diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.4999794145497962,
+  "epoch": 0.7499691218246943,
   "eval_steps": 759,
-  "global_step": 1518,
+  "global_step": 2277,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10657,6 +10657,5327 @@
       "eval_samples_per_second": 3.343,
       "eval_steps_per_second": 1.672,
       "step": 1518
+    },
+    {
+      "epoch": 0.5003087817530569,
+      "grad_norm": 3.1498970985412598,
+      "learning_rate": 5.126239470409546e-05,
+      "loss": 2.2301,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5006381489563176,
+      "grad_norm": 2.7907960414886475,
+      "learning_rate": 5.1209805415304224e-05,
+      "loss": 2.0078,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5009675161595785,
+      "grad_norm": 3.4877936840057373,
+      "learning_rate": 5.115721478734323e-05,
+      "loss": 2.7673,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5012968833628392,
+      "grad_norm": 3.1865594387054443,
+      "learning_rate": 5.1104622878426664e-05,
+      "loss": 1.9934,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5016262505660999,
+      "grad_norm": 2.964200258255005,
+      "learning_rate": 5.105202974677008e-05,
+      "loss": 1.8623,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5019556177693606,
+      "grad_norm": 3.1753077507019043,
+      "learning_rate": 5.099943545059035e-05,
+      "loss": 1.9103,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5022849849726213,
+      "grad_norm": 3.5453011989593506,
+      "learning_rate": 5.094684004810568e-05,
+      "loss": 2.223,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5026143521758821,
+      "grad_norm": 1.9202203750610352,
+      "learning_rate": 5.089424359753553e-05,
+      "loss": 2.3095,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5029437193791428,
+      "grad_norm": 2.3142552375793457,
+      "learning_rate": 5.084164615710042e-05,
+      "loss": 2.0537,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5032730865824036,
+      "grad_norm": 2.545691728591919,
+      "learning_rate": 5.078904778502206e-05,
+      "loss": 2.222,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5036024537856643,
+      "grad_norm": 2.6911139488220215,
+      "learning_rate": 5.0736448539523174e-05,
+      "loss": 2.2372,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5039318209889251,
+      "grad_norm": 2.5566017627716064,
+      "learning_rate": 5.06838484788274e-05,
+      "loss": 2.2237,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5042611881921858,
+      "grad_norm": 3.251295804977417,
+      "learning_rate": 5.063124766115933e-05,
+      "loss": 2.6693,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5045905553954465,
+      "grad_norm": 2.5430779457092285,
+      "learning_rate": 5.057864614474439e-05,
+      "loss": 2.6509,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5049199225987072,
+      "grad_norm": 2.509148597717285,
+      "learning_rate": 5.052604398780877e-05,
+      "loss": 2.2311,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5052492898019679,
+      "grad_norm": 2.435283899307251,
+      "learning_rate": 5.047344124857933e-05,
+      "loss": 2.2194,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5055786570052287,
+      "grad_norm": 2.5991721153259277,
+      "learning_rate": 5.0420837985283664e-05,
+      "loss": 2.3425,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5059080242084895,
+      "grad_norm": 2.383134126663208,
+      "learning_rate": 5.036823425614986e-05,
+      "loss": 2.3328,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5062373914117502,
+      "grad_norm": 3.121763229370117,
+      "learning_rate": 5.0315630119406565e-05,
+      "loss": 2.4681,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5065667586150109,
+      "grad_norm": 2.824033260345459,
+      "learning_rate": 5.0263025633282866e-05,
+      "loss": 2.3247,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5068961258182716,
+      "grad_norm": 2.942976713180542,
+      "learning_rate": 5.021042085600827e-05,
+      "loss": 2.254,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5072254930215324,
+      "grad_norm": 2.7724997997283936,
+      "learning_rate": 5.015781584581252e-05,
+      "loss": 2.0605,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5075548602247931,
+      "grad_norm": 2.791670322418213,
+      "learning_rate": 5.0105210660925704e-05,
+      "loss": 2.0396,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5078842274280538,
+      "grad_norm": 2.5132830142974854,
+      "learning_rate": 5.0052605359578086e-05,
+      "loss": 1.7545,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5082135946313145,
+      "grad_norm": 2.5325276851654053,
+      "learning_rate": 5e-05,
+      "loss": 1.737,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5085429618345754,
+      "grad_norm": 3.5093564987182617,
+      "learning_rate": 4.994739464042193e-05,
+      "loss": 2.3988,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5088723290378361,
+      "grad_norm": 3.0075719356536865,
+      "learning_rate": 4.989478933907431e-05,
+      "loss": 2.0844,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5092016962410968,
+      "grad_norm": 2.6146292686462402,
+      "learning_rate": 4.9842184154187486e-05,
+      "loss": 2.0929,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5095310634443575,
+      "grad_norm": 2.921995162963867,
+      "learning_rate": 4.978957914399174e-05,
+      "loss": 2.162,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5098604306476182,
+      "grad_norm": 3.16227650642395,
+      "learning_rate": 4.973697436671714e-05,
+      "loss": 2.2389,
+      "step": 1548
+    },
+    {
+      "epoch": 0.510189797850879,
+      "grad_norm": 2.8888611793518066,
+      "learning_rate": 4.9684369880593446e-05,
+      "loss": 1.9225,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5105191650541397,
+      "grad_norm": 3.1876556873321533,
+      "learning_rate": 4.963176574385015e-05,
+      "loss": 1.7554,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5108485322574005,
+      "grad_norm": 2.1988892555236816,
+      "learning_rate": 4.957916201471635e-05,
+      "loss": 2.6424,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5111778994606612,
+      "grad_norm": 2.707622766494751,
+      "learning_rate": 4.952655875142068e-05,
+      "loss": 2.4397,
+      "step": 1552
+    },
+    {
+      "epoch": 0.511507266663922,
+      "grad_norm": 2.7695624828338623,
+      "learning_rate": 4.947395601219126e-05,
+      "loss": 2.2901,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5118366338671827,
+      "grad_norm": 2.7299704551696777,
+      "learning_rate": 4.9421353855255624e-05,
+      "loss": 2.4199,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5121660010704434,
+      "grad_norm": 2.7451207637786865,
+      "learning_rate": 4.936875233884069e-05,
+      "loss": 2.1827,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5124953682737041,
+      "grad_norm": 2.116290807723999,
+      "learning_rate": 4.931615152117262e-05,
+      "loss": 1.9413,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5128247354769648,
+      "grad_norm": 2.1912240982055664,
+      "learning_rate": 4.926355146047685e-05,
+      "loss": 2.3163,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5131541026802257,
+      "grad_norm": 2.397310495376587,
+      "learning_rate": 4.9210952214977954e-05,
+      "loss": 2.022,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5134834698834864,
+      "grad_norm": 2.5802431106567383,
+      "learning_rate": 4.915835384289958e-05,
+      "loss": 2.3069,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5138128370867471,
+      "grad_norm": 2.7203660011291504,
+      "learning_rate": 4.9105756402464486e-05,
+      "loss": 2.6652,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5141422042900078,
+      "grad_norm": 2.513911247253418,
+      "learning_rate": 4.905315995189431e-05,
+      "loss": 2.012,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5144715714932686,
+      "grad_norm": 3.2579002380371094,
+      "learning_rate": 4.900056454940965e-05,
+      "loss": 2.3097,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5148009386965293,
+      "grad_norm": 3.077726364135742,
+      "learning_rate": 4.894797025322993e-05,
+      "loss": 2.1601,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51513030589979,
+      "grad_norm": 3.066802501678467,
+      "learning_rate": 4.8895377121573334e-05,
+      "loss": 2.5118,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5154596731030507,
+      "grad_norm": 2.064492702484131,
+      "learning_rate": 4.884278521265677e-05,
+      "loss": 1.8952,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5157890403063115,
+      "grad_norm": 2.6084978580474854,
+      "learning_rate": 4.879019458469579e-05,
+      "loss": 2.2112,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5161184075095723,
+      "grad_norm": 2.9014785289764404,
+      "learning_rate": 4.873760529590455e-05,
+      "loss": 2.0184,
+      "step": 1567
+    },
+    {
+      "epoch": 0.516447774712833,
+      "grad_norm": 2.735502243041992,
+      "learning_rate": 4.8685017404495683e-05,
+      "loss": 2.0125,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5167771419160937,
+      "grad_norm": 2.60123348236084,
+      "learning_rate": 4.863243096868029e-05,
+      "loss": 2.0574,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5171065091193544,
+      "grad_norm": 2.9571070671081543,
+      "learning_rate": 4.85798460466679e-05,
+      "loss": 2.1061,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5174358763226152,
+      "grad_norm": 2.7076237201690674,
+      "learning_rate": 4.85272626966663e-05,
+      "loss": 2.0312,
+      "step": 1571
+    },
+    {
+      "epoch": 0.5177652435258759,
+      "grad_norm": 2.76377010345459,
+      "learning_rate": 4.847468097688157e-05,
+      "loss": 1.9232,
+      "step": 1572
+    },
+    {
+      "epoch": 0.5180946107291367,
+      "grad_norm": 2.923283338546753,
+      "learning_rate": 4.8422100945518015e-05,
+      "loss": 2.0029,
+      "step": 1573
+    },
+    {
+      "epoch": 0.5184239779323974,
+      "grad_norm": 2.9562487602233887,
+      "learning_rate": 4.836952266077801e-05,
+      "loss": 2.1253,
+      "step": 1574
+    },
+    {
+      "epoch": 0.5187533451356581,
+      "grad_norm": 3.1904914379119873,
+      "learning_rate": 4.8316946180862036e-05,
+      "loss": 2.2682,
+      "step": 1575
+    },
+    {
+      "epoch": 0.5190827123389189,
+      "grad_norm": 2.154838800430298,
+      "learning_rate": 4.826437156396858e-05,
+      "loss": 2.6257,
+      "step": 1576
+    },
+    {
+      "epoch": 0.5194120795421796,
+      "grad_norm": 2.501347541809082,
+      "learning_rate": 4.821179886829405e-05,
+      "loss": 2.0692,
+      "step": 1577
+    },
+    {
+      "epoch": 0.5197414467454403,
+      "grad_norm": 2.355531692504883,
+      "learning_rate": 4.815922815203272e-05,
+      "loss": 2.2775,
+      "step": 1578
+    },
+    {
+      "epoch": 0.520070813948701,
+      "grad_norm": 2.39996075630188,
+      "learning_rate": 4.8106659473376695e-05,
+      "loss": 2.3885,
+      "step": 1579
+    },
+    {
+      "epoch": 0.5204001811519617,
+      "grad_norm": 2.339895486831665,
+      "learning_rate": 4.805409289051582e-05,
+      "loss": 2.2418,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5207295483552226,
+      "grad_norm": 2.3130595684051514,
+      "learning_rate": 4.80015284616376e-05,
+      "loss": 2.0223,
+      "step": 1581
+    },
+    {
+      "epoch": 0.5210589155584833,
+      "grad_norm": 2.9214603900909424,
+      "learning_rate": 4.794896624492718e-05,
+      "loss": 2.4215,
+      "step": 1582
+    },
+    {
+      "epoch": 0.521388282761744,
+      "grad_norm": 2.5012197494506836,
+      "learning_rate": 4.789640629856725e-05,
+      "loss": 2.3697,
+      "step": 1583
+    },
+    {
+      "epoch": 0.5217176499650047,
+      "grad_norm": 2.809415817260742,
+      "learning_rate": 4.7843848680737966e-05,
+      "loss": 2.1699,
+      "step": 1584
+    },
+    {
+      "epoch": 0.5220470171682655,
+      "grad_norm": 2.9632301330566406,
+      "learning_rate": 4.779129344961694e-05,
+      "loss": 2.2572,
+      "step": 1585
+    },
+    {
+      "epoch": 0.5223763843715262,
+      "grad_norm": 2.583806276321411,
+      "learning_rate": 4.7738740663379135e-05,
+      "loss": 2.1202,
+      "step": 1586
+    },
+    {
+      "epoch": 0.5227057515747869,
+      "grad_norm": 3.0589942932128906,
+      "learning_rate": 4.768619038019677e-05,
+      "loss": 2.2599,
+      "step": 1587
+    },
+    {
+      "epoch": 0.5230351187780476,
+      "grad_norm": 2.770036220550537,
+      "learning_rate": 4.7633642658239345e-05,
+      "loss": 1.9833,
+      "step": 1588
+    },
+    {
+      "epoch": 0.5233644859813084,
+      "grad_norm": 3.0664565563201904,
+      "learning_rate": 4.758109755567348e-05,
+      "loss": 2.0921,
+      "step": 1589
+    },
+    {
+      "epoch": 0.5236938531845692,
+      "grad_norm": 2.592406749725342,
+      "learning_rate": 4.752855513066293e-05,
+      "loss": 1.9637,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5240232203878299,
+      "grad_norm": 2.8975071907043457,
+      "learning_rate": 4.747601544136849e-05,
+      "loss": 2.2294,
+      "step": 1591
+    },
+    {
+      "epoch": 0.5243525875910906,
+      "grad_norm": 2.795776844024658,
+      "learning_rate": 4.742347854594791e-05,
+      "loss": 2.2874,
+      "step": 1592
+    },
+    {
+      "epoch": 0.5246819547943513,
+      "grad_norm": 2.9049065113067627,
+      "learning_rate": 4.737094450255581e-05,
+      "loss": 2.2325,
+      "step": 1593
+    },
+    {
+      "epoch": 0.5250113219976121,
+      "grad_norm": 3.0760581493377686,
+      "learning_rate": 4.731841336934372e-05,
+      "loss": 2.2926,
+      "step": 1594
+    },
+    {
+      "epoch": 0.5253406892008728,
+      "grad_norm": 2.4571144580841064,
+      "learning_rate": 4.726588520445993e-05,
+      "loss": 2.0032,
+      "step": 1595
+    },
+    {
+      "epoch": 0.5256700564041336,
+      "grad_norm": 3.1097042560577393,
+      "learning_rate": 4.721336006604941e-05,
+      "loss": 1.9375,
+      "step": 1596
+    },
+    {
+      "epoch": 0.5259994236073943,
+      "grad_norm": 2.750455617904663,
+      "learning_rate": 4.71608380122538e-05,
+      "loss": 2.1235,
+      "step": 1597
+    },
+    {
+      "epoch": 0.526328790810655,
+      "grad_norm": 2.6923322677612305,
+      "learning_rate": 4.710831910121135e-05,
+      "loss": 1.7062,
+      "step": 1598
+    },
+    {
+      "epoch": 0.5266581580139158,
+      "grad_norm": 3.282520055770874,
+      "learning_rate": 4.7055803391056795e-05,
+      "loss": 2.2095,
+      "step": 1599
+    },
+    {
+      "epoch": 0.5269875252171765,
+      "grad_norm": 2.924677610397339,
+      "learning_rate": 4.700329093992135e-05,
+      "loss": 1.6683,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5273168924204372,
+      "grad_norm": 2.091074228286743,
+      "learning_rate": 4.695078180593262e-05,
+      "loss": 2.4635,
+      "step": 1601
+    },
+    {
+      "epoch": 0.5276462596236979,
+      "grad_norm": 2.103097677230835,
+      "learning_rate": 4.689827604721451e-05,
+      "loss": 2.1667,
+      "step": 1602
+    },
+    {
+      "epoch": 0.5279756268269588,
+      "grad_norm": 2.7929534912109375,
+      "learning_rate": 4.684577372188723e-05,
+      "loss": 2.7994,
+      "step": 1603
+    },
+    {
+      "epoch": 0.5283049940302195,
+      "grad_norm": 2.534282684326172,
+      "learning_rate": 4.679327488806716e-05,
+      "loss": 2.3134,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5286343612334802,
+      "grad_norm": 2.393833875656128,
+      "learning_rate": 4.674077960386685e-05,
+      "loss": 2.3605,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5289637284367409,
+      "grad_norm": 2.5953030586242676,
+      "learning_rate": 4.668828792739487e-05,
+      "loss": 2.2843,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5292930956400016,
+      "grad_norm": 2.9071319103240967,
+      "learning_rate": 4.6635799916755836e-05,
+      "loss": 2.35,
+      "step": 1607
+    },
+    {
+      "epoch": 0.5296224628432624,
+      "grad_norm": 2.4467287063598633,
+      "learning_rate": 4.65833156300503e-05,
+      "loss": 2.1182,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5299518300465231,
+      "grad_norm": 2.2593226432800293,
+      "learning_rate": 4.653083512537467e-05,
+      "loss": 2.3638,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5302811972497838,
+      "grad_norm": 2.5257232189178467,
+      "learning_rate": 4.6478358460821184e-05,
+      "loss": 2.2619,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5306105644530446,
+      "grad_norm": 2.365927219390869,
+      "learning_rate": 4.642588569447785e-05,
+      "loss": 2.1333,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5309399316563054,
+      "grad_norm": 2.346470594406128,
+      "learning_rate": 4.637341688442831e-05,
+      "loss": 2.1835,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5312692988595661,
+      "grad_norm": 2.5479326248168945,
+      "learning_rate": 4.6320952088751876e-05,
+      "loss": 2.1302,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5315986660628268,
+      "grad_norm": 2.7286736965179443,
+      "learning_rate": 4.626849136552338e-05,
+      "loss": 2.0868,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5319280332660875,
+      "grad_norm": 2.7833032608032227,
+      "learning_rate": 4.6216034772813195e-05,
+      "loss": 2.5112,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5322574004693482,
+      "grad_norm": 2.5549001693725586,
+      "learning_rate": 4.616358236868705e-05,
+      "loss": 2.0309,
+      "step": 1616
+    },
+    {
+      "epoch": 0.532586767672609,
+      "grad_norm": 3.951969623565674,
+      "learning_rate": 4.61111342112061e-05,
+      "loss": 2.3313,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5329161348758698,
+      "grad_norm": 2.874086380004883,
+      "learning_rate": 4.605869035842677e-05,
+      "loss": 2.2732,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5332455020791305,
+      "grad_norm": 2.5527641773223877,
+      "learning_rate": 4.6006250868400726e-05,
+      "loss": 1.9412,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5335748692823912,
+      "grad_norm": 2.8807637691497803,
+      "learning_rate": 4.595381579917478e-05,
+      "loss": 2.283,
+      "step": 1620
+    },
+    {
+      "epoch": 0.533904236485652,
+      "grad_norm": 2.820483922958374,
+      "learning_rate": 4.5901385208790924e-05,
+      "loss": 2.3101,
+      "step": 1621
+    },
+    {
+      "epoch": 0.5342336036889127,
+      "grad_norm": 2.790860652923584,
+      "learning_rate": 4.584895915528611e-05,
+      "loss": 2.0345,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5345629708921734,
+      "grad_norm": 2.5604536533355713,
+      "learning_rate": 4.579653769669233e-05,
+      "loss": 1.8557,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5348923380954341,
+      "grad_norm": 3.321770668029785,
+      "learning_rate": 4.574412089103643e-05,
+      "loss": 1.9905,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5352217052986948,
+      "grad_norm": 3.51432204246521,
+      "learning_rate": 4.569170879634014e-05,
+      "loss": 1.9512,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5355510725019557,
+      "grad_norm": 2.5113067626953125,
+      "learning_rate": 4.563930147062e-05,
+      "loss": 2.393,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5358804397052164,
+      "grad_norm": 1.724272608757019,
+      "learning_rate": 4.558689897188721e-05,
+      "loss": 2.1496,
+      "step": 1627
+    },
+    {
+      "epoch": 0.5362098069084771,
+      "grad_norm": 2.2979555130004883,
+      "learning_rate": 4.553450135814768e-05,
+      "loss": 2.4137,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5365391741117378,
+      "grad_norm": 2.4890944957733154,
+      "learning_rate": 4.548210868740188e-05,
+      "loss": 2.3685,
+      "step": 1629
+    },
+    {
+      "epoch": 0.5368685413149985,
+      "grad_norm": 2.3017985820770264,
+      "learning_rate": 4.5429721017644835e-05,
+      "loss": 2.4172,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5371979085182593,
+      "grad_norm": 2.6837596893310547,
+      "learning_rate": 4.537733840686601e-05,
+      "loss": 2.424,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53752727572152,
+      "grad_norm": 2.583272695541382,
+      "learning_rate": 4.532496091304929e-05,
+      "loss": 2.5061,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5378566429247807,
+      "grad_norm": 2.5158188343048096,
+      "learning_rate": 4.5272588594172875e-05,
+      "loss": 2.3061,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5381860101280415,
+      "grad_norm": 2.558367967605591,
+      "learning_rate": 4.522022150820925e-05,
+      "loss": 2.0809,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5385153773313023,
+      "grad_norm": 2.473259687423706,
+      "learning_rate": 4.516785971312511e-05,
+      "loss": 2.2443,
+      "step": 1635
+    },
+    {
+      "epoch": 0.538844744534563,
+      "grad_norm": 2.3816640377044678,
+      "learning_rate": 4.51155032668813e-05,
+      "loss": 2.0707,
+      "step": 1636
+    },
+    {
+      "epoch": 0.5391741117378237,
+      "grad_norm": 2.75467586517334,
+      "learning_rate": 4.5063152227432705e-05,
+      "loss": 2.307,
+      "step": 1637
+    },
+    {
+      "epoch": 0.5395034789410844,
+      "grad_norm": 2.2179129123687744,
+      "learning_rate": 4.501080665272827e-05,
+      "loss": 2.1801,
+      "step": 1638
+    },
+    {
+      "epoch": 0.5398328461443451,
+      "grad_norm": 2.5906922817230225,
+      "learning_rate": 4.495846660071088e-05,
+      "loss": 2.136,
+      "step": 1639
+    },
+    {
+      "epoch": 0.540162213347606,
+      "grad_norm": 3.0321009159088135,
+      "learning_rate": 4.490613212931729e-05,
+      "loss": 2.5038,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5404915805508667,
+      "grad_norm": 2.619051694869995,
+      "learning_rate": 4.485380329647808e-05,
+      "loss": 2.2298,
+      "step": 1641
+    },
+    {
+      "epoch": 0.5408209477541274,
+      "grad_norm": 3.0019948482513428,
+      "learning_rate": 4.480148016011762e-05,
+      "loss": 2.0229,
+      "step": 1642
+    },
+    {
+      "epoch": 0.5411503149573881,
+      "grad_norm": 2.2649595737457275,
+      "learning_rate": 4.4749162778153954e-05,
+      "loss": 1.8919,
+      "step": 1643
+    },
+    {
+      "epoch": 0.5414796821606489,
+      "grad_norm": 2.849843740463257,
+      "learning_rate": 4.469685120849872e-05,
+      "loss": 2.0903,
+      "step": 1644
+    },
+    {
+      "epoch": 0.5418090493639096,
+      "grad_norm": 3.021907091140747,
+      "learning_rate": 4.4644545509057185e-05,
+      "loss": 2.3181,
+      "step": 1645
+    },
+    {
+      "epoch": 0.5421384165671703,
+      "grad_norm": 2.7080013751983643,
+      "learning_rate": 4.459224573772808e-05,
+      "loss": 1.8712,
+      "step": 1646
+    },
+    {
+      "epoch": 0.542467783770431,
+      "grad_norm": 2.6799404621124268,
+      "learning_rate": 4.453995195240358e-05,
+      "loss": 1.4717,
+      "step": 1647
+    },
+    {
+      "epoch": 0.5427971509736917,
+      "grad_norm": 3.217162609100342,
+      "learning_rate": 4.448766421096924e-05,
+      "loss": 2.263,
+      "step": 1648
+    },
+    {
+      "epoch": 0.5431265181769526,
+      "grad_norm": 3.4610049724578857,
+      "learning_rate": 4.443538257130393e-05,
+      "loss": 2.0226,
+      "step": 1649
+    },
+    {
+      "epoch": 0.5434558853802133,
+      "grad_norm": 3.7474148273468018,
+      "learning_rate": 4.4383107091279724e-05,
+      "loss": 1.9848,
+      "step": 1650
+    },
+    {
+      "epoch": 0.543785252583474,
+      "grad_norm": 1.9629192352294922,
+      "learning_rate": 4.433083782876196e-05,
+      "loss": 2.4658,
+      "step": 1651
+    },
+    {
+      "epoch": 0.5441146197867347,
+      "grad_norm": 2.639218807220459,
+      "learning_rate": 4.427857484160902e-05,
+      "loss": 2.5598,
+      "step": 1652
+    },
+    {
+      "epoch": 0.5444439869899955,
+      "grad_norm": 2.2541027069091797,
+      "learning_rate": 4.4226318187672357e-05,
+      "loss": 2.4328,
+      "step": 1653
+    },
+    {
+      "epoch": 0.5447733541932562,
+      "grad_norm": 2.1472740173339844,
+      "learning_rate": 4.4174067924796444e-05,
+      "loss": 2.1159,
+      "step": 1654
+    },
+    {
+      "epoch": 0.5451027213965169,
+      "grad_norm": 2.170586585998535,
+      "learning_rate": 4.4121824110818636e-05,
+      "loss": 2.193,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5454320885997777,
+      "grad_norm": 2.1903934478759766,
+      "learning_rate": 4.406958680356917e-05,
+      "loss": 2.0941,
+      "step": 1656
+    },
+    {
+      "epoch": 0.5457614558030384,
+      "grad_norm": 2.934429883956909,
+      "learning_rate": 4.4017356060871084e-05,
+      "loss": 2.7441,
+      "step": 1657
+    },
+    {
+      "epoch": 0.5460908230062992,
+      "grad_norm": 2.7060625553131104,
+      "learning_rate": 4.396513194054017e-05,
+      "loss": 2.2778,
+      "step": 1658
+    },
+    {
+      "epoch": 0.5464201902095599,
+      "grad_norm": 2.726825714111328,
+      "learning_rate": 4.3912914500384825e-05,
+      "loss": 2.4721,
+      "step": 1659
+    },
+    {
+      "epoch": 0.5467495574128206,
+      "grad_norm": 2.721712589263916,
+      "learning_rate": 4.386070379820612e-05,
+      "loss": 2.6526,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5470789246160813,
+      "grad_norm": 2.7384960651397705,
+      "learning_rate": 4.380849989179764e-05,
+      "loss": 2.0193,
+      "step": 1661
+    },
+    {
+      "epoch": 0.5474082918193421,
+      "grad_norm": 2.4211504459381104,
+      "learning_rate": 4.375630283894543e-05,
+      "loss": 1.86,
+      "step": 1662
+    },
+    {
+      "epoch": 0.5477376590226029,
+      "grad_norm": 2.456686019897461,
+      "learning_rate": 4.370411269742797e-05,
+      "loss": 2.2503,
+      "step": 1663
+    },
+    {
+      "epoch": 0.5480670262258636,
+      "grad_norm": 2.7705039978027344,
+      "learning_rate": 4.36519295250161e-05,
+      "loss": 2.1732,
+      "step": 1664
+    },
+    {
+      "epoch": 0.5483963934291243,
+      "grad_norm": 2.6563832759857178,
+      "learning_rate": 4.3599753379472916e-05,
+      "loss": 1.8923,
+      "step": 1665
+    },
+    {
+      "epoch": 0.548725760632385,
+      "grad_norm": 3.264117479324341,
+      "learning_rate": 4.354758431855375e-05,
+      "loss": 2.1551,
+      "step": 1666
+    },
+    {
+      "epoch": 0.5490551278356458,
+      "grad_norm": 3.068531036376953,
+      "learning_rate": 4.349542240000612e-05,
+      "loss": 2.342,
+      "step": 1667
+    },
+    {
+      "epoch": 0.5493844950389065,
+      "grad_norm": 2.412754535675049,
+      "learning_rate": 4.344326768156957e-05,
+      "loss": 1.961,
+      "step": 1668
+    },
+    {
+      "epoch": 0.5497138622421672,
+      "grad_norm": 2.5070178508758545,
+      "learning_rate": 4.339112022097574e-05,
+      "loss": 2.0561,
+      "step": 1669
+    },
+    {
+      "epoch": 0.5500432294454279,
+      "grad_norm": 3.7027029991149902,
+      "learning_rate": 4.33389800759482e-05,
+      "loss": 2.4097,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5503725966486888,
+      "grad_norm": 2.5179436206817627,
+      "learning_rate": 4.3286847304202446e-05,
+      "loss": 2.1429,
+      "step": 1671
+    },
+    {
+      "epoch": 0.5507019638519495,
+      "grad_norm": 2.777573823928833,
+      "learning_rate": 4.323472196344579e-05,
+      "loss": 1.8857,
+      "step": 1672
+    },
+    {
+      "epoch": 0.5510313310552102,
+      "grad_norm": 2.600545644760132,
+      "learning_rate": 4.3182604111377324e-05,
+      "loss": 1.7022,
+      "step": 1673
+    },
+    {
+      "epoch": 0.5513606982584709,
+      "grad_norm": 3.057023048400879,
+      "learning_rate": 4.3130493805687864e-05,
+      "loss": 2.03,
+      "step": 1674
+    },
+    {
+      "epoch": 0.5516900654617316,
+      "grad_norm": 3.4655754566192627,
+      "learning_rate": 4.3078391104059854e-05,
+      "loss": 1.8471,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5520194326649924,
+      "grad_norm": 2.065298318862915,
+      "learning_rate": 4.3026296064167334e-05,
+      "loss": 2.4574,
+      "step": 1676
+    },
+    {
+      "epoch": 0.5523487998682531,
+      "grad_norm": 2.120046615600586,
+      "learning_rate": 4.297420874367586e-05,
+      "loss": 2.1619,
+      "step": 1677
+    },
+    {
+      "epoch": 0.5526781670715138,
+      "grad_norm": 2.614112615585327,
+      "learning_rate": 4.292212920024243e-05,
+      "loss": 2.3831,
+      "step": 1678
+    },
+    {
+      "epoch": 0.5530075342747746,
+      "grad_norm": 2.4009101390838623,
+      "learning_rate": 4.287005749151546e-05,
+      "loss": 2.4063,
+      "step": 1679
+    },
+    {
+      "epoch": 0.5533369014780353,
+      "grad_norm": 2.5356178283691406,
+      "learning_rate": 4.281799367513467e-05,
+      "loss": 2.2704,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5536662686812961,
+      "grad_norm": 2.489375352859497,
+      "learning_rate": 4.276593780873105e-05,
+      "loss": 2.1472,
+      "step": 1681
+    },
+    {
+      "epoch": 0.5539956358845568,
+      "grad_norm": 2.381366491317749,
+      "learning_rate": 4.271388994992682e-05,
+      "loss": 2.0989,
+      "step": 1682
+    },
+    {
+      "epoch": 0.5543250030878175,
+      "grad_norm": 2.4626338481903076,
+      "learning_rate": 4.266185015633527e-05,
+      "loss": 2.2773,
+      "step": 1683
+    },
+    {
+      "epoch": 0.5546543702910782,
+      "grad_norm": 2.7277164459228516,
+      "learning_rate": 4.260981848556081e-05,
+      "loss": 2.3875,
+      "step": 1684
+    },
+    {
+      "epoch": 0.554983737494339,
+      "grad_norm": 2.389749765396118,
+      "learning_rate": 4.2557794995198876e-05,
+      "loss": 2.0901,
+      "step": 1685
+    },
+    {
+      "epoch": 0.5553131046975998,
+      "grad_norm": 2.2613539695739746,
+      "learning_rate": 4.25057797428358e-05,
+      "loss": 2.3888,
+      "step": 1686
+    },
+    {
+      "epoch": 0.5556424719008605,
+      "grad_norm": 2.8356738090515137,
+      "learning_rate": 4.245377278604881e-05,
+      "loss": 2.1503,
+      "step": 1687
+    },
+    {
+      "epoch": 0.5559718391041212,
+      "grad_norm": 2.471827745437622,
+      "learning_rate": 4.240177418240597e-05,
+      "loss": 2.1751,
+      "step": 1688
+    },
+    {
+      "epoch": 0.5563012063073819,
+      "grad_norm": 2.6936299800872803,
+      "learning_rate": 4.234978398946611e-05,
+      "loss": 2.2912,
+      "step": 1689
+    },
+    {
+      "epoch": 0.5566305735106427,
+      "grad_norm": 2.365645170211792,
+      "learning_rate": 4.2297802264778676e-05,
+      "loss": 2.0137,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5569599407139034,
+      "grad_norm": 2.8537073135375977,
+      "learning_rate": 4.224582906588382e-05,
+      "loss": 1.9902,
+      "step": 1691
+    },
+    {
+      "epoch": 0.5572893079171641,
+      "grad_norm": 2.6542677879333496,
+      "learning_rate": 4.2193864450312236e-05,
+      "loss": 2.0595,
+      "step": 1692
+    },
+    {
+      "epoch": 0.5576186751204248,
+      "grad_norm": 2.9382853507995605,
+      "learning_rate": 4.214190847558509e-05,
+      "loss": 2.1753,
+      "step": 1693
+    },
+    {
+      "epoch": 0.5579480423236857,
+      "grad_norm": 2.9632253646850586,
+      "learning_rate": 4.2089961199214e-05,
+      "loss": 2.0176,
+      "step": 1694
+    },
+    {
+      "epoch": 0.5582774095269464,
+      "grad_norm": 3.0138721466064453,
+      "learning_rate": 4.203802267870097e-05,
+      "loss": 1.9451,
+      "step": 1695
+    },
+    {
+      "epoch": 0.5586067767302071,
+      "grad_norm": 2.569395065307617,
+      "learning_rate": 4.1986092971538306e-05,
+      "loss": 1.8728,
+      "step": 1696
+    },
+    {
+      "epoch": 0.5589361439334678,
+      "grad_norm": 2.5671072006225586,
+      "learning_rate": 4.193417213520852e-05,
+      "loss": 1.9411,
+      "step": 1697
+    },
+    {
+      "epoch": 0.5592655111367285,
+      "grad_norm": 2.9711215496063232,
+      "learning_rate": 4.1882260227184354e-05,
+      "loss": 2.2304,
+      "step": 1698
+    },
+    {
+      "epoch": 0.5595948783399893,
+      "grad_norm": 2.82068133354187,
+      "learning_rate": 4.1830357304928664e-05,
+      "loss": 1.9563,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55992424554325,
+      "grad_norm": 3.187591314315796,
+      "learning_rate": 4.177846342589431e-05,
+      "loss": 1.8838,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5602536127465108,
+      "grad_norm": 2.2826199531555176,
+      "learning_rate": 4.17265786475242e-05,
+      "loss": 2.3018,
+      "step": 1701
+    },
+    {
+      "epoch": 0.5605829799497715,
+      "grad_norm": 2.2217764854431152,
+      "learning_rate": 4.167470302725114e-05,
+      "loss": 2.3331,
+      "step": 1702
+    },
+    {
+      "epoch": 0.5609123471530323,
+      "grad_norm": 2.2242486476898193,
+      "learning_rate": 4.16228366224978e-05,
+      "loss": 2.2806,
+      "step": 1703
+    },
+    {
+      "epoch": 0.561241714356293,
+      "grad_norm": 2.391951560974121,
+      "learning_rate": 4.157097949067664e-05,
+      "loss": 2.4716,
+      "step": 1704
+    },
+    {
+      "epoch": 0.5615710815595537,
+      "grad_norm": 2.481973886489868,
+      "learning_rate": 4.1519131689189894e-05,
+      "loss": 2.1806,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5619004487628144,
+      "grad_norm": 2.7211787700653076,
+      "learning_rate": 4.146729327542942e-05,
+      "loss": 2.3138,
+      "step": 1706
+    },
+    {
+      "epoch": 0.5622298159660751,
+      "grad_norm": 2.246211528778076,
+      "learning_rate": 4.1415464306776704e-05,
+      "loss": 2.339,
+      "step": 1707
+    },
+    {
+      "epoch": 0.562559183169336,
+      "grad_norm": 2.3714375495910645,
+      "learning_rate": 4.136364484060279e-05,
+      "loss": 2.3938,
+      "step": 1708
+    },
+    {
+      "epoch": 0.5628885503725967,
+      "grad_norm": 2.315984010696411,
+      "learning_rate": 4.131183493426819e-05,
+      "loss": 2.0192,
+      "step": 1709
+    },
+    {
+      "epoch": 0.5632179175758574,
+      "grad_norm": 3.121124029159546,
+      "learning_rate": 4.126003464512283e-05,
+      "loss": 2.5295,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5635472847791181,
+      "grad_norm": 2.801370620727539,
+      "learning_rate": 4.120824403050598e-05,
+      "loss": 2.0249,
+      "step": 1711
+    },
+    {
+      "epoch": 0.5638766519823789,
+      "grad_norm": 2.729098320007324,
+      "learning_rate": 4.115646314774626e-05,
+      "loss": 2.1085,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5642060191856396,
+      "grad_norm": 2.5545506477355957,
+      "learning_rate": 4.110469205416144e-05,
+      "loss": 2.1962,
+      "step": 1713
+    },
+    {
+      "epoch": 0.5645353863889003,
+      "grad_norm": 2.6717350482940674,
+      "learning_rate": 4.105293080705852e-05,
+      "loss": 2.124,
+      "step": 1714
+    },
+    {
+      "epoch": 0.564864753592161,
+      "grad_norm": 2.6053621768951416,
+      "learning_rate": 4.100117946373353e-05,
+      "loss": 2.4692,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5651941207954218,
+      "grad_norm": 2.648757219314575,
+      "learning_rate": 4.09494380814716e-05,
+      "loss": 2.2406,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5655234879986826,
+      "grad_norm": 2.497184991836548,
+      "learning_rate": 4.089770671754683e-05,
+      "loss": 1.9408,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5658528552019433,
+      "grad_norm": 3.321964740753174,
+      "learning_rate": 4.0845985429222156e-05,
+      "loss": 1.9579,
+      "step": 1718
+    },
+    {
+      "epoch": 0.566182222405204,
+      "grad_norm": 3.0194036960601807,
+      "learning_rate": 4.079427427374945e-05,
+      "loss": 2.1183,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5665115896084647,
+      "grad_norm": 3.1421256065368652,
+      "learning_rate": 4.0742573308369356e-05,
+      "loss": 2.041,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5668409568117255,
+      "grad_norm": 2.7292587757110596,
+      "learning_rate": 4.069088259031117e-05,
+      "loss": 1.9314,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5671703240149862,
+      "grad_norm": 2.4826438426971436,
+      "learning_rate": 4.0639202176792914e-05,
+      "loss": 1.9008,
+      "step": 1722
+    },
+    {
+      "epoch": 0.567499691218247,
+      "grad_norm": 3.0865981578826904,
+      "learning_rate": 4.0587532125021173e-05,
+      "loss": 2.1393,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5678290584215077,
+      "grad_norm": 3.053807020187378,
+      "learning_rate": 4.053587249219108e-05,
+      "loss": 2.0525,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5681584256247684,
+      "grad_norm": 3.325843572616577,
+      "learning_rate": 4.048422333548622e-05,
+      "loss": 1.8684,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5684877928280292,
+      "grad_norm": 2.406651020050049,
+      "learning_rate": 4.043258471207858e-05,
+      "loss": 2.2508,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5688171600312899,
+      "grad_norm": 2.377819061279297,
+      "learning_rate": 4.038095667912851e-05,
+      "loss": 2.4474,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5691465272345506,
+      "grad_norm": 2.348951816558838,
+      "learning_rate": 4.03293392937846e-05,
+      "loss": 2.4961,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5694758944378113,
+      "grad_norm": 2.3945837020874023,
+      "learning_rate": 4.027773261318368e-05,
+      "loss": 2.359,
+      "step": 1729
+    },
+    {
+      "epoch": 0.569805261641072,
+      "grad_norm": 2.5263919830322266,
+      "learning_rate": 4.022613669445075e-05,
+      "loss": 2.2331,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5701346288443329,
+      "grad_norm": 2.5810630321502686,
+      "learning_rate": 4.0174551594698836e-05,
+      "loss": 2.0708,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5704639960475936,
+      "grad_norm": 2.346571207046509,
+      "learning_rate": 4.012297737102903e-05,
+      "loss": 2.222,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5707933632508543,
+      "grad_norm": 2.419288158416748,
+      "learning_rate": 4.00714140805304e-05,
+      "loss": 2.1208,
+      "step": 1733
+    },
+    {
+      "epoch": 0.571122730454115,
+      "grad_norm": 2.1099019050598145,
+      "learning_rate": 4.0019861780279886e-05,
+      "loss": 2.0208,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5714520976573758,
+      "grad_norm": 2.775758981704712,
+      "learning_rate": 3.9968320527342265e-05,
+      "loss": 2.4634,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5717814648606365,
+      "grad_norm": 2.646881103515625,
+      "learning_rate": 3.991679037877008e-05,
+      "loss": 2.1785,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5721108320638972,
+      "grad_norm": 2.8634984493255615,
+      "learning_rate": 3.98652713916036e-05,
+      "loss": 2.4894,
+      "step": 1737
+    },
+    {
+      "epoch": 0.572440199267158,
+      "grad_norm": 2.8133482933044434,
+      "learning_rate": 3.981376362287072e-05,
+      "loss": 2.1222,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5727695664704187,
+      "grad_norm": 2.9048733711242676,
+      "learning_rate": 3.9762267129586934e-05,
+      "loss": 2.2615,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5730989336736795,
+      "grad_norm": 2.7857484817504883,
+      "learning_rate": 3.971078196875526e-05,
+      "loss": 2.0236,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5734283008769402,
+      "grad_norm": 2.681157350540161,
+      "learning_rate": 3.965930819736613e-05,
+      "loss": 1.9672,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5737576680802009,
+      "grad_norm": 2.8750627040863037,
+      "learning_rate": 3.960784587239741e-05,
+      "loss": 2.0623,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5740870352834616,
+      "grad_norm": 2.7466681003570557,
+      "learning_rate": 3.95563950508143e-05,
+      "loss": 2.1302,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5744164024867224,
+      "grad_norm": 3.013389825820923,
+      "learning_rate": 3.950495578956923e-05,
+      "loss": 2.1544,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5747457696899831,
+      "grad_norm": 2.87448787689209,
+      "learning_rate": 3.9453528145601875e-05,
+      "loss": 2.085,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5750751368932439,
+      "grad_norm": 2.522451162338257,
+      "learning_rate": 3.9402112175839005e-05,
+      "loss": 2.0753,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5754045040965046,
+      "grad_norm": 2.751380681991577,
+      "learning_rate": 3.9350707937194504e-05,
+      "loss": 2.3245,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5757338712997653,
+      "grad_norm": 2.8028810024261475,
+      "learning_rate": 3.929931548656925e-05,
+      "loss": 1.9075,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5760632385030261,
+      "grad_norm": 2.744546413421631,
+      "learning_rate": 3.924793488085111e-05,
+      "loss": 1.889,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5763926057062868,
+      "grad_norm": 2.8225655555725098,
+      "learning_rate": 3.9196566176914775e-05,
+      "loss": 1.7777,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5767219729095475,
+      "grad_norm": 2.291588306427002,
+      "learning_rate": 3.91452094316218e-05,
+      "loss": 2.5541,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5770513401128082,
+      "grad_norm": 2.511906862258911,
+      "learning_rate": 3.909386470182053e-05,
+      "loss": 2.2869,
+      "step": 1752
+    },
+    {
+      "epoch": 0.577380707316069,
+      "grad_norm": 2.5849432945251465,
+      "learning_rate": 3.9042532044345934e-05,
+      "loss": 2.3356,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5777100745193298,
+      "grad_norm": 2.452805280685425,
+      "learning_rate": 3.899121151601969e-05,
+      "loss": 2.3691,
+      "step": 1754
+    },
+    {
+      "epoch": 0.5780394417225905,
+      "grad_norm": 2.3843283653259277,
+      "learning_rate": 3.893990317365003e-05,
+      "loss": 2.2706,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5783688089258512,
+      "grad_norm": 2.5602409839630127,
+      "learning_rate": 3.888860707403167e-05,
+      "loss": 2.2975,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5786981761291119,
+      "grad_norm": 2.4516329765319824,
+      "learning_rate": 3.88373232739458e-05,
+      "loss": 2.2493,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5790275433323727,
+      "grad_norm": 2.7871108055114746,
+      "learning_rate": 3.878605183016001e-05,
+      "loss": 2.2575,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5793569105356334,
+      "grad_norm": 2.2499959468841553,
+      "learning_rate": 3.873479279942815e-05,
+      "loss": 2.1826,
+      "step": 1759
+    },
+    {
+      "epoch": 0.5796862777388941,
+      "grad_norm": 2.571885347366333,
+      "learning_rate": 3.8683546238490396e-05,
+      "loss": 2.0972,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5800156449421549,
+      "grad_norm": 2.976599931716919,
+      "learning_rate": 3.86323122040731e-05,
+      "loss": 2.4199,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5803450121454157,
+      "grad_norm": 2.53981876373291,
+      "learning_rate": 3.858109075288875e-05,
+      "loss": 2.0601,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5806743793486764,
+      "grad_norm": 2.889158248901367,
+      "learning_rate": 3.852988194163587e-05,
+      "loss": 2.5461,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5810037465519371,
+      "grad_norm": 2.411149740219116,
+      "learning_rate": 3.847868582699904e-05,
+      "loss": 2.027,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5813331137551978,
+      "grad_norm": 2.5908608436584473,
+      "learning_rate": 3.8427502465648776e-05,
+      "loss": 2.4121,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5816624809584585,
+      "grad_norm": 2.567713499069214,
+      "learning_rate": 3.8376331914241446e-05,
+      "loss": 1.9522,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5819918481617193,
+      "grad_norm": 2.662276029586792,
+      "learning_rate": 3.832517422941928e-05,
+      "loss": 2.1719,
+      "step": 1767
+    },
+    {
+      "epoch": 0.58232121536498,
+      "grad_norm": 2.723975419998169,
+      "learning_rate": 3.8274029467810245e-05,
+      "loss": 2.2837,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5826505825682408,
+      "grad_norm": 2.5226011276245117,
+      "learning_rate": 3.822289768602799e-05,
+      "loss": 1.8823,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5829799497715015,
+      "grad_norm": 2.713068962097168,
+      "learning_rate": 3.817177894067182e-05,
+      "loss": 2.2014,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5833093169747623,
+      "grad_norm": 2.9934639930725098,
+      "learning_rate": 3.81206732883266e-05,
+      "loss": 2.0643,
+      "step": 1771
+    },
+    {
+      "epoch": 0.583638684178023,
+      "grad_norm": 2.897000312805176,
+      "learning_rate": 3.8069580785562686e-05,
+      "loss": 2.0665,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5839680513812837,
+      "grad_norm": 2.8157401084899902,
+      "learning_rate": 3.8018501488935936e-05,
+      "loss": 1.9893,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5842974185845444,
+      "grad_norm": 3.132835865020752,
+      "learning_rate": 3.796743545498751e-05,
+      "loss": 1.8069,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5846267857878051,
+      "grad_norm": 3.8937413692474365,
+      "learning_rate": 3.791638274024394e-05,
+      "loss": 1.9874,
+      "step": 1775
+    },
+    {
+      "epoch": 0.584956152991066,
+      "grad_norm": 2.0166850090026855,
+      "learning_rate": 3.7865343401217e-05,
+      "loss": 2.5567,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5852855201943267,
+      "grad_norm": 2.2458670139312744,
+      "learning_rate": 3.781431749440365e-05,
+      "loss": 2.3101,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5856148873975874,
+      "grad_norm": 2.270277738571167,
+      "learning_rate": 3.7763305076286e-05,
+      "loss": 2.5257,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5859442546008481,
+      "grad_norm": 2.389111042022705,
+      "learning_rate": 3.7712306203331205e-05,
+      "loss": 2.4234,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5862736218041088,
+      "grad_norm": 2.423489809036255,
+      "learning_rate": 3.766132093199146e-05,
+      "loss": 2.3983,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5866029890073696,
+      "grad_norm": 2.0370967388153076,
+      "learning_rate": 3.761034931870386e-05,
+      "loss": 2.3665,
+      "step": 1781
+    },
+    {
+      "epoch": 0.5869323562106303,
+      "grad_norm": 2.6825292110443115,
+      "learning_rate": 3.7559391419890414e-05,
+      "loss": 2.561,
+      "step": 1782
+    },
+    {
+      "epoch": 0.587261723413891,
+      "grad_norm": 2.5416064262390137,
+      "learning_rate": 3.7508447291957956e-05,
+      "loss": 2.5464,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5875910906171518,
+      "grad_norm": 2.385080099105835,
+      "learning_rate": 3.7457516991298036e-05,
+      "loss": 2.143,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5879204578204126,
+      "grad_norm": 3.0276386737823486,
+      "learning_rate": 3.740660057428694e-05,
+      "loss": 2.3312,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5882498250236733,
+      "grad_norm": 2.4158527851104736,
+      "learning_rate": 3.735569809728556e-05,
+      "loss": 2.2548,
+      "step": 1786
+    },
+    {
+      "epoch": 0.588579192226934,
+      "grad_norm": 2.7139041423797607,
+      "learning_rate": 3.730480961663939e-05,
+      "loss": 2.2324,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5889085594301947,
+      "grad_norm": 2.506220579147339,
+      "learning_rate": 3.7253935188678386e-05,
+      "loss": 2.2346,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5892379266334554,
+      "grad_norm": 2.876122236251831,
+      "learning_rate": 3.720307486971697e-05,
+      "loss": 2.5066,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5895672938367162,
+      "grad_norm": 2.6854052543640137,
+      "learning_rate": 3.715222871605397e-05,
+      "loss": 2.2645,
+      "step": 1790
+    },
+    {
+      "epoch": 0.589896661039977,
+      "grad_norm": 2.4886717796325684,
+      "learning_rate": 3.710139678397249e-05,
+      "loss": 2.0398,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5902260282432377,
+      "grad_norm": 2.610877513885498,
+      "learning_rate": 3.7050579129739904e-05,
+      "loss": 2.2531,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5905553954464984,
+      "grad_norm": 2.7850184440612793,
+      "learning_rate": 3.699977580960782e-05,
+      "loss": 2.1468,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5908847626497592,
+      "grad_norm": 2.5817010402679443,
+      "learning_rate": 3.694898687981193e-05,
+      "loss": 2.2793,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5912141298530199,
+      "grad_norm": 2.5994668006896973,
+      "learning_rate": 3.689821239657202e-05,
+      "loss": 2.0205,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5915434970562806,
+      "grad_norm": 2.431635856628418,
+      "learning_rate": 3.684745241609189e-05,
+      "loss": 1.9347,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5918728642595413,
+      "grad_norm": 2.6019957065582275,
+      "learning_rate": 3.6796706994559255e-05,
+      "loss": 1.982,
+      "step": 1797
+    },
+    {
+      "epoch": 0.592202231462802,
+      "grad_norm": 2.8037219047546387,
+      "learning_rate": 3.6745976188145755e-05,
+      "loss": 1.8823,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5925315986660629,
+      "grad_norm": 2.723085641860962,
+      "learning_rate": 3.6695260053006825e-05,
+      "loss": 1.5098,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5928609658693236,
+      "grad_norm": 2.8144564628601074,
+      "learning_rate": 3.664455864528169e-05,
+      "loss": 1.6849,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5931903330725843,
+      "grad_norm": 2.097994327545166,
+      "learning_rate": 3.659387202109322e-05,
+      "loss": 2.33,
+      "step": 1801
+    },
+    {
+      "epoch": 0.593519700275845,
+      "grad_norm": 2.1393368244171143,
+      "learning_rate": 3.6543200236547956e-05,
+      "loss": 2.2933,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5938490674791058,
+      "grad_norm": 2.3984479904174805,
+      "learning_rate": 3.649254334773604e-05,
+      "loss": 2.487,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5941784346823665,
+      "grad_norm": 2.082158088684082,
+      "learning_rate": 3.6441901410731064e-05,
+      "loss": 2.2317,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5945078018856272,
+      "grad_norm": 2.287839412689209,
+      "learning_rate": 3.639127448159013e-05,
+      "loss": 2.2207,
+      "step": 1805
+    },
+    {
+      "epoch": 0.594837169088888,
+      "grad_norm": 2.7098779678344727,
+      "learning_rate": 3.634066261635366e-05,
+      "loss": 2.7025,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5951665362921487,
+      "grad_norm": 2.2493362426757812,
+      "learning_rate": 3.629006587104546e-05,
+      "loss": 2.2232,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5954959034954095,
+      "grad_norm": 2.7404730319976807,
+      "learning_rate": 3.623948430167258e-05,
+      "loss": 2.2783,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5958252706986702,
+      "grad_norm": 2.522005796432495,
+      "learning_rate": 3.6188917964225256e-05,
+      "loss": 2.1626,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5961546379019309,
+      "grad_norm": 2.4059793949127197,
+      "learning_rate": 3.613836691467688e-05,
+      "loss": 2.338,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5964840051051916,
+      "grad_norm": 2.5347237586975098,
+      "learning_rate": 3.608783120898392e-05,
+      "loss": 2.3689,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5968133723084524,
+      "grad_norm": 2.6831188201904297,
+      "learning_rate": 3.603731090308586e-05,
+      "loss": 2.1,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5971427395117131,
+      "grad_norm": 2.634315013885498,
+      "learning_rate": 3.598680605290513e-05,
+      "loss": 2.4177,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5974721067149739,
+      "grad_norm": 2.7967567443847656,
+      "learning_rate": 3.593631671434706e-05,
+      "loss": 2.406,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5978014739182346,
+      "grad_norm": 2.2884232997894287,
+      "learning_rate": 3.588584294329981e-05,
+      "loss": 1.8473,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5981308411214953,
+      "grad_norm": 2.73101806640625,
+      "learning_rate": 3.5835384795634285e-05,
+      "loss": 2.1567,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5984602083247561,
+      "grad_norm": 2.5716552734375,
+      "learning_rate": 3.578494232720413e-05,
+      "loss": 2.1229,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5987895755280168,
+      "grad_norm": 2.678837537765503,
+      "learning_rate": 3.573451559384563e-05,
+      "loss": 2.1249,
+      "step": 1818
+    },
+    {
+      "epoch": 0.5991189427312775,
+      "grad_norm": 2.7849841117858887,
+      "learning_rate": 3.568410465137762e-05,
+      "loss": 1.8091,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5994483099345382,
+      "grad_norm": 2.5732271671295166,
+      "learning_rate": 3.563370955560147e-05,
+      "loss": 2.0009,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5997776771377991,
+      "grad_norm": 2.8257884979248047,
+      "learning_rate": 3.558333036230105e-05,
+      "loss": 2.0211,
+      "step": 1821
+    },
+    {
+      "epoch": 0.6001070443410598,
+      "grad_norm": 2.7145349979400635,
+      "learning_rate": 3.553296712724256e-05,
+      "loss": 2.0294,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6004364115443205,
+      "grad_norm": 2.5253031253814697,
+      "learning_rate": 3.548261990617459e-05,
+      "loss": 1.754,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6007657787475812,
+      "grad_norm": 3.2652945518493652,
+      "learning_rate": 3.543228875482796e-05,
+      "loss": 2.1409,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6010951459508419,
+      "grad_norm": 4.042289733886719,
+      "learning_rate": 3.538197372891575e-05,
+      "loss": 2.3513,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6014245131541027,
+      "grad_norm": 2.0667660236358643,
+      "learning_rate": 3.533167488413315e-05,
+      "loss": 2.6748,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6017538803573634,
+      "grad_norm": 2.205390691757202,
+      "learning_rate": 3.528139227615744e-05,
+      "loss": 2.2744,
+      "step": 1827
+    },
+    {
+      "epoch": 0.6020832475606241,
+      "grad_norm": 2.3584322929382324,
+      "learning_rate": 3.5231125960647974e-05,
+      "loss": 2.0508,
+      "step": 1828
+    },
+    {
+      "epoch": 0.6024126147638849,
+      "grad_norm": 2.636195659637451,
+      "learning_rate": 3.5180875993246005e-05,
+      "loss": 2.3369,
+      "step": 1829
+    },
+    {
+      "epoch": 0.6027419819671456,
+      "grad_norm": 2.2127506732940674,
+      "learning_rate": 3.513064242957473e-05,
+      "loss": 2.2606,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6030713491704064,
+      "grad_norm": 2.4520390033721924,
+      "learning_rate": 3.50804253252392e-05,
+      "loss": 2.2014,
+      "step": 1831
+    },
+    {
+      "epoch": 0.6034007163736671,
+      "grad_norm": 2.1871538162231445,
+      "learning_rate": 3.503022473582619e-05,
+      "loss": 2.0991,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6037300835769278,
+      "grad_norm": 2.8282971382141113,
+      "learning_rate": 3.498004071690424e-05,
+      "loss": 2.5463,
+      "step": 1833
+    },
+    {
+      "epoch": 0.6040594507801885,
+      "grad_norm": 2.2347660064697266,
+      "learning_rate": 3.492987332402356e-05,
+      "loss": 2.189,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6043888179834493,
+      "grad_norm": 2.6401114463806152,
+      "learning_rate": 3.487972261271594e-05,
+      "loss": 2.2381,
+      "step": 1835
+    },
+    {
+      "epoch": 0.60471818518671,
+      "grad_norm": 2.1282289028167725,
+      "learning_rate": 3.482958863849467e-05,
+      "loss": 2.0018,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6050475523899708,
+      "grad_norm": 2.5486040115356445,
+      "learning_rate": 3.477947145685456e-05,
+      "loss": 2.2497,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6053769195932315,
+      "grad_norm": 2.6828370094299316,
+      "learning_rate": 3.47293711232718e-05,
+      "loss": 2.3501,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6057062867964922,
+      "grad_norm": 3.3055360317230225,
+      "learning_rate": 3.467928769320397e-05,
+      "loss": 2.3995,
+      "step": 1839
+    },
+    {
+      "epoch": 0.606035653999753,
+      "grad_norm": 2.5054500102996826,
+      "learning_rate": 3.462922122208989e-05,
+      "loss": 1.9583,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6063650212030137,
+      "grad_norm": 2.4648640155792236,
+      "learning_rate": 3.457917176534964e-05,
+      "loss": 2.1108,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6066943884062744,
+      "grad_norm": 2.776409149169922,
+      "learning_rate": 3.452913937838446e-05,
+      "loss": 2.2656,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6070237556095351,
+      "grad_norm": 2.3845038414001465,
+      "learning_rate": 3.447912411657669e-05,
+      "loss": 1.9668,
+      "step": 1843
+    },
+    {
+      "epoch": 0.607353122812796,
+      "grad_norm": 2.431183099746704,
+      "learning_rate": 3.442912603528971e-05,
+      "loss": 2.0438,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6076824900160567,
+      "grad_norm": 2.8428471088409424,
+      "learning_rate": 3.43791451898679e-05,
+      "loss": 2.3823,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6080118572193174,
+      "grad_norm": 3.1004912853240967,
+      "learning_rate": 3.432918163563654e-05,
+      "loss": 2.1599,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6083412244225781,
+      "grad_norm": 3.722520112991333,
+      "learning_rate": 3.4279235427901785e-05,
+      "loss": 2.3052,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6086705916258388,
+      "grad_norm": 2.688524007797241,
+      "learning_rate": 3.422930662195058e-05,
+      "loss": 1.7832,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6089999588290996,
+      "grad_norm": 2.663485527038574,
+      "learning_rate": 3.417939527305062e-05,
+      "loss": 1.9782,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6093293260323603,
+      "grad_norm": 3.491044759750366,
+      "learning_rate": 3.412950143645025e-05,
+      "loss": 1.9132,
+      "step": 1850
+    },
+    {
+      "epoch": 0.609658693235621,
+      "grad_norm": 1.7720582485198975,
+      "learning_rate": 3.407962516737846e-05,
+      "loss": 2.376,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6099880604388818,
+      "grad_norm": 2.126549005508423,
+      "learning_rate": 3.402976652104477e-05,
+      "loss": 1.8721,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6103174276421426,
+      "grad_norm": 2.1321988105773926,
+      "learning_rate": 3.3979925552639224e-05,
+      "loss": 2.2343,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6106467948454033,
+      "grad_norm": 2.4410877227783203,
+      "learning_rate": 3.3930102317332255e-05,
+      "loss": 2.3504,
+      "step": 1854
+    },
+    {
+      "epoch": 0.610976162048664,
+      "grad_norm": 2.425161361694336,
+      "learning_rate": 3.38802968702747e-05,
+      "loss": 2.1846,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6113055292519247,
+      "grad_norm": 2.4529266357421875,
+      "learning_rate": 3.383050926659771e-05,
+      "loss": 2.3507,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6116348964551854,
+      "grad_norm": 2.2834224700927734,
+      "learning_rate": 3.378073956141263e-05,
+      "loss": 2.2235,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6119642636584463,
+      "grad_norm": 2.8359508514404297,
+      "learning_rate": 3.3730987809811064e-05,
+      "loss": 2.3002,
+      "step": 1858
+    },
+    {
+      "epoch": 0.612293630861707,
+      "grad_norm": 2.947082042694092,
+      "learning_rate": 3.368125406686472e-05,
+      "loss": 2.2171,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6126229980649677,
+      "grad_norm": 2.474220037460327,
+      "learning_rate": 3.3631538387625325e-05,
+      "loss": 2.3359,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6129523652682284,
+      "grad_norm": 2.4733357429504395,
+      "learning_rate": 3.3581840827124665e-05,
+      "loss": 2.3127,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6132817324714892,
+      "grad_norm": 2.6637370586395264,
+      "learning_rate": 3.353216144037448e-05,
+      "loss": 2.2253,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6136110996747499,
+      "grad_norm": 2.6768898963928223,
+      "learning_rate": 3.3482500282366304e-05,
+      "loss": 2.0104,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6139404668780106,
+      "grad_norm": 2.4703683853149414,
+      "learning_rate": 3.3432857408071626e-05,
+      "loss": 2.4931,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6142698340812713,
+      "grad_norm": 2.216553211212158,
+      "learning_rate": 3.338323287244158e-05,
+      "loss": 2.0299,
+      "step": 1865
+    },
+    {
+      "epoch": 0.614599201284532,
+      "grad_norm": 2.8111774921417236,
+      "learning_rate": 3.333362673040706e-05,
+      "loss": 2.5135,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6149285684877929,
+      "grad_norm": 2.6286466121673584,
+      "learning_rate": 3.328403903687859e-05,
+      "loss": 2.1726,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6152579356910536,
+      "grad_norm": 2.4072277545928955,
+      "learning_rate": 3.323446984674627e-05,
+      "loss": 2.1707,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6155873028943143,
+      "grad_norm": 2.4936254024505615,
+      "learning_rate": 3.3184919214879696e-05,
+      "loss": 1.993,
+      "step": 1869
+    },
+    {
+      "epoch": 0.615916670097575,
+      "grad_norm": 2.5818185806274414,
+      "learning_rate": 3.313538719612796e-05,
+      "loss": 2.1305,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6162460373008357,
+      "grad_norm": 2.4668569564819336,
+      "learning_rate": 3.308587384531953e-05,
+      "loss": 1.7286,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6165754045040965,
+      "grad_norm": 2.7164247035980225,
+      "learning_rate": 3.30363792172622e-05,
+      "loss": 1.9667,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6169047717073572,
+      "grad_norm": 2.8039917945861816,
+      "learning_rate": 3.2986903366743056e-05,
+      "loss": 1.9472,
+      "step": 1873
+    },
+    {
+      "epoch": 0.617234138910618,
+      "grad_norm": 2.910883903503418,
+      "learning_rate": 3.293744634852841e-05,
+      "loss": 2.1599,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6175635061138787,
+      "grad_norm": 2.8668439388275146,
+      "learning_rate": 3.288800821736369e-05,
+      "loss": 1.6236,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6178928733171395,
+      "grad_norm": 1.9064861536026,
+      "learning_rate": 3.2838589027973444e-05,
+      "loss": 2.2323,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6182222405204002,
+      "grad_norm": 2.109698534011841,
+      "learning_rate": 3.278918883506126e-05,
+      "loss": 2.2206,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6185516077236609,
+      "grad_norm": 2.2354044914245605,
+      "learning_rate": 3.2739807693309675e-05,
+      "loss": 2.2063,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6188809749269216,
+      "grad_norm": 2.030697822570801,
+      "learning_rate": 3.269044565738014e-05,
+      "loss": 2.0355,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6192103421301823,
+      "grad_norm": 2.2695322036743164,
+      "learning_rate": 3.2641102781912994e-05,
+      "loss": 2.0768,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6195397093334432,
+      "grad_norm": 2.603289842605591,
+      "learning_rate": 3.259177912152732e-05,
+      "loss": 2.2277,
+      "step": 1881
+    },
+    {
+      "epoch": 0.6198690765367039,
+      "grad_norm": 2.7053635120391846,
+      "learning_rate": 3.254247473082094e-05,
+      "loss": 2.565,
+      "step": 1882
+    },
+    {
+      "epoch": 0.6201984437399646,
+      "grad_norm": 2.517655849456787,
+      "learning_rate": 3.249318966437037e-05,
+      "loss": 2.2004,
+      "step": 1883
+    },
+    {
+      "epoch": 0.6205278109432253,
+      "grad_norm": 2.3863282203674316,
+      "learning_rate": 3.244392397673073e-05,
+      "loss": 1.9685,
+      "step": 1884
+    },
+    {
+      "epoch": 0.6208571781464861,
+      "grad_norm": 2.740966320037842,
+      "learning_rate": 3.239467772243566e-05,
+      "loss": 2.052,
+      "step": 1885
+    },
+    {
+      "epoch": 0.6211865453497468,
+      "grad_norm": 2.55629301071167,
+      "learning_rate": 3.234545095599732e-05,
+      "loss": 2.1718,
+      "step": 1886
+    },
+    {
+      "epoch": 0.6215159125530075,
+      "grad_norm": 2.6892268657684326,
+      "learning_rate": 3.22962437319063e-05,
+      "loss": 2.1374,
+      "step": 1887
+    },
+    {
+      "epoch": 0.6218452797562682,
+      "grad_norm": 2.641112804412842,
+      "learning_rate": 3.2247056104631505e-05,
+      "loss": 2.2445,
+      "step": 1888
+    },
+    {
+      "epoch": 0.622174646959529,
+      "grad_norm": 2.7154104709625244,
+      "learning_rate": 3.2197888128620224e-05,
+      "loss": 2.1997,
+      "step": 1889
+    },
+    {
+      "epoch": 0.6225040141627898,
+      "grad_norm": 2.327333450317383,
+      "learning_rate": 3.2148739858297936e-05,
+      "loss": 2.0276,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6228333813660505,
+      "grad_norm": 2.518122434616089,
+      "learning_rate": 3.209961134806836e-05,
+      "loss": 2.1551,
+      "step": 1891
+    },
+    {
+      "epoch": 0.6231627485693112,
+      "grad_norm": 2.6456973552703857,
+      "learning_rate": 3.205050265231327e-05,
+      "loss": 2.1652,
+      "step": 1892
+    },
+    {
+      "epoch": 0.6234921157725719,
+      "grad_norm": 2.701111316680908,
+      "learning_rate": 3.2001413825392574e-05,
+      "loss": 2.2113,
+      "step": 1893
+    },
+    {
+      "epoch": 0.6238214829758327,
+      "grad_norm": 2.5059456825256348,
+      "learning_rate": 3.195234492164414e-05,
+      "loss": 2.0026,
+      "step": 1894
+    },
+    {
+      "epoch": 0.6241508501790934,
+      "grad_norm": 2.855924367904663,
+      "learning_rate": 3.190329599538382e-05,
+      "loss": 2.1053,
+      "step": 1895
+    },
+    {
+      "epoch": 0.6244802173823542,
+      "grad_norm": 2.850864887237549,
+      "learning_rate": 3.185426710090534e-05,
+      "loss": 2.3436,
+      "step": 1896
+    },
+    {
+      "epoch": 0.6248095845856149,
+      "grad_norm": 2.7715375423431396,
+      "learning_rate": 3.180525829248023e-05,
+      "loss": 1.9533,
+      "step": 1897
+    },
+    {
+      "epoch": 0.6251389517888756,
+      "grad_norm": 3.283682107925415,
+      "learning_rate": 3.1756269624357806e-05,
+      "loss": 1.8841,
+      "step": 1898
+    },
+    {
+      "epoch": 0.6254683189921364,
+      "grad_norm": 2.5927112102508545,
+      "learning_rate": 3.17073011507651e-05,
+      "loss": 1.5226,
+      "step": 1899
+    },
+    {
+      "epoch": 0.6257976861953971,
+      "grad_norm": 3.1886491775512695,
+      "learning_rate": 3.165835292590675e-05,
+      "loss": 2.0477,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6261270533986578,
+      "grad_norm": 2.009676218032837,
+      "learning_rate": 3.160942500396503e-05,
+      "loss": 2.5363,
+      "step": 1901
+    },
+    {
+      "epoch": 0.6264564206019185,
+      "grad_norm": 2.3882784843444824,
+      "learning_rate": 3.1560517439099715e-05,
+      "loss": 2.4409,
+      "step": 1902
+    },
+    {
+      "epoch": 0.6267857878051794,
+      "grad_norm": 2.1249465942382812,
+      "learning_rate": 3.151163028544804e-05,
+      "loss": 2.4448,
+      "step": 1903
+    },
+    {
+      "epoch": 0.6271151550084401,
+      "grad_norm": 2.7128207683563232,
+      "learning_rate": 3.146276359712466e-05,
+      "loss": 2.3409,
+      "step": 1904
+    },
+    {
+      "epoch": 0.6274445222117008,
+      "grad_norm": 2.4485268592834473,
+      "learning_rate": 3.141391742822156e-05,
+      "loss": 2.4835,
+      "step": 1905
+    },
+    {
+      "epoch": 0.6277738894149615,
+      "grad_norm": 2.6135315895080566,
+      "learning_rate": 3.136509183280805e-05,
+      "loss": 2.4324,
+      "step": 1906
+    },
+    {
+      "epoch": 0.6281032566182222,
+      "grad_norm": 2.25262713432312,
+      "learning_rate": 3.131628686493061e-05,
+      "loss": 2.3033,
+      "step": 1907
+    },
+    {
+      "epoch": 0.628432623821483,
+      "grad_norm": 2.3247017860412598,
+      "learning_rate": 3.1267502578612926e-05,
+      "loss": 2.1639,
+      "step": 1908
+    },
+    {
+      "epoch": 0.6287619910247437,
+      "grad_norm": 2.385807514190674,
+      "learning_rate": 3.121873902785579e-05,
+      "loss": 2.1444,
+      "step": 1909
+    },
+    {
+      "epoch": 0.6290913582280044,
+      "grad_norm": 2.7242722511291504,
+      "learning_rate": 3.116999626663701e-05,
+      "loss": 2.3064,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6294207254312651,
+      "grad_norm": 2.3983850479125977,
+      "learning_rate": 3.112127434891143e-05,
+      "loss": 2.275,
+      "step": 1911
+    },
+    {
+      "epoch": 0.629750092634526,
+      "grad_norm": 2.6150460243225098,
+      "learning_rate": 3.107257332861078e-05,
+      "loss": 2.321,
+      "step": 1912
+    },
+    {
+      "epoch": 0.6300794598377867,
+      "grad_norm": 2.5469534397125244,
+      "learning_rate": 3.1023893259643666e-05,
+      "loss": 1.8771,
+      "step": 1913
+    },
+    {
+      "epoch": 0.6304088270410474,
+      "grad_norm": 2.3631060123443604,
+      "learning_rate": 3.0975234195895526e-05,
+      "loss": 1.8878,
+      "step": 1914
+    },
+    {
+      "epoch": 0.6307381942443081,
+      "grad_norm": 2.889796495437622,
+      "learning_rate": 3.092659619122852e-05,
+      "loss": 2.1742,
+      "step": 1915
+    },
+    {
+      "epoch": 0.6310675614475688,
+      "grad_norm": 2.6340792179107666,
+      "learning_rate": 3.087797929948151e-05,
+      "loss": 1.9413,
+      "step": 1916
+    },
+    {
+      "epoch": 0.6313969286508296,
+      "grad_norm": 2.559239149093628,
+      "learning_rate": 3.0829383574469976e-05,
+      "loss": 2.0802,
+      "step": 1917
+    },
+    {
+      "epoch": 0.6317262958540903,
+      "grad_norm": 2.652400016784668,
+      "learning_rate": 3.078080906998599e-05,
+      "loss": 2.24,
+      "step": 1918
+    },
+    {
+      "epoch": 0.6320556630573511,
+      "grad_norm": 2.5503621101379395,
+      "learning_rate": 3.073225583979812e-05,
+      "loss": 1.9102,
+      "step": 1919
+    },
+    {
+      "epoch": 0.6323850302606118,
+      "grad_norm": 2.8662219047546387,
+      "learning_rate": 3.068372393765137e-05,
+      "loss": 2.2393,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6327143974638725,
+      "grad_norm": 3.3057799339294434,
+      "learning_rate": 3.063521341726717e-05,
+      "loss": 2.1511,
+      "step": 1921
+    },
+    {
+      "epoch": 0.6330437646671333,
+      "grad_norm": 2.57186222076416,
+      "learning_rate": 3.0586724332343266e-05,
+      "loss": 1.9987,
+      "step": 1922
+    },
+    {
+      "epoch": 0.633373131870394,
+      "grad_norm": 2.6434319019317627,
+      "learning_rate": 3.053825673655365e-05,
+      "loss": 1.7956,
+      "step": 1923
+    },
+    {
+      "epoch": 0.6337024990736547,
+      "grad_norm": 2.6144514083862305,
+      "learning_rate": 3.0489810683548546e-05,
+      "loss": 2.0272,
+      "step": 1924
+    },
+    {
+      "epoch": 0.6340318662769154,
+      "grad_norm": 3.186408281326294,
+      "learning_rate": 3.0441386226954372e-05,
+      "loss": 1.9882,
+      "step": 1925
+    },
+    {
+      "epoch": 0.6343612334801763,
+      "grad_norm": 1.8592625856399536,
+      "learning_rate": 3.0392983420373577e-05,
+      "loss": 2.4248,
+      "step": 1926
+    },
+    {
+      "epoch": 0.634690600683437,
+      "grad_norm": 2.230592966079712,
+      "learning_rate": 3.0344602317384695e-05,
+      "loss": 2.476,
+      "step": 1927
+    },
+    {
+      "epoch": 0.6350199678866977,
+      "grad_norm": 2.3445613384246826,
+      "learning_rate": 3.0296242971542194e-05,
+      "loss": 2.1382,
+      "step": 1928
+    },
+    {
+      "epoch": 0.6353493350899584,
+      "grad_norm": 2.5234224796295166,
+      "learning_rate": 3.024790543637648e-05,
+      "loss": 2.5123,
+      "step": 1929
+    },
+    {
+      "epoch": 0.6356787022932191,
+      "grad_norm": 2.1975531578063965,
+      "learning_rate": 3.0199589765393823e-05,
+      "loss": 2.071,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6360080694964799,
+      "grad_norm": 2.5757479667663574,
+      "learning_rate": 3.015129601207627e-05,
+      "loss": 2.2572,
+      "step": 1931
+    },
+    {
+      "epoch": 0.6363374366997406,
+      "grad_norm": 2.23809552192688,
+      "learning_rate": 3.0103024229881617e-05,
+      "loss": 1.8197,
+      "step": 1932
+    },
+    {
+      "epoch": 0.6366668039030013,
+      "grad_norm": 2.522535562515259,
+      "learning_rate": 3.0054774472243346e-05,
+      "loss": 2.4502,
+      "step": 1933
+    },
+    {
+      "epoch": 0.636996171106262,
+      "grad_norm": 2.562291145324707,
+      "learning_rate": 3.0006546792570566e-05,
+      "loss": 2.3762,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6373255383095229,
+      "grad_norm": 2.2808642387390137,
+      "learning_rate": 2.9958341244247913e-05,
+      "loss": 2.0,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6376549055127836,
+      "grad_norm": 2.5651497840881348,
+      "learning_rate": 2.991015788063556e-05,
+      "loss": 2.2833,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6379842727160443,
+      "grad_norm": 2.647606372833252,
+      "learning_rate": 2.9861996755069112e-05,
+      "loss": 2.1614,
+      "step": 1937
+    },
+    {
+      "epoch": 0.638313639919305,
+      "grad_norm": 3.004729747772217,
+      "learning_rate": 2.9813857920859544e-05,
+      "loss": 2.3966,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6386430071225657,
+      "grad_norm": 2.419417142868042,
+      "learning_rate": 2.9765741431293175e-05,
+      "loss": 2.2321,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6389723743258265,
+      "grad_norm": 2.413052558898926,
+      "learning_rate": 2.97176473396316e-05,
+      "loss": 2.2033,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6393017415290873,
+      "grad_norm": 2.419487237930298,
+      "learning_rate": 2.9669575699111575e-05,
+      "loss": 2.2214,
+      "step": 1941
+    },
+    {
+      "epoch": 0.639631108732348,
+      "grad_norm": 2.965186834335327,
+      "learning_rate": 2.9621526562945058e-05,
+      "loss": 2.3811,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6399604759356087,
+      "grad_norm": 2.3459784984588623,
+      "learning_rate": 2.9573499984319053e-05,
+      "loss": 2.0426,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6402898431388695,
+      "grad_norm": 2.5563299655914307,
+      "learning_rate": 2.9525496016395637e-05,
+      "loss": 2.0472,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6406192103421302,
+      "grad_norm": 2.555187225341797,
+      "learning_rate": 2.9477514712311803e-05,
+      "loss": 1.9391,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6409485775453909,
+      "grad_norm": 2.444321632385254,
+      "learning_rate": 2.942955612517952e-05,
+      "loss": 1.6775,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6412779447486516,
+      "grad_norm": 2.7038373947143555,
+      "learning_rate": 2.9381620308085566e-05,
+      "loss": 1.811,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6416073119519123,
+      "grad_norm": 3.07734751701355,
+      "learning_rate": 2.9333707314091525e-05,
+      "loss": 2.1634,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6419366791551732,
+      "grad_norm": 3.2388222217559814,
+      "learning_rate": 2.9285817196233722e-05,
+      "loss": 1.7891,
+      "step": 1949
+    },
+    {
+      "epoch": 0.6422660463584339,
+      "grad_norm": 3.074317216873169,
+      "learning_rate": 2.9237950007523164e-05,
+      "loss": 1.944,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6425954135616946,
+      "grad_norm": 2.3127830028533936,
+      "learning_rate": 2.919010580094546e-05,
+      "loss": 2.2439,
+      "step": 1951
+    },
+    {
+      "epoch": 0.6429247807649553,
+      "grad_norm": 2.1176540851593018,
+      "learning_rate": 2.91422846294608e-05,
+      "loss": 2.2749,
+      "step": 1952
+    },
+    {
+      "epoch": 0.6432541479682161,
+      "grad_norm": 2.5770678520202637,
+      "learning_rate": 2.9094486546003857e-05,
+      "loss": 2.2961,
+      "step": 1953
+    },
+    {
+      "epoch": 0.6435835151714768,
+      "grad_norm": 2.203704595565796,
+      "learning_rate": 2.9046711603483766e-05,
+      "loss": 2.1581,
+      "step": 1954
+    },
+    {
+      "epoch": 0.6439128823747375,
+      "grad_norm": 2.189901828765869,
+      "learning_rate": 2.8998959854784026e-05,
+      "loss": 2.4594,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6442422495779982,
+      "grad_norm": 2.659796714782715,
+      "learning_rate": 2.8951231352762486e-05,
+      "loss": 2.3197,
+      "step": 1956
+    },
+    {
+      "epoch": 0.644571616781259,
+      "grad_norm": 2.2193856239318848,
+      "learning_rate": 2.890352615025124e-05,
+      "loss": 2.2613,
+      "step": 1957
+    },
+    {
+      "epoch": 0.6449009839845198,
+      "grad_norm": 2.119441270828247,
+      "learning_rate": 2.885584430005661e-05,
+      "loss": 2.2418,
+      "step": 1958
+    },
+    {
+      "epoch": 0.6452303511877805,
+      "grad_norm": 2.3953168392181396,
+      "learning_rate": 2.8808185854959047e-05,
+      "loss": 2.2826,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6455597183910412,
+      "grad_norm": 2.431501626968384,
+      "learning_rate": 2.876055086771313e-05,
+      "loss": 2.21,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6458890855943019,
+      "grad_norm": 2.9048917293548584,
+      "learning_rate": 2.871293939104742e-05,
+      "loss": 2.5832,
+      "step": 1961
+    },
+    {
+      "epoch": 0.6462184527975627,
+      "grad_norm": 2.4806578159332275,
+      "learning_rate": 2.8665351477664492e-05,
+      "loss": 2.4109,
+      "step": 1962
+    },
+    {
+      "epoch": 0.6465478200008234,
+      "grad_norm": 2.117027521133423,
+      "learning_rate": 2.861778718024082e-05,
+      "loss": 1.8737,
+      "step": 1963
+    },
+    {
+      "epoch": 0.6468771872040842,
+      "grad_norm": 2.629340410232544,
+      "learning_rate": 2.8570246551426762e-05,
+      "loss": 2.6089,
+      "step": 1964
+    },
+    {
+      "epoch": 0.6472065544073449,
+      "grad_norm": 2.8343591690063477,
+      "learning_rate": 2.852272964384644e-05,
+      "loss": 2.2693,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6475359216106056,
+      "grad_norm": 2.3691017627716064,
+      "learning_rate": 2.8475236510097752e-05,
+      "loss": 2.2384,
+      "step": 1966
+    },
+    {
+      "epoch": 0.6478652888138664,
+      "grad_norm": 2.574763536453247,
+      "learning_rate": 2.842776720275228e-05,
+      "loss": 2.1394,
+      "step": 1967
+    },
+    {
+      "epoch": 0.6481946560171271,
+      "grad_norm": 2.5222694873809814,
+      "learning_rate": 2.838032177435518e-05,
+      "loss": 1.6567,
+      "step": 1968
+    },
+    {
+      "epoch": 0.6485240232203878,
+      "grad_norm": 2.61596941947937,
+      "learning_rate": 2.8332900277425233e-05,
+      "loss": 2.2399,
+      "step": 1969
+    },
+    {
+      "epoch": 0.6488533904236485,
+      "grad_norm": 2.659245252609253,
+      "learning_rate": 2.8285502764454703e-05,
+      "loss": 1.9834,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6491827576269092,
+      "grad_norm": 2.8906190395355225,
+      "learning_rate": 2.8238129287909314e-05,
+      "loss": 2.1137,
+      "step": 1971
+    },
+    {
+      "epoch": 0.6495121248301701,
+      "grad_norm": 2.697786331176758,
+      "learning_rate": 2.8190779900228185e-05,
+      "loss": 1.913,
+      "step": 1972
+    },
+    {
+      "epoch": 0.6498414920334308,
+      "grad_norm": 3.1825335025787354,
+      "learning_rate": 2.8143454653823787e-05,
+      "loss": 2.0227,
+      "step": 1973
+    },
+    {
+      "epoch": 0.6501708592366915,
+      "grad_norm": 3.1395328044891357,
+      "learning_rate": 2.8096153601081805e-05,
+      "loss": 2.05,
+      "step": 1974
+    },
+    {
+      "epoch": 0.6505002264399522,
+      "grad_norm": 3.7124152183532715,
+      "learning_rate": 2.8048876794361206e-05,
+      "loss": 2.0609,
+      "step": 1975
+    },
+    {
+      "epoch": 0.650829593643213,
+      "grad_norm": 1.9795833826065063,
+      "learning_rate": 2.80016242859941e-05,
+      "loss": 2.3731,
+      "step": 1976
+    },
+    {
+      "epoch": 0.6511589608464737,
+      "grad_norm": 2.223407745361328,
+      "learning_rate": 2.7954396128285698e-05,
+      "loss": 2.5673,
+      "step": 1977
+    },
+    {
+      "epoch": 0.6514883280497344,
+      "grad_norm": 1.970519781112671,
+      "learning_rate": 2.7907192373514256e-05,
+      "loss": 2.3043,
+      "step": 1978
+    },
+    {
+      "epoch": 0.6518176952529952,
+      "grad_norm": 2.7503855228424072,
+      "learning_rate": 2.7860013073931024e-05,
+      "loss": 2.5426,
+      "step": 1979
+    },
+    {
+      "epoch": 0.6521470624562559,
+      "grad_norm": 2.135634183883667,
+      "learning_rate": 2.781285828176019e-05,
+      "loss": 2.0703,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6524764296595167,
+      "grad_norm": 2.544966220855713,
+      "learning_rate": 2.7765728049198768e-05,
+      "loss": 2.2,
+      "step": 1981
+    },
+    {
+      "epoch": 0.6528057968627774,
+      "grad_norm": 2.2923483848571777,
+      "learning_rate": 2.771862242841663e-05,
+      "loss": 2.0587,
+      "step": 1982
+    },
+    {
+      "epoch": 0.6531351640660381,
+      "grad_norm": 2.587148427963257,
+      "learning_rate": 2.7671541471556404e-05,
+      "loss": 2.542,
+      "step": 1983
+    },
+    {
+      "epoch": 0.6534645312692988,
+      "grad_norm": 2.211090564727783,
+      "learning_rate": 2.7624485230733403e-05,
+      "loss": 2.2066,
+      "step": 1984
+    },
+    {
+      "epoch": 0.6537938984725596,
+      "grad_norm": 2.4215967655181885,
+      "learning_rate": 2.7577453758035588e-05,
+      "loss": 2.0838,
+      "step": 1985
+    },
+    {
+      "epoch": 0.6541232656758204,
+      "grad_norm": 2.3949782848358154,
+      "learning_rate": 2.753044710552349e-05,
+      "loss": 2.002,
+      "step": 1986
+    },
+    {
+      "epoch": 0.6544526328790811,
+      "grad_norm": 2.762983560562134,
+      "learning_rate": 2.748346532523018e-05,
+      "loss": 2.3399,
+      "step": 1987
+    },
+    {
+      "epoch": 0.6547820000823418,
+      "grad_norm": 2.6567108631134033,
+      "learning_rate": 2.7436508469161195e-05,
+      "loss": 2.2037,
+      "step": 1988
+    },
+    {
+      "epoch": 0.6551113672856025,
+      "grad_norm": 2.348445415496826,
+      "learning_rate": 2.7389576589294486e-05,
+      "loss": 2.0968,
+      "step": 1989
+    },
+    {
+      "epoch": 0.6554407344888633,
+      "grad_norm": 2.819063663482666,
+      "learning_rate": 2.734266973758034e-05,
+      "loss": 2.4161,
+      "step": 1990
+    },
+    {
+      "epoch": 0.655770101692124,
+      "grad_norm": 2.5999984741210938,
+      "learning_rate": 2.7295787965941355e-05,
+      "loss": 2.1207,
+      "step": 1991
+    },
+    {
+      "epoch": 0.6560994688953847,
+      "grad_norm": 2.612342357635498,
+      "learning_rate": 2.7248931326272386e-05,
+      "loss": 2.018,
+      "step": 1992
+    },
+    {
+      "epoch": 0.6564288360986454,
+      "grad_norm": 2.783804416656494,
+      "learning_rate": 2.720209987044041e-05,
+      "loss": 2.2325,
+      "step": 1993
+    },
+    {
+      "epoch": 0.6567582033019063,
+      "grad_norm": 2.30511212348938,
+      "learning_rate": 2.7155293650284573e-05,
+      "loss": 1.9004,
+      "step": 1994
+    },
+    {
+      "epoch": 0.657087570505167,
+      "grad_norm": 2.7158255577087402,
+      "learning_rate": 2.710851271761609e-05,
+      "loss": 1.8252,
+      "step": 1995
+    },
+    {
+      "epoch": 0.6574169377084277,
+      "grad_norm": 2.2666115760803223,
+      "learning_rate": 2.7061757124218162e-05,
+      "loss": 1.7978,
+      "step": 1996
+    },
+    {
+      "epoch": 0.6577463049116884,
+      "grad_norm": 2.959491491317749,
+      "learning_rate": 2.7015026921845952e-05,
+      "loss": 1.9785,
+      "step": 1997
+    },
+    {
+      "epoch": 0.6580756721149491,
+      "grad_norm": 3.017894983291626,
+      "learning_rate": 2.696832216222654e-05,
+      "loss": 1.828,
+      "step": 1998
+    },
+    {
+      "epoch": 0.6584050393182099,
+      "grad_norm": 3.1254472732543945,
+      "learning_rate": 2.6921642897058776e-05,
+      "loss": 2.0924,
+      "step": 1999
+    },
+    {
+      "epoch": 0.6587344065214706,
+      "grad_norm": 2.893326997756958,
+      "learning_rate": 2.6874989178013345e-05,
+      "loss": 1.7982,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6590637737247313,
+      "grad_norm": 1.9270685911178589,
+      "learning_rate": 2.6828361056732653e-05,
+      "loss": 2.466,
+      "step": 2001
+    },
+    {
+      "epoch": 0.6593931409279921,
+      "grad_norm": 1.8927141427993774,
+      "learning_rate": 2.678175858483075e-05,
+      "loss": 2.4379,
+      "step": 2002
+    },
+    {
+      "epoch": 0.6597225081312529,
+      "grad_norm": 2.279390811920166,
+      "learning_rate": 2.6735181813893318e-05,
+      "loss": 2.3186,
+      "step": 2003
+    },
+    {
+      "epoch": 0.6600518753345136,
+      "grad_norm": 2.184950828552246,
+      "learning_rate": 2.6688630795477554e-05,
+      "loss": 2.1413,
+      "step": 2004
+    },
+    {
+      "epoch": 0.6603812425377743,
+      "grad_norm": 2.0760080814361572,
+      "learning_rate": 2.664210558111221e-05,
+      "loss": 2.1418,
+      "step": 2005
+    },
+    {
+      "epoch": 0.660710609741035,
+      "grad_norm": 2.2783913612365723,
+      "learning_rate": 2.6595606222297376e-05,
+      "loss": 2.1825,
+      "step": 2006
+    },
+    {
+      "epoch": 0.6610399769442957,
+      "grad_norm": 2.2643139362335205,
+      "learning_rate": 2.6549132770504615e-05,
+      "loss": 2.1718,
+      "step": 2007
+    },
+    {
+      "epoch": 0.6613693441475565,
+      "grad_norm": 2.3673267364501953,
+      "learning_rate": 2.6502685277176765e-05,
+      "loss": 2.2925,
+      "step": 2008
+    },
+    {
+      "epoch": 0.6616987113508173,
+      "grad_norm": 2.5755388736724854,
+      "learning_rate": 2.6456263793727953e-05,
+      "loss": 2.3991,
+      "step": 2009
+    },
+    {
+      "epoch": 0.662028078554078,
+      "grad_norm": 3.254749298095703,
+      "learning_rate": 2.6409868371543506e-05,
+      "loss": 2.3747,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6623574457573387,
+      "grad_norm": 2.4208810329437256,
+      "learning_rate": 2.636349906197991e-05,
+      "loss": 2.168,
+      "step": 2011
+    },
+    {
+      "epoch": 0.6626868129605995,
+      "grad_norm": 2.396467447280884,
+      "learning_rate": 2.631715591636471e-05,
+      "loss": 2.1355,
+      "step": 2012
+    },
+    {
+      "epoch": 0.6630161801638602,
+      "grad_norm": 2.6166560649871826,
+      "learning_rate": 2.6270838985996525e-05,
+      "loss": 2.1752,
+      "step": 2013
+    },
+    {
+      "epoch": 0.6633455473671209,
+      "grad_norm": 2.6235897541046143,
+      "learning_rate": 2.6224548322144964e-05,
+      "loss": 2.3333,
+      "step": 2014
+    },
+    {
+      "epoch": 0.6636749145703816,
+      "grad_norm": 2.3000566959381104,
+      "learning_rate": 2.6178283976050532e-05,
+      "loss": 2.1201,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6640042817736423,
+      "grad_norm": 2.6367340087890625,
+      "learning_rate": 2.6132045998924616e-05,
+      "loss": 2.2484,
+      "step": 2016
+    },
+    {
+      "epoch": 0.6643336489769032,
+      "grad_norm": 2.8426151275634766,
+      "learning_rate": 2.6085834441949418e-05,
+      "loss": 2.3913,
+      "step": 2017
+    },
+    {
+      "epoch": 0.6646630161801639,
+      "grad_norm": 3.0466115474700928,
+      "learning_rate": 2.6039649356277885e-05,
+      "loss": 2.3497,
+      "step": 2018
+    },
+    {
+      "epoch": 0.6649923833834246,
+      "grad_norm": 2.5447306632995605,
+      "learning_rate": 2.599349079303367e-05,
+      "loss": 2.0447,
+      "step": 2019
+    },
+    {
+      "epoch": 0.6653217505866853,
+      "grad_norm": 2.6460211277008057,
+      "learning_rate": 2.594735880331106e-05,
+      "loss": 1.9458,
+      "step": 2020
+    },
+    {
+      "epoch": 0.665651117789946,
+      "grad_norm": 2.6749167442321777,
+      "learning_rate": 2.5901253438174938e-05,
+      "loss": 1.8064,
+      "step": 2021
+    },
+    {
+      "epoch": 0.6659804849932068,
+      "grad_norm": 2.8169589042663574,
+      "learning_rate": 2.5855174748660704e-05,
+      "loss": 2.2488,
+      "step": 2022
+    },
+    {
+      "epoch": 0.6663098521964675,
+      "grad_norm": 3.0470190048217773,
+      "learning_rate": 2.5809122785774254e-05,
+      "loss": 1.9864,
+      "step": 2023
+    },
+    {
+      "epoch": 0.6666392193997283,
+      "grad_norm": 2.856595277786255,
+      "learning_rate": 2.5763097600491847e-05,
+      "loss": 1.8953,
+      "step": 2024
+    },
+    {
+      "epoch": 0.666968586602989,
+      "grad_norm": 3.369244337081909,
+      "learning_rate": 2.5717099243760147e-05,
+      "loss": 1.5753,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6672979538062498,
+      "grad_norm": 1.7922537326812744,
+      "learning_rate": 2.5671127766496105e-05,
+      "loss": 2.545,
+      "step": 2026
+    },
+    {
+      "epoch": 0.6676273210095105,
+      "grad_norm": 2.0254573822021484,
+      "learning_rate": 2.5625183219586935e-05,
+      "loss": 2.2592,
+      "step": 2027
+    },
+    {
+      "epoch": 0.6679566882127712,
+      "grad_norm": 2.4665722846984863,
+      "learning_rate": 2.5579265653890016e-05,
+      "loss": 2.4563,
+      "step": 2028
+    },
+    {
+      "epoch": 0.6682860554160319,
+      "grad_norm": 2.5361671447753906,
+      "learning_rate": 2.5533375120232885e-05,
+      "loss": 2.3127,
+      "step": 2029
+    },
+    {
+      "epoch": 0.6686154226192926,
+      "grad_norm": 2.1621336936950684,
+      "learning_rate": 2.5487511669413143e-05,
+      "loss": 2.2734,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6689447898225535,
+      "grad_norm": 2.921595811843872,
+      "learning_rate": 2.5441675352198392e-05,
+      "loss": 2.1833,
+      "step": 2031
+    },
+    {
+      "epoch": 0.6692741570258142,
+      "grad_norm": 1.9871540069580078,
+      "learning_rate": 2.5395866219326224e-05,
+      "loss": 2.2722,
+      "step": 2032
+    },
+    {
+      "epoch": 0.6696035242290749,
+      "grad_norm": 2.5396902561187744,
+      "learning_rate": 2.5350084321504148e-05,
+      "loss": 2.3714,
+      "step": 2033
+    },
+    {
+      "epoch": 0.6699328914323356,
+      "grad_norm": 2.6981403827667236,
+      "learning_rate": 2.5304329709409508e-05,
+      "loss": 2.5758,
+      "step": 2034
+    },
+    {
+      "epoch": 0.6702622586355964,
+      "grad_norm": 2.476912498474121,
+      "learning_rate": 2.525860243368945e-05,
+      "loss": 2.1581,
+      "step": 2035
+    },
+    {
+      "epoch": 0.6705916258388571,
+      "grad_norm": 2.6519880294799805,
+      "learning_rate": 2.5212902544960882e-05,
+      "loss": 2.0924,
+      "step": 2036
+    },
+    {
+      "epoch": 0.6709209930421178,
+      "grad_norm": 2.709019660949707,
+      "learning_rate": 2.516723009381033e-05,
+      "loss": 2.3023,
+      "step": 2037
+    },
+    {
+      "epoch": 0.6712503602453785,
+      "grad_norm": 2.7613868713378906,
+      "learning_rate": 2.512158513079402e-05,
+      "loss": 2.2279,
+      "step": 2038
+    },
+    {
+      "epoch": 0.6715797274486393,
+      "grad_norm": 2.635972023010254,
+      "learning_rate": 2.507596770643772e-05,
+      "loss": 2.2506,
+      "step": 2039
+    },
+    {
+      "epoch": 0.6719090946519001,
+      "grad_norm": 2.5571701526641846,
+      "learning_rate": 2.5030377871236714e-05,
+      "loss": 1.9826,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6722384618551608,
+      "grad_norm": 2.326291799545288,
+      "learning_rate": 2.4984815675655766e-05,
+      "loss": 2.1142,
+      "step": 2041
+    },
+    {
+      "epoch": 0.6725678290584215,
+      "grad_norm": 2.433256149291992,
+      "learning_rate": 2.4939281170129015e-05,
+      "loss": 1.8394,
+      "step": 2042
+    },
+    {
+      "epoch": 0.6728971962616822,
+      "grad_norm": 2.478874921798706,
+      "learning_rate": 2.4893774405059993e-05,
+      "loss": 2.0067,
+      "step": 2043
+    },
+    {
+      "epoch": 0.673226563464943,
+      "grad_norm": 2.6590943336486816,
+      "learning_rate": 2.4848295430821455e-05,
+      "loss": 1.8746,
+      "step": 2044
+    },
+    {
+      "epoch": 0.6735559306682037,
+      "grad_norm": 2.759122610092163,
+      "learning_rate": 2.4802844297755455e-05,
+      "loss": 2.5029,
+      "step": 2045
+    },
+    {
+      "epoch": 0.6738852978714645,
+      "grad_norm": 2.439481258392334,
+      "learning_rate": 2.4757421056173184e-05,
+      "loss": 1.7643,
+      "step": 2046
+    },
+    {
+      "epoch": 0.6742146650747252,
+      "grad_norm": 2.6136834621429443,
+      "learning_rate": 2.4712025756355033e-05,
+      "loss": 2.0499,
+      "step": 2047
+    },
+    {
+      "epoch": 0.6745440322779859,
+      "grad_norm": 3.705885648727417,
+      "learning_rate": 2.466665844855041e-05,
+      "loss": 2.6908,
+      "step": 2048
+    },
+    {
+      "epoch": 0.6748733994812467,
+      "grad_norm": 3.136033773422241,
+      "learning_rate": 2.4621319182977697e-05,
+      "loss": 1.8605,
+      "step": 2049
+    },
+    {
+      "epoch": 0.6752027666845074,
+      "grad_norm": 3.363560199737549,
+      "learning_rate": 2.457600800982431e-05,
+      "loss": 1.5852,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6755321338877681,
+      "grad_norm": 2.1761481761932373,
+      "learning_rate": 2.4530724979246535e-05,
+      "loss": 2.4601,
+      "step": 2051
+    },
+    {
+      "epoch": 0.6758615010910288,
+      "grad_norm": 2.2820851802825928,
+      "learning_rate": 2.44854701413695e-05,
+      "loss": 2.2097,
+      "step": 2052
+    },
+    {
+      "epoch": 0.6761908682942896,
+      "grad_norm": 2.133443593978882,
+      "learning_rate": 2.444024354628715e-05,
+      "loss": 2.2759,
+      "step": 2053
+    },
+    {
+      "epoch": 0.6765202354975504,
+      "grad_norm": 2.1510062217712402,
+      "learning_rate": 2.4395045244062172e-05,
+      "loss": 2.1092,
+      "step": 2054
+    },
+    {
+      "epoch": 0.6768496027008111,
+      "grad_norm": 2.543733596801758,
+      "learning_rate": 2.4349875284725863e-05,
+      "loss": 2.52,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6771789699040718,
+      "grad_norm": 2.35121488571167,
+      "learning_rate": 2.430473371827824e-05,
+      "loss": 2.5922,
+      "step": 2056
+    },
+    {
+      "epoch": 0.6775083371073325,
+      "grad_norm": 2.656498432159424,
+      "learning_rate": 2.425962059468783e-05,
+      "loss": 2.6134,
+      "step": 2057
+    },
+    {
+      "epoch": 0.6778377043105933,
+      "grad_norm": 2.513655662536621,
+      "learning_rate": 2.421453596389171e-05,
+      "loss": 2.4393,
+      "step": 2058
+    },
+    {
+      "epoch": 0.678167071513854,
+      "grad_norm": 2.9910781383514404,
+      "learning_rate": 2.4169479875795396e-05,
+      "loss": 2.5802,
+      "step": 2059
+    },
+    {
+      "epoch": 0.6784964387171147,
+      "grad_norm": 2.754699230194092,
+      "learning_rate": 2.4124452380272817e-05,
+      "loss": 2.0555,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6788258059203754,
+      "grad_norm": 2.328061819076538,
+      "learning_rate": 2.4079453527166273e-05,
+      "loss": 2.4093,
+      "step": 2061
+    },
+    {
+      "epoch": 0.6791551731236363,
+      "grad_norm": 2.4432404041290283,
+      "learning_rate": 2.4034483366286305e-05,
+      "loss": 2.0952,
+      "step": 2062
+    },
+    {
+      "epoch": 0.679484540326897,
+      "grad_norm": 2.6212449073791504,
+      "learning_rate": 2.3989541947411735e-05,
+      "loss": 1.9832,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6798139075301577,
+      "grad_norm": 2.5926594734191895,
+      "learning_rate": 2.3944629320289568e-05,
+      "loss": 1.9186,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6801432747334184,
+      "grad_norm": 2.6798171997070312,
+      "learning_rate": 2.3899745534634925e-05,
+      "loss": 1.7767,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6804726419366791,
+      "grad_norm": 2.3956661224365234,
+      "learning_rate": 2.3854890640131018e-05,
+      "loss": 2.2176,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6808020091399399,
+      "grad_norm": 2.882429838180542,
+      "learning_rate": 2.3810064686429062e-05,
+      "loss": 2.1865,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6811313763432006,
+      "grad_norm": 2.6350555419921875,
+      "learning_rate": 2.3765267723148267e-05,
+      "loss": 2.0534,
+      "step": 2068
+    },
+    {
+      "epoch": 0.6814607435464614,
+      "grad_norm": 3.9625144004821777,
+      "learning_rate": 2.3720499799875677e-05,
+      "loss": 2.4213,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6817901107497221,
+      "grad_norm": 2.6471993923187256,
+      "learning_rate": 2.3675760966166276e-05,
+      "loss": 1.8502,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6821194779529828,
+      "grad_norm": 2.543469190597534,
+      "learning_rate": 2.3631051271542816e-05,
+      "loss": 1.7004,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6824488451562436,
+      "grad_norm": 3.3397600650787354,
+      "learning_rate": 2.358637076549578e-05,
+      "loss": 2.3005,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6827782123595043,
+      "grad_norm": 3.1723084449768066,
+      "learning_rate": 2.3541719497483362e-05,
+      "loss": 2.3147,
+      "step": 2073
+    },
+    {
+      "epoch": 0.683107579562765,
+      "grad_norm": 2.94470477104187,
+      "learning_rate": 2.3497097516931398e-05,
+      "loss": 2.1635,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6834369467660257,
+      "grad_norm": 3.1503822803497314,
+      "learning_rate": 2.3452504873233262e-05,
+      "loss": 1.9506,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6837663139692866,
+      "grad_norm": 2.039581298828125,
+      "learning_rate": 2.3407941615749888e-05,
+      "loss": 2.0947,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6840956811725473,
+      "grad_norm": 2.2650582790374756,
+      "learning_rate": 2.3363407793809666e-05,
+      "loss": 2.3766,
+      "step": 2077
+    },
+    {
+      "epoch": 0.684425048375808,
+      "grad_norm": 2.530653238296509,
+      "learning_rate": 2.3318903456708445e-05,
+      "loss": 2.3375,
+      "step": 2078
+    },
+    {
+      "epoch": 0.6847544155790687,
+      "grad_norm": 2.3660364151000977,
+      "learning_rate": 2.3274428653709412e-05,
+      "loss": 2.625,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6850837827823294,
+      "grad_norm": 2.313480854034424,
+      "learning_rate": 2.3229983434043006e-05,
+      "loss": 2.233,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6854131499855902,
+      "grad_norm": 2.3391940593719482,
+      "learning_rate": 2.3185567846906997e-05,
+      "loss": 2.1382,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6857425171888509,
+      "grad_norm": 2.3489909172058105,
+      "learning_rate": 2.3141181941466312e-05,
+      "loss": 1.8627,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6860718843921116,
+      "grad_norm": 2.5285491943359375,
+      "learning_rate": 2.3096825766853043e-05,
+      "loss": 2.1704,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6864012515953724,
+      "grad_norm": 2.5529744625091553,
+      "learning_rate": 2.3052499372166366e-05,
+      "loss": 2.0689,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6867306187986332,
+      "grad_norm": 2.355783462524414,
+      "learning_rate": 2.300820280647248e-05,
+      "loss": 2.1035,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6870599860018939,
+      "grad_norm": 2.845242738723755,
+      "learning_rate": 2.29639361188046e-05,
+      "loss": 2.3495,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6873893532051546,
+      "grad_norm": 2.588106632232666,
+      "learning_rate": 2.2919699358162817e-05,
+      "loss": 2.4312,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6877187204084153,
+      "grad_norm": 2.835247755050659,
+      "learning_rate": 2.2875492573514123e-05,
+      "loss": 2.4685,
+      "step": 2088
+    },
+    {
+      "epoch": 0.688048087611676,
+      "grad_norm": 2.293931722640991,
+      "learning_rate": 2.2831315813792336e-05,
+      "loss": 1.9804,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6883774548149368,
+      "grad_norm": 2.318871259689331,
+      "learning_rate": 2.2787169127898027e-05,
+      "loss": 1.7806,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6887068220181976,
+      "grad_norm": 2.732576370239258,
+      "learning_rate": 2.2743052564698487e-05,
+      "loss": 2.298,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6890361892214583,
+      "grad_norm": 2.569328784942627,
+      "learning_rate": 2.2698966173027663e-05,
+      "loss": 2.3241,
+      "step": 2092
+    },
+    {
+      "epoch": 0.689365556424719,
+      "grad_norm": 2.7927603721618652,
+      "learning_rate": 2.2654910001686076e-05,
+      "loss": 2.0825,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6896949236279798,
+      "grad_norm": 2.9026594161987305,
+      "learning_rate": 2.261088409944082e-05,
+      "loss": 2.0577,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6900242908312405,
+      "grad_norm": 3.1268606185913086,
+      "learning_rate": 2.2566888515025498e-05,
+      "loss": 2.2409,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6903536580345012,
+      "grad_norm": 2.6002843379974365,
+      "learning_rate": 2.252292329714012e-05,
+      "loss": 1.8221,
+      "step": 2096
+    },
+    {
+      "epoch": 0.6906830252377619,
+      "grad_norm": 3.102250337600708,
+      "learning_rate": 2.2478988494451102e-05,
+      "loss": 1.8374,
+      "step": 2097
+    },
+    {
+      "epoch": 0.6910123924410226,
+      "grad_norm": 2.739004611968994,
+      "learning_rate": 2.2435084155591195e-05,
+      "loss": 1.8381,
+      "step": 2098
+    },
+    {
+      "epoch": 0.6913417596442835,
+      "grad_norm": 3.433546781539917,
+      "learning_rate": 2.2391210329159433e-05,
+      "loss": 2.4603,
+      "step": 2099
+    },
+    {
+      "epoch": 0.6916711268475442,
+      "grad_norm": 3.0737321376800537,
+      "learning_rate": 2.234736706372103e-05,
+      "loss": 1.8963,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6920004940508049,
+      "grad_norm": 1.8924542665481567,
+      "learning_rate": 2.2303554407807426e-05,
+      "loss": 2.3809,
+      "step": 2101
+    },
+    {
+      "epoch": 0.6923298612540656,
+      "grad_norm": 2.463632583618164,
+      "learning_rate": 2.2259772409916153e-05,
+      "loss": 2.5391,
+      "step": 2102
+    },
+    {
+      "epoch": 0.6926592284573264,
+      "grad_norm": 2.449293375015259,
+      "learning_rate": 2.2216021118510815e-05,
+      "loss": 2.3246,
+      "step": 2103
+    },
+    {
+      "epoch": 0.6929885956605871,
+      "grad_norm": 2.3340203762054443,
+      "learning_rate": 2.2172300582021022e-05,
+      "loss": 2.3163,
+      "step": 2104
+    },
+    {
+      "epoch": 0.6933179628638478,
+      "grad_norm": 2.3864715099334717,
+      "learning_rate": 2.2128610848842336e-05,
+      "loss": 2.5004,
+      "step": 2105
+    },
+    {
+      "epoch": 0.6936473300671085,
+      "grad_norm": 2.2127561569213867,
+      "learning_rate": 2.208495196733625e-05,
+      "loss": 2.1619,
+      "step": 2106
+    },
+    {
+      "epoch": 0.6939766972703693,
+      "grad_norm": 2.7245256900787354,
+      "learning_rate": 2.2041323985830027e-05,
+      "loss": 2.3766,
+      "step": 2107
+    },
+    {
+      "epoch": 0.6943060644736301,
+      "grad_norm": 2.3172929286956787,
+      "learning_rate": 2.1997726952616836e-05,
+      "loss": 2.2989,
+      "step": 2108
+    },
+    {
+      "epoch": 0.6946354316768908,
+      "grad_norm": 2.4857938289642334,
+      "learning_rate": 2.1954160915955525e-05,
+      "loss": 2.3859,
+      "step": 2109
+    },
+    {
+      "epoch": 0.6949647988801515,
+      "grad_norm": 2.176250457763672,
+      "learning_rate": 2.1910625924070623e-05,
+      "loss": 2.0642,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6952941660834122,
+      "grad_norm": 2.457257032394409,
+      "learning_rate": 2.186712202515234e-05,
+      "loss": 2.092,
+      "step": 2111
+    },
+    {
+      "epoch": 0.695623533286673,
+      "grad_norm": 3.2521281242370605,
+      "learning_rate": 2.1823649267356412e-05,
+      "loss": 2.2814,
+      "step": 2112
+    },
+    {
+      "epoch": 0.6959529004899337,
+      "grad_norm": 2.5254688262939453,
+      "learning_rate": 2.1780207698804134e-05,
+      "loss": 2.0433,
+      "step": 2113
+    },
+    {
+      "epoch": 0.6962822676931945,
+      "grad_norm": 2.5245742797851562,
+      "learning_rate": 2.1736797367582284e-05,
+      "loss": 2.1671,
+      "step": 2114
+    },
+    {
+      "epoch": 0.6966116348964552,
+      "grad_norm": 2.781444787979126,
+      "learning_rate": 2.169341832174306e-05,
+      "loss": 2.3001,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6969410020997159,
+      "grad_norm": 2.258967876434326,
+      "learning_rate": 2.1650070609304002e-05,
+      "loss": 1.9225,
+      "step": 2116
+    },
+    {
+      "epoch": 0.6972703693029767,
+      "grad_norm": 2.339409112930298,
+      "learning_rate": 2.1606754278248025e-05,
+      "loss": 2.0533,
+      "step": 2117
+    },
+    {
+      "epoch": 0.6975997365062374,
+      "grad_norm": 2.8763694763183594,
+      "learning_rate": 2.1563469376523228e-05,
+      "loss": 1.9008,
+      "step": 2118
+    },
+    {
+      "epoch": 0.6979291037094981,
+      "grad_norm": 2.7212729454040527,
+      "learning_rate": 2.152021595204297e-05,
+      "loss": 2.1429,
+      "step": 2119
+    },
+    {
+      "epoch": 0.6982584709127588,
+      "grad_norm": 2.8974947929382324,
+      "learning_rate": 2.1476994052685766e-05,
+      "loss": 1.9576,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6985878381160195,
+      "grad_norm": 2.508470058441162,
+      "learning_rate": 2.1433803726295227e-05,
+      "loss": 1.931,
+      "step": 2121
+    },
+    {
+      "epoch": 0.6989172053192804,
+      "grad_norm": 2.526657819747925,
+      "learning_rate": 2.1390645020680006e-05,
+      "loss": 1.7588,
+      "step": 2122
+    },
+    {
+      "epoch": 0.6992465725225411,
+      "grad_norm": 2.8346455097198486,
+      "learning_rate": 2.1347517983613773e-05,
+      "loss": 2.1006,
+      "step": 2123
+    },
+    {
+      "epoch": 0.6995759397258018,
+      "grad_norm": 3.421928882598877,
+      "learning_rate": 2.1304422662835146e-05,
+      "loss": 2.4451,
+      "step": 2124
+    },
+    {
+      "epoch": 0.6999053069290625,
+      "grad_norm": 2.873753547668457,
+      "learning_rate": 2.126135910604758e-05,
+      "loss": 1.7832,
+      "step": 2125
+    },
+    {
+      "epoch": 0.7002346741323233,
+      "grad_norm": 2.716782331466675,
+      "learning_rate": 2.1218327360919438e-05,
+      "loss": 2.4388,
+      "step": 2126
+    },
+    {
+      "epoch": 0.700564041335584,
+      "grad_norm": 2.3297483921051025,
+      "learning_rate": 2.1175327475083844e-05,
+      "loss": 2.5349,
+      "step": 2127
+    },
+    {
+      "epoch": 0.7008934085388447,
+      "grad_norm": 2.592759847640991,
+      "learning_rate": 2.1132359496138648e-05,
+      "loss": 2.3805,
+      "step": 2128
+    },
+    {
+      "epoch": 0.7012227757421055,
+      "grad_norm": 2.579590082168579,
+      "learning_rate": 2.108942347164639e-05,
+      "loss": 2.4302,
+      "step": 2129
+    },
+    {
+      "epoch": 0.7015521429453662,
+      "grad_norm": 2.581580877304077,
+      "learning_rate": 2.104651944913426e-05,
+      "loss": 2.2602,
+      "step": 2130
+    },
+    {
+      "epoch": 0.701881510148627,
+      "grad_norm": 2.675879955291748,
+      "learning_rate": 2.1003647476093962e-05,
+      "loss": 2.706,
+      "step": 2131
+    },
+    {
+      "epoch": 0.7022108773518877,
+      "grad_norm": 2.9450440406799316,
+      "learning_rate": 2.0960807599981785e-05,
+      "loss": 2.5483,
+      "step": 2132
+    },
+    {
+      "epoch": 0.7025402445551484,
+      "grad_norm": 2.4769771099090576,
+      "learning_rate": 2.0917999868218457e-05,
+      "loss": 2.2321,
+      "step": 2133
+    },
+    {
+      "epoch": 0.7028696117584091,
+      "grad_norm": 2.61974835395813,
+      "learning_rate": 2.087522432818914e-05,
+      "loss": 2.3689,
+      "step": 2134
+    },
+    {
+      "epoch": 0.7031989789616699,
+      "grad_norm": 2.3466622829437256,
+      "learning_rate": 2.0832481027243357e-05,
+      "loss": 2.2311,
+      "step": 2135
+    },
+    {
+      "epoch": 0.7035283461649307,
+      "grad_norm": 2.35425066947937,
+      "learning_rate": 2.0789770012694937e-05,
+      "loss": 2.0013,
+      "step": 2136
+    },
+    {
+      "epoch": 0.7038577133681914,
+      "grad_norm": 2.2261617183685303,
+      "learning_rate": 2.0747091331822005e-05,
+      "loss": 1.9601,
+      "step": 2137
+    },
+    {
+      "epoch": 0.7041870805714521,
+      "grad_norm": 2.35845685005188,
+      "learning_rate": 2.0704445031866803e-05,
+      "loss": 1.9554,
+      "step": 2138
+    },
+    {
+      "epoch": 0.7045164477747128,
+      "grad_norm": 2.579216480255127,
+      "learning_rate": 2.066183116003586e-05,
+      "loss": 2.0605,
+      "step": 2139
+    },
+    {
+      "epoch": 0.7048458149779736,
+      "grad_norm": 2.4616470336914062,
+      "learning_rate": 2.0619249763499708e-05,
+      "loss": 2.1647,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7051751821812343,
+      "grad_norm": 2.7063252925872803,
+      "learning_rate": 2.057670088939298e-05,
+      "loss": 2.1963,
+      "step": 2141
+    },
+    {
+      "epoch": 0.705504549384495,
+      "grad_norm": 2.8242859840393066,
+      "learning_rate": 2.053418458481431e-05,
+      "loss": 2.2085,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7058339165877557,
+      "grad_norm": 3.1535305976867676,
+      "learning_rate": 2.0491700896826222e-05,
+      "loss": 2.0284,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7061632837910166,
+      "grad_norm": 2.366270065307617,
+      "learning_rate": 2.04492498724552e-05,
+      "loss": 2.1194,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7064926509942773,
+      "grad_norm": 2.758836269378662,
+      "learning_rate": 2.0406831558691552e-05,
+      "loss": 2.2551,
+      "step": 2145
+    },
+    {
+      "epoch": 0.706822018197538,
+      "grad_norm": 2.47495698928833,
+      "learning_rate": 2.0364446002489372e-05,
+      "loss": 1.9995,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7071513854007987,
+      "grad_norm": 2.167638063430786,
+      "learning_rate": 2.03220932507665e-05,
+      "loss": 1.8277,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7074807526040594,
+      "grad_norm": 2.6053552627563477,
+      "learning_rate": 2.0279773350404464e-05,
+      "loss": 1.7099,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7078101198073202,
+      "grad_norm": 2.8557825088500977,
+      "learning_rate": 2.0237486348248437e-05,
+      "loss": 1.6977,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7081394870105809,
+      "grad_norm": 3.3039448261260986,
+      "learning_rate": 2.0195232291107125e-05,
+      "loss": 1.5792,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7084688542138416,
+      "grad_norm": 1.8703773021697998,
+      "learning_rate": 2.0153011225752832e-05,
+      "loss": 2.436,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7087982214171024,
+      "grad_norm": 1.92439866065979,
+      "learning_rate": 2.0110823198921314e-05,
+      "loss": 2.1712,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7091275886203632,
+      "grad_norm": 2.654240846633911,
+      "learning_rate": 2.0068668257311752e-05,
+      "loss": 2.7122,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7094569558236239,
+      "grad_norm": 2.478987455368042,
+      "learning_rate": 2.0026546447586715e-05,
+      "loss": 2.1828,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7097863230268846,
+      "grad_norm": 2.456247091293335,
+      "learning_rate": 1.9984457816372103e-05,
+      "loss": 2.4922,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7101156902301453,
+      "grad_norm": 2.527040958404541,
+      "learning_rate": 1.994240241025705e-05,
+      "loss": 2.265,
+      "step": 2156
+    },
+    {
+      "epoch": 0.710445057433406,
+      "grad_norm": 2.492377519607544,
+      "learning_rate": 1.990038027579395e-05,
+      "loss": 2.2534,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7107744246366668,
+      "grad_norm": 3.028960943222046,
+      "learning_rate": 1.9858391459498367e-05,
+      "loss": 2.4119,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7111037918399276,
+      "grad_norm": 2.4781980514526367,
+      "learning_rate": 1.9816436007848964e-05,
+      "loss": 2.2563,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7114331590431883,
+      "grad_norm": 2.4208078384399414,
+      "learning_rate": 1.9774513967287496e-05,
+      "loss": 2.1364,
+      "step": 2160
+    },
+    {
+      "epoch": 0.711762526246449,
+      "grad_norm": 2.3853366374969482,
+      "learning_rate": 1.9732625384218705e-05,
+      "loss": 2.0286,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7120918934497097,
+      "grad_norm": 2.075531244277954,
+      "learning_rate": 1.9690770305010346e-05,
+      "loss": 2.1623,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7124212606529705,
+      "grad_norm": 2.8861477375030518,
+      "learning_rate": 1.9648948775993014e-05,
+      "loss": 2.0822,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7127506278562312,
+      "grad_norm": 2.6686360836029053,
+      "learning_rate": 1.9607160843460225e-05,
+      "loss": 2.3996,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7130799950594919,
+      "grad_norm": 2.642855644226074,
+      "learning_rate": 1.956540655366829e-05,
+      "loss": 2.1804,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7134093622627526,
+      "grad_norm": 2.506636142730713,
+      "learning_rate": 1.952368595283628e-05,
+      "loss": 1.6528,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7137387294660135,
+      "grad_norm": 2.5572757720947266,
+      "learning_rate": 1.9481999087145973e-05,
+      "loss": 1.9408,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7140680966692742,
+      "grad_norm": 2.670609474182129,
+      "learning_rate": 1.9440346002741798e-05,
+      "loss": 2.2826,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7143974638725349,
+      "grad_norm": 2.513737916946411,
+      "learning_rate": 1.939872674573081e-05,
+      "loss": 2.0831,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7147268310757956,
+      "grad_norm": 2.594437837600708,
+      "learning_rate": 1.93571413621826e-05,
+      "loss": 2.0226,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7150561982790563,
+      "grad_norm": 3.0486814975738525,
+      "learning_rate": 1.9315589898129266e-05,
+      "loss": 1.8639,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7153855654823171,
+      "grad_norm": 2.8681252002716064,
+      "learning_rate": 1.9274072399565373e-05,
+      "loss": 1.9047,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7157149326855778,
+      "grad_norm": 2.4376771450042725,
+      "learning_rate": 1.9232588912447875e-05,
+      "loss": 1.6213,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7160442998888386,
+      "grad_norm": 2.7798798084259033,
+      "learning_rate": 1.9191139482696097e-05,
+      "loss": 1.9778,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7163736670920993,
+      "grad_norm": 2.8711724281311035,
+      "learning_rate": 1.9149724156191618e-05,
+      "loss": 1.7021,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7167030342953601,
+      "grad_norm": 2.1148149967193604,
+      "learning_rate": 1.9108342978778317e-05,
+      "loss": 2.3465,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7170324014986208,
+      "grad_norm": 2.3499250411987305,
+      "learning_rate": 1.9066995996262248e-05,
+      "loss": 2.2567,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7173617687018815,
+      "grad_norm": 2.2933461666107178,
+      "learning_rate": 1.902568325441163e-05,
+      "loss": 2.1633,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7176911359051422,
+      "grad_norm": 2.383384943008423,
+      "learning_rate": 1.898440479895677e-05,
+      "loss": 2.3886,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7180205031084029,
+      "grad_norm": 2.2387309074401855,
+      "learning_rate": 1.894316067559003e-05,
+      "loss": 2.02,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7183498703116638,
+      "grad_norm": 2.4402246475219727,
+      "learning_rate": 1.890195092996573e-05,
+      "loss": 2.1891,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7186792375149245,
+      "grad_norm": 2.534137010574341,
+      "learning_rate": 1.886077560770019e-05,
+      "loss": 2.3046,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7190086047181852,
+      "grad_norm": 2.543569803237915,
+      "learning_rate": 1.88196347543716e-05,
+      "loss": 2.5529,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7193379719214459,
+      "grad_norm": 2.6904938220977783,
+      "learning_rate": 1.8778528415519998e-05,
+      "loss": 1.9117,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7196673391247067,
+      "grad_norm": 2.512737989425659,
+      "learning_rate": 1.873745663664722e-05,
+      "loss": 2.093,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7199967063279674,
+      "grad_norm": 2.5019516944885254,
+      "learning_rate": 1.869641946321684e-05,
+      "loss": 2.3069,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7203260735312281,
+      "grad_norm": 2.564549446105957,
+      "learning_rate": 1.8655416940654152e-05,
+      "loss": 1.94,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7206554407344888,
+      "grad_norm": 2.498377561569214,
+      "learning_rate": 1.8614449114346033e-05,
+      "loss": 2.317,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7209848079377495,
+      "grad_norm": 2.7334401607513428,
+      "learning_rate": 1.8573516029641015e-05,
+      "loss": 2.0478,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7213141751410104,
+      "grad_norm": 2.2970240116119385,
+      "learning_rate": 1.8532617731849144e-05,
+      "loss": 1.6288,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7216435423442711,
+      "grad_norm": 2.7691264152526855,
+      "learning_rate": 1.8491754266241973e-05,
+      "loss": 2.3471,
+      "step": 2191
+    },
+    {
+      "epoch": 0.7219729095475318,
+      "grad_norm": 2.845181941986084,
+      "learning_rate": 1.8450925678052495e-05,
+      "loss": 2.3704,
+      "step": 2192
+    },
+    {
+      "epoch": 0.7223022767507925,
+      "grad_norm": 2.658377170562744,
+      "learning_rate": 1.8410132012475094e-05,
+      "loss": 2.0011,
+      "step": 2193
+    },
+    {
+      "epoch": 0.7226316439540533,
+      "grad_norm": 2.7496695518493652,
+      "learning_rate": 1.8369373314665483e-05,
+      "loss": 1.9333,
+      "step": 2194
+    },
+    {
+      "epoch": 0.722961011157314,
+      "grad_norm": 3.0554986000061035,
+      "learning_rate": 1.8328649629740685e-05,
+      "loss": 2.1955,
+      "step": 2195
+    },
+    {
+      "epoch": 0.7232903783605747,
+      "grad_norm": 2.52851939201355,
+      "learning_rate": 1.8287961002778964e-05,
+      "loss": 2.1551,
+      "step": 2196
+    },
+    {
+      "epoch": 0.7236197455638355,
+      "grad_norm": 2.8639140129089355,
+      "learning_rate": 1.824730747881978e-05,
+      "loss": 2.0154,
+      "step": 2197
+    },
+    {
+      "epoch": 0.7239491127670962,
+      "grad_norm": 3.166149139404297,
+      "learning_rate": 1.8206689102863728e-05,
+      "loss": 2.148,
+      "step": 2198
+    },
+    {
+      "epoch": 0.724278479970357,
+      "grad_norm": 3.3647382259368896,
+      "learning_rate": 1.81661059198725e-05,
+      "loss": 2.2395,
+      "step": 2199
+    },
+    {
+      "epoch": 0.7246078471736177,
+      "grad_norm": 2.8727433681488037,
+      "learning_rate": 1.8125557974768837e-05,
+      "loss": 1.7405,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7249372143768784,
+      "grad_norm": 2.2409629821777344,
+      "learning_rate": 1.8085045312436465e-05,
+      "loss": 2.6778,
+      "step": 2201
+    },
+    {
+      "epoch": 0.7252665815801391,
+      "grad_norm": 1.9362051486968994,
+      "learning_rate": 1.804456797772006e-05,
+      "loss": 2.1103,
+      "step": 2202
+    },
+    {
+      "epoch": 0.7255959487834,
+      "grad_norm": 2.2799839973449707,
+      "learning_rate": 1.80041260154252e-05,
+      "loss": 2.3589,
+      "step": 2203
+    },
+    {
+      "epoch": 0.7259253159866607,
+      "grad_norm": 2.4302492141723633,
+      "learning_rate": 1.796371947031829e-05,
+      "loss": 2.2235,
+      "step": 2204
+    },
+    {
+      "epoch": 0.7262546831899214,
+      "grad_norm": 2.615129232406616,
+      "learning_rate": 1.7923348387126566e-05,
+      "loss": 2.3434,
+      "step": 2205
+    },
+    {
+      "epoch": 0.7265840503931821,
+      "grad_norm": 2.2985572814941406,
+      "learning_rate": 1.788301281053794e-05,
+      "loss": 2.2154,
+      "step": 2206
+    },
+    {
+      "epoch": 0.7269134175964428,
+      "grad_norm": 2.6856024265289307,
+      "learning_rate": 1.7842712785201094e-05,
+      "loss": 2.4275,
+      "step": 2207
+    },
+    {
+      "epoch": 0.7272427847997036,
+      "grad_norm": 2.337125778198242,
+      "learning_rate": 1.7802448355725322e-05,
+      "loss": 2.2931,
+      "step": 2208
+    },
+    {
+      "epoch": 0.7275721520029643,
+      "grad_norm": 2.6537628173828125,
+      "learning_rate": 1.7762219566680528e-05,
+      "loss": 2.2855,
+      "step": 2209
+    },
+    {
+      "epoch": 0.727901519206225,
+      "grad_norm": 3.049884796142578,
+      "learning_rate": 1.7722026462597153e-05,
+      "loss": 2.5902,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7282308864094857,
+      "grad_norm": 2.5096991062164307,
+      "learning_rate": 1.768186908796617e-05,
+      "loss": 2.1373,
+      "step": 2211
+    },
+    {
+      "epoch": 0.7285602536127465,
+      "grad_norm": 2.4747989177703857,
+      "learning_rate": 1.764174748723893e-05,
+      "loss": 2.626,
+      "step": 2212
+    },
+    {
+      "epoch": 0.7288896208160073,
+      "grad_norm": 2.4736328125,
+      "learning_rate": 1.7601661704827253e-05,
+      "loss": 2.303,
+      "step": 2213
+    },
+    {
+      "epoch": 0.729218988019268,
+      "grad_norm": 2.823103666305542,
+      "learning_rate": 1.7561611785103294e-05,
+      "loss": 2.1824,
+      "step": 2214
+    },
+    {
+      "epoch": 0.7295483552225287,
+      "grad_norm": 2.7458789348602295,
+      "learning_rate": 1.7521597772399496e-05,
+      "loss": 1.7698,
+      "step": 2215
+    },
+    {
+      "epoch": 0.7298777224257894,
+      "grad_norm": 2.636065721511841,
+      "learning_rate": 1.748161971100856e-05,
+      "loss": 2.2537,
+      "step": 2216
+    },
+    {
+      "epoch": 0.7302070896290502,
+      "grad_norm": 2.7386367321014404,
+      "learning_rate": 1.7441677645183413e-05,
+      "loss": 2.2158,
+      "step": 2217
+    },
+    {
+      "epoch": 0.7305364568323109,
+      "grad_norm": 2.3262336254119873,
+      "learning_rate": 1.740177161913712e-05,
+      "loss": 1.9395,
+      "step": 2218
+    },
+    {
+      "epoch": 0.7308658240355717,
+      "grad_norm": 2.5685794353485107,
+      "learning_rate": 1.736190167704283e-05,
+      "loss": 2.0057,
+      "step": 2219
+    },
+    {
+      "epoch": 0.7311951912388324,
+      "grad_norm": 2.8580093383789062,
+      "learning_rate": 1.732206786303378e-05,
+      "loss": 2.178,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7315245584420931,
+      "grad_norm": 2.475217580795288,
+      "learning_rate": 1.7282270221203213e-05,
+      "loss": 1.9048,
+      "step": 2221
+    },
+    {
+      "epoch": 0.7318539256453539,
+      "grad_norm": 2.6721620559692383,
+      "learning_rate": 1.7242508795604324e-05,
+      "loss": 1.9798,
+      "step": 2222
+    },
+    {
+      "epoch": 0.7321832928486146,
+      "grad_norm": 3.082043409347534,
+      "learning_rate": 1.720278363025022e-05,
+      "loss": 2.1289,
+      "step": 2223
+    },
+    {
+      "epoch": 0.7325126600518753,
+      "grad_norm": 3.0245370864868164,
+      "learning_rate": 1.716309476911388e-05,
+      "loss": 1.8127,
+      "step": 2224
+    },
+    {
+      "epoch": 0.732842027255136,
+      "grad_norm": 3.863074779510498,
+      "learning_rate": 1.7123442256128097e-05,
+      "loss": 1.9633,
+      "step": 2225
+    },
+    {
+      "epoch": 0.7331713944583969,
+      "grad_norm": 1.7904800176620483,
+      "learning_rate": 1.7083826135185393e-05,
+      "loss": 2.3501,
+      "step": 2226
+    },
+    {
+      "epoch": 0.7335007616616576,
+      "grad_norm": 2.111262083053589,
+      "learning_rate": 1.7044246450138053e-05,
+      "loss": 2.4573,
+      "step": 2227
+    },
+    {
+      "epoch": 0.7338301288649183,
+      "grad_norm": 2.706902503967285,
+      "learning_rate": 1.700470324479801e-05,
+      "loss": 2.0959,
+      "step": 2228
+    },
+    {
+      "epoch": 0.734159496068179,
+      "grad_norm": 2.2584450244903564,
+      "learning_rate": 1.6965196562936796e-05,
+      "loss": 2.2161,
+      "step": 2229
+    },
+    {
+      "epoch": 0.7344888632714397,
+      "grad_norm": 2.040581226348877,
+      "learning_rate": 1.6925726448285588e-05,
+      "loss": 2.0735,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7348182304747005,
+      "grad_norm": 2.670949697494507,
+      "learning_rate": 1.6886292944534994e-05,
+      "loss": 2.2781,
+      "step": 2231
+    },
+    {
+      "epoch": 0.7351475976779612,
+      "grad_norm": 2.4085488319396973,
+      "learning_rate": 1.6846896095335146e-05,
+      "loss": 2.1216,
+      "step": 2232
+    },
+    {
+      "epoch": 0.7354769648812219,
+      "grad_norm": 2.76834774017334,
+      "learning_rate": 1.680753594429559e-05,
+      "loss": 2.4634,
+      "step": 2233
+    },
+    {
+      "epoch": 0.7358063320844827,
+      "grad_norm": 2.398747444152832,
+      "learning_rate": 1.6768212534985257e-05,
+      "loss": 2.1542,
+      "step": 2234
+    },
+    {
+      "epoch": 0.7361356992877435,
+      "grad_norm": 2.2678730487823486,
+      "learning_rate": 1.672892591093239e-05,
+      "loss": 2.2039,
+      "step": 2235
+    },
+    {
+      "epoch": 0.7364650664910042,
+      "grad_norm": 2.4844462871551514,
+      "learning_rate": 1.6689676115624563e-05,
+      "loss": 2.0138,
+      "step": 2236
+    },
+    {
+      "epoch": 0.7367944336942649,
+      "grad_norm": 2.4144296646118164,
+      "learning_rate": 1.6650463192508496e-05,
+      "loss": 1.9737,
+      "step": 2237
+    },
+    {
+      "epoch": 0.7371238008975256,
+      "grad_norm": 3.1505680084228516,
+      "learning_rate": 1.6611287184990172e-05,
+      "loss": 2.4489,
+      "step": 2238
+    },
+    {
+      "epoch": 0.7374531681007863,
+      "grad_norm": 2.5178029537200928,
+      "learning_rate": 1.6572148136434678e-05,
+      "loss": 2.2807,
+      "step": 2239
+    },
+    {
+      "epoch": 0.7377825353040471,
+      "grad_norm": 2.765470266342163,
+      "learning_rate": 1.6533046090166195e-05,
+      "loss": 2.3461,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7381119025073078,
+      "grad_norm": 2.7750508785247803,
+      "learning_rate": 1.6493981089467943e-05,
+      "loss": 2.2532,
+      "step": 2241
+    },
+    {
+      "epoch": 0.7384412697105686,
+      "grad_norm": 2.7615838050842285,
+      "learning_rate": 1.645495317758214e-05,
+      "loss": 2.0369,
+      "step": 2242
+    },
+    {
+      "epoch": 0.7387706369138293,
+      "grad_norm": 2.5372447967529297,
+      "learning_rate": 1.641596239770996e-05,
+      "loss": 2.1105,
+      "step": 2243
+    },
+    {
+      "epoch": 0.7391000041170901,
+      "grad_norm": 2.304556369781494,
+      "learning_rate": 1.6377008793011433e-05,
+      "loss": 1.6536,
+      "step": 2244
+    },
+    {
+      "epoch": 0.7394293713203508,
+      "grad_norm": 2.7731292247772217,
+      "learning_rate": 1.633809240660548e-05,
+      "loss": 1.9395,
+      "step": 2245
+    },
+    {
+      "epoch": 0.7397587385236115,
+      "grad_norm": 2.9680540561676025,
+      "learning_rate": 1.6299213281569815e-05,
+      "loss": 1.7278,
+      "step": 2246
+    },
+    {
+      "epoch": 0.7400881057268722,
+      "grad_norm": 2.882687568664551,
+      "learning_rate": 1.626037146094089e-05,
+      "loss": 1.9297,
+      "step": 2247
+    },
+    {
+      "epoch": 0.7404174729301329,
+      "grad_norm": 3.0989298820495605,
+      "learning_rate": 1.6221566987713893e-05,
+      "loss": 2.0538,
+      "step": 2248
+    },
+    {
+      "epoch": 0.7407468401333938,
+      "grad_norm": 3.140981435775757,
+      "learning_rate": 1.618279990484266e-05,
+      "loss": 2.1415,
+      "step": 2249
+    },
+    {
+      "epoch": 0.7410762073366545,
+      "grad_norm": 3.206493616104126,
+      "learning_rate": 1.6144070255239597e-05,
+      "loss": 2.1208,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7414055745399152,
+      "grad_norm": 2.1097187995910645,
+      "learning_rate": 1.6105378081775735e-05,
+      "loss": 2.5872,
+      "step": 2251
+    },
+    {
+      "epoch": 0.7417349417431759,
+      "grad_norm": 2.210814952850342,
+      "learning_rate": 1.6066723427280582e-05,
+      "loss": 2.6126,
+      "step": 2252
+    },
+    {
+      "epoch": 0.7420643089464367,
+      "grad_norm": 2.039604902267456,
+      "learning_rate": 1.6028106334542144e-05,
+      "loss": 2.2881,
+      "step": 2253
+    },
+    {
+      "epoch": 0.7423936761496974,
+      "grad_norm": 2.2926268577575684,
+      "learning_rate": 1.5989526846306824e-05,
+      "loss": 2.4734,
+      "step": 2254
+    },
+    {
+      "epoch": 0.7427230433529581,
+      "grad_norm": 2.4341986179351807,
+      "learning_rate": 1.5950985005279413e-05,
+      "loss": 2.3901,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7430524105562188,
+      "grad_norm": 2.212266206741333,
+      "learning_rate": 1.5912480854123042e-05,
+      "loss": 2.5487,
+      "step": 2256
+    },
+    {
+      "epoch": 0.7433817777594796,
+      "grad_norm": 2.5183675289154053,
+      "learning_rate": 1.5874014435459068e-05,
+      "loss": 2.3042,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7437111449627404,
+      "grad_norm": 2.696521759033203,
+      "learning_rate": 1.5835585791867135e-05,
+      "loss": 2.2133,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7440405121660011,
+      "grad_norm": 2.3256208896636963,
+      "learning_rate": 1.579719496588506e-05,
+      "loss": 2.2106,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7443698793692618,
+      "grad_norm": 2.5763232707977295,
+      "learning_rate": 1.5758842000008772e-05,
+      "loss": 2.4074,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7446992465725225,
+      "grad_norm": 2.66152024269104,
+      "learning_rate": 1.572052693669237e-05,
+      "loss": 1.9538,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7450286137757832,
+      "grad_norm": 3.3669986724853516,
+      "learning_rate": 1.568224981834789e-05,
+      "loss": 2.4397,
+      "step": 2262
+    },
+    {
+      "epoch": 0.745357980979044,
+      "grad_norm": 2.784435510635376,
+      "learning_rate": 1.5644010687345433e-05,
+      "loss": 2.4693,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7456873481823048,
+      "grad_norm": 2.511077404022217,
+      "learning_rate": 1.5605809586013033e-05,
+      "loss": 2.2116,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7460167153855655,
+      "grad_norm": 2.888571262359619,
+      "learning_rate": 1.556764655663662e-05,
+      "loss": 2.5169,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7463460825888262,
+      "grad_norm": 2.7775685787200928,
+      "learning_rate": 1.552952164146001e-05,
+      "loss": 1.8714,
+      "step": 2266
+    },
+    {
+      "epoch": 0.746675449792087,
+      "grad_norm": 2.7570314407348633,
+      "learning_rate": 1.5491434882684796e-05,
+      "loss": 2.1423,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7470048169953477,
+      "grad_norm": 3.36289381980896,
+      "learning_rate": 1.545338632247037e-05,
+      "loss": 2.1092,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7473341841986084,
+      "grad_norm": 2.6311471462249756,
+      "learning_rate": 1.541537600293378e-05,
+      "loss": 1.8818,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 2.5641837120056152,
+      "learning_rate": 1.5377403966149806e-05,
+      "loss": 1.7303,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7479929186051298,
+      "grad_norm": 2.628680467605591,
+      "learning_rate": 1.533947025415083e-05,
+      "loss": 2.2235,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7483222858083907,
+      "grad_norm": 3.235398530960083,
+      "learning_rate": 1.53015749089268e-05,
+      "loss": 1.8414,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7486516530116514,
+      "grad_norm": 2.6925714015960693,
+      "learning_rate": 1.5263717972425222e-05,
+      "loss": 1.6467,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7489810202149121,
+      "grad_norm": 2.408013343811035,
+      "learning_rate": 1.5225899486551065e-05,
+      "loss": 1.8224,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7493103874181728,
+      "grad_norm": 3.5030288696289062,
+      "learning_rate": 1.5188119493166726e-05,
+      "loss": 2.0291,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7496397546214336,
+      "grad_norm": 2.0207650661468506,
+      "learning_rate": 1.515037803409201e-05,
+      "loss": 2.4186,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7499691218246943,
+      "grad_norm": 2.019650459289551,
+      "learning_rate": 1.5112675151104066e-05,
+      "loss": 2.1816,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7499691218246943,
+      "eval_loss": 2.19006085395813,
+      "eval_runtime": 765.4555,
+      "eval_samples_per_second": 3.34,
+      "eval_steps_per_second": 1.671,
+      "step": 2277
     }
   ],
   "logging_steps": 1,
@@ -10676,7 +15997,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.328291179772248e+18,
+  "total_flos": 3.492436769658372e+18,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null