diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12026 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 8564,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0005838393274170948,
+      "grad_norm": 2.8441042445199733,
+      "learning_rate": 5.827505827505828e-07,
+      "loss": 1.1342,
+      "step": 5
+    },
+    {
+      "epoch": 0.0011676786548341896,
+      "grad_norm": 2.6915962544965275,
+      "learning_rate": 1.1655011655011657e-06,
+      "loss": 1.1191,
+      "step": 10
+    },
+    {
+      "epoch": 0.0017515179822512844,
+      "grad_norm": 2.493797679678999,
+      "learning_rate": 1.7482517482517485e-06,
+      "loss": 1.1026,
+      "step": 15
+    },
+    {
+      "epoch": 0.002335357309668379,
+      "grad_norm": 1.9460529718731712,
+      "learning_rate": 2.3310023310023313e-06,
+      "loss": 1.0839,
+      "step": 20
+    },
+    {
+      "epoch": 0.002919196637085474,
+      "grad_norm": 1.3422431894949707,
+      "learning_rate": 2.9137529137529138e-06,
+      "loss": 1.0331,
+      "step": 25
+    },
+    {
+      "epoch": 0.003503035964502569,
+      "grad_norm": 1.3292095769200771,
+      "learning_rate": 3.496503496503497e-06,
+      "loss": 1.0141,
+      "step": 30
+    },
+    {
+      "epoch": 0.004086875291919664,
+      "grad_norm": 1.0501059257033536,
+      "learning_rate": 4.079254079254079e-06,
+      "loss": 0.9596,
+      "step": 35
+    },
+    {
+      "epoch": 0.004670714619336758,
+      "grad_norm": 1.0026358935033322,
+      "learning_rate": 4.662004662004663e-06,
+      "loss": 0.9244,
+      "step": 40
+    },
+    {
+      "epoch": 0.005254553946753854,
+      "grad_norm": 0.8828397424182884,
+      "learning_rate": 5.244755244755245e-06,
+      "loss": 0.9071,
+      "step": 45
+    },
+    {
+      "epoch": 0.005838393274170948,
+      "grad_norm": 0.892428474434813,
+      "learning_rate": 5.8275058275058275e-06,
+      "loss": 0.8796,
+      "step": 50
+    },
+    {
+      "epoch": 0.006422232601588043,
+      "grad_norm": 0.8645422537859606,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 0.8939,
+      "step": 55
+    },
+    {
+      "epoch": 0.007006071929005138,
+      "grad_norm": 0.9772965887821152,
+      "learning_rate": 6.993006993006994e-06,
+      "loss": 0.8939,
+      "step": 60
+    },
+    {
+      "epoch": 0.007589911256422233,
+      "grad_norm": 0.8342203095089143,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.8636,
+      "step": 65
+    },
+    {
+      "epoch": 0.008173750583839328,
+      "grad_norm": 0.838314283043167,
+      "learning_rate": 8.158508158508159e-06,
+      "loss": 0.8683,
+      "step": 70
+    },
+    {
+      "epoch": 0.008757589911256422,
+      "grad_norm": 0.8531192844964881,
+      "learning_rate": 8.741258741258741e-06,
+      "loss": 0.8741,
+      "step": 75
+    },
+    {
+      "epoch": 0.009341429238673517,
+      "grad_norm": 0.8354241808923245,
+      "learning_rate": 9.324009324009325e-06,
+      "loss": 0.8491,
+      "step": 80
+    },
+    {
+      "epoch": 0.009925268566090611,
+      "grad_norm": 0.8603802781167157,
+      "learning_rate": 9.906759906759908e-06,
+      "loss": 0.8469,
+      "step": 85
+    },
+    {
+      "epoch": 0.010509107893507707,
+      "grad_norm": 0.9094770814499542,
+      "learning_rate": 1.048951048951049e-05,
+      "loss": 0.8456,
+      "step": 90
+    },
+    {
+      "epoch": 0.011092947220924802,
+      "grad_norm": 0.8710487600121014,
+      "learning_rate": 1.1072261072261073e-05,
+      "loss": 0.8516,
+      "step": 95
+    },
+    {
+      "epoch": 0.011676786548341896,
+      "grad_norm": 0.9222268142570368,
+      "learning_rate": 1.1655011655011655e-05,
+      "loss": 0.8237,
+      "step": 100
+    },
+    {
+      "epoch": 0.01226062587575899,
+      "grad_norm": 0.9176565894712276,
+      "learning_rate": 1.2237762237762239e-05,
+      "loss": 0.8356,
+      "step": 105
+    },
+    {
+      "epoch": 0.012844465203176086,
+      "grad_norm": 0.8542848432508544,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 0.8268,
+      "step": 110
+    },
+    {
+      "epoch": 0.01342830453059318,
+      "grad_norm": 0.8547012949765272,
+      "learning_rate": 1.3403263403263406e-05,
+      "loss": 0.8179,
+      "step": 115
+    },
+    {
+      "epoch": 0.014012143858010275,
+      "grad_norm": 0.8927678257049521,
+      "learning_rate": 1.3986013986013988e-05,
+      "loss": 0.8261,
+      "step": 120
+    },
+    {
+      "epoch": 0.014595983185427371,
+      "grad_norm": 0.8771767679168678,
+      "learning_rate": 1.456876456876457e-05,
+      "loss": 0.8156,
+      "step": 125
+    },
+    {
+      "epoch": 0.015179822512844466,
+      "grad_norm": 0.9346761034876638,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.8156,
+      "step": 130
+    },
+    {
+      "epoch": 0.01576366184026156,
+      "grad_norm": 0.9967648383739597,
+      "learning_rate": 1.5734265734265734e-05,
+      "loss": 0.8236,
+      "step": 135
+    },
+    {
+      "epoch": 0.016347501167678656,
+      "grad_norm": 0.950604341293268,
+      "learning_rate": 1.6317016317016318e-05,
+      "loss": 0.8011,
+      "step": 140
+    },
+    {
+      "epoch": 0.01693134049509575,
+      "grad_norm": 0.8454797159633483,
+      "learning_rate": 1.68997668997669e-05,
+      "loss": 0.8076,
+      "step": 145
+    },
+    {
+      "epoch": 0.017515179822512845,
+      "grad_norm": 0.9570558077264374,
+      "learning_rate": 1.7482517482517483e-05,
+      "loss": 0.7957,
+      "step": 150
+    },
+    {
+      "epoch": 0.01809901914992994,
+      "grad_norm": 0.9239857096216728,
+      "learning_rate": 1.8065268065268067e-05,
+      "loss": 0.8054,
+      "step": 155
+    },
+    {
+      "epoch": 0.018682858477347034,
+      "grad_norm": 0.9907328060405487,
+      "learning_rate": 1.864801864801865e-05,
+      "loss": 0.8123,
+      "step": 160
+    },
+    {
+      "epoch": 0.01926669780476413,
+      "grad_norm": 0.9282534846828877,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.8131,
+      "step": 165
+    },
+    {
+      "epoch": 0.019850537132181222,
+      "grad_norm": 0.8627148393669002,
+      "learning_rate": 1.9813519813519816e-05,
+      "loss": 0.7845,
+      "step": 170
+    },
+    {
+      "epoch": 0.02043437645959832,
+      "grad_norm": 0.990414565208542,
+      "learning_rate": 2.0396270396270396e-05,
+      "loss": 0.8354,
+      "step": 175
+    },
+    {
+      "epoch": 0.021018215787015414,
+      "grad_norm": 1.0141975698256604,
+      "learning_rate": 2.097902097902098e-05,
+      "loss": 0.8142,
+      "step": 180
+    },
+    {
+      "epoch": 0.021602055114432507,
+      "grad_norm": 0.9843786330547074,
+      "learning_rate": 2.156177156177156e-05,
+      "loss": 0.8031,
+      "step": 185
+    },
+    {
+      "epoch": 0.022185894441849603,
+      "grad_norm": 0.934570429819127,
+      "learning_rate": 2.2144522144522145e-05,
+      "loss": 0.8138,
+      "step": 190
+    },
+    {
+      "epoch": 0.0227697337692667,
+      "grad_norm": 1.0220562837350744,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.7814,
+      "step": 195
+    },
+    {
+      "epoch": 0.023353573096683792,
+      "grad_norm": 0.961440804678633,
+      "learning_rate": 2.331002331002331e-05,
+      "loss": 0.7928,
+      "step": 200
+    },
+    {
+      "epoch": 0.023937412424100888,
+      "grad_norm": 0.94391785220549,
+      "learning_rate": 2.3892773892773894e-05,
+      "loss": 0.8014,
+      "step": 205
+    },
+    {
+      "epoch": 0.02452125175151798,
+      "grad_norm": 0.9664977176711759,
+      "learning_rate": 2.4475524475524478e-05,
+      "loss": 0.8239,
+      "step": 210
+    },
+    {
+      "epoch": 0.025105091078935077,
+      "grad_norm": 1.0561418913054852,
+      "learning_rate": 2.505827505827506e-05,
+      "loss": 0.7741,
+      "step": 215
+    },
+    {
+      "epoch": 0.025688930406352173,
+      "grad_norm": 0.9358379347102888,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 0.7983,
+      "step": 220
+    },
+    {
+      "epoch": 0.026272769733769265,
+      "grad_norm": 0.8539282155631721,
+      "learning_rate": 2.6223776223776224e-05,
+      "loss": 0.793,
+      "step": 225
+    },
+    {
+      "epoch": 0.02685660906118636,
+      "grad_norm": 0.8554815403018589,
+      "learning_rate": 2.680652680652681e-05,
+      "loss": 0.8046,
+      "step": 230
+    },
+    {
+      "epoch": 0.027440448388603458,
+      "grad_norm": 1.0012026939383618,
+      "learning_rate": 2.738927738927739e-05,
+      "loss": 0.7918,
+      "step": 235
+    },
+    {
+      "epoch": 0.02802428771602055,
+      "grad_norm": 0.9302399659722874,
+      "learning_rate": 2.7972027972027976e-05,
+      "loss": 0.7666,
+      "step": 240
+    },
+    {
+      "epoch": 0.028608127043437646,
+      "grad_norm": 0.9745631817754927,
+      "learning_rate": 2.8554778554778557e-05,
+      "loss": 0.7945,
+      "step": 245
+    },
+    {
+      "epoch": 0.029191966370854742,
+      "grad_norm": 0.9663155411494311,
+      "learning_rate": 2.913752913752914e-05,
+      "loss": 0.8208,
+      "step": 250
+    },
+    {
+      "epoch": 0.029775805698271835,
+      "grad_norm": 0.9665096896834287,
+      "learning_rate": 2.972027972027972e-05,
+      "loss": 0.8155,
+      "step": 255
+    },
+    {
+      "epoch": 0.03035964502568893,
+      "grad_norm": 0.9211131832233894,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.804,
+      "step": 260
+    },
+    {
+      "epoch": 0.030943484353106024,
+      "grad_norm": 0.9278894919237144,
+      "learning_rate": 3.088578088578088e-05,
+      "loss": 0.7997,
+      "step": 265
+    },
+    {
+      "epoch": 0.03152732368052312,
+      "grad_norm": 0.9453801400501536,
+      "learning_rate": 3.146853146853147e-05,
+      "loss": 0.782,
+      "step": 270
+    },
+    {
+      "epoch": 0.032111163007940216,
+      "grad_norm": 1.0236007828019342,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.7889,
+      "step": 275
+    },
+    {
+      "epoch": 0.03269500233535731,
+      "grad_norm": 0.959267234083067,
+      "learning_rate": 3.2634032634032635e-05,
+      "loss": 0.7898,
+      "step": 280
+    },
+    {
+      "epoch": 0.0332788416627744,
+      "grad_norm": 1.1310249700433657,
+      "learning_rate": 3.321678321678322e-05,
+      "loss": 0.8123,
+      "step": 285
+    },
+    {
+      "epoch": 0.0338626809901915,
+      "grad_norm": 1.187081207110975,
+      "learning_rate": 3.37995337995338e-05,
+      "loss": 0.8001,
+      "step": 290
+    },
+    {
+      "epoch": 0.034446520317608594,
+      "grad_norm": 0.9811581650923609,
+      "learning_rate": 3.438228438228439e-05,
+      "loss": 0.7883,
+      "step": 295
+    },
+    {
+      "epoch": 0.03503035964502569,
+      "grad_norm": 0.9889952112158437,
+      "learning_rate": 3.4965034965034965e-05,
+      "loss": 0.7819,
+      "step": 300
+    },
+    {
+      "epoch": 0.035614198972442786,
+      "grad_norm": 0.9792726994248039,
+      "learning_rate": 3.554778554778555e-05,
+      "loss": 0.7908,
+      "step": 305
+    },
+    {
+      "epoch": 0.03619803829985988,
+      "grad_norm": 1.0330116552686042,
+      "learning_rate": 3.613053613053613e-05,
+      "loss": 0.8034,
+      "step": 310
+    },
+    {
+      "epoch": 0.03678187762727697,
+      "grad_norm": 0.9791641673387794,
+      "learning_rate": 3.671328671328672e-05,
+      "loss": 0.7838,
+      "step": 315
+    },
+    {
+      "epoch": 0.03736571695469407,
+      "grad_norm": 1.087198648692349,
+      "learning_rate": 3.72960372960373e-05,
+      "loss": 0.7891,
+      "step": 320
+    },
+    {
+      "epoch": 0.03794955628211116,
+      "grad_norm": 1.1453703170207792,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.7828,
+      "step": 325
+    },
+    {
+      "epoch": 0.03853339560952826,
+      "grad_norm": 1.1352672915900581,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.7944,
+      "step": 330
+    },
+    {
+      "epoch": 0.039117234936945355,
+      "grad_norm": 1.11091425012644,
+      "learning_rate": 3.904428904428905e-05,
+      "loss": 0.8031,
+      "step": 335
+    },
+    {
+      "epoch": 0.039701074264362445,
+      "grad_norm": 0.9944507467068269,
+      "learning_rate": 3.962703962703963e-05,
+      "loss": 0.7886,
+      "step": 340
+    },
+    {
+      "epoch": 0.04028491359177954,
+      "grad_norm": 1.00813220894139,
+      "learning_rate": 4.020979020979021e-05,
+      "loss": 0.7759,
+      "step": 345
+    },
+    {
+      "epoch": 0.04086875291919664,
+      "grad_norm": 1.0770061694184807,
+      "learning_rate": 4.079254079254079e-05,
+      "loss": 0.788,
+      "step": 350
+    },
+    {
+      "epoch": 0.04145259224661373,
+      "grad_norm": 0.9596462282334578,
+      "learning_rate": 4.1375291375291377e-05,
+      "loss": 0.7962,
+      "step": 355
+    },
+    {
+      "epoch": 0.04203643157403083,
+      "grad_norm": 1.1540120979481814,
+      "learning_rate": 4.195804195804196e-05,
+      "loss": 0.797,
+      "step": 360
+    },
+    {
+      "epoch": 0.04262027090144792,
+      "grad_norm": 1.080346740497132,
+      "learning_rate": 4.254079254079254e-05,
+      "loss": 0.7954,
+      "step": 365
+    },
+    {
+      "epoch": 0.043204110228865014,
+      "grad_norm": 1.0577009933461277,
+      "learning_rate": 4.312354312354312e-05,
+      "loss": 0.7877,
+      "step": 370
+    },
+    {
+      "epoch": 0.04378794955628211,
+      "grad_norm": 0.9154501100858539,
+      "learning_rate": 4.370629370629371e-05,
+      "loss": 0.7587,
+      "step": 375
+    },
+    {
+      "epoch": 0.044371788883699206,
+      "grad_norm": 1.0086154566043717,
+      "learning_rate": 4.428904428904429e-05,
+      "loss": 0.7761,
+      "step": 380
+    },
+    {
+      "epoch": 0.0449556282111163,
+      "grad_norm": 0.9201594510651068,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.7981,
+      "step": 385
+    },
+    {
+      "epoch": 0.0455394675385334,
+      "grad_norm": 0.9666106924921596,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.7832,
+      "step": 390
+    },
+    {
+      "epoch": 0.04612330686595049,
+      "grad_norm": 0.9815206957934799,
+      "learning_rate": 4.603729603729604e-05,
+      "loss": 0.7804,
+      "step": 395
+    },
+    {
+      "epoch": 0.046707146193367584,
+      "grad_norm": 1.0289413215392271,
+      "learning_rate": 4.662004662004662e-05,
+      "loss": 0.8053,
+      "step": 400
+    },
+    {
+      "epoch": 0.04729098552078468,
+      "grad_norm": 0.9565887758710183,
+      "learning_rate": 4.7202797202797204e-05,
+      "loss": 0.7905,
+      "step": 405
+    },
+    {
+      "epoch": 0.047874824848201776,
+      "grad_norm": 1.0540380731530687,
+      "learning_rate": 4.778554778554779e-05,
+      "loss": 0.8063,
+      "step": 410
+    },
+    {
+      "epoch": 0.04845866417561887,
+      "grad_norm": 1.0638685751627128,
+      "learning_rate": 4.836829836829837e-05,
+      "loss": 0.7806,
+      "step": 415
+    },
+    {
+      "epoch": 0.04904250350303596,
+      "grad_norm": 0.9719537263872465,
+      "learning_rate": 4.8951048951048956e-05,
+      "loss": 0.7731,
+      "step": 420
+    },
+    {
+      "epoch": 0.04962634283045306,
+      "grad_norm": 1.1429570378556555,
+      "learning_rate": 4.9533799533799534e-05,
+      "loss": 0.8072,
+      "step": 425
+    },
+    {
+      "epoch": 0.050210182157870153,
+      "grad_norm": 0.9863534850571196,
+      "learning_rate": 4.999999832221176e-05,
+      "loss": 0.7794,
+      "step": 430
+    },
+    {
+      "epoch": 0.05079402148528725,
+      "grad_norm": 0.9364682491937721,
+      "learning_rate": 4.999993959964937e-05,
+      "loss": 0.7781,
+      "step": 435
+    },
+    {
+      "epoch": 0.051377860812704346,
+      "grad_norm": 0.9768522284524611,
+      "learning_rate": 4.999979698792484e-05,
+      "loss": 0.7949,
+      "step": 440
+    },
+    {
+      "epoch": 0.05196170014012144,
+      "grad_norm": 1.2907678437139853,
+      "learning_rate": 4.999957048756989e-05,
+      "loss": 0.7946,
+      "step": 445
+    },
+    {
+      "epoch": 0.05254553946753853,
+      "grad_norm": 0.9378885425942916,
+      "learning_rate": 4.999926009942899e-05,
+      "loss": 0.7881,
+      "step": 450
+    },
+    {
+      "epoch": 0.05312937879495563,
+      "grad_norm": 0.8634203598226072,
+      "learning_rate": 4.999886582465941e-05,
+      "loss": 0.8047,
+      "step": 455
+    },
+    {
+      "epoch": 0.05371321812237272,
+      "grad_norm": 0.9278097068563849,
+      "learning_rate": 4.999838766473116e-05,
+      "loss": 0.7856,
+      "step": 460
+    },
+    {
+      "epoch": 0.05429705744978982,
+      "grad_norm": 0.8609084473770569,
+      "learning_rate": 4.999782562142702e-05,
+      "loss": 0.7634,
+      "step": 465
+    },
+    {
+      "epoch": 0.054880896777206915,
+      "grad_norm": 1.0123125316299562,
+      "learning_rate": 4.999717969684254e-05,
+      "loss": 0.7955,
+      "step": 470
+    },
+    {
+      "epoch": 0.055464736104624004,
+      "grad_norm": 1.0346559365063257,
+      "learning_rate": 4.999644989338598e-05,
+      "loss": 0.8073,
+      "step": 475
+    },
+    {
+      "epoch": 0.0560485754320411,
+      "grad_norm": 1.1426053676638168,
+      "learning_rate": 4.9995636213778354e-05,
+      "loss": 0.7719,
+      "step": 480
+    },
+    {
+      "epoch": 0.0566324147594582,
+      "grad_norm": 0.9150402070927298,
+      "learning_rate": 4.99947386610534e-05,
+      "loss": 0.7764,
+      "step": 485
+    },
+    {
+      "epoch": 0.05721625408687529,
+      "grad_norm": 0.987847326887783,
+      "learning_rate": 4.9993757238557564e-05,
+      "loss": 0.7913,
+      "step": 490
+    },
+    {
+      "epoch": 0.05780009341429239,
+      "grad_norm": 0.9689685701244068,
+      "learning_rate": 4.999269194995001e-05,
+      "loss": 0.7566,
+      "step": 495
+    },
+    {
+      "epoch": 0.058383932741709485,
+      "grad_norm": 0.9521343691976072,
+      "learning_rate": 4.9991542799202574e-05,
+      "loss": 0.7682,
+      "step": 500
+    },
+    {
+      "epoch": 0.058967772069126574,
+      "grad_norm": 0.8498214034434336,
+      "learning_rate": 4.999030979059977e-05,
+      "loss": 0.7898,
+      "step": 505
+    },
+    {
+      "epoch": 0.05955161139654367,
+      "grad_norm": 0.884610027782479,
+      "learning_rate": 4.998899292873876e-05,
+      "loss": 0.7933,
+      "step": 510
+    },
+    {
+      "epoch": 0.060135450723960766,
+      "grad_norm": 0.9009920893944248,
+      "learning_rate": 4.9987592218529364e-05,
+      "loss": 0.7807,
+      "step": 515
+    },
+    {
+      "epoch": 0.06071929005137786,
+      "grad_norm": 0.982774008971455,
+      "learning_rate": 4.998610766519401e-05,
+      "loss": 0.7755,
+      "step": 520
+    },
+    {
+      "epoch": 0.06130312937879496,
+      "grad_norm": 0.9363074199246613,
+      "learning_rate": 4.9984539274267726e-05,
+      "loss": 0.7678,
+      "step": 525
+    },
+    {
+      "epoch": 0.06188696870621205,
+      "grad_norm": 0.8818074337149633,
+      "learning_rate": 4.998288705159815e-05,
+      "loss": 0.7985,
+      "step": 530
+    },
+    {
+      "epoch": 0.062470808033629144,
+      "grad_norm": 0.9582202860157079,
+      "learning_rate": 4.9981151003345436e-05,
+      "loss": 0.7718,
+      "step": 535
+    },
+    {
+      "epoch": 0.06305464736104624,
+      "grad_norm": 0.9799283014400538,
+      "learning_rate": 4.99793311359823e-05,
+      "loss": 0.7965,
+      "step": 540
+    },
+    {
+      "epoch": 0.06363848668846334,
+      "grad_norm": 1.0277101299431106,
+      "learning_rate": 4.997742745629397e-05,
+      "loss": 0.7912,
+      "step": 545
+    },
+    {
+      "epoch": 0.06422232601588043,
+      "grad_norm": 0.9196693224733947,
+      "learning_rate": 4.997543997137816e-05,
+      "loss": 0.7742,
+      "step": 550
+    },
+    {
+      "epoch": 0.06480616534329753,
+      "grad_norm": 1.1756967119581367,
+      "learning_rate": 4.9973368688645034e-05,
+      "loss": 0.7904,
+      "step": 555
+    },
+    {
+      "epoch": 0.06539000467071462,
+      "grad_norm": 0.9727112642584299,
+      "learning_rate": 4.997121361581721e-05,
+      "loss": 0.7702,
+      "step": 560
+    },
+    {
+      "epoch": 0.06597384399813172,
+      "grad_norm": 0.9340751229066998,
+      "learning_rate": 4.9968974760929694e-05,
+      "loss": 0.7611,
+      "step": 565
+    },
+    {
+      "epoch": 0.0665576833255488,
+      "grad_norm": 0.8451161838818951,
+      "learning_rate": 4.996665213232987e-05,
+      "loss": 0.7394,
+      "step": 570
+    },
+    {
+      "epoch": 0.0671415226529659,
+      "grad_norm": 0.9437782768508981,
+      "learning_rate": 4.9964245738677465e-05,
+      "loss": 0.7721,
+      "step": 575
+    },
+    {
+      "epoch": 0.067725361980383,
+      "grad_norm": 0.8666049009060005,
+      "learning_rate": 4.996175558894452e-05,
+      "loss": 0.7649,
+      "step": 580
+    },
+    {
+      "epoch": 0.06830920130780009,
+      "grad_norm": 0.9088580093169708,
+      "learning_rate": 4.9959181692415345e-05,
+      "loss": 0.7895,
+      "step": 585
+    },
+    {
+      "epoch": 0.06889304063521719,
+      "grad_norm": 1.0295119037274578,
+      "learning_rate": 4.995652405868652e-05,
+      "loss": 0.7724,
+      "step": 590
+    },
+    {
+      "epoch": 0.06947687996263428,
+      "grad_norm": 0.9432589650717397,
+      "learning_rate": 4.99537826976668e-05,
+      "loss": 0.7707,
+      "step": 595
+    },
+    {
+      "epoch": 0.07006071929005138,
+      "grad_norm": 1.1052404077350295,
+      "learning_rate": 4.9950957619577115e-05,
+      "loss": 0.7931,
+      "step": 600
+    },
+    {
+      "epoch": 0.07064455861746848,
+      "grad_norm": 0.9419128366721217,
+      "learning_rate": 4.9948048834950546e-05,
+      "loss": 0.7881,
+      "step": 605
+    },
+    {
+      "epoch": 0.07122839794488557,
+      "grad_norm": 0.954093991240324,
+      "learning_rate": 4.9945056354632255e-05,
+      "loss": 0.7806,
+      "step": 610
+    },
+    {
+      "epoch": 0.07181223727230267,
+      "grad_norm": 1.0272959906124222,
+      "learning_rate": 4.994198018977945e-05,
+      "loss": 0.7791,
+      "step": 615
+    },
+    {
+      "epoch": 0.07239607659971976,
+      "grad_norm": 0.8921172892298924,
+      "learning_rate": 4.993882035186136e-05,
+      "loss": 0.7755,
+      "step": 620
+    },
+    {
+      "epoch": 0.07297991592713685,
+      "grad_norm": 0.9518130993302121,
+      "learning_rate": 4.9935576852659175e-05,
+      "loss": 0.7655,
+      "step": 625
+    },
+    {
+      "epoch": 0.07356375525455394,
+      "grad_norm": 0.8353315117685729,
+      "learning_rate": 4.993224970426603e-05,
+      "loss": 0.7547,
+      "step": 630
+    },
+    {
+      "epoch": 0.07414759458197104,
+      "grad_norm": 0.8586464703013177,
+      "learning_rate": 4.99288389190869e-05,
+      "loss": 0.7703,
+      "step": 635
+    },
+    {
+      "epoch": 0.07473143390938813,
+      "grad_norm": 0.889200186047205,
+      "learning_rate": 4.992534450983864e-05,
+      "loss": 0.784,
+      "step": 640
+    },
+    {
+      "epoch": 0.07531527323680523,
+      "grad_norm": 1.0890280918725501,
+      "learning_rate": 4.9921766489549835e-05,
+      "loss": 0.7829,
+      "step": 645
+    },
+    {
+      "epoch": 0.07589911256422233,
+      "grad_norm": 0.8261199193795216,
+      "learning_rate": 4.991810487156087e-05,
+      "loss": 0.7644,
+      "step": 650
+    },
+    {
+      "epoch": 0.07648295189163942,
+      "grad_norm": 0.9504620852001426,
+      "learning_rate": 4.991435966952376e-05,
+      "loss": 0.7865,
+      "step": 655
+    },
+    {
+      "epoch": 0.07706679121905652,
+      "grad_norm": 0.9843549810556219,
+      "learning_rate": 4.991053089740219e-05,
+      "loss": 0.773,
+      "step": 660
+    },
+    {
+      "epoch": 0.07765063054647361,
+      "grad_norm": 1.100665136082888,
+      "learning_rate": 4.990661856947142e-05,
+      "loss": 0.7745,
+      "step": 665
+    },
+    {
+      "epoch": 0.07823446987389071,
+      "grad_norm": 1.0689962692182553,
+      "learning_rate": 4.990262270031824e-05,
+      "loss": 0.749,
+      "step": 670
+    },
+    {
+      "epoch": 0.0788183092013078,
+      "grad_norm": 0.86450476503251,
+      "learning_rate": 4.989854330484092e-05,
+      "loss": 0.7923,
+      "step": 675
+    },
+    {
+      "epoch": 0.07940214852872489,
+      "grad_norm": 0.9313631425218568,
+      "learning_rate": 4.9894380398249135e-05,
+      "loss": 0.7851,
+      "step": 680
+    },
+    {
+      "epoch": 0.07998598785614199,
+      "grad_norm": 0.9422363860930384,
+      "learning_rate": 4.989013399606396e-05,
+      "loss": 0.7589,
+      "step": 685
+    },
+    {
+      "epoch": 0.08056982718355908,
+      "grad_norm": 0.9383975558454795,
+      "learning_rate": 4.988580411411774e-05,
+      "loss": 0.7577,
+      "step": 690
+    },
+    {
+      "epoch": 0.08115366651097618,
+      "grad_norm": 0.8678732287007085,
+      "learning_rate": 4.988139076855408e-05,
+      "loss": 0.7808,
+      "step": 695
+    },
+    {
+      "epoch": 0.08173750583839327,
+      "grad_norm": 0.9443261935414679,
+      "learning_rate": 4.9876893975827774e-05,
+      "loss": 0.7678,
+      "step": 700
+    },
+    {
+      "epoch": 0.08232134516581037,
+      "grad_norm": 0.893410559184795,
+      "learning_rate": 4.987231375270475e-05,
+      "loss": 0.7859,
+      "step": 705
+    },
+    {
+      "epoch": 0.08290518449322747,
+      "grad_norm": 0.908958440524122,
+      "learning_rate": 4.9867650116261994e-05,
+      "loss": 0.7507,
+      "step": 710
+    },
+    {
+      "epoch": 0.08348902382064456,
+      "grad_norm": 0.9226106440329441,
+      "learning_rate": 4.986290308388747e-05,
+      "loss": 0.7676,
+      "step": 715
+    },
+    {
+      "epoch": 0.08407286314806166,
+      "grad_norm": 0.8181830616959214,
+      "learning_rate": 4.98580726732801e-05,
+      "loss": 0.7931,
+      "step": 720
+    },
+    {
+      "epoch": 0.08465670247547875,
+      "grad_norm": 0.9202581836834515,
+      "learning_rate": 4.985315890244969e-05,
+      "loss": 0.748,
+      "step": 725
+    },
+    {
+      "epoch": 0.08524054180289584,
+      "grad_norm": 0.8637096331068913,
+      "learning_rate": 4.9848161789716804e-05,
+      "loss": 0.7902,
+      "step": 730
+    },
+    {
+      "epoch": 0.08582438113031293,
+      "grad_norm": 1.0251325008801968,
+      "learning_rate": 4.9843081353712765e-05,
+      "loss": 0.7838,
+      "step": 735
+    },
+    {
+      "epoch": 0.08640822045773003,
+      "grad_norm": 0.9799386743626911,
+      "learning_rate": 4.983791761337958e-05,
+      "loss": 0.7528,
+      "step": 740
+    },
+    {
+      "epoch": 0.08699205978514712,
+      "grad_norm": 1.1092058160482392,
+      "learning_rate": 4.9832670587969804e-05,
+      "loss": 0.7717,
+      "step": 745
+    },
+    {
+      "epoch": 0.08757589911256422,
+      "grad_norm": 1.0895863177487297,
+      "learning_rate": 4.9827340297046546e-05,
+      "loss": 0.747,
+      "step": 750
+    },
+    {
+      "epoch": 0.08815973843998132,
+      "grad_norm": 0.8899437101345212,
+      "learning_rate": 4.9821926760483354e-05,
+      "loss": 0.772,
+      "step": 755
+    },
+    {
+      "epoch": 0.08874357776739841,
+      "grad_norm": 1.0639534479618253,
+      "learning_rate": 4.9816429998464155e-05,
+      "loss": 0.7671,
+      "step": 760
+    },
+    {
+      "epoch": 0.08932741709481551,
+      "grad_norm": 0.9208650929263082,
+      "learning_rate": 4.9810850031483155e-05,
+      "loss": 0.7498,
+      "step": 765
+    },
+    {
+      "epoch": 0.0899112564222326,
+      "grad_norm": 1.1978688347410331,
+      "learning_rate": 4.9805186880344826e-05,
+      "loss": 0.7656,
+      "step": 770
+    },
+    {
+      "epoch": 0.0904950957496497,
+      "grad_norm": 0.8388720338093995,
+      "learning_rate": 4.9799440566163726e-05,
+      "loss": 0.7634,
+      "step": 775
+    },
+    {
+      "epoch": 0.0910789350770668,
+      "grad_norm": 1.0025093358414843,
+      "learning_rate": 4.979361111036454e-05,
+      "loss": 0.779,
+      "step": 780
+    },
+    {
+      "epoch": 0.09166277440448388,
+      "grad_norm": 0.8893526188128671,
+      "learning_rate": 4.9787698534681896e-05,
+      "loss": 0.784,
+      "step": 785
+    },
+    {
+      "epoch": 0.09224661373190098,
+      "grad_norm": 0.9052342021850083,
+      "learning_rate": 4.978170286116035e-05,
+      "loss": 0.7654,
+      "step": 790
+    },
+    {
+      "epoch": 0.09283045305931807,
+      "grad_norm": 0.7721847329666197,
+      "learning_rate": 4.9775624112154275e-05,
+      "loss": 0.7434,
+      "step": 795
+    },
+    {
+      "epoch": 0.09341429238673517,
+      "grad_norm": 0.9097917170134736,
+      "learning_rate": 4.976946231032777e-05,
+      "loss": 0.7375,
+      "step": 800
+    },
+    {
+      "epoch": 0.09399813171415226,
+      "grad_norm": 0.855353795632668,
+      "learning_rate": 4.976321747865462e-05,
+      "loss": 0.7678,
+      "step": 805
+    },
+    {
+      "epoch": 0.09458197104156936,
+      "grad_norm": 0.9001694450903074,
+      "learning_rate": 4.975688964041816e-05,
+      "loss": 0.7696,
+      "step": 810
+    },
+    {
+      "epoch": 0.09516581036898646,
+      "grad_norm": 0.9036123636475458,
+      "learning_rate": 4.975047881921119e-05,
+      "loss": 0.7382,
+      "step": 815
+    },
+    {
+      "epoch": 0.09574964969640355,
+      "grad_norm": 0.7946893689315172,
+      "learning_rate": 4.974398503893596e-05,
+      "loss": 0.7701,
+      "step": 820
+    },
+    {
+      "epoch": 0.09633348902382065,
+      "grad_norm": 0.8207619623724253,
+      "learning_rate": 4.973740832380397e-05,
+      "loss": 0.7788,
+      "step": 825
+    },
+    {
+      "epoch": 0.09691732835123774,
+      "grad_norm": 0.9272781944556916,
+      "learning_rate": 4.9730748698335954e-05,
+      "loss": 0.79,
+      "step": 830
+    },
+    {
+      "epoch": 0.09750116767865484,
+      "grad_norm": 0.9100307149491746,
+      "learning_rate": 4.9724006187361794e-05,
+      "loss": 0.7823,
+      "step": 835
+    },
+    {
+      "epoch": 0.09808500700607192,
+      "grad_norm": 0.8933455668978604,
+      "learning_rate": 4.971718081602037e-05,
+      "loss": 0.7968,
+      "step": 840
+    },
+    {
+      "epoch": 0.09866884633348902,
+      "grad_norm": 0.8676070117116492,
+      "learning_rate": 4.971027260975952e-05,
+      "loss": 0.763,
+      "step": 845
+    },
+    {
+      "epoch": 0.09925268566090611,
+      "grad_norm": 0.8909257634802544,
+      "learning_rate": 4.9703281594335904e-05,
+      "loss": 0.7597,
+      "step": 850
+    },
+    {
+      "epoch": 0.09983652498832321,
+      "grad_norm": 0.8735398370337358,
+      "learning_rate": 4.969620779581497e-05,
+      "loss": 0.7643,
+      "step": 855
+    },
+    {
+      "epoch": 0.10042036431574031,
+      "grad_norm": 1.0661381628959834,
+      "learning_rate": 4.968905124057077e-05,
+      "loss": 0.7495,
+      "step": 860
+    },
+    {
+      "epoch": 0.1010042036431574,
+      "grad_norm": 0.997806166540193,
+      "learning_rate": 4.968181195528594e-05,
+      "loss": 0.7488,
+      "step": 865
+    },
+    {
+      "epoch": 0.1015880429705745,
+      "grad_norm": 0.8212097797501837,
+      "learning_rate": 4.9674489966951545e-05,
+      "loss": 0.7656,
+      "step": 870
+    },
+    {
+      "epoch": 0.1021718822979916,
+      "grad_norm": 0.9256535225832584,
+      "learning_rate": 4.9667085302867015e-05,
+      "loss": 0.7499,
+      "step": 875
+    },
+    {
+      "epoch": 0.10275572162540869,
+      "grad_norm": 0.8960932163986034,
+      "learning_rate": 4.9659597990640045e-05,
+      "loss": 0.752,
+      "step": 880
+    },
+    {
+      "epoch": 0.10333956095282579,
+      "grad_norm": 0.9177441285582005,
+      "learning_rate": 4.9652028058186435e-05,
+      "loss": 0.7639,
+      "step": 885
+    },
+    {
+      "epoch": 0.10392340028024288,
+      "grad_norm": 0.9855352847812788,
+      "learning_rate": 4.9644375533730056e-05,
+      "loss": 0.7861,
+      "step": 890
+    },
+    {
+      "epoch": 0.10450723960765997,
+      "grad_norm": 1.1755603725509243,
+      "learning_rate": 4.963664044580272e-05,
+      "loss": 0.7644,
+      "step": 895
+    },
+    {
+      "epoch": 0.10509107893507706,
+      "grad_norm": 0.9704947639839278,
+      "learning_rate": 4.9628822823244056e-05,
+      "loss": 0.7457,
+      "step": 900
+    },
+    {
+      "epoch": 0.10567491826249416,
+      "grad_norm": 0.8135916392341626,
+      "learning_rate": 4.962092269520143e-05,
+      "loss": 0.743,
+      "step": 905
+    },
+    {
+      "epoch": 0.10625875758991125,
+      "grad_norm": 0.9718963175942106,
+      "learning_rate": 4.96129400911298e-05,
+      "loss": 0.7747,
+      "step": 910
+    },
+    {
+      "epoch": 0.10684259691732835,
+      "grad_norm": 0.8056530889300901,
+      "learning_rate": 4.960487504079166e-05,
+      "loss": 0.7468,
+      "step": 915
+    },
+    {
+      "epoch": 0.10742643624474545,
+      "grad_norm": 1.0253838704385734,
+      "learning_rate": 4.959672757425688e-05,
+      "loss": 0.7689,
+      "step": 920
+    },
+    {
+      "epoch": 0.10801027557216254,
+      "grad_norm": 0.8972042104044596,
+      "learning_rate": 4.958849772190261e-05,
+      "loss": 0.7634,
+      "step": 925
+    },
+    {
+      "epoch": 0.10859411489957964,
+      "grad_norm": 0.8252991480283688,
+      "learning_rate": 4.958018551441317e-05,
+      "loss": 0.7423,
+      "step": 930
+    },
+    {
+      "epoch": 0.10917795422699673,
+      "grad_norm": 1.0182109584106913,
+      "learning_rate": 4.957179098277994e-05,
+      "loss": 0.7679,
+      "step": 935
+    },
+    {
+      "epoch": 0.10976179355441383,
+      "grad_norm": 0.8857745022475216,
+      "learning_rate": 4.956331415830125e-05,
+      "loss": 0.7367,
+      "step": 940
+    },
+    {
+      "epoch": 0.11034563288183093,
+      "grad_norm": 0.9324065138620737,
+      "learning_rate": 4.955475507258222e-05,
+      "loss": 0.7713,
+      "step": 945
+    },
+    {
+      "epoch": 0.11092947220924801,
+      "grad_norm": 0.8898078164147581,
+      "learning_rate": 4.95461137575347e-05,
+      "loss": 0.7414,
+      "step": 950
+    },
+    {
+      "epoch": 0.1115133115366651,
+      "grad_norm": 0.9976876371835823,
+      "learning_rate": 4.953739024537712e-05,
+      "loss": 0.724,
+      "step": 955
+    },
+    {
+      "epoch": 0.1120971508640822,
+      "grad_norm": 0.7955416046502354,
+      "learning_rate": 4.952858456863437e-05,
+      "loss": 0.7772,
+      "step": 960
+    },
+    {
+      "epoch": 0.1126809901914993,
+      "grad_norm": 0.9382741081225117,
+      "learning_rate": 4.951969676013768e-05,
+      "loss": 0.7523,
+      "step": 965
+    },
+    {
+      "epoch": 0.1132648295189164,
+      "grad_norm": 0.903157971998188,
+      "learning_rate": 4.951072685302452e-05,
+      "loss": 0.7503,
+      "step": 970
+    },
+    {
+      "epoch": 0.11384866884633349,
+      "grad_norm": 0.83555937924101,
+      "learning_rate": 4.950167488073844e-05,
+      "loss": 0.742,
+      "step": 975
+    },
+    {
+      "epoch": 0.11443250817375059,
+      "grad_norm": 0.9192402633470362,
+      "learning_rate": 4.949254087702896e-05,
+      "loss": 0.7509,
+      "step": 980
+    },
+    {
+      "epoch": 0.11501634750116768,
+      "grad_norm": 0.7677750625588471,
+      "learning_rate": 4.948332487595148e-05,
+      "loss": 0.7572,
+      "step": 985
+    },
+    {
+      "epoch": 0.11560018682858478,
+      "grad_norm": 0.7809723688586555,
+      "learning_rate": 4.9474026911867084e-05,
+      "loss": 0.7603,
+      "step": 990
+    },
+    {
+      "epoch": 0.11618402615600187,
+      "grad_norm": 0.8246656807985904,
+      "learning_rate": 4.9464647019442465e-05,
+      "loss": 0.7269,
+      "step": 995
+    },
+    {
+      "epoch": 0.11676786548341897,
+      "grad_norm": 0.9446736222402072,
+      "learning_rate": 4.945518523364976e-05,
+      "loss": 0.758,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11735170481083605,
+      "grad_norm": 0.8111859145900984,
+      "learning_rate": 4.944564158976647e-05,
+      "loss": 0.7619,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11793554413825315,
+      "grad_norm": 0.9039015274993903,
+      "learning_rate": 4.943601612337528e-05,
+      "loss": 0.7535,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11851938346567024,
+      "grad_norm": 1.0091119488608187,
+      "learning_rate": 4.9426308870363934e-05,
+      "loss": 0.7536,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11910322279308734,
+      "grad_norm": 0.842082203309652,
+      "learning_rate": 4.941651986692514e-05,
+      "loss": 0.7656,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11968706212050444,
+      "grad_norm": 0.8944306880913642,
+      "learning_rate": 4.940664914955637e-05,
+      "loss": 0.7566,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12027090144792153,
+      "grad_norm": 0.7868127840279026,
+      "learning_rate": 4.939669675505978e-05,
+      "loss": 0.7677,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12085474077533863,
+      "grad_norm": 0.8392305677420804,
+      "learning_rate": 4.938666272054205e-05,
+      "loss": 0.7372,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12143858010275572,
+      "grad_norm": 0.8275154165543548,
+      "learning_rate": 4.937654708341425e-05,
+      "loss": 0.7684,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12202241943017282,
+      "grad_norm": 1.0271569251545638,
+      "learning_rate": 4.93663498813917e-05,
+      "loss": 0.7559,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12260625875758992,
+      "grad_norm": 0.8001356326071442,
+      "learning_rate": 4.9356071152493815e-05,
+      "loss": 0.7479,
+      "step": 1050
+    },
+    {
+      "epoch": 0.123190098085007,
+      "grad_norm": 0.7615039077476046,
+      "learning_rate": 4.934571093504398e-05,
+      "loss": 0.7436,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1237739374124241,
+      "grad_norm": 0.8327757922786666,
+      "learning_rate": 4.933526926766943e-05,
+      "loss": 0.7756,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12435777673984119,
+      "grad_norm": 0.782215645789506,
+      "learning_rate": 4.9324746189301027e-05,
+      "loss": 0.7631,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12494161606725829,
+      "grad_norm": 0.7784532705295107,
+      "learning_rate": 4.9314141739173223e-05,
+      "loss": 0.7589,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12552545539467538,
+      "grad_norm": 0.8109850694960712,
+      "learning_rate": 4.9303455956823816e-05,
+      "loss": 0.7711,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12610929472209248,
+      "grad_norm": 0.8610019246237505,
+      "learning_rate": 4.929268888209388e-05,
+      "loss": 0.7381,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12669313404950958,
+      "grad_norm": 0.8738321504500893,
+      "learning_rate": 4.928184055512754e-05,
+      "loss": 0.7527,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12727697337692667,
+      "grad_norm": 0.7909309571669735,
+      "learning_rate": 4.927091101637189e-05,
+      "loss": 0.7435,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12786081270434377,
+      "grad_norm": 0.8289072357424887,
+      "learning_rate": 4.9259900306576825e-05,
+      "loss": 0.7299,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12844465203176086,
+      "grad_norm": 0.8249302215432155,
+      "learning_rate": 4.924880846679485e-05,
+      "loss": 0.7479,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12902849135917796,
+      "grad_norm": 0.759611385335739,
+      "learning_rate": 4.923763553838098e-05,
+      "loss": 0.7535,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12961233068659506,
+      "grad_norm": 0.7534304608019847,
+      "learning_rate": 4.9226381562992546e-05,
+      "loss": 0.7599,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13019617001401215,
+      "grad_norm": 0.7560322412131688,
+      "learning_rate": 4.9215046582589066e-05,
+      "loss": 0.7515,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13078000934142925,
+      "grad_norm": 0.8830214475388138,
+      "learning_rate": 4.9203630639432083e-05,
+      "loss": 0.7437,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13136384866884634,
+      "grad_norm": 1.0110046096851761,
+      "learning_rate": 4.919213377608499e-05,
+      "loss": 0.7612,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13194768799626344,
+      "grad_norm": 0.8362938959468142,
+      "learning_rate": 4.9180556035412876e-05,
+      "loss": 0.7444,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13253152732368054,
+      "grad_norm": 1.0345951790579992,
+      "learning_rate": 4.916889746058242e-05,
+      "loss": 0.7683,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1331153666510976,
+      "grad_norm": 0.9305435521245572,
+      "learning_rate": 4.9157158095061636e-05,
+      "loss": 0.7439,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1336992059785147,
+      "grad_norm": 0.9211532554113626,
+      "learning_rate": 4.914533798261977e-05,
+      "loss": 0.7392,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1342830453059318,
+      "grad_norm": 1.0550561983312177,
+      "learning_rate": 4.913343716732713e-05,
+      "loss": 0.7592,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1348668846333489,
+      "grad_norm": 0.8617952378635508,
+      "learning_rate": 4.912145569355495e-05,
+      "loss": 0.7462,
+      "step": 1155
+    },
+    {
+      "epoch": 0.135450723960766,
+      "grad_norm": 0.8369067397902832,
+      "learning_rate": 4.910939360597514e-05,
+      "loss": 0.7325,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13603456328818309,
+      "grad_norm": 0.8662955701532267,
+      "learning_rate": 4.909725094956019e-05,
+      "loss": 0.7572,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13661840261560018,
+      "grad_norm": 0.8781226126848558,
+      "learning_rate": 4.908502776958301e-05,
+      "loss": 0.7414,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13720224194301728,
+      "grad_norm": 1.0068828191786867,
+      "learning_rate": 4.907272411161668e-05,
+      "loss": 0.7407,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13778608127043437,
+      "grad_norm": 0.8102116162804266,
+      "learning_rate": 4.9060340021534415e-05,
+      "loss": 0.7424,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13836992059785147,
+      "grad_norm": 0.8182089508424056,
+      "learning_rate": 4.9047875545509235e-05,
+      "loss": 0.7369,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13895375992526857,
+      "grad_norm": 0.8533530387273335,
+      "learning_rate": 4.9035330730013926e-05,
+      "loss": 0.7384,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13953759925268566,
+      "grad_norm": 0.8163147298892877,
+      "learning_rate": 4.9022705621820786e-05,
+      "loss": 0.7241,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14012143858010276,
+      "grad_norm": 0.885744418567555,
+      "learning_rate": 4.901000026800148e-05,
+      "loss": 0.7528,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14070527790751985,
+      "grad_norm": 0.8738516104757809,
+      "learning_rate": 4.899721471592688e-05,
+      "loss": 0.732,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14128911723493695,
+      "grad_norm": 0.8068156315381868,
+      "learning_rate": 4.898434901326685e-05,
+      "loss": 0.726,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14187295656235405,
+      "grad_norm": 0.7856848802153881,
+      "learning_rate": 4.897140320799011e-05,
+      "loss": 0.7325,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14245679588977114,
+      "grad_norm": 0.8355251687297792,
+      "learning_rate": 4.8958377348364e-05,
+      "loss": 0.7528,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14304063521718824,
+      "grad_norm": 0.7684488878166239,
+      "learning_rate": 4.894527148295438e-05,
+      "loss": 0.7509,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14362447454460534,
+      "grad_norm": 1.0616657426243217,
+      "learning_rate": 4.8932085660625374e-05,
+      "loss": 0.7547,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14420831387202243,
+      "grad_norm": 0.9230922768754513,
+      "learning_rate": 4.8918819930539244e-05,
+      "loss": 0.7397,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14479215319943953,
+      "grad_norm": 0.9018167634804453,
+      "learning_rate": 4.8905474342156144e-05,
+      "loss": 0.7488,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1453759925268566,
+      "grad_norm": 0.8880821241737537,
+      "learning_rate": 4.889204894523401e-05,
+      "loss": 0.7489,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1459598318542737,
+      "grad_norm": 0.8306320426031331,
+      "learning_rate": 4.8878543789828314e-05,
+      "loss": 0.7403,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1465436711816908,
+      "grad_norm": 0.8903153304358166,
+      "learning_rate": 4.886495892629191e-05,
+      "loss": 0.7558,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14712751050910788,
+      "grad_norm": 0.8380421321813797,
+      "learning_rate": 4.8851294405274855e-05,
+      "loss": 0.7252,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14771134983652498,
+      "grad_norm": 0.9364679991310748,
+      "learning_rate": 4.8837550277724165e-05,
+      "loss": 0.734,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14829518916394208,
+      "grad_norm": 0.9251425941623983,
+      "learning_rate": 4.8823726594883696e-05,
+      "loss": 0.7472,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14887902849135917,
+      "grad_norm": 0.9273829246305205,
+      "learning_rate": 4.8809823408293887e-05,
+      "loss": 0.7477,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14946286781877627,
+      "grad_norm": 0.7946084300359817,
+      "learning_rate": 4.8795840769791634e-05,
+      "loss": 0.7515,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15004670714619336,
+      "grad_norm": 0.8503320421933336,
+      "learning_rate": 4.878177873151004e-05,
+      "loss": 0.7465,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15063054647361046,
+      "grad_norm": 0.8694883327865317,
+      "learning_rate": 4.876763734587825e-05,
+      "loss": 0.738,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15121438580102756,
+      "grad_norm": 0.7569044838064122,
+      "learning_rate": 4.8753416665621255e-05,
+      "loss": 0.7217,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15179822512844465,
+      "grad_norm": 0.810783326961325,
+      "learning_rate": 4.873911674375968e-05,
+      "loss": 0.7536,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15238206445586175,
+      "grad_norm": 0.8698111120094745,
+      "learning_rate": 4.87247376336096e-05,
+      "loss": 0.7481,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15296590378327884,
+      "grad_norm": 0.9178163836621003,
+      "learning_rate": 4.8710279388782345e-05,
+      "loss": 0.7401,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15354974311069594,
+      "grad_norm": 0.8785969161453483,
+      "learning_rate": 4.869574206318427e-05,
+      "loss": 0.7308,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15413358243811304,
+      "grad_norm": 0.787087494439056,
+      "learning_rate": 4.868112571101659e-05,
+      "loss": 0.7142,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15471742176553013,
+      "grad_norm": 0.9221866180118231,
+      "learning_rate": 4.866643038677519e-05,
+      "loss": 0.7439,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15530126109294723,
+      "grad_norm": 0.8861566553427299,
+      "learning_rate": 4.865165614525033e-05,
+      "loss": 0.7452,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15588510042036433,
+      "grad_norm": 0.8470239189350767,
+      "learning_rate": 4.863680304152657e-05,
+      "loss": 0.7486,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15646893974778142,
+      "grad_norm": 0.8491579579103056,
+      "learning_rate": 4.862187113098249e-05,
+      "loss": 0.7489,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15705277907519852,
+      "grad_norm": 0.9525828664204632,
+      "learning_rate": 4.8606860469290454e-05,
+      "loss": 0.7364,
+      "step": 1345
+    },
+    {
+      "epoch": 0.1576366184026156,
+      "grad_norm": 0.9121064096531482,
+      "learning_rate": 4.859177111241649e-05,
+      "loss": 0.72,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15822045773003268,
+      "grad_norm": 0.8488919401410446,
+      "learning_rate": 4.8576603116620004e-05,
+      "loss": 0.7352,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15880429705744978,
+      "grad_norm": 0.8145226432017278,
+      "learning_rate": 4.8561356538453625e-05,
+      "loss": 0.7236,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15938813638486687,
+      "grad_norm": 0.7858513627050332,
+      "learning_rate": 4.8546031434762954e-05,
+      "loss": 0.7581,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15997197571228397,
+      "grad_norm": 0.8632804578998556,
+      "learning_rate": 4.853062786268636e-05,
+      "loss": 0.7529,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16055581503970107,
+      "grad_norm": 0.8662933609475204,
+      "learning_rate": 4.85151458796548e-05,
+      "loss": 0.7196,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16113965436711816,
+      "grad_norm": 0.9008272613777886,
+      "learning_rate": 4.849958554339156e-05,
+      "loss": 0.7562,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16172349369453526,
+      "grad_norm": 1.0396019649473314,
+      "learning_rate": 4.8483946911912064e-05,
+      "loss": 0.7474,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16230733302195235,
+      "grad_norm": 0.9021657264167294,
+      "learning_rate": 4.846823004352366e-05,
+      "loss": 0.7286,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16289117234936945,
+      "grad_norm": 0.9831547552660019,
+      "learning_rate": 4.845243499682539e-05,
+      "loss": 0.7538,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16347501167678655,
+      "grad_norm": 0.8494290490564191,
+      "learning_rate": 4.8436561830707786e-05,
+      "loss": 0.7359,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16405885100420364,
+      "grad_norm": 0.8777452106207012,
+      "learning_rate": 4.842061060435261e-05,
+      "loss": 0.7323,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16464269033162074,
+      "grad_norm": 0.7876098244119992,
+      "learning_rate": 4.840458137723271e-05,
+      "loss": 0.7043,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16522652965903784,
+      "grad_norm": 0.9636798433350332,
+      "learning_rate": 4.838847420911172e-05,
+      "loss": 0.7255,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16581036898645493,
+      "grad_norm": 0.9020478525489855,
+      "learning_rate": 4.8372289160043895e-05,
+      "loss": 0.7421,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16639420831387203,
+      "grad_norm": 0.9043231536695949,
+      "learning_rate": 4.835602629037384e-05,
+      "loss": 0.7333,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16697804764128912,
+      "grad_norm": 0.8053495862344994,
+      "learning_rate": 4.8339685660736324e-05,
+      "loss": 0.7263,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16756188696870622,
+      "grad_norm": 0.9550285318543651,
+      "learning_rate": 4.8323267332056026e-05,
+      "loss": 0.7368,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16814572629612332,
+      "grad_norm": 0.9381409461089547,
+      "learning_rate": 4.830677136554733e-05,
+      "loss": 0.7341,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1687295656235404,
+      "grad_norm": 0.8683417476073385,
+      "learning_rate": 4.829019782271408e-05,
+      "loss": 0.7556,
+      "step": 1445
+    },
+    {
+      "epoch": 0.1693134049509575,
+      "grad_norm": 0.8447674256758361,
+      "learning_rate": 4.827354676534937e-05,
+      "loss": 0.7492,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1698972442783746,
+      "grad_norm": 0.8050509416825757,
+      "learning_rate": 4.825681825553527e-05,
+      "loss": 0.7421,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17048108360579167,
+      "grad_norm": 0.7649657099473817,
+      "learning_rate": 4.824001235564265e-05,
+      "loss": 0.7487,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17106492293320877,
+      "grad_norm": 0.857389296357302,
+      "learning_rate": 4.822312912833092e-05,
+      "loss": 0.733,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17164876226062586,
+      "grad_norm": 0.8424832802837008,
+      "learning_rate": 4.82061686365478e-05,
+      "loss": 0.7362,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17223260158804296,
+      "grad_norm": 0.9243836306800552,
+      "learning_rate": 4.818913094352907e-05,
+      "loss": 0.7342,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17281644091546006,
+      "grad_norm": 0.7705246634042684,
+      "learning_rate": 4.8172016112798364e-05,
+      "loss": 0.7474,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17340028024287715,
+      "grad_norm": 0.8561759097930621,
+      "learning_rate": 4.8154824208166906e-05,
+      "loss": 0.7628,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17398411957029425,
+      "grad_norm": 0.8716723514678405,
+      "learning_rate": 4.8137555293733294e-05,
+      "loss": 0.7296,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17456795889771135,
+      "grad_norm": 0.8590395423969877,
+      "learning_rate": 4.812020943388324e-05,
+      "loss": 0.7156,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17515179822512844,
+      "grad_norm": 0.8227359438250613,
+      "learning_rate": 4.810278669328935e-05,
+      "loss": 0.7458,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17573563755254554,
+      "grad_norm": 0.9291613782890741,
+      "learning_rate": 4.808528713691087e-05,
+      "loss": 0.7197,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17631947687996263,
+      "grad_norm": 0.9252768812683061,
+      "learning_rate": 4.806771082999346e-05,
+      "loss": 0.752,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17690331620737973,
+      "grad_norm": 0.828069657263595,
+      "learning_rate": 4.8050057838068904e-05,
+      "loss": 0.7454,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17748715553479683,
+      "grad_norm": 0.8547710287364533,
+      "learning_rate": 4.803232822695493e-05,
+      "loss": 0.7365,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17807099486221392,
+      "grad_norm": 0.9291014403307367,
+      "learning_rate": 4.801452206275493e-05,
+      "loss": 0.7315,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17865483418963102,
+      "grad_norm": 0.8381441138725285,
+      "learning_rate": 4.79966394118577e-05,
+      "loss": 0.7246,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1792386735170481,
+      "grad_norm": 1.012812492678937,
+      "learning_rate": 4.797868034093724e-05,
+      "loss": 0.7514,
+      "step": 1535
+    },
+    {
+      "epoch": 0.1798225128444652,
+      "grad_norm": 1.0163395309432404,
+      "learning_rate": 4.7960644916952444e-05,
+      "loss": 0.7448,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1804063521718823,
+      "grad_norm": 0.8675054579386493,
+      "learning_rate": 4.7942533207146916e-05,
+      "loss": 0.7533,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1809901914992994,
+      "grad_norm": 0.8546813403666974,
+      "learning_rate": 4.792434527904864e-05,
+      "loss": 0.7451,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1815740308267165,
+      "grad_norm": 0.7424392194511835,
+      "learning_rate": 4.7906081200469835e-05,
+      "loss": 0.7326,
+      "step": 1555
+    },
+    {
+      "epoch": 0.1821578701541336,
+      "grad_norm": 0.7825121640532524,
+      "learning_rate": 4.788774103950657e-05,
+      "loss": 0.7286,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1827417094815507,
+      "grad_norm": 0.980002561746577,
+      "learning_rate": 4.7869324864538636e-05,
+      "loss": 0.7346,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18332554880896776,
+      "grad_norm": 0.8672472363588772,
+      "learning_rate": 4.7850832744229216e-05,
+      "loss": 0.7424,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18390938813638485,
+      "grad_norm": 0.8646730016652088,
+      "learning_rate": 4.783226474752465e-05,
+      "loss": 0.7412,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18449322746380195,
+      "grad_norm": 0.8337338110769495,
+      "learning_rate": 4.781362094365417e-05,
+      "loss": 0.7218,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18507706679121905,
+      "grad_norm": 0.8339883285485024,
+      "learning_rate": 4.779490140212966e-05,
+      "loss": 0.7208,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18566090611863614,
+      "grad_norm": 0.934085237869705,
+      "learning_rate": 4.777610619274539e-05,
+      "loss": 0.7376,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18624474544605324,
+      "grad_norm": 0.755725492136055,
+      "learning_rate": 4.775723538557772e-05,
+      "loss": 0.7166,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18682858477347034,
+      "grad_norm": 0.9508762513943106,
+      "learning_rate": 4.7738289050984905e-05,
+      "loss": 0.7424,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18741242410088743,
+      "grad_norm": 0.9845480086954423,
+      "learning_rate": 4.7719267259606795e-05,
+      "loss": 0.7306,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18799626342830453,
+      "grad_norm": 1.0187787645203552,
+      "learning_rate": 4.770017008236455e-05,
+      "loss": 0.7429,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18858010275572162,
+      "grad_norm": 1.0766199613681091,
+      "learning_rate": 4.768099759046042e-05,
+      "loss": 0.7139,
+      "step": 1615
+    },
+    {
+      "epoch": 0.18916394208313872,
+      "grad_norm": 0.8018949169094243,
+      "learning_rate": 4.766174985537745e-05,
+      "loss": 0.7064,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18974778141055582,
+      "grad_norm": 0.9410002469224804,
+      "learning_rate": 4.7642426948879234e-05,
+      "loss": 0.7142,
+      "step": 1625
+    },
+    {
+      "epoch": 0.1903316207379729,
+      "grad_norm": 0.9924778731480929,
+      "learning_rate": 4.762302894300962e-05,
+      "loss": 0.7469,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19091546006539,
+      "grad_norm": 0.8928365170914925,
+      "learning_rate": 4.760355591009247e-05,
+      "loss": 0.7424,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1914992993928071,
+      "grad_norm": 0.7719462216054435,
+      "learning_rate": 4.7584007922731383e-05,
+      "loss": 0.7381,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1920831387202242,
+      "grad_norm": 0.7713047418375075,
+      "learning_rate": 4.75643850538094e-05,
+      "loss": 0.719,
+      "step": 1645
+    },
+    {
+      "epoch": 0.1926669780476413,
+      "grad_norm": 0.9075548502670161,
+      "learning_rate": 4.754468737648878e-05,
+      "loss": 0.7274,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1932508173750584,
+      "grad_norm": 0.8580706227087707,
+      "learning_rate": 4.752491496421066e-05,
+      "loss": 0.7336,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1938346567024755,
+      "grad_norm": 0.7720549473121491,
+      "learning_rate": 4.750506789069486e-05,
+      "loss": 0.7284,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19441849602989258,
+      "grad_norm": 0.9120434353934009,
+      "learning_rate": 4.7485146229939545e-05,
+      "loss": 0.726,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19500233535730968,
+      "grad_norm": 0.8772498797608074,
+      "learning_rate": 4.746515005622097e-05,
+      "loss": 0.721,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19558617468472678,
+      "grad_norm": 1.0402305568608219,
+      "learning_rate": 4.744507944409322e-05,
+      "loss": 0.7474,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19617001401214385,
+      "grad_norm": 0.9480289631067886,
+      "learning_rate": 4.742493446838791e-05,
+      "loss": 0.7127,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19675385333956094,
+      "grad_norm": 0.8549076161944649,
+      "learning_rate": 4.740471520421392e-05,
+      "loss": 0.7283,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19733769266697804,
+      "grad_norm": 0.8916814608078604,
+      "learning_rate": 4.73844217269571e-05,
+      "loss": 0.726,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19792153199439513,
+      "grad_norm": 0.719882334233512,
+      "learning_rate": 4.736405411228e-05,
+      "loss": 0.717,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19850537132181223,
+      "grad_norm": 0.9257642921113124,
+      "learning_rate": 4.7343612436121575e-05,
+      "loss": 0.765,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19908921064922933,
+      "grad_norm": 0.7913682603162452,
+      "learning_rate": 4.732309677469693e-05,
+      "loss": 0.7444,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19967304997664642,
+      "grad_norm": 0.8706125916216539,
+      "learning_rate": 4.7302507204497026e-05,
+      "loss": 0.7213,
+      "step": 1710
+    },
+    {
+      "epoch": 0.20025688930406352,
+      "grad_norm": 0.8791791140311229,
+      "learning_rate": 4.728184380228834e-05,
+      "loss": 0.7396,
+      "step": 1715
+    },
+    {
+      "epoch": 0.20084072863148061,
+      "grad_norm": 0.8625015103710599,
+      "learning_rate": 4.7261106645112677e-05,
+      "loss": 0.7301,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2014245679588977,
+      "grad_norm": 0.8861750728990333,
+      "learning_rate": 4.72402958102868e-05,
+      "loss": 0.7491,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2020084072863148,
+      "grad_norm": 0.9514113714928437,
+      "learning_rate": 4.72194113754022e-05,
+      "loss": 0.7327,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2025922466137319,
+      "grad_norm": 0.9369626542739419,
+      "learning_rate": 4.719845341832475e-05,
+      "loss": 0.6914,
+      "step": 1735
+    },
+    {
+      "epoch": 0.203176085941149,
+      "grad_norm": 1.2929308377238355,
+      "learning_rate": 4.7177422017194464e-05,
+      "loss": 0.7227,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2037599252685661,
+      "grad_norm": 1.0761261573620378,
+      "learning_rate": 4.715631725042517e-05,
+      "loss": 0.7453,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2043437645959832,
+      "grad_norm": 0.9015493784655515,
+      "learning_rate": 4.7135139196704254e-05,
+      "loss": 0.7262,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2049276039234003,
+      "grad_norm": 0.8448441635518296,
+      "learning_rate": 4.711388793499233e-05,
+      "loss": 0.7215,
+      "step": 1755
+    },
+    {
+      "epoch": 0.20551144325081738,
+      "grad_norm": 0.911123740513117,
+      "learning_rate": 4.709256354452298e-05,
+      "loss": 0.7317,
+      "step": 1760
+    },
+    {
+      "epoch": 0.20609528257823448,
+      "grad_norm": 0.9212102275736465,
+      "learning_rate": 4.7071166104802415e-05,
+      "loss": 0.7261,
+      "step": 1765
+    },
+    {
+      "epoch": 0.20667912190565157,
+      "grad_norm": 0.8280598637016623,
+      "learning_rate": 4.7049695695609224e-05,
+      "loss": 0.7123,
+      "step": 1770
+    },
+    {
+      "epoch": 0.20726296123306867,
+      "grad_norm": 0.7931744829771525,
+      "learning_rate": 4.702815239699405e-05,
+      "loss": 0.7169,
+      "step": 1775
+    },
+    {
+      "epoch": 0.20784680056048577,
+      "grad_norm": 0.8016643144301633,
+      "learning_rate": 4.7006536289279285e-05,
+      "loss": 0.7195,
+      "step": 1780
+    },
+    {
+      "epoch": 0.20843063988790284,
+      "grad_norm": 0.8263592261687366,
+      "learning_rate": 4.698484745305882e-05,
+      "loss": 0.7033,
+      "step": 1785
+    },
+    {
+      "epoch": 0.20901447921531993,
+      "grad_norm": 0.8507324933533925,
+      "learning_rate": 4.696308596919767e-05,
+      "loss": 0.7479,
+      "step": 1790
+    },
+    {
+      "epoch": 0.20959831854273703,
+      "grad_norm": 0.8328383640267244,
+      "learning_rate": 4.694125191883174e-05,
+      "loss": 0.7219,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21018215787015412,
+      "grad_norm": 0.7346869424532056,
+      "learning_rate": 4.691934538336746e-05,
+      "loss": 0.7272,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21076599719757122,
+      "grad_norm": 0.7553117082839136,
+      "learning_rate": 4.6897366444481545e-05,
+      "loss": 0.7209,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21134983652498832,
+      "grad_norm": 0.8261600749869238,
+      "learning_rate": 4.687531518412065e-05,
+      "loss": 0.733,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2119336758524054,
+      "grad_norm": 0.8631456199547245,
+      "learning_rate": 4.685319168450107e-05,
+      "loss": 0.6968,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2125175151798225,
+      "grad_norm": 0.9026029059640308,
+      "learning_rate": 4.683099602810845e-05,
+      "loss": 0.7473,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2131013545072396,
+      "grad_norm": 0.8594021123209141,
+      "learning_rate": 4.680872829769745e-05,
+      "loss": 0.7164,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2136851938346567,
+      "grad_norm": 0.9662714499957493,
+      "learning_rate": 4.6786388576291446e-05,
+      "loss": 0.7405,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2142690331620738,
+      "grad_norm": 0.8755429706535786,
+      "learning_rate": 4.6763976947182256e-05,
+      "loss": 0.7169,
+      "step": 1835
+    },
+    {
+      "epoch": 0.2148528724894909,
+      "grad_norm": 0.8295598426423224,
+      "learning_rate": 4.6741493493929794e-05,
+      "loss": 0.722,
+      "step": 1840
+    },
+    {
+      "epoch": 0.215436711816908,
+      "grad_norm": 0.8762905396970913,
+      "learning_rate": 4.671893830036174e-05,
+      "loss": 0.73,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21602055114432508,
+      "grad_norm": 0.9509644103150598,
+      "learning_rate": 4.6696311450573266e-05,
+      "loss": 0.7474,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21660439047174218,
+      "grad_norm": 0.9699081790127314,
+      "learning_rate": 4.667361302892671e-05,
+      "loss": 0.7062,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21718822979915928,
+      "grad_norm": 0.9442815326623876,
+      "learning_rate": 4.665084312005126e-05,
+      "loss": 0.6956,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21777206912657637,
+      "grad_norm": 0.8067541386553867,
+      "learning_rate": 4.662800180884263e-05,
+      "loss": 0.7417,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21835590845399347,
+      "grad_norm": 0.8697450351973937,
+      "learning_rate": 4.660508918046277e-05,
+      "loss": 0.7252,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21893974778141057,
+      "grad_norm": 0.772276879857256,
+      "learning_rate": 4.658210532033951e-05,
+      "loss": 0.7317,
+      "step": 1875
+    },
+    {
+      "epoch": 0.21952358710882766,
+      "grad_norm": 0.8192894910458268,
+      "learning_rate": 4.6559050314166264e-05,
+      "loss": 0.7196,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22010742643624476,
+      "grad_norm": 0.868446881287682,
+      "learning_rate": 4.653592424790172e-05,
+      "loss": 0.734,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22069126576366185,
+      "grad_norm": 0.8099440249953018,
+      "learning_rate": 4.6512727207769504e-05,
+      "loss": 0.7362,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22127510509107892,
+      "grad_norm": 0.7864716011858071,
+      "learning_rate": 4.6489459280257856e-05,
+      "loss": 0.7188,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22185894441849602,
+      "grad_norm": 0.8300689479187022,
+      "learning_rate": 4.646612055211933e-05,
+      "loss": 0.7418,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22244278374591311,
+      "grad_norm": 0.7697294087146005,
+      "learning_rate": 4.6442711110370424e-05,
+      "loss": 0.6993,
+      "step": 1905
+    },
+    {
+      "epoch": 0.2230266230733302,
+      "grad_norm": 0.908818261954294,
+      "learning_rate": 4.64192310422913e-05,
+      "loss": 0.7115,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2236104624007473,
+      "grad_norm": 0.7593820871173316,
+      "learning_rate": 4.639568043542548e-05,
+      "loss": 0.7061,
+      "step": 1915
+    },
+    {
+      "epoch": 0.2241943017281644,
+      "grad_norm": 0.8524203826838567,
+      "learning_rate": 4.6372059377579414e-05,
+      "loss": 0.7242,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2247781410555815,
+      "grad_norm": 0.8993830914176336,
+      "learning_rate": 4.634836795682228e-05,
+      "loss": 0.7171,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2253619803829986,
+      "grad_norm": 0.7814420865591079,
+      "learning_rate": 4.632460626148558e-05,
+      "loss": 0.7285,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2259458197104157,
+      "grad_norm": 0.74700148461869,
+      "learning_rate": 4.6300774380162825e-05,
+      "loss": 0.7325,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2265296590378328,
+      "grad_norm": 0.8530190257657563,
+      "learning_rate": 4.627687240170921e-05,
+      "loss": 0.7224,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22711349836524988,
+      "grad_norm": 0.8588558793942196,
+      "learning_rate": 4.625290041524128e-05,
+      "loss": 0.7253,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22769733769266698,
+      "grad_norm": 0.7613854196909419,
+      "learning_rate": 4.6228858510136616e-05,
+      "loss": 0.7166,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22828117702008408,
+      "grad_norm": 0.8223207615558868,
+      "learning_rate": 4.620474677603345e-05,
+      "loss": 0.726,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22886501634750117,
+      "grad_norm": 0.7697519948932519,
+      "learning_rate": 4.61805653028304e-05,
+      "loss": 0.7218,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22944885567491827,
+      "grad_norm": 0.8675067664713971,
+      "learning_rate": 4.615631418068609e-05,
+      "loss": 0.739,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23003269500233536,
+      "grad_norm": 0.8036791190421861,
+      "learning_rate": 4.613199350001881e-05,
+      "loss": 0.7338,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23061653432975246,
+      "grad_norm": 1.052515044144309,
+      "learning_rate": 4.6107603351506205e-05,
+      "loss": 0.7263,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23120037365716956,
+      "grad_norm": 0.8356991201425433,
+      "learning_rate": 4.608314382608493e-05,
+      "loss": 0.7207,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23178421298458665,
+      "grad_norm": 0.8232997169558762,
+      "learning_rate": 4.6058615014950315e-05,
+      "loss": 0.7367,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23236805231200375,
+      "grad_norm": 0.797210179690259,
+      "learning_rate": 4.6034017009555975e-05,
+      "loss": 0.73,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23295189163942084,
+      "grad_norm": 0.8241062552935673,
+      "learning_rate": 4.600934990161355e-05,
+      "loss": 0.7171,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23353573096683794,
+      "grad_norm": 0.7937008955270574,
+      "learning_rate": 4.598461378309231e-05,
+      "loss": 0.7291,
+      "step": 2000
+    },
+    {
+      "epoch": 0.234119570294255,
+      "grad_norm": 0.78035068225242,
+      "learning_rate": 4.5959808746218823e-05,
+      "loss": 0.7146,
+      "step": 2005
+    },
+    {
+      "epoch": 0.2347034096216721,
+      "grad_norm": 0.8134248627791105,
+      "learning_rate": 4.593493488347662e-05,
+      "loss": 0.7262,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2352872489490892,
+      "grad_norm": 0.8986352822533618,
+      "learning_rate": 4.590999228760583e-05,
+      "loss": 0.7224,
+      "step": 2015
+    },
+    {
+      "epoch": 0.2358710882765063,
+      "grad_norm": 0.8799624641629938,
+      "learning_rate": 4.5884981051602873e-05,
+      "loss": 0.7039,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2364549276039234,
+      "grad_norm": 0.8062832337406176,
+      "learning_rate": 4.585990126872006e-05,
+      "loss": 0.7105,
+      "step": 2025
+    },
+    {
+      "epoch": 0.2370387669313405,
+      "grad_norm": 0.8023617406960752,
+      "learning_rate": 4.583475303246527e-05,
+      "loss": 0.7082,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23762260625875758,
+      "grad_norm": 0.7858415440677039,
+      "learning_rate": 4.580953643660165e-05,
+      "loss": 0.7148,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23820644558617468,
+      "grad_norm": 0.8747517661212155,
+      "learning_rate": 4.5784251575147176e-05,
+      "loss": 0.7322,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23879028491359178,
+      "grad_norm": 0.8011757891417564,
+      "learning_rate": 4.5758898542374354e-05,
+      "loss": 0.7281,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23937412424100887,
+      "grad_norm": 0.8088704185187473,
+      "learning_rate": 4.5733477432809884e-05,
+      "loss": 0.7048,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23995796356842597,
+      "grad_norm": 0.7544430888220308,
+      "learning_rate": 4.570798834123425e-05,
+      "loss": 0.7111,
+      "step": 2055
+    },
+    {
+      "epoch": 0.24054180289584307,
+      "grad_norm": 0.8717789597793989,
+      "learning_rate": 4.5682431362681435e-05,
+      "loss": 0.7111,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24112564222326016,
+      "grad_norm": 0.7714012324044205,
+      "learning_rate": 4.565680659243851e-05,
+      "loss": 0.6974,
+      "step": 2065
+    },
+    {
+      "epoch": 0.24170948155067726,
+      "grad_norm": 0.7555562896364412,
+      "learning_rate": 4.5631114126045315e-05,
+      "loss": 0.718,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24229332087809435,
+      "grad_norm": 0.7097810232586401,
+      "learning_rate": 4.560535405929408e-05,
+      "loss": 0.7062,
+      "step": 2075
+    },
+    {
+      "epoch": 0.24287716020551145,
+      "grad_norm": 0.769470044177514,
+      "learning_rate": 4.557952648822908e-05,
+      "loss": 0.7189,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24346099953292855,
+      "grad_norm": 0.7904005906275823,
+      "learning_rate": 4.555363150914628e-05,
+      "loss": 0.7314,
+      "step": 2085
+    },
+    {
+      "epoch": 0.24404483886034564,
+      "grad_norm": 0.8463560711087779,
+      "learning_rate": 4.552766921859297e-05,
+      "loss": 0.707,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24462867818776274,
+      "grad_norm": 0.845867432164641,
+      "learning_rate": 4.5501639713367386e-05,
+      "loss": 0.6973,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24521251751517983,
+      "grad_norm": 0.8119973038192364,
+      "learning_rate": 4.547554309051839e-05,
+      "loss": 0.7359,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24579635684259693,
+      "grad_norm": 0.8828267070913842,
+      "learning_rate": 4.5449379447345084e-05,
+      "loss": 0.7217,
+      "step": 2105
+    },
+    {
+      "epoch": 0.246380196170014,
+      "grad_norm": 0.7889889781042302,
+      "learning_rate": 4.5423148881396444e-05,
+      "loss": 0.7285,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2469640354974311,
+      "grad_norm": 0.8253943240619854,
+      "learning_rate": 4.539685149047097e-05,
+      "loss": 0.7175,
+      "step": 2115
+    },
+    {
+      "epoch": 0.2475478748248482,
+      "grad_norm": 0.8463995819623495,
+      "learning_rate": 4.5370487372616285e-05,
+      "loss": 0.7088,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2481317141522653,
+      "grad_norm": 0.7557279563999925,
+      "learning_rate": 4.5344056626128847e-05,
+      "loss": 0.7273,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24871555347968238,
+      "grad_norm": 0.8227994525525993,
+      "learning_rate": 4.53175593495535e-05,
+      "loss": 0.7221,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24929939280709948,
+      "grad_norm": 0.7571552923515421,
+      "learning_rate": 4.529099564168312e-05,
+      "loss": 0.7311,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24988323213451658,
+      "grad_norm": 0.8020862994896764,
+      "learning_rate": 4.526436560155833e-05,
+      "loss": 0.7156,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2504670714619337,
+      "grad_norm": 0.8257081755099529,
+      "learning_rate": 4.5237669328467e-05,
+      "loss": 0.7039,
+      "step": 2145
+    },
+    {
+      "epoch": 0.25105091078935077,
+      "grad_norm": 0.9252224979344469,
+      "learning_rate": 4.5210906921944e-05,
+      "loss": 0.7203,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2516347501167679,
+      "grad_norm": 0.7820690558066279,
+      "learning_rate": 4.518407848177073e-05,
+      "loss": 0.7283,
+      "step": 2155
+    },
+    {
+      "epoch": 0.25221858944418496,
+      "grad_norm": 0.7900246259761322,
+      "learning_rate": 4.515718410797481e-05,
+      "loss": 0.7405,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25280242877160203,
+      "grad_norm": 0.9165560691178495,
+      "learning_rate": 4.513022390082969e-05,
+      "loss": 0.7043,
+      "step": 2165
+    },
+    {
+      "epoch": 0.25338626809901915,
+      "grad_norm": 0.7375679989625112,
+      "learning_rate": 4.510319796085428e-05,
+      "loss": 0.7343,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2539701074264362,
+      "grad_norm": 0.7827496430575249,
+      "learning_rate": 4.5076106388812534e-05,
+      "loss": 0.7266,
+      "step": 2175
+    },
+    {
+      "epoch": 0.25455394675385334,
+      "grad_norm": 0.8723807945365605,
+      "learning_rate": 4.504894928571315e-05,
+      "loss": 0.7182,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2551377860812704,
+      "grad_norm": 0.8482133158531135,
+      "learning_rate": 4.502172675280915e-05,
+      "loss": 0.7104,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25572162540868754,
+      "grad_norm": 0.8524857579835551,
+      "learning_rate": 4.4994438891597486e-05,
+      "loss": 0.7635,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2563054647361046,
+      "grad_norm": 0.778666322067883,
+      "learning_rate": 4.496708580381868e-05,
+      "loss": 0.7338,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25688930406352173,
+      "grad_norm": 0.8723980465894413,
+      "learning_rate": 4.4939667591456465e-05,
+      "loss": 0.712,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2574731433909388,
+      "grad_norm": 0.8421038493972052,
+      "learning_rate": 4.491218435673737e-05,
+      "loss": 0.722,
+      "step": 2205
+    },
+    {
+      "epoch": 0.2580569827183559,
+      "grad_norm": 0.7426905139587554,
+      "learning_rate": 4.4884636202130365e-05,
+      "loss": 0.7111,
+      "step": 2210
+    },
+    {
+      "epoch": 0.258640822045773,
+      "grad_norm": 0.8604558266845712,
+      "learning_rate": 4.485702323034647e-05,
+      "loss": 0.721,
+      "step": 2215
+    },
+    {
+      "epoch": 0.2592246613731901,
+      "grad_norm": 0.8220760507302662,
+      "learning_rate": 4.4829345544338355e-05,
+      "loss": 0.7273,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2598085007006072,
+      "grad_norm": 0.871686191616385,
+      "learning_rate": 4.480160324729998e-05,
+      "loss": 0.7092,
+      "step": 2225
+    },
+    {
+      "epoch": 0.2603923400280243,
+      "grad_norm": 0.844341787935979,
+      "learning_rate": 4.477379644266621e-05,
+      "loss": 0.7131,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2609761793554414,
+      "grad_norm": 0.7900108496693427,
+      "learning_rate": 4.47459252341124e-05,
+      "loss": 0.7559,
+      "step": 2235
+    },
+    {
+      "epoch": 0.2615600186828585,
+      "grad_norm": 0.8273038621128911,
+      "learning_rate": 4.471798972555407e-05,
+      "loss": 0.7148,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26214385801027557,
+      "grad_norm": 0.7539025667177797,
+      "learning_rate": 4.468999002114642e-05,
+      "loss": 0.7315,
+      "step": 2245
+    },
+    {
+      "epoch": 0.2627276973376927,
+      "grad_norm": 0.7613321385392502,
+      "learning_rate": 4.4661926225284057e-05,
+      "loss": 0.7127,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26331153666510976,
+      "grad_norm": 0.750959574414275,
+      "learning_rate": 4.463379844260051e-05,
+      "loss": 0.7253,
+      "step": 2255
+    },
+    {
+      "epoch": 0.2638953759925269,
+      "grad_norm": 0.7479404319087383,
+      "learning_rate": 4.460560677796788e-05,
+      "loss": 0.706,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26447921531994395,
+      "grad_norm": 0.7999080096714866,
+      "learning_rate": 4.4577351336496466e-05,
+      "loss": 0.7199,
+      "step": 2265
+    },
+    {
+      "epoch": 0.2650630546473611,
+      "grad_norm": 0.7979384420318076,
+      "learning_rate": 4.454903222353433e-05,
+      "loss": 0.7034,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26564689397477814,
+      "grad_norm": 0.8482408600249933,
+      "learning_rate": 4.4520649544666955e-05,
+      "loss": 0.7157,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2662307333021952,
+      "grad_norm": 0.7981351178418482,
+      "learning_rate": 4.4492203405716804e-05,
+      "loss": 0.7092,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26681457262961233,
+      "grad_norm": 0.7962766501017232,
+      "learning_rate": 4.4463693912742944e-05,
+      "loss": 0.7097,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2673984119570294,
+      "grad_norm": 0.7507841063947497,
+      "learning_rate": 4.4435121172040674e-05,
+      "loss": 0.7101,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2679822512844465,
+      "grad_norm": 0.7783944622822936,
+      "learning_rate": 4.4406485290141075e-05,
+      "loss": 0.7024,
+      "step": 2295
+    },
+    {
+      "epoch": 0.2685660906118636,
+      "grad_norm": 0.7543292868458843,
+      "learning_rate": 4.437778637381068e-05,
+      "loss": 0.7033,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2691499299392807,
+      "grad_norm": 0.839842240820925,
+      "learning_rate": 4.434902453005101e-05,
+      "loss": 0.7396,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2697337692666978,
+      "grad_norm": 0.7745574814360575,
+      "learning_rate": 4.4320199866098216e-05,
+      "loss": 0.7096,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2703176085941149,
+      "grad_norm": 0.8333262012405849,
+      "learning_rate": 4.4291312489422684e-05,
+      "loss": 0.7312,
+      "step": 2315
+    },
+    {
+      "epoch": 0.270901447921532,
+      "grad_norm": 0.7527817460814907,
+      "learning_rate": 4.426236250772859e-05,
+      "loss": 0.7105,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2714852872489491,
+      "grad_norm": 0.7419140402986222,
+      "learning_rate": 4.423335002895358e-05,
+      "loss": 0.7066,
+      "step": 2325
+    },
+    {
+      "epoch": 0.27206912657636617,
+      "grad_norm": 0.822685880830946,
+      "learning_rate": 4.420427516126822e-05,
+      "loss": 0.7345,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2726529659037833,
+      "grad_norm": 0.7195809448647575,
+      "learning_rate": 4.4175138013075804e-05,
+      "loss": 0.7137,
+      "step": 2335
+    },
+    {
+      "epoch": 0.27323680523120036,
+      "grad_norm": 0.8631758008930391,
+      "learning_rate": 4.4145938693011747e-05,
+      "loss": 0.7133,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2738206445586175,
+      "grad_norm": 1.0308474440769033,
+      "learning_rate": 4.4116677309943295e-05,
+      "loss": 0.7185,
+      "step": 2345
+    },
+    {
+      "epoch": 0.27440448388603456,
+      "grad_norm": 0.7570257114511578,
+      "learning_rate": 4.40873539729691e-05,
+      "loss": 0.7278,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2749883232134517,
+      "grad_norm": 0.7901622570063865,
+      "learning_rate": 4.405796879141881e-05,
+      "loss": 0.7064,
+      "step": 2355
+    },
+    {
+      "epoch": 0.27557216254086875,
+      "grad_norm": 0.7788901006471353,
+      "learning_rate": 4.402852187485262e-05,
+      "loss": 0.7224,
+      "step": 2360
+    },
+    {
+      "epoch": 0.27615600186828587,
+      "grad_norm": 0.8123579581408774,
+      "learning_rate": 4.3999013333060936e-05,
+      "loss": 0.7254,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27673984119570294,
+      "grad_norm": 0.8237555662802201,
+      "learning_rate": 4.396944327606389e-05,
+      "loss": 0.7064,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27732368052312006,
+      "grad_norm": 0.7557624979447745,
+      "learning_rate": 4.393981181411102e-05,
+      "loss": 0.7262,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27790751985053713,
+      "grad_norm": 0.7866415313414409,
+      "learning_rate": 4.3910119057680765e-05,
+      "loss": 0.6906,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2784913591779542,
+      "grad_norm": 0.8148376946961301,
+      "learning_rate": 4.3880365117480114e-05,
+      "loss": 0.725,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2790751985053713,
+      "grad_norm": 0.8034343055523648,
+      "learning_rate": 4.385055010444416e-05,
+      "loss": 0.7053,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2796590378327884,
+      "grad_norm": 0.8266078864974,
+      "learning_rate": 4.382067412973573e-05,
+      "loss": 0.7029,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2802428771602055,
+      "grad_norm": 0.7053898192924525,
+      "learning_rate": 4.3790737304744906e-05,
+      "loss": 0.7212,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2808267164876226,
+      "grad_norm": 0.7266265709842735,
+      "learning_rate": 4.376073974108866e-05,
+      "loss": 0.7253,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2814105558150397,
+      "grad_norm": 0.6798592601288477,
+      "learning_rate": 4.373068155061043e-05,
+      "loss": 0.7097,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2819943951424568,
+      "grad_norm": 0.7235954522188496,
+      "learning_rate": 4.37005628453797e-05,
+      "loss": 0.7025,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2825782344698739,
+      "grad_norm": 0.775734945025186,
+      "learning_rate": 4.367038373769155e-05,
+      "loss": 0.7324,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28316207379729097,
+      "grad_norm": 0.770747806314939,
+      "learning_rate": 4.36401443400663e-05,
+      "loss": 0.7285,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2837459131247081,
+      "grad_norm": 0.8291732339761384,
+      "learning_rate": 4.3609844765249034e-05,
+      "loss": 0.6969,
+      "step": 2430
+    },
+    {
+      "epoch": 0.28432975245212516,
+      "grad_norm": 0.8759927257734268,
+      "learning_rate": 4.357948512620922e-05,
+      "loss": 0.7069,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2849135917795423,
+      "grad_norm": 0.9555634569262756,
+      "learning_rate": 4.354906553614024e-05,
+      "loss": 0.7152,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28549743110695935,
+      "grad_norm": 0.7926964374849049,
+      "learning_rate": 4.3518586108459034e-05,
+      "loss": 0.6953,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2860812704343765,
+      "grad_norm": 0.8670597981303715,
+      "learning_rate": 4.34880469568056e-05,
+      "loss": 0.6939,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28666510976179355,
+      "grad_norm": 0.7940656148228097,
+      "learning_rate": 4.345744819504266e-05,
+      "loss": 0.714,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28724894908921067,
+      "grad_norm": 0.7161534778292359,
+      "learning_rate": 4.342678993725517e-05,
+      "loss": 0.7063,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28783278841662774,
+      "grad_norm": 0.7974259411882264,
+      "learning_rate": 4.339607229774989e-05,
+      "loss": 0.7138,
+      "step": 2465
+    },
+    {
+      "epoch": 0.28841662774404486,
+      "grad_norm": 0.7732673861806778,
+      "learning_rate": 4.3365295391054996e-05,
+      "loss": 0.7073,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28900046707146193,
+      "grad_norm": 0.7590336694258303,
+      "learning_rate": 4.333445933191964e-05,
+      "loss": 0.7202,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28958430639887905,
+      "grad_norm": 0.7575975541713273,
+      "learning_rate": 4.330356423531352e-05,
+      "loss": 0.7143,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2901681457262961,
+      "grad_norm": 0.7628200424802012,
+      "learning_rate": 4.327261021642644e-05,
+      "loss": 0.722,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2907519850537132,
+      "grad_norm": 0.7622798779523068,
+      "learning_rate": 4.32415973906679e-05,
+      "loss": 0.7105,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2913358243811303,
+      "grad_norm": 0.8001106062213871,
+      "learning_rate": 4.3210525873666656e-05,
+      "loss": 0.7084,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2919196637085474,
+      "grad_norm": 1.204280692595554,
+      "learning_rate": 4.317939578127029e-05,
+      "loss": 0.7143,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2925035030359645,
+      "grad_norm": 0.9009941077991439,
+      "learning_rate": 4.314820722954476e-05,
+      "loss": 0.7216,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2930873423633816,
+      "grad_norm": 0.8938308491195663,
+      "learning_rate": 4.3116960334774e-05,
+      "loss": 0.712,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2936711816907987,
+      "grad_norm": 0.8599772730180362,
+      "learning_rate": 4.308565521345949e-05,
+      "loss": 0.7203,
+      "step": 2515
+    },
+    {
+      "epoch": 0.29425502101821577,
+      "grad_norm": 2.102049767135387,
+      "learning_rate": 4.305429198231977e-05,
+      "loss": 0.7123,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2948388603456329,
+      "grad_norm": 0.7496640756396827,
+      "learning_rate": 4.302287075829005e-05,
+      "loss": 0.7075,
+      "step": 2525
+    },
+    {
+      "epoch": 0.29542269967304996,
+      "grad_norm": 0.664311795120198,
+      "learning_rate": 4.2991391658521765e-05,
+      "loss": 0.7018,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2960065390004671,
+      "grad_norm": 0.805681195205423,
+      "learning_rate": 4.2959854800382136e-05,
+      "loss": 0.7075,
+      "step": 2535
+    },
+    {
+      "epoch": 0.29659037832788415,
+      "grad_norm": 0.76027156646289,
+      "learning_rate": 4.292826030145372e-05,
+      "loss": 0.7134,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2971742176553013,
+      "grad_norm": 0.8764032798835537,
+      "learning_rate": 4.289660827953399e-05,
+      "loss": 0.6904,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29775805698271834,
+      "grad_norm": 0.721305727819197,
+      "learning_rate": 4.28648988526349e-05,
+      "loss": 0.7283,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29834189631013547,
+      "grad_norm": 19.423448089121035,
+      "learning_rate": 4.2833132138982415e-05,
+      "loss": 0.7238,
+      "step": 2555
+    },
+    {
+      "epoch": 0.29892573563755254,
+      "grad_norm": 0.7950100019932609,
+      "learning_rate": 4.280130825701609e-05,
+      "loss": 0.7136,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29950957496496966,
+      "grad_norm": 0.733815093607993,
+      "learning_rate": 4.276942732538866e-05,
+      "loss": 0.7032,
+      "step": 2565
+    },
+    {
+      "epoch": 0.30009341429238673,
+      "grad_norm": 0.7970021565058928,
+      "learning_rate": 4.273748946296552e-05,
+      "loss": 0.737,
+      "step": 2570
+    },
+    {
+      "epoch": 0.30067725361980385,
+      "grad_norm": 0.8628432834272957,
+      "learning_rate": 4.2705494788824345e-05,
+      "loss": 0.7149,
+      "step": 2575
+    },
+    {
+      "epoch": 0.3012610929472209,
+      "grad_norm": 0.7834419896434103,
+      "learning_rate": 4.267344342225463e-05,
+      "loss": 0.7256,
+      "step": 2580
+    },
+    {
+      "epoch": 0.30184493227463804,
+      "grad_norm": 0.7595038379537414,
+      "learning_rate": 4.264133548275725e-05,
+      "loss": 0.7034,
+      "step": 2585
+    },
+    {
+      "epoch": 0.3024287716020551,
+      "grad_norm": 0.7610589658032497,
+      "learning_rate": 4.2609171090044e-05,
+      "loss": 0.6928,
+      "step": 2590
+    },
+    {
+      "epoch": 0.30301261092947224,
+      "grad_norm": 0.7343922486197118,
+      "learning_rate": 4.257695036403714e-05,
+      "loss": 0.7057,
+      "step": 2595
+    },
+    {
+      "epoch": 0.3035964502568893,
+      "grad_norm": 0.7102829296413985,
+      "learning_rate": 4.2544673424868994e-05,
+      "loss": 0.7031,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3041802895843064,
+      "grad_norm": 0.774479875646621,
+      "learning_rate": 4.251234039288145e-05,
+      "loss": 0.7086,
+      "step": 2605
+    },
+    {
+      "epoch": 0.3047641289117235,
+      "grad_norm": 0.800845583870741,
+      "learning_rate": 4.2479951388625546e-05,
+      "loss": 0.7312,
+      "step": 2610
+    },
+    {
+      "epoch": 0.30534796823914057,
+      "grad_norm": 0.7493080904949945,
+      "learning_rate": 4.2447506532861e-05,
+      "loss": 0.7276,
+      "step": 2615
+    },
+    {
+      "epoch": 0.3059318075665577,
+      "grad_norm": 0.8127313007347643,
+      "learning_rate": 4.241500594655577e-05,
+      "loss": 0.7069,
+      "step": 2620
+    },
+    {
+      "epoch": 0.30651564689397476,
+      "grad_norm": 0.8277530192656021,
+      "learning_rate": 4.2382449750885604e-05,
+      "loss": 0.6874,
+      "step": 2625
+    },
+    {
+      "epoch": 0.3070994862213919,
+      "grad_norm": 0.7963199104898718,
+      "learning_rate": 4.2349838067233575e-05,
+      "loss": 0.6903,
+      "step": 2630
+    },
+    {
+      "epoch": 0.30768332554880895,
+      "grad_norm": 0.8146094866973611,
+      "learning_rate": 4.231717101718967e-05,
+      "loss": 0.6871,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3082671648762261,
+      "grad_norm": 0.7062438712561713,
+      "learning_rate": 4.228444872255025e-05,
+      "loss": 0.7,
+      "step": 2640
+    },
+    {
+      "epoch": 0.30885100420364314,
+      "grad_norm": 0.7536796043688521,
+      "learning_rate": 4.2251671305317696e-05,
+      "loss": 0.7114,
+      "step": 2645
+    },
+    {
+      "epoch": 0.30943484353106027,
+      "grad_norm": 0.7327638261755937,
+      "learning_rate": 4.2218838887699894e-05,
+      "loss": 0.7166,
+      "step": 2650
+    },
+    {
+      "epoch": 0.31001868285847733,
+      "grad_norm": 0.9041359970185278,
+      "learning_rate": 4.2185951592109794e-05,
+      "loss": 0.7,
+      "step": 2655
+    },
+    {
+      "epoch": 0.31060252218589446,
+      "grad_norm": 0.8351437587118097,
+      "learning_rate": 4.2153009541164965e-05,
+      "loss": 0.7162,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3111863615133115,
+      "grad_norm": 0.8334803832595723,
+      "learning_rate": 4.21200128576871e-05,
+      "loss": 0.7272,
+      "step": 2665
+    },
+    {
+      "epoch": 0.31177020084072865,
+      "grad_norm": 0.7645748557972607,
+      "learning_rate": 4.208696166470161e-05,
+      "loss": 0.7203,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3123540401681457,
+      "grad_norm": 0.8010226975358403,
+      "learning_rate": 4.2053856085437124e-05,
+      "loss": 0.7127,
+      "step": 2675
+    },
+    {
+      "epoch": 0.31293787949556284,
+      "grad_norm": 0.778105181138758,
+      "learning_rate": 4.202069624332507e-05,
+      "loss": 0.7093,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3135217188229799,
+      "grad_norm": 0.8269537229295518,
+      "learning_rate": 4.1987482261999164e-05,
+      "loss": 0.7256,
+      "step": 2685
+    },
+    {
+      "epoch": 0.31410555815039704,
+      "grad_norm": 0.7273731295392446,
+      "learning_rate": 4.1954214265294985e-05,
+      "loss": 0.6877,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3146893974778141,
+      "grad_norm": 0.6387268961733149,
+      "learning_rate": 4.192089237724951e-05,
+      "loss": 0.6936,
+      "step": 2695
+    },
+    {
+      "epoch": 0.3152732368052312,
+      "grad_norm": 0.7135748637677033,
+      "learning_rate": 4.188751672210063e-05,
+      "loss": 0.7118,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3158570761326483,
+      "grad_norm": 0.7798832279992028,
+      "learning_rate": 4.1854087424286725e-05,
+      "loss": 0.6884,
+      "step": 2705
+    },
+    {
+      "epoch": 0.31644091546006536,
+      "grad_norm": 0.9679464940606846,
+      "learning_rate": 4.182060460844615e-05,
+      "loss": 0.6979,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3170247547874825,
+      "grad_norm": 0.796914375651688,
+      "learning_rate": 4.1787068399416825e-05,
+      "loss": 0.7012,
+      "step": 2715
+    },
+    {
+      "epoch": 0.31760859411489956,
+      "grad_norm": 0.761095764820495,
+      "learning_rate": 4.175347892223572e-05,
+      "loss": 0.6793,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3181924334423167,
+      "grad_norm": 0.7258359028766036,
+      "learning_rate": 4.1719836302138426e-05,
+      "loss": 0.6994,
+      "step": 2725
+    },
+    {
+      "epoch": 0.31877627276973375,
+      "grad_norm": 0.8162191545953326,
+      "learning_rate": 4.168614066455867e-05,
+      "loss": 0.7015,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31936011209715087,
+      "grad_norm": 0.8218062352011269,
+      "learning_rate": 4.165239213512784e-05,
+      "loss": 0.6949,
+      "step": 2735
+    },
+    {
+      "epoch": 0.31994395142456794,
+      "grad_norm": 0.7739752042429793,
+      "learning_rate": 4.161859083967454e-05,
+      "loss": 0.7024,
+      "step": 2740
+    },
+    {
+      "epoch": 0.32052779075198506,
+      "grad_norm": 0.7856375360797179,
+      "learning_rate": 4.158473690422409e-05,
+      "loss": 0.6931,
+      "step": 2745
+    },
+    {
+      "epoch": 0.32111163007940213,
+      "grad_norm": 0.767076262950083,
+      "learning_rate": 4.15508304549981e-05,
+      "loss": 0.6763,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32169546940681926,
+      "grad_norm": 0.8186303246875027,
+      "learning_rate": 4.1516871618413947e-05,
+      "loss": 0.7077,
+      "step": 2755
+    },
+    {
+      "epoch": 0.3222793087342363,
+      "grad_norm": 0.7566057113699275,
+      "learning_rate": 4.148286052108436e-05,
+      "loss": 0.7095,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32286314806165345,
+      "grad_norm": 0.685642900616039,
+      "learning_rate": 4.144879728981688e-05,
+      "loss": 0.7056,
+      "step": 2765
+    },
+    {
+      "epoch": 0.3234469873890705,
+      "grad_norm": 0.7506545072890666,
+      "learning_rate": 4.141468205161345e-05,
+      "loss": 0.6991,
+      "step": 2770
+    },
+    {
+      "epoch": 0.32403082671648764,
+      "grad_norm": 0.705634756504342,
+      "learning_rate": 4.1380514933669916e-05,
+      "loss": 0.6777,
+      "step": 2775
+    },
+    {
+      "epoch": 0.3246146660439047,
+      "grad_norm": 0.7386898729822018,
+      "learning_rate": 4.134629606337555e-05,
+      "loss": 0.692,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32519850537132183,
+      "grad_norm": 0.7914830335645638,
+      "learning_rate": 4.131202556831257e-05,
+      "loss": 0.7072,
+      "step": 2785
+    },
+    {
+      "epoch": 0.3257823446987389,
+      "grad_norm": 0.7822067236686193,
+      "learning_rate": 4.1277703576255685e-05,
+      "loss": 0.692,
+      "step": 2790
+    },
+    {
+      "epoch": 0.326366184026156,
+      "grad_norm": 0.7781485290409217,
+      "learning_rate": 4.12433302151716e-05,
+      "loss": 0.6948,
+      "step": 2795
+    },
+    {
+      "epoch": 0.3269500233535731,
+      "grad_norm": 0.6879972141894926,
+      "learning_rate": 4.1208905613218547e-05,
+      "loss": 0.7009,
+      "step": 2800
+    },
+    {
+      "epoch": 0.3275338626809902,
+      "grad_norm": 0.7101221257232809,
+      "learning_rate": 4.1174429898745795e-05,
+      "loss": 0.6856,
+      "step": 2805
+    },
+    {
+      "epoch": 0.3281177020084073,
+      "grad_norm": 0.731271995376911,
+      "learning_rate": 4.113990320029321e-05,
+      "loss": 0.6805,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32870154133582435,
+      "grad_norm": 0.7196696016295292,
+      "learning_rate": 4.1105325646590714e-05,
+      "loss": 0.7144,
+      "step": 2815
+    },
+    {
+      "epoch": 0.3292853806632415,
+      "grad_norm": 0.7149670007135108,
+      "learning_rate": 4.1070697366557856e-05,
+      "loss": 0.6956,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32986921999065855,
+      "grad_norm": 0.8184512447679282,
+      "learning_rate": 4.103601848930332e-05,
+      "loss": 0.6878,
+      "step": 2825
+    },
+    {
+      "epoch": 0.33045305931807567,
+      "grad_norm": 0.8345056924813619,
+      "learning_rate": 4.100128914412442e-05,
+      "loss": 0.7327,
+      "step": 2830
+    },
+    {
+      "epoch": 0.33103689864549274,
+      "grad_norm": 0.8215005253430337,
+      "learning_rate": 4.096650946050664e-05,
+      "loss": 0.6886,
+      "step": 2835
+    },
+    {
+      "epoch": 0.33162073797290986,
+      "grad_norm": 0.7963405506325654,
+      "learning_rate": 4.0931679568123174e-05,
+      "loss": 0.7143,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33220457730032693,
+      "grad_norm": 0.753898680551099,
+      "learning_rate": 4.0896799596834365e-05,
+      "loss": 0.6857,
+      "step": 2845
+    },
+    {
+      "epoch": 0.33278841662774405,
+      "grad_norm": 0.8495093118519794,
+      "learning_rate": 4.086186967668731e-05,
+      "loss": 0.7166,
+      "step": 2850
+    },
+    {
+      "epoch": 0.3333722559551611,
+      "grad_norm": 0.8162454411979833,
+      "learning_rate": 4.082688993791533e-05,
+      "loss": 0.674,
+      "step": 2855
+    },
+    {
+      "epoch": 0.33395609528257825,
+      "grad_norm": 1.043859849113058,
+      "learning_rate": 4.079186051093747e-05,
+      "loss": 0.6954,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3345399346099953,
+      "grad_norm": 0.7987678758616251,
+      "learning_rate": 4.075678152635807e-05,
+      "loss": 0.7152,
+      "step": 2865
+    },
+    {
+      "epoch": 0.33512377393741244,
+      "grad_norm": 0.7413546464797619,
+      "learning_rate": 4.07216531149662e-05,
+      "loss": 0.7196,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3357076132648295,
+      "grad_norm": 0.7678397674426483,
+      "learning_rate": 4.068647540773524e-05,
+      "loss": 0.7076,
+      "step": 2875
+    },
+    {
+      "epoch": 0.33629145259224663,
+      "grad_norm": 0.7752892357655767,
+      "learning_rate": 4.065124853582237e-05,
+      "loss": 0.7129,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3368752919196637,
+      "grad_norm": 0.741984984049898,
+      "learning_rate": 4.0615972630568055e-05,
+      "loss": 0.7047,
+      "step": 2885
+    },
+    {
+      "epoch": 0.3374591312470808,
+      "grad_norm": 0.796527501085706,
+      "learning_rate": 4.0580647823495587e-05,
+      "loss": 0.7061,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3380429705744979,
+      "grad_norm": 0.7365729327389902,
+      "learning_rate": 4.054527424631059e-05,
+      "loss": 0.7329,
+      "step": 2895
+    },
+    {
+      "epoch": 0.338626809901915,
+      "grad_norm": 0.7723495813661937,
+      "learning_rate": 4.0509852030900506e-05,
+      "loss": 0.678,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3392106492293321,
+      "grad_norm": 0.7655325772084528,
+      "learning_rate": 4.047438130933415e-05,
+      "loss": 0.7041,
+      "step": 2905
+    },
+    {
+      "epoch": 0.3397944885567492,
+      "grad_norm": 0.8181753674636343,
+      "learning_rate": 4.043886221386117e-05,
+      "loss": 0.6914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3403783278841663,
+      "grad_norm": 0.7221488554746541,
+      "learning_rate": 4.040329487691155e-05,
+      "loss": 0.6992,
+      "step": 2915
+    },
+    {
+      "epoch": 0.34096216721158334,
+      "grad_norm": 0.9106238776241381,
+      "learning_rate": 4.036767943109519e-05,
+      "loss": 0.7062,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34154600653900047,
+      "grad_norm": 0.8667028137565969,
+      "learning_rate": 4.0332016009201315e-05,
+      "loss": 0.7055,
+      "step": 2925
+    },
+    {
+      "epoch": 0.34212984586641754,
+      "grad_norm": 0.7285032338549546,
+      "learning_rate": 4.0296304744198045e-05,
+      "loss": 0.6967,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34271368519383466,
+      "grad_norm": 0.7111450740341209,
+      "learning_rate": 4.0260545769231875e-05,
+      "loss": 0.6921,
+      "step": 2935
+    },
+    {
+      "epoch": 0.34329752452125173,
+      "grad_norm": 0.8082397092555128,
+      "learning_rate": 4.022473921762719e-05,
+      "loss": 0.7086,
+      "step": 2940
+    },
+    {
+      "epoch": 0.34388136384866885,
+      "grad_norm": 0.7588740994953772,
+      "learning_rate": 4.018888522288574e-05,
+      "loss": 0.714,
+      "step": 2945
+    },
+    {
+      "epoch": 0.3444652031760859,
+      "grad_norm": 0.757055884032176,
+      "learning_rate": 4.0152983918686175e-05,
+      "loss": 0.6955,
+      "step": 2950
+    },
+    {
+      "epoch": 0.34504904250350305,
+      "grad_norm": 0.7543938506218743,
+      "learning_rate": 4.0117035438883553e-05,
+      "loss": 0.6826,
+      "step": 2955
+    },
+    {
+      "epoch": 0.3456328818309201,
+      "grad_norm": 0.8522266982138246,
+      "learning_rate": 4.008103991750879e-05,
+      "loss": 0.7168,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34621672115833724,
+      "grad_norm": 0.6940565372247784,
+      "learning_rate": 4.004499748876821e-05,
+      "loss": 0.696,
+      "step": 2965
+    },
+    {
+      "epoch": 0.3468005604857543,
+      "grad_norm": 0.6862029386417299,
+      "learning_rate": 4.000890828704304e-05,
+      "loss": 0.6824,
+      "step": 2970
+    },
+    {
+      "epoch": 0.34738439981317143,
+      "grad_norm": 0.7315368280765282,
+      "learning_rate": 3.997277244688886e-05,
+      "loss": 0.7093,
+      "step": 2975
+    },
+    {
+      "epoch": 0.3479682391405885,
+      "grad_norm": 0.8399426230955965,
+      "learning_rate": 3.993659010303517e-05,
+      "loss": 0.7088,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3485520784680056,
+      "grad_norm": 0.7447428590500363,
+      "learning_rate": 3.9900361390384836e-05,
+      "loss": 0.6729,
+      "step": 2985
+    },
+    {
+      "epoch": 0.3491359177954227,
+      "grad_norm": 0.6894785309334255,
+      "learning_rate": 3.986408644401362e-05,
+      "loss": 0.7095,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3497197571228398,
+      "grad_norm": 0.7895286945877132,
+      "learning_rate": 3.982776539916966e-05,
+      "loss": 0.6963,
+      "step": 2995
+    },
+    {
+      "epoch": 0.3503035964502569,
+      "grad_norm": 0.7542557790149979,
+      "learning_rate": 3.979139839127296e-05,
+      "loss": 0.6805,
+      "step": 3000
+    },
+    {
+      "epoch": 0.350887435777674,
+      "grad_norm": 0.7985071306992715,
+      "learning_rate": 3.975498555591489e-05,
+      "loss": 0.6959,
+      "step": 3005
+    },
+    {
+      "epoch": 0.3514712751050911,
+      "grad_norm": 0.7109389878914494,
+      "learning_rate": 3.971852702885772e-05,
+      "loss": 0.6878,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3520551144325082,
+      "grad_norm": 0.7031748076507193,
+      "learning_rate": 3.9682022946034006e-05,
+      "loss": 0.6931,
+      "step": 3015
+    },
+    {
+      "epoch": 0.35263895375992527,
+      "grad_norm": 0.7925651121254325,
+      "learning_rate": 3.964547344354624e-05,
+      "loss": 0.697,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3532227930873424,
+      "grad_norm": 0.7244597774458819,
+      "learning_rate": 3.9608878657666195e-05,
+      "loss": 0.7022,
+      "step": 3025
+    },
+    {
+      "epoch": 0.35380663241475946,
+      "grad_norm": 0.7581238408555705,
+      "learning_rate": 3.9572238724834503e-05,
+      "loss": 0.7025,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3543904717421765,
+      "grad_norm": 0.7831980832064683,
+      "learning_rate": 3.953555378166012e-05,
+      "loss": 0.7033,
+      "step": 3035
+    },
+    {
+      "epoch": 0.35497431106959365,
+      "grad_norm": 0.8547349083576906,
+      "learning_rate": 3.9498823964919827e-05,
+      "loss": 0.7058,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3555581503970107,
+      "grad_norm": 0.7440878870880225,
+      "learning_rate": 3.94620494115577e-05,
+      "loss": 0.6895,
+      "step": 3045
+    },
+    {
+      "epoch": 0.35614198972442784,
+      "grad_norm": 0.6709320077587175,
+      "learning_rate": 3.942523025868461e-05,
+      "loss": 0.6934,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3567258290518449,
+      "grad_norm": 0.8342619568452645,
+      "learning_rate": 3.9388366643577745e-05,
+      "loss": 0.6912,
+      "step": 3055
+    },
+    {
+      "epoch": 0.35730966837926204,
+      "grad_norm": 0.7265224003909866,
+      "learning_rate": 3.9351458703680017e-05,
+      "loss": 0.7109,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3578935077066791,
+      "grad_norm": 0.7513155570253139,
+      "learning_rate": 3.931450657659963e-05,
+      "loss": 0.6973,
+      "step": 3065
+    },
+    {
+      "epoch": 0.3584773470340962,
+      "grad_norm": 0.7609685128527147,
+      "learning_rate": 3.927751040010954e-05,
+      "loss": 0.6772,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3590611863615133,
+      "grad_norm": 0.898794715325684,
+      "learning_rate": 3.924047031214691e-05,
+      "loss": 0.71,
+      "step": 3075
+    },
+    {
+      "epoch": 0.3596450256889304,
+      "grad_norm": 0.7576048114562289,
+      "learning_rate": 3.920338645081266e-05,
+      "loss": 0.6958,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3602288650163475,
+      "grad_norm": 0.7682741662829687,
+      "learning_rate": 3.916625895437089e-05,
+      "loss": 0.6701,
+      "step": 3085
+    },
+    {
+      "epoch": 0.3608127043437646,
+      "grad_norm": 0.8781282581181049,
+      "learning_rate": 3.912908796124839e-05,
+      "loss": 0.6979,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3613965436711817,
+      "grad_norm": 0.7855289086951455,
+      "learning_rate": 3.909187361003414e-05,
+      "loss": 0.6866,
+      "step": 3095
+    },
+    {
+      "epoch": 0.3619803829985988,
+      "grad_norm": 0.7648549467063727,
+      "learning_rate": 3.905461603947878e-05,
+      "loss": 0.7006,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3625642223260159,
+      "grad_norm": 0.792581012808363,
+      "learning_rate": 3.9017315388494044e-05,
+      "loss": 0.6871,
+      "step": 3105
+    },
+    {
+      "epoch": 0.363148061653433,
+      "grad_norm": 0.7754578864170861,
+      "learning_rate": 3.8979971796152346e-05,
+      "loss": 0.6933,
+      "step": 3110
+    },
+    {
+      "epoch": 0.36373190098085006,
+      "grad_norm": 0.6662434046820936,
+      "learning_rate": 3.894258540168618e-05,
+      "loss": 0.6746,
+      "step": 3115
+    },
+    {
+      "epoch": 0.3643157403082672,
+      "grad_norm": 0.7253242083521453,
+      "learning_rate": 3.89051563444876e-05,
+      "loss": 0.6802,
+      "step": 3120
+    },
+    {
+      "epoch": 0.36489957963568426,
+      "grad_norm": 0.6650194558269369,
+      "learning_rate": 3.886768476410777e-05,
+      "loss": 0.6848,
+      "step": 3125
+    },
+    {
+      "epoch": 0.3654834189631014,
+      "grad_norm": 0.8598275401566264,
+      "learning_rate": 3.883017080025638e-05,
+      "loss": 0.6858,
+      "step": 3130
+    },
+    {
+      "epoch": 0.36606725829051845,
+      "grad_norm": 0.8053667856247443,
+      "learning_rate": 3.879261459280111e-05,
+      "loss": 0.6885,
+      "step": 3135
+    },
+    {
+      "epoch": 0.3666510976179355,
+      "grad_norm": 0.7702969353671916,
+      "learning_rate": 3.875501628176719e-05,
+      "loss": 0.6956,
+      "step": 3140
+    },
+    {
+      "epoch": 0.36723493694535264,
+      "grad_norm": 0.7722892835874596,
+      "learning_rate": 3.8717376007336814e-05,
+      "loss": 0.6898,
+      "step": 3145
+    },
+    {
+      "epoch": 0.3678187762727697,
+      "grad_norm": 0.784670074007829,
+      "learning_rate": 3.867969390984862e-05,
+      "loss": 0.7156,
+      "step": 3150
+    },
+    {
+      "epoch": 0.36840261560018683,
+      "grad_norm": 0.745287207769695,
+      "learning_rate": 3.864197012979719e-05,
+      "loss": 0.7175,
+      "step": 3155
+    },
+    {
+      "epoch": 0.3689864549276039,
+      "grad_norm": 0.8145729590116079,
+      "learning_rate": 3.8604204807832516e-05,
+      "loss": 0.7291,
+      "step": 3160
+    },
+    {
+      "epoch": 0.369570294255021,
+      "grad_norm": 0.743613056899968,
+      "learning_rate": 3.856639808475947e-05,
+      "loss": 0.6812,
+      "step": 3165
+    },
+    {
+      "epoch": 0.3701541335824381,
+      "grad_norm": 0.7118971478527887,
+      "learning_rate": 3.85285501015373e-05,
+      "loss": 0.6914,
+      "step": 3170
+    },
+    {
+      "epoch": 0.3707379729098552,
+      "grad_norm": 0.7086191684890616,
+      "learning_rate": 3.8490660999279074e-05,
+      "loss": 0.6884,
+      "step": 3175
+    },
+    {
+      "epoch": 0.3713218122372723,
+      "grad_norm": 0.6696579150369846,
+      "learning_rate": 3.8452730919251174e-05,
+      "loss": 0.6986,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3719056515646894,
+      "grad_norm": 0.6900937913313506,
+      "learning_rate": 3.841476000287275e-05,
+      "loss": 0.7078,
+      "step": 3185
+    },
+    {
+      "epoch": 0.3724894908921065,
+      "grad_norm": 0.7853924347903385,
+      "learning_rate": 3.837674839171524e-05,
+      "loss": 0.6879,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3730733302195236,
+      "grad_norm": 0.7312106575367651,
+      "learning_rate": 3.833869622750177e-05,
+      "loss": 0.7089,
+      "step": 3195
+    },
+    {
+      "epoch": 0.37365716954694067,
+      "grad_norm": 0.9065474676918042,
+      "learning_rate": 3.8300603652106704e-05,
+      "loss": 0.6972,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3742410088743578,
+      "grad_norm": 0.8566499866250796,
+      "learning_rate": 3.8262470807555045e-05,
+      "loss": 0.6824,
+      "step": 3205
+    },
+    {
+      "epoch": 0.37482484820177486,
+      "grad_norm": 0.7687481924225721,
+      "learning_rate": 3.822429783602195e-05,
+      "loss": 0.7017,
+      "step": 3210
+    },
+    {
+      "epoch": 0.375408687529192,
+      "grad_norm": 0.8140836228945466,
+      "learning_rate": 3.818608487983218e-05,
+      "loss": 0.6792,
+      "step": 3215
+    },
+    {
+      "epoch": 0.37599252685660906,
+      "grad_norm": 0.733345067405923,
+      "learning_rate": 3.8147832081459574e-05,
+      "loss": 0.6881,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3765763661840262,
+      "grad_norm": 0.7744637883489761,
+      "learning_rate": 3.810953958352653e-05,
+      "loss": 0.7123,
+      "step": 3225
+    },
+    {
+      "epoch": 0.37716020551144325,
+      "grad_norm": 0.708946933223857,
+      "learning_rate": 3.807120752880346e-05,
+      "loss": 0.6907,
+      "step": 3230
+    },
+    {
+      "epoch": 0.37774404483886037,
+      "grad_norm": 0.7041806485060429,
+      "learning_rate": 3.8032836060208265e-05,
+      "loss": 0.6783,
+      "step": 3235
+    },
+    {
+      "epoch": 0.37832788416627744,
+      "grad_norm": 0.7341040187993833,
+      "learning_rate": 3.799442532080577e-05,
+      "loss": 0.6811,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3789117234936945,
+      "grad_norm": 0.8543179231859879,
+      "learning_rate": 3.795597545380724e-05,
+      "loss": 0.6942,
+      "step": 3245
+    },
+    {
+      "epoch": 0.37949556282111163,
+      "grad_norm": 0.8149701441077691,
+      "learning_rate": 3.791748660256983e-05,
+      "loss": 0.6913,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3800794021485287,
+      "grad_norm": 0.75207798303696,
+      "learning_rate": 3.787895891059603e-05,
+      "loss": 0.6746,
+      "step": 3255
+    },
+    {
+      "epoch": 0.3806632414759458,
+      "grad_norm": 0.7703254571890686,
+      "learning_rate": 3.784039252153315e-05,
+      "loss": 0.685,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3812470808033629,
+      "grad_norm": 0.8155529508918891,
+      "learning_rate": 3.780178757917278e-05,
+      "loss": 0.7064,
+      "step": 3265
+    },
+    {
+      "epoch": 0.38183092013078,
+      "grad_norm": 0.675555849154923,
+      "learning_rate": 3.776314422745026e-05,
+      "loss": 0.6764,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3824147594581971,
+      "grad_norm": 0.8278771662993144,
+      "learning_rate": 3.772446261044411e-05,
+      "loss": 0.6873,
+      "step": 3275
+    },
+    {
+      "epoch": 0.3829985987856142,
+      "grad_norm": 0.7773386206193695,
+      "learning_rate": 3.768574287237555e-05,
+      "loss": 0.6847,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3835824381130313,
+      "grad_norm": 0.8031577269955347,
+      "learning_rate": 3.7646985157607915e-05,
+      "loss": 0.692,
+      "step": 3285
+    },
+    {
+      "epoch": 0.3841662774404484,
+      "grad_norm": 0.8218626150048584,
+      "learning_rate": 3.760818961064614e-05,
+      "loss": 0.6854,
+      "step": 3290
+    },
+    {
+      "epoch": 0.38475011676786547,
+      "grad_norm": 0.930027350274466,
+      "learning_rate": 3.75693563761362e-05,
+      "loss": 0.6914,
+      "step": 3295
+    },
+    {
+      "epoch": 0.3853339560952826,
+      "grad_norm": 0.7818858106760599,
+      "learning_rate": 3.75304855988646e-05,
+      "loss": 0.703,
+      "step": 3300
+    },
+    {
+      "epoch": 0.38591779542269966,
+      "grad_norm": 0.8740352987061476,
+      "learning_rate": 3.749157742375782e-05,
+      "loss": 0.6992,
+      "step": 3305
+    },
+    {
+      "epoch": 0.3865016347501168,
+      "grad_norm": 0.802111462845047,
+      "learning_rate": 3.745263199588176e-05,
+      "loss": 0.6957,
+      "step": 3310
+    },
+    {
+      "epoch": 0.38708547407753385,
+      "grad_norm": 0.8587440611545933,
+      "learning_rate": 3.741364946044123e-05,
+      "loss": 0.6907,
+      "step": 3315
+    },
+    {
+      "epoch": 0.387669313404951,
+      "grad_norm": 0.6825891536044698,
+      "learning_rate": 3.737462996277939e-05,
+      "loss": 0.688,
+      "step": 3320
+    },
+    {
+      "epoch": 0.38825315273236805,
+      "grad_norm": 0.7500018133835387,
+      "learning_rate": 3.73355736483772e-05,
+      "loss": 0.7262,
+      "step": 3325
+    },
+    {
+      "epoch": 0.38883699205978517,
+      "grad_norm": 0.7995432384520971,
+      "learning_rate": 3.72964806628529e-05,
+      "loss": 0.681,
+      "step": 3330
+    },
+    {
+      "epoch": 0.38942083138720224,
+      "grad_norm": 0.7595096108564414,
+      "learning_rate": 3.725735115196145e-05,
+      "loss": 0.7099,
+      "step": 3335
+    },
+    {
+      "epoch": 0.39000467071461936,
+      "grad_norm": 0.7440639912876408,
+      "learning_rate": 3.7218185261593984e-05,
+      "loss": 0.6958,
+      "step": 3340
+    },
+    {
+      "epoch": 0.39058851004203643,
+      "grad_norm": 0.7373854715092809,
+      "learning_rate": 3.717898313777729e-05,
+      "loss": 0.7001,
+      "step": 3345
+    },
+    {
+      "epoch": 0.39117234936945355,
+      "grad_norm": 0.7613559077274907,
+      "learning_rate": 3.713974492667324e-05,
+      "loss": 0.698,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3917561886968706,
+      "grad_norm": 0.7015801575619542,
+      "learning_rate": 3.710047077457826e-05,
+      "loss": 0.6907,
+      "step": 3355
+    },
+    {
+      "epoch": 0.3923400280242877,
+      "grad_norm": 0.713928801821472,
+      "learning_rate": 3.706116082792276e-05,
+      "loss": 0.6829,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3929238673517048,
+      "grad_norm": 0.7154651053309964,
+      "learning_rate": 3.702181523327064e-05,
+      "loss": 0.7117,
+      "step": 3365
+    },
+    {
+      "epoch": 0.3935077066791219,
+      "grad_norm": 0.7575843994242758,
+      "learning_rate": 3.698243413731867e-05,
+      "loss": 0.6882,
+      "step": 3370
+    },
+    {
+      "epoch": 0.394091546006539,
+      "grad_norm": 0.8048906650418256,
+      "learning_rate": 3.694301768689603e-05,
+      "loss": 0.6939,
+      "step": 3375
+    },
+    {
+      "epoch": 0.3946753853339561,
+      "grad_norm": 0.801214633355065,
+      "learning_rate": 3.690356602896368e-05,
+      "loss": 0.7002,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3952592246613732,
+      "grad_norm": 0.8806970202674996,
+      "learning_rate": 3.686407931061386e-05,
+      "loss": 0.6787,
+      "step": 3385
+    },
+    {
+      "epoch": 0.39584306398879027,
+      "grad_norm": 0.9876675473391946,
+      "learning_rate": 3.682455767906954e-05,
+      "loss": 0.6949,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3964269033162074,
+      "grad_norm": 0.8131617221350074,
+      "learning_rate": 3.678500128168384e-05,
+      "loss": 0.6765,
+      "step": 3395
+    },
+    {
+      "epoch": 0.39701074264362446,
+      "grad_norm": 0.807341108618566,
+      "learning_rate": 3.674541026593952e-05,
+      "loss": 0.6889,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3975945819710416,
+      "grad_norm": 0.7925128757565136,
+      "learning_rate": 3.6705784779448405e-05,
+      "loss": 0.6795,
+      "step": 3405
+    },
+    {
+      "epoch": 0.39817842129845865,
+      "grad_norm": 0.7606764204522857,
+      "learning_rate": 3.6666124969950835e-05,
+      "loss": 0.6851,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3987622606258758,
+      "grad_norm": 0.6939460834772246,
+      "learning_rate": 3.662643098531513e-05,
+      "loss": 0.6965,
+      "step": 3415
+    },
+    {
+      "epoch": 0.39934609995329284,
+      "grad_norm": 0.8919480095390583,
+      "learning_rate": 3.6586702973537025e-05,
+      "loss": 0.6941,
+      "step": 3420
+    },
+    {
+      "epoch": 0.39992993928070997,
+      "grad_norm": 0.7608789797993931,
+      "learning_rate": 3.654694108273912e-05,
+      "loss": 0.672,
+      "step": 3425
+    },
+    {
+      "epoch": 0.40051377860812704,
+      "grad_norm": 0.7061952122457605,
+      "learning_rate": 3.6507145461170345e-05,
+      "loss": 0.6676,
+      "step": 3430
+    },
+    {
+      "epoch": 0.40109761793554416,
+      "grad_norm": 0.8011601371400675,
+      "learning_rate": 3.646731625720537e-05,
+      "loss": 0.673,
+      "step": 3435
+    },
+    {
+      "epoch": 0.40168145726296123,
+      "grad_norm": 0.8179789066041869,
+      "learning_rate": 3.642745361934408e-05,
+      "loss": 0.6968,
+      "step": 3440
+    },
+    {
+      "epoch": 0.40226529659037835,
+      "grad_norm": 0.8681890988847925,
+      "learning_rate": 3.638755769621104e-05,
+      "loss": 0.6962,
+      "step": 3445
+    },
+    {
+      "epoch": 0.4028491359177954,
+      "grad_norm": 0.7522001832365489,
+      "learning_rate": 3.634762863655487e-05,
+      "loss": 0.6842,
+      "step": 3450
+    },
+    {
+      "epoch": 0.40343297524521254,
+      "grad_norm": 0.9070294968572723,
+      "learning_rate": 3.630766658924779e-05,
+      "loss": 0.7003,
+      "step": 3455
+    },
+    {
+      "epoch": 0.4040168145726296,
+      "grad_norm": 0.7797395542657742,
+      "learning_rate": 3.6267671703284963e-05,
+      "loss": 0.6962,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4046006539000467,
+      "grad_norm": 0.9737871913004269,
+      "learning_rate": 3.6227644127784026e-05,
+      "loss": 0.701,
+      "step": 3465
+    },
+    {
+      "epoch": 0.4051844932274638,
+      "grad_norm": 0.8142196065459809,
+      "learning_rate": 3.618758401198447e-05,
+      "loss": 0.6851,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4057683325548809,
+      "grad_norm": 0.9056270042476365,
+      "learning_rate": 3.6147491505247124e-05,
+      "loss": 0.6925,
+      "step": 3475
+    },
+    {
+      "epoch": 0.406352171882298,
+      "grad_norm": 0.7315884297176146,
+      "learning_rate": 3.6107366757053586e-05,
+      "loss": 0.6938,
+      "step": 3480
+    },
+    {
+      "epoch": 0.40693601120971507,
+      "grad_norm": 0.8078588240055888,
+      "learning_rate": 3.606720991700565e-05,
+      "loss": 0.7011,
+      "step": 3485
+    },
+    {
+      "epoch": 0.4075198505371322,
+      "grad_norm": 0.7254179613186337,
+      "learning_rate": 3.602702113482477e-05,
+      "loss": 0.6944,
+      "step": 3490
+    },
+    {
+      "epoch": 0.40810368986454926,
+      "grad_norm": 0.7709082948059535,
+      "learning_rate": 3.59868005603515e-05,
+      "loss": 0.7007,
+      "step": 3495
+    },
+    {
+      "epoch": 0.4086875291919664,
+      "grad_norm": 0.7115262527043823,
+      "learning_rate": 3.5946548343544925e-05,
+      "loss": 0.6782,
+      "step": 3500
+    },
+    {
+      "epoch": 0.40927136851938345,
+      "grad_norm": 0.8008414834674444,
+      "learning_rate": 3.5906264634482084e-05,
+      "loss": 0.6783,
+      "step": 3505
+    },
+    {
+      "epoch": 0.4098552078468006,
+      "grad_norm": 0.7982721355936174,
+      "learning_rate": 3.586594958335747e-05,
+      "loss": 0.6817,
+      "step": 3510
+    },
+    {
+      "epoch": 0.41043904717421764,
+      "grad_norm": 0.7268906859455486,
+      "learning_rate": 3.582560334048241e-05,
+      "loss": 0.6771,
+      "step": 3515
+    },
+    {
+      "epoch": 0.41102288650163477,
+      "grad_norm": 0.7779505205619507,
+      "learning_rate": 3.578522605628453e-05,
+      "loss": 0.667,
+      "step": 3520
+    },
+    {
+      "epoch": 0.41160672582905183,
+      "grad_norm": 0.7265217634872664,
+      "learning_rate": 3.5744817881307184e-05,
+      "loss": 0.6823,
+      "step": 3525
+    },
+    {
+      "epoch": 0.41219056515646896,
+      "grad_norm": 0.784282429365926,
+      "learning_rate": 3.570437896620891e-05,
+      "loss": 0.6847,
+      "step": 3530
+    },
+    {
+      "epoch": 0.412774404483886,
+      "grad_norm": 0.7832014455578007,
+      "learning_rate": 3.566390946176286e-05,
+      "loss": 0.6984,
+      "step": 3535
+    },
+    {
+      "epoch": 0.41335824381130315,
+      "grad_norm": 0.8273425356102178,
+      "learning_rate": 3.5623409518856225e-05,
+      "loss": 0.7043,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4139420831387202,
+      "grad_norm": 0.6743765693532546,
+      "learning_rate": 3.55828792884897e-05,
+      "loss": 0.6963,
+      "step": 3545
+    },
+    {
+      "epoch": 0.41452592246613734,
+      "grad_norm": 0.7941581041438405,
+      "learning_rate": 3.5542318921776886e-05,
+      "loss": 0.684,
+      "step": 3550
+    },
+    {
+      "epoch": 0.4151097617935544,
+      "grad_norm": 0.7018226480438704,
+      "learning_rate": 3.5501728569943746e-05,
+      "loss": 0.6814,
+      "step": 3555
+    },
+    {
+      "epoch": 0.41569360112097153,
+      "grad_norm": 0.7685649461733042,
+      "learning_rate": 3.546110838432806e-05,
+      "loss": 0.6786,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4162774404483886,
+      "grad_norm": 0.8047778984511744,
+      "learning_rate": 3.542045851637883e-05,
+      "loss": 0.6782,
+      "step": 3565
+    },
+    {
+      "epoch": 0.41686127977580567,
+      "grad_norm": 0.9128030650660797,
+      "learning_rate": 3.53797791176557e-05,
+      "loss": 0.7038,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4174451191032228,
+      "grad_norm": 0.7706890640928363,
+      "learning_rate": 3.5339070339828466e-05,
+      "loss": 0.6882,
+      "step": 3575
+    },
+    {
+      "epoch": 0.41802895843063986,
+      "grad_norm": 0.7693351266925069,
+      "learning_rate": 3.529833233467642e-05,
+      "loss": 0.663,
+      "step": 3580
+    },
+    {
+      "epoch": 0.418612797758057,
+      "grad_norm": 0.7558947524304466,
+      "learning_rate": 3.525756525408785e-05,
+      "loss": 0.695,
+      "step": 3585
+    },
+    {
+      "epoch": 0.41919663708547406,
+      "grad_norm": 0.8022791818037412,
+      "learning_rate": 3.521676925005945e-05,
+      "loss": 0.6911,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4197804764128912,
+      "grad_norm": 0.7439983041642912,
+      "learning_rate": 3.517594447469572e-05,
+      "loss": 0.674,
+      "step": 3595
+    },
+    {
+      "epoch": 0.42036431574030825,
+      "grad_norm": 0.6965671273422072,
+      "learning_rate": 3.513509108020846e-05,
+      "loss": 0.6945,
+      "step": 3600
+    },
+    {
+      "epoch": 0.42094815506772537,
+      "grad_norm": 0.7009108611745447,
+      "learning_rate": 3.5094209218916185e-05,
+      "loss": 0.6792,
+      "step": 3605
+    },
+    {
+      "epoch": 0.42153199439514244,
+      "grad_norm": 0.6904749909673673,
+      "learning_rate": 3.505329904324351e-05,
+      "loss": 0.6739,
+      "step": 3610
+    },
+    {
+      "epoch": 0.42211583372255956,
+      "grad_norm": 0.8174685162969035,
+      "learning_rate": 3.501236070572066e-05,
+      "loss": 0.6787,
+      "step": 3615
+    },
+    {
+      "epoch": 0.42269967304997663,
+      "grad_norm": 0.7808904635403306,
+      "learning_rate": 3.497139435898283e-05,
+      "loss": 0.6861,
+      "step": 3620
+    },
+    {
+      "epoch": 0.42328351237739376,
+      "grad_norm": 0.7140829015306348,
+      "learning_rate": 3.4930400155769644e-05,
+      "loss": 0.6773,
+      "step": 3625
+    },
+    {
+      "epoch": 0.4238673517048108,
+      "grad_norm": 0.6715183649248256,
+      "learning_rate": 3.488937824892461e-05,
+      "loss": 0.6986,
+      "step": 3630
+    },
+    {
+      "epoch": 0.42445119103222795,
+      "grad_norm": 0.6645867359518857,
+      "learning_rate": 3.48483287913945e-05,
+      "loss": 0.6984,
+      "step": 3635
+    },
+    {
+      "epoch": 0.425035030359645,
+      "grad_norm": 0.7245224781750078,
+      "learning_rate": 3.480725193622881e-05,
+      "loss": 0.6768,
+      "step": 3640
+    },
+    {
+      "epoch": 0.42561886968706214,
+      "grad_norm": 0.7290628767098649,
+      "learning_rate": 3.476614783657922e-05,
+      "loss": 0.7027,
+      "step": 3645
+    },
+    {
+      "epoch": 0.4262027090144792,
+      "grad_norm": 0.7065799709469044,
+      "learning_rate": 3.472501664569894e-05,
+      "loss": 0.6906,
+      "step": 3650
+    },
+    {
+      "epoch": 0.42678654834189633,
+      "grad_norm": 0.7141779437273328,
+      "learning_rate": 3.468385851694222e-05,
+      "loss": 0.6877,
+      "step": 3655
+    },
+    {
+      "epoch": 0.4273703876693134,
+      "grad_norm": 0.763602745903948,
+      "learning_rate": 3.464267360376373e-05,
+      "loss": 0.6811,
+      "step": 3660
+    },
+    {
+      "epoch": 0.4279542269967305,
+      "grad_norm": 0.7078483922513468,
+      "learning_rate": 3.460146205971802e-05,
+      "loss": 0.6967,
+      "step": 3665
+    },
+    {
+      "epoch": 0.4285380663241476,
+      "grad_norm": 0.6864310842103045,
+      "learning_rate": 3.456022403845891e-05,
+      "loss": 0.6887,
+      "step": 3670
+    },
+    {
+      "epoch": 0.4291219056515647,
+      "grad_norm": 0.8108275144233935,
+      "learning_rate": 3.4518959693738944e-05,
+      "loss": 0.6802,
+      "step": 3675
+    },
+    {
+      "epoch": 0.4297057449789818,
+      "grad_norm": 0.7599404194217726,
+      "learning_rate": 3.4477669179408834e-05,
+      "loss": 0.6843,
+      "step": 3680
+    },
+    {
+      "epoch": 0.43028958430639885,
+      "grad_norm": 0.77809828148712,
+      "learning_rate": 3.443635264941682e-05,
+      "loss": 0.685,
+      "step": 3685
+    },
+    {
+      "epoch": 0.430873423633816,
+      "grad_norm": 0.8118714505627129,
+      "learning_rate": 3.4395010257808185e-05,
+      "loss": 0.6891,
+      "step": 3690
+    },
+    {
+      "epoch": 0.43145726296123305,
+      "grad_norm": 0.6807080465607841,
+      "learning_rate": 3.43536421587246e-05,
+      "loss": 0.698,
+      "step": 3695
+    },
+    {
+      "epoch": 0.43204110228865017,
+      "grad_norm": 0.7551098220413885,
+      "learning_rate": 3.431224850640361e-05,
+      "loss": 0.6819,
+      "step": 3700
+    },
+    {
+      "epoch": 0.43262494161606724,
+      "grad_norm": 0.7538181047973818,
+      "learning_rate": 3.427082945517801e-05,
+      "loss": 0.677,
+      "step": 3705
+    },
+    {
+      "epoch": 0.43320878094348436,
+      "grad_norm": 0.7286404689155307,
+      "learning_rate": 3.422938515947531e-05,
+      "loss": 0.6716,
+      "step": 3710
+    },
+    {
+      "epoch": 0.43379262027090143,
+      "grad_norm": 0.7398508893666071,
+      "learning_rate": 3.418791577381713e-05,
+      "loss": 0.6785,
+      "step": 3715
+    },
+    {
+      "epoch": 0.43437645959831855,
+      "grad_norm": 0.7863660031060099,
+      "learning_rate": 3.4146421452818657e-05,
+      "loss": 0.7008,
+      "step": 3720
+    },
+    {
+      "epoch": 0.4349602989257356,
+      "grad_norm": 0.7352662601655674,
+      "learning_rate": 3.4104902351188e-05,
+      "loss": 0.6811,
+      "step": 3725
+    },
+    {
+      "epoch": 0.43554413825315275,
+      "grad_norm": 0.7398589781977044,
+      "learning_rate": 3.406335862372573e-05,
+      "loss": 0.6706,
+      "step": 3730
+    },
+    {
+      "epoch": 0.4361279775805698,
+      "grad_norm": 0.7845350671727488,
+      "learning_rate": 3.402179042532417e-05,
+      "loss": 0.6916,
+      "step": 3735
+    },
+    {
+      "epoch": 0.43671181690798694,
+      "grad_norm": 0.7246822043698157,
+      "learning_rate": 3.3980197910966915e-05,
+      "loss": 0.6877,
+      "step": 3740
+    },
+    {
+      "epoch": 0.437295656235404,
+      "grad_norm": 0.7641661732006494,
+      "learning_rate": 3.3938581235728214e-05,
+      "loss": 0.6747,
+      "step": 3745
+    },
+    {
+      "epoch": 0.43787949556282113,
+      "grad_norm": 0.739738159689709,
+      "learning_rate": 3.389694055477238e-05,
+      "loss": 0.6966,
+      "step": 3750
+    },
+    {
+      "epoch": 0.4384633348902382,
+      "grad_norm": 0.7414932876347806,
+      "learning_rate": 3.385527602335327e-05,
+      "loss": 0.6482,
+      "step": 3755
+    },
+    {
+      "epoch": 0.4390471742176553,
+      "grad_norm": 0.8264691570015129,
+      "learning_rate": 3.381358779681362e-05,
+      "loss": 0.6723,
+      "step": 3760
+    },
+    {
+      "epoch": 0.4396310135450724,
+      "grad_norm": 0.8435514134920983,
+      "learning_rate": 3.377187603058454e-05,
+      "loss": 0.6882,
+      "step": 3765
+    },
+    {
+      "epoch": 0.4402148528724895,
+      "grad_norm": 0.8156187493976549,
+      "learning_rate": 3.373014088018489e-05,
+      "loss": 0.7039,
+      "step": 3770
+    },
+    {
+      "epoch": 0.4407986921999066,
+      "grad_norm": 0.7068325775458664,
+      "learning_rate": 3.3688382501220727e-05,
+      "loss": 0.6706,
+      "step": 3775
+    },
+    {
+      "epoch": 0.4413825315273237,
+      "grad_norm": 0.7900012430179307,
+      "learning_rate": 3.364660104938472e-05,
+      "loss": 0.6909,
+      "step": 3780
+    },
+    {
+      "epoch": 0.4419663708547408,
+      "grad_norm": 0.7948978228461001,
+      "learning_rate": 3.3604796680455546e-05,
+      "loss": 0.7045,
+      "step": 3785
+    },
+    {
+      "epoch": 0.44255021018215784,
+      "grad_norm": 0.6664618406737972,
+      "learning_rate": 3.356296955029733e-05,
+      "loss": 0.6803,
+      "step": 3790
+    },
+    {
+      "epoch": 0.44313404950957497,
+      "grad_norm": 0.6988062398480143,
+      "learning_rate": 3.3521119814859063e-05,
+      "loss": 0.686,
+      "step": 3795
+    },
+    {
+      "epoch": 0.44371788883699204,
+      "grad_norm": 0.6579643187279307,
+      "learning_rate": 3.347924763017403e-05,
+      "loss": 0.673,
+      "step": 3800
+    },
+    {
+      "epoch": 0.44430172816440916,
+      "grad_norm": 0.671136468708349,
+      "learning_rate": 3.3437353152359195e-05,
+      "loss": 0.6785,
+      "step": 3805
+    },
+    {
+      "epoch": 0.44488556749182623,
+      "grad_norm": 0.7125716822949324,
+      "learning_rate": 3.339543653761466e-05,
+      "loss": 0.6866,
+      "step": 3810
+    },
+    {
+      "epoch": 0.44546940681924335,
+      "grad_norm": 0.7777198124067506,
+      "learning_rate": 3.335349794222304e-05,
+      "loss": 0.675,
+      "step": 3815
+    },
+    {
+      "epoch": 0.4460532461466604,
+      "grad_norm": 0.7387167777215246,
+      "learning_rate": 3.331153752254893e-05,
+      "loss": 0.6978,
+      "step": 3820
+    },
+    {
+      "epoch": 0.44663708547407754,
+      "grad_norm": 0.8016313362554687,
+      "learning_rate": 3.326955543503827e-05,
+      "loss": 0.705,
+      "step": 3825
+    },
+    {
+      "epoch": 0.4472209248014946,
+      "grad_norm": 0.749993680363176,
+      "learning_rate": 3.322755183621779e-05,
+      "loss": 0.6772,
+      "step": 3830
+    },
+    {
+      "epoch": 0.44780476412891174,
+      "grad_norm": 0.7348760458565766,
+      "learning_rate": 3.318552688269446e-05,
+      "loss": 0.6715,
+      "step": 3835
+    },
+    {
+      "epoch": 0.4483886034563288,
+      "grad_norm": 0.8064731890629008,
+      "learning_rate": 3.314348073115481e-05,
+      "loss": 0.6858,
+      "step": 3840
+    },
+    {
+      "epoch": 0.44897244278374593,
+      "grad_norm": 0.7578691186858068,
+      "learning_rate": 3.310141353836446e-05,
+      "loss": 0.6903,
+      "step": 3845
+    },
+    {
+      "epoch": 0.449556282111163,
+      "grad_norm": 0.7460901993225614,
+      "learning_rate": 3.305932546116743e-05,
+      "loss": 0.681,
+      "step": 3850
+    },
+    {
+      "epoch": 0.4501401214385801,
+      "grad_norm": 0.7506257697761904,
+      "learning_rate": 3.301721665648566e-05,
+      "loss": 0.6984,
+      "step": 3855
+    },
+    {
+      "epoch": 0.4507239607659972,
+      "grad_norm": 0.7420341018863588,
+      "learning_rate": 3.297508728131832e-05,
+      "loss": 0.6758,
+      "step": 3860
+    },
+    {
+      "epoch": 0.4513078000934143,
+      "grad_norm": 0.7548433309336711,
+      "learning_rate": 3.29329374927413e-05,
+      "loss": 0.6858,
+      "step": 3865
+    },
+    {
+      "epoch": 0.4518916394208314,
+      "grad_norm": 0.7394699898633986,
+      "learning_rate": 3.2890767447906615e-05,
+      "loss": 0.6718,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4524754787482485,
+      "grad_norm": 0.7427273902406784,
+      "learning_rate": 3.284857730404176e-05,
+      "loss": 0.6758,
+      "step": 3875
+    },
+    {
+      "epoch": 0.4530593180756656,
+      "grad_norm": 0.7959414367342418,
+      "learning_rate": 3.2806367218449216e-05,
+      "loss": 0.6817,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4536431574030827,
+      "grad_norm": 0.7495312967063392,
+      "learning_rate": 3.2764137348505785e-05,
+      "loss": 0.6781,
+      "step": 3885
+    },
+    {
+      "epoch": 0.45422699673049977,
+      "grad_norm": 0.8033197593264897,
+      "learning_rate": 3.2721887851662044e-05,
+      "loss": 0.6708,
+      "step": 3890
+    },
+    {
+      "epoch": 0.45481083605791683,
+      "grad_norm": 0.8691456128606405,
+      "learning_rate": 3.267961888544173e-05,
+      "loss": 0.6888,
+      "step": 3895
+    },
+    {
+      "epoch": 0.45539467538533396,
+      "grad_norm": 0.6867694796227263,
+      "learning_rate": 3.263733060744121e-05,
+      "loss": 0.6782,
+      "step": 3900
+    },
+    {
+      "epoch": 0.455978514712751,
+      "grad_norm": 0.8480385798049878,
+      "learning_rate": 3.25950231753288e-05,
+      "loss": 0.6706,
+      "step": 3905
+    },
+    {
+      "epoch": 0.45656235404016815,
+      "grad_norm": 0.7507839598810029,
+      "learning_rate": 3.255269674684427e-05,
+      "loss": 0.6969,
+      "step": 3910
+    },
+    {
+      "epoch": 0.4571461933675852,
+      "grad_norm": 0.7665881732297976,
+      "learning_rate": 3.2510351479798214e-05,
+      "loss": 0.6536,
+      "step": 3915
+    },
+    {
+      "epoch": 0.45773003269500234,
+      "grad_norm": 0.892815003005755,
+      "learning_rate": 3.2467987532071436e-05,
+      "loss": 0.6769,
+      "step": 3920
+    },
+    {
+      "epoch": 0.4583138720224194,
+      "grad_norm": 0.8147955829566529,
+      "learning_rate": 3.242560506161442e-05,
+      "loss": 0.6759,
+      "step": 3925
+    },
+    {
+      "epoch": 0.45889771134983653,
+      "grad_norm": 0.7936781689898847,
+      "learning_rate": 3.2383204226446706e-05,
+      "loss": 0.6884,
+      "step": 3930
+    },
+    {
+      "epoch": 0.4594815506772536,
+      "grad_norm": 0.7769521096875983,
+      "learning_rate": 3.234078518465628e-05,
+      "loss": 0.6919,
+      "step": 3935
+    },
+    {
+      "epoch": 0.4600653900046707,
+      "grad_norm": 0.7368280819168297,
+      "learning_rate": 3.229834809439904e-05,
+      "loss": 0.6808,
+      "step": 3940
+    },
+    {
+      "epoch": 0.4606492293320878,
+      "grad_norm": 0.8226859357051185,
+      "learning_rate": 3.225589311389816e-05,
+      "loss": 0.6772,
+      "step": 3945
+    },
+    {
+      "epoch": 0.4612330686595049,
+      "grad_norm": 0.6799317529875593,
+      "learning_rate": 3.221342040144352e-05,
+      "loss": 0.6856,
+      "step": 3950
+    },
+    {
+      "epoch": 0.461816907986922,
+      "grad_norm": 0.7261626348697685,
+      "learning_rate": 3.217093011539111e-05,
+      "loss": 0.6909,
+      "step": 3955
+    },
+    {
+      "epoch": 0.4624007473143391,
+      "grad_norm": 0.8051796951442984,
+      "learning_rate": 3.2128422414162454e-05,
+      "loss": 0.6822,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4629845866417562,
+      "grad_norm": 0.7507385693708827,
+      "learning_rate": 3.2085897456243986e-05,
+      "loss": 0.6803,
+      "step": 3965
+    },
+    {
+      "epoch": 0.4635684259691733,
+      "grad_norm": 0.8035133685379359,
+      "learning_rate": 3.204335540018649e-05,
+      "loss": 0.6765,
+      "step": 3970
+    },
+    {
+      "epoch": 0.46415226529659037,
+      "grad_norm": 0.7329707765377573,
+      "learning_rate": 3.200079640460451e-05,
+      "loss": 0.6812,
+      "step": 3975
+    },
+    {
+      "epoch": 0.4647361046240075,
+      "grad_norm": 0.7136586791277427,
+      "learning_rate": 3.195822062817573e-05,
+      "loss": 0.6649,
+      "step": 3980
+    },
+    {
+      "epoch": 0.46531994395142456,
+      "grad_norm": 0.6883260567024575,
+      "learning_rate": 3.191562822964041e-05,
+      "loss": 0.7012,
+      "step": 3985
+    },
+    {
+      "epoch": 0.4659037832788417,
+      "grad_norm": 0.6991257110758196,
+      "learning_rate": 3.187301936780079e-05,
+      "loss": 0.6555,
+      "step": 3990
+    },
+    {
+      "epoch": 0.46648762260625876,
+      "grad_norm": 0.6853148064402772,
+      "learning_rate": 3.183039420152047e-05,
+      "loss": 0.6846,
+      "step": 3995
+    },
+    {
+      "epoch": 0.4670714619336759,
+      "grad_norm": 0.6637228068073139,
+      "learning_rate": 3.178775288972386e-05,
+      "loss": 0.6824,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46765530126109295,
+      "grad_norm": 0.7145647310174994,
+      "learning_rate": 3.174509559139556e-05,
+      "loss": 0.6867,
+      "step": 4005
+    },
+    {
+      "epoch": 0.46823914058851,
+      "grad_norm": 0.6478018182375663,
+      "learning_rate": 3.170242246557978e-05,
+      "loss": 0.6795,
+      "step": 4010
+    },
+    {
+      "epoch": 0.46882297991592714,
+      "grad_norm": 0.7153515691503844,
+      "learning_rate": 3.1659733671379735e-05,
+      "loss": 0.6872,
+      "step": 4015
+    },
+    {
+      "epoch": 0.4694068192433442,
+      "grad_norm": 0.6827402551177538,
+      "learning_rate": 3.1617029367957053e-05,
+      "loss": 0.6591,
+      "step": 4020
+    },
+    {
+      "epoch": 0.46999065857076133,
+      "grad_norm": 0.7576014181506056,
+      "learning_rate": 3.1574309714531195e-05,
+      "loss": 0.6825,
+      "step": 4025
+    },
+    {
+      "epoch": 0.4705744978981784,
+      "grad_norm": 0.691219585542754,
+      "learning_rate": 3.153157487037887e-05,
+      "loss": 0.6593,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4711583372255955,
+      "grad_norm": 0.7113323822516596,
+      "learning_rate": 3.1488824994833395e-05,
+      "loss": 0.6736,
+      "step": 4035
+    },
+    {
+      "epoch": 0.4717421765530126,
+      "grad_norm": 0.7106992433701441,
+      "learning_rate": 3.1446060247284134e-05,
+      "loss": 0.676,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4723260158804297,
+      "grad_norm": 0.724318171580972,
+      "learning_rate": 3.140328078717591e-05,
+      "loss": 0.6845,
+      "step": 4045
+    },
+    {
+      "epoch": 0.4729098552078468,
+      "grad_norm": 0.7803058456745933,
+      "learning_rate": 3.1360486774008415e-05,
+      "loss": 0.6804,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4734936945352639,
+      "grad_norm": 0.6717718286032255,
+      "learning_rate": 3.131767836733556e-05,
+      "loss": 0.6851,
+      "step": 4055
+    },
+    {
+      "epoch": 0.474077533862681,
+      "grad_norm": 0.702859871171385,
+      "learning_rate": 3.127485572676496e-05,
+      "loss": 0.673,
+      "step": 4060
+    },
+    {
+      "epoch": 0.4746613731900981,
+      "grad_norm": 0.7784472753189181,
+      "learning_rate": 3.1232019011957294e-05,
+      "loss": 0.6696,
+      "step": 4065
+    },
+    {
+      "epoch": 0.47524521251751517,
+      "grad_norm": 0.7599086934930439,
+      "learning_rate": 3.118916838262568e-05,
+      "loss": 0.6618,
+      "step": 4070
+    },
+    {
+      "epoch": 0.4758290518449323,
+      "grad_norm": 0.6544883749916701,
+      "learning_rate": 3.114630399853517e-05,
+      "loss": 0.6659,
+      "step": 4075
+    },
+    {
+      "epoch": 0.47641289117234936,
+      "grad_norm": 0.676166756954035,
+      "learning_rate": 3.1103426019502055e-05,
+      "loss": 0.6565,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4769967304997665,
+      "grad_norm": 0.7666409275995504,
+      "learning_rate": 3.1060534605393345e-05,
+      "loss": 0.6923,
+      "step": 4085
+    },
+    {
+      "epoch": 0.47758056982718355,
+      "grad_norm": 0.7157437169336734,
+      "learning_rate": 3.101762991612611e-05,
+      "loss": 0.6757,
+      "step": 4090
+    },
+    {
+      "epoch": 0.4781644091546007,
+      "grad_norm": 0.8118839151621046,
+      "learning_rate": 3.0974712111666935e-05,
+      "loss": 0.6867,
+      "step": 4095
+    },
+    {
+      "epoch": 0.47874824848201775,
+      "grad_norm": 0.8147402404625602,
+      "learning_rate": 3.09317813520313e-05,
+      "loss": 0.695,
+      "step": 4100
+    },
+    {
+      "epoch": 0.47933208780943487,
+      "grad_norm": 0.6964765663247126,
+      "learning_rate": 3.0888837797283005e-05,
+      "loss": 0.6835,
+      "step": 4105
+    },
+    {
+      "epoch": 0.47991592713685194,
+      "grad_norm": 0.7354320889290884,
+      "learning_rate": 3.0845881607533524e-05,
+      "loss": 0.678,
+      "step": 4110
+    },
+    {
+      "epoch": 0.480499766464269,
+      "grad_norm": 0.6947896868509368,
+      "learning_rate": 3.0802912942941453e-05,
+      "loss": 0.685,
+      "step": 4115
+    },
+    {
+      "epoch": 0.48108360579168613,
+      "grad_norm": 0.6654318510038829,
+      "learning_rate": 3.0759931963711913e-05,
+      "loss": 0.6793,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4816674451191032,
+      "grad_norm": 0.7815873141024726,
+      "learning_rate": 3.071693883009591e-05,
+      "loss": 0.6846,
+      "step": 4125
+    },
+    {
+      "epoch": 0.4822512844465203,
+      "grad_norm": 0.7091320415900886,
+      "learning_rate": 3.06739337023898e-05,
+      "loss": 0.6747,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4828351237739374,
+      "grad_norm": 0.8021203707814092,
+      "learning_rate": 3.0630916740934626e-05,
+      "loss": 0.6771,
+      "step": 4135
+    },
+    {
+      "epoch": 0.4834189631013545,
+      "grad_norm": 0.6461471445686426,
+      "learning_rate": 3.058788810611558e-05,
+      "loss": 0.6718,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4840028024287716,
+      "grad_norm": 0.814971747463668,
+      "learning_rate": 3.054484795836136e-05,
+      "loss": 0.6782,
+      "step": 4145
+    },
+    {
+      "epoch": 0.4845866417561887,
+      "grad_norm": 0.7356998398110803,
+      "learning_rate": 3.0501796458143593e-05,
+      "loss": 0.6755,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4851704810836058,
+      "grad_norm": 0.7237335237898106,
+      "learning_rate": 3.045873376597624e-05,
+      "loss": 0.6688,
+      "step": 4155
+    },
+    {
+      "epoch": 0.4857543204110229,
+      "grad_norm": 0.7662558686718588,
+      "learning_rate": 3.041566004241498e-05,
+      "loss": 0.667,
+      "step": 4160
+    },
+    {
+      "epoch": 0.48633815973843997,
+      "grad_norm": 0.7105698871150168,
+      "learning_rate": 3.037257544805661e-05,
+      "loss": 0.6706,
+      "step": 4165
+    },
+    {
+      "epoch": 0.4869219990658571,
+      "grad_norm": 0.6840148973525364,
+      "learning_rate": 3.0329480143538498e-05,
+      "loss": 0.6827,
+      "step": 4170
+    },
+    {
+      "epoch": 0.48750583839327416,
+      "grad_norm": 0.758766760947445,
+      "learning_rate": 3.0286374289537912e-05,
+      "loss": 0.6451,
+      "step": 4175
+    },
+    {
+      "epoch": 0.4880896777206913,
+      "grad_norm": 0.6952109543804652,
+      "learning_rate": 3.0243258046771446e-05,
+      "loss": 0.6685,
+      "step": 4180
+    },
+    {
+      "epoch": 0.48867351704810835,
+      "grad_norm": 0.7526199018233571,
+      "learning_rate": 3.0200131575994456e-05,
+      "loss": 0.6867,
+      "step": 4185
+    },
+    {
+      "epoch": 0.4892573563755255,
+      "grad_norm": 0.7015147433957639,
+      "learning_rate": 3.0156995038000418e-05,
+      "loss": 0.6571,
+      "step": 4190
+    },
+    {
+      "epoch": 0.48984119570294254,
+      "grad_norm": 0.733436583796095,
+      "learning_rate": 3.011384859362034e-05,
+      "loss": 0.6803,
+      "step": 4195
+    },
+    {
+      "epoch": 0.49042503503035967,
+      "grad_norm": 0.857995207926992,
+      "learning_rate": 3.0070692403722162e-05,
+      "loss": 0.6891,
+      "step": 4200
+    },
+    {
+      "epoch": 0.49100887435777674,
+      "grad_norm": 0.7292363842938806,
+      "learning_rate": 3.002752662921018e-05,
+      "loss": 0.6929,
+      "step": 4205
+    },
+    {
+      "epoch": 0.49159271368519386,
+      "grad_norm": 0.7839103814367115,
+      "learning_rate": 2.9984351431024394e-05,
+      "loss": 0.6712,
+      "step": 4210
+    },
+    {
+      "epoch": 0.49217655301261093,
+      "grad_norm": 0.6775948265557494,
+      "learning_rate": 2.9941166970139968e-05,
+      "loss": 0.648,
+      "step": 4215
+    },
+    {
+      "epoch": 0.492760392340028,
+      "grad_norm": 0.7198282951734615,
+      "learning_rate": 2.9897973407566583e-05,
+      "loss": 0.666,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4933442316674451,
+      "grad_norm": 0.7766993957049966,
+      "learning_rate": 2.985477090434786e-05,
+      "loss": 0.6892,
+      "step": 4225
+    },
+    {
+      "epoch": 0.4939280709948622,
+      "grad_norm": 0.7291729148417119,
+      "learning_rate": 2.9811559621560765e-05,
+      "loss": 0.6806,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4945119103222793,
+      "grad_norm": 0.7756444851906211,
+      "learning_rate": 2.976833972031498e-05,
+      "loss": 0.6555,
+      "step": 4235
+    },
+    {
+      "epoch": 0.4950957496496964,
+      "grad_norm": 0.693692295860524,
+      "learning_rate": 2.9725111361752333e-05,
+      "loss": 0.6962,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4956795889771135,
+      "grad_norm": 0.6950612733524342,
+      "learning_rate": 2.968187470704618e-05,
+      "loss": 0.668,
+      "step": 4245
+    },
+    {
+      "epoch": 0.4962634283045306,
+      "grad_norm": 0.7551664289622262,
+      "learning_rate": 2.9638629917400806e-05,
+      "loss": 0.6921,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4968472676319477,
+      "grad_norm": 0.7486965420977963,
+      "learning_rate": 2.9595377154050836e-05,
+      "loss": 0.6917,
+      "step": 4255
+    },
+    {
+      "epoch": 0.49743110695936477,
+      "grad_norm": 0.7007684309768127,
+      "learning_rate": 2.955211657826061e-05,
+      "loss": 0.6843,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4980149462867819,
+      "grad_norm": 0.6963601300824674,
+      "learning_rate": 2.9508848351323597e-05,
+      "loss": 0.6566,
+      "step": 4265
+    },
+    {
+      "epoch": 0.49859878561419896,
+      "grad_norm": 0.7667329518085629,
+      "learning_rate": 2.9465572634561815e-05,
+      "loss": 0.686,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4991826249416161,
+      "grad_norm": 0.7231068396927425,
+      "learning_rate": 2.9422289589325187e-05,
+      "loss": 0.6723,
+      "step": 4275
+    },
+    {
+      "epoch": 0.49976646426903315,
+      "grad_norm": 0.8284182332439602,
+      "learning_rate": 2.9378999376990958e-05,
+      "loss": 0.6464,
+      "step": 4280
+    },
+    {
+      "epoch": 0.5003503035964503,
+      "grad_norm": 0.8459390831130426,
+      "learning_rate": 2.9335702158963107e-05,
+      "loss": 0.6628,
+      "step": 4285
+    },
+    {
+      "epoch": 0.5009341429238674,
+      "grad_norm": 0.7011442779440549,
+      "learning_rate": 2.929239809667172e-05,
+      "loss": 0.6432,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5015179822512844,
+      "grad_norm": 0.8746587602987824,
+      "learning_rate": 2.9249087351572414e-05,
+      "loss": 0.6903,
+      "step": 4295
+    },
+    {
+      "epoch": 0.5021018215787015,
+      "grad_norm": 0.7257414684272304,
+      "learning_rate": 2.9205770085145716e-05,
+      "loss": 0.692,
+      "step": 4300
+    },
+    {
+      "epoch": 0.5026856609061187,
+      "grad_norm": 0.805965566100688,
+      "learning_rate": 2.916244645889647e-05,
+      "loss": 0.6764,
+      "step": 4305
+    },
+    {
+      "epoch": 0.5032695002335358,
+      "grad_norm": 0.7462663884955255,
+      "learning_rate": 2.911911663435322e-05,
+      "loss": 0.6627,
+      "step": 4310
+    },
+    {
+      "epoch": 0.5038533395609528,
+      "grad_norm": 0.7448138565708283,
+      "learning_rate": 2.9075780773067644e-05,
+      "loss": 0.6561,
+      "step": 4315
+    },
+    {
+      "epoch": 0.5044371788883699,
+      "grad_norm": 0.7391907955354905,
+      "learning_rate": 2.9032439036613907e-05,
+      "loss": 0.6938,
+      "step": 4320
+    },
+    {
+      "epoch": 0.505021018215787,
+      "grad_norm": 0.7090872369182649,
+      "learning_rate": 2.8989091586588085e-05,
+      "loss": 0.6734,
+      "step": 4325
+    },
+    {
+      "epoch": 0.5056048575432041,
+      "grad_norm": 0.6513803711091114,
+      "learning_rate": 2.894573858460755e-05,
+      "loss": 0.6759,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5061886968706212,
+      "grad_norm": 0.7084768130001255,
+      "learning_rate": 2.8902380192310392e-05,
+      "loss": 0.6649,
+      "step": 4335
+    },
+    {
+      "epoch": 0.5067725361980383,
+      "grad_norm": 0.77828284609766,
+      "learning_rate": 2.8859016571354778e-05,
+      "loss": 0.6686,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5073563755254554,
+      "grad_norm": 0.7501781315460823,
+      "learning_rate": 2.881564788341839e-05,
+      "loss": 0.6868,
+      "step": 4345
+    },
+    {
+      "epoch": 0.5079402148528724,
+      "grad_norm": 0.6544860225443073,
+      "learning_rate": 2.877227429019778e-05,
+      "loss": 0.6688,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5085240541802896,
+      "grad_norm": 0.7406104086931747,
+      "learning_rate": 2.872889595340781e-05,
+      "loss": 0.6712,
+      "step": 4355
+    },
+    {
+      "epoch": 0.5091078935077067,
+      "grad_norm": 0.6811806607091296,
+      "learning_rate": 2.8685513034781003e-05,
+      "loss": 0.6626,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5096917328351238,
+      "grad_norm": 0.767597087471059,
+      "learning_rate": 2.864212569606699e-05,
+      "loss": 0.6624,
+      "step": 4365
+    },
+    {
+      "epoch": 0.5102755721625408,
+      "grad_norm": 0.6614583652870923,
+      "learning_rate": 2.8598734099031878e-05,
+      "loss": 0.6575,
+      "step": 4370
+    },
+    {
+      "epoch": 0.510859411489958,
+      "grad_norm": 0.7740783585451079,
+      "learning_rate": 2.8555338405457628e-05,
+      "loss": 0.6863,
+      "step": 4375
+    },
+    {
+      "epoch": 0.5114432508173751,
+      "grad_norm": 0.7934060933684554,
+      "learning_rate": 2.85119387771415e-05,
+      "loss": 0.6706,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5120270901447922,
+      "grad_norm": 0.7459963061955495,
+      "learning_rate": 2.8468535375895417e-05,
+      "loss": 0.6442,
+      "step": 4385
+    },
+    {
+      "epoch": 0.5126109294722092,
+      "grad_norm": 0.6723748698282871,
+      "learning_rate": 2.8425128363545362e-05,
+      "loss": 0.6886,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5131947687996263,
+      "grad_norm": 0.669676392518392,
+      "learning_rate": 2.8381717901930792e-05,
+      "loss": 0.6547,
+      "step": 4395
+    },
+    {
+      "epoch": 0.5137786081270435,
+      "grad_norm": 0.659571997070763,
+      "learning_rate": 2.8338304152904016e-05,
+      "loss": 0.6843,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5143624474544606,
+      "grad_norm": 0.8288991445997408,
+      "learning_rate": 2.8294887278329606e-05,
+      "loss": 0.6763,
+      "step": 4405
+    },
+    {
+      "epoch": 0.5149462867818776,
+      "grad_norm": 0.707942997453321,
+      "learning_rate": 2.825146744008378e-05,
+      "loss": 0.6564,
+      "step": 4410
+    },
+    {
+      "epoch": 0.5155301261092947,
+      "grad_norm": 0.7278932584750964,
+      "learning_rate": 2.8208044800053822e-05,
+      "loss": 0.7021,
+      "step": 4415
+    },
+    {
+      "epoch": 0.5161139654367118,
+      "grad_norm": 0.6352750094631958,
+      "learning_rate": 2.8164619520137437e-05,
+      "loss": 0.6696,
+      "step": 4420
+    },
+    {
+      "epoch": 0.516697804764129,
+      "grad_norm": 0.6784043685774885,
+      "learning_rate": 2.8121191762242188e-05,
+      "loss": 0.6676,
+      "step": 4425
+    },
+    {
+      "epoch": 0.517281644091546,
+      "grad_norm": 0.6497329718856635,
+      "learning_rate": 2.8077761688284886e-05,
+      "loss": 0.6513,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5178654834189631,
+      "grad_norm": 0.6706848805024374,
+      "learning_rate": 2.803432946019095e-05,
+      "loss": 0.669,
+      "step": 4435
+    },
+    {
+      "epoch": 0.5184493227463802,
+      "grad_norm": 0.7504448419630855,
+      "learning_rate": 2.7990895239893866e-05,
+      "loss": 0.664,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5190331620737972,
+      "grad_norm": 0.6803615489829645,
+      "learning_rate": 2.7947459189334514e-05,
+      "loss": 0.6466,
+      "step": 4445
+    },
+    {
+      "epoch": 0.5196170014012144,
+      "grad_norm": 0.6974815916635871,
+      "learning_rate": 2.790402147046062e-05,
+      "loss": 0.6734,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5202008407286315,
+      "grad_norm": 0.6876467906226407,
+      "learning_rate": 2.7860582245226114e-05,
+      "loss": 0.6582,
+      "step": 4455
+    },
+    {
+      "epoch": 0.5207846800560486,
+      "grad_norm": 0.6834407069996354,
+      "learning_rate": 2.781714167559056e-05,
+      "loss": 0.6652,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5213685193834656,
+      "grad_norm": 0.8004024893171434,
+      "learning_rate": 2.7773699923518527e-05,
+      "loss": 0.68,
+      "step": 4465
+    },
+    {
+      "epoch": 0.5219523587108827,
+      "grad_norm": 0.6831945682441466,
+      "learning_rate": 2.7730257150978985e-05,
+      "loss": 0.6804,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5225361980382999,
+      "grad_norm": 0.6439654412707709,
+      "learning_rate": 2.7686813519944716e-05,
+      "loss": 0.6756,
+      "step": 4475
+    },
+    {
+      "epoch": 0.523120037365717,
+      "grad_norm": 0.6867283041103277,
+      "learning_rate": 2.7643369192391705e-05,
+      "loss": 0.6619,
+      "step": 4480
+    },
+    {
+      "epoch": 0.523703876693134,
+      "grad_norm": 0.7432268603888622,
+      "learning_rate": 2.759992433029852e-05,
+      "loss": 0.6728,
+      "step": 4485
+    },
+    {
+      "epoch": 0.5242877160205511,
+      "grad_norm": 0.7089308367934053,
+      "learning_rate": 2.7556479095645753e-05,
+      "loss": 0.6585,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5248715553479683,
+      "grad_norm": 0.7013335826205295,
+      "learning_rate": 2.7513033650415352e-05,
+      "loss": 0.683,
+      "step": 4495
+    },
+    {
+      "epoch": 0.5254553946753854,
+      "grad_norm": 0.7000494253946886,
+      "learning_rate": 2.7469588156590065e-05,
+      "loss": 0.6546,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5260392340028024,
+      "grad_norm": 0.683894527168804,
+      "learning_rate": 2.742614277615282e-05,
+      "loss": 0.6786,
+      "step": 4505
+    },
+    {
+      "epoch": 0.5266230733302195,
+      "grad_norm": 0.6302791455523522,
+      "learning_rate": 2.7382697671086115e-05,
+      "loss": 0.6775,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5272069126576366,
+      "grad_norm": 0.6529176555018806,
+      "learning_rate": 2.7339253003371434e-05,
+      "loss": 0.6769,
+      "step": 4515
+    },
+    {
+      "epoch": 0.5277907519850538,
+      "grad_norm": 0.6772272567588813,
+      "learning_rate": 2.729580893498862e-05,
+      "loss": 0.683,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5283745913124708,
+      "grad_norm": 0.6902263132262554,
+      "learning_rate": 2.725236562791529e-05,
+      "loss": 0.6591,
+      "step": 4525
+    },
+    {
+      "epoch": 0.5289584306398879,
+      "grad_norm": 0.644687802132002,
+      "learning_rate": 2.7208923244126218e-05,
+      "loss": 0.6695,
+      "step": 4530
+    },
+    {
+      "epoch": 0.529542269967305,
+      "grad_norm": 0.7687528252927972,
+      "learning_rate": 2.716548194559273e-05,
+      "loss": 0.6899,
+      "step": 4535
+    },
+    {
+      "epoch": 0.5301261092947221,
+      "grad_norm": 0.6848256355744042,
+      "learning_rate": 2.7122041894282113e-05,
+      "loss": 0.6811,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5307099486221392,
+      "grad_norm": 0.7221699347481154,
+      "learning_rate": 2.707860325215701e-05,
+      "loss": 0.6885,
+      "step": 4545
+    },
+    {
+      "epoch": 0.5312937879495563,
+      "grad_norm": 0.6872951132103544,
+      "learning_rate": 2.7035166181174786e-05,
+      "loss": 0.6732,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5318776272769734,
+      "grad_norm": 0.7183593593672226,
+      "learning_rate": 2.6991730843286985e-05,
+      "loss": 0.6701,
+      "step": 4555
+    },
+    {
+      "epoch": 0.5324614666043904,
+      "grad_norm": 0.8089135923074973,
+      "learning_rate": 2.6948297400438654e-05,
+      "loss": 0.6936,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5330453059318075,
+      "grad_norm": 0.7222352662881156,
+      "learning_rate": 2.6904866014567792e-05,
+      "loss": 0.6828,
+      "step": 4565
+    },
+    {
+      "epoch": 0.5336291452592247,
+      "grad_norm": 0.7765733283168805,
+      "learning_rate": 2.686143684760473e-05,
+      "loss": 0.6808,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5342129845866418,
+      "grad_norm": 0.6878981817242145,
+      "learning_rate": 2.6818010061471516e-05,
+      "loss": 0.655,
+      "step": 4575
+    },
+    {
+      "epoch": 0.5347968239140588,
+      "grad_norm": 0.7122473652678842,
+      "learning_rate": 2.6774585818081332e-05,
+      "loss": 0.6562,
+      "step": 4580
+    },
+    {
+      "epoch": 0.5353806632414759,
+      "grad_norm": 0.6850388051290822,
+      "learning_rate": 2.6731164279337867e-05,
+      "loss": 0.6875,
+      "step": 4585
+    },
+    {
+      "epoch": 0.535964502568893,
+      "grad_norm": 0.6854622359359065,
+      "learning_rate": 2.668774560713474e-05,
+      "loss": 0.6843,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5365483418963102,
+      "grad_norm": 0.6973189987108477,
+      "learning_rate": 2.6644329963354882e-05,
+      "loss": 0.6785,
+      "step": 4595
+    },
+    {
+      "epoch": 0.5371321812237272,
+      "grad_norm": 0.7330283715603699,
+      "learning_rate": 2.6600917509869912e-05,
+      "loss": 0.6652,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5377160205511443,
+      "grad_norm": 0.6815823319864532,
+      "learning_rate": 2.655750840853958e-05,
+      "loss": 0.6531,
+      "step": 4605
+    },
+    {
+      "epoch": 0.5382998598785614,
+      "grad_norm": 0.7460677526528949,
+      "learning_rate": 2.6514102821211117e-05,
+      "loss": 0.6843,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5388836992059786,
+      "grad_norm": 0.6836960941272519,
+      "learning_rate": 2.647070090971867e-05,
+      "loss": 0.672,
+      "step": 4615
+    },
+    {
+      "epoch": 0.5394675385333956,
+      "grad_norm": 0.7513333647671177,
+      "learning_rate": 2.6427302835882672e-05,
+      "loss": 0.6534,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5400513778608127,
+      "grad_norm": 0.7273470200485447,
+      "learning_rate": 2.6383908761509252e-05,
+      "loss": 0.6586,
+      "step": 4625
+    },
+    {
+      "epoch": 0.5406352171882298,
+      "grad_norm": 0.7056443904596775,
+      "learning_rate": 2.634051884838961e-05,
+      "loss": 0.6731,
+      "step": 4630
+    },
+    {
+      "epoch": 0.541219056515647,
+      "grad_norm": 0.6633888474887489,
+      "learning_rate": 2.629713325829946e-05,
+      "loss": 0.6865,
+      "step": 4635
+    },
+    {
+      "epoch": 0.541802895843064,
+      "grad_norm": 0.7681657660191756,
+      "learning_rate": 2.625375215299838e-05,
+      "loss": 0.674,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5423867351704811,
+      "grad_norm": 0.6753257755390405,
+      "learning_rate": 2.6210375694229227e-05,
+      "loss": 0.6872,
+      "step": 4645
+    },
+    {
+      "epoch": 0.5429705744978982,
+      "grad_norm": 0.6863217893646837,
+      "learning_rate": 2.6167004043717535e-05,
+      "loss": 0.6869,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5435544138253152,
+      "grad_norm": 0.7052451068265377,
+      "learning_rate": 2.6123637363170912e-05,
+      "loss": 0.6361,
+      "step": 4655
+    },
+    {
+      "epoch": 0.5441382531527323,
+      "grad_norm": 0.6881306528333085,
+      "learning_rate": 2.6080275814278444e-05,
+      "loss": 0.6682,
+      "step": 4660
+    },
+    {
+      "epoch": 0.5447220924801495,
+      "grad_norm": 0.6721828847442636,
+      "learning_rate": 2.6036919558710064e-05,
+      "loss": 0.6798,
+      "step": 4665
+    },
+    {
+      "epoch": 0.5453059318075666,
+      "grad_norm": 0.7165806437961335,
+      "learning_rate": 2.599356875811599e-05,
+      "loss": 0.674,
+      "step": 4670
+    },
+    {
+      "epoch": 0.5458897711349836,
+      "grad_norm": 0.7068701259435769,
+      "learning_rate": 2.595022357412609e-05,
+      "loss": 0.6427,
+      "step": 4675
+    },
+    {
+      "epoch": 0.5464736104624007,
+      "grad_norm": 0.79055893788562,
+      "learning_rate": 2.59068841683493e-05,
+      "loss": 0.6656,
+      "step": 4680
+    },
+    {
+      "epoch": 0.5470574497898179,
+      "grad_norm": 0.7125832389911771,
+      "learning_rate": 2.586355070237301e-05,
+      "loss": 0.6725,
+      "step": 4685
+    },
+    {
+      "epoch": 0.547641289117235,
+      "grad_norm": 0.6493827727959639,
+      "learning_rate": 2.5820223337762438e-05,
+      "loss": 0.6498,
+      "step": 4690
+    },
+    {
+      "epoch": 0.548225128444652,
+      "grad_norm": 0.8022416079723034,
+      "learning_rate": 2.5776902236060096e-05,
+      "loss": 0.665,
+      "step": 4695
+    },
+    {
+      "epoch": 0.5488089677720691,
+      "grad_norm": 0.6467026526515551,
+      "learning_rate": 2.5733587558785126e-05,
+      "loss": 0.6722,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5493928070994862,
+      "grad_norm": 0.7303007973795397,
+      "learning_rate": 2.569027946743271e-05,
+      "loss": 0.674,
+      "step": 4705
+    },
+    {
+      "epoch": 0.5499766464269034,
+      "grad_norm": 0.7859509288962917,
+      "learning_rate": 2.5646978123473477e-05,
+      "loss": 0.6611,
+      "step": 4710
+    },
+    {
+      "epoch": 0.5505604857543204,
+      "grad_norm": 0.7065230527690424,
+      "learning_rate": 2.560368368835291e-05,
+      "loss": 0.6643,
+      "step": 4715
+    },
+    {
+      "epoch": 0.5511443250817375,
+      "grad_norm": 0.7599848569335382,
+      "learning_rate": 2.5560396323490725e-05,
+      "loss": 0.6709,
+      "step": 4720
+    },
+    {
+      "epoch": 0.5517281644091546,
+      "grad_norm": 0.7039887949666968,
+      "learning_rate": 2.5517116190280284e-05,
+      "loss": 0.6875,
+      "step": 4725
+    },
+    {
+      "epoch": 0.5523120037365717,
+      "grad_norm": 0.6406624443500122,
+      "learning_rate": 2.547384345008797e-05,
+      "loss": 0.6778,
+      "step": 4730
+    },
+    {
+      "epoch": 0.5528958430639888,
+      "grad_norm": 0.6784449486616754,
+      "learning_rate": 2.5430578264252612e-05,
+      "loss": 0.6721,
+      "step": 4735
+    },
+    {
+      "epoch": 0.5534796823914059,
+      "grad_norm": 0.6718997692773799,
+      "learning_rate": 2.538732079408489e-05,
+      "loss": 0.6678,
+      "step": 4740
+    },
+    {
+      "epoch": 0.554063521718823,
+      "grad_norm": 0.760270195049314,
+      "learning_rate": 2.534407120086668e-05,
+      "loss": 0.6776,
+      "step": 4745
+    },
+    {
+      "epoch": 0.5546473610462401,
+      "grad_norm": 0.6944859631748597,
+      "learning_rate": 2.5300829645850533e-05,
+      "loss": 0.6511,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5552312003736571,
+      "grad_norm": 0.7573413588300381,
+      "learning_rate": 2.5257596290258983e-05,
+      "loss": 0.6439,
+      "step": 4755
+    },
+    {
+      "epoch": 0.5558150397010743,
+      "grad_norm": 0.726557474037718,
+      "learning_rate": 2.5214371295284028e-05,
+      "loss": 0.6808,
+      "step": 4760
+    },
+    {
+      "epoch": 0.5563988790284914,
+      "grad_norm": 0.748698584956097,
+      "learning_rate": 2.517115482208649e-05,
+      "loss": 0.651,
+      "step": 4765
+    },
+    {
+      "epoch": 0.5569827183559084,
+      "grad_norm": 0.693531945759846,
+      "learning_rate": 2.5127947031795397e-05,
+      "loss": 0.6613,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5575665576833255,
+      "grad_norm": 0.6803795484479034,
+      "learning_rate": 2.5084748085507432e-05,
+      "loss": 0.6682,
+      "step": 4775
+    },
+    {
+      "epoch": 0.5581503970107426,
+      "grad_norm": 0.7540637765559963,
+      "learning_rate": 2.5041558144286282e-05,
+      "loss": 0.6763,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5587342363381598,
+      "grad_norm": 0.7459185650491311,
+      "learning_rate": 2.499837736916207e-05,
+      "loss": 0.6585,
+      "step": 4785
+    },
+    {
+      "epoch": 0.5593180756655768,
+      "grad_norm": 0.7270311154763253,
+      "learning_rate": 2.495520592113074e-05,
+      "loss": 0.647,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5599019149929939,
+      "grad_norm": 0.7572319932596162,
+      "learning_rate": 2.4912043961153468e-05,
+      "loss": 0.6673,
+      "step": 4795
+    },
+    {
+      "epoch": 0.560485754320411,
+      "grad_norm": 0.7727229372888152,
+      "learning_rate": 2.486889165015604e-05,
+      "loss": 0.6651,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5610695936478282,
+      "grad_norm": 0.6497425153152417,
+      "learning_rate": 2.4825749149028277e-05,
+      "loss": 0.6601,
+      "step": 4805
+    },
+    {
+      "epoch": 0.5616534329752452,
+      "grad_norm": 0.7353593698498104,
+      "learning_rate": 2.4782616618623428e-05,
+      "loss": 0.6611,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5622372723026623,
+      "grad_norm": 0.7327606724742733,
+      "learning_rate": 2.4739494219757554e-05,
+      "loss": 0.6742,
+      "step": 4815
+    },
+    {
+      "epoch": 0.5628211116300794,
+      "grad_norm": 0.639718172618709,
+      "learning_rate": 2.4696382113208956e-05,
+      "loss": 0.6642,
+      "step": 4820
+    },
+    {
+      "epoch": 0.5634049509574965,
+      "grad_norm": 0.7019189464695904,
+      "learning_rate": 2.465328045971755e-05,
+      "loss": 0.6645,
+      "step": 4825
+    },
+    {
+      "epoch": 0.5639887902849136,
+      "grad_norm": 0.67700636805716,
+      "learning_rate": 2.4610189419984285e-05,
+      "loss": 0.6732,
+      "step": 4830
+    },
+    {
+      "epoch": 0.5645726296123307,
+      "grad_norm": 0.8010967962162209,
+      "learning_rate": 2.4567109154670542e-05,
+      "loss": 0.6597,
+      "step": 4835
+    },
+    {
+      "epoch": 0.5651564689397478,
+      "grad_norm": 0.6299080051446841,
+      "learning_rate": 2.452403982439751e-05,
+      "loss": 0.6871,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5657403082671649,
+      "grad_norm": 0.7136247413291765,
+      "learning_rate": 2.4480981589745632e-05,
+      "loss": 0.6459,
+      "step": 4845
+    },
+    {
+      "epoch": 0.5663241475945819,
+      "grad_norm": 0.6576425824086324,
+      "learning_rate": 2.4437934611253972e-05,
+      "loss": 0.6675,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5669079869219991,
+      "grad_norm": 0.6499263354460856,
+      "learning_rate": 2.4394899049419612e-05,
+      "loss": 0.6545,
+      "step": 4855
+    },
+    {
+      "epoch": 0.5674918262494162,
+      "grad_norm": 0.6386357092056523,
+      "learning_rate": 2.4351875064697093e-05,
+      "loss": 0.6604,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5680756655768333,
+      "grad_norm": 0.646440268675398,
+      "learning_rate": 2.4308862817497786e-05,
+      "loss": 0.6546,
+      "step": 4865
+    },
+    {
+      "epoch": 0.5686595049042503,
+      "grad_norm": 0.7126284723406615,
+      "learning_rate": 2.4265862468189283e-05,
+      "loss": 0.6683,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5692433442316674,
+      "grad_norm": 0.6512344053164097,
+      "learning_rate": 2.4222874177094823e-05,
+      "loss": 0.6669,
+      "step": 4875
+    },
+    {
+      "epoch": 0.5698271835590846,
+      "grad_norm": 0.6407495105162438,
+      "learning_rate": 2.4179898104492705e-05,
+      "loss": 0.6497,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5704110228865016,
+      "grad_norm": 0.6742933223157447,
+      "learning_rate": 2.4136934410615646e-05,
+      "loss": 0.663,
+      "step": 4885
+    },
+    {
+      "epoch": 0.5709948622139187,
+      "grad_norm": 0.6763547210626801,
+      "learning_rate": 2.4093983255650227e-05,
+      "loss": 0.656,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5715787015413358,
+      "grad_norm": 0.6378251514665205,
+      "learning_rate": 2.405104479973628e-05,
+      "loss": 0.6815,
+      "step": 4895
+    },
+    {
+      "epoch": 0.572162540868753,
+      "grad_norm": 0.6276578822503801,
+      "learning_rate": 2.400811920296627e-05,
+      "loss": 0.6546,
+      "step": 4900
+    },
+    {
+      "epoch": 0.57274638019617,
+      "grad_norm": 0.6804919947104875,
+      "learning_rate": 2.396520662538474e-05,
+      "loss": 0.6362,
+      "step": 4905
+    },
+    {
+      "epoch": 0.5733302195235871,
+      "grad_norm": 0.687947921848423,
+      "learning_rate": 2.3922307226987678e-05,
+      "loss": 0.6567,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5739140588510042,
+      "grad_norm": 0.7186220123458101,
+      "learning_rate": 2.3879421167721944e-05,
+      "loss": 0.6785,
+      "step": 4915
+    },
+    {
+      "epoch": 0.5744978981784213,
+      "grad_norm": 0.6464932036374117,
+      "learning_rate": 2.383654860748466e-05,
+      "loss": 0.6777,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5750817375058384,
+      "grad_norm": 0.6702554906203262,
+      "learning_rate": 2.379368970612261e-05,
+      "loss": 0.6479,
+      "step": 4925
+    },
+    {
+      "epoch": 0.5756655768332555,
+      "grad_norm": 0.6521028967880418,
+      "learning_rate": 2.375084462343167e-05,
+      "loss": 0.6458,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5762494161606726,
+      "grad_norm": 0.6446069797513071,
+      "learning_rate": 2.370801351915617e-05,
+      "loss": 0.6487,
+      "step": 4935
+    },
+    {
+      "epoch": 0.5768332554880897,
+      "grad_norm": 0.6676918348307936,
+      "learning_rate": 2.3665196552988357e-05,
+      "loss": 0.6684,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5774170948155067,
+      "grad_norm": 0.6762410280464071,
+      "learning_rate": 2.362239388456773e-05,
+      "loss": 0.6747,
+      "step": 4945
+    },
+    {
+      "epoch": 0.5780009341429239,
+      "grad_norm": 0.6347782972936462,
+      "learning_rate": 2.357960567348049e-05,
+      "loss": 0.662,
+      "step": 4950
+    },
+    {
+      "epoch": 0.578584773470341,
+      "grad_norm": 0.6669462226307387,
+      "learning_rate": 2.3536832079258952e-05,
+      "loss": 0.6617,
+      "step": 4955
+    },
+    {
+      "epoch": 0.5791686127977581,
+      "grad_norm": 0.6570631505289427,
+      "learning_rate": 2.3494073261380915e-05,
+      "loss": 0.6762,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5797524521251751,
+      "grad_norm": 0.6296806022007636,
+      "learning_rate": 2.34513293792691e-05,
+      "loss": 0.663,
+      "step": 4965
+    },
+    {
+      "epoch": 0.5803362914525922,
+      "grad_norm": 0.693156151064996,
+      "learning_rate": 2.340860059229052e-05,
+      "loss": 0.6434,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5809201307800094,
+      "grad_norm": 0.6854563034520345,
+      "learning_rate": 2.3365887059755925e-05,
+      "loss": 0.677,
+      "step": 4975
+    },
+    {
+      "epoch": 0.5815039701074264,
+      "grad_norm": 0.6839853383397014,
+      "learning_rate": 2.3323188940919188e-05,
+      "loss": 0.6547,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5820878094348435,
+      "grad_norm": 0.7313864803016533,
+      "learning_rate": 2.328050639497671e-05,
+      "loss": 0.6664,
+      "step": 4985
+    },
+    {
+      "epoch": 0.5826716487622606,
+      "grad_norm": 0.6543910450847746,
+      "learning_rate": 2.3237839581066828e-05,
+      "loss": 0.655,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5832554880896778,
+      "grad_norm": 0.7441738407990897,
+      "learning_rate": 2.3195188658269224e-05,
+      "loss": 0.6503,
+      "step": 4995
+    },
+    {
+      "epoch": 0.5838393274170948,
+      "grad_norm": 0.7099442194938861,
+      "learning_rate": 2.3152553785604336e-05,
+      "loss": 0.6605,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5844231667445119,
+      "grad_norm": 0.7147324216220688,
+      "learning_rate": 2.3109935122032754e-05,
+      "loss": 0.6482,
+      "step": 5005
+    },
+    {
+      "epoch": 0.585007006071929,
+      "grad_norm": 0.735545696956427,
+      "learning_rate": 2.3067332826454647e-05,
+      "loss": 0.665,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5855908453993461,
+      "grad_norm": 0.6164011462893052,
+      "learning_rate": 2.3024747057709132e-05,
+      "loss": 0.6779,
+      "step": 5015
+    },
+    {
+      "epoch": 0.5861746847267632,
+      "grad_norm": 0.7158026426532503,
+      "learning_rate": 2.2982177974573733e-05,
+      "loss": 0.6698,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5867585240541803,
+      "grad_norm": 0.6519119068548397,
+      "learning_rate": 2.2939625735763743e-05,
+      "loss": 0.6553,
+      "step": 5025
+    },
+    {
+      "epoch": 0.5873423633815974,
+      "grad_norm": 0.708133740672085,
+      "learning_rate": 2.2897090499931674e-05,
+      "loss": 0.6438,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5879262027090145,
+      "grad_norm": 0.7959096079436856,
+      "learning_rate": 2.285457242566662e-05,
+      "loss": 0.639,
+      "step": 5035
+    },
+    {
+      "epoch": 0.5885100420364315,
+      "grad_norm": 0.7511747413046199,
+      "learning_rate": 2.2812071671493713e-05,
+      "loss": 0.6546,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5890938813638487,
+      "grad_norm": 0.6565467442204779,
+      "learning_rate": 2.2769588395873482e-05,
+      "loss": 0.6462,
+      "step": 5045
+    },
+    {
+      "epoch": 0.5896777206912658,
+      "grad_norm": 0.6953178188769654,
+      "learning_rate": 2.272712275720132e-05,
+      "loss": 0.6675,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5902615600186829,
+      "grad_norm": 0.7005519055574339,
+      "learning_rate": 2.268467491380683e-05,
+      "loss": 0.6592,
+      "step": 5055
+    },
+    {
+      "epoch": 0.5908453993460999,
+      "grad_norm": 0.6331925247962092,
+      "learning_rate": 2.264224502395329e-05,
+      "loss": 0.6615,
+      "step": 5060
+    },
+    {
+      "epoch": 0.591429238673517,
+      "grad_norm": 0.6788925130695869,
+      "learning_rate": 2.2599833245837032e-05,
+      "loss": 0.6804,
+      "step": 5065
+    },
+    {
+      "epoch": 0.5920130780009342,
+      "grad_norm": 0.7555304492511874,
+      "learning_rate": 2.2557439737586856e-05,
+      "loss": 0.6739,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5925969173283513,
+      "grad_norm": 0.7080213191115895,
+      "learning_rate": 2.2515064657263447e-05,
+      "loss": 0.6511,
+      "step": 5075
+    },
+    {
+      "epoch": 0.5931807566557683,
+      "grad_norm": 0.7125684371696229,
+      "learning_rate": 2.2472708162858792e-05,
+      "loss": 0.6588,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5937645959831854,
+      "grad_norm": 0.6677673198930747,
+      "learning_rate": 2.2430370412295566e-05,
+      "loss": 0.6729,
+      "step": 5085
+    },
+    {
+      "epoch": 0.5943484353106026,
+      "grad_norm": 0.6833790177033672,
+      "learning_rate": 2.2388051563426577e-05,
+      "loss": 0.6915,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5949322746380196,
+      "grad_norm": 0.6961110689822065,
+      "learning_rate": 2.2345751774034135e-05,
+      "loss": 0.6573,
+      "step": 5095
+    },
+    {
+      "epoch": 0.5955161139654367,
+      "grad_norm": 0.6861149572383188,
+      "learning_rate": 2.230347120182951e-05,
+      "loss": 0.6729,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5960999532928538,
+      "grad_norm": 0.6513104746202458,
+      "learning_rate": 2.226121000445232e-05,
+      "loss": 0.6547,
+      "step": 5105
+    },
+    {
+      "epoch": 0.5966837926202709,
+      "grad_norm": 0.6794112240815948,
+      "learning_rate": 2.2218968339469932e-05,
+      "loss": 0.645,
+      "step": 5110
+    },
+    {
+      "epoch": 0.597267631947688,
+      "grad_norm": 0.659653061736524,
+      "learning_rate": 2.2176746364376904e-05,
+      "loss": 0.6581,
+      "step": 5115
+    },
+    {
+      "epoch": 0.5978514712751051,
+      "grad_norm": 0.7388240821848575,
+      "learning_rate": 2.2134544236594374e-05,
+      "loss": 0.6396,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5984353106025222,
+      "grad_norm": 0.7742865515393434,
+      "learning_rate": 2.2092362113469474e-05,
+      "loss": 0.664,
+      "step": 5125
+    },
+    {
+      "epoch": 0.5990191499299393,
+      "grad_norm": 0.6957977775421466,
+      "learning_rate": 2.2050200152274763e-05,
+      "loss": 0.6774,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5996029892573563,
+      "grad_norm": 0.6763495473519054,
+      "learning_rate": 2.2008058510207635e-05,
+      "loss": 0.6758,
+      "step": 5135
+    },
+    {
+      "epoch": 0.6001868285847735,
+      "grad_norm": 0.6311990597582967,
+      "learning_rate": 2.1965937344389692e-05,
+      "loss": 0.6529,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6007706679121906,
+      "grad_norm": 0.664405793100965,
+      "learning_rate": 2.1923836811866227e-05,
+      "loss": 0.6547,
+      "step": 5145
+    },
+    {
+      "epoch": 0.6013545072396077,
+      "grad_norm": 0.6720962869376756,
+      "learning_rate": 2.188175706960559e-05,
+      "loss": 0.6612,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6019383465670247,
+      "grad_norm": 0.6224353508799658,
+      "learning_rate": 2.1839698274498616e-05,
+      "loss": 0.6825,
+      "step": 5155
+    },
+    {
+      "epoch": 0.6025221858944418,
+      "grad_norm": 0.6880116859217059,
+      "learning_rate": 2.1797660583358032e-05,
+      "loss": 0.6597,
+      "step": 5160
+    },
+    {
+      "epoch": 0.603106025221859,
+      "grad_norm": 0.6691884184144905,
+      "learning_rate": 2.1755644152917903e-05,
+      "loss": 0.6499,
+      "step": 5165
+    },
+    {
+      "epoch": 0.6036898645492761,
+      "grad_norm": 0.6708563167435044,
+      "learning_rate": 2.1713649139833e-05,
+      "loss": 0.6734,
+      "step": 5170
+    },
+    {
+      "epoch": 0.6042737038766931,
+      "grad_norm": 0.6494737500130949,
+      "learning_rate": 2.1671675700678257e-05,
+      "loss": 0.6619,
+      "step": 5175
+    },
+    {
+      "epoch": 0.6048575432041102,
+      "grad_norm": 0.6848091657139095,
+      "learning_rate": 2.1629723991948176e-05,
+      "loss": 0.6633,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6054413825315274,
+      "grad_norm": 0.6456725587890948,
+      "learning_rate": 2.1587794170056213e-05,
+      "loss": 0.6661,
+      "step": 5185
+    },
+    {
+      "epoch": 0.6060252218589445,
+      "grad_norm": 0.6494044214651111,
+      "learning_rate": 2.154588639133425e-05,
+      "loss": 0.674,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6066090611863615,
+      "grad_norm": 0.7196799234230328,
+      "learning_rate": 2.1504000812031966e-05,
+      "loss": 0.6673,
+      "step": 5195
+    },
+    {
+      "epoch": 0.6071929005137786,
+      "grad_norm": 0.6588531600130353,
+      "learning_rate": 2.1462137588316268e-05,
+      "loss": 0.6417,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6077767398411957,
+      "grad_norm": 0.732385253858184,
+      "learning_rate": 2.142029687627074e-05,
+      "loss": 0.6505,
+      "step": 5205
+    },
+    {
+      "epoch": 0.6083605791686127,
+      "grad_norm": 0.7348856656234368,
+      "learning_rate": 2.1378478831895e-05,
+      "loss": 0.6457,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6089444184960299,
+      "grad_norm": 0.7643931748530564,
+      "learning_rate": 2.133668361110417e-05,
+      "loss": 0.6684,
+      "step": 5215
+    },
+    {
+      "epoch": 0.609528257823447,
+      "grad_norm": 0.6705244603198324,
+      "learning_rate": 2.129491136972826e-05,
+      "loss": 0.6391,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6101120971508641,
+      "grad_norm": 0.7088148504034617,
+      "learning_rate": 2.125316226351163e-05,
+      "loss": 0.6714,
+      "step": 5225
+    },
+    {
+      "epoch": 0.6106959364782811,
+      "grad_norm": 0.6894959554456558,
+      "learning_rate": 2.1211436448112356e-05,
+      "loss": 0.6594,
+      "step": 5230
+    },
+    {
+      "epoch": 0.6112797758056983,
+      "grad_norm": 0.7072742765848938,
+      "learning_rate": 2.1169734079101684e-05,
+      "loss": 0.6652,
+      "step": 5235
+    },
+    {
+      "epoch": 0.6118636151331154,
+      "grad_norm": 0.7247723315350132,
+      "learning_rate": 2.1128055311963453e-05,
+      "loss": 0.6598,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6124474544605325,
+      "grad_norm": 0.735751394291021,
+      "learning_rate": 2.1086400302093483e-05,
+      "loss": 0.6648,
+      "step": 5245
+    },
+    {
+      "epoch": 0.6130312937879495,
+      "grad_norm": 0.6587002357845835,
+      "learning_rate": 2.104476920479905e-05,
+      "loss": 0.6522,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6136151331153666,
+      "grad_norm": 0.7001904520094278,
+      "learning_rate": 2.1003162175298234e-05,
+      "loss": 0.6433,
+      "step": 5255
+    },
+    {
+      "epoch": 0.6141989724427838,
+      "grad_norm": 0.6732104941631006,
+      "learning_rate": 2.0961579368719407e-05,
+      "loss": 0.6618,
+      "step": 5260
+    },
+    {
+      "epoch": 0.6147828117702009,
+      "grad_norm": 0.6626098058920104,
+      "learning_rate": 2.0920020940100626e-05,
+      "loss": 0.6687,
+      "step": 5265
+    },
+    {
+      "epoch": 0.6153666510976179,
+      "grad_norm": 0.6655483296009259,
+      "learning_rate": 2.087848704438905e-05,
+      "loss": 0.6621,
+      "step": 5270
+    },
+    {
+      "epoch": 0.615950490425035,
+      "grad_norm": 0.686843179853701,
+      "learning_rate": 2.0836977836440364e-05,
+      "loss": 0.6427,
+      "step": 5275
+    },
+    {
+      "epoch": 0.6165343297524521,
+      "grad_norm": 0.6688371953183895,
+      "learning_rate": 2.0795493471018222e-05,
+      "loss": 0.6718,
+      "step": 5280
+    },
+    {
+      "epoch": 0.6171181690798693,
+      "grad_norm": 0.6647003616665795,
+      "learning_rate": 2.075403410279364e-05,
+      "loss": 0.6432,
+      "step": 5285
+    },
+    {
+      "epoch": 0.6177020084072863,
+      "grad_norm": 0.6859825058865614,
+      "learning_rate": 2.0712599886344447e-05,
+      "loss": 0.6481,
+      "step": 5290
+    },
+    {
+      "epoch": 0.6182858477347034,
+      "grad_norm": 0.7808965272022199,
+      "learning_rate": 2.067119097615468e-05,
+      "loss": 0.6665,
+      "step": 5295
+    },
+    {
+      "epoch": 0.6188696870621205,
+      "grad_norm": 0.6599205587793768,
+      "learning_rate": 2.0629807526614037e-05,
+      "loss": 0.6712,
+      "step": 5300
+    },
+    {
+      "epoch": 0.6194535263895375,
+      "grad_norm": 0.7272868234971551,
+      "learning_rate": 2.0588449692017287e-05,
+      "loss": 0.6522,
+      "step": 5305
+    },
+    {
+      "epoch": 0.6200373657169547,
+      "grad_norm": 0.6799580957384425,
+      "learning_rate": 2.054711762656369e-05,
+      "loss": 0.6711,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6206212050443718,
+      "grad_norm": 0.6745926690111709,
+      "learning_rate": 2.0505811484356424e-05,
+      "loss": 0.6811,
+      "step": 5315
+    },
+    {
+      "epoch": 0.6212050443717889,
+      "grad_norm": 0.757885080460303,
+      "learning_rate": 2.0464531419402026e-05,
+      "loss": 0.6521,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6217888836992059,
+      "grad_norm": 0.6362954775845878,
+      "learning_rate": 2.0423277585609806e-05,
+      "loss": 0.6591,
+      "step": 5325
+    },
+    {
+      "epoch": 0.622372723026623,
+      "grad_norm": 0.6604487088077978,
+      "learning_rate": 2.038205013679127e-05,
+      "loss": 0.6647,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6229565623540402,
+      "grad_norm": 0.6670476169325874,
+      "learning_rate": 2.034084922665953e-05,
+      "loss": 0.6463,
+      "step": 5335
+    },
+    {
+      "epoch": 0.6235404016814573,
+      "grad_norm": 0.624587303283789,
+      "learning_rate": 2.0299675008828783e-05,
+      "loss": 0.6343,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6241242410088743,
+      "grad_norm": 0.6491739846872022,
+      "learning_rate": 2.025852763681369e-05,
+      "loss": 0.6419,
+      "step": 5345
+    },
+    {
+      "epoch": 0.6247080803362914,
+      "grad_norm": 0.6944978201182703,
+      "learning_rate": 2.021740726402882e-05,
+      "loss": 0.6563,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6252919196637086,
+      "grad_norm": 0.6332299668332454,
+      "learning_rate": 2.0176314043788077e-05,
+      "loss": 0.6624,
+      "step": 5355
+    },
+    {
+      "epoch": 0.6258757589911257,
+      "grad_norm": 0.7384532382114772,
+      "learning_rate": 2.0135248129304124e-05,
+      "loss": 0.6479,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6264595983185427,
+      "grad_norm": 0.643946451086718,
+      "learning_rate": 2.009420967368784e-05,
+      "loss": 0.6397,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6270434376459598,
+      "grad_norm": 0.6322326637102691,
+      "learning_rate": 2.00531988299477e-05,
+      "loss": 0.6664,
+      "step": 5370
+    },
+    {
+      "epoch": 0.627627276973377,
+      "grad_norm": 0.6673557927836997,
+      "learning_rate": 2.0012215750989242e-05,
+      "loss": 0.653,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6282111163007941,
+      "grad_norm": 0.6706784359446668,
+      "learning_rate": 1.997126058961448e-05,
+      "loss": 0.6583,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6287949556282111,
+      "grad_norm": 0.6516995188208667,
+      "learning_rate": 1.9930333498521354e-05,
+      "loss": 0.6444,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6293787949556282,
+      "grad_norm": 0.6825646068596721,
+      "learning_rate": 1.9889434630303118e-05,
+      "loss": 0.6598,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6299626342830453,
+      "grad_norm": 0.6402118165861358,
+      "learning_rate": 1.9848564137447823e-05,
+      "loss": 0.6568,
+      "step": 5395
+    },
+    {
+      "epoch": 0.6305464736104625,
+      "grad_norm": 0.6391671575175067,
+      "learning_rate": 1.9807722172337724e-05,
+      "loss": 0.6502,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6311303129378795,
+      "grad_norm": 0.6310329791350416,
+      "learning_rate": 1.9766908887248697e-05,
+      "loss": 0.6531,
+      "step": 5405
+    },
+    {
+      "epoch": 0.6317141522652966,
+      "grad_norm": 0.6452334020484827,
+      "learning_rate": 1.9726124434349706e-05,
+      "loss": 0.6487,
+      "step": 5410
+    },
+    {
+      "epoch": 0.6322979915927137,
+      "grad_norm": 0.6445433630385896,
+      "learning_rate": 1.9685368965702204e-05,
+      "loss": 0.6711,
+      "step": 5415
+    },
+    {
+      "epoch": 0.6328818309201307,
+      "grad_norm": 0.6345788332891033,
+      "learning_rate": 1.9644642633259575e-05,
+      "loss": 0.6312,
+      "step": 5420
+    },
+    {
+      "epoch": 0.6334656702475479,
+      "grad_norm": 0.7395695651691918,
+      "learning_rate": 1.960394558886659e-05,
+      "loss": 0.6448,
+      "step": 5425
+    },
+    {
+      "epoch": 0.634049509574965,
+      "grad_norm": 0.6926534147470341,
+      "learning_rate": 1.95632779842588e-05,
+      "loss": 0.6539,
+      "step": 5430
+    },
+    {
+      "epoch": 0.6346333489023821,
+      "grad_norm": 0.673046775389808,
+      "learning_rate": 1.9522639971062008e-05,
+      "loss": 0.6704,
+      "step": 5435
+    },
+    {
+      "epoch": 0.6352171882297991,
+      "grad_norm": 0.6096765129669326,
+      "learning_rate": 1.948203170079168e-05,
+      "loss": 0.6622,
+      "step": 5440
+    },
+    {
+      "epoch": 0.6358010275572162,
+      "grad_norm": 0.6034708595881847,
+      "learning_rate": 1.9441453324852387e-05,
+      "loss": 0.6364,
+      "step": 5445
+    },
+    {
+      "epoch": 0.6363848668846334,
+      "grad_norm": 0.6812639408764771,
+      "learning_rate": 1.9400904994537257e-05,
+      "loss": 0.6528,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6369687062120505,
+      "grad_norm": 0.6436990280580072,
+      "learning_rate": 1.936038686102736e-05,
+      "loss": 0.6451,
+      "step": 5455
+    },
+    {
+      "epoch": 0.6375525455394675,
+      "grad_norm": 0.6462508555062616,
+      "learning_rate": 1.931989907539123e-05,
+      "loss": 0.6618,
+      "step": 5460
+    },
+    {
+      "epoch": 0.6381363848668846,
+      "grad_norm": 0.6640329522229098,
+      "learning_rate": 1.92794417885842e-05,
+      "loss": 0.6671,
+      "step": 5465
+    },
+    {
+      "epoch": 0.6387202241943017,
+      "grad_norm": 0.6785919614768156,
+      "learning_rate": 1.9239015151447927e-05,
+      "loss": 0.6517,
+      "step": 5470
+    },
+    {
+      "epoch": 0.6393040635217189,
+      "grad_norm": 0.7048102648283558,
+      "learning_rate": 1.919861931470978e-05,
+      "loss": 0.6591,
+      "step": 5475
+    },
+    {
+      "epoch": 0.6398879028491359,
+      "grad_norm": 0.6494301311911943,
+      "learning_rate": 1.9158254428982293e-05,
+      "loss": 0.6568,
+      "step": 5480
+    },
+    {
+      "epoch": 0.640471742176553,
+      "grad_norm": 0.6424041857418867,
+      "learning_rate": 1.9117920644762594e-05,
+      "loss": 0.6726,
+      "step": 5485
+    },
+    {
+      "epoch": 0.6410555815039701,
+      "grad_norm": 0.7077906276947327,
+      "learning_rate": 1.907761811243186e-05,
+      "loss": 0.6422,
+      "step": 5490
+    },
+    {
+      "epoch": 0.6416394208313873,
+      "grad_norm": 0.6807661269517251,
+      "learning_rate": 1.9037346982254755e-05,
+      "loss": 0.6526,
+      "step": 5495
+    },
+    {
+      "epoch": 0.6422232601588043,
+      "grad_norm": 0.6685336715529324,
+      "learning_rate": 1.8997107404378846e-05,
+      "loss": 0.6568,
+      "step": 5500
+    },
+    {
+      "epoch": 0.6428070994862214,
+      "grad_norm": 0.6269460502274804,
+      "learning_rate": 1.8956899528834065e-05,
+      "loss": 0.639,
+      "step": 5505
+    },
+    {
+      "epoch": 0.6433909388136385,
+      "grad_norm": 0.6712644610457897,
+      "learning_rate": 1.8916723505532157e-05,
+      "loss": 0.6469,
+      "step": 5510
+    },
+    {
+      "epoch": 0.6439747781410555,
+      "grad_norm": 0.6817260060318665,
+      "learning_rate": 1.8876579484266094e-05,
+      "loss": 0.6413,
+      "step": 5515
+    },
+    {
+      "epoch": 0.6445586174684726,
+      "grad_norm": 0.6709064524905491,
+      "learning_rate": 1.8836467614709535e-05,
+      "loss": 0.6603,
+      "step": 5520
+    },
+    {
+      "epoch": 0.6451424567958898,
+      "grad_norm": 0.6642156209672145,
+      "learning_rate": 1.8796388046416253e-05,
+      "loss": 0.6634,
+      "step": 5525
+    },
+    {
+      "epoch": 0.6457262961233069,
+      "grad_norm": 0.6659027841445241,
+      "learning_rate": 1.875634092881963e-05,
+      "loss": 0.6562,
+      "step": 5530
+    },
+    {
+      "epoch": 0.6463101354507239,
+      "grad_norm": 0.7108394628988333,
+      "learning_rate": 1.8716326411232016e-05,
+      "loss": 0.654,
+      "step": 5535
+    },
+    {
+      "epoch": 0.646893974778141,
+      "grad_norm": 0.6802242093433051,
+      "learning_rate": 1.8676344642844217e-05,
+      "loss": 0.6432,
+      "step": 5540
+    },
+    {
+      "epoch": 0.6474778141055582,
+      "grad_norm": 0.7111830262209046,
+      "learning_rate": 1.8636395772724952e-05,
+      "loss": 0.6526,
+      "step": 5545
+    },
+    {
+      "epoch": 0.6480616534329753,
+      "grad_norm": 0.6067533863276159,
+      "learning_rate": 1.8596479949820273e-05,
+      "loss": 0.6529,
+      "step": 5550
+    },
+    {
+      "epoch": 0.6486454927603923,
+      "grad_norm": 0.6230500326750026,
+      "learning_rate": 1.8556597322953035e-05,
+      "loss": 0.6642,
+      "step": 5555
+    },
+    {
+      "epoch": 0.6492293320878094,
+      "grad_norm": 0.6433493991498862,
+      "learning_rate": 1.8516748040822295e-05,
+      "loss": 0.6473,
+      "step": 5560
+    },
+    {
+      "epoch": 0.6498131714152265,
+      "grad_norm": 0.7761064372664181,
+      "learning_rate": 1.847693225200281e-05,
+      "loss": 0.6533,
+      "step": 5565
+    },
+    {
+      "epoch": 0.6503970107426437,
+      "grad_norm": 0.6380806598893206,
+      "learning_rate": 1.843715010494445e-05,
+      "loss": 0.6565,
+      "step": 5570
+    },
+    {
+      "epoch": 0.6509808500700607,
+      "grad_norm": 0.6596355513028087,
+      "learning_rate": 1.839740174797166e-05,
+      "loss": 0.67,
+      "step": 5575
+    },
+    {
+      "epoch": 0.6515646893974778,
+      "grad_norm": 0.6596141423952829,
+      "learning_rate": 1.8357687329282896e-05,
+      "loss": 0.6594,
+      "step": 5580
+    },
+    {
+      "epoch": 0.6521485287248949,
+      "grad_norm": 0.6720639426332328,
+      "learning_rate": 1.831800699695008e-05,
+      "loss": 0.6532,
+      "step": 5585
+    },
+    {
+      "epoch": 0.652732368052312,
+      "grad_norm": 0.6719169509066691,
+      "learning_rate": 1.827836089891805e-05,
+      "loss": 0.6606,
+      "step": 5590
+    },
+    {
+      "epoch": 0.6533162073797291,
+      "grad_norm": 0.6652163538901605,
+      "learning_rate": 1.823874918300399e-05,
+      "loss": 0.6526,
+      "step": 5595
+    },
+    {
+      "epoch": 0.6539000467071462,
+      "grad_norm": 0.6273512495673599,
+      "learning_rate": 1.8199171996896912e-05,
+      "loss": 0.6528,
+      "step": 5600
+    },
+    {
+      "epoch": 0.6544838860345633,
+      "grad_norm": 0.6956985348066894,
+      "learning_rate": 1.8159629488157082e-05,
+      "loss": 0.6711,
+      "step": 5605
+    },
+    {
+      "epoch": 0.6550677253619804,
+      "grad_norm": 0.6749270880541005,
+      "learning_rate": 1.8120121804215466e-05,
+      "loss": 0.6549,
+      "step": 5610
+    },
+    {
+      "epoch": 0.6556515646893974,
+      "grad_norm": 0.6642392459123238,
+      "learning_rate": 1.8080649092373187e-05,
+      "loss": 0.6597,
+      "step": 5615
+    },
+    {
+      "epoch": 0.6562354040168146,
+      "grad_norm": 0.693501672371854,
+      "learning_rate": 1.8041211499800992e-05,
+      "loss": 0.6495,
+      "step": 5620
+    },
+    {
+      "epoch": 0.6568192433442317,
+      "grad_norm": 0.6696143427708985,
+      "learning_rate": 1.8001809173538676e-05,
+      "loss": 0.6462,
+      "step": 5625
+    },
+    {
+      "epoch": 0.6574030826716487,
+      "grad_norm": 0.6402364490146976,
+      "learning_rate": 1.796244226049455e-05,
+      "loss": 0.6558,
+      "step": 5630
+    },
+    {
+      "epoch": 0.6579869219990658,
+      "grad_norm": 0.6518313143198281,
+      "learning_rate": 1.792311090744489e-05,
+      "loss": 0.6248,
+      "step": 5635
+    },
+    {
+      "epoch": 0.658570761326483,
+      "grad_norm": 0.6779344772697246,
+      "learning_rate": 1.7883815261033393e-05,
+      "loss": 0.6534,
+      "step": 5640
+    },
+    {
+      "epoch": 0.6591546006539001,
+      "grad_norm": 0.6534547467921898,
+      "learning_rate": 1.7844555467770624e-05,
+      "loss": 0.667,
+      "step": 5645
+    },
+    {
+      "epoch": 0.6597384399813171,
+      "grad_norm": 0.6271083968726606,
+      "learning_rate": 1.7805331674033466e-05,
+      "loss": 0.6548,
+      "step": 5650
+    },
+    {
+      "epoch": 0.6603222793087342,
+      "grad_norm": 0.6905713709592075,
+      "learning_rate": 1.776614402606459e-05,
+      "loss": 0.6531,
+      "step": 5655
+    },
+    {
+      "epoch": 0.6609061186361513,
+      "grad_norm": 0.6879979135812365,
+      "learning_rate": 1.7726992669971904e-05,
+      "loss": 0.6525,
+      "step": 5660
+    },
+    {
+      "epoch": 0.6614899579635685,
+      "grad_norm": 0.6255401526115754,
+      "learning_rate": 1.768787775172799e-05,
+      "loss": 0.6556,
+      "step": 5665
+    },
+    {
+      "epoch": 0.6620737972909855,
+      "grad_norm": 0.6633550080165926,
+      "learning_rate": 1.7648799417169588e-05,
+      "loss": 0.6552,
+      "step": 5670
+    },
+    {
+      "epoch": 0.6626576366184026,
+      "grad_norm": 0.6926249702866363,
+      "learning_rate": 1.7609757811997023e-05,
+      "loss": 0.6513,
+      "step": 5675
+    },
+    {
+      "epoch": 0.6632414759458197,
+      "grad_norm": 0.6807036634968471,
+      "learning_rate": 1.75707530817737e-05,
+      "loss": 0.6505,
+      "step": 5680
+    },
+    {
+      "epoch": 0.6638253152732368,
+      "grad_norm": 0.6560608306838643,
+      "learning_rate": 1.753178537192551e-05,
+      "loss": 0.6545,
+      "step": 5685
+    },
+    {
+      "epoch": 0.6644091546006539,
+      "grad_norm": 0.6987714592200602,
+      "learning_rate": 1.7492854827740353e-05,
+      "loss": 0.6609,
+      "step": 5690
+    },
+    {
+      "epoch": 0.664992993928071,
+      "grad_norm": 0.6772091217855885,
+      "learning_rate": 1.7453961594367528e-05,
+      "loss": 0.6534,
+      "step": 5695
+    },
+    {
+      "epoch": 0.6655768332554881,
+      "grad_norm": 0.7098236239614998,
+      "learning_rate": 1.741510581681724e-05,
+      "loss": 0.649,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6661606725829052,
+      "grad_norm": 0.7266652414785809,
+      "learning_rate": 1.737628763996005e-05,
+      "loss": 0.6679,
+      "step": 5705
+    },
+    {
+      "epoch": 0.6667445119103222,
+      "grad_norm": 0.7617695462993247,
+      "learning_rate": 1.7337507208526295e-05,
+      "loss": 0.6551,
+      "step": 5710
+    },
+    {
+      "epoch": 0.6673283512377394,
+      "grad_norm": 0.6540180478678356,
+      "learning_rate": 1.729876466710561e-05,
+      "loss": 0.6541,
+      "step": 5715
+    },
+    {
+      "epoch": 0.6679121905651565,
+      "grad_norm": 0.676683188269362,
+      "learning_rate": 1.726006016014637e-05,
+      "loss": 0.6713,
+      "step": 5720
+    },
+    {
+      "epoch": 0.6684960298925736,
+      "grad_norm": 0.6604985309242145,
+      "learning_rate": 1.7221393831955102e-05,
+      "loss": 0.6533,
+      "step": 5725
+    },
+    {
+      "epoch": 0.6690798692199906,
+      "grad_norm": 0.7549157711141019,
+      "learning_rate": 1.718276582669602e-05,
+      "loss": 0.6421,
+      "step": 5730
+    },
+    {
+      "epoch": 0.6696637085474078,
+      "grad_norm": 0.6359539861738778,
+      "learning_rate": 1.7144176288390448e-05,
+      "loss": 0.6425,
+      "step": 5735
+    },
+    {
+      "epoch": 0.6702475478748249,
+      "grad_norm": 0.6531250484152343,
+      "learning_rate": 1.7105625360916276e-05,
+      "loss": 0.6625,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6708313872022419,
+      "grad_norm": 0.6784205320156709,
+      "learning_rate": 1.7067113188007457e-05,
+      "loss": 0.6406,
+      "step": 5745
+    },
+    {
+      "epoch": 0.671415226529659,
+      "grad_norm": 0.634047717162136,
+      "learning_rate": 1.7028639913253426e-05,
+      "loss": 0.6538,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6719990658570761,
+      "grad_norm": 0.6710077949566347,
+      "learning_rate": 1.6990205680098612e-05,
+      "loss": 0.6629,
+      "step": 5755
+    },
+    {
+      "epoch": 0.6725829051844933,
+      "grad_norm": 0.6318362431313652,
+      "learning_rate": 1.695181063184187e-05,
+      "loss": 0.6607,
+      "step": 5760
+    },
+    {
+      "epoch": 0.6731667445119103,
+      "grad_norm": 0.644406552288753,
+      "learning_rate": 1.6913454911635954e-05,
+      "loss": 0.6469,
+      "step": 5765
+    },
+    {
+      "epoch": 0.6737505838393274,
+      "grad_norm": 0.6636822016723358,
+      "learning_rate": 1.6875138662486997e-05,
+      "loss": 0.6422,
+      "step": 5770
+    },
+    {
+      "epoch": 0.6743344231667445,
+      "grad_norm": 0.7029211459350307,
+      "learning_rate": 1.6836862027253963e-05,
+      "loss": 0.6354,
+      "step": 5775
+    },
+    {
+      "epoch": 0.6749182624941616,
+      "grad_norm": 0.7475098263254912,
+      "learning_rate": 1.6798625148648113e-05,
+      "loss": 0.6567,
+      "step": 5780
+    },
+    {
+      "epoch": 0.6755021018215787,
+      "grad_norm": 0.6382798450507685,
+      "learning_rate": 1.6760428169232483e-05,
+      "loss": 0.6381,
+      "step": 5785
+    },
+    {
+      "epoch": 0.6760859411489958,
+      "grad_norm": 0.6153848441637845,
+      "learning_rate": 1.672227123142136e-05,
+      "loss": 0.6588,
+      "step": 5790
+    },
+    {
+      "epoch": 0.6766697804764129,
+      "grad_norm": 0.6186569358031658,
+      "learning_rate": 1.668415447747971e-05,
+      "loss": 0.6342,
+      "step": 5795
+    },
+    {
+      "epoch": 0.67725361980383,
+      "grad_norm": 0.6656027062177072,
+      "learning_rate": 1.6646078049522706e-05,
+      "loss": 0.6396,
+      "step": 5800
+    },
+    {
+      "epoch": 0.677837459131247,
+      "grad_norm": 0.6577646859368105,
+      "learning_rate": 1.660804208951516e-05,
+      "loss": 0.6533,
+      "step": 5805
+    },
+    {
+      "epoch": 0.6784212984586642,
+      "grad_norm": 0.5927313773388592,
+      "learning_rate": 1.6570046739270988e-05,
+      "loss": 0.6531,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6790051377860813,
+      "grad_norm": 0.6091518390067749,
+      "learning_rate": 1.6532092140452725e-05,
+      "loss": 0.6553,
+      "step": 5815
+    },
+    {
+      "epoch": 0.6795889771134984,
+      "grad_norm": 0.613863353598546,
+      "learning_rate": 1.649417843457094e-05,
+      "loss": 0.6383,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6801728164409154,
+      "grad_norm": 0.7011048803452364,
+      "learning_rate": 1.6456305762983742e-05,
+      "loss": 0.6564,
+      "step": 5825
+    },
+    {
+      "epoch": 0.6807566557683326,
+      "grad_norm": 0.6584740117106023,
+      "learning_rate": 1.6418474266896257e-05,
+      "loss": 0.6436,
+      "step": 5830
+    },
+    {
+      "epoch": 0.6813404950957497,
+      "grad_norm": 0.7055538531657063,
+      "learning_rate": 1.6380684087360088e-05,
+      "loss": 0.6332,
+      "step": 5835
+    },
+    {
+      "epoch": 0.6819243344231667,
+      "grad_norm": 0.6552792858883486,
+      "learning_rate": 1.6342935365272785e-05,
+      "loss": 0.6605,
+      "step": 5840
+    },
+    {
+      "epoch": 0.6825081737505838,
+      "grad_norm": 0.6556248652540059,
+      "learning_rate": 1.6305228241377347e-05,
+      "loss": 0.6349,
+      "step": 5845
+    },
+    {
+      "epoch": 0.6830920130780009,
+      "grad_norm": 0.6321426350760702,
+      "learning_rate": 1.6267562856261638e-05,
+      "loss": 0.647,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6836758524054181,
+      "grad_norm": 0.6699560404989997,
+      "learning_rate": 1.6229939350357952e-05,
+      "loss": 0.6434,
+      "step": 5855
+    },
+    {
+      "epoch": 0.6842596917328351,
+      "grad_norm": 0.6345164186545303,
+      "learning_rate": 1.6192357863942415e-05,
+      "loss": 0.6469,
+      "step": 5860
+    },
+    {
+      "epoch": 0.6848435310602522,
+      "grad_norm": 0.6569370988654168,
+      "learning_rate": 1.615481853713448e-05,
+      "loss": 0.634,
+      "step": 5865
+    },
+    {
+      "epoch": 0.6854273703876693,
+      "grad_norm": 0.6997172396660267,
+      "learning_rate": 1.6117321509896422e-05,
+      "loss": 0.6445,
+      "step": 5870
+    },
+    {
+      "epoch": 0.6860112097150864,
+      "grad_norm": 0.632180172250705,
+      "learning_rate": 1.60798669220328e-05,
+      "loss": 0.6419,
+      "step": 5875
+    },
+    {
+      "epoch": 0.6865950490425035,
+      "grad_norm": 0.7019041464086315,
+      "learning_rate": 1.6042454913189946e-05,
+      "loss": 0.6746,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6871788883699206,
+      "grad_norm": 0.6782651171936802,
+      "learning_rate": 1.600508562285544e-05,
+      "loss": 0.672,
+      "step": 5885
+    },
+    {
+      "epoch": 0.6877627276973377,
+      "grad_norm": 0.6173469226028959,
+      "learning_rate": 1.5967759190357585e-05,
+      "loss": 0.6372,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6883465670247548,
+      "grad_norm": 0.6523808993714774,
+      "learning_rate": 1.5930475754864898e-05,
+      "loss": 0.6481,
+      "step": 5895
+    },
+    {
+      "epoch": 0.6889304063521718,
+      "grad_norm": 0.6943963703663875,
+      "learning_rate": 1.5893235455385575e-05,
+      "loss": 0.6613,
+      "step": 5900
+    },
+    {
+      "epoch": 0.689514245679589,
+      "grad_norm": 0.6205909347546901,
+      "learning_rate": 1.5856038430766994e-05,
+      "loss": 0.6504,
+      "step": 5905
+    },
+    {
+      "epoch": 0.6900980850070061,
+      "grad_norm": 0.6577250115332157,
+      "learning_rate": 1.5818884819695184e-05,
+      "loss": 0.643,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6906819243344232,
+      "grad_norm": 0.7131285746606613,
+      "learning_rate": 1.5781774760694304e-05,
+      "loss": 0.6319,
+      "step": 5915
+    },
+    {
+      "epoch": 0.6912657636618402,
+      "grad_norm": 0.6702222134956441,
+      "learning_rate": 1.5744708392126138e-05,
+      "loss": 0.6441,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6918496029892574,
+      "grad_norm": 0.6458299715715866,
+      "learning_rate": 1.5707685852189573e-05,
+      "loss": 0.6546,
+      "step": 5925
+    },
+    {
+      "epoch": 0.6924334423166745,
+      "grad_norm": 0.6467592412179095,
+      "learning_rate": 1.5670707278920084e-05,
+      "loss": 0.6529,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6930172816440916,
+      "grad_norm": 0.66541301231305,
+      "learning_rate": 1.563377281018922e-05,
+      "loss": 0.6447,
+      "step": 5935
+    },
+    {
+      "epoch": 0.6936011209715086,
+      "grad_norm": 0.6988507897107378,
+      "learning_rate": 1.5596882583704092e-05,
+      "loss": 0.6706,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6941849602989257,
+      "grad_norm": 0.6253772210358882,
+      "learning_rate": 1.5560036737006856e-05,
+      "loss": 0.6469,
+      "step": 5945
+    },
+    {
+      "epoch": 0.6947687996263429,
+      "grad_norm": 0.6636080001462308,
+      "learning_rate": 1.5523235407474195e-05,
+      "loss": 0.6516,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6953526389537599,
+      "grad_norm": 0.6904274005251589,
+      "learning_rate": 1.5486478732316827e-05,
+      "loss": 0.6692,
+      "step": 5955
+    },
+    {
+      "epoch": 0.695936478281177,
+      "grad_norm": 0.6225295732898889,
+      "learning_rate": 1.5449766848578968e-05,
+      "loss": 0.6476,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6965203176085941,
+      "grad_norm": 0.678099724634499,
+      "learning_rate": 1.541309989313784e-05,
+      "loss": 0.6688,
+      "step": 5965
+    },
+    {
+      "epoch": 0.6971041569360112,
+      "grad_norm": 0.6507846990956992,
+      "learning_rate": 1.5376478002703154e-05,
+      "loss": 0.655,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6976879962634283,
+      "grad_norm": 0.6491939428328597,
+      "learning_rate": 1.5339901313816584e-05,
+      "loss": 0.6407,
+      "step": 5975
+    },
+    {
+      "epoch": 0.6982718355908454,
+      "grad_norm": 0.6535516438873002,
+      "learning_rate": 1.5303369962851298e-05,
+      "loss": 0.6601,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6988556749182625,
+      "grad_norm": 0.6230417912205047,
+      "learning_rate": 1.5266884086011406e-05,
+      "loss": 0.6554,
+      "step": 5985
+    },
+    {
+      "epoch": 0.6994395142456796,
+      "grad_norm": 0.6135419131501751,
+      "learning_rate": 1.5230443819331492e-05,
+      "loss": 0.6445,
+      "step": 5990
+    },
+    {
+      "epoch": 0.7000233535730966,
+      "grad_norm": 0.6511954072256156,
+      "learning_rate": 1.5194049298676061e-05,
+      "loss": 0.629,
+      "step": 5995
+    },
+    {
+      "epoch": 0.7006071929005138,
+      "grad_norm": 0.6748956312601866,
+      "learning_rate": 1.515770065973907e-05,
+      "loss": 0.6444,
+      "step": 6000
+    },
+    {
+      "epoch": 0.7011910322279309,
+      "grad_norm": 0.643988678018105,
+      "learning_rate": 1.5121398038043421e-05,
+      "loss": 0.6582,
+      "step": 6005
+    },
+    {
+      "epoch": 0.701774871555348,
+      "grad_norm": 0.6643325729889801,
+      "learning_rate": 1.5085141568940419e-05,
+      "loss": 0.6415,
+      "step": 6010
+    },
+    {
+      "epoch": 0.702358710882765,
+      "grad_norm": 0.6061098072733991,
+      "learning_rate": 1.5048931387609321e-05,
+      "loss": 0.6382,
+      "step": 6015
+    },
+    {
+      "epoch": 0.7029425502101821,
+      "grad_norm": 0.6170191771001262,
+      "learning_rate": 1.501276762905679e-05,
+      "loss": 0.6531,
+      "step": 6020
+    },
+    {
+      "epoch": 0.7035263895375993,
+      "grad_norm": 0.7559185866317533,
+      "learning_rate": 1.4976650428116401e-05,
+      "loss": 0.6604,
+      "step": 6025
+    },
+    {
+      "epoch": 0.7041102288650164,
+      "grad_norm": 0.6226993559992735,
+      "learning_rate": 1.4940579919448147e-05,
+      "loss": 0.6448,
+      "step": 6030
+    },
+    {
+      "epoch": 0.7046940681924334,
+      "grad_norm": 0.6305958040557355,
+      "learning_rate": 1.4904556237537936e-05,
+      "loss": 0.683,
+      "step": 6035
+    },
+    {
+      "epoch": 0.7052779075198505,
+      "grad_norm": 0.6589110960306788,
+      "learning_rate": 1.4868579516697079e-05,
+      "loss": 0.6518,
+      "step": 6040
+    },
+    {
+      "epoch": 0.7058617468472677,
+      "grad_norm": 0.6271415611314127,
+      "learning_rate": 1.4832649891061811e-05,
+      "loss": 0.6427,
+      "step": 6045
+    },
+    {
+      "epoch": 0.7064455861746848,
+      "grad_norm": 0.6637185598724493,
+      "learning_rate": 1.4796767494592757e-05,
+      "loss": 0.6467,
+      "step": 6050
+    },
+    {
+      "epoch": 0.7070294255021018,
+      "grad_norm": 0.6167876593558868,
+      "learning_rate": 1.4760932461074467e-05,
+      "loss": 0.642,
+      "step": 6055
+    },
+    {
+      "epoch": 0.7076132648295189,
+      "grad_norm": 0.659548658248431,
+      "learning_rate": 1.4725144924114891e-05,
+      "loss": 0.6369,
+      "step": 6060
+    },
+    {
+      "epoch": 0.708197104156936,
+      "grad_norm": 0.6560272710570144,
+      "learning_rate": 1.4689405017144908e-05,
+      "loss": 0.641,
+      "step": 6065
+    },
+    {
+      "epoch": 0.708780943484353,
+      "grad_norm": 0.7199923240487227,
+      "learning_rate": 1.4653712873417796e-05,
+      "loss": 0.6585,
+      "step": 6070
+    },
+    {
+      "epoch": 0.7093647828117702,
+      "grad_norm": 0.6384648928879884,
+      "learning_rate": 1.4618068626008755e-05,
+      "loss": 0.6331,
+      "step": 6075
+    },
+    {
+      "epoch": 0.7099486221391873,
+      "grad_norm": 0.6373012832164493,
+      "learning_rate": 1.4582472407814419e-05,
+      "loss": 0.6442,
+      "step": 6080
+    },
+    {
+      "epoch": 0.7105324614666044,
+      "grad_norm": 0.5981352121846706,
+      "learning_rate": 1.4546924351552333e-05,
+      "loss": 0.6559,
+      "step": 6085
+    },
+    {
+      "epoch": 0.7111163007940214,
+      "grad_norm": 0.6394867149017587,
+      "learning_rate": 1.4511424589760486e-05,
+      "loss": 0.6361,
+      "step": 6090
+    },
+    {
+      "epoch": 0.7117001401214386,
+      "grad_norm": 0.6479792085825549,
+      "learning_rate": 1.4475973254796799e-05,
+      "loss": 0.6661,
+      "step": 6095
+    },
+    {
+      "epoch": 0.7122839794488557,
+      "grad_norm": 0.6260055427994634,
+      "learning_rate": 1.4440570478838645e-05,
+      "loss": 0.6552,
+      "step": 6100
+    },
+    {
+      "epoch": 0.7128678187762728,
+      "grad_norm": 0.6540376062658972,
+      "learning_rate": 1.440521639388233e-05,
+      "loss": 0.6459,
+      "step": 6105
+    },
+    {
+      "epoch": 0.7134516581036898,
+      "grad_norm": 0.6627728228170229,
+      "learning_rate": 1.436991113174265e-05,
+      "loss": 0.6577,
+      "step": 6110
+    },
+    {
+      "epoch": 0.714035497431107,
+      "grad_norm": 0.6265546084283484,
+      "learning_rate": 1.4334654824052351e-05,
+      "loss": 0.6368,
+      "step": 6115
+    },
+    {
+      "epoch": 0.7146193367585241,
+      "grad_norm": 0.5997797137504358,
+      "learning_rate": 1.429944760226164e-05,
+      "loss": 0.6568,
+      "step": 6120
+    },
+    {
+      "epoch": 0.7152031760859412,
+      "grad_norm": 0.6434752600173411,
+      "learning_rate": 1.4264289597637741e-05,
+      "loss": 0.6604,
+      "step": 6125
+    },
+    {
+      "epoch": 0.7157870154133582,
+      "grad_norm": 0.6311149099041024,
+      "learning_rate": 1.4229180941264364e-05,
+      "loss": 0.6358,
+      "step": 6130
+    },
+    {
+      "epoch": 0.7163708547407753,
+      "grad_norm": 0.6668547608565908,
+      "learning_rate": 1.4194121764041224e-05,
+      "loss": 0.6403,
+      "step": 6135
+    },
+    {
+      "epoch": 0.7169546940681925,
+      "grad_norm": 0.6769321842488776,
+      "learning_rate": 1.4159112196683564e-05,
+      "loss": 0.639,
+      "step": 6140
+    },
+    {
+      "epoch": 0.7175385333956096,
+      "grad_norm": 0.7170192890511556,
+      "learning_rate": 1.4124152369721655e-05,
+      "loss": 0.655,
+      "step": 6145
+    },
+    {
+      "epoch": 0.7181223727230266,
+      "grad_norm": 0.6372648272171578,
+      "learning_rate": 1.408924241350032e-05,
+      "loss": 0.622,
+      "step": 6150
+    },
+    {
+      "epoch": 0.7187062120504437,
+      "grad_norm": 0.6219636894551943,
+      "learning_rate": 1.4054382458178439e-05,
+      "loss": 0.6433,
+      "step": 6155
+    },
+    {
+      "epoch": 0.7192900513778608,
+      "grad_norm": 0.6907119907534172,
+      "learning_rate": 1.4019572633728473e-05,
+      "loss": 0.6538,
+      "step": 6160
+    },
+    {
+      "epoch": 0.7198738907052779,
+      "grad_norm": 0.6671337056917069,
+      "learning_rate": 1.3984813069935967e-05,
+      "loss": 0.6483,
+      "step": 6165
+    },
+    {
+      "epoch": 0.720457730032695,
+      "grad_norm": 0.6141381534242153,
+      "learning_rate": 1.395010389639908e-05,
+      "loss": 0.6588,
+      "step": 6170
+    },
+    {
+      "epoch": 0.7210415693601121,
+      "grad_norm": 0.5936128850781495,
+      "learning_rate": 1.391544524252808e-05,
+      "loss": 0.6411,
+      "step": 6175
+    },
+    {
+      "epoch": 0.7216254086875292,
+      "grad_norm": 0.6480723618326938,
+      "learning_rate": 1.388083723754491e-05,
+      "loss": 0.6507,
+      "step": 6180
+    },
+    {
+      "epoch": 0.7222092480149462,
+      "grad_norm": 0.6484221169199865,
+      "learning_rate": 1.384628001048264e-05,
+      "loss": 0.6429,
+      "step": 6185
+    },
+    {
+      "epoch": 0.7227930873423634,
+      "grad_norm": 0.774949721677888,
+      "learning_rate": 1.381177369018503e-05,
+      "loss": 0.6418,
+      "step": 6190
+    },
+    {
+      "epoch": 0.7233769266697805,
+      "grad_norm": 0.6723770744606823,
+      "learning_rate": 1.377731840530604e-05,
+      "loss": 0.6444,
+      "step": 6195
+    },
+    {
+      "epoch": 0.7239607659971976,
+      "grad_norm": 0.6286827888599001,
+      "learning_rate": 1.374291428430935e-05,
+      "loss": 0.6528,
+      "step": 6200
+    },
+    {
+      "epoch": 0.7245446053246146,
+      "grad_norm": 0.6690865901700592,
+      "learning_rate": 1.3708561455467872e-05,
+      "loss": 0.6472,
+      "step": 6205
+    },
+    {
+      "epoch": 0.7251284446520317,
+      "grad_norm": 0.6206607920597657,
+      "learning_rate": 1.3674260046863285e-05,
+      "loss": 0.6374,
+      "step": 6210
+    },
+    {
+      "epoch": 0.7257122839794489,
+      "grad_norm": 0.6091583021662523,
+      "learning_rate": 1.3640010186385552e-05,
+      "loss": 0.6393,
+      "step": 6215
+    },
+    {
+      "epoch": 0.726296123306866,
+      "grad_norm": 0.6621947921900122,
+      "learning_rate": 1.3605812001732444e-05,
+      "loss": 0.6546,
+      "step": 6220
+    },
+    {
+      "epoch": 0.726879962634283,
+      "grad_norm": 0.6526068291619619,
+      "learning_rate": 1.3571665620409064e-05,
+      "loss": 0.6501,
+      "step": 6225
+    },
+    {
+      "epoch": 0.7274638019617001,
+      "grad_norm": 2.4602752313470493,
+      "learning_rate": 1.3537571169727359e-05,
+      "loss": 0.6393,
+      "step": 6230
+    },
+    {
+      "epoch": 0.7280476412891173,
+      "grad_norm": 0.6356268036541916,
+      "learning_rate": 1.3503528776805676e-05,
+      "loss": 0.6306,
+      "step": 6235
+    },
+    {
+      "epoch": 0.7286314806165344,
+      "grad_norm": 0.6326305074104267,
+      "learning_rate": 1.3469538568568255e-05,
+      "loss": 0.6295,
+      "step": 6240
+    },
+    {
+      "epoch": 0.7292153199439514,
+      "grad_norm": 0.6222833616408877,
+      "learning_rate": 1.3435600671744768e-05,
+      "loss": 0.6283,
+      "step": 6245
+    },
+    {
+      "epoch": 0.7297991592713685,
+      "grad_norm": 0.6133561961949014,
+      "learning_rate": 1.3401715212869864e-05,
+      "loss": 0.6224,
+      "step": 6250
+    },
+    {
+      "epoch": 0.7303829985987856,
+      "grad_norm": 0.6037967169995397,
+      "learning_rate": 1.3367882318282666e-05,
+      "loss": 0.6486,
+      "step": 6255
+    },
+    {
+      "epoch": 0.7309668379262028,
+      "grad_norm": 0.6471850372789167,
+      "learning_rate": 1.3334102114126314e-05,
+      "loss": 0.6519,
+      "step": 6260
+    },
+    {
+      "epoch": 0.7315506772536198,
+      "grad_norm": 0.691632251325258,
+      "learning_rate": 1.330037472634752e-05,
+      "loss": 0.652,
+      "step": 6265
+    },
+    {
+      "epoch": 0.7321345165810369,
+      "grad_norm": 0.646234135333679,
+      "learning_rate": 1.3266700280696042e-05,
+      "loss": 0.6329,
+      "step": 6270
+    },
+    {
+      "epoch": 0.732718355908454,
+      "grad_norm": 0.6198426403579138,
+      "learning_rate": 1.3233078902724266e-05,
+      "loss": 0.6577,
+      "step": 6275
+    },
+    {
+      "epoch": 0.733302195235871,
+      "grad_norm": 0.7320942573246623,
+      "learning_rate": 1.3199510717786714e-05,
+      "loss": 0.6533,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7338860345632882,
+      "grad_norm": 0.6568792492349543,
+      "learning_rate": 1.3165995851039591e-05,
+      "loss": 0.6712,
+      "step": 6285
+    },
+    {
+      "epoch": 0.7344698738907053,
+      "grad_norm": 0.6837902650711637,
+      "learning_rate": 1.3132534427440301e-05,
+      "loss": 0.6426,
+      "step": 6290
+    },
+    {
+      "epoch": 0.7350537132181224,
+      "grad_norm": 0.6962511458886425,
+      "learning_rate": 1.309912657174699e-05,
+      "loss": 0.6495,
+      "step": 6295
+    },
+    {
+      "epoch": 0.7356375525455394,
+      "grad_norm": 0.6218195805719087,
+      "learning_rate": 1.3065772408518085e-05,
+      "loss": 0.6378,
+      "step": 6300
+    },
+    {
+      "epoch": 0.7362213918729565,
+      "grad_norm": 0.6577106692550687,
+      "learning_rate": 1.3032472062111823e-05,
+      "loss": 0.639,
+      "step": 6305
+    },
+    {
+      "epoch": 0.7368052312003737,
+      "grad_norm": 0.6695193352974672,
+      "learning_rate": 1.2999225656685781e-05,
+      "loss": 0.6452,
+      "step": 6310
+    },
+    {
+      "epoch": 0.7373890705277908,
+      "grad_norm": 0.6531494260665233,
+      "learning_rate": 1.2966033316196435e-05,
+      "loss": 0.6539,
+      "step": 6315
+    },
+    {
+      "epoch": 0.7379729098552078,
+      "grad_norm": 0.6108770743645784,
+      "learning_rate": 1.2932895164398684e-05,
+      "loss": 0.6473,
+      "step": 6320
+    },
+    {
+      "epoch": 0.7385567491826249,
+      "grad_norm": 0.6125085475187015,
+      "learning_rate": 1.2899811324845373e-05,
+      "loss": 0.6457,
+      "step": 6325
+    },
+    {
+      "epoch": 0.739140588510042,
+      "grad_norm": 0.6373905010866605,
+      "learning_rate": 1.2866781920886873e-05,
+      "loss": 0.6608,
+      "step": 6330
+    },
+    {
+      "epoch": 0.7397244278374592,
+      "grad_norm": 0.6187004994232301,
+      "learning_rate": 1.2833807075670564e-05,
+      "loss": 0.6462,
+      "step": 6335
+    },
+    {
+      "epoch": 0.7403082671648762,
+      "grad_norm": 0.6461021466819432,
+      "learning_rate": 1.2800886912140433e-05,
+      "loss": 0.6439,
+      "step": 6340
+    },
+    {
+      "epoch": 0.7408921064922933,
+      "grad_norm": 0.664306032806025,
+      "learning_rate": 1.2768021553036596e-05,
+      "loss": 0.6286,
+      "step": 6345
+    },
+    {
+      "epoch": 0.7414759458197104,
+      "grad_norm": 0.6983665960240255,
+      "learning_rate": 1.2735211120894813e-05,
+      "loss": 0.648,
+      "step": 6350
+    },
+    {
+      "epoch": 0.7420597851471276,
+      "grad_norm": 0.6199633025878769,
+      "learning_rate": 1.2702455738046068e-05,
+      "loss": 0.6481,
+      "step": 6355
+    },
+    {
+      "epoch": 0.7426436244745446,
+      "grad_norm": 0.6938764591430603,
+      "learning_rate": 1.2669755526616093e-05,
+      "loss": 0.6588,
+      "step": 6360
+    },
+    {
+      "epoch": 0.7432274638019617,
+      "grad_norm": 0.7073147544566707,
+      "learning_rate": 1.2637110608524916e-05,
+      "loss": 0.6221,
+      "step": 6365
+    },
+    {
+      "epoch": 0.7438113031293788,
+      "grad_norm": 0.6894625288865952,
+      "learning_rate": 1.2604521105486417e-05,
+      "loss": 0.641,
+      "step": 6370
+    },
+    {
+      "epoch": 0.7443951424567959,
+      "grad_norm": 0.6006031068931637,
+      "learning_rate": 1.2571987139007856e-05,
+      "loss": 0.6429,
+      "step": 6375
+    },
+    {
+      "epoch": 0.744978981784213,
+      "grad_norm": 0.6552582169334152,
+      "learning_rate": 1.253950883038944e-05,
+      "loss": 0.6517,
+      "step": 6380
+    },
+    {
+      "epoch": 0.7455628211116301,
+      "grad_norm": 0.6397535486475485,
+      "learning_rate": 1.2507086300723846e-05,
+      "loss": 0.6541,
+      "step": 6385
+    },
+    {
+      "epoch": 0.7461466604390472,
+      "grad_norm": 0.6902838347118688,
+      "learning_rate": 1.2474719670895796e-05,
+      "loss": 0.6468,
+      "step": 6390
+    },
+    {
+      "epoch": 0.7467304997664642,
+      "grad_norm": 0.6896130002515913,
+      "learning_rate": 1.2442409061581587e-05,
+      "loss": 0.6275,
+      "step": 6395
+    },
+    {
+      "epoch": 0.7473143390938813,
+      "grad_norm": 0.6540524402657386,
+      "learning_rate": 1.2410154593248657e-05,
+      "loss": 0.6257,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7478981784212985,
+      "grad_norm": 0.6346234250470586,
+      "learning_rate": 1.2377956386155114e-05,
+      "loss": 0.6415,
+      "step": 6405
+    },
+    {
+      "epoch": 0.7484820177487156,
+      "grad_norm": 0.6812865894965642,
+      "learning_rate": 1.2345814560349316e-05,
+      "loss": 0.6569,
+      "step": 6410
+    },
+    {
+      "epoch": 0.7490658570761326,
+      "grad_norm": 0.7266805424106589,
+      "learning_rate": 1.231372923566939e-05,
+      "loss": 0.6433,
+      "step": 6415
+    },
+    {
+      "epoch": 0.7496496964035497,
+      "grad_norm": 0.6985012316377752,
+      "learning_rate": 1.2281700531742818e-05,
+      "loss": 0.6503,
+      "step": 6420
+    },
+    {
+      "epoch": 0.7502335357309668,
+      "grad_norm": 0.6279926697643428,
+      "learning_rate": 1.2249728567985966e-05,
+      "loss": 0.6516,
+      "step": 6425
+    },
+    {
+      "epoch": 0.750817375058384,
+      "grad_norm": 0.6823315113068185,
+      "learning_rate": 1.2217813463603664e-05,
+      "loss": 0.6479,
+      "step": 6430
+    },
+    {
+      "epoch": 0.751401214385801,
+      "grad_norm": 0.6012248291651592,
+      "learning_rate": 1.2185955337588727e-05,
+      "loss": 0.6334,
+      "step": 6435
+    },
+    {
+      "epoch": 0.7519850537132181,
+      "grad_norm": 0.6823126280567675,
+      "learning_rate": 1.2154154308721546e-05,
+      "loss": 0.6418,
+      "step": 6440
+    },
+    {
+      "epoch": 0.7525688930406352,
+      "grad_norm": 0.6157410397011505,
+      "learning_rate": 1.2122410495569623e-05,
+      "loss": 0.6399,
+      "step": 6445
+    },
+    {
+      "epoch": 0.7531527323680524,
+      "grad_norm": 0.6607304490574921,
+      "learning_rate": 1.2090724016487137e-05,
+      "loss": 0.6613,
+      "step": 6450
+    },
+    {
+      "epoch": 0.7537365716954694,
+      "grad_norm": 0.6448381380250503,
+      "learning_rate": 1.2059094989614503e-05,
+      "loss": 0.6402,
+      "step": 6455
+    },
+    {
+      "epoch": 0.7543204110228865,
+      "grad_norm": 0.72890234313802,
+      "learning_rate": 1.2027523532877928e-05,
+      "loss": 0.6339,
+      "step": 6460
+    },
+    {
+      "epoch": 0.7549042503503036,
+      "grad_norm": 0.6795166242573744,
+      "learning_rate": 1.1996009763988974e-05,
+      "loss": 0.6309,
+      "step": 6465
+    },
+    {
+      "epoch": 0.7554880896777207,
+      "grad_norm": 0.6893940287221781,
+      "learning_rate": 1.1964553800444123e-05,
+      "loss": 0.6471,
+      "step": 6470
+    },
+    {
+      "epoch": 0.7560719290051378,
+      "grad_norm": 0.634194352566182,
+      "learning_rate": 1.1933155759524332e-05,
+      "loss": 0.6387,
+      "step": 6475
+    },
+    {
+      "epoch": 0.7566557683325549,
+      "grad_norm": 0.7448512153381929,
+      "learning_rate": 1.1901815758294589e-05,
+      "loss": 0.6345,
+      "step": 6480
+    },
+    {
+      "epoch": 0.757239607659972,
+      "grad_norm": 0.763372636963995,
+      "learning_rate": 1.18705339136035e-05,
+      "loss": 0.6579,
+      "step": 6485
+    },
+    {
+      "epoch": 0.757823446987389,
+      "grad_norm": 0.6862268846936076,
+      "learning_rate": 1.1839310342082835e-05,
+      "loss": 0.6688,
+      "step": 6490
+    },
+    {
+      "epoch": 0.7584072863148061,
+      "grad_norm": 0.6320332817548495,
+      "learning_rate": 1.1808145160147092e-05,
+      "loss": 0.618,
+      "step": 6495
+    },
+    {
+      "epoch": 0.7589911256422233,
+      "grad_norm": 0.6133186430419542,
+      "learning_rate": 1.1777038483993066e-05,
+      "loss": 0.6359,
+      "step": 6500
+    },
+    {
+      "epoch": 0.7595749649696404,
+      "grad_norm": 0.6508243392201695,
+      "learning_rate": 1.1745990429599439e-05,
+      "loss": 0.658,
+      "step": 6505
+    },
+    {
+      "epoch": 0.7601588042970574,
+      "grad_norm": 0.6605087610839377,
+      "learning_rate": 1.1715001112726304e-05,
+      "loss": 0.6095,
+      "step": 6510
+    },
+    {
+      "epoch": 0.7607426436244745,
+      "grad_norm": 0.629256704641095,
+      "learning_rate": 1.1684070648914763e-05,
+      "loss": 0.651,
+      "step": 6515
+    },
+    {
+      "epoch": 0.7613264829518916,
+      "grad_norm": 0.6021258354982866,
+      "learning_rate": 1.1653199153486488e-05,
+      "loss": 0.6345,
+      "step": 6520
+    },
+    {
+      "epoch": 0.7619103222793088,
+      "grad_norm": 0.6659280501504915,
+      "learning_rate": 1.1622386741543295e-05,
+      "loss": 0.6344,
+      "step": 6525
+    },
+    {
+      "epoch": 0.7624941616067258,
+      "grad_norm": 0.6831457044400984,
+      "learning_rate": 1.1591633527966713e-05,
+      "loss": 0.6467,
+      "step": 6530
+    },
+    {
+      "epoch": 0.7630780009341429,
+      "grad_norm": 0.672942340382294,
+      "learning_rate": 1.1560939627417555e-05,
+      "loss": 0.6515,
+      "step": 6535
+    },
+    {
+      "epoch": 0.76366184026156,
+      "grad_norm": 0.6744624848853943,
+      "learning_rate": 1.1530305154335482e-05,
+      "loss": 0.633,
+      "step": 6540
+    },
+    {
+      "epoch": 0.7642456795889772,
+      "grad_norm": 0.6579110785895681,
+      "learning_rate": 1.1499730222938595e-05,
+      "loss": 0.6431,
+      "step": 6545
+    },
+    {
+      "epoch": 0.7648295189163942,
+      "grad_norm": 0.6262168723984547,
+      "learning_rate": 1.1469214947222993e-05,
+      "loss": 0.6359,
+      "step": 6550
+    },
+    {
+      "epoch": 0.7654133582438113,
+      "grad_norm": 0.6528034993009432,
+      "learning_rate": 1.1438759440962353e-05,
+      "loss": 0.6242,
+      "step": 6555
+    },
+    {
+      "epoch": 0.7659971975712284,
+      "grad_norm": 0.6200492980546738,
+      "learning_rate": 1.1408363817707523e-05,
+      "loss": 0.6453,
+      "step": 6560
+    },
+    {
+      "epoch": 0.7665810368986455,
+      "grad_norm": 0.7120307987935346,
+      "learning_rate": 1.1378028190786045e-05,
+      "loss": 0.6332,
+      "step": 6565
+    },
+    {
+      "epoch": 0.7671648762260626,
+      "grad_norm": 0.6343103468365546,
+      "learning_rate": 1.134775267330181e-05,
+      "loss": 0.6217,
+      "step": 6570
+    },
+    {
+      "epoch": 0.7677487155534797,
+      "grad_norm": 0.6443793067325617,
+      "learning_rate": 1.1317537378134568e-05,
+      "loss": 0.651,
+      "step": 6575
+    },
+    {
+      "epoch": 0.7683325548808968,
+      "grad_norm": 0.618840567288472,
+      "learning_rate": 1.1287382417939555e-05,
+      "loss": 0.6548,
+      "step": 6580
+    },
+    {
+      "epoch": 0.7689163942083139,
+      "grad_norm": 0.6460389555802754,
+      "learning_rate": 1.1257287905147035e-05,
+      "loss": 0.6326,
+      "step": 6585
+    },
+    {
+      "epoch": 0.7695002335357309,
+      "grad_norm": 0.5810628693651504,
+      "learning_rate": 1.1227253951961911e-05,
+      "loss": 0.6204,
+      "step": 6590
+    },
+    {
+      "epoch": 0.7700840728631481,
+      "grad_norm": 0.5872874463618635,
+      "learning_rate": 1.1197280670363297e-05,
+      "loss": 0.6376,
+      "step": 6595
+    },
+    {
+      "epoch": 0.7706679121905652,
+      "grad_norm": 0.6086855468161991,
+      "learning_rate": 1.1167368172104084e-05,
+      "loss": 0.64,
+      "step": 6600
+    },
+    {
+      "epoch": 0.7712517515179822,
+      "grad_norm": 0.6218777515538456,
+      "learning_rate": 1.1137516568710548e-05,
+      "loss": 0.6462,
+      "step": 6605
+    },
+    {
+      "epoch": 0.7718355908453993,
+      "grad_norm": 0.6036656429455959,
+      "learning_rate": 1.1107725971481923e-05,
+      "loss": 0.6391,
+      "step": 6610
+    },
+    {
+      "epoch": 0.7724194301728164,
+      "grad_norm": 0.5855667797586916,
+      "learning_rate": 1.107799649148998e-05,
+      "loss": 0.6282,
+      "step": 6615
+    },
+    {
+      "epoch": 0.7730032695002336,
+      "grad_norm": 0.6470380891791336,
+      "learning_rate": 1.1048328239578631e-05,
+      "loss": 0.6409,
+      "step": 6620
+    },
+    {
+      "epoch": 0.7735871088276506,
+      "grad_norm": 0.6152164633750634,
+      "learning_rate": 1.1018721326363493e-05,
+      "loss": 0.629,
+      "step": 6625
+    },
+    {
+      "epoch": 0.7741709481550677,
+      "grad_norm": 0.67733200025291,
+      "learning_rate": 1.0989175862231488e-05,
+      "loss": 0.651,
+      "step": 6630
+    },
+    {
+      "epoch": 0.7747547874824848,
+      "grad_norm": 1.146412390170155,
+      "learning_rate": 1.095969195734044e-05,
+      "loss": 0.6529,
+      "step": 6635
+    },
+    {
+      "epoch": 0.775338626809902,
+      "grad_norm": 0.59037588657831,
+      "learning_rate": 1.0930269721618641e-05,
+      "loss": 0.6113,
+      "step": 6640
+    },
+    {
+      "epoch": 0.775922466137319,
+      "grad_norm": 0.6131068191439204,
+      "learning_rate": 1.0900909264764463e-05,
+      "loss": 0.6272,
+      "step": 6645
+    },
+    {
+      "epoch": 0.7765063054647361,
+      "grad_norm": 0.6254292350299556,
+      "learning_rate": 1.0871610696245941e-05,
+      "loss": 0.6588,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7770901447921532,
+      "grad_norm": 0.6513113097969633,
+      "learning_rate": 1.0842374125300364e-05,
+      "loss": 0.6484,
+      "step": 6655
+    },
+    {
+      "epoch": 0.7776739841195703,
+      "grad_norm": 0.6751648149069668,
+      "learning_rate": 1.081319966093386e-05,
+      "loss": 0.6664,
+      "step": 6660
+    },
+    {
+      "epoch": 0.7782578234469874,
+      "grad_norm": 0.6268332764660168,
+      "learning_rate": 1.0784087411921e-05,
+      "loss": 0.6215,
+      "step": 6665
+    },
+    {
+      "epoch": 0.7788416627744045,
+      "grad_norm": 0.6030514136330601,
+      "learning_rate": 1.0755037486804411e-05,
+      "loss": 0.6459,
+      "step": 6670
+    },
+    {
+      "epoch": 0.7794255021018216,
+      "grad_norm": 0.6059821182693572,
+      "learning_rate": 1.0726049993894324e-05,
+      "loss": 0.6296,
+      "step": 6675
+    },
+    {
+      "epoch": 0.7800093414292387,
+      "grad_norm": 0.6358704318338096,
+      "learning_rate": 1.0697125041268207e-05,
+      "loss": 0.6284,
+      "step": 6680
+    },
+    {
+      "epoch": 0.7805931807566557,
+      "grad_norm": 0.623055096194753,
+      "learning_rate": 1.0668262736770356e-05,
+      "loss": 0.6318,
+      "step": 6685
+    },
+    {
+      "epoch": 0.7811770200840729,
+      "grad_norm": 0.6221943017251522,
+      "learning_rate": 1.0639463188011476e-05,
+      "loss": 0.6553,
+      "step": 6690
+    },
+    {
+      "epoch": 0.78176085941149,
+      "grad_norm": 0.6274165949170496,
+      "learning_rate": 1.0610726502368303e-05,
+      "loss": 0.6499,
+      "step": 6695
+    },
+    {
+      "epoch": 0.7823446987389071,
+      "grad_norm": 0.6558366089961987,
+      "learning_rate": 1.0582052786983194e-05,
+      "loss": 0.642,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7829285380663241,
+      "grad_norm": 0.6168624390569022,
+      "learning_rate": 1.0553442148763725e-05,
+      "loss": 0.6635,
+      "step": 6705
+    },
+    {
+      "epoch": 0.7835123773937412,
+      "grad_norm": 0.660505645249963,
+      "learning_rate": 1.0524894694382284e-05,
+      "loss": 0.6347,
+      "step": 6710
+    },
+    {
+      "epoch": 0.7840962167211584,
+      "grad_norm": 0.6422313039212652,
+      "learning_rate": 1.0496410530275694e-05,
+      "loss": 0.6387,
+      "step": 6715
+    },
+    {
+      "epoch": 0.7846800560485754,
+      "grad_norm": 0.6327033532056747,
+      "learning_rate": 1.0467989762644803e-05,
+      "loss": 0.6277,
+      "step": 6720
+    },
+    {
+      "epoch": 0.7852638953759925,
+      "grad_norm": 0.6472227494580378,
+      "learning_rate": 1.0439632497454093e-05,
+      "loss": 0.643,
+      "step": 6725
+    },
+    {
+      "epoch": 0.7858477347034096,
+      "grad_norm": 0.6216051002622686,
+      "learning_rate": 1.0411338840431278e-05,
+      "loss": 0.6412,
+      "step": 6730
+    },
+    {
+      "epoch": 0.7864315740308268,
+      "grad_norm": 0.6313937791651838,
+      "learning_rate": 1.0383108897066915e-05,
+      "loss": 0.641,
+      "step": 6735
+    },
+    {
+      "epoch": 0.7870154133582438,
+      "grad_norm": 0.633226750358156,
+      "learning_rate": 1.035494277261401e-05,
+      "loss": 0.6113,
+      "step": 6740
+    },
+    {
+      "epoch": 0.7875992526856609,
+      "grad_norm": 0.6101629183833445,
+      "learning_rate": 1.0326840572087633e-05,
+      "loss": 0.627,
+      "step": 6745
+    },
+    {
+      "epoch": 0.788183092013078,
+      "grad_norm": 0.603736417034934,
+      "learning_rate": 1.0298802400264502e-05,
+      "loss": 0.6303,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7887669313404951,
+      "grad_norm": 0.6397082011815393,
+      "learning_rate": 1.0270828361682628e-05,
+      "loss": 0.6386,
+      "step": 6755
+    },
+    {
+      "epoch": 0.7893507706679121,
+      "grad_norm": 0.6666697928521588,
+      "learning_rate": 1.0242918560640893e-05,
+      "loss": 0.6172,
+      "step": 6760
+    },
+    {
+      "epoch": 0.7899346099953293,
+      "grad_norm": 0.6589671968178897,
+      "learning_rate": 1.0215073101198683e-05,
+      "loss": 0.6366,
+      "step": 6765
+    },
+    {
+      "epoch": 0.7905184493227464,
+      "grad_norm": 0.6257936502758339,
+      "learning_rate": 1.0187292087175485e-05,
+      "loss": 0.6245,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7911022886501635,
+      "grad_norm": 0.6265837040701255,
+      "learning_rate": 1.0159575622150513e-05,
+      "loss": 0.6409,
+      "step": 6775
+    },
+    {
+      "epoch": 0.7916861279775805,
+      "grad_norm": 0.6374339844464507,
+      "learning_rate": 1.0131923809462313e-05,
+      "loss": 0.6481,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7922699673049977,
+      "grad_norm": 0.6621139588212137,
+      "learning_rate": 1.0104336752208374e-05,
+      "loss": 0.6197,
+      "step": 6785
+    },
+    {
+      "epoch": 0.7928538066324148,
+      "grad_norm": 1.1720758186676021,
+      "learning_rate": 1.0076814553244762e-05,
+      "loss": 0.6477,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7934376459598319,
+      "grad_norm": 0.6112390664794839,
+      "learning_rate": 1.0049357315185711e-05,
+      "loss": 0.6368,
+      "step": 6795
+    },
+    {
+      "epoch": 0.7940214852872489,
+      "grad_norm": 0.6025289483169135,
+      "learning_rate": 1.0021965140403267e-05,
+      "loss": 0.6117,
+      "step": 6800
+    },
+    {
+      "epoch": 0.794605324614666,
+      "grad_norm": 0.6585643058560275,
+      "learning_rate": 9.99463813102688e-06,
+      "loss": 0.6222,
+      "step": 6805
+    },
+    {
+      "epoch": 0.7951891639420832,
+      "grad_norm": 0.6715809889446958,
+      "learning_rate": 9.967376388943042e-06,
+      "loss": 0.6418,
+      "step": 6810
+    },
+    {
+      "epoch": 0.7957730032695002,
+      "grad_norm": 0.6217978701228573,
+      "learning_rate": 9.940180015794908e-06,
+      "loss": 0.6577,
+      "step": 6815
+    },
+    {
+      "epoch": 0.7963568425969173,
+      "grad_norm": 0.6183066216767451,
+      "learning_rate": 9.913049112981897e-06,
+      "loss": 0.6496,
+      "step": 6820
+    },
+    {
+      "epoch": 0.7969406819243344,
+      "grad_norm": 0.6099618127843874,
+      "learning_rate": 9.885983781659332e-06,
+      "loss": 0.6558,
+      "step": 6825
+    },
+    {
+      "epoch": 0.7975245212517516,
+      "grad_norm": 0.658463790742885,
+      "learning_rate": 9.858984122738072e-06,
+      "loss": 0.627,
+      "step": 6830
+    },
+    {
+      "epoch": 0.7981083605791686,
+      "grad_norm": 0.6615276020621076,
+      "learning_rate": 9.832050236884102e-06,
+      "loss": 0.6497,
+      "step": 6835
+    },
+    {
+      "epoch": 0.7986921999065857,
+      "grad_norm": 0.6259325557193028,
+      "learning_rate": 9.805182224518186e-06,
+      "loss": 0.6414,
+      "step": 6840
+    },
+    {
+      "epoch": 0.7992760392340028,
+      "grad_norm": 0.5925039243497241,
+      "learning_rate": 9.778380185815486e-06,
+      "loss": 0.6381,
+      "step": 6845
+    },
+    {
+      "epoch": 0.7998598785614199,
+      "grad_norm": 0.6550300289531977,
+      "learning_rate": 9.751644220705187e-06,
+      "loss": 0.5937,
+      "step": 6850
+    },
+    {
+      "epoch": 0.800443717888837,
+      "grad_norm": 0.6498679334529014,
+      "learning_rate": 9.72497442887012e-06,
+      "loss": 0.6589,
+      "step": 6855
+    },
+    {
+      "epoch": 0.8010275572162541,
+      "grad_norm": 0.6258303217846111,
+      "learning_rate": 9.698370909746387e-06,
+      "loss": 0.6314,
+      "step": 6860
+    },
+    {
+      "epoch": 0.8016113965436712,
+      "grad_norm": 0.6110263661620758,
+      "learning_rate": 9.671833762523016e-06,
+      "loss": 0.6422,
+      "step": 6865
+    },
+    {
+      "epoch": 0.8021952358710883,
+      "grad_norm": 0.6604564267266404,
+      "learning_rate": 9.645363086141561e-06,
+      "loss": 0.6687,
+      "step": 6870
+    },
+    {
+      "epoch": 0.8027790751985053,
+      "grad_norm": 0.6133261669233432,
+      "learning_rate": 9.618958979295747e-06,
+      "loss": 0.6495,
+      "step": 6875
+    },
+    {
+      "epoch": 0.8033629145259225,
+      "grad_norm": 0.5805003568606534,
+      "learning_rate": 9.592621540431101e-06,
+      "loss": 0.6325,
+      "step": 6880
+    },
+    {
+      "epoch": 0.8039467538533396,
+      "grad_norm": 0.6016490003111251,
+      "learning_rate": 9.566350867744584e-06,
+      "loss": 0.6496,
+      "step": 6885
+    },
+    {
+      "epoch": 0.8045305931807567,
+      "grad_norm": 0.5860139017883728,
+      "learning_rate": 9.540147059184226e-06,
+      "loss": 0.6305,
+      "step": 6890
+    },
+    {
+      "epoch": 0.8051144325081737,
+      "grad_norm": 0.7230562270791053,
+      "learning_rate": 9.514010212448751e-06,
+      "loss": 0.6328,
+      "step": 6895
+    },
+    {
+      "epoch": 0.8056982718355908,
+      "grad_norm": 0.5980225557666627,
+      "learning_rate": 9.487940424987235e-06,
+      "loss": 0.6408,
+      "step": 6900
+    },
+    {
+      "epoch": 0.806282111163008,
+      "grad_norm": 0.6002231741684171,
+      "learning_rate": 9.461937793998723e-06,
+      "loss": 0.656,
+      "step": 6905
+    },
+    {
+      "epoch": 0.8068659504904251,
+      "grad_norm": 0.6367968204118127,
+      "learning_rate": 9.436002416431868e-06,
+      "loss": 0.6306,
+      "step": 6910
+    },
+    {
+      "epoch": 0.8074497898178421,
+      "grad_norm": 0.6065896754400874,
+      "learning_rate": 9.41013438898458e-06,
+      "loss": 0.632,
+      "step": 6915
+    },
+    {
+      "epoch": 0.8080336291452592,
+      "grad_norm": 0.6605091785427533,
+      "learning_rate": 9.384333808103656e-06,
+      "loss": 0.6321,
+      "step": 6920
+    },
+    {
+      "epoch": 0.8086174684726763,
+      "grad_norm": 0.649728718647983,
+      "learning_rate": 9.358600769984432e-06,
+      "loss": 0.6358,
+      "step": 6925
+    },
+    {
+      "epoch": 0.8092013078000934,
+      "grad_norm": 0.6052521632195271,
+      "learning_rate": 9.332935370570402e-06,
+      "loss": 0.6501,
+      "step": 6930
+    },
+    {
+      "epoch": 0.8097851471275105,
+      "grad_norm": 0.6327178716698119,
+      "learning_rate": 9.30733770555289e-06,
+      "loss": 0.6401,
+      "step": 6935
+    },
+    {
+      "epoch": 0.8103689864549276,
+      "grad_norm": 0.6531262103718444,
+      "learning_rate": 9.281807870370666e-06,
+      "loss": 0.6448,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8109528257823447,
+      "grad_norm": 0.6301419895570477,
+      "learning_rate": 9.256345960209608e-06,
+      "loss": 0.6264,
+      "step": 6945
+    },
+    {
+      "epoch": 0.8115366651097617,
+      "grad_norm": 0.6963741903466704,
+      "learning_rate": 9.23095207000234e-06,
+      "loss": 0.6741,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8121205044371789,
+      "grad_norm": 0.6464706174782056,
+      "learning_rate": 9.205626294427885e-06,
+      "loss": 0.63,
+      "step": 6955
+    },
+    {
+      "epoch": 0.812704343764596,
+      "grad_norm": 0.6334091306063494,
+      "learning_rate": 9.18036872791129e-06,
+      "loss": 0.6497,
+      "step": 6960
+    },
+    {
+      "epoch": 0.8132881830920131,
+      "grad_norm": 0.6022796874371679,
+      "learning_rate": 9.155179464623312e-06,
+      "loss": 0.6326,
+      "step": 6965
+    },
+    {
+      "epoch": 0.8138720224194301,
+      "grad_norm": 0.6427806776747196,
+      "learning_rate": 9.130058598480027e-06,
+      "loss": 0.6307,
+      "step": 6970
+    },
+    {
+      "epoch": 0.8144558617468473,
+      "grad_norm": 0.6393493440257442,
+      "learning_rate": 9.105006223142507e-06,
+      "loss": 0.6296,
+      "step": 6975
+    },
+    {
+      "epoch": 0.8150397010742644,
+      "grad_norm": 0.6183999459604124,
+      "learning_rate": 9.080022432016457e-06,
+      "loss": 0.6277,
+      "step": 6980
+    },
+    {
+      "epoch": 0.8156235404016815,
+      "grad_norm": 0.599802408180061,
+      "learning_rate": 9.05510731825188e-06,
+      "loss": 0.6452,
+      "step": 6985
+    },
+    {
+      "epoch": 0.8162073797290985,
+      "grad_norm": 0.6467508872801025,
+      "learning_rate": 9.030260974742701e-06,
+      "loss": 0.6469,
+      "step": 6990
+    },
+    {
+      "epoch": 0.8167912190565156,
+      "grad_norm": 0.6400772659223442,
+      "learning_rate": 9.005483494126474e-06,
+      "loss": 0.6382,
+      "step": 6995
+    },
+    {
+      "epoch": 0.8173750583839328,
+      "grad_norm": 0.6070945909106871,
+      "learning_rate": 8.980774968783978e-06,
+      "loss": 0.6492,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8179588977113499,
+      "grad_norm": 0.6348123346621048,
+      "learning_rate": 8.9561354908389e-06,
+      "loss": 0.6478,
+      "step": 7005
+    },
+    {
+      "epoch": 0.8185427370387669,
+      "grad_norm": 0.69411335593364,
+      "learning_rate": 8.931565152157492e-06,
+      "loss": 0.6511,
+      "step": 7010
+    },
+    {
+      "epoch": 0.819126576366184,
+      "grad_norm": 0.6676483425443996,
+      "learning_rate": 8.907064044348232e-06,
+      "loss": 0.6458,
+      "step": 7015
+    },
+    {
+      "epoch": 0.8197104156936011,
+      "grad_norm": 0.6295085789424556,
+      "learning_rate": 8.88263225876147e-06,
+      "loss": 0.6046,
+      "step": 7020
+    },
+    {
+      "epoch": 0.8202942550210183,
+      "grad_norm": 0.589645571190248,
+      "learning_rate": 8.858269886489099e-06,
+      "loss": 0.6126,
+      "step": 7025
+    },
+    {
+      "epoch": 0.8208780943484353,
+      "grad_norm": 0.6212383481031892,
+      "learning_rate": 8.8339770183642e-06,
+      "loss": 0.6433,
+      "step": 7030
+    },
+    {
+      "epoch": 0.8214619336758524,
+      "grad_norm": 0.6121504319306436,
+      "learning_rate": 8.809753744960733e-06,
+      "loss": 0.647,
+      "step": 7035
+    },
+    {
+      "epoch": 0.8220457730032695,
+      "grad_norm": 0.6020644990184818,
+      "learning_rate": 8.785600156593157e-06,
+      "loss": 0.632,
+      "step": 7040
+    },
+    {
+      "epoch": 0.8226296123306865,
+      "grad_norm": 0.612215270277333,
+      "learning_rate": 8.761516343316131e-06,
+      "loss": 0.6439,
+      "step": 7045
+    },
+    {
+      "epoch": 0.8232134516581037,
+      "grad_norm": 0.6111428805532173,
+      "learning_rate": 8.737502394924158e-06,
+      "loss": 0.6374,
+      "step": 7050
+    },
+    {
+      "epoch": 0.8237972909855208,
+      "grad_norm": 0.6324321344274677,
+      "learning_rate": 8.713558400951254e-06,
+      "loss": 0.6315,
+      "step": 7055
+    },
+    {
+      "epoch": 0.8243811303129379,
+      "grad_norm": 0.650140068168635,
+      "learning_rate": 8.689684450670627e-06,
+      "loss": 0.6447,
+      "step": 7060
+    },
+    {
+      "epoch": 0.8249649696403549,
+      "grad_norm": 0.6441498808124666,
+      "learning_rate": 8.665880633094314e-06,
+      "loss": 0.6141,
+      "step": 7065
+    },
+    {
+      "epoch": 0.825548808967772,
+      "grad_norm": 0.5710054884737357,
+      "learning_rate": 8.642147036972887e-06,
+      "loss": 0.6333,
+      "step": 7070
+    },
+    {
+      "epoch": 0.8261326482951892,
+      "grad_norm": 0.6220206555668683,
+      "learning_rate": 8.618483750795087e-06,
+      "loss": 0.6437,
+      "step": 7075
+    },
+    {
+      "epoch": 0.8267164876226063,
+      "grad_norm": 0.6003035270757918,
+      "learning_rate": 8.594890862787518e-06,
+      "loss": 0.6234,
+      "step": 7080
+    },
+    {
+      "epoch": 0.8273003269500233,
+      "grad_norm": 0.6292916954987227,
+      "learning_rate": 8.571368460914316e-06,
+      "loss": 0.6315,
+      "step": 7085
+    },
+    {
+      "epoch": 0.8278841662774404,
+      "grad_norm": 0.6174540855499432,
+      "learning_rate": 8.547916632876806e-06,
+      "loss": 0.6425,
+      "step": 7090
+    },
+    {
+      "epoch": 0.8284680056048576,
+      "grad_norm": 0.6132426662760693,
+      "learning_rate": 8.524535466113185e-06,
+      "loss": 0.6289,
+      "step": 7095
+    },
+    {
+      "epoch": 0.8290518449322747,
+      "grad_norm": 0.6175813113333095,
+      "learning_rate": 8.5012250477982e-06,
+      "loss": 0.6333,
+      "step": 7100
+    },
+    {
+      "epoch": 0.8296356842596917,
+      "grad_norm": 0.6012843126303246,
+      "learning_rate": 8.477985464842816e-06,
+      "loss": 0.6356,
+      "step": 7105
+    },
+    {
+      "epoch": 0.8302195235871088,
+      "grad_norm": 0.6689824872334442,
+      "learning_rate": 8.454816803893893e-06,
+      "loss": 0.6373,
+      "step": 7110
+    },
+    {
+      "epoch": 0.830803362914526,
+      "grad_norm": 0.6391248959131548,
+      "learning_rate": 8.431719151333864e-06,
+      "loss": 0.6247,
+      "step": 7115
+    },
+    {
+      "epoch": 0.8313872022419431,
+      "grad_norm": 0.6181793298039209,
+      "learning_rate": 8.40869259328042e-06,
+      "loss": 0.6443,
+      "step": 7120
+    },
+    {
+      "epoch": 0.8319710415693601,
+      "grad_norm": 0.636510927044935,
+      "learning_rate": 8.385737215586171e-06,
+      "loss": 0.6455,
+      "step": 7125
+    },
+    {
+      "epoch": 0.8325548808967772,
+      "grad_norm": 0.5973412707457408,
+      "learning_rate": 8.362853103838344e-06,
+      "loss": 0.6257,
+      "step": 7130
+    },
+    {
+      "epoch": 0.8331387202241943,
+      "grad_norm": 0.6492295469372185,
+      "learning_rate": 8.340040343358455e-06,
+      "loss": 0.6115,
+      "step": 7135
+    },
+    {
+      "epoch": 0.8337225595516113,
+      "grad_norm": 0.662621688734901,
+      "learning_rate": 8.317299019201996e-06,
+      "loss": 0.6387,
+      "step": 7140
+    },
+    {
+      "epoch": 0.8343063988790285,
+      "grad_norm": 0.6043337811826702,
+      "learning_rate": 8.294629216158107e-06,
+      "loss": 0.6382,
+      "step": 7145
+    },
+    {
+      "epoch": 0.8348902382064456,
+      "grad_norm": 0.6031307093474805,
+      "learning_rate": 8.272031018749272e-06,
+      "loss": 0.6442,
+      "step": 7150
+    },
+    {
+      "epoch": 0.8354740775338627,
+      "grad_norm": 0.6010727899754135,
+      "learning_rate": 8.249504511231005e-06,
+      "loss": 0.6496,
+      "step": 7155
+    },
+    {
+      "epoch": 0.8360579168612797,
+      "grad_norm": 0.6131715223251057,
+      "learning_rate": 8.227049777591516e-06,
+      "loss": 0.6277,
+      "step": 7160
+    },
+    {
+      "epoch": 0.8366417561886969,
+      "grad_norm": 0.6447679727511777,
+      "learning_rate": 8.204666901551428e-06,
+      "loss": 0.6469,
+      "step": 7165
+    },
+    {
+      "epoch": 0.837225595516114,
+      "grad_norm": 0.6122691237665745,
+      "learning_rate": 8.182355966563438e-06,
+      "loss": 0.6307,
+      "step": 7170
+    },
+    {
+      "epoch": 0.8378094348435311,
+      "grad_norm": 0.648928872074787,
+      "learning_rate": 8.160117055812019e-06,
+      "loss": 0.6491,
+      "step": 7175
+    },
+    {
+      "epoch": 0.8383932741709481,
+      "grad_norm": 0.607879109395071,
+      "learning_rate": 8.13795025221311e-06,
+      "loss": 0.6267,
+      "step": 7180
+    },
+    {
+      "epoch": 0.8389771134983652,
+      "grad_norm": 0.6059669756904347,
+      "learning_rate": 8.115855638413806e-06,
+      "loss": 0.6218,
+      "step": 7185
+    },
+    {
+      "epoch": 0.8395609528257824,
+      "grad_norm": 0.6439391595439827,
+      "learning_rate": 8.09383329679204e-06,
+      "loss": 0.627,
+      "step": 7190
+    },
+    {
+      "epoch": 0.8401447921531995,
+      "grad_norm": 0.631501317582565,
+      "learning_rate": 8.071883309456292e-06,
+      "loss": 0.6269,
+      "step": 7195
+    },
+    {
+      "epoch": 0.8407286314806165,
+      "grad_norm": 0.6122310277654351,
+      "learning_rate": 8.050005758245274e-06,
+      "loss": 0.6029,
+      "step": 7200
+    },
+    {
+      "epoch": 0.8413124708080336,
+      "grad_norm": 0.6397694501906913,
+      "learning_rate": 8.028200724727623e-06,
+      "loss": 0.6271,
+      "step": 7205
+    },
+    {
+      "epoch": 0.8418963101354507,
+      "grad_norm": 0.6195119302731571,
+      "learning_rate": 8.006468290201603e-06,
+      "loss": 0.6253,
+      "step": 7210
+    },
+    {
+      "epoch": 0.8424801494628679,
+      "grad_norm": 0.5710130842633037,
+      "learning_rate": 7.984808535694794e-06,
+      "loss": 0.6338,
+      "step": 7215
+    },
+    {
+      "epoch": 0.8430639887902849,
+      "grad_norm": 0.6170905160754272,
+      "learning_rate": 7.963221541963799e-06,
+      "loss": 0.643,
+      "step": 7220
+    },
+    {
+      "epoch": 0.843647828117702,
+      "grad_norm": 0.5929177989523172,
+      "learning_rate": 7.94170738949394e-06,
+      "loss": 0.622,
+      "step": 7225
+    },
+    {
+      "epoch": 0.8442316674451191,
+      "grad_norm": 0.6066005138729927,
+      "learning_rate": 7.920266158498948e-06,
+      "loss": 0.639,
+      "step": 7230
+    },
+    {
+      "epoch": 0.8448155067725363,
+      "grad_norm": 0.6368084808396752,
+      "learning_rate": 7.898897928920684e-06,
+      "loss": 0.6206,
+      "step": 7235
+    },
+    {
+      "epoch": 0.8453993460999533,
+      "grad_norm": 0.6065549696314979,
+      "learning_rate": 7.877602780428816e-06,
+      "loss": 0.6193,
+      "step": 7240
+    },
+    {
+      "epoch": 0.8459831854273704,
+      "grad_norm": 0.6184574053651811,
+      "learning_rate": 7.856380792420549e-06,
+      "loss": 0.631,
+      "step": 7245
+    },
+    {
+      "epoch": 0.8465670247547875,
+      "grad_norm": 0.6329384365136123,
+      "learning_rate": 7.835232044020304e-06,
+      "loss": 0.6373,
+      "step": 7250
+    },
+    {
+      "epoch": 0.8471508640822045,
+      "grad_norm": 0.6105416018238302,
+      "learning_rate": 7.81415661407944e-06,
+      "loss": 0.6242,
+      "step": 7255
+    },
+    {
+      "epoch": 0.8477347034096216,
+      "grad_norm": 0.6261425215195263,
+      "learning_rate": 7.793154581175954e-06,
+      "loss": 0.6315,
+      "step": 7260
+    },
+    {
+      "epoch": 0.8483185427370388,
+      "grad_norm": 0.5756730477657247,
+      "learning_rate": 7.772226023614185e-06,
+      "loss": 0.6348,
+      "step": 7265
+    },
+    {
+      "epoch": 0.8489023820644559,
+      "grad_norm": 0.564793355503493,
+      "learning_rate": 7.751371019424528e-06,
+      "loss": 0.6295,
+      "step": 7270
+    },
+    {
+      "epoch": 0.8494862213918729,
+      "grad_norm": 0.6778890237819897,
+      "learning_rate": 7.730589646363141e-06,
+      "loss": 0.6357,
+      "step": 7275
+    },
+    {
+      "epoch": 0.85007006071929,
+      "grad_norm": 0.5872963788822205,
+      "learning_rate": 7.709881981911648e-06,
+      "loss": 0.6263,
+      "step": 7280
+    },
+    {
+      "epoch": 0.8506539000467072,
+      "grad_norm": 0.6047046430959456,
+      "learning_rate": 7.689248103276873e-06,
+      "loss": 0.6428,
+      "step": 7285
+    },
+    {
+      "epoch": 0.8512377393741243,
+      "grad_norm": 0.6596981335365075,
+      "learning_rate": 7.668688087390509e-06,
+      "loss": 0.6219,
+      "step": 7290
+    },
+    {
+      "epoch": 0.8518215787015413,
+      "grad_norm": 0.586612617500815,
+      "learning_rate": 7.648202010908884e-06,
+      "loss": 0.6494,
+      "step": 7295
+    },
+    {
+      "epoch": 0.8524054180289584,
+      "grad_norm": 0.595787806735689,
+      "learning_rate": 7.627789950212635e-06,
+      "loss": 0.6442,
+      "step": 7300
+    },
+    {
+      "epoch": 0.8529892573563755,
+      "grad_norm": 0.5935995589497031,
+      "learning_rate": 7.607451981406441e-06,
+      "loss": 0.6407,
+      "step": 7305
+    },
+    {
+      "epoch": 0.8535730966837927,
+      "grad_norm": 0.6351732011308916,
+      "learning_rate": 7.587188180318736e-06,
+      "loss": 0.6338,
+      "step": 7310
+    },
+    {
+      "epoch": 0.8541569360112097,
+      "grad_norm": 0.6423062007453759,
+      "learning_rate": 7.5669986225014215e-06,
+      "loss": 0.6145,
+      "step": 7315
+    },
+    {
+      "epoch": 0.8547407753386268,
+      "grad_norm": 0.6104054060862208,
+      "learning_rate": 7.546883383229594e-06,
+      "loss": 0.6302,
+      "step": 7320
+    },
+    {
+      "epoch": 0.8553246146660439,
+      "grad_norm": 0.5732066016302394,
+      "learning_rate": 7.526842537501259e-06,
+      "loss": 0.6194,
+      "step": 7325
+    },
+    {
+      "epoch": 0.855908453993461,
+      "grad_norm": 0.6305709160866012,
+      "learning_rate": 7.50687616003705e-06,
+      "loss": 0.6515,
+      "step": 7330
+    },
+    {
+      "epoch": 0.8564922933208781,
+      "grad_norm": 0.5996923205852378,
+      "learning_rate": 7.486984325279956e-06,
+      "loss": 0.6349,
+      "step": 7335
+    },
+    {
+      "epoch": 0.8570761326482952,
+      "grad_norm": 0.6278616591211431,
+      "learning_rate": 7.467167107395028e-06,
+      "loss": 0.6516,
+      "step": 7340
+    },
+    {
+      "epoch": 0.8576599719757123,
+      "grad_norm": 0.6269815485337418,
+      "learning_rate": 7.44742458026913e-06,
+      "loss": 0.6456,
+      "step": 7345
+    },
+    {
+      "epoch": 0.8582438113031294,
+      "grad_norm": 0.6156463362595231,
+      "learning_rate": 7.427756817510634e-06,
+      "loss": 0.6323,
+      "step": 7350
+    },
+    {
+      "epoch": 0.8588276506305464,
+      "grad_norm": 0.5956719624976204,
+      "learning_rate": 7.408163892449172e-06,
+      "loss": 0.6574,
+      "step": 7355
+    },
+    {
+      "epoch": 0.8594114899579636,
+      "grad_norm": 0.6454289614556328,
+      "learning_rate": 7.388645878135338e-06,
+      "loss": 0.6256,
+      "step": 7360
+    },
+    {
+      "epoch": 0.8599953292853807,
+      "grad_norm": 0.6210045159647282,
+      "learning_rate": 7.369202847340432e-06,
+      "loss": 0.6354,
+      "step": 7365
+    },
+    {
+      "epoch": 0.8605791686127977,
+      "grad_norm": 0.6214434983612692,
+      "learning_rate": 7.349834872556187e-06,
+      "loss": 0.6189,
+      "step": 7370
+    },
+    {
+      "epoch": 0.8611630079402148,
+      "grad_norm": 0.6600710665798625,
+      "learning_rate": 7.330542025994495e-06,
+      "loss": 0.6404,
+      "step": 7375
+    },
+    {
+      "epoch": 0.861746847267632,
+      "grad_norm": 0.5860627523438156,
+      "learning_rate": 7.311324379587136e-06,
+      "loss": 0.6246,
+      "step": 7380
+    },
+    {
+      "epoch": 0.8623306865950491,
+      "grad_norm": 0.6212770319942422,
+      "learning_rate": 7.292182004985511e-06,
+      "loss": 0.6246,
+      "step": 7385
+    },
+    {
+      "epoch": 0.8629145259224661,
+      "grad_norm": 0.6035809127554331,
+      "learning_rate": 7.2731149735603825e-06,
+      "loss": 0.6387,
+      "step": 7390
+    },
+    {
+      "epoch": 0.8634983652498832,
+      "grad_norm": 0.5579844843636018,
+      "learning_rate": 7.254123356401597e-06,
+      "loss": 0.6445,
+      "step": 7395
+    },
+    {
+      "epoch": 0.8640822045773003,
+      "grad_norm": 0.5936151599474442,
+      "learning_rate": 7.23520722431783e-06,
+      "loss": 0.6334,
+      "step": 7400
+    },
+    {
+      "epoch": 0.8646660439047175,
+      "grad_norm": 0.650546590602176,
+      "learning_rate": 7.216366647836306e-06,
+      "loss": 0.6116,
+      "step": 7405
+    },
+    {
+      "epoch": 0.8652498832321345,
+      "grad_norm": 0.6487524123219067,
+      "learning_rate": 7.197601697202565e-06,
+      "loss": 0.6526,
+      "step": 7410
+    },
+    {
+      "epoch": 0.8658337225595516,
+      "grad_norm": 0.6535618503329066,
+      "learning_rate": 7.1789124423801645e-06,
+      "loss": 0.638,
+      "step": 7415
+    },
+    {
+      "epoch": 0.8664175618869687,
+      "grad_norm": 0.6027481431879828,
+      "learning_rate": 7.160298953050448e-06,
+      "loss": 0.6198,
+      "step": 7420
+    },
+    {
+      "epoch": 0.8670014012143858,
+      "grad_norm": 0.6369758086341853,
+      "learning_rate": 7.141761298612267e-06,
+      "loss": 0.6502,
+      "step": 7425
+    },
+    {
+      "epoch": 0.8675852405418029,
+      "grad_norm": 0.6345290792167012,
+      "learning_rate": 7.123299548181732e-06,
+      "loss": 0.6479,
+      "step": 7430
+    },
+    {
+      "epoch": 0.86816907986922,
+      "grad_norm": 0.6498287886811438,
+      "learning_rate": 7.104913770591953e-06,
+      "loss": 0.6323,
+      "step": 7435
+    },
+    {
+      "epoch": 0.8687529191966371,
+      "grad_norm": 0.610453182797401,
+      "learning_rate": 7.086604034392777e-06,
+      "loss": 0.6526,
+      "step": 7440
+    },
+    {
+      "epoch": 0.8693367585240542,
+      "grad_norm": 0.6132209789616724,
+      "learning_rate": 7.068370407850541e-06,
+      "loss": 0.6514,
+      "step": 7445
+    },
+    {
+      "epoch": 0.8699205978514712,
+      "grad_norm": 0.6042376490660476,
+      "learning_rate": 7.050212958947813e-06,
+      "loss": 0.619,
+      "step": 7450
+    },
+    {
+      "epoch": 0.8705044371788884,
+      "grad_norm": 0.6254622221090369,
+      "learning_rate": 7.032131755383134e-06,
+      "loss": 0.6131,
+      "step": 7455
+    },
+    {
+      "epoch": 0.8710882765063055,
+      "grad_norm": 0.6095513928040593,
+      "learning_rate": 7.014126864570782e-06,
+      "loss": 0.61,
+      "step": 7460
+    },
+    {
+      "epoch": 0.8716721158337225,
+      "grad_norm": 0.6000952610701883,
+      "learning_rate": 6.996198353640495e-06,
+      "loss": 0.6328,
+      "step": 7465
+    },
+    {
+      "epoch": 0.8722559551611396,
+      "grad_norm": 0.6267837879440427,
+      "learning_rate": 6.978346289437245e-06,
+      "loss": 0.6386,
+      "step": 7470
+    },
+    {
+      "epoch": 0.8728397944885568,
+      "grad_norm": 0.6099846619504717,
+      "learning_rate": 6.9605707385209755e-06,
+      "loss": 0.6322,
+      "step": 7475
+    },
+    {
+      "epoch": 0.8734236338159739,
+      "grad_norm": 0.6337055827813153,
+      "learning_rate": 6.942871767166354e-06,
+      "loss": 0.6176,
+      "step": 7480
+    },
+    {
+      "epoch": 0.8740074731433909,
+      "grad_norm": 0.6027124220327612,
+      "learning_rate": 6.925249441362533e-06,
+      "loss": 0.6454,
+      "step": 7485
+    },
+    {
+      "epoch": 0.874591312470808,
+      "grad_norm": 0.5825546360394736,
+      "learning_rate": 6.907703826812895e-06,
+      "loss": 0.6256,
+      "step": 7490
+    },
+    {
+      "epoch": 0.8751751517982251,
+      "grad_norm": 0.6364269400726253,
+      "learning_rate": 6.89023498893481e-06,
+      "loss": 0.6203,
+      "step": 7495
+    },
+    {
+      "epoch": 0.8757589911256423,
+      "grad_norm": 0.5866690674431253,
+      "learning_rate": 6.872842992859395e-06,
+      "loss": 0.6426,
+      "step": 7500
+    },
+    {
+      "epoch": 0.8763428304530593,
+      "grad_norm": 0.606473729787098,
+      "learning_rate": 6.855527903431267e-06,
+      "loss": 0.6555,
+      "step": 7505
+    },
+    {
+      "epoch": 0.8769266697804764,
+      "grad_norm": 0.6175430857724672,
+      "learning_rate": 6.838289785208303e-06,
+      "loss": 0.6228,
+      "step": 7510
+    },
+    {
+      "epoch": 0.8775105091078935,
+      "grad_norm": 0.6441178869926913,
+      "learning_rate": 6.821128702461401e-06,
+      "loss": 0.6232,
+      "step": 7515
+    },
+    {
+      "epoch": 0.8780943484353106,
+      "grad_norm": 0.6455378058084096,
+      "learning_rate": 6.804044719174235e-06,
+      "loss": 0.6397,
+      "step": 7520
+    },
+    {
+      "epoch": 0.8786781877627277,
+      "grad_norm": 0.6474729332887994,
+      "learning_rate": 6.787037899043027e-06,
+      "loss": 0.6493,
+      "step": 7525
+    },
+    {
+      "epoch": 0.8792620270901448,
+      "grad_norm": 0.6399294710875848,
+      "learning_rate": 6.770108305476293e-06,
+      "loss": 0.6339,
+      "step": 7530
+    },
+    {
+      "epoch": 0.8798458664175619,
+      "grad_norm": 0.5902882884714229,
+      "learning_rate": 6.753256001594622e-06,
+      "loss": 0.6269,
+      "step": 7535
+    },
+    {
+      "epoch": 0.880429705744979,
+      "grad_norm": 0.6258926337290766,
+      "learning_rate": 6.736481050230438e-06,
+      "loss": 0.65,
+      "step": 7540
+    },
+    {
+      "epoch": 0.881013545072396,
+      "grad_norm": 0.6781590777121034,
+      "learning_rate": 6.719783513927755e-06,
+      "loss": 0.6483,
+      "step": 7545
+    },
+    {
+      "epoch": 0.8815973843998132,
+      "grad_norm": 0.5475048535917718,
+      "learning_rate": 6.703163454941953e-06,
+      "loss": 0.6237,
+      "step": 7550
+    },
+    {
+      "epoch": 0.8821812237272303,
+      "grad_norm": 0.6058898403074587,
+      "learning_rate": 6.686620935239552e-06,
+      "loss": 0.6305,
+      "step": 7555
+    },
+    {
+      "epoch": 0.8827650630546474,
+      "grad_norm": 0.6062332874729838,
+      "learning_rate": 6.670156016497958e-06,
+      "loss": 0.6165,
+      "step": 7560
+    },
+    {
+      "epoch": 0.8833489023820644,
+      "grad_norm": 0.5993012988315325,
+      "learning_rate": 6.653768760105268e-06,
+      "loss": 0.6329,
+      "step": 7565
+    },
+    {
+      "epoch": 0.8839327417094816,
+      "grad_norm": 0.6004488967091748,
+      "learning_rate": 6.637459227160004e-06,
+      "loss": 0.6275,
+      "step": 7570
+    },
+    {
+      "epoch": 0.8845165810368987,
+      "grad_norm": 0.5954153667320861,
+      "learning_rate": 6.621227478470911e-06,
+      "loss": 0.6391,
+      "step": 7575
+    },
+    {
+      "epoch": 0.8851004203643157,
+      "grad_norm": 0.6211794389009575,
+      "learning_rate": 6.605073574556721e-06,
+      "loss": 0.6481,
+      "step": 7580
+    },
+    {
+      "epoch": 0.8856842596917328,
+      "grad_norm": 0.6117471748709564,
+      "learning_rate": 6.588997575645929e-06,
+      "loss": 0.6144,
+      "step": 7585
+    },
+    {
+      "epoch": 0.8862680990191499,
+      "grad_norm": 0.633929275434095,
+      "learning_rate": 6.572999541676563e-06,
+      "loss": 0.6475,
+      "step": 7590
+    },
+    {
+      "epoch": 0.8868519383465671,
+      "grad_norm": 0.6085055988668767,
+      "learning_rate": 6.557079532295968e-06,
+      "loss": 0.6327,
+      "step": 7595
+    },
+    {
+      "epoch": 0.8874357776739841,
+      "grad_norm": 0.5942186040054447,
+      "learning_rate": 6.541237606860582e-06,
+      "loss": 0.6254,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8880196170014012,
+      "grad_norm": 0.5846661536893463,
+      "learning_rate": 6.525473824435714e-06,
+      "loss": 0.6437,
+      "step": 7605
+    },
+    {
+      "epoch": 0.8886034563288183,
+      "grad_norm": 0.5951544079113886,
+      "learning_rate": 6.5097882437953205e-06,
+      "loss": 0.6225,
+      "step": 7610
+    },
+    {
+      "epoch": 0.8891872956562354,
+      "grad_norm": 0.6258276348637012,
+      "learning_rate": 6.49418092342179e-06,
+      "loss": 0.6452,
+      "step": 7615
+    },
+    {
+      "epoch": 0.8897711349836525,
+      "grad_norm": 0.5737411429686861,
+      "learning_rate": 6.478651921505727e-06,
+      "loss": 0.634,
+      "step": 7620
+    },
+    {
+      "epoch": 0.8903549743110696,
+      "grad_norm": 0.6039253224606503,
+      "learning_rate": 6.463201295945727e-06,
+      "loss": 0.6411,
+      "step": 7625
+    },
+    {
+      "epoch": 0.8909388136384867,
+      "grad_norm": 0.5847198844312848,
+      "learning_rate": 6.447829104348171e-06,
+      "loss": 0.6348,
+      "step": 7630
+    },
+    {
+      "epoch": 0.8915226529659038,
+      "grad_norm": 0.6030492636470951,
+      "learning_rate": 6.432535404026997e-06,
+      "loss": 0.6321,
+      "step": 7635
+    },
+    {
+      "epoch": 0.8921064922933208,
+      "grad_norm": 0.5846736812852988,
+      "learning_rate": 6.417320252003505e-06,
+      "loss": 0.6243,
+      "step": 7640
+    },
+    {
+      "epoch": 0.892690331620738,
+      "grad_norm": 0.5869240165168146,
+      "learning_rate": 6.402183705006127e-06,
+      "loss": 0.6257,
+      "step": 7645
+    },
+    {
+      "epoch": 0.8932741709481551,
+      "grad_norm": 0.6051339628375705,
+      "learning_rate": 6.387125819470231e-06,
+      "loss": 0.6286,
+      "step": 7650
+    },
+    {
+      "epoch": 0.8938580102755722,
+      "grad_norm": 0.6029543179587162,
+      "learning_rate": 6.372146651537892e-06,
+      "loss": 0.6343,
+      "step": 7655
+    },
+    {
+      "epoch": 0.8944418496029892,
+      "grad_norm": 0.5978429103009526,
+      "learning_rate": 6.3572462570576985e-06,
+      "loss": 0.6379,
+      "step": 7660
+    },
+    {
+      "epoch": 0.8950256889304063,
+      "grad_norm": 0.601501640872941,
+      "learning_rate": 6.3424246915845395e-06,
+      "loss": 0.6513,
+      "step": 7665
+    },
+    {
+      "epoch": 0.8956095282578235,
+      "grad_norm": 0.6176265112274103,
+      "learning_rate": 6.327682010379392e-06,
+      "loss": 0.6327,
+      "step": 7670
+    },
+    {
+      "epoch": 0.8961933675852406,
+      "grad_norm": 0.6347108525818637,
+      "learning_rate": 6.313018268409122e-06,
+      "loss": 0.6128,
+      "step": 7675
+    },
+    {
+      "epoch": 0.8967772069126576,
+      "grad_norm": 0.6073779195946353,
+      "learning_rate": 6.2984335203462825e-06,
+      "loss": 0.6373,
+      "step": 7680
+    },
+    {
+      "epoch": 0.8973610462400747,
+      "grad_norm": 0.575256225610877,
+      "learning_rate": 6.283927820568894e-06,
+      "loss": 0.6161,
+      "step": 7685
+    },
+    {
+      "epoch": 0.8979448855674919,
+      "grad_norm": 0.6155556881105437,
+      "learning_rate": 6.269501223160259e-06,
+      "loss": 0.6312,
+      "step": 7690
+    },
+    {
+      "epoch": 0.8985287248949089,
+      "grad_norm": 0.577643059355464,
+      "learning_rate": 6.255153781908754e-06,
+      "loss": 0.6359,
+      "step": 7695
+    },
+    {
+      "epoch": 0.899112564222326,
+      "grad_norm": 0.5977737921135581,
+      "learning_rate": 6.240885550307624e-06,
+      "loss": 0.6129,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8996964035497431,
+      "grad_norm": 0.6140981499537501,
+      "learning_rate": 6.2266965815547865e-06,
+      "loss": 0.6253,
+      "step": 7705
+    },
+    {
+      "epoch": 0.9002802428771602,
+      "grad_norm": 0.5975649843560508,
+      "learning_rate": 6.212586928552641e-06,
+      "loss": 0.6163,
+      "step": 7710
+    },
+    {
+      "epoch": 0.9008640822045773,
+      "grad_norm": 0.5950454352388025,
+      "learning_rate": 6.19855664390786e-06,
+      "loss": 0.6274,
+      "step": 7715
+    },
+    {
+      "epoch": 0.9014479215319944,
+      "grad_norm": 0.6094660507160992,
+      "learning_rate": 6.184605779931197e-06,
+      "loss": 0.6435,
+      "step": 7720
+    },
+    {
+      "epoch": 0.9020317608594115,
+      "grad_norm": 0.6119309598856335,
+      "learning_rate": 6.170734388637294e-06,
+      "loss": 0.6352,
+      "step": 7725
+    },
+    {
+      "epoch": 0.9026156001868286,
+      "grad_norm": 0.595911716393697,
+      "learning_rate": 6.156942521744484e-06,
+      "loss": 0.6194,
+      "step": 7730
+    },
+    {
+      "epoch": 0.9031994395142456,
+      "grad_norm": 0.5993346947122322,
+      "learning_rate": 6.143230230674602e-06,
+      "loss": 0.629,
+      "step": 7735
+    },
+    {
+      "epoch": 0.9037832788416628,
+      "grad_norm": 0.6044325398571229,
+      "learning_rate": 6.12959756655279e-06,
+      "loss": 0.6271,
+      "step": 7740
+    },
+    {
+      "epoch": 0.9043671181690799,
+      "grad_norm": 0.6343098779398217,
+      "learning_rate": 6.11604458020731e-06,
+      "loss": 0.6373,
+      "step": 7745
+    },
+    {
+      "epoch": 0.904950957496497,
+      "grad_norm": 0.577313456911306,
+      "learning_rate": 6.102571322169347e-06,
+      "loss": 0.6361,
+      "step": 7750
+    },
+    {
+      "epoch": 0.905534796823914,
+      "grad_norm": 0.6239547800150814,
+      "learning_rate": 6.089177842672826e-06,
+      "loss": 0.6477,
+      "step": 7755
+    },
+    {
+      "epoch": 0.9061186361513311,
+      "grad_norm": 0.6188422351793207,
+      "learning_rate": 6.075864191654231e-06,
+      "loss": 0.628,
+      "step": 7760
+    },
+    {
+      "epoch": 0.9067024754787483,
+      "grad_norm": 0.5911029453007852,
+      "learning_rate": 6.062630418752404e-06,
+      "loss": 0.6219,
+      "step": 7765
+    },
+    {
+      "epoch": 0.9072863148061654,
+      "grad_norm": 0.6284611964804068,
+      "learning_rate": 6.049476573308375e-06,
+      "loss": 0.6482,
+      "step": 7770
+    },
+    {
+      "epoch": 0.9078701541335824,
+      "grad_norm": 0.5686403476250719,
+      "learning_rate": 6.036402704365168e-06,
+      "loss": 0.6252,
+      "step": 7775
+    },
+    {
+      "epoch": 0.9084539934609995,
+      "grad_norm": 0.6248813037354658,
+      "learning_rate": 6.023408860667617e-06,
+      "loss": 0.6457,
+      "step": 7780
+    },
+    {
+      "epoch": 0.9090378327884167,
+      "grad_norm": 0.6116229301425355,
+      "learning_rate": 6.010495090662197e-06,
+      "loss": 0.6218,
+      "step": 7785
+    },
+    {
+      "epoch": 0.9096216721158337,
+      "grad_norm": 0.5982997364765981,
+      "learning_rate": 5.9976614424968245e-06,
+      "loss": 0.6238,
+      "step": 7790
+    },
+    {
+      "epoch": 0.9102055114432508,
+      "grad_norm": 0.6092146507527899,
+      "learning_rate": 5.9849079640207e-06,
+      "loss": 0.6292,
+      "step": 7795
+    },
+    {
+      "epoch": 0.9107893507706679,
+      "grad_norm": 0.6625607886398457,
+      "learning_rate": 5.972234702784106e-06,
+      "loss": 0.6401,
+      "step": 7800
+    },
+    {
+      "epoch": 0.911373190098085,
+      "grad_norm": 0.6123956879925448,
+      "learning_rate": 5.9596417060382545e-06,
+      "loss": 0.6317,
+      "step": 7805
+    },
+    {
+      "epoch": 0.911957029425502,
+      "grad_norm": 0.6042691351169877,
+      "learning_rate": 5.9471290207350925e-06,
+      "loss": 0.6133,
+      "step": 7810
+    },
+    {
+      "epoch": 0.9125408687529192,
+      "grad_norm": 0.62392183374982,
+      "learning_rate": 5.934696693527123e-06,
+      "loss": 0.6335,
+      "step": 7815
+    },
+    {
+      "epoch": 0.9131247080803363,
+      "grad_norm": 0.5796174304624362,
+      "learning_rate": 5.9223447707672564e-06,
+      "loss": 0.6163,
+      "step": 7820
+    },
+    {
+      "epoch": 0.9137085474077534,
+      "grad_norm": 0.6278698472258146,
+      "learning_rate": 5.910073298508609e-06,
+      "loss": 0.6433,
+      "step": 7825
+    },
+    {
+      "epoch": 0.9142923867351704,
+      "grad_norm": 0.5895985868489902,
+      "learning_rate": 5.8978823225043555e-06,
+      "loss": 0.6251,
+      "step": 7830
+    },
+    {
+      "epoch": 0.9148762260625876,
+      "grad_norm": 0.6008029537552508,
+      "learning_rate": 5.8857718882075325e-06,
+      "loss": 0.6469,
+      "step": 7835
+    },
+    {
+      "epoch": 0.9154600653900047,
+      "grad_norm": 0.5817527833570267,
+      "learning_rate": 5.8737420407708985e-06,
+      "loss": 0.6292,
+      "step": 7840
+    },
+    {
+      "epoch": 0.9160439047174218,
+      "grad_norm": 0.599543094939994,
+      "learning_rate": 5.861792825046739e-06,
+      "loss": 0.6255,
+      "step": 7845
+    },
+    {
+      "epoch": 0.9166277440448388,
+      "grad_norm": 0.5776097905597614,
+      "learning_rate": 5.849924285586719e-06,
+      "loss": 0.6391,
+      "step": 7850
+    },
+    {
+      "epoch": 0.917211583372256,
+      "grad_norm": 0.6107022416917586,
+      "learning_rate": 5.838136466641704e-06,
+      "loss": 0.6291,
+      "step": 7855
+    },
+    {
+      "epoch": 0.9177954226996731,
+      "grad_norm": 0.6021931223535016,
+      "learning_rate": 5.8264294121616e-06,
+      "loss": 0.6275,
+      "step": 7860
+    },
+    {
+      "epoch": 0.9183792620270902,
+      "grad_norm": 0.6246355143005593,
+      "learning_rate": 5.814803165795194e-06,
+      "loss": 0.6338,
+      "step": 7865
+    },
+    {
+      "epoch": 0.9189631013545072,
+      "grad_norm": 0.6034832752476175,
+      "learning_rate": 5.803257770889978e-06,
+      "loss": 0.6411,
+      "step": 7870
+    },
+    {
+      "epoch": 0.9195469406819243,
+      "grad_norm": 0.6092008532994692,
+      "learning_rate": 5.791793270492006e-06,
+      "loss": 0.6328,
+      "step": 7875
+    },
+    {
+      "epoch": 0.9201307800093415,
+      "grad_norm": 0.5948761862413096,
+      "learning_rate": 5.780409707345714e-06,
+      "loss": 0.6473,
+      "step": 7880
+    },
+    {
+      "epoch": 0.9207146193367586,
+      "grad_norm": 0.5708828641760395,
+      "learning_rate": 5.769107123893781e-06,
+      "loss": 0.618,
+      "step": 7885
+    },
+    {
+      "epoch": 0.9212984586641756,
+      "grad_norm": 0.5787863997950811,
+      "learning_rate": 5.757885562276948e-06,
+      "loss": 0.6221,
+      "step": 7890
+    },
+    {
+      "epoch": 0.9218822979915927,
+      "grad_norm": 0.6141574953858627,
+      "learning_rate": 5.7467450643338804e-06,
+      "loss": 0.6296,
+      "step": 7895
+    },
+    {
+      "epoch": 0.9224661373190098,
+      "grad_norm": 0.5958732143906431,
+      "learning_rate": 5.7356856716010014e-06,
+      "loss": 0.6257,
+      "step": 7900
+    },
+    {
+      "epoch": 0.9230499766464269,
+      "grad_norm": 0.6431949588838122,
+      "learning_rate": 5.724707425312344e-06,
+      "loss": 0.6255,
+      "step": 7905
+    },
+    {
+      "epoch": 0.923633815973844,
+      "grad_norm": 0.6051972855303055,
+      "learning_rate": 5.7138103663993895e-06,
+      "loss": 0.6263,
+      "step": 7910
+    },
+    {
+      "epoch": 0.9242176553012611,
+      "grad_norm": 0.6063472814698573,
+      "learning_rate": 5.70299453549092e-06,
+      "loss": 0.6331,
+      "step": 7915
+    },
+    {
+      "epoch": 0.9248014946286782,
+      "grad_norm": 0.6305201290600937,
+      "learning_rate": 5.692259972912865e-06,
+      "loss": 0.626,
+      "step": 7920
+    },
+    {
+      "epoch": 0.9253853339560952,
+      "grad_norm": 0.6033166123290059,
+      "learning_rate": 5.681606718688152e-06,
+      "loss": 0.6286,
+      "step": 7925
+    },
+    {
+      "epoch": 0.9259691732835124,
+      "grad_norm": 0.5744059957483572,
+      "learning_rate": 5.671034812536561e-06,
+      "loss": 0.6008,
+      "step": 7930
+    },
+    {
+      "epoch": 0.9265530126109295,
+      "grad_norm": 0.6197956332271047,
+      "learning_rate": 5.660544293874561e-06,
+      "loss": 0.6141,
+      "step": 7935
+    },
+    {
+      "epoch": 0.9271368519383466,
+      "grad_norm": 0.6021158964286558,
+      "learning_rate": 5.65013520181519e-06,
+      "loss": 0.6221,
+      "step": 7940
+    },
+    {
+      "epoch": 0.9277206912657636,
+      "grad_norm": 0.5656423666481871,
+      "learning_rate": 5.639807575167886e-06,
+      "loss": 0.6226,
+      "step": 7945
+    },
+    {
+      "epoch": 0.9283045305931807,
+      "grad_norm": 0.600675095444175,
+      "learning_rate": 5.6295614524383436e-06,
+      "loss": 0.6249,
+      "step": 7950
+    },
+    {
+      "epoch": 0.9288883699205979,
+      "grad_norm": 0.6133243190992899,
+      "learning_rate": 5.619396871828387e-06,
+      "loss": 0.6291,
+      "step": 7955
+    },
+    {
+      "epoch": 0.929472209248015,
+      "grad_norm": 0.6511471891410207,
+      "learning_rate": 5.6093138712358155e-06,
+      "loss": 0.6333,
+      "step": 7960
+    },
+    {
+      "epoch": 0.930056048575432,
+      "grad_norm": 0.6223717747889685,
+      "learning_rate": 5.5993124882542584e-06,
+      "loss": 0.6338,
+      "step": 7965
+    },
+    {
+      "epoch": 0.9306398879028491,
+      "grad_norm": 0.6128977080868971,
+      "learning_rate": 5.589392760173047e-06,
+      "loss": 0.6348,
+      "step": 7970
+    },
+    {
+      "epoch": 0.9312237272302663,
+      "grad_norm": 0.5952529850359788,
+      "learning_rate": 5.579554723977065e-06,
+      "loss": 0.6218,
+      "step": 7975
+    },
+    {
+      "epoch": 0.9318075665576834,
+      "grad_norm": 0.5926314655193932,
+      "learning_rate": 5.569798416346615e-06,
+      "loss": 0.6278,
+      "step": 7980
+    },
+    {
+      "epoch": 0.9323914058851004,
+      "grad_norm": 0.6364867297259194,
+      "learning_rate": 5.560123873657284e-06,
+      "loss": 0.6227,
+      "step": 7985
+    },
+    {
+      "epoch": 0.9329752452125175,
+      "grad_norm": 0.588741898129217,
+      "learning_rate": 5.550531131979804e-06,
+      "loss": 0.6322,
+      "step": 7990
+    },
+    {
+      "epoch": 0.9335590845399346,
+      "grad_norm": 0.5868391154734851,
+      "learning_rate": 5.5410202270799165e-06,
+      "loss": 0.6335,
+      "step": 7995
+    },
+    {
+      "epoch": 0.9341429238673518,
+      "grad_norm": 0.6177367560309492,
+      "learning_rate": 5.531591194418244e-06,
+      "loss": 0.6243,
+      "step": 8000
+    },
+    {
+      "epoch": 0.9347267631947688,
+      "grad_norm": 0.5846412401374685,
+      "learning_rate": 5.5222440691501534e-06,
+      "loss": 0.6239,
+      "step": 8005
+    },
+    {
+      "epoch": 0.9353106025221859,
+      "grad_norm": 0.5952668242128202,
+      "learning_rate": 5.512978886125628e-06,
+      "loss": 0.6086,
+      "step": 8010
+    },
+    {
+      "epoch": 0.935894441849603,
+      "grad_norm": 0.5975481591744062,
+      "learning_rate": 5.5037956798891345e-06,
+      "loss": 0.627,
+      "step": 8015
+    },
+    {
+      "epoch": 0.93647828117702,
+      "grad_norm": 0.6493163522370307,
+      "learning_rate": 5.494694484679501e-06,
+      "loss": 0.6105,
+      "step": 8020
+    },
+    {
+      "epoch": 0.9370621205044372,
+      "grad_norm": 0.6067098748167189,
+      "learning_rate": 5.485675334429776e-06,
+      "loss": 0.6048,
+      "step": 8025
+    },
+    {
+      "epoch": 0.9376459598318543,
+      "grad_norm": 0.6112238279178478,
+      "learning_rate": 5.476738262767116e-06,
+      "loss": 0.6409,
+      "step": 8030
+    },
+    {
+      "epoch": 0.9382297991592714,
+      "grad_norm": 0.6018698466530003,
+      "learning_rate": 5.467883303012653e-06,
+      "loss": 0.6414,
+      "step": 8035
+    },
+    {
+      "epoch": 0.9388136384866884,
+      "grad_norm": 0.5596954894197848,
+      "learning_rate": 5.459110488181373e-06,
+      "loss": 0.6177,
+      "step": 8040
+    },
+    {
+      "epoch": 0.9393974778141055,
+      "grad_norm": 0.6181282912666148,
+      "learning_rate": 5.450419850981987e-06,
+      "loss": 0.6499,
+      "step": 8045
+    },
+    {
+      "epoch": 0.9399813171415227,
+      "grad_norm": 0.600484249263648,
+      "learning_rate": 5.441811423816817e-06,
+      "loss": 0.6407,
+      "step": 8050
+    },
+    {
+      "epoch": 0.9405651564689398,
+      "grad_norm": 0.6038979619569053,
+      "learning_rate": 5.433285238781674e-06,
+      "loss": 0.6426,
+      "step": 8055
+    },
+    {
+      "epoch": 0.9411489957963568,
+      "grad_norm": 0.5930425789751672,
+      "learning_rate": 5.424841327665728e-06,
+      "loss": 0.6363,
+      "step": 8060
+    },
+    {
+      "epoch": 0.9417328351237739,
+      "grad_norm": 0.6188823703473455,
+      "learning_rate": 5.416479721951409e-06,
+      "loss": 0.6239,
+      "step": 8065
+    },
+    {
+      "epoch": 0.942316674451191,
+      "grad_norm": 0.5894093077115455,
+      "learning_rate": 5.408200452814265e-06,
+      "loss": 0.6174,
+      "step": 8070
+    },
+    {
+      "epoch": 0.9429005137786082,
+      "grad_norm": 0.5803109373806545,
+      "learning_rate": 5.400003551122871e-06,
+      "loss": 0.6368,
+      "step": 8075
+    },
+    {
+      "epoch": 0.9434843531060252,
+      "grad_norm": 0.6397959955822484,
+      "learning_rate": 5.391889047438692e-06,
+      "loss": 0.6244,
+      "step": 8080
+    },
+    {
+      "epoch": 0.9440681924334423,
+      "grad_norm": 0.6258497625424511,
+      "learning_rate": 5.383856972015984e-06,
+      "loss": 0.6226,
+      "step": 8085
+    },
+    {
+      "epoch": 0.9446520317608594,
+      "grad_norm": 0.6069468999871956,
+      "learning_rate": 5.3759073548016776e-06,
+      "loss": 0.6404,
+      "step": 8090
+    },
+    {
+      "epoch": 0.9452358710882766,
+      "grad_norm": 0.5861493386306807,
+      "learning_rate": 5.368040225435264e-06,
+      "loss": 0.6187,
+      "step": 8095
+    },
+    {
+      "epoch": 0.9458197104156936,
+      "grad_norm": 0.5634537016801693,
+      "learning_rate": 5.360255613248679e-06,
+      "loss": 0.6155,
+      "step": 8100
+    },
+    {
+      "epoch": 0.9464035497431107,
+      "grad_norm": 0.5807799609418582,
+      "learning_rate": 5.352553547266205e-06,
+      "loss": 0.6346,
+      "step": 8105
+    },
+    {
+      "epoch": 0.9469873890705278,
+      "grad_norm": 0.5985805165038645,
+      "learning_rate": 5.34493405620436e-06,
+      "loss": 0.6456,
+      "step": 8110
+    },
+    {
+      "epoch": 0.9475712283979448,
+      "grad_norm": 0.6470813090081771,
+      "learning_rate": 5.337397168471786e-06,
+      "loss": 0.6403,
+      "step": 8115
+    },
+    {
+      "epoch": 0.948155067725362,
+      "grad_norm": 0.6212004861636161,
+      "learning_rate": 5.329942912169144e-06,
+      "loss": 0.6335,
+      "step": 8120
+    },
+    {
+      "epoch": 0.9487389070527791,
+      "grad_norm": 0.5867741211903176,
+      "learning_rate": 5.322571315089009e-06,
+      "loss": 0.6264,
+      "step": 8125
+    },
+    {
+      "epoch": 0.9493227463801962,
+      "grad_norm": 0.5827684325148621,
+      "learning_rate": 5.315282404715776e-06,
+      "loss": 0.6268,
+      "step": 8130
+    },
+    {
+      "epoch": 0.9499065857076132,
+      "grad_norm": 0.5762150106079065,
+      "learning_rate": 5.308076208225538e-06,
+      "loss": 0.6261,
+      "step": 8135
+    },
+    {
+      "epoch": 0.9504904250350303,
+      "grad_norm": 0.6484971222881835,
+      "learning_rate": 5.300952752486006e-06,
+      "loss": 0.6416,
+      "step": 8140
+    },
+    {
+      "epoch": 0.9510742643624475,
+      "grad_norm": 0.5891698336069202,
+      "learning_rate": 5.293912064056394e-06,
+      "loss": 0.6229,
+      "step": 8145
+    },
+    {
+      "epoch": 0.9516581036898646,
+      "grad_norm": 0.5916315255808878,
+      "learning_rate": 5.286954169187325e-06,
+      "loss": 0.6166,
+      "step": 8150
+    },
+    {
+      "epoch": 0.9522419430172816,
+      "grad_norm": 0.6674962228781914,
+      "learning_rate": 5.280079093820737e-06,
+      "loss": 0.6451,
+      "step": 8155
+    },
+    {
+      "epoch": 0.9528257823446987,
+      "grad_norm": 0.6069069688549191,
+      "learning_rate": 5.273286863589776e-06,
+      "loss": 0.6415,
+      "step": 8160
+    },
+    {
+      "epoch": 0.9534096216721158,
+      "grad_norm": 0.5958144997860393,
+      "learning_rate": 5.266577503818708e-06,
+      "loss": 0.6527,
+      "step": 8165
+    },
+    {
+      "epoch": 0.953993460999533,
+      "grad_norm": 0.6155935484948483,
+      "learning_rate": 5.259951039522832e-06,
+      "loss": 0.6461,
+      "step": 8170
+    },
+    {
+      "epoch": 0.95457730032695,
+      "grad_norm": 0.6252376719795079,
+      "learning_rate": 5.253407495408368e-06,
+      "loss": 0.6239,
+      "step": 8175
+    },
+    {
+      "epoch": 0.9551611396543671,
+      "grad_norm": 0.5776249542863285,
+      "learning_rate": 5.24694689587238e-06,
+      "loss": 0.6237,
+      "step": 8180
+    },
+    {
+      "epoch": 0.9557449789817842,
+      "grad_norm": 0.5863217640086233,
+      "learning_rate": 5.240569265002673e-06,
+      "loss": 0.6237,
+      "step": 8185
+    },
+    {
+      "epoch": 0.9563288183092014,
+      "grad_norm": 0.6073735304248427,
+      "learning_rate": 5.234274626577723e-06,
+      "loss": 0.6423,
+      "step": 8190
+    },
+    {
+      "epoch": 0.9569126576366184,
+      "grad_norm": 0.5999967869487696,
+      "learning_rate": 5.228063004066567e-06,
+      "loss": 0.6189,
+      "step": 8195
+    },
+    {
+      "epoch": 0.9574964969640355,
+      "grad_norm": 0.6441911130662245,
+      "learning_rate": 5.22193442062872e-06,
+      "loss": 0.6232,
+      "step": 8200
+    },
+    {
+      "epoch": 0.9580803362914526,
+      "grad_norm": 0.6439126904936041,
+      "learning_rate": 5.2158888991141055e-06,
+      "loss": 0.6339,
+      "step": 8205
+    },
+    {
+      "epoch": 0.9586641756188697,
+      "grad_norm": 0.6032589048854394,
+      "learning_rate": 5.2099264620629425e-06,
+      "loss": 0.6349,
+      "step": 8210
+    },
+    {
+      "epoch": 0.9592480149462868,
+      "grad_norm": 0.5960007625552923,
+      "learning_rate": 5.204047131705689e-06,
+      "loss": 0.6204,
+      "step": 8215
+    },
+    {
+      "epoch": 0.9598318542737039,
+      "grad_norm": 0.6233901017209507,
+      "learning_rate": 5.198250929962939e-06,
+      "loss": 0.6415,
+      "step": 8220
+    },
+    {
+      "epoch": 0.960415693601121,
+      "grad_norm": 0.6066027994598999,
+      "learning_rate": 5.192537878445356e-06,
+      "loss": 0.6384,
+      "step": 8225
+    },
+    {
+      "epoch": 0.960999532928538,
+      "grad_norm": 0.5806070059656948,
+      "learning_rate": 5.186907998453573e-06,
+      "loss": 0.6247,
+      "step": 8230
+    },
+    {
+      "epoch": 0.9615833722559551,
+      "grad_norm": 0.6233770326846283,
+      "learning_rate": 5.181361310978133e-06,
+      "loss": 0.6167,
+      "step": 8235
+    },
+    {
+      "epoch": 0.9621672115833723,
+      "grad_norm": 0.5843093926235634,
+      "learning_rate": 5.175897836699403e-06,
+      "loss": 0.6435,
+      "step": 8240
+    },
+    {
+      "epoch": 0.9627510509107894,
+      "grad_norm": 0.6235011595985152,
+      "learning_rate": 5.170517595987493e-06,
+      "loss": 0.634,
+      "step": 8245
+    },
+    {
+      "epoch": 0.9633348902382064,
+      "grad_norm": 0.6556016479037008,
+      "learning_rate": 5.165220608902186e-06,
+      "loss": 0.6291,
+      "step": 8250
+    },
+    {
+      "epoch": 0.9639187295656235,
+      "grad_norm": 0.6375416786937221,
+      "learning_rate": 5.160006895192858e-06,
+      "loss": 0.6282,
+      "step": 8255
+    },
+    {
+      "epoch": 0.9645025688930406,
+      "grad_norm": 0.6137703050348912,
+      "learning_rate": 5.154876474298412e-06,
+      "loss": 0.6088,
+      "step": 8260
+    },
+    {
+      "epoch": 0.9650864082204578,
+      "grad_norm": 0.6390723634134673,
+      "learning_rate": 5.149829365347197e-06,
+      "loss": 0.6298,
+      "step": 8265
+    },
+    {
+      "epoch": 0.9656702475478748,
+      "grad_norm": 0.6067102292176847,
+      "learning_rate": 5.14486558715694e-06,
+      "loss": 0.6339,
+      "step": 8270
+    },
+    {
+      "epoch": 0.9662540868752919,
+      "grad_norm": 0.627573621858438,
+      "learning_rate": 5.139985158234677e-06,
+      "loss": 0.666,
+      "step": 8275
+    },
+    {
+      "epoch": 0.966837926202709,
+      "grad_norm": 0.6233363637808976,
+      "learning_rate": 5.135188096776682e-06,
+      "loss": 0.627,
+      "step": 8280
+    },
+    {
+      "epoch": 0.9674217655301262,
+      "grad_norm": 0.6007626857020777,
+      "learning_rate": 5.130474420668403e-06,
+      "loss": 0.6404,
+      "step": 8285
+    },
+    {
+      "epoch": 0.9680056048575432,
+      "grad_norm": 0.6298620628413639,
+      "learning_rate": 5.125844147484391e-06,
+      "loss": 0.6202,
+      "step": 8290
+    },
+    {
+      "epoch": 0.9685894441849603,
+      "grad_norm": 0.675465228558122,
+      "learning_rate": 5.121297294488237e-06,
+      "loss": 0.6518,
+      "step": 8295
+    },
+    {
+      "epoch": 0.9691732835123774,
+      "grad_norm": 0.5865659966177293,
+      "learning_rate": 5.1168338786325025e-06,
+      "loss": 0.6247,
+      "step": 8300
+    },
+    {
+      "epoch": 0.9697571228397945,
+      "grad_norm": 0.6678929540821399,
+      "learning_rate": 5.112453916558671e-06,
+      "loss": 0.6494,
+      "step": 8305
+    },
+    {
+      "epoch": 0.9703409621672116,
+      "grad_norm": 0.6009288621686822,
+      "learning_rate": 5.108157424597062e-06,
+      "loss": 0.6234,
+      "step": 8310
+    },
+    {
+      "epoch": 0.9709248014946287,
+      "grad_norm": 0.6193467021477096,
+      "learning_rate": 5.103944418766791e-06,
+      "loss": 0.635,
+      "step": 8315
+    },
+    {
+      "epoch": 0.9715086408220458,
+      "grad_norm": 0.622608684023799,
+      "learning_rate": 5.099814914775706e-06,
+      "loss": 0.6237,
+      "step": 8320
+    },
+    {
+      "epoch": 0.9720924801494629,
+      "grad_norm": 0.6052720531668783,
+      "learning_rate": 5.095768928020314e-06,
+      "loss": 0.6396,
+      "step": 8325
+    },
+    {
+      "epoch": 0.9726763194768799,
+      "grad_norm": 0.6736182968411951,
+      "learning_rate": 5.09180647358575e-06,
+      "loss": 0.6247,
+      "step": 8330
+    },
+    {
+      "epoch": 0.9732601588042971,
+      "grad_norm": 0.6389566384205301,
+      "learning_rate": 5.087927566245688e-06,
+      "loss": 0.6234,
+      "step": 8335
+    },
+    {
+      "epoch": 0.9738439981317142,
+      "grad_norm": 0.5607228041623084,
+      "learning_rate": 5.0841322204623205e-06,
+      "loss": 0.6178,
+      "step": 8340
+    },
+    {
+      "epoch": 0.9744278374591312,
+      "grad_norm": 0.5892368416358901,
+      "learning_rate": 5.080420450386274e-06,
+      "loss": 0.6296,
+      "step": 8345
+    },
+    {
+      "epoch": 0.9750116767865483,
+      "grad_norm": 0.6172458659875123,
+      "learning_rate": 5.076792269856582e-06,
+      "loss": 0.6265,
+      "step": 8350
+    },
+    {
+      "epoch": 0.9755955161139654,
+      "grad_norm": 0.6232719417629363,
+      "learning_rate": 5.073247692400609e-06,
+      "loss": 0.6198,
+      "step": 8355
+    },
+    {
+      "epoch": 0.9761793554413826,
+      "grad_norm": 0.6131296828724799,
+      "learning_rate": 5.069786731234025e-06,
+      "loss": 0.636,
+      "step": 8360
+    },
+    {
+      "epoch": 0.9767631947687996,
+      "grad_norm": 0.5852765716886096,
+      "learning_rate": 5.066409399260733e-06,
+      "loss": 0.648,
+      "step": 8365
+    },
+    {
+      "epoch": 0.9773470340962167,
+      "grad_norm": 0.5745904770680967,
+      "learning_rate": 5.063115709072837e-06,
+      "loss": 0.6363,
+      "step": 8370
+    },
+    {
+      "epoch": 0.9779308734236338,
+      "grad_norm": 0.605526140146533,
+      "learning_rate": 5.059905672950588e-06,
+      "loss": 0.6341,
+      "step": 8375
+    },
+    {
+      "epoch": 0.978514712751051,
+      "grad_norm": 0.6286995218711023,
+      "learning_rate": 5.056779302862337e-06,
+      "loss": 0.6349,
+      "step": 8380
+    },
+    {
+      "epoch": 0.979098552078468,
+      "grad_norm": 0.5967884859689654,
+      "learning_rate": 5.0537366104645e-06,
+      "loss": 0.6194,
+      "step": 8385
+    },
+    {
+      "epoch": 0.9796823914058851,
+      "grad_norm": 0.6180004875088996,
+      "learning_rate": 5.050777607101506e-06,
+      "loss": 0.6273,
+      "step": 8390
+    },
+    {
+      "epoch": 0.9802662307333022,
+      "grad_norm": 0.6310278459157885,
+      "learning_rate": 5.047902303805746e-06,
+      "loss": 0.6483,
+      "step": 8395
+    },
+    {
+      "epoch": 0.9808500700607193,
+      "grad_norm": 0.6032724694985684,
+      "learning_rate": 5.045110711297557e-06,
+      "loss": 0.6257,
+      "step": 8400
+    },
+    {
+      "epoch": 0.9814339093881363,
+      "grad_norm": 0.5781867381020175,
+      "learning_rate": 5.042402839985161e-06,
+      "loss": 0.6134,
+      "step": 8405
+    },
+    {
+      "epoch": 0.9820177487155535,
+      "grad_norm": 0.6260381778592984,
+      "learning_rate": 5.039778699964626e-06,
+      "loss": 0.6379,
+      "step": 8410
+    },
+    {
+      "epoch": 0.9826015880429706,
+      "grad_norm": 0.6005383882247752,
+      "learning_rate": 5.037238301019845e-06,
+      "loss": 0.6145,
+      "step": 8415
+    },
+    {
+      "epoch": 0.9831854273703877,
+      "grad_norm": 0.594314717849409,
+      "learning_rate": 5.034781652622484e-06,
+      "loss": 0.6172,
+      "step": 8420
+    },
+    {
+      "epoch": 0.9837692666978047,
+      "grad_norm": 0.600166199851812,
+      "learning_rate": 5.032408763931956e-06,
+      "loss": 0.6019,
+      "step": 8425
+    },
+    {
+      "epoch": 0.9843531060252219,
+      "grad_norm": 0.6307929931128271,
+      "learning_rate": 5.0301196437953755e-06,
+      "loss": 0.6229,
+      "step": 8430
+    },
+    {
+      "epoch": 0.984936945352639,
+      "grad_norm": 0.6082576984810791,
+      "learning_rate": 5.0279143007475425e-06,
+      "loss": 0.6346,
+      "step": 8435
+    },
+    {
+      "epoch": 0.985520784680056,
+      "grad_norm": 0.6326185892061758,
+      "learning_rate": 5.02579274301089e-06,
+      "loss": 0.633,
+      "step": 8440
+    },
+    {
+      "epoch": 0.9861046240074731,
+      "grad_norm": 0.6516301943360185,
+      "learning_rate": 5.0237549784954745e-06,
+      "loss": 0.6085,
+      "step": 8445
+    },
+    {
+      "epoch": 0.9866884633348902,
+      "grad_norm": 0.6213879865286841,
+      "learning_rate": 5.021801014798933e-06,
+      "loss": 0.6147,
+      "step": 8450
+    },
+    {
+      "epoch": 0.9872723026623074,
+      "grad_norm": 0.5780399611890119,
+      "learning_rate": 5.0199308592064535e-06,
+      "loss": 0.6184,
+      "step": 8455
+    },
+    {
+      "epoch": 0.9878561419897244,
+      "grad_norm": 0.601910927657346,
+      "learning_rate": 5.018144518690761e-06,
+      "loss": 0.6251,
+      "step": 8460
+    },
+    {
+      "epoch": 0.9884399813171415,
+      "grad_norm": 0.5659705673632358,
+      "learning_rate": 5.016441999912074e-06,
+      "loss": 0.6261,
+      "step": 8465
+    },
+    {
+      "epoch": 0.9890238206445586,
+      "grad_norm": 0.6329264636005011,
+      "learning_rate": 5.014823309218096e-06,
+      "loss": 0.6228,
+      "step": 8470
+    },
+    {
+      "epoch": 0.9896076599719758,
+      "grad_norm": 0.6362609489056387,
+      "learning_rate": 5.013288452643979e-06,
+      "loss": 0.6366,
+      "step": 8475
+    },
+    {
+      "epoch": 0.9901914992993928,
+      "grad_norm": 0.5997302170613081,
+      "learning_rate": 5.011837435912308e-06,
+      "loss": 0.6249,
+      "step": 8480
+    },
+    {
+      "epoch": 0.9907753386268099,
+      "grad_norm": 0.5704092338174375,
+      "learning_rate": 5.010470264433083e-06,
+      "loss": 0.633,
+      "step": 8485
+    },
+    {
+      "epoch": 0.991359177954227,
+      "grad_norm": 0.5666559915764748,
+      "learning_rate": 5.009186943303684e-06,
+      "loss": 0.6375,
+      "step": 8490
+    },
+    {
+      "epoch": 0.9919430172816441,
+      "grad_norm": 0.5759983030786257,
+      "learning_rate": 5.0079874773088735e-06,
+      "loss": 0.6313,
+      "step": 8495
+    },
+    {
+      "epoch": 0.9925268566090611,
+      "grad_norm": 0.608225948606098,
+      "learning_rate": 5.006871870920757e-06,
+      "loss": 0.6208,
+      "step": 8500
+    },
+    {
+      "epoch": 0.9931106959364783,
+      "grad_norm": 0.5628644591986458,
+      "learning_rate": 5.005840128298783e-06,
+      "loss": 0.6231,
+      "step": 8505
+    },
+    {
+      "epoch": 0.9936945352638954,
+      "grad_norm": 0.6025508957094532,
+      "learning_rate": 5.004892253289714e-06,
+      "loss": 0.6318,
+      "step": 8510
+    },
+    {
+      "epoch": 0.9942783745913125,
+      "grad_norm": 0.6076801376241487,
+      "learning_rate": 5.004028249427629e-06,
+      "loss": 0.623,
+      "step": 8515
+    },
+    {
+      "epoch": 0.9948622139187295,
+      "grad_norm": 0.569738097090106,
+      "learning_rate": 5.003248119933894e-06,
+      "loss": 0.6146,
+      "step": 8520
+    },
+    {
+      "epoch": 0.9954460532461467,
+      "grad_norm": 0.5841965142439793,
+      "learning_rate": 5.002551867717153e-06,
+      "loss": 0.6164,
+      "step": 8525
+    },
+    {
+      "epoch": 0.9960298925735638,
+      "grad_norm": 0.6266992389455439,
+      "learning_rate": 5.00193949537333e-06,
+      "loss": 0.646,
+      "step": 8530
+    },
+    {
+      "epoch": 0.9966137319009809,
+      "grad_norm": 0.6190833079470848,
+      "learning_rate": 5.0014110051856e-06,
+      "loss": 0.6542,
+      "step": 8535
+    },
+    {
+      "epoch": 0.9971975712283979,
+      "grad_norm": 0.5923866630501219,
+      "learning_rate": 5.000966399124398e-06,
+      "loss": 0.63,
+      "step": 8540
+    },
+    {
+      "epoch": 0.997781410555815,
+      "grad_norm": 0.6286317642862608,
+      "learning_rate": 5.000605678847399e-06,
+      "loss": 0.6213,
+      "step": 8545
+    },
+    {
+      "epoch": 0.9983652498832322,
+      "grad_norm": 0.5777027991615938,
+      "learning_rate": 5.000328845699522e-06,
+      "loss": 0.6262,
+      "step": 8550
+    },
+    {
+      "epoch": 0.9989490892106492,
+      "grad_norm": 0.6183899021112303,
+      "learning_rate": 5.000135900712914e-06,
+      "loss": 0.6248,
+      "step": 8555
+    },
+    {
+      "epoch": 0.9995329285380663,
+      "grad_norm": 0.5929731972464702,
+      "learning_rate": 5.000026844606953e-06,
+      "loss": 0.6319,
+      "step": 8560
+    },
+    {
+      "epoch": 1.0,
+      "step": 8564,
+      "total_flos": 487937544290304.0,
+      "train_loss": 0.687042359323604,
+      "train_runtime": 31744.1838,
+      "train_samples_per_second": 17.266,
+      "train_steps_per_second": 0.27
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 8564,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 487937544290304.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}