diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30483 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.22686392180443912,
+  "eval_steps": 500,
+  "global_step": 870000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 5.215262570216991e-05,
+      "grad_norm": 196.05113220214844,
+      "learning_rate": 2.6076639242734397e-08,
+      "loss": 19.4606,
+      "step": 200
+    },
+    {
+      "epoch": 0.00010430525140433982,
+      "grad_norm": 184.3112335205078,
+      "learning_rate": 5.2153278485468794e-08,
+      "loss": 19.0516,
+      "step": 400
+    },
+    {
+      "epoch": 0.00015645787710650975,
+      "grad_norm": 126.78282928466797,
+      "learning_rate": 7.822991772820321e-08,
+      "loss": 17.3077,
+      "step": 600
+    },
+    {
+      "epoch": 0.00020861050280867964,
+      "grad_norm": 59.83221435546875,
+      "learning_rate": 1.0430655697093759e-07,
+      "loss": 15.0746,
+      "step": 800
+    },
+    {
+      "epoch": 0.0002607631285108496,
+      "grad_norm": 28.47006607055664,
+      "learning_rate": 1.30383196213672e-07,
+      "loss": 13.063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0003129157542130195,
+      "grad_norm": 16.229631423950195,
+      "learning_rate": 1.5645983545640642e-07,
+      "loss": 11.636,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0003650683799151894,
+      "grad_norm": 14.528124809265137,
+      "learning_rate": 1.825364746991408e-07,
+      "loss": 10.7786,
+      "step": 1400
+    },
+    {
+      "epoch": 0.0004172210056173593,
+      "grad_norm": 12.064294815063477,
+      "learning_rate": 2.0861311394187518e-07,
+      "loss": 10.2434,
+      "step": 1600
+    },
+    {
+      "epoch": 0.00046937363131952924,
+      "grad_norm": 11.837329864501953,
+      "learning_rate": 2.3468975318460956e-07,
+      "loss": 9.9074,
+      "step": 1800
+    },
+    {
+      "epoch": 0.0005215262570216992,
+      "grad_norm": 10.613241195678711,
+      "learning_rate": 2.60766392427344e-07,
+      "loss": 9.6339,
+      "step": 2000
+    },
+    {
+      "epoch": 0.000573678882723869,
+      "grad_norm": 10.466361045837402,
+      "learning_rate": 2.8684303167007837e-07,
+      "loss": 9.4193,
+      "step": 2200
+    },
+    {
+      "epoch": 0.000625831508426039,
+      "grad_norm": 10.19588565826416,
+      "learning_rate": 3.1291967091281283e-07,
+      "loss": 9.2786,
+      "step": 2400
+    },
+    {
+      "epoch": 0.0006779841341282089,
+      "grad_norm": 9.127360343933105,
+      "learning_rate": 3.389963101555472e-07,
+      "loss": 9.1531,
+      "step": 2600
+    },
+    {
+      "epoch": 0.0007301367598303788,
+      "grad_norm": 10.378647804260254,
+      "learning_rate": 3.650729493982816e-07,
+      "loss": 9.0132,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0007822893855325487,
+      "grad_norm": 10.13025951385498,
+      "learning_rate": 3.91149588641016e-07,
+      "loss": 8.876,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0008344420112347186,
+      "grad_norm": 8.717490196228027,
+      "learning_rate": 4.1722622788375035e-07,
+      "loss": 8.7792,
+      "step": 3200
+    },
+    {
+      "epoch": 0.0008865946369368885,
+      "grad_norm": 8.891645431518555,
+      "learning_rate": 4.4330286712648476e-07,
+      "loss": 8.675,
+      "step": 3400
+    },
+    {
+      "epoch": 0.0009387472626390585,
+      "grad_norm": 8.681458473205566,
+      "learning_rate": 4.693795063692191e-07,
+      "loss": 8.5239,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0009908998883412284,
+      "grad_norm": 9.971908569335938,
+      "learning_rate": 4.954561456119536e-07,
+      "loss": 8.4858,
+      "step": 3800
+    },
+    {
+      "epoch": 0.0010430525140433984,
+      "grad_norm": 9.44089412689209,
+      "learning_rate": 5.21532784854688e-07,
+      "loss": 8.3296,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0010952051397455681,
+      "grad_norm": 8.453426361083984,
+      "learning_rate": 5.476094240974224e-07,
+      "loss": 8.2809,
+      "step": 4200
+    },
+    {
+      "epoch": 0.001147357765447738,
+      "grad_norm": 8.532354354858398,
+      "learning_rate": 5.736860633401567e-07,
+      "loss": 8.1715,
+      "step": 4400
+    },
+    {
+      "epoch": 0.001199510391149908,
+      "grad_norm": 8.517669677734375,
+      "learning_rate": 5.997627025828912e-07,
+      "loss": 8.0996,
+      "step": 4600
+    },
+    {
+      "epoch": 0.001251663016852078,
+      "grad_norm": 9.150565147399902,
+      "learning_rate": 6.258393418256257e-07,
+      "loss": 8.0604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.001303815642554248,
+      "grad_norm": 8.797100067138672,
+      "learning_rate": 6.519159810683599e-07,
+      "loss": 7.9397,
+      "step": 5000
+    },
+    {
+      "epoch": 0.0013559682682564179,
+      "grad_norm": 8.194568634033203,
+      "learning_rate": 6.779926203110944e-07,
+      "loss": 7.8885,
+      "step": 5200
+    },
+    {
+      "epoch": 0.0014081208939585876,
+      "grad_norm": 8.242785453796387,
+      "learning_rate": 7.040692595538287e-07,
+      "loss": 7.8287,
+      "step": 5400
+    },
+    {
+      "epoch": 0.0014602735196607576,
+      "grad_norm": 9.448251724243164,
+      "learning_rate": 7.301458987965632e-07,
+      "loss": 7.7276,
+      "step": 5600
+    },
+    {
+      "epoch": 0.0015124261453629275,
+      "grad_norm": 7.932211875915527,
+      "learning_rate": 7.562225380392975e-07,
+      "loss": 7.6658,
+      "step": 5800
+    },
+    {
+      "epoch": 0.0015645787710650975,
+      "grad_norm": 8.316240310668945,
+      "learning_rate": 7.82299177282032e-07,
+      "loss": 7.5645,
+      "step": 6000
+    },
+    {
+      "epoch": 0.0016167313967672674,
+      "grad_norm": 8.247200965881348,
+      "learning_rate": 8.083758165247665e-07,
+      "loss": 7.508,
+      "step": 6200
+    },
+    {
+      "epoch": 0.0016688840224694372,
+      "grad_norm": 8.689373970031738,
+      "learning_rate": 8.344524557675007e-07,
+      "loss": 7.4703,
+      "step": 6400
+    },
+    {
+      "epoch": 0.001721036648171607,
+      "grad_norm": 7.99875545501709,
+      "learning_rate": 8.605290950102352e-07,
+      "loss": 7.3914,
+      "step": 6600
+    },
+    {
+      "epoch": 0.001773189273873777,
+      "grad_norm": 8.472365379333496,
+      "learning_rate": 8.866057342529695e-07,
+      "loss": 7.3268,
+      "step": 6800
+    },
+    {
+      "epoch": 0.001825341899575947,
+      "grad_norm": 8.410703659057617,
+      "learning_rate": 9.12682373495704e-07,
+      "loss": 7.2791,
+      "step": 7000
+    },
+    {
+      "epoch": 0.001877494525278117,
+      "grad_norm": 9.348575592041016,
+      "learning_rate": 9.387590127384382e-07,
+      "loss": 7.2181,
+      "step": 7200
+    },
+    {
+      "epoch": 0.001929647150980287,
+      "grad_norm": 8.044520378112793,
+      "learning_rate": 9.648356519811728e-07,
+      "loss": 7.1393,
+      "step": 7400
+    },
+    {
+      "epoch": 0.001981799776682457,
+      "grad_norm": 8.222118377685547,
+      "learning_rate": 9.909122912239071e-07,
+      "loss": 7.1147,
+      "step": 7600
+    },
+    {
+      "epoch": 0.002033952402384627,
+      "grad_norm": 8.315069198608398,
+      "learning_rate": 1.0169889304666415e-06,
+      "loss": 7.0527,
+      "step": 7800
+    },
+    {
+      "epoch": 0.0020861050280867968,
+      "grad_norm": 8.228837013244629,
+      "learning_rate": 1.043065569709376e-06,
+      "loss": 6.9868,
+      "step": 8000
+    },
+    {
+      "epoch": 0.0021382576537889663,
+      "grad_norm": 8.025296211242676,
+      "learning_rate": 1.0691422089521102e-06,
+      "loss": 6.9165,
+      "step": 8200
+    },
+    {
+      "epoch": 0.0021904102794911362,
+      "grad_norm": 8.421102523803711,
+      "learning_rate": 1.0952188481948448e-06,
+      "loss": 6.8728,
+      "step": 8400
+    },
+    {
+      "epoch": 0.002242562905193306,
+      "grad_norm": 8.444592475891113,
+      "learning_rate": 1.1212954874375791e-06,
+      "loss": 6.8894,
+      "step": 8600
+    },
+    {
+      "epoch": 0.002294715530895476,
+      "grad_norm": 9.23551082611084,
+      "learning_rate": 1.1473721266803135e-06,
+      "loss": 6.8165,
+      "step": 8800
+    },
+    {
+      "epoch": 0.002346868156597646,
+      "grad_norm": 7.454519748687744,
+      "learning_rate": 1.173448765923048e-06,
+      "loss": 6.7488,
+      "step": 9000
+    },
+    {
+      "epoch": 0.002399020782299816,
+      "grad_norm": 8.644956588745117,
+      "learning_rate": 1.1995254051657824e-06,
+      "loss": 6.7656,
+      "step": 9200
+    },
+    {
+      "epoch": 0.002451173408001986,
+      "grad_norm": 8.16921615600586,
+      "learning_rate": 1.2256020444085168e-06,
+      "loss": 6.6187,
+      "step": 9400
+    },
+    {
+      "epoch": 0.002503326033704156,
+      "grad_norm": 8.748774528503418,
+      "learning_rate": 1.2516786836512513e-06,
+      "loss": 6.5993,
+      "step": 9600
+    },
+    {
+      "epoch": 0.002555478659406326,
+      "grad_norm": 7.972047328948975,
+      "learning_rate": 1.2777553228939855e-06,
+      "loss": 6.6099,
+      "step": 9800
+    },
+    {
+      "epoch": 0.002607631285108496,
+      "grad_norm": 7.958851337432861,
+      "learning_rate": 1.3038319621367198e-06,
+      "loss": 6.5708,
+      "step": 10000
+    },
+    {
+      "epoch": 0.002659783910810666,
+      "grad_norm": 8.20793628692627,
+      "learning_rate": 1.3299086013794544e-06,
+      "loss": 6.5175,
+      "step": 10200
+    },
+    {
+      "epoch": 0.0027119365365128358,
+      "grad_norm": 8.190882682800293,
+      "learning_rate": 1.3559852406221887e-06,
+      "loss": 6.4806,
+      "step": 10400
+    },
+    {
+      "epoch": 0.0027640891622150053,
+      "grad_norm": 8.604300498962402,
+      "learning_rate": 1.3820618798649233e-06,
+      "loss": 6.4768,
+      "step": 10600
+    },
+    {
+      "epoch": 0.0028162417879171752,
+      "grad_norm": 8.548392295837402,
+      "learning_rate": 1.4081385191076574e-06,
+      "loss": 6.4409,
+      "step": 10800
+    },
+    {
+      "epoch": 0.002868394413619345,
+      "grad_norm": 7.730503559112549,
+      "learning_rate": 1.4342151583503918e-06,
+      "loss": 6.4013,
+      "step": 11000
+    },
+    {
+      "epoch": 0.002920547039321515,
+      "grad_norm": 8.61767292022705,
+      "learning_rate": 1.4602917975931264e-06,
+      "loss": 6.3355,
+      "step": 11200
+    },
+    {
+      "epoch": 0.002972699665023685,
+      "grad_norm": 7.736725330352783,
+      "learning_rate": 1.4863684368358607e-06,
+      "loss": 6.2992,
+      "step": 11400
+    },
+    {
+      "epoch": 0.003024852290725855,
+      "grad_norm": 8.229381561279297,
+      "learning_rate": 1.512445076078595e-06,
+      "loss": 6.293,
+      "step": 11600
+    },
+    {
+      "epoch": 0.003077004916428025,
+      "grad_norm": 8.532485008239746,
+      "learning_rate": 1.5385217153213294e-06,
+      "loss": 6.2469,
+      "step": 11800
+    },
+    {
+      "epoch": 0.003129157542130195,
+      "grad_norm": 8.137749671936035,
+      "learning_rate": 1.564598354564064e-06,
+      "loss": 6.1848,
+      "step": 12000
+    },
+    {
+      "epoch": 0.003181310167832365,
+      "grad_norm": 7.897096157073975,
+      "learning_rate": 1.5906749938067983e-06,
+      "loss": 6.1823,
+      "step": 12200
+    },
+    {
+      "epoch": 0.003233462793534535,
+      "grad_norm": 7.656859874725342,
+      "learning_rate": 1.616751633049533e-06,
+      "loss": 6.1289,
+      "step": 12400
+    },
+    {
+      "epoch": 0.003285615419236705,
+      "grad_norm": 8.03537654876709,
+      "learning_rate": 1.642828272292267e-06,
+      "loss": 6.1108,
+      "step": 12600
+    },
+    {
+      "epoch": 0.0033377680449388743,
+      "grad_norm": 7.999448299407959,
+      "learning_rate": 1.6689049115350014e-06,
+      "loss": 6.0881,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0033899206706410443,
+      "grad_norm": 8.130730628967285,
+      "learning_rate": 1.694981550777736e-06,
+      "loss": 6.05,
+      "step": 13000
+    },
+    {
+      "epoch": 0.003442073296343214,
+      "grad_norm": 8.437195777893066,
+      "learning_rate": 1.7210581900204703e-06,
+      "loss": 6.0568,
+      "step": 13200
+    },
+    {
+      "epoch": 0.003494225922045384,
+      "grad_norm": 8.395051956176758,
+      "learning_rate": 1.747134829263205e-06,
+      "loss": 6.0069,
+      "step": 13400
+    },
+    {
+      "epoch": 0.003546378547747554,
+      "grad_norm": 8.429478645324707,
+      "learning_rate": 1.773211468505939e-06,
+      "loss": 5.9598,
+      "step": 13600
+    },
+    {
+      "epoch": 0.003598531173449724,
+      "grad_norm": 8.610182762145996,
+      "learning_rate": 1.7992881077486734e-06,
+      "loss": 5.9299,
+      "step": 13800
+    },
+    {
+      "epoch": 0.003650683799151894,
+      "grad_norm": 7.439322471618652,
+      "learning_rate": 1.825364746991408e-06,
+      "loss": 5.913,
+      "step": 14000
+    },
+    {
+      "epoch": 0.003702836424854064,
+      "grad_norm": 8.080249786376953,
+      "learning_rate": 1.8514413862341423e-06,
+      "loss": 5.9126,
+      "step": 14200
+    },
+    {
+      "epoch": 0.003754989050556234,
+      "grad_norm": 7.7550201416015625,
+      "learning_rate": 1.8775180254768765e-06,
+      "loss": 5.8244,
+      "step": 14400
+    },
+    {
+      "epoch": 0.003807141676258404,
+      "grad_norm": 8.251408576965332,
+      "learning_rate": 1.903594664719611e-06,
+      "loss": 5.8174,
+      "step": 14600
+    },
+    {
+      "epoch": 0.003859294301960574,
+      "grad_norm": 7.648636817932129,
+      "learning_rate": 1.9296713039623456e-06,
+      "loss": 5.8122,
+      "step": 14800
+    },
+    {
+      "epoch": 0.003911446927662743,
+      "grad_norm": 7.906362533569336,
+      "learning_rate": 1.95574794320508e-06,
+      "loss": 5.7618,
+      "step": 15000
+    },
+    {
+      "epoch": 0.003963599553364914,
+      "grad_norm": 7.8534698486328125,
+      "learning_rate": 1.9818245824478143e-06,
+      "loss": 5.7498,
+      "step": 15200
+    },
+    {
+      "epoch": 0.004015752179067083,
+      "grad_norm": 8.2284574508667,
+      "learning_rate": 2.0079012216905486e-06,
+      "loss": 5.7466,
+      "step": 15400
+    },
+    {
+      "epoch": 0.004067904804769254,
+      "grad_norm": 8.237747192382812,
+      "learning_rate": 2.033977860933283e-06,
+      "loss": 5.7004,
+      "step": 15600
+    },
+    {
+      "epoch": 0.004120057430471423,
+      "grad_norm": 7.435640335083008,
+      "learning_rate": 2.0600545001760174e-06,
+      "loss": 5.6729,
+      "step": 15800
+    },
+    {
+      "epoch": 0.0041722100561735935,
+      "grad_norm": 7.641366958618164,
+      "learning_rate": 2.086131139418752e-06,
+      "loss": 5.6398,
+      "step": 16000
+    },
+    {
+      "epoch": 0.004224362681875763,
+      "grad_norm": 7.673348426818848,
+      "learning_rate": 2.1122077786614865e-06,
+      "loss": 5.6584,
+      "step": 16200
+    },
+    {
+      "epoch": 0.004276515307577933,
+      "grad_norm": 8.171424865722656,
+      "learning_rate": 2.1382844179042204e-06,
+      "loss": 5.5945,
+      "step": 16400
+    },
+    {
+      "epoch": 0.004328667933280103,
+      "grad_norm": 7.752162933349609,
+      "learning_rate": 2.164361057146955e-06,
+      "loss": 5.6081,
+      "step": 16600
+    },
+    {
+      "epoch": 0.0043808205589822725,
+      "grad_norm": 7.9565958976745605,
+      "learning_rate": 2.1904376963896896e-06,
+      "loss": 5.5563,
+      "step": 16800
+    },
+    {
+      "epoch": 0.004432973184684443,
+      "grad_norm": 7.686338424682617,
+      "learning_rate": 2.216514335632424e-06,
+      "loss": 5.5785,
+      "step": 17000
+    },
+    {
+      "epoch": 0.004485125810386612,
+      "grad_norm": 7.707006931304932,
+      "learning_rate": 2.2425909748751583e-06,
+      "loss": 5.57,
+      "step": 17200
+    },
+    {
+      "epoch": 0.004537278436088783,
+      "grad_norm": 8.033032417297363,
+      "learning_rate": 2.2686676141178926e-06,
+      "loss": 5.5128,
+      "step": 17400
+    },
+    {
+      "epoch": 0.004589431061790952,
+      "grad_norm": 8.142727851867676,
+      "learning_rate": 2.294744253360627e-06,
+      "loss": 5.4853,
+      "step": 17600
+    },
+    {
+      "epoch": 0.004641583687493123,
+      "grad_norm": 7.563163757324219,
+      "learning_rate": 2.3208208926033613e-06,
+      "loss": 5.4514,
+      "step": 17800
+    },
+    {
+      "epoch": 0.004693736313195292,
+      "grad_norm": 7.709859848022461,
+      "learning_rate": 2.346897531846096e-06,
+      "loss": 5.4332,
+      "step": 18000
+    },
+    {
+      "epoch": 0.004745888938897463,
+      "grad_norm": 8.00818920135498,
+      "learning_rate": 2.37297417108883e-06,
+      "loss": 5.474,
+      "step": 18200
+    },
+    {
+      "epoch": 0.004798041564599632,
+      "grad_norm": 8.24853229522705,
+      "learning_rate": 2.399050810331565e-06,
+      "loss": 5.4155,
+      "step": 18400
+    },
+    {
+      "epoch": 0.004850194190301802,
+      "grad_norm": 8.416743278503418,
+      "learning_rate": 2.425127449574299e-06,
+      "loss": 5.3975,
+      "step": 18600
+    },
+    {
+      "epoch": 0.004902346816003972,
+      "grad_norm": 8.024255752563477,
+      "learning_rate": 2.4512040888170335e-06,
+      "loss": 5.3896,
+      "step": 18800
+    },
+    {
+      "epoch": 0.0049544994417061415,
+      "grad_norm": 7.6818132400512695,
+      "learning_rate": 2.477280728059768e-06,
+      "loss": 5.3643,
+      "step": 19000
+    },
+    {
+      "epoch": 0.005006652067408312,
+      "grad_norm": 7.544449329376221,
+      "learning_rate": 2.5033573673025026e-06,
+      "loss": 5.3326,
+      "step": 19200
+    },
+    {
+      "epoch": 0.005058804693110481,
+      "grad_norm": 7.725053787231445,
+      "learning_rate": 2.5294340065452366e-06,
+      "loss": 5.3108,
+      "step": 19400
+    },
+    {
+      "epoch": 0.005110957318812652,
+      "grad_norm": 7.843037128448486,
+      "learning_rate": 2.555510645787971e-06,
+      "loss": 5.2766,
+      "step": 19600
+    },
+    {
+      "epoch": 0.005163109944514821,
+      "grad_norm": 7.6276164054870605,
+      "learning_rate": 2.5815872850307057e-06,
+      "loss": 5.3071,
+      "step": 19800
+    },
+    {
+      "epoch": 0.005215262570216992,
+      "grad_norm": 7.907887935638428,
+      "learning_rate": 2.6076639242734396e-06,
+      "loss": 5.2333,
+      "step": 20000
+    },
+    {
+      "epoch": 0.005267415195919161,
+      "grad_norm": 7.383638858795166,
+      "learning_rate": 2.6337405635161744e-06,
+      "loss": 5.3013,
+      "step": 20200
+    },
+    {
+      "epoch": 0.005319567821621332,
+      "grad_norm": 7.0909247398376465,
+      "learning_rate": 2.6598172027589088e-06,
+      "loss": 5.222,
+      "step": 20400
+    },
+    {
+      "epoch": 0.005371720447323501,
+      "grad_norm": 8.317301750183105,
+      "learning_rate": 2.6858938420016427e-06,
+      "loss": 5.1935,
+      "step": 20600
+    },
+    {
+      "epoch": 0.0054238730730256715,
+      "grad_norm": 7.909549713134766,
+      "learning_rate": 2.7119704812443775e-06,
+      "loss": 5.1732,
+      "step": 20800
+    },
+    {
+      "epoch": 0.005476025698727841,
+      "grad_norm": 8.112817764282227,
+      "learning_rate": 2.738047120487112e-06,
+      "loss": 5.2078,
+      "step": 21000
+    },
+    {
+      "epoch": 0.0055281783244300105,
+      "grad_norm": 7.593140125274658,
+      "learning_rate": 2.7641237597298466e-06,
+      "loss": 5.1478,
+      "step": 21200
+    },
+    {
+      "epoch": 0.005580330950132181,
+      "grad_norm": 7.865172863006592,
+      "learning_rate": 2.7902003989725805e-06,
+      "loss": 5.1364,
+      "step": 21400
+    },
+    {
+      "epoch": 0.0056324835758343504,
+      "grad_norm": 7.748229503631592,
+      "learning_rate": 2.816277038215315e-06,
+      "loss": 5.1232,
+      "step": 21600
+    },
+    {
+      "epoch": 0.005684636201536521,
+      "grad_norm": 7.954893112182617,
+      "learning_rate": 2.8423536774580497e-06,
+      "loss": 5.138,
+      "step": 21800
+    },
+    {
+      "epoch": 0.00573678882723869,
+      "grad_norm": 7.68300199508667,
+      "learning_rate": 2.8684303167007836e-06,
+      "loss": 5.1083,
+      "step": 22000
+    },
+    {
+      "epoch": 0.005788941452940861,
+      "grad_norm": 9.848554611206055,
+      "learning_rate": 2.8945069559435184e-06,
+      "loss": 5.0713,
+      "step": 22200
+    },
+    {
+      "epoch": 0.00584109407864303,
+      "grad_norm": 7.352326393127441,
+      "learning_rate": 2.9205835951862527e-06,
+      "loss": 5.0239,
+      "step": 22400
+    },
+    {
+      "epoch": 0.005893246704345201,
+      "grad_norm": 8.010299682617188,
+      "learning_rate": 2.9466602344289867e-06,
+      "loss": 5.0562,
+      "step": 22600
+    },
+    {
+      "epoch": 0.00594539933004737,
+      "grad_norm": 7.50053071975708,
+      "learning_rate": 2.9727368736717214e-06,
+      "loss": 4.9805,
+      "step": 22800
+    },
+    {
+      "epoch": 0.0059975519557495405,
+      "grad_norm": 7.903580188751221,
+      "learning_rate": 2.998813512914456e-06,
+      "loss": 5.0251,
+      "step": 23000
+    },
+    {
+      "epoch": 0.00604970458145171,
+      "grad_norm": 7.933001518249512,
+      "learning_rate": 3.02489015215719e-06,
+      "loss": 5.0268,
+      "step": 23200
+    },
+    {
+      "epoch": 0.00610185720715388,
+      "grad_norm": 7.974423885345459,
+      "learning_rate": 3.0509667913999245e-06,
+      "loss": 5.0149,
+      "step": 23400
+    },
+    {
+      "epoch": 0.00615400983285605,
+      "grad_norm": 7.946493148803711,
+      "learning_rate": 3.077043430642659e-06,
+      "loss": 4.9947,
+      "step": 23600
+    },
+    {
+      "epoch": 0.0062061624585582195,
+      "grad_norm": 8.70067024230957,
+      "learning_rate": 3.1031200698853936e-06,
+      "loss": 4.9885,
+      "step": 23800
+    },
+    {
+      "epoch": 0.00625831508426039,
+      "grad_norm": 7.401217937469482,
+      "learning_rate": 3.129196709128128e-06,
+      "loss": 4.9058,
+      "step": 24000
+    },
+    {
+      "epoch": 0.006310467709962559,
+      "grad_norm": 8.371158599853516,
+      "learning_rate": 3.155273348370862e-06,
+      "loss": 4.9456,
+      "step": 24200
+    },
+    {
+      "epoch": 0.00636262033566473,
+      "grad_norm": 7.775518894195557,
+      "learning_rate": 3.1813499876135967e-06,
+      "loss": 4.9482,
+      "step": 24400
+    },
+    {
+      "epoch": 0.006414772961366899,
+      "grad_norm": 7.885232448577881,
+      "learning_rate": 3.207426626856331e-06,
+      "loss": 4.9251,
+      "step": 24600
+    },
+    {
+      "epoch": 0.00646692558706907,
+      "grad_norm": 7.657386302947998,
+      "learning_rate": 3.233503266099066e-06,
+      "loss": 4.8979,
+      "step": 24800
+    },
+    {
+      "epoch": 0.006519078212771239,
+      "grad_norm": 7.980145454406738,
+      "learning_rate": 3.2595799053417998e-06,
+      "loss": 4.8544,
+      "step": 25000
+    },
+    {
+      "epoch": 0.00657123083847341,
+      "grad_norm": 7.923119068145752,
+      "learning_rate": 3.285656544584534e-06,
+      "loss": 4.8472,
+      "step": 25200
+    },
+    {
+      "epoch": 0.006623383464175579,
+      "grad_norm": 7.500796318054199,
+      "learning_rate": 3.311733183827269e-06,
+      "loss": 4.8235,
+      "step": 25400
+    },
+    {
+      "epoch": 0.006675536089877749,
+      "grad_norm": 7.593719005584717,
+      "learning_rate": 3.337809823070003e-06,
+      "loss": 4.8552,
+      "step": 25600
+    },
+    {
+      "epoch": 0.006727688715579919,
+      "grad_norm": 7.597663879394531,
+      "learning_rate": 3.3638864623127376e-06,
+      "loss": 4.822,
+      "step": 25800
+    },
+    {
+      "epoch": 0.0067798413412820885,
+      "grad_norm": 7.924131393432617,
+      "learning_rate": 3.389963101555472e-06,
+      "loss": 4.8013,
+      "step": 26000
+    },
+    {
+      "epoch": 0.006831993966984259,
+      "grad_norm": 8.150907516479492,
+      "learning_rate": 3.416039740798206e-06,
+      "loss": 4.808,
+      "step": 26200
+    },
+    {
+      "epoch": 0.006884146592686428,
+      "grad_norm": 8.553031921386719,
+      "learning_rate": 3.4421163800409407e-06,
+      "loss": 4.7903,
+      "step": 26400
+    },
+    {
+      "epoch": 0.006936299218388599,
+      "grad_norm": 7.992140293121338,
+      "learning_rate": 3.468193019283675e-06,
+      "loss": 4.7669,
+      "step": 26600
+    },
+    {
+      "epoch": 0.006988451844090768,
+      "grad_norm": 9.123887062072754,
+      "learning_rate": 3.49426965852641e-06,
+      "loss": 4.7772,
+      "step": 26800
+    },
+    {
+      "epoch": 0.007040604469792939,
+      "grad_norm": 7.843687057495117,
+      "learning_rate": 3.5203462977691437e-06,
+      "loss": 4.7443,
+      "step": 27000
+    },
+    {
+      "epoch": 0.007092757095495108,
+      "grad_norm": 7.194400787353516,
+      "learning_rate": 3.546422937011878e-06,
+      "loss": 4.6394,
+      "step": 27200
+    },
+    {
+      "epoch": 0.007144909721197279,
+      "grad_norm": 8.256219863891602,
+      "learning_rate": 3.572499576254613e-06,
+      "loss": 4.7375,
+      "step": 27400
+    },
+    {
+      "epoch": 0.007197062346899448,
+      "grad_norm": 7.898242473602295,
+      "learning_rate": 3.5985762154973468e-06,
+      "loss": 4.6597,
+      "step": 27600
+    },
+    {
+      "epoch": 0.0072492149726016185,
+      "grad_norm": 8.30362606048584,
+      "learning_rate": 3.6246528547400816e-06,
+      "loss": 4.6724,
+      "step": 27800
+    },
+    {
+      "epoch": 0.007301367598303788,
+      "grad_norm": 7.795161247253418,
+      "learning_rate": 3.650729493982816e-06,
+      "loss": 4.6485,
+      "step": 28000
+    },
+    {
+      "epoch": 0.0073535202240059576,
+      "grad_norm": 7.712084770202637,
+      "learning_rate": 3.67680613322555e-06,
+      "loss": 4.6684,
+      "step": 28200
+    },
+    {
+      "epoch": 0.007405672849708128,
+      "grad_norm": 7.2499165534973145,
+      "learning_rate": 3.7028827724682846e-06,
+      "loss": 4.6383,
+      "step": 28400
+    },
+    {
+      "epoch": 0.0074578254754102975,
+      "grad_norm": 7.843800067901611,
+      "learning_rate": 3.728959411711019e-06,
+      "loss": 4.644,
+      "step": 28600
+    },
+    {
+      "epoch": 0.007509978101112468,
+      "grad_norm": 8.171004295349121,
+      "learning_rate": 3.755036050953753e-06,
+      "loss": 4.5964,
+      "step": 28800
+    },
+    {
+      "epoch": 0.007562130726814637,
+      "grad_norm": 8.430412292480469,
+      "learning_rate": 3.7811126901964877e-06,
+      "loss": 4.5651,
+      "step": 29000
+    },
+    {
+      "epoch": 0.007614283352516808,
+      "grad_norm": 8.003270149230957,
+      "learning_rate": 3.807189329439222e-06,
+      "loss": 4.6326,
+      "step": 29200
+    },
+    {
+      "epoch": 0.007666435978218977,
+      "grad_norm": 8.111075401306152,
+      "learning_rate": 3.833265968681957e-06,
+      "loss": 4.5397,
+      "step": 29400
+    },
+    {
+      "epoch": 0.007718588603921148,
+      "grad_norm": 8.523173332214355,
+      "learning_rate": 3.859342607924691e-06,
+      "loss": 4.5506,
+      "step": 29600
+    },
+    {
+      "epoch": 0.007770741229623317,
+      "grad_norm": 8.225560188293457,
+      "learning_rate": 3.8854192471674255e-06,
+      "loss": 4.5974,
+      "step": 29800
+    },
+    {
+      "epoch": 0.007822893855325487,
+      "grad_norm": 8.14545726776123,
+      "learning_rate": 3.91149588641016e-06,
+      "loss": 4.5563,
+      "step": 30000
+    },
+    {
+      "epoch": 0.007875046481027657,
+      "grad_norm": 8.375019073486328,
+      "learning_rate": 3.937572525652894e-06,
+      "loss": 4.513,
+      "step": 30200
+    },
+    {
+      "epoch": 0.007927199106729827,
+      "grad_norm": 7.618561744689941,
+      "learning_rate": 3.963649164895629e-06,
+      "loss": 4.5606,
+      "step": 30400
+    },
+    {
+      "epoch": 0.007979351732431996,
+      "grad_norm": 7.901875019073486,
+      "learning_rate": 3.989725804138363e-06,
+      "loss": 4.5278,
+      "step": 30600
+    },
+    {
+      "epoch": 0.008031504358134166,
+      "grad_norm": 8.019709587097168,
+      "learning_rate": 4.015802443381097e-06,
+      "loss": 4.5119,
+      "step": 30800
+    },
+    {
+      "epoch": 0.008083656983836337,
+      "grad_norm": 7.532272815704346,
+      "learning_rate": 4.041879082623832e-06,
+      "loss": 4.4828,
+      "step": 31000
+    },
+    {
+      "epoch": 0.008135809609538507,
+      "grad_norm": 8.28953742980957,
+      "learning_rate": 4.067955721866566e-06,
+      "loss": 4.516,
+      "step": 31200
+    },
+    {
+      "epoch": 0.008187962235240676,
+      "grad_norm": 8.234850883483887,
+      "learning_rate": 4.094032361109301e-06,
+      "loss": 4.5126,
+      "step": 31400
+    },
+    {
+      "epoch": 0.008240114860942846,
+      "grad_norm": 7.353629112243652,
+      "learning_rate": 4.120109000352035e-06,
+      "loss": 4.4683,
+      "step": 31600
+    },
+    {
+      "epoch": 0.008292267486645017,
+      "grad_norm": 7.64729118347168,
+      "learning_rate": 4.146185639594769e-06,
+      "loss": 4.4538,
+      "step": 31800
+    },
+    {
+      "epoch": 0.008344420112347187,
+      "grad_norm": 7.47349214553833,
+      "learning_rate": 4.172262278837504e-06,
+      "loss": 4.4283,
+      "step": 32000
+    },
+    {
+      "epoch": 0.008396572738049356,
+      "grad_norm": 8.011587142944336,
+      "learning_rate": 4.198338918080238e-06,
+      "loss": 4.414,
+      "step": 32200
+    },
+    {
+      "epoch": 0.008448725363751526,
+      "grad_norm": 8.157088279724121,
+      "learning_rate": 4.224415557322973e-06,
+      "loss": 4.4103,
+      "step": 32400
+    },
+    {
+      "epoch": 0.008500877989453696,
+      "grad_norm": 8.141148567199707,
+      "learning_rate": 4.250492196565707e-06,
+      "loss": 4.3756,
+      "step": 32600
+    },
+    {
+      "epoch": 0.008553030615155865,
+      "grad_norm": 7.892102241516113,
+      "learning_rate": 4.276568835808441e-06,
+      "loss": 4.415,
+      "step": 32800
+    },
+    {
+      "epoch": 0.008605183240858036,
+      "grad_norm": 8.264842987060547,
+      "learning_rate": 4.302645475051176e-06,
+      "loss": 4.4202,
+      "step": 33000
+    },
+    {
+      "epoch": 0.008657335866560206,
+      "grad_norm": 7.995069980621338,
+      "learning_rate": 4.32872211429391e-06,
+      "loss": 4.407,
+      "step": 33200
+    },
+    {
+      "epoch": 0.008709488492262376,
+      "grad_norm": 8.319681167602539,
+      "learning_rate": 4.354798753536645e-06,
+      "loss": 4.3725,
+      "step": 33400
+    },
+    {
+      "epoch": 0.008761641117964545,
+      "grad_norm": 7.941286087036133,
+      "learning_rate": 4.380875392779379e-06,
+      "loss": 4.3491,
+      "step": 33600
+    },
+    {
+      "epoch": 0.008813793743666715,
+      "grad_norm": 7.6486029624938965,
+      "learning_rate": 4.4069520320221135e-06,
+      "loss": 4.3385,
+      "step": 33800
+    },
+    {
+      "epoch": 0.008865946369368886,
+      "grad_norm": 8.253968238830566,
+      "learning_rate": 4.433028671264848e-06,
+      "loss": 4.3506,
+      "step": 34000
+    },
+    {
+      "epoch": 0.008918098995071056,
+      "grad_norm": 7.83550500869751,
+      "learning_rate": 4.459105310507582e-06,
+      "loss": 4.3523,
+      "step": 34200
+    },
+    {
+      "epoch": 0.008970251620773225,
+      "grad_norm": 8.142230987548828,
+      "learning_rate": 4.4851819497503165e-06,
+      "loss": 4.3497,
+      "step": 34400
+    },
+    {
+      "epoch": 0.009022404246475395,
+      "grad_norm": 8.258264541625977,
+      "learning_rate": 4.511258588993051e-06,
+      "loss": 4.334,
+      "step": 34600
+    },
+    {
+      "epoch": 0.009074556872177566,
+      "grad_norm": 7.864555358886719,
+      "learning_rate": 4.537335228235785e-06,
+      "loss": 4.3043,
+      "step": 34800
+    },
+    {
+      "epoch": 0.009126709497879734,
+      "grad_norm": 7.830788612365723,
+      "learning_rate": 4.56341186747852e-06,
+      "loss": 4.2891,
+      "step": 35000
+    },
+    {
+      "epoch": 0.009178862123581905,
+      "grad_norm": 8.586557388305664,
+      "learning_rate": 4.589488506721254e-06,
+      "loss": 4.2071,
+      "step": 35200
+    },
+    {
+      "epoch": 0.009231014749284075,
+      "grad_norm": 8.016130447387695,
+      "learning_rate": 4.615565145963988e-06,
+      "loss": 4.2689,
+      "step": 35400
+    },
+    {
+      "epoch": 0.009283167374986245,
+      "grad_norm": 7.466185092926025,
+      "learning_rate": 4.641641785206723e-06,
+      "loss": 4.313,
+      "step": 35600
+    },
+    {
+      "epoch": 0.009335320000688414,
+      "grad_norm": 7.97619104385376,
+      "learning_rate": 4.667718424449457e-06,
+      "loss": 4.2527,
+      "step": 35800
+    },
+    {
+      "epoch": 0.009387472626390584,
+      "grad_norm": 8.14663028717041,
+      "learning_rate": 4.693795063692192e-06,
+      "loss": 4.2316,
+      "step": 36000
+    },
+    {
+      "epoch": 0.009439625252092755,
+      "grad_norm": 7.500378608703613,
+      "learning_rate": 4.7198717029349265e-06,
+      "loss": 4.2477,
+      "step": 36200
+    },
+    {
+      "epoch": 0.009491777877794925,
+      "grad_norm": 7.814052104949951,
+      "learning_rate": 4.74594834217766e-06,
+      "loss": 4.2685,
+      "step": 36400
+    },
+    {
+      "epoch": 0.009543930503497094,
+      "grad_norm": 7.774244785308838,
+      "learning_rate": 4.772024981420395e-06,
+      "loss": 4.2282,
+      "step": 36600
+    },
+    {
+      "epoch": 0.009596083129199264,
+      "grad_norm": 7.567872524261475,
+      "learning_rate": 4.79810162066313e-06,
+      "loss": 4.2201,
+      "step": 36800
+    },
+    {
+      "epoch": 0.009648235754901435,
+      "grad_norm": 7.794075965881348,
+      "learning_rate": 4.824178259905864e-06,
+      "loss": 4.2564,
+      "step": 37000
+    },
+    {
+      "epoch": 0.009700388380603603,
+      "grad_norm": 7.229556560516357,
+      "learning_rate": 4.850254899148598e-06,
+      "loss": 4.197,
+      "step": 37200
+    },
+    {
+      "epoch": 0.009752541006305774,
+      "grad_norm": 7.552301406860352,
+      "learning_rate": 4.876331538391333e-06,
+      "loss": 4.1951,
+      "step": 37400
+    },
+    {
+      "epoch": 0.009804693632007944,
+      "grad_norm": 7.268792152404785,
+      "learning_rate": 4.902408177634067e-06,
+      "loss": 4.1937,
+      "step": 37600
+    },
+    {
+      "epoch": 0.009856846257710114,
+      "grad_norm": 8.427607536315918,
+      "learning_rate": 4.928484816876801e-06,
+      "loss": 4.1549,
+      "step": 37800
+    },
+    {
+      "epoch": 0.009908998883412283,
+      "grad_norm": 7.394150733947754,
+      "learning_rate": 4.954561456119536e-06,
+      "loss": 4.1654,
+      "step": 38000
+    },
+    {
+      "epoch": 0.009961151509114453,
+      "grad_norm": 8.243969917297363,
+      "learning_rate": 4.98063809536227e-06,
+      "loss": 4.1688,
+      "step": 38200
+    },
+    {
+      "epoch": 0.010013304134816624,
+      "grad_norm": 7.82209587097168,
+      "learning_rate": 5.006714734605005e-06,
+      "loss": 4.164,
+      "step": 38400
+    },
+    {
+      "epoch": 0.010065456760518794,
+      "grad_norm": 7.940906047821045,
+      "learning_rate": 5.032791373847739e-06,
+      "loss": 4.1694,
+      "step": 38600
+    },
+    {
+      "epoch": 0.010117609386220963,
+      "grad_norm": 7.952986717224121,
+      "learning_rate": 5.058868013090473e-06,
+      "loss": 4.18,
+      "step": 38800
+    },
+    {
+      "epoch": 0.010169762011923133,
+      "grad_norm": 8.000782012939453,
+      "learning_rate": 5.084944652333208e-06,
+      "loss": 4.1193,
+      "step": 39000
+    },
+    {
+      "epoch": 0.010221914637625304,
+      "grad_norm": 7.911661148071289,
+      "learning_rate": 5.111021291575942e-06,
+      "loss": 4.0802,
+      "step": 39200
+    },
+    {
+      "epoch": 0.010274067263327474,
+      "grad_norm": 7.344413757324219,
+      "learning_rate": 5.137097930818676e-06,
+      "loss": 4.0842,
+      "step": 39400
+    },
+    {
+      "epoch": 0.010326219889029643,
+      "grad_norm": 8.170016288757324,
+      "learning_rate": 5.163174570061411e-06,
+      "loss": 4.111,
+      "step": 39600
+    },
+    {
+      "epoch": 0.010378372514731813,
+      "grad_norm": 7.0969719886779785,
+      "learning_rate": 5.189251209304146e-06,
+      "loss": 4.1323,
+      "step": 39800
+    },
+    {
+      "epoch": 0.010430525140433983,
+      "grad_norm": 7.286205291748047,
+      "learning_rate": 5.215327848546879e-06,
+      "loss": 4.0863,
+      "step": 40000
+    },
+    {
+      "epoch": 0.010482677766136152,
+      "grad_norm": 7.330994129180908,
+      "learning_rate": 5.2414044877896145e-06,
+      "loss": 4.0936,
+      "step": 40200
+    },
+    {
+      "epoch": 0.010534830391838322,
+      "grad_norm": 7.573282718658447,
+      "learning_rate": 5.267481127032349e-06,
+      "loss": 4.1067,
+      "step": 40400
+    },
+    {
+      "epoch": 0.010586983017540493,
+      "grad_norm": 8.244158744812012,
+      "learning_rate": 5.293557766275082e-06,
+      "loss": 4.0502,
+      "step": 40600
+    },
+    {
+      "epoch": 0.010639135643242663,
+      "grad_norm": 7.125401496887207,
+      "learning_rate": 5.3196344055178175e-06,
+      "loss": 4.0565,
+      "step": 40800
+    },
+    {
+      "epoch": 0.010691288268944832,
+      "grad_norm": 8.068306922912598,
+      "learning_rate": 5.345711044760552e-06,
+      "loss": 4.0404,
+      "step": 41000
+    },
+    {
+      "epoch": 0.010743440894647002,
+      "grad_norm": 7.503974437713623,
+      "learning_rate": 5.371787684003285e-06,
+      "loss": 4.063,
+      "step": 41200
+    },
+    {
+      "epoch": 0.010795593520349173,
+      "grad_norm": 7.42249059677124,
+      "learning_rate": 5.397864323246021e-06,
+      "loss": 4.0053,
+      "step": 41400
+    },
+    {
+      "epoch": 0.010847746146051343,
+      "grad_norm": 7.555575370788574,
+      "learning_rate": 5.423940962488755e-06,
+      "loss": 4.045,
+      "step": 41600
+    },
+    {
+      "epoch": 0.010899898771753512,
+      "grad_norm": 7.423194885253906,
+      "learning_rate": 5.4500176017314885e-06,
+      "loss": 4.0292,
+      "step": 41800
+    },
+    {
+      "epoch": 0.010952051397455682,
+      "grad_norm": 7.335684776306152,
+      "learning_rate": 5.476094240974224e-06,
+      "loss": 3.9726,
+      "step": 42000
+    },
+    {
+      "epoch": 0.011004204023157852,
+      "grad_norm": 7.528399467468262,
+      "learning_rate": 5.502170880216958e-06,
+      "loss": 3.9907,
+      "step": 42200
+    },
+    {
+      "epoch": 0.011056356648860021,
+      "grad_norm": 7.644163131713867,
+      "learning_rate": 5.528247519459693e-06,
+      "loss": 4.034,
+      "step": 42400
+    },
+    {
+      "epoch": 0.011108509274562191,
+      "grad_norm": 8.300440788269043,
+      "learning_rate": 5.554324158702427e-06,
+      "loss": 4.0416,
+      "step": 42600
+    },
+    {
+      "epoch": 0.011160661900264362,
+      "grad_norm": 7.345607280731201,
+      "learning_rate": 5.580400797945161e-06,
+      "loss": 3.9673,
+      "step": 42800
+    },
+    {
+      "epoch": 0.011212814525966532,
+      "grad_norm": 7.909209251403809,
+      "learning_rate": 5.606477437187896e-06,
+      "loss": 3.9715,
+      "step": 43000
+    },
+    {
+      "epoch": 0.011264967151668701,
+      "grad_norm": 7.389960289001465,
+      "learning_rate": 5.63255407643063e-06,
+      "loss": 3.9649,
+      "step": 43200
+    },
+    {
+      "epoch": 0.011317119777370871,
+      "grad_norm": 7.785503387451172,
+      "learning_rate": 5.658630715673364e-06,
+      "loss": 3.9642,
+      "step": 43400
+    },
+    {
+      "epoch": 0.011369272403073042,
+      "grad_norm": 7.3766984939575195,
+      "learning_rate": 5.684707354916099e-06,
+      "loss": 4.0052,
+      "step": 43600
+    },
+    {
+      "epoch": 0.011421425028775212,
+      "grad_norm": 7.540368556976318,
+      "learning_rate": 5.710783994158833e-06,
+      "loss": 4.0034,
+      "step": 43800
+    },
+    {
+      "epoch": 0.01147357765447738,
+      "grad_norm": 8.014230728149414,
+      "learning_rate": 5.736860633401567e-06,
+      "loss": 3.9643,
+      "step": 44000
+    },
+    {
+      "epoch": 0.011525730280179551,
+      "grad_norm": 7.941577434539795,
+      "learning_rate": 5.762937272644302e-06,
+      "loss": 3.9858,
+      "step": 44200
+    },
+    {
+      "epoch": 0.011577882905881721,
+      "grad_norm": 7.780430793762207,
+      "learning_rate": 5.789013911887037e-06,
+      "loss": 3.993,
+      "step": 44400
+    },
+    {
+      "epoch": 0.01163003553158389,
+      "grad_norm": 8.022960662841797,
+      "learning_rate": 5.81509055112977e-06,
+      "loss": 3.9126,
+      "step": 44600
+    },
+    {
+      "epoch": 0.01168218815728606,
+      "grad_norm": 8.14714527130127,
+      "learning_rate": 5.8411671903725055e-06,
+      "loss": 3.9218,
+      "step": 44800
+    },
+    {
+      "epoch": 0.011734340782988231,
+      "grad_norm": 7.369551658630371,
+      "learning_rate": 5.86724382961524e-06,
+      "loss": 3.959,
+      "step": 45000
+    },
+    {
+      "epoch": 0.011786493408690401,
+      "grad_norm": 8.105989456176758,
+      "learning_rate": 5.893320468857973e-06,
+      "loss": 3.8897,
+      "step": 45200
+    },
+    {
+      "epoch": 0.01183864603439257,
+      "grad_norm": 8.191625595092773,
+      "learning_rate": 5.9193971081007085e-06,
+      "loss": 3.9187,
+      "step": 45400
+    },
+    {
+      "epoch": 0.01189079866009474,
+      "grad_norm": 7.703822135925293,
+      "learning_rate": 5.945473747343443e-06,
+      "loss": 3.8947,
+      "step": 45600
+    },
+    {
+      "epoch": 0.01194295128579691,
+      "grad_norm": 7.697946548461914,
+      "learning_rate": 5.971550386586176e-06,
+      "loss": 3.8795,
+      "step": 45800
+    },
+    {
+      "epoch": 0.011995103911499081,
+      "grad_norm": 7.603724002838135,
+      "learning_rate": 5.997627025828912e-06,
+      "loss": 3.911,
+      "step": 46000
+    },
+    {
+      "epoch": 0.01204725653720125,
+      "grad_norm": 8.207696914672852,
+      "learning_rate": 6.023703665071646e-06,
+      "loss": 3.896,
+      "step": 46200
+    },
+    {
+      "epoch": 0.01209940916290342,
+      "grad_norm": 8.124335289001465,
+      "learning_rate": 6.04978030431438e-06,
+      "loss": 3.8306,
+      "step": 46400
+    },
+    {
+      "epoch": 0.01215156178860559,
+      "grad_norm": 7.28653621673584,
+      "learning_rate": 6.075856943557115e-06,
+      "loss": 3.8624,
+      "step": 46600
+    },
+    {
+      "epoch": 0.01220371441430776,
+      "grad_norm": 7.189661026000977,
+      "learning_rate": 6.101933582799849e-06,
+      "loss": 3.8445,
+      "step": 46800
+    },
+    {
+      "epoch": 0.01225586704000993,
+      "grad_norm": 8.082289695739746,
+      "learning_rate": 6.128010222042584e-06,
+      "loss": 3.8956,
+      "step": 47000
+    },
+    {
+      "epoch": 0.0123080196657121,
+      "grad_norm": 7.882737636566162,
+      "learning_rate": 6.154086861285318e-06,
+      "loss": 3.8779,
+      "step": 47200
+    },
+    {
+      "epoch": 0.01236017229141427,
+      "grad_norm": 7.791324615478516,
+      "learning_rate": 6.180163500528052e-06,
+      "loss": 3.8339,
+      "step": 47400
+    },
+    {
+      "epoch": 0.012412324917116439,
+      "grad_norm": 7.604274272918701,
+      "learning_rate": 6.206240139770787e-06,
+      "loss": 3.8552,
+      "step": 47600
+    },
+    {
+      "epoch": 0.01246447754281861,
+      "grad_norm": 7.377195358276367,
+      "learning_rate": 6.232316779013521e-06,
+      "loss": 3.8075,
+      "step": 47800
+    },
+    {
+      "epoch": 0.01251663016852078,
+      "grad_norm": 7.8386640548706055,
+      "learning_rate": 6.258393418256256e-06,
+      "loss": 3.7798,
+      "step": 48000
+    },
+    {
+      "epoch": 0.01256878279422295,
+      "grad_norm": 7.123723983764648,
+      "learning_rate": 6.28447005749899e-06,
+      "loss": 3.8206,
+      "step": 48200
+    },
+    {
+      "epoch": 0.012620935419925119,
+      "grad_norm": 8.309353828430176,
+      "learning_rate": 6.310546696741724e-06,
+      "loss": 3.8562,
+      "step": 48400
+    },
+    {
+      "epoch": 0.01267308804562729,
+      "grad_norm": 7.630406379699707,
+      "learning_rate": 6.336623335984459e-06,
+      "loss": 3.829,
+      "step": 48600
+    },
+    {
+      "epoch": 0.01272524067132946,
+      "grad_norm": 8.362411499023438,
+      "learning_rate": 6.362699975227193e-06,
+      "loss": 3.8011,
+      "step": 48800
+    },
+    {
+      "epoch": 0.012777393297031628,
+      "grad_norm": 7.613163471221924,
+      "learning_rate": 6.388776614469928e-06,
+      "loss": 3.8119,
+      "step": 49000
+    },
+    {
+      "epoch": 0.012829545922733799,
+      "grad_norm": 8.339622497558594,
+      "learning_rate": 6.414853253712662e-06,
+      "loss": 3.8347,
+      "step": 49200
+    },
+    {
+      "epoch": 0.012881698548435969,
+      "grad_norm": 7.240667343139648,
+      "learning_rate": 6.4409298929553965e-06,
+      "loss": 3.7863,
+      "step": 49400
+    },
+    {
+      "epoch": 0.01293385117413814,
+      "grad_norm": 7.500837326049805,
+      "learning_rate": 6.467006532198132e-06,
+      "loss": 3.78,
+      "step": 49600
+    },
+    {
+      "epoch": 0.012986003799840308,
+      "grad_norm": 7.476478099822998,
+      "learning_rate": 6.493083171440865e-06,
+      "loss": 3.8424,
+      "step": 49800
+    },
+    {
+      "epoch": 0.013038156425542478,
+      "grad_norm": 7.900282382965088,
+      "learning_rate": 6.5191598106835995e-06,
+      "loss": 3.757,
+      "step": 50000
+    },
+    {
+      "epoch": 0.013090309051244649,
+      "grad_norm": 7.225290298461914,
+      "learning_rate": 6.545236449926335e-06,
+      "loss": 3.751,
+      "step": 50200
+    },
+    {
+      "epoch": 0.01314246167694682,
+      "grad_norm": 7.843364715576172,
+      "learning_rate": 6.571313089169068e-06,
+      "loss": 3.8209,
+      "step": 50400
+    },
+    {
+      "epoch": 0.013194614302648988,
+      "grad_norm": 7.373351573944092,
+      "learning_rate": 6.597389728411803e-06,
+      "loss": 3.8055,
+      "step": 50600
+    },
+    {
+      "epoch": 0.013246766928351158,
+      "grad_norm": 8.107486724853516,
+      "learning_rate": 6.623466367654538e-06,
+      "loss": 3.7116,
+      "step": 50800
+    },
+    {
+      "epoch": 0.013298919554053329,
+      "grad_norm": 7.618676662445068,
+      "learning_rate": 6.649543006897272e-06,
+      "loss": 3.7805,
+      "step": 51000
+    },
+    {
+      "epoch": 0.013351072179755497,
+      "grad_norm": 7.945600509643555,
+      "learning_rate": 6.675619646140006e-06,
+      "loss": 3.7396,
+      "step": 51200
+    },
+    {
+      "epoch": 0.013403224805457668,
+      "grad_norm": 7.414207935333252,
+      "learning_rate": 6.701696285382741e-06,
+      "loss": 3.7788,
+      "step": 51400
+    },
+    {
+      "epoch": 0.013455377431159838,
+      "grad_norm": 8.000473022460938,
+      "learning_rate": 6.727772924625475e-06,
+      "loss": 3.7445,
+      "step": 51600
+    },
+    {
+      "epoch": 0.013507530056862008,
+      "grad_norm": 7.633871555328369,
+      "learning_rate": 6.753849563868209e-06,
+      "loss": 3.7474,
+      "step": 51800
+    },
+    {
+      "epoch": 0.013559682682564177,
+      "grad_norm": 7.910715103149414,
+      "learning_rate": 6.779926203110944e-06,
+      "loss": 3.7563,
+      "step": 52000
+    },
+    {
+      "epoch": 0.013611835308266347,
+      "grad_norm": 8.315561294555664,
+      "learning_rate": 6.806002842353678e-06,
+      "loss": 3.7435,
+      "step": 52200
+    },
+    {
+      "epoch": 0.013663987933968518,
+      "grad_norm": 7.666395664215088,
+      "learning_rate": 6.832079481596412e-06,
+      "loss": 3.7198,
+      "step": 52400
+    },
+    {
+      "epoch": 0.013716140559670688,
+      "grad_norm": 7.776802062988281,
+      "learning_rate": 6.858156120839147e-06,
+      "loss": 3.7318,
+      "step": 52600
+    },
+    {
+      "epoch": 0.013768293185372857,
+      "grad_norm": 7.258547782897949,
+      "learning_rate": 6.884232760081881e-06,
+      "loss": 3.7325,
+      "step": 52800
+    },
+    {
+      "epoch": 0.013820445811075027,
+      "grad_norm": 7.867082595825195,
+      "learning_rate": 6.910309399324615e-06,
+      "loss": 3.7163,
+      "step": 53000
+    },
+    {
+      "epoch": 0.013872598436777198,
+      "grad_norm": 7.452786922454834,
+      "learning_rate": 6.93638603856735e-06,
+      "loss": 3.7157,
+      "step": 53200
+    },
+    {
+      "epoch": 0.013924751062479366,
+      "grad_norm": 7.4677042961120605,
+      "learning_rate": 6.962462677810084e-06,
+      "loss": 3.7251,
+      "step": 53400
+    },
+    {
+      "epoch": 0.013976903688181537,
+      "grad_norm": 7.686807155609131,
+      "learning_rate": 6.98853931705282e-06,
+      "loss": 3.7158,
+      "step": 53600
+    },
+    {
+      "epoch": 0.014029056313883707,
+      "grad_norm": 7.389094352722168,
+      "learning_rate": 7.014615956295553e-06,
+      "loss": 3.7139,
+      "step": 53800
+    },
+    {
+      "epoch": 0.014081208939585877,
+      "grad_norm": 7.3070526123046875,
+      "learning_rate": 7.0406925955382874e-06,
+      "loss": 3.6748,
+      "step": 54000
+    },
+    {
+      "epoch": 0.014133361565288046,
+      "grad_norm": 7.088414192199707,
+      "learning_rate": 7.066769234781023e-06,
+      "loss": 3.6817,
+      "step": 54200
+    },
+    {
+      "epoch": 0.014185514190990216,
+      "grad_norm": 7.4261088371276855,
+      "learning_rate": 7.092845874023756e-06,
+      "loss": 3.7422,
+      "step": 54400
+    },
+    {
+      "epoch": 0.014237666816692387,
+      "grad_norm": 7.321529388427734,
+      "learning_rate": 7.1189225132664905e-06,
+      "loss": 3.7282,
+      "step": 54600
+    },
+    {
+      "epoch": 0.014289819442394557,
+      "grad_norm": 7.516880035400391,
+      "learning_rate": 7.144999152509226e-06,
+      "loss": 3.6875,
+      "step": 54800
+    },
+    {
+      "epoch": 0.014341972068096726,
+      "grad_norm": 6.853364944458008,
+      "learning_rate": 7.171075791751959e-06,
+      "loss": 3.6896,
+      "step": 55000
+    },
+    {
+      "epoch": 0.014394124693798896,
+      "grad_norm": 7.774241924285889,
+      "learning_rate": 7.1971524309946936e-06,
+      "loss": 3.6278,
+      "step": 55200
+    },
+    {
+      "epoch": 0.014446277319501067,
+      "grad_norm": 8.340727806091309,
+      "learning_rate": 7.223229070237429e-06,
+      "loss": 3.6977,
+      "step": 55400
+    },
+    {
+      "epoch": 0.014498429945203237,
+      "grad_norm": 7.890402793884277,
+      "learning_rate": 7.249305709480163e-06,
+      "loss": 3.6974,
+      "step": 55600
+    },
+    {
+      "epoch": 0.014550582570905406,
+      "grad_norm": 7.747776985168457,
+      "learning_rate": 7.275382348722897e-06,
+      "loss": 3.6873,
+      "step": 55800
+    },
+    {
+      "epoch": 0.014602735196607576,
+      "grad_norm": 6.7120184898376465,
+      "learning_rate": 7.301458987965632e-06,
+      "loss": 3.6961,
+      "step": 56000
+    },
+    {
+      "epoch": 0.014654887822309746,
+      "grad_norm": 7.8532891273498535,
+      "learning_rate": 7.327535627208366e-06,
+      "loss": 3.6477,
+      "step": 56200
+    },
+    {
+      "epoch": 0.014707040448011915,
+      "grad_norm": 8.437646865844727,
+      "learning_rate": 7.3536122664511e-06,
+      "loss": 3.6398,
+      "step": 56400
+    },
+    {
+      "epoch": 0.014759193073714085,
+      "grad_norm": 7.008194446563721,
+      "learning_rate": 7.379688905693835e-06,
+      "loss": 3.6403,
+      "step": 56600
+    },
+    {
+      "epoch": 0.014811345699416256,
+      "grad_norm": 7.821798801422119,
+      "learning_rate": 7.405765544936569e-06,
+      "loss": 3.6426,
+      "step": 56800
+    },
+    {
+      "epoch": 0.014863498325118426,
+      "grad_norm": 6.9409356117248535,
+      "learning_rate": 7.431842184179303e-06,
+      "loss": 3.6323,
+      "step": 57000
+    },
+    {
+      "epoch": 0.014915650950820595,
+      "grad_norm": 7.209104537963867,
+      "learning_rate": 7.457918823422038e-06,
+      "loss": 3.6272,
+      "step": 57200
+    },
+    {
+      "epoch": 0.014967803576522765,
+      "grad_norm": 7.187128067016602,
+      "learning_rate": 7.483995462664772e-06,
+      "loss": 3.6558,
+      "step": 57400
+    },
+    {
+      "epoch": 0.015019956202224936,
+      "grad_norm": 7.077954292297363,
+      "learning_rate": 7.510072101907506e-06,
+      "loss": 3.6594,
+      "step": 57600
+    },
+    {
+      "epoch": 0.015072108827927106,
+      "grad_norm": 7.292501926422119,
+      "learning_rate": 7.536148741150241e-06,
+      "loss": 3.6413,
+      "step": 57800
+    },
+    {
+      "epoch": 0.015124261453629275,
+      "grad_norm": 7.346433162689209,
+      "learning_rate": 7.562225380392975e-06,
+      "loss": 3.6647,
+      "step": 58000
+    },
+    {
+      "epoch": 0.015176414079331445,
+      "grad_norm": 7.485527992248535,
+      "learning_rate": 7.5883020196357106e-06,
+      "loss": 3.6067,
+      "step": 58200
+    },
+    {
+      "epoch": 0.015228566705033615,
+      "grad_norm": 7.4600629806518555,
+      "learning_rate": 7.614378658878444e-06,
+      "loss": 3.6154,
+      "step": 58400
+    },
+    {
+      "epoch": 0.015280719330735784,
+      "grad_norm": 7.584334373474121,
+      "learning_rate": 7.640455298121178e-06,
+      "loss": 3.627,
+      "step": 58600
+    },
+    {
+      "epoch": 0.015332871956437955,
+      "grad_norm": 6.661746978759766,
+      "learning_rate": 7.666531937363914e-06,
+      "loss": 3.6439,
+      "step": 58800
+    },
+    {
+      "epoch": 0.015385024582140125,
+      "grad_norm": 7.055318355560303,
+      "learning_rate": 7.692608576606647e-06,
+      "loss": 3.6274,
+      "step": 59000
+    },
+    {
+      "epoch": 0.015437177207842295,
+      "grad_norm": 7.231868267059326,
+      "learning_rate": 7.718685215849382e-06,
+      "loss": 3.6557,
+      "step": 59200
+    },
+    {
+      "epoch": 0.015489329833544464,
+      "grad_norm": 7.788670063018799,
+      "learning_rate": 7.744761855092116e-06,
+      "loss": 3.6534,
+      "step": 59400
+    },
+    {
+      "epoch": 0.015541482459246634,
+      "grad_norm": 7.5201263427734375,
+      "learning_rate": 7.770838494334851e-06,
+      "loss": 3.6055,
+      "step": 59600
+    },
+    {
+      "epoch": 0.015593635084948805,
+      "grad_norm": 6.89668083190918,
+      "learning_rate": 7.796915133577585e-06,
+      "loss": 3.5638,
+      "step": 59800
+    },
+    {
+      "epoch": 0.015645787710650973,
+      "grad_norm": 7.163186550140381,
+      "learning_rate": 7.82299177282032e-06,
+      "loss": 3.6157,
+      "step": 60000
+    },
+    {
+      "epoch": 0.015697940336353144,
+      "grad_norm": 6.937023639678955,
+      "learning_rate": 7.849068412063055e-06,
+      "loss": 3.58,
+      "step": 60200
+    },
+    {
+      "epoch": 0.015750092962055314,
+      "grad_norm": 7.33142614364624,
+      "learning_rate": 7.875145051305788e-06,
+      "loss": 3.5582,
+      "step": 60400
+    },
+    {
+      "epoch": 0.015802245587757485,
+      "grad_norm": 7.158875465393066,
+      "learning_rate": 7.901221690548524e-06,
+      "loss": 3.5748,
+      "step": 60600
+    },
+    {
+      "epoch": 0.015854398213459655,
+      "grad_norm": 7.254717826843262,
+      "learning_rate": 7.927298329791257e-06,
+      "loss": 3.5821,
+      "step": 60800
+    },
+    {
+      "epoch": 0.015906550839161825,
+      "grad_norm": 7.033998966217041,
+      "learning_rate": 7.95337496903399e-06,
+      "loss": 3.5704,
+      "step": 61000
+    },
+    {
+      "epoch": 0.015958703464863992,
+      "grad_norm": 8.059544563293457,
+      "learning_rate": 7.979451608276726e-06,
+      "loss": 3.5057,
+      "step": 61200
+    },
+    {
+      "epoch": 0.016010856090566163,
+      "grad_norm": 6.929595470428467,
+      "learning_rate": 8.005528247519461e-06,
+      "loss": 3.5442,
+      "step": 61400
+    },
+    {
+      "epoch": 0.016063008716268333,
+      "grad_norm": 7.488486289978027,
+      "learning_rate": 8.031604886762195e-06,
+      "loss": 3.5421,
+      "step": 61600
+    },
+    {
+      "epoch": 0.016115161341970503,
+      "grad_norm": 7.025427341461182,
+      "learning_rate": 8.05768152600493e-06,
+      "loss": 3.5727,
+      "step": 61800
+    },
+    {
+      "epoch": 0.016167313967672674,
+      "grad_norm": 7.745316982269287,
+      "learning_rate": 8.083758165247663e-06,
+      "loss": 3.5274,
+      "step": 62000
+    },
+    {
+      "epoch": 0.016219466593374844,
+      "grad_norm": 8.315921783447266,
+      "learning_rate": 8.109834804490397e-06,
+      "loss": 3.5662,
+      "step": 62200
+    },
+    {
+      "epoch": 0.016271619219077015,
+      "grad_norm": 7.587878704071045,
+      "learning_rate": 8.135911443733132e-06,
+      "loss": 3.5412,
+      "step": 62400
+    },
+    {
+      "epoch": 0.01632377184477918,
+      "grad_norm": 7.723850250244141,
+      "learning_rate": 8.161988082975867e-06,
+      "loss": 3.5607,
+      "step": 62600
+    },
+    {
+      "epoch": 0.016375924470481352,
+      "grad_norm": 7.842193603515625,
+      "learning_rate": 8.188064722218602e-06,
+      "loss": 3.5359,
+      "step": 62800
+    },
+    {
+      "epoch": 0.016428077096183522,
+      "grad_norm": 7.071596145629883,
+      "learning_rate": 8.214141361461336e-06,
+      "loss": 3.5514,
+      "step": 63000
+    },
+    {
+      "epoch": 0.016480229721885693,
+      "grad_norm": 7.444825649261475,
+      "learning_rate": 8.24021800070407e-06,
+      "loss": 3.5449,
+      "step": 63200
+    },
+    {
+      "epoch": 0.016532382347587863,
+      "grad_norm": 7.005112648010254,
+      "learning_rate": 8.266294639946805e-06,
+      "loss": 3.5011,
+      "step": 63400
+    },
+    {
+      "epoch": 0.016584534973290033,
+      "grad_norm": 7.4845290184021,
+      "learning_rate": 8.292371279189538e-06,
+      "loss": 3.5421,
+      "step": 63600
+    },
+    {
+      "epoch": 0.016636687598992204,
+      "grad_norm": 7.57586669921875,
+      "learning_rate": 8.318447918432273e-06,
+      "loss": 3.4857,
+      "step": 63800
+    },
+    {
+      "epoch": 0.016688840224694374,
+      "grad_norm": 6.911862850189209,
+      "learning_rate": 8.344524557675009e-06,
+      "loss": 3.5319,
+      "step": 64000
+    },
+    {
+      "epoch": 0.01674099285039654,
+      "grad_norm": 6.8444061279296875,
+      "learning_rate": 8.370601196917742e-06,
+      "loss": 3.5423,
+      "step": 64200
+    },
+    {
+      "epoch": 0.01679314547609871,
+      "grad_norm": 6.937068462371826,
+      "learning_rate": 8.396677836160476e-06,
+      "loss": 3.5273,
+      "step": 64400
+    },
+    {
+      "epoch": 0.016845298101800882,
+      "grad_norm": 7.857777118682861,
+      "learning_rate": 8.42275447540321e-06,
+      "loss": 3.539,
+      "step": 64600
+    },
+    {
+      "epoch": 0.016897450727503052,
+      "grad_norm": 6.760194778442383,
+      "learning_rate": 8.448831114645946e-06,
+      "loss": 3.545,
+      "step": 64800
+    },
+    {
+      "epoch": 0.016949603353205223,
+      "grad_norm": 7.323217868804932,
+      "learning_rate": 8.47490775388868e-06,
+      "loss": 3.554,
+      "step": 65000
+    },
+    {
+      "epoch": 0.017001755978907393,
+      "grad_norm": 7.225944519042969,
+      "learning_rate": 8.500984393131415e-06,
+      "loss": 3.5109,
+      "step": 65200
+    },
+    {
+      "epoch": 0.017053908604609563,
+      "grad_norm": 7.282871246337891,
+      "learning_rate": 8.527061032374148e-06,
+      "loss": 3.5113,
+      "step": 65400
+    },
+    {
+      "epoch": 0.01710606123031173,
+      "grad_norm": 6.699639320373535,
+      "learning_rate": 8.553137671616882e-06,
+      "loss": 3.555,
+      "step": 65600
+    },
+    {
+      "epoch": 0.0171582138560139,
+      "grad_norm": 7.353262901306152,
+      "learning_rate": 8.579214310859617e-06,
+      "loss": 3.4825,
+      "step": 65800
+    },
+    {
+      "epoch": 0.01721036648171607,
+      "grad_norm": 7.2611284255981445,
+      "learning_rate": 8.605290950102352e-06,
+      "loss": 3.5099,
+      "step": 66000
+    },
+    {
+      "epoch": 0.01726251910741824,
+      "grad_norm": 7.1428680419921875,
+      "learning_rate": 8.631367589345086e-06,
+      "loss": 3.4868,
+      "step": 66200
+    },
+    {
+      "epoch": 0.017314671733120412,
+      "grad_norm": 7.161787986755371,
+      "learning_rate": 8.65744422858782e-06,
+      "loss": 3.4979,
+      "step": 66400
+    },
+    {
+      "epoch": 0.017366824358822582,
+      "grad_norm": 7.3627142906188965,
+      "learning_rate": 8.683520867830554e-06,
+      "loss": 3.4826,
+      "step": 66600
+    },
+    {
+      "epoch": 0.017418976984524753,
+      "grad_norm": 7.277604103088379,
+      "learning_rate": 8.70959750707329e-06,
+      "loss": 3.4848,
+      "step": 66800
+    },
+    {
+      "epoch": 0.017471129610226923,
+      "grad_norm": 6.910572052001953,
+      "learning_rate": 8.735674146316023e-06,
+      "loss": 3.4787,
+      "step": 67000
+    },
+    {
+      "epoch": 0.01752328223592909,
+      "grad_norm": 7.803658962249756,
+      "learning_rate": 8.761750785558758e-06,
+      "loss": 3.4327,
+      "step": 67200
+    },
+    {
+      "epoch": 0.01757543486163126,
+      "grad_norm": 7.014376640319824,
+      "learning_rate": 8.787827424801493e-06,
+      "loss": 3.4453,
+      "step": 67400
+    },
+    {
+      "epoch": 0.01762758748733343,
+      "grad_norm": 6.735202312469482,
+      "learning_rate": 8.813904064044227e-06,
+      "loss": 3.4588,
+      "step": 67600
+    },
+    {
+      "epoch": 0.0176797401130356,
+      "grad_norm": 7.092906951904297,
+      "learning_rate": 8.83998070328696e-06,
+      "loss": 3.4441,
+      "step": 67800
+    },
+    {
+      "epoch": 0.01773189273873777,
+      "grad_norm": 7.564960956573486,
+      "learning_rate": 8.866057342529696e-06,
+      "loss": 3.4894,
+      "step": 68000
+    },
+    {
+      "epoch": 0.017784045364439942,
+      "grad_norm": 7.072451591491699,
+      "learning_rate": 8.892133981772429e-06,
+      "loss": 3.4665,
+      "step": 68200
+    },
+    {
+      "epoch": 0.017836197990142112,
+      "grad_norm": 7.669380187988281,
+      "learning_rate": 8.918210621015164e-06,
+      "loss": 3.4343,
+      "step": 68400
+    },
+    {
+      "epoch": 0.01788835061584428,
+      "grad_norm": 6.885750770568848,
+      "learning_rate": 8.9442872602579e-06,
+      "loss": 3.4742,
+      "step": 68600
+    },
+    {
+      "epoch": 0.01794050324154645,
+      "grad_norm": 6.932456016540527,
+      "learning_rate": 8.970363899500633e-06,
+      "loss": 3.5029,
+      "step": 68800
+    },
+    {
+      "epoch": 0.01799265586724862,
+      "grad_norm": 7.914639472961426,
+      "learning_rate": 8.996440538743367e-06,
+      "loss": 3.4828,
+      "step": 69000
+    },
+    {
+      "epoch": 0.01804480849295079,
+      "grad_norm": 7.070594310760498,
+      "learning_rate": 9.022517177986102e-06,
+      "loss": 3.474,
+      "step": 69200
+    },
+    {
+      "epoch": 0.01809696111865296,
+      "grad_norm": 7.355897426605225,
+      "learning_rate": 9.048593817228837e-06,
+      "loss": 3.4265,
+      "step": 69400
+    },
+    {
+      "epoch": 0.01814911374435513,
+      "grad_norm": 7.521884441375732,
+      "learning_rate": 9.07467045647157e-06,
+      "loss": 3.4737,
+      "step": 69600
+    },
+    {
+      "epoch": 0.0182012663700573,
+      "grad_norm": 7.33009672164917,
+      "learning_rate": 9.100747095714306e-06,
+      "loss": 3.4509,
+      "step": 69800
+    },
+    {
+      "epoch": 0.01825341899575947,
+      "grad_norm": 7.302325248718262,
+      "learning_rate": 9.12682373495704e-06,
+      "loss": 3.4412,
+      "step": 70000
+    },
+    {
+      "epoch": 0.01830557162146164,
+      "grad_norm": 7.756179332733154,
+      "learning_rate": 9.152900374199773e-06,
+      "loss": 3.4473,
+      "step": 70200
+    },
+    {
+      "epoch": 0.01835772424716381,
+      "grad_norm": 7.5515336990356445,
+      "learning_rate": 9.178977013442508e-06,
+      "loss": 3.4144,
+      "step": 70400
+    },
+    {
+      "epoch": 0.01840987687286598,
+      "grad_norm": 7.913027286529541,
+      "learning_rate": 9.205053652685243e-06,
+      "loss": 3.4406,
+      "step": 70600
+    },
+    {
+      "epoch": 0.01846202949856815,
+      "grad_norm": 7.048377990722656,
+      "learning_rate": 9.231130291927977e-06,
+      "loss": 3.4356,
+      "step": 70800
+    },
+    {
+      "epoch": 0.01851418212427032,
+      "grad_norm": 7.313281536102295,
+      "learning_rate": 9.257206931170712e-06,
+      "loss": 3.4147,
+      "step": 71000
+    },
+    {
+      "epoch": 0.01856633474997249,
+      "grad_norm": 7.301833629608154,
+      "learning_rate": 9.283283570413445e-06,
+      "loss": 3.4198,
+      "step": 71200
+    },
+    {
+      "epoch": 0.01861848737567466,
+      "grad_norm": 6.697605133056641,
+      "learning_rate": 9.30936020965618e-06,
+      "loss": 3.4352,
+      "step": 71400
+    },
+    {
+      "epoch": 0.018670640001376828,
+      "grad_norm": 6.958870887756348,
+      "learning_rate": 9.335436848898914e-06,
+      "loss": 3.4299,
+      "step": 71600
+    },
+    {
+      "epoch": 0.018722792627079,
+      "grad_norm": 7.359804153442383,
+      "learning_rate": 9.36151348814165e-06,
+      "loss": 3.4789,
+      "step": 71800
+    },
+    {
+      "epoch": 0.01877494525278117,
+      "grad_norm": 6.766062259674072,
+      "learning_rate": 9.387590127384384e-06,
+      "loss": 3.4272,
+      "step": 72000
+    },
+    {
+      "epoch": 0.01882709787848334,
+      "grad_norm": 7.107087135314941,
+      "learning_rate": 9.413666766627118e-06,
+      "loss": 3.3802,
+      "step": 72200
+    },
+    {
+      "epoch": 0.01887925050418551,
+      "grad_norm": 7.090783596038818,
+      "learning_rate": 9.439743405869853e-06,
+      "loss": 3.4318,
+      "step": 72400
+    },
+    {
+      "epoch": 0.01893140312988768,
+      "grad_norm": 6.893887519836426,
+      "learning_rate": 9.465820045112587e-06,
+      "loss": 3.4297,
+      "step": 72600
+    },
+    {
+      "epoch": 0.01898355575558985,
+      "grad_norm": 6.953131198883057,
+      "learning_rate": 9.49189668435532e-06,
+      "loss": 3.4238,
+      "step": 72800
+    },
+    {
+      "epoch": 0.019035708381292017,
+      "grad_norm": 6.934586524963379,
+      "learning_rate": 9.517973323598055e-06,
+      "loss": 3.4275,
+      "step": 73000
+    },
+    {
+      "epoch": 0.019087861006994188,
+      "grad_norm": 7.763511657714844,
+      "learning_rate": 9.54404996284079e-06,
+      "loss": 3.3778,
+      "step": 73200
+    },
+    {
+      "epoch": 0.019140013632696358,
+      "grad_norm": 6.7908806800842285,
+      "learning_rate": 9.570126602083524e-06,
+      "loss": 3.4133,
+      "step": 73400
+    },
+    {
+      "epoch": 0.01919216625839853,
+      "grad_norm": 6.870711326599121,
+      "learning_rate": 9.59620324132626e-06,
+      "loss": 3.4281,
+      "step": 73600
+    },
+    {
+      "epoch": 0.0192443188841007,
+      "grad_norm": 7.813330173492432,
+      "learning_rate": 9.622279880568993e-06,
+      "loss": 3.424,
+      "step": 73800
+    },
+    {
+      "epoch": 0.01929647150980287,
+      "grad_norm": 7.28371524810791,
+      "learning_rate": 9.648356519811728e-06,
+      "loss": 3.4238,
+      "step": 74000
+    },
+    {
+      "epoch": 0.01934862413550504,
+      "grad_norm": 7.129782676696777,
+      "learning_rate": 9.674433159054461e-06,
+      "loss": 3.4407,
+      "step": 74200
+    },
+    {
+      "epoch": 0.019400776761207206,
+      "grad_norm": 7.057884216308594,
+      "learning_rate": 9.700509798297197e-06,
+      "loss": 3.3992,
+      "step": 74400
+    },
+    {
+      "epoch": 0.019452929386909377,
+      "grad_norm": 7.445634841918945,
+      "learning_rate": 9.726586437539932e-06,
+      "loss": 3.4258,
+      "step": 74600
+    },
+    {
+      "epoch": 0.019505082012611547,
+      "grad_norm": 7.498905658721924,
+      "learning_rate": 9.752663076782665e-06,
+      "loss": 3.4097,
+      "step": 74800
+    },
+    {
+      "epoch": 0.019557234638313718,
+      "grad_norm": 8.201305389404297,
+      "learning_rate": 9.778739716025399e-06,
+      "loss": 3.3639,
+      "step": 75000
+    },
+    {
+      "epoch": 0.019609387264015888,
+      "grad_norm": 7.614349365234375,
+      "learning_rate": 9.804816355268134e-06,
+      "loss": 3.3741,
+      "step": 75200
+    },
+    {
+      "epoch": 0.01966153988971806,
+      "grad_norm": 6.917168140411377,
+      "learning_rate": 9.830892994510868e-06,
+      "loss": 3.4018,
+      "step": 75400
+    },
+    {
+      "epoch": 0.01971369251542023,
+      "grad_norm": 6.967098236083984,
+      "learning_rate": 9.856969633753603e-06,
+      "loss": 3.3677,
+      "step": 75600
+    },
+    {
+      "epoch": 0.0197658451411224,
+      "grad_norm": 7.0407795906066895,
+      "learning_rate": 9.883046272996338e-06,
+      "loss": 3.4018,
+      "step": 75800
+    },
+    {
+      "epoch": 0.019817997766824566,
+      "grad_norm": 6.73028564453125,
+      "learning_rate": 9.909122912239071e-06,
+      "loss": 3.3787,
+      "step": 76000
+    },
+    {
+      "epoch": 0.019870150392526736,
+      "grad_norm": 6.180406093597412,
+      "learning_rate": 9.935199551481805e-06,
+      "loss": 3.3538,
+      "step": 76200
+    },
+    {
+      "epoch": 0.019922303018228907,
+      "grad_norm": 7.133777618408203,
+      "learning_rate": 9.96127619072454e-06,
+      "loss": 3.3845,
+      "step": 76400
+    },
+    {
+      "epoch": 0.019974455643931077,
+      "grad_norm": 6.977362632751465,
+      "learning_rate": 9.987352829967275e-06,
+      "loss": 3.3386,
+      "step": 76600
+    },
+    {
+      "epoch": 0.020026608269633248,
+      "grad_norm": 7.869418621063232,
+      "learning_rate": 9.999999981466639e-06,
+      "loss": 3.355,
+      "step": 76800
+    },
+    {
+      "epoch": 0.020078760895335418,
+      "grad_norm": 6.740837574005127,
+      "learning_rate": 9.999999839614538e-06,
+      "loss": 3.4371,
+      "step": 77000
+    },
+    {
+      "epoch": 0.02013091352103759,
+      "grad_norm": 7.533843994140625,
+      "learning_rate": 9.999999558006677e-06,
+      "loss": 3.3576,
+      "step": 77200
+    },
+    {
+      "epoch": 0.020183066146739755,
+      "grad_norm": 6.733776092529297,
+      "learning_rate": 9.999999136643062e-06,
+      "loss": 3.3787,
+      "step": 77400
+    },
+    {
+      "epoch": 0.020235218772441926,
+      "grad_norm": 7.021015644073486,
+      "learning_rate": 9.999998575523706e-06,
+      "loss": 3.3798,
+      "step": 77600
+    },
+    {
+      "epoch": 0.020287371398144096,
+      "grad_norm": 7.0405755043029785,
+      "learning_rate": 9.999997874648624e-06,
+      "loss": 3.3451,
+      "step": 77800
+    },
+    {
+      "epoch": 0.020339524023846266,
+      "grad_norm": 7.242607116699219,
+      "learning_rate": 9.999997034017837e-06,
+      "loss": 3.3643,
+      "step": 78000
+    },
+    {
+      "epoch": 0.020391676649548437,
+      "grad_norm": 7.021972179412842,
+      "learning_rate": 9.999996053631368e-06,
+      "loss": 3.3861,
+      "step": 78200
+    },
+    {
+      "epoch": 0.020443829275250607,
+      "grad_norm": 7.320739269256592,
+      "learning_rate": 9.999994933489244e-06,
+      "loss": 3.3503,
+      "step": 78400
+    },
+    {
+      "epoch": 0.020495981900952778,
+      "grad_norm": 7.283355712890625,
+      "learning_rate": 9.999993673591494e-06,
+      "loss": 3.393,
+      "step": 78600
+    },
+    {
+      "epoch": 0.020548134526654948,
+      "grad_norm": 7.413722515106201,
+      "learning_rate": 9.999992273938159e-06,
+      "loss": 3.3393,
+      "step": 78800
+    },
+    {
+      "epoch": 0.020600287152357115,
+      "grad_norm": 7.285412788391113,
+      "learning_rate": 9.999990734529274e-06,
+      "loss": 3.3531,
+      "step": 79000
+    },
+    {
+      "epoch": 0.020652439778059285,
+      "grad_norm": 6.342610836029053,
+      "learning_rate": 9.999989055364881e-06,
+      "loss": 3.3199,
+      "step": 79200
+    },
+    {
+      "epoch": 0.020704592403761456,
+      "grad_norm": 7.5218892097473145,
+      "learning_rate": 9.999987236445031e-06,
+      "loss": 3.3587,
+      "step": 79400
+    },
+    {
+      "epoch": 0.020756745029463626,
+      "grad_norm": 6.564230918884277,
+      "learning_rate": 9.99998527776977e-06,
+      "loss": 3.3877,
+      "step": 79600
+    },
+    {
+      "epoch": 0.020808897655165796,
+      "grad_norm": 6.8725786209106445,
+      "learning_rate": 9.999983179339158e-06,
+      "loss": 3.3915,
+      "step": 79800
+    },
+    {
+      "epoch": 0.020861050280867967,
+      "grad_norm": 7.533461093902588,
+      "learning_rate": 9.999980941153249e-06,
+      "loss": 3.3572,
+      "step": 80000
+    },
+    {
+      "epoch": 0.020913202906570137,
+      "grad_norm": 7.089059829711914,
+      "learning_rate": 9.999978563212107e-06,
+      "loss": 3.3238,
+      "step": 80200
+    },
+    {
+      "epoch": 0.020965355532272304,
+      "grad_norm": 7.241924285888672,
+      "learning_rate": 9.999976045515802e-06,
+      "loss": 3.3311,
+      "step": 80400
+    },
+    {
+      "epoch": 0.021017508157974474,
+      "grad_norm": 6.942836761474609,
+      "learning_rate": 9.9999733880644e-06,
+      "loss": 3.3374,
+      "step": 80600
+    },
+    {
+      "epoch": 0.021069660783676645,
+      "grad_norm": 7.0181884765625,
+      "learning_rate": 9.999970590857975e-06,
+      "loss": 3.3009,
+      "step": 80800
+    },
+    {
+      "epoch": 0.021121813409378815,
+      "grad_norm": 7.321265697479248,
+      "learning_rate": 9.999967653896607e-06,
+      "loss": 3.3696,
+      "step": 81000
+    },
+    {
+      "epoch": 0.021173966035080986,
+      "grad_norm": 6.867557525634766,
+      "learning_rate": 9.999964577180379e-06,
+      "loss": 3.3183,
+      "step": 81200
+    },
+    {
+      "epoch": 0.021226118660783156,
+      "grad_norm": 7.854918956756592,
+      "learning_rate": 9.999961360709376e-06,
+      "loss": 3.3517,
+      "step": 81400
+    },
+    {
+      "epoch": 0.021278271286485326,
+      "grad_norm": 7.4617438316345215,
+      "learning_rate": 9.999958004483687e-06,
+      "loss": 3.3323,
+      "step": 81600
+    },
+    {
+      "epoch": 0.021330423912187493,
+      "grad_norm": 7.94450569152832,
+      "learning_rate": 9.999954508503407e-06,
+      "loss": 3.3163,
+      "step": 81800
+    },
+    {
+      "epoch": 0.021382576537889664,
+      "grad_norm": 6.6632184982299805,
+      "learning_rate": 9.999950872768633e-06,
+      "loss": 3.3429,
+      "step": 82000
+    },
+    {
+      "epoch": 0.021434729163591834,
+      "grad_norm": 6.834831237792969,
+      "learning_rate": 9.999947097279468e-06,
+      "loss": 3.3278,
+      "step": 82200
+    },
+    {
+      "epoch": 0.021486881789294004,
+      "grad_norm": 7.85338020324707,
+      "learning_rate": 9.999943182036017e-06,
+      "loss": 3.3649,
+      "step": 82400
+    },
+    {
+      "epoch": 0.021539034414996175,
+      "grad_norm": 7.1437811851501465,
+      "learning_rate": 9.999939127038387e-06,
+      "loss": 3.3501,
+      "step": 82600
+    },
+    {
+      "epoch": 0.021591187040698345,
+      "grad_norm": 7.180891990661621,
+      "learning_rate": 9.999934932286692e-06,
+      "loss": 3.3236,
+      "step": 82800
+    },
+    {
+      "epoch": 0.021643339666400516,
+      "grad_norm": 6.1819353103637695,
+      "learning_rate": 9.999930597781054e-06,
+      "loss": 3.2915,
+      "step": 83000
+    },
+    {
+      "epoch": 0.021695492292102686,
+      "grad_norm": 6.358526229858398,
+      "learning_rate": 9.999926123521588e-06,
+      "loss": 3.2759,
+      "step": 83200
+    },
+    {
+      "epoch": 0.021747644917804853,
+      "grad_norm": 7.643764495849609,
+      "learning_rate": 9.999921509508424e-06,
+      "loss": 3.3382,
+      "step": 83400
+    },
+    {
+      "epoch": 0.021799797543507023,
+      "grad_norm": 6.65496826171875,
+      "learning_rate": 9.999916755741687e-06,
+      "loss": 3.3042,
+      "step": 83600
+    },
+    {
+      "epoch": 0.021851950169209194,
+      "grad_norm": 6.801998138427734,
+      "learning_rate": 9.999911862221512e-06,
+      "loss": 3.3193,
+      "step": 83800
+    },
+    {
+      "epoch": 0.021904102794911364,
+      "grad_norm": 7.052958011627197,
+      "learning_rate": 9.999906828948035e-06,
+      "loss": 3.2707,
+      "step": 84000
+    },
+    {
+      "epoch": 0.021956255420613534,
+      "grad_norm": 6.636064529418945,
+      "learning_rate": 9.999901655921398e-06,
+      "loss": 3.317,
+      "step": 84200
+    },
+    {
+      "epoch": 0.022008408046315705,
+      "grad_norm": 6.989995956420898,
+      "learning_rate": 9.999896343141742e-06,
+      "loss": 3.3297,
+      "step": 84400
+    },
+    {
+      "epoch": 0.022060560672017875,
+      "grad_norm": 6.694401741027832,
+      "learning_rate": 9.999890890609221e-06,
+      "loss": 3.2781,
+      "step": 84600
+    },
+    {
+      "epoch": 0.022112713297720042,
+      "grad_norm": 6.958297252655029,
+      "learning_rate": 9.999885298323984e-06,
+      "loss": 3.3058,
+      "step": 84800
+    },
+    {
+      "epoch": 0.022164865923422213,
+      "grad_norm": 7.147956371307373,
+      "learning_rate": 9.999879566286187e-06,
+      "loss": 3.3504,
+      "step": 85000
+    },
+    {
+      "epoch": 0.022217018549124383,
+      "grad_norm": 7.238255023956299,
+      "learning_rate": 9.999873694495991e-06,
+      "loss": 3.3022,
+      "step": 85200
+    },
+    {
+      "epoch": 0.022269171174826553,
+      "grad_norm": 7.0600056648254395,
+      "learning_rate": 9.999867682953562e-06,
+      "loss": 3.2688,
+      "step": 85400
+    },
+    {
+      "epoch": 0.022321323800528724,
+      "grad_norm": 6.889624118804932,
+      "learning_rate": 9.999861531659063e-06,
+      "loss": 3.2854,
+      "step": 85600
+    },
+    {
+      "epoch": 0.022373476426230894,
+      "grad_norm": 6.706833362579346,
+      "learning_rate": 9.99985524061267e-06,
+      "loss": 3.3073,
+      "step": 85800
+    },
+    {
+      "epoch": 0.022425629051933064,
+      "grad_norm": 6.433074951171875,
+      "learning_rate": 9.99984880981456e-06,
+      "loss": 3.3223,
+      "step": 86000
+    },
+    {
+      "epoch": 0.02247778167763523,
+      "grad_norm": 6.203549385070801,
+      "learning_rate": 9.99984223926491e-06,
+      "loss": 3.324,
+      "step": 86200
+    },
+    {
+      "epoch": 0.022529934303337402,
+      "grad_norm": 6.171205043792725,
+      "learning_rate": 9.999835528963905e-06,
+      "loss": 3.2961,
+      "step": 86400
+    },
+    {
+      "epoch": 0.022582086929039572,
+      "grad_norm": 7.106544017791748,
+      "learning_rate": 9.999828678911729e-06,
+      "loss": 3.2777,
+      "step": 86600
+    },
+    {
+      "epoch": 0.022634239554741743,
+      "grad_norm": 6.84821081161499,
+      "learning_rate": 9.99982168910858e-06,
+      "loss": 3.3162,
+      "step": 86800
+    },
+    {
+      "epoch": 0.022686392180443913,
+      "grad_norm": 7.026462078094482,
+      "learning_rate": 9.999814559554648e-06,
+      "loss": 3.2632,
+      "step": 87000
+    },
+    {
+      "epoch": 0.022738544806146083,
+      "grad_norm": 6.290987014770508,
+      "learning_rate": 9.999807290250133e-06,
+      "loss": 3.2957,
+      "step": 87200
+    },
+    {
+      "epoch": 0.022790697431848254,
+      "grad_norm": 6.156450271606445,
+      "learning_rate": 9.99979988119524e-06,
+      "loss": 3.2679,
+      "step": 87400
+    },
+    {
+      "epoch": 0.022842850057550424,
+      "grad_norm": 6.672393321990967,
+      "learning_rate": 9.999792332390177e-06,
+      "loss": 3.2777,
+      "step": 87600
+    },
+    {
+      "epoch": 0.02289500268325259,
+      "grad_norm": 6.625957489013672,
+      "learning_rate": 9.99978464383515e-06,
+      "loss": 3.2897,
+      "step": 87800
+    },
+    {
+      "epoch": 0.02294715530895476,
+      "grad_norm": 6.969167232513428,
+      "learning_rate": 9.99977681553038e-06,
+      "loss": 3.2615,
+      "step": 88000
+    },
+    {
+      "epoch": 0.022999307934656932,
+      "grad_norm": 6.787604331970215,
+      "learning_rate": 9.999768847476084e-06,
+      "loss": 3.2896,
+      "step": 88200
+    },
+    {
+      "epoch": 0.023051460560359102,
+      "grad_norm": 7.575397491455078,
+      "learning_rate": 9.999760739672481e-06,
+      "loss": 3.2926,
+      "step": 88400
+    },
+    {
+      "epoch": 0.023103613186061273,
+      "grad_norm": 6.810946941375732,
+      "learning_rate": 9.9997524921198e-06,
+      "loss": 3.2828,
+      "step": 88600
+    },
+    {
+      "epoch": 0.023155765811763443,
+      "grad_norm": 6.538626194000244,
+      "learning_rate": 9.999744104818275e-06,
+      "loss": 3.2672,
+      "step": 88800
+    },
+    {
+      "epoch": 0.023207918437465613,
+      "grad_norm": 6.649213790893555,
+      "learning_rate": 9.999735577768135e-06,
+      "loss": 3.3231,
+      "step": 89000
+    },
+    {
+      "epoch": 0.02326007106316778,
+      "grad_norm": 6.549954414367676,
+      "learning_rate": 9.999726910969621e-06,
+      "loss": 3.2322,
+      "step": 89200
+    },
+    {
+      "epoch": 0.02331222368886995,
+      "grad_norm": 6.339663982391357,
+      "learning_rate": 9.999718104422977e-06,
+      "loss": 3.2849,
+      "step": 89400
+    },
+    {
+      "epoch": 0.02336437631457212,
+      "grad_norm": 7.996878147125244,
+      "learning_rate": 9.999709158128444e-06,
+      "loss": 3.2478,
+      "step": 89600
+    },
+    {
+      "epoch": 0.02341652894027429,
+      "grad_norm": 6.590426921844482,
+      "learning_rate": 9.999700072086277e-06,
+      "loss": 3.257,
+      "step": 89800
+    },
+    {
+      "epoch": 0.023468681565976462,
+      "grad_norm": 6.166894435882568,
+      "learning_rate": 9.999690846296728e-06,
+      "loss": 3.266,
+      "step": 90000
+    },
+    {
+      "epoch": 0.023520834191678632,
+      "grad_norm": 6.546900749206543,
+      "learning_rate": 9.999681480760054e-06,
+      "loss": 3.3065,
+      "step": 90200
+    },
+    {
+      "epoch": 0.023572986817380803,
+      "grad_norm": 7.30242919921875,
+      "learning_rate": 9.99967197547652e-06,
+      "loss": 3.2692,
+      "step": 90400
+    },
+    {
+      "epoch": 0.02362513944308297,
+      "grad_norm": 6.420165061950684,
+      "learning_rate": 9.999662330446387e-06,
+      "loss": 3.2749,
+      "step": 90600
+    },
+    {
+      "epoch": 0.02367729206878514,
+      "grad_norm": 6.860097408294678,
+      "learning_rate": 9.999652545669926e-06,
+      "loss": 3.2887,
+      "step": 90800
+    },
+    {
+      "epoch": 0.02372944469448731,
+      "grad_norm": 6.4746599197387695,
+      "learning_rate": 9.999642621147414e-06,
+      "loss": 3.2687,
+      "step": 91000
+    },
+    {
+      "epoch": 0.02378159732018948,
+      "grad_norm": 6.279762268066406,
+      "learning_rate": 9.999632556879127e-06,
+      "loss": 3.2578,
+      "step": 91200
+    },
+    {
+      "epoch": 0.02383374994589165,
+      "grad_norm": 6.2368268966674805,
+      "learning_rate": 9.999622352865342e-06,
+      "loss": 3.2125,
+      "step": 91400
+    },
+    {
+      "epoch": 0.02388590257159382,
+      "grad_norm": 6.1834869384765625,
+      "learning_rate": 9.999612009106349e-06,
+      "loss": 3.2436,
+      "step": 91600
+    },
+    {
+      "epoch": 0.023938055197295992,
+      "grad_norm": 7.046905517578125,
+      "learning_rate": 9.999601525602436e-06,
+      "loss": 3.2105,
+      "step": 91800
+    },
+    {
+      "epoch": 0.023990207822998162,
+      "grad_norm": 6.951870918273926,
+      "learning_rate": 9.999590902353895e-06,
+      "loss": 3.2179,
+      "step": 92000
+    },
+    {
+      "epoch": 0.02404236044870033,
+      "grad_norm": 6.480534553527832,
+      "learning_rate": 9.999580139361023e-06,
+      "loss": 3.2562,
+      "step": 92200
+    },
+    {
+      "epoch": 0.0240945130744025,
+      "grad_norm": 6.898007869720459,
+      "learning_rate": 9.999569236624122e-06,
+      "loss": 3.2618,
+      "step": 92400
+    },
+    {
+      "epoch": 0.02414666570010467,
+      "grad_norm": 7.365018367767334,
+      "learning_rate": 9.999558194143497e-06,
+      "loss": 3.2709,
+      "step": 92600
+    },
+    {
+      "epoch": 0.02419881832580684,
+      "grad_norm": 6.268816947937012,
+      "learning_rate": 9.999547011919454e-06,
+      "loss": 3.2585,
+      "step": 92800
+    },
+    {
+      "epoch": 0.02425097095150901,
+      "grad_norm": 7.152459621429443,
+      "learning_rate": 9.999535689952309e-06,
+      "loss": 3.2336,
+      "step": 93000
+    },
+    {
+      "epoch": 0.02430312357721118,
+      "grad_norm": 7.479735851287842,
+      "learning_rate": 9.999524228242376e-06,
+      "loss": 3.2201,
+      "step": 93200
+    },
+    {
+      "epoch": 0.02435527620291335,
+      "grad_norm": 7.143455982208252,
+      "learning_rate": 9.999512626789977e-06,
+      "loss": 3.248,
+      "step": 93400
+    },
+    {
+      "epoch": 0.02440742882861552,
+      "grad_norm": 6.253408432006836,
+      "learning_rate": 9.999500885595435e-06,
+      "loss": 3.2142,
+      "step": 93600
+    },
+    {
+      "epoch": 0.02445958145431769,
+      "grad_norm": 7.234580039978027,
+      "learning_rate": 9.999489004659077e-06,
+      "loss": 3.2751,
+      "step": 93800
+    },
+    {
+      "epoch": 0.02451173408001986,
+      "grad_norm": 6.902135372161865,
+      "learning_rate": 9.999476983981238e-06,
+      "loss": 3.2603,
+      "step": 94000
+    },
+    {
+      "epoch": 0.02456388670572203,
+      "grad_norm": 6.845031261444092,
+      "learning_rate": 9.999464823562253e-06,
+      "loss": 3.2149,
+      "step": 94200
+    },
+    {
+      "epoch": 0.0246160393314242,
+      "grad_norm": 6.742398738861084,
+      "learning_rate": 9.999452523402461e-06,
+      "loss": 3.2196,
+      "step": 94400
+    },
+    {
+      "epoch": 0.02466819195712637,
+      "grad_norm": 7.369746208190918,
+      "learning_rate": 9.999440083502206e-06,
+      "loss": 3.2394,
+      "step": 94600
+    },
+    {
+      "epoch": 0.02472034458282854,
+      "grad_norm": 7.084136962890625,
+      "learning_rate": 9.999427503861836e-06,
+      "loss": 3.1992,
+      "step": 94800
+    },
+    {
+      "epoch": 0.02477249720853071,
+      "grad_norm": 6.0851054191589355,
+      "learning_rate": 9.999414784481705e-06,
+      "loss": 3.2615,
+      "step": 95000
+    },
+    {
+      "epoch": 0.024824649834232878,
+      "grad_norm": 6.620251178741455,
+      "learning_rate": 9.999401925362164e-06,
+      "loss": 3.2164,
+      "step": 95200
+    },
+    {
+      "epoch": 0.02487680245993505,
+      "grad_norm": 6.744817733764648,
+      "learning_rate": 9.999388926503576e-06,
+      "loss": 3.247,
+      "step": 95400
+    },
+    {
+      "epoch": 0.02492895508563722,
+      "grad_norm": 7.277287006378174,
+      "learning_rate": 9.999375787906301e-06,
+      "loss": 3.2377,
+      "step": 95600
+    },
+    {
+      "epoch": 0.02498110771133939,
+      "grad_norm": 6.504336357116699,
+      "learning_rate": 9.999362509570709e-06,
+      "loss": 3.2138,
+      "step": 95800
+    },
+    {
+      "epoch": 0.02503326033704156,
+      "grad_norm": 6.75746488571167,
+      "learning_rate": 9.999349091497173e-06,
+      "loss": 3.2087,
+      "step": 96000
+    },
+    {
+      "epoch": 0.02508541296274373,
+      "grad_norm": 6.542148590087891,
+      "learning_rate": 9.999335533686061e-06,
+      "loss": 3.1869,
+      "step": 96200
+    },
+    {
+      "epoch": 0.0251375655884459,
+      "grad_norm": 6.434966087341309,
+      "learning_rate": 9.999321836137759e-06,
+      "loss": 3.2454,
+      "step": 96400
+    },
+    {
+      "epoch": 0.025189718214148067,
+      "grad_norm": 6.355975151062012,
+      "learning_rate": 9.999307998852648e-06,
+      "loss": 3.2172,
+      "step": 96600
+    },
+    {
+      "epoch": 0.025241870839850238,
+      "grad_norm": 6.389385223388672,
+      "learning_rate": 9.999294021831112e-06,
+      "loss": 3.238,
+      "step": 96800
+    },
+    {
+      "epoch": 0.025294023465552408,
+      "grad_norm": 6.235101222991943,
+      "learning_rate": 9.999279905073544e-06,
+      "loss": 3.2511,
+      "step": 97000
+    },
+    {
+      "epoch": 0.02534617609125458,
+      "grad_norm": 7.004864692687988,
+      "learning_rate": 9.99926564858034e-06,
+      "loss": 3.1807,
+      "step": 97200
+    },
+    {
+      "epoch": 0.02539832871695675,
+      "grad_norm": 6.632132530212402,
+      "learning_rate": 9.999251252351896e-06,
+      "loss": 3.2,
+      "step": 97400
+    },
+    {
+      "epoch": 0.02545048134265892,
+      "grad_norm": 7.012777328491211,
+      "learning_rate": 9.999236716388614e-06,
+      "loss": 3.2303,
+      "step": 97600
+    },
+    {
+      "epoch": 0.02550263396836109,
+      "grad_norm": 7.067131519317627,
+      "learning_rate": 9.999222040690901e-06,
+      "loss": 3.1878,
+      "step": 97800
+    },
+    {
+      "epoch": 0.025554786594063256,
+      "grad_norm": 6.799013137817383,
+      "learning_rate": 9.99920722525917e-06,
+      "loss": 3.1748,
+      "step": 98000
+    },
+    {
+      "epoch": 0.025606939219765427,
+      "grad_norm": 6.825366497039795,
+      "learning_rate": 9.999192270093832e-06,
+      "loss": 3.2289,
+      "step": 98200
+    },
+    {
+      "epoch": 0.025659091845467597,
+      "grad_norm": 7.0657782554626465,
+      "learning_rate": 9.999177175195305e-06,
+      "loss": 3.2448,
+      "step": 98400
+    },
+    {
+      "epoch": 0.025711244471169768,
+      "grad_norm": 6.592565059661865,
+      "learning_rate": 9.99916194056401e-06,
+      "loss": 3.2417,
+      "step": 98600
+    },
+    {
+      "epoch": 0.025763397096871938,
+      "grad_norm": 6.476219177246094,
+      "learning_rate": 9.999146566200378e-06,
+      "loss": 3.1744,
+      "step": 98800
+    },
+    {
+      "epoch": 0.02581554972257411,
+      "grad_norm": 6.580258846282959,
+      "learning_rate": 9.999131052104834e-06,
+      "loss": 3.2692,
+      "step": 99000
+    },
+    {
+      "epoch": 0.02586770234827628,
+      "grad_norm": 6.556769371032715,
+      "learning_rate": 9.999115398277812e-06,
+      "loss": 3.189,
+      "step": 99200
+    },
+    {
+      "epoch": 0.02591985497397845,
+      "grad_norm": 7.036558151245117,
+      "learning_rate": 9.999099604719751e-06,
+      "loss": 3.2138,
+      "step": 99400
+    },
+    {
+      "epoch": 0.025972007599680616,
+      "grad_norm": 6.926954746246338,
+      "learning_rate": 9.999083671431092e-06,
+      "loss": 3.2066,
+      "step": 99600
+    },
+    {
+      "epoch": 0.026024160225382786,
+      "grad_norm": 6.171466827392578,
+      "learning_rate": 9.999067598412279e-06,
+      "loss": 3.1867,
+      "step": 99800
+    },
+    {
+      "epoch": 0.026076312851084957,
+      "grad_norm": 6.851698398590088,
+      "learning_rate": 9.999051385663765e-06,
+      "loss": 3.2022,
+      "step": 100000
+    },
+    {
+      "epoch": 0.026128465476787127,
+      "grad_norm": 6.199542045593262,
+      "learning_rate": 9.999035033185998e-06,
+      "loss": 3.205,
+      "step": 100200
+    },
+    {
+      "epoch": 0.026180618102489298,
+      "grad_norm": 6.503675937652588,
+      "learning_rate": 9.99901854097944e-06,
+      "loss": 3.2172,
+      "step": 100400
+    },
+    {
+      "epoch": 0.026232770728191468,
+      "grad_norm": 6.505238056182861,
+      "learning_rate": 9.999001909044548e-06,
+      "loss": 3.2047,
+      "step": 100600
+    },
+    {
+      "epoch": 0.02628492335389364,
+      "grad_norm": 6.692224979400635,
+      "learning_rate": 9.99898513738179e-06,
+      "loss": 3.1518,
+      "step": 100800
+    },
+    {
+      "epoch": 0.026337075979595805,
+      "grad_norm": 6.675695419311523,
+      "learning_rate": 9.998968225991632e-06,
+      "loss": 3.1623,
+      "step": 101000
+    },
+    {
+      "epoch": 0.026389228605297976,
+      "grad_norm": 6.617093086242676,
+      "learning_rate": 9.998951174874548e-06,
+      "loss": 3.2322,
+      "step": 101200
+    },
+    {
+      "epoch": 0.026441381231000146,
+      "grad_norm": 6.3754119873046875,
+      "learning_rate": 9.998933984031016e-06,
+      "loss": 3.2037,
+      "step": 101400
+    },
+    {
+      "epoch": 0.026493533856702316,
+      "grad_norm": 6.500361442565918,
+      "learning_rate": 9.998916653461515e-06,
+      "loss": 3.2131,
+      "step": 101600
+    },
+    {
+      "epoch": 0.026545686482404487,
+      "grad_norm": 6.8905415534973145,
+      "learning_rate": 9.998899183166529e-06,
+      "loss": 3.1613,
+      "step": 101800
+    },
+    {
+      "epoch": 0.026597839108106657,
+      "grad_norm": 6.983830451965332,
+      "learning_rate": 9.998881573146546e-06,
+      "loss": 3.1294,
+      "step": 102000
+    },
+    {
+      "epoch": 0.026649991733808828,
+      "grad_norm": 6.894038677215576,
+      "learning_rate": 9.99886382340206e-06,
+      "loss": 3.2047,
+      "step": 102200
+    },
+    {
+      "epoch": 0.026702144359510994,
+      "grad_norm": 6.187081336975098,
+      "learning_rate": 9.998845933933565e-06,
+      "loss": 3.1967,
+      "step": 102400
+    },
+    {
+      "epoch": 0.026754296985213165,
+      "grad_norm": 7.0318708419799805,
+      "learning_rate": 9.998827904741563e-06,
+      "loss": 3.2053,
+      "step": 102600
+    },
+    {
+      "epoch": 0.026806449610915335,
+      "grad_norm": 6.508965969085693,
+      "learning_rate": 9.998809735826559e-06,
+      "loss": 3.1671,
+      "step": 102800
+    },
+    {
+      "epoch": 0.026858602236617506,
+      "grad_norm": 6.52016019821167,
+      "learning_rate": 9.998791427189058e-06,
+      "loss": 3.1954,
+      "step": 103000
+    },
+    {
+      "epoch": 0.026910754862319676,
+      "grad_norm": 6.182394981384277,
+      "learning_rate": 9.998772978829571e-06,
+      "loss": 3.1369,
+      "step": 103200
+    },
+    {
+      "epoch": 0.026962907488021846,
+      "grad_norm": 6.727584362030029,
+      "learning_rate": 9.998754390748617e-06,
+      "loss": 3.1729,
+      "step": 103400
+    },
+    {
+      "epoch": 0.027015060113724017,
+      "grad_norm": 6.066624641418457,
+      "learning_rate": 9.998735662946715e-06,
+      "loss": 3.1742,
+      "step": 103600
+    },
+    {
+      "epoch": 0.027067212739426187,
+      "grad_norm": 6.274039268493652,
+      "learning_rate": 9.998716795424385e-06,
+      "loss": 3.1548,
+      "step": 103800
+    },
+    {
+      "epoch": 0.027119365365128354,
+      "grad_norm": 6.5091352462768555,
+      "learning_rate": 9.998697788182158e-06,
+      "loss": 3.1464,
+      "step": 104000
+    },
+    {
+      "epoch": 0.027171517990830524,
+      "grad_norm": 7.099887847900391,
+      "learning_rate": 9.998678641220564e-06,
+      "loss": 3.1755,
+      "step": 104200
+    },
+    {
+      "epoch": 0.027223670616532695,
+      "grad_norm": 6.529077529907227,
+      "learning_rate": 9.99865935454014e-06,
+      "loss": 3.1749,
+      "step": 104400
+    },
+    {
+      "epoch": 0.027275823242234865,
+      "grad_norm": 6.756040573120117,
+      "learning_rate": 9.998639928141422e-06,
+      "loss": 3.192,
+      "step": 104600
+    },
+    {
+      "epoch": 0.027327975867937036,
+      "grad_norm": 6.822609901428223,
+      "learning_rate": 9.998620362024954e-06,
+      "loss": 3.2029,
+      "step": 104800
+    },
+    {
+      "epoch": 0.027380128493639206,
+      "grad_norm": 6.252408504486084,
+      "learning_rate": 9.998600656191284e-06,
+      "loss": 3.1596,
+      "step": 105000
+    },
+    {
+      "epoch": 0.027432281119341376,
+      "grad_norm": 6.114434719085693,
+      "learning_rate": 9.99858081064096e-06,
+      "loss": 3.2181,
+      "step": 105200
+    },
+    {
+      "epoch": 0.027484433745043543,
+      "grad_norm": 6.77893590927124,
+      "learning_rate": 9.998560825374542e-06,
+      "loss": 3.1836,
+      "step": 105400
+    },
+    {
+      "epoch": 0.027536586370745714,
+      "grad_norm": 5.703882217407227,
+      "learning_rate": 9.998540700392583e-06,
+      "loss": 3.1661,
+      "step": 105600
+    },
+    {
+      "epoch": 0.027588738996447884,
+      "grad_norm": 6.503817558288574,
+      "learning_rate": 9.99852043569565e-06,
+      "loss": 3.1212,
+      "step": 105800
+    },
+    {
+      "epoch": 0.027640891622150054,
+      "grad_norm": 6.397846221923828,
+      "learning_rate": 9.998500031284304e-06,
+      "loss": 3.1341,
+      "step": 106000
+    },
+    {
+      "epoch": 0.027693044247852225,
+      "grad_norm": 6.377574443817139,
+      "learning_rate": 9.998479487159121e-06,
+      "loss": 3.1652,
+      "step": 106200
+    },
+    {
+      "epoch": 0.027745196873554395,
+      "grad_norm": 6.59285831451416,
+      "learning_rate": 9.998458803320671e-06,
+      "loss": 3.1326,
+      "step": 106400
+    },
+    {
+      "epoch": 0.027797349499256566,
+      "grad_norm": 5.924184322357178,
+      "learning_rate": 9.998437979769536e-06,
+      "loss": 3.1707,
+      "step": 106600
+    },
+    {
+      "epoch": 0.027849502124958733,
+      "grad_norm": 6.608452320098877,
+      "learning_rate": 9.998417016506296e-06,
+      "loss": 3.176,
+      "step": 106800
+    },
+    {
+      "epoch": 0.027901654750660903,
+      "grad_norm": 7.678200721740723,
+      "learning_rate": 9.998395913531536e-06,
+      "loss": 3.1778,
+      "step": 107000
+    },
+    {
+      "epoch": 0.027953807376363073,
+      "grad_norm": 7.167779922485352,
+      "learning_rate": 9.998374670845847e-06,
+      "loss": 3.181,
+      "step": 107200
+    },
+    {
+      "epoch": 0.028005960002065244,
+      "grad_norm": 6.5178937911987305,
+      "learning_rate": 9.998353288449823e-06,
+      "loss": 3.1868,
+      "step": 107400
+    },
+    {
+      "epoch": 0.028058112627767414,
+      "grad_norm": 5.7756218910217285,
+      "learning_rate": 9.998331766344062e-06,
+      "loss": 3.1752,
+      "step": 107600
+    },
+    {
+      "epoch": 0.028110265253469584,
+      "grad_norm": 6.773250579833984,
+      "learning_rate": 9.998310104529163e-06,
+      "loss": 3.1601,
+      "step": 107800
+    },
+    {
+      "epoch": 0.028162417879171755,
+      "grad_norm": 7.053867816925049,
+      "learning_rate": 9.998288303005734e-06,
+      "loss": 3.181,
+      "step": 108000
+    },
+    {
+      "epoch": 0.028214570504873925,
+      "grad_norm": 6.301112174987793,
+      "learning_rate": 9.998266361774385e-06,
+      "loss": 3.1717,
+      "step": 108200
+    },
+    {
+      "epoch": 0.028266723130576092,
+      "grad_norm": 6.55330753326416,
+      "learning_rate": 9.998244280835728e-06,
+      "loss": 3.1113,
+      "step": 108400
+    },
+    {
+      "epoch": 0.028318875756278263,
+      "grad_norm": 6.428907871246338,
+      "learning_rate": 9.998222060190377e-06,
+      "loss": 3.1395,
+      "step": 108600
+    },
+    {
+      "epoch": 0.028371028381980433,
+      "grad_norm": 6.372659683227539,
+      "learning_rate": 9.998199699838959e-06,
+      "loss": 3.2008,
+      "step": 108800
+    },
+    {
+      "epoch": 0.028423181007682603,
+      "grad_norm": 6.7776689529418945,
+      "learning_rate": 9.998177199782095e-06,
+      "loss": 3.214,
+      "step": 109000
+    },
+    {
+      "epoch": 0.028475333633384774,
+      "grad_norm": 6.657548427581787,
+      "learning_rate": 9.998154560020417e-06,
+      "loss": 3.1759,
+      "step": 109200
+    },
+    {
+      "epoch": 0.028527486259086944,
+      "grad_norm": 7.199187278747559,
+      "learning_rate": 9.998131780554554e-06,
+      "loss": 3.1263,
+      "step": 109400
+    },
+    {
+      "epoch": 0.028579638884789114,
+      "grad_norm": 6.797530651092529,
+      "learning_rate": 9.998108861385145e-06,
+      "loss": 3.1394,
+      "step": 109600
+    },
+    {
+      "epoch": 0.02863179151049128,
+      "grad_norm": 7.414548873901367,
+      "learning_rate": 9.998085802512832e-06,
+      "loss": 3.116,
+      "step": 109800
+    },
+    {
+      "epoch": 0.028683944136193452,
+      "grad_norm": 6.82701301574707,
+      "learning_rate": 9.998062603938255e-06,
+      "loss": 3.1589,
+      "step": 110000
+    },
+    {
+      "epoch": 0.028736096761895622,
+      "grad_norm": 6.446861267089844,
+      "learning_rate": 9.998039265662067e-06,
+      "loss": 3.1494,
+      "step": 110200
+    },
+    {
+      "epoch": 0.028788249387597793,
+      "grad_norm": 6.727176189422607,
+      "learning_rate": 9.998015787684919e-06,
+      "loss": 3.1162,
+      "step": 110400
+    },
+    {
+      "epoch": 0.028840402013299963,
+      "grad_norm": 7.0100226402282715,
+      "learning_rate": 9.997992170007464e-06,
+      "loss": 3.1154,
+      "step": 110600
+    },
+    {
+      "epoch": 0.028892554639002133,
+      "grad_norm": 5.6385722160339355,
+      "learning_rate": 9.997968412630368e-06,
+      "loss": 3.1144,
+      "step": 110800
+    },
+    {
+      "epoch": 0.028944707264704304,
+      "grad_norm": 6.636441230773926,
+      "learning_rate": 9.997944515554291e-06,
+      "loss": 3.1642,
+      "step": 111000
+    },
+    {
+      "epoch": 0.028996859890406474,
+      "grad_norm": 6.764395236968994,
+      "learning_rate": 9.997920478779901e-06,
+      "loss": 3.1758,
+      "step": 111200
+    },
+    {
+      "epoch": 0.02904901251610864,
+      "grad_norm": 7.620389938354492,
+      "learning_rate": 9.997896302307872e-06,
+      "loss": 3.1518,
+      "step": 111400
+    },
+    {
+      "epoch": 0.02910116514181081,
+      "grad_norm": 6.4152021408081055,
+      "learning_rate": 9.997871986138878e-06,
+      "loss": 3.1043,
+      "step": 111600
+    },
+    {
+      "epoch": 0.029153317767512982,
+      "grad_norm": 5.856719493865967,
+      "learning_rate": 9.9978475302736e-06,
+      "loss": 3.1493,
+      "step": 111800
+    },
+    {
+      "epoch": 0.029205470393215152,
+      "grad_norm": 6.708147048950195,
+      "learning_rate": 9.99782293471272e-06,
+      "loss": 3.1646,
+      "step": 112000
+    },
+    {
+      "epoch": 0.029257623018917323,
+      "grad_norm": 6.783796310424805,
+      "learning_rate": 9.997798199456927e-06,
+      "loss": 3.1477,
+      "step": 112200
+    },
+    {
+      "epoch": 0.029309775644619493,
+      "grad_norm": 6.652106761932373,
+      "learning_rate": 9.997773324506912e-06,
+      "loss": 3.107,
+      "step": 112400
+    },
+    {
+      "epoch": 0.029361928270321663,
+      "grad_norm": 6.451571464538574,
+      "learning_rate": 9.997748309863367e-06,
+      "loss": 3.1245,
+      "step": 112600
+    },
+    {
+      "epoch": 0.02941408089602383,
+      "grad_norm": 6.6614203453063965,
+      "learning_rate": 9.997723155526998e-06,
+      "loss": 3.1216,
+      "step": 112800
+    },
+    {
+      "epoch": 0.029466233521726,
+      "grad_norm": 6.864782333374023,
+      "learning_rate": 9.9976978614985e-06,
+      "loss": 3.1465,
+      "step": 113000
+    },
+    {
+      "epoch": 0.02951838614742817,
+      "grad_norm": 7.249753952026367,
+      "learning_rate": 9.997672427778588e-06,
+      "loss": 3.1112,
+      "step": 113200
+    },
+    {
+      "epoch": 0.02957053877313034,
+      "grad_norm": 6.135322093963623,
+      "learning_rate": 9.997646854367968e-06,
+      "loss": 3.1491,
+      "step": 113400
+    },
+    {
+      "epoch": 0.029622691398832512,
+      "grad_norm": 6.486145496368408,
+      "learning_rate": 9.997621141267355e-06,
+      "loss": 3.1672,
+      "step": 113600
+    },
+    {
+      "epoch": 0.029674844024534682,
+      "grad_norm": 7.196221828460693,
+      "learning_rate": 9.99759528847747e-06,
+      "loss": 3.1266,
+      "step": 113800
+    },
+    {
+      "epoch": 0.029726996650236853,
+      "grad_norm": 6.458970546722412,
+      "learning_rate": 9.997569295999032e-06,
+      "loss": 3.1566,
+      "step": 114000
+    },
+    {
+      "epoch": 0.02977914927593902,
+      "grad_norm": 6.437240123748779,
+      "learning_rate": 9.997543163832772e-06,
+      "loss": 3.0992,
+      "step": 114200
+    },
+    {
+      "epoch": 0.02983130190164119,
+      "grad_norm": 6.01683235168457,
+      "learning_rate": 9.997516891979418e-06,
+      "loss": 3.1501,
+      "step": 114400
+    },
+    {
+      "epoch": 0.02988345452734336,
+      "grad_norm": 6.238487720489502,
+      "learning_rate": 9.997490480439705e-06,
+      "loss": 3.1628,
+      "step": 114600
+    },
+    {
+      "epoch": 0.02993560715304553,
+      "grad_norm": 6.446498870849609,
+      "learning_rate": 9.997463929214368e-06,
+      "loss": 3.1472,
+      "step": 114800
+    },
+    {
+      "epoch": 0.0299877597787477,
+      "grad_norm": 7.348095417022705,
+      "learning_rate": 9.997437238304154e-06,
+      "loss": 3.171,
+      "step": 115000
+    },
+    {
+      "epoch": 0.03003991240444987,
+      "grad_norm": 6.736360549926758,
+      "learning_rate": 9.997410407709806e-06,
+      "loss": 3.1841,
+      "step": 115200
+    },
+    {
+      "epoch": 0.030092065030152042,
+      "grad_norm": 6.533570289611816,
+      "learning_rate": 9.997383437432075e-06,
+      "loss": 3.1261,
+      "step": 115400
+    },
+    {
+      "epoch": 0.030144217655854212,
+      "grad_norm": 6.483672618865967,
+      "learning_rate": 9.997356327471716e-06,
+      "loss": 3.1736,
+      "step": 115600
+    },
+    {
+      "epoch": 0.03019637028155638,
+      "grad_norm": 5.92385721206665,
+      "learning_rate": 9.997329077829484e-06,
+      "loss": 3.0937,
+      "step": 115800
+    },
+    {
+      "epoch": 0.03024852290725855,
+      "grad_norm": 6.332182884216309,
+      "learning_rate": 9.997301688506143e-06,
+      "loss": 3.0874,
+      "step": 116000
+    },
+    {
+      "epoch": 0.03030067553296072,
+      "grad_norm": 6.6715521812438965,
+      "learning_rate": 9.997274159502457e-06,
+      "loss": 3.1354,
+      "step": 116200
+    },
+    {
+      "epoch": 0.03035282815866289,
+      "grad_norm": 6.579791069030762,
+      "learning_rate": 9.997246490819197e-06,
+      "loss": 3.1421,
+      "step": 116400
+    },
+    {
+      "epoch": 0.03040498078436506,
+      "grad_norm": 7.069493770599365,
+      "learning_rate": 9.997218682457135e-06,
+      "loss": 3.1171,
+      "step": 116600
+    },
+    {
+      "epoch": 0.03045713341006723,
+      "grad_norm": 6.911351680755615,
+      "learning_rate": 9.997190734417048e-06,
+      "loss": 3.0969,
+      "step": 116800
+    },
+    {
+      "epoch": 0.0305092860357694,
+      "grad_norm": 6.206057548522949,
+      "learning_rate": 9.99716264669972e-06,
+      "loss": 3.1368,
+      "step": 117000
+    },
+    {
+      "epoch": 0.03056143866147157,
+      "grad_norm": 6.447131156921387,
+      "learning_rate": 9.997134419305933e-06,
+      "loss": 3.1165,
+      "step": 117200
+    },
+    {
+      "epoch": 0.03061359128717374,
+      "grad_norm": 6.3493804931640625,
+      "learning_rate": 9.997106052236475e-06,
+      "loss": 3.1207,
+      "step": 117400
+    },
+    {
+      "epoch": 0.03066574391287591,
+      "grad_norm": 6.453786849975586,
+      "learning_rate": 9.997077545492144e-06,
+      "loss": 3.1396,
+      "step": 117600
+    },
+    {
+      "epoch": 0.03071789653857808,
+      "grad_norm": 6.4193010330200195,
+      "learning_rate": 9.997048899073734e-06,
+      "loss": 3.129,
+      "step": 117800
+    },
+    {
+      "epoch": 0.03077004916428025,
+      "grad_norm": 6.173766136169434,
+      "learning_rate": 9.997020112982043e-06,
+      "loss": 3.157,
+      "step": 118000
+    },
+    {
+      "epoch": 0.03082220178998242,
+      "grad_norm": 6.576676368713379,
+      "learning_rate": 9.99699118721788e-06,
+      "loss": 3.1641,
+      "step": 118200
+    },
+    {
+      "epoch": 0.03087435441568459,
+      "grad_norm": 6.761925220489502,
+      "learning_rate": 9.99696212178205e-06,
+      "loss": 3.101,
+      "step": 118400
+    },
+    {
+      "epoch": 0.030926507041386758,
+      "grad_norm": 6.417123317718506,
+      "learning_rate": 9.996932916675368e-06,
+      "loss": 3.0993,
+      "step": 118600
+    },
+    {
+      "epoch": 0.030978659667088928,
+      "grad_norm": 6.432227611541748,
+      "learning_rate": 9.996903571898649e-06,
+      "loss": 3.1012,
+      "step": 118800
+    },
+    {
+      "epoch": 0.0310308122927911,
+      "grad_norm": 6.400918006896973,
+      "learning_rate": 9.996874087452714e-06,
+      "loss": 3.1017,
+      "step": 119000
+    },
+    {
+      "epoch": 0.03108296491849327,
+      "grad_norm": 6.3425984382629395,
+      "learning_rate": 9.996844463338387e-06,
+      "loss": 3.0497,
+      "step": 119200
+    },
+    {
+      "epoch": 0.03113511754419544,
+      "grad_norm": 6.788375377655029,
+      "learning_rate": 9.996814699556494e-06,
+      "loss": 3.1498,
+      "step": 119400
+    },
+    {
+      "epoch": 0.03118727016989761,
+      "grad_norm": 6.5473761558532715,
+      "learning_rate": 9.99678479610787e-06,
+      "loss": 3.0913,
+      "step": 119600
+    },
+    {
+      "epoch": 0.03123942279559978,
+      "grad_norm": 6.9028191566467285,
+      "learning_rate": 9.996754752993348e-06,
+      "loss": 3.1111,
+      "step": 119800
+    },
+    {
+      "epoch": 0.03129157542130195,
+      "grad_norm": 6.383526802062988,
+      "learning_rate": 9.99672457021377e-06,
+      "loss": 3.1439,
+      "step": 120000
+    },
+    {
+      "epoch": 0.03134372804700412,
+      "grad_norm": 6.976934909820557,
+      "learning_rate": 9.996694247769979e-06,
+      "loss": 3.1025,
+      "step": 120200
+    },
+    {
+      "epoch": 0.03139588067270629,
+      "grad_norm": 6.561999320983887,
+      "learning_rate": 9.996663785662823e-06,
+      "loss": 3.1209,
+      "step": 120400
+    },
+    {
+      "epoch": 0.03144803329840846,
+      "grad_norm": 6.6694817543029785,
+      "learning_rate": 9.996633183893152e-06,
+      "loss": 3.0911,
+      "step": 120600
+    },
+    {
+      "epoch": 0.03150018592411063,
+      "grad_norm": 6.184081077575684,
+      "learning_rate": 9.996602442461823e-06,
+      "loss": 3.1038,
+      "step": 120800
+    },
+    {
+      "epoch": 0.0315523385498128,
+      "grad_norm": 6.33581018447876,
+      "learning_rate": 9.996571561369692e-06,
+      "loss": 3.08,
+      "step": 121000
+    },
+    {
+      "epoch": 0.03160449117551497,
+      "grad_norm": 6.361224174499512,
+      "learning_rate": 9.996540540617628e-06,
+      "loss": 3.1018,
+      "step": 121200
+    },
+    {
+      "epoch": 0.03165664380121714,
+      "grad_norm": 5.576674461364746,
+      "learning_rate": 9.996509380206491e-06,
+      "loss": 3.1165,
+      "step": 121400
+    },
+    {
+      "epoch": 0.03170879642691931,
+      "grad_norm": 6.4147844314575195,
+      "learning_rate": 9.996478080137158e-06,
+      "loss": 3.0525,
+      "step": 121600
+    },
+    {
+      "epoch": 0.03176094905262148,
+      "grad_norm": 7.201076030731201,
+      "learning_rate": 9.996446640410502e-06,
+      "loss": 3.0917,
+      "step": 121800
+    },
+    {
+      "epoch": 0.03181310167832365,
+      "grad_norm": 6.715845584869385,
+      "learning_rate": 9.9964150610274e-06,
+      "loss": 3.0954,
+      "step": 122000
+    },
+    {
+      "epoch": 0.03186525430402582,
+      "grad_norm": 6.023102283477783,
+      "learning_rate": 9.996383341988736e-06,
+      "loss": 3.0728,
+      "step": 122200
+    },
+    {
+      "epoch": 0.031917406929727984,
+      "grad_norm": 7.259410381317139,
+      "learning_rate": 9.996351483295396e-06,
+      "loss": 3.0708,
+      "step": 122400
+    },
+    {
+      "epoch": 0.031969559555430155,
+      "grad_norm": 6.183089256286621,
+      "learning_rate": 9.996319484948273e-06,
+      "loss": 3.1023,
+      "step": 122600
+    },
+    {
+      "epoch": 0.032021712181132325,
+      "grad_norm": 6.622340202331543,
+      "learning_rate": 9.996287346948258e-06,
+      "loss": 3.0743,
+      "step": 122800
+    },
+    {
+      "epoch": 0.032073864806834496,
+      "grad_norm": 5.829206943511963,
+      "learning_rate": 9.996255069296251e-06,
+      "loss": 3.1051,
+      "step": 123000
+    },
+    {
+      "epoch": 0.032126017432536666,
+      "grad_norm": 6.266514301300049,
+      "learning_rate": 9.996222651993153e-06,
+      "loss": 3.0887,
+      "step": 123200
+    },
+    {
+      "epoch": 0.032178170058238836,
+      "grad_norm": 5.771326065063477,
+      "learning_rate": 9.996190095039874e-06,
+      "loss": 3.0605,
+      "step": 123400
+    },
+    {
+      "epoch": 0.03223032268394101,
+      "grad_norm": 6.461459636688232,
+      "learning_rate": 9.996157398437319e-06,
+      "loss": 3.089,
+      "step": 123600
+    },
+    {
+      "epoch": 0.03228247530964318,
+      "grad_norm": 6.701711177825928,
+      "learning_rate": 9.996124562186402e-06,
+      "loss": 3.0984,
+      "step": 123800
+    },
+    {
+      "epoch": 0.03233462793534535,
+      "grad_norm": 6.251786708831787,
+      "learning_rate": 9.996091586288045e-06,
+      "loss": 3.0881,
+      "step": 124000
+    },
+    {
+      "epoch": 0.03238678056104752,
+      "grad_norm": 6.822304725646973,
+      "learning_rate": 9.996058470743167e-06,
+      "loss": 3.1225,
+      "step": 124200
+    },
+    {
+      "epoch": 0.03243893318674969,
+      "grad_norm": 6.3806257247924805,
+      "learning_rate": 9.996025215552694e-06,
+      "loss": 3.0939,
+      "step": 124400
+    },
+    {
+      "epoch": 0.03249108581245186,
+      "grad_norm": 7.634413242340088,
+      "learning_rate": 9.995991820717557e-06,
+      "loss": 3.0543,
+      "step": 124600
+    },
+    {
+      "epoch": 0.03254323843815403,
+      "grad_norm": 7.31152868270874,
+      "learning_rate": 9.995958286238685e-06,
+      "loss": 3.1145,
+      "step": 124800
+    },
+    {
+      "epoch": 0.0325953910638562,
+      "grad_norm": 6.632618427276611,
+      "learning_rate": 9.99592461211702e-06,
+      "loss": 3.0899,
+      "step": 125000
+    },
+    {
+      "epoch": 0.03264754368955836,
+      "grad_norm": 6.440008640289307,
+      "learning_rate": 9.995890798353503e-06,
+      "loss": 3.0536,
+      "step": 125200
+    },
+    {
+      "epoch": 0.03269969631526053,
+      "grad_norm": 6.341098785400391,
+      "learning_rate": 9.995856844949075e-06,
+      "loss": 3.1024,
+      "step": 125400
+    },
+    {
+      "epoch": 0.032751848940962704,
+      "grad_norm": 6.386812686920166,
+      "learning_rate": 9.99582275190469e-06,
+      "loss": 3.0769,
+      "step": 125600
+    },
+    {
+      "epoch": 0.032804001566664874,
+      "grad_norm": 6.581667900085449,
+      "learning_rate": 9.995788519221297e-06,
+      "loss": 3.0737,
+      "step": 125800
+    },
+    {
+      "epoch": 0.032856154192367044,
+      "grad_norm": 7.009296894073486,
+      "learning_rate": 9.995754146899856e-06,
+      "loss": 3.0732,
+      "step": 126000
+    },
+    {
+      "epoch": 0.032908306818069215,
+      "grad_norm": 6.533790588378906,
+      "learning_rate": 9.995719634941325e-06,
+      "loss": 3.1034,
+      "step": 126200
+    },
+    {
+      "epoch": 0.032960459443771385,
+      "grad_norm": 6.344036102294922,
+      "learning_rate": 9.99568498334667e-06,
+      "loss": 3.0929,
+      "step": 126400
+    },
+    {
+      "epoch": 0.033012612069473556,
+      "grad_norm": 6.279763221740723,
+      "learning_rate": 9.995650192116862e-06,
+      "loss": 3.068,
+      "step": 126600
+    },
+    {
+      "epoch": 0.033064764695175726,
+      "grad_norm": 7.0318922996521,
+      "learning_rate": 9.995615261252868e-06,
+      "loss": 3.061,
+      "step": 126800
+    },
+    {
+      "epoch": 0.033116917320877896,
+      "grad_norm": 6.694097518920898,
+      "learning_rate": 9.995580190755667e-06,
+      "loss": 3.0943,
+      "step": 127000
+    },
+    {
+      "epoch": 0.03316906994658007,
+      "grad_norm": 5.494612693786621,
+      "learning_rate": 9.995544980626241e-06,
+      "loss": 3.1181,
+      "step": 127200
+    },
+    {
+      "epoch": 0.03322122257228224,
+      "grad_norm": 6.504891395568848,
+      "learning_rate": 9.995509630865573e-06,
+      "loss": 3.0556,
+      "step": 127400
+    },
+    {
+      "epoch": 0.03327337519798441,
+      "grad_norm": 6.207190036773682,
+      "learning_rate": 9.99547414147465e-06,
+      "loss": 3.106,
+      "step": 127600
+    },
+    {
+      "epoch": 0.03332552782368658,
+      "grad_norm": 6.687653064727783,
+      "learning_rate": 9.995438512454465e-06,
+      "loss": 3.1383,
+      "step": 127800
+    },
+    {
+      "epoch": 0.03337768044938875,
+      "grad_norm": 5.720125198364258,
+      "learning_rate": 9.995402743806012e-06,
+      "loss": 3.062,
+      "step": 128000
+    },
+    {
+      "epoch": 0.03342983307509091,
+      "grad_norm": 5.79355001449585,
+      "learning_rate": 9.995366835530295e-06,
+      "loss": 3.0938,
+      "step": 128200
+    },
+    {
+      "epoch": 0.03348198570079308,
+      "grad_norm": 6.339210510253906,
+      "learning_rate": 9.995330787628315e-06,
+      "loss": 3.0733,
+      "step": 128400
+    },
+    {
+      "epoch": 0.03353413832649525,
+      "grad_norm": 6.438317775726318,
+      "learning_rate": 9.995294600101077e-06,
+      "loss": 3.0625,
+      "step": 128600
+    },
+    {
+      "epoch": 0.03358629095219742,
+      "grad_norm": 6.823408126831055,
+      "learning_rate": 9.995258272949597e-06,
+      "loss": 3.1021,
+      "step": 128800
+    },
+    {
+      "epoch": 0.03363844357789959,
+      "grad_norm": 5.777518272399902,
+      "learning_rate": 9.995221806174888e-06,
+      "loss": 3.0813,
+      "step": 129000
+    },
+    {
+      "epoch": 0.033690596203601764,
+      "grad_norm": 6.753819942474365,
+      "learning_rate": 9.99518519977797e-06,
+      "loss": 3.0556,
+      "step": 129200
+    },
+    {
+      "epoch": 0.033742748829303934,
+      "grad_norm": 6.781210422515869,
+      "learning_rate": 9.995148453759866e-06,
+      "loss": 3.1216,
+      "step": 129400
+    },
+    {
+      "epoch": 0.033794901455006104,
+      "grad_norm": 6.243967533111572,
+      "learning_rate": 9.995111568121605e-06,
+      "loss": 3.0294,
+      "step": 129600
+    },
+    {
+      "epoch": 0.033847054080708275,
+      "grad_norm": 6.602694511413574,
+      "learning_rate": 9.995074542864215e-06,
+      "loss": 3.0701,
+      "step": 129800
+    },
+    {
+      "epoch": 0.033899206706410445,
+      "grad_norm": 6.756824016571045,
+      "learning_rate": 9.99503737798873e-06,
+      "loss": 3.0907,
+      "step": 130000
+    },
+    {
+      "epoch": 0.033951359332112616,
+      "grad_norm": 6.042007923126221,
+      "learning_rate": 9.995000073496192e-06,
+      "loss": 3.0582,
+      "step": 130200
+    },
+    {
+      "epoch": 0.034003511957814786,
+      "grad_norm": 7.078050136566162,
+      "learning_rate": 9.994962629387643e-06,
+      "loss": 3.0474,
+      "step": 130400
+    },
+    {
+      "epoch": 0.034055664583516956,
+      "grad_norm": 5.917229652404785,
+      "learning_rate": 9.994925045664127e-06,
+      "loss": 3.0569,
+      "step": 130600
+    },
+    {
+      "epoch": 0.03410781720921913,
+      "grad_norm": 6.29770565032959,
+      "learning_rate": 9.994887322326698e-06,
+      "loss": 3.0739,
+      "step": 130800
+    },
+    {
+      "epoch": 0.0341599698349213,
+      "grad_norm": 5.848755836486816,
+      "learning_rate": 9.99484945937641e-06,
+      "loss": 3.0496,
+      "step": 131000
+    },
+    {
+      "epoch": 0.03421212246062346,
+      "grad_norm": 6.648880481719971,
+      "learning_rate": 9.99481145681432e-06,
+      "loss": 3.0691,
+      "step": 131200
+    },
+    {
+      "epoch": 0.03426427508632563,
+      "grad_norm": 6.360013961791992,
+      "learning_rate": 9.99477331464149e-06,
+      "loss": 3.084,
+      "step": 131400
+    },
+    {
+      "epoch": 0.0343164277120278,
+      "grad_norm": 6.162940979003906,
+      "learning_rate": 9.994735032858987e-06,
+      "loss": 3.1258,
+      "step": 131600
+    },
+    {
+      "epoch": 0.03436858033772997,
+      "grad_norm": 6.413846492767334,
+      "learning_rate": 9.99469661146788e-06,
+      "loss": 3.0983,
+      "step": 131800
+    },
+    {
+      "epoch": 0.03442073296343214,
+      "grad_norm": 6.364840030670166,
+      "learning_rate": 9.994658050469243e-06,
+      "loss": 3.0567,
+      "step": 132000
+    },
+    {
+      "epoch": 0.03447288558913431,
+      "grad_norm": 6.655250072479248,
+      "learning_rate": 9.994619349864156e-06,
+      "loss": 3.0799,
+      "step": 132200
+    },
+    {
+      "epoch": 0.03452503821483648,
+      "grad_norm": 5.781773090362549,
+      "learning_rate": 9.9945805096537e-06,
+      "loss": 3.0514,
+      "step": 132400
+    },
+    {
+      "epoch": 0.03457719084053865,
+      "grad_norm": 7.000985145568848,
+      "learning_rate": 9.994541529838957e-06,
+      "loss": 3.0704,
+      "step": 132600
+    },
+    {
+      "epoch": 0.034629343466240824,
+      "grad_norm": 6.9071550369262695,
+      "learning_rate": 9.99450241042102e-06,
+      "loss": 3.0856,
+      "step": 132800
+    },
+    {
+      "epoch": 0.034681496091942994,
+      "grad_norm": 5.725367546081543,
+      "learning_rate": 9.994463151400984e-06,
+      "loss": 3.0851,
+      "step": 133000
+    },
+    {
+      "epoch": 0.034733648717645164,
+      "grad_norm": 6.823086738586426,
+      "learning_rate": 9.994423752779942e-06,
+      "loss": 3.0728,
+      "step": 133200
+    },
+    {
+      "epoch": 0.034785801343347335,
+      "grad_norm": 6.22132682800293,
+      "learning_rate": 9.994384214558999e-06,
+      "loss": 3.0625,
+      "step": 133400
+    },
+    {
+      "epoch": 0.034837953969049505,
+      "grad_norm": 6.658827304840088,
+      "learning_rate": 9.994344536739256e-06,
+      "loss": 3.0398,
+      "step": 133600
+    },
+    {
+      "epoch": 0.034890106594751676,
+      "grad_norm": 6.231532096862793,
+      "learning_rate": 9.994304719321825e-06,
+      "loss": 3.0462,
+      "step": 133800
+    },
+    {
+      "epoch": 0.034942259220453846,
+      "grad_norm": 6.183959484100342,
+      "learning_rate": 9.994264762307821e-06,
+      "loss": 3.0558,
+      "step": 134000
+    },
+    {
+      "epoch": 0.03499441184615601,
+      "grad_norm": 6.413631439208984,
+      "learning_rate": 9.994224665698356e-06,
+      "loss": 3.0879,
+      "step": 134200
+    },
+    {
+      "epoch": 0.03504656447185818,
+      "grad_norm": 6.981194496154785,
+      "learning_rate": 9.994184429494554e-06,
+      "loss": 3.0688,
+      "step": 134400
+    },
+    {
+      "epoch": 0.03509871709756035,
+      "grad_norm": 6.0282135009765625,
+      "learning_rate": 9.994144053697539e-06,
+      "loss": 3.0768,
+      "step": 134600
+    },
+    {
+      "epoch": 0.03515086972326252,
+      "grad_norm": 5.871272563934326,
+      "learning_rate": 9.994103538308437e-06,
+      "loss": 3.0281,
+      "step": 134800
+    },
+    {
+      "epoch": 0.03520302234896469,
+      "grad_norm": 6.554361343383789,
+      "learning_rate": 9.994062883328384e-06,
+      "loss": 3.054,
+      "step": 135000
+    },
+    {
+      "epoch": 0.03525517497466686,
+      "grad_norm": 6.451972007751465,
+      "learning_rate": 9.994022088758515e-06,
+      "loss": 3.0884,
+      "step": 135200
+    },
+    {
+      "epoch": 0.03530732760036903,
+      "grad_norm": 6.5166168212890625,
+      "learning_rate": 9.993981154599972e-06,
+      "loss": 3.034,
+      "step": 135400
+    },
+    {
+      "epoch": 0.0353594802260712,
+      "grad_norm": 6.459281921386719,
+      "learning_rate": 9.993940080853895e-06,
+      "loss": 3.0411,
+      "step": 135600
+    },
+    {
+      "epoch": 0.03541163285177337,
+      "grad_norm": 6.400155544281006,
+      "learning_rate": 9.993898867521435e-06,
+      "loss": 3.0376,
+      "step": 135800
+    },
+    {
+      "epoch": 0.03546378547747554,
+      "grad_norm": 6.980037212371826,
+      "learning_rate": 9.993857514603744e-06,
+      "loss": 3.0522,
+      "step": 136000
+    },
+    {
+      "epoch": 0.03551593810317771,
+      "grad_norm": 7.635961532592773,
+      "learning_rate": 9.99381602210198e-06,
+      "loss": 3.0653,
+      "step": 136200
+    },
+    {
+      "epoch": 0.035568090728879884,
+      "grad_norm": 7.077882289886475,
+      "learning_rate": 9.993774390017294e-06,
+      "loss": 3.0062,
+      "step": 136400
+    },
+    {
+      "epoch": 0.035620243354582054,
+      "grad_norm": 6.433791637420654,
+      "learning_rate": 9.99373261835086e-06,
+      "loss": 3.0285,
+      "step": 136600
+    },
+    {
+      "epoch": 0.035672395980284224,
+      "grad_norm": 6.266335964202881,
+      "learning_rate": 9.99369070710384e-06,
+      "loss": 3.0393,
+      "step": 136800
+    },
+    {
+      "epoch": 0.03572454860598639,
+      "grad_norm": 7.584403038024902,
+      "learning_rate": 9.993648656277409e-06,
+      "loss": 3.0379,
+      "step": 137000
+    },
+    {
+      "epoch": 0.03577670123168856,
+      "grad_norm": 5.1457133293151855,
+      "learning_rate": 9.993606465872737e-06,
+      "loss": 3.025,
+      "step": 137200
+    },
+    {
+      "epoch": 0.03582885385739073,
+      "grad_norm": 6.755367279052734,
+      "learning_rate": 9.993564135891007e-06,
+      "loss": 3.039,
+      "step": 137400
+    },
+    {
+      "epoch": 0.0358810064830929,
+      "grad_norm": 6.786823272705078,
+      "learning_rate": 9.993521666333404e-06,
+      "loss": 3.0242,
+      "step": 137600
+    },
+    {
+      "epoch": 0.03593315910879507,
+      "grad_norm": 6.774286270141602,
+      "learning_rate": 9.99347905720111e-06,
+      "loss": 3.0289,
+      "step": 137800
+    },
+    {
+      "epoch": 0.03598531173449724,
+      "grad_norm": 6.3922200202941895,
+      "learning_rate": 9.99343630849532e-06,
+      "loss": 3.0061,
+      "step": 138000
+    },
+    {
+      "epoch": 0.03603746436019941,
+      "grad_norm": 6.802679061889648,
+      "learning_rate": 9.993393420217229e-06,
+      "loss": 3.0451,
+      "step": 138200
+    },
+    {
+      "epoch": 0.03608961698590158,
+      "grad_norm": 6.436954021453857,
+      "learning_rate": 9.993350392368031e-06,
+      "loss": 3.0221,
+      "step": 138400
+    },
+    {
+      "epoch": 0.03614176961160375,
+      "grad_norm": 6.1146721839904785,
+      "learning_rate": 9.993307224948934e-06,
+      "loss": 3.0119,
+      "step": 138600
+    },
+    {
+      "epoch": 0.03619392223730592,
+      "grad_norm": 6.813344478607178,
+      "learning_rate": 9.993263917961142e-06,
+      "loss": 3.0613,
+      "step": 138800
+    },
+    {
+      "epoch": 0.03624607486300809,
+      "grad_norm": 5.840065002441406,
+      "learning_rate": 9.993220471405866e-06,
+      "loss": 3.0639,
+      "step": 139000
+    },
+    {
+      "epoch": 0.03629822748871026,
+      "grad_norm": 7.216462135314941,
+      "learning_rate": 9.99317688528432e-06,
+      "loss": 3.0472,
+      "step": 139200
+    },
+    {
+      "epoch": 0.03635038011441243,
+      "grad_norm": 6.292632102966309,
+      "learning_rate": 9.993133159597724e-06,
+      "loss": 3.0252,
+      "step": 139400
+    },
+    {
+      "epoch": 0.0364025327401146,
+      "grad_norm": 6.57106351852417,
+      "learning_rate": 9.993089294347297e-06,
+      "loss": 3.0342,
+      "step": 139600
+    },
+    {
+      "epoch": 0.03645468536581677,
+      "grad_norm": 6.309388160705566,
+      "learning_rate": 9.99304528953427e-06,
+      "loss": 3.0658,
+      "step": 139800
+    },
+    {
+      "epoch": 0.03650683799151894,
+      "grad_norm": 6.843193054199219,
+      "learning_rate": 9.993001145159867e-06,
+      "loss": 3.0457,
+      "step": 140000
+    },
+    {
+      "epoch": 0.03655899061722111,
+      "grad_norm": 6.428710460662842,
+      "learning_rate": 9.992956861225325e-06,
+      "loss": 3.108,
+      "step": 140200
+    },
+    {
+      "epoch": 0.03661114324292328,
+      "grad_norm": 5.9581499099731445,
+      "learning_rate": 9.992912437731884e-06,
+      "loss": 3.0241,
+      "step": 140400
+    },
+    {
+      "epoch": 0.03666329586862545,
+      "grad_norm": 6.154149055480957,
+      "learning_rate": 9.99286787468078e-06,
+      "loss": 3.0459,
+      "step": 140600
+    },
+    {
+      "epoch": 0.03671544849432762,
+      "grad_norm": 6.878948211669922,
+      "learning_rate": 9.992823172073264e-06,
+      "loss": 3.0133,
+      "step": 140800
+    },
+    {
+      "epoch": 0.03676760112002979,
+      "grad_norm": 7.1989264488220215,
+      "learning_rate": 9.992778329910585e-06,
+      "loss": 3.1056,
+      "step": 141000
+    },
+    {
+      "epoch": 0.03681975374573196,
+      "grad_norm": 6.319765567779541,
+      "learning_rate": 9.992733348193993e-06,
+      "loss": 2.9999,
+      "step": 141200
+    },
+    {
+      "epoch": 0.03687190637143413,
+      "grad_norm": 6.223499298095703,
+      "learning_rate": 9.992688226924747e-06,
+      "loss": 3.0706,
+      "step": 141400
+    },
+    {
+      "epoch": 0.0369240589971363,
+      "grad_norm": 5.92350435256958,
+      "learning_rate": 9.992642966104107e-06,
+      "loss": 3.0198,
+      "step": 141600
+    },
+    {
+      "epoch": 0.03697621162283847,
+      "grad_norm": 6.555150032043457,
+      "learning_rate": 9.992597565733341e-06,
+      "loss": 3.0331,
+      "step": 141800
+    },
+    {
+      "epoch": 0.03702836424854064,
+      "grad_norm": 6.806456089019775,
+      "learning_rate": 9.992552025813716e-06,
+      "loss": 3.042,
+      "step": 142000
+    },
+    {
+      "epoch": 0.03708051687424281,
+      "grad_norm": 6.848278045654297,
+      "learning_rate": 9.992506346346505e-06,
+      "loss": 3.0455,
+      "step": 142200
+    },
+    {
+      "epoch": 0.03713266949994498,
+      "grad_norm": 6.233341693878174,
+      "learning_rate": 9.992460527332986e-06,
+      "loss": 3.0627,
+      "step": 142400
+    },
+    {
+      "epoch": 0.03718482212564715,
+      "grad_norm": 6.317975044250488,
+      "learning_rate": 9.992414568774438e-06,
+      "loss": 3.0447,
+      "step": 142600
+    },
+    {
+      "epoch": 0.03723697475134932,
+      "grad_norm": 6.190425395965576,
+      "learning_rate": 9.992368470672146e-06,
+      "loss": 3.0495,
+      "step": 142800
+    },
+    {
+      "epoch": 0.037289127377051486,
+      "grad_norm": 6.360509395599365,
+      "learning_rate": 9.992322233027398e-06,
+      "loss": 2.9939,
+      "step": 143000
+    },
+    {
+      "epoch": 0.037341280002753656,
+      "grad_norm": 6.74845552444458,
+      "learning_rate": 9.99227585584149e-06,
+      "loss": 2.9787,
+      "step": 143200
+    },
+    {
+      "epoch": 0.037393432628455826,
+      "grad_norm": 6.610714912414551,
+      "learning_rate": 9.992229339115714e-06,
+      "loss": 3.0016,
+      "step": 143400
+    },
+    {
+      "epoch": 0.037445585254158,
+      "grad_norm": 6.472560405731201,
+      "learning_rate": 9.99218268285137e-06,
+      "loss": 2.9786,
+      "step": 143600
+    },
+    {
+      "epoch": 0.03749773787986017,
+      "grad_norm": 7.01295804977417,
+      "learning_rate": 9.992135887049766e-06,
+      "loss": 3.0622,
+      "step": 143800
+    },
+    {
+      "epoch": 0.03754989050556234,
+      "grad_norm": 5.686848163604736,
+      "learning_rate": 9.992088951712207e-06,
+      "loss": 3.0528,
+      "step": 144000
+    },
+    {
+      "epoch": 0.03760204313126451,
+      "grad_norm": 6.806542873382568,
+      "learning_rate": 9.992041876840007e-06,
+      "loss": 3.0256,
+      "step": 144200
+    },
+    {
+      "epoch": 0.03765419575696668,
+      "grad_norm": 6.5922088623046875,
+      "learning_rate": 9.991994662434478e-06,
+      "loss": 3.046,
+      "step": 144400
+    },
+    {
+      "epoch": 0.03770634838266885,
+      "grad_norm": 6.281486988067627,
+      "learning_rate": 9.991947308496945e-06,
+      "loss": 3.0209,
+      "step": 144600
+    },
+    {
+      "epoch": 0.03775850100837102,
+      "grad_norm": 6.36431360244751,
+      "learning_rate": 9.991899815028728e-06,
+      "loss": 2.9928,
+      "step": 144800
+    },
+    {
+      "epoch": 0.03781065363407319,
+      "grad_norm": 6.694343090057373,
+      "learning_rate": 9.991852182031153e-06,
+      "loss": 3.0287,
+      "step": 145000
+    },
+    {
+      "epoch": 0.03786280625977536,
+      "grad_norm": 6.216729164123535,
+      "learning_rate": 9.991804409505557e-06,
+      "loss": 3.0095,
+      "step": 145200
+    },
+    {
+      "epoch": 0.03791495888547753,
+      "grad_norm": 6.089059352874756,
+      "learning_rate": 9.99175649745327e-06,
+      "loss": 3.0268,
+      "step": 145400
+    },
+    {
+      "epoch": 0.0379671115111797,
+      "grad_norm": 5.32235050201416,
+      "learning_rate": 9.991708445875633e-06,
+      "loss": 3.066,
+      "step": 145600
+    },
+    {
+      "epoch": 0.03801926413688187,
+      "grad_norm": 7.004727363586426,
+      "learning_rate": 9.991660254773989e-06,
+      "loss": 3.0154,
+      "step": 145800
+    },
+    {
+      "epoch": 0.038071416762584034,
+      "grad_norm": 5.930303573608398,
+      "learning_rate": 9.991611924149687e-06,
+      "loss": 3.0313,
+      "step": 146000
+    },
+    {
+      "epoch": 0.038123569388286205,
+      "grad_norm": 6.0669145584106445,
+      "learning_rate": 9.991563454004076e-06,
+      "loss": 3.0479,
+      "step": 146200
+    },
+    {
+      "epoch": 0.038175722013988375,
+      "grad_norm": 6.367830276489258,
+      "learning_rate": 9.991514844338509e-06,
+      "loss": 3.0386,
+      "step": 146400
+    },
+    {
+      "epoch": 0.038227874639690546,
+      "grad_norm": 6.8309221267700195,
+      "learning_rate": 9.991466095154348e-06,
+      "loss": 3.0073,
+      "step": 146600
+    },
+    {
+      "epoch": 0.038280027265392716,
+      "grad_norm": 5.97960090637207,
+      "learning_rate": 9.991417206452953e-06,
+      "loss": 3.0035,
+      "step": 146800
+    },
+    {
+      "epoch": 0.038332179891094886,
+      "grad_norm": 6.310183048248291,
+      "learning_rate": 9.991368178235695e-06,
+      "loss": 3.0623,
+      "step": 147000
+    },
+    {
+      "epoch": 0.03838433251679706,
+      "grad_norm": 5.791366100311279,
+      "learning_rate": 9.991319010503938e-06,
+      "loss": 2.9937,
+      "step": 147200
+    },
+    {
+      "epoch": 0.03843648514249923,
+      "grad_norm": 6.281528949737549,
+      "learning_rate": 9.991269703259061e-06,
+      "loss": 3.0342,
+      "step": 147400
+    },
+    {
+      "epoch": 0.0384886377682014,
+      "grad_norm": 6.502980709075928,
+      "learning_rate": 9.99122025650244e-06,
+      "loss": 3.0311,
+      "step": 147600
+    },
+    {
+      "epoch": 0.03854079039390357,
+      "grad_norm": 6.228078842163086,
+      "learning_rate": 9.991170670235456e-06,
+      "loss": 3.0259,
+      "step": 147800
+    },
+    {
+      "epoch": 0.03859294301960574,
+      "grad_norm": 6.6315083503723145,
+      "learning_rate": 9.9911209444595e-06,
+      "loss": 3.0557,
+      "step": 148000
+    },
+    {
+      "epoch": 0.03864509564530791,
+      "grad_norm": 5.9421916007995605,
+      "learning_rate": 9.991071079175958e-06,
+      "loss": 3.0471,
+      "step": 148200
+    },
+    {
+      "epoch": 0.03869724827101008,
+      "grad_norm": 5.794018745422363,
+      "learning_rate": 9.991021074386222e-06,
+      "loss": 3.0512,
+      "step": 148400
+    },
+    {
+      "epoch": 0.03874940089671225,
+      "grad_norm": 6.507724761962891,
+      "learning_rate": 9.990970930091695e-06,
+      "loss": 3.0823,
+      "step": 148600
+    },
+    {
+      "epoch": 0.03880155352241441,
+      "grad_norm": 6.405786514282227,
+      "learning_rate": 9.990920646293773e-06,
+      "loss": 2.9923,
+      "step": 148800
+    },
+    {
+      "epoch": 0.03885370614811658,
+      "grad_norm": 6.925950527191162,
+      "learning_rate": 9.990870222993867e-06,
+      "loss": 3.003,
+      "step": 149000
+    },
+    {
+      "epoch": 0.038905858773818754,
+      "grad_norm": 6.7786030769348145,
+      "learning_rate": 9.990819660193383e-06,
+      "loss": 2.9691,
+      "step": 149200
+    },
+    {
+      "epoch": 0.038958011399520924,
+      "grad_norm": 6.323172569274902,
+      "learning_rate": 9.990768957893732e-06,
+      "loss": 3.0457,
+      "step": 149400
+    },
+    {
+      "epoch": 0.039010164025223094,
+      "grad_norm": 5.784226894378662,
+      "learning_rate": 9.990718116096336e-06,
+      "loss": 3.0644,
+      "step": 149600
+    },
+    {
+      "epoch": 0.039062316650925265,
+      "grad_norm": 6.236383438110352,
+      "learning_rate": 9.990667134802616e-06,
+      "loss": 3.0147,
+      "step": 149800
+    },
+    {
+      "epoch": 0.039114469276627435,
+      "grad_norm": 6.505099773406982,
+      "learning_rate": 9.990616014013992e-06,
+      "loss": 3.0315,
+      "step": 150000
+    },
+    {
+      "epoch": 0.039166621902329606,
+      "grad_norm": 5.979858875274658,
+      "learning_rate": 9.990564753731898e-06,
+      "loss": 2.9863,
+      "step": 150200
+    },
+    {
+      "epoch": 0.039218774528031776,
+      "grad_norm": 6.28549861907959,
+      "learning_rate": 9.990513353957765e-06,
+      "loss": 3.0117,
+      "step": 150400
+    },
+    {
+      "epoch": 0.039270927153733946,
+      "grad_norm": 6.456376552581787,
+      "learning_rate": 9.990461814693028e-06,
+      "loss": 3.0182,
+      "step": 150600
+    },
+    {
+      "epoch": 0.03932307977943612,
+      "grad_norm": 6.814133644104004,
+      "learning_rate": 9.99041013593913e-06,
+      "loss": 2.9962,
+      "step": 150800
+    },
+    {
+      "epoch": 0.03937523240513829,
+      "grad_norm": 5.923751354217529,
+      "learning_rate": 9.990358317697513e-06,
+      "loss": 2.9773,
+      "step": 151000
+    },
+    {
+      "epoch": 0.03942738503084046,
+      "grad_norm": 6.386137962341309,
+      "learning_rate": 9.990306359969629e-06,
+      "loss": 3.0187,
+      "step": 151200
+    },
+    {
+      "epoch": 0.03947953765654263,
+      "grad_norm": 6.146199703216553,
+      "learning_rate": 9.990254262756926e-06,
+      "loss": 3.0153,
+      "step": 151400
+    },
+    {
+      "epoch": 0.0395316902822448,
+      "grad_norm": 6.410252571105957,
+      "learning_rate": 9.990202026060864e-06,
+      "loss": 2.9887,
+      "step": 151600
+    },
+    {
+      "epoch": 0.03958384290794696,
+      "grad_norm": 5.882859230041504,
+      "learning_rate": 9.990149649882902e-06,
+      "loss": 3.026,
+      "step": 151800
+    },
+    {
+      "epoch": 0.03963599553364913,
+      "grad_norm": 6.177847862243652,
+      "learning_rate": 9.990097134224503e-06,
+      "loss": 2.9983,
+      "step": 152000
+    },
+    {
+      "epoch": 0.0396881481593513,
+      "grad_norm": 6.150598049163818,
+      "learning_rate": 9.990044479087134e-06,
+      "loss": 2.9855,
+      "step": 152200
+    },
+    {
+      "epoch": 0.03974030078505347,
+      "grad_norm": 6.874467372894287,
+      "learning_rate": 9.989991684472269e-06,
+      "loss": 3.0343,
+      "step": 152400
+    },
+    {
+      "epoch": 0.03979245341075564,
+      "grad_norm": 6.260432720184326,
+      "learning_rate": 9.989938750381383e-06,
+      "loss": 3.0507,
+      "step": 152600
+    },
+    {
+      "epoch": 0.039844606036457814,
+      "grad_norm": 6.114156246185303,
+      "learning_rate": 9.989885676815955e-06,
+      "loss": 3.0064,
+      "step": 152800
+    },
+    {
+      "epoch": 0.039896758662159984,
+      "grad_norm": 6.27534294128418,
+      "learning_rate": 9.989832463777469e-06,
+      "loss": 2.9889,
+      "step": 153000
+    },
+    {
+      "epoch": 0.039948911287862154,
+      "grad_norm": 6.149242877960205,
+      "learning_rate": 9.989779111267411e-06,
+      "loss": 2.9928,
+      "step": 153200
+    },
+    {
+      "epoch": 0.040001063913564325,
+      "grad_norm": 6.631002902984619,
+      "learning_rate": 9.989725619287276e-06,
+      "loss": 2.9901,
+      "step": 153400
+    },
+    {
+      "epoch": 0.040053216539266495,
+      "grad_norm": 6.320735454559326,
+      "learning_rate": 9.989671987838554e-06,
+      "loss": 3.0054,
+      "step": 153600
+    },
+    {
+      "epoch": 0.040105369164968666,
+      "grad_norm": 6.417031764984131,
+      "learning_rate": 9.989618216922747e-06,
+      "loss": 3.0356,
+      "step": 153800
+    },
+    {
+      "epoch": 0.040157521790670836,
+      "grad_norm": 5.854264736175537,
+      "learning_rate": 9.989564306541359e-06,
+      "loss": 2.9719,
+      "step": 154000
+    },
+    {
+      "epoch": 0.040209674416373006,
+      "grad_norm": 6.106746196746826,
+      "learning_rate": 9.989510256695893e-06,
+      "loss": 3.0148,
+      "step": 154200
+    },
+    {
+      "epoch": 0.04026182704207518,
+      "grad_norm": 6.044825553894043,
+      "learning_rate": 9.989456067387864e-06,
+      "loss": 2.9664,
+      "step": 154400
+    },
+    {
+      "epoch": 0.04031397966777735,
+      "grad_norm": 6.020196437835693,
+      "learning_rate": 9.989401738618785e-06,
+      "loss": 2.9737,
+      "step": 154600
+    },
+    {
+      "epoch": 0.04036613229347951,
+      "grad_norm": 5.970254898071289,
+      "learning_rate": 9.989347270390174e-06,
+      "loss": 3.0247,
+      "step": 154800
+    },
+    {
+      "epoch": 0.04041828491918168,
+      "grad_norm": 6.415491104125977,
+      "learning_rate": 9.989292662703554e-06,
+      "loss": 3.0039,
+      "step": 155000
+    },
+    {
+      "epoch": 0.04047043754488385,
+      "grad_norm": 6.5930094718933105,
+      "learning_rate": 9.98923791556045e-06,
+      "loss": 2.9937,
+      "step": 155200
+    },
+    {
+      "epoch": 0.04052259017058602,
+      "grad_norm": 6.674596786499023,
+      "learning_rate": 9.989183028962395e-06,
+      "loss": 3.0017,
+      "step": 155400
+    },
+    {
+      "epoch": 0.04057474279628819,
+      "grad_norm": 6.060914039611816,
+      "learning_rate": 9.989128002910922e-06,
+      "loss": 3.0152,
+      "step": 155600
+    },
+    {
+      "epoch": 0.04062689542199036,
+      "grad_norm": 5.70841121673584,
+      "learning_rate": 9.989072837407567e-06,
+      "loss": 3.0362,
+      "step": 155800
+    },
+    {
+      "epoch": 0.04067904804769253,
+      "grad_norm": 5.961905479431152,
+      "learning_rate": 9.989017532453876e-06,
+      "loss": 2.9848,
+      "step": 156000
+    },
+    {
+      "epoch": 0.0407312006733947,
+      "grad_norm": 5.478226184844971,
+      "learning_rate": 9.988962088051389e-06,
+      "loss": 2.9478,
+      "step": 156200
+    },
+    {
+      "epoch": 0.040783353299096874,
+      "grad_norm": 6.810139179229736,
+      "learning_rate": 9.98890650420166e-06,
+      "loss": 3.0032,
+      "step": 156400
+    },
+    {
+      "epoch": 0.040835505924799044,
+      "grad_norm": 6.793480396270752,
+      "learning_rate": 9.988850780906242e-06,
+      "loss": 2.9757,
+      "step": 156600
+    },
+    {
+      "epoch": 0.040887658550501214,
+      "grad_norm": 6.606081962585449,
+      "learning_rate": 9.988794918166695e-06,
+      "loss": 3.0407,
+      "step": 156800
+    },
+    {
+      "epoch": 0.040939811176203385,
+      "grad_norm": 6.329009532928467,
+      "learning_rate": 9.988738915984575e-06,
+      "loss": 2.9683,
+      "step": 157000
+    },
+    {
+      "epoch": 0.040991963801905555,
+      "grad_norm": 6.828449249267578,
+      "learning_rate": 9.988682774361451e-06,
+      "loss": 3.0212,
+      "step": 157200
+    },
+    {
+      "epoch": 0.041044116427607726,
+      "grad_norm": 7.731756210327148,
+      "learning_rate": 9.98862649329889e-06,
+      "loss": 2.9905,
+      "step": 157400
+    },
+    {
+      "epoch": 0.041096269053309896,
+      "grad_norm": 6.351221561431885,
+      "learning_rate": 9.98857007279847e-06,
+      "loss": 2.9857,
+      "step": 157600
+    },
+    {
+      "epoch": 0.04114842167901206,
+      "grad_norm": 6.527481555938721,
+      "learning_rate": 9.988513512861761e-06,
+      "loss": 2.9985,
+      "step": 157800
+    },
+    {
+      "epoch": 0.04120057430471423,
+      "grad_norm": 6.648956298828125,
+      "learning_rate": 9.988456813490348e-06,
+      "loss": 2.972,
+      "step": 158000
+    },
+    {
+      "epoch": 0.0412527269304164,
+      "grad_norm": 6.268485069274902,
+      "learning_rate": 9.988399974685815e-06,
+      "loss": 2.9893,
+      "step": 158200
+    },
+    {
+      "epoch": 0.04130487955611857,
+      "grad_norm": 6.1307806968688965,
+      "learning_rate": 9.988342996449751e-06,
+      "loss": 2.9647,
+      "step": 158400
+    },
+    {
+      "epoch": 0.04135703218182074,
+      "grad_norm": 7.105797290802002,
+      "learning_rate": 9.988285878783748e-06,
+      "loss": 2.9828,
+      "step": 158600
+    },
+    {
+      "epoch": 0.04140918480752291,
+      "grad_norm": 6.388288974761963,
+      "learning_rate": 9.988228621689403e-06,
+      "loss": 3.0058,
+      "step": 158800
+    },
+    {
+      "epoch": 0.04146133743322508,
+      "grad_norm": 6.2273850440979,
+      "learning_rate": 9.988171225168318e-06,
+      "loss": 2.9956,
+      "step": 159000
+    },
+    {
+      "epoch": 0.04151349005892725,
+      "grad_norm": 6.558915138244629,
+      "learning_rate": 9.988113689222094e-06,
+      "loss": 2.9879,
+      "step": 159200
+    },
+    {
+      "epoch": 0.04156564268462942,
+      "grad_norm": 5.836716651916504,
+      "learning_rate": 9.988056013852343e-06,
+      "loss": 2.9391,
+      "step": 159400
+    },
+    {
+      "epoch": 0.04161779531033159,
+      "grad_norm": 6.490838050842285,
+      "learning_rate": 9.987998199060674e-06,
+      "loss": 3.0215,
+      "step": 159600
+    },
+    {
+      "epoch": 0.04166994793603376,
+      "grad_norm": 6.8759589195251465,
+      "learning_rate": 9.987940244848701e-06,
+      "loss": 3.0393,
+      "step": 159800
+    },
+    {
+      "epoch": 0.041722100561735934,
+      "grad_norm": 6.900877952575684,
+      "learning_rate": 9.987882151218052e-06,
+      "loss": 2.9726,
+      "step": 160000
+    },
+    {
+      "epoch": 0.041774253187438104,
+      "grad_norm": 6.450899124145508,
+      "learning_rate": 9.987823918170343e-06,
+      "loss": 3.0096,
+      "step": 160200
+    },
+    {
+      "epoch": 0.041826405813140274,
+      "grad_norm": 6.636816024780273,
+      "learning_rate": 9.987765545707202e-06,
+      "loss": 2.9626,
+      "step": 160400
+    },
+    {
+      "epoch": 0.04187855843884244,
+      "grad_norm": 6.257472515106201,
+      "learning_rate": 9.987707033830266e-06,
+      "loss": 2.9946,
+      "step": 160600
+    },
+    {
+      "epoch": 0.04193071106454461,
+      "grad_norm": 7.453077793121338,
+      "learning_rate": 9.987648382541167e-06,
+      "loss": 3.0467,
+      "step": 160800
+    },
+    {
+      "epoch": 0.04198286369024678,
+      "grad_norm": 6.978682041168213,
+      "learning_rate": 9.987589591841545e-06,
+      "loss": 2.9799,
+      "step": 161000
+    },
+    {
+      "epoch": 0.04203501631594895,
+      "grad_norm": 6.9115095138549805,
+      "learning_rate": 9.98753066173304e-06,
+      "loss": 3.0193,
+      "step": 161200
+    },
+    {
+      "epoch": 0.04208716894165112,
+      "grad_norm": 6.3997392654418945,
+      "learning_rate": 9.987471592217306e-06,
+      "loss": 3.0198,
+      "step": 161400
+    },
+    {
+      "epoch": 0.04213932156735329,
+      "grad_norm": 5.600056171417236,
+      "learning_rate": 9.987412383295988e-06,
+      "loss": 3.0084,
+      "step": 161600
+    },
+    {
+      "epoch": 0.04219147419305546,
+      "grad_norm": 6.3253984451293945,
+      "learning_rate": 9.987353034970743e-06,
+      "loss": 2.964,
+      "step": 161800
+    },
+    {
+      "epoch": 0.04224362681875763,
+      "grad_norm": 7.4514384269714355,
+      "learning_rate": 9.987293547243231e-06,
+      "loss": 3.0007,
+      "step": 162000
+    },
+    {
+      "epoch": 0.0422957794444598,
+      "grad_norm": 6.729308605194092,
+      "learning_rate": 9.987233920115114e-06,
+      "loss": 2.9636,
+      "step": 162200
+    },
+    {
+      "epoch": 0.04234793207016197,
+      "grad_norm": 6.6917195320129395,
+      "learning_rate": 9.987174153588058e-06,
+      "loss": 2.9993,
+      "step": 162400
+    },
+    {
+      "epoch": 0.04240008469586414,
+      "grad_norm": 6.379083156585693,
+      "learning_rate": 9.987114247663734e-06,
+      "loss": 2.9828,
+      "step": 162600
+    },
+    {
+      "epoch": 0.04245223732156631,
+      "grad_norm": 7.085773944854736,
+      "learning_rate": 9.987054202343817e-06,
+      "loss": 2.9738,
+      "step": 162800
+    },
+    {
+      "epoch": 0.04250438994726848,
+      "grad_norm": 6.782082557678223,
+      "learning_rate": 9.986994017629983e-06,
+      "loss": 2.9808,
+      "step": 163000
+    },
+    {
+      "epoch": 0.04255654257297065,
+      "grad_norm": 5.9306182861328125,
+      "learning_rate": 9.986933693523919e-06,
+      "loss": 2.981,
+      "step": 163200
+    },
+    {
+      "epoch": 0.04260869519867282,
+      "grad_norm": 6.100393295288086,
+      "learning_rate": 9.986873230027305e-06,
+      "loss": 2.9452,
+      "step": 163400
+    },
+    {
+      "epoch": 0.04266084782437499,
+      "grad_norm": 6.214852809906006,
+      "learning_rate": 9.986812627141836e-06,
+      "loss": 2.9452,
+      "step": 163600
+    },
+    {
+      "epoch": 0.04271300045007716,
+      "grad_norm": 6.460691928863525,
+      "learning_rate": 9.986751884869204e-06,
+      "loss": 2.9781,
+      "step": 163800
+    },
+    {
+      "epoch": 0.04276515307577933,
+      "grad_norm": 5.98699951171875,
+      "learning_rate": 9.986691003211106e-06,
+      "loss": 2.9453,
+      "step": 164000
+    },
+    {
+      "epoch": 0.0428173057014815,
+      "grad_norm": 6.0534281730651855,
+      "learning_rate": 9.986629982169244e-06,
+      "loss": 2.978,
+      "step": 164200
+    },
+    {
+      "epoch": 0.04286945832718367,
+      "grad_norm": 6.20578145980835,
+      "learning_rate": 9.986568821745327e-06,
+      "loss": 2.9576,
+      "step": 164400
+    },
+    {
+      "epoch": 0.04292161095288584,
+      "grad_norm": 6.547789573669434,
+      "learning_rate": 9.986507521941058e-06,
+      "loss": 3.0062,
+      "step": 164600
+    },
+    {
+      "epoch": 0.04297376357858801,
+      "grad_norm": 6.358675956726074,
+      "learning_rate": 9.986446082758157e-06,
+      "loss": 2.9771,
+      "step": 164800
+    },
+    {
+      "epoch": 0.04302591620429018,
+      "grad_norm": 6.335031509399414,
+      "learning_rate": 9.986384504198336e-06,
+      "loss": 2.9796,
+      "step": 165000
+    },
+    {
+      "epoch": 0.04307806882999235,
+      "grad_norm": 6.8711090087890625,
+      "learning_rate": 9.98632278626332e-06,
+      "loss": 2.9605,
+      "step": 165200
+    },
+    {
+      "epoch": 0.04313022145569452,
+      "grad_norm": 5.9179911613464355,
+      "learning_rate": 9.986260928954833e-06,
+      "loss": 2.9654,
+      "step": 165400
+    },
+    {
+      "epoch": 0.04318237408139669,
+      "grad_norm": 6.011270046234131,
+      "learning_rate": 9.986198932274601e-06,
+      "loss": 2.9727,
+      "step": 165600
+    },
+    {
+      "epoch": 0.04323452670709886,
+      "grad_norm": 6.755000591278076,
+      "learning_rate": 9.986136796224363e-06,
+      "loss": 2.9796,
+      "step": 165800
+    },
+    {
+      "epoch": 0.04328667933280103,
+      "grad_norm": 6.37379264831543,
+      "learning_rate": 9.986074520805853e-06,
+      "loss": 3.0062,
+      "step": 166000
+    },
+    {
+      "epoch": 0.0433388319585032,
+      "grad_norm": 6.6591877937316895,
+      "learning_rate": 9.986012106020808e-06,
+      "loss": 2.9866,
+      "step": 166200
+    },
+    {
+      "epoch": 0.04339098458420537,
+      "grad_norm": 6.0572614669799805,
+      "learning_rate": 9.985949551870977e-06,
+      "loss": 3.014,
+      "step": 166400
+    },
+    {
+      "epoch": 0.043443137209907536,
+      "grad_norm": 5.694347381591797,
+      "learning_rate": 9.985886858358105e-06,
+      "loss": 2.9979,
+      "step": 166600
+    },
+    {
+      "epoch": 0.043495289835609706,
+      "grad_norm": 6.983020782470703,
+      "learning_rate": 9.98582402548395e-06,
+      "loss": 2.9455,
+      "step": 166800
+    },
+    {
+      "epoch": 0.043547442461311876,
+      "grad_norm": 5.710437297821045,
+      "learning_rate": 9.985761053250261e-06,
+      "loss": 2.9718,
+      "step": 167000
+    },
+    {
+      "epoch": 0.04359959508701405,
+      "grad_norm": 6.083151817321777,
+      "learning_rate": 9.985697941658803e-06,
+      "loss": 2.9651,
+      "step": 167200
+    },
+    {
+      "epoch": 0.04365174771271622,
+      "grad_norm": 6.084566116333008,
+      "learning_rate": 9.985634690711339e-06,
+      "loss": 2.9767,
+      "step": 167400
+    },
+    {
+      "epoch": 0.04370390033841839,
+      "grad_norm": 5.805023193359375,
+      "learning_rate": 9.985571300409637e-06,
+      "loss": 2.9637,
+      "step": 167600
+    },
+    {
+      "epoch": 0.04375605296412056,
+      "grad_norm": 6.559878826141357,
+      "learning_rate": 9.985507770755469e-06,
+      "loss": 2.9828,
+      "step": 167800
+    },
+    {
+      "epoch": 0.04380820558982273,
+      "grad_norm": 6.1664347648620605,
+      "learning_rate": 9.98544410175061e-06,
+      "loss": 2.967,
+      "step": 168000
+    },
+    {
+      "epoch": 0.0438603582155249,
+      "grad_norm": 6.0472564697265625,
+      "learning_rate": 9.985380293396839e-06,
+      "loss": 2.9384,
+      "step": 168200
+    },
+    {
+      "epoch": 0.04391251084122707,
+      "grad_norm": 6.2360029220581055,
+      "learning_rate": 9.985316345695941e-06,
+      "loss": 2.9687,
+      "step": 168400
+    },
+    {
+      "epoch": 0.04396466346692924,
+      "grad_norm": 6.714407920837402,
+      "learning_rate": 9.985252258649702e-06,
+      "loss": 2.9763,
+      "step": 168600
+    },
+    {
+      "epoch": 0.04401681609263141,
+      "grad_norm": 7.117247104644775,
+      "learning_rate": 9.985188032259917e-06,
+      "loss": 2.9749,
+      "step": 168800
+    },
+    {
+      "epoch": 0.04406896871833358,
+      "grad_norm": 5.8086981773376465,
+      "learning_rate": 9.985123666528376e-06,
+      "loss": 2.9501,
+      "step": 169000
+    },
+    {
+      "epoch": 0.04412112134403575,
+      "grad_norm": 6.109192848205566,
+      "learning_rate": 9.98505916145688e-06,
+      "loss": 3.0017,
+      "step": 169200
+    },
+    {
+      "epoch": 0.044173273969737914,
+      "grad_norm": 6.363615989685059,
+      "learning_rate": 9.984994517047234e-06,
+      "loss": 2.9637,
+      "step": 169400
+    },
+    {
+      "epoch": 0.044225426595440084,
+      "grad_norm": 5.879502773284912,
+      "learning_rate": 9.984929733301243e-06,
+      "loss": 2.9474,
+      "step": 169600
+    },
+    {
+      "epoch": 0.044277579221142255,
+      "grad_norm": 6.141928672790527,
+      "learning_rate": 9.984864810220717e-06,
+      "loss": 2.982,
+      "step": 169800
+    },
+    {
+      "epoch": 0.044329731846844425,
+      "grad_norm": 6.232560157775879,
+      "learning_rate": 9.984799747807473e-06,
+      "loss": 2.9348,
+      "step": 170000
+    },
+    {
+      "epoch": 0.044381884472546596,
+      "grad_norm": 6.826907634735107,
+      "learning_rate": 9.984734546063328e-06,
+      "loss": 3.0119,
+      "step": 170200
+    },
+    {
+      "epoch": 0.044434037098248766,
+      "grad_norm": 6.180233955383301,
+      "learning_rate": 9.984669204990104e-06,
+      "loss": 2.9685,
+      "step": 170400
+    },
+    {
+      "epoch": 0.044486189723950936,
+      "grad_norm": 6.817476272583008,
+      "learning_rate": 9.98460372458963e-06,
+      "loss": 2.9706,
+      "step": 170600
+    },
+    {
+      "epoch": 0.04453834234965311,
+      "grad_norm": 6.815362930297852,
+      "learning_rate": 9.984538104863732e-06,
+      "loss": 2.9585,
+      "step": 170800
+    },
+    {
+      "epoch": 0.04459049497535528,
+      "grad_norm": 6.279226779937744,
+      "learning_rate": 9.98447234581425e-06,
+      "loss": 2.9819,
+      "step": 171000
+    },
+    {
+      "epoch": 0.04464264760105745,
+      "grad_norm": 5.946432590484619,
+      "learning_rate": 9.984406447443015e-06,
+      "loss": 2.92,
+      "step": 171200
+    },
+    {
+      "epoch": 0.04469480022675962,
+      "grad_norm": 6.094913482666016,
+      "learning_rate": 9.984340409751875e-06,
+      "loss": 2.9627,
+      "step": 171400
+    },
+    {
+      "epoch": 0.04474695285246179,
+      "grad_norm": 6.792775630950928,
+      "learning_rate": 9.984274232742673e-06,
+      "loss": 2.9564,
+      "step": 171600
+    },
+    {
+      "epoch": 0.04479910547816396,
+      "grad_norm": 6.441338062286377,
+      "learning_rate": 9.984207916417257e-06,
+      "loss": 2.9794,
+      "step": 171800
+    },
+    {
+      "epoch": 0.04485125810386613,
+      "grad_norm": 6.825250625610352,
+      "learning_rate": 9.984141460777484e-06,
+      "loss": 2.9504,
+      "step": 172000
+    },
+    {
+      "epoch": 0.0449034107295683,
+      "grad_norm": 6.110622882843018,
+      "learning_rate": 9.98407486582521e-06,
+      "loss": 2.9836,
+      "step": 172200
+    },
+    {
+      "epoch": 0.04495556335527046,
+      "grad_norm": 7.110909461975098,
+      "learning_rate": 9.984008131562299e-06,
+      "loss": 2.9565,
+      "step": 172400
+    },
+    {
+      "epoch": 0.04500771598097263,
+      "grad_norm": 6.6831488609313965,
+      "learning_rate": 9.983941257990611e-06,
+      "loss": 2.9093,
+      "step": 172600
+    },
+    {
+      "epoch": 0.045059868606674804,
+      "grad_norm": 6.000646114349365,
+      "learning_rate": 9.983874245112019e-06,
+      "loss": 2.9815,
+      "step": 172800
+    },
+    {
+      "epoch": 0.045112021232376974,
+      "grad_norm": 5.678177356719971,
+      "learning_rate": 9.983807092928396e-06,
+      "loss": 3.0084,
+      "step": 173000
+    },
+    {
+      "epoch": 0.045164173858079144,
+      "grad_norm": 5.743597030639648,
+      "learning_rate": 9.983739801441618e-06,
+      "loss": 2.937,
+      "step": 173200
+    },
+    {
+      "epoch": 0.045216326483781315,
+      "grad_norm": 6.317743301391602,
+      "learning_rate": 9.983672370653565e-06,
+      "loss": 2.9439,
+      "step": 173400
+    },
+    {
+      "epoch": 0.045268479109483485,
+      "grad_norm": 6.100225448608398,
+      "learning_rate": 9.983604800566124e-06,
+      "loss": 2.9455,
+      "step": 173600
+    },
+    {
+      "epoch": 0.045320631735185656,
+      "grad_norm": 5.574644088745117,
+      "learning_rate": 9.98353709118118e-06,
+      "loss": 2.9886,
+      "step": 173800
+    },
+    {
+      "epoch": 0.045372784360887826,
+      "grad_norm": 6.011674880981445,
+      "learning_rate": 9.983469242500632e-06,
+      "loss": 2.9362,
+      "step": 174000
+    },
+    {
+      "epoch": 0.045424936986589996,
+      "grad_norm": 6.598609447479248,
+      "learning_rate": 9.98340125452637e-06,
+      "loss": 2.9851,
+      "step": 174200
+    },
+    {
+      "epoch": 0.04547708961229217,
+      "grad_norm": 7.741209030151367,
+      "learning_rate": 9.983333127260299e-06,
+      "loss": 2.917,
+      "step": 174400
+    },
+    {
+      "epoch": 0.04552924223799434,
+      "grad_norm": 6.492852687835693,
+      "learning_rate": 9.98326486070432e-06,
+      "loss": 2.936,
+      "step": 174600
+    },
+    {
+      "epoch": 0.04558139486369651,
+      "grad_norm": 6.744225978851318,
+      "learning_rate": 9.98319645486034e-06,
+      "loss": 2.953,
+      "step": 174800
+    },
+    {
+      "epoch": 0.04563354748939868,
+      "grad_norm": 6.86396598815918,
+      "learning_rate": 9.983127909730275e-06,
+      "loss": 2.9551,
+      "step": 175000
+    },
+    {
+      "epoch": 0.04568570011510085,
+      "grad_norm": 6.352712631225586,
+      "learning_rate": 9.983059225316042e-06,
+      "loss": 2.9466,
+      "step": 175200
+    },
+    {
+      "epoch": 0.04573785274080301,
+      "grad_norm": 6.346860408782959,
+      "learning_rate": 9.982990401619554e-06,
+      "loss": 2.9497,
+      "step": 175400
+    },
+    {
+      "epoch": 0.04579000536650518,
+      "grad_norm": 7.074236869812012,
+      "learning_rate": 9.982921438642739e-06,
+      "loss": 2.9437,
+      "step": 175600
+    },
+    {
+      "epoch": 0.04584215799220735,
+      "grad_norm": 5.964405536651611,
+      "learning_rate": 9.982852336387525e-06,
+      "loss": 2.9576,
+      "step": 175800
+    },
+    {
+      "epoch": 0.04589431061790952,
+      "grad_norm": 6.961352348327637,
+      "learning_rate": 9.982783094855844e-06,
+      "loss": 2.9587,
+      "step": 176000
+    },
+    {
+      "epoch": 0.04594646324361169,
+      "grad_norm": 6.538314342498779,
+      "learning_rate": 9.982713714049627e-06,
+      "loss": 2.996,
+      "step": 176200
+    },
+    {
+      "epoch": 0.045998615869313864,
+      "grad_norm": 5.937180519104004,
+      "learning_rate": 9.98264419397082e-06,
+      "loss": 2.9536,
+      "step": 176400
+    },
+    {
+      "epoch": 0.046050768495016034,
+      "grad_norm": 5.624795913696289,
+      "learning_rate": 9.98257453462136e-06,
+      "loss": 2.9593,
+      "step": 176600
+    },
+    {
+      "epoch": 0.046102921120718204,
+      "grad_norm": 6.610598087310791,
+      "learning_rate": 9.982504736003198e-06,
+      "loss": 2.9338,
+      "step": 176800
+    },
+    {
+      "epoch": 0.046155073746420375,
+      "grad_norm": 6.419288158416748,
+      "learning_rate": 9.982434798118283e-06,
+      "loss": 2.9388,
+      "step": 177000
+    },
+    {
+      "epoch": 0.046207226372122545,
+      "grad_norm": 6.405141830444336,
+      "learning_rate": 9.982364720968572e-06,
+      "loss": 2.9473,
+      "step": 177200
+    },
+    {
+      "epoch": 0.046259378997824716,
+      "grad_norm": 6.353682994842529,
+      "learning_rate": 9.98229450455602e-06,
+      "loss": 2.9132,
+      "step": 177400
+    },
+    {
+      "epoch": 0.046311531623526886,
+      "grad_norm": 5.974079608917236,
+      "learning_rate": 9.982224148882593e-06,
+      "loss": 2.9748,
+      "step": 177600
+    },
+    {
+      "epoch": 0.046363684249229056,
+      "grad_norm": 5.714940071105957,
+      "learning_rate": 9.982153653950254e-06,
+      "loss": 2.9446,
+      "step": 177800
+    },
+    {
+      "epoch": 0.04641583687493123,
+      "grad_norm": 6.13021183013916,
+      "learning_rate": 9.982083019760978e-06,
+      "loss": 2.9442,
+      "step": 178000
+    },
+    {
+      "epoch": 0.0464679895006334,
+      "grad_norm": 6.623483180999756,
+      "learning_rate": 9.982012246316737e-06,
+      "loss": 2.9269,
+      "step": 178200
+    },
+    {
+      "epoch": 0.04652014212633556,
+      "grad_norm": 6.2891364097595215,
+      "learning_rate": 9.981941333619509e-06,
+      "loss": 2.9979,
+      "step": 178400
+    },
+    {
+      "epoch": 0.04657229475203773,
+      "grad_norm": 6.528567314147949,
+      "learning_rate": 9.981870281671277e-06,
+      "loss": 2.9436,
+      "step": 178600
+    },
+    {
+      "epoch": 0.0466244473777399,
+      "grad_norm": 6.393155574798584,
+      "learning_rate": 9.981799090474024e-06,
+      "loss": 2.9763,
+      "step": 178800
+    },
+    {
+      "epoch": 0.04667660000344207,
+      "grad_norm": 6.487037181854248,
+      "learning_rate": 9.981727760029745e-06,
+      "loss": 2.9561,
+      "step": 179000
+    },
+    {
+      "epoch": 0.04672875262914424,
+      "grad_norm": 5.926519393920898,
+      "learning_rate": 9.981656290340429e-06,
+      "loss": 2.9715,
+      "step": 179200
+    },
+    {
+      "epoch": 0.04678090525484641,
+      "grad_norm": 6.406266689300537,
+      "learning_rate": 9.981584681408076e-06,
+      "loss": 2.9343,
+      "step": 179400
+    },
+    {
+      "epoch": 0.04683305788054858,
+      "grad_norm": 6.141971588134766,
+      "learning_rate": 9.981512933234688e-06,
+      "loss": 2.9437,
+      "step": 179600
+    },
+    {
+      "epoch": 0.04688521050625075,
+      "grad_norm": 5.708194732666016,
+      "learning_rate": 9.98144104582227e-06,
+      "loss": 2.927,
+      "step": 179800
+    },
+    {
+      "epoch": 0.046937363131952924,
+      "grad_norm": 6.583145618438721,
+      "learning_rate": 9.981369019172829e-06,
+      "loss": 2.9649,
+      "step": 180000
+    },
+    {
+      "epoch": 0.046989515757655094,
+      "grad_norm": 6.8080363273620605,
+      "learning_rate": 9.981296853288382e-06,
+      "loss": 2.9718,
+      "step": 180200
+    },
+    {
+      "epoch": 0.047041668383357264,
+      "grad_norm": 5.960728645324707,
+      "learning_rate": 9.981224548170942e-06,
+      "loss": 2.9552,
+      "step": 180400
+    },
+    {
+      "epoch": 0.047093821009059435,
+      "grad_norm": 6.71892786026001,
+      "learning_rate": 9.981152103822535e-06,
+      "loss": 2.878,
+      "step": 180600
+    },
+    {
+      "epoch": 0.047145973634761605,
+      "grad_norm": 6.228333950042725,
+      "learning_rate": 9.98107952024518e-06,
+      "loss": 2.9596,
+      "step": 180800
+    },
+    {
+      "epoch": 0.047198126260463776,
+      "grad_norm": 6.656292915344238,
+      "learning_rate": 9.981006797440913e-06,
+      "loss": 2.9251,
+      "step": 181000
+    },
+    {
+      "epoch": 0.04725027888616594,
+      "grad_norm": 6.142068386077881,
+      "learning_rate": 9.98093393541176e-06,
+      "loss": 2.9004,
+      "step": 181200
+    },
+    {
+      "epoch": 0.04730243151186811,
+      "grad_norm": 6.923412799835205,
+      "learning_rate": 9.980860934159761e-06,
+      "loss": 2.9182,
+      "step": 181400
+    },
+    {
+      "epoch": 0.04735458413757028,
+      "grad_norm": 7.025545597076416,
+      "learning_rate": 9.980787793686955e-06,
+      "loss": 2.9484,
+      "step": 181600
+    },
+    {
+      "epoch": 0.04740673676327245,
+      "grad_norm": 6.994638442993164,
+      "learning_rate": 9.980714513995389e-06,
+      "loss": 2.9093,
+      "step": 181800
+    },
+    {
+      "epoch": 0.04745888938897462,
+      "grad_norm": 6.052518367767334,
+      "learning_rate": 9.980641095087107e-06,
+      "loss": 2.9469,
+      "step": 182000
+    },
+    {
+      "epoch": 0.04751104201467679,
+      "grad_norm": 6.930753707885742,
+      "learning_rate": 9.980567536964166e-06,
+      "loss": 2.916,
+      "step": 182200
+    },
+    {
+      "epoch": 0.04756319464037896,
+      "grad_norm": 5.495729923248291,
+      "learning_rate": 9.980493839628618e-06,
+      "loss": 2.9307,
+      "step": 182400
+    },
+    {
+      "epoch": 0.04761534726608113,
+      "grad_norm": 5.72206974029541,
+      "learning_rate": 9.980420003082524e-06,
+      "loss": 2.9098,
+      "step": 182600
+    },
+    {
+      "epoch": 0.0476674998917833,
+      "grad_norm": 6.242524147033691,
+      "learning_rate": 9.980346027327948e-06,
+      "loss": 2.889,
+      "step": 182800
+    },
+    {
+      "epoch": 0.04771965251748547,
+      "grad_norm": 6.146420955657959,
+      "learning_rate": 9.980271912366959e-06,
+      "loss": 2.942,
+      "step": 183000
+    },
+    {
+      "epoch": 0.04777180514318764,
+      "grad_norm": 6.690853595733643,
+      "learning_rate": 9.980197658201627e-06,
+      "loss": 2.9214,
+      "step": 183200
+    },
+    {
+      "epoch": 0.04782395776888981,
+      "grad_norm": 5.759744644165039,
+      "learning_rate": 9.98012326483403e-06,
+      "loss": 2.9074,
+      "step": 183400
+    },
+    {
+      "epoch": 0.047876110394591984,
+      "grad_norm": 7.025085926055908,
+      "learning_rate": 9.980048732266243e-06,
+      "loss": 2.9566,
+      "step": 183600
+    },
+    {
+      "epoch": 0.047928263020294154,
+      "grad_norm": 6.400143623352051,
+      "learning_rate": 9.979974060500353e-06,
+      "loss": 2.9725,
+      "step": 183800
+    },
+    {
+      "epoch": 0.047980415645996324,
+      "grad_norm": 6.825561046600342,
+      "learning_rate": 9.979899249538445e-06,
+      "loss": 2.8797,
+      "step": 184000
+    },
+    {
+      "epoch": 0.04803256827169849,
+      "grad_norm": 6.444281578063965,
+      "learning_rate": 9.979824299382612e-06,
+      "loss": 2.9194,
+      "step": 184200
+    },
+    {
+      "epoch": 0.04808472089740066,
+      "grad_norm": 5.93040132522583,
+      "learning_rate": 9.979749210034948e-06,
+      "loss": 2.9068,
+      "step": 184400
+    },
+    {
+      "epoch": 0.04813687352310283,
+      "grad_norm": 5.90882682800293,
+      "learning_rate": 9.97967398149755e-06,
+      "loss": 2.894,
+      "step": 184600
+    },
+    {
+      "epoch": 0.048189026148805,
+      "grad_norm": 5.31546688079834,
+      "learning_rate": 9.979598613772523e-06,
+      "loss": 2.9244,
+      "step": 184800
+    },
+    {
+      "epoch": 0.04824117877450717,
+      "grad_norm": 6.048388957977295,
+      "learning_rate": 9.979523106861974e-06,
+      "loss": 2.8599,
+      "step": 185000
+    },
+    {
+      "epoch": 0.04829333140020934,
+      "grad_norm": 6.296876430511475,
+      "learning_rate": 9.979447460768012e-06,
+      "loss": 2.9213,
+      "step": 185200
+    },
+    {
+      "epoch": 0.04834548402591151,
+      "grad_norm": 6.221283435821533,
+      "learning_rate": 9.979371675492753e-06,
+      "loss": 2.9058,
+      "step": 185400
+    },
+    {
+      "epoch": 0.04839763665161368,
+      "grad_norm": 7.359024524688721,
+      "learning_rate": 9.979295751038313e-06,
+      "loss": 2.9192,
+      "step": 185600
+    },
+    {
+      "epoch": 0.04844978927731585,
+      "grad_norm": 6.607453346252441,
+      "learning_rate": 9.979219687406816e-06,
+      "loss": 3.0031,
+      "step": 185800
+    },
+    {
+      "epoch": 0.04850194190301802,
+      "grad_norm": 6.028778553009033,
+      "learning_rate": 9.979143484600387e-06,
+      "loss": 2.9068,
+      "step": 186000
+    },
+    {
+      "epoch": 0.04855409452872019,
+      "grad_norm": 5.630479335784912,
+      "learning_rate": 9.979067142621157e-06,
+      "loss": 2.9278,
+      "step": 186200
+    },
+    {
+      "epoch": 0.04860624715442236,
+      "grad_norm": 6.049453258514404,
+      "learning_rate": 9.978990661471257e-06,
+      "loss": 2.891,
+      "step": 186400
+    },
+    {
+      "epoch": 0.04865839978012453,
+      "grad_norm": 6.467658519744873,
+      "learning_rate": 9.97891404115283e-06,
+      "loss": 2.9207,
+      "step": 186600
+    },
+    {
+      "epoch": 0.0487105524058267,
+      "grad_norm": 6.942977428436279,
+      "learning_rate": 9.978837281668013e-06,
+      "loss": 2.9211,
+      "step": 186800
+    },
+    {
+      "epoch": 0.04876270503152887,
+      "grad_norm": 6.686459541320801,
+      "learning_rate": 9.978760383018953e-06,
+      "loss": 2.9207,
+      "step": 187000
+    },
+    {
+      "epoch": 0.04881485765723104,
+      "grad_norm": 6.126353740692139,
+      "learning_rate": 9.978683345207802e-06,
+      "loss": 2.9331,
+      "step": 187200
+    },
+    {
+      "epoch": 0.04886701028293321,
+      "grad_norm": 6.489861488342285,
+      "learning_rate": 9.97860616823671e-06,
+      "loss": 2.9333,
+      "step": 187400
+    },
+    {
+      "epoch": 0.04891916290863538,
+      "grad_norm": 6.95447301864624,
+      "learning_rate": 9.978528852107833e-06,
+      "loss": 2.9839,
+      "step": 187600
+    },
+    {
+      "epoch": 0.04897131553433755,
+      "grad_norm": 5.979517459869385,
+      "learning_rate": 9.978451396823334e-06,
+      "loss": 2.9585,
+      "step": 187800
+    },
+    {
+      "epoch": 0.04902346816003972,
+      "grad_norm": 6.67393684387207,
+      "learning_rate": 9.97837380238538e-06,
+      "loss": 2.9367,
+      "step": 188000
+    },
+    {
+      "epoch": 0.04907562078574189,
+      "grad_norm": 5.634337425231934,
+      "learning_rate": 9.978296068796138e-06,
+      "loss": 2.9027,
+      "step": 188200
+    },
+    {
+      "epoch": 0.04912777341144406,
+      "grad_norm": 6.696192264556885,
+      "learning_rate": 9.97821819605778e-06,
+      "loss": 2.8881,
+      "step": 188400
+    },
+    {
+      "epoch": 0.04917992603714623,
+      "grad_norm": 5.69584321975708,
+      "learning_rate": 9.978140184172482e-06,
+      "loss": 2.9149,
+      "step": 188600
+    },
+    {
+      "epoch": 0.0492320786628484,
+      "grad_norm": 6.718496322631836,
+      "learning_rate": 9.978062033142429e-06,
+      "loss": 2.8973,
+      "step": 188800
+    },
+    {
+      "epoch": 0.04928423128855057,
+      "grad_norm": 6.430253505706787,
+      "learning_rate": 9.977983742969798e-06,
+      "loss": 2.9127,
+      "step": 189000
+    },
+    {
+      "epoch": 0.04933638391425274,
+      "grad_norm": 6.769885063171387,
+      "learning_rate": 9.977905313656785e-06,
+      "loss": 2.9418,
+      "step": 189200
+    },
+    {
+      "epoch": 0.04938853653995491,
+      "grad_norm": 6.998106956481934,
+      "learning_rate": 9.977826745205578e-06,
+      "loss": 2.9484,
+      "step": 189400
+    },
+    {
+      "epoch": 0.04944068916565708,
+      "grad_norm": 6.269783020019531,
+      "learning_rate": 9.977748037618374e-06,
+      "loss": 2.9273,
+      "step": 189600
+    },
+    {
+      "epoch": 0.04949284179135925,
+      "grad_norm": 6.623769283294678,
+      "learning_rate": 9.977669190897372e-06,
+      "loss": 2.9459,
+      "step": 189800
+    },
+    {
+      "epoch": 0.04954499441706142,
+      "grad_norm": 6.445021152496338,
+      "learning_rate": 9.977590205044776e-06,
+      "loss": 2.932,
+      "step": 190000
+    },
+    {
+      "epoch": 0.049597147042763585,
+      "grad_norm": 6.26726770401001,
+      "learning_rate": 9.977511080062797e-06,
+      "loss": 2.9177,
+      "step": 190200
+    },
+    {
+      "epoch": 0.049649299668465756,
+      "grad_norm": 5.96261739730835,
+      "learning_rate": 9.977431815953642e-06,
+      "loss": 2.9476,
+      "step": 190400
+    },
+    {
+      "epoch": 0.049701452294167926,
+      "grad_norm": 6.362253189086914,
+      "learning_rate": 9.977352412719528e-06,
+      "loss": 2.9598,
+      "step": 190600
+    },
+    {
+      "epoch": 0.0497536049198701,
+      "grad_norm": 6.155995845794678,
+      "learning_rate": 9.977272870362676e-06,
+      "loss": 2.8895,
+      "step": 190800
+    },
+    {
+      "epoch": 0.04980575754557227,
+      "grad_norm": 6.854986667633057,
+      "learning_rate": 9.977193188885307e-06,
+      "loss": 2.9139,
+      "step": 191000
+    },
+    {
+      "epoch": 0.04985791017127444,
+      "grad_norm": 5.611915588378906,
+      "learning_rate": 9.97711336828965e-06,
+      "loss": 2.9157,
+      "step": 191200
+    },
+    {
+      "epoch": 0.04991006279697661,
+      "grad_norm": 6.135472774505615,
+      "learning_rate": 9.977033408577936e-06,
+      "loss": 2.8925,
+      "step": 191400
+    },
+    {
+      "epoch": 0.04996221542267878,
+      "grad_norm": 6.853118419647217,
+      "learning_rate": 9.976953309752401e-06,
+      "loss": 2.9362,
+      "step": 191600
+    },
+    {
+      "epoch": 0.05001436804838095,
+      "grad_norm": 6.357486248016357,
+      "learning_rate": 9.97687307181528e-06,
+      "loss": 2.9344,
+      "step": 191800
+    },
+    {
+      "epoch": 0.05006652067408312,
+      "grad_norm": 6.773046016693115,
+      "learning_rate": 9.97679269476882e-06,
+      "loss": 2.9625,
+      "step": 192000
+    },
+    {
+      "epoch": 0.05011867329978529,
+      "grad_norm": 6.14764404296875,
+      "learning_rate": 9.976712178615264e-06,
+      "loss": 2.9519,
+      "step": 192200
+    },
+    {
+      "epoch": 0.05017082592548746,
+      "grad_norm": 6.529992580413818,
+      "learning_rate": 9.976631523356866e-06,
+      "loss": 2.9013,
+      "step": 192400
+    },
+    {
+      "epoch": 0.05022297855118963,
+      "grad_norm": 5.915500164031982,
+      "learning_rate": 9.976550728995877e-06,
+      "loss": 2.9096,
+      "step": 192600
+    },
+    {
+      "epoch": 0.0502751311768918,
+      "grad_norm": 6.39959716796875,
+      "learning_rate": 9.976469795534557e-06,
+      "loss": 2.9478,
+      "step": 192800
+    },
+    {
+      "epoch": 0.050327283802593964,
+      "grad_norm": 6.220710277557373,
+      "learning_rate": 9.976388722975169e-06,
+      "loss": 2.9282,
+      "step": 193000
+    },
+    {
+      "epoch": 0.050379436428296134,
+      "grad_norm": 5.63620662689209,
+      "learning_rate": 9.976307511319979e-06,
+      "loss": 2.9344,
+      "step": 193200
+    },
+    {
+      "epoch": 0.050431589053998305,
+      "grad_norm": 6.4898600578308105,
+      "learning_rate": 9.976226160571254e-06,
+      "loss": 2.9223,
+      "step": 193400
+    },
+    {
+      "epoch": 0.050483741679700475,
+      "grad_norm": 5.993746757507324,
+      "learning_rate": 9.97614467073127e-06,
+      "loss": 2.9314,
+      "step": 193600
+    },
+    {
+      "epoch": 0.050535894305402645,
+      "grad_norm": 6.51394510269165,
+      "learning_rate": 9.976063041802306e-06,
+      "loss": 2.9471,
+      "step": 193800
+    },
+    {
+      "epoch": 0.050588046931104816,
+      "grad_norm": 5.766570091247559,
+      "learning_rate": 9.975981273786643e-06,
+      "loss": 2.954,
+      "step": 194000
+    },
+    {
+      "epoch": 0.050640199556806986,
+      "grad_norm": 6.516523838043213,
+      "learning_rate": 9.975899366686564e-06,
+      "loss": 2.9265,
+      "step": 194200
+    },
+    {
+      "epoch": 0.05069235218250916,
+      "grad_norm": 6.924788475036621,
+      "learning_rate": 9.975817320504362e-06,
+      "loss": 2.8861,
+      "step": 194400
+    },
+    {
+      "epoch": 0.05074450480821133,
+      "grad_norm": 6.611748218536377,
+      "learning_rate": 9.975735135242328e-06,
+      "loss": 2.9391,
+      "step": 194600
+    },
+    {
+      "epoch": 0.0507966574339135,
+      "grad_norm": 6.635772228240967,
+      "learning_rate": 9.975652810902759e-06,
+      "loss": 2.9112,
+      "step": 194800
+    },
+    {
+      "epoch": 0.05084881005961567,
+      "grad_norm": 6.5964741706848145,
+      "learning_rate": 9.975570347487958e-06,
+      "loss": 2.8749,
+      "step": 195000
+    },
+    {
+      "epoch": 0.05090096268531784,
+      "grad_norm": 5.7668986320495605,
+      "learning_rate": 9.975487745000228e-06,
+      "loss": 2.9284,
+      "step": 195200
+    },
+    {
+      "epoch": 0.05095311531102001,
+      "grad_norm": 6.671104907989502,
+      "learning_rate": 9.975405003441877e-06,
+      "loss": 2.924,
+      "step": 195400
+    },
+    {
+      "epoch": 0.05100526793672218,
+      "grad_norm": 6.551841735839844,
+      "learning_rate": 9.975322122815221e-06,
+      "loss": 2.9267,
+      "step": 195600
+    },
+    {
+      "epoch": 0.05105742056242435,
+      "grad_norm": 6.587834358215332,
+      "learning_rate": 9.975239103122576e-06,
+      "loss": 2.9144,
+      "step": 195800
+    },
+    {
+      "epoch": 0.05110957318812651,
+      "grad_norm": 6.745659828186035,
+      "learning_rate": 9.97515594436626e-06,
+      "loss": 2.9037,
+      "step": 196000
+    },
+    {
+      "epoch": 0.05116172581382868,
+      "grad_norm": 6.943085670471191,
+      "learning_rate": 9.975072646548597e-06,
+      "loss": 2.9468,
+      "step": 196200
+    },
+    {
+      "epoch": 0.051213878439530854,
+      "grad_norm": 6.360815525054932,
+      "learning_rate": 9.97498920967192e-06,
+      "loss": 2.9095,
+      "step": 196400
+    },
+    {
+      "epoch": 0.051266031065233024,
+      "grad_norm": 6.90670108795166,
+      "learning_rate": 9.974905633738559e-06,
+      "loss": 2.8835,
+      "step": 196600
+    },
+    {
+      "epoch": 0.051318183690935194,
+      "grad_norm": 5.816658020019531,
+      "learning_rate": 9.974821918750846e-06,
+      "loss": 2.9313,
+      "step": 196800
+    },
+    {
+      "epoch": 0.051370336316637365,
+      "grad_norm": 6.233535289764404,
+      "learning_rate": 9.974738064711125e-06,
+      "loss": 2.9081,
+      "step": 197000
+    },
+    {
+      "epoch": 0.051422488942339535,
+      "grad_norm": 6.274383068084717,
+      "learning_rate": 9.97465407162174e-06,
+      "loss": 2.9143,
+      "step": 197200
+    },
+    {
+      "epoch": 0.051474641568041705,
+      "grad_norm": 6.0712738037109375,
+      "learning_rate": 9.974569939485038e-06,
+      "loss": 2.9224,
+      "step": 197400
+    },
+    {
+      "epoch": 0.051526794193743876,
+      "grad_norm": 6.007919788360596,
+      "learning_rate": 9.974485668303369e-06,
+      "loss": 2.9035,
+      "step": 197600
+    },
+    {
+      "epoch": 0.051578946819446046,
+      "grad_norm": 6.441709518432617,
+      "learning_rate": 9.974401258079092e-06,
+      "loss": 2.938,
+      "step": 197800
+    },
+    {
+      "epoch": 0.05163109944514822,
+      "grad_norm": 5.86509895324707,
+      "learning_rate": 9.974316708814562e-06,
+      "loss": 2.9291,
+      "step": 198000
+    },
+    {
+      "epoch": 0.05168325207085039,
+      "grad_norm": 6.994657039642334,
+      "learning_rate": 9.974232020512144e-06,
+      "loss": 2.8551,
+      "step": 198200
+    },
+    {
+      "epoch": 0.05173540469655256,
+      "grad_norm": 6.415815353393555,
+      "learning_rate": 9.974147193174205e-06,
+      "loss": 2.9468,
+      "step": 198400
+    },
+    {
+      "epoch": 0.05178755732225473,
+      "grad_norm": 5.555944919586182,
+      "learning_rate": 9.97406222680312e-06,
+      "loss": 2.8942,
+      "step": 198600
+    },
+    {
+      "epoch": 0.0518397099479569,
+      "grad_norm": 6.089178085327148,
+      "learning_rate": 9.973977121401258e-06,
+      "loss": 2.9051,
+      "step": 198800
+    },
+    {
+      "epoch": 0.05189186257365906,
+      "grad_norm": 6.0938496589660645,
+      "learning_rate": 9.973891876971e-06,
+      "loss": 2.9038,
+      "step": 199000
+    },
+    {
+      "epoch": 0.05194401519936123,
+      "grad_norm": 6.132476806640625,
+      "learning_rate": 9.97380649351473e-06,
+      "loss": 2.912,
+      "step": 199200
+    },
+    {
+      "epoch": 0.0519961678250634,
+      "grad_norm": 6.650424957275391,
+      "learning_rate": 9.973720971034834e-06,
+      "loss": 2.9345,
+      "step": 199400
+    },
+    {
+      "epoch": 0.05204832045076557,
+      "grad_norm": 5.777918338775635,
+      "learning_rate": 9.9736353095337e-06,
+      "loss": 2.9157,
+      "step": 199600
+    },
+    {
+      "epoch": 0.05210047307646774,
+      "grad_norm": 6.441890239715576,
+      "learning_rate": 9.973549509013727e-06,
+      "loss": 2.8985,
+      "step": 199800
+    },
+    {
+      "epoch": 0.052152625702169914,
+      "grad_norm": 6.862900733947754,
+      "learning_rate": 9.973463569477309e-06,
+      "loss": 2.9326,
+      "step": 200000
+    },
+    {
+      "epoch": 0.052204778327872084,
+      "grad_norm": 6.423320770263672,
+      "learning_rate": 9.973377490926848e-06,
+      "loss": 2.8986,
+      "step": 200200
+    },
+    {
+      "epoch": 0.052256930953574254,
+      "grad_norm": 6.303603649139404,
+      "learning_rate": 9.973291273364754e-06,
+      "loss": 2.885,
+      "step": 200400
+    },
+    {
+      "epoch": 0.052309083579276425,
+      "grad_norm": 6.71627140045166,
+      "learning_rate": 9.973204916793433e-06,
+      "loss": 2.8799,
+      "step": 200600
+    },
+    {
+      "epoch": 0.052361236204978595,
+      "grad_norm": 6.168727397918701,
+      "learning_rate": 9.9731184212153e-06,
+      "loss": 2.8999,
+      "step": 200800
+    },
+    {
+      "epoch": 0.052413388830680765,
+      "grad_norm": 6.115705966949463,
+      "learning_rate": 9.973031786632773e-06,
+      "loss": 2.9156,
+      "step": 201000
+    },
+    {
+      "epoch": 0.052465541456382936,
+      "grad_norm": 5.8287200927734375,
+      "learning_rate": 9.972945013048275e-06,
+      "loss": 2.8945,
+      "step": 201200
+    },
+    {
+      "epoch": 0.052517694082085106,
+      "grad_norm": 6.276069164276123,
+      "learning_rate": 9.972858100464229e-06,
+      "loss": 2.9383,
+      "step": 201400
+    },
+    {
+      "epoch": 0.05256984670778728,
+      "grad_norm": 5.857385635375977,
+      "learning_rate": 9.972771048883065e-06,
+      "loss": 2.895,
+      "step": 201600
+    },
+    {
+      "epoch": 0.05262199933348944,
+      "grad_norm": 7.052547931671143,
+      "learning_rate": 9.972683858307217e-06,
+      "loss": 2.8879,
+      "step": 201800
+    },
+    {
+      "epoch": 0.05267415195919161,
+      "grad_norm": 6.282922267913818,
+      "learning_rate": 9.97259652873912e-06,
+      "loss": 2.8945,
+      "step": 202000
+    },
+    {
+      "epoch": 0.05272630458489378,
+      "grad_norm": 6.320275783538818,
+      "learning_rate": 9.972509060181218e-06,
+      "loss": 2.9122,
+      "step": 202200
+    },
+    {
+      "epoch": 0.05277845721059595,
+      "grad_norm": 5.877612113952637,
+      "learning_rate": 9.972421452635954e-06,
+      "loss": 2.8862,
+      "step": 202400
+    },
+    {
+      "epoch": 0.05283060983629812,
+      "grad_norm": 6.659661769866943,
+      "learning_rate": 9.972333706105777e-06,
+      "loss": 2.9418,
+      "step": 202600
+    },
+    {
+      "epoch": 0.05288276246200029,
+      "grad_norm": 6.702484130859375,
+      "learning_rate": 9.972245820593138e-06,
+      "loss": 2.9173,
+      "step": 202800
+    },
+    {
+      "epoch": 0.05293491508770246,
+      "grad_norm": 6.761118412017822,
+      "learning_rate": 9.972157796100497e-06,
+      "loss": 2.9208,
+      "step": 203000
+    },
+    {
+      "epoch": 0.05298706771340463,
+      "grad_norm": 5.8187642097473145,
+      "learning_rate": 9.972069632630312e-06,
+      "loss": 2.8982,
+      "step": 203200
+    },
+    {
+      "epoch": 0.0530392203391068,
+      "grad_norm": 6.023776531219482,
+      "learning_rate": 9.971981330185047e-06,
+      "loss": 2.9081,
+      "step": 203400
+    },
+    {
+      "epoch": 0.053091372964808974,
+      "grad_norm": 6.4479451179504395,
+      "learning_rate": 9.971892888767172e-06,
+      "loss": 2.9146,
+      "step": 203600
+    },
+    {
+      "epoch": 0.053143525590511144,
+      "grad_norm": 6.822802543640137,
+      "learning_rate": 9.971804308379156e-06,
+      "loss": 2.8861,
+      "step": 203800
+    },
+    {
+      "epoch": 0.053195678216213314,
+      "grad_norm": 6.120795726776123,
+      "learning_rate": 9.971715589023478e-06,
+      "loss": 2.8868,
+      "step": 204000
+    },
+    {
+      "epoch": 0.053247830841915485,
+      "grad_norm": 5.959011077880859,
+      "learning_rate": 9.971626730702617e-06,
+      "loss": 2.886,
+      "step": 204200
+    },
+    {
+      "epoch": 0.053299983467617655,
+      "grad_norm": 5.855833530426025,
+      "learning_rate": 9.971537733419057e-06,
+      "loss": 2.9224,
+      "step": 204400
+    },
+    {
+      "epoch": 0.053352136093319825,
+      "grad_norm": 6.570242881774902,
+      "learning_rate": 9.971448597175284e-06,
+      "loss": 2.9392,
+      "step": 204600
+    },
+    {
+      "epoch": 0.05340428871902199,
+      "grad_norm": 5.748819828033447,
+      "learning_rate": 9.97135932197379e-06,
+      "loss": 2.8955,
+      "step": 204800
+    },
+    {
+      "epoch": 0.05345644134472416,
+      "grad_norm": 6.4593000411987305,
+      "learning_rate": 9.97126990781707e-06,
+      "loss": 2.878,
+      "step": 205000
+    },
+    {
+      "epoch": 0.05350859397042633,
+      "grad_norm": 5.999743461608887,
+      "learning_rate": 9.971180354707627e-06,
+      "loss": 2.9366,
+      "step": 205200
+    },
+    {
+      "epoch": 0.0535607465961285,
+      "grad_norm": 6.419029235839844,
+      "learning_rate": 9.97109066264796e-06,
+      "loss": 2.9191,
+      "step": 205400
+    },
+    {
+      "epoch": 0.05361289922183067,
+      "grad_norm": 6.093469142913818,
+      "learning_rate": 9.971000831640576e-06,
+      "loss": 2.925,
+      "step": 205600
+    },
+    {
+      "epoch": 0.05366505184753284,
+      "grad_norm": 6.597023010253906,
+      "learning_rate": 9.970910861687988e-06,
+      "loss": 2.905,
+      "step": 205800
+    },
+    {
+      "epoch": 0.05371720447323501,
+      "grad_norm": 6.599836349487305,
+      "learning_rate": 9.97082075279271e-06,
+      "loss": 2.8901,
+      "step": 206000
+    },
+    {
+      "epoch": 0.05376935709893718,
+      "grad_norm": 7.024128437042236,
+      "learning_rate": 9.970730504957258e-06,
+      "loss": 2.8819,
+      "step": 206200
+    },
+    {
+      "epoch": 0.05382150972463935,
+      "grad_norm": 5.909964561462402,
+      "learning_rate": 9.970640118184158e-06,
+      "loss": 2.9062,
+      "step": 206400
+    },
+    {
+      "epoch": 0.05387366235034152,
+      "grad_norm": 6.118581771850586,
+      "learning_rate": 9.970549592475936e-06,
+      "loss": 2.9099,
+      "step": 206600
+    },
+    {
+      "epoch": 0.05392581497604369,
+      "grad_norm": 5.8928327560424805,
+      "learning_rate": 9.970458927835122e-06,
+      "loss": 2.8966,
+      "step": 206800
+    },
+    {
+      "epoch": 0.05397796760174586,
+      "grad_norm": 6.539867877960205,
+      "learning_rate": 9.970368124264249e-06,
+      "loss": 2.9375,
+      "step": 207000
+    },
+    {
+      "epoch": 0.054030120227448034,
+      "grad_norm": 6.851650714874268,
+      "learning_rate": 9.970277181765858e-06,
+      "loss": 2.863,
+      "step": 207200
+    },
+    {
+      "epoch": 0.054082272853150204,
+      "grad_norm": 6.393698215484619,
+      "learning_rate": 9.970186100342486e-06,
+      "loss": 2.9317,
+      "step": 207400
+    },
+    {
+      "epoch": 0.054134425478852374,
+      "grad_norm": 5.992748260498047,
+      "learning_rate": 9.970094879996683e-06,
+      "loss": 2.8793,
+      "step": 207600
+    },
+    {
+      "epoch": 0.05418657810455454,
+      "grad_norm": 5.8636298179626465,
+      "learning_rate": 9.970003520730997e-06,
+      "loss": 2.925,
+      "step": 207800
+    },
+    {
+      "epoch": 0.05423873073025671,
+      "grad_norm": 6.160617351531982,
+      "learning_rate": 9.96991202254798e-06,
+      "loss": 2.9411,
+      "step": 208000
+    },
+    {
+      "epoch": 0.05429088335595888,
+      "grad_norm": 7.532545566558838,
+      "learning_rate": 9.969820385450195e-06,
+      "loss": 2.8771,
+      "step": 208200
+    },
+    {
+      "epoch": 0.05434303598166105,
+      "grad_norm": 6.879082202911377,
+      "learning_rate": 9.969728609440197e-06,
+      "loss": 2.9334,
+      "step": 208400
+    },
+    {
+      "epoch": 0.05439518860736322,
+      "grad_norm": 6.399229049682617,
+      "learning_rate": 9.969636694520556e-06,
+      "loss": 2.8777,
+      "step": 208600
+    },
+    {
+      "epoch": 0.05444734123306539,
+      "grad_norm": 6.303351402282715,
+      "learning_rate": 9.969544640693838e-06,
+      "loss": 2.8949,
+      "step": 208800
+    },
+    {
+      "epoch": 0.05449949385876756,
+      "grad_norm": 6.815507888793945,
+      "learning_rate": 9.969452447962617e-06,
+      "loss": 2.876,
+      "step": 209000
+    },
+    {
+      "epoch": 0.05455164648446973,
+      "grad_norm": 6.317996978759766,
+      "learning_rate": 9.969360116329472e-06,
+      "loss": 2.8672,
+      "step": 209200
+    },
+    {
+      "epoch": 0.0546037991101719,
+      "grad_norm": 6.783117294311523,
+      "learning_rate": 9.96926764579698e-06,
+      "loss": 2.8434,
+      "step": 209400
+    },
+    {
+      "epoch": 0.05465595173587407,
+      "grad_norm": 6.445932865142822,
+      "learning_rate": 9.969175036367728e-06,
+      "loss": 2.9331,
+      "step": 209600
+    },
+    {
+      "epoch": 0.05470810436157624,
+      "grad_norm": 6.53240966796875,
+      "learning_rate": 9.969082288044304e-06,
+      "loss": 2.9213,
+      "step": 209800
+    },
+    {
+      "epoch": 0.05476025698727841,
+      "grad_norm": 6.389587879180908,
+      "learning_rate": 9.968989400829301e-06,
+      "loss": 2.9206,
+      "step": 210000
+    },
+    {
+      "epoch": 0.05481240961298058,
+      "grad_norm": 6.929988384246826,
+      "learning_rate": 9.968896374725314e-06,
+      "loss": 2.8684,
+      "step": 210200
+    },
+    {
+      "epoch": 0.05486456223868275,
+      "grad_norm": 6.338830471038818,
+      "learning_rate": 9.968803209734944e-06,
+      "loss": 2.8601,
+      "step": 210400
+    },
+    {
+      "epoch": 0.05491671486438492,
+      "grad_norm": 6.379602909088135,
+      "learning_rate": 9.968709905860796e-06,
+      "loss": 2.8824,
+      "step": 210600
+    },
+    {
+      "epoch": 0.05496886749008709,
+      "grad_norm": 6.1573710441589355,
+      "learning_rate": 9.968616463105476e-06,
+      "loss": 2.85,
+      "step": 210800
+    },
+    {
+      "epoch": 0.05502102011578926,
+      "grad_norm": 5.6570611000061035,
+      "learning_rate": 9.968522881471599e-06,
+      "loss": 2.8345,
+      "step": 211000
+    },
+    {
+      "epoch": 0.05507317274149143,
+      "grad_norm": 6.473025798797607,
+      "learning_rate": 9.968429160961776e-06,
+      "loss": 2.9062,
+      "step": 211200
+    },
+    {
+      "epoch": 0.0551253253671936,
+      "grad_norm": 5.675154209136963,
+      "learning_rate": 9.968335301578629e-06,
+      "loss": 2.9005,
+      "step": 211400
+    },
+    {
+      "epoch": 0.05517747799289577,
+      "grad_norm": 6.401944160461426,
+      "learning_rate": 9.968241303324783e-06,
+      "loss": 2.8383,
+      "step": 211600
+    },
+    {
+      "epoch": 0.05522963061859794,
+      "grad_norm": 5.85010290145874,
+      "learning_rate": 9.968147166202864e-06,
+      "loss": 2.8892,
+      "step": 211800
+    },
+    {
+      "epoch": 0.05528178324430011,
+      "grad_norm": 5.4949517250061035,
+      "learning_rate": 9.968052890215502e-06,
+      "loss": 2.8813,
+      "step": 212000
+    },
+    {
+      "epoch": 0.05533393587000228,
+      "grad_norm": 6.284722328186035,
+      "learning_rate": 9.967958475365334e-06,
+      "loss": 2.877,
+      "step": 212200
+    },
+    {
+      "epoch": 0.05538608849570445,
+      "grad_norm": 6.463416576385498,
+      "learning_rate": 9.967863921654998e-06,
+      "loss": 2.8947,
+      "step": 212400
+    },
+    {
+      "epoch": 0.05543824112140662,
+      "grad_norm": 6.696000576019287,
+      "learning_rate": 9.967769229087138e-06,
+      "loss": 2.8707,
+      "step": 212600
+    },
+    {
+      "epoch": 0.05549039374710879,
+      "grad_norm": 6.760770320892334,
+      "learning_rate": 9.9676743976644e-06,
+      "loss": 2.8864,
+      "step": 212800
+    },
+    {
+      "epoch": 0.05554254637281096,
+      "grad_norm": 5.885373115539551,
+      "learning_rate": 9.967579427389434e-06,
+      "loss": 2.8919,
+      "step": 213000
+    },
+    {
+      "epoch": 0.05559469899851313,
+      "grad_norm": 5.777275085449219,
+      "learning_rate": 9.967484318264895e-06,
+      "loss": 2.8508,
+      "step": 213200
+    },
+    {
+      "epoch": 0.0556468516242153,
+      "grad_norm": 6.474213123321533,
+      "learning_rate": 9.967389070293442e-06,
+      "loss": 2.875,
+      "step": 213400
+    },
+    {
+      "epoch": 0.055699004249917465,
+      "grad_norm": 6.556694030761719,
+      "learning_rate": 9.967293683477737e-06,
+      "loss": 2.9205,
+      "step": 213600
+    },
+    {
+      "epoch": 0.055751156875619635,
+      "grad_norm": 6.6564202308654785,
+      "learning_rate": 9.967198157820445e-06,
+      "loss": 2.8501,
+      "step": 213800
+    },
+    {
+      "epoch": 0.055803309501321806,
+      "grad_norm": 6.465703964233398,
+      "learning_rate": 9.967102493324239e-06,
+      "loss": 2.9072,
+      "step": 214000
+    },
+    {
+      "epoch": 0.055855462127023976,
+      "grad_norm": 6.2116780281066895,
+      "learning_rate": 9.96700668999179e-06,
+      "loss": 2.8542,
+      "step": 214200
+    },
+    {
+      "epoch": 0.05590761475272615,
+      "grad_norm": 6.195470333099365,
+      "learning_rate": 9.966910747825775e-06,
+      "loss": 2.8891,
+      "step": 214400
+    },
+    {
+      "epoch": 0.05595976737842832,
+      "grad_norm": 6.002094745635986,
+      "learning_rate": 9.96681466682888e-06,
+      "loss": 2.8885,
+      "step": 214600
+    },
+    {
+      "epoch": 0.05601192000413049,
+      "grad_norm": 6.3701653480529785,
+      "learning_rate": 9.966718447003787e-06,
+      "loss": 2.9023,
+      "step": 214800
+    },
+    {
+      "epoch": 0.05606407262983266,
+      "grad_norm": 6.410655975341797,
+      "learning_rate": 9.966622088353189e-06,
+      "loss": 2.9003,
+      "step": 215000
+    },
+    {
+      "epoch": 0.05611622525553483,
+      "grad_norm": 6.293005466461182,
+      "learning_rate": 9.966525590879774e-06,
+      "loss": 2.8997,
+      "step": 215200
+    },
+    {
+      "epoch": 0.056168377881237,
+      "grad_norm": 7.010239124298096,
+      "learning_rate": 9.966428954586243e-06,
+      "loss": 2.9075,
+      "step": 215400
+    },
+    {
+      "epoch": 0.05622053050693917,
+      "grad_norm": 7.058774948120117,
+      "learning_rate": 9.966332179475296e-06,
+      "loss": 2.8649,
+      "step": 215600
+    },
+    {
+      "epoch": 0.05627268313264134,
+      "grad_norm": 6.797397613525391,
+      "learning_rate": 9.966235265549637e-06,
+      "loss": 2.8828,
+      "step": 215800
+    },
+    {
+      "epoch": 0.05632483575834351,
+      "grad_norm": 6.849272727966309,
+      "learning_rate": 9.96613821281198e-06,
+      "loss": 2.8903,
+      "step": 216000
+    },
+    {
+      "epoch": 0.05637698838404568,
+      "grad_norm": 5.813775062561035,
+      "learning_rate": 9.96604102126503e-06,
+      "loss": 2.8662,
+      "step": 216200
+    },
+    {
+      "epoch": 0.05642914100974785,
+      "grad_norm": 6.535881996154785,
+      "learning_rate": 9.965943690911509e-06,
+      "loss": 2.8643,
+      "step": 216400
+    },
+    {
+      "epoch": 0.056481293635450014,
+      "grad_norm": 6.4796576499938965,
+      "learning_rate": 9.965846221754135e-06,
+      "loss": 2.8863,
+      "step": 216600
+    },
+    {
+      "epoch": 0.056533446261152184,
+      "grad_norm": 6.315826892852783,
+      "learning_rate": 9.965748613795633e-06,
+      "loss": 2.8721,
+      "step": 216800
+    },
+    {
+      "epoch": 0.056585598886854355,
+      "grad_norm": 6.6356635093688965,
+      "learning_rate": 9.965650867038732e-06,
+      "loss": 2.8823,
+      "step": 217000
+    },
+    {
+      "epoch": 0.056637751512556525,
+      "grad_norm": 6.826320648193359,
+      "learning_rate": 9.965552981486163e-06,
+      "loss": 2.8792,
+      "step": 217200
+    },
+    {
+      "epoch": 0.056689904138258695,
+      "grad_norm": 5.991021633148193,
+      "learning_rate": 9.965454957140665e-06,
+      "loss": 2.8717,
+      "step": 217400
+    },
+    {
+      "epoch": 0.056742056763960866,
+      "grad_norm": 5.6707048416137695,
+      "learning_rate": 9.965356794004974e-06,
+      "loss": 2.898,
+      "step": 217600
+    },
+    {
+      "epoch": 0.056794209389663036,
+      "grad_norm": 6.135042667388916,
+      "learning_rate": 9.965258492081835e-06,
+      "loss": 2.862,
+      "step": 217800
+    },
+    {
+      "epoch": 0.05684636201536521,
+      "grad_norm": 6.86878776550293,
+      "learning_rate": 9.965160051373996e-06,
+      "loss": 2.8626,
+      "step": 218000
+    },
+    {
+      "epoch": 0.05689851464106738,
+      "grad_norm": 6.301908016204834,
+      "learning_rate": 9.96506147188421e-06,
+      "loss": 2.865,
+      "step": 218200
+    },
+    {
+      "epoch": 0.05695066726676955,
+      "grad_norm": 6.453223705291748,
+      "learning_rate": 9.96496275361523e-06,
+      "loss": 2.8613,
+      "step": 218400
+    },
+    {
+      "epoch": 0.05700281989247172,
+      "grad_norm": 6.500892639160156,
+      "learning_rate": 9.964863896569816e-06,
+      "loss": 2.8533,
+      "step": 218600
+    },
+    {
+      "epoch": 0.05705497251817389,
+      "grad_norm": 5.618515491485596,
+      "learning_rate": 9.964764900750731e-06,
+      "loss": 2.872,
+      "step": 218800
+    },
+    {
+      "epoch": 0.05710712514387606,
+      "grad_norm": 6.596179485321045,
+      "learning_rate": 9.964665766160744e-06,
+      "loss": 2.8752,
+      "step": 219000
+    },
+    {
+      "epoch": 0.05715927776957823,
+      "grad_norm": 6.309721946716309,
+      "learning_rate": 9.964566492802623e-06,
+      "loss": 2.8882,
+      "step": 219200
+    },
+    {
+      "epoch": 0.0572114303952804,
+      "grad_norm": 6.222306251525879,
+      "learning_rate": 9.964467080679145e-06,
+      "loss": 2.8676,
+      "step": 219400
+    },
+    {
+      "epoch": 0.05726358302098256,
+      "grad_norm": 6.931632041931152,
+      "learning_rate": 9.964367529793086e-06,
+      "loss": 2.8618,
+      "step": 219600
+    },
+    {
+      "epoch": 0.05731573564668473,
+      "grad_norm": 6.057002544403076,
+      "learning_rate": 9.964267840147232e-06,
+      "loss": 2.8378,
+      "step": 219800
+    },
+    {
+      "epoch": 0.057367888272386904,
+      "grad_norm": 6.065264701843262,
+      "learning_rate": 9.964168011744367e-06,
+      "loss": 2.9028,
+      "step": 220000
+    },
+    {
+      "epoch": 0.057420040898089074,
+      "grad_norm": 6.861353874206543,
+      "learning_rate": 9.964068044587283e-06,
+      "loss": 2.8933,
+      "step": 220200
+    },
+    {
+      "epoch": 0.057472193523791244,
+      "grad_norm": 6.993216037750244,
+      "learning_rate": 9.963967938678774e-06,
+      "loss": 2.8741,
+      "step": 220400
+    },
+    {
+      "epoch": 0.057524346149493415,
+      "grad_norm": 6.941470146179199,
+      "learning_rate": 9.963867694021637e-06,
+      "loss": 2.8534,
+      "step": 220600
+    },
+    {
+      "epoch": 0.057576498775195585,
+      "grad_norm": 6.341226100921631,
+      "learning_rate": 9.963767310618673e-06,
+      "loss": 2.8837,
+      "step": 220800
+    },
+    {
+      "epoch": 0.057628651400897755,
+      "grad_norm": 6.165623188018799,
+      "learning_rate": 9.96366678847269e-06,
+      "loss": 2.9066,
+      "step": 221000
+    },
+    {
+      "epoch": 0.057680804026599926,
+      "grad_norm": 5.758754730224609,
+      "learning_rate": 9.963566127586497e-06,
+      "loss": 2.8871,
+      "step": 221200
+    },
+    {
+      "epoch": 0.057732956652302096,
+      "grad_norm": 6.3193206787109375,
+      "learning_rate": 9.963465327962907e-06,
+      "loss": 2.852,
+      "step": 221400
+    },
+    {
+      "epoch": 0.05778510927800427,
+      "grad_norm": 6.58033561706543,
+      "learning_rate": 9.963364389604739e-06,
+      "loss": 2.8514,
+      "step": 221600
+    },
+    {
+      "epoch": 0.05783726190370644,
+      "grad_norm": 5.843617916107178,
+      "learning_rate": 9.963263312514812e-06,
+      "loss": 2.8494,
+      "step": 221800
+    },
+    {
+      "epoch": 0.05788941452940861,
+      "grad_norm": 6.093498706817627,
+      "learning_rate": 9.963162096695954e-06,
+      "loss": 2.9139,
+      "step": 222000
+    },
+    {
+      "epoch": 0.05794156715511078,
+      "grad_norm": 6.31635856628418,
+      "learning_rate": 9.963060742150992e-06,
+      "loss": 2.8631,
+      "step": 222200
+    },
+    {
+      "epoch": 0.05799371978081295,
+      "grad_norm": 6.643308639526367,
+      "learning_rate": 9.96295924888276e-06,
+      "loss": 2.9046,
+      "step": 222400
+    },
+    {
+      "epoch": 0.05804587240651511,
+      "grad_norm": 6.348966121673584,
+      "learning_rate": 9.962857616894095e-06,
+      "loss": 2.87,
+      "step": 222600
+    },
+    {
+      "epoch": 0.05809802503221728,
+      "grad_norm": 6.411705017089844,
+      "learning_rate": 9.962755846187834e-06,
+      "loss": 2.8712,
+      "step": 222800
+    },
+    {
+      "epoch": 0.05815017765791945,
+      "grad_norm": 6.178708553314209,
+      "learning_rate": 9.962653936766828e-06,
+      "loss": 2.8583,
+      "step": 223000
+    },
+    {
+      "epoch": 0.05820233028362162,
+      "grad_norm": 6.825628757476807,
+      "learning_rate": 9.962551888633923e-06,
+      "loss": 2.8627,
+      "step": 223200
+    },
+    {
+      "epoch": 0.05825448290932379,
+      "grad_norm": 5.721690654754639,
+      "learning_rate": 9.96244970179197e-06,
+      "loss": 2.8824,
+      "step": 223400
+    },
+    {
+      "epoch": 0.058306635535025964,
+      "grad_norm": 6.0849127769470215,
+      "learning_rate": 9.962347376243824e-06,
+      "loss": 2.8654,
+      "step": 223600
+    },
+    {
+      "epoch": 0.058358788160728134,
+      "grad_norm": 6.45623254776001,
+      "learning_rate": 9.962244911992347e-06,
+      "loss": 2.9017,
+      "step": 223800
+    },
+    {
+      "epoch": 0.058410940786430304,
+      "grad_norm": 6.159735679626465,
+      "learning_rate": 9.962142309040405e-06,
+      "loss": 2.9074,
+      "step": 224000
+    },
+    {
+      "epoch": 0.058463093412132475,
+      "grad_norm": 6.511104106903076,
+      "learning_rate": 9.962039567390863e-06,
+      "loss": 2.8768,
+      "step": 224200
+    },
+    {
+      "epoch": 0.058515246037834645,
+      "grad_norm": 6.441580772399902,
+      "learning_rate": 9.961936687046595e-06,
+      "loss": 2.8928,
+      "step": 224400
+    },
+    {
+      "epoch": 0.058567398663536815,
+      "grad_norm": 5.743595123291016,
+      "learning_rate": 9.961833668010473e-06,
+      "loss": 2.9014,
+      "step": 224600
+    },
+    {
+      "epoch": 0.058619551289238986,
+      "grad_norm": 6.290615558624268,
+      "learning_rate": 9.961730510285379e-06,
+      "loss": 2.8954,
+      "step": 224800
+    },
+    {
+      "epoch": 0.058671703914941156,
+      "grad_norm": 6.227867126464844,
+      "learning_rate": 9.961627213874198e-06,
+      "loss": 2.858,
+      "step": 225000
+    },
+    {
+      "epoch": 0.05872385654064333,
+      "grad_norm": 6.234389305114746,
+      "learning_rate": 9.961523778779814e-06,
+      "loss": 2.8905,
+      "step": 225200
+    },
+    {
+      "epoch": 0.05877600916634549,
+      "grad_norm": 6.24118709564209,
+      "learning_rate": 9.961420205005118e-06,
+      "loss": 2.8632,
+      "step": 225400
+    },
+    {
+      "epoch": 0.05882816179204766,
+      "grad_norm": 6.4152750968933105,
+      "learning_rate": 9.96131649255301e-06,
+      "loss": 2.857,
+      "step": 225600
+    },
+    {
+      "epoch": 0.05888031441774983,
+      "grad_norm": 7.344254493713379,
+      "learning_rate": 9.961212641426384e-06,
+      "loss": 2.8742,
+      "step": 225800
+    },
+    {
+      "epoch": 0.058932467043452,
+      "grad_norm": 6.300400257110596,
+      "learning_rate": 9.961108651628144e-06,
+      "loss": 2.893,
+      "step": 226000
+    },
+    {
+      "epoch": 0.05898461966915417,
+      "grad_norm": 5.992100715637207,
+      "learning_rate": 9.961004523161197e-06,
+      "loss": 2.8256,
+      "step": 226200
+    },
+    {
+      "epoch": 0.05903677229485634,
+      "grad_norm": 5.90500545501709,
+      "learning_rate": 9.960900256028453e-06,
+      "loss": 2.9001,
+      "step": 226400
+    },
+    {
+      "epoch": 0.05908892492055851,
+      "grad_norm": 6.329569339752197,
+      "learning_rate": 9.960795850232827e-06,
+      "loss": 2.8679,
+      "step": 226600
+    },
+    {
+      "epoch": 0.05914107754626068,
+      "grad_norm": 6.701547622680664,
+      "learning_rate": 9.960691305777235e-06,
+      "loss": 2.8826,
+      "step": 226800
+    },
+    {
+      "epoch": 0.05919323017196285,
+      "grad_norm": 7.074747085571289,
+      "learning_rate": 9.960586622664603e-06,
+      "loss": 2.8816,
+      "step": 227000
+    },
+    {
+      "epoch": 0.059245382797665024,
+      "grad_norm": 5.853117942810059,
+      "learning_rate": 9.960481800897855e-06,
+      "loss": 2.8941,
+      "step": 227200
+    },
+    {
+      "epoch": 0.059297535423367194,
+      "grad_norm": 6.21675968170166,
+      "learning_rate": 9.960376840479922e-06,
+      "loss": 2.9143,
+      "step": 227400
+    },
+    {
+      "epoch": 0.059349688049069364,
+      "grad_norm": 6.886922836303711,
+      "learning_rate": 9.960271741413737e-06,
+      "loss": 2.8215,
+      "step": 227600
+    },
+    {
+      "epoch": 0.059401840674771535,
+      "grad_norm": 6.113399028778076,
+      "learning_rate": 9.960166503702234e-06,
+      "loss": 2.8652,
+      "step": 227800
+    },
+    {
+      "epoch": 0.059453993300473705,
+      "grad_norm": 6.7419304847717285,
+      "learning_rate": 9.96006112734836e-06,
+      "loss": 2.8997,
+      "step": 228000
+    },
+    {
+      "epoch": 0.059506145926175875,
+      "grad_norm": 5.719422817230225,
+      "learning_rate": 9.959955612355059e-06,
+      "loss": 2.8397,
+      "step": 228200
+    },
+    {
+      "epoch": 0.05955829855187804,
+      "grad_norm": 6.3306732177734375,
+      "learning_rate": 9.959849958725278e-06,
+      "loss": 2.8576,
+      "step": 228400
+    },
+    {
+      "epoch": 0.05961045117758021,
+      "grad_norm": 6.876055717468262,
+      "learning_rate": 9.959744166461973e-06,
+      "loss": 2.8646,
+      "step": 228600
+    },
+    {
+      "epoch": 0.05966260380328238,
+      "grad_norm": 6.750349998474121,
+      "learning_rate": 9.9596382355681e-06,
+      "loss": 2.8445,
+      "step": 228800
+    },
+    {
+      "epoch": 0.05971475642898455,
+      "grad_norm": 5.776258945465088,
+      "learning_rate": 9.959532166046619e-06,
+      "loss": 2.8702,
+      "step": 229000
+    },
+    {
+      "epoch": 0.05976690905468672,
+      "grad_norm": 6.348162651062012,
+      "learning_rate": 9.959425957900497e-06,
+      "loss": 2.8475,
+      "step": 229200
+    },
+    {
+      "epoch": 0.05981906168038889,
+      "grad_norm": 6.766139507293701,
+      "learning_rate": 9.959319611132698e-06,
+      "loss": 2.8688,
+      "step": 229400
+    },
+    {
+      "epoch": 0.05987121430609106,
+      "grad_norm": 6.318508625030518,
+      "learning_rate": 9.959213125746198e-06,
+      "loss": 2.8977,
+      "step": 229600
+    },
+    {
+      "epoch": 0.05992336693179323,
+      "grad_norm": 6.733820915222168,
+      "learning_rate": 9.959106501743975e-06,
+      "loss": 2.8547,
+      "step": 229800
+    },
+    {
+      "epoch": 0.0599755195574954,
+      "grad_norm": 7.468286037445068,
+      "learning_rate": 9.958999739129006e-06,
+      "loss": 2.9255,
+      "step": 230000
+    },
+    {
+      "epoch": 0.06002767218319757,
+      "grad_norm": 6.781874179840088,
+      "learning_rate": 9.958892837904277e-06,
+      "loss": 2.8473,
+      "step": 230200
+    },
+    {
+      "epoch": 0.06007982480889974,
+      "grad_norm": 6.273848056793213,
+      "learning_rate": 9.958785798072775e-06,
+      "loss": 2.8736,
+      "step": 230400
+    },
+    {
+      "epoch": 0.06013197743460191,
+      "grad_norm": 6.0436320304870605,
+      "learning_rate": 9.958678619637491e-06,
+      "loss": 2.8747,
+      "step": 230600
+    },
+    {
+      "epoch": 0.060184130060304084,
+      "grad_norm": 6.5935378074646,
+      "learning_rate": 9.958571302601425e-06,
+      "loss": 2.8568,
+      "step": 230800
+    },
+    {
+      "epoch": 0.060236282686006254,
+      "grad_norm": 6.040035724639893,
+      "learning_rate": 9.958463846967572e-06,
+      "loss": 2.8901,
+      "step": 231000
+    },
+    {
+      "epoch": 0.060288435311708424,
+      "grad_norm": 6.27731466293335,
+      "learning_rate": 9.958356252738937e-06,
+      "loss": 2.8297,
+      "step": 231200
+    },
+    {
+      "epoch": 0.06034058793741059,
+      "grad_norm": 6.9736199378967285,
+      "learning_rate": 9.958248519918527e-06,
+      "loss": 2.8432,
+      "step": 231400
+    },
+    {
+      "epoch": 0.06039274056311276,
+      "grad_norm": 6.537654876708984,
+      "learning_rate": 9.958140648509354e-06,
+      "loss": 2.907,
+      "step": 231600
+    },
+    {
+      "epoch": 0.06044489318881493,
+      "grad_norm": 6.521362781524658,
+      "learning_rate": 9.958032638514435e-06,
+      "loss": 2.832,
+      "step": 231800
+    },
+    {
+      "epoch": 0.0604970458145171,
+      "grad_norm": 6.518667697906494,
+      "learning_rate": 9.957924489936783e-06,
+      "loss": 2.8741,
+      "step": 232000
+    },
+    {
+      "epoch": 0.06054919844021927,
+      "grad_norm": 6.39730167388916,
+      "learning_rate": 9.957816202779429e-06,
+      "loss": 2.841,
+      "step": 232200
+    },
+    {
+      "epoch": 0.06060135106592144,
+      "grad_norm": 6.15274715423584,
+      "learning_rate": 9.957707777045392e-06,
+      "loss": 2.8499,
+      "step": 232400
+    },
+    {
+      "epoch": 0.06065350369162361,
+      "grad_norm": 6.834228515625,
+      "learning_rate": 9.957599212737707e-06,
+      "loss": 2.8709,
+      "step": 232600
+    },
+    {
+      "epoch": 0.06070565631732578,
+      "grad_norm": 6.3925251960754395,
+      "learning_rate": 9.957490509859409e-06,
+      "loss": 2.8808,
+      "step": 232800
+    },
+    {
+      "epoch": 0.06075780894302795,
+      "grad_norm": 6.036910533905029,
+      "learning_rate": 9.957381668413535e-06,
+      "loss": 2.8857,
+      "step": 233000
+    },
+    {
+      "epoch": 0.06080996156873012,
+      "grad_norm": 6.22501802444458,
+      "learning_rate": 9.957272688403126e-06,
+      "loss": 2.8427,
+      "step": 233200
+    },
+    {
+      "epoch": 0.06086211419443229,
+      "grad_norm": 5.928175449371338,
+      "learning_rate": 9.95716356983123e-06,
+      "loss": 2.8687,
+      "step": 233400
+    },
+    {
+      "epoch": 0.06091426682013446,
+      "grad_norm": 6.415838241577148,
+      "learning_rate": 9.957054312700897e-06,
+      "loss": 2.8375,
+      "step": 233600
+    },
+    {
+      "epoch": 0.06096641944583663,
+      "grad_norm": 6.5070013999938965,
+      "learning_rate": 9.956944917015179e-06,
+      "loss": 2.8811,
+      "step": 233800
+    },
+    {
+      "epoch": 0.0610185720715388,
+      "grad_norm": 5.6311821937561035,
+      "learning_rate": 9.956835382777137e-06,
+      "loss": 2.8823,
+      "step": 234000
+    },
+    {
+      "epoch": 0.061070724697240966,
+      "grad_norm": 5.70000696182251,
+      "learning_rate": 9.956725709989829e-06,
+      "loss": 2.881,
+      "step": 234200
+    },
+    {
+      "epoch": 0.06112287732294314,
+      "grad_norm": 6.649327754974365,
+      "learning_rate": 9.956615898656322e-06,
+      "loss": 2.8617,
+      "step": 234400
+    },
+    {
+      "epoch": 0.06117502994864531,
+      "grad_norm": 6.379722595214844,
+      "learning_rate": 9.956505948779687e-06,
+      "loss": 2.8547,
+      "step": 234600
+    },
+    {
+      "epoch": 0.06122718257434748,
+      "grad_norm": 6.3186116218566895,
+      "learning_rate": 9.956395860362992e-06,
+      "loss": 2.8306,
+      "step": 234800
+    },
+    {
+      "epoch": 0.06127933520004965,
+      "grad_norm": 6.583649158477783,
+      "learning_rate": 9.956285633409322e-06,
+      "loss": 2.8816,
+      "step": 235000
+    },
+    {
+      "epoch": 0.06133148782575182,
+      "grad_norm": 6.297317981719971,
+      "learning_rate": 9.956175267921752e-06,
+      "loss": 2.8649,
+      "step": 235200
+    },
+    {
+      "epoch": 0.06138364045145399,
+      "grad_norm": 7.1909871101379395,
+      "learning_rate": 9.956064763903368e-06,
+      "loss": 2.8988,
+      "step": 235400
+    },
+    {
+      "epoch": 0.06143579307715616,
+      "grad_norm": 6.038774490356445,
+      "learning_rate": 9.955954121357262e-06,
+      "loss": 2.8585,
+      "step": 235600
+    },
+    {
+      "epoch": 0.06148794570285833,
+      "grad_norm": 6.198418140411377,
+      "learning_rate": 9.95584334028652e-06,
+      "loss": 2.8393,
+      "step": 235800
+    },
+    {
+      "epoch": 0.0615400983285605,
+      "grad_norm": 6.438840389251709,
+      "learning_rate": 9.955732420694244e-06,
+      "loss": 2.8219,
+      "step": 236000
+    },
+    {
+      "epoch": 0.06159225095426267,
+      "grad_norm": 6.378972053527832,
+      "learning_rate": 9.955621362583533e-06,
+      "loss": 2.8501,
+      "step": 236200
+    },
+    {
+      "epoch": 0.06164440357996484,
+      "grad_norm": 6.684047222137451,
+      "learning_rate": 9.95551016595749e-06,
+      "loss": 2.8735,
+      "step": 236400
+    },
+    {
+      "epoch": 0.06169655620566701,
+      "grad_norm": 6.312347412109375,
+      "learning_rate": 9.955398830819225e-06,
+      "loss": 2.852,
+      "step": 236600
+    },
+    {
+      "epoch": 0.06174870883136918,
+      "grad_norm": 6.520938873291016,
+      "learning_rate": 9.955287357171848e-06,
+      "loss": 2.8494,
+      "step": 236800
+    },
+    {
+      "epoch": 0.06180086145707135,
+      "grad_norm": 6.856761932373047,
+      "learning_rate": 9.955175745018477e-06,
+      "loss": 2.8285,
+      "step": 237000
+    },
+    {
+      "epoch": 0.061853014082773515,
+      "grad_norm": 6.1591620445251465,
+      "learning_rate": 9.955063994362229e-06,
+      "loss": 2.8291,
+      "step": 237200
+    },
+    {
+      "epoch": 0.061905166708475685,
+      "grad_norm": 5.8291850090026855,
+      "learning_rate": 9.95495210520623e-06,
+      "loss": 2.8288,
+      "step": 237400
+    },
+    {
+      "epoch": 0.061957319334177856,
+      "grad_norm": 6.742002964019775,
+      "learning_rate": 9.954840077553604e-06,
+      "loss": 2.8435,
+      "step": 237600
+    },
+    {
+      "epoch": 0.062009471959880026,
+      "grad_norm": 6.073098659515381,
+      "learning_rate": 9.954727911407489e-06,
+      "loss": 2.8344,
+      "step": 237800
+    },
+    {
+      "epoch": 0.0620616245855822,
+      "grad_norm": 7.29561710357666,
+      "learning_rate": 9.954615606771012e-06,
+      "loss": 2.8352,
+      "step": 238000
+    },
+    {
+      "epoch": 0.06211377721128437,
+      "grad_norm": 6.251704216003418,
+      "learning_rate": 9.954503163647319e-06,
+      "loss": 2.8576,
+      "step": 238200
+    },
+    {
+      "epoch": 0.06216592983698654,
+      "grad_norm": 5.640725612640381,
+      "learning_rate": 9.954390582039545e-06,
+      "loss": 2.8439,
+      "step": 238400
+    },
+    {
+      "epoch": 0.06221808246268871,
+      "grad_norm": 6.10871696472168,
+      "learning_rate": 9.954277861950847e-06,
+      "loss": 2.8231,
+      "step": 238600
+    },
+    {
+      "epoch": 0.06227023508839088,
+      "grad_norm": 7.135128974914551,
+      "learning_rate": 9.954165003384367e-06,
+      "loss": 2.857,
+      "step": 238800
+    },
+    {
+      "epoch": 0.06232238771409305,
+      "grad_norm": 6.5773773193359375,
+      "learning_rate": 9.954052006343264e-06,
+      "loss": 2.8329,
+      "step": 239000
+    },
+    {
+      "epoch": 0.06237454033979522,
+      "grad_norm": 6.622820854187012,
+      "learning_rate": 9.953938870830696e-06,
+      "loss": 2.8687,
+      "step": 239200
+    },
+    {
+      "epoch": 0.06242669296549739,
+      "grad_norm": 6.241696834564209,
+      "learning_rate": 9.953825596849823e-06,
+      "loss": 2.8318,
+      "step": 239400
+    },
+    {
+      "epoch": 0.06247884559119956,
+      "grad_norm": 5.970136642456055,
+      "learning_rate": 9.953712184403811e-06,
+      "loss": 2.8651,
+      "step": 239600
+    },
+    {
+      "epoch": 0.06253099821690172,
+      "grad_norm": 5.942645072937012,
+      "learning_rate": 9.953598633495835e-06,
+      "loss": 2.8463,
+      "step": 239800
+    },
+    {
+      "epoch": 0.0625831508426039,
+      "grad_norm": 5.821661949157715,
+      "learning_rate": 9.953484944129064e-06,
+      "loss": 2.8955,
+      "step": 240000
+    },
+    {
+      "epoch": 0.06263530346830606,
+      "grad_norm": 6.606387138366699,
+      "learning_rate": 9.953371116306678e-06,
+      "loss": 2.8606,
+      "step": 240200
+    },
+    {
+      "epoch": 0.06268745609400823,
+      "grad_norm": 6.661681175231934,
+      "learning_rate": 9.953257150031857e-06,
+      "loss": 2.8399,
+      "step": 240400
+    },
+    {
+      "epoch": 0.0627396087197104,
+      "grad_norm": 6.761116027832031,
+      "learning_rate": 9.953143045307788e-06,
+      "loss": 2.8202,
+      "step": 240600
+    },
+    {
+      "epoch": 0.06279176134541258,
+      "grad_norm": 6.392307281494141,
+      "learning_rate": 9.953028802137658e-06,
+      "loss": 2.8542,
+      "step": 240800
+    },
+    {
+      "epoch": 0.06284391397111475,
+      "grad_norm": 6.313865661621094,
+      "learning_rate": 9.952914420524663e-06,
+      "loss": 2.8388,
+      "step": 241000
+    },
+    {
+      "epoch": 0.06289606659681692,
+      "grad_norm": 6.371045112609863,
+      "learning_rate": 9.952799900472e-06,
+      "loss": 2.8371,
+      "step": 241200
+    },
+    {
+      "epoch": 0.06294821922251909,
+      "grad_norm": 5.998501777648926,
+      "learning_rate": 9.952685241982867e-06,
+      "loss": 2.7792,
+      "step": 241400
+    },
+    {
+      "epoch": 0.06300037184822126,
+      "grad_norm": 6.239037036895752,
+      "learning_rate": 9.952570445060472e-06,
+      "loss": 2.8392,
+      "step": 241600
+    },
+    {
+      "epoch": 0.06305252447392343,
+      "grad_norm": 6.2942304611206055,
+      "learning_rate": 9.95245550970802e-06,
+      "loss": 2.8724,
+      "step": 241800
+    },
+    {
+      "epoch": 0.0631046770996256,
+      "grad_norm": 6.233669281005859,
+      "learning_rate": 9.95234043592873e-06,
+      "loss": 2.8326,
+      "step": 242000
+    },
+    {
+      "epoch": 0.06315682972532777,
+      "grad_norm": 6.2911577224731445,
+      "learning_rate": 9.952225223725814e-06,
+      "loss": 2.8716,
+      "step": 242200
+    },
+    {
+      "epoch": 0.06320898235102994,
+      "grad_norm": 6.646729946136475,
+      "learning_rate": 9.95210987310249e-06,
+      "loss": 2.8266,
+      "step": 242400
+    },
+    {
+      "epoch": 0.06326113497673211,
+      "grad_norm": 6.116643905639648,
+      "learning_rate": 9.951994384061988e-06,
+      "loss": 2.8853,
+      "step": 242600
+    },
+    {
+      "epoch": 0.06331328760243428,
+      "grad_norm": 6.70005464553833,
+      "learning_rate": 9.951878756607532e-06,
+      "loss": 2.8247,
+      "step": 242800
+    },
+    {
+      "epoch": 0.06336544022813645,
+      "grad_norm": 5.81242036819458,
+      "learning_rate": 9.951762990742356e-06,
+      "loss": 2.8165,
+      "step": 243000
+    },
+    {
+      "epoch": 0.06341759285383862,
+      "grad_norm": 6.2581329345703125,
+      "learning_rate": 9.951647086469694e-06,
+      "loss": 2.883,
+      "step": 243200
+    },
+    {
+      "epoch": 0.06346974547954079,
+      "grad_norm": 6.079429626464844,
+      "learning_rate": 9.951531043792785e-06,
+      "loss": 2.8643,
+      "step": 243400
+    },
+    {
+      "epoch": 0.06352189810524296,
+      "grad_norm": 6.490660667419434,
+      "learning_rate": 9.951414862714877e-06,
+      "loss": 2.8605,
+      "step": 243600
+    },
+    {
+      "epoch": 0.06357405073094513,
+      "grad_norm": 6.3001909255981445,
+      "learning_rate": 9.951298543239213e-06,
+      "loss": 2.8898,
+      "step": 243800
+    },
+    {
+      "epoch": 0.0636262033566473,
+      "grad_norm": 6.262456893920898,
+      "learning_rate": 9.951182085369044e-06,
+      "loss": 2.8698,
+      "step": 244000
+    },
+    {
+      "epoch": 0.06367835598234947,
+      "grad_norm": 7.45053243637085,
+      "learning_rate": 9.951065489107628e-06,
+      "loss": 2.8388,
+      "step": 244200
+    },
+    {
+      "epoch": 0.06373050860805164,
+      "grad_norm": 6.364867210388184,
+      "learning_rate": 9.950948754458222e-06,
+      "loss": 2.8592,
+      "step": 244400
+    },
+    {
+      "epoch": 0.0637826612337538,
+      "grad_norm": 6.993120193481445,
+      "learning_rate": 9.950831881424092e-06,
+      "loss": 2.824,
+      "step": 244600
+    },
+    {
+      "epoch": 0.06383481385945597,
+      "grad_norm": 6.528937339782715,
+      "learning_rate": 9.9507148700085e-06,
+      "loss": 2.8447,
+      "step": 244800
+    },
+    {
+      "epoch": 0.06388696648515814,
+      "grad_norm": 6.995550155639648,
+      "learning_rate": 9.950597720214721e-06,
+      "loss": 2.8018,
+      "step": 245000
+    },
+    {
+      "epoch": 0.06393911911086031,
+      "grad_norm": 6.352587699890137,
+      "learning_rate": 9.950480432046025e-06,
+      "loss": 2.8619,
+      "step": 245200
+    },
+    {
+      "epoch": 0.06399127173656248,
+      "grad_norm": 6.427020072937012,
+      "learning_rate": 9.950363005505695e-06,
+      "loss": 2.854,
+      "step": 245400
+    },
+    {
+      "epoch": 0.06404342436226465,
+      "grad_norm": 6.2600603103637695,
+      "learning_rate": 9.950245440597011e-06,
+      "loss": 2.8246,
+      "step": 245600
+    },
+    {
+      "epoch": 0.06409557698796682,
+      "grad_norm": 5.7311906814575195,
+      "learning_rate": 9.950127737323258e-06,
+      "loss": 2.8132,
+      "step": 245800
+    },
+    {
+      "epoch": 0.06414772961366899,
+      "grad_norm": 5.9210920333862305,
+      "learning_rate": 9.950009895687727e-06,
+      "loss": 2.8446,
+      "step": 246000
+    },
+    {
+      "epoch": 0.06419988223937116,
+      "grad_norm": 5.30364990234375,
+      "learning_rate": 9.949891915693712e-06,
+      "loss": 2.8175,
+      "step": 246200
+    },
+    {
+      "epoch": 0.06425203486507333,
+      "grad_norm": 6.701354503631592,
+      "learning_rate": 9.949773797344511e-06,
+      "loss": 2.8503,
+      "step": 246400
+    },
+    {
+      "epoch": 0.0643041874907755,
+      "grad_norm": 6.559759140014648,
+      "learning_rate": 9.949655540643425e-06,
+      "loss": 2.8999,
+      "step": 246600
+    },
+    {
+      "epoch": 0.06435634011647767,
+      "grad_norm": 6.051071643829346,
+      "learning_rate": 9.949537145593759e-06,
+      "loss": 2.8777,
+      "step": 246800
+    },
+    {
+      "epoch": 0.06440849274217984,
+      "grad_norm": 5.7465972900390625,
+      "learning_rate": 9.949418612198822e-06,
+      "loss": 2.8398,
+      "step": 247000
+    },
+    {
+      "epoch": 0.06446064536788201,
+      "grad_norm": 6.2163872718811035,
+      "learning_rate": 9.94929994046193e-06,
+      "loss": 2.8245,
+      "step": 247200
+    },
+    {
+      "epoch": 0.06451279799358418,
+      "grad_norm": 6.6872639656066895,
+      "learning_rate": 9.949181130386396e-06,
+      "loss": 2.8519,
+      "step": 247400
+    },
+    {
+      "epoch": 0.06456495061928635,
+      "grad_norm": 5.855112552642822,
+      "learning_rate": 9.949062181975544e-06,
+      "loss": 2.8286,
+      "step": 247600
+    },
+    {
+      "epoch": 0.06461710324498852,
+      "grad_norm": 6.136778354644775,
+      "learning_rate": 9.948943095232697e-06,
+      "loss": 2.8164,
+      "step": 247800
+    },
+    {
+      "epoch": 0.0646692558706907,
+      "grad_norm": 5.666250705718994,
+      "learning_rate": 9.948823870161184e-06,
+      "loss": 2.8247,
+      "step": 248000
+    },
+    {
+      "epoch": 0.06472140849639287,
+      "grad_norm": 6.700359344482422,
+      "learning_rate": 9.948704506764336e-06,
+      "loss": 2.7989,
+      "step": 248200
+    },
+    {
+      "epoch": 0.06477356112209504,
+      "grad_norm": 6.757811546325684,
+      "learning_rate": 9.948585005045495e-06,
+      "loss": 2.868,
+      "step": 248400
+    },
+    {
+      "epoch": 0.0648257137477972,
+      "grad_norm": 6.108405113220215,
+      "learning_rate": 9.948465365007995e-06,
+      "loss": 2.85,
+      "step": 248600
+    },
+    {
+      "epoch": 0.06487786637349938,
+      "grad_norm": 6.390014171600342,
+      "learning_rate": 9.948345586655181e-06,
+      "loss": 2.8826,
+      "step": 248800
+    },
+    {
+      "epoch": 0.06493001899920155,
+      "grad_norm": 6.4399733543396,
+      "learning_rate": 9.948225669990404e-06,
+      "loss": 2.7958,
+      "step": 249000
+    },
+    {
+      "epoch": 0.06498217162490372,
+      "grad_norm": 6.0152058601379395,
+      "learning_rate": 9.948105615017014e-06,
+      "loss": 2.8615,
+      "step": 249200
+    },
+    {
+      "epoch": 0.06503432425060589,
+      "grad_norm": 6.281088352203369,
+      "learning_rate": 9.947985421738366e-06,
+      "loss": 2.8323,
+      "step": 249400
+    },
+    {
+      "epoch": 0.06508647687630806,
+      "grad_norm": 7.041154384613037,
+      "learning_rate": 9.94786509015782e-06,
+      "loss": 2.812,
+      "step": 249600
+    },
+    {
+      "epoch": 0.06513862950201023,
+      "grad_norm": 6.751956462860107,
+      "learning_rate": 9.94774462027874e-06,
+      "loss": 2.8045,
+      "step": 249800
+    },
+    {
+      "epoch": 0.0651907821277124,
+      "grad_norm": 6.53183126449585,
+      "learning_rate": 9.947624012104494e-06,
+      "loss": 2.8629,
+      "step": 250000
+    },
+    {
+      "epoch": 0.06524293475341457,
+      "grad_norm": 6.1272969245910645,
+      "learning_rate": 9.947503265638449e-06,
+      "loss": 2.8009,
+      "step": 250200
+    },
+    {
+      "epoch": 0.06529508737911673,
+      "grad_norm": 6.2636942863464355,
+      "learning_rate": 9.947382380883985e-06,
+      "loss": 2.8502,
+      "step": 250400
+    },
+    {
+      "epoch": 0.0653472400048189,
+      "grad_norm": 5.978120803833008,
+      "learning_rate": 9.94726135784448e-06,
+      "loss": 2.8581,
+      "step": 250600
+    },
+    {
+      "epoch": 0.06539939263052107,
+      "grad_norm": 6.1595778465271,
+      "learning_rate": 9.947140196523312e-06,
+      "loss": 2.8445,
+      "step": 250800
+    },
+    {
+      "epoch": 0.06545154525622324,
+      "grad_norm": 6.8264923095703125,
+      "learning_rate": 9.947018896923875e-06,
+      "loss": 2.8439,
+      "step": 251000
+    },
+    {
+      "epoch": 0.06550369788192541,
+      "grad_norm": 5.948680877685547,
+      "learning_rate": 9.946897459049553e-06,
+      "loss": 2.8293,
+      "step": 251200
+    },
+    {
+      "epoch": 0.06555585050762758,
+      "grad_norm": 6.657248497009277,
+      "learning_rate": 9.946775882903745e-06,
+      "loss": 2.8726,
+      "step": 251400
+    },
+    {
+      "epoch": 0.06560800313332975,
+      "grad_norm": 5.8180155754089355,
+      "learning_rate": 9.946654168489847e-06,
+      "loss": 2.8181,
+      "step": 251600
+    },
+    {
+      "epoch": 0.06566015575903192,
+      "grad_norm": 6.403216361999512,
+      "learning_rate": 9.946532315811262e-06,
+      "loss": 2.806,
+      "step": 251800
+    },
+    {
+      "epoch": 0.06571230838473409,
+      "grad_norm": 6.092254638671875,
+      "learning_rate": 9.946410324871394e-06,
+      "loss": 2.8513,
+      "step": 252000
+    },
+    {
+      "epoch": 0.06576446101043626,
+      "grad_norm": 5.990058422088623,
+      "learning_rate": 9.946288195673654e-06,
+      "loss": 2.8147,
+      "step": 252200
+    },
+    {
+      "epoch": 0.06581661363613843,
+      "grad_norm": 6.334691047668457,
+      "learning_rate": 9.946165928221456e-06,
+      "loss": 2.8539,
+      "step": 252400
+    },
+    {
+      "epoch": 0.0658687662618406,
+      "grad_norm": 5.580613613128662,
+      "learning_rate": 9.946043522518217e-06,
+      "loss": 2.8393,
+      "step": 252600
+    },
+    {
+      "epoch": 0.06592091888754277,
+      "grad_norm": 6.962843418121338,
+      "learning_rate": 9.945920978567357e-06,
+      "loss": 2.8269,
+      "step": 252800
+    },
+    {
+      "epoch": 0.06597307151324494,
+      "grad_norm": 5.977869510650635,
+      "learning_rate": 9.945798296372303e-06,
+      "loss": 2.8493,
+      "step": 253000
+    },
+    {
+      "epoch": 0.06602522413894711,
+      "grad_norm": 5.998539447784424,
+      "learning_rate": 9.945675475936487e-06,
+      "loss": 2.8413,
+      "step": 253200
+    },
+    {
+      "epoch": 0.06607737676464928,
+      "grad_norm": 6.014648914337158,
+      "learning_rate": 9.945552517263336e-06,
+      "loss": 2.7819,
+      "step": 253400
+    },
+    {
+      "epoch": 0.06612952939035145,
+      "grad_norm": 6.409533977508545,
+      "learning_rate": 9.945429420356291e-06,
+      "loss": 2.8191,
+      "step": 253600
+    },
+    {
+      "epoch": 0.06618168201605362,
+      "grad_norm": 7.013908386230469,
+      "learning_rate": 9.94530618521879e-06,
+      "loss": 2.8302,
+      "step": 253800
+    },
+    {
+      "epoch": 0.06623383464175579,
+      "grad_norm": 6.836598873138428,
+      "learning_rate": 9.945182811854282e-06,
+      "loss": 2.8531,
+      "step": 254000
+    },
+    {
+      "epoch": 0.06628598726745796,
+      "grad_norm": 5.761704921722412,
+      "learning_rate": 9.945059300266209e-06,
+      "loss": 2.8364,
+      "step": 254200
+    },
+    {
+      "epoch": 0.06633813989316013,
+      "grad_norm": 5.95949125289917,
+      "learning_rate": 9.944935650458028e-06,
+      "loss": 2.8268,
+      "step": 254400
+    },
+    {
+      "epoch": 0.0663902925188623,
+      "grad_norm": 7.2780985832214355,
+      "learning_rate": 9.944811862433194e-06,
+      "loss": 2.8539,
+      "step": 254600
+    },
+    {
+      "epoch": 0.06644244514456447,
+      "grad_norm": 6.372018814086914,
+      "learning_rate": 9.944687936195168e-06,
+      "loss": 2.8406,
+      "step": 254800
+    },
+    {
+      "epoch": 0.06649459777026664,
+      "grad_norm": 7.005834579467773,
+      "learning_rate": 9.94456387174741e-06,
+      "loss": 2.8495,
+      "step": 255000
+    },
+    {
+      "epoch": 0.06654675039596882,
+      "grad_norm": 6.090085983276367,
+      "learning_rate": 9.944439669093393e-06,
+      "loss": 2.8185,
+      "step": 255200
+    },
+    {
+      "epoch": 0.06659890302167099,
+      "grad_norm": 6.5275750160217285,
+      "learning_rate": 9.944315328236585e-06,
+      "loss": 2.8444,
+      "step": 255400
+    },
+    {
+      "epoch": 0.06665105564737316,
+      "grad_norm": 6.6855316162109375,
+      "learning_rate": 9.944190849180464e-06,
+      "loss": 2.8106,
+      "step": 255600
+    },
+    {
+      "epoch": 0.06670320827307533,
+      "grad_norm": 5.903449058532715,
+      "learning_rate": 9.944066231928506e-06,
+      "loss": 2.8174,
+      "step": 255800
+    },
+    {
+      "epoch": 0.0667553608987775,
+      "grad_norm": 5.807356357574463,
+      "learning_rate": 9.943941476484195e-06,
+      "loss": 2.8591,
+      "step": 256000
+    },
+    {
+      "epoch": 0.06680751352447967,
+      "grad_norm": 6.048740386962891,
+      "learning_rate": 9.943816582851023e-06,
+      "loss": 2.8281,
+      "step": 256200
+    },
+    {
+      "epoch": 0.06685966615018182,
+      "grad_norm": 5.910772800445557,
+      "learning_rate": 9.943691551032472e-06,
+      "loss": 2.8024,
+      "step": 256400
+    },
+    {
+      "epoch": 0.066911818775884,
+      "grad_norm": 6.064548492431641,
+      "learning_rate": 9.943566381032047e-06,
+      "loss": 2.8232,
+      "step": 256600
+    },
+    {
+      "epoch": 0.06696397140158616,
+      "grad_norm": 6.429322719573975,
+      "learning_rate": 9.943441072853239e-06,
+      "loss": 2.8604,
+      "step": 256800
+    },
+    {
+      "epoch": 0.06701612402728833,
+      "grad_norm": 6.18205451965332,
+      "learning_rate": 9.943315626499552e-06,
+      "loss": 2.8166,
+      "step": 257000
+    },
+    {
+      "epoch": 0.0670682766529905,
+      "grad_norm": 6.421455383300781,
+      "learning_rate": 9.943190041974495e-06,
+      "loss": 2.8616,
+      "step": 257200
+    },
+    {
+      "epoch": 0.06712042927869268,
+      "grad_norm": 6.15507173538208,
+      "learning_rate": 9.943064319281576e-06,
+      "loss": 2.878,
+      "step": 257400
+    },
+    {
+      "epoch": 0.06717258190439485,
+      "grad_norm": 6.6916728019714355,
+      "learning_rate": 9.94293845842431e-06,
+      "loss": 2.8272,
+      "step": 257600
+    },
+    {
+      "epoch": 0.06722473453009702,
+      "grad_norm": 6.15769100189209,
+      "learning_rate": 9.942812459406215e-06,
+      "loss": 2.8024,
+      "step": 257800
+    },
+    {
+      "epoch": 0.06727688715579919,
+      "grad_norm": 6.359096527099609,
+      "learning_rate": 9.942686322230812e-06,
+      "loss": 2.8265,
+      "step": 258000
+    },
+    {
+      "epoch": 0.06732903978150136,
+      "grad_norm": 5.924700736999512,
+      "learning_rate": 9.942560046901628e-06,
+      "loss": 2.8314,
+      "step": 258200
+    },
+    {
+      "epoch": 0.06738119240720353,
+      "grad_norm": 6.2472968101501465,
+      "learning_rate": 9.94243363342219e-06,
+      "loss": 2.8348,
+      "step": 258400
+    },
+    {
+      "epoch": 0.0674333450329057,
+      "grad_norm": 6.63815450668335,
+      "learning_rate": 9.942307081796034e-06,
+      "loss": 2.8363,
+      "step": 258600
+    },
+    {
+      "epoch": 0.06748549765860787,
+      "grad_norm": 7.617677211761475,
+      "learning_rate": 9.942180392026697e-06,
+      "loss": 2.8584,
+      "step": 258800
+    },
+    {
+      "epoch": 0.06753765028431004,
+      "grad_norm": 6.234862327575684,
+      "learning_rate": 9.942053564117718e-06,
+      "loss": 2.8885,
+      "step": 259000
+    },
+    {
+      "epoch": 0.06758980291001221,
+      "grad_norm": 6.803557395935059,
+      "learning_rate": 9.941926598072644e-06,
+      "loss": 2.8659,
+      "step": 259200
+    },
+    {
+      "epoch": 0.06764195553571438,
+      "grad_norm": 6.141203880310059,
+      "learning_rate": 9.941799493895024e-06,
+      "loss": 2.8241,
+      "step": 259400
+    },
+    {
+      "epoch": 0.06769410816141655,
+      "grad_norm": 6.708746910095215,
+      "learning_rate": 9.94167225158841e-06,
+      "loss": 2.8405,
+      "step": 259600
+    },
+    {
+      "epoch": 0.06774626078711872,
+      "grad_norm": 6.132614612579346,
+      "learning_rate": 9.94154487115636e-06,
+      "loss": 2.8347,
+      "step": 259800
+    },
+    {
+      "epoch": 0.06779841341282089,
+      "grad_norm": 6.152980327606201,
+      "learning_rate": 9.941417352602429e-06,
+      "loss": 2.795,
+      "step": 260000
+    },
+    {
+      "epoch": 0.06785056603852306,
+      "grad_norm": 6.715810775756836,
+      "learning_rate": 9.941289695930188e-06,
+      "loss": 2.8301,
+      "step": 260200
+    },
+    {
+      "epoch": 0.06790271866422523,
+      "grad_norm": 6.222926139831543,
+      "learning_rate": 9.941161901143203e-06,
+      "loss": 2.8402,
+      "step": 260400
+    },
+    {
+      "epoch": 0.0679548712899274,
+      "grad_norm": 5.573058128356934,
+      "learning_rate": 9.941033968245045e-06,
+      "loss": 2.8325,
+      "step": 260600
+    },
+    {
+      "epoch": 0.06800702391562957,
+      "grad_norm": 5.889303207397461,
+      "learning_rate": 9.940905897239289e-06,
+      "loss": 2.8836,
+      "step": 260800
+    },
+    {
+      "epoch": 0.06805917654133174,
+      "grad_norm": 6.204052448272705,
+      "learning_rate": 9.940777688129517e-06,
+      "loss": 2.8051,
+      "step": 261000
+    },
+    {
+      "epoch": 0.06811132916703391,
+      "grad_norm": 6.2043585777282715,
+      "learning_rate": 9.940649340919313e-06,
+      "loss": 2.8156,
+      "step": 261200
+    },
+    {
+      "epoch": 0.06816348179273608,
+      "grad_norm": 6.628237724304199,
+      "learning_rate": 9.940520855612262e-06,
+      "loss": 2.8152,
+      "step": 261400
+    },
+    {
+      "epoch": 0.06821563441843825,
+      "grad_norm": 5.52294397354126,
+      "learning_rate": 9.940392232211955e-06,
+      "loss": 2.8127,
+      "step": 261600
+    },
+    {
+      "epoch": 0.06826778704414042,
+      "grad_norm": 6.279348373413086,
+      "learning_rate": 9.94026347072199e-06,
+      "loss": 2.8359,
+      "step": 261800
+    },
+    {
+      "epoch": 0.0683199396698426,
+      "grad_norm": 5.843539237976074,
+      "learning_rate": 9.940134571145966e-06,
+      "loss": 2.8432,
+      "step": 262000
+    },
+    {
+      "epoch": 0.06837209229554475,
+      "grad_norm": 6.166245460510254,
+      "learning_rate": 9.940005533487483e-06,
+      "loss": 2.8402,
+      "step": 262200
+    },
+    {
+      "epoch": 0.06842424492124692,
+      "grad_norm": 5.798083305358887,
+      "learning_rate": 9.93987635775015e-06,
+      "loss": 2.8465,
+      "step": 262400
+    },
+    {
+      "epoch": 0.06847639754694909,
+      "grad_norm": 6.206599712371826,
+      "learning_rate": 9.939747043937575e-06,
+      "loss": 2.8503,
+      "step": 262600
+    },
+    {
+      "epoch": 0.06852855017265126,
+      "grad_norm": 6.811930179595947,
+      "learning_rate": 9.939617592053376e-06,
+      "loss": 2.7846,
+      "step": 262800
+    },
+    {
+      "epoch": 0.06858070279835343,
+      "grad_norm": 6.215068340301514,
+      "learning_rate": 9.93948800210117e-06,
+      "loss": 2.8258,
+      "step": 263000
+    },
+    {
+      "epoch": 0.0686328554240556,
+      "grad_norm": 6.74024772644043,
+      "learning_rate": 9.939358274084578e-06,
+      "loss": 2.8517,
+      "step": 263200
+    },
+    {
+      "epoch": 0.06868500804975777,
+      "grad_norm": 6.108590602874756,
+      "learning_rate": 9.939228408007227e-06,
+      "loss": 2.8154,
+      "step": 263400
+    },
+    {
+      "epoch": 0.06873716067545994,
+      "grad_norm": 7.009735584259033,
+      "learning_rate": 9.939098403872747e-06,
+      "loss": 2.8149,
+      "step": 263600
+    },
+    {
+      "epoch": 0.06878931330116211,
+      "grad_norm": 6.192783832550049,
+      "learning_rate": 9.938968261684771e-06,
+      "loss": 2.8519,
+      "step": 263800
+    },
+    {
+      "epoch": 0.06884146592686428,
+      "grad_norm": 6.660505771636963,
+      "learning_rate": 9.938837981446939e-06,
+      "loss": 2.8706,
+      "step": 264000
+    },
+    {
+      "epoch": 0.06889361855256645,
+      "grad_norm": 6.753753185272217,
+      "learning_rate": 9.938707563162888e-06,
+      "loss": 2.7983,
+      "step": 264200
+    },
+    {
+      "epoch": 0.06894577117826863,
+      "grad_norm": 5.999804973602295,
+      "learning_rate": 9.938577006836268e-06,
+      "loss": 2.8527,
+      "step": 264400
+    },
+    {
+      "epoch": 0.0689979238039708,
+      "grad_norm": 6.876380443572998,
+      "learning_rate": 9.938446312470726e-06,
+      "loss": 2.8108,
+      "step": 264600
+    },
+    {
+      "epoch": 0.06905007642967297,
+      "grad_norm": 6.110062122344971,
+      "learning_rate": 9.938315480069916e-06,
+      "loss": 2.8565,
+      "step": 264800
+    },
+    {
+      "epoch": 0.06910222905537514,
+      "grad_norm": 6.41756534576416,
+      "learning_rate": 9.938184509637494e-06,
+      "loss": 2.8097,
+      "step": 265000
+    },
+    {
+      "epoch": 0.0691543816810773,
+      "grad_norm": 6.181519985198975,
+      "learning_rate": 9.93805340117712e-06,
+      "loss": 2.7903,
+      "step": 265200
+    },
+    {
+      "epoch": 0.06920653430677948,
+      "grad_norm": 6.470280170440674,
+      "learning_rate": 9.93792215469246e-06,
+      "loss": 2.7889,
+      "step": 265400
+    },
+    {
+      "epoch": 0.06925868693248165,
+      "grad_norm": 6.139248371124268,
+      "learning_rate": 9.937790770187182e-06,
+      "loss": 2.8096,
+      "step": 265600
+    },
+    {
+      "epoch": 0.06931083955818382,
+      "grad_norm": 5.582498073577881,
+      "learning_rate": 9.937659247664959e-06,
+      "loss": 2.8197,
+      "step": 265800
+    },
+    {
+      "epoch": 0.06936299218388599,
+      "grad_norm": 6.460460186004639,
+      "learning_rate": 9.937527587129468e-06,
+      "loss": 2.78,
+      "step": 266000
+    },
+    {
+      "epoch": 0.06941514480958816,
+      "grad_norm": 6.965539455413818,
+      "learning_rate": 9.937395788584384e-06,
+      "loss": 2.8352,
+      "step": 266200
+    },
+    {
+      "epoch": 0.06946729743529033,
+      "grad_norm": 6.817525386810303,
+      "learning_rate": 9.937263852033399e-06,
+      "loss": 2.8182,
+      "step": 266400
+    },
+    {
+      "epoch": 0.0695194500609925,
+      "grad_norm": 5.603514671325684,
+      "learning_rate": 9.937131777480194e-06,
+      "loss": 2.8105,
+      "step": 266600
+    },
+    {
+      "epoch": 0.06957160268669467,
+      "grad_norm": 6.385352611541748,
+      "learning_rate": 9.936999564928462e-06,
+      "loss": 2.8152,
+      "step": 266800
+    },
+    {
+      "epoch": 0.06962375531239684,
+      "grad_norm": 5.736618518829346,
+      "learning_rate": 9.936867214381902e-06,
+      "loss": 2.785,
+      "step": 267000
+    },
+    {
+      "epoch": 0.06967590793809901,
+      "grad_norm": 7.078679084777832,
+      "learning_rate": 9.93673472584421e-06,
+      "loss": 2.8278,
+      "step": 267200
+    },
+    {
+      "epoch": 0.06972806056380118,
+      "grad_norm": 6.587845802307129,
+      "learning_rate": 9.93660209931909e-06,
+      "loss": 2.7866,
+      "step": 267400
+    },
+    {
+      "epoch": 0.06978021318950335,
+      "grad_norm": 6.5862717628479,
+      "learning_rate": 9.93646933481025e-06,
+      "loss": 2.7899,
+      "step": 267600
+    },
+    {
+      "epoch": 0.06983236581520552,
+      "grad_norm": 6.024425506591797,
+      "learning_rate": 9.936336432321398e-06,
+      "loss": 2.8223,
+      "step": 267800
+    },
+    {
+      "epoch": 0.06988451844090769,
+      "grad_norm": 6.144232749938965,
+      "learning_rate": 9.936203391856254e-06,
+      "loss": 2.8156,
+      "step": 268000
+    },
+    {
+      "epoch": 0.06993667106660985,
+      "grad_norm": 6.591801166534424,
+      "learning_rate": 9.936070213418532e-06,
+      "loss": 2.7847,
+      "step": 268200
+    },
+    {
+      "epoch": 0.06998882369231202,
+      "grad_norm": 6.024001121520996,
+      "learning_rate": 9.935936897011955e-06,
+      "loss": 2.808,
+      "step": 268400
+    },
+    {
+      "epoch": 0.07004097631801419,
+      "grad_norm": 6.559907913208008,
+      "learning_rate": 9.93580344264025e-06,
+      "loss": 2.8023,
+      "step": 268600
+    },
+    {
+      "epoch": 0.07009312894371636,
+      "grad_norm": 6.3227033615112305,
+      "learning_rate": 9.93566985030715e-06,
+      "loss": 2.7682,
+      "step": 268800
+    },
+    {
+      "epoch": 0.07014528156941853,
+      "grad_norm": 5.811648368835449,
+      "learning_rate": 9.935536120016386e-06,
+      "loss": 2.8045,
+      "step": 269000
+    },
+    {
+      "epoch": 0.0701974341951207,
+      "grad_norm": 6.381039619445801,
+      "learning_rate": 9.935402251771696e-06,
+      "loss": 2.822,
+      "step": 269200
+    },
+    {
+      "epoch": 0.07024958682082287,
+      "grad_norm": 6.522758960723877,
+      "learning_rate": 9.935268245576822e-06,
+      "loss": 2.8439,
+      "step": 269400
+    },
+    {
+      "epoch": 0.07030173944652504,
+      "grad_norm": 6.2131428718566895,
+      "learning_rate": 9.935134101435511e-06,
+      "loss": 2.8601,
+      "step": 269600
+    },
+    {
+      "epoch": 0.07035389207222721,
+      "grad_norm": 6.588796615600586,
+      "learning_rate": 9.93499981935151e-06,
+      "loss": 2.8367,
+      "step": 269800
+    },
+    {
+      "epoch": 0.07040604469792938,
+      "grad_norm": 6.30739688873291,
+      "learning_rate": 9.934865399328575e-06,
+      "loss": 2.8663,
+      "step": 270000
+    },
+    {
+      "epoch": 0.07045819732363155,
+      "grad_norm": 6.022554397583008,
+      "learning_rate": 9.934730841370461e-06,
+      "loss": 2.8205,
+      "step": 270200
+    },
+    {
+      "epoch": 0.07051034994933372,
+      "grad_norm": 6.7225341796875,
+      "learning_rate": 9.93459614548093e-06,
+      "loss": 2.8732,
+      "step": 270400
+    },
+    {
+      "epoch": 0.0705625025750359,
+      "grad_norm": 5.783280849456787,
+      "learning_rate": 9.934461311663748e-06,
+      "loss": 2.7827,
+      "step": 270600
+    },
+    {
+      "epoch": 0.07061465520073806,
+      "grad_norm": 6.880871772766113,
+      "learning_rate": 9.934326339922682e-06,
+      "loss": 2.7892,
+      "step": 270800
+    },
+    {
+      "epoch": 0.07066680782644023,
+      "grad_norm": 7.110273361206055,
+      "learning_rate": 9.934191230261506e-06,
+      "loss": 2.8391,
+      "step": 271000
+    },
+    {
+      "epoch": 0.0707189604521424,
+      "grad_norm": 6.837869167327881,
+      "learning_rate": 9.934055982683995e-06,
+      "loss": 2.824,
+      "step": 271200
+    },
+    {
+      "epoch": 0.07077111307784457,
+      "grad_norm": 5.97743558883667,
+      "learning_rate": 9.933920597193932e-06,
+      "loss": 2.805,
+      "step": 271400
+    },
+    {
+      "epoch": 0.07082326570354674,
+      "grad_norm": 6.262125015258789,
+      "learning_rate": 9.933785073795096e-06,
+      "loss": 2.7591,
+      "step": 271600
+    },
+    {
+      "epoch": 0.07087541832924892,
+      "grad_norm": 6.377274036407471,
+      "learning_rate": 9.93364941249128e-06,
+      "loss": 2.8213,
+      "step": 271800
+    },
+    {
+      "epoch": 0.07092757095495109,
+      "grad_norm": 6.591363906860352,
+      "learning_rate": 9.933513613286273e-06,
+      "loss": 2.8096,
+      "step": 272000
+    },
+    {
+      "epoch": 0.07097972358065326,
+      "grad_norm": 6.756320476531982,
+      "learning_rate": 9.933377676183875e-06,
+      "loss": 2.7908,
+      "step": 272200
+    },
+    {
+      "epoch": 0.07103187620635543,
+      "grad_norm": 5.893064975738525,
+      "learning_rate": 9.93324160118788e-06,
+      "loss": 2.8085,
+      "step": 272400
+    },
+    {
+      "epoch": 0.0710840288320576,
+      "grad_norm": 6.648096561431885,
+      "learning_rate": 9.933105388302094e-06,
+      "loss": 2.8344,
+      "step": 272600
+    },
+    {
+      "epoch": 0.07113618145775977,
+      "grad_norm": 5.9668049812316895,
+      "learning_rate": 9.932969037530325e-06,
+      "loss": 2.8419,
+      "step": 272800
+    },
+    {
+      "epoch": 0.07118833408346194,
+      "grad_norm": 7.359618186950684,
+      "learning_rate": 9.932832548876384e-06,
+      "loss": 2.7993,
+      "step": 273000
+    },
+    {
+      "epoch": 0.07124048670916411,
+      "grad_norm": 7.143940448760986,
+      "learning_rate": 9.932695922344085e-06,
+      "loss": 2.8044,
+      "step": 273200
+    },
+    {
+      "epoch": 0.07129263933486628,
+      "grad_norm": 7.557311534881592,
+      "learning_rate": 9.932559157937247e-06,
+      "loss": 2.7901,
+      "step": 273400
+    },
+    {
+      "epoch": 0.07134479196056845,
+      "grad_norm": 6.371750354766846,
+      "learning_rate": 9.932422255659692e-06,
+      "loss": 2.8085,
+      "step": 273600
+    },
+    {
+      "epoch": 0.07139694458627062,
+      "grad_norm": 6.6176838874816895,
+      "learning_rate": 9.93228521551525e-06,
+      "loss": 2.8022,
+      "step": 273800
+    },
+    {
+      "epoch": 0.07144909721197278,
+      "grad_norm": 6.3366851806640625,
+      "learning_rate": 9.93214803750775e-06,
+      "loss": 2.7898,
+      "step": 274000
+    },
+    {
+      "epoch": 0.07150124983767495,
+      "grad_norm": 5.762211799621582,
+      "learning_rate": 9.932010721641022e-06,
+      "loss": 2.8627,
+      "step": 274200
+    },
+    {
+      "epoch": 0.07155340246337712,
+      "grad_norm": 6.653818607330322,
+      "learning_rate": 9.931873267918908e-06,
+      "loss": 2.7655,
+      "step": 274400
+    },
+    {
+      "epoch": 0.07160555508907929,
+      "grad_norm": 6.697924613952637,
+      "learning_rate": 9.931735676345252e-06,
+      "loss": 2.7799,
+      "step": 274600
+    },
+    {
+      "epoch": 0.07165770771478146,
+      "grad_norm": 6.904690742492676,
+      "learning_rate": 9.931597946923895e-06,
+      "loss": 2.8291,
+      "step": 274800
+    },
+    {
+      "epoch": 0.07170986034048363,
+      "grad_norm": 6.095543384552002,
+      "learning_rate": 9.93146007965869e-06,
+      "loss": 2.8048,
+      "step": 275000
+    },
+    {
+      "epoch": 0.0717620129661858,
+      "grad_norm": 6.26823616027832,
+      "learning_rate": 9.931322074553488e-06,
+      "loss": 2.7887,
+      "step": 275200
+    },
+    {
+      "epoch": 0.07181416559188797,
+      "grad_norm": 5.967849254608154,
+      "learning_rate": 9.931183931612151e-06,
+      "loss": 2.7847,
+      "step": 275400
+    },
+    {
+      "epoch": 0.07186631821759014,
+      "grad_norm": 6.218311786651611,
+      "learning_rate": 9.931045650838536e-06,
+      "loss": 2.7812,
+      "step": 275600
+    },
+    {
+      "epoch": 0.07191847084329231,
+      "grad_norm": 5.455589771270752,
+      "learning_rate": 9.930907232236508e-06,
+      "loss": 2.8283,
+      "step": 275800
+    },
+    {
+      "epoch": 0.07197062346899448,
+      "grad_norm": 5.432662010192871,
+      "learning_rate": 9.930768675809939e-06,
+      "loss": 2.8221,
+      "step": 276000
+    },
+    {
+      "epoch": 0.07202277609469665,
+      "grad_norm": 6.456852436065674,
+      "learning_rate": 9.930629981562698e-06,
+      "loss": 2.8253,
+      "step": 276200
+    },
+    {
+      "epoch": 0.07207492872039882,
+      "grad_norm": 6.429563999176025,
+      "learning_rate": 9.930491149498667e-06,
+      "loss": 2.831,
+      "step": 276400
+    },
+    {
+      "epoch": 0.07212708134610099,
+      "grad_norm": 6.186648368835449,
+      "learning_rate": 9.930352179621721e-06,
+      "loss": 2.7932,
+      "step": 276600
+    },
+    {
+      "epoch": 0.07217923397180316,
+      "grad_norm": 6.750385761260986,
+      "learning_rate": 9.930213071935746e-06,
+      "loss": 2.748,
+      "step": 276800
+    },
+    {
+      "epoch": 0.07223138659750533,
+      "grad_norm": 6.330704689025879,
+      "learning_rate": 9.930073826444634e-06,
+      "loss": 2.8346,
+      "step": 277000
+    },
+    {
+      "epoch": 0.0722835392232075,
+      "grad_norm": 6.233454704284668,
+      "learning_rate": 9.929934443152272e-06,
+      "loss": 2.7588,
+      "step": 277200
+    },
+    {
+      "epoch": 0.07233569184890967,
+      "grad_norm": 6.775033473968506,
+      "learning_rate": 9.929794922062556e-06,
+      "loss": 2.8368,
+      "step": 277400
+    },
+    {
+      "epoch": 0.07238784447461184,
+      "grad_norm": 6.234418869018555,
+      "learning_rate": 9.929655263179389e-06,
+      "loss": 2.7867,
+      "step": 277600
+    },
+    {
+      "epoch": 0.07243999710031401,
+      "grad_norm": 7.00774621963501,
+      "learning_rate": 9.929515466506675e-06,
+      "loss": 2.8419,
+      "step": 277800
+    },
+    {
+      "epoch": 0.07249214972601618,
+      "grad_norm": 6.623013019561768,
+      "learning_rate": 9.929375532048318e-06,
+      "loss": 2.7692,
+      "step": 278000
+    },
+    {
+      "epoch": 0.07254430235171835,
+      "grad_norm": 5.8589582443237305,
+      "learning_rate": 9.929235459808233e-06,
+      "loss": 2.8091,
+      "step": 278200
+    },
+    {
+      "epoch": 0.07259645497742052,
+      "grad_norm": 5.785161018371582,
+      "learning_rate": 9.92909524979033e-06,
+      "loss": 2.8113,
+      "step": 278400
+    },
+    {
+      "epoch": 0.0726486076031227,
+      "grad_norm": 6.4824018478393555,
+      "learning_rate": 9.928954901998535e-06,
+      "loss": 2.7695,
+      "step": 278600
+    },
+    {
+      "epoch": 0.07270076022882486,
+      "grad_norm": 5.669132232666016,
+      "learning_rate": 9.928814416436764e-06,
+      "loss": 2.8168,
+      "step": 278800
+    },
+    {
+      "epoch": 0.07275291285452704,
+      "grad_norm": 5.815093040466309,
+      "learning_rate": 9.92867379310895e-06,
+      "loss": 2.8186,
+      "step": 279000
+    },
+    {
+      "epoch": 0.0728050654802292,
+      "grad_norm": 6.025607585906982,
+      "learning_rate": 9.928533032019018e-06,
+      "loss": 2.7955,
+      "step": 279200
+    },
+    {
+      "epoch": 0.07285721810593138,
+      "grad_norm": 6.561540126800537,
+      "learning_rate": 9.928392133170906e-06,
+      "loss": 2.8035,
+      "step": 279400
+    },
+    {
+      "epoch": 0.07290937073163355,
+      "grad_norm": 6.505258560180664,
+      "learning_rate": 9.928251096568551e-06,
+      "loss": 2.8136,
+      "step": 279600
+    },
+    {
+      "epoch": 0.07296152335733572,
+      "grad_norm": 6.027585506439209,
+      "learning_rate": 9.928109922215895e-06,
+      "loss": 2.819,
+      "step": 279800
+    },
+    {
+      "epoch": 0.07301367598303787,
+      "grad_norm": 6.295257091522217,
+      "learning_rate": 9.927968610116885e-06,
+      "loss": 2.7694,
+      "step": 280000
+    },
+    {
+      "epoch": 0.07306582860874004,
+      "grad_norm": 5.988527297973633,
+      "learning_rate": 9.92782716027547e-06,
+      "loss": 2.8237,
+      "step": 280200
+    },
+    {
+      "epoch": 0.07311798123444221,
+      "grad_norm": 6.000006198883057,
+      "learning_rate": 9.927685572695602e-06,
+      "loss": 2.7769,
+      "step": 280400
+    },
+    {
+      "epoch": 0.07317013386014438,
+      "grad_norm": 5.901209831237793,
+      "learning_rate": 9.927543847381242e-06,
+      "loss": 2.8113,
+      "step": 280600
+    },
+    {
+      "epoch": 0.07322228648584655,
+      "grad_norm": 6.257493019104004,
+      "learning_rate": 9.927401984336351e-06,
+      "loss": 2.803,
+      "step": 280800
+    },
+    {
+      "epoch": 0.07327443911154873,
+      "grad_norm": 6.857747554779053,
+      "learning_rate": 9.927259983564892e-06,
+      "loss": 2.7732,
+      "step": 281000
+    },
+    {
+      "epoch": 0.0733265917372509,
+      "grad_norm": 6.075799465179443,
+      "learning_rate": 9.927117845070834e-06,
+      "loss": 2.8093,
+      "step": 281200
+    },
+    {
+      "epoch": 0.07337874436295307,
+      "grad_norm": 6.6486663818359375,
+      "learning_rate": 9.926975568858152e-06,
+      "loss": 2.7942,
+      "step": 281400
+    },
+    {
+      "epoch": 0.07343089698865524,
+      "grad_norm": 6.643108367919922,
+      "learning_rate": 9.926833154930823e-06,
+      "loss": 2.7644,
+      "step": 281600
+    },
+    {
+      "epoch": 0.0734830496143574,
+      "grad_norm": 6.577000141143799,
+      "learning_rate": 9.926690603292825e-06,
+      "loss": 2.8179,
+      "step": 281800
+    },
+    {
+      "epoch": 0.07353520224005958,
+      "grad_norm": 6.4486083984375,
+      "learning_rate": 9.926547913948146e-06,
+      "loss": 2.8009,
+      "step": 282000
+    },
+    {
+      "epoch": 0.07358735486576175,
+      "grad_norm": 6.277480602264404,
+      "learning_rate": 9.92640508690077e-06,
+      "loss": 2.8281,
+      "step": 282200
+    },
+    {
+      "epoch": 0.07363950749146392,
+      "grad_norm": 5.925256729125977,
+      "learning_rate": 9.926262122154692e-06,
+      "loss": 2.7898,
+      "step": 282400
+    },
+    {
+      "epoch": 0.07369166011716609,
+      "grad_norm": 6.8221049308776855,
+      "learning_rate": 9.926119019713908e-06,
+      "loss": 2.7954,
+      "step": 282600
+    },
+    {
+      "epoch": 0.07374381274286826,
+      "grad_norm": 6.1247429847717285,
+      "learning_rate": 9.925975779582417e-06,
+      "loss": 2.7843,
+      "step": 282800
+    },
+    {
+      "epoch": 0.07379596536857043,
+      "grad_norm": 6.930478572845459,
+      "learning_rate": 9.925832401764222e-06,
+      "loss": 2.7843,
+      "step": 283000
+    },
+    {
+      "epoch": 0.0738481179942726,
+      "grad_norm": 6.8114190101623535,
+      "learning_rate": 9.925688886263333e-06,
+      "loss": 2.7745,
+      "step": 283200
+    },
+    {
+      "epoch": 0.07390027061997477,
+      "grad_norm": 6.889013767242432,
+      "learning_rate": 9.92554523308376e-06,
+      "loss": 2.8041,
+      "step": 283400
+    },
+    {
+      "epoch": 0.07395242324567694,
+      "grad_norm": 5.969966411590576,
+      "learning_rate": 9.925401442229518e-06,
+      "loss": 2.7588,
+      "step": 283600
+    },
+    {
+      "epoch": 0.07400457587137911,
+      "grad_norm": 6.67039155960083,
+      "learning_rate": 9.925257513704627e-06,
+      "loss": 2.7643,
+      "step": 283800
+    },
+    {
+      "epoch": 0.07405672849708128,
+      "grad_norm": 6.659237861633301,
+      "learning_rate": 9.925113447513108e-06,
+      "loss": 2.7798,
+      "step": 284000
+    },
+    {
+      "epoch": 0.07410888112278345,
+      "grad_norm": 5.949248790740967,
+      "learning_rate": 9.924969243658991e-06,
+      "loss": 2.7853,
+      "step": 284200
+    },
+    {
+      "epoch": 0.07416103374848562,
+      "grad_norm": 6.595699310302734,
+      "learning_rate": 9.924824902146304e-06,
+      "loss": 2.8293,
+      "step": 284400
+    },
+    {
+      "epoch": 0.07421318637418779,
+      "grad_norm": 6.128589630126953,
+      "learning_rate": 9.924680422979082e-06,
+      "loss": 2.8168,
+      "step": 284600
+    },
+    {
+      "epoch": 0.07426533899988996,
+      "grad_norm": 6.020960330963135,
+      "learning_rate": 9.924535806161367e-06,
+      "loss": 2.7953,
+      "step": 284800
+    },
+    {
+      "epoch": 0.07431749162559213,
+      "grad_norm": 6.3884358406066895,
+      "learning_rate": 9.924391051697194e-06,
+      "loss": 2.8002,
+      "step": 285000
+    },
+    {
+      "epoch": 0.0743696442512943,
+      "grad_norm": 6.152309417724609,
+      "learning_rate": 9.924246159590614e-06,
+      "loss": 2.7912,
+      "step": 285200
+    },
+    {
+      "epoch": 0.07442179687699647,
+      "grad_norm": 6.479878902435303,
+      "learning_rate": 9.924101129845678e-06,
+      "loss": 2.7759,
+      "step": 285400
+    },
+    {
+      "epoch": 0.07447394950269864,
+      "grad_norm": 6.872642517089844,
+      "learning_rate": 9.923955962466437e-06,
+      "loss": 2.801,
+      "step": 285600
+    },
+    {
+      "epoch": 0.0745261021284008,
+      "grad_norm": 6.354683876037598,
+      "learning_rate": 9.923810657456947e-06,
+      "loss": 2.7839,
+      "step": 285800
+    },
+    {
+      "epoch": 0.07457825475410297,
+      "grad_norm": 6.509720325469971,
+      "learning_rate": 9.923665214821274e-06,
+      "loss": 2.7437,
+      "step": 286000
+    },
+    {
+      "epoch": 0.07463040737980514,
+      "grad_norm": 7.049443244934082,
+      "learning_rate": 9.92351963456348e-06,
+      "loss": 2.8347,
+      "step": 286200
+    },
+    {
+      "epoch": 0.07468256000550731,
+      "grad_norm": 5.861264705657959,
+      "learning_rate": 9.923373916687634e-06,
+      "loss": 2.8464,
+      "step": 286400
+    },
+    {
+      "epoch": 0.07473471263120948,
+      "grad_norm": 6.284032821655273,
+      "learning_rate": 9.923228061197814e-06,
+      "loss": 2.8014,
+      "step": 286600
+    },
+    {
+      "epoch": 0.07478686525691165,
+      "grad_norm": 7.213873386383057,
+      "learning_rate": 9.92308206809809e-06,
+      "loss": 2.8036,
+      "step": 286800
+    },
+    {
+      "epoch": 0.07483901788261382,
+      "grad_norm": 5.673182010650635,
+      "learning_rate": 9.922935937392545e-06,
+      "loss": 2.7957,
+      "step": 287000
+    },
+    {
+      "epoch": 0.074891170508316,
+      "grad_norm": 6.418299198150635,
+      "learning_rate": 9.922789669085266e-06,
+      "loss": 2.7754,
+      "step": 287200
+    },
+    {
+      "epoch": 0.07494332313401816,
+      "grad_norm": 6.251279830932617,
+      "learning_rate": 9.92264326318034e-06,
+      "loss": 2.8254,
+      "step": 287400
+    },
+    {
+      "epoch": 0.07499547575972033,
+      "grad_norm": 11.405594825744629,
+      "learning_rate": 9.922496719681858e-06,
+      "loss": 2.8577,
+      "step": 287600
+    },
+    {
+      "epoch": 0.0750476283854225,
+      "grad_norm": 5.6194963455200195,
+      "learning_rate": 9.922350038593918e-06,
+      "loss": 2.7584,
+      "step": 287800
+    },
+    {
+      "epoch": 0.07509978101112467,
+      "grad_norm": 6.875600814819336,
+      "learning_rate": 9.922203219920617e-06,
+      "loss": 2.7921,
+      "step": 288000
+    },
+    {
+      "epoch": 0.07515193363682685,
+      "grad_norm": 5.844818592071533,
+      "learning_rate": 9.92205626366606e-06,
+      "loss": 2.8321,
+      "step": 288200
+    },
+    {
+      "epoch": 0.07520408626252902,
+      "grad_norm": 6.084321975708008,
+      "learning_rate": 9.921909169834357e-06,
+      "loss": 2.7855,
+      "step": 288400
+    },
+    {
+      "epoch": 0.07525623888823119,
+      "grad_norm": 6.810995101928711,
+      "learning_rate": 9.921761938429615e-06,
+      "loss": 2.7948,
+      "step": 288600
+    },
+    {
+      "epoch": 0.07530839151393336,
+      "grad_norm": 6.287647724151611,
+      "learning_rate": 9.921614569455956e-06,
+      "loss": 2.815,
+      "step": 288800
+    },
+    {
+      "epoch": 0.07536054413963553,
+      "grad_norm": 6.69083833694458,
+      "learning_rate": 9.921467062917492e-06,
+      "loss": 2.7916,
+      "step": 289000
+    },
+    {
+      "epoch": 0.0754126967653377,
+      "grad_norm": 6.045042514801025,
+      "learning_rate": 9.921319418818347e-06,
+      "loss": 2.7645,
+      "step": 289200
+    },
+    {
+      "epoch": 0.07546484939103987,
+      "grad_norm": 6.019128799438477,
+      "learning_rate": 9.921171637162654e-06,
+      "loss": 2.813,
+      "step": 289400
+    },
+    {
+      "epoch": 0.07551700201674204,
+      "grad_norm": 6.279255390167236,
+      "learning_rate": 9.921023717954537e-06,
+      "loss": 2.8184,
+      "step": 289600
+    },
+    {
+      "epoch": 0.07556915464244421,
+      "grad_norm": 5.646428108215332,
+      "learning_rate": 9.920875661198132e-06,
+      "loss": 2.7849,
+      "step": 289800
+    },
+    {
+      "epoch": 0.07562130726814638,
+      "grad_norm": 6.301240921020508,
+      "learning_rate": 9.92072746689758e-06,
+      "loss": 2.7739,
+      "step": 290000
+    },
+    {
+      "epoch": 0.07567345989384855,
+      "grad_norm": 6.263399600982666,
+      "learning_rate": 9.920579135057022e-06,
+      "loss": 2.8212,
+      "step": 290200
+    },
+    {
+      "epoch": 0.07572561251955072,
+      "grad_norm": 6.300200939178467,
+      "learning_rate": 9.920430665680602e-06,
+      "loss": 2.8088,
+      "step": 290400
+    },
+    {
+      "epoch": 0.07577776514525289,
+      "grad_norm": 6.656348705291748,
+      "learning_rate": 9.920282058772472e-06,
+      "loss": 2.8058,
+      "step": 290600
+    },
+    {
+      "epoch": 0.07582991777095506,
+      "grad_norm": 6.327916622161865,
+      "learning_rate": 9.920133314336785e-06,
+      "loss": 2.8079,
+      "step": 290800
+    },
+    {
+      "epoch": 0.07588207039665723,
+      "grad_norm": 6.084270477294922,
+      "learning_rate": 9.919984432377698e-06,
+      "loss": 2.789,
+      "step": 291000
+    },
+    {
+      "epoch": 0.0759342230223594,
+      "grad_norm": 6.9517388343811035,
+      "learning_rate": 9.919835412899375e-06,
+      "loss": 2.7942,
+      "step": 291200
+    },
+    {
+      "epoch": 0.07598637564806157,
+      "grad_norm": 6.387750148773193,
+      "learning_rate": 9.919686255905978e-06,
+      "loss": 2.7857,
+      "step": 291400
+    },
+    {
+      "epoch": 0.07603852827376374,
+      "grad_norm": 6.805675506591797,
+      "learning_rate": 9.91953696140168e-06,
+      "loss": 2.8233,
+      "step": 291600
+    },
+    {
+      "epoch": 0.0760906808994659,
+      "grad_norm": 6.1650214195251465,
+      "learning_rate": 9.919387529390648e-06,
+      "loss": 2.8102,
+      "step": 291800
+    },
+    {
+      "epoch": 0.07614283352516807,
+      "grad_norm": 6.393313407897949,
+      "learning_rate": 9.919237959877064e-06,
+      "loss": 2.8349,
+      "step": 292000
+    },
+    {
+      "epoch": 0.07619498615087024,
+      "grad_norm": 6.139355659484863,
+      "learning_rate": 9.919088252865107e-06,
+      "loss": 2.7698,
+      "step": 292200
+    },
+    {
+      "epoch": 0.07624713877657241,
+      "grad_norm": 6.505699157714844,
+      "learning_rate": 9.918938408358961e-06,
+      "loss": 2.8034,
+      "step": 292400
+    },
+    {
+      "epoch": 0.07629929140227458,
+      "grad_norm": 6.276296138763428,
+      "learning_rate": 9.918788426362816e-06,
+      "loss": 2.7709,
+      "step": 292600
+    },
+    {
+      "epoch": 0.07635144402797675,
+      "grad_norm": 5.510497570037842,
+      "learning_rate": 9.918638306880861e-06,
+      "loss": 2.7805,
+      "step": 292800
+    },
+    {
+      "epoch": 0.07640359665367892,
+      "grad_norm": 6.827136993408203,
+      "learning_rate": 9.918488049917294e-06,
+      "loss": 2.7736,
+      "step": 293000
+    },
+    {
+      "epoch": 0.07645574927938109,
+      "grad_norm": 6.214416980743408,
+      "learning_rate": 9.918337655476315e-06,
+      "loss": 2.8018,
+      "step": 293200
+    },
+    {
+      "epoch": 0.07650790190508326,
+      "grad_norm": 6.037956237792969,
+      "learning_rate": 9.91818712356213e-06,
+      "loss": 2.7831,
+      "step": 293400
+    },
+    {
+      "epoch": 0.07656005453078543,
+      "grad_norm": 7.491139888763428,
+      "learning_rate": 9.918036454178942e-06,
+      "loss": 2.8054,
+      "step": 293600
+    },
+    {
+      "epoch": 0.0766122071564876,
+      "grad_norm": 7.301893711090088,
+      "learning_rate": 9.917885647330963e-06,
+      "loss": 2.8469,
+      "step": 293800
+    },
+    {
+      "epoch": 0.07666435978218977,
+      "grad_norm": 6.116480350494385,
+      "learning_rate": 9.917734703022411e-06,
+      "loss": 2.7881,
+      "step": 294000
+    },
+    {
+      "epoch": 0.07671651240789194,
+      "grad_norm": 6.379162788391113,
+      "learning_rate": 9.917583621257504e-06,
+      "loss": 2.7695,
+      "step": 294200
+    },
+    {
+      "epoch": 0.07676866503359411,
+      "grad_norm": 7.193239212036133,
+      "learning_rate": 9.917432402040463e-06,
+      "loss": 2.7596,
+      "step": 294400
+    },
+    {
+      "epoch": 0.07682081765929628,
+      "grad_norm": 6.380529880523682,
+      "learning_rate": 9.917281045375518e-06,
+      "loss": 2.8354,
+      "step": 294600
+    },
+    {
+      "epoch": 0.07687297028499845,
+      "grad_norm": 6.858937740325928,
+      "learning_rate": 9.917129551266897e-06,
+      "loss": 2.7872,
+      "step": 294800
+    },
+    {
+      "epoch": 0.07692512291070062,
+      "grad_norm": 6.391117572784424,
+      "learning_rate": 9.916977919718837e-06,
+      "loss": 2.8193,
+      "step": 295000
+    },
+    {
+      "epoch": 0.0769772755364028,
+      "grad_norm": 6.219594478607178,
+      "learning_rate": 9.916826150735573e-06,
+      "loss": 2.8042,
+      "step": 295200
+    },
+    {
+      "epoch": 0.07702942816210497,
+      "grad_norm": 6.027685642242432,
+      "learning_rate": 9.91667424432135e-06,
+      "loss": 2.8282,
+      "step": 295400
+    },
+    {
+      "epoch": 0.07708158078780714,
+      "grad_norm": 5.730934143066406,
+      "learning_rate": 9.916522200480412e-06,
+      "loss": 2.766,
+      "step": 295600
+    },
+    {
+      "epoch": 0.0771337334135093,
+      "grad_norm": 6.417144298553467,
+      "learning_rate": 9.91637001921701e-06,
+      "loss": 2.8094,
+      "step": 295800
+    },
+    {
+      "epoch": 0.07718588603921148,
+      "grad_norm": 6.070188522338867,
+      "learning_rate": 9.916217700535395e-06,
+      "loss": 2.7917,
+      "step": 296000
+    },
+    {
+      "epoch": 0.07723803866491365,
+      "grad_norm": 5.724254131317139,
+      "learning_rate": 9.91606524443983e-06,
+      "loss": 2.8104,
+      "step": 296200
+    },
+    {
+      "epoch": 0.07729019129061582,
+      "grad_norm": 5.841590404510498,
+      "learning_rate": 9.91591265093457e-06,
+      "loss": 2.7847,
+      "step": 296400
+    },
+    {
+      "epoch": 0.07734234391631799,
+      "grad_norm": 6.103632926940918,
+      "learning_rate": 9.915759920023886e-06,
+      "loss": 2.7978,
+      "step": 296600
+    },
+    {
+      "epoch": 0.07739449654202016,
+      "grad_norm": 6.138303756713867,
+      "learning_rate": 9.915607051712042e-06,
+      "loss": 2.7617,
+      "step": 296800
+    },
+    {
+      "epoch": 0.07744664916772233,
+      "grad_norm": 6.288359642028809,
+      "learning_rate": 9.915454046003312e-06,
+      "loss": 2.7719,
+      "step": 297000
+    },
+    {
+      "epoch": 0.0774988017934245,
+      "grad_norm": 7.5223002433776855,
+      "learning_rate": 9.915300902901976e-06,
+      "loss": 2.8224,
+      "step": 297200
+    },
+    {
+      "epoch": 0.07755095441912667,
+      "grad_norm": 6.539451599121094,
+      "learning_rate": 9.915147622412311e-06,
+      "loss": 2.7655,
+      "step": 297400
+    },
+    {
+      "epoch": 0.07760310704482883,
+      "grad_norm": 6.718863487243652,
+      "learning_rate": 9.914994204538603e-06,
+      "loss": 2.8393,
+      "step": 297600
+    },
+    {
+      "epoch": 0.077655259670531,
+      "grad_norm": 6.5553669929504395,
+      "learning_rate": 9.914840649285142e-06,
+      "loss": 2.7846,
+      "step": 297800
+    },
+    {
+      "epoch": 0.07770741229623317,
+      "grad_norm": 5.977219104766846,
+      "learning_rate": 9.914686956656214e-06,
+      "loss": 2.7882,
+      "step": 298000
+    },
+    {
+      "epoch": 0.07775956492193534,
+      "grad_norm": 6.6102752685546875,
+      "learning_rate": 9.91453312665612e-06,
+      "loss": 2.7676,
+      "step": 298200
+    },
+    {
+      "epoch": 0.07781171754763751,
+      "grad_norm": 6.522745609283447,
+      "learning_rate": 9.91437915928916e-06,
+      "loss": 2.7544,
+      "step": 298400
+    },
+    {
+      "epoch": 0.07786387017333968,
+      "grad_norm": 6.287838935852051,
+      "learning_rate": 9.914225054559636e-06,
+      "loss": 2.8284,
+      "step": 298600
+    },
+    {
+      "epoch": 0.07791602279904185,
+      "grad_norm": 6.7383270263671875,
+      "learning_rate": 9.914070812471853e-06,
+      "loss": 2.7425,
+      "step": 298800
+    },
+    {
+      "epoch": 0.07796817542474402,
+      "grad_norm": 6.654935359954834,
+      "learning_rate": 9.913916433030126e-06,
+      "loss": 2.7467,
+      "step": 299000
+    },
+    {
+      "epoch": 0.07802032805044619,
+      "grad_norm": 5.797543525695801,
+      "learning_rate": 9.91376191623877e-06,
+      "loss": 2.789,
+      "step": 299200
+    },
+    {
+      "epoch": 0.07807248067614836,
+      "grad_norm": 8.258265495300293,
+      "learning_rate": 9.913607262102101e-06,
+      "loss": 2.7822,
+      "step": 299400
+    },
+    {
+      "epoch": 0.07812463330185053,
+      "grad_norm": 5.814871788024902,
+      "learning_rate": 9.913452470624443e-06,
+      "loss": 2.763,
+      "step": 299600
+    },
+    {
+      "epoch": 0.0781767859275527,
+      "grad_norm": 6.059165954589844,
+      "learning_rate": 9.913297541810123e-06,
+      "loss": 2.7726,
+      "step": 299800
+    },
+    {
+      "epoch": 0.07822893855325487,
+      "grad_norm": 6.355923175811768,
+      "learning_rate": 9.913142475663472e-06,
+      "loss": 2.8118,
+      "step": 300000
+    },
+    {
+      "epoch": 0.07828109117895704,
+      "grad_norm": 6.16987419128418,
+      "learning_rate": 9.912987272188826e-06,
+      "loss": 2.7481,
+      "step": 300200
+    },
+    {
+      "epoch": 0.07833324380465921,
+      "grad_norm": 6.023130416870117,
+      "learning_rate": 9.912831931390518e-06,
+      "loss": 2.7463,
+      "step": 300400
+    },
+    {
+      "epoch": 0.07838539643036138,
+      "grad_norm": 5.977993011474609,
+      "learning_rate": 9.912676453272894e-06,
+      "loss": 2.845,
+      "step": 300600
+    },
+    {
+      "epoch": 0.07843754905606355,
+      "grad_norm": 6.411383152008057,
+      "learning_rate": 9.912520837840297e-06,
+      "loss": 2.7646,
+      "step": 300800
+    },
+    {
+      "epoch": 0.07848970168176572,
+      "grad_norm": 6.466489791870117,
+      "learning_rate": 9.91236508509708e-06,
+      "loss": 2.7944,
+      "step": 301000
+    },
+    {
+      "epoch": 0.07854185430746789,
+      "grad_norm": 6.878299713134766,
+      "learning_rate": 9.912209195047594e-06,
+      "loss": 2.798,
+      "step": 301200
+    },
+    {
+      "epoch": 0.07859400693317006,
+      "grad_norm": 6.08614444732666,
+      "learning_rate": 9.912053167696195e-06,
+      "loss": 2.7649,
+      "step": 301400
+    },
+    {
+      "epoch": 0.07864615955887223,
+      "grad_norm": 6.1333909034729,
+      "learning_rate": 9.911897003047251e-06,
+      "loss": 2.7795,
+      "step": 301600
+    },
+    {
+      "epoch": 0.0786983121845744,
+      "grad_norm": 6.349955081939697,
+      "learning_rate": 9.91174070110512e-06,
+      "loss": 2.7147,
+      "step": 301800
+    },
+    {
+      "epoch": 0.07875046481027657,
+      "grad_norm": 6.769674301147461,
+      "learning_rate": 9.91158426187417e-06,
+      "loss": 2.8079,
+      "step": 302000
+    },
+    {
+      "epoch": 0.07880261743597874,
+      "grad_norm": 6.519297122955322,
+      "learning_rate": 9.91142768535878e-06,
+      "loss": 2.7955,
+      "step": 302200
+    },
+    {
+      "epoch": 0.07885477006168091,
+      "grad_norm": 7.310003757476807,
+      "learning_rate": 9.911270971563322e-06,
+      "loss": 2.7827,
+      "step": 302400
+    },
+    {
+      "epoch": 0.07890692268738309,
+      "grad_norm": 5.958581924438477,
+      "learning_rate": 9.911114120492177e-06,
+      "loss": 2.7466,
+      "step": 302600
+    },
+    {
+      "epoch": 0.07895907531308526,
+      "grad_norm": 5.792890548706055,
+      "learning_rate": 9.91095713214973e-06,
+      "loss": 2.7629,
+      "step": 302800
+    },
+    {
+      "epoch": 0.07901122793878743,
+      "grad_norm": 6.76243782043457,
+      "learning_rate": 9.910800006540368e-06,
+      "loss": 2.8411,
+      "step": 303000
+    },
+    {
+      "epoch": 0.0790633805644896,
+      "grad_norm": 6.586005687713623,
+      "learning_rate": 9.910642743668486e-06,
+      "loss": 2.7527,
+      "step": 303200
+    },
+    {
+      "epoch": 0.07911553319019177,
+      "grad_norm": 6.768370628356934,
+      "learning_rate": 9.910485343538474e-06,
+      "loss": 2.7507,
+      "step": 303400
+    },
+    {
+      "epoch": 0.07916768581589392,
+      "grad_norm": 6.578641891479492,
+      "learning_rate": 9.910327806154735e-06,
+      "loss": 2.8037,
+      "step": 303600
+    },
+    {
+      "epoch": 0.0792198384415961,
+      "grad_norm": 6.524317741394043,
+      "learning_rate": 9.910170131521675e-06,
+      "loss": 2.7402,
+      "step": 303800
+    },
+    {
+      "epoch": 0.07927199106729826,
+      "grad_norm": 6.295195579528809,
+      "learning_rate": 9.910012319643696e-06,
+      "loss": 2.8275,
+      "step": 304000
+    },
+    {
+      "epoch": 0.07932414369300043,
+      "grad_norm": 7.033209800720215,
+      "learning_rate": 9.909854370525212e-06,
+      "loss": 2.7591,
+      "step": 304200
+    },
+    {
+      "epoch": 0.0793762963187026,
+      "grad_norm": 6.22572660446167,
+      "learning_rate": 9.909696284170636e-06,
+      "loss": 2.7604,
+      "step": 304400
+    },
+    {
+      "epoch": 0.07942844894440478,
+      "grad_norm": 6.447390556335449,
+      "learning_rate": 9.909538060584387e-06,
+      "loss": 2.799,
+      "step": 304600
+    },
+    {
+      "epoch": 0.07948060157010695,
+      "grad_norm": 6.915390491485596,
+      "learning_rate": 9.909379699770889e-06,
+      "loss": 2.792,
+      "step": 304800
+    },
+    {
+      "epoch": 0.07953275419580912,
+      "grad_norm": 6.417814254760742,
+      "learning_rate": 9.909221201734568e-06,
+      "loss": 2.7755,
+      "step": 305000
+    },
+    {
+      "epoch": 0.07958490682151129,
+      "grad_norm": 6.955142974853516,
+      "learning_rate": 9.909062566479854e-06,
+      "loss": 2.7464,
+      "step": 305200
+    },
+    {
+      "epoch": 0.07963705944721346,
+      "grad_norm": 6.3183393478393555,
+      "learning_rate": 9.908903794011183e-06,
+      "loss": 2.7509,
+      "step": 305400
+    },
+    {
+      "epoch": 0.07968921207291563,
+      "grad_norm": 6.242965221405029,
+      "learning_rate": 9.908744884332988e-06,
+      "loss": 2.8254,
+      "step": 305600
+    },
+    {
+      "epoch": 0.0797413646986178,
+      "grad_norm": 6.4110941886901855,
+      "learning_rate": 9.908585837449714e-06,
+      "loss": 2.7494,
+      "step": 305800
+    },
+    {
+      "epoch": 0.07979351732431997,
+      "grad_norm": 6.298842430114746,
+      "learning_rate": 9.908426653365805e-06,
+      "loss": 2.7913,
+      "step": 306000
+    },
+    {
+      "epoch": 0.07984566995002214,
+      "grad_norm": 6.478790760040283,
+      "learning_rate": 9.908267332085712e-06,
+      "loss": 2.7974,
+      "step": 306200
+    },
+    {
+      "epoch": 0.07989782257572431,
+      "grad_norm": 6.1008453369140625,
+      "learning_rate": 9.908107873613888e-06,
+      "loss": 2.79,
+      "step": 306400
+    },
+    {
+      "epoch": 0.07994997520142648,
+      "grad_norm": 6.282758712768555,
+      "learning_rate": 9.90794827795479e-06,
+      "loss": 2.7768,
+      "step": 306600
+    },
+    {
+      "epoch": 0.08000212782712865,
+      "grad_norm": 6.637455463409424,
+      "learning_rate": 9.907788545112879e-06,
+      "loss": 2.7501,
+      "step": 306800
+    },
+    {
+      "epoch": 0.08005428045283082,
+      "grad_norm": 6.408087730407715,
+      "learning_rate": 9.907628675092618e-06,
+      "loss": 2.8239,
+      "step": 307000
+    },
+    {
+      "epoch": 0.08010643307853299,
+      "grad_norm": 5.750133991241455,
+      "learning_rate": 9.907468667898478e-06,
+      "loss": 2.7664,
+      "step": 307200
+    },
+    {
+      "epoch": 0.08015858570423516,
+      "grad_norm": 6.643690586090088,
+      "learning_rate": 9.90730852353493e-06,
+      "loss": 2.7875,
+      "step": 307400
+    },
+    {
+      "epoch": 0.08021073832993733,
+      "grad_norm": 6.663003921508789,
+      "learning_rate": 9.907148242006451e-06,
+      "loss": 2.758,
+      "step": 307600
+    },
+    {
+      "epoch": 0.0802628909556395,
+      "grad_norm": 7.2608113288879395,
+      "learning_rate": 9.906987823317517e-06,
+      "loss": 2.7628,
+      "step": 307800
+    },
+    {
+      "epoch": 0.08031504358134167,
+      "grad_norm": 5.9124274253845215,
+      "learning_rate": 9.906827267472619e-06,
+      "loss": 2.766,
+      "step": 308000
+    },
+    {
+      "epoch": 0.08036719620704384,
+      "grad_norm": 6.0799241065979,
+      "learning_rate": 9.90666657447624e-06,
+      "loss": 2.7949,
+      "step": 308200
+    },
+    {
+      "epoch": 0.08041934883274601,
+      "grad_norm": 6.578120231628418,
+      "learning_rate": 9.906505744332873e-06,
+      "loss": 2.7463,
+      "step": 308400
+    },
+    {
+      "epoch": 0.08047150145844818,
+      "grad_norm": 6.545755386352539,
+      "learning_rate": 9.906344777047012e-06,
+      "loss": 2.771,
+      "step": 308600
+    },
+    {
+      "epoch": 0.08052365408415035,
+      "grad_norm": 6.288052082061768,
+      "learning_rate": 9.906183672623158e-06,
+      "loss": 2.787,
+      "step": 308800
+    },
+    {
+      "epoch": 0.08057580670985252,
+      "grad_norm": 7.15657377243042,
+      "learning_rate": 9.906022431065814e-06,
+      "loss": 2.7635,
+      "step": 309000
+    },
+    {
+      "epoch": 0.0806279593355547,
+      "grad_norm": 6.247620582580566,
+      "learning_rate": 9.905861052379484e-06,
+      "loss": 2.8279,
+      "step": 309200
+    },
+    {
+      "epoch": 0.08068011196125685,
+      "grad_norm": 6.499161243438721,
+      "learning_rate": 9.905699536568682e-06,
+      "loss": 2.8063,
+      "step": 309400
+    },
+    {
+      "epoch": 0.08073226458695902,
+      "grad_norm": 5.619341850280762,
+      "learning_rate": 9.905537883637923e-06,
+      "loss": 2.776,
+      "step": 309600
+    },
+    {
+      "epoch": 0.08078441721266119,
+      "grad_norm": 6.434546947479248,
+      "learning_rate": 9.905376093591722e-06,
+      "loss": 2.8069,
+      "step": 309800
+    },
+    {
+      "epoch": 0.08083656983836336,
+      "grad_norm": 5.9819207191467285,
+      "learning_rate": 9.905214166434605e-06,
+      "loss": 2.7757,
+      "step": 310000
+    },
+    {
+      "epoch": 0.08088872246406553,
+      "grad_norm": 6.195422649383545,
+      "learning_rate": 9.905052102171093e-06,
+      "loss": 2.7516,
+      "step": 310200
+    },
+    {
+      "epoch": 0.0809408750897677,
+      "grad_norm": 6.815351486206055,
+      "learning_rate": 9.904889900805721e-06,
+      "loss": 2.7657,
+      "step": 310400
+    },
+    {
+      "epoch": 0.08099302771546987,
+      "grad_norm": 6.130044937133789,
+      "learning_rate": 9.904727562343021e-06,
+      "loss": 2.8082,
+      "step": 310600
+    },
+    {
+      "epoch": 0.08104518034117204,
+      "grad_norm": 6.268271446228027,
+      "learning_rate": 9.90456508678753e-06,
+      "loss": 2.7433,
+      "step": 310800
+    },
+    {
+      "epoch": 0.08109733296687421,
+      "grad_norm": 6.54405403137207,
+      "learning_rate": 9.904402474143789e-06,
+      "loss": 2.7833,
+      "step": 311000
+    },
+    {
+      "epoch": 0.08114948559257638,
+      "grad_norm": 6.206225872039795,
+      "learning_rate": 9.904239724416345e-06,
+      "loss": 2.7515,
+      "step": 311200
+    },
+    {
+      "epoch": 0.08120163821827855,
+      "grad_norm": 6.3456645011901855,
+      "learning_rate": 9.904076837609745e-06,
+      "loss": 2.8183,
+      "step": 311400
+    },
+    {
+      "epoch": 0.08125379084398072,
+      "grad_norm": 6.351884841918945,
+      "learning_rate": 9.903913813728543e-06,
+      "loss": 2.7917,
+      "step": 311600
+    },
+    {
+      "epoch": 0.0813059434696829,
+      "grad_norm": 6.585855960845947,
+      "learning_rate": 9.903750652777296e-06,
+      "loss": 2.7465,
+      "step": 311800
+    },
+    {
+      "epoch": 0.08135809609538507,
+      "grad_norm": 6.618515491485596,
+      "learning_rate": 9.903587354760562e-06,
+      "loss": 2.787,
+      "step": 312000
+    },
+    {
+      "epoch": 0.08141024872108724,
+      "grad_norm": 6.154926300048828,
+      "learning_rate": 9.90342391968291e-06,
+      "loss": 2.7838,
+      "step": 312200
+    },
+    {
+      "epoch": 0.0814624013467894,
+      "grad_norm": 6.921688079833984,
+      "learning_rate": 9.903260347548904e-06,
+      "loss": 2.7479,
+      "step": 312400
+    },
+    {
+      "epoch": 0.08151455397249158,
+      "grad_norm": 7.639387130737305,
+      "learning_rate": 9.903096638363119e-06,
+      "loss": 2.752,
+      "step": 312600
+    },
+    {
+      "epoch": 0.08156670659819375,
+      "grad_norm": 6.138737201690674,
+      "learning_rate": 9.902932792130127e-06,
+      "loss": 2.8019,
+      "step": 312800
+    },
+    {
+      "epoch": 0.08161885922389592,
+      "grad_norm": 6.133929252624512,
+      "learning_rate": 9.902768808854513e-06,
+      "loss": 2.7514,
+      "step": 313000
+    },
+    {
+      "epoch": 0.08167101184959809,
+      "grad_norm": 6.8112359046936035,
+      "learning_rate": 9.902604688540855e-06,
+      "loss": 2.8003,
+      "step": 313200
+    },
+    {
+      "epoch": 0.08172316447530026,
+      "grad_norm": 6.365881443023682,
+      "learning_rate": 9.902440431193744e-06,
+      "loss": 2.7725,
+      "step": 313400
+    },
+    {
+      "epoch": 0.08177531710100243,
+      "grad_norm": 6.9227447509765625,
+      "learning_rate": 9.902276036817772e-06,
+      "loss": 2.7692,
+      "step": 313600
+    },
+    {
+      "epoch": 0.0818274697267046,
+      "grad_norm": 7.009827136993408,
+      "learning_rate": 9.90211150541753e-06,
+      "loss": 2.7507,
+      "step": 313800
+    },
+    {
+      "epoch": 0.08187962235240677,
+      "grad_norm": 6.5645976066589355,
+      "learning_rate": 9.90194683699762e-06,
+      "loss": 2.7632,
+      "step": 314000
+    },
+    {
+      "epoch": 0.08193177497810894,
+      "grad_norm": 6.148155689239502,
+      "learning_rate": 9.901782031562643e-06,
+      "loss": 2.7872,
+      "step": 314200
+    },
+    {
+      "epoch": 0.08198392760381111,
+      "grad_norm": 6.268352031707764,
+      "learning_rate": 9.901617089117209e-06,
+      "loss": 2.7974,
+      "step": 314400
+    },
+    {
+      "epoch": 0.08203608022951328,
+      "grad_norm": 5.926452159881592,
+      "learning_rate": 9.901452009665922e-06,
+      "loss": 2.7486,
+      "step": 314600
+    },
+    {
+      "epoch": 0.08208823285521545,
+      "grad_norm": 6.640249729156494,
+      "learning_rate": 9.901286793213402e-06,
+      "loss": 2.7851,
+      "step": 314800
+    },
+    {
+      "epoch": 0.08214038548091762,
+      "grad_norm": 7.127112865447998,
+      "learning_rate": 9.901121439764263e-06,
+      "loss": 2.7037,
+      "step": 315000
+    },
+    {
+      "epoch": 0.08219253810661979,
+      "grad_norm": 7.389196872711182,
+      "learning_rate": 9.90095594932313e-06,
+      "loss": 2.7886,
+      "step": 315200
+    },
+    {
+      "epoch": 0.08224469073232195,
+      "grad_norm": 6.33438777923584,
+      "learning_rate": 9.900790321894627e-06,
+      "loss": 2.7621,
+      "step": 315400
+    },
+    {
+      "epoch": 0.08229684335802412,
+      "grad_norm": 6.59637451171875,
+      "learning_rate": 9.900624557483383e-06,
+      "loss": 2.7757,
+      "step": 315600
+    },
+    {
+      "epoch": 0.08234899598372629,
+      "grad_norm": 6.34528923034668,
+      "learning_rate": 9.900458656094031e-06,
+      "loss": 2.7377,
+      "step": 315800
+    },
+    {
+      "epoch": 0.08240114860942846,
+      "grad_norm": 6.219143390655518,
+      "learning_rate": 9.90029261773121e-06,
+      "loss": 2.7665,
+      "step": 316000
+    },
+    {
+      "epoch": 0.08245330123513063,
+      "grad_norm": 6.358184337615967,
+      "learning_rate": 9.900126442399562e-06,
+      "loss": 2.7857,
+      "step": 316200
+    },
+    {
+      "epoch": 0.0825054538608328,
+      "grad_norm": 6.92657470703125,
+      "learning_rate": 9.899960130103728e-06,
+      "loss": 2.7823,
+      "step": 316400
+    },
+    {
+      "epoch": 0.08255760648653497,
+      "grad_norm": 6.66539192199707,
+      "learning_rate": 9.89979368084836e-06,
+      "loss": 2.7912,
+      "step": 316600
+    },
+    {
+      "epoch": 0.08260975911223714,
+      "grad_norm": 6.198975563049316,
+      "learning_rate": 9.899627094638107e-06,
+      "loss": 2.7537,
+      "step": 316800
+    },
+    {
+      "epoch": 0.08266191173793931,
+      "grad_norm": 5.979285717010498,
+      "learning_rate": 9.899460371477629e-06,
+      "loss": 2.7273,
+      "step": 317000
+    },
+    {
+      "epoch": 0.08271406436364148,
+      "grad_norm": 6.5081071853637695,
+      "learning_rate": 9.899293511371582e-06,
+      "loss": 2.7514,
+      "step": 317200
+    },
+    {
+      "epoch": 0.08276621698934365,
+      "grad_norm": 7.048851490020752,
+      "learning_rate": 9.899126514324636e-06,
+      "loss": 2.7624,
+      "step": 317400
+    },
+    {
+      "epoch": 0.08281836961504582,
+      "grad_norm": 6.467654228210449,
+      "learning_rate": 9.89895938034145e-06,
+      "loss": 2.7895,
+      "step": 317600
+    },
+    {
+      "epoch": 0.08287052224074799,
+      "grad_norm": 6.6041646003723145,
+      "learning_rate": 9.898792109426705e-06,
+      "loss": 2.7708,
+      "step": 317800
+    },
+    {
+      "epoch": 0.08292267486645016,
+      "grad_norm": 6.265130996704102,
+      "learning_rate": 9.898624701585069e-06,
+      "loss": 2.7361,
+      "step": 318000
+    },
+    {
+      "epoch": 0.08297482749215233,
+      "grad_norm": 6.6074724197387695,
+      "learning_rate": 9.898457156821226e-06,
+      "loss": 2.7643,
+      "step": 318200
+    },
+    {
+      "epoch": 0.0830269801178545,
+      "grad_norm": 6.601407527923584,
+      "learning_rate": 9.898289475139857e-06,
+      "loss": 2.7805,
+      "step": 318400
+    },
+    {
+      "epoch": 0.08307913274355667,
+      "grad_norm": 6.4573845863342285,
+      "learning_rate": 9.89812165654565e-06,
+      "loss": 2.8072,
+      "step": 318600
+    },
+    {
+      "epoch": 0.08313128536925884,
+      "grad_norm": 6.106098175048828,
+      "learning_rate": 9.897953701043292e-06,
+      "loss": 2.7642,
+      "step": 318800
+    },
+    {
+      "epoch": 0.08318343799496102,
+      "grad_norm": 6.303261756896973,
+      "learning_rate": 9.897785608637484e-06,
+      "loss": 2.7252,
+      "step": 319000
+    },
+    {
+      "epoch": 0.08323559062066319,
+      "grad_norm": 6.720864772796631,
+      "learning_rate": 9.897617379332919e-06,
+      "loss": 2.8183,
+      "step": 319200
+    },
+    {
+      "epoch": 0.08328774324636536,
+      "grad_norm": 6.235776424407959,
+      "learning_rate": 9.897449013134301e-06,
+      "loss": 2.7476,
+      "step": 319400
+    },
+    {
+      "epoch": 0.08333989587206753,
+      "grad_norm": 7.237074851989746,
+      "learning_rate": 9.897280510046337e-06,
+      "loss": 2.7326,
+      "step": 319600
+    },
+    {
+      "epoch": 0.0833920484977697,
+      "grad_norm": 6.430893421173096,
+      "learning_rate": 9.897111870073736e-06,
+      "loss": 2.7449,
+      "step": 319800
+    },
+    {
+      "epoch": 0.08344420112347187,
+      "grad_norm": 6.649140357971191,
+      "learning_rate": 9.896943093221212e-06,
+      "loss": 2.778,
+      "step": 320000
+    },
+    {
+      "epoch": 0.08349635374917404,
+      "grad_norm": 6.112555980682373,
+      "learning_rate": 9.89677417949348e-06,
+      "loss": 2.7041,
+      "step": 320200
+    },
+    {
+      "epoch": 0.08354850637487621,
+      "grad_norm": 6.915252685546875,
+      "learning_rate": 9.896605128895265e-06,
+      "loss": 2.7855,
+      "step": 320400
+    },
+    {
+      "epoch": 0.08360065900057838,
+      "grad_norm": 6.36874532699585,
+      "learning_rate": 9.896435941431291e-06,
+      "loss": 2.7738,
+      "step": 320600
+    },
+    {
+      "epoch": 0.08365281162628055,
+      "grad_norm": 7.028889179229736,
+      "learning_rate": 9.896266617106285e-06,
+      "loss": 2.8031,
+      "step": 320800
+    },
+    {
+      "epoch": 0.08370496425198272,
+      "grad_norm": 6.048758029937744,
+      "learning_rate": 9.896097155924983e-06,
+      "loss": 2.7842,
+      "step": 321000
+    },
+    {
+      "epoch": 0.08375711687768488,
+      "grad_norm": 6.261166095733643,
+      "learning_rate": 9.89592755789212e-06,
+      "loss": 2.7553,
+      "step": 321200
+    },
+    {
+      "epoch": 0.08380926950338705,
+      "grad_norm": 6.835334300994873,
+      "learning_rate": 9.895757823012435e-06,
+      "loss": 2.7636,
+      "step": 321400
+    },
+    {
+      "epoch": 0.08386142212908922,
+      "grad_norm": 6.209692478179932,
+      "learning_rate": 9.895587951290675e-06,
+      "loss": 2.763,
+      "step": 321600
+    },
+    {
+      "epoch": 0.08391357475479139,
+      "grad_norm": 6.277283191680908,
+      "learning_rate": 9.895417942731589e-06,
+      "loss": 2.8042,
+      "step": 321800
+    },
+    {
+      "epoch": 0.08396572738049356,
+      "grad_norm": 6.796102523803711,
+      "learning_rate": 9.895247797339925e-06,
+      "loss": 2.7538,
+      "step": 322000
+    },
+    {
+      "epoch": 0.08401788000619573,
+      "grad_norm": 6.786470890045166,
+      "learning_rate": 9.895077515120439e-06,
+      "loss": 2.7547,
+      "step": 322200
+    },
+    {
+      "epoch": 0.0840700326318979,
+      "grad_norm": 6.2995429039001465,
+      "learning_rate": 9.894907096077893e-06,
+      "loss": 2.7658,
+      "step": 322400
+    },
+    {
+      "epoch": 0.08412218525760007,
+      "grad_norm": 7.4601521492004395,
+      "learning_rate": 9.894736540217052e-06,
+      "loss": 2.7518,
+      "step": 322600
+    },
+    {
+      "epoch": 0.08417433788330224,
+      "grad_norm": 7.032712459564209,
+      "learning_rate": 9.89456584754268e-06,
+      "loss": 2.7879,
+      "step": 322800
+    },
+    {
+      "epoch": 0.08422649050900441,
+      "grad_norm": 6.522708892822266,
+      "learning_rate": 9.894395018059546e-06,
+      "loss": 2.782,
+      "step": 323000
+    },
+    {
+      "epoch": 0.08427864313470658,
+      "grad_norm": 6.497341632843018,
+      "learning_rate": 9.894224051772429e-06,
+      "loss": 2.7811,
+      "step": 323200
+    },
+    {
+      "epoch": 0.08433079576040875,
+      "grad_norm": 5.980485916137695,
+      "learning_rate": 9.894052948686108e-06,
+      "loss": 2.7278,
+      "step": 323400
+    },
+    {
+      "epoch": 0.08438294838611092,
+      "grad_norm": 6.1288628578186035,
+      "learning_rate": 9.893881708805364e-06,
+      "loss": 2.7643,
+      "step": 323600
+    },
+    {
+      "epoch": 0.08443510101181309,
+      "grad_norm": 6.619753360748291,
+      "learning_rate": 9.893710332134982e-06,
+      "loss": 2.7793,
+      "step": 323800
+    },
+    {
+      "epoch": 0.08448725363751526,
+      "grad_norm": 6.239343166351318,
+      "learning_rate": 9.893538818679754e-06,
+      "loss": 2.7257,
+      "step": 324000
+    },
+    {
+      "epoch": 0.08453940626321743,
+      "grad_norm": 6.4836883544921875,
+      "learning_rate": 9.893367168444474e-06,
+      "loss": 2.7462,
+      "step": 324200
+    },
+    {
+      "epoch": 0.0845915588889196,
+      "grad_norm": 6.484433650970459,
+      "learning_rate": 9.89319538143394e-06,
+      "loss": 2.7384,
+      "step": 324400
+    },
+    {
+      "epoch": 0.08464371151462177,
+      "grad_norm": 6.930042743682861,
+      "learning_rate": 9.893023457652951e-06,
+      "loss": 2.7905,
+      "step": 324600
+    },
+    {
+      "epoch": 0.08469586414032394,
+      "grad_norm": 5.899946212768555,
+      "learning_rate": 9.892851397106316e-06,
+      "loss": 2.7162,
+      "step": 324800
+    },
+    {
+      "epoch": 0.08474801676602611,
+      "grad_norm": 5.355585098266602,
+      "learning_rate": 9.892679199798843e-06,
+      "loss": 2.7872,
+      "step": 325000
+    },
+    {
+      "epoch": 0.08480016939172828,
+      "grad_norm": 7.755842685699463,
+      "learning_rate": 9.892506865735344e-06,
+      "loss": 2.7645,
+      "step": 325200
+    },
+    {
+      "epoch": 0.08485232201743045,
+      "grad_norm": 6.277362823486328,
+      "learning_rate": 9.892334394920638e-06,
+      "loss": 2.7905,
+      "step": 325400
+    },
+    {
+      "epoch": 0.08490447464313262,
+      "grad_norm": 6.218322277069092,
+      "learning_rate": 9.892161787359544e-06,
+      "loss": 2.7879,
+      "step": 325600
+    },
+    {
+      "epoch": 0.0849566272688348,
+      "grad_norm": 6.325297832489014,
+      "learning_rate": 9.891989043056886e-06,
+      "loss": 2.7732,
+      "step": 325800
+    },
+    {
+      "epoch": 0.08500877989453696,
+      "grad_norm": 6.6276445388793945,
+      "learning_rate": 9.891816162017495e-06,
+      "loss": 2.7643,
+      "step": 326000
+    },
+    {
+      "epoch": 0.08506093252023914,
+      "grad_norm": 6.802475929260254,
+      "learning_rate": 9.891643144246202e-06,
+      "loss": 2.7389,
+      "step": 326200
+    },
+    {
+      "epoch": 0.0851130851459413,
+      "grad_norm": 6.825811862945557,
+      "learning_rate": 9.89146998974784e-06,
+      "loss": 2.7736,
+      "step": 326400
+    },
+    {
+      "epoch": 0.08516523777164348,
+      "grad_norm": 6.184617519378662,
+      "learning_rate": 9.891296698527255e-06,
+      "loss": 2.7717,
+      "step": 326600
+    },
+    {
+      "epoch": 0.08521739039734565,
+      "grad_norm": 7.5419158935546875,
+      "learning_rate": 9.891123270589285e-06,
+      "loss": 2.7393,
+      "step": 326800
+    },
+    {
+      "epoch": 0.0852695430230478,
+      "grad_norm": 7.02357292175293,
+      "learning_rate": 9.89094970593878e-06,
+      "loss": 2.7573,
+      "step": 327000
+    },
+    {
+      "epoch": 0.08532169564874997,
+      "grad_norm": 6.308924674987793,
+      "learning_rate": 9.890776004580595e-06,
+      "loss": 2.7554,
+      "step": 327200
+    },
+    {
+      "epoch": 0.08537384827445214,
+      "grad_norm": 6.975244522094727,
+      "learning_rate": 9.890602166519578e-06,
+      "loss": 2.7893,
+      "step": 327400
+    },
+    {
+      "epoch": 0.08542600090015431,
+      "grad_norm": 6.773929119110107,
+      "learning_rate": 9.890428191760593e-06,
+      "loss": 2.7138,
+      "step": 327600
+    },
+    {
+      "epoch": 0.08547815352585648,
+      "grad_norm": 5.785812854766846,
+      "learning_rate": 9.890254080308498e-06,
+      "loss": 2.7643,
+      "step": 327800
+    },
+    {
+      "epoch": 0.08553030615155865,
+      "grad_norm": 6.477972507476807,
+      "learning_rate": 9.890079832168167e-06,
+      "loss": 2.7599,
+      "step": 328000
+    },
+    {
+      "epoch": 0.08558245877726083,
+      "grad_norm": 7.792716026306152,
+      "learning_rate": 9.889905447344463e-06,
+      "loss": 2.7773,
+      "step": 328200
+    },
+    {
+      "epoch": 0.085634611402963,
+      "grad_norm": 5.729294300079346,
+      "learning_rate": 9.889730925842266e-06,
+      "loss": 2.7661,
+      "step": 328400
+    },
+    {
+      "epoch": 0.08568676402866517,
+      "grad_norm": 5.776681900024414,
+      "learning_rate": 9.889556267666449e-06,
+      "loss": 2.714,
+      "step": 328600
+    },
+    {
+      "epoch": 0.08573891665436734,
+      "grad_norm": 5.912201404571533,
+      "learning_rate": 9.8893814728219e-06,
+      "loss": 2.781,
+      "step": 328800
+    },
+    {
+      "epoch": 0.0857910692800695,
+      "grad_norm": 6.655936241149902,
+      "learning_rate": 9.889206541313498e-06,
+      "loss": 2.752,
+      "step": 329000
+    },
+    {
+      "epoch": 0.08584322190577168,
+      "grad_norm": 6.418868064880371,
+      "learning_rate": 9.889031473146136e-06,
+      "loss": 2.8002,
+      "step": 329200
+    },
+    {
+      "epoch": 0.08589537453147385,
+      "grad_norm": 6.111396312713623,
+      "learning_rate": 9.888856268324707e-06,
+      "loss": 2.7168,
+      "step": 329400
+    },
+    {
+      "epoch": 0.08594752715717602,
+      "grad_norm": 6.363344669342041,
+      "learning_rate": 9.88868092685411e-06,
+      "loss": 2.7737,
+      "step": 329600
+    },
+    {
+      "epoch": 0.08599967978287819,
+      "grad_norm": 6.287228107452393,
+      "learning_rate": 9.888505448739243e-06,
+      "loss": 2.7443,
+      "step": 329800
+    },
+    {
+      "epoch": 0.08605183240858036,
+      "grad_norm": 7.1249542236328125,
+      "learning_rate": 9.888329833985012e-06,
+      "loss": 2.7436,
+      "step": 330000
+    },
+    {
+      "epoch": 0.08610398503428253,
+      "grad_norm": 6.433495998382568,
+      "learning_rate": 9.888154082596326e-06,
+      "loss": 2.7575,
+      "step": 330200
+    },
+    {
+      "epoch": 0.0861561376599847,
+      "grad_norm": 5.9914937019348145,
+      "learning_rate": 9.887978194578097e-06,
+      "loss": 2.7368,
+      "step": 330400
+    },
+    {
+      "epoch": 0.08620829028568687,
+      "grad_norm": 6.580966472625732,
+      "learning_rate": 9.88780216993524e-06,
+      "loss": 2.7948,
+      "step": 330600
+    },
+    {
+      "epoch": 0.08626044291138904,
+      "grad_norm": 6.364790439605713,
+      "learning_rate": 9.88762600867268e-06,
+      "loss": 2.7685,
+      "step": 330800
+    },
+    {
+      "epoch": 0.08631259553709121,
+      "grad_norm": 6.688601016998291,
+      "learning_rate": 9.887449710795333e-06,
+      "loss": 2.7504,
+      "step": 331000
+    },
+    {
+      "epoch": 0.08636474816279338,
+      "grad_norm": 6.046875953674316,
+      "learning_rate": 9.887273276308135e-06,
+      "loss": 2.7278,
+      "step": 331200
+    },
+    {
+      "epoch": 0.08641690078849555,
+      "grad_norm": 6.681288719177246,
+      "learning_rate": 9.887096705216011e-06,
+      "loss": 2.7646,
+      "step": 331400
+    },
+    {
+      "epoch": 0.08646905341419772,
+      "grad_norm": 6.149299621582031,
+      "learning_rate": 9.8869199975239e-06,
+      "loss": 2.727,
+      "step": 331600
+    },
+    {
+      "epoch": 0.08652120603989989,
+      "grad_norm": 6.459449291229248,
+      "learning_rate": 9.886743153236741e-06,
+      "loss": 2.7715,
+      "step": 331800
+    },
+    {
+      "epoch": 0.08657335866560206,
+      "grad_norm": 7.189176082611084,
+      "learning_rate": 9.886566172359475e-06,
+      "loss": 2.7691,
+      "step": 332000
+    },
+    {
+      "epoch": 0.08662551129130423,
+      "grad_norm": 6.436927318572998,
+      "learning_rate": 9.886389054897051e-06,
+      "loss": 2.7188,
+      "step": 332200
+    },
+    {
+      "epoch": 0.0866776639170064,
+      "grad_norm": 6.325084686279297,
+      "learning_rate": 9.886211800854419e-06,
+      "loss": 2.7552,
+      "step": 332400
+    },
+    {
+      "epoch": 0.08672981654270857,
+      "grad_norm": 5.9876251220703125,
+      "learning_rate": 9.886034410236533e-06,
+      "loss": 2.7334,
+      "step": 332600
+    },
+    {
+      "epoch": 0.08678196916841074,
+      "grad_norm": 6.643549919128418,
+      "learning_rate": 9.88585688304835e-06,
+      "loss": 2.7637,
+      "step": 332800
+    },
+    {
+      "epoch": 0.0868341217941129,
+      "grad_norm": 6.4676971435546875,
+      "learning_rate": 9.885679219294835e-06,
+      "loss": 2.7169,
+      "step": 333000
+    },
+    {
+      "epoch": 0.08688627441981507,
+      "grad_norm": 6.548349380493164,
+      "learning_rate": 9.885501418980953e-06,
+      "loss": 2.7825,
+      "step": 333200
+    },
+    {
+      "epoch": 0.08693842704551724,
+      "grad_norm": 5.312566757202148,
+      "learning_rate": 9.885323482111671e-06,
+      "loss": 2.7209,
+      "step": 333400
+    },
+    {
+      "epoch": 0.08699057967121941,
+      "grad_norm": 6.16527795791626,
+      "learning_rate": 9.885145408691965e-06,
+      "loss": 2.713,
+      "step": 333600
+    },
+    {
+      "epoch": 0.08704273229692158,
+      "grad_norm": 6.8437652587890625,
+      "learning_rate": 9.884967198726814e-06,
+      "loss": 2.7806,
+      "step": 333800
+    },
+    {
+      "epoch": 0.08709488492262375,
+      "grad_norm": 6.265858173370361,
+      "learning_rate": 9.884788852221195e-06,
+      "loss": 2.789,
+      "step": 334000
+    },
+    {
+      "epoch": 0.08714703754832592,
+      "grad_norm": 6.236710548400879,
+      "learning_rate": 9.884610369180097e-06,
+      "loss": 2.7499,
+      "step": 334200
+    },
+    {
+      "epoch": 0.0871991901740281,
+      "grad_norm": 6.732831001281738,
+      "learning_rate": 9.884431749608505e-06,
+      "loss": 2.7575,
+      "step": 334400
+    },
+    {
+      "epoch": 0.08725134279973026,
+      "grad_norm": 6.163575649261475,
+      "learning_rate": 9.884252993511415e-06,
+      "loss": 2.7529,
+      "step": 334600
+    },
+    {
+      "epoch": 0.08730349542543243,
+      "grad_norm": 6.205536365509033,
+      "learning_rate": 9.884074100893821e-06,
+      "loss": 2.7299,
+      "step": 334800
+    },
+    {
+      "epoch": 0.0873556480511346,
+      "grad_norm": 6.635568618774414,
+      "learning_rate": 9.883895071760726e-06,
+      "loss": 2.7082,
+      "step": 335000
+    },
+    {
+      "epoch": 0.08740780067683677,
+      "grad_norm": 7.192404270172119,
+      "learning_rate": 9.883715906117132e-06,
+      "loss": 2.7596,
+      "step": 335200
+    },
+    {
+      "epoch": 0.08745995330253895,
+      "grad_norm": 6.824044704437256,
+      "learning_rate": 9.883536603968047e-06,
+      "loss": 2.7477,
+      "step": 335400
+    },
+    {
+      "epoch": 0.08751210592824112,
+      "grad_norm": 7.502521991729736,
+      "learning_rate": 9.883357165318483e-06,
+      "loss": 2.7955,
+      "step": 335600
+    },
+    {
+      "epoch": 0.08756425855394329,
+      "grad_norm": 6.514331817626953,
+      "learning_rate": 9.883177590173454e-06,
+      "loss": 2.7597,
+      "step": 335800
+    },
+    {
+      "epoch": 0.08761641117964546,
+      "grad_norm": 5.833806037902832,
+      "learning_rate": 9.882997878537984e-06,
+      "loss": 2.7714,
+      "step": 336000
+    },
+    {
+      "epoch": 0.08766856380534763,
+      "grad_norm": 6.365836143493652,
+      "learning_rate": 9.88281803041709e-06,
+      "loss": 2.7577,
+      "step": 336200
+    },
+    {
+      "epoch": 0.0877207164310498,
+      "grad_norm": 6.60760498046875,
+      "learning_rate": 9.882638045815804e-06,
+      "loss": 2.7664,
+      "step": 336400
+    },
+    {
+      "epoch": 0.08777286905675197,
+      "grad_norm": 6.648801326751709,
+      "learning_rate": 9.882457924739153e-06,
+      "loss": 2.7739,
+      "step": 336600
+    },
+    {
+      "epoch": 0.08782502168245414,
+      "grad_norm": 6.417782306671143,
+      "learning_rate": 9.882277667192175e-06,
+      "loss": 2.777,
+      "step": 336800
+    },
+    {
+      "epoch": 0.08787717430815631,
+      "grad_norm": 7.36497163772583,
+      "learning_rate": 9.882097273179904e-06,
+      "loss": 2.7481,
+      "step": 337000
+    },
+    {
+      "epoch": 0.08792932693385848,
+      "grad_norm": 6.485381603240967,
+      "learning_rate": 9.881916742707389e-06,
+      "loss": 2.7754,
+      "step": 337200
+    },
+    {
+      "epoch": 0.08798147955956065,
+      "grad_norm": 6.6873369216918945,
+      "learning_rate": 9.88173607577967e-06,
+      "loss": 2.7517,
+      "step": 337400
+    },
+    {
+      "epoch": 0.08803363218526282,
+      "grad_norm": 7.222953796386719,
+      "learning_rate": 9.881555272401797e-06,
+      "loss": 2.7745,
+      "step": 337600
+    },
+    {
+      "epoch": 0.08808578481096499,
+      "grad_norm": 6.487031936645508,
+      "learning_rate": 9.881374332578829e-06,
+      "loss": 2.762,
+      "step": 337800
+    },
+    {
+      "epoch": 0.08813793743666716,
+      "grad_norm": 6.065067768096924,
+      "learning_rate": 9.881193256315816e-06,
+      "loss": 2.7632,
+      "step": 338000
+    },
+    {
+      "epoch": 0.08819009006236933,
+      "grad_norm": 6.604552745819092,
+      "learning_rate": 9.881012043617826e-06,
+      "loss": 2.7412,
+      "step": 338200
+    },
+    {
+      "epoch": 0.0882422426880715,
+      "grad_norm": 6.458118915557861,
+      "learning_rate": 9.88083069448992e-06,
+      "loss": 2.7364,
+      "step": 338400
+    },
+    {
+      "epoch": 0.08829439531377367,
+      "grad_norm": 7.036708354949951,
+      "learning_rate": 9.88064920893717e-06,
+      "loss": 2.779,
+      "step": 338600
+    },
+    {
+      "epoch": 0.08834654793947583,
+      "grad_norm": 6.828887462615967,
+      "learning_rate": 9.880467586964646e-06,
+      "loss": 2.7671,
+      "step": 338800
+    },
+    {
+      "epoch": 0.088398700565178,
+      "grad_norm": 7.044036388397217,
+      "learning_rate": 9.880285828577426e-06,
+      "loss": 2.7701,
+      "step": 339000
+    },
+    {
+      "epoch": 0.08845085319088017,
+      "grad_norm": 6.224187850952148,
+      "learning_rate": 9.880103933780589e-06,
+      "loss": 2.7674,
+      "step": 339200
+    },
+    {
+      "epoch": 0.08850300581658234,
+      "grad_norm": 5.749971389770508,
+      "learning_rate": 9.879921902579219e-06,
+      "loss": 2.7177,
+      "step": 339400
+    },
+    {
+      "epoch": 0.08855515844228451,
+      "grad_norm": 6.637483596801758,
+      "learning_rate": 9.879739734978408e-06,
+      "loss": 2.7148,
+      "step": 339600
+    },
+    {
+      "epoch": 0.08860731106798668,
+      "grad_norm": 6.52347469329834,
+      "learning_rate": 9.879557430983242e-06,
+      "loss": 2.7768,
+      "step": 339800
+    },
+    {
+      "epoch": 0.08865946369368885,
+      "grad_norm": 7.416553497314453,
+      "learning_rate": 9.879374990598821e-06,
+      "loss": 2.7519,
+      "step": 340000
+    },
+    {
+      "epoch": 0.08871161631939102,
+      "grad_norm": 7.152632713317871,
+      "learning_rate": 9.879192413830244e-06,
+      "loss": 2.761,
+      "step": 340200
+    },
+    {
+      "epoch": 0.08876376894509319,
+      "grad_norm": 6.972984313964844,
+      "learning_rate": 9.879009700682611e-06,
+      "loss": 2.7335,
+      "step": 340400
+    },
+    {
+      "epoch": 0.08881592157079536,
+      "grad_norm": 6.66843318939209,
+      "learning_rate": 9.878826851161032e-06,
+      "loss": 2.7011,
+      "step": 340600
+    },
+    {
+      "epoch": 0.08886807419649753,
+      "grad_norm": 6.625106334686279,
+      "learning_rate": 9.878643865270617e-06,
+      "loss": 2.7638,
+      "step": 340800
+    },
+    {
+      "epoch": 0.0889202268221997,
+      "grad_norm": 5.736109733581543,
+      "learning_rate": 9.87846074301648e-06,
+      "loss": 2.6837,
+      "step": 341000
+    },
+    {
+      "epoch": 0.08897237944790187,
+      "grad_norm": 6.541588306427002,
+      "learning_rate": 9.878277484403742e-06,
+      "loss": 2.785,
+      "step": 341200
+    },
+    {
+      "epoch": 0.08902453207360404,
+      "grad_norm": 6.787119388580322,
+      "learning_rate": 9.878094089437523e-06,
+      "loss": 2.768,
+      "step": 341400
+    },
+    {
+      "epoch": 0.08907668469930621,
+      "grad_norm": 6.477860927581787,
+      "learning_rate": 9.877910558122948e-06,
+      "loss": 2.7734,
+      "step": 341600
+    },
+    {
+      "epoch": 0.08912883732500838,
+      "grad_norm": 6.601665496826172,
+      "learning_rate": 9.87772689046515e-06,
+      "loss": 2.7677,
+      "step": 341800
+    },
+    {
+      "epoch": 0.08918098995071055,
+      "grad_norm": 6.276432514190674,
+      "learning_rate": 9.87754308646926e-06,
+      "loss": 2.7615,
+      "step": 342000
+    },
+    {
+      "epoch": 0.08923314257641272,
+      "grad_norm": 6.968433380126953,
+      "learning_rate": 9.877359146140416e-06,
+      "loss": 2.7536,
+      "step": 342200
+    },
+    {
+      "epoch": 0.0892852952021149,
+      "grad_norm": 6.095193862915039,
+      "learning_rate": 9.877175069483762e-06,
+      "loss": 2.727,
+      "step": 342400
+    },
+    {
+      "epoch": 0.08933744782781707,
+      "grad_norm": 6.783011436462402,
+      "learning_rate": 9.87699085650444e-06,
+      "loss": 2.7556,
+      "step": 342600
+    },
+    {
+      "epoch": 0.08938960045351924,
+      "grad_norm": 6.720550537109375,
+      "learning_rate": 9.876806507207601e-06,
+      "loss": 2.7567,
+      "step": 342800
+    },
+    {
+      "epoch": 0.0894417530792214,
+      "grad_norm": 7.076850891113281,
+      "learning_rate": 9.876622021598396e-06,
+      "loss": 2.735,
+      "step": 343000
+    },
+    {
+      "epoch": 0.08949390570492358,
+      "grad_norm": 6.559168815612793,
+      "learning_rate": 9.876437399681983e-06,
+      "loss": 2.7888,
+      "step": 343200
+    },
+    {
+      "epoch": 0.08954605833062575,
+      "grad_norm": 6.786516189575195,
+      "learning_rate": 9.876252641463522e-06,
+      "loss": 2.7834,
+      "step": 343400
+    },
+    {
+      "epoch": 0.08959821095632792,
+      "grad_norm": 7.153230667114258,
+      "learning_rate": 9.876067746948176e-06,
+      "loss": 2.724,
+      "step": 343600
+    },
+    {
+      "epoch": 0.08965036358203009,
+      "grad_norm": 6.643190860748291,
+      "learning_rate": 9.875882716141116e-06,
+      "loss": 2.7588,
+      "step": 343800
+    },
+    {
+      "epoch": 0.08970251620773226,
+      "grad_norm": 5.990670680999756,
+      "learning_rate": 9.875697549047511e-06,
+      "loss": 2.7544,
+      "step": 344000
+    },
+    {
+      "epoch": 0.08975466883343443,
+      "grad_norm": 6.413429260253906,
+      "learning_rate": 9.875512245672538e-06,
+      "loss": 2.755,
+      "step": 344200
+    },
+    {
+      "epoch": 0.0898068214591366,
+      "grad_norm": 6.6495361328125,
+      "learning_rate": 9.875326806021377e-06,
+      "loss": 2.7291,
+      "step": 344400
+    },
+    {
+      "epoch": 0.08985897408483877,
+      "grad_norm": 6.507997512817383,
+      "learning_rate": 9.875141230099209e-06,
+      "loss": 2.7421,
+      "step": 344600
+    },
+    {
+      "epoch": 0.08991112671054093,
+      "grad_norm": 6.5377726554870605,
+      "learning_rate": 9.874955517911223e-06,
+      "loss": 2.7329,
+      "step": 344800
+    },
+    {
+      "epoch": 0.0899632793362431,
+      "grad_norm": 7.305025577545166,
+      "learning_rate": 9.874769669462608e-06,
+      "loss": 2.7645,
+      "step": 345000
+    },
+    {
+      "epoch": 0.09001543196194527,
+      "grad_norm": 6.375731945037842,
+      "learning_rate": 9.87458368475856e-06,
+      "loss": 2.7613,
+      "step": 345200
+    },
+    {
+      "epoch": 0.09006758458764744,
+      "grad_norm": 6.055731773376465,
+      "learning_rate": 9.87439756380428e-06,
+      "loss": 2.7406,
+      "step": 345400
+    },
+    {
+      "epoch": 0.09011973721334961,
+      "grad_norm": 6.731203556060791,
+      "learning_rate": 9.874211306604966e-06,
+      "loss": 2.7252,
+      "step": 345600
+    },
+    {
+      "epoch": 0.09017188983905178,
+      "grad_norm": 7.642002582550049,
+      "learning_rate": 9.874024913165825e-06,
+      "loss": 2.7897,
+      "step": 345800
+    },
+    {
+      "epoch": 0.09022404246475395,
+      "grad_norm": 6.946300983428955,
+      "learning_rate": 9.873838383492069e-06,
+      "loss": 2.7631,
+      "step": 346000
+    },
+    {
+      "epoch": 0.09027619509045612,
+      "grad_norm": 6.039514064788818,
+      "learning_rate": 9.873651717588909e-06,
+      "loss": 2.7506,
+      "step": 346200
+    },
+    {
+      "epoch": 0.09032834771615829,
+      "grad_norm": 6.728977203369141,
+      "learning_rate": 9.873464915461566e-06,
+      "loss": 2.747,
+      "step": 346400
+    },
+    {
+      "epoch": 0.09038050034186046,
+      "grad_norm": 5.914807319641113,
+      "learning_rate": 9.873277977115256e-06,
+      "loss": 2.7535,
+      "step": 346600
+    },
+    {
+      "epoch": 0.09043265296756263,
+      "grad_norm": 7.127989768981934,
+      "learning_rate": 9.87309090255521e-06,
+      "loss": 2.7418,
+      "step": 346800
+    },
+    {
+      "epoch": 0.0904848055932648,
+      "grad_norm": 6.722864151000977,
+      "learning_rate": 9.872903691786655e-06,
+      "loss": 2.753,
+      "step": 347000
+    },
+    {
+      "epoch": 0.09053695821896697,
+      "grad_norm": 7.177712440490723,
+      "learning_rate": 9.87271634481482e-06,
+      "loss": 2.7418,
+      "step": 347200
+    },
+    {
+      "epoch": 0.09058911084466914,
+      "grad_norm": 6.4624552726745605,
+      "learning_rate": 9.872528861644947e-06,
+      "loss": 2.79,
+      "step": 347400
+    },
+    {
+      "epoch": 0.09064126347037131,
+      "grad_norm": 6.420626163482666,
+      "learning_rate": 9.872341242282274e-06,
+      "loss": 2.7962,
+      "step": 347600
+    },
+    {
+      "epoch": 0.09069341609607348,
+      "grad_norm": 5.890540599822998,
+      "learning_rate": 9.872153486732045e-06,
+      "loss": 2.7281,
+      "step": 347800
+    },
+    {
+      "epoch": 0.09074556872177565,
+      "grad_norm": 6.1183013916015625,
+      "learning_rate": 9.871965594999509e-06,
+      "loss": 2.7133,
+      "step": 348000
+    },
+    {
+      "epoch": 0.09079772134747782,
+      "grad_norm": 6.170734405517578,
+      "learning_rate": 9.871777567089916e-06,
+      "loss": 2.7438,
+      "step": 348200
+    },
+    {
+      "epoch": 0.09084987397317999,
+      "grad_norm": 6.432772636413574,
+      "learning_rate": 9.871589403008524e-06,
+      "loss": 2.7667,
+      "step": 348400
+    },
+    {
+      "epoch": 0.09090202659888216,
+      "grad_norm": 6.09767484664917,
+      "learning_rate": 9.87140110276059e-06,
+      "loss": 2.7265,
+      "step": 348600
+    },
+    {
+      "epoch": 0.09095417922458433,
+      "grad_norm": 6.650676727294922,
+      "learning_rate": 9.871212666351378e-06,
+      "loss": 2.7454,
+      "step": 348800
+    },
+    {
+      "epoch": 0.0910063318502865,
+      "grad_norm": 6.796266078948975,
+      "learning_rate": 9.871024093786154e-06,
+      "loss": 2.7593,
+      "step": 349000
+    },
+    {
+      "epoch": 0.09105848447598867,
+      "grad_norm": 6.901579856872559,
+      "learning_rate": 9.870835385070191e-06,
+      "loss": 2.7642,
+      "step": 349200
+    },
+    {
+      "epoch": 0.09111063710169084,
+      "grad_norm": 7.005277633666992,
+      "learning_rate": 9.870646540208763e-06,
+      "loss": 2.6938,
+      "step": 349400
+    },
+    {
+      "epoch": 0.09116278972739301,
+      "grad_norm": 7.363898277282715,
+      "learning_rate": 9.87045755920715e-06,
+      "loss": 2.7215,
+      "step": 349600
+    },
+    {
+      "epoch": 0.09121494235309519,
+      "grad_norm": 6.584136009216309,
+      "learning_rate": 9.870268442070629e-06,
+      "loss": 2.7236,
+      "step": 349800
+    },
+    {
+      "epoch": 0.09126709497879736,
+      "grad_norm": 6.563084602355957,
+      "learning_rate": 9.870079188804492e-06,
+      "loss": 2.7463,
+      "step": 350000
+    },
+    {
+      "epoch": 0.09131924760449953,
+      "grad_norm": 6.536829471588135,
+      "learning_rate": 9.869889799414026e-06,
+      "loss": 2.814,
+      "step": 350200
+    },
+    {
+      "epoch": 0.0913714002302017,
+      "grad_norm": 6.699076175689697,
+      "learning_rate": 9.869700273904524e-06,
+      "loss": 2.7248,
+      "step": 350400
+    },
+    {
+      "epoch": 0.09142355285590385,
+      "grad_norm": 6.6044111251831055,
+      "learning_rate": 9.869510612281284e-06,
+      "loss": 2.7594,
+      "step": 350600
+    },
+    {
+      "epoch": 0.09147570548160602,
+      "grad_norm": 6.523010730743408,
+      "learning_rate": 9.869320814549608e-06,
+      "loss": 2.7255,
+      "step": 350800
+    },
+    {
+      "epoch": 0.0915278581073082,
+      "grad_norm": 6.2166218757629395,
+      "learning_rate": 9.869130880714801e-06,
+      "loss": 2.7329,
+      "step": 351000
+    },
+    {
+      "epoch": 0.09158001073301036,
+      "grad_norm": 7.409599304199219,
+      "learning_rate": 9.868940810782172e-06,
+      "loss": 2.7557,
+      "step": 351200
+    },
+    {
+      "epoch": 0.09163216335871253,
+      "grad_norm": 6.580782890319824,
+      "learning_rate": 9.868750604757034e-06,
+      "loss": 2.7538,
+      "step": 351400
+    },
+    {
+      "epoch": 0.0916843159844147,
+      "grad_norm": 6.399933815002441,
+      "learning_rate": 9.868560262644701e-06,
+      "loss": 2.7467,
+      "step": 351600
+    },
+    {
+      "epoch": 0.09173646861011688,
+      "grad_norm": 6.934614181518555,
+      "learning_rate": 9.868369784450496e-06,
+      "loss": 2.7431,
+      "step": 351800
+    },
+    {
+      "epoch": 0.09178862123581905,
+      "grad_norm": 6.793469429016113,
+      "learning_rate": 9.868179170179742e-06,
+      "loss": 2.7429,
+      "step": 352000
+    },
+    {
+      "epoch": 0.09184077386152122,
+      "grad_norm": 6.359063148498535,
+      "learning_rate": 9.867988419837765e-06,
+      "loss": 2.7431,
+      "step": 352200
+    },
+    {
+      "epoch": 0.09189292648722339,
+      "grad_norm": 6.626612663269043,
+      "learning_rate": 9.8677975334299e-06,
+      "loss": 2.7322,
+      "step": 352400
+    },
+    {
+      "epoch": 0.09194507911292556,
+      "grad_norm": 7.432502746582031,
+      "learning_rate": 9.867606510961482e-06,
+      "loss": 2.7438,
+      "step": 352600
+    },
+    {
+      "epoch": 0.09199723173862773,
+      "grad_norm": 6.605191707611084,
+      "learning_rate": 9.867415352437849e-06,
+      "loss": 2.7487,
+      "step": 352800
+    },
+    {
+      "epoch": 0.0920493843643299,
+      "grad_norm": 7.211688995361328,
+      "learning_rate": 9.867224057864344e-06,
+      "loss": 2.721,
+      "step": 353000
+    },
+    {
+      "epoch": 0.09210153699003207,
+      "grad_norm": 6.4896135330200195,
+      "learning_rate": 9.867032627246315e-06,
+      "loss": 2.7345,
+      "step": 353200
+    },
+    {
+      "epoch": 0.09215368961573424,
+      "grad_norm": 7.31119966506958,
+      "learning_rate": 9.866841060589113e-06,
+      "loss": 2.7576,
+      "step": 353400
+    },
+    {
+      "epoch": 0.09220584224143641,
+      "grad_norm": 7.012179851531982,
+      "learning_rate": 9.866649357898089e-06,
+      "loss": 2.72,
+      "step": 353600
+    },
+    {
+      "epoch": 0.09225799486713858,
+      "grad_norm": 6.721981048583984,
+      "learning_rate": 9.866457519178605e-06,
+      "loss": 2.7204,
+      "step": 353800
+    },
+    {
+      "epoch": 0.09231014749284075,
+      "grad_norm": 6.36168098449707,
+      "learning_rate": 9.866265544436024e-06,
+      "loss": 2.7532,
+      "step": 354000
+    },
+    {
+      "epoch": 0.09236230011854292,
+      "grad_norm": 6.767820835113525,
+      "learning_rate": 9.866073433675709e-06,
+      "loss": 2.7872,
+      "step": 354200
+    },
+    {
+      "epoch": 0.09241445274424509,
+      "grad_norm": 6.610678672790527,
+      "learning_rate": 9.86588118690303e-06,
+      "loss": 2.687,
+      "step": 354400
+    },
+    {
+      "epoch": 0.09246660536994726,
+      "grad_norm": 6.138505458831787,
+      "learning_rate": 9.865688804123361e-06,
+      "loss": 2.7339,
+      "step": 354600
+    },
+    {
+      "epoch": 0.09251875799564943,
+      "grad_norm": 6.75731897354126,
+      "learning_rate": 9.865496285342079e-06,
+      "loss": 2.7374,
+      "step": 354800
+    },
+    {
+      "epoch": 0.0925709106213516,
+      "grad_norm": 6.869755268096924,
+      "learning_rate": 9.865303630564569e-06,
+      "loss": 2.7248,
+      "step": 355000
+    },
+    {
+      "epoch": 0.09262306324705377,
+      "grad_norm": 6.32354211807251,
+      "learning_rate": 9.86511083979621e-06,
+      "loss": 2.7101,
+      "step": 355200
+    },
+    {
+      "epoch": 0.09267521587275594,
+      "grad_norm": 6.488105297088623,
+      "learning_rate": 9.864917913042393e-06,
+      "loss": 2.7202,
+      "step": 355400
+    },
+    {
+      "epoch": 0.09272736849845811,
+      "grad_norm": 6.858880043029785,
+      "learning_rate": 9.864724850308513e-06,
+      "loss": 2.7165,
+      "step": 355600
+    },
+    {
+      "epoch": 0.09277952112416028,
+      "grad_norm": 7.065299034118652,
+      "learning_rate": 9.864531651599963e-06,
+      "loss": 2.759,
+      "step": 355800
+    },
+    {
+      "epoch": 0.09283167374986245,
+      "grad_norm": 6.129859924316406,
+      "learning_rate": 9.864338316922145e-06,
+      "loss": 2.7342,
+      "step": 356000
+    },
+    {
+      "epoch": 0.09288382637556462,
+      "grad_norm": 6.4933648109436035,
+      "learning_rate": 9.864144846280461e-06,
+      "loss": 2.7194,
+      "step": 356200
+    },
+    {
+      "epoch": 0.0929359790012668,
+      "grad_norm": 6.755699634552002,
+      "learning_rate": 9.86395123968032e-06,
+      "loss": 2.7097,
+      "step": 356400
+    },
+    {
+      "epoch": 0.09298813162696895,
+      "grad_norm": 6.979990005493164,
+      "learning_rate": 9.863757497127134e-06,
+      "loss": 2.6846,
+      "step": 356600
+    },
+    {
+      "epoch": 0.09304028425267112,
+      "grad_norm": 7.256637096405029,
+      "learning_rate": 9.863563618626317e-06,
+      "loss": 2.7368,
+      "step": 356800
+    },
+    {
+      "epoch": 0.09309243687837329,
+      "grad_norm": 6.8452253341674805,
+      "learning_rate": 9.86336960418329e-06,
+      "loss": 2.75,
+      "step": 357000
+    },
+    {
+      "epoch": 0.09314458950407546,
+      "grad_norm": 5.944540977478027,
+      "learning_rate": 9.863175453803476e-06,
+      "loss": 2.7395,
+      "step": 357200
+    },
+    {
+      "epoch": 0.09319674212977763,
+      "grad_norm": 7.049804210662842,
+      "learning_rate": 9.862981167492298e-06,
+      "loss": 2.7871,
+      "step": 357400
+    },
+    {
+      "epoch": 0.0932488947554798,
+      "grad_norm": 6.192124843597412,
+      "learning_rate": 9.862786745255191e-06,
+      "loss": 2.7268,
+      "step": 357600
+    },
+    {
+      "epoch": 0.09330104738118197,
+      "grad_norm": 6.69340181350708,
+      "learning_rate": 9.862592187097587e-06,
+      "loss": 2.7712,
+      "step": 357800
+    },
+    {
+      "epoch": 0.09335320000688414,
+      "grad_norm": 6.953091144561768,
+      "learning_rate": 9.862397493024925e-06,
+      "loss": 2.7733,
+      "step": 358000
+    },
+    {
+      "epoch": 0.09340535263258631,
+      "grad_norm": 7.205249309539795,
+      "learning_rate": 9.862202663042647e-06,
+      "loss": 2.691,
+      "step": 358200
+    },
+    {
+      "epoch": 0.09345750525828848,
+      "grad_norm": 6.435111999511719,
+      "learning_rate": 9.862007697156195e-06,
+      "loss": 2.7516,
+      "step": 358400
+    },
+    {
+      "epoch": 0.09350965788399065,
+      "grad_norm": 6.6230669021606445,
+      "learning_rate": 9.861812595371026e-06,
+      "loss": 2.7293,
+      "step": 358600
+    },
+    {
+      "epoch": 0.09356181050969282,
+      "grad_norm": 6.619906425476074,
+      "learning_rate": 9.861617357692585e-06,
+      "loss": 2.7295,
+      "step": 358800
+    },
+    {
+      "epoch": 0.093613963135395,
+      "grad_norm": 6.5545268058776855,
+      "learning_rate": 9.861421984126335e-06,
+      "loss": 2.72,
+      "step": 359000
+    },
+    {
+      "epoch": 0.09366611576109717,
+      "grad_norm": 7.116846561431885,
+      "learning_rate": 9.861226474677737e-06,
+      "loss": 2.7458,
+      "step": 359200
+    },
+    {
+      "epoch": 0.09371826838679934,
+      "grad_norm": 7.412746429443359,
+      "learning_rate": 9.861030829352252e-06,
+      "loss": 2.7375,
+      "step": 359400
+    },
+    {
+      "epoch": 0.0937704210125015,
+      "grad_norm": 7.558204174041748,
+      "learning_rate": 9.86083504815535e-06,
+      "loss": 2.7256,
+      "step": 359600
+    },
+    {
+      "epoch": 0.09382257363820368,
+      "grad_norm": 6.926628589630127,
+      "learning_rate": 9.860639131092504e-06,
+      "loss": 2.7327,
+      "step": 359800
+    },
+    {
+      "epoch": 0.09387472626390585,
+      "grad_norm": 6.14035177230835,
+      "learning_rate": 9.86044307816919e-06,
+      "loss": 2.7614,
+      "step": 360000
+    },
+    {
+      "epoch": 0.09392687888960802,
+      "grad_norm": 7.323028087615967,
+      "learning_rate": 9.860246889390888e-06,
+      "loss": 2.7235,
+      "step": 360200
+    },
+    {
+      "epoch": 0.09397903151531019,
+      "grad_norm": 6.862823009490967,
+      "learning_rate": 9.86005056476308e-06,
+      "loss": 2.75,
+      "step": 360400
+    },
+    {
+      "epoch": 0.09403118414101236,
+      "grad_norm": 6.618449687957764,
+      "learning_rate": 9.859854104291256e-06,
+      "loss": 2.7222,
+      "step": 360600
+    },
+    {
+      "epoch": 0.09408333676671453,
+      "grad_norm": 5.8654608726501465,
+      "learning_rate": 9.859657507980907e-06,
+      "loss": 2.7302,
+      "step": 360800
+    },
+    {
+      "epoch": 0.0941354893924167,
+      "grad_norm": 6.324068546295166,
+      "learning_rate": 9.859460775837526e-06,
+      "loss": 2.7301,
+      "step": 361000
+    },
+    {
+      "epoch": 0.09418764201811887,
+      "grad_norm": 6.8306732177734375,
+      "learning_rate": 9.859263907866612e-06,
+      "loss": 2.7369,
+      "step": 361200
+    },
+    {
+      "epoch": 0.09423979464382104,
+      "grad_norm": 6.69620943069458,
+      "learning_rate": 9.859066904073671e-06,
+      "loss": 2.7301,
+      "step": 361400
+    },
+    {
+      "epoch": 0.09429194726952321,
+      "grad_norm": 6.917806148529053,
+      "learning_rate": 9.858869764464208e-06,
+      "loss": 2.7438,
+      "step": 361600
+    },
+    {
+      "epoch": 0.09434409989522538,
+      "grad_norm": 7.454828262329102,
+      "learning_rate": 9.85867248904373e-06,
+      "loss": 2.729,
+      "step": 361800
+    },
+    {
+      "epoch": 0.09439625252092755,
+      "grad_norm": 6.281285762786865,
+      "learning_rate": 9.858475077817756e-06,
+      "loss": 2.7598,
+      "step": 362000
+    },
+    {
+      "epoch": 0.09444840514662972,
+      "grad_norm": 6.297204971313477,
+      "learning_rate": 9.8582775307918e-06,
+      "loss": 2.741,
+      "step": 362200
+    },
+    {
+      "epoch": 0.09450055777233188,
+      "grad_norm": 6.6822590827941895,
+      "learning_rate": 9.858079847971384e-06,
+      "loss": 2.715,
+      "step": 362400
+    },
+    {
+      "epoch": 0.09455271039803405,
+      "grad_norm": 6.762353897094727,
+      "learning_rate": 9.857882029362036e-06,
+      "loss": 2.727,
+      "step": 362600
+    },
+    {
+      "epoch": 0.09460486302373622,
+      "grad_norm": 7.140170574188232,
+      "learning_rate": 9.857684074969285e-06,
+      "loss": 2.714,
+      "step": 362800
+    },
+    {
+      "epoch": 0.09465701564943839,
+      "grad_norm": 6.909539222717285,
+      "learning_rate": 9.85748598479866e-06,
+      "loss": 2.7019,
+      "step": 363000
+    },
+    {
+      "epoch": 0.09470916827514056,
+      "grad_norm": 7.183372497558594,
+      "learning_rate": 9.857287758855705e-06,
+      "loss": 2.7792,
+      "step": 363200
+    },
+    {
+      "epoch": 0.09476132090084273,
+      "grad_norm": 6.247964859008789,
+      "learning_rate": 9.857089397145954e-06,
+      "loss": 2.7433,
+      "step": 363400
+    },
+    {
+      "epoch": 0.0948134735265449,
+      "grad_norm": 7.346426010131836,
+      "learning_rate": 9.856890899674954e-06,
+      "loss": 2.6835,
+      "step": 363600
+    },
+    {
+      "epoch": 0.09486562615224707,
+      "grad_norm": 7.062637805938721,
+      "learning_rate": 9.856692266448254e-06,
+      "loss": 2.7761,
+      "step": 363800
+    },
+    {
+      "epoch": 0.09491777877794924,
+      "grad_norm": 7.02108907699585,
+      "learning_rate": 9.856493497471405e-06,
+      "loss": 2.7348,
+      "step": 364000
+    },
+    {
+      "epoch": 0.09496993140365141,
+      "grad_norm": 6.802475929260254,
+      "learning_rate": 9.856294592749963e-06,
+      "loss": 2.7111,
+      "step": 364200
+    },
+    {
+      "epoch": 0.09502208402935358,
+      "grad_norm": 6.6680402755737305,
+      "learning_rate": 9.856095552289487e-06,
+      "loss": 2.7501,
+      "step": 364400
+    },
+    {
+      "epoch": 0.09507423665505575,
+      "grad_norm": 6.280881404876709,
+      "learning_rate": 9.85589637609554e-06,
+      "loss": 2.7415,
+      "step": 364600
+    },
+    {
+      "epoch": 0.09512638928075792,
+      "grad_norm": 6.65505313873291,
+      "learning_rate": 9.85569706417369e-06,
+      "loss": 2.7116,
+      "step": 364800
+    },
+    {
+      "epoch": 0.09517854190646009,
+      "grad_norm": 6.576626777648926,
+      "learning_rate": 9.855497616529511e-06,
+      "loss": 2.7489,
+      "step": 365000
+    },
+    {
+      "epoch": 0.09523069453216226,
+      "grad_norm": 7.567196369171143,
+      "learning_rate": 9.855298033168575e-06,
+      "loss": 2.7427,
+      "step": 365200
+    },
+    {
+      "epoch": 0.09528284715786443,
+      "grad_norm": 6.262023448944092,
+      "learning_rate": 9.855098314096459e-06,
+      "loss": 2.7202,
+      "step": 365400
+    },
+    {
+      "epoch": 0.0953349997835666,
+      "grad_norm": 6.940713405609131,
+      "learning_rate": 9.854898459318748e-06,
+      "loss": 2.7366,
+      "step": 365600
+    },
+    {
+      "epoch": 0.09538715240926877,
+      "grad_norm": 7.290978908538818,
+      "learning_rate": 9.854698468841024e-06,
+      "loss": 2.732,
+      "step": 365800
+    },
+    {
+      "epoch": 0.09543930503497094,
+      "grad_norm": 6.973400592803955,
+      "learning_rate": 9.854498342668883e-06,
+      "loss": 2.7675,
+      "step": 366000
+    },
+    {
+      "epoch": 0.09549145766067312,
+      "grad_norm": 6.667946815490723,
+      "learning_rate": 9.854298080807917e-06,
+      "loss": 2.7709,
+      "step": 366200
+    },
+    {
+      "epoch": 0.09554361028637529,
+      "grad_norm": 6.699540615081787,
+      "learning_rate": 9.854097683263719e-06,
+      "loss": 2.7282,
+      "step": 366400
+    },
+    {
+      "epoch": 0.09559576291207746,
+      "grad_norm": 6.533851146697998,
+      "learning_rate": 9.853897150041896e-06,
+      "loss": 2.6941,
+      "step": 366600
+    },
+    {
+      "epoch": 0.09564791553777963,
+      "grad_norm": 6.288170337677002,
+      "learning_rate": 9.85369648114805e-06,
+      "loss": 2.7054,
+      "step": 366800
+    },
+    {
+      "epoch": 0.0957000681634818,
+      "grad_norm": 6.440084457397461,
+      "learning_rate": 9.853495676587791e-06,
+      "loss": 2.7668,
+      "step": 367000
+    },
+    {
+      "epoch": 0.09575222078918397,
+      "grad_norm": 6.943708896636963,
+      "learning_rate": 9.853294736366732e-06,
+      "loss": 2.758,
+      "step": 367200
+    },
+    {
+      "epoch": 0.09580437341488614,
+      "grad_norm": 7.354572296142578,
+      "learning_rate": 9.85309366049049e-06,
+      "loss": 2.7425,
+      "step": 367400
+    },
+    {
+      "epoch": 0.09585652604058831,
+      "grad_norm": 6.695876121520996,
+      "learning_rate": 9.852892448964682e-06,
+      "loss": 2.7463,
+      "step": 367600
+    },
+    {
+      "epoch": 0.09590867866629048,
+      "grad_norm": 6.828666687011719,
+      "learning_rate": 9.852691101794937e-06,
+      "loss": 2.74,
+      "step": 367800
+    },
+    {
+      "epoch": 0.09596083129199265,
+      "grad_norm": 6.269225120544434,
+      "learning_rate": 9.852489618986878e-06,
+      "loss": 2.718,
+      "step": 368000
+    },
+    {
+      "epoch": 0.09601298391769482,
+      "grad_norm": 6.452040672302246,
+      "learning_rate": 9.85228800054614e-06,
+      "loss": 2.7166,
+      "step": 368200
+    },
+    {
+      "epoch": 0.09606513654339698,
+      "grad_norm": 6.910484790802002,
+      "learning_rate": 9.852086246478358e-06,
+      "loss": 2.744,
+      "step": 368400
+    },
+    {
+      "epoch": 0.09611728916909915,
+      "grad_norm": 8.016570091247559,
+      "learning_rate": 9.85188435678917e-06,
+      "loss": 2.7105,
+      "step": 368600
+    },
+    {
+      "epoch": 0.09616944179480132,
+      "grad_norm": 6.72928524017334,
+      "learning_rate": 9.85168233148422e-06,
+      "loss": 2.7281,
+      "step": 368800
+    },
+    {
+      "epoch": 0.09622159442050349,
+      "grad_norm": 7.141217231750488,
+      "learning_rate": 9.851480170569155e-06,
+      "loss": 2.7017,
+      "step": 369000
+    },
+    {
+      "epoch": 0.09627374704620566,
+      "grad_norm": 7.320204734802246,
+      "learning_rate": 9.851277874049624e-06,
+      "loss": 2.7614,
+      "step": 369200
+    },
+    {
+      "epoch": 0.09632589967190783,
+      "grad_norm": 6.801264762878418,
+      "learning_rate": 9.851075441931285e-06,
+      "loss": 2.7137,
+      "step": 369400
+    },
+    {
+      "epoch": 0.09637805229761,
+      "grad_norm": 6.665674686431885,
+      "learning_rate": 9.850872874219792e-06,
+      "loss": 2.7348,
+      "step": 369600
+    },
+    {
+      "epoch": 0.09643020492331217,
+      "grad_norm": 7.685469627380371,
+      "learning_rate": 9.85067017092081e-06,
+      "loss": 2.7101,
+      "step": 369800
+    },
+    {
+      "epoch": 0.09648235754901434,
+      "grad_norm": 7.03206729888916,
+      "learning_rate": 9.850467332040003e-06,
+      "loss": 2.7283,
+      "step": 370000
+    },
+    {
+      "epoch": 0.09653451017471651,
+      "grad_norm": 6.763091087341309,
+      "learning_rate": 9.850264357583042e-06,
+      "loss": 2.721,
+      "step": 370200
+    },
+    {
+      "epoch": 0.09658666280041868,
+      "grad_norm": 7.308859348297119,
+      "learning_rate": 9.850061247555598e-06,
+      "loss": 2.7411,
+      "step": 370400
+    },
+    {
+      "epoch": 0.09663881542612085,
+      "grad_norm": 6.72197961807251,
+      "learning_rate": 9.84985800196335e-06,
+      "loss": 2.7493,
+      "step": 370600
+    },
+    {
+      "epoch": 0.09669096805182302,
+      "grad_norm": 7.104571342468262,
+      "learning_rate": 9.849654620811981e-06,
+      "loss": 2.7556,
+      "step": 370800
+    },
+    {
+      "epoch": 0.09674312067752519,
+      "grad_norm": 6.912766456604004,
+      "learning_rate": 9.849451104107172e-06,
+      "loss": 2.7397,
+      "step": 371000
+    },
+    {
+      "epoch": 0.09679527330322736,
+      "grad_norm": 7.259765148162842,
+      "learning_rate": 9.849247451854614e-06,
+      "loss": 2.6929,
+      "step": 371200
+    },
+    {
+      "epoch": 0.09684742592892953,
+      "grad_norm": 7.524007320404053,
+      "learning_rate": 9.849043664059996e-06,
+      "loss": 2.7068,
+      "step": 371400
+    },
+    {
+      "epoch": 0.0968995785546317,
+      "grad_norm": 6.4871392250061035,
+      "learning_rate": 9.848839740729018e-06,
+      "loss": 2.752,
+      "step": 371600
+    },
+    {
+      "epoch": 0.09695173118033387,
+      "grad_norm": 6.942018985748291,
+      "learning_rate": 9.848635681867377e-06,
+      "loss": 2.7455,
+      "step": 371800
+    },
+    {
+      "epoch": 0.09700388380603604,
+      "grad_norm": 6.313770294189453,
+      "learning_rate": 9.84843148748078e-06,
+      "loss": 2.7225,
+      "step": 372000
+    },
+    {
+      "epoch": 0.09705603643173821,
+      "grad_norm": 7.180691242218018,
+      "learning_rate": 9.848227157574932e-06,
+      "loss": 2.7707,
+      "step": 372200
+    },
+    {
+      "epoch": 0.09710818905744038,
+      "grad_norm": 6.999627113342285,
+      "learning_rate": 9.848022692155544e-06,
+      "loss": 2.7113,
+      "step": 372400
+    },
+    {
+      "epoch": 0.09716034168314255,
+      "grad_norm": 6.8867950439453125,
+      "learning_rate": 9.847818091228332e-06,
+      "loss": 2.7619,
+      "step": 372600
+    },
+    {
+      "epoch": 0.09721249430884472,
+      "grad_norm": 6.820121765136719,
+      "learning_rate": 9.847613354799014e-06,
+      "loss": 2.7334,
+      "step": 372800
+    },
+    {
+      "epoch": 0.0972646469345469,
+      "grad_norm": 6.528743267059326,
+      "learning_rate": 9.847408482873316e-06,
+      "loss": 2.7372,
+      "step": 373000
+    },
+    {
+      "epoch": 0.09731679956024906,
+      "grad_norm": 7.362827301025391,
+      "learning_rate": 9.847203475456959e-06,
+      "loss": 2.7489,
+      "step": 373200
+    },
+    {
+      "epoch": 0.09736895218595124,
+      "grad_norm": 7.119494438171387,
+      "learning_rate": 9.846998332555676e-06,
+      "loss": 2.7094,
+      "step": 373400
+    },
+    {
+      "epoch": 0.0974211048116534,
+      "grad_norm": 7.4878387451171875,
+      "learning_rate": 9.8467930541752e-06,
+      "loss": 2.7318,
+      "step": 373600
+    },
+    {
+      "epoch": 0.09747325743735558,
+      "grad_norm": 7.716200828552246,
+      "learning_rate": 9.846587640321273e-06,
+      "loss": 2.8075,
+      "step": 373800
+    },
+    {
+      "epoch": 0.09752541006305775,
+      "grad_norm": 6.292218208312988,
+      "learning_rate": 9.84638209099963e-06,
+      "loss": 2.7333,
+      "step": 374000
+    },
+    {
+      "epoch": 0.0975775626887599,
+      "grad_norm": 6.929902076721191,
+      "learning_rate": 9.846176406216019e-06,
+      "loss": 2.7248,
+      "step": 374200
+    },
+    {
+      "epoch": 0.09762971531446207,
+      "grad_norm": 7.24159049987793,
+      "learning_rate": 9.845970585976192e-06,
+      "loss": 2.749,
+      "step": 374400
+    },
+    {
+      "epoch": 0.09768186794016424,
+      "grad_norm": 7.5973663330078125,
+      "learning_rate": 9.8457646302859e-06,
+      "loss": 2.7228,
+      "step": 374600
+    },
+    {
+      "epoch": 0.09773402056586641,
+      "grad_norm": 7.047082424163818,
+      "learning_rate": 9.845558539150895e-06,
+      "loss": 2.7224,
+      "step": 374800
+    },
+    {
+      "epoch": 0.09778617319156858,
+      "grad_norm": 6.793696880340576,
+      "learning_rate": 9.845352312576946e-06,
+      "loss": 2.7347,
+      "step": 375000
+    },
+    {
+      "epoch": 0.09783832581727075,
+      "grad_norm": 7.282424449920654,
+      "learning_rate": 9.84514595056981e-06,
+      "loss": 2.7181,
+      "step": 375200
+    },
+    {
+      "epoch": 0.09789047844297293,
+      "grad_norm": 7.171825885772705,
+      "learning_rate": 9.84493945313526e-06,
+      "loss": 2.7174,
+      "step": 375400
+    },
+    {
+      "epoch": 0.0979426310686751,
+      "grad_norm": 7.00319766998291,
+      "learning_rate": 9.844732820279064e-06,
+      "loss": 2.7221,
+      "step": 375600
+    },
+    {
+      "epoch": 0.09799478369437727,
+      "grad_norm": 7.138320446014404,
+      "learning_rate": 9.844526052007001e-06,
+      "loss": 2.7306,
+      "step": 375800
+    },
+    {
+      "epoch": 0.09804693632007944,
+      "grad_norm": 6.744442939758301,
+      "learning_rate": 9.84431914832485e-06,
+      "loss": 2.6985,
+      "step": 376000
+    },
+    {
+      "epoch": 0.0980990889457816,
+      "grad_norm": 7.503383636474609,
+      "learning_rate": 9.84411210923839e-06,
+      "loss": 2.7451,
+      "step": 376200
+    },
+    {
+      "epoch": 0.09815124157148378,
+      "grad_norm": 6.678633689880371,
+      "learning_rate": 9.843904934753414e-06,
+      "loss": 2.7316,
+      "step": 376400
+    },
+    {
+      "epoch": 0.09820339419718595,
+      "grad_norm": 6.815631866455078,
+      "learning_rate": 9.843697624875708e-06,
+      "loss": 2.7003,
+      "step": 376600
+    },
+    {
+      "epoch": 0.09825554682288812,
+      "grad_norm": 6.820425987243652,
+      "learning_rate": 9.843490179611069e-06,
+      "loss": 2.7462,
+      "step": 376800
+    },
+    {
+      "epoch": 0.09830769944859029,
+      "grad_norm": 7.071511268615723,
+      "learning_rate": 9.843282598965293e-06,
+      "loss": 2.7266,
+      "step": 377000
+    },
+    {
+      "epoch": 0.09835985207429246,
+      "grad_norm": 6.579829692840576,
+      "learning_rate": 9.843074882944188e-06,
+      "loss": 2.7225,
+      "step": 377200
+    },
+    {
+      "epoch": 0.09841200469999463,
+      "grad_norm": 7.302144527435303,
+      "learning_rate": 9.842867031553551e-06,
+      "loss": 2.7498,
+      "step": 377400
+    },
+    {
+      "epoch": 0.0984641573256968,
+      "grad_norm": 7.175357341766357,
+      "learning_rate": 9.8426590447992e-06,
+      "loss": 2.7374,
+      "step": 377600
+    },
+    {
+      "epoch": 0.09851630995139897,
+      "grad_norm": 7.382828712463379,
+      "learning_rate": 9.842450922686944e-06,
+      "loss": 2.6983,
+      "step": 377800
+    },
+    {
+      "epoch": 0.09856846257710114,
+      "grad_norm": 6.3378190994262695,
+      "learning_rate": 9.8422426652226e-06,
+      "loss": 2.7448,
+      "step": 378000
+    },
+    {
+      "epoch": 0.09862061520280331,
+      "grad_norm": 5.941269397735596,
+      "learning_rate": 9.84203427241199e-06,
+      "loss": 2.7119,
+      "step": 378200
+    },
+    {
+      "epoch": 0.09867276782850548,
+      "grad_norm": 7.9518842697143555,
+      "learning_rate": 9.84182574426094e-06,
+      "loss": 2.7104,
+      "step": 378400
+    },
+    {
+      "epoch": 0.09872492045420765,
+      "grad_norm": 7.776278018951416,
+      "learning_rate": 9.841617080775278e-06,
+      "loss": 2.7247,
+      "step": 378600
+    },
+    {
+      "epoch": 0.09877707307990982,
+      "grad_norm": 7.218269348144531,
+      "learning_rate": 9.841408281960836e-06,
+      "loss": 2.7109,
+      "step": 378800
+    },
+    {
+      "epoch": 0.09882922570561199,
+      "grad_norm": 6.931689739227295,
+      "learning_rate": 9.841199347823448e-06,
+      "loss": 2.7071,
+      "step": 379000
+    },
+    {
+      "epoch": 0.09888137833131416,
+      "grad_norm": 6.874160289764404,
+      "learning_rate": 9.840990278368957e-06,
+      "loss": 2.7048,
+      "step": 379200
+    },
+    {
+      "epoch": 0.09893353095701633,
+      "grad_norm": 7.4862823486328125,
+      "learning_rate": 9.840781073603208e-06,
+      "loss": 2.7754,
+      "step": 379400
+    },
+    {
+      "epoch": 0.0989856835827185,
+      "grad_norm": 6.68509578704834,
+      "learning_rate": 9.840571733532044e-06,
+      "loss": 2.7133,
+      "step": 379600
+    },
+    {
+      "epoch": 0.09903783620842067,
+      "grad_norm": 7.173377990722656,
+      "learning_rate": 9.840362258161322e-06,
+      "loss": 2.7296,
+      "step": 379800
+    },
+    {
+      "epoch": 0.09908998883412284,
+      "grad_norm": 6.899176120758057,
+      "learning_rate": 9.84015264749689e-06,
+      "loss": 2.7376,
+      "step": 380000
+    },
+    {
+      "epoch": 0.099142141459825,
+      "grad_norm": 7.660719394683838,
+      "learning_rate": 9.839942901544612e-06,
+      "loss": 2.6995,
+      "step": 380200
+    },
+    {
+      "epoch": 0.09919429408552717,
+      "grad_norm": 6.3884687423706055,
+      "learning_rate": 9.83973302031035e-06,
+      "loss": 2.7238,
+      "step": 380400
+    },
+    {
+      "epoch": 0.09924644671122934,
+      "grad_norm": 7.147850513458252,
+      "learning_rate": 9.839523003799969e-06,
+      "loss": 2.7093,
+      "step": 380600
+    },
+    {
+      "epoch": 0.09929859933693151,
+      "grad_norm": 7.06928014755249,
+      "learning_rate": 9.839312852019337e-06,
+      "loss": 2.6814,
+      "step": 380800
+    },
+    {
+      "epoch": 0.09935075196263368,
+      "grad_norm": 6.980913162231445,
+      "learning_rate": 9.839102564974336e-06,
+      "loss": 2.6875,
+      "step": 381000
+    },
+    {
+      "epoch": 0.09940290458833585,
+      "grad_norm": 7.352754592895508,
+      "learning_rate": 9.838892142670834e-06,
+      "loss": 2.6892,
+      "step": 381200
+    },
+    {
+      "epoch": 0.09945505721403802,
+      "grad_norm": 7.018017768859863,
+      "learning_rate": 9.838681585114721e-06,
+      "loss": 2.7277,
+      "step": 381400
+    },
+    {
+      "epoch": 0.0995072098397402,
+      "grad_norm": 7.365713119506836,
+      "learning_rate": 9.838470892311876e-06,
+      "loss": 2.7038,
+      "step": 381600
+    },
+    {
+      "epoch": 0.09955936246544236,
+      "grad_norm": 6.4748711585998535,
+      "learning_rate": 9.838260064268192e-06,
+      "loss": 2.727,
+      "step": 381800
+    },
+    {
+      "epoch": 0.09961151509114453,
+      "grad_norm": 7.302340507507324,
+      "learning_rate": 9.83804910098956e-06,
+      "loss": 2.7436,
+      "step": 382000
+    },
+    {
+      "epoch": 0.0996636677168467,
+      "grad_norm": 6.783388614654541,
+      "learning_rate": 9.837838002481876e-06,
+      "loss": 2.6953,
+      "step": 382200
+    },
+    {
+      "epoch": 0.09971582034254887,
+      "grad_norm": 7.748034477233887,
+      "learning_rate": 9.837626768751043e-06,
+      "loss": 2.737,
+      "step": 382400
+    },
+    {
+      "epoch": 0.09976797296825105,
+      "grad_norm": 7.11197566986084,
+      "learning_rate": 9.837415399802962e-06,
+      "loss": 2.7801,
+      "step": 382600
+    },
+    {
+      "epoch": 0.09982012559395322,
+      "grad_norm": 7.256601333618164,
+      "learning_rate": 9.837203895643546e-06,
+      "loss": 2.7174,
+      "step": 382800
+    },
+    {
+      "epoch": 0.09987227821965539,
+      "grad_norm": 7.089999675750732,
+      "learning_rate": 9.8369922562787e-06,
+      "loss": 2.7441,
+      "step": 383000
+    },
+    {
+      "epoch": 0.09992443084535756,
+      "grad_norm": 6.694625377655029,
+      "learning_rate": 9.836780481714348e-06,
+      "loss": 2.6984,
+      "step": 383200
+    },
+    {
+      "epoch": 0.09997658347105973,
+      "grad_norm": 7.070290565490723,
+      "learning_rate": 9.8365685719564e-06,
+      "loss": 2.7062,
+      "step": 383400
+    },
+    {
+      "epoch": 0.1000287360967619,
+      "grad_norm": 7.1637959480285645,
+      "learning_rate": 9.836356527010785e-06,
+      "loss": 2.7344,
+      "step": 383600
+    },
+    {
+      "epoch": 0.10008088872246407,
+      "grad_norm": 6.960715293884277,
+      "learning_rate": 9.83614434688343e-06,
+      "loss": 2.7294,
+      "step": 383800
+    },
+    {
+      "epoch": 0.10013304134816624,
+      "grad_norm": 6.766301155090332,
+      "learning_rate": 9.835932031580262e-06,
+      "loss": 2.7184,
+      "step": 384000
+    },
+    {
+      "epoch": 0.10018519397386841,
+      "grad_norm": 6.3087358474731445,
+      "learning_rate": 9.83571958110722e-06,
+      "loss": 2.6987,
+      "step": 384200
+    },
+    {
+      "epoch": 0.10023734659957058,
+      "grad_norm": 7.121251583099365,
+      "learning_rate": 9.835506995470237e-06,
+      "loss": 2.7211,
+      "step": 384400
+    },
+    {
+      "epoch": 0.10028949922527275,
+      "grad_norm": 6.787637710571289,
+      "learning_rate": 9.83529427467526e-06,
+      "loss": 2.7046,
+      "step": 384600
+    },
+    {
+      "epoch": 0.10034165185097492,
+      "grad_norm": 6.736263751983643,
+      "learning_rate": 9.835081418728232e-06,
+      "loss": 2.7612,
+      "step": 384800
+    },
+    {
+      "epoch": 0.10039380447667709,
+      "grad_norm": 7.012972831726074,
+      "learning_rate": 9.834868427635102e-06,
+      "loss": 2.6529,
+      "step": 385000
+    },
+    {
+      "epoch": 0.10044595710237926,
+      "grad_norm": 7.4089531898498535,
+      "learning_rate": 9.834655301401827e-06,
+      "loss": 2.7274,
+      "step": 385200
+    },
+    {
+      "epoch": 0.10049810972808143,
+      "grad_norm": 6.628983497619629,
+      "learning_rate": 9.83444204003436e-06,
+      "loss": 2.7027,
+      "step": 385400
+    },
+    {
+      "epoch": 0.1005502623537836,
+      "grad_norm": 6.684551239013672,
+      "learning_rate": 9.834228643538665e-06,
+      "loss": 2.72,
+      "step": 385600
+    },
+    {
+      "epoch": 0.10060241497948577,
+      "grad_norm": 7.784319877624512,
+      "learning_rate": 9.834015111920705e-06,
+      "loss": 2.7083,
+      "step": 385800
+    },
+    {
+      "epoch": 0.10065456760518793,
+      "grad_norm": 6.074433326721191,
+      "learning_rate": 9.833801445186448e-06,
+      "loss": 2.7258,
+      "step": 386000
+    },
+    {
+      "epoch": 0.1007067202308901,
+      "grad_norm": 7.0298171043396,
+      "learning_rate": 9.833587643341869e-06,
+      "loss": 2.7326,
+      "step": 386200
+    },
+    {
+      "epoch": 0.10075887285659227,
+      "grad_norm": 6.243155002593994,
+      "learning_rate": 9.83337370639294e-06,
+      "loss": 2.6969,
+      "step": 386400
+    },
+    {
+      "epoch": 0.10081102548229444,
+      "grad_norm": 6.764981269836426,
+      "learning_rate": 9.833159634345643e-06,
+      "loss": 2.709,
+      "step": 386600
+    },
+    {
+      "epoch": 0.10086317810799661,
+      "grad_norm": 6.844011306762695,
+      "learning_rate": 9.832945427205963e-06,
+      "loss": 2.738,
+      "step": 386800
+    },
+    {
+      "epoch": 0.10091533073369878,
+      "grad_norm": 7.266912460327148,
+      "learning_rate": 9.832731084979883e-06,
+      "loss": 2.7677,
+      "step": 387000
+    },
+    {
+      "epoch": 0.10096748335940095,
+      "grad_norm": 7.799858093261719,
+      "learning_rate": 9.8325166076734e-06,
+      "loss": 2.7008,
+      "step": 387200
+    },
+    {
+      "epoch": 0.10101963598510312,
+      "grad_norm": 6.45143461227417,
+      "learning_rate": 9.832301995292502e-06,
+      "loss": 2.7465,
+      "step": 387400
+    },
+    {
+      "epoch": 0.10107178861080529,
+      "grad_norm": 6.146852970123291,
+      "learning_rate": 9.832087247843194e-06,
+      "loss": 2.6801,
+      "step": 387600
+    },
+    {
+      "epoch": 0.10112394123650746,
+      "grad_norm": 6.76859188079834,
+      "learning_rate": 9.831872365331475e-06,
+      "loss": 2.7365,
+      "step": 387800
+    },
+    {
+      "epoch": 0.10117609386220963,
+      "grad_norm": 6.941719055175781,
+      "learning_rate": 9.831657347763353e-06,
+      "loss": 2.7285,
+      "step": 388000
+    },
+    {
+      "epoch": 0.1012282464879118,
+      "grad_norm": 7.166417121887207,
+      "learning_rate": 9.831442195144836e-06,
+      "loss": 2.7176,
+      "step": 388200
+    },
+    {
+      "epoch": 0.10128039911361397,
+      "grad_norm": 8.05933952331543,
+      "learning_rate": 9.83122690748194e-06,
+      "loss": 2.6899,
+      "step": 388400
+    },
+    {
+      "epoch": 0.10133255173931614,
+      "grad_norm": 6.222696781158447,
+      "learning_rate": 9.831011484780679e-06,
+      "loss": 2.7254,
+      "step": 388600
+    },
+    {
+      "epoch": 0.10138470436501831,
+      "grad_norm": 7.399394512176514,
+      "learning_rate": 9.830795927047078e-06,
+      "loss": 2.7333,
+      "step": 388800
+    },
+    {
+      "epoch": 0.10143685699072048,
+      "grad_norm": 7.602478504180908,
+      "learning_rate": 9.83058023428716e-06,
+      "loss": 2.7245,
+      "step": 389000
+    },
+    {
+      "epoch": 0.10148900961642265,
+      "grad_norm": 6.894057273864746,
+      "learning_rate": 9.830364406506956e-06,
+      "loss": 2.7472,
+      "step": 389200
+    },
+    {
+      "epoch": 0.10154116224212482,
+      "grad_norm": 6.566694259643555,
+      "learning_rate": 9.830148443712498e-06,
+      "loss": 2.6904,
+      "step": 389400
+    },
+    {
+      "epoch": 0.101593314867827,
+      "grad_norm": 6.913676738739014,
+      "learning_rate": 9.829932345909819e-06,
+      "loss": 2.7406,
+      "step": 389600
+    },
+    {
+      "epoch": 0.10164546749352917,
+      "grad_norm": 6.623051166534424,
+      "learning_rate": 9.829716113104964e-06,
+      "loss": 2.71,
+      "step": 389800
+    },
+    {
+      "epoch": 0.10169762011923134,
+      "grad_norm": 6.746899127960205,
+      "learning_rate": 9.829499745303972e-06,
+      "loss": 2.7284,
+      "step": 390000
+    },
+    {
+      "epoch": 0.1017497727449335,
+      "grad_norm": 7.177034854888916,
+      "learning_rate": 9.829283242512896e-06,
+      "loss": 2.7186,
+      "step": 390200
+    },
+    {
+      "epoch": 0.10180192537063568,
+      "grad_norm": 7.524350166320801,
+      "learning_rate": 9.829066604737784e-06,
+      "loss": 2.692,
+      "step": 390400
+    },
+    {
+      "epoch": 0.10185407799633785,
+      "grad_norm": 6.514491558074951,
+      "learning_rate": 9.828849831984693e-06,
+      "loss": 2.7025,
+      "step": 390600
+    },
+    {
+      "epoch": 0.10190623062204002,
+      "grad_norm": 7.224747180938721,
+      "learning_rate": 9.82863292425968e-06,
+      "loss": 2.7793,
+      "step": 390800
+    },
+    {
+      "epoch": 0.10195838324774219,
+      "grad_norm": 7.076272010803223,
+      "learning_rate": 9.82841588156881e-06,
+      "loss": 2.7168,
+      "step": 391000
+    },
+    {
+      "epoch": 0.10201053587344436,
+      "grad_norm": 7.147103786468506,
+      "learning_rate": 9.828198703918148e-06,
+      "loss": 2.7419,
+      "step": 391200
+    },
+    {
+      "epoch": 0.10206268849914653,
+      "grad_norm": 6.700717926025391,
+      "learning_rate": 9.827981391313766e-06,
+      "loss": 2.668,
+      "step": 391400
+    },
+    {
+      "epoch": 0.1021148411248487,
+      "grad_norm": 6.936060428619385,
+      "learning_rate": 9.827763943761736e-06,
+      "loss": 2.709,
+      "step": 391600
+    },
+    {
+      "epoch": 0.10216699375055086,
+      "grad_norm": 6.975337982177734,
+      "learning_rate": 9.827546361268139e-06,
+      "loss": 2.7271,
+      "step": 391800
+    },
+    {
+      "epoch": 0.10221914637625303,
+      "grad_norm": 7.4715118408203125,
+      "learning_rate": 9.827328643839054e-06,
+      "loss": 2.7067,
+      "step": 392000
+    },
+    {
+      "epoch": 0.1022712990019552,
+      "grad_norm": 8.031476974487305,
+      "learning_rate": 9.827110791480568e-06,
+      "loss": 2.7094,
+      "step": 392200
+    },
+    {
+      "epoch": 0.10232345162765737,
+      "grad_norm": 6.826619625091553,
+      "learning_rate": 9.82689280419877e-06,
+      "loss": 2.6987,
+      "step": 392400
+    },
+    {
+      "epoch": 0.10237560425335954,
+      "grad_norm": 7.765749931335449,
+      "learning_rate": 9.826674681999749e-06,
+      "loss": 2.7286,
+      "step": 392600
+    },
+    {
+      "epoch": 0.10242775687906171,
+      "grad_norm": 7.1541666984558105,
+      "learning_rate": 9.826456424889608e-06,
+      "loss": 2.7061,
+      "step": 392800
+    },
+    {
+      "epoch": 0.10247990950476388,
+      "grad_norm": 6.534503936767578,
+      "learning_rate": 9.826238032874445e-06,
+      "loss": 2.7177,
+      "step": 393000
+    },
+    {
+      "epoch": 0.10253206213046605,
+      "grad_norm": 7.043641090393066,
+      "learning_rate": 9.826019505960363e-06,
+      "loss": 2.7428,
+      "step": 393200
+    },
+    {
+      "epoch": 0.10258421475616822,
+      "grad_norm": 6.870479583740234,
+      "learning_rate": 9.82580084415347e-06,
+      "loss": 2.73,
+      "step": 393400
+    },
+    {
+      "epoch": 0.10263636738187039,
+      "grad_norm": 7.458539962768555,
+      "learning_rate": 9.825582047459883e-06,
+      "loss": 2.6956,
+      "step": 393600
+    },
+    {
+      "epoch": 0.10268852000757256,
+      "grad_norm": 7.4572038650512695,
+      "learning_rate": 9.825363115885711e-06,
+      "loss": 2.7238,
+      "step": 393800
+    },
+    {
+      "epoch": 0.10274067263327473,
+      "grad_norm": 6.63360595703125,
+      "learning_rate": 9.825144049437077e-06,
+      "loss": 2.752,
+      "step": 394000
+    },
+    {
+      "epoch": 0.1027928252589769,
+      "grad_norm": 6.744916915893555,
+      "learning_rate": 9.824924848120101e-06,
+      "loss": 2.694,
+      "step": 394200
+    },
+    {
+      "epoch": 0.10284497788467907,
+      "grad_norm": 7.818175315856934,
+      "learning_rate": 9.824705511940914e-06,
+      "loss": 2.7334,
+      "step": 394400
+    },
+    {
+      "epoch": 0.10289713051038124,
+      "grad_norm": 6.705904483795166,
+      "learning_rate": 9.824486040905646e-06,
+      "loss": 2.7112,
+      "step": 394600
+    },
+    {
+      "epoch": 0.10294928313608341,
+      "grad_norm": 6.82873010635376,
+      "learning_rate": 9.824266435020428e-06,
+      "loss": 2.6945,
+      "step": 394800
+    },
+    {
+      "epoch": 0.10300143576178558,
+      "grad_norm": 7.144585132598877,
+      "learning_rate": 9.824046694291401e-06,
+      "loss": 2.7079,
+      "step": 395000
+    },
+    {
+      "epoch": 0.10305358838748775,
+      "grad_norm": 7.308895587921143,
+      "learning_rate": 9.823826818724707e-06,
+      "loss": 2.6953,
+      "step": 395200
+    },
+    {
+      "epoch": 0.10310574101318992,
+      "grad_norm": 6.582967281341553,
+      "learning_rate": 9.823606808326491e-06,
+      "loss": 2.7148,
+      "step": 395400
+    },
+    {
+      "epoch": 0.10315789363889209,
+      "grad_norm": 7.389861106872559,
+      "learning_rate": 9.823386663102902e-06,
+      "loss": 2.6945,
+      "step": 395600
+    },
+    {
+      "epoch": 0.10321004626459426,
+      "grad_norm": 7.0465006828308105,
+      "learning_rate": 9.823166383060096e-06,
+      "loss": 2.7263,
+      "step": 395800
+    },
+    {
+      "epoch": 0.10326219889029643,
+      "grad_norm": 7.7955322265625,
+      "learning_rate": 9.82294596820423e-06,
+      "loss": 2.7256,
+      "step": 396000
+    },
+    {
+      "epoch": 0.1033143515159986,
+      "grad_norm": 7.1410322189331055,
+      "learning_rate": 9.82272541854146e-06,
+      "loss": 2.6904,
+      "step": 396200
+    },
+    {
+      "epoch": 0.10336650414170077,
+      "grad_norm": 7.031514644622803,
+      "learning_rate": 9.822504734077953e-06,
+      "loss": 2.6924,
+      "step": 396400
+    },
+    {
+      "epoch": 0.10341865676740294,
+      "grad_norm": 7.500546932220459,
+      "learning_rate": 9.822283914819882e-06,
+      "loss": 2.6723,
+      "step": 396600
+    },
+    {
+      "epoch": 0.10347080939310511,
+      "grad_norm": 7.630676746368408,
+      "learning_rate": 9.82206296077341e-06,
+      "loss": 2.7074,
+      "step": 396800
+    },
+    {
+      "epoch": 0.10352296201880729,
+      "grad_norm": 6.640892028808594,
+      "learning_rate": 9.821841871944723e-06,
+      "loss": 2.7259,
+      "step": 397000
+    },
+    {
+      "epoch": 0.10357511464450946,
+      "grad_norm": 7.342281818389893,
+      "learning_rate": 9.821620648339993e-06,
+      "loss": 2.6999,
+      "step": 397200
+    },
+    {
+      "epoch": 0.10362726727021163,
+      "grad_norm": 7.83634614944458,
+      "learning_rate": 9.821399289965408e-06,
+      "loss": 2.7423,
+      "step": 397400
+    },
+    {
+      "epoch": 0.1036794198959138,
+      "grad_norm": 7.7452826499938965,
+      "learning_rate": 9.821177796827152e-06,
+      "loss": 2.7492,
+      "step": 397600
+    },
+    {
+      "epoch": 0.10373157252161595,
+      "grad_norm": 7.283862113952637,
+      "learning_rate": 9.820956168931418e-06,
+      "loss": 2.6983,
+      "step": 397800
+    },
+    {
+      "epoch": 0.10378372514731812,
+      "grad_norm": 7.292904376983643,
+      "learning_rate": 9.820734406284402e-06,
+      "loss": 2.7275,
+      "step": 398000
+    },
+    {
+      "epoch": 0.1038358777730203,
+      "grad_norm": 7.42738151550293,
+      "learning_rate": 9.820512508892299e-06,
+      "loss": 2.673,
+      "step": 398200
+    },
+    {
+      "epoch": 0.10388803039872246,
+      "grad_norm": 7.42814302444458,
+      "learning_rate": 9.820290476761314e-06,
+      "loss": 2.7334,
+      "step": 398400
+    },
+    {
+      "epoch": 0.10394018302442463,
+      "grad_norm": 7.708849906921387,
+      "learning_rate": 9.820068309897653e-06,
+      "loss": 2.6999,
+      "step": 398600
+    },
+    {
+      "epoch": 0.1039923356501268,
+      "grad_norm": 7.1372551918029785,
+      "learning_rate": 9.819846008307523e-06,
+      "loss": 2.6921,
+      "step": 398800
+    },
+    {
+      "epoch": 0.10404448827582898,
+      "grad_norm": 6.49362850189209,
+      "learning_rate": 9.81962357199714e-06,
+      "loss": 2.7561,
+      "step": 399000
+    },
+    {
+      "epoch": 0.10409664090153115,
+      "grad_norm": 7.146295547485352,
+      "learning_rate": 9.819401000972721e-06,
+      "loss": 2.7148,
+      "step": 399200
+    },
+    {
+      "epoch": 0.10414879352723332,
+      "grad_norm": 7.127407550811768,
+      "learning_rate": 9.819178295240488e-06,
+      "loss": 2.7187,
+      "step": 399400
+    },
+    {
+      "epoch": 0.10420094615293549,
+      "grad_norm": 7.735341548919678,
+      "learning_rate": 9.818955454806664e-06,
+      "loss": 2.7221,
+      "step": 399600
+    },
+    {
+      "epoch": 0.10425309877863766,
+      "grad_norm": 7.073373317718506,
+      "learning_rate": 9.81873247967748e-06,
+      "loss": 2.7069,
+      "step": 399800
+    },
+    {
+      "epoch": 0.10430525140433983,
+      "grad_norm": 6.615175247192383,
+      "learning_rate": 9.818509369859165e-06,
+      "loss": 2.6807,
+      "step": 400000
+    },
+    {
+      "epoch": 0.104357404030042,
+      "grad_norm": 7.4061174392700195,
+      "learning_rate": 9.818286125357958e-06,
+      "loss": 2.671,
+      "step": 400200
+    },
+    {
+      "epoch": 0.10440955665574417,
+      "grad_norm": 7.3244524002075195,
+      "learning_rate": 9.818062746180098e-06,
+      "loss": 2.7053,
+      "step": 400400
+    },
+    {
+      "epoch": 0.10446170928144634,
+      "grad_norm": 7.333294868469238,
+      "learning_rate": 9.81783923233183e-06,
+      "loss": 2.7226,
+      "step": 400600
+    },
+    {
+      "epoch": 0.10451386190714851,
+      "grad_norm": 8.426340103149414,
+      "learning_rate": 9.817615583819397e-06,
+      "loss": 2.7258,
+      "step": 400800
+    },
+    {
+      "epoch": 0.10456601453285068,
+      "grad_norm": 7.6054558753967285,
+      "learning_rate": 9.817391800649056e-06,
+      "loss": 2.6957,
+      "step": 401000
+    },
+    {
+      "epoch": 0.10461816715855285,
+      "grad_norm": 7.174556732177734,
+      "learning_rate": 9.81716788282706e-06,
+      "loss": 2.6637,
+      "step": 401200
+    },
+    {
+      "epoch": 0.10467031978425502,
+      "grad_norm": 7.382963180541992,
+      "learning_rate": 9.816943830359666e-06,
+      "loss": 2.6776,
+      "step": 401400
+    },
+    {
+      "epoch": 0.10472247240995719,
+      "grad_norm": 7.8001203536987305,
+      "learning_rate": 9.816719643253138e-06,
+      "loss": 2.7124,
+      "step": 401600
+    },
+    {
+      "epoch": 0.10477462503565936,
+      "grad_norm": 7.5196733474731445,
+      "learning_rate": 9.816495321513744e-06,
+      "loss": 2.6645,
+      "step": 401800
+    },
+    {
+      "epoch": 0.10482677766136153,
+      "grad_norm": 6.8788628578186035,
+      "learning_rate": 9.81627086514775e-06,
+      "loss": 2.7302,
+      "step": 402000
+    },
+    {
+      "epoch": 0.1048789302870637,
+      "grad_norm": 7.129185199737549,
+      "learning_rate": 9.816046274161432e-06,
+      "loss": 2.6782,
+      "step": 402200
+    },
+    {
+      "epoch": 0.10493108291276587,
+      "grad_norm": 7.342111587524414,
+      "learning_rate": 9.815821548561069e-06,
+      "loss": 2.711,
+      "step": 402400
+    },
+    {
+      "epoch": 0.10498323553846804,
+      "grad_norm": 5.888463020324707,
+      "learning_rate": 9.81559668835294e-06,
+      "loss": 2.6887,
+      "step": 402600
+    },
+    {
+      "epoch": 0.10503538816417021,
+      "grad_norm": 7.24282169342041,
+      "learning_rate": 9.81537169354333e-06,
+      "loss": 2.746,
+      "step": 402800
+    },
+    {
+      "epoch": 0.10508754078987238,
+      "grad_norm": 7.5223822593688965,
+      "learning_rate": 9.81514656413853e-06,
+      "loss": 2.7138,
+      "step": 403000
+    },
+    {
+      "epoch": 0.10513969341557455,
+      "grad_norm": 6.787247180938721,
+      "learning_rate": 9.814921300144831e-06,
+      "loss": 2.7083,
+      "step": 403200
+    },
+    {
+      "epoch": 0.10519184604127672,
+      "grad_norm": 8.027321815490723,
+      "learning_rate": 9.814695901568529e-06,
+      "loss": 2.7397,
+      "step": 403400
+    },
+    {
+      "epoch": 0.10524399866697888,
+      "grad_norm": 7.712162017822266,
+      "learning_rate": 9.814470368415925e-06,
+      "loss": 2.6962,
+      "step": 403600
+    },
+    {
+      "epoch": 0.10529615129268105,
+      "grad_norm": 7.541025638580322,
+      "learning_rate": 9.814244700693325e-06,
+      "loss": 2.717,
+      "step": 403800
+    },
+    {
+      "epoch": 0.10534830391838322,
+      "grad_norm": 7.47944974899292,
+      "learning_rate": 9.814018898407034e-06,
+      "loss": 2.6941,
+      "step": 404000
+    },
+    {
+      "epoch": 0.10540045654408539,
+      "grad_norm": 7.1561737060546875,
+      "learning_rate": 9.813792961563363e-06,
+      "loss": 2.6746,
+      "step": 404200
+    },
+    {
+      "epoch": 0.10545260916978756,
+      "grad_norm": 7.5194220542907715,
+      "learning_rate": 9.81356689016863e-06,
+      "loss": 2.6951,
+      "step": 404400
+    },
+    {
+      "epoch": 0.10550476179548973,
+      "grad_norm": 7.529834747314453,
+      "learning_rate": 9.813340684229148e-06,
+      "loss": 2.6781,
+      "step": 404600
+    },
+    {
+      "epoch": 0.1055569144211919,
+      "grad_norm": 7.3788533210754395,
+      "learning_rate": 9.813114343751248e-06,
+      "loss": 2.6886,
+      "step": 404800
+    },
+    {
+      "epoch": 0.10560906704689407,
+      "grad_norm": 7.26027250289917,
+      "learning_rate": 9.81288786874125e-06,
+      "loss": 2.7073,
+      "step": 405000
+    },
+    {
+      "epoch": 0.10566121967259624,
+      "grad_norm": 7.290615081787109,
+      "learning_rate": 9.812661259205489e-06,
+      "loss": 2.72,
+      "step": 405200
+    },
+    {
+      "epoch": 0.10571337229829841,
+      "grad_norm": 7.432085037231445,
+      "learning_rate": 9.812434515150294e-06,
+      "loss": 2.6673,
+      "step": 405400
+    },
+    {
+      "epoch": 0.10576552492400058,
+      "grad_norm": 7.36521577835083,
+      "learning_rate": 9.812207636582007e-06,
+      "loss": 2.7188,
+      "step": 405600
+    },
+    {
+      "epoch": 0.10581767754970275,
+      "grad_norm": 6.904064655303955,
+      "learning_rate": 9.811980623506967e-06,
+      "loss": 2.6936,
+      "step": 405800
+    },
+    {
+      "epoch": 0.10586983017540492,
+      "grad_norm": 6.532309055328369,
+      "learning_rate": 9.81175347593152e-06,
+      "loss": 2.6753,
+      "step": 406000
+    },
+    {
+      "epoch": 0.1059219828011071,
+      "grad_norm": 7.805009365081787,
+      "learning_rate": 9.811526193862017e-06,
+      "loss": 2.7279,
+      "step": 406200
+    },
+    {
+      "epoch": 0.10597413542680927,
+      "grad_norm": 6.72572135925293,
+      "learning_rate": 9.811298777304807e-06,
+      "loss": 2.6603,
+      "step": 406400
+    },
+    {
+      "epoch": 0.10602628805251144,
+      "grad_norm": 6.9229559898376465,
+      "learning_rate": 9.811071226266248e-06,
+      "loss": 2.6847,
+      "step": 406600
+    },
+    {
+      "epoch": 0.1060784406782136,
+      "grad_norm": 7.181591033935547,
+      "learning_rate": 9.810843540752703e-06,
+      "loss": 2.7129,
+      "step": 406800
+    },
+    {
+      "epoch": 0.10613059330391578,
+      "grad_norm": 7.120484828948975,
+      "learning_rate": 9.810615720770533e-06,
+      "loss": 2.7121,
+      "step": 407000
+    },
+    {
+      "epoch": 0.10618274592961795,
+      "grad_norm": 8.234344482421875,
+      "learning_rate": 9.810387766326108e-06,
+      "loss": 2.6959,
+      "step": 407200
+    },
+    {
+      "epoch": 0.10623489855532012,
+      "grad_norm": 7.5896100997924805,
+      "learning_rate": 9.810159677425797e-06,
+      "loss": 2.7096,
+      "step": 407400
+    },
+    {
+      "epoch": 0.10628705118102229,
+      "grad_norm": 6.704971790313721,
+      "learning_rate": 9.809931454075976e-06,
+      "loss": 2.7057,
+      "step": 407600
+    },
+    {
+      "epoch": 0.10633920380672446,
+      "grad_norm": 8.379301071166992,
+      "learning_rate": 9.809703096283025e-06,
+      "loss": 2.7059,
+      "step": 407800
+    },
+    {
+      "epoch": 0.10639135643242663,
+      "grad_norm": 8.338081359863281,
+      "learning_rate": 9.809474604053329e-06,
+      "loss": 2.7231,
+      "step": 408000
+    },
+    {
+      "epoch": 0.1064435090581288,
+      "grad_norm": 7.795104026794434,
+      "learning_rate": 9.809245977393268e-06,
+      "loss": 2.7019,
+      "step": 408200
+    },
+    {
+      "epoch": 0.10649566168383097,
+      "grad_norm": 6.772091865539551,
+      "learning_rate": 9.80901721630924e-06,
+      "loss": 2.7151,
+      "step": 408400
+    },
+    {
+      "epoch": 0.10654781430953314,
+      "grad_norm": 7.263775825500488,
+      "learning_rate": 9.808788320807636e-06,
+      "loss": 2.7085,
+      "step": 408600
+    },
+    {
+      "epoch": 0.10659996693523531,
+      "grad_norm": 7.627434730529785,
+      "learning_rate": 9.808559290894852e-06,
+      "loss": 2.6663,
+      "step": 408800
+    },
+    {
+      "epoch": 0.10665211956093748,
+      "grad_norm": 7.4458465576171875,
+      "learning_rate": 9.808330126577293e-06,
+      "loss": 2.6748,
+      "step": 409000
+    },
+    {
+      "epoch": 0.10670427218663965,
+      "grad_norm": 7.223238468170166,
+      "learning_rate": 9.808100827861361e-06,
+      "loss": 2.6919,
+      "step": 409200
+    },
+    {
+      "epoch": 0.10675642481234182,
+      "grad_norm": 7.894956111907959,
+      "learning_rate": 9.807871394753468e-06,
+      "loss": 2.6707,
+      "step": 409400
+    },
+    {
+      "epoch": 0.10680857743804398,
+      "grad_norm": 7.548394203186035,
+      "learning_rate": 9.807641827260027e-06,
+      "loss": 2.6829,
+      "step": 409600
+    },
+    {
+      "epoch": 0.10686073006374615,
+      "grad_norm": 6.605945587158203,
+      "learning_rate": 9.807412125387449e-06,
+      "loss": 2.7355,
+      "step": 409800
+    },
+    {
+      "epoch": 0.10691288268944832,
+      "grad_norm": 6.510282516479492,
+      "learning_rate": 9.807182289142163e-06,
+      "loss": 2.726,
+      "step": 410000
+    },
+    {
+      "epoch": 0.10696503531515049,
+      "grad_norm": 6.95159387588501,
+      "learning_rate": 9.806952318530589e-06,
+      "loss": 2.7377,
+      "step": 410200
+    },
+    {
+      "epoch": 0.10701718794085266,
+      "grad_norm": 8.0206298828125,
+      "learning_rate": 9.806722213559153e-06,
+      "loss": 2.7229,
+      "step": 410400
+    },
+    {
+      "epoch": 0.10706934056655483,
+      "grad_norm": 7.942624568939209,
+      "learning_rate": 9.806491974234291e-06,
+      "loss": 2.7275,
+      "step": 410600
+    },
+    {
+      "epoch": 0.107121493192257,
+      "grad_norm": 8.461700439453125,
+      "learning_rate": 9.806261600562434e-06,
+      "loss": 2.6925,
+      "step": 410800
+    },
+    {
+      "epoch": 0.10717364581795917,
+      "grad_norm": 7.115810871124268,
+      "learning_rate": 9.806031092550025e-06,
+      "loss": 2.6673,
+      "step": 411000
+    },
+    {
+      "epoch": 0.10722579844366134,
+      "grad_norm": 7.023535251617432,
+      "learning_rate": 9.805800450203504e-06,
+      "loss": 2.768,
+      "step": 411200
+    },
+    {
+      "epoch": 0.10727795106936351,
+      "grad_norm": 7.725059986114502,
+      "learning_rate": 9.80556967352932e-06,
+      "loss": 2.6898,
+      "step": 411400
+    },
+    {
+      "epoch": 0.10733010369506568,
+      "grad_norm": 7.8010573387146,
+      "learning_rate": 9.805338762533924e-06,
+      "loss": 2.7011,
+      "step": 411600
+    },
+    {
+      "epoch": 0.10738225632076785,
+      "grad_norm": 7.579238414764404,
+      "learning_rate": 9.805107717223767e-06,
+      "loss": 2.7297,
+      "step": 411800
+    },
+    {
+      "epoch": 0.10743440894647002,
+      "grad_norm": 8.260748863220215,
+      "learning_rate": 9.80487653760531e-06,
+      "loss": 2.6825,
+      "step": 412000
+    },
+    {
+      "epoch": 0.10748656157217219,
+      "grad_norm": 7.9606547355651855,
+      "learning_rate": 9.804645223685012e-06,
+      "loss": 2.7009,
+      "step": 412200
+    },
+    {
+      "epoch": 0.10753871419787436,
+      "grad_norm": 8.684904098510742,
+      "learning_rate": 9.804413775469342e-06,
+      "loss": 2.7123,
+      "step": 412400
+    },
+    {
+      "epoch": 0.10759086682357653,
+      "grad_norm": 7.265321731567383,
+      "learning_rate": 9.804182192964765e-06,
+      "loss": 2.664,
+      "step": 412600
+    },
+    {
+      "epoch": 0.1076430194492787,
+      "grad_norm": 7.080298900604248,
+      "learning_rate": 9.803950476177757e-06,
+      "loss": 2.705,
+      "step": 412800
+    },
+    {
+      "epoch": 0.10769517207498087,
+      "grad_norm": 7.625601768493652,
+      "learning_rate": 9.803718625114796e-06,
+      "loss": 2.7355,
+      "step": 413000
+    },
+    {
+      "epoch": 0.10774732470068304,
+      "grad_norm": 7.863028526306152,
+      "learning_rate": 9.803486639782357e-06,
+      "loss": 2.7024,
+      "step": 413200
+    },
+    {
+      "epoch": 0.10779947732638522,
+      "grad_norm": 7.483668327331543,
+      "learning_rate": 9.803254520186932e-06,
+      "loss": 2.7058,
+      "step": 413400
+    },
+    {
+      "epoch": 0.10785162995208739,
+      "grad_norm": 7.60951042175293,
+      "learning_rate": 9.803022266335e-06,
+      "loss": 2.689,
+      "step": 413600
+    },
+    {
+      "epoch": 0.10790378257778956,
+      "grad_norm": 7.47837495803833,
+      "learning_rate": 9.80278987823306e-06,
+      "loss": 2.6998,
+      "step": 413800
+    },
+    {
+      "epoch": 0.10795593520349173,
+      "grad_norm": 7.093349933624268,
+      "learning_rate": 9.802557355887607e-06,
+      "loss": 2.7003,
+      "step": 414000
+    },
+    {
+      "epoch": 0.1080080878291939,
+      "grad_norm": 7.799093723297119,
+      "learning_rate": 9.802324699305136e-06,
+      "loss": 2.7044,
+      "step": 414200
+    },
+    {
+      "epoch": 0.10806024045489607,
+      "grad_norm": 7.693939208984375,
+      "learning_rate": 9.802091908492153e-06,
+      "loss": 2.6745,
+      "step": 414400
+    },
+    {
+      "epoch": 0.10811239308059824,
+      "grad_norm": 7.931087493896484,
+      "learning_rate": 9.801858983455164e-06,
+      "loss": 2.703,
+      "step": 414600
+    },
+    {
+      "epoch": 0.10816454570630041,
+      "grad_norm": 7.6536407470703125,
+      "learning_rate": 9.801625924200679e-06,
+      "loss": 2.6576,
+      "step": 414800
+    },
+    {
+      "epoch": 0.10821669833200258,
+      "grad_norm": 7.151515483856201,
+      "learning_rate": 9.801392730735214e-06,
+      "loss": 2.7166,
+      "step": 415000
+    },
+    {
+      "epoch": 0.10826885095770475,
+      "grad_norm": 8.142443656921387,
+      "learning_rate": 9.801159403065286e-06,
+      "loss": 2.6738,
+      "step": 415200
+    },
+    {
+      "epoch": 0.1083210035834069,
+      "grad_norm": 7.3941121101379395,
+      "learning_rate": 9.800925941197415e-06,
+      "loss": 2.7265,
+      "step": 415400
+    },
+    {
+      "epoch": 0.10837315620910908,
+      "grad_norm": 7.601657390594482,
+      "learning_rate": 9.80069234513813e-06,
+      "loss": 2.6959,
+      "step": 415600
+    },
+    {
+      "epoch": 0.10842530883481125,
+      "grad_norm": 7.414285182952881,
+      "learning_rate": 9.800458614893958e-06,
+      "loss": 2.7261,
+      "step": 415800
+    },
+    {
+      "epoch": 0.10847746146051342,
+      "grad_norm": 7.267521858215332,
+      "learning_rate": 9.800224750471434e-06,
+      "loss": 2.6731,
+      "step": 416000
+    },
+    {
+      "epoch": 0.10852961408621559,
+      "grad_norm": 7.712509632110596,
+      "learning_rate": 9.79999075187709e-06,
+      "loss": 2.7288,
+      "step": 416200
+    },
+    {
+      "epoch": 0.10858176671191776,
+      "grad_norm": 7.250224590301514,
+      "learning_rate": 9.799756619117473e-06,
+      "loss": 2.7043,
+      "step": 416400
+    },
+    {
+      "epoch": 0.10863391933761993,
+      "grad_norm": 7.441985130310059,
+      "learning_rate": 9.799522352199122e-06,
+      "loss": 2.6993,
+      "step": 416600
+    },
+    {
+      "epoch": 0.1086860719633221,
+      "grad_norm": 7.325314044952393,
+      "learning_rate": 9.79928795112859e-06,
+      "loss": 2.6846,
+      "step": 416800
+    },
+    {
+      "epoch": 0.10873822458902427,
+      "grad_norm": 7.375991344451904,
+      "learning_rate": 9.799053415912422e-06,
+      "loss": 2.7052,
+      "step": 417000
+    },
+    {
+      "epoch": 0.10879037721472644,
+      "grad_norm": 7.841715335845947,
+      "learning_rate": 9.798818746557182e-06,
+      "loss": 2.7222,
+      "step": 417200
+    },
+    {
+      "epoch": 0.10884252984042861,
+      "grad_norm": 7.707851886749268,
+      "learning_rate": 9.798583943069422e-06,
+      "loss": 2.6917,
+      "step": 417400
+    },
+    {
+      "epoch": 0.10889468246613078,
+      "grad_norm": 8.474435806274414,
+      "learning_rate": 9.798349005455707e-06,
+      "loss": 2.6629,
+      "step": 417600
+    },
+    {
+      "epoch": 0.10894683509183295,
+      "grad_norm": 7.427988529205322,
+      "learning_rate": 9.798113933722606e-06,
+      "loss": 2.6719,
+      "step": 417800
+    },
+    {
+      "epoch": 0.10899898771753512,
+      "grad_norm": 7.597637176513672,
+      "learning_rate": 9.797878727876689e-06,
+      "loss": 2.6991,
+      "step": 418000
+    },
+    {
+      "epoch": 0.10905114034323729,
+      "grad_norm": 8.201911926269531,
+      "learning_rate": 9.797643387924529e-06,
+      "loss": 2.6592,
+      "step": 418200
+    },
+    {
+      "epoch": 0.10910329296893946,
+      "grad_norm": 8.380842208862305,
+      "learning_rate": 9.797407913872703e-06,
+      "loss": 2.7308,
+      "step": 418400
+    },
+    {
+      "epoch": 0.10915544559464163,
+      "grad_norm": 8.024592399597168,
+      "learning_rate": 9.797172305727796e-06,
+      "loss": 2.6984,
+      "step": 418600
+    },
+    {
+      "epoch": 0.1092075982203438,
+      "grad_norm": 7.051395893096924,
+      "learning_rate": 9.79693656349639e-06,
+      "loss": 2.7273,
+      "step": 418800
+    },
+    {
+      "epoch": 0.10925975084604597,
+      "grad_norm": 8.769367218017578,
+      "learning_rate": 9.796700687185076e-06,
+      "loss": 2.7199,
+      "step": 419000
+    },
+    {
+      "epoch": 0.10931190347174814,
+      "grad_norm": 6.581594467163086,
+      "learning_rate": 9.796464676800448e-06,
+      "loss": 2.7053,
+      "step": 419200
+    },
+    {
+      "epoch": 0.10936405609745031,
+      "grad_norm": 7.729722499847412,
+      "learning_rate": 9.796228532349102e-06,
+      "loss": 2.6976,
+      "step": 419400
+    },
+    {
+      "epoch": 0.10941620872315248,
+      "grad_norm": 6.908459186553955,
+      "learning_rate": 9.795992253837638e-06,
+      "loss": 2.704,
+      "step": 419600
+    },
+    {
+      "epoch": 0.10946836134885465,
+      "grad_norm": 7.443035125732422,
+      "learning_rate": 9.79575584127266e-06,
+      "loss": 2.6812,
+      "step": 419800
+    },
+    {
+      "epoch": 0.10952051397455682,
+      "grad_norm": 7.436124801635742,
+      "learning_rate": 9.795519294660775e-06,
+      "loss": 2.6817,
+      "step": 420000
+    },
+    {
+      "epoch": 0.109572666600259,
+      "grad_norm": 7.913905620574951,
+      "learning_rate": 9.795282614008598e-06,
+      "loss": 2.7261,
+      "step": 420200
+    },
+    {
+      "epoch": 0.10962481922596116,
+      "grad_norm": 7.672698497772217,
+      "learning_rate": 9.795045799322744e-06,
+      "loss": 2.7001,
+      "step": 420400
+    },
+    {
+      "epoch": 0.10967697185166334,
+      "grad_norm": 7.240036487579346,
+      "learning_rate": 9.794808850609828e-06,
+      "loss": 2.6813,
+      "step": 420600
+    },
+    {
+      "epoch": 0.1097291244773655,
+      "grad_norm": 7.6813812255859375,
+      "learning_rate": 9.79457176787648e-06,
+      "loss": 2.6612,
+      "step": 420800
+    },
+    {
+      "epoch": 0.10978127710306768,
+      "grad_norm": 7.365231513977051,
+      "learning_rate": 9.79433455112932e-06,
+      "loss": 2.7116,
+      "step": 421000
+    },
+    {
+      "epoch": 0.10983342972876985,
+      "grad_norm": 8.09189510345459,
+      "learning_rate": 9.794097200374981e-06,
+      "loss": 2.6995,
+      "step": 421200
+    },
+    {
+      "epoch": 0.109885582354472,
+      "grad_norm": 7.653886318206787,
+      "learning_rate": 9.7938597156201e-06,
+      "loss": 2.6798,
+      "step": 421400
+    },
+    {
+      "epoch": 0.10993773498017417,
+      "grad_norm": 6.208761692047119,
+      "learning_rate": 9.79362209687131e-06,
+      "loss": 2.6639,
+      "step": 421600
+    },
+    {
+      "epoch": 0.10998988760587634,
+      "grad_norm": 7.84056282043457,
+      "learning_rate": 9.793384344135254e-06,
+      "loss": 2.6726,
+      "step": 421800
+    },
+    {
+      "epoch": 0.11004204023157851,
+      "grad_norm": 6.51863431930542,
+      "learning_rate": 9.793146457418581e-06,
+      "loss": 2.6799,
+      "step": 422000
+    },
+    {
+      "epoch": 0.11009419285728068,
+      "grad_norm": 7.248006820678711,
+      "learning_rate": 9.792908436727937e-06,
+      "loss": 2.7188,
+      "step": 422200
+    },
+    {
+      "epoch": 0.11014634548298285,
+      "grad_norm": 7.784631252288818,
+      "learning_rate": 9.792670282069976e-06,
+      "loss": 2.6932,
+      "step": 422400
+    },
+    {
+      "epoch": 0.11019849810868503,
+      "grad_norm": 7.782918930053711,
+      "learning_rate": 9.792431993451355e-06,
+      "loss": 2.6632,
+      "step": 422600
+    },
+    {
+      "epoch": 0.1102506507343872,
+      "grad_norm": 6.995285511016846,
+      "learning_rate": 9.792193570878733e-06,
+      "loss": 2.7118,
+      "step": 422800
+    },
+    {
+      "epoch": 0.11030280336008937,
+      "grad_norm": 8.329195976257324,
+      "learning_rate": 9.791955014358774e-06,
+      "loss": 2.6919,
+      "step": 423000
+    },
+    {
+      "epoch": 0.11035495598579154,
+      "grad_norm": 7.235878944396973,
+      "learning_rate": 9.791716323898149e-06,
+      "loss": 2.6916,
+      "step": 423200
+    },
+    {
+      "epoch": 0.1104071086114937,
+      "grad_norm": 7.546281814575195,
+      "learning_rate": 9.791477499503528e-06,
+      "loss": 2.695,
+      "step": 423400
+    },
+    {
+      "epoch": 0.11045926123719588,
+      "grad_norm": 7.888071537017822,
+      "learning_rate": 9.791238541181584e-06,
+      "loss": 2.708,
+      "step": 423600
+    },
+    {
+      "epoch": 0.11051141386289805,
+      "grad_norm": 8.009730339050293,
+      "learning_rate": 9.790999448939001e-06,
+      "loss": 2.6739,
+      "step": 423800
+    },
+    {
+      "epoch": 0.11056356648860022,
+      "grad_norm": 7.215407371520996,
+      "learning_rate": 9.790760222782455e-06,
+      "loss": 2.6846,
+      "step": 424000
+    },
+    {
+      "epoch": 0.11061571911430239,
+      "grad_norm": 7.5460638999938965,
+      "learning_rate": 9.79052086271864e-06,
+      "loss": 2.6856,
+      "step": 424200
+    },
+    {
+      "epoch": 0.11066787174000456,
+      "grad_norm": 7.906571865081787,
+      "learning_rate": 9.790281368754245e-06,
+      "loss": 2.6999,
+      "step": 424400
+    },
+    {
+      "epoch": 0.11072002436570673,
+      "grad_norm": 7.218746662139893,
+      "learning_rate": 9.790041740895958e-06,
+      "loss": 2.7122,
+      "step": 424600
+    },
+    {
+      "epoch": 0.1107721769914089,
+      "grad_norm": 9.075844764709473,
+      "learning_rate": 9.789801979150483e-06,
+      "loss": 2.6593,
+      "step": 424800
+    },
+    {
+      "epoch": 0.11082432961711107,
+      "grad_norm": 8.792651176452637,
+      "learning_rate": 9.78956208352452e-06,
+      "loss": 2.7338,
+      "step": 425000
+    },
+    {
+      "epoch": 0.11087648224281324,
+      "grad_norm": 7.27746057510376,
+      "learning_rate": 9.789322054024774e-06,
+      "loss": 2.6658,
+      "step": 425200
+    },
+    {
+      "epoch": 0.11092863486851541,
+      "grad_norm": 7.638511657714844,
+      "learning_rate": 9.789081890657955e-06,
+      "loss": 2.7166,
+      "step": 425400
+    },
+    {
+      "epoch": 0.11098078749421758,
+      "grad_norm": 7.376978874206543,
+      "learning_rate": 9.788841593430775e-06,
+      "loss": 2.6812,
+      "step": 425600
+    },
+    {
+      "epoch": 0.11103294011991975,
+      "grad_norm": 7.535377025604248,
+      "learning_rate": 9.788601162349949e-06,
+      "loss": 2.6784,
+      "step": 425800
+    },
+    {
+      "epoch": 0.11108509274562192,
+      "grad_norm": 8.068282127380371,
+      "learning_rate": 9.7883605974222e-06,
+      "loss": 2.7089,
+      "step": 426000
+    },
+    {
+      "epoch": 0.11113724537132409,
+      "grad_norm": 8.093073844909668,
+      "learning_rate": 9.788119898654253e-06,
+      "loss": 2.6628,
+      "step": 426200
+    },
+    {
+      "epoch": 0.11118939799702626,
+      "grad_norm": 7.169579029083252,
+      "learning_rate": 9.787879066052833e-06,
+      "loss": 2.7379,
+      "step": 426400
+    },
+    {
+      "epoch": 0.11124155062272843,
+      "grad_norm": 7.598651885986328,
+      "learning_rate": 9.78763809962467e-06,
+      "loss": 2.6831,
+      "step": 426600
+    },
+    {
+      "epoch": 0.1112937032484306,
+      "grad_norm": 8.083666801452637,
+      "learning_rate": 9.787396999376503e-06,
+      "loss": 2.7267,
+      "step": 426800
+    },
+    {
+      "epoch": 0.11134585587413277,
+      "grad_norm": 8.373312950134277,
+      "learning_rate": 9.787155765315071e-06,
+      "loss": 2.6836,
+      "step": 427000
+    },
+    {
+      "epoch": 0.11139800849983493,
+      "grad_norm": 8.161797523498535,
+      "learning_rate": 9.786914397447116e-06,
+      "loss": 2.6782,
+      "step": 427200
+    },
+    {
+      "epoch": 0.1114501611255371,
+      "grad_norm": 8.155803680419922,
+      "learning_rate": 9.786672895779382e-06,
+      "loss": 2.661,
+      "step": 427400
+    },
+    {
+      "epoch": 0.11150231375123927,
+      "grad_norm": 8.83053207397461,
+      "learning_rate": 9.786431260318624e-06,
+      "loss": 2.6674,
+      "step": 427600
+    },
+    {
+      "epoch": 0.11155446637694144,
+      "grad_norm": 6.54095983505249,
+      "learning_rate": 9.78618949107159e-06,
+      "loss": 2.7123,
+      "step": 427800
+    },
+    {
+      "epoch": 0.11160661900264361,
+      "grad_norm": 7.278541088104248,
+      "learning_rate": 9.785947588045044e-06,
+      "loss": 2.6894,
+      "step": 428000
+    },
+    {
+      "epoch": 0.11165877162834578,
+      "grad_norm": 8.509937286376953,
+      "learning_rate": 9.785705551245741e-06,
+      "loss": 2.6652,
+      "step": 428200
+    },
+    {
+      "epoch": 0.11171092425404795,
+      "grad_norm": 8.071086883544922,
+      "learning_rate": 9.785463380680454e-06,
+      "loss": 2.6859,
+      "step": 428400
+    },
+    {
+      "epoch": 0.11176307687975012,
+      "grad_norm": 8.377106666564941,
+      "learning_rate": 9.785221076355944e-06,
+      "loss": 2.7175,
+      "step": 428600
+    },
+    {
+      "epoch": 0.1118152295054523,
+      "grad_norm": 7.088006496429443,
+      "learning_rate": 9.784978638278989e-06,
+      "loss": 2.6663,
+      "step": 428800
+    },
+    {
+      "epoch": 0.11186738213115446,
+      "grad_norm": 8.437390327453613,
+      "learning_rate": 9.784736066456363e-06,
+      "loss": 2.7336,
+      "step": 429000
+    },
+    {
+      "epoch": 0.11191953475685663,
+      "grad_norm": 8.951874732971191,
+      "learning_rate": 9.784493360894849e-06,
+      "loss": 2.732,
+      "step": 429200
+    },
+    {
+      "epoch": 0.1119716873825588,
+      "grad_norm": 7.750514030456543,
+      "learning_rate": 9.784250521601226e-06,
+      "loss": 2.6847,
+      "step": 429400
+    },
+    {
+      "epoch": 0.11202384000826097,
+      "grad_norm": 8.56009292602539,
+      "learning_rate": 9.784007548582287e-06,
+      "loss": 2.6984,
+      "step": 429600
+    },
+    {
+      "epoch": 0.11207599263396315,
+      "grad_norm": 7.063655376434326,
+      "learning_rate": 9.783764441844819e-06,
+      "loss": 2.6662,
+      "step": 429800
+    },
+    {
+      "epoch": 0.11212814525966532,
+      "grad_norm": 7.597879886627197,
+      "learning_rate": 9.78352120139562e-06,
+      "loss": 2.6689,
+      "step": 430000
+    },
+    {
+      "epoch": 0.11218029788536749,
+      "grad_norm": 7.35047721862793,
+      "learning_rate": 9.783277827241486e-06,
+      "loss": 2.7019,
+      "step": 430200
+    },
+    {
+      "epoch": 0.11223245051106966,
+      "grad_norm": 7.905674457550049,
+      "learning_rate": 9.783034319389223e-06,
+      "loss": 2.6588,
+      "step": 430400
+    },
+    {
+      "epoch": 0.11228460313677183,
+      "grad_norm": 8.24709415435791,
+      "learning_rate": 9.782790677845638e-06,
+      "loss": 2.6809,
+      "step": 430600
+    },
+    {
+      "epoch": 0.112336755762474,
+      "grad_norm": 7.832298278808594,
+      "learning_rate": 9.782546902617535e-06,
+      "loss": 2.6982,
+      "step": 430800
+    },
+    {
+      "epoch": 0.11238890838817617,
+      "grad_norm": 7.27885103225708,
+      "learning_rate": 9.782302993711733e-06,
+      "loss": 2.7151,
+      "step": 431000
+    },
+    {
+      "epoch": 0.11244106101387834,
+      "grad_norm": 7.919535160064697,
+      "learning_rate": 9.782058951135047e-06,
+      "loss": 2.6966,
+      "step": 431200
+    },
+    {
+      "epoch": 0.11249321363958051,
+      "grad_norm": 7.770106315612793,
+      "learning_rate": 9.781814774894302e-06,
+      "loss": 2.7011,
+      "step": 431400
+    },
+    {
+      "epoch": 0.11254536626528268,
+      "grad_norm": 8.116305351257324,
+      "learning_rate": 9.781570464996318e-06,
+      "loss": 2.6831,
+      "step": 431600
+    },
+    {
+      "epoch": 0.11259751889098485,
+      "grad_norm": 8.72941780090332,
+      "learning_rate": 9.781326021447928e-06,
+      "loss": 2.6948,
+      "step": 431800
+    },
+    {
+      "epoch": 0.11264967151668702,
+      "grad_norm": 8.299556732177734,
+      "learning_rate": 9.781081444255962e-06,
+      "loss": 2.7151,
+      "step": 432000
+    },
+    {
+      "epoch": 0.11270182414238919,
+      "grad_norm": 7.460847854614258,
+      "learning_rate": 9.780836733427255e-06,
+      "loss": 2.6989,
+      "step": 432200
+    },
+    {
+      "epoch": 0.11275397676809136,
+      "grad_norm": 8.605817794799805,
+      "learning_rate": 9.780591888968652e-06,
+      "loss": 2.6516,
+      "step": 432400
+    },
+    {
+      "epoch": 0.11280612939379353,
+      "grad_norm": 7.617114067077637,
+      "learning_rate": 9.780346910886991e-06,
+      "loss": 2.6607,
+      "step": 432600
+    },
+    {
+      "epoch": 0.1128582820194957,
+      "grad_norm": 7.732776641845703,
+      "learning_rate": 9.78010179918912e-06,
+      "loss": 2.6997,
+      "step": 432800
+    },
+    {
+      "epoch": 0.11291043464519787,
+      "grad_norm": 7.245545864105225,
+      "learning_rate": 9.779856553881897e-06,
+      "loss": 2.6959,
+      "step": 433000
+    },
+    {
+      "epoch": 0.11296258727090003,
+      "grad_norm": 7.128195285797119,
+      "learning_rate": 9.779611174972169e-06,
+      "loss": 2.656,
+      "step": 433200
+    },
+    {
+      "epoch": 0.1130147398966022,
+      "grad_norm": 7.312046051025391,
+      "learning_rate": 9.779365662466798e-06,
+      "loss": 2.6691,
+      "step": 433400
+    },
+    {
+      "epoch": 0.11306689252230437,
+      "grad_norm": 8.948485374450684,
+      "learning_rate": 9.779120016372646e-06,
+      "loss": 2.7061,
+      "step": 433600
+    },
+    {
+      "epoch": 0.11311904514800654,
+      "grad_norm": 7.606996059417725,
+      "learning_rate": 9.77887423669658e-06,
+      "loss": 2.6527,
+      "step": 433800
+    },
+    {
+      "epoch": 0.11317119777370871,
+      "grad_norm": 8.001826286315918,
+      "learning_rate": 9.778628323445467e-06,
+      "loss": 2.663,
+      "step": 434000
+    },
+    {
+      "epoch": 0.11322335039941088,
+      "grad_norm": 8.344403266906738,
+      "learning_rate": 9.778382276626183e-06,
+      "loss": 2.6678,
+      "step": 434200
+    },
+    {
+      "epoch": 0.11327550302511305,
+      "grad_norm": 8.27188491821289,
+      "learning_rate": 9.778136096245605e-06,
+      "loss": 2.7192,
+      "step": 434400
+    },
+    {
+      "epoch": 0.11332765565081522,
+      "grad_norm": 8.35887336730957,
+      "learning_rate": 9.777889782310614e-06,
+      "loss": 2.6627,
+      "step": 434600
+    },
+    {
+      "epoch": 0.11337980827651739,
+      "grad_norm": 7.3902201652526855,
+      "learning_rate": 9.777643334828094e-06,
+      "loss": 2.7158,
+      "step": 434800
+    },
+    {
+      "epoch": 0.11343196090221956,
+      "grad_norm": 8.090744972229004,
+      "learning_rate": 9.777396753804935e-06,
+      "loss": 2.7031,
+      "step": 435000
+    },
+    {
+      "epoch": 0.11348411352792173,
+      "grad_norm": 7.8479084968566895,
+      "learning_rate": 9.777150039248027e-06,
+      "loss": 2.6811,
+      "step": 435200
+    },
+    {
+      "epoch": 0.1135362661536239,
+      "grad_norm": 7.591457366943359,
+      "learning_rate": 9.776903191164267e-06,
+      "loss": 2.6928,
+      "step": 435400
+    },
+    {
+      "epoch": 0.11358841877932607,
+      "grad_norm": 8.277278900146484,
+      "learning_rate": 9.776656209560556e-06,
+      "loss": 2.6798,
+      "step": 435600
+    },
+    {
+      "epoch": 0.11364057140502824,
+      "grad_norm": 8.627765655517578,
+      "learning_rate": 9.776409094443796e-06,
+      "loss": 2.6743,
+      "step": 435800
+    },
+    {
+      "epoch": 0.11369272403073041,
+      "grad_norm": 7.7793288230896,
+      "learning_rate": 9.776161845820894e-06,
+      "loss": 2.6513,
+      "step": 436000
+    },
+    {
+      "epoch": 0.11374487665643258,
+      "grad_norm": 6.800634384155273,
+      "learning_rate": 9.775914463698758e-06,
+      "loss": 2.7216,
+      "step": 436200
+    },
+    {
+      "epoch": 0.11379702928213475,
+      "grad_norm": 7.648603439331055,
+      "learning_rate": 9.77566694808431e-06,
+      "loss": 2.6822,
+      "step": 436400
+    },
+    {
+      "epoch": 0.11384918190783692,
+      "grad_norm": 7.305148124694824,
+      "learning_rate": 9.775419298984463e-06,
+      "loss": 2.6863,
+      "step": 436600
+    },
+    {
+      "epoch": 0.1139013345335391,
+      "grad_norm": 7.8893513679504395,
+      "learning_rate": 9.775171516406142e-06,
+      "loss": 2.6731,
+      "step": 436800
+    },
+    {
+      "epoch": 0.11395348715924127,
+      "grad_norm": 7.8577775955200195,
+      "learning_rate": 9.77492360035627e-06,
+      "loss": 2.6812,
+      "step": 437000
+    },
+    {
+      "epoch": 0.11400563978494344,
+      "grad_norm": 8.009392738342285,
+      "learning_rate": 9.774675550841776e-06,
+      "loss": 2.7047,
+      "step": 437200
+    },
+    {
+      "epoch": 0.1140577924106456,
+      "grad_norm": 7.961512088775635,
+      "learning_rate": 9.774427367869597e-06,
+      "loss": 2.6934,
+      "step": 437400
+    },
+    {
+      "epoch": 0.11410994503634778,
+      "grad_norm": 7.57539701461792,
+      "learning_rate": 9.774179051446667e-06,
+      "loss": 2.667,
+      "step": 437600
+    },
+    {
+      "epoch": 0.11416209766204995,
+      "grad_norm": 7.794543266296387,
+      "learning_rate": 9.773930601579928e-06,
+      "loss": 2.6578,
+      "step": 437800
+    },
+    {
+      "epoch": 0.11421425028775212,
+      "grad_norm": 8.647083282470703,
+      "learning_rate": 9.773682018276325e-06,
+      "loss": 2.7059,
+      "step": 438000
+    },
+    {
+      "epoch": 0.11426640291345429,
+      "grad_norm": 6.988203048706055,
+      "learning_rate": 9.773433301542805e-06,
+      "loss": 2.6848,
+      "step": 438200
+    },
+    {
+      "epoch": 0.11431855553915646,
+      "grad_norm": 7.052835941314697,
+      "learning_rate": 9.77318445138632e-06,
+      "loss": 2.696,
+      "step": 438400
+    },
+    {
+      "epoch": 0.11437070816485863,
+      "grad_norm": 8.016982078552246,
+      "learning_rate": 9.772935467813825e-06,
+      "loss": 2.6304,
+      "step": 438600
+    },
+    {
+      "epoch": 0.1144228607905608,
+      "grad_norm": 8.256826400756836,
+      "learning_rate": 9.77268635083228e-06,
+      "loss": 2.6369,
+      "step": 438800
+    },
+    {
+      "epoch": 0.11447501341626296,
+      "grad_norm": 9.124797821044922,
+      "learning_rate": 9.772437100448649e-06,
+      "loss": 2.6915,
+      "step": 439000
+    },
+    {
+      "epoch": 0.11452716604196513,
+      "grad_norm": 6.831801414489746,
+      "learning_rate": 9.772187716669898e-06,
+      "loss": 2.6886,
+      "step": 439200
+    },
+    {
+      "epoch": 0.1145793186676673,
+      "grad_norm": 8.428510665893555,
+      "learning_rate": 9.771938199503e-06,
+      "loss": 2.7035,
+      "step": 439400
+    },
+    {
+      "epoch": 0.11463147129336947,
+      "grad_norm": 7.8952717781066895,
+      "learning_rate": 9.771688548954923e-06,
+      "loss": 2.7098,
+      "step": 439600
+    },
+    {
+      "epoch": 0.11468362391907164,
+      "grad_norm": 6.776541233062744,
+      "learning_rate": 9.771438765032653e-06,
+      "loss": 2.6655,
+      "step": 439800
+    },
+    {
+      "epoch": 0.11473577654477381,
+      "grad_norm": 7.783594608306885,
+      "learning_rate": 9.771188847743167e-06,
+      "loss": 2.7002,
+      "step": 440000
+    },
+    {
+      "epoch": 0.11478792917047598,
+      "grad_norm": 7.315393924713135,
+      "learning_rate": 9.77093879709345e-06,
+      "loss": 2.6525,
+      "step": 440200
+    },
+    {
+      "epoch": 0.11484008179617815,
+      "grad_norm": 8.225870132446289,
+      "learning_rate": 9.770688613090492e-06,
+      "loss": 2.6535,
+      "step": 440400
+    },
+    {
+      "epoch": 0.11489223442188032,
+      "grad_norm": 7.509810447692871,
+      "learning_rate": 9.770438295741289e-06,
+      "loss": 2.7409,
+      "step": 440600
+    },
+    {
+      "epoch": 0.11494438704758249,
+      "grad_norm": 7.6700592041015625,
+      "learning_rate": 9.770187845052833e-06,
+      "loss": 2.6997,
+      "step": 440800
+    },
+    {
+      "epoch": 0.11499653967328466,
+      "grad_norm": 6.8598527908325195,
+      "learning_rate": 9.769937261032126e-06,
+      "loss": 2.6779,
+      "step": 441000
+    },
+    {
+      "epoch": 0.11504869229898683,
+      "grad_norm": 8.12748908996582,
+      "learning_rate": 9.769686543686174e-06,
+      "loss": 2.717,
+      "step": 441200
+    },
+    {
+      "epoch": 0.115100844924689,
+      "grad_norm": 8.549765586853027,
+      "learning_rate": 9.769435693021982e-06,
+      "loss": 2.663,
+      "step": 441400
+    },
+    {
+      "epoch": 0.11515299755039117,
+      "grad_norm": 6.6760783195495605,
+      "learning_rate": 9.769184709046565e-06,
+      "loss": 2.6786,
+      "step": 441600
+    },
+    {
+      "epoch": 0.11520515017609334,
+      "grad_norm": 7.719881057739258,
+      "learning_rate": 9.768933591766935e-06,
+      "loss": 2.6848,
+      "step": 441800
+    },
+    {
+      "epoch": 0.11525730280179551,
+      "grad_norm": 7.913510799407959,
+      "learning_rate": 9.768682341190114e-06,
+      "loss": 2.7145,
+      "step": 442000
+    },
+    {
+      "epoch": 0.11530945542749768,
+      "grad_norm": 7.543449401855469,
+      "learning_rate": 9.768430957323122e-06,
+      "loss": 2.6923,
+      "step": 442200
+    },
+    {
+      "epoch": 0.11536160805319985,
+      "grad_norm": 8.190464973449707,
+      "learning_rate": 9.768179440172985e-06,
+      "loss": 2.6409,
+      "step": 442400
+    },
+    {
+      "epoch": 0.11541376067890202,
+      "grad_norm": 8.82253360748291,
+      "learning_rate": 9.767927789746736e-06,
+      "loss": 2.6654,
+      "step": 442600
+    },
+    {
+      "epoch": 0.11546591330460419,
+      "grad_norm": 8.999987602233887,
+      "learning_rate": 9.767676006051409e-06,
+      "loss": 2.7038,
+      "step": 442800
+    },
+    {
+      "epoch": 0.11551806593030636,
+      "grad_norm": 8.303157806396484,
+      "learning_rate": 9.767424089094039e-06,
+      "loss": 2.6908,
+      "step": 443000
+    },
+    {
+      "epoch": 0.11557021855600853,
+      "grad_norm": 6.951584339141846,
+      "learning_rate": 9.767172038881669e-06,
+      "loss": 2.623,
+      "step": 443200
+    },
+    {
+      "epoch": 0.1156223711817107,
+      "grad_norm": 8.04922866821289,
+      "learning_rate": 9.766919855421343e-06,
+      "loss": 2.6307,
+      "step": 443400
+    },
+    {
+      "epoch": 0.11567452380741287,
+      "grad_norm": 7.755749225616455,
+      "learning_rate": 9.76666753872011e-06,
+      "loss": 2.662,
+      "step": 443600
+    },
+    {
+      "epoch": 0.11572667643311504,
+      "grad_norm": 9.410798072814941,
+      "learning_rate": 9.766415088785026e-06,
+      "loss": 2.7075,
+      "step": 443800
+    },
+    {
+      "epoch": 0.11577882905881721,
+      "grad_norm": 9.511571884155273,
+      "learning_rate": 9.76616250562314e-06,
+      "loss": 2.6914,
+      "step": 444000
+    },
+    {
+      "epoch": 0.11583098168451939,
+      "grad_norm": 8.075037956237793,
+      "learning_rate": 9.765909789241521e-06,
+      "loss": 2.6813,
+      "step": 444200
+    },
+    {
+      "epoch": 0.11588313431022156,
+      "grad_norm": 7.264787197113037,
+      "learning_rate": 9.765656939647225e-06,
+      "loss": 2.6665,
+      "step": 444400
+    },
+    {
+      "epoch": 0.11593528693592373,
+      "grad_norm": 8.67489242553711,
+      "learning_rate": 9.765403956847323e-06,
+      "loss": 2.7131,
+      "step": 444600
+    },
+    {
+      "epoch": 0.1159874395616259,
+      "grad_norm": 7.648482799530029,
+      "learning_rate": 9.765150840848888e-06,
+      "loss": 2.7141,
+      "step": 444800
+    },
+    {
+      "epoch": 0.11603959218732805,
+      "grad_norm": 7.76444149017334,
+      "learning_rate": 9.764897591658989e-06,
+      "loss": 2.6504,
+      "step": 445000
+    },
+    {
+      "epoch": 0.11609174481303022,
+      "grad_norm": 7.081927299499512,
+      "learning_rate": 9.764644209284711e-06,
+      "loss": 2.7151,
+      "step": 445200
+    },
+    {
+      "epoch": 0.1161438974387324,
+      "grad_norm": 8.81956958770752,
+      "learning_rate": 9.764390693733132e-06,
+      "loss": 2.6997,
+      "step": 445400
+    },
+    {
+      "epoch": 0.11619605006443456,
+      "grad_norm": 7.683972358703613,
+      "learning_rate": 9.764137045011339e-06,
+      "loss": 2.7042,
+      "step": 445600
+    },
+    {
+      "epoch": 0.11624820269013673,
+      "grad_norm": 8.693424224853516,
+      "learning_rate": 9.763883263126423e-06,
+      "loss": 2.7078,
+      "step": 445800
+    },
+    {
+      "epoch": 0.1163003553158389,
+      "grad_norm": 8.919655799865723,
+      "learning_rate": 9.763629348085478e-06,
+      "loss": 2.6838,
+      "step": 446000
+    },
+    {
+      "epoch": 0.11635250794154108,
+      "grad_norm": 9.183130264282227,
+      "learning_rate": 9.763375299895598e-06,
+      "loss": 2.7268,
+      "step": 446200
+    },
+    {
+      "epoch": 0.11640466056724325,
+      "grad_norm": 8.304872512817383,
+      "learning_rate": 9.763121118563888e-06,
+      "loss": 2.6786,
+      "step": 446400
+    },
+    {
+      "epoch": 0.11645681319294542,
+      "grad_norm": 8.370039939880371,
+      "learning_rate": 9.76286680409745e-06,
+      "loss": 2.6958,
+      "step": 446600
+    },
+    {
+      "epoch": 0.11650896581864759,
+      "grad_norm": 9.099044799804688,
+      "learning_rate": 9.762612356503394e-06,
+      "loss": 2.6626,
+      "step": 446800
+    },
+    {
+      "epoch": 0.11656111844434976,
+      "grad_norm": 7.61244535446167,
+      "learning_rate": 9.762357775788829e-06,
+      "loss": 2.6916,
+      "step": 447000
+    },
+    {
+      "epoch": 0.11661327107005193,
+      "grad_norm": 7.555819034576416,
+      "learning_rate": 9.762103061960874e-06,
+      "loss": 2.6282,
+      "step": 447200
+    },
+    {
+      "epoch": 0.1166654236957541,
+      "grad_norm": 8.290244102478027,
+      "learning_rate": 9.761848215026647e-06,
+      "loss": 2.6645,
+      "step": 447400
+    },
+    {
+      "epoch": 0.11671757632145627,
+      "grad_norm": 7.066608905792236,
+      "learning_rate": 9.761593234993273e-06,
+      "loss": 2.6977,
+      "step": 447600
+    },
+    {
+      "epoch": 0.11676972894715844,
+      "grad_norm": 8.447099685668945,
+      "learning_rate": 9.761338121867876e-06,
+      "loss": 2.7276,
+      "step": 447800
+    },
+    {
+      "epoch": 0.11682188157286061,
+      "grad_norm": 8.586709976196289,
+      "learning_rate": 9.76108287565759e-06,
+      "loss": 2.7122,
+      "step": 448000
+    },
+    {
+      "epoch": 0.11687403419856278,
+      "grad_norm": 8.613016128540039,
+      "learning_rate": 9.760827496369547e-06,
+      "loss": 2.6843,
+      "step": 448200
+    },
+    {
+      "epoch": 0.11692618682426495,
+      "grad_norm": 7.8855719566345215,
+      "learning_rate": 9.760571984010887e-06,
+      "loss": 2.6395,
+      "step": 448400
+    },
+    {
+      "epoch": 0.11697833944996712,
+      "grad_norm": 9.135221481323242,
+      "learning_rate": 9.76031633858875e-06,
+      "loss": 2.667,
+      "step": 448600
+    },
+    {
+      "epoch": 0.11703049207566929,
+      "grad_norm": 8.805586814880371,
+      "learning_rate": 9.760060560110283e-06,
+      "loss": 2.6881,
+      "step": 448800
+    },
+    {
+      "epoch": 0.11708264470137146,
+      "grad_norm": 8.323440551757812,
+      "learning_rate": 9.759804648582633e-06,
+      "loss": 2.656,
+      "step": 449000
+    },
+    {
+      "epoch": 0.11713479732707363,
+      "grad_norm": 7.569894790649414,
+      "learning_rate": 9.759548604012959e-06,
+      "loss": 2.6998,
+      "step": 449200
+    },
+    {
+      "epoch": 0.1171869499527758,
+      "grad_norm": 8.15855598449707,
+      "learning_rate": 9.759292426408408e-06,
+      "loss": 2.6691,
+      "step": 449400
+    },
+    {
+      "epoch": 0.11723910257847797,
+      "grad_norm": 7.520715713500977,
+      "learning_rate": 9.75903611577615e-06,
+      "loss": 2.6674,
+      "step": 449600
+    },
+    {
+      "epoch": 0.11729125520418014,
+      "grad_norm": 8.77229118347168,
+      "learning_rate": 9.758779672123344e-06,
+      "loss": 2.6771,
+      "step": 449800
+    },
+    {
+      "epoch": 0.11734340782988231,
+      "grad_norm": 8.861888885498047,
+      "learning_rate": 9.75852309545716e-06,
+      "loss": 2.7158,
+      "step": 450000
+    },
+    {
+      "epoch": 0.11739556045558448,
+      "grad_norm": 7.774789810180664,
+      "learning_rate": 9.758266385784767e-06,
+      "loss": 2.6555,
+      "step": 450200
+    },
+    {
+      "epoch": 0.11744771308128665,
+      "grad_norm": 7.58319091796875,
+      "learning_rate": 9.758009543113343e-06,
+      "loss": 2.6737,
+      "step": 450400
+    },
+    {
+      "epoch": 0.11749986570698882,
+      "grad_norm": 8.081302642822266,
+      "learning_rate": 9.757752567450065e-06,
+      "loss": 2.6793,
+      "step": 450600
+    },
+    {
+      "epoch": 0.11755201833269098,
+      "grad_norm": 7.114253044128418,
+      "learning_rate": 9.757495458802119e-06,
+      "loss": 2.694,
+      "step": 450800
+    },
+    {
+      "epoch": 0.11760417095839315,
+      "grad_norm": 7.961178779602051,
+      "learning_rate": 9.757238217176687e-06,
+      "loss": 2.6656,
+      "step": 451000
+    },
+    {
+      "epoch": 0.11765632358409532,
+      "grad_norm": 7.241010665893555,
+      "learning_rate": 9.756980842580963e-06,
+      "loss": 2.6901,
+      "step": 451200
+    },
+    {
+      "epoch": 0.11770847620979749,
+      "grad_norm": 9.523578643798828,
+      "learning_rate": 9.75672333502214e-06,
+      "loss": 2.6705,
+      "step": 451400
+    },
+    {
+      "epoch": 0.11776062883549966,
+      "grad_norm": 7.9366230964660645,
+      "learning_rate": 9.756465694507413e-06,
+      "loss": 2.6603,
+      "step": 451600
+    },
+    {
+      "epoch": 0.11781278146120183,
+      "grad_norm": 8.917617797851562,
+      "learning_rate": 9.756207921043988e-06,
+      "loss": 2.7168,
+      "step": 451800
+    },
+    {
+      "epoch": 0.117864934086904,
+      "grad_norm": 8.094945907592773,
+      "learning_rate": 9.755950014639065e-06,
+      "loss": 2.6525,
+      "step": 452000
+    },
+    {
+      "epoch": 0.11791708671260617,
+      "grad_norm": 8.251530647277832,
+      "learning_rate": 9.755691975299857e-06,
+      "loss": 2.6882,
+      "step": 452200
+    },
+    {
+      "epoch": 0.11796923933830834,
+      "grad_norm": 7.913506031036377,
+      "learning_rate": 9.755433803033574e-06,
+      "loss": 2.6975,
+      "step": 452400
+    },
+    {
+      "epoch": 0.11802139196401051,
+      "grad_norm": 8.891002655029297,
+      "learning_rate": 9.755175497847434e-06,
+      "loss": 2.6518,
+      "step": 452600
+    },
+    {
+      "epoch": 0.11807354458971268,
+      "grad_norm": 9.051965713500977,
+      "learning_rate": 9.754917059748654e-06,
+      "loss": 2.715,
+      "step": 452800
+    },
+    {
+      "epoch": 0.11812569721541485,
+      "grad_norm": 8.167122840881348,
+      "learning_rate": 9.75465848874446e-06,
+      "loss": 2.7022,
+      "step": 453000
+    },
+    {
+      "epoch": 0.11817784984111702,
+      "grad_norm": 7.507691860198975,
+      "learning_rate": 9.754399784842079e-06,
+      "loss": 2.6796,
+      "step": 453200
+    },
+    {
+      "epoch": 0.1182300024668192,
+      "grad_norm": 7.737783908843994,
+      "learning_rate": 9.754140948048743e-06,
+      "loss": 2.6774,
+      "step": 453400
+    },
+    {
+      "epoch": 0.11828215509252137,
+      "grad_norm": 8.166272163391113,
+      "learning_rate": 9.753881978371685e-06,
+      "loss": 2.7361,
+      "step": 453600
+    },
+    {
+      "epoch": 0.11833430771822354,
+      "grad_norm": 7.494532585144043,
+      "learning_rate": 9.753622875818145e-06,
+      "loss": 2.6566,
+      "step": 453800
+    },
+    {
+      "epoch": 0.1183864603439257,
+      "grad_norm": 8.62021255493164,
+      "learning_rate": 9.753363640395364e-06,
+      "loss": 2.6363,
+      "step": 454000
+    },
+    {
+      "epoch": 0.11843861296962788,
+      "grad_norm": 8.136500358581543,
+      "learning_rate": 9.753104272110587e-06,
+      "loss": 2.7016,
+      "step": 454200
+    },
+    {
+      "epoch": 0.11849076559533005,
+      "grad_norm": 6.9378662109375,
+      "learning_rate": 9.752844770971067e-06,
+      "loss": 2.6903,
+      "step": 454400
+    },
+    {
+      "epoch": 0.11854291822103222,
+      "grad_norm": 9.27441692352295,
+      "learning_rate": 9.752585136984055e-06,
+      "loss": 2.6897,
+      "step": 454600
+    },
+    {
+      "epoch": 0.11859507084673439,
+      "grad_norm": 8.395896911621094,
+      "learning_rate": 9.752325370156808e-06,
+      "loss": 2.6787,
+      "step": 454800
+    },
+    {
+      "epoch": 0.11864722347243656,
+      "grad_norm": 6.82293701171875,
+      "learning_rate": 9.752065470496587e-06,
+      "loss": 2.6816,
+      "step": 455000
+    },
+    {
+      "epoch": 0.11869937609813873,
+      "grad_norm": 9.240446090698242,
+      "learning_rate": 9.751805438010658e-06,
+      "loss": 2.7092,
+      "step": 455200
+    },
+    {
+      "epoch": 0.1187515287238409,
+      "grad_norm": 7.735306262969971,
+      "learning_rate": 9.751545272706287e-06,
+      "loss": 2.6778,
+      "step": 455400
+    },
+    {
+      "epoch": 0.11880368134954307,
+      "grad_norm": 7.703882694244385,
+      "learning_rate": 9.751284974590746e-06,
+      "loss": 2.6471,
+      "step": 455600
+    },
+    {
+      "epoch": 0.11885583397524524,
+      "grad_norm": 7.228545665740967,
+      "learning_rate": 9.751024543671315e-06,
+      "loss": 2.617,
+      "step": 455800
+    },
+    {
+      "epoch": 0.11890798660094741,
+      "grad_norm": 8.354397773742676,
+      "learning_rate": 9.750763979955267e-06,
+      "loss": 2.6919,
+      "step": 456000
+    },
+    {
+      "epoch": 0.11896013922664958,
+      "grad_norm": 7.73373556137085,
+      "learning_rate": 9.750503283449888e-06,
+      "loss": 2.7015,
+      "step": 456200
+    },
+    {
+      "epoch": 0.11901229185235175,
+      "grad_norm": 8.244440078735352,
+      "learning_rate": 9.750242454162465e-06,
+      "loss": 2.6825,
+      "step": 456400
+    },
+    {
+      "epoch": 0.11906444447805391,
+      "grad_norm": 8.330254554748535,
+      "learning_rate": 9.749981492100289e-06,
+      "loss": 2.6659,
+      "step": 456600
+    },
+    {
+      "epoch": 0.11911659710375608,
+      "grad_norm": 8.01742935180664,
+      "learning_rate": 9.749720397270652e-06,
+      "loss": 2.6779,
+      "step": 456800
+    },
+    {
+      "epoch": 0.11916874972945825,
+      "grad_norm": 8.478322982788086,
+      "learning_rate": 9.749459169680855e-06,
+      "loss": 2.6841,
+      "step": 457000
+    },
+    {
+      "epoch": 0.11922090235516042,
+      "grad_norm": 7.7415080070495605,
+      "learning_rate": 9.749197809338195e-06,
+      "loss": 2.6466,
+      "step": 457200
+    },
+    {
+      "epoch": 0.11927305498086259,
+      "grad_norm": 8.464523315429688,
+      "learning_rate": 9.748936316249982e-06,
+      "loss": 2.6419,
+      "step": 457400
+    },
+    {
+      "epoch": 0.11932520760656476,
+      "grad_norm": 8.651251792907715,
+      "learning_rate": 9.748674690423524e-06,
+      "loss": 2.684,
+      "step": 457600
+    },
+    {
+      "epoch": 0.11937736023226693,
+      "grad_norm": 8.235745429992676,
+      "learning_rate": 9.748412931866132e-06,
+      "loss": 2.6694,
+      "step": 457800
+    },
+    {
+      "epoch": 0.1194295128579691,
+      "grad_norm": 7.739285469055176,
+      "learning_rate": 9.748151040585123e-06,
+      "loss": 2.6602,
+      "step": 458000
+    },
+    {
+      "epoch": 0.11948166548367127,
+      "grad_norm": 9.078447341918945,
+      "learning_rate": 9.747889016587819e-06,
+      "loss": 2.6301,
+      "step": 458200
+    },
+    {
+      "epoch": 0.11953381810937344,
+      "grad_norm": 7.439968109130859,
+      "learning_rate": 9.747626859881542e-06,
+      "loss": 2.6789,
+      "step": 458400
+    },
+    {
+      "epoch": 0.11958597073507561,
+      "grad_norm": 8.582900047302246,
+      "learning_rate": 9.74736457047362e-06,
+      "loss": 2.6473,
+      "step": 458600
+    },
+    {
+      "epoch": 0.11963812336077778,
+      "grad_norm": 8.283160209655762,
+      "learning_rate": 9.747102148371384e-06,
+      "loss": 2.6383,
+      "step": 458800
+    },
+    {
+      "epoch": 0.11969027598647995,
+      "grad_norm": 9.031500816345215,
+      "learning_rate": 9.74683959358217e-06,
+      "loss": 2.6908,
+      "step": 459000
+    },
+    {
+      "epoch": 0.11974242861218212,
+      "grad_norm": 7.941998481750488,
+      "learning_rate": 9.746576906113314e-06,
+      "loss": 2.6518,
+      "step": 459200
+    },
+    {
+      "epoch": 0.11979458123788429,
+      "grad_norm": 9.694351196289062,
+      "learning_rate": 9.746314085972162e-06,
+      "loss": 2.6758,
+      "step": 459400
+    },
+    {
+      "epoch": 0.11984673386358646,
+      "grad_norm": 7.908174991607666,
+      "learning_rate": 9.746051133166059e-06,
+      "loss": 2.6652,
+      "step": 459600
+    },
+    {
+      "epoch": 0.11989888648928863,
+      "grad_norm": 7.092363357543945,
+      "learning_rate": 9.745788047702354e-06,
+      "loss": 2.6604,
+      "step": 459800
+    },
+    {
+      "epoch": 0.1199510391149908,
+      "grad_norm": 7.674038887023926,
+      "learning_rate": 9.7455248295884e-06,
+      "loss": 2.6792,
+      "step": 460000
+    },
+    {
+      "epoch": 0.12000319174069297,
+      "grad_norm": 8.855584144592285,
+      "learning_rate": 9.745261478831555e-06,
+      "loss": 2.6561,
+      "step": 460200
+    },
+    {
+      "epoch": 0.12005534436639514,
+      "grad_norm": 7.2290120124816895,
+      "learning_rate": 9.744997995439184e-06,
+      "loss": 2.6789,
+      "step": 460400
+    },
+    {
+      "epoch": 0.12010749699209732,
+      "grad_norm": 8.269558906555176,
+      "learning_rate": 9.744734379418644e-06,
+      "loss": 2.6772,
+      "step": 460600
+    },
+    {
+      "epoch": 0.12015964961779949,
+      "grad_norm": 9.019466400146484,
+      "learning_rate": 9.744470630777307e-06,
+      "loss": 2.685,
+      "step": 460800
+    },
+    {
+      "epoch": 0.12021180224350166,
+      "grad_norm": 8.867697715759277,
+      "learning_rate": 9.744206749522547e-06,
+      "loss": 2.6937,
+      "step": 461000
+    },
+    {
+      "epoch": 0.12026395486920383,
+      "grad_norm": 8.083369255065918,
+      "learning_rate": 9.743942735661738e-06,
+      "loss": 2.6659,
+      "step": 461200
+    },
+    {
+      "epoch": 0.120316107494906,
+      "grad_norm": 8.023956298828125,
+      "learning_rate": 9.743678589202258e-06,
+      "loss": 2.6797,
+      "step": 461400
+    },
+    {
+      "epoch": 0.12036826012060817,
+      "grad_norm": 8.839423179626465,
+      "learning_rate": 9.743414310151494e-06,
+      "loss": 2.7316,
+      "step": 461600
+    },
+    {
+      "epoch": 0.12042041274631034,
+      "grad_norm": 8.470783233642578,
+      "learning_rate": 9.74314989851683e-06,
+      "loss": 2.6571,
+      "step": 461800
+    },
+    {
+      "epoch": 0.12047256537201251,
+      "grad_norm": 8.4376802444458,
+      "learning_rate": 9.742885354305657e-06,
+      "loss": 2.649,
+      "step": 462000
+    },
+    {
+      "epoch": 0.12052471799771468,
+      "grad_norm": 7.8062424659729,
+      "learning_rate": 9.74262067752537e-06,
+      "loss": 2.6703,
+      "step": 462200
+    },
+    {
+      "epoch": 0.12057687062341685,
+      "grad_norm": 8.25600814819336,
+      "learning_rate": 9.742355868183365e-06,
+      "loss": 2.7073,
+      "step": 462400
+    },
+    {
+      "epoch": 0.120629023249119,
+      "grad_norm": 8.327945709228516,
+      "learning_rate": 9.742090926287047e-06,
+      "loss": 2.6969,
+      "step": 462600
+    },
+    {
+      "epoch": 0.12068117587482118,
+      "grad_norm": 8.068275451660156,
+      "learning_rate": 9.741825851843819e-06,
+      "loss": 2.7028,
+      "step": 462800
+    },
+    {
+      "epoch": 0.12073332850052335,
+      "grad_norm": 10.301764488220215,
+      "learning_rate": 9.741560644861091e-06,
+      "loss": 2.6903,
+      "step": 463000
+    },
+    {
+      "epoch": 0.12078548112622552,
+      "grad_norm": 7.792321681976318,
+      "learning_rate": 9.741295305346276e-06,
+      "loss": 2.6443,
+      "step": 463200
+    },
+    {
+      "epoch": 0.12083763375192769,
+      "grad_norm": 8.240435600280762,
+      "learning_rate": 9.74102983330679e-06,
+      "loss": 2.6793,
+      "step": 463400
+    },
+    {
+      "epoch": 0.12088978637762986,
+      "grad_norm": 8.414860725402832,
+      "learning_rate": 9.740764228750053e-06,
+      "loss": 2.6749,
+      "step": 463600
+    },
+    {
+      "epoch": 0.12094193900333203,
+      "grad_norm": 8.185742378234863,
+      "learning_rate": 9.74049849168349e-06,
+      "loss": 2.6609,
+      "step": 463800
+    },
+    {
+      "epoch": 0.1209940916290342,
+      "grad_norm": 8.727797508239746,
+      "learning_rate": 9.74023262211453e-06,
+      "loss": 2.6738,
+      "step": 464000
+    },
+    {
+      "epoch": 0.12104624425473637,
+      "grad_norm": 8.264991760253906,
+      "learning_rate": 9.739966620050601e-06,
+      "loss": 2.6553,
+      "step": 464200
+    },
+    {
+      "epoch": 0.12109839688043854,
+      "grad_norm": 8.572891235351562,
+      "learning_rate": 9.739700485499139e-06,
+      "loss": 2.689,
+      "step": 464400
+    },
+    {
+      "epoch": 0.12115054950614071,
+      "grad_norm": 8.757530212402344,
+      "learning_rate": 9.739434218467583e-06,
+      "loss": 2.6752,
+      "step": 464600
+    },
+    {
+      "epoch": 0.12120270213184288,
+      "grad_norm": 8.559809684753418,
+      "learning_rate": 9.739167818963378e-06,
+      "loss": 2.6809,
+      "step": 464800
+    },
+    {
+      "epoch": 0.12125485475754505,
+      "grad_norm": 8.356423377990723,
+      "learning_rate": 9.738901286993965e-06,
+      "loss": 2.6699,
+      "step": 465000
+    },
+    {
+      "epoch": 0.12130700738324722,
+      "grad_norm": 7.829847812652588,
+      "learning_rate": 9.738634622566799e-06,
+      "loss": 2.6739,
+      "step": 465200
+    },
+    {
+      "epoch": 0.12135916000894939,
+      "grad_norm": 8.362701416015625,
+      "learning_rate": 9.73836782568933e-06,
+      "loss": 2.6874,
+      "step": 465400
+    },
+    {
+      "epoch": 0.12141131263465156,
+      "grad_norm": 9.722343444824219,
+      "learning_rate": 9.738100896369019e-06,
+      "loss": 2.6827,
+      "step": 465600
+    },
+    {
+      "epoch": 0.12146346526035373,
+      "grad_norm": 8.5645112991333,
+      "learning_rate": 9.737833834613322e-06,
+      "loss": 2.6557,
+      "step": 465800
+    },
+    {
+      "epoch": 0.1215156178860559,
+      "grad_norm": 8.088398933410645,
+      "learning_rate": 9.737566640429707e-06,
+      "loss": 2.6685,
+      "step": 466000
+    },
+    {
+      "epoch": 0.12156777051175807,
+      "grad_norm": 8.453975677490234,
+      "learning_rate": 9.737299313825644e-06,
+      "loss": 2.6338,
+      "step": 466200
+    },
+    {
+      "epoch": 0.12161992313746024,
+      "grad_norm": 7.445858478546143,
+      "learning_rate": 9.737031854808601e-06,
+      "loss": 2.6752,
+      "step": 466400
+    },
+    {
+      "epoch": 0.12167207576316241,
+      "grad_norm": 9.255156517028809,
+      "learning_rate": 9.736764263386057e-06,
+      "loss": 2.7129,
+      "step": 466600
+    },
+    {
+      "epoch": 0.12172422838886458,
+      "grad_norm": 9.095346450805664,
+      "learning_rate": 9.736496539565488e-06,
+      "loss": 2.6596,
+      "step": 466800
+    },
+    {
+      "epoch": 0.12177638101456675,
+      "grad_norm": 8.963985443115234,
+      "learning_rate": 9.73622868335438e-06,
+      "loss": 2.6857,
+      "step": 467000
+    },
+    {
+      "epoch": 0.12182853364026892,
+      "grad_norm": 9.29616641998291,
+      "learning_rate": 9.73596069476022e-06,
+      "loss": 2.6297,
+      "step": 467200
+    },
+    {
+      "epoch": 0.1218806862659711,
+      "grad_norm": 7.7633795738220215,
+      "learning_rate": 9.735692573790498e-06,
+      "loss": 2.6467,
+      "step": 467400
+    },
+    {
+      "epoch": 0.12193283889167326,
+      "grad_norm": 9.138328552246094,
+      "learning_rate": 9.735424320452707e-06,
+      "loss": 2.7386,
+      "step": 467600
+    },
+    {
+      "epoch": 0.12198499151737544,
+      "grad_norm": 7.824506759643555,
+      "learning_rate": 9.735155934754347e-06,
+      "loss": 2.6447,
+      "step": 467800
+    },
+    {
+      "epoch": 0.1220371441430776,
+      "grad_norm": 9.13517951965332,
+      "learning_rate": 9.734887416702919e-06,
+      "loss": 2.6749,
+      "step": 468000
+    },
+    {
+      "epoch": 0.12208929676877978,
+      "grad_norm": 8.389497756958008,
+      "learning_rate": 9.734618766305928e-06,
+      "loss": 2.6867,
+      "step": 468200
+    },
+    {
+      "epoch": 0.12214144939448193,
+      "grad_norm": 8.484880447387695,
+      "learning_rate": 9.734349983570882e-06,
+      "loss": 2.6819,
+      "step": 468400
+    },
+    {
+      "epoch": 0.1221936020201841,
+      "grad_norm": 8.580875396728516,
+      "learning_rate": 9.734081068505296e-06,
+      "loss": 2.6606,
+      "step": 468600
+    },
+    {
+      "epoch": 0.12224575464588627,
+      "grad_norm": 8.64186954498291,
+      "learning_rate": 9.733812021116687e-06,
+      "loss": 2.711,
+      "step": 468800
+    },
+    {
+      "epoch": 0.12229790727158844,
+      "grad_norm": 8.544197082519531,
+      "learning_rate": 9.733542841412571e-06,
+      "loss": 2.6684,
+      "step": 469000
+    },
+    {
+      "epoch": 0.12235005989729061,
+      "grad_norm": 7.277874946594238,
+      "learning_rate": 9.733273529400476e-06,
+      "loss": 2.6726,
+      "step": 469200
+    },
+    {
+      "epoch": 0.12240221252299278,
+      "grad_norm": 7.846140384674072,
+      "learning_rate": 9.733004085087926e-06,
+      "loss": 2.6603,
+      "step": 469400
+    },
+    {
+      "epoch": 0.12245436514869495,
+      "grad_norm": 7.558496475219727,
+      "learning_rate": 9.732734508482458e-06,
+      "loss": 2.6833,
+      "step": 469600
+    },
+    {
+      "epoch": 0.12250651777439713,
+      "grad_norm": 8.439860343933105,
+      "learning_rate": 9.732464799591603e-06,
+      "loss": 2.6507,
+      "step": 469800
+    },
+    {
+      "epoch": 0.1225586704000993,
+      "grad_norm": 9.658734321594238,
+      "learning_rate": 9.732194958422898e-06,
+      "loss": 2.7056,
+      "step": 470000
+    },
+    {
+      "epoch": 0.12261082302580147,
+      "grad_norm": 7.978820323944092,
+      "learning_rate": 9.731924984983889e-06,
+      "loss": 2.704,
+      "step": 470200
+    },
+    {
+      "epoch": 0.12266297565150364,
+      "grad_norm": 7.804432392120361,
+      "learning_rate": 9.73165487928212e-06,
+      "loss": 2.6903,
+      "step": 470400
+    },
+    {
+      "epoch": 0.1227151282772058,
+      "grad_norm": 8.81722640991211,
+      "learning_rate": 9.73138464132514e-06,
+      "loss": 2.7032,
+      "step": 470600
+    },
+    {
+      "epoch": 0.12276728090290798,
+      "grad_norm": 9.00986385345459,
+      "learning_rate": 9.731114271120505e-06,
+      "loss": 2.6696,
+      "step": 470800
+    },
+    {
+      "epoch": 0.12281943352861015,
+      "grad_norm": 8.249679565429688,
+      "learning_rate": 9.730843768675772e-06,
+      "loss": 2.6984,
+      "step": 471000
+    },
+    {
+      "epoch": 0.12287158615431232,
+      "grad_norm": 8.557232856750488,
+      "learning_rate": 9.7305731339985e-06,
+      "loss": 2.6371,
+      "step": 471200
+    },
+    {
+      "epoch": 0.12292373878001449,
+      "grad_norm": 8.401301383972168,
+      "learning_rate": 9.730302367096254e-06,
+      "loss": 2.6919,
+      "step": 471400
+    },
+    {
+      "epoch": 0.12297589140571666,
+      "grad_norm": 7.651167392730713,
+      "learning_rate": 9.730031467976601e-06,
+      "loss": 2.7055,
+      "step": 471600
+    },
+    {
+      "epoch": 0.12302804403141883,
+      "grad_norm": 8.762410163879395,
+      "learning_rate": 9.729760436647118e-06,
+      "loss": 2.6395,
+      "step": 471800
+    },
+    {
+      "epoch": 0.123080196657121,
+      "grad_norm": 8.571608543395996,
+      "learning_rate": 9.729489273115376e-06,
+      "loss": 2.651,
+      "step": 472000
+    },
+    {
+      "epoch": 0.12313234928282317,
+      "grad_norm": 8.2172269821167,
+      "learning_rate": 9.729217977388954e-06,
+      "loss": 2.6385,
+      "step": 472200
+    },
+    {
+      "epoch": 0.12318450190852534,
+      "grad_norm": 9.237062454223633,
+      "learning_rate": 9.728946549475438e-06,
+      "loss": 2.718,
+      "step": 472400
+    },
+    {
+      "epoch": 0.12323665453422751,
+      "grad_norm": 8.741727828979492,
+      "learning_rate": 9.728674989382412e-06,
+      "loss": 2.6775,
+      "step": 472600
+    },
+    {
+      "epoch": 0.12328880715992968,
+      "grad_norm": 7.985013008117676,
+      "learning_rate": 9.728403297117469e-06,
+      "loss": 2.6703,
+      "step": 472800
+    },
+    {
+      "epoch": 0.12334095978563185,
+      "grad_norm": 8.41588020324707,
+      "learning_rate": 9.7281314726882e-06,
+      "loss": 2.6602,
+      "step": 473000
+    },
+    {
+      "epoch": 0.12339311241133402,
+      "grad_norm": 10.485689163208008,
+      "learning_rate": 9.727859516102208e-06,
+      "loss": 2.6371,
+      "step": 473200
+    },
+    {
+      "epoch": 0.12344526503703619,
+      "grad_norm": 8.508200645446777,
+      "learning_rate": 9.727587427367088e-06,
+      "loss": 2.7213,
+      "step": 473400
+    },
+    {
+      "epoch": 0.12349741766273836,
+      "grad_norm": 7.656301498413086,
+      "learning_rate": 9.72731520649045e-06,
+      "loss": 2.6225,
+      "step": 473600
+    },
+    {
+      "epoch": 0.12354957028844053,
+      "grad_norm": 7.055719375610352,
+      "learning_rate": 9.7270428534799e-06,
+      "loss": 2.6489,
+      "step": 473800
+    },
+    {
+      "epoch": 0.1236017229141427,
+      "grad_norm": 7.498048782348633,
+      "learning_rate": 9.72677036834305e-06,
+      "loss": 2.679,
+      "step": 474000
+    },
+    {
+      "epoch": 0.12365387553984487,
+      "grad_norm": 8.528254508972168,
+      "learning_rate": 9.726497751087522e-06,
+      "loss": 2.6907,
+      "step": 474200
+    },
+    {
+      "epoch": 0.12370602816554703,
+      "grad_norm": 8.69704818725586,
+      "learning_rate": 9.72622500172093e-06,
+      "loss": 2.666,
+      "step": 474400
+    },
+    {
+      "epoch": 0.1237581807912492,
+      "grad_norm": 8.989028930664062,
+      "learning_rate": 9.7259521202509e-06,
+      "loss": 2.6891,
+      "step": 474600
+    },
+    {
+      "epoch": 0.12381033341695137,
+      "grad_norm": 8.570834159851074,
+      "learning_rate": 9.725679106685058e-06,
+      "loss": 2.6665,
+      "step": 474800
+    },
+    {
+      "epoch": 0.12386248604265354,
+      "grad_norm": 8.544633865356445,
+      "learning_rate": 9.725405961031038e-06,
+      "loss": 2.6161,
+      "step": 475000
+    },
+    {
+      "epoch": 0.12391463866835571,
+      "grad_norm": 8.439756393432617,
+      "learning_rate": 9.725132683296471e-06,
+      "loss": 2.6956,
+      "step": 475200
+    },
+    {
+      "epoch": 0.12396679129405788,
+      "grad_norm": 8.925586700439453,
+      "learning_rate": 9.724859273488999e-06,
+      "loss": 2.6371,
+      "step": 475400
+    },
+    {
+      "epoch": 0.12401894391976005,
+      "grad_norm": 8.36025333404541,
+      "learning_rate": 9.72458573161626e-06,
+      "loss": 2.6451,
+      "step": 475600
+    },
+    {
+      "epoch": 0.12407109654546222,
+      "grad_norm": 8.255962371826172,
+      "learning_rate": 9.724312057685902e-06,
+      "loss": 2.6251,
+      "step": 475800
+    },
+    {
+      "epoch": 0.1241232491711644,
+      "grad_norm": 9.104443550109863,
+      "learning_rate": 9.724038251705577e-06,
+      "loss": 2.6458,
+      "step": 476000
+    },
+    {
+      "epoch": 0.12417540179686656,
+      "grad_norm": 8.984846115112305,
+      "learning_rate": 9.723764313682936e-06,
+      "loss": 2.6515,
+      "step": 476200
+    },
+    {
+      "epoch": 0.12422755442256873,
+      "grad_norm": 8.620625495910645,
+      "learning_rate": 9.723490243625634e-06,
+      "loss": 2.6752,
+      "step": 476400
+    },
+    {
+      "epoch": 0.1242797070482709,
+      "grad_norm": 8.312873840332031,
+      "learning_rate": 9.723216041541334e-06,
+      "loss": 2.6998,
+      "step": 476600
+    },
+    {
+      "epoch": 0.12433185967397307,
+      "grad_norm": 8.567025184631348,
+      "learning_rate": 9.7229417074377e-06,
+      "loss": 2.6366,
+      "step": 476800
+    },
+    {
+      "epoch": 0.12438401229967525,
+      "grad_norm": 9.091941833496094,
+      "learning_rate": 9.7226672413224e-06,
+      "loss": 2.6761,
+      "step": 477000
+    },
+    {
+      "epoch": 0.12443616492537742,
+      "grad_norm": 8.620004653930664,
+      "learning_rate": 9.722392643203103e-06,
+      "loss": 2.6256,
+      "step": 477200
+    },
+    {
+      "epoch": 0.12448831755107959,
+      "grad_norm": 8.094307899475098,
+      "learning_rate": 9.722117913087488e-06,
+      "loss": 2.6586,
+      "step": 477400
+    },
+    {
+      "epoch": 0.12454047017678176,
+      "grad_norm": 9.369007110595703,
+      "learning_rate": 9.72184305098323e-06,
+      "loss": 2.6223,
+      "step": 477600
+    },
+    {
+      "epoch": 0.12459262280248393,
+      "grad_norm": 9.586237907409668,
+      "learning_rate": 9.721568056898018e-06,
+      "loss": 2.6983,
+      "step": 477800
+    },
+    {
+      "epoch": 0.1246447754281861,
+      "grad_norm": 7.8790106773376465,
+      "learning_rate": 9.721292930839532e-06,
+      "loss": 2.6527,
+      "step": 478000
+    },
+    {
+      "epoch": 0.12469692805388827,
+      "grad_norm": 9.248851776123047,
+      "learning_rate": 9.721017672815467e-06,
+      "loss": 2.6589,
+      "step": 478200
+    },
+    {
+      "epoch": 0.12474908067959044,
+      "grad_norm": 9.875216484069824,
+      "learning_rate": 9.720742282833513e-06,
+      "loss": 2.6659,
+      "step": 478400
+    },
+    {
+      "epoch": 0.12480123330529261,
+      "grad_norm": 9.6345796585083,
+      "learning_rate": 9.720466760901368e-06,
+      "loss": 2.676,
+      "step": 478600
+    },
+    {
+      "epoch": 0.12485338593099478,
+      "grad_norm": 8.288617134094238,
+      "learning_rate": 9.720191107026735e-06,
+      "loss": 2.661,
+      "step": 478800
+    },
+    {
+      "epoch": 0.12490553855669695,
+      "grad_norm": 9.437127113342285,
+      "learning_rate": 9.719915321217319e-06,
+      "loss": 2.6597,
+      "step": 479000
+    },
+    {
+      "epoch": 0.12495769118239912,
+      "grad_norm": 8.890555381774902,
+      "learning_rate": 9.719639403480827e-06,
+      "loss": 2.6872,
+      "step": 479200
+    },
+    {
+      "epoch": 0.1250098438081013,
+      "grad_norm": 8.575922966003418,
+      "learning_rate": 9.71936335382497e-06,
+      "loss": 2.6705,
+      "step": 479400
+    },
+    {
+      "epoch": 0.12506199643380345,
+      "grad_norm": 8.43736457824707,
+      "learning_rate": 9.719087172257468e-06,
+      "loss": 2.6723,
+      "step": 479600
+    },
+    {
+      "epoch": 0.12511414905950563,
+      "grad_norm": 7.534491539001465,
+      "learning_rate": 9.71881085878604e-06,
+      "loss": 2.6203,
+      "step": 479800
+    },
+    {
+      "epoch": 0.1251663016852078,
+      "grad_norm": 9.604147911071777,
+      "learning_rate": 9.718534413418404e-06,
+      "loss": 2.6615,
+      "step": 480000
+    },
+    {
+      "epoch": 0.12521845431090997,
+      "grad_norm": 8.846263885498047,
+      "learning_rate": 9.71825783616229e-06,
+      "loss": 2.6699,
+      "step": 480200
+    },
+    {
+      "epoch": 0.12527060693661213,
+      "grad_norm": 8.537163734436035,
+      "learning_rate": 9.71798112702543e-06,
+      "loss": 2.6123,
+      "step": 480400
+    },
+    {
+      "epoch": 0.1253227595623143,
+      "grad_norm": 9.015103340148926,
+      "learning_rate": 9.71770428601556e-06,
+      "loss": 2.6473,
+      "step": 480600
+    },
+    {
+      "epoch": 0.12537491218801647,
+      "grad_norm": 8.286890029907227,
+      "learning_rate": 9.717427313140412e-06,
+      "loss": 2.6601,
+      "step": 480800
+    },
+    {
+      "epoch": 0.12542706481371865,
+      "grad_norm": 8.386801719665527,
+      "learning_rate": 9.717150208407733e-06,
+      "loss": 2.6411,
+      "step": 481000
+    },
+    {
+      "epoch": 0.1254792174394208,
+      "grad_norm": 9.405022621154785,
+      "learning_rate": 9.716872971825265e-06,
+      "loss": 2.6861,
+      "step": 481200
+    },
+    {
+      "epoch": 0.125531370065123,
+      "grad_norm": 9.213386535644531,
+      "learning_rate": 9.71659560340076e-06,
+      "loss": 2.6431,
+      "step": 481400
+    },
+    {
+      "epoch": 0.12558352269082515,
+      "grad_norm": 9.34145736694336,
+      "learning_rate": 9.71631810314197e-06,
+      "loss": 2.6649,
+      "step": 481600
+    },
+    {
+      "epoch": 0.12563567531652733,
+      "grad_norm": 9.355376243591309,
+      "learning_rate": 9.71604047105665e-06,
+      "loss": 2.6769,
+      "step": 481800
+    },
+    {
+      "epoch": 0.1256878279422295,
+      "grad_norm": 9.561544418334961,
+      "learning_rate": 9.715762707152561e-06,
+      "loss": 2.6069,
+      "step": 482000
+    },
+    {
+      "epoch": 0.12573998056793168,
+      "grad_norm": 9.205405235290527,
+      "learning_rate": 9.715484811437468e-06,
+      "loss": 2.6307,
+      "step": 482200
+    },
+    {
+      "epoch": 0.12579213319363383,
+      "grad_norm": 9.112932205200195,
+      "learning_rate": 9.715206783919136e-06,
+      "loss": 2.6701,
+      "step": 482400
+    },
+    {
+      "epoch": 0.12584428581933602,
+      "grad_norm": 8.555209159851074,
+      "learning_rate": 9.714928624605337e-06,
+      "loss": 2.6885,
+      "step": 482600
+    },
+    {
+      "epoch": 0.12589643844503817,
+      "grad_norm": 8.725874900817871,
+      "learning_rate": 9.714650333503848e-06,
+      "loss": 2.663,
+      "step": 482800
+    },
+    {
+      "epoch": 0.12594859107074036,
+      "grad_norm": 8.564589500427246,
+      "learning_rate": 9.714371910622445e-06,
+      "loss": 2.6496,
+      "step": 483000
+    },
+    {
+      "epoch": 0.1260007436964425,
+      "grad_norm": 9.859867095947266,
+      "learning_rate": 9.714093355968913e-06,
+      "loss": 2.6103,
+      "step": 483200
+    },
+    {
+      "epoch": 0.12605289632214467,
+      "grad_norm": 8.561869621276855,
+      "learning_rate": 9.713814669551034e-06,
+      "loss": 2.6476,
+      "step": 483400
+    },
+    {
+      "epoch": 0.12610504894784685,
+      "grad_norm": 9.05595874786377,
+      "learning_rate": 9.7135358513766e-06,
+      "loss": 2.6854,
+      "step": 483600
+    },
+    {
+      "epoch": 0.126157201573549,
+      "grad_norm": 8.561729431152344,
+      "learning_rate": 9.713256901453405e-06,
+      "loss": 2.6784,
+      "step": 483800
+    },
+    {
+      "epoch": 0.1262093541992512,
+      "grad_norm": 9.193467140197754,
+      "learning_rate": 9.712977819789244e-06,
+      "loss": 2.6623,
+      "step": 484000
+    },
+    {
+      "epoch": 0.12626150682495335,
+      "grad_norm": 8.660527229309082,
+      "learning_rate": 9.71269860639192e-06,
+      "loss": 2.682,
+      "step": 484200
+    },
+    {
+      "epoch": 0.12631365945065554,
+      "grad_norm": 8.973142623901367,
+      "learning_rate": 9.712419261269234e-06,
+      "loss": 2.6837,
+      "step": 484400
+    },
+    {
+      "epoch": 0.1263658120763577,
+      "grad_norm": 9.378042221069336,
+      "learning_rate": 9.712139784428998e-06,
+      "loss": 2.6662,
+      "step": 484600
+    },
+    {
+      "epoch": 0.12641796470205988,
+      "grad_norm": 7.6433424949646,
+      "learning_rate": 9.71186017587902e-06,
+      "loss": 2.6865,
+      "step": 484800
+    },
+    {
+      "epoch": 0.12647011732776203,
+      "grad_norm": 8.966972351074219,
+      "learning_rate": 9.711580435627118e-06,
+      "loss": 2.671,
+      "step": 485000
+    },
+    {
+      "epoch": 0.12652226995346422,
+      "grad_norm": 9.110841751098633,
+      "learning_rate": 9.71130056368111e-06,
+      "loss": 2.6932,
+      "step": 485200
+    },
+    {
+      "epoch": 0.12657442257916637,
+      "grad_norm": 8.476582527160645,
+      "learning_rate": 9.711020560048819e-06,
+      "loss": 2.6634,
+      "step": 485400
+    },
+    {
+      "epoch": 0.12662657520486856,
+      "grad_norm": 8.898265838623047,
+      "learning_rate": 9.710740424738072e-06,
+      "loss": 2.6732,
+      "step": 485600
+    },
+    {
+      "epoch": 0.12667872783057071,
+      "grad_norm": 9.571182250976562,
+      "learning_rate": 9.710460157756698e-06,
+      "loss": 2.6777,
+      "step": 485800
+    },
+    {
+      "epoch": 0.1267308804562729,
+      "grad_norm": 9.099579811096191,
+      "learning_rate": 9.710179759112531e-06,
+      "loss": 2.6837,
+      "step": 486000
+    },
+    {
+      "epoch": 0.12678303308197506,
+      "grad_norm": 8.9638671875,
+      "learning_rate": 9.709899228813408e-06,
+      "loss": 2.6928,
+      "step": 486200
+    },
+    {
+      "epoch": 0.12683518570767724,
+      "grad_norm": 9.293763160705566,
+      "learning_rate": 9.709618566867173e-06,
+      "loss": 2.6583,
+      "step": 486400
+    },
+    {
+      "epoch": 0.1268873383333794,
+      "grad_norm": 7.878268241882324,
+      "learning_rate": 9.709337773281666e-06,
+      "loss": 2.6466,
+      "step": 486600
+    },
+    {
+      "epoch": 0.12693949095908158,
+      "grad_norm": 8.64033317565918,
+      "learning_rate": 9.70905684806474e-06,
+      "loss": 2.6509,
+      "step": 486800
+    },
+    {
+      "epoch": 0.12699164358478374,
+      "grad_norm": 9.131351470947266,
+      "learning_rate": 9.708775791224246e-06,
+      "loss": 2.6382,
+      "step": 487000
+    },
+    {
+      "epoch": 0.12704379621048592,
+      "grad_norm": 8.987448692321777,
+      "learning_rate": 9.708494602768038e-06,
+      "loss": 2.6429,
+      "step": 487200
+    },
+    {
+      "epoch": 0.12709594883618808,
+      "grad_norm": 9.501585960388184,
+      "learning_rate": 9.708213282703975e-06,
+      "loss": 2.6805,
+      "step": 487400
+    },
+    {
+      "epoch": 0.12714810146189026,
+      "grad_norm": 9.123064994812012,
+      "learning_rate": 9.707931831039926e-06,
+      "loss": 2.6948,
+      "step": 487600
+    },
+    {
+      "epoch": 0.12720025408759242,
+      "grad_norm": 9.532994270324707,
+      "learning_rate": 9.70765024778375e-06,
+      "loss": 2.6968,
+      "step": 487800
+    },
+    {
+      "epoch": 0.1272524067132946,
+      "grad_norm": 9.438848495483398,
+      "learning_rate": 9.707368532943324e-06,
+      "loss": 2.669,
+      "step": 488000
+    },
+    {
+      "epoch": 0.12730455933899676,
+      "grad_norm": 9.121577262878418,
+      "learning_rate": 9.707086686526518e-06,
+      "loss": 2.6328,
+      "step": 488200
+    },
+    {
+      "epoch": 0.12735671196469894,
+      "grad_norm": 9.110394477844238,
+      "learning_rate": 9.706804708541211e-06,
+      "loss": 2.6712,
+      "step": 488400
+    },
+    {
+      "epoch": 0.1274088645904011,
+      "grad_norm": 7.500752925872803,
+      "learning_rate": 9.706522598995288e-06,
+      "loss": 2.6645,
+      "step": 488600
+    },
+    {
+      "epoch": 0.12746101721610328,
+      "grad_norm": 9.652029037475586,
+      "learning_rate": 9.706240357896631e-06,
+      "loss": 2.6274,
+      "step": 488800
+    },
+    {
+      "epoch": 0.12751316984180544,
+      "grad_norm": 9.1790189743042,
+      "learning_rate": 9.705957985253128e-06,
+      "loss": 2.6722,
+      "step": 489000
+    },
+    {
+      "epoch": 0.1275653224675076,
+      "grad_norm": 10.447507858276367,
+      "learning_rate": 9.705675481072673e-06,
+      "loss": 2.6685,
+      "step": 489200
+    },
+    {
+      "epoch": 0.12761747509320978,
+      "grad_norm": 9.254826545715332,
+      "learning_rate": 9.705392845363164e-06,
+      "loss": 2.6453,
+      "step": 489400
+    },
+    {
+      "epoch": 0.12766962771891194,
+      "grad_norm": 8.488991737365723,
+      "learning_rate": 9.705110078132498e-06,
+      "loss": 2.6604,
+      "step": 489600
+    },
+    {
+      "epoch": 0.12772178034461412,
+      "grad_norm": 8.976917266845703,
+      "learning_rate": 9.704827179388581e-06,
+      "loss": 2.6436,
+      "step": 489800
+    },
+    {
+      "epoch": 0.12777393297031628,
+      "grad_norm": 9.096659660339355,
+      "learning_rate": 9.704544149139319e-06,
+      "loss": 2.6794,
+      "step": 490000
+    },
+    {
+      "epoch": 0.12782608559601846,
+      "grad_norm": 9.129677772521973,
+      "learning_rate": 9.704260987392624e-06,
+      "loss": 2.6928,
+      "step": 490200
+    },
+    {
+      "epoch": 0.12787823822172062,
+      "grad_norm": 8.751590728759766,
+      "learning_rate": 9.70397769415641e-06,
+      "loss": 2.6381,
+      "step": 490400
+    },
+    {
+      "epoch": 0.1279303908474228,
+      "grad_norm": 9.134084701538086,
+      "learning_rate": 9.703694269438596e-06,
+      "loss": 2.6475,
+      "step": 490600
+    },
+    {
+      "epoch": 0.12798254347312496,
+      "grad_norm": 9.106485366821289,
+      "learning_rate": 9.703410713247102e-06,
+      "loss": 2.6657,
+      "step": 490800
+    },
+    {
+      "epoch": 0.12803469609882714,
+      "grad_norm": 8.137367248535156,
+      "learning_rate": 9.703127025589856e-06,
+      "loss": 2.6392,
+      "step": 491000
+    },
+    {
+      "epoch": 0.1280868487245293,
+      "grad_norm": 8.145878791809082,
+      "learning_rate": 9.702843206474788e-06,
+      "loss": 2.6477,
+      "step": 491200
+    },
+    {
+      "epoch": 0.12813900135023149,
+      "grad_norm": 8.465982437133789,
+      "learning_rate": 9.70255925590983e-06,
+      "loss": 2.6539,
+      "step": 491400
+    },
+    {
+      "epoch": 0.12819115397593364,
+      "grad_norm": 8.704390525817871,
+      "learning_rate": 9.702275173902915e-06,
+      "loss": 2.656,
+      "step": 491600
+    },
+    {
+      "epoch": 0.12824330660163583,
+      "grad_norm": 9.234391212463379,
+      "learning_rate": 9.70199096046199e-06,
+      "loss": 2.6496,
+      "step": 491800
+    },
+    {
+      "epoch": 0.12829545922733798,
+      "grad_norm": 8.896957397460938,
+      "learning_rate": 9.701706615594996e-06,
+      "loss": 2.5983,
+      "step": 492000
+    },
+    {
+      "epoch": 0.12834761185304017,
+      "grad_norm": 8.498276710510254,
+      "learning_rate": 9.70142213930988e-06,
+      "loss": 2.6514,
+      "step": 492200
+    },
+    {
+      "epoch": 0.12839976447874232,
+      "grad_norm": 8.999076843261719,
+      "learning_rate": 9.701137531614595e-06,
+      "loss": 2.7108,
+      "step": 492400
+    },
+    {
+      "epoch": 0.1284519171044445,
+      "grad_norm": 8.710136413574219,
+      "learning_rate": 9.700852792517095e-06,
+      "loss": 2.6373,
+      "step": 492600
+    },
+    {
+      "epoch": 0.12850406973014666,
+      "grad_norm": 8.627429008483887,
+      "learning_rate": 9.70056792202534e-06,
+      "loss": 2.6688,
+      "step": 492800
+    },
+    {
+      "epoch": 0.12855622235584885,
+      "grad_norm": 8.266898155212402,
+      "learning_rate": 9.700282920147288e-06,
+      "loss": 2.6819,
+      "step": 493000
+    },
+    {
+      "epoch": 0.128608374981551,
+      "grad_norm": 8.928200721740723,
+      "learning_rate": 9.699997786890911e-06,
+      "loss": 2.6685,
+      "step": 493200
+    },
+    {
+      "epoch": 0.1286605276072532,
+      "grad_norm": 9.263498306274414,
+      "learning_rate": 9.699712522264179e-06,
+      "loss": 2.6548,
+      "step": 493400
+    },
+    {
+      "epoch": 0.12871268023295535,
+      "grad_norm": 8.80411434173584,
+      "learning_rate": 9.69942712627506e-06,
+      "loss": 2.6194,
+      "step": 493600
+    },
+    {
+      "epoch": 0.12876483285865753,
+      "grad_norm": 9.677428245544434,
+      "learning_rate": 9.699141598931536e-06,
+      "loss": 2.6491,
+      "step": 493800
+    },
+    {
+      "epoch": 0.1288169854843597,
+      "grad_norm": 8.916584968566895,
+      "learning_rate": 9.698855940241584e-06,
+      "loss": 2.6393,
+      "step": 494000
+    },
+    {
+      "epoch": 0.12886913811006187,
+      "grad_norm": 8.620122909545898,
+      "learning_rate": 9.698570150213192e-06,
+      "loss": 2.7015,
+      "step": 494200
+    },
+    {
+      "epoch": 0.12892129073576403,
+      "grad_norm": 9.354145050048828,
+      "learning_rate": 9.698284228854346e-06,
+      "loss": 2.6735,
+      "step": 494400
+    },
+    {
+      "epoch": 0.1289734433614662,
+      "grad_norm": 9.221833229064941,
+      "learning_rate": 9.69799817617304e-06,
+      "loss": 2.639,
+      "step": 494600
+    },
+    {
+      "epoch": 0.12902559598716837,
+      "grad_norm": 8.699933052062988,
+      "learning_rate": 9.697711992177266e-06,
+      "loss": 2.6933,
+      "step": 494800
+    },
+    {
+      "epoch": 0.12907774861287052,
+      "grad_norm": 8.266427993774414,
+      "learning_rate": 9.697425676875026e-06,
+      "loss": 2.6265,
+      "step": 495000
+    },
+    {
+      "epoch": 0.1291299012385727,
+      "grad_norm": 9.16121768951416,
+      "learning_rate": 9.697139230274322e-06,
+      "loss": 2.6737,
+      "step": 495200
+    },
+    {
+      "epoch": 0.12918205386427487,
+      "grad_norm": 9.327274322509766,
+      "learning_rate": 9.69685265238316e-06,
+      "loss": 2.6599,
+      "step": 495400
+    },
+    {
+      "epoch": 0.12923420648997705,
+      "grad_norm": 8.239920616149902,
+      "learning_rate": 9.696565943209551e-06,
+      "loss": 2.6745,
+      "step": 495600
+    },
+    {
+      "epoch": 0.1292863591156792,
+      "grad_norm": 10.874256134033203,
+      "learning_rate": 9.696279102761508e-06,
+      "loss": 2.639,
+      "step": 495800
+    },
+    {
+      "epoch": 0.1293385117413814,
+      "grad_norm": 8.510387420654297,
+      "learning_rate": 9.69599213104705e-06,
+      "loss": 2.6401,
+      "step": 496000
+    },
+    {
+      "epoch": 0.12939066436708355,
+      "grad_norm": 9.092486381530762,
+      "learning_rate": 9.695705028074197e-06,
+      "loss": 2.6394,
+      "step": 496200
+    },
+    {
+      "epoch": 0.12944281699278573,
+      "grad_norm": 9.543620109558105,
+      "learning_rate": 9.695417793850974e-06,
+      "loss": 2.6777,
+      "step": 496400
+    },
+    {
+      "epoch": 0.1294949696184879,
+      "grad_norm": 8.863883018493652,
+      "learning_rate": 9.69513042838541e-06,
+      "loss": 2.6614,
+      "step": 496600
+    },
+    {
+      "epoch": 0.12954712224419007,
+      "grad_norm": 8.839035034179688,
+      "learning_rate": 9.694842931685536e-06,
+      "loss": 2.6575,
+      "step": 496800
+    },
+    {
+      "epoch": 0.12959927486989223,
+      "grad_norm": 9.193402290344238,
+      "learning_rate": 9.69455530375939e-06,
+      "loss": 2.676,
+      "step": 497000
+    },
+    {
+      "epoch": 0.1296514274955944,
+      "grad_norm": 8.699831008911133,
+      "learning_rate": 9.69426754461501e-06,
+      "loss": 2.654,
+      "step": 497200
+    },
+    {
+      "epoch": 0.12970358012129657,
+      "grad_norm": 9.69882869720459,
+      "learning_rate": 9.693979654260439e-06,
+      "loss": 2.6663,
+      "step": 497400
+    },
+    {
+      "epoch": 0.12975573274699875,
+      "grad_norm": 10.164560317993164,
+      "learning_rate": 9.693691632703724e-06,
+      "loss": 2.671,
+      "step": 497600
+    },
+    {
+      "epoch": 0.1298078853727009,
+      "grad_norm": 9.261726379394531,
+      "learning_rate": 9.693403479952914e-06,
+      "loss": 2.6322,
+      "step": 497800
+    },
+    {
+      "epoch": 0.1298600379984031,
+      "grad_norm": 8.697185516357422,
+      "learning_rate": 9.693115196016068e-06,
+      "loss": 2.6649,
+      "step": 498000
+    },
+    {
+      "epoch": 0.12991219062410525,
+      "grad_norm": 9.236394882202148,
+      "learning_rate": 9.69282678090124e-06,
+      "loss": 2.6507,
+      "step": 498200
+    },
+    {
+      "epoch": 0.12996434324980743,
+      "grad_norm": 9.209188461303711,
+      "learning_rate": 9.692538234616491e-06,
+      "loss": 2.6858,
+      "step": 498400
+    },
+    {
+      "epoch": 0.1300164958755096,
+      "grad_norm": 7.591978073120117,
+      "learning_rate": 9.69224955716989e-06,
+      "loss": 2.6934,
+      "step": 498600
+    },
+    {
+      "epoch": 0.13006864850121178,
+      "grad_norm": 9.55863094329834,
+      "learning_rate": 9.691960748569503e-06,
+      "loss": 2.6473,
+      "step": 498800
+    },
+    {
+      "epoch": 0.13012080112691393,
+      "grad_norm": 9.636316299438477,
+      "learning_rate": 9.691671808823403e-06,
+      "loss": 2.7021,
+      "step": 499000
+    },
+    {
+      "epoch": 0.13017295375261612,
+      "grad_norm": 9.680166244506836,
+      "learning_rate": 9.691382737939665e-06,
+      "loss": 2.6483,
+      "step": 499200
+    },
+    {
+      "epoch": 0.13022510637831827,
+      "grad_norm": 8.427440643310547,
+      "learning_rate": 9.69109353592637e-06,
+      "loss": 2.6132,
+      "step": 499400
+    },
+    {
+      "epoch": 0.13027725900402046,
+      "grad_norm": 9.360069274902344,
+      "learning_rate": 9.690804202791603e-06,
+      "loss": 2.6739,
+      "step": 499600
+    },
+    {
+      "epoch": 0.1303294116297226,
+      "grad_norm": 8.880375862121582,
+      "learning_rate": 9.69051473854345e-06,
+      "loss": 2.6774,
+      "step": 499800
+    },
+    {
+      "epoch": 0.1303815642554248,
+      "grad_norm": 8.440618515014648,
+      "learning_rate": 9.69022514319e-06,
+      "loss": 2.6508,
+      "step": 500000
+    },
+    {
+      "epoch": 0.13043371688112695,
+      "grad_norm": 9.217138290405273,
+      "learning_rate": 9.68993541673935e-06,
+      "loss": 2.6629,
+      "step": 500200
+    },
+    {
+      "epoch": 0.13048586950682914,
+      "grad_norm": 9.152853012084961,
+      "learning_rate": 9.6896455591996e-06,
+      "loss": 2.6395,
+      "step": 500400
+    },
+    {
+      "epoch": 0.1305380221325313,
+      "grad_norm": 9.89096736907959,
+      "learning_rate": 9.689355570578845e-06,
+      "loss": 2.6171,
+      "step": 500600
+    },
+    {
+      "epoch": 0.13059017475823345,
+      "grad_norm": 8.032015800476074,
+      "learning_rate": 9.689065450885198e-06,
+      "loss": 2.6471,
+      "step": 500800
+    },
+    {
+      "epoch": 0.13064232738393564,
+      "grad_norm": 8.966907501220703,
+      "learning_rate": 9.688775200126763e-06,
+      "loss": 2.6885,
+      "step": 501000
+    },
+    {
+      "epoch": 0.1306944800096378,
+      "grad_norm": 9.194870948791504,
+      "learning_rate": 9.688484818311654e-06,
+      "loss": 2.6357,
+      "step": 501200
+    },
+    {
+      "epoch": 0.13074663263533998,
+      "grad_norm": 10.05204963684082,
+      "learning_rate": 9.68819430544799e-06,
+      "loss": 2.6813,
+      "step": 501400
+    },
+    {
+      "epoch": 0.13079878526104213,
+      "grad_norm": 9.8402738571167,
+      "learning_rate": 9.68790366154389e-06,
+      "loss": 2.6826,
+      "step": 501600
+    },
+    {
+      "epoch": 0.13085093788674432,
+      "grad_norm": 8.931715965270996,
+      "learning_rate": 9.687612886607477e-06,
+      "loss": 2.6638,
+      "step": 501800
+    },
+    {
+      "epoch": 0.13090309051244647,
+      "grad_norm": 9.182944297790527,
+      "learning_rate": 9.687321980646877e-06,
+      "loss": 2.6094,
+      "step": 502000
+    },
+    {
+      "epoch": 0.13095524313814866,
+      "grad_norm": 8.474428176879883,
+      "learning_rate": 9.687030943670224e-06,
+      "loss": 2.6104,
+      "step": 502200
+    },
+    {
+      "epoch": 0.13100739576385081,
+      "grad_norm": 9.128947257995605,
+      "learning_rate": 9.686739775685653e-06,
+      "loss": 2.6591,
+      "step": 502400
+    },
+    {
+      "epoch": 0.131059548389553,
+      "grad_norm": 8.498746871948242,
+      "learning_rate": 9.6864484767013e-06,
+      "loss": 2.6751,
+      "step": 502600
+    },
+    {
+      "epoch": 0.13111170101525516,
+      "grad_norm": 9.038972854614258,
+      "learning_rate": 9.686157046725308e-06,
+      "loss": 2.6486,
+      "step": 502800
+    },
+    {
+      "epoch": 0.13116385364095734,
+      "grad_norm": 7.748460292816162,
+      "learning_rate": 9.685865485765824e-06,
+      "loss": 2.6387,
+      "step": 503000
+    },
+    {
+      "epoch": 0.1312160062666595,
+      "grad_norm": 9.406355857849121,
+      "learning_rate": 9.685573793830996e-06,
+      "loss": 2.66,
+      "step": 503200
+    },
+    {
+      "epoch": 0.13126815889236168,
+      "grad_norm": 10.351799011230469,
+      "learning_rate": 9.68528197092898e-06,
+      "loss": 2.6431,
+      "step": 503400
+    },
+    {
+      "epoch": 0.13132031151806384,
+      "grad_norm": 10.22693920135498,
+      "learning_rate": 9.684990017067928e-06,
+      "loss": 2.6585,
+      "step": 503600
+    },
+    {
+      "epoch": 0.13137246414376602,
+      "grad_norm": 9.691548347473145,
+      "learning_rate": 9.684697932256004e-06,
+      "loss": 2.6583,
+      "step": 503800
+    },
+    {
+      "epoch": 0.13142461676946818,
+      "grad_norm": 9.202975273132324,
+      "learning_rate": 9.68440571650137e-06,
+      "loss": 2.6798,
+      "step": 504000
+    },
+    {
+      "epoch": 0.13147676939517036,
+      "grad_norm": 10.292789459228516,
+      "learning_rate": 9.684113369812198e-06,
+      "loss": 2.6643,
+      "step": 504200
+    },
+    {
+      "epoch": 0.13152892202087252,
+      "grad_norm": 7.787657737731934,
+      "learning_rate": 9.683820892196654e-06,
+      "loss": 2.631,
+      "step": 504400
+    },
+    {
+      "epoch": 0.1315810746465747,
+      "grad_norm": 8.93878173828125,
+      "learning_rate": 9.683528283662916e-06,
+      "loss": 2.6179,
+      "step": 504600
+    },
+    {
+      "epoch": 0.13163322727227686,
+      "grad_norm": 7.960376739501953,
+      "learning_rate": 9.683235544219161e-06,
+      "loss": 2.6363,
+      "step": 504800
+    },
+    {
+      "epoch": 0.13168537989797904,
+      "grad_norm": 8.953857421875,
+      "learning_rate": 9.682942673873572e-06,
+      "loss": 2.642,
+      "step": 505000
+    },
+    {
+      "epoch": 0.1317375325236812,
+      "grad_norm": 8.703145027160645,
+      "learning_rate": 9.682649672634337e-06,
+      "loss": 2.639,
+      "step": 505200
+    },
+    {
+      "epoch": 0.13178968514938338,
+      "grad_norm": 9.463846206665039,
+      "learning_rate": 9.682356540509645e-06,
+      "loss": 2.6506,
+      "step": 505400
+    },
+    {
+      "epoch": 0.13184183777508554,
+      "grad_norm": 9.317432403564453,
+      "learning_rate": 9.68206327750769e-06,
+      "loss": 2.6546,
+      "step": 505600
+    },
+    {
+      "epoch": 0.13189399040078773,
+      "grad_norm": 8.97231674194336,
+      "learning_rate": 9.681769883636664e-06,
+      "loss": 2.6196,
+      "step": 505800
+    },
+    {
+      "epoch": 0.13194614302648988,
+      "grad_norm": 9.235013961791992,
+      "learning_rate": 9.681476358904773e-06,
+      "loss": 2.6777,
+      "step": 506000
+    },
+    {
+      "epoch": 0.13199829565219207,
+      "grad_norm": 9.837597846984863,
+      "learning_rate": 9.68118270332022e-06,
+      "loss": 2.6295,
+      "step": 506200
+    },
+    {
+      "epoch": 0.13205044827789422,
+      "grad_norm": 9.842741966247559,
+      "learning_rate": 9.680888916891213e-06,
+      "loss": 2.6498,
+      "step": 506400
+    },
+    {
+      "epoch": 0.1321026009035964,
+      "grad_norm": 9.616024017333984,
+      "learning_rate": 9.680594999625964e-06,
+      "loss": 2.6751,
+      "step": 506600
+    },
+    {
+      "epoch": 0.13215475352929856,
+      "grad_norm": 8.86656665802002,
+      "learning_rate": 9.680300951532686e-06,
+      "loss": 2.6171,
+      "step": 506800
+    },
+    {
+      "epoch": 0.13220690615500072,
+      "grad_norm": 8.192331314086914,
+      "learning_rate": 9.6800067726196e-06,
+      "loss": 2.6294,
+      "step": 507000
+    },
+    {
+      "epoch": 0.1322590587807029,
+      "grad_norm": 8.469772338867188,
+      "learning_rate": 9.679712462894931e-06,
+      "loss": 2.6976,
+      "step": 507200
+    },
+    {
+      "epoch": 0.13231121140640506,
+      "grad_norm": 7.870084762573242,
+      "learning_rate": 9.679418022366903e-06,
+      "loss": 2.6391,
+      "step": 507400
+    },
+    {
+      "epoch": 0.13236336403210724,
+      "grad_norm": 9.129673957824707,
+      "learning_rate": 9.679123451043743e-06,
+      "loss": 2.6854,
+      "step": 507600
+    },
+    {
+      "epoch": 0.1324155166578094,
+      "grad_norm": 9.1187744140625,
+      "learning_rate": 9.678828748933689e-06,
+      "loss": 2.6234,
+      "step": 507800
+    },
+    {
+      "epoch": 0.13246766928351159,
+      "grad_norm": 9.26366901397705,
+      "learning_rate": 9.678533916044974e-06,
+      "loss": 2.64,
+      "step": 508000
+    },
+    {
+      "epoch": 0.13251982190921374,
+      "grad_norm": 8.143036842346191,
+      "learning_rate": 9.678238952385844e-06,
+      "loss": 2.6068,
+      "step": 508200
+    },
+    {
+      "epoch": 0.13257197453491593,
+      "grad_norm": 9.583711624145508,
+      "learning_rate": 9.67794385796454e-06,
+      "loss": 2.6514,
+      "step": 508400
+    },
+    {
+      "epoch": 0.13262412716061808,
+      "grad_norm": 8.947481155395508,
+      "learning_rate": 9.67764863278931e-06,
+      "loss": 2.6757,
+      "step": 508600
+    },
+    {
+      "epoch": 0.13267627978632027,
+      "grad_norm": 9.801794052124023,
+      "learning_rate": 9.67735327686841e-06,
+      "loss": 2.6304,
+      "step": 508800
+    },
+    {
+      "epoch": 0.13272843241202242,
+      "grad_norm": 8.388129234313965,
+      "learning_rate": 9.677057790210093e-06,
+      "loss": 2.6435,
+      "step": 509000
+    },
+    {
+      "epoch": 0.1327805850377246,
+      "grad_norm": 9.845732688903809,
+      "learning_rate": 9.676762172822615e-06,
+      "loss": 2.6488,
+      "step": 509200
+    },
+    {
+      "epoch": 0.13283273766342676,
+      "grad_norm": 8.10527515411377,
+      "learning_rate": 9.676466424714244e-06,
+      "loss": 2.6525,
+      "step": 509400
+    },
+    {
+      "epoch": 0.13288489028912895,
+      "grad_norm": 9.184629440307617,
+      "learning_rate": 9.676170545893245e-06,
+      "loss": 2.62,
+      "step": 509600
+    },
+    {
+      "epoch": 0.1329370429148311,
+      "grad_norm": 9.896688461303711,
+      "learning_rate": 9.675874536367883e-06,
+      "loss": 2.6222,
+      "step": 509800
+    },
+    {
+      "epoch": 0.1329891955405333,
+      "grad_norm": 9.492822647094727,
+      "learning_rate": 9.675578396146441e-06,
+      "loss": 2.6601,
+      "step": 510000
+    },
+    {
+      "epoch": 0.13304134816623545,
+      "grad_norm": 8.227307319641113,
+      "learning_rate": 9.67528212523719e-06,
+      "loss": 2.6555,
+      "step": 510200
+    },
+    {
+      "epoch": 0.13309350079193763,
+      "grad_norm": 8.036877632141113,
+      "learning_rate": 9.674985723648413e-06,
+      "loss": 2.648,
+      "step": 510400
+    },
+    {
+      "epoch": 0.1331456534176398,
+      "grad_norm": 10.020105361938477,
+      "learning_rate": 9.674689191388393e-06,
+      "loss": 2.6916,
+      "step": 510600
+    },
+    {
+      "epoch": 0.13319780604334197,
+      "grad_norm": 9.248919486999512,
+      "learning_rate": 9.674392528465421e-06,
+      "loss": 2.621,
+      "step": 510800
+    },
+    {
+      "epoch": 0.13324995866904413,
+      "grad_norm": 10.11646842956543,
+      "learning_rate": 9.67409573488779e-06,
+      "loss": 2.6674,
+      "step": 511000
+    },
+    {
+      "epoch": 0.1333021112947463,
+      "grad_norm": 8.71399974822998,
+      "learning_rate": 9.67379881066379e-06,
+      "loss": 2.6399,
+      "step": 511200
+    },
+    {
+      "epoch": 0.13335426392044847,
+      "grad_norm": 9.745195388793945,
+      "learning_rate": 9.673501755801726e-06,
+      "loss": 2.6376,
+      "step": 511400
+    },
+    {
+      "epoch": 0.13340641654615065,
+      "grad_norm": 9.938071250915527,
+      "learning_rate": 9.673204570309899e-06,
+      "loss": 2.6562,
+      "step": 511600
+    },
+    {
+      "epoch": 0.1334585691718528,
+      "grad_norm": 9.505668640136719,
+      "learning_rate": 9.672907254196618e-06,
+      "loss": 2.6752,
+      "step": 511800
+    },
+    {
+      "epoch": 0.133510721797555,
+      "grad_norm": 9.43238353729248,
+      "learning_rate": 9.672609807470187e-06,
+      "loss": 2.6658,
+      "step": 512000
+    },
+    {
+      "epoch": 0.13356287442325715,
+      "grad_norm": 9.957179069519043,
+      "learning_rate": 9.672312230138926e-06,
+      "loss": 2.6154,
+      "step": 512200
+    },
+    {
+      "epoch": 0.13361502704895933,
+      "grad_norm": 10.3068265914917,
+      "learning_rate": 9.672014522211153e-06,
+      "loss": 2.6855,
+      "step": 512400
+    },
+    {
+      "epoch": 0.1336671796746615,
+      "grad_norm": 10.027313232421875,
+      "learning_rate": 9.671716683695184e-06,
+      "loss": 2.6288,
+      "step": 512600
+    },
+    {
+      "epoch": 0.13371933230036365,
+      "grad_norm": 8.177754402160645,
+      "learning_rate": 9.67141871459935e-06,
+      "loss": 2.6548,
+      "step": 512800
+    },
+    {
+      "epoch": 0.13377148492606583,
+      "grad_norm": 7.8090500831604,
+      "learning_rate": 9.671120614931975e-06,
+      "loss": 2.6628,
+      "step": 513000
+    },
+    {
+      "epoch": 0.133823637551768,
+      "grad_norm": 9.168340682983398,
+      "learning_rate": 9.670822384701393e-06,
+      "loss": 2.6837,
+      "step": 513200
+    },
+    {
+      "epoch": 0.13387579017747017,
+      "grad_norm": 9.05853271484375,
+      "learning_rate": 9.670524023915939e-06,
+      "loss": 2.619,
+      "step": 513400
+    },
+    {
+      "epoch": 0.13392794280317233,
+      "grad_norm": 9.077415466308594,
+      "learning_rate": 9.670225532583954e-06,
+      "loss": 2.6538,
+      "step": 513600
+    },
+    {
+      "epoch": 0.1339800954288745,
+      "grad_norm": 10.212859153747559,
+      "learning_rate": 9.66992691071378e-06,
+      "loss": 2.6305,
+      "step": 513800
+    },
+    {
+      "epoch": 0.13403224805457667,
+      "grad_norm": 8.626009941101074,
+      "learning_rate": 9.669628158313764e-06,
+      "loss": 2.6326,
+      "step": 514000
+    },
+    {
+      "epoch": 0.13408440068027885,
+      "grad_norm": 9.030853271484375,
+      "learning_rate": 9.669329275392256e-06,
+      "loss": 2.6567,
+      "step": 514200
+    },
+    {
+      "epoch": 0.134136553305981,
+      "grad_norm": 10.016151428222656,
+      "learning_rate": 9.669030261957613e-06,
+      "loss": 2.6399,
+      "step": 514400
+    },
+    {
+      "epoch": 0.1341887059316832,
+      "grad_norm": 9.281685829162598,
+      "learning_rate": 9.668731118018189e-06,
+      "loss": 2.6532,
+      "step": 514600
+    },
+    {
+      "epoch": 0.13424085855738535,
+      "grad_norm": 9.473543167114258,
+      "learning_rate": 9.668431843582347e-06,
+      "loss": 2.6171,
+      "step": 514800
+    },
+    {
+      "epoch": 0.13429301118308754,
+      "grad_norm": 11.197412490844727,
+      "learning_rate": 9.668132438658452e-06,
+      "loss": 2.6532,
+      "step": 515000
+    },
+    {
+      "epoch": 0.1343451638087897,
+      "grad_norm": 8.542428016662598,
+      "learning_rate": 9.667832903254873e-06,
+      "loss": 2.6227,
+      "step": 515200
+    },
+    {
+      "epoch": 0.13439731643449188,
+      "grad_norm": 8.935531616210938,
+      "learning_rate": 9.667533237379983e-06,
+      "loss": 2.663,
+      "step": 515400
+    },
+    {
+      "epoch": 0.13444946906019403,
+      "grad_norm": 8.45200252532959,
+      "learning_rate": 9.667233441042156e-06,
+      "loss": 2.6751,
+      "step": 515600
+    },
+    {
+      "epoch": 0.13450162168589622,
+      "grad_norm": 10.14538288116455,
+      "learning_rate": 9.666933514249773e-06,
+      "loss": 2.6057,
+      "step": 515800
+    },
+    {
+      "epoch": 0.13455377431159837,
+      "grad_norm": 9.2836332321167,
+      "learning_rate": 9.666633457011216e-06,
+      "loss": 2.6303,
+      "step": 516000
+    },
+    {
+      "epoch": 0.13460592693730056,
+      "grad_norm": 10.366707801818848,
+      "learning_rate": 9.666333269334876e-06,
+      "loss": 2.693,
+      "step": 516200
+    },
+    {
+      "epoch": 0.1346580795630027,
+      "grad_norm": 10.014763832092285,
+      "learning_rate": 9.666032951229138e-06,
+      "loss": 2.6208,
+      "step": 516400
+    },
+    {
+      "epoch": 0.1347102321887049,
+      "grad_norm": 8.867531776428223,
+      "learning_rate": 9.665732502702401e-06,
+      "loss": 2.6653,
+      "step": 516600
+    },
+    {
+      "epoch": 0.13476238481440705,
+      "grad_norm": 9.882331848144531,
+      "learning_rate": 9.665431923763059e-06,
+      "loss": 2.6914,
+      "step": 516800
+    },
+    {
+      "epoch": 0.13481453744010924,
+      "grad_norm": 9.857499122619629,
+      "learning_rate": 9.665131214419516e-06,
+      "loss": 2.6641,
+      "step": 517000
+    },
+    {
+      "epoch": 0.1348666900658114,
+      "grad_norm": 9.31207275390625,
+      "learning_rate": 9.664830374680176e-06,
+      "loss": 2.6546,
+      "step": 517200
+    },
+    {
+      "epoch": 0.13491884269151358,
+      "grad_norm": 10.368709564208984,
+      "learning_rate": 9.664529404553449e-06,
+      "loss": 2.6633,
+      "step": 517400
+    },
+    {
+      "epoch": 0.13497099531721574,
+      "grad_norm": 9.503002166748047,
+      "learning_rate": 9.664228304047746e-06,
+      "loss": 2.6746,
+      "step": 517600
+    },
+    {
+      "epoch": 0.13502314794291792,
+      "grad_norm": 8.640365600585938,
+      "learning_rate": 9.663927073171485e-06,
+      "loss": 2.6339,
+      "step": 517800
+    },
+    {
+      "epoch": 0.13507530056862008,
+      "grad_norm": 10.261897087097168,
+      "learning_rate": 9.663625711933083e-06,
+      "loss": 2.6787,
+      "step": 518000
+    },
+    {
+      "epoch": 0.13512745319432226,
+      "grad_norm": 10.067601203918457,
+      "learning_rate": 9.663324220340964e-06,
+      "loss": 2.6697,
+      "step": 518200
+    },
+    {
+      "epoch": 0.13517960582002442,
+      "grad_norm": 8.839470863342285,
+      "learning_rate": 9.66302259840356e-06,
+      "loss": 2.6629,
+      "step": 518400
+    },
+    {
+      "epoch": 0.13523175844572657,
+      "grad_norm": 9.251420974731445,
+      "learning_rate": 9.662720846129295e-06,
+      "loss": 2.6574,
+      "step": 518600
+    },
+    {
+      "epoch": 0.13528391107142876,
+      "grad_norm": 9.32863998413086,
+      "learning_rate": 9.662418963526605e-06,
+      "loss": 2.6484,
+      "step": 518800
+    },
+    {
+      "epoch": 0.13533606369713091,
+      "grad_norm": 9.35538387298584,
+      "learning_rate": 9.662116950603932e-06,
+      "loss": 2.6542,
+      "step": 519000
+    },
+    {
+      "epoch": 0.1353882163228331,
+      "grad_norm": 9.882723808288574,
+      "learning_rate": 9.661814807369713e-06,
+      "loss": 2.6853,
+      "step": 519200
+    },
+    {
+      "epoch": 0.13544036894853526,
+      "grad_norm": 9.342474937438965,
+      "learning_rate": 9.661512533832395e-06,
+      "loss": 2.6694,
+      "step": 519400
+    },
+    {
+      "epoch": 0.13549252157423744,
+      "grad_norm": 8.824580192565918,
+      "learning_rate": 9.661210130000425e-06,
+      "loss": 2.6505,
+      "step": 519600
+    },
+    {
+      "epoch": 0.1355446741999396,
+      "grad_norm": 9.28695011138916,
+      "learning_rate": 9.660907595882261e-06,
+      "loss": 2.6744,
+      "step": 519800
+    },
+    {
+      "epoch": 0.13559682682564178,
+      "grad_norm": 9.04250431060791,
+      "learning_rate": 9.660604931486353e-06,
+      "loss": 2.6515,
+      "step": 520000
+    },
+    {
+      "epoch": 0.13564897945134394,
+      "grad_norm": 10.339582443237305,
+      "learning_rate": 9.660302136821163e-06,
+      "loss": 2.6614,
+      "step": 520200
+    },
+    {
+      "epoch": 0.13570113207704612,
+      "grad_norm": 8.636717796325684,
+      "learning_rate": 9.659999211895156e-06,
+      "loss": 2.6249,
+      "step": 520400
+    },
+    {
+      "epoch": 0.13575328470274828,
+      "grad_norm": 8.852993965148926,
+      "learning_rate": 9.659696156716797e-06,
+      "loss": 2.6207,
+      "step": 520600
+    },
+    {
+      "epoch": 0.13580543732845046,
+      "grad_norm": 9.094849586486816,
+      "learning_rate": 9.659392971294558e-06,
+      "loss": 2.6388,
+      "step": 520800
+    },
+    {
+      "epoch": 0.13585758995415262,
+      "grad_norm": 8.45062255859375,
+      "learning_rate": 9.659089655636914e-06,
+      "loss": 2.6394,
+      "step": 521000
+    },
+    {
+      "epoch": 0.1359097425798548,
+      "grad_norm": 10.011489868164062,
+      "learning_rate": 9.65878620975234e-06,
+      "loss": 2.6376,
+      "step": 521200
+    },
+    {
+      "epoch": 0.13596189520555696,
+      "grad_norm": 9.696266174316406,
+      "learning_rate": 9.65848263364932e-06,
+      "loss": 2.6553,
+      "step": 521400
+    },
+    {
+      "epoch": 0.13601404783125914,
+      "grad_norm": 8.92795181274414,
+      "learning_rate": 9.65817892733634e-06,
+      "loss": 2.6496,
+      "step": 521600
+    },
+    {
+      "epoch": 0.1360662004569613,
+      "grad_norm": 9.415807723999023,
+      "learning_rate": 9.657875090821886e-06,
+      "loss": 2.5864,
+      "step": 521800
+    },
+    {
+      "epoch": 0.13611835308266348,
+      "grad_norm": 10.63681697845459,
+      "learning_rate": 9.657571124114455e-06,
+      "loss": 2.6961,
+      "step": 522000
+    },
+    {
+      "epoch": 0.13617050570836564,
+      "grad_norm": 9.53947639465332,
+      "learning_rate": 9.657267027222539e-06,
+      "loss": 2.6501,
+      "step": 522200
+    },
+    {
+      "epoch": 0.13622265833406783,
+      "grad_norm": 10.805290222167969,
+      "learning_rate": 9.656962800154641e-06,
+      "loss": 2.6548,
+      "step": 522400
+    },
+    {
+      "epoch": 0.13627481095976998,
+      "grad_norm": 7.7652764320373535,
+      "learning_rate": 9.656658442919261e-06,
+      "loss": 2.6244,
+      "step": 522600
+    },
+    {
+      "epoch": 0.13632696358547217,
+      "grad_norm": 9.094009399414062,
+      "learning_rate": 9.65635395552491e-06,
+      "loss": 2.6543,
+      "step": 522800
+    },
+    {
+      "epoch": 0.13637911621117432,
+      "grad_norm": 9.510210990905762,
+      "learning_rate": 9.656049337980096e-06,
+      "loss": 2.6013,
+      "step": 523000
+    },
+    {
+      "epoch": 0.1364312688368765,
+      "grad_norm": 9.347325325012207,
+      "learning_rate": 9.655744590293334e-06,
+      "loss": 2.617,
+      "step": 523200
+    },
+    {
+      "epoch": 0.13648342146257866,
+      "grad_norm": 10.486848831176758,
+      "learning_rate": 9.655439712473143e-06,
+      "loss": 2.6461,
+      "step": 523400
+    },
+    {
+      "epoch": 0.13653557408828085,
+      "grad_norm": 9.532522201538086,
+      "learning_rate": 9.655134704528044e-06,
+      "loss": 2.6238,
+      "step": 523600
+    },
+    {
+      "epoch": 0.136587726713983,
+      "grad_norm": 9.079800605773926,
+      "learning_rate": 9.654829566466564e-06,
+      "loss": 2.6561,
+      "step": 523800
+    },
+    {
+      "epoch": 0.1366398793396852,
+      "grad_norm": 9.900603294372559,
+      "learning_rate": 9.654524298297228e-06,
+      "loss": 2.6736,
+      "step": 524000
+    },
+    {
+      "epoch": 0.13669203196538735,
+      "grad_norm": 10.008954048156738,
+      "learning_rate": 9.65421890002857e-06,
+      "loss": 2.5807,
+      "step": 524200
+    },
+    {
+      "epoch": 0.1367441845910895,
+      "grad_norm": 9.736871719360352,
+      "learning_rate": 9.653913371669131e-06,
+      "loss": 2.6297,
+      "step": 524400
+    },
+    {
+      "epoch": 0.13679633721679169,
+      "grad_norm": 9.787025451660156,
+      "learning_rate": 9.653607713227445e-06,
+      "loss": 2.6355,
+      "step": 524600
+    },
+    {
+      "epoch": 0.13684848984249384,
+      "grad_norm": 9.512006759643555,
+      "learning_rate": 9.65330192471206e-06,
+      "loss": 2.6131,
+      "step": 524800
+    },
+    {
+      "epoch": 0.13690064246819603,
+      "grad_norm": 9.525954246520996,
+      "learning_rate": 9.652996006131517e-06,
+      "loss": 2.6761,
+      "step": 525000
+    },
+    {
+      "epoch": 0.13695279509389818,
+      "grad_norm": 9.713210105895996,
+      "learning_rate": 9.652689957494371e-06,
+      "loss": 2.6427,
+      "step": 525200
+    },
+    {
+      "epoch": 0.13700494771960037,
+      "grad_norm": 7.939797878265381,
+      "learning_rate": 9.652383778809178e-06,
+      "loss": 2.6448,
+      "step": 525400
+    },
+    {
+      "epoch": 0.13705710034530252,
+      "grad_norm": 9.904065132141113,
+      "learning_rate": 9.652077470084492e-06,
+      "loss": 2.616,
+      "step": 525600
+    },
+    {
+      "epoch": 0.1371092529710047,
+      "grad_norm": 8.858099937438965,
+      "learning_rate": 9.651771031328878e-06,
+      "loss": 2.6253,
+      "step": 525800
+    },
+    {
+      "epoch": 0.13716140559670686,
+      "grad_norm": 9.863373756408691,
+      "learning_rate": 9.651464462550902e-06,
+      "loss": 2.6251,
+      "step": 526000
+    },
+    {
+      "epoch": 0.13721355822240905,
+      "grad_norm": 9.789137840270996,
+      "learning_rate": 9.651157763759126e-06,
+      "loss": 2.6234,
+      "step": 526200
+    },
+    {
+      "epoch": 0.1372657108481112,
+      "grad_norm": 10.034318923950195,
+      "learning_rate": 9.650850934962132e-06,
+      "loss": 2.6572,
+      "step": 526400
+    },
+    {
+      "epoch": 0.1373178634738134,
+      "grad_norm": 9.916723251342773,
+      "learning_rate": 9.65054397616849e-06,
+      "loss": 2.632,
+      "step": 526600
+    },
+    {
+      "epoch": 0.13737001609951555,
+      "grad_norm": 8.940628051757812,
+      "learning_rate": 9.65023688738678e-06,
+      "loss": 2.6402,
+      "step": 526800
+    },
+    {
+      "epoch": 0.13742216872521773,
+      "grad_norm": 9.220625877380371,
+      "learning_rate": 9.649929668625589e-06,
+      "loss": 2.6262,
+      "step": 527000
+    },
+    {
+      "epoch": 0.1374743213509199,
+      "grad_norm": 8.476348876953125,
+      "learning_rate": 9.649622319893502e-06,
+      "loss": 2.6734,
+      "step": 527200
+    },
+    {
+      "epoch": 0.13752647397662207,
+      "grad_norm": 8.9732084274292,
+      "learning_rate": 9.649314841199109e-06,
+      "loss": 2.6198,
+      "step": 527400
+    },
+    {
+      "epoch": 0.13757862660232423,
+      "grad_norm": 9.643847465515137,
+      "learning_rate": 9.649007232551006e-06,
+      "loss": 2.672,
+      "step": 527600
+    },
+    {
+      "epoch": 0.1376307792280264,
+      "grad_norm": 9.115861892700195,
+      "learning_rate": 9.648699493957792e-06,
+      "loss": 2.6423,
+      "step": 527800
+    },
+    {
+      "epoch": 0.13768293185372857,
+      "grad_norm": 10.297475814819336,
+      "learning_rate": 9.648391625428064e-06,
+      "loss": 2.6552,
+      "step": 528000
+    },
+    {
+      "epoch": 0.13773508447943075,
+      "grad_norm": 9.935859680175781,
+      "learning_rate": 9.648083626970432e-06,
+      "loss": 2.6381,
+      "step": 528200
+    },
+    {
+      "epoch": 0.1377872371051329,
+      "grad_norm": 10.028658866882324,
+      "learning_rate": 9.647775498593502e-06,
+      "loss": 2.6879,
+      "step": 528400
+    },
+    {
+      "epoch": 0.1378393897308351,
+      "grad_norm": 10.135675430297852,
+      "learning_rate": 9.647467240305888e-06,
+      "loss": 2.6395,
+      "step": 528600
+    },
+    {
+      "epoch": 0.13789154235653725,
+      "grad_norm": 9.688112258911133,
+      "learning_rate": 9.647158852116207e-06,
+      "loss": 2.6373,
+      "step": 528800
+    },
+    {
+      "epoch": 0.13794369498223943,
+      "grad_norm": 9.650997161865234,
+      "learning_rate": 9.646850334033078e-06,
+      "loss": 2.623,
+      "step": 529000
+    },
+    {
+      "epoch": 0.1379958476079416,
+      "grad_norm": 8.232544898986816,
+      "learning_rate": 9.646541686065122e-06,
+      "loss": 2.6484,
+      "step": 529200
+    },
+    {
+      "epoch": 0.13804800023364378,
+      "grad_norm": 9.324143409729004,
+      "learning_rate": 9.64623290822097e-06,
+      "loss": 2.6221,
+      "step": 529400
+    },
+    {
+      "epoch": 0.13810015285934593,
+      "grad_norm": 9.4964599609375,
+      "learning_rate": 9.64592400050925e-06,
+      "loss": 2.6273,
+      "step": 529600
+    },
+    {
+      "epoch": 0.13815230548504812,
+      "grad_norm": 10.16766357421875,
+      "learning_rate": 9.6456149629386e-06,
+      "loss": 2.6316,
+      "step": 529800
+    },
+    {
+      "epoch": 0.13820445811075027,
+      "grad_norm": 9.12529468536377,
+      "learning_rate": 9.64530579551765e-06,
+      "loss": 2.6704,
+      "step": 530000
+    },
+    {
+      "epoch": 0.13825661073645246,
+      "grad_norm": 10.48526382446289,
+      "learning_rate": 9.64499649825505e-06,
+      "loss": 2.6643,
+      "step": 530200
+    },
+    {
+      "epoch": 0.1383087633621546,
+      "grad_norm": 8.55528736114502,
+      "learning_rate": 9.644687071159442e-06,
+      "loss": 2.6488,
+      "step": 530400
+    },
+    {
+      "epoch": 0.13836091598785677,
+      "grad_norm": 9.063948631286621,
+      "learning_rate": 9.644377514239473e-06,
+      "loss": 2.6316,
+      "step": 530600
+    },
+    {
+      "epoch": 0.13841306861355895,
+      "grad_norm": 8.189355850219727,
+      "learning_rate": 9.6440678275038e-06,
+      "loss": 2.6458,
+      "step": 530800
+    },
+    {
+      "epoch": 0.1384652212392611,
+      "grad_norm": 10.547919273376465,
+      "learning_rate": 9.643758010961075e-06,
+      "loss": 2.6643,
+      "step": 531000
+    },
+    {
+      "epoch": 0.1385173738649633,
+      "grad_norm": 8.875152587890625,
+      "learning_rate": 9.643448064619958e-06,
+      "loss": 2.6343,
+      "step": 531200
+    },
+    {
+      "epoch": 0.13856952649066545,
+      "grad_norm": 10.155411720275879,
+      "learning_rate": 9.643137988489115e-06,
+      "loss": 2.6402,
+      "step": 531400
+    },
+    {
+      "epoch": 0.13862167911636764,
+      "grad_norm": 9.877212524414062,
+      "learning_rate": 9.64282778257721e-06,
+      "loss": 2.6482,
+      "step": 531600
+    },
+    {
+      "epoch": 0.1386738317420698,
+      "grad_norm": 8.81261157989502,
+      "learning_rate": 9.642517446892919e-06,
+      "loss": 2.6458,
+      "step": 531800
+    },
+    {
+      "epoch": 0.13872598436777198,
+      "grad_norm": 9.765755653381348,
+      "learning_rate": 9.642206981444908e-06,
+      "loss": 2.6615,
+      "step": 532000
+    },
+    {
+      "epoch": 0.13877813699347413,
+      "grad_norm": 9.739371299743652,
+      "learning_rate": 9.641896386241861e-06,
+      "loss": 2.6525,
+      "step": 532200
+    },
+    {
+      "epoch": 0.13883028961917632,
+      "grad_norm": 10.315713882446289,
+      "learning_rate": 9.641585661292457e-06,
+      "loss": 2.659,
+      "step": 532400
+    },
+    {
+      "epoch": 0.13888244224487847,
+      "grad_norm": 9.760064125061035,
+      "learning_rate": 9.641274806605384e-06,
+      "loss": 2.6282,
+      "step": 532600
+    },
+    {
+      "epoch": 0.13893459487058066,
+      "grad_norm": 9.685067176818848,
+      "learning_rate": 9.640963822189327e-06,
+      "loss": 2.6185,
+      "step": 532800
+    },
+    {
+      "epoch": 0.13898674749628281,
+      "grad_norm": 10.004737854003906,
+      "learning_rate": 9.640652708052978e-06,
+      "loss": 2.6327,
+      "step": 533000
+    },
+    {
+      "epoch": 0.139038900121985,
+      "grad_norm": 8.516515731811523,
+      "learning_rate": 9.640341464205037e-06,
+      "loss": 2.6649,
+      "step": 533200
+    },
+    {
+      "epoch": 0.13909105274768715,
+      "grad_norm": 10.124285697937012,
+      "learning_rate": 9.640030090654202e-06,
+      "loss": 2.6205,
+      "step": 533400
+    },
+    {
+      "epoch": 0.13914320537338934,
+      "grad_norm": 10.336373329162598,
+      "learning_rate": 9.639718587409174e-06,
+      "loss": 2.6386,
+      "step": 533600
+    },
+    {
+      "epoch": 0.1391953579990915,
+      "grad_norm": 10.257096290588379,
+      "learning_rate": 9.639406954478663e-06,
+      "loss": 2.643,
+      "step": 533800
+    },
+    {
+      "epoch": 0.13924751062479368,
+      "grad_norm": 9.125519752502441,
+      "learning_rate": 9.639095191871379e-06,
+      "loss": 2.6576,
+      "step": 534000
+    },
+    {
+      "epoch": 0.13929966325049584,
+      "grad_norm": 8.960975646972656,
+      "learning_rate": 9.638783299596033e-06,
+      "loss": 2.6339,
+      "step": 534200
+    },
+    {
+      "epoch": 0.13935181587619802,
+      "grad_norm": 10.509865760803223,
+      "learning_rate": 9.638471277661347e-06,
+      "loss": 2.6326,
+      "step": 534400
+    },
+    {
+      "epoch": 0.13940396850190018,
+      "grad_norm": 9.875240325927734,
+      "learning_rate": 9.63815912607604e-06,
+      "loss": 2.6501,
+      "step": 534600
+    },
+    {
+      "epoch": 0.13945612112760236,
+      "grad_norm": 8.997678756713867,
+      "learning_rate": 9.637846844848838e-06,
+      "loss": 2.6376,
+      "step": 534800
+    },
+    {
+      "epoch": 0.13950827375330452,
+      "grad_norm": 9.591438293457031,
+      "learning_rate": 9.637534433988467e-06,
+      "loss": 2.6116,
+      "step": 535000
+    },
+    {
+      "epoch": 0.1395604263790067,
+      "grad_norm": 9.580354690551758,
+      "learning_rate": 9.637221893503662e-06,
+      "loss": 2.6207,
+      "step": 535200
+    },
+    {
+      "epoch": 0.13961257900470886,
+      "grad_norm": 8.961016654968262,
+      "learning_rate": 9.63690922340316e-06,
+      "loss": 2.6337,
+      "step": 535400
+    },
+    {
+      "epoch": 0.13966473163041104,
+      "grad_norm": 8.852940559387207,
+      "learning_rate": 9.636596423695697e-06,
+      "loss": 2.6134,
+      "step": 535600
+    },
+    {
+      "epoch": 0.1397168842561132,
+      "grad_norm": 9.002432823181152,
+      "learning_rate": 9.636283494390017e-06,
+      "loss": 2.6261,
+      "step": 535800
+    },
+    {
+      "epoch": 0.13976903688181538,
+      "grad_norm": 9.397136688232422,
+      "learning_rate": 9.635970435494868e-06,
+      "loss": 2.6458,
+      "step": 536000
+    },
+    {
+      "epoch": 0.13982118950751754,
+      "grad_norm": 9.2616548538208,
+      "learning_rate": 9.635657247019001e-06,
+      "loss": 2.6269,
+      "step": 536200
+    },
+    {
+      "epoch": 0.1398733421332197,
+      "grad_norm": 9.671730041503906,
+      "learning_rate": 9.635343928971167e-06,
+      "loss": 2.6507,
+      "step": 536400
+    },
+    {
+      "epoch": 0.13992549475892188,
+      "grad_norm": 9.459322929382324,
+      "learning_rate": 9.635030481360129e-06,
+      "loss": 2.6104,
+      "step": 536600
+    },
+    {
+      "epoch": 0.13997764738462404,
+      "grad_norm": 9.426004409790039,
+      "learning_rate": 9.63471690419464e-06,
+      "loss": 2.63,
+      "step": 536800
+    },
+    {
+      "epoch": 0.14002980001032622,
+      "grad_norm": 9.91978931427002,
+      "learning_rate": 9.634403197483472e-06,
+      "loss": 2.6297,
+      "step": 537000
+    },
+    {
+      "epoch": 0.14008195263602838,
+      "grad_norm": 10.36573600769043,
+      "learning_rate": 9.634089361235391e-06,
+      "loss": 2.6198,
+      "step": 537200
+    },
+    {
+      "epoch": 0.14013410526173056,
+      "grad_norm": 12.126402854919434,
+      "learning_rate": 9.633775395459169e-06,
+      "loss": 2.6285,
+      "step": 537400
+    },
+    {
+      "epoch": 0.14018625788743272,
+      "grad_norm": 8.916886329650879,
+      "learning_rate": 9.633461300163582e-06,
+      "loss": 2.6479,
+      "step": 537600
+    },
+    {
+      "epoch": 0.1402384105131349,
+      "grad_norm": 10.035049438476562,
+      "learning_rate": 9.633147075357409e-06,
+      "loss": 2.6763,
+      "step": 537800
+    },
+    {
+      "epoch": 0.14029056313883706,
+      "grad_norm": 9.93181324005127,
+      "learning_rate": 9.632832721049432e-06,
+      "loss": 2.6169,
+      "step": 538000
+    },
+    {
+      "epoch": 0.14034271576453924,
+      "grad_norm": 10.294628143310547,
+      "learning_rate": 9.632518237248439e-06,
+      "loss": 2.6648,
+      "step": 538200
+    },
+    {
+      "epoch": 0.1403948683902414,
+      "grad_norm": 9.200163841247559,
+      "learning_rate": 9.63220362396322e-06,
+      "loss": 2.6165,
+      "step": 538400
+    },
+    {
+      "epoch": 0.14044702101594359,
+      "grad_norm": 9.44741153717041,
+      "learning_rate": 9.631888881202569e-06,
+      "loss": 2.6108,
+      "step": 538600
+    },
+    {
+      "epoch": 0.14049917364164574,
+      "grad_norm": 9.307562828063965,
+      "learning_rate": 9.631574008975281e-06,
+      "loss": 2.6276,
+      "step": 538800
+    },
+    {
+      "epoch": 0.14055132626734793,
+      "grad_norm": 8.72498893737793,
+      "learning_rate": 9.631259007290162e-06,
+      "loss": 2.6432,
+      "step": 539000
+    },
+    {
+      "epoch": 0.14060347889305008,
+      "grad_norm": 10.48019027709961,
+      "learning_rate": 9.63094387615601e-06,
+      "loss": 2.6292,
+      "step": 539200
+    },
+    {
+      "epoch": 0.14065563151875227,
+      "grad_norm": 9.463690757751465,
+      "learning_rate": 9.63062861558164e-06,
+      "loss": 2.6205,
+      "step": 539400
+    },
+    {
+      "epoch": 0.14070778414445442,
+      "grad_norm": 9.219118118286133,
+      "learning_rate": 9.630313225575862e-06,
+      "loss": 2.6125,
+      "step": 539600
+    },
+    {
+      "epoch": 0.1407599367701566,
+      "grad_norm": 11.975398063659668,
+      "learning_rate": 9.629997706147488e-06,
+      "loss": 2.6088,
+      "step": 539800
+    },
+    {
+      "epoch": 0.14081208939585876,
+      "grad_norm": 8.592241287231445,
+      "learning_rate": 9.629682057305341e-06,
+      "loss": 2.626,
+      "step": 540000
+    },
+    {
+      "epoch": 0.14086424202156095,
+      "grad_norm": 9.817209243774414,
+      "learning_rate": 9.62936627905824e-06,
+      "loss": 2.647,
+      "step": 540200
+    },
+    {
+      "epoch": 0.1409163946472631,
+      "grad_norm": 10.493924140930176,
+      "learning_rate": 9.629050371415015e-06,
+      "loss": 2.622,
+      "step": 540400
+    },
+    {
+      "epoch": 0.1409685472729653,
+      "grad_norm": 9.412971496582031,
+      "learning_rate": 9.628734334384495e-06,
+      "loss": 2.6524,
+      "step": 540600
+    },
+    {
+      "epoch": 0.14102069989866745,
+      "grad_norm": 10.187376976013184,
+      "learning_rate": 9.628418167975512e-06,
+      "loss": 2.6041,
+      "step": 540800
+    },
+    {
+      "epoch": 0.14107285252436963,
+      "grad_norm": 9.966575622558594,
+      "learning_rate": 9.628101872196907e-06,
+      "loss": 2.664,
+      "step": 541000
+    },
+    {
+      "epoch": 0.1411250051500718,
+      "grad_norm": 9.64553451538086,
+      "learning_rate": 9.627785447057517e-06,
+      "loss": 2.686,
+      "step": 541200
+    },
+    {
+      "epoch": 0.14117715777577397,
+      "grad_norm": 11.089734077453613,
+      "learning_rate": 9.627468892566186e-06,
+      "loss": 2.6556,
+      "step": 541400
+    },
+    {
+      "epoch": 0.14122931040147613,
+      "grad_norm": 9.867591857910156,
+      "learning_rate": 9.627152208731765e-06,
+      "loss": 2.6004,
+      "step": 541600
+    },
+    {
+      "epoch": 0.1412814630271783,
+      "grad_norm": 9.480960845947266,
+      "learning_rate": 9.626835395563104e-06,
+      "loss": 2.6388,
+      "step": 541800
+    },
+    {
+      "epoch": 0.14133361565288047,
+      "grad_norm": 9.904638290405273,
+      "learning_rate": 9.62651845306906e-06,
+      "loss": 2.6437,
+      "step": 542000
+    },
+    {
+      "epoch": 0.14138576827858262,
+      "grad_norm": 10.75538158416748,
+      "learning_rate": 9.626201381258488e-06,
+      "loss": 2.6249,
+      "step": 542200
+    },
+    {
+      "epoch": 0.1414379209042848,
+      "grad_norm": 9.948746681213379,
+      "learning_rate": 9.625884180140255e-06,
+      "loss": 2.6459,
+      "step": 542400
+    },
+    {
+      "epoch": 0.14149007352998696,
+      "grad_norm": 8.870070457458496,
+      "learning_rate": 9.625566849723227e-06,
+      "loss": 2.6629,
+      "step": 542600
+    },
+    {
+      "epoch": 0.14154222615568915,
+      "grad_norm": 10.196480751037598,
+      "learning_rate": 9.62524939001627e-06,
+      "loss": 2.6402,
+      "step": 542800
+    },
+    {
+      "epoch": 0.1415943787813913,
+      "grad_norm": 9.863924026489258,
+      "learning_rate": 9.624931801028258e-06,
+      "loss": 2.6385,
+      "step": 543000
+    },
+    {
+      "epoch": 0.1416465314070935,
+      "grad_norm": 9.387092590332031,
+      "learning_rate": 9.624614082768073e-06,
+      "loss": 2.6503,
+      "step": 543200
+    },
+    {
+      "epoch": 0.14169868403279565,
+      "grad_norm": 11.089527130126953,
+      "learning_rate": 9.62429623524459e-06,
+      "loss": 2.6537,
+      "step": 543400
+    },
+    {
+      "epoch": 0.14175083665849783,
+      "grad_norm": 10.329263687133789,
+      "learning_rate": 9.623978258466696e-06,
+      "loss": 2.6778,
+      "step": 543600
+    },
+    {
+      "epoch": 0.1418029892842,
+      "grad_norm": 10.49349308013916,
+      "learning_rate": 9.623660152443277e-06,
+      "loss": 2.6065,
+      "step": 543800
+    },
+    {
+      "epoch": 0.14185514190990217,
+      "grad_norm": 9.694085121154785,
+      "learning_rate": 9.623341917183227e-06,
+      "loss": 2.627,
+      "step": 544000
+    },
+    {
+      "epoch": 0.14190729453560433,
+      "grad_norm": 10.875962257385254,
+      "learning_rate": 9.623023552695438e-06,
+      "loss": 2.653,
+      "step": 544200
+    },
+    {
+      "epoch": 0.1419594471613065,
+      "grad_norm": 8.842427253723145,
+      "learning_rate": 9.62270505898881e-06,
+      "loss": 2.6317,
+      "step": 544400
+    },
+    {
+      "epoch": 0.14201159978700867,
+      "grad_norm": 9.090118408203125,
+      "learning_rate": 9.622386436072246e-06,
+      "loss": 2.6197,
+      "step": 544600
+    },
+    {
+      "epoch": 0.14206375241271085,
+      "grad_norm": 10.754172325134277,
+      "learning_rate": 9.622067683954651e-06,
+      "loss": 2.5924,
+      "step": 544800
+    },
+    {
+      "epoch": 0.142115905038413,
+      "grad_norm": 9.412361145019531,
+      "learning_rate": 9.621748802644934e-06,
+      "loss": 2.6484,
+      "step": 545000
+    },
+    {
+      "epoch": 0.1421680576641152,
+      "grad_norm": 10.36867618560791,
+      "learning_rate": 9.62142979215201e-06,
+      "loss": 2.6376,
+      "step": 545200
+    },
+    {
+      "epoch": 0.14222021028981735,
+      "grad_norm": 9.003300666809082,
+      "learning_rate": 9.621110652484794e-06,
+      "loss": 2.6281,
+      "step": 545400
+    },
+    {
+      "epoch": 0.14227236291551953,
+      "grad_norm": 11.816666603088379,
+      "learning_rate": 9.62079138365221e-06,
+      "loss": 2.6324,
+      "step": 545600
+    },
+    {
+      "epoch": 0.1423245155412217,
+      "grad_norm": 9.261345863342285,
+      "learning_rate": 9.620471985663175e-06,
+      "loss": 2.6267,
+      "step": 545800
+    },
+    {
+      "epoch": 0.14237666816692388,
+      "grad_norm": 8.67002010345459,
+      "learning_rate": 9.620152458526622e-06,
+      "loss": 2.6205,
+      "step": 546000
+    },
+    {
+      "epoch": 0.14242882079262603,
+      "grad_norm": 10.905217170715332,
+      "learning_rate": 9.61983280225148e-06,
+      "loss": 2.6096,
+      "step": 546200
+    },
+    {
+      "epoch": 0.14248097341832822,
+      "grad_norm": 10.862191200256348,
+      "learning_rate": 9.619513016846685e-06,
+      "loss": 2.6368,
+      "step": 546400
+    },
+    {
+      "epoch": 0.14253312604403037,
+      "grad_norm": 9.508679389953613,
+      "learning_rate": 9.619193102321174e-06,
+      "loss": 2.6533,
+      "step": 546600
+    },
+    {
+      "epoch": 0.14258527866973256,
+      "grad_norm": 9.74131965637207,
+      "learning_rate": 9.61887305868389e-06,
+      "loss": 2.6502,
+      "step": 546800
+    },
+    {
+      "epoch": 0.1426374312954347,
+      "grad_norm": 9.636917114257812,
+      "learning_rate": 9.618552885943777e-06,
+      "loss": 2.6608,
+      "step": 547000
+    },
+    {
+      "epoch": 0.1426895839211369,
+      "grad_norm": 9.495331764221191,
+      "learning_rate": 9.618232584109788e-06,
+      "loss": 2.6519,
+      "step": 547200
+    },
+    {
+      "epoch": 0.14274173654683905,
+      "grad_norm": 9.559065818786621,
+      "learning_rate": 9.617912153190871e-06,
+      "loss": 2.6379,
+      "step": 547400
+    },
+    {
+      "epoch": 0.14279388917254124,
+      "grad_norm": 9.602763175964355,
+      "learning_rate": 9.617591593195987e-06,
+      "loss": 2.6304,
+      "step": 547600
+    },
+    {
+      "epoch": 0.1428460417982434,
+      "grad_norm": 9.828039169311523,
+      "learning_rate": 9.617270904134091e-06,
+      "loss": 2.6251,
+      "step": 547800
+    },
+    {
+      "epoch": 0.14289819442394555,
+      "grad_norm": 9.119592666625977,
+      "learning_rate": 9.616950086014152e-06,
+      "loss": 2.6231,
+      "step": 548000
+    },
+    {
+      "epoch": 0.14295034704964774,
+      "grad_norm": 10.890464782714844,
+      "learning_rate": 9.616629138845132e-06,
+      "loss": 2.649,
+      "step": 548200
+    },
+    {
+      "epoch": 0.1430024996753499,
+      "grad_norm": 8.780187606811523,
+      "learning_rate": 9.616308062636006e-06,
+      "loss": 2.6794,
+      "step": 548400
+    },
+    {
+      "epoch": 0.14305465230105208,
+      "grad_norm": 9.78468132019043,
+      "learning_rate": 9.615986857395746e-06,
+      "loss": 2.6462,
+      "step": 548600
+    },
+    {
+      "epoch": 0.14310680492675423,
+      "grad_norm": 8.558454513549805,
+      "learning_rate": 9.615665523133331e-06,
+      "loss": 2.6206,
+      "step": 548800
+    },
+    {
+      "epoch": 0.14315895755245642,
+      "grad_norm": 8.706876754760742,
+      "learning_rate": 9.615344059857743e-06,
+      "loss": 2.6301,
+      "step": 549000
+    },
+    {
+      "epoch": 0.14321111017815857,
+      "grad_norm": 9.68352222442627,
+      "learning_rate": 9.615022467577966e-06,
+      "loss": 2.6818,
+      "step": 549200
+    },
+    {
+      "epoch": 0.14326326280386076,
+      "grad_norm": 11.24920654296875,
+      "learning_rate": 9.614700746302991e-06,
+      "loss": 2.6596,
+      "step": 549400
+    },
+    {
+      "epoch": 0.14331541542956291,
+      "grad_norm": 9.429580688476562,
+      "learning_rate": 9.614378896041808e-06,
+      "loss": 2.6525,
+      "step": 549600
+    },
+    {
+      "epoch": 0.1433675680552651,
+      "grad_norm": 10.127577781677246,
+      "learning_rate": 9.614056916803415e-06,
+      "loss": 2.6228,
+      "step": 549800
+    },
+    {
+      "epoch": 0.14341972068096726,
+      "grad_norm": 10.32852840423584,
+      "learning_rate": 9.61373480859681e-06,
+      "loss": 2.6263,
+      "step": 550000
+    },
+    {
+      "epoch": 0.14347187330666944,
+      "grad_norm": 10.721578598022461,
+      "learning_rate": 9.613412571430998e-06,
+      "loss": 2.6296,
+      "step": 550200
+    },
+    {
+      "epoch": 0.1435240259323716,
+      "grad_norm": 8.454923629760742,
+      "learning_rate": 9.613090205314985e-06,
+      "loss": 2.656,
+      "step": 550400
+    },
+    {
+      "epoch": 0.14357617855807378,
+      "grad_norm": 9.832836151123047,
+      "learning_rate": 9.612767710257782e-06,
+      "loss": 2.6442,
+      "step": 550600
+    },
+    {
+      "epoch": 0.14362833118377594,
+      "grad_norm": 9.232677459716797,
+      "learning_rate": 9.612445086268403e-06,
+      "loss": 2.6341,
+      "step": 550800
+    },
+    {
+      "epoch": 0.14368048380947812,
+      "grad_norm": 11.069082260131836,
+      "learning_rate": 9.612122333355865e-06,
+      "loss": 2.6251,
+      "step": 551000
+    },
+    {
+      "epoch": 0.14373263643518028,
+      "grad_norm": 10.895528793334961,
+      "learning_rate": 9.611799451529189e-06,
+      "loss": 2.6328,
+      "step": 551200
+    },
+    {
+      "epoch": 0.14378478906088246,
+      "grad_norm": 9.884492874145508,
+      "learning_rate": 9.611476440797403e-06,
+      "loss": 2.5978,
+      "step": 551400
+    },
+    {
+      "epoch": 0.14383694168658462,
+      "grad_norm": 9.460986137390137,
+      "learning_rate": 9.61115330116953e-06,
+      "loss": 2.6392,
+      "step": 551600
+    },
+    {
+      "epoch": 0.1438890943122868,
+      "grad_norm": 9.917821884155273,
+      "learning_rate": 9.610830032654607e-06,
+      "loss": 2.6414,
+      "step": 551800
+    },
+    {
+      "epoch": 0.14394124693798896,
+      "grad_norm": 8.92539119720459,
+      "learning_rate": 9.61050663526167e-06,
+      "loss": 2.606,
+      "step": 552000
+    },
+    {
+      "epoch": 0.14399339956369114,
+      "grad_norm": 10.357989311218262,
+      "learning_rate": 9.610183108999754e-06,
+      "loss": 2.5919,
+      "step": 552200
+    },
+    {
+      "epoch": 0.1440455521893933,
+      "grad_norm": 11.38748550415039,
+      "learning_rate": 9.609859453877906e-06,
+      "loss": 2.6614,
+      "step": 552400
+    },
+    {
+      "epoch": 0.14409770481509548,
+      "grad_norm": 8.538469314575195,
+      "learning_rate": 9.60953566990517e-06,
+      "loss": 2.6751,
+      "step": 552600
+    },
+    {
+      "epoch": 0.14414985744079764,
+      "grad_norm": 10.609665870666504,
+      "learning_rate": 9.609211757090596e-06,
+      "loss": 2.6102,
+      "step": 552800
+    },
+    {
+      "epoch": 0.14420201006649983,
+      "grad_norm": 9.466772079467773,
+      "learning_rate": 9.60888771544324e-06,
+      "loss": 2.6074,
+      "step": 553000
+    },
+    {
+      "epoch": 0.14425416269220198,
+      "grad_norm": 10.574810981750488,
+      "learning_rate": 9.608563544972159e-06,
+      "loss": 2.6195,
+      "step": 553200
+    },
+    {
+      "epoch": 0.14430631531790417,
+      "grad_norm": 8.984042167663574,
+      "learning_rate": 9.60823924568641e-06,
+      "loss": 2.6586,
+      "step": 553400
+    },
+    {
+      "epoch": 0.14435846794360632,
+      "grad_norm": 9.159278869628906,
+      "learning_rate": 9.607914817595062e-06,
+      "loss": 2.5936,
+      "step": 553600
+    },
+    {
+      "epoch": 0.1444106205693085,
+      "grad_norm": 10.08705997467041,
+      "learning_rate": 9.607590260707182e-06,
+      "loss": 2.6402,
+      "step": 553800
+    },
+    {
+      "epoch": 0.14446277319501066,
+      "grad_norm": 9.373217582702637,
+      "learning_rate": 9.607265575031843e-06,
+      "loss": 2.6356,
+      "step": 554000
+    },
+    {
+      "epoch": 0.14451492582071282,
+      "grad_norm": 9.303369522094727,
+      "learning_rate": 9.606940760578118e-06,
+      "loss": 2.6004,
+      "step": 554200
+    },
+    {
+      "epoch": 0.144567078446415,
+      "grad_norm": 10.792057991027832,
+      "learning_rate": 9.606615817355084e-06,
+      "loss": 2.619,
+      "step": 554400
+    },
+    {
+      "epoch": 0.14461923107211716,
+      "grad_norm": 9.816146850585938,
+      "learning_rate": 9.60629074537183e-06,
+      "loss": 2.5744,
+      "step": 554600
+    },
+    {
+      "epoch": 0.14467138369781934,
+      "grad_norm": 9.5195894241333,
+      "learning_rate": 9.605965544637437e-06,
+      "loss": 2.6646,
+      "step": 554800
+    },
+    {
+      "epoch": 0.1447235363235215,
+      "grad_norm": 9.027356147766113,
+      "learning_rate": 9.605640215160996e-06,
+      "loss": 2.5848,
+      "step": 555000
+    },
+    {
+      "epoch": 0.14477568894922369,
+      "grad_norm": 9.522998809814453,
+      "learning_rate": 9.6053147569516e-06,
+      "loss": 2.631,
+      "step": 555200
+    },
+    {
+      "epoch": 0.14482784157492584,
+      "grad_norm": 10.631654739379883,
+      "learning_rate": 9.604989170018347e-06,
+      "loss": 2.6315,
+      "step": 555400
+    },
+    {
+      "epoch": 0.14487999420062803,
+      "grad_norm": 9.056488037109375,
+      "learning_rate": 9.604663454370338e-06,
+      "loss": 2.5954,
+      "step": 555600
+    },
+    {
+      "epoch": 0.14493214682633018,
+      "grad_norm": 8.656085014343262,
+      "learning_rate": 9.604337610016674e-06,
+      "loss": 2.619,
+      "step": 555800
+    },
+    {
+      "epoch": 0.14498429945203237,
+      "grad_norm": 10.284260749816895,
+      "learning_rate": 9.604011636966466e-06,
+      "loss": 2.6313,
+      "step": 556000
+    },
+    {
+      "epoch": 0.14503645207773452,
+      "grad_norm": 8.677928924560547,
+      "learning_rate": 9.603685535228823e-06,
+      "loss": 2.6198,
+      "step": 556200
+    },
+    {
+      "epoch": 0.1450886047034367,
+      "grad_norm": 9.978887557983398,
+      "learning_rate": 9.603359304812863e-06,
+      "loss": 2.5997,
+      "step": 556400
+    },
+    {
+      "epoch": 0.14514075732913886,
+      "grad_norm": 10.44409465789795,
+      "learning_rate": 9.6030329457277e-06,
+      "loss": 2.6256,
+      "step": 556600
+    },
+    {
+      "epoch": 0.14519290995484105,
+      "grad_norm": 10.068913459777832,
+      "learning_rate": 9.60270645798246e-06,
+      "loss": 2.5958,
+      "step": 556800
+    },
+    {
+      "epoch": 0.1452450625805432,
+      "grad_norm": 9.706381797790527,
+      "learning_rate": 9.602379841586269e-06,
+      "loss": 2.5839,
+      "step": 557000
+    },
+    {
+      "epoch": 0.1452972152062454,
+      "grad_norm": 8.250496864318848,
+      "learning_rate": 9.602053096548251e-06,
+      "loss": 2.6462,
+      "step": 557200
+    },
+    {
+      "epoch": 0.14534936783194755,
+      "grad_norm": 10.653825759887695,
+      "learning_rate": 9.601726222877546e-06,
+      "loss": 2.6536,
+      "step": 557400
+    },
+    {
+      "epoch": 0.14540152045764973,
+      "grad_norm": 9.490010261535645,
+      "learning_rate": 9.601399220583285e-06,
+      "loss": 2.6482,
+      "step": 557600
+    },
+    {
+      "epoch": 0.1454536730833519,
+      "grad_norm": 9.792756080627441,
+      "learning_rate": 9.601072089674613e-06,
+      "loss": 2.6215,
+      "step": 557800
+    },
+    {
+      "epoch": 0.14550582570905407,
+      "grad_norm": 11.322357177734375,
+      "learning_rate": 9.600744830160667e-06,
+      "loss": 2.6084,
+      "step": 558000
+    },
+    {
+      "epoch": 0.14555797833475623,
+      "grad_norm": 10.1467924118042,
+      "learning_rate": 9.6004174420506e-06,
+      "loss": 2.6443,
+      "step": 558200
+    },
+    {
+      "epoch": 0.1456101309604584,
+      "grad_norm": 10.210265159606934,
+      "learning_rate": 9.600089925353562e-06,
+      "loss": 2.6163,
+      "step": 558400
+    },
+    {
+      "epoch": 0.14566228358616057,
+      "grad_norm": 12.064213752746582,
+      "learning_rate": 9.599762280078705e-06,
+      "loss": 2.6455,
+      "step": 558600
+    },
+    {
+      "epoch": 0.14571443621186275,
+      "grad_norm": 8.971247673034668,
+      "learning_rate": 9.599434506235188e-06,
+      "loss": 2.6323,
+      "step": 558800
+    },
+    {
+      "epoch": 0.1457665888375649,
+      "grad_norm": 9.186264991760254,
+      "learning_rate": 9.599106603832176e-06,
+      "loss": 2.6271,
+      "step": 559000
+    },
+    {
+      "epoch": 0.1458187414632671,
+      "grad_norm": 10.96983814239502,
+      "learning_rate": 9.598778572878828e-06,
+      "loss": 2.6766,
+      "step": 559200
+    },
+    {
+      "epoch": 0.14587089408896925,
+      "grad_norm": 9.236897468566895,
+      "learning_rate": 9.59845041338432e-06,
+      "loss": 2.6698,
+      "step": 559400
+    },
+    {
+      "epoch": 0.14592304671467143,
+      "grad_norm": 9.946492195129395,
+      "learning_rate": 9.598122125357817e-06,
+      "loss": 2.6006,
+      "step": 559600
+    },
+    {
+      "epoch": 0.1459751993403736,
+      "grad_norm": 10.890714645385742,
+      "learning_rate": 9.597793708808501e-06,
+      "loss": 2.6319,
+      "step": 559800
+    },
+    {
+      "epoch": 0.14602735196607575,
+      "grad_norm": 10.893404006958008,
+      "learning_rate": 9.597465163745548e-06,
+      "loss": 2.6361,
+      "step": 560000
+    },
+    {
+      "epoch": 0.14607950459177793,
+      "grad_norm": 10.29994010925293,
+      "learning_rate": 9.597136490178145e-06,
+      "loss": 2.6443,
+      "step": 560200
+    },
+    {
+      "epoch": 0.1461316572174801,
+      "grad_norm": 10.262697219848633,
+      "learning_rate": 9.596807688115474e-06,
+      "loss": 2.6291,
+      "step": 560400
+    },
+    {
+      "epoch": 0.14618380984318227,
+      "grad_norm": 10.052336692810059,
+      "learning_rate": 9.596478757566729e-06,
+      "loss": 2.6116,
+      "step": 560600
+    },
+    {
+      "epoch": 0.14623596246888443,
+      "grad_norm": 9.586968421936035,
+      "learning_rate": 9.596149698541102e-06,
+      "loss": 2.6361,
+      "step": 560800
+    },
+    {
+      "epoch": 0.1462881150945866,
+      "grad_norm": 10.506963729858398,
+      "learning_rate": 9.595820511047791e-06,
+      "loss": 2.6558,
+      "step": 561000
+    },
+    {
+      "epoch": 0.14634026772028877,
+      "grad_norm": 11.302355766296387,
+      "learning_rate": 9.595491195095998e-06,
+      "loss": 2.6385,
+      "step": 561200
+    },
+    {
+      "epoch": 0.14639242034599095,
+      "grad_norm": 10.142350196838379,
+      "learning_rate": 9.595161750694927e-06,
+      "loss": 2.6049,
+      "step": 561400
+    },
+    {
+      "epoch": 0.1464445729716931,
+      "grad_norm": 8.8736572265625,
+      "learning_rate": 9.594832177853787e-06,
+      "loss": 2.606,
+      "step": 561600
+    },
+    {
+      "epoch": 0.1464967255973953,
+      "grad_norm": 10.957189559936523,
+      "learning_rate": 9.59450247658179e-06,
+      "loss": 2.6365,
+      "step": 561800
+    },
+    {
+      "epoch": 0.14654887822309745,
+      "grad_norm": 10.375482559204102,
+      "learning_rate": 9.594172646888151e-06,
+      "loss": 2.6283,
+      "step": 562000
+    },
+    {
+      "epoch": 0.14660103084879963,
+      "grad_norm": 10.932716369628906,
+      "learning_rate": 9.59384268878209e-06,
+      "loss": 2.658,
+      "step": 562200
+    },
+    {
+      "epoch": 0.1466531834745018,
+      "grad_norm": 9.20926284790039,
+      "learning_rate": 9.593512602272828e-06,
+      "loss": 2.6566,
+      "step": 562400
+    },
+    {
+      "epoch": 0.14670533610020398,
+      "grad_norm": 10.561077117919922,
+      "learning_rate": 9.593182387369592e-06,
+      "loss": 2.6381,
+      "step": 562600
+    },
+    {
+      "epoch": 0.14675748872590613,
+      "grad_norm": 9.518685340881348,
+      "learning_rate": 9.592852044081614e-06,
+      "loss": 2.5938,
+      "step": 562800
+    },
+    {
+      "epoch": 0.14680964135160832,
+      "grad_norm": 10.072381973266602,
+      "learning_rate": 9.592521572418123e-06,
+      "loss": 2.665,
+      "step": 563000
+    },
+    {
+      "epoch": 0.14686179397731047,
+      "grad_norm": 10.22624683380127,
+      "learning_rate": 9.59219097238836e-06,
+      "loss": 2.563,
+      "step": 563200
+    },
+    {
+      "epoch": 0.14691394660301266,
+      "grad_norm": 10.61490249633789,
+      "learning_rate": 9.591860244001563e-06,
+      "loss": 2.6146,
+      "step": 563400
+    },
+    {
+      "epoch": 0.1469660992287148,
+      "grad_norm": 9.314760208129883,
+      "learning_rate": 9.59152938726698e-06,
+      "loss": 2.6327,
+      "step": 563600
+    },
+    {
+      "epoch": 0.147018251854417,
+      "grad_norm": 10.206314086914062,
+      "learning_rate": 9.591198402193854e-06,
+      "loss": 2.6513,
+      "step": 563800
+    },
+    {
+      "epoch": 0.14707040448011915,
+      "grad_norm": 11.540358543395996,
+      "learning_rate": 9.59086728879144e-06,
+      "loss": 2.6168,
+      "step": 564000
+    },
+    {
+      "epoch": 0.14712255710582134,
+      "grad_norm": 10.943085670471191,
+      "learning_rate": 9.590536047068992e-06,
+      "loss": 2.6184,
+      "step": 564200
+    },
+    {
+      "epoch": 0.1471747097315235,
+      "grad_norm": 11.087271690368652,
+      "learning_rate": 9.590204677035769e-06,
+      "loss": 2.6272,
+      "step": 564400
+    },
+    {
+      "epoch": 0.14722686235722568,
+      "grad_norm": 10.729355812072754,
+      "learning_rate": 9.58987317870103e-06,
+      "loss": 2.6478,
+      "step": 564600
+    },
+    {
+      "epoch": 0.14727901498292784,
+      "grad_norm": 10.2618408203125,
+      "learning_rate": 9.589541552074044e-06,
+      "loss": 2.6489,
+      "step": 564800
+    },
+    {
+      "epoch": 0.14733116760863002,
+      "grad_norm": 8.719612121582031,
+      "learning_rate": 9.589209797164082e-06,
+      "loss": 2.6471,
+      "step": 565000
+    },
+    {
+      "epoch": 0.14738332023433218,
+      "grad_norm": 9.179375648498535,
+      "learning_rate": 9.588877913980411e-06,
+      "loss": 2.6204,
+      "step": 565200
+    },
+    {
+      "epoch": 0.14743547286003436,
+      "grad_norm": 8.887311935424805,
+      "learning_rate": 9.588545902532316e-06,
+      "loss": 2.627,
+      "step": 565400
+    },
+    {
+      "epoch": 0.14748762548573652,
+      "grad_norm": 10.409414291381836,
+      "learning_rate": 9.588213762829068e-06,
+      "loss": 2.6492,
+      "step": 565600
+    },
+    {
+      "epoch": 0.14753977811143867,
+      "grad_norm": 9.475335121154785,
+      "learning_rate": 9.587881494879956e-06,
+      "loss": 2.6261,
+      "step": 565800
+    },
+    {
+      "epoch": 0.14759193073714086,
+      "grad_norm": 10.67691707611084,
+      "learning_rate": 9.587549098694268e-06,
+      "loss": 2.6279,
+      "step": 566000
+    },
+    {
+      "epoch": 0.14764408336284301,
+      "grad_norm": 9.022562980651855,
+      "learning_rate": 9.58721657428129e-06,
+      "loss": 2.6222,
+      "step": 566200
+    },
+    {
+      "epoch": 0.1476962359885452,
+      "grad_norm": 9.80457878112793,
+      "learning_rate": 9.586883921650322e-06,
+      "loss": 2.6257,
+      "step": 566400
+    },
+    {
+      "epoch": 0.14774838861424736,
+      "grad_norm": 10.465949058532715,
+      "learning_rate": 9.586551140810659e-06,
+      "loss": 2.6321,
+      "step": 566600
+    },
+    {
+      "epoch": 0.14780054123994954,
+      "grad_norm": 8.894414901733398,
+      "learning_rate": 9.586218231771602e-06,
+      "loss": 2.6289,
+      "step": 566800
+    },
+    {
+      "epoch": 0.1478526938656517,
+      "grad_norm": 9.935043334960938,
+      "learning_rate": 9.58588519454246e-06,
+      "loss": 2.6112,
+      "step": 567000
+    },
+    {
+      "epoch": 0.14790484649135388,
+      "grad_norm": 10.814425468444824,
+      "learning_rate": 9.585552029132537e-06,
+      "loss": 2.6257,
+      "step": 567200
+    },
+    {
+      "epoch": 0.14795699911705604,
+      "grad_norm": 10.175069808959961,
+      "learning_rate": 9.585218735551147e-06,
+      "loss": 2.6417,
+      "step": 567400
+    },
+    {
+      "epoch": 0.14800915174275822,
+      "grad_norm": 9.798264503479004,
+      "learning_rate": 9.584885313807607e-06,
+      "loss": 2.6016,
+      "step": 567600
+    },
+    {
+      "epoch": 0.14806130436846038,
+      "grad_norm": 9.588288307189941,
+      "learning_rate": 9.584551763911236e-06,
+      "loss": 2.6012,
+      "step": 567800
+    },
+    {
+      "epoch": 0.14811345699416256,
+      "grad_norm": 9.886007308959961,
+      "learning_rate": 9.584218085871358e-06,
+      "loss": 2.6357,
+      "step": 568000
+    },
+    {
+      "epoch": 0.14816560961986472,
+      "grad_norm": 9.629053115844727,
+      "learning_rate": 9.583884279697297e-06,
+      "loss": 2.6257,
+      "step": 568200
+    },
+    {
+      "epoch": 0.1482177622455669,
+      "grad_norm": 9.915539741516113,
+      "learning_rate": 9.583550345398385e-06,
+      "loss": 2.6651,
+      "step": 568400
+    },
+    {
+      "epoch": 0.14826991487126906,
+      "grad_norm": 10.119706153869629,
+      "learning_rate": 9.583216282983955e-06,
+      "loss": 2.6624,
+      "step": 568600
+    },
+    {
+      "epoch": 0.14832206749697124,
+      "grad_norm": 10.616450309753418,
+      "learning_rate": 9.582882092463348e-06,
+      "loss": 2.6166,
+      "step": 568800
+    },
+    {
+      "epoch": 0.1483742201226734,
+      "grad_norm": 10.009668350219727,
+      "learning_rate": 9.582547773845901e-06,
+      "loss": 2.6372,
+      "step": 569000
+    },
+    {
+      "epoch": 0.14842637274837558,
+      "grad_norm": 9.56704330444336,
+      "learning_rate": 9.582213327140958e-06,
+      "loss": 2.648,
+      "step": 569200
+    },
+    {
+      "epoch": 0.14847852537407774,
+      "grad_norm": 12.643216133117676,
+      "learning_rate": 9.581878752357871e-06,
+      "loss": 2.6362,
+      "step": 569400
+    },
+    {
+      "epoch": 0.14853067799977993,
+      "grad_norm": 9.96658992767334,
+      "learning_rate": 9.581544049505991e-06,
+      "loss": 2.6044,
+      "step": 569600
+    },
+    {
+      "epoch": 0.14858283062548208,
+      "grad_norm": 9.937406539916992,
+      "learning_rate": 9.58120921859467e-06,
+      "loss": 2.6604,
+      "step": 569800
+    },
+    {
+      "epoch": 0.14863498325118427,
+      "grad_norm": 10.475345611572266,
+      "learning_rate": 9.58087425963327e-06,
+      "loss": 2.6185,
+      "step": 570000
+    },
+    {
+      "epoch": 0.14868713587688642,
+      "grad_norm": 9.268296241760254,
+      "learning_rate": 9.580539172631152e-06,
+      "loss": 2.6081,
+      "step": 570200
+    },
+    {
+      "epoch": 0.1487392885025886,
+      "grad_norm": 9.733953475952148,
+      "learning_rate": 9.580203957597683e-06,
+      "loss": 2.6293,
+      "step": 570400
+    },
+    {
+      "epoch": 0.14879144112829076,
+      "grad_norm": 10.396161079406738,
+      "learning_rate": 9.579868614542231e-06,
+      "loss": 2.6485,
+      "step": 570600
+    },
+    {
+      "epoch": 0.14884359375399295,
+      "grad_norm": 10.291255950927734,
+      "learning_rate": 9.579533143474172e-06,
+      "loss": 2.6246,
+      "step": 570800
+    },
+    {
+      "epoch": 0.1488957463796951,
+      "grad_norm": 10.212169647216797,
+      "learning_rate": 9.579197544402881e-06,
+      "loss": 2.5966,
+      "step": 571000
+    },
+    {
+      "epoch": 0.1489478990053973,
+      "grad_norm": 9.417984962463379,
+      "learning_rate": 9.578861817337736e-06,
+      "loss": 2.628,
+      "step": 571200
+    },
+    {
+      "epoch": 0.14900005163109944,
+      "grad_norm": 11.286150932312012,
+      "learning_rate": 9.578525962288127e-06,
+      "loss": 2.6472,
+      "step": 571400
+    },
+    {
+      "epoch": 0.1490522042568016,
+      "grad_norm": 9.67978286743164,
+      "learning_rate": 9.578189979263438e-06,
+      "loss": 2.6192,
+      "step": 571600
+    },
+    {
+      "epoch": 0.14910435688250379,
+      "grad_norm": 10.592140197753906,
+      "learning_rate": 9.577853868273057e-06,
+      "loss": 2.6086,
+      "step": 571800
+    },
+    {
+      "epoch": 0.14915650950820594,
+      "grad_norm": 10.959556579589844,
+      "learning_rate": 9.577517629326385e-06,
+      "loss": 2.5899,
+      "step": 572000
+    },
+    {
+      "epoch": 0.14920866213390813,
+      "grad_norm": 11.056251525878906,
+      "learning_rate": 9.577181262432815e-06,
+      "loss": 2.655,
+      "step": 572200
+    },
+    {
+      "epoch": 0.14926081475961028,
+      "grad_norm": 10.780267715454102,
+      "learning_rate": 9.576844767601753e-06,
+      "loss": 2.6094,
+      "step": 572400
+    },
+    {
+      "epoch": 0.14931296738531247,
+      "grad_norm": 10.068269729614258,
+      "learning_rate": 9.576508144842603e-06,
+      "loss": 2.6275,
+      "step": 572600
+    },
+    {
+      "epoch": 0.14936512001101462,
+      "grad_norm": 10.287019729614258,
+      "learning_rate": 9.57617139416477e-06,
+      "loss": 2.6343,
+      "step": 572800
+    },
+    {
+      "epoch": 0.1494172726367168,
+      "grad_norm": 8.428248405456543,
+      "learning_rate": 9.575834515577673e-06,
+      "loss": 2.597,
+      "step": 573000
+    },
+    {
+      "epoch": 0.14946942526241896,
+      "grad_norm": 10.998594284057617,
+      "learning_rate": 9.575497509090723e-06,
+      "loss": 2.6543,
+      "step": 573200
+    },
+    {
+      "epoch": 0.14952157788812115,
+      "grad_norm": 9.181807518005371,
+      "learning_rate": 9.575160374713344e-06,
+      "loss": 2.6463,
+      "step": 573400
+    },
+    {
+      "epoch": 0.1495737305138233,
+      "grad_norm": 10.688831329345703,
+      "learning_rate": 9.574823112454957e-06,
+      "loss": 2.6277,
+      "step": 573600
+    },
+    {
+      "epoch": 0.1496258831395255,
+      "grad_norm": 9.806432723999023,
+      "learning_rate": 9.574485722324986e-06,
+      "loss": 2.6082,
+      "step": 573800
+    },
+    {
+      "epoch": 0.14967803576522765,
+      "grad_norm": 10.59438419342041,
+      "learning_rate": 9.574148204332868e-06,
+      "loss": 2.6137,
+      "step": 574000
+    },
+    {
+      "epoch": 0.14973018839092983,
+      "grad_norm": 11.886221885681152,
+      "learning_rate": 9.573810558488032e-06,
+      "loss": 2.6198,
+      "step": 574200
+    },
+    {
+      "epoch": 0.149782341016632,
+      "grad_norm": 9.97330379486084,
+      "learning_rate": 9.573472784799918e-06,
+      "loss": 2.6259,
+      "step": 574400
+    },
+    {
+      "epoch": 0.14983449364233417,
+      "grad_norm": 10.082623481750488,
+      "learning_rate": 9.573134883277966e-06,
+      "loss": 2.6735,
+      "step": 574600
+    },
+    {
+      "epoch": 0.14988664626803633,
+      "grad_norm": 10.873509407043457,
+      "learning_rate": 9.572796853931619e-06,
+      "loss": 2.6497,
+      "step": 574800
+    },
+    {
+      "epoch": 0.1499387988937385,
+      "grad_norm": 10.863423347473145,
+      "learning_rate": 9.57245869677033e-06,
+      "loss": 2.6382,
+      "step": 575000
+    },
+    {
+      "epoch": 0.14999095151944067,
+      "grad_norm": 9.81516170501709,
+      "learning_rate": 9.572120411803547e-06,
+      "loss": 2.5861,
+      "step": 575200
+    },
+    {
+      "epoch": 0.15004310414514285,
+      "grad_norm": 10.294231414794922,
+      "learning_rate": 9.571781999040726e-06,
+      "loss": 2.5945,
+      "step": 575400
+    },
+    {
+      "epoch": 0.150095256770845,
+      "grad_norm": 10.843868255615234,
+      "learning_rate": 9.571443458491327e-06,
+      "loss": 2.6269,
+      "step": 575600
+    },
+    {
+      "epoch": 0.1501474093965472,
+      "grad_norm": 8.99116039276123,
+      "learning_rate": 9.571104790164814e-06,
+      "loss": 2.6215,
+      "step": 575800
+    },
+    {
+      "epoch": 0.15019956202224935,
+      "grad_norm": 10.161559104919434,
+      "learning_rate": 9.57076599407065e-06,
+      "loss": 2.6052,
+      "step": 576000
+    },
+    {
+      "epoch": 0.15025171464795153,
+      "grad_norm": 9.422714233398438,
+      "learning_rate": 9.570427070218306e-06,
+      "loss": 2.6268,
+      "step": 576200
+    },
+    {
+      "epoch": 0.1503038672736537,
+      "grad_norm": 9.956077575683594,
+      "learning_rate": 9.570088018617255e-06,
+      "loss": 2.6241,
+      "step": 576400
+    },
+    {
+      "epoch": 0.15035601989935587,
+      "grad_norm": 10.455558776855469,
+      "learning_rate": 9.569748839276975e-06,
+      "loss": 2.6146,
+      "step": 576600
+    },
+    {
+      "epoch": 0.15040817252505803,
+      "grad_norm": 9.470061302185059,
+      "learning_rate": 9.569409532206945e-06,
+      "loss": 2.6122,
+      "step": 576800
+    },
+    {
+      "epoch": 0.15046032515076022,
+      "grad_norm": 9.468082427978516,
+      "learning_rate": 9.56907009741665e-06,
+      "loss": 2.6527,
+      "step": 577000
+    },
+    {
+      "epoch": 0.15051247777646237,
+      "grad_norm": 10.123863220214844,
+      "learning_rate": 9.56873053491558e-06,
+      "loss": 2.6252,
+      "step": 577200
+    },
+    {
+      "epoch": 0.15056463040216453,
+      "grad_norm": 9.473311424255371,
+      "learning_rate": 9.568390844713221e-06,
+      "loss": 2.622,
+      "step": 577400
+    },
+    {
+      "epoch": 0.1506167830278667,
+      "grad_norm": 10.261393547058105,
+      "learning_rate": 9.568051026819072e-06,
+      "loss": 2.6369,
+      "step": 577600
+    },
+    {
+      "epoch": 0.15066893565356887,
+      "grad_norm": 7.631354331970215,
+      "learning_rate": 9.567711081242628e-06,
+      "loss": 2.6039,
+      "step": 577800
+    },
+    {
+      "epoch": 0.15072108827927105,
+      "grad_norm": 11.26151180267334,
+      "learning_rate": 9.567371007993394e-06,
+      "loss": 2.6304,
+      "step": 578000
+    },
+    {
+      "epoch": 0.1507732409049732,
+      "grad_norm": 10.592456817626953,
+      "learning_rate": 9.567030807080872e-06,
+      "loss": 2.6539,
+      "step": 578200
+    },
+    {
+      "epoch": 0.1508253935306754,
+      "grad_norm": 9.679903030395508,
+      "learning_rate": 9.566690478514574e-06,
+      "loss": 2.621,
+      "step": 578400
+    },
+    {
+      "epoch": 0.15087754615637755,
+      "grad_norm": 9.416767120361328,
+      "learning_rate": 9.566350022304013e-06,
+      "loss": 2.6417,
+      "step": 578600
+    },
+    {
+      "epoch": 0.15092969878207974,
+      "grad_norm": 11.1919527053833,
+      "learning_rate": 9.566009438458701e-06,
+      "loss": 2.5929,
+      "step": 578800
+    },
+    {
+      "epoch": 0.1509818514077819,
+      "grad_norm": 10.748730659484863,
+      "learning_rate": 9.565668726988161e-06,
+      "loss": 2.6213,
+      "step": 579000
+    },
+    {
+      "epoch": 0.15103400403348408,
+      "grad_norm": 11.362808227539062,
+      "learning_rate": 9.565327887901918e-06,
+      "loss": 2.65,
+      "step": 579200
+    },
+    {
+      "epoch": 0.15108615665918623,
+      "grad_norm": 10.134300231933594,
+      "learning_rate": 9.564986921209493e-06,
+      "loss": 2.5866,
+      "step": 579400
+    },
+    {
+      "epoch": 0.15113830928488842,
+      "grad_norm": 10.209643363952637,
+      "learning_rate": 9.564645826920422e-06,
+      "loss": 2.5839,
+      "step": 579600
+    },
+    {
+      "epoch": 0.15119046191059057,
+      "grad_norm": 11.057905197143555,
+      "learning_rate": 9.564304605044239e-06,
+      "loss": 2.5859,
+      "step": 579800
+    },
+    {
+      "epoch": 0.15124261453629276,
+      "grad_norm": 11.382614135742188,
+      "learning_rate": 9.563963255590476e-06,
+      "loss": 2.5781,
+      "step": 580000
+    },
+    {
+      "epoch": 0.15129476716199491,
+      "grad_norm": 9.950427055358887,
+      "learning_rate": 9.563621778568679e-06,
+      "loss": 2.6076,
+      "step": 580200
+    },
+    {
+      "epoch": 0.1513469197876971,
+      "grad_norm": 9.879427909851074,
+      "learning_rate": 9.56328017398839e-06,
+      "loss": 2.5791,
+      "step": 580400
+    },
+    {
+      "epoch": 0.15139907241339925,
+      "grad_norm": 9.203298568725586,
+      "learning_rate": 9.56293844185916e-06,
+      "loss": 2.6405,
+      "step": 580600
+    },
+    {
+      "epoch": 0.15145122503910144,
+      "grad_norm": 10.701813697814941,
+      "learning_rate": 9.562596582190539e-06,
+      "loss": 2.6244,
+      "step": 580800
+    },
+    {
+      "epoch": 0.1515033776648036,
+      "grad_norm": 10.396395683288574,
+      "learning_rate": 9.562254594992082e-06,
+      "loss": 2.5892,
+      "step": 581000
+    },
+    {
+      "epoch": 0.15155553029050578,
+      "grad_norm": 8.901702880859375,
+      "learning_rate": 9.561912480273348e-06,
+      "loss": 2.631,
+      "step": 581200
+    },
+    {
+      "epoch": 0.15160768291620794,
+      "grad_norm": 10.96087646484375,
+      "learning_rate": 9.561570238043902e-06,
+      "loss": 2.6193,
+      "step": 581400
+    },
+    {
+      "epoch": 0.15165983554191012,
+      "grad_norm": 10.27599048614502,
+      "learning_rate": 9.561227868313306e-06,
+      "loss": 2.6072,
+      "step": 581600
+    },
+    {
+      "epoch": 0.15171198816761228,
+      "grad_norm": 9.896045684814453,
+      "learning_rate": 9.560885371091134e-06,
+      "loss": 2.6083,
+      "step": 581800
+    },
+    {
+      "epoch": 0.15176414079331446,
+      "grad_norm": 10.777628898620605,
+      "learning_rate": 9.560542746386955e-06,
+      "loss": 2.5971,
+      "step": 582000
+    },
+    {
+      "epoch": 0.15181629341901662,
+      "grad_norm": 9.133131980895996,
+      "learning_rate": 9.560199994210349e-06,
+      "loss": 2.6302,
+      "step": 582200
+    },
+    {
+      "epoch": 0.1518684460447188,
+      "grad_norm": 11.197638511657715,
+      "learning_rate": 9.559857114570893e-06,
+      "loss": 2.6054,
+      "step": 582400
+    },
+    {
+      "epoch": 0.15192059867042096,
+      "grad_norm": 10.832380294799805,
+      "learning_rate": 9.559514107478177e-06,
+      "loss": 2.5988,
+      "step": 582600
+    },
+    {
+      "epoch": 0.15197275129612314,
+      "grad_norm": 10.314582824707031,
+      "learning_rate": 9.559170972941782e-06,
+      "loss": 2.5965,
+      "step": 582800
+    },
+    {
+      "epoch": 0.1520249039218253,
+      "grad_norm": 11.280247688293457,
+      "learning_rate": 9.558827710971302e-06,
+      "loss": 2.631,
+      "step": 583000
+    },
+    {
+      "epoch": 0.15207705654752748,
+      "grad_norm": 10.295195579528809,
+      "learning_rate": 9.558484321576329e-06,
+      "loss": 2.6085,
+      "step": 583200
+    },
+    {
+      "epoch": 0.15212920917322964,
+      "grad_norm": 9.563957214355469,
+      "learning_rate": 9.558140804766464e-06,
+      "loss": 2.6147,
+      "step": 583400
+    },
+    {
+      "epoch": 0.1521813617989318,
+      "grad_norm": 10.042048454284668,
+      "learning_rate": 9.55779716055131e-06,
+      "loss": 2.6046,
+      "step": 583600
+    },
+    {
+      "epoch": 0.15223351442463398,
+      "grad_norm": 11.11385440826416,
+      "learning_rate": 9.557453388940468e-06,
+      "loss": 2.6468,
+      "step": 583800
+    },
+    {
+      "epoch": 0.15228566705033614,
+      "grad_norm": 10.20233154296875,
+      "learning_rate": 9.557109489943548e-06,
+      "loss": 2.6675,
+      "step": 584000
+    },
+    {
+      "epoch": 0.15233781967603832,
+      "grad_norm": 10.255865097045898,
+      "learning_rate": 9.556765463570164e-06,
+      "loss": 2.6439,
+      "step": 584200
+    },
+    {
+      "epoch": 0.15238997230174048,
+      "grad_norm": 9.48961067199707,
+      "learning_rate": 9.55642130982993e-06,
+      "loss": 2.6361,
+      "step": 584400
+    },
+    {
+      "epoch": 0.15244212492744266,
+      "grad_norm": 9.803009986877441,
+      "learning_rate": 9.55607702873247e-06,
+      "loss": 2.6297,
+      "step": 584600
+    },
+    {
+      "epoch": 0.15249427755314482,
+      "grad_norm": 10.121756553649902,
+      "learning_rate": 9.5557326202874e-06,
+      "loss": 2.61,
+      "step": 584800
+    },
+    {
+      "epoch": 0.152546430178847,
+      "grad_norm": 9.789412498474121,
+      "learning_rate": 9.555388084504353e-06,
+      "loss": 2.6089,
+      "step": 585000
+    },
+    {
+      "epoch": 0.15259858280454916,
+      "grad_norm": 10.218076705932617,
+      "learning_rate": 9.555043421392955e-06,
+      "loss": 2.5853,
+      "step": 585200
+    },
+    {
+      "epoch": 0.15265073543025134,
+      "grad_norm": 10.865063667297363,
+      "learning_rate": 9.554698630962841e-06,
+      "loss": 2.6124,
+      "step": 585400
+    },
+    {
+      "epoch": 0.1527028880559535,
+      "grad_norm": 10.640392303466797,
+      "learning_rate": 9.55435371322365e-06,
+      "loss": 2.6186,
+      "step": 585600
+    },
+    {
+      "epoch": 0.15275504068165568,
+      "grad_norm": 11.789724349975586,
+      "learning_rate": 9.55400866818502e-06,
+      "loss": 2.6226,
+      "step": 585800
+    },
+    {
+      "epoch": 0.15280719330735784,
+      "grad_norm": 8.585991859436035,
+      "learning_rate": 9.553663495856598e-06,
+      "loss": 2.6153,
+      "step": 586000
+    },
+    {
+      "epoch": 0.15285934593306003,
+      "grad_norm": 12.74191665649414,
+      "learning_rate": 9.55331819624803e-06,
+      "loss": 2.6273,
+      "step": 586200
+    },
+    {
+      "epoch": 0.15291149855876218,
+      "grad_norm": 11.765472412109375,
+      "learning_rate": 9.552972769368969e-06,
+      "loss": 2.6034,
+      "step": 586400
+    },
+    {
+      "epoch": 0.15296365118446437,
+      "grad_norm": 8.79220962524414,
+      "learning_rate": 9.552627215229067e-06,
+      "loss": 2.624,
+      "step": 586600
+    },
+    {
+      "epoch": 0.15301580381016652,
+      "grad_norm": 9.72270679473877,
+      "learning_rate": 9.552281533837988e-06,
+      "loss": 2.6228,
+      "step": 586800
+    },
+    {
+      "epoch": 0.1530679564358687,
+      "grad_norm": 11.258645057678223,
+      "learning_rate": 9.55193572520539e-06,
+      "loss": 2.6312,
+      "step": 587000
+    },
+    {
+      "epoch": 0.15312010906157086,
+      "grad_norm": 10.41986083984375,
+      "learning_rate": 9.551589789340939e-06,
+      "loss": 2.5924,
+      "step": 587200
+    },
+    {
+      "epoch": 0.15317226168727305,
+      "grad_norm": 10.305185317993164,
+      "learning_rate": 9.551243726254304e-06,
+      "loss": 2.6179,
+      "step": 587400
+    },
+    {
+      "epoch": 0.1532244143129752,
+      "grad_norm": 10.521088600158691,
+      "learning_rate": 9.550897535955161e-06,
+      "loss": 2.6227,
+      "step": 587600
+    },
+    {
+      "epoch": 0.1532765669386774,
+      "grad_norm": 10.343436241149902,
+      "learning_rate": 9.550551218453186e-06,
+      "loss": 2.6194,
+      "step": 587800
+    },
+    {
+      "epoch": 0.15332871956437955,
+      "grad_norm": 9.396683692932129,
+      "learning_rate": 9.550204773758055e-06,
+      "loss": 2.6268,
+      "step": 588000
+    },
+    {
+      "epoch": 0.15338087219008173,
+      "grad_norm": 11.622981071472168,
+      "learning_rate": 9.549858201879456e-06,
+      "loss": 2.6188,
+      "step": 588200
+    },
+    {
+      "epoch": 0.15343302481578389,
+      "grad_norm": 9.679468154907227,
+      "learning_rate": 9.549511502827071e-06,
+      "loss": 2.5841,
+      "step": 588400
+    },
+    {
+      "epoch": 0.15348517744148607,
+      "grad_norm": 10.439282417297363,
+      "learning_rate": 9.549164676610596e-06,
+      "loss": 2.626,
+      "step": 588600
+    },
+    {
+      "epoch": 0.15353733006718823,
+      "grad_norm": 10.602144241333008,
+      "learning_rate": 9.548817723239723e-06,
+      "loss": 2.6081,
+      "step": 588800
+    },
+    {
+      "epoch": 0.1535894826928904,
+      "grad_norm": 9.781011581420898,
+      "learning_rate": 9.548470642724148e-06,
+      "loss": 2.5835,
+      "step": 589000
+    },
+    {
+      "epoch": 0.15364163531859257,
+      "grad_norm": 10.308581352233887,
+      "learning_rate": 9.548123435073575e-06,
+      "loss": 2.6292,
+      "step": 589200
+    },
+    {
+      "epoch": 0.15369378794429472,
+      "grad_norm": 11.796847343444824,
+      "learning_rate": 9.547776100297708e-06,
+      "loss": 2.6315,
+      "step": 589400
+    },
+    {
+      "epoch": 0.1537459405699969,
+      "grad_norm": 10.120135307312012,
+      "learning_rate": 9.547428638406255e-06,
+      "loss": 2.6352,
+      "step": 589600
+    },
+    {
+      "epoch": 0.15379809319569906,
+      "grad_norm": 10.831207275390625,
+      "learning_rate": 9.547081049408928e-06,
+      "loss": 2.6174,
+      "step": 589800
+    },
+    {
+      "epoch": 0.15385024582140125,
+      "grad_norm": 9.786165237426758,
+      "learning_rate": 9.546733333315444e-06,
+      "loss": 2.6019,
+      "step": 590000
+    },
+    {
+      "epoch": 0.1539023984471034,
+      "grad_norm": 10.431953430175781,
+      "learning_rate": 9.54638549013552e-06,
+      "loss": 2.5921,
+      "step": 590200
+    },
+    {
+      "epoch": 0.1539545510728056,
+      "grad_norm": 10.351963996887207,
+      "learning_rate": 9.546037519878878e-06,
+      "loss": 2.6084,
+      "step": 590400
+    },
+    {
+      "epoch": 0.15400670369850775,
+      "grad_norm": 13.948197364807129,
+      "learning_rate": 9.545689422555246e-06,
+      "loss": 2.6002,
+      "step": 590600
+    },
+    {
+      "epoch": 0.15405885632420993,
+      "grad_norm": 9.773366928100586,
+      "learning_rate": 9.545341198174355e-06,
+      "loss": 2.6003,
+      "step": 590800
+    },
+    {
+      "epoch": 0.1541110089499121,
+      "grad_norm": 9.792441368103027,
+      "learning_rate": 9.544992846745936e-06,
+      "loss": 2.6275,
+      "step": 591000
+    },
+    {
+      "epoch": 0.15416316157561427,
+      "grad_norm": 10.63184642791748,
+      "learning_rate": 9.544644368279724e-06,
+      "loss": 2.6105,
+      "step": 591200
+    },
+    {
+      "epoch": 0.15421531420131643,
+      "grad_norm": 11.19427490234375,
+      "learning_rate": 9.544295762785464e-06,
+      "loss": 2.6435,
+      "step": 591400
+    },
+    {
+      "epoch": 0.1542674668270186,
+      "grad_norm": 9.695633888244629,
+      "learning_rate": 9.543947030272897e-06,
+      "loss": 2.6084,
+      "step": 591600
+    },
+    {
+      "epoch": 0.15431961945272077,
+      "grad_norm": 10.136941909790039,
+      "learning_rate": 9.543598170751772e-06,
+      "loss": 2.6144,
+      "step": 591800
+    },
+    {
+      "epoch": 0.15437177207842295,
+      "grad_norm": 10.13197135925293,
+      "learning_rate": 9.543249184231837e-06,
+      "loss": 2.6069,
+      "step": 592000
+    },
+    {
+      "epoch": 0.1544239247041251,
+      "grad_norm": 10.096322059631348,
+      "learning_rate": 9.54290007072285e-06,
+      "loss": 2.6032,
+      "step": 592200
+    },
+    {
+      "epoch": 0.1544760773298273,
+      "grad_norm": 8.74548053741455,
+      "learning_rate": 9.542550830234568e-06,
+      "loss": 2.6174,
+      "step": 592400
+    },
+    {
+      "epoch": 0.15452822995552945,
+      "grad_norm": 13.123319625854492,
+      "learning_rate": 9.542201462776752e-06,
+      "loss": 2.6112,
+      "step": 592600
+    },
+    {
+      "epoch": 0.15458038258123163,
+      "grad_norm": 11.316987037658691,
+      "learning_rate": 9.541851968359169e-06,
+      "loss": 2.6042,
+      "step": 592800
+    },
+    {
+      "epoch": 0.1546325352069338,
+      "grad_norm": 9.933026313781738,
+      "learning_rate": 9.541502346991586e-06,
+      "loss": 2.6165,
+      "step": 593000
+    },
+    {
+      "epoch": 0.15468468783263598,
+      "grad_norm": 9.549524307250977,
+      "learning_rate": 9.541152598683776e-06,
+      "loss": 2.5907,
+      "step": 593200
+    },
+    {
+      "epoch": 0.15473684045833813,
+      "grad_norm": 10.845429420471191,
+      "learning_rate": 9.540802723445513e-06,
+      "loss": 2.6287,
+      "step": 593400
+    },
+    {
+      "epoch": 0.15478899308404032,
+      "grad_norm": 9.898306846618652,
+      "learning_rate": 9.54045272128658e-06,
+      "loss": 2.6095,
+      "step": 593600
+    },
+    {
+      "epoch": 0.15484114570974247,
+      "grad_norm": 11.204236030578613,
+      "learning_rate": 9.540102592216757e-06,
+      "loss": 2.6506,
+      "step": 593800
+    },
+    {
+      "epoch": 0.15489329833544466,
+      "grad_norm": 9.741256713867188,
+      "learning_rate": 9.539752336245834e-06,
+      "loss": 2.5979,
+      "step": 594000
+    },
+    {
+      "epoch": 0.1549454509611468,
+      "grad_norm": 9.863458633422852,
+      "learning_rate": 9.539401953383595e-06,
+      "loss": 2.559,
+      "step": 594200
+    },
+    {
+      "epoch": 0.154997603586849,
+      "grad_norm": 11.014803886413574,
+      "learning_rate": 9.53905144363984e-06,
+      "loss": 2.647,
+      "step": 594400
+    },
+    {
+      "epoch": 0.15504975621255115,
+      "grad_norm": 10.70952033996582,
+      "learning_rate": 9.538700807024363e-06,
+      "loss": 2.6143,
+      "step": 594600
+    },
+    {
+      "epoch": 0.15510190883825334,
+      "grad_norm": 10.817914962768555,
+      "learning_rate": 9.538350043546965e-06,
+      "loss": 2.5906,
+      "step": 594800
+    },
+    {
+      "epoch": 0.1551540614639555,
+      "grad_norm": 10.918020248413086,
+      "learning_rate": 9.53799915321745e-06,
+      "loss": 2.6202,
+      "step": 595000
+    },
+    {
+      "epoch": 0.15520621408965765,
+      "grad_norm": 9.967550277709961,
+      "learning_rate": 9.537648136045626e-06,
+      "loss": 2.6368,
+      "step": 595200
+    },
+    {
+      "epoch": 0.15525836671535984,
+      "grad_norm": 10.808151245117188,
+      "learning_rate": 9.537296992041303e-06,
+      "loss": 2.6267,
+      "step": 595400
+    },
+    {
+      "epoch": 0.155310519341062,
+      "grad_norm": 10.046856880187988,
+      "learning_rate": 9.536945721214299e-06,
+      "loss": 2.6376,
+      "step": 595600
+    },
+    {
+      "epoch": 0.15536267196676418,
+      "grad_norm": 10.47192668914795,
+      "learning_rate": 9.536594323574432e-06,
+      "loss": 2.5965,
+      "step": 595800
+    },
+    {
+      "epoch": 0.15541482459246633,
+      "grad_norm": 10.471516609191895,
+      "learning_rate": 9.536242799131522e-06,
+      "loss": 2.6248,
+      "step": 596000
+    },
+    {
+      "epoch": 0.15546697721816852,
+      "grad_norm": 9.41662311553955,
+      "learning_rate": 9.535891147895395e-06,
+      "loss": 2.568,
+      "step": 596200
+    },
+    {
+      "epoch": 0.15551912984387067,
+      "grad_norm": 10.728153228759766,
+      "learning_rate": 9.53553936987588e-06,
+      "loss": 2.615,
+      "step": 596400
+    },
+    {
+      "epoch": 0.15557128246957286,
+      "grad_norm": 11.101585388183594,
+      "learning_rate": 9.535187465082809e-06,
+      "loss": 2.6044,
+      "step": 596600
+    },
+    {
+      "epoch": 0.15562343509527501,
+      "grad_norm": 10.208956718444824,
+      "learning_rate": 9.534835433526021e-06,
+      "loss": 2.5869,
+      "step": 596800
+    },
+    {
+      "epoch": 0.1556755877209772,
+      "grad_norm": 9.980835914611816,
+      "learning_rate": 9.534483275215353e-06,
+      "loss": 2.6418,
+      "step": 597000
+    },
+    {
+      "epoch": 0.15572774034667936,
+      "grad_norm": 10.444533348083496,
+      "learning_rate": 9.534130990160649e-06,
+      "loss": 2.6149,
+      "step": 597200
+    },
+    {
+      "epoch": 0.15577989297238154,
+      "grad_norm": 9.33448600769043,
+      "learning_rate": 9.533778578371755e-06,
+      "loss": 2.6071,
+      "step": 597400
+    },
+    {
+      "epoch": 0.1558320455980837,
+      "grad_norm": 11.563689231872559,
+      "learning_rate": 9.533426039858523e-06,
+      "loss": 2.5906,
+      "step": 597600
+    },
+    {
+      "epoch": 0.15588419822378588,
+      "grad_norm": 10.888169288635254,
+      "learning_rate": 9.533073374630805e-06,
+      "loss": 2.6418,
+      "step": 597800
+    },
+    {
+      "epoch": 0.15593635084948804,
+      "grad_norm": 10.408832550048828,
+      "learning_rate": 9.53272058269846e-06,
+      "loss": 2.5685,
+      "step": 598000
+    },
+    {
+      "epoch": 0.15598850347519022,
+      "grad_norm": 11.165427207946777,
+      "learning_rate": 9.532367664071349e-06,
+      "loss": 2.6015,
+      "step": 598200
+    },
+    {
+      "epoch": 0.15604065610089238,
+      "grad_norm": 11.39417552947998,
+      "learning_rate": 9.532014618759336e-06,
+      "loss": 2.5874,
+      "step": 598400
+    },
+    {
+      "epoch": 0.15609280872659456,
+      "grad_norm": 10.001264572143555,
+      "learning_rate": 9.531661446772287e-06,
+      "loss": 2.5927,
+      "step": 598600
+    },
+    {
+      "epoch": 0.15614496135229672,
+      "grad_norm": 10.457130432128906,
+      "learning_rate": 9.531308148120077e-06,
+      "loss": 2.5764,
+      "step": 598800
+    },
+    {
+      "epoch": 0.1561971139779989,
+      "grad_norm": 8.923357963562012,
+      "learning_rate": 9.530954722812577e-06,
+      "loss": 2.6262,
+      "step": 599000
+    },
+    {
+      "epoch": 0.15624926660370106,
+      "grad_norm": 11.424574851989746,
+      "learning_rate": 9.530601170859672e-06,
+      "loss": 2.6137,
+      "step": 599200
+    },
+    {
+      "epoch": 0.15630141922940324,
+      "grad_norm": 11.1590576171875,
+      "learning_rate": 9.530247492271237e-06,
+      "loss": 2.6372,
+      "step": 599400
+    },
+    {
+      "epoch": 0.1563535718551054,
+      "grad_norm": 10.053206443786621,
+      "learning_rate": 9.529893687057163e-06,
+      "loss": 2.6275,
+      "step": 599600
+    },
+    {
+      "epoch": 0.15640572448080758,
+      "grad_norm": 10.665011405944824,
+      "learning_rate": 9.529539755227336e-06,
+      "loss": 2.5987,
+      "step": 599800
+    },
+    {
+      "epoch": 0.15645787710650974,
+      "grad_norm": 10.702035903930664,
+      "learning_rate": 9.529185696791651e-06,
+      "loss": 2.5779,
+      "step": 600000
+    },
+    {
+      "epoch": 0.15651002973221192,
+      "grad_norm": 9.53780746459961,
+      "learning_rate": 9.528831511760003e-06,
+      "loss": 2.6361,
+      "step": 600200
+    },
+    {
+      "epoch": 0.15656218235791408,
+      "grad_norm": 11.620895385742188,
+      "learning_rate": 9.528477200142292e-06,
+      "loss": 2.6155,
+      "step": 600400
+    },
+    {
+      "epoch": 0.15661433498361627,
+      "grad_norm": 9.885120391845703,
+      "learning_rate": 9.528122761948421e-06,
+      "loss": 2.6552,
+      "step": 600600
+    },
+    {
+      "epoch": 0.15666648760931842,
+      "grad_norm": 11.59696102142334,
+      "learning_rate": 9.527768197188299e-06,
+      "loss": 2.6134,
+      "step": 600800
+    },
+    {
+      "epoch": 0.15671864023502058,
+      "grad_norm": 10.598244667053223,
+      "learning_rate": 9.527413505871836e-06,
+      "loss": 2.5894,
+      "step": 601000
+    },
+    {
+      "epoch": 0.15677079286072276,
+      "grad_norm": 11.047648429870605,
+      "learning_rate": 9.527058688008943e-06,
+      "loss": 2.6413,
+      "step": 601200
+    },
+    {
+      "epoch": 0.15682294548642492,
+      "grad_norm": 10.149199485778809,
+      "learning_rate": 9.526703743609543e-06,
+      "loss": 2.6164,
+      "step": 601400
+    },
+    {
+      "epoch": 0.1568750981121271,
+      "grad_norm": 10.866960525512695,
+      "learning_rate": 9.52634867268355e-06,
+      "loss": 2.5793,
+      "step": 601600
+    },
+    {
+      "epoch": 0.15692725073782926,
+      "grad_norm": 9.808792114257812,
+      "learning_rate": 9.525993475240894e-06,
+      "loss": 2.6329,
+      "step": 601800
+    },
+    {
+      "epoch": 0.15697940336353144,
+      "grad_norm": 9.316394805908203,
+      "learning_rate": 9.525638151291504e-06,
+      "loss": 2.6382,
+      "step": 602000
+    },
+    {
+      "epoch": 0.1570315559892336,
+      "grad_norm": 9.754195213317871,
+      "learning_rate": 9.525282700845305e-06,
+      "loss": 2.6143,
+      "step": 602200
+    },
+    {
+      "epoch": 0.15708370861493579,
+      "grad_norm": 11.504837036132812,
+      "learning_rate": 9.52492712391224e-06,
+      "loss": 2.6187,
+      "step": 602400
+    },
+    {
+      "epoch": 0.15713586124063794,
+      "grad_norm": 9.503823280334473,
+      "learning_rate": 9.524571420502243e-06,
+      "loss": 2.638,
+      "step": 602600
+    },
+    {
+      "epoch": 0.15718801386634013,
+      "grad_norm": 11.267765998840332,
+      "learning_rate": 9.524215590625258e-06,
+      "loss": 2.6037,
+      "step": 602800
+    },
+    {
+      "epoch": 0.15724016649204228,
+      "grad_norm": 12.685362815856934,
+      "learning_rate": 9.52385963429123e-06,
+      "loss": 2.5829,
+      "step": 603000
+    },
+    {
+      "epoch": 0.15729231911774447,
+      "grad_norm": 9.462133407592773,
+      "learning_rate": 9.523503551510112e-06,
+      "loss": 2.6734,
+      "step": 603200
+    },
+    {
+      "epoch": 0.15734447174344662,
+      "grad_norm": 10.139527320861816,
+      "learning_rate": 9.52314734229185e-06,
+      "loss": 2.6098,
+      "step": 603400
+    },
+    {
+      "epoch": 0.1573966243691488,
+      "grad_norm": 10.855724334716797,
+      "learning_rate": 9.522791006646407e-06,
+      "loss": 2.5865,
+      "step": 603600
+    },
+    {
+      "epoch": 0.15744877699485096,
+      "grad_norm": 10.368106842041016,
+      "learning_rate": 9.522434544583738e-06,
+      "loss": 2.621,
+      "step": 603800
+    },
+    {
+      "epoch": 0.15750092962055315,
+      "grad_norm": 11.346565246582031,
+      "learning_rate": 9.522077956113812e-06,
+      "loss": 2.6346,
+      "step": 604000
+    },
+    {
+      "epoch": 0.1575530822462553,
+      "grad_norm": 11.265953063964844,
+      "learning_rate": 9.521721241246591e-06,
+      "loss": 2.6319,
+      "step": 604200
+    },
+    {
+      "epoch": 0.1576052348719575,
+      "grad_norm": 10.973451614379883,
+      "learning_rate": 9.521364399992048e-06,
+      "loss": 2.6154,
+      "step": 604400
+    },
+    {
+      "epoch": 0.15765738749765965,
+      "grad_norm": 10.001646995544434,
+      "learning_rate": 9.521007432360157e-06,
+      "loss": 2.5999,
+      "step": 604600
+    },
+    {
+      "epoch": 0.15770954012336183,
+      "grad_norm": 9.617718696594238,
+      "learning_rate": 9.520650338360896e-06,
+      "loss": 2.6654,
+      "step": 604800
+    },
+    {
+      "epoch": 0.157761692749064,
+      "grad_norm": 9.889549255371094,
+      "learning_rate": 9.520293118004243e-06,
+      "loss": 2.6504,
+      "step": 605000
+    },
+    {
+      "epoch": 0.15781384537476617,
+      "grad_norm": 12.019047737121582,
+      "learning_rate": 9.519935771300188e-06,
+      "loss": 2.619,
+      "step": 605200
+    },
+    {
+      "epoch": 0.15786599800046833,
+      "grad_norm": 9.76209831237793,
+      "learning_rate": 9.519578298258714e-06,
+      "loss": 2.5828,
+      "step": 605400
+    },
+    {
+      "epoch": 0.1579181506261705,
+      "grad_norm": 11.635560989379883,
+      "learning_rate": 9.519220698889817e-06,
+      "loss": 2.5842,
+      "step": 605600
+    },
+    {
+      "epoch": 0.15797030325187267,
+      "grad_norm": 11.385933876037598,
+      "learning_rate": 9.518862973203491e-06,
+      "loss": 2.57,
+      "step": 605800
+    },
+    {
+      "epoch": 0.15802245587757485,
+      "grad_norm": 10.346911430358887,
+      "learning_rate": 9.518505121209733e-06,
+      "loss": 2.6275,
+      "step": 606000
+    },
+    {
+      "epoch": 0.158074608503277,
+      "grad_norm": 11.468281745910645,
+      "learning_rate": 9.518147142918548e-06,
+      "loss": 2.5972,
+      "step": 606200
+    },
+    {
+      "epoch": 0.1581267611289792,
+      "grad_norm": 10.361491203308105,
+      "learning_rate": 9.517789038339939e-06,
+      "loss": 2.6139,
+      "step": 606400
+    },
+    {
+      "epoch": 0.15817891375468135,
+      "grad_norm": 10.77645206451416,
+      "learning_rate": 9.517430807483919e-06,
+      "loss": 2.6024,
+      "step": 606600
+    },
+    {
+      "epoch": 0.15823106638038353,
+      "grad_norm": 10.53165054321289,
+      "learning_rate": 9.517072450360499e-06,
+      "loss": 2.5576,
+      "step": 606800
+    },
+    {
+      "epoch": 0.1582832190060857,
+      "grad_norm": 10.261857986450195,
+      "learning_rate": 9.516713966979694e-06,
+      "loss": 2.6202,
+      "step": 607000
+    },
+    {
+      "epoch": 0.15833537163178785,
+      "grad_norm": 10.302119255065918,
+      "learning_rate": 9.516355357351528e-06,
+      "loss": 2.617,
+      "step": 607200
+    },
+    {
+      "epoch": 0.15838752425749003,
+      "grad_norm": 10.450231552124023,
+      "learning_rate": 9.515996621486021e-06,
+      "loss": 2.6169,
+      "step": 607400
+    },
+    {
+      "epoch": 0.1584396768831922,
+      "grad_norm": 11.078191757202148,
+      "learning_rate": 9.515637759393201e-06,
+      "loss": 2.6044,
+      "step": 607600
+    },
+    {
+      "epoch": 0.15849182950889437,
+      "grad_norm": 11.261504173278809,
+      "learning_rate": 9.515278771083098e-06,
+      "loss": 2.6508,
+      "step": 607800
+    },
+    {
+      "epoch": 0.15854398213459653,
+      "grad_norm": 11.22635555267334,
+      "learning_rate": 9.514919656565749e-06,
+      "loss": 2.595,
+      "step": 608000
+    },
+    {
+      "epoch": 0.1585961347602987,
+      "grad_norm": 9.812448501586914,
+      "learning_rate": 9.514560415851189e-06,
+      "loss": 2.6281,
+      "step": 608200
+    },
+    {
+      "epoch": 0.15864828738600087,
+      "grad_norm": 11.062207221984863,
+      "learning_rate": 9.51420104894946e-06,
+      "loss": 2.628,
+      "step": 608400
+    },
+    {
+      "epoch": 0.15870044001170305,
+      "grad_norm": 11.523585319519043,
+      "learning_rate": 9.513841555870607e-06,
+      "loss": 2.6367,
+      "step": 608600
+    },
+    {
+      "epoch": 0.1587525926374052,
+      "grad_norm": 9.39417839050293,
+      "learning_rate": 9.513481936624677e-06,
+      "loss": 2.63,
+      "step": 608800
+    },
+    {
+      "epoch": 0.1588047452631074,
+      "grad_norm": 10.997912406921387,
+      "learning_rate": 9.513122191221722e-06,
+      "loss": 2.5634,
+      "step": 609000
+    },
+    {
+      "epoch": 0.15885689788880955,
+      "grad_norm": 10.309552192687988,
+      "learning_rate": 9.5127623196718e-06,
+      "loss": 2.6526,
+      "step": 609200
+    },
+    {
+      "epoch": 0.15890905051451173,
+      "grad_norm": 10.835392951965332,
+      "learning_rate": 9.512402321984966e-06,
+      "loss": 2.6135,
+      "step": 609400
+    },
+    {
+      "epoch": 0.1589612031402139,
+      "grad_norm": 10.294236183166504,
+      "learning_rate": 9.512042198171286e-06,
+      "loss": 2.6536,
+      "step": 609600
+    },
+    {
+      "epoch": 0.15901335576591608,
+      "grad_norm": 11.246005058288574,
+      "learning_rate": 9.51168194824082e-06,
+      "loss": 2.6102,
+      "step": 609800
+    },
+    {
+      "epoch": 0.15906550839161823,
+      "grad_norm": 11.818621635437012,
+      "learning_rate": 9.511321572203645e-06,
+      "loss": 2.6259,
+      "step": 610000
+    },
+    {
+      "epoch": 0.15911766101732042,
+      "grad_norm": 9.672310829162598,
+      "learning_rate": 9.510961070069829e-06,
+      "loss": 2.5676,
+      "step": 610200
+    },
+    {
+      "epoch": 0.15916981364302257,
+      "grad_norm": 11.772770881652832,
+      "learning_rate": 9.510600441849451e-06,
+      "loss": 2.5989,
+      "step": 610400
+    },
+    {
+      "epoch": 0.15922196626872476,
+      "grad_norm": 10.861141204833984,
+      "learning_rate": 9.510239687552589e-06,
+      "loss": 2.5797,
+      "step": 610600
+    },
+    {
+      "epoch": 0.1592741188944269,
+      "grad_norm": 9.267687797546387,
+      "learning_rate": 9.509878807189327e-06,
+      "loss": 2.5731,
+      "step": 610800
+    },
+    {
+      "epoch": 0.1593262715201291,
+      "grad_norm": 9.538235664367676,
+      "learning_rate": 9.509517800769752e-06,
+      "loss": 2.6158,
+      "step": 611000
+    },
+    {
+      "epoch": 0.15937842414583125,
+      "grad_norm": 11.026472091674805,
+      "learning_rate": 9.509156668303956e-06,
+      "loss": 2.6308,
+      "step": 611200
+    },
+    {
+      "epoch": 0.15943057677153344,
+      "grad_norm": 10.019219398498535,
+      "learning_rate": 9.50879540980203e-06,
+      "loss": 2.6177,
+      "step": 611400
+    },
+    {
+      "epoch": 0.1594827293972356,
+      "grad_norm": 9.39755916595459,
+      "learning_rate": 9.508434025274074e-06,
+      "loss": 2.5825,
+      "step": 611600
+    },
+    {
+      "epoch": 0.15953488202293778,
+      "grad_norm": 9.156355857849121,
+      "learning_rate": 9.508072514730189e-06,
+      "loss": 2.6348,
+      "step": 611800
+    },
+    {
+      "epoch": 0.15958703464863994,
+      "grad_norm": 10.577462196350098,
+      "learning_rate": 9.50771087818048e-06,
+      "loss": 2.627,
+      "step": 612000
+    },
+    {
+      "epoch": 0.15963918727434212,
+      "grad_norm": 10.779411315917969,
+      "learning_rate": 9.507349115635053e-06,
+      "loss": 2.6185,
+      "step": 612200
+    },
+    {
+      "epoch": 0.15969133990004428,
+      "grad_norm": 12.37381362915039,
+      "learning_rate": 9.506987227104021e-06,
+      "loss": 2.6284,
+      "step": 612400
+    },
+    {
+      "epoch": 0.15974349252574646,
+      "grad_norm": 9.256758689880371,
+      "learning_rate": 9.5066252125975e-06,
+      "loss": 2.6302,
+      "step": 612600
+    },
+    {
+      "epoch": 0.15979564515144862,
+      "grad_norm": 10.987409591674805,
+      "learning_rate": 9.506263072125607e-06,
+      "loss": 2.5929,
+      "step": 612800
+    },
+    {
+      "epoch": 0.15984779777715077,
+      "grad_norm": 9.691329002380371,
+      "learning_rate": 9.505900805698468e-06,
+      "loss": 2.6097,
+      "step": 613000
+    },
+    {
+      "epoch": 0.15989995040285296,
+      "grad_norm": 10.040337562561035,
+      "learning_rate": 9.505538413326203e-06,
+      "loss": 2.6048,
+      "step": 613200
+    },
+    {
+      "epoch": 0.15995210302855511,
+      "grad_norm": 8.556117057800293,
+      "learning_rate": 9.505175895018946e-06,
+      "loss": 2.5706,
+      "step": 613400
+    },
+    {
+      "epoch": 0.1600042556542573,
+      "grad_norm": 11.223811149597168,
+      "learning_rate": 9.504813250786826e-06,
+      "loss": 2.5724,
+      "step": 613600
+    },
+    {
+      "epoch": 0.16005640827995946,
+      "grad_norm": 9.269017219543457,
+      "learning_rate": 9.504450480639984e-06,
+      "loss": 2.6118,
+      "step": 613800
+    },
+    {
+      "epoch": 0.16010856090566164,
+      "grad_norm": 10.954751968383789,
+      "learning_rate": 9.504087584588554e-06,
+      "loss": 2.622,
+      "step": 614000
+    },
+    {
+      "epoch": 0.1601607135313638,
+      "grad_norm": 10.984915733337402,
+      "learning_rate": 9.503724562642686e-06,
+      "loss": 2.5626,
+      "step": 614200
+    },
+    {
+      "epoch": 0.16021286615706598,
+      "grad_norm": 10.41535758972168,
+      "learning_rate": 9.503361414812522e-06,
+      "loss": 2.6036,
+      "step": 614400
+    },
+    {
+      "epoch": 0.16026501878276814,
+      "grad_norm": 9.983470916748047,
+      "learning_rate": 9.502998141108215e-06,
+      "loss": 2.6439,
+      "step": 614600
+    },
+    {
+      "epoch": 0.16031717140847032,
+      "grad_norm": 9.90550708770752,
+      "learning_rate": 9.502634741539916e-06,
+      "loss": 2.5825,
+      "step": 614800
+    },
+    {
+      "epoch": 0.16036932403417248,
+      "grad_norm": 10.278912544250488,
+      "learning_rate": 9.502271216117784e-06,
+      "loss": 2.5804,
+      "step": 615000
+    },
+    {
+      "epoch": 0.16042147665987466,
+      "grad_norm": 10.6985445022583,
+      "learning_rate": 9.501907564851982e-06,
+      "loss": 2.6131,
+      "step": 615200
+    },
+    {
+      "epoch": 0.16047362928557682,
+      "grad_norm": 10.952655792236328,
+      "learning_rate": 9.501543787752672e-06,
+      "loss": 2.6165,
+      "step": 615400
+    },
+    {
+      "epoch": 0.160525781911279,
+      "grad_norm": 10.143887519836426,
+      "learning_rate": 9.501179884830021e-06,
+      "loss": 2.6062,
+      "step": 615600
+    },
+    {
+      "epoch": 0.16057793453698116,
+      "grad_norm": 11.375021934509277,
+      "learning_rate": 9.500815856094204e-06,
+      "loss": 2.6002,
+      "step": 615800
+    },
+    {
+      "epoch": 0.16063008716268334,
+      "grad_norm": 10.76820182800293,
+      "learning_rate": 9.500451701555394e-06,
+      "loss": 2.5903,
+      "step": 616000
+    },
+    {
+      "epoch": 0.1606822397883855,
+      "grad_norm": 9.969457626342773,
+      "learning_rate": 9.500087421223769e-06,
+      "loss": 2.6216,
+      "step": 616200
+    },
+    {
+      "epoch": 0.16073439241408768,
+      "grad_norm": 11.117074012756348,
+      "learning_rate": 9.499723015109514e-06,
+      "loss": 2.6253,
+      "step": 616400
+    },
+    {
+      "epoch": 0.16078654503978984,
+      "grad_norm": 10.098209381103516,
+      "learning_rate": 9.499358483222809e-06,
+      "loss": 2.6172,
+      "step": 616600
+    },
+    {
+      "epoch": 0.16083869766549203,
+      "grad_norm": 10.3887939453125,
+      "learning_rate": 9.49899382557385e-06,
+      "loss": 2.619,
+      "step": 616800
+    },
+    {
+      "epoch": 0.16089085029119418,
+      "grad_norm": 11.24229621887207,
+      "learning_rate": 9.498629042172824e-06,
+      "loss": 2.6362,
+      "step": 617000
+    },
+    {
+      "epoch": 0.16094300291689637,
+      "grad_norm": 11.60924243927002,
+      "learning_rate": 9.498264133029928e-06,
+      "loss": 2.6756,
+      "step": 617200
+    },
+    {
+      "epoch": 0.16099515554259852,
+      "grad_norm": 11.145500183105469,
+      "learning_rate": 9.497899098155364e-06,
+      "loss": 2.6094,
+      "step": 617400
+    },
+    {
+      "epoch": 0.1610473081683007,
+      "grad_norm": 11.344988822937012,
+      "learning_rate": 9.497533937559335e-06,
+      "loss": 2.5854,
+      "step": 617600
+    },
+    {
+      "epoch": 0.16109946079400286,
+      "grad_norm": 11.063643455505371,
+      "learning_rate": 9.497168651252044e-06,
+      "loss": 2.6057,
+      "step": 617800
+    },
+    {
+      "epoch": 0.16115161341970505,
+      "grad_norm": 10.891105651855469,
+      "learning_rate": 9.496803239243706e-06,
+      "loss": 2.5996,
+      "step": 618000
+    },
+    {
+      "epoch": 0.1612037660454072,
+      "grad_norm": 10.276857376098633,
+      "learning_rate": 9.496437701544533e-06,
+      "loss": 2.6236,
+      "step": 618200
+    },
+    {
+      "epoch": 0.1612559186711094,
+      "grad_norm": 10.791783332824707,
+      "learning_rate": 9.49607203816474e-06,
+      "loss": 2.6193,
+      "step": 618400
+    },
+    {
+      "epoch": 0.16130807129681154,
+      "grad_norm": 9.532054901123047,
+      "learning_rate": 9.495706249114549e-06,
+      "loss": 2.5948,
+      "step": 618600
+    },
+    {
+      "epoch": 0.1613602239225137,
+      "grad_norm": 9.186354637145996,
+      "learning_rate": 9.495340334404186e-06,
+      "loss": 2.595,
+      "step": 618800
+    },
+    {
+      "epoch": 0.16141237654821589,
+      "grad_norm": 9.525355339050293,
+      "learning_rate": 9.494974294043878e-06,
+      "loss": 2.551,
+      "step": 619000
+    },
+    {
+      "epoch": 0.16146452917391804,
+      "grad_norm": 9.88534164428711,
+      "learning_rate": 9.494608128043852e-06,
+      "loss": 2.5548,
+      "step": 619200
+    },
+    {
+      "epoch": 0.16151668179962023,
+      "grad_norm": 11.86941146850586,
+      "learning_rate": 9.49424183641435e-06,
+      "loss": 2.6126,
+      "step": 619400
+    },
+    {
+      "epoch": 0.16156883442532238,
+      "grad_norm": 9.518510818481445,
+      "learning_rate": 9.493875419165605e-06,
+      "loss": 2.6217,
+      "step": 619600
+    },
+    {
+      "epoch": 0.16162098705102457,
+      "grad_norm": 11.503427505493164,
+      "learning_rate": 9.49350887630786e-06,
+      "loss": 2.6139,
+      "step": 619800
+    },
+    {
+      "epoch": 0.16167313967672672,
+      "grad_norm": 11.120110511779785,
+      "learning_rate": 9.493142207851362e-06,
+      "loss": 2.6086,
+      "step": 620000
+    },
+    {
+      "epoch": 0.1617252923024289,
+      "grad_norm": 9.80033016204834,
+      "learning_rate": 9.492775413806357e-06,
+      "loss": 2.5888,
+      "step": 620200
+    },
+    {
+      "epoch": 0.16177744492813106,
+      "grad_norm": 9.526128768920898,
+      "learning_rate": 9.4924084941831e-06,
+      "loss": 2.5898,
+      "step": 620400
+    },
+    {
+      "epoch": 0.16182959755383325,
+      "grad_norm": 9.477727890014648,
+      "learning_rate": 9.492041448991845e-06,
+      "loss": 2.5997,
+      "step": 620600
+    },
+    {
+      "epoch": 0.1618817501795354,
+      "grad_norm": 11.579855918884277,
+      "learning_rate": 9.491674278242852e-06,
+      "loss": 2.6183,
+      "step": 620800
+    },
+    {
+      "epoch": 0.1619339028052376,
+      "grad_norm": 10.910771369934082,
+      "learning_rate": 9.491306981946385e-06,
+      "loss": 2.5857,
+      "step": 621000
+    },
+    {
+      "epoch": 0.16198605543093975,
+      "grad_norm": 10.065752983093262,
+      "learning_rate": 9.490939560112709e-06,
+      "loss": 2.6177,
+      "step": 621200
+    },
+    {
+      "epoch": 0.16203820805664193,
+      "grad_norm": 11.171086311340332,
+      "learning_rate": 9.490572012752093e-06,
+      "loss": 2.5936,
+      "step": 621400
+    },
+    {
+      "epoch": 0.1620903606823441,
+      "grad_norm": 11.971680641174316,
+      "learning_rate": 9.490204339874811e-06,
+      "loss": 2.6105,
+      "step": 621600
+    },
+    {
+      "epoch": 0.16214251330804627,
+      "grad_norm": 11.462523460388184,
+      "learning_rate": 9.489836541491142e-06,
+      "loss": 2.571,
+      "step": 621800
+    },
+    {
+      "epoch": 0.16219466593374843,
+      "grad_norm": 10.30375862121582,
+      "learning_rate": 9.489468617611363e-06,
+      "loss": 2.6248,
+      "step": 622000
+    },
+    {
+      "epoch": 0.1622468185594506,
+      "grad_norm": 12.754924774169922,
+      "learning_rate": 9.48910056824576e-06,
+      "loss": 2.626,
+      "step": 622200
+    },
+    {
+      "epoch": 0.16229897118515277,
+      "grad_norm": 10.41318416595459,
+      "learning_rate": 9.48873239340462e-06,
+      "loss": 2.6197,
+      "step": 622400
+    },
+    {
+      "epoch": 0.16235112381085495,
+      "grad_norm": 10.452898025512695,
+      "learning_rate": 9.488364093098235e-06,
+      "loss": 2.6034,
+      "step": 622600
+    },
+    {
+      "epoch": 0.1624032764365571,
+      "grad_norm": 10.983317375183105,
+      "learning_rate": 9.487995667336897e-06,
+      "loss": 2.6084,
+      "step": 622800
+    },
+    {
+      "epoch": 0.1624554290622593,
+      "grad_norm": 12.018777847290039,
+      "learning_rate": 9.487627116130907e-06,
+      "loss": 2.5987,
+      "step": 623000
+    },
+    {
+      "epoch": 0.16250758168796145,
+      "grad_norm": 10.180075645446777,
+      "learning_rate": 9.487258439490563e-06,
+      "loss": 2.578,
+      "step": 623200
+    },
+    {
+      "epoch": 0.16255973431366363,
+      "grad_norm": 10.785030364990234,
+      "learning_rate": 9.486889637426171e-06,
+      "loss": 2.5835,
+      "step": 623400
+    },
+    {
+      "epoch": 0.1626118869393658,
+      "grad_norm": 12.28320026397705,
+      "learning_rate": 9.486520709948042e-06,
+      "loss": 2.5512,
+      "step": 623600
+    },
+    {
+      "epoch": 0.16266403956506797,
+      "grad_norm": 10.83224868774414,
+      "learning_rate": 9.486151657066485e-06,
+      "loss": 2.621,
+      "step": 623800
+    },
+    {
+      "epoch": 0.16271619219077013,
+      "grad_norm": 11.643034934997559,
+      "learning_rate": 9.485782478791817e-06,
+      "loss": 2.6352,
+      "step": 624000
+    },
+    {
+      "epoch": 0.16276834481647232,
+      "grad_norm": 11.283828735351562,
+      "learning_rate": 9.485413175134355e-06,
+      "loss": 2.6173,
+      "step": 624200
+    },
+    {
+      "epoch": 0.16282049744217447,
+      "grad_norm": 11.758881568908691,
+      "learning_rate": 9.485043746104424e-06,
+      "loss": 2.6289,
+      "step": 624400
+    },
+    {
+      "epoch": 0.16287265006787663,
+      "grad_norm": 10.56377124786377,
+      "learning_rate": 9.484674191712348e-06,
+      "loss": 2.5545,
+      "step": 624600
+    },
+    {
+      "epoch": 0.1629248026935788,
+      "grad_norm": 10.32989501953125,
+      "learning_rate": 9.48430451196846e-06,
+      "loss": 2.5994,
+      "step": 624800
+    },
+    {
+      "epoch": 0.16297695531928097,
+      "grad_norm": 10.324919700622559,
+      "learning_rate": 9.483934706883086e-06,
+      "loss": 2.6167,
+      "step": 625000
+    },
+    {
+      "epoch": 0.16302910794498315,
+      "grad_norm": 10.184969902038574,
+      "learning_rate": 9.48356477646657e-06,
+      "loss": 2.5771,
+      "step": 625200
+    },
+    {
+      "epoch": 0.1630812605706853,
+      "grad_norm": 10.589273452758789,
+      "learning_rate": 9.483194720729247e-06,
+      "loss": 2.6033,
+      "step": 625400
+    },
+    {
+      "epoch": 0.1631334131963875,
+      "grad_norm": 10.217822074890137,
+      "learning_rate": 9.482824539681463e-06,
+      "loss": 2.6012,
+      "step": 625600
+    },
+    {
+      "epoch": 0.16318556582208965,
+      "grad_norm": 10.657920837402344,
+      "learning_rate": 9.482454233333564e-06,
+      "loss": 2.6132,
+      "step": 625800
+    },
+    {
+      "epoch": 0.16323771844779184,
+      "grad_norm": 12.07504653930664,
+      "learning_rate": 9.482083801695902e-06,
+      "loss": 2.6208,
+      "step": 626000
+    },
+    {
+      "epoch": 0.163289871073494,
+      "grad_norm": 10.118162155151367,
+      "learning_rate": 9.481713244778829e-06,
+      "loss": 2.6053,
+      "step": 626200
+    },
+    {
+      "epoch": 0.16334202369919618,
+      "grad_norm": 10.543754577636719,
+      "learning_rate": 9.481342562592702e-06,
+      "loss": 2.6122,
+      "step": 626400
+    },
+    {
+      "epoch": 0.16339417632489833,
+      "grad_norm": 11.980351448059082,
+      "learning_rate": 9.480971755147884e-06,
+      "loss": 2.5896,
+      "step": 626600
+    },
+    {
+      "epoch": 0.16344632895060052,
+      "grad_norm": 9.149388313293457,
+      "learning_rate": 9.48060082245474e-06,
+      "loss": 2.62,
+      "step": 626800
+    },
+    {
+      "epoch": 0.16349848157630267,
+      "grad_norm": 10.801803588867188,
+      "learning_rate": 9.480229764523634e-06,
+      "loss": 2.5989,
+      "step": 627000
+    },
+    {
+      "epoch": 0.16355063420200486,
+      "grad_norm": 9.936863899230957,
+      "learning_rate": 9.479858581364942e-06,
+      "loss": 2.5892,
+      "step": 627200
+    },
+    {
+      "epoch": 0.16360278682770701,
+      "grad_norm": 12.720370292663574,
+      "learning_rate": 9.479487272989036e-06,
+      "loss": 2.6159,
+      "step": 627400
+    },
+    {
+      "epoch": 0.1636549394534092,
+      "grad_norm": 8.66869831085205,
+      "learning_rate": 9.479115839406296e-06,
+      "loss": 2.6255,
+      "step": 627600
+    },
+    {
+      "epoch": 0.16370709207911135,
+      "grad_norm": 11.05100154876709,
+      "learning_rate": 9.478744280627106e-06,
+      "loss": 2.6127,
+      "step": 627800
+    },
+    {
+      "epoch": 0.16375924470481354,
+      "grad_norm": 11.327431678771973,
+      "learning_rate": 9.478372596661847e-06,
+      "loss": 2.6414,
+      "step": 628000
+    },
+    {
+      "epoch": 0.1638113973305157,
+      "grad_norm": 10.757742881774902,
+      "learning_rate": 9.47800078752091e-06,
+      "loss": 2.6034,
+      "step": 628200
+    },
+    {
+      "epoch": 0.16386354995621788,
+      "grad_norm": 9.70084285736084,
+      "learning_rate": 9.477628853214689e-06,
+      "loss": 2.582,
+      "step": 628400
+    },
+    {
+      "epoch": 0.16391570258192004,
+      "grad_norm": 11.229771614074707,
+      "learning_rate": 9.477256793753578e-06,
+      "loss": 2.6069,
+      "step": 628600
+    },
+    {
+      "epoch": 0.16396785520762222,
+      "grad_norm": 10.731244087219238,
+      "learning_rate": 9.476884609147976e-06,
+      "loss": 2.5613,
+      "step": 628800
+    },
+    {
+      "epoch": 0.16402000783332438,
+      "grad_norm": 9.405749320983887,
+      "learning_rate": 9.476512299408287e-06,
+      "loss": 2.5655,
+      "step": 629000
+    },
+    {
+      "epoch": 0.16407216045902656,
+      "grad_norm": 12.480544090270996,
+      "learning_rate": 9.47613986454492e-06,
+      "loss": 2.6047,
+      "step": 629200
+    },
+    {
+      "epoch": 0.16412431308472872,
+      "grad_norm": 10.722823143005371,
+      "learning_rate": 9.47576730456828e-06,
+      "loss": 2.5775,
+      "step": 629400
+    },
+    {
+      "epoch": 0.1641764657104309,
+      "grad_norm": 10.385933876037598,
+      "learning_rate": 9.475394619488785e-06,
+      "loss": 2.6002,
+      "step": 629600
+    },
+    {
+      "epoch": 0.16422861833613306,
+      "grad_norm": 8.670275688171387,
+      "learning_rate": 9.47502180931685e-06,
+      "loss": 2.552,
+      "step": 629800
+    },
+    {
+      "epoch": 0.16428077096183524,
+      "grad_norm": 11.322480201721191,
+      "learning_rate": 9.474648874062894e-06,
+      "loss": 2.6183,
+      "step": 630000
+    },
+    {
+      "epoch": 0.1643329235875374,
+      "grad_norm": 10.096075057983398,
+      "learning_rate": 9.474275813737344e-06,
+      "loss": 2.582,
+      "step": 630200
+    },
+    {
+      "epoch": 0.16438507621323958,
+      "grad_norm": 10.714835166931152,
+      "learning_rate": 9.473902628350624e-06,
+      "loss": 2.6248,
+      "step": 630400
+    },
+    {
+      "epoch": 0.16443722883894174,
+      "grad_norm": 9.770901679992676,
+      "learning_rate": 9.473529317913169e-06,
+      "loss": 2.6214,
+      "step": 630600
+    },
+    {
+      "epoch": 0.1644893814646439,
+      "grad_norm": 11.455342292785645,
+      "learning_rate": 9.47315588243541e-06,
+      "loss": 2.5787,
+      "step": 630800
+    },
+    {
+      "epoch": 0.16454153409034608,
+      "grad_norm": 11.288113594055176,
+      "learning_rate": 9.472782321927786e-06,
+      "loss": 2.6009,
+      "step": 631000
+    },
+    {
+      "epoch": 0.16459368671604824,
+      "grad_norm": 9.5037260055542,
+      "learning_rate": 9.472408636400742e-06,
+      "loss": 2.579,
+      "step": 631200
+    },
+    {
+      "epoch": 0.16464583934175042,
+      "grad_norm": 11.143527030944824,
+      "learning_rate": 9.472034825864718e-06,
+      "loss": 2.585,
+      "step": 631400
+    },
+    {
+      "epoch": 0.16469799196745258,
+      "grad_norm": 10.50413703918457,
+      "learning_rate": 9.471660890330163e-06,
+      "loss": 2.5971,
+      "step": 631600
+    },
+    {
+      "epoch": 0.16475014459315476,
+      "grad_norm": 11.185114860534668,
+      "learning_rate": 9.471286829807531e-06,
+      "loss": 2.6326,
+      "step": 631800
+    },
+    {
+      "epoch": 0.16480229721885692,
+      "grad_norm": 10.260226249694824,
+      "learning_rate": 9.470912644307276e-06,
+      "loss": 2.6244,
+      "step": 632000
+    },
+    {
+      "epoch": 0.1648544498445591,
+      "grad_norm": 11.488462448120117,
+      "learning_rate": 9.470538333839858e-06,
+      "loss": 2.598,
+      "step": 632200
+    },
+    {
+      "epoch": 0.16490660247026126,
+      "grad_norm": 11.961732864379883,
+      "learning_rate": 9.470163898415738e-06,
+      "loss": 2.5907,
+      "step": 632400
+    },
+    {
+      "epoch": 0.16495875509596344,
+      "grad_norm": 12.416627883911133,
+      "learning_rate": 9.469789338045383e-06,
+      "loss": 2.5822,
+      "step": 632600
+    },
+    {
+      "epoch": 0.1650109077216656,
+      "grad_norm": 10.282339096069336,
+      "learning_rate": 9.469414652739262e-06,
+      "loss": 2.5707,
+      "step": 632800
+    },
+    {
+      "epoch": 0.16506306034736778,
+      "grad_norm": 10.362985610961914,
+      "learning_rate": 9.46903984250785e-06,
+      "loss": 2.6038,
+      "step": 633000
+    },
+    {
+      "epoch": 0.16511521297306994,
+      "grad_norm": 11.942998886108398,
+      "learning_rate": 9.468664907361618e-06,
+      "loss": 2.6165,
+      "step": 633200
+    },
+    {
+      "epoch": 0.16516736559877213,
+      "grad_norm": 11.504192352294922,
+      "learning_rate": 9.468289847311052e-06,
+      "loss": 2.6164,
+      "step": 633400
+    },
+    {
+      "epoch": 0.16521951822447428,
+      "grad_norm": 11.131896018981934,
+      "learning_rate": 9.467914662366632e-06,
+      "loss": 2.5774,
+      "step": 633600
+    },
+    {
+      "epoch": 0.16527167085017647,
+      "grad_norm": 10.091717720031738,
+      "learning_rate": 9.467539352538845e-06,
+      "loss": 2.6165,
+      "step": 633800
+    },
+    {
+      "epoch": 0.16532382347587862,
+      "grad_norm": 9.936517715454102,
+      "learning_rate": 9.467163917838183e-06,
+      "loss": 2.5712,
+      "step": 634000
+    },
+    {
+      "epoch": 0.1653759761015808,
+      "grad_norm": 11.358585357666016,
+      "learning_rate": 9.466788358275136e-06,
+      "loss": 2.5891,
+      "step": 634200
+    },
+    {
+      "epoch": 0.16542812872728296,
+      "grad_norm": 10.875038146972656,
+      "learning_rate": 9.466412673860206e-06,
+      "loss": 2.6178,
+      "step": 634400
+    },
+    {
+      "epoch": 0.16548028135298515,
+      "grad_norm": 10.750335693359375,
+      "learning_rate": 9.466036864603893e-06,
+      "loss": 2.6026,
+      "step": 634600
+    },
+    {
+      "epoch": 0.1655324339786873,
+      "grad_norm": 9.322654724121094,
+      "learning_rate": 9.465660930516698e-06,
+      "loss": 2.5806,
+      "step": 634800
+    },
+    {
+      "epoch": 0.1655845866043895,
+      "grad_norm": 11.006738662719727,
+      "learning_rate": 9.465284871609132e-06,
+      "loss": 2.5799,
+      "step": 635000
+    },
+    {
+      "epoch": 0.16563673923009165,
+      "grad_norm": 11.173583030700684,
+      "learning_rate": 9.464908687891704e-06,
+      "loss": 2.6159,
+      "step": 635200
+    },
+    {
+      "epoch": 0.16568889185579383,
+      "grad_norm": 9.691333770751953,
+      "learning_rate": 9.46453237937493e-06,
+      "loss": 2.5984,
+      "step": 635400
+    },
+    {
+      "epoch": 0.16574104448149599,
+      "grad_norm": 10.239128112792969,
+      "learning_rate": 9.464155946069329e-06,
+      "loss": 2.6037,
+      "step": 635600
+    },
+    {
+      "epoch": 0.16579319710719817,
+      "grad_norm": 11.106518745422363,
+      "learning_rate": 9.463779387985422e-06,
+      "loss": 2.5693,
+      "step": 635800
+    },
+    {
+      "epoch": 0.16584534973290033,
+      "grad_norm": 11.097664833068848,
+      "learning_rate": 9.463402705133735e-06,
+      "loss": 2.567,
+      "step": 636000
+    },
+    {
+      "epoch": 0.1658975023586025,
+      "grad_norm": 10.165889739990234,
+      "learning_rate": 9.463025897524794e-06,
+      "loss": 2.6039,
+      "step": 636200
+    },
+    {
+      "epoch": 0.16594965498430467,
+      "grad_norm": 10.918819427490234,
+      "learning_rate": 9.462648965169133e-06,
+      "loss": 2.6199,
+      "step": 636400
+    },
+    {
+      "epoch": 0.16600180761000682,
+      "grad_norm": 10.222068786621094,
+      "learning_rate": 9.46227190807729e-06,
+      "loss": 2.5954,
+      "step": 636600
+    },
+    {
+      "epoch": 0.166053960235709,
+      "grad_norm": 10.973430633544922,
+      "learning_rate": 9.4618947262598e-06,
+      "loss": 2.5924,
+      "step": 636800
+    },
+    {
+      "epoch": 0.16610611286141116,
+      "grad_norm": 12.050614356994629,
+      "learning_rate": 9.461517419727208e-06,
+      "loss": 2.5885,
+      "step": 637000
+    },
+    {
+      "epoch": 0.16615826548711335,
+      "grad_norm": 10.913813591003418,
+      "learning_rate": 9.46113998849006e-06,
+      "loss": 2.5929,
+      "step": 637200
+    },
+    {
+      "epoch": 0.1662104181128155,
+      "grad_norm": 11.964225769042969,
+      "learning_rate": 9.460762432558905e-06,
+      "loss": 2.5462,
+      "step": 637400
+    },
+    {
+      "epoch": 0.1662625707385177,
+      "grad_norm": 12.226649284362793,
+      "learning_rate": 9.460384751944298e-06,
+      "loss": 2.5944,
+      "step": 637600
+    },
+    {
+      "epoch": 0.16631472336421985,
+      "grad_norm": 11.865588188171387,
+      "learning_rate": 9.460006946656794e-06,
+      "loss": 2.5778,
+      "step": 637800
+    },
+    {
+      "epoch": 0.16636687598992203,
+      "grad_norm": 10.3606538772583,
+      "learning_rate": 9.459629016706952e-06,
+      "loss": 2.5963,
+      "step": 638000
+    },
+    {
+      "epoch": 0.1664190286156242,
+      "grad_norm": 9.922835350036621,
+      "learning_rate": 9.459250962105337e-06,
+      "loss": 2.606,
+      "step": 638200
+    },
+    {
+      "epoch": 0.16647118124132637,
+      "grad_norm": 11.583536148071289,
+      "learning_rate": 9.458872782862516e-06,
+      "loss": 2.5747,
+      "step": 638400
+    },
+    {
+      "epoch": 0.16652333386702853,
+      "grad_norm": 9.941807746887207,
+      "learning_rate": 9.458494478989059e-06,
+      "loss": 2.5838,
+      "step": 638600
+    },
+    {
+      "epoch": 0.1665754864927307,
+      "grad_norm": 11.269411087036133,
+      "learning_rate": 9.458116050495543e-06,
+      "loss": 2.6033,
+      "step": 638800
+    },
+    {
+      "epoch": 0.16662763911843287,
+      "grad_norm": 11.465824127197266,
+      "learning_rate": 9.457737497392541e-06,
+      "loss": 2.5794,
+      "step": 639000
+    },
+    {
+      "epoch": 0.16667979174413505,
+      "grad_norm": 11.089844703674316,
+      "learning_rate": 9.457358819690636e-06,
+      "loss": 2.591,
+      "step": 639200
+    },
+    {
+      "epoch": 0.1667319443698372,
+      "grad_norm": 11.80525016784668,
+      "learning_rate": 9.456980017400413e-06,
+      "loss": 2.5967,
+      "step": 639400
+    },
+    {
+      "epoch": 0.1667840969955394,
+      "grad_norm": 11.84639835357666,
+      "learning_rate": 9.456601090532458e-06,
+      "loss": 2.5895,
+      "step": 639600
+    },
+    {
+      "epoch": 0.16683624962124155,
+      "grad_norm": 10.74103832244873,
+      "learning_rate": 9.456222039097365e-06,
+      "loss": 2.5974,
+      "step": 639800
+    },
+    {
+      "epoch": 0.16688840224694373,
+      "grad_norm": 11.63347053527832,
+      "learning_rate": 9.455842863105728e-06,
+      "loss": 2.6094,
+      "step": 640000
+    },
+    {
+      "epoch": 0.1669405548726459,
+      "grad_norm": 10.795598983764648,
+      "learning_rate": 9.455463562568143e-06,
+      "loss": 2.585,
+      "step": 640200
+    },
+    {
+      "epoch": 0.16699270749834808,
+      "grad_norm": 10.217510223388672,
+      "learning_rate": 9.455084137495216e-06,
+      "loss": 2.6029,
+      "step": 640400
+    },
+    {
+      "epoch": 0.16704486012405023,
+      "grad_norm": 11.189576148986816,
+      "learning_rate": 9.454704587897553e-06,
+      "loss": 2.6158,
+      "step": 640600
+    },
+    {
+      "epoch": 0.16709701274975242,
+      "grad_norm": 12.054612159729004,
+      "learning_rate": 9.454324913785756e-06,
+      "loss": 2.5981,
+      "step": 640800
+    },
+    {
+      "epoch": 0.16714916537545457,
+      "grad_norm": 10.436782836914062,
+      "learning_rate": 9.453945115170444e-06,
+      "loss": 2.5662,
+      "step": 641000
+    },
+    {
+      "epoch": 0.16720131800115676,
+      "grad_norm": 8.611639022827148,
+      "learning_rate": 9.453565192062228e-06,
+      "loss": 2.5866,
+      "step": 641200
+    },
+    {
+      "epoch": 0.1672534706268589,
+      "grad_norm": 12.085579872131348,
+      "learning_rate": 9.453185144471732e-06,
+      "loss": 2.5935,
+      "step": 641400
+    },
+    {
+      "epoch": 0.1673056232525611,
+      "grad_norm": 12.087554931640625,
+      "learning_rate": 9.452804972409576e-06,
+      "loss": 2.6145,
+      "step": 641600
+    },
+    {
+      "epoch": 0.16735777587826325,
+      "grad_norm": 11.656718254089355,
+      "learning_rate": 9.452424675886386e-06,
+      "loss": 2.6006,
+      "step": 641800
+    },
+    {
+      "epoch": 0.16740992850396544,
+      "grad_norm": 9.5364351272583,
+      "learning_rate": 9.452044254912794e-06,
+      "loss": 2.638,
+      "step": 642000
+    },
+    {
+      "epoch": 0.1674620811296676,
+      "grad_norm": 9.897908210754395,
+      "learning_rate": 9.45166370949943e-06,
+      "loss": 2.5881,
+      "step": 642200
+    },
+    {
+      "epoch": 0.16751423375536975,
+      "grad_norm": 10.919352531433105,
+      "learning_rate": 9.451283039656933e-06,
+      "loss": 2.5761,
+      "step": 642400
+    },
+    {
+      "epoch": 0.16756638638107194,
+      "grad_norm": 11.361388206481934,
+      "learning_rate": 9.450902245395943e-06,
+      "loss": 2.6329,
+      "step": 642600
+    },
+    {
+      "epoch": 0.1676185390067741,
+      "grad_norm": 11.634418487548828,
+      "learning_rate": 9.450521326727104e-06,
+      "loss": 2.5698,
+      "step": 642800
+    },
+    {
+      "epoch": 0.16767069163247628,
+      "grad_norm": 11.556270599365234,
+      "learning_rate": 9.45014028366106e-06,
+      "loss": 2.5883,
+      "step": 643000
+    },
+    {
+      "epoch": 0.16772284425817843,
+      "grad_norm": 11.184852600097656,
+      "learning_rate": 9.449759116208465e-06,
+      "loss": 2.6441,
+      "step": 643200
+    },
+    {
+      "epoch": 0.16777499688388062,
+      "grad_norm": 10.851284980773926,
+      "learning_rate": 9.449377824379973e-06,
+      "loss": 2.5469,
+      "step": 643400
+    },
+    {
+      "epoch": 0.16782714950958277,
+      "grad_norm": 11.684367179870605,
+      "learning_rate": 9.448996408186239e-06,
+      "loss": 2.6165,
+      "step": 643600
+    },
+    {
+      "epoch": 0.16787930213528496,
+      "grad_norm": 11.91071891784668,
+      "learning_rate": 9.448614867637926e-06,
+      "loss": 2.6167,
+      "step": 643800
+    },
+    {
+      "epoch": 0.16793145476098711,
+      "grad_norm": 10.489933013916016,
+      "learning_rate": 9.448233202745699e-06,
+      "loss": 2.5944,
+      "step": 644000
+    },
+    {
+      "epoch": 0.1679836073866893,
+      "grad_norm": 9.974836349487305,
+      "learning_rate": 9.447851413520222e-06,
+      "loss": 2.5532,
+      "step": 644200
+    },
+    {
+      "epoch": 0.16803576001239146,
+      "grad_norm": 12.725610733032227,
+      "learning_rate": 9.447469499972172e-06,
+      "loss": 2.5883,
+      "step": 644400
+    },
+    {
+      "epoch": 0.16808791263809364,
+      "grad_norm": 10.390010833740234,
+      "learning_rate": 9.447087462112222e-06,
+      "loss": 2.6363,
+      "step": 644600
+    },
+    {
+      "epoch": 0.1681400652637958,
+      "grad_norm": 11.040779113769531,
+      "learning_rate": 9.446705299951047e-06,
+      "loss": 2.5922,
+      "step": 644800
+    },
+    {
+      "epoch": 0.16819221788949798,
+      "grad_norm": 10.456809997558594,
+      "learning_rate": 9.446323013499333e-06,
+      "loss": 2.587,
+      "step": 645000
+    },
+    {
+      "epoch": 0.16824437051520014,
+      "grad_norm": 12.05203628540039,
+      "learning_rate": 9.445940602767764e-06,
+      "loss": 2.6597,
+      "step": 645200
+    },
+    {
+      "epoch": 0.16829652314090232,
+      "grad_norm": 12.002357482910156,
+      "learning_rate": 9.445558067767028e-06,
+      "loss": 2.6082,
+      "step": 645400
+    },
+    {
+      "epoch": 0.16834867576660448,
+      "grad_norm": 11.387803077697754,
+      "learning_rate": 9.445175408507818e-06,
+      "loss": 2.6126,
+      "step": 645600
+    },
+    {
+      "epoch": 0.16840082839230666,
+      "grad_norm": 11.261042594909668,
+      "learning_rate": 9.444792625000832e-06,
+      "loss": 2.6011,
+      "step": 645800
+    },
+    {
+      "epoch": 0.16845298101800882,
+      "grad_norm": 10.792635917663574,
+      "learning_rate": 9.444409717256766e-06,
+      "loss": 2.6017,
+      "step": 646000
+    },
+    {
+      "epoch": 0.168505133643711,
+      "grad_norm": 10.310630798339844,
+      "learning_rate": 9.444026685286323e-06,
+      "loss": 2.6158,
+      "step": 646200
+    },
+    {
+      "epoch": 0.16855728626941316,
+      "grad_norm": 11.336012840270996,
+      "learning_rate": 9.443643529100211e-06,
+      "loss": 2.5637,
+      "step": 646400
+    },
+    {
+      "epoch": 0.16860943889511534,
+      "grad_norm": 10.27746295928955,
+      "learning_rate": 9.443260248709138e-06,
+      "loss": 2.5942,
+      "step": 646600
+    },
+    {
+      "epoch": 0.1686615915208175,
+      "grad_norm": 11.652708053588867,
+      "learning_rate": 9.442876844123818e-06,
+      "loss": 2.5802,
+      "step": 646800
+    },
+    {
+      "epoch": 0.16871374414651968,
+      "grad_norm": 10.635597229003906,
+      "learning_rate": 9.442493315354966e-06,
+      "loss": 2.6022,
+      "step": 647000
+    },
+    {
+      "epoch": 0.16876589677222184,
+      "grad_norm": 12.215426445007324,
+      "learning_rate": 9.442109662413306e-06,
+      "loss": 2.603,
+      "step": 647200
+    },
+    {
+      "epoch": 0.16881804939792402,
+      "grad_norm": 10.632146835327148,
+      "learning_rate": 9.441725885309558e-06,
+      "loss": 2.5967,
+      "step": 647400
+    },
+    {
+      "epoch": 0.16887020202362618,
+      "grad_norm": 12.226157188415527,
+      "learning_rate": 9.441341984054448e-06,
+      "loss": 2.6191,
+      "step": 647600
+    },
+    {
+      "epoch": 0.16892235464932837,
+      "grad_norm": 11.939155578613281,
+      "learning_rate": 9.440957958658712e-06,
+      "loss": 2.5977,
+      "step": 647800
+    },
+    {
+      "epoch": 0.16897450727503052,
+      "grad_norm": 11.38477611541748,
+      "learning_rate": 9.440573809133078e-06,
+      "loss": 2.6372,
+      "step": 648000
+    },
+    {
+      "epoch": 0.16902665990073268,
+      "grad_norm": 10.353914260864258,
+      "learning_rate": 9.440189535488286e-06,
+      "loss": 2.6133,
+      "step": 648200
+    },
+    {
+      "epoch": 0.16907881252643486,
+      "grad_norm": 10.853445053100586,
+      "learning_rate": 9.439805137735078e-06,
+      "loss": 2.5439,
+      "step": 648400
+    },
+    {
+      "epoch": 0.16913096515213702,
+      "grad_norm": 10.450331687927246,
+      "learning_rate": 9.439420615884196e-06,
+      "loss": 2.6195,
+      "step": 648600
+    },
+    {
+      "epoch": 0.1691831177778392,
+      "grad_norm": 11.067255973815918,
+      "learning_rate": 9.43903596994639e-06,
+      "loss": 2.5792,
+      "step": 648800
+    },
+    {
+      "epoch": 0.16923527040354136,
+      "grad_norm": 10.674095153808594,
+      "learning_rate": 9.43865119993241e-06,
+      "loss": 2.5886,
+      "step": 649000
+    },
+    {
+      "epoch": 0.16928742302924354,
+      "grad_norm": 11.1282377243042,
+      "learning_rate": 9.43826630585301e-06,
+      "loss": 2.5493,
+      "step": 649200
+    },
+    {
+      "epoch": 0.1693395756549457,
+      "grad_norm": 11.285887718200684,
+      "learning_rate": 9.43788128771895e-06,
+      "loss": 2.5827,
+      "step": 649400
+    },
+    {
+      "epoch": 0.16939172828064789,
+      "grad_norm": 12.13648509979248,
+      "learning_rate": 9.43749614554099e-06,
+      "loss": 2.589,
+      "step": 649600
+    },
+    {
+      "epoch": 0.16944388090635004,
+      "grad_norm": 11.782530784606934,
+      "learning_rate": 9.437110879329897e-06,
+      "loss": 2.5266,
+      "step": 649800
+    },
+    {
+      "epoch": 0.16949603353205223,
+      "grad_norm": 10.601338386535645,
+      "learning_rate": 9.436725489096438e-06,
+      "loss": 2.6385,
+      "step": 650000
+    },
+    {
+      "epoch": 0.16954818615775438,
+      "grad_norm": 10.485753059387207,
+      "learning_rate": 9.436339974851388e-06,
+      "loss": 2.5673,
+      "step": 650200
+    },
+    {
+      "epoch": 0.16960033878345657,
+      "grad_norm": 10.410168647766113,
+      "learning_rate": 9.435954336605518e-06,
+      "loss": 2.6,
+      "step": 650400
+    },
+    {
+      "epoch": 0.16965249140915872,
+      "grad_norm": 11.205649375915527,
+      "learning_rate": 9.43556857436961e-06,
+      "loss": 2.5582,
+      "step": 650600
+    },
+    {
+      "epoch": 0.1697046440348609,
+      "grad_norm": 11.990690231323242,
+      "learning_rate": 9.435182688154447e-06,
+      "loss": 2.6067,
+      "step": 650800
+    },
+    {
+      "epoch": 0.16975679666056306,
+      "grad_norm": 12.044350624084473,
+      "learning_rate": 9.434796677970813e-06,
+      "loss": 2.5565,
+      "step": 651000
+    },
+    {
+      "epoch": 0.16980894928626525,
+      "grad_norm": 12.324272155761719,
+      "learning_rate": 9.434410543829497e-06,
+      "loss": 2.5784,
+      "step": 651200
+    },
+    {
+      "epoch": 0.1698611019119674,
+      "grad_norm": 10.339320182800293,
+      "learning_rate": 9.434024285741295e-06,
+      "loss": 2.5863,
+      "step": 651400
+    },
+    {
+      "epoch": 0.1699132545376696,
+      "grad_norm": 10.739986419677734,
+      "learning_rate": 9.433637903717e-06,
+      "loss": 2.6013,
+      "step": 651600
+    },
+    {
+      "epoch": 0.16996540716337175,
+      "grad_norm": 11.532021522521973,
+      "learning_rate": 9.433251397767415e-06,
+      "loss": 2.5864,
+      "step": 651800
+    },
+    {
+      "epoch": 0.17001755978907393,
+      "grad_norm": 11.322754859924316,
+      "learning_rate": 9.432864767903342e-06,
+      "loss": 2.5692,
+      "step": 652000
+    },
+    {
+      "epoch": 0.1700697124147761,
+      "grad_norm": 10.201240539550781,
+      "learning_rate": 9.432478014135587e-06,
+      "loss": 2.579,
+      "step": 652200
+    },
+    {
+      "epoch": 0.17012186504047827,
+      "grad_norm": 10.831350326538086,
+      "learning_rate": 9.432091136474961e-06,
+      "loss": 2.5776,
+      "step": 652400
+    },
+    {
+      "epoch": 0.17017401766618043,
+      "grad_norm": 10.44613265991211,
+      "learning_rate": 9.431704134932276e-06,
+      "loss": 2.5915,
+      "step": 652600
+    },
+    {
+      "epoch": 0.1702261702918826,
+      "grad_norm": 10.422142028808594,
+      "learning_rate": 9.431317009518352e-06,
+      "loss": 2.5365,
+      "step": 652800
+    },
+    {
+      "epoch": 0.17027832291758477,
+      "grad_norm": 12.878448486328125,
+      "learning_rate": 9.430929760244007e-06,
+      "loss": 2.6059,
+      "step": 653000
+    },
+    {
+      "epoch": 0.17033047554328695,
+      "grad_norm": 10.190254211425781,
+      "learning_rate": 9.430542387120068e-06,
+      "loss": 2.5466,
+      "step": 653200
+    },
+    {
+      "epoch": 0.1703826281689891,
+      "grad_norm": 11.413830757141113,
+      "learning_rate": 9.430154890157359e-06,
+      "loss": 2.5667,
+      "step": 653400
+    },
+    {
+      "epoch": 0.1704347807946913,
+      "grad_norm": 9.806634902954102,
+      "learning_rate": 9.429767269366712e-06,
+      "loss": 2.5951,
+      "step": 653600
+    },
+    {
+      "epoch": 0.17048693342039345,
+      "grad_norm": 11.987215042114258,
+      "learning_rate": 9.429379524758963e-06,
+      "loss": 2.5993,
+      "step": 653800
+    },
+    {
+      "epoch": 0.1705390860460956,
+      "grad_norm": 11.944783210754395,
+      "learning_rate": 9.42899165634495e-06,
+      "loss": 2.6648,
+      "step": 654000
+    },
+    {
+      "epoch": 0.1705912386717978,
+      "grad_norm": 10.974967956542969,
+      "learning_rate": 9.428603664135511e-06,
+      "loss": 2.5808,
+      "step": 654200
+    },
+    {
+      "epoch": 0.17064339129749995,
+      "grad_norm": 11.658656120300293,
+      "learning_rate": 9.428215548141495e-06,
+      "loss": 2.5701,
+      "step": 654400
+    },
+    {
+      "epoch": 0.17069554392320213,
+      "grad_norm": 10.598328590393066,
+      "learning_rate": 9.427827308373747e-06,
+      "loss": 2.6166,
+      "step": 654600
+    },
+    {
+      "epoch": 0.1707476965489043,
+      "grad_norm": 11.660055160522461,
+      "learning_rate": 9.427438944843121e-06,
+      "loss": 2.6142,
+      "step": 654800
+    },
+    {
+      "epoch": 0.17079984917460647,
+      "grad_norm": 11.008819580078125,
+      "learning_rate": 9.427050457560472e-06,
+      "loss": 2.6348,
+      "step": 655000
+    },
+    {
+      "epoch": 0.17085200180030863,
+      "grad_norm": 10.61733341217041,
+      "learning_rate": 9.42666184653666e-06,
+      "loss": 2.6307,
+      "step": 655200
+    },
+    {
+      "epoch": 0.1709041544260108,
+      "grad_norm": 10.868151664733887,
+      "learning_rate": 9.426273111782543e-06,
+      "loss": 2.5962,
+      "step": 655400
+    },
+    {
+      "epoch": 0.17095630705171297,
+      "grad_norm": 11.051012992858887,
+      "learning_rate": 9.425884253308988e-06,
+      "loss": 2.625,
+      "step": 655600
+    },
+    {
+      "epoch": 0.17100845967741515,
+      "grad_norm": 11.177716255187988,
+      "learning_rate": 9.425495271126865e-06,
+      "loss": 2.5808,
+      "step": 655800
+    },
+    {
+      "epoch": 0.1710606123031173,
+      "grad_norm": 10.845020294189453,
+      "learning_rate": 9.425106165247048e-06,
+      "loss": 2.5558,
+      "step": 656000
+    },
+    {
+      "epoch": 0.1711127649288195,
+      "grad_norm": 10.903834342956543,
+      "learning_rate": 9.42471693568041e-06,
+      "loss": 2.6147,
+      "step": 656200
+    },
+    {
+      "epoch": 0.17116491755452165,
+      "grad_norm": 9.43630599975586,
+      "learning_rate": 9.424327582437833e-06,
+      "loss": 2.6235,
+      "step": 656400
+    },
+    {
+      "epoch": 0.17121707018022383,
+      "grad_norm": 10.715693473815918,
+      "learning_rate": 9.423938105530197e-06,
+      "loss": 2.6456,
+      "step": 656600
+    },
+    {
+      "epoch": 0.171269222805926,
+      "grad_norm": 10.508771896362305,
+      "learning_rate": 9.423548504968392e-06,
+      "loss": 2.578,
+      "step": 656800
+    },
+    {
+      "epoch": 0.17132137543162818,
+      "grad_norm": 10.303911209106445,
+      "learning_rate": 9.423158780763304e-06,
+      "loss": 2.5929,
+      "step": 657000
+    },
+    {
+      "epoch": 0.17137352805733033,
+      "grad_norm": 11.929397583007812,
+      "learning_rate": 9.422768932925828e-06,
+      "loss": 2.5516,
+      "step": 657200
+    },
+    {
+      "epoch": 0.17142568068303252,
+      "grad_norm": 11.60670280456543,
+      "learning_rate": 9.422378961466863e-06,
+      "loss": 2.6122,
+      "step": 657400
+    },
+    {
+      "epoch": 0.17147783330873467,
+      "grad_norm": 10.905845642089844,
+      "learning_rate": 9.421988866397304e-06,
+      "loss": 2.5991,
+      "step": 657600
+    },
+    {
+      "epoch": 0.17152998593443686,
+      "grad_norm": 9.888165473937988,
+      "learning_rate": 9.421598647728059e-06,
+      "loss": 2.5957,
+      "step": 657800
+    },
+    {
+      "epoch": 0.171582138560139,
+      "grad_norm": 10.218853950500488,
+      "learning_rate": 9.421208305470033e-06,
+      "loss": 2.5762,
+      "step": 658000
+    },
+    {
+      "epoch": 0.1716342911858412,
+      "grad_norm": 10.51092529296875,
+      "learning_rate": 9.420817839634138e-06,
+      "loss": 2.6286,
+      "step": 658200
+    },
+    {
+      "epoch": 0.17168644381154335,
+      "grad_norm": 12.242321968078613,
+      "learning_rate": 9.420427250231286e-06,
+      "loss": 2.5824,
+      "step": 658400
+    },
+    {
+      "epoch": 0.17173859643724554,
+      "grad_norm": 9.587675094604492,
+      "learning_rate": 9.420036537272395e-06,
+      "loss": 2.6159,
+      "step": 658600
+    },
+    {
+      "epoch": 0.1717907490629477,
+      "grad_norm": 12.193526268005371,
+      "learning_rate": 9.419645700768386e-06,
+      "loss": 2.5814,
+      "step": 658800
+    },
+    {
+      "epoch": 0.17184290168864988,
+      "grad_norm": 13.005501747131348,
+      "learning_rate": 9.419254740730184e-06,
+      "loss": 2.6249,
+      "step": 659000
+    },
+    {
+      "epoch": 0.17189505431435204,
+      "grad_norm": 12.042132377624512,
+      "learning_rate": 9.418863657168717e-06,
+      "loss": 2.592,
+      "step": 659200
+    },
+    {
+      "epoch": 0.17194720694005422,
+      "grad_norm": 11.057788848876953,
+      "learning_rate": 9.418472450094915e-06,
+      "loss": 2.6023,
+      "step": 659400
+    },
+    {
+      "epoch": 0.17199935956575638,
+      "grad_norm": 11.42861557006836,
+      "learning_rate": 9.418081119519715e-06,
+      "loss": 2.6074,
+      "step": 659600
+    },
+    {
+      "epoch": 0.17205151219145856,
+      "grad_norm": 10.534041404724121,
+      "learning_rate": 9.417689665454052e-06,
+      "loss": 2.6056,
+      "step": 659800
+    },
+    {
+      "epoch": 0.17210366481716072,
+      "grad_norm": 10.17191219329834,
+      "learning_rate": 9.417298087908869e-06,
+      "loss": 2.5732,
+      "step": 660000
+    },
+    {
+      "epoch": 0.17215581744286287,
+      "grad_norm": 10.42569351196289,
+      "learning_rate": 9.41690638689511e-06,
+      "loss": 2.5551,
+      "step": 660200
+    },
+    {
+      "epoch": 0.17220797006856506,
+      "grad_norm": 10.62281608581543,
+      "learning_rate": 9.416514562423726e-06,
+      "loss": 2.5706,
+      "step": 660400
+    },
+    {
+      "epoch": 0.17226012269426721,
+      "grad_norm": 10.042311668395996,
+      "learning_rate": 9.416122614505669e-06,
+      "loss": 2.5726,
+      "step": 660600
+    },
+    {
+      "epoch": 0.1723122753199694,
+      "grad_norm": 11.639966011047363,
+      "learning_rate": 9.41573054315189e-06,
+      "loss": 2.6004,
+      "step": 660800
+    },
+    {
+      "epoch": 0.17236442794567156,
+      "grad_norm": 12.120760917663574,
+      "learning_rate": 9.415338348373354e-06,
+      "loss": 2.5984,
+      "step": 661000
+    },
+    {
+      "epoch": 0.17241658057137374,
+      "grad_norm": 10.926827430725098,
+      "learning_rate": 9.414946030181018e-06,
+      "loss": 2.5645,
+      "step": 661200
+    },
+    {
+      "epoch": 0.1724687331970759,
+      "grad_norm": 12.61771011352539,
+      "learning_rate": 9.414553588585849e-06,
+      "loss": 2.6064,
+      "step": 661400
+    },
+    {
+      "epoch": 0.17252088582277808,
+      "grad_norm": 11.024273872375488,
+      "learning_rate": 9.41416102359882e-06,
+      "loss": 2.6001,
+      "step": 661600
+    },
+    {
+      "epoch": 0.17257303844848024,
+      "grad_norm": 12.414682388305664,
+      "learning_rate": 9.413768335230897e-06,
+      "loss": 2.5993,
+      "step": 661800
+    },
+    {
+      "epoch": 0.17262519107418242,
+      "grad_norm": 12.029796600341797,
+      "learning_rate": 9.413375523493062e-06,
+      "loss": 2.542,
+      "step": 662000
+    },
+    {
+      "epoch": 0.17267734369988458,
+      "grad_norm": 10.11413288116455,
+      "learning_rate": 9.412982588396292e-06,
+      "loss": 2.5833,
+      "step": 662200
+    },
+    {
+      "epoch": 0.17272949632558676,
+      "grad_norm": 11.750227928161621,
+      "learning_rate": 9.41258952995157e-06,
+      "loss": 2.5748,
+      "step": 662400
+    },
+    {
+      "epoch": 0.17278164895128892,
+      "grad_norm": 10.92453670501709,
+      "learning_rate": 9.412196348169883e-06,
+      "loss": 2.5841,
+      "step": 662600
+    },
+    {
+      "epoch": 0.1728338015769911,
+      "grad_norm": 11.723861694335938,
+      "learning_rate": 9.411803043062222e-06,
+      "loss": 2.5925,
+      "step": 662800
+    },
+    {
+      "epoch": 0.17288595420269326,
+      "grad_norm": 10.404121398925781,
+      "learning_rate": 9.411409614639576e-06,
+      "loss": 2.5609,
+      "step": 663000
+    },
+    {
+      "epoch": 0.17293810682839544,
+      "grad_norm": 10.455796241760254,
+      "learning_rate": 9.411016062912946e-06,
+      "loss": 2.6097,
+      "step": 663200
+    },
+    {
+      "epoch": 0.1729902594540976,
+      "grad_norm": 10.422048568725586,
+      "learning_rate": 9.410622387893332e-06,
+      "loss": 2.6332,
+      "step": 663400
+    },
+    {
+      "epoch": 0.17304241207979978,
+      "grad_norm": 10.72389030456543,
+      "learning_rate": 9.410228589591735e-06,
+      "loss": 2.5613,
+      "step": 663600
+    },
+    {
+      "epoch": 0.17309456470550194,
+      "grad_norm": 10.777589797973633,
+      "learning_rate": 9.409834668019165e-06,
+      "loss": 2.56,
+      "step": 663800
+    },
+    {
+      "epoch": 0.17314671733120413,
+      "grad_norm": 11.299818992614746,
+      "learning_rate": 9.40944062318663e-06,
+      "loss": 2.6022,
+      "step": 664000
+    },
+    {
+      "epoch": 0.17319886995690628,
+      "grad_norm": 12.202249526977539,
+      "learning_rate": 9.409046455105146e-06,
+      "loss": 2.6085,
+      "step": 664200
+    },
+    {
+      "epoch": 0.17325102258260847,
+      "grad_norm": 10.913101196289062,
+      "learning_rate": 9.40865216378573e-06,
+      "loss": 2.5667,
+      "step": 664400
+    },
+    {
+      "epoch": 0.17330317520831062,
+      "grad_norm": 12.865901947021484,
+      "learning_rate": 9.408257749239402e-06,
+      "loss": 2.6148,
+      "step": 664600
+    },
+    {
+      "epoch": 0.1733553278340128,
+      "grad_norm": 11.728925704956055,
+      "learning_rate": 9.407863211477189e-06,
+      "loss": 2.5964,
+      "step": 664800
+    },
+    {
+      "epoch": 0.17340748045971496,
+      "grad_norm": 10.57736587524414,
+      "learning_rate": 9.407468550510114e-06,
+      "loss": 2.5577,
+      "step": 665000
+    },
+    {
+      "epoch": 0.17345963308541715,
+      "grad_norm": 11.458696365356445,
+      "learning_rate": 9.407073766349213e-06,
+      "loss": 2.5685,
+      "step": 665200
+    },
+    {
+      "epoch": 0.1735117857111193,
+      "grad_norm": 10.13565444946289,
+      "learning_rate": 9.406678859005518e-06,
+      "loss": 2.5543,
+      "step": 665400
+    },
+    {
+      "epoch": 0.1735639383368215,
+      "grad_norm": 11.961712837219238,
+      "learning_rate": 9.406283828490068e-06,
+      "loss": 2.5873,
+      "step": 665600
+    },
+    {
+      "epoch": 0.17361609096252364,
+      "grad_norm": 12.306097030639648,
+      "learning_rate": 9.405888674813903e-06,
+      "loss": 2.6193,
+      "step": 665800
+    },
+    {
+      "epoch": 0.1736682435882258,
+      "grad_norm": 11.791540145874023,
+      "learning_rate": 9.40549339798807e-06,
+      "loss": 2.575,
+      "step": 666000
+    },
+    {
+      "epoch": 0.17372039621392799,
+      "grad_norm": 12.87740421295166,
+      "learning_rate": 9.405097998023618e-06,
+      "loss": 2.5365,
+      "step": 666200
+    },
+    {
+      "epoch": 0.17377254883963014,
+      "grad_norm": 13.530356407165527,
+      "learning_rate": 9.404702474931598e-06,
+      "loss": 2.5866,
+      "step": 666400
+    },
+    {
+      "epoch": 0.17382470146533233,
+      "grad_norm": 10.974345207214355,
+      "learning_rate": 9.404306828723063e-06,
+      "loss": 2.6068,
+      "step": 666600
+    },
+    {
+      "epoch": 0.17387685409103448,
+      "grad_norm": 12.054535865783691,
+      "learning_rate": 9.403911059409075e-06,
+      "loss": 2.6092,
+      "step": 666800
+    },
+    {
+      "epoch": 0.17392900671673667,
+      "grad_norm": 10.9446382522583,
+      "learning_rate": 9.403515167000694e-06,
+      "loss": 2.5992,
+      "step": 667000
+    },
+    {
+      "epoch": 0.17398115934243882,
+      "grad_norm": 10.59375286102295,
+      "learning_rate": 9.403119151508988e-06,
+      "loss": 2.5542,
+      "step": 667200
+    },
+    {
+      "epoch": 0.174033311968141,
+      "grad_norm": 10.471702575683594,
+      "learning_rate": 9.402723012945023e-06,
+      "loss": 2.6148,
+      "step": 667400
+    },
+    {
+      "epoch": 0.17408546459384316,
+      "grad_norm": 11.591514587402344,
+      "learning_rate": 9.402326751319875e-06,
+      "loss": 2.5909,
+      "step": 667600
+    },
+    {
+      "epoch": 0.17413761721954535,
+      "grad_norm": 11.10065746307373,
+      "learning_rate": 9.401930366644617e-06,
+      "loss": 2.5856,
+      "step": 667800
+    },
+    {
+      "epoch": 0.1741897698452475,
+      "grad_norm": 11.09948444366455,
+      "learning_rate": 9.401533858930331e-06,
+      "loss": 2.6074,
+      "step": 668000
+    },
+    {
+      "epoch": 0.1742419224709497,
+      "grad_norm": 10.385692596435547,
+      "learning_rate": 9.401137228188098e-06,
+      "loss": 2.5725,
+      "step": 668200
+    },
+    {
+      "epoch": 0.17429407509665185,
+      "grad_norm": 8.567261695861816,
+      "learning_rate": 9.400740474429004e-06,
+      "loss": 2.5714,
+      "step": 668400
+    },
+    {
+      "epoch": 0.17434622772235403,
+      "grad_norm": 10.014311790466309,
+      "learning_rate": 9.40034359766414e-06,
+      "loss": 2.5659,
+      "step": 668600
+    },
+    {
+      "epoch": 0.1743983803480562,
+      "grad_norm": 11.485774040222168,
+      "learning_rate": 9.3999465979046e-06,
+      "loss": 2.5965,
+      "step": 668800
+    },
+    {
+      "epoch": 0.17445053297375837,
+      "grad_norm": 10.644266128540039,
+      "learning_rate": 9.399549475161476e-06,
+      "loss": 2.61,
+      "step": 669000
+    },
+    {
+      "epoch": 0.17450268559946053,
+      "grad_norm": 12.157289505004883,
+      "learning_rate": 9.399152229445874e-06,
+      "loss": 2.5701,
+      "step": 669200
+    },
+    {
+      "epoch": 0.1745548382251627,
+      "grad_norm": 9.978353500366211,
+      "learning_rate": 9.398754860768892e-06,
+      "loss": 2.5829,
+      "step": 669400
+    },
+    {
+      "epoch": 0.17460699085086487,
+      "grad_norm": 11.341612815856934,
+      "learning_rate": 9.398357369141641e-06,
+      "loss": 2.6211,
+      "step": 669600
+    },
+    {
+      "epoch": 0.17465914347656705,
+      "grad_norm": 10.040153503417969,
+      "learning_rate": 9.397959754575232e-06,
+      "loss": 2.6005,
+      "step": 669800
+    },
+    {
+      "epoch": 0.1747112961022692,
+      "grad_norm": 12.061646461486816,
+      "learning_rate": 9.397562017080774e-06,
+      "loss": 2.608,
+      "step": 670000
+    },
+    {
+      "epoch": 0.1747634487279714,
+      "grad_norm": 11.82841968536377,
+      "learning_rate": 9.39716415666939e-06,
+      "loss": 2.5988,
+      "step": 670200
+    },
+    {
+      "epoch": 0.17481560135367355,
+      "grad_norm": 9.31937313079834,
+      "learning_rate": 9.396766173352196e-06,
+      "loss": 2.585,
+      "step": 670400
+    },
+    {
+      "epoch": 0.17486775397937573,
+      "grad_norm": 11.591330528259277,
+      "learning_rate": 9.396368067140318e-06,
+      "loss": 2.5517,
+      "step": 670600
+    },
+    {
+      "epoch": 0.1749199066050779,
+      "grad_norm": 10.716523170471191,
+      "learning_rate": 9.395969838044882e-06,
+      "loss": 2.5751,
+      "step": 670800
+    },
+    {
+      "epoch": 0.17497205923078007,
+      "grad_norm": 9.377893447875977,
+      "learning_rate": 9.395571486077024e-06,
+      "loss": 2.5578,
+      "step": 671000
+    },
+    {
+      "epoch": 0.17502421185648223,
+      "grad_norm": 10.563931465148926,
+      "learning_rate": 9.395173011247871e-06,
+      "loss": 2.5933,
+      "step": 671200
+    },
+    {
+      "epoch": 0.17507636448218442,
+      "grad_norm": 12.169573783874512,
+      "learning_rate": 9.394774413568565e-06,
+      "loss": 2.5995,
+      "step": 671400
+    },
+    {
+      "epoch": 0.17512851710788657,
+      "grad_norm": 10.579012870788574,
+      "learning_rate": 9.394375693050248e-06,
+      "loss": 2.6174,
+      "step": 671600
+    },
+    {
+      "epoch": 0.17518066973358873,
+      "grad_norm": 11.727015495300293,
+      "learning_rate": 9.393976849704063e-06,
+      "loss": 2.5779,
+      "step": 671800
+    },
+    {
+      "epoch": 0.1752328223592909,
+      "grad_norm": 11.265336036682129,
+      "learning_rate": 9.393577883541158e-06,
+      "loss": 2.6134,
+      "step": 672000
+    },
+    {
+      "epoch": 0.17528497498499307,
+      "grad_norm": 10.374798774719238,
+      "learning_rate": 9.393178794572687e-06,
+      "loss": 2.6363,
+      "step": 672200
+    },
+    {
+      "epoch": 0.17533712761069525,
+      "grad_norm": 12.196139335632324,
+      "learning_rate": 9.392779582809802e-06,
+      "loss": 2.599,
+      "step": 672400
+    },
+    {
+      "epoch": 0.1753892802363974,
+      "grad_norm": 10.827107429504395,
+      "learning_rate": 9.392380248263661e-06,
+      "loss": 2.5711,
+      "step": 672600
+    },
+    {
+      "epoch": 0.1754414328620996,
+      "grad_norm": 11.606658935546875,
+      "learning_rate": 9.39198079094543e-06,
+      "loss": 2.5836,
+      "step": 672800
+    },
+    {
+      "epoch": 0.17549358548780175,
+      "grad_norm": 11.592090606689453,
+      "learning_rate": 9.39158121086627e-06,
+      "loss": 2.5763,
+      "step": 673000
+    },
+    {
+      "epoch": 0.17554573811350394,
+      "grad_norm": 9.465235710144043,
+      "learning_rate": 9.391181508037352e-06,
+      "loss": 2.5332,
+      "step": 673200
+    },
+    {
+      "epoch": 0.1755978907392061,
+      "grad_norm": 11.990010261535645,
+      "learning_rate": 9.390781682469848e-06,
+      "loss": 2.5785,
+      "step": 673400
+    },
+    {
+      "epoch": 0.17565004336490828,
+      "grad_norm": 11.225507736206055,
+      "learning_rate": 9.390381734174933e-06,
+      "loss": 2.5475,
+      "step": 673600
+    },
+    {
+      "epoch": 0.17570219599061043,
+      "grad_norm": 12.462555885314941,
+      "learning_rate": 9.389981663163786e-06,
+      "loss": 2.628,
+      "step": 673800
+    },
+    {
+      "epoch": 0.17575434861631262,
+      "grad_norm": 12.385912895202637,
+      "learning_rate": 9.389581469447591e-06,
+      "loss": 2.5793,
+      "step": 674000
+    },
+    {
+      "epoch": 0.17580650124201477,
+      "grad_norm": 10.774480819702148,
+      "learning_rate": 9.38918115303753e-06,
+      "loss": 2.5479,
+      "step": 674200
+    },
+    {
+      "epoch": 0.17585865386771696,
+      "grad_norm": 11.743460655212402,
+      "learning_rate": 9.388780713944795e-06,
+      "loss": 2.5841,
+      "step": 674400
+    },
+    {
+      "epoch": 0.17591080649341911,
+      "grad_norm": 10.81772232055664,
+      "learning_rate": 9.388380152180582e-06,
+      "loss": 2.568,
+      "step": 674600
+    },
+    {
+      "epoch": 0.1759629591191213,
+      "grad_norm": 11.486669540405273,
+      "learning_rate": 9.387979467756081e-06,
+      "loss": 2.5902,
+      "step": 674800
+    },
+    {
+      "epoch": 0.17601511174482345,
+      "grad_norm": 10.133848190307617,
+      "learning_rate": 9.387578660682495e-06,
+      "loss": 2.5696,
+      "step": 675000
+    },
+    {
+      "epoch": 0.17606726437052564,
+      "grad_norm": 11.145345687866211,
+      "learning_rate": 9.387177730971027e-06,
+      "loss": 2.5567,
+      "step": 675200
+    },
+    {
+      "epoch": 0.1761194169962278,
+      "grad_norm": 11.252696990966797,
+      "learning_rate": 9.386776678632881e-06,
+      "loss": 2.5772,
+      "step": 675400
+    },
+    {
+      "epoch": 0.17617156962192998,
+      "grad_norm": 10.816537857055664,
+      "learning_rate": 9.38637550367927e-06,
+      "loss": 2.6356,
+      "step": 675600
+    },
+    {
+      "epoch": 0.17622372224763214,
+      "grad_norm": 11.040006637573242,
+      "learning_rate": 9.385974206121407e-06,
+      "loss": 2.558,
+      "step": 675800
+    },
+    {
+      "epoch": 0.17627587487333432,
+      "grad_norm": 10.698840141296387,
+      "learning_rate": 9.385572785970505e-06,
+      "loss": 2.5948,
+      "step": 676000
+    },
+    {
+      "epoch": 0.17632802749903648,
+      "grad_norm": 11.69105339050293,
+      "learning_rate": 9.38517124323779e-06,
+      "loss": 2.6081,
+      "step": 676200
+    },
+    {
+      "epoch": 0.17638018012473866,
+      "grad_norm": 11.279681205749512,
+      "learning_rate": 9.38476957793448e-06,
+      "loss": 2.6172,
+      "step": 676400
+    },
+    {
+      "epoch": 0.17643233275044082,
+      "grad_norm": 10.574265480041504,
+      "learning_rate": 9.384367790071805e-06,
+      "loss": 2.603,
+      "step": 676600
+    },
+    {
+      "epoch": 0.176484485376143,
+      "grad_norm": 13.000247955322266,
+      "learning_rate": 9.383965879660995e-06,
+      "loss": 2.5997,
+      "step": 676800
+    },
+    {
+      "epoch": 0.17653663800184516,
+      "grad_norm": 10.329712867736816,
+      "learning_rate": 9.383563846713284e-06,
+      "loss": 2.5801,
+      "step": 677000
+    },
+    {
+      "epoch": 0.17658879062754734,
+      "grad_norm": 10.054405212402344,
+      "learning_rate": 9.383161691239909e-06,
+      "loss": 2.5559,
+      "step": 677200
+    },
+    {
+      "epoch": 0.1766409432532495,
+      "grad_norm": 10.512011528015137,
+      "learning_rate": 9.382759413252112e-06,
+      "loss": 2.5955,
+      "step": 677400
+    },
+    {
+      "epoch": 0.17669309587895166,
+      "grad_norm": 11.089954376220703,
+      "learning_rate": 9.382357012761133e-06,
+      "loss": 2.565,
+      "step": 677600
+    },
+    {
+      "epoch": 0.17674524850465384,
+      "grad_norm": 9.8038330078125,
+      "learning_rate": 9.381954489778225e-06,
+      "loss": 2.5907,
+      "step": 677800
+    },
+    {
+      "epoch": 0.176797401130356,
+      "grad_norm": 15.173497200012207,
+      "learning_rate": 9.381551844314637e-06,
+      "loss": 2.6021,
+      "step": 678000
+    },
+    {
+      "epoch": 0.17684955375605818,
+      "grad_norm": 12.605514526367188,
+      "learning_rate": 9.38114907638162e-06,
+      "loss": 2.6105,
+      "step": 678200
+    },
+    {
+      "epoch": 0.17690170638176034,
+      "grad_norm": 10.773606300354004,
+      "learning_rate": 9.380746185990435e-06,
+      "loss": 2.5503,
+      "step": 678400
+    },
+    {
+      "epoch": 0.17695385900746252,
+      "grad_norm": 12.675154685974121,
+      "learning_rate": 9.380343173152344e-06,
+      "loss": 2.5884,
+      "step": 678600
+    },
+    {
+      "epoch": 0.17700601163316468,
+      "grad_norm": 11.189080238342285,
+      "learning_rate": 9.379940037878611e-06,
+      "loss": 2.5901,
+      "step": 678800
+    },
+    {
+      "epoch": 0.17705816425886686,
+      "grad_norm": 11.658578872680664,
+      "learning_rate": 9.379536780180503e-06,
+      "loss": 2.6359,
+      "step": 679000
+    },
+    {
+      "epoch": 0.17711031688456902,
+      "grad_norm": 11.764540672302246,
+      "learning_rate": 9.379133400069292e-06,
+      "loss": 2.572,
+      "step": 679200
+    },
+    {
+      "epoch": 0.1771624695102712,
+      "grad_norm": 12.119824409484863,
+      "learning_rate": 9.378729897556253e-06,
+      "loss": 2.581,
+      "step": 679400
+    },
+    {
+      "epoch": 0.17721462213597336,
+      "grad_norm": 12.850314140319824,
+      "learning_rate": 9.378326272652666e-06,
+      "loss": 2.6052,
+      "step": 679600
+    },
+    {
+      "epoch": 0.17726677476167554,
+      "grad_norm": 11.843114852905273,
+      "learning_rate": 9.37792252536981e-06,
+      "loss": 2.6066,
+      "step": 679800
+    },
+    {
+      "epoch": 0.1773189273873777,
+      "grad_norm": 11.768165588378906,
+      "learning_rate": 9.377518655718971e-06,
+      "loss": 2.5708,
+      "step": 680000
+    },
+    {
+      "epoch": 0.17737108001307988,
+      "grad_norm": 11.470100402832031,
+      "learning_rate": 9.37711466371144e-06,
+      "loss": 2.5771,
+      "step": 680200
+    },
+    {
+      "epoch": 0.17742323263878204,
+      "grad_norm": 11.421998023986816,
+      "learning_rate": 9.376710549358505e-06,
+      "loss": 2.602,
+      "step": 680400
+    },
+    {
+      "epoch": 0.17747538526448423,
+      "grad_norm": 11.222713470458984,
+      "learning_rate": 9.376306312671465e-06,
+      "loss": 2.6073,
+      "step": 680600
+    },
+    {
+      "epoch": 0.17752753789018638,
+      "grad_norm": 11.578697204589844,
+      "learning_rate": 9.375901953661618e-06,
+      "loss": 2.6128,
+      "step": 680800
+    },
+    {
+      "epoch": 0.17757969051588857,
+      "grad_norm": 11.527877807617188,
+      "learning_rate": 9.375497472340263e-06,
+      "loss": 2.6079,
+      "step": 681000
+    },
+    {
+      "epoch": 0.17763184314159072,
+      "grad_norm": 11.338236808776855,
+      "learning_rate": 9.375092868718711e-06,
+      "loss": 2.5406,
+      "step": 681200
+    },
+    {
+      "epoch": 0.1776839957672929,
+      "grad_norm": 10.98487377166748,
+      "learning_rate": 9.374688142808268e-06,
+      "loss": 2.5617,
+      "step": 681400
+    },
+    {
+      "epoch": 0.17773614839299506,
+      "grad_norm": 10.685301780700684,
+      "learning_rate": 9.374283294620247e-06,
+      "loss": 2.5682,
+      "step": 681600
+    },
+    {
+      "epoch": 0.17778830101869725,
+      "grad_norm": 12.549797058105469,
+      "learning_rate": 9.373878324165965e-06,
+      "loss": 2.6306,
+      "step": 681800
+    },
+    {
+      "epoch": 0.1778404536443994,
+      "grad_norm": 11.423171997070312,
+      "learning_rate": 9.37347323145674e-06,
+      "loss": 2.5989,
+      "step": 682000
+    },
+    {
+      "epoch": 0.1778926062701016,
+      "grad_norm": 11.801868438720703,
+      "learning_rate": 9.373068016503894e-06,
+      "loss": 2.5902,
+      "step": 682200
+    },
+    {
+      "epoch": 0.17794475889580375,
+      "grad_norm": 10.22305965423584,
+      "learning_rate": 9.372662679318755e-06,
+      "loss": 2.5873,
+      "step": 682400
+    },
+    {
+      "epoch": 0.17799691152150593,
+      "grad_norm": 10.769112586975098,
+      "learning_rate": 9.372257219912654e-06,
+      "loss": 2.5549,
+      "step": 682600
+    },
+    {
+      "epoch": 0.17804906414720809,
+      "grad_norm": 11.621337890625,
+      "learning_rate": 9.371851638296922e-06,
+      "loss": 2.6047,
+      "step": 682800
+    },
+    {
+      "epoch": 0.17810121677291027,
+      "grad_norm": 10.884603500366211,
+      "learning_rate": 9.371445934482895e-06,
+      "loss": 2.6138,
+      "step": 683000
+    },
+    {
+      "epoch": 0.17815336939861243,
+      "grad_norm": 11.253877639770508,
+      "learning_rate": 9.371040108481915e-06,
+      "loss": 2.5755,
+      "step": 683200
+    },
+    {
+      "epoch": 0.1782055220243146,
+      "grad_norm": 13.072035789489746,
+      "learning_rate": 9.370634160305321e-06,
+      "loss": 2.5701,
+      "step": 683400
+    },
+    {
+      "epoch": 0.17825767465001677,
+      "grad_norm": 13.245817184448242,
+      "learning_rate": 9.370228089964466e-06,
+      "loss": 2.5832,
+      "step": 683600
+    },
+    {
+      "epoch": 0.17830982727571892,
+      "grad_norm": 11.859313011169434,
+      "learning_rate": 9.369821897470695e-06,
+      "loss": 2.5743,
+      "step": 683800
+    },
+    {
+      "epoch": 0.1783619799014211,
+      "grad_norm": 10.999593734741211,
+      "learning_rate": 9.369415582835365e-06,
+      "loss": 2.6368,
+      "step": 684000
+    },
+    {
+      "epoch": 0.17841413252712326,
+      "grad_norm": 12.858357429504395,
+      "learning_rate": 9.369009146069828e-06,
+      "loss": 2.5957,
+      "step": 684200
+    },
+    {
+      "epoch": 0.17846628515282545,
+      "grad_norm": 12.619375228881836,
+      "learning_rate": 9.36860258718545e-06,
+      "loss": 2.55,
+      "step": 684400
+    },
+    {
+      "epoch": 0.1785184377785276,
+      "grad_norm": 11.915067672729492,
+      "learning_rate": 9.368195906193593e-06,
+      "loss": 2.5971,
+      "step": 684600
+    },
+    {
+      "epoch": 0.1785705904042298,
+      "grad_norm": 15.2025146484375,
+      "learning_rate": 9.367789103105623e-06,
+      "loss": 2.5893,
+      "step": 684800
+    },
+    {
+      "epoch": 0.17862274302993195,
+      "grad_norm": 11.45402717590332,
+      "learning_rate": 9.367382177932911e-06,
+      "loss": 2.5893,
+      "step": 685000
+    },
+    {
+      "epoch": 0.17867489565563413,
+      "grad_norm": 12.089704513549805,
+      "learning_rate": 9.36697513068683e-06,
+      "loss": 2.5795,
+      "step": 685200
+    },
+    {
+      "epoch": 0.1787270482813363,
+      "grad_norm": 12.791685104370117,
+      "learning_rate": 9.366567961378762e-06,
+      "loss": 2.5793,
+      "step": 685400
+    },
+    {
+      "epoch": 0.17877920090703847,
+      "grad_norm": 11.997780799865723,
+      "learning_rate": 9.366160670020084e-06,
+      "loss": 2.5665,
+      "step": 685600
+    },
+    {
+      "epoch": 0.17883135353274063,
+      "grad_norm": 10.478496551513672,
+      "learning_rate": 9.365753256622178e-06,
+      "loss": 2.5858,
+      "step": 685800
+    },
+    {
+      "epoch": 0.1788835061584428,
+      "grad_norm": 9.763985633850098,
+      "learning_rate": 9.365345721196436e-06,
+      "loss": 2.5637,
+      "step": 686000
+    },
+    {
+      "epoch": 0.17893565878414497,
+      "grad_norm": 12.446106910705566,
+      "learning_rate": 9.36493806375425e-06,
+      "loss": 2.5734,
+      "step": 686200
+    },
+    {
+      "epoch": 0.17898781140984715,
+      "grad_norm": 12.356990814208984,
+      "learning_rate": 9.364530284307009e-06,
+      "loss": 2.5836,
+      "step": 686400
+    },
+    {
+      "epoch": 0.1790399640355493,
+      "grad_norm": 12.088711738586426,
+      "learning_rate": 9.364122382866117e-06,
+      "loss": 2.5795,
+      "step": 686600
+    },
+    {
+      "epoch": 0.1790921166612515,
+      "grad_norm": 10.435306549072266,
+      "learning_rate": 9.363714359442971e-06,
+      "loss": 2.6026,
+      "step": 686800
+    },
+    {
+      "epoch": 0.17914426928695365,
+      "grad_norm": 11.215888023376465,
+      "learning_rate": 9.363306214048975e-06,
+      "loss": 2.577,
+      "step": 687000
+    },
+    {
+      "epoch": 0.17919642191265583,
+      "grad_norm": 13.190821647644043,
+      "learning_rate": 9.362897946695542e-06,
+      "loss": 2.6001,
+      "step": 687200
+    },
+    {
+      "epoch": 0.179248574538358,
+      "grad_norm": 11.123869895935059,
+      "learning_rate": 9.362489557394079e-06,
+      "loss": 2.6028,
+      "step": 687400
+    },
+    {
+      "epoch": 0.17930072716406018,
+      "grad_norm": 11.40549373626709,
+      "learning_rate": 9.362081046156004e-06,
+      "loss": 2.6268,
+      "step": 687600
+    },
+    {
+      "epoch": 0.17935287978976233,
+      "grad_norm": 11.619917869567871,
+      "learning_rate": 9.361672412992734e-06,
+      "loss": 2.6188,
+      "step": 687800
+    },
+    {
+      "epoch": 0.17940503241546452,
+      "grad_norm": 11.15652084350586,
+      "learning_rate": 9.36126365791569e-06,
+      "loss": 2.5824,
+      "step": 688000
+    },
+    {
+      "epoch": 0.17945718504116667,
+      "grad_norm": 11.156540870666504,
+      "learning_rate": 9.360854780936298e-06,
+      "loss": 2.581,
+      "step": 688200
+    },
+    {
+      "epoch": 0.17950933766686886,
+      "grad_norm": 11.437117576599121,
+      "learning_rate": 9.360445782065989e-06,
+      "loss": 2.5932,
+      "step": 688400
+    },
+    {
+      "epoch": 0.179561490292571,
+      "grad_norm": 11.62105655670166,
+      "learning_rate": 9.36003666131619e-06,
+      "loss": 2.5851,
+      "step": 688600
+    },
+    {
+      "epoch": 0.1796136429182732,
+      "grad_norm": 14.763154029846191,
+      "learning_rate": 9.359627418698338e-06,
+      "loss": 2.6017,
+      "step": 688800
+    },
+    {
+      "epoch": 0.17966579554397535,
+      "grad_norm": 12.010329246520996,
+      "learning_rate": 9.359218054223876e-06,
+      "loss": 2.5934,
+      "step": 689000
+    },
+    {
+      "epoch": 0.17971794816967754,
+      "grad_norm": 12.975419998168945,
+      "learning_rate": 9.35880856790424e-06,
+      "loss": 2.5678,
+      "step": 689200
+    },
+    {
+      "epoch": 0.1797701007953797,
+      "grad_norm": 11.763967514038086,
+      "learning_rate": 9.35839895975088e-06,
+      "loss": 2.5383,
+      "step": 689400
+    },
+    {
+      "epoch": 0.17982225342108185,
+      "grad_norm": 11.407500267028809,
+      "learning_rate": 9.357989229775245e-06,
+      "loss": 2.5591,
+      "step": 689600
+    },
+    {
+      "epoch": 0.17987440604678404,
+      "grad_norm": 10.644564628601074,
+      "learning_rate": 9.357579377988786e-06,
+      "loss": 2.5843,
+      "step": 689800
+    },
+    {
+      "epoch": 0.1799265586724862,
+      "grad_norm": 11.088133811950684,
+      "learning_rate": 9.357169404402958e-06,
+      "loss": 2.592,
+      "step": 690000
+    },
+    {
+      "epoch": 0.17997871129818838,
+      "grad_norm": 11.872809410095215,
+      "learning_rate": 9.35675930902922e-06,
+      "loss": 2.5783,
+      "step": 690200
+    },
+    {
+      "epoch": 0.18003086392389053,
+      "grad_norm": 12.999768257141113,
+      "learning_rate": 9.356349091879037e-06,
+      "loss": 2.6072,
+      "step": 690400
+    },
+    {
+      "epoch": 0.18008301654959272,
+      "grad_norm": 13.067625045776367,
+      "learning_rate": 9.355938752963874e-06,
+      "loss": 2.5654,
+      "step": 690600
+    },
+    {
+      "epoch": 0.18013516917529487,
+      "grad_norm": 11.797418594360352,
+      "learning_rate": 9.355528292295199e-06,
+      "loss": 2.5624,
+      "step": 690800
+    },
+    {
+      "epoch": 0.18018732180099706,
+      "grad_norm": 10.904340744018555,
+      "learning_rate": 9.355117709884487e-06,
+      "loss": 2.5769,
+      "step": 691000
+    },
+    {
+      "epoch": 0.18023947442669921,
+      "grad_norm": 10.31658935546875,
+      "learning_rate": 9.354707005743213e-06,
+      "loss": 2.5788,
+      "step": 691200
+    },
+    {
+      "epoch": 0.1802916270524014,
+      "grad_norm": 12.880640029907227,
+      "learning_rate": 9.354296179882858e-06,
+      "loss": 2.588,
+      "step": 691400
+    },
+    {
+      "epoch": 0.18034377967810356,
+      "grad_norm": 12.985132217407227,
+      "learning_rate": 9.353885232314902e-06,
+      "loss": 2.599,
+      "step": 691600
+    },
+    {
+      "epoch": 0.18039593230380574,
+      "grad_norm": 9.16226863861084,
+      "learning_rate": 9.353474163050835e-06,
+      "loss": 2.5426,
+      "step": 691800
+    },
+    {
+      "epoch": 0.1804480849295079,
+      "grad_norm": 12.11599349975586,
+      "learning_rate": 9.353062972102146e-06,
+      "loss": 2.5754,
+      "step": 692000
+    },
+    {
+      "epoch": 0.18050023755521008,
+      "grad_norm": 12.336219787597656,
+      "learning_rate": 9.352651659480327e-06,
+      "loss": 2.5988,
+      "step": 692200
+    },
+    {
+      "epoch": 0.18055239018091224,
+      "grad_norm": 10.14101505279541,
+      "learning_rate": 9.352240225196876e-06,
+      "loss": 2.5971,
+      "step": 692400
+    },
+    {
+      "epoch": 0.18060454280661442,
+      "grad_norm": 12.834318161010742,
+      "learning_rate": 9.35182866926329e-06,
+      "loss": 2.5767,
+      "step": 692600
+    },
+    {
+      "epoch": 0.18065669543231658,
+      "grad_norm": 10.38381576538086,
+      "learning_rate": 9.351416991691078e-06,
+      "loss": 2.5789,
+      "step": 692800
+    },
+    {
+      "epoch": 0.18070884805801876,
+      "grad_norm": 11.648655891418457,
+      "learning_rate": 9.351005192491742e-06,
+      "loss": 2.5979,
+      "step": 693000
+    },
+    {
+      "epoch": 0.18076100068372092,
+      "grad_norm": 11.521854400634766,
+      "learning_rate": 9.350593271676793e-06,
+      "loss": 2.5649,
+      "step": 693200
+    },
+    {
+      "epoch": 0.1808131533094231,
+      "grad_norm": 12.451175689697266,
+      "learning_rate": 9.350181229257748e-06,
+      "loss": 2.5619,
+      "step": 693400
+    },
+    {
+      "epoch": 0.18086530593512526,
+      "grad_norm": 11.495298385620117,
+      "learning_rate": 9.34976906524612e-06,
+      "loss": 2.5701,
+      "step": 693600
+    },
+    {
+      "epoch": 0.18091745856082744,
+      "grad_norm": 12.428627967834473,
+      "learning_rate": 9.349356779653432e-06,
+      "loss": 2.5957,
+      "step": 693800
+    },
+    {
+      "epoch": 0.1809696111865296,
+      "grad_norm": 12.761686325073242,
+      "learning_rate": 9.348944372491207e-06,
+      "loss": 2.577,
+      "step": 694000
+    },
+    {
+      "epoch": 0.18102176381223178,
+      "grad_norm": 13.41728401184082,
+      "learning_rate": 9.348531843770971e-06,
+      "loss": 2.5966,
+      "step": 694200
+    },
+    {
+      "epoch": 0.18107391643793394,
+      "grad_norm": 12.403990745544434,
+      "learning_rate": 9.348119193504256e-06,
+      "loss": 2.5791,
+      "step": 694400
+    },
+    {
+      "epoch": 0.18112606906363612,
+      "grad_norm": 11.769331932067871,
+      "learning_rate": 9.347706421702598e-06,
+      "loss": 2.5963,
+      "step": 694600
+    },
+    {
+      "epoch": 0.18117822168933828,
+      "grad_norm": 12.589640617370605,
+      "learning_rate": 9.347293528377532e-06,
+      "loss": 2.6107,
+      "step": 694800
+    },
+    {
+      "epoch": 0.18123037431504047,
+      "grad_norm": 13.722719192504883,
+      "learning_rate": 9.346880513540598e-06,
+      "loss": 2.5917,
+      "step": 695000
+    },
+    {
+      "epoch": 0.18128252694074262,
+      "grad_norm": 9.757765769958496,
+      "learning_rate": 9.34646737720334e-06,
+      "loss": 2.585,
+      "step": 695200
+    },
+    {
+      "epoch": 0.18133467956644478,
+      "grad_norm": 11.891173362731934,
+      "learning_rate": 9.34605411937731e-06,
+      "loss": 2.5579,
+      "step": 695400
+    },
+    {
+      "epoch": 0.18138683219214696,
+      "grad_norm": 10.937880516052246,
+      "learning_rate": 9.345640740074055e-06,
+      "loss": 2.5802,
+      "step": 695600
+    },
+    {
+      "epoch": 0.18143898481784912,
+      "grad_norm": 11.980300903320312,
+      "learning_rate": 9.345227239305133e-06,
+      "loss": 2.5721,
+      "step": 695800
+    },
+    {
+      "epoch": 0.1814911374435513,
+      "grad_norm": 10.967789649963379,
+      "learning_rate": 9.344813617082096e-06,
+      "loss": 2.6024,
+      "step": 696000
+    },
+    {
+      "epoch": 0.18154329006925346,
+      "grad_norm": 12.861534118652344,
+      "learning_rate": 9.34439987341651e-06,
+      "loss": 2.619,
+      "step": 696200
+    },
+    {
+      "epoch": 0.18159544269495564,
+      "grad_norm": 12.319485664367676,
+      "learning_rate": 9.343986008319938e-06,
+      "loss": 2.5858,
+      "step": 696400
+    },
+    {
+      "epoch": 0.1816475953206578,
+      "grad_norm": 11.069070816040039,
+      "learning_rate": 9.343572021803948e-06,
+      "loss": 2.6071,
+      "step": 696600
+    },
+    {
+      "epoch": 0.18169974794635999,
+      "grad_norm": 12.012398719787598,
+      "learning_rate": 9.343157913880113e-06,
+      "loss": 2.6151,
+      "step": 696800
+    },
+    {
+      "epoch": 0.18175190057206214,
+      "grad_norm": 9.027801513671875,
+      "learning_rate": 9.342743684560006e-06,
+      "loss": 2.6,
+      "step": 697000
+    },
+    {
+      "epoch": 0.18180405319776433,
+      "grad_norm": 13.315918922424316,
+      "learning_rate": 9.342329333855205e-06,
+      "loss": 2.5678,
+      "step": 697200
+    },
+    {
+      "epoch": 0.18185620582346648,
+      "grad_norm": 11.573213577270508,
+      "learning_rate": 9.341914861777293e-06,
+      "loss": 2.5673,
+      "step": 697400
+    },
+    {
+      "epoch": 0.18190835844916867,
+      "grad_norm": 10.803704261779785,
+      "learning_rate": 9.341500268337853e-06,
+      "loss": 2.5895,
+      "step": 697600
+    },
+    {
+      "epoch": 0.18196051107487082,
+      "grad_norm": 11.29845142364502,
+      "learning_rate": 9.341085553548474e-06,
+      "loss": 2.5807,
+      "step": 697800
+    },
+    {
+      "epoch": 0.182012663700573,
+      "grad_norm": 12.04796314239502,
+      "learning_rate": 9.34067071742075e-06,
+      "loss": 2.5675,
+      "step": 698000
+    },
+    {
+      "epoch": 0.18206481632627516,
+      "grad_norm": 12.755510330200195,
+      "learning_rate": 9.340255759966274e-06,
+      "loss": 2.5761,
+      "step": 698200
+    },
+    {
+      "epoch": 0.18211696895197735,
+      "grad_norm": 11.835643768310547,
+      "learning_rate": 9.339840681196645e-06,
+      "loss": 2.5841,
+      "step": 698400
+    },
+    {
+      "epoch": 0.1821691215776795,
+      "grad_norm": 11.568483352661133,
+      "learning_rate": 9.339425481123465e-06,
+      "loss": 2.5586,
+      "step": 698600
+    },
+    {
+      "epoch": 0.1822212742033817,
+      "grad_norm": 13.110424041748047,
+      "learning_rate": 9.33901015975834e-06,
+      "loss": 2.6143,
+      "step": 698800
+    },
+    {
+      "epoch": 0.18227342682908385,
+      "grad_norm": 11.408985137939453,
+      "learning_rate": 9.338594717112877e-06,
+      "loss": 2.5753,
+      "step": 699000
+    },
+    {
+      "epoch": 0.18232557945478603,
+      "grad_norm": 10.737070083618164,
+      "learning_rate": 9.33817915319869e-06,
+      "loss": 2.5876,
+      "step": 699200
+    },
+    {
+      "epoch": 0.1823777320804882,
+      "grad_norm": 12.560068130493164,
+      "learning_rate": 9.337763468027392e-06,
+      "loss": 2.5513,
+      "step": 699400
+    },
+    {
+      "epoch": 0.18242988470619037,
+      "grad_norm": 11.872684478759766,
+      "learning_rate": 9.337347661610604e-06,
+      "loss": 2.5624,
+      "step": 699600
+    },
+    {
+      "epoch": 0.18248203733189253,
+      "grad_norm": 11.913339614868164,
+      "learning_rate": 9.336931733959948e-06,
+      "loss": 2.538,
+      "step": 699800
+    },
+    {
+      "epoch": 0.1825341899575947,
+      "grad_norm": 10.39101791381836,
+      "learning_rate": 9.336515685087048e-06,
+      "loss": 2.5836,
+      "step": 700000
+    },
+    {
+      "epoch": 0.18258634258329687,
+      "grad_norm": 8.983271598815918,
+      "learning_rate": 9.336099515003538e-06,
+      "loss": 2.5803,
+      "step": 700200
+    },
+    {
+      "epoch": 0.18263849520899905,
+      "grad_norm": 12.488344192504883,
+      "learning_rate": 9.335683223721044e-06,
+      "loss": 2.5908,
+      "step": 700400
+    },
+    {
+      "epoch": 0.1826906478347012,
+      "grad_norm": 10.796935081481934,
+      "learning_rate": 9.335266811251205e-06,
+      "loss": 2.6309,
+      "step": 700600
+    },
+    {
+      "epoch": 0.1827428004604034,
+      "grad_norm": 10.452614784240723,
+      "learning_rate": 9.33485027760566e-06,
+      "loss": 2.5893,
+      "step": 700800
+    },
+    {
+      "epoch": 0.18279495308610555,
+      "grad_norm": 10.865897178649902,
+      "learning_rate": 9.33443362279605e-06,
+      "loss": 2.5761,
+      "step": 701000
+    },
+    {
+      "epoch": 0.1828471057118077,
+      "grad_norm": 12.461289405822754,
+      "learning_rate": 9.334016846834024e-06,
+      "loss": 2.5851,
+      "step": 701200
+    },
+    {
+      "epoch": 0.1828992583375099,
+      "grad_norm": 10.119328498840332,
+      "learning_rate": 9.333599949731232e-06,
+      "loss": 2.5364,
+      "step": 701400
+    },
+    {
+      "epoch": 0.18295141096321205,
+      "grad_norm": 12.12794017791748,
+      "learning_rate": 9.33318293149932e-06,
+      "loss": 2.6064,
+      "step": 701600
+    },
+    {
+      "epoch": 0.18300356358891423,
+      "grad_norm": 12.383764266967773,
+      "learning_rate": 9.332765792149951e-06,
+      "loss": 2.6052,
+      "step": 701800
+    },
+    {
+      "epoch": 0.1830557162146164,
+      "grad_norm": 10.087251663208008,
+      "learning_rate": 9.332348531694783e-06,
+      "loss": 2.5666,
+      "step": 702000
+    },
+    {
+      "epoch": 0.18310786884031857,
+      "grad_norm": 12.394096374511719,
+      "learning_rate": 9.331931150145478e-06,
+      "loss": 2.6195,
+      "step": 702200
+    },
+    {
+      "epoch": 0.18316002146602073,
+      "grad_norm": 12.181896209716797,
+      "learning_rate": 9.331513647513702e-06,
+      "loss": 2.5802,
+      "step": 702400
+    },
+    {
+      "epoch": 0.1832121740917229,
+      "grad_norm": 12.717833518981934,
+      "learning_rate": 9.331096023811125e-06,
+      "loss": 2.5753,
+      "step": 702600
+    },
+    {
+      "epoch": 0.18326432671742507,
+      "grad_norm": 11.721474647521973,
+      "learning_rate": 9.33067827904942e-06,
+      "loss": 2.6129,
+      "step": 702800
+    },
+    {
+      "epoch": 0.18331647934312725,
+      "grad_norm": 10.442642211914062,
+      "learning_rate": 9.330260413240265e-06,
+      "loss": 2.614,
+      "step": 703000
+    },
+    {
+      "epoch": 0.1833686319688294,
+      "grad_norm": 10.827333450317383,
+      "learning_rate": 9.329842426395335e-06,
+      "loss": 2.5673,
+      "step": 703200
+    },
+    {
+      "epoch": 0.1834207845945316,
+      "grad_norm": 11.326786994934082,
+      "learning_rate": 9.32942431852632e-06,
+      "loss": 2.5926,
+      "step": 703400
+    },
+    {
+      "epoch": 0.18347293722023375,
+      "grad_norm": 11.326935768127441,
+      "learning_rate": 9.329006089644902e-06,
+      "loss": 2.6239,
+      "step": 703600
+    },
+    {
+      "epoch": 0.18352508984593593,
+      "grad_norm": 13.342899322509766,
+      "learning_rate": 9.328587739762774e-06,
+      "loss": 2.5926,
+      "step": 703800
+    },
+    {
+      "epoch": 0.1835772424716381,
+      "grad_norm": 10.269484519958496,
+      "learning_rate": 9.328169268891624e-06,
+      "loss": 2.5741,
+      "step": 704000
+    },
+    {
+      "epoch": 0.18362939509734028,
+      "grad_norm": 12.175603866577148,
+      "learning_rate": 9.327750677043156e-06,
+      "loss": 2.5677,
+      "step": 704200
+    },
+    {
+      "epoch": 0.18368154772304243,
+      "grad_norm": 11.766140937805176,
+      "learning_rate": 9.327331964229066e-06,
+      "loss": 2.5791,
+      "step": 704400
+    },
+    {
+      "epoch": 0.18373370034874462,
+      "grad_norm": 10.756917953491211,
+      "learning_rate": 9.326913130461056e-06,
+      "loss": 2.5927,
+      "step": 704600
+    },
+    {
+      "epoch": 0.18378585297444677,
+      "grad_norm": 11.94490909576416,
+      "learning_rate": 9.326494175750836e-06,
+      "loss": 2.6214,
+      "step": 704800
+    },
+    {
+      "epoch": 0.18383800560014896,
+      "grad_norm": 12.475340843200684,
+      "learning_rate": 9.326075100110115e-06,
+      "loss": 2.5548,
+      "step": 705000
+    },
+    {
+      "epoch": 0.1838901582258511,
+      "grad_norm": 10.366930961608887,
+      "learning_rate": 9.325655903550605e-06,
+      "loss": 2.5704,
+      "step": 705200
+    },
+    {
+      "epoch": 0.1839423108515533,
+      "grad_norm": 13.210090637207031,
+      "learning_rate": 9.325236586084028e-06,
+      "loss": 2.5656,
+      "step": 705400
+    },
+    {
+      "epoch": 0.18399446347725545,
+      "grad_norm": 12.236307144165039,
+      "learning_rate": 9.3248171477221e-06,
+      "loss": 2.549,
+      "step": 705600
+    },
+    {
+      "epoch": 0.18404661610295764,
+      "grad_norm": 12.707765579223633,
+      "learning_rate": 9.324397588476545e-06,
+      "loss": 2.5777,
+      "step": 705800
+    },
+    {
+      "epoch": 0.1840987687286598,
+      "grad_norm": 12.61831283569336,
+      "learning_rate": 9.323977908359093e-06,
+      "loss": 2.6,
+      "step": 706000
+    },
+    {
+      "epoch": 0.18415092135436198,
+      "grad_norm": 11.035533905029297,
+      "learning_rate": 9.323558107381472e-06,
+      "loss": 2.5415,
+      "step": 706200
+    },
+    {
+      "epoch": 0.18420307398006414,
+      "grad_norm": 10.974006652832031,
+      "learning_rate": 9.323138185555416e-06,
+      "loss": 2.6035,
+      "step": 706400
+    },
+    {
+      "epoch": 0.18425522660576632,
+      "grad_norm": 11.0816650390625,
+      "learning_rate": 9.322718142892663e-06,
+      "loss": 2.5626,
+      "step": 706600
+    },
+    {
+      "epoch": 0.18430737923146848,
+      "grad_norm": 11.282931327819824,
+      "learning_rate": 9.322297979404955e-06,
+      "loss": 2.5733,
+      "step": 706800
+    },
+    {
+      "epoch": 0.18435953185717063,
+      "grad_norm": 11.803282737731934,
+      "learning_rate": 9.321877695104034e-06,
+      "loss": 2.5855,
+      "step": 707000
+    },
+    {
+      "epoch": 0.18441168448287282,
+      "grad_norm": 11.176085472106934,
+      "learning_rate": 9.321457290001646e-06,
+      "loss": 2.5848,
+      "step": 707200
+    },
+    {
+      "epoch": 0.18446383710857497,
+      "grad_norm": 10.337486267089844,
+      "learning_rate": 9.321036764109546e-06,
+      "loss": 2.5328,
+      "step": 707400
+    },
+    {
+      "epoch": 0.18451598973427716,
+      "grad_norm": 11.199088096618652,
+      "learning_rate": 9.320616117439486e-06,
+      "loss": 2.5684,
+      "step": 707600
+    },
+    {
+      "epoch": 0.18456814235997931,
+      "grad_norm": 11.944901466369629,
+      "learning_rate": 9.320195350003223e-06,
+      "loss": 2.5778,
+      "step": 707800
+    },
+    {
+      "epoch": 0.1846202949856815,
+      "grad_norm": 12.067788124084473,
+      "learning_rate": 9.31977446181252e-06,
+      "loss": 2.6062,
+      "step": 708000
+    },
+    {
+      "epoch": 0.18467244761138366,
+      "grad_norm": 11.054697036743164,
+      "learning_rate": 9.319353452879139e-06,
+      "loss": 2.5903,
+      "step": 708200
+    },
+    {
+      "epoch": 0.18472460023708584,
+      "grad_norm": 12.969680786132812,
+      "learning_rate": 9.31893232321485e-06,
+      "loss": 2.5689,
+      "step": 708400
+    },
+    {
+      "epoch": 0.184776752862788,
+      "grad_norm": 12.888757705688477,
+      "learning_rate": 9.31851107283142e-06,
+      "loss": 2.593,
+      "step": 708600
+    },
+    {
+      "epoch": 0.18482890548849018,
+      "grad_norm": 11.72866153717041,
+      "learning_rate": 9.318089701740627e-06,
+      "loss": 2.5896,
+      "step": 708800
+    },
+    {
+      "epoch": 0.18488105811419234,
+      "grad_norm": 12.54987621307373,
+      "learning_rate": 9.317668209954248e-06,
+      "loss": 2.5961,
+      "step": 709000
+    },
+    {
+      "epoch": 0.18493321073989452,
+      "grad_norm": 12.830785751342773,
+      "learning_rate": 9.317246597484065e-06,
+      "loss": 2.5997,
+      "step": 709200
+    },
+    {
+      "epoch": 0.18498536336559668,
+      "grad_norm": 10.779577255249023,
+      "learning_rate": 9.31682486434186e-06,
+      "loss": 2.608,
+      "step": 709400
+    },
+    {
+      "epoch": 0.18503751599129886,
+      "grad_norm": 10.413871765136719,
+      "learning_rate": 9.316403010539424e-06,
+      "loss": 2.5584,
+      "step": 709600
+    },
+    {
+      "epoch": 0.18508966861700102,
+      "grad_norm": 12.727044105529785,
+      "learning_rate": 9.315981036088547e-06,
+      "loss": 2.5805,
+      "step": 709800
+    },
+    {
+      "epoch": 0.1851418212427032,
+      "grad_norm": 10.990044593811035,
+      "learning_rate": 9.315558941001022e-06,
+      "loss": 2.5659,
+      "step": 710000
+    },
+    {
+      "epoch": 0.18519397386840536,
+      "grad_norm": 13.053669929504395,
+      "learning_rate": 9.315136725288648e-06,
+      "loss": 2.5645,
+      "step": 710200
+    },
+    {
+      "epoch": 0.18524612649410754,
+      "grad_norm": 12.4849853515625,
+      "learning_rate": 9.314714388963229e-06,
+      "loss": 2.5712,
+      "step": 710400
+    },
+    {
+      "epoch": 0.1852982791198097,
+      "grad_norm": 13.245108604431152,
+      "learning_rate": 9.314291932036568e-06,
+      "loss": 2.589,
+      "step": 710600
+    },
+    {
+      "epoch": 0.18535043174551188,
+      "grad_norm": 11.096653938293457,
+      "learning_rate": 9.31386935452047e-06,
+      "loss": 2.5642,
+      "step": 710800
+    },
+    {
+      "epoch": 0.18540258437121404,
+      "grad_norm": 12.3607177734375,
+      "learning_rate": 9.313446656426752e-06,
+      "loss": 2.5596,
+      "step": 711000
+    },
+    {
+      "epoch": 0.18545473699691623,
+      "grad_norm": 11.200916290283203,
+      "learning_rate": 9.313023837767225e-06,
+      "loss": 2.5321,
+      "step": 711200
+    },
+    {
+      "epoch": 0.18550688962261838,
+      "grad_norm": 12.221725463867188,
+      "learning_rate": 9.31260089855371e-06,
+      "loss": 2.5827,
+      "step": 711400
+    },
+    {
+      "epoch": 0.18555904224832057,
+      "grad_norm": 11.39847469329834,
+      "learning_rate": 9.312177838798028e-06,
+      "loss": 2.581,
+      "step": 711600
+    },
+    {
+      "epoch": 0.18561119487402272,
+      "grad_norm": 11.100894927978516,
+      "learning_rate": 9.311754658512003e-06,
+      "loss": 2.5598,
+      "step": 711800
+    },
+    {
+      "epoch": 0.1856633474997249,
+      "grad_norm": 11.927329063415527,
+      "learning_rate": 9.31133135770746e-06,
+      "loss": 2.5918,
+      "step": 712000
+    },
+    {
+      "epoch": 0.18571550012542706,
+      "grad_norm": 12.195674896240234,
+      "learning_rate": 9.310907936396238e-06,
+      "loss": 2.5352,
+      "step": 712200
+    },
+    {
+      "epoch": 0.18576765275112925,
+      "grad_norm": 12.455114364624023,
+      "learning_rate": 9.310484394590168e-06,
+      "loss": 2.5571,
+      "step": 712400
+    },
+    {
+      "epoch": 0.1858198053768314,
+      "grad_norm": 12.462471961975098,
+      "learning_rate": 9.310060732301087e-06,
+      "loss": 2.5686,
+      "step": 712600
+    },
+    {
+      "epoch": 0.1858719580025336,
+      "grad_norm": 12.136844635009766,
+      "learning_rate": 9.30963694954084e-06,
+      "loss": 2.5705,
+      "step": 712800
+    },
+    {
+      "epoch": 0.18592411062823574,
+      "grad_norm": 13.6740083694458,
+      "learning_rate": 9.309213046321272e-06,
+      "loss": 2.568,
+      "step": 713000
+    },
+    {
+      "epoch": 0.1859762632539379,
+      "grad_norm": 10.739278793334961,
+      "learning_rate": 9.308789022654233e-06,
+      "loss": 2.576,
+      "step": 713200
+    },
+    {
+      "epoch": 0.18602841587964009,
+      "grad_norm": 12.809683799743652,
+      "learning_rate": 9.308364878551567e-06,
+      "loss": 2.5984,
+      "step": 713400
+    },
+    {
+      "epoch": 0.18608056850534224,
+      "grad_norm": 11.436633110046387,
+      "learning_rate": 9.30794061402514e-06,
+      "loss": 2.5605,
+      "step": 713600
+    },
+    {
+      "epoch": 0.18613272113104443,
+      "grad_norm": 10.629294395446777,
+      "learning_rate": 9.307516229086802e-06,
+      "loss": 2.5838,
+      "step": 713800
+    },
+    {
+      "epoch": 0.18618487375674658,
+      "grad_norm": 11.265149116516113,
+      "learning_rate": 9.30709172374842e-06,
+      "loss": 2.5753,
+      "step": 714000
+    },
+    {
+      "epoch": 0.18623702638244877,
+      "grad_norm": 13.775673866271973,
+      "learning_rate": 9.306667098021858e-06,
+      "loss": 2.5764,
+      "step": 714200
+    },
+    {
+      "epoch": 0.18628917900815092,
+      "grad_norm": 10.820959091186523,
+      "learning_rate": 9.306242351918983e-06,
+      "loss": 2.5516,
+      "step": 714400
+    },
+    {
+      "epoch": 0.1863413316338531,
+      "grad_norm": 11.11262035369873,
+      "learning_rate": 9.305817485451672e-06,
+      "loss": 2.5498,
+      "step": 714600
+    },
+    {
+      "epoch": 0.18639348425955526,
+      "grad_norm": 12.226655006408691,
+      "learning_rate": 9.305392498631796e-06,
+      "loss": 2.5372,
+      "step": 714800
+    },
+    {
+      "epoch": 0.18644563688525745,
+      "grad_norm": 11.85600757598877,
+      "learning_rate": 9.304967391471236e-06,
+      "loss": 2.628,
+      "step": 715000
+    },
+    {
+      "epoch": 0.1864977895109596,
+      "grad_norm": 10.384273529052734,
+      "learning_rate": 9.304542163981874e-06,
+      "loss": 2.5591,
+      "step": 715200
+    },
+    {
+      "epoch": 0.1865499421366618,
+      "grad_norm": 10.744622230529785,
+      "learning_rate": 9.304116816175593e-06,
+      "loss": 2.5466,
+      "step": 715400
+    },
+    {
+      "epoch": 0.18660209476236395,
+      "grad_norm": 14.192846298217773,
+      "learning_rate": 9.303691348064285e-06,
+      "loss": 2.6048,
+      "step": 715600
+    },
+    {
+      "epoch": 0.18665424738806613,
+      "grad_norm": 10.374147415161133,
+      "learning_rate": 9.303265759659842e-06,
+      "loss": 2.5472,
+      "step": 715800
+    },
+    {
+      "epoch": 0.1867064000137683,
+      "grad_norm": 12.727148056030273,
+      "learning_rate": 9.30284005097416e-06,
+      "loss": 2.5792,
+      "step": 716000
+    },
+    {
+      "epoch": 0.18675855263947047,
+      "grad_norm": 14.218303680419922,
+      "learning_rate": 9.302414222019136e-06,
+      "loss": 2.5839,
+      "step": 716200
+    },
+    {
+      "epoch": 0.18681070526517263,
+      "grad_norm": 11.11287784576416,
+      "learning_rate": 9.301988272806675e-06,
+      "loss": 2.5779,
+      "step": 716400
+    },
+    {
+      "epoch": 0.1868628578908748,
+      "grad_norm": 9.908751487731934,
+      "learning_rate": 9.30156220334868e-06,
+      "loss": 2.562,
+      "step": 716600
+    },
+    {
+      "epoch": 0.18691501051657697,
+      "grad_norm": 10.67510986328125,
+      "learning_rate": 9.301136013657061e-06,
+      "loss": 2.6033,
+      "step": 716800
+    },
+    {
+      "epoch": 0.18696716314227915,
+      "grad_norm": 10.921405792236328,
+      "learning_rate": 9.300709703743733e-06,
+      "loss": 2.5775,
+      "step": 717000
+    },
+    {
+      "epoch": 0.1870193157679813,
+      "grad_norm": 11.9124116897583,
+      "learning_rate": 9.300283273620607e-06,
+      "loss": 2.5652,
+      "step": 717200
+    },
+    {
+      "epoch": 0.1870714683936835,
+      "grad_norm": 13.153666496276855,
+      "learning_rate": 9.299856723299608e-06,
+      "loss": 2.5931,
+      "step": 717400
+    },
+    {
+      "epoch": 0.18712362101938565,
+      "grad_norm": 11.42650318145752,
+      "learning_rate": 9.299430052792655e-06,
+      "loss": 2.5649,
+      "step": 717600
+    },
+    {
+      "epoch": 0.18717577364508783,
+      "grad_norm": 14.207989692687988,
+      "learning_rate": 9.299003262111672e-06,
+      "loss": 2.5943,
+      "step": 717800
+    },
+    {
+      "epoch": 0.18722792627079,
+      "grad_norm": 12.080742835998535,
+      "learning_rate": 9.298576351268594e-06,
+      "loss": 2.5583,
+      "step": 718000
+    },
+    {
+      "epoch": 0.18728007889649217,
+      "grad_norm": 10.130331039428711,
+      "learning_rate": 9.298149320275349e-06,
+      "loss": 2.5811,
+      "step": 718200
+    },
+    {
+      "epoch": 0.18733223152219433,
+      "grad_norm": 11.960268020629883,
+      "learning_rate": 9.297722169143875e-06,
+      "loss": 2.5706,
+      "step": 718400
+    },
+    {
+      "epoch": 0.18738438414789652,
+      "grad_norm": 12.298662185668945,
+      "learning_rate": 9.29729489788611e-06,
+      "loss": 2.5851,
+      "step": 718600
+    },
+    {
+      "epoch": 0.18743653677359867,
+      "grad_norm": 12.417919158935547,
+      "learning_rate": 9.296867506514e-06,
+      "loss": 2.557,
+      "step": 718800
+    },
+    {
+      "epoch": 0.18748868939930083,
+      "grad_norm": 11.724634170532227,
+      "learning_rate": 9.296439995039488e-06,
+      "loss": 2.571,
+      "step": 719000
+    },
+    {
+      "epoch": 0.187540842025003,
+      "grad_norm": 13.457600593566895,
+      "learning_rate": 9.296012363474523e-06,
+      "loss": 2.6037,
+      "step": 719200
+    },
+    {
+      "epoch": 0.18759299465070517,
+      "grad_norm": 15.612174987792969,
+      "learning_rate": 9.295584611831059e-06,
+      "loss": 2.5789,
+      "step": 719400
+    },
+    {
+      "epoch": 0.18764514727640735,
+      "grad_norm": 11.381507873535156,
+      "learning_rate": 9.295156740121052e-06,
+      "loss": 2.5929,
+      "step": 719600
+    },
+    {
+      "epoch": 0.1876972999021095,
+      "grad_norm": 11.643939971923828,
+      "learning_rate": 9.294728748356463e-06,
+      "loss": 2.5697,
+      "step": 719800
+    },
+    {
+      "epoch": 0.1877494525278117,
+      "grad_norm": 12.142743110656738,
+      "learning_rate": 9.294300636549251e-06,
+      "loss": 2.5505,
+      "step": 720000
+    },
+    {
+      "epoch": 0.18780160515351385,
+      "grad_norm": 12.137763023376465,
+      "learning_rate": 9.293872404711388e-06,
+      "loss": 2.6389,
+      "step": 720200
+    },
+    {
+      "epoch": 0.18785375777921604,
+      "grad_norm": 12.277729988098145,
+      "learning_rate": 9.29344405285484e-06,
+      "loss": 2.5933,
+      "step": 720400
+    },
+    {
+      "epoch": 0.1879059104049182,
+      "grad_norm": 11.714141845703125,
+      "learning_rate": 9.293015580991577e-06,
+      "loss": 2.5719,
+      "step": 720600
+    },
+    {
+      "epoch": 0.18795806303062038,
+      "grad_norm": 11.93021297454834,
+      "learning_rate": 9.292586989133581e-06,
+      "loss": 2.5805,
+      "step": 720800
+    },
+    {
+      "epoch": 0.18801021565632253,
+      "grad_norm": 12.212942123413086,
+      "learning_rate": 9.292158277292828e-06,
+      "loss": 2.5759,
+      "step": 721000
+    },
+    {
+      "epoch": 0.18806236828202472,
+      "grad_norm": 10.348050117492676,
+      "learning_rate": 9.2917294454813e-06,
+      "loss": 2.5756,
+      "step": 721200
+    },
+    {
+      "epoch": 0.18811452090772687,
+      "grad_norm": 11.41096305847168,
+      "learning_rate": 9.29130049371099e-06,
+      "loss": 2.589,
+      "step": 721400
+    },
+    {
+      "epoch": 0.18816667353342906,
+      "grad_norm": 11.99845027923584,
+      "learning_rate": 9.290871421993881e-06,
+      "loss": 2.5932,
+      "step": 721600
+    },
+    {
+      "epoch": 0.18821882615913121,
+      "grad_norm": 13.32406234741211,
+      "learning_rate": 9.290442230341967e-06,
+      "loss": 2.5926,
+      "step": 721800
+    },
+    {
+      "epoch": 0.1882709787848334,
+      "grad_norm": 12.009814262390137,
+      "learning_rate": 9.290012918767248e-06,
+      "loss": 2.5701,
+      "step": 722000
+    },
+    {
+      "epoch": 0.18832313141053555,
+      "grad_norm": 13.789295196533203,
+      "learning_rate": 9.289583487281719e-06,
+      "loss": 2.576,
+      "step": 722200
+    },
+    {
+      "epoch": 0.18837528403623774,
+      "grad_norm": 10.698989868164062,
+      "learning_rate": 9.289153935897387e-06,
+      "loss": 2.5756,
+      "step": 722400
+    },
+    {
+      "epoch": 0.1884274366619399,
+      "grad_norm": 12.086974143981934,
+      "learning_rate": 9.288724264626256e-06,
+      "loss": 2.5829,
+      "step": 722600
+    },
+    {
+      "epoch": 0.18847958928764208,
+      "grad_norm": 11.715788841247559,
+      "learning_rate": 9.288294473480337e-06,
+      "loss": 2.5337,
+      "step": 722800
+    },
+    {
+      "epoch": 0.18853174191334424,
+      "grad_norm": 10.705584526062012,
+      "learning_rate": 9.287864562471644e-06,
+      "loss": 2.5735,
+      "step": 723000
+    },
+    {
+      "epoch": 0.18858389453904642,
+      "grad_norm": 12.674153327941895,
+      "learning_rate": 9.287434531612192e-06,
+      "loss": 2.5843,
+      "step": 723200
+    },
+    {
+      "epoch": 0.18863604716474858,
+      "grad_norm": 12.425131797790527,
+      "learning_rate": 9.287004380914e-06,
+      "loss": 2.5627,
+      "step": 723400
+    },
+    {
+      "epoch": 0.18868819979045076,
+      "grad_norm": 11.173230171203613,
+      "learning_rate": 9.286574110389094e-06,
+      "loss": 2.5609,
+      "step": 723600
+    },
+    {
+      "epoch": 0.18874035241615292,
+      "grad_norm": 12.521120071411133,
+      "learning_rate": 9.286143720049498e-06,
+      "loss": 2.5708,
+      "step": 723800
+    },
+    {
+      "epoch": 0.1887925050418551,
+      "grad_norm": 10.894947052001953,
+      "learning_rate": 9.285713209907243e-06,
+      "loss": 2.524,
+      "step": 724000
+    },
+    {
+      "epoch": 0.18884465766755726,
+      "grad_norm": 11.677881240844727,
+      "learning_rate": 9.285282579974362e-06,
+      "loss": 2.6056,
+      "step": 724200
+    },
+    {
+      "epoch": 0.18889681029325944,
+      "grad_norm": 12.158601760864258,
+      "learning_rate": 9.284851830262892e-06,
+      "loss": 2.6083,
+      "step": 724400
+    },
+    {
+      "epoch": 0.1889489629189616,
+      "grad_norm": 13.806035995483398,
+      "learning_rate": 9.284420960784873e-06,
+      "loss": 2.6437,
+      "step": 724600
+    },
+    {
+      "epoch": 0.18900111554466376,
+      "grad_norm": 12.63658332824707,
+      "learning_rate": 9.283989971552348e-06,
+      "loss": 2.5694,
+      "step": 724800
+    },
+    {
+      "epoch": 0.18905326817036594,
+      "grad_norm": 13.176187515258789,
+      "learning_rate": 9.283558862577363e-06,
+      "loss": 2.5266,
+      "step": 725000
+    },
+    {
+      "epoch": 0.1891054207960681,
+      "grad_norm": 11.850820541381836,
+      "learning_rate": 9.28312763387197e-06,
+      "loss": 2.5799,
+      "step": 725200
+    },
+    {
+      "epoch": 0.18915757342177028,
+      "grad_norm": 12.633203506469727,
+      "learning_rate": 9.282696285448219e-06,
+      "loss": 2.5638,
+      "step": 725400
+    },
+    {
+      "epoch": 0.18920972604747244,
+      "grad_norm": 11.37259292602539,
+      "learning_rate": 9.28226481731817e-06,
+      "loss": 2.5573,
+      "step": 725600
+    },
+    {
+      "epoch": 0.18926187867317462,
+      "grad_norm": 12.123799324035645,
+      "learning_rate": 9.28183322949388e-06,
+      "loss": 2.5736,
+      "step": 725800
+    },
+    {
+      "epoch": 0.18931403129887678,
+      "grad_norm": 11.90881061553955,
+      "learning_rate": 9.281401521987416e-06,
+      "loss": 2.5502,
+      "step": 726000
+    },
+    {
+      "epoch": 0.18936618392457896,
+      "grad_norm": 12.91506576538086,
+      "learning_rate": 9.280969694810844e-06,
+      "loss": 2.5229,
+      "step": 726200
+    },
+    {
+      "epoch": 0.18941833655028112,
+      "grad_norm": 11.650004386901855,
+      "learning_rate": 9.28053774797623e-06,
+      "loss": 2.5783,
+      "step": 726400
+    },
+    {
+      "epoch": 0.1894704891759833,
+      "grad_norm": 12.156107902526855,
+      "learning_rate": 9.280105681495652e-06,
+      "loss": 2.604,
+      "step": 726600
+    },
+    {
+      "epoch": 0.18952264180168546,
+      "grad_norm": 10.867297172546387,
+      "learning_rate": 9.279673495381185e-06,
+      "loss": 2.5855,
+      "step": 726800
+    },
+    {
+      "epoch": 0.18957479442738764,
+      "grad_norm": 12.994755744934082,
+      "learning_rate": 9.279241189644908e-06,
+      "loss": 2.5467,
+      "step": 727000
+    },
+    {
+      "epoch": 0.1896269470530898,
+      "grad_norm": 11.489134788513184,
+      "learning_rate": 9.278808764298906e-06,
+      "loss": 2.5654,
+      "step": 727200
+    },
+    {
+      "epoch": 0.18967909967879198,
+      "grad_norm": 12.920225143432617,
+      "learning_rate": 9.278376219355267e-06,
+      "loss": 2.5749,
+      "step": 727400
+    },
+    {
+      "epoch": 0.18973125230449414,
+      "grad_norm": 9.510746002197266,
+      "learning_rate": 9.277943554826076e-06,
+      "loss": 2.5537,
+      "step": 727600
+    },
+    {
+      "epoch": 0.18978340493019633,
+      "grad_norm": 12.295012474060059,
+      "learning_rate": 9.277510770723433e-06,
+      "loss": 2.5912,
+      "step": 727800
+    },
+    {
+      "epoch": 0.18983555755589848,
+      "grad_norm": 10.54262638092041,
+      "learning_rate": 9.277077867059432e-06,
+      "loss": 2.591,
+      "step": 728000
+    },
+    {
+      "epoch": 0.18988771018160067,
+      "grad_norm": 11.955072402954102,
+      "learning_rate": 9.276644843846171e-06,
+      "loss": 2.5709,
+      "step": 728200
+    },
+    {
+      "epoch": 0.18993986280730282,
+      "grad_norm": 10.959343910217285,
+      "learning_rate": 9.276211701095756e-06,
+      "loss": 2.5735,
+      "step": 728400
+    },
+    {
+      "epoch": 0.189992015433005,
+      "grad_norm": 12.265323638916016,
+      "learning_rate": 9.275778438820294e-06,
+      "loss": 2.5402,
+      "step": 728600
+    },
+    {
+      "epoch": 0.19004416805870716,
+      "grad_norm": 13.312240600585938,
+      "learning_rate": 9.275345057031894e-06,
+      "loss": 2.5894,
+      "step": 728800
+    },
+    {
+      "epoch": 0.19009632068440935,
+      "grad_norm": 12.820699691772461,
+      "learning_rate": 9.27491155574267e-06,
+      "loss": 2.5495,
+      "step": 729000
+    },
+    {
+      "epoch": 0.1901484733101115,
+      "grad_norm": 12.239838600158691,
+      "learning_rate": 9.27447793496474e-06,
+      "loss": 2.5772,
+      "step": 729200
+    },
+    {
+      "epoch": 0.1902006259358137,
+      "grad_norm": 12.58304214477539,
+      "learning_rate": 9.274044194710222e-06,
+      "loss": 2.5978,
+      "step": 729400
+    },
+    {
+      "epoch": 0.19025277856151585,
+      "grad_norm": 12.55810260772705,
+      "learning_rate": 9.27361033499124e-06,
+      "loss": 2.6003,
+      "step": 729600
+    },
+    {
+      "epoch": 0.19030493118721803,
+      "grad_norm": 11.778428077697754,
+      "learning_rate": 9.273176355819922e-06,
+      "loss": 2.5742,
+      "step": 729800
+    },
+    {
+      "epoch": 0.19035708381292019,
+      "grad_norm": 13.294259071350098,
+      "learning_rate": 9.272742257208399e-06,
+      "loss": 2.5477,
+      "step": 730000
+    },
+    {
+      "epoch": 0.19040923643862237,
+      "grad_norm": 13.392877578735352,
+      "learning_rate": 9.272308039168802e-06,
+      "loss": 2.535,
+      "step": 730200
+    },
+    {
+      "epoch": 0.19046138906432453,
+      "grad_norm": 11.833348274230957,
+      "learning_rate": 9.271873701713269e-06,
+      "loss": 2.5452,
+      "step": 730400
+    },
+    {
+      "epoch": 0.19051354169002668,
+      "grad_norm": 12.308874130249023,
+      "learning_rate": 9.27143924485394e-06,
+      "loss": 2.5484,
+      "step": 730600
+    },
+    {
+      "epoch": 0.19056569431572887,
+      "grad_norm": 10.478835105895996,
+      "learning_rate": 9.27100466860296e-06,
+      "loss": 2.5708,
+      "step": 730800
+    },
+    {
+      "epoch": 0.19061784694143102,
+      "grad_norm": 11.751260757446289,
+      "learning_rate": 9.270569972972475e-06,
+      "loss": 2.5554,
+      "step": 731000
+    },
+    {
+      "epoch": 0.1906699995671332,
+      "grad_norm": 13.028668403625488,
+      "learning_rate": 9.270135157974633e-06,
+      "loss": 2.5585,
+      "step": 731200
+    },
+    {
+      "epoch": 0.19072215219283536,
+      "grad_norm": 11.351934432983398,
+      "learning_rate": 9.269700223621591e-06,
+      "loss": 2.5808,
+      "step": 731400
+    },
+    {
+      "epoch": 0.19077430481853755,
+      "grad_norm": 11.560927391052246,
+      "learning_rate": 9.269265169925506e-06,
+      "loss": 2.5884,
+      "step": 731600
+    },
+    {
+      "epoch": 0.1908264574442397,
+      "grad_norm": 11.362569808959961,
+      "learning_rate": 9.268829996898537e-06,
+      "loss": 2.6066,
+      "step": 731800
+    },
+    {
+      "epoch": 0.1908786100699419,
+      "grad_norm": 12.66362476348877,
+      "learning_rate": 9.268394704552845e-06,
+      "loss": 2.5914,
+      "step": 732000
+    },
+    {
+      "epoch": 0.19093076269564405,
+      "grad_norm": 10.702469825744629,
+      "learning_rate": 9.2679592929006e-06,
+      "loss": 2.5223,
+      "step": 732200
+    },
+    {
+      "epoch": 0.19098291532134623,
+      "grad_norm": 12.114631652832031,
+      "learning_rate": 9.267523761953973e-06,
+      "loss": 2.5882,
+      "step": 732400
+    },
+    {
+      "epoch": 0.1910350679470484,
+      "grad_norm": 10.952610969543457,
+      "learning_rate": 9.267088111725135e-06,
+      "loss": 2.5453,
+      "step": 732600
+    },
+    {
+      "epoch": 0.19108722057275057,
+      "grad_norm": 11.303411483764648,
+      "learning_rate": 9.266652342226264e-06,
+      "loss": 2.5963,
+      "step": 732800
+    },
+    {
+      "epoch": 0.19113937319845273,
+      "grad_norm": 11.67508602142334,
+      "learning_rate": 9.26621645346954e-06,
+      "loss": 2.5723,
+      "step": 733000
+    },
+    {
+      "epoch": 0.1911915258241549,
+      "grad_norm": 11.651264190673828,
+      "learning_rate": 9.265780445467148e-06,
+      "loss": 2.5943,
+      "step": 733200
+    },
+    {
+      "epoch": 0.19124367844985707,
+      "grad_norm": 12.28518009185791,
+      "learning_rate": 9.265344318231274e-06,
+      "loss": 2.5283,
+      "step": 733400
+    },
+    {
+      "epoch": 0.19129583107555925,
+      "grad_norm": 13.750175476074219,
+      "learning_rate": 9.264908071774108e-06,
+      "loss": 2.5599,
+      "step": 733600
+    },
+    {
+      "epoch": 0.1913479837012614,
+      "grad_norm": 12.860754013061523,
+      "learning_rate": 9.264471706107846e-06,
+      "loss": 2.5571,
+      "step": 733800
+    },
+    {
+      "epoch": 0.1914001363269636,
+      "grad_norm": 11.767181396484375,
+      "learning_rate": 9.264035221244678e-06,
+      "loss": 2.5767,
+      "step": 734000
+    },
+    {
+      "epoch": 0.19145228895266575,
+      "grad_norm": 13.27584171295166,
+      "learning_rate": 9.263598617196812e-06,
+      "loss": 2.5659,
+      "step": 734200
+    },
+    {
+      "epoch": 0.19150444157836793,
+      "grad_norm": 11.297654151916504,
+      "learning_rate": 9.263161893976446e-06,
+      "loss": 2.5645,
+      "step": 734400
+    },
+    {
+      "epoch": 0.1915565942040701,
+      "grad_norm": 13.385966300964355,
+      "learning_rate": 9.262725051595793e-06,
+      "loss": 2.5824,
+      "step": 734600
+    },
+    {
+      "epoch": 0.19160874682977228,
+      "grad_norm": 12.745625495910645,
+      "learning_rate": 9.262288090067058e-06,
+      "loss": 2.5605,
+      "step": 734800
+    },
+    {
+      "epoch": 0.19166089945547443,
+      "grad_norm": 13.604351043701172,
+      "learning_rate": 9.261851009402457e-06,
+      "loss": 2.5457,
+      "step": 735000
+    },
+    {
+      "epoch": 0.19171305208117662,
+      "grad_norm": 14.262986183166504,
+      "learning_rate": 9.261413809614204e-06,
+      "loss": 2.5429,
+      "step": 735200
+    },
+    {
+      "epoch": 0.19176520470687877,
+      "grad_norm": 13.41905689239502,
+      "learning_rate": 9.260976490714524e-06,
+      "loss": 2.5912,
+      "step": 735400
+    },
+    {
+      "epoch": 0.19181735733258096,
+      "grad_norm": 11.548351287841797,
+      "learning_rate": 9.260539052715636e-06,
+      "loss": 2.55,
+      "step": 735600
+    },
+    {
+      "epoch": 0.1918695099582831,
+      "grad_norm": 12.51290512084961,
+      "learning_rate": 9.26010149562977e-06,
+      "loss": 2.557,
+      "step": 735800
+    },
+    {
+      "epoch": 0.1919216625839853,
+      "grad_norm": 11.825644493103027,
+      "learning_rate": 9.259663819469154e-06,
+      "loss": 2.5657,
+      "step": 736000
+    },
+    {
+      "epoch": 0.19197381520968745,
+      "grad_norm": 11.373788833618164,
+      "learning_rate": 9.259226024246024e-06,
+      "loss": 2.5175,
+      "step": 736200
+    },
+    {
+      "epoch": 0.19202596783538964,
+      "grad_norm": 13.40487003326416,
+      "learning_rate": 9.258788109972616e-06,
+      "loss": 2.5853,
+      "step": 736400
+    },
+    {
+      "epoch": 0.1920781204610918,
+      "grad_norm": 11.846600532531738,
+      "learning_rate": 9.258350076661168e-06,
+      "loss": 2.5468,
+      "step": 736600
+    },
+    {
+      "epoch": 0.19213027308679395,
+      "grad_norm": 11.58352279663086,
+      "learning_rate": 9.257911924323926e-06,
+      "loss": 2.5769,
+      "step": 736800
+    },
+    {
+      "epoch": 0.19218242571249614,
+      "grad_norm": 11.748197555541992,
+      "learning_rate": 9.257473652973137e-06,
+      "loss": 2.5518,
+      "step": 737000
+    },
+    {
+      "epoch": 0.1922345783381983,
+      "grad_norm": 10.648813247680664,
+      "learning_rate": 9.25703526262105e-06,
+      "loss": 2.5885,
+      "step": 737200
+    },
+    {
+      "epoch": 0.19228673096390048,
+      "grad_norm": 13.590847969055176,
+      "learning_rate": 9.256596753279918e-06,
+      "loss": 2.5561,
+      "step": 737400
+    },
+    {
+      "epoch": 0.19233888358960263,
+      "grad_norm": 11.725268363952637,
+      "learning_rate": 9.256158124962e-06,
+      "loss": 2.5282,
+      "step": 737600
+    },
+    {
+      "epoch": 0.19239103621530482,
+      "grad_norm": 12.270450592041016,
+      "learning_rate": 9.255719377679552e-06,
+      "loss": 2.5635,
+      "step": 737800
+    },
+    {
+      "epoch": 0.19244318884100697,
+      "grad_norm": 10.960904121398926,
+      "learning_rate": 9.255280511444842e-06,
+      "loss": 2.5165,
+      "step": 738000
+    },
+    {
+      "epoch": 0.19249534146670916,
+      "grad_norm": 11.803460121154785,
+      "learning_rate": 9.254841526270137e-06,
+      "loss": 2.5727,
+      "step": 738200
+    },
+    {
+      "epoch": 0.19254749409241131,
+      "grad_norm": 12.786358833312988,
+      "learning_rate": 9.254402422167704e-06,
+      "loss": 2.5412,
+      "step": 738400
+    },
+    {
+      "epoch": 0.1925996467181135,
+      "grad_norm": 12.822945594787598,
+      "learning_rate": 9.253963199149818e-06,
+      "loss": 2.5498,
+      "step": 738600
+    },
+    {
+      "epoch": 0.19265179934381566,
+      "grad_norm": 11.774388313293457,
+      "learning_rate": 9.253523857228754e-06,
+      "loss": 2.5893,
+      "step": 738800
+    },
+    {
+      "epoch": 0.19270395196951784,
+      "grad_norm": 12.514403343200684,
+      "learning_rate": 9.253084396416795e-06,
+      "loss": 2.6186,
+      "step": 739000
+    },
+    {
+      "epoch": 0.19275610459522,
+      "grad_norm": 11.048097610473633,
+      "learning_rate": 9.252644816726224e-06,
+      "loss": 2.5628,
+      "step": 739200
+    },
+    {
+      "epoch": 0.19280825722092218,
+      "grad_norm": 11.978404998779297,
+      "learning_rate": 9.252205118169326e-06,
+      "loss": 2.592,
+      "step": 739400
+    },
+    {
+      "epoch": 0.19286040984662434,
+      "grad_norm": 11.403801918029785,
+      "learning_rate": 9.25176530075839e-06,
+      "loss": 2.5574,
+      "step": 739600
+    },
+    {
+      "epoch": 0.19291256247232652,
+      "grad_norm": 13.35159969329834,
+      "learning_rate": 9.251325364505715e-06,
+      "loss": 2.5446,
+      "step": 739800
+    },
+    {
+      "epoch": 0.19296471509802868,
+      "grad_norm": 11.937582969665527,
+      "learning_rate": 9.250885309423592e-06,
+      "loss": 2.5404,
+      "step": 740000
+    },
+    {
+      "epoch": 0.19301686772373086,
+      "grad_norm": 12.94690990447998,
+      "learning_rate": 9.250445135524325e-06,
+      "loss": 2.5505,
+      "step": 740200
+    },
+    {
+      "epoch": 0.19306902034943302,
+      "grad_norm": 13.668988227844238,
+      "learning_rate": 9.250004842820212e-06,
+      "loss": 2.5818,
+      "step": 740400
+    },
+    {
+      "epoch": 0.1931211729751352,
+      "grad_norm": 11.1817626953125,
+      "learning_rate": 9.249564431323566e-06,
+      "loss": 2.574,
+      "step": 740600
+    },
+    {
+      "epoch": 0.19317332560083736,
+      "grad_norm": 12.859142303466797,
+      "learning_rate": 9.249123901046694e-06,
+      "loss": 2.5295,
+      "step": 740800
+    },
+    {
+      "epoch": 0.19322547822653954,
+      "grad_norm": 10.10409164428711,
+      "learning_rate": 9.248683252001909e-06,
+      "loss": 2.557,
+      "step": 741000
+    },
+    {
+      "epoch": 0.1932776308522417,
+      "grad_norm": 13.226909637451172,
+      "learning_rate": 9.248242484201528e-06,
+      "loss": 2.5851,
+      "step": 741200
+    },
+    {
+      "epoch": 0.19332978347794388,
+      "grad_norm": 12.193153381347656,
+      "learning_rate": 9.247801597657871e-06,
+      "loss": 2.5712,
+      "step": 741400
+    },
+    {
+      "epoch": 0.19338193610364604,
+      "grad_norm": 12.625770568847656,
+      "learning_rate": 9.24736059238326e-06,
+      "loss": 2.5571,
+      "step": 741600
+    },
+    {
+      "epoch": 0.19343408872934822,
+      "grad_norm": 13.078922271728516,
+      "learning_rate": 9.246919468390026e-06,
+      "loss": 2.5776,
+      "step": 741800
+    },
+    {
+      "epoch": 0.19348624135505038,
+      "grad_norm": 12.740152359008789,
+      "learning_rate": 9.246478225690496e-06,
+      "loss": 2.5985,
+      "step": 742000
+    },
+    {
+      "epoch": 0.19353839398075257,
+      "grad_norm": 13.259086608886719,
+      "learning_rate": 9.246036864297001e-06,
+      "loss": 2.5446,
+      "step": 742200
+    },
+    {
+      "epoch": 0.19359054660645472,
+      "grad_norm": 12.893516540527344,
+      "learning_rate": 9.245595384221881e-06,
+      "loss": 2.5904,
+      "step": 742400
+    },
+    {
+      "epoch": 0.19364269923215688,
+      "grad_norm": 11.758481979370117,
+      "learning_rate": 9.245153785477475e-06,
+      "loss": 2.5576,
+      "step": 742600
+    },
+    {
+      "epoch": 0.19369485185785906,
+      "grad_norm": 11.747676849365234,
+      "learning_rate": 9.244712068076125e-06,
+      "loss": 2.5636,
+      "step": 742800
+    },
+    {
+      "epoch": 0.19374700448356122,
+      "grad_norm": 12.008854866027832,
+      "learning_rate": 9.24427023203018e-06,
+      "loss": 2.5726,
+      "step": 743000
+    },
+    {
+      "epoch": 0.1937991571092634,
+      "grad_norm": 9.737689971923828,
+      "learning_rate": 9.243828277351985e-06,
+      "loss": 2.5412,
+      "step": 743200
+    },
+    {
+      "epoch": 0.19385130973496556,
+      "grad_norm": 13.055411338806152,
+      "learning_rate": 9.2433862040539e-06,
+      "loss": 2.5482,
+      "step": 743400
+    },
+    {
+      "epoch": 0.19390346236066774,
+      "grad_norm": 11.527270317077637,
+      "learning_rate": 9.242944012148273e-06,
+      "loss": 2.5932,
+      "step": 743600
+    },
+    {
+      "epoch": 0.1939556149863699,
+      "grad_norm": 11.464157104492188,
+      "learning_rate": 9.242501701647473e-06,
+      "loss": 2.5869,
+      "step": 743800
+    },
+    {
+      "epoch": 0.19400776761207209,
+      "grad_norm": 13.078601837158203,
+      "learning_rate": 9.242059272563858e-06,
+      "loss": 2.54,
+      "step": 744000
+    },
+    {
+      "epoch": 0.19405992023777424,
+      "grad_norm": 10.983396530151367,
+      "learning_rate": 9.241616724909794e-06,
+      "loss": 2.5761,
+      "step": 744200
+    },
+    {
+      "epoch": 0.19411207286347643,
+      "grad_norm": 10.754619598388672,
+      "learning_rate": 9.241174058697651e-06,
+      "loss": 2.5597,
+      "step": 744400
+    },
+    {
+      "epoch": 0.19416422548917858,
+      "grad_norm": 13.653790473937988,
+      "learning_rate": 9.240731273939804e-06,
+      "loss": 2.5787,
+      "step": 744600
+    },
+    {
+      "epoch": 0.19421637811488077,
+      "grad_norm": 11.377376556396484,
+      "learning_rate": 9.240288370648628e-06,
+      "loss": 2.5587,
+      "step": 744800
+    },
+    {
+      "epoch": 0.19426853074058292,
+      "grad_norm": 10.749224662780762,
+      "learning_rate": 9.239845348836503e-06,
+      "loss": 2.5793,
+      "step": 745000
+    },
+    {
+      "epoch": 0.1943206833662851,
+      "grad_norm": 13.052656173706055,
+      "learning_rate": 9.23940220851581e-06,
+      "loss": 2.5555,
+      "step": 745200
+    },
+    {
+      "epoch": 0.19437283599198726,
+      "grad_norm": 12.965775489807129,
+      "learning_rate": 9.23895894969894e-06,
+      "loss": 2.5797,
+      "step": 745400
+    },
+    {
+      "epoch": 0.19442498861768945,
+      "grad_norm": 13.597799301147461,
+      "learning_rate": 9.238515572398278e-06,
+      "loss": 2.5492,
+      "step": 745600
+    },
+    {
+      "epoch": 0.1944771412433916,
+      "grad_norm": 11.634117126464844,
+      "learning_rate": 9.23807207662622e-06,
+      "loss": 2.5746,
+      "step": 745800
+    },
+    {
+      "epoch": 0.1945292938690938,
+      "grad_norm": 12.693094253540039,
+      "learning_rate": 9.237628462395158e-06,
+      "loss": 2.6023,
+      "step": 746000
+    },
+    {
+      "epoch": 0.19458144649479595,
+      "grad_norm": 15.489215850830078,
+      "learning_rate": 9.237184729717496e-06,
+      "loss": 2.5555,
+      "step": 746200
+    },
+    {
+      "epoch": 0.19463359912049813,
+      "grad_norm": 13.176939010620117,
+      "learning_rate": 9.236740878605635e-06,
+      "loss": 2.5887,
+      "step": 746400
+    },
+    {
+      "epoch": 0.1946857517462003,
+      "grad_norm": 12.691610336303711,
+      "learning_rate": 9.23629690907198e-06,
+      "loss": 2.5661,
+      "step": 746600
+    },
+    {
+      "epoch": 0.19473790437190247,
+      "grad_norm": 11.665881156921387,
+      "learning_rate": 9.235852821128943e-06,
+      "loss": 2.5562,
+      "step": 746800
+    },
+    {
+      "epoch": 0.19479005699760463,
+      "grad_norm": 13.195979118347168,
+      "learning_rate": 9.235408614788937e-06,
+      "loss": 2.5584,
+      "step": 747000
+    },
+    {
+      "epoch": 0.1948422096233068,
+      "grad_norm": 11.392936706542969,
+      "learning_rate": 9.234964290064375e-06,
+      "loss": 2.5803,
+      "step": 747200
+    },
+    {
+      "epoch": 0.19489436224900897,
+      "grad_norm": 11.263978004455566,
+      "learning_rate": 9.234519846967678e-06,
+      "loss": 2.6038,
+      "step": 747400
+    },
+    {
+      "epoch": 0.19494651487471115,
+      "grad_norm": 10.900824546813965,
+      "learning_rate": 9.234075285511268e-06,
+      "loss": 2.5391,
+      "step": 747600
+    },
+    {
+      "epoch": 0.1949986675004133,
+      "grad_norm": 11.745079040527344,
+      "learning_rate": 9.233630605707573e-06,
+      "loss": 2.6114,
+      "step": 747800
+    },
+    {
+      "epoch": 0.1950508201261155,
+      "grad_norm": 11.821778297424316,
+      "learning_rate": 9.23318580756902e-06,
+      "loss": 2.6185,
+      "step": 748000
+    },
+    {
+      "epoch": 0.19510297275181765,
+      "grad_norm": 11.111929893493652,
+      "learning_rate": 9.232740891108042e-06,
+      "loss": 2.6214,
+      "step": 748200
+    },
+    {
+      "epoch": 0.1951551253775198,
+      "grad_norm": 11.941657066345215,
+      "learning_rate": 9.232295856337078e-06,
+      "loss": 2.5602,
+      "step": 748400
+    },
+    {
+      "epoch": 0.195207278003222,
+      "grad_norm": 11.378263473510742,
+      "learning_rate": 9.231850703268563e-06,
+      "loss": 2.5749,
+      "step": 748600
+    },
+    {
+      "epoch": 0.19525943062892415,
+      "grad_norm": 10.591818809509277,
+      "learning_rate": 9.231405431914942e-06,
+      "loss": 2.5177,
+      "step": 748800
+    },
+    {
+      "epoch": 0.19531158325462633,
+      "grad_norm": 12.579631805419922,
+      "learning_rate": 9.230960042288661e-06,
+      "loss": 2.5455,
+      "step": 749000
+    },
+    {
+      "epoch": 0.1953637358803285,
+      "grad_norm": 13.101773262023926,
+      "learning_rate": 9.230514534402166e-06,
+      "loss": 2.6177,
+      "step": 749200
+    },
+    {
+      "epoch": 0.19541588850603067,
+      "grad_norm": 12.09244441986084,
+      "learning_rate": 9.230068908267916e-06,
+      "loss": 2.5752,
+      "step": 749400
+    },
+    {
+      "epoch": 0.19546804113173283,
+      "grad_norm": 12.434764862060547,
+      "learning_rate": 9.229623163898358e-06,
+      "loss": 2.5857,
+      "step": 749600
+    },
+    {
+      "epoch": 0.195520193757435,
+      "grad_norm": 11.809731483459473,
+      "learning_rate": 9.229177301305957e-06,
+      "loss": 2.5606,
+      "step": 749800
+    },
+    {
+      "epoch": 0.19557234638313717,
+      "grad_norm": 12.460705757141113,
+      "learning_rate": 9.228731320503174e-06,
+      "loss": 2.574,
+      "step": 750000
+    },
+    {
+      "epoch": 0.19562449900883935,
+      "grad_norm": 12.289457321166992,
+      "learning_rate": 9.228285221502475e-06,
+      "loss": 2.5422,
+      "step": 750200
+    },
+    {
+      "epoch": 0.1956766516345415,
+      "grad_norm": 12.129727363586426,
+      "learning_rate": 9.227839004316329e-06,
+      "loss": 2.5749,
+      "step": 750400
+    },
+    {
+      "epoch": 0.1957288042602437,
+      "grad_norm": 12.88148307800293,
+      "learning_rate": 9.227392668957208e-06,
+      "loss": 2.6006,
+      "step": 750600
+    },
+    {
+      "epoch": 0.19578095688594585,
+      "grad_norm": 11.499628067016602,
+      "learning_rate": 9.226946215437586e-06,
+      "loss": 2.5805,
+      "step": 750800
+    },
+    {
+      "epoch": 0.19583310951164803,
+      "grad_norm": 11.71347427368164,
+      "learning_rate": 9.226499643769945e-06,
+      "loss": 2.587,
+      "step": 751000
+    },
+    {
+      "epoch": 0.1958852621373502,
+      "grad_norm": 11.851282119750977,
+      "learning_rate": 9.226052953966766e-06,
+      "loss": 2.5516,
+      "step": 751200
+    },
+    {
+      "epoch": 0.19593741476305238,
+      "grad_norm": 14.087993621826172,
+      "learning_rate": 9.225606146040533e-06,
+      "loss": 2.5593,
+      "step": 751400
+    },
+    {
+      "epoch": 0.19598956738875453,
+      "grad_norm": 13.594001770019531,
+      "learning_rate": 9.225159220003737e-06,
+      "loss": 2.5425,
+      "step": 751600
+    },
+    {
+      "epoch": 0.19604172001445672,
+      "grad_norm": 10.394376754760742,
+      "learning_rate": 9.224712175868866e-06,
+      "loss": 2.5569,
+      "step": 751800
+    },
+    {
+      "epoch": 0.19609387264015887,
+      "grad_norm": 13.085862159729004,
+      "learning_rate": 9.224265013648421e-06,
+      "loss": 2.5548,
+      "step": 752000
+    },
+    {
+      "epoch": 0.19614602526586106,
+      "grad_norm": 12.25311279296875,
+      "learning_rate": 9.223817733354898e-06,
+      "loss": 2.5291,
+      "step": 752200
+    },
+    {
+      "epoch": 0.1961981778915632,
+      "grad_norm": 10.312944412231445,
+      "learning_rate": 9.223370335000799e-06,
+      "loss": 2.5391,
+      "step": 752400
+    },
+    {
+      "epoch": 0.1962503305172654,
+      "grad_norm": 13.032076835632324,
+      "learning_rate": 9.22292281859863e-06,
+      "loss": 2.5997,
+      "step": 752600
+    },
+    {
+      "epoch": 0.19630248314296755,
+      "grad_norm": 11.651784896850586,
+      "learning_rate": 9.222475184160897e-06,
+      "loss": 2.5493,
+      "step": 752800
+    },
+    {
+      "epoch": 0.19635463576866974,
+      "grad_norm": 12.205024719238281,
+      "learning_rate": 9.222027431700116e-06,
+      "loss": 2.5583,
+      "step": 753000
+    },
+    {
+      "epoch": 0.1964067883943719,
+      "grad_norm": 12.745562553405762,
+      "learning_rate": 9.221579561228799e-06,
+      "loss": 2.5554,
+      "step": 753200
+    },
+    {
+      "epoch": 0.19645894102007408,
+      "grad_norm": 13.312294006347656,
+      "learning_rate": 9.221131572759464e-06,
+      "loss": 2.5625,
+      "step": 753400
+    },
+    {
+      "epoch": 0.19651109364577624,
+      "grad_norm": 11.942469596862793,
+      "learning_rate": 9.220683466304638e-06,
+      "loss": 2.5518,
+      "step": 753600
+    },
+    {
+      "epoch": 0.19656324627147842,
+      "grad_norm": 12.910618782043457,
+      "learning_rate": 9.22023524187684e-06,
+      "loss": 2.554,
+      "step": 753800
+    },
+    {
+      "epoch": 0.19661539889718058,
+      "grad_norm": 9.96152400970459,
+      "learning_rate": 9.219786899488603e-06,
+      "loss": 2.5309,
+      "step": 754000
+    },
+    {
+      "epoch": 0.19666755152288273,
+      "grad_norm": 12.984268188476562,
+      "learning_rate": 9.219338439152455e-06,
+      "loss": 2.5723,
+      "step": 754200
+    },
+    {
+      "epoch": 0.19671970414858492,
+      "grad_norm": 12.589516639709473,
+      "learning_rate": 9.218889860880932e-06,
+      "loss": 2.5655,
+      "step": 754400
+    },
+    {
+      "epoch": 0.19677185677428707,
+      "grad_norm": 13.960628509521484,
+      "learning_rate": 9.218441164686573e-06,
+      "loss": 2.5815,
+      "step": 754600
+    },
+    {
+      "epoch": 0.19682400939998926,
+      "grad_norm": 11.358234405517578,
+      "learning_rate": 9.217992350581921e-06,
+      "loss": 2.574,
+      "step": 754800
+    },
+    {
+      "epoch": 0.19687616202569141,
+      "grad_norm": 12.214129447937012,
+      "learning_rate": 9.217543418579517e-06,
+      "loss": 2.5392,
+      "step": 755000
+    },
+    {
+      "epoch": 0.1969283146513936,
+      "grad_norm": 14.097967147827148,
+      "learning_rate": 9.217094368691913e-06,
+      "loss": 2.6056,
+      "step": 755200
+    },
+    {
+      "epoch": 0.19698046727709576,
+      "grad_norm": 12.386037826538086,
+      "learning_rate": 9.216645200931657e-06,
+      "loss": 2.5794,
+      "step": 755400
+    },
+    {
+      "epoch": 0.19703261990279794,
+      "grad_norm": 12.34218692779541,
+      "learning_rate": 9.216195915311307e-06,
+      "loss": 2.5821,
+      "step": 755600
+    },
+    {
+      "epoch": 0.1970847725285001,
+      "grad_norm": 13.724196434020996,
+      "learning_rate": 9.215746511843422e-06,
+      "loss": 2.5602,
+      "step": 755800
+    },
+    {
+      "epoch": 0.19713692515420228,
+      "grad_norm": 12.456463813781738,
+      "learning_rate": 9.21529699054056e-06,
+      "loss": 2.5832,
+      "step": 756000
+    },
+    {
+      "epoch": 0.19718907777990444,
+      "grad_norm": 12.305964469909668,
+      "learning_rate": 9.214847351415283e-06,
+      "loss": 2.5294,
+      "step": 756200
+    },
+    {
+      "epoch": 0.19724123040560662,
+      "grad_norm": 11.164262771606445,
+      "learning_rate": 9.214397594480165e-06,
+      "loss": 2.5495,
+      "step": 756400
+    },
+    {
+      "epoch": 0.19729338303130878,
+      "grad_norm": 13.309433937072754,
+      "learning_rate": 9.213947719747775e-06,
+      "loss": 2.5062,
+      "step": 756600
+    },
+    {
+      "epoch": 0.19734553565701096,
+      "grad_norm": 12.617058753967285,
+      "learning_rate": 9.213497727230688e-06,
+      "loss": 2.5857,
+      "step": 756800
+    },
+    {
+      "epoch": 0.19739768828271312,
+      "grad_norm": 11.997270584106445,
+      "learning_rate": 9.21304761694148e-06,
+      "loss": 2.5594,
+      "step": 757000
+    },
+    {
+      "epoch": 0.1974498409084153,
+      "grad_norm": 11.326188087463379,
+      "learning_rate": 9.212597388892734e-06,
+      "loss": 2.5126,
+      "step": 757200
+    },
+    {
+      "epoch": 0.19750199353411746,
+      "grad_norm": 13.245183944702148,
+      "learning_rate": 9.212147043097034e-06,
+      "loss": 2.5587,
+      "step": 757400
+    },
+    {
+      "epoch": 0.19755414615981964,
+      "grad_norm": 12.528741836547852,
+      "learning_rate": 9.211696579566967e-06,
+      "loss": 2.5524,
+      "step": 757600
+    },
+    {
+      "epoch": 0.1976062987855218,
+      "grad_norm": 12.96116828918457,
+      "learning_rate": 9.211245998315124e-06,
+      "loss": 2.5411,
+      "step": 757800
+    },
+    {
+      "epoch": 0.19765845141122398,
+      "grad_norm": 14.029032707214355,
+      "learning_rate": 9.210795299354101e-06,
+      "loss": 2.5702,
+      "step": 758000
+    },
+    {
+      "epoch": 0.19771060403692614,
+      "grad_norm": 12.810486793518066,
+      "learning_rate": 9.210344482696491e-06,
+      "loss": 2.5369,
+      "step": 758200
+    },
+    {
+      "epoch": 0.19776275666262833,
+      "grad_norm": 11.713202476501465,
+      "learning_rate": 9.2098935483549e-06,
+      "loss": 2.5273,
+      "step": 758400
+    },
+    {
+      "epoch": 0.19781490928833048,
+      "grad_norm": 11.515366554260254,
+      "learning_rate": 9.209442496341932e-06,
+      "loss": 2.5773,
+      "step": 758600
+    },
+    {
+      "epoch": 0.19786706191403267,
+      "grad_norm": 12.124978065490723,
+      "learning_rate": 9.208991326670189e-06,
+      "loss": 2.5723,
+      "step": 758800
+    },
+    {
+      "epoch": 0.19791921453973482,
+      "grad_norm": 11.972982406616211,
+      "learning_rate": 9.208540039352288e-06,
+      "loss": 2.5797,
+      "step": 759000
+    },
+    {
+      "epoch": 0.197971367165437,
+      "grad_norm": 12.521751403808594,
+      "learning_rate": 9.208088634400838e-06,
+      "loss": 2.5721,
+      "step": 759200
+    },
+    {
+      "epoch": 0.19802351979113916,
+      "grad_norm": 12.26280403137207,
+      "learning_rate": 9.20763711182846e-06,
+      "loss": 2.5352,
+      "step": 759400
+    },
+    {
+      "epoch": 0.19807567241684135,
+      "grad_norm": 11.379363059997559,
+      "learning_rate": 9.207185471647774e-06,
+      "loss": 2.5372,
+      "step": 759600
+    },
+    {
+      "epoch": 0.1981278250425435,
+      "grad_norm": 11.682640075683594,
+      "learning_rate": 9.2067337138714e-06,
+      "loss": 2.584,
+      "step": 759800
+    },
+    {
+      "epoch": 0.1981799776682457,
+      "grad_norm": 10.798892974853516,
+      "learning_rate": 9.206281838511972e-06,
+      "loss": 2.5728,
+      "step": 760000
+    },
+    {
+      "epoch": 0.19823213029394784,
+      "grad_norm": 13.537489891052246,
+      "learning_rate": 9.205829845582114e-06,
+      "loss": 2.5534,
+      "step": 760200
+    },
+    {
+      "epoch": 0.19828428291965,
+      "grad_norm": 12.42953872680664,
+      "learning_rate": 9.205377735094462e-06,
+      "loss": 2.5662,
+      "step": 760400
+    },
+    {
+      "epoch": 0.19833643554535219,
+      "grad_norm": 10.437345504760742,
+      "learning_rate": 9.204925507061655e-06,
+      "loss": 2.5577,
+      "step": 760600
+    },
+    {
+      "epoch": 0.19838858817105434,
+      "grad_norm": 12.269465446472168,
+      "learning_rate": 9.204473161496331e-06,
+      "loss": 2.5762,
+      "step": 760800
+    },
+    {
+      "epoch": 0.19844074079675653,
+      "grad_norm": 11.716769218444824,
+      "learning_rate": 9.204020698411133e-06,
+      "loss": 2.5548,
+      "step": 761000
+    },
+    {
+      "epoch": 0.19849289342245868,
+      "grad_norm": 13.604849815368652,
+      "learning_rate": 9.20356811781871e-06,
+      "loss": 2.5544,
+      "step": 761200
+    },
+    {
+      "epoch": 0.19854504604816087,
+      "grad_norm": 12.914355278015137,
+      "learning_rate": 9.203115419731711e-06,
+      "loss": 2.5554,
+      "step": 761400
+    },
+    {
+      "epoch": 0.19859719867386302,
+      "grad_norm": 13.604276657104492,
+      "learning_rate": 9.20266260416279e-06,
+      "loss": 2.556,
+      "step": 761600
+    },
+    {
+      "epoch": 0.1986493512995652,
+      "grad_norm": 12.59882640838623,
+      "learning_rate": 9.202209671124605e-06,
+      "loss": 2.5409,
+      "step": 761800
+    },
+    {
+      "epoch": 0.19870150392526736,
+      "grad_norm": 12.737641334533691,
+      "learning_rate": 9.20175662062981e-06,
+      "loss": 2.5656,
+      "step": 762000
+    },
+    {
+      "epoch": 0.19875365655096955,
+      "grad_norm": 11.206287384033203,
+      "learning_rate": 9.201303452691078e-06,
+      "loss": 2.5629,
+      "step": 762200
+    },
+    {
+      "epoch": 0.1988058091766717,
+      "grad_norm": 12.095778465270996,
+      "learning_rate": 9.200850167321067e-06,
+      "loss": 2.5611,
+      "step": 762400
+    },
+    {
+      "epoch": 0.1988579618023739,
+      "grad_norm": 11.354753494262695,
+      "learning_rate": 9.20039676453245e-06,
+      "loss": 2.5851,
+      "step": 762600
+    },
+    {
+      "epoch": 0.19891011442807605,
+      "grad_norm": 10.592757225036621,
+      "learning_rate": 9.199943244337901e-06,
+      "loss": 2.5733,
+      "step": 762800
+    },
+    {
+      "epoch": 0.19896226705377823,
+      "grad_norm": 12.773603439331055,
+      "learning_rate": 9.199489606750095e-06,
+      "loss": 2.578,
+      "step": 763000
+    },
+    {
+      "epoch": 0.1990144196794804,
+      "grad_norm": 11.842935562133789,
+      "learning_rate": 9.199035851781712e-06,
+      "loss": 2.5338,
+      "step": 763200
+    },
+    {
+      "epoch": 0.19906657230518257,
+      "grad_norm": 11.150446891784668,
+      "learning_rate": 9.198581979445437e-06,
+      "loss": 2.5977,
+      "step": 763400
+    },
+    {
+      "epoch": 0.19911872493088473,
+      "grad_norm": 13.279424667358398,
+      "learning_rate": 9.198127989753953e-06,
+      "loss": 2.5638,
+      "step": 763600
+    },
+    {
+      "epoch": 0.1991708775565869,
+      "grad_norm": 12.847058296203613,
+      "learning_rate": 9.197673882719953e-06,
+      "loss": 2.5607,
+      "step": 763800
+    },
+    {
+      "epoch": 0.19922303018228907,
+      "grad_norm": 12.287805557250977,
+      "learning_rate": 9.197219658356125e-06,
+      "loss": 2.6154,
+      "step": 764000
+    },
+    {
+      "epoch": 0.19927518280799125,
+      "grad_norm": 12.280625343322754,
+      "learning_rate": 9.196765316675169e-06,
+      "loss": 2.5747,
+      "step": 764200
+    },
+    {
+      "epoch": 0.1993273354336934,
+      "grad_norm": 12.880599975585938,
+      "learning_rate": 9.196310857689785e-06,
+      "loss": 2.5721,
+      "step": 764400
+    },
+    {
+      "epoch": 0.1993794880593956,
+      "grad_norm": 12.690503120422363,
+      "learning_rate": 9.195856281412672e-06,
+      "loss": 2.5614,
+      "step": 764600
+    },
+    {
+      "epoch": 0.19943164068509775,
+      "grad_norm": 11.280840873718262,
+      "learning_rate": 9.195401587856538e-06,
+      "loss": 2.5797,
+      "step": 764800
+    },
+    {
+      "epoch": 0.19948379331079993,
+      "grad_norm": 13.591432571411133,
+      "learning_rate": 9.194946777034094e-06,
+      "loss": 2.554,
+      "step": 765000
+    },
+    {
+      "epoch": 0.1995359459365021,
+      "grad_norm": 13.786008834838867,
+      "learning_rate": 9.19449184895805e-06,
+      "loss": 2.5347,
+      "step": 765200
+    },
+    {
+      "epoch": 0.19958809856220427,
+      "grad_norm": 11.823830604553223,
+      "learning_rate": 9.19403680364112e-06,
+      "loss": 2.551,
+      "step": 765400
+    },
+    {
+      "epoch": 0.19964025118790643,
+      "grad_norm": 11.602143287658691,
+      "learning_rate": 9.193581641096028e-06,
+      "loss": 2.577,
+      "step": 765600
+    },
+    {
+      "epoch": 0.19969240381360862,
+      "grad_norm": 14.088010787963867,
+      "learning_rate": 9.193126361335491e-06,
+      "loss": 2.5477,
+      "step": 765800
+    },
+    {
+      "epoch": 0.19974455643931077,
+      "grad_norm": 13.316328048706055,
+      "learning_rate": 9.19267096437224e-06,
+      "loss": 2.5517,
+      "step": 766000
+    },
+    {
+      "epoch": 0.19979670906501293,
+      "grad_norm": 11.534945487976074,
+      "learning_rate": 9.192215450219e-06,
+      "loss": 2.5514,
+      "step": 766200
+    },
+    {
+      "epoch": 0.1998488616907151,
+      "grad_norm": 11.87644100189209,
+      "learning_rate": 9.191759818888506e-06,
+      "loss": 2.5328,
+      "step": 766400
+    },
+    {
+      "epoch": 0.19990101431641727,
+      "grad_norm": 12.74470043182373,
+      "learning_rate": 9.19130407039349e-06,
+      "loss": 2.5838,
+      "step": 766600
+    },
+    {
+      "epoch": 0.19995316694211945,
+      "grad_norm": 11.58600902557373,
+      "learning_rate": 9.190848204746694e-06,
+      "loss": 2.5657,
+      "step": 766800
+    },
+    {
+      "epoch": 0.2000053195678216,
+      "grad_norm": 12.279742240905762,
+      "learning_rate": 9.190392221960858e-06,
+      "loss": 2.5543,
+      "step": 767000
+    },
+    {
+      "epoch": 0.2000574721935238,
+      "grad_norm": 12.125825881958008,
+      "learning_rate": 9.189936122048727e-06,
+      "loss": 2.5282,
+      "step": 767200
+    },
+    {
+      "epoch": 0.20010962481922595,
+      "grad_norm": 10.70932674407959,
+      "learning_rate": 9.18947990502305e-06,
+      "loss": 2.5886,
+      "step": 767400
+    },
+    {
+      "epoch": 0.20016177744492814,
+      "grad_norm": 12.008572578430176,
+      "learning_rate": 9.189023570896579e-06,
+      "loss": 2.5466,
+      "step": 767600
+    },
+    {
+      "epoch": 0.2002139300706303,
+      "grad_norm": 10.968779563903809,
+      "learning_rate": 9.18856711968207e-06,
+      "loss": 2.5674,
+      "step": 767800
+    },
+    {
+      "epoch": 0.20026608269633248,
+      "grad_norm": 11.873457908630371,
+      "learning_rate": 9.18811055139228e-06,
+      "loss": 2.5667,
+      "step": 768000
+    },
+    {
+      "epoch": 0.20031823532203463,
+      "grad_norm": 11.809953689575195,
+      "learning_rate": 9.18765386603997e-06,
+      "loss": 2.5602,
+      "step": 768200
+    },
+    {
+      "epoch": 0.20037038794773682,
+      "grad_norm": 12.026460647583008,
+      "learning_rate": 9.187197063637906e-06,
+      "loss": 2.5693,
+      "step": 768400
+    },
+    {
+      "epoch": 0.20042254057343897,
+      "grad_norm": 12.511942863464355,
+      "learning_rate": 9.186740144198858e-06,
+      "loss": 2.5677,
+      "step": 768600
+    },
+    {
+      "epoch": 0.20047469319914116,
+      "grad_norm": 11.590380668640137,
+      "learning_rate": 9.186283107735593e-06,
+      "loss": 2.5472,
+      "step": 768800
+    },
+    {
+      "epoch": 0.20052684582484331,
+      "grad_norm": 12.04472541809082,
+      "learning_rate": 9.18582595426089e-06,
+      "loss": 2.5422,
+      "step": 769000
+    },
+    {
+      "epoch": 0.2005789984505455,
+      "grad_norm": 13.985250473022461,
+      "learning_rate": 9.185368683787526e-06,
+      "loss": 2.5152,
+      "step": 769200
+    },
+    {
+      "epoch": 0.20063115107624765,
+      "grad_norm": 12.050628662109375,
+      "learning_rate": 9.184911296328279e-06,
+      "loss": 2.5715,
+      "step": 769400
+    },
+    {
+      "epoch": 0.20068330370194984,
+      "grad_norm": 13.415891647338867,
+      "learning_rate": 9.184453791895937e-06,
+      "loss": 2.5828,
+      "step": 769600
+    },
+    {
+      "epoch": 0.200735456327652,
+      "grad_norm": 13.432297706604004,
+      "learning_rate": 9.183996170503287e-06,
+      "loss": 2.5177,
+      "step": 769800
+    },
+    {
+      "epoch": 0.20078760895335418,
+      "grad_norm": 12.210216522216797,
+      "learning_rate": 9.18353843216312e-06,
+      "loss": 2.5423,
+      "step": 770000
+    },
+    {
+      "epoch": 0.20083976157905634,
+      "grad_norm": 11.473400115966797,
+      "learning_rate": 9.183080576888228e-06,
+      "loss": 2.5561,
+      "step": 770200
+    },
+    {
+      "epoch": 0.20089191420475852,
+      "grad_norm": 12.12957763671875,
+      "learning_rate": 9.182622604691411e-06,
+      "loss": 2.5614,
+      "step": 770400
+    },
+    {
+      "epoch": 0.20094406683046068,
+      "grad_norm": 11.806977272033691,
+      "learning_rate": 9.182164515585472e-06,
+      "loss": 2.5903,
+      "step": 770600
+    },
+    {
+      "epoch": 0.20099621945616286,
+      "grad_norm": 12.727505683898926,
+      "learning_rate": 9.181706309583209e-06,
+      "loss": 2.577,
+      "step": 770800
+    },
+    {
+      "epoch": 0.20104837208186502,
+      "grad_norm": 13.943395614624023,
+      "learning_rate": 9.181247986697436e-06,
+      "loss": 2.5928,
+      "step": 771000
+    },
+    {
+      "epoch": 0.2011005247075672,
+      "grad_norm": 12.006414413452148,
+      "learning_rate": 9.18078954694096e-06,
+      "loss": 2.5669,
+      "step": 771200
+    },
+    {
+      "epoch": 0.20115267733326936,
+      "grad_norm": 11.008611679077148,
+      "learning_rate": 9.180330990326593e-06,
+      "loss": 2.5689,
+      "step": 771400
+    },
+    {
+      "epoch": 0.20120482995897154,
+      "grad_norm": 12.664291381835938,
+      "learning_rate": 9.179872316867158e-06,
+      "loss": 2.6105,
+      "step": 771600
+    },
+    {
+      "epoch": 0.2012569825846737,
+      "grad_norm": 11.425148010253906,
+      "learning_rate": 9.179413526575472e-06,
+      "loss": 2.5416,
+      "step": 771800
+    },
+    {
+      "epoch": 0.20130913521037586,
+      "grad_norm": 10.56281566619873,
+      "learning_rate": 9.178954619464357e-06,
+      "loss": 2.5577,
+      "step": 772000
+    },
+    {
+      "epoch": 0.20136128783607804,
+      "grad_norm": 12.567456245422363,
+      "learning_rate": 9.178495595546641e-06,
+      "loss": 2.5976,
+      "step": 772200
+    },
+    {
+      "epoch": 0.2014134404617802,
+      "grad_norm": 11.59072208404541,
+      "learning_rate": 9.178036454835157e-06,
+      "loss": 2.5353,
+      "step": 772400
+    },
+    {
+      "epoch": 0.20146559308748238,
+      "grad_norm": 12.048215866088867,
+      "learning_rate": 9.177577197342738e-06,
+      "loss": 2.5556,
+      "step": 772600
+    },
+    {
+      "epoch": 0.20151774571318454,
+      "grad_norm": 13.227574348449707,
+      "learning_rate": 9.177117823082217e-06,
+      "loss": 2.5473,
+      "step": 772800
+    },
+    {
+      "epoch": 0.20156989833888672,
+      "grad_norm": 11.449342727661133,
+      "learning_rate": 9.176658332066438e-06,
+      "loss": 2.5772,
+      "step": 773000
+    },
+    {
+      "epoch": 0.20162205096458888,
+      "grad_norm": 13.140538215637207,
+      "learning_rate": 9.17619872430824e-06,
+      "loss": 2.5734,
+      "step": 773200
+    },
+    {
+      "epoch": 0.20167420359029106,
+      "grad_norm": 13.485116958618164,
+      "learning_rate": 9.175738999820476e-06,
+      "loss": 2.5461,
+      "step": 773400
+    },
+    {
+      "epoch": 0.20172635621599322,
+      "grad_norm": 14.482010841369629,
+      "learning_rate": 9.175279158615989e-06,
+      "loss": 2.5771,
+      "step": 773600
+    },
+    {
+      "epoch": 0.2017785088416954,
+      "grad_norm": 11.397805213928223,
+      "learning_rate": 9.174819200707636e-06,
+      "loss": 2.5517,
+      "step": 773800
+    },
+    {
+      "epoch": 0.20183066146739756,
+      "grad_norm": 13.39069938659668,
+      "learning_rate": 9.174359126108274e-06,
+      "loss": 2.5457,
+      "step": 774000
+    },
+    {
+      "epoch": 0.20188281409309974,
+      "grad_norm": 13.59749698638916,
+      "learning_rate": 9.17389893483076e-06,
+      "loss": 2.5409,
+      "step": 774200
+    },
+    {
+      "epoch": 0.2019349667188019,
+      "grad_norm": 12.491495132446289,
+      "learning_rate": 9.173438626887958e-06,
+      "loss": 2.5758,
+      "step": 774400
+    },
+    {
+      "epoch": 0.20198711934450408,
+      "grad_norm": 12.573298454284668,
+      "learning_rate": 9.172978202292735e-06,
+      "loss": 2.5665,
+      "step": 774600
+    },
+    {
+      "epoch": 0.20203927197020624,
+      "grad_norm": 12.443755149841309,
+      "learning_rate": 9.17251766105796e-06,
+      "loss": 2.603,
+      "step": 774800
+    },
+    {
+      "epoch": 0.20209142459590843,
+      "grad_norm": 11.910422325134277,
+      "learning_rate": 9.172057003196505e-06,
+      "loss": 2.5361,
+      "step": 775000
+    },
+    {
+      "epoch": 0.20214357722161058,
+      "grad_norm": 12.631829261779785,
+      "learning_rate": 9.171596228721245e-06,
+      "loss": 2.5566,
+      "step": 775200
+    },
+    {
+      "epoch": 0.20219572984731277,
+      "grad_norm": 13.858083724975586,
+      "learning_rate": 9.171135337645061e-06,
+      "loss": 2.5467,
+      "step": 775400
+    },
+    {
+      "epoch": 0.20224788247301492,
+      "grad_norm": 11.458602905273438,
+      "learning_rate": 9.170674329980835e-06,
+      "loss": 2.5821,
+      "step": 775600
+    },
+    {
+      "epoch": 0.2023000350987171,
+      "grad_norm": 12.20776081085205,
+      "learning_rate": 9.170213205741453e-06,
+      "loss": 2.5435,
+      "step": 775800
+    },
+    {
+      "epoch": 0.20235218772441926,
+      "grad_norm": 12.886215209960938,
+      "learning_rate": 9.169751964939802e-06,
+      "loss": 2.6111,
+      "step": 776000
+    },
+    {
+      "epoch": 0.20240434035012145,
+      "grad_norm": 13.515503883361816,
+      "learning_rate": 9.169290607588776e-06,
+      "loss": 2.5066,
+      "step": 776200
+    },
+    {
+      "epoch": 0.2024564929758236,
+      "grad_norm": 12.592185974121094,
+      "learning_rate": 9.168829133701273e-06,
+      "loss": 2.5507,
+      "step": 776400
+    },
+    {
+      "epoch": 0.2025086456015258,
+      "grad_norm": 13.320043563842773,
+      "learning_rate": 9.168367543290187e-06,
+      "loss": 2.5546,
+      "step": 776600
+    },
+    {
+      "epoch": 0.20256079822722795,
+      "grad_norm": 12.519879341125488,
+      "learning_rate": 9.16790583636842e-06,
+      "loss": 2.5799,
+      "step": 776800
+    },
+    {
+      "epoch": 0.20261295085293013,
+      "grad_norm": 12.188693046569824,
+      "learning_rate": 9.16744401294888e-06,
+      "loss": 2.5855,
+      "step": 777000
+    },
+    {
+      "epoch": 0.20266510347863229,
+      "grad_norm": 14.931890487670898,
+      "learning_rate": 9.166982073044475e-06,
+      "loss": 2.5712,
+      "step": 777200
+    },
+    {
+      "epoch": 0.20271725610433447,
+      "grad_norm": 13.907896995544434,
+      "learning_rate": 9.166520016668117e-06,
+      "loss": 2.5602,
+      "step": 777400
+    },
+    {
+      "epoch": 0.20276940873003663,
+      "grad_norm": 12.67017936706543,
+      "learning_rate": 9.16605784383272e-06,
+      "loss": 2.5684,
+      "step": 777600
+    },
+    {
+      "epoch": 0.20282156135573878,
+      "grad_norm": 12.602633476257324,
+      "learning_rate": 9.165595554551204e-06,
+      "loss": 2.5424,
+      "step": 777800
+    },
+    {
+      "epoch": 0.20287371398144097,
+      "grad_norm": 11.999449729919434,
+      "learning_rate": 9.165133148836487e-06,
+      "loss": 2.5635,
+      "step": 778000
+    },
+    {
+      "epoch": 0.20292586660714312,
+      "grad_norm": 15.479057312011719,
+      "learning_rate": 9.164670626701495e-06,
+      "loss": 2.5707,
+      "step": 778200
+    },
+    {
+      "epoch": 0.2029780192328453,
+      "grad_norm": 11.809393882751465,
+      "learning_rate": 9.16420798815916e-06,
+      "loss": 2.537,
+      "step": 778400
+    },
+    {
+      "epoch": 0.20303017185854746,
+      "grad_norm": 12.722270965576172,
+      "learning_rate": 9.163745233222407e-06,
+      "loss": 2.569,
+      "step": 778600
+    },
+    {
+      "epoch": 0.20308232448424965,
+      "grad_norm": 12.335058212280273,
+      "learning_rate": 9.163282361904176e-06,
+      "loss": 2.5594,
+      "step": 778800
+    },
+    {
+      "epoch": 0.2031344771099518,
+      "grad_norm": 11.937206268310547,
+      "learning_rate": 9.162819374217403e-06,
+      "loss": 2.5612,
+      "step": 779000
+    },
+    {
+      "epoch": 0.203186629735654,
+      "grad_norm": 10.777091026306152,
+      "learning_rate": 9.162356270175026e-06,
+      "loss": 2.5397,
+      "step": 779200
+    },
+    {
+      "epoch": 0.20323878236135615,
+      "grad_norm": 13.843822479248047,
+      "learning_rate": 9.161893049789995e-06,
+      "loss": 2.5632,
+      "step": 779400
+    },
+    {
+      "epoch": 0.20329093498705833,
+      "grad_norm": 11.743480682373047,
+      "learning_rate": 9.161429713075252e-06,
+      "loss": 2.587,
+      "step": 779600
+    },
+    {
+      "epoch": 0.2033430876127605,
+      "grad_norm": 10.867159843444824,
+      "learning_rate": 9.160966260043751e-06,
+      "loss": 2.5847,
+      "step": 779800
+    },
+    {
+      "epoch": 0.20339524023846267,
+      "grad_norm": 12.39773941040039,
+      "learning_rate": 9.160502690708447e-06,
+      "loss": 2.5626,
+      "step": 780000
+    },
+    {
+      "epoch": 0.20344739286416483,
+      "grad_norm": 13.51063060760498,
+      "learning_rate": 9.160039005082291e-06,
+      "loss": 2.5048,
+      "step": 780200
+    },
+    {
+      "epoch": 0.203499545489867,
+      "grad_norm": 12.987715721130371,
+      "learning_rate": 9.159575203178253e-06,
+      "loss": 2.5538,
+      "step": 780400
+    },
+    {
+      "epoch": 0.20355169811556917,
+      "grad_norm": 12.870197296142578,
+      "learning_rate": 9.159111285009289e-06,
+      "loss": 2.5578,
+      "step": 780600
+    },
+    {
+      "epoch": 0.20360385074127135,
+      "grad_norm": 10.9700927734375,
+      "learning_rate": 9.15864725058837e-06,
+      "loss": 2.5964,
+      "step": 780800
+    },
+    {
+      "epoch": 0.2036560033669735,
+      "grad_norm": 11.876527786254883,
+      "learning_rate": 9.158183099928465e-06,
+      "loss": 2.5587,
+      "step": 781000
+    },
+    {
+      "epoch": 0.2037081559926757,
+      "grad_norm": 12.678916931152344,
+      "learning_rate": 9.157718833042548e-06,
+      "loss": 2.57,
+      "step": 781200
+    },
+    {
+      "epoch": 0.20376030861837785,
+      "grad_norm": 11.316527366638184,
+      "learning_rate": 9.157254449943594e-06,
+      "loss": 2.531,
+      "step": 781400
+    },
+    {
+      "epoch": 0.20381246124408003,
+      "grad_norm": 11.957809448242188,
+      "learning_rate": 9.156789950644587e-06,
+      "loss": 2.5381,
+      "step": 781600
+    },
+    {
+      "epoch": 0.2038646138697822,
+      "grad_norm": 12.463024139404297,
+      "learning_rate": 9.156325335158507e-06,
+      "loss": 2.5681,
+      "step": 781800
+    },
+    {
+      "epoch": 0.20391676649548438,
+      "grad_norm": 12.041062355041504,
+      "learning_rate": 9.155860603498341e-06,
+      "loss": 2.5684,
+      "step": 782000
+    },
+    {
+      "epoch": 0.20396891912118653,
+      "grad_norm": 11.98395824432373,
+      "learning_rate": 9.15539575567708e-06,
+      "loss": 2.5603,
+      "step": 782200
+    },
+    {
+      "epoch": 0.20402107174688872,
+      "grad_norm": 12.181907653808594,
+      "learning_rate": 9.154930791707714e-06,
+      "loss": 2.5406,
+      "step": 782400
+    },
+    {
+      "epoch": 0.20407322437259087,
+      "grad_norm": 12.69912338256836,
+      "learning_rate": 9.154465711603244e-06,
+      "loss": 2.5441,
+      "step": 782600
+    },
+    {
+      "epoch": 0.20412537699829306,
+      "grad_norm": 11.131607055664062,
+      "learning_rate": 9.154000515376667e-06,
+      "loss": 2.5484,
+      "step": 782800
+    },
+    {
+      "epoch": 0.2041775296239952,
+      "grad_norm": 11.983840942382812,
+      "learning_rate": 9.153535203040986e-06,
+      "loss": 2.5317,
+      "step": 783000
+    },
+    {
+      "epoch": 0.2042296822496974,
+      "grad_norm": 13.614217758178711,
+      "learning_rate": 9.153069774609205e-06,
+      "loss": 2.5382,
+      "step": 783200
+    },
+    {
+      "epoch": 0.20428183487539955,
+      "grad_norm": 13.162174224853516,
+      "learning_rate": 9.152604230094337e-06,
+      "loss": 2.5865,
+      "step": 783400
+    },
+    {
+      "epoch": 0.2043339875011017,
+      "grad_norm": 12.635634422302246,
+      "learning_rate": 9.152138569509393e-06,
+      "loss": 2.549,
+      "step": 783600
+    },
+    {
+      "epoch": 0.2043861401268039,
+      "grad_norm": 9.673964500427246,
+      "learning_rate": 9.151672792867387e-06,
+      "loss": 2.5795,
+      "step": 783800
+    },
+    {
+      "epoch": 0.20443829275250605,
+      "grad_norm": 12.545588493347168,
+      "learning_rate": 9.15120690018134e-06,
+      "loss": 2.546,
+      "step": 784000
+    },
+    {
+      "epoch": 0.20449044537820824,
+      "grad_norm": 11.919151306152344,
+      "learning_rate": 9.150740891464272e-06,
+      "loss": 2.5573,
+      "step": 784200
+    },
+    {
+      "epoch": 0.2045425980039104,
+      "grad_norm": 14.723599433898926,
+      "learning_rate": 9.150274766729213e-06,
+      "loss": 2.5681,
+      "step": 784400
+    },
+    {
+      "epoch": 0.20459475062961258,
+      "grad_norm": 11.212860107421875,
+      "learning_rate": 9.149808525989185e-06,
+      "loss": 2.5524,
+      "step": 784600
+    },
+    {
+      "epoch": 0.20464690325531473,
+      "grad_norm": 13.626235961914062,
+      "learning_rate": 9.149342169257228e-06,
+      "loss": 2.6008,
+      "step": 784800
+    },
+    {
+      "epoch": 0.20469905588101692,
+      "grad_norm": 13.193553924560547,
+      "learning_rate": 9.148875696546372e-06,
+      "loss": 2.54,
+      "step": 785000
+    },
+    {
+      "epoch": 0.20475120850671907,
+      "grad_norm": 12.401244163513184,
+      "learning_rate": 9.148409107869654e-06,
+      "loss": 2.5822,
+      "step": 785200
+    },
+    {
+      "epoch": 0.20480336113242126,
+      "grad_norm": 13.341384887695312,
+      "learning_rate": 9.14794240324012e-06,
+      "loss": 2.581,
+      "step": 785400
+    },
+    {
+      "epoch": 0.20485551375812341,
+      "grad_norm": 10.86294937133789,
+      "learning_rate": 9.147475582670813e-06,
+      "loss": 2.558,
+      "step": 785600
+    },
+    {
+      "epoch": 0.2049076663838256,
+      "grad_norm": 11.97840690612793,
+      "learning_rate": 9.14700864617478e-06,
+      "loss": 2.569,
+      "step": 785800
+    },
+    {
+      "epoch": 0.20495981900952776,
+      "grad_norm": 12.532195091247559,
+      "learning_rate": 9.146541593765075e-06,
+      "loss": 2.5214,
+      "step": 786000
+    },
+    {
+      "epoch": 0.20501197163522994,
+      "grad_norm": 13.745171546936035,
+      "learning_rate": 9.14607442545475e-06,
+      "loss": 2.5511,
+      "step": 786200
+    },
+    {
+      "epoch": 0.2050641242609321,
+      "grad_norm": 12.68714714050293,
+      "learning_rate": 9.145607141256864e-06,
+      "loss": 2.5863,
+      "step": 786400
+    },
+    {
+      "epoch": 0.20511627688663428,
+      "grad_norm": 12.79076862335205,
+      "learning_rate": 9.145139741184481e-06,
+      "loss": 2.5629,
+      "step": 786600
+    },
+    {
+      "epoch": 0.20516842951233644,
+      "grad_norm": 13.012794494628906,
+      "learning_rate": 9.144672225250661e-06,
+      "loss": 2.582,
+      "step": 786800
+    },
+    {
+      "epoch": 0.20522058213803862,
+      "grad_norm": 12.885845184326172,
+      "learning_rate": 9.144204593468473e-06,
+      "loss": 2.5376,
+      "step": 787000
+    },
+    {
+      "epoch": 0.20527273476374078,
+      "grad_norm": 11.415619850158691,
+      "learning_rate": 9.14373684585099e-06,
+      "loss": 2.5376,
+      "step": 787200
+    },
+    {
+      "epoch": 0.20532488738944296,
+      "grad_norm": 10.882523536682129,
+      "learning_rate": 9.143268982411282e-06,
+      "loss": 2.5388,
+      "step": 787400
+    },
+    {
+      "epoch": 0.20537704001514512,
+      "grad_norm": 13.310264587402344,
+      "learning_rate": 9.142801003162429e-06,
+      "loss": 2.5681,
+      "step": 787600
+    },
+    {
+      "epoch": 0.2054291926408473,
+      "grad_norm": 10.162239074707031,
+      "learning_rate": 9.142332908117512e-06,
+      "loss": 2.5519,
+      "step": 787800
+    },
+    {
+      "epoch": 0.20548134526654946,
+      "grad_norm": 10.431777000427246,
+      "learning_rate": 9.141864697289612e-06,
+      "loss": 2.5372,
+      "step": 788000
+    },
+    {
+      "epoch": 0.20553349789225164,
+      "grad_norm": 11.729969024658203,
+      "learning_rate": 9.141396370691822e-06,
+      "loss": 2.5446,
+      "step": 788200
+    },
+    {
+      "epoch": 0.2055856505179538,
+      "grad_norm": 13.751687049865723,
+      "learning_rate": 9.140927928337224e-06,
+      "loss": 2.5337,
+      "step": 788400
+    },
+    {
+      "epoch": 0.20563780314365598,
+      "grad_norm": 11.931695938110352,
+      "learning_rate": 9.140459370238919e-06,
+      "loss": 2.5327,
+      "step": 788600
+    },
+    {
+      "epoch": 0.20568995576935814,
+      "grad_norm": 12.925618171691895,
+      "learning_rate": 9.139990696409998e-06,
+      "loss": 2.5283,
+      "step": 788800
+    },
+    {
+      "epoch": 0.20574210839506032,
+      "grad_norm": 11.938955307006836,
+      "learning_rate": 9.139521906863564e-06,
+      "loss": 2.545,
+      "step": 789000
+    },
+    {
+      "epoch": 0.20579426102076248,
+      "grad_norm": 12.688498497009277,
+      "learning_rate": 9.13905300161272e-06,
+      "loss": 2.5699,
+      "step": 789200
+    },
+    {
+      "epoch": 0.20584641364646467,
+      "grad_norm": 12.349008560180664,
+      "learning_rate": 9.138583980670573e-06,
+      "loss": 2.5775,
+      "step": 789400
+    },
+    {
+      "epoch": 0.20589856627216682,
+      "grad_norm": 13.858742713928223,
+      "learning_rate": 9.13811484405023e-06,
+      "loss": 2.5536,
+      "step": 789600
+    },
+    {
+      "epoch": 0.20595071889786898,
+      "grad_norm": 12.965885162353516,
+      "learning_rate": 9.137645591764807e-06,
+      "loss": 2.5364,
+      "step": 789800
+    },
+    {
+      "epoch": 0.20600287152357116,
+      "grad_norm": 12.98017692565918,
+      "learning_rate": 9.137176223827417e-06,
+      "loss": 2.5487,
+      "step": 790000
+    },
+    {
+      "epoch": 0.20605502414927332,
+      "grad_norm": 12.841651916503906,
+      "learning_rate": 9.136706740251185e-06,
+      "loss": 2.5751,
+      "step": 790200
+    },
+    {
+      "epoch": 0.2061071767749755,
+      "grad_norm": 11.136397361755371,
+      "learning_rate": 9.136237141049226e-06,
+      "loss": 2.5773,
+      "step": 790400
+    },
+    {
+      "epoch": 0.20615932940067766,
+      "grad_norm": 13.198196411132812,
+      "learning_rate": 9.135767426234671e-06,
+      "loss": 2.529,
+      "step": 790600
+    },
+    {
+      "epoch": 0.20621148202637984,
+      "grad_norm": 13.684469223022461,
+      "learning_rate": 9.135297595820646e-06,
+      "loss": 2.5633,
+      "step": 790800
+    },
+    {
+      "epoch": 0.206263634652082,
+      "grad_norm": 12.033394813537598,
+      "learning_rate": 9.134827649820289e-06,
+      "loss": 2.602,
+      "step": 791000
+    },
+    {
+      "epoch": 0.20631578727778419,
+      "grad_norm": 10.937596321105957,
+      "learning_rate": 9.134357588246729e-06,
+      "loss": 2.5496,
+      "step": 791200
+    },
+    {
+      "epoch": 0.20636793990348634,
+      "grad_norm": 11.299455642700195,
+      "learning_rate": 9.133887411113108e-06,
+      "loss": 2.5665,
+      "step": 791400
+    },
+    {
+      "epoch": 0.20642009252918853,
+      "grad_norm": 13.532719612121582,
+      "learning_rate": 9.133417118432566e-06,
+      "loss": 2.5522,
+      "step": 791600
+    },
+    {
+      "epoch": 0.20647224515489068,
+      "grad_norm": 12.678144454956055,
+      "learning_rate": 9.13294671021825e-06,
+      "loss": 2.5615,
+      "step": 791800
+    },
+    {
+      "epoch": 0.20652439778059287,
+      "grad_norm": 12.046524047851562,
+      "learning_rate": 9.13247618648331e-06,
+      "loss": 2.562,
+      "step": 792000
+    },
+    {
+      "epoch": 0.20657655040629502,
+      "grad_norm": 12.832523345947266,
+      "learning_rate": 9.132005547240893e-06,
+      "loss": 2.5695,
+      "step": 792200
+    },
+    {
+      "epoch": 0.2066287030319972,
+      "grad_norm": 12.524669647216797,
+      "learning_rate": 9.131534792504158e-06,
+      "loss": 2.5554,
+      "step": 792400
+    },
+    {
+      "epoch": 0.20668085565769936,
+      "grad_norm": 11.63914966583252,
+      "learning_rate": 9.131063922286261e-06,
+      "loss": 2.5182,
+      "step": 792600
+    },
+    {
+      "epoch": 0.20673300828340155,
+      "grad_norm": 12.573324203491211,
+      "learning_rate": 9.130592936600364e-06,
+      "loss": 2.589,
+      "step": 792800
+    },
+    {
+      "epoch": 0.2067851609091037,
+      "grad_norm": 11.700994491577148,
+      "learning_rate": 9.130121835459633e-06,
+      "loss": 2.6005,
+      "step": 793000
+    },
+    {
+      "epoch": 0.2068373135348059,
+      "grad_norm": 10.683900833129883,
+      "learning_rate": 9.129650618877233e-06,
+      "loss": 2.5726,
+      "step": 793200
+    },
+    {
+      "epoch": 0.20688946616050805,
+      "grad_norm": 12.077662467956543,
+      "learning_rate": 9.129179286866337e-06,
+      "loss": 2.5581,
+      "step": 793400
+    },
+    {
+      "epoch": 0.20694161878621023,
+      "grad_norm": 12.618571281433105,
+      "learning_rate": 9.128707839440119e-06,
+      "loss": 2.5401,
+      "step": 793600
+    },
+    {
+      "epoch": 0.2069937714119124,
+      "grad_norm": 13.423776626586914,
+      "learning_rate": 9.128236276611757e-06,
+      "loss": 2.5562,
+      "step": 793800
+    },
+    {
+      "epoch": 0.20704592403761457,
+      "grad_norm": 11.433748245239258,
+      "learning_rate": 9.127764598394429e-06,
+      "loss": 2.5461,
+      "step": 794000
+    },
+    {
+      "epoch": 0.20709807666331673,
+      "grad_norm": 13.008195877075195,
+      "learning_rate": 9.127292804801322e-06,
+      "loss": 2.5663,
+      "step": 794200
+    },
+    {
+      "epoch": 0.2071502292890189,
+      "grad_norm": 12.282379150390625,
+      "learning_rate": 9.126820895845623e-06,
+      "loss": 2.5785,
+      "step": 794400
+    },
+    {
+      "epoch": 0.20720238191472107,
+      "grad_norm": 12.038056373596191,
+      "learning_rate": 9.12634887154052e-06,
+      "loss": 2.5305,
+      "step": 794600
+    },
+    {
+      "epoch": 0.20725453454042325,
+      "grad_norm": 10.847259521484375,
+      "learning_rate": 9.125876731899209e-06,
+      "loss": 2.5142,
+      "step": 794800
+    },
+    {
+      "epoch": 0.2073066871661254,
+      "grad_norm": 12.055601119995117,
+      "learning_rate": 9.125404476934888e-06,
+      "loss": 2.556,
+      "step": 795000
+    },
+    {
+      "epoch": 0.2073588397918276,
+      "grad_norm": 14.068562507629395,
+      "learning_rate": 9.124932106660752e-06,
+      "loss": 2.5441,
+      "step": 795200
+    },
+    {
+      "epoch": 0.20741099241752975,
+      "grad_norm": 12.50983715057373,
+      "learning_rate": 9.124459621090009e-06,
+      "loss": 2.5705,
+      "step": 795400
+    },
+    {
+      "epoch": 0.2074631450432319,
+      "grad_norm": 12.685457229614258,
+      "learning_rate": 9.123987020235863e-06,
+      "loss": 2.5693,
+      "step": 795600
+    },
+    {
+      "epoch": 0.2075152976689341,
+      "grad_norm": 12.426636695861816,
+      "learning_rate": 9.123514304111525e-06,
+      "loss": 2.555,
+      "step": 795800
+    },
+    {
+      "epoch": 0.20756745029463625,
+      "grad_norm": 11.003486633300781,
+      "learning_rate": 9.123041472730207e-06,
+      "loss": 2.5363,
+      "step": 796000
+    },
+    {
+      "epoch": 0.20761960292033843,
+      "grad_norm": 12.062921524047852,
+      "learning_rate": 9.122568526105127e-06,
+      "loss": 2.5494,
+      "step": 796200
+    },
+    {
+      "epoch": 0.2076717555460406,
+      "grad_norm": 9.995904922485352,
+      "learning_rate": 9.122095464249504e-06,
+      "loss": 2.5621,
+      "step": 796400
+    },
+    {
+      "epoch": 0.20772390817174277,
+      "grad_norm": 12.198345184326172,
+      "learning_rate": 9.121622287176557e-06,
+      "loss": 2.5531,
+      "step": 796600
+    },
+    {
+      "epoch": 0.20777606079744493,
+      "grad_norm": 11.961278915405273,
+      "learning_rate": 9.121148994899517e-06,
+      "loss": 2.5245,
+      "step": 796800
+    },
+    {
+      "epoch": 0.2078282134231471,
+      "grad_norm": 11.721890449523926,
+      "learning_rate": 9.12067558743161e-06,
+      "loss": 2.5773,
+      "step": 797000
+    },
+    {
+      "epoch": 0.20788036604884927,
+      "grad_norm": 11.148172378540039,
+      "learning_rate": 9.120202064786067e-06,
+      "loss": 2.5425,
+      "step": 797200
+    },
+    {
+      "epoch": 0.20793251867455145,
+      "grad_norm": 12.131102561950684,
+      "learning_rate": 9.119728426976129e-06,
+      "loss": 2.5342,
+      "step": 797400
+    },
+    {
+      "epoch": 0.2079846713002536,
+      "grad_norm": 12.33278751373291,
+      "learning_rate": 9.11925467401503e-06,
+      "loss": 2.5778,
+      "step": 797600
+    },
+    {
+      "epoch": 0.2080368239259558,
+      "grad_norm": 13.5529203414917,
+      "learning_rate": 9.118780805916011e-06,
+      "loss": 2.5409,
+      "step": 797800
+    },
+    {
+      "epoch": 0.20808897655165795,
+      "grad_norm": 13.351973533630371,
+      "learning_rate": 9.118306822692322e-06,
+      "loss": 2.5533,
+      "step": 798000
+    },
+    {
+      "epoch": 0.20814112917736013,
+      "grad_norm": 11.997188568115234,
+      "learning_rate": 9.117832724357208e-06,
+      "loss": 2.5669,
+      "step": 798200
+    },
+    {
+      "epoch": 0.2081932818030623,
+      "grad_norm": 12.028074264526367,
+      "learning_rate": 9.117358510923921e-06,
+      "loss": 2.5191,
+      "step": 798400
+    },
+    {
+      "epoch": 0.20824543442876448,
+      "grad_norm": 10.646809577941895,
+      "learning_rate": 9.116884182405718e-06,
+      "loss": 2.5348,
+      "step": 798600
+    },
+    {
+      "epoch": 0.20829758705446663,
+      "grad_norm": 10.890563011169434,
+      "learning_rate": 9.116409738815853e-06,
+      "loss": 2.5531,
+      "step": 798800
+    },
+    {
+      "epoch": 0.20834973968016882,
+      "grad_norm": 11.447530746459961,
+      "learning_rate": 9.11593518016759e-06,
+      "loss": 2.5651,
+      "step": 799000
+    },
+    {
+      "epoch": 0.20840189230587097,
+      "grad_norm": 13.491917610168457,
+      "learning_rate": 9.115460506474193e-06,
+      "loss": 2.5481,
+      "step": 799200
+    },
+    {
+      "epoch": 0.20845404493157316,
+      "grad_norm": 12.284280776977539,
+      "learning_rate": 9.114985717748928e-06,
+      "loss": 2.5543,
+      "step": 799400
+    },
+    {
+      "epoch": 0.2085061975572753,
+      "grad_norm": 11.280057907104492,
+      "learning_rate": 9.11451081400507e-06,
+      "loss": 2.5347,
+      "step": 799600
+    },
+    {
+      "epoch": 0.2085583501829775,
+      "grad_norm": 13.034576416015625,
+      "learning_rate": 9.114035795255888e-06,
+      "loss": 2.5667,
+      "step": 799800
+    },
+    {
+      "epoch": 0.20861050280867965,
+      "grad_norm": 10.916426658630371,
+      "learning_rate": 9.113560661514664e-06,
+      "loss": 2.5311,
+      "step": 800000
+    },
+    {
+      "epoch": 0.20866265543438184,
+      "grad_norm": 12.369806289672852,
+      "learning_rate": 9.113085412794676e-06,
+      "loss": 2.5332,
+      "step": 800200
+    },
+    {
+      "epoch": 0.208714808060084,
+      "grad_norm": 12.768467903137207,
+      "learning_rate": 9.112610049109207e-06,
+      "loss": 2.5693,
+      "step": 800400
+    },
+    {
+      "epoch": 0.20876696068578618,
+      "grad_norm": 13.05500316619873,
+      "learning_rate": 9.112134570471545e-06,
+      "loss": 2.5134,
+      "step": 800600
+    },
+    {
+      "epoch": 0.20881911331148834,
+      "grad_norm": 11.387818336486816,
+      "learning_rate": 9.111658976894982e-06,
+      "loss": 2.5834,
+      "step": 800800
+    },
+    {
+      "epoch": 0.20887126593719052,
+      "grad_norm": 10.839855194091797,
+      "learning_rate": 9.11118326839281e-06,
+      "loss": 2.5247,
+      "step": 801000
+    },
+    {
+      "epoch": 0.20892341856289268,
+      "grad_norm": 12.861638069152832,
+      "learning_rate": 9.110707444978322e-06,
+      "loss": 2.5669,
+      "step": 801200
+    },
+    {
+      "epoch": 0.20897557118859483,
+      "grad_norm": 11.404274940490723,
+      "learning_rate": 9.110231506664824e-06,
+      "loss": 2.5437,
+      "step": 801400
+    },
+    {
+      "epoch": 0.20902772381429702,
+      "grad_norm": 13.15756607055664,
+      "learning_rate": 9.109755453465615e-06,
+      "loss": 2.53,
+      "step": 801600
+    },
+    {
+      "epoch": 0.20907987643999917,
+      "grad_norm": 13.401494026184082,
+      "learning_rate": 9.109279285394003e-06,
+      "loss": 2.5359,
+      "step": 801800
+    },
+    {
+      "epoch": 0.20913202906570136,
+      "grad_norm": 13.968293190002441,
+      "learning_rate": 9.108803002463295e-06,
+      "loss": 2.5573,
+      "step": 802000
+    },
+    {
+      "epoch": 0.20918418169140351,
+      "grad_norm": 13.148781776428223,
+      "learning_rate": 9.108326604686807e-06,
+      "loss": 2.56,
+      "step": 802200
+    },
+    {
+      "epoch": 0.2092363343171057,
+      "grad_norm": 13.43651294708252,
+      "learning_rate": 9.107850092077853e-06,
+      "loss": 2.5353,
+      "step": 802400
+    },
+    {
+      "epoch": 0.20928848694280786,
+      "grad_norm": 12.484440803527832,
+      "learning_rate": 9.107373464649754e-06,
+      "loss": 2.5772,
+      "step": 802600
+    },
+    {
+      "epoch": 0.20934063956851004,
+      "grad_norm": 13.605812072753906,
+      "learning_rate": 9.106896722415829e-06,
+      "loss": 2.5525,
+      "step": 802800
+    },
+    {
+      "epoch": 0.2093927921942122,
+      "grad_norm": 13.259238243103027,
+      "learning_rate": 9.106419865389405e-06,
+      "loss": 2.596,
+      "step": 803000
+    },
+    {
+      "epoch": 0.20944494481991438,
+      "grad_norm": 11.530973434448242,
+      "learning_rate": 9.10594289358381e-06,
+      "loss": 2.5526,
+      "step": 803200
+    },
+    {
+      "epoch": 0.20949709744561654,
+      "grad_norm": 12.25853157043457,
+      "learning_rate": 9.105465807012379e-06,
+      "loss": 2.5761,
+      "step": 803400
+    },
+    {
+      "epoch": 0.20954925007131872,
+      "grad_norm": 10.755802154541016,
+      "learning_rate": 9.104988605688445e-06,
+      "loss": 2.5424,
+      "step": 803600
+    },
+    {
+      "epoch": 0.20960140269702088,
+      "grad_norm": 13.130935668945312,
+      "learning_rate": 9.104511289625345e-06,
+      "loss": 2.5453,
+      "step": 803800
+    },
+    {
+      "epoch": 0.20965355532272306,
+      "grad_norm": 13.585618019104004,
+      "learning_rate": 9.10403385883642e-06,
+      "loss": 2.4936,
+      "step": 804000
+    },
+    {
+      "epoch": 0.20970570794842522,
+      "grad_norm": 13.895092964172363,
+      "learning_rate": 9.103556313335019e-06,
+      "loss": 2.5333,
+      "step": 804200
+    },
+    {
+      "epoch": 0.2097578605741274,
+      "grad_norm": 12.45972728729248,
+      "learning_rate": 9.103078653134487e-06,
+      "loss": 2.5459,
+      "step": 804400
+    },
+    {
+      "epoch": 0.20981001319982956,
+      "grad_norm": 13.445938110351562,
+      "learning_rate": 9.102600878248176e-06,
+      "loss": 2.5226,
+      "step": 804600
+    },
+    {
+      "epoch": 0.20986216582553174,
+      "grad_norm": 12.260578155517578,
+      "learning_rate": 9.102122988689439e-06,
+      "loss": 2.5466,
+      "step": 804800
+    },
+    {
+      "epoch": 0.2099143184512339,
+      "grad_norm": 13.503302574157715,
+      "learning_rate": 9.101644984471636e-06,
+      "loss": 2.5485,
+      "step": 805000
+    },
+    {
+      "epoch": 0.20996647107693608,
+      "grad_norm": 11.830412864685059,
+      "learning_rate": 9.101166865608125e-06,
+      "loss": 2.5331,
+      "step": 805200
+    },
+    {
+      "epoch": 0.21001862370263824,
+      "grad_norm": 11.57419204711914,
+      "learning_rate": 9.100688632112272e-06,
+      "loss": 2.5551,
+      "step": 805400
+    },
+    {
+      "epoch": 0.21007077632834043,
+      "grad_norm": 13.763896942138672,
+      "learning_rate": 9.100210283997442e-06,
+      "loss": 2.5441,
+      "step": 805600
+    },
+    {
+      "epoch": 0.21012292895404258,
+      "grad_norm": 11.554506301879883,
+      "learning_rate": 9.099731821277008e-06,
+      "loss": 2.571,
+      "step": 805800
+    },
+    {
+      "epoch": 0.21017508157974477,
+      "grad_norm": 12.712745666503906,
+      "learning_rate": 9.099253243964343e-06,
+      "loss": 2.5449,
+      "step": 806000
+    },
+    {
+      "epoch": 0.21022723420544692,
+      "grad_norm": 12.259404182434082,
+      "learning_rate": 9.098774552072823e-06,
+      "loss": 2.5364,
+      "step": 806200
+    },
+    {
+      "epoch": 0.2102793868311491,
+      "grad_norm": 14.388871192932129,
+      "learning_rate": 9.098295745615826e-06,
+      "loss": 2.5435,
+      "step": 806400
+    },
+    {
+      "epoch": 0.21033153945685126,
+      "grad_norm": 11.50523853302002,
+      "learning_rate": 9.09781682460674e-06,
+      "loss": 2.5728,
+      "step": 806600
+    },
+    {
+      "epoch": 0.21038369208255345,
+      "grad_norm": 11.484689712524414,
+      "learning_rate": 9.097337789058947e-06,
+      "loss": 2.5825,
+      "step": 806800
+    },
+    {
+      "epoch": 0.2104358447082556,
+      "grad_norm": 12.460712432861328,
+      "learning_rate": 9.096858638985839e-06,
+      "loss": 2.5243,
+      "step": 807000
+    },
+    {
+      "epoch": 0.21048799733395776,
+      "grad_norm": 13.950214385986328,
+      "learning_rate": 9.09637937440081e-06,
+      "loss": 2.5351,
+      "step": 807200
+    },
+    {
+      "epoch": 0.21054014995965994,
+      "grad_norm": 12.65377140045166,
+      "learning_rate": 9.095899995317252e-06,
+      "loss": 2.5472,
+      "step": 807400
+    },
+    {
+      "epoch": 0.2105923025853621,
+      "grad_norm": 11.964714050292969,
+      "learning_rate": 9.095420501748568e-06,
+      "loss": 2.5647,
+      "step": 807600
+    },
+    {
+      "epoch": 0.21064445521106429,
+      "grad_norm": 12.092817306518555,
+      "learning_rate": 9.094940893708157e-06,
+      "loss": 2.5691,
+      "step": 807800
+    },
+    {
+      "epoch": 0.21069660783676644,
+      "grad_norm": 13.357704162597656,
+      "learning_rate": 9.094461171209427e-06,
+      "loss": 2.5075,
+      "step": 808000
+    },
+    {
+      "epoch": 0.21074876046246863,
+      "grad_norm": 11.30611801147461,
+      "learning_rate": 9.093981334265787e-06,
+      "loss": 2.509,
+      "step": 808200
+    },
+    {
+      "epoch": 0.21080091308817078,
+      "grad_norm": 12.069721221923828,
+      "learning_rate": 9.093501382890647e-06,
+      "loss": 2.5223,
+      "step": 808400
+    },
+    {
+      "epoch": 0.21085306571387297,
+      "grad_norm": 11.740452766418457,
+      "learning_rate": 9.093021317097424e-06,
+      "loss": 2.5337,
+      "step": 808600
+    },
+    {
+      "epoch": 0.21090521833957512,
+      "grad_norm": 12.168525695800781,
+      "learning_rate": 9.092541136899535e-06,
+      "loss": 2.5598,
+      "step": 808800
+    },
+    {
+      "epoch": 0.2109573709652773,
+      "grad_norm": 13.350499153137207,
+      "learning_rate": 9.092060842310404e-06,
+      "loss": 2.4994,
+      "step": 809000
+    },
+    {
+      "epoch": 0.21100952359097946,
+      "grad_norm": 12.407944679260254,
+      "learning_rate": 9.091580433343452e-06,
+      "loss": 2.4965,
+      "step": 809200
+    },
+    {
+      "epoch": 0.21106167621668165,
+      "grad_norm": 14.529518127441406,
+      "learning_rate": 9.09109991001211e-06,
+      "loss": 2.5579,
+      "step": 809400
+    },
+    {
+      "epoch": 0.2111138288423838,
+      "grad_norm": 12.541584968566895,
+      "learning_rate": 9.09061927232981e-06,
+      "loss": 2.5532,
+      "step": 809600
+    },
+    {
+      "epoch": 0.211165981468086,
+      "grad_norm": 14.252954483032227,
+      "learning_rate": 9.090138520309983e-06,
+      "loss": 2.5472,
+      "step": 809800
+    },
+    {
+      "epoch": 0.21121813409378815,
+      "grad_norm": 13.405433654785156,
+      "learning_rate": 9.089657653966069e-06,
+      "loss": 2.5333,
+      "step": 810000
+    },
+    {
+      "epoch": 0.21127028671949033,
+      "grad_norm": 10.493156433105469,
+      "learning_rate": 9.089176673311506e-06,
+      "loss": 2.5312,
+      "step": 810200
+    },
+    {
+      "epoch": 0.2113224393451925,
+      "grad_norm": 12.578083992004395,
+      "learning_rate": 9.088695578359742e-06,
+      "loss": 2.5336,
+      "step": 810400
+    },
+    {
+      "epoch": 0.21137459197089467,
+      "grad_norm": 12.994580268859863,
+      "learning_rate": 9.088214369124221e-06,
+      "loss": 2.5489,
+      "step": 810600
+    },
+    {
+      "epoch": 0.21142674459659683,
+      "grad_norm": 11.961739540100098,
+      "learning_rate": 9.087733045618396e-06,
+      "loss": 2.5717,
+      "step": 810800
+    },
+    {
+      "epoch": 0.211478897222299,
+      "grad_norm": 12.124736785888672,
+      "learning_rate": 9.087251607855718e-06,
+      "loss": 2.5273,
+      "step": 811000
+    },
+    {
+      "epoch": 0.21153104984800117,
+      "grad_norm": 14.014177322387695,
+      "learning_rate": 9.086770055849645e-06,
+      "loss": 2.5209,
+      "step": 811200
+    },
+    {
+      "epoch": 0.21158320247370335,
+      "grad_norm": 12.617166519165039,
+      "learning_rate": 9.086288389613636e-06,
+      "loss": 2.5425,
+      "step": 811400
+    },
+    {
+      "epoch": 0.2116353550994055,
+      "grad_norm": 12.897732734680176,
+      "learning_rate": 9.085806609161156e-06,
+      "loss": 2.5501,
+      "step": 811600
+    },
+    {
+      "epoch": 0.2116875077251077,
+      "grad_norm": 14.856685638427734,
+      "learning_rate": 9.085324714505669e-06,
+      "loss": 2.5208,
+      "step": 811800
+    },
+    {
+      "epoch": 0.21173966035080985,
+      "grad_norm": 11.253473281860352,
+      "learning_rate": 9.084842705660646e-06,
+      "loss": 2.5396,
+      "step": 812000
+    },
+    {
+      "epoch": 0.21179181297651203,
+      "grad_norm": 13.86609172821045,
+      "learning_rate": 9.08436058263956e-06,
+      "loss": 2.5441,
+      "step": 812200
+    },
+    {
+      "epoch": 0.2118439656022142,
+      "grad_norm": 11.815281867980957,
+      "learning_rate": 9.083878345455885e-06,
+      "loss": 2.543,
+      "step": 812400
+    },
+    {
+      "epoch": 0.21189611822791637,
+      "grad_norm": 12.452479362487793,
+      "learning_rate": 9.083395994123103e-06,
+      "loss": 2.5331,
+      "step": 812600
+    },
+    {
+      "epoch": 0.21194827085361853,
+      "grad_norm": 13.19066333770752,
+      "learning_rate": 9.082913528654694e-06,
+      "loss": 2.5344,
+      "step": 812800
+    },
+    {
+      "epoch": 0.21200042347932072,
+      "grad_norm": 11.936997413635254,
+      "learning_rate": 9.082430949064144e-06,
+      "loss": 2.5393,
+      "step": 813000
+    },
+    {
+      "epoch": 0.21205257610502287,
+      "grad_norm": 13.574031829833984,
+      "learning_rate": 9.081948255364941e-06,
+      "loss": 2.5316,
+      "step": 813200
+    },
+    {
+      "epoch": 0.21210472873072503,
+      "grad_norm": 11.525445938110352,
+      "learning_rate": 9.081465447570578e-06,
+      "loss": 2.5151,
+      "step": 813400
+    },
+    {
+      "epoch": 0.2121568813564272,
+      "grad_norm": 11.842822074890137,
+      "learning_rate": 9.08098252569455e-06,
+      "loss": 2.6006,
+      "step": 813600
+    },
+    {
+      "epoch": 0.21220903398212937,
+      "grad_norm": 12.733033180236816,
+      "learning_rate": 9.080499489750357e-06,
+      "loss": 2.569,
+      "step": 813800
+    },
+    {
+      "epoch": 0.21226118660783155,
+      "grad_norm": 13.042905807495117,
+      "learning_rate": 9.080016339751495e-06,
+      "loss": 2.5706,
+      "step": 814000
+    },
+    {
+      "epoch": 0.2123133392335337,
+      "grad_norm": 12.87771224975586,
+      "learning_rate": 9.079533075711473e-06,
+      "loss": 2.5289,
+      "step": 814200
+    },
+    {
+      "epoch": 0.2123654918592359,
+      "grad_norm": 12.961420059204102,
+      "learning_rate": 9.079049697643798e-06,
+      "loss": 2.58,
+      "step": 814400
+    },
+    {
+      "epoch": 0.21241764448493805,
+      "grad_norm": 12.723822593688965,
+      "learning_rate": 9.078566205561981e-06,
+      "loss": 2.542,
+      "step": 814600
+    },
+    {
+      "epoch": 0.21246979711064023,
+      "grad_norm": 12.822134971618652,
+      "learning_rate": 9.078082599479535e-06,
+      "loss": 2.5328,
+      "step": 814800
+    },
+    {
+      "epoch": 0.2125219497363424,
+      "grad_norm": 10.794196128845215,
+      "learning_rate": 9.077598879409977e-06,
+      "loss": 2.5756,
+      "step": 815000
+    },
+    {
+      "epoch": 0.21257410236204458,
+      "grad_norm": 11.451395034790039,
+      "learning_rate": 9.07711504536683e-06,
+      "loss": 2.5266,
+      "step": 815200
+    },
+    {
+      "epoch": 0.21262625498774673,
+      "grad_norm": 13.746084213256836,
+      "learning_rate": 9.076631097363615e-06,
+      "loss": 2.53,
+      "step": 815400
+    },
+    {
+      "epoch": 0.21267840761344892,
+      "grad_norm": 13.913445472717285,
+      "learning_rate": 9.076147035413863e-06,
+      "loss": 2.5575,
+      "step": 815600
+    },
+    {
+      "epoch": 0.21273056023915107,
+      "grad_norm": 12.547298431396484,
+      "learning_rate": 9.0756628595311e-06,
+      "loss": 2.5415,
+      "step": 815800
+    },
+    {
+      "epoch": 0.21278271286485326,
+      "grad_norm": 13.744832038879395,
+      "learning_rate": 9.07517856972886e-06,
+      "loss": 2.5073,
+      "step": 816000
+    },
+    {
+      "epoch": 0.2128348654905554,
+      "grad_norm": 12.159852027893066,
+      "learning_rate": 9.074694166020682e-06,
+      "loss": 2.5024,
+      "step": 816200
+    },
+    {
+      "epoch": 0.2128870181162576,
+      "grad_norm": 12.20659065246582,
+      "learning_rate": 9.074209648420101e-06,
+      "loss": 2.518,
+      "step": 816400
+    },
+    {
+      "epoch": 0.21293917074195975,
+      "grad_norm": 12.317503929138184,
+      "learning_rate": 9.073725016940665e-06,
+      "loss": 2.5395,
+      "step": 816600
+    },
+    {
+      "epoch": 0.21299132336766194,
+      "grad_norm": 12.387846946716309,
+      "learning_rate": 9.073240271595916e-06,
+      "loss": 2.5347,
+      "step": 816800
+    },
+    {
+      "epoch": 0.2130434759933641,
+      "grad_norm": 13.619342803955078,
+      "learning_rate": 9.072755412399406e-06,
+      "loss": 2.5815,
+      "step": 817000
+    },
+    {
+      "epoch": 0.21309562861906628,
+      "grad_norm": 12.313323020935059,
+      "learning_rate": 9.072270439364684e-06,
+      "loss": 2.5288,
+      "step": 817200
+    },
+    {
+      "epoch": 0.21314778124476844,
+      "grad_norm": 15.066256523132324,
+      "learning_rate": 9.07178535250531e-06,
+      "loss": 2.5235,
+      "step": 817400
+    },
+    {
+      "epoch": 0.21319993387047062,
+      "grad_norm": 13.39647388458252,
+      "learning_rate": 9.07130015183484e-06,
+      "loss": 2.5484,
+      "step": 817600
+    },
+    {
+      "epoch": 0.21325208649617278,
+      "grad_norm": 11.415046691894531,
+      "learning_rate": 9.070814837366836e-06,
+      "loss": 2.5879,
+      "step": 817800
+    },
+    {
+      "epoch": 0.21330423912187496,
+      "grad_norm": 10.62917709350586,
+      "learning_rate": 9.070329409114862e-06,
+      "loss": 2.5464,
+      "step": 818000
+    },
+    {
+      "epoch": 0.21335639174757712,
+      "grad_norm": 13.691715240478516,
+      "learning_rate": 9.06984386709249e-06,
+      "loss": 2.5395,
+      "step": 818200
+    },
+    {
+      "epoch": 0.2134085443732793,
+      "grad_norm": 10.864567756652832,
+      "learning_rate": 9.069358211313289e-06,
+      "loss": 2.5282,
+      "step": 818400
+    },
+    {
+      "epoch": 0.21346069699898146,
+      "grad_norm": 11.341833114624023,
+      "learning_rate": 9.068872441790832e-06,
+      "loss": 2.5543,
+      "step": 818600
+    },
+    {
+      "epoch": 0.21351284962468364,
+      "grad_norm": 11.319380760192871,
+      "learning_rate": 9.0683865585387e-06,
+      "loss": 2.5383,
+      "step": 818800
+    },
+    {
+      "epoch": 0.2135650022503858,
+      "grad_norm": 11.097443580627441,
+      "learning_rate": 9.067900561570472e-06,
+      "loss": 2.5757,
+      "step": 819000
+    },
+    {
+      "epoch": 0.21361715487608796,
+      "grad_norm": 11.91701889038086,
+      "learning_rate": 9.067414450899735e-06,
+      "loss": 2.5255,
+      "step": 819200
+    },
+    {
+      "epoch": 0.21366930750179014,
+      "grad_norm": 11.580026626586914,
+      "learning_rate": 9.066928226540073e-06,
+      "loss": 2.5549,
+      "step": 819400
+    },
+    {
+      "epoch": 0.2137214601274923,
+      "grad_norm": 11.118182182312012,
+      "learning_rate": 9.066441888505077e-06,
+      "loss": 2.5484,
+      "step": 819600
+    },
+    {
+      "epoch": 0.21377361275319448,
+      "grad_norm": 11.571749687194824,
+      "learning_rate": 9.065955436808343e-06,
+      "loss": 2.5065,
+      "step": 819800
+    },
+    {
+      "epoch": 0.21382576537889664,
+      "grad_norm": 12.736671447753906,
+      "learning_rate": 9.065468871463464e-06,
+      "loss": 2.5625,
+      "step": 820000
+    },
+    {
+      "epoch": 0.21387791800459882,
+      "grad_norm": 13.102340698242188,
+      "learning_rate": 9.064982192484042e-06,
+      "loss": 2.5556,
+      "step": 820200
+    },
+    {
+      "epoch": 0.21393007063030098,
+      "grad_norm": 13.578348159790039,
+      "learning_rate": 9.064495399883684e-06,
+      "loss": 2.5439,
+      "step": 820400
+    },
+    {
+      "epoch": 0.21398222325600316,
+      "grad_norm": 13.68409538269043,
+      "learning_rate": 9.064008493675991e-06,
+      "loss": 2.5887,
+      "step": 820600
+    },
+    {
+      "epoch": 0.21403437588170532,
+      "grad_norm": 13.699695587158203,
+      "learning_rate": 9.063521473874574e-06,
+      "loss": 2.5437,
+      "step": 820800
+    },
+    {
+      "epoch": 0.2140865285074075,
+      "grad_norm": 12.38679313659668,
+      "learning_rate": 9.063034340493048e-06,
+      "loss": 2.5402,
+      "step": 821000
+    },
+    {
+      "epoch": 0.21413868113310966,
+      "grad_norm": 11.524242401123047,
+      "learning_rate": 9.062547093545025e-06,
+      "loss": 2.5281,
+      "step": 821200
+    },
+    {
+      "epoch": 0.21419083375881184,
+      "grad_norm": 11.350960731506348,
+      "learning_rate": 9.062059733044129e-06,
+      "loss": 2.5657,
+      "step": 821400
+    },
+    {
+      "epoch": 0.214242986384514,
+      "grad_norm": 14.124442100524902,
+      "learning_rate": 9.06157225900398e-06,
+      "loss": 2.543,
+      "step": 821600
+    },
+    {
+      "epoch": 0.21429513901021618,
+      "grad_norm": 13.106842041015625,
+      "learning_rate": 9.061084671438201e-06,
+      "loss": 2.5266,
+      "step": 821800
+    },
+    {
+      "epoch": 0.21434729163591834,
+      "grad_norm": 12.565150260925293,
+      "learning_rate": 9.060596970360423e-06,
+      "loss": 2.5241,
+      "step": 822000
+    },
+    {
+      "epoch": 0.21439944426162053,
+      "grad_norm": 12.959101676940918,
+      "learning_rate": 9.060109155784278e-06,
+      "loss": 2.5496,
+      "step": 822200
+    },
+    {
+      "epoch": 0.21445159688732268,
+      "grad_norm": 12.728083610534668,
+      "learning_rate": 9.059621227723401e-06,
+      "loss": 2.5343,
+      "step": 822400
+    },
+    {
+      "epoch": 0.21450374951302487,
+      "grad_norm": 13.199159622192383,
+      "learning_rate": 9.05913318619143e-06,
+      "loss": 2.5516,
+      "step": 822600
+    },
+    {
+      "epoch": 0.21455590213872702,
+      "grad_norm": 11.12812614440918,
+      "learning_rate": 9.058645031202005e-06,
+      "loss": 2.5033,
+      "step": 822800
+    },
+    {
+      "epoch": 0.2146080547644292,
+      "grad_norm": 13.735895156860352,
+      "learning_rate": 9.058156762768773e-06,
+      "loss": 2.5754,
+      "step": 823000
+    },
+    {
+      "epoch": 0.21466020739013136,
+      "grad_norm": 14.441383361816406,
+      "learning_rate": 9.057668380905379e-06,
+      "loss": 2.5403,
+      "step": 823200
+    },
+    {
+      "epoch": 0.21471236001583355,
+      "grad_norm": 14.57379150390625,
+      "learning_rate": 9.057179885625475e-06,
+      "loss": 2.5133,
+      "step": 823400
+    },
+    {
+      "epoch": 0.2147645126415357,
+      "grad_norm": 12.929058074951172,
+      "learning_rate": 9.056691276942717e-06,
+      "loss": 2.5448,
+      "step": 823600
+    },
+    {
+      "epoch": 0.2148166652672379,
+      "grad_norm": 12.269718170166016,
+      "learning_rate": 9.056202554870759e-06,
+      "loss": 2.5746,
+      "step": 823800
+    },
+    {
+      "epoch": 0.21486881789294004,
+      "grad_norm": 13.69266414642334,
+      "learning_rate": 9.055713719423262e-06,
+      "loss": 2.5415,
+      "step": 824000
+    },
+    {
+      "epoch": 0.21492097051864223,
+      "grad_norm": 10.19147777557373,
+      "learning_rate": 9.05522477061389e-06,
+      "loss": 2.6066,
+      "step": 824200
+    },
+    {
+      "epoch": 0.21497312314434439,
+      "grad_norm": 13.49588680267334,
+      "learning_rate": 9.054735708456311e-06,
+      "loss": 2.5503,
+      "step": 824400
+    },
+    {
+      "epoch": 0.21502527577004657,
+      "grad_norm": 13.206926345825195,
+      "learning_rate": 9.054246532964194e-06,
+      "loss": 2.582,
+      "step": 824600
+    },
+    {
+      "epoch": 0.21507742839574873,
+      "grad_norm": 14.236640930175781,
+      "learning_rate": 9.053757244151211e-06,
+      "loss": 2.563,
+      "step": 824800
+    },
+    {
+      "epoch": 0.21512958102145088,
+      "grad_norm": 12.087728500366211,
+      "learning_rate": 9.053267842031038e-06,
+      "loss": 2.5586,
+      "step": 825000
+    },
+    {
+      "epoch": 0.21518173364715307,
+      "grad_norm": 13.20797348022461,
+      "learning_rate": 9.052778326617357e-06,
+      "loss": 2.5417,
+      "step": 825200
+    },
+    {
+      "epoch": 0.21523388627285522,
+      "grad_norm": 13.239849090576172,
+      "learning_rate": 9.052288697923848e-06,
+      "loss": 2.5569,
+      "step": 825400
+    },
+    {
+      "epoch": 0.2152860388985574,
+      "grad_norm": 13.600337028503418,
+      "learning_rate": 9.051798955964196e-06,
+      "loss": 2.5431,
+      "step": 825600
+    },
+    {
+      "epoch": 0.21533819152425956,
+      "grad_norm": 14.598825454711914,
+      "learning_rate": 9.051309100752093e-06,
+      "loss": 2.5203,
+      "step": 825800
+    },
+    {
+      "epoch": 0.21539034414996175,
+      "grad_norm": 11.750934600830078,
+      "learning_rate": 9.050819132301227e-06,
+      "loss": 2.5027,
+      "step": 826000
+    },
+    {
+      "epoch": 0.2154424967756639,
+      "grad_norm": 13.683958053588867,
+      "learning_rate": 9.050329050625299e-06,
+      "loss": 2.5162,
+      "step": 826200
+    },
+    {
+      "epoch": 0.2154946494013661,
+      "grad_norm": 12.847051620483398,
+      "learning_rate": 9.049838855738001e-06,
+      "loss": 2.5338,
+      "step": 826400
+    },
+    {
+      "epoch": 0.21554680202706825,
+      "grad_norm": 12.681694984436035,
+      "learning_rate": 9.049348547653038e-06,
+      "loss": 2.532,
+      "step": 826600
+    },
+    {
+      "epoch": 0.21559895465277043,
+      "grad_norm": 13.66247272491455,
+      "learning_rate": 9.048858126384115e-06,
+      "loss": 2.5436,
+      "step": 826800
+    },
+    {
+      "epoch": 0.2156511072784726,
+      "grad_norm": 14.794609069824219,
+      "learning_rate": 9.048367591944938e-06,
+      "loss": 2.57,
+      "step": 827000
+    },
+    {
+      "epoch": 0.21570325990417477,
+      "grad_norm": 13.683991432189941,
+      "learning_rate": 9.04787694434922e-06,
+      "loss": 2.5269,
+      "step": 827200
+    },
+    {
+      "epoch": 0.21575541252987693,
+      "grad_norm": 11.677818298339844,
+      "learning_rate": 9.047386183610672e-06,
+      "loss": 2.5464,
+      "step": 827400
+    },
+    {
+      "epoch": 0.2158075651555791,
+      "grad_norm": 12.510682106018066,
+      "learning_rate": 9.046895309743016e-06,
+      "loss": 2.5522,
+      "step": 827600
+    },
+    {
+      "epoch": 0.21585971778128127,
+      "grad_norm": 14.272283554077148,
+      "learning_rate": 9.046404322759968e-06,
+      "loss": 2.5624,
+      "step": 827800
+    },
+    {
+      "epoch": 0.21591187040698345,
+      "grad_norm": 13.069134712219238,
+      "learning_rate": 9.045913222675253e-06,
+      "loss": 2.576,
+      "step": 828000
+    },
+    {
+      "epoch": 0.2159640230326856,
+      "grad_norm": 12.253870964050293,
+      "learning_rate": 9.045422009502598e-06,
+      "loss": 2.5634,
+      "step": 828200
+    },
+    {
+      "epoch": 0.2160161756583878,
+      "grad_norm": 13.879683494567871,
+      "learning_rate": 9.044930683255735e-06,
+      "loss": 2.5178,
+      "step": 828400
+    },
+    {
+      "epoch": 0.21606832828408995,
+      "grad_norm": 11.61858081817627,
+      "learning_rate": 9.044439243948395e-06,
+      "loss": 2.5837,
+      "step": 828600
+    },
+    {
+      "epoch": 0.21612048090979213,
+      "grad_norm": 14.2604398727417,
+      "learning_rate": 9.043947691594314e-06,
+      "loss": 2.5271,
+      "step": 828800
+    },
+    {
+      "epoch": 0.2161726335354943,
+      "grad_norm": 13.9703950881958,
+      "learning_rate": 9.043456026207234e-06,
+      "loss": 2.5254,
+      "step": 829000
+    },
+    {
+      "epoch": 0.21622478616119647,
+      "grad_norm": 12.149286270141602,
+      "learning_rate": 9.042964247800894e-06,
+      "loss": 2.5174,
+      "step": 829200
+    },
+    {
+      "epoch": 0.21627693878689863,
+      "grad_norm": 11.892560958862305,
+      "learning_rate": 9.042472356389043e-06,
+      "loss": 2.5527,
+      "step": 829400
+    },
+    {
+      "epoch": 0.21632909141260082,
+      "grad_norm": 12.29517650604248,
+      "learning_rate": 9.041980351985427e-06,
+      "loss": 2.5139,
+      "step": 829600
+    },
+    {
+      "epoch": 0.21638124403830297,
+      "grad_norm": 12.279248237609863,
+      "learning_rate": 9.0414882346038e-06,
+      "loss": 2.5625,
+      "step": 829800
+    },
+    {
+      "epoch": 0.21643339666400516,
+      "grad_norm": 13.284655570983887,
+      "learning_rate": 9.040996004257915e-06,
+      "loss": 2.5293,
+      "step": 830000
+    },
+    {
+      "epoch": 0.2164855492897073,
+      "grad_norm": 14.251969337463379,
+      "learning_rate": 9.040503660961536e-06,
+      "loss": 2.5693,
+      "step": 830200
+    },
+    {
+      "epoch": 0.2165377019154095,
+      "grad_norm": 11.924370765686035,
+      "learning_rate": 9.040011204728419e-06,
+      "loss": 2.5473,
+      "step": 830400
+    },
+    {
+      "epoch": 0.21658985454111165,
+      "grad_norm": 10.449265480041504,
+      "learning_rate": 9.03951863557233e-06,
+      "loss": 2.5117,
+      "step": 830600
+    },
+    {
+      "epoch": 0.2166420071668138,
+      "grad_norm": 12.4495210647583,
+      "learning_rate": 9.039025953507038e-06,
+      "loss": 2.5552,
+      "step": 830800
+    },
+    {
+      "epoch": 0.216694159792516,
+      "grad_norm": 13.751075744628906,
+      "learning_rate": 9.038533158546313e-06,
+      "loss": 2.568,
+      "step": 831000
+    },
+    {
+      "epoch": 0.21674631241821815,
+      "grad_norm": 12.695106506347656,
+      "learning_rate": 9.03804025070393e-06,
+      "loss": 2.5351,
+      "step": 831200
+    },
+    {
+      "epoch": 0.21679846504392034,
+      "grad_norm": 11.535404205322266,
+      "learning_rate": 9.037547229993667e-06,
+      "loss": 2.5723,
+      "step": 831400
+    },
+    {
+      "epoch": 0.2168506176696225,
+      "grad_norm": 12.56066608428955,
+      "learning_rate": 9.037054096429301e-06,
+      "loss": 2.54,
+      "step": 831600
+    },
+    {
+      "epoch": 0.21690277029532468,
+      "grad_norm": 12.696258544921875,
+      "learning_rate": 9.03656085002462e-06,
+      "loss": 2.5412,
+      "step": 831800
+    },
+    {
+      "epoch": 0.21695492292102683,
+      "grad_norm": 12.870928764343262,
+      "learning_rate": 9.036067490793408e-06,
+      "loss": 2.5416,
+      "step": 832000
+    },
+    {
+      "epoch": 0.21700707554672902,
+      "grad_norm": 11.494585037231445,
+      "learning_rate": 9.035574018749456e-06,
+      "loss": 2.5161,
+      "step": 832200
+    },
+    {
+      "epoch": 0.21705922817243117,
+      "grad_norm": 13.567378997802734,
+      "learning_rate": 9.035080433906557e-06,
+      "loss": 2.5638,
+      "step": 832400
+    },
+    {
+      "epoch": 0.21711138079813336,
+      "grad_norm": 11.019129753112793,
+      "learning_rate": 9.034586736278507e-06,
+      "loss": 2.5199,
+      "step": 832600
+    },
+    {
+      "epoch": 0.21716353342383551,
+      "grad_norm": 11.188690185546875,
+      "learning_rate": 9.034092925879106e-06,
+      "loss": 2.5601,
+      "step": 832800
+    },
+    {
+      "epoch": 0.2172156860495377,
+      "grad_norm": 13.095340728759766,
+      "learning_rate": 9.033599002722157e-06,
+      "loss": 2.5373,
+      "step": 833000
+    },
+    {
+      "epoch": 0.21726783867523985,
+      "grad_norm": 13.44362735748291,
+      "learning_rate": 9.033104966821464e-06,
+      "loss": 2.5655,
+      "step": 833200
+    },
+    {
+      "epoch": 0.21731999130094204,
+      "grad_norm": 13.946337699890137,
+      "learning_rate": 9.032610818190836e-06,
+      "loss": 2.5533,
+      "step": 833400
+    },
+    {
+      "epoch": 0.2173721439266442,
+      "grad_norm": 13.125496864318848,
+      "learning_rate": 9.032116556844088e-06,
+      "loss": 2.5252,
+      "step": 833600
+    },
+    {
+      "epoch": 0.21742429655234638,
+      "grad_norm": 14.219965934753418,
+      "learning_rate": 9.031622182795033e-06,
+      "loss": 2.5037,
+      "step": 833800
+    },
+    {
+      "epoch": 0.21747644917804854,
+      "grad_norm": 12.486796379089355,
+      "learning_rate": 9.031127696057488e-06,
+      "loss": 2.5568,
+      "step": 834000
+    },
+    {
+      "epoch": 0.21752860180375072,
+      "grad_norm": 14.136662483215332,
+      "learning_rate": 9.030633096645277e-06,
+      "loss": 2.5376,
+      "step": 834200
+    },
+    {
+      "epoch": 0.21758075442945288,
+      "grad_norm": 13.452658653259277,
+      "learning_rate": 9.030138384572222e-06,
+      "loss": 2.5766,
+      "step": 834400
+    },
+    {
+      "epoch": 0.21763290705515506,
+      "grad_norm": 11.165623664855957,
+      "learning_rate": 9.029643559852152e-06,
+      "loss": 2.527,
+      "step": 834600
+    },
+    {
+      "epoch": 0.21768505968085722,
+      "grad_norm": 11.985844612121582,
+      "learning_rate": 9.029148622498898e-06,
+      "loss": 2.5726,
+      "step": 834800
+    },
+    {
+      "epoch": 0.2177372123065594,
+      "grad_norm": 12.202817916870117,
+      "learning_rate": 9.028653572526296e-06,
+      "loss": 2.5832,
+      "step": 835000
+    },
+    {
+      "epoch": 0.21778936493226156,
+      "grad_norm": 11.637069702148438,
+      "learning_rate": 9.028158409948181e-06,
+      "loss": 2.5166,
+      "step": 835200
+    },
+    {
+      "epoch": 0.21784151755796374,
+      "grad_norm": 12.84749698638916,
+      "learning_rate": 9.027663134778394e-06,
+      "loss": 2.5217,
+      "step": 835400
+    },
+    {
+      "epoch": 0.2178936701836659,
+      "grad_norm": 12.990008354187012,
+      "learning_rate": 9.027167747030776e-06,
+      "loss": 2.5318,
+      "step": 835600
+    },
+    {
+      "epoch": 0.21794582280936808,
+      "grad_norm": 12.376798629760742,
+      "learning_rate": 9.026672246719179e-06,
+      "loss": 2.5525,
+      "step": 835800
+    },
+    {
+      "epoch": 0.21799797543507024,
+      "grad_norm": 11.98593521118164,
+      "learning_rate": 9.02617663385745e-06,
+      "loss": 2.492,
+      "step": 836000
+    },
+    {
+      "epoch": 0.21805012806077242,
+      "grad_norm": 12.605941772460938,
+      "learning_rate": 9.025680908459437e-06,
+      "loss": 2.5542,
+      "step": 836200
+    },
+    {
+      "epoch": 0.21810228068647458,
+      "grad_norm": 12.421950340270996,
+      "learning_rate": 9.025185070539005e-06,
+      "loss": 2.5238,
+      "step": 836400
+    },
+    {
+      "epoch": 0.21815443331217677,
+      "grad_norm": 11.057263374328613,
+      "learning_rate": 9.024689120110009e-06,
+      "loss": 2.5182,
+      "step": 836600
+    },
+    {
+      "epoch": 0.21820658593787892,
+      "grad_norm": 13.192597389221191,
+      "learning_rate": 9.02419305718631e-06,
+      "loss": 2.5215,
+      "step": 836800
+    },
+    {
+      "epoch": 0.21825873856358108,
+      "grad_norm": 13.819595336914062,
+      "learning_rate": 9.023696881781776e-06,
+      "loss": 2.5339,
+      "step": 837000
+    },
+    {
+      "epoch": 0.21831089118928326,
+      "grad_norm": 12.131237983703613,
+      "learning_rate": 9.023200593910273e-06,
+      "loss": 2.5688,
+      "step": 837200
+    },
+    {
+      "epoch": 0.21836304381498542,
+      "grad_norm": 12.664222717285156,
+      "learning_rate": 9.022704193585677e-06,
+      "loss": 2.5514,
+      "step": 837400
+    },
+    {
+      "epoch": 0.2184151964406876,
+      "grad_norm": 13.142849922180176,
+      "learning_rate": 9.022207680821858e-06,
+      "loss": 2.5142,
+      "step": 837600
+    },
+    {
+      "epoch": 0.21846734906638976,
+      "grad_norm": 13.29996109008789,
+      "learning_rate": 9.021711055632698e-06,
+      "loss": 2.5353,
+      "step": 837800
+    },
+    {
+      "epoch": 0.21851950169209194,
+      "grad_norm": 12.690460205078125,
+      "learning_rate": 9.021214318032077e-06,
+      "loss": 2.5696,
+      "step": 838000
+    },
+    {
+      "epoch": 0.2185716543177941,
+      "grad_norm": 13.326699256896973,
+      "learning_rate": 9.020717468033877e-06,
+      "loss": 2.5998,
+      "step": 838200
+    },
+    {
+      "epoch": 0.21862380694349628,
+      "grad_norm": 11.772669792175293,
+      "learning_rate": 9.02022050565199e-06,
+      "loss": 2.5221,
+      "step": 838400
+    },
+    {
+      "epoch": 0.21867595956919844,
+      "grad_norm": 12.478630065917969,
+      "learning_rate": 9.019723430900304e-06,
+      "loss": 2.5427,
+      "step": 838600
+    },
+    {
+      "epoch": 0.21872811219490063,
+      "grad_norm": 11.8933687210083,
+      "learning_rate": 9.019226243792712e-06,
+      "loss": 2.5576,
+      "step": 838800
+    },
+    {
+      "epoch": 0.21878026482060278,
+      "grad_norm": 11.510830879211426,
+      "learning_rate": 9.018728944343113e-06,
+      "loss": 2.5357,
+      "step": 839000
+    },
+    {
+      "epoch": 0.21883241744630497,
+      "grad_norm": 12.072577476501465,
+      "learning_rate": 9.018231532565407e-06,
+      "loss": 2.5547,
+      "step": 839200
+    },
+    {
+      "epoch": 0.21888457007200712,
+      "grad_norm": 12.852787017822266,
+      "learning_rate": 9.017734008473495e-06,
+      "loss": 2.5539,
+      "step": 839400
+    },
+    {
+      "epoch": 0.2189367226977093,
+      "grad_norm": 15.137070655822754,
+      "learning_rate": 9.017236372081286e-06,
+      "loss": 2.5924,
+      "step": 839600
+    },
+    {
+      "epoch": 0.21898887532341146,
+      "grad_norm": 13.725959777832031,
+      "learning_rate": 9.016738623402688e-06,
+      "loss": 2.5264,
+      "step": 839800
+    },
+    {
+      "epoch": 0.21904102794911365,
+      "grad_norm": 12.524657249450684,
+      "learning_rate": 9.016240762451613e-06,
+      "loss": 2.5328,
+      "step": 840000
+    },
+    {
+      "epoch": 0.2190931805748158,
+      "grad_norm": 12.656744956970215,
+      "learning_rate": 9.015742789241979e-06,
+      "loss": 2.5303,
+      "step": 840200
+    },
+    {
+      "epoch": 0.219145333200518,
+      "grad_norm": 14.41982364654541,
+      "learning_rate": 9.015244703787704e-06,
+      "loss": 2.5192,
+      "step": 840400
+    },
+    {
+      "epoch": 0.21919748582622015,
+      "grad_norm": 13.495166778564453,
+      "learning_rate": 9.014746506102709e-06,
+      "loss": 2.5097,
+      "step": 840600
+    },
+    {
+      "epoch": 0.21924963845192233,
+      "grad_norm": 12.303117752075195,
+      "learning_rate": 9.01424819620092e-06,
+      "loss": 2.5406,
+      "step": 840800
+    },
+    {
+      "epoch": 0.21930179107762449,
+      "grad_norm": 11.897546768188477,
+      "learning_rate": 9.013749774096265e-06,
+      "loss": 2.5776,
+      "step": 841000
+    },
+    {
+      "epoch": 0.21935394370332667,
+      "grad_norm": 11.718711853027344,
+      "learning_rate": 9.013251239802676e-06,
+      "loss": 2.5363,
+      "step": 841200
+    },
+    {
+      "epoch": 0.21940609632902883,
+      "grad_norm": 14.32603931427002,
+      "learning_rate": 9.012752593334087e-06,
+      "loss": 2.528,
+      "step": 841400
+    },
+    {
+      "epoch": 0.219458248954731,
+      "grad_norm": 11.693556785583496,
+      "learning_rate": 9.012253834704438e-06,
+      "loss": 2.5022,
+      "step": 841600
+    },
+    {
+      "epoch": 0.21951040158043317,
+      "grad_norm": 10.804730415344238,
+      "learning_rate": 9.011754963927666e-06,
+      "loss": 2.5521,
+      "step": 841800
+    },
+    {
+      "epoch": 0.21956255420613535,
+      "grad_norm": 12.166260719299316,
+      "learning_rate": 9.011255981017718e-06,
+      "loss": 2.5571,
+      "step": 842000
+    },
+    {
+      "epoch": 0.2196147068318375,
+      "grad_norm": 9.505474090576172,
+      "learning_rate": 9.010756885988541e-06,
+      "loss": 2.5141,
+      "step": 842200
+    },
+    {
+      "epoch": 0.2196668594575397,
+      "grad_norm": 12.927766799926758,
+      "learning_rate": 9.010257678854081e-06,
+      "loss": 2.5813,
+      "step": 842400
+    },
+    {
+      "epoch": 0.21971901208324185,
+      "grad_norm": 12.975953102111816,
+      "learning_rate": 9.009758359628298e-06,
+      "loss": 2.5434,
+      "step": 842600
+    },
+    {
+      "epoch": 0.219771164708944,
+      "grad_norm": 12.598480224609375,
+      "learning_rate": 9.009258928325146e-06,
+      "loss": 2.52,
+      "step": 842800
+    },
+    {
+      "epoch": 0.2198233173346462,
+      "grad_norm": 11.917780876159668,
+      "learning_rate": 9.008759384958582e-06,
+      "loss": 2.5169,
+      "step": 843000
+    },
+    {
+      "epoch": 0.21987546996034835,
+      "grad_norm": 12.304062843322754,
+      "learning_rate": 9.008259729542572e-06,
+      "loss": 2.5698,
+      "step": 843200
+    },
+    {
+      "epoch": 0.21992762258605053,
+      "grad_norm": 15.084339141845703,
+      "learning_rate": 9.00775996209108e-06,
+      "loss": 2.5361,
+      "step": 843400
+    },
+    {
+      "epoch": 0.2199797752117527,
+      "grad_norm": 13.05875301361084,
+      "learning_rate": 9.007260082618077e-06,
+      "loss": 2.5248,
+      "step": 843600
+    },
+    {
+      "epoch": 0.22003192783745487,
+      "grad_norm": 13.704729080200195,
+      "learning_rate": 9.006760091137534e-06,
+      "loss": 2.5512,
+      "step": 843800
+    },
+    {
+      "epoch": 0.22008408046315703,
+      "grad_norm": 13.089705467224121,
+      "learning_rate": 9.006259987663425e-06,
+      "loss": 2.526,
+      "step": 844000
+    },
+    {
+      "epoch": 0.2201362330888592,
+      "grad_norm": 12.926041603088379,
+      "learning_rate": 9.005759772209732e-06,
+      "loss": 2.5207,
+      "step": 844200
+    },
+    {
+      "epoch": 0.22018838571456137,
+      "grad_norm": 11.693181991577148,
+      "learning_rate": 9.005259444790432e-06,
+      "loss": 2.492,
+      "step": 844400
+    },
+    {
+      "epoch": 0.22024053834026355,
+      "grad_norm": 12.62060260772705,
+      "learning_rate": 9.004759005419515e-06,
+      "loss": 2.5323,
+      "step": 844600
+    },
+    {
+      "epoch": 0.2202926909659657,
+      "grad_norm": 12.25759506225586,
+      "learning_rate": 9.004258454110966e-06,
+      "loss": 2.4841,
+      "step": 844800
+    },
+    {
+      "epoch": 0.2203448435916679,
+      "grad_norm": 13.82259750366211,
+      "learning_rate": 9.003757790878775e-06,
+      "loss": 2.5533,
+      "step": 845000
+    },
+    {
+      "epoch": 0.22039699621737005,
+      "grad_norm": 12.706326484680176,
+      "learning_rate": 9.003257015736937e-06,
+      "loss": 2.5247,
+      "step": 845200
+    },
+    {
+      "epoch": 0.22044914884307223,
+      "grad_norm": 12.647273063659668,
+      "learning_rate": 9.00275612869945e-06,
+      "loss": 2.5762,
+      "step": 845400
+    },
+    {
+      "epoch": 0.2205013014687744,
+      "grad_norm": 12.341408729553223,
+      "learning_rate": 9.002255129780313e-06,
+      "loss": 2.5289,
+      "step": 845600
+    },
+    {
+      "epoch": 0.22055345409447658,
+      "grad_norm": 13.406020164489746,
+      "learning_rate": 9.001754018993531e-06,
+      "loss": 2.4797,
+      "step": 845800
+    },
+    {
+      "epoch": 0.22060560672017873,
+      "grad_norm": 13.798885345458984,
+      "learning_rate": 9.001252796353111e-06,
+      "loss": 2.5031,
+      "step": 846000
+    },
+    {
+      "epoch": 0.22065775934588092,
+      "grad_norm": 13.883713722229004,
+      "learning_rate": 9.000751461873061e-06,
+      "loss": 2.5013,
+      "step": 846200
+    },
+    {
+      "epoch": 0.22070991197158307,
+      "grad_norm": 12.606523513793945,
+      "learning_rate": 9.000250015567394e-06,
+      "loss": 2.5392,
+      "step": 846400
+    },
+    {
+      "epoch": 0.22076206459728526,
+      "grad_norm": 13.218819618225098,
+      "learning_rate": 8.999748457450125e-06,
+      "loss": 2.5223,
+      "step": 846600
+    },
+    {
+      "epoch": 0.2208142172229874,
+      "grad_norm": 14.202933311462402,
+      "learning_rate": 8.999246787535279e-06,
+      "loss": 2.5431,
+      "step": 846800
+    },
+    {
+      "epoch": 0.2208663698486896,
+      "grad_norm": 10.781413078308105,
+      "learning_rate": 8.99874500583687e-06,
+      "loss": 2.5516,
+      "step": 847000
+    },
+    {
+      "epoch": 0.22091852247439175,
+      "grad_norm": 12.983638763427734,
+      "learning_rate": 8.99824311236893e-06,
+      "loss": 2.5079,
+      "step": 847200
+    },
+    {
+      "epoch": 0.22097067510009394,
+      "grad_norm": 12.726056098937988,
+      "learning_rate": 8.997741107145484e-06,
+      "loss": 2.4876,
+      "step": 847400
+    },
+    {
+      "epoch": 0.2210228277257961,
+      "grad_norm": 13.305671691894531,
+      "learning_rate": 8.997238990180563e-06,
+      "loss": 2.5322,
+      "step": 847600
+    },
+    {
+      "epoch": 0.22107498035149828,
+      "grad_norm": 12.6900634765625,
+      "learning_rate": 8.996736761488205e-06,
+      "loss": 2.539,
+      "step": 847800
+    },
+    {
+      "epoch": 0.22112713297720044,
+      "grad_norm": 12.353796005249023,
+      "learning_rate": 8.996234421082447e-06,
+      "loss": 2.6022,
+      "step": 848000
+    },
+    {
+      "epoch": 0.22117928560290262,
+      "grad_norm": 13.8502836227417,
+      "learning_rate": 8.995731968977327e-06,
+      "loss": 2.5678,
+      "step": 848200
+    },
+    {
+      "epoch": 0.22123143822860478,
+      "grad_norm": 13.232843399047852,
+      "learning_rate": 8.995229405186892e-06,
+      "loss": 2.5363,
+      "step": 848400
+    },
+    {
+      "epoch": 0.22128359085430693,
+      "grad_norm": 13.014341354370117,
+      "learning_rate": 8.994726729725188e-06,
+      "loss": 2.5494,
+      "step": 848600
+    },
+    {
+      "epoch": 0.22133574348000912,
+      "grad_norm": 13.17010498046875,
+      "learning_rate": 8.994223942606266e-06,
+      "loss": 2.5208,
+      "step": 848800
+    },
+    {
+      "epoch": 0.22138789610571127,
+      "grad_norm": 13.344563484191895,
+      "learning_rate": 8.99372104384418e-06,
+      "loss": 2.5356,
+      "step": 849000
+    },
+    {
+      "epoch": 0.22144004873141346,
+      "grad_norm": 11.80360221862793,
+      "learning_rate": 8.993218033452985e-06,
+      "loss": 2.5652,
+      "step": 849200
+    },
+    {
+      "epoch": 0.22149220135711561,
+      "grad_norm": 12.011212348937988,
+      "learning_rate": 8.992714911446743e-06,
+      "loss": 2.5629,
+      "step": 849400
+    },
+    {
+      "epoch": 0.2215443539828178,
+      "grad_norm": 13.381622314453125,
+      "learning_rate": 8.992211677839514e-06,
+      "loss": 2.542,
+      "step": 849600
+    },
+    {
+      "epoch": 0.22159650660851996,
+      "grad_norm": 12.223089218139648,
+      "learning_rate": 8.991708332645365e-06,
+      "loss": 2.545,
+      "step": 849800
+    },
+    {
+      "epoch": 0.22164865923422214,
+      "grad_norm": 12.903460502624512,
+      "learning_rate": 8.991204875878369e-06,
+      "loss": 2.5258,
+      "step": 850000
+    },
+    {
+      "epoch": 0.2217008118599243,
+      "grad_norm": 13.3897123336792,
+      "learning_rate": 8.990701307552589e-06,
+      "loss": 2.4829,
+      "step": 850200
+    },
+    {
+      "epoch": 0.22175296448562648,
+      "grad_norm": 14.366217613220215,
+      "learning_rate": 8.99019762768211e-06,
+      "loss": 2.5757,
+      "step": 850400
+    },
+    {
+      "epoch": 0.22180511711132864,
+      "grad_norm": 12.164314270019531,
+      "learning_rate": 8.989693836281006e-06,
+      "loss": 2.5282,
+      "step": 850600
+    },
+    {
+      "epoch": 0.22185726973703082,
+      "grad_norm": 12.974170684814453,
+      "learning_rate": 8.989189933363359e-06,
+      "loss": 2.5623,
+      "step": 850800
+    },
+    {
+      "epoch": 0.22190942236273298,
+      "grad_norm": 12.635279655456543,
+      "learning_rate": 8.988685918943252e-06,
+      "loss": 2.4967,
+      "step": 851000
+    },
+    {
+      "epoch": 0.22196157498843516,
+      "grad_norm": 11.550530433654785,
+      "learning_rate": 8.988181793034776e-06,
+      "loss": 2.5069,
+      "step": 851200
+    },
+    {
+      "epoch": 0.22201372761413732,
+      "grad_norm": 12.456903457641602,
+      "learning_rate": 8.987677555652018e-06,
+      "loss": 2.5032,
+      "step": 851400
+    },
+    {
+      "epoch": 0.2220658802398395,
+      "grad_norm": 14.79409122467041,
+      "learning_rate": 8.987173206809078e-06,
+      "loss": 2.5223,
+      "step": 851600
+    },
+    {
+      "epoch": 0.22211803286554166,
+      "grad_norm": 12.690301895141602,
+      "learning_rate": 8.986668746520048e-06,
+      "loss": 2.5552,
+      "step": 851800
+    },
+    {
+      "epoch": 0.22217018549124384,
+      "grad_norm": 14.763097763061523,
+      "learning_rate": 8.986164174799029e-06,
+      "loss": 2.5392,
+      "step": 852000
+    },
+    {
+      "epoch": 0.222222338116946,
+      "grad_norm": 13.585575103759766,
+      "learning_rate": 8.985659491660126e-06,
+      "loss": 2.5645,
+      "step": 852200
+    },
+    {
+      "epoch": 0.22227449074264818,
+      "grad_norm": 13.037672996520996,
+      "learning_rate": 8.985154697117444e-06,
+      "loss": 2.545,
+      "step": 852400
+    },
+    {
+      "epoch": 0.22232664336835034,
+      "grad_norm": 13.282525062561035,
+      "learning_rate": 8.984649791185093e-06,
+      "loss": 2.5653,
+      "step": 852600
+    },
+    {
+      "epoch": 0.22237879599405252,
+      "grad_norm": 13.44238567352295,
+      "learning_rate": 8.984144773877186e-06,
+      "loss": 2.5334,
+      "step": 852800
+    },
+    {
+      "epoch": 0.22243094861975468,
+      "grad_norm": 14.811910629272461,
+      "learning_rate": 8.983639645207839e-06,
+      "loss": 2.5435,
+      "step": 853000
+    },
+    {
+      "epoch": 0.22248310124545687,
+      "grad_norm": 11.7992582321167,
+      "learning_rate": 8.98313440519117e-06,
+      "loss": 2.5333,
+      "step": 853200
+    },
+    {
+      "epoch": 0.22253525387115902,
+      "grad_norm": 12.599976539611816,
+      "learning_rate": 8.982629053841302e-06,
+      "loss": 2.5414,
+      "step": 853400
+    },
+    {
+      "epoch": 0.2225874064968612,
+      "grad_norm": 12.891295433044434,
+      "learning_rate": 8.98212359117236e-06,
+      "loss": 2.5178,
+      "step": 853600
+    },
+    {
+      "epoch": 0.22263955912256336,
+      "grad_norm": 10.120680809020996,
+      "learning_rate": 8.981618017198472e-06,
+      "loss": 2.523,
+      "step": 853800
+    },
+    {
+      "epoch": 0.22269171174826555,
+      "grad_norm": 14.803202629089355,
+      "learning_rate": 8.98111233193377e-06,
+      "loss": 2.5104,
+      "step": 854000
+    },
+    {
+      "epoch": 0.2227438643739677,
+      "grad_norm": 11.992095947265625,
+      "learning_rate": 8.980606535392387e-06,
+      "loss": 2.5419,
+      "step": 854200
+    },
+    {
+      "epoch": 0.22279601699966986,
+      "grad_norm": 13.348030090332031,
+      "learning_rate": 8.980100627588461e-06,
+      "loss": 2.5249,
+      "step": 854400
+    },
+    {
+      "epoch": 0.22284816962537204,
+      "grad_norm": 12.748968124389648,
+      "learning_rate": 8.979594608536135e-06,
+      "loss": 2.5046,
+      "step": 854600
+    },
+    {
+      "epoch": 0.2229003222510742,
+      "grad_norm": 12.485724449157715,
+      "learning_rate": 8.97908847824955e-06,
+      "loss": 2.5059,
+      "step": 854800
+    },
+    {
+      "epoch": 0.22295247487677639,
+      "grad_norm": 13.133146286010742,
+      "learning_rate": 8.978582236742854e-06,
+      "loss": 2.5285,
+      "step": 855000
+    },
+    {
+      "epoch": 0.22300462750247854,
+      "grad_norm": 11.857640266418457,
+      "learning_rate": 8.978075884030197e-06,
+      "loss": 2.5358,
+      "step": 855200
+    },
+    {
+      "epoch": 0.22305678012818073,
+      "grad_norm": 12.643730163574219,
+      "learning_rate": 8.977569420125732e-06,
+      "loss": 2.4963,
+      "step": 855400
+    },
+    {
+      "epoch": 0.22310893275388288,
+      "grad_norm": 11.763866424560547,
+      "learning_rate": 8.977062845043616e-06,
+      "loss": 2.5002,
+      "step": 855600
+    },
+    {
+      "epoch": 0.22316108537958507,
+      "grad_norm": 11.065905570983887,
+      "learning_rate": 8.976556158798006e-06,
+      "loss": 2.5613,
+      "step": 855800
+    },
+    {
+      "epoch": 0.22321323800528722,
+      "grad_norm": 12.88422679901123,
+      "learning_rate": 8.976049361403067e-06,
+      "loss": 2.5092,
+      "step": 856000
+    },
+    {
+      "epoch": 0.2232653906309894,
+      "grad_norm": 12.495038986206055,
+      "learning_rate": 8.975542452872966e-06,
+      "loss": 2.5522,
+      "step": 856200
+    },
+    {
+      "epoch": 0.22331754325669156,
+      "grad_norm": 10.122501373291016,
+      "learning_rate": 8.975035433221867e-06,
+      "loss": 2.5214,
+      "step": 856400
+    },
+    {
+      "epoch": 0.22336969588239375,
+      "grad_norm": 10.929478645324707,
+      "learning_rate": 8.974528302463946e-06,
+      "loss": 2.5244,
+      "step": 856600
+    },
+    {
+      "epoch": 0.2234218485080959,
+      "grad_norm": 13.326924324035645,
+      "learning_rate": 8.974021060613374e-06,
+      "loss": 2.499,
+      "step": 856800
+    },
+    {
+      "epoch": 0.2234740011337981,
+      "grad_norm": 11.079444885253906,
+      "learning_rate": 8.973513707684332e-06,
+      "loss": 2.5378,
+      "step": 857000
+    },
+    {
+      "epoch": 0.22352615375950025,
+      "grad_norm": 12.476800918579102,
+      "learning_rate": 8.973006243691001e-06,
+      "loss": 2.5788,
+      "step": 857200
+    },
+    {
+      "epoch": 0.22357830638520243,
+      "grad_norm": 14.203133583068848,
+      "learning_rate": 8.972498668647565e-06,
+      "loss": 2.5313,
+      "step": 857400
+    },
+    {
+      "epoch": 0.2236304590109046,
+      "grad_norm": 12.900059700012207,
+      "learning_rate": 8.971990982568208e-06,
+      "loss": 2.5252,
+      "step": 857600
+    },
+    {
+      "epoch": 0.22368261163660677,
+      "grad_norm": 10.50400447845459,
+      "learning_rate": 8.971483185467126e-06,
+      "loss": 2.506,
+      "step": 857800
+    },
+    {
+      "epoch": 0.22373476426230893,
+      "grad_norm": 12.504179954528809,
+      "learning_rate": 8.970975277358509e-06,
+      "loss": 2.5408,
+      "step": 858000
+    },
+    {
+      "epoch": 0.2237869168880111,
+      "grad_norm": 11.387737274169922,
+      "learning_rate": 8.970467258256552e-06,
+      "loss": 2.4938,
+      "step": 858200
+    },
+    {
+      "epoch": 0.22383906951371327,
+      "grad_norm": 12.079946517944336,
+      "learning_rate": 8.96995912817546e-06,
+      "loss": 2.5486,
+      "step": 858400
+    },
+    {
+      "epoch": 0.22389122213941545,
+      "grad_norm": 11.162164688110352,
+      "learning_rate": 8.969450887129431e-06,
+      "loss": 2.5334,
+      "step": 858600
+    },
+    {
+      "epoch": 0.2239433747651176,
+      "grad_norm": 15.253430366516113,
+      "learning_rate": 8.968942535132675e-06,
+      "loss": 2.5128,
+      "step": 858800
+    },
+    {
+      "epoch": 0.2239955273908198,
+      "grad_norm": 12.579545974731445,
+      "learning_rate": 8.968434072199396e-06,
+      "loss": 2.59,
+      "step": 859000
+    },
+    {
+      "epoch": 0.22404768001652195,
+      "grad_norm": 11.06598949432373,
+      "learning_rate": 8.96792549834381e-06,
+      "loss": 2.5401,
+      "step": 859200
+    },
+    {
+      "epoch": 0.22409983264222413,
+      "grad_norm": 12.26706600189209,
+      "learning_rate": 8.967416813580132e-06,
+      "loss": 2.5401,
+      "step": 859400
+    },
+    {
+      "epoch": 0.2241519852679263,
+      "grad_norm": 12.439864158630371,
+      "learning_rate": 8.966908017922578e-06,
+      "loss": 2.5566,
+      "step": 859600
+    },
+    {
+      "epoch": 0.22420413789362847,
+      "grad_norm": 12.08425521850586,
+      "learning_rate": 8.966399111385371e-06,
+      "loss": 2.5422,
+      "step": 859800
+    },
+    {
+      "epoch": 0.22425629051933063,
+      "grad_norm": 11.494114875793457,
+      "learning_rate": 8.965890093982736e-06,
+      "loss": 2.5269,
+      "step": 860000
+    },
+    {
+      "epoch": 0.2243084431450328,
+      "grad_norm": 13.456847190856934,
+      "learning_rate": 8.9653809657289e-06,
+      "loss": 2.5331,
+      "step": 860200
+    },
+    {
+      "epoch": 0.22436059577073497,
+      "grad_norm": 12.735716819763184,
+      "learning_rate": 8.964871726638093e-06,
+      "loss": 2.5312,
+      "step": 860400
+    },
+    {
+      "epoch": 0.22441274839643713,
+      "grad_norm": 12.311686515808105,
+      "learning_rate": 8.964362376724552e-06,
+      "loss": 2.5054,
+      "step": 860600
+    },
+    {
+      "epoch": 0.2244649010221393,
+      "grad_norm": 14.03046989440918,
+      "learning_rate": 8.96385291600251e-06,
+      "loss": 2.5587,
+      "step": 860800
+    },
+    {
+      "epoch": 0.22451705364784147,
+      "grad_norm": 12.448440551757812,
+      "learning_rate": 8.963343344486208e-06,
+      "loss": 2.5317,
+      "step": 861000
+    },
+    {
+      "epoch": 0.22456920627354365,
+      "grad_norm": 12.499717712402344,
+      "learning_rate": 8.962833662189889e-06,
+      "loss": 2.5408,
+      "step": 861200
+    },
+    {
+      "epoch": 0.2246213588992458,
+      "grad_norm": 12.628153800964355,
+      "learning_rate": 8.962323869127802e-06,
+      "loss": 2.5539,
+      "step": 861400
+    },
+    {
+      "epoch": 0.224673511524948,
+      "grad_norm": 12.310242652893066,
+      "learning_rate": 8.961813965314192e-06,
+      "loss": 2.4965,
+      "step": 861600
+    },
+    {
+      "epoch": 0.22472566415065015,
+      "grad_norm": 12.289976119995117,
+      "learning_rate": 8.961303950763314e-06,
+      "loss": 2.5299,
+      "step": 861800
+    },
+    {
+      "epoch": 0.22477781677635233,
+      "grad_norm": 14.414912223815918,
+      "learning_rate": 8.960793825489425e-06,
+      "loss": 2.6051,
+      "step": 862000
+    },
+    {
+      "epoch": 0.2248299694020545,
+      "grad_norm": 12.091730117797852,
+      "learning_rate": 8.960283589506779e-06,
+      "loss": 2.5215,
+      "step": 862200
+    },
+    {
+      "epoch": 0.22488212202775668,
+      "grad_norm": 13.612672805786133,
+      "learning_rate": 8.959773242829641e-06,
+      "loss": 2.5501,
+      "step": 862400
+    },
+    {
+      "epoch": 0.22493427465345883,
+      "grad_norm": 13.881776809692383,
+      "learning_rate": 8.959262785472275e-06,
+      "loss": 2.5188,
+      "step": 862600
+    },
+    {
+      "epoch": 0.22498642727916102,
+      "grad_norm": 14.095483779907227,
+      "learning_rate": 8.95875221744895e-06,
+      "loss": 2.5083,
+      "step": 862800
+    },
+    {
+      "epoch": 0.22503857990486317,
+      "grad_norm": 12.817255020141602,
+      "learning_rate": 8.958241538773935e-06,
+      "loss": 2.5343,
+      "step": 863000
+    },
+    {
+      "epoch": 0.22509073253056536,
+      "grad_norm": 12.889725685119629,
+      "learning_rate": 8.957730749461505e-06,
+      "loss": 2.5256,
+      "step": 863200
+    },
+    {
+      "epoch": 0.2251428851562675,
+      "grad_norm": 13.904434204101562,
+      "learning_rate": 8.957219849525938e-06,
+      "loss": 2.5147,
+      "step": 863400
+    },
+    {
+      "epoch": 0.2251950377819697,
+      "grad_norm": 12.321836471557617,
+      "learning_rate": 8.956708838981512e-06,
+      "loss": 2.5378,
+      "step": 863600
+    },
+    {
+      "epoch": 0.22524719040767185,
+      "grad_norm": 12.931632041931152,
+      "learning_rate": 8.956197717842512e-06,
+      "loss": 2.4962,
+      "step": 863800
+    },
+    {
+      "epoch": 0.22529934303337404,
+      "grad_norm": 13.46435260772705,
+      "learning_rate": 8.955686486123225e-06,
+      "loss": 2.5106,
+      "step": 864000
+    },
+    {
+      "epoch": 0.2253514956590762,
+      "grad_norm": 13.250704765319824,
+      "learning_rate": 8.95517514383794e-06,
+      "loss": 2.5317,
+      "step": 864200
+    },
+    {
+      "epoch": 0.22540364828477838,
+      "grad_norm": 12.657617568969727,
+      "learning_rate": 8.954663691000947e-06,
+      "loss": 2.5211,
+      "step": 864400
+    },
+    {
+      "epoch": 0.22545580091048054,
+      "grad_norm": 13.068683624267578,
+      "learning_rate": 8.954152127626545e-06,
+      "loss": 2.4836,
+      "step": 864600
+    },
+    {
+      "epoch": 0.22550795353618272,
+      "grad_norm": 12.12160873413086,
+      "learning_rate": 8.953640453729034e-06,
+      "loss": 2.548,
+      "step": 864800
+    },
+    {
+      "epoch": 0.22556010616188488,
+      "grad_norm": 11.243968963623047,
+      "learning_rate": 8.95312866932271e-06,
+      "loss": 2.5234,
+      "step": 865000
+    },
+    {
+      "epoch": 0.22561225878758706,
+      "grad_norm": 12.38250732421875,
+      "learning_rate": 8.952616774421883e-06,
+      "loss": 2.573,
+      "step": 865200
+    },
+    {
+      "epoch": 0.22566441141328922,
+      "grad_norm": 10.652915954589844,
+      "learning_rate": 8.95210476904086e-06,
+      "loss": 2.5371,
+      "step": 865400
+    },
+    {
+      "epoch": 0.2257165640389914,
+      "grad_norm": 13.076416015625,
+      "learning_rate": 8.95159265319395e-06,
+      "loss": 2.5046,
+      "step": 865600
+    },
+    {
+      "epoch": 0.22576871666469356,
+      "grad_norm": 12.925193786621094,
+      "learning_rate": 8.951080426895473e-06,
+      "loss": 2.5507,
+      "step": 865800
+    },
+    {
+      "epoch": 0.22582086929039574,
+      "grad_norm": 14.921239852905273,
+      "learning_rate": 8.95056809015974e-06,
+      "loss": 2.5211,
+      "step": 866000
+    },
+    {
+      "epoch": 0.2258730219160979,
+      "grad_norm": 12.77281665802002,
+      "learning_rate": 8.950055643001072e-06,
+      "loss": 2.548,
+      "step": 866200
+    },
+    {
+      "epoch": 0.22592517454180006,
+      "grad_norm": 12.110180854797363,
+      "learning_rate": 8.949543085433797e-06,
+      "loss": 2.5187,
+      "step": 866400
+    },
+    {
+      "epoch": 0.22597732716750224,
+      "grad_norm": 12.40743637084961,
+      "learning_rate": 8.949030417472235e-06,
+      "loss": 2.507,
+      "step": 866600
+    },
+    {
+      "epoch": 0.2260294797932044,
+      "grad_norm": 12.984556198120117,
+      "learning_rate": 8.948517639130722e-06,
+      "loss": 2.5171,
+      "step": 866800
+    },
+    {
+      "epoch": 0.22608163241890658,
+      "grad_norm": 12.792410850524902,
+      "learning_rate": 8.948004750423589e-06,
+      "loss": 2.5155,
+      "step": 867000
+    },
+    {
+      "epoch": 0.22613378504460874,
+      "grad_norm": 13.460736274719238,
+      "learning_rate": 8.94749175136517e-06,
+      "loss": 2.5426,
+      "step": 867200
+    },
+    {
+      "epoch": 0.22618593767031092,
+      "grad_norm": 10.04277229309082,
+      "learning_rate": 8.946978641969805e-06,
+      "loss": 2.5214,
+      "step": 867400
+    },
+    {
+      "epoch": 0.22623809029601308,
+      "grad_norm": 12.798548698425293,
+      "learning_rate": 8.946465422251836e-06,
+      "loss": 2.585,
+      "step": 867600
+    },
+    {
+      "epoch": 0.22629024292171526,
+      "grad_norm": 12.13387680053711,
+      "learning_rate": 8.945952092225607e-06,
+      "loss": 2.4898,
+      "step": 867800
+    },
+    {
+      "epoch": 0.22634239554741742,
+      "grad_norm": 13.749103546142578,
+      "learning_rate": 8.945438651905469e-06,
+      "loss": 2.4992,
+      "step": 868000
+    },
+    {
+      "epoch": 0.2263945481731196,
+      "grad_norm": 12.682474136352539,
+      "learning_rate": 8.94492510130577e-06,
+      "loss": 2.534,
+      "step": 868200
+    },
+    {
+      "epoch": 0.22644670079882176,
+      "grad_norm": 14.173575401306152,
+      "learning_rate": 8.944411440440865e-06,
+      "loss": 2.5106,
+      "step": 868400
+    },
+    {
+      "epoch": 0.22649885342452394,
+      "grad_norm": 14.226604461669922,
+      "learning_rate": 8.943897669325114e-06,
+      "loss": 2.538,
+      "step": 868600
+    },
+    {
+      "epoch": 0.2265510060502261,
+      "grad_norm": 11.75302791595459,
+      "learning_rate": 8.943383787972875e-06,
+      "loss": 2.5191,
+      "step": 868800
+    },
+    {
+      "epoch": 0.22660315867592828,
+      "grad_norm": 14.185355186462402,
+      "learning_rate": 8.942869796398513e-06,
+      "loss": 2.5775,
+      "step": 869000
+    },
+    {
+      "epoch": 0.22665531130163044,
+      "grad_norm": 13.081635475158691,
+      "learning_rate": 8.942355694616393e-06,
+      "loss": 2.5,
+      "step": 869200
+    },
+    {
+      "epoch": 0.22670746392733263,
+      "grad_norm": 13.637446403503418,
+      "learning_rate": 8.941841482640885e-06,
+      "loss": 2.5123,
+      "step": 869400
+    },
+    {
+      "epoch": 0.22675961655303478,
+      "grad_norm": 12.775493621826172,
+      "learning_rate": 8.941327160486364e-06,
+      "loss": 2.5291,
+      "step": 869600
+    },
+    {
+      "epoch": 0.22681176917873697,
+      "grad_norm": 12.630292892456055,
+      "learning_rate": 8.940812728167203e-06,
+      "loss": 2.5195,
+      "step": 869800
+    },
+    {
+      "epoch": 0.22686392180443912,
+      "grad_norm": 12.8863525390625,
+      "learning_rate": 8.940298185697783e-06,
+      "loss": 2.5228,
+      "step": 870000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 3834898,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 15000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}