diff --git "a/codet5_ia3_official_0.0001/checkpoint-14718/trainer_state.json" "b/codet5_ia3_official_0.0001/checkpoint-14718/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/codet5_ia3_official_0.0001/checkpoint-14718/trainer_state.json"
@@ -0,0 +1,20656 @@
+{
+  "best_metric": 0.0021924919669198163,
+  "best_model_checkpoint": "./results-cc/code-t5/codet5_ia3_official_0.0001/checkpoint-14718",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 14718,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003397200706617747,
+      "grad_norm": 2.2195968627929688,
+      "learning_rate": 9.999575349911673e-05,
+      "loss": 10.0277,
+      "step": 5
+    },
+    {
+      "epoch": 0.0006794401413235494,
+      "grad_norm": 1.40236496925354,
+      "learning_rate": 9.999320559858677e-05,
+      "loss": 10.112,
+      "step": 10
+    },
+    {
+      "epoch": 0.0010191602119853241,
+      "grad_norm": 2.544349431991577,
+      "learning_rate": 9.99889590977035e-05,
+      "loss": 10.7856,
+      "step": 15
+    },
+    {
+      "epoch": 0.001358880282647099,
+      "grad_norm": 1.8622939586639404,
+      "learning_rate": 9.998471259682023e-05,
+      "loss": 9.73,
+      "step": 20
+    },
+    {
+      "epoch": 0.0016986003533088735,
+      "grad_norm": 2.0292515754699707,
+      "learning_rate": 9.998046609593695e-05,
+      "loss": 10.2627,
+      "step": 25
+    },
+    {
+      "epoch": 0.0020383204239706482,
+      "grad_norm": 1.862714171409607,
+      "learning_rate": 9.997621959505368e-05,
+      "loss": 9.8577,
+      "step": 30
+    },
+    {
+      "epoch": 0.002378040494632423,
+      "grad_norm": 3.0175716876983643,
+      "learning_rate": 9.997197309417041e-05,
+      "loss": 9.6583,
+      "step": 35
+    },
+    {
+      "epoch": 0.002717760565294198,
+      "grad_norm": 2.3832008838653564,
+      "learning_rate": 9.996772659328714e-05,
+      "loss": 9.8661,
+      "step": 40
+    },
+    {
+      "epoch": 0.0030574806359559724,
+      "grad_norm": 1.7646359205245972,
+      "learning_rate": 9.996348009240387e-05,
+      "loss": 10.5005,
+      "step": 45
+    },
+    {
+      "epoch": 0.003397200706617747,
+      "grad_norm": 1.7444077730178833,
+      "learning_rate": 9.996008289169725e-05,
+      "loss": 9.5758,
+      "step": 50
+    },
+    {
+      "epoch": 0.0037369207772795215,
+      "grad_norm": 1.9149781465530396,
+      "learning_rate": 9.995583639081398e-05,
+      "loss": 9.735,
+      "step": 55
+    },
+    {
+      "epoch": 0.0040766408479412965,
+      "grad_norm": 1.6488384008407593,
+      "learning_rate": 9.995158988993069e-05,
+      "loss": 9.8686,
+      "step": 60
+    },
+    {
+      "epoch": 0.0044163609186030715,
+      "grad_norm": 2.184131622314453,
+      "learning_rate": 9.994734338904743e-05,
+      "loss": 10.1822,
+      "step": 65
+    },
+    {
+      "epoch": 0.004756080989264846,
+      "grad_norm": 2.6063928604125977,
+      "learning_rate": 9.994309688816416e-05,
+      "loss": 9.9386,
+      "step": 70
+    },
+    {
+      "epoch": 0.005095801059926621,
+      "grad_norm": 1.925402045249939,
+      "learning_rate": 9.993885038728087e-05,
+      "loss": 9.3722,
+      "step": 75
+    },
+    {
+      "epoch": 0.005435521130588396,
+      "grad_norm": 3.882549524307251,
+      "learning_rate": 9.993545318657427e-05,
+      "loss": 9.4705,
+      "step": 80
+    },
+    {
+      "epoch": 0.00577524120125017,
+      "grad_norm": 2.1654701232910156,
+      "learning_rate": 9.9931206685691e-05,
+      "loss": 9.5574,
+      "step": 85
+    },
+    {
+      "epoch": 0.006114961271911945,
+      "grad_norm": 2.0329928398132324,
+      "learning_rate": 9.992696018480773e-05,
+      "loss": 9.7831,
+      "step": 90
+    },
+    {
+      "epoch": 0.006454681342573719,
+      "grad_norm": 2.408450126647949,
+      "learning_rate": 9.992271368392445e-05,
+      "loss": 10.2149,
+      "step": 95
+    },
+    {
+      "epoch": 0.006794401413235494,
+      "grad_norm": 2.039151430130005,
+      "learning_rate": 9.991846718304117e-05,
+      "loss": 10.2672,
+      "step": 100
+    },
+    {
+      "epoch": 0.007134121483897269,
+      "grad_norm": 2.3165321350097656,
+      "learning_rate": 9.991422068215791e-05,
+      "loss": 9.4804,
+      "step": 105
+    },
+    {
+      "epoch": 0.007473841554559043,
+      "grad_norm": 2.117964029312134,
+      "learning_rate": 9.990997418127464e-05,
+      "loss": 10.1318,
+      "step": 110
+    },
+    {
+      "epoch": 0.007813561625220818,
+      "grad_norm": 2.1264026165008545,
+      "learning_rate": 9.990572768039135e-05,
+      "loss": 9.6686,
+      "step": 115
+    },
+    {
+      "epoch": 0.008153281695882593,
+      "grad_norm": 1.2933008670806885,
+      "learning_rate": 9.99014811795081e-05,
+      "loss": 9.4348,
+      "step": 120
+    },
+    {
+      "epoch": 0.008493001766544368,
+      "grad_norm": 2.1963112354278564,
+      "learning_rate": 9.989723467862482e-05,
+      "loss": 9.7095,
+      "step": 125
+    },
+    {
+      "epoch": 0.008832721837206143,
+      "grad_norm": 1.4545302391052246,
+      "learning_rate": 9.989298817774154e-05,
+      "loss": 9.5571,
+      "step": 130
+    },
+    {
+      "epoch": 0.009172441907867916,
+      "grad_norm": 1.4792237281799316,
+      "learning_rate": 9.988874167685828e-05,
+      "loss": 8.9859,
+      "step": 135
+    },
+    {
+      "epoch": 0.009512161978529691,
+      "grad_norm": 1.2513773441314697,
+      "learning_rate": 9.9884495175975e-05,
+      "loss": 9.1648,
+      "step": 140
+    },
+    {
+      "epoch": 0.009851882049191466,
+      "grad_norm": 2.559937000274658,
+      "learning_rate": 9.988024867509172e-05,
+      "loss": 9.7622,
+      "step": 145
+    },
+    {
+      "epoch": 0.010191602119853241,
+      "grad_norm": 1.9447046518325806,
+      "learning_rate": 9.987600217420846e-05,
+      "loss": 9.8491,
+      "step": 150
+    },
+    {
+      "epoch": 0.010531322190515016,
+      "grad_norm": 1.3316339254379272,
+      "learning_rate": 9.987175567332519e-05,
+      "loss": 9.0772,
+      "step": 155
+    },
+    {
+      "epoch": 0.010871042261176791,
+      "grad_norm": 1.4368464946746826,
+      "learning_rate": 9.98675091724419e-05,
+      "loss": 8.9018,
+      "step": 160
+    },
+    {
+      "epoch": 0.011210762331838564,
+      "grad_norm": 1.5808727741241455,
+      "learning_rate": 9.986326267155865e-05,
+      "loss": 8.9733,
+      "step": 165
+    },
+    {
+      "epoch": 0.01155048240250034,
+      "grad_norm": 1.7572826147079468,
+      "learning_rate": 9.985901617067537e-05,
+      "loss": 8.7883,
+      "step": 170
+    },
+    {
+      "epoch": 0.011890202473162114,
+      "grad_norm": 1.6672322750091553,
+      "learning_rate": 9.985476966979209e-05,
+      "loss": 8.8474,
+      "step": 175
+    },
+    {
+      "epoch": 0.01222992254382389,
+      "grad_norm": 1.3700741529464722,
+      "learning_rate": 9.985052316890883e-05,
+      "loss": 9.4713,
+      "step": 180
+    },
+    {
+      "epoch": 0.012569642614485664,
+      "grad_norm": 1.9305622577667236,
+      "learning_rate": 9.984627666802555e-05,
+      "loss": 9.2708,
+      "step": 185
+    },
+    {
+      "epoch": 0.012909362685147438,
+      "grad_norm": 1.964107871055603,
+      "learning_rate": 9.984203016714227e-05,
+      "loss": 8.8052,
+      "step": 190
+    },
+    {
+      "epoch": 0.013249082755809213,
+      "grad_norm": 1.9907019138336182,
+      "learning_rate": 9.983778366625901e-05,
+      "loss": 9.6229,
+      "step": 195
+    },
+    {
+      "epoch": 0.013588802826470988,
+      "grad_norm": 1.7012939453125,
+      "learning_rate": 9.983353716537573e-05,
+      "loss": 8.9995,
+      "step": 200
+    },
+    {
+      "epoch": 0.013928522897132763,
+      "grad_norm": 1.7597671747207642,
+      "learning_rate": 9.982929066449246e-05,
+      "loss": 8.8808,
+      "step": 205
+    },
+    {
+      "epoch": 0.014268242967794538,
+      "grad_norm": 1.6379801034927368,
+      "learning_rate": 9.98250441636092e-05,
+      "loss": 8.8094,
+      "step": 210
+    },
+    {
+      "epoch": 0.014607963038456313,
+      "grad_norm": 1.9670891761779785,
+      "learning_rate": 9.982079766272591e-05,
+      "loss": 9.522,
+      "step": 215
+    },
+    {
+      "epoch": 0.014947683109118086,
+      "grad_norm": 1.9001445770263672,
+      "learning_rate": 9.981655116184264e-05,
+      "loss": 9.3781,
+      "step": 220
+    },
+    {
+      "epoch": 0.015287403179779861,
+      "grad_norm": 1.9136974811553955,
+      "learning_rate": 9.981230466095938e-05,
+      "loss": 8.8219,
+      "step": 225
+    },
+    {
+      "epoch": 0.015627123250441636,
+      "grad_norm": 2.4746735095977783,
+      "learning_rate": 9.98080581600761e-05,
+      "loss": 8.9191,
+      "step": 230
+    },
+    {
+      "epoch": 0.01596684332110341,
+      "grad_norm": 2.372750759124756,
+      "learning_rate": 9.980381165919283e-05,
+      "loss": 9.1705,
+      "step": 235
+    },
+    {
+      "epoch": 0.016306563391765186,
+      "grad_norm": 1.418637752532959,
+      "learning_rate": 9.980041445848621e-05,
+      "loss": 8.6098,
+      "step": 240
+    },
+    {
+      "epoch": 0.01664628346242696,
+      "grad_norm": 2.103688955307007,
+      "learning_rate": 9.979616795760294e-05,
+      "loss": 9.1948,
+      "step": 245
+    },
+    {
+      "epoch": 0.016986003533088736,
+      "grad_norm": 2.14066481590271,
+      "learning_rate": 9.979192145671968e-05,
+      "loss": 9.0493,
+      "step": 250
+    },
+    {
+      "epoch": 0.01732572360375051,
+      "grad_norm": 1.4272230863571167,
+      "learning_rate": 9.978767495583639e-05,
+      "loss": 8.7393,
+      "step": 255
+    },
+    {
+      "epoch": 0.017665443674412286,
+      "grad_norm": 1.5108071565628052,
+      "learning_rate": 9.978342845495312e-05,
+      "loss": 9.3101,
+      "step": 260
+    },
+    {
+      "epoch": 0.01800516374507406,
+      "grad_norm": 2.017267942428589,
+      "learning_rate": 9.977918195406986e-05,
+      "loss": 9.0584,
+      "step": 265
+    },
+    {
+      "epoch": 0.018344883815735832,
+      "grad_norm": 1.4429893493652344,
+      "learning_rate": 9.977493545318658e-05,
+      "loss": 8.5374,
+      "step": 270
+    },
+    {
+      "epoch": 0.01868460388639761,
+      "grad_norm": 1.7082629203796387,
+      "learning_rate": 9.97706889523033e-05,
+      "loss": 8.2905,
+      "step": 275
+    },
+    {
+      "epoch": 0.019024323957059382,
+      "grad_norm": 1.9540777206420898,
+      "learning_rate": 9.976644245142005e-05,
+      "loss": 8.9633,
+      "step": 280
+    },
+    {
+      "epoch": 0.01936404402772116,
+      "grad_norm": 1.4489926099777222,
+      "learning_rate": 9.976219595053676e-05,
+      "loss": 8.7871,
+      "step": 285
+    },
+    {
+      "epoch": 0.019703764098382932,
+      "grad_norm": 1.5926896333694458,
+      "learning_rate": 9.975794944965349e-05,
+      "loss": 8.5374,
+      "step": 290
+    },
+    {
+      "epoch": 0.020043484169044706,
+      "grad_norm": 1.8608131408691406,
+      "learning_rate": 9.975370294877022e-05,
+      "loss": 8.4816,
+      "step": 295
+    },
+    {
+      "epoch": 0.020383204239706482,
+      "grad_norm": 1.6402130126953125,
+      "learning_rate": 9.974945644788694e-05,
+      "loss": 8.3059,
+      "step": 300
+    },
+    {
+      "epoch": 0.020722924310368256,
+      "grad_norm": 1.0616756677627563,
+      "learning_rate": 9.974520994700367e-05,
+      "loss": 8.5302,
+      "step": 305
+    },
+    {
+      "epoch": 0.021062644381030032,
+      "grad_norm": 2.0486464500427246,
+      "learning_rate": 9.97409634461204e-05,
+      "loss": 8.6573,
+      "step": 310
+    },
+    {
+      "epoch": 0.021402364451691806,
+      "grad_norm": 2.764594316482544,
+      "learning_rate": 9.973671694523713e-05,
+      "loss": 9.0297,
+      "step": 315
+    },
+    {
+      "epoch": 0.021742084522353582,
+      "grad_norm": 1.457748293876648,
+      "learning_rate": 9.973247044435386e-05,
+      "loss": 8.6019,
+      "step": 320
+    },
+    {
+      "epoch": 0.022081804593015356,
+      "grad_norm": 1.5769116878509521,
+      "learning_rate": 9.972822394347058e-05,
+      "loss": 8.2577,
+      "step": 325
+    },
+    {
+      "epoch": 0.02242152466367713,
+      "grad_norm": 1.7383826971054077,
+      "learning_rate": 9.972397744258731e-05,
+      "loss": 8.1914,
+      "step": 330
+    },
+    {
+      "epoch": 0.022761244734338906,
+      "grad_norm": 1.6818331480026245,
+      "learning_rate": 9.971973094170404e-05,
+      "loss": 8.8289,
+      "step": 335
+    },
+    {
+      "epoch": 0.02310096480500068,
+      "grad_norm": 1.429895281791687,
+      "learning_rate": 9.971548444082077e-05,
+      "loss": 8.3524,
+      "step": 340
+    },
+    {
+      "epoch": 0.023440684875662456,
+      "grad_norm": 1.2669081687927246,
+      "learning_rate": 9.97112379399375e-05,
+      "loss": 8.289,
+      "step": 345
+    },
+    {
+      "epoch": 0.02378040494632423,
+      "grad_norm": 1.399524211883545,
+      "learning_rate": 9.970699143905422e-05,
+      "loss": 7.8466,
+      "step": 350
+    },
+    {
+      "epoch": 0.024120125016986002,
+      "grad_norm": 1.6960299015045166,
+      "learning_rate": 9.970274493817095e-05,
+      "loss": 8.6849,
+      "step": 355
+    },
+    {
+      "epoch": 0.02445984508764778,
+      "grad_norm": 1.7302825450897217,
+      "learning_rate": 9.969849843728768e-05,
+      "loss": 8.6366,
+      "step": 360
+    },
+    {
+      "epoch": 0.024799565158309552,
+      "grad_norm": 2.6233043670654297,
+      "learning_rate": 9.969425193640441e-05,
+      "loss": 8.3961,
+      "step": 365
+    },
+    {
+      "epoch": 0.02513928522897133,
+      "grad_norm": 1.480035424232483,
+      "learning_rate": 9.969000543552114e-05,
+      "loss": 8.6224,
+      "step": 370
+    },
+    {
+      "epoch": 0.025479005299633102,
+      "grad_norm": 1.156540870666504,
+      "learning_rate": 9.968575893463786e-05,
+      "loss": 7.9226,
+      "step": 375
+    },
+    {
+      "epoch": 0.025818725370294875,
+      "grad_norm": 1.7962318658828735,
+      "learning_rate": 9.968151243375459e-05,
+      "loss": 8.5519,
+      "step": 380
+    },
+    {
+      "epoch": 0.026158445440956652,
+      "grad_norm": 1.8737194538116455,
+      "learning_rate": 9.967726593287132e-05,
+      "loss": 8.4364,
+      "step": 385
+    },
+    {
+      "epoch": 0.026498165511618425,
+      "grad_norm": 1.6001181602478027,
+      "learning_rate": 9.967301943198805e-05,
+      "loss": 7.8641,
+      "step": 390
+    },
+    {
+      "epoch": 0.026837885582280202,
+      "grad_norm": 1.6181342601776123,
+      "learning_rate": 9.966877293110478e-05,
+      "loss": 7.3995,
+      "step": 395
+    },
+    {
+      "epoch": 0.027177605652941975,
+      "grad_norm": 1.5771849155426025,
+      "learning_rate": 9.96645264302215e-05,
+      "loss": 7.8526,
+      "step": 400
+    },
+    {
+      "epoch": 0.027517325723603752,
+      "grad_norm": 1.1884416341781616,
+      "learning_rate": 9.966027992933823e-05,
+      "loss": 8.117,
+      "step": 405
+    },
+    {
+      "epoch": 0.027857045794265525,
+      "grad_norm": 2.015026092529297,
+      "learning_rate": 9.965603342845496e-05,
+      "loss": 8.5386,
+      "step": 410
+    },
+    {
+      "epoch": 0.0281967658649273,
+      "grad_norm": 1.2226543426513672,
+      "learning_rate": 9.965178692757169e-05,
+      "loss": 8.1885,
+      "step": 415
+    },
+    {
+      "epoch": 0.028536485935589075,
+      "grad_norm": 0.8924500942230225,
+      "learning_rate": 9.964754042668842e-05,
+      "loss": 7.9768,
+      "step": 420
+    },
+    {
+      "epoch": 0.02887620600625085,
+      "grad_norm": 1.3788869380950928,
+      "learning_rate": 9.964329392580513e-05,
+      "loss": 8.2231,
+      "step": 425
+    },
+    {
+      "epoch": 0.029215926076912625,
+      "grad_norm": 1.4598475694656372,
+      "learning_rate": 9.963904742492187e-05,
+      "loss": 7.9246,
+      "step": 430
+    },
+    {
+      "epoch": 0.0295556461475744,
+      "grad_norm": 1.8520469665527344,
+      "learning_rate": 9.96348009240386e-05,
+      "loss": 8.0527,
+      "step": 435
+    },
+    {
+      "epoch": 0.029895366218236172,
+      "grad_norm": 1.4544007778167725,
+      "learning_rate": 9.963055442315531e-05,
+      "loss": 7.9365,
+      "step": 440
+    },
+    {
+      "epoch": 0.03023508628889795,
+      "grad_norm": 1.632197380065918,
+      "learning_rate": 9.962630792227206e-05,
+      "loss": 7.6173,
+      "step": 445
+    },
+    {
+      "epoch": 0.030574806359559722,
+      "grad_norm": 1.7875256538391113,
+      "learning_rate": 9.962206142138878e-05,
+      "loss": 7.8095,
+      "step": 450
+    },
+    {
+      "epoch": 0.0309145264302215,
+      "grad_norm": 1.496638298034668,
+      "learning_rate": 9.961866422068217e-05,
+      "loss": 8.3371,
+      "step": 455
+    },
+    {
+      "epoch": 0.03125424650088327,
+      "grad_norm": 1.915732741355896,
+      "learning_rate": 9.96144177197989e-05,
+      "loss": 8.5992,
+      "step": 460
+    },
+    {
+      "epoch": 0.03159396657154505,
+      "grad_norm": 1.1497036218643188,
+      "learning_rate": 9.961017121891561e-05,
+      "loss": 7.7974,
+      "step": 465
+    },
+    {
+      "epoch": 0.03193368664220682,
+      "grad_norm": 1.5241578817367554,
+      "learning_rate": 9.960592471803235e-05,
+      "loss": 7.9306,
+      "step": 470
+    },
+    {
+      "epoch": 0.032273406712868595,
+      "grad_norm": 1.8431288003921509,
+      "learning_rate": 9.960167821714908e-05,
+      "loss": 8.0421,
+      "step": 475
+    },
+    {
+      "epoch": 0.03261312678353037,
+      "grad_norm": 1.520870566368103,
+      "learning_rate": 9.959743171626579e-05,
+      "loss": 7.443,
+      "step": 480
+    },
+    {
+      "epoch": 0.03295284685419215,
+      "grad_norm": 1.9101468324661255,
+      "learning_rate": 9.959318521538253e-05,
+      "loss": 7.5383,
+      "step": 485
+    },
+    {
+      "epoch": 0.03329256692485392,
+      "grad_norm": 1.2860801219940186,
+      "learning_rate": 9.958893871449926e-05,
+      "loss": 7.4753,
+      "step": 490
+    },
+    {
+      "epoch": 0.033632286995515695,
+      "grad_norm": 1.3139928579330444,
+      "learning_rate": 9.958469221361598e-05,
+      "loss": 8.0711,
+      "step": 495
+    },
+    {
+      "epoch": 0.03397200706617747,
+      "grad_norm": 2.382516622543335,
+      "learning_rate": 9.958044571273272e-05,
+      "loss": 7.9498,
+      "step": 500
+    },
+    {
+      "epoch": 0.03431172713683924,
+      "grad_norm": 1.4085739850997925,
+      "learning_rate": 9.957619921184945e-05,
+      "loss": 7.6138,
+      "step": 505
+    },
+    {
+      "epoch": 0.03465144720750102,
+      "grad_norm": 1.34367835521698,
+      "learning_rate": 9.957195271096616e-05,
+      "loss": 7.7689,
+      "step": 510
+    },
+    {
+      "epoch": 0.034991167278162795,
+      "grad_norm": 1.3598331212997437,
+      "learning_rate": 9.95677062100829e-05,
+      "loss": 7.4086,
+      "step": 515
+    },
+    {
+      "epoch": 0.03533088734882457,
+      "grad_norm": 1.3421567678451538,
+      "learning_rate": 9.956345970919963e-05,
+      "loss": 7.8855,
+      "step": 520
+    },
+    {
+      "epoch": 0.03567060741948634,
+      "grad_norm": 1.5375795364379883,
+      "learning_rate": 9.955921320831634e-05,
+      "loss": 7.642,
+      "step": 525
+    },
+    {
+      "epoch": 0.03601032749014812,
+      "grad_norm": 1.6586527824401855,
+      "learning_rate": 9.955496670743309e-05,
+      "loss": 7.1819,
+      "step": 530
+    },
+    {
+      "epoch": 0.036350047560809895,
+      "grad_norm": 1.179348349571228,
+      "learning_rate": 9.955072020654981e-05,
+      "loss": 7.4156,
+      "step": 535
+    },
+    {
+      "epoch": 0.036689767631471665,
+      "grad_norm": 2.1848621368408203,
+      "learning_rate": 9.954647370566653e-05,
+      "loss": 7.4587,
+      "step": 540
+    },
+    {
+      "epoch": 0.03702948770213344,
+      "grad_norm": 1.3930761814117432,
+      "learning_rate": 9.954222720478327e-05,
+      "loss": 7.3459,
+      "step": 545
+    },
+    {
+      "epoch": 0.03736920777279522,
+      "grad_norm": 1.6147781610488892,
+      "learning_rate": 9.953798070389998e-05,
+      "loss": 7.913,
+      "step": 550
+    },
+    {
+      "epoch": 0.03770892784345699,
+      "grad_norm": 1.2318421602249146,
+      "learning_rate": 9.953373420301671e-05,
+      "loss": 7.4613,
+      "step": 555
+    },
+    {
+      "epoch": 0.038048647914118765,
+      "grad_norm": 14.445450782775879,
+      "learning_rate": 9.952948770213345e-05,
+      "loss": 7.4797,
+      "step": 560
+    },
+    {
+      "epoch": 0.03838836798478054,
+      "grad_norm": 1.447785496711731,
+      "learning_rate": 9.952524120125017e-05,
+      "loss": 7.663,
+      "step": 565
+    },
+    {
+      "epoch": 0.03872808805544232,
+      "grad_norm": 1.2133177518844604,
+      "learning_rate": 9.95209947003669e-05,
+      "loss": 7.5911,
+      "step": 570
+    },
+    {
+      "epoch": 0.03906780812610409,
+      "grad_norm": 1.6101973056793213,
+      "learning_rate": 9.951674819948364e-05,
+      "loss": 7.7617,
+      "step": 575
+    },
+    {
+      "epoch": 0.039407528196765865,
+      "grad_norm": 1.5707918405532837,
+      "learning_rate": 9.951250169860035e-05,
+      "loss": 7.2746,
+      "step": 580
+    },
+    {
+      "epoch": 0.03974724826742764,
+      "grad_norm": 1.4747017621994019,
+      "learning_rate": 9.950825519771708e-05,
+      "loss": 7.559,
+      "step": 585
+    },
+    {
+      "epoch": 0.04008696833808941,
+      "grad_norm": 1.2751151323318481,
+      "learning_rate": 9.950400869683382e-05,
+      "loss": 7.0742,
+      "step": 590
+    },
+    {
+      "epoch": 0.04042668840875119,
+      "grad_norm": 1.4686486721038818,
+      "learning_rate": 9.949976219595054e-05,
+      "loss": 7.179,
+      "step": 595
+    },
+    {
+      "epoch": 0.040766408479412965,
+      "grad_norm": 1.333364725112915,
+      "learning_rate": 9.949551569506727e-05,
+      "loss": 7.3349,
+      "step": 600
+    },
+    {
+      "epoch": 0.04110612855007474,
+      "grad_norm": 1.2560040950775146,
+      "learning_rate": 9.9491269194184e-05,
+      "loss": 7.2152,
+      "step": 605
+    },
+    {
+      "epoch": 0.04144584862073651,
+      "grad_norm": 5.9243011474609375,
+      "learning_rate": 9.948702269330072e-05,
+      "loss": 7.253,
+      "step": 610
+    },
+    {
+      "epoch": 0.04178556869139829,
+      "grad_norm": 1.3705462217330933,
+      "learning_rate": 9.948277619241745e-05,
+      "loss": 7.2954,
+      "step": 615
+    },
+    {
+      "epoch": 0.042125288762060065,
+      "grad_norm": 1.3280870914459229,
+      "learning_rate": 9.947852969153418e-05,
+      "loss": 7.0023,
+      "step": 620
+    },
+    {
+      "epoch": 0.042465008832721834,
+      "grad_norm": 1.5480890274047852,
+      "learning_rate": 9.94742831906509e-05,
+      "loss": 6.6209,
+      "step": 625
+    },
+    {
+      "epoch": 0.04280472890338361,
+      "grad_norm": 1.4617500305175781,
+      "learning_rate": 9.947003668976763e-05,
+      "loss": 6.6055,
+      "step": 630
+    },
+    {
+      "epoch": 0.04314444897404539,
+      "grad_norm": 1.5756878852844238,
+      "learning_rate": 9.946579018888436e-05,
+      "loss": 7.0135,
+      "step": 635
+    },
+    {
+      "epoch": 0.043484169044707165,
+      "grad_norm": 1.4289640188217163,
+      "learning_rate": 9.946154368800109e-05,
+      "loss": 7.1441,
+      "step": 640
+    },
+    {
+      "epoch": 0.043823889115368934,
+      "grad_norm": 1.3657900094985962,
+      "learning_rate": 9.945729718711782e-05,
+      "loss": 7.5154,
+      "step": 645
+    },
+    {
+      "epoch": 0.04416360918603071,
+      "grad_norm": 1.971498966217041,
+      "learning_rate": 9.94538999864112e-05,
+      "loss": 7.2665,
+      "step": 650
+    },
+    {
+      "epoch": 0.04450332925669249,
+      "grad_norm": 1.4446492195129395,
+      "learning_rate": 9.944965348552793e-05,
+      "loss": 7.0674,
+      "step": 655
+    },
+    {
+      "epoch": 0.04484304932735426,
+      "grad_norm": 1.0143150091171265,
+      "learning_rate": 9.944540698464467e-05,
+      "loss": 6.7426,
+      "step": 660
+    },
+    {
+      "epoch": 0.045182769398016034,
+      "grad_norm": 1.3732986450195312,
+      "learning_rate": 9.944116048376138e-05,
+      "loss": 7.0236,
+      "step": 665
+    },
+    {
+      "epoch": 0.04552248946867781,
+      "grad_norm": 1.5511842966079712,
+      "learning_rate": 9.943691398287811e-05,
+      "loss": 7.2107,
+      "step": 670
+    },
+    {
+      "epoch": 0.04586220953933958,
+      "grad_norm": 1.4255778789520264,
+      "learning_rate": 9.943266748199484e-05,
+      "loss": 6.817,
+      "step": 675
+    },
+    {
+      "epoch": 0.04620192961000136,
+      "grad_norm": 1.0669182538986206,
+      "learning_rate": 9.942842098111157e-05,
+      "loss": 6.9347,
+      "step": 680
+    },
+    {
+      "epoch": 0.046541649680663134,
+      "grad_norm": 1.512604832649231,
+      "learning_rate": 9.94241744802283e-05,
+      "loss": 7.0817,
+      "step": 685
+    },
+    {
+      "epoch": 0.04688136975132491,
+      "grad_norm": 1.3859061002731323,
+      "learning_rate": 9.941992797934502e-05,
+      "loss": 7.1805,
+      "step": 690
+    },
+    {
+      "epoch": 0.04722108982198668,
+      "grad_norm": 3.252913236618042,
+      "learning_rate": 9.941568147846175e-05,
+      "loss": 6.7985,
+      "step": 695
+    },
+    {
+      "epoch": 0.04756080989264846,
+      "grad_norm": 1.4156177043914795,
+      "learning_rate": 9.941143497757848e-05,
+      "loss": 6.9955,
+      "step": 700
+    },
+    {
+      "epoch": 0.047900529963310234,
+      "grad_norm": 1.5510213375091553,
+      "learning_rate": 9.940718847669521e-05,
+      "loss": 7.0235,
+      "step": 705
+    },
+    {
+      "epoch": 0.048240250033972004,
+      "grad_norm": 1.3725285530090332,
+      "learning_rate": 9.940294197581194e-05,
+      "loss": 6.9692,
+      "step": 710
+    },
+    {
+      "epoch": 0.04857997010463378,
+      "grad_norm": 1.4986199140548706,
+      "learning_rate": 9.939869547492866e-05,
+      "loss": 6.6778,
+      "step": 715
+    },
+    {
+      "epoch": 0.04891969017529556,
+      "grad_norm": 1.2320705652236938,
+      "learning_rate": 9.939444897404539e-05,
+      "loss": 6.8953,
+      "step": 720
+    },
+    {
+      "epoch": 0.049259410245957334,
+      "grad_norm": 0.9118322134017944,
+      "learning_rate": 9.939020247316212e-05,
+      "loss": 6.7496,
+      "step": 725
+    },
+    {
+      "epoch": 0.049599130316619104,
+      "grad_norm": 3.4886631965637207,
+      "learning_rate": 9.938595597227885e-05,
+      "loss": 7.0242,
+      "step": 730
+    },
+    {
+      "epoch": 0.04993885038728088,
+      "grad_norm": 0.9548838138580322,
+      "learning_rate": 9.938170947139558e-05,
+      "loss": 6.6306,
+      "step": 735
+    },
+    {
+      "epoch": 0.05027857045794266,
+      "grad_norm": 0.8389047980308533,
+      "learning_rate": 9.93774629705123e-05,
+      "loss": 7.134,
+      "step": 740
+    },
+    {
+      "epoch": 0.05061829052860443,
+      "grad_norm": 5.4491801261901855,
+      "learning_rate": 9.937321646962903e-05,
+      "loss": 6.698,
+      "step": 745
+    },
+    {
+      "epoch": 0.050958010599266204,
+      "grad_norm": 1.3063551187515259,
+      "learning_rate": 9.936896996874576e-05,
+      "loss": 7.0113,
+      "step": 750
+    },
+    {
+      "epoch": 0.05129773066992798,
+      "grad_norm": 1.470941424369812,
+      "learning_rate": 9.936472346786249e-05,
+      "loss": 6.606,
+      "step": 755
+    },
+    {
+      "epoch": 0.05163745074058975,
+      "grad_norm": 1.9392439126968384,
+      "learning_rate": 9.936047696697922e-05,
+      "loss": 7.07,
+      "step": 760
+    },
+    {
+      "epoch": 0.05197717081125153,
+      "grad_norm": 0.9688730239868164,
+      "learning_rate": 9.935623046609594e-05,
+      "loss": 6.5451,
+      "step": 765
+    },
+    {
+      "epoch": 0.052316890881913304,
+      "grad_norm": 1.4289032220840454,
+      "learning_rate": 9.935198396521267e-05,
+      "loss": 6.8784,
+      "step": 770
+    },
+    {
+      "epoch": 0.05265661095257508,
+      "grad_norm": 1.4620697498321533,
+      "learning_rate": 9.93477374643294e-05,
+      "loss": 6.5151,
+      "step": 775
+    },
+    {
+      "epoch": 0.05299633102323685,
+      "grad_norm": 2.3521432876586914,
+      "learning_rate": 9.934349096344613e-05,
+      "loss": 6.5384,
+      "step": 780
+    },
+    {
+      "epoch": 0.05333605109389863,
+      "grad_norm": 3.160248041152954,
+      "learning_rate": 9.933924446256286e-05,
+      "loss": 6.7476,
+      "step": 785
+    },
+    {
+      "epoch": 0.053675771164560404,
+      "grad_norm": 1.3147598505020142,
+      "learning_rate": 9.933499796167957e-05,
+      "loss": 6.8376,
+      "step": 790
+    },
+    {
+      "epoch": 0.054015491235222174,
+      "grad_norm": 1.6566650867462158,
+      "learning_rate": 9.933075146079631e-05,
+      "loss": 6.2506,
+      "step": 795
+    },
+    {
+      "epoch": 0.05435521130588395,
+      "grad_norm": 0.9440861344337463,
+      "learning_rate": 9.932650495991304e-05,
+      "loss": 6.9212,
+      "step": 800
+    },
+    {
+      "epoch": 0.05469493137654573,
+      "grad_norm": 1.1842477321624756,
+      "learning_rate": 9.932225845902975e-05,
+      "loss": 6.6892,
+      "step": 805
+    },
+    {
+      "epoch": 0.055034651447207504,
+      "grad_norm": 1.1736949682235718,
+      "learning_rate": 9.93180119581465e-05,
+      "loss": 6.9224,
+      "step": 810
+    },
+    {
+      "epoch": 0.055374371517869274,
+      "grad_norm": 0.7181898951530457,
+      "learning_rate": 9.931376545726322e-05,
+      "loss": 6.5164,
+      "step": 815
+    },
+    {
+      "epoch": 0.05571409158853105,
+      "grad_norm": 0.9374647736549377,
+      "learning_rate": 9.930951895637994e-05,
+      "loss": 6.5026,
+      "step": 820
+    },
+    {
+      "epoch": 0.05605381165919283,
+      "grad_norm": 1.2754137516021729,
+      "learning_rate": 9.930527245549668e-05,
+      "loss": 6.4676,
+      "step": 825
+    },
+    {
+      "epoch": 0.0563935317298546,
+      "grad_norm": 1.0159765481948853,
+      "learning_rate": 9.930102595461341e-05,
+      "loss": 6.3455,
+      "step": 830
+    },
+    {
+      "epoch": 0.056733251800516374,
+      "grad_norm": 1.0118136405944824,
+      "learning_rate": 9.929677945373012e-05,
+      "loss": 6.4984,
+      "step": 835
+    },
+    {
+      "epoch": 0.05707297187117815,
+      "grad_norm": 0.9740552306175232,
+      "learning_rate": 9.929253295284686e-05,
+      "loss": 6.665,
+      "step": 840
+    },
+    {
+      "epoch": 0.05741269194183993,
+      "grad_norm": 2.6464507579803467,
+      "learning_rate": 9.928828645196359e-05,
+      "loss": 6.5277,
+      "step": 845
+    },
+    {
+      "epoch": 0.0577524120125017,
+      "grad_norm": 1.1687380075454712,
+      "learning_rate": 9.92840399510803e-05,
+      "loss": 6.5141,
+      "step": 850
+    },
+    {
+      "epoch": 0.058092132083163474,
+      "grad_norm": 1.0684833526611328,
+      "learning_rate": 9.927979345019705e-05,
+      "loss": 6.2975,
+      "step": 855
+    },
+    {
+      "epoch": 0.05843185215382525,
+      "grad_norm": 3.8520267009735107,
+      "learning_rate": 9.927554694931378e-05,
+      "loss": 5.9187,
+      "step": 860
+    },
+    {
+      "epoch": 0.05877157222448702,
+      "grad_norm": 1.048731803894043,
+      "learning_rate": 9.927130044843049e-05,
+      "loss": 6.3076,
+      "step": 865
+    },
+    {
+      "epoch": 0.0591112922951488,
+      "grad_norm": 0.8034812808036804,
+      "learning_rate": 9.926705394754723e-05,
+      "loss": 6.5594,
+      "step": 870
+    },
+    {
+      "epoch": 0.059451012365810574,
+      "grad_norm": 0.9210667610168457,
+      "learning_rate": 9.926280744666395e-05,
+      "loss": 6.0014,
+      "step": 875
+    },
+    {
+      "epoch": 0.059790732436472344,
+      "grad_norm": 1.0861904621124268,
+      "learning_rate": 9.925856094578067e-05,
+      "loss": 6.2432,
+      "step": 880
+    },
+    {
+      "epoch": 0.06013045250713412,
+      "grad_norm": 0.8701607584953308,
+      "learning_rate": 9.925431444489742e-05,
+      "loss": 6.1582,
+      "step": 885
+    },
+    {
+      "epoch": 0.0604701725777959,
+      "grad_norm": 0.691939651966095,
+      "learning_rate": 9.925006794401413e-05,
+      "loss": 6.487,
+      "step": 890
+    },
+    {
+      "epoch": 0.060809892648457674,
+      "grad_norm": 1.1983147859573364,
+      "learning_rate": 9.924582144313086e-05,
+      "loss": 6.1766,
+      "step": 895
+    },
+    {
+      "epoch": 0.061149612719119444,
+      "grad_norm": 1.1613506078720093,
+      "learning_rate": 9.92415749422476e-05,
+      "loss": 6.5303,
+      "step": 900
+    },
+    {
+      "epoch": 0.06148933278978122,
+      "grad_norm": 0.8394651412963867,
+      "learning_rate": 9.923732844136431e-05,
+      "loss": 6.1502,
+      "step": 905
+    },
+    {
+      "epoch": 0.061829052860443,
+      "grad_norm": 1.2242004871368408,
+      "learning_rate": 9.923308194048104e-05,
+      "loss": 6.2054,
+      "step": 910
+    },
+    {
+      "epoch": 0.06216877293110477,
+      "grad_norm": 1.1255033016204834,
+      "learning_rate": 9.922883543959778e-05,
+      "loss": 6.4181,
+      "step": 915
+    },
+    {
+      "epoch": 0.06250849300176654,
+      "grad_norm": 0.7849110960960388,
+      "learning_rate": 9.92245889387145e-05,
+      "loss": 6.1411,
+      "step": 920
+    },
+    {
+      "epoch": 0.06284821307242831,
+      "grad_norm": 1.0032676458358765,
+      "learning_rate": 9.922034243783123e-05,
+      "loss": 6.4892,
+      "step": 925
+    },
+    {
+      "epoch": 0.0631879331430901,
+      "grad_norm": 1.127551555633545,
+      "learning_rate": 9.921609593694797e-05,
+      "loss": 6.3768,
+      "step": 930
+    },
+    {
+      "epoch": 0.06352765321375187,
+      "grad_norm": 1.0425925254821777,
+      "learning_rate": 9.921184943606468e-05,
+      "loss": 6.4448,
+      "step": 935
+    },
+    {
+      "epoch": 0.06386737328441364,
+      "grad_norm": 1.1642504930496216,
+      "learning_rate": 9.920760293518142e-05,
+      "loss": 5.7809,
+      "step": 940
+    },
+    {
+      "epoch": 0.06420709335507542,
+      "grad_norm": 1.8521403074264526,
+      "learning_rate": 9.920335643429814e-05,
+      "loss": 5.9249,
+      "step": 945
+    },
+    {
+      "epoch": 0.06454681342573719,
+      "grad_norm": 1.073219895362854,
+      "learning_rate": 9.919910993341487e-05,
+      "loss": 6.0477,
+      "step": 950
+    },
+    {
+      "epoch": 0.06488653349639897,
+      "grad_norm": 1.2109575271606445,
+      "learning_rate": 9.919486343253161e-05,
+      "loss": 6.0364,
+      "step": 955
+    },
+    {
+      "epoch": 0.06522625356706074,
+      "grad_norm": 1.1780409812927246,
+      "learning_rate": 9.919061693164832e-05,
+      "loss": 6.4147,
+      "step": 960
+    },
+    {
+      "epoch": 0.06556597363772251,
+      "grad_norm": 0.8810535073280334,
+      "learning_rate": 9.918637043076505e-05,
+      "loss": 6.0036,
+      "step": 965
+    },
+    {
+      "epoch": 0.0659056937083843,
+      "grad_norm": 0.7648366093635559,
+      "learning_rate": 9.918212392988179e-05,
+      "loss": 6.2567,
+      "step": 970
+    },
+    {
+      "epoch": 0.06624541377904607,
+      "grad_norm": 2.192458391189575,
+      "learning_rate": 9.91778774289985e-05,
+      "loss": 6.144,
+      "step": 975
+    },
+    {
+      "epoch": 0.06658513384970784,
+      "grad_norm": 1.2390516996383667,
+      "learning_rate": 9.917363092811523e-05,
+      "loss": 6.0287,
+      "step": 980
+    },
+    {
+      "epoch": 0.06692485392036962,
+      "grad_norm": 0.8258079886436462,
+      "learning_rate": 9.916938442723198e-05,
+      "loss": 6.2043,
+      "step": 985
+    },
+    {
+      "epoch": 0.06726457399103139,
+      "grad_norm": 0.9516924023628235,
+      "learning_rate": 9.916513792634869e-05,
+      "loss": 6.0747,
+      "step": 990
+    },
+    {
+      "epoch": 0.06760429406169316,
+      "grad_norm": 0.856916069984436,
+      "learning_rate": 9.916089142546542e-05,
+      "loss": 5.926,
+      "step": 995
+    },
+    {
+      "epoch": 0.06794401413235494,
+      "grad_norm": 0.8324723839759827,
+      "learning_rate": 9.915664492458216e-05,
+      "loss": 6.2707,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06828373420301671,
+      "grad_norm": 0.7908216714859009,
+      "learning_rate": 9.915239842369887e-05,
+      "loss": 5.967,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06862345427367848,
+      "grad_norm": 0.9094476103782654,
+      "learning_rate": 9.91481519228156e-05,
+      "loss": 5.6132,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06896317434434027,
+      "grad_norm": 0.9734240770339966,
+      "learning_rate": 9.914390542193233e-05,
+      "loss": 5.9794,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06930289441500204,
+      "grad_norm": 0.8310399055480957,
+      "learning_rate": 9.913965892104906e-05,
+      "loss": 6.0146,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0696426144856638,
+      "grad_norm": 0.9436420798301697,
+      "learning_rate": 9.913541242016579e-05,
+      "loss": 5.9812,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06998233455632559,
+      "grad_norm": 1.2283395528793335,
+      "learning_rate": 9.913116591928251e-05,
+      "loss": 5.9537,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07032205462698736,
+      "grad_norm": 0.8751355409622192,
+      "learning_rate": 9.912691941839924e-05,
+      "loss": 5.9992,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07066177469764914,
+      "grad_norm": 0.6706697344779968,
+      "learning_rate": 9.912267291751597e-05,
+      "loss": 6.0753,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07100149476831091,
+      "grad_norm": 0.7029627561569214,
+      "learning_rate": 9.91184264166327e-05,
+      "loss": 6.0136,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07134121483897268,
+      "grad_norm": 0.6171499490737915,
+      "learning_rate": 9.911417991574943e-05,
+      "loss": 6.0223,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07168093490963447,
+      "grad_norm": 2.5934255123138428,
+      "learning_rate": 9.910993341486615e-05,
+      "loss": 5.807,
+      "step": 1055
+    },
+    {
+      "epoch": 0.07202065498029624,
+      "grad_norm": 0.9291547536849976,
+      "learning_rate": 9.910568691398288e-05,
+      "loss": 5.7086,
+      "step": 1060
+    },
+    {
+      "epoch": 0.072360375050958,
+      "grad_norm": 1.4394763708114624,
+      "learning_rate": 9.910144041309961e-05,
+      "loss": 5.5023,
+      "step": 1065
+    },
+    {
+      "epoch": 0.07270009512161979,
+      "grad_norm": 0.6298092603683472,
+      "learning_rate": 9.909719391221634e-05,
+      "loss": 5.9972,
+      "step": 1070
+    },
+    {
+      "epoch": 0.07303981519228156,
+      "grad_norm": 0.6151909232139587,
+      "learning_rate": 9.909294741133307e-05,
+      "loss": 5.6389,
+      "step": 1075
+    },
+    {
+      "epoch": 0.07337953526294333,
+      "grad_norm": 1.1861008405685425,
+      "learning_rate": 9.90887009104498e-05,
+      "loss": 6.2689,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07371925533360511,
+      "grad_norm": 0.7876234650611877,
+      "learning_rate": 9.908445440956652e-05,
+      "loss": 5.5448,
+      "step": 1085
+    },
+    {
+      "epoch": 0.07405897540426688,
+      "grad_norm": 0.592897891998291,
+      "learning_rate": 9.908020790868325e-05,
+      "loss": 5.8057,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07439869547492865,
+      "grad_norm": 0.9189316034317017,
+      "learning_rate": 9.907596140779998e-05,
+      "loss": 6.0782,
+      "step": 1095
+    },
+    {
+      "epoch": 0.07473841554559044,
+      "grad_norm": 0.6605345010757446,
+      "learning_rate": 9.90717149069167e-05,
+      "loss": 5.6442,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0750781356162522,
+      "grad_norm": 0.6724756956100464,
+      "learning_rate": 9.906746840603343e-05,
+      "loss": 6.2757,
+      "step": 1105
+    },
+    {
+      "epoch": 0.07541785568691398,
+      "grad_norm": 0.8074867725372314,
+      "learning_rate": 9.906322190515016e-05,
+      "loss": 5.6868,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07575757575757576,
+      "grad_norm": 2.2021851539611816,
+      "learning_rate": 9.905897540426689e-05,
+      "loss": 5.755,
+      "step": 1115
+    },
+    {
+      "epoch": 0.07609729582823753,
+      "grad_norm": 0.7941934466362,
+      "learning_rate": 9.905472890338362e-05,
+      "loss": 5.6318,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07643701589889931,
+      "grad_norm": 0.9947513937950134,
+      "learning_rate": 9.905048240250035e-05,
+      "loss": 5.9247,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07677673596956108,
+      "grad_norm": 0.6511673927307129,
+      "learning_rate": 9.904623590161707e-05,
+      "loss": 5.6326,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07711645604022285,
+      "grad_norm": 0.6497818231582642,
+      "learning_rate": 9.90419894007338e-05,
+      "loss": 5.8753,
+      "step": 1135
+    },
+    {
+      "epoch": 0.07745617611088464,
+      "grad_norm": 0.5531424880027771,
+      "learning_rate": 9.903774289985053e-05,
+      "loss": 5.2715,
+      "step": 1140
+    },
+    {
+      "epoch": 0.0777958961815464,
+      "grad_norm": 0.7292714715003967,
+      "learning_rate": 9.903349639896725e-05,
+      "loss": 5.7824,
+      "step": 1145
+    },
+    {
+      "epoch": 0.07813561625220818,
+      "grad_norm": 0.6802114248275757,
+      "learning_rate": 9.902924989808399e-05,
+      "loss": 5.7912,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07847533632286996,
+      "grad_norm": 0.680204451084137,
+      "learning_rate": 9.902500339720071e-05,
+      "loss": 5.961,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07881505639353173,
+      "grad_norm": 0.596501350402832,
+      "learning_rate": 9.902075689631743e-05,
+      "loss": 5.5833,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0791547764641935,
+      "grad_norm": 0.622715950012207,
+      "learning_rate": 9.901651039543417e-05,
+      "loss": 5.933,
+      "step": 1165
+    },
+    {
+      "epoch": 0.07949449653485528,
+      "grad_norm": 0.9008530974388123,
+      "learning_rate": 9.90122638945509e-05,
+      "loss": 5.568,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07983421660551705,
+      "grad_norm": 0.5429263114929199,
+      "learning_rate": 9.900801739366761e-05,
+      "loss": 5.2962,
+      "step": 1175
+    },
+    {
+      "epoch": 0.08017393667617882,
+      "grad_norm": 0.6079940795898438,
+      "learning_rate": 9.900377089278435e-05,
+      "loss": 5.8575,
+      "step": 1180
+    },
+    {
+      "epoch": 0.0805136567468406,
+      "grad_norm": 0.6796315312385559,
+      "learning_rate": 9.899952439190108e-05,
+      "loss": 5.6058,
+      "step": 1185
+    },
+    {
+      "epoch": 0.08085337681750238,
+      "grad_norm": 0.6909620761871338,
+      "learning_rate": 9.89952778910178e-05,
+      "loss": 5.6206,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08119309688816416,
+      "grad_norm": 0.8451843857765198,
+      "learning_rate": 9.899103139013454e-05,
+      "loss": 5.5017,
+      "step": 1195
+    },
+    {
+      "epoch": 0.08153281695882593,
+      "grad_norm": 0.6521558165550232,
+      "learning_rate": 9.898678488925127e-05,
+      "loss": 5.6647,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0818725370294877,
+      "grad_norm": 0.4295422434806824,
+      "learning_rate": 9.898253838836798e-05,
+      "loss": 5.96,
+      "step": 1205
+    },
+    {
+      "epoch": 0.08221225710014948,
+      "grad_norm": 1.2797423601150513,
+      "learning_rate": 9.897829188748472e-05,
+      "loss": 5.5158,
+      "step": 1210
+    },
+    {
+      "epoch": 0.08255197717081125,
+      "grad_norm": 0.663374125957489,
+      "learning_rate": 9.897404538660144e-05,
+      "loss": 5.5387,
+      "step": 1215
+    },
+    {
+      "epoch": 0.08289169724147302,
+      "grad_norm": 0.6130101084709167,
+      "learning_rate": 9.896979888571817e-05,
+      "loss": 5.3825,
+      "step": 1220
+    },
+    {
+      "epoch": 0.0832314173121348,
+      "grad_norm": 0.9059043526649475,
+      "learning_rate": 9.896555238483491e-05,
+      "loss": 5.6168,
+      "step": 1225
+    },
+    {
+      "epoch": 0.08357113738279658,
+      "grad_norm": 0.9198205471038818,
+      "learning_rate": 9.896130588395162e-05,
+      "loss": 5.7126,
+      "step": 1230
+    },
+    {
+      "epoch": 0.08391085745345835,
+      "grad_norm": 0.6826533079147339,
+      "learning_rate": 9.895705938306835e-05,
+      "loss": 5.5344,
+      "step": 1235
+    },
+    {
+      "epoch": 0.08425057752412013,
+      "grad_norm": 0.6488471031188965,
+      "learning_rate": 9.895281288218509e-05,
+      "loss": 5.6311,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0845902975947819,
+      "grad_norm": 1.4997718334197998,
+      "learning_rate": 9.89485663813018e-05,
+      "loss": 5.3847,
+      "step": 1245
+    },
+    {
+      "epoch": 0.08493001766544367,
+      "grad_norm": 0.8614689111709595,
+      "learning_rate": 9.894431988041853e-05,
+      "loss": 5.7996,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08526973773610545,
+      "grad_norm": 0.910275936126709,
+      "learning_rate": 9.894007337953527e-05,
+      "loss": 5.6557,
+      "step": 1255
+    },
+    {
+      "epoch": 0.08560945780676722,
+      "grad_norm": 0.8584810495376587,
+      "learning_rate": 9.893582687865199e-05,
+      "loss": 5.6384,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08594917787742899,
+      "grad_norm": 0.5326058864593506,
+      "learning_rate": 9.893158037776872e-05,
+      "loss": 5.4521,
+      "step": 1265
+    },
+    {
+      "epoch": 0.08628889794809078,
+      "grad_norm": 1.008244276046753,
+      "learning_rate": 9.892733387688546e-05,
+      "loss": 5.6282,
+      "step": 1270
+    },
+    {
+      "epoch": 0.08662861801875255,
+      "grad_norm": 0.9059062600135803,
+      "learning_rate": 9.892308737600217e-05,
+      "loss": 5.7698,
+      "step": 1275
+    },
+    {
+      "epoch": 0.08696833808941433,
+      "grad_norm": 0.754760205745697,
+      "learning_rate": 9.891884087511891e-05,
+      "loss": 5.7735,
+      "step": 1280
+    },
+    {
+      "epoch": 0.0873080581600761,
+      "grad_norm": 0.6785455346107483,
+      "learning_rate": 9.891459437423564e-05,
+      "loss": 5.8881,
+      "step": 1285
+    },
+    {
+      "epoch": 0.08764777823073787,
+      "grad_norm": 0.8128915429115295,
+      "learning_rate": 9.891034787335236e-05,
+      "loss": 5.5169,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08798749830139965,
+      "grad_norm": 0.912551760673523,
+      "learning_rate": 9.89061013724691e-05,
+      "loss": 5.5467,
+      "step": 1295
+    },
+    {
+      "epoch": 0.08832721837206142,
+      "grad_norm": 2.0163161754608154,
+      "learning_rate": 9.890185487158581e-05,
+      "loss": 5.5448,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08866693844272319,
+      "grad_norm": 0.5310774445533752,
+      "learning_rate": 9.889760837070254e-05,
+      "loss": 5.5268,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08900665851338498,
+      "grad_norm": 0.603813886642456,
+      "learning_rate": 9.889336186981928e-05,
+      "loss": 5.6777,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08934637858404675,
+      "grad_norm": 0.431869238615036,
+      "learning_rate": 9.8889115368936e-05,
+      "loss": 5.5403,
+      "step": 1315
+    },
+    {
+      "epoch": 0.08968609865470852,
+      "grad_norm": 0.7721471190452576,
+      "learning_rate": 9.888486886805273e-05,
+      "loss": 5.2792,
+      "step": 1320
+    },
+    {
+      "epoch": 0.0900258187253703,
+      "grad_norm": 1.0440256595611572,
+      "learning_rate": 9.888062236716947e-05,
+      "loss": 5.1811,
+      "step": 1325
+    },
+    {
+      "epoch": 0.09036553879603207,
+      "grad_norm": 0.5089631080627441,
+      "learning_rate": 9.887637586628618e-05,
+      "loss": 5.5219,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09070525886669384,
+      "grad_norm": 0.5013499855995178,
+      "learning_rate": 9.887212936540291e-05,
+      "loss": 5.5286,
+      "step": 1335
+    },
+    {
+      "epoch": 0.09104497893735562,
+      "grad_norm": 0.8858599662780762,
+      "learning_rate": 9.886788286451965e-05,
+      "loss": 5.6356,
+      "step": 1340
+    },
+    {
+      "epoch": 0.09138469900801739,
+      "grad_norm": 0.45014268159866333,
+      "learning_rate": 9.886363636363637e-05,
+      "loss": 5.1792,
+      "step": 1345
+    },
+    {
+      "epoch": 0.09172441907867916,
+      "grad_norm": 0.6357929706573486,
+      "learning_rate": 9.88593898627531e-05,
+      "loss": 5.6234,
+      "step": 1350
+    },
+    {
+      "epoch": 0.09206413914934095,
+      "grad_norm": 4.850500106811523,
+      "learning_rate": 9.885514336186984e-05,
+      "loss": 5.3482,
+      "step": 1355
+    },
+    {
+      "epoch": 0.09240385922000272,
+      "grad_norm": 0.477006196975708,
+      "learning_rate": 9.885089686098655e-05,
+      "loss": 5.2999,
+      "step": 1360
+    },
+    {
+      "epoch": 0.0927435792906645,
+      "grad_norm": 0.5572239756584167,
+      "learning_rate": 9.884665036010328e-05,
+      "loss": 5.4801,
+      "step": 1365
+    },
+    {
+      "epoch": 0.09308329936132627,
+      "grad_norm": 0.3890332877635956,
+      "learning_rate": 9.884240385922e-05,
+      "loss": 5.3536,
+      "step": 1370
+    },
+    {
+      "epoch": 0.09342301943198804,
+      "grad_norm": 0.43894070386886597,
+      "learning_rate": 9.883815735833673e-05,
+      "loss": 5.6419,
+      "step": 1375
+    },
+    {
+      "epoch": 0.09376273950264982,
+      "grad_norm": 0.48212140798568726,
+      "learning_rate": 9.883391085745346e-05,
+      "loss": 5.431,
+      "step": 1380
+    },
+    {
+      "epoch": 0.09410245957331159,
+      "grad_norm": 0.5145598649978638,
+      "learning_rate": 9.882966435657019e-05,
+      "loss": 5.7753,
+      "step": 1385
+    },
+    {
+      "epoch": 0.09444217964397336,
+      "grad_norm": 0.4795299172401428,
+      "learning_rate": 9.882541785568692e-05,
+      "loss": 5.5992,
+      "step": 1390
+    },
+    {
+      "epoch": 0.09478189971463515,
+      "grad_norm": 0.4439328610897064,
+      "learning_rate": 9.882117135480365e-05,
+      "loss": 5.3292,
+      "step": 1395
+    },
+    {
+      "epoch": 0.09512161978529692,
+      "grad_norm": 0.6526133418083191,
+      "learning_rate": 9.881692485392037e-05,
+      "loss": 5.3765,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09546133985595869,
+      "grad_norm": 0.8982023596763611,
+      "learning_rate": 9.88126783530371e-05,
+      "loss": 5.4783,
+      "step": 1405
+    },
+    {
+      "epoch": 0.09580105992662047,
+      "grad_norm": 0.46160343289375305,
+      "learning_rate": 9.880843185215383e-05,
+      "loss": 5.502,
+      "step": 1410
+    },
+    {
+      "epoch": 0.09614077999728224,
+      "grad_norm": 0.6750124096870422,
+      "learning_rate": 9.880418535127056e-05,
+      "loss": 5.4252,
+      "step": 1415
+    },
+    {
+      "epoch": 0.09648050006794401,
+      "grad_norm": 0.42901554703712463,
+      "learning_rate": 9.879993885038729e-05,
+      "loss": 5.3442,
+      "step": 1420
+    },
+    {
+      "epoch": 0.09682022013860579,
+      "grad_norm": 0.6184918284416199,
+      "learning_rate": 9.879569234950401e-05,
+      "loss": 5.3717,
+      "step": 1425
+    },
+    {
+      "epoch": 0.09715994020926756,
+      "grad_norm": 0.5006517171859741,
+      "learning_rate": 9.879144584862074e-05,
+      "loss": 5.4458,
+      "step": 1430
+    },
+    {
+      "epoch": 0.09749966027992933,
+      "grad_norm": 0.4495384693145752,
+      "learning_rate": 9.878719934773747e-05,
+      "loss": 5.263,
+      "step": 1435
+    },
+    {
+      "epoch": 0.09783938035059112,
+      "grad_norm": 0.4285268783569336,
+      "learning_rate": 9.87829528468542e-05,
+      "loss": 5.2207,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09817910042125289,
+      "grad_norm": 1.0460352897644043,
+      "learning_rate": 9.877870634597093e-05,
+      "loss": 5.3448,
+      "step": 1445
+    },
+    {
+      "epoch": 0.09851882049191467,
+      "grad_norm": 0.6291869878768921,
+      "learning_rate": 9.877445984508765e-05,
+      "loss": 5.3182,
+      "step": 1450
+    },
+    {
+      "epoch": 0.09885854056257644,
+      "grad_norm": 1.0043153762817383,
+      "learning_rate": 9.877021334420438e-05,
+      "loss": 5.4226,
+      "step": 1455
+    },
+    {
+      "epoch": 0.09919826063323821,
+      "grad_norm": 0.7458539009094238,
+      "learning_rate": 9.876596684332111e-05,
+      "loss": 5.6298,
+      "step": 1460
+    },
+    {
+      "epoch": 0.09953798070389999,
+      "grad_norm": 0.45767852663993835,
+      "learning_rate": 9.876172034243784e-05,
+      "loss": 5.5378,
+      "step": 1465
+    },
+    {
+      "epoch": 0.09987770077456176,
+      "grad_norm": 0.4586849510669708,
+      "learning_rate": 9.875747384155457e-05,
+      "loss": 5.7105,
+      "step": 1470
+    },
+    {
+      "epoch": 0.10021742084522353,
+      "grad_norm": 0.37701913714408875,
+      "learning_rate": 9.87532273406713e-05,
+      "loss": 5.1507,
+      "step": 1475
+    },
+    {
+      "epoch": 0.10055714091588532,
+      "grad_norm": 0.42394065856933594,
+      "learning_rate": 9.874898083978802e-05,
+      "loss": 5.5164,
+      "step": 1480
+    },
+    {
+      "epoch": 0.10089686098654709,
+      "grad_norm": 0.6020697355270386,
+      "learning_rate": 9.874473433890475e-05,
+      "loss": 5.3247,
+      "step": 1485
+    },
+    {
+      "epoch": 0.10123658105720885,
+      "grad_norm": 0.5261949300765991,
+      "learning_rate": 9.874048783802148e-05,
+      "loss": 5.5159,
+      "step": 1490
+    },
+    {
+      "epoch": 0.10157630112787064,
+      "grad_norm": 0.4316195547580719,
+      "learning_rate": 9.87362413371382e-05,
+      "loss": 5.4615,
+      "step": 1495
+    },
+    {
+      "epoch": 0.10191602119853241,
+      "grad_norm": 0.3691781759262085,
+      "learning_rate": 9.873199483625492e-05,
+      "loss": 5.5583,
+      "step": 1500
+    },
+    {
+      "epoch": 0.10225574126919418,
+      "grad_norm": 0.5686614513397217,
+      "learning_rate": 9.872774833537166e-05,
+      "loss": 5.5936,
+      "step": 1505
+    },
+    {
+      "epoch": 0.10259546133985596,
+      "grad_norm": 0.3296063542366028,
+      "learning_rate": 9.872350183448839e-05,
+      "loss": 5.4636,
+      "step": 1510
+    },
+    {
+      "epoch": 0.10293518141051773,
+      "grad_norm": 0.46032190322875977,
+      "learning_rate": 9.87192553336051e-05,
+      "loss": 5.1794,
+      "step": 1515
+    },
+    {
+      "epoch": 0.1032749014811795,
+      "grad_norm": 0.4383249282836914,
+      "learning_rate": 9.871500883272185e-05,
+      "loss": 5.3524,
+      "step": 1520
+    },
+    {
+      "epoch": 0.10361462155184128,
+      "grad_norm": 0.5257749557495117,
+      "learning_rate": 9.871076233183857e-05,
+      "loss": 5.5045,
+      "step": 1525
+    },
+    {
+      "epoch": 0.10395434162250305,
+      "grad_norm": 0.5201256275177002,
+      "learning_rate": 9.870651583095529e-05,
+      "loss": 5.3214,
+      "step": 1530
+    },
+    {
+      "epoch": 0.10429406169316484,
+      "grad_norm": 0.7715117931365967,
+      "learning_rate": 9.870226933007203e-05,
+      "loss": 5.3583,
+      "step": 1535
+    },
+    {
+      "epoch": 0.10463378176382661,
+      "grad_norm": 0.34586212038993835,
+      "learning_rate": 9.869802282918876e-05,
+      "loss": 5.3337,
+      "step": 1540
+    },
+    {
+      "epoch": 0.10497350183448838,
+      "grad_norm": 0.45313313603401184,
+      "learning_rate": 9.869377632830547e-05,
+      "loss": 5.2788,
+      "step": 1545
+    },
+    {
+      "epoch": 0.10531322190515016,
+      "grad_norm": 0.4078027904033661,
+      "learning_rate": 9.868952982742221e-05,
+      "loss": 5.4364,
+      "step": 1550
+    },
+    {
+      "epoch": 0.10565294197581193,
+      "grad_norm": 0.4670262038707733,
+      "learning_rate": 9.868528332653894e-05,
+      "loss": 5.308,
+      "step": 1555
+    },
+    {
+      "epoch": 0.1059926620464737,
+      "grad_norm": 0.4951310455799103,
+      "learning_rate": 9.868103682565566e-05,
+      "loss": 5.5171,
+      "step": 1560
+    },
+    {
+      "epoch": 0.10633238211713548,
+      "grad_norm": 0.7351198792457581,
+      "learning_rate": 9.86767903247724e-05,
+      "loss": 5.6146,
+      "step": 1565
+    },
+    {
+      "epoch": 0.10667210218779725,
+      "grad_norm": 0.5064637660980225,
+      "learning_rate": 9.867254382388911e-05,
+      "loss": 5.3591,
+      "step": 1570
+    },
+    {
+      "epoch": 0.10701182225845902,
+      "grad_norm": 0.39143896102905273,
+      "learning_rate": 9.866829732300584e-05,
+      "loss": 5.3523,
+      "step": 1575
+    },
+    {
+      "epoch": 0.10735154232912081,
+      "grad_norm": 1.2670384645462036,
+      "learning_rate": 9.866405082212258e-05,
+      "loss": 5.2359,
+      "step": 1580
+    },
+    {
+      "epoch": 0.10769126239978258,
+      "grad_norm": 0.3745839297771454,
+      "learning_rate": 9.86598043212393e-05,
+      "loss": 5.2194,
+      "step": 1585
+    },
+    {
+      "epoch": 0.10803098247044435,
+      "grad_norm": 0.26325303316116333,
+      "learning_rate": 9.865555782035602e-05,
+      "loss": 5.1649,
+      "step": 1590
+    },
+    {
+      "epoch": 0.10837070254110613,
+      "grad_norm": 0.3311369717121124,
+      "learning_rate": 9.865131131947277e-05,
+      "loss": 5.326,
+      "step": 1595
+    },
+    {
+      "epoch": 0.1087104226117679,
+      "grad_norm": 0.4302009046077728,
+      "learning_rate": 9.864706481858948e-05,
+      "loss": 5.1757,
+      "step": 1600
+    },
+    {
+      "epoch": 0.10905014268242967,
+      "grad_norm": 0.5953149795532227,
+      "learning_rate": 9.864281831770621e-05,
+      "loss": 5.3937,
+      "step": 1605
+    },
+    {
+      "epoch": 0.10938986275309145,
+      "grad_norm": 0.4650028645992279,
+      "learning_rate": 9.863857181682295e-05,
+      "loss": 5.4321,
+      "step": 1610
+    },
+    {
+      "epoch": 0.10972958282375322,
+      "grad_norm": 1.5760172605514526,
+      "learning_rate": 9.863432531593966e-05,
+      "loss": 5.1935,
+      "step": 1615
+    },
+    {
+      "epoch": 0.11006930289441501,
+      "grad_norm": 0.4063778221607208,
+      "learning_rate": 9.86300788150564e-05,
+      "loss": 5.0892,
+      "step": 1620
+    },
+    {
+      "epoch": 0.11040902296507678,
+      "grad_norm": 0.3407844603061676,
+      "learning_rate": 9.862583231417313e-05,
+      "loss": 5.2536,
+      "step": 1625
+    },
+    {
+      "epoch": 0.11074874303573855,
+      "grad_norm": 0.5247534513473511,
+      "learning_rate": 9.862158581328985e-05,
+      "loss": 5.3708,
+      "step": 1630
+    },
+    {
+      "epoch": 0.11108846310640033,
+      "grad_norm": 1.0360844135284424,
+      "learning_rate": 9.861733931240659e-05,
+      "loss": 5.3132,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1114281831770621,
+      "grad_norm": 0.5687776803970337,
+      "learning_rate": 9.86130928115233e-05,
+      "loss": 5.2554,
+      "step": 1640
+    },
+    {
+      "epoch": 0.11176790324772387,
+      "grad_norm": 0.3441666066646576,
+      "learning_rate": 9.860884631064003e-05,
+      "loss": 5.1661,
+      "step": 1645
+    },
+    {
+      "epoch": 0.11210762331838565,
+      "grad_norm": 0.6809844970703125,
+      "learning_rate": 9.860459980975677e-05,
+      "loss": 5.3175,
+      "step": 1650
+    },
+    {
+      "epoch": 0.11244734338904742,
+      "grad_norm": 0.4880785644054413,
+      "learning_rate": 9.860035330887349e-05,
+      "loss": 5.3368,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1127870634597092,
+      "grad_norm": 0.5996628999710083,
+      "learning_rate": 9.859610680799022e-05,
+      "loss": 5.2202,
+      "step": 1660
+    },
+    {
+      "epoch": 0.11312678353037098,
+      "grad_norm": 0.3819567561149597,
+      "learning_rate": 9.859186030710696e-05,
+      "loss": 5.2171,
+      "step": 1665
+    },
+    {
+      "epoch": 0.11346650360103275,
+      "grad_norm": 0.8039321899414062,
+      "learning_rate": 9.858761380622367e-05,
+      "loss": 5.1053,
+      "step": 1670
+    },
+    {
+      "epoch": 0.11380622367169452,
+      "grad_norm": 0.7948293685913086,
+      "learning_rate": 9.85833673053404e-05,
+      "loss": 5.1902,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1141459437423563,
+      "grad_norm": 0.45758846402168274,
+      "learning_rate": 9.857912080445714e-05,
+      "loss": 5.2766,
+      "step": 1680
+    },
+    {
+      "epoch": 0.11448566381301807,
+      "grad_norm": 0.33638903498649597,
+      "learning_rate": 9.857487430357386e-05,
+      "loss": 5.1641,
+      "step": 1685
+    },
+    {
+      "epoch": 0.11482538388367985,
+      "grad_norm": 0.3370652496814728,
+      "learning_rate": 9.857062780269058e-05,
+      "loss": 4.8306,
+      "step": 1690
+    },
+    {
+      "epoch": 0.11516510395434162,
+      "grad_norm": 0.30151012539863586,
+      "learning_rate": 9.856638130180733e-05,
+      "loss": 4.9977,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1155048240250034,
+      "grad_norm": 0.4379021227359772,
+      "learning_rate": 9.856213480092404e-05,
+      "loss": 5.2388,
+      "step": 1700
+    },
+    {
+      "epoch": 0.11584454409566518,
+      "grad_norm": 0.5139544606208801,
+      "learning_rate": 9.855788830004077e-05,
+      "loss": 5.0948,
+      "step": 1705
+    },
+    {
+      "epoch": 0.11618426416632695,
+      "grad_norm": 0.2696703374385834,
+      "learning_rate": 9.855364179915751e-05,
+      "loss": 5.1955,
+      "step": 1710
+    },
+    {
+      "epoch": 0.11652398423698872,
+      "grad_norm": 0.43758052587509155,
+      "learning_rate": 9.854939529827422e-05,
+      "loss": 5.1093,
+      "step": 1715
+    },
+    {
+      "epoch": 0.1168637043076505,
+      "grad_norm": 0.45877712965011597,
+      "learning_rate": 9.854514879739095e-05,
+      "loss": 5.1859,
+      "step": 1720
+    },
+    {
+      "epoch": 0.11720342437831227,
+      "grad_norm": 0.5116316676139832,
+      "learning_rate": 9.854090229650768e-05,
+      "loss": 4.9734,
+      "step": 1725
+    },
+    {
+      "epoch": 0.11754314444897404,
+      "grad_norm": 0.3733248710632324,
+      "learning_rate": 9.853665579562441e-05,
+      "loss": 4.9164,
+      "step": 1730
+    },
+    {
+      "epoch": 0.11788286451963582,
+      "grad_norm": 0.9365966320037842,
+      "learning_rate": 9.853240929474114e-05,
+      "loss": 5.1925,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1182225845902976,
+      "grad_norm": 0.43242642283439636,
+      "learning_rate": 9.852816279385786e-05,
+      "loss": 5.2445,
+      "step": 1740
+    },
+    {
+      "epoch": 0.11856230466095936,
+      "grad_norm": 0.28928515315055847,
+      "learning_rate": 9.852391629297459e-05,
+      "loss": 5.2208,
+      "step": 1745
+    },
+    {
+      "epoch": 0.11890202473162115,
+      "grad_norm": 0.4645937383174896,
+      "learning_rate": 9.851966979209132e-05,
+      "loss": 5.1848,
+      "step": 1750
+    },
+    {
+      "epoch": 0.11924174480228292,
+      "grad_norm": 0.3553106486797333,
+      "learning_rate": 9.851542329120805e-05,
+      "loss": 5.0109,
+      "step": 1755
+    },
+    {
+      "epoch": 0.11958146487294469,
+      "grad_norm": 0.47933322191238403,
+      "learning_rate": 9.851117679032478e-05,
+      "loss": 5.2334,
+      "step": 1760
+    },
+    {
+      "epoch": 0.11992118494360647,
+      "grad_norm": 0.3383587896823883,
+      "learning_rate": 9.85069302894415e-05,
+      "loss": 5.3224,
+      "step": 1765
+    },
+    {
+      "epoch": 0.12026090501426824,
+      "grad_norm": 0.41393041610717773,
+      "learning_rate": 9.850268378855823e-05,
+      "loss": 5.062,
+      "step": 1770
+    },
+    {
+      "epoch": 0.12060062508493002,
+      "grad_norm": 0.43941059708595276,
+      "learning_rate": 9.849843728767496e-05,
+      "loss": 5.0013,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1209403451555918,
+      "grad_norm": 0.35179632902145386,
+      "learning_rate": 9.849419078679169e-05,
+      "loss": 5.1989,
+      "step": 1780
+    },
+    {
+      "epoch": 0.12128006522625356,
+      "grad_norm": 0.7730126976966858,
+      "learning_rate": 9.848994428590842e-05,
+      "loss": 5.1821,
+      "step": 1785
+    },
+    {
+      "epoch": 0.12161978529691535,
+      "grad_norm": 0.45934972167015076,
+      "learning_rate": 9.848569778502514e-05,
+      "loss": 5.2454,
+      "step": 1790
+    },
+    {
+      "epoch": 0.12195950536757712,
+      "grad_norm": 0.6049938797950745,
+      "learning_rate": 9.848145128414187e-05,
+      "loss": 5.0269,
+      "step": 1795
+    },
+    {
+      "epoch": 0.12229922543823889,
+      "grad_norm": 0.9805595874786377,
+      "learning_rate": 9.84772047832586e-05,
+      "loss": 4.885,
+      "step": 1800
+    },
+    {
+      "epoch": 0.12263894550890067,
+      "grad_norm": 0.47447869181632996,
+      "learning_rate": 9.847295828237533e-05,
+      "loss": 4.9461,
+      "step": 1805
+    },
+    {
+      "epoch": 0.12297866557956244,
+      "grad_norm": 0.4037536382675171,
+      "learning_rate": 9.846871178149206e-05,
+      "loss": 4.79,
+      "step": 1810
+    },
+    {
+      "epoch": 0.12331838565022421,
+      "grad_norm": 0.6516850590705872,
+      "learning_rate": 9.846446528060878e-05,
+      "loss": 4.8244,
+      "step": 1815
+    },
+    {
+      "epoch": 0.123658105720886,
+      "grad_norm": 0.46356433629989624,
+      "learning_rate": 9.846021877972551e-05,
+      "loss": 4.8264,
+      "step": 1820
+    },
+    {
+      "epoch": 0.12399782579154776,
+      "grad_norm": 1.0530160665512085,
+      "learning_rate": 9.845597227884224e-05,
+      "loss": 4.7702,
+      "step": 1825
+    },
+    {
+      "epoch": 0.12433754586220953,
+      "grad_norm": 1.3012051582336426,
+      "learning_rate": 9.845172577795897e-05,
+      "loss": 4.9146,
+      "step": 1830
+    },
+    {
+      "epoch": 0.12467726593287132,
+      "grad_norm": 0.6158355474472046,
+      "learning_rate": 9.84474792770757e-05,
+      "loss": 5.0082,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1250169860035331,
+      "grad_norm": 4.688101768493652,
+      "learning_rate": 9.844323277619241e-05,
+      "loss": 4.9467,
+      "step": 1840
+    },
+    {
+      "epoch": 0.12535670607419486,
+      "grad_norm": 0.5098426342010498,
+      "learning_rate": 9.843898627530915e-05,
+      "loss": 4.7443,
+      "step": 1845
+    },
+    {
+      "epoch": 0.12569642614485663,
+      "grad_norm": 0.6203608512878418,
+      "learning_rate": 9.843473977442588e-05,
+      "loss": 5.0459,
+      "step": 1850
+    },
+    {
+      "epoch": 0.12603614621551842,
+      "grad_norm": 0.5965786576271057,
+      "learning_rate": 9.84304932735426e-05,
+      "loss": 5.0615,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1263758662861802,
+      "grad_norm": 0.7298919558525085,
+      "learning_rate": 9.842624677265934e-05,
+      "loss": 4.9333,
+      "step": 1860
+    },
+    {
+      "epoch": 0.12671558635684196,
+      "grad_norm": 0.514262318611145,
+      "learning_rate": 9.842200027177606e-05,
+      "loss": 4.9152,
+      "step": 1865
+    },
+    {
+      "epoch": 0.12705530642750373,
+      "grad_norm": 0.49908140301704407,
+      "learning_rate": 9.841775377089278e-05,
+      "loss": 4.8825,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1273950264981655,
+      "grad_norm": 0.3929906487464905,
+      "learning_rate": 9.841350727000952e-05,
+      "loss": 4.6288,
+      "step": 1875
+    },
+    {
+      "epoch": 0.12773474656882727,
+      "grad_norm": 0.6221901178359985,
+      "learning_rate": 9.840926076912625e-05,
+      "loss": 4.6867,
+      "step": 1880
+    },
+    {
+      "epoch": 0.12807446663948907,
+      "grad_norm": 0.4237980544567108,
+      "learning_rate": 9.840501426824296e-05,
+      "loss": 5.0275,
+      "step": 1885
+    },
+    {
+      "epoch": 0.12841418671015084,
+      "grad_norm": 0.5076737403869629,
+      "learning_rate": 9.84007677673597e-05,
+      "loss": 4.3183,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1287539067808126,
+      "grad_norm": 0.562611997127533,
+      "learning_rate": 9.839652126647643e-05,
+      "loss": 4.8492,
+      "step": 1895
+    },
+    {
+      "epoch": 0.12909362685147438,
+      "grad_norm": 0.43838977813720703,
+      "learning_rate": 9.839227476559315e-05,
+      "loss": 4.7559,
+      "step": 1900
+    },
+    {
+      "epoch": 0.12943334692213615,
+      "grad_norm": 1.7356271743774414,
+      "learning_rate": 9.838802826470989e-05,
+      "loss": 4.4937,
+      "step": 1905
+    },
+    {
+      "epoch": 0.12977306699279795,
+      "grad_norm": 0.35975855588912964,
+      "learning_rate": 9.838378176382662e-05,
+      "loss": 4.8153,
+      "step": 1910
+    },
+    {
+      "epoch": 0.13011278706345972,
+      "grad_norm": 0.46843382716178894,
+      "learning_rate": 9.837953526294333e-05,
+      "loss": 4.7742,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1304525071341215,
+      "grad_norm": 0.49429741501808167,
+      "learning_rate": 9.837528876206007e-05,
+      "loss": 4.5403,
+      "step": 1920
+    },
+    {
+      "epoch": 0.13079222720478326,
+      "grad_norm": 0.496423602104187,
+      "learning_rate": 9.837104226117679e-05,
+      "loss": 4.8032,
+      "step": 1925
+    },
+    {
+      "epoch": 0.13113194727544503,
+      "grad_norm": 0.7953855395317078,
+      "learning_rate": 9.836679576029352e-05,
+      "loss": 4.8191,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1314716673461068,
+      "grad_norm": 0.5093162655830383,
+      "learning_rate": 9.836254925941026e-05,
+      "loss": 4.794,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1318113874167686,
+      "grad_norm": 0.37883055210113525,
+      "learning_rate": 9.835830275852697e-05,
+      "loss": 4.3129,
+      "step": 1940
+    },
+    {
+      "epoch": 0.13215110748743036,
+      "grad_norm": 0.6972466707229614,
+      "learning_rate": 9.83540562576437e-05,
+      "loss": 4.6677,
+      "step": 1945
+    },
+    {
+      "epoch": 0.13249082755809213,
+      "grad_norm": 0.4960924983024597,
+      "learning_rate": 9.834980975676044e-05,
+      "loss": 4.7554,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1328305476287539,
+      "grad_norm": 0.3313211500644684,
+      "learning_rate": 9.834556325587716e-05,
+      "loss": 4.5136,
+      "step": 1955
+    },
+    {
+      "epoch": 0.13317026769941567,
+      "grad_norm": 0.37889447808265686,
+      "learning_rate": 9.83413167549939e-05,
+      "loss": 4.6352,
+      "step": 1960
+    },
+    {
+      "epoch": 0.13350998777007744,
+      "grad_norm": 0.2897196412086487,
+      "learning_rate": 9.833707025411062e-05,
+      "loss": 4.2392,
+      "step": 1965
+    },
+    {
+      "epoch": 0.13384970784073924,
+      "grad_norm": 0.4556117653846741,
+      "learning_rate": 9.833282375322734e-05,
+      "loss": 4.4802,
+      "step": 1970
+    },
+    {
+      "epoch": 0.134189427911401,
+      "grad_norm": 0.24939770996570587,
+      "learning_rate": 9.832857725234408e-05,
+      "loss": 4.8028,
+      "step": 1975
+    },
+    {
+      "epoch": 0.13452914798206278,
+      "grad_norm": 0.5589706301689148,
+      "learning_rate": 9.832433075146081e-05,
+      "loss": 4.5484,
+      "step": 1980
+    },
+    {
+      "epoch": 0.13486886805272455,
+      "grad_norm": 0.403367817401886,
+      "learning_rate": 9.832008425057752e-05,
+      "loss": 4.6303,
+      "step": 1985
+    },
+    {
+      "epoch": 0.13520858812338632,
+      "grad_norm": 0.2891002595424652,
+      "learning_rate": 9.831583774969426e-05,
+      "loss": 4.6125,
+      "step": 1990
+    },
+    {
+      "epoch": 0.13554830819404812,
+      "grad_norm": 0.4545519948005676,
+      "learning_rate": 9.831159124881098e-05,
+      "loss": 4.6801,
+      "step": 1995
+    },
+    {
+      "epoch": 0.1358880282647099,
+      "grad_norm": 0.2752302289009094,
+      "learning_rate": 9.830734474792771e-05,
+      "loss": 4.5331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.13622774833537166,
+      "grad_norm": 0.4735427498817444,
+      "learning_rate": 9.830309824704445e-05,
+      "loss": 4.5487,
+      "step": 2005
+    },
+    {
+      "epoch": 0.13656746840603343,
+      "grad_norm": 0.2892632782459259,
+      "learning_rate": 9.829885174616116e-05,
+      "loss": 4.7872,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1369071884766952,
+      "grad_norm": 0.3587241768836975,
+      "learning_rate": 9.829460524527789e-05,
+      "loss": 4.8017,
+      "step": 2015
+    },
+    {
+      "epoch": 0.13724690854735697,
+      "grad_norm": 0.8643600940704346,
+      "learning_rate": 9.829035874439463e-05,
+      "loss": 4.9978,
+      "step": 2020
+    },
+    {
+      "epoch": 0.13758662861801876,
+      "grad_norm": 0.3995005786418915,
+      "learning_rate": 9.828611224351135e-05,
+      "loss": 4.7966,
+      "step": 2025
+    },
+    {
+      "epoch": 0.13792634868868053,
+      "grad_norm": 0.5287114381790161,
+      "learning_rate": 9.828186574262808e-05,
+      "loss": 4.6836,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1382660687593423,
+      "grad_norm": 0.356660932302475,
+      "learning_rate": 9.827761924174482e-05,
+      "loss": 4.6598,
+      "step": 2035
+    },
+    {
+      "epoch": 0.13860578883000407,
+      "grad_norm": 0.3594839572906494,
+      "learning_rate": 9.827337274086153e-05,
+      "loss": 4.7932,
+      "step": 2040
+    },
+    {
+      "epoch": 0.13894550890066584,
+      "grad_norm": 0.460989385843277,
+      "learning_rate": 9.826912623997826e-05,
+      "loss": 4.8404,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1392852289713276,
+      "grad_norm": 0.3044515550136566,
+      "learning_rate": 9.8264879739095e-05,
+      "loss": 4.4804,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1396249490419894,
+      "grad_norm": 0.2440759837627411,
+      "learning_rate": 9.826063323821172e-05,
+      "loss": 4.6584,
+      "step": 2055
+    },
+    {
+      "epoch": 0.13996466911265118,
+      "grad_norm": 0.39719679951667786,
+      "learning_rate": 9.825638673732844e-05,
+      "loss": 4.6913,
+      "step": 2060
+    },
+    {
+      "epoch": 0.14030438918331295,
+      "grad_norm": 0.2519219219684601,
+      "learning_rate": 9.825214023644517e-05,
+      "loss": 4.7914,
+      "step": 2065
+    },
+    {
+      "epoch": 0.14064410925397472,
+      "grad_norm": 0.27213895320892334,
+      "learning_rate": 9.82478937355619e-05,
+      "loss": 4.4571,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1409838293246365,
+      "grad_norm": 0.31952184438705444,
+      "learning_rate": 9.824364723467863e-05,
+      "loss": 4.4334,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1413235493952983,
+      "grad_norm": 0.2466011643409729,
+      "learning_rate": 9.823940073379536e-05,
+      "loss": 4.623,
+      "step": 2080
+    },
+    {
+      "epoch": 0.14166326946596006,
+      "grad_norm": 0.41923725605010986,
+      "learning_rate": 9.823515423291208e-05,
+      "loss": 4.5557,
+      "step": 2085
+    },
+    {
+      "epoch": 0.14200298953662183,
+      "grad_norm": 0.23959270119667053,
+      "learning_rate": 9.823090773202881e-05,
+      "loss": 4.5756,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1423427096072836,
+      "grad_norm": 0.7019773721694946,
+      "learning_rate": 9.822666123114554e-05,
+      "loss": 4.74,
+      "step": 2095
+    },
+    {
+      "epoch": 0.14268242967794537,
+      "grad_norm": 0.6014403700828552,
+      "learning_rate": 9.822241473026227e-05,
+      "loss": 4.4456,
+      "step": 2100
+    },
+    {
+      "epoch": 0.14302214974860714,
+      "grad_norm": 0.2578621804714203,
+      "learning_rate": 9.8218168229379e-05,
+      "loss": 4.6776,
+      "step": 2105
+    },
+    {
+      "epoch": 0.14336186981926893,
+      "grad_norm": 0.24368084967136383,
+      "learning_rate": 9.821392172849572e-05,
+      "loss": 4.7798,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1437015898899307,
+      "grad_norm": 0.4451867938041687,
+      "learning_rate": 9.820967522761245e-05,
+      "loss": 4.2507,
+      "step": 2115
+    },
+    {
+      "epoch": 0.14404130996059247,
+      "grad_norm": 0.27697330713272095,
+      "learning_rate": 9.820542872672918e-05,
+      "loss": 4.6886,
+      "step": 2120
+    },
+    {
+      "epoch": 0.14438103003125424,
+      "grad_norm": 0.8379690647125244,
+      "learning_rate": 9.820118222584591e-05,
+      "loss": 4.5629,
+      "step": 2125
+    },
+    {
+      "epoch": 0.144720750101916,
+      "grad_norm": 0.9834319353103638,
+      "learning_rate": 9.819693572496264e-05,
+      "loss": 4.4945,
+      "step": 2130
+    },
+    {
+      "epoch": 0.14506047017257778,
+      "grad_norm": 0.45272937417030334,
+      "learning_rate": 9.819268922407936e-05,
+      "loss": 4.6099,
+      "step": 2135
+    },
+    {
+      "epoch": 0.14540019024323958,
+      "grad_norm": 0.517729640007019,
+      "learning_rate": 9.818844272319609e-05,
+      "loss": 4.6808,
+      "step": 2140
+    },
+    {
+      "epoch": 0.14573991031390135,
+      "grad_norm": 0.26133647561073303,
+      "learning_rate": 9.818419622231282e-05,
+      "loss": 4.4916,
+      "step": 2145
+    },
+    {
+      "epoch": 0.14607963038456312,
+      "grad_norm": 0.31160035729408264,
+      "learning_rate": 9.817994972142955e-05,
+      "loss": 4.4746,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1464193504552249,
+      "grad_norm": 0.3950839936733246,
+      "learning_rate": 9.817570322054628e-05,
+      "loss": 4.8946,
+      "step": 2155
+    },
+    {
+      "epoch": 0.14675907052588666,
+      "grad_norm": 0.254171758890152,
+      "learning_rate": 9.8171456719663e-05,
+      "loss": 4.5237,
+      "step": 2160
+    },
+    {
+      "epoch": 0.14709879059654846,
+      "grad_norm": 0.4314219653606415,
+      "learning_rate": 9.816721021877973e-05,
+      "loss": 4.6116,
+      "step": 2165
+    },
+    {
+      "epoch": 0.14743851066721023,
+      "grad_norm": 0.2894288897514343,
+      "learning_rate": 9.816296371789646e-05,
+      "loss": 4.4748,
+      "step": 2170
+    },
+    {
+      "epoch": 0.147778230737872,
+      "grad_norm": 0.2681034207344055,
+      "learning_rate": 9.815871721701319e-05,
+      "loss": 4.5926,
+      "step": 2175
+    },
+    {
+      "epoch": 0.14811795080853377,
+      "grad_norm": 0.27911391854286194,
+      "learning_rate": 9.815447071612992e-05,
+      "loss": 4.5391,
+      "step": 2180
+    },
+    {
+      "epoch": 0.14845767087919554,
+      "grad_norm": 0.3182697296142578,
+      "learning_rate": 9.815022421524664e-05,
+      "loss": 4.5708,
+      "step": 2185
+    },
+    {
+      "epoch": 0.1487973909498573,
+      "grad_norm": 0.2478509396314621,
+      "learning_rate": 9.814597771436337e-05,
+      "loss": 4.3974,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1491371110205191,
+      "grad_norm": 0.3418025076389313,
+      "learning_rate": 9.814173121348009e-05,
+      "loss": 4.5312,
+      "step": 2195
+    },
+    {
+      "epoch": 0.14947683109118087,
+      "grad_norm": 0.2670694887638092,
+      "learning_rate": 9.813748471259683e-05,
+      "loss": 4.5906,
+      "step": 2200
+    },
+    {
+      "epoch": 0.14981655116184264,
+      "grad_norm": 0.29988008737564087,
+      "learning_rate": 9.813323821171356e-05,
+      "loss": 4.4151,
+      "step": 2205
+    },
+    {
+      "epoch": 0.1501562712325044,
+      "grad_norm": 0.2230396866798401,
+      "learning_rate": 9.812899171083027e-05,
+      "loss": 4.6196,
+      "step": 2210
+    },
+    {
+      "epoch": 0.15049599130316618,
+      "grad_norm": 0.2940434515476227,
+      "learning_rate": 9.812474520994701e-05,
+      "loss": 4.5765,
+      "step": 2215
+    },
+    {
+      "epoch": 0.15083571137382795,
+      "grad_norm": 0.2943139672279358,
+      "learning_rate": 9.812049870906374e-05,
+      "loss": 4.7439,
+      "step": 2220
+    },
+    {
+      "epoch": 0.15117543144448975,
+      "grad_norm": 0.5938501954078674,
+      "learning_rate": 9.811625220818045e-05,
+      "loss": 4.6233,
+      "step": 2225
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 0.29499292373657227,
+      "learning_rate": 9.81120057072972e-05,
+      "loss": 4.43,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1518548715858133,
+      "grad_norm": 0.21327312290668488,
+      "learning_rate": 9.810775920641392e-05,
+      "loss": 4.5865,
+      "step": 2235
+    },
+    {
+      "epoch": 0.15219459165647506,
+      "grad_norm": 0.4112052917480469,
+      "learning_rate": 9.810351270553064e-05,
+      "loss": 4.7013,
+      "step": 2240
+    },
+    {
+      "epoch": 0.15253431172713683,
+      "grad_norm": 0.40261027216911316,
+      "learning_rate": 9.809926620464738e-05,
+      "loss": 4.6419,
+      "step": 2245
+    },
+    {
+      "epoch": 0.15287403179779863,
+      "grad_norm": 0.2737533748149872,
+      "learning_rate": 9.809501970376411e-05,
+      "loss": 4.3994,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1532137518684604,
+      "grad_norm": 0.24050559103488922,
+      "learning_rate": 9.809077320288082e-05,
+      "loss": 4.5648,
+      "step": 2255
+    },
+    {
+      "epoch": 0.15355347193912217,
+      "grad_norm": 0.3781549036502838,
+      "learning_rate": 9.808652670199756e-05,
+      "loss": 4.4987,
+      "step": 2260
+    },
+    {
+      "epoch": 0.15389319200978394,
+      "grad_norm": 0.46098098158836365,
+      "learning_rate": 9.808228020111428e-05,
+      "loss": 4.5322,
+      "step": 2265
+    },
+    {
+      "epoch": 0.1542329120804457,
+      "grad_norm": 0.32969388365745544,
+      "learning_rate": 9.8078033700231e-05,
+      "loss": 4.59,
+      "step": 2270
+    },
+    {
+      "epoch": 0.15457263215110748,
+      "grad_norm": 0.28195780515670776,
+      "learning_rate": 9.807378719934775e-05,
+      "loss": 4.4548,
+      "step": 2275
+    },
+    {
+      "epoch": 0.15491235222176927,
+      "grad_norm": 0.2665387690067291,
+      "learning_rate": 9.806954069846446e-05,
+      "loss": 4.4353,
+      "step": 2280
+    },
+    {
+      "epoch": 0.15525207229243104,
+      "grad_norm": 0.3116438686847687,
+      "learning_rate": 9.806529419758119e-05,
+      "loss": 4.3557,
+      "step": 2285
+    },
+    {
+      "epoch": 0.1555917923630928,
+      "grad_norm": 0.42467501759529114,
+      "learning_rate": 9.806104769669793e-05,
+      "loss": 4.3241,
+      "step": 2290
+    },
+    {
+      "epoch": 0.15593151243375458,
+      "grad_norm": 0.24590204656124115,
+      "learning_rate": 9.805680119581465e-05,
+      "loss": 4.3719,
+      "step": 2295
+    },
+    {
+      "epoch": 0.15627123250441635,
+      "grad_norm": 0.7295488119125366,
+      "learning_rate": 9.805255469493139e-05,
+      "loss": 4.5891,
+      "step": 2300
+    },
+    {
+      "epoch": 0.15661095257507815,
+      "grad_norm": 0.24560780823230743,
+      "learning_rate": 9.804830819404812e-05,
+      "loss": 4.6124,
+      "step": 2305
+    },
+    {
+      "epoch": 0.15695067264573992,
+      "grad_norm": 0.2907837927341461,
+      "learning_rate": 9.804406169316483e-05,
+      "loss": 4.2532,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1572903927164017,
+      "grad_norm": 1.0109922885894775,
+      "learning_rate": 9.803981519228157e-05,
+      "loss": 4.5454,
+      "step": 2315
+    },
+    {
+      "epoch": 0.15763011278706346,
+      "grad_norm": 0.2637081444263458,
+      "learning_rate": 9.80355686913983e-05,
+      "loss": 4.5952,
+      "step": 2320
+    },
+    {
+      "epoch": 0.15796983285772523,
+      "grad_norm": 0.2559982240200043,
+      "learning_rate": 9.803132219051501e-05,
+      "loss": 4.6078,
+      "step": 2325
+    },
+    {
+      "epoch": 0.158309552928387,
+      "grad_norm": 0.4410446882247925,
+      "learning_rate": 9.802707568963176e-05,
+      "loss": 4.6202,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1586492729990488,
+      "grad_norm": 0.20168878138065338,
+      "learning_rate": 9.802282918874848e-05,
+      "loss": 4.4023,
+      "step": 2335
+    },
+    {
+      "epoch": 0.15898899306971057,
+      "grad_norm": 0.29185861349105835,
+      "learning_rate": 9.80185826878652e-05,
+      "loss": 4.543,
+      "step": 2340
+    },
+    {
+      "epoch": 0.15932871314037234,
+      "grad_norm": 0.22290275990962982,
+      "learning_rate": 9.801433618698194e-05,
+      "loss": 4.6697,
+      "step": 2345
+    },
+    {
+      "epoch": 0.1596684332110341,
+      "grad_norm": 0.7529789805412292,
+      "learning_rate": 9.801008968609865e-05,
+      "loss": 4.737,
+      "step": 2350
+    },
+    {
+      "epoch": 0.16000815328169588,
+      "grad_norm": 0.3712422549724579,
+      "learning_rate": 9.800584318521538e-05,
+      "loss": 4.6241,
+      "step": 2355
+    },
+    {
+      "epoch": 0.16034787335235764,
+      "grad_norm": 0.23941993713378906,
+      "learning_rate": 9.800159668433212e-05,
+      "loss": 4.6871,
+      "step": 2360
+    },
+    {
+      "epoch": 0.16068759342301944,
+      "grad_norm": 0.37533217668533325,
+      "learning_rate": 9.799735018344884e-05,
+      "loss": 4.467,
+      "step": 2365
+    },
+    {
+      "epoch": 0.1610273134936812,
+      "grad_norm": 0.2338525950908661,
+      "learning_rate": 9.799310368256557e-05,
+      "loss": 4.435,
+      "step": 2370
+    },
+    {
+      "epoch": 0.16136703356434298,
+      "grad_norm": 0.26814886927604675,
+      "learning_rate": 9.798885718168231e-05,
+      "loss": 4.4838,
+      "step": 2375
+    },
+    {
+      "epoch": 0.16170675363500475,
+      "grad_norm": 0.3187100887298584,
+      "learning_rate": 9.798461068079902e-05,
+      "loss": 4.354,
+      "step": 2380
+    },
+    {
+      "epoch": 0.16204647370566652,
+      "grad_norm": 0.7054830193519592,
+      "learning_rate": 9.798036417991575e-05,
+      "loss": 4.489,
+      "step": 2385
+    },
+    {
+      "epoch": 0.16238619377632832,
+      "grad_norm": 0.25023216009140015,
+      "learning_rate": 9.797611767903249e-05,
+      "loss": 4.6024,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1627259138469901,
+      "grad_norm": 0.24370110034942627,
+      "learning_rate": 9.79718711781492e-05,
+      "loss": 4.3951,
+      "step": 2395
+    },
+    {
+      "epoch": 0.16306563391765186,
+      "grad_norm": 0.23113249242305756,
+      "learning_rate": 9.796762467726593e-05,
+      "loss": 4.4352,
+      "step": 2400
+    },
+    {
+      "epoch": 0.16340535398831363,
+      "grad_norm": 0.4448549747467041,
+      "learning_rate": 9.796337817638268e-05,
+      "loss": 4.4063,
+      "step": 2405
+    },
+    {
+      "epoch": 0.1637450740589754,
+      "grad_norm": 0.20236225426197052,
+      "learning_rate": 9.795913167549939e-05,
+      "loss": 4.6175,
+      "step": 2410
+    },
+    {
+      "epoch": 0.16408479412963717,
+      "grad_norm": 0.5627440810203552,
+      "learning_rate": 9.795488517461612e-05,
+      "loss": 4.5675,
+      "step": 2415
+    },
+    {
+      "epoch": 0.16442451420029897,
+      "grad_norm": 0.28272920846939087,
+      "learning_rate": 9.795063867373285e-05,
+      "loss": 4.6146,
+      "step": 2420
+    },
+    {
+      "epoch": 0.16476423427096074,
+      "grad_norm": 0.2605418264865875,
+      "learning_rate": 9.794639217284957e-05,
+      "loss": 4.5697,
+      "step": 2425
+    },
+    {
+      "epoch": 0.1651039543416225,
+      "grad_norm": 0.23570238053798676,
+      "learning_rate": 9.79421456719663e-05,
+      "loss": 4.5072,
+      "step": 2430
+    },
+    {
+      "epoch": 0.16544367441228428,
+      "grad_norm": 0.20745481550693512,
+      "learning_rate": 9.793789917108303e-05,
+      "loss": 4.5735,
+      "step": 2435
+    },
+    {
+      "epoch": 0.16578339448294604,
+      "grad_norm": 0.23489026725292206,
+      "learning_rate": 9.793365267019976e-05,
+      "loss": 4.4731,
+      "step": 2440
+    },
+    {
+      "epoch": 0.16612311455360781,
+      "grad_norm": 0.4274902939796448,
+      "learning_rate": 9.792940616931649e-05,
+      "loss": 4.6706,
+      "step": 2445
+    },
+    {
+      "epoch": 0.1664628346242696,
+      "grad_norm": 0.25951382517814636,
+      "learning_rate": 9.792515966843321e-05,
+      "loss": 4.441,
+      "step": 2450
+    },
+    {
+      "epoch": 0.16680255469493138,
+      "grad_norm": 1.9463924169540405,
+      "learning_rate": 9.792091316754994e-05,
+      "loss": 4.4691,
+      "step": 2455
+    },
+    {
+      "epoch": 0.16714227476559315,
+      "grad_norm": 0.4177579879760742,
+      "learning_rate": 9.791666666666667e-05,
+      "loss": 4.4903,
+      "step": 2460
+    },
+    {
+      "epoch": 0.16748199483625492,
+      "grad_norm": 0.533138632774353,
+      "learning_rate": 9.79124201657834e-05,
+      "loss": 4.3311,
+      "step": 2465
+    },
+    {
+      "epoch": 0.1678217149069167,
+      "grad_norm": 0.2822255790233612,
+      "learning_rate": 9.790817366490013e-05,
+      "loss": 4.5948,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1681614349775785,
+      "grad_norm": 0.29035472869873047,
+      "learning_rate": 9.790392716401685e-05,
+      "loss": 4.5585,
+      "step": 2475
+    },
+    {
+      "epoch": 0.16850115504824026,
+      "grad_norm": 2.6457104682922363,
+      "learning_rate": 9.789968066313358e-05,
+      "loss": 4.5255,
+      "step": 2480
+    },
+    {
+      "epoch": 0.16884087511890203,
+      "grad_norm": 0.21925875544548035,
+      "learning_rate": 9.789543416225031e-05,
+      "loss": 4.5955,
+      "step": 2485
+    },
+    {
+      "epoch": 0.1691805951895638,
+      "grad_norm": 0.3095509707927704,
+      "learning_rate": 9.789118766136704e-05,
+      "loss": 4.5427,
+      "step": 2490
+    },
+    {
+      "epoch": 0.16952031526022557,
+      "grad_norm": 1.3866817951202393,
+      "learning_rate": 9.788694116048377e-05,
+      "loss": 4.3407,
+      "step": 2495
+    },
+    {
+      "epoch": 0.16986003533088734,
+      "grad_norm": 0.31529414653778076,
+      "learning_rate": 9.78826946596005e-05,
+      "loss": 4.614,
+      "step": 2500
+    },
+    {
+      "epoch": 0.17019975540154914,
+      "grad_norm": 0.25377875566482544,
+      "learning_rate": 9.787844815871722e-05,
+      "loss": 4.5838,
+      "step": 2505
+    },
+    {
+      "epoch": 0.1705394754722109,
+      "grad_norm": 0.7861871123313904,
+      "learning_rate": 9.787420165783395e-05,
+      "loss": 4.5731,
+      "step": 2510
+    },
+    {
+      "epoch": 0.17087919554287267,
+      "grad_norm": 0.19743318855762482,
+      "learning_rate": 9.786995515695068e-05,
+      "loss": 4.3947,
+      "step": 2515
+    },
+    {
+      "epoch": 0.17121891561353444,
+      "grad_norm": 0.3416430950164795,
+      "learning_rate": 9.78657086560674e-05,
+      "loss": 4.5711,
+      "step": 2520
+    },
+    {
+      "epoch": 0.17155863568419621,
+      "grad_norm": 0.3679373562335968,
+      "learning_rate": 9.786146215518413e-05,
+      "loss": 4.6518,
+      "step": 2525
+    },
+    {
+      "epoch": 0.17189835575485798,
+      "grad_norm": 0.23833996057510376,
+      "learning_rate": 9.785721565430086e-05,
+      "loss": 4.4339,
+      "step": 2530
+    },
+    {
+      "epoch": 0.17223807582551978,
+      "grad_norm": 0.25589922070503235,
+      "learning_rate": 9.785296915341759e-05,
+      "loss": 4.6889,
+      "step": 2535
+    },
+    {
+      "epoch": 0.17257779589618155,
+      "grad_norm": 0.27489981055259705,
+      "learning_rate": 9.784872265253432e-05,
+      "loss": 4.3215,
+      "step": 2540
+    },
+    {
+      "epoch": 0.17291751596684332,
+      "grad_norm": 0.23039469122886658,
+      "learning_rate": 9.784447615165105e-05,
+      "loss": 4.4543,
+      "step": 2545
+    },
+    {
+      "epoch": 0.1732572360375051,
+      "grad_norm": 0.3405773341655731,
+      "learning_rate": 9.784022965076776e-05,
+      "loss": 4.5138,
+      "step": 2550
+    },
+    {
+      "epoch": 0.17359695610816686,
+      "grad_norm": 0.8154670000076294,
+      "learning_rate": 9.78359831498845e-05,
+      "loss": 4.4716,
+      "step": 2555
+    },
+    {
+      "epoch": 0.17393667617882866,
+      "grad_norm": 0.30465012788772583,
+      "learning_rate": 9.783173664900123e-05,
+      "loss": 4.2262,
+      "step": 2560
+    },
+    {
+      "epoch": 0.17427639624949043,
+      "grad_norm": 0.3995078504085541,
+      "learning_rate": 9.782749014811795e-05,
+      "loss": 4.5862,
+      "step": 2565
+    },
+    {
+      "epoch": 0.1746161163201522,
+      "grad_norm": 0.2636319398880005,
+      "learning_rate": 9.782324364723469e-05,
+      "loss": 4.6933,
+      "step": 2570
+    },
+    {
+      "epoch": 0.17495583639081397,
+      "grad_norm": 0.3614608943462372,
+      "learning_rate": 9.781899714635141e-05,
+      "loss": 4.6504,
+      "step": 2575
+    },
+    {
+      "epoch": 0.17529555646147574,
+      "grad_norm": 0.3470248878002167,
+      "learning_rate": 9.781475064546813e-05,
+      "loss": 4.7038,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1756352765321375,
+      "grad_norm": 0.39428719878196716,
+      "learning_rate": 9.781050414458487e-05,
+      "loss": 4.2662,
+      "step": 2585
+    },
+    {
+      "epoch": 0.1759749966027993,
+      "grad_norm": 0.22955843806266785,
+      "learning_rate": 9.78062576437016e-05,
+      "loss": 4.4044,
+      "step": 2590
+    },
+    {
+      "epoch": 0.17631471667346107,
+      "grad_norm": 0.2899189293384552,
+      "learning_rate": 9.780201114281831e-05,
+      "loss": 4.4223,
+      "step": 2595
+    },
+    {
+      "epoch": 0.17665443674412284,
+      "grad_norm": 0.4230986535549164,
+      "learning_rate": 9.779776464193505e-05,
+      "loss": 4.1781,
+      "step": 2600
+    },
+    {
+      "epoch": 0.17699415681478461,
+      "grad_norm": 0.32788804173469543,
+      "learning_rate": 9.779351814105178e-05,
+      "loss": 4.616,
+      "step": 2605
+    },
+    {
+      "epoch": 0.17733387688544638,
+      "grad_norm": 0.2200581431388855,
+      "learning_rate": 9.77892716401685e-05,
+      "loss": 4.2604,
+      "step": 2610
+    },
+    {
+      "epoch": 0.17767359695610815,
+      "grad_norm": 0.30823394656181335,
+      "learning_rate": 9.778502513928524e-05,
+      "loss": 4.2978,
+      "step": 2615
+    },
+    {
+      "epoch": 0.17801331702676995,
+      "grad_norm": 0.22299472987651825,
+      "learning_rate": 9.778077863840195e-05,
+      "loss": 4.3587,
+      "step": 2620
+    },
+    {
+      "epoch": 0.17835303709743172,
+      "grad_norm": 0.22951941192150116,
+      "learning_rate": 9.777653213751868e-05,
+      "loss": 4.2573,
+      "step": 2625
+    },
+    {
+      "epoch": 0.1786927571680935,
+      "grad_norm": 0.35953882336616516,
+      "learning_rate": 9.777228563663542e-05,
+      "loss": 4.3515,
+      "step": 2630
+    },
+    {
+      "epoch": 0.17903247723875526,
+      "grad_norm": 0.4688868522644043,
+      "learning_rate": 9.776803913575214e-05,
+      "loss": 4.5713,
+      "step": 2635
+    },
+    {
+      "epoch": 0.17937219730941703,
+      "grad_norm": 0.21083256602287292,
+      "learning_rate": 9.776379263486888e-05,
+      "loss": 4.4969,
+      "step": 2640
+    },
+    {
+      "epoch": 0.17971191738007883,
+      "grad_norm": 0.36712825298309326,
+      "learning_rate": 9.775954613398561e-05,
+      "loss": 4.5619,
+      "step": 2645
+    },
+    {
+      "epoch": 0.1800516374507406,
+      "grad_norm": 0.2260504513978958,
+      "learning_rate": 9.775529963310232e-05,
+      "loss": 4.6722,
+      "step": 2650
+    },
+    {
+      "epoch": 0.18039135752140237,
+      "grad_norm": 0.36943840980529785,
+      "learning_rate": 9.775105313221906e-05,
+      "loss": 4.4934,
+      "step": 2655
+    },
+    {
+      "epoch": 0.18073107759206414,
+      "grad_norm": 0.4936888515949249,
+      "learning_rate": 9.774680663133579e-05,
+      "loss": 4.348,
+      "step": 2660
+    },
+    {
+      "epoch": 0.1810707976627259,
+      "grad_norm": 0.21958352625370026,
+      "learning_rate": 9.77425601304525e-05,
+      "loss": 4.4938,
+      "step": 2665
+    },
+    {
+      "epoch": 0.18141051773338768,
+      "grad_norm": 1.1148053407669067,
+      "learning_rate": 9.773831362956925e-05,
+      "loss": 4.2976,
+      "step": 2670
+    },
+    {
+      "epoch": 0.18175023780404947,
+      "grad_norm": 0.39846473932266235,
+      "learning_rate": 9.773406712868597e-05,
+      "loss": 4.3586,
+      "step": 2675
+    },
+    {
+      "epoch": 0.18208995787471124,
+      "grad_norm": 0.28287413716316223,
+      "learning_rate": 9.772982062780269e-05,
+      "loss": 4.4437,
+      "step": 2680
+    },
+    {
+      "epoch": 0.18242967794537301,
+      "grad_norm": 0.3402862846851349,
+      "learning_rate": 9.772557412691943e-05,
+      "loss": 4.5291,
+      "step": 2685
+    },
+    {
+      "epoch": 0.18276939801603478,
+      "grad_norm": 0.3358980715274811,
+      "learning_rate": 9.772132762603616e-05,
+      "loss": 4.3738,
+      "step": 2690
+    },
+    {
+      "epoch": 0.18310911808669655,
+      "grad_norm": 0.19017407298088074,
+      "learning_rate": 9.771708112515287e-05,
+      "loss": 4.1701,
+      "step": 2695
+    },
+    {
+      "epoch": 0.18344883815735832,
+      "grad_norm": 0.2291361540555954,
+      "learning_rate": 9.771283462426961e-05,
+      "loss": 4.6092,
+      "step": 2700
+    },
+    {
+      "epoch": 0.18378855822802012,
+      "grad_norm": 0.42033877968788147,
+      "learning_rate": 9.770858812338633e-05,
+      "loss": 4.3813,
+      "step": 2705
+    },
+    {
+      "epoch": 0.1841282782986819,
+      "grad_norm": 0.22784222662448883,
+      "learning_rate": 9.770434162250306e-05,
+      "loss": 4.4408,
+      "step": 2710
+    },
+    {
+      "epoch": 0.18446799836934366,
+      "grad_norm": 0.23395958542823792,
+      "learning_rate": 9.77000951216198e-05,
+      "loss": 4.4935,
+      "step": 2715
+    },
+    {
+      "epoch": 0.18480771844000543,
+      "grad_norm": 0.2610359191894531,
+      "learning_rate": 9.769584862073651e-05,
+      "loss": 4.4932,
+      "step": 2720
+    },
+    {
+      "epoch": 0.1851474385106672,
+      "grad_norm": 0.2646908164024353,
+      "learning_rate": 9.769160211985324e-05,
+      "loss": 4.4914,
+      "step": 2725
+    },
+    {
+      "epoch": 0.185487158581329,
+      "grad_norm": 0.31001701951026917,
+      "learning_rate": 9.768735561896998e-05,
+      "loss": 4.5656,
+      "step": 2730
+    },
+    {
+      "epoch": 0.18582687865199077,
+      "grad_norm": 0.3422091007232666,
+      "learning_rate": 9.76831091180867e-05,
+      "loss": 4.4946,
+      "step": 2735
+    },
+    {
+      "epoch": 0.18616659872265254,
+      "grad_norm": 0.4761231243610382,
+      "learning_rate": 9.767886261720343e-05,
+      "loss": 4.1494,
+      "step": 2740
+    },
+    {
+      "epoch": 0.1865063187933143,
+      "grad_norm": 0.23646193742752075,
+      "learning_rate": 9.767461611632017e-05,
+      "loss": 4.3254,
+      "step": 2745
+    },
+    {
+      "epoch": 0.18684603886397608,
+      "grad_norm": 1.6517447233200073,
+      "learning_rate": 9.767036961543688e-05,
+      "loss": 4.5333,
+      "step": 2750
+    },
+    {
+      "epoch": 0.18718575893463785,
+      "grad_norm": 0.2012016475200653,
+      "learning_rate": 9.766612311455361e-05,
+      "loss": 4.7069,
+      "step": 2755
+    },
+    {
+      "epoch": 0.18752547900529964,
+      "grad_norm": 0.20281845331192017,
+      "learning_rate": 9.766187661367035e-05,
+      "loss": 4.4399,
+      "step": 2760
+    },
+    {
+      "epoch": 0.18786519907596141,
+      "grad_norm": 0.1804925948381424,
+      "learning_rate": 9.765763011278707e-05,
+      "loss": 4.5354,
+      "step": 2765
+    },
+    {
+      "epoch": 0.18820491914662318,
+      "grad_norm": 0.4761740267276764,
+      "learning_rate": 9.76533836119038e-05,
+      "loss": 4.6633,
+      "step": 2770
+    },
+    {
+      "epoch": 0.18854463921728495,
+      "grad_norm": 0.22267234325408936,
+      "learning_rate": 9.764913711102052e-05,
+      "loss": 4.5686,
+      "step": 2775
+    },
+    {
+      "epoch": 0.18888435928794672,
+      "grad_norm": 0.5881355404853821,
+      "learning_rate": 9.764489061013725e-05,
+      "loss": 4.4554,
+      "step": 2780
+    },
+    {
+      "epoch": 0.1892240793586085,
+      "grad_norm": 0.43992605805397034,
+      "learning_rate": 9.764064410925398e-05,
+      "loss": 4.1684,
+      "step": 2785
+    },
+    {
+      "epoch": 0.1895637994292703,
+      "grad_norm": 0.21498017013072968,
+      "learning_rate": 9.76363976083707e-05,
+      "loss": 4.5047,
+      "step": 2790
+    },
+    {
+      "epoch": 0.18990351949993206,
+      "grad_norm": 0.37874165177345276,
+      "learning_rate": 9.763215110748743e-05,
+      "loss": 4.2255,
+      "step": 2795
+    },
+    {
+      "epoch": 0.19024323957059383,
+      "grad_norm": 0.2565677762031555,
+      "learning_rate": 9.762790460660416e-05,
+      "loss": 4.4333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1905829596412556,
+      "grad_norm": 0.2246963530778885,
+      "learning_rate": 9.762365810572089e-05,
+      "loss": 4.4198,
+      "step": 2805
+    },
+    {
+      "epoch": 0.19092267971191737,
+      "grad_norm": 0.946719229221344,
+      "learning_rate": 9.761941160483762e-05,
+      "loss": 3.9055,
+      "step": 2810
+    },
+    {
+      "epoch": 0.19126239978257917,
+      "grad_norm": 1.0544602870941162,
+      "learning_rate": 9.761516510395435e-05,
+      "loss": 4.4512,
+      "step": 2815
+    },
+    {
+      "epoch": 0.19160211985324094,
+      "grad_norm": 0.21298794448375702,
+      "learning_rate": 9.761091860307107e-05,
+      "loss": 4.4729,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1919418399239027,
+      "grad_norm": 1.3822523355484009,
+      "learning_rate": 9.76066721021878e-05,
+      "loss": 4.3136,
+      "step": 2825
+    },
+    {
+      "epoch": 0.19228155999456448,
+      "grad_norm": 0.1828567236661911,
+      "learning_rate": 9.760242560130453e-05,
+      "loss": 4.5116,
+      "step": 2830
+    },
+    {
+      "epoch": 0.19262128006522625,
+      "grad_norm": 0.28580307960510254,
+      "learning_rate": 9.759817910042126e-05,
+      "loss": 4.4903,
+      "step": 2835
+    },
+    {
+      "epoch": 0.19296100013588802,
+      "grad_norm": 0.39433717727661133,
+      "learning_rate": 9.759393259953799e-05,
+      "loss": 4.4487,
+      "step": 2840
+    },
+    {
+      "epoch": 0.19330072020654981,
+      "grad_norm": 0.49140483140945435,
+      "learning_rate": 9.758968609865471e-05,
+      "loss": 4.4639,
+      "step": 2845
+    },
+    {
+      "epoch": 0.19364044027721158,
+      "grad_norm": 0.3383556306362152,
+      "learning_rate": 9.758543959777144e-05,
+      "loss": 4.3328,
+      "step": 2850
+    },
+    {
+      "epoch": 0.19398016034787335,
+      "grad_norm": 0.7367972135543823,
+      "learning_rate": 9.758119309688817e-05,
+      "loss": 4.5457,
+      "step": 2855
+    },
+    {
+      "epoch": 0.19431988041853512,
+      "grad_norm": 0.19852545857429504,
+      "learning_rate": 9.75769465960049e-05,
+      "loss": 4.2916,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1946596004891969,
+      "grad_norm": 0.3379197120666504,
+      "learning_rate": 9.757270009512163e-05,
+      "loss": 4.4011,
+      "step": 2865
+    },
+    {
+      "epoch": 0.19499932055985866,
+      "grad_norm": 0.4577140212059021,
+      "learning_rate": 9.756845359423835e-05,
+      "loss": 4.2438,
+      "step": 2870
+    },
+    {
+      "epoch": 0.19533904063052046,
+      "grad_norm": 0.32074615359306335,
+      "learning_rate": 9.756420709335508e-05,
+      "loss": 4.3429,
+      "step": 2875
+    },
+    {
+      "epoch": 0.19567876070118223,
+      "grad_norm": 0.4993734359741211,
+      "learning_rate": 9.755996059247181e-05,
+      "loss": 4.6458,
+      "step": 2880
+    },
+    {
+      "epoch": 0.196018480771844,
+      "grad_norm": 0.21413934230804443,
+      "learning_rate": 9.755571409158854e-05,
+      "loss": 4.3236,
+      "step": 2885
+    },
+    {
+      "epoch": 0.19635820084250577,
+      "grad_norm": 0.39588046073913574,
+      "learning_rate": 9.755146759070527e-05,
+      "loss": 4.2725,
+      "step": 2890
+    },
+    {
+      "epoch": 0.19669792091316754,
+      "grad_norm": 0.23066450655460358,
+      "learning_rate": 9.7547221089822e-05,
+      "loss": 4.2981,
+      "step": 2895
+    },
+    {
+      "epoch": 0.19703764098382934,
+      "grad_norm": 0.24343866109848022,
+      "learning_rate": 9.754297458893872e-05,
+      "loss": 4.4485,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1973773610544911,
+      "grad_norm": 0.2774411141872406,
+      "learning_rate": 9.753872808805544e-05,
+      "loss": 4.2737,
+      "step": 2905
+    },
+    {
+      "epoch": 0.19771708112515288,
+      "grad_norm": 0.3360697329044342,
+      "learning_rate": 9.753448158717218e-05,
+      "loss": 4.2094,
+      "step": 2910
+    },
+    {
+      "epoch": 0.19805680119581465,
+      "grad_norm": 0.3886429965496063,
+      "learning_rate": 9.75302350862889e-05,
+      "loss": 4.4864,
+      "step": 2915
+    },
+    {
+      "epoch": 0.19839652126647642,
+      "grad_norm": 0.5242161154747009,
+      "learning_rate": 9.752598858540562e-05,
+      "loss": 4.3877,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1987362413371382,
+      "grad_norm": 0.2082594633102417,
+      "learning_rate": 9.752174208452236e-05,
+      "loss": 4.3006,
+      "step": 2925
+    },
+    {
+      "epoch": 0.19907596140779998,
+      "grad_norm": 1.1216654777526855,
+      "learning_rate": 9.751749558363909e-05,
+      "loss": 4.2916,
+      "step": 2930
+    },
+    {
+      "epoch": 0.19941568147846175,
+      "grad_norm": 0.1812744289636612,
+      "learning_rate": 9.75132490827558e-05,
+      "loss": 4.4246,
+      "step": 2935
+    },
+    {
+      "epoch": 0.19975540154912352,
+      "grad_norm": 0.316278874874115,
+      "learning_rate": 9.750900258187255e-05,
+      "loss": 4.2248,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2000951216197853,
+      "grad_norm": 0.2795095443725586,
+      "learning_rate": 9.750475608098927e-05,
+      "loss": 4.4285,
+      "step": 2945
+    },
+    {
+      "epoch": 0.20043484169044706,
+      "grad_norm": 0.25871169567108154,
+      "learning_rate": 9.750050958010599e-05,
+      "loss": 4.3629,
+      "step": 2950
+    },
+    {
+      "epoch": 0.20077456176110883,
+      "grad_norm": 0.3203955888748169,
+      "learning_rate": 9.749626307922273e-05,
+      "loss": 4.8022,
+      "step": 2955
+    },
+    {
+      "epoch": 0.20111428183177063,
+      "grad_norm": 0.897880494594574,
+      "learning_rate": 9.749201657833946e-05,
+      "loss": 4.444,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2014540019024324,
+      "grad_norm": 0.6095696687698364,
+      "learning_rate": 9.748777007745617e-05,
+      "loss": 4.3442,
+      "step": 2965
+    },
+    {
+      "epoch": 0.20179372197309417,
+      "grad_norm": 0.8089606761932373,
+      "learning_rate": 9.748352357657291e-05,
+      "loss": 4.3223,
+      "step": 2970
+    },
+    {
+      "epoch": 0.20213344204375594,
+      "grad_norm": 0.5481230616569519,
+      "learning_rate": 9.747927707568963e-05,
+      "loss": 4.3059,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2024731621144177,
+      "grad_norm": 0.24502769112586975,
+      "learning_rate": 9.747503057480637e-05,
+      "loss": 4.2946,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2028128821850795,
+      "grad_norm": 0.20267254114151,
+      "learning_rate": 9.74707840739231e-05,
+      "loss": 4.3506,
+      "step": 2985
+    },
+    {
+      "epoch": 0.20315260225574128,
+      "grad_norm": 1.4581079483032227,
+      "learning_rate": 9.746653757303981e-05,
+      "loss": 4.3024,
+      "step": 2990
+    },
+    {
+      "epoch": 0.20349232232640305,
+      "grad_norm": 0.3428595960140228,
+      "learning_rate": 9.746229107215655e-05,
+      "loss": 4.4748,
+      "step": 2995
+    },
+    {
+      "epoch": 0.20383204239706482,
+      "grad_norm": 0.3032056391239166,
+      "learning_rate": 9.745804457127328e-05,
+      "loss": 4.3262,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2041717624677266,
+      "grad_norm": 0.32317832112312317,
+      "learning_rate": 9.745379807039e-05,
+      "loss": 4.475,
+      "step": 3005
+    },
+    {
+      "epoch": 0.20451148253838836,
+      "grad_norm": 0.21782419085502625,
+      "learning_rate": 9.744955156950674e-05,
+      "loss": 4.5021,
+      "step": 3010
+    },
+    {
+      "epoch": 0.20485120260905015,
+      "grad_norm": 0.17983724176883698,
+      "learning_rate": 9.744530506862347e-05,
+      "loss": 4.3826,
+      "step": 3015
+    },
+    {
+      "epoch": 0.20519092267971192,
+      "grad_norm": 0.3824704587459564,
+      "learning_rate": 9.744105856774018e-05,
+      "loss": 4.4883,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2055306427503737,
+      "grad_norm": 0.6417528390884399,
+      "learning_rate": 9.743681206685692e-05,
+      "loss": 4.3623,
+      "step": 3025
+    },
+    {
+      "epoch": 0.20587036282103546,
+      "grad_norm": 0.31229910254478455,
+      "learning_rate": 9.743256556597365e-05,
+      "loss": 4.3362,
+      "step": 3030
+    },
+    {
+      "epoch": 0.20621008289169723,
+      "grad_norm": 0.35579913854599,
+      "learning_rate": 9.742831906509036e-05,
+      "loss": 4.3119,
+      "step": 3035
+    },
+    {
+      "epoch": 0.206549802962359,
+      "grad_norm": 0.21225492656230927,
+      "learning_rate": 9.74240725642071e-05,
+      "loss": 4.4159,
+      "step": 3040
+    },
+    {
+      "epoch": 0.2068895230330208,
+      "grad_norm": 0.5204954147338867,
+      "learning_rate": 9.741982606332382e-05,
+      "loss": 4.254,
+      "step": 3045
+    },
+    {
+      "epoch": 0.20722924310368257,
+      "grad_norm": 0.22649656236171722,
+      "learning_rate": 9.741557956244055e-05,
+      "loss": 4.3604,
+      "step": 3050
+    },
+    {
+      "epoch": 0.20756896317434434,
+      "grad_norm": 0.22533409297466278,
+      "learning_rate": 9.741133306155729e-05,
+      "loss": 4.1594,
+      "step": 3055
+    },
+    {
+      "epoch": 0.2079086832450061,
+      "grad_norm": 0.2681191563606262,
+      "learning_rate": 9.7407086560674e-05,
+      "loss": 4.553,
+      "step": 3060
+    },
+    {
+      "epoch": 0.20824840331566788,
+      "grad_norm": 1.2959145307540894,
+      "learning_rate": 9.740284005979073e-05,
+      "loss": 4.2071,
+      "step": 3065
+    },
+    {
+      "epoch": 0.20858812338632968,
+      "grad_norm": 0.21679522097110748,
+      "learning_rate": 9.739859355890747e-05,
+      "loss": 4.5038,
+      "step": 3070
+    },
+    {
+      "epoch": 0.20892784345699145,
+      "grad_norm": 0.36338409781455994,
+      "learning_rate": 9.739434705802419e-05,
+      "loss": 4.3356,
+      "step": 3075
+    },
+    {
+      "epoch": 0.20926756352765322,
+      "grad_norm": 0.2271890938282013,
+      "learning_rate": 9.739010055714092e-05,
+      "loss": 4.3573,
+      "step": 3080
+    },
+    {
+      "epoch": 0.209607283598315,
+      "grad_norm": 0.2753996253013611,
+      "learning_rate": 9.738585405625766e-05,
+      "loss": 4.4467,
+      "step": 3085
+    },
+    {
+      "epoch": 0.20994700366897676,
+      "grad_norm": 0.32643699645996094,
+      "learning_rate": 9.738160755537437e-05,
+      "loss": 4.2911,
+      "step": 3090
+    },
+    {
+      "epoch": 0.21028672373963853,
+      "grad_norm": 0.1794055551290512,
+      "learning_rate": 9.73773610544911e-05,
+      "loss": 4.3936,
+      "step": 3095
+    },
+    {
+      "epoch": 0.21062644381030032,
+      "grad_norm": 0.2121143937110901,
+      "learning_rate": 9.737311455360784e-05,
+      "loss": 4.4391,
+      "step": 3100
+    },
+    {
+      "epoch": 0.2109661638809621,
+      "grad_norm": 0.6584509015083313,
+      "learning_rate": 9.736886805272456e-05,
+      "loss": 4.3936,
+      "step": 3105
+    },
+    {
+      "epoch": 0.21130588395162386,
+      "grad_norm": 0.2863527834415436,
+      "learning_rate": 9.736462155184128e-05,
+      "loss": 4.4768,
+      "step": 3110
+    },
+    {
+      "epoch": 0.21164560402228563,
+      "grad_norm": 1.1741371154785156,
+      "learning_rate": 9.736037505095803e-05,
+      "loss": 4.3373,
+      "step": 3115
+    },
+    {
+      "epoch": 0.2119853240929474,
+      "grad_norm": 0.3653934597969055,
+      "learning_rate": 9.735612855007474e-05,
+      "loss": 4.6397,
+      "step": 3120
+    },
+    {
+      "epoch": 0.21232504416360917,
+      "grad_norm": 0.369391530752182,
+      "learning_rate": 9.735188204919147e-05,
+      "loss": 4.1315,
+      "step": 3125
+    },
+    {
+      "epoch": 0.21266476423427097,
+      "grad_norm": 0.22272364795207977,
+      "learning_rate": 9.73476355483082e-05,
+      "loss": 4.4174,
+      "step": 3130
+    },
+    {
+      "epoch": 0.21300448430493274,
+      "grad_norm": 0.26364120841026306,
+      "learning_rate": 9.734338904742492e-05,
+      "loss": 4.3034,
+      "step": 3135
+    },
+    {
+      "epoch": 0.2133442043755945,
+      "grad_norm": 0.2755309045314789,
+      "learning_rate": 9.733914254654165e-05,
+      "loss": 4.3209,
+      "step": 3140
+    },
+    {
+      "epoch": 0.21368392444625628,
+      "grad_norm": 0.27905556559562683,
+      "learning_rate": 9.733489604565838e-05,
+      "loss": 4.4046,
+      "step": 3145
+    },
+    {
+      "epoch": 0.21402364451691805,
+      "grad_norm": 0.25759658217430115,
+      "learning_rate": 9.733064954477511e-05,
+      "loss": 4.1602,
+      "step": 3150
+    },
+    {
+      "epoch": 0.21436336458757985,
+      "grad_norm": 1.0761340856552124,
+      "learning_rate": 9.732640304389184e-05,
+      "loss": 4.2794,
+      "step": 3155
+    },
+    {
+      "epoch": 0.21470308465824162,
+      "grad_norm": 0.18029484152793884,
+      "learning_rate": 9.732215654300856e-05,
+      "loss": 4.3554,
+      "step": 3160
+    },
+    {
+      "epoch": 0.2150428047289034,
+      "grad_norm": 0.373797208070755,
+      "learning_rate": 9.731791004212529e-05,
+      "loss": 4.3863,
+      "step": 3165
+    },
+    {
+      "epoch": 0.21538252479956516,
+      "grad_norm": 0.6202191710472107,
+      "learning_rate": 9.731366354124202e-05,
+      "loss": 4.2955,
+      "step": 3170
+    },
+    {
+      "epoch": 0.21572224487022693,
+      "grad_norm": 0.20301900804042816,
+      "learning_rate": 9.730941704035875e-05,
+      "loss": 4.2914,
+      "step": 3175
+    },
+    {
+      "epoch": 0.2160619649408887,
+      "grad_norm": 0.17571194469928741,
+      "learning_rate": 9.730517053947548e-05,
+      "loss": 4.276,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2164016850115505,
+      "grad_norm": 0.3209381401538849,
+      "learning_rate": 9.73009240385922e-05,
+      "loss": 4.3957,
+      "step": 3185
+    },
+    {
+      "epoch": 0.21674140508221226,
+      "grad_norm": 0.2638840079307556,
+      "learning_rate": 9.729667753770893e-05,
+      "loss": 4.6224,
+      "step": 3190
+    },
+    {
+      "epoch": 0.21708112515287403,
+      "grad_norm": 0.19320239126682281,
+      "learning_rate": 9.729243103682566e-05,
+      "loss": 4.3039,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2174208452235358,
+      "grad_norm": 0.43768310546875,
+      "learning_rate": 9.728818453594239e-05,
+      "loss": 4.3898,
+      "step": 3200
+    },
+    {
+      "epoch": 0.21776056529419757,
+      "grad_norm": 0.35756048560142517,
+      "learning_rate": 9.728393803505912e-05,
+      "loss": 4.5883,
+      "step": 3205
+    },
+    {
+      "epoch": 0.21810028536485934,
+      "grad_norm": 0.2380749136209488,
+      "learning_rate": 9.727969153417584e-05,
+      "loss": 4.4224,
+      "step": 3210
+    },
+    {
+      "epoch": 0.21844000543552114,
+      "grad_norm": 0.23136284947395325,
+      "learning_rate": 9.727544503329257e-05,
+      "loss": 4.3141,
+      "step": 3215
+    },
+    {
+      "epoch": 0.2187797255061829,
+      "grad_norm": 0.3109607398509979,
+      "learning_rate": 9.72711985324093e-05,
+      "loss": 4.295,
+      "step": 3220
+    },
+    {
+      "epoch": 0.21911944557684468,
+      "grad_norm": 0.4062863290309906,
+      "learning_rate": 9.726695203152603e-05,
+      "loss": 4.2421,
+      "step": 3225
+    },
+    {
+      "epoch": 0.21945916564750645,
+      "grad_norm": 0.20023144781589508,
+      "learning_rate": 9.726270553064276e-05,
+      "loss": 4.3144,
+      "step": 3230
+    },
+    {
+      "epoch": 0.21979888571816822,
+      "grad_norm": 0.8297600150108337,
+      "learning_rate": 9.725845902975948e-05,
+      "loss": 4.2986,
+      "step": 3235
+    },
+    {
+      "epoch": 0.22013860578883002,
+      "grad_norm": 0.6315371990203857,
+      "learning_rate": 9.725421252887621e-05,
+      "loss": 4.4641,
+      "step": 3240
+    },
+    {
+      "epoch": 0.22047832585949179,
+      "grad_norm": 0.2108875811100006,
+      "learning_rate": 9.724996602799293e-05,
+      "loss": 4.2764,
+      "step": 3245
+    },
+    {
+      "epoch": 0.22081804593015356,
+      "grad_norm": 0.20751313865184784,
+      "learning_rate": 9.724571952710967e-05,
+      "loss": 4.3553,
+      "step": 3250
+    },
+    {
+      "epoch": 0.22115776600081533,
+      "grad_norm": 0.24425362050533295,
+      "learning_rate": 9.72414730262264e-05,
+      "loss": 4.243,
+      "step": 3255
+    },
+    {
+      "epoch": 0.2214974860714771,
+      "grad_norm": 0.22244137525558472,
+      "learning_rate": 9.723722652534311e-05,
+      "loss": 4.2259,
+      "step": 3260
+    },
+    {
+      "epoch": 0.22183720614213887,
+      "grad_norm": 1.1119288206100464,
+      "learning_rate": 9.723298002445985e-05,
+      "loss": 4.322,
+      "step": 3265
+    },
+    {
+      "epoch": 0.22217692621280066,
+      "grad_norm": 0.3089415729045868,
+      "learning_rate": 9.722873352357658e-05,
+      "loss": 4.5175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.22251664628346243,
+      "grad_norm": 0.2517615556716919,
+      "learning_rate": 9.72244870226933e-05,
+      "loss": 4.4571,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2228563663541242,
+      "grad_norm": 0.17470265924930573,
+      "learning_rate": 9.722024052181004e-05,
+      "loss": 4.2085,
+      "step": 3280
+    },
+    {
+      "epoch": 0.22319608642478597,
+      "grad_norm": 0.22137637436389923,
+      "learning_rate": 9.721599402092676e-05,
+      "loss": 4.597,
+      "step": 3285
+    },
+    {
+      "epoch": 0.22353580649544774,
+      "grad_norm": 0.3537333309650421,
+      "learning_rate": 9.721174752004348e-05,
+      "loss": 4.4751,
+      "step": 3290
+    },
+    {
+      "epoch": 0.22387552656610954,
+      "grad_norm": 0.24241957068443298,
+      "learning_rate": 9.720750101916022e-05,
+      "loss": 4.0842,
+      "step": 3295
+    },
+    {
+      "epoch": 0.2242152466367713,
+      "grad_norm": 0.2881457805633545,
+      "learning_rate": 9.720325451827695e-05,
+      "loss": 4.1991,
+      "step": 3300
+    },
+    {
+      "epoch": 0.22455496670743308,
+      "grad_norm": 0.23095691204071045,
+      "learning_rate": 9.719900801739366e-05,
+      "loss": 4.4024,
+      "step": 3305
+    },
+    {
+      "epoch": 0.22489468677809485,
+      "grad_norm": 0.25291046500205994,
+      "learning_rate": 9.71947615165104e-05,
+      "loss": 4.1577,
+      "step": 3310
+    },
+    {
+      "epoch": 0.22523440684875662,
+      "grad_norm": 0.2241574227809906,
+      "learning_rate": 9.719051501562713e-05,
+      "loss": 4.2593,
+      "step": 3315
+    },
+    {
+      "epoch": 0.2255741269194184,
+      "grad_norm": 0.19019931554794312,
+      "learning_rate": 9.718626851474386e-05,
+      "loss": 4.1674,
+      "step": 3320
+    },
+    {
+      "epoch": 0.22591384699008019,
+      "grad_norm": 0.2490902543067932,
+      "learning_rate": 9.718202201386059e-05,
+      "loss": 4.201,
+      "step": 3325
+    },
+    {
+      "epoch": 0.22625356706074196,
+      "grad_norm": 0.2902776896953583,
+      "learning_rate": 9.71777755129773e-05,
+      "loss": 4.5505,
+      "step": 3330
+    },
+    {
+      "epoch": 0.22659328713140373,
+      "grad_norm": 0.22167052328586578,
+      "learning_rate": 9.717352901209404e-05,
+      "loss": 4.282,
+      "step": 3335
+    },
+    {
+      "epoch": 0.2269330072020655,
+      "grad_norm": 3.241713523864746,
+      "learning_rate": 9.716928251121077e-05,
+      "loss": 4.095,
+      "step": 3340
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 0.2534385323524475,
+      "learning_rate": 9.716503601032749e-05,
+      "loss": 4.325,
+      "step": 3345
+    },
+    {
+      "epoch": 0.22761244734338903,
+      "grad_norm": 0.2039516121149063,
+      "learning_rate": 9.716078950944423e-05,
+      "loss": 4.3627,
+      "step": 3350
+    },
+    {
+      "epoch": 0.22795216741405083,
+      "grad_norm": 0.20797346532344818,
+      "learning_rate": 9.715654300856096e-05,
+      "loss": 4.1518,
+      "step": 3355
+    },
+    {
+      "epoch": 0.2282918874847126,
+      "grad_norm": 0.2560058534145355,
+      "learning_rate": 9.715229650767767e-05,
+      "loss": 4.2377,
+      "step": 3360
+    },
+    {
+      "epoch": 0.22863160755537437,
+      "grad_norm": 0.20020583271980286,
+      "learning_rate": 9.714805000679441e-05,
+      "loss": 4.4088,
+      "step": 3365
+    },
+    {
+      "epoch": 0.22897132762603614,
+      "grad_norm": 0.32701200246810913,
+      "learning_rate": 9.714380350591114e-05,
+      "loss": 4.4603,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2293110476966979,
+      "grad_norm": 0.16908589005470276,
+      "learning_rate": 9.713955700502786e-05,
+      "loss": 4.4159,
+      "step": 3375
+    },
+    {
+      "epoch": 0.2296507677673597,
+      "grad_norm": 10.86708927154541,
+      "learning_rate": 9.71353105041446e-05,
+      "loss": 4.373,
+      "step": 3380
+    },
+    {
+      "epoch": 0.22999048783802148,
+      "grad_norm": 0.2162582278251648,
+      "learning_rate": 9.713106400326132e-05,
+      "loss": 4.3303,
+      "step": 3385
+    },
+    {
+      "epoch": 0.23033020790868325,
+      "grad_norm": 0.1772332489490509,
+      "learning_rate": 9.712681750237804e-05,
+      "loss": 4.2396,
+      "step": 3390
+    },
+    {
+      "epoch": 0.23066992797934502,
+      "grad_norm": 0.36134952306747437,
+      "learning_rate": 9.712257100149478e-05,
+      "loss": 4.3838,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2310096480500068,
+      "grad_norm": 0.32894113659858704,
+      "learning_rate": 9.71183245006115e-05,
+      "loss": 4.177,
+      "step": 3400
+    },
+    {
+      "epoch": 0.23134936812066856,
+      "grad_norm": 0.2267148792743683,
+      "learning_rate": 9.711407799972822e-05,
+      "loss": 4.3994,
+      "step": 3405
+    },
+    {
+      "epoch": 0.23168908819133036,
+      "grad_norm": 0.22997945547103882,
+      "learning_rate": 9.710983149884496e-05,
+      "loss": 4.2585,
+      "step": 3410
+    },
+    {
+      "epoch": 0.23202880826199213,
+      "grad_norm": 0.21913081407546997,
+      "learning_rate": 9.710558499796168e-05,
+      "loss": 4.495,
+      "step": 3415
+    },
+    {
+      "epoch": 0.2323685283326539,
+      "grad_norm": 0.2355417013168335,
+      "learning_rate": 9.710133849707841e-05,
+      "loss": 4.4619,
+      "step": 3420
+    },
+    {
+      "epoch": 0.23270824840331567,
+      "grad_norm": 0.29134589433670044,
+      "learning_rate": 9.709709199619515e-05,
+      "loss": 4.3438,
+      "step": 3425
+    },
+    {
+      "epoch": 0.23304796847397743,
+      "grad_norm": 0.4645059108734131,
+      "learning_rate": 9.709284549531186e-05,
+      "loss": 4.467,
+      "step": 3430
+    },
+    {
+      "epoch": 0.2333876885446392,
+      "grad_norm": 0.3466382324695587,
+      "learning_rate": 9.708859899442859e-05,
+      "loss": 4.4636,
+      "step": 3435
+    },
+    {
+      "epoch": 0.233727408615301,
+      "grad_norm": 0.2788010835647583,
+      "learning_rate": 9.708435249354533e-05,
+      "loss": 4.1318,
+      "step": 3440
+    },
+    {
+      "epoch": 0.23406712868596277,
+      "grad_norm": 0.4784042537212372,
+      "learning_rate": 9.708010599266205e-05,
+      "loss": 4.2089,
+      "step": 3445
+    },
+    {
+      "epoch": 0.23440684875662454,
+      "grad_norm": 0.45934122800827026,
+      "learning_rate": 9.707585949177878e-05,
+      "loss": 4.3023,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2347465688272863,
+      "grad_norm": 0.25707322359085083,
+      "learning_rate": 9.707161299089552e-05,
+      "loss": 4.2901,
+      "step": 3455
+    },
+    {
+      "epoch": 0.23508628889794808,
+      "grad_norm": 0.4797256290912628,
+      "learning_rate": 9.706736649001223e-05,
+      "loss": 4.2574,
+      "step": 3460
+    },
+    {
+      "epoch": 0.23542600896860988,
+      "grad_norm": 0.2368171215057373,
+      "learning_rate": 9.706311998912896e-05,
+      "loss": 4.3998,
+      "step": 3465
+    },
+    {
+      "epoch": 0.23576572903927165,
+      "grad_norm": 1.7958965301513672,
+      "learning_rate": 9.705887348824569e-05,
+      "loss": 4.212,
+      "step": 3470
+    },
+    {
+      "epoch": 0.23610544910993342,
+      "grad_norm": 0.24695445597171783,
+      "learning_rate": 9.705462698736242e-05,
+      "loss": 4.3325,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2364451691805952,
+      "grad_norm": 0.24877724051475525,
+      "learning_rate": 9.705038048647914e-05,
+      "loss": 4.4167,
+      "step": 3480
+    },
+    {
+      "epoch": 0.23678488925125696,
+      "grad_norm": 0.2147648185491562,
+      "learning_rate": 9.704613398559587e-05,
+      "loss": 4.3511,
+      "step": 3485
+    },
+    {
+      "epoch": 0.23712460932191873,
+      "grad_norm": 0.38735896348953247,
+      "learning_rate": 9.70418874847126e-05,
+      "loss": 4.054,
+      "step": 3490
+    },
+    {
+      "epoch": 0.23746432939258053,
+      "grad_norm": 0.28407546877861023,
+      "learning_rate": 9.703764098382933e-05,
+      "loss": 4.1299,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2378040494632423,
+      "grad_norm": 0.4963781237602234,
+      "learning_rate": 9.703339448294606e-05,
+      "loss": 4.3751,
+      "step": 3500
+    },
+    {
+      "epoch": 0.23814376953390406,
+      "grad_norm": 0.2363215535879135,
+      "learning_rate": 9.702914798206278e-05,
+      "loss": 4.3631,
+      "step": 3505
+    },
+    {
+      "epoch": 0.23848348960456583,
+      "grad_norm": 0.2752895951271057,
+      "learning_rate": 9.702490148117951e-05,
+      "loss": 4.3192,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2388232096752276,
+      "grad_norm": 0.2011261135339737,
+      "learning_rate": 9.702065498029624e-05,
+      "loss": 4.3019,
+      "step": 3515
+    },
+    {
+      "epoch": 0.23916292974588937,
+      "grad_norm": 0.30605676770210266,
+      "learning_rate": 9.701640847941297e-05,
+      "loss": 4.2733,
+      "step": 3520
+    },
+    {
+      "epoch": 0.23950264981655117,
+      "grad_norm": 0.23777063190937042,
+      "learning_rate": 9.70121619785297e-05,
+      "loss": 4.4391,
+      "step": 3525
+    },
+    {
+      "epoch": 0.23984236988721294,
+      "grad_norm": 0.19578081369400024,
+      "learning_rate": 9.700791547764642e-05,
+      "loss": 4.3464,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2401820899578747,
+      "grad_norm": 0.43479400873184204,
+      "learning_rate": 9.700366897676315e-05,
+      "loss": 4.1509,
+      "step": 3535
+    },
+    {
+      "epoch": 0.24052181002853648,
+      "grad_norm": 0.23320983350276947,
+      "learning_rate": 9.699942247587988e-05,
+      "loss": 4.2031,
+      "step": 3540
+    },
+    {
+      "epoch": 0.24086153009919825,
+      "grad_norm": 0.395224004983902,
+      "learning_rate": 9.699517597499661e-05,
+      "loss": 4.3575,
+      "step": 3545
+    },
+    {
+      "epoch": 0.24120125016986005,
+      "grad_norm": 0.22553794085979462,
+      "learning_rate": 9.699092947411334e-05,
+      "loss": 4.4234,
+      "step": 3550
+    },
+    {
+      "epoch": 0.24154097024052182,
+      "grad_norm": 0.21396693587303162,
+      "learning_rate": 9.698668297323006e-05,
+      "loss": 4.3933,
+      "step": 3555
+    },
+    {
+      "epoch": 0.2418806903111836,
+      "grad_norm": 0.2883829176425934,
+      "learning_rate": 9.698243647234679e-05,
+      "loss": 4.1656,
+      "step": 3560
+    },
+    {
+      "epoch": 0.24222041038184536,
+      "grad_norm": 0.3761749267578125,
+      "learning_rate": 9.697818997146352e-05,
+      "loss": 4.254,
+      "step": 3565
+    },
+    {
+      "epoch": 0.24256013045250713,
+      "grad_norm": 0.2654179036617279,
+      "learning_rate": 9.697394347058025e-05,
+      "loss": 4.0754,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2428998505231689,
+      "grad_norm": 0.20404711365699768,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 4.4404,
+      "step": 3575
+    },
+    {
+      "epoch": 0.2432395705938307,
+      "grad_norm": 0.661677360534668,
+      "learning_rate": 9.69654504688137e-05,
+      "loss": 4.4084,
+      "step": 3580
+    },
+    {
+      "epoch": 0.24357929066449246,
+      "grad_norm": 0.19168756902217865,
+      "learning_rate": 9.696120396793043e-05,
+      "loss": 4.3991,
+      "step": 3585
+    },
+    {
+      "epoch": 0.24391901073515423,
+      "grad_norm": 0.21689128875732422,
+      "learning_rate": 9.695695746704716e-05,
+      "loss": 4.2182,
+      "step": 3590
+    },
+    {
+      "epoch": 0.244258730805816,
+      "grad_norm": 0.1910148561000824,
+      "learning_rate": 9.695271096616389e-05,
+      "loss": 4.2426,
+      "step": 3595
+    },
+    {
+      "epoch": 0.24459845087647777,
+      "grad_norm": 0.463371604681015,
+      "learning_rate": 9.69484644652806e-05,
+      "loss": 4.1501,
+      "step": 3600
+    },
+    {
+      "epoch": 0.24493817094713954,
+      "grad_norm": 0.2187051922082901,
+      "learning_rate": 9.694421796439734e-05,
+      "loss": 4.2383,
+      "step": 3605
+    },
+    {
+      "epoch": 0.24527789101780134,
+      "grad_norm": 0.7701082229614258,
+      "learning_rate": 9.693997146351407e-05,
+      "loss": 4.2811,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2456176110884631,
+      "grad_norm": 0.2454994171857834,
+      "learning_rate": 9.693572496263079e-05,
+      "loss": 4.3994,
+      "step": 3615
+    },
+    {
+      "epoch": 0.24595733115912488,
+      "grad_norm": 0.22179093956947327,
+      "learning_rate": 9.693147846174753e-05,
+      "loss": 4.1261,
+      "step": 3620
+    },
+    {
+      "epoch": 0.24629705122978665,
+      "grad_norm": 0.23975835740566254,
+      "learning_rate": 9.692723196086426e-05,
+      "loss": 4.308,
+      "step": 3625
+    },
+    {
+      "epoch": 0.24663677130044842,
+      "grad_norm": 0.21660096943378448,
+      "learning_rate": 9.692298545998097e-05,
+      "loss": 4.2105,
+      "step": 3630
+    },
+    {
+      "epoch": 0.24697649137111022,
+      "grad_norm": 0.22534438967704773,
+      "learning_rate": 9.691873895909771e-05,
+      "loss": 4.2923,
+      "step": 3635
+    },
+    {
+      "epoch": 0.247316211441772,
+      "grad_norm": 0.19649091362953186,
+      "learning_rate": 9.691449245821444e-05,
+      "loss": 4.4036,
+      "step": 3640
+    },
+    {
+      "epoch": 0.24765593151243376,
+      "grad_norm": 0.32042601704597473,
+      "learning_rate": 9.691024595733115e-05,
+      "loss": 4.4266,
+      "step": 3645
+    },
+    {
+      "epoch": 0.24799565158309553,
+      "grad_norm": 0.6859878301620483,
+      "learning_rate": 9.69059994564479e-05,
+      "loss": 4.2232,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2483353716537573,
+      "grad_norm": 0.23352079093456268,
+      "learning_rate": 9.690175295556462e-05,
+      "loss": 4.0696,
+      "step": 3655
+    },
+    {
+      "epoch": 0.24867509172441907,
+      "grad_norm": 0.272712767124176,
+      "learning_rate": 9.689750645468135e-05,
+      "loss": 4.1759,
+      "step": 3660
+    },
+    {
+      "epoch": 0.24901481179508086,
+      "grad_norm": 0.22009974718093872,
+      "learning_rate": 9.689325995379808e-05,
+      "loss": 4.1973,
+      "step": 3665
+    },
+    {
+      "epoch": 0.24935453186574263,
+      "grad_norm": 1.4543390274047852,
+      "learning_rate": 9.68890134529148e-05,
+      "loss": 4.314,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2496942519364044,
+      "grad_norm": 0.3941098153591156,
+      "learning_rate": 9.688476695203154e-05,
+      "loss": 4.4651,
+      "step": 3675
+    },
+    {
+      "epoch": 0.2500339720070662,
+      "grad_norm": 0.28159454464912415,
+      "learning_rate": 9.688052045114826e-05,
+      "loss": 4.2579,
+      "step": 3680
+    },
+    {
+      "epoch": 0.25037369207772797,
+      "grad_norm": 0.22340060770511627,
+      "learning_rate": 9.687627395026498e-05,
+      "loss": 4.2427,
+      "step": 3685
+    },
+    {
+      "epoch": 0.2507134121483897,
+      "grad_norm": 0.24438177049160004,
+      "learning_rate": 9.687202744938172e-05,
+      "loss": 4.1173,
+      "step": 3690
+    },
+    {
+      "epoch": 0.2510531322190515,
+      "grad_norm": 0.19045932590961456,
+      "learning_rate": 9.686778094849845e-05,
+      "loss": 4.3531,
+      "step": 3695
+    },
+    {
+      "epoch": 0.25139285228971325,
+      "grad_norm": 0.21072247624397278,
+      "learning_rate": 9.686353444761516e-05,
+      "loss": 4.3211,
+      "step": 3700
+    },
+    {
+      "epoch": 0.25173257236037505,
+      "grad_norm": 0.20157082378864288,
+      "learning_rate": 9.68592879467319e-05,
+      "loss": 4.3939,
+      "step": 3705
+    },
+    {
+      "epoch": 0.25207229243103685,
+      "grad_norm": 0.23919062316417694,
+      "learning_rate": 9.685504144584863e-05,
+      "loss": 4.2216,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2524120125016986,
+      "grad_norm": 0.3379192352294922,
+      "learning_rate": 9.685079494496535e-05,
+      "loss": 4.15,
+      "step": 3715
+    },
+    {
+      "epoch": 0.2527517325723604,
+      "grad_norm": 0.2691631615161896,
+      "learning_rate": 9.684654844408209e-05,
+      "loss": 4.178,
+      "step": 3720
+    },
+    {
+      "epoch": 0.25309145264302213,
+      "grad_norm": 0.2460995614528656,
+      "learning_rate": 9.684230194319882e-05,
+      "loss": 4.1714,
+      "step": 3725
+    },
+    {
+      "epoch": 0.2534311727136839,
+      "grad_norm": 0.24896664917469025,
+      "learning_rate": 9.683805544231553e-05,
+      "loss": 4.5137,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2537708927843457,
+      "grad_norm": 0.2998896837234497,
+      "learning_rate": 9.683380894143227e-05,
+      "loss": 4.2024,
+      "step": 3735
+    },
+    {
+      "epoch": 0.25411061285500747,
+      "grad_norm": 0.2170042097568512,
+      "learning_rate": 9.6829562440549e-05,
+      "loss": 4.2967,
+      "step": 3740
+    },
+    {
+      "epoch": 0.25445033292566926,
+      "grad_norm": 0.2534918487071991,
+      "learning_rate": 9.682531593966571e-05,
+      "loss": 4.1857,
+      "step": 3745
+    },
+    {
+      "epoch": 0.254790052996331,
+      "grad_norm": 0.2045327126979828,
+      "learning_rate": 9.682106943878246e-05,
+      "loss": 3.9961,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2551297730669928,
+      "grad_norm": 0.23638245463371277,
+      "learning_rate": 9.681682293789917e-05,
+      "loss": 4.3547,
+      "step": 3755
+    },
+    {
+      "epoch": 0.25546949313765455,
+      "grad_norm": 0.22549360990524292,
+      "learning_rate": 9.68125764370159e-05,
+      "loss": 4.1726,
+      "step": 3760
+    },
+    {
+      "epoch": 0.25580921320831634,
+      "grad_norm": 0.24715493619441986,
+      "learning_rate": 9.680832993613264e-05,
+      "loss": 4.2836,
+      "step": 3765
+    },
+    {
+      "epoch": 0.25614893327897814,
+      "grad_norm": 0.33308762311935425,
+      "learning_rate": 9.680408343524935e-05,
+      "loss": 4.2667,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2564886533496399,
+      "grad_norm": 0.4240279197692871,
+      "learning_rate": 9.679983693436608e-05,
+      "loss": 4.2174,
+      "step": 3775
+    },
+    {
+      "epoch": 0.2568283734203017,
+      "grad_norm": 0.26198095083236694,
+      "learning_rate": 9.679559043348282e-05,
+      "loss": 4.33,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2571680934909634,
+      "grad_norm": 0.21898075938224792,
+      "learning_rate": 9.679134393259954e-05,
+      "loss": 4.2754,
+      "step": 3785
+    },
+    {
+      "epoch": 0.2575078135616252,
+      "grad_norm": 0.1936497837305069,
+      "learning_rate": 9.678709743171627e-05,
+      "loss": 4.2178,
+      "step": 3790
+    },
+    {
+      "epoch": 0.257847533632287,
+      "grad_norm": 0.3042401075363159,
+      "learning_rate": 9.678285093083301e-05,
+      "loss": 4.2444,
+      "step": 3795
+    },
+    {
+      "epoch": 0.25818725370294876,
+      "grad_norm": 0.22089192271232605,
+      "learning_rate": 9.677860442994972e-05,
+      "loss": 4.1209,
+      "step": 3800
+    },
+    {
+      "epoch": 0.25852697377361056,
+      "grad_norm": 0.26595672965049744,
+      "learning_rate": 9.677435792906645e-05,
+      "loss": 4.3664,
+      "step": 3805
+    },
+    {
+      "epoch": 0.2588666938442723,
+      "grad_norm": 0.38972827792167664,
+      "learning_rate": 9.677011142818319e-05,
+      "loss": 4.3291,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2592064139149341,
+      "grad_norm": 0.8308687210083008,
+      "learning_rate": 9.67658649272999e-05,
+      "loss": 4.3588,
+      "step": 3815
+    },
+    {
+      "epoch": 0.2595461339855959,
+      "grad_norm": 0.29095426201820374,
+      "learning_rate": 9.676161842641663e-05,
+      "loss": 4.2095,
+      "step": 3820
+    },
+    {
+      "epoch": 0.25988585405625764,
+      "grad_norm": 0.6666823625564575,
+      "learning_rate": 9.675737192553336e-05,
+      "loss": 4.249,
+      "step": 3825
+    },
+    {
+      "epoch": 0.26022557412691943,
+      "grad_norm": 0.2800503373146057,
+      "learning_rate": 9.675312542465009e-05,
+      "loss": 4.0724,
+      "step": 3830
+    },
+    {
+      "epoch": 0.2605652941975812,
+      "grad_norm": 0.31251639127731323,
+      "learning_rate": 9.674887892376682e-05,
+      "loss": 4.2177,
+      "step": 3835
+    },
+    {
+      "epoch": 0.260905014268243,
+      "grad_norm": 0.19203290343284607,
+      "learning_rate": 9.674463242288355e-05,
+      "loss": 4.1164,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2612447343389047,
+      "grad_norm": 0.21506604552268982,
+      "learning_rate": 9.674038592200027e-05,
+      "loss": 4.3448,
+      "step": 3845
+    },
+    {
+      "epoch": 0.2615844544095665,
+      "grad_norm": 0.2286742478609085,
+      "learning_rate": 9.6736139421117e-05,
+      "loss": 4.3154,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2619241744802283,
+      "grad_norm": 0.22341595590114594,
+      "learning_rate": 9.673189292023373e-05,
+      "loss": 4.0496,
+      "step": 3855
+    },
+    {
+      "epoch": 0.26226389455089005,
+      "grad_norm": 3.2631723880767822,
+      "learning_rate": 9.672764641935046e-05,
+      "loss": 4.2739,
+      "step": 3860
+    },
+    {
+      "epoch": 0.26260361462155185,
+      "grad_norm": 0.21692293882369995,
+      "learning_rate": 9.672339991846719e-05,
+      "loss": 4.2239,
+      "step": 3865
+    },
+    {
+      "epoch": 0.2629433346922136,
+      "grad_norm": 0.24772769212722778,
+      "learning_rate": 9.671915341758391e-05,
+      "loss": 4.0354,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2632830547628754,
+      "grad_norm": 0.2190844863653183,
+      "learning_rate": 9.671490691670064e-05,
+      "loss": 4.2554,
+      "step": 3875
+    },
+    {
+      "epoch": 0.2636227748335372,
+      "grad_norm": 0.19608178734779358,
+      "learning_rate": 9.671066041581737e-05,
+      "loss": 4.2199,
+      "step": 3880
+    },
+    {
+      "epoch": 0.26396249490419893,
+      "grad_norm": 0.22313562035560608,
+      "learning_rate": 9.67064139149341e-05,
+      "loss": 4.1976,
+      "step": 3885
+    },
+    {
+      "epoch": 0.2643022149748607,
+      "grad_norm": 0.25129613280296326,
+      "learning_rate": 9.670216741405083e-05,
+      "loss": 4.2595,
+      "step": 3890
+    },
+    {
+      "epoch": 0.26464193504552247,
+      "grad_norm": 0.19212405383586884,
+      "learning_rate": 9.669792091316755e-05,
+      "loss": 4.3704,
+      "step": 3895
+    },
+    {
+      "epoch": 0.26498165511618427,
+      "grad_norm": 0.21401169896125793,
+      "learning_rate": 9.669367441228428e-05,
+      "loss": 4.1088,
+      "step": 3900
+    },
+    {
+      "epoch": 0.26532137518684606,
+      "grad_norm": 0.2625492811203003,
+      "learning_rate": 9.668942791140101e-05,
+      "loss": 4.4765,
+      "step": 3905
+    },
+    {
+      "epoch": 0.2656610952575078,
+      "grad_norm": 0.23690305650234222,
+      "learning_rate": 9.668518141051774e-05,
+      "loss": 4.1193,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2660008153281696,
+      "grad_norm": 0.2038702368736267,
+      "learning_rate": 9.668093490963447e-05,
+      "loss": 4.4361,
+      "step": 3915
+    },
+    {
+      "epoch": 0.26634053539883135,
+      "grad_norm": 1.9976972341537476,
+      "learning_rate": 9.66766884087512e-05,
+      "loss": 4.448,
+      "step": 3920
+    },
+    {
+      "epoch": 0.26668025546949314,
+      "grad_norm": 0.2619224488735199,
+      "learning_rate": 9.667244190786792e-05,
+      "loss": 4.3061,
+      "step": 3925
+    },
+    {
+      "epoch": 0.2670199755401549,
+      "grad_norm": 0.17488695681095123,
+      "learning_rate": 9.666819540698465e-05,
+      "loss": 3.924,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2673596956108167,
+      "grad_norm": 0.3555572032928467,
+      "learning_rate": 9.666394890610138e-05,
+      "loss": 4.4889,
+      "step": 3935
+    },
+    {
+      "epoch": 0.2676994156814785,
+      "grad_norm": 0.18651026487350464,
+      "learning_rate": 9.66597024052181e-05,
+      "loss": 4.2601,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2680391357521402,
+      "grad_norm": 0.4118260145187378,
+      "learning_rate": 9.665545590433483e-05,
+      "loss": 4.0048,
+      "step": 3945
+    },
+    {
+      "epoch": 0.268378855822802,
+      "grad_norm": 0.21420472860336304,
+      "learning_rate": 9.665120940345156e-05,
+      "loss": 4.5124,
+      "step": 3950
+    },
+    {
+      "epoch": 0.26871857589346376,
+      "grad_norm": 0.25867247581481934,
+      "learning_rate": 9.664696290256828e-05,
+      "loss": 3.9366,
+      "step": 3955
+    },
+    {
+      "epoch": 0.26905829596412556,
+      "grad_norm": 0.9560242295265198,
+      "learning_rate": 9.664271640168502e-05,
+      "loss": 4.273,
+      "step": 3960
+    },
+    {
+      "epoch": 0.26939801603478736,
+      "grad_norm": 0.22547510266304016,
+      "learning_rate": 9.663846990080175e-05,
+      "loss": 4.4065,
+      "step": 3965
+    },
+    {
+      "epoch": 0.2697377361054491,
+      "grad_norm": 0.4761745035648346,
+      "learning_rate": 9.663422339991846e-05,
+      "loss": 4.0622,
+      "step": 3970
+    },
+    {
+      "epoch": 0.2700774561761109,
+      "grad_norm": 0.26078933477401733,
+      "learning_rate": 9.66299768990352e-05,
+      "loss": 4.4622,
+      "step": 3975
+    },
+    {
+      "epoch": 0.27041717624677264,
+      "grad_norm": 0.21970224380493164,
+      "learning_rate": 9.662573039815193e-05,
+      "loss": 4.2412,
+      "step": 3980
+    },
+    {
+      "epoch": 0.27075689631743444,
+      "grad_norm": 0.5211921334266663,
+      "learning_rate": 9.662148389726865e-05,
+      "loss": 4.3848,
+      "step": 3985
+    },
+    {
+      "epoch": 0.27109661638809623,
+      "grad_norm": 0.31244269013404846,
+      "learning_rate": 9.661723739638539e-05,
+      "loss": 4.3852,
+      "step": 3990
+    },
+    {
+      "epoch": 0.271436336458758,
+      "grad_norm": 0.29353293776512146,
+      "learning_rate": 9.661299089550211e-05,
+      "loss": 4.091,
+      "step": 3995
+    },
+    {
+      "epoch": 0.2717760565294198,
+      "grad_norm": 0.23753587901592255,
+      "learning_rate": 9.660874439461884e-05,
+      "loss": 4.3828,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2721157766000815,
+      "grad_norm": 0.2865026593208313,
+      "learning_rate": 9.660449789373557e-05,
+      "loss": 4.2366,
+      "step": 4005
+    },
+    {
+      "epoch": 0.2724554966707433,
+      "grad_norm": 0.32267966866493225,
+      "learning_rate": 9.66002513928523e-05,
+      "loss": 4.1455,
+      "step": 4010
+    },
+    {
+      "epoch": 0.27279521674140506,
+      "grad_norm": 0.25711655616760254,
+      "learning_rate": 9.659600489196903e-05,
+      "loss": 4.4231,
+      "step": 4015
+    },
+    {
+      "epoch": 0.27313493681206685,
+      "grad_norm": 0.21570606529712677,
+      "learning_rate": 9.659175839108575e-05,
+      "loss": 4.2092,
+      "step": 4020
+    },
+    {
+      "epoch": 0.27347465688272865,
+      "grad_norm": 0.22739212214946747,
+      "learning_rate": 9.658751189020247e-05,
+      "loss": 4.2107,
+      "step": 4025
+    },
+    {
+      "epoch": 0.2738143769533904,
+      "grad_norm": 0.22698377072811127,
+      "learning_rate": 9.658326538931921e-05,
+      "loss": 3.965,
+      "step": 4030
+    },
+    {
+      "epoch": 0.2741540970240522,
+      "grad_norm": 0.2108132541179657,
+      "learning_rate": 9.657901888843594e-05,
+      "loss": 4.3312,
+      "step": 4035
+    },
+    {
+      "epoch": 0.27449381709471393,
+      "grad_norm": 0.3985457122325897,
+      "learning_rate": 9.657477238755265e-05,
+      "loss": 4.1752,
+      "step": 4040
+    },
+    {
+      "epoch": 0.27483353716537573,
+      "grad_norm": 0.2395816147327423,
+      "learning_rate": 9.65705258866694e-05,
+      "loss": 4.6349,
+      "step": 4045
+    },
+    {
+      "epoch": 0.2751732572360375,
+      "grad_norm": 0.24473239481449127,
+      "learning_rate": 9.656627938578612e-05,
+      "loss": 4.4361,
+      "step": 4050
+    },
+    {
+      "epoch": 0.27551297730669927,
+      "grad_norm": 0.3756929636001587,
+      "learning_rate": 9.656203288490284e-05,
+      "loss": 4.2976,
+      "step": 4055
+    },
+    {
+      "epoch": 0.27585269737736107,
+      "grad_norm": 0.2284708023071289,
+      "learning_rate": 9.655778638401958e-05,
+      "loss": 4.2554,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2761924174480228,
+      "grad_norm": 0.26325544714927673,
+      "learning_rate": 9.655353988313631e-05,
+      "loss": 4.152,
+      "step": 4065
+    },
+    {
+      "epoch": 0.2765321375186846,
+      "grad_norm": 0.20146878063678741,
+      "learning_rate": 9.654929338225302e-05,
+      "loss": 4.1147,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2768718575893464,
+      "grad_norm": 0.21572470664978027,
+      "learning_rate": 9.654504688136976e-05,
+      "loss": 4.2792,
+      "step": 4075
+    },
+    {
+      "epoch": 0.27721157766000815,
+      "grad_norm": 0.32967862486839294,
+      "learning_rate": 9.654080038048649e-05,
+      "loss": 4.1379,
+      "step": 4080
+    },
+    {
+      "epoch": 0.27755129773066994,
+      "grad_norm": 0.23224560916423798,
+      "learning_rate": 9.65365538796032e-05,
+      "loss": 4.0244,
+      "step": 4085
+    },
+    {
+      "epoch": 0.2778910178013317,
+      "grad_norm": 0.2101649045944214,
+      "learning_rate": 9.653230737871995e-05,
+      "loss": 4.2301,
+      "step": 4090
+    },
+    {
+      "epoch": 0.2782307378719935,
+      "grad_norm": 0.22994142770767212,
+      "learning_rate": 9.652806087783666e-05,
+      "loss": 4.1505,
+      "step": 4095
+    },
+    {
+      "epoch": 0.2785704579426552,
+      "grad_norm": 0.23458316922187805,
+      "learning_rate": 9.652381437695339e-05,
+      "loss": 4.2603,
+      "step": 4100
+    },
+    {
+      "epoch": 0.278910178013317,
+      "grad_norm": 1.6248669624328613,
+      "learning_rate": 9.651956787607013e-05,
+      "loss": 4.1157,
+      "step": 4105
+    },
+    {
+      "epoch": 0.2792498980839788,
+      "grad_norm": 0.49504292011260986,
+      "learning_rate": 9.651532137518685e-05,
+      "loss": 4.2854,
+      "step": 4110
+    },
+    {
+      "epoch": 0.27958961815464056,
+      "grad_norm": 0.2064412534236908,
+      "learning_rate": 9.651107487430357e-05,
+      "loss": 4.2062,
+      "step": 4115
+    },
+    {
+      "epoch": 0.27992933822530236,
+      "grad_norm": 0.26098746061325073,
+      "learning_rate": 9.650682837342031e-05,
+      "loss": 4.1904,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2802690582959641,
+      "grad_norm": 1.5820512771606445,
+      "learning_rate": 9.650258187253703e-05,
+      "loss": 4.0721,
+      "step": 4125
+    },
+    {
+      "epoch": 0.2806087783666259,
+      "grad_norm": 0.28080296516418457,
+      "learning_rate": 9.649833537165376e-05,
+      "loss": 4.1324,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2809484984372877,
+      "grad_norm": 0.186203733086586,
+      "learning_rate": 9.64940888707705e-05,
+      "loss": 4.0794,
+      "step": 4135
+    },
+    {
+      "epoch": 0.28128821850794944,
+      "grad_norm": 0.3637949824333191,
+      "learning_rate": 9.648984236988721e-05,
+      "loss": 4.135,
+      "step": 4140
+    },
+    {
+      "epoch": 0.28162793857861124,
+      "grad_norm": 0.24603451788425446,
+      "learning_rate": 9.648559586900394e-05,
+      "loss": 4.3026,
+      "step": 4145
+    },
+    {
+      "epoch": 0.281967658649273,
+      "grad_norm": 0.21384142339229584,
+      "learning_rate": 9.648134936812068e-05,
+      "loss": 4.5439,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2823073787199348,
+      "grad_norm": 0.23172922432422638,
+      "learning_rate": 9.64771028672374e-05,
+      "loss": 4.2762,
+      "step": 4155
+    },
+    {
+      "epoch": 0.2826470987905966,
+      "grad_norm": 0.22065469622612,
+      "learning_rate": 9.647285636635413e-05,
+      "loss": 4.3842,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2829868188612583,
+      "grad_norm": 3.40319561958313,
+      "learning_rate": 9.646860986547087e-05,
+      "loss": 4.1898,
+      "step": 4165
+    },
+    {
+      "epoch": 0.2833265389319201,
+      "grad_norm": 0.20233049988746643,
+      "learning_rate": 9.646436336458758e-05,
+      "loss": 4.0764,
+      "step": 4170
+    },
+    {
+      "epoch": 0.28366625900258186,
+      "grad_norm": 0.586283802986145,
+      "learning_rate": 9.646011686370431e-05,
+      "loss": 4.314,
+      "step": 4175
+    },
+    {
+      "epoch": 0.28400597907324365,
+      "grad_norm": 2.9164395332336426,
+      "learning_rate": 9.645587036282104e-05,
+      "loss": 4.0146,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2843456991439054,
+      "grad_norm": 0.28996986150741577,
+      "learning_rate": 9.645162386193777e-05,
+      "loss": 4.2867,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2846854192145672,
+      "grad_norm": 0.5857749581336975,
+      "learning_rate": 9.64473773610545e-05,
+      "loss": 3.9711,
+      "step": 4190
+    },
+    {
+      "epoch": 0.285025139285229,
+      "grad_norm": 0.2920601963996887,
+      "learning_rate": 9.644313086017122e-05,
+      "loss": 4.2076,
+      "step": 4195
+    },
+    {
+      "epoch": 0.28536485935589073,
+      "grad_norm": 0.5181906223297119,
+      "learning_rate": 9.643888435928795e-05,
+      "loss": 4.2119,
+      "step": 4200
+    },
+    {
+      "epoch": 0.28570457942655253,
+      "grad_norm": 0.28782758116722107,
+      "learning_rate": 9.643463785840468e-05,
+      "loss": 4.3009,
+      "step": 4205
+    },
+    {
+      "epoch": 0.28604429949721427,
+      "grad_norm": 0.23932106792926788,
+      "learning_rate": 9.64303913575214e-05,
+      "loss": 4.2094,
+      "step": 4210
+    },
+    {
+      "epoch": 0.28638401956787607,
+      "grad_norm": 0.3413240611553192,
+      "learning_rate": 9.642614485663813e-05,
+      "loss": 4.2305,
+      "step": 4215
+    },
+    {
+      "epoch": 0.28672373963853787,
+      "grad_norm": 2.542914390563965,
+      "learning_rate": 9.642189835575486e-05,
+      "loss": 4.4582,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2870634597091996,
+      "grad_norm": 0.21945473551750183,
+      "learning_rate": 9.641765185487159e-05,
+      "loss": 4.3744,
+      "step": 4225
+    },
+    {
+      "epoch": 0.2874031797798614,
+      "grad_norm": 0.209177166223526,
+      "learning_rate": 9.641340535398832e-05,
+      "loss": 4.2808,
+      "step": 4230
+    },
+    {
+      "epoch": 0.28774289985052315,
+      "grad_norm": 0.2568071782588959,
+      "learning_rate": 9.640915885310505e-05,
+      "loss": 3.9962,
+      "step": 4235
+    },
+    {
+      "epoch": 0.28808261992118495,
+      "grad_norm": 0.31277233362197876,
+      "learning_rate": 9.640491235222177e-05,
+      "loss": 4.3054,
+      "step": 4240
+    },
+    {
+      "epoch": 0.28842233999184674,
+      "grad_norm": 0.24698586761951447,
+      "learning_rate": 9.64006658513385e-05,
+      "loss": 4.5058,
+      "step": 4245
+    },
+    {
+      "epoch": 0.2887620600625085,
+      "grad_norm": 0.23559850454330444,
+      "learning_rate": 9.639641935045523e-05,
+      "loss": 4.0805,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2891017801331703,
+      "grad_norm": 0.22997041046619415,
+      "learning_rate": 9.639217284957196e-05,
+      "loss": 3.9677,
+      "step": 4255
+    },
+    {
+      "epoch": 0.289441500203832,
+      "grad_norm": 0.2484733760356903,
+      "learning_rate": 9.638792634868869e-05,
+      "loss": 4.4475,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2897812202744938,
+      "grad_norm": 0.19737331569194794,
+      "learning_rate": 9.638367984780541e-05,
+      "loss": 4.0734,
+      "step": 4265
+    },
+    {
+      "epoch": 0.29012094034515556,
+      "grad_norm": 0.21674834191799164,
+      "learning_rate": 9.637943334692214e-05,
+      "loss": 4.2674,
+      "step": 4270
+    },
+    {
+      "epoch": 0.29046066041581736,
+      "grad_norm": 1.3918811082839966,
+      "learning_rate": 9.637518684603887e-05,
+      "loss": 4.2559,
+      "step": 4275
+    },
+    {
+      "epoch": 0.29080038048647916,
+      "grad_norm": 0.3141735792160034,
+      "learning_rate": 9.63709403451556e-05,
+      "loss": 4.2088,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2911401005571409,
+      "grad_norm": 0.17463235557079315,
+      "learning_rate": 9.636669384427233e-05,
+      "loss": 4.2172,
+      "step": 4285
+    },
+    {
+      "epoch": 0.2914798206278027,
+      "grad_norm": 0.28866469860076904,
+      "learning_rate": 9.636244734338905e-05,
+      "loss": 4.1058,
+      "step": 4290
+    },
+    {
+      "epoch": 0.29181954069846444,
+      "grad_norm": 0.19368840754032135,
+      "learning_rate": 9.635820084250577e-05,
+      "loss": 4.3816,
+      "step": 4295
+    },
+    {
+      "epoch": 0.29215926076912624,
+      "grad_norm": 0.5456646680831909,
+      "learning_rate": 9.635395434162251e-05,
+      "loss": 4.004,
+      "step": 4300
+    },
+    {
+      "epoch": 0.29249898083978804,
+      "grad_norm": 0.5791100263595581,
+      "learning_rate": 9.634970784073924e-05,
+      "loss": 3.9493,
+      "step": 4305
+    },
+    {
+      "epoch": 0.2928387009104498,
+      "grad_norm": 0.18864502012729645,
+      "learning_rate": 9.634546133985595e-05,
+      "loss": 4.2906,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2931784209811116,
+      "grad_norm": 0.6542890071868896,
+      "learning_rate": 9.63412148389727e-05,
+      "loss": 4.2943,
+      "step": 4315
+    },
+    {
+      "epoch": 0.2935181410517733,
+      "grad_norm": 0.2639864683151245,
+      "learning_rate": 9.633696833808942e-05,
+      "loss": 4.1306,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2938578611224351,
+      "grad_norm": 0.24884024262428284,
+      "learning_rate": 9.633272183720614e-05,
+      "loss": 4.1749,
+      "step": 4325
+    },
+    {
+      "epoch": 0.2941975811930969,
+      "grad_norm": 1.6146323680877686,
+      "learning_rate": 9.632847533632288e-05,
+      "loss": 4.1622,
+      "step": 4330
+    },
+    {
+      "epoch": 0.29453730126375866,
+      "grad_norm": 0.19550690054893494,
+      "learning_rate": 9.63242288354396e-05,
+      "loss": 4.2481,
+      "step": 4335
+    },
+    {
+      "epoch": 0.29487702133442045,
+      "grad_norm": 0.48053327202796936,
+      "learning_rate": 9.631998233455633e-05,
+      "loss": 4.1654,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2952167414050822,
+      "grad_norm": 0.6082022190093994,
+      "learning_rate": 9.631573583367306e-05,
+      "loss": 3.922,
+      "step": 4345
+    },
+    {
+      "epoch": 0.295556461475744,
+      "grad_norm": 0.410819411277771,
+      "learning_rate": 9.631148933278979e-05,
+      "loss": 4.1838,
+      "step": 4350
+    },
+    {
+      "epoch": 0.29589618154640573,
+      "grad_norm": 0.20050150156021118,
+      "learning_rate": 9.630724283190652e-05,
+      "loss": 4.2083,
+      "step": 4355
+    },
+    {
+      "epoch": 0.29623590161706753,
+      "grad_norm": 0.2641303539276123,
+      "learning_rate": 9.630299633102325e-05,
+      "loss": 4.0804,
+      "step": 4360
+    },
+    {
+      "epoch": 0.29657562168772933,
+      "grad_norm": 0.1961575597524643,
+      "learning_rate": 9.629874983013997e-05,
+      "loss": 4.3261,
+      "step": 4365
+    },
+    {
+      "epoch": 0.29691534175839107,
+      "grad_norm": 0.17782385647296906,
+      "learning_rate": 9.62945033292567e-05,
+      "loss": 4.3017,
+      "step": 4370
+    },
+    {
+      "epoch": 0.29725506182905287,
+      "grad_norm": 1.2139571905136108,
+      "learning_rate": 9.629025682837343e-05,
+      "loss": 4.1051,
+      "step": 4375
+    },
+    {
+      "epoch": 0.2975947818997146,
+      "grad_norm": 0.2687116265296936,
+      "learning_rate": 9.628601032749014e-05,
+      "loss": 4.0757,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2979345019703764,
+      "grad_norm": 0.19845756888389587,
+      "learning_rate": 9.628176382660689e-05,
+      "loss": 4.0948,
+      "step": 4385
+    },
+    {
+      "epoch": 0.2982742220410382,
+      "grad_norm": 0.2517881989479065,
+      "learning_rate": 9.627751732572361e-05,
+      "loss": 4.3492,
+      "step": 4390
+    },
+    {
+      "epoch": 0.29861394211169995,
+      "grad_norm": 0.19443516433238983,
+      "learning_rate": 9.627327082484033e-05,
+      "loss": 4.2639,
+      "step": 4395
+    },
+    {
+      "epoch": 0.29895366218236175,
+      "grad_norm": 0.21518000960350037,
+      "learning_rate": 9.626902432395707e-05,
+      "loss": 4.0157,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2992933822530235,
+      "grad_norm": 0.3461875915527344,
+      "learning_rate": 9.62647778230738e-05,
+      "loss": 4.2115,
+      "step": 4405
+    },
+    {
+      "epoch": 0.2996331023236853,
+      "grad_norm": 0.39930984377861023,
+      "learning_rate": 9.626053132219051e-05,
+      "loss": 4.3391,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2999728223943471,
+      "grad_norm": 0.22730666399002075,
+      "learning_rate": 9.625628482130725e-05,
+      "loss": 4.4092,
+      "step": 4415
+    },
+    {
+      "epoch": 0.3003125424650088,
+      "grad_norm": 0.2596425414085388,
+      "learning_rate": 9.625203832042398e-05,
+      "loss": 4.2127,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3006522625356706,
+      "grad_norm": 0.19453459978103638,
+      "learning_rate": 9.62477918195407e-05,
+      "loss": 4.338,
+      "step": 4425
+    },
+    {
+      "epoch": 0.30099198260633236,
+      "grad_norm": 0.5006263852119446,
+      "learning_rate": 9.624354531865744e-05,
+      "loss": 4.3887,
+      "step": 4430
+    },
+    {
+      "epoch": 0.30133170267699416,
+      "grad_norm": 0.5625196099281311,
+      "learning_rate": 9.623929881777417e-05,
+      "loss": 4.411,
+      "step": 4435
+    },
+    {
+      "epoch": 0.3016714227476559,
+      "grad_norm": 0.2295573651790619,
+      "learning_rate": 9.623505231689088e-05,
+      "loss": 4.0798,
+      "step": 4440
+    },
+    {
+      "epoch": 0.3020111428183177,
+      "grad_norm": 0.22928239405155182,
+      "learning_rate": 9.623080581600762e-05,
+      "loss": 4.2019,
+      "step": 4445
+    },
+    {
+      "epoch": 0.3023508628889795,
+      "grad_norm": 0.7567266225814819,
+      "learning_rate": 9.622655931512434e-05,
+      "loss": 4.069,
+      "step": 4450
+    },
+    {
+      "epoch": 0.30269058295964124,
+      "grad_norm": 0.3791631758213043,
+      "learning_rate": 9.622231281424106e-05,
+      "loss": 4.0739,
+      "step": 4455
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.21165654063224792,
+      "learning_rate": 9.62180663133578e-05,
+      "loss": 4.0467,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3033700231009648,
+      "grad_norm": 0.2274232804775238,
+      "learning_rate": 9.621381981247452e-05,
+      "loss": 4.1655,
+      "step": 4465
+    },
+    {
+      "epoch": 0.3037097431716266,
+      "grad_norm": 0.2882256507873535,
+      "learning_rate": 9.620957331159125e-05,
+      "loss": 4.0162,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3040494632422884,
+      "grad_norm": 0.21718928217887878,
+      "learning_rate": 9.620532681070799e-05,
+      "loss": 4.1421,
+      "step": 4475
+    },
+    {
+      "epoch": 0.3043891833129501,
+      "grad_norm": 0.20482736825942993,
+      "learning_rate": 9.62010803098247e-05,
+      "loss": 4.3067,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3047289033836119,
+      "grad_norm": 0.8644359111785889,
+      "learning_rate": 9.619683380894143e-05,
+      "loss": 4.1044,
+      "step": 4485
+    },
+    {
+      "epoch": 0.30506862345427366,
+      "grad_norm": 0.2353629469871521,
+      "learning_rate": 9.619258730805817e-05,
+      "loss": 4.6311,
+      "step": 4490
+    },
+    {
+      "epoch": 0.30540834352493546,
+      "grad_norm": 0.4746248722076416,
+      "learning_rate": 9.618834080717489e-05,
+      "loss": 4.2593,
+      "step": 4495
+    },
+    {
+      "epoch": 0.30574806359559725,
+      "grad_norm": 0.412517786026001,
+      "learning_rate": 9.618409430629162e-05,
+      "loss": 4.2203,
+      "step": 4500
+    },
+    {
+      "epoch": 0.306087783666259,
+      "grad_norm": 0.1828744113445282,
+      "learning_rate": 9.617984780540836e-05,
+      "loss": 4.0572,
+      "step": 4505
+    },
+    {
+      "epoch": 0.3064275037369208,
+      "grad_norm": 0.2963522672653198,
+      "learning_rate": 9.617560130452507e-05,
+      "loss": 4.3707,
+      "step": 4510
+    },
+    {
+      "epoch": 0.30676722380758253,
+      "grad_norm": 0.20454883575439453,
+      "learning_rate": 9.61713548036418e-05,
+      "loss": 4.1609,
+      "step": 4515
+    },
+    {
+      "epoch": 0.30710694387824433,
+      "grad_norm": 0.9719656109809875,
+      "learning_rate": 9.616710830275853e-05,
+      "loss": 4.3215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3074466639489061,
+      "grad_norm": 0.6544187664985657,
+      "learning_rate": 9.616286180187526e-05,
+      "loss": 3.9987,
+      "step": 4525
+    },
+    {
+      "epoch": 0.30778638401956787,
+      "grad_norm": 0.3224717080593109,
+      "learning_rate": 9.615861530099198e-05,
+      "loss": 4.2707,
+      "step": 4530
+    },
+    {
+      "epoch": 0.30812610409022967,
+      "grad_norm": 0.3716621398925781,
+      "learning_rate": 9.615436880010871e-05,
+      "loss": 3.9584,
+      "step": 4535
+    },
+    {
+      "epoch": 0.3084658241608914,
+      "grad_norm": 0.23874157667160034,
+      "learning_rate": 9.615012229922544e-05,
+      "loss": 4.0501,
+      "step": 4540
+    },
+    {
+      "epoch": 0.3088055442315532,
+      "grad_norm": 0.30314287543296814,
+      "learning_rate": 9.614587579834217e-05,
+      "loss": 3.8515,
+      "step": 4545
+    },
+    {
+      "epoch": 0.30914526430221495,
+      "grad_norm": 0.19416074454784393,
+      "learning_rate": 9.61416292974589e-05,
+      "loss": 4.0812,
+      "step": 4550
+    },
+    {
+      "epoch": 0.30948498437287675,
+      "grad_norm": 0.5242770910263062,
+      "learning_rate": 9.613738279657562e-05,
+      "loss": 4.37,
+      "step": 4555
+    },
+    {
+      "epoch": 0.30982470444353855,
+      "grad_norm": 0.20864085853099823,
+      "learning_rate": 9.613313629569235e-05,
+      "loss": 4.2711,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3101644245142003,
+      "grad_norm": 0.20202593505382538,
+      "learning_rate": 9.612888979480908e-05,
+      "loss": 4.2423,
+      "step": 4565
+    },
+    {
+      "epoch": 0.3105041445848621,
+      "grad_norm": 0.5426074266433716,
+      "learning_rate": 9.612464329392581e-05,
+      "loss": 4.2333,
+      "step": 4570
+    },
+    {
+      "epoch": 0.3108438646555238,
+      "grad_norm": 0.21013125777244568,
+      "learning_rate": 9.612039679304254e-05,
+      "loss": 4.0275,
+      "step": 4575
+    },
+    {
+      "epoch": 0.3111835847261856,
+      "grad_norm": 0.3289850056171417,
+      "learning_rate": 9.611615029215926e-05,
+      "loss": 4.3321,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3115233047968474,
+      "grad_norm": 0.20981498062610626,
+      "learning_rate": 9.611190379127599e-05,
+      "loss": 4.3237,
+      "step": 4585
+    },
+    {
+      "epoch": 0.31186302486750916,
+      "grad_norm": 0.2699143886566162,
+      "learning_rate": 9.610765729039272e-05,
+      "loss": 4.1158,
+      "step": 4590
+    },
+    {
+      "epoch": 0.31220274493817096,
+      "grad_norm": 0.27582621574401855,
+      "learning_rate": 9.610341078950945e-05,
+      "loss": 4.1325,
+      "step": 4595
+    },
+    {
+      "epoch": 0.3125424650088327,
+      "grad_norm": 0.2031656950712204,
+      "learning_rate": 9.609916428862618e-05,
+      "loss": 4.2928,
+      "step": 4600
+    },
+    {
+      "epoch": 0.3128821850794945,
+      "grad_norm": 0.17339491844177246,
+      "learning_rate": 9.60949177877429e-05,
+      "loss": 4.2046,
+      "step": 4605
+    },
+    {
+      "epoch": 0.3132219051501563,
+      "grad_norm": 0.2266245186328888,
+      "learning_rate": 9.609067128685963e-05,
+      "loss": 4.0126,
+      "step": 4610
+    },
+    {
+      "epoch": 0.31356162522081804,
+      "grad_norm": 0.2599181830883026,
+      "learning_rate": 9.608642478597636e-05,
+      "loss": 4.2055,
+      "step": 4615
+    },
+    {
+      "epoch": 0.31390134529147984,
+      "grad_norm": 0.27137070894241333,
+      "learning_rate": 9.608217828509309e-05,
+      "loss": 4.3499,
+      "step": 4620
+    },
+    {
+      "epoch": 0.3142410653621416,
+      "grad_norm": 0.22108152508735657,
+      "learning_rate": 9.607793178420982e-05,
+      "loss": 4.257,
+      "step": 4625
+    },
+    {
+      "epoch": 0.3145807854328034,
+      "grad_norm": 0.2058125138282776,
+      "learning_rate": 9.607368528332654e-05,
+      "loss": 3.919,
+      "step": 4630
+    },
+    {
+      "epoch": 0.3149205055034651,
+      "grad_norm": 0.4298430383205414,
+      "learning_rate": 9.606943878244327e-05,
+      "loss": 4.0876,
+      "step": 4635
+    },
+    {
+      "epoch": 0.3152602255741269,
+      "grad_norm": 0.19494764506816864,
+      "learning_rate": 9.606519228156e-05,
+      "loss": 4.2404,
+      "step": 4640
+    },
+    {
+      "epoch": 0.3155999456447887,
+      "grad_norm": 0.35915589332580566,
+      "learning_rate": 9.606094578067673e-05,
+      "loss": 4.1185,
+      "step": 4645
+    },
+    {
+      "epoch": 0.31593966571545046,
+      "grad_norm": 0.395074725151062,
+      "learning_rate": 9.605669927979344e-05,
+      "loss": 4.2395,
+      "step": 4650
+    },
+    {
+      "epoch": 0.31627938578611225,
+      "grad_norm": 0.3247855007648468,
+      "learning_rate": 9.605245277891018e-05,
+      "loss": 3.9987,
+      "step": 4655
+    },
+    {
+      "epoch": 0.316619105856774,
+      "grad_norm": 0.18057437241077423,
+      "learning_rate": 9.604820627802691e-05,
+      "loss": 4.3912,
+      "step": 4660
+    },
+    {
+      "epoch": 0.3169588259274358,
+      "grad_norm": 0.25986525416374207,
+      "learning_rate": 9.604395977714363e-05,
+      "loss": 4.2514,
+      "step": 4665
+    },
+    {
+      "epoch": 0.3172985459980976,
+      "grad_norm": 0.2058873325586319,
+      "learning_rate": 9.603971327626037e-05,
+      "loss": 4.3795,
+      "step": 4670
+    },
+    {
+      "epoch": 0.31763826606875933,
+      "grad_norm": 0.30512815713882446,
+      "learning_rate": 9.60354667753771e-05,
+      "loss": 4.2392,
+      "step": 4675
+    },
+    {
+      "epoch": 0.31797798613942113,
+      "grad_norm": 0.18412257730960846,
+      "learning_rate": 9.603122027449382e-05,
+      "loss": 4.2086,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3183177062100829,
+      "grad_norm": 0.2081792950630188,
+      "learning_rate": 9.602697377361055e-05,
+      "loss": 4.3573,
+      "step": 4685
+    },
+    {
+      "epoch": 0.31865742628074467,
+      "grad_norm": 0.3510904014110565,
+      "learning_rate": 9.602272727272728e-05,
+      "loss": 4.3618,
+      "step": 4690
+    },
+    {
+      "epoch": 0.31899714635140647,
+      "grad_norm": 0.37145447731018066,
+      "learning_rate": 9.601848077184401e-05,
+      "loss": 4.301,
+      "step": 4695
+    },
+    {
+      "epoch": 0.3193368664220682,
+      "grad_norm": 0.23616161942481995,
+      "learning_rate": 9.601423427096074e-05,
+      "loss": 4.2158,
+      "step": 4700
+    },
+    {
+      "epoch": 0.31967658649273,
+      "grad_norm": 1.332667589187622,
+      "learning_rate": 9.600998777007746e-05,
+      "loss": 4.0569,
+      "step": 4705
+    },
+    {
+      "epoch": 0.32001630656339175,
+      "grad_norm": 0.21901053190231323,
+      "learning_rate": 9.600574126919419e-05,
+      "loss": 4.0555,
+      "step": 4710
+    },
+    {
+      "epoch": 0.32035602663405355,
+      "grad_norm": 0.18509267270565033,
+      "learning_rate": 9.600149476831092e-05,
+      "loss": 4.1219,
+      "step": 4715
+    },
+    {
+      "epoch": 0.3206957467047153,
+      "grad_norm": 0.17567375302314758,
+      "learning_rate": 9.599724826742764e-05,
+      "loss": 4.2921,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3210354667753771,
+      "grad_norm": 0.17881886661052704,
+      "learning_rate": 9.599300176654438e-05,
+      "loss": 4.2577,
+      "step": 4725
+    },
+    {
+      "epoch": 0.3213751868460389,
+      "grad_norm": 0.2836432158946991,
+      "learning_rate": 9.59887552656611e-05,
+      "loss": 4.1362,
+      "step": 4730
+    },
+    {
+      "epoch": 0.3217149069167006,
+      "grad_norm": 0.43188366293907166,
+      "learning_rate": 9.598450876477782e-05,
+      "loss": 3.7939,
+      "step": 4735
+    },
+    {
+      "epoch": 0.3220546269873624,
+      "grad_norm": 0.21462103724479675,
+      "learning_rate": 9.598026226389456e-05,
+      "loss": 4.166,
+      "step": 4740
+    },
+    {
+      "epoch": 0.32239434705802417,
+      "grad_norm": 0.16883856058120728,
+      "learning_rate": 9.597601576301129e-05,
+      "loss": 3.9182,
+      "step": 4745
+    },
+    {
+      "epoch": 0.32273406712868596,
+      "grad_norm": 0.32869285345077515,
+      "learning_rate": 9.5971769262128e-05,
+      "loss": 4.0832,
+      "step": 4750
+    },
+    {
+      "epoch": 0.32307378719934776,
+      "grad_norm": 0.22861923277378082,
+      "learning_rate": 9.596752276124474e-05,
+      "loss": 4.344,
+      "step": 4755
+    },
+    {
+      "epoch": 0.3234135072700095,
+      "grad_norm": 0.3567066192626953,
+      "learning_rate": 9.596327626036147e-05,
+      "loss": 4.0191,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3237532273406713,
+      "grad_norm": 0.23706111311912537,
+      "learning_rate": 9.595902975947819e-05,
+      "loss": 4.1505,
+      "step": 4765
+    },
+    {
+      "epoch": 0.32409294741133304,
+      "grad_norm": 0.22384607791900635,
+      "learning_rate": 9.595478325859493e-05,
+      "loss": 4.1113,
+      "step": 4770
+    },
+    {
+      "epoch": 0.32443266748199484,
+      "grad_norm": 0.1979631930589676,
+      "learning_rate": 9.595053675771166e-05,
+      "loss": 4.0686,
+      "step": 4775
+    },
+    {
+      "epoch": 0.32477238755265664,
+      "grad_norm": 0.3480912148952484,
+      "learning_rate": 9.594629025682837e-05,
+      "loss": 4.2106,
+      "step": 4780
+    },
+    {
+      "epoch": 0.3251121076233184,
+      "grad_norm": 0.23982703685760498,
+      "learning_rate": 9.594204375594511e-05,
+      "loss": 4.2114,
+      "step": 4785
+    },
+    {
+      "epoch": 0.3254518276939802,
+      "grad_norm": 1.1133886575698853,
+      "learning_rate": 9.593779725506184e-05,
+      "loss": 3.894,
+      "step": 4790
+    },
+    {
+      "epoch": 0.3257915477646419,
+      "grad_norm": 0.1953042596578598,
+      "learning_rate": 9.593355075417856e-05,
+      "loss": 4.1392,
+      "step": 4795
+    },
+    {
+      "epoch": 0.3261312678353037,
+      "grad_norm": 0.24198144674301147,
+      "learning_rate": 9.59293042532953e-05,
+      "loss": 3.9054,
+      "step": 4800
+    },
+    {
+      "epoch": 0.32647098790596546,
+      "grad_norm": 0.5570387840270996,
+      "learning_rate": 9.592505775241201e-05,
+      "loss": 3.9928,
+      "step": 4805
+    },
+    {
+      "epoch": 0.32681070797662726,
+      "grad_norm": 0.3616771399974823,
+      "learning_rate": 9.592081125152874e-05,
+      "loss": 4.3527,
+      "step": 4810
+    },
+    {
+      "epoch": 0.32715042804728905,
+      "grad_norm": 0.7734106183052063,
+      "learning_rate": 9.591656475064548e-05,
+      "loss": 3.9594,
+      "step": 4815
+    },
+    {
+      "epoch": 0.3274901481179508,
+      "grad_norm": 0.23176033794879913,
+      "learning_rate": 9.59123182497622e-05,
+      "loss": 4.2269,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3278298681886126,
+      "grad_norm": 0.22131270170211792,
+      "learning_rate": 9.590807174887892e-05,
+      "loss": 4.205,
+      "step": 4825
+    },
+    {
+      "epoch": 0.32816958825927434,
+      "grad_norm": 0.1967305839061737,
+      "learning_rate": 9.590382524799566e-05,
+      "loss": 4.1081,
+      "step": 4830
+    },
+    {
+      "epoch": 0.32850930832993613,
+      "grad_norm": 0.20413081347942352,
+      "learning_rate": 9.589957874711238e-05,
+      "loss": 4.2146,
+      "step": 4835
+    },
+    {
+      "epoch": 0.32884902840059793,
+      "grad_norm": 0.237589493393898,
+      "learning_rate": 9.589533224622911e-05,
+      "loss": 4.321,
+      "step": 4840
+    },
+    {
+      "epoch": 0.3291887484712597,
+      "grad_norm": 0.3263246417045593,
+      "learning_rate": 9.589108574534585e-05,
+      "loss": 4.2572,
+      "step": 4845
+    },
+    {
+      "epoch": 0.32952846854192147,
+      "grad_norm": 0.2639991343021393,
+      "learning_rate": 9.588683924446256e-05,
+      "loss": 4.3049,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3298681886125832,
+      "grad_norm": 0.33223286271095276,
+      "learning_rate": 9.588259274357929e-05,
+      "loss": 4.3715,
+      "step": 4855
+    },
+    {
+      "epoch": 0.330207908683245,
+      "grad_norm": 0.22897298634052277,
+      "learning_rate": 9.587834624269603e-05,
+      "loss": 3.93,
+      "step": 4860
+    },
+    {
+      "epoch": 0.3305476287539068,
+      "grad_norm": 0.3667212128639221,
+      "learning_rate": 9.587409974181275e-05,
+      "loss": 4.3149,
+      "step": 4865
+    },
+    {
+      "epoch": 0.33088734882456855,
+      "grad_norm": 0.22442007064819336,
+      "learning_rate": 9.586985324092948e-05,
+      "loss": 4.4189,
+      "step": 4870
+    },
+    {
+      "epoch": 0.33122706889523035,
+      "grad_norm": 0.18334710597991943,
+      "learning_rate": 9.58656067400462e-05,
+      "loss": 3.9457,
+      "step": 4875
+    },
+    {
+      "epoch": 0.3315667889658921,
+      "grad_norm": 0.34593579173088074,
+      "learning_rate": 9.586136023916293e-05,
+      "loss": 3.9007,
+      "step": 4880
+    },
+    {
+      "epoch": 0.3319065090365539,
+      "grad_norm": 0.24884456396102905,
+      "learning_rate": 9.585711373827966e-05,
+      "loss": 4.198,
+      "step": 4885
+    },
+    {
+      "epoch": 0.33224622910721563,
+      "grad_norm": 0.6150047183036804,
+      "learning_rate": 9.585286723739639e-05,
+      "loss": 4.1738,
+      "step": 4890
+    },
+    {
+      "epoch": 0.3325859491778774,
+      "grad_norm": 0.18449634313583374,
+      "learning_rate": 9.584862073651312e-05,
+      "loss": 4.3916,
+      "step": 4895
+    },
+    {
+      "epoch": 0.3329256692485392,
+      "grad_norm": 0.45811331272125244,
+      "learning_rate": 9.584437423562984e-05,
+      "loss": 4.0894,
+      "step": 4900
+    },
+    {
+      "epoch": 0.33326538931920097,
+      "grad_norm": 0.9670056700706482,
+      "learning_rate": 9.584012773474657e-05,
+      "loss": 4.0947,
+      "step": 4905
+    },
+    {
+      "epoch": 0.33360510938986276,
+      "grad_norm": 0.2828699052333832,
+      "learning_rate": 9.58358812338633e-05,
+      "loss": 4.27,
+      "step": 4910
+    },
+    {
+      "epoch": 0.3339448294605245,
+      "grad_norm": 0.22989730536937714,
+      "learning_rate": 9.583163473298003e-05,
+      "loss": 4.2658,
+      "step": 4915
+    },
+    {
+      "epoch": 0.3342845495311863,
+      "grad_norm": 0.4018714427947998,
+      "learning_rate": 9.582738823209676e-05,
+      "loss": 4.1278,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3346242696018481,
+      "grad_norm": 0.5296480059623718,
+      "learning_rate": 9.582314173121348e-05,
+      "loss": 4.1865,
+      "step": 4925
+    },
+    {
+      "epoch": 0.33496398967250984,
+      "grad_norm": 0.2477627843618393,
+      "learning_rate": 9.581889523033021e-05,
+      "loss": 4.4925,
+      "step": 4930
+    },
+    {
+      "epoch": 0.33530370974317164,
+      "grad_norm": 0.24414370954036713,
+      "learning_rate": 9.581464872944694e-05,
+      "loss": 4.1706,
+      "step": 4935
+    },
+    {
+      "epoch": 0.3356434298138334,
+      "grad_norm": 0.15603503584861755,
+      "learning_rate": 9.581040222856367e-05,
+      "loss": 4.0303,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3359831498844952,
+      "grad_norm": 0.17460083961486816,
+      "learning_rate": 9.58061557276804e-05,
+      "loss": 4.1986,
+      "step": 4945
+    },
+    {
+      "epoch": 0.336322869955157,
+      "grad_norm": 0.2035687118768692,
+      "learning_rate": 9.580190922679712e-05,
+      "loss": 4.3514,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3366625900258187,
+      "grad_norm": 0.4037059545516968,
+      "learning_rate": 9.579766272591385e-05,
+      "loss": 4.1645,
+      "step": 4955
+    },
+    {
+      "epoch": 0.3370023100964805,
+      "grad_norm": 0.22527378797531128,
+      "learning_rate": 9.579341622503058e-05,
+      "loss": 4.1399,
+      "step": 4960
+    },
+    {
+      "epoch": 0.33734203016714226,
+      "grad_norm": 0.27611520886421204,
+      "learning_rate": 9.578916972414731e-05,
+      "loss": 4.1679,
+      "step": 4965
+    },
+    {
+      "epoch": 0.33768175023780406,
+      "grad_norm": 0.26130980253219604,
+      "learning_rate": 9.578492322326404e-05,
+      "loss": 3.9786,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3380214703084658,
+      "grad_norm": 0.20753854513168335,
+      "learning_rate": 9.578067672238076e-05,
+      "loss": 4.0629,
+      "step": 4975
+    },
+    {
+      "epoch": 0.3383611903791276,
+      "grad_norm": 0.198018416762352,
+      "learning_rate": 9.577643022149749e-05,
+      "loss": 4.0956,
+      "step": 4980
+    },
+    {
+      "epoch": 0.3387009104497894,
+      "grad_norm": 0.21650417149066925,
+      "learning_rate": 9.577218372061422e-05,
+      "loss": 4.0965,
+      "step": 4985
+    },
+    {
+      "epoch": 0.33904063052045114,
+      "grad_norm": 0.3832937777042389,
+      "learning_rate": 9.576793721973095e-05,
+      "loss": 4.1086,
+      "step": 4990
+    },
+    {
+      "epoch": 0.33938035059111293,
+      "grad_norm": 0.23973116278648376,
+      "learning_rate": 9.576369071884768e-05,
+      "loss": 4.3533,
+      "step": 4995
+    },
+    {
+      "epoch": 0.3397200706617747,
+      "grad_norm": 0.3817537724971771,
+      "learning_rate": 9.57594442179644e-05,
+      "loss": 4.2127,
+      "step": 5000
+    },
+    {
+      "epoch": 0.3400597907324365,
+      "grad_norm": 0.2038896679878235,
+      "learning_rate": 9.575519771708112e-05,
+      "loss": 4.3264,
+      "step": 5005
+    },
+    {
+      "epoch": 0.34039951080309827,
+      "grad_norm": 0.23982146382331848,
+      "learning_rate": 9.575095121619786e-05,
+      "loss": 4.0702,
+      "step": 5010
+    },
+    {
+      "epoch": 0.34073923087376,
+      "grad_norm": 0.30504170060157776,
+      "learning_rate": 9.574670471531459e-05,
+      "loss": 3.9892,
+      "step": 5015
+    },
+    {
+      "epoch": 0.3410789509444218,
+      "grad_norm": 0.2355673611164093,
+      "learning_rate": 9.574245821443132e-05,
+      "loss": 3.9423,
+      "step": 5020
+    },
+    {
+      "epoch": 0.34141867101508355,
+      "grad_norm": 0.22874650359153748,
+      "learning_rate": 9.573821171354804e-05,
+      "loss": 3.9142,
+      "step": 5025
+    },
+    {
+      "epoch": 0.34175839108574535,
+      "grad_norm": 0.21437452733516693,
+      "learning_rate": 9.573396521266477e-05,
+      "loss": 4.1365,
+      "step": 5030
+    },
+    {
+      "epoch": 0.34209811115640715,
+      "grad_norm": 0.3095625936985016,
+      "learning_rate": 9.57297187117815e-05,
+      "loss": 4.1468,
+      "step": 5035
+    },
+    {
+      "epoch": 0.3424378312270689,
+      "grad_norm": 0.2202177494764328,
+      "learning_rate": 9.572547221089823e-05,
+      "loss": 4.01,
+      "step": 5040
+    },
+    {
+      "epoch": 0.3427775512977307,
+      "grad_norm": 0.32474327087402344,
+      "learning_rate": 9.572122571001496e-05,
+      "loss": 4.0868,
+      "step": 5045
+    },
+    {
+      "epoch": 0.34311727136839243,
+      "grad_norm": 0.2622280716896057,
+      "learning_rate": 9.571697920913168e-05,
+      "loss": 4.2103,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3434569914390542,
+      "grad_norm": 0.7448468208312988,
+      "learning_rate": 9.571273270824841e-05,
+      "loss": 4.2945,
+      "step": 5055
+    },
+    {
+      "epoch": 0.34379671150971597,
+      "grad_norm": 0.2269594371318817,
+      "learning_rate": 9.570848620736514e-05,
+      "loss": 3.9127,
+      "step": 5060
+    },
+    {
+      "epoch": 0.34413643158037777,
+      "grad_norm": 0.2645524740219116,
+      "learning_rate": 9.570423970648187e-05,
+      "loss": 3.9423,
+      "step": 5065
+    },
+    {
+      "epoch": 0.34447615165103956,
+      "grad_norm": 0.246607705950737,
+      "learning_rate": 9.56999932055986e-05,
+      "loss": 4.1347,
+      "step": 5070
+    },
+    {
+      "epoch": 0.3448158717217013,
+      "grad_norm": 0.19342374801635742,
+      "learning_rate": 9.569574670471531e-05,
+      "loss": 4.0787,
+      "step": 5075
+    },
+    {
+      "epoch": 0.3451555917923631,
+      "grad_norm": 0.31122297048568726,
+      "learning_rate": 9.569150020383205e-05,
+      "loss": 4.1522,
+      "step": 5080
+    },
+    {
+      "epoch": 0.34549531186302485,
+      "grad_norm": 1.0425968170166016,
+      "learning_rate": 9.568725370294878e-05,
+      "loss": 3.995,
+      "step": 5085
+    },
+    {
+      "epoch": 0.34583503193368664,
+      "grad_norm": 0.22739467024803162,
+      "learning_rate": 9.56830072020655e-05,
+      "loss": 3.9042,
+      "step": 5090
+    },
+    {
+      "epoch": 0.34617475200434844,
+      "grad_norm": 0.21344348788261414,
+      "learning_rate": 9.567876070118224e-05,
+      "loss": 4.1351,
+      "step": 5095
+    },
+    {
+      "epoch": 0.3465144720750102,
+      "grad_norm": 0.2140883356332779,
+      "learning_rate": 9.567451420029896e-05,
+      "loss": 3.9542,
+      "step": 5100
+    },
+    {
+      "epoch": 0.346854192145672,
+      "grad_norm": 0.1795925498008728,
+      "learning_rate": 9.567026769941568e-05,
+      "loss": 4.061,
+      "step": 5105
+    },
+    {
+      "epoch": 0.3471939122163337,
+      "grad_norm": 0.37569665908813477,
+      "learning_rate": 9.566602119853242e-05,
+      "loss": 4.3051,
+      "step": 5110
+    },
+    {
+      "epoch": 0.3475336322869955,
+      "grad_norm": 0.19528646767139435,
+      "learning_rate": 9.566177469764915e-05,
+      "loss": 4.1591,
+      "step": 5115
+    },
+    {
+      "epoch": 0.3478733523576573,
+      "grad_norm": 0.1688258945941925,
+      "learning_rate": 9.565752819676586e-05,
+      "loss": 3.9876,
+      "step": 5120
+    },
+    {
+      "epoch": 0.34821307242831906,
+      "grad_norm": 0.17164097726345062,
+      "learning_rate": 9.56532816958826e-05,
+      "loss": 4.2081,
+      "step": 5125
+    },
+    {
+      "epoch": 0.34855279249898086,
+      "grad_norm": 0.18012307584285736,
+      "learning_rate": 9.564903519499933e-05,
+      "loss": 3.9195,
+      "step": 5130
+    },
+    {
+      "epoch": 0.3488925125696426,
+      "grad_norm": 0.2118469625711441,
+      "learning_rate": 9.564478869411605e-05,
+      "loss": 4.1027,
+      "step": 5135
+    },
+    {
+      "epoch": 0.3492322326403044,
+      "grad_norm": 0.21049754321575165,
+      "learning_rate": 9.564054219323279e-05,
+      "loss": 4.0254,
+      "step": 5140
+    },
+    {
+      "epoch": 0.34957195271096614,
+      "grad_norm": 0.5723572969436646,
+      "learning_rate": 9.563629569234952e-05,
+      "loss": 3.8725,
+      "step": 5145
+    },
+    {
+      "epoch": 0.34991167278162794,
+      "grad_norm": 0.22618575394153595,
+      "learning_rate": 9.563204919146623e-05,
+      "loss": 4.1794,
+      "step": 5150
+    },
+    {
+      "epoch": 0.35025139285228973,
+      "grad_norm": 0.1874348223209381,
+      "learning_rate": 9.562780269058297e-05,
+      "loss": 3.9515,
+      "step": 5155
+    },
+    {
+      "epoch": 0.3505911129229515,
+      "grad_norm": 0.2910906672477722,
+      "learning_rate": 9.562355618969969e-05,
+      "loss": 4.2863,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3509308329936133,
+      "grad_norm": 0.19384640455245972,
+      "learning_rate": 9.561930968881641e-05,
+      "loss": 4.2367,
+      "step": 5165
+    },
+    {
+      "epoch": 0.351270553064275,
+      "grad_norm": 0.15815407037734985,
+      "learning_rate": 9.561506318793316e-05,
+      "loss": 3.9907,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3516102731349368,
+      "grad_norm": 0.2070821076631546,
+      "learning_rate": 9.561081668704987e-05,
+      "loss": 3.9338,
+      "step": 5175
+    },
+    {
+      "epoch": 0.3519499932055986,
+      "grad_norm": 0.24302656948566437,
+      "learning_rate": 9.56065701861666e-05,
+      "loss": 4.2737,
+      "step": 5180
+    },
+    {
+      "epoch": 0.35228971327626035,
+      "grad_norm": 0.24706007540225983,
+      "learning_rate": 9.560232368528334e-05,
+      "loss": 4.23,
+      "step": 5185
+    },
+    {
+      "epoch": 0.35262943334692215,
+      "grad_norm": 0.5972357988357544,
+      "learning_rate": 9.559807718440005e-05,
+      "loss": 4.157,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3529691534175839,
+      "grad_norm": 1.1296205520629883,
+      "learning_rate": 9.559383068351678e-05,
+      "loss": 4.1556,
+      "step": 5195
+    },
+    {
+      "epoch": 0.3533088734882457,
+      "grad_norm": 0.19614671170711517,
+      "learning_rate": 9.558958418263352e-05,
+      "loss": 4.1069,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3536485935589075,
+      "grad_norm": 0.2329636514186859,
+      "learning_rate": 9.558533768175024e-05,
+      "loss": 4.0195,
+      "step": 5205
+    },
+    {
+      "epoch": 0.35398831362956923,
+      "grad_norm": 0.3606981635093689,
+      "learning_rate": 9.558109118086697e-05,
+      "loss": 4.0518,
+      "step": 5210
+    },
+    {
+      "epoch": 0.354328033700231,
+      "grad_norm": 0.212651789188385,
+      "learning_rate": 9.557684467998371e-05,
+      "loss": 4.1561,
+      "step": 5215
+    },
+    {
+      "epoch": 0.35466775377089277,
+      "grad_norm": 0.2813778817653656,
+      "learning_rate": 9.557259817910042e-05,
+      "loss": 4.1709,
+      "step": 5220
+    },
+    {
+      "epoch": 0.35500747384155457,
+      "grad_norm": 0.1836264729499817,
+      "learning_rate": 9.556835167821715e-05,
+      "loss": 3.9968,
+      "step": 5225
+    },
+    {
+      "epoch": 0.3553471939122163,
+      "grad_norm": 0.24313072860240936,
+      "learning_rate": 9.556410517733388e-05,
+      "loss": 4.1021,
+      "step": 5230
+    },
+    {
+      "epoch": 0.3556869139828781,
+      "grad_norm": 0.21879182755947113,
+      "learning_rate": 9.55598586764506e-05,
+      "loss": 4.0678,
+      "step": 5235
+    },
+    {
+      "epoch": 0.3560266340535399,
+      "grad_norm": 0.19957928359508514,
+      "learning_rate": 9.555561217556733e-05,
+      "loss": 4.1579,
+      "step": 5240
+    },
+    {
+      "epoch": 0.35636635412420165,
+      "grad_norm": 0.2043609321117401,
+      "learning_rate": 9.555136567468406e-05,
+      "loss": 3.9197,
+      "step": 5245
+    },
+    {
+      "epoch": 0.35670607419486344,
+      "grad_norm": 0.1743493527173996,
+      "learning_rate": 9.554711917380079e-05,
+      "loss": 4.3003,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3570457942655252,
+      "grad_norm": 0.3488079309463501,
+      "learning_rate": 9.554287267291752e-05,
+      "loss": 4.0098,
+      "step": 5255
+    },
+    {
+      "epoch": 0.357385514336187,
+      "grad_norm": 0.2585020959377289,
+      "learning_rate": 9.553862617203425e-05,
+      "loss": 4.0749,
+      "step": 5260
+    },
+    {
+      "epoch": 0.3577252344068488,
+      "grad_norm": 0.22201067209243774,
+      "learning_rate": 9.553437967115097e-05,
+      "loss": 3.7807,
+      "step": 5265
+    },
+    {
+      "epoch": 0.3580649544775105,
+      "grad_norm": 0.4632178843021393,
+      "learning_rate": 9.55301331702677e-05,
+      "loss": 4.1557,
+      "step": 5270
+    },
+    {
+      "epoch": 0.3584046745481723,
+      "grad_norm": 0.4491996765136719,
+      "learning_rate": 9.552588666938443e-05,
+      "loss": 4.2286,
+      "step": 5275
+    },
+    {
+      "epoch": 0.35874439461883406,
+      "grad_norm": 0.22126582264900208,
+      "learning_rate": 9.552164016850116e-05,
+      "loss": 3.9587,
+      "step": 5280
+    },
+    {
+      "epoch": 0.35908411468949586,
+      "grad_norm": 0.20614346861839294,
+      "learning_rate": 9.551739366761789e-05,
+      "loss": 4.1402,
+      "step": 5285
+    },
+    {
+      "epoch": 0.35942383476015766,
+      "grad_norm": 0.2311069220304489,
+      "learning_rate": 9.551314716673461e-05,
+      "loss": 4.1504,
+      "step": 5290
+    },
+    {
+      "epoch": 0.3597635548308194,
+      "grad_norm": 0.20152784883975983,
+      "learning_rate": 9.550890066585134e-05,
+      "loss": 4.2783,
+      "step": 5295
+    },
+    {
+      "epoch": 0.3601032749014812,
+      "grad_norm": 0.2334737479686737,
+      "learning_rate": 9.550465416496807e-05,
+      "loss": 4.2849,
+      "step": 5300
+    },
+    {
+      "epoch": 0.36044299497214294,
+      "grad_norm": 0.17994678020477295,
+      "learning_rate": 9.55004076640848e-05,
+      "loss": 4.3848,
+      "step": 5305
+    },
+    {
+      "epoch": 0.36078271504280474,
+      "grad_norm": 0.2141488939523697,
+      "learning_rate": 9.549616116320153e-05,
+      "loss": 4.309,
+      "step": 5310
+    },
+    {
+      "epoch": 0.3611224351134665,
+      "grad_norm": 0.2028026133775711,
+      "learning_rate": 9.549191466231825e-05,
+      "loss": 4.4349,
+      "step": 5315
+    },
+    {
+      "epoch": 0.3614621551841283,
+      "grad_norm": 0.1849725842475891,
+      "learning_rate": 9.548766816143498e-05,
+      "loss": 3.929,
+      "step": 5320
+    },
+    {
+      "epoch": 0.3618018752547901,
+      "grad_norm": 0.20538243651390076,
+      "learning_rate": 9.548342166055171e-05,
+      "loss": 4.2426,
+      "step": 5325
+    },
+    {
+      "epoch": 0.3621415953254518,
+      "grad_norm": 0.22145512700080872,
+      "learning_rate": 9.547917515966844e-05,
+      "loss": 4.1321,
+      "step": 5330
+    },
+    {
+      "epoch": 0.3624813153961136,
+      "grad_norm": 0.24293570220470428,
+      "learning_rate": 9.547492865878517e-05,
+      "loss": 4.4132,
+      "step": 5335
+    },
+    {
+      "epoch": 0.36282103546677535,
+      "grad_norm": 3.3797924518585205,
+      "learning_rate": 9.54706821579019e-05,
+      "loss": 3.9775,
+      "step": 5340
+    },
+    {
+      "epoch": 0.36316075553743715,
+      "grad_norm": 0.22349077463150024,
+      "learning_rate": 9.546643565701862e-05,
+      "loss": 4.095,
+      "step": 5345
+    },
+    {
+      "epoch": 0.36350047560809895,
+      "grad_norm": 0.22708293795585632,
+      "learning_rate": 9.546218915613535e-05,
+      "loss": 4.1219,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3638401956787607,
+      "grad_norm": 0.40382349491119385,
+      "learning_rate": 9.545794265525208e-05,
+      "loss": 3.8157,
+      "step": 5355
+    },
+    {
+      "epoch": 0.3641799157494225,
+      "grad_norm": 0.1983574777841568,
+      "learning_rate": 9.54536961543688e-05,
+      "loss": 4.0346,
+      "step": 5360
+    },
+    {
+      "epoch": 0.36451963582008423,
+      "grad_norm": 0.3324495851993561,
+      "learning_rate": 9.544944965348553e-05,
+      "loss": 4.2765,
+      "step": 5365
+    },
+    {
+      "epoch": 0.36485935589074603,
+      "grad_norm": 0.2055937945842743,
+      "learning_rate": 9.544520315260226e-05,
+      "loss": 3.8416,
+      "step": 5370
+    },
+    {
+      "epoch": 0.3651990759614078,
+      "grad_norm": 0.18161867558956146,
+      "learning_rate": 9.544095665171899e-05,
+      "loss": 4.0468,
+      "step": 5375
+    },
+    {
+      "epoch": 0.36553879603206957,
+      "grad_norm": 0.2383970320224762,
+      "learning_rate": 9.543671015083572e-05,
+      "loss": 4.0359,
+      "step": 5380
+    },
+    {
+      "epoch": 0.36587851610273137,
+      "grad_norm": 0.1611696481704712,
+      "learning_rate": 9.543246364995245e-05,
+      "loss": 4.1817,
+      "step": 5385
+    },
+    {
+      "epoch": 0.3662182361733931,
+      "grad_norm": 0.3070268929004669,
+      "learning_rate": 9.542821714906917e-05,
+      "loss": 4.0901,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3665579562440549,
+      "grad_norm": 0.17862237989902496,
+      "learning_rate": 9.54239706481859e-05,
+      "loss": 3.9993,
+      "step": 5395
+    },
+    {
+      "epoch": 0.36689767631471665,
+      "grad_norm": 0.30012592673301697,
+      "learning_rate": 9.541972414730263e-05,
+      "loss": 4.0737,
+      "step": 5400
+    },
+    {
+      "epoch": 0.36723739638537845,
+      "grad_norm": 15.268974304199219,
+      "learning_rate": 9.541547764641936e-05,
+      "loss": 3.8387,
+      "step": 5405
+    },
+    {
+      "epoch": 0.36757711645604024,
+      "grad_norm": 0.19847442209720612,
+      "learning_rate": 9.541123114553609e-05,
+      "loss": 4.0158,
+      "step": 5410
+    },
+    {
+      "epoch": 0.367916836526702,
+      "grad_norm": 1.345680832862854,
+      "learning_rate": 9.540698464465281e-05,
+      "loss": 4.1187,
+      "step": 5415
+    },
+    {
+      "epoch": 0.3682565565973638,
+      "grad_norm": 0.15424399077892303,
+      "learning_rate": 9.540273814376954e-05,
+      "loss": 4.1746,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3685962766680255,
+      "grad_norm": 0.47641104459762573,
+      "learning_rate": 9.539849164288627e-05,
+      "loss": 3.8821,
+      "step": 5425
+    },
+    {
+      "epoch": 0.3689359967386873,
+      "grad_norm": 0.27253925800323486,
+      "learning_rate": 9.539424514200299e-05,
+      "loss": 4.1306,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3692757168093491,
+      "grad_norm": 0.5784019231796265,
+      "learning_rate": 9.538999864111973e-05,
+      "loss": 4.1907,
+      "step": 5435
+    },
+    {
+      "epoch": 0.36961543688001086,
+      "grad_norm": 0.21910730004310608,
+      "learning_rate": 9.538575214023645e-05,
+      "loss": 3.9055,
+      "step": 5440
+    },
+    {
+      "epoch": 0.36995515695067266,
+      "grad_norm": 0.195495143532753,
+      "learning_rate": 9.538150563935317e-05,
+      "loss": 4.1521,
+      "step": 5445
+    },
+    {
+      "epoch": 0.3702948770213344,
+      "grad_norm": 0.20794479548931122,
+      "learning_rate": 9.537725913846991e-05,
+      "loss": 3.9962,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3706345970919962,
+      "grad_norm": 1.305681586265564,
+      "learning_rate": 9.537301263758664e-05,
+      "loss": 4.173,
+      "step": 5455
+    },
+    {
+      "epoch": 0.370974317162658,
+      "grad_norm": 0.1818116158246994,
+      "learning_rate": 9.536876613670335e-05,
+      "loss": 4.0849,
+      "step": 5460
+    },
+    {
+      "epoch": 0.37131403723331974,
+      "grad_norm": 0.38611772656440735,
+      "learning_rate": 9.53645196358201e-05,
+      "loss": 3.9796,
+      "step": 5465
+    },
+    {
+      "epoch": 0.37165375730398154,
+      "grad_norm": 0.2650381922721863,
+      "learning_rate": 9.536027313493682e-05,
+      "loss": 4.1452,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3719934773746433,
+      "grad_norm": 0.2208934873342514,
+      "learning_rate": 9.535602663405354e-05,
+      "loss": 4.2661,
+      "step": 5475
+    },
+    {
+      "epoch": 0.3723331974453051,
+      "grad_norm": 0.20266486704349518,
+      "learning_rate": 9.535178013317028e-05,
+      "loss": 4.207,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3726729175159668,
+      "grad_norm": 0.21677860617637634,
+      "learning_rate": 9.5347533632287e-05,
+      "loss": 4.3784,
+      "step": 5485
+    },
+    {
+      "epoch": 0.3730126375866286,
+      "grad_norm": 0.33356210589408875,
+      "learning_rate": 9.534328713140372e-05,
+      "loss": 4.0518,
+      "step": 5490
+    },
+    {
+      "epoch": 0.3733523576572904,
+      "grad_norm": 0.2748437225818634,
+      "learning_rate": 9.533904063052046e-05,
+      "loss": 4.0669,
+      "step": 5495
+    },
+    {
+      "epoch": 0.37369207772795215,
+      "grad_norm": 0.22416415810585022,
+      "learning_rate": 9.533479412963718e-05,
+      "loss": 4.1892,
+      "step": 5500
+    },
+    {
+      "epoch": 0.37403179779861395,
+      "grad_norm": 0.20975516736507416,
+      "learning_rate": 9.53305476287539e-05,
+      "loss": 4.2146,
+      "step": 5505
+    },
+    {
+      "epoch": 0.3743715178692757,
+      "grad_norm": 0.1820031851530075,
+      "learning_rate": 9.532630112787065e-05,
+      "loss": 4.1054,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3747112379399375,
+      "grad_norm": 0.2546124756336212,
+      "learning_rate": 9.532205462698736e-05,
+      "loss": 4.1923,
+      "step": 5515
+    },
+    {
+      "epoch": 0.3750509580105993,
+      "grad_norm": 0.5577950477600098,
+      "learning_rate": 9.531780812610409e-05,
+      "loss": 3.851,
+      "step": 5520
+    },
+    {
+      "epoch": 0.37539067808126103,
+      "grad_norm": 0.2909904718399048,
+      "learning_rate": 9.531356162522083e-05,
+      "loss": 3.8513,
+      "step": 5525
+    },
+    {
+      "epoch": 0.37573039815192283,
+      "grad_norm": 0.20286774635314941,
+      "learning_rate": 9.530931512433755e-05,
+      "loss": 3.8725,
+      "step": 5530
+    },
+    {
+      "epoch": 0.37607011822258457,
+      "grad_norm": 0.20398856699466705,
+      "learning_rate": 9.530506862345427e-05,
+      "loss": 4.3243,
+      "step": 5535
+    },
+    {
+      "epoch": 0.37640983829324637,
+      "grad_norm": 0.1849180907011032,
+      "learning_rate": 9.530082212257101e-05,
+      "loss": 4.1106,
+      "step": 5540
+    },
+    {
+      "epoch": 0.37674955836390817,
+      "grad_norm": 0.1672249287366867,
+      "learning_rate": 9.529657562168773e-05,
+      "loss": 4.4955,
+      "step": 5545
+    },
+    {
+      "epoch": 0.3770892784345699,
+      "grad_norm": 0.7186090350151062,
+      "learning_rate": 9.529232912080446e-05,
+      "loss": 4.2498,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3774289985052317,
+      "grad_norm": 0.17973625659942627,
+      "learning_rate": 9.52880826199212e-05,
+      "loss": 4.2932,
+      "step": 5555
+    },
+    {
+      "epoch": 0.37776871857589345,
+      "grad_norm": 0.23119674623012543,
+      "learning_rate": 9.528383611903791e-05,
+      "loss": 4.2597,
+      "step": 5560
+    },
+    {
+      "epoch": 0.37810843864655524,
+      "grad_norm": 0.27012819051742554,
+      "learning_rate": 9.527958961815464e-05,
+      "loss": 4.0075,
+      "step": 5565
+    },
+    {
+      "epoch": 0.378448158717217,
+      "grad_norm": 0.22133472561836243,
+      "learning_rate": 9.527534311727138e-05,
+      "loss": 4.1785,
+      "step": 5570
+    },
+    {
+      "epoch": 0.3787878787878788,
+      "grad_norm": 0.17481616139411926,
+      "learning_rate": 9.52710966163881e-05,
+      "loss": 3.9544,
+      "step": 5575
+    },
+    {
+      "epoch": 0.3791275988585406,
+      "grad_norm": 0.20295321941375732,
+      "learning_rate": 9.526685011550483e-05,
+      "loss": 3.8667,
+      "step": 5580
+    },
+    {
+      "epoch": 0.3794673189292023,
+      "grad_norm": 0.3702150881290436,
+      "learning_rate": 9.526260361462155e-05,
+      "loss": 4.012,
+      "step": 5585
+    },
+    {
+      "epoch": 0.3798070389998641,
+      "grad_norm": 0.3844399154186249,
+      "learning_rate": 9.525835711373828e-05,
+      "loss": 4.0456,
+      "step": 5590
+    },
+    {
+      "epoch": 0.38014675907052586,
+      "grad_norm": 0.27248868346214294,
+      "learning_rate": 9.525411061285501e-05,
+      "loss": 4.0001,
+      "step": 5595
+    },
+    {
+      "epoch": 0.38048647914118766,
+      "grad_norm": 0.4196895360946655,
+      "learning_rate": 9.524986411197174e-05,
+      "loss": 4.2368,
+      "step": 5600
+    },
+    {
+      "epoch": 0.38082619921184946,
+      "grad_norm": 0.2525693476200104,
+      "learning_rate": 9.524561761108847e-05,
+      "loss": 4.1713,
+      "step": 5605
+    },
+    {
+      "epoch": 0.3811659192825112,
+      "grad_norm": 0.19002725183963776,
+      "learning_rate": 9.52413711102052e-05,
+      "loss": 3.9637,
+      "step": 5610
+    },
+    {
+      "epoch": 0.381505639353173,
+      "grad_norm": 0.21603484451770782,
+      "learning_rate": 9.523712460932192e-05,
+      "loss": 4.0533,
+      "step": 5615
+    },
+    {
+      "epoch": 0.38184535942383474,
+      "grad_norm": 0.1926129311323166,
+      "learning_rate": 9.523287810843865e-05,
+      "loss": 4.0691,
+      "step": 5620
+    },
+    {
+      "epoch": 0.38218507949449654,
+      "grad_norm": 0.41377758979797363,
+      "learning_rate": 9.522863160755538e-05,
+      "loss": 4.0141,
+      "step": 5625
+    },
+    {
+      "epoch": 0.38252479956515834,
+      "grad_norm": 0.2249571830034256,
+      "learning_rate": 9.52243851066721e-05,
+      "loss": 4.1834,
+      "step": 5630
+    },
+    {
+      "epoch": 0.3828645196358201,
+      "grad_norm": 0.1857426017522812,
+      "learning_rate": 9.522013860578883e-05,
+      "loss": 4.0473,
+      "step": 5635
+    },
+    {
+      "epoch": 0.3832042397064819,
+      "grad_norm": 0.20423340797424316,
+      "learning_rate": 9.521589210490556e-05,
+      "loss": 4.1039,
+      "step": 5640
+    },
+    {
+      "epoch": 0.3835439597771436,
+      "grad_norm": 0.20530074834823608,
+      "learning_rate": 9.521164560402229e-05,
+      "loss": 4.1668,
+      "step": 5645
+    },
+    {
+      "epoch": 0.3838836798478054,
+      "grad_norm": 0.21148528158664703,
+      "learning_rate": 9.520739910313902e-05,
+      "loss": 4.0141,
+      "step": 5650
+    },
+    {
+      "epoch": 0.38422339991846716,
+      "grad_norm": 0.23851540684700012,
+      "learning_rate": 9.520315260225575e-05,
+      "loss": 4.2311,
+      "step": 5655
+    },
+    {
+      "epoch": 0.38456311998912895,
+      "grad_norm": 0.17751434445381165,
+      "learning_rate": 9.519890610137247e-05,
+      "loss": 4.0698,
+      "step": 5660
+    },
+    {
+      "epoch": 0.38490284005979075,
+      "grad_norm": 0.4216800332069397,
+      "learning_rate": 9.51946596004892e-05,
+      "loss": 4.0324,
+      "step": 5665
+    },
+    {
+      "epoch": 0.3852425601304525,
+      "grad_norm": 0.19707581400871277,
+      "learning_rate": 9.519041309960593e-05,
+      "loss": 4.0026,
+      "step": 5670
+    },
+    {
+      "epoch": 0.3855822802011143,
+      "grad_norm": 0.21056298911571503,
+      "learning_rate": 9.518616659872266e-05,
+      "loss": 4.0293,
+      "step": 5675
+    },
+    {
+      "epoch": 0.38592200027177603,
+      "grad_norm": 0.18237900733947754,
+      "learning_rate": 9.518192009783939e-05,
+      "loss": 3.9467,
+      "step": 5680
+    },
+    {
+      "epoch": 0.38626172034243783,
+      "grad_norm": 0.1838427186012268,
+      "learning_rate": 9.517767359695611e-05,
+      "loss": 3.925,
+      "step": 5685
+    },
+    {
+      "epoch": 0.38660144041309963,
+      "grad_norm": 0.21086731553077698,
+      "learning_rate": 9.517342709607284e-05,
+      "loss": 4.2034,
+      "step": 5690
+    },
+    {
+      "epoch": 0.38694116048376137,
+      "grad_norm": 0.17555493116378784,
+      "learning_rate": 9.516918059518957e-05,
+      "loss": 3.8963,
+      "step": 5695
+    },
+    {
+      "epoch": 0.38728088055442317,
+      "grad_norm": 0.23491710424423218,
+      "learning_rate": 9.51649340943063e-05,
+      "loss": 4.116,
+      "step": 5700
+    },
+    {
+      "epoch": 0.3876206006250849,
+      "grad_norm": 0.18439505994319916,
+      "learning_rate": 9.516068759342303e-05,
+      "loss": 4.1069,
+      "step": 5705
+    },
+    {
+      "epoch": 0.3879603206957467,
+      "grad_norm": 0.18807227909564972,
+      "learning_rate": 9.515644109253975e-05,
+      "loss": 4.104,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3883000407664085,
+      "grad_norm": 0.4963626265525818,
+      "learning_rate": 9.515219459165648e-05,
+      "loss": 4.1994,
+      "step": 5715
+    },
+    {
+      "epoch": 0.38863976083707025,
+      "grad_norm": 0.24339251220226288,
+      "learning_rate": 9.514794809077321e-05,
+      "loss": 4.1082,
+      "step": 5720
+    },
+    {
+      "epoch": 0.38897948090773204,
+      "grad_norm": 0.17436154186725616,
+      "learning_rate": 9.514370158988994e-05,
+      "loss": 4.1256,
+      "step": 5725
+    },
+    {
+      "epoch": 0.3893192009783938,
+      "grad_norm": 0.2445308268070221,
+      "learning_rate": 9.513945508900667e-05,
+      "loss": 4.0222,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3896589210490556,
+      "grad_norm": 0.6241475939750671,
+      "learning_rate": 9.51352085881234e-05,
+      "loss": 4.1132,
+      "step": 5735
+    },
+    {
+      "epoch": 0.3899986411197173,
+      "grad_norm": 0.16763907670974731,
+      "learning_rate": 9.513096208724012e-05,
+      "loss": 4.3379,
+      "step": 5740
+    },
+    {
+      "epoch": 0.3903383611903791,
+      "grad_norm": 0.1730974316596985,
+      "learning_rate": 9.512671558635685e-05,
+      "loss": 4.1179,
+      "step": 5745
+    },
+    {
+      "epoch": 0.3906780812610409,
+      "grad_norm": 0.19016407430171967,
+      "learning_rate": 9.512246908547358e-05,
+      "loss": 4.0632,
+      "step": 5750
+    },
+    {
+      "epoch": 0.39101780133170266,
+      "grad_norm": 0.19713403284549713,
+      "learning_rate": 9.51182225845903e-05,
+      "loss": 4.088,
+      "step": 5755
+    },
+    {
+      "epoch": 0.39135752140236446,
+      "grad_norm": 0.26910024881362915,
+      "learning_rate": 9.511397608370703e-05,
+      "loss": 4.2572,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3916972414730262,
+      "grad_norm": 0.20750823616981506,
+      "learning_rate": 9.510972958282376e-05,
+      "loss": 4.0017,
+      "step": 5765
+    },
+    {
+      "epoch": 0.392036961543688,
+      "grad_norm": 0.18255822360515594,
+      "learning_rate": 9.510548308194049e-05,
+      "loss": 4.4284,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3923766816143498,
+      "grad_norm": 0.19282065331935883,
+      "learning_rate": 9.510123658105722e-05,
+      "loss": 4.1651,
+      "step": 5775
+    },
+    {
+      "epoch": 0.39271640168501154,
+      "grad_norm": 0.22694478929042816,
+      "learning_rate": 9.509699008017395e-05,
+      "loss": 4.0152,
+      "step": 5780
+    },
+    {
+      "epoch": 0.39305612175567334,
+      "grad_norm": 0.2607446014881134,
+      "learning_rate": 9.509274357929066e-05,
+      "loss": 4.3358,
+      "step": 5785
+    },
+    {
+      "epoch": 0.3933958418263351,
+      "grad_norm": 0.22173616290092468,
+      "learning_rate": 9.50884970784074e-05,
+      "loss": 4.2362,
+      "step": 5790
+    },
+    {
+      "epoch": 0.3937355618969969,
+      "grad_norm": 0.20545057952404022,
+      "learning_rate": 9.508425057752413e-05,
+      "loss": 4.0821,
+      "step": 5795
+    },
+    {
+      "epoch": 0.3940752819676587,
+      "grad_norm": 0.23421698808670044,
+      "learning_rate": 9.508000407664084e-05,
+      "loss": 4.2178,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3944150020383204,
+      "grad_norm": 0.2095632702112198,
+      "learning_rate": 9.507575757575759e-05,
+      "loss": 4.2833,
+      "step": 5805
+    },
+    {
+      "epoch": 0.3947547221089822,
+      "grad_norm": 0.23404939472675323,
+      "learning_rate": 9.507151107487431e-05,
+      "loss": 4.0823,
+      "step": 5810
+    },
+    {
+      "epoch": 0.39509444217964396,
+      "grad_norm": 0.23966114223003387,
+      "learning_rate": 9.506726457399103e-05,
+      "loss": 3.8059,
+      "step": 5815
+    },
+    {
+      "epoch": 0.39543416225030575,
+      "grad_norm": 0.2027054876089096,
+      "learning_rate": 9.506301807310777e-05,
+      "loss": 4.2229,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3957738823209675,
+      "grad_norm": 0.18689711391925812,
+      "learning_rate": 9.50587715722245e-05,
+      "loss": 4.26,
+      "step": 5825
+    },
+    {
+      "epoch": 0.3961136023916293,
+      "grad_norm": 0.263927698135376,
+      "learning_rate": 9.505452507134121e-05,
+      "loss": 4.021,
+      "step": 5830
+    },
+    {
+      "epoch": 0.3964533224622911,
+      "grad_norm": 0.18399837613105774,
+      "learning_rate": 9.505027857045795e-05,
+      "loss": 4.218,
+      "step": 5835
+    },
+    {
+      "epoch": 0.39679304253295283,
+      "grad_norm": 0.17031966149806976,
+      "learning_rate": 9.504603206957468e-05,
+      "loss": 3.9079,
+      "step": 5840
+    },
+    {
+      "epoch": 0.39713276260361463,
+      "grad_norm": 0.3210891783237457,
+      "learning_rate": 9.50417855686914e-05,
+      "loss": 3.9466,
+      "step": 5845
+    },
+    {
+      "epoch": 0.3974724826742764,
+      "grad_norm": 0.1981404423713684,
+      "learning_rate": 9.503753906780814e-05,
+      "loss": 4.001,
+      "step": 5850
+    },
+    {
+      "epoch": 0.39781220274493817,
+      "grad_norm": 0.3136885464191437,
+      "learning_rate": 9.503329256692485e-05,
+      "loss": 4.058,
+      "step": 5855
+    },
+    {
+      "epoch": 0.39815192281559997,
+      "grad_norm": 2.190765857696533,
+      "learning_rate": 9.502904606604158e-05,
+      "loss": 4.0696,
+      "step": 5860
+    },
+    {
+      "epoch": 0.3984916428862617,
+      "grad_norm": 0.17315542697906494,
+      "learning_rate": 9.502479956515832e-05,
+      "loss": 4.2626,
+      "step": 5865
+    },
+    {
+      "epoch": 0.3988313629569235,
+      "grad_norm": 0.33235201239585876,
+      "learning_rate": 9.502055306427504e-05,
+      "loss": 3.9844,
+      "step": 5870
+    },
+    {
+      "epoch": 0.39917108302758525,
+      "grad_norm": 0.23391257226467133,
+      "learning_rate": 9.501630656339176e-05,
+      "loss": 3.8401,
+      "step": 5875
+    },
+    {
+      "epoch": 0.39951080309824705,
+      "grad_norm": 0.24006325006484985,
+      "learning_rate": 9.50120600625085e-05,
+      "loss": 4.0724,
+      "step": 5880
+    },
+    {
+      "epoch": 0.39985052316890884,
+      "grad_norm": 0.17999830842018127,
+      "learning_rate": 9.500781356162522e-05,
+      "loss": 4.2167,
+      "step": 5885
+    },
+    {
+      "epoch": 0.4001902432395706,
+      "grad_norm": 0.18070223927497864,
+      "learning_rate": 9.500356706074195e-05,
+      "loss": 4.105,
+      "step": 5890
+    },
+    {
+      "epoch": 0.4005299633102324,
+      "grad_norm": 0.19634656608104706,
+      "learning_rate": 9.499932055985869e-05,
+      "loss": 3.989,
+      "step": 5895
+    },
+    {
+      "epoch": 0.4008696833808941,
+      "grad_norm": 0.23722241818904877,
+      "learning_rate": 9.49950740589754e-05,
+      "loss": 4.1728,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4012094034515559,
+      "grad_norm": 0.19146768748760223,
+      "learning_rate": 9.499082755809213e-05,
+      "loss": 4.1732,
+      "step": 5905
+    },
+    {
+      "epoch": 0.40154912352221767,
+      "grad_norm": 0.21835100650787354,
+      "learning_rate": 9.498658105720887e-05,
+      "loss": 4.0836,
+      "step": 5910
+    },
+    {
+      "epoch": 0.40188884359287946,
+      "grad_norm": 0.17060807347297668,
+      "learning_rate": 9.498233455632559e-05,
+      "loss": 4.0794,
+      "step": 5915
+    },
+    {
+      "epoch": 0.40222856366354126,
+      "grad_norm": 0.9648451805114746,
+      "learning_rate": 9.497808805544232e-05,
+      "loss": 4.1605,
+      "step": 5920
+    },
+    {
+      "epoch": 0.402568283734203,
+      "grad_norm": 0.1983519345521927,
+      "learning_rate": 9.497384155455904e-05,
+      "loss": 3.9843,
+      "step": 5925
+    },
+    {
+      "epoch": 0.4029080038048648,
+      "grad_norm": 0.23070013523101807,
+      "learning_rate": 9.496959505367577e-05,
+      "loss": 3.9679,
+      "step": 5930
+    },
+    {
+      "epoch": 0.40324772387552654,
+      "grad_norm": 1.1979519128799438,
+      "learning_rate": 9.49653485527925e-05,
+      "loss": 4.1856,
+      "step": 5935
+    },
+    {
+      "epoch": 0.40358744394618834,
+      "grad_norm": 0.7521958947181702,
+      "learning_rate": 9.496110205190923e-05,
+      "loss": 4.082,
+      "step": 5940
+    },
+    {
+      "epoch": 0.40392716401685014,
+      "grad_norm": 0.192143052816391,
+      "learning_rate": 9.495685555102596e-05,
+      "loss": 4.2121,
+      "step": 5945
+    },
+    {
+      "epoch": 0.4042668840875119,
+      "grad_norm": 0.2611311972141266,
+      "learning_rate": 9.495260905014268e-05,
+      "loss": 4.2154,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4046066041581737,
+      "grad_norm": 0.18073415756225586,
+      "learning_rate": 9.494836254925941e-05,
+      "loss": 4.1762,
+      "step": 5955
+    },
+    {
+      "epoch": 0.4049463242288354,
+      "grad_norm": 0.1921936720609665,
+      "learning_rate": 9.494411604837614e-05,
+      "loss": 4.0705,
+      "step": 5960
+    },
+    {
+      "epoch": 0.4052860442994972,
+      "grad_norm": 0.16377374529838562,
+      "learning_rate": 9.493986954749287e-05,
+      "loss": 4.1785,
+      "step": 5965
+    },
+    {
+      "epoch": 0.405625764370159,
+      "grad_norm": 0.21104343235492706,
+      "learning_rate": 9.49356230466096e-05,
+      "loss": 4.2715,
+      "step": 5970
+    },
+    {
+      "epoch": 0.40596548444082076,
+      "grad_norm": 0.2071741223335266,
+      "learning_rate": 9.493137654572632e-05,
+      "loss": 3.9067,
+      "step": 5975
+    },
+    {
+      "epoch": 0.40630520451148255,
+      "grad_norm": 0.22247660160064697,
+      "learning_rate": 9.492713004484305e-05,
+      "loss": 4.0561,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4066449245821443,
+      "grad_norm": 0.20433616638183594,
+      "learning_rate": 9.492288354395978e-05,
+      "loss": 4.0168,
+      "step": 5985
+    },
+    {
+      "epoch": 0.4069846446528061,
+      "grad_norm": 0.2049606889486313,
+      "learning_rate": 9.491863704307651e-05,
+      "loss": 4.3927,
+      "step": 5990
+    },
+    {
+      "epoch": 0.40732436472346784,
+      "grad_norm": 0.22720476984977722,
+      "learning_rate": 9.491439054219324e-05,
+      "loss": 4.2648,
+      "step": 5995
+    },
+    {
+      "epoch": 0.40766408479412963,
+      "grad_norm": 0.25233665108680725,
+      "learning_rate": 9.491014404130996e-05,
+      "loss": 3.9343,
+      "step": 6000
+    },
+    {
+      "epoch": 0.40800380486479143,
+      "grad_norm": 0.21542391180992126,
+      "learning_rate": 9.490589754042669e-05,
+      "loss": 4.2302,
+      "step": 6005
+    },
+    {
+      "epoch": 0.4083435249354532,
+      "grad_norm": 0.8740308284759521,
+      "learning_rate": 9.490165103954342e-05,
+      "loss": 4.0408,
+      "step": 6010
+    },
+    {
+      "epoch": 0.40868324500611497,
+      "grad_norm": 0.18519490957260132,
+      "learning_rate": 9.489740453866015e-05,
+      "loss": 4.0773,
+      "step": 6015
+    },
+    {
+      "epoch": 0.4090229650767767,
+      "grad_norm": 0.2651638388633728,
+      "learning_rate": 9.489315803777688e-05,
+      "loss": 4.1498,
+      "step": 6020
+    },
+    {
+      "epoch": 0.4093626851474385,
+      "grad_norm": 0.4929540157318115,
+      "learning_rate": 9.48889115368936e-05,
+      "loss": 3.9128,
+      "step": 6025
+    },
+    {
+      "epoch": 0.4097024052181003,
+      "grad_norm": 0.20049385726451874,
+      "learning_rate": 9.488466503601033e-05,
+      "loss": 4.0097,
+      "step": 6030
+    },
+    {
+      "epoch": 0.41004212528876205,
+      "grad_norm": 0.17493902146816254,
+      "learning_rate": 9.488041853512706e-05,
+      "loss": 4.1654,
+      "step": 6035
+    },
+    {
+      "epoch": 0.41038184535942385,
+      "grad_norm": 0.30577751994132996,
+      "learning_rate": 9.487617203424379e-05,
+      "loss": 4.0563,
+      "step": 6040
+    },
+    {
+      "epoch": 0.4107215654300856,
+      "grad_norm": 0.2669510543346405,
+      "learning_rate": 9.487192553336052e-05,
+      "loss": 4.0721,
+      "step": 6045
+    },
+    {
+      "epoch": 0.4110612855007474,
+      "grad_norm": 0.18722812831401825,
+      "learning_rate": 9.486767903247724e-05,
+      "loss": 4.1749,
+      "step": 6050
+    },
+    {
+      "epoch": 0.4114010055714092,
+      "grad_norm": 0.1755664199590683,
+      "learning_rate": 9.486343253159397e-05,
+      "loss": 4.1804,
+      "step": 6055
+    },
+    {
+      "epoch": 0.4117407256420709,
+      "grad_norm": 0.27995565533638,
+      "learning_rate": 9.48591860307107e-05,
+      "loss": 3.9794,
+      "step": 6060
+    },
+    {
+      "epoch": 0.4120804457127327,
+      "grad_norm": 0.2518627345561981,
+      "learning_rate": 9.485493952982743e-05,
+      "loss": 4.2129,
+      "step": 6065
+    },
+    {
+      "epoch": 0.41242016578339447,
+      "grad_norm": 0.21856893599033356,
+      "learning_rate": 9.485069302894416e-05,
+      "loss": 4.3444,
+      "step": 6070
+    },
+    {
+      "epoch": 0.41275988585405626,
+      "grad_norm": 1.2819626331329346,
+      "learning_rate": 9.484644652806088e-05,
+      "loss": 3.6595,
+      "step": 6075
+    },
+    {
+      "epoch": 0.413099605924718,
+      "grad_norm": 0.25162968039512634,
+      "learning_rate": 9.484220002717761e-05,
+      "loss": 4.2169,
+      "step": 6080
+    },
+    {
+      "epoch": 0.4134393259953798,
+      "grad_norm": 0.2659519910812378,
+      "learning_rate": 9.483795352629434e-05,
+      "loss": 4.0789,
+      "step": 6085
+    },
+    {
+      "epoch": 0.4137790460660416,
+      "grad_norm": 0.20415237545967102,
+      "learning_rate": 9.483370702541107e-05,
+      "loss": 4.2934,
+      "step": 6090
+    },
+    {
+      "epoch": 0.41411876613670334,
+      "grad_norm": 0.33283525705337524,
+      "learning_rate": 9.48294605245278e-05,
+      "loss": 4.0536,
+      "step": 6095
+    },
+    {
+      "epoch": 0.41445848620736514,
+      "grad_norm": 0.21782909333705902,
+      "learning_rate": 9.482521402364452e-05,
+      "loss": 3.9827,
+      "step": 6100
+    },
+    {
+      "epoch": 0.4147982062780269,
+      "grad_norm": 4.150507926940918,
+      "learning_rate": 9.482096752276125e-05,
+      "loss": 3.9654,
+      "step": 6105
+    },
+    {
+      "epoch": 0.4151379263486887,
+      "grad_norm": 0.20316611230373383,
+      "learning_rate": 9.481672102187798e-05,
+      "loss": 4.045,
+      "step": 6110
+    },
+    {
+      "epoch": 0.4154776464193505,
+      "grad_norm": 1.9692164659500122,
+      "learning_rate": 9.481247452099471e-05,
+      "loss": 4.0912,
+      "step": 6115
+    },
+    {
+      "epoch": 0.4158173664900122,
+      "grad_norm": 0.21312934160232544,
+      "learning_rate": 9.480822802011144e-05,
+      "loss": 4.2765,
+      "step": 6120
+    },
+    {
+      "epoch": 0.416157086560674,
+      "grad_norm": 0.1886243224143982,
+      "learning_rate": 9.480398151922815e-05,
+      "loss": 4.0305,
+      "step": 6125
+    },
+    {
+      "epoch": 0.41649680663133576,
+      "grad_norm": 0.22211480140686035,
+      "learning_rate": 9.479973501834489e-05,
+      "loss": 3.906,
+      "step": 6130
+    },
+    {
+      "epoch": 0.41683652670199756,
+      "grad_norm": 0.24448561668395996,
+      "learning_rate": 9.479548851746162e-05,
+      "loss": 3.9652,
+      "step": 6135
+    },
+    {
+      "epoch": 0.41717624677265935,
+      "grad_norm": 0.2330089956521988,
+      "learning_rate": 9.479124201657834e-05,
+      "loss": 4.0935,
+      "step": 6140
+    },
+    {
+      "epoch": 0.4175159668433211,
+      "grad_norm": 0.354692280292511,
+      "learning_rate": 9.478699551569508e-05,
+      "loss": 4.1893,
+      "step": 6145
+    },
+    {
+      "epoch": 0.4178556869139829,
+      "grad_norm": 0.31616339087486267,
+      "learning_rate": 9.47827490148118e-05,
+      "loss": 4.1393,
+      "step": 6150
+    },
+    {
+      "epoch": 0.41819540698464464,
+      "grad_norm": 0.6363674402236938,
+      "learning_rate": 9.477850251392852e-05,
+      "loss": 4.1578,
+      "step": 6155
+    },
+    {
+      "epoch": 0.41853512705530643,
+      "grad_norm": 0.19385862350463867,
+      "learning_rate": 9.477425601304526e-05,
+      "loss": 4.0994,
+      "step": 6160
+    },
+    {
+      "epoch": 0.4188748471259682,
+      "grad_norm": 0.20381571352481842,
+      "learning_rate": 9.477000951216199e-05,
+      "loss": 4.1873,
+      "step": 6165
+    },
+    {
+      "epoch": 0.41921456719663,
+      "grad_norm": 0.20795594155788422,
+      "learning_rate": 9.47657630112787e-05,
+      "loss": 4.0544,
+      "step": 6170
+    },
+    {
+      "epoch": 0.41955428726729177,
+      "grad_norm": 0.3839801847934723,
+      "learning_rate": 9.476151651039544e-05,
+      "loss": 4.1777,
+      "step": 6175
+    },
+    {
+      "epoch": 0.4198940073379535,
+      "grad_norm": 0.2491442710161209,
+      "learning_rate": 9.475727000951217e-05,
+      "loss": 3.9298,
+      "step": 6180
+    },
+    {
+      "epoch": 0.4202337274086153,
+      "grad_norm": 0.1739528328180313,
+      "learning_rate": 9.475302350862889e-05,
+      "loss": 4.0482,
+      "step": 6185
+    },
+    {
+      "epoch": 0.42057344747927705,
+      "grad_norm": 0.19341996312141418,
+      "learning_rate": 9.474877700774563e-05,
+      "loss": 3.911,
+      "step": 6190
+    },
+    {
+      "epoch": 0.42091316754993885,
+      "grad_norm": 0.16241292655467987,
+      "learning_rate": 9.474453050686236e-05,
+      "loss": 4.0751,
+      "step": 6195
+    },
+    {
+      "epoch": 0.42125288762060065,
+      "grad_norm": 0.16985565423965454,
+      "learning_rate": 9.474028400597907e-05,
+      "loss": 3.9359,
+      "step": 6200
+    },
+    {
+      "epoch": 0.4215926076912624,
+      "grad_norm": 0.21724484860897064,
+      "learning_rate": 9.473603750509581e-05,
+      "loss": 4.0409,
+      "step": 6205
+    },
+    {
+      "epoch": 0.4219323277619242,
+      "grad_norm": 0.21480692923069,
+      "learning_rate": 9.473179100421253e-05,
+      "loss": 3.987,
+      "step": 6210
+    },
+    {
+      "epoch": 0.42227204783258593,
+      "grad_norm": 0.2604687809944153,
+      "learning_rate": 9.472754450332926e-05,
+      "loss": 3.89,
+      "step": 6215
+    },
+    {
+      "epoch": 0.4226117679032477,
+      "grad_norm": 0.22292381525039673,
+      "learning_rate": 9.4723298002446e-05,
+      "loss": 4.029,
+      "step": 6220
+    },
+    {
+      "epoch": 0.4229514879739095,
+      "grad_norm": 0.2695325016975403,
+      "learning_rate": 9.471905150156271e-05,
+      "loss": 4.0199,
+      "step": 6225
+    },
+    {
+      "epoch": 0.42329120804457127,
+      "grad_norm": 0.17921492457389832,
+      "learning_rate": 9.471480500067944e-05,
+      "loss": 4.0195,
+      "step": 6230
+    },
+    {
+      "epoch": 0.42363092811523306,
+      "grad_norm": 0.29654955863952637,
+      "learning_rate": 9.471055849979618e-05,
+      "loss": 4.2356,
+      "step": 6235
+    },
+    {
+      "epoch": 0.4239706481858948,
+      "grad_norm": 0.3091282844543457,
+      "learning_rate": 9.47063119989129e-05,
+      "loss": 3.8808,
+      "step": 6240
+    },
+    {
+      "epoch": 0.4243103682565566,
+      "grad_norm": 0.20580576360225677,
+      "learning_rate": 9.470206549802962e-05,
+      "loss": 4.0886,
+      "step": 6245
+    },
+    {
+      "epoch": 0.42465008832721834,
+      "grad_norm": 0.19273176789283752,
+      "learning_rate": 9.469781899714636e-05,
+      "loss": 3.9993,
+      "step": 6250
+    },
+    {
+      "epoch": 0.42498980839788014,
+      "grad_norm": 0.18639002740383148,
+      "learning_rate": 9.469357249626308e-05,
+      "loss": 4.0441,
+      "step": 6255
+    },
+    {
+      "epoch": 0.42532952846854194,
+      "grad_norm": 2.978999614715576,
+      "learning_rate": 9.468932599537981e-05,
+      "loss": 3.7926,
+      "step": 6260
+    },
+    {
+      "epoch": 0.4256692485392037,
+      "grad_norm": 0.17055857181549072,
+      "learning_rate": 9.468507949449655e-05,
+      "loss": 4.0499,
+      "step": 6265
+    },
+    {
+      "epoch": 0.4260089686098655,
+      "grad_norm": 1.0542086362838745,
+      "learning_rate": 9.468083299361326e-05,
+      "loss": 4.2183,
+      "step": 6270
+    },
+    {
+      "epoch": 0.4263486886805272,
+      "grad_norm": 1.1902449131011963,
+      "learning_rate": 9.467658649272999e-05,
+      "loss": 3.9083,
+      "step": 6275
+    },
+    {
+      "epoch": 0.426688408751189,
+      "grad_norm": 0.26639068126678467,
+      "learning_rate": 9.467233999184672e-05,
+      "loss": 3.9113,
+      "step": 6280
+    },
+    {
+      "epoch": 0.4270281288218508,
+      "grad_norm": 0.2732694149017334,
+      "learning_rate": 9.466809349096345e-05,
+      "loss": 4.1468,
+      "step": 6285
+    },
+    {
+      "epoch": 0.42736784889251256,
+      "grad_norm": 0.27546462416648865,
+      "learning_rate": 9.466384699008018e-05,
+      "loss": 3.9733,
+      "step": 6290
+    },
+    {
+      "epoch": 0.42770756896317436,
+      "grad_norm": 0.1994089037179947,
+      "learning_rate": 9.46596004891969e-05,
+      "loss": 4.1123,
+      "step": 6295
+    },
+    {
+      "epoch": 0.4280472890338361,
+      "grad_norm": 0.20160436630249023,
+      "learning_rate": 9.465535398831363e-05,
+      "loss": 4.0467,
+      "step": 6300
+    },
+    {
+      "epoch": 0.4283870091044979,
+      "grad_norm": 0.2255067527294159,
+      "learning_rate": 9.465110748743036e-05,
+      "loss": 3.7712,
+      "step": 6305
+    },
+    {
+      "epoch": 0.4287267291751597,
+      "grad_norm": 0.1755346655845642,
+      "learning_rate": 9.464686098654709e-05,
+      "loss": 4.3119,
+      "step": 6310
+    },
+    {
+      "epoch": 0.42906644924582144,
+      "grad_norm": 0.16779784858226776,
+      "learning_rate": 9.464261448566382e-05,
+      "loss": 4.1447,
+      "step": 6315
+    },
+    {
+      "epoch": 0.42940616931648323,
+      "grad_norm": 0.2779237926006317,
+      "learning_rate": 9.463836798478054e-05,
+      "loss": 4.1285,
+      "step": 6320
+    },
+    {
+      "epoch": 0.429745889387145,
+      "grad_norm": 0.21850286424160004,
+      "learning_rate": 9.463412148389727e-05,
+      "loss": 4.118,
+      "step": 6325
+    },
+    {
+      "epoch": 0.4300856094578068,
+      "grad_norm": 0.2127694934606552,
+      "learning_rate": 9.4629874983014e-05,
+      "loss": 4.1073,
+      "step": 6330
+    },
+    {
+      "epoch": 0.4304253295284685,
+      "grad_norm": 0.24460366368293762,
+      "learning_rate": 9.462562848213073e-05,
+      "loss": 4.0523,
+      "step": 6335
+    },
+    {
+      "epoch": 0.4307650495991303,
+      "grad_norm": 0.18039904534816742,
+      "learning_rate": 9.462138198124746e-05,
+      "loss": 4.205,
+      "step": 6340
+    },
+    {
+      "epoch": 0.4311047696697921,
+      "grad_norm": 0.23940406739711761,
+      "learning_rate": 9.461713548036418e-05,
+      "loss": 4.0222,
+      "step": 6345
+    },
+    {
+      "epoch": 0.43144448974045385,
+      "grad_norm": 0.223390132188797,
+      "learning_rate": 9.461288897948091e-05,
+      "loss": 4.0133,
+      "step": 6350
+    },
+    {
+      "epoch": 0.43178420981111565,
+      "grad_norm": 0.20645423233509064,
+      "learning_rate": 9.460864247859764e-05,
+      "loss": 3.9967,
+      "step": 6355
+    },
+    {
+      "epoch": 0.4321239298817774,
+      "grad_norm": 0.7943032383918762,
+      "learning_rate": 9.460439597771437e-05,
+      "loss": 3.927,
+      "step": 6360
+    },
+    {
+      "epoch": 0.4324636499524392,
+      "grad_norm": 0.1684955656528473,
+      "learning_rate": 9.46001494768311e-05,
+      "loss": 4.0099,
+      "step": 6365
+    },
+    {
+      "epoch": 0.432803370023101,
+      "grad_norm": 0.21439214050769806,
+      "learning_rate": 9.459590297594782e-05,
+      "loss": 3.7963,
+      "step": 6370
+    },
+    {
+      "epoch": 0.43314309009376273,
+      "grad_norm": 0.21478402614593506,
+      "learning_rate": 9.459165647506455e-05,
+      "loss": 4.0739,
+      "step": 6375
+    },
+    {
+      "epoch": 0.4334828101644245,
+      "grad_norm": 0.8936269283294678,
+      "learning_rate": 9.458740997418128e-05,
+      "loss": 4.1838,
+      "step": 6380
+    },
+    {
+      "epoch": 0.43382253023508627,
+      "grad_norm": 0.5096725821495056,
+      "learning_rate": 9.458316347329801e-05,
+      "loss": 4.047,
+      "step": 6385
+    },
+    {
+      "epoch": 0.43416225030574807,
+      "grad_norm": 0.285972535610199,
+      "learning_rate": 9.457891697241474e-05,
+      "loss": 4.1083,
+      "step": 6390
+    },
+    {
+      "epoch": 0.43450197037640986,
+      "grad_norm": 0.1650124043226242,
+      "learning_rate": 9.457467047153146e-05,
+      "loss": 4.157,
+      "step": 6395
+    },
+    {
+      "epoch": 0.4348416904470716,
+      "grad_norm": 0.19191978871822357,
+      "learning_rate": 9.457042397064819e-05,
+      "loss": 3.9201,
+      "step": 6400
+    },
+    {
+      "epoch": 0.4351814105177334,
+      "grad_norm": 0.20855942368507385,
+      "learning_rate": 9.456617746976492e-05,
+      "loss": 4.1146,
+      "step": 6405
+    },
+    {
+      "epoch": 0.43552113058839514,
+      "grad_norm": 0.17791011929512024,
+      "learning_rate": 9.456193096888165e-05,
+      "loss": 3.9971,
+      "step": 6410
+    },
+    {
+      "epoch": 0.43586085065905694,
+      "grad_norm": 0.2120276242494583,
+      "learning_rate": 9.455768446799838e-05,
+      "loss": 4.0967,
+      "step": 6415
+    },
+    {
+      "epoch": 0.4362005707297187,
+      "grad_norm": 0.20230713486671448,
+      "learning_rate": 9.45534379671151e-05,
+      "loss": 3.902,
+      "step": 6420
+    },
+    {
+      "epoch": 0.4365402908003805,
+      "grad_norm": 0.5752553343772888,
+      "learning_rate": 9.454919146623183e-05,
+      "loss": 3.9801,
+      "step": 6425
+    },
+    {
+      "epoch": 0.4368800108710423,
+      "grad_norm": 0.19792981445789337,
+      "learning_rate": 9.454494496534856e-05,
+      "loss": 4.2249,
+      "step": 6430
+    },
+    {
+      "epoch": 0.437219730941704,
+      "grad_norm": 0.20685099065303802,
+      "learning_rate": 9.454069846446529e-05,
+      "loss": 4.2953,
+      "step": 6435
+    },
+    {
+      "epoch": 0.4375594510123658,
+      "grad_norm": 0.22175006568431854,
+      "learning_rate": 9.453645196358202e-05,
+      "loss": 3.9621,
+      "step": 6440
+    },
+    {
+      "epoch": 0.43789917108302756,
+      "grad_norm": 0.4952181875705719,
+      "learning_rate": 9.453220546269874e-05,
+      "loss": 3.986,
+      "step": 6445
+    },
+    {
+      "epoch": 0.43823889115368936,
+      "grad_norm": 0.20618560910224915,
+      "learning_rate": 9.452795896181547e-05,
+      "loss": 3.9802,
+      "step": 6450
+    },
+    {
+      "epoch": 0.43857861122435116,
+      "grad_norm": 1.136326551437378,
+      "learning_rate": 9.45237124609322e-05,
+      "loss": 4.2695,
+      "step": 6455
+    },
+    {
+      "epoch": 0.4389183312950129,
+      "grad_norm": 0.22814400494098663,
+      "learning_rate": 9.451946596004893e-05,
+      "loss": 3.8791,
+      "step": 6460
+    },
+    {
+      "epoch": 0.4392580513656747,
+      "grad_norm": 0.24193866550922394,
+      "learning_rate": 9.451521945916566e-05,
+      "loss": 3.9838,
+      "step": 6465
+    },
+    {
+      "epoch": 0.43959777143633644,
+      "grad_norm": 0.25064903497695923,
+      "learning_rate": 9.451097295828238e-05,
+      "loss": 3.9355,
+      "step": 6470
+    },
+    {
+      "epoch": 0.43993749150699824,
+      "grad_norm": 1.2106342315673828,
+      "learning_rate": 9.450672645739911e-05,
+      "loss": 4.1844,
+      "step": 6475
+    },
+    {
+      "epoch": 0.44027721157766003,
+      "grad_norm": 0.16492827236652374,
+      "learning_rate": 9.450247995651583e-05,
+      "loss": 4.2447,
+      "step": 6480
+    },
+    {
+      "epoch": 0.4406169316483218,
+      "grad_norm": 0.1795361191034317,
+      "learning_rate": 9.449823345563257e-05,
+      "loss": 3.8589,
+      "step": 6485
+    },
+    {
+      "epoch": 0.44095665171898357,
+      "grad_norm": 0.19358831644058228,
+      "learning_rate": 9.44939869547493e-05,
+      "loss": 4.2963,
+      "step": 6490
+    },
+    {
+      "epoch": 0.4412963717896453,
+      "grad_norm": 0.2732826769351959,
+      "learning_rate": 9.448974045386601e-05,
+      "loss": 3.9789,
+      "step": 6495
+    },
+    {
+      "epoch": 0.4416360918603071,
+      "grad_norm": 0.23638220131397247,
+      "learning_rate": 9.448549395298275e-05,
+      "loss": 4.2448,
+      "step": 6500
+    },
+    {
+      "epoch": 0.4419758119309689,
+      "grad_norm": 0.2072085738182068,
+      "learning_rate": 9.448124745209948e-05,
+      "loss": 3.8356,
+      "step": 6505
+    },
+    {
+      "epoch": 0.44231553200163065,
+      "grad_norm": 3.1101341247558594,
+      "learning_rate": 9.44770009512162e-05,
+      "loss": 4.2411,
+      "step": 6510
+    },
+    {
+      "epoch": 0.44265525207229245,
+      "grad_norm": 0.4264751374721527,
+      "learning_rate": 9.447275445033294e-05,
+      "loss": 3.9676,
+      "step": 6515
+    },
+    {
+      "epoch": 0.4429949721429542,
+      "grad_norm": 0.20776435732841492,
+      "learning_rate": 9.446850794944966e-05,
+      "loss": 3.8493,
+      "step": 6520
+    },
+    {
+      "epoch": 0.443334692213616,
+      "grad_norm": 0.3044533133506775,
+      "learning_rate": 9.446426144856638e-05,
+      "loss": 4.1147,
+      "step": 6525
+    },
+    {
+      "epoch": 0.44367441228427773,
+      "grad_norm": 0.16665169596672058,
+      "learning_rate": 9.446001494768312e-05,
+      "loss": 3.9521,
+      "step": 6530
+    },
+    {
+      "epoch": 0.44401413235493953,
+      "grad_norm": 0.2023710161447525,
+      "learning_rate": 9.445576844679985e-05,
+      "loss": 3.9206,
+      "step": 6535
+    },
+    {
+      "epoch": 0.4443538524256013,
+      "grad_norm": 0.4145415425300598,
+      "learning_rate": 9.445152194591656e-05,
+      "loss": 4.0138,
+      "step": 6540
+    },
+    {
+      "epoch": 0.44469357249626307,
+      "grad_norm": 0.16682837903499603,
+      "learning_rate": 9.44472754450333e-05,
+      "loss": 4.0957,
+      "step": 6545
+    },
+    {
+      "epoch": 0.44503329256692487,
+      "grad_norm": 0.2003334015607834,
+      "learning_rate": 9.444302894415002e-05,
+      "loss": 3.8427,
+      "step": 6550
+    },
+    {
+      "epoch": 0.4453730126375866,
+      "grad_norm": 0.29585328698158264,
+      "learning_rate": 9.443878244326675e-05,
+      "loss": 4.2463,
+      "step": 6555
+    },
+    {
+      "epoch": 0.4457127327082484,
+      "grad_norm": 0.20154406130313873,
+      "learning_rate": 9.443453594238349e-05,
+      "loss": 4.2644,
+      "step": 6560
+    },
+    {
+      "epoch": 0.4460524527789102,
+      "grad_norm": 0.23148468136787415,
+      "learning_rate": 9.44302894415002e-05,
+      "loss": 3.9205,
+      "step": 6565
+    },
+    {
+      "epoch": 0.44639217284957194,
+      "grad_norm": 0.1762179434299469,
+      "learning_rate": 9.442604294061693e-05,
+      "loss": 4.0293,
+      "step": 6570
+    },
+    {
+      "epoch": 0.44673189292023374,
+      "grad_norm": 0.4714028835296631,
+      "learning_rate": 9.442179643973367e-05,
+      "loss": 4.2011,
+      "step": 6575
+    },
+    {
+      "epoch": 0.4470716129908955,
+      "grad_norm": 0.368407666683197,
+      "learning_rate": 9.441754993885039e-05,
+      "loss": 4.0047,
+      "step": 6580
+    },
+    {
+      "epoch": 0.4474113330615573,
+      "grad_norm": 0.28887784481048584,
+      "learning_rate": 9.441330343796711e-05,
+      "loss": 4.0332,
+      "step": 6585
+    },
+    {
+      "epoch": 0.4477510531322191,
+      "grad_norm": 0.25729164481163025,
+      "learning_rate": 9.440905693708386e-05,
+      "loss": 4.0735,
+      "step": 6590
+    },
+    {
+      "epoch": 0.4480907732028808,
+      "grad_norm": 0.1723019927740097,
+      "learning_rate": 9.440481043620057e-05,
+      "loss": 4.0399,
+      "step": 6595
+    },
+    {
+      "epoch": 0.4484304932735426,
+      "grad_norm": 0.2043658047914505,
+      "learning_rate": 9.44005639353173e-05,
+      "loss": 4.2346,
+      "step": 6600
+    },
+    {
+      "epoch": 0.44877021334420436,
+      "grad_norm": 0.15108297765254974,
+      "learning_rate": 9.439631743443404e-05,
+      "loss": 3.9109,
+      "step": 6605
+    },
+    {
+      "epoch": 0.44910993341486616,
+      "grad_norm": 0.18265971541404724,
+      "learning_rate": 9.439207093355075e-05,
+      "loss": 3.845,
+      "step": 6610
+    },
+    {
+      "epoch": 0.4494496534855279,
+      "grad_norm": 0.9520887732505798,
+      "learning_rate": 9.438782443266748e-05,
+      "loss": 4.2181,
+      "step": 6615
+    },
+    {
+      "epoch": 0.4497893735561897,
+      "grad_norm": 0.28121015429496765,
+      "learning_rate": 9.438357793178422e-05,
+      "loss": 4.0234,
+      "step": 6620
+    },
+    {
+      "epoch": 0.4501290936268515,
+      "grad_norm": 0.21010081470012665,
+      "learning_rate": 9.437933143090094e-05,
+      "loss": 4.1509,
+      "step": 6625
+    },
+    {
+      "epoch": 0.45046881369751324,
+      "grad_norm": 0.23798321187496185,
+      "learning_rate": 9.437508493001767e-05,
+      "loss": 4.2879,
+      "step": 6630
+    },
+    {
+      "epoch": 0.45080853376817503,
+      "grad_norm": 0.20470625162124634,
+      "learning_rate": 9.43708384291344e-05,
+      "loss": 4.0134,
+      "step": 6635
+    },
+    {
+      "epoch": 0.4511482538388368,
+      "grad_norm": 0.19223268330097198,
+      "learning_rate": 9.436659192825112e-05,
+      "loss": 4.1928,
+      "step": 6640
+    },
+    {
+      "epoch": 0.4514879739094986,
+      "grad_norm": 0.1938367336988449,
+      "learning_rate": 9.436234542736785e-05,
+      "loss": 3.923,
+      "step": 6645
+    },
+    {
+      "epoch": 0.45182769398016037,
+      "grad_norm": 0.2842964231967926,
+      "learning_rate": 9.435809892648458e-05,
+      "loss": 4.2081,
+      "step": 6650
+    },
+    {
+      "epoch": 0.4521674140508221,
+      "grad_norm": 0.22615915536880493,
+      "learning_rate": 9.43538524256013e-05,
+      "loss": 4.0994,
+      "step": 6655
+    },
+    {
+      "epoch": 0.4525071341214839,
+      "grad_norm": 0.23465953767299652,
+      "learning_rate": 9.434960592471803e-05,
+      "loss": 3.8209,
+      "step": 6660
+    },
+    {
+      "epoch": 0.45284685419214565,
+      "grad_norm": 0.17599263787269592,
+      "learning_rate": 9.434535942383476e-05,
+      "loss": 3.8742,
+      "step": 6665
+    },
+    {
+      "epoch": 0.45318657426280745,
+      "grad_norm": 0.5463417172431946,
+      "learning_rate": 9.434111292295149e-05,
+      "loss": 4.1574,
+      "step": 6670
+    },
+    {
+      "epoch": 0.45352629433346925,
+      "grad_norm": 0.21346516907215118,
+      "learning_rate": 9.433686642206822e-05,
+      "loss": 4.3701,
+      "step": 6675
+    },
+    {
+      "epoch": 0.453866014404131,
+      "grad_norm": 0.2235599011182785,
+      "learning_rate": 9.433261992118495e-05,
+      "loss": 4.1951,
+      "step": 6680
+    },
+    {
+      "epoch": 0.4542057344747928,
+      "grad_norm": 0.2730211615562439,
+      "learning_rate": 9.432837342030167e-05,
+      "loss": 3.8224,
+      "step": 6685
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.1868310272693634,
+      "learning_rate": 9.43241269194184e-05,
+      "loss": 3.7951,
+      "step": 6690
+    },
+    {
+      "epoch": 0.45488517461611633,
+      "grad_norm": 0.3626730442047119,
+      "learning_rate": 9.431988041853513e-05,
+      "loss": 4.1127,
+      "step": 6695
+    },
+    {
+      "epoch": 0.45522489468677807,
+      "grad_norm": 0.22474856674671173,
+      "learning_rate": 9.431563391765186e-05,
+      "loss": 3.8253,
+      "step": 6700
+    },
+    {
+      "epoch": 0.45556461475743987,
+      "grad_norm": 0.3556784689426422,
+      "learning_rate": 9.431138741676859e-05,
+      "loss": 4.0104,
+      "step": 6705
+    },
+    {
+      "epoch": 0.45590433482810166,
+      "grad_norm": 1.793366551399231,
+      "learning_rate": 9.430714091588531e-05,
+      "loss": 4.1386,
+      "step": 6710
+    },
+    {
+      "epoch": 0.4562440548987634,
+      "grad_norm": 0.18291668593883514,
+      "learning_rate": 9.430289441500204e-05,
+      "loss": 4.118,
+      "step": 6715
+    },
+    {
+      "epoch": 0.4565837749694252,
+      "grad_norm": 0.3133949339389801,
+      "learning_rate": 9.429864791411877e-05,
+      "loss": 3.9527,
+      "step": 6720
+    },
+    {
+      "epoch": 0.45692349504008695,
+      "grad_norm": 0.2146921306848526,
+      "learning_rate": 9.42944014132355e-05,
+      "loss": 3.9596,
+      "step": 6725
+    },
+    {
+      "epoch": 0.45726321511074874,
+      "grad_norm": 0.17036296427249908,
+      "learning_rate": 9.429015491235223e-05,
+      "loss": 4.0381,
+      "step": 6730
+    },
+    {
+      "epoch": 0.45760293518141054,
+      "grad_norm": 0.32481151819229126,
+      "learning_rate": 9.428590841146895e-05,
+      "loss": 4.0932,
+      "step": 6735
+    },
+    {
+      "epoch": 0.4579426552520723,
+      "grad_norm": 0.18955262005329132,
+      "learning_rate": 9.428166191058568e-05,
+      "loss": 4.1081,
+      "step": 6740
+    },
+    {
+      "epoch": 0.4582823753227341,
+      "grad_norm": 0.2482958883047104,
+      "learning_rate": 9.427741540970241e-05,
+      "loss": 4.036,
+      "step": 6745
+    },
+    {
+      "epoch": 0.4586220953933958,
+      "grad_norm": 0.1786300390958786,
+      "learning_rate": 9.427316890881914e-05,
+      "loss": 4.2296,
+      "step": 6750
+    },
+    {
+      "epoch": 0.4589618154640576,
+      "grad_norm": 0.19114576280117035,
+      "learning_rate": 9.426892240793587e-05,
+      "loss": 4.1781,
+      "step": 6755
+    },
+    {
+      "epoch": 0.4593015355347194,
+      "grad_norm": 0.1777360886335373,
+      "learning_rate": 9.42646759070526e-05,
+      "loss": 3.7195,
+      "step": 6760
+    },
+    {
+      "epoch": 0.45964125560538116,
+      "grad_norm": 0.2255825698375702,
+      "learning_rate": 9.426042940616932e-05,
+      "loss": 4.0727,
+      "step": 6765
+    },
+    {
+      "epoch": 0.45998097567604296,
+      "grad_norm": 2.2815988063812256,
+      "learning_rate": 9.425618290528605e-05,
+      "loss": 4.1177,
+      "step": 6770
+    },
+    {
+      "epoch": 0.4603206957467047,
+      "grad_norm": 0.18281744420528412,
+      "learning_rate": 9.425193640440278e-05,
+      "loss": 4.1818,
+      "step": 6775
+    },
+    {
+      "epoch": 0.4606604158173665,
+      "grad_norm": 0.1932012438774109,
+      "learning_rate": 9.42476899035195e-05,
+      "loss": 4.207,
+      "step": 6780
+    },
+    {
+      "epoch": 0.46100013588802824,
+      "grad_norm": 0.3870634138584137,
+      "learning_rate": 9.424344340263623e-05,
+      "loss": 4.0997,
+      "step": 6785
+    },
+    {
+      "epoch": 0.46133985595869004,
+      "grad_norm": 0.20440851151943207,
+      "learning_rate": 9.423919690175296e-05,
+      "loss": 3.9454,
+      "step": 6790
+    },
+    {
+      "epoch": 0.46167957602935183,
+      "grad_norm": 0.21663671731948853,
+      "learning_rate": 9.423495040086969e-05,
+      "loss": 4.013,
+      "step": 6795
+    },
+    {
+      "epoch": 0.4620192961000136,
+      "grad_norm": 0.18921007215976715,
+      "learning_rate": 9.423070389998642e-05,
+      "loss": 4.0641,
+      "step": 6800
+    },
+    {
+      "epoch": 0.4623590161706754,
+      "grad_norm": 0.21008360385894775,
+      "learning_rate": 9.422645739910315e-05,
+      "loss": 3.8814,
+      "step": 6805
+    },
+    {
+      "epoch": 0.4626987362413371,
+      "grad_norm": 1.4397491216659546,
+      "learning_rate": 9.422221089821987e-05,
+      "loss": 3.977,
+      "step": 6810
+    },
+    {
+      "epoch": 0.4630384563119989,
+      "grad_norm": 0.1650581657886505,
+      "learning_rate": 9.42179643973366e-05,
+      "loss": 3.8511,
+      "step": 6815
+    },
+    {
+      "epoch": 0.4633781763826607,
+      "grad_norm": 0.24074020981788635,
+      "learning_rate": 9.421371789645333e-05,
+      "loss": 4.4683,
+      "step": 6820
+    },
+    {
+      "epoch": 0.46371789645332245,
+      "grad_norm": 0.2204151600599289,
+      "learning_rate": 9.420947139557006e-05,
+      "loss": 4.077,
+      "step": 6825
+    },
+    {
+      "epoch": 0.46405761652398425,
+      "grad_norm": 0.24461984634399414,
+      "learning_rate": 9.420522489468679e-05,
+      "loss": 3.9247,
+      "step": 6830
+    },
+    {
+      "epoch": 0.464397336594646,
+      "grad_norm": 0.19434142112731934,
+      "learning_rate": 9.42009783938035e-05,
+      "loss": 3.8453,
+      "step": 6835
+    },
+    {
+      "epoch": 0.4647370566653078,
+      "grad_norm": 0.2689877450466156,
+      "learning_rate": 9.419673189292024e-05,
+      "loss": 4.0808,
+      "step": 6840
+    },
+    {
+      "epoch": 0.4650767767359696,
+      "grad_norm": 0.2343098372220993,
+      "learning_rate": 9.419248539203697e-05,
+      "loss": 4.0781,
+      "step": 6845
+    },
+    {
+      "epoch": 0.46541649680663133,
+      "grad_norm": 0.19415900111198425,
+      "learning_rate": 9.418823889115369e-05,
+      "loss": 4.1067,
+      "step": 6850
+    },
+    {
+      "epoch": 0.4657562168772931,
+      "grad_norm": 0.20309092104434967,
+      "learning_rate": 9.418399239027043e-05,
+      "loss": 4.0426,
+      "step": 6855
+    },
+    {
+      "epoch": 0.46609593694795487,
+      "grad_norm": 0.35404348373413086,
+      "learning_rate": 9.417974588938715e-05,
+      "loss": 4.0913,
+      "step": 6860
+    },
+    {
+      "epoch": 0.46643565701861667,
+      "grad_norm": 0.8057840466499329,
+      "learning_rate": 9.417549938850387e-05,
+      "loss": 4.0028,
+      "step": 6865
+    },
+    {
+      "epoch": 0.4667753770892784,
+      "grad_norm": 0.23458503186702728,
+      "learning_rate": 9.417125288762061e-05,
+      "loss": 3.878,
+      "step": 6870
+    },
+    {
+      "epoch": 0.4671150971599402,
+      "grad_norm": 0.2019844651222229,
+      "learning_rate": 9.416700638673734e-05,
+      "loss": 3.7162,
+      "step": 6875
+    },
+    {
+      "epoch": 0.467454817230602,
+      "grad_norm": 0.19805146753787994,
+      "learning_rate": 9.416275988585405e-05,
+      "loss": 3.8687,
+      "step": 6880
+    },
+    {
+      "epoch": 0.46779453730126375,
+      "grad_norm": 0.2559395730495453,
+      "learning_rate": 9.41585133849708e-05,
+      "loss": 4.0791,
+      "step": 6885
+    },
+    {
+      "epoch": 0.46813425737192554,
+      "grad_norm": 0.3069989085197449,
+      "learning_rate": 9.415426688408752e-05,
+      "loss": 4.1204,
+      "step": 6890
+    },
+    {
+      "epoch": 0.4684739774425873,
+      "grad_norm": 0.5808936953544617,
+      "learning_rate": 9.415002038320424e-05,
+      "loss": 4.078,
+      "step": 6895
+    },
+    {
+      "epoch": 0.4688136975132491,
+      "grad_norm": 0.2510988414287567,
+      "learning_rate": 9.414577388232098e-05,
+      "loss": 4.23,
+      "step": 6900
+    },
+    {
+      "epoch": 0.4691534175839109,
+      "grad_norm": 0.2112618386745453,
+      "learning_rate": 9.414152738143769e-05,
+      "loss": 4.1868,
+      "step": 6905
+    },
+    {
+      "epoch": 0.4694931376545726,
+      "grad_norm": 0.22074821591377258,
+      "learning_rate": 9.413728088055442e-05,
+      "loss": 4.116,
+      "step": 6910
+    },
+    {
+      "epoch": 0.4698328577252344,
+      "grad_norm": 0.18397152423858643,
+      "learning_rate": 9.413303437967116e-05,
+      "loss": 4.0599,
+      "step": 6915
+    },
+    {
+      "epoch": 0.47017257779589616,
+      "grad_norm": 0.23679019510746002,
+      "learning_rate": 9.412878787878788e-05,
+      "loss": 3.8418,
+      "step": 6920
+    },
+    {
+      "epoch": 0.47051229786655796,
+      "grad_norm": 0.16602519154548645,
+      "learning_rate": 9.41245413779046e-05,
+      "loss": 4.0308,
+      "step": 6925
+    },
+    {
+      "epoch": 0.47085201793721976,
+      "grad_norm": 0.2898738384246826,
+      "learning_rate": 9.412029487702135e-05,
+      "loss": 4.2135,
+      "step": 6930
+    },
+    {
+      "epoch": 0.4711917380078815,
+      "grad_norm": 0.21048447489738464,
+      "learning_rate": 9.411604837613806e-05,
+      "loss": 3.9954,
+      "step": 6935
+    },
+    {
+      "epoch": 0.4715314580785433,
+      "grad_norm": 0.16546538472175598,
+      "learning_rate": 9.411180187525479e-05,
+      "loss": 4.1153,
+      "step": 6940
+    },
+    {
+      "epoch": 0.47187117814920504,
+      "grad_norm": 0.5077167749404907,
+      "learning_rate": 9.410755537437153e-05,
+      "loss": 4.1654,
+      "step": 6945
+    },
+    {
+      "epoch": 0.47221089821986684,
+      "grad_norm": 0.20563165843486786,
+      "learning_rate": 9.410330887348825e-05,
+      "loss": 3.9387,
+      "step": 6950
+    },
+    {
+      "epoch": 0.4725506182905286,
+      "grad_norm": 0.2332395762205124,
+      "learning_rate": 9.409906237260497e-05,
+      "loss": 4.2803,
+      "step": 6955
+    },
+    {
+      "epoch": 0.4728903383611904,
+      "grad_norm": 0.9494916796684265,
+      "learning_rate": 9.409481587172171e-05,
+      "loss": 4.1342,
+      "step": 6960
+    },
+    {
+      "epoch": 0.4732300584318522,
+      "grad_norm": 0.4015176296234131,
+      "learning_rate": 9.409056937083843e-05,
+      "loss": 4.1532,
+      "step": 6965
+    },
+    {
+      "epoch": 0.4735697785025139,
+      "grad_norm": 0.17285263538360596,
+      "learning_rate": 9.408632286995516e-05,
+      "loss": 3.8368,
+      "step": 6970
+    },
+    {
+      "epoch": 0.4739094985731757,
+      "grad_norm": 0.24160470068454742,
+      "learning_rate": 9.408207636907189e-05,
+      "loss": 4.0496,
+      "step": 6975
+    },
+    {
+      "epoch": 0.47424921864383746,
+      "grad_norm": 0.16916899383068085,
+      "learning_rate": 9.407782986818861e-05,
+      "loss": 4.3213,
+      "step": 6980
+    },
+    {
+      "epoch": 0.47458893871449925,
+      "grad_norm": 0.1876440942287445,
+      "learning_rate": 9.407358336730534e-05,
+      "loss": 4.0836,
+      "step": 6985
+    },
+    {
+      "epoch": 0.47492865878516105,
+      "grad_norm": 0.20803870260715485,
+      "learning_rate": 9.406933686642207e-05,
+      "loss": 4.0582,
+      "step": 6990
+    },
+    {
+      "epoch": 0.4752683788558228,
+      "grad_norm": 0.5098522305488586,
+      "learning_rate": 9.40650903655388e-05,
+      "loss": 3.7007,
+      "step": 6995
+    },
+    {
+      "epoch": 0.4756080989264846,
+      "grad_norm": 0.28446561098098755,
+      "learning_rate": 9.406084386465553e-05,
+      "loss": 3.7383,
+      "step": 7000
+    },
+    {
+      "epoch": 0.47594781899714633,
+      "grad_norm": 1.344814658164978,
+      "learning_rate": 9.405659736377225e-05,
+      "loss": 4.3367,
+      "step": 7005
+    },
+    {
+      "epoch": 0.47628753906780813,
+      "grad_norm": 0.3608788549900055,
+      "learning_rate": 9.405235086288898e-05,
+      "loss": 4.0906,
+      "step": 7010
+    },
+    {
+      "epoch": 0.4766272591384699,
+      "grad_norm": 0.2733428478240967,
+      "learning_rate": 9.404810436200571e-05,
+      "loss": 4.079,
+      "step": 7015
+    },
+    {
+      "epoch": 0.47696697920913167,
+      "grad_norm": 0.2144654393196106,
+      "learning_rate": 9.404385786112244e-05,
+      "loss": 3.5828,
+      "step": 7020
+    },
+    {
+      "epoch": 0.47730669927979347,
+      "grad_norm": 0.21566329896450043,
+      "learning_rate": 9.403961136023917e-05,
+      "loss": 4.1559,
+      "step": 7025
+    },
+    {
+      "epoch": 0.4776464193504552,
+      "grad_norm": 0.19979317486286163,
+      "learning_rate": 9.40353648593559e-05,
+      "loss": 4.078,
+      "step": 7030
+    },
+    {
+      "epoch": 0.477986139421117,
+      "grad_norm": 0.179921954870224,
+      "learning_rate": 9.403111835847262e-05,
+      "loss": 3.9188,
+      "step": 7035
+    },
+    {
+      "epoch": 0.47832585949177875,
+      "grad_norm": 0.17742060124874115,
+      "learning_rate": 9.402687185758935e-05,
+      "loss": 4.023,
+      "step": 7040
+    },
+    {
+      "epoch": 0.47866557956244055,
+      "grad_norm": 0.3981129229068756,
+      "learning_rate": 9.402262535670608e-05,
+      "loss": 4.0171,
+      "step": 7045
+    },
+    {
+      "epoch": 0.47900529963310234,
+      "grad_norm": 0.206672802567482,
+      "learning_rate": 9.40183788558228e-05,
+      "loss": 4.175,
+      "step": 7050
+    },
+    {
+      "epoch": 0.4793450197037641,
+      "grad_norm": 0.1959320604801178,
+      "learning_rate": 9.401413235493953e-05,
+      "loss": 4.1638,
+      "step": 7055
+    },
+    {
+      "epoch": 0.4796847397744259,
+      "grad_norm": 0.2009037733078003,
+      "learning_rate": 9.400988585405626e-05,
+      "loss": 4.1122,
+      "step": 7060
+    },
+    {
+      "epoch": 0.4800244598450876,
+      "grad_norm": 0.17592014372348785,
+      "learning_rate": 9.400563935317299e-05,
+      "loss": 4.0237,
+      "step": 7065
+    },
+    {
+      "epoch": 0.4803641799157494,
+      "grad_norm": 0.26748034358024597,
+      "learning_rate": 9.400139285228972e-05,
+      "loss": 3.8357,
+      "step": 7070
+    },
+    {
+      "epoch": 0.4807038999864112,
+      "grad_norm": 0.16173365712165833,
+      "learning_rate": 9.399714635140645e-05,
+      "loss": 4.0592,
+      "step": 7075
+    },
+    {
+      "epoch": 0.48104362005707296,
+      "grad_norm": 0.3452107906341553,
+      "learning_rate": 9.399289985052317e-05,
+      "loss": 4.1379,
+      "step": 7080
+    },
+    {
+      "epoch": 0.48138334012773476,
+      "grad_norm": 0.20402079820632935,
+      "learning_rate": 9.39886533496399e-05,
+      "loss": 4.2539,
+      "step": 7085
+    },
+    {
+      "epoch": 0.4817230601983965,
+      "grad_norm": 0.3040589690208435,
+      "learning_rate": 9.398440684875663e-05,
+      "loss": 4.0941,
+      "step": 7090
+    },
+    {
+      "epoch": 0.4820627802690583,
+      "grad_norm": 0.18901465833187103,
+      "learning_rate": 9.398016034787336e-05,
+      "loss": 4.0795,
+      "step": 7095
+    },
+    {
+      "epoch": 0.4824025003397201,
+      "grad_norm": 0.1665019541978836,
+      "learning_rate": 9.397591384699009e-05,
+      "loss": 4.0337,
+      "step": 7100
+    },
+    {
+      "epoch": 0.48274222041038184,
+      "grad_norm": 0.42847058176994324,
+      "learning_rate": 9.397166734610681e-05,
+      "loss": 4.108,
+      "step": 7105
+    },
+    {
+      "epoch": 0.48308194048104364,
+      "grad_norm": 0.21919941902160645,
+      "learning_rate": 9.396742084522354e-05,
+      "loss": 4.2498,
+      "step": 7110
+    },
+    {
+      "epoch": 0.4834216605517054,
+      "grad_norm": 0.7178875803947449,
+      "learning_rate": 9.396317434434027e-05,
+      "loss": 3.938,
+      "step": 7115
+    },
+    {
+      "epoch": 0.4837613806223672,
+      "grad_norm": 0.15290340781211853,
+      "learning_rate": 9.3958927843457e-05,
+      "loss": 4.1507,
+      "step": 7120
+    },
+    {
+      "epoch": 0.4841011006930289,
+      "grad_norm": 0.20199231803417206,
+      "learning_rate": 9.395468134257373e-05,
+      "loss": 4.165,
+      "step": 7125
+    },
+    {
+      "epoch": 0.4844408207636907,
+      "grad_norm": 0.4050713777542114,
+      "learning_rate": 9.395043484169045e-05,
+      "loss": 4.1098,
+      "step": 7130
+    },
+    {
+      "epoch": 0.4847805408343525,
+      "grad_norm": 0.22871138155460358,
+      "learning_rate": 9.394618834080718e-05,
+      "loss": 3.8713,
+      "step": 7135
+    },
+    {
+      "epoch": 0.48512026090501426,
+      "grad_norm": 0.2092018574476242,
+      "learning_rate": 9.394194183992391e-05,
+      "loss": 4.13,
+      "step": 7140
+    },
+    {
+      "epoch": 0.48545998097567605,
+      "grad_norm": 0.23514516651630402,
+      "learning_rate": 9.393769533904064e-05,
+      "loss": 4.1008,
+      "step": 7145
+    },
+    {
+      "epoch": 0.4857997010463378,
+      "grad_norm": 0.19990748167037964,
+      "learning_rate": 9.393344883815737e-05,
+      "loss": 4.1277,
+      "step": 7150
+    },
+    {
+      "epoch": 0.4861394211169996,
+      "grad_norm": 0.23331451416015625,
+      "learning_rate": 9.39292023372741e-05,
+      "loss": 4.1307,
+      "step": 7155
+    },
+    {
+      "epoch": 0.4864791411876614,
+      "grad_norm": 0.4659624397754669,
+      "learning_rate": 9.392495583639082e-05,
+      "loss": 4.0373,
+      "step": 7160
+    },
+    {
+      "epoch": 0.48681886125832313,
+      "grad_norm": 0.350339412689209,
+      "learning_rate": 9.392070933550755e-05,
+      "loss": 4.0145,
+      "step": 7165
+    },
+    {
+      "epoch": 0.48715858132898493,
+      "grad_norm": 0.29896309971809387,
+      "learning_rate": 9.391646283462428e-05,
+      "loss": 3.7892,
+      "step": 7170
+    },
+    {
+      "epoch": 0.48749830139964667,
+      "grad_norm": 0.2809394896030426,
+      "learning_rate": 9.391221633374099e-05,
+      "loss": 4.0247,
+      "step": 7175
+    },
+    {
+      "epoch": 0.48783802147030847,
+      "grad_norm": 0.2700156271457672,
+      "learning_rate": 9.390796983285773e-05,
+      "loss": 4.0203,
+      "step": 7180
+    },
+    {
+      "epoch": 0.48817774154097027,
+      "grad_norm": 0.2099238932132721,
+      "learning_rate": 9.390372333197446e-05,
+      "loss": 3.8992,
+      "step": 7185
+    },
+    {
+      "epoch": 0.488517461611632,
+      "grad_norm": 0.1657472550868988,
+      "learning_rate": 9.389947683109118e-05,
+      "loss": 4.0689,
+      "step": 7190
+    },
+    {
+      "epoch": 0.4888571816822938,
+      "grad_norm": 2.5052669048309326,
+      "learning_rate": 9.389523033020792e-05,
+      "loss": 4.2311,
+      "step": 7195
+    },
+    {
+      "epoch": 0.48919690175295555,
+      "grad_norm": 0.18674081563949585,
+      "learning_rate": 9.389098382932465e-05,
+      "loss": 4.0554,
+      "step": 7200
+    },
+    {
+      "epoch": 0.48953662182361735,
+      "grad_norm": 0.17581969499588013,
+      "learning_rate": 9.388673732844136e-05,
+      "loss": 3.9992,
+      "step": 7205
+    },
+    {
+      "epoch": 0.4898763418942791,
+      "grad_norm": 0.2378435879945755,
+      "learning_rate": 9.38824908275581e-05,
+      "loss": 4.1965,
+      "step": 7210
+    },
+    {
+      "epoch": 0.4902160619649409,
+      "grad_norm": 0.21895438432693481,
+      "learning_rate": 9.387824432667483e-05,
+      "loss": 3.9994,
+      "step": 7215
+    },
+    {
+      "epoch": 0.4905557820356027,
+      "grad_norm": 0.18622034788131714,
+      "learning_rate": 9.387399782579154e-05,
+      "loss": 4.2384,
+      "step": 7220
+    },
+    {
+      "epoch": 0.4908955021062644,
+      "grad_norm": 0.21598079800605774,
+      "learning_rate": 9.386975132490829e-05,
+      "loss": 4.2434,
+      "step": 7225
+    },
+    {
+      "epoch": 0.4912352221769262,
+      "grad_norm": 1.403287410736084,
+      "learning_rate": 9.386550482402501e-05,
+      "loss": 3.9085,
+      "step": 7230
+    },
+    {
+      "epoch": 0.49157494224758796,
+      "grad_norm": 0.31629684567451477,
+      "learning_rate": 9.386125832314173e-05,
+      "loss": 4.0834,
+      "step": 7235
+    },
+    {
+      "epoch": 0.49191466231824976,
+      "grad_norm": 0.503648042678833,
+      "learning_rate": 9.385701182225847e-05,
+      "loss": 4.0346,
+      "step": 7240
+    },
+    {
+      "epoch": 0.49225438238891156,
+      "grad_norm": 0.16960012912750244,
+      "learning_rate": 9.38527653213752e-05,
+      "loss": 4.0295,
+      "step": 7245
+    },
+    {
+      "epoch": 0.4925941024595733,
+      "grad_norm": 0.2992265224456787,
+      "learning_rate": 9.384851882049191e-05,
+      "loss": 4.147,
+      "step": 7250
+    },
+    {
+      "epoch": 0.4929338225302351,
+      "grad_norm": 0.37070542573928833,
+      "learning_rate": 9.384427231960865e-05,
+      "loss": 3.6991,
+      "step": 7255
+    },
+    {
+      "epoch": 0.49327354260089684,
+      "grad_norm": 0.2252090573310852,
+      "learning_rate": 9.384002581872537e-05,
+      "loss": 4.2128,
+      "step": 7260
+    },
+    {
+      "epoch": 0.49361326267155864,
+      "grad_norm": 0.1695706844329834,
+      "learning_rate": 9.38357793178421e-05,
+      "loss": 4.0321,
+      "step": 7265
+    },
+    {
+      "epoch": 0.49395298274222044,
+      "grad_norm": 0.23216302692890167,
+      "learning_rate": 9.383153281695884e-05,
+      "loss": 3.9758,
+      "step": 7270
+    },
+    {
+      "epoch": 0.4942927028128822,
+      "grad_norm": 0.9034651517868042,
+      "learning_rate": 9.382728631607555e-05,
+      "loss": 3.9329,
+      "step": 7275
+    },
+    {
+      "epoch": 0.494632422883544,
+      "grad_norm": 0.21417531371116638,
+      "learning_rate": 9.382303981519228e-05,
+      "loss": 3.8799,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4949721429542057,
+      "grad_norm": 0.18411661684513092,
+      "learning_rate": 9.381879331430902e-05,
+      "loss": 4.2956,
+      "step": 7285
+    },
+    {
+      "epoch": 0.4953118630248675,
+      "grad_norm": 0.23416036367416382,
+      "learning_rate": 9.381454681342574e-05,
+      "loss": 3.8779,
+      "step": 7290
+    },
+    {
+      "epoch": 0.49565158309552926,
+      "grad_norm": 0.2349618524312973,
+      "learning_rate": 9.381030031254246e-05,
+      "loss": 4.027,
+      "step": 7295
+    },
+    {
+      "epoch": 0.49599130316619106,
+      "grad_norm": 0.2522861659526825,
+      "learning_rate": 9.38060538116592e-05,
+      "loss": 4.1115,
+      "step": 7300
+    },
+    {
+      "epoch": 0.49633102323685285,
+      "grad_norm": 0.18916946649551392,
+      "learning_rate": 9.380180731077592e-05,
+      "loss": 4.1002,
+      "step": 7305
+    },
+    {
+      "epoch": 0.4966707433075146,
+      "grad_norm": 0.20227526128292084,
+      "learning_rate": 9.379756080989265e-05,
+      "loss": 4.0984,
+      "step": 7310
+    },
+    {
+      "epoch": 0.4970104633781764,
+      "grad_norm": 0.20442931354045868,
+      "learning_rate": 9.379331430900939e-05,
+      "loss": 4.0096,
+      "step": 7315
+    },
+    {
+      "epoch": 0.49735018344883813,
+      "grad_norm": 0.19708792865276337,
+      "learning_rate": 9.37890678081261e-05,
+      "loss": 4.2264,
+      "step": 7320
+    },
+    {
+      "epoch": 0.49768990351949993,
+      "grad_norm": 0.2058994174003601,
+      "learning_rate": 9.378482130724283e-05,
+      "loss": 3.9661,
+      "step": 7325
+    },
+    {
+      "epoch": 0.49802962359016173,
+      "grad_norm": 0.18831300735473633,
+      "learning_rate": 9.378057480635956e-05,
+      "loss": 4.0001,
+      "step": 7330
+    },
+    {
+      "epoch": 0.49836934366082347,
+      "grad_norm": 0.5251606702804565,
+      "learning_rate": 9.377632830547629e-05,
+      "loss": 4.1706,
+      "step": 7335
+    },
+    {
+      "epoch": 0.49870906373148527,
+      "grad_norm": 0.17007534205913544,
+      "learning_rate": 9.377208180459302e-05,
+      "loss": 4.1699,
+      "step": 7340
+    },
+    {
+      "epoch": 0.499048783802147,
+      "grad_norm": 0.3484830856323242,
+      "learning_rate": 9.376783530370974e-05,
+      "loss": 3.944,
+      "step": 7345
+    },
+    {
+      "epoch": 0.4993885038728088,
+      "grad_norm": 0.20382869243621826,
+      "learning_rate": 9.376358880282647e-05,
+      "loss": 3.7016,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4997282239434706,
+      "grad_norm": 0.2002745270729065,
+      "learning_rate": 9.37593423019432e-05,
+      "loss": 4.1238,
+      "step": 7355
+    },
+    {
+      "epoch": 0.5000679440141323,
+      "grad_norm": 0.17399045825004578,
+      "learning_rate": 9.375509580105993e-05,
+      "loss": 4.0909,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5004076640847941,
+      "grad_norm": 0.24848084151744843,
+      "learning_rate": 9.375084930017666e-05,
+      "loss": 4.1379,
+      "step": 7365
+    },
+    {
+      "epoch": 0.5007473841554559,
+      "grad_norm": 0.5024029016494751,
+      "learning_rate": 9.374660279929338e-05,
+      "loss": 4.2591,
+      "step": 7370
+    },
+    {
+      "epoch": 0.5010871042261177,
+      "grad_norm": 0.22552575170993805,
+      "learning_rate": 9.374235629841011e-05,
+      "loss": 4.049,
+      "step": 7375
+    },
+    {
+      "epoch": 0.5014268242967794,
+      "grad_norm": 0.2249031662940979,
+      "learning_rate": 9.373810979752684e-05,
+      "loss": 4.2421,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5017665443674413,
+      "grad_norm": 0.22408431768417358,
+      "learning_rate": 9.373386329664357e-05,
+      "loss": 4.3666,
+      "step": 7385
+    },
+    {
+      "epoch": 0.502106264438103,
+      "grad_norm": 0.16393537819385529,
+      "learning_rate": 9.37296167957603e-05,
+      "loss": 3.7884,
+      "step": 7390
+    },
+    {
+      "epoch": 0.5024459845087648,
+      "grad_norm": 0.25391802191734314,
+      "learning_rate": 9.372537029487702e-05,
+      "loss": 4.1085,
+      "step": 7395
+    },
+    {
+      "epoch": 0.5027857045794265,
+      "grad_norm": 0.25248852372169495,
+      "learning_rate": 9.372112379399375e-05,
+      "loss": 4.0558,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5031254246500884,
+      "grad_norm": 0.2197033017873764,
+      "learning_rate": 9.371687729311048e-05,
+      "loss": 3.979,
+      "step": 7405
+    },
+    {
+      "epoch": 0.5034651447207501,
+      "grad_norm": 0.20195040106773376,
+      "learning_rate": 9.371263079222721e-05,
+      "loss": 3.8918,
+      "step": 7410
+    },
+    {
+      "epoch": 0.5038048647914118,
+      "grad_norm": 0.1969507336616516,
+      "learning_rate": 9.370838429134394e-05,
+      "loss": 3.9211,
+      "step": 7415
+    },
+    {
+      "epoch": 0.5041445848620737,
+      "grad_norm": 0.26221612095832825,
+      "learning_rate": 9.370413779046066e-05,
+      "loss": 4.067,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5044843049327354,
+      "grad_norm": 0.20401246845722198,
+      "learning_rate": 9.369989128957739e-05,
+      "loss": 3.9535,
+      "step": 7425
+    },
+    {
+      "epoch": 0.5048240250033972,
+      "grad_norm": 0.5490508675575256,
+      "learning_rate": 9.369564478869412e-05,
+      "loss": 3.9772,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5051637450740589,
+      "grad_norm": 0.1551188975572586,
+      "learning_rate": 9.369139828781085e-05,
+      "loss": 4.1125,
+      "step": 7435
+    },
+    {
+      "epoch": 0.5055034651447208,
+      "grad_norm": 0.31335705518722534,
+      "learning_rate": 9.368715178692758e-05,
+      "loss": 4.0064,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5058431852153825,
+      "grad_norm": 0.23974861204624176,
+      "learning_rate": 9.36829052860443e-05,
+      "loss": 3.9679,
+      "step": 7445
+    },
+    {
+      "epoch": 0.5061829052860443,
+      "grad_norm": 0.15649579465389252,
+      "learning_rate": 9.367865878516103e-05,
+      "loss": 4.005,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5065226253567061,
+      "grad_norm": 0.18585532903671265,
+      "learning_rate": 9.367441228427776e-05,
+      "loss": 3.8728,
+      "step": 7455
+    },
+    {
+      "epoch": 0.5068623454273679,
+      "grad_norm": 0.2581709623336792,
+      "learning_rate": 9.367016578339449e-05,
+      "loss": 4.0771,
+      "step": 7460
+    },
+    {
+      "epoch": 0.5072020654980296,
+      "grad_norm": 0.2468833178281784,
+      "learning_rate": 9.366591928251122e-05,
+      "loss": 4.054,
+      "step": 7465
+    },
+    {
+      "epoch": 0.5075417855686915,
+      "grad_norm": 0.2019713968038559,
+      "learning_rate": 9.366167278162794e-05,
+      "loss": 3.6126,
+      "step": 7470
+    },
+    {
+      "epoch": 0.5078815056393532,
+      "grad_norm": 0.25231555104255676,
+      "learning_rate": 9.365742628074467e-05,
+      "loss": 4.4004,
+      "step": 7475
+    },
+    {
+      "epoch": 0.5082212257100149,
+      "grad_norm": 0.14785149693489075,
+      "learning_rate": 9.36531797798614e-05,
+      "loss": 3.7965,
+      "step": 7480
+    },
+    {
+      "epoch": 0.5085609457806767,
+      "grad_norm": 0.21854764223098755,
+      "learning_rate": 9.364893327897813e-05,
+      "loss": 4.3081,
+      "step": 7485
+    },
+    {
+      "epoch": 0.5089006658513385,
+      "grad_norm": 0.35712555050849915,
+      "learning_rate": 9.364468677809486e-05,
+      "loss": 3.7848,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5092403859220003,
+      "grad_norm": 0.9170289039611816,
+      "learning_rate": 9.364044027721158e-05,
+      "loss": 4.0845,
+      "step": 7495
+    },
+    {
+      "epoch": 0.509580105992662,
+      "grad_norm": 0.19155257940292358,
+      "learning_rate": 9.363619377632831e-05,
+      "loss": 4.0807,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5099198260633239,
+      "grad_norm": 0.20362383127212524,
+      "learning_rate": 9.363194727544504e-05,
+      "loss": 4.1181,
+      "step": 7505
+    },
+    {
+      "epoch": 0.5102595461339856,
+      "grad_norm": 1.610526204109192,
+      "learning_rate": 9.362770077456177e-05,
+      "loss": 4.1332,
+      "step": 7510
+    },
+    {
+      "epoch": 0.5105992662046474,
+      "grad_norm": 0.36077892780303955,
+      "learning_rate": 9.36234542736785e-05,
+      "loss": 4.0379,
+      "step": 7515
+    },
+    {
+      "epoch": 0.5109389862753091,
+      "grad_norm": 0.34184160828590393,
+      "learning_rate": 9.361920777279522e-05,
+      "loss": 4.0309,
+      "step": 7520
+    },
+    {
+      "epoch": 0.511278706345971,
+      "grad_norm": 0.266156941652298,
+      "learning_rate": 9.361496127191195e-05,
+      "loss": 4.0715,
+      "step": 7525
+    },
+    {
+      "epoch": 0.5116184264166327,
+      "grad_norm": 0.21247410774230957,
+      "learning_rate": 9.361071477102867e-05,
+      "loss": 4.1461,
+      "step": 7530
+    },
+    {
+      "epoch": 0.5119581464872944,
+      "grad_norm": 0.3173115849494934,
+      "learning_rate": 9.360646827014541e-05,
+      "loss": 4.0909,
+      "step": 7535
+    },
+    {
+      "epoch": 0.5122978665579563,
+      "grad_norm": 0.1932353973388672,
+      "learning_rate": 9.360222176926214e-05,
+      "loss": 4.0699,
+      "step": 7540
+    },
+    {
+      "epoch": 0.512637586628618,
+      "grad_norm": 0.34887808561325073,
+      "learning_rate": 9.359797526837885e-05,
+      "loss": 3.8442,
+      "step": 7545
+    },
+    {
+      "epoch": 0.5129773066992798,
+      "grad_norm": 0.1603212207555771,
+      "learning_rate": 9.359372876749559e-05,
+      "loss": 3.9252,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5133170267699416,
+      "grad_norm": 0.18673382699489594,
+      "learning_rate": 9.358948226661232e-05,
+      "loss": 4.132,
+      "step": 7555
+    },
+    {
+      "epoch": 0.5136567468406034,
+      "grad_norm": 0.17931464314460754,
+      "learning_rate": 9.358523576572904e-05,
+      "loss": 4.2045,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5139964669112651,
+      "grad_norm": 0.20832332968711853,
+      "learning_rate": 9.358098926484578e-05,
+      "loss": 4.1134,
+      "step": 7565
+    },
+    {
+      "epoch": 0.5143361869819268,
+      "grad_norm": 0.1900588423013687,
+      "learning_rate": 9.35767427639625e-05,
+      "loss": 4.1737,
+      "step": 7570
+    },
+    {
+      "epoch": 0.5146759070525887,
+      "grad_norm": 0.25555694103240967,
+      "learning_rate": 9.357249626307922e-05,
+      "loss": 3.9879,
+      "step": 7575
+    },
+    {
+      "epoch": 0.5150156271232504,
+      "grad_norm": 0.3741958737373352,
+      "learning_rate": 9.356824976219596e-05,
+      "loss": 4.0922,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5153553471939122,
+      "grad_norm": 0.25290772318840027,
+      "learning_rate": 9.356400326131269e-05,
+      "loss": 4.1101,
+      "step": 7585
+    },
+    {
+      "epoch": 0.515695067264574,
+      "grad_norm": 0.23055055737495422,
+      "learning_rate": 9.35597567604294e-05,
+      "loss": 3.9841,
+      "step": 7590
+    },
+    {
+      "epoch": 0.5160347873352358,
+      "grad_norm": 0.2004847675561905,
+      "learning_rate": 9.355551025954614e-05,
+      "loss": 4.0573,
+      "step": 7595
+    },
+    {
+      "epoch": 0.5163745074058975,
+      "grad_norm": 0.17430374026298523,
+      "learning_rate": 9.355126375866287e-05,
+      "loss": 4.1038,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5167142274765593,
+      "grad_norm": 0.19129951298236847,
+      "learning_rate": 9.354701725777959e-05,
+      "loss": 3.9517,
+      "step": 7605
+    },
+    {
+      "epoch": 0.5170539475472211,
+      "grad_norm": 0.24866996705532074,
+      "learning_rate": 9.354277075689633e-05,
+      "loss": 3.9927,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5173936676178829,
+      "grad_norm": 0.2479562759399414,
+      "learning_rate": 9.353852425601304e-05,
+      "loss": 3.824,
+      "step": 7615
+    },
+    {
+      "epoch": 0.5177333876885446,
+      "grad_norm": 0.25535717606544495,
+      "learning_rate": 9.353427775512977e-05,
+      "loss": 4.1203,
+      "step": 7620
+    },
+    {
+      "epoch": 0.5180731077592065,
+      "grad_norm": 0.19401293992996216,
+      "learning_rate": 9.353003125424651e-05,
+      "loss": 4.2723,
+      "step": 7625
+    },
+    {
+      "epoch": 0.5184128278298682,
+      "grad_norm": 0.25242581963539124,
+      "learning_rate": 9.352578475336323e-05,
+      "loss": 3.7278,
+      "step": 7630
+    },
+    {
+      "epoch": 0.5187525479005299,
+      "grad_norm": 0.24895191192626953,
+      "learning_rate": 9.352153825247996e-05,
+      "loss": 4.0693,
+      "step": 7635
+    },
+    {
+      "epoch": 0.5190922679711918,
+      "grad_norm": 0.208522230386734,
+      "learning_rate": 9.35172917515967e-05,
+      "loss": 4.105,
+      "step": 7640
+    },
+    {
+      "epoch": 0.5194319880418535,
+      "grad_norm": 0.21266648173332214,
+      "learning_rate": 9.351304525071341e-05,
+      "loss": 4.0295,
+      "step": 7645
+    },
+    {
+      "epoch": 0.5197717081125153,
+      "grad_norm": 0.1788320690393448,
+      "learning_rate": 9.350879874983014e-05,
+      "loss": 3.9959,
+      "step": 7650
+    },
+    {
+      "epoch": 0.520111428183177,
+      "grad_norm": 0.36194828152656555,
+      "learning_rate": 9.350455224894688e-05,
+      "loss": 4.0003,
+      "step": 7655
+    },
+    {
+      "epoch": 0.5204511482538389,
+      "grad_norm": 0.22471646964550018,
+      "learning_rate": 9.35003057480636e-05,
+      "loss": 4.0001,
+      "step": 7660
+    },
+    {
+      "epoch": 0.5207908683245006,
+      "grad_norm": 0.20140118896961212,
+      "learning_rate": 9.349605924718032e-05,
+      "loss": 3.802,
+      "step": 7665
+    },
+    {
+      "epoch": 0.5211305883951624,
+      "grad_norm": 0.2610224485397339,
+      "learning_rate": 9.349181274629706e-05,
+      "loss": 4.0378,
+      "step": 7670
+    },
+    {
+      "epoch": 0.5214703084658242,
+      "grad_norm": 0.19528135657310486,
+      "learning_rate": 9.348756624541378e-05,
+      "loss": 4.059,
+      "step": 7675
+    },
+    {
+      "epoch": 0.521810028536486,
+      "grad_norm": 0.1775432527065277,
+      "learning_rate": 9.348331974453051e-05,
+      "loss": 4.0532,
+      "step": 7680
+    },
+    {
+      "epoch": 0.5221497486071477,
+      "grad_norm": 4.0261311531066895,
+      "learning_rate": 9.347907324364724e-05,
+      "loss": 4.1247,
+      "step": 7685
+    },
+    {
+      "epoch": 0.5224894686778094,
+      "grad_norm": 0.18927231431007385,
+      "learning_rate": 9.347482674276396e-05,
+      "loss": 4.1597,
+      "step": 7690
+    },
+    {
+      "epoch": 0.5228291887484713,
+      "grad_norm": 0.17525963485240936,
+      "learning_rate": 9.347058024188069e-05,
+      "loss": 3.7783,
+      "step": 7695
+    },
+    {
+      "epoch": 0.523168908819133,
+      "grad_norm": 0.15147970616817474,
+      "learning_rate": 9.346633374099742e-05,
+      "loss": 4.017,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5235086288897948,
+      "grad_norm": 0.17710480093955994,
+      "learning_rate": 9.346208724011415e-05,
+      "loss": 4.3239,
+      "step": 7705
+    },
+    {
+      "epoch": 0.5238483489604566,
+      "grad_norm": 0.1875525414943695,
+      "learning_rate": 9.345784073923088e-05,
+      "loss": 3.9417,
+      "step": 7710
+    },
+    {
+      "epoch": 0.5241880690311184,
+      "grad_norm": 0.2575678825378418,
+      "learning_rate": 9.34535942383476e-05,
+      "loss": 4.1485,
+      "step": 7715
+    },
+    {
+      "epoch": 0.5245277891017801,
+      "grad_norm": 0.8326651453971863,
+      "learning_rate": 9.344934773746433e-05,
+      "loss": 4.0598,
+      "step": 7720
+    },
+    {
+      "epoch": 0.524867509172442,
+      "grad_norm": 0.1673835813999176,
+      "learning_rate": 9.344510123658106e-05,
+      "loss": 4.1255,
+      "step": 7725
+    },
+    {
+      "epoch": 0.5252072292431037,
+      "grad_norm": 0.764521598815918,
+      "learning_rate": 9.344085473569779e-05,
+      "loss": 3.9019,
+      "step": 7730
+    },
+    {
+      "epoch": 0.5255469493137654,
+      "grad_norm": 0.17800885438919067,
+      "learning_rate": 9.343660823481452e-05,
+      "loss": 3.7869,
+      "step": 7735
+    },
+    {
+      "epoch": 0.5258866693844272,
+      "grad_norm": 0.1920061856508255,
+      "learning_rate": 9.343236173393124e-05,
+      "loss": 4.1229,
+      "step": 7740
+    },
+    {
+      "epoch": 0.526226389455089,
+      "grad_norm": 3.5604405403137207,
+      "learning_rate": 9.342811523304797e-05,
+      "loss": 3.9478,
+      "step": 7745
+    },
+    {
+      "epoch": 0.5265661095257508,
+      "grad_norm": 0.194805309176445,
+      "learning_rate": 9.34238687321647e-05,
+      "loss": 4.2643,
+      "step": 7750
+    },
+    {
+      "epoch": 0.5269058295964125,
+      "grad_norm": 0.22051791846752167,
+      "learning_rate": 9.341962223128143e-05,
+      "loss": 4.1483,
+      "step": 7755
+    },
+    {
+      "epoch": 0.5272455496670744,
+      "grad_norm": 0.22619082033634186,
+      "learning_rate": 9.341622503057481e-05,
+      "loss": 3.5109,
+      "step": 7760
+    },
+    {
+      "epoch": 0.5275852697377361,
+      "grad_norm": 0.2676302194595337,
+      "learning_rate": 9.341197852969154e-05,
+      "loss": 4.174,
+      "step": 7765
+    },
+    {
+      "epoch": 0.5279249898083979,
+      "grad_norm": 0.25767549872398376,
+      "learning_rate": 9.340773202880827e-05,
+      "loss": 3.9739,
+      "step": 7770
+    },
+    {
+      "epoch": 0.5282647098790596,
+      "grad_norm": 0.18924906849861145,
+      "learning_rate": 9.3403485527925e-05,
+      "loss": 4.2547,
+      "step": 7775
+    },
+    {
+      "epoch": 0.5286044299497215,
+      "grad_norm": 0.1977754682302475,
+      "learning_rate": 9.339923902704172e-05,
+      "loss": 3.8264,
+      "step": 7780
+    },
+    {
+      "epoch": 0.5289441500203832,
+      "grad_norm": 0.1865202784538269,
+      "learning_rate": 9.339499252615845e-05,
+      "loss": 4.2391,
+      "step": 7785
+    },
+    {
+      "epoch": 0.5292838700910449,
+      "grad_norm": 0.22030463814735413,
+      "learning_rate": 9.339074602527518e-05,
+      "loss": 4.1641,
+      "step": 7790
+    },
+    {
+      "epoch": 0.5296235901617068,
+      "grad_norm": 0.1868383139371872,
+      "learning_rate": 9.33864995243919e-05,
+      "loss": 4.247,
+      "step": 7795
+    },
+    {
+      "epoch": 0.5299633102323685,
+      "grad_norm": 0.19364790618419647,
+      "learning_rate": 9.338225302350863e-05,
+      "loss": 3.9474,
+      "step": 7800
+    },
+    {
+      "epoch": 0.5303030303030303,
+      "grad_norm": 0.1819707751274109,
+      "learning_rate": 9.337800652262536e-05,
+      "loss": 4.0559,
+      "step": 7805
+    },
+    {
+      "epoch": 0.5306427503736921,
+      "grad_norm": 0.25897395610809326,
+      "learning_rate": 9.337376002174209e-05,
+      "loss": 3.9502,
+      "step": 7810
+    },
+    {
+      "epoch": 0.5309824704443539,
+      "grad_norm": 0.23874245584011078,
+      "learning_rate": 9.336951352085882e-05,
+      "loss": 3.89,
+      "step": 7815
+    },
+    {
+      "epoch": 0.5313221905150156,
+      "grad_norm": 0.16747428476810455,
+      "learning_rate": 9.336526701997555e-05,
+      "loss": 4.0011,
+      "step": 7820
+    },
+    {
+      "epoch": 0.5316619105856774,
+      "grad_norm": 0.17043693363666534,
+      "learning_rate": 9.336102051909227e-05,
+      "loss": 4.0938,
+      "step": 7825
+    },
+    {
+      "epoch": 0.5320016306563392,
+      "grad_norm": 0.19424232840538025,
+      "learning_rate": 9.3356774018209e-05,
+      "loss": 3.9479,
+      "step": 7830
+    },
+    {
+      "epoch": 0.532341350727001,
+      "grad_norm": 0.1900462657213211,
+      "learning_rate": 9.335252751732573e-05,
+      "loss": 4.2105,
+      "step": 7835
+    },
+    {
+      "epoch": 0.5326810707976627,
+      "grad_norm": 0.2120985984802246,
+      "learning_rate": 9.334828101644246e-05,
+      "loss": 4.2686,
+      "step": 7840
+    },
+    {
+      "epoch": 0.5330207908683245,
+      "grad_norm": 0.24349243938922882,
+      "learning_rate": 9.334403451555919e-05,
+      "loss": 4.09,
+      "step": 7845
+    },
+    {
+      "epoch": 0.5333605109389863,
+      "grad_norm": 0.22167713940143585,
+      "learning_rate": 9.333978801467591e-05,
+      "loss": 4.0405,
+      "step": 7850
+    },
+    {
+      "epoch": 0.533700231009648,
+      "grad_norm": 0.2967727780342102,
+      "learning_rate": 9.333554151379263e-05,
+      "loss": 4.3056,
+      "step": 7855
+    },
+    {
+      "epoch": 0.5340399510803098,
+      "grad_norm": 0.1543935239315033,
+      "learning_rate": 9.333129501290937e-05,
+      "loss": 4.0946,
+      "step": 7860
+    },
+    {
+      "epoch": 0.5343796711509716,
+      "grad_norm": 0.27082520723342896,
+      "learning_rate": 9.33270485120261e-05,
+      "loss": 3.9952,
+      "step": 7865
+    },
+    {
+      "epoch": 0.5347193912216334,
+      "grad_norm": 0.27814018726348877,
+      "learning_rate": 9.332280201114281e-05,
+      "loss": 4.2333,
+      "step": 7870
+    },
+    {
+      "epoch": 0.5350591112922951,
+      "grad_norm": 0.18961313366889954,
+      "learning_rate": 9.331855551025955e-05,
+      "loss": 4.1017,
+      "step": 7875
+    },
+    {
+      "epoch": 0.535398831362957,
+      "grad_norm": 0.19121582806110382,
+      "learning_rate": 9.331430900937628e-05,
+      "loss": 4.1663,
+      "step": 7880
+    },
+    {
+      "epoch": 0.5357385514336187,
+      "grad_norm": 0.20209085941314697,
+      "learning_rate": 9.3310062508493e-05,
+      "loss": 3.9602,
+      "step": 7885
+    },
+    {
+      "epoch": 0.5360782715042804,
+      "grad_norm": 0.222623810172081,
+      "learning_rate": 9.330581600760974e-05,
+      "loss": 4.208,
+      "step": 7890
+    },
+    {
+      "epoch": 0.5364179915749423,
+      "grad_norm": 0.2834756672382355,
+      "learning_rate": 9.330156950672647e-05,
+      "loss": 3.8017,
+      "step": 7895
+    },
+    {
+      "epoch": 0.536757711645604,
+      "grad_norm": 0.26975885033607483,
+      "learning_rate": 9.329732300584318e-05,
+      "loss": 3.8618,
+      "step": 7900
+    },
+    {
+      "epoch": 0.5370974317162658,
+      "grad_norm": 0.24554960429668427,
+      "learning_rate": 9.329307650495992e-05,
+      "loss": 4.1339,
+      "step": 7905
+    },
+    {
+      "epoch": 0.5374371517869275,
+      "grad_norm": 0.19073788821697235,
+      "learning_rate": 9.328883000407665e-05,
+      "loss": 4.1798,
+      "step": 7910
+    },
+    {
+      "epoch": 0.5377768718575894,
+      "grad_norm": 0.1700531542301178,
+      "learning_rate": 9.328458350319336e-05,
+      "loss": 4.1152,
+      "step": 7915
+    },
+    {
+      "epoch": 0.5381165919282511,
+      "grad_norm": 0.2178450971841812,
+      "learning_rate": 9.32803370023101e-05,
+      "loss": 3.9352,
+      "step": 7920
+    },
+    {
+      "epoch": 0.5384563119989129,
+      "grad_norm": 0.20611537992954254,
+      "learning_rate": 9.327609050142682e-05,
+      "loss": 4.0871,
+      "step": 7925
+    },
+    {
+      "epoch": 0.5387960320695747,
+      "grad_norm": 0.20074544847011566,
+      "learning_rate": 9.327184400054355e-05,
+      "loss": 4.1689,
+      "step": 7930
+    },
+    {
+      "epoch": 0.5391357521402365,
+      "grad_norm": 0.18785405158996582,
+      "learning_rate": 9.326759749966029e-05,
+      "loss": 4.0912,
+      "step": 7935
+    },
+    {
+      "epoch": 0.5394754722108982,
+      "grad_norm": 0.2205738127231598,
+      "learning_rate": 9.3263350998777e-05,
+      "loss": 4.0251,
+      "step": 7940
+    },
+    {
+      "epoch": 0.5398151922815599,
+      "grad_norm": 0.2106117606163025,
+      "learning_rate": 9.325910449789373e-05,
+      "loss": 3.9616,
+      "step": 7945
+    },
+    {
+      "epoch": 0.5401549123522218,
+      "grad_norm": 0.20841112732887268,
+      "learning_rate": 9.325485799701047e-05,
+      "loss": 4.1365,
+      "step": 7950
+    },
+    {
+      "epoch": 0.5404946324228835,
+      "grad_norm": 0.20364391803741455,
+      "learning_rate": 9.325061149612719e-05,
+      "loss": 3.8211,
+      "step": 7955
+    },
+    {
+      "epoch": 0.5408343524935453,
+      "grad_norm": 0.1856883019208908,
+      "learning_rate": 9.324636499524393e-05,
+      "loss": 4.102,
+      "step": 7960
+    },
+    {
+      "epoch": 0.5411740725642071,
+      "grad_norm": 0.22017249464988708,
+      "learning_rate": 9.324211849436066e-05,
+      "loss": 3.9317,
+      "step": 7965
+    },
+    {
+      "epoch": 0.5415137926348689,
+      "grad_norm": 0.2391664832830429,
+      "learning_rate": 9.323787199347737e-05,
+      "loss": 3.975,
+      "step": 7970
+    },
+    {
+      "epoch": 0.5418535127055306,
+      "grad_norm": 0.15632979571819305,
+      "learning_rate": 9.323362549259411e-05,
+      "loss": 4.1384,
+      "step": 7975
+    },
+    {
+      "epoch": 0.5421932327761925,
+      "grad_norm": 0.18496164679527283,
+      "learning_rate": 9.322937899171084e-05,
+      "loss": 4.1871,
+      "step": 7980
+    },
+    {
+      "epoch": 0.5425329528468542,
+      "grad_norm": 0.18015219271183014,
+      "learning_rate": 9.322513249082756e-05,
+      "loss": 3.9501,
+      "step": 7985
+    },
+    {
+      "epoch": 0.542872672917516,
+      "grad_norm": 0.19946110248565674,
+      "learning_rate": 9.32208859899443e-05,
+      "loss": 3.928,
+      "step": 7990
+    },
+    {
+      "epoch": 0.5432123929881777,
+      "grad_norm": 0.18122997879981995,
+      "learning_rate": 9.321663948906103e-05,
+      "loss": 4.0794,
+      "step": 7995
+    },
+    {
+      "epoch": 0.5435521130588395,
+      "grad_norm": 0.2898665964603424,
+      "learning_rate": 9.321239298817774e-05,
+      "loss": 4.1946,
+      "step": 8000
+    },
+    {
+      "epoch": 0.5438918331295013,
+      "grad_norm": 0.19237551093101501,
+      "learning_rate": 9.320814648729448e-05,
+      "loss": 3.8687,
+      "step": 8005
+    },
+    {
+      "epoch": 0.544231553200163,
+      "grad_norm": 0.18704599142074585,
+      "learning_rate": 9.32038999864112e-05,
+      "loss": 4.0483,
+      "step": 8010
+    },
+    {
+      "epoch": 0.5445712732708249,
+      "grad_norm": 0.16550405323505402,
+      "learning_rate": 9.319965348552792e-05,
+      "loss": 3.927,
+      "step": 8015
+    },
+    {
+      "epoch": 0.5449109933414866,
+      "grad_norm": 0.20201174914836884,
+      "learning_rate": 9.319540698464467e-05,
+      "loss": 4.1593,
+      "step": 8020
+    },
+    {
+      "epoch": 0.5452507134121484,
+      "grad_norm": 0.19555409252643585,
+      "learning_rate": 9.319116048376138e-05,
+      "loss": 4.2645,
+      "step": 8025
+    },
+    {
+      "epoch": 0.5455904334828101,
+      "grad_norm": 0.20266516506671906,
+      "learning_rate": 9.318691398287811e-05,
+      "loss": 4.165,
+      "step": 8030
+    },
+    {
+      "epoch": 0.545930153553472,
+      "grad_norm": 0.18069981038570404,
+      "learning_rate": 9.318266748199485e-05,
+      "loss": 4.0198,
+      "step": 8035
+    },
+    {
+      "epoch": 0.5462698736241337,
+      "grad_norm": 0.1856188029050827,
+      "learning_rate": 9.317842098111156e-05,
+      "loss": 3.9087,
+      "step": 8040
+    },
+    {
+      "epoch": 0.5466095936947954,
+      "grad_norm": 0.19204466044902802,
+      "learning_rate": 9.317417448022829e-05,
+      "loss": 4.0723,
+      "step": 8045
+    },
+    {
+      "epoch": 0.5469493137654573,
+      "grad_norm": 0.19473916292190552,
+      "learning_rate": 9.316992797934503e-05,
+      "loss": 3.9512,
+      "step": 8050
+    },
+    {
+      "epoch": 0.547289033836119,
+      "grad_norm": 0.23411086201667786,
+      "learning_rate": 9.316568147846175e-05,
+      "loss": 4.019,
+      "step": 8055
+    },
+    {
+      "epoch": 0.5476287539067808,
+      "grad_norm": 0.17357757687568665,
+      "learning_rate": 9.316143497757848e-05,
+      "loss": 3.9088,
+      "step": 8060
+    },
+    {
+      "epoch": 0.5479684739774426,
+      "grad_norm": 0.19055502116680145,
+      "learning_rate": 9.315718847669522e-05,
+      "loss": 3.9099,
+      "step": 8065
+    },
+    {
+      "epoch": 0.5483081940481044,
+      "grad_norm": 0.21026012301445007,
+      "learning_rate": 9.315294197581193e-05,
+      "loss": 4.0825,
+      "step": 8070
+    },
+    {
+      "epoch": 0.5486479141187661,
+      "grad_norm": 0.1714550107717514,
+      "learning_rate": 9.314869547492866e-05,
+      "loss": 4.1415,
+      "step": 8075
+    },
+    {
+      "epoch": 0.5489876341894279,
+      "grad_norm": 0.18270696699619293,
+      "learning_rate": 9.314444897404539e-05,
+      "loss": 4.0709,
+      "step": 8080
+    },
+    {
+      "epoch": 0.5493273542600897,
+      "grad_norm": 0.1924707442522049,
+      "learning_rate": 9.314020247316212e-05,
+      "loss": 4.0765,
+      "step": 8085
+    },
+    {
+      "epoch": 0.5496670743307515,
+      "grad_norm": 1.1790002584457397,
+      "learning_rate": 9.313595597227884e-05,
+      "loss": 4.0,
+      "step": 8090
+    },
+    {
+      "epoch": 0.5500067944014132,
+      "grad_norm": 0.2330000251531601,
+      "learning_rate": 9.313170947139557e-05,
+      "loss": 4.0367,
+      "step": 8095
+    },
+    {
+      "epoch": 0.550346514472075,
+      "grad_norm": 0.21324478089809418,
+      "learning_rate": 9.31274629705123e-05,
+      "loss": 3.9346,
+      "step": 8100
+    },
+    {
+      "epoch": 0.5506862345427368,
+      "grad_norm": 0.1989852339029312,
+      "learning_rate": 9.312321646962903e-05,
+      "loss": 3.883,
+      "step": 8105
+    },
+    {
+      "epoch": 0.5510259546133985,
+      "grad_norm": 0.2093362659215927,
+      "learning_rate": 9.311896996874576e-05,
+      "loss": 3.945,
+      "step": 8110
+    },
+    {
+      "epoch": 0.5513656746840603,
+      "grad_norm": 0.17465944588184357,
+      "learning_rate": 9.311472346786248e-05,
+      "loss": 3.9291,
+      "step": 8115
+    },
+    {
+      "epoch": 0.5517053947547221,
+      "grad_norm": 0.7979787588119507,
+      "learning_rate": 9.311047696697921e-05,
+      "loss": 3.9622,
+      "step": 8120
+    },
+    {
+      "epoch": 0.5520451148253839,
+      "grad_norm": 0.20573332905769348,
+      "learning_rate": 9.310623046609594e-05,
+      "loss": 4.0717,
+      "step": 8125
+    },
+    {
+      "epoch": 0.5523848348960456,
+      "grad_norm": 0.8882215023040771,
+      "learning_rate": 9.310198396521267e-05,
+      "loss": 4.045,
+      "step": 8130
+    },
+    {
+      "epoch": 0.5527245549667075,
+      "grad_norm": 0.1780032366514206,
+      "learning_rate": 9.30977374643294e-05,
+      "loss": 3.8284,
+      "step": 8135
+    },
+    {
+      "epoch": 0.5530642750373692,
+      "grad_norm": 0.19671432673931122,
+      "learning_rate": 9.309349096344612e-05,
+      "loss": 4.0132,
+      "step": 8140
+    },
+    {
+      "epoch": 0.553403995108031,
+      "grad_norm": 0.17597247660160065,
+      "learning_rate": 9.308924446256285e-05,
+      "loss": 4.0527,
+      "step": 8145
+    },
+    {
+      "epoch": 0.5537437151786928,
+      "grad_norm": 0.19633089005947113,
+      "learning_rate": 9.308499796167958e-05,
+      "loss": 3.8693,
+      "step": 8150
+    },
+    {
+      "epoch": 0.5540834352493546,
+      "grad_norm": 0.19133026897907257,
+      "learning_rate": 9.308075146079631e-05,
+      "loss": 3.9765,
+      "step": 8155
+    },
+    {
+      "epoch": 0.5544231553200163,
+      "grad_norm": 0.18762901425361633,
+      "learning_rate": 9.307650495991304e-05,
+      "loss": 3.9366,
+      "step": 8160
+    },
+    {
+      "epoch": 0.554762875390678,
+      "grad_norm": 0.21953530609607697,
+      "learning_rate": 9.307225845902976e-05,
+      "loss": 4.1145,
+      "step": 8165
+    },
+    {
+      "epoch": 0.5551025954613399,
+      "grad_norm": 0.1631409376859665,
+      "learning_rate": 9.306801195814649e-05,
+      "loss": 4.3299,
+      "step": 8170
+    },
+    {
+      "epoch": 0.5554423155320016,
+      "grad_norm": 0.1782941222190857,
+      "learning_rate": 9.306376545726322e-05,
+      "loss": 4.0921,
+      "step": 8175
+    },
+    {
+      "epoch": 0.5557820356026634,
+      "grad_norm": 0.18272286653518677,
+      "learning_rate": 9.305951895637995e-05,
+      "loss": 4.0609,
+      "step": 8180
+    },
+    {
+      "epoch": 0.5561217556733252,
+      "grad_norm": 0.5140985250473022,
+      "learning_rate": 9.305527245549668e-05,
+      "loss": 3.9297,
+      "step": 8185
+    },
+    {
+      "epoch": 0.556461475743987,
+      "grad_norm": 1.4922987222671509,
+      "learning_rate": 9.30510259546134e-05,
+      "loss": 4.0238,
+      "step": 8190
+    },
+    {
+      "epoch": 0.5568011958146487,
+      "grad_norm": 3.8372085094451904,
+      "learning_rate": 9.304677945373013e-05,
+      "loss": 4.1705,
+      "step": 8195
+    },
+    {
+      "epoch": 0.5571409158853105,
+      "grad_norm": 0.1523323357105255,
+      "learning_rate": 9.304253295284686e-05,
+      "loss": 4.0109,
+      "step": 8200
+    },
+    {
+      "epoch": 0.5574806359559723,
+      "grad_norm": 0.2263563573360443,
+      "learning_rate": 9.303828645196359e-05,
+      "loss": 4.0453,
+      "step": 8205
+    },
+    {
+      "epoch": 0.557820356026634,
+      "grad_norm": 0.20858174562454224,
+      "learning_rate": 9.30340399510803e-05,
+      "loss": 4.0197,
+      "step": 8210
+    },
+    {
+      "epoch": 0.5581600760972958,
+      "grad_norm": 0.2111169844865799,
+      "learning_rate": 9.302979345019704e-05,
+      "loss": 3.9775,
+      "step": 8215
+    },
+    {
+      "epoch": 0.5584997961679576,
+      "grad_norm": 0.16785529255867004,
+      "learning_rate": 9.302554694931377e-05,
+      "loss": 4.1782,
+      "step": 8220
+    },
+    {
+      "epoch": 0.5588395162386194,
+      "grad_norm": 0.21160061657428741,
+      "learning_rate": 9.302130044843049e-05,
+      "loss": 3.7809,
+      "step": 8225
+    },
+    {
+      "epoch": 0.5591792363092811,
+      "grad_norm": 0.34842145442962646,
+      "learning_rate": 9.301705394754723e-05,
+      "loss": 4.0454,
+      "step": 8230
+    },
+    {
+      "epoch": 0.559518956379943,
+      "grad_norm": 0.22332549095153809,
+      "learning_rate": 9.301280744666396e-05,
+      "loss": 3.9447,
+      "step": 8235
+    },
+    {
+      "epoch": 0.5598586764506047,
+      "grad_norm": 0.33598944544792175,
+      "learning_rate": 9.300856094578067e-05,
+      "loss": 4.2742,
+      "step": 8240
+    },
+    {
+      "epoch": 0.5601983965212665,
+      "grad_norm": 0.18852943181991577,
+      "learning_rate": 9.300431444489741e-05,
+      "loss": 3.9494,
+      "step": 8245
+    },
+    {
+      "epoch": 0.5605381165919282,
+      "grad_norm": 0.1902894675731659,
+      "learning_rate": 9.300006794401414e-05,
+      "loss": 4.0114,
+      "step": 8250
+    },
+    {
+      "epoch": 0.5608778366625901,
+      "grad_norm": 0.3319913446903229,
+      "learning_rate": 9.299582144313086e-05,
+      "loss": 4.1077,
+      "step": 8255
+    },
+    {
+      "epoch": 0.5612175567332518,
+      "grad_norm": 1.5811398029327393,
+      "learning_rate": 9.29915749422476e-05,
+      "loss": 3.9531,
+      "step": 8260
+    },
+    {
+      "epoch": 0.5615572768039135,
+      "grad_norm": 0.18648011982440948,
+      "learning_rate": 9.298732844136433e-05,
+      "loss": 4.0406,
+      "step": 8265
+    },
+    {
+      "epoch": 0.5618969968745754,
+      "grad_norm": 0.20678003132343292,
+      "learning_rate": 9.298308194048104e-05,
+      "loss": 4.0141,
+      "step": 8270
+    },
+    {
+      "epoch": 0.5622367169452371,
+      "grad_norm": 0.284509539604187,
+      "learning_rate": 9.297883543959778e-05,
+      "loss": 3.7511,
+      "step": 8275
+    },
+    {
+      "epoch": 0.5625764370158989,
+      "grad_norm": 0.23593567311763763,
+      "learning_rate": 9.29745889387145e-05,
+      "loss": 3.9993,
+      "step": 8280
+    },
+    {
+      "epoch": 0.5629161570865606,
+      "grad_norm": 0.221805140376091,
+      "learning_rate": 9.297034243783122e-05,
+      "loss": 4.0023,
+      "step": 8285
+    },
+    {
+      "epoch": 0.5632558771572225,
+      "grad_norm": 0.1965785026550293,
+      "learning_rate": 9.296609593694797e-05,
+      "loss": 4.0849,
+      "step": 8290
+    },
+    {
+      "epoch": 0.5635955972278842,
+      "grad_norm": 0.3942621648311615,
+      "learning_rate": 9.296184943606468e-05,
+      "loss": 4.0986,
+      "step": 8295
+    },
+    {
+      "epoch": 0.563935317298546,
+      "grad_norm": 0.14950452744960785,
+      "learning_rate": 9.295760293518142e-05,
+      "loss": 4.0598,
+      "step": 8300
+    },
+    {
+      "epoch": 0.5642750373692078,
+      "grad_norm": 0.19888624548912048,
+      "learning_rate": 9.295335643429815e-05,
+      "loss": 4.1756,
+      "step": 8305
+    },
+    {
+      "epoch": 0.5646147574398696,
+      "grad_norm": 0.18374282121658325,
+      "learning_rate": 9.294910993341486e-05,
+      "loss": 4.0213,
+      "step": 8310
+    },
+    {
+      "epoch": 0.5649544775105313,
+      "grad_norm": 0.19329077005386353,
+      "learning_rate": 9.29448634325316e-05,
+      "loss": 3.8402,
+      "step": 8315
+    },
+    {
+      "epoch": 0.5652941975811931,
+      "grad_norm": 0.4161980152130127,
+      "learning_rate": 9.294061693164833e-05,
+      "loss": 3.8012,
+      "step": 8320
+    },
+    {
+      "epoch": 0.5656339176518549,
+      "grad_norm": 0.20671139657497406,
+      "learning_rate": 9.293637043076505e-05,
+      "loss": 4.335,
+      "step": 8325
+    },
+    {
+      "epoch": 0.5659736377225166,
+      "grad_norm": 0.20330487191677094,
+      "learning_rate": 9.293212392988179e-05,
+      "loss": 4.0325,
+      "step": 8330
+    },
+    {
+      "epoch": 0.5663133577931784,
+      "grad_norm": 0.19175422191619873,
+      "learning_rate": 9.292787742899852e-05,
+      "loss": 4.1178,
+      "step": 8335
+    },
+    {
+      "epoch": 0.5666530778638402,
+      "grad_norm": 0.22935859858989716,
+      "learning_rate": 9.292363092811523e-05,
+      "loss": 4.1405,
+      "step": 8340
+    },
+    {
+      "epoch": 0.566992797934502,
+      "grad_norm": 0.17749212682247162,
+      "learning_rate": 9.291938442723197e-05,
+      "loss": 3.9575,
+      "step": 8345
+    },
+    {
+      "epoch": 0.5673325180051637,
+      "grad_norm": 1.0764920711517334,
+      "learning_rate": 9.291513792634869e-05,
+      "loss": 4.3775,
+      "step": 8350
+    },
+    {
+      "epoch": 0.5676722380758256,
+      "grad_norm": 0.15739497542381287,
+      "learning_rate": 9.291089142546542e-05,
+      "loss": 4.1305,
+      "step": 8355
+    },
+    {
+      "epoch": 0.5680119581464873,
+      "grad_norm": 0.19351720809936523,
+      "learning_rate": 9.290664492458216e-05,
+      "loss": 3.8318,
+      "step": 8360
+    },
+    {
+      "epoch": 0.568351678217149,
+      "grad_norm": 0.16042912006378174,
+      "learning_rate": 9.290239842369887e-05,
+      "loss": 4.1028,
+      "step": 8365
+    },
+    {
+      "epoch": 0.5686913982878108,
+      "grad_norm": 0.1768733263015747,
+      "learning_rate": 9.28981519228156e-05,
+      "loss": 3.8895,
+      "step": 8370
+    },
+    {
+      "epoch": 0.5690311183584726,
+      "grad_norm": 0.15216295421123505,
+      "learning_rate": 9.289390542193234e-05,
+      "loss": 4.1829,
+      "step": 8375
+    },
+    {
+      "epoch": 0.5693708384291344,
+      "grad_norm": 0.20485453307628632,
+      "learning_rate": 9.288965892104906e-05,
+      "loss": 4.1128,
+      "step": 8380
+    },
+    {
+      "epoch": 0.5697105584997961,
+      "grad_norm": 0.2530359625816345,
+      "learning_rate": 9.288541242016578e-05,
+      "loss": 4.0308,
+      "step": 8385
+    },
+    {
+      "epoch": 0.570050278570458,
+      "grad_norm": 0.24390393495559692,
+      "learning_rate": 9.288116591928253e-05,
+      "loss": 3.9455,
+      "step": 8390
+    },
+    {
+      "epoch": 0.5703899986411197,
+      "grad_norm": 0.15447057783603668,
+      "learning_rate": 9.287691941839924e-05,
+      "loss": 4.2323,
+      "step": 8395
+    },
+    {
+      "epoch": 0.5707297187117815,
+      "grad_norm": 0.262494295835495,
+      "learning_rate": 9.287267291751597e-05,
+      "loss": 3.8537,
+      "step": 8400
+    },
+    {
+      "epoch": 0.5710694387824433,
+      "grad_norm": 0.2734646499156952,
+      "learning_rate": 9.286842641663271e-05,
+      "loss": 4.0975,
+      "step": 8405
+    },
+    {
+      "epoch": 0.5714091588531051,
+      "grad_norm": 0.3060397803783417,
+      "learning_rate": 9.286417991574942e-05,
+      "loss": 4.1002,
+      "step": 8410
+    },
+    {
+      "epoch": 0.5717488789237668,
+      "grad_norm": 0.194105863571167,
+      "learning_rate": 9.285993341486615e-05,
+      "loss": 3.7785,
+      "step": 8415
+    },
+    {
+      "epoch": 0.5720885989944285,
+      "grad_norm": 0.19388654828071594,
+      "learning_rate": 9.28556869139829e-05,
+      "loss": 4.0577,
+      "step": 8420
+    },
+    {
+      "epoch": 0.5724283190650904,
+      "grad_norm": 0.4632340669631958,
+      "learning_rate": 9.285144041309961e-05,
+      "loss": 4.0301,
+      "step": 8425
+    },
+    {
+      "epoch": 0.5727680391357521,
+      "grad_norm": 0.23817621171474457,
+      "learning_rate": 9.284719391221634e-05,
+      "loss": 3.9506,
+      "step": 8430
+    },
+    {
+      "epoch": 0.5731077592064139,
+      "grad_norm": 0.26604166626930237,
+      "learning_rate": 9.284294741133306e-05,
+      "loss": 3.7005,
+      "step": 8435
+    },
+    {
+      "epoch": 0.5734474792770757,
+      "grad_norm": 0.21348132193088531,
+      "learning_rate": 9.283870091044979e-05,
+      "loss": 4.0274,
+      "step": 8440
+    },
+    {
+      "epoch": 0.5737871993477375,
+      "grad_norm": 0.3157108724117279,
+      "learning_rate": 9.283445440956652e-05,
+      "loss": 3.8789,
+      "step": 8445
+    },
+    {
+      "epoch": 0.5741269194183992,
+      "grad_norm": 0.6310774087905884,
+      "learning_rate": 9.283020790868325e-05,
+      "loss": 4.0599,
+      "step": 8450
+    },
+    {
+      "epoch": 0.574466639489061,
+      "grad_norm": 0.18738152086734772,
+      "learning_rate": 9.282596140779998e-05,
+      "loss": 3.8024,
+      "step": 8455
+    },
+    {
+      "epoch": 0.5748063595597228,
+      "grad_norm": 0.3210128843784332,
+      "learning_rate": 9.28217149069167e-05,
+      "loss": 4.1008,
+      "step": 8460
+    },
+    {
+      "epoch": 0.5751460796303846,
+      "grad_norm": 0.180195152759552,
+      "learning_rate": 9.281746840603343e-05,
+      "loss": 4.0769,
+      "step": 8465
+    },
+    {
+      "epoch": 0.5754857997010463,
+      "grad_norm": 0.17868779599666595,
+      "learning_rate": 9.281322190515016e-05,
+      "loss": 3.9685,
+      "step": 8470
+    },
+    {
+      "epoch": 0.5758255197717081,
+      "grad_norm": 0.5827326774597168,
+      "learning_rate": 9.280897540426689e-05,
+      "loss": 3.9109,
+      "step": 8475
+    },
+    {
+      "epoch": 0.5761652398423699,
+      "grad_norm": 0.3124859631061554,
+      "learning_rate": 9.280472890338362e-05,
+      "loss": 3.8527,
+      "step": 8480
+    },
+    {
+      "epoch": 0.5765049599130316,
+      "grad_norm": 0.42251700162887573,
+      "learning_rate": 9.280048240250034e-05,
+      "loss": 4.1552,
+      "step": 8485
+    },
+    {
+      "epoch": 0.5768446799836935,
+      "grad_norm": 0.1950671374797821,
+      "learning_rate": 9.279623590161707e-05,
+      "loss": 3.836,
+      "step": 8490
+    },
+    {
+      "epoch": 0.5771844000543552,
+      "grad_norm": 0.8381164073944092,
+      "learning_rate": 9.27919894007338e-05,
+      "loss": 4.0999,
+      "step": 8495
+    },
+    {
+      "epoch": 0.577524120125017,
+      "grad_norm": 0.15643952786922455,
+      "learning_rate": 9.278774289985053e-05,
+      "loss": 3.9804,
+      "step": 8500
+    },
+    {
+      "epoch": 0.5778638401956787,
+      "grad_norm": 0.21564146876335144,
+      "learning_rate": 9.278349639896726e-05,
+      "loss": 3.9151,
+      "step": 8505
+    },
+    {
+      "epoch": 0.5782035602663406,
+      "grad_norm": 0.21207468211650848,
+      "learning_rate": 9.277924989808398e-05,
+      "loss": 3.9481,
+      "step": 8510
+    },
+    {
+      "epoch": 0.5785432803370023,
+      "grad_norm": 0.19761881232261658,
+      "learning_rate": 9.277500339720071e-05,
+      "loss": 4.0216,
+      "step": 8515
+    },
+    {
+      "epoch": 0.578883000407664,
+      "grad_norm": 0.18729913234710693,
+      "learning_rate": 9.277075689631744e-05,
+      "loss": 3.9112,
+      "step": 8520
+    },
+    {
+      "epoch": 0.5792227204783259,
+      "grad_norm": 0.13949252665042877,
+      "learning_rate": 9.276651039543417e-05,
+      "loss": 3.928,
+      "step": 8525
+    },
+    {
+      "epoch": 0.5795624405489876,
+      "grad_norm": 0.1481368988752365,
+      "learning_rate": 9.27622638945509e-05,
+      "loss": 4.1387,
+      "step": 8530
+    },
+    {
+      "epoch": 0.5799021606196494,
+      "grad_norm": 0.2295740246772766,
+      "learning_rate": 9.275801739366762e-05,
+      "loss": 4.0267,
+      "step": 8535
+    },
+    {
+      "epoch": 0.5802418806903111,
+      "grad_norm": 0.22337546944618225,
+      "learning_rate": 9.275377089278435e-05,
+      "loss": 3.927,
+      "step": 8540
+    },
+    {
+      "epoch": 0.580581600760973,
+      "grad_norm": 0.41308972239494324,
+      "learning_rate": 9.274952439190108e-05,
+      "loss": 4.1804,
+      "step": 8545
+    },
+    {
+      "epoch": 0.5809213208316347,
+      "grad_norm": 0.15988604724407196,
+      "learning_rate": 9.27452778910178e-05,
+      "loss": 3.8497,
+      "step": 8550
+    },
+    {
+      "epoch": 0.5812610409022965,
+      "grad_norm": 0.17153027653694153,
+      "learning_rate": 9.274103139013454e-05,
+      "loss": 4.1164,
+      "step": 8555
+    },
+    {
+      "epoch": 0.5816007609729583,
+      "grad_norm": 0.2415970116853714,
+      "learning_rate": 9.273678488925126e-05,
+      "loss": 4.1781,
+      "step": 8560
+    },
+    {
+      "epoch": 0.5819404810436201,
+      "grad_norm": 2.764481544494629,
+      "learning_rate": 9.273253838836798e-05,
+      "loss": 3.9708,
+      "step": 8565
+    },
+    {
+      "epoch": 0.5822802011142818,
+      "grad_norm": 0.1980726718902588,
+      "learning_rate": 9.272829188748472e-05,
+      "loss": 4.0011,
+      "step": 8570
+    },
+    {
+      "epoch": 0.5826199211849437,
+      "grad_norm": 0.23337212204933167,
+      "learning_rate": 9.272404538660145e-05,
+      "loss": 3.9122,
+      "step": 8575
+    },
+    {
+      "epoch": 0.5829596412556054,
+      "grad_norm": 0.1616441160440445,
+      "learning_rate": 9.271979888571816e-05,
+      "loss": 3.8085,
+      "step": 8580
+    },
+    {
+      "epoch": 0.5832993613262671,
+      "grad_norm": 0.19544613361358643,
+      "learning_rate": 9.27155523848349e-05,
+      "loss": 4.0634,
+      "step": 8585
+    },
+    {
+      "epoch": 0.5836390813969289,
+      "grad_norm": 0.19798190891742706,
+      "learning_rate": 9.271130588395163e-05,
+      "loss": 3.9989,
+      "step": 8590
+    },
+    {
+      "epoch": 0.5839788014675907,
+      "grad_norm": 0.1608157753944397,
+      "learning_rate": 9.270705938306835e-05,
+      "loss": 4.3514,
+      "step": 8595
+    },
+    {
+      "epoch": 0.5843185215382525,
+      "grad_norm": 0.23190398514270782,
+      "learning_rate": 9.270281288218509e-05,
+      "loss": 3.9989,
+      "step": 8600
+    },
+    {
+      "epoch": 0.5846582416089142,
+      "grad_norm": 0.694850742816925,
+      "learning_rate": 9.269856638130182e-05,
+      "loss": 4.3087,
+      "step": 8605
+    },
+    {
+      "epoch": 0.5849979616795761,
+      "grad_norm": 0.2711631655693054,
+      "learning_rate": 9.269431988041853e-05,
+      "loss": 3.9979,
+      "step": 8610
+    },
+    {
+      "epoch": 0.5853376817502378,
+      "grad_norm": 0.17717806994915009,
+      "learning_rate": 9.269007337953527e-05,
+      "loss": 3.9771,
+      "step": 8615
+    },
+    {
+      "epoch": 0.5856774018208996,
+      "grad_norm": 0.258022665977478,
+      "learning_rate": 9.2685826878652e-05,
+      "loss": 3.9518,
+      "step": 8620
+    },
+    {
+      "epoch": 0.5860171218915613,
+      "grad_norm": 1.0233123302459717,
+      "learning_rate": 9.268158037776871e-05,
+      "loss": 4.0393,
+      "step": 8625
+    },
+    {
+      "epoch": 0.5863568419622232,
+      "grad_norm": 0.19119100272655487,
+      "learning_rate": 9.267733387688546e-05,
+      "loss": 4.2572,
+      "step": 8630
+    },
+    {
+      "epoch": 0.5866965620328849,
+      "grad_norm": 0.16495579481124878,
+      "learning_rate": 9.267308737600217e-05,
+      "loss": 4.0212,
+      "step": 8635
+    },
+    {
+      "epoch": 0.5870362821035466,
+      "grad_norm": 0.3087138831615448,
+      "learning_rate": 9.266884087511891e-05,
+      "loss": 3.6527,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5873760021742085,
+      "grad_norm": 0.16957888007164001,
+      "learning_rate": 9.266459437423564e-05,
+      "loss": 4.2809,
+      "step": 8645
+    },
+    {
+      "epoch": 0.5877157222448702,
+      "grad_norm": 0.15367551147937775,
+      "learning_rate": 9.266034787335235e-05,
+      "loss": 4.2469,
+      "step": 8650
+    },
+    {
+      "epoch": 0.588055442315532,
+      "grad_norm": 0.2356184720993042,
+      "learning_rate": 9.26561013724691e-05,
+      "loss": 3.8713,
+      "step": 8655
+    },
+    {
+      "epoch": 0.5883951623861938,
+      "grad_norm": 6.177441120147705,
+      "learning_rate": 9.265185487158582e-05,
+      "loss": 3.8173,
+      "step": 8660
+    },
+    {
+      "epoch": 0.5887348824568556,
+      "grad_norm": 0.15291577577590942,
+      "learning_rate": 9.264760837070254e-05,
+      "loss": 4.0039,
+      "step": 8665
+    },
+    {
+      "epoch": 0.5890746025275173,
+      "grad_norm": 0.1881755292415619,
+      "learning_rate": 9.264336186981928e-05,
+      "loss": 4.1521,
+      "step": 8670
+    },
+    {
+      "epoch": 0.589414322598179,
+      "grad_norm": 0.28705185651779175,
+      "learning_rate": 9.263911536893601e-05,
+      "loss": 4.1136,
+      "step": 8675
+    },
+    {
+      "epoch": 0.5897540426688409,
+      "grad_norm": 0.8784993886947632,
+      "learning_rate": 9.263486886805272e-05,
+      "loss": 4.0763,
+      "step": 8680
+    },
+    {
+      "epoch": 0.5900937627395026,
+      "grad_norm": 0.5077646970748901,
+      "learning_rate": 9.263062236716946e-05,
+      "loss": 3.818,
+      "step": 8685
+    },
+    {
+      "epoch": 0.5904334828101644,
+      "grad_norm": 0.17957673966884613,
+      "learning_rate": 9.262637586628619e-05,
+      "loss": 3.8784,
+      "step": 8690
+    },
+    {
+      "epoch": 0.5907732028808262,
+      "grad_norm": 0.2007267326116562,
+      "learning_rate": 9.26221293654029e-05,
+      "loss": 4.0291,
+      "step": 8695
+    },
+    {
+      "epoch": 0.591112922951488,
+      "grad_norm": 0.3630281686782837,
+      "learning_rate": 9.261788286451965e-05,
+      "loss": 4.0589,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5914526430221497,
+      "grad_norm": 0.33113589882850647,
+      "learning_rate": 9.261363636363636e-05,
+      "loss": 4.0724,
+      "step": 8705
+    },
+    {
+      "epoch": 0.5917923630928115,
+      "grad_norm": 0.9370392560958862,
+      "learning_rate": 9.260938986275309e-05,
+      "loss": 3.9571,
+      "step": 8710
+    },
+    {
+      "epoch": 0.5921320831634733,
+      "grad_norm": 0.7348127365112305,
+      "learning_rate": 9.260514336186983e-05,
+      "loss": 3.8521,
+      "step": 8715
+    },
+    {
+      "epoch": 0.5924718032341351,
+      "grad_norm": 0.20114554464817047,
+      "learning_rate": 9.260089686098655e-05,
+      "loss": 4.0328,
+      "step": 8720
+    },
+    {
+      "epoch": 0.5928115233047968,
+      "grad_norm": 0.375806987285614,
+      "learning_rate": 9.259665036010327e-05,
+      "loss": 4.045,
+      "step": 8725
+    },
+    {
+      "epoch": 0.5931512433754587,
+      "grad_norm": 0.17481490969657898,
+      "learning_rate": 9.259240385922002e-05,
+      "loss": 4.1616,
+      "step": 8730
+    },
+    {
+      "epoch": 0.5934909634461204,
+      "grad_norm": 0.3022591471672058,
+      "learning_rate": 9.258815735833673e-05,
+      "loss": 4.0886,
+      "step": 8735
+    },
+    {
+      "epoch": 0.5938306835167821,
+      "grad_norm": 0.17564785480499268,
+      "learning_rate": 9.258391085745346e-05,
+      "loss": 3.9194,
+      "step": 8740
+    },
+    {
+      "epoch": 0.594170403587444,
+      "grad_norm": 0.19886420667171478,
+      "learning_rate": 9.25796643565702e-05,
+      "loss": 3.9606,
+      "step": 8745
+    },
+    {
+      "epoch": 0.5945101236581057,
+      "grad_norm": 0.18436846137046814,
+      "learning_rate": 9.257541785568691e-05,
+      "loss": 4.0952,
+      "step": 8750
+    },
+    {
+      "epoch": 0.5948498437287675,
+      "grad_norm": 0.39597442746162415,
+      "learning_rate": 9.257117135480364e-05,
+      "loss": 4.0044,
+      "step": 8755
+    },
+    {
+      "epoch": 0.5951895637994292,
+      "grad_norm": 0.2468959391117096,
+      "learning_rate": 9.256692485392038e-05,
+      "loss": 4.2667,
+      "step": 8760
+    },
+    {
+      "epoch": 0.5955292838700911,
+      "grad_norm": 0.21869252622127533,
+      "learning_rate": 9.25626783530371e-05,
+      "loss": 3.862,
+      "step": 8765
+    },
+    {
+      "epoch": 0.5958690039407528,
+      "grad_norm": 0.18693305552005768,
+      "learning_rate": 9.255843185215383e-05,
+      "loss": 4.1253,
+      "step": 8770
+    },
+    {
+      "epoch": 0.5962087240114146,
+      "grad_norm": 0.2016768604516983,
+      "learning_rate": 9.255418535127055e-05,
+      "loss": 4.0373,
+      "step": 8775
+    },
+    {
+      "epoch": 0.5965484440820764,
+      "grad_norm": 0.20323170721530914,
+      "learning_rate": 9.254993885038728e-05,
+      "loss": 3.9236,
+      "step": 8780
+    },
+    {
+      "epoch": 0.5968881641527382,
+      "grad_norm": 0.17881441116333008,
+      "learning_rate": 9.254569234950401e-05,
+      "loss": 3.9083,
+      "step": 8785
+    },
+    {
+      "epoch": 0.5972278842233999,
+      "grad_norm": 0.22243642807006836,
+      "learning_rate": 9.254144584862074e-05,
+      "loss": 4.0484,
+      "step": 8790
+    },
+    {
+      "epoch": 0.5975676042940616,
+      "grad_norm": 0.160291388630867,
+      "learning_rate": 9.253719934773747e-05,
+      "loss": 4.0718,
+      "step": 8795
+    },
+    {
+      "epoch": 0.5979073243647235,
+      "grad_norm": 0.18238534033298492,
+      "learning_rate": 9.25329528468542e-05,
+      "loss": 3.8954,
+      "step": 8800
+    },
+    {
+      "epoch": 0.5982470444353852,
+      "grad_norm": 0.2048415243625641,
+      "learning_rate": 9.252870634597092e-05,
+      "loss": 3.9912,
+      "step": 8805
+    },
+    {
+      "epoch": 0.598586764506047,
+      "grad_norm": 0.17319513857364655,
+      "learning_rate": 9.252445984508765e-05,
+      "loss": 4.0986,
+      "step": 8810
+    },
+    {
+      "epoch": 0.5989264845767088,
+      "grad_norm": 0.24065633118152618,
+      "learning_rate": 9.252021334420438e-05,
+      "loss": 4.1444,
+      "step": 8815
+    },
+    {
+      "epoch": 0.5992662046473706,
+      "grad_norm": 0.20234987139701843,
+      "learning_rate": 9.251596684332111e-05,
+      "loss": 3.8242,
+      "step": 8820
+    },
+    {
+      "epoch": 0.5996059247180323,
+      "grad_norm": 0.21660032868385315,
+      "learning_rate": 9.251172034243783e-05,
+      "loss": 3.8864,
+      "step": 8825
+    },
+    {
+      "epoch": 0.5999456447886942,
+      "grad_norm": 0.17871896922588348,
+      "learning_rate": 9.250747384155456e-05,
+      "loss": 3.8432,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6002853648593559,
+      "grad_norm": 0.16359539330005646,
+      "learning_rate": 9.250322734067129e-05,
+      "loss": 4.0026,
+      "step": 8835
+    },
+    {
+      "epoch": 0.6006250849300176,
+      "grad_norm": 0.2683166265487671,
+      "learning_rate": 9.249898083978802e-05,
+      "loss": 4.1686,
+      "step": 8840
+    },
+    {
+      "epoch": 0.6009648050006794,
+      "grad_norm": 0.16088762879371643,
+      "learning_rate": 9.249473433890475e-05,
+      "loss": 3.9448,
+      "step": 8845
+    },
+    {
+      "epoch": 0.6013045250713412,
+      "grad_norm": 0.21767516434192657,
+      "learning_rate": 9.249048783802147e-05,
+      "loss": 4.1642,
+      "step": 8850
+    },
+    {
+      "epoch": 0.601644245142003,
+      "grad_norm": 1.050563097000122,
+      "learning_rate": 9.24862413371382e-05,
+      "loss": 4.0446,
+      "step": 8855
+    },
+    {
+      "epoch": 0.6019839652126647,
+      "grad_norm": 0.16180701553821564,
+      "learning_rate": 9.248199483625493e-05,
+      "loss": 4.014,
+      "step": 8860
+    },
+    {
+      "epoch": 0.6023236852833266,
+      "grad_norm": 0.1821652054786682,
+      "learning_rate": 9.247774833537166e-05,
+      "loss": 4.0171,
+      "step": 8865
+    },
+    {
+      "epoch": 0.6026634053539883,
+      "grad_norm": 1.0722553730010986,
+      "learning_rate": 9.247350183448839e-05,
+      "loss": 3.8606,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6030031254246501,
+      "grad_norm": 0.22805406153202057,
+      "learning_rate": 9.246925533360511e-05,
+      "loss": 3.993,
+      "step": 8875
+    },
+    {
+      "epoch": 0.6033428454953118,
+      "grad_norm": 0.2157077044248581,
+      "learning_rate": 9.246500883272184e-05,
+      "loss": 3.7777,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6036825655659737,
+      "grad_norm": 0.18143793940544128,
+      "learning_rate": 9.246076233183857e-05,
+      "loss": 4.127,
+      "step": 8885
+    },
+    {
+      "epoch": 0.6040222856366354,
+      "grad_norm": 0.18271377682685852,
+      "learning_rate": 9.24565158309553e-05,
+      "loss": 4.0305,
+      "step": 8890
+    },
+    {
+      "epoch": 0.6043620057072971,
+      "grad_norm": 0.2080451250076294,
+      "learning_rate": 9.245226933007203e-05,
+      "loss": 3.8488,
+      "step": 8895
+    },
+    {
+      "epoch": 0.604701725777959,
+      "grad_norm": 1.494086742401123,
+      "learning_rate": 9.244802282918875e-05,
+      "loss": 4.0619,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6050414458486207,
+      "grad_norm": 0.2166607528924942,
+      "learning_rate": 9.244377632830547e-05,
+      "loss": 3.8448,
+      "step": 8905
+    },
+    {
+      "epoch": 0.6053811659192825,
+      "grad_norm": 0.24948440492153168,
+      "learning_rate": 9.243952982742221e-05,
+      "loss": 3.9839,
+      "step": 8910
+    },
+    {
+      "epoch": 0.6057208859899443,
+      "grad_norm": 0.33411622047424316,
+      "learning_rate": 9.243528332653894e-05,
+      "loss": 4.0442,
+      "step": 8915
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.17864017188549042,
+      "learning_rate": 9.243103682565565e-05,
+      "loss": 3.996,
+      "step": 8920
+    },
+    {
+      "epoch": 0.6064003261312678,
+      "grad_norm": 0.18109479546546936,
+      "learning_rate": 9.24267903247724e-05,
+      "loss": 4.1308,
+      "step": 8925
+    },
+    {
+      "epoch": 0.6067400462019296,
+      "grad_norm": 0.21822641789913177,
+      "learning_rate": 9.242254382388912e-05,
+      "loss": 4.1508,
+      "step": 8930
+    },
+    {
+      "epoch": 0.6070797662725914,
+      "grad_norm": 0.37898629903793335,
+      "learning_rate": 9.241829732300584e-05,
+      "loss": 4.1584,
+      "step": 8935
+    },
+    {
+      "epoch": 0.6074194863432532,
+      "grad_norm": 0.30412939190864563,
+      "learning_rate": 9.241405082212258e-05,
+      "loss": 3.948,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6077592064139149,
+      "grad_norm": 0.20068703591823578,
+      "learning_rate": 9.240980432123931e-05,
+      "loss": 4.1206,
+      "step": 8945
+    },
+    {
+      "epoch": 0.6080989264845768,
+      "grad_norm": 0.25400644540786743,
+      "learning_rate": 9.240555782035602e-05,
+      "loss": 4.3133,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6084386465552385,
+      "grad_norm": 0.19139589369297028,
+      "learning_rate": 9.240131131947276e-05,
+      "loss": 3.9386,
+      "step": 8955
+    },
+    {
+      "epoch": 0.6087783666259002,
+      "grad_norm": 0.16936640441417694,
+      "learning_rate": 9.239706481858949e-05,
+      "loss": 4.0571,
+      "step": 8960
+    },
+    {
+      "epoch": 0.609118086696562,
+      "grad_norm": 0.20352940261363983,
+      "learning_rate": 9.23928183177062e-05,
+      "loss": 3.7657,
+      "step": 8965
+    },
+    {
+      "epoch": 0.6094578067672238,
+      "grad_norm": 0.1573779433965683,
+      "learning_rate": 9.238857181682295e-05,
+      "loss": 4.1298,
+      "step": 8970
+    },
+    {
+      "epoch": 0.6097975268378856,
+      "grad_norm": 0.20279887318611145,
+      "learning_rate": 9.238432531593966e-05,
+      "loss": 3.9271,
+      "step": 8975
+    },
+    {
+      "epoch": 0.6101372469085473,
+      "grad_norm": 16.52877426147461,
+      "learning_rate": 9.23800788150564e-05,
+      "loss": 3.8319,
+      "step": 8980
+    },
+    {
+      "epoch": 0.6104769669792092,
+      "grad_norm": 0.17606152594089508,
+      "learning_rate": 9.237583231417313e-05,
+      "loss": 3.7453,
+      "step": 8985
+    },
+    {
+      "epoch": 0.6108166870498709,
+      "grad_norm": 0.23678942024707794,
+      "learning_rate": 9.237158581328985e-05,
+      "loss": 3.9744,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6111564071205327,
+      "grad_norm": 0.21576398611068726,
+      "learning_rate": 9.236733931240659e-05,
+      "loss": 4.1691,
+      "step": 8995
+    },
+    {
+      "epoch": 0.6114961271911945,
+      "grad_norm": 0.2256469577550888,
+      "learning_rate": 9.236309281152332e-05,
+      "loss": 3.9419,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6118358472618562,
+      "grad_norm": 0.23562292754650116,
+      "learning_rate": 9.235884631064003e-05,
+      "loss": 4.1049,
+      "step": 9005
+    },
+    {
+      "epoch": 0.612175567332518,
+      "grad_norm": 0.2894531190395355,
+      "learning_rate": 9.235459980975677e-05,
+      "loss": 3.9535,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6125152874031797,
+      "grad_norm": 0.22967609763145447,
+      "learning_rate": 9.23503533088735e-05,
+      "loss": 3.9537,
+      "step": 9015
+    },
+    {
+      "epoch": 0.6128550074738416,
+      "grad_norm": 1.3449379205703735,
+      "learning_rate": 9.234610680799021e-05,
+      "loss": 4.08,
+      "step": 9020
+    },
+    {
+      "epoch": 0.6131947275445033,
+      "grad_norm": 0.2513696849346161,
+      "learning_rate": 9.234186030710696e-05,
+      "loss": 3.8626,
+      "step": 9025
+    },
+    {
+      "epoch": 0.6135344476151651,
+      "grad_norm": 0.19834820926189423,
+      "learning_rate": 9.233761380622368e-05,
+      "loss": 3.9427,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6138741676858269,
+      "grad_norm": 0.2118111550807953,
+      "learning_rate": 9.23333673053404e-05,
+      "loss": 3.8768,
+      "step": 9035
+    },
+    {
+      "epoch": 0.6142138877564887,
+      "grad_norm": 0.15969218313694,
+      "learning_rate": 9.232912080445714e-05,
+      "loss": 4.1679,
+      "step": 9040
+    },
+    {
+      "epoch": 0.6145536078271504,
+      "grad_norm": 0.184014692902565,
+      "learning_rate": 9.232487430357387e-05,
+      "loss": 4.1939,
+      "step": 9045
+    },
+    {
+      "epoch": 0.6148933278978121,
+      "grad_norm": 0.17408449947834015,
+      "learning_rate": 9.232062780269058e-05,
+      "loss": 4.0627,
+      "step": 9050
+    },
+    {
+      "epoch": 0.615233047968474,
+      "grad_norm": 0.3706364929676056,
+      "learning_rate": 9.231638130180732e-05,
+      "loss": 4.1244,
+      "step": 9055
+    },
+    {
+      "epoch": 0.6155727680391357,
+      "grad_norm": 0.2606452703475952,
+      "learning_rate": 9.231213480092404e-05,
+      "loss": 4.0637,
+      "step": 9060
+    },
+    {
+      "epoch": 0.6159124881097975,
+      "grad_norm": 0.2730484902858734,
+      "learning_rate": 9.230788830004077e-05,
+      "loss": 4.227,
+      "step": 9065
+    },
+    {
+      "epoch": 0.6162522081804593,
+      "grad_norm": 0.17146620154380798,
+      "learning_rate": 9.230364179915751e-05,
+      "loss": 3.9837,
+      "step": 9070
+    },
+    {
+      "epoch": 0.6165919282511211,
+      "grad_norm": 0.23479421436786652,
+      "learning_rate": 9.229939529827422e-05,
+      "loss": 3.9549,
+      "step": 9075
+    },
+    {
+      "epoch": 0.6169316483217828,
+      "grad_norm": 0.17351080477237701,
+      "learning_rate": 9.229514879739095e-05,
+      "loss": 4.0837,
+      "step": 9080
+    },
+    {
+      "epoch": 0.6172713683924447,
+      "grad_norm": 0.18759216368198395,
+      "learning_rate": 9.229090229650769e-05,
+      "loss": 4.1507,
+      "step": 9085
+    },
+    {
+      "epoch": 0.6176110884631064,
+      "grad_norm": 0.373471736907959,
+      "learning_rate": 9.22866557956244e-05,
+      "loss": 3.8382,
+      "step": 9090
+    },
+    {
+      "epoch": 0.6179508085337682,
+      "grad_norm": 0.16654084622859955,
+      "learning_rate": 9.228240929474113e-05,
+      "loss": 3.9211,
+      "step": 9095
+    },
+    {
+      "epoch": 0.6182905286044299,
+      "grad_norm": 2.6412665843963623,
+      "learning_rate": 9.227816279385788e-05,
+      "loss": 4.1437,
+      "step": 9100
+    },
+    {
+      "epoch": 0.6186302486750918,
+      "grad_norm": 0.22232688963413239,
+      "learning_rate": 9.227391629297459e-05,
+      "loss": 4.3322,
+      "step": 9105
+    },
+    {
+      "epoch": 0.6189699687457535,
+      "grad_norm": 0.19792711734771729,
+      "learning_rate": 9.226966979209132e-05,
+      "loss": 3.6801,
+      "step": 9110
+    },
+    {
+      "epoch": 0.6193096888164152,
+      "grad_norm": 0.19452767074108124,
+      "learning_rate": 9.226542329120806e-05,
+      "loss": 4.1374,
+      "step": 9115
+    },
+    {
+      "epoch": 0.6196494088870771,
+      "grad_norm": 0.17766821384429932,
+      "learning_rate": 9.226117679032477e-05,
+      "loss": 3.998,
+      "step": 9120
+    },
+    {
+      "epoch": 0.6199891289577388,
+      "grad_norm": 0.21171297132968903,
+      "learning_rate": 9.22569302894415e-05,
+      "loss": 4.0408,
+      "step": 9125
+    },
+    {
+      "epoch": 0.6203288490284006,
+      "grad_norm": 0.2353495955467224,
+      "learning_rate": 9.225268378855823e-05,
+      "loss": 4.2569,
+      "step": 9130
+    },
+    {
+      "epoch": 0.6206685690990623,
+      "grad_norm": 0.1814018189907074,
+      "learning_rate": 9.224843728767496e-05,
+      "loss": 4.1783,
+      "step": 9135
+    },
+    {
+      "epoch": 0.6210082891697242,
+      "grad_norm": 0.16691021621227264,
+      "learning_rate": 9.224419078679169e-05,
+      "loss": 3.95,
+      "step": 9140
+    },
+    {
+      "epoch": 0.6213480092403859,
+      "grad_norm": 0.24587292969226837,
+      "learning_rate": 9.223994428590841e-05,
+      "loss": 4.187,
+      "step": 9145
+    },
+    {
+      "epoch": 0.6216877293110477,
+      "grad_norm": 0.12846866250038147,
+      "learning_rate": 9.223569778502514e-05,
+      "loss": 3.7833,
+      "step": 9150
+    },
+    {
+      "epoch": 0.6220274493817095,
+      "grad_norm": 0.15816563367843628,
+      "learning_rate": 9.223145128414187e-05,
+      "loss": 3.9622,
+      "step": 9155
+    },
+    {
+      "epoch": 0.6223671694523712,
+      "grad_norm": 0.5960243344306946,
+      "learning_rate": 9.22272047832586e-05,
+      "loss": 4.0261,
+      "step": 9160
+    },
+    {
+      "epoch": 0.622706889523033,
+      "grad_norm": 0.21092690527439117,
+      "learning_rate": 9.222295828237533e-05,
+      "loss": 4.0417,
+      "step": 9165
+    },
+    {
+      "epoch": 0.6230466095936948,
+      "grad_norm": 0.21227766573429108,
+      "learning_rate": 9.221871178149205e-05,
+      "loss": 4.0396,
+      "step": 9170
+    },
+    {
+      "epoch": 0.6233863296643566,
+      "grad_norm": 0.1947152018547058,
+      "learning_rate": 9.221446528060878e-05,
+      "loss": 3.9731,
+      "step": 9175
+    },
+    {
+      "epoch": 0.6237260497350183,
+      "grad_norm": 0.264658123254776,
+      "learning_rate": 9.221021877972551e-05,
+      "loss": 3.7214,
+      "step": 9180
+    },
+    {
+      "epoch": 0.6240657698056801,
+      "grad_norm": 0.14752690494060516,
+      "learning_rate": 9.220597227884224e-05,
+      "loss": 4.262,
+      "step": 9185
+    },
+    {
+      "epoch": 0.6244054898763419,
+      "grad_norm": 0.20559826493263245,
+      "learning_rate": 9.220172577795897e-05,
+      "loss": 4.0443,
+      "step": 9190
+    },
+    {
+      "epoch": 0.6247452099470037,
+      "grad_norm": 0.17190653085708618,
+      "learning_rate": 9.21974792770757e-05,
+      "loss": 4.246,
+      "step": 9195
+    },
+    {
+      "epoch": 0.6250849300176654,
+      "grad_norm": 0.20981894433498383,
+      "learning_rate": 9.219323277619242e-05,
+      "loss": 4.0482,
+      "step": 9200
+    },
+    {
+      "epoch": 0.6254246500883273,
+      "grad_norm": 0.20660968124866486,
+      "learning_rate": 9.218898627530915e-05,
+      "loss": 3.9868,
+      "step": 9205
+    },
+    {
+      "epoch": 0.625764370158989,
+      "grad_norm": 0.21324321627616882,
+      "learning_rate": 9.218473977442588e-05,
+      "loss": 4.3044,
+      "step": 9210
+    },
+    {
+      "epoch": 0.6261040902296507,
+      "grad_norm": 0.45024943351745605,
+      "learning_rate": 9.21804932735426e-05,
+      "loss": 3.945,
+      "step": 9215
+    },
+    {
+      "epoch": 0.6264438103003126,
+      "grad_norm": 0.19029852747917175,
+      "learning_rate": 9.217624677265933e-05,
+      "loss": 4.2293,
+      "step": 9220
+    },
+    {
+      "epoch": 0.6267835303709743,
+      "grad_norm": 0.17948591709136963,
+      "learning_rate": 9.217200027177606e-05,
+      "loss": 4.4357,
+      "step": 9225
+    },
+    {
+      "epoch": 0.6271232504416361,
+      "grad_norm": 0.30317583680152893,
+      "learning_rate": 9.216775377089279e-05,
+      "loss": 3.9639,
+      "step": 9230
+    },
+    {
+      "epoch": 0.6274629705122978,
+      "grad_norm": 0.1628941148519516,
+      "learning_rate": 9.216350727000952e-05,
+      "loss": 4.0654,
+      "step": 9235
+    },
+    {
+      "epoch": 0.6278026905829597,
+      "grad_norm": 0.20330828428268433,
+      "learning_rate": 9.215926076912625e-05,
+      "loss": 4.0975,
+      "step": 9240
+    },
+    {
+      "epoch": 0.6281424106536214,
+      "grad_norm": 0.19045989215373993,
+      "learning_rate": 9.215501426824297e-05,
+      "loss": 4.0935,
+      "step": 9245
+    },
+    {
+      "epoch": 0.6284821307242832,
+      "grad_norm": 1.2690106630325317,
+      "learning_rate": 9.21507677673597e-05,
+      "loss": 3.8794,
+      "step": 9250
+    },
+    {
+      "epoch": 0.628821850794945,
+      "grad_norm": 0.3011123538017273,
+      "learning_rate": 9.214652126647643e-05,
+      "loss": 3.8597,
+      "step": 9255
+    },
+    {
+      "epoch": 0.6291615708656068,
+      "grad_norm": 0.17079856991767883,
+      "learning_rate": 9.214227476559314e-05,
+      "loss": 3.9687,
+      "step": 9260
+    },
+    {
+      "epoch": 0.6295012909362685,
+      "grad_norm": 0.18377001583576202,
+      "learning_rate": 9.213802826470989e-05,
+      "loss": 3.8344,
+      "step": 9265
+    },
+    {
+      "epoch": 0.6298410110069302,
+      "grad_norm": 0.26580023765563965,
+      "learning_rate": 9.213378176382661e-05,
+      "loss": 3.9417,
+      "step": 9270
+    },
+    {
+      "epoch": 0.6301807310775921,
+      "grad_norm": 0.24509429931640625,
+      "learning_rate": 9.212953526294333e-05,
+      "loss": 3.9923,
+      "step": 9275
+    },
+    {
+      "epoch": 0.6305204511482538,
+      "grad_norm": 0.1840851753950119,
+      "learning_rate": 9.212528876206007e-05,
+      "loss": 4.1023,
+      "step": 9280
+    },
+    {
+      "epoch": 0.6308601712189156,
+      "grad_norm": 0.16701363027095795,
+      "learning_rate": 9.21210422611768e-05,
+      "loss": 4.1101,
+      "step": 9285
+    },
+    {
+      "epoch": 0.6311998912895774,
+      "grad_norm": 0.5241246819496155,
+      "learning_rate": 9.211679576029351e-05,
+      "loss": 3.5958,
+      "step": 9290
+    },
+    {
+      "epoch": 0.6315396113602392,
+      "grad_norm": 0.17940564453601837,
+      "learning_rate": 9.211254925941025e-05,
+      "loss": 3.9681,
+      "step": 9295
+    },
+    {
+      "epoch": 0.6318793314309009,
+      "grad_norm": 0.2497144490480423,
+      "learning_rate": 9.210830275852698e-05,
+      "loss": 4.0491,
+      "step": 9300
+    },
+    {
+      "epoch": 0.6322190515015628,
+      "grad_norm": 0.48636677861213684,
+      "learning_rate": 9.21040562576437e-05,
+      "loss": 3.9699,
+      "step": 9305
+    },
+    {
+      "epoch": 0.6325587715722245,
+      "grad_norm": 0.2682591676712036,
+      "learning_rate": 9.209980975676044e-05,
+      "loss": 3.8142,
+      "step": 9310
+    },
+    {
+      "epoch": 0.6328984916428863,
+      "grad_norm": 0.5112836360931396,
+      "learning_rate": 9.209556325587717e-05,
+      "loss": 3.8701,
+      "step": 9315
+    },
+    {
+      "epoch": 0.633238211713548,
+      "grad_norm": 0.37686988711357117,
+      "learning_rate": 9.20913167549939e-05,
+      "loss": 3.9299,
+      "step": 9320
+    },
+    {
+      "epoch": 0.6335779317842098,
+      "grad_norm": 0.3224041759967804,
+      "learning_rate": 9.208707025411062e-05,
+      "loss": 4.0601,
+      "step": 9325
+    },
+    {
+      "epoch": 0.6339176518548716,
+      "grad_norm": 0.241369366645813,
+      "learning_rate": 9.208282375322734e-05,
+      "loss": 4.1394,
+      "step": 9330
+    },
+    {
+      "epoch": 0.6342573719255333,
+      "grad_norm": 0.1902744024991989,
+      "learning_rate": 9.207857725234408e-05,
+      "loss": 4.1937,
+      "step": 9335
+    },
+    {
+      "epoch": 0.6345970919961952,
+      "grad_norm": 0.20103438198566437,
+      "learning_rate": 9.20743307514608e-05,
+      "loss": 4.1394,
+      "step": 9340
+    },
+    {
+      "epoch": 0.6349368120668569,
+      "grad_norm": 0.17396235466003418,
+      "learning_rate": 9.207008425057752e-05,
+      "loss": 4.0629,
+      "step": 9345
+    },
+    {
+      "epoch": 0.6352765321375187,
+      "grad_norm": 0.1853020042181015,
+      "learning_rate": 9.206583774969426e-05,
+      "loss": 4.0948,
+      "step": 9350
+    },
+    {
+      "epoch": 0.6356162522081804,
+      "grad_norm": 0.15830761194229126,
+      "learning_rate": 9.206159124881099e-05,
+      "loss": 4.0189,
+      "step": 9355
+    },
+    {
+      "epoch": 0.6359559722788423,
+      "grad_norm": 0.18493899703025818,
+      "learning_rate": 9.20573447479277e-05,
+      "loss": 3.9398,
+      "step": 9360
+    },
+    {
+      "epoch": 0.636295692349504,
+      "grad_norm": 0.19994007050991058,
+      "learning_rate": 9.205309824704445e-05,
+      "loss": 4.1339,
+      "step": 9365
+    },
+    {
+      "epoch": 0.6366354124201657,
+      "grad_norm": 0.16409240663051605,
+      "learning_rate": 9.204885174616117e-05,
+      "loss": 4.3045,
+      "step": 9370
+    },
+    {
+      "epoch": 0.6369751324908276,
+      "grad_norm": 0.34056273102760315,
+      "learning_rate": 9.204460524527789e-05,
+      "loss": 3.9817,
+      "step": 9375
+    },
+    {
+      "epoch": 0.6373148525614893,
+      "grad_norm": 0.19899314641952515,
+      "learning_rate": 9.204035874439463e-05,
+      "loss": 3.8643,
+      "step": 9380
+    },
+    {
+      "epoch": 0.6376545726321511,
+      "grad_norm": 0.2605150043964386,
+      "learning_rate": 9.203611224351136e-05,
+      "loss": 4.0556,
+      "step": 9385
+    },
+    {
+      "epoch": 0.6379942927028129,
+      "grad_norm": 0.23088324069976807,
+      "learning_rate": 9.203186574262807e-05,
+      "loss": 3.9551,
+      "step": 9390
+    },
+    {
+      "epoch": 0.6383340127734747,
+      "grad_norm": 0.181942418217659,
+      "learning_rate": 9.202761924174481e-05,
+      "loss": 4.0324,
+      "step": 9395
+    },
+    {
+      "epoch": 0.6386737328441364,
+      "grad_norm": 0.1608920693397522,
+      "learning_rate": 9.202337274086154e-05,
+      "loss": 3.8843,
+      "step": 9400
+    },
+    {
+      "epoch": 0.6390134529147982,
+      "grad_norm": 0.23045985400676727,
+      "learning_rate": 9.201912623997826e-05,
+      "loss": 3.8905,
+      "step": 9405
+    },
+    {
+      "epoch": 0.63935317298546,
+      "grad_norm": 0.20380303263664246,
+      "learning_rate": 9.2014879739095e-05,
+      "loss": 3.988,
+      "step": 9410
+    },
+    {
+      "epoch": 0.6396928930561218,
+      "grad_norm": 0.6223300695419312,
+      "learning_rate": 9.201063323821171e-05,
+      "loss": 3.9791,
+      "step": 9415
+    },
+    {
+      "epoch": 0.6400326131267835,
+      "grad_norm": 0.8067218065261841,
+      "learning_rate": 9.200638673732844e-05,
+      "loss": 4.1077,
+      "step": 9420
+    },
+    {
+      "epoch": 0.6403723331974454,
+      "grad_norm": 0.20715036988258362,
+      "learning_rate": 9.200214023644518e-05,
+      "loss": 4.138,
+      "step": 9425
+    },
+    {
+      "epoch": 0.6407120532681071,
+      "grad_norm": 0.20288419723510742,
+      "learning_rate": 9.19978937355619e-05,
+      "loss": 3.9553,
+      "step": 9430
+    },
+    {
+      "epoch": 0.6410517733387688,
+      "grad_norm": 0.18636272847652435,
+      "learning_rate": 9.199364723467862e-05,
+      "loss": 3.9276,
+      "step": 9435
+    },
+    {
+      "epoch": 0.6413914934094306,
+      "grad_norm": 0.18782015144824982,
+      "learning_rate": 9.198940073379537e-05,
+      "loss": 4.0336,
+      "step": 9440
+    },
+    {
+      "epoch": 0.6417312134800924,
+      "grad_norm": 0.19050072133541107,
+      "learning_rate": 9.198515423291208e-05,
+      "loss": 4.0227,
+      "step": 9445
+    },
+    {
+      "epoch": 0.6420709335507542,
+      "grad_norm": 0.30288147926330566,
+      "learning_rate": 9.198090773202881e-05,
+      "loss": 3.8769,
+      "step": 9450
+    },
+    {
+      "epoch": 0.6424106536214159,
+      "grad_norm": 0.22393420338630676,
+      "learning_rate": 9.197666123114555e-05,
+      "loss": 3.9509,
+      "step": 9455
+    },
+    {
+      "epoch": 0.6427503736920778,
+      "grad_norm": 0.21114152669906616,
+      "learning_rate": 9.197241473026226e-05,
+      "loss": 4.1008,
+      "step": 9460
+    },
+    {
+      "epoch": 0.6430900937627395,
+      "grad_norm": 0.238412544131279,
+      "learning_rate": 9.196816822937899e-05,
+      "loss": 3.8772,
+      "step": 9465
+    },
+    {
+      "epoch": 0.6434298138334013,
+      "grad_norm": 0.6189041137695312,
+      "learning_rate": 9.196392172849573e-05,
+      "loss": 3.9286,
+      "step": 9470
+    },
+    {
+      "epoch": 0.6437695339040631,
+      "grad_norm": 0.26222628355026245,
+      "learning_rate": 9.195967522761245e-05,
+      "loss": 4.1328,
+      "step": 9475
+    },
+    {
+      "epoch": 0.6441092539747248,
+      "grad_norm": 0.21858513355255127,
+      "learning_rate": 9.195542872672918e-05,
+      "loss": 3.9626,
+      "step": 9480
+    },
+    {
+      "epoch": 0.6444489740453866,
+      "grad_norm": 0.21814042329788208,
+      "learning_rate": 9.19511822258459e-05,
+      "loss": 4.3128,
+      "step": 9485
+    },
+    {
+      "epoch": 0.6447886941160483,
+      "grad_norm": 0.19709447026252747,
+      "learning_rate": 9.194693572496263e-05,
+      "loss": 3.9969,
+      "step": 9490
+    },
+    {
+      "epoch": 0.6451284141867102,
+      "grad_norm": 0.172958642244339,
+      "learning_rate": 9.194268922407936e-05,
+      "loss": 4.0359,
+      "step": 9495
+    },
+    {
+      "epoch": 0.6454681342573719,
+      "grad_norm": 0.17147018015384674,
+      "learning_rate": 9.193844272319609e-05,
+      "loss": 3.8998,
+      "step": 9500
+    },
+    {
+      "epoch": 0.6458078543280337,
+      "grad_norm": 0.16436798870563507,
+      "learning_rate": 9.193419622231282e-05,
+      "loss": 3.8929,
+      "step": 9505
+    },
+    {
+      "epoch": 0.6461475743986955,
+      "grad_norm": 0.17490500211715698,
+      "learning_rate": 9.192994972142954e-05,
+      "loss": 3.9468,
+      "step": 9510
+    },
+    {
+      "epoch": 0.6464872944693573,
+      "grad_norm": 0.5229065418243408,
+      "learning_rate": 9.192570322054627e-05,
+      "loss": 4.1358,
+      "step": 9515
+    },
+    {
+      "epoch": 0.646827014540019,
+      "grad_norm": 0.1972028911113739,
+      "learning_rate": 9.1921456719663e-05,
+      "loss": 4.1462,
+      "step": 9520
+    },
+    {
+      "epoch": 0.6471667346106807,
+      "grad_norm": 0.18450862169265747,
+      "learning_rate": 9.191721021877973e-05,
+      "loss": 4.0512,
+      "step": 9525
+    },
+    {
+      "epoch": 0.6475064546813426,
+      "grad_norm": 0.24547728896141052,
+      "learning_rate": 9.191296371789646e-05,
+      "loss": 4.0998,
+      "step": 9530
+    },
+    {
+      "epoch": 0.6478461747520043,
+      "grad_norm": 0.19123290479183197,
+      "learning_rate": 9.190871721701318e-05,
+      "loss": 3.8672,
+      "step": 9535
+    },
+    {
+      "epoch": 0.6481858948226661,
+      "grad_norm": 0.22196923196315765,
+      "learning_rate": 9.190447071612991e-05,
+      "loss": 4.1529,
+      "step": 9540
+    },
+    {
+      "epoch": 0.6485256148933279,
+      "grad_norm": 0.21963347494602203,
+      "learning_rate": 9.190022421524664e-05,
+      "loss": 4.1048,
+      "step": 9545
+    },
+    {
+      "epoch": 0.6488653349639897,
+      "grad_norm": 0.19842089712619781,
+      "learning_rate": 9.189597771436337e-05,
+      "loss": 4.0297,
+      "step": 9550
+    },
+    {
+      "epoch": 0.6492050550346514,
+      "grad_norm": 0.2944156527519226,
+      "learning_rate": 9.18917312134801e-05,
+      "loss": 4.0261,
+      "step": 9555
+    },
+    {
+      "epoch": 0.6495447751053133,
+      "grad_norm": 0.1940944492816925,
+      "learning_rate": 9.188748471259682e-05,
+      "loss": 3.9962,
+      "step": 9560
+    },
+    {
+      "epoch": 0.649884495175975,
+      "grad_norm": 0.1459953486919403,
+      "learning_rate": 9.188323821171355e-05,
+      "loss": 4.0297,
+      "step": 9565
+    },
+    {
+      "epoch": 0.6502242152466368,
+      "grad_norm": 0.18244166672229767,
+      "learning_rate": 9.187899171083028e-05,
+      "loss": 3.9038,
+      "step": 9570
+    },
+    {
+      "epoch": 0.6505639353172985,
+      "grad_norm": 0.5632447004318237,
+      "learning_rate": 9.187474520994701e-05,
+      "loss": 3.8968,
+      "step": 9575
+    },
+    {
+      "epoch": 0.6509036553879604,
+      "grad_norm": 0.21160653233528137,
+      "learning_rate": 9.187049870906374e-05,
+      "loss": 4.0296,
+      "step": 9580
+    },
+    {
+      "epoch": 0.6512433754586221,
+      "grad_norm": 0.5150845646858215,
+      "learning_rate": 9.186625220818046e-05,
+      "loss": 3.9448,
+      "step": 9585
+    },
+    {
+      "epoch": 0.6515830955292838,
+      "grad_norm": 0.16149458289146423,
+      "learning_rate": 9.186200570729719e-05,
+      "loss": 3.9315,
+      "step": 9590
+    },
+    {
+      "epoch": 0.6519228155999457,
+      "grad_norm": 0.19759464263916016,
+      "learning_rate": 9.185775920641392e-05,
+      "loss": 3.9934,
+      "step": 9595
+    },
+    {
+      "epoch": 0.6522625356706074,
+      "grad_norm": 0.4384717047214508,
+      "learning_rate": 9.185351270553065e-05,
+      "loss": 3.932,
+      "step": 9600
+    },
+    {
+      "epoch": 0.6526022557412692,
+      "grad_norm": 0.17127420008182526,
+      "learning_rate": 9.184926620464738e-05,
+      "loss": 4.038,
+      "step": 9605
+    },
+    {
+      "epoch": 0.6529419758119309,
+      "grad_norm": 0.1805383861064911,
+      "learning_rate": 9.18450197037641e-05,
+      "loss": 4.077,
+      "step": 9610
+    },
+    {
+      "epoch": 0.6532816958825928,
+      "grad_norm": 0.17218123376369476,
+      "learning_rate": 9.184077320288082e-05,
+      "loss": 4.0027,
+      "step": 9615
+    },
+    {
+      "epoch": 0.6536214159532545,
+      "grad_norm": 0.21027123928070068,
+      "learning_rate": 9.183652670199756e-05,
+      "loss": 4.004,
+      "step": 9620
+    },
+    {
+      "epoch": 0.6539611360239163,
+      "grad_norm": 0.18752577900886536,
+      "learning_rate": 9.183228020111429e-05,
+      "loss": 4.1132,
+      "step": 9625
+    },
+    {
+      "epoch": 0.6543008560945781,
+      "grad_norm": 0.2274181991815567,
+      "learning_rate": 9.1828033700231e-05,
+      "loss": 4.0189,
+      "step": 9630
+    },
+    {
+      "epoch": 0.6546405761652399,
+      "grad_norm": 0.5568689107894897,
+      "learning_rate": 9.182378719934774e-05,
+      "loss": 3.9176,
+      "step": 9635
+    },
+    {
+      "epoch": 0.6549802962359016,
+      "grad_norm": 0.20141251385211945,
+      "learning_rate": 9.181954069846447e-05,
+      "loss": 4.1433,
+      "step": 9640
+    },
+    {
+      "epoch": 0.6553200163065634,
+      "grad_norm": 0.1734774261713028,
+      "learning_rate": 9.181529419758119e-05,
+      "loss": 3.8714,
+      "step": 9645
+    },
+    {
+      "epoch": 0.6556597363772252,
+      "grad_norm": 0.20372764766216278,
+      "learning_rate": 9.181104769669793e-05,
+      "loss": 3.6352,
+      "step": 9650
+    },
+    {
+      "epoch": 0.6559994564478869,
+      "grad_norm": 0.22679565846920013,
+      "learning_rate": 9.180680119581466e-05,
+      "loss": 4.0213,
+      "step": 9655
+    },
+    {
+      "epoch": 0.6563391765185487,
+      "grad_norm": 0.19201093912124634,
+      "learning_rate": 9.180255469493138e-05,
+      "loss": 4.3365,
+      "step": 9660
+    },
+    {
+      "epoch": 0.6566788965892105,
+      "grad_norm": 0.23552648723125458,
+      "learning_rate": 9.179830819404811e-05,
+      "loss": 3.827,
+      "step": 9665
+    },
+    {
+      "epoch": 0.6570186166598723,
+      "grad_norm": 0.2363472580909729,
+      "learning_rate": 9.179406169316484e-05,
+      "loss": 3.5363,
+      "step": 9670
+    },
+    {
+      "epoch": 0.657358336730534,
+      "grad_norm": 0.1936025321483612,
+      "learning_rate": 9.178981519228157e-05,
+      "loss": 3.6846,
+      "step": 9675
+    },
+    {
+      "epoch": 0.6576980568011959,
+      "grad_norm": 0.16396436095237732,
+      "learning_rate": 9.17855686913983e-05,
+      "loss": 4.2385,
+      "step": 9680
+    },
+    {
+      "epoch": 0.6580377768718576,
+      "grad_norm": 0.19219662249088287,
+      "learning_rate": 9.178132219051501e-05,
+      "loss": 3.8246,
+      "step": 9685
+    },
+    {
+      "epoch": 0.6583774969425193,
+      "grad_norm": 0.3243776857852936,
+      "learning_rate": 9.177707568963175e-05,
+      "loss": 3.9252,
+      "step": 9690
+    },
+    {
+      "epoch": 0.6587172170131811,
+      "grad_norm": 0.2118336260318756,
+      "learning_rate": 9.177282918874848e-05,
+      "loss": 3.9518,
+      "step": 9695
+    },
+    {
+      "epoch": 0.6590569370838429,
+      "grad_norm": 0.2181885987520218,
+      "learning_rate": 9.17685826878652e-05,
+      "loss": 3.8817,
+      "step": 9700
+    },
+    {
+      "epoch": 0.6593966571545047,
+      "grad_norm": 0.2063203752040863,
+      "learning_rate": 9.176433618698194e-05,
+      "loss": 4.0715,
+      "step": 9705
+    },
+    {
+      "epoch": 0.6597363772251664,
+      "grad_norm": 0.16221602261066437,
+      "learning_rate": 9.176008968609867e-05,
+      "loss": 3.8972,
+      "step": 9710
+    },
+    {
+      "epoch": 0.6600760972958283,
+      "grad_norm": 0.2043311893939972,
+      "learning_rate": 9.175584318521538e-05,
+      "loss": 3.9982,
+      "step": 9715
+    },
+    {
+      "epoch": 0.66041581736649,
+      "grad_norm": 0.15358102321624756,
+      "learning_rate": 9.175159668433212e-05,
+      "loss": 4.4382,
+      "step": 9720
+    },
+    {
+      "epoch": 0.6607555374371518,
+      "grad_norm": 0.1804971694946289,
+      "learning_rate": 9.174735018344885e-05,
+      "loss": 4.0575,
+      "step": 9725
+    },
+    {
+      "epoch": 0.6610952575078136,
+      "grad_norm": 0.8969916701316833,
+      "learning_rate": 9.174310368256556e-05,
+      "loss": 3.8002,
+      "step": 9730
+    },
+    {
+      "epoch": 0.6614349775784754,
+      "grad_norm": 0.17219315469264984,
+      "learning_rate": 9.17388571816823e-05,
+      "loss": 3.8753,
+      "step": 9735
+    },
+    {
+      "epoch": 0.6617746976491371,
+      "grad_norm": 1.4561220407485962,
+      "learning_rate": 9.173461068079903e-05,
+      "loss": 3.8808,
+      "step": 9740
+    },
+    {
+      "epoch": 0.6621144177197988,
+      "grad_norm": 0.3004949688911438,
+      "learning_rate": 9.173036417991575e-05,
+      "loss": 4.01,
+      "step": 9745
+    },
+    {
+      "epoch": 0.6624541377904607,
+      "grad_norm": 0.4193626642227173,
+      "learning_rate": 9.172611767903249e-05,
+      "loss": 4.1539,
+      "step": 9750
+    },
+    {
+      "epoch": 0.6627938578611224,
+      "grad_norm": 0.5232547521591187,
+      "learning_rate": 9.17218711781492e-05,
+      "loss": 4.1406,
+      "step": 9755
+    },
+    {
+      "epoch": 0.6631335779317842,
+      "grad_norm": 0.2067098319530487,
+      "learning_rate": 9.171762467726593e-05,
+      "loss": 3.9981,
+      "step": 9760
+    },
+    {
+      "epoch": 0.663473298002446,
+      "grad_norm": 0.2100781351327896,
+      "learning_rate": 9.171337817638267e-05,
+      "loss": 3.7371,
+      "step": 9765
+    },
+    {
+      "epoch": 0.6638130180731078,
+      "grad_norm": 0.28894108533859253,
+      "learning_rate": 9.170913167549939e-05,
+      "loss": 4.1151,
+      "step": 9770
+    },
+    {
+      "epoch": 0.6641527381437695,
+      "grad_norm": 0.2366105318069458,
+      "learning_rate": 9.170488517461612e-05,
+      "loss": 3.7963,
+      "step": 9775
+    },
+    {
+      "epoch": 0.6644924582144313,
+      "grad_norm": 0.15720832347869873,
+      "learning_rate": 9.170063867373286e-05,
+      "loss": 4.1177,
+      "step": 9780
+    },
+    {
+      "epoch": 0.6648321782850931,
+      "grad_norm": 0.16979362070560455,
+      "learning_rate": 9.169639217284957e-05,
+      "loss": 4.1974,
+      "step": 9785
+    },
+    {
+      "epoch": 0.6651718983557549,
+      "grad_norm": 0.1718185693025589,
+      "learning_rate": 9.16921456719663e-05,
+      "loss": 3.8146,
+      "step": 9790
+    },
+    {
+      "epoch": 0.6655116184264166,
+      "grad_norm": 0.1998133510351181,
+      "learning_rate": 9.168789917108304e-05,
+      "loss": 3.8062,
+      "step": 9795
+    },
+    {
+      "epoch": 0.6658513384970784,
+      "grad_norm": 0.19915224611759186,
+      "learning_rate": 9.168365267019976e-05,
+      "loss": 3.8217,
+      "step": 9800
+    },
+    {
+      "epoch": 0.6661910585677402,
+      "grad_norm": 0.19672515988349915,
+      "learning_rate": 9.167940616931648e-05,
+      "loss": 3.9699,
+      "step": 9805
+    },
+    {
+      "epoch": 0.6665307786384019,
+      "grad_norm": 0.18894067406654358,
+      "learning_rate": 9.167515966843323e-05,
+      "loss": 3.9818,
+      "step": 9810
+    },
+    {
+      "epoch": 0.6668704987090638,
+      "grad_norm": 0.2761344611644745,
+      "learning_rate": 9.167091316754994e-05,
+      "loss": 4.0251,
+      "step": 9815
+    },
+    {
+      "epoch": 0.6672102187797255,
+      "grad_norm": 0.20272742211818695,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 4.1925,
+      "step": 9820
+    },
+    {
+      "epoch": 0.6675499388503873,
+      "grad_norm": 1.1285771131515503,
+      "learning_rate": 9.166242016578341e-05,
+      "loss": 4.1825,
+      "step": 9825
+    },
+    {
+      "epoch": 0.667889658921049,
+      "grad_norm": 0.1726224571466446,
+      "learning_rate": 9.165817366490012e-05,
+      "loss": 3.9708,
+      "step": 9830
+    },
+    {
+      "epoch": 0.6682293789917109,
+      "grad_norm": 0.1987651288509369,
+      "learning_rate": 9.165392716401685e-05,
+      "loss": 4.1101,
+      "step": 9835
+    },
+    {
+      "epoch": 0.6685690990623726,
+      "grad_norm": 0.19084063172340393,
+      "learning_rate": 9.164968066313358e-05,
+      "loss": 3.9633,
+      "step": 9840
+    },
+    {
+      "epoch": 0.6689088191330343,
+      "grad_norm": 0.20953474938869476,
+      "learning_rate": 9.164543416225031e-05,
+      "loss": 4.0714,
+      "step": 9845
+    },
+    {
+      "epoch": 0.6692485392036962,
+      "grad_norm": 0.1689091920852661,
+      "learning_rate": 9.164118766136704e-05,
+      "loss": 4.0451,
+      "step": 9850
+    },
+    {
+      "epoch": 0.6695882592743579,
+      "grad_norm": 0.38470450043678284,
+      "learning_rate": 9.163694116048376e-05,
+      "loss": 4.2373,
+      "step": 9855
+    },
+    {
+      "epoch": 0.6699279793450197,
+      "grad_norm": 0.18749617040157318,
+      "learning_rate": 9.163269465960049e-05,
+      "loss": 4.0489,
+      "step": 9860
+    },
+    {
+      "epoch": 0.6702676994156814,
+      "grad_norm": 0.1558235138654709,
+      "learning_rate": 9.162844815871722e-05,
+      "loss": 4.0883,
+      "step": 9865
+    },
+    {
+      "epoch": 0.6706074194863433,
+      "grad_norm": 0.17656366527080536,
+      "learning_rate": 9.162420165783395e-05,
+      "loss": 3.8735,
+      "step": 9870
+    },
+    {
+      "epoch": 0.670947139557005,
+      "grad_norm": 0.16513271629810333,
+      "learning_rate": 9.161995515695068e-05,
+      "loss": 4.3651,
+      "step": 9875
+    },
+    {
+      "epoch": 0.6712868596276668,
+      "grad_norm": 0.369552880525589,
+      "learning_rate": 9.16157086560674e-05,
+      "loss": 4.1642,
+      "step": 9880
+    },
+    {
+      "epoch": 0.6716265796983286,
+      "grad_norm": 0.19037148356437683,
+      "learning_rate": 9.161146215518413e-05,
+      "loss": 4.0021,
+      "step": 9885
+    },
+    {
+      "epoch": 0.6719662997689904,
+      "grad_norm": 0.1722508817911148,
+      "learning_rate": 9.160721565430086e-05,
+      "loss": 3.8246,
+      "step": 9890
+    },
+    {
+      "epoch": 0.6723060198396521,
+      "grad_norm": 0.16816957294940948,
+      "learning_rate": 9.160296915341759e-05,
+      "loss": 3.7951,
+      "step": 9895
+    },
+    {
+      "epoch": 0.672645739910314,
+      "grad_norm": 0.18479801714420319,
+      "learning_rate": 9.159872265253432e-05,
+      "loss": 4.0533,
+      "step": 9900
+    },
+    {
+      "epoch": 0.6729854599809757,
+      "grad_norm": 0.1568441092967987,
+      "learning_rate": 9.159447615165104e-05,
+      "loss": 3.7568,
+      "step": 9905
+    },
+    {
+      "epoch": 0.6733251800516374,
+      "grad_norm": 0.15381227433681488,
+      "learning_rate": 9.159022965076777e-05,
+      "loss": 4.0221,
+      "step": 9910
+    },
+    {
+      "epoch": 0.6736649001222992,
+      "grad_norm": 0.22633537650108337,
+      "learning_rate": 9.15859831498845e-05,
+      "loss": 4.0213,
+      "step": 9915
+    },
+    {
+      "epoch": 0.674004620192961,
+      "grad_norm": 0.22000354528427124,
+      "learning_rate": 9.158173664900123e-05,
+      "loss": 3.9787,
+      "step": 9920
+    },
+    {
+      "epoch": 0.6743443402636228,
+      "grad_norm": 0.18316060304641724,
+      "learning_rate": 9.157749014811796e-05,
+      "loss": 3.939,
+      "step": 9925
+    },
+    {
+      "epoch": 0.6746840603342845,
+      "grad_norm": 0.3888174891471863,
+      "learning_rate": 9.157324364723468e-05,
+      "loss": 3.9585,
+      "step": 9930
+    },
+    {
+      "epoch": 0.6750237804049464,
+      "grad_norm": 0.24031803011894226,
+      "learning_rate": 9.156899714635141e-05,
+      "loss": 3.9616,
+      "step": 9935
+    },
+    {
+      "epoch": 0.6753635004756081,
+      "grad_norm": 0.22554604709148407,
+      "learning_rate": 9.156475064546814e-05,
+      "loss": 4.13,
+      "step": 9940
+    },
+    {
+      "epoch": 0.6757032205462699,
+      "grad_norm": 0.15670569241046906,
+      "learning_rate": 9.156050414458487e-05,
+      "loss": 4.0036,
+      "step": 9945
+    },
+    {
+      "epoch": 0.6760429406169316,
+      "grad_norm": 0.18224500119686127,
+      "learning_rate": 9.15562576437016e-05,
+      "loss": 4.1177,
+      "step": 9950
+    },
+    {
+      "epoch": 0.6763826606875935,
+      "grad_norm": 0.46049654483795166,
+      "learning_rate": 9.155201114281831e-05,
+      "loss": 3.8908,
+      "step": 9955
+    },
+    {
+      "epoch": 0.6767223807582552,
+      "grad_norm": 0.1925237625837326,
+      "learning_rate": 9.154776464193505e-05,
+      "loss": 4.0503,
+      "step": 9960
+    },
+    {
+      "epoch": 0.6770621008289169,
+      "grad_norm": 0.2237987518310547,
+      "learning_rate": 9.154351814105178e-05,
+      "loss": 4.0872,
+      "step": 9965
+    },
+    {
+      "epoch": 0.6774018208995788,
+      "grad_norm": 0.5446917414665222,
+      "learning_rate": 9.15392716401685e-05,
+      "loss": 3.9465,
+      "step": 9970
+    },
+    {
+      "epoch": 0.6777415409702405,
+      "grad_norm": 0.19259484112262726,
+      "learning_rate": 9.153502513928524e-05,
+      "loss": 4.0308,
+      "step": 9975
+    },
+    {
+      "epoch": 0.6780812610409023,
+      "grad_norm": 0.16928361356258392,
+      "learning_rate": 9.153077863840196e-05,
+      "loss": 3.8983,
+      "step": 9980
+    },
+    {
+      "epoch": 0.6784209811115641,
+      "grad_norm": 0.20402824878692627,
+      "learning_rate": 9.152653213751868e-05,
+      "loss": 4.0315,
+      "step": 9985
+    },
+    {
+      "epoch": 0.6787607011822259,
+      "grad_norm": 0.18891318142414093,
+      "learning_rate": 9.152228563663542e-05,
+      "loss": 3.9693,
+      "step": 9990
+    },
+    {
+      "epoch": 0.6791004212528876,
+      "grad_norm": 0.21101170778274536,
+      "learning_rate": 9.151803913575215e-05,
+      "loss": 3.8452,
+      "step": 9995
+    },
+    {
+      "epoch": 0.6794401413235494,
+      "grad_norm": 0.20866774022579193,
+      "learning_rate": 9.151379263486888e-05,
+      "loss": 3.7796,
+      "step": 10000
+    },
+    {
+      "epoch": 0.6797798613942112,
+      "grad_norm": 0.22108693420886993,
+      "learning_rate": 9.15095461339856e-05,
+      "loss": 4.1394,
+      "step": 10005
+    },
+    {
+      "epoch": 0.680119581464873,
+      "grad_norm": 0.18283884227275848,
+      "learning_rate": 9.150529963310233e-05,
+      "loss": 4.0837,
+      "step": 10010
+    },
+    {
+      "epoch": 0.6804593015355347,
+      "grad_norm": 0.3496660888195038,
+      "learning_rate": 9.150105313221906e-05,
+      "loss": 3.9772,
+      "step": 10015
+    },
+    {
+      "epoch": 0.6807990216061965,
+      "grad_norm": 0.21944838762283325,
+      "learning_rate": 9.149680663133579e-05,
+      "loss": 3.9497,
+      "step": 10020
+    },
+    {
+      "epoch": 0.6811387416768583,
+      "grad_norm": 0.9018047451972961,
+      "learning_rate": 9.149256013045252e-05,
+      "loss": 4.1269,
+      "step": 10025
+    },
+    {
+      "epoch": 0.68147846174752,
+      "grad_norm": 0.1882868856191635,
+      "learning_rate": 9.148831362956924e-05,
+      "loss": 3.9114,
+      "step": 10030
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.9261885285377502,
+      "learning_rate": 9.148406712868597e-05,
+      "loss": 4.1938,
+      "step": 10035
+    },
+    {
+      "epoch": 0.6821579018888436,
+      "grad_norm": 0.20487643778324127,
+      "learning_rate": 9.147982062780269e-05,
+      "loss": 4.029,
+      "step": 10040
+    },
+    {
+      "epoch": 0.6824976219595054,
+      "grad_norm": 0.20628653466701508,
+      "learning_rate": 9.147557412691943e-05,
+      "loss": 3.9453,
+      "step": 10045
+    },
+    {
+      "epoch": 0.6828373420301671,
+      "grad_norm": 0.18048962950706482,
+      "learning_rate": 9.147132762603616e-05,
+      "loss": 3.7238,
+      "step": 10050
+    },
+    {
+      "epoch": 0.683177062100829,
+      "grad_norm": 0.20143994688987732,
+      "learning_rate": 9.146708112515287e-05,
+      "loss": 4.0817,
+      "step": 10055
+    },
+    {
+      "epoch": 0.6835167821714907,
+      "grad_norm": 0.1772017776966095,
+      "learning_rate": 9.146283462426961e-05,
+      "loss": 4.3308,
+      "step": 10060
+    },
+    {
+      "epoch": 0.6838565022421524,
+      "grad_norm": 0.17534330487251282,
+      "learning_rate": 9.145858812338634e-05,
+      "loss": 3.9769,
+      "step": 10065
+    },
+    {
+      "epoch": 0.6841962223128143,
+      "grad_norm": 0.3508715331554413,
+      "learning_rate": 9.145434162250305e-05,
+      "loss": 4.1162,
+      "step": 10070
+    },
+    {
+      "epoch": 0.684535942383476,
+      "grad_norm": 0.22710181772708893,
+      "learning_rate": 9.14500951216198e-05,
+      "loss": 3.7313,
+      "step": 10075
+    },
+    {
+      "epoch": 0.6848756624541378,
+      "grad_norm": 0.1621914505958557,
+      "learning_rate": 9.144584862073652e-05,
+      "loss": 4.0611,
+      "step": 10080
+    },
+    {
+      "epoch": 0.6852153825247995,
+      "grad_norm": 0.26334255933761597,
+      "learning_rate": 9.144160211985324e-05,
+      "loss": 4.0556,
+      "step": 10085
+    },
+    {
+      "epoch": 0.6855551025954614,
+      "grad_norm": 0.6280850768089294,
+      "learning_rate": 9.143735561896998e-05,
+      "loss": 4.0588,
+      "step": 10090
+    },
+    {
+      "epoch": 0.6858948226661231,
+      "grad_norm": 0.3040946424007416,
+      "learning_rate": 9.143310911808671e-05,
+      "loss": 3.9212,
+      "step": 10095
+    },
+    {
+      "epoch": 0.6862345427367849,
+      "grad_norm": 0.30600547790527344,
+      "learning_rate": 9.142886261720342e-05,
+      "loss": 4.0689,
+      "step": 10100
+    },
+    {
+      "epoch": 0.6865742628074467,
+      "grad_norm": 0.18050602078437805,
+      "learning_rate": 9.142461611632016e-05,
+      "loss": 3.9464,
+      "step": 10105
+    },
+    {
+      "epoch": 0.6869139828781085,
+      "grad_norm": 0.48490941524505615,
+      "learning_rate": 9.142036961543688e-05,
+      "loss": 4.025,
+      "step": 10110
+    },
+    {
+      "epoch": 0.6872537029487702,
+      "grad_norm": 0.24022653698921204,
+      "learning_rate": 9.14161231145536e-05,
+      "loss": 4.0635,
+      "step": 10115
+    },
+    {
+      "epoch": 0.6875934230194319,
+      "grad_norm": 0.5546215772628784,
+      "learning_rate": 9.141187661367035e-05,
+      "loss": 3.9159,
+      "step": 10120
+    },
+    {
+      "epoch": 0.6879331430900938,
+      "grad_norm": 0.1747526228427887,
+      "learning_rate": 9.140763011278706e-05,
+      "loss": 4.0271,
+      "step": 10125
+    },
+    {
+      "epoch": 0.6882728631607555,
+      "grad_norm": 0.8281142115592957,
+      "learning_rate": 9.140338361190379e-05,
+      "loss": 3.8248,
+      "step": 10130
+    },
+    {
+      "epoch": 0.6886125832314173,
+      "grad_norm": 0.19988751411437988,
+      "learning_rate": 9.139913711102053e-05,
+      "loss": 4.022,
+      "step": 10135
+    },
+    {
+      "epoch": 0.6889523033020791,
+      "grad_norm": 0.5465024709701538,
+      "learning_rate": 9.139489061013725e-05,
+      "loss": 3.9433,
+      "step": 10140
+    },
+    {
+      "epoch": 0.6892920233727409,
+      "grad_norm": 0.704890787601471,
+      "learning_rate": 9.139064410925397e-05,
+      "loss": 4.2021,
+      "step": 10145
+    },
+    {
+      "epoch": 0.6896317434434026,
+      "grad_norm": 0.1683177798986435,
+      "learning_rate": 9.138639760837072e-05,
+      "loss": 4.0005,
+      "step": 10150
+    },
+    {
+      "epoch": 0.6899714635140645,
+      "grad_norm": 0.2499643862247467,
+      "learning_rate": 9.138215110748743e-05,
+      "loss": 4.0452,
+      "step": 10155
+    },
+    {
+      "epoch": 0.6903111835847262,
+      "grad_norm": 0.20261327922344208,
+      "learning_rate": 9.137790460660416e-05,
+      "loss": 3.9332,
+      "step": 10160
+    },
+    {
+      "epoch": 0.690650903655388,
+      "grad_norm": 0.19001875817775726,
+      "learning_rate": 9.13736581057209e-05,
+      "loss": 3.9008,
+      "step": 10165
+    },
+    {
+      "epoch": 0.6909906237260497,
+      "grad_norm": 0.18063665926456451,
+      "learning_rate": 9.136941160483761e-05,
+      "loss": 3.8843,
+      "step": 10170
+    },
+    {
+      "epoch": 0.6913303437967115,
+      "grad_norm": 0.1968679428100586,
+      "learning_rate": 9.136516510395434e-05,
+      "loss": 4.0298,
+      "step": 10175
+    },
+    {
+      "epoch": 0.6916700638673733,
+      "grad_norm": 0.21212510764598846,
+      "learning_rate": 9.136091860307107e-05,
+      "loss": 4.1819,
+      "step": 10180
+    },
+    {
+      "epoch": 0.692009783938035,
+      "grad_norm": 0.2317853420972824,
+      "learning_rate": 9.13566721021878e-05,
+      "loss": 3.8627,
+      "step": 10185
+    },
+    {
+      "epoch": 0.6923495040086969,
+      "grad_norm": 0.17582780122756958,
+      "learning_rate": 9.135242560130453e-05,
+      "loss": 4.044,
+      "step": 10190
+    },
+    {
+      "epoch": 0.6926892240793586,
+      "grad_norm": 0.1663304716348648,
+      "learning_rate": 9.134817910042125e-05,
+      "loss": 3.9946,
+      "step": 10195
+    },
+    {
+      "epoch": 0.6930289441500204,
+      "grad_norm": 0.2554979920387268,
+      "learning_rate": 9.134393259953798e-05,
+      "loss": 3.9845,
+      "step": 10200
+    },
+    {
+      "epoch": 0.6933686642206821,
+      "grad_norm": 0.19808262586593628,
+      "learning_rate": 9.133968609865471e-05,
+      "loss": 3.7411,
+      "step": 10205
+    },
+    {
+      "epoch": 0.693708384291344,
+      "grad_norm": 0.16207602620124817,
+      "learning_rate": 9.133543959777144e-05,
+      "loss": 3.9169,
+      "step": 10210
+    },
+    {
+      "epoch": 0.6940481043620057,
+      "grad_norm": 0.19255559146404266,
+      "learning_rate": 9.133119309688817e-05,
+      "loss": 4.0015,
+      "step": 10215
+    },
+    {
+      "epoch": 0.6943878244326674,
+      "grad_norm": 0.40374597907066345,
+      "learning_rate": 9.13269465960049e-05,
+      "loss": 4.0568,
+      "step": 10220
+    },
+    {
+      "epoch": 0.6947275445033293,
+      "grad_norm": 0.5542440414428711,
+      "learning_rate": 9.132270009512162e-05,
+      "loss": 4.0383,
+      "step": 10225
+    },
+    {
+      "epoch": 0.695067264573991,
+      "grad_norm": 0.2855238914489746,
+      "learning_rate": 9.131845359423835e-05,
+      "loss": 3.957,
+      "step": 10230
+    },
+    {
+      "epoch": 0.6954069846446528,
+      "grad_norm": 0.30983904004096985,
+      "learning_rate": 9.131420709335508e-05,
+      "loss": 3.9347,
+      "step": 10235
+    },
+    {
+      "epoch": 0.6957467047153146,
+      "grad_norm": 0.1820448637008667,
+      "learning_rate": 9.13099605924718e-05,
+      "loss": 4.0424,
+      "step": 10240
+    },
+    {
+      "epoch": 0.6960864247859764,
+      "grad_norm": 0.1688624769449234,
+      "learning_rate": 9.130571409158853e-05,
+      "loss": 4.1492,
+      "step": 10245
+    },
+    {
+      "epoch": 0.6964261448566381,
+      "grad_norm": 0.18944065272808075,
+      "learning_rate": 9.130146759070526e-05,
+      "loss": 4.0306,
+      "step": 10250
+    },
+    {
+      "epoch": 0.6967658649272999,
+      "grad_norm": 0.23642663657665253,
+      "learning_rate": 9.129722108982199e-05,
+      "loss": 3.9713,
+      "step": 10255
+    },
+    {
+      "epoch": 0.6971055849979617,
+      "grad_norm": 0.30841895937919617,
+      "learning_rate": 9.129297458893872e-05,
+      "loss": 4.0823,
+      "step": 10260
+    },
+    {
+      "epoch": 0.6974453050686235,
+      "grad_norm": 0.20149889588356018,
+      "learning_rate": 9.128872808805545e-05,
+      "loss": 3.9524,
+      "step": 10265
+    },
+    {
+      "epoch": 0.6977850251392852,
+      "grad_norm": 0.18833234906196594,
+      "learning_rate": 9.128448158717217e-05,
+      "loss": 4.1513,
+      "step": 10270
+    },
+    {
+      "epoch": 0.698124745209947,
+      "grad_norm": 0.1727929264307022,
+      "learning_rate": 9.12802350862889e-05,
+      "loss": 3.928,
+      "step": 10275
+    },
+    {
+      "epoch": 0.6984644652806088,
+      "grad_norm": 0.19599127769470215,
+      "learning_rate": 9.127598858540563e-05,
+      "loss": 3.9861,
+      "step": 10280
+    },
+    {
+      "epoch": 0.6988041853512705,
+      "grad_norm": 0.19696658849716187,
+      "learning_rate": 9.127174208452236e-05,
+      "loss": 3.9878,
+      "step": 10285
+    },
+    {
+      "epoch": 0.6991439054219323,
+      "grad_norm": 0.24747943878173828,
+      "learning_rate": 9.126749558363909e-05,
+      "loss": 3.9343,
+      "step": 10290
+    },
+    {
+      "epoch": 0.6994836254925941,
+      "grad_norm": 0.23848794400691986,
+      "learning_rate": 9.126324908275581e-05,
+      "loss": 3.7876,
+      "step": 10295
+    },
+    {
+      "epoch": 0.6998233455632559,
+      "grad_norm": 0.18184241652488708,
+      "learning_rate": 9.125900258187254e-05,
+      "loss": 4.003,
+      "step": 10300
+    },
+    {
+      "epoch": 0.7001630656339176,
+      "grad_norm": 0.21852192282676697,
+      "learning_rate": 9.125475608098927e-05,
+      "loss": 3.9115,
+      "step": 10305
+    },
+    {
+      "epoch": 0.7005027857045795,
+      "grad_norm": 0.18718090653419495,
+      "learning_rate": 9.125050958010599e-05,
+      "loss": 3.8846,
+      "step": 10310
+    },
+    {
+      "epoch": 0.7008425057752412,
+      "grad_norm": 0.4584771394729614,
+      "learning_rate": 9.124626307922273e-05,
+      "loss": 3.9954,
+      "step": 10315
+    },
+    {
+      "epoch": 0.701182225845903,
+      "grad_norm": 0.19998236000537872,
+      "learning_rate": 9.124201657833945e-05,
+      "loss": 3.9594,
+      "step": 10320
+    },
+    {
+      "epoch": 0.7015219459165648,
+      "grad_norm": 0.1826915442943573,
+      "learning_rate": 9.123777007745617e-05,
+      "loss": 3.8949,
+      "step": 10325
+    },
+    {
+      "epoch": 0.7018616659872265,
+      "grad_norm": 0.31630200147628784,
+      "learning_rate": 9.123352357657291e-05,
+      "loss": 4.0863,
+      "step": 10330
+    },
+    {
+      "epoch": 0.7022013860578883,
+      "grad_norm": 0.22190065681934357,
+      "learning_rate": 9.122927707568964e-05,
+      "loss": 3.9006,
+      "step": 10335
+    },
+    {
+      "epoch": 0.70254110612855,
+      "grad_norm": 0.14430834352970123,
+      "learning_rate": 9.122503057480637e-05,
+      "loss": 3.7831,
+      "step": 10340
+    },
+    {
+      "epoch": 0.7028808261992119,
+      "grad_norm": 0.23812465369701385,
+      "learning_rate": 9.12207840739231e-05,
+      "loss": 4.07,
+      "step": 10345
+    },
+    {
+      "epoch": 0.7032205462698736,
+      "grad_norm": 0.26677510142326355,
+      "learning_rate": 9.121653757303982e-05,
+      "loss": 4.0027,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7035602663405354,
+      "grad_norm": 1.3501478433609009,
+      "learning_rate": 9.121229107215655e-05,
+      "loss": 3.987,
+      "step": 10355
+    },
+    {
+      "epoch": 0.7038999864111972,
+      "grad_norm": 0.20116499066352844,
+      "learning_rate": 9.120804457127328e-05,
+      "loss": 4.2462,
+      "step": 10360
+    },
+    {
+      "epoch": 0.704239706481859,
+      "grad_norm": 0.22384878993034363,
+      "learning_rate": 9.120379807039001e-05,
+      "loss": 3.7601,
+      "step": 10365
+    },
+    {
+      "epoch": 0.7045794265525207,
+      "grad_norm": 0.2325512170791626,
+      "learning_rate": 9.119955156950673e-05,
+      "loss": 4.0985,
+      "step": 10370
+    },
+    {
+      "epoch": 0.7049191466231824,
+      "grad_norm": 0.16118580102920532,
+      "learning_rate": 9.119530506862346e-05,
+      "loss": 4.0038,
+      "step": 10375
+    },
+    {
+      "epoch": 0.7052588666938443,
+      "grad_norm": 0.16809016466140747,
+      "learning_rate": 9.119105856774018e-05,
+      "loss": 4.0822,
+      "step": 10380
+    },
+    {
+      "epoch": 0.705598586764506,
+      "grad_norm": 0.29214194416999817,
+      "learning_rate": 9.118681206685692e-05,
+      "loss": 3.9853,
+      "step": 10385
+    },
+    {
+      "epoch": 0.7059383068351678,
+      "grad_norm": 0.17380909621715546,
+      "learning_rate": 9.118256556597365e-05,
+      "loss": 3.9326,
+      "step": 10390
+    },
+    {
+      "epoch": 0.7062780269058296,
+      "grad_norm": 0.19394385814666748,
+      "learning_rate": 9.117831906509036e-05,
+      "loss": 4.0416,
+      "step": 10395
+    },
+    {
+      "epoch": 0.7066177469764914,
+      "grad_norm": 0.16878750920295715,
+      "learning_rate": 9.11740725642071e-05,
+      "loss": 4.2154,
+      "step": 10400
+    },
+    {
+      "epoch": 0.7069574670471531,
+      "grad_norm": 0.17969490587711334,
+      "learning_rate": 9.116982606332383e-05,
+      "loss": 3.8792,
+      "step": 10405
+    },
+    {
+      "epoch": 0.707297187117815,
+      "grad_norm": 0.18729190528392792,
+      "learning_rate": 9.116557956244055e-05,
+      "loss": 4.0986,
+      "step": 10410
+    },
+    {
+      "epoch": 0.7076369071884767,
+      "grad_norm": 0.16963765025138855,
+      "learning_rate": 9.116133306155729e-05,
+      "loss": 4.062,
+      "step": 10415
+    },
+    {
+      "epoch": 0.7079766272591385,
+      "grad_norm": 0.2884555757045746,
+      "learning_rate": 9.115708656067402e-05,
+      "loss": 4.078,
+      "step": 10420
+    },
+    {
+      "epoch": 0.7083163473298002,
+      "grad_norm": 0.29540711641311646,
+      "learning_rate": 9.115284005979073e-05,
+      "loss": 4.0736,
+      "step": 10425
+    },
+    {
+      "epoch": 0.708656067400462,
+      "grad_norm": 0.41383740305900574,
+      "learning_rate": 9.114859355890747e-05,
+      "loss": 4.2034,
+      "step": 10430
+    },
+    {
+      "epoch": 0.7089957874711238,
+      "grad_norm": 0.17644955217838287,
+      "learning_rate": 9.11443470580242e-05,
+      "loss": 4.1137,
+      "step": 10435
+    },
+    {
+      "epoch": 0.7093355075417855,
+      "grad_norm": 0.1741337925195694,
+      "learning_rate": 9.114010055714091e-05,
+      "loss": 4.0893,
+      "step": 10440
+    },
+    {
+      "epoch": 0.7096752276124474,
+      "grad_norm": 0.2527826130390167,
+      "learning_rate": 9.113585405625766e-05,
+      "loss": 4.091,
+      "step": 10445
+    },
+    {
+      "epoch": 0.7100149476831091,
+      "grad_norm": 0.1837528795003891,
+      "learning_rate": 9.113160755537438e-05,
+      "loss": 4.0449,
+      "step": 10450
+    },
+    {
+      "epoch": 0.7103546677537709,
+      "grad_norm": 0.18858109414577484,
+      "learning_rate": 9.11273610544911e-05,
+      "loss": 4.0575,
+      "step": 10455
+    },
+    {
+      "epoch": 0.7106943878244326,
+      "grad_norm": 0.1998942494392395,
+      "learning_rate": 9.112311455360784e-05,
+      "loss": 3.9053,
+      "step": 10460
+    },
+    {
+      "epoch": 0.7110341078950945,
+      "grad_norm": 0.20997202396392822,
+      "learning_rate": 9.111886805272455e-05,
+      "loss": 3.8818,
+      "step": 10465
+    },
+    {
+      "epoch": 0.7113738279657562,
+      "grad_norm": 0.17449168860912323,
+      "learning_rate": 9.111462155184128e-05,
+      "loss": 4.0812,
+      "step": 10470
+    },
+    {
+      "epoch": 0.711713548036418,
+      "grad_norm": 0.20188404619693756,
+      "learning_rate": 9.111037505095802e-05,
+      "loss": 3.918,
+      "step": 10475
+    },
+    {
+      "epoch": 0.7120532681070798,
+      "grad_norm": 0.16575513780117035,
+      "learning_rate": 9.110612855007474e-05,
+      "loss": 4.053,
+      "step": 10480
+    },
+    {
+      "epoch": 0.7123929881777415,
+      "grad_norm": 0.20615115761756897,
+      "learning_rate": 9.110188204919147e-05,
+      "loss": 3.7282,
+      "step": 10485
+    },
+    {
+      "epoch": 0.7127327082484033,
+      "grad_norm": 0.17066192626953125,
+      "learning_rate": 9.109763554830821e-05,
+      "loss": 4.0129,
+      "step": 10490
+    },
+    {
+      "epoch": 0.7130724283190651,
+      "grad_norm": 0.2495145946741104,
+      "learning_rate": 9.109338904742492e-05,
+      "loss": 3.9959,
+      "step": 10495
+    },
+    {
+      "epoch": 0.7134121483897269,
+      "grad_norm": 0.19648021459579468,
+      "learning_rate": 9.108914254654165e-05,
+      "loss": 4.0106,
+      "step": 10500
+    },
+    {
+      "epoch": 0.7137518684603886,
+      "grad_norm": 0.20270267128944397,
+      "learning_rate": 9.108489604565839e-05,
+      "loss": 4.0734,
+      "step": 10505
+    },
+    {
+      "epoch": 0.7140915885310504,
+      "grad_norm": 0.1632177233695984,
+      "learning_rate": 9.10806495447751e-05,
+      "loss": 4.2488,
+      "step": 10510
+    },
+    {
+      "epoch": 0.7144313086017122,
+      "grad_norm": 0.1604064404964447,
+      "learning_rate": 9.107640304389183e-05,
+      "loss": 3.8896,
+      "step": 10515
+    },
+    {
+      "epoch": 0.714771028672374,
+      "grad_norm": 0.21193253993988037,
+      "learning_rate": 9.107215654300858e-05,
+      "loss": 3.9983,
+      "step": 10520
+    },
+    {
+      "epoch": 0.7151107487430357,
+      "grad_norm": 0.3716839551925659,
+      "learning_rate": 9.106791004212529e-05,
+      "loss": 3.9367,
+      "step": 10525
+    },
+    {
+      "epoch": 0.7154504688136976,
+      "grad_norm": 0.1587960124015808,
+      "learning_rate": 9.106366354124202e-05,
+      "loss": 3.7641,
+      "step": 10530
+    },
+    {
+      "epoch": 0.7157901888843593,
+      "grad_norm": 4.4356184005737305,
+      "learning_rate": 9.105941704035875e-05,
+      "loss": 4.0203,
+      "step": 10535
+    },
+    {
+      "epoch": 0.716129908955021,
+      "grad_norm": 0.2456178516149521,
+      "learning_rate": 9.105517053947547e-05,
+      "loss": 4.0491,
+      "step": 10540
+    },
+    {
+      "epoch": 0.7164696290256828,
+      "grad_norm": 0.5795451402664185,
+      "learning_rate": 9.10509240385922e-05,
+      "loss": 3.8471,
+      "step": 10545
+    },
+    {
+      "epoch": 0.7168093490963446,
+      "grad_norm": 0.20285823941230774,
+      "learning_rate": 9.104667753770893e-05,
+      "loss": 4.1314,
+      "step": 10550
+    },
+    {
+      "epoch": 0.7171490691670064,
+      "grad_norm": 0.16531841456890106,
+      "learning_rate": 9.104243103682566e-05,
+      "loss": 4.1928,
+      "step": 10555
+    },
+    {
+      "epoch": 0.7174887892376681,
+      "grad_norm": 0.1814994513988495,
+      "learning_rate": 9.103818453594239e-05,
+      "loss": 4.0319,
+      "step": 10560
+    },
+    {
+      "epoch": 0.71782850930833,
+      "grad_norm": 0.16023671627044678,
+      "learning_rate": 9.103393803505911e-05,
+      "loss": 4.1279,
+      "step": 10565
+    },
+    {
+      "epoch": 0.7181682293789917,
+      "grad_norm": 0.2011050581932068,
+      "learning_rate": 9.102969153417584e-05,
+      "loss": 4.0957,
+      "step": 10570
+    },
+    {
+      "epoch": 0.7185079494496535,
+      "grad_norm": 0.21786335110664368,
+      "learning_rate": 9.102544503329257e-05,
+      "loss": 3.9609,
+      "step": 10575
+    },
+    {
+      "epoch": 0.7188476695203153,
+      "grad_norm": 0.22446627914905548,
+      "learning_rate": 9.10211985324093e-05,
+      "loss": 3.9967,
+      "step": 10580
+    },
+    {
+      "epoch": 0.719187389590977,
+      "grad_norm": 0.301862508058548,
+      "learning_rate": 9.101695203152603e-05,
+      "loss": 3.7026,
+      "step": 10585
+    },
+    {
+      "epoch": 0.7195271096616388,
+      "grad_norm": 0.19402875006198883,
+      "learning_rate": 9.101270553064275e-05,
+      "loss": 4.1842,
+      "step": 10590
+    },
+    {
+      "epoch": 0.7198668297323005,
+      "grad_norm": 0.2339441329240799,
+      "learning_rate": 9.100845902975948e-05,
+      "loss": 3.8018,
+      "step": 10595
+    },
+    {
+      "epoch": 0.7202065498029624,
+      "grad_norm": 0.1565333753824234,
+      "learning_rate": 9.100421252887621e-05,
+      "loss": 3.8915,
+      "step": 10600
+    },
+    {
+      "epoch": 0.7205462698736241,
+      "grad_norm": 0.16925933957099915,
+      "learning_rate": 9.099996602799294e-05,
+      "loss": 4.1202,
+      "step": 10605
+    },
+    {
+      "epoch": 0.7208859899442859,
+      "grad_norm": 0.13388794660568237,
+      "learning_rate": 9.099571952710967e-05,
+      "loss": 3.9878,
+      "step": 10610
+    },
+    {
+      "epoch": 0.7212257100149477,
+      "grad_norm": 0.15455901622772217,
+      "learning_rate": 9.09914730262264e-05,
+      "loss": 3.9719,
+      "step": 10615
+    },
+    {
+      "epoch": 0.7215654300856095,
+      "grad_norm": 0.19704410433769226,
+      "learning_rate": 9.098722652534312e-05,
+      "loss": 3.9981,
+      "step": 10620
+    },
+    {
+      "epoch": 0.7219051501562712,
+      "grad_norm": 0.1932808756828308,
+      "learning_rate": 9.098298002445985e-05,
+      "loss": 4.0945,
+      "step": 10625
+    },
+    {
+      "epoch": 0.722244870226933,
+      "grad_norm": 0.2587969899177551,
+      "learning_rate": 9.097873352357658e-05,
+      "loss": 4.0047,
+      "step": 10630
+    },
+    {
+      "epoch": 0.7225845902975948,
+      "grad_norm": 0.24855893850326538,
+      "learning_rate": 9.09744870226933e-05,
+      "loss": 3.927,
+      "step": 10635
+    },
+    {
+      "epoch": 0.7229243103682566,
+      "grad_norm": 0.2058570235967636,
+      "learning_rate": 9.097024052181003e-05,
+      "loss": 4.0545,
+      "step": 10640
+    },
+    {
+      "epoch": 0.7232640304389183,
+      "grad_norm": 0.20533056557178497,
+      "learning_rate": 9.096599402092676e-05,
+      "loss": 4.0696,
+      "step": 10645
+    },
+    {
+      "epoch": 0.7236037505095801,
+      "grad_norm": 0.2629019320011139,
+      "learning_rate": 9.096174752004349e-05,
+      "loss": 3.9769,
+      "step": 10650
+    },
+    {
+      "epoch": 0.7239434705802419,
+      "grad_norm": 0.2127770483493805,
+      "learning_rate": 9.095750101916022e-05,
+      "loss": 3.9245,
+      "step": 10655
+    },
+    {
+      "epoch": 0.7242831906509036,
+      "grad_norm": 0.1867874562740326,
+      "learning_rate": 9.095325451827695e-05,
+      "loss": 3.5818,
+      "step": 10660
+    },
+    {
+      "epoch": 0.7246229107215655,
+      "grad_norm": 0.25175002217292786,
+      "learning_rate": 9.094900801739366e-05,
+      "loss": 4.0028,
+      "step": 10665
+    },
+    {
+      "epoch": 0.7249626307922272,
+      "grad_norm": 0.18207040429115295,
+      "learning_rate": 9.09447615165104e-05,
+      "loss": 3.9293,
+      "step": 10670
+    },
+    {
+      "epoch": 0.725302350862889,
+      "grad_norm": 0.2175348699092865,
+      "learning_rate": 9.094051501562713e-05,
+      "loss": 4.116,
+      "step": 10675
+    },
+    {
+      "epoch": 0.7256420709335507,
+      "grad_norm": 0.1736600250005722,
+      "learning_rate": 9.093626851474386e-05,
+      "loss": 4.0422,
+      "step": 10680
+    },
+    {
+      "epoch": 0.7259817910042126,
+      "grad_norm": 0.2036193609237671,
+      "learning_rate": 9.093202201386059e-05,
+      "loss": 4.0416,
+      "step": 10685
+    },
+    {
+      "epoch": 0.7263215110748743,
+      "grad_norm": 0.5849753618240356,
+      "learning_rate": 9.092777551297731e-05,
+      "loss": 3.9602,
+      "step": 10690
+    },
+    {
+      "epoch": 0.726661231145536,
+      "grad_norm": 0.15708011388778687,
+      "learning_rate": 9.092352901209404e-05,
+      "loss": 3.9746,
+      "step": 10695
+    },
+    {
+      "epoch": 0.7270009512161979,
+      "grad_norm": 0.21586155891418457,
+      "learning_rate": 9.091928251121077e-05,
+      "loss": 4.1214,
+      "step": 10700
+    },
+    {
+      "epoch": 0.7273406712868596,
+      "grad_norm": 0.19140325486660004,
+      "learning_rate": 9.09150360103275e-05,
+      "loss": 4.0866,
+      "step": 10705
+    },
+    {
+      "epoch": 0.7276803913575214,
+      "grad_norm": 0.6538243889808655,
+      "learning_rate": 9.091078950944423e-05,
+      "loss": 4.1316,
+      "step": 10710
+    },
+    {
+      "epoch": 0.7280201114281831,
+      "grad_norm": 0.1779155284166336,
+      "learning_rate": 9.090654300856095e-05,
+      "loss": 4.0354,
+      "step": 10715
+    },
+    {
+      "epoch": 0.728359831498845,
+      "grad_norm": 0.24357867240905762,
+      "learning_rate": 9.090229650767768e-05,
+      "loss": 3.7766,
+      "step": 10720
+    },
+    {
+      "epoch": 0.7286995515695067,
+      "grad_norm": 0.15026213228702545,
+      "learning_rate": 9.089805000679441e-05,
+      "loss": 3.8955,
+      "step": 10725
+    },
+    {
+      "epoch": 0.7290392716401685,
+      "grad_norm": 0.19349145889282227,
+      "learning_rate": 9.089380350591114e-05,
+      "loss": 3.7735,
+      "step": 10730
+    },
+    {
+      "epoch": 0.7293789917108303,
+      "grad_norm": 0.25646016001701355,
+      "learning_rate": 9.088955700502785e-05,
+      "loss": 4.0811,
+      "step": 10735
+    },
+    {
+      "epoch": 0.7297187117814921,
+      "grad_norm": 0.17351272702217102,
+      "learning_rate": 9.08853105041446e-05,
+      "loss": 4.1555,
+      "step": 10740
+    },
+    {
+      "epoch": 0.7300584318521538,
+      "grad_norm": 0.17100609838962555,
+      "learning_rate": 9.088106400326132e-05,
+      "loss": 4.0126,
+      "step": 10745
+    },
+    {
+      "epoch": 0.7303981519228157,
+      "grad_norm": 0.25536659359931946,
+      "learning_rate": 9.087681750237804e-05,
+      "loss": 4.0176,
+      "step": 10750
+    },
+    {
+      "epoch": 0.7307378719934774,
+      "grad_norm": 0.2601194977760315,
+      "learning_rate": 9.087257100149478e-05,
+      "loss": 4.0722,
+      "step": 10755
+    },
+    {
+      "epoch": 0.7310775920641391,
+      "grad_norm": 0.19794826209545135,
+      "learning_rate": 9.08683245006115e-05,
+      "loss": 4.2004,
+      "step": 10760
+    },
+    {
+      "epoch": 0.7314173121348009,
+      "grad_norm": 0.2230055183172226,
+      "learning_rate": 9.086407799972822e-05,
+      "loss": 3.9865,
+      "step": 10765
+    },
+    {
+      "epoch": 0.7317570322054627,
+      "grad_norm": 0.24480870366096497,
+      "learning_rate": 9.085983149884496e-05,
+      "loss": 3.7441,
+      "step": 10770
+    },
+    {
+      "epoch": 0.7320967522761245,
+      "grad_norm": 0.15868893265724182,
+      "learning_rate": 9.085558499796169e-05,
+      "loss": 4.0704,
+      "step": 10775
+    },
+    {
+      "epoch": 0.7324364723467862,
+      "grad_norm": 0.3648226857185364,
+      "learning_rate": 9.08513384970784e-05,
+      "loss": 4.0741,
+      "step": 10780
+    },
+    {
+      "epoch": 0.7327761924174481,
+      "grad_norm": 1.1779170036315918,
+      "learning_rate": 9.084709199619515e-05,
+      "loss": 4.0508,
+      "step": 10785
+    },
+    {
+      "epoch": 0.7331159124881098,
+      "grad_norm": 0.5466019511222839,
+      "learning_rate": 9.084284549531187e-05,
+      "loss": 3.7824,
+      "step": 10790
+    },
+    {
+      "epoch": 0.7334556325587716,
+      "grad_norm": 0.2697416841983795,
+      "learning_rate": 9.083859899442859e-05,
+      "loss": 3.8887,
+      "step": 10795
+    },
+    {
+      "epoch": 0.7337953526294333,
+      "grad_norm": 0.1877298504114151,
+      "learning_rate": 9.083435249354533e-05,
+      "loss": 4.0495,
+      "step": 10800
+    },
+    {
+      "epoch": 0.7341350727000951,
+      "grad_norm": 0.21535362303256989,
+      "learning_rate": 9.083010599266204e-05,
+      "loss": 4.1075,
+      "step": 10805
+    },
+    {
+      "epoch": 0.7344747927707569,
+      "grad_norm": 0.15432433784008026,
+      "learning_rate": 9.082585949177877e-05,
+      "loss": 4.1321,
+      "step": 10810
+    },
+    {
+      "epoch": 0.7348145128414186,
+      "grad_norm": 0.17613931000232697,
+      "learning_rate": 9.082161299089551e-05,
+      "loss": 4.0152,
+      "step": 10815
+    },
+    {
+      "epoch": 0.7351542329120805,
+      "grad_norm": 0.17943201959133148,
+      "learning_rate": 9.081736649001223e-05,
+      "loss": 4.0291,
+      "step": 10820
+    },
+    {
+      "epoch": 0.7354939529827422,
+      "grad_norm": 0.17358125746250153,
+      "learning_rate": 9.081311998912896e-05,
+      "loss": 4.0228,
+      "step": 10825
+    },
+    {
+      "epoch": 0.735833673053404,
+      "grad_norm": 0.16327157616615295,
+      "learning_rate": 9.08088734882457e-05,
+      "loss": 3.8393,
+      "step": 10830
+    },
+    {
+      "epoch": 0.7361733931240658,
+      "grad_norm": 0.38127797842025757,
+      "learning_rate": 9.080462698736241e-05,
+      "loss": 4.0771,
+      "step": 10835
+    },
+    {
+      "epoch": 0.7365131131947276,
+      "grad_norm": 0.17917278409004211,
+      "learning_rate": 9.080038048647914e-05,
+      "loss": 3.9258,
+      "step": 10840
+    },
+    {
+      "epoch": 0.7368528332653893,
+      "grad_norm": 0.1688838005065918,
+      "learning_rate": 9.079613398559588e-05,
+      "loss": 3.8387,
+      "step": 10845
+    },
+    {
+      "epoch": 0.737192553336051,
+      "grad_norm": 0.1823907047510147,
+      "learning_rate": 9.07918874847126e-05,
+      "loss": 3.9959,
+      "step": 10850
+    },
+    {
+      "epoch": 0.7375322734067129,
+      "grad_norm": 0.20357432961463928,
+      "learning_rate": 9.078764098382932e-05,
+      "loss": 3.99,
+      "step": 10855
+    },
+    {
+      "epoch": 0.7378719934773746,
+      "grad_norm": 0.18823125958442688,
+      "learning_rate": 9.078339448294607e-05,
+      "loss": 4.0908,
+      "step": 10860
+    },
+    {
+      "epoch": 0.7382117135480364,
+      "grad_norm": 0.16174191236495972,
+      "learning_rate": 9.077914798206278e-05,
+      "loss": 4.0641,
+      "step": 10865
+    },
+    {
+      "epoch": 0.7385514336186982,
+      "grad_norm": 0.1720336228609085,
+      "learning_rate": 9.077490148117951e-05,
+      "loss": 3.8454,
+      "step": 10870
+    },
+    {
+      "epoch": 0.73889115368936,
+      "grad_norm": 0.23603537678718567,
+      "learning_rate": 9.077065498029625e-05,
+      "loss": 3.9752,
+      "step": 10875
+    },
+    {
+      "epoch": 0.7392308737600217,
+      "grad_norm": 0.20614124834537506,
+      "learning_rate": 9.076640847941296e-05,
+      "loss": 4.045,
+      "step": 10880
+    },
+    {
+      "epoch": 0.7395705938306835,
+      "grad_norm": 0.30947062373161316,
+      "learning_rate": 9.076216197852969e-05,
+      "loss": 3.9429,
+      "step": 10885
+    },
+    {
+      "epoch": 0.7399103139013453,
+      "grad_norm": 0.2017713338136673,
+      "learning_rate": 9.075791547764642e-05,
+      "loss": 3.9205,
+      "step": 10890
+    },
+    {
+      "epoch": 0.7402500339720071,
+      "grad_norm": 1.3917611837387085,
+      "learning_rate": 9.075366897676315e-05,
+      "loss": 4.0456,
+      "step": 10895
+    },
+    {
+      "epoch": 0.7405897540426688,
+      "grad_norm": 0.4103597104549408,
+      "learning_rate": 9.074942247587988e-05,
+      "loss": 3.9107,
+      "step": 10900
+    },
+    {
+      "epoch": 0.7409294741133307,
+      "grad_norm": 0.5144510269165039,
+      "learning_rate": 9.07451759749966e-05,
+      "loss": 4.0178,
+      "step": 10905
+    },
+    {
+      "epoch": 0.7412691941839924,
+      "grad_norm": 0.16965581476688385,
+      "learning_rate": 9.074092947411333e-05,
+      "loss": 3.9729,
+      "step": 10910
+    },
+    {
+      "epoch": 0.7416089142546541,
+      "grad_norm": 0.38037505745887756,
+      "learning_rate": 9.073668297323006e-05,
+      "loss": 4.2043,
+      "step": 10915
+    },
+    {
+      "epoch": 0.741948634325316,
+      "grad_norm": 0.26255086064338684,
+      "learning_rate": 9.073243647234679e-05,
+      "loss": 3.8663,
+      "step": 10920
+    },
+    {
+      "epoch": 0.7422883543959777,
+      "grad_norm": 0.3262099623680115,
+      "learning_rate": 9.072818997146352e-05,
+      "loss": 4.0803,
+      "step": 10925
+    },
+    {
+      "epoch": 0.7426280744666395,
+      "grad_norm": 0.21773168444633484,
+      "learning_rate": 9.072394347058024e-05,
+      "loss": 3.9992,
+      "step": 10930
+    },
+    {
+      "epoch": 0.7429677945373012,
+      "grad_norm": 0.22857216000556946,
+      "learning_rate": 9.071969696969697e-05,
+      "loss": 3.8582,
+      "step": 10935
+    },
+    {
+      "epoch": 0.7433075146079631,
+      "grad_norm": 0.312259316444397,
+      "learning_rate": 9.07154504688137e-05,
+      "loss": 4.0318,
+      "step": 10940
+    },
+    {
+      "epoch": 0.7436472346786248,
+      "grad_norm": 0.1695690155029297,
+      "learning_rate": 9.071120396793043e-05,
+      "loss": 3.9077,
+      "step": 10945
+    },
+    {
+      "epoch": 0.7439869547492866,
+      "grad_norm": 0.29498061537742615,
+      "learning_rate": 9.070695746704716e-05,
+      "loss": 4.3087,
+      "step": 10950
+    },
+    {
+      "epoch": 0.7443266748199484,
+      "grad_norm": 0.24566805362701416,
+      "learning_rate": 9.070271096616388e-05,
+      "loss": 3.8372,
+      "step": 10955
+    },
+    {
+      "epoch": 0.7446663948906102,
+      "grad_norm": 0.163113072514534,
+      "learning_rate": 9.069846446528061e-05,
+      "loss": 4.0402,
+      "step": 10960
+    },
+    {
+      "epoch": 0.7450061149612719,
+      "grad_norm": 0.18011754751205444,
+      "learning_rate": 9.069421796439734e-05,
+      "loss": 4.1377,
+      "step": 10965
+    },
+    {
+      "epoch": 0.7453458350319336,
+      "grad_norm": 0.8807979822158813,
+      "learning_rate": 9.068997146351407e-05,
+      "loss": 3.8495,
+      "step": 10970
+    },
+    {
+      "epoch": 0.7456855551025955,
+      "grad_norm": 0.22865957021713257,
+      "learning_rate": 9.06857249626308e-05,
+      "loss": 4.1146,
+      "step": 10975
+    },
+    {
+      "epoch": 0.7460252751732572,
+      "grad_norm": 0.2118086814880371,
+      "learning_rate": 9.068147846174752e-05,
+      "loss": 3.8964,
+      "step": 10980
+    },
+    {
+      "epoch": 0.746364995243919,
+      "grad_norm": 0.18207617104053497,
+      "learning_rate": 9.067723196086425e-05,
+      "loss": 3.9706,
+      "step": 10985
+    },
+    {
+      "epoch": 0.7467047153145808,
+      "grad_norm": 0.16859905421733856,
+      "learning_rate": 9.067298545998098e-05,
+      "loss": 3.7855,
+      "step": 10990
+    },
+    {
+      "epoch": 0.7470444353852426,
+      "grad_norm": 0.16500097513198853,
+      "learning_rate": 9.066873895909771e-05,
+      "loss": 3.7684,
+      "step": 10995
+    },
+    {
+      "epoch": 0.7473841554559043,
+      "grad_norm": 0.1520179957151413,
+      "learning_rate": 9.066449245821444e-05,
+      "loss": 3.8084,
+      "step": 11000
+    },
+    {
+      "epoch": 0.7477238755265662,
+      "grad_norm": 0.21755331754684448,
+      "learning_rate": 9.066024595733115e-05,
+      "loss": 3.8863,
+      "step": 11005
+    },
+    {
+      "epoch": 0.7480635955972279,
+      "grad_norm": 0.20671890676021576,
+      "learning_rate": 9.065599945644789e-05,
+      "loss": 3.9006,
+      "step": 11010
+    },
+    {
+      "epoch": 0.7484033156678896,
+      "grad_norm": 0.16787393391132355,
+      "learning_rate": 9.065175295556462e-05,
+      "loss": 3.7611,
+      "step": 11015
+    },
+    {
+      "epoch": 0.7487430357385514,
+      "grad_norm": 0.22157283127307892,
+      "learning_rate": 9.064750645468135e-05,
+      "loss": 4.1048,
+      "step": 11020
+    },
+    {
+      "epoch": 0.7490827558092132,
+      "grad_norm": 0.22022277116775513,
+      "learning_rate": 9.064325995379808e-05,
+      "loss": 3.8746,
+      "step": 11025
+    },
+    {
+      "epoch": 0.749422475879875,
+      "grad_norm": 0.2435934692621231,
+      "learning_rate": 9.06390134529148e-05,
+      "loss": 3.9949,
+      "step": 11030
+    },
+    {
+      "epoch": 0.7497621959505367,
+      "grad_norm": 0.18187767267227173,
+      "learning_rate": 9.063476695203153e-05,
+      "loss": 4.0584,
+      "step": 11035
+    },
+    {
+      "epoch": 0.7501019160211986,
+      "grad_norm": 0.18477857112884521,
+      "learning_rate": 9.063052045114826e-05,
+      "loss": 4.1217,
+      "step": 11040
+    },
+    {
+      "epoch": 0.7504416360918603,
+      "grad_norm": 0.1471758335828781,
+      "learning_rate": 9.062627395026499e-05,
+      "loss": 4.0513,
+      "step": 11045
+    },
+    {
+      "epoch": 0.7507813561625221,
+      "grad_norm": 0.20632903277873993,
+      "learning_rate": 9.062202744938172e-05,
+      "loss": 3.9865,
+      "step": 11050
+    },
+    {
+      "epoch": 0.7511210762331838,
+      "grad_norm": 0.21105721592903137,
+      "learning_rate": 9.061778094849844e-05,
+      "loss": 3.8277,
+      "step": 11055
+    },
+    {
+      "epoch": 0.7514607963038457,
+      "grad_norm": 0.19280456006526947,
+      "learning_rate": 9.061353444761517e-05,
+      "loss": 4.0091,
+      "step": 11060
+    },
+    {
+      "epoch": 0.7518005163745074,
+      "grad_norm": 0.1918146163225174,
+      "learning_rate": 9.06092879467319e-05,
+      "loss": 4.0648,
+      "step": 11065
+    },
+    {
+      "epoch": 0.7521402364451691,
+      "grad_norm": 0.22963494062423706,
+      "learning_rate": 9.060504144584863e-05,
+      "loss": 3.9061,
+      "step": 11070
+    },
+    {
+      "epoch": 0.752479956515831,
+      "grad_norm": 0.16479997336864471,
+      "learning_rate": 9.060079494496536e-05,
+      "loss": 4.0303,
+      "step": 11075
+    },
+    {
+      "epoch": 0.7528196765864927,
+      "grad_norm": 0.18432816863059998,
+      "learning_rate": 9.059654844408208e-05,
+      "loss": 3.8773,
+      "step": 11080
+    },
+    {
+      "epoch": 0.7531593966571545,
+      "grad_norm": 0.22336050868034363,
+      "learning_rate": 9.059230194319881e-05,
+      "loss": 3.7997,
+      "step": 11085
+    },
+    {
+      "epoch": 0.7534991167278163,
+      "grad_norm": 0.242068812251091,
+      "learning_rate": 9.058805544231553e-05,
+      "loss": 3.9268,
+      "step": 11090
+    },
+    {
+      "epoch": 0.7538388367984781,
+      "grad_norm": 0.14753904938697815,
+      "learning_rate": 9.058380894143227e-05,
+      "loss": 4.0985,
+      "step": 11095
+    },
+    {
+      "epoch": 0.7541785568691398,
+      "grad_norm": 0.19245490431785583,
+      "learning_rate": 9.0579562440549e-05,
+      "loss": 3.8995,
+      "step": 11100
+    },
+    {
+      "epoch": 0.7545182769398016,
+      "grad_norm": 0.18615277111530304,
+      "learning_rate": 9.057531593966571e-05,
+      "loss": 4.0072,
+      "step": 11105
+    },
+    {
+      "epoch": 0.7548579970104634,
+      "grad_norm": 0.19581812620162964,
+      "learning_rate": 9.057106943878245e-05,
+      "loss": 3.8758,
+      "step": 11110
+    },
+    {
+      "epoch": 0.7551977170811252,
+      "grad_norm": 0.15949614346027374,
+      "learning_rate": 9.056682293789918e-05,
+      "loss": 3.9473,
+      "step": 11115
+    },
+    {
+      "epoch": 0.7555374371517869,
+      "grad_norm": 1.359558343887329,
+      "learning_rate": 9.05625764370159e-05,
+      "loss": 3.9416,
+      "step": 11120
+    },
+    {
+      "epoch": 0.7558771572224487,
+      "grad_norm": 0.1593676656484604,
+      "learning_rate": 9.055832993613264e-05,
+      "loss": 4.0518,
+      "step": 11125
+    },
+    {
+      "epoch": 0.7562168772931105,
+      "grad_norm": 0.1715662181377411,
+      "learning_rate": 9.055408343524937e-05,
+      "loss": 3.982,
+      "step": 11130
+    },
+    {
+      "epoch": 0.7565565973637722,
+      "grad_norm": 0.1934783011674881,
+      "learning_rate": 9.054983693436608e-05,
+      "loss": 4.1628,
+      "step": 11135
+    },
+    {
+      "epoch": 0.756896317434434,
+      "grad_norm": 0.16365660727024078,
+      "learning_rate": 9.054559043348282e-05,
+      "loss": 4.0758,
+      "step": 11140
+    },
+    {
+      "epoch": 0.7572360375050958,
+      "grad_norm": 0.2995030879974365,
+      "learning_rate": 9.054134393259955e-05,
+      "loss": 3.915,
+      "step": 11145
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.22450530529022217,
+      "learning_rate": 9.053709743171626e-05,
+      "loss": 4.2514,
+      "step": 11150
+    },
+    {
+      "epoch": 0.7579154776464193,
+      "grad_norm": 3.4520628452301025,
+      "learning_rate": 9.0532850930833e-05,
+      "loss": 4.0989,
+      "step": 11155
+    },
+    {
+      "epoch": 0.7582551977170812,
+      "grad_norm": 0.31930363178253174,
+      "learning_rate": 9.052860442994972e-05,
+      "loss": 4.0048,
+      "step": 11160
+    },
+    {
+      "epoch": 0.7585949177877429,
+      "grad_norm": 0.21884280443191528,
+      "learning_rate": 9.052435792906645e-05,
+      "loss": 3.8112,
+      "step": 11165
+    },
+    {
+      "epoch": 0.7589346378584046,
+      "grad_norm": 0.1630697101354599,
+      "learning_rate": 9.052011142818319e-05,
+      "loss": 3.9614,
+      "step": 11170
+    },
+    {
+      "epoch": 0.7592743579290665,
+      "grad_norm": 0.2021295428276062,
+      "learning_rate": 9.05158649272999e-05,
+      "loss": 3.9433,
+      "step": 11175
+    },
+    {
+      "epoch": 0.7596140779997282,
+      "grad_norm": 0.21674887835979462,
+      "learning_rate": 9.051161842641663e-05,
+      "loss": 3.9271,
+      "step": 11180
+    },
+    {
+      "epoch": 0.75995379807039,
+      "grad_norm": 0.2624286413192749,
+      "learning_rate": 9.050737192553337e-05,
+      "loss": 3.8687,
+      "step": 11185
+    },
+    {
+      "epoch": 0.7602935181410517,
+      "grad_norm": 0.15679609775543213,
+      "learning_rate": 9.050312542465009e-05,
+      "loss": 4.0386,
+      "step": 11190
+    },
+    {
+      "epoch": 0.7606332382117136,
+      "grad_norm": 0.298999547958374,
+      "learning_rate": 9.049887892376682e-05,
+      "loss": 4.039,
+      "step": 11195
+    },
+    {
+      "epoch": 0.7609729582823753,
+      "grad_norm": 0.1916528344154358,
+      "learning_rate": 9.049463242288356e-05,
+      "loss": 3.9924,
+      "step": 11200
+    },
+    {
+      "epoch": 0.7613126783530371,
+      "grad_norm": 0.20813806354999542,
+      "learning_rate": 9.049038592200027e-05,
+      "loss": 3.9728,
+      "step": 11205
+    },
+    {
+      "epoch": 0.7616523984236989,
+      "grad_norm": 0.26955127716064453,
+      "learning_rate": 9.0486139421117e-05,
+      "loss": 3.9563,
+      "step": 11210
+    },
+    {
+      "epoch": 0.7619921184943607,
+      "grad_norm": 0.22610753774642944,
+      "learning_rate": 9.048189292023374e-05,
+      "loss": 3.9385,
+      "step": 11215
+    },
+    {
+      "epoch": 0.7623318385650224,
+      "grad_norm": 0.1515149027109146,
+      "learning_rate": 9.047764641935046e-05,
+      "loss": 4.0009,
+      "step": 11220
+    },
+    {
+      "epoch": 0.7626715586356841,
+      "grad_norm": 0.15270265936851501,
+      "learning_rate": 9.047339991846718e-05,
+      "loss": 3.8132,
+      "step": 11225
+    },
+    {
+      "epoch": 0.763011278706346,
+      "grad_norm": 0.256085067987442,
+      "learning_rate": 9.046915341758391e-05,
+      "loss": 4.1206,
+      "step": 11230
+    },
+    {
+      "epoch": 0.7633509987770077,
+      "grad_norm": 0.19662001729011536,
+      "learning_rate": 9.046490691670064e-05,
+      "loss": 4.0799,
+      "step": 11235
+    },
+    {
+      "epoch": 0.7636907188476695,
+      "grad_norm": 0.18421690165996552,
+      "learning_rate": 9.046066041581737e-05,
+      "loss": 4.0328,
+      "step": 11240
+    },
+    {
+      "epoch": 0.7640304389183313,
+      "grad_norm": 0.19748954474925995,
+      "learning_rate": 9.04564139149341e-05,
+      "loss": 3.7995,
+      "step": 11245
+    },
+    {
+      "epoch": 0.7643701589889931,
+      "grad_norm": 0.15954630076885223,
+      "learning_rate": 9.045216741405082e-05,
+      "loss": 3.8286,
+      "step": 11250
+    },
+    {
+      "epoch": 0.7647098790596548,
+      "grad_norm": 0.5489984154701233,
+      "learning_rate": 9.044792091316755e-05,
+      "loss": 4.2052,
+      "step": 11255
+    },
+    {
+      "epoch": 0.7650495991303167,
+      "grad_norm": 0.44495344161987305,
+      "learning_rate": 9.044367441228428e-05,
+      "loss": 4.1036,
+      "step": 11260
+    },
+    {
+      "epoch": 0.7653893192009784,
+      "grad_norm": 0.16555814445018768,
+      "learning_rate": 9.043942791140101e-05,
+      "loss": 3.9737,
+      "step": 11265
+    },
+    {
+      "epoch": 0.7657290392716402,
+      "grad_norm": 0.21692036092281342,
+      "learning_rate": 9.043518141051774e-05,
+      "loss": 4.0437,
+      "step": 11270
+    },
+    {
+      "epoch": 0.7660687593423019,
+      "grad_norm": 1.8791022300720215,
+      "learning_rate": 9.043093490963446e-05,
+      "loss": 4.0974,
+      "step": 11275
+    },
+    {
+      "epoch": 0.7664084794129638,
+      "grad_norm": 0.19241303205490112,
+      "learning_rate": 9.042668840875119e-05,
+      "loss": 3.9209,
+      "step": 11280
+    },
+    {
+      "epoch": 0.7667481994836255,
+      "grad_norm": 0.17803336679935455,
+      "learning_rate": 9.042244190786792e-05,
+      "loss": 3.842,
+      "step": 11285
+    },
+    {
+      "epoch": 0.7670879195542872,
+      "grad_norm": 0.20129168033599854,
+      "learning_rate": 9.041819540698465e-05,
+      "loss": 4.0462,
+      "step": 11290
+    },
+    {
+      "epoch": 0.7674276396249491,
+      "grad_norm": 0.18283264338970184,
+      "learning_rate": 9.041394890610138e-05,
+      "loss": 3.9094,
+      "step": 11295
+    },
+    {
+      "epoch": 0.7677673596956108,
+      "grad_norm": 0.20721754431724548,
+      "learning_rate": 9.04097024052181e-05,
+      "loss": 3.9013,
+      "step": 11300
+    },
+    {
+      "epoch": 0.7681070797662726,
+      "grad_norm": 0.43089064955711365,
+      "learning_rate": 9.040545590433483e-05,
+      "loss": 3.964,
+      "step": 11305
+    },
+    {
+      "epoch": 0.7684467998369343,
+      "grad_norm": 0.24873760342597961,
+      "learning_rate": 9.040120940345156e-05,
+      "loss": 4.0062,
+      "step": 11310
+    },
+    {
+      "epoch": 0.7687865199075962,
+      "grad_norm": 0.22243714332580566,
+      "learning_rate": 9.039696290256829e-05,
+      "loss": 3.8801,
+      "step": 11315
+    },
+    {
+      "epoch": 0.7691262399782579,
+      "grad_norm": 0.41750073432922363,
+      "learning_rate": 9.039271640168502e-05,
+      "loss": 4.2682,
+      "step": 11320
+    },
+    {
+      "epoch": 0.7694659600489197,
+      "grad_norm": 0.369742214679718,
+      "learning_rate": 9.038846990080174e-05,
+      "loss": 3.8898,
+      "step": 11325
+    },
+    {
+      "epoch": 0.7698056801195815,
+      "grad_norm": 0.26420828700065613,
+      "learning_rate": 9.038422339991847e-05,
+      "loss": 3.9038,
+      "step": 11330
+    },
+    {
+      "epoch": 0.7701454001902432,
+      "grad_norm": 0.2597283720970154,
+      "learning_rate": 9.03799768990352e-05,
+      "loss": 3.9406,
+      "step": 11335
+    },
+    {
+      "epoch": 0.770485120260905,
+      "grad_norm": 0.17518769204616547,
+      "learning_rate": 9.037573039815193e-05,
+      "loss": 3.6741,
+      "step": 11340
+    },
+    {
+      "epoch": 0.7708248403315668,
+      "grad_norm": 0.6777191758155823,
+      "learning_rate": 9.037148389726866e-05,
+      "loss": 4.0673,
+      "step": 11345
+    },
+    {
+      "epoch": 0.7711645604022286,
+      "grad_norm": 0.201960951089859,
+      "learning_rate": 9.036723739638538e-05,
+      "loss": 4.0074,
+      "step": 11350
+    },
+    {
+      "epoch": 0.7715042804728903,
+      "grad_norm": 0.4381665587425232,
+      "learning_rate": 9.036299089550211e-05,
+      "loss": 3.8382,
+      "step": 11355
+    },
+    {
+      "epoch": 0.7718440005435521,
+      "grad_norm": 0.1966671347618103,
+      "learning_rate": 9.035874439461884e-05,
+      "loss": 4.1766,
+      "step": 11360
+    },
+    {
+      "epoch": 0.7721837206142139,
+      "grad_norm": 0.16876500844955444,
+      "learning_rate": 9.035449789373557e-05,
+      "loss": 3.9402,
+      "step": 11365
+    },
+    {
+      "epoch": 0.7725234406848757,
+      "grad_norm": 4.147640705108643,
+      "learning_rate": 9.03502513928523e-05,
+      "loss": 4.109,
+      "step": 11370
+    },
+    {
+      "epoch": 0.7728631607555374,
+      "grad_norm": 0.2072206437587738,
+      "learning_rate": 9.034600489196902e-05,
+      "loss": 4.0296,
+      "step": 11375
+    },
+    {
+      "epoch": 0.7732028808261993,
+      "grad_norm": 0.2016468346118927,
+      "learning_rate": 9.034175839108575e-05,
+      "loss": 3.7331,
+      "step": 11380
+    },
+    {
+      "epoch": 0.773542600896861,
+      "grad_norm": 0.47726747393608093,
+      "learning_rate": 9.033751189020248e-05,
+      "loss": 4.0178,
+      "step": 11385
+    },
+    {
+      "epoch": 0.7738823209675227,
+      "grad_norm": 0.17172425985336304,
+      "learning_rate": 9.033326538931921e-05,
+      "loss": 4.1782,
+      "step": 11390
+    },
+    {
+      "epoch": 0.7742220410381845,
+      "grad_norm": 0.1911281943321228,
+      "learning_rate": 9.032901888843594e-05,
+      "loss": 3.9972,
+      "step": 11395
+    },
+    {
+      "epoch": 0.7745617611088463,
+      "grad_norm": 0.14127899706363678,
+      "learning_rate": 9.032477238755266e-05,
+      "loss": 3.9405,
+      "step": 11400
+    },
+    {
+      "epoch": 0.7749014811795081,
+      "grad_norm": 0.1841440349817276,
+      "learning_rate": 9.032052588666939e-05,
+      "loss": 4.0964,
+      "step": 11405
+    },
+    {
+      "epoch": 0.7752412012501698,
+      "grad_norm": 0.17826926708221436,
+      "learning_rate": 9.031627938578612e-05,
+      "loss": 4.2658,
+      "step": 11410
+    },
+    {
+      "epoch": 0.7755809213208317,
+      "grad_norm": 0.2882947325706482,
+      "learning_rate": 9.031203288490285e-05,
+      "loss": 4.1508,
+      "step": 11415
+    },
+    {
+      "epoch": 0.7759206413914934,
+      "grad_norm": 0.21743571758270264,
+      "learning_rate": 9.030778638401958e-05,
+      "loss": 4.1659,
+      "step": 11420
+    },
+    {
+      "epoch": 0.7762603614621552,
+      "grad_norm": 0.23792824149131775,
+      "learning_rate": 9.03035398831363e-05,
+      "loss": 3.8291,
+      "step": 11425
+    },
+    {
+      "epoch": 0.776600081532817,
+      "grad_norm": 0.2318025827407837,
+      "learning_rate": 9.029929338225302e-05,
+      "loss": 3.8444,
+      "step": 11430
+    },
+    {
+      "epoch": 0.7769398016034788,
+      "grad_norm": 0.17949531972408295,
+      "learning_rate": 9.029504688136976e-05,
+      "loss": 3.9841,
+      "step": 11435
+    },
+    {
+      "epoch": 0.7772795216741405,
+      "grad_norm": 0.22435292601585388,
+      "learning_rate": 9.029080038048649e-05,
+      "loss": 3.9821,
+      "step": 11440
+    },
+    {
+      "epoch": 0.7776192417448022,
+      "grad_norm": 0.1865406632423401,
+      "learning_rate": 9.02865538796032e-05,
+      "loss": 3.935,
+      "step": 11445
+    },
+    {
+      "epoch": 0.7779589618154641,
+      "grad_norm": 0.2090293914079666,
+      "learning_rate": 9.028230737871994e-05,
+      "loss": 3.9991,
+      "step": 11450
+    },
+    {
+      "epoch": 0.7782986818861258,
+      "grad_norm": 0.18024842441082,
+      "learning_rate": 9.027806087783667e-05,
+      "loss": 4.076,
+      "step": 11455
+    },
+    {
+      "epoch": 0.7786384019567876,
+      "grad_norm": 0.17997018992900848,
+      "learning_rate": 9.027381437695339e-05,
+      "loss": 3.7585,
+      "step": 11460
+    },
+    {
+      "epoch": 0.7789781220274494,
+      "grad_norm": 0.16544857621192932,
+      "learning_rate": 9.026956787607013e-05,
+      "loss": 4.2184,
+      "step": 11465
+    },
+    {
+      "epoch": 0.7793178420981112,
+      "grad_norm": 0.17606359720230103,
+      "learning_rate": 9.026532137518686e-05,
+      "loss": 3.9412,
+      "step": 11470
+    },
+    {
+      "epoch": 0.7796575621687729,
+      "grad_norm": 0.2205812931060791,
+      "learning_rate": 9.026107487430357e-05,
+      "loss": 4.0843,
+      "step": 11475
+    },
+    {
+      "epoch": 0.7799972822394347,
+      "grad_norm": 0.25740867853164673,
+      "learning_rate": 9.025682837342031e-05,
+      "loss": 3.9544,
+      "step": 11480
+    },
+    {
+      "epoch": 0.7803370023100965,
+      "grad_norm": 0.14909543097019196,
+      "learning_rate": 9.025258187253704e-05,
+      "loss": 3.8342,
+      "step": 11485
+    },
+    {
+      "epoch": 0.7806767223807582,
+      "grad_norm": 0.24682089686393738,
+      "learning_rate": 9.024833537165375e-05,
+      "loss": 3.9944,
+      "step": 11490
+    },
+    {
+      "epoch": 0.78101644245142,
+      "grad_norm": 0.15707463026046753,
+      "learning_rate": 9.02440888707705e-05,
+      "loss": 3.9675,
+      "step": 11495
+    },
+    {
+      "epoch": 0.7813561625220818,
+      "grad_norm": 0.22718797624111176,
+      "learning_rate": 9.023984236988722e-05,
+      "loss": 3.6624,
+      "step": 11500
+    },
+    {
+      "epoch": 0.7816958825927436,
+      "grad_norm": 0.15948626399040222,
+      "learning_rate": 9.023559586900394e-05,
+      "loss": 3.9947,
+      "step": 11505
+    },
+    {
+      "epoch": 0.7820356026634053,
+      "grad_norm": 0.16061913967132568,
+      "learning_rate": 9.023134936812068e-05,
+      "loss": 4.1845,
+      "step": 11510
+    },
+    {
+      "epoch": 0.7823753227340672,
+      "grad_norm": 0.25919288396835327,
+      "learning_rate": 9.02271028672374e-05,
+      "loss": 3.9301,
+      "step": 11515
+    },
+    {
+      "epoch": 0.7827150428047289,
+      "grad_norm": 0.21657872200012207,
+      "learning_rate": 9.022285636635412e-05,
+      "loss": 4.0299,
+      "step": 11520
+    },
+    {
+      "epoch": 0.7830547628753907,
+      "grad_norm": 0.18826550245285034,
+      "learning_rate": 9.021860986547086e-05,
+      "loss": 4.212,
+      "step": 11525
+    },
+    {
+      "epoch": 0.7833944829460524,
+      "grad_norm": 0.2549474835395813,
+      "learning_rate": 9.021436336458758e-05,
+      "loss": 4.0705,
+      "step": 11530
+    },
+    {
+      "epoch": 0.7837342030167143,
+      "grad_norm": 0.6155955195426941,
+      "learning_rate": 9.02101168637043e-05,
+      "loss": 3.79,
+      "step": 11535
+    },
+    {
+      "epoch": 0.784073923087376,
+      "grad_norm": 0.1635499894618988,
+      "learning_rate": 9.020587036282105e-05,
+      "loss": 4.1024,
+      "step": 11540
+    },
+    {
+      "epoch": 0.7844136431580377,
+      "grad_norm": 0.15726587176322937,
+      "learning_rate": 9.020162386193776e-05,
+      "loss": 3.9499,
+      "step": 11545
+    },
+    {
+      "epoch": 0.7847533632286996,
+      "grad_norm": 0.16258913278579712,
+      "learning_rate": 9.019737736105449e-05,
+      "loss": 3.9322,
+      "step": 11550
+    },
+    {
+      "epoch": 0.7850930832993613,
+      "grad_norm": 0.2376587688922882,
+      "learning_rate": 9.019313086017123e-05,
+      "loss": 3.9946,
+      "step": 11555
+    },
+    {
+      "epoch": 0.7854328033700231,
+      "grad_norm": 0.1641000360250473,
+      "learning_rate": 9.018888435928795e-05,
+      "loss": 4.1131,
+      "step": 11560
+    },
+    {
+      "epoch": 0.7857725234406848,
+      "grad_norm": 0.18432609736919403,
+      "learning_rate": 9.018463785840467e-05,
+      "loss": 3.8617,
+      "step": 11565
+    },
+    {
+      "epoch": 0.7861122435113467,
+      "grad_norm": 0.31025978922843933,
+      "learning_rate": 9.018039135752142e-05,
+      "loss": 3.7286,
+      "step": 11570
+    },
+    {
+      "epoch": 0.7864519635820084,
+      "grad_norm": 0.18590706586837769,
+      "learning_rate": 9.017614485663813e-05,
+      "loss": 3.9636,
+      "step": 11575
+    },
+    {
+      "epoch": 0.7867916836526702,
+      "grad_norm": 0.18814896047115326,
+      "learning_rate": 9.017189835575486e-05,
+      "loss": 4.3321,
+      "step": 11580
+    },
+    {
+      "epoch": 0.787131403723332,
+      "grad_norm": 0.17569060623645782,
+      "learning_rate": 9.016765185487159e-05,
+      "loss": 4.0207,
+      "step": 11585
+    },
+    {
+      "epoch": 0.7874711237939938,
+      "grad_norm": 0.8084515333175659,
+      "learning_rate": 9.016340535398831e-05,
+      "loss": 3.8702,
+      "step": 11590
+    },
+    {
+      "epoch": 0.7878108438646555,
+      "grad_norm": 0.1719738245010376,
+      "learning_rate": 9.015915885310504e-05,
+      "loss": 3.9472,
+      "step": 11595
+    },
+    {
+      "epoch": 0.7881505639353173,
+      "grad_norm": 0.3733132481575012,
+      "learning_rate": 9.015491235222177e-05,
+      "loss": 3.9621,
+      "step": 11600
+    },
+    {
+      "epoch": 0.7884902840059791,
+      "grad_norm": 0.1931469440460205,
+      "learning_rate": 9.01506658513385e-05,
+      "loss": 4.1482,
+      "step": 11605
+    },
+    {
+      "epoch": 0.7888300040766408,
+      "grad_norm": 0.6097820997238159,
+      "learning_rate": 9.014641935045523e-05,
+      "loss": 4.2861,
+      "step": 11610
+    },
+    {
+      "epoch": 0.7891697241473026,
+      "grad_norm": 0.23092709481716156,
+      "learning_rate": 9.014217284957195e-05,
+      "loss": 3.7435,
+      "step": 11615
+    },
+    {
+      "epoch": 0.7895094442179644,
+      "grad_norm": 0.20437659323215485,
+      "learning_rate": 9.013792634868868e-05,
+      "loss": 4.0259,
+      "step": 11620
+    },
+    {
+      "epoch": 0.7898491642886262,
+      "grad_norm": 0.19561974704265594,
+      "learning_rate": 9.013367984780541e-05,
+      "loss": 3.8396,
+      "step": 11625
+    },
+    {
+      "epoch": 0.7901888843592879,
+      "grad_norm": 0.22799140214920044,
+      "learning_rate": 9.012943334692214e-05,
+      "loss": 4.0335,
+      "step": 11630
+    },
+    {
+      "epoch": 0.7905286044299498,
+      "grad_norm": 0.1820353865623474,
+      "learning_rate": 9.012518684603887e-05,
+      "loss": 3.9942,
+      "step": 11635
+    },
+    {
+      "epoch": 0.7908683245006115,
+      "grad_norm": 0.217819482088089,
+      "learning_rate": 9.01209403451556e-05,
+      "loss": 4.1374,
+      "step": 11640
+    },
+    {
+      "epoch": 0.7912080445712733,
+      "grad_norm": 0.20061899721622467,
+      "learning_rate": 9.011669384427232e-05,
+      "loss": 3.6462,
+      "step": 11645
+    },
+    {
+      "epoch": 0.791547764641935,
+      "grad_norm": 0.21914707124233246,
+      "learning_rate": 9.011244734338905e-05,
+      "loss": 4.0377,
+      "step": 11650
+    },
+    {
+      "epoch": 0.7918874847125968,
+      "grad_norm": 0.2225886732339859,
+      "learning_rate": 9.010820084250578e-05,
+      "loss": 4.1587,
+      "step": 11655
+    },
+    {
+      "epoch": 0.7922272047832586,
+      "grad_norm": 0.23360738158226013,
+      "learning_rate": 9.01039543416225e-05,
+      "loss": 4.0995,
+      "step": 11660
+    },
+    {
+      "epoch": 0.7925669248539203,
+      "grad_norm": 0.20647506415843964,
+      "learning_rate": 9.009970784073923e-05,
+      "loss": 4.0943,
+      "step": 11665
+    },
+    {
+      "epoch": 0.7929066449245822,
+      "grad_norm": 0.17202545702457428,
+      "learning_rate": 9.009546133985596e-05,
+      "loss": 3.8883,
+      "step": 11670
+    },
+    {
+      "epoch": 0.7932463649952439,
+      "grad_norm": 1.380285382270813,
+      "learning_rate": 9.009121483897269e-05,
+      "loss": 4.0137,
+      "step": 11675
+    },
+    {
+      "epoch": 0.7935860850659057,
+      "grad_norm": 0.23098598420619965,
+      "learning_rate": 9.008696833808942e-05,
+      "loss": 4.0737,
+      "step": 11680
+    },
+    {
+      "epoch": 0.7939258051365675,
+      "grad_norm": 0.17068329453468323,
+      "learning_rate": 9.008272183720615e-05,
+      "loss": 4.2746,
+      "step": 11685
+    },
+    {
+      "epoch": 0.7942655252072293,
+      "grad_norm": 0.23422260582447052,
+      "learning_rate": 9.007847533632287e-05,
+      "loss": 3.7582,
+      "step": 11690
+    },
+    {
+      "epoch": 0.794605245277891,
+      "grad_norm": 0.1885872483253479,
+      "learning_rate": 9.00742288354396e-05,
+      "loss": 4.0525,
+      "step": 11695
+    },
+    {
+      "epoch": 0.7949449653485527,
+      "grad_norm": 0.18177750706672668,
+      "learning_rate": 9.006998233455633e-05,
+      "loss": 4.0782,
+      "step": 11700
+    },
+    {
+      "epoch": 0.7952846854192146,
+      "grad_norm": 2.576247453689575,
+      "learning_rate": 9.006573583367306e-05,
+      "loss": 3.8726,
+      "step": 11705
+    },
+    {
+      "epoch": 0.7956244054898763,
+      "grad_norm": 0.16896361112594604,
+      "learning_rate": 9.006148933278979e-05,
+      "loss": 3.8062,
+      "step": 11710
+    },
+    {
+      "epoch": 0.7959641255605381,
+      "grad_norm": 0.1680668294429779,
+      "learning_rate": 9.005724283190651e-05,
+      "loss": 4.0136,
+      "step": 11715
+    },
+    {
+      "epoch": 0.7963038456311999,
+      "grad_norm": 0.18001356720924377,
+      "learning_rate": 9.005299633102324e-05,
+      "loss": 4.1243,
+      "step": 11720
+    },
+    {
+      "epoch": 0.7966435657018617,
+      "grad_norm": 0.19494907557964325,
+      "learning_rate": 9.004874983013997e-05,
+      "loss": 3.9437,
+      "step": 11725
+    },
+    {
+      "epoch": 0.7969832857725234,
+      "grad_norm": 0.18916480243206024,
+      "learning_rate": 9.00445033292567e-05,
+      "loss": 4.0779,
+      "step": 11730
+    },
+    {
+      "epoch": 0.7973230058431852,
+      "grad_norm": 0.211675226688385,
+      "learning_rate": 9.004025682837343e-05,
+      "loss": 3.8902,
+      "step": 11735
+    },
+    {
+      "epoch": 0.797662725913847,
+      "grad_norm": 0.2676939368247986,
+      "learning_rate": 9.003601032749015e-05,
+      "loss": 4.0427,
+      "step": 11740
+    },
+    {
+      "epoch": 0.7980024459845088,
+      "grad_norm": 0.20862559974193573,
+      "learning_rate": 9.003176382660688e-05,
+      "loss": 4.2145,
+      "step": 11745
+    },
+    {
+      "epoch": 0.7983421660551705,
+      "grad_norm": 0.20464570820331573,
+      "learning_rate": 9.002751732572361e-05,
+      "loss": 4.113,
+      "step": 11750
+    },
+    {
+      "epoch": 0.7986818861258324,
+      "grad_norm": 0.17028920352458954,
+      "learning_rate": 9.002327082484034e-05,
+      "loss": 3.6957,
+      "step": 11755
+    },
+    {
+      "epoch": 0.7990216061964941,
+      "grad_norm": 0.24813637137413025,
+      "learning_rate": 9.001902432395707e-05,
+      "loss": 4.1226,
+      "step": 11760
+    },
+    {
+      "epoch": 0.7993613262671558,
+      "grad_norm": 1.6223915815353394,
+      "learning_rate": 9.00147778230738e-05,
+      "loss": 3.9943,
+      "step": 11765
+    },
+    {
+      "epoch": 0.7997010463378177,
+      "grad_norm": 0.1639162003993988,
+      "learning_rate": 9.001053132219052e-05,
+      "loss": 3.9359,
+      "step": 11770
+    },
+    {
+      "epoch": 0.8000407664084794,
+      "grad_norm": 0.16888979077339172,
+      "learning_rate": 9.000628482130725e-05,
+      "loss": 3.8101,
+      "step": 11775
+    },
+    {
+      "epoch": 0.8003804864791412,
+      "grad_norm": 0.1576785147190094,
+      "learning_rate": 9.000203832042398e-05,
+      "loss": 4.062,
+      "step": 11780
+    },
+    {
+      "epoch": 0.8007202065498029,
+      "grad_norm": 0.19945354759693146,
+      "learning_rate": 8.99977918195407e-05,
+      "loss": 4.2053,
+      "step": 11785
+    },
+    {
+      "epoch": 0.8010599266204648,
+      "grad_norm": 0.13953137397766113,
+      "learning_rate": 8.999354531865743e-05,
+      "loss": 3.6578,
+      "step": 11790
+    },
+    {
+      "epoch": 0.8013996466911265,
+      "grad_norm": 0.1995120495557785,
+      "learning_rate": 8.998929881777416e-05,
+      "loss": 3.9994,
+      "step": 11795
+    },
+    {
+      "epoch": 0.8017393667617883,
+      "grad_norm": 0.22360244393348694,
+      "learning_rate": 8.998505231689088e-05,
+      "loss": 4.1224,
+      "step": 11800
+    },
+    {
+      "epoch": 0.8020790868324501,
+      "grad_norm": 0.20481501519680023,
+      "learning_rate": 8.998080581600762e-05,
+      "loss": 4.2824,
+      "step": 11805
+    },
+    {
+      "epoch": 0.8024188069031118,
+      "grad_norm": 0.39974507689476013,
+      "learning_rate": 8.997655931512435e-05,
+      "loss": 4.1875,
+      "step": 11810
+    },
+    {
+      "epoch": 0.8027585269737736,
+      "grad_norm": 0.32297125458717346,
+      "learning_rate": 8.997231281424106e-05,
+      "loss": 3.8753,
+      "step": 11815
+    },
+    {
+      "epoch": 0.8030982470444353,
+      "grad_norm": 0.2076197862625122,
+      "learning_rate": 8.996891561353446e-05,
+      "loss": 4.0354,
+      "step": 11820
+    },
+    {
+      "epoch": 0.8034379671150972,
+      "grad_norm": 0.17974001169204712,
+      "learning_rate": 8.996466911265119e-05,
+      "loss": 4.0395,
+      "step": 11825
+    },
+    {
+      "epoch": 0.8037776871857589,
+      "grad_norm": 0.18468719720840454,
+      "learning_rate": 8.99604226117679e-05,
+      "loss": 4.0589,
+      "step": 11830
+    },
+    {
+      "epoch": 0.8041174072564207,
+      "grad_norm": 0.2775026857852936,
+      "learning_rate": 8.995617611088464e-05,
+      "loss": 4.0899,
+      "step": 11835
+    },
+    {
+      "epoch": 0.8044571273270825,
+      "grad_norm": 0.20769546926021576,
+      "learning_rate": 8.995192961000136e-05,
+      "loss": 4.1438,
+      "step": 11840
+    },
+    {
+      "epoch": 0.8047968473977443,
+      "grad_norm": 0.18603581190109253,
+      "learning_rate": 8.994768310911808e-05,
+      "loss": 3.9822,
+      "step": 11845
+    },
+    {
+      "epoch": 0.805136567468406,
+      "grad_norm": 0.3347295820713043,
+      "learning_rate": 8.994343660823483e-05,
+      "loss": 4.042,
+      "step": 11850
+    },
+    {
+      "epoch": 0.8054762875390679,
+      "grad_norm": 0.26557305455207825,
+      "learning_rate": 8.993919010735154e-05,
+      "loss": 3.9212,
+      "step": 11855
+    },
+    {
+      "epoch": 0.8058160076097296,
+      "grad_norm": 0.27433109283447266,
+      "learning_rate": 8.993494360646827e-05,
+      "loss": 3.7039,
+      "step": 11860
+    },
+    {
+      "epoch": 0.8061557276803913,
+      "grad_norm": 0.1835566610097885,
+      "learning_rate": 8.993069710558501e-05,
+      "loss": 3.8467,
+      "step": 11865
+    },
+    {
+      "epoch": 0.8064954477510531,
+      "grad_norm": 0.15933853387832642,
+      "learning_rate": 8.992645060470172e-05,
+      "loss": 4.035,
+      "step": 11870
+    },
+    {
+      "epoch": 0.8068351678217149,
+      "grad_norm": 0.1779545098543167,
+      "learning_rate": 8.992220410381845e-05,
+      "loss": 4.0,
+      "step": 11875
+    },
+    {
+      "epoch": 0.8071748878923767,
+      "grad_norm": 0.19771164655685425,
+      "learning_rate": 8.99179576029352e-05,
+      "loss": 3.5028,
+      "step": 11880
+    },
+    {
+      "epoch": 0.8075146079630384,
+      "grad_norm": 0.17675349116325378,
+      "learning_rate": 8.991371110205191e-05,
+      "loss": 3.9989,
+      "step": 11885
+    },
+    {
+      "epoch": 0.8078543280337003,
+      "grad_norm": 0.23120000958442688,
+      "learning_rate": 8.990946460116864e-05,
+      "loss": 3.7091,
+      "step": 11890
+    },
+    {
+      "epoch": 0.808194048104362,
+      "grad_norm": 0.18149258196353912,
+      "learning_rate": 8.990521810028538e-05,
+      "loss": 3.9628,
+      "step": 11895
+    },
+    {
+      "epoch": 0.8085337681750238,
+      "grad_norm": 0.2674315571784973,
+      "learning_rate": 8.990097159940209e-05,
+      "loss": 3.7765,
+      "step": 11900
+    },
+    {
+      "epoch": 0.8088734882456855,
+      "grad_norm": 0.21212173998355865,
+      "learning_rate": 8.989672509851883e-05,
+      "loss": 3.8405,
+      "step": 11905
+    },
+    {
+      "epoch": 0.8092132083163474,
+      "grad_norm": 0.16879509389400482,
+      "learning_rate": 8.989247859763555e-05,
+      "loss": 4.0809,
+      "step": 11910
+    },
+    {
+      "epoch": 0.8095529283870091,
+      "grad_norm": 0.14125306904315948,
+      "learning_rate": 8.988823209675228e-05,
+      "loss": 4.0725,
+      "step": 11915
+    },
+    {
+      "epoch": 0.8098926484576708,
+      "grad_norm": 0.1498613953590393,
+      "learning_rate": 8.988398559586902e-05,
+      "loss": 3.7912,
+      "step": 11920
+    },
+    {
+      "epoch": 0.8102323685283327,
+      "grad_norm": 0.16456682980060577,
+      "learning_rate": 8.987973909498573e-05,
+      "loss": 4.0187,
+      "step": 11925
+    },
+    {
+      "epoch": 0.8105720885989944,
+      "grad_norm": 0.3114604949951172,
+      "learning_rate": 8.987549259410246e-05,
+      "loss": 4.0009,
+      "step": 11930
+    },
+    {
+      "epoch": 0.8109118086696562,
+      "grad_norm": 0.5615077018737793,
+      "learning_rate": 8.98712460932192e-05,
+      "loss": 4.0609,
+      "step": 11935
+    },
+    {
+      "epoch": 0.811251528740318,
+      "grad_norm": 0.27753254771232605,
+      "learning_rate": 8.986699959233592e-05,
+      "loss": 3.8107,
+      "step": 11940
+    },
+    {
+      "epoch": 0.8115912488109798,
+      "grad_norm": 0.21950267255306244,
+      "learning_rate": 8.986275309145264e-05,
+      "loss": 3.8093,
+      "step": 11945
+    },
+    {
+      "epoch": 0.8119309688816415,
+      "grad_norm": 0.17988736927509308,
+      "learning_rate": 8.985850659056939e-05,
+      "loss": 3.9343,
+      "step": 11950
+    },
+    {
+      "epoch": 0.8122706889523033,
+      "grad_norm": 0.23350049555301666,
+      "learning_rate": 8.98542600896861e-05,
+      "loss": 4.0691,
+      "step": 11955
+    },
+    {
+      "epoch": 0.8126104090229651,
+      "grad_norm": 0.19277788698673248,
+      "learning_rate": 8.985001358880283e-05,
+      "loss": 3.9519,
+      "step": 11960
+    },
+    {
+      "epoch": 0.8129501290936268,
+      "grad_norm": 0.21622268855571747,
+      "learning_rate": 8.984576708791957e-05,
+      "loss": 3.8689,
+      "step": 11965
+    },
+    {
+      "epoch": 0.8132898491642886,
+      "grad_norm": 1.2102338075637817,
+      "learning_rate": 8.984152058703628e-05,
+      "loss": 3.9296,
+      "step": 11970
+    },
+    {
+      "epoch": 0.8136295692349504,
+      "grad_norm": 0.2097243219614029,
+      "learning_rate": 8.983727408615301e-05,
+      "loss": 4.2595,
+      "step": 11975
+    },
+    {
+      "epoch": 0.8139692893056122,
+      "grad_norm": 0.3595362603664398,
+      "learning_rate": 8.983302758526974e-05,
+      "loss": 3.9542,
+      "step": 11980
+    },
+    {
+      "epoch": 0.8143090093762739,
+      "grad_norm": 0.18622025847434998,
+      "learning_rate": 8.982878108438647e-05,
+      "loss": 3.8882,
+      "step": 11985
+    },
+    {
+      "epoch": 0.8146487294469357,
+      "grad_norm": 0.19790107011795044,
+      "learning_rate": 8.98245345835032e-05,
+      "loss": 4.1604,
+      "step": 11990
+    },
+    {
+      "epoch": 0.8149884495175975,
+      "grad_norm": 0.21050450205802917,
+      "learning_rate": 8.982028808261992e-05,
+      "loss": 3.7831,
+      "step": 11995
+    },
+    {
+      "epoch": 0.8153281695882593,
+      "grad_norm": 0.2178838849067688,
+      "learning_rate": 8.981604158173665e-05,
+      "loss": 3.8846,
+      "step": 12000
+    },
+    {
+      "epoch": 0.815667889658921,
+      "grad_norm": 0.20060613751411438,
+      "learning_rate": 8.981179508085338e-05,
+      "loss": 4.0856,
+      "step": 12005
+    },
+    {
+      "epoch": 0.8160076097295829,
+      "grad_norm": 0.19663146138191223,
+      "learning_rate": 8.980754857997011e-05,
+      "loss": 3.9376,
+      "step": 12010
+    },
+    {
+      "epoch": 0.8163473298002446,
+      "grad_norm": 0.36938565969467163,
+      "learning_rate": 8.980330207908684e-05,
+      "loss": 4.121,
+      "step": 12015
+    },
+    {
+      "epoch": 0.8166870498709063,
+      "grad_norm": 0.17913353443145752,
+      "learning_rate": 8.979905557820356e-05,
+      "loss": 3.9507,
+      "step": 12020
+    },
+    {
+      "epoch": 0.8170267699415682,
+      "grad_norm": 0.18103277683258057,
+      "learning_rate": 8.979480907732029e-05,
+      "loss": 3.651,
+      "step": 12025
+    },
+    {
+      "epoch": 0.8173664900122299,
+      "grad_norm": 0.1673816740512848,
+      "learning_rate": 8.979056257643702e-05,
+      "loss": 3.9309,
+      "step": 12030
+    },
+    {
+      "epoch": 0.8177062100828917,
+      "grad_norm": 0.4948117434978485,
+      "learning_rate": 8.978631607555375e-05,
+      "loss": 3.9102,
+      "step": 12035
+    },
+    {
+      "epoch": 0.8180459301535534,
+      "grad_norm": 0.16142868995666504,
+      "learning_rate": 8.978206957467048e-05,
+      "loss": 3.8234,
+      "step": 12040
+    },
+    {
+      "epoch": 0.8183856502242153,
+      "grad_norm": 0.2791318893432617,
+      "learning_rate": 8.97778230737872e-05,
+      "loss": 4.1417,
+      "step": 12045
+    },
+    {
+      "epoch": 0.818725370294877,
+      "grad_norm": 0.17257554829120636,
+      "learning_rate": 8.977357657290393e-05,
+      "loss": 4.043,
+      "step": 12050
+    },
+    {
+      "epoch": 0.8190650903655388,
+      "grad_norm": 0.2123369425535202,
+      "learning_rate": 8.976933007202066e-05,
+      "loss": 3.9297,
+      "step": 12055
+    },
+    {
+      "epoch": 0.8194048104362006,
+      "grad_norm": 0.19854533672332764,
+      "learning_rate": 8.976508357113739e-05,
+      "loss": 3.8441,
+      "step": 12060
+    },
+    {
+      "epoch": 0.8197445305068624,
+      "grad_norm": 0.24193492531776428,
+      "learning_rate": 8.976083707025412e-05,
+      "loss": 4.0607,
+      "step": 12065
+    },
+    {
+      "epoch": 0.8200842505775241,
+      "grad_norm": 0.1961483359336853,
+      "learning_rate": 8.975659056937084e-05,
+      "loss": 3.8928,
+      "step": 12070
+    },
+    {
+      "epoch": 0.8204239706481858,
+      "grad_norm": 0.14383824169635773,
+      "learning_rate": 8.975234406848757e-05,
+      "loss": 4.0603,
+      "step": 12075
+    },
+    {
+      "epoch": 0.8207636907188477,
+      "grad_norm": 0.2458658516407013,
+      "learning_rate": 8.97480975676043e-05,
+      "loss": 3.8334,
+      "step": 12080
+    },
+    {
+      "epoch": 0.8211034107895094,
+      "grad_norm": 0.17008869349956512,
+      "learning_rate": 8.974385106672103e-05,
+      "loss": 3.9803,
+      "step": 12085
+    },
+    {
+      "epoch": 0.8214431308601712,
+      "grad_norm": 0.20078590512275696,
+      "learning_rate": 8.973960456583776e-05,
+      "loss": 4.1641,
+      "step": 12090
+    },
+    {
+      "epoch": 0.821782850930833,
+      "grad_norm": 0.1937909722328186,
+      "learning_rate": 8.973535806495448e-05,
+      "loss": 4.0954,
+      "step": 12095
+    },
+    {
+      "epoch": 0.8221225710014948,
+      "grad_norm": 0.18328414857387543,
+      "learning_rate": 8.973111156407121e-05,
+      "loss": 3.9427,
+      "step": 12100
+    },
+    {
+      "epoch": 0.8224622910721565,
+      "grad_norm": 0.2016650289297104,
+      "learning_rate": 8.972686506318794e-05,
+      "loss": 3.9728,
+      "step": 12105
+    },
+    {
+      "epoch": 0.8228020111428184,
+      "grad_norm": 0.17548047006130219,
+      "learning_rate": 8.972261856230465e-05,
+      "loss": 4.1828,
+      "step": 12110
+    },
+    {
+      "epoch": 0.8231417312134801,
+      "grad_norm": 0.39229270815849304,
+      "learning_rate": 8.97183720614214e-05,
+      "loss": 4.0582,
+      "step": 12115
+    },
+    {
+      "epoch": 0.8234814512841419,
+      "grad_norm": 0.18692153692245483,
+      "learning_rate": 8.971412556053812e-05,
+      "loss": 3.9979,
+      "step": 12120
+    },
+    {
+      "epoch": 0.8238211713548036,
+      "grad_norm": 0.22566412389278412,
+      "learning_rate": 8.970987905965484e-05,
+      "loss": 4.0853,
+      "step": 12125
+    },
+    {
+      "epoch": 0.8241608914254654,
+      "grad_norm": 0.2699925899505615,
+      "learning_rate": 8.970563255877158e-05,
+      "loss": 4.2611,
+      "step": 12130
+    },
+    {
+      "epoch": 0.8245006114961272,
+      "grad_norm": 0.2766724228858948,
+      "learning_rate": 8.970138605788831e-05,
+      "loss": 3.8676,
+      "step": 12135
+    },
+    {
+      "epoch": 0.8248403315667889,
+      "grad_norm": 0.149053156375885,
+      "learning_rate": 8.969713955700502e-05,
+      "loss": 4.0457,
+      "step": 12140
+    },
+    {
+      "epoch": 0.8251800516374508,
+      "grad_norm": 0.29666438698768616,
+      "learning_rate": 8.969289305612176e-05,
+      "loss": 4.0984,
+      "step": 12145
+    },
+    {
+      "epoch": 0.8255197717081125,
+      "grad_norm": 0.1891719549894333,
+      "learning_rate": 8.968864655523849e-05,
+      "loss": 4.009,
+      "step": 12150
+    },
+    {
+      "epoch": 0.8258594917787743,
+      "grad_norm": 0.1801346093416214,
+      "learning_rate": 8.968440005435521e-05,
+      "loss": 4.1132,
+      "step": 12155
+    },
+    {
+      "epoch": 0.826199211849436,
+      "grad_norm": 1.4226734638214111,
+      "learning_rate": 8.968015355347195e-05,
+      "loss": 3.9887,
+      "step": 12160
+    },
+    {
+      "epoch": 0.8265389319200979,
+      "grad_norm": 0.3138851523399353,
+      "learning_rate": 8.967590705258868e-05,
+      "loss": 4.0721,
+      "step": 12165
+    },
+    {
+      "epoch": 0.8268786519907596,
+      "grad_norm": 0.19921836256980896,
+      "learning_rate": 8.967166055170539e-05,
+      "loss": 4.0086,
+      "step": 12170
+    },
+    {
+      "epoch": 0.8272183720614213,
+      "grad_norm": 0.2232120782136917,
+      "learning_rate": 8.966741405082213e-05,
+      "loss": 3.8876,
+      "step": 12175
+    },
+    {
+      "epoch": 0.8275580921320832,
+      "grad_norm": 0.21191275119781494,
+      "learning_rate": 8.966316754993885e-05,
+      "loss": 3.9509,
+      "step": 12180
+    },
+    {
+      "epoch": 0.8278978122027449,
+      "grad_norm": 1.0071362257003784,
+      "learning_rate": 8.965892104905557e-05,
+      "loss": 4.0208,
+      "step": 12185
+    },
+    {
+      "epoch": 0.8282375322734067,
+      "grad_norm": 0.30778366327285767,
+      "learning_rate": 8.965467454817232e-05,
+      "loss": 4.0989,
+      "step": 12190
+    },
+    {
+      "epoch": 0.8285772523440685,
+      "grad_norm": 0.15304256975650787,
+      "learning_rate": 8.965042804728903e-05,
+      "loss": 4.1485,
+      "step": 12195
+    },
+    {
+      "epoch": 0.8289169724147303,
+      "grad_norm": 0.1691897213459015,
+      "learning_rate": 8.964618154640576e-05,
+      "loss": 4.1178,
+      "step": 12200
+    },
+    {
+      "epoch": 0.829256692485392,
+      "grad_norm": 0.2017151266336441,
+      "learning_rate": 8.96419350455225e-05,
+      "loss": 4.1713,
+      "step": 12205
+    },
+    {
+      "epoch": 0.8295964125560538,
+      "grad_norm": 0.45046570897102356,
+      "learning_rate": 8.963768854463921e-05,
+      "loss": 3.8627,
+      "step": 12210
+    },
+    {
+      "epoch": 0.8299361326267156,
+      "grad_norm": 0.21693216264247894,
+      "learning_rate": 8.963344204375594e-05,
+      "loss": 3.969,
+      "step": 12215
+    },
+    {
+      "epoch": 0.8302758526973774,
+      "grad_norm": 0.22828508913516998,
+      "learning_rate": 8.962919554287268e-05,
+      "loss": 3.9584,
+      "step": 12220
+    },
+    {
+      "epoch": 0.8306155727680391,
+      "grad_norm": 0.2518628239631653,
+      "learning_rate": 8.96249490419894e-05,
+      "loss": 4.0381,
+      "step": 12225
+    },
+    {
+      "epoch": 0.830955292838701,
+      "grad_norm": 0.1994330883026123,
+      "learning_rate": 8.962070254110613e-05,
+      "loss": 3.8909,
+      "step": 12230
+    },
+    {
+      "epoch": 0.8312950129093627,
+      "grad_norm": 0.1634039431810379,
+      "learning_rate": 8.961645604022287e-05,
+      "loss": 3.5102,
+      "step": 12235
+    },
+    {
+      "epoch": 0.8316347329800244,
+      "grad_norm": 0.17670801281929016,
+      "learning_rate": 8.961220953933958e-05,
+      "loss": 3.8131,
+      "step": 12240
+    },
+    {
+      "epoch": 0.8319744530506862,
+      "grad_norm": 0.19512879848480225,
+      "learning_rate": 8.960796303845632e-05,
+      "loss": 3.5444,
+      "step": 12245
+    },
+    {
+      "epoch": 0.832314173121348,
+      "grad_norm": 0.5432287454605103,
+      "learning_rate": 8.960371653757305e-05,
+      "loss": 3.8723,
+      "step": 12250
+    },
+    {
+      "epoch": 0.8326538931920098,
+      "grad_norm": 0.21648722887039185,
+      "learning_rate": 8.959947003668977e-05,
+      "loss": 3.7991,
+      "step": 12255
+    },
+    {
+      "epoch": 0.8329936132626715,
+      "grad_norm": 0.3105649948120117,
+      "learning_rate": 8.959522353580651e-05,
+      "loss": 3.8407,
+      "step": 12260
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.32569703459739685,
+      "learning_rate": 8.959097703492322e-05,
+      "loss": 4.2322,
+      "step": 12265
+    },
+    {
+      "epoch": 0.8336730534039951,
+      "grad_norm": 0.2628330588340759,
+      "learning_rate": 8.958673053403995e-05,
+      "loss": 3.599,
+      "step": 12270
+    },
+    {
+      "epoch": 0.8340127734746569,
+      "grad_norm": 0.15857040882110596,
+      "learning_rate": 8.958248403315669e-05,
+      "loss": 4.0897,
+      "step": 12275
+    },
+    {
+      "epoch": 0.8343524935453187,
+      "grad_norm": 0.20659282803535461,
+      "learning_rate": 8.957823753227341e-05,
+      "loss": 3.8562,
+      "step": 12280
+    },
+    {
+      "epoch": 0.8346922136159804,
+      "grad_norm": 0.21449913084506989,
+      "learning_rate": 8.957399103139014e-05,
+      "loss": 3.9756,
+      "step": 12285
+    },
+    {
+      "epoch": 0.8350319336866422,
+      "grad_norm": 0.1901203691959381,
+      "learning_rate": 8.956974453050688e-05,
+      "loss": 4.025,
+      "step": 12290
+    },
+    {
+      "epoch": 0.8353716537573039,
+      "grad_norm": 0.20290473103523254,
+      "learning_rate": 8.956549802962359e-05,
+      "loss": 3.8794,
+      "step": 12295
+    },
+    {
+      "epoch": 0.8357113738279658,
+      "grad_norm": 0.1798480749130249,
+      "learning_rate": 8.956125152874032e-05,
+      "loss": 3.9994,
+      "step": 12300
+    },
+    {
+      "epoch": 0.8360510938986275,
+      "grad_norm": 0.25437653064727783,
+      "learning_rate": 8.955700502785706e-05,
+      "loss": 3.6909,
+      "step": 12305
+    },
+    {
+      "epoch": 0.8363908139692893,
+      "grad_norm": 0.3736377954483032,
+      "learning_rate": 8.955275852697378e-05,
+      "loss": 4.0622,
+      "step": 12310
+    },
+    {
+      "epoch": 0.8367305340399511,
+      "grad_norm": 0.18843521177768707,
+      "learning_rate": 8.95485120260905e-05,
+      "loss": 3.7766,
+      "step": 12315
+    },
+    {
+      "epoch": 0.8370702541106129,
+      "grad_norm": 0.18586871027946472,
+      "learning_rate": 8.954426552520724e-05,
+      "loss": 4.0302,
+      "step": 12320
+    },
+    {
+      "epoch": 0.8374099741812746,
+      "grad_norm": 0.17058813571929932,
+      "learning_rate": 8.954001902432396e-05,
+      "loss": 4.0756,
+      "step": 12325
+    },
+    {
+      "epoch": 0.8377496942519363,
+      "grad_norm": 0.21250483393669128,
+      "learning_rate": 8.953577252344069e-05,
+      "loss": 3.6402,
+      "step": 12330
+    },
+    {
+      "epoch": 0.8380894143225982,
+      "grad_norm": 0.18693235516548157,
+      "learning_rate": 8.953152602255742e-05,
+      "loss": 4.2171,
+      "step": 12335
+    },
+    {
+      "epoch": 0.83842913439326,
+      "grad_norm": 0.20142552256584167,
+      "learning_rate": 8.952727952167414e-05,
+      "loss": 4.0116,
+      "step": 12340
+    },
+    {
+      "epoch": 0.8387688544639217,
+      "grad_norm": 0.5936012864112854,
+      "learning_rate": 8.952303302079087e-05,
+      "loss": 4.0355,
+      "step": 12345
+    },
+    {
+      "epoch": 0.8391085745345835,
+      "grad_norm": 0.3252449631690979,
+      "learning_rate": 8.95187865199076e-05,
+      "loss": 4.0143,
+      "step": 12350
+    },
+    {
+      "epoch": 0.8394482946052453,
+      "grad_norm": 0.18693962693214417,
+      "learning_rate": 8.951454001902433e-05,
+      "loss": 3.938,
+      "step": 12355
+    },
+    {
+      "epoch": 0.839788014675907,
+      "grad_norm": 0.36720210313796997,
+      "learning_rate": 8.951029351814106e-05,
+      "loss": 3.9352,
+      "step": 12360
+    },
+    {
+      "epoch": 0.8401277347465689,
+      "grad_norm": 0.14825104176998138,
+      "learning_rate": 8.950604701725778e-05,
+      "loss": 4.2168,
+      "step": 12365
+    },
+    {
+      "epoch": 0.8404674548172306,
+      "grad_norm": 0.23677104711532593,
+      "learning_rate": 8.950180051637451e-05,
+      "loss": 3.967,
+      "step": 12370
+    },
+    {
+      "epoch": 0.8408071748878924,
+      "grad_norm": 0.5124611258506775,
+      "learning_rate": 8.949755401549124e-05,
+      "loss": 3.9194,
+      "step": 12375
+    },
+    {
+      "epoch": 0.8411468949585541,
+      "grad_norm": 0.3029448688030243,
+      "learning_rate": 8.949330751460797e-05,
+      "loss": 3.9354,
+      "step": 12380
+    },
+    {
+      "epoch": 0.841486615029216,
+      "grad_norm": 0.20730510354042053,
+      "learning_rate": 8.94890610137247e-05,
+      "loss": 4.1058,
+      "step": 12385
+    },
+    {
+      "epoch": 0.8418263350998777,
+      "grad_norm": 0.22315102815628052,
+      "learning_rate": 8.948481451284142e-05,
+      "loss": 3.893,
+      "step": 12390
+    },
+    {
+      "epoch": 0.8421660551705394,
+      "grad_norm": 0.47029411792755127,
+      "learning_rate": 8.948056801195815e-05,
+      "loss": 3.8572,
+      "step": 12395
+    },
+    {
+      "epoch": 0.8425057752412013,
+      "grad_norm": 0.19684122502803802,
+      "learning_rate": 8.947632151107488e-05,
+      "loss": 4.0835,
+      "step": 12400
+    },
+    {
+      "epoch": 0.842845495311863,
+      "grad_norm": 0.18742690980434418,
+      "learning_rate": 8.947207501019161e-05,
+      "loss": 3.9798,
+      "step": 12405
+    },
+    {
+      "epoch": 0.8431852153825248,
+      "grad_norm": 0.177710622549057,
+      "learning_rate": 8.946782850930834e-05,
+      "loss": 4.1085,
+      "step": 12410
+    },
+    {
+      "epoch": 0.8435249354531865,
+      "grad_norm": 0.18476006388664246,
+      "learning_rate": 8.946358200842506e-05,
+      "loss": 4.0936,
+      "step": 12415
+    },
+    {
+      "epoch": 0.8438646555238484,
+      "grad_norm": 0.16293834149837494,
+      "learning_rate": 8.946018480771845e-05,
+      "loss": 3.9854,
+      "step": 12420
+    },
+    {
+      "epoch": 0.8442043755945101,
+      "grad_norm": 0.22308799624443054,
+      "learning_rate": 8.945593830683517e-05,
+      "loss": 4.1042,
+      "step": 12425
+    },
+    {
+      "epoch": 0.8445440956651719,
+      "grad_norm": 0.16046442091464996,
+      "learning_rate": 8.94516918059519e-05,
+      "loss": 4.0012,
+      "step": 12430
+    },
+    {
+      "epoch": 0.8448838157358337,
+      "grad_norm": 0.26154249906539917,
+      "learning_rate": 8.944744530506862e-05,
+      "loss": 4.0459,
+      "step": 12435
+    },
+    {
+      "epoch": 0.8452235358064955,
+      "grad_norm": 0.23383556306362152,
+      "learning_rate": 8.944319880418536e-05,
+      "loss": 4.0728,
+      "step": 12440
+    },
+    {
+      "epoch": 0.8455632558771572,
+      "grad_norm": 0.22498470544815063,
+      "learning_rate": 8.943895230330209e-05,
+      "loss": 3.979,
+      "step": 12445
+    },
+    {
+      "epoch": 0.845902975947819,
+      "grad_norm": 0.18164518475532532,
+      "learning_rate": 8.943470580241881e-05,
+      "loss": 3.8648,
+      "step": 12450
+    },
+    {
+      "epoch": 0.8462426960184808,
+      "grad_norm": 0.21877458691596985,
+      "learning_rate": 8.943045930153554e-05,
+      "loss": 4.2495,
+      "step": 12455
+    },
+    {
+      "epoch": 0.8465824160891425,
+      "grad_norm": 0.1924646645784378,
+      "learning_rate": 8.942621280065227e-05,
+      "loss": 3.8453,
+      "step": 12460
+    },
+    {
+      "epoch": 0.8469221361598043,
+      "grad_norm": 0.20620742440223694,
+      "learning_rate": 8.9421966299769e-05,
+      "loss": 3.9962,
+      "step": 12465
+    },
+    {
+      "epoch": 0.8472618562304661,
+      "grad_norm": 0.290763795375824,
+      "learning_rate": 8.941771979888573e-05,
+      "loss": 3.8139,
+      "step": 12470
+    },
+    {
+      "epoch": 0.8476015763011279,
+      "grad_norm": 0.16620713472366333,
+      "learning_rate": 8.941347329800245e-05,
+      "loss": 3.7297,
+      "step": 12475
+    },
+    {
+      "epoch": 0.8479412963717896,
+      "grad_norm": 2.936108112335205,
+      "learning_rate": 8.940922679711918e-05,
+      "loss": 3.8085,
+      "step": 12480
+    },
+    {
+      "epoch": 0.8482810164424515,
+      "grad_norm": 0.1852518916130066,
+      "learning_rate": 8.940498029623591e-05,
+      "loss": 3.9015,
+      "step": 12485
+    },
+    {
+      "epoch": 0.8486207365131132,
+      "grad_norm": 0.159807950258255,
+      "learning_rate": 8.940073379535264e-05,
+      "loss": 3.9683,
+      "step": 12490
+    },
+    {
+      "epoch": 0.848960456583775,
+      "grad_norm": 0.15564261376857758,
+      "learning_rate": 8.939648729446937e-05,
+      "loss": 4.0309,
+      "step": 12495
+    },
+    {
+      "epoch": 0.8493001766544367,
+      "grad_norm": 0.18451645970344543,
+      "learning_rate": 8.93922407935861e-05,
+      "loss": 4.002,
+      "step": 12500
+    },
+    {
+      "epoch": 0.8496398967250985,
+      "grad_norm": 0.4409978687763214,
+      "learning_rate": 8.938799429270281e-05,
+      "loss": 3.9582,
+      "step": 12505
+    },
+    {
+      "epoch": 0.8499796167957603,
+      "grad_norm": 0.21767017245292664,
+      "learning_rate": 8.938374779181955e-05,
+      "loss": 3.7942,
+      "step": 12510
+    },
+    {
+      "epoch": 0.850319336866422,
+      "grad_norm": 0.29611897468566895,
+      "learning_rate": 8.937950129093628e-05,
+      "loss": 3.8641,
+      "step": 12515
+    },
+    {
+      "epoch": 0.8506590569370839,
+      "grad_norm": 0.15617810189723969,
+      "learning_rate": 8.937525479005299e-05,
+      "loss": 4.0887,
+      "step": 12520
+    },
+    {
+      "epoch": 0.8509987770077456,
+      "grad_norm": 0.19923017919063568,
+      "learning_rate": 8.937100828916973e-05,
+      "loss": 3.9796,
+      "step": 12525
+    },
+    {
+      "epoch": 0.8513384970784074,
+      "grad_norm": 0.178538978099823,
+      "learning_rate": 8.936676178828646e-05,
+      "loss": 4.1373,
+      "step": 12530
+    },
+    {
+      "epoch": 0.8516782171490692,
+      "grad_norm": 0.20157839357852936,
+      "learning_rate": 8.936251528740318e-05,
+      "loss": 4.1656,
+      "step": 12535
+    },
+    {
+      "epoch": 0.852017937219731,
+      "grad_norm": 0.21917836368083954,
+      "learning_rate": 8.935826878651992e-05,
+      "loss": 3.9735,
+      "step": 12540
+    },
+    {
+      "epoch": 0.8523576572903927,
+      "grad_norm": 0.1743486225605011,
+      "learning_rate": 8.935402228563665e-05,
+      "loss": 4.0561,
+      "step": 12545
+    },
+    {
+      "epoch": 0.8526973773610544,
+      "grad_norm": 0.3735666573047638,
+      "learning_rate": 8.934977578475336e-05,
+      "loss": 3.9863,
+      "step": 12550
+    },
+    {
+      "epoch": 0.8530370974317163,
+      "grad_norm": 0.21841804683208466,
+      "learning_rate": 8.93455292838701e-05,
+      "loss": 4.0283,
+      "step": 12555
+    },
+    {
+      "epoch": 0.853376817502378,
+      "grad_norm": 0.4104025065898895,
+      "learning_rate": 8.934128278298683e-05,
+      "loss": 3.7465,
+      "step": 12560
+    },
+    {
+      "epoch": 0.8537165375730398,
+      "grad_norm": 0.15292176604270935,
+      "learning_rate": 8.933703628210354e-05,
+      "loss": 4.0306,
+      "step": 12565
+    },
+    {
+      "epoch": 0.8540562576437016,
+      "grad_norm": 0.30871346592903137,
+      "learning_rate": 8.933278978122029e-05,
+      "loss": 3.9504,
+      "step": 12570
+    },
+    {
+      "epoch": 0.8543959777143634,
+      "grad_norm": 0.18270432949066162,
+      "learning_rate": 8.9328543280337e-05,
+      "loss": 4.1066,
+      "step": 12575
+    },
+    {
+      "epoch": 0.8547356977850251,
+      "grad_norm": 0.5479381680488586,
+      "learning_rate": 8.932429677945373e-05,
+      "loss": 3.9433,
+      "step": 12580
+    },
+    {
+      "epoch": 0.8550754178556869,
+      "grad_norm": 0.19208583235740662,
+      "learning_rate": 8.932005027857047e-05,
+      "loss": 3.7949,
+      "step": 12585
+    },
+    {
+      "epoch": 0.8554151379263487,
+      "grad_norm": 0.19413863122463226,
+      "learning_rate": 8.931580377768718e-05,
+      "loss": 4.07,
+      "step": 12590
+    },
+    {
+      "epoch": 0.8557548579970105,
+      "grad_norm": 0.19963692128658295,
+      "learning_rate": 8.931155727680391e-05,
+      "loss": 4.038,
+      "step": 12595
+    },
+    {
+      "epoch": 0.8560945780676722,
+      "grad_norm": 0.18308939039707184,
+      "learning_rate": 8.930731077592065e-05,
+      "loss": 4.0086,
+      "step": 12600
+    },
+    {
+      "epoch": 0.856434298138334,
+      "grad_norm": 0.23906344175338745,
+      "learning_rate": 8.930306427503737e-05,
+      "loss": 3.835,
+      "step": 12605
+    },
+    {
+      "epoch": 0.8567740182089958,
+      "grad_norm": 0.15061314404010773,
+      "learning_rate": 8.92988177741541e-05,
+      "loss": 3.9051,
+      "step": 12610
+    },
+    {
+      "epoch": 0.8571137382796575,
+      "grad_norm": 0.4929114282131195,
+      "learning_rate": 8.929457127327084e-05,
+      "loss": 3.8619,
+      "step": 12615
+    },
+    {
+      "epoch": 0.8574534583503194,
+      "grad_norm": 0.23578637838363647,
+      "learning_rate": 8.929032477238755e-05,
+      "loss": 4.2782,
+      "step": 12620
+    },
+    {
+      "epoch": 0.8577931784209811,
+      "grad_norm": 0.2066326141357422,
+      "learning_rate": 8.928607827150428e-05,
+      "loss": 4.2725,
+      "step": 12625
+    },
+    {
+      "epoch": 0.8581328984916429,
+      "grad_norm": 0.22157415747642517,
+      "learning_rate": 8.928183177062102e-05,
+      "loss": 4.0701,
+      "step": 12630
+    },
+    {
+      "epoch": 0.8584726185623046,
+      "grad_norm": 0.9750187397003174,
+      "learning_rate": 8.927758526973774e-05,
+      "loss": 3.8094,
+      "step": 12635
+    },
+    {
+      "epoch": 0.8588123386329665,
+      "grad_norm": 0.17541570961475372,
+      "learning_rate": 8.927333876885446e-05,
+      "loss": 4.1766,
+      "step": 12640
+    },
+    {
+      "epoch": 0.8591520587036282,
+      "grad_norm": 0.1566866785287857,
+      "learning_rate": 8.92690922679712e-05,
+      "loss": 4.0965,
+      "step": 12645
+    },
+    {
+      "epoch": 0.85949177877429,
+      "grad_norm": 0.18229223787784576,
+      "learning_rate": 8.926484576708792e-05,
+      "loss": 3.9081,
+      "step": 12650
+    },
+    {
+      "epoch": 0.8598314988449518,
+      "grad_norm": 0.19233596324920654,
+      "learning_rate": 8.926059926620465e-05,
+      "loss": 3.9146,
+      "step": 12655
+    },
+    {
+      "epoch": 0.8601712189156135,
+      "grad_norm": 0.26399555802345276,
+      "learning_rate": 8.925635276532138e-05,
+      "loss": 3.8794,
+      "step": 12660
+    },
+    {
+      "epoch": 0.8605109389862753,
+      "grad_norm": 0.18803419172763824,
+      "learning_rate": 8.92521062644381e-05,
+      "loss": 4.2288,
+      "step": 12665
+    },
+    {
+      "epoch": 0.860850659056937,
+      "grad_norm": 0.2003091722726822,
+      "learning_rate": 8.924785976355483e-05,
+      "loss": 3.9376,
+      "step": 12670
+    },
+    {
+      "epoch": 0.8611903791275989,
+      "grad_norm": 2.6071043014526367,
+      "learning_rate": 8.924361326267156e-05,
+      "loss": 3.9786,
+      "step": 12675
+    },
+    {
+      "epoch": 0.8615300991982606,
+      "grad_norm": 0.21313896775245667,
+      "learning_rate": 8.923936676178829e-05,
+      "loss": 4.0317,
+      "step": 12680
+    },
+    {
+      "epoch": 0.8618698192689224,
+      "grad_norm": 0.17100752890110016,
+      "learning_rate": 8.923512026090502e-05,
+      "loss": 3.9988,
+      "step": 12685
+    },
+    {
+      "epoch": 0.8622095393395842,
+      "grad_norm": 0.23430535197257996,
+      "learning_rate": 8.923087376002174e-05,
+      "loss": 3.7328,
+      "step": 12690
+    },
+    {
+      "epoch": 0.862549259410246,
+      "grad_norm": 0.1643848717212677,
+      "learning_rate": 8.922662725913847e-05,
+      "loss": 4.109,
+      "step": 12695
+    },
+    {
+      "epoch": 0.8628889794809077,
+      "grad_norm": 0.2526448369026184,
+      "learning_rate": 8.92223807582552e-05,
+      "loss": 4.0189,
+      "step": 12700
+    },
+    {
+      "epoch": 0.8632286995515696,
+      "grad_norm": 0.24552328884601593,
+      "learning_rate": 8.921813425737193e-05,
+      "loss": 3.9226,
+      "step": 12705
+    },
+    {
+      "epoch": 0.8635684196222313,
+      "grad_norm": 0.17898543179035187,
+      "learning_rate": 8.921388775648866e-05,
+      "loss": 3.7268,
+      "step": 12710
+    },
+    {
+      "epoch": 0.863908139692893,
+      "grad_norm": 1.15958833694458,
+      "learning_rate": 8.920964125560538e-05,
+      "loss": 3.9694,
+      "step": 12715
+    },
+    {
+      "epoch": 0.8642478597635548,
+      "grad_norm": 0.15424911677837372,
+      "learning_rate": 8.920539475472211e-05,
+      "loss": 4.0695,
+      "step": 12720
+    },
+    {
+      "epoch": 0.8645875798342166,
+      "grad_norm": 0.1732960045337677,
+      "learning_rate": 8.920114825383884e-05,
+      "loss": 4.0694,
+      "step": 12725
+    },
+    {
+      "epoch": 0.8649272999048784,
+      "grad_norm": 0.25387898087501526,
+      "learning_rate": 8.919690175295557e-05,
+      "loss": 3.9112,
+      "step": 12730
+    },
+    {
+      "epoch": 0.8652670199755401,
+      "grad_norm": 0.18854272365570068,
+      "learning_rate": 8.91926552520723e-05,
+      "loss": 4.1747,
+      "step": 12735
+    },
+    {
+      "epoch": 0.865606740046202,
+      "grad_norm": 0.18318019807338715,
+      "learning_rate": 8.918840875118902e-05,
+      "loss": 4.0418,
+      "step": 12740
+    },
+    {
+      "epoch": 0.8659464601168637,
+      "grad_norm": 0.1997986137866974,
+      "learning_rate": 8.918416225030575e-05,
+      "loss": 3.9741,
+      "step": 12745
+    },
+    {
+      "epoch": 0.8662861801875255,
+      "grad_norm": 0.1834951937198639,
+      "learning_rate": 8.917991574942248e-05,
+      "loss": 4.0931,
+      "step": 12750
+    },
+    {
+      "epoch": 0.8666259002581872,
+      "grad_norm": 0.1741950660943985,
+      "learning_rate": 8.917566924853921e-05,
+      "loss": 3.8176,
+      "step": 12755
+    },
+    {
+      "epoch": 0.866965620328849,
+      "grad_norm": 0.18136630952358246,
+      "learning_rate": 8.917142274765594e-05,
+      "loss": 4.0065,
+      "step": 12760
+    },
+    {
+      "epoch": 0.8673053403995108,
+      "grad_norm": 0.22858762741088867,
+      "learning_rate": 8.916717624677266e-05,
+      "loss": 3.8951,
+      "step": 12765
+    },
+    {
+      "epoch": 0.8676450604701725,
+      "grad_norm": 0.15833310782909393,
+      "learning_rate": 8.916292974588939e-05,
+      "loss": 3.9966,
+      "step": 12770
+    },
+    {
+      "epoch": 0.8679847805408344,
+      "grad_norm": 0.16821962594985962,
+      "learning_rate": 8.915868324500611e-05,
+      "loss": 4.1885,
+      "step": 12775
+    },
+    {
+      "epoch": 0.8683245006114961,
+      "grad_norm": 0.16304543614387512,
+      "learning_rate": 8.915443674412285e-05,
+      "loss": 4.054,
+      "step": 12780
+    },
+    {
+      "epoch": 0.8686642206821579,
+      "grad_norm": 0.19145479798316956,
+      "learning_rate": 8.915019024323958e-05,
+      "loss": 4.0134,
+      "step": 12785
+    },
+    {
+      "epoch": 0.8690039407528197,
+      "grad_norm": 0.20590396225452423,
+      "learning_rate": 8.91459437423563e-05,
+      "loss": 3.9,
+      "step": 12790
+    },
+    {
+      "epoch": 0.8693436608234815,
+      "grad_norm": 0.1521267145872116,
+      "learning_rate": 8.914169724147303e-05,
+      "loss": 3.9639,
+      "step": 12795
+    },
+    {
+      "epoch": 0.8696833808941432,
+      "grad_norm": 1.8004759550094604,
+      "learning_rate": 8.913745074058976e-05,
+      "loss": 4.1324,
+      "step": 12800
+    },
+    {
+      "epoch": 0.870023100964805,
+      "grad_norm": 0.1751425862312317,
+      "learning_rate": 8.913320423970649e-05,
+      "loss": 4.0504,
+      "step": 12805
+    },
+    {
+      "epoch": 0.8703628210354668,
+      "grad_norm": 0.21332374215126038,
+      "learning_rate": 8.912895773882322e-05,
+      "loss": 4.095,
+      "step": 12810
+    },
+    {
+      "epoch": 0.8707025411061285,
+      "grad_norm": 0.2132454216480255,
+      "learning_rate": 8.912471123793994e-05,
+      "loss": 3.9654,
+      "step": 12815
+    },
+    {
+      "epoch": 0.8710422611767903,
+      "grad_norm": 0.1583162248134613,
+      "learning_rate": 8.912046473705667e-05,
+      "loss": 3.9617,
+      "step": 12820
+    },
+    {
+      "epoch": 0.8713819812474521,
+      "grad_norm": 0.19870373606681824,
+      "learning_rate": 8.91162182361734e-05,
+      "loss": 4.2326,
+      "step": 12825
+    },
+    {
+      "epoch": 0.8717217013181139,
+      "grad_norm": 0.18757444620132446,
+      "learning_rate": 8.911197173529013e-05,
+      "loss": 3.9562,
+      "step": 12830
+    },
+    {
+      "epoch": 0.8720614213887756,
+      "grad_norm": 0.15151342749595642,
+      "learning_rate": 8.910772523440686e-05,
+      "loss": 3.8253,
+      "step": 12835
+    },
+    {
+      "epoch": 0.8724011414594374,
+      "grad_norm": 0.23465317487716675,
+      "learning_rate": 8.910347873352358e-05,
+      "loss": 3.8039,
+      "step": 12840
+    },
+    {
+      "epoch": 0.8727408615300992,
+      "grad_norm": 0.3286290168762207,
+      "learning_rate": 8.909923223264031e-05,
+      "loss": 3.9344,
+      "step": 12845
+    },
+    {
+      "epoch": 0.873080581600761,
+      "grad_norm": 0.3796158730983734,
+      "learning_rate": 8.909498573175704e-05,
+      "loss": 4.2968,
+      "step": 12850
+    },
+    {
+      "epoch": 0.8734203016714227,
+      "grad_norm": 0.17161943018436432,
+      "learning_rate": 8.909073923087377e-05,
+      "loss": 4.0475,
+      "step": 12855
+    },
+    {
+      "epoch": 0.8737600217420846,
+      "grad_norm": 0.18927636742591858,
+      "learning_rate": 8.908649272999048e-05,
+      "loss": 3.9,
+      "step": 12860
+    },
+    {
+      "epoch": 0.8740997418127463,
+      "grad_norm": 0.17746715247631073,
+      "learning_rate": 8.908224622910722e-05,
+      "loss": 4.0236,
+      "step": 12865
+    },
+    {
+      "epoch": 0.874439461883408,
+      "grad_norm": 0.2432946115732193,
+      "learning_rate": 8.907799972822395e-05,
+      "loss": 3.595,
+      "step": 12870
+    },
+    {
+      "epoch": 0.8747791819540699,
+      "grad_norm": 0.18916042149066925,
+      "learning_rate": 8.907375322734067e-05,
+      "loss": 3.9668,
+      "step": 12875
+    },
+    {
+      "epoch": 0.8751189020247316,
+      "grad_norm": 0.18429867923259735,
+      "learning_rate": 8.906950672645741e-05,
+      "loss": 3.8215,
+      "step": 12880
+    },
+    {
+      "epoch": 0.8754586220953934,
+      "grad_norm": 0.30447253584861755,
+      "learning_rate": 8.906526022557414e-05,
+      "loss": 3.94,
+      "step": 12885
+    },
+    {
+      "epoch": 0.8757983421660551,
+      "grad_norm": 0.1514458954334259,
+      "learning_rate": 8.906101372469085e-05,
+      "loss": 3.8559,
+      "step": 12890
+    },
+    {
+      "epoch": 0.876138062236717,
+      "grad_norm": 0.1644824743270874,
+      "learning_rate": 8.905676722380759e-05,
+      "loss": 4.136,
+      "step": 12895
+    },
+    {
+      "epoch": 0.8764777823073787,
+      "grad_norm": 0.1758906990289688,
+      "learning_rate": 8.905252072292432e-05,
+      "loss": 4.2008,
+      "step": 12900
+    },
+    {
+      "epoch": 0.8768175023780405,
+      "grad_norm": 0.19972476363182068,
+      "learning_rate": 8.904827422204104e-05,
+      "loss": 3.9077,
+      "step": 12905
+    },
+    {
+      "epoch": 0.8771572224487023,
+      "grad_norm": 0.21778126060962677,
+      "learning_rate": 8.904402772115778e-05,
+      "loss": 3.9808,
+      "step": 12910
+    },
+    {
+      "epoch": 0.877496942519364,
+      "grad_norm": 0.19557218253612518,
+      "learning_rate": 8.90397812202745e-05,
+      "loss": 3.7706,
+      "step": 12915
+    },
+    {
+      "epoch": 0.8778366625900258,
+      "grad_norm": 0.1903916746377945,
+      "learning_rate": 8.903553471939122e-05,
+      "loss": 3.8328,
+      "step": 12920
+    },
+    {
+      "epoch": 0.8781763826606876,
+      "grad_norm": 0.21190743148326874,
+      "learning_rate": 8.903128821850796e-05,
+      "loss": 3.8016,
+      "step": 12925
+    },
+    {
+      "epoch": 0.8785161027313494,
+      "grad_norm": 0.22646445035934448,
+      "learning_rate": 8.902704171762468e-05,
+      "loss": 3.9702,
+      "step": 12930
+    },
+    {
+      "epoch": 0.8788558228020111,
+      "grad_norm": 0.2211994081735611,
+      "learning_rate": 8.90227952167414e-05,
+      "loss": 4.0118,
+      "step": 12935
+    },
+    {
+      "epoch": 0.8791955428726729,
+      "grad_norm": 0.23225241899490356,
+      "learning_rate": 8.901854871585814e-05,
+      "loss": 4.1305,
+      "step": 12940
+    },
+    {
+      "epoch": 0.8795352629433347,
+      "grad_norm": 0.20378831028938293,
+      "learning_rate": 8.901430221497486e-05,
+      "loss": 3.9627,
+      "step": 12945
+    },
+    {
+      "epoch": 0.8798749830139965,
+      "grad_norm": 0.15453274548053741,
+      "learning_rate": 8.901005571409159e-05,
+      "loss": 3.8527,
+      "step": 12950
+    },
+    {
+      "epoch": 0.8802147030846582,
+      "grad_norm": 0.19792801141738892,
+      "learning_rate": 8.900580921320833e-05,
+      "loss": 4.0273,
+      "step": 12955
+    },
+    {
+      "epoch": 0.8805544231553201,
+      "grad_norm": 0.17211174964904785,
+      "learning_rate": 8.900156271232504e-05,
+      "loss": 3.9116,
+      "step": 12960
+    },
+    {
+      "epoch": 0.8808941432259818,
+      "grad_norm": 0.5672011375427246,
+      "learning_rate": 8.899731621144177e-05,
+      "loss": 3.9642,
+      "step": 12965
+    },
+    {
+      "epoch": 0.8812338632966435,
+      "grad_norm": 0.18356992304325104,
+      "learning_rate": 8.899306971055851e-05,
+      "loss": 3.9994,
+      "step": 12970
+    },
+    {
+      "epoch": 0.8815735833673053,
+      "grad_norm": 0.18002454936504364,
+      "learning_rate": 8.898882320967523e-05,
+      "loss": 3.9536,
+      "step": 12975
+    },
+    {
+      "epoch": 0.8819133034379671,
+      "grad_norm": 0.18387572467327118,
+      "learning_rate": 8.898457670879196e-05,
+      "loss": 4.0552,
+      "step": 12980
+    },
+    {
+      "epoch": 0.8822530235086289,
+      "grad_norm": 0.27319806814193726,
+      "learning_rate": 8.89803302079087e-05,
+      "loss": 3.9619,
+      "step": 12985
+    },
+    {
+      "epoch": 0.8825927435792906,
+      "grad_norm": 0.2080732136964798,
+      "learning_rate": 8.897608370702541e-05,
+      "loss": 4.1625,
+      "step": 12990
+    },
+    {
+      "epoch": 0.8829324636499525,
+      "grad_norm": 0.17268094420433044,
+      "learning_rate": 8.897183720614214e-05,
+      "loss": 3.9614,
+      "step": 12995
+    },
+    {
+      "epoch": 0.8832721837206142,
+      "grad_norm": 0.15917377173900604,
+      "learning_rate": 8.896759070525888e-05,
+      "loss": 3.9529,
+      "step": 13000
+    },
+    {
+      "epoch": 0.883611903791276,
+      "grad_norm": 0.22826112806797028,
+      "learning_rate": 8.89633442043756e-05,
+      "loss": 4.0541,
+      "step": 13005
+    },
+    {
+      "epoch": 0.8839516238619378,
+      "grad_norm": 0.2643020749092102,
+      "learning_rate": 8.895909770349232e-05,
+      "loss": 3.8896,
+      "step": 13010
+    },
+    {
+      "epoch": 0.8842913439325996,
+      "grad_norm": 0.19719457626342773,
+      "learning_rate": 8.895485120260905e-05,
+      "loss": 3.8283,
+      "step": 13015
+    },
+    {
+      "epoch": 0.8846310640032613,
+      "grad_norm": 0.18295039236545563,
+      "learning_rate": 8.895060470172578e-05,
+      "loss": 3.8955,
+      "step": 13020
+    },
+    {
+      "epoch": 0.884970784073923,
+      "grad_norm": 0.23282389342784882,
+      "learning_rate": 8.894635820084251e-05,
+      "loss": 3.9094,
+      "step": 13025
+    },
+    {
+      "epoch": 0.8853105041445849,
+      "grad_norm": 0.1835237592458725,
+      "learning_rate": 8.894211169995924e-05,
+      "loss": 3.9402,
+      "step": 13030
+    },
+    {
+      "epoch": 0.8856502242152466,
+      "grad_norm": 0.2141278088092804,
+      "learning_rate": 8.893786519907596e-05,
+      "loss": 4.0951,
+      "step": 13035
+    },
+    {
+      "epoch": 0.8859899442859084,
+      "grad_norm": 0.16725867986679077,
+      "learning_rate": 8.893361869819269e-05,
+      "loss": 3.8523,
+      "step": 13040
+    },
+    {
+      "epoch": 0.8863296643565702,
+      "grad_norm": 0.15616454184055328,
+      "learning_rate": 8.892937219730942e-05,
+      "loss": 3.9671,
+      "step": 13045
+    },
+    {
+      "epoch": 0.886669384427232,
+      "grad_norm": 0.17522963881492615,
+      "learning_rate": 8.892512569642615e-05,
+      "loss": 4.0624,
+      "step": 13050
+    },
+    {
+      "epoch": 0.8870091044978937,
+      "grad_norm": 0.19032767415046692,
+      "learning_rate": 8.892087919554288e-05,
+      "loss": 3.8977,
+      "step": 13055
+    },
+    {
+      "epoch": 0.8873488245685555,
+      "grad_norm": 0.4839637279510498,
+      "learning_rate": 8.89166326946596e-05,
+      "loss": 3.9498,
+      "step": 13060
+    },
+    {
+      "epoch": 0.8876885446392173,
+      "grad_norm": 0.15772879123687744,
+      "learning_rate": 8.891238619377633e-05,
+      "loss": 3.9909,
+      "step": 13065
+    },
+    {
+      "epoch": 0.8880282647098791,
+      "grad_norm": 0.23280905187129974,
+      "learning_rate": 8.890813969289306e-05,
+      "loss": 3.9899,
+      "step": 13070
+    },
+    {
+      "epoch": 0.8883679847805408,
+      "grad_norm": 0.21939218044281006,
+      "learning_rate": 8.890389319200979e-05,
+      "loss": 4.0157,
+      "step": 13075
+    },
+    {
+      "epoch": 0.8887077048512027,
+      "grad_norm": 0.19345150887966156,
+      "learning_rate": 8.889964669112652e-05,
+      "loss": 3.8884,
+      "step": 13080
+    },
+    {
+      "epoch": 0.8890474249218644,
+      "grad_norm": 0.18044187128543854,
+      "learning_rate": 8.889540019024324e-05,
+      "loss": 3.9464,
+      "step": 13085
+    },
+    {
+      "epoch": 0.8893871449925261,
+      "grad_norm": 0.7762069702148438,
+      "learning_rate": 8.889115368935997e-05,
+      "loss": 4.0273,
+      "step": 13090
+    },
+    {
+      "epoch": 0.889726865063188,
+      "grad_norm": 0.20319564640522003,
+      "learning_rate": 8.88869071884767e-05,
+      "loss": 3.943,
+      "step": 13095
+    },
+    {
+      "epoch": 0.8900665851338497,
+      "grad_norm": 0.4176552891731262,
+      "learning_rate": 8.888266068759343e-05,
+      "loss": 3.9463,
+      "step": 13100
+    },
+    {
+      "epoch": 0.8904063052045115,
+      "grad_norm": 0.6137194633483887,
+      "learning_rate": 8.887841418671016e-05,
+      "loss": 3.9732,
+      "step": 13105
+    },
+    {
+      "epoch": 0.8907460252751732,
+      "grad_norm": 0.3522084355354309,
+      "learning_rate": 8.887416768582688e-05,
+      "loss": 4.083,
+      "step": 13110
+    },
+    {
+      "epoch": 0.8910857453458351,
+      "grad_norm": 0.1649700552225113,
+      "learning_rate": 8.886992118494361e-05,
+      "loss": 3.8638,
+      "step": 13115
+    },
+    {
+      "epoch": 0.8914254654164968,
+      "grad_norm": 0.1624513566493988,
+      "learning_rate": 8.886567468406034e-05,
+      "loss": 4.1418,
+      "step": 13120
+    },
+    {
+      "epoch": 0.8917651854871586,
+      "grad_norm": 0.18356308341026306,
+      "learning_rate": 8.886142818317707e-05,
+      "loss": 3.8366,
+      "step": 13125
+    },
+    {
+      "epoch": 0.8921049055578204,
+      "grad_norm": 0.22682762145996094,
+      "learning_rate": 8.88571816822938e-05,
+      "loss": 3.8805,
+      "step": 13130
+    },
+    {
+      "epoch": 0.8924446256284821,
+      "grad_norm": 0.3890301287174225,
+      "learning_rate": 8.885293518141052e-05,
+      "loss": 3.8615,
+      "step": 13135
+    },
+    {
+      "epoch": 0.8927843456991439,
+      "grad_norm": 0.1815677434206009,
+      "learning_rate": 8.884868868052725e-05,
+      "loss": 3.9861,
+      "step": 13140
+    },
+    {
+      "epoch": 0.8931240657698056,
+      "grad_norm": 0.5793675780296326,
+      "learning_rate": 8.884444217964398e-05,
+      "loss": 4.0116,
+      "step": 13145
+    },
+    {
+      "epoch": 0.8934637858404675,
+      "grad_norm": 0.30733972787857056,
+      "learning_rate": 8.884019567876071e-05,
+      "loss": 3.8564,
+      "step": 13150
+    },
+    {
+      "epoch": 0.8938035059111292,
+      "grad_norm": 0.15819989144802094,
+      "learning_rate": 8.883594917787744e-05,
+      "loss": 4.2131,
+      "step": 13155
+    },
+    {
+      "epoch": 0.894143225981791,
+      "grad_norm": 0.1839209347963333,
+      "learning_rate": 8.883170267699416e-05,
+      "loss": 3.8485,
+      "step": 13160
+    },
+    {
+      "epoch": 0.8944829460524528,
+      "grad_norm": 0.2315215915441513,
+      "learning_rate": 8.882745617611089e-05,
+      "loss": 3.7713,
+      "step": 13165
+    },
+    {
+      "epoch": 0.8948226661231146,
+      "grad_norm": 0.18673592805862427,
+      "learning_rate": 8.882320967522762e-05,
+      "loss": 3.7431,
+      "step": 13170
+    },
+    {
+      "epoch": 0.8951623861937763,
+      "grad_norm": 0.1436724215745926,
+      "learning_rate": 8.881896317434435e-05,
+      "loss": 4.0778,
+      "step": 13175
+    },
+    {
+      "epoch": 0.8955021062644382,
+      "grad_norm": 0.23739652335643768,
+      "learning_rate": 8.881471667346108e-05,
+      "loss": 3.925,
+      "step": 13180
+    },
+    {
+      "epoch": 0.8958418263350999,
+      "grad_norm": 0.17830795049667358,
+      "learning_rate": 8.88104701725778e-05,
+      "loss": 4.0997,
+      "step": 13185
+    },
+    {
+      "epoch": 0.8961815464057616,
+      "grad_norm": 0.20703446865081787,
+      "learning_rate": 8.880622367169453e-05,
+      "loss": 3.8833,
+      "step": 13190
+    },
+    {
+      "epoch": 0.8965212664764234,
+      "grad_norm": 0.18058006465435028,
+      "learning_rate": 8.880197717081126e-05,
+      "loss": 4.1206,
+      "step": 13195
+    },
+    {
+      "epoch": 0.8968609865470852,
+      "grad_norm": 0.16602760553359985,
+      "learning_rate": 8.879773066992799e-05,
+      "loss": 4.0043,
+      "step": 13200
+    },
+    {
+      "epoch": 0.897200706617747,
+      "grad_norm": 0.2073521465063095,
+      "learning_rate": 8.879348416904472e-05,
+      "loss": 3.8992,
+      "step": 13205
+    },
+    {
+      "epoch": 0.8975404266884087,
+      "grad_norm": 0.1674661785364151,
+      "learning_rate": 8.878923766816144e-05,
+      "loss": 3.9651,
+      "step": 13210
+    },
+    {
+      "epoch": 0.8978801467590706,
+      "grad_norm": 0.19709919393062592,
+      "learning_rate": 8.878499116727816e-05,
+      "loss": 3.8494,
+      "step": 13215
+    },
+    {
+      "epoch": 0.8982198668297323,
+      "grad_norm": 0.18262673914432526,
+      "learning_rate": 8.87807446663949e-05,
+      "loss": 3.9929,
+      "step": 13220
+    },
+    {
+      "epoch": 0.8985595869003941,
+      "grad_norm": 0.1922248899936676,
+      "learning_rate": 8.877649816551163e-05,
+      "loss": 4.1698,
+      "step": 13225
+    },
+    {
+      "epoch": 0.8988993069710558,
+      "grad_norm": 0.22797107696533203,
+      "learning_rate": 8.877225166462834e-05,
+      "loss": 4.1415,
+      "step": 13230
+    },
+    {
+      "epoch": 0.8992390270417177,
+      "grad_norm": 0.21377937495708466,
+      "learning_rate": 8.876800516374508e-05,
+      "loss": 3.7365,
+      "step": 13235
+    },
+    {
+      "epoch": 0.8995787471123794,
+      "grad_norm": 0.15349645912647247,
+      "learning_rate": 8.876375866286181e-05,
+      "loss": 4.21,
+      "step": 13240
+    },
+    {
+      "epoch": 0.8999184671830411,
+      "grad_norm": 0.17594188451766968,
+      "learning_rate": 8.875951216197853e-05,
+      "loss": 3.6277,
+      "step": 13245
+    },
+    {
+      "epoch": 0.900258187253703,
+      "grad_norm": 0.21915188431739807,
+      "learning_rate": 8.875526566109527e-05,
+      "loss": 3.8465,
+      "step": 13250
+    },
+    {
+      "epoch": 0.9005979073243647,
+      "grad_norm": 0.1609984189271927,
+      "learning_rate": 8.8751019160212e-05,
+      "loss": 3.9646,
+      "step": 13255
+    },
+    {
+      "epoch": 0.9009376273950265,
+      "grad_norm": 0.22858203947544098,
+      "learning_rate": 8.874677265932871e-05,
+      "loss": 3.9217,
+      "step": 13260
+    },
+    {
+      "epoch": 0.9012773474656883,
+      "grad_norm": 0.1865098774433136,
+      "learning_rate": 8.874252615844545e-05,
+      "loss": 4.0665,
+      "step": 13265
+    },
+    {
+      "epoch": 0.9016170675363501,
+      "grad_norm": 0.27133429050445557,
+      "learning_rate": 8.873827965756218e-05,
+      "loss": 4.1901,
+      "step": 13270
+    },
+    {
+      "epoch": 0.9019567876070118,
+      "grad_norm": 0.26253220438957214,
+      "learning_rate": 8.87340331566789e-05,
+      "loss": 4.0598,
+      "step": 13275
+    },
+    {
+      "epoch": 0.9022965076776736,
+      "grad_norm": 0.19666416943073273,
+      "learning_rate": 8.872978665579564e-05,
+      "loss": 4.1986,
+      "step": 13280
+    },
+    {
+      "epoch": 0.9026362277483354,
+      "grad_norm": 0.16205628216266632,
+      "learning_rate": 8.872554015491235e-05,
+      "loss": 3.9244,
+      "step": 13285
+    },
+    {
+      "epoch": 0.9029759478189971,
+      "grad_norm": 0.21423132717609406,
+      "learning_rate": 8.872129365402908e-05,
+      "loss": 3.9477,
+      "step": 13290
+    },
+    {
+      "epoch": 0.9033156678896589,
+      "grad_norm": 0.18254421651363373,
+      "learning_rate": 8.871704715314582e-05,
+      "loss": 4.0589,
+      "step": 13295
+    },
+    {
+      "epoch": 0.9036553879603207,
+      "grad_norm": 0.17389804124832153,
+      "learning_rate": 8.871280065226253e-05,
+      "loss": 4.0397,
+      "step": 13300
+    },
+    {
+      "epoch": 0.9039951080309825,
+      "grad_norm": 0.23967847228050232,
+      "learning_rate": 8.870855415137926e-05,
+      "loss": 3.9713,
+      "step": 13305
+    },
+    {
+      "epoch": 0.9043348281016442,
+      "grad_norm": 0.15660813450813293,
+      "learning_rate": 8.8704307650496e-05,
+      "loss": 4.155,
+      "step": 13310
+    },
+    {
+      "epoch": 0.904674548172306,
+      "grad_norm": 0.1497335135936737,
+      "learning_rate": 8.870006114961272e-05,
+      "loss": 4.2737,
+      "step": 13315
+    },
+    {
+      "epoch": 0.9050142682429678,
+      "grad_norm": 0.16438940167427063,
+      "learning_rate": 8.869581464872945e-05,
+      "loss": 3.7956,
+      "step": 13320
+    },
+    {
+      "epoch": 0.9053539883136296,
+      "grad_norm": 0.19728811085224152,
+      "learning_rate": 8.869156814784619e-05,
+      "loss": 3.8597,
+      "step": 13325
+    },
+    {
+      "epoch": 0.9056937083842913,
+      "grad_norm": 0.19347457587718964,
+      "learning_rate": 8.86873216469629e-05,
+      "loss": 3.9375,
+      "step": 13330
+    },
+    {
+      "epoch": 0.9060334284549532,
+      "grad_norm": 0.1782982498407364,
+      "learning_rate": 8.868307514607963e-05,
+      "loss": 4.1215,
+      "step": 13335
+    },
+    {
+      "epoch": 0.9063731485256149,
+      "grad_norm": 0.16826027631759644,
+      "learning_rate": 8.867882864519637e-05,
+      "loss": 3.9993,
+      "step": 13340
+    },
+    {
+      "epoch": 0.9067128685962766,
+      "grad_norm": 0.1599569171667099,
+      "learning_rate": 8.867458214431309e-05,
+      "loss": 3.8991,
+      "step": 13345
+    },
+    {
+      "epoch": 0.9070525886669385,
+      "grad_norm": 2.387441396713257,
+      "learning_rate": 8.867033564342981e-05,
+      "loss": 4.0009,
+      "step": 13350
+    },
+    {
+      "epoch": 0.9073923087376002,
+      "grad_norm": 0.16353747248649597,
+      "learning_rate": 8.866608914254654e-05,
+      "loss": 3.8368,
+      "step": 13355
+    },
+    {
+      "epoch": 0.907732028808262,
+      "grad_norm": 0.18139252066612244,
+      "learning_rate": 8.866184264166327e-05,
+      "loss": 3.8567,
+      "step": 13360
+    },
+    {
+      "epoch": 0.9080717488789237,
+      "grad_norm": 0.28488507866859436,
+      "learning_rate": 8.865759614078e-05,
+      "loss": 3.9918,
+      "step": 13365
+    },
+    {
+      "epoch": 0.9084114689495856,
+      "grad_norm": 0.193324476480484,
+      "learning_rate": 8.865334963989673e-05,
+      "loss": 4.0164,
+      "step": 13370
+    },
+    {
+      "epoch": 0.9087511890202473,
+      "grad_norm": 0.18897615373134613,
+      "learning_rate": 8.864910313901345e-05,
+      "loss": 3.6622,
+      "step": 13375
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.18190960586071014,
+      "learning_rate": 8.864485663813018e-05,
+      "loss": 4.0802,
+      "step": 13380
+    },
+    {
+      "epoch": 0.9094306291615709,
+      "grad_norm": 0.3872554302215576,
+      "learning_rate": 8.864061013724691e-05,
+      "loss": 4.1349,
+      "step": 13385
+    },
+    {
+      "epoch": 0.9097703492322327,
+      "grad_norm": 0.18860095739364624,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 4.1308,
+      "step": 13390
+    },
+    {
+      "epoch": 0.9101100693028944,
+      "grad_norm": 0.1607290804386139,
+      "learning_rate": 8.863211713548037e-05,
+      "loss": 3.9187,
+      "step": 13395
+    },
+    {
+      "epoch": 0.9104497893735561,
+      "grad_norm": 0.24063172936439514,
+      "learning_rate": 8.86278706345971e-05,
+      "loss": 3.7071,
+      "step": 13400
+    },
+    {
+      "epoch": 0.910789509444218,
+      "grad_norm": 0.2261088341474533,
+      "learning_rate": 8.862362413371382e-05,
+      "loss": 4.3313,
+      "step": 13405
+    },
+    {
+      "epoch": 0.9111292295148797,
+      "grad_norm": 0.2024299055337906,
+      "learning_rate": 8.861937763283055e-05,
+      "loss": 3.7482,
+      "step": 13410
+    },
+    {
+      "epoch": 0.9114689495855415,
+      "grad_norm": 0.32624539732933044,
+      "learning_rate": 8.861513113194728e-05,
+      "loss": 3.9679,
+      "step": 13415
+    },
+    {
+      "epoch": 0.9118086696562033,
+      "grad_norm": 0.2082969844341278,
+      "learning_rate": 8.8610884631064e-05,
+      "loss": 4.1913,
+      "step": 13420
+    },
+    {
+      "epoch": 0.9121483897268651,
+      "grad_norm": 0.20776890218257904,
+      "learning_rate": 8.860663813018073e-05,
+      "loss": 4.0744,
+      "step": 13425
+    },
+    {
+      "epoch": 0.9124881097975268,
+      "grad_norm": 0.16512270271778107,
+      "learning_rate": 8.860239162929746e-05,
+      "loss": 4.0043,
+      "step": 13430
+    },
+    {
+      "epoch": 0.9128278298681887,
+      "grad_norm": 1.4505500793457031,
+      "learning_rate": 8.859814512841419e-05,
+      "loss": 3.8782,
+      "step": 13435
+    },
+    {
+      "epoch": 0.9131675499388504,
+      "grad_norm": 0.25911930203437805,
+      "learning_rate": 8.859389862753092e-05,
+      "loss": 3.6646,
+      "step": 13440
+    },
+    {
+      "epoch": 0.9135072700095122,
+      "grad_norm": 0.23000332713127136,
+      "learning_rate": 8.858965212664765e-05,
+      "loss": 4.0076,
+      "step": 13445
+    },
+    {
+      "epoch": 0.9138469900801739,
+      "grad_norm": 0.19737814366817474,
+      "learning_rate": 8.858540562576437e-05,
+      "loss": 3.8124,
+      "step": 13450
+    },
+    {
+      "epoch": 0.9141867101508357,
+      "grad_norm": 0.1637062281370163,
+      "learning_rate": 8.85811591248811e-05,
+      "loss": 3.8238,
+      "step": 13455
+    },
+    {
+      "epoch": 0.9145264302214975,
+      "grad_norm": 0.24078230559825897,
+      "learning_rate": 8.857691262399783e-05,
+      "loss": 4.2022,
+      "step": 13460
+    },
+    {
+      "epoch": 0.9148661502921592,
+      "grad_norm": 0.17801740765571594,
+      "learning_rate": 8.857266612311456e-05,
+      "loss": 4.11,
+      "step": 13465
+    },
+    {
+      "epoch": 0.9152058703628211,
+      "grad_norm": 0.18943698704242706,
+      "learning_rate": 8.856841962223129e-05,
+      "loss": 4.0061,
+      "step": 13470
+    },
+    {
+      "epoch": 0.9155455904334828,
+      "grad_norm": 0.1784680187702179,
+      "learning_rate": 8.856417312134801e-05,
+      "loss": 3.9399,
+      "step": 13475
+    },
+    {
+      "epoch": 0.9158853105041446,
+      "grad_norm": 0.19242218136787415,
+      "learning_rate": 8.855992662046474e-05,
+      "loss": 3.7147,
+      "step": 13480
+    },
+    {
+      "epoch": 0.9162250305748063,
+      "grad_norm": 0.1983332335948944,
+      "learning_rate": 8.855568011958147e-05,
+      "loss": 3.7828,
+      "step": 13485
+    },
+    {
+      "epoch": 0.9165647506454682,
+      "grad_norm": 0.1877221167087555,
+      "learning_rate": 8.85514336186982e-05,
+      "loss": 3.9979,
+      "step": 13490
+    },
+    {
+      "epoch": 0.9169044707161299,
+      "grad_norm": 0.287514328956604,
+      "learning_rate": 8.854718711781493e-05,
+      "loss": 4.0574,
+      "step": 13495
+    },
+    {
+      "epoch": 0.9172441907867916,
+      "grad_norm": 0.1971914917230606,
+      "learning_rate": 8.854294061693165e-05,
+      "loss": 3.8822,
+      "step": 13500
+    },
+    {
+      "epoch": 0.9175839108574535,
+      "grad_norm": 0.19974425435066223,
+      "learning_rate": 8.853869411604838e-05,
+      "loss": 4.0866,
+      "step": 13505
+    },
+    {
+      "epoch": 0.9179236309281152,
+      "grad_norm": 0.1471916139125824,
+      "learning_rate": 8.853444761516511e-05,
+      "loss": 4.2775,
+      "step": 13510
+    },
+    {
+      "epoch": 0.918263350998777,
+      "grad_norm": 0.2695204019546509,
+      "learning_rate": 8.853020111428184e-05,
+      "loss": 3.5593,
+      "step": 13515
+    },
+    {
+      "epoch": 0.9186030710694388,
+      "grad_norm": 0.3328780233860016,
+      "learning_rate": 8.852595461339857e-05,
+      "loss": 4.0474,
+      "step": 13520
+    },
+    {
+      "epoch": 0.9189427911401006,
+      "grad_norm": 0.22376228868961334,
+      "learning_rate": 8.85217081125153e-05,
+      "loss": 3.8726,
+      "step": 13525
+    },
+    {
+      "epoch": 0.9192825112107623,
+      "grad_norm": 1.0047399997711182,
+      "learning_rate": 8.851746161163202e-05,
+      "loss": 3.6569,
+      "step": 13530
+    },
+    {
+      "epoch": 0.9196222312814241,
+      "grad_norm": 0.20537297427654266,
+      "learning_rate": 8.851321511074875e-05,
+      "loss": 4.0576,
+      "step": 13535
+    },
+    {
+      "epoch": 0.9199619513520859,
+      "grad_norm": 0.19384372234344482,
+      "learning_rate": 8.850896860986548e-05,
+      "loss": 4.1421,
+      "step": 13540
+    },
+    {
+      "epoch": 0.9203016714227477,
+      "grad_norm": 0.24288725852966309,
+      "learning_rate": 8.85047221089822e-05,
+      "loss": 3.9712,
+      "step": 13545
+    },
+    {
+      "epoch": 0.9206413914934094,
+      "grad_norm": 0.17010895907878876,
+      "learning_rate": 8.850047560809893e-05,
+      "loss": 4.1179,
+      "step": 13550
+    },
+    {
+      "epoch": 0.9209811115640713,
+      "grad_norm": 0.2580501139163971,
+      "learning_rate": 8.849622910721565e-05,
+      "loss": 3.963,
+      "step": 13555
+    },
+    {
+      "epoch": 0.921320831634733,
+      "grad_norm": 0.19051308929920197,
+      "learning_rate": 8.849198260633239e-05,
+      "loss": 3.9626,
+      "step": 13560
+    },
+    {
+      "epoch": 0.9216605517053947,
+      "grad_norm": 0.18489578366279602,
+      "learning_rate": 8.848773610544912e-05,
+      "loss": 4.0235,
+      "step": 13565
+    },
+    {
+      "epoch": 0.9220002717760565,
+      "grad_norm": 0.1751706898212433,
+      "learning_rate": 8.848348960456583e-05,
+      "loss": 4.03,
+      "step": 13570
+    },
+    {
+      "epoch": 0.9223399918467183,
+      "grad_norm": 0.15869379043579102,
+      "learning_rate": 8.847924310368257e-05,
+      "loss": 3.9887,
+      "step": 13575
+    },
+    {
+      "epoch": 0.9226797119173801,
+      "grad_norm": 0.18910722434520721,
+      "learning_rate": 8.84749966027993e-05,
+      "loss": 3.9694,
+      "step": 13580
+    },
+    {
+      "epoch": 0.9230194319880418,
+      "grad_norm": 0.18458014726638794,
+      "learning_rate": 8.847075010191602e-05,
+      "loss": 3.7103,
+      "step": 13585
+    },
+    {
+      "epoch": 0.9233591520587037,
+      "grad_norm": 0.15150383114814758,
+      "learning_rate": 8.846650360103276e-05,
+      "loss": 3.9727,
+      "step": 13590
+    },
+    {
+      "epoch": 0.9236988721293654,
+      "grad_norm": 0.1747668832540512,
+      "learning_rate": 8.846225710014949e-05,
+      "loss": 4.0081,
+      "step": 13595
+    },
+    {
+      "epoch": 0.9240385922000272,
+      "grad_norm": 0.1757059097290039,
+      "learning_rate": 8.84580105992662e-05,
+      "loss": 3.8936,
+      "step": 13600
+    },
+    {
+      "epoch": 0.924378312270689,
+      "grad_norm": 0.17729350924491882,
+      "learning_rate": 8.845376409838294e-05,
+      "loss": 4.1143,
+      "step": 13605
+    },
+    {
+      "epoch": 0.9247180323413507,
+      "grad_norm": 0.18305832147598267,
+      "learning_rate": 8.844951759749967e-05,
+      "loss": 4.1294,
+      "step": 13610
+    },
+    {
+      "epoch": 0.9250577524120125,
+      "grad_norm": 1.0373708009719849,
+      "learning_rate": 8.844527109661639e-05,
+      "loss": 4.0552,
+      "step": 13615
+    },
+    {
+      "epoch": 0.9253974724826742,
+      "grad_norm": 0.18093042075634003,
+      "learning_rate": 8.844102459573313e-05,
+      "loss": 4.0405,
+      "step": 13620
+    },
+    {
+      "epoch": 0.9257371925533361,
+      "grad_norm": 0.1686364710330963,
+      "learning_rate": 8.843677809484985e-05,
+      "loss": 4.0729,
+      "step": 13625
+    },
+    {
+      "epoch": 0.9260769126239978,
+      "grad_norm": 0.12190663069486618,
+      "learning_rate": 8.843253159396657e-05,
+      "loss": 3.958,
+      "step": 13630
+    },
+    {
+      "epoch": 0.9264166326946596,
+      "grad_norm": 0.18013213574886322,
+      "learning_rate": 8.842828509308331e-05,
+      "loss": 4.0457,
+      "step": 13635
+    },
+    {
+      "epoch": 0.9267563527653214,
+      "grad_norm": 0.47123172879219055,
+      "learning_rate": 8.842403859220003e-05,
+      "loss": 3.9378,
+      "step": 13640
+    },
+    {
+      "epoch": 0.9270960728359832,
+      "grad_norm": 0.17332005500793457,
+      "learning_rate": 8.841979209131675e-05,
+      "loss": 3.9994,
+      "step": 13645
+    },
+    {
+      "epoch": 0.9274357929066449,
+      "grad_norm": 0.2330494374036789,
+      "learning_rate": 8.84155455904335e-05,
+      "loss": 4.2059,
+      "step": 13650
+    },
+    {
+      "epoch": 0.9277755129773066,
+      "grad_norm": 0.1844603419303894,
+      "learning_rate": 8.841129908955021e-05,
+      "loss": 3.9922,
+      "step": 13655
+    },
+    {
+      "epoch": 0.9281152330479685,
+      "grad_norm": 0.19840815663337708,
+      "learning_rate": 8.840705258866694e-05,
+      "loss": 4.0601,
+      "step": 13660
+    },
+    {
+      "epoch": 0.9284549531186302,
+      "grad_norm": 0.25802430510520935,
+      "learning_rate": 8.840280608778368e-05,
+      "loss": 3.8963,
+      "step": 13665
+    },
+    {
+      "epoch": 0.928794673189292,
+      "grad_norm": 0.1573478877544403,
+      "learning_rate": 8.83985595869004e-05,
+      "loss": 4.1013,
+      "step": 13670
+    },
+    {
+      "epoch": 0.9291343932599538,
+      "grad_norm": 0.20257075130939484,
+      "learning_rate": 8.839431308601712e-05,
+      "loss": 4.0064,
+      "step": 13675
+    },
+    {
+      "epoch": 0.9294741133306156,
+      "grad_norm": 0.2046387791633606,
+      "learning_rate": 8.839006658513386e-05,
+      "loss": 3.948,
+      "step": 13680
+    },
+    {
+      "epoch": 0.9298138334012773,
+      "grad_norm": 0.29030993580818176,
+      "learning_rate": 8.838582008425058e-05,
+      "loss": 3.7141,
+      "step": 13685
+    },
+    {
+      "epoch": 0.9301535534719392,
+      "grad_norm": 0.1701250672340393,
+      "learning_rate": 8.83815735833673e-05,
+      "loss": 3.8852,
+      "step": 13690
+    },
+    {
+      "epoch": 0.9304932735426009,
+      "grad_norm": 0.18545231223106384,
+      "learning_rate": 8.837732708248405e-05,
+      "loss": 3.8857,
+      "step": 13695
+    },
+    {
+      "epoch": 0.9308329936132627,
+      "grad_norm": 0.22790156304836273,
+      "learning_rate": 8.837308058160076e-05,
+      "loss": 3.9376,
+      "step": 13700
+    },
+    {
+      "epoch": 0.9311727136839244,
+      "grad_norm": 0.2060774713754654,
+      "learning_rate": 8.836883408071749e-05,
+      "loss": 4.0047,
+      "step": 13705
+    },
+    {
+      "epoch": 0.9315124337545863,
+      "grad_norm": 0.17100512981414795,
+      "learning_rate": 8.836458757983422e-05,
+      "loss": 4.0347,
+      "step": 13710
+    },
+    {
+      "epoch": 0.931852153825248,
+      "grad_norm": 0.18439409136772156,
+      "learning_rate": 8.836034107895095e-05,
+      "loss": 4.0597,
+      "step": 13715
+    },
+    {
+      "epoch": 0.9321918738959097,
+      "grad_norm": 0.20171356201171875,
+      "learning_rate": 8.835609457806767e-05,
+      "loss": 3.8904,
+      "step": 13720
+    },
+    {
+      "epoch": 0.9325315939665716,
+      "grad_norm": 0.2539232075214386,
+      "learning_rate": 8.83518480771844e-05,
+      "loss": 3.7127,
+      "step": 13725
+    },
+    {
+      "epoch": 0.9328713140372333,
+      "grad_norm": 0.303066611289978,
+      "learning_rate": 8.834760157630113e-05,
+      "loss": 3.858,
+      "step": 13730
+    },
+    {
+      "epoch": 0.9332110341078951,
+      "grad_norm": 10.150884628295898,
+      "learning_rate": 8.834335507541786e-05,
+      "loss": 4.214,
+      "step": 13735
+    },
+    {
+      "epoch": 0.9335507541785568,
+      "grad_norm": 0.17438891530036926,
+      "learning_rate": 8.833910857453459e-05,
+      "loss": 3.9798,
+      "step": 13740
+    },
+    {
+      "epoch": 0.9338904742492187,
+      "grad_norm": 0.1963614672422409,
+      "learning_rate": 8.833486207365131e-05,
+      "loss": 4.0225,
+      "step": 13745
+    },
+    {
+      "epoch": 0.9342301943198804,
+      "grad_norm": 0.33438101410865784,
+      "learning_rate": 8.833061557276804e-05,
+      "loss": 3.8237,
+      "step": 13750
+    },
+    {
+      "epoch": 0.9345699143905422,
+      "grad_norm": 0.15952834486961365,
+      "learning_rate": 8.832636907188477e-05,
+      "loss": 4.0374,
+      "step": 13755
+    },
+    {
+      "epoch": 0.934909634461204,
+      "grad_norm": 0.20450946688652039,
+      "learning_rate": 8.83221225710015e-05,
+      "loss": 3.8739,
+      "step": 13760
+    },
+    {
+      "epoch": 0.9352493545318658,
+      "grad_norm": 0.2039669156074524,
+      "learning_rate": 8.831787607011823e-05,
+      "loss": 3.953,
+      "step": 13765
+    },
+    {
+      "epoch": 0.9355890746025275,
+      "grad_norm": 0.15678437054157257,
+      "learning_rate": 8.831362956923495e-05,
+      "loss": 3.9398,
+      "step": 13770
+    },
+    {
+      "epoch": 0.9359287946731893,
+      "grad_norm": 0.5224528312683105,
+      "learning_rate": 8.830938306835168e-05,
+      "loss": 3.8759,
+      "step": 13775
+    },
+    {
+      "epoch": 0.9362685147438511,
+      "grad_norm": 0.2511097192764282,
+      "learning_rate": 8.830513656746841e-05,
+      "loss": 3.9613,
+      "step": 13780
+    },
+    {
+      "epoch": 0.9366082348145128,
+      "grad_norm": 0.14598341286182404,
+      "learning_rate": 8.830089006658514e-05,
+      "loss": 4.0519,
+      "step": 13785
+    },
+    {
+      "epoch": 0.9369479548851746,
+      "grad_norm": 0.18947695195674896,
+      "learning_rate": 8.829664356570187e-05,
+      "loss": 3.9679,
+      "step": 13790
+    },
+    {
+      "epoch": 0.9372876749558364,
+      "grad_norm": 0.19946783781051636,
+      "learning_rate": 8.82923970648186e-05,
+      "loss": 4.0397,
+      "step": 13795
+    },
+    {
+      "epoch": 0.9376273950264982,
+      "grad_norm": 0.1777060627937317,
+      "learning_rate": 8.828815056393532e-05,
+      "loss": 3.8039,
+      "step": 13800
+    },
+    {
+      "epoch": 0.9379671150971599,
+      "grad_norm": 0.2687058746814728,
+      "learning_rate": 8.828390406305205e-05,
+      "loss": 3.9514,
+      "step": 13805
+    },
+    {
+      "epoch": 0.9383068351678218,
+      "grad_norm": 0.17168082296848297,
+      "learning_rate": 8.827965756216878e-05,
+      "loss": 3.6143,
+      "step": 13810
+    },
+    {
+      "epoch": 0.9386465552384835,
+      "grad_norm": 0.2019256055355072,
+      "learning_rate": 8.82754110612855e-05,
+      "loss": 3.9093,
+      "step": 13815
+    },
+    {
+      "epoch": 0.9389862753091452,
+      "grad_norm": 0.15410295128822327,
+      "learning_rate": 8.827116456040223e-05,
+      "loss": 4.0334,
+      "step": 13820
+    },
+    {
+      "epoch": 0.939325995379807,
+      "grad_norm": 0.18953141570091248,
+      "learning_rate": 8.826691805951896e-05,
+      "loss": 3.5909,
+      "step": 13825
+    },
+    {
+      "epoch": 0.9396657154504688,
+      "grad_norm": 0.4388710856437683,
+      "learning_rate": 8.826267155863569e-05,
+      "loss": 4.0782,
+      "step": 13830
+    },
+    {
+      "epoch": 0.9400054355211306,
+      "grad_norm": 0.5807581543922424,
+      "learning_rate": 8.825842505775242e-05,
+      "loss": 3.7892,
+      "step": 13835
+    },
+    {
+      "epoch": 0.9403451555917923,
+      "grad_norm": 0.14411011338233948,
+      "learning_rate": 8.825417855686915e-05,
+      "loss": 3.9576,
+      "step": 13840
+    },
+    {
+      "epoch": 0.9406848756624542,
+      "grad_norm": 0.21537935733795166,
+      "learning_rate": 8.824993205598587e-05,
+      "loss": 3.8762,
+      "step": 13845
+    },
+    {
+      "epoch": 0.9410245957331159,
+      "grad_norm": 1.785944938659668,
+      "learning_rate": 8.82456855551026e-05,
+      "loss": 4.103,
+      "step": 13850
+    },
+    {
+      "epoch": 0.9413643158037777,
+      "grad_norm": 0.13651463389396667,
+      "learning_rate": 8.824143905421933e-05,
+      "loss": 4.032,
+      "step": 13855
+    },
+    {
+      "epoch": 0.9417040358744395,
+      "grad_norm": 0.16791968047618866,
+      "learning_rate": 8.823719255333606e-05,
+      "loss": 3.9523,
+      "step": 13860
+    },
+    {
+      "epoch": 0.9420437559451013,
+      "grad_norm": 0.22517231106758118,
+      "learning_rate": 8.823294605245279e-05,
+      "loss": 3.8512,
+      "step": 13865
+    },
+    {
+      "epoch": 0.942383476015763,
+      "grad_norm": 0.2402997463941574,
+      "learning_rate": 8.822869955156951e-05,
+      "loss": 3.9759,
+      "step": 13870
+    },
+    {
+      "epoch": 0.9427231960864247,
+      "grad_norm": 0.20395690202713013,
+      "learning_rate": 8.822445305068624e-05,
+      "loss": 3.9377,
+      "step": 13875
+    },
+    {
+      "epoch": 0.9430629161570866,
+      "grad_norm": 0.24165527522563934,
+      "learning_rate": 8.822020654980297e-05,
+      "loss": 3.9313,
+      "step": 13880
+    },
+    {
+      "epoch": 0.9434026362277483,
+      "grad_norm": 0.3844189941883087,
+      "learning_rate": 8.82159600489197e-05,
+      "loss": 3.7314,
+      "step": 13885
+    },
+    {
+      "epoch": 0.9437423562984101,
+      "grad_norm": 0.2007877379655838,
+      "learning_rate": 8.821171354803643e-05,
+      "loss": 4.0631,
+      "step": 13890
+    },
+    {
+      "epoch": 0.9440820763690719,
+      "grad_norm": 0.18726769089698792,
+      "learning_rate": 8.820746704715315e-05,
+      "loss": 4.055,
+      "step": 13895
+    },
+    {
+      "epoch": 0.9444217964397337,
+      "grad_norm": 0.38226640224456787,
+      "learning_rate": 8.820322054626988e-05,
+      "loss": 4.0028,
+      "step": 13900
+    },
+    {
+      "epoch": 0.9447615165103954,
+      "grad_norm": 0.1904405802488327,
+      "learning_rate": 8.819897404538661e-05,
+      "loss": 3.9834,
+      "step": 13905
+    },
+    {
+      "epoch": 0.9451012365810572,
+      "grad_norm": 0.1809813529253006,
+      "learning_rate": 8.819472754450332e-05,
+      "loss": 4.0637,
+      "step": 13910
+    },
+    {
+      "epoch": 0.945440956651719,
+      "grad_norm": 0.4116685092449188,
+      "learning_rate": 8.819048104362007e-05,
+      "loss": 3.9916,
+      "step": 13915
+    },
+    {
+      "epoch": 0.9457806767223808,
+      "grad_norm": 1.9458621740341187,
+      "learning_rate": 8.81862345427368e-05,
+      "loss": 4.0563,
+      "step": 13920
+    },
+    {
+      "epoch": 0.9461203967930425,
+      "grad_norm": 0.15543721616268158,
+      "learning_rate": 8.818198804185351e-05,
+      "loss": 4.0744,
+      "step": 13925
+    },
+    {
+      "epoch": 0.9464601168637043,
+      "grad_norm": 0.42934080958366394,
+      "learning_rate": 8.817774154097025e-05,
+      "loss": 3.8221,
+      "step": 13930
+    },
+    {
+      "epoch": 0.9467998369343661,
+      "grad_norm": 0.228485107421875,
+      "learning_rate": 8.817349504008698e-05,
+      "loss": 3.9268,
+      "step": 13935
+    },
+    {
+      "epoch": 0.9471395570050278,
+      "grad_norm": 0.21295882761478424,
+      "learning_rate": 8.816924853920369e-05,
+      "loss": 4.2737,
+      "step": 13940
+    },
+    {
+      "epoch": 0.9474792770756897,
+      "grad_norm": 0.2023978978395462,
+      "learning_rate": 8.816500203832043e-05,
+      "loss": 4.0087,
+      "step": 13945
+    },
+    {
+      "epoch": 0.9478189971463514,
+      "grad_norm": 0.1767956167459488,
+      "learning_rate": 8.816075553743716e-05,
+      "loss": 3.9129,
+      "step": 13950
+    },
+    {
+      "epoch": 0.9481587172170132,
+      "grad_norm": 1.3790664672851562,
+      "learning_rate": 8.815650903655388e-05,
+      "loss": 3.6768,
+      "step": 13955
+    },
+    {
+      "epoch": 0.9484984372876749,
+      "grad_norm": 0.583037793636322,
+      "learning_rate": 8.815226253567062e-05,
+      "loss": 3.8518,
+      "step": 13960
+    },
+    {
+      "epoch": 0.9488381573583368,
+      "grad_norm": 0.16457735002040863,
+      "learning_rate": 8.814801603478735e-05,
+      "loss": 3.7035,
+      "step": 13965
+    },
+    {
+      "epoch": 0.9491778774289985,
+      "grad_norm": 0.2307073473930359,
+      "learning_rate": 8.814376953390406e-05,
+      "loss": 4.1654,
+      "step": 13970
+    },
+    {
+      "epoch": 0.9495175974996602,
+      "grad_norm": 0.3080320358276367,
+      "learning_rate": 8.81395230330208e-05,
+      "loss": 3.975,
+      "step": 13975
+    },
+    {
+      "epoch": 0.9498573175703221,
+      "grad_norm": 0.17688848078250885,
+      "learning_rate": 8.813527653213752e-05,
+      "loss": 4.0859,
+      "step": 13980
+    },
+    {
+      "epoch": 0.9501970376409838,
+      "grad_norm": 0.24346010386943817,
+      "learning_rate": 8.813103003125424e-05,
+      "loss": 3.9282,
+      "step": 13985
+    },
+    {
+      "epoch": 0.9505367577116456,
+      "grad_norm": 0.1712096929550171,
+      "learning_rate": 8.812678353037099e-05,
+      "loss": 3.9035,
+      "step": 13990
+    },
+    {
+      "epoch": 0.9508764777823073,
+      "grad_norm": 0.19020821154117584,
+      "learning_rate": 8.81225370294877e-05,
+      "loss": 4.1025,
+      "step": 13995
+    },
+    {
+      "epoch": 0.9512161978529692,
+      "grad_norm": 0.21272438764572144,
+      "learning_rate": 8.811829052860443e-05,
+      "loss": 3.9025,
+      "step": 14000
+    },
+    {
+      "epoch": 0.9515559179236309,
+      "grad_norm": 0.14927330613136292,
+      "learning_rate": 8.811404402772117e-05,
+      "loss": 3.6392,
+      "step": 14005
+    },
+    {
+      "epoch": 0.9518956379942927,
+      "grad_norm": 0.14234548807144165,
+      "learning_rate": 8.810979752683788e-05,
+      "loss": 3.7538,
+      "step": 14010
+    },
+    {
+      "epoch": 0.9522353580649545,
+      "grad_norm": 0.19329407811164856,
+      "learning_rate": 8.810555102595461e-05,
+      "loss": 4.0138,
+      "step": 14015
+    },
+    {
+      "epoch": 0.9525750781356163,
+      "grad_norm": 1.4322816133499146,
+      "learning_rate": 8.810130452507135e-05,
+      "loss": 3.8935,
+      "step": 14020
+    },
+    {
+      "epoch": 0.952914798206278,
+      "grad_norm": 0.19049431383609772,
+      "learning_rate": 8.809705802418807e-05,
+      "loss": 4.0516,
+      "step": 14025
+    },
+    {
+      "epoch": 0.9532545182769399,
+      "grad_norm": 0.2139282375574112,
+      "learning_rate": 8.80928115233048e-05,
+      "loss": 3.8331,
+      "step": 14030
+    },
+    {
+      "epoch": 0.9535942383476016,
+      "grad_norm": 0.17371760308742523,
+      "learning_rate": 8.808856502242154e-05,
+      "loss": 3.9821,
+      "step": 14035
+    },
+    {
+      "epoch": 0.9539339584182633,
+      "grad_norm": 0.16707171499729156,
+      "learning_rate": 8.808431852153825e-05,
+      "loss": 4.1538,
+      "step": 14040
+    },
+    {
+      "epoch": 0.9542736784889251,
+      "grad_norm": 0.2354259192943573,
+      "learning_rate": 8.808007202065498e-05,
+      "loss": 3.9828,
+      "step": 14045
+    },
+    {
+      "epoch": 0.9546133985595869,
+      "grad_norm": 0.26454171538352966,
+      "learning_rate": 8.807582551977172e-05,
+      "loss": 4.1493,
+      "step": 14050
+    },
+    {
+      "epoch": 0.9549531186302487,
+      "grad_norm": 0.26322662830352783,
+      "learning_rate": 8.807157901888844e-05,
+      "loss": 4.0161,
+      "step": 14055
+    },
+    {
+      "epoch": 0.9552928387009104,
+      "grad_norm": 0.5517749786376953,
+      "learning_rate": 8.806733251800516e-05,
+      "loss": 3.8552,
+      "step": 14060
+    },
+    {
+      "epoch": 0.9556325587715723,
+      "grad_norm": 0.34178662300109863,
+      "learning_rate": 8.806308601712189e-05,
+      "loss": 3.9374,
+      "step": 14065
+    },
+    {
+      "epoch": 0.955972278842234,
+      "grad_norm": 0.4318985044956207,
+      "learning_rate": 8.805883951623862e-05,
+      "loss": 3.9312,
+      "step": 14070
+    },
+    {
+      "epoch": 0.9563119989128958,
+      "grad_norm": 0.7835567593574524,
+      "learning_rate": 8.805459301535535e-05,
+      "loss": 4.0294,
+      "step": 14075
+    },
+    {
+      "epoch": 0.9566517189835575,
+      "grad_norm": 0.248532235622406,
+      "learning_rate": 8.805034651447208e-05,
+      "loss": 4.0144,
+      "step": 14080
+    },
+    {
+      "epoch": 0.9569914390542194,
+      "grad_norm": 0.17486423254013062,
+      "learning_rate": 8.80461000135888e-05,
+      "loss": 3.8262,
+      "step": 14085
+    },
+    {
+      "epoch": 0.9573311591248811,
+      "grad_norm": 0.1894778460264206,
+      "learning_rate": 8.804185351270553e-05,
+      "loss": 3.989,
+      "step": 14090
+    },
+    {
+      "epoch": 0.9576708791955428,
+      "grad_norm": 0.1900128871202469,
+      "learning_rate": 8.803760701182226e-05,
+      "loss": 4.0652,
+      "step": 14095
+    },
+    {
+      "epoch": 0.9580105992662047,
+      "grad_norm": 0.21038229763507843,
+      "learning_rate": 8.803336051093899e-05,
+      "loss": 3.7708,
+      "step": 14100
+    },
+    {
+      "epoch": 0.9583503193368664,
+      "grad_norm": 0.22659200429916382,
+      "learning_rate": 8.802911401005572e-05,
+      "loss": 3.701,
+      "step": 14105
+    },
+    {
+      "epoch": 0.9586900394075282,
+      "grad_norm": 0.18774689733982086,
+      "learning_rate": 8.802486750917244e-05,
+      "loss": 3.8036,
+      "step": 14110
+    },
+    {
+      "epoch": 0.95902975947819,
+      "grad_norm": 0.7720076441764832,
+      "learning_rate": 8.802062100828917e-05,
+      "loss": 4.0017,
+      "step": 14115
+    },
+    {
+      "epoch": 0.9593694795488518,
+      "grad_norm": 0.17789320647716522,
+      "learning_rate": 8.80163745074059e-05,
+      "loss": 4.3506,
+      "step": 14120
+    },
+    {
+      "epoch": 0.9597091996195135,
+      "grad_norm": 0.23013809323310852,
+      "learning_rate": 8.801212800652263e-05,
+      "loss": 4.1502,
+      "step": 14125
+    },
+    {
+      "epoch": 0.9600489196901753,
+      "grad_norm": 0.20013386011123657,
+      "learning_rate": 8.800788150563936e-05,
+      "loss": 4.002,
+      "step": 14130
+    },
+    {
+      "epoch": 0.9603886397608371,
+      "grad_norm": 0.331853449344635,
+      "learning_rate": 8.800363500475608e-05,
+      "loss": 3.9042,
+      "step": 14135
+    },
+    {
+      "epoch": 0.9607283598314988,
+      "grad_norm": 0.19631457328796387,
+      "learning_rate": 8.799938850387281e-05,
+      "loss": 4.1109,
+      "step": 14140
+    },
+    {
+      "epoch": 0.9610680799021606,
+      "grad_norm": 0.38952094316482544,
+      "learning_rate": 8.799514200298954e-05,
+      "loss": 3.7741,
+      "step": 14145
+    },
+    {
+      "epoch": 0.9614077999728224,
+      "grad_norm": 0.15865278244018555,
+      "learning_rate": 8.799089550210627e-05,
+      "loss": 4.0511,
+      "step": 14150
+    },
+    {
+      "epoch": 0.9617475200434842,
+      "grad_norm": 0.20270100235939026,
+      "learning_rate": 8.7986649001223e-05,
+      "loss": 3.9072,
+      "step": 14155
+    },
+    {
+      "epoch": 0.9620872401141459,
+      "grad_norm": 0.17968709766864777,
+      "learning_rate": 8.798240250033972e-05,
+      "loss": 4.0201,
+      "step": 14160
+    },
+    {
+      "epoch": 0.9624269601848077,
+      "grad_norm": 0.19091679155826569,
+      "learning_rate": 8.797815599945645e-05,
+      "loss": 4.0135,
+      "step": 14165
+    },
+    {
+      "epoch": 0.9627666802554695,
+      "grad_norm": 0.18477921187877655,
+      "learning_rate": 8.797390949857318e-05,
+      "loss": 4.0118,
+      "step": 14170
+    },
+    {
+      "epoch": 0.9631064003261313,
+      "grad_norm": 0.1884998232126236,
+      "learning_rate": 8.796966299768991e-05,
+      "loss": 3.9669,
+      "step": 14175
+    },
+    {
+      "epoch": 0.963446120396793,
+      "grad_norm": 0.18764863908290863,
+      "learning_rate": 8.796541649680664e-05,
+      "loss": 3.9139,
+      "step": 14180
+    },
+    {
+      "epoch": 0.9637858404674549,
+      "grad_norm": 0.3251302242279053,
+      "learning_rate": 8.796116999592336e-05,
+      "loss": 4.1504,
+      "step": 14185
+    },
+    {
+      "epoch": 0.9641255605381166,
+      "grad_norm": 0.17968013882637024,
+      "learning_rate": 8.795692349504009e-05,
+      "loss": 3.8325,
+      "step": 14190
+    },
+    {
+      "epoch": 0.9644652806087783,
+      "grad_norm": 0.28362056612968445,
+      "learning_rate": 8.795267699415682e-05,
+      "loss": 3.945,
+      "step": 14195
+    },
+    {
+      "epoch": 0.9648050006794402,
+      "grad_norm": 0.18230807781219482,
+      "learning_rate": 8.794843049327355e-05,
+      "loss": 3.8641,
+      "step": 14200
+    },
+    {
+      "epoch": 0.9651447207501019,
+      "grad_norm": 0.20012174546718597,
+      "learning_rate": 8.794418399239028e-05,
+      "loss": 3.8536,
+      "step": 14205
+    },
+    {
+      "epoch": 0.9654844408207637,
+      "grad_norm": 0.16306108236312866,
+      "learning_rate": 8.7939937491507e-05,
+      "loss": 3.9824,
+      "step": 14210
+    },
+    {
+      "epoch": 0.9658241608914254,
+      "grad_norm": 0.19044755399227142,
+      "learning_rate": 8.793569099062373e-05,
+      "loss": 4.0477,
+      "step": 14215
+    },
+    {
+      "epoch": 0.9661638809620873,
+      "grad_norm": 0.9334998726844788,
+      "learning_rate": 8.793144448974046e-05,
+      "loss": 4.0839,
+      "step": 14220
+    },
+    {
+      "epoch": 0.966503601032749,
+      "grad_norm": 0.2080710530281067,
+      "learning_rate": 8.792719798885719e-05,
+      "loss": 3.935,
+      "step": 14225
+    },
+    {
+      "epoch": 0.9668433211034108,
+      "grad_norm": 0.20228326320648193,
+      "learning_rate": 8.792295148797392e-05,
+      "loss": 3.8579,
+      "step": 14230
+    },
+    {
+      "epoch": 0.9671830411740726,
+      "grad_norm": 0.14868243038654327,
+      "learning_rate": 8.791870498709064e-05,
+      "loss": 3.9681,
+      "step": 14235
+    },
+    {
+      "epoch": 0.9675227612447344,
+      "grad_norm": 0.3080695569515228,
+      "learning_rate": 8.791445848620737e-05,
+      "loss": 3.7428,
+      "step": 14240
+    },
+    {
+      "epoch": 0.9678624813153961,
+      "grad_norm": 0.3069162666797638,
+      "learning_rate": 8.79102119853241e-05,
+      "loss": 3.9151,
+      "step": 14245
+    },
+    {
+      "epoch": 0.9682022013860578,
+      "grad_norm": 0.18564817309379578,
+      "learning_rate": 8.790596548444083e-05,
+      "loss": 3.9755,
+      "step": 14250
+    },
+    {
+      "epoch": 0.9685419214567197,
+      "grad_norm": 0.26427149772644043,
+      "learning_rate": 8.790171898355756e-05,
+      "loss": 4.0877,
+      "step": 14255
+    },
+    {
+      "epoch": 0.9688816415273814,
+      "grad_norm": 0.17514817416667938,
+      "learning_rate": 8.789747248267428e-05,
+      "loss": 3.9761,
+      "step": 14260
+    },
+    {
+      "epoch": 0.9692213615980432,
+      "grad_norm": 0.17348946630954742,
+      "learning_rate": 8.7893225981791e-05,
+      "loss": 3.9038,
+      "step": 14265
+    },
+    {
+      "epoch": 0.969561081668705,
+      "grad_norm": 0.20074446499347687,
+      "learning_rate": 8.788897948090774e-05,
+      "loss": 3.9772,
+      "step": 14270
+    },
+    {
+      "epoch": 0.9699008017393668,
+      "grad_norm": 0.24490071833133698,
+      "learning_rate": 8.788473298002447e-05,
+      "loss": 4.1249,
+      "step": 14275
+    },
+    {
+      "epoch": 0.9702405218100285,
+      "grad_norm": 0.2058936506509781,
+      "learning_rate": 8.788048647914118e-05,
+      "loss": 3.8624,
+      "step": 14280
+    },
+    {
+      "epoch": 0.9705802418806904,
+      "grad_norm": 0.16198603808879852,
+      "learning_rate": 8.787623997825792e-05,
+      "loss": 3.9449,
+      "step": 14285
+    },
+    {
+      "epoch": 0.9709199619513521,
+      "grad_norm": 0.15939363837242126,
+      "learning_rate": 8.787199347737465e-05,
+      "loss": 3.9548,
+      "step": 14290
+    },
+    {
+      "epoch": 0.9712596820220138,
+      "grad_norm": 0.15633496642112732,
+      "learning_rate": 8.786774697649137e-05,
+      "loss": 3.8646,
+      "step": 14295
+    },
+    {
+      "epoch": 0.9715994020926756,
+      "grad_norm": 0.18158230185508728,
+      "learning_rate": 8.786350047560811e-05,
+      "loss": 4.0045,
+      "step": 14300
+    },
+    {
+      "epoch": 0.9719391221633374,
+      "grad_norm": 0.2111271321773529,
+      "learning_rate": 8.785925397472484e-05,
+      "loss": 4.1019,
+      "step": 14305
+    },
+    {
+      "epoch": 0.9722788422339992,
+      "grad_norm": 0.19582590460777283,
+      "learning_rate": 8.785500747384155e-05,
+      "loss": 4.0289,
+      "step": 14310
+    },
+    {
+      "epoch": 0.9726185623046609,
+      "grad_norm": 0.18569405376911163,
+      "learning_rate": 8.785076097295829e-05,
+      "loss": 3.8372,
+      "step": 14315
+    },
+    {
+      "epoch": 0.9729582823753228,
+      "grad_norm": 0.1685798466205597,
+      "learning_rate": 8.784651447207502e-05,
+      "loss": 3.9108,
+      "step": 14320
+    },
+    {
+      "epoch": 0.9732980024459845,
+      "grad_norm": 0.236485555768013,
+      "learning_rate": 8.784226797119174e-05,
+      "loss": 3.6307,
+      "step": 14325
+    },
+    {
+      "epoch": 0.9736377225166463,
+      "grad_norm": 0.17849500477313995,
+      "learning_rate": 8.783802147030848e-05,
+      "loss": 4.0166,
+      "step": 14330
+    },
+    {
+      "epoch": 0.973977442587308,
+      "grad_norm": 0.15316098928451538,
+      "learning_rate": 8.783377496942519e-05,
+      "loss": 4.0223,
+      "step": 14335
+    },
+    {
+      "epoch": 0.9743171626579699,
+      "grad_norm": 0.14150911569595337,
+      "learning_rate": 8.782952846854192e-05,
+      "loss": 3.9861,
+      "step": 14340
+    },
+    {
+      "epoch": 0.9746568827286316,
+      "grad_norm": 0.15742090344429016,
+      "learning_rate": 8.782528196765866e-05,
+      "loss": 4.0853,
+      "step": 14345
+    },
+    {
+      "epoch": 0.9749966027992933,
+      "grad_norm": 0.1783253401517868,
+      "learning_rate": 8.782103546677538e-05,
+      "loss": 3.8919,
+      "step": 14350
+    },
+    {
+      "epoch": 0.9753363228699552,
+      "grad_norm": 0.16757941246032715,
+      "learning_rate": 8.78167889658921e-05,
+      "loss": 3.5916,
+      "step": 14355
+    },
+    {
+      "epoch": 0.9756760429406169,
+      "grad_norm": 0.25573256611824036,
+      "learning_rate": 8.781254246500884e-05,
+      "loss": 3.9848,
+      "step": 14360
+    },
+    {
+      "epoch": 0.9760157630112787,
+      "grad_norm": 0.1471344381570816,
+      "learning_rate": 8.780829596412556e-05,
+      "loss": 3.9848,
+      "step": 14365
+    },
+    {
+      "epoch": 0.9763554830819405,
+      "grad_norm": 0.17102879285812378,
+      "learning_rate": 8.780404946324229e-05,
+      "loss": 3.9778,
+      "step": 14370
+    },
+    {
+      "epoch": 0.9766952031526023,
+      "grad_norm": 0.21928320825099945,
+      "learning_rate": 8.779980296235903e-05,
+      "loss": 3.7909,
+      "step": 14375
+    },
+    {
+      "epoch": 0.977034923223264,
+      "grad_norm": 0.1851445883512497,
+      "learning_rate": 8.779555646147574e-05,
+      "loss": 4.0371,
+      "step": 14380
+    },
+    {
+      "epoch": 0.9773746432939258,
+      "grad_norm": 0.15043723583221436,
+      "learning_rate": 8.779130996059247e-05,
+      "loss": 3.7731,
+      "step": 14385
+    },
+    {
+      "epoch": 0.9777143633645876,
+      "grad_norm": 0.19180312752723694,
+      "learning_rate": 8.778706345970921e-05,
+      "loss": 3.8016,
+      "step": 14390
+    },
+    {
+      "epoch": 0.9780540834352494,
+      "grad_norm": 0.18517717719078064,
+      "learning_rate": 8.778281695882593e-05,
+      "loss": 4.0766,
+      "step": 14395
+    },
+    {
+      "epoch": 0.9783938035059111,
+      "grad_norm": 0.2040787637233734,
+      "learning_rate": 8.777857045794266e-05,
+      "loss": 3.9668,
+      "step": 14400
+    },
+    {
+      "epoch": 0.978733523576573,
+      "grad_norm": 0.17522788047790527,
+      "learning_rate": 8.777432395705938e-05,
+      "loss": 3.7057,
+      "step": 14405
+    },
+    {
+      "epoch": 0.9790732436472347,
+      "grad_norm": 0.258577823638916,
+      "learning_rate": 8.777007745617611e-05,
+      "loss": 3.8998,
+      "step": 14410
+    },
+    {
+      "epoch": 0.9794129637178964,
+      "grad_norm": 0.16101132333278656,
+      "learning_rate": 8.776583095529284e-05,
+      "loss": 4.024,
+      "step": 14415
+    },
+    {
+      "epoch": 0.9797526837885582,
+      "grad_norm": 0.23143424093723297,
+      "learning_rate": 8.776158445440957e-05,
+      "loss": 3.876,
+      "step": 14420
+    },
+    {
+      "epoch": 0.98009240385922,
+      "grad_norm": 0.5347188711166382,
+      "learning_rate": 8.77573379535263e-05,
+      "loss": 3.721,
+      "step": 14425
+    },
+    {
+      "epoch": 0.9804321239298818,
+      "grad_norm": 3.5879626274108887,
+      "learning_rate": 8.775309145264302e-05,
+      "loss": 3.9406,
+      "step": 14430
+    },
+    {
+      "epoch": 0.9807718440005435,
+      "grad_norm": 0.48771944642066956,
+      "learning_rate": 8.774884495175975e-05,
+      "loss": 3.8004,
+      "step": 14435
+    },
+    {
+      "epoch": 0.9811115640712054,
+      "grad_norm": 0.1710396111011505,
+      "learning_rate": 8.774459845087648e-05,
+      "loss": 3.9633,
+      "step": 14440
+    },
+    {
+      "epoch": 0.9814512841418671,
+      "grad_norm": 0.16899172961711884,
+      "learning_rate": 8.774035194999321e-05,
+      "loss": 4.0779,
+      "step": 14445
+    },
+    {
+      "epoch": 0.9817910042125289,
+      "grad_norm": 0.18617889285087585,
+      "learning_rate": 8.773610544910994e-05,
+      "loss": 3.8466,
+      "step": 14450
+    },
+    {
+      "epoch": 0.9821307242831907,
+      "grad_norm": 0.13299688696861267,
+      "learning_rate": 8.773185894822666e-05,
+      "loss": 4.0135,
+      "step": 14455
+    },
+    {
+      "epoch": 0.9824704443538524,
+      "grad_norm": 0.20678380131721497,
+      "learning_rate": 8.772761244734339e-05,
+      "loss": 3.8453,
+      "step": 14460
+    },
+    {
+      "epoch": 0.9828101644245142,
+      "grad_norm": 0.14222410321235657,
+      "learning_rate": 8.772336594646012e-05,
+      "loss": 3.9648,
+      "step": 14465
+    },
+    {
+      "epoch": 0.9831498844951759,
+      "grad_norm": 0.21488900482654572,
+      "learning_rate": 8.771911944557685e-05,
+      "loss": 3.9613,
+      "step": 14470
+    },
+    {
+      "epoch": 0.9834896045658378,
+      "grad_norm": 0.2058933824300766,
+      "learning_rate": 8.771487294469358e-05,
+      "loss": 3.7448,
+      "step": 14475
+    },
+    {
+      "epoch": 0.9838293246364995,
+      "grad_norm": 0.25969386100769043,
+      "learning_rate": 8.77106264438103e-05,
+      "loss": 3.8523,
+      "step": 14480
+    },
+    {
+      "epoch": 0.9841690447071613,
+      "grad_norm": 0.17804904282093048,
+      "learning_rate": 8.770637994292703e-05,
+      "loss": 4.0136,
+      "step": 14485
+    },
+    {
+      "epoch": 0.9845087647778231,
+      "grad_norm": 0.17092011868953705,
+      "learning_rate": 8.770213344204376e-05,
+      "loss": 3.9277,
+      "step": 14490
+    },
+    {
+      "epoch": 0.9848484848484849,
+      "grad_norm": 0.26749852299690247,
+      "learning_rate": 8.769788694116049e-05,
+      "loss": 4.1337,
+      "step": 14495
+    },
+    {
+      "epoch": 0.9851882049191466,
+      "grad_norm": 1.6659806966781616,
+      "learning_rate": 8.769364044027722e-05,
+      "loss": 3.9058,
+      "step": 14500
+    },
+    {
+      "epoch": 0.9855279249898083,
+      "grad_norm": 0.16872169077396393,
+      "learning_rate": 8.768939393939394e-05,
+      "loss": 3.9551,
+      "step": 14505
+    },
+    {
+      "epoch": 0.9858676450604702,
+      "grad_norm": 0.25871741771698,
+      "learning_rate": 8.768514743851067e-05,
+      "loss": 3.9576,
+      "step": 14510
+    },
+    {
+      "epoch": 0.9862073651311319,
+      "grad_norm": 0.2599276006221771,
+      "learning_rate": 8.76809009376274e-05,
+      "loss": 4.0486,
+      "step": 14515
+    },
+    {
+      "epoch": 0.9865470852017937,
+      "grad_norm": 0.17734144628047943,
+      "learning_rate": 8.767665443674413e-05,
+      "loss": 3.9271,
+      "step": 14520
+    },
+    {
+      "epoch": 0.9868868052724555,
+      "grad_norm": 0.15988993644714355,
+      "learning_rate": 8.767240793586086e-05,
+      "loss": 3.7398,
+      "step": 14525
+    },
+    {
+      "epoch": 0.9872265253431173,
+      "grad_norm": 0.3864258825778961,
+      "learning_rate": 8.766816143497758e-05,
+      "loss": 4.1109,
+      "step": 14530
+    },
+    {
+      "epoch": 0.987566245413779,
+      "grad_norm": 0.1917543262243271,
+      "learning_rate": 8.766391493409431e-05,
+      "loss": 3.9459,
+      "step": 14535
+    },
+    {
+      "epoch": 0.9879059654844409,
+      "grad_norm": 0.2340976893901825,
+      "learning_rate": 8.765966843321104e-05,
+      "loss": 3.9126,
+      "step": 14540
+    },
+    {
+      "epoch": 0.9882456855551026,
+      "grad_norm": 0.1463710069656372,
+      "learning_rate": 8.765542193232777e-05,
+      "loss": 3.8255,
+      "step": 14545
+    },
+    {
+      "epoch": 0.9885854056257644,
+      "grad_norm": 0.20803597569465637,
+      "learning_rate": 8.76511754314445e-05,
+      "loss": 3.9285,
+      "step": 14550
+    },
+    {
+      "epoch": 0.9889251256964261,
+      "grad_norm": 0.23730699717998505,
+      "learning_rate": 8.764692893056122e-05,
+      "loss": 3.9045,
+      "step": 14555
+    },
+    {
+      "epoch": 0.989264845767088,
+      "grad_norm": 0.17138421535491943,
+      "learning_rate": 8.764268242967795e-05,
+      "loss": 3.8542,
+      "step": 14560
+    },
+    {
+      "epoch": 0.9896045658377497,
+      "grad_norm": 0.16893406212329865,
+      "learning_rate": 8.763843592879468e-05,
+      "loss": 4.0535,
+      "step": 14565
+    },
+    {
+      "epoch": 0.9899442859084114,
+      "grad_norm": 0.13037355244159698,
+      "learning_rate": 8.763418942791141e-05,
+      "loss": 3.7079,
+      "step": 14570
+    },
+    {
+      "epoch": 0.9902840059790733,
+      "grad_norm": 0.1940174400806427,
+      "learning_rate": 8.762994292702814e-05,
+      "loss": 3.9701,
+      "step": 14575
+    },
+    {
+      "epoch": 0.990623726049735,
+      "grad_norm": 0.19718214869499207,
+      "learning_rate": 8.762569642614486e-05,
+      "loss": 3.8213,
+      "step": 14580
+    },
+    {
+      "epoch": 0.9909634461203968,
+      "grad_norm": 0.6371154189109802,
+      "learning_rate": 8.762144992526159e-05,
+      "loss": 3.8541,
+      "step": 14585
+    },
+    {
+      "epoch": 0.9913031661910585,
+      "grad_norm": 0.15423880517482758,
+      "learning_rate": 8.761720342437832e-05,
+      "loss": 4.0511,
+      "step": 14590
+    },
+    {
+      "epoch": 0.9916428862617204,
+      "grad_norm": 0.1736377775669098,
+      "learning_rate": 8.761295692349505e-05,
+      "loss": 3.8275,
+      "step": 14595
+    },
+    {
+      "epoch": 0.9919826063323821,
+      "grad_norm": 0.19258588552474976,
+      "learning_rate": 8.760871042261178e-05,
+      "loss": 4.1035,
+      "step": 14600
+    },
+    {
+      "epoch": 0.9923223264030439,
+      "grad_norm": 0.43297508358955383,
+      "learning_rate": 8.760446392172849e-05,
+      "loss": 4.0978,
+      "step": 14605
+    },
+    {
+      "epoch": 0.9926620464737057,
+      "grad_norm": 0.20580710470676422,
+      "learning_rate": 8.760021742084523e-05,
+      "loss": 3.8437,
+      "step": 14610
+    },
+    {
+      "epoch": 0.9930017665443674,
+      "grad_norm": 1.218690037727356,
+      "learning_rate": 8.759597091996196e-05,
+      "loss": 4.0635,
+      "step": 14615
+    },
+    {
+      "epoch": 0.9933414866150292,
+      "grad_norm": 0.1623927801847458,
+      "learning_rate": 8.759172441907867e-05,
+      "loss": 3.8147,
+      "step": 14620
+    },
+    {
+      "epoch": 0.993681206685691,
+      "grad_norm": 0.17431679368019104,
+      "learning_rate": 8.758747791819542e-05,
+      "loss": 3.9406,
+      "step": 14625
+    },
+    {
+      "epoch": 0.9940209267563528,
+      "grad_norm": 0.18215017020702362,
+      "learning_rate": 8.758323141731214e-05,
+      "loss": 4.1441,
+      "step": 14630
+    },
+    {
+      "epoch": 0.9943606468270145,
+      "grad_norm": 0.20073819160461426,
+      "learning_rate": 8.757898491642886e-05,
+      "loss": 3.9045,
+      "step": 14635
+    },
+    {
+      "epoch": 0.9947003668976763,
+      "grad_norm": 0.17208696901798248,
+      "learning_rate": 8.75747384155456e-05,
+      "loss": 4.223,
+      "step": 14640
+    },
+    {
+      "epoch": 0.9950400869683381,
+      "grad_norm": 0.18610428273677826,
+      "learning_rate": 8.757049191466233e-05,
+      "loss": 3.9652,
+      "step": 14645
+    },
+    {
+      "epoch": 0.9953798070389999,
+      "grad_norm": 0.2461862713098526,
+      "learning_rate": 8.756624541377904e-05,
+      "loss": 3.8623,
+      "step": 14650
+    },
+    {
+      "epoch": 0.9957195271096616,
+      "grad_norm": 0.37638500332832336,
+      "learning_rate": 8.756199891289578e-05,
+      "loss": 3.896,
+      "step": 14655
+    },
+    {
+      "epoch": 0.9960592471803235,
+      "grad_norm": 0.2814083993434906,
+      "learning_rate": 8.755775241201251e-05,
+      "loss": 3.8634,
+      "step": 14660
+    },
+    {
+      "epoch": 0.9963989672509852,
+      "grad_norm": 0.2636764645576477,
+      "learning_rate": 8.755350591112923e-05,
+      "loss": 4.1024,
+      "step": 14665
+    },
+    {
+      "epoch": 0.9967386873216469,
+      "grad_norm": 0.1730240136384964,
+      "learning_rate": 8.754925941024597e-05,
+      "loss": 4.0616,
+      "step": 14670
+    },
+    {
+      "epoch": 0.9970784073923087,
+      "grad_norm": 0.17876474559307098,
+      "learning_rate": 8.75450129093627e-05,
+      "loss": 4.0613,
+      "step": 14675
+    },
+    {
+      "epoch": 0.9974181274629705,
+      "grad_norm": 0.1803043931722641,
+      "learning_rate": 8.754076640847941e-05,
+      "loss": 4.0982,
+      "step": 14680
+    },
+    {
+      "epoch": 0.9977578475336323,
+      "grad_norm": 0.2912727892398834,
+      "learning_rate": 8.753651990759615e-05,
+      "loss": 3.8865,
+      "step": 14685
+    },
+    {
+      "epoch": 0.998097567604294,
+      "grad_norm": 0.21436181664466858,
+      "learning_rate": 8.753227340671287e-05,
+      "loss": 4.0897,
+      "step": 14690
+    },
+    {
+      "epoch": 0.9984372876749559,
+      "grad_norm": 0.1772686094045639,
+      "learning_rate": 8.75280269058296e-05,
+      "loss": 4.1523,
+      "step": 14695
+    },
+    {
+      "epoch": 0.9987770077456176,
+      "grad_norm": 0.269540935754776,
+      "learning_rate": 8.752378040494634e-05,
+      "loss": 3.9891,
+      "step": 14700
+    },
+    {
+      "epoch": 0.9991167278162794,
+      "grad_norm": 0.20457632839679718,
+      "learning_rate": 8.751953390406305e-05,
+      "loss": 3.8832,
+      "step": 14705
+    },
+    {
+      "epoch": 0.9994564478869412,
+      "grad_norm": 0.2145458459854126,
+      "learning_rate": 8.751528740317978e-05,
+      "loss": 4.0273,
+      "step": 14710
+    },
+    {
+      "epoch": 0.999796167957603,
+      "grad_norm": 0.18700024485588074,
+      "learning_rate": 8.751104090229652e-05,
+      "loss": 4.1366,
+      "step": 14715
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": {
+        "f1": 0.8525333878305233,
+        "precision": 0.8753549892468697,
+        "recall": 0.8312223081395127
+      },
+      "eval_bleu_4": 0.0021924919669198163,
+      "eval_exact_match": 0.0,
+      "eval_loss": 3.751943349838257,
+      "eval_meteor": 0.07837825616589945,
+      "eval_rouge": {
+        "rouge1": 0.12766942289604644,
+        "rouge2": 0.01466477971693661,
+        "rougeL": 0.11110403147048781,
+        "rougeLsum": 0.11113999503571864
+      },
+      "eval_runtime": 404.0965,
+      "eval_samples_per_second": 25.536,
+      "eval_steps_per_second": 3.192,
+      "step": 14718
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 117744,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.174476031013683e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}