diff --git "a/final/trainer_state.json" "b/final/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/final/trainer_state.json"
@@ -0,0 +1,4185 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7532681812412743,
+  "eval_steps": 500,
+  "global_step": 5935,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001269196598553116,
+      "grad_norm": 64.06603546200458,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 1.496615505218506,
+      "step": 10
+    },
+    {
+      "epoch": 0.002538393197106232,
+      "grad_norm": 61.17677058430513,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 0.9656316757202148,
+      "step": 20
+    },
+    {
+      "epoch": 0.0038075897956593475,
+      "grad_norm": 53.26206127683936,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.8455992698669433,
+      "step": 30
+    },
+    {
+      "epoch": 0.005076786394212464,
+      "grad_norm": 35.42148973971469,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.833045482635498,
+      "step": 40
+    },
+    {
+      "epoch": 0.006345982992765579,
+      "grad_norm": 28.31039728631522,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.8395201683044433,
+      "step": 50
+    },
+    {
+      "epoch": 0.007615179591318695,
+      "grad_norm": 92.00417899138034,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.7459193229675293,
+      "step": 60
+    },
+    {
+      "epoch": 0.00888437618987181,
+      "grad_norm": 21.56423431358976,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.7551187515258789,
+      "step": 70
+    },
+    {
+      "epoch": 0.010153572788424927,
+      "grad_norm": 24.279111717697603,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.7326831817626953,
+      "step": 80
+    },
+    {
+      "epoch": 0.011422769386978043,
+      "grad_norm": 7.92340537047251,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7629533290863038,
+      "step": 90
+    },
+    {
+      "epoch": 0.012691965985531158,
+      "grad_norm": 25.68185994635364,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.7467505455017089,
+      "step": 100
+    },
+    {
+      "epoch": 0.013961162584084275,
+      "grad_norm": 95.24574877398771,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.7687312126159668,
+      "step": 110
+    },
+    {
+      "epoch": 0.01523035918263739,
+      "grad_norm": 139.8090383590867,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.7320384979248047,
+      "step": 120
+    },
+    {
+      "epoch": 0.016499555781190507,
+      "grad_norm": 19.751796793598576,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.7771665573120117,
+      "step": 130
+    },
+    {
+      "epoch": 0.01776875237974362,
+      "grad_norm": 8.431699053867712,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 0.8182608604431152,
+      "step": 140
+    },
+    {
+      "epoch": 0.019037948978296738,
+      "grad_norm": 18.361714667650983,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.7524012565612793,
+      "step": 150
+    },
+    {
+      "epoch": 0.020307145576849855,
+      "grad_norm": 22.221439953917375,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.7908215999603272,
+      "step": 160
+    },
+    {
+      "epoch": 0.02157634217540297,
+      "grad_norm": 6.171916519477009,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.7817217826843261,
+      "step": 170
+    },
+    {
+      "epoch": 0.022845538773956085,
+      "grad_norm": 386.2647117794816,
+      "learning_rate": 2e-05,
+      "loss": 0.7678230285644532,
+      "step": 180
+    },
+    {
+      "epoch": 0.024114735372509202,
+      "grad_norm": 46.79698458675453,
+      "learning_rate": 2e-05,
+      "loss": 0.8175263404846191,
+      "step": 190
+    },
+    {
+      "epoch": 0.025383931971062316,
+      "grad_norm": 21.564731067927138,
+      "learning_rate": 2e-05,
+      "loss": 0.8099200248718261,
+      "step": 200
+    },
+    {
+      "epoch": 0.026653128569615433,
+      "grad_norm": 10.112660799094582,
+      "learning_rate": 2e-05,
+      "loss": 0.7557328701019287,
+      "step": 210
+    },
+    {
+      "epoch": 0.02792232516816855,
+      "grad_norm": 70.87770991444113,
+      "learning_rate": 2e-05,
+      "loss": 0.799714469909668,
+      "step": 220
+    },
+    {
+      "epoch": 0.029191521766721667,
+      "grad_norm": 29.61142712899928,
+      "learning_rate": 2e-05,
+      "loss": 0.7821331024169922,
+      "step": 230
+    },
+    {
+      "epoch": 0.03046071836527478,
+      "grad_norm": 6.028570925206176,
+      "learning_rate": 2e-05,
+      "loss": 0.794653844833374,
+      "step": 240
+    },
+    {
+      "epoch": 0.0317299149638279,
+      "grad_norm": 7.356925131714258,
+      "learning_rate": 2e-05,
+      "loss": 0.7670344352722168,
+      "step": 250
+    },
+    {
+      "epoch": 0.032999111562381014,
+      "grad_norm": 8.50292238259149,
+      "learning_rate": 2e-05,
+      "loss": 0.7702371120452881,
+      "step": 260
+    },
+    {
+      "epoch": 0.03426830816093413,
+      "grad_norm": 5.738770293267178,
+      "learning_rate": 2e-05,
+      "loss": 0.7589911937713623,
+      "step": 270
+    },
+    {
+      "epoch": 0.03553750475948724,
+      "grad_norm": 13.780459778070338,
+      "learning_rate": 2e-05,
+      "loss": 0.7504461288452149,
+      "step": 280
+    },
+    {
+      "epoch": 0.03680670135804036,
+      "grad_norm": 7.037259039593961,
+      "learning_rate": 2e-05,
+      "loss": 0.7812541961669922,
+      "step": 290
+    },
+    {
+      "epoch": 0.038075897956593475,
+      "grad_norm": 47.53464818478786,
+      "learning_rate": 2e-05,
+      "loss": 0.7693456649780274,
+      "step": 300
+    },
+    {
+      "epoch": 0.03934509455514659,
+      "grad_norm": 20.810012535447804,
+      "learning_rate": 2e-05,
+      "loss": 0.8223299026489258,
+      "step": 310
+    },
+    {
+      "epoch": 0.04061429115369971,
+      "grad_norm": 21.127414301668168,
+      "learning_rate": 2e-05,
+      "loss": 0.870754623413086,
+      "step": 320
+    },
+    {
+      "epoch": 0.041883487752252826,
+      "grad_norm": 3.106504462577958,
+      "learning_rate": 2e-05,
+      "loss": 0.8264900207519531,
+      "step": 330
+    },
+    {
+      "epoch": 0.04315268435080594,
+      "grad_norm": 18.852907786721996,
+      "learning_rate": 2e-05,
+      "loss": 0.7927096366882325,
+      "step": 340
+    },
+    {
+      "epoch": 0.044421880949359054,
+      "grad_norm": 3.352542910995253,
+      "learning_rate": 2e-05,
+      "loss": 0.7883159637451171,
+      "step": 350
+    },
+    {
+      "epoch": 0.04569107754791217,
+      "grad_norm": 12.669725383780657,
+      "learning_rate": 2e-05,
+      "loss": 0.7765592098236084,
+      "step": 360
+    },
+    {
+      "epoch": 0.04696027414646529,
+      "grad_norm": 25.877518535303064,
+      "learning_rate": 2e-05,
+      "loss": 0.7736199855804443,
+      "step": 370
+    },
+    {
+      "epoch": 0.048229470745018405,
+      "grad_norm": 10.91816434794243,
+      "learning_rate": 2e-05,
+      "loss": 0.7494498729705811,
+      "step": 380
+    },
+    {
+      "epoch": 0.04949866734357152,
+      "grad_norm": 13.67406209259735,
+      "learning_rate": 2e-05,
+      "loss": 0.7499611854553223,
+      "step": 390
+    },
+    {
+      "epoch": 0.05076786394212463,
+      "grad_norm": 3.8147882194852087,
+      "learning_rate": 2e-05,
+      "loss": 0.7557431697845459,
+      "step": 400
+    },
+    {
+      "epoch": 0.05203706054067775,
+      "grad_norm": 2.7753971814485774,
+      "learning_rate": 2e-05,
+      "loss": 0.7824067592620849,
+      "step": 410
+    },
+    {
+      "epoch": 0.053306257139230866,
+      "grad_norm": 3.1414993701491425,
+      "learning_rate": 2e-05,
+      "loss": 0.7844434261322022,
+      "step": 420
+    },
+    {
+      "epoch": 0.05457545373778398,
+      "grad_norm": 2.650376019691899,
+      "learning_rate": 2e-05,
+      "loss": 0.7791579246520997,
+      "step": 430
+    },
+    {
+      "epoch": 0.0558446503363371,
+      "grad_norm": 2.3505776158955114,
+      "learning_rate": 2e-05,
+      "loss": 0.7677821159362793,
+      "step": 440
+    },
+    {
+      "epoch": 0.05711384693489022,
+      "grad_norm": 2.4167304649154544,
+      "learning_rate": 2e-05,
+      "loss": 0.7416158199310303,
+      "step": 450
+    },
+    {
+      "epoch": 0.058383043533443334,
+      "grad_norm": 20.628088081908153,
+      "learning_rate": 2e-05,
+      "loss": 0.786812162399292,
+      "step": 460
+    },
+    {
+      "epoch": 0.059652240131996444,
+      "grad_norm": 14.036840748904096,
+      "learning_rate": 2e-05,
+      "loss": 0.806132698059082,
+      "step": 470
+    },
+    {
+      "epoch": 0.06092143673054956,
+      "grad_norm": 3.6108523197876727,
+      "learning_rate": 2e-05,
+      "loss": 0.7933418273925781,
+      "step": 480
+    },
+    {
+      "epoch": 0.06219063332910268,
+      "grad_norm": 2.8383392566788004,
+      "learning_rate": 2e-05,
+      "loss": 0.783383846282959,
+      "step": 490
+    },
+    {
+      "epoch": 0.0634598299276558,
+      "grad_norm": 3.311630151785513,
+      "learning_rate": 2e-05,
+      "loss": 0.752874755859375,
+      "step": 500
+    },
+    {
+      "epoch": 0.0647290265262089,
+      "grad_norm": 18.404185102869576,
+      "learning_rate": 2e-05,
+      "loss": 0.7731963634490967,
+      "step": 510
+    },
+    {
+      "epoch": 0.06599822312476203,
+      "grad_norm": 5.722365541276341,
+      "learning_rate": 2e-05,
+      "loss": 0.7650835990905762,
+      "step": 520
+    },
+    {
+      "epoch": 0.06726741972331514,
+      "grad_norm": 2.882286599410323,
+      "learning_rate": 2e-05,
+      "loss": 0.7476066589355469,
+      "step": 530
+    },
+    {
+      "epoch": 0.06853661632186826,
+      "grad_norm": 3.524305738605355,
+      "learning_rate": 2e-05,
+      "loss": 0.7588428020477295,
+      "step": 540
+    },
+    {
+      "epoch": 0.06980581292042137,
+      "grad_norm": 4.1229386234958385,
+      "learning_rate": 2e-05,
+      "loss": 0.751226806640625,
+      "step": 550
+    },
+    {
+      "epoch": 0.07107500951897448,
+      "grad_norm": 2.53194141785262,
+      "learning_rate": 2e-05,
+      "loss": 0.7618921279907227,
+      "step": 560
+    },
+    {
+      "epoch": 0.0723442061175276,
+      "grad_norm": 4.802129594690749,
+      "learning_rate": 2e-05,
+      "loss": 0.7337077617645263,
+      "step": 570
+    },
+    {
+      "epoch": 0.07361340271608072,
+      "grad_norm": 3.0181946999414118,
+      "learning_rate": 2e-05,
+      "loss": 0.7823495388031005,
+      "step": 580
+    },
+    {
+      "epoch": 0.07488259931463384,
+      "grad_norm": 3.2204673609998435,
+      "learning_rate": 2e-05,
+      "loss": 0.8138880729675293,
+      "step": 590
+    },
+    {
+      "epoch": 0.07615179591318695,
+      "grad_norm": 8.428231557915897,
+      "learning_rate": 2e-05,
+      "loss": 0.7558135986328125,
+      "step": 600
+    },
+    {
+      "epoch": 0.07742099251174007,
+      "grad_norm": 4.133422858758047,
+      "learning_rate": 2e-05,
+      "loss": 0.7770913600921631,
+      "step": 610
+    },
+    {
+      "epoch": 0.07869018911029318,
+      "grad_norm": 2.911616377790855,
+      "learning_rate": 2e-05,
+      "loss": 0.763375997543335,
+      "step": 620
+    },
+    {
+      "epoch": 0.0799593857088463,
+      "grad_norm": 7.607499235319638,
+      "learning_rate": 2e-05,
+      "loss": 0.7343894004821777,
+      "step": 630
+    },
+    {
+      "epoch": 0.08122858230739942,
+      "grad_norm": 54.5393573873479,
+      "learning_rate": 2e-05,
+      "loss": 0.7464728832244873,
+      "step": 640
+    },
+    {
+      "epoch": 0.08249777890595253,
+      "grad_norm": 2.858288244018835,
+      "learning_rate": 2e-05,
+      "loss": 0.7510586738586426,
+      "step": 650
+    },
+    {
+      "epoch": 0.08376697550450565,
+      "grad_norm": 2.1845652772113864,
+      "learning_rate": 2e-05,
+      "loss": 0.7360637664794922,
+      "step": 660
+    },
+    {
+      "epoch": 0.08503617210305876,
+      "grad_norm": 1.9250890561063947,
+      "learning_rate": 2e-05,
+      "loss": 0.7429749965667725,
+      "step": 670
+    },
+    {
+      "epoch": 0.08630536870161187,
+      "grad_norm": 2.8579361424025373,
+      "learning_rate": 2e-05,
+      "loss": 0.7204679012298584,
+      "step": 680
+    },
+    {
+      "epoch": 0.087574565300165,
+      "grad_norm": 1.9912246670172478,
+      "learning_rate": 2e-05,
+      "loss": 0.7139169692993164,
+      "step": 690
+    },
+    {
+      "epoch": 0.08884376189871811,
+      "grad_norm": 2.6208348162106097,
+      "learning_rate": 2e-05,
+      "loss": 0.7403483390808105,
+      "step": 700
+    },
+    {
+      "epoch": 0.09011295849727123,
+      "grad_norm": 1.7679064456764806,
+      "learning_rate": 2e-05,
+      "loss": 0.761525821685791,
+      "step": 710
+    },
+    {
+      "epoch": 0.09138215509582434,
+      "grad_norm": 2.298026459953476,
+      "learning_rate": 2e-05,
+      "loss": 0.7347946166992188,
+      "step": 720
+    },
+    {
+      "epoch": 0.09265135169437747,
+      "grad_norm": 2.406757933836833,
+      "learning_rate": 2e-05,
+      "loss": 0.7573525905609131,
+      "step": 730
+    },
+    {
+      "epoch": 0.09392054829293058,
+      "grad_norm": 2.870561511277864,
+      "learning_rate": 2e-05,
+      "loss": 0.7275297164916992,
+      "step": 740
+    },
+    {
+      "epoch": 0.09518974489148369,
+      "grad_norm": 1.7343538454888145,
+      "learning_rate": 2e-05,
+      "loss": 0.7347771644592285,
+      "step": 750
+    },
+    {
+      "epoch": 0.09645894149003681,
+      "grad_norm": 2.3967764817589803,
+      "learning_rate": 2e-05,
+      "loss": 0.7204466819763183,
+      "step": 760
+    },
+    {
+      "epoch": 0.09772813808858992,
+      "grad_norm": 1.8719845811174207,
+      "learning_rate": 2e-05,
+      "loss": 0.7233079433441162,
+      "step": 770
+    },
+    {
+      "epoch": 0.09899733468714304,
+      "grad_norm": 41.56740308804347,
+      "learning_rate": 2e-05,
+      "loss": 0.7606782436370849,
+      "step": 780
+    },
+    {
+      "epoch": 0.10026653128569615,
+      "grad_norm": 2.849413874131603,
+      "learning_rate": 2e-05,
+      "loss": 0.7183428764343261,
+      "step": 790
+    },
+    {
+      "epoch": 0.10153572788424926,
+      "grad_norm": 1.5961048711629258,
+      "learning_rate": 2e-05,
+      "loss": 0.7446324825286865,
+      "step": 800
+    },
+    {
+      "epoch": 0.10280492448280239,
+      "grad_norm": 1.1936376998091593,
+      "learning_rate": 2e-05,
+      "loss": 0.7654414653778077,
+      "step": 810
+    },
+    {
+      "epoch": 0.1040741210813555,
+      "grad_norm": 1.2135419498926854,
+      "learning_rate": 2e-05,
+      "loss": 0.7101289749145507,
+      "step": 820
+    },
+    {
+      "epoch": 0.10534331767990862,
+      "grad_norm": 1.9651816139447085,
+      "learning_rate": 2e-05,
+      "loss": 0.7488377571105957,
+      "step": 830
+    },
+    {
+      "epoch": 0.10661251427846173,
+      "grad_norm": 1.6974295850201375,
+      "learning_rate": 2e-05,
+      "loss": 0.7266185760498047,
+      "step": 840
+    },
+    {
+      "epoch": 0.10788171087701486,
+      "grad_norm": 1.1494939530042951,
+      "learning_rate": 2e-05,
+      "loss": 0.7002715587615966,
+      "step": 850
+    },
+    {
+      "epoch": 0.10915090747556797,
+      "grad_norm": 1.627649054989265,
+      "learning_rate": 2e-05,
+      "loss": 0.7101876735687256,
+      "step": 860
+    },
+    {
+      "epoch": 0.11042010407412108,
+      "grad_norm": 2.4673513413307586,
+      "learning_rate": 2e-05,
+      "loss": 0.7189567565917969,
+      "step": 870
+    },
+    {
+      "epoch": 0.1116893006726742,
+      "grad_norm": 1.8914407958898596,
+      "learning_rate": 2e-05,
+      "loss": 0.6732224941253662,
+      "step": 880
+    },
+    {
+      "epoch": 0.11295849727122731,
+      "grad_norm": 1.25737054869581,
+      "learning_rate": 2e-05,
+      "loss": 0.7188662528991699,
+      "step": 890
+    },
+    {
+      "epoch": 0.11422769386978043,
+      "grad_norm": 2.7970593339713377,
+      "learning_rate": 2e-05,
+      "loss": 0.768404483795166,
+      "step": 900
+    },
+    {
+      "epoch": 0.11549689046833354,
+      "grad_norm": 1.2100883151026398,
+      "learning_rate": 2e-05,
+      "loss": 0.7342836380004882,
+      "step": 910
+    },
+    {
+      "epoch": 0.11676608706688667,
+      "grad_norm": 1.7689464812633282,
+      "learning_rate": 2e-05,
+      "loss": 0.6932127952575684,
+      "step": 920
+    },
+    {
+      "epoch": 0.11803528366543978,
+      "grad_norm": 24.46312019806331,
+      "learning_rate": 2e-05,
+      "loss": 0.7131558418273926,
+      "step": 930
+    },
+    {
+      "epoch": 0.11930448026399289,
+      "grad_norm": 1.4401629868679464,
+      "learning_rate": 2e-05,
+      "loss": 0.7198721885681152,
+      "step": 940
+    },
+    {
+      "epoch": 0.12057367686254601,
+      "grad_norm": 2.8881340201173344,
+      "learning_rate": 2e-05,
+      "loss": 0.7238938808441162,
+      "step": 950
+    },
+    {
+      "epoch": 0.12184287346109912,
+      "grad_norm": 3.8069447232771423,
+      "learning_rate": 2e-05,
+      "loss": 0.735845947265625,
+      "step": 960
+    },
+    {
+      "epoch": 0.12311207005965225,
+      "grad_norm": 1.65977234932993,
+      "learning_rate": 2e-05,
+      "loss": 0.7196044445037841,
+      "step": 970
+    },
+    {
+      "epoch": 0.12438126665820536,
+      "grad_norm": 2.3582192388773278,
+      "learning_rate": 2e-05,
+      "loss": 0.7296237945556641,
+      "step": 980
+    },
+    {
+      "epoch": 0.12565046325675847,
+      "grad_norm": 1.3096772145164088,
+      "learning_rate": 2e-05,
+      "loss": 0.7319690704345703,
+      "step": 990
+    },
+    {
+      "epoch": 0.1269196598553116,
+      "grad_norm": 2.768378793994153,
+      "learning_rate": 2e-05,
+      "loss": 0.7171053886413574,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1281888564538647,
+      "grad_norm": 21.13021693144351,
+      "learning_rate": 2e-05,
+      "loss": 0.7210842132568359,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1294580530524178,
+      "grad_norm": 6.461783847400385,
+      "learning_rate": 2e-05,
+      "loss": 0.719035530090332,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13072724965097093,
+      "grad_norm": 2.668930292931086,
+      "learning_rate": 2e-05,
+      "loss": 0.7092215538024902,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13199644624952406,
+      "grad_norm": 1.6702243495228197,
+      "learning_rate": 2e-05,
+      "loss": 0.6995366096496582,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13326564284807715,
+      "grad_norm": 121.66250271016467,
+      "learning_rate": 2e-05,
+      "loss": 0.706237506866455,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13453483944663028,
+      "grad_norm": 2.267211319195803,
+      "learning_rate": 2e-05,
+      "loss": 0.7493322372436524,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1358040360451834,
+      "grad_norm": 6.368905095822087,
+      "learning_rate": 2e-05,
+      "loss": 0.769202709197998,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13707323264373653,
+      "grad_norm": 1.9611489768271684,
+      "learning_rate": 2e-05,
+      "loss": 0.7350125312805176,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13834242924228962,
+      "grad_norm": 16.708776429157798,
+      "learning_rate": 2e-05,
+      "loss": 0.7035937309265137,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13961162584084275,
+      "grad_norm": 11.46126539981295,
+      "learning_rate": 2e-05,
+      "loss": 0.7125218391418457,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14088082243939587,
+      "grad_norm": 12.23564157742202,
+      "learning_rate": 2e-05,
+      "loss": 0.7346056938171387,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14215001903794897,
+      "grad_norm": 10.994067160786324,
+      "learning_rate": 2e-05,
+      "loss": 0.7709558486938477,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1434192156365021,
+      "grad_norm": 1.6932275523600715,
+      "learning_rate": 2e-05,
+      "loss": 0.7574703216552734,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1446884122350552,
+      "grad_norm": 3.481606105035328,
+      "learning_rate": 2e-05,
+      "loss": 0.7068160533905029,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14595760883360834,
+      "grad_norm": 1.287350901683337,
+      "learning_rate": 2e-05,
+      "loss": 0.7185220718383789,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14722680543216143,
+      "grad_norm": 36.72744854503198,
+      "learning_rate": 2e-05,
+      "loss": 0.6944140434265137,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14849600203071456,
+      "grad_norm": 3.7813106070325135,
+      "learning_rate": 2e-05,
+      "loss": 0.7398875236511231,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14976519862926768,
+      "grad_norm": 2.480273963594583,
+      "learning_rate": 2e-05,
+      "loss": 0.6865447044372559,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15103439522782078,
+      "grad_norm": 4.445880813084278,
+      "learning_rate": 2e-05,
+      "loss": 0.7280641555786133,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1523035918263739,
+      "grad_norm": 5.239603018280949,
+      "learning_rate": 2e-05,
+      "loss": 0.7084729194641113,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15357278842492703,
+      "grad_norm": 13.609702001094654,
+      "learning_rate": 2e-05,
+      "loss": 0.7493104457855224,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15484198502348015,
+      "grad_norm": 2.7200161742937934,
+      "learning_rate": 2e-05,
+      "loss": 0.7335978507995605,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15611118162203325,
+      "grad_norm": 1.2145687940635885,
+      "learning_rate": 2e-05,
+      "loss": 0.7125667572021485,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15738037822058637,
+      "grad_norm": 1.2241463795106138,
+      "learning_rate": 2e-05,
+      "loss": 0.7185767650604248,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1586495748191395,
+      "grad_norm": 3.555885891493168,
+      "learning_rate": 2e-05,
+      "loss": 0.7141647338867188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1599187714176926,
+      "grad_norm": 1.9194540312284558,
+      "learning_rate": 2e-05,
+      "loss": 0.7142346858978271,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16118796801624571,
+      "grad_norm": 8.967819198009728,
+      "learning_rate": 2e-05,
+      "loss": 0.6929870128631592,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16245716461479884,
+      "grad_norm": 4.579908780898307,
+      "learning_rate": 2e-05,
+      "loss": 0.6866648674011231,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16372636121335196,
+      "grad_norm": 2.391259306915726,
+      "learning_rate": 2e-05,
+      "loss": 0.7132150173187256,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16499555781190506,
+      "grad_norm": 3.2435823893612716,
+      "learning_rate": 2e-05,
+      "loss": 0.7147534847259521,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16626475441045818,
+      "grad_norm": 9.967220102904227,
+      "learning_rate": 2e-05,
+      "loss": 0.7580214500427246,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1675339510090113,
+      "grad_norm": 2.032934180828253,
+      "learning_rate": 2e-05,
+      "loss": 0.7365469932556152,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1688031476075644,
+      "grad_norm": 4.482568717701066,
+      "learning_rate": 2e-05,
+      "loss": 0.7797821044921875,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17007234420611753,
+      "grad_norm": 9.705663878033436,
+      "learning_rate": 2e-05,
+      "loss": 0.7290232658386231,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17134154080467065,
+      "grad_norm": 4.830320252791453,
+      "learning_rate": 2e-05,
+      "loss": 0.6980529308319092,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17261073740322375,
+      "grad_norm": 5.712592100673355,
+      "learning_rate": 2e-05,
+      "loss": 0.7129548072814942,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17387993400177687,
+      "grad_norm": 3.6652260686016613,
+      "learning_rate": 2e-05,
+      "loss": 0.6861123085021973,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17514913060033,
+      "grad_norm": 1.830090338778074,
+      "learning_rate": 2e-05,
+      "loss": 0.7101086616516114,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17641832719888312,
+      "grad_norm": 3.2006859786261965,
+      "learning_rate": 2e-05,
+      "loss": 0.7196810722351075,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17768752379743621,
+      "grad_norm": 58.529642886881184,
+      "learning_rate": 2e-05,
+      "loss": 0.6799466609954834,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17895672039598934,
+      "grad_norm": 5.315650680940421,
+      "learning_rate": 2e-05,
+      "loss": 0.7009426116943359,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18022591699454246,
+      "grad_norm": 2.5968114372436206,
+      "learning_rate": 2e-05,
+      "loss": 0.7230406761169433,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18149511359309556,
+      "grad_norm": 20.342084989596863,
+      "learning_rate": 2e-05,
+      "loss": 0.7067214488983155,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18276431019164868,
+      "grad_norm": 1.5290124379496879,
+      "learning_rate": 2e-05,
+      "loss": 0.6898145198822021,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1840335067902018,
+      "grad_norm": 5.574739707168099,
+      "learning_rate": 2e-05,
+      "loss": 0.7231745719909668,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18530270338875493,
+      "grad_norm": 1.5001724313188949,
+      "learning_rate": 2e-05,
+      "loss": 0.6858955383300781,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18657189998730803,
+      "grad_norm": 1.700018550443666,
+      "learning_rate": 2e-05,
+      "loss": 0.705136775970459,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18784109658586115,
+      "grad_norm": 0.9469945674471103,
+      "learning_rate": 2e-05,
+      "loss": 0.6926360130310059,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18911029318441427,
+      "grad_norm": 1.3538378422387347,
+      "learning_rate": 2e-05,
+      "loss": 0.6977663993835449,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19037948978296737,
+      "grad_norm": 1.455383203281115,
+      "learning_rate": 2e-05,
+      "loss": 0.7223066329956055,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1916486863815205,
+      "grad_norm": 3.972199629523166,
+      "learning_rate": 2e-05,
+      "loss": 0.6782344341278076,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19291788298007362,
+      "grad_norm": 3.0969720182407725,
+      "learning_rate": 2e-05,
+      "loss": 0.7165674209594727,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19418707957862674,
+      "grad_norm": 3.0527900599029163,
+      "learning_rate": 2e-05,
+      "loss": 0.7175450325012207,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19545627617717984,
+      "grad_norm": 3.8341495567014436,
+      "learning_rate": 2e-05,
+      "loss": 0.723912525177002,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19672547277573296,
+      "grad_norm": 1.947508519994872,
+      "learning_rate": 2e-05,
+      "loss": 0.7383893966674805,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1979946693742861,
+      "grad_norm": 1.8462321700632978,
+      "learning_rate": 2e-05,
+      "loss": 0.7208988189697265,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19926386597283918,
+      "grad_norm": 1.7198903280777138,
+      "learning_rate": 2e-05,
+      "loss": 0.6724112510681153,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2005330625713923,
+      "grad_norm": 2.1405605175306124,
+      "learning_rate": 2e-05,
+      "loss": 0.7294898986816406,
+      "step": 1580
+    },
+    {
+      "epoch": 0.20180225916994543,
+      "grad_norm": 4.227746703868925,
+      "learning_rate": 2e-05,
+      "loss": 0.7287868022918701,
+      "step": 1590
+    },
+    {
+      "epoch": 0.20307145576849853,
+      "grad_norm": 9.768285261777095,
+      "learning_rate": 2e-05,
+      "loss": 0.7028857231140136,
+      "step": 1600
+    },
+    {
+      "epoch": 0.20434065236705165,
+      "grad_norm": 1.7792917952969463,
+      "learning_rate": 2e-05,
+      "loss": 0.7140511989593505,
+      "step": 1610
+    },
+    {
+      "epoch": 0.20560984896560477,
+      "grad_norm": 1.3040899219700945,
+      "learning_rate": 2e-05,
+      "loss": 0.7118062019348145,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2068790455641579,
+      "grad_norm": 1.7007055621957476,
+      "learning_rate": 2e-05,
+      "loss": 0.707486629486084,
+      "step": 1630
+    },
+    {
+      "epoch": 0.208148242162711,
+      "grad_norm": 1.8769292710864678,
+      "learning_rate": 2e-05,
+      "loss": 0.720918607711792,
+      "step": 1640
+    },
+    {
+      "epoch": 0.20941743876126412,
+      "grad_norm": 6.129630406825217,
+      "learning_rate": 2e-05,
+      "loss": 0.7186423301696777,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21068663535981724,
+      "grad_norm": 1.1703801257670128,
+      "learning_rate": 2e-05,
+      "loss": 0.7236575126647949,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21195583195837034,
+      "grad_norm": 2.70328737308837,
+      "learning_rate": 2e-05,
+      "loss": 0.6925942897796631,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21322502855692346,
+      "grad_norm": 6.676030371040207,
+      "learning_rate": 2e-05,
+      "loss": 0.6980341911315918,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2144942251554766,
+      "grad_norm": 1.9703840027872128,
+      "learning_rate": 2e-05,
+      "loss": 0.6997060298919677,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2157634217540297,
+      "grad_norm": 1.930151191536304,
+      "learning_rate": 2e-05,
+      "loss": 0.7096672058105469,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2170326183525828,
+      "grad_norm": 1.5971423989640974,
+      "learning_rate": 2e-05,
+      "loss": 0.6999159812927246,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21830181495113593,
+      "grad_norm": 1.2439067162544901,
+      "learning_rate": 2e-05,
+      "loss": 0.6927071571350097,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21957101154968905,
+      "grad_norm": 1.405834319587726,
+      "learning_rate": 2e-05,
+      "loss": 0.6896047115325927,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22084020814824215,
+      "grad_norm": 1.6450831477120642,
+      "learning_rate": 2e-05,
+      "loss": 0.6765172481536865,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22210940474679527,
+      "grad_norm": 31.152990133251816,
+      "learning_rate": 2e-05,
+      "loss": 0.7346570491790771,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2233786013453484,
+      "grad_norm": 2.20122696950214,
+      "learning_rate": 2e-05,
+      "loss": 0.7049872875213623,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22464779794390152,
+      "grad_norm": 1.8812044407136512,
+      "learning_rate": 2e-05,
+      "loss": 0.6898641109466552,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22591699454245462,
+      "grad_norm": 3.5344547403887256,
+      "learning_rate": 2e-05,
+      "loss": 0.7044347286224365,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22718619114100774,
+      "grad_norm": 1.0188535973596087,
+      "learning_rate": 2e-05,
+      "loss": 0.703303337097168,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22845538773956087,
+      "grad_norm": 1.3593555737841707,
+      "learning_rate": 2e-05,
+      "loss": 0.7034189224243164,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22972458433811396,
+      "grad_norm": 0.9623009318480035,
+      "learning_rate": 2e-05,
+      "loss": 0.7039454936981201,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2309937809366671,
+      "grad_norm": 1.3101226548361318,
+      "learning_rate": 2e-05,
+      "loss": 0.7383696556091308,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2322629775352202,
+      "grad_norm": 2.4893933260316006,
+      "learning_rate": 2e-05,
+      "loss": 0.6845382690429688,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23353217413377333,
+      "grad_norm": 2.4841465697734977,
+      "learning_rate": 2e-05,
+      "loss": 0.6853599071502685,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23480137073232643,
+      "grad_norm": 1.6581272802864346,
+      "learning_rate": 2e-05,
+      "loss": 0.7452334403991699,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23607056733087955,
+      "grad_norm": 1.37088665101461,
+      "learning_rate": 2e-05,
+      "loss": 0.7337687492370606,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23733976392943268,
+      "grad_norm": 1.3647513469209192,
+      "learning_rate": 2e-05,
+      "loss": 0.6960929870605469,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23860896052798578,
+      "grad_norm": 2.9263951391809275,
+      "learning_rate": 2e-05,
+      "loss": 0.706198263168335,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2398781571265389,
+      "grad_norm": 1.8982911221139078,
+      "learning_rate": 2e-05,
+      "loss": 0.6878005027770996,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24114735372509202,
+      "grad_norm": 2.6965000137847204,
+      "learning_rate": 2e-05,
+      "loss": 0.716560697555542,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24241655032364512,
+      "grad_norm": 2.27600674641753,
+      "learning_rate": 2e-05,
+      "loss": 0.6992623805999756,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24368574692219824,
+      "grad_norm": 7.658063725746519,
+      "learning_rate": 2e-05,
+      "loss": 0.6911978721618652,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24495494352075137,
+      "grad_norm": 2.3897175022207544,
+      "learning_rate": 2e-05,
+      "loss": 0.7327141761779785,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2462241401193045,
+      "grad_norm": 2.611194695101572,
+      "learning_rate": 2e-05,
+      "loss": 0.7019572257995605,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2474933367178576,
+      "grad_norm": 1.6903599382055103,
+      "learning_rate": 2e-05,
+      "loss": 0.6935071468353271,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2487625333164107,
+      "grad_norm": 2.112217002863749,
+      "learning_rate": 2e-05,
+      "loss": 0.7182205200195313,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2500317299149638,
+      "grad_norm": 1.6176341129959053,
+      "learning_rate": 2e-05,
+      "loss": 0.7326358795166016,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25130092651351693,
+      "grad_norm": 2.911543907127,
+      "learning_rate": 2e-05,
+      "loss": 0.7045173168182373,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25257012311207006,
+      "grad_norm": 4.355306061158963,
+      "learning_rate": 2e-05,
+      "loss": 0.6572384834289551,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2538393197106232,
+      "grad_norm": 1.4703323146292895,
+      "learning_rate": 2e-05,
+      "loss": 0.7193952560424804,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2551085163091763,
+      "grad_norm": 1.393791899012489,
+      "learning_rate": 2e-05,
+      "loss": 0.6960308074951171,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2563777129077294,
+      "grad_norm": 1.2070352349594737,
+      "learning_rate": 2e-05,
+      "loss": 0.7007707118988037,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2576469095062825,
+      "grad_norm": 3.308002263089678,
+      "learning_rate": 2e-05,
+      "loss": 0.6798537254333497,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2589161061048356,
+      "grad_norm": 1.7174636866260846,
+      "learning_rate": 2e-05,
+      "loss": 0.7235496520996094,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26018530270338874,
+      "grad_norm": 1.1019618883142333,
+      "learning_rate": 2e-05,
+      "loss": 0.6935445308685303,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26145449930194187,
+      "grad_norm": 2.263169001519526,
+      "learning_rate": 2e-05,
+      "loss": 0.7165854930877685,
+      "step": 2060
+    },
+    {
+      "epoch": 0.262723695900495,
+      "grad_norm": 3.579553888433581,
+      "learning_rate": 2e-05,
+      "loss": 0.6986983299255372,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2639928924990481,
+      "grad_norm": 1.4044134605617704,
+      "learning_rate": 2e-05,
+      "loss": 0.7135456085205079,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26526208909760124,
+      "grad_norm": 1.5099254645004807,
+      "learning_rate": 2e-05,
+      "loss": 0.6784690856933594,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2665312856961543,
+      "grad_norm": 4.764055855609546,
+      "learning_rate": 2e-05,
+      "loss": 0.6540004253387451,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26780048229470743,
+      "grad_norm": 2.05845014191107,
+      "learning_rate": 2e-05,
+      "loss": 0.6922139167785645,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26906967889326056,
+      "grad_norm": 4.623917747555655,
+      "learning_rate": 2e-05,
+      "loss": 0.6600709438323975,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2703388754918137,
+      "grad_norm": 1.6779219678721877,
+      "learning_rate": 2e-05,
+      "loss": 0.6826789855957032,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2716080720903668,
+      "grad_norm": 2.6153490839563305,
+      "learning_rate": 2e-05,
+      "loss": 0.6859559535980224,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2728772686889199,
+      "grad_norm": 1.0538264330816605,
+      "learning_rate": 2e-05,
+      "loss": 0.674953842163086,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27414646528747305,
+      "grad_norm": 1.5121055768325138,
+      "learning_rate": 2e-05,
+      "loss": 0.696351146697998,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2754156618860261,
+      "grad_norm": 0.9657635797707542,
+      "learning_rate": 2e-05,
+      "loss": 0.690283727645874,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27668485848457924,
+      "grad_norm": 1.0478341533882702,
+      "learning_rate": 2e-05,
+      "loss": 0.6650715827941894,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27795405508313237,
+      "grad_norm": 0.9276953664799223,
+      "learning_rate": 2e-05,
+      "loss": 0.6776557922363281,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2792232516816855,
+      "grad_norm": 1.1265445142605153,
+      "learning_rate": 2e-05,
+      "loss": 0.7132327079772949,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2804924482802386,
+      "grad_norm": 1.0835782433582077,
+      "learning_rate": 2e-05,
+      "loss": 0.7109849452972412,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28176164487879174,
+      "grad_norm": 3.5059431763261197,
+      "learning_rate": 2e-05,
+      "loss": 0.673506498336792,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28303084147734486,
+      "grad_norm": 1.4436123816094562,
+      "learning_rate": 2e-05,
+      "loss": 0.7144210338592529,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28430003807589793,
+      "grad_norm": 1.1354595553700741,
+      "learning_rate": 2e-05,
+      "loss": 0.7044991970062255,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28556923467445106,
+      "grad_norm": 2.7230622295989235,
+      "learning_rate": 2e-05,
+      "loss": 0.6636761665344239,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2868384312730042,
+      "grad_norm": 1.1694418398010178,
+      "learning_rate": 2e-05,
+      "loss": 0.6927776336669922,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2881076278715573,
+      "grad_norm": 1.5107851197045172,
+      "learning_rate": 2e-05,
+      "loss": 0.6771825313568115,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2893768244701104,
+      "grad_norm": 1.1621441645335486,
+      "learning_rate": 2e-05,
+      "loss": 0.6916715621948242,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29064602106866355,
+      "grad_norm": 6.434299294995004,
+      "learning_rate": 2e-05,
+      "loss": 0.6960276603698731,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2919152176672167,
+      "grad_norm": 1.385368773405726,
+      "learning_rate": 2e-05,
+      "loss": 0.6719958782196045,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29318441426576974,
+      "grad_norm": 1.475324341939836,
+      "learning_rate": 2e-05,
+      "loss": 0.6970046997070313,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29445361086432287,
+      "grad_norm": 1.5266438742509716,
+      "learning_rate": 2e-05,
+      "loss": 0.7041642665863037,
+      "step": 2320
+    },
+    {
+      "epoch": 0.295722807462876,
+      "grad_norm": 1.0521598059140391,
+      "learning_rate": 2e-05,
+      "loss": 0.6700316429138183,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2969920040614291,
+      "grad_norm": 3.329736804526033,
+      "learning_rate": 2e-05,
+      "loss": 0.6452198028564453,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29826120065998224,
+      "grad_norm": 0.8914849561576785,
+      "learning_rate": 2e-05,
+      "loss": 0.6761260509490967,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29953039725853536,
+      "grad_norm": 1.3013071683478892,
+      "learning_rate": 2e-05,
+      "loss": 0.7069483757019043,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3007995938570885,
+      "grad_norm": 1.1687388959814224,
+      "learning_rate": 2e-05,
+      "loss": 0.678680419921875,
+      "step": 2370
+    },
+    {
+      "epoch": 0.30206879045564156,
+      "grad_norm": 5.859527819940056,
+      "learning_rate": 2e-05,
+      "loss": 0.6810003280639648,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3033379870541947,
+      "grad_norm": 1.0473349930521416,
+      "learning_rate": 2e-05,
+      "loss": 0.6612431526184082,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3046071836527478,
+      "grad_norm": 1.7245826010161895,
+      "learning_rate": 2e-05,
+      "loss": 0.6682802200317383,
+      "step": 2400
+    },
+    {
+      "epoch": 0.30587638025130093,
+      "grad_norm": 1.1583809423320401,
+      "learning_rate": 2e-05,
+      "loss": 0.6659842491149902,
+      "step": 2410
+    },
+    {
+      "epoch": 0.30714557684985405,
+      "grad_norm": 1.1684708242768307,
+      "learning_rate": 2e-05,
+      "loss": 0.7101010799407959,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3084147734484072,
+      "grad_norm": 4.1292567683192445,
+      "learning_rate": 2e-05,
+      "loss": 0.7205910682678223,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3096839700469603,
+      "grad_norm": 1.5202070208893206,
+      "learning_rate": 2e-05,
+      "loss": 0.7077850341796875,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31095316664551337,
+      "grad_norm": 1.4698618199042586,
+      "learning_rate": 2e-05,
+      "loss": 0.7106716156005859,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3122223632440665,
+      "grad_norm": 1.3496778085998071,
+      "learning_rate": 2e-05,
+      "loss": 0.6891602516174317,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3134915598426196,
+      "grad_norm": 1.1329774594004545,
+      "learning_rate": 2e-05,
+      "loss": 0.6777769088745117,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31476075644117274,
+      "grad_norm": 1.6485738913799488,
+      "learning_rate": 2e-05,
+      "loss": 0.6918991088867188,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31602995303972586,
+      "grad_norm": 1.0719459285843456,
+      "learning_rate": 2e-05,
+      "loss": 0.6974703788757324,
+      "step": 2490
+    },
+    {
+      "epoch": 0.317299149638279,
+      "grad_norm": 3.1086933062760385,
+      "learning_rate": 2e-05,
+      "loss": 0.6644510269165039,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3185683462368321,
+      "grad_norm": 3.333155602980405,
+      "learning_rate": 2e-05,
+      "loss": 0.6809342861175537,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3198375428353852,
+      "grad_norm": 5.09675191350906,
+      "learning_rate": 2e-05,
+      "loss": 0.6938048362731933,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3211067394339383,
+      "grad_norm": 1.1926062071351522,
+      "learning_rate": 2e-05,
+      "loss": 0.6757203102111816,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32237593603249143,
+      "grad_norm": 1.3450080147153045,
+      "learning_rate": 2e-05,
+      "loss": 0.6961421966552734,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32364513263104455,
+      "grad_norm": 0.8805718583868404,
+      "learning_rate": 2e-05,
+      "loss": 0.6860201835632325,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3249143292295977,
+      "grad_norm": 1.7164579296686397,
+      "learning_rate": 2e-05,
+      "loss": 0.6963025569915772,
+      "step": 2560
+    },
+    {
+      "epoch": 0.3261835258281508,
+      "grad_norm": 2.9359824868976157,
+      "learning_rate": 2e-05,
+      "loss": 0.6743741035461426,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3274527224267039,
+      "grad_norm": 1.3714596548370839,
+      "learning_rate": 2e-05,
+      "loss": 0.6874894142150879,
+      "step": 2580
+    },
+    {
+      "epoch": 0.328721919025257,
+      "grad_norm": 1.2720754750449736,
+      "learning_rate": 2e-05,
+      "loss": 0.6704371452331543,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3299911156238101,
+      "grad_norm": 1.7654176749273565,
+      "learning_rate": 2e-05,
+      "loss": 0.6946074485778808,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33126031222236324,
+      "grad_norm": 2.0337505533665796,
+      "learning_rate": 2e-05,
+      "loss": 0.7077631950378418,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33252950882091636,
+      "grad_norm": 1.3901659488720537,
+      "learning_rate": 2e-05,
+      "loss": 0.6855005264282227,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3337987054194695,
+      "grad_norm": 1.079015131588736,
+      "learning_rate": 2e-05,
+      "loss": 0.692991828918457,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3350679020180226,
+      "grad_norm": 1.7984440075888581,
+      "learning_rate": 2e-05,
+      "loss": 0.6902318000793457,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3363370986165757,
+      "grad_norm": 0.9559881062010754,
+      "learning_rate": 2e-05,
+      "loss": 0.7011756896972656,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3376062952151288,
+      "grad_norm": 0.9148181231313194,
+      "learning_rate": 2e-05,
+      "loss": 0.7119962215423584,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33887549181368193,
+      "grad_norm": 0.8836447446795664,
+      "learning_rate": 2e-05,
+      "loss": 0.6828885555267334,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34014468841223505,
+      "grad_norm": 1.1652549359037911,
+      "learning_rate": 2e-05,
+      "loss": 0.7115286827087403,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3414138850107882,
+      "grad_norm": 1.0536924423547303,
+      "learning_rate": 2e-05,
+      "loss": 0.6770869731903076,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3426830816093413,
+      "grad_norm": 1.5685533485109324,
+      "learning_rate": 2e-05,
+      "loss": 0.6682273864746093,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3439522782078944,
+      "grad_norm": 7.558739469427474,
+      "learning_rate": 2e-05,
+      "loss": 0.6884564876556396,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3452214748064475,
+      "grad_norm": 2.0552499665410315,
+      "learning_rate": 2e-05,
+      "loss": 0.7139716625213623,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3464906714050006,
+      "grad_norm": 2.0606999610278227,
+      "learning_rate": 2e-05,
+      "loss": 0.709437894821167,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34775986800355374,
+      "grad_norm": 1.0854772520724743,
+      "learning_rate": 2e-05,
+      "loss": 0.7189342498779296,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34902906460210686,
+      "grad_norm": 2.878269739462293,
+      "learning_rate": 2e-05,
+      "loss": 0.666627311706543,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35029826120066,
+      "grad_norm": 0.8706798739593159,
+      "learning_rate": 2e-05,
+      "loss": 0.684064245223999,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3515674577992131,
+      "grad_norm": 3.6276987754270924,
+      "learning_rate": 2e-05,
+      "loss": 0.6672313690185547,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35283665439776624,
+      "grad_norm": 4.676065559417869,
+      "learning_rate": 2e-05,
+      "loss": 0.6832221508026123,
+      "step": 2780
+    },
+    {
+      "epoch": 0.3541058509963193,
+      "grad_norm": 1.2950325624067205,
+      "learning_rate": 2e-05,
+      "loss": 0.6508336544036866,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35537504759487243,
+      "grad_norm": 2.2002291538849397,
+      "learning_rate": 2e-05,
+      "loss": 0.6765498161315918,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35664424419342555,
+      "grad_norm": 4.487843593764293,
+      "learning_rate": 2e-05,
+      "loss": 0.6740169048309326,
+      "step": 2810
+    },
+    {
+      "epoch": 0.3579134407919787,
+      "grad_norm": 3.7491254660875843,
+      "learning_rate": 2e-05,
+      "loss": 0.7060698509216309,
+      "step": 2820
+    },
+    {
+      "epoch": 0.3591826373905318,
+      "grad_norm": 1.647193301159518,
+      "learning_rate": 2e-05,
+      "loss": 0.6926600933074951,
+      "step": 2830
+    },
+    {
+      "epoch": 0.3604518339890849,
+      "grad_norm": 1.1879874584999572,
+      "learning_rate": 2e-05,
+      "loss": 0.6904709815979004,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36172103058763805,
+      "grad_norm": 10.200424834542355,
+      "learning_rate": 2e-05,
+      "loss": 0.6667781352996827,
+      "step": 2850
+    },
+    {
+      "epoch": 0.3629902271861911,
+      "grad_norm": 10.869498876941115,
+      "learning_rate": 2e-05,
+      "loss": 0.7024850845336914,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36425942378474424,
+      "grad_norm": 2.063792639311769,
+      "learning_rate": 2e-05,
+      "loss": 0.6900119304656982,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36552862038329736,
+      "grad_norm": 1.8964268898112402,
+      "learning_rate": 2e-05,
+      "loss": 0.6867493629455567,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3667978169818505,
+      "grad_norm": 3.0265625846176083,
+      "learning_rate": 2e-05,
+      "loss": 0.6911677360534668,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3680670135804036,
+      "grad_norm": 1.6484545709122091,
+      "learning_rate": 2e-05,
+      "loss": 0.6593696117401123,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36933621017895674,
+      "grad_norm": 4.425382869217627,
+      "learning_rate": 2e-05,
+      "loss": 0.6771161079406738,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37060540677750986,
+      "grad_norm": 2.0482551068701955,
+      "learning_rate": 2e-05,
+      "loss": 0.6813270092010498,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37187460337606293,
+      "grad_norm": 0.9444837649644567,
+      "learning_rate": 2e-05,
+      "loss": 0.6603889465332031,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37314379997461605,
+      "grad_norm": 3.538295589794316,
+      "learning_rate": 2e-05,
+      "loss": 0.6703418731689453,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3744129965731692,
+      "grad_norm": 1.7846470661859684,
+      "learning_rate": 2e-05,
+      "loss": 0.6912987709045411,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3756821931717223,
+      "grad_norm": 1.009543729163236,
+      "learning_rate": 2e-05,
+      "loss": 0.6839200496673584,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3769513897702754,
+      "grad_norm": 1.100208023132109,
+      "learning_rate": 2e-05,
+      "loss": 0.7183089256286621,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37822058636882855,
+      "grad_norm": 1.097883082370207,
+      "learning_rate": 2e-05,
+      "loss": 0.7069741249084472,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3794897829673817,
+      "grad_norm": 0.9760377918691916,
+      "learning_rate": 2e-05,
+      "loss": 0.67308349609375,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38075897956593474,
+      "grad_norm": 3.886618836597359,
+      "learning_rate": 2e-05,
+      "loss": 0.68440260887146,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38202817616448786,
+      "grad_norm": 2.0480615813428313,
+      "learning_rate": 2e-05,
+      "loss": 0.6767269134521484,
+      "step": 3010
+    },
+    {
+      "epoch": 0.383297372763041,
+      "grad_norm": 2.7691044182546403,
+      "learning_rate": 2e-05,
+      "loss": 0.687132453918457,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3845665693615941,
+      "grad_norm": 1.4106810016844282,
+      "learning_rate": 2e-05,
+      "loss": 0.6832622051239013,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38583576596014724,
+      "grad_norm": 9.5717249257556,
+      "learning_rate": 2e-05,
+      "loss": 0.6606577873229981,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38710496255870036,
+      "grad_norm": 1.0827375398062002,
+      "learning_rate": 2e-05,
+      "loss": 0.6780203819274903,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3883741591572535,
+      "grad_norm": 1.1914725392078453,
+      "learning_rate": 2e-05,
+      "loss": 0.6991460800170899,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38964335575580655,
+      "grad_norm": 1.5671263365843944,
+      "learning_rate": 2e-05,
+      "loss": 0.6901996612548829,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3909125523543597,
+      "grad_norm": 3.1252728473876594,
+      "learning_rate": 2e-05,
+      "loss": 0.6899737358093262,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3921817489529128,
+      "grad_norm": 8.074692348088018,
+      "learning_rate": 2e-05,
+      "loss": 0.6660305976867675,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3934509455514659,
+      "grad_norm": 1.2643158691705767,
+      "learning_rate": 2e-05,
+      "loss": 0.6814273834228516,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39472014215001905,
+      "grad_norm": 2.0587419337769925,
+      "learning_rate": 2e-05,
+      "loss": 0.6974873542785645,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3959893387485722,
+      "grad_norm": 1.9893594820305398,
+      "learning_rate": 2e-05,
+      "loss": 0.6941835403442382,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3972585353471253,
+      "grad_norm": 3.470588344579962,
+      "learning_rate": 2e-05,
+      "loss": 0.6767202377319336,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39852773194567837,
+      "grad_norm": 1.143477576825162,
+      "learning_rate": 2e-05,
+      "loss": 0.6489578247070312,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3997969285442315,
+      "grad_norm": 2.16894018021262,
+      "learning_rate": 2e-05,
+      "loss": 0.670732069015503,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4010661251427846,
+      "grad_norm": 1.5063741733253424,
+      "learning_rate": 2e-05,
+      "loss": 0.7047544002532959,
+      "step": 3160
+    },
+    {
+      "epoch": 0.40233532174133774,
+      "grad_norm": 1.0404996145221876,
+      "learning_rate": 2e-05,
+      "loss": 0.6804914474487305,
+      "step": 3170
+    },
+    {
+      "epoch": 0.40360451833989086,
+      "grad_norm": 0.9914769320122442,
+      "learning_rate": 2e-05,
+      "loss": 0.6699023246765137,
+      "step": 3180
+    },
+    {
+      "epoch": 0.404873714938444,
+      "grad_norm": 6.255494273615712,
+      "learning_rate": 2e-05,
+      "loss": 0.6754027843475342,
+      "step": 3190
+    },
+    {
+      "epoch": 0.40614291153699705,
+      "grad_norm": 0.9755628914259246,
+      "learning_rate": 2e-05,
+      "loss": 0.6523271083831788,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4074121081355502,
+      "grad_norm": 2.8056587226083383,
+      "learning_rate": 2e-05,
+      "loss": 0.66622953414917,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4086813047341033,
+      "grad_norm": 1.1078944565652615,
+      "learning_rate": 2e-05,
+      "loss": 0.7013818264007569,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4099505013326564,
+      "grad_norm": 1.2995178890788006,
+      "learning_rate": 2e-05,
+      "loss": 0.679383373260498,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41121969793120955,
+      "grad_norm": 0.9732937999250203,
+      "learning_rate": 2e-05,
+      "loss": 0.6548215389251709,
+      "step": 3240
+    },
+    {
+      "epoch": 0.4124888945297627,
+      "grad_norm": 0.8280904001656607,
+      "learning_rate": 2e-05,
+      "loss": 0.6616822719573975,
+      "step": 3250
+    },
+    {
+      "epoch": 0.4137580911283158,
+      "grad_norm": 1.0132115720739858,
+      "learning_rate": 2e-05,
+      "loss": 0.6739038467407227,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41502728772686887,
+      "grad_norm": 1.4275553990453107,
+      "learning_rate": 2e-05,
+      "loss": 0.6831706047058106,
+      "step": 3270
+    },
+    {
+      "epoch": 0.416296484325422,
+      "grad_norm": 1.11366062485083,
+      "learning_rate": 2e-05,
+      "loss": 0.6794371604919434,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4175656809239751,
+      "grad_norm": 1.6367363566145334,
+      "learning_rate": 2e-05,
+      "loss": 0.667444372177124,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41883487752252824,
+      "grad_norm": 1.1259273791563325,
+      "learning_rate": 2e-05,
+      "loss": 0.6491257667541503,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42010407412108136,
+      "grad_norm": 1.3225045372977915,
+      "learning_rate": 2e-05,
+      "loss": 0.6859857082366944,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4213732707196345,
+      "grad_norm": 1.0263610829433494,
+      "learning_rate": 2e-05,
+      "loss": 0.675250768661499,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4226424673181876,
+      "grad_norm": 1.112060132258793,
+      "learning_rate": 2e-05,
+      "loss": 0.669788122177124,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4239116639167407,
+      "grad_norm": 0.937871889847132,
+      "learning_rate": 2e-05,
+      "loss": 0.6511672973632813,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4251808605152938,
+      "grad_norm": 1.3083411221827117,
+      "learning_rate": 2e-05,
+      "loss": 0.6796665191650391,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4264500571138469,
+      "grad_norm": 1.2420989378323404,
+      "learning_rate": 2e-05,
+      "loss": 0.6779007911682129,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42771925371240005,
+      "grad_norm": 0.9463317832348107,
+      "learning_rate": 2e-05,
+      "loss": 0.666592788696289,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4289884503109532,
+      "grad_norm": 9.631116855120194,
+      "learning_rate": 2e-05,
+      "loss": 0.7053398132324219,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4302576469095063,
+      "grad_norm": 1.9879068252592467,
+      "learning_rate": 2e-05,
+      "loss": 0.665035629272461,
+      "step": 3390
+    },
+    {
+      "epoch": 0.4315268435080594,
+      "grad_norm": 1.7073658842152728,
+      "learning_rate": 2e-05,
+      "loss": 0.6601058006286621,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4327960401066125,
+      "grad_norm": 0.9325694936991373,
+      "learning_rate": 2e-05,
+      "loss": 0.6808363914489746,
+      "step": 3410
+    },
+    {
+      "epoch": 0.4340652367051656,
+      "grad_norm": 0.9299194906404906,
+      "learning_rate": 2e-05,
+      "loss": 0.6839866161346435,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43533443330371874,
+      "grad_norm": 3.1325600884803944,
+      "learning_rate": 2e-05,
+      "loss": 0.6725717067718506,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43660362990227186,
+      "grad_norm": 1.360113372186893,
+      "learning_rate": 2e-05,
+      "loss": 0.6823818206787109,
+      "step": 3440
+    },
+    {
+      "epoch": 0.437872826500825,
+      "grad_norm": 1.8055815044982761,
+      "learning_rate": 2e-05,
+      "loss": 0.6350967407226562,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4391420230993781,
+      "grad_norm": 0.9190673776440563,
+      "learning_rate": 2e-05,
+      "loss": 0.6603227138519288,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44041121969793123,
+      "grad_norm": 1.1527631097976916,
+      "learning_rate": 2e-05,
+      "loss": 0.6740958213806152,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4416804162964843,
+      "grad_norm": 1.113729722191755,
+      "learning_rate": 2e-05,
+      "loss": 0.6700681209564209,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4429496128950374,
+      "grad_norm": 0.9331153462168471,
+      "learning_rate": 2e-05,
+      "loss": 0.6802886009216309,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44421880949359055,
+      "grad_norm": 1.8014819327891058,
+      "learning_rate": 2e-05,
+      "loss": 0.6852154731750488,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4454880060921437,
+      "grad_norm": 1.8544824676929599,
+      "learning_rate": 2e-05,
+      "loss": 0.6642695426940918,
+      "step": 3510
+    },
+    {
+      "epoch": 0.4467572026906968,
+      "grad_norm": 8.241901615016506,
+      "learning_rate": 2e-05,
+      "loss": 0.6711222648620605,
+      "step": 3520
+    },
+    {
+      "epoch": 0.4480263992892499,
+      "grad_norm": 1.0216875256184714,
+      "learning_rate": 2e-05,
+      "loss": 0.6691722869873047,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44929559588780305,
+      "grad_norm": 1.0332395098861513,
+      "learning_rate": 2e-05,
+      "loss": 0.6635716438293457,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4505647924863561,
+      "grad_norm": 0.8169787053838757,
+      "learning_rate": 2e-05,
+      "loss": 0.6858432769775391,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45183398908490924,
+      "grad_norm": 1.1963729168466504,
+      "learning_rate": 2e-05,
+      "loss": 0.6611463546752929,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45310318568346236,
+      "grad_norm": 2.899041698105585,
+      "learning_rate": 2e-05,
+      "loss": 0.6728678703308105,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4543723822820155,
+      "grad_norm": 1.4447055426447943,
+      "learning_rate": 2e-05,
+      "loss": 0.6572350978851318,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4556415788805686,
+      "grad_norm": 1.2257263469590438,
+      "learning_rate": 2e-05,
+      "loss": 0.660037899017334,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45691077547912173,
+      "grad_norm": 0.8875832767608117,
+      "learning_rate": 2e-05,
+      "loss": 0.6560229301452637,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45817997207767486,
+      "grad_norm": 1.3834398976583584,
+      "learning_rate": 2e-05,
+      "loss": 0.6705333709716796,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4594491686762279,
+      "grad_norm": 0.8924400055799993,
+      "learning_rate": 2e-05,
+      "loss": 0.6631116390228271,
+      "step": 3620
+    },
+    {
+      "epoch": 0.46071836527478105,
+      "grad_norm": 0.9622066909037335,
+      "learning_rate": 2e-05,
+      "loss": 0.6677485466003418,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4619875618733342,
+      "grad_norm": 0.8370602505716267,
+      "learning_rate": 2e-05,
+      "loss": 0.6631430625915528,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4632567584718873,
+      "grad_norm": 1.1510116221975755,
+      "learning_rate": 2e-05,
+      "loss": 0.6609024047851563,
+      "step": 3650
+    },
+    {
+      "epoch": 0.4645259550704404,
+      "grad_norm": 0.9547789996858147,
+      "learning_rate": 2e-05,
+      "loss": 0.6723182201385498,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46579515166899355,
+      "grad_norm": 0.8774471716110702,
+      "learning_rate": 2e-05,
+      "loss": 0.6970244407653808,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46706434826754667,
+      "grad_norm": 0.9391226961296547,
+      "learning_rate": 2e-05,
+      "loss": 0.6631155967712402,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46833354486609974,
+      "grad_norm": 0.7199333948971468,
+      "learning_rate": 2e-05,
+      "loss": 0.6857652187347412,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46960274146465286,
+      "grad_norm": 1.697734165889419,
+      "learning_rate": 2e-05,
+      "loss": 0.6637660026550293,
+      "step": 3700
+    },
+    {
+      "epoch": 0.470871938063206,
+      "grad_norm": 2.000484156217958,
+      "learning_rate": 2e-05,
+      "loss": 0.6643031597137451,
+      "step": 3710
+    },
+    {
+      "epoch": 0.4721411346617591,
+      "grad_norm": 1.0394751963617215,
+      "learning_rate": 2e-05,
+      "loss": 0.6828313827514648,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47341033126031223,
+      "grad_norm": 0.8220340848360204,
+      "learning_rate": 2e-05,
+      "loss": 0.6378774642944336,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47467952785886536,
+      "grad_norm": 0.9174739032624212,
+      "learning_rate": 2e-05,
+      "loss": 0.6632115840911865,
+      "step": 3740
+    },
+    {
+      "epoch": 0.4759487244574185,
+      "grad_norm": 1.0379837728884982,
+      "learning_rate": 2e-05,
+      "loss": 0.6518317222595215,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47721792105597155,
+      "grad_norm": 1.2314900719783382,
+      "learning_rate": 2e-05,
+      "loss": 0.6436082363128662,
+      "step": 3760
+    },
+    {
+      "epoch": 0.4784871176545247,
+      "grad_norm": 1.3417102898250968,
+      "learning_rate": 2e-05,
+      "loss": 0.704498291015625,
+      "step": 3770
+    },
+    {
+      "epoch": 0.4797563142530778,
+      "grad_norm": 0.9374545647678044,
+      "learning_rate": 2e-05,
+      "loss": 0.6673612594604492,
+      "step": 3780
+    },
+    {
+      "epoch": 0.4810255108516309,
+      "grad_norm": 0.8985159261033764,
+      "learning_rate": 2e-05,
+      "loss": 0.6671756744384766,
+      "step": 3790
+    },
+    {
+      "epoch": 0.48229470745018405,
+      "grad_norm": 0.8035375350741052,
+      "learning_rate": 2e-05,
+      "loss": 0.6436275482177735,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48356390404873717,
+      "grad_norm": 1.0224411054924063,
+      "learning_rate": 2e-05,
+      "loss": 0.6502698421478271,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48483310064729024,
+      "grad_norm": 0.7405550083278435,
+      "learning_rate": 2e-05,
+      "loss": 0.636743688583374,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48610229724584336,
+      "grad_norm": 0.9001682824014395,
+      "learning_rate": 2e-05,
+      "loss": 0.6850669860839844,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4873714938443965,
+      "grad_norm": 0.9370390606737176,
+      "learning_rate": 2e-05,
+      "loss": 0.6699307441711426,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4886406904429496,
+      "grad_norm": 0.8180283253221624,
+      "learning_rate": 2e-05,
+      "loss": 0.7221845149993896,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48990988704150273,
+      "grad_norm": 0.7708892842834187,
+      "learning_rate": 2e-05,
+      "loss": 0.6508908748626709,
+      "step": 3860
+    },
+    {
+      "epoch": 0.49117908364005586,
+      "grad_norm": 0.9547707513578754,
+      "learning_rate": 2e-05,
+      "loss": 0.6534225463867187,
+      "step": 3870
+    },
+    {
+      "epoch": 0.492448280238609,
+      "grad_norm": 0.8008576324520927,
+      "learning_rate": 2e-05,
+      "loss": 0.6713773727416992,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49371747683716205,
+      "grad_norm": 0.8054868068052583,
+      "learning_rate": 2e-05,
+      "loss": 0.6752046585083008,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4949866734357152,
+      "grad_norm": 0.8689792716922524,
+      "learning_rate": 2e-05,
+      "loss": 0.6691486358642578,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4962558700342683,
+      "grad_norm": 0.8155272544750918,
+      "learning_rate": 2e-05,
+      "loss": 0.6512347221374511,
+      "step": 3910
+    },
+    {
+      "epoch": 0.4975250666328214,
+      "grad_norm": 1.5885438014726774,
+      "learning_rate": 2e-05,
+      "loss": 0.6545228004455567,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49879426323137455,
+      "grad_norm": 0.7890400931686108,
+      "learning_rate": 2e-05,
+      "loss": 0.6719659328460693,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5000634598299276,
+      "grad_norm": 0.8813337160304279,
+      "learning_rate": 2e-05,
+      "loss": 0.6548175811767578,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5013326564284808,
+      "grad_norm": 1.0117720207674732,
+      "learning_rate": 2e-05,
+      "loss": 0.6846891403198242,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5026018530270339,
+      "grad_norm": 2.9748774846520587,
+      "learning_rate": 2e-05,
+      "loss": 0.6784535884857178,
+      "step": 3960
+    },
+    {
+      "epoch": 0.503871049625587,
+      "grad_norm": 1.145991218292352,
+      "learning_rate": 2e-05,
+      "loss": 0.6626908779144287,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5051402462241401,
+      "grad_norm": 1.9424593758227897,
+      "learning_rate": 2e-05,
+      "loss": 0.6995798110961914,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5064094428226933,
+      "grad_norm": 1.1991268648858209,
+      "learning_rate": 2e-05,
+      "loss": 0.6744263648986817,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5076786394212464,
+      "grad_norm": 0.9109917902573351,
+      "learning_rate": 2e-05,
+      "loss": 0.6499448299407959,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5089478360197994,
+      "grad_norm": 1.5842944011008318,
+      "learning_rate": 2e-05,
+      "loss": 0.6573305130004883,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5102170326183526,
+      "grad_norm": 1.3618202880490566,
+      "learning_rate": 2e-05,
+      "loss": 0.6500113487243653,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5114862292169057,
+      "grad_norm": 1.7444044072911193,
+      "learning_rate": 2e-05,
+      "loss": 0.7040845870971679,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5127554258154589,
+      "grad_norm": 0.816133645382111,
+      "learning_rate": 2e-05,
+      "loss": 0.6877236843109131,
+      "step": 4040
+    },
+    {
+      "epoch": 0.5140246224140119,
+      "grad_norm": 1.0958242132353944,
+      "learning_rate": 2e-05,
+      "loss": 0.6477513790130616,
+      "step": 4050
+    },
+    {
+      "epoch": 0.515293819012565,
+      "grad_norm": 0.9205865303270933,
+      "learning_rate": 2e-05,
+      "loss": 0.6891566276550293,
+      "step": 4060
+    },
+    {
+      "epoch": 0.5165630156111182,
+      "grad_norm": 2.059334992064026,
+      "learning_rate": 2e-05,
+      "loss": 0.6567224025726318,
+      "step": 4070
+    },
+    {
+      "epoch": 0.5178322122096712,
+      "grad_norm": 0.8171192633422016,
+      "learning_rate": 2e-05,
+      "loss": 0.6940532207489014,
+      "step": 4080
+    },
+    {
+      "epoch": 0.5191014088082244,
+      "grad_norm": 0.9126128831032772,
+      "learning_rate": 2e-05,
+      "loss": 0.6564689636230469,
+      "step": 4090
+    },
+    {
+      "epoch": 0.5203706054067775,
+      "grad_norm": 1.2171292958883924,
+      "learning_rate": 2e-05,
+      "loss": 0.6521645545959472,
+      "step": 4100
+    },
+    {
+      "epoch": 0.5216398020053307,
+      "grad_norm": 0.9805561424977111,
+      "learning_rate": 2e-05,
+      "loss": 0.6488425254821777,
+      "step": 4110
+    },
+    {
+      "epoch": 0.5229089986038837,
+      "grad_norm": 3.77511750075388,
+      "learning_rate": 2e-05,
+      "loss": 0.6658992767333984,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5241781952024368,
+      "grad_norm": 1.7930444706846342,
+      "learning_rate": 2e-05,
+      "loss": 0.6585663795471192,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52544739180099,
+      "grad_norm": 1.344889286001087,
+      "learning_rate": 2e-05,
+      "loss": 0.6788478851318359,
+      "step": 4140
+    },
+    {
+      "epoch": 0.526716588399543,
+      "grad_norm": 2.34241456575822,
+      "learning_rate": 2e-05,
+      "loss": 0.6609466552734375,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5279857849980962,
+      "grad_norm": 4.9413231840419565,
+      "learning_rate": 2e-05,
+      "loss": 0.6757864952087402,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5292549815966493,
+      "grad_norm": 2.007866114527804,
+      "learning_rate": 2e-05,
+      "loss": 0.6651741027832031,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5305241781952025,
+      "grad_norm": 9.239137436220071,
+      "learning_rate": 2e-05,
+      "loss": 0.6791013240814209,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5317933747937555,
+      "grad_norm": 1.9216636205089819,
+      "learning_rate": 2e-05,
+      "loss": 0.6357375144958496,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5330625713923086,
+      "grad_norm": 1.0243815958436338,
+      "learning_rate": 2e-05,
+      "loss": 0.6532147884368896,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5343317679908618,
+      "grad_norm": 1.7309692238091086,
+      "learning_rate": 2e-05,
+      "loss": 0.6577208518981934,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5356009645894149,
+      "grad_norm": 2.4699523230891836,
+      "learning_rate": 2e-05,
+      "loss": 0.6561010360717774,
+      "step": 4220
+    },
+    {
+      "epoch": 0.536870161187968,
+      "grad_norm": 0.8496332932410408,
+      "learning_rate": 2e-05,
+      "loss": 0.631166410446167,
+      "step": 4230
+    },
+    {
+      "epoch": 0.5381393577865211,
+      "grad_norm": 1.4382704853165937,
+      "learning_rate": 2e-05,
+      "loss": 0.6738145351409912,
+      "step": 4240
+    },
+    {
+      "epoch": 0.5394085543850743,
+      "grad_norm": 2.087680426024538,
+      "learning_rate": 2e-05,
+      "loss": 0.6434807777404785,
+      "step": 4250
+    },
+    {
+      "epoch": 0.5406777509836274,
+      "grad_norm": 1.1533761937772942,
+      "learning_rate": 2e-05,
+      "loss": 0.6724715232849121,
+      "step": 4260
+    },
+    {
+      "epoch": 0.5419469475821804,
+      "grad_norm": 1.5241042801664644,
+      "learning_rate": 2e-05,
+      "loss": 0.6800670146942138,
+      "step": 4270
+    },
+    {
+      "epoch": 0.5432161441807336,
+      "grad_norm": 1.3033593373925005,
+      "learning_rate": 2e-05,
+      "loss": 0.6184725761413574,
+      "step": 4280
+    },
+    {
+      "epoch": 0.5444853407792867,
+      "grad_norm": 1.8425667317078225,
+      "learning_rate": 2e-05,
+      "loss": 0.6519538879394531,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5457545373778399,
+      "grad_norm": 1.0139101215462223,
+      "learning_rate": 2e-05,
+      "loss": 0.641317892074585,
+      "step": 4300
+    },
+    {
+      "epoch": 0.5470237339763929,
+      "grad_norm": 2.2376742895884814,
+      "learning_rate": 2e-05,
+      "loss": 0.6605411529541015,
+      "step": 4310
+    },
+    {
+      "epoch": 0.5482929305749461,
+      "grad_norm": 0.9111448146973758,
+      "learning_rate": 2e-05,
+      "loss": 0.6893904685974122,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5495621271734992,
+      "grad_norm": 0.996917610485439,
+      "learning_rate": 2e-05,
+      "loss": 0.6462752342224121,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5508313237720522,
+      "grad_norm": 1.1928612171629684,
+      "learning_rate": 2e-05,
+      "loss": 0.6619662284851074,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5521005203706054,
+      "grad_norm": 2.676822342541805,
+      "learning_rate": 2e-05,
+      "loss": 0.6752418518066406,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5533697169691585,
+      "grad_norm": 1.5716874322915775,
+      "learning_rate": 2e-05,
+      "loss": 0.6672055721282959,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5546389135677117,
+      "grad_norm": 1.3677400410583507,
+      "learning_rate": 2e-05,
+      "loss": 0.6644375801086426,
+      "step": 4370
+    },
+    {
+      "epoch": 0.5559081101662647,
+      "grad_norm": 2.7361439067249154,
+      "learning_rate": 2e-05,
+      "loss": 0.6488841056823731,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5571773067648179,
+      "grad_norm": 1.2494001672145691,
+      "learning_rate": 2e-05,
+      "loss": 0.6313636302947998,
+      "step": 4390
+    },
+    {
+      "epoch": 0.558446503363371,
+      "grad_norm": 2.690631001054337,
+      "learning_rate": 2e-05,
+      "loss": 0.6764656066894531,
+      "step": 4400
+    },
+    {
+      "epoch": 0.559715699961924,
+      "grad_norm": 1.5487911266181829,
+      "learning_rate": 2e-05,
+      "loss": 0.650812292098999,
+      "step": 4410
+    },
+    {
+      "epoch": 0.5609848965604772,
+      "grad_norm": 1.0520374663779364,
+      "learning_rate": 2e-05,
+      "loss": 0.6898282527923584,
+      "step": 4420
+    },
+    {
+      "epoch": 0.5622540931590303,
+      "grad_norm": 1.7368840136155566,
+      "learning_rate": 2e-05,
+      "loss": 0.6611525535583496,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5635232897575835,
+      "grad_norm": 3.619439614441826,
+      "learning_rate": 2e-05,
+      "loss": 0.6676709175109863,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5647924863561365,
+      "grad_norm": 1.0113395680023953,
+      "learning_rate": 2e-05,
+      "loss": 0.6662263870239258,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5660616829546897,
+      "grad_norm": 4.406971073036479,
+      "learning_rate": 2e-05,
+      "loss": 0.673825740814209,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5673308795532428,
+      "grad_norm": 0.9890001393360089,
+      "learning_rate": 2e-05,
+      "loss": 0.655239486694336,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5686000761517959,
+      "grad_norm": 0.9564732140436093,
+      "learning_rate": 2e-05,
+      "loss": 0.672548246383667,
+      "step": 4480
+    },
+    {
+      "epoch": 0.569869272750349,
+      "grad_norm": 0.8095435554791974,
+      "learning_rate": 2e-05,
+      "loss": 0.6666032791137695,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5711384693489021,
+      "grad_norm": 1.955650431556263,
+      "learning_rate": 2e-05,
+      "loss": 0.6729245662689209,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5724076659474553,
+      "grad_norm": 2.223437960301981,
+      "learning_rate": 2e-05,
+      "loss": 0.66112380027771,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5736768625460084,
+      "grad_norm": 3.502929060626195,
+      "learning_rate": 2e-05,
+      "loss": 0.6619402885437011,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5749460591445615,
+      "grad_norm": 10.126664742345836,
+      "learning_rate": 2e-05,
+      "loss": 0.6840590953826904,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5762152557431146,
+      "grad_norm": 0.8555577132581583,
+      "learning_rate": 2e-05,
+      "loss": 0.6468152046203614,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5774844523416677,
+      "grad_norm": 0.8342677923106495,
+      "learning_rate": 2e-05,
+      "loss": 0.6477110862731934,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5787536489402209,
+      "grad_norm": 0.9861809968676206,
+      "learning_rate": 2e-05,
+      "loss": 0.6491549491882325,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5800228455387739,
+      "grad_norm": 1.5031982318481734,
+      "learning_rate": 2e-05,
+      "loss": 0.6709595680236816,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5812920421373271,
+      "grad_norm": 1.1756867668243183,
+      "learning_rate": 2e-05,
+      "loss": 0.6685933589935302,
+      "step": 4580
+    },
+    {
+      "epoch": 0.5825612387358802,
+      "grad_norm": 0.9010771098533525,
+      "learning_rate": 2e-05,
+      "loss": 0.6743881225585937,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5838304353344334,
+      "grad_norm": 1.4174056682438874,
+      "learning_rate": 2e-05,
+      "loss": 0.6973422050476075,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5850996319329864,
+      "grad_norm": 1.3520825847501865,
+      "learning_rate": 2e-05,
+      "loss": 0.6840449333190918,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5863688285315395,
+      "grad_norm": 1.01165447942195,
+      "learning_rate": 2e-05,
+      "loss": 0.6610830307006836,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5876380251300927,
+      "grad_norm": 1.842481227557108,
+      "learning_rate": 2e-05,
+      "loss": 0.6683774471282959,
+      "step": 4630
+    },
+    {
+      "epoch": 0.5889072217286457,
+      "grad_norm": 1.0852101191116865,
+      "learning_rate": 2e-05,
+      "loss": 0.6860476493835449,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5901764183271989,
+      "grad_norm": 1.8751724380099173,
+      "learning_rate": 2e-05,
+      "loss": 0.6589637756347656,
+      "step": 4650
+    },
+    {
+      "epoch": 0.591445614925752,
+      "grad_norm": 1.3447510319326075,
+      "learning_rate": 2e-05,
+      "loss": 0.6443521499633789,
+      "step": 4660
+    },
+    {
+      "epoch": 0.5927148115243052,
+      "grad_norm": 1.0705938674689184,
+      "learning_rate": 2e-05,
+      "loss": 0.6436320304870605,
+      "step": 4670
+    },
+    {
+      "epoch": 0.5939840081228582,
+      "grad_norm": 2.0447881612880545,
+      "learning_rate": 2e-05,
+      "loss": 0.6901763439178467,
+      "step": 4680
+    },
+    {
+      "epoch": 0.5952532047214113,
+      "grad_norm": 0.8322329443522019,
+      "learning_rate": 2e-05,
+      "loss": 0.6353757381439209,
+      "step": 4690
+    },
+    {
+      "epoch": 0.5965224013199645,
+      "grad_norm": 0.8708390424578789,
+      "learning_rate": 2e-05,
+      "loss": 0.6609786987304688,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5977915979185175,
+      "grad_norm": 1.1811929772368541,
+      "learning_rate": 2e-05,
+      "loss": 0.6537248611450195,
+      "step": 4710
+    },
+    {
+      "epoch": 0.5990607945170707,
+      "grad_norm": 0.9017640480151989,
+      "learning_rate": 2e-05,
+      "loss": 0.6676509380340576,
+      "step": 4720
+    },
+    {
+      "epoch": 0.6003299911156238,
+      "grad_norm": 1.452169053324432,
+      "learning_rate": 2e-05,
+      "loss": 0.6839349269866943,
+      "step": 4730
+    },
+    {
+      "epoch": 0.601599187714177,
+      "grad_norm": 0.83661626737386,
+      "learning_rate": 2e-05,
+      "loss": 0.6576152324676514,
+      "step": 4740
+    },
+    {
+      "epoch": 0.60286838431273,
+      "grad_norm": 1.2660753235191882,
+      "learning_rate": 2e-05,
+      "loss": 0.6698710918426514,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6041375809112831,
+      "grad_norm": 2.20208086386603,
+      "learning_rate": 2e-05,
+      "loss": 0.6689995765686035,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6054067775098363,
+      "grad_norm": 1.6234935426146342,
+      "learning_rate": 2e-05,
+      "loss": 0.6765145778656005,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6066759741083894,
+      "grad_norm": 1.5243956677912458,
+      "learning_rate": 2e-05,
+      "loss": 0.6844106197357178,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6079451707069425,
+      "grad_norm": 1.205662604884683,
+      "learning_rate": 2e-05,
+      "loss": 0.6594111919403076,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6092143673054956,
+      "grad_norm": 1.4227986311713559,
+      "learning_rate": 2e-05,
+      "loss": 0.6675002574920654,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6104835639040488,
+      "grad_norm": 0.959006030800747,
+      "learning_rate": 2e-05,
+      "loss": 0.6699732780456543,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6117527605026019,
+      "grad_norm": 1.9544658607102774,
+      "learning_rate": 2e-05,
+      "loss": 0.6544156551361084,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6130219571011549,
+      "grad_norm": 0.8332610012650475,
+      "learning_rate": 2e-05,
+      "loss": 0.6524645805358886,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6142911536997081,
+      "grad_norm": 0.9317501525601399,
+      "learning_rate": 2e-05,
+      "loss": 0.668744421005249,
+      "step": 4840
+    },
+    {
+      "epoch": 0.6155603502982612,
+      "grad_norm": 0.9406917134675316,
+      "learning_rate": 2e-05,
+      "loss": 0.6673583984375,
+      "step": 4850
+    },
+    {
+      "epoch": 0.6168295468968144,
+      "grad_norm": 1.0941740332752838,
+      "learning_rate": 2e-05,
+      "loss": 0.6725406169891357,
+      "step": 4860
+    },
+    {
+      "epoch": 0.6180987434953674,
+      "grad_norm": 1.3752774484881154,
+      "learning_rate": 2e-05,
+      "loss": 0.6741479396820068,
+      "step": 4870
+    },
+    {
+      "epoch": 0.6193679400939206,
+      "grad_norm": 1.3559231942904544,
+      "learning_rate": 2e-05,
+      "loss": 0.654120922088623,
+      "step": 4880
+    },
+    {
+      "epoch": 0.6206371366924737,
+      "grad_norm": 1.0511035204644243,
+      "learning_rate": 2e-05,
+      "loss": 0.6851940155029297,
+      "step": 4890
+    },
+    {
+      "epoch": 0.6219063332910267,
+      "grad_norm": 1.374958209297334,
+      "learning_rate": 2e-05,
+      "loss": 0.6714517593383789,
+      "step": 4900
+    },
+    {
+      "epoch": 0.6231755298895799,
+      "grad_norm": 1.1113145288196622,
+      "learning_rate": 2e-05,
+      "loss": 0.6819391250610352,
+      "step": 4910
+    },
+    {
+      "epoch": 0.624444726488133,
+      "grad_norm": 0.8999625092371784,
+      "learning_rate": 2e-05,
+      "loss": 0.6368520259857178,
+      "step": 4920
+    },
+    {
+      "epoch": 0.6257139230866862,
+      "grad_norm": 3.7057394205188845,
+      "learning_rate": 2e-05,
+      "loss": 0.6511198997497558,
+      "step": 4930
+    },
+    {
+      "epoch": 0.6269831196852392,
+      "grad_norm": 0.8349705086647643,
+      "learning_rate": 2e-05,
+      "loss": 0.6547498226165771,
+      "step": 4940
+    },
+    {
+      "epoch": 0.6282523162837924,
+      "grad_norm": 1.4175558211724757,
+      "learning_rate": 2e-05,
+      "loss": 0.6429747581481934,
+      "step": 4950
+    },
+    {
+      "epoch": 0.6295215128823455,
+      "grad_norm": 0.9651049753181332,
+      "learning_rate": 2e-05,
+      "loss": 0.6735240936279296,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6307907094808985,
+      "grad_norm": 1.552293642594364,
+      "learning_rate": 2e-05,
+      "loss": 0.6254170894622803,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6320599060794517,
+      "grad_norm": 1.8449116761514837,
+      "learning_rate": 2e-05,
+      "loss": 0.6590550899505615,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6333291026780048,
+      "grad_norm": 0.8914894333342914,
+      "learning_rate": 2e-05,
+      "loss": 0.6918347835540771,
+      "step": 4990
+    },
+    {
+      "epoch": 0.634598299276558,
+      "grad_norm": 0.954605194910724,
+      "learning_rate": 2e-05,
+      "loss": 0.6701622009277344,
+      "step": 5000
+    },
+    {
+      "epoch": 0.635867495875111,
+      "grad_norm": 1.0917482241013625,
+      "learning_rate": 2e-05,
+      "loss": 0.6577583312988281,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6371366924736642,
+      "grad_norm": 1.3172820335715711,
+      "learning_rate": 2e-05,
+      "loss": 0.6451882362365723,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6384058890722173,
+      "grad_norm": 0.8150746540069186,
+      "learning_rate": 2e-05,
+      "loss": 0.6584060668945313,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6396750856707704,
+      "grad_norm": 0.9520161951156947,
+      "learning_rate": 2e-05,
+      "loss": 0.672630500793457,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6409442822693235,
+      "grad_norm": 9.073185581258697,
+      "learning_rate": 2e-05,
+      "loss": 0.6395196437835693,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6422134788678766,
+      "grad_norm": 1.5868436234524692,
+      "learning_rate": 2e-05,
+      "loss": 0.6540477752685547,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6434826754664298,
+      "grad_norm": 0.8244218618251271,
+      "learning_rate": 2e-05,
+      "loss": 0.6300366878509521,
+      "step": 5070
+    },
+    {
+      "epoch": 0.6447518720649829,
+      "grad_norm": 0.9805177503635651,
+      "learning_rate": 2e-05,
+      "loss": 0.679721736907959,
+      "step": 5080
+    },
+    {
+      "epoch": 0.646021068663536,
+      "grad_norm": 1.1246944022139553,
+      "learning_rate": 2e-05,
+      "loss": 0.6822329521179199,
+      "step": 5090
+    },
+    {
+      "epoch": 0.6472902652620891,
+      "grad_norm": 0.8549741305350386,
+      "learning_rate": 2e-05,
+      "loss": 0.6687555313110352,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6485594618606422,
+      "grad_norm": 0.9818918566313932,
+      "learning_rate": 2e-05,
+      "loss": 0.6532404899597168,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6498286584591954,
+      "grad_norm": 0.9440827339677342,
+      "learning_rate": 2e-05,
+      "loss": 0.6677381038665772,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6510978550577484,
+      "grad_norm": 1.876384946425362,
+      "learning_rate": 2e-05,
+      "loss": 0.6486550807952881,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6523670516563016,
+      "grad_norm": 0.9005147166192696,
+      "learning_rate": 2e-05,
+      "loss": 0.664516544342041,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6536362482548547,
+      "grad_norm": 1.3539397311697965,
+      "learning_rate": 2e-05,
+      "loss": 0.6377689361572265,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6549054448534078,
+      "grad_norm": 2.9455619627566256,
+      "learning_rate": 2e-05,
+      "loss": 0.6450087547302246,
+      "step": 5160
+    },
+    {
+      "epoch": 0.6561746414519609,
+      "grad_norm": 1.037980544029311,
+      "learning_rate": 2e-05,
+      "loss": 0.6405978202819824,
+      "step": 5170
+    },
+    {
+      "epoch": 0.657443838050514,
+      "grad_norm": 4.912507709585592,
+      "learning_rate": 2e-05,
+      "loss": 0.649745512008667,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6587130346490672,
+      "grad_norm": 1.0609959385719505,
+      "learning_rate": 2e-05,
+      "loss": 0.6527548789978027,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6599822312476202,
+      "grad_norm": 0.9045390372378351,
+      "learning_rate": 2e-05,
+      "loss": 0.6748480796813965,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6612514278461734,
+      "grad_norm": 2.8217328805208015,
+      "learning_rate": 2e-05,
+      "loss": 0.6403985500335694,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6625206244447265,
+      "grad_norm": 2.73929693613192,
+      "learning_rate": 2e-05,
+      "loss": 0.6728531837463378,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6637898210432795,
+      "grad_norm": 1.3461944898984592,
+      "learning_rate": 2e-05,
+      "loss": 0.6563349723815918,
+      "step": 5230
+    },
+    {
+      "epoch": 0.6650590176418327,
+      "grad_norm": 0.939901984678927,
+      "learning_rate": 2e-05,
+      "loss": 0.6568197250366211,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6663282142403858,
+      "grad_norm": 1.0154186530780362,
+      "learning_rate": 2e-05,
+      "loss": 0.6752130508422851,
+      "step": 5250
+    },
+    {
+      "epoch": 0.667597410838939,
+      "grad_norm": 1.176337065487465,
+      "learning_rate": 2e-05,
+      "loss": 0.6836812973022461,
+      "step": 5260
+    },
+    {
+      "epoch": 0.668866607437492,
+      "grad_norm": 1.1931652271682958,
+      "learning_rate": 2e-05,
+      "loss": 0.671379280090332,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6701358040360452,
+      "grad_norm": 0.8848592952917024,
+      "learning_rate": 2e-05,
+      "loss": 0.6358261108398438,
+      "step": 5280
+    },
+    {
+      "epoch": 0.6714050006345983,
+      "grad_norm": 2.093835704492863,
+      "learning_rate": 2e-05,
+      "loss": 0.6444013118743896,
+      "step": 5290
+    },
+    {
+      "epoch": 0.6726741972331514,
+      "grad_norm": 0.8162402513147938,
+      "learning_rate": 2e-05,
+      "loss": 0.6532706260681153,
+      "step": 5300
+    },
+    {
+      "epoch": 0.6739433938317045,
+      "grad_norm": 1.3064661222485678,
+      "learning_rate": 2e-05,
+      "loss": 0.6681886196136475,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6752125904302576,
+      "grad_norm": 0.8650547339352569,
+      "learning_rate": 2e-05,
+      "loss": 0.6813479900360108,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6764817870288108,
+      "grad_norm": 0.9620521416011913,
+      "learning_rate": 2e-05,
+      "loss": 0.659577465057373,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6777509836273639,
+      "grad_norm": 1.0783033616040365,
+      "learning_rate": 2e-05,
+      "loss": 0.6426406860351562,
+      "step": 5340
+    },
+    {
+      "epoch": 0.679020180225917,
+      "grad_norm": 1.0963641431668159,
+      "learning_rate": 1.999194520388485e-05,
+      "loss": 0.6458285808563232,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6802893768244701,
+      "grad_norm": 1.2704346143879532,
+      "learning_rate": 1.9959247303157763e-05,
+      "loss": 0.6372426509857178,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6815585734230232,
+      "grad_norm": 0.9754558511159297,
+      "learning_rate": 1.990149424459817e-05,
+      "loss": 0.6500587463378906,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6828277700215764,
+      "grad_norm": 0.8456695535755101,
+      "learning_rate": 1.9818847538648688e-05,
+      "loss": 0.6528340816497803,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6840969666201294,
+      "grad_norm": 1.0686319907727386,
+      "learning_rate": 1.971153831256721e-05,
+      "loss": 0.6316551208496094,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6853661632186826,
+      "grad_norm": 1.225389194522042,
+      "learning_rate": 1.9579866664063435e-05,
+      "loss": 0.6384797096252441,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6866353598172357,
+      "grad_norm": 1.2794727884413368,
+      "learning_rate": 1.9424200822054782e-05,
+      "loss": 0.6397587776184082,
+      "step": 5410
+    },
+    {
+      "epoch": 0.6879045564157888,
+      "grad_norm": 1.006208237267993,
+      "learning_rate": 1.9244976116888626e-05,
+      "loss": 0.651089096069336,
+      "step": 5420
+    },
+    {
+      "epoch": 0.6891737530143419,
+      "grad_norm": 2.3764113815801053,
+      "learning_rate": 1.904269376291071e-05,
+      "loss": 0.6748910903930664,
+      "step": 5430
+    },
+    {
+      "epoch": 0.690442949612895,
+      "grad_norm": 3.1057535137845362,
+      "learning_rate": 1.88179194567844e-05,
+      "loss": 0.6195881843566895,
+      "step": 5440
+    },
+    {
+      "epoch": 0.6917121462114482,
+      "grad_norm": 1.0379617465087594,
+      "learning_rate": 1.8571281795480632e-05,
+      "loss": 0.6239624977111816,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6929813428100012,
+      "grad_norm": 1.291783757271844,
+      "learning_rate": 1.8303470518362808e-05,
+      "loss": 0.658579158782959,
+      "step": 5460
+    },
+    {
+      "epoch": 0.6942505394085544,
+      "grad_norm": 2.224989887203341,
+      "learning_rate": 1.801523457828271e-05,
+      "loss": 0.6622195720672608,
+      "step": 5470
+    },
+    {
+      "epoch": 0.6955197360071075,
+      "grad_norm": 0.8687392787271337,
+      "learning_rate": 1.7707380047081796e-05,
+      "loss": 0.646207332611084,
+      "step": 5480
+    },
+    {
+      "epoch": 0.6967889326056607,
+      "grad_norm": 1.3420255730329926,
+      "learning_rate": 1.7380767861355252e-05,
+      "loss": 0.6533578395843506,
+      "step": 5490
+    },
+    {
+      "epoch": 0.6980581292042137,
+      "grad_norm": 1.6731300052366367,
+      "learning_rate": 1.703631141478299e-05,
+      "loss": 0.6319987297058105,
+      "step": 5500
+    },
+    {
+      "epoch": 0.6993273258027668,
+      "grad_norm": 0.824845616688245,
+      "learning_rate": 1.6674974003760705e-05,
+      "loss": 0.685127592086792,
+      "step": 5510
+    },
+    {
+      "epoch": 0.70059652240132,
+      "grad_norm": 0.7476574205300913,
+      "learning_rate": 1.6297766133474556e-05,
+      "loss": 0.6838325023651123,
+      "step": 5520
+    },
+    {
+      "epoch": 0.701865718999873,
+      "grad_norm": 0.9855693379414168,
+      "learning_rate": 1.5905742691953186e-05,
+      "loss": 0.6698818206787109,
+      "step": 5530
+    },
+    {
+      "epoch": 0.7031349155984262,
+      "grad_norm": 0.885122475341566,
+      "learning_rate": 1.55e-05,
+      "loss": 0.6396905422210694,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7044041121969793,
+      "grad_norm": 1.0669167683586287,
+      "learning_rate": 1.5081672745255862e-05,
+      "loss": 0.6441478729248047,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7056733087955325,
+      "grad_norm": 1.088148852952991,
+      "learning_rate": 1.4651930808966206e-05,
+      "loss": 0.6560922622680664,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7069425053940855,
+      "grad_norm": 0.955229708551378,
+      "learning_rate": 1.4211975994326848e-05,
+      "loss": 0.6540224075317382,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7082117019926386,
+      "grad_norm": 2.43346491287166,
+      "learning_rate": 1.376303866555779e-05,
+      "loss": 0.6365595817565918,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7094808985911918,
+      "grad_norm": 1.485628027068409,
+      "learning_rate": 1.3306374307104164e-05,
+      "loss": 0.6573779106140136,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7107500951897449,
+      "grad_norm": 0.9204769101734288,
+      "learning_rate": 1.2843260012586718e-05,
+      "loss": 0.6073714256286621,
+      "step": 5600
+    },
+    {
+      "epoch": 0.712019291788298,
+      "grad_norm": 0.973733365643115,
+      "learning_rate": 1.2374990913320726e-05,
+      "loss": 0.6591270446777344,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7132884883868511,
+      "grad_norm": 0.9464949869853067,
+      "learning_rate": 1.1902876556391247e-05,
+      "loss": 0.6291365623474121,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7145576849854043,
+      "grad_norm": 1.0310427703362444,
+      "learning_rate": 1.1428237242413685e-05,
+      "loss": 0.6285597801208496,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7158268815839574,
+      "grad_norm": 74.24516096880984,
+      "learning_rate": 1.095240033322132e-05,
+      "loss": 0.6464774131774902,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7170960781825104,
+      "grad_norm": 0.91661412733031,
+      "learning_rate": 1.047669653980572e-05,
+      "loss": 0.6213871002197265,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7183652747810636,
+      "grad_norm": 0.8448338332091152,
+      "learning_rate": 1.0002456200890902e-05,
+      "loss": 0.6397408485412598,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7196344713796167,
+      "grad_norm": 0.9008731858316706,
+      "learning_rate": 9.531005562548668e-06,
+      "loss": 0.6388719558715821,
+      "step": 5670
+    },
+    {
+      "epoch": 0.7209036679781698,
+      "grad_norm": 1.3049177174679707,
+      "learning_rate": 9.06366306925925e-06,
+      "loss": 0.677491283416748,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7221728645767229,
+      "grad_norm": 1.319910168446297,
+      "learning_rate": 8.601735676789686e-06,
+      "loss": 0.6340361595153808,
+      "step": 5690
+    },
+    {
+      "epoch": 0.7234420611752761,
+      "grad_norm": 0.7372026627578863,
+      "learning_rate": 8.14651519720118e-06,
+      "loss": 0.6337122917175293,
+      "step": 5700
+    },
+    {
+      "epoch": 0.7247112577738292,
+      "grad_norm": 1.2177065772175075,
+      "learning_rate": 7.699274686206866e-06,
+      "loss": 0.6361382961273193,
+      "step": 5710
+    },
+    {
+      "epoch": 0.7259804543723822,
+      "grad_norm": 0.8454556818215734,
+      "learning_rate": 7.261264882983024e-06,
+      "loss": 0.5935632228851319,
+      "step": 5720
+    },
+    {
+      "epoch": 0.7272496509709354,
+      "grad_norm": 1.2130222872128056,
+      "learning_rate": 6.8337107123900386e-06,
+      "loss": 0.6379383087158204,
+      "step": 5730
+    },
+    {
+      "epoch": 0.7285188475694885,
+      "grad_norm": 1.436306750160683,
+      "learning_rate": 6.417807859384881e-06,
+      "loss": 0.639438533782959,
+      "step": 5740
+    },
+    {
+      "epoch": 0.7297880441680417,
+      "grad_norm": 0.7609574446255283,
+      "learning_rate": 6.014719425205009e-06,
+      "loss": 0.634493637084961,
+      "step": 5750
+    },
+    {
+      "epoch": 0.7310572407665947,
+      "grad_norm": 1.2995854290920883,
+      "learning_rate": 5.6255726746749275e-06,
+      "loss": 0.631171703338623,
+      "step": 5760
+    },
+    {
+      "epoch": 0.7323264373651479,
+      "grad_norm": 0.9123520822181704,
+      "learning_rate": 5.251455883731728e-06,
+      "loss": 0.6408793926239014,
+      "step": 5770
+    },
+    {
+      "epoch": 0.733595633963701,
+      "grad_norm": 0.8658561983322264,
+      "learning_rate": 4.893415295985812e-06,
+      "loss": 0.6286262512207031,
+      "step": 5780
+    },
+    {
+      "epoch": 0.734864830562254,
+      "grad_norm": 0.8164953875027156,
+      "learning_rate": 4.552452196827885e-06,
+      "loss": 0.6547434806823731,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7361340271608072,
+      "grad_norm": 1.4641510503033233,
+      "learning_rate": 4.229520113264785e-06,
+      "loss": 0.6368844985961915,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7374032237593603,
+      "grad_norm": 0.9598098245028448,
+      "learning_rate": 3.925522147314915e-06,
+      "loss": 0.6294554710388184,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7386724203579135,
+      "grad_norm": 1.009356441345964,
+      "learning_rate": 3.641308450420675e-06,
+      "loss": 0.6177220821380616,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7399416169564665,
+      "grad_norm": 0.8240647509132678,
+      "learning_rate": 3.37767384594087e-06,
+      "loss": 0.6307158470153809,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7412108135550197,
+      "grad_norm": 1.1542174689447895,
+      "learning_rate": 3.1353556063719357e-06,
+      "loss": 0.6274588584899903,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7424800101535728,
+      "grad_norm": 0.7939759615957755,
+      "learning_rate": 2.9150313915141474e-06,
+      "loss": 0.6363517284393311,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7437492067521259,
+      "grad_norm": 0.8609215208550008,
+      "learning_rate": 2.717317353348891e-06,
+      "loss": 0.6226596832275391,
+      "step": 5860
+    },
+    {
+      "epoch": 0.745018403350679,
+      "grad_norm": 2.479869419769789,
+      "learning_rate": 2.5427664129268253e-06,
+      "loss": 0.6489467144012451,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7462875999492321,
+      "grad_norm": 1.0016503169304862,
+      "learning_rate": 2.391866714085717e-06,
+      "loss": 0.6111717224121094,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7475567965477853,
+      "grad_norm": 2.5951990557188416,
+      "learning_rate": 2.2650402583222753e-06,
+      "loss": 0.6319608211517334,
+      "step": 5890
+    },
+    {
+      "epoch": 0.7488259931463384,
+      "grad_norm": 1.7302486268183335,
+      "learning_rate": 2.1626417246356398e-06,
+      "loss": 0.6457314491271973,
+      "step": 5900
+    },
+    {
+      "epoch": 0.7500951897448915,
+      "grad_norm": 2.754484373117495,
+      "learning_rate": 2.084957477642894e-06,
+      "loss": 0.6203114509582519,
+      "step": 5910
+    },
+    {
+      "epoch": 0.7513643863434446,
+      "grad_norm": 1.2247650932887257,
+      "learning_rate": 2.0322047667405147e-06,
+      "loss": 0.6380536556243896,
+      "step": 5920
+    },
+    {
+      "epoch": 0.7526335829419977,
+      "grad_norm": 0.7716109314316574,
+      "learning_rate": 2.0045311185513342e-06,
+      "loss": 0.6163107872009277,
+      "step": 5930
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5935,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2035,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2810342056919040.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}