diff --git "a/checkpoint-27600/trainer_state.json" "b/checkpoint-27600/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-27600/trainer_state.json"
@@ -0,0 +1,7954 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.176287051482059,
+  "eval_steps": 3450,
+  "global_step": 27600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006500260010400416,
+      "grad_norm": 0.17750905454158783,
+      "learning_rate": 1.7391304347826088e-07,
+      "loss": 0.004283378422260284,
+      "step": 25
+    },
+    {
+      "epoch": 0.013000520020800831,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5507246376811595e-07,
+      "loss": 0.003918839395046234,
+      "step": 50
+    },
+    {
+      "epoch": 0.01950078003120125,
+      "grad_norm": 0.16697220504283905,
+      "learning_rate": 5.36231884057971e-07,
+      "loss": 0.003418746292591095,
+      "step": 75
+    },
+    {
+      "epoch": 0.026001040041601663,
+      "grad_norm": 0.1167321726679802,
+      "learning_rate": 7.17391304347826e-07,
+      "loss": 0.002580061852931976,
+      "step": 100
+    },
+    {
+      "epoch": 0.03250130005200208,
+      "grad_norm": 0.105965755879879,
+      "learning_rate": 8.985507246376812e-07,
+      "loss": 0.0020868690311908722,
+      "step": 125
+    },
+    {
+      "epoch": 0.0390015600624025,
+      "grad_norm": 0.07271800190210342,
+      "learning_rate": 1.0797101449275362e-06,
+      "loss": 0.0017790715396404267,
+      "step": 150
+    },
+    {
+      "epoch": 0.045501820072802915,
+      "grad_norm": 0.03961865231394768,
+      "learning_rate": 1.2608695652173913e-06,
+      "loss": 0.0012882563471794128,
+      "step": 175
+    },
+    {
+      "epoch": 0.052002080083203325,
+      "grad_norm": 0.03000999428331852,
+      "learning_rate": 1.4420289855072464e-06,
+      "loss": 0.0009397192299365997,
+      "step": 200
+    },
+    {
+      "epoch": 0.05850234009360374,
+      "grad_norm": 0.0296466201543808,
+      "learning_rate": 1.6231884057971016e-06,
+      "loss": 0.0007933838665485382,
+      "step": 225
+    },
+    {
+      "epoch": 0.06500260010400416,
+      "grad_norm": 0.017398398369550705,
+      "learning_rate": 1.8043478260869565e-06,
+      "loss": 0.0007029397785663604,
+      "step": 250
+    },
+    {
+      "epoch": 0.07150286011440457,
+      "grad_norm": 0.01113984640687704,
+      "learning_rate": 1.9855072463768116e-06,
+      "loss": 0.0006421895325183868,
+      "step": 275
+    },
+    {
+      "epoch": 0.078003120124805,
+      "grad_norm": 0.004796209745109081,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 0.0006597068160772324,
+      "step": 300
+    },
+    {
+      "epoch": 0.0845033801352054,
+      "grad_norm": 0.0022946936078369617,
+      "learning_rate": 2.347826086956522e-06,
+      "loss": 0.0005539159476757049,
+      "step": 325
+    },
+    {
+      "epoch": 0.09100364014560583,
+      "grad_norm": 0.006128576118499041,
+      "learning_rate": 2.5289855072463768e-06,
+      "loss": 0.0005852743610739708,
+      "step": 350
+    },
+    {
+      "epoch": 0.09750390015600624,
+      "grad_norm": 0.002111990237608552,
+      "learning_rate": 2.710144927536232e-06,
+      "loss": 0.0005731129273772239,
+      "step": 375
+    },
+    {
+      "epoch": 0.10400416016640665,
+      "grad_norm": 0.0030836218502372503,
+      "learning_rate": 2.8913043478260875e-06,
+      "loss": 0.0005443268269300461,
+      "step": 400
+    },
+    {
+      "epoch": 0.11050442017680707,
+      "grad_norm": 0.0027108530048280954,
+      "learning_rate": 3.072463768115942e-06,
+      "loss": 0.000556994080543518,
+      "step": 425
+    },
+    {
+      "epoch": 0.11700468018720749,
+      "grad_norm": 0.002011853503063321,
+      "learning_rate": 3.2536231884057973e-06,
+      "loss": 0.0005222153663635254,
+      "step": 450
+    },
+    {
+      "epoch": 0.12350494019760791,
+      "grad_norm": 0.003067562822252512,
+      "learning_rate": 3.4347826086956526e-06,
+      "loss": 0.00048564128577709197,
+      "step": 475
+    },
+    {
+      "epoch": 0.13000520020800832,
+      "grad_norm": 0.002368070650845766,
+      "learning_rate": 3.6159420289855076e-06,
+      "loss": 0.00047893021255731585,
+      "step": 500
+    },
+    {
+      "epoch": 0.13650546021840873,
+      "grad_norm": 0.001583599136210978,
+      "learning_rate": 3.797101449275362e-06,
+      "loss": 0.00047833062708377836,
+      "step": 525
+    },
+    {
+      "epoch": 0.14300572022880914,
+      "grad_norm": 0.0014594887616112828,
+      "learning_rate": 3.978260869565217e-06,
+      "loss": 0.0005091005936264992,
+      "step": 550
+    },
+    {
+      "epoch": 0.14950598023920958,
+      "grad_norm": 0.0014220334123820066,
+      "learning_rate": 4.159420289855073e-06,
+      "loss": 0.0005074851214885712,
+      "step": 575
+    },
+    {
+      "epoch": 0.15600624024961,
+      "grad_norm": 0.00207452941685915,
+      "learning_rate": 4.340579710144928e-06,
+      "loss": 0.000527675561606884,
+      "step": 600
+    },
+    {
+      "epoch": 0.1625065002600104,
+      "grad_norm": 0.0017482496332377195,
+      "learning_rate": 4.5217391304347826e-06,
+      "loss": 0.00044770020991563797,
+      "step": 625
+    },
+    {
+      "epoch": 0.1690067602704108,
+      "grad_norm": 0.0013480493798851967,
+      "learning_rate": 4.702898550724638e-06,
+      "loss": 0.000461682491004467,
+      "step": 650
+    },
+    {
+      "epoch": 0.17550702028081122,
+      "grad_norm": 0.0017523127608001232,
+      "learning_rate": 4.884057971014493e-06,
+      "loss": 0.000520368255674839,
+      "step": 675
+    },
+    {
+      "epoch": 0.18200728029121166,
+      "grad_norm": 0.004605201538652182,
+      "learning_rate": 5.065217391304348e-06,
+      "loss": 0.0004663777351379395,
+      "step": 700
+    },
+    {
+      "epoch": 0.18850754030161207,
+      "grad_norm": 0.0011810408905148506,
+      "learning_rate": 5.246376811594203e-06,
+      "loss": 0.0004738373309373856,
+      "step": 725
+    },
+    {
+      "epoch": 0.19500780031201248,
+      "grad_norm": 0.0024364716373384,
+      "learning_rate": 5.427536231884058e-06,
+      "loss": 0.000477650985121727,
+      "step": 750
+    },
+    {
+      "epoch": 0.2015080603224129,
+      "grad_norm": 0.0014733612770214677,
+      "learning_rate": 5.608695652173914e-06,
+      "loss": 0.00044887635856866837,
+      "step": 775
+    },
+    {
+      "epoch": 0.2080083203328133,
+      "grad_norm": 0.0022185698617249727,
+      "learning_rate": 5.789855072463768e-06,
+      "loss": 0.0004545997828245163,
+      "step": 800
+    },
+    {
+      "epoch": 0.21450858034321374,
+      "grad_norm": 0.001337498426437378,
+      "learning_rate": 5.9710144927536236e-06,
+      "loss": 0.00044511511921882627,
+      "step": 825
+    },
+    {
+      "epoch": 0.22100884035361415,
+      "grad_norm": 0.0011507358867675066,
+      "learning_rate": 6.152173913043478e-06,
+      "loss": 0.0004143805801868439,
+      "step": 850
+    },
+    {
+      "epoch": 0.22750910036401456,
+      "grad_norm": 0.0007577093783766031,
+      "learning_rate": 6.333333333333334e-06,
+      "loss": 0.00043120242655277254,
+      "step": 875
+    },
+    {
+      "epoch": 0.23400936037441497,
+      "grad_norm": 0.0011255984427407384,
+      "learning_rate": 6.514492753623189e-06,
+      "loss": 0.00042323727160692215,
+      "step": 900
+    },
+    {
+      "epoch": 0.24050962038481538,
+      "grad_norm": 0.0013750302605330944,
+      "learning_rate": 6.695652173913043e-06,
+      "loss": 0.00038128703832626345,
+      "step": 925
+    },
+    {
+      "epoch": 0.24700988039521582,
+      "grad_norm": 0.0006967707304283977,
+      "learning_rate": 6.876811594202899e-06,
+      "loss": 0.00040074307471513746,
+      "step": 950
+    },
+    {
+      "epoch": 0.2535101404056162,
+      "grad_norm": 0.0013500290224328637,
+      "learning_rate": 7.057971014492754e-06,
+      "loss": 0.0003993678465485573,
+      "step": 975
+    },
+    {
+      "epoch": 0.26001040041601664,
+      "grad_norm": 0.0015217582695186138,
+      "learning_rate": 7.239130434782609e-06,
+      "loss": 0.0003542538732290268,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2665106604264171,
+      "grad_norm": 0.0009631581488065422,
+      "learning_rate": 7.420289855072464e-06,
+      "loss": 0.000437321700155735,
+      "step": 1025
+    },
+    {
+      "epoch": 0.27301092043681746,
+      "grad_norm": 0.0005817293422296643,
+      "learning_rate": 7.60144927536232e-06,
+      "loss": 0.0003608403727412224,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2795111804472179,
+      "grad_norm": 0.0005613359971903265,
+      "learning_rate": 7.782608695652174e-06,
+      "loss": 0.00035999853163957594,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2860114404576183,
+      "grad_norm": 0.0007700200658291578,
+      "learning_rate": 7.963768115942028e-06,
+      "loss": 0.0004147729650139809,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2925117004680187,
+      "grad_norm": 0.0013384182238951325,
+      "learning_rate": 8.144927536231884e-06,
+      "loss": 0.00035684410482645035,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29901196047841916,
+      "grad_norm": 0.0005455451901070774,
+      "learning_rate": 8.326086956521739e-06,
+      "loss": 0.0003590710461139679,
+      "step": 1150
+    },
+    {
+      "epoch": 0.30551222048881954,
+      "grad_norm": 0.0006808220059610903,
+      "learning_rate": 8.507246376811595e-06,
+      "loss": 0.0003543094918131828,
+      "step": 1175
+    },
+    {
+      "epoch": 0.31201248049922,
+      "grad_norm": 0.0007319796131923795,
+      "learning_rate": 8.68840579710145e-06,
+      "loss": 0.0004256186634302139,
+      "step": 1200
+    },
+    {
+      "epoch": 0.31851274050962036,
+      "grad_norm": 0.00041355780558660626,
+      "learning_rate": 8.869565217391304e-06,
+      "loss": 0.0003611164167523384,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3250130005200208,
+      "grad_norm": 0.0009358125389553607,
+      "learning_rate": 9.05072463768116e-06,
+      "loss": 0.0003898784518241882,
+      "step": 1250
+    },
+    {
+      "epoch": 0.33151326053042124,
+      "grad_norm": 0.0009471450466662645,
+      "learning_rate": 9.231884057971015e-06,
+      "loss": 0.00036198627203702926,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3380135205408216,
+      "grad_norm": 0.0005847045686095953,
+      "learning_rate": 9.41304347826087e-06,
+      "loss": 0.0003970867022871971,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34451378055122206,
+      "grad_norm": 0.0011201045708730817,
+      "learning_rate": 9.594202898550725e-06,
+      "loss": 0.0003872425854206085,
+      "step": 1325
+    },
+    {
+      "epoch": 0.35101404056162244,
+      "grad_norm": 0.00048665812937542796,
+      "learning_rate": 9.775362318840581e-06,
+      "loss": 0.0003776940330862999,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3575143005720229,
+      "grad_norm": 0.0005485154106281698,
+      "learning_rate": 9.956521739130436e-06,
+      "loss": 0.0003538001701235771,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3640145605824233,
+      "grad_norm": 0.0006577415624633431,
+      "learning_rate": 1.013768115942029e-05,
+      "loss": 0.0003705337271094322,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3705148205928237,
+      "grad_norm": 0.0006318407249636948,
+      "learning_rate": 1.0318840579710145e-05,
+      "loss": 0.0003729483112692833,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37701508060322414,
+      "grad_norm": 0.0007680612616240978,
+      "learning_rate": 1.05e-05,
+      "loss": 0.0003821665421128273,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3835153406136245,
+      "grad_norm": 0.0006065071211196482,
+      "learning_rate": 1.0681159420289856e-05,
+      "loss": 0.0003943805769085884,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39001560062402496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.086231884057971e-05,
+      "loss": 0.0003467845171689987,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3965158606344254,
+      "grad_norm": 0.0021487465128302574,
+      "learning_rate": 1.1043478260869566e-05,
+      "loss": 0.00034700412303209307,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4030161206448258,
+      "grad_norm": 0.000630208698567003,
+      "learning_rate": 1.122463768115942e-05,
+      "loss": 0.0003760438784956932,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4095163806552262,
+      "grad_norm": 0.000845056667458266,
+      "learning_rate": 1.1405797101449275e-05,
+      "loss": 0.0003565104678273201,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4160166406656266,
+      "grad_norm": 0.0014387870905920863,
+      "learning_rate": 1.1586956521739131e-05,
+      "loss": 0.000367693230509758,
+      "step": 1600
+    },
+    {
+      "epoch": 0.42251690067602704,
+      "grad_norm": 0.0012101867469027638,
+      "learning_rate": 1.1768115942028986e-05,
+      "loss": 0.0003509974479675293,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4290171606864275,
+      "grad_norm": 0.0007229935727082193,
+      "learning_rate": 1.1949275362318842e-05,
+      "loss": 0.00032424140721559524,
+      "step": 1650
+    },
+    {
+      "epoch": 0.43551742069682786,
+      "grad_norm": 0.0007024598307907581,
+      "learning_rate": 1.2130434782608697e-05,
+      "loss": 0.00034356161952018736,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4420176807072283,
+      "grad_norm": 0.0014094120124354959,
+      "learning_rate": 1.2311594202898551e-05,
+      "loss": 0.00032688610255718234,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4485179407176287,
+      "grad_norm": 0.0004223083960823715,
+      "learning_rate": 1.2492753623188407e-05,
+      "loss": 0.0003457293659448624,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4550182007280291,
+      "grad_norm": 0.0009581646299920976,
+      "learning_rate": 1.2673913043478262e-05,
+      "loss": 0.0003291889280080795,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46151846073842956,
+      "grad_norm": 0.002243791939690709,
+      "learning_rate": 1.2855072463768116e-05,
+      "loss": 0.00037662502378225325,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46801872074882994,
+      "grad_norm": 0.000704752397723496,
+      "learning_rate": 1.3036231884057972e-05,
+      "loss": 0.0003469870612025261,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4745189807592304,
+      "grad_norm": 0.000755470129661262,
+      "learning_rate": 1.3217391304347825e-05,
+      "loss": 0.0003220204263925552,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48101924076963076,
+      "grad_norm": 0.0014091039774939418,
+      "learning_rate": 1.3398550724637681e-05,
+      "loss": 0.00035315047949552536,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4875195007800312,
+      "grad_norm": 0.001953344326466322,
+      "learning_rate": 1.3579710144927538e-05,
+      "loss": 0.0003283596411347389,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49401976079043164,
+      "grad_norm": 0.0007954083266668022,
+      "learning_rate": 1.376086956521739e-05,
+      "loss": 0.0003253992274403572,
+      "step": 1900
+    },
+    {
+      "epoch": 0.500520020800832,
+      "grad_norm": 0.0008482153061777353,
+      "learning_rate": 1.3942028985507247e-05,
+      "loss": 0.0003893987089395523,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5070202808112324,
+      "grad_norm": 0.000624845502898097,
+      "learning_rate": 1.4123188405797103e-05,
+      "loss": 0.0003261050209403038,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5135205408216329,
+      "grad_norm": 0.001078785746358335,
+      "learning_rate": 1.4304347826086959e-05,
+      "loss": 0.0003268422186374664,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5200208008320333,
+      "grad_norm": 0.0012631005374714732,
+      "learning_rate": 1.4485507246376812e-05,
+      "loss": 0.00033046893775463105,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5265210608424337,
+      "grad_norm": 0.001895563444122672,
+      "learning_rate": 1.4666666666666668e-05,
+      "loss": 0.0003267909586429596,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5330213208528342,
+      "grad_norm": 0.00041873074951581657,
+      "learning_rate": 1.4847826086956524e-05,
+      "loss": 0.0003187931701540947,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5395215808632345,
+      "grad_norm": 0.0010419194586575031,
+      "learning_rate": 1.5028985507246377e-05,
+      "loss": 0.00034294985234737394,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5460218408736349,
+      "grad_norm": 0.0019357349956408143,
+      "learning_rate": 1.5210144927536233e-05,
+      "loss": 0.0003515840321779251,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5525221008840354,
+      "grad_norm": 0.0013144762488082051,
+      "learning_rate": 1.5391304347826088e-05,
+      "loss": 0.00035412784665822983,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5590223608944358,
+      "grad_norm": 0.0006452434463426471,
+      "learning_rate": 1.5572463768115944e-05,
+      "loss": 0.00033226653933525084,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5655226209048362,
+      "grad_norm": 0.0012123215710744262,
+      "learning_rate": 1.5753623188405797e-05,
+      "loss": 0.0003139122948050499,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5720228809152366,
+      "grad_norm": 0.0006717830547131598,
+      "learning_rate": 1.5934782608695653e-05,
+      "loss": 0.00033644989132881165,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5785231409256371,
+      "grad_norm": 0.0004599529202096164,
+      "learning_rate": 1.611594202898551e-05,
+      "loss": 0.0003589295595884323,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5850234009360374,
+      "grad_norm": 0.0004877887840848416,
+      "learning_rate": 1.6297101449275362e-05,
+      "loss": 0.00031096141785383227,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5915236609464378,
+      "grad_norm": 0.0002835657214745879,
+      "learning_rate": 1.6478260869565218e-05,
+      "loss": 0.00033451490104198457,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5980239209568383,
+      "grad_norm": 0.0010292587103322148,
+      "learning_rate": 1.6659420289855074e-05,
+      "loss": 0.0003324819728732109,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6045241809672387,
+      "grad_norm": 0.0020344816148281097,
+      "learning_rate": 1.6840579710144927e-05,
+      "loss": 0.00030768170952796933,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6110244409776391,
+      "grad_norm": 0.0008710804395377636,
+      "learning_rate": 1.7021739130434783e-05,
+      "loss": 0.00029756918549537657,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6175247009880396,
+      "grad_norm": 0.0010297639528289437,
+      "learning_rate": 1.720289855072464e-05,
+      "loss": 0.0003132735565304756,
+      "step": 2375
+    },
+    {
+      "epoch": 0.62402496099844,
+      "grad_norm": 0.0010929155396297574,
+      "learning_rate": 1.7384057971014496e-05,
+      "loss": 0.0003162991255521774,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6305252210088403,
+      "grad_norm": 0.0005164285539649427,
+      "learning_rate": 1.756521739130435e-05,
+      "loss": 0.0003191165253520012,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6370254810192407,
+      "grad_norm": 0.0014058813685551286,
+      "learning_rate": 1.7746376811594205e-05,
+      "loss": 0.00029166329652071,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6435257410296412,
+      "grad_norm": 0.00036998544237576425,
+      "learning_rate": 1.792753623188406e-05,
+      "loss": 0.00031204935163259506,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6500260010400416,
+      "grad_norm": 0.0005906142177991569,
+      "learning_rate": 1.8108695652173913e-05,
+      "loss": 0.00031461842358112337,
+      "step": 2500
+    },
+    {
+      "epoch": 0.656526261050442,
+      "grad_norm": 0.0019112491281703115,
+      "learning_rate": 1.828985507246377e-05,
+      "loss": 0.0003568634390830994,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6630265210608425,
+      "grad_norm": 0.0005400856025516987,
+      "learning_rate": 1.8471014492753626e-05,
+      "loss": 0.00031186673790216445,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6695267810712429,
+      "grad_norm": 0.0014024826232343912,
+      "learning_rate": 1.865217391304348e-05,
+      "loss": 0.00029229436069726943,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6760270410816432,
+      "grad_norm": 0.0009240263025276363,
+      "learning_rate": 1.8833333333333335e-05,
+      "loss": 0.00031984806060791015,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6825273010920437,
+      "grad_norm": 0.001019714749418199,
+      "learning_rate": 1.9014492753623188e-05,
+      "loss": 0.0002883545309305191,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6890275611024441,
+      "grad_norm": 0.0007436838932335377,
+      "learning_rate": 1.9195652173913044e-05,
+      "loss": 0.00031793177127838136,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6955278211128445,
+      "grad_norm": 0.004361623898148537,
+      "learning_rate": 1.9376811594202897e-05,
+      "loss": 0.0003340127319097519,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7020280811232449,
+      "grad_norm": 0.002120215678587556,
+      "learning_rate": 1.9557971014492753e-05,
+      "loss": 0.0003347470238804817,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7085283411336454,
+      "grad_norm": 0.0009742141119204462,
+      "learning_rate": 1.973913043478261e-05,
+      "loss": 0.0002991884760558605,
+      "step": 2725
+    },
+    {
+      "epoch": 0.7150286011440458,
+      "grad_norm": 0.0017010399606078863,
+      "learning_rate": 1.9920289855072465e-05,
+      "loss": 0.0003556852787733078,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7215288611544461,
+      "grad_norm": 0.0007170455064624548,
+      "learning_rate": 2.0101449275362318e-05,
+      "loss": 0.0002682306058704853,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7280291211648466,
+      "grad_norm": 0.00114281359128654,
+      "learning_rate": 2.0282608695652174e-05,
+      "loss": 0.0003118632733821869,
+      "step": 2800
+    },
+    {
+      "epoch": 0.734529381175247,
+      "grad_norm": 0.0010286562610417604,
+      "learning_rate": 2.046376811594203e-05,
+      "loss": 0.00028729531913995743,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7410296411856474,
+      "grad_norm": 0.001677649561315775,
+      "learning_rate": 2.0644927536231883e-05,
+      "loss": 0.0003165191411972046,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7475299011960479,
+      "grad_norm": 0.0006778730312362313,
+      "learning_rate": 2.082608695652174e-05,
+      "loss": 0.0002633074298501015,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7540301612064483,
+      "grad_norm": 0.0007009701803326607,
+      "learning_rate": 2.1007246376811596e-05,
+      "loss": 0.00035017453134059904,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7605304212168487,
+      "grad_norm": 0.002753955079242587,
+      "learning_rate": 2.1188405797101452e-05,
+      "loss": 0.00035684075206518176,
+      "step": 2925
+    },
+    {
+      "epoch": 0.767030681227249,
+      "grad_norm": 0.0009396990644745529,
+      "learning_rate": 2.1369565217391305e-05,
+      "loss": 0.00032220348715782164,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7735309412376495,
+      "grad_norm": 0.0005787694244645536,
+      "learning_rate": 2.155072463768116e-05,
+      "loss": 0.0002845560945570469,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7800312012480499,
+      "grad_norm": 0.001705740811303258,
+      "learning_rate": 2.1731884057971017e-05,
+      "loss": 0.0003118696250021458,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7865314612584503,
+      "grad_norm": 0.0016730790957808495,
+      "learning_rate": 2.191304347826087e-05,
+      "loss": 0.00028975291177630425,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7930317212688508,
+      "grad_norm": 0.0006580168846994638,
+      "learning_rate": 2.2094202898550726e-05,
+      "loss": 0.00031032718718051913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7995319812792512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2275362318840582e-05,
+      "loss": 0.0002661898173391819,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8060322412896516,
+      "grad_norm": 0.004355259705334902,
+      "learning_rate": 2.2456521739130438e-05,
+      "loss": 0.00029955208301544187,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8125325013000521,
+      "grad_norm": 0.0021052362862974405,
+      "learning_rate": 2.263768115942029e-05,
+      "loss": 0.00029472261667251587,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8190327613104524,
+      "grad_norm": 0.0020532577764242887,
+      "learning_rate": 2.2818840579710147e-05,
+      "loss": 0.00032611183822154997,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8255330213208528,
+      "grad_norm": 0.001252846559509635,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.0002973646484315395,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8320332813312532,
+      "grad_norm": 0.0025867847725749016,
+      "learning_rate": 2.3181159420289856e-05,
+      "loss": 0.0003094041533768177,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8385335413416537,
+      "grad_norm": 0.0043645896948874,
+      "learning_rate": 2.3362318840579712e-05,
+      "loss": 0.00031783077865839006,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8450338013520541,
+      "grad_norm": 0.0006059009465388954,
+      "learning_rate": 2.354347826086957e-05,
+      "loss": 0.00027763141319155695,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8515340613624545,
+      "grad_norm": 0.0023931902833282948,
+      "learning_rate": 2.372463768115942e-05,
+      "loss": 0.0002982875518500805,
+      "step": 3275
+    },
+    {
+      "epoch": 0.858034321372855,
+      "grad_norm": 0.0011626601917669177,
+      "learning_rate": 2.3905797101449278e-05,
+      "loss": 0.00029426002874970435,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8645345813832553,
+      "grad_norm": 0.0014857512433081865,
+      "learning_rate": 2.408695652173913e-05,
+      "loss": 0.0003030484728515148,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8710348413936557,
+      "grad_norm": 0.0026861918158829212,
+      "learning_rate": 2.4268115942028987e-05,
+      "loss": 0.00026893947273492813,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8775351014040562,
+      "grad_norm": 0.00171413563657552,
+      "learning_rate": 2.444927536231884e-05,
+      "loss": 0.00031247254461050036,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8840353614144566,
+      "grad_norm": 0.0013334028190001845,
+      "learning_rate": 2.4630434782608696e-05,
+      "loss": 0.0002977939695119858,
+      "step": 3400
+    },
+    {
+      "epoch": 0.890535621424857,
+      "grad_norm": 0.0015155395958572626,
+      "learning_rate": 2.4811594202898552e-05,
+      "loss": 0.0003019925206899643,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8970358814352574,
+      "grad_norm": 0.0019004828063771129,
+      "learning_rate": 2.4992753623188405e-05,
+      "loss": 0.0002948020212352276,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8970358814352574,
+      "eval_loss": 0.0013163755647838116,
+      "eval_runtime": 7.2747,
+      "eval_samples_per_second": 687.314,
+      "eval_steps_per_second": 3.711,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8970358814352574,
+      "eval_holdout_avg_strict_f1": 0.01979999249818373,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.041815838464843186,
+      "eval_holdout_nemotron_pii_strict_f1": 0.04583859376517432,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.07483772432226041,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.055307760927743095,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9035361414456579,
+      "grad_norm": 0.0011671730317175388,
+      "learning_rate": 2.5173913043478264e-05,
+      "loss": 0.00026905197650194167,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9100364014560582,
+      "grad_norm": 0.0010831784456968307,
+      "learning_rate": 2.5355072463768113e-05,
+      "loss": 0.0003184381127357483,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9165366614664586,
+      "grad_norm": 0.0027733801398426294,
+      "learning_rate": 2.553623188405797e-05,
+      "loss": 0.00031777344644069673,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9230369214768591,
+      "grad_norm": 0.006140136159956455,
+      "learning_rate": 2.5717391304347826e-05,
+      "loss": 0.00030992023646831513,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9295371814872595,
+      "grad_norm": 0.004694697447121143,
+      "learning_rate": 2.5898550724637682e-05,
+      "loss": 0.0003017419762909412,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9360374414976599,
+      "grad_norm": 0.0031786097679287195,
+      "learning_rate": 2.6079710144927538e-05,
+      "loss": 0.0003097628243267536,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9425377015080604,
+      "grad_norm": 0.0016794903203845024,
+      "learning_rate": 2.6260869565217394e-05,
+      "loss": 0.0002991587668657303,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9490379615184608,
+      "grad_norm": 0.0023807429242879152,
+      "learning_rate": 2.644202898550725e-05,
+      "loss": 0.00031139522790908815,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9555382215288611,
+      "grad_norm": 0.003003185149282217,
+      "learning_rate": 2.66231884057971e-05,
+      "loss": 0.0002772625721991062,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9620384815392615,
+      "grad_norm": 0.000796256004832685,
+      "learning_rate": 2.6804347826086956e-05,
+      "loss": 0.0003119866922497749,
+      "step": 3700
+    },
+    {
+      "epoch": 0.968538741549662,
+      "grad_norm": 0.0021845803130418062,
+      "learning_rate": 2.6985507246376812e-05,
+      "loss": 0.000316128209233284,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9750390015600624,
+      "grad_norm": 0.00112715526483953,
+      "learning_rate": 2.716666666666667e-05,
+      "loss": 0.00029152233153581617,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9815392615704628,
+      "grad_norm": 0.0024580543395131826,
+      "learning_rate": 2.7347826086956525e-05,
+      "loss": 0.0002828400582075119,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9880395215808633,
+      "grad_norm": 0.0031969628762453794,
+      "learning_rate": 2.752898550724638e-05,
+      "loss": 0.0003035380691289902,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9945397815912637,
+      "grad_norm": 0.015604513697326183,
+      "learning_rate": 2.7710144927536237e-05,
+      "loss": 0.0003056460060179234,
+      "step": 3825
+    },
+    {
+      "epoch": 1.001040041601664,
+      "grad_norm": 0.0015990090323612094,
+      "learning_rate": 2.7891304347826087e-05,
+      "loss": 0.00030797651037573815,
+      "step": 3850
+    },
+    {
+      "epoch": 1.0075403016120645,
+      "grad_norm": 0.0006027631461620331,
+      "learning_rate": 2.8072463768115943e-05,
+      "loss": 0.0002863717824220657,
+      "step": 3875
+    },
+    {
+      "epoch": 1.0140405616224648,
+      "grad_norm": 0.008887791074812412,
+      "learning_rate": 2.82536231884058e-05,
+      "loss": 0.00033138148486614225,
+      "step": 3900
+    },
+    {
+      "epoch": 1.0205408216328653,
+      "grad_norm": 0.0011880005476996303,
+      "learning_rate": 2.8434782608695655e-05,
+      "loss": 0.0003792601823806763,
+      "step": 3925
+    },
+    {
+      "epoch": 1.0270410816432658,
+      "grad_norm": 0.0044065662659704685,
+      "learning_rate": 2.8615942028985508e-05,
+      "loss": 0.0003368314728140831,
+      "step": 3950
+    },
+    {
+      "epoch": 1.033541341653666,
+      "grad_norm": 0.004491528961807489,
+      "learning_rate": 2.8797101449275364e-05,
+      "loss": 0.00031654369086027146,
+      "step": 3975
+    },
+    {
+      "epoch": 1.0400416016640666,
+      "grad_norm": 0.0022014344576746225,
+      "learning_rate": 2.897826086956522e-05,
+      "loss": 0.000343184731900692,
+      "step": 4000
+    },
+    {
+      "epoch": 1.046541861674467,
+      "grad_norm": 0.0020901078823953867,
+      "learning_rate": 2.9159420289855073e-05,
+      "loss": 0.0003895349428057671,
+      "step": 4025
+    },
+    {
+      "epoch": 1.0530421216848673,
+      "grad_norm": 0.006787622347474098,
+      "learning_rate": 2.934057971014493e-05,
+      "loss": 0.000345432460308075,
+      "step": 4050
+    },
+    {
+      "epoch": 1.0595423816952678,
+      "grad_norm": 0.001472746254876256,
+      "learning_rate": 2.9521739130434782e-05,
+      "loss": 0.00033133041113615037,
+      "step": 4075
+    },
+    {
+      "epoch": 1.0660426417056683,
+      "grad_norm": 0.00274971523322165,
+      "learning_rate": 2.9702898550724638e-05,
+      "loss": 0.0003213922679424286,
+      "step": 4100
+    },
+    {
+      "epoch": 1.0725429017160686,
+      "grad_norm": 0.003665718249976635,
+      "learning_rate": 2.9884057971014494e-05,
+      "loss": 0.00037232354283332823,
+      "step": 4125
+    },
+    {
+      "epoch": 1.079043161726469,
+      "grad_norm": 0.0026763668283820152,
+      "learning_rate": 3.006521739130435e-05,
+      "loss": 0.0003335591405630112,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0855434217368696,
+      "grad_norm": 0.004248565994203091,
+      "learning_rate": 3.0246376811594207e-05,
+      "loss": 0.0003483542054891586,
+      "step": 4175
+    },
+    {
+      "epoch": 1.0920436817472698,
+      "grad_norm": 0.0028635687194764614,
+      "learning_rate": 3.0427536231884056e-05,
+      "loss": 0.00031913988292217257,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0985439417576703,
+      "grad_norm": 0.002372679766267538,
+      "learning_rate": 3.0608695652173916e-05,
+      "loss": 0.0003160311281681061,
+      "step": 4225
+    },
+    {
+      "epoch": 1.1050442017680706,
+      "grad_norm": 0.004170225001871586,
+      "learning_rate": 3.078985507246377e-05,
+      "loss": 0.0002849435992538929,
+      "step": 4250
+    },
+    {
+      "epoch": 1.111544461778471,
+      "grad_norm": 0.0,
+      "learning_rate": 3.097101449275363e-05,
+      "loss": 0.0003091808594763279,
+      "step": 4275
+    },
+    {
+      "epoch": 1.1180447217888716,
+      "grad_norm": 0.0010796879651024938,
+      "learning_rate": 3.115217391304348e-05,
+      "loss": 0.0003037399798631668,
+      "step": 4300
+    },
+    {
+      "epoch": 1.1245449817992719,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1333333333333334e-05,
+      "loss": 0.00032576750963926316,
+      "step": 4325
+    },
+    {
+      "epoch": 1.1310452418096724,
+      "grad_norm": 0.0064179119653999805,
+      "learning_rate": 3.151449275362319e-05,
+      "loss": 0.0003202928230166435,
+      "step": 4350
+    },
+    {
+      "epoch": 1.1375455018200729,
+      "grad_norm": 0.006963932886719704,
+      "learning_rate": 3.1695652173913046e-05,
+      "loss": 0.00034561607986688614,
+      "step": 4375
+    },
+    {
+      "epoch": 1.1440457618304731,
+      "grad_norm": 0.0019319987623021007,
+      "learning_rate": 3.18768115942029e-05,
+      "loss": 0.00034509383141994476,
+      "step": 4400
+    },
+    {
+      "epoch": 1.1505460218408736,
+      "grad_norm": 0.004242972936481237,
+      "learning_rate": 3.205797101449275e-05,
+      "loss": 0.00032770711928606036,
+      "step": 4425
+    },
+    {
+      "epoch": 1.1570462818512741,
+      "grad_norm": 0.0014416216872632504,
+      "learning_rate": 3.223913043478261e-05,
+      "loss": 0.00033207762986421586,
+      "step": 4450
+    },
+    {
+      "epoch": 1.1635465418616744,
+      "grad_norm": 0.003044139128178358,
+      "learning_rate": 3.2420289855072464e-05,
+      "loss": 0.00028797827661037446,
+      "step": 4475
+    },
+    {
+      "epoch": 1.1700468018720749,
+      "grad_norm": 0.004789582919329405,
+      "learning_rate": 3.2601449275362324e-05,
+      "loss": 0.00030507130548357964,
+      "step": 4500
+    },
+    {
+      "epoch": 1.1765470618824754,
+      "grad_norm": 0.011801498010754585,
+      "learning_rate": 3.278260869565217e-05,
+      "loss": 0.00031583432108163835,
+      "step": 4525
+    },
+    {
+      "epoch": 1.1830473218928756,
+      "grad_norm": 0.002333596348762512,
+      "learning_rate": 3.296376811594203e-05,
+      "loss": 0.0002924199961125851,
+      "step": 4550
+    },
+    {
+      "epoch": 1.1895475819032761,
+      "grad_norm": 0.003968964330852032,
+      "learning_rate": 3.314492753623188e-05,
+      "loss": 0.00034423466771841047,
+      "step": 4575
+    },
+    {
+      "epoch": 1.1960478419136766,
+      "grad_norm": 0.0037589913699775934,
+      "learning_rate": 3.332608695652174e-05,
+      "loss": 0.0002912192791700363,
+      "step": 4600
+    },
+    {
+      "epoch": 1.202548101924077,
+      "grad_norm": 0.005469044204801321,
+      "learning_rate": 3.3507246376811594e-05,
+      "loss": 0.000267421193420887,
+      "step": 4625
+    },
+    {
+      "epoch": 1.2090483619344774,
+      "grad_norm": 0.0013472590362653136,
+      "learning_rate": 3.3688405797101454e-05,
+      "loss": 0.0002920958027243614,
+      "step": 4650
+    },
+    {
+      "epoch": 1.215548621944878,
+      "grad_norm": 0.0010584670817479491,
+      "learning_rate": 3.386956521739131e-05,
+      "loss": 0.0003347089141607285,
+      "step": 4675
+    },
+    {
+      "epoch": 1.2220488819552782,
+      "grad_norm": 0.003385865828022361,
+      "learning_rate": 3.405072463768116e-05,
+      "loss": 0.00033686719834804537,
+      "step": 4700
+    },
+    {
+      "epoch": 1.2285491419656787,
+      "grad_norm": 0.0015582754276692867,
+      "learning_rate": 3.423188405797101e-05,
+      "loss": 0.00030806563794612886,
+      "step": 4725
+    },
+    {
+      "epoch": 1.2350494019760792,
+      "grad_norm": 0.006736577954143286,
+      "learning_rate": 3.441304347826087e-05,
+      "loss": 0.00030645377933979034,
+      "step": 4750
+    },
+    {
+      "epoch": 1.2415496619864794,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4594202898550725e-05,
+      "loss": 0.0002596873417496681,
+      "step": 4775
+    },
+    {
+      "epoch": 1.24804992199688,
+      "grad_norm": 0.0014598104171454906,
+      "learning_rate": 3.4775362318840584e-05,
+      "loss": 0.00031735524535179136,
+      "step": 4800
+    },
+    {
+      "epoch": 1.2545501820072804,
+      "grad_norm": 0.0012760423123836517,
+      "learning_rate": 3.495652173913044e-05,
+      "loss": 0.00034475263208150864,
+      "step": 4825
+    },
+    {
+      "epoch": 1.2610504420176807,
+      "grad_norm": 0.00399738596752286,
+      "learning_rate": 3.51376811594203e-05,
+      "loss": 0.0003468216210603714,
+      "step": 4850
+    },
+    {
+      "epoch": 1.2675507020280812,
+      "grad_norm": 0.000710926775354892,
+      "learning_rate": 3.531884057971014e-05,
+      "loss": 0.0003094401024281979,
+      "step": 4875
+    },
+    {
+      "epoch": 1.2740509620384817,
+      "grad_norm": 0.001548282103613019,
+      "learning_rate": 3.55e-05,
+      "loss": 0.00032061580568552015,
+      "step": 4900
+    },
+    {
+      "epoch": 1.280551222048882,
+      "grad_norm": 0.0024284820538014174,
+      "learning_rate": 3.5681159420289855e-05,
+      "loss": 0.00030055604875087736,
+      "step": 4925
+    },
+    {
+      "epoch": 1.2870514820592824,
+      "grad_norm": 0.004141749814152718,
+      "learning_rate": 3.5862318840579715e-05,
+      "loss": 0.0003376200422644615,
+      "step": 4950
+    },
+    {
+      "epoch": 1.2935517420696827,
+      "grad_norm": 0.0016009368700906634,
+      "learning_rate": 3.604347826086957e-05,
+      "loss": 0.00035102896392345427,
+      "step": 4975
+    },
+    {
+      "epoch": 1.3000520020800832,
+      "grad_norm": 0.002298409352079034,
+      "learning_rate": 3.622463768115942e-05,
+      "loss": 0.00029324846342206,
+      "step": 5000
+    },
+    {
+      "epoch": 1.3065522620904837,
+      "grad_norm": 0.002268020762130618,
+      "learning_rate": 3.640579710144928e-05,
+      "loss": 0.0003086787275969982,
+      "step": 5025
+    },
+    {
+      "epoch": 1.313052522100884,
+      "grad_norm": 0.0012281977105885744,
+      "learning_rate": 3.658695652173913e-05,
+      "loss": 0.0003048296086490154,
+      "step": 5050
+    },
+    {
+      "epoch": 1.3195527821112845,
+      "grad_norm": 0.003418742213398218,
+      "learning_rate": 3.6768115942028985e-05,
+      "loss": 0.00032037079334259035,
+      "step": 5075
+    },
+    {
+      "epoch": 1.3260530421216847,
+      "grad_norm": 0.004171136766672134,
+      "learning_rate": 3.694927536231884e-05,
+      "loss": 0.0002728208154439926,
+      "step": 5100
+    },
+    {
+      "epoch": 1.3325533021320852,
+      "grad_norm": 0.0014296929584816098,
+      "learning_rate": 3.71304347826087e-05,
+      "loss": 0.00031274981796741485,
+      "step": 5125
+    },
+    {
+      "epoch": 1.3390535621424857,
+      "grad_norm": 0.006193948443979025,
+      "learning_rate": 3.731159420289855e-05,
+      "loss": 0.00031706057488918306,
+      "step": 5150
+    },
+    {
+      "epoch": 1.345553822152886,
+      "grad_norm": 0.0031706327572464943,
+      "learning_rate": 3.749275362318841e-05,
+      "loss": 0.0002851133979856968,
+      "step": 5175
+    },
+    {
+      "epoch": 1.3520540821632865,
+      "grad_norm": 0.0012347496813163161,
+      "learning_rate": 3.767391304347826e-05,
+      "loss": 0.0003035278618335724,
+      "step": 5200
+    },
+    {
+      "epoch": 1.358554342173687,
+      "grad_norm": 0.005003906786441803,
+      "learning_rate": 3.7855072463768116e-05,
+      "loss": 0.0003134777024388313,
+      "step": 5225
+    },
+    {
+      "epoch": 1.3650546021840873,
+      "grad_norm": 0.0019250995246693492,
+      "learning_rate": 3.803623188405797e-05,
+      "loss": 0.00032027017325162886,
+      "step": 5250
+    },
+    {
+      "epoch": 1.3715548621944877,
+      "grad_norm": 0.0020014136098325253,
+      "learning_rate": 3.821739130434783e-05,
+      "loss": 0.0003059590607881546,
+      "step": 5275
+    },
+    {
+      "epoch": 1.3780551222048882,
+      "grad_norm": 0.0023152746725827456,
+      "learning_rate": 3.839855072463768e-05,
+      "loss": 0.00031596384942531584,
+      "step": 5300
+    },
+    {
+      "epoch": 1.3845553822152885,
+      "grad_norm": 0.002091446192935109,
+      "learning_rate": 3.857971014492754e-05,
+      "loss": 0.0003000116907060146,
+      "step": 5325
+    },
+    {
+      "epoch": 1.391055642225689,
+      "grad_norm": 0.0017833883175626397,
+      "learning_rate": 3.876086956521739e-05,
+      "loss": 0.0003047176636755466,
+      "step": 5350
+    },
+    {
+      "epoch": 1.3975559022360895,
+      "grad_norm": 0.0011746827512979507,
+      "learning_rate": 3.894202898550725e-05,
+      "loss": 0.00030552351847290993,
+      "step": 5375
+    },
+    {
+      "epoch": 1.4040561622464898,
+      "grad_norm": 0.0025903950445353985,
+      "learning_rate": 3.91231884057971e-05,
+      "loss": 0.0002794661931693554,
+      "step": 5400
+    },
+    {
+      "epoch": 1.4105564222568903,
+      "grad_norm": 0.0011848583817481995,
+      "learning_rate": 3.930434782608696e-05,
+      "loss": 0.00027751639485359194,
+      "step": 5425
+    },
+    {
+      "epoch": 1.4170566822672908,
+      "grad_norm": 0.0022143861278891563,
+      "learning_rate": 3.948550724637681e-05,
+      "loss": 0.0003024022839963436,
+      "step": 5450
+    },
+    {
+      "epoch": 1.423556942277691,
+      "grad_norm": 0.0025171921588480473,
+      "learning_rate": 3.966666666666667e-05,
+      "loss": 0.00028847914189100265,
+      "step": 5475
+    },
+    {
+      "epoch": 1.4300572022880915,
+      "grad_norm": 0.0015157477464526892,
+      "learning_rate": 3.9847826086956524e-05,
+      "loss": 0.00025584105402231216,
+      "step": 5500
+    },
+    {
+      "epoch": 1.436557462298492,
+      "grad_norm": 0.0020852391608059406,
+      "learning_rate": 4.002898550724638e-05,
+      "loss": 0.0003291204199194908,
+      "step": 5525
+    },
+    {
+      "epoch": 1.4430577223088923,
+      "grad_norm": 0.003646504133939743,
+      "learning_rate": 4.0210144927536236e-05,
+      "loss": 0.00030050855129957197,
+      "step": 5550
+    },
+    {
+      "epoch": 1.4495579823192928,
+      "grad_norm": 0.001804452738724649,
+      "learning_rate": 4.039130434782609e-05,
+      "loss": 0.00027791304513812064,
+      "step": 5575
+    },
+    {
+      "epoch": 1.4560582423296933,
+      "grad_norm": 0.0014344255905598402,
+      "learning_rate": 4.057246376811594e-05,
+      "loss": 0.00030676303431391715,
+      "step": 5600
+    },
+    {
+      "epoch": 1.4625585023400935,
+      "grad_norm": 0.005396896041929722,
+      "learning_rate": 4.07536231884058e-05,
+      "loss": 0.0002854640409350395,
+      "step": 5625
+    },
+    {
+      "epoch": 1.469058762350494,
+      "grad_norm": 0.00498207937926054,
+      "learning_rate": 4.0934782608695654e-05,
+      "loss": 0.00028169915080070493,
+      "step": 5650
+    },
+    {
+      "epoch": 1.4755590223608945,
+      "grad_norm": 0.0011505525326356292,
+      "learning_rate": 4.111594202898551e-05,
+      "loss": 0.00032645512372255327,
+      "step": 5675
+    },
+    {
+      "epoch": 1.4820592823712948,
+      "grad_norm": 0.007459944114089012,
+      "learning_rate": 4.1297101449275366e-05,
+      "loss": 0.0002781330794095993,
+      "step": 5700
+    },
+    {
+      "epoch": 1.4885595423816953,
+      "grad_norm": 0.002072904258966446,
+      "learning_rate": 4.147826086956522e-05,
+      "loss": 0.00028281256556510925,
+      "step": 5725
+    },
+    {
+      "epoch": 1.4950598023920958,
+      "grad_norm": 0.0022271168418228626,
+      "learning_rate": 4.165942028985507e-05,
+      "loss": 0.00030599093064665795,
+      "step": 5750
+    },
+    {
+      "epoch": 1.501560062402496,
+      "grad_norm": 0.0026721549220383167,
+      "learning_rate": 4.184057971014493e-05,
+      "loss": 0.0002837671712040901,
+      "step": 5775
+    },
+    {
+      "epoch": 1.5080603224128966,
+      "grad_norm": 0.001238685566931963,
+      "learning_rate": 4.2021739130434784e-05,
+      "loss": 0.00027412747964262964,
+      "step": 5800
+    },
+    {
+      "epoch": 1.514560582423297,
+      "grad_norm": 0.0008851462625898421,
+      "learning_rate": 4.220289855072464e-05,
+      "loss": 0.0002843565493822098,
+      "step": 5825
+    },
+    {
+      "epoch": 1.5210608424336973,
+      "grad_norm": 0.004694326780736446,
+      "learning_rate": 4.23840579710145e-05,
+      "loss": 0.0002548668347299099,
+      "step": 5850
+    },
+    {
+      "epoch": 1.5275611024440978,
+      "grad_norm": 0.0029512110631912947,
+      "learning_rate": 4.256521739130435e-05,
+      "loss": 0.00030956786125898363,
+      "step": 5875
+    },
+    {
+      "epoch": 1.5340613624544983,
+      "grad_norm": 0.003379747737199068,
+      "learning_rate": 4.274637681159421e-05,
+      "loss": 0.00031427673995494844,
+      "step": 5900
+    },
+    {
+      "epoch": 1.5405616224648986,
+      "grad_norm": 0.005398206412792206,
+      "learning_rate": 4.2927536231884055e-05,
+      "loss": 0.0002831252105534077,
+      "step": 5925
+    },
+    {
+      "epoch": 1.5470618824752989,
+      "grad_norm": 0.009443123824894428,
+      "learning_rate": 4.3108695652173915e-05,
+      "loss": 0.0003003663755953312,
+      "step": 5950
+    },
+    {
+      "epoch": 1.5535621424856996,
+      "grad_norm": 0.005164314527064562,
+      "learning_rate": 4.328985507246377e-05,
+      "loss": 0.00030718352645635604,
+      "step": 5975
+    },
+    {
+      "epoch": 1.5600624024960998,
+      "grad_norm": 0.00410798192024231,
+      "learning_rate": 4.347101449275363e-05,
+      "loss": 0.0003109707310795784,
+      "step": 6000
+    },
+    {
+      "epoch": 1.5665626625065001,
+      "grad_norm": 0.004693225026130676,
+      "learning_rate": 4.365217391304348e-05,
+      "loss": 0.0002868692018091679,
+      "step": 6025
+    },
+    {
+      "epoch": 1.5730629225169008,
+      "grad_norm": 0.001609576283954084,
+      "learning_rate": 4.383333333333334e-05,
+      "loss": 0.00031623709946870804,
+      "step": 6050
+    },
+    {
+      "epoch": 1.579563182527301,
+      "grad_norm": 0.0011194883845746517,
+      "learning_rate": 4.401449275362319e-05,
+      "loss": 0.00029221972450613974,
+      "step": 6075
+    },
+    {
+      "epoch": 1.5860634425377014,
+      "grad_norm": 0.0020942233968526125,
+      "learning_rate": 4.4195652173913045e-05,
+      "loss": 0.00028448238968849183,
+      "step": 6100
+    },
+    {
+      "epoch": 1.592563702548102,
+      "grad_norm": 0.0007156800711527467,
+      "learning_rate": 4.43768115942029e-05,
+      "loss": 0.0003088047541677952,
+      "step": 6125
+    },
+    {
+      "epoch": 1.5990639625585024,
+      "grad_norm": 0.0010163492988795042,
+      "learning_rate": 4.455797101449276e-05,
+      "loss": 0.00024831485003232954,
+      "step": 6150
+    },
+    {
+      "epoch": 1.6055642225689026,
+      "grad_norm": 0.0011129551567137241,
+      "learning_rate": 4.473913043478261e-05,
+      "loss": 0.0002939954400062561,
+      "step": 6175
+    },
+    {
+      "epoch": 1.6120644825793031,
+      "grad_norm": 0.0014835961628705263,
+      "learning_rate": 4.492028985507247e-05,
+      "loss": 0.00027918266132473943,
+      "step": 6200
+    },
+    {
+      "epoch": 1.6185647425897036,
+      "grad_norm": 0.005272570997476578,
+      "learning_rate": 4.510144927536232e-05,
+      "loss": 0.0003057974763214588,
+      "step": 6225
+    },
+    {
+      "epoch": 1.625065002600104,
+      "grad_norm": 0.004790041130036116,
+      "learning_rate": 4.5282608695652175e-05,
+      "loss": 0.0002860172651708126,
+      "step": 6250
+    },
+    {
+      "epoch": 1.6315652626105044,
+      "grad_norm": 0.0022453030105680227,
+      "learning_rate": 4.546376811594203e-05,
+      "loss": 0.0002870689146220684,
+      "step": 6275
+    },
+    {
+      "epoch": 1.6380655226209049,
+      "grad_norm": 0.0024950592778623104,
+      "learning_rate": 4.564492753623189e-05,
+      "loss": 0.0003029079921543598,
+      "step": 6300
+    },
+    {
+      "epoch": 1.6445657826313052,
+      "grad_norm": 0.0020117126405239105,
+      "learning_rate": 4.582608695652174e-05,
+      "loss": 0.0003067331947386265,
+      "step": 6325
+    },
+    {
+      "epoch": 1.6510660426417056,
+      "grad_norm": 0.0009332145564258099,
+      "learning_rate": 4.60072463768116e-05,
+      "loss": 0.00027948716655373573,
+      "step": 6350
+    },
+    {
+      "epoch": 1.6575663026521061,
+      "grad_norm": 0.0,
+      "learning_rate": 4.618840579710145e-05,
+      "loss": 0.00028807364404201505,
+      "step": 6375
+    },
+    {
+      "epoch": 1.6640665626625064,
+      "grad_norm": 0.0032374351285398006,
+      "learning_rate": 4.6369565217391306e-05,
+      "loss": 0.00027499660849571227,
+      "step": 6400
+    },
+    {
+      "epoch": 1.670566822672907,
+      "grad_norm": 0.0030327781569212675,
+      "learning_rate": 4.655072463768116e-05,
+      "loss": 0.0002995528653264046,
+      "step": 6425
+    },
+    {
+      "epoch": 1.6770670826833074,
+      "grad_norm": 0.0018715931801125407,
+      "learning_rate": 4.673188405797102e-05,
+      "loss": 0.00024274324998259543,
+      "step": 6450
+    },
+    {
+      "epoch": 1.6835673426937077,
+      "grad_norm": 0.001741762738674879,
+      "learning_rate": 4.691304347826087e-05,
+      "loss": 0.00027031151577830317,
+      "step": 6475
+    },
+    {
+      "epoch": 1.6900676027041082,
+      "grad_norm": 0.0029014807660132647,
+      "learning_rate": 4.7094202898550724e-05,
+      "loss": 0.0002902892790734768,
+      "step": 6500
+    },
+    {
+      "epoch": 1.6965678627145087,
+      "grad_norm": 0.0006979657919146121,
+      "learning_rate": 4.727536231884058e-05,
+      "loss": 0.0002635803818702698,
+      "step": 6525
+    },
+    {
+      "epoch": 1.703068122724909,
+      "grad_norm": 0.0022528229746967554,
+      "learning_rate": 4.7456521739130436e-05,
+      "loss": 0.0003087851218879223,
+      "step": 6550
+    },
+    {
+      "epoch": 1.7095683827353094,
+      "grad_norm": 0.005399159621447325,
+      "learning_rate": 4.7637681159420296e-05,
+      "loss": 0.0002844073623418808,
+      "step": 6575
+    },
+    {
+      "epoch": 1.71606864274571,
+      "grad_norm": 0.0027920063585042953,
+      "learning_rate": 4.781884057971014e-05,
+      "loss": 0.00029874764382839205,
+      "step": 6600
+    },
+    {
+      "epoch": 1.7225689027561102,
+      "grad_norm": 0.0016516833566129208,
+      "learning_rate": 4.8e-05,
+      "loss": 0.0002489457279443741,
+      "step": 6625
+    },
+    {
+      "epoch": 1.7290691627665107,
+      "grad_norm": 0.0023696268908679485,
+      "learning_rate": 4.8181159420289854e-05,
+      "loss": 0.00027289777994155885,
+      "step": 6650
+    },
+    {
+      "epoch": 1.7355694227769112,
+      "grad_norm": 0.0017222384922206402,
+      "learning_rate": 4.8362318840579714e-05,
+      "loss": 0.0002775443904101849,
+      "step": 6675
+    },
+    {
+      "epoch": 1.7420696827873114,
+      "grad_norm": 0.0018073809333145618,
+      "learning_rate": 4.8543478260869566e-05,
+      "loss": 0.000254177488386631,
+      "step": 6700
+    },
+    {
+      "epoch": 1.748569942797712,
+      "grad_norm": 0.0030928237829357386,
+      "learning_rate": 4.8724637681159426e-05,
+      "loss": 0.00029973674565553666,
+      "step": 6725
+    },
+    {
+      "epoch": 1.7550702028081124,
+      "grad_norm": 0.003188478061929345,
+      "learning_rate": 4.890579710144928e-05,
+      "loss": 0.00026500694453716276,
+      "step": 6750
+    },
+    {
+      "epoch": 1.7615704628185127,
+      "grad_norm": 0.0006110169924795628,
+      "learning_rate": 4.908695652173913e-05,
+      "loss": 0.00028070401400327684,
+      "step": 6775
+    },
+    {
+      "epoch": 1.7680707228289132,
+      "grad_norm": 0.002211988903582096,
+      "learning_rate": 4.9268115942028984e-05,
+      "loss": 0.00031072892248630525,
+      "step": 6800
+    },
+    {
+      "epoch": 1.7745709828393137,
+      "grad_norm": 0.002339594531804323,
+      "learning_rate": 4.9449275362318844e-05,
+      "loss": 0.00026881076395511626,
+      "step": 6825
+    },
+    {
+      "epoch": 1.781071242849714,
+      "grad_norm": 0.001855787937529385,
+      "learning_rate": 4.96304347826087e-05,
+      "loss": 0.00028889000415802005,
+      "step": 6850
+    },
+    {
+      "epoch": 1.7875715028601145,
+      "grad_norm": 0.004768841899931431,
+      "learning_rate": 4.9811594202898556e-05,
+      "loss": 0.0002878274954855442,
+      "step": 6875
+    },
+    {
+      "epoch": 1.794071762870515,
+      "grad_norm": 0.00204999977722764,
+      "learning_rate": 4.999275362318841e-05,
+      "loss": 0.00026362594217062,
+      "step": 6900
+    },
+    {
+      "epoch": 1.794071762870515,
+      "eval_loss": 0.001298407674767077,
+      "eval_runtime": 7.2552,
+      "eval_samples_per_second": 689.164,
+      "eval_steps_per_second": 3.721,
+      "step": 6900
+    },
+    {
+      "epoch": 1.794071762870515,
+      "eval_holdout_avg_strict_f1": 0.015627198422668045,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.03821876351059231,
+      "eval_holdout_nemotron_pii_strict_f1": 0.029488364277913136,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.06010272101409682,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.044089333846746244,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 6900
+    },
+    {
+      "epoch": 1.8005720228809152,
+      "grad_norm": 0.0017703310586512089,
+      "learning_rate": 4.999998157323963e-05,
+      "loss": 0.00029657723382115365,
+      "step": 6925
+    },
+    {
+      "epoch": 1.8070722828913155,
+      "grad_norm": 0.002401393372565508,
+      "learning_rate": 4.999992318987079e-05,
+      "loss": 0.0002728748507797718,
+      "step": 6950
+    },
+    {
+      "epoch": 1.8135725429017162,
+      "grad_norm": 0.0010186949511989951,
+      "learning_rate": 4.999982481799595e-05,
+      "loss": 0.0002660237066447735,
+      "step": 6975
+    },
+    {
+      "epoch": 1.8200728029121165,
+      "grad_norm": 0.002337829675525427,
+      "learning_rate": 4.999968645777246e-05,
+      "loss": 0.0002816925011575222,
+      "step": 7000
+    },
+    {
+      "epoch": 1.8265730629225168,
+      "grad_norm": 0.001261275727301836,
+      "learning_rate": 4.999950810942164e-05,
+      "loss": 0.0002926921471953392,
+      "step": 7025
+    },
+    {
+      "epoch": 1.8330733229329175,
+      "grad_norm": 0.001112476922571659,
+      "learning_rate": 4.999928977322876e-05,
+      "loss": 0.00028484620153903964,
+      "step": 7050
+    },
+    {
+      "epoch": 1.8395735829433177,
+      "grad_norm": 0.004276876803487539,
+      "learning_rate": 4.999903144954307e-05,
+      "loss": 0.0002890905737876892,
+      "step": 7075
+    },
+    {
+      "epoch": 1.846073842953718,
+      "grad_norm": 0.001109086791984737,
+      "learning_rate": 4.9998733138777766e-05,
+      "loss": 0.00028899595141410826,
+      "step": 7100
+    },
+    {
+      "epoch": 1.8525741029641187,
+      "grad_norm": 0.003312828252092004,
+      "learning_rate": 4.999839484140999e-05,
+      "loss": 0.00028695842251181603,
+      "step": 7125
+    },
+    {
+      "epoch": 1.859074362974519,
+      "grad_norm": 0.0036718384362757206,
+      "learning_rate": 4.9998016557980896e-05,
+      "loss": 0.0002957011759281158,
+      "step": 7150
+    },
+    {
+      "epoch": 1.8655746229849193,
+      "grad_norm": 0.0010820156894624233,
+      "learning_rate": 4.999759828909555e-05,
+      "loss": 0.0002718496508896351,
+      "step": 7175
+    },
+    {
+      "epoch": 1.8720748829953198,
+      "grad_norm": 0.0010525485267862678,
+      "learning_rate": 4.9997140035422984e-05,
+      "loss": 0.0002739377319812775,
+      "step": 7200
+    },
+    {
+      "epoch": 1.8785751430057203,
+      "grad_norm": 0.003618677845224738,
+      "learning_rate": 4.9996641797696206e-05,
+      "loss": 0.00028557723388075826,
+      "step": 7225
+    },
+    {
+      "epoch": 1.8850754030161205,
+      "grad_norm": 0.0034078890457749367,
+      "learning_rate": 4.999610357671217e-05,
+      "loss": 0.0002824321761727333,
+      "step": 7250
+    },
+    {
+      "epoch": 1.891575663026521,
+      "grad_norm": 0.0026714326813817024,
+      "learning_rate": 4.999552537333178e-05,
+      "loss": 0.000270571056753397,
+      "step": 7275
+    },
+    {
+      "epoch": 1.8980759230369215,
+      "grad_norm": 0.0030453635845333338,
+      "learning_rate": 4.99949071884799e-05,
+      "loss": 0.0002762207575142384,
+      "step": 7300
+    },
+    {
+      "epoch": 1.9045761830473218,
+      "grad_norm": 0.002129437867552042,
+      "learning_rate": 4.999424902314534e-05,
+      "loss": 0.00029886599630117415,
+      "step": 7325
+    },
+    {
+      "epoch": 1.9110764430577223,
+      "grad_norm": 0.0017199000576511025,
+      "learning_rate": 4.9993550878380876e-05,
+      "loss": 0.00024105872958898544,
+      "step": 7350
+    },
+    {
+      "epoch": 1.9175767030681228,
+      "grad_norm": 0.0019791550002992153,
+      "learning_rate": 4.999281275530321e-05,
+      "loss": 0.0002700541727244854,
+      "step": 7375
+    },
+    {
+      "epoch": 1.924076963078523,
+      "grad_norm": 0.002077782526612282,
+      "learning_rate": 4.9992034655093016e-05,
+      "loss": 0.0002733784168958664,
+      "step": 7400
+    },
+    {
+      "epoch": 1.9305772230889235,
+      "grad_norm": 0.0021692560985684395,
+      "learning_rate": 4.9991216578994894e-05,
+      "loss": 0.0002755324728786945,
+      "step": 7425
+    },
+    {
+      "epoch": 1.937077483099324,
+      "grad_norm": 0.0012318184599280357,
+      "learning_rate": 4.999035852831739e-05,
+      "loss": 0.00027664078399538996,
+      "step": 7450
+    },
+    {
+      "epoch": 1.9435777431097243,
+      "grad_norm": 0.0022035727743059397,
+      "learning_rate": 4.998946050443299e-05,
+      "loss": 0.00027742795646190643,
+      "step": 7475
+    },
+    {
+      "epoch": 1.9500780031201248,
+      "grad_norm": 0.003872137051075697,
+      "learning_rate": 4.9988522508778144e-05,
+      "loss": 0.0002809068188071251,
+      "step": 7500
+    },
+    {
+      "epoch": 1.9565782631305253,
+      "grad_norm": 0.005694707855582237,
+      "learning_rate": 4.9987544542853207e-05,
+      "loss": 0.0002876455523073673,
+      "step": 7525
+    },
+    {
+      "epoch": 1.9630785231409256,
+      "grad_norm": 0.0018619212787598372,
+      "learning_rate": 4.9986526608222475e-05,
+      "loss": 0.0002589203044772148,
+      "step": 7550
+    },
+    {
+      "epoch": 1.969578783151326,
+      "grad_norm": 0.0019440543837845325,
+      "learning_rate": 4.998546870651418e-05,
+      "loss": 0.0002764608152210712,
+      "step": 7575
+    },
+    {
+      "epoch": 1.9760790431617266,
+      "grad_norm": 0.0015528312651440501,
+      "learning_rate": 4.998437083942049e-05,
+      "loss": 0.00025359803810715676,
+      "step": 7600
+    },
+    {
+      "epoch": 1.9825793031721268,
+      "grad_norm": 0.004921973682940006,
+      "learning_rate": 4.998323300869749e-05,
+      "loss": 0.0002727708593010902,
+      "step": 7625
+    },
+    {
+      "epoch": 1.9890795631825273,
+      "grad_norm": 0.001523703569546342,
+      "learning_rate": 4.99820552161652e-05,
+      "loss": 0.0002619338594377041,
+      "step": 7650
+    },
+    {
+      "epoch": 1.9955798231929278,
+      "grad_norm": 0.004311833530664444,
+      "learning_rate": 4.998083746370754e-05,
+      "loss": 0.0002619973011314869,
+      "step": 7675
+    },
+    {
+      "epoch": 2.002080083203328,
+      "grad_norm": 0.0030685316305607557,
+      "learning_rate": 4.9979579753272364e-05,
+      "loss": 0.00028862494975328445,
+      "step": 7700
+    },
+    {
+      "epoch": 2.0085803432137284,
+      "grad_norm": 0.0011595230316743255,
+      "learning_rate": 4.997828208687144e-05,
+      "loss": 0.00024254759773612022,
+      "step": 7725
+    },
+    {
+      "epoch": 2.015080603224129,
+      "grad_norm": 0.0019829794764518738,
+      "learning_rate": 4.997694446658044e-05,
+      "loss": 0.000305801797658205,
+      "step": 7750
+    },
+    {
+      "epoch": 2.0215808632345293,
+      "grad_norm": 0.002637671772390604,
+      "learning_rate": 4.997556689453895e-05,
+      "loss": 0.00028775529935956003,
+      "step": 7775
+    },
+    {
+      "epoch": 2.0280811232449296,
+      "grad_norm": 0.004940721672028303,
+      "learning_rate": 4.997414937295046e-05,
+      "loss": 0.000274547953158617,
+      "step": 7800
+    },
+    {
+      "epoch": 2.0345813832553303,
+      "grad_norm": 0.0036540755536407232,
+      "learning_rate": 4.997269190408235e-05,
+      "loss": 0.00027498628944158557,
+      "step": 7825
+    },
+    {
+      "epoch": 2.0410816432657306,
+      "grad_norm": 0.004672015551477671,
+      "learning_rate": 4.997119449026592e-05,
+      "loss": 0.00029294028878211974,
+      "step": 7850
+    },
+    {
+      "epoch": 2.047581903276131,
+      "grad_norm": 0.00043306767474859953,
+      "learning_rate": 4.996965713389635e-05,
+      "loss": 0.0002886112034320831,
+      "step": 7875
+    },
+    {
+      "epoch": 2.0540821632865316,
+      "grad_norm": 0.001710674841888249,
+      "learning_rate": 4.99680798374327e-05,
+      "loss": 0.00029880210757255555,
+      "step": 7900
+    },
+    {
+      "epoch": 2.060582423296932,
+      "grad_norm": 0.00074931321432814,
+      "learning_rate": 4.9966462603397946e-05,
+      "loss": 0.0002872174419462681,
+      "step": 7925
+    },
+    {
+      "epoch": 2.067082683307332,
+      "grad_norm": 0.0014533158391714096,
+      "learning_rate": 4.9964805434378906e-05,
+      "loss": 0.00029754893854260444,
+      "step": 7950
+    },
+    {
+      "epoch": 2.073582943317733,
+      "grad_norm": 0.0011532707139849663,
+      "learning_rate": 4.996310833302631e-05,
+      "loss": 0.000272640660405159,
+      "step": 7975
+    },
+    {
+      "epoch": 2.080083203328133,
+      "grad_norm": 0.002353905700147152,
+      "learning_rate": 4.996137130205474e-05,
+      "loss": 0.0002785234898328781,
+      "step": 8000
+    },
+    {
+      "epoch": 2.0865834633385334,
+      "grad_norm": 0.00506283063441515,
+      "learning_rate": 4.995959434424265e-05,
+      "loss": 0.00026953674852848053,
+      "step": 8025
+    },
+    {
+      "epoch": 2.093083723348934,
+      "grad_norm": 0.006331348791718483,
+      "learning_rate": 4.9957777462432394e-05,
+      "loss": 0.0002770484797656536,
+      "step": 8050
+    },
+    {
+      "epoch": 2.0995839833593344,
+      "grad_norm": 0.0030248507391661406,
+      "learning_rate": 4.9955920659530124e-05,
+      "loss": 0.00031356040388345717,
+      "step": 8075
+    },
+    {
+      "epoch": 2.1060842433697347,
+      "grad_norm": 0.0006748432642780244,
+      "learning_rate": 4.99540239385059e-05,
+      "loss": 0.0002919316105544567,
+      "step": 8100
+    },
+    {
+      "epoch": 2.1125845033801354,
+      "grad_norm": 0.006296483334153891,
+      "learning_rate": 4.99520873023936e-05,
+      "loss": 0.0002602156065404415,
+      "step": 8125
+    },
+    {
+      "epoch": 2.1190847633905356,
+      "grad_norm": 0.0015592221170663834,
+      "learning_rate": 4.9950110754290966e-05,
+      "loss": 0.0002915428206324577,
+      "step": 8150
+    },
+    {
+      "epoch": 2.125585023400936,
+      "grad_norm": 0.0007437528111040592,
+      "learning_rate": 4.994809429735957e-05,
+      "loss": 0.00032635748386383056,
+      "step": 8175
+    },
+    {
+      "epoch": 2.1320852834113366,
+      "grad_norm": 0.0034404215402901173,
+      "learning_rate": 4.994603793482484e-05,
+      "loss": 0.00027954017743468283,
+      "step": 8200
+    },
+    {
+      "epoch": 2.138585543421737,
+      "grad_norm": 0.0014827263075858355,
+      "learning_rate": 4.994394166997601e-05,
+      "loss": 0.0002614803984761238,
+      "step": 8225
+    },
+    {
+      "epoch": 2.145085803432137,
+      "grad_norm": 0.001980932429432869,
+      "learning_rate": 4.994180550616615e-05,
+      "loss": 0.0002945771440863609,
+      "step": 8250
+    },
+    {
+      "epoch": 2.151586063442538,
+      "grad_norm": 0.0009216663311235607,
+      "learning_rate": 4.993962944681216e-05,
+      "loss": 0.00029664896428585055,
+      "step": 8275
+    },
+    {
+      "epoch": 2.158086323452938,
+      "grad_norm": 0.0025570986326783895,
+      "learning_rate": 4.993741349539474e-05,
+      "loss": 0.00026415424421429636,
+      "step": 8300
+    },
+    {
+      "epoch": 2.1645865834633384,
+      "grad_norm": 0.0017446415731683373,
+      "learning_rate": 4.99351576554584e-05,
+      "loss": 0.0002811905182898045,
+      "step": 8325
+    },
+    {
+      "epoch": 2.171086843473739,
+      "grad_norm": 0.002114599570631981,
+      "learning_rate": 4.9932861930611454e-05,
+      "loss": 0.00027634453028440475,
+      "step": 8350
+    },
+    {
+      "epoch": 2.1775871034841394,
+      "grad_norm": 0.0073842755518853664,
+      "learning_rate": 4.993052632452603e-05,
+      "loss": 0.00027546945959329607,
+      "step": 8375
+    },
+    {
+      "epoch": 2.1840873634945397,
+      "grad_norm": 0.0021200620103627443,
+      "learning_rate": 4.9928150840938034e-05,
+      "loss": 0.00025118796154856684,
+      "step": 8400
+    },
+    {
+      "epoch": 2.19058762350494,
+      "grad_norm": 0.007495725993067026,
+      "learning_rate": 4.9925735483647144e-05,
+      "loss": 0.00026308244094252586,
+      "step": 8425
+    },
+    {
+      "epoch": 2.1970878835153407,
+      "grad_norm": 0.0017843242967501283,
+      "learning_rate": 4.9923280256516856e-05,
+      "loss": 0.0002609914168715477,
+      "step": 8450
+    },
+    {
+      "epoch": 2.203588143525741,
+      "grad_norm": 0.0025775767862796783,
+      "learning_rate": 4.99207851634744e-05,
+      "loss": 0.0002497840486466885,
+      "step": 8475
+    },
+    {
+      "epoch": 2.210088403536141,
+      "grad_norm": 0.0014296849258244038,
+      "learning_rate": 4.991825020851079e-05,
+      "loss": 0.00028953062370419504,
+      "step": 8500
+    },
+    {
+      "epoch": 2.216588663546542,
+      "grad_norm": 0.0021426023449748755,
+      "learning_rate": 4.991567539568081e-05,
+      "loss": 0.0002593442238867283,
+      "step": 8525
+    },
+    {
+      "epoch": 2.223088923556942,
+      "grad_norm": 0.0016822812613099813,
+      "learning_rate": 4.991306072910299e-05,
+      "loss": 0.0002584538049995899,
+      "step": 8550
+    },
+    {
+      "epoch": 2.2295891835673425,
+      "grad_norm": 0.0011391935404390097,
+      "learning_rate": 4.991040621295959e-05,
+      "loss": 0.00028433525934815404,
+      "step": 8575
+    },
+    {
+      "epoch": 2.236089443577743,
+      "grad_norm": 0.0033121451269835234,
+      "learning_rate": 4.9907711851496655e-05,
+      "loss": 0.0002397463284432888,
+      "step": 8600
+    },
+    {
+      "epoch": 2.2425897035881435,
+      "grad_norm": 0.0016361881280317903,
+      "learning_rate": 4.9904977649023906e-05,
+      "loss": 0.0002411177009344101,
+      "step": 8625
+    },
+    {
+      "epoch": 2.2490899635985437,
+      "grad_norm": 0.0020404800307005644,
+      "learning_rate": 4.9902203609914855e-05,
+      "loss": 0.0002576117590069771,
+      "step": 8650
+    },
+    {
+      "epoch": 2.2555902236089445,
+      "grad_norm": 0.0017065717838704586,
+      "learning_rate": 4.989938973860668e-05,
+      "loss": 0.000222860686480999,
+      "step": 8675
+    },
+    {
+      "epoch": 2.2620904836193447,
+      "grad_norm": 0.0007460704655386508,
+      "learning_rate": 4.989653603960032e-05,
+      "loss": 0.00022550849243998527,
+      "step": 8700
+    },
+    {
+      "epoch": 2.268590743629745,
+      "grad_norm": 0.0027334203477948904,
+      "learning_rate": 4.989364251746036e-05,
+      "loss": 0.0002485204301774502,
+      "step": 8725
+    },
+    {
+      "epoch": 2.2750910036401457,
+      "grad_norm": 0.0027240319177508354,
+      "learning_rate": 4.989070917681515e-05,
+      "loss": 0.00024812601506710055,
+      "step": 8750
+    },
+    {
+      "epoch": 2.281591263650546,
+      "grad_norm": 0.00138418551068753,
+      "learning_rate": 4.988773602235669e-05,
+      "loss": 0.00022993290796875952,
+      "step": 8775
+    },
+    {
+      "epoch": 2.2880915236609463,
+      "grad_norm": 0.002026743022724986,
+      "learning_rate": 4.9884723058840665e-05,
+      "loss": 0.00023552784696221352,
+      "step": 8800
+    },
+    {
+      "epoch": 2.294591783671347,
+      "grad_norm": 0.0021410661283880472,
+      "learning_rate": 4.988167029108646e-05,
+      "loss": 0.00024767260998487473,
+      "step": 8825
+    },
+    {
+      "epoch": 2.3010920436817472,
+      "grad_norm": 0.0016540430951863527,
+      "learning_rate": 4.9878577723977116e-05,
+      "loss": 0.0002473931759595871,
+      "step": 8850
+    },
+    {
+      "epoch": 2.3075923036921475,
+      "grad_norm": 0.0019779433496296406,
+      "learning_rate": 4.987544536245931e-05,
+      "loss": 0.00025823218747973443,
+      "step": 8875
+    },
+    {
+      "epoch": 2.3140925637025482,
+      "grad_norm": 0.0012516207061707973,
+      "learning_rate": 4.987227321154344e-05,
+      "loss": 0.00022903639823198318,
+      "step": 8900
+    },
+    {
+      "epoch": 2.3205928237129485,
+      "grad_norm": 0.003492312040179968,
+      "learning_rate": 4.986906127630346e-05,
+      "loss": 0.00022529695183038712,
+      "step": 8925
+    },
+    {
+      "epoch": 2.3270930837233488,
+      "grad_norm": 0.009140203706920147,
+      "learning_rate": 4.986580956187703e-05,
+      "loss": 0.0002398119121789932,
+      "step": 8950
+    },
+    {
+      "epoch": 2.3335933437337495,
+      "grad_norm": 0.0006524514756165445,
+      "learning_rate": 4.9862518073465405e-05,
+      "loss": 0.00024563953280448914,
+      "step": 8975
+    },
+    {
+      "epoch": 2.3400936037441498,
+      "grad_norm": 0.0013254332588985562,
+      "learning_rate": 4.985918681633348e-05,
+      "loss": 0.00023173581808805466,
+      "step": 9000
+    },
+    {
+      "epoch": 2.34659386375455,
+      "grad_norm": 0.0008085084264166653,
+      "learning_rate": 4.985581579580973e-05,
+      "loss": 0.00023689467459917069,
+      "step": 9025
+    },
+    {
+      "epoch": 2.3530941237649508,
+      "grad_norm": 0.001455951132811606,
+      "learning_rate": 4.985240501728628e-05,
+      "loss": 0.00023865673691034317,
+      "step": 9050
+    },
+    {
+      "epoch": 2.359594383775351,
+      "grad_norm": 0.002361190738156438,
+      "learning_rate": 4.984895448621881e-05,
+      "loss": 0.00022536056116223335,
+      "step": 9075
+    },
+    {
+      "epoch": 2.3660946437857513,
+      "grad_norm": 0.0020702260080724955,
+      "learning_rate": 4.98454642081266e-05,
+      "loss": 0.00024512186646461485,
+      "step": 9100
+    },
+    {
+      "epoch": 2.372594903796152,
+      "grad_norm": 0.0050494009628891945,
+      "learning_rate": 4.984193418859251e-05,
+      "loss": 0.00023177556693553924,
+      "step": 9125
+    },
+    {
+      "epoch": 2.3790951638065523,
+      "grad_norm": 0.004237701650708914,
+      "learning_rate": 4.9838364433262965e-05,
+      "loss": 0.00028020873665809634,
+      "step": 9150
+    },
+    {
+      "epoch": 2.3855954238169526,
+      "grad_norm": 0.0010891571873798966,
+      "learning_rate": 4.983475494784795e-05,
+      "loss": 0.00028782278299331664,
+      "step": 9175
+    },
+    {
+      "epoch": 2.3920956838273533,
+      "grad_norm": 0.0018918045097962022,
+      "learning_rate": 4.9831105738120995e-05,
+      "loss": 0.00023671358823776245,
+      "step": 9200
+    },
+    {
+      "epoch": 2.3985959438377535,
+      "grad_norm": 0.004604646936058998,
+      "learning_rate": 4.982741680991919e-05,
+      "loss": 0.00025982240214943885,
+      "step": 9225
+    },
+    {
+      "epoch": 2.405096203848154,
+      "grad_norm": 0.012237738817930222,
+      "learning_rate": 4.982368816914314e-05,
+      "loss": 0.00020902009680867195,
+      "step": 9250
+    },
+    {
+      "epoch": 2.4115964638585545,
+      "grad_norm": 0.0015675468603149056,
+      "learning_rate": 4.981991982175695e-05,
+      "loss": 0.0002525036036968231,
+      "step": 9275
+    },
+    {
+      "epoch": 2.418096723868955,
+      "grad_norm": 0.0030031027272343636,
+      "learning_rate": 4.981611177378829e-05,
+      "loss": 0.0002297206223011017,
+      "step": 9300
+    },
+    {
+      "epoch": 2.424596983879355,
+      "grad_norm": 0.002510670106858015,
+      "learning_rate": 4.9812264031328285e-05,
+      "loss": 0.00025444598868489264,
+      "step": 9325
+    },
+    {
+      "epoch": 2.431097243889756,
+      "grad_norm": 0.0013309629866853356,
+      "learning_rate": 4.980837660053159e-05,
+      "loss": 0.00024253139272332192,
+      "step": 9350
+    },
+    {
+      "epoch": 2.437597503900156,
+      "grad_norm": 0.0027411924675107002,
+      "learning_rate": 4.980444948761631e-05,
+      "loss": 0.00024229643866419793,
+      "step": 9375
+    },
+    {
+      "epoch": 2.4440977639105563,
+      "grad_norm": 0.006203151773661375,
+      "learning_rate": 4.980048269886406e-05,
+      "loss": 0.00027989933267235755,
+      "step": 9400
+    },
+    {
+      "epoch": 2.450598023920957,
+      "grad_norm": 0.0031860952731221914,
+      "learning_rate": 4.979647624061987e-05,
+      "loss": 0.00025709541514515875,
+      "step": 9425
+    },
+    {
+      "epoch": 2.4570982839313573,
+      "grad_norm": 0.0018132183467969298,
+      "learning_rate": 4.9792430119292275e-05,
+      "loss": 0.0002631310559809208,
+      "step": 9450
+    },
+    {
+      "epoch": 2.4635985439417576,
+      "grad_norm": 0.0027568559162318707,
+      "learning_rate": 4.978834434135322e-05,
+      "loss": 0.00023871993646025656,
+      "step": 9475
+    },
+    {
+      "epoch": 2.4700988039521583,
+      "grad_norm": 0.0012175653828307986,
+      "learning_rate": 4.9784218913338096e-05,
+      "loss": 0.0002557317726314068,
+      "step": 9500
+    },
+    {
+      "epoch": 2.4765990639625586,
+      "grad_norm": 0.0024774980265647173,
+      "learning_rate": 4.9780053841845697e-05,
+      "loss": 0.0002469548024237156,
+      "step": 9525
+    },
+    {
+      "epoch": 2.483099323972959,
+      "grad_norm": 0.0024396195076406,
+      "learning_rate": 4.9775849133538254e-05,
+      "loss": 0.0002535165287554264,
+      "step": 9550
+    },
+    {
+      "epoch": 2.4895995839833596,
+      "grad_norm": 0.005489069037139416,
+      "learning_rate": 4.9771604795141385e-05,
+      "loss": 0.0002929968759417534,
+      "step": 9575
+    },
+    {
+      "epoch": 2.49609984399376,
+      "grad_norm": 0.003155334386974573,
+      "learning_rate": 4.976732083344411e-05,
+      "loss": 0.000257662869989872,
+      "step": 9600
+    },
+    {
+      "epoch": 2.50260010400416,
+      "grad_norm": 0.002911205869168043,
+      "learning_rate": 4.97629972552988e-05,
+      "loss": 0.000229572132229805,
+      "step": 9625
+    },
+    {
+      "epoch": 2.509100364014561,
+      "grad_norm": 0.0023384687956422567,
+      "learning_rate": 4.975863406762124e-05,
+      "loss": 0.00022791551426053048,
+      "step": 9650
+    },
+    {
+      "epoch": 2.515600624024961,
+      "grad_norm": 0.002526897471398115,
+      "learning_rate": 4.9754231277390514e-05,
+      "loss": 0.00026006758213043213,
+      "step": 9675
+    },
+    {
+      "epoch": 2.5221008840353614,
+      "grad_norm": 0.0011523400899022818,
+      "learning_rate": 4.974978889164911e-05,
+      "loss": 0.00023126760497689248,
+      "step": 9700
+    },
+    {
+      "epoch": 2.528601144045762,
+      "grad_norm": 0.0017160821007564664,
+      "learning_rate": 4.9745306917502805e-05,
+      "loss": 0.00023550502955913544,
+      "step": 9725
+    },
+    {
+      "epoch": 2.5351014040561624,
+      "grad_norm": 0.0010612220503389835,
+      "learning_rate": 4.9740785362120736e-05,
+      "loss": 0.00021429264917969705,
+      "step": 9750
+    },
+    {
+      "epoch": 2.5416016640665626,
+      "grad_norm": 0.00163923273794353,
+      "learning_rate": 4.973622423273533e-05,
+      "loss": 0.00024077022448182106,
+      "step": 9775
+    },
+    {
+      "epoch": 2.5481019240769633,
+      "grad_norm": 0.0012632396537810564,
+      "learning_rate": 4.973162353664231e-05,
+      "loss": 0.00023677242919802665,
+      "step": 9800
+    },
+    {
+      "epoch": 2.5546021840873636,
+      "grad_norm": 0.005001894664019346,
+      "learning_rate": 4.972698328120071e-05,
+      "loss": 0.00024106081575155257,
+      "step": 9825
+    },
+    {
+      "epoch": 2.561102444097764,
+      "grad_norm": 0.003063301322981715,
+      "learning_rate": 4.9722303473832824e-05,
+      "loss": 0.00022191539406776428,
+      "step": 9850
+    },
+    {
+      "epoch": 2.5676027041081646,
+      "grad_norm": 0.0019900454208254814,
+      "learning_rate": 4.97175841220242e-05,
+      "loss": 0.0002374211885035038,
+      "step": 9875
+    },
+    {
+      "epoch": 2.574102964118565,
+      "grad_norm": 0.001180443912744522,
+      "learning_rate": 4.971282523332368e-05,
+      "loss": 0.00026023676618933676,
+      "step": 9900
+    },
+    {
+      "epoch": 2.580603224128965,
+      "grad_norm": 0.002654899377375841,
+      "learning_rate": 4.9708026815343314e-05,
+      "loss": 0.0002488452754914761,
+      "step": 9925
+    },
+    {
+      "epoch": 2.5871034841393654,
+      "grad_norm": 0.0021628928370773792,
+      "learning_rate": 4.970318887575837e-05,
+      "loss": 0.00024352485314011574,
+      "step": 9950
+    },
+    {
+      "epoch": 2.593603744149766,
+      "grad_norm": 0.0020067659206688404,
+      "learning_rate": 4.969831142230737e-05,
+      "loss": 0.00021531779319047928,
+      "step": 9975
+    },
+    {
+      "epoch": 2.6001040041601664,
+      "grad_norm": 0.0031929572578519583,
+      "learning_rate": 4.969339446279202e-05,
+      "loss": 0.000210524033755064,
+      "step": 10000
+    },
+    {
+      "epoch": 2.6066042641705667,
+      "grad_norm": 0.004822343587875366,
+      "learning_rate": 4.968843800507721e-05,
+      "loss": 0.0002182518132030964,
+      "step": 10025
+    },
+    {
+      "epoch": 2.6131045241809674,
+      "grad_norm": 0.0032660351134836674,
+      "learning_rate": 4.9683442057091024e-05,
+      "loss": 0.00021552788093686103,
+      "step": 10050
+    },
+    {
+      "epoch": 2.6196047841913677,
+      "grad_norm": 0.003720888402312994,
+      "learning_rate": 4.96784066268247e-05,
+      "loss": 0.00020926753059029578,
+      "step": 10075
+    },
+    {
+      "epoch": 2.626105044201768,
+      "grad_norm": 0.0016928133554756641,
+      "learning_rate": 4.9673331722332644e-05,
+      "loss": 0.00022818347439169885,
+      "step": 10100
+    },
+    {
+      "epoch": 2.6326053042121687,
+      "grad_norm": 0.0012767099542543292,
+      "learning_rate": 4.9668217351732396e-05,
+      "loss": 0.00019867463037371636,
+      "step": 10125
+    },
+    {
+      "epoch": 2.639105564222569,
+      "grad_norm": 0.0027798812370747328,
+      "learning_rate": 4.966306352320461e-05,
+      "loss": 0.00021723633632063865,
+      "step": 10150
+    },
+    {
+      "epoch": 2.645605824232969,
+      "grad_norm": 0.002425914164632559,
+      "learning_rate": 4.9657870244993074e-05,
+      "loss": 0.00021136820316314698,
+      "step": 10175
+    },
+    {
+      "epoch": 2.6521060842433695,
+      "grad_norm": 0.0028046544175595045,
+      "learning_rate": 4.965263752540468e-05,
+      "loss": 0.00021517733111977577,
+      "step": 10200
+    },
+    {
+      "epoch": 2.65860634425377,
+      "grad_norm": 0.0010617782827466726,
+      "learning_rate": 4.964736537280939e-05,
+      "loss": 0.00019327972084283828,
+      "step": 10225
+    },
+    {
+      "epoch": 2.6651066042641705,
+      "grad_norm": 0.002738635055720806,
+      "learning_rate": 4.9642053795640255e-05,
+      "loss": 0.00022597866132855414,
+      "step": 10250
+    },
+    {
+      "epoch": 2.6716068642745707,
+      "grad_norm": 0.002575536025688052,
+      "learning_rate": 4.963670280239337e-05,
+      "loss": 0.00021602431312203407,
+      "step": 10275
+    },
+    {
+      "epoch": 2.6781071242849714,
+      "grad_norm": 0.0026268346700817347,
+      "learning_rate": 4.96313124016279e-05,
+      "loss": 0.0002253362722694874,
+      "step": 10300
+    },
+    {
+      "epoch": 2.6846073842953717,
+      "grad_norm": 0.0016526305116713047,
+      "learning_rate": 4.962588260196605e-05,
+      "loss": 0.0002079196460545063,
+      "step": 10325
+    },
+    {
+      "epoch": 2.691107644305772,
+      "grad_norm": 0.0038073293399065733,
+      "learning_rate": 4.9620413412093e-05,
+      "loss": 0.0002385610155761242,
+      "step": 10350
+    },
+    {
+      "epoch": 2.691107644305772,
+      "eval_loss": 0.0009969116654247046,
+      "eval_runtime": 7.3897,
+      "eval_samples_per_second": 676.621,
+      "eval_steps_per_second": 3.654,
+      "step": 10350
+    },
+    {
+      "epoch": 2.691107644305772,
+      "eval_holdout_avg_strict_f1": 0.0246094939046087,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.07336084364970197,
+      "eval_holdout_nemotron_pii_strict_f1": 0.05948357442206299,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.08300842661300464,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.05485158826592606,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 10350
+    },
+    {
+      "epoch": 2.6976079043161727,
+      "grad_norm": 0.0015660177450627089,
+      "learning_rate": 4.9614904840756985e-05,
+      "loss": 0.00020033180713653566,
+      "step": 10375
+    },
+    {
+      "epoch": 2.704108164326573,
+      "grad_norm": 0.001537105068564415,
+      "learning_rate": 4.96093568967692e-05,
+      "loss": 0.00020766632631421088,
+      "step": 10400
+    },
+    {
+      "epoch": 2.7106084243369732,
+      "grad_norm": 0.0023187182378023863,
+      "learning_rate": 4.960376958900385e-05,
+      "loss": 0.00022665463387966156,
+      "step": 10425
+    },
+    {
+      "epoch": 2.717108684347374,
+      "grad_norm": 0.0021124405320733786,
+      "learning_rate": 4.9598142926398075e-05,
+      "loss": 0.0002423793449997902,
+      "step": 10450
+    },
+    {
+      "epoch": 2.7236089443577742,
+      "grad_norm": 0.002342729829251766,
+      "learning_rate": 4.959247691795198e-05,
+      "loss": 0.00021285438910126686,
+      "step": 10475
+    },
+    {
+      "epoch": 2.7301092043681745,
+      "grad_norm": 0.0011349227279424667,
+      "learning_rate": 4.9586771572728597e-05,
+      "loss": 0.00023826278746128082,
+      "step": 10500
+    },
+    {
+      "epoch": 2.736609464378575,
+      "grad_norm": 0.006559950765222311,
+      "learning_rate": 4.958102689985388e-05,
+      "loss": 0.00026176514104008674,
+      "step": 10525
+    },
+    {
+      "epoch": 2.7431097243889755,
+      "grad_norm": 0.002465988975018263,
+      "learning_rate": 4.9575242908516705e-05,
+      "loss": 0.00022005362436175346,
+      "step": 10550
+    },
+    {
+      "epoch": 2.7496099843993758,
+      "grad_norm": 0.0,
+      "learning_rate": 4.956941960796881e-05,
+      "loss": 0.00023344876244664193,
+      "step": 10575
+    },
+    {
+      "epoch": 2.7561102444097765,
+      "grad_norm": 0.004045927431434393,
+      "learning_rate": 4.9563557007524854e-05,
+      "loss": 0.00024110456928610801,
+      "step": 10600
+    },
+    {
+      "epoch": 2.7626105044201767,
+      "grad_norm": 0.0031332962680608034,
+      "learning_rate": 4.95576551165623e-05,
+      "loss": 0.0002464031986892223,
+      "step": 10625
+    },
+    {
+      "epoch": 2.769110764430577,
+      "grad_norm": 0.003062512492761016,
+      "learning_rate": 4.955171394452151e-05,
+      "loss": 0.0002591661736369133,
+      "step": 10650
+    },
+    {
+      "epoch": 2.7756110244409777,
+      "grad_norm": 0.0021324511617422104,
+      "learning_rate": 4.954573350090565e-05,
+      "loss": 0.00022760441526770592,
+      "step": 10675
+    },
+    {
+      "epoch": 2.782111284451378,
+      "grad_norm": 0.006765362340956926,
+      "learning_rate": 4.953971379528071e-05,
+      "loss": 0.00021498167887330056,
+      "step": 10700
+    },
+    {
+      "epoch": 2.7886115444617783,
+      "grad_norm": 0.0029858271591365337,
+      "learning_rate": 4.953365483727548e-05,
+      "loss": 0.00021410981193184853,
+      "step": 10725
+    },
+    {
+      "epoch": 2.795111804472179,
+      "grad_norm": 0.006151528097689152,
+      "learning_rate": 4.952755663658154e-05,
+      "loss": 0.00022366801276803018,
+      "step": 10750
+    },
+    {
+      "epoch": 2.8016120644825793,
+      "grad_norm": 0.003916145768016577,
+      "learning_rate": 4.9521419202953236e-05,
+      "loss": 0.0002202354371547699,
+      "step": 10775
+    },
+    {
+      "epoch": 2.8081123244929795,
+      "grad_norm": 0.0029758187010884285,
+      "learning_rate": 4.951524254620766e-05,
+      "loss": 0.00022540690377354623,
+      "step": 10800
+    },
+    {
+      "epoch": 2.8146125845033803,
+      "grad_norm": 0.002156576607376337,
+      "learning_rate": 4.950902667622468e-05,
+      "loss": 0.00022879617288708686,
+      "step": 10825
+    },
+    {
+      "epoch": 2.8211128445137805,
+      "grad_norm": 0.004176854155957699,
+      "learning_rate": 4.950277160294683e-05,
+      "loss": 0.000242316797375679,
+      "step": 10850
+    },
+    {
+      "epoch": 2.827613104524181,
+      "grad_norm": 0.0014948262833058834,
+      "learning_rate": 4.94964773363794e-05,
+      "loss": 0.0002415275201201439,
+      "step": 10875
+    },
+    {
+      "epoch": 2.8341133645345815,
+      "grad_norm": 0.0019357949495315552,
+      "learning_rate": 4.9490143886590355e-05,
+      "loss": 0.00022052757441997527,
+      "step": 10900
+    },
+    {
+      "epoch": 2.840613624544982,
+      "grad_norm": 0.0026212057564407587,
+      "learning_rate": 4.948377126371032e-05,
+      "loss": 0.000219243373721838,
+      "step": 10925
+    },
+    {
+      "epoch": 2.847113884555382,
+      "grad_norm": 0.0007852587732486427,
+      "learning_rate": 4.9477359477932604e-05,
+      "loss": 0.0002223828434944153,
+      "step": 10950
+    },
+    {
+      "epoch": 2.8536141445657828,
+      "grad_norm": 0.004293252248317003,
+      "learning_rate": 4.947090853951315e-05,
+      "loss": 0.00024185044690966606,
+      "step": 10975
+    },
+    {
+      "epoch": 2.860114404576183,
+      "grad_norm": 0.0025315084494650364,
+      "learning_rate": 4.9464418458770516e-05,
+      "loss": 0.00024910921230912207,
+      "step": 11000
+    },
+    {
+      "epoch": 2.8666146645865833,
+      "grad_norm": 0.0007604174315929413,
+      "learning_rate": 4.945788924608589e-05,
+      "loss": 0.0002356138825416565,
+      "step": 11025
+    },
+    {
+      "epoch": 2.873114924596984,
+      "grad_norm": 0.0015335583593696356,
+      "learning_rate": 4.9451320911903027e-05,
+      "loss": 0.0002243470400571823,
+      "step": 11050
+    },
+    {
+      "epoch": 2.8796151846073843,
+      "grad_norm": 0.0012435317039489746,
+      "learning_rate": 4.9444713466728286e-05,
+      "loss": 0.0002243451587855816,
+      "step": 11075
+    },
+    {
+      "epoch": 2.8861154446177846,
+      "grad_norm": 0.0007130259764380753,
+      "learning_rate": 4.943806692113058e-05,
+      "loss": 0.0002097889594733715,
+      "step": 11100
+    },
+    {
+      "epoch": 2.8926157046281853,
+      "grad_norm": 0.005351623985916376,
+      "learning_rate": 4.943138128574134e-05,
+      "loss": 0.00022439273074269293,
+      "step": 11125
+    },
+    {
+      "epoch": 2.8991159646385856,
+      "grad_norm": 0.0022237093653529882,
+      "learning_rate": 4.942465657125455e-05,
+      "loss": 0.00023205667734146117,
+      "step": 11150
+    },
+    {
+      "epoch": 2.905616224648986,
+      "grad_norm": 0.0023445715196430683,
+      "learning_rate": 4.94178927884267e-05,
+      "loss": 0.00024087470024824142,
+      "step": 11175
+    },
+    {
+      "epoch": 2.9121164846593866,
+      "grad_norm": 0.002685474930331111,
+      "learning_rate": 4.941108994807676e-05,
+      "loss": 0.00021706270053982734,
+      "step": 11200
+    },
+    {
+      "epoch": 2.918616744669787,
+      "grad_norm": 0.001236864016391337,
+      "learning_rate": 4.940424806108619e-05,
+      "loss": 0.00022023888304829599,
+      "step": 11225
+    },
+    {
+      "epoch": 2.925117004680187,
+      "grad_norm": 0.008599478751420975,
+      "learning_rate": 4.939736713839889e-05,
+      "loss": 0.0002023216336965561,
+      "step": 11250
+    },
+    {
+      "epoch": 2.931617264690588,
+      "grad_norm": 0.001540121273137629,
+      "learning_rate": 4.93904471910212e-05,
+      "loss": 0.00019916126504540443,
+      "step": 11275
+    },
+    {
+      "epoch": 2.938117524700988,
+      "grad_norm": 0.0052956342697143555,
+      "learning_rate": 4.938348823002189e-05,
+      "loss": 0.00021275838837027549,
+      "step": 11300
+    },
+    {
+      "epoch": 2.9446177847113884,
+      "grad_norm": 0.0017920131795108318,
+      "learning_rate": 4.937649026653215e-05,
+      "loss": 0.00025085536763072016,
+      "step": 11325
+    },
+    {
+      "epoch": 2.951118044721789,
+      "grad_norm": 0.0021888134069740772,
+      "learning_rate": 4.9369453311745516e-05,
+      "loss": 0.00024143418297171594,
+      "step": 11350
+    },
+    {
+      "epoch": 2.9576183047321893,
+      "grad_norm": 0.0015382303390651941,
+      "learning_rate": 4.9362377376917936e-05,
+      "loss": 0.00022566163912415504,
+      "step": 11375
+    },
+    {
+      "epoch": 2.9641185647425896,
+      "grad_norm": 0.0030028631445020437,
+      "learning_rate": 4.935526247336767e-05,
+      "loss": 0.0002589407004415989,
+      "step": 11400
+    },
+    {
+      "epoch": 2.9706188247529903,
+      "grad_norm": 0.004402895923703909,
+      "learning_rate": 4.934810861247533e-05,
+      "loss": 0.00023730933666229247,
+      "step": 11425
+    },
+    {
+      "epoch": 2.9771190847633906,
+      "grad_norm": 0.0029234548565000296,
+      "learning_rate": 4.934091580568384e-05,
+      "loss": 0.00021670525893568993,
+      "step": 11450
+    },
+    {
+      "epoch": 2.983619344773791,
+      "grad_norm": 0.0029379078187048435,
+      "learning_rate": 4.933368406449843e-05,
+      "loss": 0.00024921443313360216,
+      "step": 11475
+    },
+    {
+      "epoch": 2.9901196047841916,
+      "grad_norm": 0.002386544831097126,
+      "learning_rate": 4.932641340048658e-05,
+      "loss": 0.00023454571142792702,
+      "step": 11500
+    },
+    {
+      "epoch": 2.996619864794592,
+      "grad_norm": 0.0027743966784328222,
+      "learning_rate": 4.931910382527806e-05,
+      "loss": 0.00022950667887926103,
+      "step": 11525
+    },
+    {
+      "epoch": 3.003120124804992,
+      "grad_norm": 0.0017982296412810683,
+      "learning_rate": 4.931175535056486e-05,
+      "loss": 0.00020994417369365692,
+      "step": 11550
+    },
+    {
+      "epoch": 3.0096203848153924,
+      "grad_norm": 0.00301287486217916,
+      "learning_rate": 4.9304367988101194e-05,
+      "loss": 0.0002242743782699108,
+      "step": 11575
+    },
+    {
+      "epoch": 3.016120644825793,
+      "grad_norm": 0.0020963498391211033,
+      "learning_rate": 4.9296941749703486e-05,
+      "loss": 0.00022736454382538796,
+      "step": 11600
+    },
+    {
+      "epoch": 3.0226209048361934,
+      "grad_norm": 0.002201492665335536,
+      "learning_rate": 4.9289476647250344e-05,
+      "loss": 0.00023226510733366013,
+      "step": 11625
+    },
+    {
+      "epoch": 3.0291211648465937,
+      "grad_norm": 0.0034691316541284323,
+      "learning_rate": 4.928197269268252e-05,
+      "loss": 0.0002492295950651169,
+      "step": 11650
+    },
+    {
+      "epoch": 3.0356214248569944,
+      "grad_norm": 0.001661220914684236,
+      "learning_rate": 4.927442989800294e-05,
+      "loss": 0.0002529814094305038,
+      "step": 11675
+    },
+    {
+      "epoch": 3.0421216848673946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.926684827527665e-05,
+      "loss": 0.00018309568986296654,
+      "step": 11700
+    },
+    {
+      "epoch": 3.048621944877795,
+      "grad_norm": 0.003122541354969144,
+      "learning_rate": 4.925922783663079e-05,
+      "loss": 0.00023760529235005378,
+      "step": 11725
+    },
+    {
+      "epoch": 3.0551222048881956,
+      "grad_norm": 0.001743837958201766,
+      "learning_rate": 4.925156859425459e-05,
+      "loss": 0.00023260578513145447,
+      "step": 11750
+    },
+    {
+      "epoch": 3.061622464898596,
+      "grad_norm": 0.0015707521233707666,
+      "learning_rate": 4.924387056039937e-05,
+      "loss": 0.0002434580959379673,
+      "step": 11775
+    },
+    {
+      "epoch": 3.068122724908996,
+      "grad_norm": 0.004237363114953041,
+      "learning_rate": 4.9236133747378475e-05,
+      "loss": 0.00022610457614064217,
+      "step": 11800
+    },
+    {
+      "epoch": 3.074622984919397,
+      "grad_norm": 0.00282191950827837,
+      "learning_rate": 4.922835816756728e-05,
+      "loss": 0.00022933052852749824,
+      "step": 11825
+    },
+    {
+      "epoch": 3.081123244929797,
+      "grad_norm": 0.003238640259951353,
+      "learning_rate": 4.922054383340319e-05,
+      "loss": 0.0002453051507472992,
+      "step": 11850
+    },
+    {
+      "epoch": 3.0876235049401974,
+      "grad_norm": 0.0011007124558091164,
+      "learning_rate": 4.9212690757385565e-05,
+      "loss": 0.00019693251699209213,
+      "step": 11875
+    },
+    {
+      "epoch": 3.094123764950598,
+      "grad_norm": 0.003405121387913823,
+      "learning_rate": 4.920479895207577e-05,
+      "loss": 0.00020770594477653503,
+      "step": 11900
+    },
+    {
+      "epoch": 3.1006240249609984,
+      "grad_norm": 0.0027633283752948046,
+      "learning_rate": 4.919686843009711e-05,
+      "loss": 0.00021515555679798125,
+      "step": 11925
+    },
+    {
+      "epoch": 3.1071242849713987,
+      "grad_norm": 0.0016431858530268073,
+      "learning_rate": 4.9188899204134794e-05,
+      "loss": 0.00019752735272049904,
+      "step": 11950
+    },
+    {
+      "epoch": 3.1136245449817994,
+      "grad_norm": 0.002135417889803648,
+      "learning_rate": 4.918089128693597e-05,
+      "loss": 0.00020202819257974624,
+      "step": 11975
+    },
+    {
+      "epoch": 3.1201248049921997,
+      "grad_norm": 0.0,
+      "learning_rate": 4.917284469130966e-05,
+      "loss": 0.00021086450666189194,
+      "step": 12000
+    },
+    {
+      "epoch": 3.1266250650026,
+      "grad_norm": 0.001049010781571269,
+      "learning_rate": 4.916475943012676e-05,
+      "loss": 0.00020914655178785324,
+      "step": 12025
+    },
+    {
+      "epoch": 3.1331253250130007,
+      "grad_norm": 0.0032548336312174797,
+      "learning_rate": 4.915663551632002e-05,
+      "loss": 0.00021471098065376282,
+      "step": 12050
+    },
+    {
+      "epoch": 3.139625585023401,
+      "grad_norm": 0.004152526613324881,
+      "learning_rate": 4.914847296288398e-05,
+      "loss": 0.00020719435065984725,
+      "step": 12075
+    },
+    {
+      "epoch": 3.146125845033801,
+      "grad_norm": 0.002509004669263959,
+      "learning_rate": 4.9140271782875035e-05,
+      "loss": 0.00019310545176267623,
+      "step": 12100
+    },
+    {
+      "epoch": 3.152626105044202,
+      "grad_norm": 0.0033038260880857706,
+      "learning_rate": 4.913203198941134e-05,
+      "loss": 0.0001863062009215355,
+      "step": 12125
+    },
+    {
+      "epoch": 3.159126365054602,
+      "grad_norm": 0.0020433571189641953,
+      "learning_rate": 4.912375359567281e-05,
+      "loss": 0.00021097654476761818,
+      "step": 12150
+    },
+    {
+      "epoch": 3.1656266250650025,
+      "grad_norm": 0.0014055912615731359,
+      "learning_rate": 4.9115436614901114e-05,
+      "loss": 0.0001880042627453804,
+      "step": 12175
+    },
+    {
+      "epoch": 3.172126885075403,
+      "grad_norm": 0.004213460721075535,
+      "learning_rate": 4.9107081060399627e-05,
+      "loss": 0.00020551171153783797,
+      "step": 12200
+    },
+    {
+      "epoch": 3.1786271450858035,
+      "grad_norm": 0.0012788143940269947,
+      "learning_rate": 4.909868694553346e-05,
+      "loss": 0.0001973356120288372,
+      "step": 12225
+    },
+    {
+      "epoch": 3.1851274050962037,
+      "grad_norm": 0.003932163119316101,
+      "learning_rate": 4.909025428372936e-05,
+      "loss": 0.00020261868834495545,
+      "step": 12250
+    },
+    {
+      "epoch": 3.1916276651066045,
+      "grad_norm": 0.00136783707421273,
+      "learning_rate": 4.908178308847575e-05,
+      "loss": 0.00019613714888691902,
+      "step": 12275
+    },
+    {
+      "epoch": 3.1981279251170047,
+      "grad_norm": 0.0025099413469433784,
+      "learning_rate": 4.907327337332271e-05,
+      "loss": 0.00019794836640357971,
+      "step": 12300
+    },
+    {
+      "epoch": 3.204628185127405,
+      "grad_norm": 0.006327567622065544,
+      "learning_rate": 4.906472515188188e-05,
+      "loss": 0.0002284386567771435,
+      "step": 12325
+    },
+    {
+      "epoch": 3.2111284451378057,
+      "grad_norm": 0.004143077880144119,
+      "learning_rate": 4.905613843782655e-05,
+      "loss": 0.0002003065124154091,
+      "step": 12350
+    },
+    {
+      "epoch": 3.217628705148206,
+      "grad_norm": 0.0021839463151991367,
+      "learning_rate": 4.904751324489155e-05,
+      "loss": 0.00019245944917201997,
+      "step": 12375
+    },
+    {
+      "epoch": 3.2241289651586063,
+      "grad_norm": 0.0034279569517821074,
+      "learning_rate": 4.903884958687327e-05,
+      "loss": 0.00020418841391801835,
+      "step": 12400
+    },
+    {
+      "epoch": 3.230629225169007,
+      "grad_norm": 0.0022887536324560642,
+      "learning_rate": 4.903014747762961e-05,
+      "loss": 0.00020190199837088584,
+      "step": 12425
+    },
+    {
+      "epoch": 3.2371294851794072,
+      "grad_norm": 0.003737232880666852,
+      "learning_rate": 4.902140693108e-05,
+      "loss": 0.00020541302859783172,
+      "step": 12450
+    },
+    {
+      "epoch": 3.2436297451898075,
+      "grad_norm": 0.0021527439821511507,
+      "learning_rate": 4.901262796120534e-05,
+      "loss": 0.0002332279458642006,
+      "step": 12475
+    },
+    {
+      "epoch": 3.250130005200208,
+      "grad_norm": 0.0021792054176330566,
+      "learning_rate": 4.900381058204798e-05,
+      "loss": 0.00020191103219985963,
+      "step": 12500
+    },
+    {
+      "epoch": 3.2566302652106085,
+      "grad_norm": 0.002642333973199129,
+      "learning_rate": 4.8994954807711715e-05,
+      "loss": 0.00017438501119613648,
+      "step": 12525
+    },
+    {
+      "epoch": 3.2631305252210088,
+      "grad_norm": 0.002192799700424075,
+      "learning_rate": 4.898606065236176e-05,
+      "loss": 0.00019916070625185966,
+      "step": 12550
+    },
+    {
+      "epoch": 3.269630785231409,
+      "grad_norm": 0.004767266567796469,
+      "learning_rate": 4.897712813022473e-05,
+      "loss": 0.00019739609211683274,
+      "step": 12575
+    },
+    {
+      "epoch": 3.2761310452418098,
+      "grad_norm": 0.0038899958599358797,
+      "learning_rate": 4.896815725558857e-05,
+      "loss": 0.00020079337060451507,
+      "step": 12600
+    },
+    {
+      "epoch": 3.28263130525221,
+      "grad_norm": 0.004382239654660225,
+      "learning_rate": 4.895914804280262e-05,
+      "loss": 0.00022807007655501366,
+      "step": 12625
+    },
+    {
+      "epoch": 3.2891315652626103,
+      "grad_norm": 0.0020285255741328,
+      "learning_rate": 4.895010050627752e-05,
+      "loss": 0.00021267659962177277,
+      "step": 12650
+    },
+    {
+      "epoch": 3.295631825273011,
+      "grad_norm": 0.005049275700002909,
+      "learning_rate": 4.8941014660485206e-05,
+      "loss": 0.00019947847351431847,
+      "step": 12675
+    },
+    {
+      "epoch": 3.3021320852834113,
+      "grad_norm": 0.001262821489945054,
+      "learning_rate": 4.893189051995891e-05,
+      "loss": 0.00019506225362420082,
+      "step": 12700
+    },
+    {
+      "epoch": 3.3086323452938116,
+      "grad_norm": 0.004890700336545706,
+      "learning_rate": 4.892272809929309e-05,
+      "loss": 0.00020861361175775528,
+      "step": 12725
+    },
+    {
+      "epoch": 3.3151326053042123,
+      "grad_norm": 0.0016499614575877786,
+      "learning_rate": 4.891352741314347e-05,
+      "loss": 0.00019455984234809876,
+      "step": 12750
+    },
+    {
+      "epoch": 3.3216328653146125,
+      "grad_norm": 0.003458687337115407,
+      "learning_rate": 4.8904288476226954e-05,
+      "loss": 0.00017893066629767418,
+      "step": 12775
+    },
+    {
+      "epoch": 3.328133125325013,
+      "grad_norm": 0.003017697250470519,
+      "learning_rate": 4.889501130332163e-05,
+      "loss": 0.00019013486802577973,
+      "step": 12800
+    },
+    {
+      "epoch": 3.3346333853354135,
+      "grad_norm": 0.0025098652113229036,
+      "learning_rate": 4.8885695909266774e-05,
+      "loss": 0.00017980284988880157,
+      "step": 12825
+    },
+    {
+      "epoch": 3.341133645345814,
+      "grad_norm": 0.005720977205783129,
+      "learning_rate": 4.887634230896276e-05,
+      "loss": 0.0002051236853003502,
+      "step": 12850
+    },
+    {
+      "epoch": 3.347633905356214,
+      "grad_norm": 0.002619906095787883,
+      "learning_rate": 4.886695051737112e-05,
+      "loss": 0.00021421480923891067,
+      "step": 12875
+    },
+    {
+      "epoch": 3.354134165366615,
+      "grad_norm": 0.0018695854814723134,
+      "learning_rate": 4.885752054951442e-05,
+      "loss": 0.00018555115908384324,
+      "step": 12900
+    },
+    {
+      "epoch": 3.360634425377015,
+      "grad_norm": 0.0019466273952275515,
+      "learning_rate": 4.884805242047633e-05,
+      "loss": 0.0001966596581041813,
+      "step": 12925
+    },
+    {
+      "epoch": 3.3671346853874153,
+      "grad_norm": 0.0010263555450364947,
+      "learning_rate": 4.8838546145401556e-05,
+      "loss": 0.0001837818883359432,
+      "step": 12950
+    },
+    {
+      "epoch": 3.373634945397816,
+      "grad_norm": 0.0010368929943069816,
+      "learning_rate": 4.882900173949581e-05,
+      "loss": 0.00018719714134931564,
+      "step": 12975
+    },
+    {
+      "epoch": 3.3801352054082163,
+      "grad_norm": 0.002077738055959344,
+      "learning_rate": 4.88194192180258e-05,
+      "loss": 0.00019075836986303329,
+      "step": 13000
+    },
+    {
+      "epoch": 3.3866354654186166,
+      "grad_norm": 0.0018235064344480634,
+      "learning_rate": 4.88097985963192e-05,
+      "loss": 0.00015992471948266028,
+      "step": 13025
+    },
+    {
+      "epoch": 3.3931357254290173,
+      "grad_norm": 0.00295456824824214,
+      "learning_rate": 4.8800139889764635e-05,
+      "loss": 0.00019619416445493699,
+      "step": 13050
+    },
+    {
+      "epoch": 3.3996359854394176,
+      "grad_norm": 0.002058732323348522,
+      "learning_rate": 4.879044311381163e-05,
+      "loss": 0.00018507666885852814,
+      "step": 13075
+    },
+    {
+      "epoch": 3.406136245449818,
+      "grad_norm": 0.0011640433222055435,
+      "learning_rate": 4.8780708283970635e-05,
+      "loss": 0.00019303014501929284,
+      "step": 13100
+    },
+    {
+      "epoch": 3.4126365054602186,
+      "grad_norm": 0.004538402892649174,
+      "learning_rate": 4.877093541581293e-05,
+      "loss": 0.00022105732932686806,
+      "step": 13125
+    },
+    {
+      "epoch": 3.419136765470619,
+      "grad_norm": 0.004582549445331097,
+      "learning_rate": 4.876112452497067e-05,
+      "loss": 0.00019006865099072457,
+      "step": 13150
+    },
+    {
+      "epoch": 3.425637025481019,
+      "grad_norm": 0.0015251049771904945,
+      "learning_rate": 4.875127562713682e-05,
+      "loss": 0.00019551951438188554,
+      "step": 13175
+    },
+    {
+      "epoch": 3.43213728549142,
+      "grad_norm": 0.0026452955789864063,
+      "learning_rate": 4.874138873806513e-05,
+      "loss": 0.00019827393814921378,
+      "step": 13200
+    },
+    {
+      "epoch": 3.43863754550182,
+      "grad_norm": 0.0037624940741807222,
+      "learning_rate": 4.873146387357012e-05,
+      "loss": 0.00021021952852606773,
+      "step": 13225
+    },
+    {
+      "epoch": 3.4451378055122204,
+      "grad_norm": 0.001287887105718255,
+      "learning_rate": 4.8721501049527074e-05,
+      "loss": 0.00017457578331232072,
+      "step": 13250
+    },
+    {
+      "epoch": 3.451638065522621,
+      "grad_norm": 0.004461884032934904,
+      "learning_rate": 4.871150028187197e-05,
+      "loss": 0.0001941155456006527,
+      "step": 13275
+    },
+    {
+      "epoch": 3.4581383255330214,
+      "grad_norm": 0.0016347732162103057,
+      "learning_rate": 4.8701461586601495e-05,
+      "loss": 0.0001958230137825012,
+      "step": 13300
+    },
+    {
+      "epoch": 3.4646385855434216,
+      "grad_norm": 0.001056948327459395,
+      "learning_rate": 4.869138497977298e-05,
+      "loss": 0.00017797507345676422,
+      "step": 13325
+    },
+    {
+      "epoch": 3.4711388455538223,
+      "grad_norm": 0.000918197853025049,
+      "learning_rate": 4.8681270477504425e-05,
+      "loss": 0.00016535835340619087,
+      "step": 13350
+    },
+    {
+      "epoch": 3.4776391055642226,
+      "grad_norm": 0.004509458784013987,
+      "learning_rate": 4.8671118095974435e-05,
+      "loss": 0.00019067727029323578,
+      "step": 13375
+    },
+    {
+      "epoch": 3.484139365574623,
+      "grad_norm": 0.0028125063981860876,
+      "learning_rate": 4.86609278514222e-05,
+      "loss": 0.00020383823662996293,
+      "step": 13400
+    },
+    {
+      "epoch": 3.4906396255850236,
+      "grad_norm": 0.002086355583742261,
+      "learning_rate": 4.865069976014747e-05,
+      "loss": 0.00017648333683609963,
+      "step": 13425
+    },
+    {
+      "epoch": 3.497139885595424,
+      "grad_norm": 0.008011997677385807,
+      "learning_rate": 4.864043383851054e-05,
+      "loss": 0.00019700530916452408,
+      "step": 13450
+    },
+    {
+      "epoch": 3.503640145605824,
+      "grad_norm": 0.001153540681116283,
+      "learning_rate": 4.8630130102932226e-05,
+      "loss": 0.00019348829984664918,
+      "step": 13475
+    },
+    {
+      "epoch": 3.510140405616225,
+      "grad_norm": 0.0018860353156924248,
+      "learning_rate": 4.86197885698938e-05,
+      "loss": 0.00017718693241477012,
+      "step": 13500
+    },
+    {
+      "epoch": 3.516640665626625,
+      "grad_norm": 0.001993714366108179,
+      "learning_rate": 4.860940925593703e-05,
+      "loss": 0.00019930761307477952,
+      "step": 13525
+    },
+    {
+      "epoch": 3.5231409256370254,
+      "grad_norm": 0.00442202715203166,
+      "learning_rate": 4.859899217766408e-05,
+      "loss": 0.00020368916913866996,
+      "step": 13550
+    },
+    {
+      "epoch": 3.529641185647426,
+      "grad_norm": 0.00381108233705163,
+      "learning_rate": 4.858853735173756e-05,
+      "loss": 0.00019909050315618514,
+      "step": 13575
+    },
+    {
+      "epoch": 3.5361414456578264,
+      "grad_norm": 0.002255432540550828,
+      "learning_rate": 4.857804479488042e-05,
+      "loss": 0.00019707290455698968,
+      "step": 13600
+    },
+    {
+      "epoch": 3.5426417056682267,
+      "grad_norm": 0.0012507532956078649,
+      "learning_rate": 4.856751452387597e-05,
+      "loss": 0.00019065476953983307,
+      "step": 13625
+    },
+    {
+      "epoch": 3.5491419656786274,
+      "grad_norm": 0.0036212389823049307,
+      "learning_rate": 4.8556946555567875e-05,
+      "loss": 0.00017694346606731415,
+      "step": 13650
+    },
+    {
+      "epoch": 3.5556422256890277,
+      "grad_norm": 0.00439284835010767,
+      "learning_rate": 4.854634090686005e-05,
+      "loss": 0.00019004333764314652,
+      "step": 13675
+    },
+    {
+      "epoch": 3.562142485699428,
+      "grad_norm": 0.0031485799700021744,
+      "learning_rate": 4.853569759471674e-05,
+      "loss": 0.00019272057339549066,
+      "step": 13700
+    },
+    {
+      "epoch": 3.5686427457098286,
+      "grad_norm": 0.002816871739923954,
+      "learning_rate": 4.85250166361624e-05,
+      "loss": 0.00018054230138659478,
+      "step": 13725
+    },
+    {
+      "epoch": 3.575143005720229,
+      "grad_norm": 0.0011562134604901075,
+      "learning_rate": 4.851429804828168e-05,
+      "loss": 0.00018591873347759247,
+      "step": 13750
+    },
+    {
+      "epoch": 3.581643265730629,
+      "grad_norm": 0.0016186692519113421,
+      "learning_rate": 4.8503541848219464e-05,
+      "loss": 0.0001849461905658245,
+      "step": 13775
+    },
+    {
+      "epoch": 3.58814352574103,
+      "grad_norm": 0.003737223334610462,
+      "learning_rate": 4.8492748053180776e-05,
+      "loss": 0.00019492663443088532,
+      "step": 13800
+    },
+    {
+      "epoch": 3.58814352574103,
+      "eval_loss": 0.0007769459625706077,
+      "eval_runtime": 6.7276,
+      "eval_samples_per_second": 743.204,
+      "eval_steps_per_second": 4.013,
+      "step": 13800
+    },
+    {
+      "epoch": 3.58814352574103,
+      "eval_holdout_avg_strict_f1": 0.03755268702500873,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.11696513539233497,
+      "eval_holdout_nemotron_pii_strict_f1": 0.10866026828310041,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.11537326645028033,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.07208088714938031,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 13800
+    },
+    {
+      "epoch": 3.59464378575143,
+      "grad_norm": 0.002106535015627742,
+      "learning_rate": 4.848191668043078e-05,
+      "loss": 0.00018692363053560257,
+      "step": 13825
+    },
+    {
+      "epoch": 3.6011440457618304,
+      "grad_norm": 0.0009807189926505089,
+      "learning_rate": 4.847104774729474e-05,
+      "loss": 0.00018487952649593354,
+      "step": 13850
+    },
+    {
+      "epoch": 3.607644305772231,
+      "grad_norm": 0.0013675274094566703,
+      "learning_rate": 4.8460141271158014e-05,
+      "loss": 0.00020114419981837273,
+      "step": 13875
+    },
+    {
+      "epoch": 3.6141445657826314,
+      "grad_norm": 0.001720141852274537,
+      "learning_rate": 4.8449197269466e-05,
+      "loss": 0.0001868550106883049,
+      "step": 13900
+    },
+    {
+      "epoch": 3.6206448257930317,
+      "grad_norm": 0.002226258162409067,
+      "learning_rate": 4.843821575972411e-05,
+      "loss": 0.00018215464428067207,
+      "step": 13925
+    },
+    {
+      "epoch": 3.627145085803432,
+      "grad_norm": 0.0027116078417748213,
+      "learning_rate": 4.8427196759497796e-05,
+      "loss": 0.000181178729981184,
+      "step": 13950
+    },
+    {
+      "epoch": 3.6336453458138327,
+      "grad_norm": 0.0016104019014164805,
+      "learning_rate": 4.8416140286412416e-05,
+      "loss": 0.00018801089376211167,
+      "step": 13975
+    },
+    {
+      "epoch": 3.640145605824233,
+      "grad_norm": 0.0023415798787027597,
+      "learning_rate": 4.840504635815331e-05,
+      "loss": 0.00016814099624752997,
+      "step": 14000
+    },
+    {
+      "epoch": 3.6466458658346332,
+      "grad_norm": 0.0018245171522721648,
+      "learning_rate": 4.8393914992465715e-05,
+      "loss": 0.00019202623516321183,
+      "step": 14025
+    },
+    {
+      "epoch": 3.653146125845034,
+      "grad_norm": 0.001894753542728722,
+      "learning_rate": 4.838274620715475e-05,
+      "loss": 0.00016656076535582543,
+      "step": 14050
+    },
+    {
+      "epoch": 3.6596463858554342,
+      "grad_norm": 0.002777579938992858,
+      "learning_rate": 4.8371540020085406e-05,
+      "loss": 0.00018895357847213744,
+      "step": 14075
+    },
+    {
+      "epoch": 3.6661466458658345,
+      "grad_norm": 0.003976226318627596,
+      "learning_rate": 4.836029644918246e-05,
+      "loss": 0.00016790941357612609,
+      "step": 14100
+    },
+    {
+      "epoch": 3.672646905876235,
+      "grad_norm": 0.0019418034935370088,
+      "learning_rate": 4.834901551243054e-05,
+      "loss": 0.00015508588403463365,
+      "step": 14125
+    },
+    {
+      "epoch": 3.6791471658866355,
+      "grad_norm": 0.004138902761042118,
+      "learning_rate": 4.833769722787399e-05,
+      "loss": 0.00017922932282090186,
+      "step": 14150
+    },
+    {
+      "epoch": 3.6856474258970358,
+      "grad_norm": 0.0012737332144752145,
+      "learning_rate": 4.832634161361693e-05,
+      "loss": 0.0001669163815677166,
+      "step": 14175
+    },
+    {
+      "epoch": 3.692147685907436,
+      "grad_norm": 0.0013900664635002613,
+      "learning_rate": 4.831494868782318e-05,
+      "loss": 0.0001726311631500721,
+      "step": 14200
+    },
+    {
+      "epoch": 3.6986479459178367,
+      "grad_norm": 0.004070568364113569,
+      "learning_rate": 4.830351846871623e-05,
+      "loss": 0.00017538364976644517,
+      "step": 14225
+    },
+    {
+      "epoch": 3.705148205928237,
+      "grad_norm": 0.0032449776772409678,
+      "learning_rate": 4.829205097457923e-05,
+      "loss": 0.00017785368487238884,
+      "step": 14250
+    },
+    {
+      "epoch": 3.7116484659386373,
+      "grad_norm": 0.0047850823029875755,
+      "learning_rate": 4.8280546223754967e-05,
+      "loss": 0.00018772052600979805,
+      "step": 14275
+    },
+    {
+      "epoch": 3.718148725949038,
+      "grad_norm": 0.0022762890439480543,
+      "learning_rate": 4.8269004234645805e-05,
+      "loss": 0.00019112121313810348,
+      "step": 14300
+    },
+    {
+      "epoch": 3.7246489859594383,
+      "grad_norm": 0.0018947579665109515,
+      "learning_rate": 4.825742502571366e-05,
+      "loss": 0.00018218569457530975,
+      "step": 14325
+    },
+    {
+      "epoch": 3.7311492459698385,
+      "grad_norm": 0.004367785062640905,
+      "learning_rate": 4.824580861548004e-05,
+      "loss": 0.00019441751763224602,
+      "step": 14350
+    },
+    {
+      "epoch": 3.7376495059802393,
+      "grad_norm": 0.0,
+      "learning_rate": 4.823415502252587e-05,
+      "loss": 0.0001856766641139984,
+      "step": 14375
+    },
+    {
+      "epoch": 3.7441497659906395,
+      "grad_norm": 0.00402756268158555,
+      "learning_rate": 4.8222464265491624e-05,
+      "loss": 0.0001696070656180382,
+      "step": 14400
+    },
+    {
+      "epoch": 3.75065002600104,
+      "grad_norm": 0.002447994891554117,
+      "learning_rate": 4.821073636307719e-05,
+      "loss": 0.0001848655380308628,
+      "step": 14425
+    },
+    {
+      "epoch": 3.7571502860114405,
+      "grad_norm": 0.0013418401358649135,
+      "learning_rate": 4.819897133404189e-05,
+      "loss": 0.00018463864922523498,
+      "step": 14450
+    },
+    {
+      "epoch": 3.763650546021841,
+      "grad_norm": 0.0013997508212924004,
+      "learning_rate": 4.8187169197204395e-05,
+      "loss": 0.00017585938796401023,
+      "step": 14475
+    },
+    {
+      "epoch": 3.770150806032241,
+      "grad_norm": 0.0013077299809083343,
+      "learning_rate": 4.8175329971442774e-05,
+      "loss": 0.00017360333353281021,
+      "step": 14500
+    },
+    {
+      "epoch": 3.776651066042642,
+      "grad_norm": 0.0018581555923447013,
+      "learning_rate": 4.8163453675694396e-05,
+      "loss": 0.00017287006601691247,
+      "step": 14525
+    },
+    {
+      "epoch": 3.783151326053042,
+      "grad_norm": 0.0011628908105194569,
+      "learning_rate": 4.8151540328955935e-05,
+      "loss": 0.00017978239804506302,
+      "step": 14550
+    },
+    {
+      "epoch": 3.7896515860634423,
+      "grad_norm": 0.002391796326264739,
+      "learning_rate": 4.813958995028333e-05,
+      "loss": 0.00017921293154358865,
+      "step": 14575
+    },
+    {
+      "epoch": 3.796151846073843,
+      "grad_norm": 0.0029841905925422907,
+      "learning_rate": 4.812760255879175e-05,
+      "loss": 0.00018638083711266517,
+      "step": 14600
+    },
+    {
+      "epoch": 3.8026521060842433,
+      "grad_norm": 0.0021617060992866755,
+      "learning_rate": 4.811557817365556e-05,
+      "loss": 0.00018230058252811433,
+      "step": 14625
+    },
+    {
+      "epoch": 3.8091523660946436,
+      "grad_norm": 0.0005080679547972977,
+      "learning_rate": 4.810351681410832e-05,
+      "loss": 0.00017180234193801879,
+      "step": 14650
+    },
+    {
+      "epoch": 3.8156526261050443,
+      "grad_norm": 0.00275943661108613,
+      "learning_rate": 4.8091418499442705e-05,
+      "loss": 0.0001853370852768421,
+      "step": 14675
+    },
+    {
+      "epoch": 3.8221528861154446,
+      "grad_norm": 0.0016223345883190632,
+      "learning_rate": 4.8079283249010535e-05,
+      "loss": 0.00017199864611029626,
+      "step": 14700
+    },
+    {
+      "epoch": 3.828653146125845,
+      "grad_norm": 0.0014000064693391323,
+      "learning_rate": 4.806711108222267e-05,
+      "loss": 0.00017740676179528236,
+      "step": 14725
+    },
+    {
+      "epoch": 3.8351534061362456,
+      "grad_norm": 0.007425474934279919,
+      "learning_rate": 4.805490201854905e-05,
+      "loss": 0.00015938971191644668,
+      "step": 14750
+    },
+    {
+      "epoch": 3.841653666146646,
+      "grad_norm": 0.002419322030618787,
+      "learning_rate": 4.8042656077518625e-05,
+      "loss": 0.0001897728256881237,
+      "step": 14775
+    },
+    {
+      "epoch": 3.848153926157046,
+      "grad_norm": 0.0017272852128371596,
+      "learning_rate": 4.8030373278719335e-05,
+      "loss": 0.00019258715212345124,
+      "step": 14800
+    },
+    {
+      "epoch": 3.854654186167447,
+      "grad_norm": 0.0023506644647568464,
+      "learning_rate": 4.801805364179806e-05,
+      "loss": 0.0001934950426220894,
+      "step": 14825
+    },
+    {
+      "epoch": 3.861154446177847,
+      "grad_norm": 0.0008850769954733551,
+      "learning_rate": 4.800569718646062e-05,
+      "loss": 0.00019235765561461448,
+      "step": 14850
+    },
+    {
+      "epoch": 3.8676547061882474,
+      "grad_norm": 0.0015340980608016253,
+      "learning_rate": 4.7993303932471726e-05,
+      "loss": 0.0001842624507844448,
+      "step": 14875
+    },
+    {
+      "epoch": 3.874154966198648,
+      "grad_norm": 0.002366200787946582,
+      "learning_rate": 4.798087389965495e-05,
+      "loss": 0.00019244063645601272,
+      "step": 14900
+    },
+    {
+      "epoch": 3.8806552262090483,
+      "grad_norm": 0.004188060760498047,
+      "learning_rate": 4.7968407107892684e-05,
+      "loss": 0.0001769442670047283,
+      "step": 14925
+    },
+    {
+      "epoch": 3.8871554862194486,
+      "grad_norm": 0.0020440025255084038,
+      "learning_rate": 4.795590357712613e-05,
+      "loss": 0.00019140174612402916,
+      "step": 14950
+    },
+    {
+      "epoch": 3.8936557462298493,
+      "grad_norm": 0.0037882053293287754,
+      "learning_rate": 4.794336332735524e-05,
+      "loss": 0.00022109247744083404,
+      "step": 14975
+    },
+    {
+      "epoch": 3.9001560062402496,
+      "grad_norm": 0.0,
+      "learning_rate": 4.793078637863872e-05,
+      "loss": 0.00018556270748376847,
+      "step": 15000
+    },
+    {
+      "epoch": 3.90665626625065,
+      "grad_norm": 0.00209584878757596,
+      "learning_rate": 4.791817275109396e-05,
+      "loss": 0.00018636154010891914,
+      "step": 15025
+    },
+    {
+      "epoch": 3.9131565262610506,
+      "grad_norm": 0.0017556521343067288,
+      "learning_rate": 4.790552246489703e-05,
+      "loss": 0.0001750059239566326,
+      "step": 15050
+    },
+    {
+      "epoch": 3.919656786271451,
+      "grad_norm": 0.002784095238894224,
+      "learning_rate": 4.789283554028264e-05,
+      "loss": 0.00018199097365140916,
+      "step": 15075
+    },
+    {
+      "epoch": 3.926157046281851,
+      "grad_norm": 0.0010626687435433269,
+      "learning_rate": 4.7880111997544094e-05,
+      "loss": 0.00017649631947278977,
+      "step": 15100
+    },
+    {
+      "epoch": 3.932657306292252,
+      "grad_norm": 0.00348655809648335,
+      "learning_rate": 4.786735185703327e-05,
+      "loss": 0.0001511754747480154,
+      "step": 15125
+    },
+    {
+      "epoch": 3.939157566302652,
+      "grad_norm": 0.019497273489832878,
+      "learning_rate": 4.785455513916059e-05,
+      "loss": 0.00016047555953264237,
+      "step": 15150
+    },
+    {
+      "epoch": 3.9456578263130524,
+      "grad_norm": 0.0030943872407078743,
+      "learning_rate": 4.784172186439499e-05,
+      "loss": 0.00019644493237137796,
+      "step": 15175
+    },
+    {
+      "epoch": 3.952158086323453,
+      "grad_norm": 0.0007395350839942694,
+      "learning_rate": 4.7828852053263865e-05,
+      "loss": 0.00019324317574501037,
+      "step": 15200
+    },
+    {
+      "epoch": 3.9586583463338534,
+      "grad_norm": 0.0026222425512969494,
+      "learning_rate": 4.781594572635306e-05,
+      "loss": 0.00019622469320893288,
+      "step": 15225
+    },
+    {
+      "epoch": 3.9651586063442537,
+      "grad_norm": 0.0024738628417253494,
+      "learning_rate": 4.780300290430682e-05,
+      "loss": 0.0002056974731385708,
+      "step": 15250
+    },
+    {
+      "epoch": 3.9716588663546544,
+      "grad_norm": 0.0013616640353575349,
+      "learning_rate": 4.779002360782779e-05,
+      "loss": 0.0001998567208647728,
+      "step": 15275
+    },
+    {
+      "epoch": 3.9781591263650546,
+      "grad_norm": 0.0008051333134062588,
+      "learning_rate": 4.777700785767694e-05,
+      "loss": 0.0001793583855032921,
+      "step": 15300
+    },
+    {
+      "epoch": 3.984659386375455,
+      "grad_norm": 0.001919957809150219,
+      "learning_rate": 4.776395567467353e-05,
+      "loss": 0.00017205262556672097,
+      "step": 15325
+    },
+    {
+      "epoch": 3.9911596463858556,
+      "grad_norm": 0.0017199747962877154,
+      "learning_rate": 4.775086707969513e-05,
+      "loss": 0.00018198668956756592,
+      "step": 15350
+    },
+    {
+      "epoch": 3.997659906396256,
+      "grad_norm": 0.001567529165185988,
+      "learning_rate": 4.773774209367753e-05,
+      "loss": 0.00020061776041984557,
+      "step": 15375
+    },
+    {
+      "epoch": 4.004160166406656,
+      "grad_norm": 0.0015432513318955898,
+      "learning_rate": 4.7724580737614755e-05,
+      "loss": 0.00017593447118997575,
+      "step": 15400
+    },
+    {
+      "epoch": 4.010660426417057,
+      "grad_norm": 0.002256323117762804,
+      "learning_rate": 4.7711383032558976e-05,
+      "loss": 0.00020126234740018844,
+      "step": 15425
+    },
+    {
+      "epoch": 4.017160686427457,
+      "grad_norm": 0.0012787890154868364,
+      "learning_rate": 4.769814899962052e-05,
+      "loss": 0.00018624329939484596,
+      "step": 15450
+    },
+    {
+      "epoch": 4.023660946437857,
+      "grad_norm": 0.002444961341097951,
+      "learning_rate": 4.7684878659967816e-05,
+      "loss": 0.0001855655387043953,
+      "step": 15475
+    },
+    {
+      "epoch": 4.030161206448258,
+      "grad_norm": 0.001980958040803671,
+      "learning_rate": 4.7671572034827375e-05,
+      "loss": 0.00017914535477757453,
+      "step": 15500
+    },
+    {
+      "epoch": 4.036661466458658,
+      "grad_norm": 0.0038619190454483032,
+      "learning_rate": 4.765822914548375e-05,
+      "loss": 0.00015805648639798163,
+      "step": 15525
+    },
+    {
+      "epoch": 4.043161726469059,
+      "grad_norm": 0.00157670967746526,
+      "learning_rate": 4.764485001327949e-05,
+      "loss": 0.00017007097601890565,
+      "step": 15550
+    },
+    {
+      "epoch": 4.049661986479459,
+      "grad_norm": 0.0024021840654313564,
+      "learning_rate": 4.763143465961512e-05,
+      "loss": 0.00016745181754231452,
+      "step": 15575
+    },
+    {
+      "epoch": 4.056162246489859,
+      "grad_norm": 0.0011933110654354095,
+      "learning_rate": 4.761798310594911e-05,
+      "loss": 0.00016522496938705444,
+      "step": 15600
+    },
+    {
+      "epoch": 4.06266250650026,
+      "grad_norm": 0.0032834820449352264,
+      "learning_rate": 4.760449537379782e-05,
+      "loss": 0.00017104774713516235,
+      "step": 15625
+    },
+    {
+      "epoch": 4.069162766510661,
+      "grad_norm": 0.0014264112105593085,
+      "learning_rate": 4.7590971484735495e-05,
+      "loss": 0.00016997601836919783,
+      "step": 15650
+    },
+    {
+      "epoch": 4.0756630265210605,
+      "grad_norm": 0.0032719115260988474,
+      "learning_rate": 4.757741146039419e-05,
+      "loss": 0.0001667327620089054,
+      "step": 15675
+    },
+    {
+      "epoch": 4.082163286531461,
+      "grad_norm": 0.0029537843074649572,
+      "learning_rate": 4.756381532246379e-05,
+      "loss": 0.0001581120491027832,
+      "step": 15700
+    },
+    {
+      "epoch": 4.088663546541862,
+      "grad_norm": 0.0018526166677474976,
+      "learning_rate": 4.755018309269194e-05,
+      "loss": 0.00017901845276355744,
+      "step": 15725
+    },
+    {
+      "epoch": 4.095163806552262,
+      "grad_norm": 0.0019269533222541213,
+      "learning_rate": 4.753651479288398e-05,
+      "loss": 0.0001612401194870472,
+      "step": 15750
+    },
+    {
+      "epoch": 4.1016640665626625,
+      "grad_norm": 0.0019432856934145093,
+      "learning_rate": 4.7522810444903e-05,
+      "loss": 0.00017771489918231965,
+      "step": 15775
+    },
+    {
+      "epoch": 4.108164326573063,
+      "grad_norm": 0.004724913742393255,
+      "learning_rate": 4.75090700706697e-05,
+      "loss": 0.0001797553151845932,
+      "step": 15800
+    },
+    {
+      "epoch": 4.114664586583463,
+      "grad_norm": 0.0033014630898833275,
+      "learning_rate": 4.749529369216246e-05,
+      "loss": 0.0001592009700834751,
+      "step": 15825
+    },
+    {
+      "epoch": 4.121164846593864,
+      "grad_norm": 0.0007016555755399168,
+      "learning_rate": 4.748148133141718e-05,
+      "loss": 0.00016692623496055602,
+      "step": 15850
+    },
+    {
+      "epoch": 4.1276651066042644,
+      "grad_norm": 0.002613243879750371,
+      "learning_rate": 4.746763301052739e-05,
+      "loss": 0.00016209758818149566,
+      "step": 15875
+    },
+    {
+      "epoch": 4.134165366614664,
+      "grad_norm": 0.0029490189626812935,
+      "learning_rate": 4.7453748751644075e-05,
+      "loss": 0.00015947986394166947,
+      "step": 15900
+    },
+    {
+      "epoch": 4.140665626625065,
+      "grad_norm": 0.0014137791004031897,
+      "learning_rate": 4.7439828576975766e-05,
+      "loss": 0.00016196394339203834,
+      "step": 15925
+    },
+    {
+      "epoch": 4.147165886635466,
+      "grad_norm": 0.0013701192801818252,
+      "learning_rate": 4.7425872508788384e-05,
+      "loss": 0.0001614590920507908,
+      "step": 15950
+    },
+    {
+      "epoch": 4.1536661466458655,
+      "grad_norm": 0.0011157949920743704,
+      "learning_rate": 4.74118805694053e-05,
+      "loss": 0.00014938462525606154,
+      "step": 15975
+    },
+    {
+      "epoch": 4.160166406656266,
+      "grad_norm": 0.0019668089225888252,
+      "learning_rate": 4.739785278120725e-05,
+      "loss": 0.00016474539414048194,
+      "step": 16000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 0.004336381331086159,
+      "learning_rate": 4.738378916663233e-05,
+      "loss": 0.00016380157321691514,
+      "step": 16025
+    },
+    {
+      "epoch": 4.173166926677067,
+      "grad_norm": 0.0012645269744098186,
+      "learning_rate": 4.73696897481759e-05,
+      "loss": 0.00016380883753299713,
+      "step": 16050
+    },
+    {
+      "epoch": 4.1796671866874675,
+      "grad_norm": 0.0014034017222002149,
+      "learning_rate": 4.7355554548390636e-05,
+      "loss": 0.00015050990507006645,
+      "step": 16075
+    },
+    {
+      "epoch": 4.186167446697868,
+      "grad_norm": 0.0010972560849040747,
+      "learning_rate": 4.7341383589886423e-05,
+      "loss": 0.00016104785725474358,
+      "step": 16100
+    },
+    {
+      "epoch": 4.192667706708268,
+      "grad_norm": 0.0023443165700882673,
+      "learning_rate": 4.732717689533035e-05,
+      "loss": 0.0001543428935110569,
+      "step": 16125
+    },
+    {
+      "epoch": 4.199167966718669,
+      "grad_norm": 0.005482005886733532,
+      "learning_rate": 4.731293448744666e-05,
+      "loss": 0.00016352515667676925,
+      "step": 16150
+    },
+    {
+      "epoch": 4.2056682267290695,
+      "grad_norm": 0.0015973112313076854,
+      "learning_rate": 4.7298656389016735e-05,
+      "loss": 0.0001575806923210621,
+      "step": 16175
+    },
+    {
+      "epoch": 4.212168486739469,
+      "grad_norm": 0.0020761583000421524,
+      "learning_rate": 4.728434262287904e-05,
+      "loss": 0.00017347386106848718,
+      "step": 16200
+    },
+    {
+      "epoch": 4.21866874674987,
+      "grad_norm": 0.002122585428878665,
+      "learning_rate": 4.726999321192908e-05,
+      "loss": 0.00017534645274281502,
+      "step": 16225
+    },
+    {
+      "epoch": 4.225169006760271,
+      "grad_norm": 0.004099489189684391,
+      "learning_rate": 4.725560817911939e-05,
+      "loss": 0.00018511157482862473,
+      "step": 16250
+    },
+    {
+      "epoch": 4.231669266770671,
+      "grad_norm": 0.005462062079459429,
+      "learning_rate": 4.724118754745949e-05,
+      "loss": 0.00017231382429599762,
+      "step": 16275
+    },
+    {
+      "epoch": 4.238169526781071,
+      "grad_norm": 0.002726222388446331,
+      "learning_rate": 4.7226731340015814e-05,
+      "loss": 0.00016330508515238762,
+      "step": 16300
+    },
+    {
+      "epoch": 4.244669786791472,
+      "grad_norm": 0.0012742180842906237,
+      "learning_rate": 4.721223957991174e-05,
+      "loss": 0.0001804397813975811,
+      "step": 16325
+    },
+    {
+      "epoch": 4.251170046801872,
+      "grad_norm": 0.001461735344491899,
+      "learning_rate": 4.7197712290327465e-05,
+      "loss": 0.00016668090596795083,
+      "step": 16350
+    },
+    {
+      "epoch": 4.2576703068122725,
+      "grad_norm": 0.001352659659460187,
+      "learning_rate": 4.718314949450007e-05,
+      "loss": 0.00015542961657047272,
+      "step": 16375
+    },
+    {
+      "epoch": 4.264170566822673,
+      "grad_norm": 0.0031707535963505507,
+      "learning_rate": 4.7168551215723395e-05,
+      "loss": 0.0001780588924884796,
+      "step": 16400
+    },
+    {
+      "epoch": 4.270670826833073,
+      "grad_norm": 0.0021144773345440626,
+      "learning_rate": 4.715391747734804e-05,
+      "loss": 0.00016632189974188805,
+      "step": 16425
+    },
+    {
+      "epoch": 4.277171086843474,
+      "grad_norm": 0.006771764252334833,
+      "learning_rate": 4.7139248302781335e-05,
+      "loss": 0.00016728438436985015,
+      "step": 16450
+    },
+    {
+      "epoch": 4.2836713468538745,
+      "grad_norm": 0.001838906668126583,
+      "learning_rate": 4.712454371548729e-05,
+      "loss": 0.00017785703763365746,
+      "step": 16475
+    },
+    {
+      "epoch": 4.290171606864274,
+      "grad_norm": 0.009326976723968983,
+      "learning_rate": 4.710980373898656e-05,
+      "loss": 0.00017834877595305444,
+      "step": 16500
+    },
+    {
+      "epoch": 4.296671866874675,
+      "grad_norm": 0.0019171081949025393,
+      "learning_rate": 4.709502839685639e-05,
+      "loss": 0.0002019973285496235,
+      "step": 16525
+    },
+    {
+      "epoch": 4.303172126885076,
+      "grad_norm": 0.003684645751491189,
+      "learning_rate": 4.708021771273061e-05,
+      "loss": 0.00017412833869457245,
+      "step": 16550
+    },
+    {
+      "epoch": 4.309672386895476,
+      "grad_norm": 0.006142088677734137,
+      "learning_rate": 4.706537171029959e-05,
+      "loss": 0.00018382873386144639,
+      "step": 16575
+    },
+    {
+      "epoch": 4.316172646905876,
+      "grad_norm": 0.0008659467566758394,
+      "learning_rate": 4.7050490413310166e-05,
+      "loss": 0.00016139013692736625,
+      "step": 16600
+    },
+    {
+      "epoch": 4.322672906916277,
+      "grad_norm": 0.0020545271690934896,
+      "learning_rate": 4.703557384556566e-05,
+      "loss": 0.00017395377159118653,
+      "step": 16625
+    },
+    {
+      "epoch": 4.329173166926677,
+      "grad_norm": 0.0026616626419126987,
+      "learning_rate": 4.7020622030925785e-05,
+      "loss": 0.00017738474532961847,
+      "step": 16650
+    },
+    {
+      "epoch": 4.335673426937078,
+      "grad_norm": 0.0012004738673567772,
+      "learning_rate": 4.700563499330664e-05,
+      "loss": 0.00016084292903542519,
+      "step": 16675
+    },
+    {
+      "epoch": 4.342173686947478,
+      "grad_norm": 0.0014764935476705432,
+      "learning_rate": 4.699061275668068e-05,
+      "loss": 0.00016870886087417603,
+      "step": 16700
+    },
+    {
+      "epoch": 4.348673946957878,
+      "grad_norm": 0.0017060237005352974,
+      "learning_rate": 4.6975555345076644e-05,
+      "loss": 0.0001657457835972309,
+      "step": 16725
+    },
+    {
+      "epoch": 4.355174206968279,
+      "grad_norm": 0.0006398553377948701,
+      "learning_rate": 4.696046278257954e-05,
+      "loss": 0.00020843859761953353,
+      "step": 16750
+    },
+    {
+      "epoch": 4.3616744669786796,
+      "grad_norm": 0.0010869739344343543,
+      "learning_rate": 4.694533509333061e-05,
+      "loss": 0.00017780004069209098,
+      "step": 16775
+    },
+    {
+      "epoch": 4.368174726989079,
+      "grad_norm": 0.0008869327721185982,
+      "learning_rate": 4.693017230152727e-05,
+      "loss": 0.00016541866585612297,
+      "step": 16800
+    },
+    {
+      "epoch": 4.37467498699948,
+      "grad_norm": 0.0008592939120717347,
+      "learning_rate": 4.6914974431423095e-05,
+      "loss": 0.00016656942665576935,
+      "step": 16825
+    },
+    {
+      "epoch": 4.38117524700988,
+      "grad_norm": 0.0007334901602007449,
+      "learning_rate": 4.689974150732776e-05,
+      "loss": 0.0001574445515871048,
+      "step": 16850
+    },
+    {
+      "epoch": 4.387675507020281,
+      "grad_norm": 0.0012354288483038545,
+      "learning_rate": 4.688447355360702e-05,
+      "loss": 0.00018074676394462584,
+      "step": 16875
+    },
+    {
+      "epoch": 4.394175767030681,
+      "grad_norm": 0.0029802690260112286,
+      "learning_rate": 4.686917059468265e-05,
+      "loss": 0.00016910672187805175,
+      "step": 16900
+    },
+    {
+      "epoch": 4.400676027041081,
+      "grad_norm": 0.0034686995204538107,
+      "learning_rate": 4.685383265503243e-05,
+      "loss": 0.00015295635908842088,
+      "step": 16925
+    },
+    {
+      "epoch": 4.407176287051482,
+      "grad_norm": 0.0010052589932456613,
+      "learning_rate": 4.683845975919008e-05,
+      "loss": 0.00016116408631205558,
+      "step": 16950
+    },
+    {
+      "epoch": 4.413676547061883,
+      "grad_norm": 0.0013952176086604595,
+      "learning_rate": 4.682305193174524e-05,
+      "loss": 0.0001619725301861763,
+      "step": 16975
+    },
+    {
+      "epoch": 4.420176807072282,
+      "grad_norm": 0.002685805084183812,
+      "learning_rate": 4.6807609197343444e-05,
+      "loss": 0.00015479259192943573,
+      "step": 17000
+    },
+    {
+      "epoch": 4.426677067082683,
+      "grad_norm": 0.0017541280249133706,
+      "learning_rate": 4.679213158068602e-05,
+      "loss": 0.00015131274238228799,
+      "step": 17025
+    },
+    {
+      "epoch": 4.433177327093084,
+      "grad_norm": 0.0015597807941958308,
+      "learning_rate": 4.677661910653013e-05,
+      "loss": 0.00016712935641407967,
+      "step": 17050
+    },
+    {
+      "epoch": 4.439677587103484,
+      "grad_norm": 0.0024185015354305506,
+      "learning_rate": 4.676107179968867e-05,
+      "loss": 0.00017396846786141396,
+      "step": 17075
+    },
+    {
+      "epoch": 4.446177847113884,
+      "grad_norm": 0.0012252889573574066,
+      "learning_rate": 4.674548968503026e-05,
+      "loss": 0.00015929028391838075,
+      "step": 17100
+    },
+    {
+      "epoch": 4.452678107124285,
+      "grad_norm": 0.0,
+      "learning_rate": 4.672987278747919e-05,
+      "loss": 0.00014780063182115555,
+      "step": 17125
+    },
+    {
+      "epoch": 4.459178367134685,
+      "grad_norm": 0.0015239381464198232,
+      "learning_rate": 4.671422113201541e-05,
+      "loss": 0.00015520074404776096,
+      "step": 17150
+    },
+    {
+      "epoch": 4.465678627145086,
+      "grad_norm": 0.005024355836212635,
+      "learning_rate": 4.6698534743674436e-05,
+      "loss": 0.0001642940752208233,
+      "step": 17175
+    },
+    {
+      "epoch": 4.472178887155486,
+      "grad_norm": 0.0040642754174768925,
+      "learning_rate": 4.668281364754735e-05,
+      "loss": 0.00016543280333280563,
+      "step": 17200
+    },
+    {
+      "epoch": 4.478679147165886,
+      "grad_norm": 0.0019486952805891633,
+      "learning_rate": 4.666705786878077e-05,
+      "loss": 0.00015313019044697285,
+      "step": 17225
+    },
+    {
+      "epoch": 4.485179407176287,
+      "grad_norm": 0.0022806357592344284,
+      "learning_rate": 4.665126743257677e-05,
+      "loss": 0.0001513226330280304,
+      "step": 17250
+    },
+    {
+      "epoch": 4.485179407176287,
+      "eval_loss": 0.000658437900710851,
+      "eval_runtime": 6.8934,
+      "eval_samples_per_second": 725.334,
+      "eval_steps_per_second": 3.917,
+      "step": 17250
+    },
+    {
+      "epoch": 4.485179407176287,
+      "eval_holdout_avg_strict_f1": 0.050783621623763826,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.19670152855993564,
+      "eval_holdout_nemotron_pii_strict_f1": 0.13723134427980244,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.12369285927696444,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.10099410574469957,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 17250
+    },
+    {
+      "epoch": 4.491679667186688,
+      "grad_norm": 0.00196244427934289,
+      "learning_rate": 4.663544236419286e-05,
+      "loss": 0.00017497438937425614,
+      "step": 17275
+    },
+    {
+      "epoch": 4.4981799271970875,
+      "grad_norm": 0.0015371248591691256,
+      "learning_rate": 4.6619582688941967e-05,
+      "loss": 0.00015914950519800187,
+      "step": 17300
+    },
+    {
+      "epoch": 4.504680187207488,
+      "grad_norm": 0.0008240469032898545,
+      "learning_rate": 4.660368843219233e-05,
+      "loss": 0.00016264503821730614,
+      "step": 17325
+    },
+    {
+      "epoch": 4.511180447217889,
+      "grad_norm": 0.0010795125272125006,
+      "learning_rate": 4.6587759619367556e-05,
+      "loss": 0.00017988884821534156,
+      "step": 17350
+    },
+    {
+      "epoch": 4.517680707228289,
+      "grad_norm": 0.0008447085274383426,
+      "learning_rate": 4.657179627594649e-05,
+      "loss": 0.00015538561157882213,
+      "step": 17375
+    },
+    {
+      "epoch": 4.5241809672386895,
+      "grad_norm": 0.0007679208065383136,
+      "learning_rate": 4.6555798427463214e-05,
+      "loss": 0.0001585719734430313,
+      "step": 17400
+    },
+    {
+      "epoch": 4.53068122724909,
+      "grad_norm": 0.0023282505571842194,
+      "learning_rate": 4.653976609950702e-05,
+      "loss": 0.00015766914933919906,
+      "step": 17425
+    },
+    {
+      "epoch": 4.53718148725949,
+      "grad_norm": 0.0025958914775401354,
+      "learning_rate": 4.6523699317722344e-05,
+      "loss": 0.00016013462096452712,
+      "step": 17450
+    },
+    {
+      "epoch": 4.543681747269891,
+      "grad_norm": 0.0014435028424486518,
+      "learning_rate": 4.650759810780872e-05,
+      "loss": 0.0001541033573448658,
+      "step": 17475
+    },
+    {
+      "epoch": 4.550182007280291,
+      "grad_norm": 0.0009686989360488951,
+      "learning_rate": 4.6491462495520755e-05,
+      "loss": 0.00014713797718286515,
+      "step": 17500
+    },
+    {
+      "epoch": 4.556682267290691,
+      "grad_norm": 0.002669141162186861,
+      "learning_rate": 4.64752925066681e-05,
+      "loss": 0.0001501713041216135,
+      "step": 17525
+    },
+    {
+      "epoch": 4.563182527301092,
+      "grad_norm": 0.001345828059129417,
+      "learning_rate": 4.645908816711538e-05,
+      "loss": 0.00015742238610982895,
+      "step": 17550
+    },
+    {
+      "epoch": 4.569682787311493,
+      "grad_norm": 0.0014919483801349998,
+      "learning_rate": 4.644284950278217e-05,
+      "loss": 0.00014388660900294781,
+      "step": 17575
+    },
+    {
+      "epoch": 4.5761830473218925,
+      "grad_norm": 0.0015580441104248166,
+      "learning_rate": 4.6426576539642944e-05,
+      "loss": 0.0001566757448017597,
+      "step": 17600
+    },
+    {
+      "epoch": 4.582683307332293,
+      "grad_norm": 0.000979835633188486,
+      "learning_rate": 4.641026930372703e-05,
+      "loss": 0.00015022841282188892,
+      "step": 17625
+    },
+    {
+      "epoch": 4.589183567342694,
+      "grad_norm": 0.002203001407906413,
+      "learning_rate": 4.639392782111861e-05,
+      "loss": 0.00016715886071324348,
+      "step": 17650
+    },
+    {
+      "epoch": 4.595683827353094,
+      "grad_norm": 0.0007987716817297041,
+      "learning_rate": 4.6377552117956605e-05,
+      "loss": 0.00013450652360916138,
+      "step": 17675
+    },
+    {
+      "epoch": 4.6021840873634945,
+      "grad_norm": 0.004143223632127047,
+      "learning_rate": 4.636114222043471e-05,
+      "loss": 0.0001484651118516922,
+      "step": 17700
+    },
+    {
+      "epoch": 4.608684347373895,
+      "grad_norm": 0.0023716369178146124,
+      "learning_rate": 4.634469815480128e-05,
+      "loss": 0.00015488354489207268,
+      "step": 17725
+    },
+    {
+      "epoch": 4.615184607384295,
+      "grad_norm": 0.0012466327752918005,
+      "learning_rate": 4.6328219947359345e-05,
+      "loss": 0.0001378494407981634,
+      "step": 17750
+    },
+    {
+      "epoch": 4.621684867394696,
+      "grad_norm": 0.0010982052190229297,
+      "learning_rate": 4.631170762446654e-05,
+      "loss": 0.00016040455549955367,
+      "step": 17775
+    },
+    {
+      "epoch": 4.6281851274050965,
+      "grad_norm": 0.0019075711024925113,
+      "learning_rate": 4.6295161212535085e-05,
+      "loss": 0.0001590457744896412,
+      "step": 17800
+    },
+    {
+      "epoch": 4.634685387415496,
+      "grad_norm": 0.0032204068265855312,
+      "learning_rate": 4.62785807380317e-05,
+      "loss": 0.00015350147150456906,
+      "step": 17825
+    },
+    {
+      "epoch": 4.641185647425897,
+      "grad_norm": 0.001020675990730524,
+      "learning_rate": 4.62619662274776e-05,
+      "loss": 0.00014694424346089362,
+      "step": 17850
+    },
+    {
+      "epoch": 4.647685907436298,
+      "grad_norm": 0.0016116098267957568,
+      "learning_rate": 4.624531770744845e-05,
+      "loss": 0.0001344368513673544,
+      "step": 17875
+    },
+    {
+      "epoch": 4.6541861674466976,
+      "grad_norm": 0.0036316344048827887,
+      "learning_rate": 4.622863520457431e-05,
+      "loss": 0.0001422207895666361,
+      "step": 17900
+    },
+    {
+      "epoch": 4.660686427457098,
+      "grad_norm": 0.0025543614756315947,
+      "learning_rate": 4.621191874553961e-05,
+      "loss": 0.00013731281273066998,
+      "step": 17925
+    },
+    {
+      "epoch": 4.667186687467499,
+      "grad_norm": 0.0033108708448708057,
+      "learning_rate": 4.619516835708305e-05,
+      "loss": 0.0001466973777860403,
+      "step": 17950
+    },
+    {
+      "epoch": 4.673686947477899,
+      "grad_norm": 0.002512441948056221,
+      "learning_rate": 4.617838406599766e-05,
+      "loss": 0.00015655793249607085,
+      "step": 17975
+    },
+    {
+      "epoch": 4.6801872074882995,
+      "grad_norm": 0.0017554826335981488,
+      "learning_rate": 4.616156589913066e-05,
+      "loss": 0.0001465380098670721,
+      "step": 18000
+    },
+    {
+      "epoch": 4.6866874674987,
+      "grad_norm": 0.0014095176011323929,
+      "learning_rate": 4.614471388338346e-05,
+      "loss": 0.00014162415638566017,
+      "step": 18025
+    },
+    {
+      "epoch": 4.6931877275091,
+      "grad_norm": 0.001687633222900331,
+      "learning_rate": 4.612782804571164e-05,
+      "loss": 0.0001689901389181614,
+      "step": 18050
+    },
+    {
+      "epoch": 4.699687987519501,
+      "grad_norm": 0.008750260807573795,
+      "learning_rate": 4.6110908413124844e-05,
+      "loss": 0.00015635639429092407,
+      "step": 18075
+    },
+    {
+      "epoch": 4.7061882475299015,
+      "grad_norm": 0.002474795328453183,
+      "learning_rate": 4.60939550126868e-05,
+      "loss": 0.0001487763598561287,
+      "step": 18100
+    },
+    {
+      "epoch": 4.712688507540301,
+      "grad_norm": 0.0008304418297484517,
+      "learning_rate": 4.607696787151522e-05,
+      "loss": 0.00015151412226259707,
+      "step": 18125
+    },
+    {
+      "epoch": 4.719188767550702,
+      "grad_norm": 0.0018005773890763521,
+      "learning_rate": 4.605994701678182e-05,
+      "loss": 0.0001472252421081066,
+      "step": 18150
+    },
+    {
+      "epoch": 4.725689027561103,
+      "grad_norm": 0.004015025682747364,
+      "learning_rate": 4.604289247571222e-05,
+      "loss": 0.00014193103648722173,
+      "step": 18175
+    },
+    {
+      "epoch": 4.732189287571503,
+      "grad_norm": 0.004501794930547476,
+      "learning_rate": 4.6025804275585923e-05,
+      "loss": 0.000141292754560709,
+      "step": 18200
+    },
+    {
+      "epoch": 4.738689547581903,
+      "grad_norm": 0.003402718808501959,
+      "learning_rate": 4.600868244373629e-05,
+      "loss": 0.0001660127192735672,
+      "step": 18225
+    },
+    {
+      "epoch": 4.745189807592304,
+      "grad_norm": 0.002020718529820442,
+      "learning_rate": 4.599152700755045e-05,
+      "loss": 0.00014321424067020416,
+      "step": 18250
+    },
+    {
+      "epoch": 4.751690067602704,
+      "grad_norm": 0.0010505187092348933,
+      "learning_rate": 4.597433799446931e-05,
+      "loss": 0.00013493100181221962,
+      "step": 18275
+    },
+    {
+      "epoch": 4.758190327613105,
+      "grad_norm": 0.001982392743229866,
+      "learning_rate": 4.5957115431987456e-05,
+      "loss": 0.00015888415277004242,
+      "step": 18300
+    },
+    {
+      "epoch": 4.764690587623505,
+      "grad_norm": 0.0012624519877135754,
+      "learning_rate": 4.593985934765317e-05,
+      "loss": 0.00015005415305495263,
+      "step": 18325
+    },
+    {
+      "epoch": 4.771190847633905,
+      "grad_norm": 0.0010668979957699776,
+      "learning_rate": 4.5922569769068337e-05,
+      "loss": 0.0001418204139918089,
+      "step": 18350
+    },
+    {
+      "epoch": 4.777691107644306,
+      "grad_norm": 0.001417837687768042,
+      "learning_rate": 4.590524672388842e-05,
+      "loss": 0.00015240835025906563,
+      "step": 18375
+    },
+    {
+      "epoch": 4.7841913676547065,
+      "grad_norm": 0.00182001618668437,
+      "learning_rate": 4.5887890239822394e-05,
+      "loss": 0.00014681240543723106,
+      "step": 18400
+    },
+    {
+      "epoch": 4.790691627665106,
+      "grad_norm": 0.002036181977018714,
+      "learning_rate": 4.587050034463276e-05,
+      "loss": 0.00013212449848651887,
+      "step": 18425
+    },
+    {
+      "epoch": 4.797191887675507,
+      "grad_norm": 0.0024265011306852102,
+      "learning_rate": 4.5853077066135425e-05,
+      "loss": 0.00015206038020551204,
+      "step": 18450
+    },
+    {
+      "epoch": 4.803692147685908,
+      "grad_norm": 0.002515725791454315,
+      "learning_rate": 4.583562043219972e-05,
+      "loss": 0.0001493960525840521,
+      "step": 18475
+    },
+    {
+      "epoch": 4.810192407696308,
+      "grad_norm": 0.0023409551940858364,
+      "learning_rate": 4.5818130470748315e-05,
+      "loss": 0.00015377653762698172,
+      "step": 18500
+    },
+    {
+      "epoch": 4.816692667706708,
+      "grad_norm": 0.002436652546748519,
+      "learning_rate": 4.5800607209757195e-05,
+      "loss": 0.0001548993121832609,
+      "step": 18525
+    },
+    {
+      "epoch": 4.823192927717109,
+      "grad_norm": 0.0018788946326822042,
+      "learning_rate": 4.578305067725561e-05,
+      "loss": 0.00014670987613499164,
+      "step": 18550
+    },
+    {
+      "epoch": 4.829693187727509,
+      "grad_norm": 0.0037026950158178806,
+      "learning_rate": 4.5765460901326016e-05,
+      "loss": 0.00014252387918531895,
+      "step": 18575
+    },
+    {
+      "epoch": 4.83619344773791,
+      "grad_norm": 0.002747404621914029,
+      "learning_rate": 4.574783791010407e-05,
+      "loss": 0.00015695074573159217,
+      "step": 18600
+    },
+    {
+      "epoch": 4.84269370774831,
+      "grad_norm": 0.0011979611590504646,
+      "learning_rate": 4.573018173177853e-05,
+      "loss": 0.000147454421967268,
+      "step": 18625
+    },
+    {
+      "epoch": 4.84919396775871,
+      "grad_norm": 0.001963973045349121,
+      "learning_rate": 4.5712492394591245e-05,
+      "loss": 0.00014441316016018392,
+      "step": 18650
+    },
+    {
+      "epoch": 4.855694227769111,
+      "grad_norm": 0.001429482363164425,
+      "learning_rate": 4.569476992683712e-05,
+      "loss": 0.0001541359443217516,
+      "step": 18675
+    },
+    {
+      "epoch": 4.862194487779512,
+      "grad_norm": 0.001821653451770544,
+      "learning_rate": 4.567701435686404e-05,
+      "loss": 0.00013285648077726364,
+      "step": 18700
+    },
+    {
+      "epoch": 4.868694747789911,
+      "grad_norm": 0.003140270709991455,
+      "learning_rate": 4.565922571307285e-05,
+      "loss": 0.00015356378629803656,
+      "step": 18725
+    },
+    {
+      "epoch": 4.875195007800312,
+      "grad_norm": 0.0016336648259311914,
+      "learning_rate": 4.5641404023917274e-05,
+      "loss": 0.00015476999804377555,
+      "step": 18750
+    },
+    {
+      "epoch": 4.881695267810713,
+      "grad_norm": 0.00110739900264889,
+      "learning_rate": 4.5623549317903915e-05,
+      "loss": 0.00015768682584166527,
+      "step": 18775
+    },
+    {
+      "epoch": 4.888195527821113,
+      "grad_norm": 0.002892926102504134,
+      "learning_rate": 4.560566162359217e-05,
+      "loss": 0.00015479933470487594,
+      "step": 18800
+    },
+    {
+      "epoch": 4.894695787831513,
+      "grad_norm": 0.0010525796096771955,
+      "learning_rate": 4.558774096959423e-05,
+      "loss": 0.00014340215362608434,
+      "step": 18825
+    },
+    {
+      "epoch": 4.901196047841914,
+      "grad_norm": 0.0016999448416754603,
+      "learning_rate": 4.556978738457499e-05,
+      "loss": 0.00014264050871133803,
+      "step": 18850
+    },
+    {
+      "epoch": 4.907696307852314,
+      "grad_norm": 0.0014307581586763263,
+      "learning_rate": 4.555180089725201e-05,
+      "loss": 0.0001373931858688593,
+      "step": 18875
+    },
+    {
+      "epoch": 4.914196567862715,
+      "grad_norm": 0.0008670628885738552,
+      "learning_rate": 4.5533781536395495e-05,
+      "loss": 0.00014224116690456868,
+      "step": 18900
+    },
+    {
+      "epoch": 4.920696827873115,
+      "grad_norm": 0.0031871842220425606,
+      "learning_rate": 4.551572933082822e-05,
+      "loss": 0.00014263702556490898,
+      "step": 18925
+    },
+    {
+      "epoch": 4.927197087883515,
+      "grad_norm": 0.0009962792973965406,
+      "learning_rate": 4.5497644309425516e-05,
+      "loss": 0.00013574679382145406,
+      "step": 18950
+    },
+    {
+      "epoch": 4.933697347893916,
+      "grad_norm": 0.0030755933839827776,
+      "learning_rate": 4.547952650111517e-05,
+      "loss": 0.0001388239860534668,
+      "step": 18975
+    },
+    {
+      "epoch": 4.940197607904317,
+      "grad_norm": 0.002856396371498704,
+      "learning_rate": 4.546137593487746e-05,
+      "loss": 0.00013483071699738503,
+      "step": 19000
+    },
+    {
+      "epoch": 4.946697867914716,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5443192639745015e-05,
+      "loss": 0.00013662678189575673,
+      "step": 19025
+    },
+    {
+      "epoch": 4.953198127925117,
+      "grad_norm": 0.002188618527725339,
+      "learning_rate": 4.542497664480284e-05,
+      "loss": 0.00014385654591023923,
+      "step": 19050
+    },
+    {
+      "epoch": 4.959698387935518,
+      "grad_norm": 0.0012306453427299857,
+      "learning_rate": 4.5406727979188255e-05,
+      "loss": 0.0001329208817332983,
+      "step": 19075
+    },
+    {
+      "epoch": 4.966198647945918,
+      "grad_norm": 0.0013932990841567516,
+      "learning_rate": 4.5388446672090814e-05,
+      "loss": 0.00014986141584813595,
+      "step": 19100
+    },
+    {
+      "epoch": 4.972698907956318,
+      "grad_norm": 0.0017995877424255013,
+      "learning_rate": 4.5370132752752294e-05,
+      "loss": 0.00016053756698966026,
+      "step": 19125
+    },
+    {
+      "epoch": 4.979199167966719,
+      "grad_norm": 0.0038302044849842787,
+      "learning_rate": 4.5351786250466636e-05,
+      "loss": 0.0001371334958821535,
+      "step": 19150
+    },
+    {
+      "epoch": 4.985699427977119,
+      "grad_norm": 0.0007338288123719394,
+      "learning_rate": 4.53334071945799e-05,
+      "loss": 0.00013917976059019566,
+      "step": 19175
+    },
+    {
+      "epoch": 4.99219968798752,
+      "grad_norm": 0.0012923972681164742,
+      "learning_rate": 4.531499561449022e-05,
+      "loss": 0.00014735984615981578,
+      "step": 19200
+    },
+    {
+      "epoch": 4.99869994799792,
+      "grad_norm": 0.003122801659628749,
+      "learning_rate": 4.529655153964775e-05,
+      "loss": 0.0001434572786092758,
+      "step": 19225
+    },
+    {
+      "epoch": 5.00520020800832,
+      "grad_norm": 0.000923612155020237,
+      "learning_rate": 4.52780749995546e-05,
+      "loss": 0.00012164958752691746,
+      "step": 19250
+    },
+    {
+      "epoch": 5.011700468018721,
+      "grad_norm": 0.0016798863653093576,
+      "learning_rate": 4.525956602376486e-05,
+      "loss": 0.0001315576024353504,
+      "step": 19275
+    },
+    {
+      "epoch": 5.018200728029121,
+      "grad_norm": 0.001232946990057826,
+      "learning_rate": 4.5241024641884456e-05,
+      "loss": 0.00013162718154489993,
+      "step": 19300
+    },
+    {
+      "epoch": 5.0247009880395215,
+      "grad_norm": 0.0009978599846363068,
+      "learning_rate": 4.522245088357116e-05,
+      "loss": 0.00013904462568461895,
+      "step": 19325
+    },
+    {
+      "epoch": 5.031201248049922,
+      "grad_norm": 0.0010107139823958278,
+      "learning_rate": 4.5203844778534554e-05,
+      "loss": 0.00015151292085647584,
+      "step": 19350
+    },
+    {
+      "epoch": 5.037701508060322,
+      "grad_norm": 0.0012251532170921564,
+      "learning_rate": 4.518520635653594e-05,
+      "loss": 0.0001429096981883049,
+      "step": 19375
+    },
+    {
+      "epoch": 5.044201768070723,
+      "grad_norm": 0.0012166014639660716,
+      "learning_rate": 4.5166535647388296e-05,
+      "loss": 0.0001403882261365652,
+      "step": 19400
+    },
+    {
+      "epoch": 5.0507020280811235,
+      "grad_norm": 0.0014541230630129576,
+      "learning_rate": 4.5147832680956284e-05,
+      "loss": 0.0001278739608824253,
+      "step": 19425
+    },
+    {
+      "epoch": 5.057202288091523,
+      "grad_norm": 0.0011057070223614573,
+      "learning_rate": 4.512909748715613e-05,
+      "loss": 0.00013740201480686665,
+      "step": 19450
+    },
+    {
+      "epoch": 5.063702548101924,
+      "grad_norm": 0.0014885697746649384,
+      "learning_rate": 4.511033009595564e-05,
+      "loss": 0.00015405756421387195,
+      "step": 19475
+    },
+    {
+      "epoch": 5.070202808112325,
+      "grad_norm": 0.001101053785532713,
+      "learning_rate": 4.50915305373741e-05,
+      "loss": 0.00013779288157820703,
+      "step": 19500
+    },
+    {
+      "epoch": 5.0767030681227245,
+      "grad_norm": 0.0032690393272787333,
+      "learning_rate": 4.507269884148224e-05,
+      "loss": 0.00013352164067327975,
+      "step": 19525
+    },
+    {
+      "epoch": 5.083203328133125,
+      "grad_norm": 0.0009752037585712969,
+      "learning_rate": 4.505383503840221e-05,
+      "loss": 0.0001376598235219717,
+      "step": 19550
+    },
+    {
+      "epoch": 5.089703588143526,
+      "grad_norm": 0.002196341287344694,
+      "learning_rate": 4.503493915830753e-05,
+      "loss": 0.00013712641783058643,
+      "step": 19575
+    },
+    {
+      "epoch": 5.096203848153926,
+      "grad_norm": 0.0010258402908220887,
+      "learning_rate": 4.5016011231422995e-05,
+      "loss": 0.0001346300169825554,
+      "step": 19600
+    },
+    {
+      "epoch": 5.1027041081643265,
+      "grad_norm": 0.003978818655014038,
+      "learning_rate": 4.4997051288024686e-05,
+      "loss": 0.00014454452320933342,
+      "step": 19625
+    },
+    {
+      "epoch": 5.109204368174727,
+      "grad_norm": 0.0037584619130939245,
+      "learning_rate": 4.497805935843989e-05,
+      "loss": 0.00013516144827008247,
+      "step": 19650
+    },
+    {
+      "epoch": 5.115704628185127,
+      "grad_norm": 0.0009517311118543148,
+      "learning_rate": 4.4959035473047054e-05,
+      "loss": 0.00013424252159893513,
+      "step": 19675
+    },
+    {
+      "epoch": 5.122204888195528,
+      "grad_norm": 0.004111779388040304,
+      "learning_rate": 4.4939979662275745e-05,
+      "loss": 0.0001388068776577711,
+      "step": 19700
+    },
+    {
+      "epoch": 5.1287051482059285,
+      "grad_norm": 0.0012805311707779765,
+      "learning_rate": 4.4920891956606594e-05,
+      "loss": 0.00013941220939159393,
+      "step": 19725
+    },
+    {
+      "epoch": 5.135205408216328,
+      "grad_norm": 0.004675137344747782,
+      "learning_rate": 4.490177238657124e-05,
+      "loss": 0.00013734254986047745,
+      "step": 19750
+    },
+    {
+      "epoch": 5.141705668226729,
+      "grad_norm": 0.002303802641108632,
+      "learning_rate": 4.48826209827523e-05,
+      "loss": 0.00016402501612901687,
+      "step": 19775
+    },
+    {
+      "epoch": 5.14820592823713,
+      "grad_norm": 0.0019201309187337756,
+      "learning_rate": 4.4863437775783323e-05,
+      "loss": 0.00014541741460561753,
+      "step": 19800
+    },
+    {
+      "epoch": 5.15470618824753,
+      "grad_norm": 0.0026621189899742603,
+      "learning_rate": 4.48442227963487e-05,
+      "loss": 0.0001568194106221199,
+      "step": 19825
+    },
+    {
+      "epoch": 5.16120644825793,
+      "grad_norm": 0.001147376955486834,
+      "learning_rate": 4.4824976075183654e-05,
+      "loss": 0.0001450801733881235,
+      "step": 19850
+    },
+    {
+      "epoch": 5.167706708268331,
+      "grad_norm": 0.005741590168327093,
+      "learning_rate": 4.4805697643074206e-05,
+      "loss": 0.00014287268742918969,
+      "step": 19875
+    },
+    {
+      "epoch": 5.174206968278731,
+      "grad_norm": 0.0028361014556139708,
+      "learning_rate": 4.4786387530857056e-05,
+      "loss": 0.00015186586417257787,
+      "step": 19900
+    },
+    {
+      "epoch": 5.1807072282891315,
+      "grad_norm": 0.0014331071870401502,
+      "learning_rate": 4.476704576941961e-05,
+      "loss": 0.00015548578463494777,
+      "step": 19925
+    },
+    {
+      "epoch": 5.187207488299532,
+      "grad_norm": 0.0030039548873901367,
+      "learning_rate": 4.47476723896999e-05,
+      "loss": 0.00014972995035350323,
+      "step": 19950
+    },
+    {
+      "epoch": 5.193707748309932,
+      "grad_norm": 0.0022576020564883947,
+      "learning_rate": 4.4728267422686496e-05,
+      "loss": 0.0001554740034043789,
+      "step": 19975
+    },
+    {
+      "epoch": 5.200208008320333,
+      "grad_norm": 0.0012061097659170628,
+      "learning_rate": 4.4708830899418544e-05,
+      "loss": 0.0001298661343753338,
+      "step": 20000
+    },
+    {
+      "epoch": 5.2067082683307335,
+      "grad_norm": 0.002239755354821682,
+      "learning_rate": 4.4689362850985625e-05,
+      "loss": 0.00015579594299197197,
+      "step": 20025
+    },
+    {
+      "epoch": 5.213208528341133,
+      "grad_norm": 0.00401849253103137,
+      "learning_rate": 4.466986330852777e-05,
+      "loss": 0.00016034215688705443,
+      "step": 20050
+    },
+    {
+      "epoch": 5.219708788351534,
+      "grad_norm": 0.002211707876995206,
+      "learning_rate": 4.465033230323538e-05,
+      "loss": 0.0001427256502211094,
+      "step": 20075
+    },
+    {
+      "epoch": 5.226209048361935,
+      "grad_norm": 0.002577380510047078,
+      "learning_rate": 4.463076986634916e-05,
+      "loss": 0.00012760492973029613,
+      "step": 20100
+    },
+    {
+      "epoch": 5.232709308372335,
+      "grad_norm": 0.0013343971222639084,
+      "learning_rate": 4.461117602916014e-05,
+      "loss": 0.00014281315729022025,
+      "step": 20125
+    },
+    {
+      "epoch": 5.239209568382735,
+      "grad_norm": 0.0012126417132094502,
+      "learning_rate": 4.459155082300952e-05,
+      "loss": 0.00015582663938403128,
+      "step": 20150
+    },
+    {
+      "epoch": 5.245709828393136,
+      "grad_norm": 0.0,
+      "learning_rate": 4.457189427928872e-05,
+      "loss": 0.00013873948715627195,
+      "step": 20175
+    },
+    {
+      "epoch": 5.252210088403536,
+      "grad_norm": 0.000978428521193564,
+      "learning_rate": 4.455220642943927e-05,
+      "loss": 0.00013329820707440375,
+      "step": 20200
+    },
+    {
+      "epoch": 5.258710348413937,
+      "grad_norm": 0.0025654276832938194,
+      "learning_rate": 4.4532487304952764e-05,
+      "loss": 0.00014376726001501085,
+      "step": 20225
+    },
+    {
+      "epoch": 5.265210608424337,
+      "grad_norm": 0.0027144309133291245,
+      "learning_rate": 4.4512736937370844e-05,
+      "loss": 0.00014224441722035408,
+      "step": 20250
+    },
+    {
+      "epoch": 5.271710868434737,
+      "grad_norm": 0.001510527217760682,
+      "learning_rate": 4.4492955358285104e-05,
+      "loss": 0.00013470981270074844,
+      "step": 20275
+    },
+    {
+      "epoch": 5.278211128445138,
+      "grad_norm": 0.0009118269081227481,
+      "learning_rate": 4.447314259933708e-05,
+      "loss": 0.00013923139311373234,
+      "step": 20300
+    },
+    {
+      "epoch": 5.284711388455539,
+      "grad_norm": 0.0012834982480853796,
+      "learning_rate": 4.445329869221817e-05,
+      "loss": 0.00013785173185169698,
+      "step": 20325
+    },
+    {
+      "epoch": 5.291211648465938,
+      "grad_norm": 0.0020495106000453234,
+      "learning_rate": 4.4433423668669604e-05,
+      "loss": 0.0001358055043965578,
+      "step": 20350
+    },
+    {
+      "epoch": 5.297711908476339,
+      "grad_norm": 0.0019294389057904482,
+      "learning_rate": 4.4413517560482376e-05,
+      "loss": 0.00013429016806185246,
+      "step": 20375
+    },
+    {
+      "epoch": 5.30421216848674,
+      "grad_norm": 0.0020139359403401613,
+      "learning_rate": 4.4393580399497214e-05,
+      "loss": 0.00014911632053554057,
+      "step": 20400
+    },
+    {
+      "epoch": 5.31071242849714,
+      "grad_norm": 0.0017949657049030066,
+      "learning_rate": 4.4373612217604496e-05,
+      "loss": 0.000138550391420722,
+      "step": 20425
+    },
+    {
+      "epoch": 5.31721268850754,
+      "grad_norm": 0.0008366036345250905,
+      "learning_rate": 4.435361304674423e-05,
+      "loss": 0.00013641472905874253,
+      "step": 20450
+    },
+    {
+      "epoch": 5.323712948517941,
+      "grad_norm": 0.002636739518493414,
+      "learning_rate": 4.4333582918906e-05,
+      "loss": 0.0001424773782491684,
+      "step": 20475
+    },
+    {
+      "epoch": 5.330213208528341,
+      "grad_norm": 0.0007566487183794379,
+      "learning_rate": 4.431352186612889e-05,
+      "loss": 0.00013931132853031158,
+      "step": 20500
+    },
+    {
+      "epoch": 5.336713468538742,
+      "grad_norm": 0.001040862058289349,
+      "learning_rate": 4.429342992050148e-05,
+      "loss": 0.0001413393020629883,
+      "step": 20525
+    },
+    {
+      "epoch": 5.343213728549142,
+      "grad_norm": 0.0007505036774091423,
+      "learning_rate": 4.427330711416172e-05,
+      "loss": 0.0001287364587187767,
+      "step": 20550
+    },
+    {
+      "epoch": 5.349713988559542,
+      "grad_norm": 0.003622201969847083,
+      "learning_rate": 4.4253153479296964e-05,
+      "loss": 0.0001598481647670269,
+      "step": 20575
+    },
+    {
+      "epoch": 5.356214248569943,
+      "grad_norm": 0.0020210405346006155,
+      "learning_rate": 4.423296904814386e-05,
+      "loss": 0.00013407637365162373,
+      "step": 20600
+    },
+    {
+      "epoch": 5.362714508580344,
+      "grad_norm": 0.0033002665732055902,
+      "learning_rate": 4.4212753852988296e-05,
+      "loss": 0.0001285996474325657,
+      "step": 20625
+    },
+    {
+      "epoch": 5.369214768590743,
+      "grad_norm": 0.002899797400459647,
+      "learning_rate": 4.419250792616542e-05,
+      "loss": 0.00014401165768504142,
+      "step": 20650
+    },
+    {
+      "epoch": 5.375715028601144,
+      "grad_norm": 0.0007184264832176268,
+      "learning_rate": 4.4172231300059485e-05,
+      "loss": 0.0001254305336624384,
+      "step": 20675
+    },
+    {
+      "epoch": 5.382215288611545,
+      "grad_norm": 0.0011705051874741912,
+      "learning_rate": 4.415192400710388e-05,
+      "loss": 0.0001362087856978178,
+      "step": 20700
+    },
+    {
+      "epoch": 5.382215288611545,
+      "eval_loss": 0.0005826261476613581,
+      "eval_runtime": 7.6309,
+      "eval_samples_per_second": 655.227,
+      "eval_steps_per_second": 3.538,
+      "step": 20700
+    },
+    {
+      "epoch": 5.382215288611545,
+      "eval_holdout_avg_strict_f1": 0.07663434321377455,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.29186602870813394,
+      "eval_holdout_nemotron_pii_strict_f1": 0.18598039215686274,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.20000000000000004,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.16513135448652336,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 20700
+    },
+    {
+      "epoch": 5.388715548621945,
+      "grad_norm": 0.0011600384023040533,
+      "learning_rate": 4.413158607978104e-05,
+      "loss": 0.0001383385993540287,
+      "step": 20725
+    },
+    {
+      "epoch": 5.395215808632345,
+      "grad_norm": 0.0012965621426701546,
+      "learning_rate": 4.4111217550622385e-05,
+      "loss": 0.00013235370628535747,
+      "step": 20750
+    },
+    {
+      "epoch": 5.401716068642746,
+      "grad_norm": 0.004435330163687468,
+      "learning_rate": 4.409081845220831e-05,
+      "loss": 0.00014182785525918007,
+      "step": 20775
+    },
+    {
+      "epoch": 5.408216328653146,
+      "grad_norm": 0.0034650543238967657,
+      "learning_rate": 4.407038881716808e-05,
+      "loss": 0.00014904932118952273,
+      "step": 20800
+    },
+    {
+      "epoch": 5.414716588663547,
+      "grad_norm": 0.0009779005777090788,
+      "learning_rate": 4.404992867817983e-05,
+      "loss": 0.00013652848079800605,
+      "step": 20825
+    },
+    {
+      "epoch": 5.4212168486739465,
+      "grad_norm": 0.0036383962724357843,
+      "learning_rate": 4.402943806797048e-05,
+      "loss": 0.00014459606260061264,
+      "step": 20850
+    },
+    {
+      "epoch": 5.427717108684347,
+      "grad_norm": 0.001130777527578175,
+      "learning_rate": 4.4008917019315676e-05,
+      "loss": 0.00013321375474333762,
+      "step": 20875
+    },
+    {
+      "epoch": 5.434217368694748,
+      "grad_norm": 0.00240878714248538,
+      "learning_rate": 4.398836556503976e-05,
+      "loss": 0.00013470495119690895,
+      "step": 20900
+    },
+    {
+      "epoch": 5.440717628705148,
+      "grad_norm": 0.0017072060145437717,
+      "learning_rate": 4.3967783738015724e-05,
+      "loss": 0.000141152273863554,
+      "step": 20925
+    },
+    {
+      "epoch": 5.4472178887155485,
+      "grad_norm": 0.00344785675406456,
+      "learning_rate": 4.394717157116512e-05,
+      "loss": 0.000126673337072134,
+      "step": 20950
+    },
+    {
+      "epoch": 5.453718148725949,
+      "grad_norm": 0.0014222816098481417,
+      "learning_rate": 4.392652909745804e-05,
+      "loss": 0.00013333518989384175,
+      "step": 20975
+    },
+    {
+      "epoch": 5.460218408736349,
+      "grad_norm": 0.0015250548021867871,
+      "learning_rate": 4.390585634991305e-05,
+      "loss": 0.00014791802503168582,
+      "step": 21000
+    },
+    {
+      "epoch": 5.46671866874675,
+      "grad_norm": 0.0023152055218815804,
+      "learning_rate": 4.388515336159716e-05,
+      "loss": 0.00013493741862475873,
+      "step": 21025
+    },
+    {
+      "epoch": 5.47321892875715,
+      "grad_norm": 0.002872672863304615,
+      "learning_rate": 4.386442016562572e-05,
+      "loss": 0.0001456017978489399,
+      "step": 21050
+    },
+    {
+      "epoch": 5.47971918876755,
+      "grad_norm": 0.001714720157906413,
+      "learning_rate": 4.38436567951624e-05,
+      "loss": 0.00012950732372701167,
+      "step": 21075
+    },
+    {
+      "epoch": 5.486219448777951,
+      "grad_norm": 0.0011554666562005877,
+      "learning_rate": 4.382286328341918e-05,
+      "loss": 0.00012936241924762726,
+      "step": 21100
+    },
+    {
+      "epoch": 5.492719708788352,
+      "grad_norm": 0.001156545476987958,
+      "learning_rate": 4.38020396636562e-05,
+      "loss": 0.00012953244149684905,
+      "step": 21125
+    },
+    {
+      "epoch": 5.4992199687987515,
+      "grad_norm": 0.0025892690755426884,
+      "learning_rate": 4.378118596918179e-05,
+      "loss": 0.00014081483706831932,
+      "step": 21150
+    },
+    {
+      "epoch": 5.505720228809152,
+      "grad_norm": 0.006588731426745653,
+      "learning_rate": 4.376030223335237e-05,
+      "loss": 0.00014686424285173416,
+      "step": 21175
+    },
+    {
+      "epoch": 5.512220488819553,
+      "grad_norm": 0.0020245795603841543,
+      "learning_rate": 4.3739388489572416e-05,
+      "loss": 0.00013585050590336322,
+      "step": 21200
+    },
+    {
+      "epoch": 5.518720748829953,
+      "grad_norm": 0.0038208116311579943,
+      "learning_rate": 4.371844477129441e-05,
+      "loss": 0.0001333075761795044,
+      "step": 21225
+    },
+    {
+      "epoch": 5.5252210088403535,
+      "grad_norm": 0.002841449109837413,
+      "learning_rate": 4.369747111201877e-05,
+      "loss": 0.00012721997685730458,
+      "step": 21250
+    },
+    {
+      "epoch": 5.531721268850754,
+      "grad_norm": 0.0024051882792264223,
+      "learning_rate": 4.367646754529383e-05,
+      "loss": 0.00013284836895763875,
+      "step": 21275
+    },
+    {
+      "epoch": 5.538221528861154,
+      "grad_norm": 0.005994176957756281,
+      "learning_rate": 4.365543410471571e-05,
+      "loss": 0.00013940009288489818,
+      "step": 21300
+    },
+    {
+      "epoch": 5.544721788871555,
+      "grad_norm": 0.00116148276720196,
+      "learning_rate": 4.363437082392838e-05,
+      "loss": 0.00014780103228986263,
+      "step": 21325
+    },
+    {
+      "epoch": 5.5512220488819555,
+      "grad_norm": 0.0018392236670479178,
+      "learning_rate": 4.361327773662349e-05,
+      "loss": 0.00014781424775719642,
+      "step": 21350
+    },
+    {
+      "epoch": 5.557722308892355,
+      "grad_norm": 0.0016857964219525456,
+      "learning_rate": 4.359215487654039e-05,
+      "loss": 0.00013778201304376125,
+      "step": 21375
+    },
+    {
+      "epoch": 5.564222568902756,
+      "grad_norm": 0.002162114717066288,
+      "learning_rate": 4.357100227746606e-05,
+      "loss": 0.00015323886647820474,
+      "step": 21400
+    },
+    {
+      "epoch": 5.570722828913157,
+      "grad_norm": 0.0030271727591753006,
+      "learning_rate": 4.354981997323501e-05,
+      "loss": 0.00014950009062886238,
+      "step": 21425
+    },
+    {
+      "epoch": 5.577223088923557,
+      "grad_norm": 0.002122987527400255,
+      "learning_rate": 4.352860799772931e-05,
+      "loss": 0.00014159180223941804,
+      "step": 21450
+    },
+    {
+      "epoch": 5.583723348933957,
+      "grad_norm": 0.0017169964266940951,
+      "learning_rate": 4.3507366384878486e-05,
+      "loss": 0.00014035237953066826,
+      "step": 21475
+    },
+    {
+      "epoch": 5.590223608944358,
+      "grad_norm": 0.0028544568922370672,
+      "learning_rate": 4.3486095168659435e-05,
+      "loss": 0.00013027112931013107,
+      "step": 21500
+    },
+    {
+      "epoch": 5.596723868954758,
+      "grad_norm": 0.0,
+      "learning_rate": 4.346479438309643e-05,
+      "loss": 0.00016134755685925483,
+      "step": 21525
+    },
+    {
+      "epoch": 5.6032241289651585,
+      "grad_norm": 0.0013462648494169116,
+      "learning_rate": 4.3443464062261055e-05,
+      "loss": 0.0001466800831258297,
+      "step": 21550
+    },
+    {
+      "epoch": 5.609724388975559,
+      "grad_norm": 0.0016618708614259958,
+      "learning_rate": 4.34221042402721e-05,
+      "loss": 0.00014012696221470832,
+      "step": 21575
+    },
+    {
+      "epoch": 5.616224648985959,
+      "grad_norm": 0.0015136783476918936,
+      "learning_rate": 4.3400714951295586e-05,
+      "loss": 0.00016626695170998573,
+      "step": 21600
+    },
+    {
+      "epoch": 5.62272490899636,
+      "grad_norm": 0.0022124568931758404,
+      "learning_rate": 4.337929622954463e-05,
+      "loss": 0.00014641952700912951,
+      "step": 21625
+    },
+    {
+      "epoch": 5.6292251690067605,
+      "grad_norm": 0.002614477416500449,
+      "learning_rate": 4.335784810927945e-05,
+      "loss": 0.00014230445027351379,
+      "step": 21650
+    },
+    {
+      "epoch": 5.63572542901716,
+      "grad_norm": 0.0027004072908312082,
+      "learning_rate": 4.3336370624807286e-05,
+      "loss": 0.0001326928474009037,
+      "step": 21675
+    },
+    {
+      "epoch": 5.642225689027561,
+      "grad_norm": 0.002002907684072852,
+      "learning_rate": 4.3314863810482344e-05,
+      "loss": 0.00015296720899641515,
+      "step": 21700
+    },
+    {
+      "epoch": 5.648725949037962,
+      "grad_norm": 0.0015064626932144165,
+      "learning_rate": 4.3293327700705744e-05,
+      "loss": 0.00013795320875942707,
+      "step": 21725
+    },
+    {
+      "epoch": 5.655226209048362,
+      "grad_norm": 0.002079053083434701,
+      "learning_rate": 4.3271762329925456e-05,
+      "loss": 0.00013431180268526078,
+      "step": 21750
+    },
+    {
+      "epoch": 5.661726469058762,
+      "grad_norm": 0.0014507503947243094,
+      "learning_rate": 4.325016773263628e-05,
+      "loss": 0.00014116465114057065,
+      "step": 21775
+    },
+    {
+      "epoch": 5.668226729069163,
+      "grad_norm": 0.0013049120316281915,
+      "learning_rate": 4.322854394337974e-05,
+      "loss": 0.00012939125299453734,
+      "step": 21800
+    },
+    {
+      "epoch": 5.674726989079563,
+      "grad_norm": 0.0018046763725578785,
+      "learning_rate": 4.320689099674406e-05,
+      "loss": 0.0001381920464336872,
+      "step": 21825
+    },
+    {
+      "epoch": 5.681227249089964,
+      "grad_norm": 0.0009468544158153236,
+      "learning_rate": 4.318520892736412e-05,
+      "loss": 0.00011906111612915992,
+      "step": 21850
+    },
+    {
+      "epoch": 5.687727509100364,
+      "grad_norm": 0.0017206099582836032,
+      "learning_rate": 4.316349776992136e-05,
+      "loss": 0.00013601195067167282,
+      "step": 21875
+    },
+    {
+      "epoch": 5.694227769110764,
+      "grad_norm": 0.0010455433512106538,
+      "learning_rate": 4.314175755914375e-05,
+      "loss": 0.0001409765612334013,
+      "step": 21900
+    },
+    {
+      "epoch": 5.700728029121165,
+      "grad_norm": 0.0017494077328592539,
+      "learning_rate": 4.311998832980576e-05,
+      "loss": 0.00013409361243247987,
+      "step": 21925
+    },
+    {
+      "epoch": 5.7072282891315655,
+      "grad_norm": 0.0020472300238907337,
+      "learning_rate": 4.309819011672823e-05,
+      "loss": 0.00014232087880373002,
+      "step": 21950
+    },
+    {
+      "epoch": 5.713728549141965,
+      "grad_norm": 0.0020290531683713198,
+      "learning_rate": 4.3076362954778393e-05,
+      "loss": 0.00012993237003684045,
+      "step": 21975
+    },
+    {
+      "epoch": 5.720228809152366,
+      "grad_norm": 0.003339259186759591,
+      "learning_rate": 4.3054506878869794e-05,
+      "loss": 0.0001337669137865305,
+      "step": 22000
+    },
+    {
+      "epoch": 5.726729069162767,
+      "grad_norm": 0.0016429167008027434,
+      "learning_rate": 4.3032621923962203e-05,
+      "loss": 0.00012040069326758384,
+      "step": 22025
+    },
+    {
+      "epoch": 5.733229329173167,
+      "grad_norm": 0.0021118680015206337,
+      "learning_rate": 4.3010708125061596e-05,
+      "loss": 0.00012498789466917516,
+      "step": 22050
+    },
+    {
+      "epoch": 5.739729589183567,
+      "grad_norm": 0.004076567478477955,
+      "learning_rate": 4.298876551722008e-05,
+      "loss": 0.00014244080521166324,
+      "step": 22075
+    },
+    {
+      "epoch": 5.746229849193968,
+      "grad_norm": 0.0009202787186950445,
+      "learning_rate": 4.296679413553584e-05,
+      "loss": 0.00013342386111617088,
+      "step": 22100
+    },
+    {
+      "epoch": 5.752730109204368,
+      "grad_norm": 0.0016398931620642543,
+      "learning_rate": 4.2944794015153114e-05,
+      "loss": 0.00012856223620474338,
+      "step": 22125
+    },
+    {
+      "epoch": 5.759230369214769,
+      "grad_norm": 0.001380627858452499,
+      "learning_rate": 4.292276519126207e-05,
+      "loss": 0.00013237960636615752,
+      "step": 22150
+    },
+    {
+      "epoch": 5.765730629225169,
+      "grad_norm": 0.003768360475078225,
+      "learning_rate": 4.290070769909882e-05,
+      "loss": 0.00013401787728071213,
+      "step": 22175
+    },
+    {
+      "epoch": 5.772230889235569,
+      "grad_norm": 0.0012056480627506971,
+      "learning_rate": 4.287862157394531e-05,
+      "loss": 0.00012289749458432197,
+      "step": 22200
+    },
+    {
+      "epoch": 5.77873114924597,
+      "grad_norm": 0.0010526994010433555,
+      "learning_rate": 4.285650685112929e-05,
+      "loss": 0.000126950703561306,
+      "step": 22225
+    },
+    {
+      "epoch": 5.785231409256371,
+      "grad_norm": 0.002675502561032772,
+      "learning_rate": 4.283436356602426e-05,
+      "loss": 0.00014487255364656448,
+      "step": 22250
+    },
+    {
+      "epoch": 5.79173166926677,
+      "grad_norm": 0.0016502812504768372,
+      "learning_rate": 4.2812191754049416e-05,
+      "loss": 0.00012364027090370656,
+      "step": 22275
+    },
+    {
+      "epoch": 5.798231929277171,
+      "grad_norm": 0.0022894584108144045,
+      "learning_rate": 4.278999145066956e-05,
+      "loss": 0.00014711578376591205,
+      "step": 22300
+    },
+    {
+      "epoch": 5.804732189287572,
+      "grad_norm": 0.0037363553419709206,
+      "learning_rate": 4.27677626913951e-05,
+      "loss": 0.00011959685012698173,
+      "step": 22325
+    },
+    {
+      "epoch": 5.811232449297972,
+      "grad_norm": 0.003776401514187455,
+      "learning_rate": 4.274550551178191e-05,
+      "loss": 0.0001354140229523182,
+      "step": 22350
+    },
+    {
+      "epoch": 5.817732709308372,
+      "grad_norm": 0.0012568400707095861,
+      "learning_rate": 4.272321994743137e-05,
+      "loss": 0.00013568734750151634,
+      "step": 22375
+    },
+    {
+      "epoch": 5.824232969318773,
+      "grad_norm": 0.000644272833596915,
+      "learning_rate": 4.2700906033990246e-05,
+      "loss": 0.0001478616986423731,
+      "step": 22400
+    },
+    {
+      "epoch": 5.830733229329173,
+      "grad_norm": 0.0016588791040703654,
+      "learning_rate": 4.267856380715065e-05,
+      "loss": 0.00014744569547474383,
+      "step": 22425
+    },
+    {
+      "epoch": 5.837233489339574,
+      "grad_norm": 0.0012692208401858807,
+      "learning_rate": 4.2656193302649975e-05,
+      "loss": 0.00012710162438452243,
+      "step": 22450
+    },
+    {
+      "epoch": 5.843733749349974,
+      "grad_norm": 0.004288715310394764,
+      "learning_rate": 4.2633794556270865e-05,
+      "loss": 0.00012563006952404977,
+      "step": 22475
+    },
+    {
+      "epoch": 5.850234009360374,
+      "grad_norm": 0.0016791289672255516,
+      "learning_rate": 4.261136760384112e-05,
+      "loss": 0.00011552553623914719,
+      "step": 22500
+    },
+    {
+      "epoch": 5.856734269370775,
+      "grad_norm": 0.005380165297538042,
+      "learning_rate": 4.2588912481233666e-05,
+      "loss": 0.00013402472250163556,
+      "step": 22525
+    },
+    {
+      "epoch": 5.863234529381176,
+      "grad_norm": 0.002106114523485303,
+      "learning_rate": 4.256642922436648e-05,
+      "loss": 0.0001407475583255291,
+      "step": 22550
+    },
+    {
+      "epoch": 5.8697347893915754,
+      "grad_norm": 0.0011061398545280099,
+      "learning_rate": 4.2543917869202555e-05,
+      "loss": 0.00013997879810631275,
+      "step": 22575
+    },
+    {
+      "epoch": 5.876235049401976,
+      "grad_norm": 0.0019175371853634715,
+      "learning_rate": 4.2521378451749805e-05,
+      "loss": 0.00013676457107067108,
+      "step": 22600
+    },
+    {
+      "epoch": 5.882735309412377,
+      "grad_norm": 0.0016208944143727422,
+      "learning_rate": 4.249881100806107e-05,
+      "loss": 0.0001434333622455597,
+      "step": 22625
+    },
+    {
+      "epoch": 5.889235569422777,
+      "grad_norm": 0.0018686995608732104,
+      "learning_rate": 4.247621557423397e-05,
+      "loss": 0.00013396822847425938,
+      "step": 22650
+    },
+    {
+      "epoch": 5.895735829433177,
+      "grad_norm": 0.0016240967670455575,
+      "learning_rate": 4.245359218641094e-05,
+      "loss": 0.00014730295166373253,
+      "step": 22675
+    },
+    {
+      "epoch": 5.902236089443578,
+      "grad_norm": 0.0008497955859638751,
+      "learning_rate": 4.2430940880779096e-05,
+      "loss": 0.00014496689662337302,
+      "step": 22700
+    },
+    {
+      "epoch": 5.908736349453978,
+      "grad_norm": 0.0012186749372631311,
+      "learning_rate": 4.240826169357024e-05,
+      "loss": 0.0001465201936662197,
+      "step": 22725
+    },
+    {
+      "epoch": 5.915236609464379,
+      "grad_norm": 0.002044511493295431,
+      "learning_rate": 4.238555466106074e-05,
+      "loss": 0.00013051839545369148,
+      "step": 22750
+    },
+    {
+      "epoch": 5.921736869474779,
+      "grad_norm": 0.0014217623975127935,
+      "learning_rate": 4.236281981957153e-05,
+      "loss": 0.00012539124116301537,
+      "step": 22775
+    },
+    {
+      "epoch": 5.928237129485179,
+      "grad_norm": 0.002617568476125598,
+      "learning_rate": 4.234005720546801e-05,
+      "loss": 0.0001501511223614216,
+      "step": 22800
+    },
+    {
+      "epoch": 5.93473738949558,
+      "grad_norm": 0.0041812388226389885,
+      "learning_rate": 4.231726685516002e-05,
+      "loss": 0.0001412770338356495,
+      "step": 22825
+    },
+    {
+      "epoch": 5.941237649505981,
+      "grad_norm": 0.0018407449824735522,
+      "learning_rate": 4.229444880510174e-05,
+      "loss": 0.00013424738310277462,
+      "step": 22850
+    },
+    {
+      "epoch": 5.9477379095163805,
+      "grad_norm": 0.0006838127737864852,
+      "learning_rate": 4.227160309179168e-05,
+      "loss": 0.0001266089640557766,
+      "step": 22875
+    },
+    {
+      "epoch": 5.954238169526781,
+      "grad_norm": 0.0015300296945497394,
+      "learning_rate": 4.2248729751772586e-05,
+      "loss": 0.00013262062333524227,
+      "step": 22900
+    },
+    {
+      "epoch": 5.960738429537182,
+      "grad_norm": 0.0013057185569778085,
+      "learning_rate": 4.22258288216314e-05,
+      "loss": 0.00012688847258687019,
+      "step": 22925
+    },
+    {
+      "epoch": 5.967238689547582,
+      "grad_norm": 0.0009290992747992277,
+      "learning_rate": 4.2202900337999196e-05,
+      "loss": 0.00012667604722082614,
+      "step": 22950
+    },
+    {
+      "epoch": 5.9737389495579825,
+      "grad_norm": 0.0057319290935993195,
+      "learning_rate": 4.217994433755111e-05,
+      "loss": 0.00012275479733943938,
+      "step": 22975
+    },
+    {
+      "epoch": 5.980239209568383,
+      "grad_norm": 0.0018983723130077124,
+      "learning_rate": 4.215696085700632e-05,
+      "loss": 0.00011097787879407406,
+      "step": 23000
+    },
+    {
+      "epoch": 5.986739469578783,
+      "grad_norm": 0.001285347156226635,
+      "learning_rate": 4.213394993312792e-05,
+      "loss": 0.00012497798539698124,
+      "step": 23025
+    },
+    {
+      "epoch": 5.993239729589184,
+      "grad_norm": 0.00106144102755934,
+      "learning_rate": 4.2110911602722937e-05,
+      "loss": 0.00011190829798579216,
+      "step": 23050
+    },
+    {
+      "epoch": 5.999739989599584,
+      "grad_norm": 0.0015664062229916453,
+      "learning_rate": 4.208784590264221e-05,
+      "loss": 0.00011443838477134704,
+      "step": 23075
+    },
+    {
+      "epoch": 6.006240249609984,
+      "grad_norm": 0.004631684627383947,
+      "learning_rate": 4.206475286978037e-05,
+      "loss": 0.000141043346375227,
+      "step": 23100
+    },
+    {
+      "epoch": 6.012740509620385,
+      "grad_norm": 0.0012806013692170382,
+      "learning_rate": 4.204163254107577e-05,
+      "loss": 0.00011649413034319877,
+      "step": 23125
+    },
+    {
+      "epoch": 6.019240769630785,
+      "grad_norm": 0.0010893320431932807,
+      "learning_rate": 4.201848495351042e-05,
+      "loss": 0.00011832174845039844,
+      "step": 23150
+    },
+    {
+      "epoch": 6.0257410296411855,
+      "grad_norm": 0.0015831352211534977,
+      "learning_rate": 4.1995310144109925e-05,
+      "loss": 0.00014070321805775165,
+      "step": 23175
+    },
+    {
+      "epoch": 6.032241289651586,
+      "grad_norm": 0.002874738536775112,
+      "learning_rate": 4.1972108149943445e-05,
+      "loss": 0.00012726476415991783,
+      "step": 23200
+    },
+    {
+      "epoch": 6.038741549661986,
+      "grad_norm": 0.0033063048031181097,
+      "learning_rate": 4.194887900812361e-05,
+      "loss": 0.0001378224976360798,
+      "step": 23225
+    },
+    {
+      "epoch": 6.045241809672387,
+      "grad_norm": 0.0020392874721437693,
+      "learning_rate": 4.19256227558065e-05,
+      "loss": 0.0001405619364231825,
+      "step": 23250
+    },
+    {
+      "epoch": 6.0517420696827875,
+      "grad_norm": 0.0021701862569898367,
+      "learning_rate": 4.190233943019152e-05,
+      "loss": 0.00013308810070157052,
+      "step": 23275
+    },
+    {
+      "epoch": 6.058242329693187,
+      "grad_norm": 0.0027431275229901075,
+      "learning_rate": 4.187902906852141e-05,
+      "loss": 0.00014646329917013644,
+      "step": 23300
+    },
+    {
+      "epoch": 6.064742589703588,
+      "grad_norm": 0.0033975490368902683,
+      "learning_rate": 4.185569170808215e-05,
+      "loss": 0.00012525195255875588,
+      "step": 23325
+    },
+    {
+      "epoch": 6.071242849713989,
+      "grad_norm": 0.0028715208172798157,
+      "learning_rate": 4.1832327386202887e-05,
+      "loss": 0.00012716676108539103,
+      "step": 23350
+    },
+    {
+      "epoch": 6.077743109724389,
+      "grad_norm": 0.0012630614219233394,
+      "learning_rate": 4.1808936140255914e-05,
+      "loss": 0.00013919188641011714,
+      "step": 23375
+    },
+    {
+      "epoch": 6.084243369734789,
+      "grad_norm": 0.002235237741842866,
+      "learning_rate": 4.17855180076566e-05,
+      "loss": 0.0001459474954754114,
+      "step": 23400
+    },
+    {
+      "epoch": 6.09074362974519,
+      "grad_norm": 0.0030322587117552757,
+      "learning_rate": 4.176207302586329e-05,
+      "loss": 0.00013226771727204323,
+      "step": 23425
+    },
+    {
+      "epoch": 6.09724388975559,
+      "grad_norm": 0.0033617664594203234,
+      "learning_rate": 4.173860123237729e-05,
+      "loss": 0.00015032600611448289,
+      "step": 23450
+    },
+    {
+      "epoch": 6.103744149765991,
+      "grad_norm": 0.0022996291518211365,
+      "learning_rate": 4.171510266474281e-05,
+      "loss": 0.0001298911590129137,
+      "step": 23475
+    },
+    {
+      "epoch": 6.110244409776391,
+      "grad_norm": 0.00320655875839293,
+      "learning_rate": 4.1691577360546855e-05,
+      "loss": 0.00013250798918306827,
+      "step": 23500
+    },
+    {
+      "epoch": 6.116744669786791,
+      "grad_norm": 0.0028464498464018106,
+      "learning_rate": 4.1668025357419216e-05,
+      "loss": 0.00013406586833298207,
+      "step": 23525
+    },
+    {
+      "epoch": 6.123244929797192,
+      "grad_norm": 0.0014294130960479379,
+      "learning_rate": 4.164444669303238e-05,
+      "loss": 0.00014320195652544498,
+      "step": 23550
+    },
+    {
+      "epoch": 6.1297451898075925,
+      "grad_norm": 0.0019119908101856709,
+      "learning_rate": 4.1620841405101495e-05,
+      "loss": 0.00012864221818745136,
+      "step": 23575
+    },
+    {
+      "epoch": 6.136245449817992,
+      "grad_norm": 0.0031773310620337725,
+      "learning_rate": 4.159720953138428e-05,
+      "loss": 0.0001443338580429554,
+      "step": 23600
+    },
+    {
+      "epoch": 6.142745709828393,
+      "grad_norm": 0.0018979604355990887,
+      "learning_rate": 4.1573551109680986e-05,
+      "loss": 0.00013210905715823173,
+      "step": 23625
+    },
+    {
+      "epoch": 6.149245969838794,
+      "grad_norm": 0.004290254320949316,
+      "learning_rate": 4.154986617783433e-05,
+      "loss": 0.00012187777087092399,
+      "step": 23650
+    },
+    {
+      "epoch": 6.155746229849194,
+      "grad_norm": 0.0014123201835900545,
+      "learning_rate": 4.152615477372941e-05,
+      "loss": 0.00011678831651806831,
+      "step": 23675
+    },
+    {
+      "epoch": 6.162246489859594,
+      "grad_norm": 0.0014078370295464993,
+      "learning_rate": 4.1502416935293714e-05,
+      "loss": 0.00013603556901216506,
+      "step": 23700
+    },
+    {
+      "epoch": 6.168746749869995,
+      "grad_norm": 0.003061910392716527,
+      "learning_rate": 4.147865270049698e-05,
+      "loss": 0.00012292655184864998,
+      "step": 23725
+    },
+    {
+      "epoch": 6.175247009880395,
+      "grad_norm": 0.0016806441126391292,
+      "learning_rate": 4.145486210735116e-05,
+      "loss": 0.0001419335324317217,
+      "step": 23750
+    },
+    {
+      "epoch": 6.181747269890796,
+      "grad_norm": 0.0017229362856596708,
+      "learning_rate": 4.143104519391039e-05,
+      "loss": 0.00012309485115110874,
+      "step": 23775
+    },
+    {
+      "epoch": 6.188247529901196,
+      "grad_norm": 0.0009782557608559728,
+      "learning_rate": 4.1407201998270905e-05,
+      "loss": 0.00011569835245609284,
+      "step": 23800
+    },
+    {
+      "epoch": 6.194747789911596,
+      "grad_norm": 0.0014956933446228504,
+      "learning_rate": 4.1383332558570966e-05,
+      "loss": 0.00012629065662622452,
+      "step": 23825
+    },
+    {
+      "epoch": 6.201248049921997,
+      "grad_norm": 0.0021856180392205715,
+      "learning_rate": 4.135943691299082e-05,
+      "loss": 0.00012633978389203548,
+      "step": 23850
+    },
+    {
+      "epoch": 6.207748309932398,
+      "grad_norm": 0.0016348492354154587,
+      "learning_rate": 4.133551509975264e-05,
+      "loss": 0.00011823375709354877,
+      "step": 23875
+    },
+    {
+      "epoch": 6.214248569942797,
+      "grad_norm": 0.003681580303236842,
+      "learning_rate": 4.131156715712043e-05,
+      "loss": 0.0001365287508815527,
+      "step": 23900
+    },
+    {
+      "epoch": 6.220748829953198,
+      "grad_norm": 0.0019721677526831627,
+      "learning_rate": 4.1287593123400024e-05,
+      "loss": 0.00010654201731085777,
+      "step": 23925
+    },
+    {
+      "epoch": 6.227249089963599,
+      "grad_norm": 0.0012722479877993464,
+      "learning_rate": 4.126359303693896e-05,
+      "loss": 0.0001127240527421236,
+      "step": 23950
+    },
+    {
+      "epoch": 6.233749349973999,
+      "grad_norm": 0.0015793109778314829,
+      "learning_rate": 4.123956693612646e-05,
+      "loss": 0.00012294868007302286,
+      "step": 23975
+    },
+    {
+      "epoch": 6.240249609984399,
+      "grad_norm": 0.0008132465300150216,
+      "learning_rate": 4.1215514859393364e-05,
+      "loss": 0.00010871563106775284,
+      "step": 24000
+    },
+    {
+      "epoch": 6.2467498699948,
+      "grad_norm": 0.0019460894400253892,
+      "learning_rate": 4.119143684521204e-05,
+      "loss": 0.00010960295796394348,
+      "step": 24025
+    },
+    {
+      "epoch": 6.2532501300052,
+      "grad_norm": 0.0021695338655263186,
+      "learning_rate": 4.116733293209638e-05,
+      "loss": 0.00012794556096196175,
+      "step": 24050
+    },
+    {
+      "epoch": 6.259750390015601,
+      "grad_norm": 0.0005786272813566029,
+      "learning_rate": 4.1143203158601665e-05,
+      "loss": 0.00011649612337350845,
+      "step": 24075
+    },
+    {
+      "epoch": 6.266250650026001,
+      "grad_norm": 0.0029142231214791536,
+      "learning_rate": 4.1119047563324566e-05,
+      "loss": 0.00011053699068725109,
+      "step": 24100
+    },
+    {
+      "epoch": 6.272750910036401,
+      "grad_norm": 0.001933974795974791,
+      "learning_rate": 4.109486618490305e-05,
+      "loss": 0.00010253974236547948,
+      "step": 24125
+    },
+    {
+      "epoch": 6.279251170046802,
+      "grad_norm": 0.0012837565736845136,
+      "learning_rate": 4.107065906201631e-05,
+      "loss": 0.00011141752824187279,
+      "step": 24150
+    },
+    {
+      "epoch": 6.279251170046802,
+      "eval_loss": 0.0004750777152366936,
+      "eval_runtime": 7.999,
+      "eval_samples_per_second": 625.079,
+      "eval_steps_per_second": 3.375,
+      "step": 24150
+    },
+    {
+      "epoch": 6.279251170046802,
+      "eval_holdout_avg_strict_f1": 0.08667084089610556,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.3328488372093023,
+      "eval_holdout_nemotron_pii_strict_f1": 0.2043583988028433,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.23449907078898463,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.18167294305603102,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 24150
+    },
+    {
+      "epoch": 6.285751430057203,
+      "grad_norm": 0.0015662575606256723,
+      "learning_rate": 4.104642623338474e-05,
+      "loss": 0.00010832840576767922,
+      "step": 24175
+    },
+    {
+      "epoch": 6.292251690067602,
+      "grad_norm": 0.002608800306916237,
+      "learning_rate": 4.1022167737769856e-05,
+      "loss": 0.00010405355133116245,
+      "step": 24200
+    },
+    {
+      "epoch": 6.298751950078003,
+      "grad_norm": 0.0015294577460736036,
+      "learning_rate": 4.09978836139742e-05,
+      "loss": 0.0001112870778888464,
+      "step": 24225
+    },
+    {
+      "epoch": 6.305252210088404,
+      "grad_norm": 0.0015903846360743046,
+      "learning_rate": 4.097357390084133e-05,
+      "loss": 0.00011478011496365071,
+      "step": 24250
+    },
+    {
+      "epoch": 6.311752470098804,
+      "grad_norm": 0.001592286629602313,
+      "learning_rate": 4.0949238637255726e-05,
+      "loss": 0.00010644377209246159,
+      "step": 24275
+    },
+    {
+      "epoch": 6.318252730109204,
+      "grad_norm": 0.0020193499512970448,
+      "learning_rate": 4.092487786214275e-05,
+      "loss": 0.0001119605079293251,
+      "step": 24300
+    },
+    {
+      "epoch": 6.324752990119605,
+      "grad_norm": 0.0016956543549895287,
+      "learning_rate": 4.0900491614468553e-05,
+      "loss": 0.00010874100960791111,
+      "step": 24325
+    },
+    {
+      "epoch": 6.331253250130005,
+      "grad_norm": 0.0029434727039188147,
+      "learning_rate": 4.087607993324005e-05,
+      "loss": 0.00012604793533682823,
+      "step": 24350
+    },
+    {
+      "epoch": 6.337753510140406,
+      "grad_norm": 0.0025831100065261126,
+      "learning_rate": 4.085164285750481e-05,
+      "loss": 0.00010682960972189903,
+      "step": 24375
+    },
+    {
+      "epoch": 6.344253770150806,
+      "grad_norm": 0.0012894815299659967,
+      "learning_rate": 4.0827180426351066e-05,
+      "loss": 0.00010824393481016158,
+      "step": 24400
+    },
+    {
+      "epoch": 6.350754030161206,
+      "grad_norm": 0.0012475516414269805,
+      "learning_rate": 4.080269267890755e-05,
+      "loss": 0.00010632259771227837,
+      "step": 24425
+    },
+    {
+      "epoch": 6.357254290171607,
+      "grad_norm": 0.0030106082558631897,
+      "learning_rate": 4.077817965434354e-05,
+      "loss": 0.00010344186797738075,
+      "step": 24450
+    },
+    {
+      "epoch": 6.363754550182008,
+      "grad_norm": 0.0018284962279722095,
+      "learning_rate": 4.075364139186872e-05,
+      "loss": 0.00010096374899148941,
+      "step": 24475
+    },
+    {
+      "epoch": 6.3702548101924075,
+      "grad_norm": 0.0012710350565612316,
+      "learning_rate": 4.072907793073316e-05,
+      "loss": 9.118602611124516e-05,
+      "step": 24500
+    },
+    {
+      "epoch": 6.376755070202808,
+      "grad_norm": 0.00146900350227952,
+      "learning_rate": 4.07044893102272e-05,
+      "loss": 0.00011363954283297062,
+      "step": 24525
+    },
+    {
+      "epoch": 6.383255330213209,
+      "grad_norm": 0.002900431863963604,
+      "learning_rate": 4.067987556968147e-05,
+      "loss": 0.00011475393548607827,
+      "step": 24550
+    },
+    {
+      "epoch": 6.389755590223609,
+      "grad_norm": 0.00224672956392169,
+      "learning_rate": 4.0655236748466754e-05,
+      "loss": 0.00011210255324840545,
+      "step": 24575
+    },
+    {
+      "epoch": 6.396255850234009,
+      "grad_norm": 0.002814608858898282,
+      "learning_rate": 4.063057288599395e-05,
+      "loss": 0.00012261774390935897,
+      "step": 24600
+    },
+    {
+      "epoch": 6.40275611024441,
+      "grad_norm": 0.001213836600072682,
+      "learning_rate": 4.0605884021714035e-05,
+      "loss": 0.00010759948752820492,
+      "step": 24625
+    },
+    {
+      "epoch": 6.40925637025481,
+      "grad_norm": 0.0008372975862585008,
+      "learning_rate": 4.0581170195117944e-05,
+      "loss": 0.00011210777796804905,
+      "step": 24650
+    },
+    {
+      "epoch": 6.415756630265211,
+      "grad_norm": 0.0014346669195219874,
+      "learning_rate": 4.055643144573658e-05,
+      "loss": 0.00010934955440461636,
+      "step": 24675
+    },
+    {
+      "epoch": 6.422256890275611,
+      "grad_norm": 0.0009425717871636152,
+      "learning_rate": 4.053166781314068e-05,
+      "loss": 0.00010890507139265537,
+      "step": 24700
+    },
+    {
+      "epoch": 6.428757150286011,
+      "grad_norm": 0.002245101844891906,
+      "learning_rate": 4.050687933694078e-05,
+      "loss": 0.00010580516420304775,
+      "step": 24725
+    },
+    {
+      "epoch": 6.435257410296412,
+      "grad_norm": 0.0010284578893333673,
+      "learning_rate": 4.048206605678719e-05,
+      "loss": 0.00011216386221349239,
+      "step": 24750
+    },
+    {
+      "epoch": 6.441757670306813,
+      "grad_norm": 0.001580017269589007,
+      "learning_rate": 4.045722801236986e-05,
+      "loss": 0.00012316817417740822,
+      "step": 24775
+    },
+    {
+      "epoch": 6.4482579303172125,
+      "grad_norm": 0.0019763698801398277,
+      "learning_rate": 4.043236524341836e-05,
+      "loss": 0.00011748719029128552,
+      "step": 24800
+    },
+    {
+      "epoch": 6.454758190327613,
+      "grad_norm": 0.0005627158097922802,
+      "learning_rate": 4.0407477789701806e-05,
+      "loss": 0.00012304619885981083,
+      "step": 24825
+    },
+    {
+      "epoch": 6.461258450338014,
+      "grad_norm": 0.0016607515281066298,
+      "learning_rate": 4.038256569102883e-05,
+      "loss": 0.00011632999405264855,
+      "step": 24850
+    },
+    {
+      "epoch": 6.467758710348414,
+      "grad_norm": 0.003028704784810543,
+      "learning_rate": 4.035762898724744e-05,
+      "loss": 0.00011094081215560436,
+      "step": 24875
+    },
+    {
+      "epoch": 6.4742589703588145,
+      "grad_norm": 0.0011620358563959599,
+      "learning_rate": 4.0332667718245014e-05,
+      "loss": 0.0001129049714654684,
+      "step": 24900
+    },
+    {
+      "epoch": 6.480759230369214,
+      "grad_norm": 0.001956784166395664,
+      "learning_rate": 4.030768192394824e-05,
+      "loss": 0.0001184759009629488,
+      "step": 24925
+    },
+    {
+      "epoch": 6.487259490379615,
+      "grad_norm": 0.001993892714381218,
+      "learning_rate": 4.028267164432302e-05,
+      "loss": 0.0001015638466924429,
+      "step": 24950
+    },
+    {
+      "epoch": 6.493759750390016,
+      "grad_norm": 0.0014928573509678245,
+      "learning_rate": 4.025763691937443e-05,
+      "loss": 0.00010811746120452881,
+      "step": 24975
+    },
+    {
+      "epoch": 6.500260010400416,
+      "grad_norm": 0.002178949536755681,
+      "learning_rate": 4.023257778914664e-05,
+      "loss": 0.00011847011744976044,
+      "step": 25000
+    },
+    {
+      "epoch": 6.506760270410816,
+      "grad_norm": 0.0012920384760946035,
+      "learning_rate": 4.020749429372286e-05,
+      "loss": 0.00011507784016430378,
+      "step": 25025
+    },
+    {
+      "epoch": 6.513260530421217,
+      "grad_norm": 0.0013235692167654634,
+      "learning_rate": 4.018238647322528e-05,
+      "loss": 0.00010819977149367332,
+      "step": 25050
+    },
+    {
+      "epoch": 6.519760790431617,
+      "grad_norm": 0.0020488635636866093,
+      "learning_rate": 4.0157254367814974e-05,
+      "loss": 0.00011900385841727257,
+      "step": 25075
+    },
+    {
+      "epoch": 6.5262610504420175,
+      "grad_norm": 0.001664237817749381,
+      "learning_rate": 4.013209801769189e-05,
+      "loss": 0.00011807081289589406,
+      "step": 25100
+    },
+    {
+      "epoch": 6.532761310452418,
+      "grad_norm": 0.003178318962454796,
+      "learning_rate": 4.010691746309476e-05,
+      "loss": 0.0001173883117735386,
+      "step": 25125
+    },
+    {
+      "epoch": 6.539261570462818,
+      "grad_norm": 0.0013092603767290711,
+      "learning_rate": 4.0081712744300996e-05,
+      "loss": 0.000108729163184762,
+      "step": 25150
+    },
+    {
+      "epoch": 6.545761830473219,
+      "grad_norm": 0.0006098266458138824,
+      "learning_rate": 4.005648390162669e-05,
+      "loss": 0.00012521205469965936,
+      "step": 25175
+    },
+    {
+      "epoch": 6.5522620904836195,
+      "grad_norm": 0.0013972645392641425,
+      "learning_rate": 4.003123097542654e-05,
+      "loss": 0.000113962572067976,
+      "step": 25200
+    },
+    {
+      "epoch": 6.558762350494019,
+      "grad_norm": 0.001288426574319601,
+      "learning_rate": 4.00059540060937e-05,
+      "loss": 0.00010716897435486316,
+      "step": 25225
+    },
+    {
+      "epoch": 6.56526261050442,
+      "grad_norm": 0.0013647075975313783,
+      "learning_rate": 3.998065303405985e-05,
+      "loss": 0.00012669201008975507,
+      "step": 25250
+    },
+    {
+      "epoch": 6.571762870514821,
+      "grad_norm": 0.0007535081240348518,
+      "learning_rate": 3.995532809979503e-05,
+      "loss": 0.00011134160682559013,
+      "step": 25275
+    },
+    {
+      "epoch": 6.578263130525221,
+      "grad_norm": 0.0020200912840664387,
+      "learning_rate": 3.992997924380763e-05,
+      "loss": 0.00010556045919656753,
+      "step": 25300
+    },
+    {
+      "epoch": 6.584763390535621,
+      "grad_norm": 0.0019679542165249586,
+      "learning_rate": 3.990460650664426e-05,
+      "loss": 0.00011368278414011002,
+      "step": 25325
+    },
+    {
+      "epoch": 6.591263650546022,
+      "grad_norm": 0.003311831271275878,
+      "learning_rate": 3.9879209928889785e-05,
+      "loss": 0.000104788513854146,
+      "step": 25350
+    },
+    {
+      "epoch": 6.597763910556422,
+      "grad_norm": 0.002349372487515211,
+      "learning_rate": 3.985378955116716e-05,
+      "loss": 0.00010274632833898067,
+      "step": 25375
+    },
+    {
+      "epoch": 6.604264170566823,
+      "grad_norm": 0.0023579709231853485,
+      "learning_rate": 3.9828345414137436e-05,
+      "loss": 0.00011161249130964279,
+      "step": 25400
+    },
+    {
+      "epoch": 6.610764430577223,
+      "grad_norm": 0.007348076440393925,
+      "learning_rate": 3.980287755849966e-05,
+      "loss": 0.00012961586937308312,
+      "step": 25425
+    },
+    {
+      "epoch": 6.617264690587623,
+      "grad_norm": 0.0011249575763940811,
+      "learning_rate": 3.9777386024990824e-05,
+      "loss": 0.00011946114711463451,
+      "step": 25450
+    },
+    {
+      "epoch": 6.623764950598024,
+      "grad_norm": 0.0013109352439641953,
+      "learning_rate": 3.9751870854385775e-05,
+      "loss": 0.00012387875467538833,
+      "step": 25475
+    },
+    {
+      "epoch": 6.6302652106084246,
+      "grad_norm": 0.0014741946943104267,
+      "learning_rate": 3.972633208749719e-05,
+      "loss": 0.00011541242711246013,
+      "step": 25500
+    },
+    {
+      "epoch": 6.636765470618824,
+      "grad_norm": 0.0008637275896035135,
+      "learning_rate": 3.970076976517546e-05,
+      "loss": 0.00011342433281242847,
+      "step": 25525
+    },
+    {
+      "epoch": 6.643265730629225,
+      "grad_norm": 0.001253096736036241,
+      "learning_rate": 3.967518392830869e-05,
+      "loss": 0.00011184891685843468,
+      "step": 25550
+    },
+    {
+      "epoch": 6.649765990639626,
+      "grad_norm": 0.0009482137975282967,
+      "learning_rate": 3.964957461782259e-05,
+      "loss": 0.00011849512346088886,
+      "step": 25575
+    },
+    {
+      "epoch": 6.656266250650026,
+      "grad_norm": 0.003740509506314993,
+      "learning_rate": 3.962394187468039e-05,
+      "loss": 0.000108328927308321,
+      "step": 25600
+    },
+    {
+      "epoch": 6.662766510660426,
+      "grad_norm": 0.0027206242084503174,
+      "learning_rate": 3.959828573988283e-05,
+      "loss": 0.0001143067330121994,
+      "step": 25625
+    },
+    {
+      "epoch": 6.669266770670827,
+      "grad_norm": 0.0015034364769235253,
+      "learning_rate": 3.957260625446805e-05,
+      "loss": 0.00011748811230063438,
+      "step": 25650
+    },
+    {
+      "epoch": 6.675767030681227,
+      "grad_norm": 0.0,
+      "learning_rate": 3.954690345951155e-05,
+      "loss": 0.00011198337189853191,
+      "step": 25675
+    },
+    {
+      "epoch": 6.682267290691628,
+      "grad_norm": 0.002175846602767706,
+      "learning_rate": 3.952117739612612e-05,
+      "loss": 0.00011297490447759629,
+      "step": 25700
+    },
+    {
+      "epoch": 6.688767550702028,
+      "grad_norm": 0.0038169934414327145,
+      "learning_rate": 3.9495428105461746e-05,
+      "loss": 0.00010661475360393524,
+      "step": 25725
+    },
+    {
+      "epoch": 6.695267810712428,
+      "grad_norm": 0.0022700561676174402,
+      "learning_rate": 3.9469655628705587e-05,
+      "loss": 0.00012003203853964805,
+      "step": 25750
+    },
+    {
+      "epoch": 6.701768070722829,
+      "grad_norm": 0.0009079027222469449,
+      "learning_rate": 3.944386000708189e-05,
+      "loss": 0.0001005709357559681,
+      "step": 25775
+    },
+    {
+      "epoch": 6.70826833073323,
+      "grad_norm": 0.0029594795778393745,
+      "learning_rate": 3.9418041281851894e-05,
+      "loss": 0.00010305088944733143,
+      "step": 25800
+    },
+    {
+      "epoch": 6.714768590743629,
+      "grad_norm": 0.0022919829934835434,
+      "learning_rate": 3.939219949431384e-05,
+      "loss": 0.00011034698225557804,
+      "step": 25825
+    },
+    {
+      "epoch": 6.72126885075403,
+      "grad_norm": 0.0008221949101425707,
+      "learning_rate": 3.9366334685802824e-05,
+      "loss": 0.00010054687038064003,
+      "step": 25850
+    },
+    {
+      "epoch": 6.727769110764431,
+      "grad_norm": 0.001326465979218483,
+      "learning_rate": 3.9340446897690775e-05,
+      "loss": 0.00011152955703437328,
+      "step": 25875
+    },
+    {
+      "epoch": 6.734269370774831,
+      "grad_norm": 0.002211722545325756,
+      "learning_rate": 3.9314536171386386e-05,
+      "loss": 0.00012038430199027062,
+      "step": 25900
+    },
+    {
+      "epoch": 6.740769630785231,
+      "grad_norm": 0.003697257721796632,
+      "learning_rate": 3.9288602548335026e-05,
+      "loss": 0.0001094141136854887,
+      "step": 25925
+    },
+    {
+      "epoch": 6.747269890795632,
+      "grad_norm": 0.0013182315742596984,
+      "learning_rate": 3.9262646070018696e-05,
+      "loss": 0.0001205064170062542,
+      "step": 25950
+    },
+    {
+      "epoch": 6.753770150806032,
+      "grad_norm": 0.0024832915514707565,
+      "learning_rate": 3.923666677795596e-05,
+      "loss": 0.00012537592090666294,
+      "step": 25975
+    },
+    {
+      "epoch": 6.760270410816433,
+      "grad_norm": 0.0033784189727157354,
+      "learning_rate": 3.921066471370186e-05,
+      "loss": 0.00012734243646264076,
+      "step": 26000
+    },
+    {
+      "epoch": 6.766770670826833,
+      "grad_norm": 0.0017894685734063387,
+      "learning_rate": 3.918463991884789e-05,
+      "loss": 0.00011924908496439457,
+      "step": 26025
+    },
+    {
+      "epoch": 6.773270930837233,
+      "grad_norm": 0.0009700027876533568,
+      "learning_rate": 3.9158592435021864e-05,
+      "loss": 0.00010192785412073135,
+      "step": 26050
+    },
+    {
+      "epoch": 6.779771190847634,
+      "grad_norm": 0.001378193381242454,
+      "learning_rate": 3.913252230388792e-05,
+      "loss": 0.00012017076835036278,
+      "step": 26075
+    },
+    {
+      "epoch": 6.786271450858035,
+      "grad_norm": 0.0031117494218051434,
+      "learning_rate": 3.91064295671464e-05,
+      "loss": 0.00010474678128957748,
+      "step": 26100
+    },
+    {
+      "epoch": 6.7927717108684345,
+      "grad_norm": 0.0033372053876519203,
+      "learning_rate": 3.908031426653383e-05,
+      "loss": 0.00012944124639034272,
+      "step": 26125
+    },
+    {
+      "epoch": 6.799271970878835,
+      "grad_norm": 0.0015742985997349024,
+      "learning_rate": 3.90541764438228e-05,
+      "loss": 0.00012753351591527463,
+      "step": 26150
+    },
+    {
+      "epoch": 6.805772230889236,
+      "grad_norm": 0.0012687500566244125,
+      "learning_rate": 3.902801614082194e-05,
+      "loss": 0.00010754807852208614,
+      "step": 26175
+    },
+    {
+      "epoch": 6.812272490899636,
+      "grad_norm": 0.0014284363714978099,
+      "learning_rate": 3.9001833399375845e-05,
+      "loss": 0.00011043894104659558,
+      "step": 26200
+    },
+    {
+      "epoch": 6.818772750910036,
+      "grad_norm": 0.0027165126521140337,
+      "learning_rate": 3.8975628261364975e-05,
+      "loss": 0.00011792697943747044,
+      "step": 26225
+    },
+    {
+      "epoch": 6.825273010920437,
+      "grad_norm": 0.0013110070722177625,
+      "learning_rate": 3.8949400768705646e-05,
+      "loss": 0.00012383973225951194,
+      "step": 26250
+    },
+    {
+      "epoch": 6.831773270930837,
+      "grad_norm": 0.002058715559542179,
+      "learning_rate": 3.892315096334992e-05,
+      "loss": 0.00010601886548101902,
+      "step": 26275
+    },
+    {
+      "epoch": 6.838273530941238,
+      "grad_norm": 0.002516526263207197,
+      "learning_rate": 3.889687888728553e-05,
+      "loss": 0.00011323167942464351,
+      "step": 26300
+    },
+    {
+      "epoch": 6.844773790951638,
+      "grad_norm": 0.001380758942104876,
+      "learning_rate": 3.8870584582535854e-05,
+      "loss": 0.00011060774326324463,
+      "step": 26325
+    },
+    {
+      "epoch": 6.851274050962038,
+      "grad_norm": 0.001599041628651321,
+      "learning_rate": 3.884426809115982e-05,
+      "loss": 0.00011652723886072635,
+      "step": 26350
+    },
+    {
+      "epoch": 6.857774310972439,
+      "grad_norm": 0.0015828153118491173,
+      "learning_rate": 3.8817929455251856e-05,
+      "loss": 0.00011761505156755447,
+      "step": 26375
+    },
+    {
+      "epoch": 6.86427457098284,
+      "grad_norm": 0.002911556512117386,
+      "learning_rate": 3.879156871694179e-05,
+      "loss": 0.00012177078053355217,
+      "step": 26400
+    },
+    {
+      "epoch": 6.8707748309932395,
+      "grad_norm": 0.0010526723926886916,
+      "learning_rate": 3.87651859183948e-05,
+      "loss": 0.0001139861810952425,
+      "step": 26425
+    },
+    {
+      "epoch": 6.87727509100364,
+      "grad_norm": 0.0009642353979870677,
+      "learning_rate": 3.8738781101811386e-05,
+      "loss": 0.00010670073330402374,
+      "step": 26450
+    },
+    {
+      "epoch": 6.883775351014041,
+      "grad_norm": 0.004540348891168833,
+      "learning_rate": 3.8712354309427234e-05,
+      "loss": 0.00012120610103011131,
+      "step": 26475
+    },
+    {
+      "epoch": 6.890275611024441,
+      "grad_norm": 0.0038952853064984083,
+      "learning_rate": 3.8685905583513186e-05,
+      "loss": 0.00011744635179638862,
+      "step": 26500
+    },
+    {
+      "epoch": 6.8967758710348415,
+      "grad_norm": 0.001943692215718329,
+      "learning_rate": 3.8659434966375184e-05,
+      "loss": 0.00011857620440423488,
+      "step": 26525
+    },
+    {
+      "epoch": 6.903276131045242,
+      "grad_norm": 0.004994726274162531,
+      "learning_rate": 3.8632942500354175e-05,
+      "loss": 0.00011114344000816345,
+      "step": 26550
+    },
+    {
+      "epoch": 6.909776391055642,
+      "grad_norm": 0.0013535808539018035,
+      "learning_rate": 3.8606428227826045e-05,
+      "loss": 0.00011538634076714516,
+      "step": 26575
+    },
+    {
+      "epoch": 6.916276651066043,
+      "grad_norm": 0.001590519561432302,
+      "learning_rate": 3.8579892191201585e-05,
+      "loss": 0.00014122212305665017,
+      "step": 26600
+    },
+    {
+      "epoch": 6.922776911076443,
+      "grad_norm": 0.0024080509319901466,
+      "learning_rate": 3.855333443292637e-05,
+      "loss": 0.00014358939602971077,
+      "step": 26625
+    },
+    {
+      "epoch": 6.929277171086843,
+      "grad_norm": 0.001678533386439085,
+      "learning_rate": 3.852675499548073e-05,
+      "loss": 0.00012503803707659245,
+      "step": 26650
+    },
+    {
+      "epoch": 6.935777431097244,
+      "grad_norm": 0.0018341480754315853,
+      "learning_rate": 3.850015392137969e-05,
+      "loss": 9.755505248904228e-05,
+      "step": 26675
+    },
+    {
+      "epoch": 6.942277691107645,
+      "grad_norm": 0.0014395725447684526,
+      "learning_rate": 3.847353125317287e-05,
+      "loss": 0.00011185365729033947,
+      "step": 26700
+    },
+    {
+      "epoch": 6.9487779511180445,
+      "grad_norm": 0.001719315187074244,
+      "learning_rate": 3.844688703344441e-05,
+      "loss": 0.0001119834091514349,
+      "step": 26725
+    },
+    {
+      "epoch": 6.955278211128445,
+      "grad_norm": 0.002458140254020691,
+      "learning_rate": 3.8420221304812964e-05,
+      "loss": 0.00011097895912826061,
+      "step": 26750
+    },
+    {
+      "epoch": 6.961778471138846,
+      "grad_norm": 0.0010102568194270134,
+      "learning_rate": 3.839353410993156e-05,
+      "loss": 0.00012165975756943226,
+      "step": 26775
+    },
+    {
+      "epoch": 6.968278731149246,
+      "grad_norm": 0.0011886069551110268,
+      "learning_rate": 3.8366825491487576e-05,
+      "loss": 0.00011145750992000103,
+      "step": 26800
+    },
+    {
+      "epoch": 6.9747789911596465,
+      "grad_norm": 0.0021879642736166716,
+      "learning_rate": 3.834009549220264e-05,
+      "loss": 0.00011545484885573387,
+      "step": 26825
+    },
+    {
+      "epoch": 6.981279251170047,
+      "grad_norm": 0.0031086443923413754,
+      "learning_rate": 3.8313344154832607e-05,
+      "loss": 0.00011009662412106991,
+      "step": 26850
+    },
+    {
+      "epoch": 6.987779511180447,
+      "grad_norm": 0.0015444220043718815,
+      "learning_rate": 3.8286571522167434e-05,
+      "loss": 0.00010692393407225609,
+      "step": 26875
+    },
+    {
+      "epoch": 6.994279771190848,
+      "grad_norm": 0.002990622306242585,
+      "learning_rate": 3.825977763703118e-05,
+      "loss": 0.00011025475338101387,
+      "step": 26900
+    },
+    {
+      "epoch": 7.0007800312012485,
+      "grad_norm": 0.002712846966460347,
+      "learning_rate": 3.8232962542281844e-05,
+      "loss": 0.00011795877479016781,
+      "step": 26925
+    },
+    {
+      "epoch": 7.007280291211648,
+      "grad_norm": 0.004657026380300522,
+      "learning_rate": 3.8206126280811417e-05,
+      "loss": 0.00010386927053332329,
+      "step": 26950
+    },
+    {
+      "epoch": 7.013780551222049,
+      "grad_norm": 0.0009838445112109184,
+      "learning_rate": 3.817926889554567e-05,
+      "loss": 0.0001006865594536066,
+      "step": 26975
+    },
+    {
+      "epoch": 7.02028081123245,
+      "grad_norm": 0.000906344095710665,
+      "learning_rate": 3.815239042944424e-05,
+      "loss": 9.951076470315456e-05,
+      "step": 27000
+    },
+    {
+      "epoch": 7.02678107124285,
+      "grad_norm": 0.0009494044352322817,
+      "learning_rate": 3.8125490925500425e-05,
+      "loss": 0.00011620162054896354,
+      "step": 27025
+    },
+    {
+      "epoch": 7.03328133125325,
+      "grad_norm": 0.0010471625719219446,
+      "learning_rate": 3.8098570426741195e-05,
+      "loss": 0.00011458195745944977,
+      "step": 27050
+    },
+    {
+      "epoch": 7.03978159126365,
+      "grad_norm": 0.0015998894814401865,
+      "learning_rate": 3.807162897622712e-05,
+      "loss": 0.00011363034136593341,
+      "step": 27075
+    },
+    {
+      "epoch": 7.046281851274051,
+      "grad_norm": 0.0013887033564969897,
+      "learning_rate": 3.804466661705226e-05,
+      "loss": 0.00012030398473143577,
+      "step": 27100
+    },
+    {
+      "epoch": 7.0527821112844515,
+      "grad_norm": 0.0010439151665195823,
+      "learning_rate": 3.8017683392344117e-05,
+      "loss": 9.753520600497723e-05,
+      "step": 27125
+    },
+    {
+      "epoch": 7.059282371294851,
+      "grad_norm": 0.0016576464986428618,
+      "learning_rate": 3.799067934526358e-05,
+      "loss": 0.00012017428874969483,
+      "step": 27150
+    },
+    {
+      "epoch": 7.065782631305252,
+      "grad_norm": 0.0014531930210068822,
+      "learning_rate": 3.7963654519004854e-05,
+      "loss": 0.00010396531783044339,
+      "step": 27175
+    },
+    {
+      "epoch": 7.072282891315653,
+      "grad_norm": 0.0022538285702466965,
+      "learning_rate": 3.793660895679536e-05,
+      "loss": 0.00011012404225766658,
+      "step": 27200
+    },
+    {
+      "epoch": 7.078783151326053,
+      "grad_norm": 0.000809481309261173,
+      "learning_rate": 3.79095427018957e-05,
+      "loss": 0.00010953256860375404,
+      "step": 27225
+    },
+    {
+      "epoch": 7.085283411336453,
+      "grad_norm": 0.0014863279648125172,
+      "learning_rate": 3.7882455797599554e-05,
+      "loss": 0.0001140495203435421,
+      "step": 27250
+    },
+    {
+      "epoch": 7.091783671346854,
+      "grad_norm": 0.0011069235624745488,
+      "learning_rate": 3.7855348287233664e-05,
+      "loss": 0.0001098285149782896,
+      "step": 27275
+    },
+    {
+      "epoch": 7.098283931357254,
+      "grad_norm": 0.0024071892257779837,
+      "learning_rate": 3.782822021415772e-05,
+      "loss": 0.00011785288341343404,
+      "step": 27300
+    },
+    {
+      "epoch": 7.104784191367655,
+      "grad_norm": 0.0011146635515615344,
+      "learning_rate": 3.780107162176429e-05,
+      "loss": 0.00011987052857875824,
+      "step": 27325
+    },
+    {
+      "epoch": 7.111284451378055,
+      "grad_norm": 0.0013401745818555355,
+      "learning_rate": 3.777390255347878e-05,
+      "loss": 0.00011973907239735127,
+      "step": 27350
+    },
+    {
+      "epoch": 7.117784711388455,
+      "grad_norm": 0.001196415163576603,
+      "learning_rate": 3.774671305275934e-05,
+      "loss": 0.000108198132365942,
+      "step": 27375
+    },
+    {
+      "epoch": 7.124284971398856,
+      "grad_norm": 0.0019441972253844142,
+      "learning_rate": 3.77195031630968e-05,
+      "loss": 0.0001106107421219349,
+      "step": 27400
+    },
+    {
+      "epoch": 7.130785231409257,
+      "grad_norm": 0.0013329897774383426,
+      "learning_rate": 3.769227292801462e-05,
+      "loss": 0.00012577936984598637,
+      "step": 27425
+    },
+    {
+      "epoch": 7.137285491419656,
+      "grad_norm": 0.0025628148578107357,
+      "learning_rate": 3.766502239106877e-05,
+      "loss": 0.00012169907800853252,
+      "step": 27450
+    },
+    {
+      "epoch": 7.143785751430057,
+      "grad_norm": 0.001455930876545608,
+      "learning_rate": 3.763775159584773e-05,
+      "loss": 0.00011937427334487438,
+      "step": 27475
+    },
+    {
+      "epoch": 7.150286011440458,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7610460585972364e-05,
+      "loss": 0.00010603815317153931,
+      "step": 27500
+    },
+    {
+      "epoch": 7.156786271450858,
+      "grad_norm": 0.0016526292311027646,
+      "learning_rate": 3.7583149405095875e-05,
+      "loss": 0.00011348278261721135,
+      "step": 27525
+    },
+    {
+      "epoch": 7.163286531461258,
+      "grad_norm": 0.0009249432478100061,
+      "learning_rate": 3.755581809690373e-05,
+      "loss": 0.0001146230474114418,
+      "step": 27550
+    },
+    {
+      "epoch": 7.169786791471659,
+      "grad_norm": 0.0015359838725998998,
+      "learning_rate": 3.7528466705113585e-05,
+      "loss": 0.00012477039359509945,
+      "step": 27575
+    },
+    {
+      "epoch": 7.176287051482059,
+      "grad_norm": 0.0014865921111777425,
+      "learning_rate": 3.750109527347524e-05,
+      "loss": 0.00011578558012843133,
+      "step": 27600
+    },
+    {
+      "epoch": 7.176287051482059,
+      "eval_loss": 0.0004892141441814601,
+      "eval_runtime": 7.9271,
+      "eval_samples_per_second": 630.747,
+      "eval_steps_per_second": 3.406,
+      "step": 27600
+    },
+    {
+      "epoch": 7.176287051482059,
+      "eval_holdout_avg_strict_f1": 0.09704957518230674,
+      "eval_holdout_enron_spam_bvk_strict_f1": 0.0,
+      "eval_holdout_enron_spam_setfit_strict_f1": 0.0,
+      "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.36040729339332234,
+      "eval_holdout_nemotron_pii_strict_f1": 0.24904813043053792,
+      "eval_holdout_open_pii_masking_500k_strict_f1": 0.25095188646590516,
+      "eval_holdout_phishing_darkknight_strict_f1": 0.0,
+      "eval_holdout_phishing_zefang_strict_f1": 0.0,
+      "eval_holdout_pii_masking_400k_strict_f1": 0.20713801671560875,
+      "eval_holdout_sms_spam_multilingual_strict_f1": 0.0,
+      "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0,
+      "eval_holdout_spamassassin_strict_f1": 0.0,
+      "step": 27600
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 69000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 18,
+  "save_steps": 3450,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 256,
+  "trial_name": null,
+  "trial_params": null
+}