diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10477 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 7440,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 2405.744001883295,
+      "learning_rate": 1.3440860215053765e-08,
+      "loss": 16.4013,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3034.488329915047,
+      "learning_rate": 6.720430107526882e-08,
+      "loss": 15.2227,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2194.668070304958,
+      "learning_rate": 1.3440860215053764e-07,
+      "loss": 14.9719,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1109.7592265856106,
+      "learning_rate": 2.0161290322580645e-07,
+      "loss": 12.059,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 943.00305621849,
+      "learning_rate": 2.688172043010753e-07,
+      "loss": 9.3417,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 478.42527281589776,
+      "learning_rate": 3.360215053763441e-07,
+      "loss": 8.6733,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 565.5993534396201,
+      "learning_rate": 4.032258064516129e-07,
+      "loss": 7.3474,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 416.3498595234287,
+      "learning_rate": 4.7043010752688173e-07,
+      "loss": 6.0082,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 144.83609879706117,
+      "learning_rate": 5.376344086021506e-07,
+      "loss": 5.2806,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 198.8755236355333,
+      "learning_rate": 6.048387096774194e-07,
+      "loss": 4.911,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 141.3997064644455,
+      "learning_rate": 6.720430107526882e-07,
+      "loss": 4.6813,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 133.55678587453863,
+      "learning_rate": 7.392473118279571e-07,
+      "loss": 4.4999,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 89.5221361098514,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 4.1593,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 67.067444507024,
+      "learning_rate": 8.736559139784947e-07,
+      "loss": 4.0657,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 71.8736929708043,
+      "learning_rate": 9.408602150537635e-07,
+      "loss": 3.8795,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 63.35970895428552,
+      "learning_rate": 1.0080645161290323e-06,
+      "loss": 3.6566,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 64.04418691396123,
+      "learning_rate": 1.0752688172043011e-06,
+      "loss": 3.5426,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 72.15811883738871,
+      "learning_rate": 1.14247311827957e-06,
+      "loss": 3.432,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 35.0840730802353,
+      "learning_rate": 1.2096774193548388e-06,
+      "loss": 3.3836,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 35.986671417112184,
+      "learning_rate": 1.2768817204301076e-06,
+      "loss": 3.2688,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 44.94992141846718,
+      "learning_rate": 1.3440860215053765e-06,
+      "loss": 3.1197,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 45.223776066588805,
+      "learning_rate": 1.4112903225806455e-06,
+      "loss": 3.0585,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 31.407139694755763,
+      "learning_rate": 1.4784946236559141e-06,
+      "loss": 2.9057,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 36.20253411854206,
+      "learning_rate": 1.545698924731183e-06,
+      "loss": 2.9106,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 28.700364183150768,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 2.7861,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 27.06342518674971,
+      "learning_rate": 1.6801075268817204e-06,
+      "loss": 2.7283,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 28.540690319879598,
+      "learning_rate": 1.7473118279569895e-06,
+      "loss": 2.6426,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 34.620976558663195,
+      "learning_rate": 1.8145161290322583e-06,
+      "loss": 2.5803,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 26.543690537377664,
+      "learning_rate": 1.881720430107527e-06,
+      "loss": 2.531,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 28.879788866937925,
+      "learning_rate": 1.9489247311827958e-06,
+      "loss": 2.5599,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 27.156547394379697,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 2.4819,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 28.848376888372684,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 2.4144,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 42.21204775186985,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 2.3185,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 30.68767009110495,
+      "learning_rate": 2.217741935483871e-06,
+      "loss": 2.2708,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 26.92156915117187,
+      "learning_rate": 2.28494623655914e-06,
+      "loss": 2.3781,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 36.07878414291202,
+      "learning_rate": 2.3521505376344088e-06,
+      "loss": 2.2721,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 49.41562385012835,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 2.2575,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 80.19360462795755,
+      "learning_rate": 2.4865591397849464e-06,
+      "loss": 2.1993,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 31.355161769761775,
+      "learning_rate": 2.5537634408602153e-06,
+      "loss": 2.1638,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 20.163644022319186,
+      "learning_rate": 2.620967741935484e-06,
+      "loss": 2.1039,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 23.891485968211146,
+      "learning_rate": 2.688172043010753e-06,
+      "loss": 2.16,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 21.854996810869945,
+      "learning_rate": 2.7553763440860214e-06,
+      "loss": 2.0699,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 22.673720996455838,
+      "learning_rate": 2.822580645161291e-06,
+      "loss": 2.1116,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 21.156272800674458,
+      "learning_rate": 2.8897849462365594e-06,
+      "loss": 2.0847,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 44.53918518006611,
+      "learning_rate": 2.9569892473118283e-06,
+      "loss": 2.0735,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 19.749956551312987,
+      "learning_rate": 3.024193548387097e-06,
+      "loss": 2.0162,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 18.9094022713334,
+      "learning_rate": 3.091397849462366e-06,
+      "loss": 2.0357,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 95.50205264471555,
+      "learning_rate": 3.1586021505376348e-06,
+      "loss": 2.0166,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 61.03310995042449,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 1.9872,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 148.06519658847031,
+      "learning_rate": 3.293010752688172e-06,
+      "loss": 1.9794,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 71.88304348970674,
+      "learning_rate": 3.360215053763441e-06,
+      "loss": 1.9394,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 42.55817917403798,
+      "learning_rate": 3.4274193548387097e-06,
+      "loss": 1.9042,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 62.632740823396254,
+      "learning_rate": 3.494623655913979e-06,
+      "loss": 1.9356,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 82.62949499885727,
+      "learning_rate": 3.5618279569892478e-06,
+      "loss": 1.9217,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 106.05019040841613,
+      "learning_rate": 3.6290322580645166e-06,
+      "loss": 1.887,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 55.06301980988493,
+      "learning_rate": 3.6962365591397855e-06,
+      "loss": 1.9451,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 130.87435186526554,
+      "learning_rate": 3.763440860215054e-06,
+      "loss": 1.9274,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 32.26813401369511,
+      "learning_rate": 3.830645161290323e-06,
+      "loss": 1.9515,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 21.287058712378297,
+      "learning_rate": 3.8978494623655915e-06,
+      "loss": 1.8073,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 43.30590780035467,
+      "learning_rate": 3.96505376344086e-06,
+      "loss": 1.8384,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 27.725152849500137,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 1.8629,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 72.58433132209495,
+      "learning_rate": 4.0994623655913985e-06,
+      "loss": 1.8542,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 53.68398067383799,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.9029,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 80.57174908929451,
+      "learning_rate": 4.233870967741936e-06,
+      "loss": 1.8513,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 29.328609119184353,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 1.8422,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 158.61973350833483,
+      "learning_rate": 4.368279569892474e-06,
+      "loss": 1.8495,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 89.79309662280859,
+      "learning_rate": 4.435483870967742e-06,
+      "loss": 1.8471,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 34.585935518827434,
+      "learning_rate": 4.502688172043011e-06,
+      "loss": 1.8231,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 29.047033753427808,
+      "learning_rate": 4.56989247311828e-06,
+      "loss": 1.8467,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 29.36779393455815,
+      "learning_rate": 4.637096774193548e-06,
+      "loss": 1.7917,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 38.27320279386626,
+      "learning_rate": 4.7043010752688175e-06,
+      "loss": 1.801,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 52.440356986032434,
+      "learning_rate": 4.771505376344087e-06,
+      "loss": 1.7392,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 23.469211235073434,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 1.6948,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 122.04808746479898,
+      "learning_rate": 4.9059139784946245e-06,
+      "loss": 1.7054,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 27.353719593612908,
+      "learning_rate": 4.973118279569893e-06,
+      "loss": 1.8008,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 75.58332564395842,
+      "learning_rate": 5.040322580645161e-06,
+      "loss": 1.7638,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 115.84385456972525,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 1.7685,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 20.815649499633817,
+      "learning_rate": 5.174731182795699e-06,
+      "loss": 1.7133,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 56.00497633312478,
+      "learning_rate": 5.241935483870968e-06,
+      "loss": 1.7583,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 22.142864138843443,
+      "learning_rate": 5.309139784946237e-06,
+      "loss": 1.6969,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 47.28452231044601,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 1.6606,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 67.4075196733354,
+      "learning_rate": 5.443548387096774e-06,
+      "loss": 1.7388,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 34.46202571621722,
+      "learning_rate": 5.510752688172043e-06,
+      "loss": 1.6617,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 74.27692687457844,
+      "learning_rate": 5.577956989247312e-06,
+      "loss": 1.6395,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 40.14434097759086,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 1.6555,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 41.14692244976057,
+      "learning_rate": 5.7123655913978505e-06,
+      "loss": 1.6652,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 22.312716185219983,
+      "learning_rate": 5.779569892473119e-06,
+      "loss": 1.6506,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 19.256139604521387,
+      "learning_rate": 5.846774193548388e-06,
+      "loss": 1.7371,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 25.700910574817517,
+      "learning_rate": 5.9139784946236566e-06,
+      "loss": 1.6608,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 22.821561364463633,
+      "learning_rate": 5.981182795698926e-06,
+      "loss": 1.6313,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 19.870552743864323,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 1.6161,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 53.6444075279863,
+      "learning_rate": 6.115591397849463e-06,
+      "loss": 1.6859,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 28.8794203264216,
+      "learning_rate": 6.182795698924732e-06,
+      "loss": 1.6631,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 120.22205680776511,
+      "learning_rate": 6.25e-06,
+      "loss": 1.6452,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 316.225179424074,
+      "learning_rate": 6.3172043010752696e-06,
+      "loss": 1.6546,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 121.16674122762446,
+      "learning_rate": 6.384408602150538e-06,
+      "loss": 1.6831,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 169.5442481771858,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 1.7142,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 125.73244041154908,
+      "learning_rate": 6.518817204301076e-06,
+      "loss": 1.6934,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 108.16723580470715,
+      "learning_rate": 6.586021505376344e-06,
+      "loss": 1.6705,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 30.40652462650454,
+      "learning_rate": 6.653225806451613e-06,
+      "loss": 1.6102,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 31.568179891075236,
+      "learning_rate": 6.720430107526882e-06,
+      "loss": 1.6252,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 43.45226932043275,
+      "learning_rate": 6.787634408602151e-06,
+      "loss": 1.5751,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 149.7079881614224,
+      "learning_rate": 6.854838709677419e-06,
+      "loss": 1.5947,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 26.137998998569532,
+      "learning_rate": 6.9220430107526895e-06,
+      "loss": 1.5897,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 88.0669237337174,
+      "learning_rate": 6.989247311827958e-06,
+      "loss": 1.6109,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 73.8257754559226,
+      "learning_rate": 7.056451612903227e-06,
+      "loss": 1.5878,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 147.4377223256405,
+      "learning_rate": 7.1236559139784956e-06,
+      "loss": 1.6054,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 22.359997880253086,
+      "learning_rate": 7.190860215053764e-06,
+      "loss": 1.6279,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 90.04737946524673,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 1.6141,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 184.74521353082778,
+      "learning_rate": 7.325268817204302e-06,
+      "loss": 1.6657,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 87.6341554878999,
+      "learning_rate": 7.392473118279571e-06,
+      "loss": 1.6697,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 124.15647148280911,
+      "learning_rate": 7.459677419354839e-06,
+      "loss": 1.5679,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 71.76277477811094,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 1.5602,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 69.18166743319745,
+      "learning_rate": 7.594086021505377e-06,
+      "loss": 1.5971,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 28.78194054444829,
+      "learning_rate": 7.661290322580646e-06,
+      "loss": 1.519,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 19.091851411246502,
+      "learning_rate": 7.728494623655915e-06,
+      "loss": 1.5402,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 92.97374658503595,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 1.615,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 27.083399847961495,
+      "learning_rate": 7.862903225806451e-06,
+      "loss": 1.5675,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 101.26575622689107,
+      "learning_rate": 7.93010752688172e-06,
+      "loss": 1.5725,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 105.00439760891088,
+      "learning_rate": 7.99731182795699e-06,
+      "loss": 1.5167,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 29.372122848753133,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 1.5785,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 125.12997709099443,
+      "learning_rate": 8.131720430107529e-06,
+      "loss": 1.5055,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 116.86754106699597,
+      "learning_rate": 8.198924731182797e-06,
+      "loss": 1.552,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 25.816198312391126,
+      "learning_rate": 8.266129032258065e-06,
+      "loss": 1.5503,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 64.77990675037309,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.5202,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 65.8080879982294,
+      "learning_rate": 8.400537634408604e-06,
+      "loss": 1.5369,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 49.62811502509354,
+      "learning_rate": 8.467741935483872e-06,
+      "loss": 1.5436,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 20.67002495077415,
+      "learning_rate": 8.53494623655914e-06,
+      "loss": 1.5056,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 23.4427438772852,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 1.5478,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 19.455471057884907,
+      "learning_rate": 8.669354838709677e-06,
+      "loss": 1.5214,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 33.71020525063305,
+      "learning_rate": 8.736559139784948e-06,
+      "loss": 1.5368,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 18.244470463362397,
+      "learning_rate": 8.803763440860216e-06,
+      "loss": 1.5367,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 80.7680024231703,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 1.5209,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 69.12004977405029,
+      "learning_rate": 8.938172043010753e-06,
+      "loss": 1.5009,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 54.23759857820231,
+      "learning_rate": 9.005376344086021e-06,
+      "loss": 1.4909,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 92.40748934567165,
+      "learning_rate": 9.072580645161291e-06,
+      "loss": 1.5055,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 143.31769519651678,
+      "learning_rate": 9.13978494623656e-06,
+      "loss": 1.4875,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 97.36193747154425,
+      "learning_rate": 9.206989247311828e-06,
+      "loss": 1.4945,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 29.740643098875434,
+      "learning_rate": 9.274193548387097e-06,
+      "loss": 1.5051,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 49.29389525654887,
+      "learning_rate": 9.341397849462367e-06,
+      "loss": 1.5113,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 39.24115135672808,
+      "learning_rate": 9.408602150537635e-06,
+      "loss": 1.48,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 43.22837226234346,
+      "learning_rate": 9.475806451612905e-06,
+      "loss": 1.4782,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 98.36424217860439,
+      "learning_rate": 9.543010752688174e-06,
+      "loss": 1.4411,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 62.56138052343926,
+      "learning_rate": 9.610215053763442e-06,
+      "loss": 1.445,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 190.25287378962327,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 1.4729,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 92.32226152575208,
+      "learning_rate": 9.744623655913979e-06,
+      "loss": 1.4519,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 116.81280536649015,
+      "learning_rate": 9.811827956989249e-06,
+      "loss": 1.513,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 291.52022690918443,
+      "learning_rate": 9.879032258064517e-06,
+      "loss": 1.6681,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 356.3376654572282,
+      "learning_rate": 9.946236559139786e-06,
+      "loss": 1.6312,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 220.45441043367958,
+      "learning_rate": 9.999999449688162e-06,
+      "loss": 1.5773,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 78.05237614822047,
+      "learning_rate": 9.999980188786511e-06,
+      "loss": 1.6092,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 135.03111642896593,
+      "learning_rate": 9.99993341241404e-06,
+      "loss": 1.5106,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 105.52754668242709,
+      "learning_rate": 9.999859120828162e-06,
+      "loss": 1.5305,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 22.459152616619647,
+      "learning_rate": 9.999757314437716e-06,
+      "loss": 1.4181,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 78.86727263490747,
+      "learning_rate": 9.999627993802953e-06,
+      "loss": 1.4567,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 96.97487963294047,
+      "learning_rate": 9.999471159635538e-06,
+      "loss": 1.4171,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 66.73676955331372,
+      "learning_rate": 9.99928681279855e-06,
+      "loss": 1.4588,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 26.962955872031625,
+      "learning_rate": 9.99907495430647e-06,
+      "loss": 1.4622,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 71.73804484311708,
+      "learning_rate": 9.998835585325178e-06,
+      "loss": 1.5276,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 34.831739971575175,
+      "learning_rate": 9.998568707171951e-06,
+      "loss": 1.4745,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 65.78359574681531,
+      "learning_rate": 9.998274321315453e-06,
+      "loss": 1.4481,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 31.706957349839072,
+      "learning_rate": 9.99795242937572e-06,
+      "loss": 1.4591,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 14.376703907334546,
+      "learning_rate": 9.997603033124159e-06,
+      "loss": 1.4533,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 32.30990965535323,
+      "learning_rate": 9.997226134483542e-06,
+      "loss": 1.4228,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 76.3300278384829,
+      "learning_rate": 9.996821735527986e-06,
+      "loss": 1.4161,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 82.54740323475892,
+      "learning_rate": 9.996389838482942e-06,
+      "loss": 1.4087,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 41.92538272672517,
+      "learning_rate": 9.995930445725192e-06,
+      "loss": 1.4019,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 91.90061356954347,
+      "learning_rate": 9.995443559782826e-06,
+      "loss": 1.4318,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 23.127523580087114,
+      "learning_rate": 9.994929183335237e-06,
+      "loss": 1.4262,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 30.533571028127273,
+      "learning_rate": 9.994387319213096e-06,
+      "loss": 1.3977,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 124.51166821611204,
+      "learning_rate": 9.993817970398343e-06,
+      "loss": 1.4285,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 43.75518738413468,
+      "learning_rate": 9.993221140024172e-06,
+      "loss": 1.3578,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 104.75907081855853,
+      "learning_rate": 9.99259683137501e-06,
+      "loss": 1.4132,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 15.485578305808174,
+      "learning_rate": 9.9919450478865e-06,
+      "loss": 1.4205,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 51.78660153786826,
+      "learning_rate": 9.99126579314548e-06,
+      "loss": 1.4,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 38.198018117609955,
+      "learning_rate": 9.99055907088997e-06,
+      "loss": 1.4281,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 16.405157688165122,
+      "learning_rate": 9.989824885009142e-06,
+      "loss": 1.373,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 37.33734419965699,
+      "learning_rate": 9.98906323954331e-06,
+      "loss": 1.3937,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 31.54354697375666,
+      "learning_rate": 9.988274138683896e-06,
+      "loss": 1.3792,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 30.95108771662825,
+      "learning_rate": 9.98745758677341e-06,
+      "loss": 1.3644,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 29.735370838772337,
+      "learning_rate": 9.986613588305435e-06,
+      "loss": 1.4352,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 53.76876219975497,
+      "learning_rate": 9.985742147924594e-06,
+      "loss": 1.4251,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 32.85464865616201,
+      "learning_rate": 9.98484327042652e-06,
+      "loss": 1.3833,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 45.56089408991015,
+      "learning_rate": 9.983916960757842e-06,
+      "loss": 1.3962,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 22.073641499227346,
+      "learning_rate": 9.982963224016152e-06,
+      "loss": 1.3967,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 71.3983312798029,
+      "learning_rate": 9.98198206544997e-06,
+      "loss": 1.4279,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 114.78070373157898,
+      "learning_rate": 9.980973490458728e-06,
+      "loss": 1.3915,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 108.02209218468816,
+      "learning_rate": 9.979937504592731e-06,
+      "loss": 1.3479,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 82.1549015848553,
+      "learning_rate": 9.97887411355313e-06,
+      "loss": 1.3823,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 27.545592321574446,
+      "learning_rate": 9.977783323191885e-06,
+      "loss": 1.3568,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 18.85495269820905,
+      "learning_rate": 9.976665139511747e-06,
+      "loss": 1.3817,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 26.225586592183646,
+      "learning_rate": 9.975519568666208e-06,
+      "loss": 1.3611,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 16.606612034755415,
+      "learning_rate": 9.974346616959476e-06,
+      "loss": 1.3375,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 16.294572009334328,
+      "learning_rate": 9.973146290846444e-06,
+      "loss": 1.3749,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 28.496884185291073,
+      "learning_rate": 9.971918596932642e-06,
+      "loss": 1.3865,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 35.30571084433214,
+      "learning_rate": 9.970663541974216e-06,
+      "loss": 1.3589,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 49.15304817138109,
+      "learning_rate": 9.969381132877873e-06,
+      "loss": 1.3296,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 55.32282620037942,
+      "learning_rate": 9.968071376700865e-06,
+      "loss": 1.3671,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 17.823341685824705,
+      "learning_rate": 9.96673428065093e-06,
+      "loss": 1.3479,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 58.59615740509754,
+      "learning_rate": 9.965369852086259e-06,
+      "loss": 1.2876,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 37.75057737791426,
+      "learning_rate": 9.963978098515468e-06,
+      "loss": 1.3618,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 30.26048440538982,
+      "learning_rate": 9.96255902759753e-06,
+      "loss": 1.3614,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 35.587405533117405,
+      "learning_rate": 9.961112647141765e-06,
+      "loss": 1.3281,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 65.91103463522634,
+      "learning_rate": 9.959638965107767e-06,
+      "loss": 1.3363,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 109.41009141202416,
+      "learning_rate": 9.95813798960538e-06,
+      "loss": 1.3955,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 25.788301271590747,
+      "learning_rate": 9.956609728894648e-06,
+      "loss": 1.3587,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 37.81687531410579,
+      "learning_rate": 9.955054191385764e-06,
+      "loss": 1.3584,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 44.454331958590565,
+      "learning_rate": 9.953471385639035e-06,
+      "loss": 1.2848,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 18.607449172766433,
+      "learning_rate": 9.951861320364822e-06,
+      "loss": 1.3328,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 47.820356043232294,
+      "learning_rate": 9.950224004423502e-06,
+      "loss": 1.3214,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 54.68385445449125,
+      "learning_rate": 9.948559446825412e-06,
+      "loss": 1.2909,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 23.93641581026352,
+      "learning_rate": 9.94686765673081e-06,
+      "loss": 1.2826,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 38.61137909099191,
+      "learning_rate": 9.945148643449809e-06,
+      "loss": 1.3475,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 51.58880200991109,
+      "learning_rate": 9.94340241644234e-06,
+      "loss": 1.3366,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 78.84714564786627,
+      "learning_rate": 9.941628985318094e-06,
+      "loss": 1.367,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 33.10275420828958,
+      "learning_rate": 9.939828359836468e-06,
+      "loss": 1.2973,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 57.000986979043475,
+      "learning_rate": 9.938000549906509e-06,
+      "loss": 1.3115,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 12.260954765550334,
+      "learning_rate": 9.936145565586871e-06,
+      "loss": 1.3591,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 24.184895415543437,
+      "learning_rate": 9.934263417085746e-06,
+      "loss": 1.3529,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 28.38759761728956,
+      "learning_rate": 9.932354114760819e-06,
+      "loss": 1.3278,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 28.265439590291777,
+      "learning_rate": 9.930417669119195e-06,
+      "loss": 1.2881,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 19.79257726694249,
+      "learning_rate": 9.928454090817366e-06,
+      "loss": 1.3244,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 18.81138887973412,
+      "learning_rate": 9.926463390661127e-06,
+      "loss": 1.3163,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 15.45271589032639,
+      "learning_rate": 9.924445579605534e-06,
+      "loss": 1.3339,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 35.67290775317024,
+      "learning_rate": 9.922400668754833e-06,
+      "loss": 1.336,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 36.016211519831224,
+      "learning_rate": 9.920328669362408e-06,
+      "loss": 1.3309,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 41.136663477634876,
+      "learning_rate": 9.91822959283071e-06,
+      "loss": 1.2888,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 71.39059456429649,
+      "learning_rate": 9.916103450711203e-06,
+      "loss": 1.2846,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 14.308431453840171,
+      "learning_rate": 9.913950254704291e-06,
+      "loss": 1.3203,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 19.683299521176988,
+      "learning_rate": 9.911770016659263e-06,
+      "loss": 1.3101,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 57.69862211953297,
+      "learning_rate": 9.909562748574219e-06,
+      "loss": 1.3216,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 80.37479888512688,
+      "learning_rate": 9.907328462596015e-06,
+      "loss": 1.3281,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 93.55250241336888,
+      "learning_rate": 9.905067171020185e-06,
+      "loss": 1.3178,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 126.4183263179252,
+      "learning_rate": 9.902778886290877e-06,
+      "loss": 1.3276,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 20.98202085655841,
+      "learning_rate": 9.900463621000789e-06,
+      "loss": 1.3429,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 83.50991308572252,
+      "learning_rate": 9.898121387891093e-06,
+      "loss": 1.307,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 117.549332724898,
+      "learning_rate": 9.895752199851372e-06,
+      "loss": 1.3501,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 18.381523104894782,
+      "learning_rate": 9.893356069919538e-06,
+      "loss": 1.3263,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 19.331193749134517,
+      "learning_rate": 9.890933011281777e-06,
+      "loss": 1.3478,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 18.049439306686804,
+      "learning_rate": 9.888483037272455e-06,
+      "loss": 1.2933,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 39.59046179151359,
+      "learning_rate": 9.88600616137407e-06,
+      "loss": 1.2854,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 26.130533942662304,
+      "learning_rate": 9.883502397217156e-06,
+      "loss": 1.2798,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 22.071697667770607,
+      "learning_rate": 9.880971758580215e-06,
+      "loss": 1.3116,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 12.72020076792302,
+      "learning_rate": 9.878414259389646e-06,
+      "loss": 1.3055,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 20.237947004394226,
+      "learning_rate": 9.875829913719664e-06,
+      "loss": 1.3189,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 31.873048673900332,
+      "learning_rate": 9.873218735792222e-06,
+      "loss": 1.2765,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 27.088563725925333,
+      "learning_rate": 9.870580739976936e-06,
+      "loss": 1.3102,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 13.531456471383404,
+      "learning_rate": 9.867915940791002e-06,
+      "loss": 1.3141,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 27.865087251416963,
+      "learning_rate": 9.86522435289912e-06,
+      "loss": 1.2778,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 53.98402677705456,
+      "learning_rate": 9.862505991113409e-06,
+      "loss": 1.3094,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 27.88968143492497,
+      "learning_rate": 9.85976087039333e-06,
+      "loss": 1.254,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 25.85297515433478,
+      "learning_rate": 9.8569890058456e-06,
+      "loss": 1.2451,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 26.84638324463917,
+      "learning_rate": 9.854190412724114e-06,
+      "loss": 1.3181,
+      "step": 1260
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 15.851942478228562,
+      "learning_rate": 9.85136510642985e-06,
+      "loss": 1.2916,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 22.671976970541067,
+      "learning_rate": 9.8485131025108e-06,
+      "loss": 1.2341,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 18.437688673195737,
+      "learning_rate": 9.845634416661867e-06,
+      "loss": 1.3045,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 15.703812477381069,
+      "learning_rate": 9.8427290647248e-06,
+      "loss": 1.3057,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 16.211393182922986,
+      "learning_rate": 9.839797062688084e-06,
+      "loss": 1.3173,
+      "step": 1285
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 22.041910721041404,
+      "learning_rate": 9.836838426686865e-06,
+      "loss": 1.2584,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 31.985533853613575,
+      "learning_rate": 9.833853173002862e-06,
+      "loss": 1.2427,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 43.2359613692205,
+      "learning_rate": 9.830841318064273e-06,
+      "loss": 1.3166,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 33.25595187233199,
+      "learning_rate": 9.82780287844568e-06,
+      "loss": 1.3012,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 19.93101882119115,
+      "learning_rate": 9.824737870867977e-06,
+      "loss": 1.2803,
+      "step": 1310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 61.8929882355421,
+      "learning_rate": 9.82164631219825e-06,
+      "loss": 1.2567,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 17.35420830500758,
+      "learning_rate": 9.818528219449705e-06,
+      "loss": 1.2797,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 46.85772955287638,
+      "learning_rate": 9.81538360978157e-06,
+      "loss": 1.2805,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 14.261225266706827,
+      "learning_rate": 9.812212500498997e-06,
+      "loss": 1.2951,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 41.95393995518883,
+      "learning_rate": 9.809014909052966e-06,
+      "loss": 1.3067,
+      "step": 1335
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 32.442857600519396,
+      "learning_rate": 9.805790853040196e-06,
+      "loss": 1.2862,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 48.33715508637013,
+      "learning_rate": 9.802540350203039e-06,
+      "loss": 1.282,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 119.1461084806285,
+      "learning_rate": 9.79926341842939e-06,
+      "loss": 1.2686,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 19.41390747179324,
+      "learning_rate": 9.795960075752584e-06,
+      "loss": 1.2404,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 23.593253599284026,
+      "learning_rate": 9.792630340351301e-06,
+      "loss": 1.3194,
+      "step": 1360
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 99.72956707812463,
+      "learning_rate": 9.789274230549456e-06,
+      "loss": 1.3039,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 183.51250735641264,
+      "learning_rate": 9.785891764816117e-06,
+      "loss": 1.2991,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 111.31189190292812,
+      "learning_rate": 9.782482961765383e-06,
+      "loss": 1.3393,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 57.621513450395966,
+      "learning_rate": 9.779047840156288e-06,
+      "loss": 1.3496,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 224.84362627842353,
+      "learning_rate": 9.77558641889271e-06,
+      "loss": 1.374,
+      "step": 1385
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 108.64616430086639,
+      "learning_rate": 9.772098717023251e-06,
+      "loss": 1.2826,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 80.49628966307156,
+      "learning_rate": 9.768584753741134e-06,
+      "loss": 1.285,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 114.56926153475547,
+      "learning_rate": 9.765044548384113e-06,
+      "loss": 1.2996,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 56.84323238972865,
+      "learning_rate": 9.761478120434346e-06,
+      "loss": 1.3199,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 15.150030759879591,
+      "learning_rate": 9.757885489518296e-06,
+      "loss": 1.3021,
+      "step": 1410
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 137.76855197950997,
+      "learning_rate": 9.754266675406635e-06,
+      "loss": 1.288,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 58.6931471963433,
+      "learning_rate": 9.75062169801411e-06,
+      "loss": 1.2575,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 11.812252416929596,
+      "learning_rate": 9.746950577399458e-06,
+      "loss": 1.2666,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 96.19518185816071,
+      "learning_rate": 9.74325333376528e-06,
+      "loss": 1.2562,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 41.41343310410979,
+      "learning_rate": 9.739529987457937e-06,
+      "loss": 1.2906,
+      "step": 1435
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 61.86287366206095,
+      "learning_rate": 9.735780558967434e-06,
+      "loss": 1.3053,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 41.78455096075733,
+      "learning_rate": 9.732005068927314e-06,
+      "loss": 1.2722,
+      "step": 1445
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 19.96119388415896,
+      "learning_rate": 9.728203538114533e-06,
+      "loss": 1.2972,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 27.53156133860246,
+      "learning_rate": 9.72437598744936e-06,
+      "loss": 1.2622,
+      "step": 1455
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 27.46859648803268,
+      "learning_rate": 9.720522437995246e-06,
+      "loss": 1.2431,
+      "step": 1460
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 16.990826404472397,
+      "learning_rate": 9.716642910958723e-06,
+      "loss": 1.2525,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 22.88378620636613,
+      "learning_rate": 9.712737427689275e-06,
+      "loss": 1.2587,
+      "step": 1470
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 12.442134327263519,
+      "learning_rate": 9.708806009679234e-06,
+      "loss": 1.244,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 81.96482418927604,
+      "learning_rate": 9.70484867856365e-06,
+      "loss": 1.2888,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 30.119061996487957,
+      "learning_rate": 9.700865456120168e-06,
+      "loss": 1.2899,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 15.530972405955056,
+      "learning_rate": 9.696856364268928e-06,
+      "loss": 1.2586,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 17.031699636462967,
+      "learning_rate": 9.692821425072428e-06,
+      "loss": 1.2408,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 16.689591527147797,
+      "learning_rate": 9.688760660735403e-06,
+      "loss": 1.2146,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 44.120684544668464,
+      "learning_rate": 9.684674093604713e-06,
+      "loss": 1.2497,
+      "step": 1505
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 31.145849380952164,
+      "learning_rate": 9.680561746169211e-06,
+      "loss": 1.2484,
+      "step": 1510
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 12.784383990756531,
+      "learning_rate": 9.676423641059617e-06,
+      "loss": 1.2573,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 12.892352528526105,
+      "learning_rate": 9.67225980104841e-06,
+      "loss": 1.2482,
+      "step": 1520
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 41.336648198731716,
+      "learning_rate": 9.668070249049682e-06,
+      "loss": 1.2636,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 18.994035338368224,
+      "learning_rate": 9.66385500811902e-06,
+      "loss": 1.2286,
+      "step": 1530
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 34.785160539270905,
+      "learning_rate": 9.65961410145339e-06,
+      "loss": 1.252,
+      "step": 1535
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 19.142178309672364,
+      "learning_rate": 9.655347552390987e-06,
+      "loss": 1.2856,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 34.07491704273605,
+      "learning_rate": 9.65105538441113e-06,
+      "loss": 1.2688,
+      "step": 1545
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 29.062155407795853,
+      "learning_rate": 9.646737621134112e-06,
+      "loss": 1.215,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 37.75134880000602,
+      "learning_rate": 9.642394286321092e-06,
+      "loss": 1.252,
+      "step": 1555
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 24.144174972799604,
+      "learning_rate": 9.638025403873939e-06,
+      "loss": 1.2844,
+      "step": 1560
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 71.47772818886311,
+      "learning_rate": 9.633630997835125e-06,
+      "loss": 1.2526,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 23.73415171894665,
+      "learning_rate": 9.629211092387568e-06,
+      "loss": 1.2784,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 15.037257351444415,
+      "learning_rate": 9.624765711854531e-06,
+      "loss": 1.271,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 23.03882395507045,
+      "learning_rate": 9.62029488069945e-06,
+      "loss": 1.2505,
+      "step": 1580
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 9.210592478807614,
+      "learning_rate": 9.615798623525831e-06,
+      "loss": 1.2356,
+      "step": 1585
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 14.183678897776636,
+      "learning_rate": 9.611276965077099e-06,
+      "loss": 1.2805,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 45.64866322625938,
+      "learning_rate": 9.606729930236463e-06,
+      "loss": 1.2781,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 32.5896663980658,
+      "learning_rate": 9.602157544026785e-06,
+      "loss": 1.231,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 33.30146493427689,
+      "learning_rate": 9.597559831610435e-06,
+      "loss": 1.2276,
+      "step": 1605
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 69.26817624038185,
+      "learning_rate": 9.592936818289158e-06,
+      "loss": 1.2456,
+      "step": 1610
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 66.73362945811155,
+      "learning_rate": 9.58828852950393e-06,
+      "loss": 1.2474,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 22.970426868509886,
+      "learning_rate": 9.58361499083483e-06,
+      "loss": 1.2376,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 29.54147666440491,
+      "learning_rate": 9.578916228000876e-06,
+      "loss": 1.2636,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 13.839530834648398,
+      "learning_rate": 9.574192266859909e-06,
+      "loss": 1.2322,
+      "step": 1630
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 44.91473533547081,
+      "learning_rate": 9.569443133408434e-06,
+      "loss": 1.283,
+      "step": 1635
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 69.09130264548097,
+      "learning_rate": 9.564668853781483e-06,
+      "loss": 1.2767,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 34.76730315570345,
+      "learning_rate": 9.55986945425247e-06,
+      "loss": 1.2581,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 67.13936878197774,
+      "learning_rate": 9.555044961233047e-06,
+      "loss": 1.1957,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 93.33158887794684,
+      "learning_rate": 9.55019540127296e-06,
+      "loss": 1.2471,
+      "step": 1655
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 72.92634997428354,
+      "learning_rate": 9.5453208010599e-06,
+      "loss": 1.2644,
+      "step": 1660
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 71.84144410962492,
+      "learning_rate": 9.540421187419355e-06,
+      "loss": 1.2184,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 105.83362024344693,
+      "learning_rate": 9.535496587314468e-06,
+      "loss": 1.2135,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 97.36345614802303,
+      "learning_rate": 9.530547027845886e-06,
+      "loss": 1.2912,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 74.28181401831145,
+      "learning_rate": 9.525572536251608e-06,
+      "loss": 1.2026,
+      "step": 1680
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 49.820474017113035,
+      "learning_rate": 9.520573139906835e-06,
+      "loss": 1.2335,
+      "step": 1685
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 68.20267751489494,
+      "learning_rate": 9.51554886632383e-06,
+      "loss": 1.2165,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 118.0720632731834,
+      "learning_rate": 9.510499743151751e-06,
+      "loss": 1.2526,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 26.705492104039806,
+      "learning_rate": 9.505425798176506e-06,
+      "loss": 1.199,
+      "step": 1700
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 60.06112754213599,
+      "learning_rate": 9.500327059320606e-06,
+      "loss": 1.2293,
+      "step": 1705
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 24.183247547681844,
+      "learning_rate": 9.495203554643001e-06,
+      "loss": 1.2443,
+      "step": 1710
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 24.32499961073832,
+      "learning_rate": 9.490055312338933e-06,
+      "loss": 1.1944,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 18.20393601452291,
+      "learning_rate": 9.484882360739772e-06,
+      "loss": 1.2184,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 40.866092970463804,
+      "learning_rate": 9.479684728312875e-06,
+      "loss": 1.2389,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 92.35475245959559,
+      "learning_rate": 9.474462443661414e-06,
+      "loss": 1.2319,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 41.519813441456535,
+      "learning_rate": 9.469215535524228e-06,
+      "loss": 1.208,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 47.23986610846689,
+      "learning_rate": 9.46394403277566e-06,
+      "loss": 1.2013,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 27.51877463083975,
+      "learning_rate": 9.4586479644254e-06,
+      "loss": 1.2269,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 23.596945501836345,
+      "learning_rate": 9.45332735961833e-06,
+      "loss": 1.2363,
+      "step": 1750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 63.94255905141467,
+      "learning_rate": 9.44798224763435e-06,
+      "loss": 1.2354,
+      "step": 1755
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 59.78800938724632,
+      "learning_rate": 9.442612657888237e-06,
+      "loss": 1.2401,
+      "step": 1760
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 89.16456759582186,
+      "learning_rate": 9.437218619929461e-06,
+      "loss": 1.2247,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 17.952864484722834,
+      "learning_rate": 9.431800163442043e-06,
+      "loss": 1.2396,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 13.73846058731808,
+      "learning_rate": 9.426357318244372e-06,
+      "loss": 1.2091,
+      "step": 1775
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 72.76269270079044,
+      "learning_rate": 9.42089011428906e-06,
+      "loss": 1.2101,
+      "step": 1780
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 41.1494452737733,
+      "learning_rate": 9.415398581662766e-06,
+      "loss": 1.2116,
+      "step": 1785
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 40.01241452374857,
+      "learning_rate": 9.409882750586027e-06,
+      "loss": 1.1853,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 27.221899797256327,
+      "learning_rate": 9.404342651413103e-06,
+      "loss": 1.2194,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 91.82911823396265,
+      "learning_rate": 9.398778314631801e-06,
+      "loss": 1.214,
+      "step": 1800
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 72.85176951438244,
+      "learning_rate": 9.393189770863313e-06,
+      "loss": 1.2686,
+      "step": 1805
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 28.382099335836124,
+      "learning_rate": 9.387577050862043e-06,
+      "loss": 1.2328,
+      "step": 1810
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 47.78619571105184,
+      "learning_rate": 9.38194018551544e-06,
+      "loss": 1.2266,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 21.50835726202906,
+      "learning_rate": 9.376279205843827e-06,
+      "loss": 1.2205,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 18.25140872912014,
+      "learning_rate": 9.370594143000234e-06,
+      "loss": 1.17,
+      "step": 1825
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 11.796348568361772,
+      "learning_rate": 9.364885028270219e-06,
+      "loss": 1.1944,
+      "step": 1830
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 70.58248149980511,
+      "learning_rate": 9.359151893071701e-06,
+      "loss": 1.2284,
+      "step": 1835
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 81.69301427989976,
+      "learning_rate": 9.353394768954791e-06,
+      "loss": 1.2216,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 65.0523619329761,
+      "learning_rate": 9.347613687601611e-06,
+      "loss": 1.2125,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 39.19694472155119,
+      "learning_rate": 9.341808680826119e-06,
+      "loss": 1.2081,
+      "step": 1850
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 85.6467307326097,
+      "learning_rate": 9.335979780573942e-06,
+      "loss": 1.2182,
+      "step": 1855
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 76.74276719526604,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 1.2297,
+      "step": 1860
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 12.54421395508913,
+      "learning_rate": 9.324250428079303e-06,
+      "loss": 1.185,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 22.64501494887728,
+      "learning_rate": 9.318350040384825e-06,
+      "loss": 1.2319,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 35.310515985194684,
+      "learning_rate": 9.312425888309282e-06,
+      "loss": 1.2294,
+      "step": 1875
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 11.406451300971566,
+      "learning_rate": 9.30647800445397e-06,
+      "loss": 1.2343,
+      "step": 1880
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 54.73585951186893,
+      "learning_rate": 9.300506421550781e-06,
+      "loss": 1.2269,
+      "step": 1885
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 11.59997902279277,
+      "learning_rate": 9.29451117246203e-06,
+      "loss": 1.1945,
+      "step": 1890
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 22.89675471898756,
+      "learning_rate": 9.288492290180266e-06,
+      "loss": 1.2454,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 46.55068281863006,
+      "learning_rate": 9.2824498078281e-06,
+      "loss": 1.2058,
+      "step": 1900
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 79.02174211630475,
+      "learning_rate": 9.27638375865801e-06,
+      "loss": 1.1856,
+      "step": 1905
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 51.19315475520796,
+      "learning_rate": 9.27029417605217e-06,
+      "loss": 1.1968,
+      "step": 1910
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 65.39683060095506,
+      "learning_rate": 9.264181093522258e-06,
+      "loss": 1.2445,
+      "step": 1915
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 25.01539534644911,
+      "learning_rate": 9.258044544709276e-06,
+      "loss": 1.2034,
+      "step": 1920
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 27.573144653965674,
+      "learning_rate": 9.251884563383363e-06,
+      "loss": 1.2061,
+      "step": 1925
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 31.302322831757646,
+      "learning_rate": 9.245701183443612e-06,
+      "loss": 1.2363,
+      "step": 1930
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 77.48271654234222,
+      "learning_rate": 9.239494438917881e-06,
+      "loss": 1.2077,
+      "step": 1935
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 109.2806382548179,
+      "learning_rate": 9.233264363962602e-06,
+      "loss": 1.2325,
+      "step": 1940
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 111.22392412722785,
+      "learning_rate": 9.227010992862602e-06,
+      "loss": 1.2232,
+      "step": 1945
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 24.201003951280565,
+      "learning_rate": 9.220734360030908e-06,
+      "loss": 1.2251,
+      "step": 1950
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 98.21487054636425,
+      "learning_rate": 9.214434500008555e-06,
+      "loss": 1.2425,
+      "step": 1955
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 34.46247299364145,
+      "learning_rate": 9.208111447464407e-06,
+      "loss": 1.2403,
+      "step": 1960
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 51.79925138895671,
+      "learning_rate": 9.201765237194955e-06,
+      "loss": 1.211,
+      "step": 1965
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 18.6823580389484,
+      "learning_rate": 9.195395904124126e-06,
+      "loss": 1.2336,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 10.265915792652455,
+      "learning_rate": 9.189003483303104e-06,
+      "loss": 1.2386,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 19.723674009137426,
+      "learning_rate": 9.182588009910119e-06,
+      "loss": 1.2283,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 18.150689642053763,
+      "learning_rate": 9.176149519250264e-06,
+      "loss": 1.2019,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 36.71799741057715,
+      "learning_rate": 9.169688046755302e-06,
+      "loss": 1.2372,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 20.819677998506364,
+      "learning_rate": 9.163203627983468e-06,
+      "loss": 1.2141,
+      "step": 1995
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 18.688397366071406,
+      "learning_rate": 9.156696298619266e-06,
+      "loss": 1.2151,
+      "step": 2000
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 38.84444336197096,
+      "learning_rate": 9.150166094473282e-06,
+      "loss": 1.1898,
+      "step": 2005
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 16.559502112235748,
+      "learning_rate": 9.143613051481994e-06,
+      "loss": 1.2,
+      "step": 2010
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 24.348427932448423,
+      "learning_rate": 9.137037205707552e-06,
+      "loss": 1.1948,
+      "step": 2015
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 13.569601089502072,
+      "learning_rate": 9.130438593337603e-06,
+      "loss": 1.1547,
+      "step": 2020
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 26.60756353902665,
+      "learning_rate": 9.12381725068507e-06,
+      "loss": 1.1944,
+      "step": 2025
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 11.640717034955413,
+      "learning_rate": 9.117173214187972e-06,
+      "loss": 1.2149,
+      "step": 2030
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 50.68489988886974,
+      "learning_rate": 9.110506520409213e-06,
+      "loss": 1.1866,
+      "step": 2035
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 38.78614229270772,
+      "learning_rate": 9.103817206036383e-06,
+      "loss": 1.2105,
+      "step": 2040
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 22.243967064566867,
+      "learning_rate": 9.097105307881552e-06,
+      "loss": 1.201,
+      "step": 2045
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 16.19663532790498,
+      "learning_rate": 9.090370862881074e-06,
+      "loss": 1.1868,
+      "step": 2050
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 18.228565591182583,
+      "learning_rate": 9.083613908095386e-06,
+      "loss": 1.2985,
+      "step": 2055
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 28.080970000905847,
+      "learning_rate": 9.076834480708788e-06,
+      "loss": 1.2218,
+      "step": 2060
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 68.97254875036558,
+      "learning_rate": 9.070032618029256e-06,
+      "loss": 1.267,
+      "step": 2065
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 55.587757182551115,
+      "learning_rate": 9.06320835748823e-06,
+      "loss": 1.1915,
+      "step": 2070
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 29.8106436840182,
+      "learning_rate": 9.056361736640407e-06,
+      "loss": 1.2097,
+      "step": 2075
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 37.83325544510067,
+      "learning_rate": 9.049492793163539e-06,
+      "loss": 1.1808,
+      "step": 2080
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 10.152194020824325,
+      "learning_rate": 9.042601564858214e-06,
+      "loss": 1.2113,
+      "step": 2085
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 17.81441360493928,
+      "learning_rate": 9.03568808964766e-06,
+      "loss": 1.2142,
+      "step": 2090
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 10.963255004587548,
+      "learning_rate": 9.028752405577535e-06,
+      "loss": 1.1636,
+      "step": 2095
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 19.94462584855531,
+      "learning_rate": 9.021794550815713e-06,
+      "loss": 1.2378,
+      "step": 2100
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 40.99194638099434,
+      "learning_rate": 9.014814563652077e-06,
+      "loss": 1.2137,
+      "step": 2105
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 74.55679120071602,
+      "learning_rate": 9.007812482498302e-06,
+      "loss": 1.1863,
+      "step": 2110
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 32.70450389121002,
+      "learning_rate": 9.000788345887655e-06,
+      "loss": 1.2127,
+      "step": 2115
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 29.773704970689955,
+      "learning_rate": 8.993742192474773e-06,
+      "loss": 1.2217,
+      "step": 2120
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 21.22031863340936,
+      "learning_rate": 8.986674061035456e-06,
+      "loss": 1.2323,
+      "step": 2125
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 27.68350129099609,
+      "learning_rate": 8.979583990466454e-06,
+      "loss": 1.248,
+      "step": 2130
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 30.386416668864385,
+      "learning_rate": 8.972472019785244e-06,
+      "loss": 1.2092,
+      "step": 2135
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 19.517795800460053,
+      "learning_rate": 8.965338188129823e-06,
+      "loss": 1.1808,
+      "step": 2140
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 13.553988632026039,
+      "learning_rate": 8.958182534758502e-06,
+      "loss": 1.1733,
+      "step": 2145
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 27.869427429755273,
+      "learning_rate": 8.951005099049663e-06,
+      "loss": 1.1844,
+      "step": 2150
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 39.68582477498169,
+      "learning_rate": 8.94380592050157e-06,
+      "loss": 1.1786,
+      "step": 2155
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 92.22511839711535,
+      "learning_rate": 8.936585038732143e-06,
+      "loss": 1.2144,
+      "step": 2160
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 137.73864765286555,
+      "learning_rate": 8.929342493478721e-06,
+      "loss": 1.1941,
+      "step": 2165
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 93.65382199294945,
+      "learning_rate": 8.92207832459788e-06,
+      "loss": 1.1957,
+      "step": 2170
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 132.76276423721396,
+      "learning_rate": 8.914792572065178e-06,
+      "loss": 1.2219,
+      "step": 2175
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 59.69241023885523,
+      "learning_rate": 8.907485275974959e-06,
+      "loss": 1.2649,
+      "step": 2180
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 68.86723598658776,
+      "learning_rate": 8.900156476540119e-06,
+      "loss": 1.1914,
+      "step": 2185
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 68.88903296874858,
+      "learning_rate": 8.892806214091894e-06,
+      "loss": 1.2206,
+      "step": 2190
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 144.9718517912028,
+      "learning_rate": 8.885434529079626e-06,
+      "loss": 1.2057,
+      "step": 2195
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 98.20156012388937,
+      "learning_rate": 8.878041462070556e-06,
+      "loss": 1.2649,
+      "step": 2200
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 76.38888144711643,
+      "learning_rate": 8.87062705374959e-06,
+      "loss": 1.2159,
+      "step": 2205
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 27.817297330819205,
+      "learning_rate": 8.863191344919074e-06,
+      "loss": 1.2353,
+      "step": 2210
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 33.6536090571804,
+      "learning_rate": 8.855734376498576e-06,
+      "loss": 1.2008,
+      "step": 2215
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 12.033556316415826,
+      "learning_rate": 8.848256189524661e-06,
+      "loss": 1.1971,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 22.095611956106953,
+      "learning_rate": 8.840756825150659e-06,
+      "loss": 1.1679,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 23.29990032198999,
+      "learning_rate": 8.833236324646437e-06,
+      "loss": 1.1755,
+      "step": 2230
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 37.16525637778255,
+      "learning_rate": 8.825694729398187e-06,
+      "loss": 1.1941,
+      "step": 2235
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 24.418057882118365,
+      "learning_rate": 8.818132080908178e-06,
+      "loss": 1.1951,
+      "step": 2240
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 10.866303596839401,
+      "learning_rate": 8.810548420794543e-06,
+      "loss": 1.2037,
+      "step": 2245
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 14.613749692453576,
+      "learning_rate": 8.802943790791045e-06,
+      "loss": 1.1664,
+      "step": 2250
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 17.078484071937776,
+      "learning_rate": 8.795318232746844e-06,
+      "loss": 1.1895,
+      "step": 2255
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 24.372266680343394,
+      "learning_rate": 8.78767178862627e-06,
+      "loss": 1.1974,
+      "step": 2260
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 13.49033393287541,
+      "learning_rate": 8.780004500508587e-06,
+      "loss": 1.216,
+      "step": 2265
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 25.24602077065757,
+      "learning_rate": 8.77231641058778e-06,
+      "loss": 1.1642,
+      "step": 2270
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 57.54455368044807,
+      "learning_rate": 8.764607561172292e-06,
+      "loss": 1.2057,
+      "step": 2275
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 15.087989021875776,
+      "learning_rate": 8.756877994684818e-06,
+      "loss": 1.1605,
+      "step": 2280
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 15.21317699168746,
+      "learning_rate": 8.749127753662059e-06,
+      "loss": 1.1809,
+      "step": 2285
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 27.828075829213894,
+      "learning_rate": 8.74135688075449e-06,
+      "loss": 1.1868,
+      "step": 2290
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 19.28153095603385,
+      "learning_rate": 8.733565418726126e-06,
+      "loss": 1.2211,
+      "step": 2295
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 22.48790510139669,
+      "learning_rate": 8.725753410454286e-06,
+      "loss": 1.1788,
+      "step": 2300
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 54.68299195731748,
+      "learning_rate": 8.717920898929355e-06,
+      "loss": 1.1703,
+      "step": 2305
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 50.21262012359284,
+      "learning_rate": 8.710067927254555e-06,
+      "loss": 1.1563,
+      "step": 2310
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 15.039924211844733,
+      "learning_rate": 8.702194538645698e-06,
+      "loss": 1.1606,
+      "step": 2315
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 11.3320184175023,
+      "learning_rate": 8.694300776430958e-06,
+      "loss": 1.1716,
+      "step": 2320
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 52.97658708505161,
+      "learning_rate": 8.68638668405062e-06,
+      "loss": 1.2033,
+      "step": 2325
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 60.16455692219241,
+      "learning_rate": 8.678452305056855e-06,
+      "loss": 1.1351,
+      "step": 2330
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 54.5323334555245,
+      "learning_rate": 8.67049768311347e-06,
+      "loss": 1.2015,
+      "step": 2335
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 10.668608422184063,
+      "learning_rate": 8.66252286199567e-06,
+      "loss": 1.17,
+      "step": 2340
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 47.98752570948412,
+      "learning_rate": 8.654527885589823e-06,
+      "loss": 1.1876,
+      "step": 2345
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 104.19257200772952,
+      "learning_rate": 8.646512797893209e-06,
+      "loss": 1.1882,
+      "step": 2350
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 69.20118488487903,
+      "learning_rate": 8.638477643013789e-06,
+      "loss": 1.1759,
+      "step": 2355
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 28.27492834861394,
+      "learning_rate": 8.630422465169947e-06,
+      "loss": 1.1884,
+      "step": 2360
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 21.920246857166433,
+      "learning_rate": 8.622347308690264e-06,
+      "loss": 1.1569,
+      "step": 2365
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 46.93097177469271,
+      "learning_rate": 8.614252218013264e-06,
+      "loss": 1.1845,
+      "step": 2370
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 31.169777809227252,
+      "learning_rate": 8.606137237687166e-06,
+      "loss": 1.2099,
+      "step": 2375
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 24.86557434591264,
+      "learning_rate": 8.598002412369649e-06,
+      "loss": 1.1903,
+      "step": 2380
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 40.31682653584947,
+      "learning_rate": 8.589847786827601e-06,
+      "loss": 1.1729,
+      "step": 2385
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 47.80468052861838,
+      "learning_rate": 8.581673405936873e-06,
+      "loss": 1.1828,
+      "step": 2390
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 23.326084945768947,
+      "learning_rate": 8.573479314682028e-06,
+      "loss": 1.1916,
+      "step": 2395
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 12.416316769308711,
+      "learning_rate": 8.565265558156101e-06,
+      "loss": 1.196,
+      "step": 2400
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 35.72772435636289,
+      "learning_rate": 8.557032181560352e-06,
+      "loss": 1.1694,
+      "step": 2405
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.619986907617468,
+      "learning_rate": 8.548779230204e-06,
+      "loss": 1.1651,
+      "step": 2410
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 32.761372418831634,
+      "learning_rate": 8.540506749503997e-06,
+      "loss": 1.1715,
+      "step": 2415
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 61.3055493387315,
+      "learning_rate": 8.532214784984763e-06,
+      "loss": 1.1499,
+      "step": 2420
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 59.330525489071,
+      "learning_rate": 8.52390338227794e-06,
+      "loss": 1.1566,
+      "step": 2425
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 9.388547993206364,
+      "learning_rate": 8.515572587122143e-06,
+      "loss": 1.1916,
+      "step": 2430
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 19.591377355238617,
+      "learning_rate": 8.507222445362703e-06,
+      "loss": 1.172,
+      "step": 2435
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 12.587047679753722,
+      "learning_rate": 8.498853002951414e-06,
+      "loss": 1.1557,
+      "step": 2440
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 29.959799715131666,
+      "learning_rate": 8.490464305946296e-06,
+      "loss": 1.2,
+      "step": 2445
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 14.975482685862213,
+      "learning_rate": 8.482056400511316e-06,
+      "loss": 1.1559,
+      "step": 2450
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 15.40648000652942,
+      "learning_rate": 8.473629332916153e-06,
+      "loss": 1.2285,
+      "step": 2455
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 30.382411953297986,
+      "learning_rate": 8.465183149535939e-06,
+      "loss": 1.1756,
+      "step": 2460
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 64.89866462433648,
+      "learning_rate": 8.456717896851e-06,
+      "loss": 1.1531,
+      "step": 2465
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 128.6662406240173,
+      "learning_rate": 8.448233621446603e-06,
+      "loss": 1.1618,
+      "step": 2470
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 83.25907938124904,
+      "learning_rate": 8.4397303700127e-06,
+      "loss": 1.2396,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 89.08170449864967,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 1.1994,
+      "step": 2480
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.1953928470611572,
+      "eval_runtime": 166.8488,
+      "eval_samples_per_second": 4.831,
+      "eval_steps_per_second": 0.605,
+      "step": 2480
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 34.162842183997434,
+      "learning_rate": 8.422667126338057e-06,
+      "loss": 1.0506,
+      "step": 2485
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 196.56600927450077,
+      "learning_rate": 8.414107227998329e-06,
+      "loss": 1.0977,
+      "step": 2490
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 72.35802116188461,
+      "learning_rate": 8.405528541430595e-06,
+      "loss": 1.0621,
+      "step": 2495
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 98.7370524213735,
+      "learning_rate": 8.396931113844363e-06,
+      "loss": 1.0701,
+      "step": 2500
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 63.55008798348666,
+      "learning_rate": 8.388314992552271e-06,
+      "loss": 1.1214,
+      "step": 2505
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 108.19756404008838,
+      "learning_rate": 8.379680224969837e-06,
+      "loss": 1.0671,
+      "step": 2510
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 34.652316236237034,
+      "learning_rate": 8.371026858615189e-06,
+      "loss": 1.041,
+      "step": 2515
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 23.740621030455095,
+      "learning_rate": 8.362354941108803e-06,
+      "loss": 1.005,
+      "step": 2520
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 12.921809635549243,
+      "learning_rate": 8.353664520173247e-06,
+      "loss": 1.0433,
+      "step": 2525
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 77.45000375098243,
+      "learning_rate": 8.344955643632917e-06,
+      "loss": 1.0412,
+      "step": 2530
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 111.3918351387325,
+      "learning_rate": 8.336228359413769e-06,
+      "loss": 1.0256,
+      "step": 2535
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 124.63288622011407,
+      "learning_rate": 8.327482715543061e-06,
+      "loss": 1.0492,
+      "step": 2540
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 77.63619439839776,
+      "learning_rate": 8.318718760149085e-06,
+      "loss": 1.0673,
+      "step": 2545
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 27.6475228241581,
+      "learning_rate": 8.309936541460904e-06,
+      "loss": 1.0622,
+      "step": 2550
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 22.43391172428548,
+      "learning_rate": 8.301136107808086e-06,
+      "loss": 1.0277,
+      "step": 2555
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 61.184395234688864,
+      "learning_rate": 8.292317507620438e-06,
+      "loss": 0.995,
+      "step": 2560
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 96.99317001807856,
+      "learning_rate": 8.28348078942774e-06,
+      "loss": 1.0,
+      "step": 2565
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 45.44063987471112,
+      "learning_rate": 8.274626001859476e-06,
+      "loss": 1.0494,
+      "step": 2570
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 42.38019507055514,
+      "learning_rate": 8.265753193644569e-06,
+      "loss": 1.0128,
+      "step": 2575
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 39.65736839998161,
+      "learning_rate": 8.256862413611113e-06,
+      "loss": 1.0105,
+      "step": 2580
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 11.12223640835322,
+      "learning_rate": 8.247953710686099e-06,
+      "loss": 1.0436,
+      "step": 2585
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 47.36603114667974,
+      "learning_rate": 8.239027133895154e-06,
+      "loss": 1.0014,
+      "step": 2590
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 13.858763075141361,
+      "learning_rate": 8.230082732362264e-06,
+      "loss": 1.0101,
+      "step": 2595
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 9.543214017262684,
+      "learning_rate": 8.221120555309511e-06,
+      "loss": 1.0363,
+      "step": 2600
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 29.76464741844112,
+      "learning_rate": 8.21214065205679e-06,
+      "loss": 0.9995,
+      "step": 2605
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 14.846741343057671,
+      "learning_rate": 8.203143072021555e-06,
+      "loss": 0.9912,
+      "step": 2610
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 19.294904748253614,
+      "learning_rate": 8.19412786471853e-06,
+      "loss": 1.0365,
+      "step": 2615
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 11.008570732808002,
+      "learning_rate": 8.185095079759445e-06,
+      "loss": 1.0262,
+      "step": 2620
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 24.980255505082216,
+      "learning_rate": 8.176044766852766e-06,
+      "loss": 0.9791,
+      "step": 2625
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 18.28633718898899,
+      "learning_rate": 8.166976975803412e-06,
+      "loss": 1.0545,
+      "step": 2630
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 14.487130364956482,
+      "learning_rate": 8.157891756512488e-06,
+      "loss": 0.9748,
+      "step": 2635
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 10.272864231392472,
+      "learning_rate": 8.148789158977012e-06,
+      "loss": 0.9948,
+      "step": 2640
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 142.10905301854768,
+      "learning_rate": 8.139669233289634e-06,
+      "loss": 1.093,
+      "step": 2645
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 90.87270254055976,
+      "learning_rate": 8.13053202963836e-06,
+      "loss": 0.981,
+      "step": 2650
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 22.874478900522877,
+      "learning_rate": 8.121377598306283e-06,
+      "loss": 1.0392,
+      "step": 2655
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 21.139536470601552,
+      "learning_rate": 8.112205989671302e-06,
+      "loss": 1.035,
+      "step": 2660
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 16.017310065218556,
+      "learning_rate": 8.10301725420584e-06,
+      "loss": 1.0404,
+      "step": 2665
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 11.28250680363922,
+      "learning_rate": 8.093811442476573e-06,
+      "loss": 1.046,
+      "step": 2670
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 28.46863219914913,
+      "learning_rate": 8.084588605144155e-06,
+      "loss": 1.0135,
+      "step": 2675
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 69.43922503494132,
+      "learning_rate": 8.075348792962924e-06,
+      "loss": 1.0162,
+      "step": 2680
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 81.98036315710506,
+      "learning_rate": 8.066092056780644e-06,
+      "loss": 1.1267,
+      "step": 2685
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 28.157761337346354,
+      "learning_rate": 8.056818447538202e-06,
+      "loss": 1.0422,
+      "step": 2690
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 39.816962768629224,
+      "learning_rate": 8.047528016269348e-06,
+      "loss": 1.0681,
+      "step": 2695
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 35.9771780464358,
+      "learning_rate": 8.038220814100403e-06,
+      "loss": 1.0562,
+      "step": 2700
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 13.149616552660088,
+      "learning_rate": 8.02889689224998e-06,
+      "loss": 1.0051,
+      "step": 2705
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 12.704218651246096,
+      "learning_rate": 8.019556302028701e-06,
+      "loss": 1.0397,
+      "step": 2710
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 38.87306738675905,
+      "learning_rate": 8.010199094838915e-06,
+      "loss": 1.0211,
+      "step": 2715
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 19.279397420889897,
+      "learning_rate": 8.000825322174424e-06,
+      "loss": 1.0242,
+      "step": 2720
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 29.94854241006509,
+      "learning_rate": 7.99143503562018e-06,
+      "loss": 0.9867,
+      "step": 2725
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 11.596270671590164,
+      "learning_rate": 7.982028286852025e-06,
+      "loss": 1.0154,
+      "step": 2730
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 81.92677337752968,
+      "learning_rate": 7.972605127636384e-06,
+      "loss": 1.0576,
+      "step": 2735
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 15.569772707933453,
+      "learning_rate": 7.963165609829994e-06,
+      "loss": 1.0648,
+      "step": 2740
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 16.57278954689521,
+      "learning_rate": 7.953709785379622e-06,
+      "loss": 1.0513,
+      "step": 2745
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 17.41146317058611,
+      "learning_rate": 7.944237706321761e-06,
+      "loss": 1.0234,
+      "step": 2750
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 16.588173657125395,
+      "learning_rate": 7.934749424782362e-06,
+      "loss": 1.0348,
+      "step": 2755
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 16.088632338078156,
+      "learning_rate": 7.925244992976538e-06,
+      "loss": 1.03,
+      "step": 2760
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 10.415152584990919,
+      "learning_rate": 7.915724463208283e-06,
+      "loss": 1.0379,
+      "step": 2765
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 8.805924813348835,
+      "learning_rate": 7.906187887870173e-06,
+      "loss": 0.9873,
+      "step": 2770
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 9.792215691660319,
+      "learning_rate": 7.896635319443088e-06,
+      "loss": 1.0352,
+      "step": 2775
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 17.88307238664097,
+      "learning_rate": 7.88706681049592e-06,
+      "loss": 1.016,
+      "step": 2780
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 21.95408549846677,
+      "learning_rate": 7.877482413685287e-06,
+      "loss": 1.0313,
+      "step": 2785
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 9.683449995182132,
+      "learning_rate": 7.86788218175523e-06,
+      "loss": 1.0362,
+      "step": 2790
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 29.754639376164715,
+      "learning_rate": 7.858266167536945e-06,
+      "loss": 1.0072,
+      "step": 2795
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 16.850525931963297,
+      "learning_rate": 7.848634423948468e-06,
+      "loss": 1.0047,
+      "step": 2800
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 10.50735126486331,
+      "learning_rate": 7.838987003994404e-06,
+      "loss": 1.0223,
+      "step": 2805
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 14.931595107488798,
+      "learning_rate": 7.829323960765624e-06,
+      "loss": 1.0186,
+      "step": 2810
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 17.625068008673907,
+      "learning_rate": 7.819645347438973e-06,
+      "loss": 1.0527,
+      "step": 2815
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 10.059602672275243,
+      "learning_rate": 7.809951217276986e-06,
+      "loss": 1.0057,
+      "step": 2820
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 18.984752466743547,
+      "learning_rate": 7.800241623627583e-06,
+      "loss": 1.0239,
+      "step": 2825
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 9.695343896153208,
+      "learning_rate": 7.790516619923783e-06,
+      "loss": 1.0114,
+      "step": 2830
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 24.686354093950506,
+      "learning_rate": 7.780776259683411e-06,
+      "loss": 1.0449,
+      "step": 2835
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 43.546553146832146,
+      "learning_rate": 7.7710205965088e-06,
+      "loss": 1.0132,
+      "step": 2840
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 41.07607126812267,
+      "learning_rate": 7.761249684086494e-06,
+      "loss": 1.0388,
+      "step": 2845
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 71.18001851205797,
+      "learning_rate": 7.751463576186957e-06,
+      "loss": 1.0002,
+      "step": 2850
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 29.767681166160923,
+      "learning_rate": 7.741662326664277e-06,
+      "loss": 1.0243,
+      "step": 2855
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 49.613935895354246,
+      "learning_rate": 7.731845989455866e-06,
+      "loss": 1.0219,
+      "step": 2860
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 36.85736897806983,
+      "learning_rate": 7.722014618582166e-06,
+      "loss": 1.0334,
+      "step": 2865
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 38.85812381135648,
+      "learning_rate": 7.712168268146351e-06,
+      "loss": 1.0234,
+      "step": 2870
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 22.14152765307725,
+      "learning_rate": 7.70230699233403e-06,
+      "loss": 1.0313,
+      "step": 2875
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 54.80053198000988,
+      "learning_rate": 7.692430845412946e-06,
+      "loss": 1.0366,
+      "step": 2880
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 33.25896849653014,
+      "learning_rate": 7.682539881732682e-06,
+      "loss": 1.0491,
+      "step": 2885
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 61.010403744942316,
+      "learning_rate": 7.672634155724356e-06,
+      "loss": 1.0377,
+      "step": 2890
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 33.90715314788795,
+      "learning_rate": 7.662713721900332e-06,
+      "loss": 1.0147,
+      "step": 2895
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 11.615749348338921,
+      "learning_rate": 7.652778634853903e-06,
+      "loss": 1.0664,
+      "step": 2900
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 20.05877428776725,
+      "learning_rate": 7.642828949259006e-06,
+      "loss": 1.0458,
+      "step": 2905
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 34.50900627296251,
+      "learning_rate": 7.632864719869916e-06,
+      "loss": 1.0418,
+      "step": 2910
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 11.262994092173233,
+      "learning_rate": 7.6228860015209446e-06,
+      "loss": 1.0116,
+      "step": 2915
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 8.89312506665108,
+      "learning_rate": 7.612892849126132e-06,
+      "loss": 1.0067,
+      "step": 2920
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 37.86739512723313,
+      "learning_rate": 7.6028853176789576e-06,
+      "loss": 1.0429,
+      "step": 2925
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 17.534257965092515,
+      "learning_rate": 7.592863462252025e-06,
+      "loss": 1.0485,
+      "step": 2930
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 73.39467612348243,
+      "learning_rate": 7.582827337996769e-06,
+      "loss": 1.0391,
+      "step": 2935
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 20.37485245176674,
+      "learning_rate": 7.572777000143145e-06,
+      "loss": 0.9954,
+      "step": 2940
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 13.892625297288765,
+      "learning_rate": 7.562712503999327e-06,
+      "loss": 0.986,
+      "step": 2945
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 13.32267801986881,
+      "learning_rate": 7.552633904951405e-06,
+      "loss": 1.0534,
+      "step": 2950
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 20.018475105671392,
+      "learning_rate": 7.542541258463076e-06,
+      "loss": 1.0105,
+      "step": 2955
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 14.012742126815223,
+      "learning_rate": 7.532434620075349e-06,
+      "loss": 1.0426,
+      "step": 2960
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 22.867551791549793,
+      "learning_rate": 7.522314045406223e-06,
+      "loss": 1.0492,
+      "step": 2965
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 52.25996674758155,
+      "learning_rate": 7.5121795901503955e-06,
+      "loss": 1.0276,
+      "step": 2970
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 16.033916975805806,
+      "learning_rate": 7.50203131007895e-06,
+      "loss": 1.0094,
+      "step": 2975
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 30.70850892382383,
+      "learning_rate": 7.491869261039046e-06,
+      "loss": 0.9891,
+      "step": 2980
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 16.955498632329324,
+      "learning_rate": 7.481693498953621e-06,
+      "loss": 1.0155,
+      "step": 2985
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 28.332497486212482,
+      "learning_rate": 7.47150407982107e-06,
+      "loss": 1.0121,
+      "step": 2990
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 14.689911880407493,
+      "learning_rate": 7.461301059714952e-06,
+      "loss": 1.0512,
+      "step": 2995
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 52.81123599794421,
+      "learning_rate": 7.451084494783668e-06,
+      "loss": 0.9742,
+      "step": 3000
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 9.788808844659156,
+      "learning_rate": 7.44085444125016e-06,
+      "loss": 0.9937,
+      "step": 3005
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 45.96420512116728,
+      "learning_rate": 7.430610955411597e-06,
+      "loss": 0.9991,
+      "step": 3010
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 9.647180632597168,
+      "learning_rate": 7.420354093639074e-06,
+      "loss": 1.0072,
+      "step": 3015
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 21.013082658386075,
+      "learning_rate": 7.410083912377286e-06,
+      "loss": 1.0577,
+      "step": 3020
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 26.694976684324654,
+      "learning_rate": 7.399800468144235e-06,
+      "loss": 1.066,
+      "step": 3025
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 9.134573687381785,
+      "learning_rate": 7.389503817530905e-06,
+      "loss": 1.081,
+      "step": 3030
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 15.324015002761898,
+      "learning_rate": 7.379194017200959e-06,
+      "loss": 1.0386,
+      "step": 3035
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 23.616248520746904,
+      "learning_rate": 7.368871123890425e-06,
+      "loss": 1.0137,
+      "step": 3040
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 8.492882191328237,
+      "learning_rate": 7.35853519440738e-06,
+      "loss": 1.0343,
+      "step": 3045
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 11.451780948360682,
+      "learning_rate": 7.3481862856316455e-06,
+      "loss": 1.0275,
+      "step": 3050
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 14.42888350543538,
+      "learning_rate": 7.337824454514463e-06,
+      "loss": 1.0401,
+      "step": 3055
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 52.4027830988403,
+      "learning_rate": 7.327449758078194e-06,
+      "loss": 1.0224,
+      "step": 3060
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 18.220099875809783,
+      "learning_rate": 7.317062253415993e-06,
+      "loss": 1.029,
+      "step": 3065
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 182.64642392561194,
+      "learning_rate": 7.306661997691504e-06,
+      "loss": 1.0237,
+      "step": 3070
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 111.91648918440882,
+      "learning_rate": 7.296249048138543e-06,
+      "loss": 1.0727,
+      "step": 3075
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 65.49347355019428,
+      "learning_rate": 7.285823462060776e-06,
+      "loss": 1.0913,
+      "step": 3080
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 60.72831282404967,
+      "learning_rate": 7.275385296831414e-06,
+      "loss": 1.0728,
+      "step": 3085
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 125.13963886874934,
+      "learning_rate": 7.264934609892889e-06,
+      "loss": 1.0591,
+      "step": 3090
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 50.206153033985714,
+      "learning_rate": 7.254471458756544e-06,
+      "loss": 1.021,
+      "step": 3095
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 40.26635860112713,
+      "learning_rate": 7.243995901002312e-06,
+      "loss": 0.9967,
+      "step": 3100
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 31.944898456289675,
+      "learning_rate": 7.233507994278404e-06,
+      "loss": 1.0126,
+      "step": 3105
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 16.10051422613548,
+      "learning_rate": 7.2230077963009856e-06,
+      "loss": 1.0382,
+      "step": 3110
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 17.676061585956354,
+      "learning_rate": 7.2124953648538665e-06,
+      "loss": 1.0039,
+      "step": 3115
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 76.04395600658047,
+      "learning_rate": 7.201970757788172e-06,
+      "loss": 1.0114,
+      "step": 3120
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 10.923038710009777,
+      "learning_rate": 7.1914340330220385e-06,
+      "loss": 1.0299,
+      "step": 3125
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 24.208429948176793,
+      "learning_rate": 7.180885248540283e-06,
+      "loss": 1.026,
+      "step": 3130
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 37.104049346106265,
+      "learning_rate": 7.170324462394091e-06,
+      "loss": 1.0138,
+      "step": 3135
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 35.71354882719401,
+      "learning_rate": 7.1597517327006905e-06,
+      "loss": 1.0246,
+      "step": 3140
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 11.47162580883148,
+      "learning_rate": 7.149167117643043e-06,
+      "loss": 0.9976,
+      "step": 3145
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 15.175307981675164,
+      "learning_rate": 7.138570675469511e-06,
+      "loss": 1.0409,
+      "step": 3150
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 22.40837102976381,
+      "learning_rate": 7.127962464493543e-06,
+      "loss": 0.9756,
+      "step": 3155
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 16.517731913854483,
+      "learning_rate": 7.117342543093358e-06,
+      "loss": 1.0055,
+      "step": 3160
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 9.584701704481905,
+      "learning_rate": 7.10671096971161e-06,
+      "loss": 1.0084,
+      "step": 3165
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 50.1736681110015,
+      "learning_rate": 7.096067802855082e-06,
+      "loss": 0.9805,
+      "step": 3170
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 10.20329405219765,
+      "learning_rate": 7.0854131010943586e-06,
+      "loss": 0.9828,
+      "step": 3175
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 42.24651081997016,
+      "learning_rate": 7.074746923063497e-06,
+      "loss": 1.0023,
+      "step": 3180
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 72.2170024343786,
+      "learning_rate": 7.06406932745971e-06,
+      "loss": 1.0317,
+      "step": 3185
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 39.67245253902288,
+      "learning_rate": 7.0533803730430495e-06,
+      "loss": 0.9937,
+      "step": 3190
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 72.3620422568804,
+      "learning_rate": 7.042680118636066e-06,
+      "loss": 1.0393,
+      "step": 3195
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 67.10224702167315,
+      "learning_rate": 7.031968623123503e-06,
+      "loss": 1.0017,
+      "step": 3200
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 15.88506575459156,
+      "learning_rate": 7.0212459454519614e-06,
+      "loss": 1.0125,
+      "step": 3205
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 18.664088363268984,
+      "learning_rate": 7.01051214462958e-06,
+      "loss": 1.0318,
+      "step": 3210
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 37.083225109615974,
+      "learning_rate": 6.999767279725711e-06,
+      "loss": 1.0245,
+      "step": 3215
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 20.31803824773978,
+      "learning_rate": 6.989011409870591e-06,
+      "loss": 0.9835,
+      "step": 3220
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 14.605843803080104,
+      "learning_rate": 6.978244594255019e-06,
+      "loss": 1.0263,
+      "step": 3225
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 8.71390640581468,
+      "learning_rate": 6.96746689213003e-06,
+      "loss": 1.0009,
+      "step": 3230
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 11.280622775265291,
+      "learning_rate": 6.95667836280657e-06,
+      "loss": 1.027,
+      "step": 3235
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 9.252222719080446,
+      "learning_rate": 6.945879065655164e-06,
+      "loss": 0.9908,
+      "step": 3240
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 30.5015739583395,
+      "learning_rate": 6.9350690601055995e-06,
+      "loss": 1.0306,
+      "step": 3245
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 11.963380785590312,
+      "learning_rate": 6.924248405646588e-06,
+      "loss": 0.9768,
+      "step": 3250
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 12.793911762164084,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 1.0326,
+      "step": 3255
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 9.828796230585477,
+      "learning_rate": 6.902575388247773e-06,
+      "loss": 1.0224,
+      "step": 3260
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 20.425729253702407,
+      "learning_rate": 6.891723144577093e-06,
+      "loss": 0.9763,
+      "step": 3265
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 19.442762154988007,
+      "learning_rate": 6.8808604905345685e-06,
+      "loss": 0.986,
+      "step": 3270
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 16.537840749312895,
+      "learning_rate": 6.869987485898644e-06,
+      "loss": 0.9929,
+      "step": 3275
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 18.164900342901575,
+      "learning_rate": 6.859104190504725e-06,
+      "loss": 1.0538,
+      "step": 3280
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 7.92089405703362,
+      "learning_rate": 6.848210664244846e-06,
+      "loss": 1.0272,
+      "step": 3285
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 43.157682545104855,
+      "learning_rate": 6.8373069670673486e-06,
+      "loss": 1.0391,
+      "step": 3290
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 12.315156613440807,
+      "learning_rate": 6.826393158976541e-06,
+      "loss": 1.0284,
+      "step": 3295
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 32.77579146126162,
+      "learning_rate": 6.815469300032374e-06,
+      "loss": 1.0371,
+      "step": 3300
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 44.7313527771387,
+      "learning_rate": 6.804535450350111e-06,
+      "loss": 1.0243,
+      "step": 3305
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 41.62704905117679,
+      "learning_rate": 6.793591670099995e-06,
+      "loss": 1.0057,
+      "step": 3310
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 12.268675277610308,
+      "learning_rate": 6.782638019506917e-06,
+      "loss": 0.9895,
+      "step": 3315
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 11.55767059946778,
+      "learning_rate": 6.771674558850088e-06,
+      "loss": 0.9805,
+      "step": 3320
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 8.727069994209206,
+      "learning_rate": 6.760701348462704e-06,
+      "loss": 1.0078,
+      "step": 3325
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 14.428014718419858,
+      "learning_rate": 6.749718448731611e-06,
+      "loss": 1.0481,
+      "step": 3330
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 32.5138121103151,
+      "learning_rate": 6.738725920096983e-06,
+      "loss": 0.9777,
+      "step": 3335
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 14.722147479682125,
+      "learning_rate": 6.7277238230519796e-06,
+      "loss": 1.0086,
+      "step": 3340
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 9.042706993469462,
+      "learning_rate": 6.716712218142414e-06,
+      "loss": 0.9802,
+      "step": 3345
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 14.781446121189488,
+      "learning_rate": 6.705691165966426e-06,
+      "loss": 1.0274,
+      "step": 3350
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 10.501428779050258,
+      "learning_rate": 6.694660727174147e-06,
+      "loss": 1.017,
+      "step": 3355
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 11.60204172587971,
+      "learning_rate": 6.6836209624673575e-06,
+      "loss": 1.0602,
+      "step": 3360
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 33.9331745173826,
+      "learning_rate": 6.6725719325991645e-06,
+      "loss": 1.0273,
+      "step": 3365
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 46.1327279223193,
+      "learning_rate": 6.661513698373661e-06,
+      "loss": 0.9787,
+      "step": 3370
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 40.39384753474129,
+      "learning_rate": 6.650446320645589e-06,
+      "loss": 1.0146,
+      "step": 3375
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 26.75895144989692,
+      "learning_rate": 6.6393698603200165e-06,
+      "loss": 0.9889,
+      "step": 3380
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 26.55779833587494,
+      "learning_rate": 6.628284378351986e-06,
+      "loss": 0.9953,
+      "step": 3385
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 17.57260411550591,
+      "learning_rate": 6.617189935746191e-06,
+      "loss": 1.0001,
+      "step": 3390
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 8.946018733146417,
+      "learning_rate": 6.606086593556636e-06,
+      "loss": 1.0251,
+      "step": 3395
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 23.37502565948009,
+      "learning_rate": 6.5949744128863026e-06,
+      "loss": 1.0246,
+      "step": 3400
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 37.34677089943191,
+      "learning_rate": 6.583853454886806e-06,
+      "loss": 1.0113,
+      "step": 3405
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 11.832867291526817,
+      "learning_rate": 6.572723780758069e-06,
+      "loss": 1.0067,
+      "step": 3410
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 42.31374848868545,
+      "learning_rate": 6.561585451747981e-06,
+      "loss": 1.033,
+      "step": 3415
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 15.07282714472137,
+      "learning_rate": 6.5504385291520554e-06,
+      "loss": 1.0155,
+      "step": 3420
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 32.668565361751206,
+      "learning_rate": 6.539283074313102e-06,
+      "loss": 1.0162,
+      "step": 3425
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 15.621424247001702,
+      "learning_rate": 6.528119148620883e-06,
+      "loss": 0.9917,
+      "step": 3430
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 59.93545393665107,
+      "learning_rate": 6.5169468135117745e-06,
+      "loss": 1.0222,
+      "step": 3435
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 52.6336157587358,
+      "learning_rate": 6.5057661304684314e-06,
+      "loss": 1.0158,
+      "step": 3440
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 62.51808798926133,
+      "learning_rate": 6.4945771610194505e-06,
+      "loss": 1.0143,
+      "step": 3445
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 52.45112003357092,
+      "learning_rate": 6.483379966739029e-06,
+      "loss": 0.9871,
+      "step": 3450
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 18.809313384654185,
+      "learning_rate": 6.472174609246626e-06,
+      "loss": 1.0078,
+      "step": 3455
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 63.907403724635465,
+      "learning_rate": 6.46096115020662e-06,
+      "loss": 1.0359,
+      "step": 3460
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 35.947489893637105,
+      "learning_rate": 6.449739651327979e-06,
+      "loss": 0.9788,
+      "step": 3465
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 9.283745969766212,
+      "learning_rate": 6.438510174363913e-06,
+      "loss": 1.0097,
+      "step": 3470
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 13.05462947670369,
+      "learning_rate": 6.427272781111534e-06,
+      "loss": 0.994,
+      "step": 3475
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 37.75632382596889,
+      "learning_rate": 6.41602753341152e-06,
+      "loss": 1.0221,
+      "step": 3480
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 26.683152997518274,
+      "learning_rate": 6.404774493147777e-06,
+      "loss": 1.0158,
+      "step": 3485
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 17.097367427386654,
+      "learning_rate": 6.393513722247089e-06,
+      "loss": 1.0,
+      "step": 3490
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 16.229825022601784,
+      "learning_rate": 6.38224528267878e-06,
+      "loss": 0.9858,
+      "step": 3495
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 39.78386897358033,
+      "learning_rate": 6.370969236454386e-06,
+      "loss": 1.0153,
+      "step": 3500
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 14.913446960568821,
+      "learning_rate": 6.359685645627294e-06,
+      "loss": 1.0335,
+      "step": 3505
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 41.148861233824114,
+      "learning_rate": 6.348394572292411e-06,
+      "loss": 1.0291,
+      "step": 3510
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 28.4468675762153,
+      "learning_rate": 6.337096078585825e-06,
+      "loss": 1.0056,
+      "step": 3515
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 14.14930386957616,
+      "learning_rate": 6.32579022668446e-06,
+      "loss": 1.0125,
+      "step": 3520
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 10.466233425152792,
+      "learning_rate": 6.314477078805724e-06,
+      "loss": 1.0182,
+      "step": 3525
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 15.895922380493161,
+      "learning_rate": 6.303156697207187e-06,
+      "loss": 0.9797,
+      "step": 3530
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 8.830828441980106,
+      "learning_rate": 6.291829144186219e-06,
+      "loss": 1.005,
+      "step": 3535
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 10.299260057122133,
+      "learning_rate": 6.2804944820796596e-06,
+      "loss": 1.0452,
+      "step": 3540
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 8.868215083199106,
+      "learning_rate": 6.269152773263467e-06,
+      "loss": 0.9859,
+      "step": 3545
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 11.289857957709584,
+      "learning_rate": 6.2578040801523795e-06,
+      "loss": 0.9823,
+      "step": 3550
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 38.43130747175126,
+      "learning_rate": 6.246448465199576e-06,
+      "loss": 1.0305,
+      "step": 3555
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 29.964044462196668,
+      "learning_rate": 6.235085990896317e-06,
+      "loss": 0.9457,
+      "step": 3560
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 15.691622516866989,
+      "learning_rate": 6.2237167197716195e-06,
+      "loss": 0.9822,
+      "step": 3565
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 14.810723686957973,
+      "learning_rate": 6.2123407143919e-06,
+      "loss": 1.0236,
+      "step": 3570
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 25.650290850908604,
+      "learning_rate": 6.200958037360636e-06,
+      "loss": 1.0135,
+      "step": 3575
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 10.475355579445026,
+      "learning_rate": 6.189568751318018e-06,
+      "loss": 1.0078,
+      "step": 3580
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 27.56934465571091,
+      "learning_rate": 6.178172918940608e-06,
+      "loss": 1.0175,
+      "step": 3585
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 11.627943861121338,
+      "learning_rate": 6.166770602940995e-06,
+      "loss": 0.9988,
+      "step": 3590
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 35.75718836112594,
+      "learning_rate": 6.155361866067445e-06,
+      "loss": 1.0348,
+      "step": 3595
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 11.272611161416208,
+      "learning_rate": 6.143946771103561e-06,
+      "loss": 0.9882,
+      "step": 3600
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 13.138140147782178,
+      "learning_rate": 6.13252538086793e-06,
+      "loss": 1.0237,
+      "step": 3605
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 50.65571533833276,
+      "learning_rate": 6.121097758213793e-06,
+      "loss": 1.0189,
+      "step": 3610
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 19.379659921982515,
+      "learning_rate": 6.109663966028681e-06,
+      "loss": 0.9949,
+      "step": 3615
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 12.40755438116134,
+      "learning_rate": 6.098224067234076e-06,
+      "loss": 1.0043,
+      "step": 3620
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 8.855751692951607,
+      "learning_rate": 6.086778124785069e-06,
+      "loss": 1.0178,
+      "step": 3625
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 25.5987304467699,
+      "learning_rate": 6.075326201670011e-06,
+      "loss": 1.0417,
+      "step": 3630
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 13.26639277937442,
+      "learning_rate": 6.063868360910159e-06,
+      "loss": 1.0462,
+      "step": 3635
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 16.337013641809335,
+      "learning_rate": 6.052404665559342e-06,
+      "loss": 1.0216,
+      "step": 3640
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 28.95391095055042,
+      "learning_rate": 6.040935178703604e-06,
+      "loss": 0.9843,
+      "step": 3645
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 22.163048156045544,
+      "learning_rate": 6.029459963460861e-06,
+      "loss": 1.0003,
+      "step": 3650
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 11.29890727872941,
+      "learning_rate": 6.017979082980554e-06,
+      "loss": 1.0431,
+      "step": 3655
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 17.381953688004256,
+      "learning_rate": 6.006492600443301e-06,
+      "loss": 1.0317,
+      "step": 3660
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 13.480009070615699,
+      "learning_rate": 5.995000579060546e-06,
+      "loss": 0.9892,
+      "step": 3665
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 31.544902517748252,
+      "learning_rate": 5.983503082074213e-06,
+      "loss": 1.0224,
+      "step": 3670
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 22.53715405623453,
+      "learning_rate": 5.972000172756367e-06,
+      "loss": 1.0617,
+      "step": 3675
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 85.97422780944547,
+      "learning_rate": 5.960491914408846e-06,
+      "loss": 1.0165,
+      "step": 3680
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 38.65337696076255,
+      "learning_rate": 5.9489783703629335e-06,
+      "loss": 1.0712,
+      "step": 3685
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 56.61853413357405,
+      "learning_rate": 5.937459603978997e-06,
+      "loss": 0.9899,
+      "step": 3690
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 45.84469836512608,
+      "learning_rate": 5.925935678646144e-06,
+      "loss": 1.0007,
+      "step": 3695
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 17.14463555179784,
+      "learning_rate": 5.914406657781871e-06,
+      "loss": 0.9999,
+      "step": 3700
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 65.91198999951662,
+      "learning_rate": 5.9028726048317194e-06,
+      "loss": 1.0136,
+      "step": 3705
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 37.51578425266073,
+      "learning_rate": 5.891333583268917e-06,
+      "loss": 1.0183,
+      "step": 3710
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 20.016122161200215,
+      "learning_rate": 5.879789656594038e-06,
+      "loss": 1.0077,
+      "step": 3715
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 26.792681395042763,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 1.0235,
+      "step": 3720
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 25.544726776408446,
+      "learning_rate": 5.856687342044969e-06,
+      "loss": 0.9809,
+      "step": 3725
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 10.579345539325837,
+      "learning_rate": 5.845129081305493e-06,
+      "loss": 0.9958,
+      "step": 3730
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 8.9271726806215,
+      "learning_rate": 5.833566169722674e-06,
+      "loss": 0.9935,
+      "step": 3735
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 19.194138662694023,
+      "learning_rate": 5.821998670928556e-06,
+      "loss": 1.0227,
+      "step": 3740
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 8.788343862868038,
+      "learning_rate": 5.810426648580425e-06,
+      "loss": 1.0262,
+      "step": 3745
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 10.022574392079015,
+      "learning_rate": 5.798850166360461e-06,
+      "loss": 1.0009,
+      "step": 3750
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 40.635736884626596,
+      "learning_rate": 5.787269287975392e-06,
+      "loss": 1.0038,
+      "step": 3755
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 16.6432090652126,
+      "learning_rate": 5.775684077156133e-06,
+      "loss": 0.9908,
+      "step": 3760
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 17.435151435826274,
+      "learning_rate": 5.76409459765744e-06,
+      "loss": 0.9885,
+      "step": 3765
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 28.284253338560653,
+      "learning_rate": 5.752500913257568e-06,
+      "loss": 0.9986,
+      "step": 3770
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 9.773215148597064,
+      "learning_rate": 5.740903087757902e-06,
+      "loss": 1.0141,
+      "step": 3775
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 8.472902745997331,
+      "learning_rate": 5.729301184982622e-06,
+      "loss": 0.9482,
+      "step": 3780
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 16.365524571343943,
+      "learning_rate": 5.717695268778345e-06,
+      "loss": 1.0115,
+      "step": 3785
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 19.33560398015478,
+      "learning_rate": 5.706085403013774e-06,
+      "loss": 0.9768,
+      "step": 3790
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 9.631578863850336,
+      "learning_rate": 5.694471651579346e-06,
+      "loss": 0.9786,
+      "step": 3795
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 73.45622305825805,
+      "learning_rate": 5.682854078386882e-06,
+      "loss": 1.0234,
+      "step": 3800
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 8.403903307690225,
+      "learning_rate": 5.671232747369236e-06,
+      "loss": 0.9814,
+      "step": 3805
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 19.492909280614263,
+      "learning_rate": 5.659607722479938e-06,
+      "loss": 0.9736,
+      "step": 3810
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 9.716992670899291,
+      "learning_rate": 5.647979067692851e-06,
+      "loss": 1.0212,
+      "step": 3815
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 11.841418133532775,
+      "learning_rate": 5.636346847001807e-06,
+      "loss": 0.9801,
+      "step": 3820
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 19.55357083027445,
+      "learning_rate": 5.624711124420269e-06,
+      "loss": 0.9906,
+      "step": 3825
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 20.81765110380046,
+      "learning_rate": 5.613071963980966e-06,
+      "loss": 0.9913,
+      "step": 3830
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 21.975390663868655,
+      "learning_rate": 5.6014294297355475e-06,
+      "loss": 1.0213,
+      "step": 3835
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 21.490026352526726,
+      "learning_rate": 5.5897835857542315e-06,
+      "loss": 0.9829,
+      "step": 3840
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 11.93270966131888,
+      "learning_rate": 5.5781344961254445e-06,
+      "loss": 0.9981,
+      "step": 3845
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 18.77263276363035,
+      "learning_rate": 5.566482224955481e-06,
+      "loss": 0.9691,
+      "step": 3850
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 27.510040332158336,
+      "learning_rate": 5.554826836368139e-06,
+      "loss": 0.9842,
+      "step": 3855
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 16.73208516633144,
+      "learning_rate": 5.543168394504374e-06,
+      "loss": 1.0064,
+      "step": 3860
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 27.517124532057508,
+      "learning_rate": 5.531506963521943e-06,
+      "loss": 0.9898,
+      "step": 3865
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 32.73440555010284,
+      "learning_rate": 5.519842607595055e-06,
+      "loss": 0.9808,
+      "step": 3870
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 21.201827334876604,
+      "learning_rate": 5.50817539091401e-06,
+      "loss": 1.0106,
+      "step": 3875
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 12.541690640702148,
+      "learning_rate": 5.496505377684858e-06,
+      "loss": 1.0062,
+      "step": 3880
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 17.856302468130952,
+      "learning_rate": 5.484832632129031e-06,
+      "loss": 1.0019,
+      "step": 3885
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 18.091246932044104,
+      "learning_rate": 5.473157218483006e-06,
+      "loss": 0.9969,
+      "step": 3890
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 10.01125236254292,
+      "learning_rate": 5.461479200997935e-06,
+      "loss": 1.0063,
+      "step": 3895
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 8.515528512214727,
+      "learning_rate": 5.449798643939305e-06,
+      "loss": 1.0179,
+      "step": 3900
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 12.231393309911725,
+      "learning_rate": 5.438115611586575e-06,
+      "loss": 0.997,
+      "step": 3905
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 14.743216803723987,
+      "learning_rate": 5.426430168232826e-06,
+      "loss": 0.9601,
+      "step": 3910
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 10.99337082685551,
+      "learning_rate": 5.414742378184409e-06,
+      "loss": 0.9793,
+      "step": 3915
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 9.782287149993518,
+      "learning_rate": 5.4030523057605865e-06,
+      "loss": 0.9938,
+      "step": 3920
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 28.655669847166596,
+      "learning_rate": 5.391360015293184e-06,
+      "loss": 1.0049,
+      "step": 3925
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 12.814431075178419,
+      "learning_rate": 5.379665571126232e-06,
+      "loss": 1.0185,
+      "step": 3930
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 7.608007304163836,
+      "learning_rate": 5.367969037615614e-06,
+      "loss": 1.0029,
+      "step": 3935
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 26.896324489688606,
+      "learning_rate": 5.356270479128707e-06,
+      "loss": 0.9805,
+      "step": 3940
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 12.451204995416903,
+      "learning_rate": 5.344569960044039e-06,
+      "loss": 1.0125,
+      "step": 3945
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 38.254860430992125,
+      "learning_rate": 5.33286754475092e-06,
+      "loss": 0.9856,
+      "step": 3950
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 9.594675533888049,
+      "learning_rate": 5.3211632976491e-06,
+      "loss": 0.9971,
+      "step": 3955
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 9.738083084918026,
+      "learning_rate": 5.30945728314841e-06,
+      "loss": 1.0169,
+      "step": 3960
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 8.941756704243613,
+      "learning_rate": 5.2977495656684016e-06,
+      "loss": 0.9862,
+      "step": 3965
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 45.983958205105644,
+      "learning_rate": 5.286040209638005e-06,
+      "loss": 1.0016,
+      "step": 3970
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 23.110146191421364,
+      "learning_rate": 5.2743292794951616e-06,
+      "loss": 1.0055,
+      "step": 3975
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 61.923466632850655,
+      "learning_rate": 5.26261683968648e-06,
+      "loss": 1.0225,
+      "step": 3980
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 51.186851210172534,
+      "learning_rate": 5.250902954666875e-06,
+      "loss": 0.9664,
+      "step": 3985
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 22.662294122712844,
+      "learning_rate": 5.2391876888992135e-06,
+      "loss": 1.0228,
+      "step": 3990
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 47.00814260437848,
+      "learning_rate": 5.22747110685396e-06,
+      "loss": 1.0307,
+      "step": 3995
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 53.67263273425259,
+      "learning_rate": 5.215753273008828e-06,
+      "loss": 1.0431,
+      "step": 4000
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 44.485911782334775,
+      "learning_rate": 5.204034251848413e-06,
+      "loss": 0.9961,
+      "step": 4005
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 36.525690266517735,
+      "learning_rate": 5.192314107863848e-06,
+      "loss": 1.0039,
+      "step": 4010
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 83.21181541526505,
+      "learning_rate": 5.180592905552448e-06,
+      "loss": 1.0047,
+      "step": 4015
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 11.821627823465693,
+      "learning_rate": 5.168870709417342e-06,
+      "loss": 0.9928,
+      "step": 4020
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 25.861983277409138,
+      "learning_rate": 5.157147583967139e-06,
+      "loss": 1.0085,
+      "step": 4025
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 20.019058269985877,
+      "learning_rate": 5.145423593715558e-06,
+      "loss": 1.0193,
+      "step": 4030
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 9.382589868562954,
+      "learning_rate": 5.133698803181075e-06,
+      "loss": 0.9822,
+      "step": 4035
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 10.841985815459434,
+      "learning_rate": 5.1219732768865744e-06,
+      "loss": 1.0203,
+      "step": 4040
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 9.307384900366749,
+      "learning_rate": 5.110247079358986e-06,
+      "loss": 0.9782,
+      "step": 4045
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 14.16352552118682,
+      "learning_rate": 5.098520275128934e-06,
+      "loss": 0.9868,
+      "step": 4050
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 28.35614435399522,
+      "learning_rate": 5.086792928730383e-06,
+      "loss": 1.0282,
+      "step": 4055
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 40.589123536982136,
+      "learning_rate": 5.075065104700281e-06,
+      "loss": 1.0282,
+      "step": 4060
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 17.62987826434959,
+      "learning_rate": 5.063336867578201e-06,
+      "loss": 1.0308,
+      "step": 4065
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 11.628754535382729,
+      "learning_rate": 5.051608281905994e-06,
+      "loss": 1.0079,
+      "step": 4070
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 8.307010805846195,
+      "learning_rate": 5.039879412227428e-06,
+      "loss": 0.9974,
+      "step": 4075
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 15.442567328580099,
+      "learning_rate": 5.0281503230878304e-06,
+      "loss": 0.9965,
+      "step": 4080
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 36.02159707646093,
+      "learning_rate": 5.01642107903374e-06,
+      "loss": 0.9726,
+      "step": 4085
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 41.60960110203915,
+      "learning_rate": 5.004691744612547e-06,
+      "loss": 1.0071,
+      "step": 4090
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 11.786403980665508,
+      "learning_rate": 4.992962384372142e-06,
+      "loss": 1.0157,
+      "step": 4095
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 39.1557632356577,
+      "learning_rate": 4.981233062860548e-06,
+      "loss": 1.0113,
+      "step": 4100
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 35.19409024432306,
+      "learning_rate": 4.9695038446255885e-06,
+      "loss": 1.0053,
+      "step": 4105
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 16.09755722418478,
+      "learning_rate": 4.957774794214509e-06,
+      "loss": 0.9309,
+      "step": 4110
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 10.215737034545892,
+      "learning_rate": 4.946045976173632e-06,
+      "loss": 1.0102,
+      "step": 4115
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 34.637930518716615,
+      "learning_rate": 4.934317455048005e-06,
+      "loss": 0.9648,
+      "step": 4120
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 44.9478395784996,
+      "learning_rate": 4.92258929538104e-06,
+      "loss": 0.9693,
+      "step": 4125
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 21.530014467497892,
+      "learning_rate": 4.910861561714161e-06,
+      "loss": 0.9829,
+      "step": 4130
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 10.71925681650583,
+      "learning_rate": 4.8991343185864445e-06,
+      "loss": 1.0127,
+      "step": 4135
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 37.06813122448028,
+      "learning_rate": 4.887407630534271e-06,
+      "loss": 0.9995,
+      "step": 4140
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 16.693036216888956,
+      "learning_rate": 4.875681562090962e-06,
+      "loss": 0.9857,
+      "step": 4145
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 9.004349402834453,
+      "learning_rate": 4.8639561777864355e-06,
+      "loss": 1.0546,
+      "step": 4150
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 17.319489609030885,
+      "learning_rate": 4.852231542146839e-06,
+      "loss": 1.006,
+      "step": 4155
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 43.36487067882959,
+      "learning_rate": 4.840507719694202e-06,
+      "loss": 0.9722,
+      "step": 4160
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 25.17142402779811,
+      "learning_rate": 4.8287847749460785e-06,
+      "loss": 0.9824,
+      "step": 4165
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 10.048713766624477,
+      "learning_rate": 4.817062772415196e-06,
+      "loss": 0.99,
+      "step": 4170
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 53.25267612505607,
+      "learning_rate": 4.805341776609092e-06,
+      "loss": 0.9958,
+      "step": 4175
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 35.56079618212311,
+      "learning_rate": 4.793621852029767e-06,
+      "loss": 1.0025,
+      "step": 4180
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 30.656384954066542,
+      "learning_rate": 4.781903063173321e-06,
+      "loss": 0.9731,
+      "step": 4185
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 53.310108890643264,
+      "learning_rate": 4.770185474529611e-06,
+      "loss": 0.9948,
+      "step": 4190
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 34.19572860470345,
+      "learning_rate": 4.758469150581887e-06,
+      "loss": 1.0009,
+      "step": 4195
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 19.496116909194217,
+      "learning_rate": 4.746754155806437e-06,
+      "loss": 0.9602,
+      "step": 4200
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 50.517897273832304,
+      "learning_rate": 4.7350405546722375e-06,
+      "loss": 0.9706,
+      "step": 4205
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 32.68098812951939,
+      "learning_rate": 4.72332841164059e-06,
+      "loss": 1.0161,
+      "step": 4210
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 32.602984226324224,
+      "learning_rate": 4.711617791164781e-06,
+      "loss": 0.9576,
+      "step": 4215
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 20.895596814038374,
+      "learning_rate": 4.699908757689709e-06,
+      "loss": 1.0067,
+      "step": 4220
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 12.007303358830825,
+      "learning_rate": 4.688201375651545e-06,
+      "loss": 0.9745,
+      "step": 4225
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 35.577170489212946,
+      "learning_rate": 4.6764957094773704e-06,
+      "loss": 1.0228,
+      "step": 4230
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 15.319645442064518,
+      "learning_rate": 4.6647918235848215e-06,
+      "loss": 1.0076,
+      "step": 4235
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 10.150742823806933,
+      "learning_rate": 4.6530897823817425e-06,
+      "loss": 0.9596,
+      "step": 4240
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 11.674813818686681,
+      "learning_rate": 4.6413896502658215e-06,
+      "loss": 1.0058,
+      "step": 4245
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 22.414842802144854,
+      "learning_rate": 4.6296914916242434e-06,
+      "loss": 1.0004,
+      "step": 4250
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 43.263891122381494,
+      "learning_rate": 4.617995370833329e-06,
+      "loss": 0.9935,
+      "step": 4255
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 11.76080822704277,
+      "learning_rate": 4.606301352258192e-06,
+      "loss": 0.9844,
+      "step": 4260
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 14.936200948426407,
+      "learning_rate": 4.5946095002523686e-06,
+      "loss": 0.9491,
+      "step": 4265
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 27.500241741161222,
+      "learning_rate": 4.582919879157479e-06,
+      "loss": 0.9909,
+      "step": 4270
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 16.484820797461804,
+      "learning_rate": 4.571232553302861e-06,
+      "loss": 1.0062,
+      "step": 4275
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 35.307120079758334,
+      "learning_rate": 4.559547587005227e-06,
+      "loss": 0.9906,
+      "step": 4280
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 48.09110177500657,
+      "learning_rate": 4.5478650445683015e-06,
+      "loss": 0.9922,
+      "step": 4285
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 25.58111167712098,
+      "learning_rate": 4.5361849902824675e-06,
+      "loss": 0.9505,
+      "step": 4290
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 13.706479123182746,
+      "learning_rate": 4.524507488424422e-06,
+      "loss": 0.9928,
+      "step": 4295
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 22.41767364911632,
+      "learning_rate": 4.512832603256807e-06,
+      "loss": 0.9679,
+      "step": 4300
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 52.507167267681126,
+      "learning_rate": 4.501160399027874e-06,
+      "loss": 0.9624,
+      "step": 4305
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 11.492683655916354,
+      "learning_rate": 4.489490939971117e-06,
+      "loss": 1.0085,
+      "step": 4310
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 26.062024557348025,
+      "learning_rate": 4.477824290304919e-06,
+      "loss": 0.9863,
+      "step": 4315
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 56.71755781997285,
+      "learning_rate": 4.466160514232206e-06,
+      "loss": 0.9589,
+      "step": 4320
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 15.968281495466721,
+      "learning_rate": 4.454499675940094e-06,
+      "loss": 0.9786,
+      "step": 4325
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 21.054689397186266,
+      "learning_rate": 4.442841839599525e-06,
+      "loss": 0.9968,
+      "step": 4330
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 22.34304013469586,
+      "learning_rate": 4.431187069364927e-06,
+      "loss": 0.9549,
+      "step": 4335
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 15.401013625746206,
+      "learning_rate": 4.4195354293738484e-06,
+      "loss": 0.9974,
+      "step": 4340
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 19.5883959837736,
+      "learning_rate": 4.40788698374662e-06,
+      "loss": 1.0144,
+      "step": 4345
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 10.314509785502024,
+      "learning_rate": 4.396241796585987e-06,
+      "loss": 0.968,
+      "step": 4350
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 21.63930495269253,
+      "learning_rate": 4.384599931976764e-06,
+      "loss": 1.0094,
+      "step": 4355
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 9.359569416879197,
+      "learning_rate": 4.3729614539854815e-06,
+      "loss": 0.9796,
+      "step": 4360
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 9.39151630144433,
+      "learning_rate": 4.361326426660035e-06,
+      "loss": 1.0042,
+      "step": 4365
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 20.479042824969,
+      "learning_rate": 4.3496949140293285e-06,
+      "loss": 0.9858,
+      "step": 4370
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 8.655115252867901,
+      "learning_rate": 4.3380669801029265e-06,
+      "loss": 1.006,
+      "step": 4375
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 11.94304831924867,
+      "learning_rate": 4.326442688870697e-06,
+      "loss": 1.0027,
+      "step": 4380
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 15.338465500775095,
+      "learning_rate": 4.314822104302459e-06,
+      "loss": 0.9889,
+      "step": 4385
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 9.329249928306815,
+      "learning_rate": 4.3032052903476436e-06,
+      "loss": 1.0262,
+      "step": 4390
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 7.86378064663073,
+      "learning_rate": 4.29159231093492e-06,
+      "loss": 0.9829,
+      "step": 4395
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 11.169777398812325,
+      "learning_rate": 4.279983229971863e-06,
+      "loss": 0.994,
+      "step": 4400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 15.517339484404777,
+      "learning_rate": 4.268378111344593e-06,
+      "loss": 0.9591,
+      "step": 4405
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.011668196020228,
+      "learning_rate": 4.256777018917418e-06,
+      "loss": 0.9712,
+      "step": 4410
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.144580946340453,
+      "learning_rate": 4.245180016532501e-06,
+      "loss": 0.9592,
+      "step": 4415
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.637705163099548,
+      "learning_rate": 4.233587168009489e-06,
+      "loss": 0.9618,
+      "step": 4420
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 14.189533731874635,
+      "learning_rate": 4.221998537145169e-06,
+      "loss": 0.9846,
+      "step": 4425
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 17.033419382931804,
+      "learning_rate": 4.210414187713124e-06,
+      "loss": 1.0003,
+      "step": 4430
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 15.76135941495352,
+      "learning_rate": 4.1988341834633714e-06,
+      "loss": 1.0146,
+      "step": 4435
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 34.88708649681162,
+      "learning_rate": 4.187258588122019e-06,
+      "loss": 0.9936,
+      "step": 4440
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 29.174557212173497,
+      "learning_rate": 4.175687465390908e-06,
+      "loss": 0.9885,
+      "step": 4445
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 33.751248168264986,
+      "learning_rate": 4.164120878947268e-06,
+      "loss": 0.9782,
+      "step": 4450
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 20.1719560139535,
+      "learning_rate": 4.152558892443369e-06,
+      "loss": 0.975,
+      "step": 4455
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 9.785430509469538,
+      "learning_rate": 4.141001569506161e-06,
+      "loss": 0.9952,
+      "step": 4460
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 10.47350835689242,
+      "learning_rate": 4.1294489737369325e-06,
+      "loss": 0.9771,
+      "step": 4465
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 12.558380433822101,
+      "learning_rate": 4.11790116871096e-06,
+      "loss": 0.9981,
+      "step": 4470
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 32.049499620776984,
+      "learning_rate": 4.106358217977148e-06,
+      "loss": 1.0045,
+      "step": 4475
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 13.266466932459126,
+      "learning_rate": 4.094820185057701e-06,
+      "loss": 1.0025,
+      "step": 4480
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 18.662166973808628,
+      "learning_rate": 4.083287133447747e-06,
+      "loss": 0.9752,
+      "step": 4485
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 22.150116609199884,
+      "learning_rate": 4.0717591266150085e-06,
+      "loss": 0.9519,
+      "step": 4490
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 15.38212411747071,
+      "learning_rate": 4.060236227999441e-06,
+      "loss": 1.0076,
+      "step": 4495
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 12.315249686944153,
+      "learning_rate": 4.048718501012895e-06,
+      "loss": 0.9671,
+      "step": 4500
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 9.679261033729183,
+      "learning_rate": 4.037206009038758e-06,
+      "loss": 0.973,
+      "step": 4505
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 8.676674731769216,
+      "learning_rate": 4.025698815431607e-06,
+      "loss": 1.0108,
+      "step": 4510
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 14.507021196439057,
+      "learning_rate": 4.014196983516862e-06,
+      "loss": 0.9515,
+      "step": 4515
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 21.37357830376474,
+      "learning_rate": 4.002700576590441e-06,
+      "loss": 0.9944,
+      "step": 4520
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 32.602575451830994,
+      "learning_rate": 3.991209657918402e-06,
+      "loss": 0.9476,
+      "step": 4525
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 16.66752171528357,
+      "learning_rate": 3.979724290736603e-06,
+      "loss": 0.9851,
+      "step": 4530
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 9.74844029535486,
+      "learning_rate": 3.968244538250354e-06,
+      "loss": 0.9715,
+      "step": 4535
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 58.2432376431901,
+      "learning_rate": 3.956770463634059e-06,
+      "loss": 0.9397,
+      "step": 4540
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 30.35617745293608,
+      "learning_rate": 3.945302130030888e-06,
+      "loss": 0.9598,
+      "step": 4545
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 49.80326384646548,
+      "learning_rate": 3.933839600552407e-06,
+      "loss": 0.9974,
+      "step": 4550
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 46.15846010522385,
+      "learning_rate": 3.922382938278245e-06,
+      "loss": 0.9818,
+      "step": 4555
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 10.749329727288327,
+      "learning_rate": 3.910932206255742e-06,
+      "loss": 0.9787,
+      "step": 4560
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 15.949999913542994,
+      "learning_rate": 3.899487467499609e-06,
+      "loss": 0.9771,
+      "step": 4565
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 9.986861991661971,
+      "learning_rate": 3.8880487849915645e-06,
+      "loss": 0.9646,
+      "step": 4570
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 10.073554565547601,
+      "learning_rate": 3.8766162216800105e-06,
+      "loss": 0.9505,
+      "step": 4575
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 18.078489086458575,
+      "learning_rate": 3.865189840479665e-06,
+      "loss": 0.9675,
+      "step": 4580
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 8.984138132698236,
+      "learning_rate": 3.853769704271229e-06,
+      "loss": 0.9442,
+      "step": 4585
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 15.681176939934998,
+      "learning_rate": 3.842355875901038e-06,
+      "loss": 0.9776,
+      "step": 4590
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 11.775138108771696,
+      "learning_rate": 3.830948418180713e-06,
+      "loss": 0.9819,
+      "step": 4595
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 12.31354222453272,
+      "learning_rate": 3.819547393886816e-06,
+      "loss": 0.9553,
+      "step": 4600
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 28.40748231930582,
+      "learning_rate": 3.8081528657605045e-06,
+      "loss": 0.9948,
+      "step": 4605
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 51.323862996897546,
+      "learning_rate": 3.796764896507192e-06,
+      "loss": 0.9652,
+      "step": 4610
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 17.640625496427386,
+      "learning_rate": 3.7853835487961913e-06,
+      "loss": 0.9808,
+      "step": 4615
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 8.625985883550005,
+      "learning_rate": 3.77400888526038e-06,
+      "loss": 0.9642,
+      "step": 4620
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 22.59805259795483,
+      "learning_rate": 3.7626409684958477e-06,
+      "loss": 0.9831,
+      "step": 4625
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 22.91412923609426,
+      "learning_rate": 3.7512798610615624e-06,
+      "loss": 0.999,
+      "step": 4630
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 11.000489369889706,
+      "learning_rate": 3.7399256254790133e-06,
+      "loss": 0.9808,
+      "step": 4635
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 13.829730046627517,
+      "learning_rate": 3.7285783242318773e-06,
+      "loss": 0.9669,
+      "step": 4640
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 9.99894628910545,
+      "learning_rate": 3.717238019765669e-06,
+      "loss": 0.9439,
+      "step": 4645
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 19.530225291338915,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 0.9649,
+      "step": 4650
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 10.19043012847544,
+      "learning_rate": 3.694578650765226e-06,
+      "loss": 0.9844,
+      "step": 4655
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 40.66541834435717,
+      "learning_rate": 3.683259710928129e-06,
+      "loss": 1.0084,
+      "step": 4660
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 21.429955386516756,
+      "learning_rate": 3.671948017265542e-06,
+      "loss": 0.983,
+      "step": 4665
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 9.571041387180902,
+      "learning_rate": 3.660643632027029e-06,
+      "loss": 0.9722,
+      "step": 4670
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 11.135011856218295,
+      "learning_rate": 3.6493466174219316e-06,
+      "loss": 0.9612,
+      "step": 4675
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 10.19220828586926,
+      "learning_rate": 3.6380570356190346e-06,
+      "loss": 0.9647,
+      "step": 4680
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 22.17611186385138,
+      "learning_rate": 3.6267749487462126e-06,
+      "loss": 0.9628,
+      "step": 4685
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 36.264622952792806,
+      "learning_rate": 3.6155004188900995e-06,
+      "loss": 0.9809,
+      "step": 4690
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 35.18861018766551,
+      "learning_rate": 3.6042335080957373e-06,
+      "loss": 0.9707,
+      "step": 4695
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 14.677993191535572,
+      "learning_rate": 3.592974278366248e-06,
+      "loss": 0.9735,
+      "step": 4700
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 17.101889527887568,
+      "learning_rate": 3.5817227916624748e-06,
+      "loss": 0.9553,
+      "step": 4705
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 9.873539362898917,
+      "learning_rate": 3.5704791099026558e-06,
+      "loss": 0.9739,
+      "step": 4710
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 7.548761304889987,
+      "learning_rate": 3.559243294962073e-06,
+      "loss": 0.9481,
+      "step": 4715
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 8.589846921664401,
+      "learning_rate": 3.548015408672723e-06,
+      "loss": 0.9652,
+      "step": 4720
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 11.508984080584233,
+      "learning_rate": 3.536795512822964e-06,
+      "loss": 0.9609,
+      "step": 4725
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 41.251796525753996,
+      "learning_rate": 3.5255836691571845e-06,
+      "loss": 0.9721,
+      "step": 4730
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 12.14028985277054,
+      "learning_rate": 3.514379939375457e-06,
+      "loss": 0.9862,
+      "step": 4735
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 9.546709810008233,
+      "learning_rate": 3.5031843851332105e-06,
+      "loss": 1.0029,
+      "step": 4740
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 10.700204738107006,
+      "learning_rate": 3.4919970680408767e-06,
+      "loss": 0.9948,
+      "step": 4745
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 29.891681249576166,
+      "learning_rate": 3.480818049663559e-06,
+      "loss": 0.9465,
+      "step": 4750
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 13.155476354862126,
+      "learning_rate": 3.4696473915206918e-06,
+      "loss": 0.9889,
+      "step": 4755
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 38.59253438385496,
+      "learning_rate": 3.4584851550857007e-06,
+      "loss": 0.9702,
+      "step": 4760
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 20.13290393233704,
+      "learning_rate": 3.447331401785671e-06,
+      "loss": 0.97,
+      "step": 4765
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 35.75418799455426,
+      "learning_rate": 3.4361861930009966e-06,
+      "loss": 0.9829,
+      "step": 4770
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 8.78033743901871,
+      "learning_rate": 3.425049590065057e-06,
+      "loss": 0.955,
+      "step": 4775
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 10.875319981025177,
+      "learning_rate": 3.4139216542638665e-06,
+      "loss": 0.9241,
+      "step": 4780
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 22.696375641845567,
+      "learning_rate": 3.40280244683575e-06,
+      "loss": 1.0508,
+      "step": 4785
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 13.094779283891201,
+      "learning_rate": 3.3916920289709936e-06,
+      "loss": 0.9386,
+      "step": 4790
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 21.3991739229068,
+      "learning_rate": 3.380590461811516e-06,
+      "loss": 0.9774,
+      "step": 4795
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 15.275029457940944,
+      "learning_rate": 3.3694978064505258e-06,
+      "loss": 0.9803,
+      "step": 4800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 12.774697743450275,
+      "learning_rate": 3.3584141239321953e-06,
+      "loss": 0.9519,
+      "step": 4805
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 32.04675443101915,
+      "learning_rate": 3.3473394752513133e-06,
+      "loss": 0.9934,
+      "step": 4810
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 16.358611962826902,
+      "learning_rate": 3.3362739213529572e-06,
+      "loss": 0.9608,
+      "step": 4815
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 23.546881423236798,
+      "learning_rate": 3.3252175231321516e-06,
+      "loss": 0.9922,
+      "step": 4820
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 23.60367253955802,
+      "learning_rate": 3.3141703414335367e-06,
+      "loss": 0.9853,
+      "step": 4825
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 8.028245975063125,
+      "learning_rate": 3.3031324370510397e-06,
+      "loss": 0.994,
+      "step": 4830
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 14.899072710524097,
+      "learning_rate": 3.2921038707275243e-06,
+      "loss": 0.9794,
+      "step": 4835
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 36.27594479358489,
+      "learning_rate": 3.2810847031544703e-06,
+      "loss": 0.9896,
+      "step": 4840
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 17.067401742985304,
+      "learning_rate": 3.270074994971635e-06,
+      "loss": 0.9934,
+      "step": 4845
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 19.201614639786936,
+      "learning_rate": 3.2590748067667204e-06,
+      "loss": 0.9928,
+      "step": 4850
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 41.253187808383565,
+      "learning_rate": 3.2480841990750376e-06,
+      "loss": 0.9563,
+      "step": 4855
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 39.02357449517357,
+      "learning_rate": 3.2371032323791757e-06,
+      "loss": 0.9668,
+      "step": 4860
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 16.207285531426546,
+      "learning_rate": 3.2261319671086654e-06,
+      "loss": 1.0069,
+      "step": 4865
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 10.154528182144999,
+      "learning_rate": 3.2151704636396537e-06,
+      "loss": 0.9707,
+      "step": 4870
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 15.725470434474367,
+      "learning_rate": 3.204218782294565e-06,
+      "loss": 0.9695,
+      "step": 4875
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 29.466778077374947,
+      "learning_rate": 3.193276983341773e-06,
+      "loss": 0.9684,
+      "step": 4880
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 41.44285638457885,
+      "learning_rate": 3.1823451269952654e-06,
+      "loss": 0.9966,
+      "step": 4885
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 54.440789198493434,
+      "learning_rate": 3.1714232734143125e-06,
+      "loss": 0.9733,
+      "step": 4890
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 43.816361024248685,
+      "learning_rate": 3.160511482703147e-06,
+      "loss": 0.9882,
+      "step": 4895
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 36.74369170157289,
+      "learning_rate": 3.1496098149106147e-06,
+      "loss": 0.9828,
+      "step": 4900
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 35.628079479155026,
+      "learning_rate": 3.1387183300298574e-06,
+      "loss": 0.9717,
+      "step": 4905
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 25.078865420594674,
+      "learning_rate": 3.1278370879979813e-06,
+      "loss": 0.9456,
+      "step": 4910
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 31.887918641790915,
+      "learning_rate": 3.1169661486957215e-06,
+      "loss": 1.0266,
+      "step": 4915
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 16.833480681950846,
+      "learning_rate": 3.10610557194712e-06,
+      "loss": 0.9655,
+      "step": 4920
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 9.659776691132217,
+      "learning_rate": 3.0952554175191895e-06,
+      "loss": 0.9717,
+      "step": 4925
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 16.672819916651907,
+      "learning_rate": 3.084415745121586e-06,
+      "loss": 0.9804,
+      "step": 4930
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 33.195815487526886,
+      "learning_rate": 3.073586614406284e-06,
+      "loss": 0.9627,
+      "step": 4935
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 19.211257978880255,
+      "learning_rate": 3.0627680849672493e-06,
+      "loss": 0.9548,
+      "step": 4940
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 21.9290292175729,
+      "learning_rate": 3.0519602163400997e-06,
+      "loss": 0.9744,
+      "step": 4945
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 11.362805750104576,
+      "learning_rate": 3.0411630680017924e-06,
+      "loss": 0.9942,
+      "step": 4950
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 35.099045351919415,
+      "learning_rate": 3.0303766993702844e-06,
+      "loss": 0.9706,
+      "step": 4955
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 30.23998872907586,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 1.0039,
+      "step": 4960
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.0974189043045044,
+      "eval_runtime": 164.7878,
+      "eval_samples_per_second": 4.891,
+      "eval_steps_per_second": 0.613,
+      "step": 4960
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 25.60673447961065,
+      "learning_rate": 3.008836538602575e-06,
+      "loss": 0.7319,
+      "step": 4965
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 11.320426623405694,
+      "learning_rate": 2.9980828650043742e-06,
+      "loss": 0.7158,
+      "step": 4970
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 14.696590711778382,
+      "learning_rate": 2.987340208188328e-06,
+      "loss": 0.7169,
+      "step": 4975
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 34.649343494173586,
+      "learning_rate": 2.97660862727252e-06,
+      "loss": 0.7143,
+      "step": 4980
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 24.22309324830745,
+      "learning_rate": 2.9658881813140884e-06,
+      "loss": 0.7311,
+      "step": 4985
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 9.699910586219252,
+      "learning_rate": 2.9551789293088886e-06,
+      "loss": 0.7306,
+      "step": 4990
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 8.778692413295019,
+      "learning_rate": 2.9444809301911758e-06,
+      "loss": 0.7273,
+      "step": 4995
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 7.342909124898869,
+      "learning_rate": 2.9337942428332787e-06,
+      "loss": 0.7462,
+      "step": 5000
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 8.460649917741181,
+      "learning_rate": 2.923118926045281e-06,
+      "loss": 0.734,
+      "step": 5005
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 11.444046192891937,
+      "learning_rate": 2.912455038574686e-06,
+      "loss": 0.7115,
+      "step": 5010
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 7.955867192944722,
+      "learning_rate": 2.9018026391061026e-06,
+      "loss": 0.6799,
+      "step": 5015
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 15.626065706533977,
+      "learning_rate": 2.8911617862609226e-06,
+      "loss": 0.6959,
+      "step": 5020
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 18.70458625881847,
+      "learning_rate": 2.8805325385969916e-06,
+      "loss": 0.7006,
+      "step": 5025
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 12.144570172996847,
+      "learning_rate": 2.8699149546082934e-06,
+      "loss": 0.723,
+      "step": 5030
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 18.639101110384757,
+      "learning_rate": 2.8593090927246247e-06,
+      "loss": 0.7263,
+      "step": 5035
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 15.7002852063242,
+      "learning_rate": 2.848715011311271e-06,
+      "loss": 0.7472,
+      "step": 5040
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 10.133589092455146,
+      "learning_rate": 2.8381327686686917e-06,
+      "loss": 0.7427,
+      "step": 5045
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 14.143118839238095,
+      "learning_rate": 2.8275624230321953e-06,
+      "loss": 0.6868,
+      "step": 5050
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 16.033920564274712,
+      "learning_rate": 2.817004032571622e-06,
+      "loss": 0.702,
+      "step": 5055
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 9.021193561065502,
+      "learning_rate": 2.806457655391015e-06,
+      "loss": 0.6874,
+      "step": 5060
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 9.998979773050994,
+      "learning_rate": 2.795923349528311e-06,
+      "loss": 0.7213,
+      "step": 5065
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 9.917838960805637,
+      "learning_rate": 2.7854011729550225e-06,
+      "loss": 0.7072,
+      "step": 5070
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 27.358870585228118,
+      "learning_rate": 2.7748911835759017e-06,
+      "loss": 0.7245,
+      "step": 5075
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 17.840184673053667,
+      "learning_rate": 2.764393439228643e-06,
+      "loss": 0.718,
+      "step": 5080
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 11.190345510960688,
+      "learning_rate": 2.7539079976835447e-06,
+      "loss": 0.7137,
+      "step": 5085
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 8.13294279192623,
+      "learning_rate": 2.743434916643216e-06,
+      "loss": 0.7047,
+      "step": 5090
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 8.592635342931136,
+      "learning_rate": 2.73297425374223e-06,
+      "loss": 0.7278,
+      "step": 5095
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 15.171805707120754,
+      "learning_rate": 2.72252606654683e-06,
+      "loss": 0.7096,
+      "step": 5100
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 10.177855688362342,
+      "learning_rate": 2.712090412554604e-06,
+      "loss": 0.7144,
+      "step": 5105
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 15.062260827232874,
+      "learning_rate": 2.7016673491941594e-06,
+      "loss": 0.6819,
+      "step": 5110
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 23.600490651613107,
+      "learning_rate": 2.6912569338248317e-06,
+      "loss": 0.7262,
+      "step": 5115
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 14.497758759389454,
+      "learning_rate": 2.6808592237363364e-06,
+      "loss": 0.7625,
+      "step": 5120
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 8.7168967183322,
+      "learning_rate": 2.6704742761484814e-06,
+      "loss": 0.7077,
+      "step": 5125
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 18.657108516954423,
+      "learning_rate": 2.6601021482108374e-06,
+      "loss": 0.7244,
+      "step": 5130
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 22.58396850965322,
+      "learning_rate": 2.649742897002426e-06,
+      "loss": 0.7168,
+      "step": 5135
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 14.411269054170244,
+      "learning_rate": 2.6393965795314124e-06,
+      "loss": 0.6912,
+      "step": 5140
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 30.321795489118244,
+      "learning_rate": 2.629063252734775e-06,
+      "loss": 0.7177,
+      "step": 5145
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 26.558823064996645,
+      "learning_rate": 2.6187429734780145e-06,
+      "loss": 0.7193,
+      "step": 5150
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 15.349485553172165,
+      "learning_rate": 2.6084357985548192e-06,
+      "loss": 0.6807,
+      "step": 5155
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 12.787628647907104,
+      "learning_rate": 2.5981417846867753e-06,
+      "loss": 0.6963,
+      "step": 5160
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 7.1783644247257445,
+      "learning_rate": 2.5878609885230278e-06,
+      "loss": 0.7037,
+      "step": 5165
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 7.294931754599566,
+      "learning_rate": 2.577593466639994e-06,
+      "loss": 0.71,
+      "step": 5170
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 13.395255931323153,
+      "learning_rate": 2.567339275541037e-06,
+      "loss": 0.7007,
+      "step": 5175
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 36.56934428157861,
+      "learning_rate": 2.5570984716561597e-06,
+      "loss": 0.7209,
+      "step": 5180
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 7.579962432685944,
+      "learning_rate": 2.5468711113416944e-06,
+      "loss": 0.6964,
+      "step": 5185
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 22.91956890739734,
+      "learning_rate": 2.5366572508799884e-06,
+      "loss": 0.6755,
+      "step": 5190
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 14.180788527605213,
+      "learning_rate": 2.5264569464791023e-06,
+      "loss": 0.7073,
+      "step": 5195
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 8.74433804300584,
+      "learning_rate": 2.5162702542724924e-06,
+      "loss": 0.7038,
+      "step": 5200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 8.453744534656876,
+      "learning_rate": 2.506097230318709e-06,
+      "loss": 0.7291,
+      "step": 5205
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 11.561061370648575,
+      "learning_rate": 2.4959379306010845e-06,
+      "loss": 0.7155,
+      "step": 5210
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 13.085171124759647,
+      "learning_rate": 2.4857924110274185e-06,
+      "loss": 0.7291,
+      "step": 5215
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 7.580777868791533,
+      "learning_rate": 2.4756607274296844e-06,
+      "loss": 0.6896,
+      "step": 5220
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 8.124400652263825,
+      "learning_rate": 2.4655429355637122e-06,
+      "loss": 0.7194,
+      "step": 5225
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 9.209667691796083,
+      "learning_rate": 2.4554390911088834e-06,
+      "loss": 0.6867,
+      "step": 5230
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 9.235335051262112,
+      "learning_rate": 2.445349249667828e-06,
+      "loss": 0.6856,
+      "step": 5235
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 18.573953714651054,
+      "learning_rate": 2.4352734667661073e-06,
+      "loss": 0.6992,
+      "step": 5240
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 7.231459671524343,
+      "learning_rate": 2.425211797851931e-06,
+      "loss": 0.6882,
+      "step": 5245
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 7.173355505618337,
+      "learning_rate": 2.4151642982958246e-06,
+      "loss": 0.7009,
+      "step": 5250
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 10.748327760080251,
+      "learning_rate": 2.4051310233903467e-06,
+      "loss": 0.7158,
+      "step": 5255
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 7.368133746887829,
+      "learning_rate": 2.395112028349767e-06,
+      "loss": 0.7022,
+      "step": 5260
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 10.073393070220808,
+      "learning_rate": 2.3851073683097852e-06,
+      "loss": 0.7133,
+      "step": 5265
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 10.369530915498508,
+      "learning_rate": 2.3751170983272e-06,
+      "loss": 0.7186,
+      "step": 5270
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 7.487117503067416,
+      "learning_rate": 2.3651412733796267e-06,
+      "loss": 0.7053,
+      "step": 5275
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 14.468727653866845,
+      "learning_rate": 2.3551799483651894e-06,
+      "loss": 0.6895,
+      "step": 5280
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 7.9047271546778655,
+      "learning_rate": 2.3452331781022092e-06,
+      "loss": 0.7109,
+      "step": 5285
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 7.59176364109064,
+      "learning_rate": 2.335301017328924e-06,
+      "loss": 0.7365,
+      "step": 5290
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 13.290169739630931,
+      "learning_rate": 2.3253835207031607e-06,
+      "loss": 0.73,
+      "step": 5295
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 8.739040399686104,
+      "learning_rate": 2.315480742802054e-06,
+      "loss": 0.7072,
+      "step": 5300
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 24.51870180490624,
+      "learning_rate": 2.305592738121741e-06,
+      "loss": 0.7247,
+      "step": 5305
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 9.829239075741722,
+      "learning_rate": 2.2957195610770576e-06,
+      "loss": 0.7227,
+      "step": 5310
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 10.921773906431268,
+      "learning_rate": 2.2858612660012438e-06,
+      "loss": 0.7194,
+      "step": 5315
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 17.80895529612371,
+      "learning_rate": 2.2760179071456356e-06,
+      "loss": 0.7302,
+      "step": 5320
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 7.606101503962172,
+      "learning_rate": 2.2661895386793826e-06,
+      "loss": 0.6781,
+      "step": 5325
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 10.548213861789673,
+      "learning_rate": 2.256376214689134e-06,
+      "loss": 0.6998,
+      "step": 5330
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 14.751772651717873,
+      "learning_rate": 2.2465779891787514e-06,
+      "loss": 0.6899,
+      "step": 5335
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 7.683939964961461,
+      "learning_rate": 2.236794916069007e-06,
+      "loss": 0.7034,
+      "step": 5340
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 12.164444172892702,
+      "learning_rate": 2.2270270491972835e-06,
+      "loss": 0.7326,
+      "step": 5345
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 8.853758277588964,
+      "learning_rate": 2.2172744423172864e-06,
+      "loss": 0.7088,
+      "step": 5350
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 8.264379895297916,
+      "learning_rate": 2.2075371490987425e-06,
+      "loss": 0.7185,
+      "step": 5355
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 7.247789903804749,
+      "learning_rate": 2.1978152231271077e-06,
+      "loss": 0.6888,
+      "step": 5360
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 8.160749370126286,
+      "learning_rate": 2.1881087179032645e-06,
+      "loss": 0.6841,
+      "step": 5365
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 13.82247142057415,
+      "learning_rate": 2.1784176868432376e-06,
+      "loss": 0.7513,
+      "step": 5370
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 20.10218867227893,
+      "learning_rate": 2.1687421832778966e-06,
+      "loss": 0.7107,
+      "step": 5375
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 12.17553016304403,
+      "learning_rate": 2.159082260452658e-06,
+      "loss": 0.7049,
+      "step": 5380
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 7.439591359866398,
+      "learning_rate": 2.1494379715272e-06,
+      "loss": 0.7101,
+      "step": 5385
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 19.254748498294624,
+      "learning_rate": 2.139809369575159e-06,
+      "loss": 0.6986,
+      "step": 5390
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 9.689935355786089,
+      "learning_rate": 2.1301965075838497e-06,
+      "loss": 0.7289,
+      "step": 5395
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 7.650899934128293,
+      "learning_rate": 2.120599438453968e-06,
+      "loss": 0.6992,
+      "step": 5400
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 22.2918533549259,
+      "learning_rate": 2.1110182149992964e-06,
+      "loss": 0.7254,
+      "step": 5405
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 11.390936408485016,
+      "learning_rate": 2.1014528899464215e-06,
+      "loss": 0.7148,
+      "step": 5410
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 9.362105641436626,
+      "learning_rate": 2.09190351593443e-06,
+      "loss": 0.726,
+      "step": 5415
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 8.74619979223555,
+      "learning_rate": 2.0823701455146434e-06,
+      "loss": 0.7012,
+      "step": 5420
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 9.254192015124952,
+      "learning_rate": 2.0728528311502977e-06,
+      "loss": 0.6679,
+      "step": 5425
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 17.314780848984693,
+      "learning_rate": 2.0633516252162803e-06,
+      "loss": 0.7266,
+      "step": 5430
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 8.50296257020781,
+      "learning_rate": 2.0538665799988305e-06,
+      "loss": 0.7441,
+      "step": 5435
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 17.40546352872344,
+      "learning_rate": 2.044397747695247e-06,
+      "loss": 0.7274,
+      "step": 5440
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 14.204020920659485,
+      "learning_rate": 2.0349451804136195e-06,
+      "loss": 0.7302,
+      "step": 5445
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 12.657320357914022,
+      "learning_rate": 2.025508930172516e-06,
+      "loss": 0.6961,
+      "step": 5450
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 7.202625840549151,
+      "learning_rate": 2.0160890489007197e-06,
+      "loss": 0.6958,
+      "step": 5455
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 15.01779205289609,
+      "learning_rate": 2.0066855884369246e-06,
+      "loss": 0.7242,
+      "step": 5460
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 7.957219335339912,
+      "learning_rate": 1.997298600529471e-06,
+      "loss": 0.7034,
+      "step": 5465
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 7.494798736835006,
+      "learning_rate": 1.987928136836036e-06,
+      "loss": 0.7312,
+      "step": 5470
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 13.537277919470428,
+      "learning_rate": 1.978574248923371e-06,
+      "loss": 0.7016,
+      "step": 5475
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 9.410344187792107,
+      "learning_rate": 1.969236988267005e-06,
+      "loss": 0.701,
+      "step": 5480
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 20.58650190857061,
+      "learning_rate": 1.9599164062509666e-06,
+      "loss": 0.7182,
+      "step": 5485
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 16.671784857429135,
+      "learning_rate": 1.950612554167502e-06,
+      "loss": 0.6935,
+      "step": 5490
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 18.79962351839764,
+      "learning_rate": 1.9413254832167834e-06,
+      "loss": 0.684,
+      "step": 5495
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 16.856262957963185,
+      "learning_rate": 1.932055244506641e-06,
+      "loss": 0.7068,
+      "step": 5500
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 10.576929533911413,
+      "learning_rate": 1.9228018890522742e-06,
+      "loss": 0.7186,
+      "step": 5505
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 18.17968626510052,
+      "learning_rate": 1.913565467775971e-06,
+      "loss": 0.6849,
+      "step": 5510
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 28.724452600407343,
+      "learning_rate": 1.9043460315068308e-06,
+      "loss": 0.7055,
+      "step": 5515
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 15.318639965568057,
+      "learning_rate": 1.8951436309804766e-06,
+      "loss": 0.7168,
+      "step": 5520
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 23.419729430364512,
+      "learning_rate": 1.885958316838788e-06,
+      "loss": 0.7057,
+      "step": 5525
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 10.027843069519783,
+      "learning_rate": 1.8767901396296128e-06,
+      "loss": 0.686,
+      "step": 5530
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 20.117806117644886,
+      "learning_rate": 1.867639149806495e-06,
+      "loss": 0.7166,
+      "step": 5535
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 7.895514629559552,
+      "learning_rate": 1.8585053977283933e-06,
+      "loss": 0.6841,
+      "step": 5540
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 9.363146927768579,
+      "learning_rate": 1.8493889336593995e-06,
+      "loss": 0.718,
+      "step": 5545
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 7.8610033918578095,
+      "learning_rate": 1.8402898077684806e-06,
+      "loss": 0.7023,
+      "step": 5550
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 11.524388899422258,
+      "learning_rate": 1.831208070129175e-06,
+      "loss": 0.6782,
+      "step": 5555
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 10.335299630858202,
+      "learning_rate": 1.8221437707193424e-06,
+      "loss": 0.717,
+      "step": 5560
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 14.116987390941619,
+      "learning_rate": 1.8130969594208703e-06,
+      "loss": 0.7153,
+      "step": 5565
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 8.85283101812693,
+      "learning_rate": 1.8040676860194118e-06,
+      "loss": 0.6878,
+      "step": 5570
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 8.589942345683562,
+      "learning_rate": 1.7950560002041051e-06,
+      "loss": 0.707,
+      "step": 5575
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 8.431013513048278,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 0.7208,
+      "step": 5580
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 7.390020158191169,
+      "learning_rate": 1.7770855896043004e-06,
+      "loss": 0.6934,
+      "step": 5585
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 8.227435360273066,
+      "learning_rate": 1.7681269637130522e-06,
+      "loss": 0.6883,
+      "step": 5590
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 8.081270933436889,
+      "learning_rate": 1.7591861231939216e-06,
+      "loss": 0.7016,
+      "step": 5595
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 11.88112262855669,
+      "learning_rate": 1.7502631172493878e-06,
+      "loss": 0.7176,
+      "step": 5600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 8.428836470772556,
+      "learning_rate": 1.7413579949837873e-06,
+      "loss": 0.7202,
+      "step": 5605
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 7.101090657459622,
+      "learning_rate": 1.732470805403043e-06,
+      "loss": 0.6931,
+      "step": 5610
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 8.393720846249298,
+      "learning_rate": 1.723601597414385e-06,
+      "loss": 0.6873,
+      "step": 5615
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 6.777383863552303,
+      "learning_rate": 1.7147504198260995e-06,
+      "loss": 0.7116,
+      "step": 5620
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 21.731672527581406,
+      "learning_rate": 1.7059173213472385e-06,
+      "loss": 0.7139,
+      "step": 5625
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 7.774732801415944,
+      "learning_rate": 1.6971023505873706e-06,
+      "loss": 0.7015,
+      "step": 5630
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 8.076311745093234,
+      "learning_rate": 1.6883055560562977e-06,
+      "loss": 0.7029,
+      "step": 5635
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 8.206810445200425,
+      "learning_rate": 1.6795269861638041e-06,
+      "loss": 0.7013,
+      "step": 5640
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 10.097989200722806,
+      "learning_rate": 1.6707666892193791e-06,
+      "loss": 0.6847,
+      "step": 5645
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 7.376391242183933,
+      "learning_rate": 1.662024713431949e-06,
+      "loss": 0.6913,
+      "step": 5650
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 7.771559239769423,
+      "learning_rate": 1.6533011069096216e-06,
+      "loss": 0.6898,
+      "step": 5655
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 16.042798155906937,
+      "learning_rate": 1.6445959176594156e-06,
+      "loss": 0.7091,
+      "step": 5660
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 8.51539326328401,
+      "learning_rate": 1.6359091935869986e-06,
+      "loss": 0.71,
+      "step": 5665
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 6.997772689634818,
+      "learning_rate": 1.6272409824964169e-06,
+      "loss": 0.6849,
+      "step": 5670
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.791553414840887,
+      "learning_rate": 1.6185913320898439e-06,
+      "loss": 0.6864,
+      "step": 5675
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.94642792448608,
+      "learning_rate": 1.6099602899673083e-06,
+      "loss": 0.6741,
+      "step": 5680
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.850867578468772,
+      "learning_rate": 1.6013479036264358e-06,
+      "loss": 0.7058,
+      "step": 5685
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.327887090480733,
+      "learning_rate": 1.5927542204621888e-06,
+      "loss": 0.6891,
+      "step": 5690
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 12.291356019659574,
+      "learning_rate": 1.5841792877665995e-06,
+      "loss": 0.756,
+      "step": 5695
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 11.296251458222326,
+      "learning_rate": 1.5756231527285181e-06,
+      "loss": 0.7155,
+      "step": 5700
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 11.231039396100408,
+      "learning_rate": 1.567085862433348e-06,
+      "loss": 0.7159,
+      "step": 5705
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 9.288385826690003,
+      "learning_rate": 1.5585674638627884e-06,
+      "loss": 0.715,
+      "step": 5710
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 9.624034679264133,
+      "learning_rate": 1.5500680038945753e-06,
+      "loss": 0.7163,
+      "step": 5715
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 6.88924826118366,
+      "learning_rate": 1.5415875293022181e-06,
+      "loss": 0.6975,
+      "step": 5720
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 18.428049490277107,
+      "learning_rate": 1.5331260867547582e-06,
+      "loss": 0.7147,
+      "step": 5725
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 8.971913030351825,
+      "learning_rate": 1.5246837228164907e-06,
+      "loss": 0.7286,
+      "step": 5730
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 23.495834492054406,
+      "learning_rate": 1.5162604839467265e-06,
+      "loss": 0.704,
+      "step": 5735
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 16.848727747480694,
+      "learning_rate": 1.5078564164995225e-06,
+      "loss": 0.7271,
+      "step": 5740
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 12.230638138972909,
+      "learning_rate": 1.4994715667234366e-06,
+      "loss": 0.7055,
+      "step": 5745
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 7.230658248515549,
+      "learning_rate": 1.4911059807612743e-06,
+      "loss": 0.6628,
+      "step": 5750
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 8.127010195894611,
+      "learning_rate": 1.482759704649821e-06,
+      "loss": 0.7104,
+      "step": 5755
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 10.853401864935572,
+      "learning_rate": 1.4744327843196043e-06,
+      "loss": 0.6745,
+      "step": 5760
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 6.9379408550625365,
+      "learning_rate": 1.4661252655946284e-06,
+      "loss": 0.7204,
+      "step": 5765
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 10.07034132824172,
+      "learning_rate": 1.4578371941921383e-06,
+      "loss": 0.6996,
+      "step": 5770
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 14.024834806221602,
+      "learning_rate": 1.4495686157223476e-06,
+      "loss": 0.6822,
+      "step": 5775
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 7.013039491524748,
+      "learning_rate": 1.4413195756882037e-06,
+      "loss": 0.7091,
+      "step": 5780
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 14.144498174052684,
+      "learning_rate": 1.4330901194851321e-06,
+      "loss": 0.7001,
+      "step": 5785
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 9.228028072708394,
+      "learning_rate": 1.4248802924007804e-06,
+      "loss": 0.6877,
+      "step": 5790
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 9.103143640063841,
+      "learning_rate": 1.4166901396147852e-06,
+      "loss": 0.6929,
+      "step": 5795
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 8.436262396139144,
+      "learning_rate": 1.4085197061985022e-06,
+      "loss": 0.7415,
+      "step": 5800
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 8.71042792436878,
+      "learning_rate": 1.4003690371147755e-06,
+      "loss": 0.7129,
+      "step": 5805
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 7.7086665572226405,
+      "learning_rate": 1.392238177217683e-06,
+      "loss": 0.6941,
+      "step": 5810
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 8.015548875035755,
+      "learning_rate": 1.3841271712522886e-06,
+      "loss": 0.7162,
+      "step": 5815
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 8.876482645557113,
+      "learning_rate": 1.3760360638544012e-06,
+      "loss": 0.6753,
+      "step": 5820
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 12.829621165002179,
+      "learning_rate": 1.3679648995503191e-06,
+      "loss": 0.695,
+      "step": 5825
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 8.15856019740538,
+      "learning_rate": 1.3599137227565983e-06,
+      "loss": 0.6871,
+      "step": 5830
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 10.60356450695432,
+      "learning_rate": 1.3518825777797973e-06,
+      "loss": 0.671,
+      "step": 5835
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 7.500495803815079,
+      "learning_rate": 1.3438715088162403e-06,
+      "loss": 0.6792,
+      "step": 5840
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 8.955269506917622,
+      "learning_rate": 1.3358805599517655e-06,
+      "loss": 0.6819,
+      "step": 5845
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 8.035499527843996,
+      "learning_rate": 1.3279097751614928e-06,
+      "loss": 0.6943,
+      "step": 5850
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 6.826809104561465,
+      "learning_rate": 1.319959198309575e-06,
+      "loss": 0.6607,
+      "step": 5855
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 10.289508662371126,
+      "learning_rate": 1.3120288731489593e-06,
+      "loss": 0.6834,
+      "step": 5860
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 8.677865561549089,
+      "learning_rate": 1.304118843321146e-06,
+      "loss": 0.7068,
+      "step": 5865
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 14.048173371644863,
+      "learning_rate": 1.2962291523559428e-06,
+      "loss": 0.709,
+      "step": 5870
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 7.591034809299726,
+      "learning_rate": 1.288359843671237e-06,
+      "loss": 0.6765,
+      "step": 5875
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 15.047890302018441,
+      "learning_rate": 1.280510960572745e-06,
+      "loss": 0.6778,
+      "step": 5880
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 7.433657042047171,
+      "learning_rate": 1.2726825462537822e-06,
+      "loss": 0.71,
+      "step": 5885
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 18.217825688468125,
+      "learning_rate": 1.264874643795021e-06,
+      "loss": 0.7062,
+      "step": 5890
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 11.471285424059117,
+      "learning_rate": 1.2570872961642523e-06,
+      "loss": 0.6939,
+      "step": 5895
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 13.217245935701765,
+      "learning_rate": 1.2493205462161533e-06,
+      "loss": 0.6796,
+      "step": 5900
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 6.940856716689683,
+      "learning_rate": 1.2415744366920506e-06,
+      "loss": 0.6914,
+      "step": 5905
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 8.305386306019244,
+      "learning_rate": 1.2338490102196827e-06,
+      "loss": 0.6908,
+      "step": 5910
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 9.624149922390462,
+      "learning_rate": 1.2261443093129693e-06,
+      "loss": 0.7012,
+      "step": 5915
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 8.841023467812798,
+      "learning_rate": 1.2184603763717684e-06,
+      "loss": 0.6741,
+      "step": 5920
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 19.88249915582039,
+      "learning_rate": 1.2107972536816597e-06,
+      "loss": 0.698,
+      "step": 5925
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 14.487483894255304,
+      "learning_rate": 1.2031549834136924e-06,
+      "loss": 0.6745,
+      "step": 5930
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 8.062689979650553,
+      "learning_rate": 1.195533607624168e-06,
+      "loss": 0.6843,
+      "step": 5935
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 8.553635815257165,
+      "learning_rate": 1.1879331682543972e-06,
+      "loss": 0.6989,
+      "step": 5940
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 14.111776882579814,
+      "learning_rate": 1.1803537071304855e-06,
+      "loss": 0.6739,
+      "step": 5945
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 10.353991889949414,
+      "learning_rate": 1.172795265963082e-06,
+      "loss": 0.7074,
+      "step": 5950
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 7.929082880285211,
+      "learning_rate": 1.1652578863471664e-06,
+      "loss": 0.7038,
+      "step": 5955
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 13.55998270164541,
+      "learning_rate": 1.1577416097618138e-06,
+      "loss": 0.6913,
+      "step": 5960
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 7.208036588311436,
+      "learning_rate": 1.1502464775699619e-06,
+      "loss": 0.6773,
+      "step": 5965
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 17.777919169893906,
+      "learning_rate": 1.1427725310181982e-06,
+      "loss": 0.6838,
+      "step": 5970
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 7.720118007955193,
+      "learning_rate": 1.1353198112365133e-06,
+      "loss": 0.6983,
+      "step": 5975
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 7.141061885275735,
+      "learning_rate": 1.127888359238089e-06,
+      "loss": 0.6691,
+      "step": 5980
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 16.620150804924627,
+      "learning_rate": 1.1204782159190686e-06,
+      "loss": 0.7046,
+      "step": 5985
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 7.969373704721411,
+      "learning_rate": 1.1130894220583294e-06,
+      "loss": 0.6793,
+      "step": 5990
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 7.177320784722896,
+      "learning_rate": 1.1057220183172624e-06,
+      "loss": 0.6948,
+      "step": 5995
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 22.00401576107171,
+      "learning_rate": 1.0983760452395415e-06,
+      "loss": 0.7062,
+      "step": 6000
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 11.116453422441834,
+      "learning_rate": 1.0910515432509105e-06,
+      "loss": 0.7139,
+      "step": 6005
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 8.392865813607877,
+      "learning_rate": 1.0837485526589536e-06,
+      "loss": 0.689,
+      "step": 6010
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 7.916414734802656,
+      "learning_rate": 1.0764671136528742e-06,
+      "loss": 0.7104,
+      "step": 6015
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 7.197325104102647,
+      "learning_rate": 1.0692072663032771e-06,
+      "loss": 0.691,
+      "step": 6020
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 7.700014284452919,
+      "learning_rate": 1.0619690505619417e-06,
+      "loss": 0.6838,
+      "step": 6025
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 9.532064185210691,
+      "learning_rate": 1.0547525062616092e-06,
+      "loss": 0.663,
+      "step": 6030
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 8.172850502490537,
+      "learning_rate": 1.047557673115761e-06,
+      "loss": 0.6586,
+      "step": 6035
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 10.363207739603784,
+      "learning_rate": 1.040384590718399e-06,
+      "loss": 0.6898,
+      "step": 6040
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 8.670819751427343,
+      "learning_rate": 1.0332332985438248e-06,
+      "loss": 0.7059,
+      "step": 6045
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 10.973700515873551,
+      "learning_rate": 1.0261038359464304e-06,
+      "loss": 0.7076,
+      "step": 6050
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 13.842798651731552,
+      "learning_rate": 1.0189962421604738e-06,
+      "loss": 0.711,
+      "step": 6055
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 11.476103249170656,
+      "learning_rate": 1.01191055629987e-06,
+      "loss": 0.7292,
+      "step": 6060
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 7.600975072140781,
+      "learning_rate": 1.0048468173579695e-06,
+      "loss": 0.7072,
+      "step": 6065
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 16.85895201885889,
+      "learning_rate": 9.97805064207345e-07,
+      "loss": 0.6958,
+      "step": 6070
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 18.45110956773347,
+      "learning_rate": 9.907853355995818e-07,
+      "loss": 0.6949,
+      "step": 6075
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 8.974445511822958,
+      "learning_rate": 9.837876701650606e-07,
+      "loss": 0.7379,
+      "step": 6080
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 9.084116565720729,
+      "learning_rate": 9.768121064127462e-07,
+      "loss": 0.7109,
+      "step": 6085
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 8.589509628775673,
+      "learning_rate": 9.69858682729976e-07,
+      "loss": 0.7114,
+      "step": 6090
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 12.51093714626655,
+      "learning_rate": 9.629274373822423e-07,
+      "loss": 0.6928,
+      "step": 6095
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 12.663929219550615,
+      "learning_rate": 9.560184085129982e-07,
+      "loss": 0.7059,
+      "step": 6100
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 11.108973364813883,
+      "learning_rate": 9.491316341434264e-07,
+      "loss": 0.6841,
+      "step": 6105
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 8.430080678409125,
+      "learning_rate": 9.42267152172246e-07,
+      "loss": 0.6787,
+      "step": 6110
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 6.960533415526076,
+      "learning_rate": 9.354250003754983e-07,
+      "loss": 0.6881,
+      "step": 6115
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 7.086989797649012,
+      "learning_rate": 9.286052164063369e-07,
+      "loss": 0.689,
+      "step": 6120
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 12.175338378153151,
+      "learning_rate": 9.218078377948258e-07,
+      "loss": 0.6643,
+      "step": 6125
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 7.010238048011958,
+      "learning_rate": 9.150329019477255e-07,
+      "loss": 0.6802,
+      "step": 6130
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 8.003069949326946,
+      "learning_rate": 9.082804461482952e-07,
+      "loss": 0.7048,
+      "step": 6135
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 11.039749701379492,
+      "learning_rate": 9.015505075560799e-07,
+      "loss": 0.6936,
+      "step": 6140
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 7.477570093080713,
+      "learning_rate": 8.948431232067157e-07,
+      "loss": 0.7112,
+      "step": 6145
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 8.58743324892904,
+      "learning_rate": 8.881583300117136e-07,
+      "loss": 0.7217,
+      "step": 6150
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 22.804401422449356,
+      "learning_rate": 8.814961647582671e-07,
+      "loss": 0.7016,
+      "step": 6155
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 8.212024124224245,
+      "learning_rate": 8.748566641090433e-07,
+      "loss": 0.7179,
+      "step": 6160
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 9.517744104048571,
+      "learning_rate": 8.682398646019858e-07,
+      "loss": 0.6968,
+      "step": 6165
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 10.017391145857028,
+      "learning_rate": 8.61645802650109e-07,
+      "loss": 0.6814,
+      "step": 6170
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 7.151780421280677,
+      "learning_rate": 8.550745145412997e-07,
+      "loss": 0.6903,
+      "step": 6175
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 7.336742010971506,
+      "learning_rate": 8.485260364381187e-07,
+      "loss": 0.6975,
+      "step": 6180
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 10.109681497012197,
+      "learning_rate": 8.420004043776003e-07,
+      "loss": 0.6894,
+      "step": 6185
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 13.997305563658804,
+      "learning_rate": 8.354976542710558e-07,
+      "loss": 0.6817,
+      "step": 6190
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 7.759921099351603,
+      "learning_rate": 8.290178219038724e-07,
+      "loss": 0.7047,
+      "step": 6195
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 11.587020704329706,
+      "learning_rate": 8.225609429353187e-07,
+      "loss": 0.7226,
+      "step": 6200
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 7.505892642680992,
+      "learning_rate": 8.161270528983478e-07,
+      "loss": 0.7116,
+      "step": 6205
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 14.387670785851986,
+      "learning_rate": 8.097161871994042e-07,
+      "loss": 0.6901,
+      "step": 6210
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 8.860085820738941,
+      "learning_rate": 8.033283811182247e-07,
+      "loss": 0.7013,
+      "step": 6215
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 9.567448170081905,
+      "learning_rate": 7.969636698076477e-07,
+      "loss": 0.7122,
+      "step": 6220
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 7.883406283218387,
+      "learning_rate": 7.906220882934146e-07,
+      "loss": 0.6803,
+      "step": 6225
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 10.496750852129724,
+      "learning_rate": 7.843036714739888e-07,
+      "loss": 0.6788,
+      "step": 6230
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 8.0517863588975,
+      "learning_rate": 7.780084541203464e-07,
+      "loss": 0.6668,
+      "step": 6235
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 9.373203719623788,
+      "learning_rate": 7.717364708758024e-07,
+      "loss": 0.6704,
+      "step": 6240
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 7.245295491811536,
+      "learning_rate": 7.654877562558038e-07,
+      "loss": 0.6755,
+      "step": 6245
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 8.33051493384858,
+      "learning_rate": 7.592623446477543e-07,
+      "loss": 0.6804,
+      "step": 6250
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 10.03296372665666,
+      "learning_rate": 7.530602703108159e-07,
+      "loss": 0.6897,
+      "step": 6255
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 6.796926435441726,
+      "learning_rate": 7.468815673757218e-07,
+      "loss": 0.6693,
+      "step": 6260
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 7.8148484438509875,
+      "learning_rate": 7.407262698445932e-07,
+      "loss": 0.6722,
+      "step": 6265
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 11.161037506477513,
+      "learning_rate": 7.345944115907421e-07,
+      "loss": 0.6919,
+      "step": 6270
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 10.251223754865906,
+      "learning_rate": 7.284860263585003e-07,
+      "loss": 0.7192,
+      "step": 6275
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 8.416513873799278,
+      "learning_rate": 7.224011477630166e-07,
+      "loss": 0.6928,
+      "step": 6280
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 7.54571664772697,
+      "learning_rate": 7.163398092900853e-07,
+      "loss": 0.6742,
+      "step": 6285
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 7.630952619319009,
+      "learning_rate": 7.103020442959541e-07,
+      "loss": 0.7067,
+      "step": 6290
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 7.415041807839196,
+      "learning_rate": 7.04287886007144e-07,
+      "loss": 0.6889,
+      "step": 6295
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 8.325897258274617,
+      "learning_rate": 6.982973675202676e-07,
+      "loss": 0.672,
+      "step": 6300
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 14.829942805859428,
+      "learning_rate": 6.923305218018395e-07,
+      "loss": 0.6624,
+      "step": 6305
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 7.8483093268929975,
+      "learning_rate": 6.863873816881061e-07,
+      "loss": 0.7021,
+      "step": 6310
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 6.92498216547613,
+      "learning_rate": 6.80467979884854e-07,
+      "loss": 0.6915,
+      "step": 6315
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 8.838108298902782,
+      "learning_rate": 6.745723489672412e-07,
+      "loss": 0.7,
+      "step": 6320
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 7.613133145194472,
+      "learning_rate": 6.687005213796072e-07,
+      "loss": 0.697,
+      "step": 6325
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 12.62935627779005,
+      "learning_rate": 6.628525294352989e-07,
+      "loss": 0.7083,
+      "step": 6330
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 6.774865982386047,
+      "learning_rate": 6.570284053164943e-07,
+      "loss": 0.6816,
+      "step": 6335
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 8.118843212773909,
+      "learning_rate": 6.512281810740251e-07,
+      "loss": 0.6882,
+      "step": 6340
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 10.630712949397768,
+      "learning_rate": 6.454518886271982e-07,
+      "loss": 0.6984,
+      "step": 6345
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 8.428110096616827,
+      "learning_rate": 6.396995597636191e-07,
+      "loss": 0.6644,
+      "step": 6350
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 7.780288857084446,
+      "learning_rate": 6.339712261390213e-07,
+      "loss": 0.6848,
+      "step": 6355
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 21.409925250361773,
+      "learning_rate": 6.282669192770896e-07,
+      "loss": 0.6967,
+      "step": 6360
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.240745644008053,
+      "learning_rate": 6.225866705692857e-07,
+      "loss": 0.6883,
+      "step": 6365
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.674756280883551,
+      "learning_rate": 6.169305112746777e-07,
+      "loss": 0.6951,
+      "step": 6370
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 12.048397050723365,
+      "learning_rate": 6.112984725197646e-07,
+      "loss": 0.7429,
+      "step": 6375
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 11.577785729229298,
+      "learning_rate": 6.056905852983103e-07,
+      "loss": 0.6849,
+      "step": 6380
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 13.26700761340119,
+      "learning_rate": 6.001068804711674e-07,
+      "loss": 0.6761,
+      "step": 6385
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 7.293043200627963,
+      "learning_rate": 5.945473887661118e-07,
+      "loss": 0.6724,
+      "step": 6390
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 8.457648865570288,
+      "learning_rate": 5.890121407776716e-07,
+      "loss": 0.6528,
+      "step": 6395
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 9.492245777716876,
+      "learning_rate": 5.83501166966956e-07,
+      "loss": 0.6592,
+      "step": 6400
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 7.626565378949079,
+      "learning_rate": 5.780144976614971e-07,
+      "loss": 0.67,
+      "step": 6405
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 9.903758819130118,
+      "learning_rate": 5.725521630550684e-07,
+      "loss": 0.7089,
+      "step": 6410
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 11.221012430791403,
+      "learning_rate": 5.671141932075336e-07,
+      "loss": 0.6768,
+      "step": 6415
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 7.0216988076201154,
+      "learning_rate": 5.617006180446688e-07,
+      "loss": 0.68,
+      "step": 6420
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 10.261197273236373,
+      "learning_rate": 5.563114673580062e-07,
+      "loss": 0.7065,
+      "step": 6425
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 10.444019940639068,
+      "learning_rate": 5.509467708046706e-07,
+      "loss": 0.6883,
+      "step": 6430
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 10.814008582648402,
+      "learning_rate": 5.456065579072051e-07,
+      "loss": 0.6791,
+      "step": 6435
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 12.625740069874405,
+      "learning_rate": 5.402908580534233e-07,
+      "loss": 0.7143,
+      "step": 6440
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 32.28495620280846,
+      "learning_rate": 5.349997004962348e-07,
+      "loss": 0.6818,
+      "step": 6445
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 8.018590214419145,
+      "learning_rate": 5.297331143534973e-07,
+      "loss": 0.6683,
+      "step": 6450
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 10.843690308266353,
+      "learning_rate": 5.244911286078419e-07,
+      "loss": 0.6646,
+      "step": 6455
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 7.448060701438149,
+      "learning_rate": 5.192737721065255e-07,
+      "loss": 0.6647,
+      "step": 6460
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 7.497660606213866,
+      "learning_rate": 5.140810735612656e-07,
+      "loss": 0.7232,
+      "step": 6465
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 13.747703268296267,
+      "learning_rate": 5.089130615480842e-07,
+      "loss": 0.7328,
+      "step": 6470
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 10.980931578514912,
+      "learning_rate": 5.037697645071527e-07,
+      "loss": 0.6819,
+      "step": 6475
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 11.102550489688186,
+      "learning_rate": 4.986512107426283e-07,
+      "loss": 0.6665,
+      "step": 6480
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 7.309435947152377,
+      "learning_rate": 4.93557428422507e-07,
+      "loss": 0.6694,
+      "step": 6485
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 6.934226738544792,
+      "learning_rate": 4.884884455784644e-07,
+      "loss": 0.6875,
+      "step": 6490
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 6.6508005982102825,
+      "learning_rate": 4.834442901057001e-07,
+      "loss": 0.6796,
+      "step": 6495
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 7.608629277763512,
+      "learning_rate": 4.78424989762788e-07,
+      "loss": 0.7415,
+      "step": 6500
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 7.050556142459157,
+      "learning_rate": 4.734305721715182e-07,
+      "loss": 0.6769,
+      "step": 6505
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 8.091611569372494,
+      "learning_rate": 4.6846106481675035e-07,
+      "loss": 0.6615,
+      "step": 6510
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 12.15038336162669,
+      "learning_rate": 4.6351649504626007e-07,
+      "loss": 0.6822,
+      "step": 6515
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 9.24527503269406,
+      "learning_rate": 4.5859689007058896e-07,
+      "loss": 0.6917,
+      "step": 6520
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 7.303872356527945,
+      "learning_rate": 4.537022769628924e-07,
+      "loss": 0.6727,
+      "step": 6525
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 6.8918656083018925,
+      "learning_rate": 4.488326826587941e-07,
+      "loss": 0.6839,
+      "step": 6530
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 7.73491402517389,
+      "learning_rate": 4.439881339562363e-07,
+      "loss": 0.6316,
+      "step": 6535
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 11.439001983592874,
+      "learning_rate": 4.3916865751533313e-07,
+      "loss": 0.6758,
+      "step": 6540
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 10.5221379590623,
+      "learning_rate": 4.343742798582229e-07,
+      "loss": 0.6607,
+      "step": 6545
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 7.646233632679172,
+      "learning_rate": 4.296050273689195e-07,
+      "loss": 0.6707,
+      "step": 6550
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 7.671968491727378,
+      "learning_rate": 4.2486092629317354e-07,
+      "loss": 0.6687,
+      "step": 6555
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 8.46067866415217,
+      "learning_rate": 4.2014200273832406e-07,
+      "loss": 0.7214,
+      "step": 6560
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 9.092465339705312,
+      "learning_rate": 4.154482826731543e-07,
+      "loss": 0.65,
+      "step": 6565
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 7.019529590276667,
+      "learning_rate": 4.1077979192775065e-07,
+      "loss": 0.6985,
+      "step": 6570
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 10.084817313774266,
+      "learning_rate": 4.0613655619335667e-07,
+      "loss": 0.6887,
+      "step": 6575
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 10.511653607166885,
+      "learning_rate": 4.015186010222405e-07,
+      "loss": 0.6639,
+      "step": 6580
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 7.463141604936293,
+      "learning_rate": 3.969259518275431e-07,
+      "loss": 0.6306,
+      "step": 6585
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 7.79565484434834,
+      "learning_rate": 3.923586338831453e-07,
+      "loss": 0.689,
+      "step": 6590
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 6.712370198831204,
+      "learning_rate": 3.878166723235288e-07,
+      "loss": 0.6938,
+      "step": 6595
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 12.790220661302872,
+      "learning_rate": 3.8330009214363197e-07,
+      "loss": 0.6758,
+      "step": 6600
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 10.327767164889268,
+      "learning_rate": 3.7880891819872365e-07,
+      "loss": 0.6591,
+      "step": 6605
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 7.5003444196363755,
+      "learning_rate": 3.743431752042509e-07,
+      "loss": 0.708,
+      "step": 6610
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 9.2620020592344,
+      "learning_rate": 3.6990288773571856e-07,
+      "loss": 0.6975,
+      "step": 6615
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 10.860966745902388,
+      "learning_rate": 3.6548808022853933e-07,
+      "loss": 0.6732,
+      "step": 6620
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 6.781412412422652,
+      "learning_rate": 3.610987769779156e-07,
+      "loss": 0.6818,
+      "step": 6625
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 7.029500433738177,
+      "learning_rate": 3.567350021386895e-07,
+      "loss": 0.6908,
+      "step": 6630
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 7.368106695431497,
+      "learning_rate": 3.5239677972522037e-07,
+      "loss": 0.6717,
+      "step": 6635
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 6.990527222441549,
+      "learning_rate": 3.4808413361125004e-07,
+      "loss": 0.6901,
+      "step": 6640
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 8.153704370737962,
+      "learning_rate": 3.4379708752976804e-07,
+      "loss": 0.6713,
+      "step": 6645
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 7.367420549219586,
+      "learning_rate": 3.395356650728893e-07,
+      "loss": 0.6869,
+      "step": 6650
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 8.373370537605064,
+      "learning_rate": 3.352998896917131e-07,
+      "loss": 0.688,
+      "step": 6655
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 8.322681648516124,
+      "learning_rate": 3.310897846962041e-07,
+      "loss": 0.687,
+      "step": 6660
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 7.713115881099371,
+      "learning_rate": 3.269053732550581e-07,
+      "loss": 0.6526,
+      "step": 6665
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 7.2157260593853705,
+      "learning_rate": 3.2274667839557574e-07,
+      "loss": 0.7072,
+      "step": 6670
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 8.581960613658934,
+      "learning_rate": 3.1861372300353856e-07,
+      "loss": 0.6515,
+      "step": 6675
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 8.859775154512494,
+      "learning_rate": 3.1450652982307815e-07,
+      "loss": 0.6791,
+      "step": 6680
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 7.821560290050431,
+      "learning_rate": 3.1042512145655456e-07,
+      "loss": 0.6657,
+      "step": 6685
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 10.406315452535901,
+      "learning_rate": 3.063695203644329e-07,
+      "loss": 0.6792,
+      "step": 6690
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 8.088514615395225,
+      "learning_rate": 3.023397488651547e-07,
+      "loss": 0.679,
+      "step": 6695
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 8.139576090720407,
+      "learning_rate": 2.9833582913502136e-07,
+      "loss": 0.7124,
+      "step": 6700
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 6.951908590242519,
+      "learning_rate": 2.943577832080663e-07,
+      "loss": 0.6536,
+      "step": 6705
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 9.634692866480727,
+      "learning_rate": 2.90405632975938e-07,
+      "loss": 0.6841,
+      "step": 6710
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 7.583020996780692,
+      "learning_rate": 2.864794001877774e-07,
+      "loss": 0.6584,
+      "step": 6715
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 7.133337233983223,
+      "learning_rate": 2.8257910645009935e-07,
+      "loss": 0.6895,
+      "step": 6720
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 8.487491873845302,
+      "learning_rate": 2.7870477322667176e-07,
+      "loss": 0.6684,
+      "step": 6725
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 7.275742235305359,
+      "learning_rate": 2.7485642183840036e-07,
+      "loss": 0.6737,
+      "step": 6730
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 10.234808397687504,
+      "learning_rate": 2.710340734632083e-07,
+      "loss": 0.683,
+      "step": 6735
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 13.39492396596042,
+      "learning_rate": 2.6723774913592294e-07,
+      "loss": 0.7026,
+      "step": 6740
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 7.688204833170652,
+      "learning_rate": 2.634674697481576e-07,
+      "loss": 0.6781,
+      "step": 6745
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 7.015038753163292,
+      "learning_rate": 2.597232560481949e-07,
+      "loss": 0.6382,
+      "step": 6750
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 8.246976015566766,
+      "learning_rate": 2.5600512864087935e-07,
+      "loss": 0.655,
+      "step": 6755
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 7.9917906454794645,
+      "learning_rate": 2.523131079874963e-07,
+      "loss": 0.6937,
+      "step": 6760
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 7.030953216673749,
+      "learning_rate": 2.486472144056634e-07,
+      "loss": 0.6729,
+      "step": 6765
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 8.008501775119932,
+      "learning_rate": 2.450074680692188e-07,
+      "loss": 0.6966,
+      "step": 6770
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 7.535600491082128,
+      "learning_rate": 2.413938890081069e-07,
+      "loss": 0.6866,
+      "step": 6775
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 9.984049476041328,
+      "learning_rate": 2.3780649710827552e-07,
+      "loss": 0.6991,
+      "step": 6780
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 6.637614427528961,
+      "learning_rate": 2.3424531211155565e-07,
+      "loss": 0.6528,
+      "step": 6785
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 16.226665886178726,
+      "learning_rate": 2.3071035361556194e-07,
+      "loss": 0.6732,
+      "step": 6790
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 8.68628996022435,
+      "learning_rate": 2.272016410735811e-07,
+      "loss": 0.6689,
+      "step": 6795
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 8.516187854598774,
+      "learning_rate": 2.2371919379446495e-07,
+      "loss": 0.6586,
+      "step": 6800
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 8.620085418870987,
+      "learning_rate": 2.2026303094252632e-07,
+      "loss": 0.7165,
+      "step": 6805
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 9.201811624782207,
+      "learning_rate": 2.1683317153742777e-07,
+      "loss": 0.6939,
+      "step": 6810
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.195464331596205,
+      "learning_rate": 2.1342963445408483e-07,
+      "loss": 0.6664,
+      "step": 6815
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.5236578927542475,
+      "learning_rate": 2.1005243842255552e-07,
+      "loss": 0.6884,
+      "step": 6820
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.70860379052487,
+      "learning_rate": 2.067016020279433e-07,
+      "loss": 0.6657,
+      "step": 6825
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 8.771313327660119,
+      "learning_rate": 2.033771437102877e-07,
+      "loss": 0.6688,
+      "step": 6830
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 9.764299285160561,
+      "learning_rate": 2.0007908176446879e-07,
+      "loss": 0.6725,
+      "step": 6835
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 6.933795763348559,
+      "learning_rate": 1.9680743434010385e-07,
+      "loss": 0.6874,
+      "step": 6840
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 12.074043349607445,
+      "learning_rate": 1.9356221944144826e-07,
+      "loss": 0.7027,
+      "step": 6845
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 6.81518182102359,
+      "learning_rate": 1.90343454927297e-07,
+      "loss": 0.6924,
+      "step": 6850
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 9.204048705819153,
+      "learning_rate": 1.8715115851088263e-07,
+      "loss": 0.6515,
+      "step": 6855
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 7.244875656674812,
+      "learning_rate": 1.8398534775978305e-07,
+      "loss": 0.6369,
+      "step": 6860
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 7.778287344019572,
+      "learning_rate": 1.8084604009582229e-07,
+      "loss": 0.6434,
+      "step": 6865
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 8.821616561739027,
+      "learning_rate": 1.777332527949749e-07,
+      "loss": 0.6892,
+      "step": 6870
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 7.148975540163136,
+      "learning_rate": 1.7464700298727044e-07,
+      "loss": 0.6897,
+      "step": 6875
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 10.13058116392989,
+      "learning_rate": 1.7158730765669817e-07,
+      "loss": 0.6861,
+      "step": 6880
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 6.651987875873002,
+      "learning_rate": 1.6855418364111698e-07,
+      "loss": 0.6858,
+      "step": 6885
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.8895295822966816,
+      "learning_rate": 1.6554764763216047e-07,
+      "loss": 0.696,
+      "step": 6890
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.250140261393451,
+      "learning_rate": 1.6256771617514487e-07,
+      "loss": 0.6439,
+      "step": 6895
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.716682557279987,
+      "learning_rate": 1.5961440566897913e-07,
+      "loss": 0.6786,
+      "step": 6900
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.595444153908226,
+      "learning_rate": 1.5668773236607092e-07,
+      "loss": 0.6746,
+      "step": 6905
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 7.328748971928146,
+      "learning_rate": 1.5378771237224643e-07,
+      "loss": 0.6635,
+      "step": 6910
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 6.669088243571489,
+      "learning_rate": 1.5091436164664963e-07,
+      "loss": 0.6479,
+      "step": 6915
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 10.050157025742639,
+      "learning_rate": 1.480676960016636e-07,
+      "loss": 0.6696,
+      "step": 6920
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 7.251173160471616,
+      "learning_rate": 1.4524773110281888e-07,
+      "loss": 0.6662,
+      "step": 6925
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 7.465237655188784,
+      "learning_rate": 1.424544824687102e-07,
+      "loss": 0.6851,
+      "step": 6930
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 8.16883255403586,
+      "learning_rate": 1.3968796547090768e-07,
+      "loss": 0.6696,
+      "step": 6935
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 8.089671615730229,
+      "learning_rate": 1.369481953338764e-07,
+      "loss": 0.6713,
+      "step": 6940
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 10.260450087341056,
+      "learning_rate": 1.342351871348885e-07,
+      "loss": 0.6776,
+      "step": 6945
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 11.258479357560187,
+      "learning_rate": 1.315489558039418e-07,
+      "loss": 0.6898,
+      "step": 6950
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 7.380202432487591,
+      "learning_rate": 1.2888951612368084e-07,
+      "loss": 0.6698,
+      "step": 6955
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 8.12243172392476,
+      "learning_rate": 1.2625688272930925e-07,
+      "loss": 0.652,
+      "step": 6960
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 7.166583155520059,
+      "learning_rate": 1.2365107010851417e-07,
+      "loss": 0.6529,
+      "step": 6965
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 7.0987346872586885,
+      "learning_rate": 1.2107209260138474e-07,
+      "loss": 0.6855,
+      "step": 6970
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 7.296487392678461,
+      "learning_rate": 1.185199644003332e-07,
+      "loss": 0.6811,
+      "step": 6975
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 7.115812289725834,
+      "learning_rate": 1.1599469955001719e-07,
+      "loss": 0.6822,
+      "step": 6980
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 7.055262412503819,
+      "learning_rate": 1.1349631194726151e-07,
+      "loss": 0.6837,
+      "step": 6985
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 7.884264180319673,
+      "learning_rate": 1.1102481534098375e-07,
+      "loss": 0.6693,
+      "step": 6990
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 7.534291720280173,
+      "learning_rate": 1.0858022333211481e-07,
+      "loss": 0.6641,
+      "step": 6995
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 7.300432052216407,
+      "learning_rate": 1.0616254937352966e-07,
+      "loss": 0.6698,
+      "step": 7000
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 9.667278163540502,
+      "learning_rate": 1.0377180676996834e-07,
+      "loss": 0.7191,
+      "step": 7005
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 8.933023403124968,
+      "learning_rate": 1.014080086779634e-07,
+      "loss": 0.678,
+      "step": 7010
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 7.17753973528174,
+      "learning_rate": 9.907116810577045e-08,
+      "loss": 0.6863,
+      "step": 7015
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 7.868822264873724,
+      "learning_rate": 9.676129791329481e-08,
+      "loss": 0.6254,
+      "step": 7020
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 6.508467484246546,
+      "learning_rate": 9.447841081201948e-08,
+      "loss": 0.6573,
+      "step": 7025
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 8.38472848445943,
+      "learning_rate": 9.222251936493675e-08,
+      "loss": 0.6897,
+      "step": 7030
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 7.074584969820798,
+      "learning_rate": 8.999363598647893e-08,
+      "loss": 0.6878,
+      "step": 7035
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 7.298711666179283,
+      "learning_rate": 8.779177294245044e-08,
+      "loss": 0.7161,
+      "step": 7040
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 7.952323915272186,
+      "learning_rate": 8.561694234995754e-08,
+      "loss": 0.7145,
+      "step": 7045
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 7.033240486657288,
+      "learning_rate": 8.346915617734707e-08,
+      "loss": 0.7128,
+      "step": 7050
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 8.340030079448628,
+      "learning_rate": 8.134842624413387e-08,
+      "loss": 0.7025,
+      "step": 7055
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 7.581166478719366,
+      "learning_rate": 7.925476422094125e-08,
+      "loss": 0.7033,
+      "step": 7060
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 6.987430203222376,
+      "learning_rate": 7.718818162943398e-08,
+      "loss": 0.6567,
+      "step": 7065
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 7.992039783034009,
+      "learning_rate": 7.514868984225599e-08,
+      "loss": 0.6677,
+      "step": 7070
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 7.134161947829603,
+      "learning_rate": 7.313630008296713e-08,
+      "loss": 0.6692,
+      "step": 7075
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 8.221096961441273,
+      "learning_rate": 7.115102342598101e-08,
+      "loss": 0.6671,
+      "step": 7080
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 7.376520411371209,
+      "learning_rate": 6.919287079650616e-08,
+      "loss": 0.6762,
+      "step": 7085
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 6.845969768343702,
+      "learning_rate": 6.726185297048326e-08,
+      "loss": 0.6694,
+      "step": 7090
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 6.238211345432872,
+      "learning_rate": 6.535798057452691e-08,
+      "loss": 0.6376,
+      "step": 7095
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 7.249286502161536,
+      "learning_rate": 6.34812640858684e-08,
+      "loss": 0.6945,
+      "step": 7100
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 6.775723266678111,
+      "learning_rate": 6.163171383229527e-08,
+      "loss": 0.6679,
+      "step": 7105
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 7.759292429954733,
+      "learning_rate": 5.980933999209793e-08,
+      "loss": 0.662,
+      "step": 7110
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 8.912249199662226,
+      "learning_rate": 5.801415259401033e-08,
+      "loss": 0.6757,
+      "step": 7115
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 8.013436529211443,
+      "learning_rate": 5.6246161517158336e-08,
+      "loss": 0.6905,
+      "step": 7120
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 8.848628631696203,
+      "learning_rate": 5.450537649099918e-08,
+      "loss": 0.6779,
+      "step": 7125
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 7.843003604456921,
+      "learning_rate": 5.279180709527765e-08,
+      "loss": 0.6703,
+      "step": 7130
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 7.1693855907424275,
+      "learning_rate": 5.110546275996275e-08,
+      "loss": 0.6847,
+      "step": 7135
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 8.695889856273157,
+      "learning_rate": 4.944635276520393e-08,
+      "loss": 0.6283,
+      "step": 7140
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 8.10405457925902,
+      "learning_rate": 4.781448624127605e-08,
+      "loss": 0.7114,
+      "step": 7145
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 7.346064882938857,
+      "learning_rate": 4.6209872168528905e-08,
+      "loss": 0.6772,
+      "step": 7150
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 6.821512731867321,
+      "learning_rate": 4.463251937734059e-08,
+      "loss": 0.6725,
+      "step": 7155
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 7.70332817085616,
+      "learning_rate": 4.308243654806643e-08,
+      "loss": 0.6837,
+      "step": 7160
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 7.043942123939794,
+      "learning_rate": 4.155963221099124e-08,
+      "loss": 0.6752,
+      "step": 7165
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 6.883979197861966,
+      "learning_rate": 4.006411474628491e-08,
+      "loss": 0.6633,
+      "step": 7170
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 7.702076002410021,
+      "learning_rate": 3.8595892383953e-08,
+      "loss": 0.6617,
+      "step": 7175
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 8.638589076794691,
+      "learning_rate": 3.7154973203793466e-08,
+      "loss": 0.6888,
+      "step": 7180
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 7.24263452036519,
+      "learning_rate": 3.5741365135351646e-08,
+      "loss": 0.6735,
+      "step": 7185
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 6.886376403840186,
+      "learning_rate": 3.435507595787646e-08,
+      "loss": 0.665,
+      "step": 7190
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 7.192528171033981,
+      "learning_rate": 3.299611330027819e-08,
+      "loss": 0.6523,
+      "step": 7195
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 6.863720633968301,
+      "learning_rate": 3.166448464108629e-08,
+      "loss": 0.673,
+      "step": 7200
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 6.98578157726086,
+      "learning_rate": 3.036019730840667e-08,
+      "loss": 0.6927,
+      "step": 7205
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 6.963184873488591,
+      "learning_rate": 2.9083258479885023e-08,
+      "loss": 0.6784,
+      "step": 7210
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 6.921141305787604,
+      "learning_rate": 2.7833675182663e-08,
+      "loss": 0.6901,
+      "step": 7215
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 7.151454839165128,
+      "learning_rate": 2.661145429334322e-08,
+      "loss": 0.7008,
+      "step": 7220
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 6.924299978808564,
+      "learning_rate": 2.5416602537948197e-08,
+      "loss": 0.7242,
+      "step": 7225
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 8.602824717701267,
+      "learning_rate": 2.4249126491885934e-08,
+      "loss": 0.6975,
+      "step": 7230
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 7.264360226535887,
+      "learning_rate": 2.310903257991215e-08,
+      "loss": 0.6565,
+      "step": 7235
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 7.681180289656343,
+      "learning_rate": 2.1996327076096446e-08,
+      "loss": 0.6966,
+      "step": 7240
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 6.947285450893149,
+      "learning_rate": 2.091101610378621e-08,
+      "loss": 0.697,
+      "step": 7245
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 7.720434569800489,
+      "learning_rate": 1.98531056355733e-08,
+      "loss": 0.6616,
+      "step": 7250
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.386434329519245,
+      "learning_rate": 1.8822601493261316e-08,
+      "loss": 0.6691,
+      "step": 7255
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 6.6596322168750435,
+      "learning_rate": 1.781950934783505e-08,
+      "loss": 0.6516,
+      "step": 7260
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.262432991908425,
+      "learning_rate": 1.684383471942608e-08,
+      "loss": 0.6575,
+      "step": 7265
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.311890451065669,
+      "learning_rate": 1.589558297728555e-08,
+      "loss": 0.6636,
+      "step": 7270
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.906000744788898,
+      "learning_rate": 1.497475933975312e-08,
+      "loss": 0.696,
+      "step": 7275
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 12.045630756085984,
+      "learning_rate": 1.4081368874226398e-08,
+      "loss": 0.6992,
+      "step": 7280
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 7.405966176562505,
+      "learning_rate": 1.3215416497138756e-08,
+      "loss": 0.7323,
+      "step": 7285
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 9.309630509701096,
+      "learning_rate": 1.2376906973924907e-08,
+      "loss": 0.6933,
+      "step": 7290
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 7.507448689843324,
+      "learning_rate": 1.156584491900148e-08,
+      "loss": 0.6805,
+      "step": 7295
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 8.272436921961551,
+      "learning_rate": 1.0782234795735368e-08,
+      "loss": 0.6938,
+      "step": 7300
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 6.618968260399834,
+      "learning_rate": 1.0026080916425429e-08,
+      "loss": 0.676,
+      "step": 7305
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 9.009224943936573,
+      "learning_rate": 9.297387442273598e-09,
+      "loss": 0.6493,
+      "step": 7310
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 7.405492273791612,
+      "learning_rate": 8.596158383364361e-09,
+      "loss": 0.6822,
+      "step": 7315
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 6.855916339254849,
+      "learning_rate": 7.922397598642551e-09,
+      "loss": 0.6763,
+      "step": 7320
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 8.964068833091947,
+      "learning_rate": 7.276108795892245e-09,
+      "loss": 0.7172,
+      "step": 7325
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 7.736986286585069,
+      "learning_rate": 6.657295531715679e-09,
+      "loss": 0.6524,
+      "step": 7330
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 7.023965236554041,
+      "learning_rate": 6.06596121151326e-09,
+      "loss": 0.6782,
+      "step": 7335
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 6.972496079634721,
+      "learning_rate": 5.5021090894674665e-09,
+      "loss": 0.6694,
+      "step": 7340
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 6.836599421823032,
+      "learning_rate": 4.965742268521201e-09,
+      "loss": 0.6867,
+      "step": 7345
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 7.067217726173067,
+      "learning_rate": 4.456863700363357e-09,
+      "loss": 0.6688,
+      "step": 7350
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 7.689340391484074,
+      "learning_rate": 3.975476185411608e-09,
+      "loss": 0.658,
+      "step": 7355
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 7.420542436912137,
+      "learning_rate": 3.5215823727974274e-09,
+      "loss": 0.6736,
+      "step": 7360
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 7.109044382994916,
+      "learning_rate": 3.0951847603516437e-09,
+      "loss": 0.6536,
+      "step": 7365
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 7.020917382250396,
+      "learning_rate": 2.696285694588352e-09,
+      "loss": 0.663,
+      "step": 7370
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 6.523996927191407,
+      "learning_rate": 2.3248873706971374e-09,
+      "loss": 0.6983,
+      "step": 7375
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 6.9865517555009005,
+      "learning_rate": 1.980991832524759e-09,
+      "loss": 0.6815,
+      "step": 7380
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 9.119159209725249,
+      "learning_rate": 1.6646009725684864e-09,
+      "loss": 0.6788,
+      "step": 7385
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 7.7265477482194855,
+      "learning_rate": 1.3757165319644439e-09,
+      "loss": 0.6629,
+      "step": 7390
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 7.662052856250405,
+      "learning_rate": 1.1143401004765075e-09,
+      "loss": 0.6561,
+      "step": 7395
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 9.781011008340089,
+      "learning_rate": 8.804731164901991e-10,
+      "loss": 0.6837,
+      "step": 7400
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 6.9540258783343,
+      "learning_rate": 6.741168670021392e-10,
+      "loss": 0.6504,
+      "step": 7405
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 7.210146643455515,
+      "learning_rate": 4.952724876150505e-10,
+      "loss": 0.6599,
+      "step": 7410
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 7.3052428055619165,
+      "learning_rate": 3.439409625294321e-10,
+      "loss": 0.6877,
+      "step": 7415
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 10.38665088184887,
+      "learning_rate": 2.2012312454133822e-10,
+      "loss": 0.6678,
+      "step": 7420
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 8.391998417326777,
+      "learning_rate": 1.2381965503460712e-10,
+      "loss": 0.6748,
+      "step": 7425
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 8.835244351989205,
+      "learning_rate": 5.503108397753032e-11,
+      "loss": 0.7143,
+      "step": 7430
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 6.941822447796259,
+      "learning_rate": 1.375778992229737e-11,
+      "loss": 0.6986,
+      "step": 7435
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 7.077387432765926,
+      "learning_rate": 0.0,
+      "loss": 0.6836,
+      "step": 7440
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 1.1355550289154053,
+      "eval_runtime": 164.9125,
+      "eval_samples_per_second": 4.887,
+      "eval_steps_per_second": 0.612,
+      "step": 7440
+    },
+    {
+      "epoch": 3.0,
+      "step": 7440,
+      "total_flos": 256118529982464.0,
+      "train_loss": 1.1025839365938659,
+      "train_runtime": 66549.14,
+      "train_samples_per_second": 1.789,
+      "train_steps_per_second": 0.112
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7440,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 256118529982464.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}