diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9051 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 6438,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004659832246039142,
+      "grad_norm": 35.50701112538932,
+      "learning_rate": 3.8819875776397516e-07,
+      "loss": 4.8954,
+      "step": 5
+    },
+    {
+      "epoch": 0.009319664492078284,
+      "grad_norm": 36.069560254366,
+      "learning_rate": 7.763975155279503e-07,
+      "loss": 4.8747,
+      "step": 10
+    },
+    {
+      "epoch": 0.013979496738117428,
+      "grad_norm": 35.774397237004045,
+      "learning_rate": 1.1645962732919254e-06,
+      "loss": 4.8001,
+      "step": 15
+    },
+    {
+      "epoch": 0.01863932898415657,
+      "grad_norm": 19.47714488611424,
+      "learning_rate": 1.5527950310559006e-06,
+      "loss": 4.4763,
+      "step": 20
+    },
+    {
+      "epoch": 0.023299161230195712,
+      "grad_norm": 8.063023854835496,
+      "learning_rate": 1.940993788819876e-06,
+      "loss": 4.0988,
+      "step": 25
+    },
+    {
+      "epoch": 0.027958993476234855,
+      "grad_norm": 7.57117336682603,
+      "learning_rate": 2.329192546583851e-06,
+      "loss": 3.8527,
+      "step": 30
+    },
+    {
+      "epoch": 0.032618825722273995,
+      "grad_norm": 8.627958190031924,
+      "learning_rate": 2.7173913043478263e-06,
+      "loss": 3.5734,
+      "step": 35
+    },
+    {
+      "epoch": 0.03727865796831314,
+      "grad_norm": 4.250772744712094,
+      "learning_rate": 3.1055900621118013e-06,
+      "loss": 3.468,
+      "step": 40
+    },
+    {
+      "epoch": 0.04193849021435228,
+      "grad_norm": 4.247065348246596,
+      "learning_rate": 3.4937888198757763e-06,
+      "loss": 3.199,
+      "step": 45
+    },
+    {
+      "epoch": 0.046598322460391424,
+      "grad_norm": 3.134537193905115,
+      "learning_rate": 3.881987577639752e-06,
+      "loss": 2.9552,
+      "step": 50
+    },
+    {
+      "epoch": 0.05125815470643057,
+      "grad_norm": 3.160581324791657,
+      "learning_rate": 4.270186335403727e-06,
+      "loss": 2.6003,
+      "step": 55
+    },
+    {
+      "epoch": 0.05591798695246971,
+      "grad_norm": 2.3105843738565084,
+      "learning_rate": 4.658385093167702e-06,
+      "loss": 2.2491,
+      "step": 60
+    },
+    {
+      "epoch": 0.06057781919850885,
+      "grad_norm": 2.3001534631052896,
+      "learning_rate": 5.046583850931677e-06,
+      "loss": 1.9459,
+      "step": 65
+    },
+    {
+      "epoch": 0.06523765144454799,
+      "grad_norm": 1.7095476007689887,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 1.5873,
+      "step": 70
+    },
+    {
+      "epoch": 0.06989748369058714,
+      "grad_norm": 1.3742305794978262,
+      "learning_rate": 5.8229813664596275e-06,
+      "loss": 1.346,
+      "step": 75
+    },
+    {
+      "epoch": 0.07455731593662628,
+      "grad_norm": 1.1391591725551033,
+      "learning_rate": 6.2111801242236025e-06,
+      "loss": 1.1771,
+      "step": 80
+    },
+    {
+      "epoch": 0.07921714818266543,
+      "grad_norm": 0.8422530514111145,
+      "learning_rate": 6.5993788819875775e-06,
+      "loss": 1.0568,
+      "step": 85
+    },
+    {
+      "epoch": 0.08387698042870456,
+      "grad_norm": 0.8626061215392534,
+      "learning_rate": 6.9875776397515525e-06,
+      "loss": 0.9826,
+      "step": 90
+    },
+    {
+      "epoch": 0.08853681267474371,
+      "grad_norm": 1.1742928188519286,
+      "learning_rate": 7.375776397515528e-06,
+      "loss": 0.8945,
+      "step": 95
+    },
+    {
+      "epoch": 0.09319664492078285,
+      "grad_norm": 0.6100653556742394,
+      "learning_rate": 7.763975155279503e-06,
+      "loss": 0.8261,
+      "step": 100
+    },
+    {
+      "epoch": 0.097856477166822,
+      "grad_norm": 0.6697063598231574,
+      "learning_rate": 8.15217391304348e-06,
+      "loss": 0.7856,
+      "step": 105
+    },
+    {
+      "epoch": 0.10251630941286113,
+      "grad_norm": 0.6186819616185586,
+      "learning_rate": 8.540372670807453e-06,
+      "loss": 0.7616,
+      "step": 110
+    },
+    {
+      "epoch": 0.10717614165890028,
+      "grad_norm": 0.5415166228535374,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.7381,
+      "step": 115
+    },
+    {
+      "epoch": 0.11183597390493942,
+      "grad_norm": 0.4748784369597874,
+      "learning_rate": 9.316770186335403e-06,
+      "loss": 0.7016,
+      "step": 120
+    },
+    {
+      "epoch": 0.11649580615097857,
+      "grad_norm": 0.4408664459103081,
+      "learning_rate": 9.70496894409938e-06,
+      "loss": 0.684,
+      "step": 125
+    },
+    {
+      "epoch": 0.1211556383970177,
+      "grad_norm": 0.4886387253109222,
+      "learning_rate": 1.0093167701863353e-05,
+      "loss": 0.6754,
+      "step": 130
+    },
+    {
+      "epoch": 0.12581547064305684,
+      "grad_norm": 0.5075757149274733,
+      "learning_rate": 1.048136645962733e-05,
+      "loss": 0.6614,
+      "step": 135
+    },
+    {
+      "epoch": 0.13047530288909598,
+      "grad_norm": 0.445441334990142,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.6316,
+      "step": 140
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 0.39921865575468074,
+      "learning_rate": 1.1257763975155281e-05,
+      "loss": 0.6527,
+      "step": 145
+    },
+    {
+      "epoch": 0.13979496738117428,
+      "grad_norm": 0.4456719280991211,
+      "learning_rate": 1.1645962732919255e-05,
+      "loss": 0.6158,
+      "step": 150
+    },
+    {
+      "epoch": 0.14445479962721341,
+      "grad_norm": 0.43647970354094884,
+      "learning_rate": 1.2034161490683231e-05,
+      "loss": 0.6161,
+      "step": 155
+    },
+    {
+      "epoch": 0.14911463187325255,
+      "grad_norm": 0.37334760287232616,
+      "learning_rate": 1.2422360248447205e-05,
+      "loss": 0.5887,
+      "step": 160
+    },
+    {
+      "epoch": 0.15377446411929171,
+      "grad_norm": 0.37820422077569116,
+      "learning_rate": 1.2810559006211181e-05,
+      "loss": 0.5949,
+      "step": 165
+    },
+    {
+      "epoch": 0.15843429636533085,
+      "grad_norm": 0.3956164887876263,
+      "learning_rate": 1.3198757763975155e-05,
+      "loss": 0.5672,
+      "step": 170
+    },
+    {
+      "epoch": 0.16309412861137,
+      "grad_norm": 0.4125116008657431,
+      "learning_rate": 1.3586956521739131e-05,
+      "loss": 0.5748,
+      "step": 175
+    },
+    {
+      "epoch": 0.16775396085740912,
+      "grad_norm": 0.4042523084597145,
+      "learning_rate": 1.3975155279503105e-05,
+      "loss": 0.5618,
+      "step": 180
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 0.3968428244208969,
+      "learning_rate": 1.4363354037267083e-05,
+      "loss": 0.5613,
+      "step": 185
+    },
+    {
+      "epoch": 0.17707362534948742,
+      "grad_norm": 0.42186563467630556,
+      "learning_rate": 1.4751552795031057e-05,
+      "loss": 0.5689,
+      "step": 190
+    },
+    {
+      "epoch": 0.18173345759552656,
+      "grad_norm": 0.40374622437171653,
+      "learning_rate": 1.5139751552795031e-05,
+      "loss": 0.5547,
+      "step": 195
+    },
+    {
+      "epoch": 0.1863932898415657,
+      "grad_norm": 0.36118013072907545,
+      "learning_rate": 1.5527950310559007e-05,
+      "loss": 0.5518,
+      "step": 200
+    },
+    {
+      "epoch": 0.19105312208760486,
+      "grad_norm": 0.40313235870073283,
+      "learning_rate": 1.5916149068322984e-05,
+      "loss": 0.5374,
+      "step": 205
+    },
+    {
+      "epoch": 0.195712954333644,
+      "grad_norm": 0.4599929793107047,
+      "learning_rate": 1.630434782608696e-05,
+      "loss": 0.5501,
+      "step": 210
+    },
+    {
+      "epoch": 0.20037278657968313,
+      "grad_norm": 0.3865610634297696,
+      "learning_rate": 1.6692546583850933e-05,
+      "loss": 0.5287,
+      "step": 215
+    },
+    {
+      "epoch": 0.20503261882572227,
+      "grad_norm": 0.7294272539970464,
+      "learning_rate": 1.7080745341614907e-05,
+      "loss": 0.533,
+      "step": 220
+    },
+    {
+      "epoch": 0.2096924510717614,
+      "grad_norm": 0.5569825730733867,
+      "learning_rate": 1.7468944099378884e-05,
+      "loss": 0.5127,
+      "step": 225
+    },
+    {
+      "epoch": 0.21435228331780057,
+      "grad_norm": 0.39174687817823234,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.5283,
+      "step": 230
+    },
+    {
+      "epoch": 0.2190121155638397,
+      "grad_norm": 0.341241624060807,
+      "learning_rate": 1.8245341614906833e-05,
+      "loss": 0.5187,
+      "step": 235
+    },
+    {
+      "epoch": 0.22367194780987884,
+      "grad_norm": 0.35015493809257703,
+      "learning_rate": 1.8633540372670807e-05,
+      "loss": 0.5169,
+      "step": 240
+    },
+    {
+      "epoch": 0.22833178005591798,
+      "grad_norm": 0.33867637901607595,
+      "learning_rate": 1.9021739130434784e-05,
+      "loss": 0.5198,
+      "step": 245
+    },
+    {
+      "epoch": 0.23299161230195714,
+      "grad_norm": 0.3658576170334555,
+      "learning_rate": 1.940993788819876e-05,
+      "loss": 0.5086,
+      "step": 250
+    },
+    {
+      "epoch": 0.23765144454799628,
+      "grad_norm": 0.40481958813279273,
+      "learning_rate": 1.9798136645962733e-05,
+      "loss": 0.5171,
+      "step": 255
+    },
+    {
+      "epoch": 0.2423112767940354,
+      "grad_norm": 0.38666223513358283,
+      "learning_rate": 2.0186335403726707e-05,
+      "loss": 0.5094,
+      "step": 260
+    },
+    {
+      "epoch": 0.24697110904007455,
+      "grad_norm": 0.4189928929883536,
+      "learning_rate": 2.0574534161490684e-05,
+      "loss": 0.5067,
+      "step": 265
+    },
+    {
+      "epoch": 0.2516309412861137,
+      "grad_norm": 0.3786446496366556,
+      "learning_rate": 2.096273291925466e-05,
+      "loss": 0.5044,
+      "step": 270
+    },
+    {
+      "epoch": 0.25629077353215285,
+      "grad_norm": 0.41635025378590107,
+      "learning_rate": 2.1350931677018636e-05,
+      "loss": 0.507,
+      "step": 275
+    },
+    {
+      "epoch": 0.26095060577819196,
+      "grad_norm": 0.3520883653479603,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 0.499,
+      "step": 280
+    },
+    {
+      "epoch": 0.2656104380242311,
+      "grad_norm": 0.5125339616308159,
+      "learning_rate": 2.2127329192546584e-05,
+      "loss": 0.5006,
+      "step": 285
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.48299702812351186,
+      "learning_rate": 2.2515527950310562e-05,
+      "loss": 0.5047,
+      "step": 290
+    },
+    {
+      "epoch": 0.2749301025163094,
+      "grad_norm": 0.5096289990172574,
+      "learning_rate": 2.2903726708074536e-05,
+      "loss": 0.506,
+      "step": 295
+    },
+    {
+      "epoch": 0.27958993476234856,
+      "grad_norm": 0.6040482063049619,
+      "learning_rate": 2.329192546583851e-05,
+      "loss": 0.5051,
+      "step": 300
+    },
+    {
+      "epoch": 0.2842497670083877,
+      "grad_norm": 0.38310144316154465,
+      "learning_rate": 2.3680124223602484e-05,
+      "loss": 0.4933,
+      "step": 305
+    },
+    {
+      "epoch": 0.28890959925442683,
+      "grad_norm": 0.45394844679336827,
+      "learning_rate": 2.4068322981366462e-05,
+      "loss": 0.4856,
+      "step": 310
+    },
+    {
+      "epoch": 0.293569431500466,
+      "grad_norm": 0.4152400236577921,
+      "learning_rate": 2.4456521739130436e-05,
+      "loss": 0.4922,
+      "step": 315
+    },
+    {
+      "epoch": 0.2982292637465051,
+      "grad_norm": 0.34695000251329294,
+      "learning_rate": 2.484472049689441e-05,
+      "loss": 0.476,
+      "step": 320
+    },
+    {
+      "epoch": 0.30288909599254427,
+      "grad_norm": 0.3376273705719003,
+      "learning_rate": 2.5232919254658388e-05,
+      "loss": 0.4738,
+      "step": 325
+    },
+    {
+      "epoch": 0.30754892823858343,
+      "grad_norm": 0.33320562433694106,
+      "learning_rate": 2.5621118012422362e-05,
+      "loss": 0.4835,
+      "step": 330
+    },
+    {
+      "epoch": 0.31220876048462254,
+      "grad_norm": 4.662940474437422,
+      "learning_rate": 2.6009316770186336e-05,
+      "loss": 0.4984,
+      "step": 335
+    },
+    {
+      "epoch": 0.3168685927306617,
+      "grad_norm": 0.37667182943636984,
+      "learning_rate": 2.639751552795031e-05,
+      "loss": 0.4864,
+      "step": 340
+    },
+    {
+      "epoch": 0.32152842497670087,
+      "grad_norm": 0.40944974993031563,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.4796,
+      "step": 345
+    },
+    {
+      "epoch": 0.32618825722274,
+      "grad_norm": 0.41403227124386566,
+      "learning_rate": 2.7173913043478262e-05,
+      "loss": 0.4922,
+      "step": 350
+    },
+    {
+      "epoch": 0.33084808946877914,
+      "grad_norm": 0.42503410465754166,
+      "learning_rate": 2.7562111801242236e-05,
+      "loss": 0.4806,
+      "step": 355
+    },
+    {
+      "epoch": 0.33550792171481825,
+      "grad_norm": 0.46405536114113854,
+      "learning_rate": 2.795031055900621e-05,
+      "loss": 0.4743,
+      "step": 360
+    },
+    {
+      "epoch": 0.3401677539608574,
+      "grad_norm": 0.42178448744386976,
+      "learning_rate": 2.833850931677019e-05,
+      "loss": 0.4712,
+      "step": 365
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 0.41157240463837597,
+      "learning_rate": 2.8726708074534165e-05,
+      "loss": 0.4834,
+      "step": 370
+    },
+    {
+      "epoch": 0.3494874184529357,
+      "grad_norm": 0.33825470690326404,
+      "learning_rate": 2.911490683229814e-05,
+      "loss": 0.4675,
+      "step": 375
+    },
+    {
+      "epoch": 0.35414725069897485,
+      "grad_norm": 0.4076222579327731,
+      "learning_rate": 2.9503105590062114e-05,
+      "loss": 0.4821,
+      "step": 380
+    },
+    {
+      "epoch": 0.35880708294501396,
+      "grad_norm": 0.3715996726621956,
+      "learning_rate": 2.9891304347826088e-05,
+      "loss": 0.4841,
+      "step": 385
+    },
+    {
+      "epoch": 0.3634669151910531,
+      "grad_norm": 0.37466284025324836,
+      "learning_rate": 3.0279503105590062e-05,
+      "loss": 0.4682,
+      "step": 390
+    },
+    {
+      "epoch": 0.3681267474370923,
+      "grad_norm": 0.3758484725966801,
+      "learning_rate": 3.066770186335404e-05,
+      "loss": 0.4735,
+      "step": 395
+    },
+    {
+      "epoch": 0.3727865796831314,
+      "grad_norm": 0.3671638770621295,
+      "learning_rate": 3.1055900621118014e-05,
+      "loss": 0.4598,
+      "step": 400
+    },
+    {
+      "epoch": 0.37744641192917056,
+      "grad_norm": 0.3648473442855245,
+      "learning_rate": 3.1444099378881995e-05,
+      "loss": 0.4774,
+      "step": 405
+    },
+    {
+      "epoch": 0.3821062441752097,
+      "grad_norm": 0.3490647842689985,
+      "learning_rate": 3.183229813664597e-05,
+      "loss": 0.4725,
+      "step": 410
+    },
+    {
+      "epoch": 0.38676607642124883,
+      "grad_norm": 0.34593640455605407,
+      "learning_rate": 3.222049689440994e-05,
+      "loss": 0.458,
+      "step": 415
+    },
+    {
+      "epoch": 0.391425908667288,
+      "grad_norm": 0.42820613387711126,
+      "learning_rate": 3.260869565217392e-05,
+      "loss": 0.4757,
+      "step": 420
+    },
+    {
+      "epoch": 0.3960857409133271,
+      "grad_norm": 0.4358817121702933,
+      "learning_rate": 3.299689440993789e-05,
+      "loss": 0.4749,
+      "step": 425
+    },
+    {
+      "epoch": 0.40074557315936626,
+      "grad_norm": 0.4202207163358899,
+      "learning_rate": 3.3385093167701865e-05,
+      "loss": 0.4618,
+      "step": 430
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 0.357109766113718,
+      "learning_rate": 3.377329192546584e-05,
+      "loss": 0.4739,
+      "step": 435
+    },
+    {
+      "epoch": 0.41006523765144454,
+      "grad_norm": 0.3128342092948851,
+      "learning_rate": 3.4161490683229814e-05,
+      "loss": 0.4691,
+      "step": 440
+    },
+    {
+      "epoch": 0.4147250698974837,
+      "grad_norm": 0.34964364379003354,
+      "learning_rate": 3.4549689440993795e-05,
+      "loss": 0.4621,
+      "step": 445
+    },
+    {
+      "epoch": 0.4193849021435228,
+      "grad_norm": 0.39221756466087493,
+      "learning_rate": 3.493788819875777e-05,
+      "loss": 0.4545,
+      "step": 450
+    },
+    {
+      "epoch": 0.424044734389562,
+      "grad_norm": 0.4050444049964907,
+      "learning_rate": 3.532608695652174e-05,
+      "loss": 0.4612,
+      "step": 455
+    },
+    {
+      "epoch": 0.42870456663560114,
+      "grad_norm": 0.3634320321019683,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.4571,
+      "step": 460
+    },
+    {
+      "epoch": 0.43336439888164024,
+      "grad_norm": 0.42739366454744665,
+      "learning_rate": 3.610248447204969e-05,
+      "loss": 0.4515,
+      "step": 465
+    },
+    {
+      "epoch": 0.4380242311276794,
+      "grad_norm": 0.2913069345751138,
+      "learning_rate": 3.6490683229813665e-05,
+      "loss": 0.4673,
+      "step": 470
+    },
+    {
+      "epoch": 0.4426840633737186,
+      "grad_norm": 0.3098561941006096,
+      "learning_rate": 3.687888198757764e-05,
+      "loss": 0.4555,
+      "step": 475
+    },
+    {
+      "epoch": 0.4473438956197577,
+      "grad_norm": 0.357408910931571,
+      "learning_rate": 3.7267080745341614e-05,
+      "loss": 0.4672,
+      "step": 480
+    },
+    {
+      "epoch": 0.45200372786579684,
+      "grad_norm": 0.38007381160509457,
+      "learning_rate": 3.765527950310559e-05,
+      "loss": 0.4585,
+      "step": 485
+    },
+    {
+      "epoch": 0.45666356011183595,
+      "grad_norm": 0.32640167895211697,
+      "learning_rate": 3.804347826086957e-05,
+      "loss": 0.4554,
+      "step": 490
+    },
+    {
+      "epoch": 0.4613233923578751,
+      "grad_norm": 0.46229728689792227,
+      "learning_rate": 3.843167701863354e-05,
+      "loss": 0.4581,
+      "step": 495
+    },
+    {
+      "epoch": 0.4659832246039143,
+      "grad_norm": 0.362881155784984,
+      "learning_rate": 3.881987577639752e-05,
+      "loss": 0.4617,
+      "step": 500
+    },
+    {
+      "epoch": 0.4706430568499534,
+      "grad_norm": 0.36917807120287993,
+      "learning_rate": 3.920807453416149e-05,
+      "loss": 0.4482,
+      "step": 505
+    },
+    {
+      "epoch": 0.47530288909599255,
+      "grad_norm": 0.3245633798838294,
+      "learning_rate": 3.9596273291925465e-05,
+      "loss": 0.456,
+      "step": 510
+    },
+    {
+      "epoch": 0.47996272134203166,
+      "grad_norm": 0.29798132767356683,
+      "learning_rate": 3.998447204968944e-05,
+      "loss": 0.4535,
+      "step": 515
+    },
+    {
+      "epoch": 0.4846225535880708,
+      "grad_norm": 0.28016723890790424,
+      "learning_rate": 4.0372670807453414e-05,
+      "loss": 0.4562,
+      "step": 520
+    },
+    {
+      "epoch": 0.48928238583411,
+      "grad_norm": 0.4212412359229258,
+      "learning_rate": 4.076086956521739e-05,
+      "loss": 0.4575,
+      "step": 525
+    },
+    {
+      "epoch": 0.4939422180801491,
+      "grad_norm": 0.44105825258174725,
+      "learning_rate": 4.114906832298137e-05,
+      "loss": 0.4603,
+      "step": 530
+    },
+    {
+      "epoch": 0.49860205032618826,
+      "grad_norm": 0.364172013976295,
+      "learning_rate": 4.153726708074534e-05,
+      "loss": 0.4596,
+      "step": 535
+    },
+    {
+      "epoch": 0.5032618825722274,
+      "grad_norm": 0.37270027960845126,
+      "learning_rate": 4.192546583850932e-05,
+      "loss": 0.4485,
+      "step": 540
+    },
+    {
+      "epoch": 0.5079217148182665,
+      "grad_norm": 0.3328158564293457,
+      "learning_rate": 4.23136645962733e-05,
+      "loss": 0.4524,
+      "step": 545
+    },
+    {
+      "epoch": 0.5125815470643057,
+      "grad_norm": 0.33740077501332855,
+      "learning_rate": 4.270186335403727e-05,
+      "loss": 0.4514,
+      "step": 550
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.3177505203418901,
+      "learning_rate": 4.3090062111801246e-05,
+      "loss": 0.4437,
+      "step": 555
+    },
+    {
+      "epoch": 0.5219012115563839,
+      "grad_norm": 0.26294805335839155,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 0.4665,
+      "step": 560
+    },
+    {
+      "epoch": 0.5265610438024231,
+      "grad_norm": 0.2852645546169492,
+      "learning_rate": 4.3866459627329195e-05,
+      "loss": 0.4419,
+      "step": 565
+    },
+    {
+      "epoch": 0.5312208760484622,
+      "grad_norm": 0.2845447628207681,
+      "learning_rate": 4.425465838509317e-05,
+      "loss": 0.4427,
+      "step": 570
+    },
+    {
+      "epoch": 0.5358807082945014,
+      "grad_norm": 0.36535338229261943,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.4469,
+      "step": 575
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.31452296599079493,
+      "learning_rate": 4.5031055900621124e-05,
+      "loss": 0.4475,
+      "step": 580
+    },
+    {
+      "epoch": 0.5452003727865797,
+      "grad_norm": 0.29735223555272217,
+      "learning_rate": 4.54192546583851e-05,
+      "loss": 0.4458,
+      "step": 585
+    },
+    {
+      "epoch": 0.5498602050326188,
+      "grad_norm": 0.3126214078301511,
+      "learning_rate": 4.580745341614907e-05,
+      "loss": 0.4408,
+      "step": 590
+    },
+    {
+      "epoch": 0.554520037278658,
+      "grad_norm": 0.3210604166529035,
+      "learning_rate": 4.6195652173913046e-05,
+      "loss": 0.4432,
+      "step": 595
+    },
+    {
+      "epoch": 0.5591798695246971,
+      "grad_norm": 0.31891094827819777,
+      "learning_rate": 4.658385093167702e-05,
+      "loss": 0.4424,
+      "step": 600
+    },
+    {
+      "epoch": 0.5638397017707363,
+      "grad_norm": 0.25733195147107646,
+      "learning_rate": 4.6972049689440995e-05,
+      "loss": 0.4422,
+      "step": 605
+    },
+    {
+      "epoch": 0.5684995340167754,
+      "grad_norm": 0.3371390886138988,
+      "learning_rate": 4.736024844720497e-05,
+      "loss": 0.4525,
+      "step": 610
+    },
+    {
+      "epoch": 0.5731593662628145,
+      "grad_norm": 0.30637429801921895,
+      "learning_rate": 4.774844720496895e-05,
+      "loss": 0.4401,
+      "step": 615
+    },
+    {
+      "epoch": 0.5778191985088537,
+      "grad_norm": 0.29587694161905936,
+      "learning_rate": 4.8136645962732924e-05,
+      "loss": 0.4348,
+      "step": 620
+    },
+    {
+      "epoch": 0.5824790307548928,
+      "grad_norm": 0.2891558519070029,
+      "learning_rate": 4.85248447204969e-05,
+      "loss": 0.4574,
+      "step": 625
+    },
+    {
+      "epoch": 0.587138863000932,
+      "grad_norm": 0.2643030329040585,
+      "learning_rate": 4.891304347826087e-05,
+      "loss": 0.4499,
+      "step": 630
+    },
+    {
+      "epoch": 0.5917986952469712,
+      "grad_norm": 0.290133014155197,
+      "learning_rate": 4.9301242236024846e-05,
+      "loss": 0.4521,
+      "step": 635
+    },
+    {
+      "epoch": 0.5964585274930102,
+      "grad_norm": 0.3297963615333832,
+      "learning_rate": 4.968944099378882e-05,
+      "loss": 0.4432,
+      "step": 640
+    },
+    {
+      "epoch": 0.6011183597390494,
+      "grad_norm": 0.3893299119407896,
+      "learning_rate": 4.999137038315499e-05,
+      "loss": 0.4437,
+      "step": 645
+    },
+    {
+      "epoch": 0.6057781919850885,
+      "grad_norm": 0.2870391941851256,
+      "learning_rate": 4.994822229892993e-05,
+      "loss": 0.4509,
+      "step": 650
+    },
+    {
+      "epoch": 0.6104380242311277,
+      "grad_norm": 0.2715802093164516,
+      "learning_rate": 4.9905074214704875e-05,
+      "loss": 0.4487,
+      "step": 655
+    },
+    {
+      "epoch": 0.6150978564771669,
+      "grad_norm": 0.3285791192177138,
+      "learning_rate": 4.986192613047981e-05,
+      "loss": 0.4319,
+      "step": 660
+    },
+    {
+      "epoch": 0.6197576887232059,
+      "grad_norm": 0.33495396782671516,
+      "learning_rate": 4.9818778046254746e-05,
+      "loss": 0.4314,
+      "step": 665
+    },
+    {
+      "epoch": 0.6244175209692451,
+      "grad_norm": 0.29843444679839276,
+      "learning_rate": 4.977562996202969e-05,
+      "loss": 0.4307,
+      "step": 670
+    },
+    {
+      "epoch": 0.6290773532152842,
+      "grad_norm": 0.3501798273958651,
+      "learning_rate": 4.9732481877804624e-05,
+      "loss": 0.4375,
+      "step": 675
+    },
+    {
+      "epoch": 0.6337371854613234,
+      "grad_norm": 0.3427153919468158,
+      "learning_rate": 4.968933379357957e-05,
+      "loss": 0.4425,
+      "step": 680
+    },
+    {
+      "epoch": 0.6383970177073626,
+      "grad_norm": 0.3504930257944694,
+      "learning_rate": 4.964618570935451e-05,
+      "loss": 0.4504,
+      "step": 685
+    },
+    {
+      "epoch": 0.6430568499534017,
+      "grad_norm": 0.29000948040848384,
+      "learning_rate": 4.9603037625129445e-05,
+      "loss": 0.432,
+      "step": 690
+    },
+    {
+      "epoch": 0.6477166821994408,
+      "grad_norm": 0.24755545050269823,
+      "learning_rate": 4.955988954090439e-05,
+      "loss": 0.4369,
+      "step": 695
+    },
+    {
+      "epoch": 0.65237651444548,
+      "grad_norm": 0.2954176485945787,
+      "learning_rate": 4.951674145667933e-05,
+      "loss": 0.4463,
+      "step": 700
+    },
+    {
+      "epoch": 0.6570363466915191,
+      "grad_norm": 0.28932089879966616,
+      "learning_rate": 4.9473593372454266e-05,
+      "loss": 0.4404,
+      "step": 705
+    },
+    {
+      "epoch": 0.6616961789375583,
+      "grad_norm": 0.2881289318776839,
+      "learning_rate": 4.94304452882292e-05,
+      "loss": 0.4409,
+      "step": 710
+    },
+    {
+      "epoch": 0.6663560111835974,
+      "grad_norm": 0.30966677832024303,
+      "learning_rate": 4.9387297204004144e-05,
+      "loss": 0.4283,
+      "step": 715
+    },
+    {
+      "epoch": 0.6710158434296365,
+      "grad_norm": 0.34794708795559215,
+      "learning_rate": 4.934414911977908e-05,
+      "loss": 0.4387,
+      "step": 720
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.2700084450136511,
+      "learning_rate": 4.930100103555402e-05,
+      "loss": 0.4377,
+      "step": 725
+    },
+    {
+      "epoch": 0.6803355079217148,
+      "grad_norm": 0.2827457096119229,
+      "learning_rate": 4.9257852951328965e-05,
+      "loss": 0.4334,
+      "step": 730
+    },
+    {
+      "epoch": 0.684995340167754,
+      "grad_norm": 0.25971959728748645,
+      "learning_rate": 4.921470486710391e-05,
+      "loss": 0.4367,
+      "step": 735
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.22491817023977273,
+      "learning_rate": 4.917155678287884e-05,
+      "loss": 0.4297,
+      "step": 740
+    },
+    {
+      "epoch": 0.6943150046598322,
+      "grad_norm": 0.26902266376810086,
+      "learning_rate": 4.9128408698653785e-05,
+      "loss": 0.4257,
+      "step": 745
+    },
+    {
+      "epoch": 0.6989748369058714,
+      "grad_norm": 0.2629969267846459,
+      "learning_rate": 4.908526061442872e-05,
+      "loss": 0.4292,
+      "step": 750
+    },
+    {
+      "epoch": 0.7036346691519105,
+      "grad_norm": 0.2926854504645197,
+      "learning_rate": 4.904211253020366e-05,
+      "loss": 0.4325,
+      "step": 755
+    },
+    {
+      "epoch": 0.7082945013979497,
+      "grad_norm": 0.3104189920307263,
+      "learning_rate": 4.89989644459786e-05,
+      "loss": 0.4362,
+      "step": 760
+    },
+    {
+      "epoch": 0.7129543336439889,
+      "grad_norm": 0.254928139438877,
+      "learning_rate": 4.895581636175354e-05,
+      "loss": 0.4186,
+      "step": 765
+    },
+    {
+      "epoch": 0.7176141658900279,
+      "grad_norm": 0.25240112313887436,
+      "learning_rate": 4.891266827752848e-05,
+      "loss": 0.4285,
+      "step": 770
+    },
+    {
+      "epoch": 0.7222739981360671,
+      "grad_norm": 0.24714733800627392,
+      "learning_rate": 4.886952019330342e-05,
+      "loss": 0.4411,
+      "step": 775
+    },
+    {
+      "epoch": 0.7269338303821062,
+      "grad_norm": 0.24228318234317853,
+      "learning_rate": 4.882637210907836e-05,
+      "loss": 0.4256,
+      "step": 780
+    },
+    {
+      "epoch": 0.7315936626281454,
+      "grad_norm": 0.2667179700426176,
+      "learning_rate": 4.87832240248533e-05,
+      "loss": 0.431,
+      "step": 785
+    },
+    {
+      "epoch": 0.7362534948741846,
+      "grad_norm": 0.24095926762318678,
+      "learning_rate": 4.874007594062824e-05,
+      "loss": 0.4358,
+      "step": 790
+    },
+    {
+      "epoch": 0.7409133271202236,
+      "grad_norm": 0.24206178812577334,
+      "learning_rate": 4.8696927856403176e-05,
+      "loss": 0.4305,
+      "step": 795
+    },
+    {
+      "epoch": 0.7455731593662628,
+      "grad_norm": 0.2598632865022014,
+      "learning_rate": 4.865377977217811e-05,
+      "loss": 0.428,
+      "step": 800
+    },
+    {
+      "epoch": 0.750232991612302,
+      "grad_norm": 0.2818536749930896,
+      "learning_rate": 4.8610631687953055e-05,
+      "loss": 0.4272,
+      "step": 805
+    },
+    {
+      "epoch": 0.7548928238583411,
+      "grad_norm": 0.3008302888331436,
+      "learning_rate": 4.8567483603728e-05,
+      "loss": 0.4454,
+      "step": 810
+    },
+    {
+      "epoch": 0.7595526561043803,
+      "grad_norm": 0.3100087821853064,
+      "learning_rate": 4.852433551950293e-05,
+      "loss": 0.4202,
+      "step": 815
+    },
+    {
+      "epoch": 0.7642124883504194,
+      "grad_norm": 0.3074315721619132,
+      "learning_rate": 4.8481187435277875e-05,
+      "loss": 0.4362,
+      "step": 820
+    },
+    {
+      "epoch": 0.7688723205964585,
+      "grad_norm": 0.2966994001237217,
+      "learning_rate": 4.843803935105282e-05,
+      "loss": 0.4283,
+      "step": 825
+    },
+    {
+      "epoch": 0.7735321528424977,
+      "grad_norm": 0.2468427548772633,
+      "learning_rate": 4.839489126682776e-05,
+      "loss": 0.4232,
+      "step": 830
+    },
+    {
+      "epoch": 0.7781919850885368,
+      "grad_norm": 0.239445391381232,
+      "learning_rate": 4.8351743182602696e-05,
+      "loss": 0.4286,
+      "step": 835
+    },
+    {
+      "epoch": 0.782851817334576,
+      "grad_norm": 0.28150948918472163,
+      "learning_rate": 4.830859509837763e-05,
+      "loss": 0.4327,
+      "step": 840
+    },
+    {
+      "epoch": 0.7875116495806151,
+      "grad_norm": 0.3734765126890979,
+      "learning_rate": 4.8265447014152574e-05,
+      "loss": 0.4407,
+      "step": 845
+    },
+    {
+      "epoch": 0.7921714818266542,
+      "grad_norm": 0.29043000955747134,
+      "learning_rate": 4.822229892992751e-05,
+      "loss": 0.4299,
+      "step": 850
+    },
+    {
+      "epoch": 0.7968313140726934,
+      "grad_norm": 0.2611412425146056,
+      "learning_rate": 4.817915084570245e-05,
+      "loss": 0.4318,
+      "step": 855
+    },
+    {
+      "epoch": 0.8014911463187325,
+      "grad_norm": 0.22623048551154368,
+      "learning_rate": 4.8136002761477395e-05,
+      "loss": 0.4395,
+      "step": 860
+    },
+    {
+      "epoch": 0.8061509785647717,
+      "grad_norm": 0.24710642344488037,
+      "learning_rate": 4.809285467725233e-05,
+      "loss": 0.4195,
+      "step": 865
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.2221937124982761,
+      "learning_rate": 4.804970659302727e-05,
+      "loss": 0.434,
+      "step": 870
+    },
+    {
+      "epoch": 0.8154706430568499,
+      "grad_norm": 0.23466507192816327,
+      "learning_rate": 4.8006558508802216e-05,
+      "loss": 0.4292,
+      "step": 875
+    },
+    {
+      "epoch": 0.8201304753028891,
+      "grad_norm": 0.2248157749350788,
+      "learning_rate": 4.796341042457715e-05,
+      "loss": 0.4201,
+      "step": 880
+    },
+    {
+      "epoch": 0.8247903075489282,
+      "grad_norm": 0.24586022261362458,
+      "learning_rate": 4.792026234035209e-05,
+      "loss": 0.4341,
+      "step": 885
+    },
+    {
+      "epoch": 0.8294501397949674,
+      "grad_norm": 0.25929743995571014,
+      "learning_rate": 4.787711425612703e-05,
+      "loss": 0.4214,
+      "step": 890
+    },
+    {
+      "epoch": 0.8341099720410066,
+      "grad_norm": 0.25430868339864937,
+      "learning_rate": 4.7833966171901966e-05,
+      "loss": 0.441,
+      "step": 895
+    },
+    {
+      "epoch": 0.8387698042870456,
+      "grad_norm": 0.218152703390182,
+      "learning_rate": 4.779081808767691e-05,
+      "loss": 0.4231,
+      "step": 900
+    },
+    {
+      "epoch": 0.8434296365330848,
+      "grad_norm": 0.22569219066377355,
+      "learning_rate": 4.774767000345185e-05,
+      "loss": 0.4243,
+      "step": 905
+    },
+    {
+      "epoch": 0.848089468779124,
+      "grad_norm": 0.2296437868718205,
+      "learning_rate": 4.770452191922679e-05,
+      "loss": 0.4297,
+      "step": 910
+    },
+    {
+      "epoch": 0.8527493010251631,
+      "grad_norm": 0.2724473449205512,
+      "learning_rate": 4.766137383500173e-05,
+      "loss": 0.4235,
+      "step": 915
+    },
+    {
+      "epoch": 0.8574091332712023,
+      "grad_norm": 0.26893401021329133,
+      "learning_rate": 4.761822575077667e-05,
+      "loss": 0.4309,
+      "step": 920
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.24994479469769898,
+      "learning_rate": 4.757507766655161e-05,
+      "loss": 0.4179,
+      "step": 925
+    },
+    {
+      "epoch": 0.8667287977632805,
+      "grad_norm": 0.2905519215722696,
+      "learning_rate": 4.753192958232654e-05,
+      "loss": 0.4148,
+      "step": 930
+    },
+    {
+      "epoch": 0.8713886300093197,
+      "grad_norm": 0.23313449743539438,
+      "learning_rate": 4.7488781498101485e-05,
+      "loss": 0.427,
+      "step": 935
+    },
+    {
+      "epoch": 0.8760484622553588,
+      "grad_norm": 0.2391978515901403,
+      "learning_rate": 4.744563341387643e-05,
+      "loss": 0.424,
+      "step": 940
+    },
+    {
+      "epoch": 0.880708294501398,
+      "grad_norm": 0.20062245456807687,
+      "learning_rate": 4.7402485329651363e-05,
+      "loss": 0.4258,
+      "step": 945
+    },
+    {
+      "epoch": 0.8853681267474371,
+      "grad_norm": 0.21218216322801703,
+      "learning_rate": 4.7359337245426306e-05,
+      "loss": 0.4333,
+      "step": 950
+    },
+    {
+      "epoch": 0.8900279589934762,
+      "grad_norm": 0.24606784096035242,
+      "learning_rate": 4.731618916120125e-05,
+      "loss": 0.4316,
+      "step": 955
+    },
+    {
+      "epoch": 0.8946877912395154,
+      "grad_norm": 0.23119956517851986,
+      "learning_rate": 4.7273041076976184e-05,
+      "loss": 0.4267,
+      "step": 960
+    },
+    {
+      "epoch": 0.8993476234855545,
+      "grad_norm": 0.25653893194920113,
+      "learning_rate": 4.722989299275113e-05,
+      "loss": 0.4193,
+      "step": 965
+    },
+    {
+      "epoch": 0.9040074557315937,
+      "grad_norm": 0.23417875926682785,
+      "learning_rate": 4.718674490852606e-05,
+      "loss": 0.4164,
+      "step": 970
+    },
+    {
+      "epoch": 0.9086672879776329,
+      "grad_norm": 0.25956801747128566,
+      "learning_rate": 4.7143596824301e-05,
+      "loss": 0.4358,
+      "step": 975
+    },
+    {
+      "epoch": 0.9133271202236719,
+      "grad_norm": 0.2745626337435571,
+      "learning_rate": 4.710044874007594e-05,
+      "loss": 0.4227,
+      "step": 980
+    },
+    {
+      "epoch": 0.9179869524697111,
+      "grad_norm": 0.2838916621597491,
+      "learning_rate": 4.705730065585088e-05,
+      "loss": 0.4229,
+      "step": 985
+    },
+    {
+      "epoch": 0.9226467847157502,
+      "grad_norm": 0.2551505455839328,
+      "learning_rate": 4.7014152571625826e-05,
+      "loss": 0.4215,
+      "step": 990
+    },
+    {
+      "epoch": 0.9273066169617894,
+      "grad_norm": 0.25012739329275585,
+      "learning_rate": 4.697100448740076e-05,
+      "loss": 0.417,
+      "step": 995
+    },
+    {
+      "epoch": 0.9319664492078286,
+      "grad_norm": 0.2880442638315601,
+      "learning_rate": 4.6927856403175704e-05,
+      "loss": 0.4194,
+      "step": 1000
+    },
+    {
+      "epoch": 0.9366262814538676,
+      "grad_norm": 0.23599077885697287,
+      "learning_rate": 4.6884708318950646e-05,
+      "loss": 0.4244,
+      "step": 1005
+    },
+    {
+      "epoch": 0.9412861136999068,
+      "grad_norm": 0.28072119580408117,
+      "learning_rate": 4.684156023472558e-05,
+      "loss": 0.4173,
+      "step": 1010
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.2635634256596414,
+      "learning_rate": 4.679841215050052e-05,
+      "loss": 0.4273,
+      "step": 1015
+    },
+    {
+      "epoch": 0.9506057781919851,
+      "grad_norm": 0.23120843950786468,
+      "learning_rate": 4.675526406627546e-05,
+      "loss": 0.4193,
+      "step": 1020
+    },
+    {
+      "epoch": 0.9552656104380243,
+      "grad_norm": 0.2416655060234862,
+      "learning_rate": 4.6712115982050396e-05,
+      "loss": 0.4244,
+      "step": 1025
+    },
+    {
+      "epoch": 0.9599254426840633,
+      "grad_norm": 0.23294322281602764,
+      "learning_rate": 4.666896789782534e-05,
+      "loss": 0.4271,
+      "step": 1030
+    },
+    {
+      "epoch": 0.9645852749301025,
+      "grad_norm": 0.24531344399536567,
+      "learning_rate": 4.662581981360028e-05,
+      "loss": 0.4211,
+      "step": 1035
+    },
+    {
+      "epoch": 0.9692451071761417,
+      "grad_norm": 0.24303315355213737,
+      "learning_rate": 4.658267172937522e-05,
+      "loss": 0.4183,
+      "step": 1040
+    },
+    {
+      "epoch": 0.9739049394221808,
+      "grad_norm": 0.22776901759480672,
+      "learning_rate": 4.653952364515016e-05,
+      "loss": 0.4101,
+      "step": 1045
+    },
+    {
+      "epoch": 0.97856477166822,
+      "grad_norm": 0.2286845974873945,
+      "learning_rate": 4.64963755609251e-05,
+      "loss": 0.4398,
+      "step": 1050
+    },
+    {
+      "epoch": 0.983224603914259,
+      "grad_norm": 0.23116427577249352,
+      "learning_rate": 4.645322747670004e-05,
+      "loss": 0.4182,
+      "step": 1055
+    },
+    {
+      "epoch": 0.9878844361602982,
+      "grad_norm": 0.23826208735123147,
+      "learning_rate": 4.641007939247497e-05,
+      "loss": 0.4163,
+      "step": 1060
+    },
+    {
+      "epoch": 0.9925442684063374,
+      "grad_norm": 0.21093557220222217,
+      "learning_rate": 4.6366931308249916e-05,
+      "loss": 0.4206,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9972041006523765,
+      "grad_norm": 0.20952993245488266,
+      "learning_rate": 4.632378322402486e-05,
+      "loss": 0.4209,
+      "step": 1070
+    },
+    {
+      "epoch": 1.0018639328984156,
+      "grad_norm": 0.2542085635079821,
+      "learning_rate": 4.6280635139799794e-05,
+      "loss": 0.3906,
+      "step": 1075
+    },
+    {
+      "epoch": 1.0065237651444547,
+      "grad_norm": 0.23525440980099246,
+      "learning_rate": 4.6237487055574736e-05,
+      "loss": 0.3786,
+      "step": 1080
+    },
+    {
+      "epoch": 1.011183597390494,
+      "grad_norm": 0.28154418552416793,
+      "learning_rate": 4.619433897134968e-05,
+      "loss": 0.3842,
+      "step": 1085
+    },
+    {
+      "epoch": 1.015843429636533,
+      "grad_norm": 0.25827181472583155,
+      "learning_rate": 4.6151190887124615e-05,
+      "loss": 0.3925,
+      "step": 1090
+    },
+    {
+      "epoch": 1.0205032618825722,
+      "grad_norm": 0.22387177284004287,
+      "learning_rate": 4.610804280289955e-05,
+      "loss": 0.3781,
+      "step": 1095
+    },
+    {
+      "epoch": 1.0251630941286114,
+      "grad_norm": 0.2063596321184128,
+      "learning_rate": 4.606489471867449e-05,
+      "loss": 0.3674,
+      "step": 1100
+    },
+    {
+      "epoch": 1.0298229263746506,
+      "grad_norm": 0.2163976644741488,
+      "learning_rate": 4.602174663444943e-05,
+      "loss": 0.3756,
+      "step": 1105
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 0.23965601626912958,
+      "learning_rate": 4.597859855022437e-05,
+      "loss": 0.3749,
+      "step": 1110
+    },
+    {
+      "epoch": 1.0391425908667289,
+      "grad_norm": 0.21809408759172885,
+      "learning_rate": 4.5935450465999314e-05,
+      "loss": 0.3746,
+      "step": 1115
+    },
+    {
+      "epoch": 1.0438024231127678,
+      "grad_norm": 0.2146681212359368,
+      "learning_rate": 4.589230238177425e-05,
+      "loss": 0.3738,
+      "step": 1120
+    },
+    {
+      "epoch": 1.048462255358807,
+      "grad_norm": 0.2164808135770368,
+      "learning_rate": 4.584915429754919e-05,
+      "loss": 0.381,
+      "step": 1125
+    },
+    {
+      "epoch": 1.0531220876048462,
+      "grad_norm": 0.20038845101621153,
+      "learning_rate": 4.5806006213324134e-05,
+      "loss": 0.3755,
+      "step": 1130
+    },
+    {
+      "epoch": 1.0577819198508853,
+      "grad_norm": 0.22908928446036173,
+      "learning_rate": 4.576285812909907e-05,
+      "loss": 0.3937,
+      "step": 1135
+    },
+    {
+      "epoch": 1.0624417520969245,
+      "grad_norm": 0.21262252219726624,
+      "learning_rate": 4.5719710044874006e-05,
+      "loss": 0.3855,
+      "step": 1140
+    },
+    {
+      "epoch": 1.0671015843429636,
+      "grad_norm": 0.2434452660202187,
+      "learning_rate": 4.567656196064895e-05,
+      "loss": 0.3753,
+      "step": 1145
+    },
+    {
+      "epoch": 1.0717614165890028,
+      "grad_norm": 0.22521632895216864,
+      "learning_rate": 4.563341387642389e-05,
+      "loss": 0.3737,
+      "step": 1150
+    },
+    {
+      "epoch": 1.076421248835042,
+      "grad_norm": 0.2600605074683848,
+      "learning_rate": 4.5590265792198827e-05,
+      "loss": 0.3831,
+      "step": 1155
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.2476145059761402,
+      "learning_rate": 4.554711770797377e-05,
+      "loss": 0.376,
+      "step": 1160
+    },
+    {
+      "epoch": 1.0857409133271203,
+      "grad_norm": 0.2610823774589846,
+      "learning_rate": 4.550396962374871e-05,
+      "loss": 0.388,
+      "step": 1165
+    },
+    {
+      "epoch": 1.0904007455731595,
+      "grad_norm": 0.2533322111505516,
+      "learning_rate": 4.546082153952365e-05,
+      "loss": 0.3903,
+      "step": 1170
+    },
+    {
+      "epoch": 1.0950605778191984,
+      "grad_norm": 0.22394262454966646,
+      "learning_rate": 4.541767345529859e-05,
+      "loss": 0.3737,
+      "step": 1175
+    },
+    {
+      "epoch": 1.0997204100652376,
+      "grad_norm": 0.23178884867275992,
+      "learning_rate": 4.5374525371073526e-05,
+      "loss": 0.3792,
+      "step": 1180
+    },
+    {
+      "epoch": 1.1043802423112767,
+      "grad_norm": 0.22252120261609648,
+      "learning_rate": 4.533137728684846e-05,
+      "loss": 0.3769,
+      "step": 1185
+    },
+    {
+      "epoch": 1.109040074557316,
+      "grad_norm": 0.20526095591805865,
+      "learning_rate": 4.5288229202623404e-05,
+      "loss": 0.3769,
+      "step": 1190
+    },
+    {
+      "epoch": 1.113699906803355,
+      "grad_norm": 0.21845670980292947,
+      "learning_rate": 4.5245081118398346e-05,
+      "loss": 0.3913,
+      "step": 1195
+    },
+    {
+      "epoch": 1.1183597390493942,
+      "grad_norm": 0.25921479143094495,
+      "learning_rate": 4.520193303417328e-05,
+      "loss": 0.3784,
+      "step": 1200
+    },
+    {
+      "epoch": 1.1230195712954334,
+      "grad_norm": 0.26881562520931584,
+      "learning_rate": 4.5158784949948224e-05,
+      "loss": 0.3791,
+      "step": 1205
+    },
+    {
+      "epoch": 1.1276794035414726,
+      "grad_norm": 0.2549148746109543,
+      "learning_rate": 4.511563686572317e-05,
+      "loss": 0.3764,
+      "step": 1210
+    },
+    {
+      "epoch": 1.1323392357875117,
+      "grad_norm": 0.19585298960229888,
+      "learning_rate": 4.50724887814981e-05,
+      "loss": 0.3875,
+      "step": 1215
+    },
+    {
+      "epoch": 1.1369990680335509,
+      "grad_norm": 0.22758121062666856,
+      "learning_rate": 4.5029340697273045e-05,
+      "loss": 0.3834,
+      "step": 1220
+    },
+    {
+      "epoch": 1.14165890027959,
+      "grad_norm": 0.22485639482824113,
+      "learning_rate": 4.498619261304798e-05,
+      "loss": 0.3799,
+      "step": 1225
+    },
+    {
+      "epoch": 1.146318732525629,
+      "grad_norm": 0.21996606801680868,
+      "learning_rate": 4.4943044528822923e-05,
+      "loss": 0.3821,
+      "step": 1230
+    },
+    {
+      "epoch": 1.1509785647716682,
+      "grad_norm": 0.2317463824272281,
+      "learning_rate": 4.489989644459786e-05,
+      "loss": 0.3765,
+      "step": 1235
+    },
+    {
+      "epoch": 1.1556383970177073,
+      "grad_norm": 0.21233493420387,
+      "learning_rate": 4.48567483603728e-05,
+      "loss": 0.377,
+      "step": 1240
+    },
+    {
+      "epoch": 1.1602982292637465,
+      "grad_norm": 0.19334974160350707,
+      "learning_rate": 4.4813600276147744e-05,
+      "loss": 0.3853,
+      "step": 1245
+    },
+    {
+      "epoch": 1.1649580615097856,
+      "grad_norm": 0.22270099209315078,
+      "learning_rate": 4.477045219192268e-05,
+      "loss": 0.3792,
+      "step": 1250
+    },
+    {
+      "epoch": 1.1696178937558248,
+      "grad_norm": 0.19918268379484572,
+      "learning_rate": 4.472730410769762e-05,
+      "loss": 0.3542,
+      "step": 1255
+    },
+    {
+      "epoch": 1.174277726001864,
+      "grad_norm": 0.19894227060205655,
+      "learning_rate": 4.4684156023472565e-05,
+      "loss": 0.3791,
+      "step": 1260
+    },
+    {
+      "epoch": 1.1789375582479031,
+      "grad_norm": 0.21857921210060932,
+      "learning_rate": 4.46410079392475e-05,
+      "loss": 0.373,
+      "step": 1265
+    },
+    {
+      "epoch": 1.1835973904939423,
+      "grad_norm": 0.23152513440746553,
+      "learning_rate": 4.4597859855022436e-05,
+      "loss": 0.3809,
+      "step": 1270
+    },
+    {
+      "epoch": 1.1882572227399812,
+      "grad_norm": 0.2144192282979416,
+      "learning_rate": 4.455471177079738e-05,
+      "loss": 0.3745,
+      "step": 1275
+    },
+    {
+      "epoch": 1.1929170549860204,
+      "grad_norm": 0.22022007346824968,
+      "learning_rate": 4.4511563686572315e-05,
+      "loss": 0.3892,
+      "step": 1280
+    },
+    {
+      "epoch": 1.1975768872320596,
+      "grad_norm": 0.21919158015058546,
+      "learning_rate": 4.446841560234726e-05,
+      "loss": 0.3894,
+      "step": 1285
+    },
+    {
+      "epoch": 1.2022367194780987,
+      "grad_norm": 0.2056004661138724,
+      "learning_rate": 4.44252675181222e-05,
+      "loss": 0.3705,
+      "step": 1290
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 0.2159415587868541,
+      "learning_rate": 4.4382119433897135e-05,
+      "loss": 0.3769,
+      "step": 1295
+    },
+    {
+      "epoch": 1.211556383970177,
+      "grad_norm": 0.2803474793282514,
+      "learning_rate": 4.433897134967208e-05,
+      "loss": 0.3835,
+      "step": 1300
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.25289170186439075,
+      "learning_rate": 4.429582326544702e-05,
+      "loss": 0.3836,
+      "step": 1305
+    },
+    {
+      "epoch": 1.2208760484622554,
+      "grad_norm": 0.22675032169351944,
+      "learning_rate": 4.4252675181221956e-05,
+      "loss": 0.3722,
+      "step": 1310
+    },
+    {
+      "epoch": 1.2255358807082946,
+      "grad_norm": 0.24330994532254344,
+      "learning_rate": 4.420952709699689e-05,
+      "loss": 0.3933,
+      "step": 1315
+    },
+    {
+      "epoch": 1.2301957129543337,
+      "grad_norm": 0.21354016911735535,
+      "learning_rate": 4.4166379012771834e-05,
+      "loss": 0.3795,
+      "step": 1320
+    },
+    {
+      "epoch": 1.2348555452003729,
+      "grad_norm": 0.20747141402328617,
+      "learning_rate": 4.412323092854678e-05,
+      "loss": 0.3868,
+      "step": 1325
+    },
+    {
+      "epoch": 1.2395153774464118,
+      "grad_norm": 0.20919740267084352,
+      "learning_rate": 4.408008284432171e-05,
+      "loss": 0.3738,
+      "step": 1330
+    },
+    {
+      "epoch": 1.244175209692451,
+      "grad_norm": 0.22293752898621816,
+      "learning_rate": 4.4036934760096655e-05,
+      "loss": 0.3746,
+      "step": 1335
+    },
+    {
+      "epoch": 1.2488350419384902,
+      "grad_norm": 0.2518579841737606,
+      "learning_rate": 4.39937866758716e-05,
+      "loss": 0.3828,
+      "step": 1340
+    },
+    {
+      "epoch": 1.2534948741845293,
+      "grad_norm": 0.2290885571287781,
+      "learning_rate": 4.395063859164653e-05,
+      "loss": 0.3694,
+      "step": 1345
+    },
+    {
+      "epoch": 1.2581547064305685,
+      "grad_norm": 0.20605301346454027,
+      "learning_rate": 4.3907490507421476e-05,
+      "loss": 0.3769,
+      "step": 1350
+    },
+    {
+      "epoch": 1.2628145386766076,
+      "grad_norm": 0.23140803414726607,
+      "learning_rate": 4.386434242319641e-05,
+      "loss": 0.3865,
+      "step": 1355
+    },
+    {
+      "epoch": 1.2674743709226468,
+      "grad_norm": 0.23977627991613779,
+      "learning_rate": 4.382119433897135e-05,
+      "loss": 0.3755,
+      "step": 1360
+    },
+    {
+      "epoch": 1.272134203168686,
+      "grad_norm": 0.19621458265540112,
+      "learning_rate": 4.377804625474629e-05,
+      "loss": 0.3817,
+      "step": 1365
+    },
+    {
+      "epoch": 1.2767940354147251,
+      "grad_norm": 0.21105116341518093,
+      "learning_rate": 4.373489817052123e-05,
+      "loss": 0.3818,
+      "step": 1370
+    },
+    {
+      "epoch": 1.281453867660764,
+      "grad_norm": 0.19753350779049403,
+      "learning_rate": 4.369175008629617e-05,
+      "loss": 0.3705,
+      "step": 1375
+    },
+    {
+      "epoch": 1.2861136999068035,
+      "grad_norm": 0.20903672915282,
+      "learning_rate": 4.364860200207111e-05,
+      "loss": 0.385,
+      "step": 1380
+    },
+    {
+      "epoch": 1.2907735321528424,
+      "grad_norm": 0.23707970076420615,
+      "learning_rate": 4.360545391784605e-05,
+      "loss": 0.3854,
+      "step": 1385
+    },
+    {
+      "epoch": 1.2954333643988816,
+      "grad_norm": 0.18617341901760584,
+      "learning_rate": 4.356230583362099e-05,
+      "loss": 0.3792,
+      "step": 1390
+    },
+    {
+      "epoch": 1.3000931966449207,
+      "grad_norm": 0.20405354057137057,
+      "learning_rate": 4.351915774939593e-05,
+      "loss": 0.3824,
+      "step": 1395
+    },
+    {
+      "epoch": 1.30475302889096,
+      "grad_norm": 0.20869904243474333,
+      "learning_rate": 4.347600966517087e-05,
+      "loss": 0.3747,
+      "step": 1400
+    },
+    {
+      "epoch": 1.309412861136999,
+      "grad_norm": 0.23106729315097563,
+      "learning_rate": 4.343286158094581e-05,
+      "loss": 0.3853,
+      "step": 1405
+    },
+    {
+      "epoch": 1.3140726933830382,
+      "grad_norm": 0.22149589619643079,
+      "learning_rate": 4.3389713496720745e-05,
+      "loss": 0.3742,
+      "step": 1410
+    },
+    {
+      "epoch": 1.3187325256290774,
+      "grad_norm": 0.19247405518518898,
+      "learning_rate": 4.334656541249569e-05,
+      "loss": 0.3806,
+      "step": 1415
+    },
+    {
+      "epoch": 1.3233923578751166,
+      "grad_norm": 0.19999051098061882,
+      "learning_rate": 4.330341732827063e-05,
+      "loss": 0.3694,
+      "step": 1420
+    },
+    {
+      "epoch": 1.3280521901211557,
+      "grad_norm": 0.19291445318883707,
+      "learning_rate": 4.3260269244045566e-05,
+      "loss": 0.3774,
+      "step": 1425
+    },
+    {
+      "epoch": 1.3327120223671947,
+      "grad_norm": 0.19541548501759143,
+      "learning_rate": 4.321712115982051e-05,
+      "loss": 0.3845,
+      "step": 1430
+    },
+    {
+      "epoch": 1.337371854613234,
+      "grad_norm": 0.250609158947476,
+      "learning_rate": 4.317397307559545e-05,
+      "loss": 0.3839,
+      "step": 1435
+    },
+    {
+      "epoch": 1.342031686859273,
+      "grad_norm": 0.24110178472064014,
+      "learning_rate": 4.3130824991370387e-05,
+      "loss": 0.3818,
+      "step": 1440
+    },
+    {
+      "epoch": 1.3466915191053122,
+      "grad_norm": 0.20375144928928238,
+      "learning_rate": 4.308767690714532e-05,
+      "loss": 0.3871,
+      "step": 1445
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.20060485323561683,
+      "learning_rate": 4.3044528822920265e-05,
+      "loss": 0.3732,
+      "step": 1450
+    },
+    {
+      "epoch": 1.3560111835973905,
+      "grad_norm": 0.20326455690653897,
+      "learning_rate": 4.30013807386952e-05,
+      "loss": 0.3798,
+      "step": 1455
+    },
+    {
+      "epoch": 1.3606710158434296,
+      "grad_norm": 0.25390373781780784,
+      "learning_rate": 4.295823265447014e-05,
+      "loss": 0.3853,
+      "step": 1460
+    },
+    {
+      "epoch": 1.3653308480894688,
+      "grad_norm": 0.21217867916654828,
+      "learning_rate": 4.2915084570245085e-05,
+      "loss": 0.3794,
+      "step": 1465
+    },
+    {
+      "epoch": 1.369990680335508,
+      "grad_norm": 0.19154324062422584,
+      "learning_rate": 4.287193648602002e-05,
+      "loss": 0.3745,
+      "step": 1470
+    },
+    {
+      "epoch": 1.3746505125815471,
+      "grad_norm": 0.1890211144792994,
+      "learning_rate": 4.2828788401794964e-05,
+      "loss": 0.378,
+      "step": 1475
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 0.190827313112808,
+      "learning_rate": 4.27856403175699e-05,
+      "loss": 0.3612,
+      "step": 1480
+    },
+    {
+      "epoch": 1.3839701770736252,
+      "grad_norm": 0.20166073798658482,
+      "learning_rate": 4.274249223334484e-05,
+      "loss": 0.3812,
+      "step": 1485
+    },
+    {
+      "epoch": 1.3886300093196646,
+      "grad_norm": 0.23888461713071446,
+      "learning_rate": 4.269934414911978e-05,
+      "loss": 0.3836,
+      "step": 1490
+    },
+    {
+      "epoch": 1.3932898415657036,
+      "grad_norm": 0.24503693315398012,
+      "learning_rate": 4.265619606489472e-05,
+      "loss": 0.3947,
+      "step": 1495
+    },
+    {
+      "epoch": 1.3979496738117427,
+      "grad_norm": 0.2254793552087811,
+      "learning_rate": 4.261304798066966e-05,
+      "loss": 0.3896,
+      "step": 1500
+    },
+    {
+      "epoch": 1.402609506057782,
+      "grad_norm": 0.22280588670933843,
+      "learning_rate": 4.25698998964446e-05,
+      "loss": 0.3847,
+      "step": 1505
+    },
+    {
+      "epoch": 1.407269338303821,
+      "grad_norm": 0.1890299209833747,
+      "learning_rate": 4.252675181221954e-05,
+      "loss": 0.3784,
+      "step": 1510
+    },
+    {
+      "epoch": 1.4119291705498602,
+      "grad_norm": 0.20018547148888208,
+      "learning_rate": 4.248360372799448e-05,
+      "loss": 0.3813,
+      "step": 1515
+    },
+    {
+      "epoch": 1.4165890027958994,
+      "grad_norm": 0.20939261041651627,
+      "learning_rate": 4.244045564376942e-05,
+      "loss": 0.3792,
+      "step": 1520
+    },
+    {
+      "epoch": 1.4212488350419386,
+      "grad_norm": 0.19982422224236887,
+      "learning_rate": 4.2397307559544355e-05,
+      "loss": 0.3784,
+      "step": 1525
+    },
+    {
+      "epoch": 1.4259086672879777,
+      "grad_norm": 0.21965866261326533,
+      "learning_rate": 4.23541594753193e-05,
+      "loss": 0.375,
+      "step": 1530
+    },
+    {
+      "epoch": 1.4305684995340169,
+      "grad_norm": 0.24099895516334108,
+      "learning_rate": 4.231101139109423e-05,
+      "loss": 0.3818,
+      "step": 1535
+    },
+    {
+      "epoch": 1.4352283317800558,
+      "grad_norm": 0.19773025441071193,
+      "learning_rate": 4.2267863306869176e-05,
+      "loss": 0.3759,
+      "step": 1540
+    },
+    {
+      "epoch": 1.439888164026095,
+      "grad_norm": 0.23090799921472505,
+      "learning_rate": 4.222471522264412e-05,
+      "loss": 0.3769,
+      "step": 1545
+    },
+    {
+      "epoch": 1.4445479962721341,
+      "grad_norm": 0.20614180834735402,
+      "learning_rate": 4.2181567138419054e-05,
+      "loss": 0.3743,
+      "step": 1550
+    },
+    {
+      "epoch": 1.4492078285181733,
+      "grad_norm": 0.2263505943709229,
+      "learning_rate": 4.2138419054193996e-05,
+      "loss": 0.3732,
+      "step": 1555
+    },
+    {
+      "epoch": 1.4538676607642125,
+      "grad_norm": 0.2042004636551584,
+      "learning_rate": 4.209527096996894e-05,
+      "loss": 0.3747,
+      "step": 1560
+    },
+    {
+      "epoch": 1.4585274930102516,
+      "grad_norm": 0.2153268967782406,
+      "learning_rate": 4.2052122885743875e-05,
+      "loss": 0.3742,
+      "step": 1565
+    },
+    {
+      "epoch": 1.4631873252562908,
+      "grad_norm": 0.22371872830623132,
+      "learning_rate": 4.200897480151881e-05,
+      "loss": 0.3866,
+      "step": 1570
+    },
+    {
+      "epoch": 1.46784715750233,
+      "grad_norm": 0.26071707581068004,
+      "learning_rate": 4.196582671729375e-05,
+      "loss": 0.3832,
+      "step": 1575
+    },
+    {
+      "epoch": 1.4725069897483691,
+      "grad_norm": 0.20858250838038425,
+      "learning_rate": 4.1922678633068695e-05,
+      "loss": 0.3809,
+      "step": 1580
+    },
+    {
+      "epoch": 1.477166821994408,
+      "grad_norm": 0.20779166311982555,
+      "learning_rate": 4.187953054884363e-05,
+      "loss": 0.3764,
+      "step": 1585
+    },
+    {
+      "epoch": 1.4818266542404475,
+      "grad_norm": 0.20686638103388302,
+      "learning_rate": 4.1836382464618573e-05,
+      "loss": 0.3878,
+      "step": 1590
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 0.2177476212828853,
+      "learning_rate": 4.1793234380393516e-05,
+      "loss": 0.3775,
+      "step": 1595
+    },
+    {
+      "epoch": 1.4911463187325256,
+      "grad_norm": 0.22956941207438086,
+      "learning_rate": 4.175008629616845e-05,
+      "loss": 0.3788,
+      "step": 1600
+    },
+    {
+      "epoch": 1.4958061509785647,
+      "grad_norm": 0.20938925853887305,
+      "learning_rate": 4.1706938211943394e-05,
+      "loss": 0.3838,
+      "step": 1605
+    },
+    {
+      "epoch": 1.500465983224604,
+      "grad_norm": 0.2208691208615957,
+      "learning_rate": 4.166379012771833e-05,
+      "loss": 0.3792,
+      "step": 1610
+    },
+    {
+      "epoch": 1.505125815470643,
+      "grad_norm": 0.26766506324989775,
+      "learning_rate": 4.1620642043493266e-05,
+      "loss": 0.3756,
+      "step": 1615
+    },
+    {
+      "epoch": 1.5097856477166822,
+      "grad_norm": 0.17590780783791074,
+      "learning_rate": 4.157749395926821e-05,
+      "loss": 0.3766,
+      "step": 1620
+    },
+    {
+      "epoch": 1.5144454799627214,
+      "grad_norm": 0.18001665388570595,
+      "learning_rate": 4.153434587504315e-05,
+      "loss": 0.3692,
+      "step": 1625
+    },
+    {
+      "epoch": 1.5191053122087603,
+      "grad_norm": 0.20178084777330935,
+      "learning_rate": 4.1491197790818086e-05,
+      "loss": 0.3629,
+      "step": 1630
+    },
+    {
+      "epoch": 1.5237651444547997,
+      "grad_norm": 0.19825160517671855,
+      "learning_rate": 4.144804970659303e-05,
+      "loss": 0.3715,
+      "step": 1635
+    },
+    {
+      "epoch": 1.5284249767008387,
+      "grad_norm": 0.2014612455557746,
+      "learning_rate": 4.140490162236797e-05,
+      "loss": 0.3749,
+      "step": 1640
+    },
+    {
+      "epoch": 1.533084808946878,
+      "grad_norm": 0.18722738338673067,
+      "learning_rate": 4.1361753538142914e-05,
+      "loss": 0.3917,
+      "step": 1645
+    },
+    {
+      "epoch": 1.537744641192917,
+      "grad_norm": 0.20306745024848047,
+      "learning_rate": 4.131860545391785e-05,
+      "loss": 0.3786,
+      "step": 1650
+    },
+    {
+      "epoch": 1.5424044734389561,
+      "grad_norm": 0.18133323583465574,
+      "learning_rate": 4.1275457369692785e-05,
+      "loss": 0.3857,
+      "step": 1655
+    },
+    {
+      "epoch": 1.5470643056849953,
+      "grad_norm": 0.20076911019219085,
+      "learning_rate": 4.123230928546773e-05,
+      "loss": 0.377,
+      "step": 1660
+    },
+    {
+      "epoch": 1.5517241379310345,
+      "grad_norm": 0.18284857376554714,
+      "learning_rate": 4.1189161201242664e-05,
+      "loss": 0.3682,
+      "step": 1665
+    },
+    {
+      "epoch": 1.5563839701770736,
+      "grad_norm": 0.2179976981974033,
+      "learning_rate": 4.1146013117017606e-05,
+      "loss": 0.3747,
+      "step": 1670
+    },
+    {
+      "epoch": 1.5610438024231128,
+      "grad_norm": 0.22071564131096041,
+      "learning_rate": 4.110286503279255e-05,
+      "loss": 0.3833,
+      "step": 1675
+    },
+    {
+      "epoch": 1.565703634669152,
+      "grad_norm": 0.19999665153685364,
+      "learning_rate": 4.1059716948567484e-05,
+      "loss": 0.3715,
+      "step": 1680
+    },
+    {
+      "epoch": 1.570363466915191,
+      "grad_norm": 0.19710080251830372,
+      "learning_rate": 4.101656886434243e-05,
+      "loss": 0.3712,
+      "step": 1685
+    },
+    {
+      "epoch": 1.5750232991612303,
+      "grad_norm": 0.19636495751773445,
+      "learning_rate": 4.097342078011737e-05,
+      "loss": 0.3834,
+      "step": 1690
+    },
+    {
+      "epoch": 1.5796831314072692,
+      "grad_norm": 0.1899257980794914,
+      "learning_rate": 4.0930272695892305e-05,
+      "loss": 0.3877,
+      "step": 1695
+    },
+    {
+      "epoch": 1.5843429636533086,
+      "grad_norm": 0.21337467837096968,
+      "learning_rate": 4.088712461166724e-05,
+      "loss": 0.3805,
+      "step": 1700
+    },
+    {
+      "epoch": 1.5890027958993476,
+      "grad_norm": 0.2045759256375423,
+      "learning_rate": 4.084397652744218e-05,
+      "loss": 0.3778,
+      "step": 1705
+    },
+    {
+      "epoch": 1.5936626281453867,
+      "grad_norm": 0.21527836655351887,
+      "learning_rate": 4.080082844321712e-05,
+      "loss": 0.3806,
+      "step": 1710
+    },
+    {
+      "epoch": 1.598322460391426,
+      "grad_norm": 0.17287637957366325,
+      "learning_rate": 4.075768035899206e-05,
+      "loss": 0.3775,
+      "step": 1715
+    },
+    {
+      "epoch": 1.602982292637465,
+      "grad_norm": 0.18716859674461575,
+      "learning_rate": 4.0714532274767004e-05,
+      "loss": 0.3777,
+      "step": 1720
+    },
+    {
+      "epoch": 1.6076421248835042,
+      "grad_norm": 0.18453381537381053,
+      "learning_rate": 4.0671384190541946e-05,
+      "loss": 0.377,
+      "step": 1725
+    },
+    {
+      "epoch": 1.6123019571295434,
+      "grad_norm": 0.2136870691876367,
+      "learning_rate": 4.062823610631688e-05,
+      "loss": 0.3697,
+      "step": 1730
+    },
+    {
+      "epoch": 1.6169617893755825,
+      "grad_norm": 0.18258540626594325,
+      "learning_rate": 4.0585088022091825e-05,
+      "loss": 0.384,
+      "step": 1735
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.2054363878853284,
+      "learning_rate": 4.054193993786676e-05,
+      "loss": 0.3776,
+      "step": 1740
+    },
+    {
+      "epoch": 1.6262814538676609,
+      "grad_norm": 0.21068221669581286,
+      "learning_rate": 4.0498791853641696e-05,
+      "loss": 0.3828,
+      "step": 1745
+    },
+    {
+      "epoch": 1.6309412861136998,
+      "grad_norm": 0.22202254062256316,
+      "learning_rate": 4.045564376941664e-05,
+      "loss": 0.379,
+      "step": 1750
+    },
+    {
+      "epoch": 1.6356011183597392,
+      "grad_norm": 0.21994054167730592,
+      "learning_rate": 4.041249568519158e-05,
+      "loss": 0.3735,
+      "step": 1755
+    },
+    {
+      "epoch": 1.6402609506057781,
+      "grad_norm": 0.19967588056411845,
+      "learning_rate": 4.036934760096652e-05,
+      "loss": 0.3812,
+      "step": 1760
+    },
+    {
+      "epoch": 1.6449207828518173,
+      "grad_norm": 0.18564448301179218,
+      "learning_rate": 4.032619951674146e-05,
+      "loss": 0.3859,
+      "step": 1765
+    },
+    {
+      "epoch": 1.6495806150978565,
+      "grad_norm": 0.20235188153279768,
+      "learning_rate": 4.02830514325164e-05,
+      "loss": 0.3746,
+      "step": 1770
+    },
+    {
+      "epoch": 1.6542404473438956,
+      "grad_norm": 0.18087011879737688,
+      "learning_rate": 4.023990334829134e-05,
+      "loss": 0.3666,
+      "step": 1775
+    },
+    {
+      "epoch": 1.6589002795899348,
+      "grad_norm": 0.19564143801128964,
+      "learning_rate": 4.019675526406628e-05,
+      "loss": 0.3791,
+      "step": 1780
+    },
+    {
+      "epoch": 1.6635601118359737,
+      "grad_norm": 0.18367191768487362,
+      "learning_rate": 4.0153607179841216e-05,
+      "loss": 0.3735,
+      "step": 1785
+    },
+    {
+      "epoch": 1.6682199440820131,
+      "grad_norm": 0.18850898947062436,
+      "learning_rate": 4.011045909561615e-05,
+      "loss": 0.374,
+      "step": 1790
+    },
+    {
+      "epoch": 1.672879776328052,
+      "grad_norm": 0.19015348608753815,
+      "learning_rate": 4.0067311011391094e-05,
+      "loss": 0.3678,
+      "step": 1795
+    },
+    {
+      "epoch": 1.6775396085740915,
+      "grad_norm": 0.1812016370420152,
+      "learning_rate": 4.0024162927166037e-05,
+      "loss": 0.3681,
+      "step": 1800
+    },
+    {
+      "epoch": 1.6821994408201304,
+      "grad_norm": 0.18991124262146164,
+      "learning_rate": 3.998101484294098e-05,
+      "loss": 0.3735,
+      "step": 1805
+    },
+    {
+      "epoch": 1.6868592730661698,
+      "grad_norm": 0.19335865383616155,
+      "learning_rate": 3.9937866758715915e-05,
+      "loss": 0.3737,
+      "step": 1810
+    },
+    {
+      "epoch": 1.6915191053122087,
+      "grad_norm": 0.17558578779107908,
+      "learning_rate": 3.989471867449086e-05,
+      "loss": 0.3607,
+      "step": 1815
+    },
+    {
+      "epoch": 1.696178937558248,
+      "grad_norm": 0.17931233189821733,
+      "learning_rate": 3.98515705902658e-05,
+      "loss": 0.3725,
+      "step": 1820
+    },
+    {
+      "epoch": 1.700838769804287,
+      "grad_norm": 0.19542620999701227,
+      "learning_rate": 3.9808422506040736e-05,
+      "loss": 0.3795,
+      "step": 1825
+    },
+    {
+      "epoch": 1.7054986020503262,
+      "grad_norm": 0.1883870749348063,
+      "learning_rate": 3.976527442181567e-05,
+      "loss": 0.3853,
+      "step": 1830
+    },
+    {
+      "epoch": 1.7101584342963654,
+      "grad_norm": 0.2550973674666963,
+      "learning_rate": 3.9722126337590614e-05,
+      "loss": 0.3781,
+      "step": 1835
+    },
+    {
+      "epoch": 1.7148182665424043,
+      "grad_norm": 0.21597478472458898,
+      "learning_rate": 3.967897825336555e-05,
+      "loss": 0.3754,
+      "step": 1840
+    },
+    {
+      "epoch": 1.7194780987884437,
+      "grad_norm": 0.17695936144071578,
+      "learning_rate": 3.963583016914049e-05,
+      "loss": 0.3732,
+      "step": 1845
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 0.18864285350121257,
+      "learning_rate": 3.9592682084915434e-05,
+      "loss": 0.3603,
+      "step": 1850
+    },
+    {
+      "epoch": 1.728797763280522,
+      "grad_norm": 0.1986816272577181,
+      "learning_rate": 3.954953400069037e-05,
+      "loss": 0.37,
+      "step": 1855
+    },
+    {
+      "epoch": 1.733457595526561,
+      "grad_norm": 0.1825186899419743,
+      "learning_rate": 3.950638591646531e-05,
+      "loss": 0.3727,
+      "step": 1860
+    },
+    {
+      "epoch": 1.7381174277726001,
+      "grad_norm": 0.19597110550624666,
+      "learning_rate": 3.9463237832240255e-05,
+      "loss": 0.3722,
+      "step": 1865
+    },
+    {
+      "epoch": 1.7427772600186393,
+      "grad_norm": 0.23066653709167062,
+      "learning_rate": 3.942008974801519e-05,
+      "loss": 0.3727,
+      "step": 1870
+    },
+    {
+      "epoch": 1.7474370922646785,
+      "grad_norm": 0.17570155300517393,
+      "learning_rate": 3.937694166379013e-05,
+      "loss": 0.3795,
+      "step": 1875
+    },
+    {
+      "epoch": 1.7520969245107176,
+      "grad_norm": 0.1854692266655587,
+      "learning_rate": 3.933379357956507e-05,
+      "loss": 0.3759,
+      "step": 1880
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.2093877179591599,
+      "learning_rate": 3.9290645495340005e-05,
+      "loss": 0.3768,
+      "step": 1885
+    },
+    {
+      "epoch": 1.761416589002796,
+      "grad_norm": 0.18504230989179457,
+      "learning_rate": 3.924749741111495e-05,
+      "loss": 0.3756,
+      "step": 1890
+    },
+    {
+      "epoch": 1.766076421248835,
+      "grad_norm": 0.2156876632435152,
+      "learning_rate": 3.920434932688989e-05,
+      "loss": 0.3755,
+      "step": 1895
+    },
+    {
+      "epoch": 1.7707362534948743,
+      "grad_norm": 0.19009588981775974,
+      "learning_rate": 3.916120124266483e-05,
+      "loss": 0.364,
+      "step": 1900
+    },
+    {
+      "epoch": 1.7753960857409132,
+      "grad_norm": 0.1719822595056424,
+      "learning_rate": 3.911805315843977e-05,
+      "loss": 0.3657,
+      "step": 1905
+    },
+    {
+      "epoch": 1.7800559179869526,
+      "grad_norm": 0.19341489396522193,
+      "learning_rate": 3.9074905074214704e-05,
+      "loss": 0.3783,
+      "step": 1910
+    },
+    {
+      "epoch": 1.7847157502329916,
+      "grad_norm": 0.17278073328033802,
+      "learning_rate": 3.9031756989989646e-05,
+      "loss": 0.3612,
+      "step": 1915
+    },
+    {
+      "epoch": 1.7893755824790307,
+      "grad_norm": 0.18445457576270444,
+      "learning_rate": 3.898860890576458e-05,
+      "loss": 0.3809,
+      "step": 1920
+    },
+    {
+      "epoch": 1.7940354147250699,
+      "grad_norm": 0.17669353474609195,
+      "learning_rate": 3.8945460821539525e-05,
+      "loss": 0.3734,
+      "step": 1925
+    },
+    {
+      "epoch": 1.798695246971109,
+      "grad_norm": 0.19819377030199534,
+      "learning_rate": 3.890231273731447e-05,
+      "loss": 0.3709,
+      "step": 1930
+    },
+    {
+      "epoch": 1.8033550792171482,
+      "grad_norm": 0.19916482341507918,
+      "learning_rate": 3.88591646530894e-05,
+      "loss": 0.3761,
+      "step": 1935
+    },
+    {
+      "epoch": 1.8080149114631874,
+      "grad_norm": 0.19018430374495865,
+      "learning_rate": 3.8816016568864345e-05,
+      "loss": 0.3627,
+      "step": 1940
+    },
+    {
+      "epoch": 1.8126747437092265,
+      "grad_norm": 0.20734412941959443,
+      "learning_rate": 3.877286848463929e-05,
+      "loss": 0.3807,
+      "step": 1945
+    },
+    {
+      "epoch": 1.8173345759552655,
+      "grad_norm": 0.17260754747628748,
+      "learning_rate": 3.8729720400414224e-05,
+      "loss": 0.3632,
+      "step": 1950
+    },
+    {
+      "epoch": 1.8219944082013049,
+      "grad_norm": 0.2014756770864653,
+      "learning_rate": 3.868657231618916e-05,
+      "loss": 0.369,
+      "step": 1955
+    },
+    {
+      "epoch": 1.8266542404473438,
+      "grad_norm": 0.1783424454799133,
+      "learning_rate": 3.86434242319641e-05,
+      "loss": 0.3722,
+      "step": 1960
+    },
+    {
+      "epoch": 1.8313140726933832,
+      "grad_norm": 0.1995943707626634,
+      "learning_rate": 3.860027614773904e-05,
+      "loss": 0.3761,
+      "step": 1965
+    },
+    {
+      "epoch": 1.8359739049394221,
+      "grad_norm": 0.188888953762579,
+      "learning_rate": 3.855712806351398e-05,
+      "loss": 0.3738,
+      "step": 1970
+    },
+    {
+      "epoch": 1.8406337371854613,
+      "grad_norm": 0.196683058381539,
+      "learning_rate": 3.851397997928892e-05,
+      "loss": 0.3822,
+      "step": 1975
+    },
+    {
+      "epoch": 1.8452935694315005,
+      "grad_norm": 0.17779651274065691,
+      "learning_rate": 3.8470831895063865e-05,
+      "loss": 0.3709,
+      "step": 1980
+    },
+    {
+      "epoch": 1.8499534016775396,
+      "grad_norm": 0.18729601550966157,
+      "learning_rate": 3.84276838108388e-05,
+      "loss": 0.3698,
+      "step": 1985
+    },
+    {
+      "epoch": 1.8546132339235788,
+      "grad_norm": 0.18868182095331562,
+      "learning_rate": 3.838453572661374e-05,
+      "loss": 0.387,
+      "step": 1990
+    },
+    {
+      "epoch": 1.8592730661696177,
+      "grad_norm": 0.18913662887440688,
+      "learning_rate": 3.834138764238868e-05,
+      "loss": 0.372,
+      "step": 1995
+    },
+    {
+      "epoch": 1.8639328984156571,
+      "grad_norm": 0.18575167715922308,
+      "learning_rate": 3.8298239558163615e-05,
+      "loss": 0.3715,
+      "step": 2000
+    },
+    {
+      "epoch": 1.868592730661696,
+      "grad_norm": 0.18723732068649998,
+      "learning_rate": 3.825509147393856e-05,
+      "loss": 0.3652,
+      "step": 2005
+    },
+    {
+      "epoch": 1.8732525629077355,
+      "grad_norm": 0.1796141336345043,
+      "learning_rate": 3.82119433897135e-05,
+      "loss": 0.3816,
+      "step": 2010
+    },
+    {
+      "epoch": 1.8779123951537744,
+      "grad_norm": 0.18352436775237538,
+      "learning_rate": 3.8168795305488435e-05,
+      "loss": 0.3653,
+      "step": 2015
+    },
+    {
+      "epoch": 1.8825722273998136,
+      "grad_norm": 0.19993902338434422,
+      "learning_rate": 3.812564722126338e-05,
+      "loss": 0.3631,
+      "step": 2020
+    },
+    {
+      "epoch": 1.8872320596458527,
+      "grad_norm": 0.1924271424950404,
+      "learning_rate": 3.808249913703832e-05,
+      "loss": 0.3675,
+      "step": 2025
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.18336319884702126,
+      "learning_rate": 3.8039351052813256e-05,
+      "loss": 0.3753,
+      "step": 2030
+    },
+    {
+      "epoch": 1.896551724137931,
+      "grad_norm": 0.1852822757449342,
+      "learning_rate": 3.79962029685882e-05,
+      "loss": 0.3665,
+      "step": 2035
+    },
+    {
+      "epoch": 1.9012115563839702,
+      "grad_norm": 0.18017930014053968,
+      "learning_rate": 3.7953054884363134e-05,
+      "loss": 0.3725,
+      "step": 2040
+    },
+    {
+      "epoch": 1.9058713886300094,
+      "grad_norm": 0.18821891623969317,
+      "learning_rate": 3.790990680013807e-05,
+      "loss": 0.3884,
+      "step": 2045
+    },
+    {
+      "epoch": 1.9105312208760483,
+      "grad_norm": 0.19184571394580444,
+      "learning_rate": 3.786675871591301e-05,
+      "loss": 0.3779,
+      "step": 2050
+    },
+    {
+      "epoch": 1.9151910531220877,
+      "grad_norm": 0.19111696432125175,
+      "learning_rate": 3.7823610631687955e-05,
+      "loss": 0.372,
+      "step": 2055
+    },
+    {
+      "epoch": 1.9198508853681266,
+      "grad_norm": 0.1871965423980423,
+      "learning_rate": 3.77804625474629e-05,
+      "loss": 0.3799,
+      "step": 2060
+    },
+    {
+      "epoch": 1.924510717614166,
+      "grad_norm": 0.16969140230119115,
+      "learning_rate": 3.773731446323783e-05,
+      "loss": 0.3676,
+      "step": 2065
+    },
+    {
+      "epoch": 1.929170549860205,
+      "grad_norm": 0.18088263813145852,
+      "learning_rate": 3.7694166379012776e-05,
+      "loss": 0.3773,
+      "step": 2070
+    },
+    {
+      "epoch": 1.9338303821062441,
+      "grad_norm": 0.19367671214925725,
+      "learning_rate": 3.765101829478772e-05,
+      "loss": 0.3761,
+      "step": 2075
+    },
+    {
+      "epoch": 1.9384902143522833,
+      "grad_norm": 0.1848076689446235,
+      "learning_rate": 3.7607870210562654e-05,
+      "loss": 0.3758,
+      "step": 2080
+    },
+    {
+      "epoch": 1.9431500465983225,
+      "grad_norm": 0.21383054955581582,
+      "learning_rate": 3.756472212633759e-05,
+      "loss": 0.3746,
+      "step": 2085
+    },
+    {
+      "epoch": 1.9478098788443616,
+      "grad_norm": 0.19430030248894598,
+      "learning_rate": 3.752157404211253e-05,
+      "loss": 0.3792,
+      "step": 2090
+    },
+    {
+      "epoch": 1.9524697110904008,
+      "grad_norm": 0.21773126400538895,
+      "learning_rate": 3.747842595788747e-05,
+      "loss": 0.3622,
+      "step": 2095
+    },
+    {
+      "epoch": 1.95712954333644,
+      "grad_norm": 0.19433345670714272,
+      "learning_rate": 3.743527787366241e-05,
+      "loss": 0.378,
+      "step": 2100
+    },
+    {
+      "epoch": 1.961789375582479,
+      "grad_norm": 0.19455093963968728,
+      "learning_rate": 3.739212978943735e-05,
+      "loss": 0.3709,
+      "step": 2105
+    },
+    {
+      "epoch": 1.9664492078285183,
+      "grad_norm": 0.18620135684303868,
+      "learning_rate": 3.734898170521229e-05,
+      "loss": 0.3792,
+      "step": 2110
+    },
+    {
+      "epoch": 1.9711090400745572,
+      "grad_norm": 0.19855983718748652,
+      "learning_rate": 3.730583362098723e-05,
+      "loss": 0.3842,
+      "step": 2115
+    },
+    {
+      "epoch": 1.9757688723205966,
+      "grad_norm": 0.2121174010445428,
+      "learning_rate": 3.7262685536762174e-05,
+      "loss": 0.3704,
+      "step": 2120
+    },
+    {
+      "epoch": 1.9804287045666356,
+      "grad_norm": 0.17845134105199764,
+      "learning_rate": 3.721953745253711e-05,
+      "loss": 0.3734,
+      "step": 2125
+    },
+    {
+      "epoch": 1.9850885368126747,
+      "grad_norm": 0.18567962692066017,
+      "learning_rate": 3.7176389368312045e-05,
+      "loss": 0.3769,
+      "step": 2130
+    },
+    {
+      "epoch": 1.9897483690587139,
+      "grad_norm": 0.1810250879265219,
+      "learning_rate": 3.713324128408699e-05,
+      "loss": 0.3635,
+      "step": 2135
+    },
+    {
+      "epoch": 1.994408201304753,
+      "grad_norm": 0.183290868705745,
+      "learning_rate": 3.709009319986193e-05,
+      "loss": 0.3703,
+      "step": 2140
+    },
+    {
+      "epoch": 1.9990680335507922,
+      "grad_norm": 0.19107933751902184,
+      "learning_rate": 3.7046945115636866e-05,
+      "loss": 0.3751,
+      "step": 2145
+    },
+    {
+      "epoch": 2.003727865796831,
+      "grad_norm": 0.2625735473639505,
+      "learning_rate": 3.700379703141181e-05,
+      "loss": 0.3115,
+      "step": 2150
+    },
+    {
+      "epoch": 2.0083876980428705,
+      "grad_norm": 0.22532373463177166,
+      "learning_rate": 3.696064894718675e-05,
+      "loss": 0.3167,
+      "step": 2155
+    },
+    {
+      "epoch": 2.0130475302889095,
+      "grad_norm": 0.20281814646651494,
+      "learning_rate": 3.6917500862961687e-05,
+      "loss": 0.3054,
+      "step": 2160
+    },
+    {
+      "epoch": 2.017707362534949,
+      "grad_norm": 0.20627186916369916,
+      "learning_rate": 3.687435277873663e-05,
+      "loss": 0.3108,
+      "step": 2165
+    },
+    {
+      "epoch": 2.022367194780988,
+      "grad_norm": 0.18934751924967644,
+      "learning_rate": 3.6831204694511565e-05,
+      "loss": 0.3104,
+      "step": 2170
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 0.16946500530050818,
+      "learning_rate": 3.67880566102865e-05,
+      "loss": 0.3047,
+      "step": 2175
+    },
+    {
+      "epoch": 2.031686859273066,
+      "grad_norm": 0.18518101829907888,
+      "learning_rate": 3.674490852606144e-05,
+      "loss": 0.3081,
+      "step": 2180
+    },
+    {
+      "epoch": 2.0363466915191055,
+      "grad_norm": 0.1730449176101433,
+      "learning_rate": 3.6701760441836386e-05,
+      "loss": 0.306,
+      "step": 2185
+    },
+    {
+      "epoch": 2.0410065237651445,
+      "grad_norm": 0.18312753641847493,
+      "learning_rate": 3.665861235761132e-05,
+      "loss": 0.3055,
+      "step": 2190
+    },
+    {
+      "epoch": 2.0456663560111834,
+      "grad_norm": 0.18738857958545455,
+      "learning_rate": 3.6615464273386264e-05,
+      "loss": 0.3147,
+      "step": 2195
+    },
+    {
+      "epoch": 2.050326188257223,
+      "grad_norm": 0.1707933461498227,
+      "learning_rate": 3.6572316189161206e-05,
+      "loss": 0.3145,
+      "step": 2200
+    },
+    {
+      "epoch": 2.0549860205032617,
+      "grad_norm": 0.17824648371749702,
+      "learning_rate": 3.652916810493614e-05,
+      "loss": 0.3049,
+      "step": 2205
+    },
+    {
+      "epoch": 2.059645852749301,
+      "grad_norm": 0.18122903793667575,
+      "learning_rate": 3.6486020020711085e-05,
+      "loss": 0.3044,
+      "step": 2210
+    },
+    {
+      "epoch": 2.06430568499534,
+      "grad_norm": 0.2101581052454426,
+      "learning_rate": 3.644287193648602e-05,
+      "loss": 0.3168,
+      "step": 2215
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 0.17308640872418218,
+      "learning_rate": 3.639972385226096e-05,
+      "loss": 0.3004,
+      "step": 2220
+    },
+    {
+      "epoch": 2.0736253494874184,
+      "grad_norm": 0.18707622699563384,
+      "learning_rate": 3.63565757680359e-05,
+      "loss": 0.3013,
+      "step": 2225
+    },
+    {
+      "epoch": 2.0782851817334578,
+      "grad_norm": 0.17409669082399934,
+      "learning_rate": 3.631342768381084e-05,
+      "loss": 0.3015,
+      "step": 2230
+    },
+    {
+      "epoch": 2.0829450139794967,
+      "grad_norm": 0.1793991552841305,
+      "learning_rate": 3.6270279599585783e-05,
+      "loss": 0.3011,
+      "step": 2235
+    },
+    {
+      "epoch": 2.0876048462255357,
+      "grad_norm": 0.19265026750389214,
+      "learning_rate": 3.622713151536072e-05,
+      "loss": 0.3183,
+      "step": 2240
+    },
+    {
+      "epoch": 2.092264678471575,
+      "grad_norm": 0.18567668059408807,
+      "learning_rate": 3.618398343113566e-05,
+      "loss": 0.3056,
+      "step": 2245
+    },
+    {
+      "epoch": 2.096924510717614,
+      "grad_norm": 0.18873582886595616,
+      "learning_rate": 3.6140835346910604e-05,
+      "loss": 0.3109,
+      "step": 2250
+    },
+    {
+      "epoch": 2.1015843429636534,
+      "grad_norm": 0.19384445755488322,
+      "learning_rate": 3.609768726268554e-05,
+      "loss": 0.3172,
+      "step": 2255
+    },
+    {
+      "epoch": 2.1062441752096923,
+      "grad_norm": 0.15758725651128513,
+      "learning_rate": 3.6054539178460476e-05,
+      "loss": 0.3093,
+      "step": 2260
+    },
+    {
+      "epoch": 2.1109040074557317,
+      "grad_norm": 0.17166613383236495,
+      "learning_rate": 3.601139109423542e-05,
+      "loss": 0.3102,
+      "step": 2265
+    },
+    {
+      "epoch": 2.1155638397017706,
+      "grad_norm": 0.1878108313147635,
+      "learning_rate": 3.5968243010010354e-05,
+      "loss": 0.3153,
+      "step": 2270
+    },
+    {
+      "epoch": 2.12022367194781,
+      "grad_norm": 0.16283679916861268,
+      "learning_rate": 3.5925094925785296e-05,
+      "loss": 0.3052,
+      "step": 2275
+    },
+    {
+      "epoch": 2.124883504193849,
+      "grad_norm": 0.20627248846312002,
+      "learning_rate": 3.588194684156024e-05,
+      "loss": 0.3147,
+      "step": 2280
+    },
+    {
+      "epoch": 2.1295433364398884,
+      "grad_norm": 0.17074096203713113,
+      "learning_rate": 3.5838798757335175e-05,
+      "loss": 0.3046,
+      "step": 2285
+    },
+    {
+      "epoch": 2.1342031686859273,
+      "grad_norm": 0.16135171632176104,
+      "learning_rate": 3.579565067311012e-05,
+      "loss": 0.3081,
+      "step": 2290
+    },
+    {
+      "epoch": 2.1388630009319662,
+      "grad_norm": 0.1856809267678432,
+      "learning_rate": 3.575250258888505e-05,
+      "loss": 0.3096,
+      "step": 2295
+    },
+    {
+      "epoch": 2.1435228331780056,
+      "grad_norm": 0.1934047761818321,
+      "learning_rate": 3.5709354504659995e-05,
+      "loss": 0.3121,
+      "step": 2300
+    },
+    {
+      "epoch": 2.1481826654240446,
+      "grad_norm": 0.1805885337281842,
+      "learning_rate": 3.566620642043493e-05,
+      "loss": 0.3047,
+      "step": 2305
+    },
+    {
+      "epoch": 2.152842497670084,
+      "grad_norm": 0.16562579241179268,
+      "learning_rate": 3.5623058336209874e-05,
+      "loss": 0.3102,
+      "step": 2310
+    },
+    {
+      "epoch": 2.157502329916123,
+      "grad_norm": 0.2024472721177785,
+      "learning_rate": 3.5579910251984816e-05,
+      "loss": 0.311,
+      "step": 2315
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.16667079963583561,
+      "learning_rate": 3.553676216775975e-05,
+      "loss": 0.3037,
+      "step": 2320
+    },
+    {
+      "epoch": 2.1668219944082012,
+      "grad_norm": 0.19642400131287788,
+      "learning_rate": 3.5493614083534694e-05,
+      "loss": 0.3151,
+      "step": 2325
+    },
+    {
+      "epoch": 2.1714818266542406,
+      "grad_norm": 0.18198095875835987,
+      "learning_rate": 3.545046599930964e-05,
+      "loss": 0.3056,
+      "step": 2330
+    },
+    {
+      "epoch": 2.1761416589002796,
+      "grad_norm": 0.1584822823429212,
+      "learning_rate": 3.540731791508457e-05,
+      "loss": 0.3087,
+      "step": 2335
+    },
+    {
+      "epoch": 2.180801491146319,
+      "grad_norm": 0.1711859889942154,
+      "learning_rate": 3.536416983085951e-05,
+      "loss": 0.3046,
+      "step": 2340
+    },
+    {
+      "epoch": 2.185461323392358,
+      "grad_norm": 0.1736698740621732,
+      "learning_rate": 3.532102174663445e-05,
+      "loss": 0.3194,
+      "step": 2345
+    },
+    {
+      "epoch": 2.190121155638397,
+      "grad_norm": 0.1719462935072656,
+      "learning_rate": 3.5277873662409386e-05,
+      "loss": 0.2986,
+      "step": 2350
+    },
+    {
+      "epoch": 2.194780987884436,
+      "grad_norm": 0.18762433756276262,
+      "learning_rate": 3.523472557818433e-05,
+      "loss": 0.3128,
+      "step": 2355
+    },
+    {
+      "epoch": 2.199440820130475,
+      "grad_norm": 0.17925578450983062,
+      "learning_rate": 3.519157749395927e-05,
+      "loss": 0.3147,
+      "step": 2360
+    },
+    {
+      "epoch": 2.2041006523765145,
+      "grad_norm": 0.17796876149755342,
+      "learning_rate": 3.514842940973421e-05,
+      "loss": 0.3033,
+      "step": 2365
+    },
+    {
+      "epoch": 2.2087604846225535,
+      "grad_norm": 0.16480169661875405,
+      "learning_rate": 3.510528132550915e-05,
+      "loss": 0.2991,
+      "step": 2370
+    },
+    {
+      "epoch": 2.213420316868593,
+      "grad_norm": 0.17037691006845565,
+      "learning_rate": 3.506213324128409e-05,
+      "loss": 0.3045,
+      "step": 2375
+    },
+    {
+      "epoch": 2.218080149114632,
+      "grad_norm": 0.16290876632422666,
+      "learning_rate": 3.501898515705903e-05,
+      "loss": 0.3133,
+      "step": 2380
+    },
+    {
+      "epoch": 2.222739981360671,
+      "grad_norm": 0.16316178782217536,
+      "learning_rate": 3.4975837072833964e-05,
+      "loss": 0.3018,
+      "step": 2385
+    },
+    {
+      "epoch": 2.22739981360671,
+      "grad_norm": 0.19291921983267185,
+      "learning_rate": 3.4932688988608906e-05,
+      "loss": 0.3075,
+      "step": 2390
+    },
+    {
+      "epoch": 2.2320596458527495,
+      "grad_norm": 0.1822186185456547,
+      "learning_rate": 3.488954090438385e-05,
+      "loss": 0.3133,
+      "step": 2395
+    },
+    {
+      "epoch": 2.2367194780987885,
+      "grad_norm": 0.1684958463720547,
+      "learning_rate": 3.4846392820158784e-05,
+      "loss": 0.3081,
+      "step": 2400
+    },
+    {
+      "epoch": 2.2413793103448274,
+      "grad_norm": 0.17034587488940908,
+      "learning_rate": 3.480324473593373e-05,
+      "loss": 0.297,
+      "step": 2405
+    },
+    {
+      "epoch": 2.246039142590867,
+      "grad_norm": 0.21549412935143467,
+      "learning_rate": 3.476009665170867e-05,
+      "loss": 0.3173,
+      "step": 2410
+    },
+    {
+      "epoch": 2.2506989748369057,
+      "grad_norm": 0.1693480215288796,
+      "learning_rate": 3.4716948567483605e-05,
+      "loss": 0.3132,
+      "step": 2415
+    },
+    {
+      "epoch": 2.255358807082945,
+      "grad_norm": 0.18110530139120207,
+      "learning_rate": 3.467380048325855e-05,
+      "loss": 0.3102,
+      "step": 2420
+    },
+    {
+      "epoch": 2.260018639328984,
+      "grad_norm": 0.18357998552845683,
+      "learning_rate": 3.463065239903348e-05,
+      "loss": 0.3128,
+      "step": 2425
+    },
+    {
+      "epoch": 2.2646784715750234,
+      "grad_norm": 0.16853744676196655,
+      "learning_rate": 3.458750431480842e-05,
+      "loss": 0.3073,
+      "step": 2430
+    },
+    {
+      "epoch": 2.2693383038210624,
+      "grad_norm": 0.1945953054088613,
+      "learning_rate": 3.454435623058336e-05,
+      "loss": 0.3099,
+      "step": 2435
+    },
+    {
+      "epoch": 2.2739981360671018,
+      "grad_norm": 0.1752903682443434,
+      "learning_rate": 3.4501208146358304e-05,
+      "loss": 0.311,
+      "step": 2440
+    },
+    {
+      "epoch": 2.2786579683131407,
+      "grad_norm": 0.18566276292750983,
+      "learning_rate": 3.445806006213324e-05,
+      "loss": 0.3157,
+      "step": 2445
+    },
+    {
+      "epoch": 2.28331780055918,
+      "grad_norm": 0.19760147925238,
+      "learning_rate": 3.441491197790818e-05,
+      "loss": 0.3095,
+      "step": 2450
+    },
+    {
+      "epoch": 2.287977632805219,
+      "grad_norm": 0.19339458116046718,
+      "learning_rate": 3.4371763893683125e-05,
+      "loss": 0.3119,
+      "step": 2455
+    },
+    {
+      "epoch": 2.292637465051258,
+      "grad_norm": 0.1771775343782336,
+      "learning_rate": 3.432861580945806e-05,
+      "loss": 0.3124,
+      "step": 2460
+    },
+    {
+      "epoch": 2.2972972972972974,
+      "grad_norm": 0.17324128399235456,
+      "learning_rate": 3.4285467725233e-05,
+      "loss": 0.3073,
+      "step": 2465
+    },
+    {
+      "epoch": 2.3019571295433363,
+      "grad_norm": 0.18123119343034538,
+      "learning_rate": 3.424231964100794e-05,
+      "loss": 0.3095,
+      "step": 2470
+    },
+    {
+      "epoch": 2.3066169617893757,
+      "grad_norm": 0.17597535974315281,
+      "learning_rate": 3.419917155678288e-05,
+      "loss": 0.3126,
+      "step": 2475
+    },
+    {
+      "epoch": 2.3112767940354146,
+      "grad_norm": 0.1740493113166357,
+      "learning_rate": 3.415602347255782e-05,
+      "loss": 0.3042,
+      "step": 2480
+    },
+    {
+      "epoch": 2.315936626281454,
+      "grad_norm": 0.17404391232385405,
+      "learning_rate": 3.411287538833276e-05,
+      "loss": 0.3108,
+      "step": 2485
+    },
+    {
+      "epoch": 2.320596458527493,
+      "grad_norm": 0.17278804925207308,
+      "learning_rate": 3.40697273041077e-05,
+      "loss": 0.3067,
+      "step": 2490
+    },
+    {
+      "epoch": 2.325256290773532,
+      "grad_norm": 0.1794733807567191,
+      "learning_rate": 3.402657921988264e-05,
+      "loss": 0.3232,
+      "step": 2495
+    },
+    {
+      "epoch": 2.3299161230195713,
+      "grad_norm": 0.18094813481009225,
+      "learning_rate": 3.398343113565758e-05,
+      "loss": 0.3078,
+      "step": 2500
+    },
+    {
+      "epoch": 2.3345759552656107,
+      "grad_norm": 0.1960174749220055,
+      "learning_rate": 3.394028305143252e-05,
+      "loss": 0.3117,
+      "step": 2505
+    },
+    {
+      "epoch": 2.3392357875116496,
+      "grad_norm": 0.1686396435263158,
+      "learning_rate": 3.389713496720746e-05,
+      "loss": 0.3199,
+      "step": 2510
+    },
+    {
+      "epoch": 2.3438956197576886,
+      "grad_norm": 0.1767534970735736,
+      "learning_rate": 3.3853986882982394e-05,
+      "loss": 0.3153,
+      "step": 2515
+    },
+    {
+      "epoch": 2.348555452003728,
+      "grad_norm": 0.17281536228238867,
+      "learning_rate": 3.381083879875734e-05,
+      "loss": 0.3133,
+      "step": 2520
+    },
+    {
+      "epoch": 2.353215284249767,
+      "grad_norm": 0.1851314788857115,
+      "learning_rate": 3.376769071453227e-05,
+      "loss": 0.3085,
+      "step": 2525
+    },
+    {
+      "epoch": 2.3578751164958063,
+      "grad_norm": 0.18301194584780625,
+      "learning_rate": 3.3724542630307215e-05,
+      "loss": 0.3153,
+      "step": 2530
+    },
+    {
+      "epoch": 2.362534948741845,
+      "grad_norm": 0.17476176822004574,
+      "learning_rate": 3.368139454608216e-05,
+      "loss": 0.3171,
+      "step": 2535
+    },
+    {
+      "epoch": 2.3671947809878846,
+      "grad_norm": 0.18610366000260867,
+      "learning_rate": 3.363824646185709e-05,
+      "loss": 0.3029,
+      "step": 2540
+    },
+    {
+      "epoch": 2.3718546132339235,
+      "grad_norm": 0.16481555194983327,
+      "learning_rate": 3.3595098377632036e-05,
+      "loss": 0.3165,
+      "step": 2545
+    },
+    {
+      "epoch": 2.3765144454799625,
+      "grad_norm": 0.1814569333200653,
+      "learning_rate": 3.355195029340698e-05,
+      "loss": 0.3025,
+      "step": 2550
+    },
+    {
+      "epoch": 2.381174277726002,
+      "grad_norm": 0.1791080922541904,
+      "learning_rate": 3.3508802209181914e-05,
+      "loss": 0.3048,
+      "step": 2555
+    },
+    {
+      "epoch": 2.385834109972041,
+      "grad_norm": 0.1624359518087968,
+      "learning_rate": 3.346565412495685e-05,
+      "loss": 0.3096,
+      "step": 2560
+    },
+    {
+      "epoch": 2.39049394221808,
+      "grad_norm": 0.18363118970792072,
+      "learning_rate": 3.342250604073179e-05,
+      "loss": 0.3244,
+      "step": 2565
+    },
+    {
+      "epoch": 2.395153774464119,
+      "grad_norm": 0.18594811314031923,
+      "learning_rate": 3.3379357956506735e-05,
+      "loss": 0.3154,
+      "step": 2570
+    },
+    {
+      "epoch": 2.3998136067101585,
+      "grad_norm": 0.16528576056676783,
+      "learning_rate": 3.333620987228167e-05,
+      "loss": 0.3089,
+      "step": 2575
+    },
+    {
+      "epoch": 2.4044734389561975,
+      "grad_norm": 0.16226338426475218,
+      "learning_rate": 3.329306178805661e-05,
+      "loss": 0.3125,
+      "step": 2580
+    },
+    {
+      "epoch": 2.409133271202237,
+      "grad_norm": 0.1636399119668061,
+      "learning_rate": 3.3249913703831555e-05,
+      "loss": 0.3263,
+      "step": 2585
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 0.17084314544691023,
+      "learning_rate": 3.320676561960649e-05,
+      "loss": 0.3074,
+      "step": 2590
+    },
+    {
+      "epoch": 2.418452935694315,
+      "grad_norm": 0.1761880956312914,
+      "learning_rate": 3.3163617535381434e-05,
+      "loss": 0.3132,
+      "step": 2595
+    },
+    {
+      "epoch": 2.423112767940354,
+      "grad_norm": 0.17485896375984325,
+      "learning_rate": 3.312046945115637e-05,
+      "loss": 0.3135,
+      "step": 2600
+    },
+    {
+      "epoch": 2.427772600186393,
+      "grad_norm": 0.17351782499400142,
+      "learning_rate": 3.3077321366931305e-05,
+      "loss": 0.3177,
+      "step": 2605
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 0.16809046960458454,
+      "learning_rate": 3.303417328270625e-05,
+      "loss": 0.3099,
+      "step": 2610
+    },
+    {
+      "epoch": 2.4370922646784714,
+      "grad_norm": 0.18227169849451783,
+      "learning_rate": 3.299102519848119e-05,
+      "loss": 0.3072,
+      "step": 2615
+    },
+    {
+      "epoch": 2.441752096924511,
+      "grad_norm": 0.17819378298295488,
+      "learning_rate": 3.2947877114256126e-05,
+      "loss": 0.3114,
+      "step": 2620
+    },
+    {
+      "epoch": 2.4464119291705497,
+      "grad_norm": 0.21309510415594726,
+      "learning_rate": 3.290472903003107e-05,
+      "loss": 0.3105,
+      "step": 2625
+    },
+    {
+      "epoch": 2.451071761416589,
+      "grad_norm": 0.17597163145791375,
+      "learning_rate": 3.286158094580601e-05,
+      "loss": 0.3077,
+      "step": 2630
+    },
+    {
+      "epoch": 2.455731593662628,
+      "grad_norm": 0.17594324137864278,
+      "learning_rate": 3.281843286158095e-05,
+      "loss": 0.3052,
+      "step": 2635
+    },
+    {
+      "epoch": 2.4603914259086674,
+      "grad_norm": 0.18409143948602258,
+      "learning_rate": 3.277528477735589e-05,
+      "loss": 0.3085,
+      "step": 2640
+    },
+    {
+      "epoch": 2.4650512581547064,
+      "grad_norm": 0.15552257302687467,
+      "learning_rate": 3.2732136693130825e-05,
+      "loss": 0.3033,
+      "step": 2645
+    },
+    {
+      "epoch": 2.4697110904007458,
+      "grad_norm": 0.18020955785312615,
+      "learning_rate": 3.268898860890577e-05,
+      "loss": 0.3211,
+      "step": 2650
+    },
+    {
+      "epoch": 2.4743709226467847,
+      "grad_norm": 0.18401127408829257,
+      "learning_rate": 3.26458405246807e-05,
+      "loss": 0.3138,
+      "step": 2655
+    },
+    {
+      "epoch": 2.4790307548928237,
+      "grad_norm": 0.1749490933206319,
+      "learning_rate": 3.2602692440455645e-05,
+      "loss": 0.3094,
+      "step": 2660
+    },
+    {
+      "epoch": 2.483690587138863,
+      "grad_norm": 0.16129358334342486,
+      "learning_rate": 3.255954435623059e-05,
+      "loss": 0.3123,
+      "step": 2665
+    },
+    {
+      "epoch": 2.488350419384902,
+      "grad_norm": 0.17111413396682817,
+      "learning_rate": 3.2516396272005524e-05,
+      "loss": 0.3105,
+      "step": 2670
+    },
+    {
+      "epoch": 2.4930102516309414,
+      "grad_norm": 0.17906516637232242,
+      "learning_rate": 3.2473248187780466e-05,
+      "loss": 0.318,
+      "step": 2675
+    },
+    {
+      "epoch": 2.4976700838769803,
+      "grad_norm": 0.195800861139648,
+      "learning_rate": 3.243010010355541e-05,
+      "loss": 0.3143,
+      "step": 2680
+    },
+    {
+      "epoch": 2.5023299161230197,
+      "grad_norm": 0.1838119033143121,
+      "learning_rate": 3.2386952019330344e-05,
+      "loss": 0.3117,
+      "step": 2685
+    },
+    {
+      "epoch": 2.5069897483690586,
+      "grad_norm": 0.17404109762885403,
+      "learning_rate": 3.234380393510528e-05,
+      "loss": 0.3209,
+      "step": 2690
+    },
+    {
+      "epoch": 2.511649580615098,
+      "grad_norm": 0.17205726782500166,
+      "learning_rate": 3.230065585088022e-05,
+      "loss": 0.3126,
+      "step": 2695
+    },
+    {
+      "epoch": 2.516309412861137,
+      "grad_norm": 0.17145358810821057,
+      "learning_rate": 3.225750776665516e-05,
+      "loss": 0.3077,
+      "step": 2700
+    },
+    {
+      "epoch": 2.5209692451071763,
+      "grad_norm": 0.19171331809488926,
+      "learning_rate": 3.22143596824301e-05,
+      "loss": 0.3141,
+      "step": 2705
+    },
+    {
+      "epoch": 2.5256290773532153,
+      "grad_norm": 0.1801887999215754,
+      "learning_rate": 3.217121159820504e-05,
+      "loss": 0.3132,
+      "step": 2710
+    },
+    {
+      "epoch": 2.5302889095992542,
+      "grad_norm": 0.18951827876093824,
+      "learning_rate": 3.2128063513979986e-05,
+      "loss": 0.3117,
+      "step": 2715
+    },
+    {
+      "epoch": 2.5349487418452936,
+      "grad_norm": 0.16971943180505167,
+      "learning_rate": 3.208491542975492e-05,
+      "loss": 0.3178,
+      "step": 2720
+    },
+    {
+      "epoch": 2.5396085740913326,
+      "grad_norm": 0.1827754799353055,
+      "learning_rate": 3.204176734552986e-05,
+      "loss": 0.3032,
+      "step": 2725
+    },
+    {
+      "epoch": 2.544268406337372,
+      "grad_norm": 0.16515965792510373,
+      "learning_rate": 3.19986192613048e-05,
+      "loss": 0.3056,
+      "step": 2730
+    },
+    {
+      "epoch": 2.548928238583411,
+      "grad_norm": 0.15504721082936385,
+      "learning_rate": 3.1955471177079735e-05,
+      "loss": 0.311,
+      "step": 2735
+    },
+    {
+      "epoch": 2.5535880708294503,
+      "grad_norm": 0.1581520881369321,
+      "learning_rate": 3.191232309285468e-05,
+      "loss": 0.3056,
+      "step": 2740
+    },
+    {
+      "epoch": 2.558247903075489,
+      "grad_norm": 0.17428528300028548,
+      "learning_rate": 3.186917500862962e-05,
+      "loss": 0.3112,
+      "step": 2745
+    },
+    {
+      "epoch": 2.562907735321528,
+      "grad_norm": 0.16979995669094822,
+      "learning_rate": 3.1826026924404556e-05,
+      "loss": 0.3129,
+      "step": 2750
+    },
+    {
+      "epoch": 2.5675675675675675,
+      "grad_norm": 0.16630597611331704,
+      "learning_rate": 3.17828788401795e-05,
+      "loss": 0.3093,
+      "step": 2755
+    },
+    {
+      "epoch": 2.572227399813607,
+      "grad_norm": 0.1830994032842025,
+      "learning_rate": 3.173973075595444e-05,
+      "loss": 0.3167,
+      "step": 2760
+    },
+    {
+      "epoch": 2.576887232059646,
+      "grad_norm": 0.17918939309089452,
+      "learning_rate": 3.169658267172938e-05,
+      "loss": 0.3184,
+      "step": 2765
+    },
+    {
+      "epoch": 2.581547064305685,
+      "grad_norm": 0.16348536461345298,
+      "learning_rate": 3.165343458750431e-05,
+      "loss": 0.3052,
+      "step": 2770
+    },
+    {
+      "epoch": 2.586206896551724,
+      "grad_norm": 0.1872898581886334,
+      "learning_rate": 3.1610286503279255e-05,
+      "loss": 0.3079,
+      "step": 2775
+    },
+    {
+      "epoch": 2.590866728797763,
+      "grad_norm": 0.17462658439783693,
+      "learning_rate": 3.156713841905419e-05,
+      "loss": 0.3165,
+      "step": 2780
+    },
+    {
+      "epoch": 2.5955265610438025,
+      "grad_norm": 0.16767554292020126,
+      "learning_rate": 3.152399033482913e-05,
+      "loss": 0.3111,
+      "step": 2785
+    },
+    {
+      "epoch": 2.6001863932898415,
+      "grad_norm": 0.1686444036008289,
+      "learning_rate": 3.1480842250604076e-05,
+      "loss": 0.3176,
+      "step": 2790
+    },
+    {
+      "epoch": 2.604846225535881,
+      "grad_norm": 0.1727209958840771,
+      "learning_rate": 3.143769416637902e-05,
+      "loss": 0.3176,
+      "step": 2795
+    },
+    {
+      "epoch": 2.60950605778192,
+      "grad_norm": 0.16053190657473493,
+      "learning_rate": 3.1394546082153954e-05,
+      "loss": 0.3079,
+      "step": 2800
+    },
+    {
+      "epoch": 2.6141658900279587,
+      "grad_norm": 0.16816508302132707,
+      "learning_rate": 3.13513979979289e-05,
+      "loss": 0.3153,
+      "step": 2805
+    },
+    {
+      "epoch": 2.618825722273998,
+      "grad_norm": 0.16540599884205437,
+      "learning_rate": 3.130824991370383e-05,
+      "loss": 0.31,
+      "step": 2810
+    },
+    {
+      "epoch": 2.6234855545200375,
+      "grad_norm": 0.17281287968963135,
+      "learning_rate": 3.126510182947877e-05,
+      "loss": 0.3104,
+      "step": 2815
+    },
+    {
+      "epoch": 2.6281453867660765,
+      "grad_norm": 0.1764310159826407,
+      "learning_rate": 3.122195374525371e-05,
+      "loss": 0.3231,
+      "step": 2820
+    },
+    {
+      "epoch": 2.6328052190121154,
+      "grad_norm": 0.18069512990945666,
+      "learning_rate": 3.117880566102865e-05,
+      "loss": 0.317,
+      "step": 2825
+    },
+    {
+      "epoch": 2.637465051258155,
+      "grad_norm": 0.17959445224755632,
+      "learning_rate": 3.113565757680359e-05,
+      "loss": 0.3134,
+      "step": 2830
+    },
+    {
+      "epoch": 2.6421248835041937,
+      "grad_norm": 0.1824329884089278,
+      "learning_rate": 3.109250949257853e-05,
+      "loss": 0.3174,
+      "step": 2835
+    },
+    {
+      "epoch": 2.646784715750233,
+      "grad_norm": 0.16568936270250936,
+      "learning_rate": 3.1049361408353474e-05,
+      "loss": 0.3135,
+      "step": 2840
+    },
+    {
+      "epoch": 2.651444547996272,
+      "grad_norm": 0.16643964656072052,
+      "learning_rate": 3.100621332412841e-05,
+      "loss": 0.3085,
+      "step": 2845
+    },
+    {
+      "epoch": 2.6561043802423114,
+      "grad_norm": 0.16578958054899806,
+      "learning_rate": 3.096306523990335e-05,
+      "loss": 0.3184,
+      "step": 2850
+    },
+    {
+      "epoch": 2.6607642124883504,
+      "grad_norm": 0.16661775365156622,
+      "learning_rate": 3.091991715567829e-05,
+      "loss": 0.314,
+      "step": 2855
+    },
+    {
+      "epoch": 2.6654240447343893,
+      "grad_norm": 0.18262857909179542,
+      "learning_rate": 3.0876769071453223e-05,
+      "loss": 0.3207,
+      "step": 2860
+    },
+    {
+      "epoch": 2.6700838769804287,
+      "grad_norm": 0.18280694396177918,
+      "learning_rate": 3.0833620987228166e-05,
+      "loss": 0.3148,
+      "step": 2865
+    },
+    {
+      "epoch": 2.674743709226468,
+      "grad_norm": 0.1628149562591223,
+      "learning_rate": 3.079047290300311e-05,
+      "loss": 0.311,
+      "step": 2870
+    },
+    {
+      "epoch": 2.679403541472507,
+      "grad_norm": 0.1837135993541515,
+      "learning_rate": 3.074732481877805e-05,
+      "loss": 0.3265,
+      "step": 2875
+    },
+    {
+      "epoch": 2.684063373718546,
+      "grad_norm": 0.17109185985854608,
+      "learning_rate": 3.070417673455299e-05,
+      "loss": 0.3203,
+      "step": 2880
+    },
+    {
+      "epoch": 2.6887232059645854,
+      "grad_norm": 0.1695240523454995,
+      "learning_rate": 3.066102865032793e-05,
+      "loss": 0.311,
+      "step": 2885
+    },
+    {
+      "epoch": 2.6933830382106243,
+      "grad_norm": 0.1645222716761149,
+      "learning_rate": 3.061788056610287e-05,
+      "loss": 0.3167,
+      "step": 2890
+    },
+    {
+      "epoch": 2.6980428704566637,
+      "grad_norm": 0.17034984416923166,
+      "learning_rate": 3.057473248187781e-05,
+      "loss": 0.3149,
+      "step": 2895
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 0.17875185943225533,
+      "learning_rate": 3.053158439765274e-05,
+      "loss": 0.3079,
+      "step": 2900
+    },
+    {
+      "epoch": 2.707362534948742,
+      "grad_norm": 0.1606845277084022,
+      "learning_rate": 3.048843631342769e-05,
+      "loss": 0.3123,
+      "step": 2905
+    },
+    {
+      "epoch": 2.712022367194781,
+      "grad_norm": 0.17875654117697398,
+      "learning_rate": 3.0445288229202625e-05,
+      "loss": 0.3211,
+      "step": 2910
+    },
+    {
+      "epoch": 2.71668219944082,
+      "grad_norm": 0.17562313964018886,
+      "learning_rate": 3.0402140144977564e-05,
+      "loss": 0.3104,
+      "step": 2915
+    },
+    {
+      "epoch": 2.7213420316868593,
+      "grad_norm": 0.16461722825931616,
+      "learning_rate": 3.0358992060752506e-05,
+      "loss": 0.3062,
+      "step": 2920
+    },
+    {
+      "epoch": 2.7260018639328987,
+      "grad_norm": 0.16508665529851527,
+      "learning_rate": 3.0315843976527442e-05,
+      "loss": 0.3111,
+      "step": 2925
+    },
+    {
+      "epoch": 2.7306616961789376,
+      "grad_norm": 0.1581356300054825,
+      "learning_rate": 3.027269589230238e-05,
+      "loss": 0.3221,
+      "step": 2930
+    },
+    {
+      "epoch": 2.7353215284249766,
+      "grad_norm": 0.16226326532104743,
+      "learning_rate": 3.0229547808077324e-05,
+      "loss": 0.3119,
+      "step": 2935
+    },
+    {
+      "epoch": 2.739981360671016,
+      "grad_norm": 0.171964561283215,
+      "learning_rate": 3.018639972385226e-05,
+      "loss": 0.3158,
+      "step": 2940
+    },
+    {
+      "epoch": 2.744641192917055,
+      "grad_norm": 0.15323025906122206,
+      "learning_rate": 3.0143251639627202e-05,
+      "loss": 0.3137,
+      "step": 2945
+    },
+    {
+      "epoch": 2.7493010251630943,
+      "grad_norm": 0.15868039227761774,
+      "learning_rate": 3.010010355540214e-05,
+      "loss": 0.3207,
+      "step": 2950
+    },
+    {
+      "epoch": 2.753960857409133,
+      "grad_norm": 0.16482110591448124,
+      "learning_rate": 3.0056955471177077e-05,
+      "loss": 0.3064,
+      "step": 2955
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 0.1564002298655694,
+      "learning_rate": 3.001380738695202e-05,
+      "loss": 0.3107,
+      "step": 2960
+    },
+    {
+      "epoch": 2.7632805219012115,
+      "grad_norm": 0.177273932192958,
+      "learning_rate": 2.9970659302726962e-05,
+      "loss": 0.3152,
+      "step": 2965
+    },
+    {
+      "epoch": 2.7679403541472505,
+      "grad_norm": 0.1757334682599843,
+      "learning_rate": 2.99275112185019e-05,
+      "loss": 0.3155,
+      "step": 2970
+    },
+    {
+      "epoch": 2.77260018639329,
+      "grad_norm": 0.17452806838266255,
+      "learning_rate": 2.9884363134276837e-05,
+      "loss": 0.3151,
+      "step": 2975
+    },
+    {
+      "epoch": 2.7772600186393293,
+      "grad_norm": 0.17434340983966004,
+      "learning_rate": 2.984121505005178e-05,
+      "loss": 0.3093,
+      "step": 2980
+    },
+    {
+      "epoch": 2.781919850885368,
+      "grad_norm": 0.1587038934920305,
+      "learning_rate": 2.979806696582672e-05,
+      "loss": 0.3179,
+      "step": 2985
+    },
+    {
+      "epoch": 2.786579683131407,
+      "grad_norm": 0.1631092008094617,
+      "learning_rate": 2.9754918881601657e-05,
+      "loss": 0.3179,
+      "step": 2990
+    },
+    {
+      "epoch": 2.7912395153774465,
+      "grad_norm": 0.16856667017163138,
+      "learning_rate": 2.9711770797376596e-05,
+      "loss": 0.3081,
+      "step": 2995
+    },
+    {
+      "epoch": 2.7958993476234855,
+      "grad_norm": 0.17525203814865234,
+      "learning_rate": 2.966862271315154e-05,
+      "loss": 0.3108,
+      "step": 3000
+    },
+    {
+      "epoch": 2.800559179869525,
+      "grad_norm": 0.1732841028082698,
+      "learning_rate": 2.9625474628926475e-05,
+      "loss": 0.3202,
+      "step": 3005
+    },
+    {
+      "epoch": 2.805219012115564,
+      "grad_norm": 0.17301327453768614,
+      "learning_rate": 2.9582326544701417e-05,
+      "loss": 0.3244,
+      "step": 3010
+    },
+    {
+      "epoch": 2.809878844361603,
+      "grad_norm": 0.2110602094247485,
+      "learning_rate": 2.9539178460476356e-05,
+      "loss": 0.31,
+      "step": 3015
+    },
+    {
+      "epoch": 2.814538676607642,
+      "grad_norm": 0.18828189059777564,
+      "learning_rate": 2.9496030376251292e-05,
+      "loss": 0.3089,
+      "step": 3020
+    },
+    {
+      "epoch": 2.819198508853681,
+      "grad_norm": 0.17262372934115236,
+      "learning_rate": 2.9452882292026235e-05,
+      "loss": 0.3205,
+      "step": 3025
+    },
+    {
+      "epoch": 2.8238583410997204,
+      "grad_norm": 0.18989419460546014,
+      "learning_rate": 2.9409734207801177e-05,
+      "loss": 0.3266,
+      "step": 3030
+    },
+    {
+      "epoch": 2.8285181733457594,
+      "grad_norm": 0.17193455837772903,
+      "learning_rate": 2.9366586123576113e-05,
+      "loss": 0.318,
+      "step": 3035
+    },
+    {
+      "epoch": 2.8331780055917988,
+      "grad_norm": 0.1604581102891857,
+      "learning_rate": 2.9323438039351052e-05,
+      "loss": 0.3125,
+      "step": 3040
+    },
+    {
+      "epoch": 2.8378378378378377,
+      "grad_norm": 0.19256215986431852,
+      "learning_rate": 2.9280289955125994e-05,
+      "loss": 0.3121,
+      "step": 3045
+    },
+    {
+      "epoch": 2.842497670083877,
+      "grad_norm": 0.16030978927704514,
+      "learning_rate": 2.9237141870900937e-05,
+      "loss": 0.3188,
+      "step": 3050
+    },
+    {
+      "epoch": 2.847157502329916,
+      "grad_norm": 0.1888198271840261,
+      "learning_rate": 2.9193993786675873e-05,
+      "loss": 0.3132,
+      "step": 3055
+    },
+    {
+      "epoch": 2.8518173345759554,
+      "grad_norm": 0.17199373227215894,
+      "learning_rate": 2.9150845702450812e-05,
+      "loss": 0.3105,
+      "step": 3060
+    },
+    {
+      "epoch": 2.8564771668219944,
+      "grad_norm": 0.17164804507088174,
+      "learning_rate": 2.9107697618225754e-05,
+      "loss": 0.3119,
+      "step": 3065
+    },
+    {
+      "epoch": 2.8611369990680338,
+      "grad_norm": 0.1700771208717647,
+      "learning_rate": 2.906454953400069e-05,
+      "loss": 0.3111,
+      "step": 3070
+    },
+    {
+      "epoch": 2.8657968313140727,
+      "grad_norm": 0.18174101237461388,
+      "learning_rate": 2.9021401449775632e-05,
+      "loss": 0.314,
+      "step": 3075
+    },
+    {
+      "epoch": 2.8704566635601116,
+      "grad_norm": 0.17955692094843687,
+      "learning_rate": 2.897825336555057e-05,
+      "loss": 0.3203,
+      "step": 3080
+    },
+    {
+      "epoch": 2.875116495806151,
+      "grad_norm": 0.15699548212494752,
+      "learning_rate": 2.8935105281325507e-05,
+      "loss": 0.3096,
+      "step": 3085
+    },
+    {
+      "epoch": 2.87977632805219,
+      "grad_norm": 0.15654181784175508,
+      "learning_rate": 2.889195719710045e-05,
+      "loss": 0.3094,
+      "step": 3090
+    },
+    {
+      "epoch": 2.8844361602982294,
+      "grad_norm": 0.15218420487617018,
+      "learning_rate": 2.8848809112875392e-05,
+      "loss": 0.3125,
+      "step": 3095
+    },
+    {
+      "epoch": 2.8890959925442683,
+      "grad_norm": 0.1667155587002056,
+      "learning_rate": 2.8805661028650328e-05,
+      "loss": 0.3143,
+      "step": 3100
+    },
+    {
+      "epoch": 2.8937558247903077,
+      "grad_norm": 0.16227453364908237,
+      "learning_rate": 2.8762512944425267e-05,
+      "loss": 0.3071,
+      "step": 3105
+    },
+    {
+      "epoch": 2.8984156570363466,
+      "grad_norm": 0.1670953481765833,
+      "learning_rate": 2.871936486020021e-05,
+      "loss": 0.3221,
+      "step": 3110
+    },
+    {
+      "epoch": 2.9030754892823856,
+      "grad_norm": 0.17544367630819677,
+      "learning_rate": 2.8676216775975145e-05,
+      "loss": 0.3063,
+      "step": 3115
+    },
+    {
+      "epoch": 2.907735321528425,
+      "grad_norm": 0.1786736622307439,
+      "learning_rate": 2.8633068691750088e-05,
+      "loss": 0.3145,
+      "step": 3120
+    },
+    {
+      "epoch": 2.9123951537744643,
+      "grad_norm": 0.1693553097612915,
+      "learning_rate": 2.8589920607525027e-05,
+      "loss": 0.3072,
+      "step": 3125
+    },
+    {
+      "epoch": 2.9170549860205033,
+      "grad_norm": 0.16506798196657724,
+      "learning_rate": 2.854677252329997e-05,
+      "loss": 0.3068,
+      "step": 3130
+    },
+    {
+      "epoch": 2.9217148182665422,
+      "grad_norm": 0.17322310238616498,
+      "learning_rate": 2.8503624439074905e-05,
+      "loss": 0.3136,
+      "step": 3135
+    },
+    {
+      "epoch": 2.9263746505125816,
+      "grad_norm": 0.1568521315258698,
+      "learning_rate": 2.8460476354849848e-05,
+      "loss": 0.3086,
+      "step": 3140
+    },
+    {
+      "epoch": 2.9310344827586206,
+      "grad_norm": 0.15725039764767065,
+      "learning_rate": 2.8417328270624787e-05,
+      "loss": 0.3155,
+      "step": 3145
+    },
+    {
+      "epoch": 2.93569431500466,
+      "grad_norm": 0.17174407106022443,
+      "learning_rate": 2.8374180186399723e-05,
+      "loss": 0.3166,
+      "step": 3150
+    },
+    {
+      "epoch": 2.940354147250699,
+      "grad_norm": 0.16448285180151676,
+      "learning_rate": 2.8331032102174665e-05,
+      "loss": 0.3145,
+      "step": 3155
+    },
+    {
+      "epoch": 2.9450139794967383,
+      "grad_norm": 0.15903502283752102,
+      "learning_rate": 2.8287884017949608e-05,
+      "loss": 0.3139,
+      "step": 3160
+    },
+    {
+      "epoch": 2.949673811742777,
+      "grad_norm": 0.15727223435810367,
+      "learning_rate": 2.8244735933724543e-05,
+      "loss": 0.3085,
+      "step": 3165
+    },
+    {
+      "epoch": 2.954333643988816,
+      "grad_norm": 0.1795957737871983,
+      "learning_rate": 2.8201587849499482e-05,
+      "loss": 0.3126,
+      "step": 3170
+    },
+    {
+      "epoch": 2.9589934762348555,
+      "grad_norm": 0.16910588602096008,
+      "learning_rate": 2.8158439765274425e-05,
+      "loss": 0.3166,
+      "step": 3175
+    },
+    {
+      "epoch": 2.963653308480895,
+      "grad_norm": 0.16738271828448753,
+      "learning_rate": 2.811529168104936e-05,
+      "loss": 0.3171,
+      "step": 3180
+    },
+    {
+      "epoch": 2.968313140726934,
+      "grad_norm": 0.15422488596323689,
+      "learning_rate": 2.8072143596824303e-05,
+      "loss": 0.3125,
+      "step": 3185
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 0.16119523023251425,
+      "learning_rate": 2.8028995512599242e-05,
+      "loss": 0.3117,
+      "step": 3190
+    },
+    {
+      "epoch": 2.977632805219012,
+      "grad_norm": 0.17092546387743232,
+      "learning_rate": 2.7985847428374178e-05,
+      "loss": 0.3129,
+      "step": 3195
+    },
+    {
+      "epoch": 2.982292637465051,
+      "grad_norm": 0.17102548098833978,
+      "learning_rate": 2.794269934414912e-05,
+      "loss": 0.3181,
+      "step": 3200
+    },
+    {
+      "epoch": 2.9869524697110905,
+      "grad_norm": 0.1633906148610829,
+      "learning_rate": 2.7899551259924063e-05,
+      "loss": 0.3137,
+      "step": 3205
+    },
+    {
+      "epoch": 2.9916123019571295,
+      "grad_norm": 0.16288844482054976,
+      "learning_rate": 2.7856403175699002e-05,
+      "loss": 0.3152,
+      "step": 3210
+    },
+    {
+      "epoch": 2.996272134203169,
+      "grad_norm": 0.17198579707450976,
+      "learning_rate": 2.7813255091473938e-05,
+      "loss": 0.3154,
+      "step": 3215
+    },
+    {
+      "epoch": 3.000931966449208,
+      "grad_norm": 0.2839752409330015,
+      "learning_rate": 2.777010700724888e-05,
+      "loss": 0.3043,
+      "step": 3220
+    },
+    {
+      "epoch": 3.005591798695247,
+      "grad_norm": 0.23245958339160072,
+      "learning_rate": 2.7726958923023823e-05,
+      "loss": 0.2367,
+      "step": 3225
+    },
+    {
+      "epoch": 3.010251630941286,
+      "grad_norm": 0.21054198940734953,
+      "learning_rate": 2.768381083879876e-05,
+      "loss": 0.2365,
+      "step": 3230
+    },
+    {
+      "epoch": 3.014911463187325,
+      "grad_norm": 0.18832957302346418,
+      "learning_rate": 2.7640662754573698e-05,
+      "loss": 0.2338,
+      "step": 3235
+    },
+    {
+      "epoch": 3.0195712954333644,
+      "grad_norm": 0.18778075802367614,
+      "learning_rate": 2.759751467034864e-05,
+      "loss": 0.2308,
+      "step": 3240
+    },
+    {
+      "epoch": 3.0242311276794034,
+      "grad_norm": 0.17246925923622156,
+      "learning_rate": 2.7554366586123576e-05,
+      "loss": 0.2338,
+      "step": 3245
+    },
+    {
+      "epoch": 3.0288909599254428,
+      "grad_norm": 0.18991043473015579,
+      "learning_rate": 2.751121850189852e-05,
+      "loss": 0.2339,
+      "step": 3250
+    },
+    {
+      "epoch": 3.0335507921714817,
+      "grad_norm": 0.1773276473764673,
+      "learning_rate": 2.7468070417673457e-05,
+      "loss": 0.2394,
+      "step": 3255
+    },
+    {
+      "epoch": 3.038210624417521,
+      "grad_norm": 0.1812721632411487,
+      "learning_rate": 2.7424922333448393e-05,
+      "loss": 0.2313,
+      "step": 3260
+    },
+    {
+      "epoch": 3.04287045666356,
+      "grad_norm": 0.17841383344952372,
+      "learning_rate": 2.7381774249223336e-05,
+      "loss": 0.2388,
+      "step": 3265
+    },
+    {
+      "epoch": 3.0475302889095994,
+      "grad_norm": 0.1803910464930852,
+      "learning_rate": 2.7338626164998278e-05,
+      "loss": 0.2379,
+      "step": 3270
+    },
+    {
+      "epoch": 3.0521901211556384,
+      "grad_norm": 0.17857483311326022,
+      "learning_rate": 2.7295478080773214e-05,
+      "loss": 0.2385,
+      "step": 3275
+    },
+    {
+      "epoch": 3.0568499534016778,
+      "grad_norm": 0.17823088138062881,
+      "learning_rate": 2.7252329996548153e-05,
+      "loss": 0.2363,
+      "step": 3280
+    },
+    {
+      "epoch": 3.0615097856477167,
+      "grad_norm": 0.178995781350824,
+      "learning_rate": 2.7209181912323096e-05,
+      "loss": 0.2341,
+      "step": 3285
+    },
+    {
+      "epoch": 3.0661696178937556,
+      "grad_norm": 0.19245010648095198,
+      "learning_rate": 2.7166033828098038e-05,
+      "loss": 0.2381,
+      "step": 3290
+    },
+    {
+      "epoch": 3.070829450139795,
+      "grad_norm": 0.18337334340233233,
+      "learning_rate": 2.7122885743872974e-05,
+      "loss": 0.236,
+      "step": 3295
+    },
+    {
+      "epoch": 3.075489282385834,
+      "grad_norm": 0.18827968760378397,
+      "learning_rate": 2.7079737659647913e-05,
+      "loss": 0.2376,
+      "step": 3300
+    },
+    {
+      "epoch": 3.0801491146318734,
+      "grad_norm": 0.17801817716802906,
+      "learning_rate": 2.7036589575422855e-05,
+      "loss": 0.2278,
+      "step": 3305
+    },
+    {
+      "epoch": 3.0848089468779123,
+      "grad_norm": 0.18756117753741589,
+      "learning_rate": 2.699344149119779e-05,
+      "loss": 0.2358,
+      "step": 3310
+    },
+    {
+      "epoch": 3.0894687791239517,
+      "grad_norm": 0.19062620075055187,
+      "learning_rate": 2.6950293406972734e-05,
+      "loss": 0.2312,
+      "step": 3315
+    },
+    {
+      "epoch": 3.0941286113699906,
+      "grad_norm": 0.17686693618083657,
+      "learning_rate": 2.6907145322747673e-05,
+      "loss": 0.2384,
+      "step": 3320
+    },
+    {
+      "epoch": 3.09878844361603,
+      "grad_norm": 0.17568861727824264,
+      "learning_rate": 2.686399723852261e-05,
+      "loss": 0.2349,
+      "step": 3325
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 0.1847026203820198,
+      "learning_rate": 2.682084915429755e-05,
+      "loss": 0.2423,
+      "step": 3330
+    },
+    {
+      "epoch": 3.108108108108108,
+      "grad_norm": 0.18737394306934632,
+      "learning_rate": 2.677770107007249e-05,
+      "loss": 0.2424,
+      "step": 3335
+    },
+    {
+      "epoch": 3.1127679403541473,
+      "grad_norm": 0.1711574871713759,
+      "learning_rate": 2.673455298584743e-05,
+      "loss": 0.2403,
+      "step": 3340
+    },
+    {
+      "epoch": 3.117427772600186,
+      "grad_norm": 0.17676054425761437,
+      "learning_rate": 2.6691404901622368e-05,
+      "loss": 0.2412,
+      "step": 3345
+    },
+    {
+      "epoch": 3.1220876048462256,
+      "grad_norm": 0.18318634766270822,
+      "learning_rate": 2.664825681739731e-05,
+      "loss": 0.235,
+      "step": 3350
+    },
+    {
+      "epoch": 3.1267474370922645,
+      "grad_norm": 0.17262783983608546,
+      "learning_rate": 2.6605108733172247e-05,
+      "loss": 0.2401,
+      "step": 3355
+    },
+    {
+      "epoch": 3.131407269338304,
+      "grad_norm": 0.1753471525314905,
+      "learning_rate": 2.6561960648947186e-05,
+      "loss": 0.2325,
+      "step": 3360
+    },
+    {
+      "epoch": 3.136067101584343,
+      "grad_norm": 0.1725084723960369,
+      "learning_rate": 2.6518812564722128e-05,
+      "loss": 0.2323,
+      "step": 3365
+    },
+    {
+      "epoch": 3.1407269338303823,
+      "grad_norm": 0.1750741883746647,
+      "learning_rate": 2.647566448049707e-05,
+      "loss": 0.24,
+      "step": 3370
+    },
+    {
+      "epoch": 3.145386766076421,
+      "grad_norm": 0.17164136539827254,
+      "learning_rate": 2.6432516396272006e-05,
+      "loss": 0.2372,
+      "step": 3375
+    },
+    {
+      "epoch": 3.1500465983224606,
+      "grad_norm": 0.18707853882473635,
+      "learning_rate": 2.6389368312046945e-05,
+      "loss": 0.2414,
+      "step": 3380
+    },
+    {
+      "epoch": 3.1547064305684995,
+      "grad_norm": 0.1769712333624102,
+      "learning_rate": 2.6346220227821888e-05,
+      "loss": 0.2355,
+      "step": 3385
+    },
+    {
+      "epoch": 3.1593662628145385,
+      "grad_norm": 0.18458837536463837,
+      "learning_rate": 2.6303072143596824e-05,
+      "loss": 0.233,
+      "step": 3390
+    },
+    {
+      "epoch": 3.164026095060578,
+      "grad_norm": 0.1750058698128996,
+      "learning_rate": 2.6259924059371766e-05,
+      "loss": 0.2382,
+      "step": 3395
+    },
+    {
+      "epoch": 3.168685927306617,
+      "grad_norm": 0.17066718595713276,
+      "learning_rate": 2.6216775975146705e-05,
+      "loss": 0.2366,
+      "step": 3400
+    },
+    {
+      "epoch": 3.173345759552656,
+      "grad_norm": 0.1706931819948248,
+      "learning_rate": 2.617362789092164e-05,
+      "loss": 0.2324,
+      "step": 3405
+    },
+    {
+      "epoch": 3.178005591798695,
+      "grad_norm": 0.17351982499671245,
+      "learning_rate": 2.6130479806696584e-05,
+      "loss": 0.2331,
+      "step": 3410
+    },
+    {
+      "epoch": 3.1826654240447345,
+      "grad_norm": 0.19087932191903775,
+      "learning_rate": 2.6087331722471526e-05,
+      "loss": 0.24,
+      "step": 3415
+    },
+    {
+      "epoch": 3.1873252562907735,
+      "grad_norm": 0.1719169971864863,
+      "learning_rate": 2.6044183638246462e-05,
+      "loss": 0.2377,
+      "step": 3420
+    },
+    {
+      "epoch": 3.191985088536813,
+      "grad_norm": 0.17237052028440908,
+      "learning_rate": 2.60010355540214e-05,
+      "loss": 0.2392,
+      "step": 3425
+    },
+    {
+      "epoch": 3.196644920782852,
+      "grad_norm": 0.18726164695570094,
+      "learning_rate": 2.5957887469796343e-05,
+      "loss": 0.2397,
+      "step": 3430
+    },
+    {
+      "epoch": 3.201304753028891,
+      "grad_norm": 0.179790681644883,
+      "learning_rate": 2.591473938557128e-05,
+      "loss": 0.2419,
+      "step": 3435
+    },
+    {
+      "epoch": 3.20596458527493,
+      "grad_norm": 0.17907091189671123,
+      "learning_rate": 2.587159130134622e-05,
+      "loss": 0.24,
+      "step": 3440
+    },
+    {
+      "epoch": 3.210624417520969,
+      "grad_norm": 0.1790348904623648,
+      "learning_rate": 2.582844321712116e-05,
+      "loss": 0.2349,
+      "step": 3445
+    },
+    {
+      "epoch": 3.2152842497670084,
+      "grad_norm": 0.18346894707783884,
+      "learning_rate": 2.5785295132896096e-05,
+      "loss": 0.2441,
+      "step": 3450
+    },
+    {
+      "epoch": 3.2199440820130474,
+      "grad_norm": 0.1754209529675665,
+      "learning_rate": 2.574214704867104e-05,
+      "loss": 0.241,
+      "step": 3455
+    },
+    {
+      "epoch": 3.2246039142590868,
+      "grad_norm": 0.18180144933964731,
+      "learning_rate": 2.569899896444598e-05,
+      "loss": 0.238,
+      "step": 3460
+    },
+    {
+      "epoch": 3.2292637465051257,
+      "grad_norm": 0.18418079330200668,
+      "learning_rate": 2.565585088022092e-05,
+      "loss": 0.2428,
+      "step": 3465
+    },
+    {
+      "epoch": 3.233923578751165,
+      "grad_norm": 0.17870173108567952,
+      "learning_rate": 2.5612702795995856e-05,
+      "loss": 0.237,
+      "step": 3470
+    },
+    {
+      "epoch": 3.238583410997204,
+      "grad_norm": 0.1740997965622895,
+      "learning_rate": 2.55695547117708e-05,
+      "loss": 0.2398,
+      "step": 3475
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 0.18333939100518043,
+      "learning_rate": 2.552640662754574e-05,
+      "loss": 0.2416,
+      "step": 3480
+    },
+    {
+      "epoch": 3.2479030754892824,
+      "grad_norm": 0.188564395122458,
+      "learning_rate": 2.5483258543320677e-05,
+      "loss": 0.2453,
+      "step": 3485
+    },
+    {
+      "epoch": 3.2525629077353218,
+      "grad_norm": 0.18701508493480223,
+      "learning_rate": 2.5440110459095616e-05,
+      "loss": 0.239,
+      "step": 3490
+    },
+    {
+      "epoch": 3.2572227399813607,
+      "grad_norm": 0.1908162901853877,
+      "learning_rate": 2.539696237487056e-05,
+      "loss": 0.2414,
+      "step": 3495
+    },
+    {
+      "epoch": 3.2618825722273996,
+      "grad_norm": 0.17412328643938918,
+      "learning_rate": 2.5353814290645494e-05,
+      "loss": 0.239,
+      "step": 3500
+    },
+    {
+      "epoch": 3.266542404473439,
+      "grad_norm": 0.17284121391777552,
+      "learning_rate": 2.5310666206420437e-05,
+      "loss": 0.2385,
+      "step": 3505
+    },
+    {
+      "epoch": 3.271202236719478,
+      "grad_norm": 0.18027858450554357,
+      "learning_rate": 2.5267518122195376e-05,
+      "loss": 0.2424,
+      "step": 3510
+    },
+    {
+      "epoch": 3.2758620689655173,
+      "grad_norm": 0.1790517831306251,
+      "learning_rate": 2.5224370037970312e-05,
+      "loss": 0.2428,
+      "step": 3515
+    },
+    {
+      "epoch": 3.2805219012115563,
+      "grad_norm": 0.1828719210027198,
+      "learning_rate": 2.5181221953745254e-05,
+      "loss": 0.2464,
+      "step": 3520
+    },
+    {
+      "epoch": 3.2851817334575957,
+      "grad_norm": 0.17896345083562346,
+      "learning_rate": 2.5138073869520197e-05,
+      "loss": 0.2393,
+      "step": 3525
+    },
+    {
+      "epoch": 3.2898415657036346,
+      "grad_norm": 0.20738514948648956,
+      "learning_rate": 2.5094925785295132e-05,
+      "loss": 0.2363,
+      "step": 3530
+    },
+    {
+      "epoch": 3.294501397949674,
+      "grad_norm": 0.19125391532855057,
+      "learning_rate": 2.505177770107007e-05,
+      "loss": 0.2396,
+      "step": 3535
+    },
+    {
+      "epoch": 3.299161230195713,
+      "grad_norm": 0.205044215381975,
+      "learning_rate": 2.5008629616845014e-05,
+      "loss": 0.2471,
+      "step": 3540
+    },
+    {
+      "epoch": 3.3038210624417523,
+      "grad_norm": 0.17524455538023537,
+      "learning_rate": 2.4965481532619953e-05,
+      "loss": 0.2425,
+      "step": 3545
+    },
+    {
+      "epoch": 3.3084808946877913,
+      "grad_norm": 0.17515673322281508,
+      "learning_rate": 2.4922333448394892e-05,
+      "loss": 0.2387,
+      "step": 3550
+    },
+    {
+      "epoch": 3.31314072693383,
+      "grad_norm": 0.17359414660798325,
+      "learning_rate": 2.487918536416983e-05,
+      "loss": 0.2385,
+      "step": 3555
+    },
+    {
+      "epoch": 3.3178005591798696,
+      "grad_norm": 0.17438564953744223,
+      "learning_rate": 2.483603727994477e-05,
+      "loss": 0.2441,
+      "step": 3560
+    },
+    {
+      "epoch": 3.3224603914259085,
+      "grad_norm": 0.17883001103175256,
+      "learning_rate": 2.4792889195719713e-05,
+      "loss": 0.2415,
+      "step": 3565
+    },
+    {
+      "epoch": 3.327120223671948,
+      "grad_norm": 0.171407691874097,
+      "learning_rate": 2.4749741111494652e-05,
+      "loss": 0.2375,
+      "step": 3570
+    },
+    {
+      "epoch": 3.331780055917987,
+      "grad_norm": 0.17489170096204273,
+      "learning_rate": 2.4706593027269588e-05,
+      "loss": 0.2422,
+      "step": 3575
+    },
+    {
+      "epoch": 3.3364398881640263,
+      "grad_norm": 0.18672256534067577,
+      "learning_rate": 2.466344494304453e-05,
+      "loss": 0.2427,
+      "step": 3580
+    },
+    {
+      "epoch": 3.341099720410065,
+      "grad_norm": 0.16875095674600468,
+      "learning_rate": 2.462029685881947e-05,
+      "loss": 0.2367,
+      "step": 3585
+    },
+    {
+      "epoch": 3.345759552656104,
+      "grad_norm": 0.17418142268618275,
+      "learning_rate": 2.457714877459441e-05,
+      "loss": 0.2408,
+      "step": 3590
+    },
+    {
+      "epoch": 3.3504193849021435,
+      "grad_norm": 0.16870447218973614,
+      "learning_rate": 2.4534000690369348e-05,
+      "loss": 0.242,
+      "step": 3595
+    },
+    {
+      "epoch": 3.355079217148183,
+      "grad_norm": 0.18044523116802536,
+      "learning_rate": 2.4490852606144287e-05,
+      "loss": 0.2389,
+      "step": 3600
+    },
+    {
+      "epoch": 3.359739049394222,
+      "grad_norm": 0.1811184395442793,
+      "learning_rate": 2.444770452191923e-05,
+      "loss": 0.2384,
+      "step": 3605
+    },
+    {
+      "epoch": 3.364398881640261,
+      "grad_norm": 0.17356154981479185,
+      "learning_rate": 2.440455643769417e-05,
+      "loss": 0.2428,
+      "step": 3610
+    },
+    {
+      "epoch": 3.3690587138863,
+      "grad_norm": 0.18172713575620875,
+      "learning_rate": 2.4361408353469108e-05,
+      "loss": 0.2367,
+      "step": 3615
+    },
+    {
+      "epoch": 3.373718546132339,
+      "grad_norm": 0.18643247396536974,
+      "learning_rate": 2.4318260269244047e-05,
+      "loss": 0.2423,
+      "step": 3620
+    },
+    {
+      "epoch": 3.3783783783783785,
+      "grad_norm": 0.16740086237428323,
+      "learning_rate": 2.4275112185018986e-05,
+      "loss": 0.2397,
+      "step": 3625
+    },
+    {
+      "epoch": 3.3830382106244175,
+      "grad_norm": 0.1798558488762979,
+      "learning_rate": 2.4231964100793925e-05,
+      "loss": 0.2368,
+      "step": 3630
+    },
+    {
+      "epoch": 3.387698042870457,
+      "grad_norm": 0.16580491618899695,
+      "learning_rate": 2.4188816016568867e-05,
+      "loss": 0.2445,
+      "step": 3635
+    },
+    {
+      "epoch": 3.392357875116496,
+      "grad_norm": 0.17306497814688196,
+      "learning_rate": 2.4145667932343803e-05,
+      "loss": 0.2377,
+      "step": 3640
+    },
+    {
+      "epoch": 3.3970177073625347,
+      "grad_norm": 0.18299140163219002,
+      "learning_rate": 2.4102519848118746e-05,
+      "loss": 0.2442,
+      "step": 3645
+    },
+    {
+      "epoch": 3.401677539608574,
+      "grad_norm": 0.1781994767350486,
+      "learning_rate": 2.4059371763893685e-05,
+      "loss": 0.2433,
+      "step": 3650
+    },
+    {
+      "epoch": 3.406337371854613,
+      "grad_norm": 0.1804845735883304,
+      "learning_rate": 2.4016223679668624e-05,
+      "loss": 0.2435,
+      "step": 3655
+    },
+    {
+      "epoch": 3.4109972041006524,
+      "grad_norm": 0.183858648764894,
+      "learning_rate": 2.3973075595443563e-05,
+      "loss": 0.243,
+      "step": 3660
+    },
+    {
+      "epoch": 3.4156570363466914,
+      "grad_norm": 0.17073660376525726,
+      "learning_rate": 2.3929927511218502e-05,
+      "loss": 0.2459,
+      "step": 3665
+    },
+    {
+      "epoch": 3.4203168685927308,
+      "grad_norm": 0.17994066905719286,
+      "learning_rate": 2.388677942699344e-05,
+      "loss": 0.2465,
+      "step": 3670
+    },
+    {
+      "epoch": 3.4249767008387697,
+      "grad_norm": 0.18781582042692557,
+      "learning_rate": 2.3843631342768384e-05,
+      "loss": 0.241,
+      "step": 3675
+    },
+    {
+      "epoch": 3.429636533084809,
+      "grad_norm": 0.17549822936083914,
+      "learning_rate": 2.3800483258543323e-05,
+      "loss": 0.2401,
+      "step": 3680
+    },
+    {
+      "epoch": 3.434296365330848,
+      "grad_norm": 0.1751311846783813,
+      "learning_rate": 2.3757335174318262e-05,
+      "loss": 0.2395,
+      "step": 3685
+    },
+    {
+      "epoch": 3.4389561975768874,
+      "grad_norm": 0.17941128180881236,
+      "learning_rate": 2.37141870900932e-05,
+      "loss": 0.2383,
+      "step": 3690
+    },
+    {
+      "epoch": 3.4436160298229264,
+      "grad_norm": 0.20188875487775104,
+      "learning_rate": 2.367103900586814e-05,
+      "loss": 0.2407,
+      "step": 3695
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 0.17827107031441444,
+      "learning_rate": 2.3627890921643083e-05,
+      "loss": 0.2362,
+      "step": 3700
+    },
+    {
+      "epoch": 3.4529356943150047,
+      "grad_norm": 0.17692739762500873,
+      "learning_rate": 2.358474283741802e-05,
+      "loss": 0.2405,
+      "step": 3705
+    },
+    {
+      "epoch": 3.4575955265610436,
+      "grad_norm": 0.19043155079946036,
+      "learning_rate": 2.3541594753192957e-05,
+      "loss": 0.2472,
+      "step": 3710
+    },
+    {
+      "epoch": 3.462255358807083,
+      "grad_norm": 0.1760566996963653,
+      "learning_rate": 2.34984466689679e-05,
+      "loss": 0.2425,
+      "step": 3715
+    },
+    {
+      "epoch": 3.466915191053122,
+      "grad_norm": 0.16450274949437593,
+      "learning_rate": 2.345529858474284e-05,
+      "loss": 0.2364,
+      "step": 3720
+    },
+    {
+      "epoch": 3.4715750232991613,
+      "grad_norm": 0.1804246297106457,
+      "learning_rate": 2.3412150500517778e-05,
+      "loss": 0.2362,
+      "step": 3725
+    },
+    {
+      "epoch": 3.4762348555452003,
+      "grad_norm": 0.1859147814073279,
+      "learning_rate": 2.3369002416292717e-05,
+      "loss": 0.2367,
+      "step": 3730
+    },
+    {
+      "epoch": 3.4808946877912397,
+      "grad_norm": 0.1730254126109042,
+      "learning_rate": 2.3325854332067656e-05,
+      "loss": 0.2415,
+      "step": 3735
+    },
+    {
+      "epoch": 3.4855545200372786,
+      "grad_norm": 0.18213454126067116,
+      "learning_rate": 2.32827062478426e-05,
+      "loss": 0.2335,
+      "step": 3740
+    },
+    {
+      "epoch": 3.490214352283318,
+      "grad_norm": 0.17510294542149088,
+      "learning_rate": 2.3239558163617538e-05,
+      "loss": 0.243,
+      "step": 3745
+    },
+    {
+      "epoch": 3.494874184529357,
+      "grad_norm": 0.18317427795648974,
+      "learning_rate": 2.3196410079392474e-05,
+      "loss": 0.2424,
+      "step": 3750
+    },
+    {
+      "epoch": 3.499534016775396,
+      "grad_norm": 0.17901344372400704,
+      "learning_rate": 2.3153261995167416e-05,
+      "loss": 0.2433,
+      "step": 3755
+    },
+    {
+      "epoch": 3.5041938490214353,
+      "grad_norm": 0.16395137430154183,
+      "learning_rate": 2.3110113910942355e-05,
+      "loss": 0.2402,
+      "step": 3760
+    },
+    {
+      "epoch": 3.508853681267474,
+      "grad_norm": 0.1693430939002628,
+      "learning_rate": 2.3066965826717294e-05,
+      "loss": 0.245,
+      "step": 3765
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 0.19825043127422068,
+      "learning_rate": 2.3023817742492234e-05,
+      "loss": 0.2424,
+      "step": 3770
+    },
+    {
+      "epoch": 3.5181733457595525,
+      "grad_norm": 0.19584534281838226,
+      "learning_rate": 2.2980669658267173e-05,
+      "loss": 0.2487,
+      "step": 3775
+    },
+    {
+      "epoch": 3.522833178005592,
+      "grad_norm": 0.1735193882848878,
+      "learning_rate": 2.2937521574042115e-05,
+      "loss": 0.2418,
+      "step": 3780
+    },
+    {
+      "epoch": 3.527493010251631,
+      "grad_norm": 0.18909643323844744,
+      "learning_rate": 2.2894373489817054e-05,
+      "loss": 0.2476,
+      "step": 3785
+    },
+    {
+      "epoch": 3.53215284249767,
+      "grad_norm": 0.18893033571033546,
+      "learning_rate": 2.285122540559199e-05,
+      "loss": 0.2428,
+      "step": 3790
+    },
+    {
+      "epoch": 3.536812674743709,
+      "grad_norm": 0.1845874329989991,
+      "learning_rate": 2.2808077321366933e-05,
+      "loss": 0.2379,
+      "step": 3795
+    },
+    {
+      "epoch": 3.5414725069897486,
+      "grad_norm": 0.18855044564176923,
+      "learning_rate": 2.276492923714187e-05,
+      "loss": 0.2409,
+      "step": 3800
+    },
+    {
+      "epoch": 3.5461323392357875,
+      "grad_norm": 0.180459845736056,
+      "learning_rate": 2.2721781152916814e-05,
+      "loss": 0.2484,
+      "step": 3805
+    },
+    {
+      "epoch": 3.5507921714818265,
+      "grad_norm": 0.19210070643291682,
+      "learning_rate": 2.267863306869175e-05,
+      "loss": 0.2484,
+      "step": 3810
+    },
+    {
+      "epoch": 3.555452003727866,
+      "grad_norm": 0.18048823351655285,
+      "learning_rate": 2.263548498446669e-05,
+      "loss": 0.2433,
+      "step": 3815
+    },
+    {
+      "epoch": 3.560111835973905,
+      "grad_norm": 0.1672334956686539,
+      "learning_rate": 2.259233690024163e-05,
+      "loss": 0.2373,
+      "step": 3820
+    },
+    {
+      "epoch": 3.564771668219944,
+      "grad_norm": 0.17719378101985395,
+      "learning_rate": 2.254918881601657e-05,
+      "loss": 0.2467,
+      "step": 3825
+    },
+    {
+      "epoch": 3.569431500465983,
+      "grad_norm": 0.18004025975710075,
+      "learning_rate": 2.250604073179151e-05,
+      "loss": 0.2408,
+      "step": 3830
+    },
+    {
+      "epoch": 3.5740913327120225,
+      "grad_norm": 0.1757202725302393,
+      "learning_rate": 2.246289264756645e-05,
+      "loss": 0.2483,
+      "step": 3835
+    },
+    {
+      "epoch": 3.5787511649580614,
+      "grad_norm": 0.17712269105674527,
+      "learning_rate": 2.2419744563341388e-05,
+      "loss": 0.2484,
+      "step": 3840
+    },
+    {
+      "epoch": 3.5834109972041004,
+      "grad_norm": 0.17076337936197014,
+      "learning_rate": 2.237659647911633e-05,
+      "loss": 0.2466,
+      "step": 3845
+    },
+    {
+      "epoch": 3.5880708294501398,
+      "grad_norm": 0.1830565616803644,
+      "learning_rate": 2.233344839489127e-05,
+      "loss": 0.2407,
+      "step": 3850
+    },
+    {
+      "epoch": 3.592730661696179,
+      "grad_norm": 0.17078249206308213,
+      "learning_rate": 2.2290300310666205e-05,
+      "loss": 0.2405,
+      "step": 3855
+    },
+    {
+      "epoch": 3.597390493942218,
+      "grad_norm": 0.18806887444897402,
+      "learning_rate": 2.2247152226441148e-05,
+      "loss": 0.2458,
+      "step": 3860
+    },
+    {
+      "epoch": 3.602050326188257,
+      "grad_norm": 0.17595047234610883,
+      "learning_rate": 2.2204004142216087e-05,
+      "loss": 0.2405,
+      "step": 3865
+    },
+    {
+      "epoch": 3.6067101584342964,
+      "grad_norm": 0.1930728045672143,
+      "learning_rate": 2.2160856057991026e-05,
+      "loss": 0.2418,
+      "step": 3870
+    },
+    {
+      "epoch": 3.6113699906803354,
+      "grad_norm": 0.181171281482145,
+      "learning_rate": 2.2117707973765965e-05,
+      "loss": 0.2473,
+      "step": 3875
+    },
+    {
+      "epoch": 3.6160298229263748,
+      "grad_norm": 0.16879567229051679,
+      "learning_rate": 2.2074559889540904e-05,
+      "loss": 0.2418,
+      "step": 3880
+    },
+    {
+      "epoch": 3.6206896551724137,
+      "grad_norm": 0.1784453749414881,
+      "learning_rate": 2.2031411805315847e-05,
+      "loss": 0.2525,
+      "step": 3885
+    },
+    {
+      "epoch": 3.625349487418453,
+      "grad_norm": 0.17362910714067598,
+      "learning_rate": 2.1988263721090786e-05,
+      "loss": 0.2411,
+      "step": 3890
+    },
+    {
+      "epoch": 3.630009319664492,
+      "grad_norm": 0.17166057558545225,
+      "learning_rate": 2.1945115636865725e-05,
+      "loss": 0.2374,
+      "step": 3895
+    },
+    {
+      "epoch": 3.634669151910531,
+      "grad_norm": 0.16852361342823868,
+      "learning_rate": 2.1901967552640664e-05,
+      "loss": 0.2403,
+      "step": 3900
+    },
+    {
+      "epoch": 3.6393289841565704,
+      "grad_norm": 0.17866787506419696,
+      "learning_rate": 2.1858819468415603e-05,
+      "loss": 0.2414,
+      "step": 3905
+    },
+    {
+      "epoch": 3.6439888164026097,
+      "grad_norm": 0.17564116442017555,
+      "learning_rate": 2.1815671384190542e-05,
+      "loss": 0.2441,
+      "step": 3910
+    },
+    {
+      "epoch": 3.6486486486486487,
+      "grad_norm": 0.18492146106442667,
+      "learning_rate": 2.1772523299965485e-05,
+      "loss": 0.2381,
+      "step": 3915
+    },
+    {
+      "epoch": 3.6533084808946876,
+      "grad_norm": 0.17236609439089842,
+      "learning_rate": 2.172937521574042e-05,
+      "loss": 0.2405,
+      "step": 3920
+    },
+    {
+      "epoch": 3.657968313140727,
+      "grad_norm": 0.1729528244279502,
+      "learning_rate": 2.1686227131515363e-05,
+      "loss": 0.2357,
+      "step": 3925
+    },
+    {
+      "epoch": 3.662628145386766,
+      "grad_norm": 0.17329257629608796,
+      "learning_rate": 2.1643079047290302e-05,
+      "loss": 0.243,
+      "step": 3930
+    },
+    {
+      "epoch": 3.6672879776328053,
+      "grad_norm": 0.17690406017346702,
+      "learning_rate": 2.159993096306524e-05,
+      "loss": 0.2462,
+      "step": 3935
+    },
+    {
+      "epoch": 3.6719478098788443,
+      "grad_norm": 0.1804232140601883,
+      "learning_rate": 2.155678287884018e-05,
+      "loss": 0.2466,
+      "step": 3940
+    },
+    {
+      "epoch": 3.6766076421248837,
+      "grad_norm": 0.18498877131788877,
+      "learning_rate": 2.151363479461512e-05,
+      "loss": 0.2446,
+      "step": 3945
+    },
+    {
+      "epoch": 3.6812674743709226,
+      "grad_norm": 0.18170878162012966,
+      "learning_rate": 2.147048671039006e-05,
+      "loss": 0.2395,
+      "step": 3950
+    },
+    {
+      "epoch": 3.6859273066169616,
+      "grad_norm": 0.16629132376466255,
+      "learning_rate": 2.1427338626165e-05,
+      "loss": 0.2469,
+      "step": 3955
+    },
+    {
+      "epoch": 3.690587138863001,
+      "grad_norm": 0.1722302850090895,
+      "learning_rate": 2.1384190541939937e-05,
+      "loss": 0.2451,
+      "step": 3960
+    },
+    {
+      "epoch": 3.6952469711090403,
+      "grad_norm": 0.1823882350266332,
+      "learning_rate": 2.134104245771488e-05,
+      "loss": 0.2419,
+      "step": 3965
+    },
+    {
+      "epoch": 3.6999068033550793,
+      "grad_norm": 0.17150884617360115,
+      "learning_rate": 2.129789437348982e-05,
+      "loss": 0.2451,
+      "step": 3970
+    },
+    {
+      "epoch": 3.704566635601118,
+      "grad_norm": 0.18101415189353548,
+      "learning_rate": 2.1254746289264758e-05,
+      "loss": 0.2485,
+      "step": 3975
+    },
+    {
+      "epoch": 3.7092264678471576,
+      "grad_norm": 0.16521844050383383,
+      "learning_rate": 2.1211598205039697e-05,
+      "loss": 0.2388,
+      "step": 3980
+    },
+    {
+      "epoch": 3.7138863000931965,
+      "grad_norm": 0.16844691659657765,
+      "learning_rate": 2.1168450120814636e-05,
+      "loss": 0.2475,
+      "step": 3985
+    },
+    {
+      "epoch": 3.718546132339236,
+      "grad_norm": 0.17915194232448164,
+      "learning_rate": 2.1125302036589575e-05,
+      "loss": 0.2438,
+      "step": 3990
+    },
+    {
+      "epoch": 3.723205964585275,
+      "grad_norm": 0.1776989301975763,
+      "learning_rate": 2.1082153952364517e-05,
+      "loss": 0.2404,
+      "step": 3995
+    },
+    {
+      "epoch": 3.7278657968313142,
+      "grad_norm": 0.16479423952194158,
+      "learning_rate": 2.1039005868139457e-05,
+      "loss": 0.2428,
+      "step": 4000
+    },
+    {
+      "epoch": 3.732525629077353,
+      "grad_norm": 0.16765042467844662,
+      "learning_rate": 2.0995857783914392e-05,
+      "loss": 0.2389,
+      "step": 4005
+    },
+    {
+      "epoch": 3.737185461323392,
+      "grad_norm": 0.1818779748740014,
+      "learning_rate": 2.0952709699689335e-05,
+      "loss": 0.2456,
+      "step": 4010
+    },
+    {
+      "epoch": 3.7418452935694315,
+      "grad_norm": 0.16832284546085408,
+      "learning_rate": 2.0909561615464274e-05,
+      "loss": 0.2388,
+      "step": 4015
+    },
+    {
+      "epoch": 3.746505125815471,
+      "grad_norm": 0.16674902473297593,
+      "learning_rate": 2.0866413531239216e-05,
+      "loss": 0.2447,
+      "step": 4020
+    },
+    {
+      "epoch": 3.75116495806151,
+      "grad_norm": 0.17285213216871598,
+      "learning_rate": 2.0823265447014152e-05,
+      "loss": 0.2466,
+      "step": 4025
+    },
+    {
+      "epoch": 3.755824790307549,
+      "grad_norm": 0.17571433599098618,
+      "learning_rate": 2.078011736278909e-05,
+      "loss": 0.2427,
+      "step": 4030
+    },
+    {
+      "epoch": 3.760484622553588,
+      "grad_norm": 0.18267251905511547,
+      "learning_rate": 2.0736969278564034e-05,
+      "loss": 0.2492,
+      "step": 4035
+    },
+    {
+      "epoch": 3.765144454799627,
+      "grad_norm": 0.17338917256855824,
+      "learning_rate": 2.0693821194338973e-05,
+      "loss": 0.241,
+      "step": 4040
+    },
+    {
+      "epoch": 3.7698042870456665,
+      "grad_norm": 0.17633421449991354,
+      "learning_rate": 2.0650673110113912e-05,
+      "loss": 0.2463,
+      "step": 4045
+    },
+    {
+      "epoch": 3.7744641192917054,
+      "grad_norm": 0.16372799835353208,
+      "learning_rate": 2.060752502588885e-05,
+      "loss": 0.2477,
+      "step": 4050
+    },
+    {
+      "epoch": 3.779123951537745,
+      "grad_norm": 0.18291265875387921,
+      "learning_rate": 2.056437694166379e-05,
+      "loss": 0.243,
+      "step": 4055
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 0.1846384305413761,
+      "learning_rate": 2.0521228857438733e-05,
+      "loss": 0.2438,
+      "step": 4060
+    },
+    {
+      "epoch": 3.7884436160298227,
+      "grad_norm": 0.1735590888130186,
+      "learning_rate": 2.0478080773213672e-05,
+      "loss": 0.2456,
+      "step": 4065
+    },
+    {
+      "epoch": 3.793103448275862,
+      "grad_norm": 0.1786172294620634,
+      "learning_rate": 2.0434932688988608e-05,
+      "loss": 0.2502,
+      "step": 4070
+    },
+    {
+      "epoch": 3.7977632805219015,
+      "grad_norm": 0.6289659086273054,
+      "learning_rate": 2.039178460476355e-05,
+      "loss": 0.2404,
+      "step": 4075
+    },
+    {
+      "epoch": 3.8024231127679404,
+      "grad_norm": 0.1810689675798759,
+      "learning_rate": 2.034863652053849e-05,
+      "loss": 0.2435,
+      "step": 4080
+    },
+    {
+      "epoch": 3.8070829450139794,
+      "grad_norm": 0.16298761691454225,
+      "learning_rate": 2.0305488436313428e-05,
+      "loss": 0.2429,
+      "step": 4085
+    },
+    {
+      "epoch": 3.8117427772600188,
+      "grad_norm": 0.17178327034225416,
+      "learning_rate": 2.0262340352088367e-05,
+      "loss": 0.2412,
+      "step": 4090
+    },
+    {
+      "epoch": 3.8164026095060577,
+      "grad_norm": 0.17594330528969726,
+      "learning_rate": 2.0219192267863306e-05,
+      "loss": 0.2464,
+      "step": 4095
+    },
+    {
+      "epoch": 3.821062441752097,
+      "grad_norm": 0.16811474010725866,
+      "learning_rate": 2.017604418363825e-05,
+      "loss": 0.2402,
+      "step": 4100
+    },
+    {
+      "epoch": 3.825722273998136,
+      "grad_norm": 0.18273481874299854,
+      "learning_rate": 2.0132896099413188e-05,
+      "loss": 0.2493,
+      "step": 4105
+    },
+    {
+      "epoch": 3.8303821062441754,
+      "grad_norm": 0.17305336891155346,
+      "learning_rate": 2.0089748015188127e-05,
+      "loss": 0.2413,
+      "step": 4110
+    },
+    {
+      "epoch": 3.8350419384902144,
+      "grad_norm": 0.17229977039271094,
+      "learning_rate": 2.0046599930963066e-05,
+      "loss": 0.2502,
+      "step": 4115
+    },
+    {
+      "epoch": 3.8397017707362533,
+      "grad_norm": 0.18162651366701693,
+      "learning_rate": 2.0003451846738005e-05,
+      "loss": 0.2439,
+      "step": 4120
+    },
+    {
+      "epoch": 3.8443616029822927,
+      "grad_norm": 0.17440571739371485,
+      "learning_rate": 1.9960303762512945e-05,
+      "loss": 0.2428,
+      "step": 4125
+    },
+    {
+      "epoch": 3.8490214352283316,
+      "grad_norm": 0.17987372561506818,
+      "learning_rate": 1.9917155678287887e-05,
+      "loss": 0.2459,
+      "step": 4130
+    },
+    {
+      "epoch": 3.853681267474371,
+      "grad_norm": 0.17870385201460737,
+      "learning_rate": 1.9874007594062823e-05,
+      "loss": 0.2448,
+      "step": 4135
+    },
+    {
+      "epoch": 3.85834109972041,
+      "grad_norm": 0.19188827995834903,
+      "learning_rate": 1.9830859509837765e-05,
+      "loss": 0.2446,
+      "step": 4140
+    },
+    {
+      "epoch": 3.8630009319664493,
+      "grad_norm": 0.1713701977888893,
+      "learning_rate": 1.9787711425612704e-05,
+      "loss": 0.2455,
+      "step": 4145
+    },
+    {
+      "epoch": 3.8676607642124883,
+      "grad_norm": 0.1659287276893002,
+      "learning_rate": 1.9744563341387643e-05,
+      "loss": 0.2413,
+      "step": 4150
+    },
+    {
+      "epoch": 3.872320596458527,
+      "grad_norm": 0.1773843681086169,
+      "learning_rate": 1.9701415257162583e-05,
+      "loss": 0.2459,
+      "step": 4155
+    },
+    {
+      "epoch": 3.8769804287045666,
+      "grad_norm": 0.17235610154056527,
+      "learning_rate": 1.9658267172937522e-05,
+      "loss": 0.2414,
+      "step": 4160
+    },
+    {
+      "epoch": 3.881640260950606,
+      "grad_norm": 0.17726878982183544,
+      "learning_rate": 1.961511908871246e-05,
+      "loss": 0.2434,
+      "step": 4165
+    },
+    {
+      "epoch": 3.886300093196645,
+      "grad_norm": 0.17693115120454214,
+      "learning_rate": 1.9571971004487403e-05,
+      "loss": 0.2397,
+      "step": 4170
+    },
+    {
+      "epoch": 3.890959925442684,
+      "grad_norm": 0.17679183961180675,
+      "learning_rate": 1.952882292026234e-05,
+      "loss": 0.248,
+      "step": 4175
+    },
+    {
+      "epoch": 3.8956197576887233,
+      "grad_norm": 0.17028096252230804,
+      "learning_rate": 1.948567483603728e-05,
+      "loss": 0.241,
+      "step": 4180
+    },
+    {
+      "epoch": 3.900279589934762,
+      "grad_norm": 0.18357263119812048,
+      "learning_rate": 1.944252675181222e-05,
+      "loss": 0.2475,
+      "step": 4185
+    },
+    {
+      "epoch": 3.9049394221808016,
+      "grad_norm": 0.17359189650285525,
+      "learning_rate": 1.939937866758716e-05,
+      "loss": 0.2461,
+      "step": 4190
+    },
+    {
+      "epoch": 3.9095992544268405,
+      "grad_norm": 0.17068861191647797,
+      "learning_rate": 1.93562305833621e-05,
+      "loss": 0.2398,
+      "step": 4195
+    },
+    {
+      "epoch": 3.91425908667288,
+      "grad_norm": 0.16977244734915517,
+      "learning_rate": 1.9313082499137038e-05,
+      "loss": 0.2405,
+      "step": 4200
+    },
+    {
+      "epoch": 3.918918918918919,
+      "grad_norm": 0.17538407646338203,
+      "learning_rate": 1.9269934414911977e-05,
+      "loss": 0.2408,
+      "step": 4205
+    },
+    {
+      "epoch": 3.923578751164958,
+      "grad_norm": 0.1801113456186898,
+      "learning_rate": 1.922678633068692e-05,
+      "loss": 0.2471,
+      "step": 4210
+    },
+    {
+      "epoch": 3.928238583410997,
+      "grad_norm": 0.18223962174172267,
+      "learning_rate": 1.918363824646186e-05,
+      "loss": 0.2482,
+      "step": 4215
+    },
+    {
+      "epoch": 3.9328984156570366,
+      "grad_norm": 0.18278563758016664,
+      "learning_rate": 1.9140490162236798e-05,
+      "loss": 0.2493,
+      "step": 4220
+    },
+    {
+      "epoch": 3.9375582479030755,
+      "grad_norm": 0.16796115440621007,
+      "learning_rate": 1.9097342078011737e-05,
+      "loss": 0.2431,
+      "step": 4225
+    },
+    {
+      "epoch": 3.9422180801491145,
+      "grad_norm": 0.18756729750732218,
+      "learning_rate": 1.9054193993786676e-05,
+      "loss": 0.2442,
+      "step": 4230
+    },
+    {
+      "epoch": 3.946877912395154,
+      "grad_norm": 0.17741995308016661,
+      "learning_rate": 1.901104590956162e-05,
+      "loss": 0.2489,
+      "step": 4235
+    },
+    {
+      "epoch": 3.951537744641193,
+      "grad_norm": 0.1700774040262742,
+      "learning_rate": 1.8967897825336554e-05,
+      "loss": 0.2374,
+      "step": 4240
+    },
+    {
+      "epoch": 3.956197576887232,
+      "grad_norm": 0.17604124186064382,
+      "learning_rate": 1.8924749741111493e-05,
+      "loss": 0.2464,
+      "step": 4245
+    },
+    {
+      "epoch": 3.960857409133271,
+      "grad_norm": 0.17405973754932416,
+      "learning_rate": 1.8881601656886436e-05,
+      "loss": 0.2381,
+      "step": 4250
+    },
+    {
+      "epoch": 3.9655172413793105,
+      "grad_norm": 0.18390350018702917,
+      "learning_rate": 1.8838453572661375e-05,
+      "loss": 0.2486,
+      "step": 4255
+    },
+    {
+      "epoch": 3.9701770736253494,
+      "grad_norm": 0.17342279187475101,
+      "learning_rate": 1.8795305488436314e-05,
+      "loss": 0.241,
+      "step": 4260
+    },
+    {
+      "epoch": 3.9748369058713884,
+      "grad_norm": 0.18104051865899928,
+      "learning_rate": 1.8752157404211253e-05,
+      "loss": 0.2417,
+      "step": 4265
+    },
+    {
+      "epoch": 3.9794967381174278,
+      "grad_norm": 0.18852601817601009,
+      "learning_rate": 1.8709009319986192e-05,
+      "loss": 0.247,
+      "step": 4270
+    },
+    {
+      "epoch": 3.984156570363467,
+      "grad_norm": 0.1738016973957961,
+      "learning_rate": 1.8665861235761135e-05,
+      "loss": 0.2372,
+      "step": 4275
+    },
+    {
+      "epoch": 3.988816402609506,
+      "grad_norm": 0.16612912684223705,
+      "learning_rate": 1.8622713151536074e-05,
+      "loss": 0.2422,
+      "step": 4280
+    },
+    {
+      "epoch": 3.993476234855545,
+      "grad_norm": 0.17125170875271,
+      "learning_rate": 1.857956506731101e-05,
+      "loss": 0.2388,
+      "step": 4285
+    },
+    {
+      "epoch": 3.9981360671015844,
+      "grad_norm": 0.17974828668145335,
+      "learning_rate": 1.8536416983085952e-05,
+      "loss": 0.247,
+      "step": 4290
+    },
+    {
+      "epoch": 4.002795899347624,
+      "grad_norm": 0.22127221128589092,
+      "learning_rate": 1.849326889886089e-05,
+      "loss": 0.2004,
+      "step": 4295
+    },
+    {
+      "epoch": 4.007455731593662,
+      "grad_norm": 0.2584514280454837,
+      "learning_rate": 1.8450120814635834e-05,
+      "loss": 0.172,
+      "step": 4300
+    },
+    {
+      "epoch": 4.012115563839702,
+      "grad_norm": 0.25454147569854596,
+      "learning_rate": 1.840697273041077e-05,
+      "loss": 0.1638,
+      "step": 4305
+    },
+    {
+      "epoch": 4.016775396085741,
+      "grad_norm": 0.21530889305803796,
+      "learning_rate": 1.836382464618571e-05,
+      "loss": 0.166,
+      "step": 4310
+    },
+    {
+      "epoch": 4.0214352283317805,
+      "grad_norm": 0.22061695259749842,
+      "learning_rate": 1.832067656196065e-05,
+      "loss": 0.1657,
+      "step": 4315
+    },
+    {
+      "epoch": 4.026095060577819,
+      "grad_norm": 0.1974575231863919,
+      "learning_rate": 1.827752847773559e-05,
+      "loss": 0.1658,
+      "step": 4320
+    },
+    {
+      "epoch": 4.030754892823858,
+      "grad_norm": 0.25902706727451763,
+      "learning_rate": 1.823438039351053e-05,
+      "loss": 0.1654,
+      "step": 4325
+    },
+    {
+      "epoch": 4.035414725069898,
+      "grad_norm": 0.1966646815598998,
+      "learning_rate": 1.819123230928547e-05,
+      "loss": 0.1652,
+      "step": 4330
+    },
+    {
+      "epoch": 4.040074557315936,
+      "grad_norm": 0.19747160344091164,
+      "learning_rate": 1.8148084225060408e-05,
+      "loss": 0.1635,
+      "step": 4335
+    },
+    {
+      "epoch": 4.044734389561976,
+      "grad_norm": 0.20556565489181802,
+      "learning_rate": 1.810493614083535e-05,
+      "loss": 0.163,
+      "step": 4340
+    },
+    {
+      "epoch": 4.049394221808015,
+      "grad_norm": 0.20071839296347127,
+      "learning_rate": 1.806178805661029e-05,
+      "loss": 0.1666,
+      "step": 4345
+    },
+    {
+      "epoch": 4.054054054054054,
+      "grad_norm": 0.20130752286509923,
+      "learning_rate": 1.8018639972385225e-05,
+      "loss": 0.1677,
+      "step": 4350
+    },
+    {
+      "epoch": 4.058713886300093,
+      "grad_norm": 0.19083838127338368,
+      "learning_rate": 1.7975491888160167e-05,
+      "loss": 0.1661,
+      "step": 4355
+    },
+    {
+      "epoch": 4.063373718546132,
+      "grad_norm": 0.19268739185173436,
+      "learning_rate": 1.7932343803935107e-05,
+      "loss": 0.1682,
+      "step": 4360
+    },
+    {
+      "epoch": 4.068033550792172,
+      "grad_norm": 0.22311504885584618,
+      "learning_rate": 1.7889195719710046e-05,
+      "loss": 0.1673,
+      "step": 4365
+    },
+    {
+      "epoch": 4.072693383038211,
+      "grad_norm": 0.20285816389621258,
+      "learning_rate": 1.7846047635484985e-05,
+      "loss": 0.1658,
+      "step": 4370
+    },
+    {
+      "epoch": 4.0773532152842495,
+      "grad_norm": 0.1844463766549495,
+      "learning_rate": 1.7802899551259924e-05,
+      "loss": 0.1618,
+      "step": 4375
+    },
+    {
+      "epoch": 4.082013047530289,
+      "grad_norm": 0.19552840707339897,
+      "learning_rate": 1.7759751467034866e-05,
+      "loss": 0.1669,
+      "step": 4380
+    },
+    {
+      "epoch": 4.086672879776328,
+      "grad_norm": 0.1946698839772263,
+      "learning_rate": 1.7716603382809806e-05,
+      "loss": 0.1686,
+      "step": 4385
+    },
+    {
+      "epoch": 4.091332712022367,
+      "grad_norm": 0.18766336791308946,
+      "learning_rate": 1.767345529858474e-05,
+      "loss": 0.1619,
+      "step": 4390
+    },
+    {
+      "epoch": 4.095992544268406,
+      "grad_norm": 0.18649086410112825,
+      "learning_rate": 1.7630307214359684e-05,
+      "loss": 0.1642,
+      "step": 4395
+    },
+    {
+      "epoch": 4.100652376514446,
+      "grad_norm": 0.1913680154423023,
+      "learning_rate": 1.7587159130134623e-05,
+      "loss": 0.1686,
+      "step": 4400
+    },
+    {
+      "epoch": 4.105312208760485,
+      "grad_norm": 0.18907836783187165,
+      "learning_rate": 1.7544011045909562e-05,
+      "loss": 0.1617,
+      "step": 4405
+    },
+    {
+      "epoch": 4.1099720410065235,
+      "grad_norm": 0.19739578700376867,
+      "learning_rate": 1.75008629616845e-05,
+      "loss": 0.166,
+      "step": 4410
+    },
+    {
+      "epoch": 4.114631873252563,
+      "grad_norm": 0.2037759705596476,
+      "learning_rate": 1.745771487745944e-05,
+      "loss": 0.1649,
+      "step": 4415
+    },
+    {
+      "epoch": 4.119291705498602,
+      "grad_norm": 0.1902813045174069,
+      "learning_rate": 1.7414566793234383e-05,
+      "loss": 0.1651,
+      "step": 4420
+    },
+    {
+      "epoch": 4.123951537744642,
+      "grad_norm": 0.20257508157918924,
+      "learning_rate": 1.7371418709009322e-05,
+      "loss": 0.1664,
+      "step": 4425
+    },
+    {
+      "epoch": 4.12861136999068,
+      "grad_norm": 0.19659878055481086,
+      "learning_rate": 1.732827062478426e-05,
+      "loss": 0.1652,
+      "step": 4430
+    },
+    {
+      "epoch": 4.1332712022367195,
+      "grad_norm": 0.21345754309857443,
+      "learning_rate": 1.72851225405592e-05,
+      "loss": 0.1671,
+      "step": 4435
+    },
+    {
+      "epoch": 4.137931034482759,
+      "grad_norm": 0.18888107048503852,
+      "learning_rate": 1.724197445633414e-05,
+      "loss": 0.1669,
+      "step": 4440
+    },
+    {
+      "epoch": 4.142590866728797,
+      "grad_norm": 0.19381663176560335,
+      "learning_rate": 1.7198826372109078e-05,
+      "loss": 0.1684,
+      "step": 4445
+    },
+    {
+      "epoch": 4.147250698974837,
+      "grad_norm": 0.19217131684468863,
+      "learning_rate": 1.715567828788402e-05,
+      "loss": 0.1651,
+      "step": 4450
+    },
+    {
+      "epoch": 4.151910531220876,
+      "grad_norm": 0.2009099425795485,
+      "learning_rate": 1.7112530203658957e-05,
+      "loss": 0.1656,
+      "step": 4455
+    },
+    {
+      "epoch": 4.1565703634669156,
+      "grad_norm": 0.1951381605902706,
+      "learning_rate": 1.70693821194339e-05,
+      "loss": 0.1663,
+      "step": 4460
+    },
+    {
+      "epoch": 4.161230195712954,
+      "grad_norm": 0.19057885704628816,
+      "learning_rate": 1.7026234035208838e-05,
+      "loss": 0.1637,
+      "step": 4465
+    },
+    {
+      "epoch": 4.165890027958993,
+      "grad_norm": 0.1883882629997917,
+      "learning_rate": 1.6983085950983777e-05,
+      "loss": 0.1653,
+      "step": 4470
+    },
+    {
+      "epoch": 4.170549860205033,
+      "grad_norm": 0.19086649334784356,
+      "learning_rate": 1.6939937866758716e-05,
+      "loss": 0.1624,
+      "step": 4475
+    },
+    {
+      "epoch": 4.175209692451071,
+      "grad_norm": 0.19839421436405796,
+      "learning_rate": 1.6896789782533655e-05,
+      "loss": 0.1654,
+      "step": 4480
+    },
+    {
+      "epoch": 4.179869524697111,
+      "grad_norm": 0.19762025651107734,
+      "learning_rate": 1.6853641698308595e-05,
+      "loss": 0.1672,
+      "step": 4485
+    },
+    {
+      "epoch": 4.18452935694315,
+      "grad_norm": 0.20375910235246886,
+      "learning_rate": 1.6810493614083537e-05,
+      "loss": 0.1657,
+      "step": 4490
+    },
+    {
+      "epoch": 4.1891891891891895,
+      "grad_norm": 0.19861606166598217,
+      "learning_rate": 1.6767345529858476e-05,
+      "loss": 0.1704,
+      "step": 4495
+    },
+    {
+      "epoch": 4.193849021435228,
+      "grad_norm": 0.2097411578260557,
+      "learning_rate": 1.6724197445633415e-05,
+      "loss": 0.1663,
+      "step": 4500
+    },
+    {
+      "epoch": 4.198508853681267,
+      "grad_norm": 0.18884637950525104,
+      "learning_rate": 1.6681049361408354e-05,
+      "loss": 0.1682,
+      "step": 4505
+    },
+    {
+      "epoch": 4.203168685927307,
+      "grad_norm": 0.1989441272054952,
+      "learning_rate": 1.6637901277183294e-05,
+      "loss": 0.1672,
+      "step": 4510
+    },
+    {
+      "epoch": 4.207828518173346,
+      "grad_norm": 0.20747901783784725,
+      "learning_rate": 1.6594753192958236e-05,
+      "loss": 0.1699,
+      "step": 4515
+    },
+    {
+      "epoch": 4.212488350419385,
+      "grad_norm": 0.20363740419099968,
+      "learning_rate": 1.6551605108733172e-05,
+      "loss": 0.1656,
+      "step": 4520
+    },
+    {
+      "epoch": 4.217148182665424,
+      "grad_norm": 0.19708635834240143,
+      "learning_rate": 1.650845702450811e-05,
+      "loss": 0.1695,
+      "step": 4525
+    },
+    {
+      "epoch": 4.221808014911463,
+      "grad_norm": 0.19023209568008417,
+      "learning_rate": 1.6465308940283053e-05,
+      "loss": 0.1656,
+      "step": 4530
+    },
+    {
+      "epoch": 4.226467847157502,
+      "grad_norm": 0.19284299096767635,
+      "learning_rate": 1.6422160856057992e-05,
+      "loss": 0.1695,
+      "step": 4535
+    },
+    {
+      "epoch": 4.231127679403541,
+      "grad_norm": 0.19614232977396992,
+      "learning_rate": 1.637901277183293e-05,
+      "loss": 0.1652,
+      "step": 4540
+    },
+    {
+      "epoch": 4.235787511649581,
+      "grad_norm": 0.19007667749881513,
+      "learning_rate": 1.633586468760787e-05,
+      "loss": 0.1686,
+      "step": 4545
+    },
+    {
+      "epoch": 4.24044734389562,
+      "grad_norm": 0.19365401637599486,
+      "learning_rate": 1.629271660338281e-05,
+      "loss": 0.1684,
+      "step": 4550
+    },
+    {
+      "epoch": 4.245107176141659,
+      "grad_norm": 0.19998126695054724,
+      "learning_rate": 1.6249568519157752e-05,
+      "loss": 0.1664,
+      "step": 4555
+    },
+    {
+      "epoch": 4.249767008387698,
+      "grad_norm": 0.1949899930381957,
+      "learning_rate": 1.620642043493269e-05,
+      "loss": 0.1679,
+      "step": 4560
+    },
+    {
+      "epoch": 4.254426840633737,
+      "grad_norm": 0.18863589718186724,
+      "learning_rate": 1.6163272350707627e-05,
+      "loss": 0.1665,
+      "step": 4565
+    },
+    {
+      "epoch": 4.259086672879777,
+      "grad_norm": 0.18993081246610233,
+      "learning_rate": 1.612012426648257e-05,
+      "loss": 0.1684,
+      "step": 4570
+    },
+    {
+      "epoch": 4.263746505125815,
+      "grad_norm": 0.19129033380442018,
+      "learning_rate": 1.607697618225751e-05,
+      "loss": 0.1677,
+      "step": 4575
+    },
+    {
+      "epoch": 4.268406337371855,
+      "grad_norm": 0.20139076609873743,
+      "learning_rate": 1.6033828098032448e-05,
+      "loss": 0.1639,
+      "step": 4580
+    },
+    {
+      "epoch": 4.273066169617894,
+      "grad_norm": 0.2039867141554659,
+      "learning_rate": 1.5990680013807387e-05,
+      "loss": 0.1641,
+      "step": 4585
+    },
+    {
+      "epoch": 4.2777260018639325,
+      "grad_norm": 0.1899080913111299,
+      "learning_rate": 1.5947531929582326e-05,
+      "loss": 0.1681,
+      "step": 4590
+    },
+    {
+      "epoch": 4.282385834109972,
+      "grad_norm": 0.1930301922542164,
+      "learning_rate": 1.590438384535727e-05,
+      "loss": 0.1722,
+      "step": 4595
+    },
+    {
+      "epoch": 4.287045666356011,
+      "grad_norm": 0.1922247012266805,
+      "learning_rate": 1.5861235761132208e-05,
+      "loss": 0.1656,
+      "step": 4600
+    },
+    {
+      "epoch": 4.291705498602051,
+      "grad_norm": 0.19786452984406597,
+      "learning_rate": 1.5818087676907143e-05,
+      "loss": 0.17,
+      "step": 4605
+    },
+    {
+      "epoch": 4.296365330848089,
+      "grad_norm": 0.19239375689796356,
+      "learning_rate": 1.5774939592682086e-05,
+      "loss": 0.1697,
+      "step": 4610
+    },
+    {
+      "epoch": 4.3010251630941285,
+      "grad_norm": 0.1887896348660511,
+      "learning_rate": 1.5731791508457025e-05,
+      "loss": 0.1665,
+      "step": 4615
+    },
+    {
+      "epoch": 4.305684995340168,
+      "grad_norm": 0.18953404236910507,
+      "learning_rate": 1.5688643424231964e-05,
+      "loss": 0.1639,
+      "step": 4620
+    },
+    {
+      "epoch": 4.310344827586207,
+      "grad_norm": 0.1894775321239529,
+      "learning_rate": 1.5645495340006903e-05,
+      "loss": 0.1687,
+      "step": 4625
+    },
+    {
+      "epoch": 4.315004659832246,
+      "grad_norm": 0.19566648846011822,
+      "learning_rate": 1.5602347255781842e-05,
+      "loss": 0.1686,
+      "step": 4630
+    },
+    {
+      "epoch": 4.319664492078285,
+      "grad_norm": 0.19855444724548998,
+      "learning_rate": 1.5559199171556785e-05,
+      "loss": 0.1695,
+      "step": 4635
+    },
+    {
+      "epoch": 4.324324324324325,
+      "grad_norm": 0.2002308008354795,
+      "learning_rate": 1.5516051087331724e-05,
+      "loss": 0.1648,
+      "step": 4640
+    },
+    {
+      "epoch": 4.328984156570363,
+      "grad_norm": 0.19356659217945885,
+      "learning_rate": 1.5472903003106663e-05,
+      "loss": 0.1653,
+      "step": 4645
+    },
+    {
+      "epoch": 4.3336439888164024,
+      "grad_norm": 0.18356546650925176,
+      "learning_rate": 1.5429754918881602e-05,
+      "loss": 0.1716,
+      "step": 4650
+    },
+    {
+      "epoch": 4.338303821062442,
+      "grad_norm": 0.19594821267228185,
+      "learning_rate": 1.538660683465654e-05,
+      "loss": 0.17,
+      "step": 4655
+    },
+    {
+      "epoch": 4.342963653308481,
+      "grad_norm": 0.18312564428026198,
+      "learning_rate": 1.534345875043148e-05,
+      "loss": 0.1664,
+      "step": 4660
+    },
+    {
+      "epoch": 4.34762348555452,
+      "grad_norm": 0.20831216776721556,
+      "learning_rate": 1.5300310666206423e-05,
+      "loss": 0.1682,
+      "step": 4665
+    },
+    {
+      "epoch": 4.352283317800559,
+      "grad_norm": 0.18546860917725988,
+      "learning_rate": 1.525716258198136e-05,
+      "loss": 0.1694,
+      "step": 4670
+    },
+    {
+      "epoch": 4.3569431500465985,
+      "grad_norm": 0.19268919616077404,
+      "learning_rate": 1.5214014497756301e-05,
+      "loss": 0.1674,
+      "step": 4675
+    },
+    {
+      "epoch": 4.361602982292638,
+      "grad_norm": 0.19708583387761888,
+      "learning_rate": 1.517086641353124e-05,
+      "loss": 0.1737,
+      "step": 4680
+    },
+    {
+      "epoch": 4.366262814538676,
+      "grad_norm": 0.19719351493587853,
+      "learning_rate": 1.5127718329306178e-05,
+      "loss": 0.1699,
+      "step": 4685
+    },
+    {
+      "epoch": 4.370922646784716,
+      "grad_norm": 0.19336281917633694,
+      "learning_rate": 1.508457024508112e-05,
+      "loss": 0.1694,
+      "step": 4690
+    },
+    {
+      "epoch": 4.375582479030755,
+      "grad_norm": 0.19136053003240552,
+      "learning_rate": 1.5041422160856058e-05,
+      "loss": 0.1685,
+      "step": 4695
+    },
+    {
+      "epoch": 4.380242311276794,
+      "grad_norm": 0.19575640578314735,
+      "learning_rate": 1.4998274076630997e-05,
+      "loss": 0.1645,
+      "step": 4700
+    },
+    {
+      "epoch": 4.384902143522833,
+      "grad_norm": 0.18744561677490928,
+      "learning_rate": 1.4955125992405938e-05,
+      "loss": 0.168,
+      "step": 4705
+    },
+    {
+      "epoch": 4.389561975768872,
+      "grad_norm": 0.19053532651050714,
+      "learning_rate": 1.4911977908180877e-05,
+      "loss": 0.1689,
+      "step": 4710
+    },
+    {
+      "epoch": 4.394221808014912,
+      "grad_norm": 0.18298821312357938,
+      "learning_rate": 1.4868829823955818e-05,
+      "loss": 0.169,
+      "step": 4715
+    },
+    {
+      "epoch": 4.39888164026095,
+      "grad_norm": 0.19298750614099838,
+      "learning_rate": 1.4825681739730757e-05,
+      "loss": 0.1644,
+      "step": 4720
+    },
+    {
+      "epoch": 4.40354147250699,
+      "grad_norm": 0.1923104097562937,
+      "learning_rate": 1.4782533655505696e-05,
+      "loss": 0.162,
+      "step": 4725
+    },
+    {
+      "epoch": 4.408201304753029,
+      "grad_norm": 0.19894820832150945,
+      "learning_rate": 1.4739385571280637e-05,
+      "loss": 0.166,
+      "step": 4730
+    },
+    {
+      "epoch": 4.412861136999068,
+      "grad_norm": 0.19790465663192633,
+      "learning_rate": 1.4696237487055576e-05,
+      "loss": 0.1648,
+      "step": 4735
+    },
+    {
+      "epoch": 4.417520969245107,
+      "grad_norm": 0.19086526188736963,
+      "learning_rate": 1.4653089402830513e-05,
+      "loss": 0.1681,
+      "step": 4740
+    },
+    {
+      "epoch": 4.422180801491146,
+      "grad_norm": 0.20150832035765703,
+      "learning_rate": 1.4609941318605456e-05,
+      "loss": 0.168,
+      "step": 4745
+    },
+    {
+      "epoch": 4.426840633737186,
+      "grad_norm": 0.20101999846091315,
+      "learning_rate": 1.4566793234380393e-05,
+      "loss": 0.1746,
+      "step": 4750
+    },
+    {
+      "epoch": 4.431500465983224,
+      "grad_norm": 0.19020207270586906,
+      "learning_rate": 1.4523645150155336e-05,
+      "loss": 0.1639,
+      "step": 4755
+    },
+    {
+      "epoch": 4.436160298229264,
+      "grad_norm": 0.1888017758081102,
+      "learning_rate": 1.4480497065930273e-05,
+      "loss": 0.1663,
+      "step": 4760
+    },
+    {
+      "epoch": 4.440820130475303,
+      "grad_norm": 0.1925701796535439,
+      "learning_rate": 1.4437348981705212e-05,
+      "loss": 0.1726,
+      "step": 4765
+    },
+    {
+      "epoch": 4.445479962721342,
+      "grad_norm": 0.1943968775621875,
+      "learning_rate": 1.4394200897480153e-05,
+      "loss": 0.1654,
+      "step": 4770
+    },
+    {
+      "epoch": 4.450139794967381,
+      "grad_norm": 0.19633741664249002,
+      "learning_rate": 1.4351052813255092e-05,
+      "loss": 0.1718,
+      "step": 4775
+    },
+    {
+      "epoch": 4.45479962721342,
+      "grad_norm": 0.19264106377527027,
+      "learning_rate": 1.4307904729030031e-05,
+      "loss": 0.1673,
+      "step": 4780
+    },
+    {
+      "epoch": 4.45945945945946,
+      "grad_norm": 0.1895556058665603,
+      "learning_rate": 1.4264756644804972e-05,
+      "loss": 0.1636,
+      "step": 4785
+    },
+    {
+      "epoch": 4.464119291705499,
+      "grad_norm": 0.20659688152159755,
+      "learning_rate": 1.4221608560579911e-05,
+      "loss": 0.1689,
+      "step": 4790
+    },
+    {
+      "epoch": 4.4687791239515375,
+      "grad_norm": 0.2264821369743272,
+      "learning_rate": 1.4178460476354852e-05,
+      "loss": 0.1675,
+      "step": 4795
+    },
+    {
+      "epoch": 4.473438956197577,
+      "grad_norm": 0.2006713742923623,
+      "learning_rate": 1.4135312392129791e-05,
+      "loss": 0.1709,
+      "step": 4800
+    },
+    {
+      "epoch": 4.478098788443616,
+      "grad_norm": 0.19630282572512783,
+      "learning_rate": 1.4092164307904728e-05,
+      "loss": 0.1641,
+      "step": 4805
+    },
+    {
+      "epoch": 4.482758620689655,
+      "grad_norm": 0.19434406280464486,
+      "learning_rate": 1.404901622367967e-05,
+      "loss": 0.1683,
+      "step": 4810
+    },
+    {
+      "epoch": 4.487418452935694,
+      "grad_norm": 0.20272935628438785,
+      "learning_rate": 1.4005868139454608e-05,
+      "loss": 0.1709,
+      "step": 4815
+    },
+    {
+      "epoch": 4.492078285181734,
+      "grad_norm": 0.20436914406256568,
+      "learning_rate": 1.3962720055229547e-05,
+      "loss": 0.1691,
+      "step": 4820
+    },
+    {
+      "epoch": 4.496738117427773,
+      "grad_norm": 0.20604016497442953,
+      "learning_rate": 1.3919571971004488e-05,
+      "loss": 0.1691,
+      "step": 4825
+    },
+    {
+      "epoch": 4.5013979496738115,
+      "grad_norm": 0.19864601061101336,
+      "learning_rate": 1.3876423886779427e-05,
+      "loss": 0.168,
+      "step": 4830
+    },
+    {
+      "epoch": 4.506057781919851,
+      "grad_norm": 0.19511168727307873,
+      "learning_rate": 1.3833275802554368e-05,
+      "loss": 0.1675,
+      "step": 4835
+    },
+    {
+      "epoch": 4.51071761416589,
+      "grad_norm": 0.19275527430346479,
+      "learning_rate": 1.3790127718329307e-05,
+      "loss": 0.1665,
+      "step": 4840
+    },
+    {
+      "epoch": 4.515377446411929,
+      "grad_norm": 0.1932607591464089,
+      "learning_rate": 1.3746979634104246e-05,
+      "loss": 0.1684,
+      "step": 4845
+    },
+    {
+      "epoch": 4.520037278657968,
+      "grad_norm": 0.19543125343348094,
+      "learning_rate": 1.3703831549879187e-05,
+      "loss": 0.1677,
+      "step": 4850
+    },
+    {
+      "epoch": 4.5246971109040075,
+      "grad_norm": 0.1996351024621225,
+      "learning_rate": 1.3660683465654126e-05,
+      "loss": 0.1662,
+      "step": 4855
+    },
+    {
+      "epoch": 4.529356943150047,
+      "grad_norm": 0.19960560834008312,
+      "learning_rate": 1.3617535381429064e-05,
+      "loss": 0.1687,
+      "step": 4860
+    },
+    {
+      "epoch": 4.534016775396085,
+      "grad_norm": 0.18772946753950534,
+      "learning_rate": 1.3574387297204006e-05,
+      "loss": 0.1724,
+      "step": 4865
+    },
+    {
+      "epoch": 4.538676607642125,
+      "grad_norm": 0.19162796496254633,
+      "learning_rate": 1.3531239212978944e-05,
+      "loss": 0.169,
+      "step": 4870
+    },
+    {
+      "epoch": 4.543336439888164,
+      "grad_norm": 0.21064418959547979,
+      "learning_rate": 1.3488091128753886e-05,
+      "loss": 0.1694,
+      "step": 4875
+    },
+    {
+      "epoch": 4.5479962721342035,
+      "grad_norm": 0.19125575178420415,
+      "learning_rate": 1.3444943044528824e-05,
+      "loss": 0.1696,
+      "step": 4880
+    },
+    {
+      "epoch": 4.552656104380242,
+      "grad_norm": 0.19157388010620663,
+      "learning_rate": 1.3401794960303763e-05,
+      "loss": 0.1675,
+      "step": 4885
+    },
+    {
+      "epoch": 4.557315936626281,
+      "grad_norm": 0.18730108030515977,
+      "learning_rate": 1.3358646876078703e-05,
+      "loss": 0.1636,
+      "step": 4890
+    },
+    {
+      "epoch": 4.561975768872321,
+      "grad_norm": 0.19345439842211715,
+      "learning_rate": 1.3315498791853643e-05,
+      "loss": 0.1706,
+      "step": 4895
+    },
+    {
+      "epoch": 4.56663560111836,
+      "grad_norm": 0.20033469669258958,
+      "learning_rate": 1.327235070762858e-05,
+      "loss": 0.1708,
+      "step": 4900
+    },
+    {
+      "epoch": 4.571295433364399,
+      "grad_norm": 0.19183795234386822,
+      "learning_rate": 1.3229202623403522e-05,
+      "loss": 0.1727,
+      "step": 4905
+    },
+    {
+      "epoch": 4.575955265610438,
+      "grad_norm": 0.18941466920797387,
+      "learning_rate": 1.318605453917846e-05,
+      "loss": 0.1679,
+      "step": 4910
+    },
+    {
+      "epoch": 4.5806150978564775,
+      "grad_norm": 0.18133665983081315,
+      "learning_rate": 1.3142906454953402e-05,
+      "loss": 0.1683,
+      "step": 4915
+    },
+    {
+      "epoch": 4.585274930102516,
+      "grad_norm": 0.20206203148872737,
+      "learning_rate": 1.309975837072834e-05,
+      "loss": 0.1687,
+      "step": 4920
+    },
+    {
+      "epoch": 4.589934762348555,
+      "grad_norm": 0.1984368388111654,
+      "learning_rate": 1.3056610286503279e-05,
+      "loss": 0.1692,
+      "step": 4925
+    },
+    {
+      "epoch": 4.594594594594595,
+      "grad_norm": 0.1937163064119492,
+      "learning_rate": 1.301346220227822e-05,
+      "loss": 0.1677,
+      "step": 4930
+    },
+    {
+      "epoch": 4.599254426840634,
+      "grad_norm": 0.1901143657250384,
+      "learning_rate": 1.2970314118053159e-05,
+      "loss": 0.1716,
+      "step": 4935
+    },
+    {
+      "epoch": 4.603914259086673,
+      "grad_norm": 0.1939934103037522,
+      "learning_rate": 1.2927166033828098e-05,
+      "loss": 0.1681,
+      "step": 4940
+    },
+    {
+      "epoch": 4.608574091332712,
+      "grad_norm": 0.18477377187104493,
+      "learning_rate": 1.2884017949603039e-05,
+      "loss": 0.1661,
+      "step": 4945
+    },
+    {
+      "epoch": 4.613233923578751,
+      "grad_norm": 0.18634001017746094,
+      "learning_rate": 1.2840869865377978e-05,
+      "loss": 0.1706,
+      "step": 4950
+    },
+    {
+      "epoch": 4.61789375582479,
+      "grad_norm": 0.19652492348889217,
+      "learning_rate": 1.2797721781152919e-05,
+      "loss": 0.1711,
+      "step": 4955
+    },
+    {
+      "epoch": 4.622553588070829,
+      "grad_norm": 0.19251244754759048,
+      "learning_rate": 1.2754573696927858e-05,
+      "loss": 0.1664,
+      "step": 4960
+    },
+    {
+      "epoch": 4.627213420316869,
+      "grad_norm": 0.18795100960164077,
+      "learning_rate": 1.2711425612702795e-05,
+      "loss": 0.1704,
+      "step": 4965
+    },
+    {
+      "epoch": 4.631873252562908,
+      "grad_norm": 0.1867369000401008,
+      "learning_rate": 1.2668277528477738e-05,
+      "loss": 0.1697,
+      "step": 4970
+    },
+    {
+      "epoch": 4.6365330848089465,
+      "grad_norm": 0.18724624164790069,
+      "learning_rate": 1.2625129444252675e-05,
+      "loss": 0.166,
+      "step": 4975
+    },
+    {
+      "epoch": 4.641192917054986,
+      "grad_norm": 0.19262724266902048,
+      "learning_rate": 1.2581981360027614e-05,
+      "loss": 0.1704,
+      "step": 4980
+    },
+    {
+      "epoch": 4.645852749301025,
+      "grad_norm": 0.1894637830400957,
+      "learning_rate": 1.2538833275802555e-05,
+      "loss": 0.1633,
+      "step": 4985
+    },
+    {
+      "epoch": 4.650512581547064,
+      "grad_norm": 0.19971516071811193,
+      "learning_rate": 1.2495685191577494e-05,
+      "loss": 0.1713,
+      "step": 4990
+    },
+    {
+      "epoch": 4.655172413793103,
+      "grad_norm": 0.19130291918443493,
+      "learning_rate": 1.2452537107352433e-05,
+      "loss": 0.1693,
+      "step": 4995
+    },
+    {
+      "epoch": 4.659832246039143,
+      "grad_norm": 0.1902584446560321,
+      "learning_rate": 1.2409389023127374e-05,
+      "loss": 0.1697,
+      "step": 5000
+    },
+    {
+      "epoch": 4.664492078285182,
+      "grad_norm": 0.18919250009392766,
+      "learning_rate": 1.2366240938902313e-05,
+      "loss": 0.1673,
+      "step": 5005
+    },
+    {
+      "epoch": 4.669151910531221,
+      "grad_norm": 0.19640932777574088,
+      "learning_rate": 1.2323092854677252e-05,
+      "loss": 0.1663,
+      "step": 5010
+    },
+    {
+      "epoch": 4.67381174277726,
+      "grad_norm": 0.1984671127943998,
+      "learning_rate": 1.2279944770452193e-05,
+      "loss": 0.1722,
+      "step": 5015
+    },
+    {
+      "epoch": 4.678471575023299,
+      "grad_norm": 0.1886848882572527,
+      "learning_rate": 1.2236796686227132e-05,
+      "loss": 0.1721,
+      "step": 5020
+    },
+    {
+      "epoch": 4.683131407269339,
+      "grad_norm": 0.19772808282886978,
+      "learning_rate": 1.2193648602002073e-05,
+      "loss": 0.1677,
+      "step": 5025
+    },
+    {
+      "epoch": 4.687791239515377,
+      "grad_norm": 0.18838202528401776,
+      "learning_rate": 1.215050051777701e-05,
+      "loss": 0.1688,
+      "step": 5030
+    },
+    {
+      "epoch": 4.6924510717614165,
+      "grad_norm": 0.19093743637683047,
+      "learning_rate": 1.2107352433551951e-05,
+      "loss": 0.171,
+      "step": 5035
+    },
+    {
+      "epoch": 4.697110904007456,
+      "grad_norm": 0.19439345615986633,
+      "learning_rate": 1.206420434932689e-05,
+      "loss": 0.1704,
+      "step": 5040
+    },
+    {
+      "epoch": 4.701770736253495,
+      "grad_norm": 0.1908768292019991,
+      "learning_rate": 1.202105626510183e-05,
+      "loss": 0.1674,
+      "step": 5045
+    },
+    {
+      "epoch": 4.706430568499534,
+      "grad_norm": 0.1900643916145769,
+      "learning_rate": 1.1977908180876769e-05,
+      "loss": 0.1734,
+      "step": 5050
+    },
+    {
+      "epoch": 4.711090400745573,
+      "grad_norm": 0.1948804617862936,
+      "learning_rate": 1.193476009665171e-05,
+      "loss": 0.1688,
+      "step": 5055
+    },
+    {
+      "epoch": 4.715750232991613,
+      "grad_norm": 0.1872605924282601,
+      "learning_rate": 1.1891612012426649e-05,
+      "loss": 0.1674,
+      "step": 5060
+    },
+    {
+      "epoch": 4.720410065237651,
+      "grad_norm": 0.20054638149418375,
+      "learning_rate": 1.1848463928201588e-05,
+      "loss": 0.1695,
+      "step": 5065
+    },
+    {
+      "epoch": 4.72506989748369,
+      "grad_norm": 0.189184532355845,
+      "learning_rate": 1.1805315843976528e-05,
+      "loss": 0.1652,
+      "step": 5070
+    },
+    {
+      "epoch": 4.72972972972973,
+      "grad_norm": 0.19910409309254315,
+      "learning_rate": 1.1762167759751468e-05,
+      "loss": 0.1724,
+      "step": 5075
+    },
+    {
+      "epoch": 4.734389561975769,
+      "grad_norm": 0.19311367289414105,
+      "learning_rate": 1.1719019675526408e-05,
+      "loss": 0.1705,
+      "step": 5080
+    },
+    {
+      "epoch": 4.739049394221808,
+      "grad_norm": 0.18346160536856643,
+      "learning_rate": 1.1675871591301346e-05,
+      "loss": 0.1652,
+      "step": 5085
+    },
+    {
+      "epoch": 4.743709226467847,
+      "grad_norm": 0.19927446486864325,
+      "learning_rate": 1.1632723507076287e-05,
+      "loss": 0.1655,
+      "step": 5090
+    },
+    {
+      "epoch": 4.7483690587138865,
+      "grad_norm": 0.18107884962830795,
+      "learning_rate": 1.1589575422851226e-05,
+      "loss": 0.1663,
+      "step": 5095
+    },
+    {
+      "epoch": 4.753028890959925,
+      "grad_norm": 0.19537113885063645,
+      "learning_rate": 1.1546427338626167e-05,
+      "loss": 0.1735,
+      "step": 5100
+    },
+    {
+      "epoch": 4.757688723205964,
+      "grad_norm": 0.20519980628812387,
+      "learning_rate": 1.1503279254401104e-05,
+      "loss": 0.1724,
+      "step": 5105
+    },
+    {
+      "epoch": 4.762348555452004,
+      "grad_norm": 0.20335701400617157,
+      "learning_rate": 1.1460131170176045e-05,
+      "loss": 0.1708,
+      "step": 5110
+    },
+    {
+      "epoch": 4.767008387698043,
+      "grad_norm": 0.1956160365725887,
+      "learning_rate": 1.1416983085950984e-05,
+      "loss": 0.1749,
+      "step": 5115
+    },
+    {
+      "epoch": 4.771668219944082,
+      "grad_norm": 0.18623275275420806,
+      "learning_rate": 1.1373835001725925e-05,
+      "loss": 0.1694,
+      "step": 5120
+    },
+    {
+      "epoch": 4.776328052190121,
+      "grad_norm": 0.19466321528731798,
+      "learning_rate": 1.1330686917500862e-05,
+      "loss": 0.1711,
+      "step": 5125
+    },
+    {
+      "epoch": 4.78098788443616,
+      "grad_norm": 0.19630175085990537,
+      "learning_rate": 1.1287538833275803e-05,
+      "loss": 0.1682,
+      "step": 5130
+    },
+    {
+      "epoch": 4.7856477166822,
+      "grad_norm": 0.18797651494844678,
+      "learning_rate": 1.1244390749050742e-05,
+      "loss": 0.1678,
+      "step": 5135
+    },
+    {
+      "epoch": 4.790307548928238,
+      "grad_norm": 0.18259644871037034,
+      "learning_rate": 1.1201242664825683e-05,
+      "loss": 0.1694,
+      "step": 5140
+    },
+    {
+      "epoch": 4.794967381174278,
+      "grad_norm": 0.1933172970018142,
+      "learning_rate": 1.1158094580600622e-05,
+      "loss": 0.167,
+      "step": 5145
+    },
+    {
+      "epoch": 4.799627213420317,
+      "grad_norm": 0.1850964735654659,
+      "learning_rate": 1.1114946496375561e-05,
+      "loss": 0.1685,
+      "step": 5150
+    },
+    {
+      "epoch": 4.8042870456663564,
+      "grad_norm": 0.18939557625817566,
+      "learning_rate": 1.1071798412150502e-05,
+      "loss": 0.1678,
+      "step": 5155
+    },
+    {
+      "epoch": 4.808946877912395,
+      "grad_norm": 0.1973076227331028,
+      "learning_rate": 1.1028650327925441e-05,
+      "loss": 0.1659,
+      "step": 5160
+    },
+    {
+      "epoch": 4.813606710158434,
+      "grad_norm": 0.1851938060170737,
+      "learning_rate": 1.098550224370038e-05,
+      "loss": 0.169,
+      "step": 5165
+    },
+    {
+      "epoch": 4.818266542404474,
+      "grad_norm": 0.17856786057172777,
+      "learning_rate": 1.094235415947532e-05,
+      "loss": 0.166,
+      "step": 5170
+    },
+    {
+      "epoch": 4.822926374650512,
+      "grad_norm": 0.18590696231601592,
+      "learning_rate": 1.089920607525026e-05,
+      "loss": 0.1663,
+      "step": 5175
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 0.18628126674632692,
+      "learning_rate": 1.0856057991025199e-05,
+      "loss": 0.1678,
+      "step": 5180
+    },
+    {
+      "epoch": 4.832246039142591,
+      "grad_norm": 0.20200282167703737,
+      "learning_rate": 1.0812909906800138e-05,
+      "loss": 0.1691,
+      "step": 5185
+    },
+    {
+      "epoch": 4.83690587138863,
+      "grad_norm": 0.18392093254905442,
+      "learning_rate": 1.0769761822575077e-05,
+      "loss": 0.1653,
+      "step": 5190
+    },
+    {
+      "epoch": 4.841565703634669,
+      "grad_norm": 0.19532944597023272,
+      "learning_rate": 1.0726613738350018e-05,
+      "loss": 0.1709,
+      "step": 5195
+    },
+    {
+      "epoch": 4.846225535880708,
+      "grad_norm": 0.1872944357427929,
+      "learning_rate": 1.0683465654124957e-05,
+      "loss": 0.1679,
+      "step": 5200
+    },
+    {
+      "epoch": 4.850885368126748,
+      "grad_norm": 0.19288381532560925,
+      "learning_rate": 1.0640317569899896e-05,
+      "loss": 0.1664,
+      "step": 5205
+    },
+    {
+      "epoch": 4.855545200372786,
+      "grad_norm": 0.19498094802321386,
+      "learning_rate": 1.0597169485674835e-05,
+      "loss": 0.1732,
+      "step": 5210
+    },
+    {
+      "epoch": 4.8602050326188255,
+      "grad_norm": 0.18157495486945277,
+      "learning_rate": 1.0554021401449776e-05,
+      "loss": 0.1705,
+      "step": 5215
+    },
+    {
+      "epoch": 4.864864864864865,
+      "grad_norm": 0.18875450387611267,
+      "learning_rate": 1.0510873317224715e-05,
+      "loss": 0.1692,
+      "step": 5220
+    },
+    {
+      "epoch": 4.869524697110904,
+      "grad_norm": 0.19635754606195516,
+      "learning_rate": 1.0467725232999655e-05,
+      "loss": 0.1709,
+      "step": 5225
+    },
+    {
+      "epoch": 4.874184529356943,
+      "grad_norm": 0.18932979033299804,
+      "learning_rate": 1.0424577148774595e-05,
+      "loss": 0.1663,
+      "step": 5230
+    },
+    {
+      "epoch": 4.878844361602982,
+      "grad_norm": 0.18982899791441588,
+      "learning_rate": 1.0381429064549534e-05,
+      "loss": 0.1694,
+      "step": 5235
+    },
+    {
+      "epoch": 4.883504193849022,
+      "grad_norm": 0.18566360718681515,
+      "learning_rate": 1.0338280980324475e-05,
+      "loss": 0.172,
+      "step": 5240
+    },
+    {
+      "epoch": 4.888164026095061,
+      "grad_norm": 0.18701870969241546,
+      "learning_rate": 1.0295132896099413e-05,
+      "loss": 0.17,
+      "step": 5245
+    },
+    {
+      "epoch": 4.8928238583410995,
+      "grad_norm": 0.19756905622281792,
+      "learning_rate": 1.0251984811874353e-05,
+      "loss": 0.1709,
+      "step": 5250
+    },
+    {
+      "epoch": 4.897483690587139,
+      "grad_norm": 0.19590894407604892,
+      "learning_rate": 1.0208836727649293e-05,
+      "loss": 0.1685,
+      "step": 5255
+    },
+    {
+      "epoch": 4.902143522833178,
+      "grad_norm": 0.1908492151619134,
+      "learning_rate": 1.0165688643424233e-05,
+      "loss": 0.1698,
+      "step": 5260
+    },
+    {
+      "epoch": 4.906803355079218,
+      "grad_norm": 0.19241916642629986,
+      "learning_rate": 1.012254055919917e-05,
+      "loss": 0.167,
+      "step": 5265
+    },
+    {
+      "epoch": 4.911463187325256,
+      "grad_norm": 0.19392646388504445,
+      "learning_rate": 1.0079392474974112e-05,
+      "loss": 0.1699,
+      "step": 5270
+    },
+    {
+      "epoch": 4.9161230195712955,
+      "grad_norm": 0.18360749491917772,
+      "learning_rate": 1.003624439074905e-05,
+      "loss": 0.169,
+      "step": 5275
+    },
+    {
+      "epoch": 4.920782851817335,
+      "grad_norm": 0.1819460136350066,
+      "learning_rate": 9.993096306523992e-06,
+      "loss": 0.1702,
+      "step": 5280
+    },
+    {
+      "epoch": 4.925442684063373,
+      "grad_norm": 0.19076171487454013,
+      "learning_rate": 9.94994822229893e-06,
+      "loss": 0.1649,
+      "step": 5285
+    },
+    {
+      "epoch": 4.930102516309413,
+      "grad_norm": 0.18846673775216605,
+      "learning_rate": 9.90680013807387e-06,
+      "loss": 0.1697,
+      "step": 5290
+    },
+    {
+      "epoch": 4.934762348555452,
+      "grad_norm": 0.18331394723845146,
+      "learning_rate": 9.86365205384881e-06,
+      "loss": 0.172,
+      "step": 5295
+    },
+    {
+      "epoch": 4.9394221808014915,
+      "grad_norm": 0.19760604747972085,
+      "learning_rate": 9.82050396962375e-06,
+      "loss": 0.1677,
+      "step": 5300
+    },
+    {
+      "epoch": 4.94408201304753,
+      "grad_norm": 0.18503128967697524,
+      "learning_rate": 9.777355885398689e-06,
+      "loss": 0.1711,
+      "step": 5305
+    },
+    {
+      "epoch": 4.948741845293569,
+      "grad_norm": 0.19245056273271058,
+      "learning_rate": 9.734207801173628e-06,
+      "loss": 0.1716,
+      "step": 5310
+    },
+    {
+      "epoch": 4.953401677539609,
+      "grad_norm": 0.19250607218526558,
+      "learning_rate": 9.691059716948569e-06,
+      "loss": 0.1695,
+      "step": 5315
+    },
+    {
+      "epoch": 4.958061509785647,
+      "grad_norm": 0.19749608997858184,
+      "learning_rate": 9.647911632723508e-06,
+      "loss": 0.1736,
+      "step": 5320
+    },
+    {
+      "epoch": 4.962721342031687,
+      "grad_norm": 0.19881578448270026,
+      "learning_rate": 9.604763548498447e-06,
+      "loss": 0.1706,
+      "step": 5325
+    },
+    {
+      "epoch": 4.967381174277726,
+      "grad_norm": 0.18656258254010424,
+      "learning_rate": 9.561615464273386e-06,
+      "loss": 0.1671,
+      "step": 5330
+    },
+    {
+      "epoch": 4.9720410065237655,
+      "grad_norm": 0.1915374391438505,
+      "learning_rate": 9.518467380048327e-06,
+      "loss": 0.1705,
+      "step": 5335
+    },
+    {
+      "epoch": 4.976700838769804,
+      "grad_norm": 0.1895051673973669,
+      "learning_rate": 9.475319295823266e-06,
+      "loss": 0.1709,
+      "step": 5340
+    },
+    {
+      "epoch": 4.981360671015843,
+      "grad_norm": 0.1914223058338611,
+      "learning_rate": 9.432171211598205e-06,
+      "loss": 0.1683,
+      "step": 5345
+    },
+    {
+      "epoch": 4.986020503261883,
+      "grad_norm": 0.18935057219138235,
+      "learning_rate": 9.389023127373144e-06,
+      "loss": 0.1643,
+      "step": 5350
+    },
+    {
+      "epoch": 4.990680335507921,
+      "grad_norm": 0.1888892695105428,
+      "learning_rate": 9.345875043148085e-06,
+      "loss": 0.1728,
+      "step": 5355
+    },
+    {
+      "epoch": 4.995340167753961,
+      "grad_norm": 0.18866390381515022,
+      "learning_rate": 9.302726958923024e-06,
+      "loss": 0.1672,
+      "step": 5360
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.18669508493513562,
+      "learning_rate": 9.259578874697963e-06,
+      "loss": 0.1694,
+      "step": 5365
+    },
+    {
+      "epoch": 5.004659832246039,
+      "grad_norm": 0.26084661299931095,
+      "learning_rate": 9.216430790472904e-06,
+      "loss": 0.1152,
+      "step": 5370
+    },
+    {
+      "epoch": 5.009319664492078,
+      "grad_norm": 0.2163143224159846,
+      "learning_rate": 9.173282706247843e-06,
+      "loss": 0.1137,
+      "step": 5375
+    },
+    {
+      "epoch": 5.013979496738117,
+      "grad_norm": 0.21332421115483335,
+      "learning_rate": 9.130134622022784e-06,
+      "loss": 0.1139,
+      "step": 5380
+    },
+    {
+      "epoch": 5.018639328984157,
+      "grad_norm": 0.20796161743376443,
+      "learning_rate": 9.086986537797721e-06,
+      "loss": 0.1108,
+      "step": 5385
+    },
+    {
+      "epoch": 5.023299161230196,
+      "grad_norm": 0.19698353468579985,
+      "learning_rate": 9.043838453572662e-06,
+      "loss": 0.1094,
+      "step": 5390
+    },
+    {
+      "epoch": 5.0279589934762345,
+      "grad_norm": 0.1955764881561846,
+      "learning_rate": 9.000690369347601e-06,
+      "loss": 0.1109,
+      "step": 5395
+    },
+    {
+      "epoch": 5.032618825722274,
+      "grad_norm": 0.18540020242147104,
+      "learning_rate": 8.957542285122542e-06,
+      "loss": 0.1067,
+      "step": 5400
+    },
+    {
+      "epoch": 5.037278657968313,
+      "grad_norm": 0.19576016779282843,
+      "learning_rate": 8.91439420089748e-06,
+      "loss": 0.11,
+      "step": 5405
+    },
+    {
+      "epoch": 5.041938490214353,
+      "grad_norm": 0.18546517950353272,
+      "learning_rate": 8.87124611667242e-06,
+      "loss": 0.1085,
+      "step": 5410
+    },
+    {
+      "epoch": 5.046598322460391,
+      "grad_norm": 0.188166470917478,
+      "learning_rate": 8.82809803244736e-06,
+      "loss": 0.1113,
+      "step": 5415
+    },
+    {
+      "epoch": 5.051258154706431,
+      "grad_norm": 0.18923544602153616,
+      "learning_rate": 8.7849499482223e-06,
+      "loss": 0.1089,
+      "step": 5420
+    },
+    {
+      "epoch": 5.05591798695247,
+      "grad_norm": 0.18924494579112888,
+      "learning_rate": 8.741801863997238e-06,
+      "loss": 0.1115,
+      "step": 5425
+    },
+    {
+      "epoch": 5.0605778191985085,
+      "grad_norm": 0.21162607694819963,
+      "learning_rate": 8.698653779772179e-06,
+      "loss": 0.1117,
+      "step": 5430
+    },
+    {
+      "epoch": 5.065237651444548,
+      "grad_norm": 0.19075058913317822,
+      "learning_rate": 8.655505695547118e-06,
+      "loss": 0.1092,
+      "step": 5435
+    },
+    {
+      "epoch": 5.069897483690587,
+      "grad_norm": 0.1979976118876634,
+      "learning_rate": 8.612357611322058e-06,
+      "loss": 0.1106,
+      "step": 5440
+    },
+    {
+      "epoch": 5.074557315936627,
+      "grad_norm": 0.19134239359593805,
+      "learning_rate": 8.569209527096998e-06,
+      "loss": 0.1079,
+      "step": 5445
+    },
+    {
+      "epoch": 5.079217148182665,
+      "grad_norm": 0.19353176180323886,
+      "learning_rate": 8.526061442871937e-06,
+      "loss": 0.1124,
+      "step": 5450
+    },
+    {
+      "epoch": 5.0838769804287045,
+      "grad_norm": 0.18205409800423908,
+      "learning_rate": 8.482913358646877e-06,
+      "loss": 0.1116,
+      "step": 5455
+    },
+    {
+      "epoch": 5.088536812674744,
+      "grad_norm": 0.19258329686657596,
+      "learning_rate": 8.439765274421817e-06,
+      "loss": 0.1081,
+      "step": 5460
+    },
+    {
+      "epoch": 5.093196644920783,
+      "grad_norm": 0.20130150576979933,
+      "learning_rate": 8.396617190196756e-06,
+      "loss": 0.1101,
+      "step": 5465
+    },
+    {
+      "epoch": 5.097856477166822,
+      "grad_norm": 0.17304866526288862,
+      "learning_rate": 8.353469105971695e-06,
+      "loss": 0.107,
+      "step": 5470
+    },
+    {
+      "epoch": 5.102516309412861,
+      "grad_norm": 0.19919930664308813,
+      "learning_rate": 8.310321021746636e-06,
+      "loss": 0.1144,
+      "step": 5475
+    },
+    {
+      "epoch": 5.1071761416589005,
+      "grad_norm": 0.18128825047755825,
+      "learning_rate": 8.267172937521575e-06,
+      "loss": 0.1082,
+      "step": 5480
+    },
+    {
+      "epoch": 5.111835973904939,
+      "grad_norm": 0.19520755045798827,
+      "learning_rate": 8.224024853296514e-06,
+      "loss": 0.1123,
+      "step": 5485
+    },
+    {
+      "epoch": 5.116495806150978,
+      "grad_norm": 0.19636509216304285,
+      "learning_rate": 8.180876769071453e-06,
+      "loss": 0.1098,
+      "step": 5490
+    },
+    {
+      "epoch": 5.121155638397018,
+      "grad_norm": 0.19056325716733158,
+      "learning_rate": 8.137728684846394e-06,
+      "loss": 0.1126,
+      "step": 5495
+    },
+    {
+      "epoch": 5.125815470643057,
+      "grad_norm": 0.19604709685974214,
+      "learning_rate": 8.094580600621333e-06,
+      "loss": 0.1124,
+      "step": 5500
+    },
+    {
+      "epoch": 5.130475302889096,
+      "grad_norm": 0.19444697342773062,
+      "learning_rate": 8.051432516396272e-06,
+      "loss": 0.1125,
+      "step": 5505
+    },
+    {
+      "epoch": 5.135135135135135,
+      "grad_norm": 0.18761438314389767,
+      "learning_rate": 8.008284432171211e-06,
+      "loss": 0.1121,
+      "step": 5510
+    },
+    {
+      "epoch": 5.1397949673811745,
+      "grad_norm": 0.19234916597098617,
+      "learning_rate": 7.965136347946152e-06,
+      "loss": 0.1109,
+      "step": 5515
+    },
+    {
+      "epoch": 5.144454799627214,
+      "grad_norm": 0.18170539482879025,
+      "learning_rate": 7.921988263721091e-06,
+      "loss": 0.1077,
+      "step": 5520
+    },
+    {
+      "epoch": 5.149114631873252,
+      "grad_norm": 0.19644587070825442,
+      "learning_rate": 7.87884017949603e-06,
+      "loss": 0.1107,
+      "step": 5525
+    },
+    {
+      "epoch": 5.153774464119292,
+      "grad_norm": 0.185942592108051,
+      "learning_rate": 7.835692095270971e-06,
+      "loss": 0.1088,
+      "step": 5530
+    },
+    {
+      "epoch": 5.158434296365331,
+      "grad_norm": 0.22344098638623303,
+      "learning_rate": 7.79254401104591e-06,
+      "loss": 0.1144,
+      "step": 5535
+    },
+    {
+      "epoch": 5.16309412861137,
+      "grad_norm": 0.18927110172737308,
+      "learning_rate": 7.749395926820851e-06,
+      "loss": 0.1121,
+      "step": 5540
+    },
+    {
+      "epoch": 5.167753960857409,
+      "grad_norm": 0.186731338930193,
+      "learning_rate": 7.706247842595788e-06,
+      "loss": 0.1122,
+      "step": 5545
+    },
+    {
+      "epoch": 5.172413793103448,
+      "grad_norm": 0.18721292512081325,
+      "learning_rate": 7.663099758370729e-06,
+      "loss": 0.1087,
+      "step": 5550
+    },
+    {
+      "epoch": 5.177073625349488,
+      "grad_norm": 0.1929232117920852,
+      "learning_rate": 7.619951674145669e-06,
+      "loss": 0.11,
+      "step": 5555
+    },
+    {
+      "epoch": 5.181733457595526,
+      "grad_norm": 0.18721086683573093,
+      "learning_rate": 7.576803589920608e-06,
+      "loss": 0.1134,
+      "step": 5560
+    },
+    {
+      "epoch": 5.186393289841566,
+      "grad_norm": 0.19406754634874676,
+      "learning_rate": 7.533655505695547e-06,
+      "loss": 0.1118,
+      "step": 5565
+    },
+    {
+      "epoch": 5.191053122087605,
+      "grad_norm": 0.18428076974807517,
+      "learning_rate": 7.490507421470487e-06,
+      "loss": 0.1105,
+      "step": 5570
+    },
+    {
+      "epoch": 5.1957129543336436,
+      "grad_norm": 0.19485090534169874,
+      "learning_rate": 7.447359337245427e-06,
+      "loss": 0.1099,
+      "step": 5575
+    },
+    {
+      "epoch": 5.200372786579683,
+      "grad_norm": 0.18589817377871912,
+      "learning_rate": 7.4042112530203655e-06,
+      "loss": 0.1096,
+      "step": 5580
+    },
+    {
+      "epoch": 5.205032618825722,
+      "grad_norm": 0.19011282286421433,
+      "learning_rate": 7.361063168795305e-06,
+      "loss": 0.1098,
+      "step": 5585
+    },
+    {
+      "epoch": 5.209692451071762,
+      "grad_norm": 0.1905147622206683,
+      "learning_rate": 7.317915084570245e-06,
+      "loss": 0.1098,
+      "step": 5590
+    },
+    {
+      "epoch": 5.2143522833178,
+      "grad_norm": 0.189574857113662,
+      "learning_rate": 7.274767000345185e-06,
+      "loss": 0.1087,
+      "step": 5595
+    },
+    {
+      "epoch": 5.21901211556384,
+      "grad_norm": 0.19524921966403216,
+      "learning_rate": 7.231618916120124e-06,
+      "loss": 0.1112,
+      "step": 5600
+    },
+    {
+      "epoch": 5.223671947809879,
+      "grad_norm": 0.18541710965904684,
+      "learning_rate": 7.1884708318950636e-06,
+      "loss": 0.1106,
+      "step": 5605
+    },
+    {
+      "epoch": 5.228331780055918,
+      "grad_norm": 0.18821900561364988,
+      "learning_rate": 7.1453227476700035e-06,
+      "loss": 0.1089,
+      "step": 5610
+    },
+    {
+      "epoch": 5.232991612301957,
+      "grad_norm": 0.19600537555981504,
+      "learning_rate": 7.1021746634449435e-06,
+      "loss": 0.1102,
+      "step": 5615
+    },
+    {
+      "epoch": 5.237651444547996,
+      "grad_norm": 0.18967322530499686,
+      "learning_rate": 7.059026579219883e-06,
+      "loss": 0.1128,
+      "step": 5620
+    },
+    {
+      "epoch": 5.242311276794036,
+      "grad_norm": 0.19879899222093034,
+      "learning_rate": 7.0158784949948226e-06,
+      "loss": 0.1113,
+      "step": 5625
+    },
+    {
+      "epoch": 5.246971109040074,
+      "grad_norm": 0.1956950880048718,
+      "learning_rate": 6.9727304107697625e-06,
+      "loss": 0.1126,
+      "step": 5630
+    },
+    {
+      "epoch": 5.2516309412861135,
+      "grad_norm": 0.1945444768014536,
+      "learning_rate": 6.9295823265447025e-06,
+      "loss": 0.1138,
+      "step": 5635
+    },
+    {
+      "epoch": 5.256290773532153,
+      "grad_norm": 0.1963116353897705,
+      "learning_rate": 6.886434242319641e-06,
+      "loss": 0.1108,
+      "step": 5640
+    },
+    {
+      "epoch": 5.260950605778192,
+      "grad_norm": 0.1895519865735857,
+      "learning_rate": 6.843286158094581e-06,
+      "loss": 0.1096,
+      "step": 5645
+    },
+    {
+      "epoch": 5.265610438024231,
+      "grad_norm": 0.18703098114102523,
+      "learning_rate": 6.800138073869521e-06,
+      "loss": 0.1116,
+      "step": 5650
+    },
+    {
+      "epoch": 5.27027027027027,
+      "grad_norm": 0.1928971690558433,
+      "learning_rate": 6.756989989644461e-06,
+      "loss": 0.1113,
+      "step": 5655
+    },
+    {
+      "epoch": 5.27493010251631,
+      "grad_norm": 0.19120637931005804,
+      "learning_rate": 6.713841905419399e-06,
+      "loss": 0.1116,
+      "step": 5660
+    },
+    {
+      "epoch": 5.279589934762349,
+      "grad_norm": 0.19442107906254544,
+      "learning_rate": 6.670693821194339e-06,
+      "loss": 0.1098,
+      "step": 5665
+    },
+    {
+      "epoch": 5.284249767008387,
+      "grad_norm": 0.18940590131013618,
+      "learning_rate": 6.627545736969279e-06,
+      "loss": 0.106,
+      "step": 5670
+    },
+    {
+      "epoch": 5.288909599254427,
+      "grad_norm": 0.1883145235792858,
+      "learning_rate": 6.584397652744219e-06,
+      "loss": 0.1143,
+      "step": 5675
+    },
+    {
+      "epoch": 5.293569431500466,
+      "grad_norm": 0.19270738453639655,
+      "learning_rate": 6.541249568519157e-06,
+      "loss": 0.1123,
+      "step": 5680
+    },
+    {
+      "epoch": 5.298229263746505,
+      "grad_norm": 0.19473687895554245,
+      "learning_rate": 6.498101484294097e-06,
+      "loss": 0.1079,
+      "step": 5685
+    },
+    {
+      "epoch": 5.302889095992544,
+      "grad_norm": 0.1964386528389667,
+      "learning_rate": 6.454953400069037e-06,
+      "loss": 0.1094,
+      "step": 5690
+    },
+    {
+      "epoch": 5.3075489282385835,
+      "grad_norm": 0.18934048532394884,
+      "learning_rate": 6.411805315843977e-06,
+      "loss": 0.108,
+      "step": 5695
+    },
+    {
+      "epoch": 5.312208760484623,
+      "grad_norm": 0.20202840031733965,
+      "learning_rate": 6.368657231618916e-06,
+      "loss": 0.1097,
+      "step": 5700
+    },
+    {
+      "epoch": 5.316868592730661,
+      "grad_norm": 0.1874212195367486,
+      "learning_rate": 6.325509147393856e-06,
+      "loss": 0.1124,
+      "step": 5705
+    },
+    {
+      "epoch": 5.321528424976701,
+      "grad_norm": 0.19531759889204642,
+      "learning_rate": 6.282361063168796e-06,
+      "loss": 0.111,
+      "step": 5710
+    },
+    {
+      "epoch": 5.32618825722274,
+      "grad_norm": 0.18767423632127567,
+      "learning_rate": 6.239212978943735e-06,
+      "loss": 0.1121,
+      "step": 5715
+    },
+    {
+      "epoch": 5.3308480894687795,
+      "grad_norm": 0.1910831532620871,
+      "learning_rate": 6.196064894718675e-06,
+      "loss": 0.1097,
+      "step": 5720
+    },
+    {
+      "epoch": 5.335507921714818,
+      "grad_norm": 0.19613772675458707,
+      "learning_rate": 6.152916810493614e-06,
+      "loss": 0.1095,
+      "step": 5725
+    },
+    {
+      "epoch": 5.340167753960857,
+      "grad_norm": 0.19046986901334315,
+      "learning_rate": 6.109768726268554e-06,
+      "loss": 0.1113,
+      "step": 5730
+    },
+    {
+      "epoch": 5.344827586206897,
+      "grad_norm": 0.19199024386903832,
+      "learning_rate": 6.066620642043493e-06,
+      "loss": 0.1137,
+      "step": 5735
+    },
+    {
+      "epoch": 5.349487418452935,
+      "grad_norm": 0.19526119019780067,
+      "learning_rate": 6.023472557818433e-06,
+      "loss": 0.1107,
+      "step": 5740
+    },
+    {
+      "epoch": 5.354147250698975,
+      "grad_norm": 0.191036748390845,
+      "learning_rate": 5.980324473593372e-06,
+      "loss": 0.1097,
+      "step": 5745
+    },
+    {
+      "epoch": 5.358807082945014,
+      "grad_norm": 0.18743318501635242,
+      "learning_rate": 5.937176389368312e-06,
+      "loss": 0.1093,
+      "step": 5750
+    },
+    {
+      "epoch": 5.3634669151910535,
+      "grad_norm": 0.18459565932018152,
+      "learning_rate": 5.894028305143251e-06,
+      "loss": 0.1114,
+      "step": 5755
+    },
+    {
+      "epoch": 5.368126747437092,
+      "grad_norm": 0.19127413420508943,
+      "learning_rate": 5.850880220918191e-06,
+      "loss": 0.1091,
+      "step": 5760
+    },
+    {
+      "epoch": 5.372786579683131,
+      "grad_norm": 0.20078199024773524,
+      "learning_rate": 5.807732136693131e-06,
+      "loss": 0.1105,
+      "step": 5765
+    },
+    {
+      "epoch": 5.377446411929171,
+      "grad_norm": 0.1974403526837057,
+      "learning_rate": 5.76458405246807e-06,
+      "loss": 0.1108,
+      "step": 5770
+    },
+    {
+      "epoch": 5.38210624417521,
+      "grad_norm": 0.19832050433471762,
+      "learning_rate": 5.72143596824301e-06,
+      "loss": 0.11,
+      "step": 5775
+    },
+    {
+      "epoch": 5.386766076421249,
+      "grad_norm": 0.19256437503131105,
+      "learning_rate": 5.67828788401795e-06,
+      "loss": 0.1112,
+      "step": 5780
+    },
+    {
+      "epoch": 5.391425908667288,
+      "grad_norm": 0.18771544661256917,
+      "learning_rate": 5.6351397997928894e-06,
+      "loss": 0.1065,
+      "step": 5785
+    },
+    {
+      "epoch": 5.396085740913327,
+      "grad_norm": 0.18571182384228033,
+      "learning_rate": 5.591991715567829e-06,
+      "loss": 0.1125,
+      "step": 5790
+    },
+    {
+      "epoch": 5.400745573159366,
+      "grad_norm": 0.1936797805157484,
+      "learning_rate": 5.5488436313427685e-06,
+      "loss": 0.1142,
+      "step": 5795
+    },
+    {
+      "epoch": 5.405405405405405,
+      "grad_norm": 0.18724632228463917,
+      "learning_rate": 5.5056955471177085e-06,
+      "loss": 0.111,
+      "step": 5800
+    },
+    {
+      "epoch": 5.410065237651445,
+      "grad_norm": 0.18184941175147953,
+      "learning_rate": 5.462547462892648e-06,
+      "loss": 0.1067,
+      "step": 5805
+    },
+    {
+      "epoch": 5.414725069897484,
+      "grad_norm": 0.19289237880580454,
+      "learning_rate": 5.4193993786675876e-06,
+      "loss": 0.113,
+      "step": 5810
+    },
+    {
+      "epoch": 5.4193849021435225,
+      "grad_norm": 0.19080499066020176,
+      "learning_rate": 5.376251294442527e-06,
+      "loss": 0.1087,
+      "step": 5815
+    },
+    {
+      "epoch": 5.424044734389562,
+      "grad_norm": 0.195432772414175,
+      "learning_rate": 5.333103210217467e-06,
+      "loss": 0.1107,
+      "step": 5820
+    },
+    {
+      "epoch": 5.428704566635601,
+      "grad_norm": 0.18221090111762125,
+      "learning_rate": 5.289955125992406e-06,
+      "loss": 0.113,
+      "step": 5825
+    },
+    {
+      "epoch": 5.43336439888164,
+      "grad_norm": 0.18528053155701307,
+      "learning_rate": 5.246807041767346e-06,
+      "loss": 0.1124,
+      "step": 5830
+    },
+    {
+      "epoch": 5.438024231127679,
+      "grad_norm": 0.19395176877815692,
+      "learning_rate": 5.203658957542285e-06,
+      "loss": 0.1114,
+      "step": 5835
+    },
+    {
+      "epoch": 5.442684063373719,
+      "grad_norm": 0.18932110605866143,
+      "learning_rate": 5.160510873317225e-06,
+      "loss": 0.1102,
+      "step": 5840
+    },
+    {
+      "epoch": 5.447343895619758,
+      "grad_norm": 0.19064195327253813,
+      "learning_rate": 5.117362789092165e-06,
+      "loss": 0.1104,
+      "step": 5845
+    },
+    {
+      "epoch": 5.4520037278657965,
+      "grad_norm": 0.19966251744496039,
+      "learning_rate": 5.074214704867105e-06,
+      "loss": 0.1096,
+      "step": 5850
+    },
+    {
+      "epoch": 5.456663560111836,
+      "grad_norm": 0.18889287506266514,
+      "learning_rate": 5.031066620642044e-06,
+      "loss": 0.112,
+      "step": 5855
+    },
+    {
+      "epoch": 5.461323392357875,
+      "grad_norm": 0.18884540230737232,
+      "learning_rate": 4.987918536416984e-06,
+      "loss": 0.1075,
+      "step": 5860
+    },
+    {
+      "epoch": 5.465983224603915,
+      "grad_norm": 0.19587450190618333,
+      "learning_rate": 4.944770452191923e-06,
+      "loss": 0.1052,
+      "step": 5865
+    },
+    {
+      "epoch": 5.470643056849953,
+      "grad_norm": 0.19626674120131438,
+      "learning_rate": 4.901622367966863e-06,
+      "loss": 0.1086,
+      "step": 5870
+    },
+    {
+      "epoch": 5.4753028890959925,
+      "grad_norm": 0.18788875591714232,
+      "learning_rate": 4.858474283741802e-06,
+      "loss": 0.1107,
+      "step": 5875
+    },
+    {
+      "epoch": 5.479962721342032,
+      "grad_norm": 0.19293010013020173,
+      "learning_rate": 4.815326199516742e-06,
+      "loss": 0.1111,
+      "step": 5880
+    },
+    {
+      "epoch": 5.484622553588071,
+      "grad_norm": 0.18642835454050677,
+      "learning_rate": 4.772178115291681e-06,
+      "loss": 0.1077,
+      "step": 5885
+    },
+    {
+      "epoch": 5.48928238583411,
+      "grad_norm": 0.19357378100687805,
+      "learning_rate": 4.729030031066621e-06,
+      "loss": 0.1118,
+      "step": 5890
+    },
+    {
+      "epoch": 5.493942218080149,
+      "grad_norm": 0.19233706722531116,
+      "learning_rate": 4.68588194684156e-06,
+      "loss": 0.1119,
+      "step": 5895
+    },
+    {
+      "epoch": 5.4986020503261885,
+      "grad_norm": 0.18934257460071613,
+      "learning_rate": 4.6427338626165e-06,
+      "loss": 0.109,
+      "step": 5900
+    },
+    {
+      "epoch": 5.503261882572227,
+      "grad_norm": 0.18835634062740506,
+      "learning_rate": 4.599585778391439e-06,
+      "loss": 0.1093,
+      "step": 5905
+    },
+    {
+      "epoch": 5.507921714818266,
+      "grad_norm": 0.1803569285666244,
+      "learning_rate": 4.556437694166379e-06,
+      "loss": 0.1127,
+      "step": 5910
+    },
+    {
+      "epoch": 5.512581547064306,
+      "grad_norm": 0.18011848077935255,
+      "learning_rate": 4.513289609941319e-06,
+      "loss": 0.1108,
+      "step": 5915
+    },
+    {
+      "epoch": 5.517241379310345,
+      "grad_norm": 0.18801360152472635,
+      "learning_rate": 4.470141525716258e-06,
+      "loss": 0.1122,
+      "step": 5920
+    },
+    {
+      "epoch": 5.521901211556384,
+      "grad_norm": 0.20605747977783032,
+      "learning_rate": 4.426993441491198e-06,
+      "loss": 0.113,
+      "step": 5925
+    },
+    {
+      "epoch": 5.526561043802423,
+      "grad_norm": 0.18217610415910357,
+      "learning_rate": 4.383845357266138e-06,
+      "loss": 0.111,
+      "step": 5930
+    },
+    {
+      "epoch": 5.5312208760484625,
+      "grad_norm": 0.19085596158295426,
+      "learning_rate": 4.340697273041077e-06,
+      "loss": 0.1088,
+      "step": 5935
+    },
+    {
+      "epoch": 5.535880708294501,
+      "grad_norm": 0.19481488950705642,
+      "learning_rate": 4.297549188816017e-06,
+      "loss": 0.1083,
+      "step": 5940
+    },
+    {
+      "epoch": 5.54054054054054,
+      "grad_norm": 0.19021567735743644,
+      "learning_rate": 4.254401104590956e-06,
+      "loss": 0.1126,
+      "step": 5945
+    },
+    {
+      "epoch": 5.54520037278658,
+      "grad_norm": 0.18628551936388948,
+      "learning_rate": 4.211253020365896e-06,
+      "loss": 0.1074,
+      "step": 5950
+    },
+    {
+      "epoch": 5.549860205032619,
+      "grad_norm": 0.18551309875932837,
+      "learning_rate": 4.168104936140835e-06,
+      "loss": 0.1094,
+      "step": 5955
+    },
+    {
+      "epoch": 5.554520037278658,
+      "grad_norm": 0.18423380733297137,
+      "learning_rate": 4.124956851915775e-06,
+      "loss": 0.1108,
+      "step": 5960
+    },
+    {
+      "epoch": 5.559179869524697,
+      "grad_norm": 0.18980712272128641,
+      "learning_rate": 4.0818087676907145e-06,
+      "loss": 0.1118,
+      "step": 5965
+    },
+    {
+      "epoch": 5.563839701770736,
+      "grad_norm": 0.18255253978500785,
+      "learning_rate": 4.0386606834656544e-06,
+      "loss": 0.108,
+      "step": 5970
+    },
+    {
+      "epoch": 5.568499534016776,
+      "grad_norm": 0.17916719499595,
+      "learning_rate": 3.9955125992405935e-06,
+      "loss": 0.1118,
+      "step": 5975
+    },
+    {
+      "epoch": 5.573159366262814,
+      "grad_norm": 0.1831167783491531,
+      "learning_rate": 3.9523645150155335e-06,
+      "loss": 0.1115,
+      "step": 5980
+    },
+    {
+      "epoch": 5.577819198508854,
+      "grad_norm": 0.18507008472366945,
+      "learning_rate": 3.909216430790473e-06,
+      "loss": 0.1087,
+      "step": 5985
+    },
+    {
+      "epoch": 5.582479030754893,
+      "grad_norm": 0.18876081489336802,
+      "learning_rate": 3.8660683465654126e-06,
+      "loss": 0.1102,
+      "step": 5990
+    },
+    {
+      "epoch": 5.587138863000932,
+      "grad_norm": 0.18062202136064281,
+      "learning_rate": 3.8229202623403525e-06,
+      "loss": 0.1095,
+      "step": 5995
+    },
+    {
+      "epoch": 5.591798695246971,
+      "grad_norm": 0.18607246460941362,
+      "learning_rate": 3.779772178115292e-06,
+      "loss": 0.1108,
+      "step": 6000
+    },
+    {
+      "epoch": 5.59645852749301,
+      "grad_norm": 0.1865686000569821,
+      "learning_rate": 3.736624093890231e-06,
+      "loss": 0.1073,
+      "step": 6005
+    },
+    {
+      "epoch": 5.60111835973905,
+      "grad_norm": 0.1922576586269823,
+      "learning_rate": 3.693476009665171e-06,
+      "loss": 0.1083,
+      "step": 6010
+    },
+    {
+      "epoch": 5.605778191985088,
+      "grad_norm": 0.1893476983705235,
+      "learning_rate": 3.6503279254401107e-06,
+      "loss": 0.1129,
+      "step": 6015
+    },
+    {
+      "epoch": 5.610438024231128,
+      "grad_norm": 0.20199636907601048,
+      "learning_rate": 3.6071798412150506e-06,
+      "loss": 0.1116,
+      "step": 6020
+    },
+    {
+      "epoch": 5.615097856477167,
+      "grad_norm": 0.18596745254968966,
+      "learning_rate": 3.5640317569899898e-06,
+      "loss": 0.1125,
+      "step": 6025
+    },
+    {
+      "epoch": 5.619757688723206,
+      "grad_norm": 0.19574687506018168,
+      "learning_rate": 3.5208836727649297e-06,
+      "loss": 0.1105,
+      "step": 6030
+    },
+    {
+      "epoch": 5.624417520969245,
+      "grad_norm": 0.1895332032398572,
+      "learning_rate": 3.477735588539869e-06,
+      "loss": 0.1104,
+      "step": 6035
+    },
+    {
+      "epoch": 5.629077353215284,
+      "grad_norm": 0.18636834683097744,
+      "learning_rate": 3.434587504314809e-06,
+      "loss": 0.111,
+      "step": 6040
+    },
+    {
+      "epoch": 5.633737185461324,
+      "grad_norm": 0.18061918218183914,
+      "learning_rate": 3.3914394200897483e-06,
+      "loss": 0.1117,
+      "step": 6045
+    },
+    {
+      "epoch": 5.638397017707362,
+      "grad_norm": 0.18689608431668087,
+      "learning_rate": 3.348291335864688e-06,
+      "loss": 0.109,
+      "step": 6050
+    },
+    {
+      "epoch": 5.6430568499534015,
+      "grad_norm": 0.18877691950307657,
+      "learning_rate": 3.3051432516396274e-06,
+      "loss": 0.1097,
+      "step": 6055
+    },
+    {
+      "epoch": 5.647716682199441,
+      "grad_norm": 0.19305741084177863,
+      "learning_rate": 3.2619951674145674e-06,
+      "loss": 0.1077,
+      "step": 6060
+    },
+    {
+      "epoch": 5.65237651444548,
+      "grad_norm": 0.19086284647205096,
+      "learning_rate": 3.2188470831895065e-06,
+      "loss": 0.1057,
+      "step": 6065
+    },
+    {
+      "epoch": 5.657036346691519,
+      "grad_norm": 0.18658262827246316,
+      "learning_rate": 3.1756989989644464e-06,
+      "loss": 0.1075,
+      "step": 6070
+    },
+    {
+      "epoch": 5.661696178937558,
+      "grad_norm": 0.19926020645148987,
+      "learning_rate": 3.1325509147393856e-06,
+      "loss": 0.1087,
+      "step": 6075
+    },
+    {
+      "epoch": 5.6663560111835976,
+      "grad_norm": 0.19349025995686786,
+      "learning_rate": 3.089402830514325e-06,
+      "loss": 0.1066,
+      "step": 6080
+    },
+    {
+      "epoch": 5.671015843429636,
+      "grad_norm": 0.1933648834033394,
+      "learning_rate": 3.046254746289265e-06,
+      "loss": 0.1084,
+      "step": 6085
+    },
+    {
+      "epoch": 5.675675675675675,
+      "grad_norm": 0.18434772343863393,
+      "learning_rate": 3.0031066620642046e-06,
+      "loss": 0.1078,
+      "step": 6090
+    },
+    {
+      "epoch": 5.680335507921715,
+      "grad_norm": 0.19446074772313918,
+      "learning_rate": 2.959958577839144e-06,
+      "loss": 0.1088,
+      "step": 6095
+    },
+    {
+      "epoch": 5.684995340167754,
+      "grad_norm": 0.1899690356848603,
+      "learning_rate": 2.9168104936140837e-06,
+      "loss": 0.1098,
+      "step": 6100
+    },
+    {
+      "epoch": 5.689655172413794,
+      "grad_norm": 0.18704312181293242,
+      "learning_rate": 2.873662409389023e-06,
+      "loss": 0.1125,
+      "step": 6105
+    },
+    {
+      "epoch": 5.694315004659832,
+      "grad_norm": 0.19060251253823574,
+      "learning_rate": 2.8305143251639627e-06,
+      "loss": 0.1093,
+      "step": 6110
+    },
+    {
+      "epoch": 5.6989748369058715,
+      "grad_norm": 0.18007639551981863,
+      "learning_rate": 2.7873662409389023e-06,
+      "loss": 0.1101,
+      "step": 6115
+    },
+    {
+      "epoch": 5.703634669151911,
+      "grad_norm": 0.18529227030870118,
+      "learning_rate": 2.7442181567138422e-06,
+      "loss": 0.1075,
+      "step": 6120
+    },
+    {
+      "epoch": 5.708294501397949,
+      "grad_norm": 0.19240164933962267,
+      "learning_rate": 2.7010700724887818e-06,
+      "loss": 0.1085,
+      "step": 6125
+    },
+    {
+      "epoch": 5.712954333643989,
+      "grad_norm": 0.18864094001007692,
+      "learning_rate": 2.6579219882637213e-06,
+      "loss": 0.1125,
+      "step": 6130
+    },
+    {
+      "epoch": 5.717614165890028,
+      "grad_norm": 0.18789984364146883,
+      "learning_rate": 2.614773904038661e-06,
+      "loss": 0.1143,
+      "step": 6135
+    },
+    {
+      "epoch": 5.7222739981360675,
+      "grad_norm": 0.1878224539536749,
+      "learning_rate": 2.5716258198136004e-06,
+      "loss": 0.1095,
+      "step": 6140
+    },
+    {
+      "epoch": 5.726933830382106,
+      "grad_norm": 0.179525012444654,
+      "learning_rate": 2.52847773558854e-06,
+      "loss": 0.1082,
+      "step": 6145
+    },
+    {
+      "epoch": 5.731593662628145,
+      "grad_norm": 0.1879775285348369,
+      "learning_rate": 2.4853296513634795e-06,
+      "loss": 0.1074,
+      "step": 6150
+    },
+    {
+      "epoch": 5.736253494874185,
+      "grad_norm": 0.18959048853860078,
+      "learning_rate": 2.442181567138419e-06,
+      "loss": 0.1105,
+      "step": 6155
+    },
+    {
+      "epoch": 5.740913327120223,
+      "grad_norm": 0.18095845307586522,
+      "learning_rate": 2.399033482913359e-06,
+      "loss": 0.1077,
+      "step": 6160
+    },
+    {
+      "epoch": 5.745573159366263,
+      "grad_norm": 0.1754585697312625,
+      "learning_rate": 2.3558853986882985e-06,
+      "loss": 0.1078,
+      "step": 6165
+    },
+    {
+      "epoch": 5.750232991612302,
+      "grad_norm": 0.19499601432261485,
+      "learning_rate": 2.312737314463238e-06,
+      "loss": 0.1084,
+      "step": 6170
+    },
+    {
+      "epoch": 5.754892823858341,
+      "grad_norm": 0.19458000897542455,
+      "learning_rate": 2.2695892302381776e-06,
+      "loss": 0.1116,
+      "step": 6175
+    },
+    {
+      "epoch": 5.75955265610438,
+      "grad_norm": 0.20105144763780067,
+      "learning_rate": 2.226441146013117e-06,
+      "loss": 0.1108,
+      "step": 6180
+    },
+    {
+      "epoch": 5.764212488350419,
+      "grad_norm": 0.1833921647983291,
+      "learning_rate": 2.1832930617880566e-06,
+      "loss": 0.1042,
+      "step": 6185
+    },
+    {
+      "epoch": 5.768872320596459,
+      "grad_norm": 0.19196547870939493,
+      "learning_rate": 2.140144977562996e-06,
+      "loss": 0.1092,
+      "step": 6190
+    },
+    {
+      "epoch": 5.773532152842497,
+      "grad_norm": 0.1858222683645061,
+      "learning_rate": 2.096996893337936e-06,
+      "loss": 0.1097,
+      "step": 6195
+    },
+    {
+      "epoch": 5.778191985088537,
+      "grad_norm": 0.18565312334805076,
+      "learning_rate": 2.0538488091128757e-06,
+      "loss": 0.1084,
+      "step": 6200
+    },
+    {
+      "epoch": 5.782851817334576,
+      "grad_norm": 0.1883037775331338,
+      "learning_rate": 2.010700724887815e-06,
+      "loss": 0.11,
+      "step": 6205
+    },
+    {
+      "epoch": 5.787511649580615,
+      "grad_norm": 0.1794743482490075,
+      "learning_rate": 1.9675526406627547e-06,
+      "loss": 0.1059,
+      "step": 6210
+    },
+    {
+      "epoch": 5.792171481826654,
+      "grad_norm": 0.18522145940165735,
+      "learning_rate": 1.9244045564376943e-06,
+      "loss": 0.1107,
+      "step": 6215
+    },
+    {
+      "epoch": 5.796831314072693,
+      "grad_norm": 0.17809358592155075,
+      "learning_rate": 1.8812564722126338e-06,
+      "loss": 0.1097,
+      "step": 6220
+    },
+    {
+      "epoch": 5.801491146318733,
+      "grad_norm": 0.18702588394588376,
+      "learning_rate": 1.8381083879875736e-06,
+      "loss": 0.1089,
+      "step": 6225
+    },
+    {
+      "epoch": 5.806150978564772,
+      "grad_norm": 0.19536290575377574,
+      "learning_rate": 1.794960303762513e-06,
+      "loss": 0.112,
+      "step": 6230
+    },
+    {
+      "epoch": 5.8108108108108105,
+      "grad_norm": 0.17794830430019243,
+      "learning_rate": 1.7518122195374526e-06,
+      "loss": 0.1065,
+      "step": 6235
+    },
+    {
+      "epoch": 5.81547064305685,
+      "grad_norm": 0.18665065083644808,
+      "learning_rate": 1.7086641353123924e-06,
+      "loss": 0.1067,
+      "step": 6240
+    },
+    {
+      "epoch": 5.820130475302889,
+      "grad_norm": 0.1877339032475885,
+      "learning_rate": 1.665516051087332e-06,
+      "loss": 0.1126,
+      "step": 6245
+    },
+    {
+      "epoch": 5.824790307548929,
+      "grad_norm": 0.18486665123583973,
+      "learning_rate": 1.6223679668622715e-06,
+      "loss": 0.1107,
+      "step": 6250
+    },
+    {
+      "epoch": 5.829450139794967,
+      "grad_norm": 0.17788359051430597,
+      "learning_rate": 1.579219882637211e-06,
+      "loss": 0.1122,
+      "step": 6255
+    },
+    {
+      "epoch": 5.834109972041007,
+      "grad_norm": 0.1876061937029791,
+      "learning_rate": 1.5360717984121505e-06,
+      "loss": 0.1129,
+      "step": 6260
+    },
+    {
+      "epoch": 5.838769804287046,
+      "grad_norm": 0.18633257412736237,
+      "learning_rate": 1.49292371418709e-06,
+      "loss": 0.1093,
+      "step": 6265
+    },
+    {
+      "epoch": 5.8434296365330844,
+      "grad_norm": 0.1901617560888031,
+      "learning_rate": 1.4497756299620296e-06,
+      "loss": 0.1069,
+      "step": 6270
+    },
+    {
+      "epoch": 5.848089468779124,
+      "grad_norm": 0.18146936218746637,
+      "learning_rate": 1.4066275457369694e-06,
+      "loss": 0.1063,
+      "step": 6275
+    },
+    {
+      "epoch": 5.852749301025163,
+      "grad_norm": 0.18191729005838062,
+      "learning_rate": 1.363479461511909e-06,
+      "loss": 0.108,
+      "step": 6280
+    },
+    {
+      "epoch": 5.857409133271203,
+      "grad_norm": 0.18751990960518036,
+      "learning_rate": 1.3203313772868484e-06,
+      "loss": 0.1103,
+      "step": 6285
+    },
+    {
+      "epoch": 5.862068965517241,
+      "grad_norm": 0.2274758435860074,
+      "learning_rate": 1.277183293061788e-06,
+      "loss": 0.1081,
+      "step": 6290
+    },
+    {
+      "epoch": 5.8667287977632805,
+      "grad_norm": 0.19558704543341096,
+      "learning_rate": 1.2340352088367277e-06,
+      "loss": 0.1097,
+      "step": 6295
+    },
+    {
+      "epoch": 5.87138863000932,
+      "grad_norm": 0.18133850006739877,
+      "learning_rate": 1.1908871246116673e-06,
+      "loss": 0.1067,
+      "step": 6300
+    },
+    {
+      "epoch": 5.876048462255358,
+      "grad_norm": 0.18722537724714527,
+      "learning_rate": 1.1477390403866068e-06,
+      "loss": 0.1082,
+      "step": 6305
+    },
+    {
+      "epoch": 5.880708294501398,
+      "grad_norm": 0.18422794552236618,
+      "learning_rate": 1.1045909561615463e-06,
+      "loss": 0.1069,
+      "step": 6310
+    },
+    {
+      "epoch": 5.885368126747437,
+      "grad_norm": 0.18847428070686295,
+      "learning_rate": 1.061442871936486e-06,
+      "loss": 0.1107,
+      "step": 6315
+    },
+    {
+      "epoch": 5.8900279589934765,
+      "grad_norm": 0.17947448425757695,
+      "learning_rate": 1.0182947877114256e-06,
+      "loss": 0.1089,
+      "step": 6320
+    },
+    {
+      "epoch": 5.894687791239515,
+      "grad_norm": 0.1891919674642366,
+      "learning_rate": 9.751467034863652e-07,
+      "loss": 0.1091,
+      "step": 6325
+    },
+    {
+      "epoch": 5.899347623485554,
+      "grad_norm": 0.1843850767663366,
+      "learning_rate": 9.319986192613048e-07,
+      "loss": 0.1073,
+      "step": 6330
+    },
+    {
+      "epoch": 5.904007455731594,
+      "grad_norm": 0.18417708688455778,
+      "learning_rate": 8.888505350362444e-07,
+      "loss": 0.1082,
+      "step": 6335
+    },
+    {
+      "epoch": 5.908667287977633,
+      "grad_norm": 0.18667968087319464,
+      "learning_rate": 8.45702450811184e-07,
+      "loss": 0.1113,
+      "step": 6340
+    },
+    {
+      "epoch": 5.913327120223672,
+      "grad_norm": 0.1931366010220463,
+      "learning_rate": 8.025543665861236e-07,
+      "loss": 0.1126,
+      "step": 6345
+    },
+    {
+      "epoch": 5.917986952469711,
+      "grad_norm": 0.1795082035832971,
+      "learning_rate": 7.594062823610632e-07,
+      "loss": 0.1118,
+      "step": 6350
+    },
+    {
+      "epoch": 5.9226467847157505,
+      "grad_norm": 0.18883886926773777,
+      "learning_rate": 7.162581981360028e-07,
+      "loss": 0.1081,
+      "step": 6355
+    },
+    {
+      "epoch": 5.92730661696179,
+      "grad_norm": 0.1845372042242747,
+      "learning_rate": 6.731101139109423e-07,
+      "loss": 0.107,
+      "step": 6360
+    },
+    {
+      "epoch": 5.931966449207828,
+      "grad_norm": 0.18650131602072892,
+      "learning_rate": 6.29962029685882e-07,
+      "loss": 0.1101,
+      "step": 6365
+    },
+    {
+      "epoch": 5.936626281453868,
+      "grad_norm": 0.17502453367742662,
+      "learning_rate": 5.868139454608215e-07,
+      "loss": 0.1101,
+      "step": 6370
+    },
+    {
+      "epoch": 5.941286113699907,
+      "grad_norm": 0.1837258556554783,
+      "learning_rate": 5.436658612357612e-07,
+      "loss": 0.11,
+      "step": 6375
+    },
+    {
+      "epoch": 5.945945945945946,
+      "grad_norm": 0.18712443672010842,
+      "learning_rate": 5.005177770107007e-07,
+      "loss": 0.1052,
+      "step": 6380
+    },
+    {
+      "epoch": 5.950605778191985,
+      "grad_norm": 0.185778274245323,
+      "learning_rate": 4.5736969278564034e-07,
+      "loss": 0.1121,
+      "step": 6385
+    },
+    {
+      "epoch": 5.955265610438024,
+      "grad_norm": 0.18196881493434902,
+      "learning_rate": 4.142216085605799e-07,
+      "loss": 0.1113,
+      "step": 6390
+    },
+    {
+      "epoch": 5.959925442684064,
+      "grad_norm": 0.19319305183066607,
+      "learning_rate": 3.710735243355195e-07,
+      "loss": 0.1095,
+      "step": 6395
+    },
+    {
+      "epoch": 5.964585274930102,
+      "grad_norm": 0.190821130519731,
+      "learning_rate": 3.279254401104591e-07,
+      "loss": 0.1077,
+      "step": 6400
+    },
+    {
+      "epoch": 5.969245107176142,
+      "grad_norm": 0.185945698685377,
+      "learning_rate": 2.847773558853987e-07,
+      "loss": 0.1101,
+      "step": 6405
+    },
+    {
+      "epoch": 5.973904939422181,
+      "grad_norm": 0.1840724230583126,
+      "learning_rate": 2.416292716603383e-07,
+      "loss": 0.1089,
+      "step": 6410
+    },
+    {
+      "epoch": 5.9785647716682195,
+      "grad_norm": 0.17985002823768823,
+      "learning_rate": 1.984811874352779e-07,
+      "loss": 0.1102,
+      "step": 6415
+    },
+    {
+      "epoch": 5.983224603914259,
+      "grad_norm": 0.18315605325083412,
+      "learning_rate": 1.5533310321021747e-07,
+      "loss": 0.1133,
+      "step": 6420
+    },
+    {
+      "epoch": 5.987884436160298,
+      "grad_norm": 0.17288738849873067,
+      "learning_rate": 1.1218501898515707e-07,
+      "loss": 0.1057,
+      "step": 6425
+    },
+    {
+      "epoch": 5.992544268406338,
+      "grad_norm": 0.18145253138810077,
+      "learning_rate": 6.903693476009665e-08,
+      "loss": 0.1072,
+      "step": 6430
+    },
+    {
+      "epoch": 5.997204100652376,
+      "grad_norm": 0.18169111394575085,
+      "learning_rate": 2.5888850535036245e-08,
+      "loss": 0.109,
+      "step": 6435
+    },
+    {
+      "epoch": 6.0,
+      "step": 6438,
+      "total_flos": 5.511853464944116e+18,
+      "train_loss": 0.31514360057841034,
+      "train_runtime": 84482.0666,
+      "train_samples_per_second": 1.219,
+      "train_steps_per_second": 0.076
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6438,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.511853464944116e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}