss": 2.63112211227417,
      "step": 8720
    },
    {
      "epoch": 0.7071885919015272,
      "grad_norm": 0.023888010531663895,
      "learning_rate": 0.0012,
      "loss": 2.6713509559631348,
      "step": 8722
    },
    {
      "epoch": 0.7073507539267281,
      "grad_norm": 0.02919260784983635,
      "learning_rate": 0.0012,
      "loss": 2.6307449340820312,
      "step": 8724
    },
    {
      "epoch": 0.7075129159519291,
      "grad_norm": 0.041572585701942444,
      "learning_rate": 0.0012,
      "loss": 2.7026124000549316,
      "step": 8726
    },
    {
      "epoch": 0.7076750779771301,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0012,
      "loss": 2.7041800022125244,
      "step": 8728
    },
    {
      "epoch": 0.7078372400023311,
      "grad_norm": 0.03134343773126602,
      "learning_rate": 0.0012,
      "loss": 2.6935551166534424,
      "step": 8730
    },
    {
      "epoch": 0.7079994020275321,
      "grad_norm": 0.026853544637560844,
      "learning_rate": 0.0012,
      "loss": 2.6789469718933105,
      "step": 8732
    },
    {
      "epoch": 0.7081615640527331,
      "grad_norm": 0.02570568397641182,
      "learning_rate": 0.0012,
      "loss": 2.708606243133545,
      "step": 8734
    },
    {
      "epoch": 0.708323726077934,
      "grad_norm": 0.02510865218937397,
      "learning_rate": 0.0012,
      "loss": 2.6181325912475586,
      "step": 8736
    },
    {
      "epoch": 0.7084858881031351,
      "grad_norm": 0.025275159627199173,
      "learning_rate": 0.0012,
      "loss": 2.6941967010498047,
      "step": 8738
    },
    {
      "epoch": 0.7086480501283361,
      "grad_norm": 0.02937457710504532,
      "learning_rate": 0.0012,
      "loss": 2.6896934509277344,
      "step": 8740
    },
    {
      "epoch": 0.708810212153537,
      "grad_norm": 0.02679615281522274,
      "learning_rate": 0.0012,
      "loss": 2.5955920219421387,
      "step": 8742
    },
    {
      "epoch": 0.708972374178738,
      "grad_norm": 0.05134962871670723,
      "learning_rate": 0.0012,
      "loss": 2.6435036659240723,
      "step": 8744
    },
    {
      "epoch": 0.7091345362039391,
      "grad_norm": 0.06453116983175278,
      "learning_rate": 0.0012,
      "loss": 2.690962791442871,
      "step": 8746
    },
    {
      "epoch": 0.70929669822914,
      "grad_norm": 0.028858955949544907,
      "learning_rate": 0.0012,
      "loss": 2.602438449859619,
      "step": 8748
    },
    {
      "epoch": 0.709458860254341,
      "grad_norm": 0.041096948087215424,
      "learning_rate": 0.0012,
      "loss": 2.6243221759796143,
      "step": 8750
    },
    {
      "epoch": 0.709621022279542,
      "grad_norm": 0.04052338749170303,
      "learning_rate": 0.0012,
      "loss": 2.662022590637207,
      "step": 8752
    },
    {
      "epoch": 0.709783184304743,
      "grad_norm": 0.03554660081863403,
      "learning_rate": 0.0012,
      "loss": 2.6637845039367676,
      "step": 8754
    },
    {
      "epoch": 0.709945346329944,
      "grad_norm": 0.027388600632548332,
      "learning_rate": 0.0012,
      "loss": 2.653128147125244,
      "step": 8756
    },
    {
      "epoch": 0.710107508355145,
      "grad_norm": 0.04184501990675926,
      "learning_rate": 0.0012,
      "loss": 2.7070865631103516,
      "step": 8758
    },
    {
      "epoch": 0.7102696703803459,
      "grad_norm": 0.037382274866104126,
      "learning_rate": 0.0012,
      "loss": 2.7385964393615723,
      "step": 8760
    },
    {
      "epoch": 0.710431832405547,
      "grad_norm": 0.03833373636007309,
      "learning_rate": 0.0012,
      "loss": 2.6522719860076904,
      "step": 8762
    },
    {
      "epoch": 0.710593994430748,
      "grad_norm": 0.02665957435965538,
      "learning_rate": 0.0012,
      "loss": 2.701023578643799,
      "step": 8764
    },
    {
      "epoch": 0.7107561564559489,
      "grad_norm": 0.03575578331947327,
      "learning_rate": 0.0012,
      "loss": 2.6858155727386475,
      "step": 8766
    },
    {
      "epoch": 0.71091831848115,
      "grad_norm": 0.04260304942727089,
      "learning_rate": 0.0012,
      "loss": 2.674577236175537,
      "step": 8768
    },
    {
      "epoch": 0.711080480506351,
      "grad_norm": 0.023979296907782555,
      "learning_rate": 0.0012,
      "loss": 2.7211661338806152,
      "step": 8770
    },
    {
      "epoch": 0.7112426425315519,
      "grad_norm": 0.05790523439645767,
      "learning_rate": 0.0012,
      "loss": 2.6001930236816406,
      "step": 8772
    },
    {
      "epoch": 0.7114048045567529,
      "grad_norm": 0.05977874994277954,
      "learning_rate": 0.0012,
      "loss": 2.6719164848327637,
      "step": 8774
    },
    {
      "epoch": 0.711566966581954,
      "grad_norm": 0.03805151954293251,
      "learning_rate": 0.0012,
      "loss": 2.709477663040161,
      "step": 8776
    },
    {
      "epoch": 0.7117291286071549,
      "grad_norm": 0.02434527315199375,
      "learning_rate": 0.0012,
      "loss": 2.6970386505126953,
      "step": 8778
    },
    {
      "epoch": 0.7118912906323559,
      "grad_norm": 0.026942649856209755,
      "learning_rate": 0.0012,
      "loss": 2.6202683448791504,
      "step": 8780
    },
    {
      "epoch": 0.7120534526575569,
      "grad_norm": 0.025718890130519867,
      "learning_rate": 0.0012,
      "loss": 2.702274799346924,
      "step": 8782
    },
    {
      "epoch": 0.7122156146827578,
      "grad_norm": 0.02237187698483467,
      "learning_rate": 0.0012,
      "loss": 2.6602654457092285,
      "step": 8784
    },
    {
      "epoch": 0.7123777767079589,
      "grad_norm": 0.03925709053874016,
      "learning_rate": 0.0012,
      "loss": 2.6119282245635986,
      "step": 8786
    },
    {
      "epoch": 0.7125399387331599,
      "grad_norm": 0.02924446202814579,
      "learning_rate": 0.0012,
      "loss": 2.651475429534912,
      "step": 8788
    },
    {
      "epoch": 0.7127021007583608,
      "grad_norm": 0.02643553540110588,
      "learning_rate": 0.0012,
      "loss": 2.674370765686035,
      "step": 8790
    },
    {
      "epoch": 0.7128642627835619,
      "grad_norm": 0.024590203538537025,
      "learning_rate": 0.0012,
      "loss": 2.6905899047851562,
      "step": 8792
    },
    {
      "epoch": 0.7130264248087629,
      "grad_norm": 0.05198843032121658,
      "learning_rate": 0.0012,
      "loss": 2.727264881134033,
      "step": 8794
    },
    {
      "epoch": 0.7131885868339638,
      "grad_norm": 0.048136018216609955,
      "learning_rate": 0.0012,
      "loss": 2.7227816581726074,
      "step": 8796
    },
    {
      "epoch": 0.7133507488591648,
      "grad_norm": 0.05448898673057556,
      "learning_rate": 0.0012,
      "loss": 2.674065589904785,
      "step": 8798
    },
    {
      "epoch": 0.7135129108843659,
      "grad_norm": 0.042521052062511444,
      "learning_rate": 0.0012,
      "loss": 2.639023780822754,
      "step": 8800
    },
    {
      "epoch": 0.7136750729095668,
      "grad_norm": 0.026325708255171776,
      "learning_rate": 0.0012,
      "loss": 2.6755051612854004,
      "step": 8802
    },
    {
      "epoch": 0.7138372349347678,
      "grad_norm": 0.0450839027762413,
      "learning_rate": 0.0012,
      "loss": 2.6532936096191406,
      "step": 8804
    },
    {
      "epoch": 0.7139993969599688,
      "grad_norm": 0.029921824112534523,
      "learning_rate": 0.0012,
      "loss": 2.644577741622925,
      "step": 8806
    },
    {
      "epoch": 0.7141615589851698,
      "grad_norm": 0.022967180237174034,
      "learning_rate": 0.0012,
      "loss": 2.6730313301086426,
      "step": 8808
    },
    {
      "epoch": 0.7143237210103708,
      "grad_norm": 0.03657757490873337,
      "learning_rate": 0.0012,
      "loss": 2.631755828857422,
      "step": 8810
    },
    {
      "epoch": 0.7144858830355718,
      "grad_norm": 0.032681673765182495,
      "learning_rate": 0.0012,
      "loss": 2.635988235473633,
      "step": 8812
    },
    {
      "epoch": 0.7146480450607727,
      "grad_norm": 0.03559032082557678,
      "learning_rate": 0.0012,
      "loss": 2.673553228378296,
      "step": 8814
    },
    {
      "epoch": 0.7148102070859738,
      "grad_norm": 0.036689694970846176,
      "learning_rate": 0.0012,
      "loss": 2.708876132965088,
      "step": 8816
    },
    {
      "epoch": 0.7149723691111748,
      "grad_norm": 0.029014553874731064,
      "learning_rate": 0.0012,
      "loss": 2.7018890380859375,
      "step": 8818
    },
    {
      "epoch": 0.7151345311363757,
      "grad_norm": 0.024738289415836334,
      "learning_rate": 0.0012,
      "loss": 2.6608405113220215,
      "step": 8820
    },
    {
      "epoch": 0.7152966931615767,
      "grad_norm": 0.027581362053751945,
      "learning_rate": 0.0012,
      "loss": 2.629901647567749,
      "step": 8822
    },
    {
      "epoch": 0.7154588551867778,
      "grad_norm": 0.02327108196914196,
      "learning_rate": 0.0012,
      "loss": 2.6115784645080566,
      "step": 8824
    },
    {
      "epoch": 0.7156210172119787,
      "grad_norm": 0.02497825399041176,
      "learning_rate": 0.0012,
      "loss": 2.6790175437927246,
      "step": 8826
    },
    {
      "epoch": 0.7157831792371797,
      "grad_norm": 0.024849766865372658,
      "learning_rate": 0.0012,
      "loss": 2.6175098419189453,
      "step": 8828
    },
    {
      "epoch": 0.7159453412623807,
      "grad_norm": 0.044369202107191086,
      "learning_rate": 0.0012,
      "loss": 2.6664538383483887,
      "step": 8830
    },
    {
      "epoch": 0.7161075032875817,
      "grad_norm": 0.05106721818447113,
      "learning_rate": 0.0012,
      "loss": 2.7160868644714355,
      "step": 8832
    },
    {
      "epoch": 0.7162696653127827,
      "grad_norm": 0.037341322749853134,
      "learning_rate": 0.0012,
      "loss": 2.695560932159424,
      "step": 8834
    },
    {
      "epoch": 0.7164318273379837,
      "grad_norm": 0.029253849759697914,
      "learning_rate": 0.0012,
      "loss": 2.6728711128234863,
      "step": 8836
    },
    {
      "epoch": 0.7165939893631846,
      "grad_norm": 0.02638118900358677,
      "learning_rate": 0.0012,
      "loss": 2.7056775093078613,
      "step": 8838
    },
    {
      "epoch": 0.7167561513883857,
      "grad_norm": 0.03451872989535332,
      "learning_rate": 0.0012,
      "loss": 2.7282705307006836,
      "step": 8840
    },
    {
      "epoch": 0.7169183134135867,
      "grad_norm": 0.030410829931497574,
      "learning_rate": 0.0012,
      "loss": 2.666317939758301,
      "step": 8842
    },
    {
      "epoch": 0.7170804754387876,
      "grad_norm": 0.029710479080677032,
      "learning_rate": 0.0012,
      "loss": 2.719992160797119,
      "step": 8844
    },
    {
      "epoch": 0.7172426374639886,
      "grad_norm": 0.02775103971362114,
      "learning_rate": 0.0012,
      "loss": 2.6212148666381836,
      "step": 8846
    },
    {
      "epoch": 0.7174047994891897,
      "grad_norm": 0.023621805012226105,
      "learning_rate": 0.0012,
      "loss": 2.6572303771972656,
      "step": 8848
    },
    {
      "epoch": 0.7175669615143906,
      "grad_norm": 0.06474201381206512,
      "learning_rate": 0.0012,
      "loss": 2.7526254653930664,
      "step": 8850
    },
    {
      "epoch": 0.7177291235395916,
      "grad_norm": 0.0460171140730381,
      "learning_rate": 0.0012,
      "loss": 2.6560044288635254,
      "step": 8852
    },
    {
      "epoch": 0.7178912855647926,
      "grad_norm": 0.02586180716753006,
      "learning_rate": 0.0012,
      "loss": 2.629964828491211,
      "step": 8854
    },
    {
      "epoch": 0.7180534475899936,
      "grad_norm": 0.028489673510193825,
      "learning_rate": 0.0012,
      "loss": 2.736300468444824,
      "step": 8856
    },
    {
      "epoch": 0.7182156096151946,
      "grad_norm": 0.045422837138175964,
      "learning_rate": 0.0012,
      "loss": 2.677198648452759,
      "step": 8858
    },
    {
      "epoch": 0.7183777716403956,
      "grad_norm": 0.04931988939642906,
      "learning_rate": 0.0012,
      "loss": 2.613539934158325,
      "step": 8860
    },
    {
      "epoch": 0.7185399336655965,
      "grad_norm": 0.0662846490740776,
      "learning_rate": 0.0012,
      "loss": 2.6678614616394043,
      "step": 8862
    },
    {
      "epoch": 0.7187020956907976,
      "grad_norm": 0.04322737827897072,
      "learning_rate": 0.0012,
      "loss": 2.68438458442688,
      "step": 8864
    },
    {
      "epoch": 0.7188642577159986,
      "grad_norm": 0.027875015512108803,
      "learning_rate": 0.0012,
      "loss": 2.6578526496887207,
      "step": 8866
    },
    {
      "epoch": 0.7190264197411995,
      "grad_norm": 0.02806466817855835,
      "learning_rate": 0.0012,
      "loss": 2.704618453979492,
      "step": 8868
    },
    {
      "epoch": 0.7191885817664005,
      "grad_norm": 0.029963651672005653,
      "learning_rate": 0.0012,
      "loss": 2.6741878986358643,
      "step": 8870
    },
    {
      "epoch": 0.7193507437916016,
      "grad_norm": 0.025599943473935127,
      "learning_rate": 0.0012,
      "loss": 2.6542654037475586,
      "step": 8872
    },
    {
      "epoch": 0.7195129058168025,
      "grad_norm": 0.026875631883740425,
      "learning_rate": 0.0012,
      "loss": 2.6980032920837402,
      "step": 8874
    },
    {
      "epoch": 0.7196750678420035,
      "grad_norm": 0.02404828555881977,
      "learning_rate": 0.0012,
      "loss": 2.702625036239624,
      "step": 8876
    },
    {
      "epoch": 0.7198372298672046,
      "grad_norm": 0.038078878074884415,
      "learning_rate": 0.0012,
      "loss": 2.6358749866485596,
      "step": 8878
    },
    {
      "epoch": 0.7199993918924055,
      "grad_norm": 0.03902093693614006,
      "learning_rate": 0.0012,
      "loss": 2.655695915222168,
      "step": 8880
    },
    {
      "epoch": 0.7201615539176065,
      "grad_norm": 0.028261663392186165,
      "learning_rate": 0.0012,
      "loss": 2.694770574569702,
      "step": 8882
    },
    {
      "epoch": 0.7203237159428075,
      "grad_norm": 0.050694789737463,
      "learning_rate": 0.0012,
      "loss": 2.632167339324951,
      "step": 8884
    },
    {
      "epoch": 0.7204858779680084,
      "grad_norm": 0.02998553030192852,
      "learning_rate": 0.0012,
      "loss": 2.6345014572143555,
      "step": 8886
    },
    {
      "epoch": 0.7206480399932095,
      "grad_norm": 0.03457975387573242,
      "learning_rate": 0.0012,
      "loss": 2.683516263961792,
      "step": 8888
    },
    {
      "epoch": 0.7208102020184105,
      "grad_norm": 0.026584472507238388,
      "learning_rate": 0.0012,
      "loss": 2.7107322216033936,
      "step": 8890
    },
    {
      "epoch": 0.7209723640436114,
      "grad_norm": 0.025932691991329193,
      "learning_rate": 0.0012,
      "loss": 2.65767502784729,
      "step": 8892
    },
    {
      "epoch": 0.7211345260688125,
      "grad_norm": 0.02961260825395584,
      "learning_rate": 0.0012,
      "loss": 2.6767306327819824,
      "step": 8894
    },
    {
      "epoch": 0.7212966880940135,
      "grad_norm": 0.0381023995578289,
      "learning_rate": 0.0012,
      "loss": 2.639253854751587,
      "step": 8896
    },
    {
      "epoch": 0.7214588501192144,
      "grad_norm": 0.040155041962862015,
      "learning_rate": 0.0012,
      "loss": 2.6344733238220215,
      "step": 8898
    },
    {
      "epoch": 0.7216210121444154,
      "grad_norm": 0.03889737278223038,
      "learning_rate": 0.0012,
      "loss": 2.6251189708709717,
      "step": 8900
    },
    {
      "epoch": 0.7217831741696165,
      "grad_norm": 0.024733806028962135,
      "learning_rate": 0.0012,
      "loss": 2.724432945251465,
      "step": 8902
    },
    {
      "epoch": 0.7219453361948174,
      "grad_norm": 0.02482658624649048,
      "learning_rate": 0.0012,
      "loss": 2.681790590286255,
      "step": 8904
    },
    {
      "epoch": 0.7221074982200184,
      "grad_norm": 0.042229846119880676,
      "learning_rate": 0.0012,
      "loss": 2.7008824348449707,
      "step": 8906
    },
    {
      "epoch": 0.7222696602452194,
      "grad_norm": 0.04229925945401192,
      "learning_rate": 0.0012,
      "loss": 2.648798704147339,
      "step": 8908
    },
    {
      "epoch": 0.7224318222704204,
      "grad_norm": 0.02815607190132141,
      "learning_rate": 0.0012,
      "loss": 2.615175247192383,
      "step": 8910
    },
    {
      "epoch": 0.7225939842956214,
      "grad_norm": 0.026716170832514763,
      "learning_rate": 0.0012,
      "loss": 2.628390312194824,
      "step": 8912
    },
    {
      "epoch": 0.7227561463208224,
      "grad_norm": 0.04229775443673134,
      "learning_rate": 0.0012,
      "loss": 2.632451295852661,
      "step": 8914
    },
    {
      "epoch": 0.7229183083460233,
      "grad_norm": 0.04120687022805214,
      "learning_rate": 0.0012,
      "loss": 2.658308982849121,
      "step": 8916
    },
    {
      "epoch": 0.7230804703712244,
      "grad_norm": 0.0538850836455822,
      "learning_rate": 0.0012,
      "loss": 2.6318869590759277,
      "step": 8918
    },
    {
      "epoch": 0.7232426323964254,
      "grad_norm": 0.026861008256673813,
      "learning_rate": 0.0012,
      "loss": 2.6729531288146973,
      "step": 8920
    },
    {
      "epoch": 0.7234047944216263,
      "grad_norm": 0.04203268885612488,
      "learning_rate": 0.0012,
      "loss": 2.620727300643921,
      "step": 8922
    },
    {
      "epoch": 0.7235669564468273,
      "grad_norm": 0.03428163379430771,
      "learning_rate": 0.0012,
      "loss": 2.614684581756592,
      "step": 8924
    },
    {
      "epoch": 0.7237291184720284,
      "grad_norm": 0.025608958676457405,
      "learning_rate": 0.0012,
      "loss": 2.6480929851531982,
      "step": 8926
    },
    {
      "epoch": 0.7238912804972293,
      "grad_norm": 0.03094377927482128,
      "learning_rate": 0.0012,
      "loss": 2.7154157161712646,
      "step": 8928
    },
    {
      "epoch": 0.7240534425224303,
      "grad_norm": 0.03767908364534378,
      "learning_rate": 0.0012,
      "loss": 2.676835775375366,
      "step": 8930
    },
    {
      "epoch": 0.7242156045476313,
      "grad_norm": 0.04938361421227455,
      "learning_rate": 0.0012,
      "loss": 2.7172765731811523,
      "step": 8932
    },
    {
      "epoch": 0.7243777665728323,
      "grad_norm": 0.02369101718068123,
      "learning_rate": 0.0012,
      "loss": 2.6189637184143066,
      "step": 8934
    },
    {
      "epoch": 0.7245399285980333,
      "grad_norm": 0.027761822566390038,
      "learning_rate": 0.0012,
      "loss": 2.6716017723083496,
      "step": 8936
    },
    {
      "epoch": 0.7247020906232343,
      "grad_norm": 0.053287751972675323,
      "learning_rate": 0.0012,
      "loss": 2.598752498626709,
      "step": 8938
    },
    {
      "epoch": 0.7248642526484352,
      "grad_norm": 0.06872156262397766,
      "learning_rate": 0.0012,
      "loss": 2.6997687816619873,
      "step": 8940
    },
    {
      "epoch": 0.7250264146736363,
      "grad_norm": 0.03789858892560005,
      "learning_rate": 0.0012,
      "loss": 2.6133620738983154,
      "step": 8942
    },
    {
      "epoch": 0.7251885766988373,
      "grad_norm": 0.04320390895009041,
      "learning_rate": 0.0012,
      "loss": 2.6997995376586914,
      "step": 8944
    },
    {
      "epoch": 0.7253507387240382,
      "grad_norm": 0.08343791216611862,
      "learning_rate": 0.0012,
      "loss": 2.632516860961914,
      "step": 8946
    },
    {
      "epoch": 0.7255129007492392,
      "grad_norm": 0.025180600583553314,
      "learning_rate": 0.0012,
      "loss": 2.6762123107910156,
      "step": 8948
    },
    {
      "epoch": 0.7256750627744403,
      "grad_norm": 0.07115805149078369,
      "learning_rate": 0.0012,
      "loss": 2.612468719482422,
      "step": 8950
    },
    {
      "epoch": 0.7258372247996412,
      "grad_norm": 0.06466493010520935,
      "learning_rate": 0.0012,
      "loss": 2.6994071006774902,
      "step": 8952
    },
    {
      "epoch": 0.7259993868248422,
      "grad_norm": 0.03395286574959755,
      "learning_rate": 0.0012,
      "loss": 2.646397590637207,
      "step": 8954
    },
    {
      "epoch": 0.7261615488500432,
      "grad_norm": 0.0748041570186615,
      "learning_rate": 0.0012,
      "loss": 2.6755363941192627,
      "step": 8956
    },
    {
      "epoch": 0.7263237108752442,
      "grad_norm": 0.027461083605885506,
      "learning_rate": 0.0012,
      "loss": 2.75111722946167,
      "step": 8958
    },
    {
      "epoch": 0.7264858729004452,
      "grad_norm": 0.0414281003177166,
      "learning_rate": 0.0012,
      "loss": 2.6772708892822266,
      "step": 8960
    },
    {
      "epoch": 0.7266480349256462,
      "grad_norm": 0.050725121051073074,
      "learning_rate": 0.0012,
      "loss": 2.684685230255127,
      "step": 8962
    },
    {
      "epoch": 0.7268101969508471,
      "grad_norm": 0.030702397227287292,
      "learning_rate": 0.0012,
      "loss": 2.592407703399658,
      "step": 8964
    },
    {
      "epoch": 0.7269723589760482,
      "grad_norm": 0.029474422335624695,
      "learning_rate": 0.0012,
      "loss": 2.6362009048461914,
      "step": 8966
    },
    {
      "epoch": 0.7271345210012492,
      "grad_norm": 0.024028392508625984,
      "learning_rate": 0.0012,
      "loss": 2.7103705406188965,
      "step": 8968
    },
    {
      "epoch": 0.7272966830264501,
      "grad_norm": 0.023728234693408012,
      "learning_rate": 0.0012,
      "loss": 2.6669163703918457,
      "step": 8970
    },
    {
      "epoch": 0.7274588450516511,
      "grad_norm": 0.036235880106687546,
      "learning_rate": 0.0012,
      "loss": 2.705291986465454,
      "step": 8972
    },
    {
      "epoch": 0.7276210070768522,
      "grad_norm": 0.0229126438498497,
      "learning_rate": 0.0012,
      "loss": 2.633545398712158,
      "step": 8974
    },
    {
      "epoch": 0.7277831691020531,
      "grad_norm": 0.03710463270545006,
      "learning_rate": 0.0012,
      "loss": 2.589390993118286,
      "step": 8976
    },
    {
      "epoch": 0.7279453311272541,
      "grad_norm": 0.03382264822721481,
      "learning_rate": 0.0012,
      "loss": 2.6929214000701904,
      "step": 8978
    },
    {
      "epoch": 0.7281074931524552,
      "grad_norm": 0.027814475819468498,
      "learning_rate": 0.0012,
      "loss": 2.6665761470794678,
      "step": 8980
    },
    {
      "epoch": 0.7282696551776561,
      "grad_norm": 0.031194591894745827,
      "learning_rate": 0.0012,
      "loss": 2.6313323974609375,
      "step": 8982
    },
    {
      "epoch": 0.7284318172028571,
      "grad_norm": 0.026580914855003357,
      "learning_rate": 0.0012,
      "loss": 2.6967549324035645,
      "step": 8984
    },
    {
      "epoch": 0.7285939792280581,
      "grad_norm": 0.03447329252958298,
      "learning_rate": 0.0012,
      "loss": 2.644557476043701,
      "step": 8986
    },
    {
      "epoch": 0.728756141253259,
      "grad_norm": 0.07663948833942413,
      "learning_rate": 0.0012,
      "loss": 2.7048749923706055,
      "step": 8988
    },
    {
      "epoch": 0.7289183032784601,
      "grad_norm": 0.07130274176597595,
      "learning_rate": 0.0012,
      "loss": 2.6077349185943604,
      "step": 8990
    },
    {
      "epoch": 0.7290804653036611,
      "grad_norm": 0.07147082686424255,
      "learning_rate": 0.0012,
      "loss": 2.6280765533447266,
      "step": 8992
    },
    {
      "epoch": 0.729242627328862,
      "grad_norm": 0.05305178090929985,
      "learning_rate": 0.0012,
      "loss": 2.6899056434631348,
      "step": 8994
    },
    {
      "epoch": 0.729404789354063,
      "grad_norm": 0.0449523963034153,
      "learning_rate": 0.0012,
      "loss": 2.7008349895477295,
      "step": 8996
    },
    {
      "epoch": 0.7295669513792641,
      "grad_norm": 0.05644380673766136,
      "learning_rate": 0.0012,
      "loss": 2.6564903259277344,
      "step": 8998
    },
    {
      "epoch": 0.729729113404465,
      "grad_norm": 0.03587336465716362,
      "learning_rate": 0.0012,
      "loss": 2.6967504024505615,
      "step": 9000
    },
    {
      "epoch": 0.729891275429666,
      "grad_norm": 0.03193652257323265,
      "learning_rate": 0.0012,
      "loss": 2.639758825302124,
      "step": 9002
    },
    {
      "epoch": 0.7300534374548671,
      "grad_norm": 0.03692487254738808,
      "learning_rate": 0.0012,
      "loss": 2.7287564277648926,
      "step": 9004
    },
    {
      "epoch": 0.730215599480068,
      "grad_norm": 0.02921161614358425,
      "learning_rate": 0.0012,
      "loss": 2.711639404296875,
      "step": 9006
    },
    {
      "epoch": 0.730377761505269,
      "grad_norm": 0.027210913598537445,
      "learning_rate": 0.0012,
      "loss": 2.6469578742980957,
      "step": 9008
    },
    {
      "epoch": 0.73053992353047,
      "grad_norm": 0.04128560051321983,
      "learning_rate": 0.0012,
      "loss": 2.5901427268981934,
      "step": 9010
    },
    {
      "epoch": 0.730702085555671,
      "grad_norm": 0.049668487161397934,
      "learning_rate": 0.0012,
      "loss": 2.6852893829345703,
      "step": 9012
    },
    {
      "epoch": 0.730864247580872,
      "grad_norm": 0.026038864627480507,
      "learning_rate": 0.0012,
      "loss": 2.6707139015197754,
      "step": 9014
    },
    {
      "epoch": 0.731026409606073,
      "grad_norm": 0.030876200646162033,
      "learning_rate": 0.0012,
      "loss": 2.6414237022399902,
      "step": 9016
    },
    {
      "epoch": 0.7311885716312739,
      "grad_norm": 0.025559060275554657,
      "learning_rate": 0.0012,
      "loss": 2.653014659881592,
      "step": 9018
    },
    {
      "epoch": 0.731350733656475,
      "grad_norm": 0.02662133052945137,
      "learning_rate": 0.0012,
      "loss": 2.66080641746521,
      "step": 9020
    },
    {
      "epoch": 0.731512895681676,
      "grad_norm": 0.030024239793419838,
      "learning_rate": 0.0012,
      "loss": 2.6168251037597656,
      "step": 9022
    },
    {
      "epoch": 0.7316750577068769,
      "grad_norm": 0.031597357243299484,
      "learning_rate": 0.0012,
      "loss": 2.7079153060913086,
      "step": 9024
    },
    {
      "epoch": 0.7318372197320779,
      "grad_norm": 0.026479097083210945,
      "learning_rate": 0.0012,
      "loss": 2.6124842166900635,
      "step": 9026
    },
    {
      "epoch": 0.731999381757279,
      "grad_norm": 0.029401147738099098,
      "learning_rate": 0.0012,
      "loss": 2.670543909072876,
      "step": 9028
    },
    {
      "epoch": 0.7321615437824799,
      "grad_norm": 0.024492863565683365,
      "learning_rate": 0.0012,
      "loss": 2.7047343254089355,
      "step": 9030
    },
    {
      "epoch": 0.7323237058076809,
      "grad_norm": 0.029109932482242584,
      "learning_rate": 0.0012,
      "loss": 2.6953775882720947,
      "step": 9032
    },
    {
      "epoch": 0.7324858678328819,
      "grad_norm": 0.027554761618375778,
      "learning_rate": 0.0012,
      "loss": 2.693880319595337,
      "step": 9034
    },
    {
      "epoch": 0.7326480298580829,
      "grad_norm": 0.03847280144691467,
      "learning_rate": 0.0012,
      "loss": 2.7175731658935547,
      "step": 9036
    },
    {
      "epoch": 0.7328101918832839,
      "grad_norm": 0.027745956555008888,
      "learning_rate": 0.0012,
      "loss": 2.60479736328125,
      "step": 9038
    },
    {
      "epoch": 0.7329723539084849,
      "grad_norm": 0.028606368228793144,
      "learning_rate": 0.0012,
      "loss": 2.633723258972168,
      "step": 9040
    },
    {
      "epoch": 0.7331345159336858,
      "grad_norm": 0.03277192637324333,
      "learning_rate": 0.0012,
      "loss": 2.66507887840271,
      "step": 9042
    },
    {
      "epoch": 0.7332966779588869,
      "grad_norm": 0.03086058236658573,
      "learning_rate": 0.0012,
      "loss": 2.6763863563537598,
      "step": 9044
    },
    {
      "epoch": 0.7334588399840879,
      "grad_norm": 0.024160290136933327,
      "learning_rate": 0.0012,
      "loss": 2.7386035919189453,
      "step": 9046
    },
    {
      "epoch": 0.7336210020092888,
      "grad_norm": 0.026214024052023888,
      "learning_rate": 0.0012,
      "loss": 2.6258201599121094,
      "step": 9048
    },
    {
      "epoch": 0.7337831640344898,
      "grad_norm": 0.035366155207157135,
      "learning_rate": 0.0012,
      "loss": 2.6604833602905273,
      "step": 9050
    },
    {
      "epoch": 0.7339453260596909,
      "grad_norm": 0.030510984361171722,
      "learning_rate": 0.0012,
      "loss": 2.6583099365234375,
      "step": 9052
    },
    {
      "epoch": 0.7341074880848918,
      "grad_norm": 0.07023417204618454,
      "learning_rate": 0.0012,
      "loss": 2.691119432449341,
      "step": 9054
    },
    {
      "epoch": 0.7342696501100928,
      "grad_norm": 0.06349441409111023,
      "learning_rate": 0.0012,
      "loss": 2.639767646789551,
      "step": 9056
    },
    {
      "epoch": 0.7344318121352938,
      "grad_norm": 0.03422055020928383,
      "learning_rate": 0.0012,
      "loss": 2.5637435913085938,
      "step": 9058
    },
    {
      "epoch": 0.7345939741604948,
      "grad_norm": 0.03386635705828667,
      "learning_rate": 0.0012,
      "loss": 2.703606605529785,
      "step": 9060
    },
    {
      "epoch": 0.7347561361856958,
      "grad_norm": 0.06253267824649811,
      "learning_rate": 0.0012,
      "loss": 2.683342933654785,
      "step": 9062
    },
    {
      "epoch": 0.7349182982108968,
      "grad_norm": 0.046255920082330704,
      "learning_rate": 0.0012,
      "loss": 2.64892578125,
      "step": 9064
    },
    {
      "epoch": 0.7350804602360977,
      "grad_norm": 0.029440348967909813,
      "learning_rate": 0.0012,
      "loss": 2.6218326091766357,
      "step": 9066
    },
    {
      "epoch": 0.7352426222612988,
      "grad_norm": 0.0547582171857357,
      "learning_rate": 0.0012,
      "loss": 2.6387438774108887,
      "step": 9068
    },
    {
      "epoch": 0.7354047842864998,
      "grad_norm": 0.03453134745359421,
      "learning_rate": 0.0012,
      "loss": 2.7511844635009766,
      "step": 9070
    },
    {
      "epoch": 0.7355669463117007,
      "grad_norm": 0.02562636323273182,
      "learning_rate": 0.0012,
      "loss": 2.6360726356506348,
      "step": 9072
    },
    {
      "epoch": 0.7357291083369017,
      "grad_norm": 0.024288635700941086,
      "learning_rate": 0.0012,
      "loss": 2.6764798164367676,
      "step": 9074
    },
    {
      "epoch": 0.7358912703621028,
      "grad_norm": 0.03458630293607712,
      "learning_rate": 0.0012,
      "loss": 2.6575992107391357,
      "step": 9076
    },
    {
      "epoch": 0.7360534323873037,
      "grad_norm": 0.03325461223721504,
      "learning_rate": 0.0012,
      "loss": 2.6552438735961914,
      "step": 9078
    },
    {
      "epoch": 0.7362155944125047,
      "grad_norm": 0.033168621361255646,
      "learning_rate": 0.0012,
      "loss": 2.6770873069763184,
      "step": 9080
    },
    {
      "epoch": 0.7363777564377058,
      "grad_norm": 0.03635849431157112,
      "learning_rate": 0.0012,
      "loss": 2.679434299468994,
      "step": 9082
    },
    {
      "epoch": 0.7365399184629067,
      "grad_norm": 0.07457394152879715,
      "learning_rate": 0.0012,
      "loss": 2.6763792037963867,
      "step": 9084
    },
    {
      "epoch": 0.7367020804881077,
      "grad_norm": 0.04085304215550423,
      "learning_rate": 0.0012,
      "loss": 2.6387526988983154,
      "step": 9086
    },
    {
      "epoch": 0.7368642425133087,
      "grad_norm": 0.024113735184073448,
      "learning_rate": 0.0012,
      "loss": 2.689146041870117,
      "step": 9088
    },
    {
      "epoch": 0.7370264045385096,
      "grad_norm": 0.023824287578463554,
      "learning_rate": 0.0012,
      "loss": 2.7107841968536377,
      "step": 9090
    },
    {
      "epoch": 0.7371885665637107,
      "grad_norm": 0.043022580444812775,
      "learning_rate": 0.0012,
      "loss": 2.649684429168701,
      "step": 9092
    },
    {
      "epoch": 0.7373507285889117,
      "grad_norm": 0.033936310559511185,
      "learning_rate": 0.0012,
      "loss": 2.6650285720825195,
      "step": 9094
    },
    {
      "epoch": 0.7375128906141126,
      "grad_norm": 0.02479413151741028,
      "learning_rate": 0.0012,
      "loss": 2.756251335144043,
      "step": 9096
    },
    {
      "epoch": 0.7376750526393137,
      "grad_norm": 0.04217628017067909,
      "learning_rate": 0.0012,
      "loss": 2.696465492248535,
      "step": 9098
    },
    {
      "epoch": 0.7378372146645147,
      "grad_norm": 0.043723415583372116,
      "learning_rate": 0.0012,
      "loss": 2.7241640090942383,
      "step": 9100
    },
    {
      "epoch": 0.7379993766897156,
      "grad_norm": 0.027138298377394676,
      "learning_rate": 0.0012,
      "loss": 2.6947221755981445,
      "step": 9102
    },
    {
      "epoch": 0.7381615387149166,
      "grad_norm": 0.02429349161684513,
      "learning_rate": 0.0012,
      "loss": 2.7103395462036133,
      "step": 9104
    },
    {
      "epoch": 0.7383237007401177,
      "grad_norm": 0.05222051218152046,
      "learning_rate": 0.0012,
      "loss": 2.6415698528289795,
      "step": 9106
    },
    {
      "epoch": 0.7384858627653186,
      "grad_norm": 0.05113638564944267,
      "learning_rate": 0.0012,
      "loss": 2.662370204925537,
      "step": 9108
    },
    {
      "epoch": 0.7386480247905196,
      "grad_norm": 0.059879567474126816,
      "learning_rate": 0.0012,
      "loss": 2.688760757446289,
      "step": 9110
    },
    {
      "epoch": 0.7388101868157206,
      "grad_norm": 0.02668190933763981,
      "learning_rate": 0.0012,
      "loss": 2.594794273376465,
      "step": 9112
    },
    {
      "epoch": 0.7389723488409216,
      "grad_norm": 0.024826930835843086,
      "learning_rate": 0.0012,
      "loss": 2.6996405124664307,
      "step": 9114
    },
    {
      "epoch": 0.7391345108661226,
      "grad_norm": 0.02489348128437996,
      "learning_rate": 0.0012,
      "loss": 2.6483817100524902,
      "step": 9116
    },
    {
      "epoch": 0.7392966728913236,
      "grad_norm": 0.024619271978735924,
      "learning_rate": 0.0012,
      "loss": 2.680604934692383,
      "step": 9118
    },
    {
      "epoch": 0.7394588349165245,
      "grad_norm": 0.030346887186169624,
      "learning_rate": 0.0012,
      "loss": 2.682053565979004,
      "step": 9120
    },
    {
      "epoch": 0.7396209969417256,
      "grad_norm": 0.03549396991729736,
      "learning_rate": 0.0012,
      "loss": 2.6738972663879395,
      "step": 9122
    },
    {
      "epoch": 0.7397831589669266,
      "grad_norm": 0.02767488732933998,
      "learning_rate": 0.0012,
      "loss": 2.701751470565796,
      "step": 9124
    },
    {
      "epoch": 0.7399453209921275,
      "grad_norm": 0.03827529028058052,
      "learning_rate": 0.0012,
      "loss": 2.6973795890808105,
      "step": 9126
    },
    {
      "epoch": 0.7401074830173285,
      "grad_norm": 0.03769810497760773,
      "learning_rate": 0.0012,
      "loss": 2.664235830307007,
      "step": 9128
    },
    {
      "epoch": 0.7402696450425296,
      "grad_norm": 0.02463446743786335,
      "learning_rate": 0.0012,
      "loss": 2.6426098346710205,
      "step": 9130
    },
    {
      "epoch": 0.7404318070677305,
      "grad_norm": 0.031032199040055275,
      "learning_rate": 0.0012,
      "loss": 2.66390323638916,
      "step": 9132
    },
    {
      "epoch": 0.7405939690929315,
      "grad_norm": 0.034385278820991516,
      "learning_rate": 0.0012,
      "loss": 2.7262332439422607,
      "step": 9134
    },
    {
      "epoch": 0.7407561311181325,
      "grad_norm": 0.02334737218916416,
      "learning_rate": 0.0012,
      "loss": 2.5832386016845703,
      "step": 9136
    },
    {
      "epoch": 0.7409182931433335,
      "grad_norm": 0.03834664449095726,
      "learning_rate": 0.0012,
      "loss": 2.695746898651123,
      "step": 9138
    },
    {
      "epoch": 0.7410804551685345,
      "grad_norm": 0.036248281598091125,
      "learning_rate": 0.0012,
      "loss": 2.6504037380218506,
      "step": 9140
    },
    {
      "epoch": 0.7412426171937355,
      "grad_norm": 0.030370451509952545,
      "learning_rate": 0.0012,
      "loss": 2.6651456356048584,
      "step": 9142
    },
    {
      "epoch": 0.7414047792189364,
      "grad_norm": 0.03242216631770134,
      "learning_rate": 0.0012,
      "loss": 2.6223020553588867,
      "step": 9144
    },
    {
      "epoch": 0.7415669412441375,
      "grad_norm": 0.024145975708961487,
      "learning_rate": 0.0012,
      "loss": 2.6655020713806152,
      "step": 9146
    },
    {
      "epoch": 0.7417291032693385,
      "grad_norm": 0.0299733467400074,
      "learning_rate": 0.0012,
      "loss": 2.7057952880859375,
      "step": 9148
    },
    {
      "epoch": 0.7418912652945394,
      "grad_norm": 0.03862719237804413,
      "learning_rate": 0.0012,
      "loss": 2.7058815956115723,
      "step": 9150
    },
    {
      "epoch": 0.7420534273197404,
      "grad_norm": 0.03392815589904785,
      "learning_rate": 0.0012,
      "loss": 2.63435697555542,
      "step": 9152
    },
    {
      "epoch": 0.7422155893449415,
      "grad_norm": 0.06214327737689018,
      "learning_rate": 0.0012,
      "loss": 2.661466121673584,
      "step": 9154
    },
    {
      "epoch": 0.7423777513701424,
      "grad_norm": 0.06254151463508606,
      "learning_rate": 0.0012,
      "loss": 2.5793042182922363,
      "step": 9156
    },
    {
      "epoch": 0.7425399133953434,
      "grad_norm": 0.03615499287843704,
      "learning_rate": 0.0012,
      "loss": 2.690216541290283,
      "step": 9158
    },
    {
      "epoch": 0.7427020754205444,
      "grad_norm": 0.04002392292022705,
      "learning_rate": 0.0012,
      "loss": 2.654142379760742,
      "step": 9160
    },
    {
      "epoch": 0.7428642374457454,
      "grad_norm": 0.04396485164761543,
      "learning_rate": 0.0012,
      "loss": 2.690584182739258,
      "step": 9162
    },
    {
      "epoch": 0.7430263994709464,
      "grad_norm": 0.041542552411556244,
      "learning_rate": 0.0012,
      "loss": 2.63482403755188,
      "step": 9164
    },
    {
      "epoch": 0.7431885614961474,
      "grad_norm": 0.023814978078007698,
      "learning_rate": 0.0012,
      "loss": 2.592665672302246,
      "step": 9166
    },
    {
      "epoch": 0.7433507235213483,
      "grad_norm": 0.057962313294410706,
      "learning_rate": 0.0012,
      "loss": 2.5901474952697754,
      "step": 9168
    },
    {
      "epoch": 0.7435128855465494,
      "grad_norm": 0.030012313276529312,
      "learning_rate": 0.0012,
      "loss": 2.643664836883545,
      "step": 9170
    },
    {
      "epoch": 0.7436750475717504,
      "grad_norm": 0.02691221982240677,
      "learning_rate": 0.0012,
      "loss": 2.6383585929870605,
      "step": 9172
    },
    {
      "epoch": 0.7438372095969513,
      "grad_norm": 0.036436617374420166,
      "learning_rate": 0.0012,
      "loss": 2.619554281234741,
      "step": 9174
    },
    {
      "epoch": 0.7439993716221524,
      "grad_norm": 0.026578163728117943,
      "learning_rate": 0.0012,
      "loss": 2.623727798461914,
      "step": 9176
    },
    {
      "epoch": 0.7441615336473534,
      "grad_norm": 0.02620839886367321,
      "learning_rate": 0.0012,
      "loss": 2.6597843170166016,
      "step": 9178
    },
    {
      "epoch": 0.7443236956725543,
      "grad_norm": 0.03934147208929062,
      "learning_rate": 0.0012,
      "loss": 2.6515114307403564,
      "step": 9180
    },
    {
      "epoch": 0.7444858576977553,
      "grad_norm": 0.05039828643202782,
      "learning_rate": 0.0012,
      "loss": 2.688342332839966,
      "step": 9182
    },
    {
      "epoch": 0.7446480197229564,
      "grad_norm": 0.03851708769798279,
      "learning_rate": 0.0012,
      "loss": 2.665656805038452,
      "step": 9184
    },
    {
      "epoch": 0.7448101817481573,
      "grad_norm": 0.03792840242385864,
      "learning_rate": 0.0012,
      "loss": 2.6556200981140137,
      "step": 9186
    },
    {
      "epoch": 0.7449723437733583,
      "grad_norm": 0.027161303907632828,
      "learning_rate": 0.0012,
      "loss": 2.670783042907715,
      "step": 9188
    },
    {
      "epoch": 0.7451345057985593,
      "grad_norm": 0.04121796786785126,
      "learning_rate": 0.0012,
      "loss": 2.6323065757751465,
      "step": 9190
    },
    {
      "epoch": 0.7452966678237603,
      "grad_norm": 0.03151979297399521,
      "learning_rate": 0.0012,
      "loss": 2.63338041305542,
      "step": 9192
    },
    {
      "epoch": 0.7454588298489613,
      "grad_norm": 0.0269197728484869,
      "learning_rate": 0.0012,
      "loss": 2.660057783126831,
      "step": 9194
    },
    {
      "epoch": 0.7456209918741623,
      "grad_norm": 0.04223715141415596,
      "learning_rate": 0.0012,
      "loss": 2.6963002681732178,
      "step": 9196
    },
    {
      "epoch": 0.7457831538993632,
      "grad_norm": 0.03831716626882553,
      "learning_rate": 0.0012,
      "loss": 2.6533260345458984,
      "step": 9198
    },
    {
      "epoch": 0.7459453159245643,
      "grad_norm": 0.0531466044485569,
      "learning_rate": 0.0012,
      "loss": 2.700404405593872,
      "step": 9200
    },
    {
      "epoch": 0.7461074779497653,
      "grad_norm": 0.047526534646749496,
      "learning_rate": 0.0012,
      "loss": 2.5830416679382324,
      "step": 9202
    },
    {
      "epoch": 0.7462696399749662,
      "grad_norm": 0.02440623566508293,
      "learning_rate": 0.0012,
      "loss": 2.6386795043945312,
      "step": 9204
    },
    {
      "epoch": 0.7464318020001672,
      "grad_norm": 0.02723446488380432,
      "learning_rate": 0.0012,
      "loss": 2.6537318229675293,
      "step": 9206
    },
    {
      "epoch": 0.7465939640253683,
      "grad_norm": 0.029141709208488464,
      "learning_rate": 0.0012,
      "loss": 2.709733724594116,
      "step": 9208
    },
    {
      "epoch": 0.7467561260505692,
      "grad_norm": 0.024242183193564415,
      "learning_rate": 0.0012,
      "loss": 2.6638236045837402,
      "step": 9210
    },
    {
      "epoch": 0.7469182880757702,
      "grad_norm": 0.02448657713830471,
      "learning_rate": 0.0012,
      "loss": 2.6340324878692627,
      "step": 9212
    },
    {
      "epoch": 0.7470804501009712,
      "grad_norm": 0.04470985010266304,
      "learning_rate": 0.0012,
      "loss": 2.657618761062622,
      "step": 9214
    },
    {
      "epoch": 0.7472426121261722,
      "grad_norm": 0.04249253123998642,
      "learning_rate": 0.0012,
      "loss": 2.627014636993408,
      "step": 9216
    },
    {
      "epoch": 0.7474047741513732,
      "grad_norm": 0.04065296798944473,
      "learning_rate": 0.0012,
      "loss": 2.612736701965332,
      "step": 9218
    },
    {
      "epoch": 0.7475669361765742,
      "grad_norm": 0.030262775719165802,
      "learning_rate": 0.0012,
      "loss": 2.681468963623047,
      "step": 9220
    },
    {
      "epoch": 0.7477290982017751,
      "grad_norm": 0.028434323146939278,
      "learning_rate": 0.0012,
      "loss": 2.6185665130615234,
      "step": 9222
    },
    {
      "epoch": 0.7478912602269762,
      "grad_norm": 0.028870169073343277,
      "learning_rate": 0.0012,
      "loss": 2.598982334136963,
      "step": 9224
    },
    {
      "epoch": 0.7480534222521772,
      "grad_norm": 0.025222228839993477,
      "learning_rate": 0.0012,
      "loss": 2.668687582015991,
      "step": 9226
    },
    {
      "epoch": 0.7482155842773781,
      "grad_norm": 0.03260356932878494,
      "learning_rate": 0.0012,
      "loss": 2.5500288009643555,
      "step": 9228
    },
    {
      "epoch": 0.7483777463025791,
      "grad_norm": 0.03141386806964874,
      "learning_rate": 0.0012,
      "loss": 2.617983341217041,
      "step": 9230
    },
    {
      "epoch": 0.7485399083277802,
      "grad_norm": 0.028105992823839188,
      "learning_rate": 0.0012,
      "loss": 2.6764450073242188,
      "step": 9232
    },
    {
      "epoch": 0.7487020703529811,
      "grad_norm": 0.029146363958716393,
      "learning_rate": 0.0012,
      "loss": 2.6457157135009766,
      "step": 9234
    },
    {
      "epoch": 0.7488642323781821,
      "grad_norm": 0.029161138460040092,
      "learning_rate": 0.0012,
      "loss": 2.659365653991699,
      "step": 9236
    },
    {
      "epoch": 0.7490263944033831,
      "grad_norm": 0.054964859038591385,
      "learning_rate": 0.0012,
      "loss": 2.709752082824707,
      "step": 9238
    },
    {
      "epoch": 0.7491885564285841,
      "grad_norm": 0.0808100476861,
      "learning_rate": 0.0012,
      "loss": 2.6477842330932617,
      "step": 9240
    },
    {
      "epoch": 0.7493507184537851,
      "grad_norm": 0.05112648010253906,
      "learning_rate": 0.0012,
      "loss": 2.7183988094329834,
      "step": 9242
    },
    {
      "epoch": 0.7495128804789861,
      "grad_norm": 0.045990172773599625,
      "learning_rate": 0.0012,
      "loss": 2.590847969055176,
      "step": 9244
    },
    {
      "epoch": 0.749675042504187,
      "grad_norm": 0.06986477226018906,
      "learning_rate": 0.0012,
      "loss": 2.6961381435394287,
      "step": 9246
    },
    {
      "epoch": 0.7498372045293881,
      "grad_norm": 0.05615553632378578,
      "learning_rate": 0.0012,
      "loss": 2.669726848602295,
      "step": 9248
    },
    {
      "epoch": 0.7499993665545891,
      "grad_norm": 0.028860511258244514,
      "learning_rate": 0.0012,
      "loss": 2.586146354675293,
      "step": 9250
    },
    {
      "epoch": 0.75016152857979,
      "grad_norm": 0.0521184466779232,
      "learning_rate": 0.0012,
      "loss": 2.6900200843811035,
      "step": 9252
    },
    {
      "epoch": 0.75016152857979,
      "eval_loss": 2.6659395694732666,
      "eval_runtime": 1148.9147,
      "eval_samples_per_second": 27.759,
      "eval_steps_per_second": 6.94,
      "step": 9252
    },
    {
      "epoch": 0.750323690604991,
      "grad_norm": 0.022893136367201805,
      "learning_rate": 0.0012,
      "loss": 2.643247604370117,
      "step": 9254
    },
    {
      "epoch": 0.7504858526301921,
      "grad_norm": 0.0531049482524395,
      "learning_rate": 0.0012,
      "loss": 2.6628823280334473,
      "step": 9256
    },
    {
      "epoch": 0.750648014655393,
      "grad_norm": 0.05756595730781555,
      "learning_rate": 0.0012,
      "loss": 2.6808700561523438,
      "step": 9258
    },
    {
      "epoch": 0.750810176680594,
      "grad_norm": 0.02854207158088684,
      "learning_rate": 0.0012,
      "loss": 2.5968058109283447,
      "step": 9260
    },
    {
      "epoch": 0.750972338705795,
      "grad_norm": 0.034345805644989014,
      "learning_rate": 0.0012,
      "loss": 2.6287951469421387,
      "step": 9262
    },
    {
      "epoch": 0.751134500730996,
      "grad_norm": 0.05970115587115288,
      "learning_rate": 0.0012,
      "loss": 2.657470226287842,
      "step": 9264
    },
    {
      "epoch": 0.751296662756197,
      "grad_norm": 0.02480914257466793,
      "learning_rate": 0.0012,
      "loss": 2.7384140491485596,
      "step": 9266
    },
    {
      "epoch": 0.751458824781398,
      "grad_norm": 0.04782954603433609,
      "learning_rate": 0.0012,
      "loss": 2.631850242614746,
      "step": 9268
    },
    {
      "epoch": 0.751620986806599,
      "grad_norm": 0.03518272936344147,
      "learning_rate": 0.0012,
      "loss": 2.640596628189087,
      "step": 9270
    },
    {
      "epoch": 0.7517831488318,
      "grad_norm": 0.027830297127366066,
      "learning_rate": 0.0012,
      "loss": 2.6643223762512207,
      "step": 9272
    },
    {
      "epoch": 0.751945310857001,
      "grad_norm": 0.024838315322995186,
      "learning_rate": 0.0012,
      "loss": 2.638322353363037,
      "step": 9274
    },
    {
      "epoch": 0.7521074728822019,
      "grad_norm": 0.052717406302690506,
      "learning_rate": 0.0012,
      "loss": 2.6376612186431885,
      "step": 9276
    },
    {
      "epoch": 0.752269634907403,
      "grad_norm": 0.03929677605628967,
      "learning_rate": 0.0012,
      "loss": 2.7116894721984863,
      "step": 9278
    },
    {
      "epoch": 0.752431796932604,
      "grad_norm": 0.05738448351621628,
      "learning_rate": 0.0012,
      "loss": 2.606793165206909,
      "step": 9280
    },
    {
      "epoch": 0.7525939589578049,
      "grad_norm": 0.029824841767549515,
      "learning_rate": 0.0012,
      "loss": 2.6628293991088867,
      "step": 9282
    },
    {
      "epoch": 0.7527561209830059,
      "grad_norm": 0.02486390247941017,
      "learning_rate": 0.0012,
      "loss": 2.6410748958587646,
      "step": 9284
    },
    {
      "epoch": 0.752918283008207,
      "grad_norm": 0.028832659125328064,
      "learning_rate": 0.0012,
      "loss": 2.6465420722961426,
      "step": 9286
    },
    {
      "epoch": 0.7530804450334079,
      "grad_norm": 0.03631141409277916,
      "learning_rate": 0.0012,
      "loss": 2.681222915649414,
      "step": 9288
    },
    {
      "epoch": 0.7532426070586089,
      "grad_norm": 0.036490436643362045,
      "learning_rate": 0.0012,
      "loss": 2.647294044494629,
      "step": 9290
    },
    {
      "epoch": 0.7534047690838099,
      "grad_norm": 0.028911545872688293,
      "learning_rate": 0.0012,
      "loss": 2.6401190757751465,
      "step": 9292
    },
    {
      "epoch": 0.7535669311090109,
      "grad_norm": 0.045060303062200546,
      "learning_rate": 0.0012,
      "loss": 2.687312126159668,
      "step": 9294
    },
    {
      "epoch": 0.7537290931342119,
      "grad_norm": 0.024772722274065018,
      "learning_rate": 0.0012,
      "loss": 2.613676071166992,
      "step": 9296
    },
    {
      "epoch": 0.7538912551594129,
      "grad_norm": 0.02516285516321659,
      "learning_rate": 0.0012,
      "loss": 2.713721752166748,
      "step": 9298
    },
    {
      "epoch": 0.7540534171846138,
      "grad_norm": 0.03218274563550949,
      "learning_rate": 0.0012,
      "loss": 2.6038289070129395,
      "step": 9300
    },
    {
      "epoch": 0.7542155792098149,
      "grad_norm": 0.03582766652107239,
      "learning_rate": 0.0012,
      "loss": 2.687371253967285,
      "step": 9302
    },
    {
      "epoch": 0.7543777412350159,
      "grad_norm": 0.029010022059082985,
      "learning_rate": 0.0012,
      "loss": 2.7392971515655518,
      "step": 9304
    },
    {
      "epoch": 0.7545399032602168,
      "grad_norm": 0.023968849331140518,
      "learning_rate": 0.0012,
      "loss": 2.6314003467559814,
      "step": 9306
    },
    {
      "epoch": 0.7547020652854178,
      "grad_norm": 0.03439465910196304,
      "learning_rate": 0.0012,
      "loss": 2.6375765800476074,
      "step": 9308
    },
    {
      "epoch": 0.7548642273106189,
      "grad_norm": 0.0466441884636879,
      "learning_rate": 0.0012,
      "loss": 2.6433002948760986,
      "step": 9310
    },
    {
      "epoch": 0.7550263893358198,
      "grad_norm": 0.04906356707215309,
      "learning_rate": 0.0012,
      "loss": 2.64373779296875,
      "step": 9312
    },
    {
      "epoch": 0.7551885513610208,
      "grad_norm": 0.033746927976608276,
      "learning_rate": 0.0012,
      "loss": 2.633333444595337,
      "step": 9314
    },
    {
      "epoch": 0.7553507133862218,
      "grad_norm": 0.02481168508529663,
      "learning_rate": 0.0012,
      "loss": 2.6471283435821533,
      "step": 9316
    },
    {
      "epoch": 0.7555128754114228,
      "grad_norm": 0.026111755520105362,
      "learning_rate": 0.0012,
      "loss": 2.5930566787719727,
      "step": 9318
    },
    {
      "epoch": 0.7556750374366238,
      "grad_norm": 0.026131317019462585,
      "learning_rate": 0.0012,
      "loss": 2.6801795959472656,
      "step": 9320
    },
    {
      "epoch": 0.7558371994618248,
      "grad_norm": 0.04290035367012024,
      "learning_rate": 0.0012,
      "loss": 2.668900728225708,
      "step": 9322
    },
    {
      "epoch": 0.7559993614870257,
      "grad_norm": 0.04951997846364975,
      "learning_rate": 0.0012,
      "loss": 2.635176181793213,
      "step": 9324
    },
    {
      "epoch": 0.7561615235122268,
      "grad_norm": 0.06683845072984695,
      "learning_rate": 0.0012,
      "loss": 2.6129703521728516,
      "step": 9326
    },
    {
      "epoch": 0.7563236855374278,
      "grad_norm": 0.07684135437011719,
      "learning_rate": 0.0012,
      "loss": 2.667849063873291,
      "step": 9328
    },
    {
      "epoch": 0.7564858475626287,
      "grad_norm": 0.027303235605359077,
      "learning_rate": 0.0012,
      "loss": 2.62037992477417,
      "step": 9330
    },
    {
      "epoch": 0.7566480095878297,
      "grad_norm": 0.05194947496056557,
      "learning_rate": 0.0012,
      "loss": 2.6719579696655273,
      "step": 9332
    },
    {
      "epoch": 0.7568101716130308,
      "grad_norm": 0.04990995302796364,
      "learning_rate": 0.0012,
      "loss": 2.7261805534362793,
      "step": 9334
    },
    {
      "epoch": 0.7569723336382317,
      "grad_norm": 0.027103105559945107,
      "learning_rate": 0.0012,
      "loss": 2.587911605834961,
      "step": 9336
    },
    {
      "epoch": 0.7571344956634327,
      "grad_norm": 0.04107770696282387,
      "learning_rate": 0.0012,
      "loss": 2.6947500705718994,
      "step": 9338
    },
    {
      "epoch": 0.7572966576886337,
      "grad_norm": 0.04639193415641785,
      "learning_rate": 0.0012,
      "loss": 2.6190898418426514,
      "step": 9340
    },
    {
      "epoch": 0.7574588197138347,
      "grad_norm": 0.029338853433728218,
      "learning_rate": 0.0012,
      "loss": 2.7169876098632812,
      "step": 9342
    },
    {
      "epoch": 0.7576209817390357,
      "grad_norm": 0.02944025956094265,
      "learning_rate": 0.0012,
      "loss": 2.642357349395752,
      "step": 9344
    },
    {
      "epoch": 0.7577831437642367,
      "grad_norm": 0.03758031502366066,
      "learning_rate": 0.0012,
      "loss": 2.6849045753479004,
      "step": 9346
    },
    {
      "epoch": 0.7579453057894376,
      "grad_norm": 0.03532253950834274,
      "learning_rate": 0.0012,
      "loss": 2.6525416374206543,
      "step": 9348
    },
    {
      "epoch": 0.7581074678146387,
      "grad_norm": 0.038905415683984756,
      "learning_rate": 0.0012,
      "loss": 2.6795473098754883,
      "step": 9350
    },
    {
      "epoch": 0.7582696298398397,
      "grad_norm": 0.02767259068787098,
      "learning_rate": 0.0012,
      "loss": 2.702498435974121,
      "step": 9352
    },
    {
      "epoch": 0.7584317918650406,
      "grad_norm": 0.026526302099227905,
      "learning_rate": 0.0012,
      "loss": 2.729067802429199,
      "step": 9354
    },
    {
      "epoch": 0.7585939538902416,
      "grad_norm": 0.024526681751012802,
      "learning_rate": 0.0012,
      "loss": 2.689577102661133,
      "step": 9356
    },
    {
      "epoch": 0.7587561159154427,
      "grad_norm": 0.02689945697784424,
      "learning_rate": 0.0012,
      "loss": 2.6487598419189453,
      "step": 9358
    },
    {
      "epoch": 0.7589182779406436,
      "grad_norm": 0.0322907455265522,
      "learning_rate": 0.0012,
      "loss": 2.6478638648986816,
      "step": 9360
    },
    {
      "epoch": 0.7590804399658446,
      "grad_norm": 0.026760797947645187,
      "learning_rate": 0.0012,
      "loss": 2.6284470558166504,
      "step": 9362
    },
    {
      "epoch": 0.7592426019910457,
      "grad_norm": 0.04381531849503517,
      "learning_rate": 0.0012,
      "loss": 2.6535143852233887,
      "step": 9364
    },
    {
      "epoch": 0.7594047640162466,
      "grad_norm": 0.03608185797929764,
      "learning_rate": 0.0012,
      "loss": 2.677198886871338,
      "step": 9366
    },
    {
      "epoch": 0.7595669260414476,
      "grad_norm": 0.032977715134620667,
      "learning_rate": 0.0012,
      "loss": 2.605893611907959,
      "step": 9368
    },
    {
      "epoch": 0.7597290880666486,
      "grad_norm": 0.024564692750573158,
      "learning_rate": 0.0012,
      "loss": 2.7591300010681152,
      "step": 9370
    },
    {
      "epoch": 0.7598912500918495,
      "grad_norm": 0.029908301308751106,
      "learning_rate": 0.0012,
      "loss": 2.6227381229400635,
      "step": 9372
    },
    {
      "epoch": 0.7600534121170506,
      "grad_norm": 0.027064897119998932,
      "learning_rate": 0.0012,
      "loss": 2.7046449184417725,
      "step": 9374
    },
    {
      "epoch": 0.7602155741422516,
      "grad_norm": 0.030386321246623993,
      "learning_rate": 0.0012,
      "loss": 2.643106460571289,
      "step": 9376
    },
    {
      "epoch": 0.7603777361674525,
      "grad_norm": 0.03883172944188118,
      "learning_rate": 0.0012,
      "loss": 2.6641697883605957,
      "step": 9378
    },
    {
      "epoch": 0.7605398981926536,
      "grad_norm": 0.052260395139455795,
      "learning_rate": 0.0012,
      "loss": 2.6560277938842773,
      "step": 9380
    },
    {
      "epoch": 0.7607020602178546,
      "grad_norm": 0.027542315423488617,
      "learning_rate": 0.0012,
      "loss": 2.6575212478637695,
      "step": 9382
    },
    {
      "epoch": 0.7608642222430555,
      "grad_norm": 0.030781127512454987,
      "learning_rate": 0.0012,
      "loss": 2.6546876430511475,
      "step": 9384
    },
    {
      "epoch": 0.7610263842682565,
      "grad_norm": 0.029682185500860214,
      "learning_rate": 0.0012,
      "loss": 2.677436351776123,
      "step": 9386
    },
    {
      "epoch": 0.7611885462934576,
      "grad_norm": 0.03472258150577545,
      "learning_rate": 0.0012,
      "loss": 2.686983585357666,
      "step": 9388
    },
    {
      "epoch": 0.7613507083186585,
      "grad_norm": 0.022639311850070953,
      "learning_rate": 0.0012,
      "loss": 2.697421073913574,
      "step": 9390
    },
    {
      "epoch": 0.7615128703438595,
      "grad_norm": 0.05024553835391998,
      "learning_rate": 0.0012,
      "loss": 2.721975564956665,
      "step": 9392
    },
    {
      "epoch": 0.7616750323690605,
      "grad_norm": 0.04923105239868164,
      "learning_rate": 0.0012,
      "loss": 2.6166110038757324,
      "step": 9394
    },
    {
      "epoch": 0.7618371943942615,
      "grad_norm": 0.051235638558864594,
      "learning_rate": 0.0012,
      "loss": 2.7196784019470215,
      "step": 9396
    },
    {
      "epoch": 0.7619993564194625,
      "grad_norm": 0.03321798890829086,
      "learning_rate": 0.0012,
      "loss": 2.666140556335449,
      "step": 9398
    },
    {
      "epoch": 0.7621615184446635,
      "grad_norm": 0.04300227388739586,
      "learning_rate": 0.0012,
      "loss": 2.6273255348205566,
      "step": 9400
    },
    {
      "epoch": 0.7623236804698644,
      "grad_norm": 0.03716682642698288,
      "learning_rate": 0.0012,
      "loss": 2.6836681365966797,
      "step": 9402
    },
    {
      "epoch": 0.7624858424950655,
      "grad_norm": 0.0386650376021862,
      "learning_rate": 0.0012,
      "loss": 2.660360097885132,
      "step": 9404
    },
    {
      "epoch": 0.7626480045202665,
      "grad_norm": 0.050131600350141525,
      "learning_rate": 0.0012,
      "loss": 2.6352853775024414,
      "step": 9406
    },
    {
      "epoch": 0.7628101665454674,
      "grad_norm": 0.026186319068074226,
      "learning_rate": 0.0012,
      "loss": 2.6494266986846924,
      "step": 9408
    },
    {
      "epoch": 0.7629723285706684,
      "grad_norm": 0.025662029162049294,
      "learning_rate": 0.0012,
      "loss": 2.644179105758667,
      "step": 9410
    },
    {
      "epoch": 0.7631344905958695,
      "grad_norm": 0.02770663984119892,
      "learning_rate": 0.0012,
      "loss": 2.6543173789978027,
      "step": 9412
    },
    {
      "epoch": 0.7632966526210704,
      "grad_norm": 0.03782191127538681,
      "learning_rate": 0.0012,
      "loss": 2.6119465827941895,
      "step": 9414
    },
    {
      "epoch": 0.7634588146462714,
      "grad_norm": 0.02786666341125965,
      "learning_rate": 0.0012,
      "loss": 2.7038402557373047,
      "step": 9416
    },
    {
      "epoch": 0.7636209766714724,
      "grad_norm": 0.03181063011288643,
      "learning_rate": 0.0012,
      "loss": 2.646233320236206,
      "step": 9418
    },
    {
      "epoch": 0.7637831386966734,
      "grad_norm": 0.027126338332891464,
      "learning_rate": 0.0012,
      "loss": 2.6051692962646484,
      "step": 9420
    },
    {
      "epoch": 0.7639453007218744,
      "grad_norm": 0.05827576294541359,
      "learning_rate": 0.0012,
      "loss": 2.634376049041748,
      "step": 9422
    },
    {
      "epoch": 0.7641074627470754,
      "grad_norm": 0.049090173095464706,
      "learning_rate": 0.0012,
      "loss": 2.63000226020813,
      "step": 9424
    },
    {
      "epoch": 0.7642696247722763,
      "grad_norm": 0.03802117705345154,
      "learning_rate": 0.0012,
      "loss": 2.645475387573242,
      "step": 9426
    },
    {
      "epoch": 0.7644317867974774,
      "grad_norm": 0.027509016916155815,
      "learning_rate": 0.0012,
      "loss": 2.6527321338653564,
      "step": 9428
    },
    {
      "epoch": 0.7645939488226784,
      "grad_norm": 0.029796866700053215,
      "learning_rate": 0.0012,
      "loss": 2.6275243759155273,
      "step": 9430
    },
    {
      "epoch": 0.7647561108478793,
      "grad_norm": 0.03202444314956665,
      "learning_rate": 0.0012,
      "loss": 2.6562576293945312,
      "step": 9432
    },
    {
      "epoch": 0.7649182728730803,
      "grad_norm": 0.04632434993982315,
      "learning_rate": 0.0012,
      "loss": 2.6280763149261475,
      "step": 9434
    },
    {
      "epoch": 0.7650804348982814,
      "grad_norm": 0.049089767038822174,
      "learning_rate": 0.0012,
      "loss": 2.62638521194458,
      "step": 9436
    },
    {
      "epoch": 0.7652425969234823,
      "grad_norm": 0.025704503059387207,
      "learning_rate": 0.0012,
      "loss": 2.62764310836792,
      "step": 9438
    },
    {
      "epoch": 0.7654047589486833,
      "grad_norm": 0.02520086243748665,
      "learning_rate": 0.0012,
      "loss": 2.729024887084961,
      "step": 9440
    },
    {
      "epoch": 0.7655669209738843,
      "grad_norm": 0.04364878684282303,
      "learning_rate": 0.0012,
      "loss": 2.743967056274414,
      "step": 9442
    },
    {
      "epoch": 0.7657290829990853,
      "grad_norm": 0.02483200468122959,
      "learning_rate": 0.0012,
      "loss": 2.611349582672119,
      "step": 9444
    },
    {
      "epoch": 0.7658912450242863,
      "grad_norm": 0.05087961256504059,
      "learning_rate": 0.0012,
      "loss": 2.667469024658203,
      "step": 9446
    },
    {
      "epoch": 0.7660534070494873,
      "grad_norm": 0.03358987346291542,
      "learning_rate": 0.0012,
      "loss": 2.619309425354004,
      "step": 9448
    },
    {
      "epoch": 0.7662155690746882,
      "grad_norm": 0.03316621482372284,
      "learning_rate": 0.0012,
      "loss": 2.5845589637756348,
      "step": 9450
    },
    {
      "epoch": 0.7663777310998893,
      "grad_norm": 0.03381721302866936,
      "learning_rate": 0.0012,
      "loss": 2.6892967224121094,
      "step": 9452
    },
    {
      "epoch": 0.7665398931250903,
      "grad_norm": 0.036166395992040634,
      "learning_rate": 0.0012,
      "loss": 2.628999710083008,
      "step": 9454
    },
    {
      "epoch": 0.7667020551502912,
      "grad_norm": 0.030548643320798874,
      "learning_rate": 0.0012,
      "loss": 2.6606643199920654,
      "step": 9456
    },
    {
      "epoch": 0.7668642171754922,
      "grad_norm": 0.05580200254917145,
      "learning_rate": 0.0012,
      "loss": 2.605380058288574,
      "step": 9458
    },
    {
      "epoch": 0.7670263792006933,
      "grad_norm": 0.02523045800626278,
      "learning_rate": 0.0012,
      "loss": 2.6168746948242188,
      "step": 9460
    },
    {
      "epoch": 0.7671885412258942,
      "grad_norm": 0.02933035045862198,
      "learning_rate": 0.0012,
      "loss": 2.6598734855651855,
      "step": 9462
    },
    {
      "epoch": 0.7673507032510952,
      "grad_norm": 0.06195637211203575,
      "learning_rate": 0.0012,
      "loss": 2.644397258758545,
      "step": 9464
    },
    {
      "epoch": 0.7675128652762963,
      "grad_norm": 0.03796984627842903,
      "learning_rate": 0.0012,
      "loss": 2.656770706176758,
      "step": 9466
    },
    {
      "epoch": 0.7676750273014972,
      "grad_norm": 0.029292147606611252,
      "learning_rate": 0.0012,
      "loss": 2.6657774448394775,
      "step": 9468
    },
    {
      "epoch": 0.7678371893266982,
      "grad_norm": 0.07188896089792252,
      "learning_rate": 0.0012,
      "loss": 2.591456413269043,
      "step": 9470
    },
    {
      "epoch": 0.7679993513518992,
      "grad_norm": 0.042273811995983124,
      "learning_rate": 0.0012,
      "loss": 2.6607666015625,
      "step": 9472
    },
    {
      "epoch": 0.7681615133771001,
      "grad_norm": 0.03958675265312195,
      "learning_rate": 0.0012,
      "loss": 2.6274936199188232,
      "step": 9474
    },
    {
      "epoch": 0.7683236754023012,
      "grad_norm": 0.031437359750270844,
      "learning_rate": 0.0012,
      "loss": 2.616342544555664,
      "step": 9476
    },
    {
      "epoch": 0.7684858374275022,
      "grad_norm": 0.02523067221045494,
      "learning_rate": 0.0012,
      "loss": 2.644907236099243,
      "step": 9478
    },
    {
      "epoch": 0.7686479994527031,
      "grad_norm": 0.03296936675906181,
      "learning_rate": 0.0012,
      "loss": 2.6441030502319336,
      "step": 9480
    },
    {
      "epoch": 0.7688101614779042,
      "grad_norm": 0.03898452967405319,
      "learning_rate": 0.0012,
      "loss": 2.657081127166748,
      "step": 9482
    },
    {
      "epoch": 0.7689723235031052,
      "grad_norm": 0.03575950115919113,
      "learning_rate": 0.0012,
      "loss": 2.680959701538086,
      "step": 9484
    },
    {
      "epoch": 0.7691344855283061,
      "grad_norm": 0.06511750817298889,
      "learning_rate": 0.0012,
      "loss": 2.651679754257202,
      "step": 9486
    },
    {
      "epoch": 0.7692966475535071,
      "grad_norm": 0.04431906342506409,
      "learning_rate": 0.0012,
      "loss": 2.6876988410949707,
      "step": 9488
    },
    {
      "epoch": 0.7694588095787082,
      "grad_norm": 0.028731591999530792,
      "learning_rate": 0.0012,
      "loss": 2.642364501953125,
      "step": 9490
    },
    {
      "epoch": 0.7696209716039091,
      "grad_norm": 0.05041513219475746,
      "learning_rate": 0.0012,
      "loss": 2.636519432067871,
      "step": 9492
    },
    {
      "epoch": 0.7697831336291101,
      "grad_norm": 0.029651015996932983,
      "learning_rate": 0.0012,
      "loss": 2.650991916656494,
      "step": 9494
    },
    {
      "epoch": 0.7699452956543111,
      "grad_norm": 0.028326386585831642,
      "learning_rate": 0.0012,
      "loss": 2.6348891258239746,
      "step": 9496
    },
    {
      "epoch": 0.770107457679512,
      "grad_norm": 0.03923293948173523,
      "learning_rate": 0.0012,
      "loss": 2.6383423805236816,
      "step": 9498
    },
    {
      "epoch": 0.7702696197047131,
      "grad_norm": 0.03412853926420212,
      "learning_rate": 0.0012,
      "loss": 2.678445339202881,
      "step": 9500
    },
    {
      "epoch": 0.7704317817299141,
      "grad_norm": 0.029896950349211693,
      "learning_rate": 0.0012,
      "loss": 2.6878745555877686,
      "step": 9502
    },
    {
      "epoch": 0.770593943755115,
      "grad_norm": 0.03638007491827011,
      "learning_rate": 0.0012,
      "loss": 2.652695894241333,
      "step": 9504
    },
    {
      "epoch": 0.7707561057803161,
      "grad_norm": 0.025138627737760544,
      "learning_rate": 0.0012,
      "loss": 2.630190372467041,
      "step": 9506
    },
    {
      "epoch": 0.7709182678055171,
      "grad_norm": 0.027232492342591286,
      "learning_rate": 0.0012,
      "loss": 2.6719675064086914,
      "step": 9508
    },
    {
      "epoch": 0.771080429830718,
      "grad_norm": 0.03845582529902458,
      "learning_rate": 0.0012,
      "loss": 2.6414341926574707,
      "step": 9510
    },
    {
      "epoch": 0.771242591855919,
      "grad_norm": 0.04394492134451866,
      "learning_rate": 0.0012,
      "loss": 2.623845338821411,
      "step": 9512
    },
    {
      "epoch": 0.7714047538811201,
      "grad_norm": 0.0281525868922472,
      "learning_rate": 0.0012,
      "loss": 2.6360716819763184,
      "step": 9514
    },
    {
      "epoch": 0.771566915906321,
      "grad_norm": 0.0349125862121582,
      "learning_rate": 0.0012,
      "loss": 2.670957565307617,
      "step": 9516
    },
    {
      "epoch": 0.771729077931522,
      "grad_norm": 0.07639529556035995,
      "learning_rate": 0.0012,
      "loss": 2.629842758178711,
      "step": 9518
    },
    {
      "epoch": 0.771891239956723,
      "grad_norm": 0.03341196849942207,
      "learning_rate": 0.0012,
      "loss": 2.650888442993164,
      "step": 9520
    },
    {
      "epoch": 0.772053401981924,
      "grad_norm": 0.0322178490459919,
      "learning_rate": 0.0012,
      "loss": 2.7535245418548584,
      "step": 9522
    },
    {
      "epoch": 0.772215564007125,
      "grad_norm": 0.037463992834091187,
      "learning_rate": 0.0012,
      "loss": 2.7506837844848633,
      "step": 9524
    },
    {
      "epoch": 0.772377726032326,
      "grad_norm": 0.02641293779015541,
      "learning_rate": 0.0012,
      "loss": 2.703479290008545,
      "step": 9526
    },
    {
      "epoch": 0.7725398880575269,
      "grad_norm": 0.05221620202064514,
      "learning_rate": 0.0012,
      "loss": 2.70416259765625,
      "step": 9528
    },
    {
      "epoch": 0.772702050082728,
      "grad_norm": 0.039568204432725906,
      "learning_rate": 0.0012,
      "loss": 2.5975470542907715,
      "step": 9530
    },
    {
      "epoch": 0.772864212107929,
      "grad_norm": 0.02865748666226864,
      "learning_rate": 0.0012,
      "loss": 2.6594080924987793,
      "step": 9532
    },
    {
      "epoch": 0.7730263741331299,
      "grad_norm": 0.02645178511738777,
      "learning_rate": 0.0012,
      "loss": 2.6286098957061768,
      "step": 9534
    },
    {
      "epoch": 0.7731885361583309,
      "grad_norm": 0.03286132588982582,
      "learning_rate": 0.0012,
      "loss": 2.6123929023742676,
      "step": 9536
    },
    {
      "epoch": 0.773350698183532,
      "grad_norm": 0.023958854377269745,
      "learning_rate": 0.0012,
      "loss": 2.691554307937622,
      "step": 9538
    },
    {
      "epoch": 0.7735128602087329,
      "grad_norm": 0.029209354892373085,
      "learning_rate": 0.0012,
      "loss": 2.683382511138916,
      "step": 9540
    },
    {
      "epoch": 0.7736750222339339,
      "grad_norm": 0.03297377750277519,
      "learning_rate": 0.0012,
      "loss": 2.6253550052642822,
      "step": 9542
    },
    {
      "epoch": 0.773837184259135,
      "grad_norm": 0.027869023382663727,
      "learning_rate": 0.0012,
      "loss": 2.6081013679504395,
      "step": 9544
    },
    {
      "epoch": 0.7739993462843359,
      "grad_norm": 0.04669741541147232,
      "learning_rate": 0.0012,
      "loss": 2.7005434036254883,
      "step": 9546
    },
    {
      "epoch": 0.7741615083095369,
      "grad_norm": 0.02577287331223488,
      "learning_rate": 0.0012,
      "loss": 2.679750442504883,
      "step": 9548
    },
    {
      "epoch": 0.7743236703347379,
      "grad_norm": 0.030850380659103394,
      "learning_rate": 0.0012,
      "loss": 2.683534622192383,
      "step": 9550
    },
    {
      "epoch": 0.7744858323599388,
      "grad_norm": 0.025618625804781914,
      "learning_rate": 0.0012,
      "loss": 2.6371665000915527,
      "step": 9552
    },
    {
      "epoch": 0.7746479943851399,
      "grad_norm": 0.0350211076438427,
      "learning_rate": 0.0012,
      "loss": 2.6577608585357666,
      "step": 9554
    },
    {
      "epoch": 0.7748101564103409,
      "grad_norm": 0.040963415056467056,
      "learning_rate": 0.0012,
      "loss": 2.6772451400756836,
      "step": 9556
    },
    {
      "epoch": 0.7749723184355418,
      "grad_norm": 0.025556832551956177,
      "learning_rate": 0.0012,
      "loss": 2.63637113571167,
      "step": 9558
    },
    {
      "epoch": 0.7751344804607428,
      "grad_norm": 0.02497340552508831,
      "learning_rate": 0.0012,
      "loss": 2.652113199234009,
      "step": 9560
    },
    {
      "epoch": 0.7752966424859439,
      "grad_norm": 0.02538461610674858,
      "learning_rate": 0.0012,
      "loss": 2.650387763977051,
      "step": 9562
    },
    {
      "epoch": 0.7754588045111448,
      "grad_norm": 0.025537269189953804,
      "learning_rate": 0.0012,
      "loss": 2.649181365966797,
      "step": 9564
    },
    {
      "epoch": 0.7756209665363458,
      "grad_norm": 0.03190247714519501,
      "learning_rate": 0.0012,
      "loss": 2.6225790977478027,
      "step": 9566
    },
    {
      "epoch": 0.7757831285615469,
      "grad_norm": 0.0396827831864357,
      "learning_rate": 0.0012,
      "loss": 2.664701223373413,
      "step": 9568
    },
    {
      "epoch": 0.7759452905867478,
      "grad_norm": 0.053706731647253036,
      "learning_rate": 0.0012,
      "loss": 2.6770248413085938,
      "step": 9570
    },
    {
      "epoch": 0.7761074526119488,
      "grad_norm": 0.028105873614549637,
      "learning_rate": 0.0012,
      "loss": 2.6369714736938477,
      "step": 9572
    },
    {
      "epoch": 0.7762696146371498,
      "grad_norm": 0.028150862082839012,
      "learning_rate": 0.0012,
      "loss": 2.704181671142578,
      "step": 9574
    },
    {
      "epoch": 0.7764317766623507,
      "grad_norm": 0.08676321059465408,
      "learning_rate": 0.0012,
      "loss": 2.668858528137207,
      "step": 9576
    },
    {
      "epoch": 0.7765939386875518,
      "grad_norm": 0.04062314331531525,
      "learning_rate": 0.0012,
      "loss": 2.714132785797119,
      "step": 9578
    },
    {
      "epoch": 0.7767561007127528,
      "grad_norm": 0.03234368562698364,
      "learning_rate": 0.0012,
      "loss": 2.63460636138916,
      "step": 9580
    },
    {
      "epoch": 0.7769182627379537,
      "grad_norm": 0.030271533876657486,
      "learning_rate": 0.0012,
      "loss": 2.693812370300293,
      "step": 9582
    },
    {
      "epoch": 0.7770804247631548,
      "grad_norm": 0.037179648876190186,
      "learning_rate": 0.0012,
      "loss": 2.6395280361175537,
      "step": 9584
    },
    {
      "epoch": 0.7772425867883558,
      "grad_norm": 0.03974525257945061,
      "learning_rate": 0.0012,
      "loss": 2.6329104900360107,
      "step": 9586
    },
    {
      "epoch": 0.7774047488135567,
      "grad_norm": 0.03780752792954445,
      "learning_rate": 0.0012,
      "loss": 2.6616249084472656,
      "step": 9588
    },
    {
      "epoch": 0.7775669108387577,
      "grad_norm": 0.028776543214917183,
      "learning_rate": 0.0012,
      "loss": 2.654162883758545,
      "step": 9590
    },
    {
      "epoch": 0.7777290728639588,
      "grad_norm": 0.04700135439634323,
      "learning_rate": 0.0012,
      "loss": 2.628736972808838,
      "step": 9592
    },
    {
      "epoch": 0.7778912348891597,
      "grad_norm": 0.034621793776750565,
      "learning_rate": 0.0012,
      "loss": 2.700518846511841,
      "step": 9594
    },
    {
      "epoch": 0.7780533969143607,
      "grad_norm": 0.04238909110426903,
      "learning_rate": 0.0012,
      "loss": 2.6426453590393066,
      "step": 9596
    },
    {
      "epoch": 0.7782155589395617,
      "grad_norm": 0.04331514239311218,
      "learning_rate": 0.0012,
      "loss": 2.6227118968963623,
      "step": 9598
    },
    {
      "epoch": 0.7783777209647627,
      "grad_norm": 0.027354417368769646,
      "learning_rate": 0.0012,
      "loss": 2.6643567085266113,
      "step": 9600
    },
    {
      "epoch": 0.7785398829899637,
      "grad_norm": 0.023310955613851547,
      "learning_rate": 0.0012,
      "loss": 2.6742749214172363,
      "step": 9602
    },
    {
      "epoch": 0.7787020450151647,
      "grad_norm": 0.023778919130563736,
      "learning_rate": 0.0012,
      "loss": 2.6854310035705566,
      "step": 9604
    },
    {
      "epoch": 0.7788642070403656,
      "grad_norm": 0.025214264169335365,
      "learning_rate": 0.0012,
      "loss": 2.6482481956481934,
      "step": 9606
    },
    {
      "epoch": 0.7790263690655667,
      "grad_norm": 0.02403286099433899,
      "learning_rate": 0.0012,
      "loss": 2.6429600715637207,
      "step": 9608
    },
    {
      "epoch": 0.7791885310907677,
      "grad_norm": 0.02694893442094326,
      "learning_rate": 0.0012,
      "loss": 2.5631186962127686,
      "step": 9610
    },
    {
      "epoch": 0.7793506931159686,
      "grad_norm": 0.027689985930919647,
      "learning_rate": 0.0012,
      "loss": 2.6735668182373047,
      "step": 9612
    },
    {
      "epoch": 0.7795128551411696,
      "grad_norm": 0.02550623007118702,
      "learning_rate": 0.0012,
      "loss": 2.696458101272583,
      "step": 9614
    },
    {
      "epoch": 0.7796750171663707,
      "grad_norm": 0.023115862160921097,
      "learning_rate": 0.0012,
      "loss": 2.6530659198760986,
      "step": 9616
    },
    {
      "epoch": 0.7798371791915716,
      "grad_norm": 0.02720983326435089,
      "learning_rate": 0.0012,
      "loss": 2.5734567642211914,
      "step": 9618
    },
    {
      "epoch": 0.7799993412167726,
      "grad_norm": 0.02553495019674301,
      "learning_rate": 0.0012,
      "loss": 2.6773033142089844,
      "step": 9620
    },
    {
      "epoch": 0.7801615032419736,
      "grad_norm": 0.027729958295822144,
      "learning_rate": 0.0012,
      "loss": 2.6246023178100586,
      "step": 9622
    },
    {
      "epoch": 0.7803236652671746,
      "grad_norm": 0.027583535760641098,
      "learning_rate": 0.0012,
      "loss": 2.6634230613708496,
      "step": 9624
    },
    {
      "epoch": 0.7804858272923756,
      "grad_norm": 0.024182282388210297,
      "learning_rate": 0.0012,
      "loss": 2.6185123920440674,
      "step": 9626
    },
    {
      "epoch": 0.7806479893175766,
      "grad_norm": 0.024541100487113,
      "learning_rate": 0.0012,
      "loss": 2.658738136291504,
      "step": 9628
    },
    {
      "epoch": 0.7808101513427775,
      "grad_norm": 0.051042284816503525,
      "learning_rate": 0.0012,
      "loss": 2.6338558197021484,
      "step": 9630
    },
    {
      "epoch": 0.7809723133679786,
      "grad_norm": 0.0628514215350151,
      "learning_rate": 0.0012,
      "loss": 2.6785130500793457,
      "step": 9632
    },
    {
      "epoch": 0.7811344753931796,
      "grad_norm": 0.027543602511286736,
      "learning_rate": 0.0012,
      "loss": 2.619886875152588,
      "step": 9634
    },
    {
      "epoch": 0.7812966374183805,
      "grad_norm": 0.03251085802912712,
      "learning_rate": 0.0012,
      "loss": 2.5913124084472656,
      "step": 9636
    },
    {
      "epoch": 0.7814587994435815,
      "grad_norm": 0.029242269694805145,
      "learning_rate": 0.0012,
      "loss": 2.729214906692505,
      "step": 9638
    },
    {
      "epoch": 0.7816209614687826,
      "grad_norm": 0.03072386234998703,
      "learning_rate": 0.0012,
      "loss": 2.6643614768981934,
      "step": 9640
    },
    {
      "epoch": 0.7817831234939835,
      "grad_norm": 0.0253499336540699,
      "learning_rate": 0.0012,
      "loss": 2.669431447982788,
      "step": 9642
    },
    {
      "epoch": 0.7819452855191845,
      "grad_norm": 0.02603083848953247,
      "learning_rate": 0.0012,
      "loss": 2.6175553798675537,
      "step": 9644
    },
    {
      "epoch": 0.7821074475443855,
      "grad_norm": 0.029698414728045464,
      "learning_rate": 0.0012,
      "loss": 2.7088794708251953,
      "step": 9646
    },
    {
      "epoch": 0.7822696095695865,
      "grad_norm": 0.03601466864347458,
      "learning_rate": 0.0012,
      "loss": 2.6938841342926025,
      "step": 9648
    },
    {
      "epoch": 0.7824317715947875,
      "grad_norm": 0.06133940443396568,
      "learning_rate": 0.0012,
      "loss": 2.669773578643799,
      "step": 9650
    },
    {
      "epoch": 0.7825939336199885,
      "grad_norm": 0.046449486166238785,
      "learning_rate": 0.0012,
      "loss": 2.634824752807617,
      "step": 9652
    },
    {
      "epoch": 0.7827560956451894,
      "grad_norm": 0.03564751520752907,
      "learning_rate": 0.0012,
      "loss": 2.6316421031951904,
      "step": 9654
    },
    {
      "epoch": 0.7829182576703905,
      "grad_norm": 0.024807196110486984,
      "learning_rate": 0.0012,
      "loss": 2.6585092544555664,
      "step": 9656
    },
    {
      "epoch": 0.7830804196955915,
      "grad_norm": 0.026102259755134583,
      "learning_rate": 0.0012,
      "loss": 2.715579032897949,
      "step": 9658
    },
    {
      "epoch": 0.7832425817207924,
      "grad_norm": 0.03559070825576782,
      "learning_rate": 0.0012,
      "loss": 2.6685619354248047,
      "step": 9660
    },
    {
      "epoch": 0.7834047437459934,
      "grad_norm": 0.04341878369450569,
      "learning_rate": 0.0012,
      "loss": 2.628077983856201,
      "step": 9662
    },
    {
      "epoch": 0.7835669057711945,
      "grad_norm": 0.03223796188831329,
      "learning_rate": 0.0012,
      "loss": 2.6638681888580322,
      "step": 9664
    },
    {
      "epoch": 0.7837290677963954,
      "grad_norm": 0.028088251128792763,
      "learning_rate": 0.0012,
      "loss": 2.598598003387451,
      "step": 9666
    },
    {
      "epoch": 0.7838912298215964,
      "grad_norm": 0.02613641694188118,
      "learning_rate": 0.0012,
      "loss": 2.7035961151123047,
      "step": 9668
    },
    {
      "epoch": 0.7840533918467975,
      "grad_norm": 0.03808344155550003,
      "learning_rate": 0.0012,
      "loss": 2.672102451324463,
      "step": 9670
    },
    {
      "epoch": 0.7842155538719984,
      "grad_norm": 0.03766528144478798,
      "learning_rate": 0.0012,
      "loss": 2.6717491149902344,
      "step": 9672
    },
    {
      "epoch": 0.7843777158971994,
      "grad_norm": 0.036697909235954285,
      "learning_rate": 0.0012,
      "loss": 2.614652156829834,
      "step": 9674
    },
    {
      "epoch": 0.7845398779224004,
      "grad_norm": 0.026880202814936638,
      "learning_rate": 0.0012,
      "loss": 2.6736981868743896,
      "step": 9676
    },
    {
      "epoch": 0.7847020399476013,
      "grad_norm": 0.060649510473012924,
      "learning_rate": 0.0012,
      "loss": 2.6009368896484375,
      "step": 9678
    },
    {
      "epoch": 0.7848642019728024,
      "grad_norm": 0.0448935441672802,
      "learning_rate": 0.0012,
      "loss": 2.683320999145508,
      "step": 9680
    },
    {
      "epoch": 0.7850263639980034,
      "grad_norm": 0.0363609716296196,
      "learning_rate": 0.0012,
      "loss": 2.6386661529541016,
      "step": 9682
    },
    {
      "epoch": 0.7851885260232043,
      "grad_norm": 0.02550467662513256,
      "learning_rate": 0.0012,
      "loss": 2.6235995292663574,
      "step": 9684
    },
    {
      "epoch": 0.7853506880484054,
      "grad_norm": 0.05191199854016304,
      "learning_rate": 0.0012,
      "loss": 2.6515321731567383,
      "step": 9686
    },
    {
      "epoch": 0.7855128500736064,
      "grad_norm": 0.031650081276893616,
      "learning_rate": 0.0012,
      "loss": 2.636188507080078,
      "step": 9688
    },
    {
      "epoch": 0.7856750120988073,
      "grad_norm": 0.03313637897372246,
      "learning_rate": 0.0012,
      "loss": 2.7209959030151367,
      "step": 9690
    },
    {
      "epoch": 0.7858371741240083,
      "grad_norm": 0.02625713311135769,
      "learning_rate": 0.0012,
      "loss": 2.5580244064331055,
      "step": 9692
    },
    {
      "epoch": 0.7859993361492094,
      "grad_norm": 0.025363247841596603,
      "learning_rate": 0.0012,
      "loss": 2.6847147941589355,
      "step": 9694
    },
    {
      "epoch": 0.7861614981744103,
      "grad_norm": 0.033036742359399796,
      "learning_rate": 0.0012,
      "loss": 2.6738545894622803,
      "step": 9696
    },
    {
      "epoch": 0.7863236601996113,
      "grad_norm": 0.029086139053106308,
      "learning_rate": 0.0012,
      "loss": 2.6576790809631348,
      "step": 9698
    },
    {
      "epoch": 0.7864858222248123,
      "grad_norm": 0.040166258811950684,
      "learning_rate": 0.0012,
      "loss": 2.655071496963501,
      "step": 9700
    },
    {
      "epoch": 0.7866479842500133,
      "grad_norm": 0.048837997019290924,
      "learning_rate": 0.0012,
      "loss": 2.659579277038574,
      "step": 9702
    },
    {
      "epoch": 0.7868101462752143,
      "grad_norm": 0.07646498084068298,
      "learning_rate": 0.0012,
      "loss": 2.714688777923584,
      "step": 9704
    },
    {
      "epoch": 0.7869723083004153,
      "grad_norm": 0.043853938579559326,
      "learning_rate": 0.0012,
      "loss": 2.7146942615509033,
      "step": 9706
    },
    {
      "epoch": 0.7871344703256162,
      "grad_norm": 0.023881368339061737,
      "learning_rate": 0.0012,
      "loss": 2.7043588161468506,
      "step": 9708
    },
    {
      "epoch": 0.7872966323508173,
      "grad_norm": 0.0515604205429554,
      "learning_rate": 0.0012,
      "loss": 2.6892263889312744,
      "step": 9710
    },
    {
      "epoch": 0.7874587943760183,
      "grad_norm": 0.05787713825702667,
      "learning_rate": 0.0012,
      "loss": 2.684128999710083,
      "step": 9712
    },
    {
      "epoch": 0.7876209564012192,
      "grad_norm": 0.04442562907934189,
      "learning_rate": 0.0012,
      "loss": 2.654999017715454,
      "step": 9714
    },
    {
      "epoch": 0.7877831184264202,
      "grad_norm": 0.026372501626610756,
      "learning_rate": 0.0012,
      "loss": 2.6625049114227295,
      "step": 9716
    },
    {
      "epoch": 0.7879452804516213,
      "grad_norm": 0.04329940676689148,
      "learning_rate": 0.0012,
      "loss": 2.653411865234375,
      "step": 9718
    },
    {
      "epoch": 0.7881074424768222,
      "grad_norm": 0.07914422452449799,
      "learning_rate": 0.0012,
      "loss": 2.732595920562744,
      "step": 9720
    },
    {
      "epoch": 0.7882696045020232,
      "grad_norm": 0.0268681813031435,
      "learning_rate": 0.0012,
      "loss": 2.6631016731262207,
      "step": 9722
    },
    {
      "epoch": 0.7884317665272242,
      "grad_norm": 0.03210645169019699,
      "learning_rate": 0.0012,
      "loss": 2.6910743713378906,
      "step": 9724
    },
    {
      "epoch": 0.7885939285524252,
      "grad_norm": 0.027887284755706787,
      "learning_rate": 0.0012,
      "loss": 2.61380672454834,
      "step": 9726
    },
    {
      "epoch": 0.7887560905776262,
      "grad_norm": 0.02428358420729637,
      "learning_rate": 0.0012,
      "loss": 2.640291690826416,
      "step": 9728
    },
    {
      "epoch": 0.7889182526028272,
      "grad_norm": 0.028155993670225143,
      "learning_rate": 0.0012,
      "loss": 2.588564395904541,
      "step": 9730
    },
    {
      "epoch": 0.7890804146280281,
      "grad_norm": 0.03478361666202545,
      "learning_rate": 0.0012,
      "loss": 2.7315993309020996,
      "step": 9732
    },
    {
      "epoch": 0.7892425766532292,
      "grad_norm": 0.02448839507997036,
      "learning_rate": 0.0012,
      "loss": 2.6705002784729004,
      "step": 9734
    },
    {
      "epoch": 0.7894047386784302,
      "grad_norm": 0.026164405047893524,
      "learning_rate": 0.0012,
      "loss": 2.670651435852051,
      "step": 9736
    },
    {
      "epoch": 0.7895669007036311,
      "grad_norm": 0.02867026813328266,
      "learning_rate": 0.0012,
      "loss": 2.709372043609619,
      "step": 9738
    },
    {
      "epoch": 0.7897290627288321,
      "grad_norm": 0.03842611610889435,
      "learning_rate": 0.0012,
      "loss": 2.637329339981079,
      "step": 9740
    },
    {
      "epoch": 0.7898912247540332,
      "grad_norm": 0.03629661351442337,
      "learning_rate": 0.0012,
      "loss": 2.6319122314453125,
      "step": 9742
    },
    {
      "epoch": 0.7900533867792341,
      "grad_norm": 0.03782528266310692,
      "learning_rate": 0.0012,
      "loss": 2.6529927253723145,
      "step": 9744
    },
    {
      "epoch": 0.7902155488044351,
      "grad_norm": 0.02839243970811367,
      "learning_rate": 0.0012,
      "loss": 2.7000985145568848,
      "step": 9746
    },
    {
      "epoch": 0.7903777108296361,
      "grad_norm": 0.026311799883842468,
      "learning_rate": 0.0012,
      "loss": 2.6918911933898926,
      "step": 9748
    },
    {
      "epoch": 0.7905398728548371,
      "grad_norm": 0.03964696079492569,
      "learning_rate": 0.0012,
      "loss": 2.6427793502807617,
      "step": 9750
    },
    {
      "epoch": 0.7907020348800381,
      "grad_norm": 0.036189038306474686,
      "learning_rate": 0.0012,
      "loss": 2.692821979522705,
      "step": 9752
    },
    {
      "epoch": 0.7908641969052391,
      "grad_norm": 0.042179640382528305,
      "learning_rate": 0.0012,
      "loss": 2.633915901184082,
      "step": 9754
    },
    {
      "epoch": 0.79102635893044,
      "grad_norm": 0.04623764380812645,
      "learning_rate": 0.0012,
      "loss": 2.6685609817504883,
      "step": 9756
    },
    {
      "epoch": 0.7911885209556411,
      "grad_norm": 0.025144554674625397,
      "learning_rate": 0.0012,
      "loss": 2.6221988201141357,
      "step": 9758
    },
    {
      "epoch": 0.7913506829808421,
      "grad_norm": 0.029733806848526,
      "learning_rate": 0.0012,
      "loss": 2.67672061920166,
      "step": 9760
    },
    {
      "epoch": 0.791512845006043,
      "grad_norm": 0.031016720458865166,
      "learning_rate": 0.0012,
      "loss": 2.6830403804779053,
      "step": 9762
    },
    {
      "epoch": 0.791675007031244,
      "grad_norm": 0.034470461308956146,
      "learning_rate": 0.0012,
      "loss": 2.6670970916748047,
      "step": 9764
    },
    {
      "epoch": 0.7918371690564451,
      "grad_norm": 0.036130424588918686,
      "learning_rate": 0.0012,
      "loss": 2.671633720397949,
      "step": 9766
    },
    {
      "epoch": 0.791999331081646,
      "grad_norm": 0.05913977697491646,
      "learning_rate": 0.0012,
      "loss": 2.6447973251342773,
      "step": 9768
    },
    {
      "epoch": 0.792161493106847,
      "grad_norm": 0.08227912336587906,
      "learning_rate": 0.0012,
      "loss": 2.6986989974975586,
      "step": 9770
    },
    {
      "epoch": 0.792323655132048,
      "grad_norm": 0.03649021312594414,
      "learning_rate": 0.0012,
      "loss": 2.599574089050293,
      "step": 9772
    },
    {
      "epoch": 0.792485817157249,
      "grad_norm": 0.046859435737133026,
      "learning_rate": 0.0012,
      "loss": 2.655752182006836,
      "step": 9774
    },
    {
      "epoch": 0.79264797918245,
      "grad_norm": 0.05610016733407974,
      "learning_rate": 0.0012,
      "loss": 2.666111946105957,
      "step": 9776
    },
    {
      "epoch": 0.792810141207651,
      "grad_norm": 0.04778391495347023,
      "learning_rate": 0.0012,
      "loss": 2.5984115600585938,
      "step": 9778
    },
    {
      "epoch": 0.792972303232852,
      "grad_norm": 0.028632869943976402,
      "learning_rate": 0.0012,
      "loss": 2.6469740867614746,
      "step": 9780
    },
    {
      "epoch": 0.793134465258053,
      "grad_norm": 0.06076015532016754,
      "learning_rate": 0.0012,
      "loss": 2.6659092903137207,
      "step": 9782
    },
    {
      "epoch": 0.793296627283254,
      "grad_norm": 0.04902735725045204,
      "learning_rate": 0.0012,
      "loss": 2.685762405395508,
      "step": 9784
    },
    {
      "epoch": 0.7934587893084549,
      "grad_norm": 0.049666568636894226,
      "learning_rate": 0.0012,
      "loss": 2.686411142349243,
      "step": 9786
    },
    {
      "epoch": 0.793620951333656,
      "grad_norm": 0.026542024686932564,
      "learning_rate": 0.0012,
      "loss": 2.6622304916381836,
      "step": 9788
    },
    {
      "epoch": 0.793783113358857,
      "grad_norm": 0.03571649640798569,
      "learning_rate": 0.0012,
      "loss": 2.6236824989318848,
      "step": 9790
    },
    {
      "epoch": 0.7939452753840579,
      "grad_norm": 0.027981432154774666,
      "learning_rate": 0.0012,
      "loss": 2.660613536834717,
      "step": 9792
    },
    {
      "epoch": 0.7941074374092589,
      "grad_norm": 0.041257958859205246,
      "learning_rate": 0.0012,
      "loss": 2.7032790184020996,
      "step": 9794
    },
    {
      "epoch": 0.79426959943446,
      "grad_norm": 0.048123303800821304,
      "learning_rate": 0.0012,
      "loss": 2.6354336738586426,
      "step": 9796
    },
    {
      "epoch": 0.7944317614596609,
      "grad_norm": 0.03485047444701195,
      "learning_rate": 0.0012,
      "loss": 2.678628921508789,
      "step": 9798
    },
    {
      "epoch": 0.7945939234848619,
      "grad_norm": 0.024915048852562904,
      "learning_rate": 0.0012,
      "loss": 2.592543363571167,
      "step": 9800
    },
    {
      "epoch": 0.7947560855100629,
      "grad_norm": 0.06541505455970764,
      "learning_rate": 0.0012,
      "loss": 2.706625461578369,
      "step": 9802
    },
    {
      "epoch": 0.7949182475352639,
      "grad_norm": 0.052340004593133926,
      "learning_rate": 0.0012,
      "loss": 2.761000633239746,
      "step": 9804
    },
    {
      "epoch": 0.7950804095604649,
      "grad_norm": 0.027156496420502663,
      "learning_rate": 0.0012,
      "loss": 2.6195900440216064,
      "step": 9806
    },
    {
      "epoch": 0.7952425715856659,
      "grad_norm": 0.04535486176609993,
      "learning_rate": 0.0012,
      "loss": 2.6855006217956543,
      "step": 9808
    },
    {
      "epoch": 0.7954047336108668,
      "grad_norm": 0.042150117456912994,
      "learning_rate": 0.0012,
      "loss": 2.664659023284912,
      "step": 9810
    },
    {
      "epoch": 0.7955668956360679,
      "grad_norm": 0.028376474976539612,
      "learning_rate": 0.0012,
      "loss": 2.6694369316101074,
      "step": 9812
    },
    {
      "epoch": 0.7957290576612689,
      "grad_norm": 0.02590375766158104,
      "learning_rate": 0.0012,
      "loss": 2.6125943660736084,
      "step": 9814
    },
    {
      "epoch": 0.7958912196864698,
      "grad_norm": 0.04485686495900154,
      "learning_rate": 0.0012,
      "loss": 2.650102138519287,
      "step": 9816
    },
    {
      "epoch": 0.7960533817116708,
      "grad_norm": 0.03034697286784649,
      "learning_rate": 0.0012,
      "loss": 2.685514211654663,
      "step": 9818
    },
    {
      "epoch": 0.7962155437368719,
      "grad_norm": 0.029457952827215195,
      "learning_rate": 0.0012,
      "loss": 2.6339166164398193,
      "step": 9820
    },
    {
      "epoch": 0.7963777057620728,
      "grad_norm": 0.06166757643222809,
      "learning_rate": 0.0012,
      "loss": 2.666276216506958,
      "step": 9822
    },
    {
      "epoch": 0.7965398677872738,
      "grad_norm": 0.06761802732944489,
      "learning_rate": 0.0012,
      "loss": 2.7056684494018555,
      "step": 9824
    },
    {
      "epoch": 0.7967020298124748,
      "grad_norm": 0.028144823387265205,
      "learning_rate": 0.0012,
      "loss": 2.68034291267395,
      "step": 9826
    },
    {
      "epoch": 0.7968641918376758,
      "grad_norm": 0.03236342594027519,
      "learning_rate": 0.0012,
      "loss": 2.6033358573913574,
      "step": 9828
    },
    {
      "epoch": 0.7970263538628768,
      "grad_norm": 0.051006365567445755,
      "learning_rate": 0.0012,
      "loss": 2.687283515930176,
      "step": 9830
    },
    {
      "epoch": 0.7971885158880778,
      "grad_norm": 0.02466234751045704,
      "learning_rate": 0.0012,
      "loss": 2.7022111415863037,
      "step": 9832
    },
    {
      "epoch": 0.7973506779132787,
      "grad_norm": 0.04784243553876877,
      "learning_rate": 0.0012,
      "loss": 2.6570239067077637,
      "step": 9834
    },
    {
      "epoch": 0.7975128399384798,
      "grad_norm": 0.04228174313902855,
      "learning_rate": 0.0012,
      "loss": 2.6382036209106445,
      "step": 9836
    },
    {
      "epoch": 0.7976750019636808,
      "grad_norm": 0.03283889219164848,
      "learning_rate": 0.0012,
      "loss": 2.616352081298828,
      "step": 9838
    },
    {
      "epoch": 0.7978371639888817,
      "grad_norm": 0.08032308518886566,
      "learning_rate": 0.0012,
      "loss": 2.6637842655181885,
      "step": 9840
    },
    {
      "epoch": 0.7979993260140827,
      "grad_norm": 0.08065615594387054,
      "learning_rate": 0.0012,
      "loss": 2.6955747604370117,
      "step": 9842
    },
    {
      "epoch": 0.7981614880392838,
      "grad_norm": 0.027895813807845116,
      "learning_rate": 0.0012,
      "loss": 2.657665729522705,
      "step": 9844
    },
    {
      "epoch": 0.7983236500644847,
      "grad_norm": 0.045080944895744324,
      "learning_rate": 0.0012,
      "loss": 2.6159608364105225,
      "step": 9846
    },
    {
      "epoch": 0.7984858120896857,
      "grad_norm": 0.05267564207315445,
      "learning_rate": 0.0012,
      "loss": 2.7044262886047363,
      "step": 9848
    },
    {
      "epoch": 0.7986479741148867,
      "grad_norm": 0.031611792743206024,
      "learning_rate": 0.0012,
      "loss": 2.6946568489074707,
      "step": 9850
    },
    {
      "epoch": 0.7988101361400877,
      "grad_norm": 0.036134425550699234,
      "learning_rate": 0.0012,
      "loss": 2.624284029006958,
      "step": 9852
    },
    {
      "epoch": 0.7989722981652887,
      "grad_norm": 0.03139821067452431,
      "learning_rate": 0.0012,
      "loss": 2.6560540199279785,
      "step": 9854
    },
    {
      "epoch": 0.7991344601904897,
      "grad_norm": 0.025146683678030968,
      "learning_rate": 0.0012,
      "loss": 2.625818967819214,
      "step": 9856
    },
    {
      "epoch": 0.7992966222156906,
      "grad_norm": 0.03673255443572998,
      "learning_rate": 0.0012,
      "loss": 2.6289310455322266,
      "step": 9858
    },
    {
      "epoch": 0.7994587842408917,
      "grad_norm": 0.037706345319747925,
      "learning_rate": 0.0012,
      "loss": 2.6979997158050537,
      "step": 9860
    },
    {
      "epoch": 0.7996209462660927,
      "grad_norm": 0.02893657051026821,
      "learning_rate": 0.0012,
      "loss": 2.720585346221924,
      "step": 9862
    },
    {
      "epoch": 0.7997831082912936,
      "grad_norm": 0.06593672186136246,
      "learning_rate": 0.0012,
      "loss": 2.674192428588867,
      "step": 9864
    },
    {
      "epoch": 0.7999452703164946,
      "grad_norm": 0.03234926983714104,
      "learning_rate": 0.0012,
      "loss": 2.6715407371520996,
      "step": 9866
    },
    {
      "epoch": 0.8001074323416957,
      "grad_norm": 0.026658551767468452,
      "learning_rate": 0.0012,
      "loss": 2.657161235809326,
      "step": 9868
    },
    {
      "epoch": 0.8002695943668966,
      "grad_norm": 0.02526722103357315,
      "learning_rate": 0.0012,
      "loss": 2.6806111335754395,
      "step": 9870
    },
    {
      "epoch": 0.8004317563920976,
      "grad_norm": 0.04027284309267998,
      "learning_rate": 0.0012,
      "loss": 2.6985509395599365,
      "step": 9872
    },
    {
      "epoch": 0.8005939184172987,
      "grad_norm": 0.028270140290260315,
      "learning_rate": 0.0012,
      "loss": 2.672981023788452,
      "step": 9874
    },
    {
      "epoch": 0.8007560804424996,
      "grad_norm": 0.038027744740247726,
      "learning_rate": 0.0012,
      "loss": 2.696408748626709,
      "step": 9876
    },
    {
      "epoch": 0.8009182424677006,
      "grad_norm": 0.039852336049079895,
      "learning_rate": 0.0012,
      "loss": 2.6698122024536133,
      "step": 9878
    },
    {
      "epoch": 0.8010804044929016,
      "grad_norm": 0.03204211965203285,
      "learning_rate": 0.0012,
      "loss": 2.6544575691223145,
      "step": 9880
    },
    {
      "epoch": 0.8012425665181026,
      "grad_norm": 0.08151978999376297,
      "learning_rate": 0.0012,
      "loss": 2.6438801288604736,
      "step": 9882
    },
    {
      "epoch": 0.8014047285433036,
      "grad_norm": 0.03800063580274582,
      "learning_rate": 0.0012,
      "loss": 2.6804680824279785,
      "step": 9884
    },
    {
      "epoch": 0.8015668905685046,
      "grad_norm": 0.030029086396098137,
      "learning_rate": 0.0012,
      "loss": 2.6315431594848633,
      "step": 9886
    },
    {
      "epoch": 0.8017290525937055,
      "grad_norm": 0.07336274534463882,
      "learning_rate": 0.0012,
      "loss": 2.6883254051208496,
      "step": 9888
    },
    {
      "epoch": 0.8018912146189066,
      "grad_norm": 0.06719404458999634,
      "learning_rate": 0.0012,
      "loss": 2.667876720428467,
      "step": 9890
    },
    {
      "epoch": 0.8020533766441076,
      "grad_norm": 0.02511332556605339,
      "learning_rate": 0.0012,
      "loss": 2.6508870124816895,
      "step": 9892
    },
    {
      "epoch": 0.8022155386693085,
      "grad_norm": 0.04703506827354431,
      "learning_rate": 0.0012,
      "loss": 2.658616781234741,
      "step": 9894
    },
    {
      "epoch": 0.8023777006945095,
      "grad_norm": 0.0323619470000267,
      "learning_rate": 0.0012,
      "loss": 2.6385746002197266,
      "step": 9896
    },
    {
      "epoch": 0.8025398627197106,
      "grad_norm": 0.026990646496415138,
      "learning_rate": 0.0012,
      "loss": 2.666747808456421,
      "step": 9898
    },
    {
      "epoch": 0.8027020247449115,
      "grad_norm": 0.03183416277170181,
      "learning_rate": 0.0012,
      "loss": 2.649266242980957,
      "step": 9900
    },
    {
      "epoch": 0.8028641867701125,
      "grad_norm": 0.035162195563316345,
      "learning_rate": 0.0012,
      "loss": 2.7459840774536133,
      "step": 9902
    },
    {
      "epoch": 0.8030263487953135,
      "grad_norm": 0.027071699500083923,
      "learning_rate": 0.0012,
      "loss": 2.619365692138672,
      "step": 9904
    },
    {
      "epoch": 0.8031885108205145,
      "grad_norm": 0.042829230427742004,
      "learning_rate": 0.0012,
      "loss": 2.7037124633789062,
      "step": 9906
    },
    {
      "epoch": 0.8033506728457155,
      "grad_norm": 0.036438580602407455,
      "learning_rate": 0.0012,
      "loss": 2.686173915863037,
      "step": 9908
    },
    {
      "epoch": 0.8035128348709165,
      "grad_norm": 0.026506932452321053,
      "learning_rate": 0.0012,
      "loss": 2.6102070808410645,
      "step": 9910
    },
    {
      "epoch": 0.8036749968961174,
      "grad_norm": 0.024562669917941093,
      "learning_rate": 0.0012,
      "loss": 2.6570491790771484,
      "step": 9912
    },
    {
      "epoch": 0.8038371589213185,
      "grad_norm": 0.02538737840950489,
      "learning_rate": 0.0012,
      "loss": 2.6337695121765137,
      "step": 9914
    },
    {
      "epoch": 0.8039993209465195,
      "grad_norm": 0.02400907501578331,
      "learning_rate": 0.0012,
      "loss": 2.6239943504333496,
      "step": 9916
    },
    {
      "epoch": 0.8041614829717204,
      "grad_norm": 0.02387930452823639,
      "learning_rate": 0.0012,
      "loss": 2.6376752853393555,
      "step": 9918
    },
    {
      "epoch": 0.8043236449969214,
      "grad_norm": 0.0236417967826128,
      "learning_rate": 0.0012,
      "loss": 2.655355453491211,
      "step": 9920
    },
    {
      "epoch": 0.8044858070221225,
      "grad_norm": 0.029094507917761803,
      "learning_rate": 0.0012,
      "loss": 2.6738033294677734,
      "step": 9922
    },
    {
      "epoch": 0.8046479690473234,
      "grad_norm": 0.02710038423538208,
      "learning_rate": 0.0012,
      "loss": 2.673257350921631,
      "step": 9924
    },
    {
      "epoch": 0.8048101310725244,
      "grad_norm": 0.023159505799412727,
      "learning_rate": 0.0012,
      "loss": 2.610086441040039,
      "step": 9926
    },
    {
      "epoch": 0.8049722930977254,
      "grad_norm": 0.0346599817276001,
      "learning_rate": 0.0012,
      "loss": 2.6836001873016357,
      "step": 9928
    },
    {
      "epoch": 0.8051344551229264,
      "grad_norm": 0.02351435273885727,
      "learning_rate": 0.0012,
      "loss": 2.7325661182403564,
      "step": 9930
    },
    {
      "epoch": 0.8052966171481274,
      "grad_norm": 0.02741461992263794,
      "learning_rate": 0.0012,
      "loss": 2.678865432739258,
      "step": 9932
    },
    {
      "epoch": 0.8054587791733284,
      "grad_norm": 0.03567569702863693,
      "learning_rate": 0.0012,
      "loss": 2.6956911087036133,
      "step": 9934
    },
    {
      "epoch": 0.8056209411985293,
      "grad_norm": 0.0258750319480896,
      "learning_rate": 0.0012,
      "loss": 2.710728406906128,
      "step": 9936
    },
    {
      "epoch": 0.8057831032237304,
      "grad_norm": 0.028896473348140717,
      "learning_rate": 0.0012,
      "loss": 2.6694765090942383,
      "step": 9938
    },
    {
      "epoch": 0.8059452652489314,
      "grad_norm": 0.0259323101490736,
      "learning_rate": 0.0012,
      "loss": 2.5707480907440186,
      "step": 9940
    },
    {
      "epoch": 0.8061074272741323,
      "grad_norm": 0.046680428087711334,
      "learning_rate": 0.0012,
      "loss": 2.7262768745422363,
      "step": 9942
    },
    {
      "epoch": 0.8062695892993333,
      "grad_norm": 0.05565629154443741,
      "learning_rate": 0.0012,
      "loss": 2.652968406677246,
      "step": 9944
    },
    {
      "epoch": 0.8064317513245344,
      "grad_norm": 0.0264577679336071,
      "learning_rate": 0.0012,
      "loss": 2.681441068649292,
      "step": 9946
    },
    {
      "epoch": 0.8065939133497353,
      "grad_norm": 0.029123643413186073,
      "learning_rate": 0.0012,
      "loss": 2.6767592430114746,
      "step": 9948
    },
    {
      "epoch": 0.8067560753749363,
      "grad_norm": 0.028868285939097404,
      "learning_rate": 0.0012,
      "loss": 2.657043933868408,
      "step": 9950
    },
    {
      "epoch": 0.8069182374001374,
      "grad_norm": 0.024538282305002213,
      "learning_rate": 0.0012,
      "loss": 2.6110963821411133,
      "step": 9952
    },
    {
      "epoch": 0.8070803994253383,
      "grad_norm": 0.02745942957699299,
      "learning_rate": 0.0012,
      "loss": 2.659317970275879,
      "step": 9954
    },
    {
      "epoch": 0.8072425614505393,
      "grad_norm": 0.029719777405261993,
      "learning_rate": 0.0012,
      "loss": 2.6140260696411133,
      "step": 9956
    },
    {
      "epoch": 0.8074047234757403,
      "grad_norm": 0.03958693891763687,
      "learning_rate": 0.0012,
      "loss": 2.6829371452331543,
      "step": 9958
    },
    {
      "epoch": 0.8075668855009412,
      "grad_norm": 0.025820527225732803,
      "learning_rate": 0.0012,
      "loss": 2.6868929862976074,
      "step": 9960
    },
    {
      "epoch": 0.8077290475261423,
      "grad_norm": 0.03561558201909065,
      "learning_rate": 0.0012,
      "loss": 2.6666336059570312,
      "step": 9962
    },
    {
      "epoch": 0.8078912095513433,
      "grad_norm": 0.046382393687963486,
      "learning_rate": 0.0012,
      "loss": 2.654203414916992,
      "step": 9964
    },
    {
      "epoch": 0.8080533715765442,
      "grad_norm": 0.09560346603393555,
      "learning_rate": 0.0012,
      "loss": 2.666840076446533,
      "step": 9966
    },
    {
      "epoch": 0.8082155336017453,
      "grad_norm": 0.06160050258040428,
      "learning_rate": 0.0012,
      "loss": 2.643707752227783,
      "step": 9968
    },
    {
      "epoch": 0.8083776956269463,
      "grad_norm": 0.029040995985269547,
      "learning_rate": 0.0012,
      "loss": 2.6353063583374023,
      "step": 9970
    },
    {
      "epoch": 0.8085398576521472,
      "grad_norm": 0.047528188675642014,
      "learning_rate": 0.0012,
      "loss": 2.6970760822296143,
      "step": 9972
    },
    {
      "epoch": 0.8087020196773482,
      "grad_norm": 0.043272681534290314,
      "learning_rate": 0.0012,
      "loss": 2.643852710723877,
      "step": 9974
    },
    {
      "epoch": 0.8088641817025493,
      "grad_norm": 0.02444421499967575,
      "learning_rate": 0.0012,
      "loss": 2.733724594116211,
      "step": 9976
    },
    {
      "epoch": 0.8090263437277502,
      "grad_norm": 0.0390874482691288,
      "learning_rate": 0.0012,
      "loss": 2.6552181243896484,
      "step": 9978
    },
    {
      "epoch": 0.8091885057529512,
      "grad_norm": 0.04132160171866417,
      "learning_rate": 0.0012,
      "loss": 2.662904739379883,
      "step": 9980
    },
    {
      "epoch": 0.8093506677781522,
      "grad_norm": 0.05249716714024544,
      "learning_rate": 0.0012,
      "loss": 2.7257041931152344,
      "step": 9982
    },
    {
      "epoch": 0.8095128298033533,
      "grad_norm": 0.024600252509117126,
      "learning_rate": 0.0012,
      "loss": 2.6369924545288086,
      "step": 9984
    },
    {
      "epoch": 0.8096749918285542,
      "grad_norm": 0.04375624284148216,
      "learning_rate": 0.0012,
      "loss": 2.631556749343872,
      "step": 9986
    },
    {
      "epoch": 0.8098371538537552,
      "grad_norm": 0.042234744876623154,
      "learning_rate": 0.0012,
      "loss": 2.653717517852783,
      "step": 9988
    },
    {
      "epoch": 0.8099993158789562,
      "grad_norm": 0.07071632146835327,
      "learning_rate": 0.0012,
      "loss": 2.7071666717529297,
      "step": 9990
    },
    {
      "epoch": 0.8101614779041572,
      "grad_norm": 0.035430774092674255,
      "learning_rate": 0.0012,
      "loss": 2.610172748565674,
      "step": 9992
    },
    {
      "epoch": 0.8103236399293582,
      "grad_norm": 0.032151222229003906,
      "learning_rate": 0.0012,
      "loss": 2.646671772003174,
      "step": 9994
    },
    {
      "epoch": 0.8104858019545592,
      "grad_norm": 0.032417599111795425,
      "learning_rate": 0.0012,
      "loss": 2.66805362701416,
      "step": 9996
    },
    {
      "epoch": 0.8106479639797601,
      "grad_norm": 0.03615803271532059,
      "learning_rate": 0.0012,
      "loss": 2.6318984031677246,
      "step": 9998
    },
    {
      "epoch": 0.8108101260049612,
      "grad_norm": 0.025426998734474182,
      "learning_rate": 0.0012,
      "loss": 2.6946678161621094,
      "step": 10000
    },
    {
      "epoch": 0.8109722880301622,
      "grad_norm": 0.035054706037044525,
      "learning_rate": 0.0012,
      "loss": 2.6649270057678223,
      "step": 10002
    },
    {
      "epoch": 0.8111344500553631,
      "grad_norm": 0.040434885770082474,
      "learning_rate": 0.0012,
      "loss": 2.7041501998901367,
      "step": 10004
    },
    {
      "epoch": 0.8112966120805641,
      "grad_norm": 0.030108492821455002,
      "learning_rate": 0.0012,
      "loss": 2.6217970848083496,
      "step": 10006
    },
    {
      "epoch": 0.8114587741057652,
      "grad_norm": 0.024342410266399384,
      "learning_rate": 0.0012,
      "loss": 2.6343302726745605,
      "step": 10008
    },
    {
      "epoch": 0.8116209361309661,
      "grad_norm": 0.04552619904279709,
      "learning_rate": 0.0012,
      "loss": 2.6154589653015137,
      "step": 10010
    },
    {
      "epoch": 0.8117830981561671,
      "grad_norm": 0.04426977410912514,
      "learning_rate": 0.0012,
      "loss": 2.641644239425659,
      "step": 10012
    },
    {
      "epoch": 0.8119452601813681,
      "grad_norm": 0.03141864389181137,
      "learning_rate": 0.0012,
      "loss": 2.6369199752807617,
      "step": 10014
    },
    {
      "epoch": 0.8121074222065691,
      "grad_norm": 0.02726357989013195,
      "learning_rate": 0.0012,
      "loss": 2.6281566619873047,
      "step": 10016
    },
    {
      "epoch": 0.8122695842317701,
      "grad_norm": 0.027004674077033997,
      "learning_rate": 0.0012,
      "loss": 2.716111183166504,
      "step": 10018
    },
    {
      "epoch": 0.8124317462569711,
      "grad_norm": 0.029973290860652924,
      "learning_rate": 0.0012,
      "loss": 2.6641969680786133,
      "step": 10020
    },
    {
      "epoch": 0.812593908282172,
      "grad_norm": 0.05974457785487175,
      "learning_rate": 0.0012,
      "loss": 2.655874252319336,
      "step": 10022
    },
    {
      "epoch": 0.8127560703073731,
      "grad_norm": 0.05278881639242172,
      "learning_rate": 0.0012,
      "loss": 2.640195846557617,
      "step": 10024
    },
    {
      "epoch": 0.8129182323325741,
      "grad_norm": 0.024599455296993256,
      "learning_rate": 0.0012,
      "loss": 2.6180734634399414,
      "step": 10026
    },
    {
      "epoch": 0.813080394357775,
      "grad_norm": 0.034229326993227005,
      "learning_rate": 0.0012,
      "loss": 2.593355894088745,
      "step": 10028
    },
    {
      "epoch": 0.813242556382976,
      "grad_norm": 0.0482928566634655,
      "learning_rate": 0.0012,
      "loss": 2.640653133392334,
      "step": 10030
    },
    {
      "epoch": 0.8134047184081771,
      "grad_norm": 0.04154109209775925,
      "learning_rate": 0.0012,
      "loss": 2.6337361335754395,
      "step": 10032
    },
    {
      "epoch": 0.813566880433378,
      "grad_norm": 0.026047205552458763,
      "learning_rate": 0.0012,
      "loss": 2.6673407554626465,
      "step": 10034
    },
    {
      "epoch": 0.813729042458579,
      "grad_norm": 0.028079001232981682,
      "learning_rate": 0.0012,
      "loss": 2.695956230163574,
      "step": 10036
    },
    {
      "epoch": 0.81389120448378,
      "grad_norm": 0.0364665761590004,
      "learning_rate": 0.0012,
      "loss": 2.7164878845214844,
      "step": 10038
    },
    {
      "epoch": 0.814053366508981,
      "grad_norm": 0.023705249652266502,
      "learning_rate": 0.0012,
      "loss": 2.652555227279663,
      "step": 10040
    },
    {
      "epoch": 0.814215528534182,
      "grad_norm": 0.038186460733413696,
      "learning_rate": 0.0012,
      "loss": 2.6748950481414795,
      "step": 10042
    },
    {
      "epoch": 0.814377690559383,
      "grad_norm": 0.02926536835730076,
      "learning_rate": 0.0012,
      "loss": 2.6782407760620117,
      "step": 10044
    },
    {
      "epoch": 0.814539852584584,
      "grad_norm": 0.023815253749489784,
      "learning_rate": 0.0012,
      "loss": 2.649451732635498,
      "step": 10046
    },
    {
      "epoch": 0.814702014609785,
      "grad_norm": 0.025201568379998207,
      "learning_rate": 0.0012,
      "loss": 2.659579277038574,
      "step": 10048
    },
    {
      "epoch": 0.814864176634986,
      "grad_norm": 0.0260101780295372,
      "learning_rate": 0.0012,
      "loss": 2.6762523651123047,
      "step": 10050
    },
    {
      "epoch": 0.8150263386601869,
      "grad_norm": 0.0249590165913105,
      "learning_rate": 0.0012,
      "loss": 2.637972593307495,
      "step": 10052
    },
    {
      "epoch": 0.815188500685388,
      "grad_norm": 0.02899274043738842,
      "learning_rate": 0.0012,
      "loss": 2.667901039123535,
      "step": 10054
    },
    {
      "epoch": 0.815350662710589,
      "grad_norm": 0.03627385199069977,
      "learning_rate": 0.0012,
      "loss": 2.6852948665618896,
      "step": 10056
    },
    {
      "epoch": 0.8155128247357899,
      "grad_norm": 0.030632607638835907,
      "learning_rate": 0.0012,
      "loss": 2.6072373390197754,
      "step": 10058
    },
    {
      "epoch": 0.8156749867609909,
      "grad_norm": 0.034700583666563034,
      "learning_rate": 0.0012,
      "loss": 2.6050589084625244,
      "step": 10060
    },
    {
      "epoch": 0.815837148786192,
      "grad_norm": 0.025989465415477753,
      "learning_rate": 0.0012,
      "loss": 2.6566953659057617,
      "step": 10062
    },
    {
      "epoch": 0.8159993108113929,
      "grad_norm": 0.029145218431949615,
      "learning_rate": 0.0012,
      "loss": 2.7296676635742188,
      "step": 10064
    },
    {
      "epoch": 0.8161614728365939,
      "grad_norm": 0.03149498999118805,
      "learning_rate": 0.0012,
      "loss": 2.6558384895324707,
      "step": 10066
    },
    {
      "epoch": 0.8163236348617949,
      "grad_norm": 0.036014847457408905,
      "learning_rate": 0.0012,
      "loss": 2.6319150924682617,
      "step": 10068
    },
    {
      "epoch": 0.8164857968869959,
      "grad_norm": 0.036283817142248154,
      "learning_rate": 0.0012,
      "loss": 2.6355628967285156,
      "step": 10070
    },
    {
      "epoch": 0.8166479589121969,
      "grad_norm": 0.03296751528978348,
      "learning_rate": 0.0012,
      "loss": 2.690690755844116,
      "step": 10072
    },
    {
      "epoch": 0.8168101209373979,
      "grad_norm": 0.030865652486681938,
      "learning_rate": 0.0012,
      "loss": 2.7182860374450684,
      "step": 10074
    },
    {
      "epoch": 0.8169722829625988,
      "grad_norm": 0.024023139849305153,
      "learning_rate": 0.0012,
      "loss": 2.6255292892456055,
      "step": 10076
    },
    {
      "epoch": 0.8171344449877999,
      "grad_norm": 0.027064936235547066,
      "learning_rate": 0.0012,
      "loss": 2.58335542678833,
      "step": 10078
    },
    {
      "epoch": 0.8172966070130009,
      "grad_norm": 0.026487557217478752,
      "learning_rate": 0.0012,
      "loss": 2.668883800506592,
      "step": 10080
    },
    {
      "epoch": 0.8174587690382018,
      "grad_norm": 0.045703090727329254,
      "learning_rate": 0.0012,
      "loss": 2.640697956085205,
      "step": 10082
    },
    {
      "epoch": 0.8176209310634028,
      "grad_norm": 0.025724386796355247,
      "learning_rate": 0.0012,
      "loss": 2.65118670463562,
      "step": 10084
    },
    {
      "epoch": 0.8177830930886039,
      "grad_norm": 0.039640169590711594,
      "learning_rate": 0.0012,
      "loss": 2.6690354347229004,
      "step": 10086
    },
    {
      "epoch": 0.8179452551138048,
      "grad_norm": 0.03412935137748718,
      "learning_rate": 0.0012,
      "loss": 2.6886210441589355,
      "step": 10088
    },
    {
      "epoch": 0.8181074171390058,
      "grad_norm": 0.025934264063835144,
      "learning_rate": 0.0012,
      "loss": 2.6097238063812256,
      "step": 10090
    },
    {
      "epoch": 0.8182695791642068,
      "grad_norm": 0.049029555171728134,
      "learning_rate": 0.0012,
      "loss": 2.598626136779785,
      "step": 10092
    },
    {
      "epoch": 0.8184317411894078,
      "grad_norm": 0.033034101128578186,
      "learning_rate": 0.0012,
      "loss": 2.710578203201294,
      "step": 10094
    },
    {
      "epoch": 0.8185939032146088,
      "grad_norm": 0.02664872445166111,
      "learning_rate": 0.0012,
      "loss": 2.686721086502075,
      "step": 10096
    },
    {
      "epoch": 0.8187560652398098,
      "grad_norm": 0.03170020505785942,
      "learning_rate": 0.0012,
      "loss": 2.7281503677368164,
      "step": 10098
    },
    {
      "epoch": 0.8189182272650107,
      "grad_norm": 0.02643272466957569,
      "learning_rate": 0.0012,
      "loss": 2.666314125061035,
      "step": 10100
    },
    {
      "epoch": 0.8190803892902118,
      "grad_norm": 0.02549782022833824,
      "learning_rate": 0.0012,
      "loss": 2.6560802459716797,
      "step": 10102
    },
    {
      "epoch": 0.8192425513154128,
      "grad_norm": 0.0470380038022995,
      "learning_rate": 0.0012,
      "loss": 2.710864782333374,
      "step": 10104
    },
    {
      "epoch": 0.8194047133406137,
      "grad_norm": 0.043533917516469955,
      "learning_rate": 0.0012,
      "loss": 2.6070826053619385,
      "step": 10106
    },
    {
      "epoch": 0.8195668753658147,
      "grad_norm": 0.03001108020544052,
      "learning_rate": 0.0012,
      "loss": 2.656031608581543,
      "step": 10108
    },
    {
      "epoch": 0.8197290373910158,
      "grad_norm": 0.028773730620741844,
      "learning_rate": 0.0012,
      "loss": 2.6452717781066895,
      "step": 10110
    },
    {
      "epoch": 0.8198911994162167,
      "grad_norm": 0.027619315311312675,
      "learning_rate": 0.0012,
      "loss": 2.673755168914795,
      "step": 10112
    },
    {
      "epoch": 0.8200533614414177,
      "grad_norm": 0.03845413029193878,
      "learning_rate": 0.0012,
      "loss": 2.6728925704956055,
      "step": 10114
    },
    {
      "epoch": 0.8202155234666187,
      "grad_norm": 0.04609404504299164,
      "learning_rate": 0.0012,
      "loss": 2.635821580886841,
      "step": 10116
    },
    {
      "epoch": 0.8203776854918197,
      "grad_norm": 0.05644385144114494,
      "learning_rate": 0.0012,
      "loss": 2.6674928665161133,
      "step": 10118
    },
    {
      "epoch": 0.8205398475170207,
      "grad_norm": 0.042385075241327286,
      "learning_rate": 0.0012,
      "loss": 2.5848846435546875,
      "step": 10120
    },
    {
      "epoch": 0.8207020095422217,
      "grad_norm": 0.026519933715462685,
      "learning_rate": 0.0012,
      "loss": 2.623527765274048,
      "step": 10122
    },
    {
      "epoch": 0.8208641715674226,
      "grad_norm": 0.0487542450428009,
      "learning_rate": 0.0012,
      "loss": 2.6243085861206055,
      "step": 10124
    },
    {
      "epoch": 0.8210263335926237,
      "grad_norm": 0.05431690812110901,
      "learning_rate": 0.0012,
      "loss": 2.6326916217803955,
      "step": 10126
    },
    {
      "epoch": 0.8211884956178247,
      "grad_norm": 0.026909328997135162,
      "learning_rate": 0.0012,
      "loss": 2.617884635925293,
      "step": 10128
    },
    {
      "epoch": 0.8213506576430256,
      "grad_norm": 0.04167446494102478,
      "learning_rate": 0.0012,
      "loss": 2.5959558486938477,
      "step": 10130
    },
    {
      "epoch": 0.8215128196682266,
      "grad_norm": 0.029127879068255424,
      "learning_rate": 0.0012,
      "loss": 2.697486639022827,
      "step": 10132
    },
    {
      "epoch": 0.8216749816934277,
      "grad_norm": 0.0481230802834034,
      "learning_rate": 0.0012,
      "loss": 2.596853494644165,
      "step": 10134
    },
    {
      "epoch": 0.8218371437186286,
      "grad_norm": 0.04381050914525986,
      "learning_rate": 0.0012,
      "loss": 2.6662325859069824,
      "step": 10136
    },
    {
      "epoch": 0.8219993057438296,
      "grad_norm": 0.037050507962703705,
      "learning_rate": 0.0012,
      "loss": 2.6340689659118652,
      "step": 10138
    },
    {
      "epoch": 0.8221614677690307,
      "grad_norm": 0.02955588325858116,
      "learning_rate": 0.0012,
      "loss": 2.679093837738037,
      "step": 10140
    },
    {
      "epoch": 0.8223236297942316,
      "grad_norm": 0.056712739169597626,
      "learning_rate": 0.0012,
      "loss": 2.676389217376709,
      "step": 10142
    },
    {
      "epoch": 0.8224857918194326,
      "grad_norm": 0.045335955917835236,
      "learning_rate": 0.0012,
      "loss": 2.6605052947998047,
      "step": 10144
    },
    {
      "epoch": 0.8226479538446336,
      "grad_norm": 0.02721523307263851,
      "learning_rate": 0.0012,
      "loss": 2.637845039367676,
      "step": 10146
    },
    {
      "epoch": 0.8228101158698345,
      "grad_norm": 0.04601876437664032,
      "learning_rate": 0.0012,
      "loss": 2.685452938079834,
      "step": 10148
    },
    {
      "epoch": 0.8229722778950356,
      "grad_norm": 0.04630709066987038,
      "learning_rate": 0.0012,
      "loss": 2.6881752014160156,
      "step": 10150
    },
    {
      "epoch": 0.8231344399202366,
      "grad_norm": 0.027651546522974968,
      "learning_rate": 0.0012,
      "loss": 2.646664619445801,
      "step": 10152
    },
    {
      "epoch": 0.8232966019454375,
      "grad_norm": 0.024217696860432625,
      "learning_rate": 0.0012,
      "loss": 2.7320477962493896,
      "step": 10154
    },
    {
      "epoch": 0.8234587639706386,
      "grad_norm": 0.02999688684940338,
      "learning_rate": 0.0012,
      "loss": 2.6584692001342773,
      "step": 10156
    },
    {
      "epoch": 0.8236209259958396,
      "grad_norm": 0.03771334886550903,
      "learning_rate": 0.0012,
      "loss": 2.6742801666259766,
      "step": 10158
    },
    {
      "epoch": 0.8237830880210405,
      "grad_norm": 0.03929399326443672,
      "learning_rate": 0.0012,
      "loss": 2.646106719970703,
      "step": 10160
    },
    {
      "epoch": 0.8239452500462415,
      "grad_norm": 0.024347268044948578,
      "learning_rate": 0.0012,
      "loss": 2.7007904052734375,
      "step": 10162
    },
    {
      "epoch": 0.8241074120714426,
      "grad_norm": 0.043139103800058365,
      "learning_rate": 0.0012,
      "loss": 2.632714033126831,
      "step": 10164
    },
    {
      "epoch": 0.8242695740966435,
      "grad_norm": 0.03234386816620827,
      "learning_rate": 0.0012,
      "loss": 2.6809792518615723,
      "step": 10166
    },
    {
      "epoch": 0.8244317361218445,
      "grad_norm": 0.02677248977124691,
      "learning_rate": 0.0012,
      "loss": 2.6069846153259277,
      "step": 10168
    },
    {
      "epoch": 0.8245938981470455,
      "grad_norm": 0.032570719718933105,
      "learning_rate": 0.0012,
      "loss": 2.697103500366211,
      "step": 10170
    },
    {
      "epoch": 0.8247560601722465,
      "grad_norm": 0.03983451798558235,
      "learning_rate": 0.0012,
      "loss": 2.6522350311279297,
      "step": 10172
    },
    {
      "epoch": 0.8249182221974475,
      "grad_norm": 0.03768815100193024,
      "learning_rate": 0.0012,
      "loss": 2.5426406860351562,
      "step": 10174
    },
    {
      "epoch": 0.8250803842226485,
      "grad_norm": 0.026055049151182175,
      "learning_rate": 0.0012,
      "loss": 2.654313564300537,
      "step": 10176
    },
    {
      "epoch": 0.8252425462478494,
      "grad_norm": 0.03249823674559593,
      "learning_rate": 0.0012,
      "loss": 2.6668310165405273,
      "step": 10178
    },
    {
      "epoch": 0.8254047082730505,
      "grad_norm": 0.028891516849398613,
      "learning_rate": 0.0012,
      "loss": 2.6575710773468018,
      "step": 10180
    },
    {
      "epoch": 0.8255668702982515,
      "grad_norm": 0.02668793685734272,
      "learning_rate": 0.0012,
      "loss": 2.5788512229919434,
      "step": 10182
    },
    {
      "epoch": 0.8257290323234524,
      "grad_norm": 0.02471478469669819,
      "learning_rate": 0.0012,
      "loss": 2.6294662952423096,
      "step": 10184
    },
    {
      "epoch": 0.8258911943486534,
      "grad_norm": 0.03228875994682312,
      "learning_rate": 0.0012,
      "loss": 2.6601402759552,
      "step": 10186
    },
    {
      "epoch": 0.8260533563738545,
      "grad_norm": 0.033578116446733475,
      "learning_rate": 0.0012,
      "loss": 2.6672353744506836,
      "step": 10188
    },
    {
      "epoch": 0.8262155183990554,
      "grad_norm": 0.02762840874493122,
      "learning_rate": 0.0012,
      "loss": 2.677530527114868,
      "step": 10190
    },
    {
      "epoch": 0.8263776804242564,
      "grad_norm": 0.03947677090764046,
      "learning_rate": 0.0012,
      "loss": 2.645936965942383,
      "step": 10192
    },
    {
      "epoch": 0.8265398424494574,
      "grad_norm": 0.02597847580909729,
      "learning_rate": 0.0012,
      "loss": 2.7003066539764404,
      "step": 10194
    },
    {
      "epoch": 0.8267020044746584,
      "grad_norm": 0.02313290722668171,
      "learning_rate": 0.0012,
      "loss": 2.754030704498291,
      "step": 10196
    },
    {
      "epoch": 0.8268641664998594,
      "grad_norm": 0.03001842088997364,
      "learning_rate": 0.0012,
      "loss": 2.670717716217041,
      "step": 10198
    },
    {
      "epoch": 0.8270263285250604,
      "grad_norm": 0.024660443887114525,
      "learning_rate": 0.0012,
      "loss": 2.6523196697235107,
      "step": 10200
    },
    {
      "epoch": 0.8271884905502613,
      "grad_norm": 0.024302829056978226,
      "learning_rate": 0.0012,
      "loss": 2.6407973766326904,
      "step": 10202
    },
    {
      "epoch": 0.8273506525754624,
      "grad_norm": 0.030372967943549156,
      "learning_rate": 0.0012,
      "loss": 2.6211159229278564,
      "step": 10204
    },
    {
      "epoch": 0.8275128146006634,
      "grad_norm": 0.04565395414829254,
      "learning_rate": 0.0012,
      "loss": 2.6233129501342773,
      "step": 10206
    },
    {
      "epoch": 0.8276749766258643,
      "grad_norm": 0.06873153895139694,
      "learning_rate": 0.0012,
      "loss": 2.732581615447998,
      "step": 10208
    },
    {
      "epoch": 0.8278371386510653,
      "grad_norm": 0.07026107609272003,
      "learning_rate": 0.0012,
      "loss": 2.6303133964538574,
      "step": 10210
    },
    {
      "epoch": 0.8279993006762664,
      "grad_norm": 0.06426500529050827,
      "learning_rate": 0.0012,
      "loss": 2.644963026046753,
      "step": 10212
    },
    {
      "epoch": 0.8281614627014673,
      "grad_norm": 0.028461363166570663,
      "learning_rate": 0.0012,
      "loss": 2.640631675720215,
      "step": 10214
    },
    {
      "epoch": 0.8283236247266683,
      "grad_norm": 0.04755562171339989,
      "learning_rate": 0.0012,
      "loss": 2.737452507019043,
      "step": 10216
    },
    {
      "epoch": 0.8284857867518693,
      "grad_norm": 0.04833776503801346,
      "learning_rate": 0.0012,
      "loss": 2.650080680847168,
      "step": 10218
    },
    {
      "epoch": 0.8286479487770703,
      "grad_norm": 0.03140414506196976,
      "learning_rate": 0.0012,
      "loss": 2.6150262355804443,
      "step": 10220
    },
    {
      "epoch": 0.8288101108022713,
      "grad_norm": 0.02989353984594345,
      "learning_rate": 0.0012,
      "loss": 2.678737163543701,
      "step": 10222
    },
    {
      "epoch": 0.8289722728274723,
      "grad_norm": 0.027164651080965996,
      "learning_rate": 0.0012,
      "loss": 2.7635669708251953,
      "step": 10224
    },
    {
      "epoch": 0.8291344348526732,
      "grad_norm": 0.05747466906905174,
      "learning_rate": 0.0012,
      "loss": 2.680521011352539,
      "step": 10226
    },
    {
      "epoch": 0.8292965968778743,
      "grad_norm": 0.032937824726104736,
      "learning_rate": 0.0012,
      "loss": 2.6220040321350098,
      "step": 10228
    },
    {
      "epoch": 0.8294587589030753,
      "grad_norm": 0.037693411111831665,
      "learning_rate": 0.0012,
      "loss": 2.67832088470459,
      "step": 10230
    },
    {
      "epoch": 0.8296209209282762,
      "grad_norm": 0.025030165910720825,
      "learning_rate": 0.0012,
      "loss": 2.643677234649658,
      "step": 10232
    },
    {
      "epoch": 0.8297830829534772,
      "grad_norm": 0.03582018241286278,
      "learning_rate": 0.0012,
      "loss": 2.6387569904327393,
      "step": 10234
    },
    {
      "epoch": 0.8299452449786783,
      "grad_norm": 0.036661092191934586,
      "learning_rate": 0.0012,
      "loss": 2.7066268920898438,
      "step": 10236
    },
    {
      "epoch": 0.8301074070038792,
      "grad_norm": 0.026590008288621902,
      "learning_rate": 0.0012,
      "loss": 2.7037389278411865,
      "step": 10238
    },
    {
      "epoch": 0.8302695690290802,
      "grad_norm": 0.04166460037231445,
      "learning_rate": 0.0012,
      "loss": 2.617814302444458,
      "step": 10240
    },
    {
      "epoch": 0.8304317310542813,
      "grad_norm": 0.0603296272456646,
      "learning_rate": 0.0012,
      "loss": 2.709341526031494,
      "step": 10242
    },
    {
      "epoch": 0.8305938930794822,
      "grad_norm": 0.025932222604751587,
      "learning_rate": 0.0012,
      "loss": 2.6943657398223877,
      "step": 10244
    },
    {
      "epoch": 0.8307560551046832,
      "grad_norm": 0.02892306074500084,
      "learning_rate": 0.0012,
      "loss": 2.7133102416992188,
      "step": 10246
    },
    {
      "epoch": 0.8309182171298842,
      "grad_norm": 0.029195699840784073,
      "learning_rate": 0.0012,
      "loss": 2.6883347034454346,
      "step": 10248
    },
    {
      "epoch": 0.8310803791550851,
      "grad_norm": 0.02768135815858841,
      "learning_rate": 0.0012,
      "loss": 2.654933452606201,
      "step": 10250
    },
    {
      "epoch": 0.8312425411802862,
      "grad_norm": 0.033509448170661926,
      "learning_rate": 0.0012,
      "loss": 2.6422886848449707,
      "step": 10252
    },
    {
      "epoch": 0.8314047032054872,
      "grad_norm": 0.03974926099181175,
      "learning_rate": 0.0012,
      "loss": 2.6819236278533936,
      "step": 10254
    },
    {
      "epoch": 0.8315668652306881,
      "grad_norm": 0.027782795950770378,
      "learning_rate": 0.0012,
      "loss": 2.688114881515503,
      "step": 10256
    },
    {
      "epoch": 0.8317290272558892,
      "grad_norm": 0.030895721167325974,
      "learning_rate": 0.0012,
      "loss": 2.6782121658325195,
      "step": 10258
    },
    {
      "epoch": 0.8318911892810902,
      "grad_norm": 0.04302801936864853,
      "learning_rate": 0.0012,
      "loss": 2.655215263366699,
      "step": 10260
    },
    {
      "epoch": 0.8320533513062911,
      "grad_norm": 0.05268239602446556,
      "learning_rate": 0.0012,
      "loss": 2.671252727508545,
      "step": 10262
    },
    {
      "epoch": 0.8322155133314921,
      "grad_norm": 0.0294251199811697,
      "learning_rate": 0.0012,
      "loss": 2.645659923553467,
      "step": 10264
    },
    {
      "epoch": 0.8323776753566932,
      "grad_norm": 0.04022127017378807,
      "learning_rate": 0.0012,
      "loss": 2.7224435806274414,
      "step": 10266
    },
    {
      "epoch": 0.8325398373818941,
      "grad_norm": 0.03991300240159035,
      "learning_rate": 0.0012,
      "loss": 2.682368755340576,
      "step": 10268
    },
    {
      "epoch": 0.8327019994070951,
      "grad_norm": 0.030688602477312088,
      "learning_rate": 0.0012,
      "loss": 2.6652183532714844,
      "step": 10270
    },
    {
      "epoch": 0.8328641614322961,
      "grad_norm": 0.026116790249943733,
      "learning_rate": 0.0012,
      "loss": 2.650418281555176,
      "step": 10272
    },
    {
      "epoch": 0.833026323457497,
      "grad_norm": 0.028283238410949707,
      "learning_rate": 0.0012,
      "loss": 2.6537933349609375,
      "step": 10274
    },
    {
      "epoch": 0.8331884854826981,
      "grad_norm": 0.034767281264066696,
      "learning_rate": 0.0012,
      "loss": 2.655210018157959,
      "step": 10276
    },
    {
      "epoch": 0.8333506475078991,
      "grad_norm": 0.02833813987672329,
      "learning_rate": 0.0012,
      "loss": 2.710350513458252,
      "step": 10278
    },
    {
      "epoch": 0.8335128095331,
      "grad_norm": 0.02826671116054058,
      "learning_rate": 0.0012,
      "loss": 2.6358704566955566,
      "step": 10280
    },
    {
      "epoch": 0.8336749715583011,
      "grad_norm": 0.022602520883083344,
      "learning_rate": 0.0012,
      "loss": 2.655632495880127,
      "step": 10282
    },
    {
      "epoch": 0.8338371335835021,
      "grad_norm": 0.03663930296897888,
      "learning_rate": 0.0012,
      "loss": 2.6601648330688477,
      "step": 10284
    },
    {
      "epoch": 0.833999295608703,
      "grad_norm": 0.031306374818086624,
      "learning_rate": 0.0012,
      "loss": 2.659099578857422,
      "step": 10286
    },
    {
      "epoch": 0.834161457633904,
      "grad_norm": 0.029680700972676277,
      "learning_rate": 0.0012,
      "loss": 2.618990898132324,
      "step": 10288
    },
    {
      "epoch": 0.8343236196591051,
      "grad_norm": 0.040422867983579636,
      "learning_rate": 0.0012,
      "loss": 2.674734592437744,
      "step": 10290
    },
    {
      "epoch": 0.834485781684306,
      "grad_norm": 0.05585333704948425,
      "learning_rate": 0.0012,
      "loss": 2.6465954780578613,
      "step": 10292
    },
    {
      "epoch": 0.834647943709507,
      "grad_norm": 0.056875310838222504,
      "learning_rate": 0.0012,
      "loss": 2.6460256576538086,
      "step": 10294
    },
    {
      "epoch": 0.834810105734708,
      "grad_norm": 0.05592784285545349,
      "learning_rate": 0.0012,
      "loss": 2.6356780529022217,
      "step": 10296
    },
    {
      "epoch": 0.834972267759909,
      "grad_norm": 0.04532213881611824,
      "learning_rate": 0.0012,
      "loss": 2.668602466583252,
      "step": 10298
    },
    {
      "epoch": 0.83513442978511,
      "grad_norm": 0.027850085869431496,
      "learning_rate": 0.0012,
      "loss": 2.6650543212890625,
      "step": 10300
    },
    {
      "epoch": 0.835296591810311,
      "grad_norm": 0.0656648576259613,
      "learning_rate": 0.0012,
      "loss": 2.6443121433258057,
      "step": 10302
    },
    {
      "epoch": 0.8354587538355119,
      "grad_norm": 0.06746945530176163,
      "learning_rate": 0.0012,
      "loss": 2.704200267791748,
      "step": 10304
    },
    {
      "epoch": 0.835620915860713,
      "grad_norm": 0.0281851626932621,
      "learning_rate": 0.0012,
      "loss": 2.6731715202331543,
      "step": 10306
    },
    {
      "epoch": 0.835783077885914,
      "grad_norm": 0.02661210298538208,
      "learning_rate": 0.0012,
      "loss": 2.6501364707946777,
      "step": 10308
    },
    {
      "epoch": 0.8359452399111149,
      "grad_norm": 0.030188770964741707,
      "learning_rate": 0.0012,
      "loss": 2.6864638328552246,
      "step": 10310
    },
    {
      "epoch": 0.8361074019363159,
      "grad_norm": 0.028417350724339485,
      "learning_rate": 0.0012,
      "loss": 2.7295751571655273,
      "step": 10312
    },
    {
      "epoch": 0.836269563961517,
      "grad_norm": 0.02608252689242363,
      "learning_rate": 0.0012,
      "loss": 2.6515092849731445,
      "step": 10314
    },
    {
      "epoch": 0.8364317259867179,
      "grad_norm": 0.03596780449151993,
      "learning_rate": 0.0012,
      "loss": 2.579259157180786,
      "step": 10316
    },
    {
      "epoch": 0.8365938880119189,
      "grad_norm": 0.04605334997177124,
      "learning_rate": 0.0012,
      "loss": 2.600468635559082,
      "step": 10318
    },
    {
      "epoch": 0.83675605003712,
      "grad_norm": 0.05520500987768173,
      "learning_rate": 0.0012,
      "loss": 2.680431365966797,
      "step": 10320
    },
    {
      "epoch": 0.8369182120623209,
      "grad_norm": 0.042418669909238815,
      "learning_rate": 0.0012,
      "loss": 2.569841146469116,
      "step": 10322
    },
    {
      "epoch": 0.8370803740875219,
      "grad_norm": 0.03376374393701553,
      "learning_rate": 0.0012,
      "loss": 2.704244613647461,
      "step": 10324
    },
    {
      "epoch": 0.8372425361127229,
      "grad_norm": 0.03546707332134247,
      "learning_rate": 0.0012,
      "loss": 2.6733009815216064,
      "step": 10326
    },
    {
      "epoch": 0.8374046981379238,
      "grad_norm": 0.03430262953042984,
      "learning_rate": 0.0012,
      "loss": 2.619940757751465,
      "step": 10328
    },
    {
      "epoch": 0.8375668601631249,
      "grad_norm": 0.061129726469516754,
      "learning_rate": 0.0012,
      "loss": 2.683228015899658,
      "step": 10330
    },
    {
      "epoch": 0.8377290221883259,
      "grad_norm": 0.052510302513837814,
      "learning_rate": 0.0012,
      "loss": 2.721895694732666,
      "step": 10332
    },
    {
      "epoch": 0.8378911842135268,
      "grad_norm": 0.028620276600122452,
      "learning_rate": 0.0012,
      "loss": 2.6872053146362305,
      "step": 10334
    },
    {
      "epoch": 0.8380533462387278,
      "grad_norm": 0.028249302878975868,
      "learning_rate": 0.0012,
      "loss": 2.6246800422668457,
      "step": 10336
    },
    {
      "epoch": 0.8382155082639289,
      "grad_norm": 0.04856298863887787,
      "learning_rate": 0.0012,
      "loss": 2.698603868484497,
      "step": 10338
    },
    {
      "epoch": 0.8383776702891298,
      "grad_norm": 0.048622261732816696,
      "learning_rate": 0.0012,
      "loss": 2.6389551162719727,
      "step": 10340
    },
    {
      "epoch": 0.8385398323143308,
      "grad_norm": 0.04115966334939003,
      "learning_rate": 0.0012,
      "loss": 2.696150302886963,
      "step": 10342
    },
    {
      "epoch": 0.8387019943395319,
      "grad_norm": 0.022975048050284386,
      "learning_rate": 0.0012,
      "loss": 2.7027034759521484,
      "step": 10344
    },
    {
      "epoch": 0.8388641563647328,
      "grad_norm": 0.03643566370010376,
      "learning_rate": 0.0012,
      "loss": 2.622976303100586,
      "step": 10346
    },
    {
      "epoch": 0.8390263183899338,
      "grad_norm": 0.035669080913066864,
      "learning_rate": 0.0012,
      "loss": 2.642963409423828,
      "step": 10348
    },
    {
      "epoch": 0.8391884804151348,
      "grad_norm": 0.025413095951080322,
      "learning_rate": 0.0012,
      "loss": 2.612835168838501,
      "step": 10350
    },
    {
      "epoch": 0.8393506424403357,
      "grad_norm": 0.035528551787137985,
      "learning_rate": 0.0012,
      "loss": 2.6442294120788574,
      "step": 10352
    },
    {
      "epoch": 0.8395128044655368,
      "grad_norm": 0.02581573650240898,
      "learning_rate": 0.0012,
      "loss": 2.6757946014404297,
      "step": 10354
    },
    {
      "epoch": 0.8396749664907378,
      "grad_norm": 0.03354598209261894,
      "learning_rate": 0.0012,
      "loss": 2.7625699043273926,
      "step": 10356
    },
    {
      "epoch": 0.8398371285159387,
      "grad_norm": 0.03808089718222618,
      "learning_rate": 0.0012,
      "loss": 2.6798911094665527,
      "step": 10358
    },
    {
      "epoch": 0.8399992905411398,
      "grad_norm": 0.04401814565062523,
      "learning_rate": 0.0012,
      "loss": 2.6869006156921387,
      "step": 10360
    },
    {
      "epoch": 0.8401614525663408,
      "grad_norm": 0.0374382883310318,
      "learning_rate": 0.0012,
      "loss": 2.692732095718384,
      "step": 10362
    },
    {
      "epoch": 0.8403236145915417,
      "grad_norm": 0.026977185159921646,
      "learning_rate": 0.0012,
      "loss": 2.6611671447753906,
      "step": 10364
    },
    {
      "epoch": 0.8404857766167427,
      "grad_norm": 0.049570482224226,
      "learning_rate": 0.0012,
      "loss": 2.5951578617095947,
      "step": 10366
    },
    {
      "epoch": 0.8406479386419438,
      "grad_norm": 0.07423996925354004,
      "learning_rate": 0.0012,
      "loss": 2.6437363624572754,
      "step": 10368
    },
    {
      "epoch": 0.8408101006671447,
      "grad_norm": 0.06878964602947235,
      "learning_rate": 0.0012,
      "loss": 2.706339120864868,
      "step": 10370
    },
    {
      "epoch": 0.8409722626923457,
      "grad_norm": 0.027071474120020866,
      "learning_rate": 0.0012,
      "loss": 2.640190601348877,
      "step": 10372
    },
    {
      "epoch": 0.8411344247175467,
      "grad_norm": 0.06602447479963303,
      "learning_rate": 0.0012,
      "loss": 2.6480941772460938,
      "step": 10374
    },
    {
      "epoch": 0.8412965867427477,
      "grad_norm": 0.06827229261398315,
      "learning_rate": 0.0012,
      "loss": 2.628685235977173,
      "step": 10376
    },
    {
      "epoch": 0.8414587487679487,
      "grad_norm": 0.025927986949682236,
      "learning_rate": 0.0012,
      "loss": 2.7090911865234375,
      "step": 10378
    },
    {
      "epoch": 0.8416209107931497,
      "grad_norm": 0.0832291916012764,
      "learning_rate": 0.0012,
      "loss": 2.684054374694824,
      "step": 10380
    },
    {
      "epoch": 0.8417830728183506,
      "grad_norm": 0.04414386674761772,
      "learning_rate": 0.0012,
      "loss": 2.6234793663024902,
      "step": 10382
    },
    {
      "epoch": 0.8419452348435517,
      "grad_norm": 0.025156881660223007,
      "learning_rate": 0.0012,
      "loss": 2.678252696990967,
      "step": 10384
    },
    {
      "epoch": 0.8421073968687527,
      "grad_norm": 0.0589231513440609,
      "learning_rate": 0.0012,
      "loss": 2.6728713512420654,
      "step": 10386
    },
    {
      "epoch": 0.8422695588939536,
      "grad_norm": 0.03963610902428627,
      "learning_rate": 0.0012,
      "loss": 2.677562713623047,
      "step": 10388
    },
    {
      "epoch": 0.8424317209191546,
      "grad_norm": 0.02664426527917385,
      "learning_rate": 0.0012,
      "loss": 2.6942625045776367,
      "step": 10390
    },
    {
      "epoch": 0.8425938829443557,
      "grad_norm": 0.04634057730436325,
      "learning_rate": 0.0012,
      "loss": 2.5979971885681152,
      "step": 10392
    },
    {
      "epoch": 0.8427560449695566,
      "grad_norm": 0.0448312945663929,
      "learning_rate": 0.0012,
      "loss": 2.6193394660949707,
      "step": 10394
    },
    {
      "epoch": 0.8429182069947576,
      "grad_norm": 0.02564338967204094,
      "learning_rate": 0.0012,
      "loss": 2.614272117614746,
      "step": 10396
    },
    {
      "epoch": 0.8430803690199586,
      "grad_norm": 0.05405130609869957,
      "learning_rate": 0.0012,
      "loss": 2.669445037841797,
      "step": 10398
    },
    {
      "epoch": 0.8432425310451596,
      "grad_norm": 0.035884566605091095,
      "learning_rate": 0.0012,
      "loss": 2.6948294639587402,
      "step": 10400
    },
    {
      "epoch": 0.8434046930703606,
      "grad_norm": 0.04953140765428543,
      "learning_rate": 0.0012,
      "loss": 2.644317150115967,
      "step": 10402
    },
    {
      "epoch": 0.8435668550955616,
      "grad_norm": 0.03602002561092377,
      "learning_rate": 0.0012,
      "loss": 2.7126717567443848,
      "step": 10404
    },
    {
      "epoch": 0.8437290171207625,
      "grad_norm": 0.03042621724307537,
      "learning_rate": 0.0012,
      "loss": 2.706373691558838,
      "step": 10406
    },
    {
      "epoch": 0.8438911791459636,
      "grad_norm": 0.025377774611115456,
      "learning_rate": 0.0012,
      "loss": 2.719241142272949,
      "step": 10408
    },
    {
      "epoch": 0.8440533411711646,
      "grad_norm": 0.04529775306582451,
      "learning_rate": 0.0012,
      "loss": 2.6519827842712402,
      "step": 10410
    },
    {
      "epoch": 0.8442155031963655,
      "grad_norm": 0.03500954061746597,
      "learning_rate": 0.0012,
      "loss": 2.6825273036956787,
      "step": 10412
    },
    {
      "epoch": 0.8443776652215665,
      "grad_norm": 0.02489713951945305,
      "learning_rate": 0.0012,
      "loss": 2.6746034622192383,
      "step": 10414
    },
    {
      "epoch": 0.8445398272467676,
      "grad_norm": 0.026429645717144012,
      "learning_rate": 0.0012,
      "loss": 2.6778368949890137,
      "step": 10416
    },
    {
      "epoch": 0.8447019892719685,
      "grad_norm": 0.031119007617235184,
      "learning_rate": 0.0012,
      "loss": 2.7026150226593018,
      "step": 10418
    },
    {
      "epoch": 0.8448641512971695,
      "grad_norm": 0.029550544917583466,
      "learning_rate": 0.0012,
      "loss": 2.702481746673584,
      "step": 10420
    },
    {
      "epoch": 0.8450263133223705,
      "grad_norm": 0.027964407578110695,
      "learning_rate": 0.0012,
      "loss": 2.684662103652954,
      "step": 10422
    },
    {
      "epoch": 0.8451884753475715,
      "grad_norm": 0.029505429789423943,
      "learning_rate": 0.0012,
      "loss": 2.6704509258270264,
      "step": 10424
    },
    {
      "epoch": 0.8453506373727725,
      "grad_norm": 0.035790059715509415,
      "learning_rate": 0.0012,
      "loss": 2.6637630462646484,
      "step": 10426
    },
    {
      "epoch": 0.8455127993979735,
      "grad_norm": 0.02393343858420849,
      "learning_rate": 0.0012,
      "loss": 2.7002792358398438,
      "step": 10428
    },
    {
      "epoch": 0.8456749614231744,
      "grad_norm": 0.05082600936293602,
      "learning_rate": 0.0012,
      "loss": 2.6301589012145996,
      "step": 10430
    },
    {
      "epoch": 0.8458371234483755,
      "grad_norm": 0.06513181328773499,
      "learning_rate": 0.0012,
      "loss": 2.650573253631592,
      "step": 10432
    },
    {
      "epoch": 0.8459992854735765,
      "grad_norm": 0.041201140731573105,
      "learning_rate": 0.0012,
      "loss": 2.6856274604797363,
      "step": 10434
    },
    {
      "epoch": 0.8461614474987774,
      "grad_norm": 0.05089627951383591,
      "learning_rate": 0.0012,
      "loss": 2.694983959197998,
      "step": 10436
    },
    {
      "epoch": 0.8463236095239784,
      "grad_norm": 0.0448421947658062,
      "learning_rate": 0.0012,
      "loss": 2.6672892570495605,
      "step": 10438
    },
    {
      "epoch": 0.8464857715491795,
      "grad_norm": 0.03237171471118927,
      "learning_rate": 0.0012,
      "loss": 2.637373924255371,
      "step": 10440
    },
    {
      "epoch": 0.8466479335743804,
      "grad_norm": 0.03276897221803665,
      "learning_rate": 0.0012,
      "loss": 2.631633758544922,
      "step": 10442
    },
    {
      "epoch": 0.8468100955995814,
      "grad_norm": 0.04324664548039436,
      "learning_rate": 0.0012,
      "loss": 2.669929027557373,
      "step": 10444
    },
    {
      "epoch": 0.8469722576247825,
      "grad_norm": 0.0313311405479908,
      "learning_rate": 0.0012,
      "loss": 2.6954240798950195,
      "step": 10446
    },
    {
      "epoch": 0.8471344196499834,
      "grad_norm": 0.026803823187947273,
      "learning_rate": 0.0012,
      "loss": 2.7143378257751465,
      "step": 10448
    },
    {
      "epoch": 0.8472965816751844,
      "grad_norm": 0.026799796149134636,
      "learning_rate": 0.0012,
      "loss": 2.7089266777038574,
      "step": 10450
    },
    {
      "epoch": 0.8474587437003854,
      "grad_norm": 0.024671562016010284,
      "learning_rate": 0.0012,
      "loss": 2.660938262939453,
      "step": 10452
    },
    {
      "epoch": 0.8476209057255863,
      "grad_norm": 0.024123841896653175,
      "learning_rate": 0.0012,
      "loss": 2.6870436668395996,
      "step": 10454
    },
    {
      "epoch": 0.8477830677507874,
      "grad_norm": 0.04569348692893982,
      "learning_rate": 0.0012,
      "loss": 2.6781272888183594,
      "step": 10456
    },
    {
      "epoch": 0.8479452297759884,
      "grad_norm": 0.04769454523921013,
      "learning_rate": 0.0012,
      "loss": 2.658362865447998,
      "step": 10458
    },
    {
      "epoch": 0.8481073918011893,
      "grad_norm": 0.02669248729944229,
      "learning_rate": 0.0012,
      "loss": 2.725754976272583,
      "step": 10460
    },
    {
      "epoch": 0.8482695538263904,
      "grad_norm": 0.032914016395807266,
      "learning_rate": 0.0012,
      "loss": 2.671919107437134,
      "step": 10462
    },
    {
      "epoch": 0.8484317158515914,
      "grad_norm": 0.030827490612864494,
      "learning_rate": 0.0012,
      "loss": 2.6780824661254883,
      "step": 10464
    },
    {
      "epoch": 0.8485938778767923,
      "grad_norm": 0.025676965713500977,
      "learning_rate": 0.0012,
      "loss": 2.706209182739258,
      "step": 10466
    },
    {
      "epoch": 0.8487560399019933,
      "grad_norm": 0.027027975767850876,
      "learning_rate": 0.0012,
      "loss": 2.635434627532959,
      "step": 10468
    },
    {
      "epoch": 0.8489182019271944,
      "grad_norm": 0.028670258820056915,
      "learning_rate": 0.0012,
      "loss": 2.6670572757720947,
      "step": 10470
    },
    {
      "epoch": 0.8490803639523953,
      "grad_norm": 0.02976754866540432,
      "learning_rate": 0.0012,
      "loss": 2.6755857467651367,
      "step": 10472
    },
    {
      "epoch": 0.8492425259775963,
      "grad_norm": 0.04128289595246315,
      "learning_rate": 0.0012,
      "loss": 2.6765787601470947,
      "step": 10474
    },
    {
      "epoch": 0.8494046880027973,
      "grad_norm": 0.027657778933644295,
      "learning_rate": 0.0012,
      "loss": 2.6139297485351562,
      "step": 10476
    },
    {
      "epoch": 0.8495668500279983,
      "grad_norm": 0.031673744320869446,
      "learning_rate": 0.0012,
      "loss": 2.6275391578674316,
      "step": 10478
    },
    {
      "epoch": 0.8497290120531993,
      "grad_norm": 0.031113654375076294,
      "learning_rate": 0.0012,
      "loss": 2.655932903289795,
      "step": 10480
    },
    {
      "epoch": 0.8498911740784003,
      "grad_norm": 0.03071192465722561,
      "learning_rate": 0.0012,
      "loss": 2.6269752979278564,
      "step": 10482
    },
    {
      "epoch": 0.8500533361036012,
      "grad_norm": 0.03950653597712517,
      "learning_rate": 0.0012,
      "loss": 2.656114339828491,
      "step": 10484
    },
    {
      "epoch": 0.8502154981288023,
      "grad_norm": 0.028108254075050354,
      "learning_rate": 0.0012,
      "loss": 2.6365034580230713,
      "step": 10486
    },
    {
      "epoch": 0.8503776601540033,
      "grad_norm": 0.03430657833814621,
      "learning_rate": 0.0012,
      "loss": 2.648648738861084,
      "step": 10488
    },
    {
      "epoch": 0.8505398221792042,
      "grad_norm": 0.06085526570677757,
      "learning_rate": 0.0012,
      "loss": 2.6368026733398438,
      "step": 10490
    },
    {
      "epoch": 0.8507019842044052,
      "grad_norm": 0.08060155808925629,
      "learning_rate": 0.0012,
      "loss": 2.6512484550476074,
      "step": 10492
    },
    {
      "epoch": 0.8508641462296063,
      "grad_norm": 0.03308883681893349,
      "learning_rate": 0.0012,
      "loss": 2.6237525939941406,
      "step": 10494
    },
    {
      "epoch": 0.8510263082548072,
      "grad_norm": 0.035548605024814606,
      "learning_rate": 0.0012,
      "loss": 2.6254160404205322,
      "step": 10496
    },
    {
      "epoch": 0.8511884702800082,
      "grad_norm": 0.05152931809425354,
      "learning_rate": 0.0012,
      "loss": 2.654496192932129,
      "step": 10498
    },
    {
      "epoch": 0.8513506323052092,
      "grad_norm": 0.032801080495119095,
      "learning_rate": 0.0012,
      "loss": 2.6145639419555664,
      "step": 10500
    },
    {
      "epoch": 0.8515127943304102,
      "grad_norm": 0.04015464708209038,
      "learning_rate": 0.0012,
      "loss": 2.624281883239746,
      "step": 10502
    },
    {
      "epoch": 0.8516749563556112,
      "grad_norm": 0.05392368137836456,
      "learning_rate": 0.0012,
      "loss": 2.642230272293091,
      "step": 10504
    },
    {
      "epoch": 0.8518371183808122,
      "grad_norm": 0.0321844145655632,
      "learning_rate": 0.0012,
      "loss": 2.626749038696289,
      "step": 10506
    },
    {
      "epoch": 0.8519992804060131,
      "grad_norm": 0.027568772435188293,
      "learning_rate": 0.0012,
      "loss": 2.719522476196289,
      "step": 10508
    },
    {
      "epoch": 0.8521614424312142,
      "grad_norm": 0.04762861877679825,
      "learning_rate": 0.0012,
      "loss": 2.6192750930786133,
      "step": 10510
    },
    {
      "epoch": 0.8523236044564152,
      "grad_norm": 0.0353546142578125,
      "learning_rate": 0.0012,
      "loss": 2.630462646484375,
      "step": 10512
    },
    {
      "epoch": 0.8524857664816161,
      "grad_norm": 0.027290424332022667,
      "learning_rate": 0.0012,
      "loss": 2.640338897705078,
      "step": 10514
    },
    {
      "epoch": 0.8526479285068171,
      "grad_norm": 0.043236397206783295,
      "learning_rate": 0.0012,
      "loss": 2.694387435913086,
      "step": 10516
    },
    {
      "epoch": 0.8528100905320182,
      "grad_norm": 0.03606991469860077,
      "learning_rate": 0.0012,
      "loss": 2.6922354698181152,
      "step": 10518
    },
    {
      "epoch": 0.8529722525572191,
      "grad_norm": 0.032384391874074936,
      "learning_rate": 0.0012,
      "loss": 2.6722092628479004,
      "step": 10520
    },
    {
      "epoch": 0.8531344145824201,
      "grad_norm": 0.046463530510663986,
      "learning_rate": 0.0012,
      "loss": 2.6210103034973145,
      "step": 10522
    },
    {
      "epoch": 0.8532965766076211,
      "grad_norm": 0.0261174775660038,
      "learning_rate": 0.0012,
      "loss": 2.651392936706543,
      "step": 10524
    },
    {
      "epoch": 0.8534587386328221,
      "grad_norm": 0.033367179334163666,
      "learning_rate": 0.0012,
      "loss": 2.639181137084961,
      "step": 10526
    },
    {
      "epoch": 0.8536209006580231,
      "grad_norm": 0.04391628876328468,
      "learning_rate": 0.0012,
      "loss": 2.638002395629883,
      "step": 10528
    },
    {
      "epoch": 0.8537830626832241,
      "grad_norm": 0.050880152732133865,
      "learning_rate": 0.0012,
      "loss": 2.6325888633728027,
      "step": 10530
    },
    {
      "epoch": 0.853945224708425,
      "grad_norm": 0.031522028148174286,
      "learning_rate": 0.0012,
      "loss": 2.560598373413086,
      "step": 10532
    },
    {
      "epoch": 0.8541073867336261,
      "grad_norm": 0.023654324933886528,
      "learning_rate": 0.0012,
      "loss": 2.5821971893310547,
      "step": 10534
    },
    {
      "epoch": 0.8542695487588271,
      "grad_norm": 0.025771422311663628,
      "learning_rate": 0.0012,
      "loss": 2.643655776977539,
      "step": 10536
    },
    {
      "epoch": 0.854431710784028,
      "grad_norm": 0.030463794246315956,
      "learning_rate": 0.0012,
      "loss": 2.689297914505005,
      "step": 10538
    },
    {
      "epoch": 0.854593872809229,
      "grad_norm": 0.03706200048327446,
      "learning_rate": 0.0012,
      "loss": 2.6601805686950684,
      "step": 10540
    },
    {
      "epoch": 0.8547560348344301,
      "grad_norm": 0.03996396064758301,
      "learning_rate": 0.0012,
      "loss": 2.681332588195801,
      "step": 10542
    },
    {
      "epoch": 0.854918196859631,
      "grad_norm": 0.04699377715587616,
      "learning_rate": 0.0012,
      "loss": 2.6502537727355957,
      "step": 10544
    },
    {
      "epoch": 0.855080358884832,
      "grad_norm": 0.02749411016702652,
      "learning_rate": 0.0012,
      "loss": 2.719130516052246,
      "step": 10546
    },
    {
      "epoch": 0.855242520910033,
      "grad_norm": 0.04143669083714485,
      "learning_rate": 0.0012,
      "loss": 2.656005859375,
      "step": 10548
    },
    {
      "epoch": 0.855404682935234,
      "grad_norm": 0.055371496826410294,
      "learning_rate": 0.0012,
      "loss": 2.641824722290039,
      "step": 10550
    },
    {
      "epoch": 0.855566844960435,
      "grad_norm": 0.03992273285984993,
      "learning_rate": 0.0012,
      "loss": 2.748404026031494,
      "step": 10552
    },
    {
      "epoch": 0.855729006985636,
      "grad_norm": 0.026368340477347374,
      "learning_rate": 0.0012,
      "loss": 2.687835693359375,
      "step": 10554
    },
    {
      "epoch": 0.855891169010837,
      "grad_norm": 0.0278308242559433,
      "learning_rate": 0.0012,
      "loss": 2.7263574600219727,
      "step": 10556
    },
    {
      "epoch": 0.856053331036038,
      "grad_norm": 0.03884035348892212,
      "learning_rate": 0.0012,
      "loss": 2.6922249794006348,
      "step": 10558
    },
    {
      "epoch": 0.856215493061239,
      "grad_norm": 0.0749986544251442,
      "learning_rate": 0.0012,
      "loss": 2.6126255989074707,
      "step": 10560
    },
    {
      "epoch": 0.8563776550864399,
      "grad_norm": 0.0871691107749939,
      "learning_rate": 0.0012,
      "loss": 2.724222183227539,
      "step": 10562
    },
    {
      "epoch": 0.856539817111641,
      "grad_norm": 0.04537944495677948,
      "learning_rate": 0.0012,
      "loss": 2.6283226013183594,
      "step": 10564
    },
    {
      "epoch": 0.856701979136842,
      "grad_norm": 0.029371917247772217,
      "learning_rate": 0.0012,
      "loss": 2.649394989013672,
      "step": 10566
    },
    {
      "epoch": 0.8568641411620429,
      "grad_norm": 0.04094978794455528,
      "learning_rate": 0.0012,
      "loss": 2.68229341506958,
      "step": 10568
    },
    {
      "epoch": 0.8570263031872439,
      "grad_norm": 0.04271988570690155,
      "learning_rate": 0.0012,
      "loss": 2.6512393951416016,
      "step": 10570
    },
    {
      "epoch": 0.857188465212445,
      "grad_norm": 0.03594940900802612,
      "learning_rate": 0.0012,
      "loss": 2.6221940517425537,
      "step": 10572
    },
    {
      "epoch": 0.8573506272376459,
      "grad_norm": 0.02982877753674984,
      "learning_rate": 0.0012,
      "loss": 2.636319637298584,
      "step": 10574
    },
    {
      "epoch": 0.8575127892628469,
      "grad_norm": 0.02923402562737465,
      "learning_rate": 0.0012,
      "loss": 2.7100884914398193,
      "step": 10576
    },
    {
      "epoch": 0.8576749512880479,
      "grad_norm": 0.02484891563653946,
      "learning_rate": 0.0012,
      "loss": 2.6007885932922363,
      "step": 10578
    },
    {
      "epoch": 0.8578371133132489,
      "grad_norm": 0.04391096904873848,
      "learning_rate": 0.0012,
      "loss": 2.652304172515869,
      "step": 10580
    },
    {
      "epoch": 0.8579992753384499,
      "grad_norm": 0.03316733241081238,
      "learning_rate": 0.0012,
      "loss": 2.603818893432617,
      "step": 10582
    },
    {
      "epoch": 0.8581614373636509,
      "grad_norm": 0.032952405512332916,
      "learning_rate": 0.0012,
      "loss": 2.629920482635498,
      "step": 10584
    },
    {
      "epoch": 0.8583235993888518,
      "grad_norm": 0.03281385824084282,
      "learning_rate": 0.0012,
      "loss": 2.6671152114868164,
      "step": 10586
    },
    {
      "epoch": 0.8584857614140529,
      "grad_norm": 0.03103972040116787,
      "learning_rate": 0.0012,
      "loss": 2.620211124420166,
      "step": 10588
    },
    {
      "epoch": 0.8586479234392539,
      "grad_norm": 0.04435543715953827,
      "learning_rate": 0.0012,
      "loss": 2.6895008087158203,
      "step": 10590
    },
    {
      "epoch": 0.8588100854644548,
      "grad_norm": 0.03592633828520775,
      "learning_rate": 0.0012,
      "loss": 2.6490345001220703,
      "step": 10592
    },
    {
      "epoch": 0.8589722474896558,
      "grad_norm": 0.029652582481503487,
      "learning_rate": 0.0012,
      "loss": 2.6654052734375,
      "step": 10594
    },
    {
      "epoch": 0.8591344095148569,
      "grad_norm": 0.025642044842243195,
      "learning_rate": 0.0012,
      "loss": 2.6681485176086426,
      "step": 10596
    },
    {
      "epoch": 0.8592965715400578,
      "grad_norm": 0.03712867945432663,
      "learning_rate": 0.0012,
      "loss": 2.613130569458008,
      "step": 10598
    },
    {
      "epoch": 0.8594587335652588,
      "grad_norm": 0.053151752799749374,
      "learning_rate": 0.0012,
      "loss": 2.663094997406006,
      "step": 10600
    },
    {
      "epoch": 0.8596208955904598,
      "grad_norm": 0.02574864774942398,
      "learning_rate": 0.0012,
      "loss": 2.6323578357696533,
      "step": 10602
    },
    {
      "epoch": 0.8597830576156608,
      "grad_norm": 0.02645597979426384,
      "learning_rate": 0.0012,
      "loss": 2.6583805084228516,
      "step": 10604
    },
    {
      "epoch": 0.8599452196408618,
      "grad_norm": 0.024421028792858124,
      "learning_rate": 0.0012,
      "loss": 2.613039493560791,
      "step": 10606
    },
    {
      "epoch": 0.8601073816660628,
      "grad_norm": 0.027267049998044968,
      "learning_rate": 0.0012,
      "loss": 2.618842601776123,
      "step": 10608
    },
    {
      "epoch": 0.8602695436912637,
      "grad_norm": 0.04157089442014694,
      "learning_rate": 0.0012,
      "loss": 2.6712806224823,
      "step": 10610
    },
    {
      "epoch": 0.8604317057164648,
      "grad_norm": 0.04845055565237999,
      "learning_rate": 0.0012,
      "loss": 2.665238380432129,
      "step": 10612
    },
    {
      "epoch": 0.8605938677416658,
      "grad_norm": 0.0283551886677742,
      "learning_rate": 0.0012,
      "loss": 2.6960790157318115,
      "step": 10614
    },
    {
      "epoch": 0.8607560297668667,
      "grad_norm": 0.02897179313004017,
      "learning_rate": 0.0012,
      "loss": 2.7040460109710693,
      "step": 10616
    },
    {
      "epoch": 0.8609181917920677,
      "grad_norm": 0.044164665043354034,
      "learning_rate": 0.0012,
      "loss": 2.6866228580474854,
      "step": 10618
    },
    {
      "epoch": 0.8610803538172688,
      "grad_norm": 0.05979525297880173,
      "learning_rate": 0.0012,
      "loss": 2.6381659507751465,
      "step": 10620
    },
    {
      "epoch": 0.8612425158424697,
      "grad_norm": 0.07358389347791672,
      "learning_rate": 0.0012,
      "loss": 2.6148345470428467,
      "step": 10622
    },
    {
      "epoch": 0.8614046778676707,
      "grad_norm": 0.039499927312135696,
      "learning_rate": 0.0012,
      "loss": 2.71229887008667,
      "step": 10624
    },
    {
      "epoch": 0.8615668398928717,
      "grad_norm": 0.02313259430229664,
      "learning_rate": 0.0012,
      "loss": 2.6963062286376953,
      "step": 10626
    },
    {
      "epoch": 0.8617290019180727,
      "grad_norm": 0.05442367494106293,
      "learning_rate": 0.0012,
      "loss": 2.6768381595611572,
      "step": 10628
    },
    {
      "epoch": 0.8618911639432737,
      "grad_norm": 0.04226117953658104,
      "learning_rate": 0.0012,
      "loss": 2.6701669692993164,
      "step": 10630
    },
    {
      "epoch": 0.8620533259684747,
      "grad_norm": 0.028997192159295082,
      "learning_rate": 0.0012,
      "loss": 2.6417768001556396,
      "step": 10632
    },
    {
      "epoch": 0.8622154879936756,
      "grad_norm": 0.04344933107495308,
      "learning_rate": 0.0012,
      "loss": 2.6686394214630127,
      "step": 10634
    },
    {
      "epoch": 0.8623776500188767,
      "grad_norm": 0.04053162783384323,
      "learning_rate": 0.0012,
      "loss": 2.644001007080078,
      "step": 10636
    },
    {
      "epoch": 0.8625398120440777,
      "grad_norm": 0.027897614985704422,
      "learning_rate": 0.0012,
      "loss": 2.6216304302215576,
      "step": 10638
    },
    {
      "epoch": 0.8627019740692786,
      "grad_norm": 0.03477257490158081,
      "learning_rate": 0.0012,
      "loss": 2.647303581237793,
      "step": 10640
    },
    {
      "epoch": 0.8628641360944797,
      "grad_norm": 0.034872036427259445,
      "learning_rate": 0.0012,
      "loss": 2.678948402404785,
      "step": 10642
    },
    {
      "epoch": 0.8630262981196807,
      "grad_norm": 0.056733835488557816,
      "learning_rate": 0.0012,
      "loss": 2.6612110137939453,
      "step": 10644
    },
    {
      "epoch": 0.8631884601448816,
      "grad_norm": 0.03223969787359238,
      "learning_rate": 0.0012,
      "loss": 2.684995412826538,
      "step": 10646
    },
    {
      "epoch": 0.8633506221700826,
      "grad_norm": 0.0274045430123806,
      "learning_rate": 0.0012,
      "loss": 2.676302433013916,
      "step": 10648
    },
    {
      "epoch": 0.8635127841952837,
      "grad_norm": 0.023616855964064598,
      "learning_rate": 0.0012,
      "loss": 2.729391098022461,
      "step": 10650
    },
    {
      "epoch": 0.8636749462204846,
      "grad_norm": 0.033760957419872284,
      "learning_rate": 0.0012,
      "loss": 2.620184898376465,
      "step": 10652
    },
    {
      "epoch": 0.8638371082456856,
      "grad_norm": 0.023046189919114113,
      "learning_rate": 0.0012,
      "loss": 2.616635322570801,
      "step": 10654
    },
    {
      "epoch": 0.8639992702708866,
      "grad_norm": 0.02288314327597618,
      "learning_rate": 0.0012,
      "loss": 2.6306657791137695,
      "step": 10656
    },
    {
      "epoch": 0.8641614322960876,
      "grad_norm": 0.025934230536222458,
      "learning_rate": 0.0012,
      "loss": 2.6388540267944336,
      "step": 10658
    },
    {
      "epoch": 0.8643235943212886,
      "grad_norm": 0.024399064481258392,
      "learning_rate": 0.0012,
      "loss": 2.6611990928649902,
      "step": 10660
    },
    {
      "epoch": 0.8644857563464896,
      "grad_norm": 0.025609856471419334,
      "learning_rate": 0.0012,
      "loss": 2.643238067626953,
      "step": 10662
    },
    {
      "epoch": 0.8646479183716905,
      "grad_norm": 0.02442837879061699,
      "learning_rate": 0.0012,
      "loss": 2.6604790687561035,
      "step": 10664
    },
    {
      "epoch": 0.8648100803968916,
      "grad_norm": 0.03730470687150955,
      "learning_rate": 0.0012,
      "loss": 2.6415939331054688,
      "step": 10666
    },
    {
      "epoch": 0.8649722424220926,
      "grad_norm": 0.02966475300490856,
      "learning_rate": 0.0012,
      "loss": 2.648320198059082,
      "step": 10668
    },
    {
      "epoch": 0.8651344044472935,
      "grad_norm": 0.02344663254916668,
      "learning_rate": 0.0012,
      "loss": 2.637284278869629,
      "step": 10670
    },
    {
      "epoch": 0.8652965664724945,
      "grad_norm": 0.03609570488333702,
      "learning_rate": 0.0012,
      "loss": 2.671668529510498,
      "step": 10672
    },
    {
      "epoch": 0.8654587284976956,
      "grad_norm": 0.055638477206230164,
      "learning_rate": 0.0012,
      "loss": 2.659151077270508,
      "step": 10674
    },
    {
      "epoch": 0.8656208905228965,
      "grad_norm": 0.058752886950969696,
      "learning_rate": 0.0012,
      "loss": 2.6459527015686035,
      "step": 10676
    },
    {
      "epoch": 0.8657830525480975,
      "grad_norm": 0.04854206368327141,
      "learning_rate": 0.0012,
      "loss": 2.5887041091918945,
      "step": 10678
    },
    {
      "epoch": 0.8659452145732985,
      "grad_norm": 0.03180943429470062,
      "learning_rate": 0.0012,
      "loss": 2.6444907188415527,
      "step": 10680
    },
    {
      "epoch": 0.8661073765984995,
      "grad_norm": 0.061902422457933426,
      "learning_rate": 0.0012,
      "loss": 2.641995429992676,
      "step": 10682
    },
    {
      "epoch": 0.8662695386237005,
      "grad_norm": 0.037194713950157166,
      "learning_rate": 0.0012,
      "loss": 2.601313591003418,
      "step": 10684
    },
    {
      "epoch": 0.8664317006489015,
      "grad_norm": 0.04364653304219246,
      "learning_rate": 0.0012,
      "loss": 2.6778063774108887,
      "step": 10686
    },
    {
      "epoch": 0.8665938626741024,
      "grad_norm": 0.05403335019946098,
      "learning_rate": 0.0012,
      "loss": 2.6403489112854004,
      "step": 10688
    },
    {
      "epoch": 0.8667560246993035,
      "grad_norm": 0.029515516012907028,
      "learning_rate": 0.0012,
      "loss": 2.6837637424468994,
      "step": 10690
    },
    {
      "epoch": 0.8669181867245045,
      "grad_norm": 0.0407450906932354,
      "learning_rate": 0.0012,
      "loss": 2.715935230255127,
      "step": 10692
    },
    {
      "epoch": 0.8670803487497054,
      "grad_norm": 0.05101694539189339,
      "learning_rate": 0.0012,
      "loss": 2.683352470397949,
      "step": 10694
    },
    {
      "epoch": 0.8672425107749064,
      "grad_norm": 0.0475538931787014,
      "learning_rate": 0.0012,
      "loss": 2.6885921955108643,
      "step": 10696
    },
    {
      "epoch": 0.8674046728001075,
      "grad_norm": 0.03239462152123451,
      "learning_rate": 0.0012,
      "loss": 2.722825288772583,
      "step": 10698
    },
    {
      "epoch": 0.8675668348253084,
      "grad_norm": 0.030191410332918167,
      "learning_rate": 0.0012,
      "loss": 2.7076122760772705,
      "step": 10700
    },
    {
      "epoch": 0.8677289968505094,
      "grad_norm": 0.0313849076628685,
      "learning_rate": 0.0012,
      "loss": 2.554074287414551,
      "step": 10702
    },
    {
      "epoch": 0.8678911588757104,
      "grad_norm": 0.028663665056228638,
      "learning_rate": 0.0012,
      "loss": 2.672800064086914,
      "step": 10704
    },
    {
      "epoch": 0.8680533209009114,
      "grad_norm": 0.030457941815257072,
      "learning_rate": 0.0012,
      "loss": 2.6905250549316406,
      "step": 10706
    },
    {
      "epoch": 0.8682154829261124,
      "grad_norm": 0.030491789802908897,
      "learning_rate": 0.0012,
      "loss": 2.5953993797302246,
      "step": 10708
    },
    {
      "epoch": 0.8683776449513134,
      "grad_norm": 0.025876663625240326,
      "learning_rate": 0.0012,
      "loss": 2.6928224563598633,
      "step": 10710
    },
    {
      "epoch": 0.8685398069765143,
      "grad_norm": 0.03028196468949318,
      "learning_rate": 0.0012,
      "loss": 2.678389072418213,
      "step": 10712
    },
    {
      "epoch": 0.8687019690017154,
      "grad_norm": 0.03298383578658104,
      "learning_rate": 0.0012,
      "loss": 2.65627384185791,
      "step": 10714
    },
    {
      "epoch": 0.8688641310269164,
      "grad_norm": 0.05261733755469322,
      "learning_rate": 0.0012,
      "loss": 2.616746425628662,
      "step": 10716
    },
    {
      "epoch": 0.8690262930521173,
      "grad_norm": 0.030353834852576256,
      "learning_rate": 0.0012,
      "loss": 2.7172482013702393,
      "step": 10718
    },
    {
      "epoch": 0.8691884550773183,
      "grad_norm": 0.026345156133174896,
      "learning_rate": 0.0012,
      "loss": 2.6733641624450684,
      "step": 10720
    },
    {
      "epoch": 0.8693506171025194,
      "grad_norm": 0.037569597363471985,
      "learning_rate": 0.0012,
      "loss": 2.6380224227905273,
      "step": 10722
    },
    {
      "epoch": 0.8695127791277203,
      "grad_norm": 0.06445734202861786,
      "learning_rate": 0.0012,
      "loss": 2.672863245010376,
      "step": 10724
    },
    {
      "epoch": 0.8696749411529213,
      "grad_norm": 0.06523755937814713,
      "learning_rate": 0.0012,
      "loss": 2.7051422595977783,
      "step": 10726
    },
    {
      "epoch": 0.8698371031781224,
      "grad_norm": 0.026935573667287827,
      "learning_rate": 0.0012,
      "loss": 2.639321804046631,
      "step": 10728
    },
    {
      "epoch": 0.8699992652033233,
      "grad_norm": 0.06481961905956268,
      "learning_rate": 0.0012,
      "loss": 2.5774149894714355,
      "step": 10730
    },
    {
      "epoch": 0.8701614272285243,
      "grad_norm": 0.06147518381476402,
      "learning_rate": 0.0012,
      "loss": 2.676804542541504,
      "step": 10732
    },
    {
      "epoch": 0.8703235892537253,
      "grad_norm": 0.036059487611055374,
      "learning_rate": 0.0012,
      "loss": 2.6499831676483154,
      "step": 10734
    },
    {
      "epoch": 0.8704857512789262,
      "grad_norm": 0.039733707904815674,
      "learning_rate": 0.0012,
      "loss": 2.6725914478302,
      "step": 10736
    },
    {
      "epoch": 0.8706479133041273,
      "grad_norm": 0.05660710111260414,
      "learning_rate": 0.0012,
      "loss": 2.6991536617279053,
      "step": 10738
    },
    {
      "epoch": 0.8708100753293283,
      "grad_norm": 0.031107960268855095,
      "learning_rate": 0.0012,
      "loss": 2.707054615020752,
      "step": 10740
    },
    {
      "epoch": 0.8709722373545292,
      "grad_norm": 0.03484917804598808,
      "learning_rate": 0.0012,
      "loss": 2.6770424842834473,
      "step": 10742
    },
    {
      "epoch": 0.8711343993797303,
      "grad_norm": 0.025281600654125214,
      "learning_rate": 0.0012,
      "loss": 2.712686061859131,
      "step": 10744
    },
    {
      "epoch": 0.8712965614049313,
      "grad_norm": 0.02985203266143799,
      "learning_rate": 0.0012,
      "loss": 2.6298463344573975,
      "step": 10746
    },
    {
      "epoch": 0.8714587234301322,
      "grad_norm": 0.04446684941649437,
      "learning_rate": 0.0012,
      "loss": 2.6566531658172607,
      "step": 10748
    },
    {
      "epoch": 0.8716208854553332,
      "grad_norm": 0.03408156335353851,
      "learning_rate": 0.0012,
      "loss": 2.676732063293457,
      "step": 10750
    },
    {
      "epoch": 0.8717830474805343,
      "grad_norm": 0.0244534220546484,
      "learning_rate": 0.0012,
      "loss": 2.6123852729797363,
      "step": 10752
    },
    {
      "epoch": 0.8719452095057352,
      "grad_norm": 0.0248312558978796,
      "learning_rate": 0.0012,
      "loss": 2.610654830932617,
      "step": 10754
    },
    {
      "epoch": 0.8721073715309362,
      "grad_norm": 0.02512338012456894,
      "learning_rate": 0.0012,
      "loss": 2.6872644424438477,
      "step": 10756
    },
    {
      "epoch": 0.8722695335561372,
      "grad_norm": 0.02930818684399128,
      "learning_rate": 0.0012,
      "loss": 2.6674365997314453,
      "step": 10758
    },
    {
      "epoch": 0.8724316955813382,
      "grad_norm": 0.03554282709956169,
      "learning_rate": 0.0012,
      "loss": 2.6726233959198,
      "step": 10760
    },
    {
      "epoch": 0.8725938576065392,
      "grad_norm": 0.02738306298851967,
      "learning_rate": 0.0012,
      "loss": 2.6970877647399902,
      "step": 10762
    },
    {
      "epoch": 0.8727560196317402,
      "grad_norm": 0.05901135876774788,
      "learning_rate": 0.0012,
      "loss": 2.640122890472412,
      "step": 10764
    },
    {
      "epoch": 0.8729181816569411,
      "grad_norm": 0.04834631457924843,
      "learning_rate": 0.0012,
      "loss": 2.6311933994293213,
      "step": 10766
    },
    {
      "epoch": 0.8730803436821422,
      "grad_norm": 0.0312400683760643,
      "learning_rate": 0.0012,
      "loss": 2.651965618133545,
      "step": 10768
    },
    {
      "epoch": 0.8732425057073432,
      "grad_norm": 0.049655746668577194,
      "learning_rate": 0.0012,
      "loss": 2.6773972511291504,
      "step": 10770
    },
    {
      "epoch": 0.8734046677325441,
      "grad_norm": 0.050953589379787445,
      "learning_rate": 0.0012,
      "loss": 2.689028024673462,
      "step": 10772
    },
    {
      "epoch": 0.8735668297577451,
      "grad_norm": 0.062282200902700424,
      "learning_rate": 0.0012,
      "loss": 2.6073551177978516,
      "step": 10774
    },
    {
      "epoch": 0.8737289917829462,
      "grad_norm": 0.03632522001862526,
      "learning_rate": 0.0012,
      "loss": 2.608036518096924,
      "step": 10776
    },
    {
      "epoch": 0.8738911538081471,
      "grad_norm": 0.04095935821533203,
      "learning_rate": 0.0012,
      "loss": 2.6180238723754883,
      "step": 10778
    },
    {
      "epoch": 0.8740533158333481,
      "grad_norm": 0.045082949101924896,
      "learning_rate": 0.0012,
      "loss": 2.654132843017578,
      "step": 10780
    },
    {
      "epoch": 0.8742154778585491,
      "grad_norm": 0.049537137150764465,
      "learning_rate": 0.0012,
      "loss": 2.6690330505371094,
      "step": 10782
    },
    {
      "epoch": 0.8743776398837501,
      "grad_norm": 0.02464883215725422,
      "learning_rate": 0.0012,
      "loss": 2.7107033729553223,
      "step": 10784
    },
    {
      "epoch": 0.8745398019089511,
      "grad_norm": 0.04509636387228966,
      "learning_rate": 0.0012,
      "loss": 2.652341842651367,
      "step": 10786
    },
    {
      "epoch": 0.8747019639341521,
      "grad_norm": 0.02908645197749138,
      "learning_rate": 0.0012,
      "loss": 2.6355435848236084,
      "step": 10788
    },
    {
      "epoch": 0.874864125959353,
      "grad_norm": 0.024551646783947945,
      "learning_rate": 0.0012,
      "loss": 2.658940076828003,
      "step": 10790
    },
    {
      "epoch": 0.8750262879845541,
      "grad_norm": 0.030678313225507736,
      "learning_rate": 0.0012,
      "loss": 2.6619162559509277,
      "step": 10792
    },
    {
      "epoch": 0.8751884500097551,
      "grad_norm": 0.043315671384334564,
      "learning_rate": 0.0012,
      "loss": 2.6817476749420166,
      "step": 10794
    },
    {
      "epoch": 0.875350612034956,
      "grad_norm": 0.031540147960186005,
      "learning_rate": 0.0012,
      "loss": 2.692789077758789,
      "step": 10796
    },
    {
      "epoch": 0.875512774060157,
      "grad_norm": 0.026258761063218117,
      "learning_rate": 0.0012,
      "loss": 2.6958694458007812,
      "step": 10798
    },
    {
      "epoch": 0.8756749360853581,
      "grad_norm": 0.027024511247873306,
      "learning_rate": 0.0012,
      "loss": 2.7314157485961914,
      "step": 10800
    },
    {
      "epoch": 0.875837098110559,
      "grad_norm": 0.025531502440571785,
      "learning_rate": 0.0012,
      "loss": 2.6830997467041016,
      "step": 10802
    },
    {
      "epoch": 0.87599926013576,
      "grad_norm": 0.025853268802165985,
      "learning_rate": 0.0012,
      "loss": 2.67667293548584,
      "step": 10804
    },
    {
      "epoch": 0.876161422160961,
      "grad_norm": 0.028050830587744713,
      "learning_rate": 0.0012,
      "loss": 2.6390271186828613,
      "step": 10806
    },
    {
      "epoch": 0.876323584186162,
      "grad_norm": 0.02780330926179886,
      "learning_rate": 0.0012,
      "loss": 2.657616138458252,
      "step": 10808
    },
    {
      "epoch": 0.876485746211363,
      "grad_norm": 0.03537152335047722,
      "learning_rate": 0.0012,
      "loss": 2.665961265563965,
      "step": 10810
    },
    {
      "epoch": 0.876647908236564,
      "grad_norm": 0.026184935122728348,
      "learning_rate": 0.0012,
      "loss": 2.676988124847412,
      "step": 10812
    },
    {
      "epoch": 0.8768100702617649,
      "grad_norm": 0.02999011054635048,
      "learning_rate": 0.0012,
      "loss": 2.6874895095825195,
      "step": 10814
    },
    {
      "epoch": 0.876972232286966,
      "grad_norm": 0.044021014124155045,
      "learning_rate": 0.0012,
      "loss": 2.628237724304199,
      "step": 10816
    },
    {
      "epoch": 0.877134394312167,
      "grad_norm": 0.04018241539597511,
      "learning_rate": 0.0012,
      "loss": 2.599064826965332,
      "step": 10818
    },
    {
      "epoch": 0.8772965563373679,
      "grad_norm": 0.026001349091529846,
      "learning_rate": 0.0012,
      "loss": 2.6846566200256348,
      "step": 10820
    },
    {
      "epoch": 0.877458718362569,
      "grad_norm": 0.06318832188844681,
      "learning_rate": 0.0012,
      "loss": 2.5737316608428955,
      "step": 10822
    },
    {
      "epoch": 0.87762088038777,
      "grad_norm": 0.06313878297805786,
      "learning_rate": 0.0012,
      "loss": 2.648503303527832,
      "step": 10824
    },
    {
      "epoch": 0.8777830424129709,
      "grad_norm": 0.03148777037858963,
      "learning_rate": 0.0012,
      "loss": 2.66194486618042,
      "step": 10826
    },
    {
      "epoch": 0.8779452044381719,
      "grad_norm": 0.028747595846652985,
      "learning_rate": 0.0012,
      "loss": 2.657285690307617,
      "step": 10828
    },
    {
      "epoch": 0.878107366463373,
      "grad_norm": 0.02943436987698078,
      "learning_rate": 0.0012,
      "loss": 2.6404213905334473,
      "step": 10830
    },
    {
      "epoch": 0.8782695284885739,
      "grad_norm": 0.046665292233228683,
      "learning_rate": 0.0012,
      "loss": 2.7188658714294434,
      "step": 10832
    },
    {
      "epoch": 0.8784316905137749,
      "grad_norm": 0.027858983725309372,
      "learning_rate": 0.0012,
      "loss": 2.6173737049102783,
      "step": 10834
    },
    {
      "epoch": 0.8785938525389759,
      "grad_norm": 0.048543766140937805,
      "learning_rate": 0.0012,
      "loss": 2.6812243461608887,
      "step": 10836
    },
    {
      "epoch": 0.8787560145641768,
      "grad_norm": 0.04679590091109276,
      "learning_rate": 0.0012,
      "loss": 2.6382851600646973,
      "step": 10838
    },
    {
      "epoch": 0.8789181765893779,
      "grad_norm": 0.023622434586286545,
      "learning_rate": 0.0012,
      "loss": 2.757983446121216,
      "step": 10840
    },
    {
      "epoch": 0.8790803386145789,
      "grad_norm": 0.042660508304834366,
      "learning_rate": 0.0012,
      "loss": 2.5821785926818848,
      "step": 10842
    },
    {
      "epoch": 0.8792425006397798,
      "grad_norm": 0.06508813053369522,
      "learning_rate": 0.0012,
      "loss": 2.639021158218384,
      "step": 10844
    },
    {
      "epoch": 0.8794046626649809,
      "grad_norm": 0.024717796593904495,
      "learning_rate": 0.0012,
      "loss": 2.604186534881592,
      "step": 10846
    },
    {
      "epoch": 0.8795668246901819,
      "grad_norm": 0.031057219952344894,
      "learning_rate": 0.0012,
      "loss": 2.6035470962524414,
      "step": 10848
    },
    {
      "epoch": 0.8797289867153828,
      "grad_norm": 0.025771018117666245,
      "learning_rate": 0.0012,
      "loss": 2.6243956089019775,
      "step": 10850
    },
    {
      "epoch": 0.8798911487405838,
      "grad_norm": 0.02770451456308365,
      "learning_rate": 0.0012,
      "loss": 2.6845240592956543,
      "step": 10852
    },
    {
      "epoch": 0.8800533107657849,
      "grad_norm": 0.02564157359302044,
      "learning_rate": 0.0012,
      "loss": 2.595071315765381,
      "step": 10854
    },
    {
      "epoch": 0.8802154727909858,
      "grad_norm": 0.02589300088584423,
      "learning_rate": 0.0012,
      "loss": 2.605969190597534,
      "step": 10856
    },
    {
      "epoch": 0.8803776348161868,
      "grad_norm": 0.02517584338784218,
      "learning_rate": 0.0012,
      "loss": 2.6791329383850098,
      "step": 10858
    },
    {
      "epoch": 0.8805397968413878,
      "grad_norm": 0.024552110582590103,
      "learning_rate": 0.0012,
      "loss": 2.593465805053711,
      "step": 10860
    },
    {
      "epoch": 0.8807019588665888,
      "grad_norm": 0.02346634306013584,
      "learning_rate": 0.0012,
      "loss": 2.6389174461364746,
      "step": 10862
    },
    {
      "epoch": 0.8808641208917898,
      "grad_norm": 0.02612270973622799,
      "learning_rate": 0.0012,
      "loss": 2.6386609077453613,
      "step": 10864
    },
    {
      "epoch": 0.8810262829169908,
      "grad_norm": 0.025027671828866005,
      "learning_rate": 0.0012,
      "loss": 2.698275566101074,
      "step": 10866
    },
    {
      "epoch": 0.8811884449421917,
      "grad_norm": 0.03578514605760574,
      "learning_rate": 0.0012,
      "loss": 2.5992066860198975,
      "step": 10868
    },
    {
      "epoch": 0.8813506069673928,
      "grad_norm": 0.03351902589201927,
      "learning_rate": 0.0012,
      "loss": 2.6570398807525635,
      "step": 10870
    },
    {
      "epoch": 0.8815127689925938,
      "grad_norm": 0.03410440683364868,
      "learning_rate": 0.0012,
      "loss": 2.7336883544921875,
      "step": 10872
    },
    {
      "epoch": 0.8816749310177947,
      "grad_norm": 0.0337831936776638,
      "learning_rate": 0.0012,
      "loss": 2.737752676010132,
      "step": 10874
    },
    {
      "epoch": 0.8818370930429957,
      "grad_norm": 0.037010446190834045,
      "learning_rate": 0.0012,
      "loss": 2.6886987686157227,
      "step": 10876
    },
    {
      "epoch": 0.8819992550681968,
      "grad_norm": 0.028789807111024857,
      "learning_rate": 0.0012,
      "loss": 2.56544828414917,
      "step": 10878
    },
    {
      "epoch": 0.8821614170933977,
      "grad_norm": 0.031867723912000656,
      "learning_rate": 0.0012,
      "loss": 2.6935911178588867,
      "step": 10880
    },
    {
      "epoch": 0.8823235791185987,
      "grad_norm": 0.04907330125570297,
      "learning_rate": 0.0012,
      "loss": 2.6727588176727295,
      "step": 10882
    },
    {
      "epoch": 0.8824857411437997,
      "grad_norm": 0.025552917271852493,
      "learning_rate": 0.0012,
      "loss": 2.6871120929718018,
      "step": 10884
    },
    {
      "epoch": 0.8826479031690007,
      "grad_norm": 0.028734246268868446,
      "learning_rate": 0.0012,
      "loss": 2.6067800521850586,
      "step": 10886
    },
    {
      "epoch": 0.8828100651942017,
      "grad_norm": 0.040895745158195496,
      "learning_rate": 0.0012,
      "loss": 2.592933416366577,
      "step": 10888
    },
    {
      "epoch": 0.8829722272194027,
      "grad_norm": 0.06460537761449814,
      "learning_rate": 0.0012,
      "loss": 2.6586503982543945,
      "step": 10890
    },
    {
      "epoch": 0.8831343892446036,
      "grad_norm": 0.03615601733326912,
      "learning_rate": 0.0012,
      "loss": 2.6700751781463623,
      "step": 10892
    },
    {
      "epoch": 0.8832965512698047,
      "grad_norm": 0.03327789157629013,
      "learning_rate": 0.0012,
      "loss": 2.681990623474121,
      "step": 10894
    },
    {
      "epoch": 0.8834587132950057,
      "grad_norm": 0.03051908314228058,
      "learning_rate": 0.0012,
      "loss": 2.623623847961426,
      "step": 10896
    },
    {
      "epoch": 0.8836208753202066,
      "grad_norm": 0.02712186984717846,
      "learning_rate": 0.0012,
      "loss": 2.6619114875793457,
      "step": 10898
    },
    {
      "epoch": 0.8837830373454076,
      "grad_norm": 0.03303850442171097,
      "learning_rate": 0.0012,
      "loss": 2.679922580718994,
      "step": 10900
    },
    {
      "epoch": 0.8839451993706087,
      "grad_norm": 0.06738936901092529,
      "learning_rate": 0.0012,
      "loss": 2.6281070709228516,
      "step": 10902
    },
    {
      "epoch": 0.8841073613958096,
      "grad_norm": 0.06672210246324539,
      "learning_rate": 0.0012,
      "loss": 2.6003406047821045,
      "step": 10904
    },
    {
      "epoch": 0.8842695234210106,
      "grad_norm": 0.030761169269680977,
      "learning_rate": 0.0012,
      "loss": 2.672396659851074,
      "step": 10906
    },
    {
      "epoch": 0.8844316854462116,
      "grad_norm": 0.039161793887615204,
      "learning_rate": 0.0012,
      "loss": 2.6598706245422363,
      "step": 10908
    },
    {
      "epoch": 0.8845938474714126,
      "grad_norm": 0.05492941662669182,
      "learning_rate": 0.0012,
      "loss": 2.631899356842041,
      "step": 10910
    },
    {
      "epoch": 0.8847560094966136,
      "grad_norm": 0.023717958480119705,
      "learning_rate": 0.0012,
      "loss": 2.6205694675445557,
      "step": 10912
    },
    {
      "epoch": 0.8849181715218146,
      "grad_norm": 0.027437346056103706,
      "learning_rate": 0.0012,
      "loss": 2.7468278408050537,
      "step": 10914
    },
    {
      "epoch": 0.8850803335470155,
      "grad_norm": 0.02602507919073105,
      "learning_rate": 0.0012,
      "loss": 2.6874923706054688,
      "step": 10916
    },
    {
      "epoch": 0.8852424955722166,
      "grad_norm": 0.02531374618411064,
      "learning_rate": 0.0012,
      "loss": 2.714940309524536,
      "step": 10918
    },
    {
      "epoch": 0.8854046575974176,
      "grad_norm": 0.04036043584346771,
      "learning_rate": 0.0012,
      "loss": 2.7070305347442627,
      "step": 10920
    },
    {
      "epoch": 0.8855668196226185,
      "grad_norm": 0.08319214731454849,
      "learning_rate": 0.0012,
      "loss": 2.6694536209106445,
      "step": 10922
    },
    {
      "epoch": 0.8857289816478195,
      "grad_norm": 0.04673445597290993,
      "learning_rate": 0.0012,
      "loss": 2.663753032684326,
      "step": 10924
    },
    {
      "epoch": 0.8858911436730206,
      "grad_norm": 0.026679212227463722,
      "learning_rate": 0.0012,
      "loss": 2.698303699493408,
      "step": 10926
    },
    {
      "epoch": 0.8860533056982215,
      "grad_norm": 0.06046805903315544,
      "learning_rate": 0.0012,
      "loss": 2.6283702850341797,
      "step": 10928
    },
    {
      "epoch": 0.8862154677234225,
      "grad_norm": 0.04495222121477127,
      "learning_rate": 0.0012,
      "loss": 2.681640148162842,
      "step": 10930
    },
    {
      "epoch": 0.8863776297486236,
      "grad_norm": 0.024687940254807472,
      "learning_rate": 0.0012,
      "loss": 2.620622396469116,
      "step": 10932
    },
    {
      "epoch": 0.8865397917738245,
      "grad_norm": 0.034171540290117264,
      "learning_rate": 0.0012,
      "loss": 2.643048048019409,
      "step": 10934
    },
    {
      "epoch": 0.8867019537990255,
      "grad_norm": 0.0546254999935627,
      "learning_rate": 0.0012,
      "loss": 2.7057442665100098,
      "step": 10936
    },
    {
      "epoch": 0.8868641158242265,
      "grad_norm": 0.032305438071489334,
      "learning_rate": 0.0012,
      "loss": 2.6460256576538086,
      "step": 10938
    },
    {
      "epoch": 0.8870262778494274,
      "grad_norm": 0.024299724027514458,
      "learning_rate": 0.0012,
      "loss": 2.6005020141601562,
      "step": 10940
    },
    {
      "epoch": 0.8871884398746285,
      "grad_norm": 0.026210160925984383,
      "learning_rate": 0.0012,
      "loss": 2.7338972091674805,
      "step": 10942
    },
    {
      "epoch": 0.8873506018998295,
      "grad_norm": 0.0503024198114872,
      "learning_rate": 0.0012,
      "loss": 2.6381125450134277,
      "step": 10944
    },
    {
      "epoch": 0.8875127639250304,
      "grad_norm": 0.04650400951504707,
      "learning_rate": 0.0012,
      "loss": 2.6608293056488037,
      "step": 10946
    },
    {
      "epoch": 0.8876749259502315,
      "grad_norm": 0.02534005604684353,
      "learning_rate": 0.0012,
      "loss": 2.6565966606140137,
      "step": 10948
    },
    {
      "epoch": 0.8878370879754325,
      "grad_norm": 0.05166751146316528,
      "learning_rate": 0.0012,
      "loss": 2.6036224365234375,
      "step": 10950
    },
    {
      "epoch": 0.8879992500006334,
      "grad_norm": 0.06779090315103531,
      "learning_rate": 0.0012,
      "loss": 2.6723239421844482,
      "step": 10952
    },
    {
      "epoch": 0.8881614120258344,
      "grad_norm": 0.02779216505587101,
      "learning_rate": 0.0012,
      "loss": 2.633810043334961,
      "step": 10954
    },
    {
      "epoch": 0.8883235740510355,
      "grad_norm": 0.04269394651055336,
      "learning_rate": 0.0012,
      "loss": 2.639549732208252,
      "step": 10956
    },
    {
      "epoch": 0.8884857360762364,
      "grad_norm": 0.06658411771059036,
      "learning_rate": 0.0012,
      "loss": 2.6384902000427246,
      "step": 10958
    },
    {
      "epoch": 0.8886478981014374,
      "grad_norm": 0.03921263664960861,
      "learning_rate": 0.0012,
      "loss": 2.6626980304718018,
      "step": 10960
    },
    {
      "epoch": 0.8888100601266384,
      "grad_norm": 0.02350567653775215,
      "learning_rate": 0.0012,
      "loss": 2.6823067665100098,
      "step": 10962
    },
    {
      "epoch": 0.8889722221518394,
      "grad_norm": 0.03189942613244057,
      "learning_rate": 0.0012,
      "loss": 2.6174697875976562,
      "step": 10964
    },
    {
      "epoch": 0.8891343841770404,
      "grad_norm": 0.02340407483279705,
      "learning_rate": 0.0012,
      "loss": 2.6210598945617676,
      "step": 10966
    },
    {
      "epoch": 0.8892965462022414,
      "grad_norm": 0.028734738007187843,
      "learning_rate": 0.0012,
      "loss": 2.6542649269104004,
      "step": 10968
    },
    {
      "epoch": 0.8894587082274423,
      "grad_norm": 0.025756746530532837,
      "learning_rate": 0.0012,
      "loss": 2.644883871078491,
      "step": 10970
    },
    {
      "epoch": 0.8896208702526434,
      "grad_norm": 0.0281473807990551,
      "learning_rate": 0.0012,
      "loss": 2.6593666076660156,
      "step": 10972
    },
    {
      "epoch": 0.8897830322778444,
      "grad_norm": 0.025791984051465988,
      "learning_rate": 0.0012,
      "loss": 2.6776275634765625,
      "step": 10974
    },
    {
      "epoch": 0.8899451943030453,
      "grad_norm": 0.037713054567575455,
      "learning_rate": 0.0012,
      "loss": 2.6945102214813232,
      "step": 10976
    },
    {
      "epoch": 0.8901073563282463,
      "grad_norm": 0.05170557647943497,
      "learning_rate": 0.0012,
      "loss": 2.6680006980895996,
      "step": 10978
    },
    {
      "epoch": 0.8902695183534474,
      "grad_norm": 0.05599929764866829,
      "learning_rate": 0.0012,
      "loss": 2.6297271251678467,
      "step": 10980
    },
    {
      "epoch": 0.8904316803786483,
      "grad_norm": 0.03724345564842224,
      "learning_rate": 0.0012,
      "loss": 2.665215015411377,
      "step": 10982
    },
    {
      "epoch": 0.8905938424038493,
      "grad_norm": 0.025032920762896538,
      "learning_rate": 0.0012,
      "loss": 2.6071524620056152,
      "step": 10984
    },
    {
      "epoch": 0.8907560044290503,
      "grad_norm": 0.029435504227876663,
      "learning_rate": 0.0012,
      "loss": 2.6456239223480225,
      "step": 10986
    },
    {
      "epoch": 0.8909181664542513,
      "grad_norm": 0.04609863460063934,
      "learning_rate": 0.0012,
      "loss": 2.6806693077087402,
      "step": 10988
    },
    {
      "epoch": 0.8910803284794523,
      "grad_norm": 0.032281238585710526,
      "learning_rate": 0.0012,
      "loss": 2.629631996154785,
      "step": 10990
    },
    {
      "epoch": 0.8912424905046533,
      "grad_norm": 0.026844115927815437,
      "learning_rate": 0.0011997289972899728,
      "loss": 2.6487178802490234,
      "step": 10992
    },
    {
      "epoch": 0.8914046525298542,
      "grad_norm": 0.03195367753505707,
      "learning_rate": 0.0011991869918699185,
      "loss": 2.6816961765289307,
      "step": 10994
    },
    {
      "epoch": 0.8915668145550553,
      "grad_norm": 0.03281693905591965,
      "learning_rate": 0.0011986449864498643,
      "loss": 2.69077730178833,
      "step": 10996
    },
    {
      "epoch": 0.8917289765802563,
      "grad_norm": 0.03857310861349106,
      "learning_rate": 0.0011981029810298103,
      "loss": 2.69455885887146,
      "step": 10998
    },
    {
      "epoch": 0.8918911386054572,
      "grad_norm": 0.0288342647254467,
      "learning_rate": 0.001197560975609756,
      "loss": 2.6930909156799316,
      "step": 11000
    },
    {
      "epoch": 0.8920533006306582,
      "grad_norm": 0.027259226888418198,
      "learning_rate": 0.0011970189701897018,
      "loss": 2.6536061763763428,
      "step": 11002
    },
    {
      "epoch": 0.8922154626558593,
      "grad_norm": 0.03342244774103165,
      "learning_rate": 0.0011964769647696476,
      "loss": 2.6615545749664307,
      "step": 11004
    },
    {
      "epoch": 0.8923776246810602,
      "grad_norm": 0.024301864206790924,
      "learning_rate": 0.0011959349593495933,
      "loss": 2.6569676399230957,
      "step": 11006
    },
    {
      "epoch": 0.8925397867062612,
      "grad_norm": 0.031368426978588104,
      "learning_rate": 0.001195392953929539,
      "loss": 2.649658203125,
      "step": 11008
    },
    {
      "epoch": 0.8927019487314622,
      "grad_norm": 0.026999456807971,
      "learning_rate": 0.001194850948509485,
      "loss": 2.550304412841797,
      "step": 11010
    },
    {
      "epoch": 0.8928641107566632,
      "grad_norm": 0.029232477769255638,
      "learning_rate": 0.0011943089430894308,
      "loss": 2.649362087249756,
      "step": 11012
    },
    {
      "epoch": 0.8930262727818642,
      "grad_norm": 0.030563170090317726,
      "learning_rate": 0.0011937669376693766,
      "loss": 2.621260643005371,
      "step": 11014
    },
    {
      "epoch": 0.8931884348070652,
      "grad_norm": 0.02517833188176155,
      "learning_rate": 0.0011932249322493223,
      "loss": 2.599428176879883,
      "step": 11016
    },
    {
      "epoch": 0.8933505968322661,
      "grad_norm": 0.026665296405553818,
      "learning_rate": 0.001192682926829268,
      "loss": 2.6443850994110107,
      "step": 11018
    },
    {
      "epoch": 0.8935127588574672,
      "grad_norm": 0.029758863151073456,
      "learning_rate": 0.0011921409214092139,
      "loss": 2.6916394233703613,
      "step": 11020
    },
    {
      "epoch": 0.8936749208826682,
      "grad_norm": 0.026742352172732353,
      "learning_rate": 0.0011915989159891598,
      "loss": 2.6243810653686523,
      "step": 11022
    },
    {
      "epoch": 0.8938370829078691,
      "grad_norm": 0.03749086707830429,
      "learning_rate": 0.0011910569105691056,
      "loss": 2.645503520965576,
      "step": 11024
    },
    {
      "epoch": 0.8939992449330701,
      "grad_norm": 0.05187178775668144,
      "learning_rate": 0.0011905149051490514,
      "loss": 2.6203830242156982,
      "step": 11026
    },
    {
      "epoch": 0.8941614069582712,
      "grad_norm": 0.04985440894961357,
      "learning_rate": 0.0011899728997289971,
      "loss": 2.6476199626922607,
      "step": 11028
    },
    {
      "epoch": 0.8943235689834721,
      "grad_norm": 0.06842823326587677,
      "learning_rate": 0.0011894308943089429,
      "loss": 2.6737489700317383,
      "step": 11030
    },
    {
      "epoch": 0.8944857310086731,
      "grad_norm": 0.04292386397719383,
      "learning_rate": 0.0011888888888888886,
      "loss": 2.674057960510254,
      "step": 11032
    },
    {
      "epoch": 0.8946478930338742,
      "grad_norm": 0.029162529855966568,
      "learning_rate": 0.0011883468834688346,
      "loss": 2.6561532020568848,
      "step": 11034
    },
    {
      "epoch": 0.8948100550590751,
      "grad_norm": 0.04922806844115257,
      "learning_rate": 0.0011878048780487804,
      "loss": 2.5825424194335938,
      "step": 11036
    },
    {
      "epoch": 0.8949722170842761,
      "grad_norm": 0.06048431992530823,
      "learning_rate": 0.0011872628726287261,
      "loss": 2.6264867782592773,
      "step": 11038
    },
    {
      "epoch": 0.8951343791094771,
      "grad_norm": 0.047033652663230896,
      "learning_rate": 0.001186720867208672,
      "loss": 2.666085720062256,
      "step": 11040
    },
    {
      "epoch": 0.895296541134678,
      "grad_norm": 0.025731299072504044,
      "learning_rate": 0.0011861788617886177,
      "loss": 2.669975996017456,
      "step": 11042
    },
    {
      "epoch": 0.8954587031598791,
      "grad_norm": 0.02788332849740982,
      "learning_rate": 0.0011856368563685636,
      "loss": 2.654798984527588,
      "step": 11044
    },
    {
      "epoch": 0.8956208651850801,
      "grad_norm": 0.029392346739768982,
      "learning_rate": 0.0011850948509485094,
      "loss": 2.6351819038391113,
      "step": 11046
    },
    {
      "epoch": 0.895783027210281,
      "grad_norm": 0.027321549132466316,
      "learning_rate": 0.0011845528455284552,
      "loss": 2.669172525405884,
      "step": 11048
    },
    {
      "epoch": 0.895945189235482,
      "grad_norm": 0.025715474039316177,
      "learning_rate": 0.001184010840108401,
      "loss": 2.676607608795166,
      "step": 11050
    },
    {
      "epoch": 0.8961073512606831,
      "grad_norm": 0.031594038009643555,
      "learning_rate": 0.0011834688346883467,
      "loss": 2.698178768157959,
      "step": 11052
    },
    {
      "epoch": 0.896269513285884,
      "grad_norm": 0.047316450625658035,
      "learning_rate": 0.0011829268292682924,
      "loss": 2.67952823638916,
      "step": 11054
    },
    {
      "epoch": 0.896431675311085,
      "grad_norm": 0.03469435125589371,
      "learning_rate": 0.0011823848238482384,
      "loss": 2.7001965045928955,
      "step": 11056
    },
    {
      "epoch": 0.8965938373362861,
      "grad_norm": 0.027889633551239967,
      "learning_rate": 0.0011818428184281842,
      "loss": 2.655463218688965,
      "step": 11058
    },
    {
      "epoch": 0.896755999361487,
      "grad_norm": 0.05300725996494293,
      "learning_rate": 0.00118130081300813,
      "loss": 2.663041114807129,
      "step": 11060
    },
    {
      "epoch": 0.896918161386688,
      "grad_norm": 0.0364573635160923,
      "learning_rate": 0.0011807588075880757,
      "loss": 2.6089940071105957,
      "step": 11062
    },
    {
      "epoch": 0.897080323411889,
      "grad_norm": 0.024661418050527573,
      "learning_rate": 0.0011802168021680215,
      "loss": 2.6324357986450195,
      "step": 11064
    },
    {
      "epoch": 0.89724248543709,
      "grad_norm": 0.08462685346603394,
      "learning_rate": 0.0011796747967479674,
      "loss": 2.6621575355529785,
      "step": 11066
    },
    {
      "epoch": 0.897404647462291,
      "grad_norm": 0.09504001587629318,
      "learning_rate": 0.0011791327913279132,
      "loss": 2.6431450843811035,
      "step": 11068
    },
    {
      "epoch": 0.897566809487492,
      "grad_norm": 0.024691203609108925,
      "learning_rate": 0.001178590785907859,
      "loss": 2.665109634399414,
      "step": 11070
    },
    {
      "epoch": 0.8977289715126929,
      "grad_norm": 0.0627012774348259,
      "learning_rate": 0.0011780487804878047,
      "loss": 2.651793956756592,
      "step": 11072
    },
    {
      "epoch": 0.897891133537894,
      "grad_norm": 0.04193783178925514,
      "learning_rate": 0.0011775067750677505,
      "loss": 2.633129596710205,
      "step": 11074
    },
    {
      "epoch": 0.898053295563095,
      "grad_norm": 0.03299489617347717,
      "learning_rate": 0.0011769647696476964,
      "loss": 2.624150276184082,
      "step": 11076
    },
    {
      "epoch": 0.8982154575882959,
      "grad_norm": 0.03339770808815956,
      "learning_rate": 0.0011764227642276422,
      "loss": 2.617997169494629,
      "step": 11078
    },
    {
      "epoch": 0.8983776196134969,
      "grad_norm": 0.039736852049827576,
      "learning_rate": 0.001175880758807588,
      "loss": 2.6718499660491943,
      "step": 11080
    },
    {
      "epoch": 0.898539781638698,
      "grad_norm": 0.07877648621797562,
      "learning_rate": 0.0011753387533875337,
      "loss": 2.651123523712158,
      "step": 11082
    },
    {
      "epoch": 0.8987019436638989,
      "grad_norm": 0.044307105243206024,
      "learning_rate": 0.0011747967479674795,
      "loss": 2.6385531425476074,
      "step": 11084
    },
    {
      "epoch": 0.8988641056890999,
      "grad_norm": 0.05551261082291603,
      "learning_rate": 0.0011742547425474252,
      "loss": 2.6287918090820312,
      "step": 11086
    },
    {
      "epoch": 0.8990262677143009,
      "grad_norm": 0.05379801243543625,
      "learning_rate": 0.0011737127371273712,
      "loss": 2.6826891899108887,
      "step": 11088
    },
    {
      "epoch": 0.8991884297395019,
      "grad_norm": 0.024725396186113358,
      "learning_rate": 0.001173170731707317,
      "loss": 2.6086835861206055,
      "step": 11090
    },
    {
      "epoch": 0.8993505917647029,
      "grad_norm": 0.050198037177324295,
      "learning_rate": 0.0011726287262872627,
      "loss": 2.6821041107177734,
      "step": 11092
    },
    {
      "epoch": 0.8995127537899039,
      "grad_norm": 0.04787140712141991,
      "learning_rate": 0.0011720867208672085,
      "loss": 2.681000232696533,
      "step": 11094
    },
    {
      "epoch": 0.8996749158151048,
      "grad_norm": 0.029663825407624245,
      "learning_rate": 0.0011715447154471543,
      "loss": 2.6549465656280518,
      "step": 11096
    },
    {
      "epoch": 0.8998370778403059,
      "grad_norm": 0.0548357218503952,
      "learning_rate": 0.0011710027100271002,
      "loss": 2.660388946533203,
      "step": 11098
    },
    {
      "epoch": 0.8999992398655069,
      "grad_norm": 0.03451484069228172,
      "learning_rate": 0.001170460704607046,
      "loss": 2.6799228191375732,
      "step": 11100
    },
    {
      "epoch": 0.9001614018907078,
      "grad_norm": 0.031111430376768112,
      "learning_rate": 0.0011699186991869918,
      "loss": 2.5998992919921875,
      "step": 11102
    },
    {
      "epoch": 0.9003235639159088,
      "grad_norm": 0.05148565396666527,
      "learning_rate": 0.0011693766937669375,
      "loss": 2.6810355186462402,
      "step": 11104
    },
    {
      "epoch": 0.9004857259411099,
      "grad_norm": 0.031026555225253105,
      "learning_rate": 0.0011688346883468833,
      "loss": 2.649834632873535,
      "step": 11106
    },
    {
      "epoch": 0.9006478879663108,
      "grad_norm": 0.04356495663523674,
      "learning_rate": 0.0011682926829268293,
      "loss": 2.6535592079162598,
      "step": 11108
    },
    {
      "epoch": 0.9008100499915118,
      "grad_norm": 0.033501774072647095,
      "learning_rate": 0.001167750677506775,
      "loss": 2.638432741165161,
      "step": 11110
    },
    {
      "epoch": 0.9009722120167128,
      "grad_norm": 0.02699403278529644,
      "learning_rate": 0.0011672086720867208,
      "loss": 2.6485233306884766,
      "step": 11112
    },
    {
      "epoch": 0.9011343740419138,
      "grad_norm": 0.03659229725599289,
      "learning_rate": 0.0011666666666666665,
      "loss": 2.6430046558380127,
      "step": 11114
    },
    {
      "epoch": 0.9012965360671148,
      "grad_norm": 0.05346805602312088,
      "learning_rate": 0.0011661246612466123,
      "loss": 2.6559762954711914,
      "step": 11116
    },
    {
      "epoch": 0.9014586980923158,
      "grad_norm": 0.026005661115050316,
      "learning_rate": 0.001165582655826558,
      "loss": 2.5851082801818848,
      "step": 11118
    },
    {
      "epoch": 0.9016208601175167,
      "grad_norm": 0.034055907279253006,
      "learning_rate": 0.001165040650406504,
      "loss": 2.5916919708251953,
      "step": 11120
    },
    {
      "epoch": 0.9017830221427178,
      "grad_norm": 0.025810400024056435,
      "learning_rate": 0.0011644986449864498,
      "loss": 2.643932342529297,
      "step": 11122
    },
    {
      "epoch": 0.9019451841679188,
      "grad_norm": 0.028676409274339676,
      "learning_rate": 0.0011639566395663956,
      "loss": 2.6381096839904785,
      "step": 11124
    },
    {
      "epoch": 0.9021073461931197,
      "grad_norm": 0.024243680760264397,
      "learning_rate": 0.0011634146341463413,
      "loss": 2.66127347946167,
      "step": 11126
    },
    {
      "epoch": 0.9022695082183207,
      "grad_norm": 0.040778547525405884,
      "learning_rate": 0.001162872628726287,
      "loss": 2.5624279975891113,
      "step": 11128
    },
    {
      "epoch": 0.9024316702435218,
      "grad_norm": 0.024845276027917862,
      "learning_rate": 0.001162330623306233,
      "loss": 2.6549768447875977,
      "step": 11130
    },
    {
      "epoch": 0.9025938322687227,
      "grad_norm": 0.026141418144106865,
      "learning_rate": 0.0011617886178861788,
      "loss": 2.646463632583618,
      "step": 11132
    },
    {
      "epoch": 0.9027559942939237,
      "grad_norm": 0.026104534044861794,
      "learning_rate": 0.0011612466124661246,
      "loss": 2.655503988265991,
      "step": 11134
    },
    {
      "epoch": 0.9029181563191248,
      "grad_norm": 0.041829317808151245,
      "learning_rate": 0.0011607046070460703,
      "loss": 2.646902561187744,
      "step": 11136
    },
    {
      "epoch": 0.9030803183443257,
      "grad_norm": 0.037474311888217926,
      "learning_rate": 0.001160162601626016,
      "loss": 2.646122694015503,
      "step": 11138
    },
    {
      "epoch": 0.9032424803695267,
      "grad_norm": 0.027452880516648293,
      "learning_rate": 0.001159620596205962,
      "loss": 2.607236385345459,
      "step": 11140
    },
    {
      "epoch": 0.9034046423947277,
      "grad_norm": 0.039022047072649,
      "learning_rate": 0.0011590785907859078,
      "loss": 2.6543126106262207,
      "step": 11142
    },
    {
      "epoch": 0.9035668044199286,
      "grad_norm": 0.04451137036085129,
      "learning_rate": 0.0011585365853658536,
      "loss": 2.695587158203125,
      "step": 11144
    },
    {
      "epoch": 0.9037289664451297,
      "grad_norm": 0.05873934552073479,
      "learning_rate": 0.0011579945799457994,
      "loss": 2.6140999794006348,
      "step": 11146
    },
    {
      "epoch": 0.9038911284703307,
      "grad_norm": 0.043280281126499176,
      "learning_rate": 0.0011574525745257451,
      "loss": 2.5931785106658936,
      "step": 11148
    },
    {
      "epoch": 0.9040532904955316,
      "grad_norm": 0.025227690115571022,
      "learning_rate": 0.001156910569105691,
      "loss": 2.7561824321746826,
      "step": 11150
    },
    {
      "epoch": 0.9042154525207327,
      "grad_norm": 0.03769110515713692,
      "learning_rate": 0.0011563685636856369,
      "loss": 2.618732213973999,
      "step": 11152
    },
    {
      "epoch": 0.9043776145459337,
      "grad_norm": 0.06561446189880371,
      "learning_rate": 0.0011558265582655826,
      "loss": 2.6644957065582275,
      "step": 11154
    },
    {
      "epoch": 0.9045397765711346,
      "grad_norm": 0.023186925798654556,
      "learning_rate": 0.0011552845528455284,
      "loss": 2.6927690505981445,
      "step": 11156
    },
    {
      "epoch": 0.9047019385963356,
      "grad_norm": 0.06639531999826431,
      "learning_rate": 0.0011547425474254741,
      "loss": 2.594862461090088,
      "step": 11158
    },
    {
      "epoch": 0.9048641006215367,
      "grad_norm": 0.05446270480751991,
      "learning_rate": 0.00115420054200542,
      "loss": 2.6521217823028564,
      "step": 11160
    },
    {
      "epoch": 0.9050262626467376,
      "grad_norm": 0.032260600477457047,
      "learning_rate": 0.0011536585365853659,
      "loss": 2.6760239601135254,
      "step": 11162
    },
    {
      "epoch": 0.9051884246719386,
      "grad_norm": 0.055813662707805634,
      "learning_rate": 0.0011531165311653116,
      "loss": 2.6289517879486084,
      "step": 11164
    },
    {
      "epoch": 0.9053505866971396,
      "grad_norm": 0.024427788332104683,
      "learning_rate": 0.0011525745257452574,
      "loss": 2.694662094116211,
      "step": 11166
    },
    {
      "epoch": 0.9055127487223406,
      "grad_norm": 0.045530352741479874,
      "learning_rate": 0.0011520325203252032,
      "loss": 2.666553020477295,
      "step": 11168
    },
    {
      "epoch": 0.9056749107475416,
      "grad_norm": 0.04710822552442551,
      "learning_rate": 0.001151490514905149,
      "loss": 2.613837480545044,
      "step": 11170
    },
    {
      "epoch": 0.9058370727727426,
      "grad_norm": 0.030720841139554977,
      "learning_rate": 0.001150948509485095,
      "loss": 2.585395336151123,
      "step": 11172
    },
    {
      "epoch": 0.9059992347979435,
      "grad_norm": 0.027337992563843727,
      "learning_rate": 0.0011504065040650407,
      "loss": 2.6291990280151367,
      "step": 11174
    },
    {
      "epoch": 0.9061613968231446,
      "grad_norm": 0.02479574643075466,
      "learning_rate": 0.0011498644986449864,
      "loss": 2.6275742053985596,
      "step": 11176
    },
    {
      "epoch": 0.9063235588483456,
      "grad_norm": 0.025047050788998604,
      "learning_rate": 0.0011493224932249322,
      "loss": 2.5876004695892334,
      "step": 11178
    },
    {
      "epoch": 0.9064857208735465,
      "grad_norm": 0.027393197640776634,
      "learning_rate": 0.001148780487804878,
      "loss": 2.660768985748291,
      "step": 11180
    },
    {
      "epoch": 0.9066478828987475,
      "grad_norm": 0.04732023552060127,
      "learning_rate": 0.0011482384823848237,
      "loss": 2.7029621601104736,
      "step": 11182
    },
    {
      "epoch": 0.9068100449239486,
      "grad_norm": 0.02800411731004715,
      "learning_rate": 0.0011476964769647697,
      "loss": 2.6092963218688965,
      "step": 11184
    },
    {
      "epoch": 0.9069722069491495,
      "grad_norm": 0.030626438558101654,
      "learning_rate": 0.0011471544715447154,
      "loss": 2.660754442214966,
      "step": 11186
    },
    {
      "epoch": 0.9071343689743505,
      "grad_norm": 0.04219692200422287,
      "learning_rate": 0.0011466124661246612,
      "loss": 2.587571620941162,
      "step": 11188
    },
    {
      "epoch": 0.9072965309995515,
      "grad_norm": 0.0266142338514328,
      "learning_rate": 0.001146070460704607,
      "loss": 2.6997151374816895,
      "step": 11190
    },
    {
      "epoch": 0.9074586930247525,
      "grad_norm": 0.03788625821471214,
      "learning_rate": 0.0011455284552845527,
      "loss": 2.6977925300598145,
      "step": 11192
    },
    {
      "epoch": 0.9076208550499535,
      "grad_norm": 0.02663324400782585,
      "learning_rate": 0.0011449864498644985,
      "loss": 2.6563825607299805,
      "step": 11194
    },
    {
      "epoch": 0.9077830170751545,
      "grad_norm": 0.04563462734222412,
      "learning_rate": 0.0011444444444444445,
      "loss": 2.7017006874084473,
      "step": 11196
    },
    {
      "epoch": 0.9079451791003554,
      "grad_norm": 0.05265957862138748,
      "learning_rate": 0.0011439024390243902,
      "loss": 2.632061719894409,
      "step": 11198
    },
    {
      "epoch": 0.9081073411255565,
      "grad_norm": 0.029055463150143623,
      "learning_rate": 0.001143360433604336,
      "loss": 2.660989761352539,
      "step": 11200
    },
    {
      "epoch": 0.9082695031507575,
      "grad_norm": 0.02959148958325386,
      "learning_rate": 0.0011428184281842817,
      "loss": 2.6442813873291016,
      "step": 11202
    },
    {
      "epoch": 0.9084316651759584,
      "grad_norm": 0.04511315003037453,
      "learning_rate": 0.0011422764227642275,
      "loss": 2.6044747829437256,
      "step": 11204
    },
    {
      "epoch": 0.9085938272011594,
      "grad_norm": 0.05178837105631828,
      "learning_rate": 0.0011417344173441733,
      "loss": 2.6595163345336914,
      "step": 11206
    },
    {
      "epoch": 0.9087559892263605,
      "grad_norm": 0.02456456981599331,
      "learning_rate": 0.0011411924119241192,
      "loss": 2.6806559562683105,
      "step": 11208
    },
    {
      "epoch": 0.9089181512515614,
      "grad_norm": 0.052426207810640335,
      "learning_rate": 0.001140650406504065,
      "loss": 2.700202465057373,
      "step": 11210
    },
    {
      "epoch": 0.9090803132767624,
      "grad_norm": 0.0709402784705162,
      "learning_rate": 0.0011401084010840108,
      "loss": 2.658888339996338,
      "step": 11212
    },
    {
      "epoch": 0.9092424753019634,
      "grad_norm": 0.02452078089118004,
      "learning_rate": 0.0011395663956639565,
      "loss": 2.654392719268799,
      "step": 11214
    },
    {
      "epoch": 0.9094046373271644,
      "grad_norm": 0.061966631561517715,
      "learning_rate": 0.0011390243902439023,
      "loss": 2.742982864379883,
      "step": 11216
    },
    {
      "epoch": 0.9095667993523654,
      "grad_norm": 0.07354535907506943,
      "learning_rate": 0.001138482384823848,
      "loss": 2.684535503387451,
      "step": 11218
    },
    {
      "epoch": 0.9097289613775664,
      "grad_norm": 0.027775665745139122,
      "learning_rate": 0.001137940379403794,
      "loss": 2.671910285949707,
      "step": 11220
    },
    {
      "epoch": 0.9098911234027673,
      "grad_norm": 0.036384467035532,
      "learning_rate": 0.0011373983739837398,
      "loss": 2.627087116241455,
      "step": 11222
    },
    {
      "epoch": 0.9100532854279684,
      "grad_norm": 0.028936829417943954,
      "learning_rate": 0.0011368563685636855,
      "loss": 2.643425464630127,
      "step": 11224
    },
    {
      "epoch": 0.9102154474531694,
      "grad_norm": 0.03273051232099533,
      "learning_rate": 0.0011363143631436313,
      "loss": 2.686896800994873,
      "step": 11226
    },
    {
      "epoch": 0.9103776094783703,
      "grad_norm": 0.04893738403916359,
      "learning_rate": 0.001135772357723577,
      "loss": 2.6110002994537354,
      "step": 11228
    },
    {
      "epoch": 0.9105397715035713,
      "grad_norm": 0.026186589151620865,
      "learning_rate": 0.0011352303523035228,
      "loss": 2.6222376823425293,
      "step": 11230
    },
    {
      "epoch": 0.9107019335287724,
      "grad_norm": 0.023482900112867355,
      "learning_rate": 0.0011346883468834688,
      "loss": 2.6648311614990234,
      "step": 11232
    },
    {
      "epoch": 0.9108640955539733,
      "grad_norm": 0.023408854380249977,
      "learning_rate": 0.0011341463414634146,
      "loss": 2.734200954437256,
      "step": 11234
    },
    {
      "epoch": 0.9110262575791743,
      "grad_norm": 0.03317952901124954,
      "learning_rate": 0.0011336043360433603,
      "loss": 2.5653762817382812,
      "step": 11236
    },
    {
      "epoch": 0.9111884196043754,
      "grad_norm": 0.04064216837286949,
      "learning_rate": 0.001133062330623306,
      "loss": 2.6297574043273926,
      "step": 11238
    },
    {
      "epoch": 0.9113505816295763,
      "grad_norm": 0.02809704653918743,
      "learning_rate": 0.0011325203252032518,
      "loss": 2.692091941833496,
      "step": 11240
    },
    {
      "epoch": 0.9115127436547773,
      "grad_norm": 0.05367521941661835,
      "learning_rate": 0.0011319783197831976,
      "loss": 2.635989189147949,
      "step": 11242
    },
    {
      "epoch": 0.9116749056799783,
      "grad_norm": 0.04592041298747063,
      "learning_rate": 0.0011314363143631436,
      "loss": 2.6552181243896484,
      "step": 11244
    },
    {
      "epoch": 0.9118370677051792,
      "grad_norm": 0.030322272330522537,
      "learning_rate": 0.0011308943089430893,
      "loss": 2.712113857269287,
      "step": 11246
    },
    {
      "epoch": 0.9119992297303803,
      "grad_norm": 0.024057965725660324,
      "learning_rate": 0.001130352303523035,
      "loss": 2.62802791595459,
      "step": 11248
    },
    {
      "epoch": 0.9121613917555813,
      "grad_norm": 0.024815132841467857,
      "learning_rate": 0.0011298102981029809,
      "loss": 2.6792383193969727,
      "step": 11250
    },
    {
      "epoch": 0.9123235537807822,
      "grad_norm": 0.028840504586696625,
      "learning_rate": 0.0011292682926829266,
      "loss": 2.620223045349121,
      "step": 11252
    },
    {
      "epoch": 0.9124857158059833,
      "grad_norm": 0.03745029494166374,
      "learning_rate": 0.0011287262872628724,
      "loss": 2.6592679023742676,
      "step": 11254
    },
    {
      "epoch": 0.9126478778311843,
      "grad_norm": 0.03183932229876518,
      "learning_rate": 0.0011281842818428184,
      "loss": 2.603799819946289,
      "step": 11256
    },
    {
      "epoch": 0.9128100398563852,
      "grad_norm": 0.028396468609571457,
      "learning_rate": 0.0011276422764227641,
      "loss": 2.6734390258789062,
      "step": 11258
    },
    {
      "epoch": 0.9129722018815862,
      "grad_norm": 0.04924379661679268,
      "learning_rate": 0.0011271002710027099,
      "loss": 2.651515245437622,
      "step": 11260
    },
    {
      "epoch": 0.9131343639067873,
      "grad_norm": 0.06382068991661072,
      "learning_rate": 0.0011265582655826556,
      "loss": 2.613110065460205,
      "step": 11262
    },
    {
      "epoch": 0.9132965259319882,
      "grad_norm": 0.035668790340423584,
      "learning_rate": 0.0011260162601626014,
      "loss": 2.6486892700195312,
      "step": 11264
    },
    {
      "epoch": 0.9134586879571892,
      "grad_norm": 0.06493192166090012,
      "learning_rate": 0.0011254742547425472,
      "loss": 2.6594672203063965,
      "step": 11266
    },
    {
      "epoch": 0.9136208499823902,
      "grad_norm": 0.05659659206867218,
      "learning_rate": 0.0011249322493224931,
      "loss": 2.6886727809906006,
      "step": 11268
    },
    {
      "epoch": 0.9137830120075912,
      "grad_norm": 0.0258766058832407,
      "learning_rate": 0.001124390243902439,
      "loss": 2.619436740875244,
      "step": 11270
    },
    {
      "epoch": 0.9139451740327922,
      "grad_norm": 0.05610670894384384,
      "learning_rate": 0.0011238482384823847,
      "loss": 2.6261518001556396,
      "step": 11272
    },
    {
      "epoch": 0.9141073360579932,
      "grad_norm": 0.034552235156297684,
      "learning_rate": 0.0011233062330623304,
      "loss": 2.671755313873291,
      "step": 11274
    },
    {
      "epoch": 0.9142694980831941,
      "grad_norm": 0.030672581866383553,
      "learning_rate": 0.0011227642276422762,
      "loss": 2.7257871627807617,
      "step": 11276
    },
    {
      "epoch": 0.9144316601083952,
      "grad_norm": 0.06939415633678436,
      "learning_rate": 0.0011222222222222222,
      "loss": 2.6682159900665283,
      "step": 11278
    },
    {
      "epoch": 0.9145938221335962,
      "grad_norm": 0.04637368395924568,
      "learning_rate": 0.001121680216802168,
      "loss": 2.613950729370117,
      "step": 11280
    },
    {
      "epoch": 0.9147559841587971,
      "grad_norm": 0.025269851088523865,
      "learning_rate": 0.0011211382113821137,
      "loss": 2.6622536182403564,
      "step": 11282
    },
    {
      "epoch": 0.9149181461839981,
      "grad_norm": 0.02548152022063732,
      "learning_rate": 0.0011205962059620594,
      "loss": 2.6487555503845215,
      "step": 11284
    },
    {
      "epoch": 0.9150803082091992,
      "grad_norm": 0.02377154491841793,
      "learning_rate": 0.0011200542005420052,
      "loss": 2.6593005657196045,
      "step": 11286
    },
    {
      "epoch": 0.9152424702344001,
      "grad_norm": 0.03685295954346657,
      "learning_rate": 0.0011195121951219512,
      "loss": 2.65341854095459,
      "step": 11288
    },
    {
      "epoch": 0.9154046322596011,
      "grad_norm": 0.03839182108640671,
      "learning_rate": 0.001118970189701897,
      "loss": 2.6191272735595703,
      "step": 11290
    },
    {
      "epoch": 0.9155667942848021,
      "grad_norm": 0.024788981303572655,
      "learning_rate": 0.0011184281842818427,
      "loss": 2.687343120574951,
      "step": 11292
    },
    {
      "epoch": 0.9157289563100031,
      "grad_norm": 0.026321904733777046,
      "learning_rate": 0.0011178861788617885,
      "loss": 2.731570243835449,
      "step": 11294
    },
    {
      "epoch": 0.9158911183352041,
      "grad_norm": 0.02404925599694252,
      "learning_rate": 0.0011173441734417342,
      "loss": 2.671146869659424,
      "step": 11296
    },
    {
      "epoch": 0.9160532803604051,
      "grad_norm": 0.034028325229883194,
      "learning_rate": 0.00111680216802168,
      "loss": 2.6612467765808105,
      "step": 11298
    },
    {
      "epoch": 0.916215442385606,
      "grad_norm": 0.031850703060626984,
      "learning_rate": 0.001116260162601626,
      "loss": 2.678809404373169,
      "step": 11300
    },
    {
      "epoch": 0.9163776044108071,
      "grad_norm": 0.02938586287200451,
      "learning_rate": 0.0011157181571815717,
      "loss": 2.6449029445648193,
      "step": 11302
    },
    {
      "epoch": 0.9165397664360081,
      "grad_norm": 0.03897496685385704,
      "learning_rate": 0.0011151761517615175,
      "loss": 2.6611328125,
      "step": 11304
    },
    {
      "epoch": 0.916701928461209,
      "grad_norm": 0.034982211887836456,
      "learning_rate": 0.0011146341463414632,
      "loss": 2.671971559524536,
      "step": 11306
    },
    {
      "epoch": 0.91686409048641,
      "grad_norm": 0.03624209389090538,
      "learning_rate": 0.001114092140921409,
      "loss": 2.694629192352295,
      "step": 11308
    },
    {
      "epoch": 0.9170262525116111,
      "grad_norm": 0.026020485907793045,
      "learning_rate": 0.001113550135501355,
      "loss": 2.6621809005737305,
      "step": 11310
    },
    {
      "epoch": 0.917188414536812,
      "grad_norm": 0.026593003422021866,
      "learning_rate": 0.0011130081300813007,
      "loss": 2.661482334136963,
      "step": 11312
    },
    {
      "epoch": 0.917350576562013,
      "grad_norm": 0.025974199175834656,
      "learning_rate": 0.0011124661246612465,
      "loss": 2.7453360557556152,
      "step": 11314
    },
    {
      "epoch": 0.917512738587214,
      "grad_norm": 0.024065211415290833,
      "learning_rate": 0.0011119241192411923,
      "loss": 2.6540780067443848,
      "step": 11316
    },
    {
      "epoch": 0.9176749006124151,
      "grad_norm": 0.025758979842066765,
      "learning_rate": 0.001111382113821138,
      "loss": 2.6510095596313477,
      "step": 11318
    },
    {
      "epoch": 0.917837062637616,
      "grad_norm": 0.02586134523153305,
      "learning_rate": 0.001110840108401084,
      "loss": 2.653803825378418,
      "step": 11320
    },
    {
      "epoch": 0.917999224662817,
      "grad_norm": 0.02322201430797577,
      "learning_rate": 0.0011102981029810298,
      "loss": 2.6360530853271484,
      "step": 11322
    },
    {
      "epoch": 0.918161386688018,
      "grad_norm": 0.02389395423233509,
      "learning_rate": 0.0011097560975609755,
      "loss": 2.731729030609131,
      "step": 11324
    },
    {
      "epoch": 0.918323548713219,
      "grad_norm": 0.03209315612912178,
      "learning_rate": 0.0011092140921409213,
      "loss": 2.66306209564209,
      "step": 11326
    },
    {
      "epoch": 0.91848571073842,
      "grad_norm": 0.02703963778913021,
      "learning_rate": 0.001108672086720867,
      "loss": 2.526460647583008,
      "step": 11328
    },
    {
      "epoch": 0.918647872763621,
      "grad_norm": 0.024472761899232864,
      "learning_rate": 0.001108130081300813,
      "loss": 2.6630964279174805,
      "step": 11330
    },
    {
      "epoch": 0.918810034788822,
      "grad_norm": 0.03316411003470421,
      "learning_rate": 0.0011075880758807588,
      "loss": 2.6284213066101074,
      "step": 11332
    },
    {
      "epoch": 0.918972196814023,
      "grad_norm": 0.034354887902736664,
      "learning_rate": 0.0011070460704607045,
      "loss": 2.6325888633728027,
      "step": 11334
    },
    {
      "epoch": 0.919134358839224,
      "grad_norm": 0.04980611428618431,
      "learning_rate": 0.0011065040650406503,
      "loss": 2.724043846130371,
      "step": 11336
    },
    {
      "epoch": 0.9192965208644249,
      "grad_norm": 0.033755749464035034,
      "learning_rate": 0.001105962059620596,
      "loss": 2.642911672592163,
      "step": 11338
    },
    {
      "epoch": 0.919458682889626,
      "grad_norm": 0.025055857375264168,
      "learning_rate": 0.0011054200542005418,
      "loss": 2.723376750946045,
      "step": 11340
    },
    {
      "epoch": 0.919620844914827,
      "grad_norm": 0.033734988421201706,
      "learning_rate": 0.0011048780487804878,
      "loss": 2.675445079803467,
      "step": 11342
    },
    {
      "epoch": 0.9197830069400279,
      "grad_norm": 0.024991218000650406,
      "learning_rate": 0.0011043360433604335,
      "loss": 2.600480556488037,
      "step": 11344
    },
    {
      "epoch": 0.9199451689652289,
      "grad_norm": 0.02574424259364605,
      "learning_rate": 0.0011037940379403793,
      "loss": 2.661801815032959,
      "step": 11346
    },
    {
      "epoch": 0.92010733099043,
      "grad_norm": 0.02703353576362133,
      "learning_rate": 0.001103252032520325,
      "loss": 2.6812806129455566,
      "step": 11348
    },
    {
      "epoch": 0.9202694930156309,
      "grad_norm": 0.040459003299474716,
      "learning_rate": 0.0011027100271002708,
      "loss": 2.632171154022217,
      "step": 11350
    },
    {
      "epoch": 0.9204316550408319,
      "grad_norm": 0.05527844280004501,
      "learning_rate": 0.0011021680216802168,
      "loss": 2.6642494201660156,
      "step": 11352
    },
    {
      "epoch": 0.9205938170660329,
      "grad_norm": 0.025888744741678238,
      "learning_rate": 0.0011016260162601626,
      "loss": 2.601719379425049,
      "step": 11354
    },
    {
      "epoch": 0.9207559790912339,
      "grad_norm": 0.02527719922363758,
      "learning_rate": 0.0011010840108401083,
      "loss": 2.6193737983703613,
      "step": 11356
    },
    {
      "epoch": 0.9209181411164349,
      "grad_norm": 0.047858722507953644,
      "learning_rate": 0.001100542005420054,
      "loss": 2.666375160217285,
      "step": 11358
    },
    {
      "epoch": 0.9210803031416359,
      "grad_norm": 0.029847895726561546,
      "learning_rate": 0.0010999999999999998,
      "loss": 2.6094706058502197,
      "step": 11360
    },
    {
      "epoch": 0.9212424651668368,
      "grad_norm": 0.03174791485071182,
      "learning_rate": 0.0010994579945799458,
      "loss": 2.618223190307617,
      "step": 11362
    },
    {
      "epoch": 0.9214046271920379,
      "grad_norm": 0.02545003779232502,
      "learning_rate": 0.0010989159891598916,
      "loss": 2.6229088306427,
      "step": 11364
    },
    {
      "epoch": 0.9215667892172389,
      "grad_norm": 0.03015412762761116,
      "learning_rate": 0.0010983739837398373,
      "loss": 2.627830982208252,
      "step": 11366
    },
    {
      "epoch": 0.9217289512424398,
      "grad_norm": 0.025390472263097763,
      "learning_rate": 0.001097831978319783,
      "loss": 2.6368966102600098,
      "step": 11368
    },
    {
      "epoch": 0.9218911132676408,
      "grad_norm": 0.024217523634433746,
      "learning_rate": 0.0010972899728997289,
      "loss": 2.6741249561309814,
      "step": 11370
    },
    {
      "epoch": 0.9220532752928419,
      "grad_norm": 0.024788374081254005,
      "learning_rate": 0.0010967479674796746,
      "loss": 2.7085845470428467,
      "step": 11372
    },
    {
      "epoch": 0.9222154373180428,
      "grad_norm": 0.03024129569530487,
      "learning_rate": 0.0010962059620596206,
      "loss": 2.6565589904785156,
      "step": 11374
    },
    {
      "epoch": 0.9223775993432438,
      "grad_norm": 0.02981507033109665,
      "learning_rate": 0.0010956639566395664,
      "loss": 2.6778922080993652,
      "step": 11376
    },
    {
      "epoch": 0.9225397613684448,
      "grad_norm": 0.04541495442390442,
      "learning_rate": 0.0010951219512195121,
      "loss": 2.6316781044006348,
      "step": 11378
    },
    {
      "epoch": 0.9227019233936458,
      "grad_norm": 0.023127492517232895,
      "learning_rate": 0.0010945799457994579,
      "loss": 2.639678478240967,
      "step": 11380
    },
    {
      "epoch": 0.9228640854188468,
      "grad_norm": 0.02363434061408043,
      "learning_rate": 0.0010940379403794036,
      "loss": 2.6144723892211914,
      "step": 11382
    },
    {
      "epoch": 0.9230262474440478,
      "grad_norm": 0.039431821554899216,
      "learning_rate": 0.0010934959349593496,
      "loss": 2.6463866233825684,
      "step": 11384
    },
    {
      "epoch": 0.9231884094692487,
      "grad_norm": 0.04944028705358505,
      "learning_rate": 0.0010929539295392954,
      "loss": 2.692251682281494,
      "step": 11386
    },
    {
      "epoch": 0.9233505714944498,
      "grad_norm": 0.04770369827747345,
      "learning_rate": 0.0010924119241192411,
      "loss": 2.682236671447754,
      "step": 11388
    },
    {
      "epoch": 0.9235127335196508,
      "grad_norm": 0.061946142464876175,
      "learning_rate": 0.001091869918699187,
      "loss": 2.6663336753845215,
      "step": 11390
    },
    {
      "epoch": 0.9236748955448517,
      "grad_norm": 0.04437609016895294,
      "learning_rate": 0.0010913279132791327,
      "loss": 2.6662845611572266,
      "step": 11392
    },
    {
      "epoch": 0.9238370575700527,
      "grad_norm": 0.028412871062755585,
      "learning_rate": 0.0010907859078590786,
      "loss": 2.649758815765381,
      "step": 11394
    },
    {
      "epoch": 0.9239992195952538,
      "grad_norm": 0.025764012709259987,
      "learning_rate": 0.0010902439024390244,
      "loss": 2.6153669357299805,
      "step": 11396
    },
    {
      "epoch": 0.9241613816204547,
      "grad_norm": 0.02424730360507965,
      "learning_rate": 0.0010897018970189702,
      "loss": 2.6734681129455566,
      "step": 11398
    },
    {
      "epoch": 0.9243235436456557,
      "grad_norm": 0.03057757578790188,
      "learning_rate": 0.001089159891598916,
      "loss": 2.6089563369750977,
      "step": 11400
    },
    {
      "epoch": 0.9244857056708567,
      "grad_norm": 0.026794711127877235,
      "learning_rate": 0.0010886178861788617,
      "loss": 2.711963653564453,
      "step": 11402
    },
    {
      "epoch": 0.9246478676960577,
      "grad_norm": 0.0288597010076046,
      "learning_rate": 0.0010880758807588074,
      "loss": 2.686229705810547,
      "step": 11404
    },
    {
      "epoch": 0.9248100297212587,
      "grad_norm": 0.02743823267519474,
      "learning_rate": 0.0010875338753387534,
      "loss": 2.6229400634765625,
      "step": 11406
    },
    {
      "epoch": 0.9249721917464597,
      "grad_norm": 0.025605270639061928,
      "learning_rate": 0.0010869918699186992,
      "loss": 2.638261556625366,
      "step": 11408
    },
    {
      "epoch": 0.9251343537716606,
      "grad_norm": 0.029616381973028183,
      "learning_rate": 0.001086449864498645,
      "loss": 2.6802074909210205,
      "step": 11410
    },
    {
      "epoch": 0.9252965157968617,
      "grad_norm": 0.03988727927207947,
      "learning_rate": 0.0010859078590785907,
      "loss": 2.6897785663604736,
      "step": 11412
    },
    {
      "epoch": 0.9254586778220627,
      "grad_norm": 0.023987270891666412,
      "learning_rate": 0.0010853658536585365,
      "loss": 2.5926599502563477,
      "step": 11414
    },
    {
      "epoch": 0.9256208398472636,
      "grad_norm": 0.026790428906679153,
      "learning_rate": 0.0010848238482384824,
      "loss": 2.6749212741851807,
      "step": 11416
    },
    {
      "epoch": 0.9257830018724647,
      "grad_norm": 0.039658285677433014,
      "learning_rate": 0.0010842818428184282,
      "loss": 2.668919086456299,
      "step": 11418
    },
    {
      "epoch": 0.9259451638976657,
      "grad_norm": 0.056744206696748734,
      "learning_rate": 0.001083739837398374,
      "loss": 2.630993366241455,
      "step": 11420
    },
    {
      "epoch": 0.9261073259228666,
      "grad_norm": 0.0728098675608635,
      "learning_rate": 0.0010831978319783197,
      "loss": 2.628014326095581,
      "step": 11422
    },
    {
      "epoch": 0.9262694879480676,
      "grad_norm": 0.03406589478254318,
      "learning_rate": 0.0010826558265582655,
      "loss": 2.6409878730773926,
      "step": 11424
    },
    {
      "epoch": 0.9264316499732687,
      "grad_norm": 0.03628862649202347,
      "learning_rate": 0.0010821138211382112,
      "loss": 2.621349811553955,
      "step": 11426
    },
    {
      "epoch": 0.9265938119984696,
      "grad_norm": 0.0285029336810112,
      "learning_rate": 0.0010815718157181572,
      "loss": 2.6102101802825928,
      "step": 11428
    },
    {
      "epoch": 0.9267559740236706,
      "grad_norm": 0.036640334874391556,
      "learning_rate": 0.001081029810298103,
      "loss": 2.614450454711914,
      "step": 11430
    },
    {
      "epoch": 0.9269181360488716,
      "grad_norm": 0.026890628039836884,
      "learning_rate": 0.0010804878048780487,
      "loss": 2.651029586791992,
      "step": 11432
    },
    {
      "epoch": 0.9270802980740726,
      "grad_norm": 0.047736164182424545,
      "learning_rate": 0.0010799457994579945,
      "loss": 2.6863198280334473,
      "step": 11434
    },
    {
      "epoch": 0.9272424600992736,
      "grad_norm": 0.04278913140296936,
      "learning_rate": 0.0010794037940379403,
      "loss": 2.62400484085083,
      "step": 11436
    },
    {
      "epoch": 0.9274046221244746,
      "grad_norm": 0.027761101722717285,
      "learning_rate": 0.001078861788617886,
      "loss": 2.6602087020874023,
      "step": 11438
    },
    {
      "epoch": 0.9275667841496755,
      "grad_norm": 0.02838210016489029,
      "learning_rate": 0.001078319783197832,
      "loss": 2.688779354095459,
      "step": 11440
    },
    {
      "epoch": 0.9277289461748766,
      "grad_norm": 0.02640346810221672,
      "learning_rate": 0.0010777777777777778,
      "loss": 2.69915771484375,
      "step": 11442
    },
    {
      "epoch": 0.9278911082000776,
      "grad_norm": 0.026404282078146935,
      "learning_rate": 0.0010772357723577235,
      "loss": 2.737114906311035,
      "step": 11444
    },
    {
      "epoch": 0.9280532702252785,
      "grad_norm": 0.0240380447357893,
      "learning_rate": 0.0010766937669376693,
      "loss": 2.68406343460083,
      "step": 11446
    },
    {
      "epoch": 0.9282154322504795,
      "grad_norm": 0.04398747906088829,
      "learning_rate": 0.001076151761517615,
      "loss": 2.6414713859558105,
      "step": 11448
    },
    {
      "epoch": 0.9283775942756806,
      "grad_norm": 0.03218860179185867,
      "learning_rate": 0.0010756097560975608,
      "loss": 2.62947416305542,
      "step": 11450
    },
    {
      "epoch": 0.9285397563008815,
      "grad_norm": 0.026556195691227913,
      "learning_rate": 0.0010750677506775068,
      "loss": 2.7048377990722656,
      "step": 11452
    },
    {
      "epoch": 0.9287019183260825,
      "grad_norm": 0.02752045728266239,
      "learning_rate": 0.0010745257452574525,
      "loss": 2.6895618438720703,
      "step": 11454
    },
    {
      "epoch": 0.9288640803512835,
      "grad_norm": 0.024033470079302788,
      "learning_rate": 0.0010739837398373983,
      "loss": 2.636970043182373,
      "step": 11456
    },
    {
      "epoch": 0.9290262423764845,
      "grad_norm": 0.029035858809947968,
      "learning_rate": 0.001073441734417344,
      "loss": 2.609222412109375,
      "step": 11458
    },
    {
      "epoch": 0.9291884044016855,
      "grad_norm": 0.023145766928792,
      "learning_rate": 0.0010728997289972898,
      "loss": 2.6437883377075195,
      "step": 11460
    },
    {
      "epoch": 0.9293505664268865,
      "grad_norm": 0.025236645713448524,
      "learning_rate": 0.0010723577235772356,
      "loss": 2.6403303146362305,
      "step": 11462
    },
    {
      "epoch": 0.9295127284520874,
      "grad_norm": 0.059462666511535645,
      "learning_rate": 0.0010718157181571816,
      "loss": 2.6868748664855957,
      "step": 11464
    },
    {
      "epoch": 0.9296748904772885,
      "grad_norm": 0.040260713547468185,
      "learning_rate": 0.0010712737127371273,
      "loss": 2.639681577682495,
      "step": 11466
    },
    {
      "epoch": 0.9298370525024895,
      "grad_norm": 0.025181109085679054,
      "learning_rate": 0.001070731707317073,
      "loss": 2.6833534240722656,
      "step": 11468
    },
    {
      "epoch": 0.9299992145276904,
      "grad_norm": 0.05728946626186371,
      "learning_rate": 0.0010701897018970188,
      "loss": 2.630429744720459,
      "step": 11470
    },
    {
      "epoch": 0.9301613765528914,
      "grad_norm": 0.07972104847431183,
      "learning_rate": 0.0010696476964769646,
      "loss": 2.6530823707580566,
      "step": 11472
    },
    {
      "epoch": 0.9303235385780925,
      "grad_norm": 0.031689126044511795,
      "learning_rate": 0.0010691056910569104,
      "loss": 2.6675374507904053,
      "step": 11474
    },
    {
      "epoch": 0.9304857006032934,
      "grad_norm": 0.04180451110005379,
      "learning_rate": 0.0010685636856368563,
      "loss": 2.5711817741394043,
      "step": 11476
    },
    {
      "epoch": 0.9306478626284944,
      "grad_norm": 0.06511947512626648,
      "learning_rate": 0.001068021680216802,
      "loss": 2.7224576473236084,
      "step": 11478
    },
    {
      "epoch": 0.9308100246536954,
      "grad_norm": 0.042427632957696915,
      "learning_rate": 0.0010674796747967479,
      "loss": 2.6625919342041016,
      "step": 11480
    },
    {
      "epoch": 0.9309721866788964,
      "grad_norm": 0.0401419997215271,
      "learning_rate": 0.0010669376693766936,
      "loss": 2.6984195709228516,
      "step": 11482
    },
    {
      "epoch": 0.9311343487040974,
      "grad_norm": 0.07301852107048035,
      "learning_rate": 0.0010663956639566394,
      "loss": 2.6809182167053223,
      "step": 11484
    },
    {
      "epoch": 0.9312965107292984,
      "grad_norm": 0.03061525523662567,
      "learning_rate": 0.0010658536585365851,
      "loss": 2.587900161743164,
      "step": 11486
    },
    {
      "epoch": 0.9314586727544993,
      "grad_norm": 0.05619337782263756,
      "learning_rate": 0.0010653116531165311,
      "loss": 2.6163992881774902,
      "step": 11488
    },
    {
      "epoch": 0.9316208347797004,
      "grad_norm": 0.04202960804104805,
      "learning_rate": 0.0010647696476964769,
      "loss": 2.686128616333008,
      "step": 11490
    },
    {
      "epoch": 0.9317829968049014,
      "grad_norm": 0.023800522089004517,
      "learning_rate": 0.0010642276422764226,
      "loss": 2.6531901359558105,
      "step": 11492
    },
    {
      "epoch": 0.9319451588301023,
      "grad_norm": 0.0265100859105587,
      "learning_rate": 0.0010636856368563684,
      "loss": 2.6908822059631348,
      "step": 11494
    },
    {
      "epoch": 0.9321073208553033,
      "grad_norm": 0.03065306879580021,
      "learning_rate": 0.0010631436314363142,
      "loss": 2.649022102355957,
      "step": 11496
    },
    {
      "epoch": 0.9322694828805044,
      "grad_norm": 0.027258362621068954,
      "learning_rate": 0.00106260162601626,
      "loss": 2.678649425506592,
      "step": 11498
    },
    {
      "epoch": 0.9324316449057053,
      "grad_norm": 0.02982533909380436,
      "learning_rate": 0.001062059620596206,
      "loss": 2.71675968170166,
      "step": 11500
    },
    {
      "epoch": 0.9325938069309063,
      "grad_norm": 0.02381153404712677,
      "learning_rate": 0.0010615176151761517,
      "loss": 2.617763042449951,
      "step": 11502
    },
    {
      "epoch": 0.9327559689561074,
      "grad_norm": 0.024547133594751358,
      "learning_rate": 0.0010609756097560974,
      "loss": 2.6446518898010254,
      "step": 11504
    },
    {
      "epoch": 0.9329181309813083,
      "grad_norm": 0.045584067702293396,
      "learning_rate": 0.0010604336043360432,
      "loss": 2.6457486152648926,
      "step": 11506
    },
    {
      "epoch": 0.9330802930065093,
      "grad_norm": 0.03144770860671997,
      "learning_rate": 0.001059891598915989,
      "loss": 2.619950771331787,
      "step": 11508
    },
    {
      "epoch": 0.9332424550317103,
      "grad_norm": 0.031187688931822777,
      "learning_rate": 0.0010593495934959347,
      "loss": 2.6869797706604004,
      "step": 11510
    },
    {
      "epoch": 0.9334046170569112,
      "grad_norm": 0.029131298884749413,
      "learning_rate": 0.0010588075880758807,
      "loss": 2.6764049530029297,
      "step": 11512
    },
    {
      "epoch": 0.9335667790821123,
      "grad_norm": 0.02641988918185234,
      "learning_rate": 0.0010582655826558264,
      "loss": 2.673936367034912,
      "step": 11514
    },
    {
      "epoch": 0.9337289411073133,
      "grad_norm": 0.029614293947815895,
      "learning_rate": 0.0010577235772357722,
      "loss": 2.586857318878174,
      "step": 11516
    },
    {
      "epoch": 0.9338911031325142,
      "grad_norm": 0.03557489439845085,
      "learning_rate": 0.001057181571815718,
      "loss": 2.660975217819214,
      "step": 11518
    },
    {
      "epoch": 0.9340532651577153,
      "grad_norm": 0.03232504054903984,
      "learning_rate": 0.0010566395663956637,
      "loss": 2.6847879886627197,
      "step": 11520
    },
    {
      "epoch": 0.9342154271829163,
      "grad_norm": 0.033686161041259766,
      "learning_rate": 0.0010560975609756097,
      "loss": 2.658803701400757,
      "step": 11522
    },
    {
      "epoch": 0.9343775892081172,
      "grad_norm": 0.02946348302066326,
      "learning_rate": 0.0010555555555555555,
      "loss": 2.7024753093719482,
      "step": 11524
    },
    {
      "epoch": 0.9345397512333182,
      "grad_norm": 0.036523349583148956,
      "learning_rate": 0.0010550135501355012,
      "loss": 2.618814468383789,
      "step": 11526
    },
    {
      "epoch": 0.9347019132585193,
      "grad_norm": 0.057799603790044785,
      "learning_rate": 0.001054471544715447,
      "loss": 2.6865029335021973,
      "step": 11528
    },
    {
      "epoch": 0.9348640752837202,
      "grad_norm": 0.05439191684126854,
      "learning_rate": 0.0010539295392953927,
      "loss": 2.629910469055176,
      "step": 11530
    },
    {
      "epoch": 0.9350262373089212,
      "grad_norm": 0.02924889326095581,
      "learning_rate": 0.0010533875338753387,
      "loss": 2.6160545349121094,
      "step": 11532
    },
    {
      "epoch": 0.9351883993341222,
      "grad_norm": 0.057535573840141296,
      "learning_rate": 0.0010528455284552845,
      "loss": 2.6956682205200195,
      "step": 11534
    },
    {
      "epoch": 0.9353505613593232,
      "grad_norm": 0.05860369652509689,
      "learning_rate": 0.0010523035230352302,
      "loss": 2.7118115425109863,
      "step": 11536
    },
    {
      "epoch": 0.9355127233845242,
      "grad_norm": 0.04113089665770531,
      "learning_rate": 0.001051761517615176,
      "loss": 2.6714727878570557,
      "step": 11538
    },
    {
      "epoch": 0.9356748854097252,
      "grad_norm": 0.06253303587436676,
      "learning_rate": 0.0010512195121951218,
      "loss": 2.6554160118103027,
      "step": 11540
    },
    {
      "epoch": 0.9358370474349261,
      "grad_norm": 0.053250450640916824,
      "learning_rate": 0.0010506775067750677,
      "loss": 2.6513876914978027,
      "step": 11542
    },
    {
      "epoch": 0.9359992094601272,
      "grad_norm": 0.024961451068520546,
      "learning_rate": 0.0010501355013550135,
      "loss": 2.65847110748291,
      "step": 11544
    },
    {
      "epoch": 0.9361613714853282,
      "grad_norm": 0.04936720430850983,
      "learning_rate": 0.0010495934959349593,
      "loss": 2.669275999069214,
      "step": 11546
    },
    {
      "epoch": 0.9363235335105291,
      "grad_norm": 0.025744665414094925,
      "learning_rate": 0.001049051490514905,
      "loss": 2.692232608795166,
      "step": 11548
    },
    {
      "epoch": 0.9364856955357301,
      "grad_norm": 0.0564151369035244,
      "learning_rate": 0.0010485094850948508,
      "loss": 2.633121967315674,
      "step": 11550
    },
    {
      "epoch": 0.9366478575609312,
      "grad_norm": 0.05354185029864311,
      "learning_rate": 0.0010479674796747965,
      "loss": 2.6854758262634277,
      "step": 11552
    },
    {
      "epoch": 0.9368100195861321,
      "grad_norm": 0.03150342404842377,
      "learning_rate": 0.0010474254742547425,
      "loss": 2.6742656230926514,
      "step": 11554
    },
    {
      "epoch": 0.9369721816113331,
      "grad_norm": 0.03333459049463272,
      "learning_rate": 0.0010468834688346883,
      "loss": 2.6547975540161133,
      "step": 11556
    },
    {
      "epoch": 0.9371343436365341,
      "grad_norm": 0.03644736111164093,
      "learning_rate": 0.001046341463414634,
      "loss": 2.5367658138275146,
      "step": 11558
    },
    {
      "epoch": 0.9372965056617351,
      "grad_norm": 0.02673722431063652,
      "learning_rate": 0.0010457994579945798,
      "loss": 2.677469491958618,
      "step": 11560
    },
    {
      "epoch": 0.9374586676869361,
      "grad_norm": 0.042726337909698486,
      "learning_rate": 0.0010452574525745256,
      "loss": 2.7058634757995605,
      "step": 11562
    },
    {
      "epoch": 0.9376208297121371,
      "grad_norm": 0.025275345891714096,
      "learning_rate": 0.0010447154471544715,
      "loss": 2.619020938873291,
      "step": 11564
    },
    {
      "epoch": 0.937782991737338,
      "grad_norm": 0.03265144303441048,
      "learning_rate": 0.0010441734417344173,
      "loss": 2.6576876640319824,
      "step": 11566
    },
    {
      "epoch": 0.9379451537625391,
      "grad_norm": 0.036767613142728806,
      "learning_rate": 0.001043631436314363,
      "loss": 2.6416897773742676,
      "step": 11568
    },
    {
      "epoch": 0.9381073157877401,
      "grad_norm": 0.025341322645545006,
      "learning_rate": 0.0010430894308943088,
      "loss": 2.643014430999756,
      "step": 11570
    },
    {
      "epoch": 0.938269477812941,
      "grad_norm": 0.048076946288347244,
      "learning_rate": 0.0010425474254742546,
      "loss": 2.6687121391296387,
      "step": 11572
    },
    {
      "epoch": 0.938431639838142,
      "grad_norm": 0.040657345205545425,
      "learning_rate": 0.0010420054200542006,
      "loss": 2.730207681655884,
      "step": 11574
    },
    {
      "epoch": 0.9385938018633431,
      "grad_norm": 0.03000197745859623,
      "learning_rate": 0.0010414634146341463,
      "loss": 2.6304268836975098,
      "step": 11576
    },
    {
      "epoch": 0.938755963888544,
      "grad_norm": 0.03583608940243721,
      "learning_rate": 0.001040921409214092,
      "loss": 2.6165971755981445,
      "step": 11578
    },
    {
      "epoch": 0.938918125913745,
      "grad_norm": 0.02502564899623394,
      "learning_rate": 0.0010403794037940378,
      "loss": 2.6037988662719727,
      "step": 11580
    },
    {
      "epoch": 0.939080287938946,
      "grad_norm": 0.03580991551280022,
      "learning_rate": 0.0010398373983739836,
      "loss": 2.6348514556884766,
      "step": 11582
    },
    {
      "epoch": 0.939242449964147,
      "grad_norm": 0.045407865196466446,
      "learning_rate": 0.0010392953929539294,
      "loss": 2.637955904006958,
      "step": 11584
    },
    {
      "epoch": 0.939404611989348,
      "grad_norm": 0.0398552268743515,
      "learning_rate": 0.0010387533875338753,
      "loss": 2.6895642280578613,
      "step": 11586
    },
    {
      "epoch": 0.939566774014549,
      "grad_norm": 0.03575260937213898,
      "learning_rate": 0.001038211382113821,
      "loss": 2.724975824356079,
      "step": 11588
    },
    {
      "epoch": 0.9397289360397499,
      "grad_norm": 0.043637778609991074,
      "learning_rate": 0.0010376693766937669,
      "loss": 2.6326420307159424,
      "step": 11590
    },
    {
      "epoch": 0.939891098064951,
      "grad_norm": 0.0343000590801239,
      "learning_rate": 0.0010371273712737126,
      "loss": 2.6138272285461426,
      "step": 11592
    },
    {
      "epoch": 0.940053260090152,
      "grad_norm": 0.04229454696178436,
      "learning_rate": 0.0010365853658536584,
      "loss": 2.6634702682495117,
      "step": 11594
    },
    {
      "epoch": 0.9402154221153529,
      "grad_norm": 0.05206628143787384,
      "learning_rate": 0.0010360433604336044,
      "loss": 2.6348984241485596,
      "step": 11596
    },
    {
      "epoch": 0.940377584140554,
      "grad_norm": 0.02860487438738346,
      "learning_rate": 0.0010355013550135501,
      "loss": 2.665172576904297,
      "step": 11598
    },
    {
      "epoch": 0.940539746165755,
      "grad_norm": 0.047014180570840836,
      "learning_rate": 0.0010349593495934959,
      "loss": 2.723337173461914,
      "step": 11600
    },
    {
      "epoch": 0.9407019081909559,
      "grad_norm": 0.059636782854795456,
      "learning_rate": 0.0010344173441734416,
      "loss": 2.6611642837524414,
      "step": 11602
    },
    {
      "epoch": 0.9408640702161569,
      "grad_norm": 0.027361690998077393,
      "learning_rate": 0.0010338753387533874,
      "loss": 2.6705307960510254,
      "step": 11604
    },
    {
      "epoch": 0.941026232241358,
      "grad_norm": 0.04238593205809593,
      "learning_rate": 0.0010333333333333334,
      "loss": 2.6232094764709473,
      "step": 11606
    },
    {
      "epoch": 0.9411883942665589,
      "grad_norm": 0.04462633654475212,
      "learning_rate": 0.0010327913279132791,
      "loss": 2.6590287685394287,
      "step": 11608
    },
    {
      "epoch": 0.9413505562917599,
      "grad_norm": 0.033605873584747314,
      "learning_rate": 0.0010322493224932249,
      "loss": 2.6568994522094727,
      "step": 11610
    },
    {
      "epoch": 0.9415127183169609,
      "grad_norm": 0.05275816097855568,
      "learning_rate": 0.0010317073170731707,
      "loss": 2.6269664764404297,
      "step": 11612
    },
    {
      "epoch": 0.9416748803421618,
      "grad_norm": 0.02410031482577324,
      "learning_rate": 0.0010311653116531164,
      "loss": 2.608968734741211,
      "step": 11614
    },
    {
      "epoch": 0.9418370423673629,
      "grad_norm": 0.031261686235666275,
      "learning_rate": 0.0010306233062330622,
      "loss": 2.6385855674743652,
      "step": 11616
    },
    {
      "epoch": 0.9419992043925639,
      "grad_norm": 0.02458411268889904,
      "learning_rate": 0.0010300813008130081,
      "loss": 2.6592135429382324,
      "step": 11618
    },
    {
      "epoch": 0.9421613664177648,
      "grad_norm": 0.030284326523542404,
      "learning_rate": 0.001029539295392954,
      "loss": 2.6985080242156982,
      "step": 11620
    },
    {
      "epoch": 0.9423235284429659,
      "grad_norm": 0.026930736377835274,
      "learning_rate": 0.0010289972899728997,
      "loss": 2.6261816024780273,
      "step": 11622
    },
    {
      "epoch": 0.9424856904681669,
      "grad_norm": 0.02557341940701008,
      "learning_rate": 0.0010284552845528454,
      "loss": 2.6854097843170166,
      "step": 11624
    },
    {
      "epoch": 0.9426478524933678,
      "grad_norm": 0.026721930131316185,
      "learning_rate": 0.0010279132791327912,
      "loss": 2.678510904312134,
      "step": 11626
    },
    {
      "epoch": 0.9428100145185688,
      "grad_norm": 0.023858236148953438,
      "learning_rate": 0.0010273712737127372,
      "loss": 2.593900203704834,
      "step": 11628
    },
    {
      "epoch": 0.9429721765437699,
      "grad_norm": 0.03103775717318058,
      "learning_rate": 0.001026829268292683,
      "loss": 2.6492364406585693,
      "step": 11630
    },
    {
      "epoch": 0.9431343385689708,
      "grad_norm": 0.04066286236047745,
      "learning_rate": 0.0010262872628726287,
      "loss": 2.648364305496216,
      "step": 11632
    },
    {
      "epoch": 0.9432965005941718,
      "grad_norm": 0.02833704464137554,
      "learning_rate": 0.0010257452574525744,
      "loss": 2.620173454284668,
      "step": 11634
    },
    {
      "epoch": 0.9434586626193728,
      "grad_norm": 0.02626861073076725,
      "learning_rate": 0.0010252032520325202,
      "loss": 2.704723358154297,
      "step": 11636
    },
    {
      "epoch": 0.9436208246445738,
      "grad_norm": 0.03088536486029625,
      "learning_rate": 0.0010246612466124662,
      "loss": 2.6402487754821777,
      "step": 11638
    },
    {
      "epoch": 0.9437829866697748,
      "grad_norm": 0.027859199792146683,
      "learning_rate": 0.001024119241192412,
      "loss": 2.675999879837036,
      "step": 11640
    },
    {
      "epoch": 0.9439451486949758,
      "grad_norm": 0.05062833055853844,
      "learning_rate": 0.0010235772357723577,
      "loss": 2.722729444503784,
      "step": 11642
    },
    {
      "epoch": 0.9441073107201767,
      "grad_norm": 0.05420329421758652,
      "learning_rate": 0.0010230352303523035,
      "loss": 2.6557812690734863,
      "step": 11644
    },
    {
      "epoch": 0.9442694727453778,
      "grad_norm": 0.025544634088873863,
      "learning_rate": 0.0010224932249322492,
      "loss": 2.670619010925293,
      "step": 11646
    },
    {
      "epoch": 0.9444316347705788,
      "grad_norm": 0.04864715039730072,
      "learning_rate": 0.001021951219512195,
      "loss": 2.6854777336120605,
      "step": 11648
    },
    {
      "epoch": 0.9445937967957797,
      "grad_norm": 0.04699272662401199,
      "learning_rate": 0.001021409214092141,
      "loss": 2.634847640991211,
      "step": 11650
    },
    {
      "epoch": 0.9447559588209807,
      "grad_norm": 0.027559969574213028,
      "learning_rate": 0.0010208672086720867,
      "loss": 2.7036361694335938,
      "step": 11652
    },
    {
      "epoch": 0.9449181208461818,
      "grad_norm": 0.059054333716630936,
      "learning_rate": 0.0010203252032520325,
      "loss": 2.6557533740997314,
      "step": 11654
    },
    {
      "epoch": 0.9450802828713827,
      "grad_norm": 0.04379011318087578,
      "learning_rate": 0.0010197831978319782,
      "loss": 2.638338088989258,
      "step": 11656
    },
    {
      "epoch": 0.9452424448965837,
      "grad_norm": 0.03392767533659935,
      "learning_rate": 0.001019241192411924,
      "loss": 2.601177453994751,
      "step": 11658
    },
    {
      "epoch": 0.9454046069217847,
      "grad_norm": 0.047247763723134995,
      "learning_rate": 0.0010186991869918698,
      "loss": 2.667802333831787,
      "step": 11660
    },
    {
      "epoch": 0.9455667689469857,
      "grad_norm": 0.023915590718388557,
      "learning_rate": 0.0010181571815718157,
      "loss": 2.7160654067993164,
      "step": 11662
    },
    {
      "epoch": 0.9457289309721867,
      "grad_norm": 0.025878820568323135,
      "learning_rate": 0.0010176151761517615,
      "loss": 2.6098685264587402,
      "step": 11664
    },
    {
      "epoch": 0.9458910929973877,
      "grad_norm": 0.029755135998129845,
      "learning_rate": 0.0010170731707317073,
      "loss": 2.6619839668273926,
      "step": 11666
    },
    {
      "epoch": 0.9460532550225886,
      "grad_norm": 0.027240199968218803,
      "learning_rate": 0.001016531165311653,
      "loss": 2.6375250816345215,
      "step": 11668
    },
    {
      "epoch": 0.9462154170477897,
      "grad_norm": 0.02658885344862938,
      "learning_rate": 0.0010159891598915988,
      "loss": 2.6631078720092773,
      "step": 11670
    },
    {
      "epoch": 0.9463775790729907,
      "grad_norm": 0.03771700710058212,
      "learning_rate": 0.0010154471544715445,
      "loss": 2.700932025909424,
      "step": 11672
    },
    {
      "epoch": 0.9465397410981916,
      "grad_norm": 0.041352540254592896,
      "learning_rate": 0.0010149051490514905,
      "loss": 2.5629801750183105,
      "step": 11674
    },
    {
      "epoch": 0.9467019031233926,
      "grad_norm": 0.02359360083937645,
      "learning_rate": 0.0010143631436314363,
      "loss": 2.7032036781311035,
      "step": 11676
    },
    {
      "epoch": 0.9468640651485937,
      "grad_norm": 0.043586645275354385,
      "learning_rate": 0.001013821138211382,
      "loss": 2.6192097663879395,
      "step": 11678
    },
    {
      "epoch": 0.9470262271737946,
      "grad_norm": 0.037464167922735214,
      "learning_rate": 0.0010132791327913278,
      "loss": 2.6318130493164062,
      "step": 11680
    },
    {
      "epoch": 0.9471883891989956,
      "grad_norm": 0.022547483444213867,
      "learning_rate": 0.0010127371273712736,
      "loss": 2.614470958709717,
      "step": 11682
    },
    {
      "epoch": 0.9473505512241966,
      "grad_norm": 0.05329741910099983,
      "learning_rate": 0.0010121951219512193,
      "loss": 2.6168317794799805,
      "step": 11684
    },
    {
      "epoch": 0.9475127132493976,
      "grad_norm": 0.06656701117753983,
      "learning_rate": 0.0010116531165311653,
      "loss": 2.66933274269104,
      "step": 11686
    },
    {
      "epoch": 0.9476748752745986,
      "grad_norm": 0.032640937715768814,
      "learning_rate": 0.001011111111111111,
      "loss": 2.6808414459228516,
      "step": 11688
    },
    {
      "epoch": 0.9478370372997996,
      "grad_norm": 0.03168867528438568,
      "learning_rate": 0.0010105691056910568,
      "loss": 2.6114931106567383,
      "step": 11690
    },
    {
      "epoch": 0.9479991993250005,
      "grad_norm": 0.029136693105101585,
      "learning_rate": 0.0010100271002710026,
      "loss": 2.6802940368652344,
      "step": 11692
    },
    {
      "epoch": 0.9481613613502016,
      "grad_norm": 0.024958981201052666,
      "learning_rate": 0.0010094850948509483,
      "loss": 2.6384425163269043,
      "step": 11694
    },
    {
      "epoch": 0.9483235233754026,
      "grad_norm": 0.027377087622880936,
      "learning_rate": 0.001008943089430894,
      "loss": 2.5806522369384766,
      "step": 11696
    },
    {
      "epoch": 0.9484856854006035,
      "grad_norm": 0.04230114817619324,
      "learning_rate": 0.00100840108401084,
      "loss": 2.6921849250793457,
      "step": 11698
    },
    {
      "epoch": 0.9486478474258045,
      "grad_norm": 0.03444362059235573,
      "learning_rate": 0.0010078590785907858,
      "loss": 2.6943297386169434,
      "step": 11700
    },
    {
      "epoch": 0.9488100094510056,
      "grad_norm": 0.02916163206100464,
      "learning_rate": 0.0010073170731707316,
      "loss": 2.648270606994629,
      "step": 11702
    },
    {
      "epoch": 0.9489721714762065,
      "grad_norm": 0.03239402920007706,
      "learning_rate": 0.0010067750677506774,
      "loss": 2.6446120738983154,
      "step": 11704
    },
    {
      "epoch": 0.9491343335014075,
      "grad_norm": 0.037064336240291595,
      "learning_rate": 0.0010062330623306231,
      "loss": 2.6299662590026855,
      "step": 11706
    },
    {
      "epoch": 0.9492964955266086,
      "grad_norm": 0.035664964467287064,
      "learning_rate": 0.0010056910569105689,
      "loss": 2.6518402099609375,
      "step": 11708
    },
    {
      "epoch": 0.9494586575518095,
      "grad_norm": 0.023350581526756287,
      "learning_rate": 0.0010051490514905149,
      "loss": 2.6779096126556396,
      "step": 11710
    },
    {
      "epoch": 0.9496208195770105,
      "grad_norm": 0.024368811398744583,
      "learning_rate": 0.0010046070460704606,
      "loss": 2.7483081817626953,
      "step": 11712
    },
    {
      "epoch": 0.9497829816022115,
      "grad_norm": 0.024200009182095528,
      "learning_rate": 0.0010040650406504064,
      "loss": 2.671909809112549,
      "step": 11714
    },
    {
      "epoch": 0.9499451436274124,
      "grad_norm": 0.03023422881960869,
      "learning_rate": 0.0010035230352303521,
      "loss": 2.615020751953125,
      "step": 11716
    },
    {
      "epoch": 0.9501073056526135,
      "grad_norm": 0.024164482951164246,
      "learning_rate": 0.001002981029810298,
      "loss": 2.637117624282837,
      "step": 11718
    },
    {
      "epoch": 0.9502694676778145,
      "grad_norm": 0.03033696487545967,
      "learning_rate": 0.0010024390243902437,
      "loss": 2.603252410888672,
      "step": 11720
    },
    {
      "epoch": 0.9504316297030154,
      "grad_norm": 0.025360004976391792,
      "learning_rate": 0.0010018970189701896,
      "loss": 2.653348922729492,
      "step": 11722
    },
    {
      "epoch": 0.9505937917282165,
      "grad_norm": 0.029511261731386185,
      "learning_rate": 0.0010013550135501354,
      "loss": 2.685267448425293,
      "step": 11724
    },
    {
      "epoch": 0.9507559537534175,
      "grad_norm": 0.04643676057457924,
      "learning_rate": 0.0010008130081300812,
      "loss": 2.670759916305542,
      "step": 11726
    },
    {
      "epoch": 0.9509181157786184,
      "grad_norm": 0.049367506057024,
      "learning_rate": 0.001000271002710027,
      "loss": 2.65855073928833,
      "step": 11728
    },
    {
      "epoch": 0.9510802778038194,
      "grad_norm": 0.04136735945940018,
      "learning_rate": 0.0009997289972899727,
      "loss": 2.6550416946411133,
      "step": 11730
    },
    {
      "epoch": 0.9512424398290205,
      "grad_norm": 0.02882442995905876,
      "learning_rate": 0.0009991869918699184,
      "loss": 2.672483444213867,
      "step": 11732
    },
    {
      "epoch": 0.9514046018542214,
      "grad_norm": 0.027472665533423424,
      "learning_rate": 0.0009986449864498644,
      "loss": 2.6691932678222656,
      "step": 11734
    },
    {
      "epoch": 0.9515667638794224,
      "grad_norm": 0.029464086517691612,
      "learning_rate": 0.0009981029810298102,
      "loss": 2.6595473289489746,
      "step": 11736
    },
    {
      "epoch": 0.9517289259046234,
      "grad_norm": 0.033297713845968246,
      "learning_rate": 0.000997560975609756,
      "loss": 2.625911235809326,
      "step": 11738
    },
    {
      "epoch": 0.9518910879298244,
      "grad_norm": 0.03160769119858742,
      "learning_rate": 0.0009970189701897017,
      "loss": 2.6989731788635254,
      "step": 11740
    },
    {
      "epoch": 0.9520532499550254,
      "grad_norm": 0.024521782994270325,
      "learning_rate": 0.0009964769647696475,
      "loss": 2.6754150390625,
      "step": 11742
    },
    {
      "epoch": 0.9522154119802264,
      "grad_norm": 0.030599527060985565,
      "learning_rate": 0.0009959349593495934,
      "loss": 2.6461949348449707,
      "step": 11744
    },
    {
      "epoch": 0.9523775740054273,
      "grad_norm": 0.03869258239865303,
      "learning_rate": 0.0009953929539295392,
      "loss": 2.6649866104125977,
      "step": 11746
    },
    {
      "epoch": 0.9525397360306284,
      "grad_norm": 0.02838987484574318,
      "learning_rate": 0.000994850948509485,
      "loss": 2.6303651332855225,
      "step": 11748
    },
    {
      "epoch": 0.9527018980558294,
      "grad_norm": 0.041621144860982895,
      "learning_rate": 0.0009943089430894307,
      "loss": 2.6172142028808594,
      "step": 11750
    },
    {
      "epoch": 0.9528640600810303,
      "grad_norm": 0.023762565106153488,
      "learning_rate": 0.0009937669376693765,
      "loss": 2.6129629611968994,
      "step": 11752
    },
    {
      "epoch": 0.9530262221062313,
      "grad_norm": 0.024753322824835777,
      "learning_rate": 0.0009932249322493225,
      "loss": 2.64548921585083,
      "step": 11754
    },
    {
      "epoch": 0.9531883841314324,
      "grad_norm": 0.0380735881626606,
      "learning_rate": 0.0009926829268292682,
      "loss": 2.6865618228912354,
      "step": 11756
    },
    {
      "epoch": 0.9533505461566333,
      "grad_norm": 0.03862697631120682,
      "learning_rate": 0.000992140921409214,
      "loss": 2.5714142322540283,
      "step": 11758
    },
    {
      "epoch": 0.9535127081818343,
      "grad_norm": 0.0731125921010971,
      "learning_rate": 0.0009915989159891597,
      "loss": 2.6509954929351807,
      "step": 11760
    },
    {
      "epoch": 0.9536748702070353,
      "grad_norm": 0.04156839847564697,
      "learning_rate": 0.0009910569105691055,
      "loss": 2.6367545127868652,
      "step": 11762
    },
    {
      "epoch": 0.9538370322322363,
      "grad_norm": 0.027050750330090523,
      "learning_rate": 0.0009905149051490513,
      "loss": 2.7308993339538574,
      "step": 11764
    },
    {
      "epoch": 0.9539991942574373,
      "grad_norm": 0.052043382078409195,
      "learning_rate": 0.0009899728997289972,
      "loss": 2.6401166915893555,
      "step": 11766
    },
    {
      "epoch": 0.9541613562826383,
      "grad_norm": 0.04337325319647789,
      "learning_rate": 0.000989430894308943,
      "loss": 2.657461643218994,
      "step": 11768
    },
    {
      "epoch": 0.9543235183078392,
      "grad_norm": 0.023907558992505074,
      "learning_rate": 0.0009888888888888888,
      "loss": 2.6755900382995605,
      "step": 11770
    },
    {
      "epoch": 0.9544856803330403,
      "grad_norm": 0.04709495231509209,
      "learning_rate": 0.0009883468834688345,
      "loss": 2.697974443435669,
      "step": 11772
    },
    {
      "epoch": 0.9546478423582413,
      "grad_norm": 0.0338531993329525,
      "learning_rate": 0.0009878048780487803,
      "loss": 2.657487630844116,
      "step": 11774
    },
    {
      "epoch": 0.9548100043834422,
      "grad_norm": 0.02688683569431305,
      "learning_rate": 0.0009872628726287263,
      "loss": 2.656116485595703,
      "step": 11776
    },
    {
      "epoch": 0.9549721664086432,
      "grad_norm": 0.030405644327402115,
      "learning_rate": 0.000986720867208672,
      "loss": 2.6771984100341797,
      "step": 11778
    },
    {
      "epoch": 0.9551343284338443,
      "grad_norm": 0.037452176213264465,
      "learning_rate": 0.0009861788617886178,
      "loss": 2.7193117141723633,
      "step": 11780
    },
    {
      "epoch": 0.9552964904590452,
      "grad_norm": 0.03804720565676689,
      "learning_rate": 0.0009856368563685635,
      "loss": 2.621739387512207,
      "step": 11782
    },
    {
      "epoch": 0.9554586524842462,
      "grad_norm": 0.04657045379281044,
      "learning_rate": 0.0009850948509485093,
      "loss": 2.587993860244751,
      "step": 11784
    },
    {
      "epoch": 0.9556208145094472,
      "grad_norm": 0.05484028905630112,
      "learning_rate": 0.0009845528455284553,
      "loss": 2.66007661819458,
      "step": 11786
    },
    {
      "epoch": 0.9557829765346482,
      "grad_norm": 0.037868089973926544,
      "learning_rate": 0.000984010840108401,
      "loss": 2.670085906982422,
      "step": 11788
    },
    {
      "epoch": 0.9559451385598492,
      "grad_norm": 0.031021788716316223,
      "learning_rate": 0.0009834688346883468,
      "loss": 2.6927924156188965,
      "step": 11790
    },
    {
      "epoch": 0.9561073005850502,
      "grad_norm": 0.04280811920762062,
      "learning_rate": 0.0009829268292682926,
      "loss": 2.6861839294433594,
      "step": 11792
    },
    {
      "epoch": 0.9562694626102511,
      "grad_norm": 0.052859578281641006,
      "learning_rate": 0.0009823848238482383,
      "loss": 2.635129451751709,
      "step": 11794
    },
    {
      "epoch": 0.9564316246354522,
      "grad_norm": 0.029154622927308083,
      "learning_rate": 0.000981842818428184,
      "loss": 2.6058030128479004,
      "step": 11796
    },
    {
      "epoch": 0.9565937866606532,
      "grad_norm": 0.027220504358410835,
      "learning_rate": 0.00098130081300813,
      "loss": 2.7136621475219727,
      "step": 11798
    },
    {
      "epoch": 0.9567559486858541,
      "grad_norm": 0.02626686543226242,
      "learning_rate": 0.0009807588075880758,
      "loss": 2.701012134552002,
      "step": 11800
    },
    {
      "epoch": 0.9569181107110551,
      "grad_norm": 0.02638453245162964,
      "learning_rate": 0.0009802168021680216,
      "loss": 2.62009859085083,
      "step": 11802
    },
    {
      "epoch": 0.9570802727362562,
      "grad_norm": 0.02401050180196762,
      "learning_rate": 0.0009796747967479673,
      "loss": 2.61207914352417,
      "step": 11804
    },
    {
      "epoch": 0.9572424347614571,
      "grad_norm": 0.026039760559797287,
      "learning_rate": 0.000979132791327913,
      "loss": 2.6466574668884277,
      "step": 11806
    },
    {
      "epoch": 0.9574045967866581,
      "grad_norm": 0.025333762168884277,
      "learning_rate": 0.000978590785907859,
      "loss": 2.6320078372955322,
      "step": 11808
    },
    {
      "epoch": 0.9575667588118592,
      "grad_norm": 0.030782412737607956,
      "learning_rate": 0.0009780487804878048,
      "loss": 2.6582982540130615,
      "step": 11810
    },
    {
      "epoch": 0.9577289208370601,
      "grad_norm": 0.026400597766041756,
      "learning_rate": 0.0009775067750677506,
      "loss": 2.637014389038086,
      "step": 11812
    },
    {
      "epoch": 0.9578910828622611,
      "grad_norm": 0.04124266281723976,
      "learning_rate": 0.0009769647696476964,
      "loss": 2.603046417236328,
      "step": 11814
    },
    {
      "epoch": 0.9580532448874621,
      "grad_norm": 0.02405397780239582,
      "learning_rate": 0.0009764227642276421,
      "loss": 2.650878429412842,
      "step": 11816
    },
    {
      "epoch": 0.958215406912663,
      "grad_norm": 0.03763377666473389,
      "learning_rate": 0.0009758807588075881,
      "loss": 2.6674389839172363,
      "step": 11818
    },
    {
      "epoch": 0.9583775689378641,
      "grad_norm": 0.027452297508716583,
      "learning_rate": 0.0009753387533875339,
      "loss": 2.641500234603882,
      "step": 11820
    },
    {
      "epoch": 0.9585397309630651,
      "grad_norm": 0.028634564951062202,
      "learning_rate": 0.0009747967479674796,
      "loss": 2.655819892883301,
      "step": 11822
    },
    {
      "epoch": 0.958701892988266,
      "grad_norm": 0.048093412071466446,
      "learning_rate": 0.0009742547425474254,
      "loss": 2.6044178009033203,
      "step": 11824
    },
    {
      "epoch": 0.958864055013467,
      "grad_norm": 0.03527548536658287,
      "learning_rate": 0.0009737127371273711,
      "loss": 2.6453475952148438,
      "step": 11826
    },
    {
      "epoch": 0.9590262170386681,
      "grad_norm": 0.033977147191762924,
      "learning_rate": 0.000973170731707317,
      "loss": 2.639394998550415,
      "step": 11828
    },
    {
      "epoch": 0.959188379063869,
      "grad_norm": 0.02486306056380272,
      "learning_rate": 0.0009726287262872629,
      "loss": 2.662121295928955,
      "step": 11830
    },
    {
      "epoch": 0.95935054108907,
      "grad_norm": 0.028659474104642868,
      "learning_rate": 0.0009720867208672086,
      "loss": 2.7046055793762207,
      "step": 11832
    },
    {
      "epoch": 0.9595127031142711,
      "grad_norm": 0.0329878069460392,
      "learning_rate": 0.0009715447154471544,
      "loss": 2.657451629638672,
      "step": 11834
    },
    {
      "epoch": 0.959674865139472,
      "grad_norm": 0.03125752881169319,
      "learning_rate": 0.0009710027100271002,
      "loss": 2.6709532737731934,
      "step": 11836
    },
    {
      "epoch": 0.959837027164673,
      "grad_norm": 0.026411738246679306,
      "learning_rate": 0.0009704607046070459,
      "loss": 2.674919843673706,
      "step": 11838
    },
    {
      "epoch": 0.959999189189874,
      "grad_norm": 0.025471782311797142,
      "learning_rate": 0.0009699186991869918,
      "loss": 2.5770578384399414,
      "step": 11840
    },
    {
      "epoch": 0.960161351215075,
      "grad_norm": 0.033635254949331284,
      "learning_rate": 0.0009693766937669377,
      "loss": 2.6480765342712402,
      "step": 11842
    },
    {
      "epoch": 0.960323513240276,
      "grad_norm": 0.052245739847421646,
      "learning_rate": 0.0009688346883468834,
      "loss": 2.676464557647705,
      "step": 11844
    },
    {
      "epoch": 0.960485675265477,
      "grad_norm": 0.029014185070991516,
      "learning_rate": 0.0009682926829268292,
      "loss": 2.6774439811706543,
      "step": 11846
    },
    {
      "epoch": 0.9606478372906779,
      "grad_norm": 0.03140914440155029,
      "learning_rate": 0.0009677506775067749,
      "loss": 2.7019691467285156,
      "step": 11848
    },
    {
      "epoch": 0.960809999315879,
      "grad_norm": 0.03375516086816788,
      "learning_rate": 0.0009672086720867208,
      "loss": 2.698782444000244,
      "step": 11850
    },
    {
      "epoch": 0.96097216134108,
      "grad_norm": 0.045352742075920105,
      "learning_rate": 0.0009666666666666666,
      "loss": 2.6840591430664062,
      "step": 11852
    },
    {
      "epoch": 0.9611343233662809,
      "grad_norm": 0.054348140954971313,
      "learning_rate": 0.0009661246612466124,
      "loss": 2.6311843395233154,
      "step": 11854
    },
    {
      "epoch": 0.9612964853914819,
      "grad_norm": 0.023855913430452347,
      "learning_rate": 0.0009655826558265582,
      "loss": 2.707765579223633,
      "step": 11856
    },
    {
      "epoch": 0.961458647416683,
      "grad_norm": 0.057242125272750854,
      "learning_rate": 0.000965040650406504,
      "loss": 2.6685290336608887,
      "step": 11858
    },
    {
      "epoch": 0.9616208094418839,
      "grad_norm": 0.07150266319513321,
      "learning_rate": 0.0009644986449864498,
      "loss": 2.6124258041381836,
      "step": 11860
    },
    {
      "epoch": 0.9617829714670849,
      "grad_norm": 0.023973504081368446,
      "learning_rate": 0.0009639566395663956,
      "loss": 2.7110815048217773,
      "step": 11862
    },
    {
      "epoch": 0.9619451334922859,
      "grad_norm": 0.0726916715502739,
      "learning_rate": 0.0009634146341463413,
      "loss": 2.66195011138916,
      "step": 11864
    },
    {
      "epoch": 0.9621072955174869,
      "grad_norm": 0.0382252000272274,
      "learning_rate": 0.0009628726287262872,
      "loss": 2.6479153633117676,
      "step": 11866
    },
    {
      "epoch": 0.9622694575426879,
      "grad_norm": 0.0339784100651741,
      "learning_rate": 0.000962330623306233,
      "loss": 2.6003074645996094,
      "step": 11868
    },
    {
      "epoch": 0.9624316195678889,
      "grad_norm": 0.057332418859004974,
      "learning_rate": 0.0009617886178861787,
      "loss": 2.6520113945007324,
      "step": 11870
    },
    {
      "epoch": 0.9625937815930898,
      "grad_norm": 0.0280393585562706,
      "learning_rate": 0.0009612466124661246,
      "loss": 2.704562187194824,
      "step": 11872
    },
    {
      "epoch": 0.9627559436182909,
      "grad_norm": 0.0387326255440712,
      "learning_rate": 0.0009607046070460704,
      "loss": 2.7253150939941406,
      "step": 11874
    },
    {
      "epoch": 0.9629181056434919,
      "grad_norm": 0.05075753107666969,
      "learning_rate": 0.0009601626016260161,
      "loss": 2.677001953125,
      "step": 11876
    },
    {
      "epoch": 0.9630802676686928,
      "grad_norm": 0.02409222722053528,
      "learning_rate": 0.000959620596205962,
      "loss": 2.5621187686920166,
      "step": 11878
    },
    {
      "epoch": 0.9632424296938938,
      "grad_norm": 0.04846561327576637,
      "learning_rate": 0.0009590785907859078,
      "loss": 2.6450119018554688,
      "step": 11880
    },
    {
      "epoch": 0.9634045917190949,
      "grad_norm": 0.04478650167584419,
      "learning_rate": 0.0009585365853658536,
      "loss": 2.6208691596984863,
      "step": 11882
    },
    {
      "epoch": 0.9635667537442958,
      "grad_norm": 0.02606899105012417,
      "learning_rate": 0.0009579945799457994,
      "loss": 2.6977295875549316,
      "step": 11884
    },
    {
      "epoch": 0.9637289157694968,
      "grad_norm": 0.04646744579076767,
      "learning_rate": 0.0009574525745257451,
      "loss": 2.622915744781494,
      "step": 11886
    },
    {
      "epoch": 0.9638910777946978,
      "grad_norm": 0.03357356786727905,
      "learning_rate": 0.0009569105691056909,
      "loss": 2.635761260986328,
      "step": 11888
    },
    {
      "epoch": 0.9640532398198988,
      "grad_norm": 0.02407718263566494,
      "learning_rate": 0.0009563685636856368,
      "loss": 2.641617774963379,
      "step": 11890
    },
    {
      "epoch": 0.9642154018450998,
      "grad_norm": 0.04928215965628624,
      "learning_rate": 0.0009558265582655826,
      "loss": 2.628239154815674,
      "step": 11892
    },
    {
      "epoch": 0.9643775638703008,
      "grad_norm": 0.03777390718460083,
      "learning_rate": 0.0009552845528455284,
      "loss": 2.677643060684204,
      "step": 11894
    },
    {
      "epoch": 0.9645397258955017,
      "grad_norm": 0.024609657004475594,
      "learning_rate": 0.0009547425474254742,
      "loss": 2.6571528911590576,
      "step": 11896
    },
    {
      "epoch": 0.9647018879207028,
      "grad_norm": 0.04972552880644798,
      "learning_rate": 0.0009542005420054199,
      "loss": 2.621786117553711,
      "step": 11898
    },
    {
      "epoch": 0.9648640499459038,
      "grad_norm": 0.06721260398626328,
      "learning_rate": 0.0009536585365853657,
      "loss": 2.599827766418457,
      "step": 11900
    },
    {
      "epoch": 0.9650262119711047,
      "grad_norm": 0.03228020295500755,
      "learning_rate": 0.0009531165311653116,
      "loss": 2.6584720611572266,
      "step": 11902
    },
    {
      "epoch": 0.9651883739963057,
      "grad_norm": 0.03865686058998108,
      "learning_rate": 0.0009525745257452574,
      "loss": 2.6839165687561035,
      "step": 11904
    },
    {
      "epoch": 0.9653505360215068,
      "grad_norm": 0.030838442966341972,
      "learning_rate": 0.0009520325203252032,
      "loss": 2.628452777862549,
      "step": 11906
    },
    {
      "epoch": 0.9655126980467077,
      "grad_norm": 0.0284223984926939,
      "learning_rate": 0.0009514905149051489,
      "loss": 2.6301636695861816,
      "step": 11908
    },
    {
      "epoch": 0.9656748600719087,
      "grad_norm": 0.03460818529129028,
      "learning_rate": 0.0009509485094850947,
      "loss": 2.624372959136963,
      "step": 11910
    },
    {
      "epoch": 0.9658370220971098,
      "grad_norm": 0.025938067585229874,
      "learning_rate": 0.0009504065040650405,
      "loss": 2.65023136138916,
      "step": 11912
    },
    {
      "epoch": 0.9659991841223107,
      "grad_norm": 0.0251492690294981,
      "learning_rate": 0.0009498644986449864,
      "loss": 2.6755423545837402,
      "step": 11914
    },
    {
      "epoch": 0.9661613461475117,
      "grad_norm": 0.03998899832367897,
      "learning_rate": 0.0009493224932249322,
      "loss": 2.628323554992676,
      "step": 11916
    },
    {
      "epoch": 0.9663235081727127,
      "grad_norm": 0.02502640336751938,
      "learning_rate": 0.000948780487804878,
      "loss": 2.6816463470458984,
      "step": 11918
    },
    {
      "epoch": 0.9664856701979136,
      "grad_norm": 0.03462338447570801,
      "learning_rate": 0.0009482384823848237,
      "loss": 2.704057216644287,
      "step": 11920
    },
    {
      "epoch": 0.9666478322231147,
      "grad_norm": 0.03807378187775612,
      "learning_rate": 0.0009476964769647695,
      "loss": 2.6690948009490967,
      "step": 11922
    },
    {
      "epoch": 0.9668099942483157,
      "grad_norm": 0.035117149353027344,
      "learning_rate": 0.0009471544715447155,
      "loss": 2.658895969390869,
      "step": 11924
    },
    {
      "epoch": 0.9669721562735166,
      "grad_norm": 0.04008201137185097,
      "learning_rate": 0.0009466124661246612,
      "loss": 2.6397290229797363,
      "step": 11926
    },
    {
      "epoch": 0.9671343182987177,
      "grad_norm": 0.04440319538116455,
      "learning_rate": 0.000946070460704607,
      "loss": 2.642482280731201,
      "step": 11928
    },
    {
      "epoch": 0.9672964803239187,
      "grad_norm": 0.029505418613553047,
      "learning_rate": 0.0009455284552845527,
      "loss": 2.639111280441284,
      "step": 11930
    },
    {
      "epoch": 0.9674586423491196,
      "grad_norm": 0.050583481788635254,
      "learning_rate": 0.0009449864498644985,
      "loss": 2.657057285308838,
      "step": 11932
    },
    {
      "epoch": 0.9676208043743206,
      "grad_norm": 0.02424854040145874,
      "learning_rate": 0.0009444444444444445,
      "loss": 2.691380500793457,
      "step": 11934
    },
    {
      "epoch": 0.9677829663995217,
      "grad_norm": 0.05768221989274025,
      "learning_rate": 0.0009439024390243902,
      "loss": 2.626725196838379,
      "step": 11936
    },
    {
      "epoch": 0.9679451284247226,
      "grad_norm": 0.05091079697012901,
      "learning_rate": 0.000943360433604336,
      "loss": 2.67665433883667,
      "step": 11938
    },
    {
      "epoch": 0.9681072904499236,
      "grad_norm": 0.026060421019792557,
      "learning_rate": 0.0009428184281842818,
      "loss": 2.6399755477905273,
      "step": 11940
    },
    {
      "epoch": 0.9682694524751246,
      "grad_norm": 0.04445699602365494,
      "learning_rate": 0.0009422764227642275,
      "loss": 2.6134753227233887,
      "step": 11942
    },
    {
      "epoch": 0.9684316145003256,
      "grad_norm": 0.07600291818380356,
      "learning_rate": 0.0009417344173441733,
      "loss": 2.7113070487976074,
      "step": 11944
    },
    {
      "epoch": 0.9685937765255266,
      "grad_norm": 0.02674132026731968,
      "learning_rate": 0.0009411924119241193,
      "loss": 2.63482666015625,
      "step": 11946
    },
    {
      "epoch": 0.9687559385507276,
      "grad_norm": 0.053147830069065094,
      "learning_rate": 0.000940650406504065,
      "loss": 2.664078712463379,
      "step": 11948
    },
    {
      "epoch": 0.9689181005759285,
      "grad_norm": 0.0333804190158844,
      "learning_rate": 0.0009401084010840108,
      "loss": 2.6305246353149414,
      "step": 11950
    },
    {
      "epoch": 0.9690802626011296,
      "grad_norm": 0.06156730279326439,
      "learning_rate": 0.0009395663956639565,
      "loss": 2.670435905456543,
      "step": 11952
    },
    {
      "epoch": 0.9692424246263306,
      "grad_norm": 0.05770261213183403,
      "learning_rate": 0.0009390243902439023,
      "loss": 2.6684648990631104,
      "step": 11954
    },
    {
      "epoch": 0.9694045866515315,
      "grad_norm": 0.03796371817588806,
      "learning_rate": 0.0009384823848238482,
      "loss": 2.604991912841797,
      "step": 11956
    },
    {
      "epoch": 0.9695667486767325,
      "grad_norm": 0.038690924644470215,
      "learning_rate": 0.000937940379403794,
      "loss": 2.73984956741333,
      "step": 11958
    },
    {
      "epoch": 0.9697289107019336,
      "grad_norm": 0.02918170392513275,
      "learning_rate": 0.0009373983739837398,
      "loss": 2.6938881874084473,
      "step": 11960
    },
    {
      "epoch": 0.9698910727271345,
      "grad_norm": 0.03140373155474663,
      "learning_rate": 0.0009368563685636856,
      "loss": 2.6210460662841797,
      "step": 11962
    },
    {
      "epoch": 0.9700532347523355,
      "grad_norm": 0.02435198426246643,
      "learning_rate": 0.0009363143631436313,
      "loss": 2.6496567726135254,
      "step": 11964
    },
    {
      "epoch": 0.9702153967775365,
      "grad_norm": 0.03760472312569618,
      "learning_rate": 0.0009357723577235772,
      "loss": 2.6525354385375977,
      "step": 11966
    },
    {
      "epoch": 0.9703775588027375,
      "grad_norm": 0.024870168417692184,
      "learning_rate": 0.0009352303523035229,
      "loss": 2.6173155307769775,
      "step": 11968
    },
    {
      "epoch": 0.9705397208279385,
      "grad_norm": 0.06035342440009117,
      "learning_rate": 0.0009346883468834688,
      "loss": 2.6556637287139893,
      "step": 11970
    },
    {
      "epoch": 0.9707018828531395,
      "grad_norm": 0.06191597133874893,
      "learning_rate": 0.0009341463414634146,
      "loss": 2.6544790267944336,
      "step": 11972
    },
    {
      "epoch": 0.9708640448783404,
      "grad_norm": 0.03675086051225662,
      "learning_rate": 0.0009336043360433603,
      "loss": 2.6109466552734375,
      "step": 11974
    },
    {
      "epoch": 0.9710262069035415,
      "grad_norm": 0.05534983426332474,
      "learning_rate": 0.0009330623306233061,
      "loss": 2.6806812286376953,
      "step": 11976
    },
    {
      "epoch": 0.9711883689287425,
      "grad_norm": 0.026701537892222404,
      "learning_rate": 0.000932520325203252,
      "loss": 2.692368507385254,
      "step": 11978
    },
    {
      "epoch": 0.9713505309539434,
      "grad_norm": 0.047607339918613434,
      "learning_rate": 0.0009319783197831977,
      "loss": 2.738640308380127,
      "step": 11980
    },
    {
      "epoch": 0.9715126929791444,
      "grad_norm": 0.03178601339459419,
      "learning_rate": 0.0009314363143631436,
      "loss": 2.627440929412842,
      "step": 11982
    },
    {
      "epoch": 0.9716748550043455,
      "grad_norm": 0.03143298253417015,
      "learning_rate": 0.0009308943089430894,
      "loss": 2.697681188583374,
      "step": 11984
    },
    {
      "epoch": 0.9718370170295464,
      "grad_norm": 0.02996136248111725,
      "learning_rate": 0.0009303523035230351,
      "loss": 2.615492820739746,
      "step": 11986
    },
    {
      "epoch": 0.9719991790547474,
      "grad_norm": 0.024934392422437668,
      "learning_rate": 0.000929810298102981,
      "loss": 2.6512880325317383,
      "step": 11988
    },
    {
      "epoch": 0.9721613410799484,
      "grad_norm": 0.02756170742213726,
      "learning_rate": 0.0009292682926829267,
      "loss": 2.6098380088806152,
      "step": 11990
    },
    {
      "epoch": 0.9723235031051494,
      "grad_norm": 0.02706162817776203,
      "learning_rate": 0.0009287262872628725,
      "loss": 2.6742539405822754,
      "step": 11992
    },
    {
      "epoch": 0.9724856651303504,
      "grad_norm": 0.025933949276804924,
      "learning_rate": 0.0009281842818428184,
      "loss": 2.6413800716400146,
      "step": 11994
    },
    {
      "epoch": 0.9726478271555514,
      "grad_norm": 0.02501506172120571,
      "learning_rate": 0.0009276422764227641,
      "loss": 2.6891303062438965,
      "step": 11996
    },
    {
      "epoch": 0.9728099891807523,
      "grad_norm": 0.030168281868100166,
      "learning_rate": 0.00092710027100271,
      "loss": 2.658665657043457,
      "step": 11998
    },
    {
      "epoch": 0.9729721512059534,
      "grad_norm": 0.03321559354662895,
      "learning_rate": 0.0009265582655826558,
      "loss": 2.710130214691162,
      "step": 12000
    },
    {
      "epoch": 0.9731343132311544,
      "grad_norm": 0.025363465771079063,
      "learning_rate": 0.0009260162601626015,
      "loss": 2.7103629112243652,
      "step": 12002
    },
    {
      "epoch": 0.9732964752563553,
      "grad_norm": 0.02746608480811119,
      "learning_rate": 0.0009254742547425473,
      "loss": 2.5989954471588135,
      "step": 12004
    },
    {
      "epoch": 0.9734586372815563,
      "grad_norm": 0.02576526440680027,
      "learning_rate": 0.0009249322493224932,
      "loss": 2.6496827602386475,
      "step": 12006
    },
    {
      "epoch": 0.9736207993067574,
      "grad_norm": 0.02692592516541481,
      "learning_rate": 0.000924390243902439,
      "loss": 2.6470718383789062,
      "step": 12008
    },
    {
      "epoch": 0.9737829613319583,
      "grad_norm": 0.03178763389587402,
      "learning_rate": 0.0009238482384823848,
      "loss": 2.6843631267547607,
      "step": 12010
    },
    {
      "epoch": 0.9739451233571593,
      "grad_norm": 0.027791732922196388,
      "learning_rate": 0.0009233062330623305,
      "loss": 2.6422348022460938,
      "step": 12012
    },
    {
      "epoch": 0.9741072853823604,
      "grad_norm": 0.03540303185582161,
      "learning_rate": 0.0009227642276422763,
      "loss": 2.666076183319092,
      "step": 12014
    },
    {
      "epoch": 0.9742694474075613,
      "grad_norm": 0.04687872529029846,
      "learning_rate": 0.0009222222222222221,
      "loss": 2.66068172454834,
      "step": 12016
    },
    {
      "epoch": 0.9744316094327623,
      "grad_norm": 0.033265452831983566,
      "learning_rate": 0.0009216802168021679,
      "loss": 2.6432676315307617,
      "step": 12018
    },
    {
      "epoch": 0.9745937714579633,
      "grad_norm": 0.02782902494072914,
      "learning_rate": 0.0009211382113821138,
      "loss": 2.6281869411468506,
      "step": 12020
    },
    {
      "epoch": 0.9747559334831642,
      "grad_norm": 0.024660013616085052,
      "learning_rate": 0.0009205962059620596,
      "loss": 2.680083751678467,
      "step": 12022
    },
    {
      "epoch": 0.9749180955083653,
      "grad_norm": 0.028825970366597176,
      "learning_rate": 0.0009200542005420053,
      "loss": 2.5965185165405273,
      "step": 12024
    },
    {
      "epoch": 0.9750802575335663,
      "grad_norm": 0.03849039599299431,
      "learning_rate": 0.0009195121951219511,
      "loss": 2.7024765014648438,
      "step": 12026
    },
    {
      "epoch": 0.9752424195587672,
      "grad_norm": 0.04156889393925667,
      "learning_rate": 0.0009189701897018968,
      "loss": 2.642930746078491,
      "step": 12028
    },
    {
      "epoch": 0.9754045815839683,
      "grad_norm": 0.028087330982089043,
      "learning_rate": 0.0009184281842818428,
      "loss": 2.662599563598633,
      "step": 12030
    },
    {
      "epoch": 0.9755667436091693,
      "grad_norm": 0.04252581670880318,
      "learning_rate": 0.0009178861788617886,
      "loss": 2.6728830337524414,
      "step": 12032
    },
    {
      "epoch": 0.9757289056343702,
      "grad_norm": 0.02952791005373001,
      "learning_rate": 0.0009173441734417343,
      "loss": 2.686067581176758,
      "step": 12034
    },
    {
      "epoch": 0.9758910676595712,
      "grad_norm": 0.029036428779363632,
      "learning_rate": 0.0009168021680216801,
      "loss": 2.6373748779296875,
      "step": 12036
    },
    {
      "epoch": 0.9760532296847723,
      "grad_norm": 0.04985431581735611,
      "learning_rate": 0.0009162601626016259,
      "loss": 2.6439597606658936,
      "step": 12038
    },
    {
      "epoch": 0.9762153917099732,
      "grad_norm": 0.03348122164607048,
      "learning_rate": 0.0009157181571815718,
      "loss": 2.6734137535095215,
      "step": 12040
    },
    {
      "epoch": 0.9763775537351742,
      "grad_norm": 0.027866648510098457,
      "learning_rate": 0.0009151761517615176,
      "loss": 2.642733573913574,
      "step": 12042
    },
    {
      "epoch": 0.9765397157603752,
      "grad_norm": 0.024228988215327263,
      "learning_rate": 0.0009146341463414634,
      "loss": 2.629518985748291,
      "step": 12044
    },
    {
      "epoch": 0.9767018777855762,
      "grad_norm": 0.03679553046822548,
      "learning_rate": 0.0009140921409214091,
      "loss": 2.671339988708496,
      "step": 12046
    },
    {
      "epoch": 0.9768640398107772,
      "grad_norm": 0.027245361357927322,
      "learning_rate": 0.0009135501355013549,
      "loss": 2.6714630126953125,
      "step": 12048
    },
    {
      "epoch": 0.9770262018359782,
      "grad_norm": 0.032815683633089066,
      "learning_rate": 0.0009130081300813006,
      "loss": 2.6631462574005127,
      "step": 12050
    },
    {
      "epoch": 0.9771883638611791,
      "grad_norm": 0.039507944136857986,
      "learning_rate": 0.0009124661246612466,
      "loss": 2.625303030014038,
      "step": 12052
    },
    {
      "epoch": 0.9773505258863802,
      "grad_norm": 0.030851153656840324,
      "learning_rate": 0.0009119241192411924,
      "loss": 2.591279983520508,
      "step": 12054
    },
    {
      "epoch": 0.9775126879115812,
      "grad_norm": 0.0262528695166111,
      "learning_rate": 0.0009113821138211381,
      "loss": 2.6321654319763184,
      "step": 12056
    },
    {
      "epoch": 0.9776748499367821,
      "grad_norm": 0.057269465178251266,
      "learning_rate": 0.0009108401084010839,
      "loss": 2.727774143218994,
      "step": 12058
    },
    {
      "epoch": 0.9778370119619831,
      "grad_norm": 0.042490337044000626,
      "learning_rate": 0.0009102981029810297,
      "loss": 2.700509548187256,
      "step": 12060
    },
    {
      "epoch": 0.9779991739871842,
      "grad_norm": 0.023489071056246758,
      "learning_rate": 0.0009097560975609756,
      "loss": 2.6958556175231934,
      "step": 12062
    },
    {
      "epoch": 0.9781613360123851,
      "grad_norm": 0.03679446503520012,
      "learning_rate": 0.0009092140921409214,
      "loss": 2.6028051376342773,
      "step": 12064
    },
    {
      "epoch": 0.9783234980375861,
      "grad_norm": 0.038280777633190155,
      "learning_rate": 0.0009086720867208672,
      "loss": 2.592534303665161,
      "step": 12066
    },
    {
      "epoch": 0.9784856600627871,
      "grad_norm": 0.041793093085289,
      "learning_rate": 0.0009081300813008129,
      "loss": 2.6781253814697266,
      "step": 12068
    },
    {
      "epoch": 0.9786478220879881,
      "grad_norm": 0.034133180975914,
      "learning_rate": 0.0009075880758807587,
      "loss": 2.6797537803649902,
      "step": 12070
    },
    {
      "epoch": 0.9788099841131891,
      "grad_norm": 0.03987056761980057,
      "learning_rate": 0.0009070460704607045,
      "loss": 2.666598320007324,
      "step": 12072
    },
    {
      "epoch": 0.9789721461383901,
      "grad_norm": 0.042468298226594925,
      "learning_rate": 0.0009065040650406504,
      "loss": 2.6670901775360107,
      "step": 12074
    },
    {
      "epoch": 0.979134308163591,
      "grad_norm": 0.023152215406298637,
      "learning_rate": 0.0009059620596205962,
      "loss": 2.6435413360595703,
      "step": 12076
    },
    {
      "epoch": 0.9792964701887921,
      "grad_norm": 0.024857165291905403,
      "learning_rate": 0.0009054200542005419,
      "loss": 2.6399102210998535,
      "step": 12078
    },
    {
      "epoch": 0.9794586322139931,
      "grad_norm": 0.02728443592786789,
      "learning_rate": 0.0009048780487804877,
      "loss": 2.6501564979553223,
      "step": 12080
    },
    {
      "epoch": 0.979620794239194,
      "grad_norm": 0.023662136867642403,
      "learning_rate": 0.0009043360433604335,
      "loss": 2.5968849658966064,
      "step": 12082
    },
    {
      "epoch": 0.979782956264395,
      "grad_norm": 0.02512369304895401,
      "learning_rate": 0.0009037940379403793,
      "loss": 2.657933235168457,
      "step": 12084
    },
    {
      "epoch": 0.9799451182895961,
      "grad_norm": 0.024104513227939606,
      "learning_rate": 0.0009032520325203252,
      "loss": 2.6116857528686523,
      "step": 12086
    },
    {
      "epoch": 0.980107280314797,
      "grad_norm": 0.028581004589796066,
      "learning_rate": 0.000902710027100271,
      "loss": 2.652723789215088,
      "step": 12088
    },
    {
      "epoch": 0.980269442339998,
      "grad_norm": 0.026918938383460045,
      "learning_rate": 0.0009021680216802167,
      "loss": 2.6698365211486816,
      "step": 12090
    },
    {
      "epoch": 0.980431604365199,
      "grad_norm": 0.02268073335289955,
      "learning_rate": 0.0009016260162601625,
      "loss": 2.627634048461914,
      "step": 12092
    },
    {
      "epoch": 0.9805937663904,
      "grad_norm": 0.025901786983013153,
      "learning_rate": 0.0009010840108401083,
      "loss": 2.6187267303466797,
      "step": 12094
    },
    {
      "epoch": 0.980755928415601,
      "grad_norm": 0.027088304981589317,
      "learning_rate": 0.0009005420054200541,
      "loss": 2.5912859439849854,
      "step": 12096
    },
    {
      "epoch": 0.980918090440802,
      "grad_norm": 0.029428737238049507,
      "learning_rate": 0.0009,
      "loss": 2.676334857940674,
      "step": 12098
    },
    {
      "epoch": 0.9810802524660029,
      "grad_norm": 0.03418608382344246,
      "learning_rate": 0.0008994579945799457,
      "loss": 2.6500635147094727,
      "step": 12100
    },
    {
      "epoch": 0.981242414491204,
      "grad_norm": 0.02333039604127407,
      "learning_rate": 0.0008989159891598915,
      "loss": 2.5996570587158203,
      "step": 12102
    },
    {
      "epoch": 0.981404576516405,
      "grad_norm": 0.023293213918805122,
      "learning_rate": 0.0008983739837398374,
      "loss": 2.677335500717163,
      "step": 12104
    },
    {
      "epoch": 0.9815667385416059,
      "grad_norm": 0.030683958902955055,
      "learning_rate": 0.0008978319783197831,
      "loss": 2.6375062465667725,
      "step": 12106
    },
    {
      "epoch": 0.981728900566807,
      "grad_norm": 0.02451406419277191,
      "learning_rate": 0.0008972899728997289,
      "loss": 2.6744842529296875,
      "step": 12108
    },
    {
      "epoch": 0.981891062592008,
      "grad_norm": 0.022752130404114723,
      "learning_rate": 0.0008967479674796746,
      "loss": 2.6335697174072266,
      "step": 12110
    },
    {
      "epoch": 0.9820532246172089,
      "grad_norm": 0.02698848955333233,
      "learning_rate": 0.0008962059620596205,
      "loss": 2.725338935852051,
      "step": 12112
    },
    {
      "epoch": 0.9822153866424099,
      "grad_norm": 0.03052990697324276,
      "learning_rate": 0.0008956639566395664,
      "loss": 2.6852469444274902,
      "step": 12114
    },
    {
      "epoch": 0.982377548667611,
      "grad_norm": 0.023782674223184586,
      "learning_rate": 0.0008951219512195121,
      "loss": 2.680678367614746,
      "step": 12116
    },
    {
      "epoch": 0.9825397106928119,
      "grad_norm": 0.026630936190485954,
      "learning_rate": 0.0008945799457994579,
      "loss": 2.6522343158721924,
      "step": 12118
    },
    {
      "epoch": 0.9827018727180129,
      "grad_norm": 0.026110054925084114,
      "learning_rate": 0.0008940379403794037,
      "loss": 2.6732304096221924,
      "step": 12120
    },
    {
      "epoch": 0.9828640347432139,
      "grad_norm": 0.0326547771692276,
      "learning_rate": 0.0008934959349593494,
      "loss": 2.63944149017334,
      "step": 12122
    },
    {
      "epoch": 0.9830261967684149,
      "grad_norm": 0.02910480462014675,
      "learning_rate": 0.0008929539295392953,
      "loss": 2.618818759918213,
      "step": 12124
    },
    {
      "epoch": 0.9831883587936159,
      "grad_norm": 0.04695800691843033,
      "learning_rate": 0.0008924119241192412,
      "loss": 2.6891674995422363,
      "step": 12126
    },
    {
      "epoch": 0.9833505208188169,
      "grad_norm": 0.038459766656160355,
      "learning_rate": 0.0008918699186991869,
      "loss": 2.6260242462158203,
      "step": 12128
    },
    {
      "epoch": 0.9835126828440178,
      "grad_norm": 0.024722333997488022,
      "learning_rate": 0.0008913279132791327,
      "loss": 2.7095532417297363,
      "step": 12130
    },
    {
      "epoch": 0.9836748448692189,
      "grad_norm": 0.02639002911746502,
      "learning_rate": 0.0008907859078590784,
      "loss": 2.6395184993743896,
      "step": 12132
    },
    {
      "epoch": 0.9838370068944199,
      "grad_norm": 0.0494588166475296,
      "learning_rate": 0.0008902439024390242,
      "loss": 2.6710517406463623,
      "step": 12134
    },
    {
      "epoch": 0.9839991689196208,
      "grad_norm": 0.026853682473301888,
      "learning_rate": 0.0008897018970189702,
      "loss": 2.636263370513916,
      "step": 12136
    },
    {
      "epoch": 0.9841613309448218,
      "grad_norm": 0.029710037633776665,
      "learning_rate": 0.0008891598915989159,
      "loss": 2.6605334281921387,
      "step": 12138
    },
    {
      "epoch": 0.9843234929700229,
      "grad_norm": 0.048928916454315186,
      "learning_rate": 0.0008886178861788617,
      "loss": 2.6403250694274902,
      "step": 12140
    },
    {
      "epoch": 0.9844856549952238,
      "grad_norm": 0.02340661734342575,
      "learning_rate": 0.0008880758807588075,
      "loss": 2.596987247467041,
      "step": 12142
    },
    {
      "epoch": 0.9846478170204248,
      "grad_norm": 0.02465207874774933,
      "learning_rate": 0.0008875338753387532,
      "loss": 2.7019360065460205,
      "step": 12144
    },
    {
      "epoch": 0.9848099790456258,
      "grad_norm": 0.035425614565610886,
      "learning_rate": 0.0008869918699186992,
      "loss": 2.636955499649048,
      "step": 12146
    },
    {
      "epoch": 0.9849721410708268,
      "grad_norm": 0.030564315617084503,
      "learning_rate": 0.000886449864498645,
      "loss": 2.6844396591186523,
      "step": 12148
    },
    {
      "epoch": 0.9851343030960278,
      "grad_norm": 0.027087440714240074,
      "learning_rate": 0.0008859078590785907,
      "loss": 2.60856294631958,
      "step": 12150
    },
    {
      "epoch": 0.9852964651212288,
      "grad_norm": 0.023967579007148743,
      "learning_rate": 0.0008853658536585365,
      "loss": 2.603853225708008,
      "step": 12152
    },
    {
      "epoch": 0.9854586271464297,
      "grad_norm": 0.05664756894111633,
      "learning_rate": 0.0008848238482384822,
      "loss": 2.6245639324188232,
      "step": 12154
    },
    {
      "epoch": 0.9856207891716308,
      "grad_norm": 0.0387977734208107,
      "learning_rate": 0.000884281842818428,
      "loss": 2.6765036582946777,
      "step": 12156
    },
    {
      "epoch": 0.9857829511968318,
      "grad_norm": 0.026337627321481705,
      "learning_rate": 0.000883739837398374,
      "loss": 2.661505699157715,
      "step": 12158
    },
    {
      "epoch": 0.9859451132220327,
      "grad_norm": 0.03982257470488548,
      "learning_rate": 0.0008831978319783197,
      "loss": 2.6524581909179688,
      "step": 12160
    },
    {
      "epoch": 0.9861072752472337,
      "grad_norm": 0.033084820955991745,
      "learning_rate": 0.0008826558265582655,
      "loss": 2.7161903381347656,
      "step": 12162
    },
    {
      "epoch": 0.9862694372724348,
      "grad_norm": 0.023122701793909073,
      "learning_rate": 0.0008821138211382113,
      "loss": 2.641139030456543,
      "step": 12164
    },
    {
      "epoch": 0.9864315992976357,
      "grad_norm": 0.024036644026637077,
      "learning_rate": 0.000881571815718157,
      "loss": 2.7374215126037598,
      "step": 12166
    },
    {
      "epoch": 0.9865937613228367,
      "grad_norm": 0.026939185336232185,
      "learning_rate": 0.000881029810298103,
      "loss": 2.6900739669799805,
      "step": 12168
    },
    {
      "epoch": 0.9867559233480377,
      "grad_norm": 0.039163414388895035,
      "learning_rate": 0.0008804878048780488,
      "loss": 2.604452610015869,
      "step": 12170
    },
    {
      "epoch": 0.9869180853732387,
      "grad_norm": 0.024264751002192497,
      "learning_rate": 0.0008799457994579945,
      "loss": 2.63358736038208,
      "step": 12172
    },
    {
      "epoch": 0.9870802473984397,
      "grad_norm": 0.0263130534440279,
      "learning_rate": 0.0008794037940379403,
      "loss": 2.6230978965759277,
      "step": 12174
    },
    {
      "epoch": 0.9872424094236407,
      "grad_norm": 0.042375314980745316,
      "learning_rate": 0.000878861788617886,
      "loss": 2.6681220531463623,
      "step": 12176
    },
    {
      "epoch": 0.9874045714488416,
      "grad_norm": 0.030201559886336327,
      "learning_rate": 0.0008783197831978319,
      "loss": 2.647793769836426,
      "step": 12178
    },
    {
      "epoch": 0.9875667334740427,
      "grad_norm": 0.025635654106736183,
      "learning_rate": 0.0008777777777777778,
      "loss": 2.636502742767334,
      "step": 12180
    },
    {
      "epoch": 0.9877288954992437,
      "grad_norm": 0.024404143914580345,
      "learning_rate": 0.0008772357723577235,
      "loss": 2.6383633613586426,
      "step": 12182
    },
    {
      "epoch": 0.9878910575244446,
      "grad_norm": 0.038048919290304184,
      "learning_rate": 0.0008766937669376693,
      "loss": 2.6132068634033203,
      "step": 12184
    },
    {
      "epoch": 0.9880532195496456,
      "grad_norm": 0.023963693529367447,
      "learning_rate": 0.0008761517615176151,
      "loss": 2.6489295959472656,
      "step": 12186
    },
    {
      "epoch": 0.9882153815748467,
      "grad_norm": 0.035370271652936935,
      "learning_rate": 0.0008756097560975608,
      "loss": 2.7031688690185547,
      "step": 12188
    },
    {
      "epoch": 0.9883775436000476,
      "grad_norm": 0.04249725863337517,
      "learning_rate": 0.0008750677506775067,
      "loss": 2.6818084716796875,
      "step": 12190
    },
    {
      "epoch": 0.9885397056252486,
      "grad_norm": 0.03524264320731163,
      "learning_rate": 0.0008745257452574526,
      "loss": 2.6581954956054688,
      "step": 12192
    },
    {
      "epoch": 0.9887018676504497,
      "grad_norm": 0.0379025973379612,
      "learning_rate": 0.0008739837398373983,
      "loss": 2.5821657180786133,
      "step": 12194
    },
    {
      "epoch": 0.9888640296756506,
      "grad_norm": 0.06050550937652588,
      "learning_rate": 0.0008734417344173441,
      "loss": 2.674360752105713,
      "step": 12196
    },
    {
      "epoch": 0.9890261917008516,
      "grad_norm": 0.0281661506742239,
      "learning_rate": 0.0008728997289972898,
      "loss": 2.714510440826416,
      "step": 12198
    },
    {
      "epoch": 0.9891883537260526,
      "grad_norm": 0.04770194739103317,
      "learning_rate": 0.0008723577235772357,
      "loss": 2.6801347732543945,
      "step": 12200
    },
    {
      "epoch": 0.9893505157512535,
      "grad_norm": 0.04433393478393555,
      "learning_rate": 0.0008718157181571815,
      "loss": 2.654851198196411,
      "step": 12202
    },
    {
      "epoch": 0.9895126777764546,
      "grad_norm": 0.022557571530342102,
      "learning_rate": 0.0008712737127371273,
      "loss": 2.7105188369750977,
      "step": 12204
    },
    {
      "epoch": 0.9896748398016556,
      "grad_norm": 0.03418521210551262,
      "learning_rate": 0.0008707317073170731,
      "loss": 2.6180520057678223,
      "step": 12206
    },
    {
      "epoch": 0.9898370018268565,
      "grad_norm": 0.022409481927752495,
      "learning_rate": 0.0008701897018970189,
      "loss": 2.6751420497894287,
      "step": 12208
    },
    {
      "epoch": 0.9899991638520576,
      "grad_norm": 0.030351024121046066,
      "learning_rate": 0.0008696476964769647,
      "loss": 2.6412782669067383,
      "step": 12210
    },
    {
      "epoch": 0.9901613258772586,
      "grad_norm": 0.044479209929704666,
      "learning_rate": 0.0008691056910569105,
      "loss": 2.6462316513061523,
      "step": 12212
    },
    {
      "epoch": 0.9903234879024595,
      "grad_norm": 0.025118645280599594,
      "learning_rate": 0.0008685636856368562,
      "loss": 2.604891300201416,
      "step": 12214
    },
    {
      "epoch": 0.9904856499276605,
      "grad_norm": 0.032186783850193024,
      "learning_rate": 0.0008680216802168021,
      "loss": 2.604489326477051,
      "step": 12216
    },
    {
      "epoch": 0.9906478119528616,
      "grad_norm": 0.051047809422016144,
      "learning_rate": 0.0008674796747967479,
      "loss": 2.666903018951416,
      "step": 12218
    },
    {
      "epoch": 0.9908099739780625,
      "grad_norm": 0.04423607513308525,
      "learning_rate": 0.0008669376693766937,
      "loss": 2.610520839691162,
      "step": 12220
    },
    {
      "epoch": 0.9909721360032635,
      "grad_norm": 0.025677509605884552,
      "learning_rate": 0.0008663956639566395,
      "loss": 2.6952476501464844,
      "step": 12222
    },
    {
      "epoch": 0.9911342980284645,
      "grad_norm": 0.02432718127965927,
      "learning_rate": 0.0008658536585365853,
      "loss": 2.647830009460449,
      "step": 12224
    },
    {
      "epoch": 0.9912964600536655,
      "grad_norm": 0.034661319106817245,
      "learning_rate": 0.000865311653116531,
      "loss": 2.649723529815674,
      "step": 12226
    },
    {
      "epoch": 0.9914586220788665,
      "grad_norm": 0.03026571311056614,
      "learning_rate": 0.0008647696476964769,
      "loss": 2.707446575164795,
      "step": 12228
    },
    {
      "epoch": 0.9916207841040675,
      "grad_norm": 0.04551996290683746,
      "learning_rate": 0.0008642276422764227,
      "loss": 2.592190742492676,
      "step": 12230
    },
    {
      "epoch": 0.9917829461292684,
      "grad_norm": 0.03670252487063408,
      "learning_rate": 0.0008636856368563685,
      "loss": 2.6369495391845703,
      "step": 12232
    },
    {
      "epoch": 0.9919451081544695,
      "grad_norm": 0.030394652858376503,
      "learning_rate": 0.0008631436314363143,
      "loss": 2.630505323410034,
      "step": 12234
    },
    {
      "epoch": 0.9921072701796705,
      "grad_norm": 0.047484587877988815,
      "learning_rate": 0.00086260162601626,
      "loss": 2.6262497901916504,
      "step": 12236
    },
    {
      "epoch": 0.9922694322048714,
      "grad_norm": 0.025303173810243607,
      "learning_rate": 0.0008620596205962058,
      "loss": 2.603376865386963,
      "step": 12238
    },
    {
      "epoch": 0.9924315942300724,
      "grad_norm": 0.029260406270623207,
      "learning_rate": 0.0008615176151761517,
      "loss": 2.784874439239502,
      "step": 12240
    },
    {
      "epoch": 0.9925937562552735,
      "grad_norm": 0.06556902080774307,
      "learning_rate": 0.0008609756097560975,
      "loss": 2.6690263748168945,
      "step": 12242
    },
    {
      "epoch": 0.9927559182804744,
      "grad_norm": 0.026567354798316956,
      "learning_rate": 0.0008604336043360433,
      "loss": 2.6511943340301514,
      "step": 12244
    },
    {
      "epoch": 0.9929180803056754,
      "grad_norm": 0.0457654669880867,
      "learning_rate": 0.0008598915989159891,
      "loss": 2.689469337463379,
      "step": 12246
    },
    {
      "epoch": 0.9930802423308764,
      "grad_norm": 0.04591895267367363,
      "learning_rate": 0.0008593495934959348,
      "loss": 2.6949784755706787,
      "step": 12248
    },
    {
      "epoch": 0.9932424043560774,
      "grad_norm": 0.021690525114536285,
      "learning_rate": 0.0008588075880758806,
      "loss": 2.6671929359436035,
      "step": 12250
    },
    {
      "epoch": 0.9934045663812784,
      "grad_norm": 0.0374511182308197,
      "learning_rate": 0.0008582655826558266,
      "loss": 2.69584321975708,
      "step": 12252
    },
    {
      "epoch": 0.9935667284064794,
      "grad_norm": 0.024455076083540916,
      "learning_rate": 0.0008577235772357723,
      "loss": 2.6354827880859375,
      "step": 12254
    },
    {
      "epoch": 0.9937288904316803,
      "grad_norm": 0.026025082916021347,
      "learning_rate": 0.0008571815718157181,
      "loss": 2.5807478427886963,
      "step": 12256
    },
    {
      "epoch": 0.9938910524568814,
      "grad_norm": 0.033368874341249466,
      "learning_rate": 0.0008566395663956638,
      "loss": 2.673483371734619,
      "step": 12258
    },
    {
      "epoch": 0.9940532144820824,
      "grad_norm": 0.02910059690475464,
      "learning_rate": 0.0008560975609756096,
      "loss": 2.684380054473877,
      "step": 12260
    },
    {
      "epoch": 0.9942153765072833,
      "grad_norm": 0.025701886042952538,
      "learning_rate": 0.0008555555555555554,
      "loss": 2.678161859512329,
      "step": 12262
    },
    {
      "epoch": 0.9943775385324843,
      "grad_norm": 0.023339081555604935,
      "learning_rate": 0.0008550135501355013,
      "loss": 2.7282960414886475,
      "step": 12264
    },
    {
      "epoch": 0.9945397005576854,
      "grad_norm": 0.0244143083691597,
      "learning_rate": 0.0008544715447154471,
      "loss": 2.660642147064209,
      "step": 12266
    },
    {
      "epoch": 0.9947018625828863,
      "grad_norm": 0.024432657286524773,
      "learning_rate": 0.0008539295392953929,
      "loss": 2.7013988494873047,
      "step": 12268
    },
    {
      "epoch": 0.9948640246080873,
      "grad_norm": 0.026502635329961777,
      "learning_rate": 0.0008533875338753386,
      "loss": 2.650359630584717,
      "step": 12270
    },
    {
      "epoch": 0.9950261866332883,
      "grad_norm": 0.028679251670837402,
      "learning_rate": 0.0008528455284552844,
      "loss": 2.7011966705322266,
      "step": 12272
    },
    {
      "epoch": 0.9951883486584893,
      "grad_norm": 0.028403136879205704,
      "learning_rate": 0.0008523035230352304,
      "loss": 2.5962352752685547,
      "step": 12274
    },
    {
      "epoch": 0.9953505106836903,
      "grad_norm": 0.031592901796102524,
      "learning_rate": 0.0008517615176151761,
      "loss": 2.753170967102051,
      "step": 12276
    },
    {
      "epoch": 0.9955126727088913,
      "grad_norm": 0.040489472448825836,
      "learning_rate": 0.0008512195121951219,
      "loss": 2.6240549087524414,
      "step": 12278
    },
    {
      "epoch": 0.9956748347340922,
      "grad_norm": 0.05993494391441345,
      "learning_rate": 0.0008506775067750676,
      "loss": 2.681548595428467,
      "step": 12280
    },
    {
      "epoch": 0.9958369967592933,
      "grad_norm": 0.05389665812253952,
      "learning_rate": 0.0008501355013550134,
      "loss": 2.716763734817505,
      "step": 12282
    },
    {
      "epoch": 0.9959991587844943,
      "grad_norm": 0.022279052063822746,
      "learning_rate": 0.0008495934959349594,
      "loss": 2.6638011932373047,
      "step": 12284
    },
    {
      "epoch": 0.9961613208096952,
      "grad_norm": 0.031833235174417496,
      "learning_rate": 0.0008490514905149051,
      "loss": 2.6393187046051025,
      "step": 12286
    },
    {
      "epoch": 0.9963234828348962,
      "grad_norm": 0.03360108658671379,
      "learning_rate": 0.0008485094850948509,
      "loss": 2.7096943855285645,
      "step": 12288
    },
    {
      "epoch": 0.9964856448600973,
      "grad_norm": 0.027252327650785446,
      "learning_rate": 0.0008479674796747967,
      "loss": 2.751828193664551,
      "step": 12290
    },
    {
      "epoch": 0.9966478068852982,
      "grad_norm": 0.042135920375585556,
      "learning_rate": 0.0008474254742547424,
      "loss": 2.631910800933838,
      "step": 12292
    },
    {
      "epoch": 0.9968099689104992,
      "grad_norm": 0.028693191707134247,
      "learning_rate": 0.0008468834688346883,
      "loss": 2.7046141624450684,
      "step": 12294
    },
    {
      "epoch": 0.9969721309357003,
      "grad_norm": 0.02582079917192459,
      "learning_rate": 0.0008463414634146342,
      "loss": 2.7258009910583496,
      "step": 12296
    },
    {
      "epoch": 0.9971342929609012,
      "grad_norm": 0.02805597521364689,
      "learning_rate": 0.0008457994579945799,
      "loss": 2.6894044876098633,
      "step": 12298
    },
    {
      "epoch": 0.9972964549861022,
      "grad_norm": 0.028356250375509262,
      "learning_rate": 0.0008452574525745257,
      "loss": 2.705886125564575,
      "step": 12300
    },
    {
      "epoch": 0.9974586170113032,
      "grad_norm": 0.022718602791428566,
      "learning_rate": 0.0008447154471544714,
      "loss": 2.6804046630859375,
      "step": 12302
    },
    {
      "epoch": 0.9976207790365041,
      "grad_norm": 0.036905791610479355,
      "learning_rate": 0.0008441734417344172,
      "loss": 2.7255921363830566,
      "step": 12304
    },
    {
      "epoch": 0.9977829410617052,
      "grad_norm": 0.04353989288210869,
      "learning_rate": 0.0008436314363143631,
      "loss": 2.6802046298980713,
      "step": 12306
    },
    {
      "epoch": 0.9979451030869062,
      "grad_norm": 0.037466369569301605,
      "learning_rate": 0.0008430894308943089,
      "loss": 2.676889419555664,
      "step": 12308
    },
    {
      "epoch": 0.9981072651121071,
      "grad_norm": 0.023219861090183258,
      "learning_rate": 0.0008425474254742547,
      "loss": 2.651834487915039,
      "step": 12310
    },
    {
      "epoch": 0.9982694271373082,
      "grad_norm": 0.03617582842707634,
      "learning_rate": 0.0008420054200542005,
      "loss": 2.6181015968322754,
      "step": 12312
    },
    {
      "epoch": 0.9984315891625092,
      "grad_norm": 0.049093469977378845,
      "learning_rate": 0.0008414634146341462,
      "loss": 2.69926118850708,
      "step": 12314
    },
    {
      "epoch": 0.9985937511877101,
      "grad_norm": 0.02806774526834488,
      "learning_rate": 0.0008409214092140921,
      "loss": 2.64633846282959,
      "step": 12316
    },
    {
      "epoch": 0.9987559132129111,
      "grad_norm": 0.04648776724934578,
      "learning_rate": 0.0008403794037940379,
      "loss": 2.694744110107422,
      "step": 12318
    },
    {
      "epoch": 0.9989180752381122,
      "grad_norm": 0.055591996759176254,
      "learning_rate": 0.0008398373983739837,
      "loss": 2.665167808532715,
      "step": 12320
    },
    {
      "epoch": 0.9990802372633131,
      "grad_norm": 0.023938195779919624,
      "learning_rate": 0.0008392953929539295,
      "loss": 2.6963772773742676,
      "step": 12322
    },
    {
      "epoch": 0.9992423992885141,
      "grad_norm": 0.053962863981723785,
      "learning_rate": 0.0008387533875338752,
      "loss": 2.678426742553711,
      "step": 12324
    },
    {
      "epoch": 0.9994045613137151,
      "grad_norm": 0.0332353338599205,
      "learning_rate": 0.0008382113821138211,
      "loss": 2.6230480670928955,
      "step": 12326
    },
    {
      "epoch": 0.999566723338916,
      "grad_norm": 0.06313833594322205,
      "learning_rate": 0.0008376693766937669,
      "loss": 2.656700372695923,
      "step": 12328
    },
    {
      "epoch": 0.9997288853641171,
      "grad_norm": 0.06582508236169815,
      "learning_rate": 0.0008371273712737126,
      "loss": 2.658054828643799,
      "step": 12330
    },
    {
      "epoch": 0.9998910473893181,
      "grad_norm": 0.024370497092604637,
      "learning_rate": 0.0008365853658536585,
      "loss": 2.650052785873413,
      "step": 12332
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0658532902598381,
      "learning_rate": 0.0008360433604336043,
      "loss": 2.6459388732910156,
      "step": 12334
    },
    {
      "epoch": 1.000162162025201,
      "grad_norm": 0.03311123326420784,
      "learning_rate": 0.00083550135501355,
      "loss": 2.656437873840332,
      "step": 12336
    },
    {
      "epoch": 1.000162162025201,
      "eval_loss": 2.658768653869629,
      "eval_runtime": 1149.2334,
      "eval_samples_per_second": 27.752,
      "eval_steps_per_second": 6.939,
      "step": 12336
    }
  ],
  "logging_steps": 2,
  "max_steps": 15418,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1542,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 7.904657733642407e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}