diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2844 +1,12761 @@
 {
-  "best_metric": 0.8456375838926175,
-  "best_model_checkpoint": "ctsinov1/checkpoint-3148",
-  "epoch": 4.2,
+  "best_global_step": 9800,
+  "best_metric": 0.8585858585858586,
+  "best_model_checkpoint": "ctsinov1/checkpoint-9800",
+  "epoch": 49.02,
   "eval_steps": 500,
-  "global_step": 3935,
+  "global_step": 17500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
-      "grad_norm": 6.577972888946533,
-      "learning_rate": 2.538071065989848e-07,
-      "loss": 0.7076,
+      "epoch": 0.0005714285714285715,
+      "grad_norm": 6.513674259185791,
+      "learning_rate": 5.142857142857143e-08,
+      "loss": 0.696,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "grad_norm": 6.098381519317627,
-      "learning_rate": 5.076142131979696e-07,
-      "loss": 0.7072,
+      "epoch": 0.001142857142857143,
+      "grad_norm": 2.7451493740081787,
+      "learning_rate": 1.0857142857142857e-07,
+      "loss": 0.6997,
       "step": 20
     },
     {
-      "epoch": 0.01,
-      "grad_norm": 6.236210346221924,
-      "learning_rate": 7.614213197969544e-07,
-      "loss": 0.7278,
+      "epoch": 0.0017142857142857142,
+      "grad_norm": 2.719616174697876,
+      "learning_rate": 1.657142857142857e-07,
+      "loss": 0.6917,
       "step": 30
     },
     {
-      "epoch": 0.01,
-      "grad_norm": 7.610434532165527,
-      "learning_rate": 1.0152284263959392e-06,
-      "loss": 0.6914,
+      "epoch": 0.002285714285714286,
+      "grad_norm": 2.852513074874878,
+      "learning_rate": 2.228571428571429e-07,
+      "loss": 0.6825,
       "step": 40
     },
     {
-      "epoch": 0.01,
-      "grad_norm": 5.927430152893066,
-      "learning_rate": 1.2690355329949238e-06,
-      "loss": 0.7105,
+      "epoch": 0.002857142857142857,
+      "grad_norm": 6.225112438201904,
+      "learning_rate": 2.8e-07,
+      "loss": 0.685,
       "step": 50
     },
     {
-      "epoch": 0.02,
-      "grad_norm": 22.82501220703125,
-      "learning_rate": 1.5228426395939088e-06,
-      "loss": 0.694,
+      "epoch": 0.0034285714285714284,
+      "grad_norm": 2.038743495941162,
+      "learning_rate": 3.371428571428572e-07,
+      "loss": 0.691,
       "step": 60
     },
     {
-      "epoch": 0.02,
-      "grad_norm": 24.948528289794922,
-      "learning_rate": 1.7766497461928936e-06,
-      "loss": 0.7014,
+      "epoch": 0.004,
+      "grad_norm": 2.501828908920288,
+      "learning_rate": 3.9428571428571436e-07,
+      "loss": 0.701,
       "step": 70
     },
     {
-      "epoch": 0.02,
-      "grad_norm": 12.509531021118164,
-      "learning_rate": 2.0304568527918785e-06,
-      "loss": 0.6915,
+      "epoch": 0.004571428571428572,
+      "grad_norm": 5.308904647827148,
+      "learning_rate": 4.514285714285715e-07,
+      "loss": 0.6841,
       "step": 80
     },
     {
-      "epoch": 0.02,
-      "grad_norm": 10.74423885345459,
-      "learning_rate": 2.284263959390863e-06,
-      "loss": 0.6788,
+      "epoch": 0.005142857142857143,
+      "grad_norm": 2.171583890914917,
+      "learning_rate": 5.085714285714286e-07,
+      "loss": 0.6845,
       "step": 90
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 16.888263702392578,
-      "learning_rate": 2.5380710659898476e-06,
-      "loss": 0.6601,
+      "epoch": 0.005714285714285714,
+      "grad_norm": 1.9283939599990845,
+      "learning_rate": 5.657142857142857e-07,
+      "loss": 0.6901,
       "step": 100
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 10.467493057250977,
-      "learning_rate": 2.7918781725888327e-06,
-      "loss": 0.6162,
+      "epoch": 0.006285714285714286,
+      "grad_norm": 5.542513370513916,
+      "learning_rate": 6.228571428571429e-07,
+      "loss": 0.6768,
       "step": 110
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 13.966683387756348,
-      "learning_rate": 3.0456852791878177e-06,
-      "loss": 0.5912,
+      "epoch": 0.006857142857142857,
+      "grad_norm": 6.506889820098877,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.666,
       "step": 120
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 5.591104030609131,
-      "learning_rate": 3.2994923857868023e-06,
-      "loss": 0.4726,
+      "epoch": 0.0074285714285714285,
+      "grad_norm": 3.021516799926758,
+      "learning_rate": 7.371428571428572e-07,
+      "loss": 0.6799,
       "step": 130
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 8.37885856628418,
-      "learning_rate": 3.5532994923857873e-06,
-      "loss": 1.0922,
+      "epoch": 0.008,
+      "grad_norm": 3.173678398132324,
+      "learning_rate": 7.942857142857144e-07,
+      "loss": 0.6749,
       "step": 140
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 24.369611740112305,
-      "learning_rate": 3.8071065989847715e-06,
-      "loss": 1.0916,
+      "epoch": 0.008571428571428572,
+      "grad_norm": 6.398750305175781,
+      "learning_rate": 8.514285714285716e-07,
+      "loss": 0.7071,
       "step": 150
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 12.621369361877441,
-      "learning_rate": 4.060913705583757e-06,
-      "loss": 0.6395,
+      "epoch": 0.009142857142857144,
+      "grad_norm": 4.193422317504883,
+      "learning_rate": 9.085714285714286e-07,
+      "loss": 0.7018,
       "step": 160
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 7.74204683303833,
-      "learning_rate": 4.3147208121827415e-06,
-      "loss": 0.6146,
+      "epoch": 0.009714285714285713,
+      "grad_norm": 6.120445728302002,
+      "learning_rate": 9.657142857142857e-07,
+      "loss": 0.6685,
       "step": 170
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 29.357898712158203,
-      "learning_rate": 4.568527918781726e-06,
-      "loss": 0.6861,
+      "epoch": 0.010285714285714285,
+      "grad_norm": 9.130088806152344,
+      "learning_rate": 1.0228571428571429e-06,
+      "loss": 0.6791,
       "step": 180
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 14.954898834228516,
-      "learning_rate": 4.822335025380711e-06,
-      "loss": 0.7809,
+      "epoch": 0.010857142857142857,
+      "grad_norm": 4.139962673187256,
+      "learning_rate": 1.08e-06,
+      "loss": 0.6318,
       "step": 190
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 12.208268165588379,
-      "learning_rate": 5.076142131979695e-06,
-      "loss": 0.5812,
+      "epoch": 0.011428571428571429,
+      "grad_norm": 10.299067497253418,
+      "learning_rate": 1.1371428571428572e-06,
+      "loss": 0.6648,
       "step": 200
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 75.62841033935547,
-      "learning_rate": 5.329949238578681e-06,
-      "loss": 0.7408,
+      "epoch": 0.012,
+      "grad_norm": 8.751834869384766,
+      "learning_rate": 1.1942857142857144e-06,
+      "loss": 0.7039,
       "step": 210
     },
     {
-      "epoch": 0.06,
-      "grad_norm": 17.70330047607422,
-      "learning_rate": 5.583756345177665e-06,
-      "loss": 0.8077,
+      "epoch": 0.012571428571428572,
+      "grad_norm": 6.685519695281982,
+      "learning_rate": 1.2514285714285715e-06,
+      "loss": 0.6488,
       "step": 220
     },
     {
-      "epoch": 0.06,
-      "grad_norm": 26.325668334960938,
-      "learning_rate": 5.83756345177665e-06,
-      "loss": 0.6845,
+      "epoch": 0.013142857142857144,
+      "grad_norm": 12.093446731567383,
+      "learning_rate": 1.3085714285714287e-06,
+      "loss": 0.6134,
       "step": 230
     },
     {
-      "epoch": 0.06,
-      "grad_norm": 30.602115631103516,
-      "learning_rate": 6.091370558375635e-06,
-      "loss": 0.7371,
+      "epoch": 0.013714285714285714,
+      "grad_norm": 10.574247360229492,
+      "learning_rate": 1.3657142857142857e-06,
+      "loss": 0.6786,
       "step": 240
     },
     {
-      "epoch": 0.06,
-      "grad_norm": 17.599937438964844,
-      "learning_rate": 6.34517766497462e-06,
-      "loss": 0.6351,
+      "epoch": 0.014285714285714285,
+      "grad_norm": 12.45152473449707,
+      "learning_rate": 1.422857142857143e-06,
+      "loss": 0.6112,
       "step": 250
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 27.942121505737305,
-      "learning_rate": 6.5989847715736045e-06,
-      "loss": 0.6677,
+      "epoch": 0.014857142857142857,
+      "grad_norm": 20.439712524414062,
+      "learning_rate": 1.48e-06,
+      "loss": 0.6244,
       "step": 260
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 22.718555450439453,
-      "learning_rate": 6.852791878172589e-06,
-      "loss": 0.6294,
+      "epoch": 0.015428571428571429,
+      "grad_norm": 14.572081565856934,
+      "learning_rate": 1.5371428571428574e-06,
+      "loss": 0.6225,
       "step": 270
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 15.279131889343262,
-      "learning_rate": 7.106598984771575e-06,
-      "loss": 0.7051,
+      "epoch": 0.016,
+      "grad_norm": 12.138855934143066,
+      "learning_rate": 1.5942857142857144e-06,
+      "loss": 0.6541,
       "step": 280
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 2.718256950378418,
-      "learning_rate": 7.360406091370559e-06,
-      "loss": 0.2624,
+      "epoch": 0.01657142857142857,
+      "grad_norm": 10.430656433105469,
+      "learning_rate": 1.6514285714285715e-06,
+      "loss": 0.6444,
       "step": 290
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 7.366211414337158,
-      "learning_rate": 7.614213197969543e-06,
-      "loss": 1.0217,
+      "epoch": 0.017142857142857144,
+      "grad_norm": 20.285198211669922,
+      "learning_rate": 1.7085714285714287e-06,
+      "loss": 0.6046,
       "step": 300
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 88.110595703125,
-      "learning_rate": 7.86802030456853e-06,
-      "loss": 0.97,
+      "epoch": 0.017714285714285714,
+      "grad_norm": 14.162407875061035,
+      "learning_rate": 1.7657142857142859e-06,
+      "loss": 0.5698,
       "step": 310
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 8.04466438293457,
-      "learning_rate": 8.121827411167514e-06,
-      "loss": 0.7516,
+      "epoch": 0.018285714285714287,
+      "grad_norm": 12.920883178710938,
+      "learning_rate": 1.8228571428571428e-06,
+      "loss": 0.5905,
       "step": 320
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 9.804755210876465,
-      "learning_rate": 8.375634517766498e-06,
-      "loss": 0.6807,
+      "epoch": 0.018857142857142857,
+      "grad_norm": 13.900638580322266,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.5498,
       "step": 330
     },
     {
-      "epoch": 0.09,
-      "grad_norm": 8.710782051086426,
-      "learning_rate": 8.629441624365483e-06,
-      "loss": 0.431,
+      "epoch": 0.019428571428571427,
+      "grad_norm": 12.032271385192871,
+      "learning_rate": 1.9371428571428576e-06,
+      "loss": 0.5398,
       "step": 340
     },
     {
-      "epoch": 0.09,
-      "grad_norm": 6.989062786102295,
-      "learning_rate": 8.883248730964468e-06,
-      "loss": 0.4437,
+      "epoch": 0.02,
+      "grad_norm": 10.851804733276367,
+      "learning_rate": 1.9942857142857146e-06,
+      "loss": 0.5787,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "eval_accuracy": 0.7104377104377104,
+      "eval_loss": 0.5787273049354553,
+      "eval_runtime": 141.3287,
+      "eval_samples_per_second": 2.101,
+      "eval_steps_per_second": 1.054,
       "step": 350
     },
     {
-      "epoch": 0.09,
-      "grad_norm": 0.23465396463871002,
-      "learning_rate": 9.137055837563452e-06,
-      "loss": 0.6021,
+      "epoch": 1.0005714285714287,
+      "grad_norm": 21.635120391845703,
+      "learning_rate": 2.0514285714285715e-06,
+      "loss": 0.5574,
       "step": 360
     },
     {
-      "epoch": 0.09,
-      "grad_norm": 4.12635612487793,
-      "learning_rate": 9.390862944162438e-06,
-      "loss": 1.1584,
+      "epoch": 1.0011428571428571,
+      "grad_norm": 8.378313064575195,
+      "learning_rate": 2.108571428571429e-06,
+      "loss": 0.6273,
       "step": 370
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 67.80364990234375,
-      "learning_rate": 9.644670050761421e-06,
-      "loss": 1.3129,
+      "epoch": 1.0017142857142858,
+      "grad_norm": 25.1311092376709,
+      "learning_rate": 2.165714285714286e-06,
+      "loss": 0.4157,
       "step": 380
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 1.1369630098342896,
-      "learning_rate": 9.898477157360406e-06,
-      "loss": 0.4667,
+      "epoch": 1.0022857142857142,
+      "grad_norm": 68.14948272705078,
+      "learning_rate": 2.222857142857143e-06,
+      "loss": 0.6408,
       "step": 390
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 0.777511477470398,
-      "learning_rate": 9.983055634001694e-06,
-      "loss": 1.0065,
+      "epoch": 1.002857142857143,
+      "grad_norm": 7.816642761230469,
+      "learning_rate": 2.28e-06,
+      "loss": 0.5284,
       "step": 400
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 5.83087158203125,
-      "learning_rate": 9.95481502400452e-06,
-      "loss": 1.6977,
+      "epoch": 1.0034285714285713,
+      "grad_norm": 23.304162979125977,
+      "learning_rate": 2.337142857142857e-06,
+      "loss": 0.6163,
       "step": 410
     },
     {
-      "epoch": 0.11,
-      "grad_norm": 0.4953118562698364,
-      "learning_rate": 9.926574414007344e-06,
-      "loss": 0.7726,
+      "epoch": 1.004,
+      "grad_norm": 48.6783332824707,
+      "learning_rate": 2.3942857142857145e-06,
+      "loss": 0.5474,
       "step": 420
     },
     {
-      "epoch": 0.11,
-      "grad_norm": 15.068619728088379,
-      "learning_rate": 9.898333804010167e-06,
-      "loss": 0.6931,
+      "epoch": 1.0045714285714287,
+      "grad_norm": 17.72844696044922,
+      "learning_rate": 2.4514285714285715e-06,
+      "loss": 0.5792,
       "step": 430
     },
     {
-      "epoch": 0.11,
-      "grad_norm": 9.647561073303223,
-      "learning_rate": 9.870093194012991e-06,
-      "loss": 0.6504,
+      "epoch": 1.0051428571428571,
+      "grad_norm": 8.830244064331055,
+      "learning_rate": 2.5085714285714285e-06,
+      "loss": 0.5057,
       "step": 440
     },
     {
-      "epoch": 0.11,
-      "grad_norm": 5.142507076263428,
-      "learning_rate": 9.841852584015815e-06,
-      "loss": 0.4246,
+      "epoch": 1.0057142857142858,
+      "grad_norm": 63.260135650634766,
+      "learning_rate": 2.565714285714286e-06,
+      "loss": 0.5394,
       "step": 450
     },
     {
-      "epoch": 0.12,
-      "grad_norm": 44.15083694458008,
-      "learning_rate": 9.81361197401864e-06,
-      "loss": 1.5019,
+      "epoch": 1.0062857142857142,
+      "grad_norm": 3.5656652450561523,
+      "learning_rate": 2.6228571428571432e-06,
+      "loss": 0.4402,
       "step": 460
     },
     {
-      "epoch": 0.12,
-      "grad_norm": 73.28787231445312,
-      "learning_rate": 9.785371364021462e-06,
-      "loss": 0.9851,
+      "epoch": 1.006857142857143,
+      "grad_norm": 31.8543758392334,
+      "learning_rate": 2.68e-06,
+      "loss": 0.5913,
       "step": 470
     },
     {
-      "epoch": 0.12,
-      "grad_norm": 5.464163780212402,
-      "learning_rate": 9.757130754024288e-06,
-      "loss": 0.8729,
+      "epoch": 1.0074285714285713,
+      "grad_norm": 3.7504777908325195,
+      "learning_rate": 2.737142857142857e-06,
+      "loss": 0.579,
       "step": 480
     },
     {
-      "epoch": 0.12,
-      "grad_norm": 0.4673428535461426,
-      "learning_rate": 9.728890144027113e-06,
-      "loss": 1.3793,
+      "epoch": 1.008,
+      "grad_norm": 13.457418441772461,
+      "learning_rate": 2.7942857142857145e-06,
+      "loss": 0.3719,
       "step": 490
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 68.62584686279297,
-      "learning_rate": 9.700649534029935e-06,
-      "loss": 1.7333,
+      "epoch": 1.0085714285714287,
+      "grad_norm": 8.717528343200684,
+      "learning_rate": 2.8514285714285715e-06,
+      "loss": 0.5286,
       "step": 500
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 43.09763717651367,
-      "learning_rate": 9.67240892403276e-06,
-      "loss": 0.9818,
+      "epoch": 1.0091428571428571,
+      "grad_norm": 51.75168991088867,
+      "learning_rate": 2.908571428571429e-06,
+      "loss": 0.5389,
       "step": 510
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 1.3706504106521606,
-      "learning_rate": 9.644168314035584e-06,
-      "loss": 0.6076,
+      "epoch": 1.0097142857142858,
+      "grad_norm": 4.350738048553467,
+      "learning_rate": 2.9657142857142862e-06,
+      "loss": 0.1809,
       "step": 520
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 0.3608866333961487,
-      "learning_rate": 9.615927704038408e-06,
-      "loss": 0.011,
+      "epoch": 1.0102857142857142,
+      "grad_norm": 51.596439361572266,
+      "learning_rate": 3.0228571428571428e-06,
+      "loss": 0.4441,
       "step": 530
     },
     {
-      "epoch": 0.14,
-      "grad_norm": 2.5606987476348877,
-      "learning_rate": 9.587687094041232e-06,
-      "loss": 1.1129,
+      "epoch": 1.010857142857143,
+      "grad_norm": 60.40998840332031,
+      "learning_rate": 3.08e-06,
+      "loss": 0.6152,
       "step": 540
     },
     {
-      "epoch": 0.14,
-      "grad_norm": 0.4630231559276581,
-      "learning_rate": 9.559446484044057e-06,
-      "loss": 0.5555,
+      "epoch": 1.0114285714285713,
+      "grad_norm": 13.335647583007812,
+      "learning_rate": 3.1371428571428575e-06,
+      "loss": 0.3943,
       "step": 550
     },
     {
-      "epoch": 0.14,
-      "grad_norm": 2.497171401977539,
-      "learning_rate": 9.531205874046881e-06,
-      "loss": 0.5851,
+      "epoch": 1.012,
+      "grad_norm": 0.6494603157043457,
+      "learning_rate": 3.194285714285715e-06,
+      "loss": 0.4774,
       "step": 560
     },
     {
-      "epoch": 0.14,
-      "grad_norm": 72.00778198242188,
-      "learning_rate": 9.502965264049704e-06,
-      "loss": 1.1757,
+      "epoch": 1.0125714285714287,
+      "grad_norm": 2.483837127685547,
+      "learning_rate": 3.2514285714285715e-06,
+      "loss": 0.5635,
       "step": 570
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 51.60250473022461,
-      "learning_rate": 9.474724654052528e-06,
-      "loss": 0.5766,
+      "epoch": 1.0131428571428571,
+      "grad_norm": 79.77306365966797,
+      "learning_rate": 3.308571428571429e-06,
+      "loss": 1.0499,
       "step": 580
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 0.111382856965065,
-      "learning_rate": 9.446484044055352e-06,
-      "loss": 0.0039,
+      "epoch": 1.0137142857142858,
+      "grad_norm": 1.0996540784835815,
+      "learning_rate": 3.3657142857142862e-06,
+      "loss": 0.5509,
       "step": 590
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 0.24558886885643005,
-      "learning_rate": 9.418243434058176e-06,
-      "loss": 0.0022,
+      "epoch": 1.0142857142857142,
+      "grad_norm": 48.051490783691406,
+      "learning_rate": 3.422857142857143e-06,
+      "loss": 1.0613,
       "step": 600
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.46412792801856995,
-      "learning_rate": 9.390002824061e-06,
-      "loss": 2.3804,
+      "epoch": 1.014857142857143,
+      "grad_norm": 1.7596064805984497,
+      "learning_rate": 3.48e-06,
+      "loss": 0.2783,
       "step": 610
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.6763355731964111,
-      "learning_rate": 9.361762214063825e-06,
-      "loss": 1.9181,
+      "epoch": 1.0154285714285713,
+      "grad_norm": 0.31064021587371826,
+      "learning_rate": 3.5371428571428575e-06,
+      "loss": 0.649,
       "step": 620
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.18327564001083374,
-      "learning_rate": 9.33352160406665e-06,
-      "loss": 1.68,
+      "epoch": 1.016,
+      "grad_norm": 0.5567959547042847,
+      "learning_rate": 3.5942857142857145e-06,
+      "loss": 0.2289,
       "step": 630
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.0858134999871254,
-      "learning_rate": 9.305280994069472e-06,
-      "loss": 0.4881,
+      "epoch": 1.0165714285714285,
+      "grad_norm": 2.9928736686706543,
+      "learning_rate": 3.651428571428572e-06,
+      "loss": 0.2232,
       "step": 640
     },
     {
-      "epoch": 0.17,
-      "grad_norm": 43.90891647338867,
-      "learning_rate": 9.277040384072296e-06,
-      "loss": 0.9335,
+      "epoch": 1.0171428571428571,
+      "grad_norm": 56.99592208862305,
+      "learning_rate": 3.7085714285714284e-06,
+      "loss": 0.3615,
       "step": 650
     },
     {
-      "epoch": 0.17,
-      "grad_norm": 0.740903913974762,
-      "learning_rate": 9.24879977407512e-06,
-      "loss": 1.8727,
+      "epoch": 1.0177142857142858,
+      "grad_norm": 1.2344051599502563,
+      "learning_rate": 3.7657142857142858e-06,
+      "loss": 1.8052,
       "step": 660
     },
     {
-      "epoch": 0.17,
-      "grad_norm": 41.42824172973633,
-      "learning_rate": 9.220559164077945e-06,
-      "loss": 1.0477,
+      "epoch": 1.0182857142857142,
+      "grad_norm": 44.120845794677734,
+      "learning_rate": 3.822857142857143e-06,
+      "loss": 0.7242,
       "step": 670
     },
     {
-      "epoch": 0.17,
-      "grad_norm": 0.6841148138046265,
-      "learning_rate": 9.192318554080769e-06,
-      "loss": 0.0093,
+      "epoch": 1.018857142857143,
+      "grad_norm": 1.3891487121582031,
+      "learning_rate": 3.88e-06,
+      "loss": 0.737,
       "step": 680
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 0.15574102103710175,
-      "learning_rate": 9.164077944083593e-06,
-      "loss": 0.0061,
+      "epoch": 1.0194285714285714,
+      "grad_norm": 31.26248550415039,
+      "learning_rate": 3.937142857142858e-06,
+      "loss": 0.5545,
       "step": 690
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 30.2320613861084,
-      "learning_rate": 9.135837334086418e-06,
-      "loss": 1.2614,
+      "epoch": 1.02,
+      "grad_norm": 39.20292663574219,
+      "learning_rate": 3.994285714285714e-06,
+      "loss": 0.5175,
+      "step": 700
+    },
+    {
+      "epoch": 1.02,
+      "eval_accuracy": 0.8080808080808081,
+      "eval_loss": 0.7401928901672363,
+      "eval_runtime": 137.8773,
+      "eval_samples_per_second": 2.154,
+      "eval_steps_per_second": 1.081,
       "step": 700
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 0.1653001755475998,
-      "learning_rate": 9.107596724089242e-06,
-      "loss": 1.4186,
+      "epoch": 2.0005714285714284,
+      "grad_norm": 0.3124562203884125,
+      "learning_rate": 4.051428571428572e-06,
+      "loss": 1.2603,
       "step": 710
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 5.986731052398682,
-      "learning_rate": 9.079356114092065e-06,
-      "loss": 0.5271,
+      "epoch": 2.0011428571428573,
+      "grad_norm": 20.57735252380371,
+      "learning_rate": 4.108571428571429e-06,
+      "loss": 0.693,
       "step": 720
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 0.5438697338104248,
-      "learning_rate": 9.051115504094889e-06,
-      "loss": 1.6235,
+      "epoch": 2.001714285714286,
+      "grad_norm": 49.15703582763672,
+      "learning_rate": 4.165714285714287e-06,
+      "loss": 1.4011,
       "step": 730
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 132.91490173339844,
-      "learning_rate": 9.022874894097713e-06,
-      "loss": 0.8793,
+      "epoch": 2.0022857142857142,
+      "grad_norm": 21.79818344116211,
+      "learning_rate": 4.222857142857143e-06,
+      "loss": 1.7675,
       "step": 740
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 222.04197692871094,
-      "learning_rate": 8.994634284100537e-06,
-      "loss": 0.0887,
+      "epoch": 2.0028571428571427,
+      "grad_norm": 8.412939071655273,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.7266,
       "step": 750
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 55.007293701171875,
-      "learning_rate": 8.96639367410336e-06,
-      "loss": 3.5521,
+      "epoch": 2.0034285714285716,
+      "grad_norm": 16.968673706054688,
+      "learning_rate": 4.3371428571428575e-06,
+      "loss": 0.7732,
       "step": 760
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 43.21747589111328,
-      "learning_rate": 8.938153064106186e-06,
-      "loss": 0.8727,
+      "epoch": 2.004,
+      "grad_norm": 70.78853607177734,
+      "learning_rate": 4.3942857142857144e-06,
+      "loss": 0.7847,
       "step": 770
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 52.716983795166016,
-      "learning_rate": 8.90991245410901e-06,
-      "loss": 1.8305,
+      "epoch": 2.0045714285714284,
+      "grad_norm": 6.835943698883057,
+      "learning_rate": 4.451428571428571e-06,
+      "loss": 0.8395,
       "step": 780
     },
     {
-      "epoch": 0.2,
-      "eval_accuracy": 0.6023489932885906,
-      "eval_loss": 1.2154779434204102,
-      "eval_runtime": 323.3556,
-      "eval_samples_per_second": 1.843,
-      "eval_steps_per_second": 1.843,
-      "step": 787
-    },
-    {
-      "epoch": 1.0,
-      "grad_norm": 130.62637329101562,
-      "learning_rate": 8.881671844111833e-06,
-      "loss": 1.5259,
+      "epoch": 2.0051428571428573,
+      "grad_norm": 24.16145133972168,
+      "learning_rate": 4.508571428571429e-06,
+      "loss": 0.531,
       "step": 790
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 28.14907455444336,
-      "learning_rate": 8.853431234114657e-06,
-      "loss": 0.9381,
+      "epoch": 2.005714285714286,
+      "grad_norm": 1.41634202003479,
+      "learning_rate": 4.565714285714286e-06,
+      "loss": 0.4976,
       "step": 800
     },
     {
-      "epoch": 1.01,
-      "grad_norm": 4.458512783050537,
-      "learning_rate": 8.825190624117481e-06,
-      "loss": 0.6209,
+      "epoch": 2.0062857142857142,
+      "grad_norm": 2.940735340118408,
+      "learning_rate": 4.622857142857143e-06,
+      "loss": 0.4175,
       "step": 810
     },
     {
-      "epoch": 1.01,
-      "grad_norm": 2.9990813732147217,
-      "learning_rate": 8.796950014120306e-06,
-      "loss": 0.3972,
+      "epoch": 2.0068571428571427,
+      "grad_norm": 2.9641330242156982,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.2843,
       "step": 820
     },
     {
-      "epoch": 1.01,
-      "grad_norm": 317.2392272949219,
-      "learning_rate": 8.76870940412313e-06,
-      "loss": 1.7995,
+      "epoch": 2.0074285714285716,
+      "grad_norm": 46.827030181884766,
+      "learning_rate": 4.737142857142857e-06,
+      "loss": 1.0676,
       "step": 830
     },
     {
-      "epoch": 1.01,
-      "grad_norm": 0.08616624772548676,
-      "learning_rate": 8.740468794125954e-06,
-      "loss": 0.94,
+      "epoch": 2.008,
+      "grad_norm": 92.20220184326172,
+      "learning_rate": 4.794285714285715e-06,
+      "loss": 0.789,
       "step": 840
     },
     {
-      "epoch": 1.02,
-      "grad_norm": 0.08287107199430466,
-      "learning_rate": 8.712228184128779e-06,
-      "loss": 0.0513,
+      "epoch": 2.0085714285714285,
+      "grad_norm": 181.356201171875,
+      "learning_rate": 4.851428571428572e-06,
+      "loss": 0.6685,
       "step": 850
     },
     {
-      "epoch": 1.02,
-      "grad_norm": 177.66172790527344,
-      "learning_rate": 8.683987574131601e-06,
-      "loss": 0.7323,
+      "epoch": 2.0091428571428573,
+      "grad_norm": 0.2601974904537201,
+      "learning_rate": 4.90857142857143e-06,
+      "loss": 0.6834,
       "step": 860
     },
     {
-      "epoch": 1.02,
-      "grad_norm": 79.82263946533203,
-      "learning_rate": 8.655746964134425e-06,
-      "loss": 1.6152,
+      "epoch": 2.009714285714286,
+      "grad_norm": 123.33201599121094,
+      "learning_rate": 4.965714285714286e-06,
+      "loss": 0.8586,
       "step": 870
     },
     {
-      "epoch": 1.02,
-      "grad_norm": 0.3191066086292267,
-      "learning_rate": 8.62750635413725e-06,
-      "loss": 1.4359,
+      "epoch": 2.0102857142857142,
+      "grad_norm": 143.15835571289062,
+      "learning_rate": 5.0228571428571435e-06,
+      "loss": 0.8327,
       "step": 880
     },
     {
-      "epoch": 1.03,
-      "grad_norm": 172.41537475585938,
-      "learning_rate": 8.599265744140074e-06,
-      "loss": 1.473,
+      "epoch": 2.0108571428571427,
+      "grad_norm": 2.790085792541504,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 1.6946,
       "step": 890
     },
     {
-      "epoch": 1.03,
-      "grad_norm": 9.612916946411133,
-      "learning_rate": 8.571025134142898e-06,
-      "loss": 1.0049,
+      "epoch": 2.0114285714285716,
+      "grad_norm": 50.61666488647461,
+      "learning_rate": 5.1371428571428574e-06,
+      "loss": 0.3813,
       "step": 900
     },
     {
-      "epoch": 1.03,
-      "grad_norm": 27.31446647644043,
-      "learning_rate": 8.542784524145723e-06,
-      "loss": 0.7613,
+      "epoch": 2.012,
+      "grad_norm": 4.110306739807129,
+      "learning_rate": 5.194285714285715e-06,
+      "loss": 0.5825,
       "step": 910
     },
     {
-      "epoch": 1.03,
-      "grad_norm": 0.6966761350631714,
-      "learning_rate": 8.514543914148547e-06,
-      "loss": 0.3961,
+      "epoch": 2.0125714285714285,
+      "grad_norm": 2.2944891452789307,
+      "learning_rate": 5.251428571428571e-06,
+      "loss": 0.9097,
       "step": 920
     },
     {
-      "epoch": 1.04,
-      "grad_norm": 0.1365099400281906,
-      "learning_rate": 8.48630330415137e-06,
-      "loss": 0.1293,
+      "epoch": 2.0131428571428573,
+      "grad_norm": 0.1945544183254242,
+      "learning_rate": 5.308571428571428e-06,
+      "loss": 0.2092,
       "step": 930
     },
     {
-      "epoch": 1.04,
-      "grad_norm": 0.0884900763630867,
-      "learning_rate": 8.458062694154194e-06,
-      "loss": 0.5977,
+      "epoch": 2.013714285714286,
+      "grad_norm": 0.31246262788772583,
+      "learning_rate": 5.365714285714286e-06,
+      "loss": 0.6408,
       "step": 940
     },
     {
-      "epoch": 1.04,
-      "grad_norm": 2.8027291297912598,
-      "learning_rate": 8.429822084157018e-06,
-      "loss": 1.2093,
+      "epoch": 2.0142857142857142,
+      "grad_norm": 0.27790567278862,
+      "learning_rate": 5.422857142857143e-06,
+      "loss": 0.2627,
       "step": 950
     },
     {
-      "epoch": 1.04,
-      "grad_norm": 75.59962463378906,
-      "learning_rate": 8.401581474159842e-06,
-      "loss": 0.4025,
+      "epoch": 2.0148571428571427,
+      "grad_norm": 0.11815852671861649,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.7804,
       "step": 960
     },
     {
-      "epoch": 1.05,
-      "grad_norm": 2.279308795928955,
-      "learning_rate": 8.373340864162667e-06,
-      "loss": 2.1111,
+      "epoch": 2.0154285714285716,
+      "grad_norm": 81.77935028076172,
+      "learning_rate": 5.537142857142858e-06,
+      "loss": 0.9296,
       "step": 970
     },
     {
-      "epoch": 1.05,
-      "grad_norm": 0.3451221287250519,
-      "learning_rate": 8.345100254165491e-06,
-      "loss": 0.8048,
+      "epoch": 2.016,
+      "grad_norm": 0.16326524317264557,
+      "learning_rate": 5.594285714285714e-06,
+      "loss": 0.811,
       "step": 980
     },
     {
-      "epoch": 1.05,
-      "grad_norm": 0.8533955812454224,
-      "learning_rate": 8.316859644168315e-06,
-      "loss": 0.9632,
+      "epoch": 2.0165714285714285,
+      "grad_norm": 0.6108279228210449,
+      "learning_rate": 5.651428571428572e-06,
+      "loss": 0.0187,
       "step": 990
     },
     {
-      "epoch": 1.05,
-      "grad_norm": 55.09623718261719,
-      "learning_rate": 8.28861903417114e-06,
-      "loss": 0.5962,
+      "epoch": 2.0171428571428573,
+      "grad_norm": 0.0926850363612175,
+      "learning_rate": 5.708571428571429e-06,
+      "loss": 1.1172,
       "step": 1000
     },
     {
-      "epoch": 1.06,
-      "grad_norm": 0.5793609023094177,
-      "learning_rate": 8.260378424173962e-06,
-      "loss": 0.8425,
+      "epoch": 2.017714285714286,
+      "grad_norm": 27.080734252929688,
+      "learning_rate": 5.7657142857142865e-06,
+      "loss": 0.2747,
       "step": 1010
     },
     {
-      "epoch": 1.06,
-      "grad_norm": 0.020316464826464653,
-      "learning_rate": 8.232137814176786e-06,
-      "loss": 0.4848,
+      "epoch": 2.0182857142857142,
+      "grad_norm": 32.981544494628906,
+      "learning_rate": 5.8228571428571435e-06,
+      "loss": 0.754,
       "step": 1020
     },
     {
-      "epoch": 1.06,
-      "grad_norm": 0.025828547775745392,
-      "learning_rate": 8.20389720417961e-06,
-      "loss": 0.4477,
+      "epoch": 2.0188571428571427,
+      "grad_norm": 30.81633186340332,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.3928,
       "step": 1030
     },
     {
-      "epoch": 1.06,
-      "grad_norm": 0.02087996155023575,
-      "learning_rate": 8.175656594182435e-06,
-      "loss": 0.391,
+      "epoch": 2.0194285714285716,
+      "grad_norm": 0.45116323232650757,
+      "learning_rate": 5.937142857142858e-06,
+      "loss": 0.3754,
       "step": 1040
     },
     {
-      "epoch": 1.07,
-      "grad_norm": 172.80055236816406,
-      "learning_rate": 8.14741598418526e-06,
-      "loss": 1.9783,
+      "epoch": 2.02,
+      "grad_norm": 0.12361976504325867,
+      "learning_rate": 5.994285714285714e-06,
+      "loss": 0.4062,
+      "step": 1050
+    },
+    {
+      "epoch": 2.02,
+      "eval_accuracy": 0.8282828282828283,
+      "eval_loss": 0.8531781435012817,
+      "eval_runtime": 137.2114,
+      "eval_samples_per_second": 2.165,
+      "eval_steps_per_second": 1.086,
       "step": 1050
     },
     {
-      "epoch": 1.07,
-      "grad_norm": 0.15524566173553467,
-      "learning_rate": 8.119175374188084e-06,
-      "loss": 0.3193,
+      "epoch": 3.0005714285714284,
+      "grad_norm": 0.03133957087993622,
+      "learning_rate": 6.051428571428571e-06,
+      "loss": 0.9268,
       "step": 1060
     },
     {
-      "epoch": 1.07,
-      "grad_norm": 0.34669268131256104,
-      "learning_rate": 8.090934764190908e-06,
-      "loss": 0.5599,
+      "epoch": 3.0011428571428573,
+      "grad_norm": 2.317265748977661,
+      "learning_rate": 6.108571428571429e-06,
+      "loss": 0.602,
       "step": 1070
     },
     {
-      "epoch": 1.07,
-      "grad_norm": 0.14205870032310486,
-      "learning_rate": 8.06269415419373e-06,
-      "loss": 1.6577,
+      "epoch": 3.001714285714286,
+      "grad_norm": 0.7146720290184021,
+      "learning_rate": 6.165714285714286e-06,
+      "loss": 0.2092,
       "step": 1080
     },
     {
-      "epoch": 1.08,
-      "grad_norm": 0.0540863536298275,
-      "learning_rate": 8.034453544196555e-06,
-      "loss": 0.4903,
+      "epoch": 3.0022857142857142,
+      "grad_norm": 0.33893364667892456,
+      "learning_rate": 6.222857142857144e-06,
+      "loss": 1.658,
       "step": 1090
     },
     {
-      "epoch": 1.08,
-      "grad_norm": 3.833791494369507,
-      "learning_rate": 8.006212934199379e-06,
-      "loss": 0.3037,
+      "epoch": 3.0028571428571427,
+      "grad_norm": 45.28017807006836,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.2583,
       "step": 1100
     },
     {
-      "epoch": 1.08,
-      "grad_norm": 37.181514739990234,
-      "learning_rate": 7.977972324202203e-06,
-      "loss": 1.1881,
+      "epoch": 3.0034285714285716,
+      "grad_norm": 0.11307206004858017,
+      "learning_rate": 6.337142857142857e-06,
+      "loss": 0.381,
       "step": 1110
     },
     {
-      "epoch": 1.08,
-      "grad_norm": 66.44300079345703,
-      "learning_rate": 7.949731714205028e-06,
-      "loss": 0.5832,
+      "epoch": 3.004,
+      "grad_norm": 0.7531479597091675,
+      "learning_rate": 6.394285714285715e-06,
+      "loss": 0.5401,
       "step": 1120
     },
     {
-      "epoch": 1.09,
-      "grad_norm": 154.6601104736328,
-      "learning_rate": 7.921491104207852e-06,
-      "loss": 2.519,
+      "epoch": 3.0045714285714284,
+      "grad_norm": 0.3606512248516083,
+      "learning_rate": 6.451428571428572e-06,
+      "loss": 0.8567,
       "step": 1130
     },
     {
-      "epoch": 1.09,
-      "grad_norm": 30.22860336303711,
-      "learning_rate": 7.893250494210676e-06,
-      "loss": 0.9667,
+      "epoch": 3.0051428571428573,
+      "grad_norm": 1.0117719173431396,
+      "learning_rate": 6.5085714285714295e-06,
+      "loss": 0.8051,
       "step": 1140
     },
     {
-      "epoch": 1.09,
-      "grad_norm": 0.13177183270454407,
-      "learning_rate": 7.865009884213499e-06,
-      "loss": 1.0848,
+      "epoch": 3.005714285714286,
+      "grad_norm": 26.470552444458008,
+      "learning_rate": 6.5657142857142865e-06,
+      "loss": 0.8478,
       "step": 1150
     },
     {
-      "epoch": 1.09,
-      "grad_norm": 39.34396743774414,
-      "learning_rate": 7.836769274216323e-06,
-      "loss": 0.5417,
+      "epoch": 3.0062857142857142,
+      "grad_norm": 0.5472680926322937,
+      "learning_rate": 6.6228571428571435e-06,
+      "loss": 0.2333,
       "step": 1160
     },
     {
-      "epoch": 1.1,
-      "grad_norm": 0.5694972276687622,
-      "learning_rate": 7.808528664219147e-06,
-      "loss": 1.0765,
+      "epoch": 3.0068571428571427,
+      "grad_norm": 56.5042839050293,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 1.0095,
       "step": 1170
     },
     {
-      "epoch": 1.1,
-      "grad_norm": 0.16328032314777374,
-      "learning_rate": 7.780288054221972e-06,
-      "loss": 0.01,
+      "epoch": 3.0074285714285716,
+      "grad_norm": 0.2309829443693161,
+      "learning_rate": 6.737142857142857e-06,
+      "loss": 0.495,
       "step": 1180
     },
     {
-      "epoch": 1.1,
-      "grad_norm": 0.15245263278484344,
-      "learning_rate": 7.752047444224796e-06,
-      "loss": 0.6003,
+      "epoch": 3.008,
+      "grad_norm": 0.2706996202468872,
+      "learning_rate": 6.794285714285714e-06,
+      "loss": 0.7256,
       "step": 1190
     },
     {
-      "epoch": 1.1,
-      "grad_norm": 0.11677064746618271,
-      "learning_rate": 7.72380683422762e-06,
-      "loss": 0.0037,
+      "epoch": 3.0085714285714285,
+      "grad_norm": 31.313552856445312,
+      "learning_rate": 6.851428571428572e-06,
+      "loss": 1.3175,
       "step": 1200
     },
     {
-      "epoch": 1.11,
-      "grad_norm": 0.06681229919195175,
-      "learning_rate": 7.695566224230445e-06,
-      "loss": 0.0045,
+      "epoch": 3.0091428571428573,
+      "grad_norm": 37.99323272705078,
+      "learning_rate": 6.908571428571429e-06,
+      "loss": 0.7056,
       "step": 1210
     },
     {
-      "epoch": 1.11,
-      "grad_norm": 0.03700481355190277,
-      "learning_rate": 7.667325614233269e-06,
-      "loss": 0.6897,
+      "epoch": 3.009714285714286,
+      "grad_norm": 34.16461181640625,
+      "learning_rate": 6.965714285714287e-06,
+      "loss": 0.4093,
       "step": 1220
     },
     {
-      "epoch": 1.11,
-      "grad_norm": 0.4864102005958557,
-      "learning_rate": 7.639085004236091e-06,
-      "loss": 2.1175,
+      "epoch": 3.0102857142857142,
+      "grad_norm": 0.45271262526512146,
+      "learning_rate": 7.022857142857144e-06,
+      "loss": 0.3888,
       "step": 1230
     },
     {
-      "epoch": 1.12,
-      "grad_norm": 0.16693036258220673,
-      "learning_rate": 7.610844394238917e-06,
-      "loss": 1.1404,
+      "epoch": 3.0108571428571427,
+      "grad_norm": 15.342572212219238,
+      "learning_rate": 7.08e-06,
+      "loss": 0.6555,
       "step": 1240
     },
     {
-      "epoch": 1.12,
-      "grad_norm": 0.8580232858657837,
-      "learning_rate": 7.58260378424174e-06,
-      "loss": 0.0049,
+      "epoch": 3.0114285714285716,
+      "grad_norm": 209.43014526367188,
+      "learning_rate": 7.137142857142858e-06,
+      "loss": 1.0105,
       "step": 1250
     },
     {
-      "epoch": 1.12,
-      "grad_norm": 0.25582146644592285,
-      "learning_rate": 7.554363174244564e-06,
-      "loss": 0.0043,
+      "epoch": 3.012,
+      "grad_norm": 46.23904800415039,
+      "learning_rate": 7.194285714285715e-06,
+      "loss": 1.2347,
       "step": 1260
     },
     {
-      "epoch": 1.12,
-      "grad_norm": 0.11917036771774292,
-      "learning_rate": 7.526122564247388e-06,
-      "loss": 0.2731,
+      "epoch": 3.0125714285714285,
+      "grad_norm": 120.28668975830078,
+      "learning_rate": 7.251428571428572e-06,
+      "loss": 1.229,
       "step": 1270
     },
     {
-      "epoch": 1.13,
-      "grad_norm": 46.90564727783203,
-      "learning_rate": 7.497881954250212e-06,
-      "loss": 0.7342,
+      "epoch": 3.0131428571428573,
+      "grad_norm": 15.921452522277832,
+      "learning_rate": 7.3085714285714295e-06,
+      "loss": 0.8367,
       "step": 1280
     },
     {
-      "epoch": 1.13,
-      "grad_norm": 24.272871017456055,
-      "learning_rate": 7.469641344253037e-06,
-      "loss": 1.1628,
+      "epoch": 3.013714285714286,
+      "grad_norm": 0.2926520109176636,
+      "learning_rate": 7.365714285714286e-06,
+      "loss": 0.2382,
       "step": 1290
     },
     {
-      "epoch": 1.13,
-      "grad_norm": 0.12642551958560944,
-      "learning_rate": 7.441400734255861e-06,
-      "loss": 0.4991,
+      "epoch": 3.0142857142857142,
+      "grad_norm": 0.1254614293575287,
+      "learning_rate": 7.422857142857144e-06,
+      "loss": 0.857,
       "step": 1300
     },
     {
-      "epoch": 1.13,
-      "grad_norm": 26.394733428955078,
-      "learning_rate": 7.413160124258685e-06,
-      "loss": 1.6511,
+      "epoch": 3.0148571428571427,
+      "grad_norm": 0.10389392822980881,
+      "learning_rate": 7.48e-06,
+      "loss": 0.0128,
       "step": 1310
     },
     {
-      "epoch": 1.14,
-      "grad_norm": 0.33993953466415405,
-      "learning_rate": 7.384919514261508e-06,
-      "loss": 0.4879,
+      "epoch": 3.0154285714285716,
+      "grad_norm": 0.1118193119764328,
+      "learning_rate": 7.537142857142857e-06,
+      "loss": 0.01,
       "step": 1320
     },
     {
-      "epoch": 1.14,
-      "grad_norm": 0.2504374086856842,
-      "learning_rate": 7.356678904264333e-06,
-      "loss": 0.4509,
+      "epoch": 3.016,
+      "grad_norm": 0.023473914712667465,
+      "learning_rate": 7.594285714285715e-06,
+      "loss": 1.3965,
       "step": 1330
     },
     {
-      "epoch": 1.14,
-      "grad_norm": 0.05574103072285652,
-      "learning_rate": 7.328438294267157e-06,
-      "loss": 0.4486,
+      "epoch": 3.0165714285714285,
+      "grad_norm": 79.61312103271484,
+      "learning_rate": 7.651428571428571e-06,
+      "loss": 1.6107,
       "step": 1340
     },
     {
-      "epoch": 1.14,
-      "grad_norm": 0.23328031599521637,
-      "learning_rate": 7.30019768426998e-06,
-      "loss": 0.5054,
+      "epoch": 3.0171428571428573,
+      "grad_norm": 4.563874244689941,
+      "learning_rate": 7.708571428571429e-06,
+      "loss": 0.3567,
       "step": 1350
     },
     {
-      "epoch": 1.15,
-      "grad_norm": 0.010512278415262699,
-      "learning_rate": 7.2719570742728055e-06,
-      "loss": 1.1599,
+      "epoch": 3.017714285714286,
+      "grad_norm": 33.95674514770508,
+      "learning_rate": 7.765714285714287e-06,
+      "loss": 1.1828,
       "step": 1360
     },
     {
-      "epoch": 1.15,
-      "grad_norm": 0.042285043746232986,
-      "learning_rate": 7.243716464275629e-06,
-      "loss": 0.4782,
+      "epoch": 3.0182857142857142,
+      "grad_norm": 36.74417495727539,
+      "learning_rate": 7.822857142857143e-06,
+      "loss": 0.9229,
       "step": 1370
     },
     {
-      "epoch": 1.15,
-      "grad_norm": 0.012322720140218735,
-      "learning_rate": 7.215475854278453e-06,
-      "loss": 1.0997,
+      "epoch": 3.0188571428571427,
+      "grad_norm": 63.20321273803711,
+      "learning_rate": 7.88e-06,
+      "loss": 0.3935,
       "step": 1380
     },
     {
-      "epoch": 1.15,
-      "grad_norm": 0.014798709191381931,
-      "learning_rate": 7.1872352442812775e-06,
-      "loss": 1.2561,
+      "epoch": 3.0194285714285716,
+      "grad_norm": 29.39291000366211,
+      "learning_rate": 7.937142857142857e-06,
+      "loss": 0.7986,
       "step": 1390
     },
     {
-      "epoch": 1.16,
-      "grad_norm": 0.45606276392936707,
-      "learning_rate": 7.158994634284101e-06,
-      "loss": 1.2503,
+      "epoch": 3.02,
+      "grad_norm": 29.3126277923584,
+      "learning_rate": 7.994285714285715e-06,
+      "loss": 0.7962,
+      "step": 1400
+    },
+    {
+      "epoch": 3.02,
+      "eval_accuracy": 0.8114478114478114,
+      "eval_loss": 0.7183520197868347,
+      "eval_runtime": 137.6694,
+      "eval_samples_per_second": 2.157,
+      "eval_steps_per_second": 1.082,
       "step": 1400
     },
     {
-      "epoch": 1.16,
-      "grad_norm": 0.04647226259112358,
-      "learning_rate": 7.130754024286925e-06,
-      "loss": 1.2027,
+      "epoch": 4.000571428571429,
+      "grad_norm": 2.059375047683716,
+      "learning_rate": 8.051428571428573e-06,
+      "loss": 0.7828,
       "step": 1410
     },
     {
-      "epoch": 1.16,
-      "grad_norm": 0.34346136450767517,
-      "learning_rate": 7.102513414289749e-06,
-      "loss": 0.0049,
+      "epoch": 4.001142857142857,
+      "grad_norm": 47.22036361694336,
+      "learning_rate": 8.108571428571429e-06,
+      "loss": 0.9877,
       "step": 1420
     },
     {
-      "epoch": 1.16,
-      "grad_norm": 23.320772171020508,
-      "learning_rate": 7.074272804292574e-06,
-      "loss": 0.8748,
+      "epoch": 4.001714285714286,
+      "grad_norm": 0.17464332282543182,
+      "learning_rate": 8.165714285714286e-06,
+      "loss": 0.6497,
       "step": 1430
     },
     {
-      "epoch": 1.17,
-      "grad_norm": 0.4166504144668579,
-      "learning_rate": 7.0460321942953965e-06,
-      "loss": 0.8503,
+      "epoch": 4.002285714285715,
+      "grad_norm": 1.4610955715179443,
+      "learning_rate": 8.222857142857144e-06,
+      "loss": 0.5103,
       "step": 1440
     },
     {
-      "epoch": 1.17,
-      "grad_norm": 0.028790142387151718,
-      "learning_rate": 7.017791584298222e-06,
-      "loss": 0.755,
+      "epoch": 4.002857142857143,
+      "grad_norm": 0.10568273067474365,
+      "learning_rate": 8.28e-06,
+      "loss": 0.0078,
       "step": 1450
     },
     {
-      "epoch": 1.17,
-      "grad_norm": 0.10875601321458817,
-      "learning_rate": 6.989550974301046e-06,
-      "loss": 1.2674,
+      "epoch": 4.003428571428572,
+      "grad_norm": 0.04568742215633392,
+      "learning_rate": 8.337142857142858e-06,
+      "loss": 0.0033,
       "step": 1460
     },
     {
-      "epoch": 1.17,
-      "grad_norm": 0.24065102636814117,
-      "learning_rate": 6.961310364303869e-06,
-      "loss": 1.0944,
+      "epoch": 4.004,
+      "grad_norm": 0.1109299287199974,
+      "learning_rate": 8.394285714285714e-06,
+      "loss": 0.8707,
       "step": 1470
     },
     {
-      "epoch": 1.18,
-      "grad_norm": 0.17569658160209656,
-      "learning_rate": 6.933069754306694e-06,
-      "loss": 1.0381,
+      "epoch": 4.0045714285714284,
+      "grad_norm": 0.08762586861848831,
+      "learning_rate": 8.451428571428572e-06,
+      "loss": 0.7401,
       "step": 1480
     },
     {
-      "epoch": 1.18,
-      "grad_norm": 130.54307556152344,
-      "learning_rate": 6.904829144309517e-06,
-      "loss": 1.0751,
+      "epoch": 4.005142857142857,
+      "grad_norm": 0.16827107965946198,
+      "learning_rate": 8.50857142857143e-06,
+      "loss": 0.6692,
       "step": 1490
     },
     {
-      "epoch": 1.18,
-      "grad_norm": 0.1745329350233078,
-      "learning_rate": 6.876588534312341e-06,
-      "loss": 0.97,
+      "epoch": 4.005714285714285,
+      "grad_norm": 0.9331682324409485,
+      "learning_rate": 8.565714285714286e-06,
+      "loss": 0.005,
       "step": 1500
     },
     {
-      "epoch": 1.18,
-      "grad_norm": 24.403528213500977,
-      "learning_rate": 6.8483479243151665e-06,
-      "loss": 0.8853,
+      "epoch": 4.006285714285714,
+      "grad_norm": 0.14198555052280426,
+      "learning_rate": 8.622857142857144e-06,
+      "loss": 0.0065,
       "step": 1510
     },
     {
-      "epoch": 1.19,
-      "grad_norm": 3.1027746200561523,
-      "learning_rate": 6.82010731431799e-06,
-      "loss": 0.6651,
+      "epoch": 4.006857142857143,
+      "grad_norm": 0.1513282060623169,
+      "learning_rate": 8.68e-06,
+      "loss": 0.5806,
       "step": 1520
     },
     {
-      "epoch": 1.19,
-      "grad_norm": 5.846127033233643,
-      "learning_rate": 6.791866704320814e-06,
-      "loss": 0.4536,
+      "epoch": 4.007428571428571,
+      "grad_norm": 0.06693244725465775,
+      "learning_rate": 8.737142857142858e-06,
+      "loss": 0.7874,
       "step": 1530
     },
     {
-      "epoch": 1.19,
-      "grad_norm": 0.017441660165786743,
-      "learning_rate": 6.763626094323638e-06,
-      "loss": 1.0821,
+      "epoch": 4.008,
+      "grad_norm": 0.12991634011268616,
+      "learning_rate": 8.794285714285716e-06,
+      "loss": 0.2847,
       "step": 1540
     },
     {
-      "epoch": 1.19,
-      "grad_norm": 0.20883318781852722,
-      "learning_rate": 6.735385484326462e-06,
-      "loss": 0.0086,
+      "epoch": 4.008571428571429,
+      "grad_norm": 19.66016960144043,
+      "learning_rate": 8.851428571428572e-06,
+      "loss": 1.0498,
       "step": 1550
     },
     {
-      "epoch": 1.2,
-      "grad_norm": 0.15525928139686584,
-      "learning_rate": 6.707144874329285e-06,
-      "loss": 0.5036,
+      "epoch": 4.009142857142857,
+      "grad_norm": 2.3184783458709717,
+      "learning_rate": 8.90857142857143e-06,
+      "loss": 0.2343,
       "step": 1560
     },
     {
-      "epoch": 1.2,
-      "grad_norm": 0.15142841637134552,
-      "learning_rate": 6.67890426433211e-06,
-      "loss": 0.7465,
+      "epoch": 4.009714285714286,
+      "grad_norm": 0.2929948568344116,
+      "learning_rate": 8.965714285714287e-06,
+      "loss": 0.5086,
       "step": 1570
     },
     {
-      "epoch": 1.2,
-      "eval_accuracy": 0.8338926174496645,
-      "eval_loss": 0.7484979629516602,
-      "eval_runtime": 324.9358,
-      "eval_samples_per_second": 1.834,
-      "eval_steps_per_second": 1.834,
-      "step": 1574
-    },
-    {
-      "epoch": 2.0,
-      "grad_norm": 0.13457632064819336,
-      "learning_rate": 6.650663654334935e-06,
-      "loss": 1.7798,
+      "epoch": 4.010285714285715,
+      "grad_norm": 22.23305892944336,
+      "learning_rate": 9.022857142857143e-06,
+      "loss": 0.5249,
       "step": 1580
     },
     {
-      "epoch": 2.0,
-      "grad_norm": 0.16015511751174927,
-      "learning_rate": 6.622423044337758e-06,
-      "loss": 0.3209,
+      "epoch": 4.010857142857143,
+      "grad_norm": 15.748601913452148,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 1.3767,
       "step": 1590
     },
     {
-      "epoch": 2.01,
-      "grad_norm": 0.16651424765586853,
-      "learning_rate": 6.5941824343405826e-06,
-      "loss": 1.5177,
+      "epoch": 4.011428571428572,
+      "grad_norm": 0.976803183555603,
+      "learning_rate": 9.137142857142857e-06,
+      "loss": 0.4042,
       "step": 1600
     },
     {
-      "epoch": 2.01,
-      "grad_norm": 0.35417357087135315,
-      "learning_rate": 6.565941824343406e-06,
-      "loss": 0.515,
+      "epoch": 4.012,
+      "grad_norm": 0.44382333755493164,
+      "learning_rate": 9.194285714285715e-06,
+      "loss": 0.4972,
       "step": 1610
     },
     {
-      "epoch": 2.01,
-      "grad_norm": 0.05276577174663544,
-      "learning_rate": 6.53770121434623e-06,
-      "loss": 1.9682,
+      "epoch": 4.0125714285714285,
+      "grad_norm": 27.20537757873535,
+      "learning_rate": 9.251428571428573e-06,
+      "loss": 0.9029,
       "step": 1620
     },
     {
-      "epoch": 2.01,
-      "grad_norm": 0.23119059205055237,
-      "learning_rate": 6.5094606043490546e-06,
-      "loss": 0.4395,
+      "epoch": 4.013142857142857,
+      "grad_norm": 14.854856491088867,
+      "learning_rate": 9.308571428571429e-06,
+      "loss": 1.1477,
       "step": 1630
     },
     {
-      "epoch": 2.02,
-      "grad_norm": 0.4055440425872803,
-      "learning_rate": 6.481219994351878e-06,
-      "loss": 0.0106,
+      "epoch": 4.013714285714285,
+      "grad_norm": 14.011804580688477,
+      "learning_rate": 9.365714285714287e-06,
+      "loss": 0.4896,
       "step": 1640
     },
     {
-      "epoch": 2.02,
-      "grad_norm": 0.15100878477096558,
-      "learning_rate": 6.452979384354703e-06,
-      "loss": 0.0059,
+      "epoch": 4.014285714285714,
+      "grad_norm": 0.38275304436683655,
+      "learning_rate": 9.422857142857143e-06,
+      "loss": 0.8805,
       "step": 1650
     },
     {
-      "epoch": 2.02,
-      "grad_norm": 0.14264227449893951,
-      "learning_rate": 6.424738774357527e-06,
-      "loss": 1.2383,
+      "epoch": 4.014857142857143,
+      "grad_norm": 28.851638793945312,
+      "learning_rate": 9.48e-06,
+      "loss": 1.3966,
       "step": 1660
     },
     {
-      "epoch": 2.02,
-      "grad_norm": 0.16001996397972107,
-      "learning_rate": 6.396498164360351e-06,
-      "loss": 1.02,
+      "epoch": 4.015428571428571,
+      "grad_norm": 35.203712463378906,
+      "learning_rate": 9.537142857142859e-06,
+      "loss": 0.613,
       "step": 1670
     },
     {
-      "epoch": 2.03,
-      "grad_norm": 0.15914620459079742,
-      "learning_rate": 6.368257554363175e-06,
-      "loss": 0.005,
+      "epoch": 4.016,
+      "grad_norm": 0.1584346443414688,
+      "learning_rate": 9.594285714285715e-06,
+      "loss": 0.2728,
       "step": 1680
     },
     {
-      "epoch": 2.03,
-      "grad_norm": 0.11949492990970612,
-      "learning_rate": 6.340016944365999e-06,
-      "loss": 0.5224,
+      "epoch": 4.016571428571429,
+      "grad_norm": 0.10078814625740051,
+      "learning_rate": 9.651428571428572e-06,
+      "loss": 0.4286,
       "step": 1690
     },
     {
-      "epoch": 2.03,
-      "grad_norm": 68.48541259765625,
-      "learning_rate": 6.311776334368823e-06,
-      "loss": 0.7585,
+      "epoch": 4.017142857142857,
+      "grad_norm": 13.510937690734863,
+      "learning_rate": 9.70857142857143e-06,
+      "loss": 0.6424,
       "step": 1700
     },
     {
-      "epoch": 2.03,
-      "grad_norm": 78.09130859375,
-      "learning_rate": 6.283535724371646e-06,
-      "loss": 1.5627,
+      "epoch": 4.017714285714286,
+      "grad_norm": 17.916301727294922,
+      "learning_rate": 9.765714285714286e-06,
+      "loss": 0.993,
       "step": 1710
     },
     {
-      "epoch": 2.04,
-      "grad_norm": 0.06310901045799255,
-      "learning_rate": 6.2552951143744715e-06,
-      "loss": 1.0456,
+      "epoch": 4.018285714285715,
+      "grad_norm": 35.681949615478516,
+      "learning_rate": 9.822857142857144e-06,
+      "loss": 1.2221,
       "step": 1720
     },
     {
-      "epoch": 2.04,
-      "grad_norm": 0.7593610882759094,
-      "learning_rate": 6.227054504377295e-06,
-      "loss": 0.6762,
+      "epoch": 4.018857142857143,
+      "grad_norm": 15.966533660888672,
+      "learning_rate": 9.88e-06,
+      "loss": 0.7781,
       "step": 1730
     },
     {
-      "epoch": 2.04,
-      "grad_norm": 0.1845823973417282,
-      "learning_rate": 6.198813894380119e-06,
-      "loss": 0.9624,
+      "epoch": 4.019428571428572,
+      "grad_norm": 6.339288234710693,
+      "learning_rate": 9.937142857142858e-06,
+      "loss": 0.4475,
       "step": 1740
     },
     {
-      "epoch": 2.04,
-      "grad_norm": 0.49727919697761536,
-      "learning_rate": 6.1705732843829435e-06,
-      "loss": 1.1605,
+      "epoch": 4.02,
+      "grad_norm": 145.03355407714844,
+      "learning_rate": 9.994285714285716e-06,
+      "loss": 0.8225,
+      "step": 1750
+    },
+    {
+      "epoch": 4.02,
+      "eval_accuracy": 0.5656565656565656,
+      "eval_loss": 1.686766266822815,
+      "eval_runtime": 137.779,
+      "eval_samples_per_second": 2.156,
+      "eval_steps_per_second": 1.081,
       "step": 1750
     },
     {
-      "epoch": 2.05,
-      "grad_norm": 25.680139541625977,
-      "learning_rate": 6.142332674385767e-06,
-      "loss": 1.5904,
+      "epoch": 5.000571428571429,
+      "grad_norm": 9.777188301086426,
+      "learning_rate": 9.994285714285716e-06,
+      "loss": 1.4855,
       "step": 1760
     },
     {
-      "epoch": 2.05,
-      "grad_norm": 0.20278450846672058,
-      "learning_rate": 6.114092064388591e-06,
-      "loss": 0.3835,
+      "epoch": 5.001142857142857,
+      "grad_norm": 39.02730178833008,
+      "learning_rate": 9.987936507936509e-06,
+      "loss": 1.214,
       "step": 1770
     },
     {
-      "epoch": 2.05,
-      "grad_norm": 0.2322162538766861,
-      "learning_rate": 6.085851454391415e-06,
-      "loss": 0.9161,
+      "epoch": 5.001714285714286,
+      "grad_norm": 40.402408599853516,
+      "learning_rate": 9.981587301587303e-06,
+      "loss": 0.5619,
       "step": 1780
     },
     {
-      "epoch": 2.05,
-      "grad_norm": 0.2889300286769867,
-      "learning_rate": 6.05761084439424e-06,
-      "loss": 0.2446,
+      "epoch": 5.002285714285715,
+      "grad_norm": 13.327921867370605,
+      "learning_rate": 9.975238095238095e-06,
+      "loss": 0.6943,
       "step": 1790
     },
     {
-      "epoch": 2.06,
-      "grad_norm": 0.21073293685913086,
-      "learning_rate": 6.029370234397064e-06,
-      "loss": 0.6002,
+      "epoch": 5.002857142857143,
+      "grad_norm": 7.069613456726074,
+      "learning_rate": 9.96888888888889e-06,
+      "loss": 0.5466,
       "step": 1800
     },
     {
-      "epoch": 2.06,
-      "grad_norm": 0.2775285243988037,
-      "learning_rate": 6.0011296243998876e-06,
-      "loss": 0.0199,
+      "epoch": 5.003428571428572,
+      "grad_norm": 81.00372314453125,
+      "learning_rate": 9.962539682539684e-06,
+      "loss": 1.0447,
       "step": 1810
     },
     {
-      "epoch": 2.06,
-      "grad_norm": 0.13404442369937897,
-      "learning_rate": 5.972889014402712e-06,
-      "loss": 0.0098,
+      "epoch": 5.004,
+      "grad_norm": 0.9917150735855103,
+      "learning_rate": 9.956190476190477e-06,
+      "loss": 0.5806,
       "step": 1820
     },
     {
-      "epoch": 2.07,
-      "grad_norm": 0.19267532229423523,
-      "learning_rate": 5.944648404405535e-06,
-      "loss": 2.1965,
+      "epoch": 5.0045714285714284,
+      "grad_norm": 0.22874107956886292,
+      "learning_rate": 9.949841269841271e-06,
+      "loss": 0.4238,
       "step": 1830
     },
     {
-      "epoch": 2.07,
-      "grad_norm": 0.314384788274765,
-      "learning_rate": 5.9164077944083596e-06,
-      "loss": 1.875,
+      "epoch": 5.005142857142857,
+      "grad_norm": 13.64168643951416,
+      "learning_rate": 9.943492063492064e-06,
+      "loss": 0.7042,
       "step": 1840
     },
     {
-      "epoch": 2.07,
-      "grad_norm": 0.410067081451416,
-      "learning_rate": 5.888167184411185e-06,
-      "loss": 0.01,
+      "epoch": 5.005714285714285,
+      "grad_norm": 0.5768861174583435,
+      "learning_rate": 9.937142857142858e-06,
+      "loss": 0.4288,
       "step": 1850
     },
     {
-      "epoch": 2.07,
-      "grad_norm": 26.344202041625977,
-      "learning_rate": 5.859926574414007e-06,
-      "loss": 0.8549,
+      "epoch": 5.006285714285714,
+      "grad_norm": 31.998321533203125,
+      "learning_rate": 9.930793650793652e-06,
+      "loss": 1.2366,
       "step": 1860
     },
     {
-      "epoch": 2.08,
-      "grad_norm": 0.2887156307697296,
-      "learning_rate": 5.8316859644168324e-06,
-      "loss": 0.0071,
+      "epoch": 5.006857142857143,
+      "grad_norm": 13.20121955871582,
+      "learning_rate": 9.924444444444445e-06,
+      "loss": 0.6062,
       "step": 1870
     },
     {
-      "epoch": 2.08,
-      "grad_norm": 0.2524619698524475,
-      "learning_rate": 5.803445354419656e-06,
-      "loss": 0.4629,
+      "epoch": 5.007428571428571,
+      "grad_norm": 0.0679963082075119,
+      "learning_rate": 9.91809523809524e-06,
+      "loss": 0.5082,
       "step": 1880
     },
     {
-      "epoch": 2.08,
-      "grad_norm": 0.22347721457481384,
-      "learning_rate": 5.77520474442248e-06,
-      "loss": 0.454,
+      "epoch": 5.008,
+      "grad_norm": 0.2697165608406067,
+      "learning_rate": 9.911746031746032e-06,
+      "loss": 0.8584,
       "step": 1890
     },
     {
-      "epoch": 2.08,
-      "grad_norm": 0.19489675760269165,
-      "learning_rate": 5.746964134425304e-06,
-      "loss": 0.0037,
+      "epoch": 5.008571428571429,
+      "grad_norm": 12.50837230682373,
+      "learning_rate": 9.905396825396826e-06,
+      "loss": 0.5164,
       "step": 1900
     },
     {
-      "epoch": 2.09,
-      "grad_norm": 25.353939056396484,
-      "learning_rate": 5.718723524428128e-06,
-      "loss": 0.5369,
+      "epoch": 5.009142857142857,
+      "grad_norm": 28.091419219970703,
+      "learning_rate": 9.89904761904762e-06,
+      "loss": 0.9162,
       "step": 1910
     },
     {
-      "epoch": 2.09,
-      "grad_norm": 0.01046634092926979,
-      "learning_rate": 5.690482914430952e-06,
-      "loss": 0.0017,
+      "epoch": 5.009714285714286,
+      "grad_norm": 0.1433819681406021,
+      "learning_rate": 9.892698412698413e-06,
+      "loss": 0.39,
       "step": 1920
     },
     {
-      "epoch": 2.09,
-      "grad_norm": 0.016024693846702576,
-      "learning_rate": 5.662242304433776e-06,
-      "loss": 0.6874,
+      "epoch": 5.010285714285715,
+      "grad_norm": 0.6932072043418884,
+      "learning_rate": 9.886349206349208e-06,
+      "loss": 0.5809,
       "step": 1930
     },
     {
-      "epoch": 2.09,
-      "grad_norm": 0.13052043318748474,
-      "learning_rate": 5.634001694436601e-06,
-      "loss": 0.5459,
+      "epoch": 5.010857142857143,
+      "grad_norm": 0.5466342568397522,
+      "learning_rate": 9.88e-06,
+      "loss": 0.5664,
       "step": 1940
     },
     {
-      "epoch": 2.1,
-      "grad_norm": 22.755247116088867,
-      "learning_rate": 5.605761084439424e-06,
-      "loss": 1.063,
+      "epoch": 5.011428571428572,
+      "grad_norm": 10.084834098815918,
+      "learning_rate": 9.873650793650795e-06,
+      "loss": 0.4083,
       "step": 1950
     },
     {
-      "epoch": 2.1,
-      "grad_norm": 0.017313385382294655,
-      "learning_rate": 5.5775204744422485e-06,
-      "loss": 1.3419,
+      "epoch": 5.012,
+      "grad_norm": 0.201524555683136,
+      "learning_rate": 9.867301587301587e-06,
+      "loss": 0.9247,
       "step": 1960
     },
     {
-      "epoch": 2.1,
-      "grad_norm": 0.27627086639404297,
-      "learning_rate": 5.549279864445073e-06,
-      "loss": 0.7342,
+      "epoch": 5.0125714285714285,
+      "grad_norm": 49.69109344482422,
+      "learning_rate": 9.860952380952382e-06,
+      "loss": 0.4362,
       "step": 1970
     },
     {
-      "epoch": 2.1,
-      "grad_norm": 0.02726130001246929,
-      "learning_rate": 5.521039254447896e-06,
-      "loss": 1.1784,
+      "epoch": 5.013142857142857,
+      "grad_norm": 2.8617489337921143,
+      "learning_rate": 9.854603174603176e-06,
+      "loss": 0.009,
       "step": 1980
     },
     {
-      "epoch": 2.11,
-      "grad_norm": 0.3983820378780365,
-      "learning_rate": 5.4927986444507205e-06,
-      "loss": 0.018,
+      "epoch": 5.013714285714285,
+      "grad_norm": 0.3022560775279999,
+      "learning_rate": 9.848253968253969e-06,
+      "loss": 0.849,
       "step": 1990
     },
     {
-      "epoch": 2.11,
-      "grad_norm": 0.13953474164009094,
-      "learning_rate": 5.464558034453544e-06,
-      "loss": 0.0044,
+      "epoch": 5.014285714285714,
+      "grad_norm": 0.2493607997894287,
+      "learning_rate": 9.841904761904763e-06,
+      "loss": 0.438,
       "step": 2000
     },
     {
-      "epoch": 2.11,
-      "grad_norm": 0.04043055698275566,
-      "learning_rate": 5.436317424456369e-06,
-      "loss": 0.5193,
+      "epoch": 5.014857142857143,
+      "grad_norm": 0.21475115418434143,
+      "learning_rate": 9.835555555555556e-06,
+      "loss": 0.846,
       "step": 2010
     },
     {
-      "epoch": 2.11,
-      "grad_norm": 22.795866012573242,
-      "learning_rate": 5.408076814459193e-06,
-      "loss": 0.823,
+      "epoch": 5.015428571428571,
+      "grad_norm": 0.6425090432167053,
+      "learning_rate": 9.82920634920635e-06,
+      "loss": 0.5353,
       "step": 2020
     },
     {
-      "epoch": 2.12,
-      "grad_norm": 23.257352828979492,
-      "learning_rate": 5.379836204462017e-06,
-      "loss": 1.0949,
+      "epoch": 5.016,
+      "grad_norm": 0.29937419295310974,
+      "learning_rate": 9.822857142857144e-06,
+      "loss": 0.8212,
       "step": 2030
     },
     {
-      "epoch": 2.12,
-      "grad_norm": 22.406156539916992,
-      "learning_rate": 5.351595594464841e-06,
-      "loss": 2.1405,
+      "epoch": 5.016571428571429,
+      "grad_norm": 5.154245853424072,
+      "learning_rate": 9.816507936507937e-06,
+      "loss": 0.9711,
       "step": 2040
     },
     {
-      "epoch": 2.12,
-      "grad_norm": 0.4050818979740143,
-      "learning_rate": 5.323354984467665e-06,
-      "loss": 1.4621,
+      "epoch": 5.017142857142857,
+      "grad_norm": 0.5371858477592468,
+      "learning_rate": 9.810158730158731e-06,
+      "loss": 0.4229,
       "step": 2050
     },
     {
-      "epoch": 2.12,
-      "grad_norm": 0.4241192638874054,
-      "learning_rate": 5.295114374470489e-06,
-      "loss": 0.4477,
+      "epoch": 5.017714285714286,
+      "grad_norm": 11.923803329467773,
+      "learning_rate": 9.803809523809524e-06,
+      "loss": 1.5601,
       "step": 2060
     },
     {
-      "epoch": 2.13,
-      "grad_norm": 0.30280259251594543,
-      "learning_rate": 5.266873764473312e-06,
-      "loss": 0.0106,
+      "epoch": 5.018285714285715,
+      "grad_norm": 25.41405487060547,
+      "learning_rate": 9.797460317460318e-06,
+      "loss": 0.374,
       "step": 2070
     },
     {
-      "epoch": 2.13,
-      "grad_norm": 0.18719686567783356,
-      "learning_rate": 5.2386331544761374e-06,
-      "loss": 0.4244,
+      "epoch": 5.018857142857143,
+      "grad_norm": 12.134407043457031,
+      "learning_rate": 9.791111111111112e-06,
+      "loss": 1.0097,
       "step": 2080
     },
     {
-      "epoch": 2.13,
-      "grad_norm": 0.04479646682739258,
-      "learning_rate": 5.210392544478962e-06,
-      "loss": 0.6732,
+      "epoch": 5.019428571428572,
+      "grad_norm": 191.27777099609375,
+      "learning_rate": 9.784761904761905e-06,
+      "loss": 0.8402,
       "step": 2090
     },
     {
-      "epoch": 2.13,
-      "grad_norm": 22.563093185424805,
-      "learning_rate": 5.182151934481785e-06,
-      "loss": 0.5233,
+      "epoch": 5.02,
+      "grad_norm": 48.240779876708984,
+      "learning_rate": 9.7784126984127e-06,
+      "loss": 0.724,
+      "step": 2100
+    },
+    {
+      "epoch": 5.02,
+      "eval_accuracy": 0.7508417508417509,
+      "eval_loss": 1.0066299438476562,
+      "eval_runtime": 136.2857,
+      "eval_samples_per_second": 2.179,
+      "eval_steps_per_second": 1.093,
       "step": 2100
     },
     {
-      "epoch": 2.14,
-      "grad_norm": 0.0441102497279644,
-      "learning_rate": 5.1539113244846095e-06,
-      "loss": 1.7924,
+      "epoch": 6.000571428571429,
+      "grad_norm": 16.156768798828125,
+      "learning_rate": 9.772063492063492e-06,
+      "loss": 0.4888,
       "step": 2110
     },
     {
-      "epoch": 2.14,
-      "grad_norm": 0.046370625495910645,
-      "learning_rate": 5.125670714487433e-06,
-      "loss": 0.0049,
+      "epoch": 6.001142857142857,
+      "grad_norm": 3.711266040802002,
+      "learning_rate": 9.765714285714286e-06,
+      "loss": 0.92,
       "step": 2120
     },
     {
-      "epoch": 2.14,
-      "grad_norm": 1.073113203048706,
-      "learning_rate": 5.097430104490257e-06,
-      "loss": 0.5907,
+      "epoch": 6.001714285714286,
+      "grad_norm": 13.800188064575195,
+      "learning_rate": 9.75936507936508e-06,
+      "loss": 1.3366,
       "step": 2130
     },
     {
-      "epoch": 2.14,
-      "grad_norm": 0.08737409859895706,
-      "learning_rate": 5.069189494493082e-06,
-      "loss": 1.5069,
+      "epoch": 6.002285714285715,
+      "grad_norm": 42.781002044677734,
+      "learning_rate": 9.753015873015873e-06,
+      "loss": 0.4288,
       "step": 2140
     },
     {
-      "epoch": 2.15,
-      "grad_norm": 0.058251939713954926,
-      "learning_rate": 5.040948884495906e-06,
-      "loss": 0.0395,
+      "epoch": 6.002857142857143,
+      "grad_norm": 35.87137985229492,
+      "learning_rate": 9.746666666666668e-06,
+      "loss": 0.3171,
       "step": 2150
     },
     {
-      "epoch": 2.15,
-      "grad_norm": 39.82133102416992,
-      "learning_rate": 5.01270827449873e-06,
-      "loss": 0.01,
+      "epoch": 6.003428571428572,
+      "grad_norm": 0.34640470147132874,
+      "learning_rate": 9.74031746031746e-06,
+      "loss": 0.6315,
       "step": 2160
     },
     {
-      "epoch": 2.15,
-      "grad_norm": 0.18180860579013824,
-      "learning_rate": 4.9844676645015535e-06,
-      "loss": 0.0042,
+      "epoch": 6.004,
+      "grad_norm": 0.15693581104278564,
+      "learning_rate": 9.733968253968255e-06,
+      "loss": 0.9196,
       "step": 2170
     },
     {
-      "epoch": 2.15,
-      "grad_norm": 0.10672131180763245,
-      "learning_rate": 4.956227054504378e-06,
-      "loss": 0.01,
+      "epoch": 6.0045714285714284,
+      "grad_norm": 0.8390889763832092,
+      "learning_rate": 9.727619047619047e-06,
+      "loss": 0.5796,
       "step": 2180
     },
     {
-      "epoch": 2.16,
-      "grad_norm": 28.28722381591797,
-      "learning_rate": 4.927986444507202e-06,
-      "loss": 1.221,
+      "epoch": 6.005142857142857,
+      "grad_norm": 0.38496437668800354,
+      "learning_rate": 9.721269841269843e-06,
+      "loss": 0.7525,
       "step": 2190
     },
     {
-      "epoch": 2.16,
-      "grad_norm": 0.12183073163032532,
-      "learning_rate": 4.8997458345100255e-06,
-      "loss": 1.1702,
+      "epoch": 6.005714285714285,
+      "grad_norm": 0.7874132990837097,
+      "learning_rate": 9.714920634920636e-06,
+      "loss": 0.5311,
       "step": 2200
     },
     {
-      "epoch": 2.16,
-      "grad_norm": 0.17091098427772522,
-      "learning_rate": 4.87150522451285e-06,
-      "loss": 1.9466,
+      "epoch": 6.006285714285714,
+      "grad_norm": 0.2642306983470917,
+      "learning_rate": 9.70857142857143e-06,
+      "loss": 0.3915,
       "step": 2210
     },
     {
-      "epoch": 2.16,
-      "grad_norm": 24.53551483154297,
-      "learning_rate": 4.843264614515674e-06,
-      "loss": 1.0011,
+      "epoch": 6.006857142857143,
+      "grad_norm": 0.1062416136264801,
+      "learning_rate": 9.702222222222223e-06,
+      "loss": 0.4608,
       "step": 2220
     },
     {
-      "epoch": 2.17,
-      "grad_norm": 33.58413314819336,
-      "learning_rate": 4.815024004518498e-06,
-      "loss": 1.1844,
+      "epoch": 6.007428571428571,
+      "grad_norm": 8.830906867980957,
+      "learning_rate": 9.695873015873016e-06,
+      "loss": 0.4464,
       "step": 2230
     },
     {
-      "epoch": 2.17,
-      "grad_norm": 24.298912048339844,
-      "learning_rate": 4.786783394521322e-06,
-      "loss": 1.477,
+      "epoch": 6.008,
+      "grad_norm": 69.506103515625,
+      "learning_rate": 9.68952380952381e-06,
+      "loss": 0.7923,
       "step": 2240
     },
     {
-      "epoch": 2.17,
-      "grad_norm": 0.2964635193347931,
-      "learning_rate": 4.758542784524146e-06,
-      "loss": 0.5349,
+      "epoch": 6.008571428571429,
+      "grad_norm": 42.58530807495117,
+      "learning_rate": 9.683174603174604e-06,
+      "loss": 0.2295,
       "step": 2250
     },
     {
-      "epoch": 2.17,
-      "grad_norm": 0.5078187584877014,
-      "learning_rate": 4.7303021745269704e-06,
-      "loss": 0.9149,
+      "epoch": 6.009142857142857,
+      "grad_norm": 3.4575247764587402,
+      "learning_rate": 9.676825396825399e-06,
+      "loss": 0.0089,
       "step": 2260
     },
     {
-      "epoch": 2.18,
-      "grad_norm": 0.2942535877227783,
-      "learning_rate": 4.702061564529794e-06,
-      "loss": 0.9105,
+      "epoch": 6.009714285714286,
+      "grad_norm": 12.615706443786621,
+      "learning_rate": 9.670476190476191e-06,
+      "loss": 0.5269,
       "step": 2270
     },
     {
-      "epoch": 2.18,
-      "grad_norm": 0.2372201532125473,
-      "learning_rate": 4.673820954532618e-06,
-      "loss": 0.4805,
+      "epoch": 6.010285714285715,
+      "grad_norm": 0.2601805627346039,
+      "learning_rate": 9.664126984126985e-06,
+      "loss": 0.5875,
       "step": 2280
     },
     {
-      "epoch": 2.18,
-      "grad_norm": 1.4744486808776855,
-      "learning_rate": 4.6455803445354425e-06,
-      "loss": 0.0077,
+      "epoch": 6.010857142857143,
+      "grad_norm": 0.12038259208202362,
+      "learning_rate": 9.657777777777778e-06,
+      "loss": 0.2318,
       "step": 2290
     },
     {
-      "epoch": 2.18,
-      "grad_norm": 3.4638776779174805,
-      "learning_rate": 4.617339734538267e-06,
-      "loss": 0.8804,
+      "epoch": 6.011428571428572,
+      "grad_norm": 0.11833731830120087,
+      "learning_rate": 9.651428571428572e-06,
+      "loss": 0.4668,
       "step": 2300
     },
     {
-      "epoch": 2.19,
-      "grad_norm": 0.03575224056839943,
-      "learning_rate": 4.58909912454109e-06,
-      "loss": 0.4857,
+      "epoch": 6.012,
+      "grad_norm": 13.778709411621094,
+      "learning_rate": 9.645079365079367e-06,
+      "loss": 1.5556,
       "step": 2310
     },
     {
-      "epoch": 2.19,
-      "grad_norm": 595.8480834960938,
-      "learning_rate": 4.5608585145439145e-06,
-      "loss": 0.2092,
+      "epoch": 6.0125714285714285,
+      "grad_norm": 12.374515533447266,
+      "learning_rate": 9.63873015873016e-06,
+      "loss": 0.7705,
       "step": 2320
     },
     {
-      "epoch": 2.19,
-      "grad_norm": 0.17693261802196503,
-      "learning_rate": 4.532617904546738e-06,
-      "loss": 1.4817,
+      "epoch": 6.013142857142857,
+      "grad_norm": 12.057218551635742,
+      "learning_rate": 9.632380952380954e-06,
+      "loss": 1.2163,
       "step": 2330
     },
     {
-      "epoch": 2.19,
-      "grad_norm": 0.1713973432779312,
-      "learning_rate": 4.504377294549563e-06,
-      "loss": 0.4704,
+      "epoch": 6.013714285714285,
+      "grad_norm": 0.848735511302948,
+      "learning_rate": 9.626031746031746e-06,
+      "loss": 0.3669,
       "step": 2340
     },
     {
-      "epoch": 2.2,
-      "grad_norm": 0.17562945187091827,
-      "learning_rate": 4.4761366845523865e-06,
-      "loss": 0.4608,
+      "epoch": 6.014285714285714,
+      "grad_norm": 0.11727745085954666,
+      "learning_rate": 9.61968253968254e-06,
+      "loss": 0.6778,
       "step": 2350
     },
     {
-      "epoch": 2.2,
-      "grad_norm": 0.04520571604371071,
-      "learning_rate": 4.447896074555211e-06,
-      "loss": 0.5066,
+      "epoch": 6.014857142857143,
+      "grad_norm": 0.6909122467041016,
+      "learning_rate": 9.613333333333335e-06,
+      "loss": 0.2815,
       "step": 2360
     },
     {
-      "epoch": 2.2,
-      "eval_accuracy": 0.8104026845637584,
-      "eval_loss": 0.8862009644508362,
-      "eval_runtime": 319.8982,
-      "eval_samples_per_second": 1.863,
-      "eval_steps_per_second": 1.863,
-      "step": 2361
-    },
-    {
-      "epoch": 3.0,
-      "grad_norm": 0.06622195243835449,
-      "learning_rate": 4.419655464558035e-06,
-      "loss": 0.4614,
+      "epoch": 6.015428571428571,
+      "grad_norm": 0.31187936663627625,
+      "learning_rate": 9.606984126984128e-06,
+      "loss": 0.418,
       "step": 2370
     },
     {
-      "epoch": 3.0,
-      "grad_norm": 24.007558822631836,
-      "learning_rate": 4.3914148545608585e-06,
-      "loss": 0.463,
+      "epoch": 6.016,
+      "grad_norm": 12.6905517578125,
+      "learning_rate": 9.600634920634922e-06,
+      "loss": 0.6797,
       "step": 2380
     },
     {
-      "epoch": 3.01,
-      "grad_norm": 74.15673065185547,
-      "learning_rate": 4.363174244563683e-06,
-      "loss": 0.4436,
+      "epoch": 6.016571428571429,
+      "grad_norm": 14.10185432434082,
+      "learning_rate": 9.594285714285715e-06,
+      "loss": 0.2239,
       "step": 2390
     },
     {
-      "epoch": 3.01,
-      "grad_norm": 0.021963810548186302,
-      "learning_rate": 4.334933634566507e-06,
-      "loss": 0.0038,
+      "epoch": 6.017142857142857,
+      "grad_norm": 0.6114095449447632,
+      "learning_rate": 9.587936507936509e-06,
+      "loss": 0.4369,
       "step": 2400
     },
     {
-      "epoch": 3.01,
-      "grad_norm": 0.015897750854492188,
-      "learning_rate": 4.306693024569331e-06,
-      "loss": 0.0044,
+      "epoch": 6.017714285714286,
+      "grad_norm": 86.14923095703125,
+      "learning_rate": 9.581587301587303e-06,
+      "loss": 0.7966,
       "step": 2410
     },
     {
-      "epoch": 3.01,
-      "grad_norm": 0.19425620138645172,
-      "learning_rate": 4.278452414572155e-06,
-      "loss": 1.5548,
+      "epoch": 6.018285714285715,
+      "grad_norm": 136.2497100830078,
+      "learning_rate": 9.575238095238096e-06,
+      "loss": 0.8023,
       "step": 2420
     },
     {
-      "epoch": 3.02,
-      "grad_norm": 0.2513945400714874,
-      "learning_rate": 4.250211804574979e-06,
-      "loss": 0.5131,
+      "epoch": 6.018857142857143,
+      "grad_norm": 22.98550796508789,
+      "learning_rate": 9.56888888888889e-06,
+      "loss": 0.2369,
       "step": 2430
     },
     {
-      "epoch": 3.02,
-      "grad_norm": 0.010541065596044064,
-      "learning_rate": 4.221971194577803e-06,
-      "loss": 0.0067,
+      "epoch": 6.019428571428572,
+      "grad_norm": 0.261092871427536,
+      "learning_rate": 9.562539682539683e-06,
+      "loss": 0.5792,
       "step": 2440
     },
     {
-      "epoch": 3.02,
-      "grad_norm": 0.016293305903673172,
-      "learning_rate": 4.193730584580628e-06,
-      "loss": 0.9886,
+      "epoch": 6.02,
+      "grad_norm": 0.12631256878376007,
+      "learning_rate": 9.556190476190477e-06,
+      "loss": 0.1468,
       "step": 2450
     },
     {
-      "epoch": 3.03,
-      "grad_norm": 0.009771335870027542,
-      "learning_rate": 4.165489974583451e-06,
-      "loss": 0.0052,
+      "epoch": 6.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 0.770273745059967,
+      "eval_runtime": 137.4635,
+      "eval_samples_per_second": 2.161,
+      "eval_steps_per_second": 1.084,
+      "step": 2450
+    },
+    {
+      "epoch": 7.000571428571429,
+      "grad_norm": 0.15533775091171265,
+      "learning_rate": 9.54984126984127e-06,
+      "loss": 0.6611,
       "step": 2460
     },
     {
-      "epoch": 3.03,
-      "grad_norm": 22.732620239257812,
-      "learning_rate": 4.1372493645862754e-06,
-      "loss": 0.9754,
+      "epoch": 7.001142857142857,
+      "grad_norm": 15.251904487609863,
+      "learning_rate": 9.543492063492064e-06,
+      "loss": 0.7168,
       "step": 2470
     },
     {
-      "epoch": 3.03,
-      "grad_norm": 0.01741587370634079,
-      "learning_rate": 4.1090087545891e-06,
-      "loss": 0.007,
+      "epoch": 7.001714285714286,
+      "grad_norm": 0.15305683016777039,
+      "learning_rate": 9.537142857142859e-06,
+      "loss": 0.2702,
       "step": 2480
     },
     {
-      "epoch": 3.03,
-      "grad_norm": 0.011250239796936512,
-      "learning_rate": 4.080768144591923e-06,
-      "loss": 1.2522,
+      "epoch": 7.002285714285715,
+      "grad_norm": 0.5726995468139648,
+      "learning_rate": 9.530793650793651e-06,
+      "loss": 0.2093,
       "step": 2490
     },
     {
-      "epoch": 3.04,
-      "grad_norm": 35.29149627685547,
-      "learning_rate": 4.0525275345947475e-06,
-      "loss": 1.0829,
+      "epoch": 7.002857142857143,
+      "grad_norm": 0.23029519617557526,
+      "learning_rate": 9.524444444444445e-06,
+      "loss": 0.6895,
       "step": 2500
     },
     {
-      "epoch": 3.04,
-      "grad_norm": 0.2253457009792328,
-      "learning_rate": 4.024286924597572e-06,
-      "loss": 2.0893,
+      "epoch": 7.003428571428572,
+      "grad_norm": 0.13955456018447876,
+      "learning_rate": 9.518095238095238e-06,
+      "loss": 0.8783,
       "step": 2510
     },
     {
-      "epoch": 3.04,
-      "grad_norm": 0.07997630536556244,
-      "learning_rate": 3.996046314600396e-06,
-      "loss": 0.5229,
+      "epoch": 7.004,
+      "grad_norm": 0.01913376897573471,
+      "learning_rate": 9.511746031746032e-06,
+      "loss": 0.1524,
       "step": 2520
     },
     {
-      "epoch": 3.04,
-      "grad_norm": 0.24757777154445648,
-      "learning_rate": 3.9678057046032195e-06,
-      "loss": 0.0078,
+      "epoch": 7.0045714285714284,
+      "grad_norm": 0.20704668760299683,
+      "learning_rate": 9.505396825396827e-06,
+      "loss": 1.607,
       "step": 2530
     },
     {
-      "epoch": 3.05,
-      "grad_norm": 0.20268729329109192,
-      "learning_rate": 3.939565094606044e-06,
-      "loss": 0.007,
+      "epoch": 7.005142857142857,
+      "grad_norm": 12.19189167022705,
+      "learning_rate": 9.49904761904762e-06,
+      "loss": 0.5308,
       "step": 2540
     },
     {
-      "epoch": 3.05,
-      "grad_norm": 0.18007797002792358,
-      "learning_rate": 3.911324484608868e-06,
-      "loss": 0.0057,
+      "epoch": 7.005714285714285,
+      "grad_norm": 0.30487680435180664,
+      "learning_rate": 9.492698412698414e-06,
+      "loss": 0.4346,
       "step": 2550
     },
     {
-      "epoch": 3.05,
-      "grad_norm": 0.04412233456969261,
-      "learning_rate": 3.8830838746116915e-06,
-      "loss": 0.6502,
+      "epoch": 7.006285714285714,
+      "grad_norm": 0.047250378876924515,
+      "learning_rate": 9.486349206349206e-06,
+      "loss": 0.4239,
       "step": 2560
     },
     {
-      "epoch": 3.05,
-      "grad_norm": 0.2677147686481476,
-      "learning_rate": 3.854843264614516e-06,
-      "loss": 1.0556,
+      "epoch": 7.006857142857143,
+      "grad_norm": 0.45693501830101013,
+      "learning_rate": 9.48e-06,
+      "loss": 0.477,
       "step": 2570
     },
     {
-      "epoch": 3.06,
-      "grad_norm": 0.13710559904575348,
-      "learning_rate": 3.82660265461734e-06,
-      "loss": 0.6048,
+      "epoch": 7.007428571428571,
+      "grad_norm": 0.449886292219162,
+      "learning_rate": 9.473650793650795e-06,
+      "loss": 0.8766,
       "step": 2580
     },
     {
-      "epoch": 3.06,
-      "grad_norm": 0.14919081330299377,
-      "learning_rate": 3.798362044620164e-06,
-      "loss": 1.0146,
+      "epoch": 7.008,
+      "grad_norm": 0.1275578886270523,
+      "learning_rate": 9.467301587301588e-06,
+      "loss": 0.4449,
       "step": 2590
     },
     {
-      "epoch": 3.06,
-      "grad_norm": 0.06460587680339813,
-      "learning_rate": 3.7701214346229882e-06,
-      "loss": 0.0037,
+      "epoch": 7.008571428571429,
+      "grad_norm": 0.08520308881998062,
+      "learning_rate": 9.460952380952382e-06,
+      "loss": 0.4128,
       "step": 2600
     },
     {
-      "epoch": 3.06,
-      "grad_norm": 22.712013244628906,
-      "learning_rate": 3.741880824625812e-06,
-      "loss": 1.1229,
+      "epoch": 7.009142857142857,
+      "grad_norm": 18.56478500366211,
+      "learning_rate": 9.454603174603175e-06,
+      "loss": 0.6254,
       "step": 2610
     },
     {
-      "epoch": 3.07,
-      "grad_norm": 0.03939145430922508,
-      "learning_rate": 3.7136402146286364e-06,
-      "loss": 0.5016,
+      "epoch": 7.009714285714286,
+      "grad_norm": 0.32579270005226135,
+      "learning_rate": 9.448253968253969e-06,
+      "loss": 0.1916,
       "step": 2620
     },
     {
-      "epoch": 3.07,
-      "grad_norm": 0.179279625415802,
-      "learning_rate": 3.6853996046314607e-06,
-      "loss": 0.991,
+      "epoch": 7.010285714285715,
+      "grad_norm": 0.2691892087459564,
+      "learning_rate": 9.441904761904762e-06,
+      "loss": 0.4455,
       "step": 2630
     },
     {
-      "epoch": 3.07,
-      "grad_norm": 0.16804170608520508,
-      "learning_rate": 3.6571589946342845e-06,
-      "loss": 0.0079,
+      "epoch": 7.010857142857143,
+      "grad_norm": 0.18024860322475433,
+      "learning_rate": 9.435555555555556e-06,
+      "loss": 0.2448,
       "step": 2640
     },
     {
-      "epoch": 3.07,
-      "grad_norm": 0.20420299470424652,
-      "learning_rate": 3.6289183846371084e-06,
-      "loss": 0.9568,
+      "epoch": 7.011428571428572,
+      "grad_norm": 12.539464950561523,
+      "learning_rate": 9.42920634920635e-06,
+      "loss": 0.6483,
       "step": 2650
     },
     {
-      "epoch": 3.08,
-      "grad_norm": 0.22263668477535248,
-      "learning_rate": 3.6006777746399323e-06,
-      "loss": 0.4772,
+      "epoch": 7.012,
+      "grad_norm": 24.065021514892578,
+      "learning_rate": 9.422857142857143e-06,
+      "loss": 1.1989,
       "step": 2660
     },
     {
-      "epoch": 3.08,
-      "grad_norm": 22.77140998840332,
-      "learning_rate": 3.5724371646427566e-06,
-      "loss": 1.9533,
+      "epoch": 7.0125714285714285,
+      "grad_norm": 0.15510013699531555,
+      "learning_rate": 9.416507936507937e-06,
+      "loss": 0.4549,
       "step": 2670
     },
     {
-      "epoch": 3.08,
-      "grad_norm": 0.27024608850479126,
-      "learning_rate": 3.544196554645581e-06,
-      "loss": 2.2664,
+      "epoch": 7.013142857142857,
+      "grad_norm": 0.21933165192604065,
+      "learning_rate": 9.41015873015873e-06,
+      "loss": 0.4212,
       "step": 2680
     },
     {
-      "epoch": 3.08,
-      "grad_norm": 0.4369601905345917,
-      "learning_rate": 3.5159559446484047e-06,
-      "loss": 1.1402,
+      "epoch": 7.013714285714285,
+      "grad_norm": 46.31543731689453,
+      "learning_rate": 9.403809523809526e-06,
+      "loss": 0.6977,
       "step": 2690
     },
     {
-      "epoch": 3.09,
-      "grad_norm": 0.055966880172491074,
-      "learning_rate": 3.4877153346512286e-06,
-      "loss": 0.0093,
+      "epoch": 7.014285714285714,
+      "grad_norm": 0.49816974997520447,
+      "learning_rate": 9.397460317460319e-06,
+      "loss": 0.7104,
       "step": 2700
     },
     {
-      "epoch": 3.09,
-      "grad_norm": 0.31162434816360474,
-      "learning_rate": 3.459474724654053e-06,
-      "loss": 0.0104,
+      "epoch": 7.014857142857143,
+      "grad_norm": 0.6536113023757935,
+      "learning_rate": 9.391111111111111e-06,
+      "loss": 0.7153,
       "step": 2710
     },
     {
-      "epoch": 3.09,
-      "grad_norm": 0.09564479440450668,
-      "learning_rate": 3.4312341146568767e-06,
-      "loss": 0.4793,
+      "epoch": 7.015428571428571,
+      "grad_norm": 40.60562515258789,
+      "learning_rate": 9.384761904761906e-06,
+      "loss": 0.3319,
       "step": 2720
     },
     {
-      "epoch": 3.09,
-      "grad_norm": 0.20361103117465973,
-      "learning_rate": 3.4029935046597006e-06,
-      "loss": 0.9354,
+      "epoch": 7.016,
+      "grad_norm": 13.018871307373047,
+      "learning_rate": 9.378412698412698e-06,
+      "loss": 0.2629,
       "step": 2730
     },
     {
-      "epoch": 3.1,
-      "grad_norm": 0.03742867708206177,
-      "learning_rate": 3.3747528946625253e-06,
-      "loss": 0.0062,
+      "epoch": 7.016571428571429,
+      "grad_norm": 0.34966105222702026,
+      "learning_rate": 9.372063492063492e-06,
+      "loss": 0.9186,
       "step": 2740
     },
     {
-      "epoch": 3.1,
-      "grad_norm": 0.17758674919605255,
-      "learning_rate": 3.346512284665349e-06,
-      "loss": 0.9682,
+      "epoch": 7.017142857142857,
+      "grad_norm": 57.22317123413086,
+      "learning_rate": 9.365714285714287e-06,
+      "loss": 1.1176,
       "step": 2750
     },
     {
-      "epoch": 3.1,
-      "grad_norm": 0.04006683826446533,
-      "learning_rate": 3.318271674668173e-06,
-      "loss": 0.6539,
+      "epoch": 7.017714285714286,
+      "grad_norm": 2.8083252906799316,
+      "learning_rate": 9.359365079365081e-06,
+      "loss": 0.5711,
       "step": 2760
     },
     {
-      "epoch": 3.1,
-      "grad_norm": 0.2262236326932907,
-      "learning_rate": 3.290031064670997e-06,
-      "loss": 0.472,
+      "epoch": 7.018285714285715,
+      "grad_norm": 0.2513306140899658,
+      "learning_rate": 9.353015873015874e-06,
+      "loss": 0.3423,
       "step": 2770
     },
     {
-      "epoch": 3.11,
-      "grad_norm": 0.021157417446374893,
-      "learning_rate": 3.2617904546738212e-06,
-      "loss": 0.4522,
+      "epoch": 7.018857142857143,
+      "grad_norm": 0.24741508066654205,
+      "learning_rate": 9.346666666666666e-06,
+      "loss": 0.412,
       "step": 2780
     },
     {
-      "epoch": 3.11,
-      "grad_norm": 0.20334932208061218,
-      "learning_rate": 3.233549844676645e-06,
-      "loss": 0.0057,
+      "epoch": 7.019428571428572,
+      "grad_norm": 0.17685921490192413,
+      "learning_rate": 9.34031746031746e-06,
+      "loss": 0.8817,
       "step": 2790
     },
     {
-      "epoch": 3.11,
-      "grad_norm": 0.18717893958091736,
-      "learning_rate": 3.2053092346794694e-06,
-      "loss": 0.0058,
+      "epoch": 7.02,
+      "grad_norm": 0.2248346507549286,
+      "learning_rate": 9.333968253968255e-06,
+      "loss": 0.8406,
+      "step": 2800
+    },
+    {
+      "epoch": 7.02,
+      "eval_accuracy": 0.8484848484848485,
+      "eval_loss": 0.5862956047058105,
+      "eval_runtime": 137.2695,
+      "eval_samples_per_second": 2.164,
+      "eval_steps_per_second": 1.085,
       "step": 2800
     },
     {
-      "epoch": 3.11,
-      "grad_norm": 0.015843726694583893,
-      "learning_rate": 3.1770686246822937e-06,
-      "loss": 0.9782,
+      "epoch": 8.000571428571428,
+      "grad_norm": 0.10712999105453491,
+      "learning_rate": 9.32761904761905e-06,
+      "loss": 0.5924,
       "step": 2810
     },
     {
-      "epoch": 3.12,
-      "grad_norm": 304.4051208496094,
-      "learning_rate": 3.1488280146851175e-06,
-      "loss": 0.3817,
+      "epoch": 8.001142857142858,
+      "grad_norm": 0.37382298707962036,
+      "learning_rate": 9.321269841269842e-06,
+      "loss": 0.2847,
       "step": 2820
     },
     {
-      "epoch": 3.12,
-      "grad_norm": 22.5703067779541,
-      "learning_rate": 3.1205874046879414e-06,
-      "loss": 0.9501,
+      "epoch": 8.001714285714286,
+      "grad_norm": 0.08458230644464493,
+      "learning_rate": 9.314920634920636e-06,
+      "loss": 1.0296,
       "step": 2830
     },
     {
-      "epoch": 3.12,
-      "grad_norm": 0.015223432332277298,
-      "learning_rate": 3.0923467946907653e-06,
-      "loss": 1.3262,
+      "epoch": 8.002285714285714,
+      "grad_norm": 2.4656920433044434,
+      "learning_rate": 9.308571428571429e-06,
+      "loss": 0.4784,
       "step": 2840
     },
     {
-      "epoch": 3.12,
-      "grad_norm": 22.17841339111328,
-      "learning_rate": 3.06410618469359e-06,
-      "loss": 0.4495,
+      "epoch": 8.002857142857144,
+      "grad_norm": 10.618043899536133,
+      "learning_rate": 9.302222222222223e-06,
+      "loss": 0.733,
       "step": 2850
     },
     {
-      "epoch": 3.13,
-      "grad_norm": 0.2771930694580078,
-      "learning_rate": 3.035865574696414e-06,
-      "loss": 0.005,
+      "epoch": 8.003428571428572,
+      "grad_norm": 0.1708851009607315,
+      "learning_rate": 9.295873015873018e-06,
+      "loss": 0.2806,
       "step": 2860
     },
     {
-      "epoch": 3.13,
-      "grad_norm": 0.24659463763237,
-      "learning_rate": 3.0076249646992377e-06,
-      "loss": 1.0779,
+      "epoch": 8.004,
+      "grad_norm": 0.06309555470943451,
+      "learning_rate": 9.28952380952381e-06,
+      "loss": 0.5611,
       "step": 2870
     },
     {
-      "epoch": 3.13,
-      "grad_norm": 0.017306260764598846,
-      "learning_rate": 2.9793843547020616e-06,
-      "loss": 0.0052,
+      "epoch": 8.00457142857143,
+      "grad_norm": 0.46451622247695923,
+      "learning_rate": 9.283174603174605e-06,
+      "loss": 0.1565,
       "step": 2880
     },
     {
-      "epoch": 3.13,
-      "grad_norm": 0.16755558550357819,
-      "learning_rate": 2.951143744704886e-06,
-      "loss": 0.4679,
+      "epoch": 8.005142857142857,
+      "grad_norm": 13.138236999511719,
+      "learning_rate": 9.276825396825397e-06,
+      "loss": 0.7403,
       "step": 2890
     },
     {
-      "epoch": 3.14,
-      "grad_norm": 22.35555076599121,
-      "learning_rate": 2.9229031347077097e-06,
-      "loss": 1.6778,
+      "epoch": 8.005714285714285,
+      "grad_norm": 1.0157630443572998,
+      "learning_rate": 9.270476190476192e-06,
+      "loss": 0.3141,
       "step": 2900
     },
     {
-      "epoch": 3.14,
-      "grad_norm": 848.5310668945312,
-      "learning_rate": 2.894662524710534e-06,
-      "loss": 0.368,
+      "epoch": 8.006285714285715,
+      "grad_norm": 0.22003336250782013,
+      "learning_rate": 9.264126984126986e-06,
+      "loss": 0.7244,
       "step": 2910
     },
     {
-      "epoch": 3.14,
-      "grad_norm": 0.12664149701595306,
-      "learning_rate": 2.8664219147133583e-06,
-      "loss": 1.8596,
+      "epoch": 8.006857142857143,
+      "grad_norm": 0.22460892796516418,
+      "learning_rate": 9.257777777777779e-06,
+      "loss": 0.397,
       "step": 2920
     },
     {
-      "epoch": 3.14,
-      "grad_norm": 0.15273146331310272,
-      "learning_rate": 2.838181304716182e-06,
-      "loss": 0.0056,
+      "epoch": 8.007428571428571,
+      "grad_norm": 1.3341542482376099,
+      "learning_rate": 9.251428571428573e-06,
+      "loss": 0.9219,
       "step": 2930
     },
     {
-      "epoch": 3.15,
-      "grad_norm": 22.398298263549805,
-      "learning_rate": 2.809940694719006e-06,
-      "loss": 1.4325,
+      "epoch": 8.008,
+      "grad_norm": 0.180324986577034,
+      "learning_rate": 9.245079365079366e-06,
+      "loss": 1.0728,
       "step": 2940
     },
     {
-      "epoch": 3.15,
-      "grad_norm": 413.4339294433594,
-      "learning_rate": 2.78170008472183e-06,
-      "loss": 0.3167,
+      "epoch": 8.008571428571429,
+      "grad_norm": 3.0605599880218506,
+      "learning_rate": 9.23873015873016e-06,
+      "loss": 0.6721,
       "step": 2950
     },
     {
-      "epoch": 3.15,
-      "grad_norm": 0.19188903272151947,
-      "learning_rate": 2.753459474724654e-06,
-      "loss": 0.904,
+      "epoch": 8.009142857142857,
+      "grad_norm": 0.17587581276893616,
+      "learning_rate": 9.232380952380952e-06,
+      "loss": 0.3343,
       "step": 2960
     },
     {
-      "epoch": 3.15,
-      "grad_norm": 3.69321608543396,
-      "learning_rate": 2.7252188647274785e-06,
-      "loss": 0.007,
+      "epoch": 8.009714285714285,
+      "grad_norm": 1.3076510429382324,
+      "learning_rate": 9.226031746031747e-06,
+      "loss": 0.5037,
       "step": 2970
     },
     {
-      "epoch": 3.16,
-      "grad_norm": 0.2725941240787506,
-      "learning_rate": 2.6969782547303024e-06,
-      "loss": 0.9046,
+      "epoch": 8.010285714285715,
+      "grad_norm": 0.15406930446624756,
+      "learning_rate": 9.219682539682541e-06,
+      "loss": 0.4566,
       "step": 2980
     },
     {
-      "epoch": 3.16,
-      "grad_norm": 0.2501017451286316,
-      "learning_rate": 2.6687376447331266e-06,
-      "loss": 0.0085,
+      "epoch": 8.010857142857143,
+      "grad_norm": 12.370487213134766,
+      "learning_rate": 9.213333333333334e-06,
+      "loss": 0.6773,
       "step": 2990
     },
     {
-      "epoch": 3.16,
-      "grad_norm": 0.08356527239084244,
-      "learning_rate": 2.6404970347359505e-06,
-      "loss": 0.6352,
+      "epoch": 8.01142857142857,
+      "grad_norm": 24.417396545410156,
+      "learning_rate": 9.206984126984128e-06,
+      "loss": 0.8717,
       "step": 3000
     },
     {
-      "epoch": 3.16,
-      "grad_norm": 0.05544191226363182,
-      "learning_rate": 2.6122564247387744e-06,
-      "loss": 1.566,
+      "epoch": 8.012,
+      "grad_norm": 0.2499029040336609,
+      "learning_rate": 9.20063492063492e-06,
+      "loss": 0.9279,
       "step": 3010
     },
     {
-      "epoch": 3.17,
-      "grad_norm": 0.2636459469795227,
-      "learning_rate": 2.584015814741599e-06,
-      "loss": 0.9192,
+      "epoch": 8.012571428571428,
+      "grad_norm": 8.220014572143555,
+      "learning_rate": 9.194285714285715e-06,
+      "loss": 0.0187,
       "step": 3020
     },
     {
-      "epoch": 3.17,
-      "grad_norm": 0.05203928053379059,
-      "learning_rate": 2.555775204744423e-06,
-      "loss": 0.4324,
+      "epoch": 8.013142857142856,
+      "grad_norm": 0.12472938001155853,
+      "learning_rate": 9.18793650793651e-06,
+      "loss": 0.1618,
       "step": 3030
     },
     {
-      "epoch": 3.17,
-      "grad_norm": 0.16593405604362488,
-      "learning_rate": 2.527534594747247e-06,
-      "loss": 0.8602,
+      "epoch": 8.013714285714286,
+      "grad_norm": 12.041219711303711,
+      "learning_rate": 9.181587301587302e-06,
+      "loss": 0.7642,
       "step": 3040
     },
     {
-      "epoch": 3.18,
-      "grad_norm": 0.02371574193239212,
-      "learning_rate": 2.4992939847500707e-06,
-      "loss": 0.5104,
+      "epoch": 8.014285714285714,
+      "grad_norm": 0.26260077953338623,
+      "learning_rate": 9.175238095238096e-06,
+      "loss": 0.586,
       "step": 3050
     },
     {
-      "epoch": 3.18,
-      "grad_norm": 4.215577125549316,
-      "learning_rate": 2.4710533747528946e-06,
-      "loss": 0.6802,
+      "epoch": 8.014857142857142,
+      "grad_norm": 0.13079126179218292,
+      "learning_rate": 9.168888888888889e-06,
+      "loss": 0.4182,
       "step": 3060
     },
     {
-      "epoch": 3.18,
-      "grad_norm": 0.1291200816631317,
-      "learning_rate": 2.442812764755719e-06,
-      "loss": 0.0066,
+      "epoch": 8.015428571428572,
+      "grad_norm": 0.1269873082637787,
+      "learning_rate": 9.162539682539683e-06,
+      "loss": 0.2187,
       "step": 3070
     },
     {
-      "epoch": 3.18,
-      "grad_norm": 0.31712397933006287,
-      "learning_rate": 2.414572154758543e-06,
-      "loss": 0.876,
+      "epoch": 8.016,
+      "grad_norm": 0.14042362570762634,
+      "learning_rate": 9.156190476190478e-06,
+      "loss": 0.3016,
       "step": 3080
     },
     {
-      "epoch": 3.19,
-      "grad_norm": 320.600341796875,
-      "learning_rate": 2.386331544761367e-06,
-      "loss": 0.9575,
+      "epoch": 8.016571428571428,
+      "grad_norm": 13.570809364318848,
+      "learning_rate": 9.14984126984127e-06,
+      "loss": 0.8131,
       "step": 3090
     },
     {
-      "epoch": 3.19,
-      "grad_norm": 0.16467230021953583,
-      "learning_rate": 2.3580909347641913e-06,
-      "loss": 0.884,
+      "epoch": 8.017142857142858,
+      "grad_norm": 0.28256142139434814,
+      "learning_rate": 9.143492063492065e-06,
+      "loss": 0.2524,
       "step": 3100
     },
     {
-      "epoch": 3.19,
-      "grad_norm": 0.028689516708254814,
-      "learning_rate": 2.329850324767015e-06,
-      "loss": 0.0067,
+      "epoch": 8.017714285714286,
+      "grad_norm": 0.08613581210374832,
+      "learning_rate": 9.137142857142857e-06,
+      "loss": 0.9875,
       "step": 3110
     },
     {
-      "epoch": 3.19,
-      "grad_norm": 0.3429773151874542,
-      "learning_rate": 2.3016097147698394e-06,
-      "loss": 1.0181,
+      "epoch": 8.018285714285714,
+      "grad_norm": 29.147151947021484,
+      "learning_rate": 9.130793650793652e-06,
+      "loss": 0.6377,
       "step": 3120
     },
     {
-      "epoch": 3.2,
-      "grad_norm": 0.039928894490003586,
-      "learning_rate": 2.2733691047726633e-06,
-      "loss": 1.1309,
+      "epoch": 8.018857142857144,
+      "grad_norm": 119.630126953125,
+      "learning_rate": 9.124444444444444e-06,
+      "loss": 0.1833,
       "step": 3130
     },
     {
-      "epoch": 3.2,
-      "grad_norm": 0.0716266855597496,
-      "learning_rate": 2.245128494775487e-06,
-      "loss": 0.0055,
+      "epoch": 8.019428571428572,
+      "grad_norm": 0.17084644734859467,
+      "learning_rate": 9.118095238095239e-06,
+      "loss": 0.9565,
       "step": 3140
     },
     {
-      "epoch": 3.2,
-      "eval_accuracy": 0.8456375838926175,
-      "eval_loss": 0.7402271628379822,
-      "eval_runtime": 327.8476,
-      "eval_samples_per_second": 1.818,
-      "eval_steps_per_second": 1.818,
-      "step": 3148
+      "epoch": 8.02,
+      "grad_norm": 12.522076606750488,
+      "learning_rate": 9.111746031746033e-06,
+      "loss": 0.4485,
+      "step": 3150
     },
     {
-      "epoch": 4.0,
-      "grad_norm": 0.2870597243309021,
-      "learning_rate": 2.2168878847783115e-06,
-      "loss": 0.9632,
+      "epoch": 8.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.6601792573928833,
+      "eval_runtime": 137.764,
+      "eval_samples_per_second": 2.156,
+      "eval_steps_per_second": 1.082,
       "step": 3150
     },
     {
-      "epoch": 4.0,
-      "grad_norm": 0.29839763045310974,
-      "learning_rate": 2.1886472747811353e-06,
-      "loss": 0.5113,
+      "epoch": 9.000571428571428,
+      "grad_norm": 0.1897997260093689,
+      "learning_rate": 9.105396825396826e-06,
+      "loss": 0.0101,
       "step": 3160
     },
     {
-      "epoch": 4.01,
-      "grad_norm": 0.14275802671909332,
-      "learning_rate": 2.1604066647839596e-06,
-      "loss": 1.0579,
+      "epoch": 9.001142857142858,
+      "grad_norm": 26.98958396911621,
+      "learning_rate": 9.09904761904762e-06,
+      "loss": 1.1643,
       "step": 3170
     },
     {
-      "epoch": 4.01,
-      "grad_norm": 25.018110275268555,
-      "learning_rate": 2.1321660547867835e-06,
-      "loss": 1.0512,
+      "epoch": 9.001714285714286,
+      "grad_norm": 0.6562146544456482,
+      "learning_rate": 9.092698412698412e-06,
+      "loss": 0.5971,
       "step": 3180
     },
     {
-      "epoch": 4.01,
-      "grad_norm": 0.07139433175325394,
-      "learning_rate": 2.1039254447896078e-06,
-      "loss": 0.0045,
+      "epoch": 9.002285714285714,
+      "grad_norm": 6.7756476402282715,
+      "learning_rate": 9.086349206349207e-06,
+      "loss": 0.448,
       "step": 3190
     },
     {
-      "epoch": 4.01,
-      "grad_norm": 0.07226335257291794,
-      "learning_rate": 2.0756848347924316e-06,
-      "loss": 0.1377,
+      "epoch": 9.002857142857144,
+      "grad_norm": 0.3257388472557068,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.3954,
       "step": 3200
     },
     {
-      "epoch": 4.02,
-      "grad_norm": 0.05169367790222168,
-      "learning_rate": 2.047444224795256e-06,
-      "loss": 0.0114,
+      "epoch": 9.003428571428572,
+      "grad_norm": 0.04351123794913292,
+      "learning_rate": 9.073650793650794e-06,
+      "loss": 0.2449,
       "step": 3210
     },
     {
-      "epoch": 4.02,
-      "grad_norm": 0.2018321454524994,
-      "learning_rate": 2.01920361479808e-06,
-      "loss": 0.5378,
+      "epoch": 9.004,
+      "grad_norm": 0.12862706184387207,
+      "learning_rate": 9.067301587301588e-06,
+      "loss": 0.238,
       "step": 3220
     },
     {
-      "epoch": 4.02,
-      "grad_norm": 0.023068850859999657,
-      "learning_rate": 1.9909630048009037e-06,
-      "loss": 0.5436,
+      "epoch": 9.00457142857143,
+      "grad_norm": 0.30041778087615967,
+      "learning_rate": 9.06095238095238e-06,
+      "loss": 0.0238,
       "step": 3230
     },
     {
-      "epoch": 4.02,
-      "grad_norm": 0.12592564523220062,
-      "learning_rate": 1.962722394803728e-06,
-      "loss": 1.2217,
+      "epoch": 9.005142857142857,
+      "grad_norm": 12.429978370666504,
+      "learning_rate": 9.054603174603175e-06,
+      "loss": 1.2523,
       "step": 3240
     },
     {
-      "epoch": 4.03,
-      "grad_norm": 81.48845672607422,
-      "learning_rate": 1.934481784806552e-06,
-      "loss": 0.9742,
+      "epoch": 9.005714285714285,
+      "grad_norm": 12.339615821838379,
+      "learning_rate": 9.04825396825397e-06,
+      "loss": 0.2253,
       "step": 3250
     },
     {
-      "epoch": 4.03,
-      "grad_norm": 0.19448934495449066,
-      "learning_rate": 1.906241174809376e-06,
-      "loss": 0.4926,
+      "epoch": 9.006285714285715,
+      "grad_norm": 0.3683285117149353,
+      "learning_rate": 9.041904761904762e-06,
+      "loss": 0.5474,
       "step": 3260
     },
     {
-      "epoch": 4.03,
-      "grad_norm": 0.3121252655982971,
-      "learning_rate": 1.8780005648122002e-06,
-      "loss": 0.4974,
+      "epoch": 9.006857142857143,
+      "grad_norm": 0.29264554381370544,
+      "learning_rate": 9.035555555555556e-06,
+      "loss": 0.0913,
       "step": 3270
     },
     {
-      "epoch": 4.03,
-      "grad_norm": 0.3977165222167969,
-      "learning_rate": 1.849759954815024e-06,
-      "loss": 0.5231,
+      "epoch": 9.007428571428571,
+      "grad_norm": 0.14638611674308777,
+      "learning_rate": 9.029206349206349e-06,
+      "loss": 0.0075,
       "step": 3280
     },
     {
-      "epoch": 4.04,
-      "grad_norm": 0.19376535713672638,
-      "learning_rate": 1.8215193448178481e-06,
-      "loss": 1.344,
+      "epoch": 9.008,
+      "grad_norm": 0.13496576249599457,
+      "learning_rate": 9.022857142857143e-06,
+      "loss": 0.3115,
       "step": 3290
     },
     {
-      "epoch": 4.04,
-      "grad_norm": 0.12233041226863861,
-      "learning_rate": 1.7932787348206724e-06,
-      "loss": 0.0042,
+      "epoch": 9.008571428571429,
+      "grad_norm": 0.09361585974693298,
+      "learning_rate": 9.016507936507938e-06,
+      "loss": 0.5494,
       "step": 3300
     },
     {
-      "epoch": 4.04,
-      "grad_norm": 1.1810427904129028,
-      "learning_rate": 1.7650381248234963e-06,
-      "loss": 1.0738,
+      "epoch": 9.009142857142857,
+      "grad_norm": 0.09114635735750198,
+      "learning_rate": 9.010158730158732e-06,
+      "loss": 0.4454,
       "step": 3310
     },
     {
-      "epoch": 4.04,
-      "grad_norm": 0.048179544508457184,
-      "learning_rate": 1.7367975148263202e-06,
-      "loss": 0.4614,
+      "epoch": 9.009714285714285,
+      "grad_norm": 107.50813293457031,
+      "learning_rate": 9.003809523809525e-06,
+      "loss": 0.8347,
       "step": 3320
     },
     {
-      "epoch": 4.05,
-      "grad_norm": 0.058793701231479645,
-      "learning_rate": 1.7085569048291444e-06,
-      "loss": 0.4894,
+      "epoch": 9.010285714285715,
+      "grad_norm": 0.29273521900177,
+      "learning_rate": 8.997460317460317e-06,
+      "loss": 0.5854,
       "step": 3330
     },
     {
-      "epoch": 4.05,
-      "grad_norm": 0.033707864582538605,
-      "learning_rate": 1.6803162948319685e-06,
-      "loss": 1.4924,
+      "epoch": 9.010857142857143,
+      "grad_norm": 18.072633743286133,
+      "learning_rate": 8.991111111111112e-06,
+      "loss": 0.7517,
       "step": 3340
     },
     {
-      "epoch": 4.05,
-      "grad_norm": 0.12816235423088074,
-      "learning_rate": 1.6520756848347926e-06,
-      "loss": 0.0034,
+      "epoch": 9.01142857142857,
+      "grad_norm": 0.3977367877960205,
+      "learning_rate": 8.984761904761904e-06,
+      "loss": 0.7607,
       "step": 3350
     },
     {
-      "epoch": 4.05,
-      "grad_norm": 0.07232780009508133,
-      "learning_rate": 1.6238350748376167e-06,
-      "loss": 0.4661,
+      "epoch": 9.012,
+      "grad_norm": 0.37464699149131775,
+      "learning_rate": 8.9784126984127e-06,
+      "loss": 0.659,
       "step": 3360
     },
     {
-      "epoch": 4.06,
-      "grad_norm": 0.027900898829102516,
-      "learning_rate": 1.5955944648404405e-06,
-      "loss": 0.671,
+      "epoch": 9.012571428571428,
+      "grad_norm": 0.13560600578784943,
+      "learning_rate": 8.972063492063493e-06,
+      "loss": 0.3875,
       "step": 3370
     },
     {
-      "epoch": 4.06,
-      "grad_norm": 0.1532682329416275,
-      "learning_rate": 1.5673538548432648e-06,
-      "loss": 0.4657,
+      "epoch": 9.013142857142856,
+      "grad_norm": 0.609217643737793,
+      "learning_rate": 8.965714285714287e-06,
+      "loss": 0.8813,
       "step": 3380
     },
     {
-      "epoch": 4.06,
-      "grad_norm": 0.3767637014389038,
-      "learning_rate": 1.539113244846089e-06,
-      "loss": 0.0093,
+      "epoch": 9.013714285714286,
+      "grad_norm": 0.6446564793586731,
+      "learning_rate": 8.95936507936508e-06,
+      "loss": 0.3148,
       "step": 3390
     },
     {
-      "epoch": 4.06,
-      "grad_norm": 23.31383514404297,
-      "learning_rate": 1.5108726348489128e-06,
-      "loss": 1.4704,
+      "epoch": 9.014285714285714,
+      "grad_norm": 48.195072174072266,
+      "learning_rate": 8.953015873015874e-06,
+      "loss": 1.2333,
       "step": 3400
     },
     {
-      "epoch": 4.07,
-      "grad_norm": 113.44857788085938,
-      "learning_rate": 1.482632024851737e-06,
-      "loss": 1.4151,
+      "epoch": 9.014857142857142,
+      "grad_norm": 0.19967958331108093,
+      "learning_rate": 8.946666666666669e-06,
+      "loss": 0.3988,
       "step": 3410
     },
     {
-      "epoch": 4.07,
-      "grad_norm": 0.1415228694677353,
-      "learning_rate": 1.454391414854561e-06,
-      "loss": 0.0069,
+      "epoch": 9.015428571428572,
+      "grad_norm": 0.22969119250774384,
+      "learning_rate": 8.940317460317461e-06,
+      "loss": 0.2877,
       "step": 3420
     },
     {
-      "epoch": 4.07,
-      "grad_norm": 0.12466636300086975,
-      "learning_rate": 1.426150804857385e-06,
-      "loss": 0.4532,
+      "epoch": 9.016,
+      "grad_norm": 14.168766975402832,
+      "learning_rate": 8.933968253968256e-06,
+      "loss": 1.2265,
       "step": 3430
     },
     {
-      "epoch": 4.07,
-      "grad_norm": 0.2385646253824234,
-      "learning_rate": 1.397910194860209e-06,
-      "loss": 0.0064,
+      "epoch": 9.016571428571428,
+      "grad_norm": 0.15207406878471375,
+      "learning_rate": 8.927619047619048e-06,
+      "loss": 0.4666,
       "step": 3440
     },
     {
-      "epoch": 4.08,
-      "grad_norm": 0.0742715522646904,
-      "learning_rate": 1.3696695848630332e-06,
-      "loss": 0.0047,
+      "epoch": 9.017142857142858,
+      "grad_norm": 0.09754037857055664,
+      "learning_rate": 8.921269841269842e-06,
+      "loss": 0.7216,
       "step": 3450
     },
     {
-      "epoch": 4.08,
-      "grad_norm": 1417.8148193359375,
-      "learning_rate": 1.341428974865857e-06,
-      "loss": 0.2286,
+      "epoch": 9.017714285714286,
+      "grad_norm": 13.291885375976562,
+      "learning_rate": 8.914920634920635e-06,
+      "loss": 0.4949,
       "step": 3460
     },
     {
-      "epoch": 4.08,
-      "grad_norm": 0.026701288297772408,
-      "learning_rate": 1.3131883648686813e-06,
-      "loss": 0.4781,
+      "epoch": 9.018285714285714,
+      "grad_norm": 0.16420067846775055,
+      "learning_rate": 8.90857142857143e-06,
+      "loss": 0.1193,
       "step": 3470
     },
     {
-      "epoch": 4.08,
-      "grad_norm": 24.302400588989258,
-      "learning_rate": 1.2849477548715054e-06,
-      "loss": 1.4506,
+      "epoch": 9.018857142857144,
+      "grad_norm": 1.3726239204406738,
+      "learning_rate": 8.902222222222224e-06,
+      "loss": 0.4908,
       "step": 3480
     },
     {
-      "epoch": 4.09,
-      "grad_norm": 0.20589368045330048,
-      "learning_rate": 1.2567071448743293e-06,
-      "loss": 0.5483,
+      "epoch": 9.019428571428572,
+      "grad_norm": 12.551621437072754,
+      "learning_rate": 8.895873015873016e-06,
+      "loss": 1.2967,
       "step": 3490
     },
     {
-      "epoch": 4.09,
-      "grad_norm": 23.55211067199707,
-      "learning_rate": 1.2284665348771536e-06,
-      "loss": 1.104,
+      "epoch": 9.02,
+      "grad_norm": 0.05716940015554428,
+      "learning_rate": 8.88952380952381e-06,
+      "loss": 0.0134,
+      "step": 3500
+    },
+    {
+      "epoch": 9.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 0.6907000541687012,
+      "eval_runtime": 137.8377,
+      "eval_samples_per_second": 2.155,
+      "eval_steps_per_second": 1.081,
       "step": 3500
     },
     {
-      "epoch": 4.09,
-      "grad_norm": 0.11621620506048203,
-      "learning_rate": 1.2002259248799774e-06,
-      "loss": 0.6127,
+      "epoch": 10.000571428571428,
+      "grad_norm": 63.84096145629883,
+      "learning_rate": 8.883174603174603e-06,
+      "loss": 1.0078,
       "step": 3510
     },
     {
-      "epoch": 4.09,
-      "grad_norm": 0.12657718360424042,
-      "learning_rate": 1.1719853148828015e-06,
-      "loss": 0.4753,
+      "epoch": 10.001142857142858,
+      "grad_norm": 25.32430076599121,
+      "learning_rate": 8.876825396825398e-06,
+      "loss": 1.2722,
       "step": 3520
     },
     {
-      "epoch": 4.1,
-      "grad_norm": 62.24040222167969,
-      "learning_rate": 1.1437447048856256e-06,
-      "loss": 0.9395,
+      "epoch": 10.001714285714286,
+      "grad_norm": 126.55713653564453,
+      "learning_rate": 8.870476190476192e-06,
+      "loss": 0.5043,
       "step": 3530
     },
     {
-      "epoch": 4.1,
-      "grad_norm": 0.11141040176153183,
-      "learning_rate": 1.1155040948884497e-06,
-      "loss": 0.4753,
+      "epoch": 10.002285714285714,
+      "grad_norm": 0.31926584243774414,
+      "learning_rate": 8.864126984126985e-06,
+      "loss": 0.122,
       "step": 3540
     },
     {
-      "epoch": 4.1,
-      "grad_norm": 23.662633895874023,
-      "learning_rate": 1.0872634848912737e-06,
-      "loss": 0.472,
+      "epoch": 10.002857142857144,
+      "grad_norm": 75.52194213867188,
+      "learning_rate": 8.857777777777779e-06,
+      "loss": 0.579,
       "step": 3550
     },
     {
-      "epoch": 4.1,
-      "grad_norm": 0.029395530000329018,
-      "learning_rate": 1.0590228748940978e-06,
-      "loss": 0.0064,
+      "epoch": 10.003428571428572,
+      "grad_norm": 59.787635803222656,
+      "learning_rate": 8.851428571428572e-06,
+      "loss": 0.5681,
       "step": 3560
     },
     {
-      "epoch": 4.11,
-      "grad_norm": 0.03742952272295952,
-      "learning_rate": 1.0307822648969219e-06,
-      "loss": 0.469,
+      "epoch": 10.004,
+      "grad_norm": 0.1938324123620987,
+      "learning_rate": 8.845079365079366e-06,
+      "loss": 0.7633,
       "step": 3570
     },
     {
-      "epoch": 4.11,
-      "grad_norm": 0.08150259405374527,
-      "learning_rate": 1.002541654899746e-06,
-      "loss": 0.4674,
+      "epoch": 10.00457142857143,
+      "grad_norm": 12.319451332092285,
+      "learning_rate": 8.83873015873016e-06,
+      "loss": 1.1426,
       "step": 3580
     },
     {
-      "epoch": 4.11,
-      "grad_norm": 0.05578187480568886,
-      "learning_rate": 9.7430104490257e-07,
-      "loss": 0.5241,
+      "epoch": 10.005142857142857,
+      "grad_norm": 12.19789981842041,
+      "learning_rate": 8.832380952380953e-06,
+      "loss": 0.4935,
       "step": 3590
     },
     {
-      "epoch": 4.11,
-      "grad_norm": 56.26320266723633,
-      "learning_rate": 9.460604349053941e-07,
-      "loss": 1.0151,
+      "epoch": 10.005714285714285,
+      "grad_norm": 0.08483655005693436,
+      "learning_rate": 8.826031746031747e-06,
+      "loss": 0.4319,
       "step": 3600
     },
     {
-      "epoch": 4.12,
-      "grad_norm": 0.1536935567855835,
-      "learning_rate": 9.178198249082181e-07,
-      "loss": 1.9666,
+      "epoch": 10.006285714285715,
+      "grad_norm": 0.2407812774181366,
+      "learning_rate": 8.81968253968254e-06,
+      "loss": 0.4388,
       "step": 3610
     },
     {
-      "epoch": 4.12,
-      "grad_norm": 2.32802677154541,
-      "learning_rate": 8.895792149110422e-07,
-      "loss": 0.4655,
+      "epoch": 10.006857142857143,
+      "grad_norm": 0.20790864527225494,
+      "learning_rate": 8.813333333333334e-06,
+      "loss": 0.6393,
       "step": 3620
     },
     {
-      "epoch": 4.12,
-      "grad_norm": 619.9100952148438,
-      "learning_rate": 8.613386049138663e-07,
-      "loss": 0.788,
+      "epoch": 10.007428571428571,
+      "grad_norm": 6.273917198181152,
+      "learning_rate": 8.806984126984127e-06,
+      "loss": 0.2051,
       "step": 3630
     },
     {
-      "epoch": 4.13,
-      "grad_norm": 34.0890007019043,
-      "learning_rate": 8.330979949166902e-07,
-      "loss": 1.3822,
+      "epoch": 10.008,
+      "grad_norm": 0.6706716418266296,
+      "learning_rate": 8.800634920634921e-06,
+      "loss": 0.6428,
       "step": 3640
     },
     {
-      "epoch": 4.13,
-      "grad_norm": 0.2547508180141449,
-      "learning_rate": 8.048573849195143e-07,
-      "loss": 0.8994,
+      "epoch": 10.008571428571429,
+      "grad_norm": 0.3309767544269562,
+      "learning_rate": 8.794285714285716e-06,
+      "loss": 0.2073,
       "step": 3650
     },
     {
-      "epoch": 4.13,
-      "grad_norm": 0.27960023283958435,
-      "learning_rate": 7.766167749223384e-07,
-      "loss": 0.5559,
+      "epoch": 10.009142857142857,
+      "grad_norm": 0.3042340576648712,
+      "learning_rate": 8.787936507936508e-06,
+      "loss": 0.7154,
       "step": 3660
     },
     {
-      "epoch": 4.13,
-      "grad_norm": 0.05300937965512276,
-      "learning_rate": 7.483761649251624e-07,
-      "loss": 0.4556,
+      "epoch": 10.009714285714285,
+      "grad_norm": 0.23488643765449524,
+      "learning_rate": 8.781587301587302e-06,
+      "loss": 0.2957,
       "step": 3670
     },
     {
-      "epoch": 4.14,
-      "grad_norm": 0.024076635017991066,
-      "learning_rate": 7.201355549279864e-07,
-      "loss": 0.0023,
+      "epoch": 10.010285714285715,
+      "grad_norm": 0.09738802909851074,
+      "learning_rate": 8.775238095238095e-06,
+      "loss": 0.0073,
       "step": 3680
     },
     {
-      "epoch": 4.14,
-      "grad_norm": 0.27708545327186584,
-      "learning_rate": 6.918949449308106e-07,
-      "loss": 1.2558,
+      "epoch": 10.010857142857143,
+      "grad_norm": 0.10398265719413757,
+      "learning_rate": 8.76888888888889e-06,
+      "loss": 0.8261,
       "step": 3690
     },
     {
-      "epoch": 4.14,
-      "grad_norm": 25.573650360107422,
-      "learning_rate": 6.636543349336347e-07,
-      "loss": 2.7475,
+      "epoch": 10.01142857142857,
+      "grad_norm": 0.1489027887582779,
+      "learning_rate": 8.762539682539684e-06,
+      "loss": 0.8143,
       "step": 3700
     },
     {
-      "epoch": 4.14,
-      "grad_norm": 0.28637179732322693,
-      "learning_rate": 6.354137249364587e-07,
-      "loss": 0.8946,
+      "epoch": 10.012,
+      "grad_norm": 0.2872755527496338,
+      "learning_rate": 8.756190476190476e-06,
+      "loss": 0.7231,
       "step": 3710
     },
     {
-      "epoch": 4.15,
-      "grad_norm": 0.1498030722141266,
-      "learning_rate": 6.071731149392827e-07,
-      "loss": 0.0068,
+      "epoch": 10.012571428571428,
+      "grad_norm": 0.08701366931200027,
+      "learning_rate": 8.74984126984127e-06,
+      "loss": 0.4884,
       "step": 3720
     },
     {
-      "epoch": 4.15,
-      "grad_norm": 0.03544044494628906,
-      "learning_rate": 5.789325049421068e-07,
-      "loss": 0.1509,
+      "epoch": 10.013142857142856,
+      "grad_norm": 0.18288986384868622,
+      "learning_rate": 8.743492063492063e-06,
+      "loss": 0.2193,
       "step": 3730
     },
     {
-      "epoch": 4.15,
-      "grad_norm": 0.2865348756313324,
-      "learning_rate": 5.506918949449309e-07,
-      "loss": 0.0056,
+      "epoch": 10.013714285714286,
+      "grad_norm": 0.14304602146148682,
+      "learning_rate": 8.737142857142858e-06,
+      "loss": 0.4951,
       "step": 3740
     },
     {
-      "epoch": 4.15,
-      "grad_norm": 0.6191418170928955,
-      "learning_rate": 5.224512849477549e-07,
-      "loss": 0.987,
+      "epoch": 10.014285714285714,
+      "grad_norm": 0.23201577365398407,
+      "learning_rate": 8.730793650793652e-06,
+      "loss": 0.3121,
       "step": 3750
     },
     {
-      "epoch": 4.16,
-      "grad_norm": 0.03451582416892052,
-      "learning_rate": 4.942106749505791e-07,
-      "loss": 0.5444,
+      "epoch": 10.014857142857142,
+      "grad_norm": 196.32400512695312,
+      "learning_rate": 8.724444444444445e-06,
+      "loss": 0.5922,
       "step": 3760
     },
     {
-      "epoch": 4.16,
-      "grad_norm": 0.3244670331478119,
-      "learning_rate": 4.6597006495340303e-07,
-      "loss": 0.6433,
+      "epoch": 10.015428571428572,
+      "grad_norm": 0.09482351690530777,
+      "learning_rate": 8.718095238095239e-06,
+      "loss": 0.4892,
       "step": 3770
     },
     {
-      "epoch": 4.16,
-      "grad_norm": 23.457420349121094,
-      "learning_rate": 4.3772945495622706e-07,
-      "loss": 1.4575,
+      "epoch": 10.016,
+      "grad_norm": 0.1551242470741272,
+      "learning_rate": 8.711746031746032e-06,
+      "loss": 0.523,
       "step": 3780
     },
     {
-      "epoch": 4.16,
-      "grad_norm": 0.2671697735786438,
-      "learning_rate": 4.094888449590512e-07,
-      "loss": 0.0069,
+      "epoch": 10.016571428571428,
+      "grad_norm": 0.15693677961826324,
+      "learning_rate": 8.705396825396826e-06,
+      "loss": 0.9376,
       "step": 3790
     },
     {
-      "epoch": 4.17,
-      "grad_norm": 0.0375777892768383,
-      "learning_rate": 3.812482349618752e-07,
-      "loss": 0.5398,
+      "epoch": 10.017142857142858,
+      "grad_norm": 0.38993221521377563,
+      "learning_rate": 8.69904761904762e-06,
+      "loss": 0.8241,
       "step": 3800
     },
     {
-      "epoch": 4.17,
-      "grad_norm": 0.7031461596488953,
-      "learning_rate": 3.530076249646993e-07,
-      "loss": 0.0067,
+      "epoch": 10.017714285714286,
+      "grad_norm": 0.5978397727012634,
+      "learning_rate": 8.692698412698413e-06,
+      "loss": 0.806,
       "step": 3810
     },
     {
-      "epoch": 4.17,
-      "grad_norm": 0.027200641110539436,
-      "learning_rate": 3.247670149675233e-07,
-      "loss": 0.0063,
+      "epoch": 10.018285714285714,
+      "grad_norm": 0.41364625096321106,
+      "learning_rate": 8.686349206349207e-06,
+      "loss": 0.5164,
       "step": 3820
     },
     {
-      "epoch": 4.17,
-      "grad_norm": 0.1259397715330124,
-      "learning_rate": 2.965264049703474e-07,
-      "loss": 0.619,
+      "epoch": 10.018857142857144,
+      "grad_norm": 0.32526934146881104,
+      "learning_rate": 8.68e-06,
+      "loss": 0.5054,
       "step": 3830
     },
     {
-      "epoch": 4.18,
-      "grad_norm": 0.12452813237905502,
-      "learning_rate": 2.682857949731714e-07,
-      "loss": 0.0046,
+      "epoch": 10.019428571428572,
+      "grad_norm": 0.6359074711799622,
+      "learning_rate": 8.673650793650794e-06,
+      "loss": 0.31,
       "step": 3840
     },
     {
-      "epoch": 4.18,
-      "grad_norm": 0.25807440280914307,
-      "learning_rate": 2.400451849759955e-07,
-      "loss": 0.6994,
+      "epoch": 10.02,
+      "grad_norm": 85.47923278808594,
+      "learning_rate": 8.667301587301587e-06,
+      "loss": 0.11,
+      "step": 3850
+    },
+    {
+      "epoch": 10.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 0.7097596526145935,
+      "eval_runtime": 137.5142,
+      "eval_samples_per_second": 2.16,
+      "eval_steps_per_second": 1.084,
       "step": 3850
     },
     {
-      "epoch": 4.18,
-      "grad_norm": 0.13341209292411804,
-      "learning_rate": 2.1180457497881955e-07,
-      "loss": 0.5316,
+      "epoch": 11.000571428571428,
+      "grad_norm": 0.1688612848520279,
+      "learning_rate": 8.660952380952383e-06,
+      "loss": 0.4141,
       "step": 3860
     },
     {
-      "epoch": 4.18,
-      "grad_norm": 22.768707275390625,
-      "learning_rate": 1.8356396498164363e-07,
-      "loss": 1.5915,
+      "epoch": 11.001142857142858,
+      "grad_norm": 12.029210090637207,
+      "learning_rate": 8.654603174603176e-06,
+      "loss": 0.8705,
       "step": 3870
     },
     {
-      "epoch": 4.19,
-      "grad_norm": 0.050129234790802,
-      "learning_rate": 1.5532335498446768e-07,
-      "loss": 1.065,
+      "epoch": 11.001714285714286,
+      "grad_norm": 0.19282205402851105,
+      "learning_rate": 8.648253968253968e-06,
+      "loss": 0.3451,
       "step": 3880
     },
     {
-      "epoch": 4.19,
-      "grad_norm": 0.10654575377702713,
-      "learning_rate": 1.2708274498729173e-07,
-      "loss": 1.1427,
+      "epoch": 11.002285714285714,
+      "grad_norm": 12.639680862426758,
+      "learning_rate": 8.641904761904762e-06,
+      "loss": 0.9504,
       "step": 3890
     },
     {
-      "epoch": 4.19,
-      "grad_norm": 0.04145923629403114,
-      "learning_rate": 9.884213499011579e-08,
-      "loss": 0.0048,
+      "epoch": 11.002857142857144,
+      "grad_norm": 0.34344059228897095,
+      "learning_rate": 8.635555555555555e-06,
+      "loss": 0.3582,
       "step": 3900
     },
     {
-      "epoch": 4.19,
-      "grad_norm": 35.09569549560547,
-      "learning_rate": 7.060152499293986e-08,
-      "loss": 1.6677,
+      "epoch": 11.003428571428572,
+      "grad_norm": 193.6134033203125,
+      "learning_rate": 8.62920634920635e-06,
+      "loss": 0.1138,
       "step": 3910
     },
     {
-      "epoch": 4.2,
-      "grad_norm": 0.2241126149892807,
-      "learning_rate": 4.236091499576391e-08,
-      "loss": 0.4727,
+      "epoch": 11.004,
+      "grad_norm": 0.36762064695358276,
+      "learning_rate": 8.622857142857144e-06,
+      "loss": 0.556,
       "step": 3920
     },
     {
-      "epoch": 4.2,
-      "grad_norm": 0.043600454926490784,
-      "learning_rate": 1.4120304998587971e-08,
-      "loss": 0.0042,
+      "epoch": 11.00457142857143,
+      "grad_norm": 0.19313932955265045,
+      "learning_rate": 8.616507936507938e-06,
+      "loss": 0.5323,
       "step": 3930
     },
     {
-      "epoch": 4.2,
-      "eval_accuracy": 0.8104026845637584,
-      "eval_loss": 0.8931716680526733,
-      "eval_runtime": 327.1353,
-      "eval_samples_per_second": 1.822,
-      "eval_steps_per_second": 1.822,
-      "step": 3935
-    },
-    {
-      "epoch": 4.2,
-      "step": 3935,
-      "total_flos": 1.7278688550692782e+19,
-      "train_loss": 0.7462392025486373,
-      "train_runtime": 7487.9132,
-      "train_samples_per_second": 0.526,
-      "train_steps_per_second": 0.526
-    },
-    {
-      "epoch": 4.2,
-      "eval_accuracy": 0.8456375838926175,
-      "eval_loss": 0.7402271628379822,
-      "eval_runtime": 321.7128,
-      "eval_samples_per_second": 1.853,
-      "eval_steps_per_second": 1.853,
-      "step": 3935
-    },
-    {
-      "epoch": 4.2,
-      "eval_accuracy": 0.8456375838926175,
-      "eval_loss": 0.7402271032333374,
-      "eval_runtime": 328.0839,
-      "eval_samples_per_second": 1.817,
-      "eval_steps_per_second": 1.817,
-      "step": 3935
-    }
-  ],
-  "logging_steps": 10,
-  "max_steps": 3935,
-  "num_input_tokens_seen": 0,
-  "num_train_epochs": 9223372036854775807,
-  "save_steps": 500,
-  "total_flos": 1.7278688550692782e+19,
-  "train_batch_size": 1,
+      "epoch": 11.005142857142857,
+      "grad_norm": 0.26507726311683655,
+      "learning_rate": 8.61015873015873e-06,
+      "loss": 0.4584,
+      "step": 3940
+    },
+    {
+      "epoch": 11.005714285714285,
+      "grad_norm": 0.09229809045791626,
+      "learning_rate": 8.603809523809525e-06,
+      "loss": 0.4409,
+      "step": 3950
+    },
+    {
+      "epoch": 11.006285714285715,
+      "grad_norm": 0.14882345497608185,
+      "learning_rate": 8.597460317460318e-06,
+      "loss": 0.0062,
+      "step": 3960
+    },
+    {
+      "epoch": 11.006857142857143,
+      "grad_norm": 12.128185272216797,
+      "learning_rate": 8.591111111111112e-06,
+      "loss": 0.2344,
+      "step": 3970
+    },
+    {
+      "epoch": 11.007428571428571,
+      "grad_norm": 0.22297517955303192,
+      "learning_rate": 8.584761904761906e-06,
+      "loss": 0.1457,
+      "step": 3980
+    },
+    {
+      "epoch": 11.008,
+      "grad_norm": 119.80109405517578,
+      "learning_rate": 8.578412698412699e-06,
+      "loss": 0.1712,
+      "step": 3990
+    },
+    {
+      "epoch": 11.008571428571429,
+      "grad_norm": 0.09167554974555969,
+      "learning_rate": 8.572063492063493e-06,
+      "loss": 0.5816,
+      "step": 4000
+    },
+    {
+      "epoch": 11.009142857142857,
+      "grad_norm": 21.545888900756836,
+      "learning_rate": 8.565714285714286e-06,
+      "loss": 0.9225,
+      "step": 4010
+    },
+    {
+      "epoch": 11.009714285714285,
+      "grad_norm": 0.07726138085126877,
+      "learning_rate": 8.55936507936508e-06,
+      "loss": 0.3713,
+      "step": 4020
+    },
+    {
+      "epoch": 11.010285714285715,
+      "grad_norm": 0.14826586842536926,
+      "learning_rate": 8.553015873015875e-06,
+      "loss": 0.2916,
+      "step": 4030
+    },
+    {
+      "epoch": 11.010857142857143,
+      "grad_norm": 13.104598045349121,
+      "learning_rate": 8.546666666666667e-06,
+      "loss": 1.1717,
+      "step": 4040
+    },
+    {
+      "epoch": 11.01142857142857,
+      "grad_norm": 0.14416873455047607,
+      "learning_rate": 8.540317460317462e-06,
+      "loss": 0.5383,
+      "step": 4050
+    },
+    {
+      "epoch": 11.012,
+      "grad_norm": 0.22047214210033417,
+      "learning_rate": 8.533968253968254e-06,
+      "loss": 0.4027,
+      "step": 4060
+    },
+    {
+      "epoch": 11.012571428571428,
+      "grad_norm": 1.8622279167175293,
+      "learning_rate": 8.527619047619049e-06,
+      "loss": 0.4155,
+      "step": 4070
+    },
+    {
+      "epoch": 11.013142857142856,
+      "grad_norm": 18.07794952392578,
+      "learning_rate": 8.521269841269843e-06,
+      "loss": 0.2927,
+      "step": 4080
+    },
+    {
+      "epoch": 11.013714285714286,
+      "grad_norm": 30.948429107666016,
+      "learning_rate": 8.514920634920636e-06,
+      "loss": 0.7502,
+      "step": 4090
+    },
+    {
+      "epoch": 11.014285714285714,
+      "grad_norm": 12.959928512573242,
+      "learning_rate": 8.50857142857143e-06,
+      "loss": 0.2332,
+      "step": 4100
+    },
+    {
+      "epoch": 11.014857142857142,
+      "grad_norm": 147.560546875,
+      "learning_rate": 8.502222222222223e-06,
+      "loss": 0.3589,
+      "step": 4110
+    },
+    {
+      "epoch": 11.015428571428572,
+      "grad_norm": 0.5723152756690979,
+      "learning_rate": 8.495873015873017e-06,
+      "loss": 0.7663,
+      "step": 4120
+    },
+    {
+      "epoch": 11.016,
+      "grad_norm": 0.13386109471321106,
+      "learning_rate": 8.48952380952381e-06,
+      "loss": 1.0655,
+      "step": 4130
+    },
+    {
+      "epoch": 11.016571428571428,
+      "grad_norm": 0.2565602660179138,
+      "learning_rate": 8.483174603174604e-06,
+      "loss": 0.6372,
+      "step": 4140
+    },
+    {
+      "epoch": 11.017142857142858,
+      "grad_norm": 0.517414391040802,
+      "learning_rate": 8.476825396825398e-06,
+      "loss": 0.2303,
+      "step": 4150
+    },
+    {
+      "epoch": 11.017714285714286,
+      "grad_norm": 0.18709756433963776,
+      "learning_rate": 8.47047619047619e-06,
+      "loss": 0.8909,
+      "step": 4160
+    },
+    {
+      "epoch": 11.018285714285714,
+      "grad_norm": 241.76470947265625,
+      "learning_rate": 8.464126984126985e-06,
+      "loss": 1.3007,
+      "step": 4170
+    },
+    {
+      "epoch": 11.018857142857144,
+      "grad_norm": 0.4402504861354828,
+      "learning_rate": 8.457777777777778e-06,
+      "loss": 0.1889,
+      "step": 4180
+    },
+    {
+      "epoch": 11.019428571428572,
+      "grad_norm": 0.13897043466567993,
+      "learning_rate": 8.451428571428572e-06,
+      "loss": 0.5257,
+      "step": 4190
+    },
+    {
+      "epoch": 11.02,
+      "grad_norm": 12.089799880981445,
+      "learning_rate": 8.445079365079366e-06,
+      "loss": 0.6557,
+      "step": 4200
+    },
+    {
+      "epoch": 11.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.650736927986145,
+      "eval_runtime": 126.7199,
+      "eval_samples_per_second": 2.344,
+      "eval_steps_per_second": 1.176,
+      "step": 4200
+    },
+    {
+      "epoch": 12.000571428571428,
+      "grad_norm": 0.39973193407058716,
+      "learning_rate": 8.438730158730159e-06,
+      "loss": 0.4371,
+      "step": 4210
+    },
+    {
+      "epoch": 12.001142857142858,
+      "grad_norm": 11.976874351501465,
+      "learning_rate": 8.432380952380953e-06,
+      "loss": 0.4166,
+      "step": 4220
+    },
+    {
+      "epoch": 12.001714285714286,
+      "grad_norm": 0.3415261209011078,
+      "learning_rate": 8.426031746031746e-06,
+      "loss": 0.3675,
+      "step": 4230
+    },
+    {
+      "epoch": 12.002285714285714,
+      "grad_norm": 0.4809642732143402,
+      "learning_rate": 8.41968253968254e-06,
+      "loss": 0.2013,
+      "step": 4240
+    },
+    {
+      "epoch": 12.002857142857144,
+      "grad_norm": 0.682577908039093,
+      "learning_rate": 8.413333333333335e-06,
+      "loss": 0.2326,
+      "step": 4250
+    },
+    {
+      "epoch": 12.003428571428572,
+      "grad_norm": 0.03049817495048046,
+      "learning_rate": 8.406984126984127e-06,
+      "loss": 0.7267,
+      "step": 4260
+    },
+    {
+      "epoch": 12.004,
+      "grad_norm": 0.10638611763715744,
+      "learning_rate": 8.400634920634922e-06,
+      "loss": 0.8069,
+      "step": 4270
+    },
+    {
+      "epoch": 12.00457142857143,
+      "grad_norm": 0.041142355650663376,
+      "learning_rate": 8.394285714285714e-06,
+      "loss": 0.1764,
+      "step": 4280
+    },
+    {
+      "epoch": 12.005142857142857,
+      "grad_norm": 0.5415995121002197,
+      "learning_rate": 8.387936507936509e-06,
+      "loss": 1.0413,
+      "step": 4290
+    },
+    {
+      "epoch": 12.005714285714285,
+      "grad_norm": 0.4838305711746216,
+      "learning_rate": 8.381587301587303e-06,
+      "loss": 0.4062,
+      "step": 4300
+    },
+    {
+      "epoch": 12.006285714285715,
+      "grad_norm": 23.351930618286133,
+      "learning_rate": 8.375238095238096e-06,
+      "loss": 0.4338,
+      "step": 4310
+    },
+    {
+      "epoch": 12.006857142857143,
+      "grad_norm": 15.27731704711914,
+      "learning_rate": 8.36888888888889e-06,
+      "loss": 0.682,
+      "step": 4320
+    },
+    {
+      "epoch": 12.007428571428571,
+      "grad_norm": 0.21492832899093628,
+      "learning_rate": 8.362539682539683e-06,
+      "loss": 0.2962,
+      "step": 4330
+    },
+    {
+      "epoch": 12.008,
+      "grad_norm": 0.39623013138771057,
+      "learning_rate": 8.356190476190477e-06,
+      "loss": 0.4511,
+      "step": 4340
+    },
+    {
+      "epoch": 12.008571428571429,
+      "grad_norm": 0.8621640205383301,
+      "learning_rate": 8.34984126984127e-06,
+      "loss": 0.7211,
+      "step": 4350
+    },
+    {
+      "epoch": 12.009142857142857,
+      "grad_norm": 0.17645128071308136,
+      "learning_rate": 8.343492063492064e-06,
+      "loss": 0.609,
+      "step": 4360
+    },
+    {
+      "epoch": 12.009714285714285,
+      "grad_norm": 1.3618069887161255,
+      "learning_rate": 8.337142857142858e-06,
+      "loss": 0.1656,
+      "step": 4370
+    },
+    {
+      "epoch": 12.010285714285715,
+      "grad_norm": 12.654424667358398,
+      "learning_rate": 8.33079365079365e-06,
+      "loss": 0.715,
+      "step": 4380
+    },
+    {
+      "epoch": 12.010857142857143,
+      "grad_norm": 14.740551948547363,
+      "learning_rate": 8.324444444444445e-06,
+      "loss": 1.23,
+      "step": 4390
+    },
+    {
+      "epoch": 12.01142857142857,
+      "grad_norm": 33.95640182495117,
+      "learning_rate": 8.318095238095238e-06,
+      "loss": 0.6692,
+      "step": 4400
+    },
+    {
+      "epoch": 12.012,
+      "grad_norm": 7.696796417236328,
+      "learning_rate": 8.311746031746032e-06,
+      "loss": 0.8213,
+      "step": 4410
+    },
+    {
+      "epoch": 12.012571428571428,
+      "grad_norm": 0.5786144733428955,
+      "learning_rate": 8.305396825396826e-06,
+      "loss": 0.2544,
+      "step": 4420
+    },
+    {
+      "epoch": 12.013142857142856,
+      "grad_norm": 0.296314537525177,
+      "learning_rate": 8.29904761904762e-06,
+      "loss": 0.0166,
+      "step": 4430
+    },
+    {
+      "epoch": 12.013714285714286,
+      "grad_norm": 0.23542876541614532,
+      "learning_rate": 8.292698412698413e-06,
+      "loss": 0.2817,
+      "step": 4440
+    },
+    {
+      "epoch": 12.014285714285714,
+      "grad_norm": 16.194673538208008,
+      "learning_rate": 8.286349206349206e-06,
+      "loss": 0.7401,
+      "step": 4450
+    },
+    {
+      "epoch": 12.014857142857142,
+      "grad_norm": 0.1849094033241272,
+      "learning_rate": 8.28e-06,
+      "loss": 0.5863,
+      "step": 4460
+    },
+    {
+      "epoch": 12.015428571428572,
+      "grad_norm": 0.14007309079170227,
+      "learning_rate": 8.273650793650795e-06,
+      "loss": 1.4727,
+      "step": 4470
+    },
+    {
+      "epoch": 12.016,
+      "grad_norm": 74.8141098022461,
+      "learning_rate": 8.267301587301589e-06,
+      "loss": 0.4532,
+      "step": 4480
+    },
+    {
+      "epoch": 12.016571428571428,
+      "grad_norm": 0.1595589816570282,
+      "learning_rate": 8.260952380952382e-06,
+      "loss": 0.2142,
+      "step": 4490
+    },
+    {
+      "epoch": 12.017142857142858,
+      "grad_norm": 23.603548049926758,
+      "learning_rate": 8.254603174603176e-06,
+      "loss": 0.2174,
+      "step": 4500
+    },
+    {
+      "epoch": 12.017714285714286,
+      "grad_norm": 0.32662734389305115,
+      "learning_rate": 8.248253968253969e-06,
+      "loss": 0.4129,
+      "step": 4510
+    },
+    {
+      "epoch": 12.018285714285714,
+      "grad_norm": 15.74620532989502,
+      "learning_rate": 8.241904761904761e-06,
+      "loss": 0.5215,
+      "step": 4520
+    },
+    {
+      "epoch": 12.018857142857144,
+      "grad_norm": 0.10295995324850082,
+      "learning_rate": 8.235555555555557e-06,
+      "loss": 0.2909,
+      "step": 4530
+    },
+    {
+      "epoch": 12.019428571428572,
+      "grad_norm": 0.2319801300764084,
+      "learning_rate": 8.22920634920635e-06,
+      "loss": 1.1379,
+      "step": 4540
+    },
+    {
+      "epoch": 12.02,
+      "grad_norm": 0.1083206757903099,
+      "learning_rate": 8.222857142857144e-06,
+      "loss": 0.2642,
+      "step": 4550
+    },
+    {
+      "epoch": 12.02,
+      "eval_accuracy": 0.8518518518518519,
+      "eval_loss": 0.6555494070053101,
+      "eval_runtime": 127.2639,
+      "eval_samples_per_second": 2.334,
+      "eval_steps_per_second": 1.171,
+      "step": 4550
+    },
+    {
+      "epoch": 13.000571428571428,
+      "grad_norm": 0.09389404952526093,
+      "learning_rate": 8.216507936507937e-06,
+      "loss": 0.2617,
+      "step": 4560
+    },
+    {
+      "epoch": 13.001142857142858,
+      "grad_norm": 0.11154793202877045,
+      "learning_rate": 8.210158730158731e-06,
+      "loss": 0.0051,
+      "step": 4570
+    },
+    {
+      "epoch": 13.001714285714286,
+      "grad_norm": 0.1510355919599533,
+      "learning_rate": 8.203809523809526e-06,
+      "loss": 0.0096,
+      "step": 4580
+    },
+    {
+      "epoch": 13.002285714285714,
+      "grad_norm": 14.562790870666504,
+      "learning_rate": 8.197460317460318e-06,
+      "loss": 1.4925,
+      "step": 4590
+    },
+    {
+      "epoch": 13.002857142857144,
+      "grad_norm": 0.1331201046705246,
+      "learning_rate": 8.191111111111112e-06,
+      "loss": 0.5703,
+      "step": 4600
+    },
+    {
+      "epoch": 13.003428571428572,
+      "grad_norm": 0.2739226520061493,
+      "learning_rate": 8.184761904761905e-06,
+      "loss": 0.8987,
+      "step": 4610
+    },
+    {
+      "epoch": 13.004,
+      "grad_norm": 0.46435093879699707,
+      "learning_rate": 8.1784126984127e-06,
+      "loss": 0.4531,
+      "step": 4620
+    },
+    {
+      "epoch": 13.00457142857143,
+      "grad_norm": 0.17606890201568604,
+      "learning_rate": 8.172063492063492e-06,
+      "loss": 0.7506,
+      "step": 4630
+    },
+    {
+      "epoch": 13.005142857142857,
+      "grad_norm": 0.20931817591190338,
+      "learning_rate": 8.165714285714286e-06,
+      "loss": 0.2208,
+      "step": 4640
+    },
+    {
+      "epoch": 13.005714285714285,
+      "grad_norm": 0.1586388796567917,
+      "learning_rate": 8.15936507936508e-06,
+      "loss": 0.1573,
+      "step": 4650
+    },
+    {
+      "epoch": 13.006285714285715,
+      "grad_norm": 0.2133455127477646,
+      "learning_rate": 8.153015873015873e-06,
+      "loss": 1.0149,
+      "step": 4660
+    },
+    {
+      "epoch": 13.006857142857143,
+      "grad_norm": 0.10580915212631226,
+      "learning_rate": 8.146666666666668e-06,
+      "loss": 0.6209,
+      "step": 4670
+    },
+    {
+      "epoch": 13.007428571428571,
+      "grad_norm": 0.05809102952480316,
+      "learning_rate": 8.14031746031746e-06,
+      "loss": 0.4834,
+      "step": 4680
+    },
+    {
+      "epoch": 13.008,
+      "grad_norm": 0.18141770362854004,
+      "learning_rate": 8.133968253968255e-06,
+      "loss": 0.9549,
+      "step": 4690
+    },
+    {
+      "epoch": 13.008571428571429,
+      "grad_norm": 0.3100875914096832,
+      "learning_rate": 8.127619047619049e-06,
+      "loss": 0.4513,
+      "step": 4700
+    },
+    {
+      "epoch": 13.009142857142857,
+      "grad_norm": 12.605453491210938,
+      "learning_rate": 8.121269841269842e-06,
+      "loss": 0.9802,
+      "step": 4710
+    },
+    {
+      "epoch": 13.009714285714285,
+      "grad_norm": 18.329814910888672,
+      "learning_rate": 8.114920634920636e-06,
+      "loss": 0.4474,
+      "step": 4720
+    },
+    {
+      "epoch": 13.010285714285715,
+      "grad_norm": 0.18632696568965912,
+      "learning_rate": 8.108571428571429e-06,
+      "loss": 0.5227,
+      "step": 4730
+    },
+    {
+      "epoch": 13.010857142857143,
+      "grad_norm": 0.15728500485420227,
+      "learning_rate": 8.102222222222223e-06,
+      "loss": 0.3175,
+      "step": 4740
+    },
+    {
+      "epoch": 13.01142857142857,
+      "grad_norm": 16.559619903564453,
+      "learning_rate": 8.095873015873017e-06,
+      "loss": 0.4173,
+      "step": 4750
+    },
+    {
+      "epoch": 13.012,
+      "grad_norm": 0.04901667311787605,
+      "learning_rate": 8.08952380952381e-06,
+      "loss": 0.3976,
+      "step": 4760
+    },
+    {
+      "epoch": 13.012571428571428,
+      "grad_norm": 0.22993479669094086,
+      "learning_rate": 8.083174603174604e-06,
+      "loss": 0.1813,
+      "step": 4770
+    },
+    {
+      "epoch": 13.013142857142856,
+      "grad_norm": 0.3767968416213989,
+      "learning_rate": 8.076825396825397e-06,
+      "loss": 0.7901,
+      "step": 4780
+    },
+    {
+      "epoch": 13.013714285714286,
+      "grad_norm": 0.03023025207221508,
+      "learning_rate": 8.070476190476191e-06,
+      "loss": 0.1889,
+      "step": 4790
+    },
+    {
+      "epoch": 13.014285714285714,
+      "grad_norm": 0.9322307109832764,
+      "learning_rate": 8.064126984126984e-06,
+      "loss": 1.3871,
+      "step": 4800
+    },
+    {
+      "epoch": 13.014857142857142,
+      "grad_norm": 25.164409637451172,
+      "learning_rate": 8.057777777777778e-06,
+      "loss": 0.9618,
+      "step": 4810
+    },
+    {
+      "epoch": 13.015428571428572,
+      "grad_norm": 94.88408660888672,
+      "learning_rate": 8.051428571428573e-06,
+      "loss": 0.632,
+      "step": 4820
+    },
+    {
+      "epoch": 13.016,
+      "grad_norm": 0.15524975955486298,
+      "learning_rate": 8.045079365079365e-06,
+      "loss": 0.7624,
+      "step": 4830
+    },
+    {
+      "epoch": 13.016571428571428,
+      "grad_norm": 0.06509774178266525,
+      "learning_rate": 8.03873015873016e-06,
+      "loss": 0.2143,
+      "step": 4840
+    },
+    {
+      "epoch": 13.017142857142858,
+      "grad_norm": 0.24536223709583282,
+      "learning_rate": 8.032380952380952e-06,
+      "loss": 0.2895,
+      "step": 4850
+    },
+    {
+      "epoch": 13.017714285714286,
+      "grad_norm": 58.50824737548828,
+      "learning_rate": 8.026031746031746e-06,
+      "loss": 0.0836,
+      "step": 4860
+    },
+    {
+      "epoch": 13.018285714285714,
+      "grad_norm": 0.14553256332874298,
+      "learning_rate": 8.01968253968254e-06,
+      "loss": 0.672,
+      "step": 4870
+    },
+    {
+      "epoch": 13.018857142857144,
+      "grad_norm": 0.06745839864015579,
+      "learning_rate": 8.013333333333333e-06,
+      "loss": 0.6554,
+      "step": 4880
+    },
+    {
+      "epoch": 13.019428571428572,
+      "grad_norm": 0.15893854200839996,
+      "learning_rate": 8.006984126984128e-06,
+      "loss": 0.6895,
+      "step": 4890
+    },
+    {
+      "epoch": 13.02,
+      "grad_norm": 0.23196399211883545,
+      "learning_rate": 8.00063492063492e-06,
+      "loss": 0.2413,
+      "step": 4900
+    },
+    {
+      "epoch": 13.02,
+      "eval_accuracy": 0.8518518518518519,
+      "eval_loss": 0.6480634808540344,
+      "eval_runtime": 126.6312,
+      "eval_samples_per_second": 2.345,
+      "eval_steps_per_second": 1.177,
+      "step": 4900
+    },
+    {
+      "epoch": 14.000571428571428,
+      "grad_norm": 0.1645181030035019,
+      "learning_rate": 7.994285714285715e-06,
+      "loss": 0.6897,
+      "step": 4910
+    },
+    {
+      "epoch": 14.001142857142858,
+      "grad_norm": 168.658203125,
+      "learning_rate": 7.987936507936509e-06,
+      "loss": 0.8059,
+      "step": 4920
+    },
+    {
+      "epoch": 14.001714285714286,
+      "grad_norm": 0.17724451422691345,
+      "learning_rate": 7.981587301587302e-06,
+      "loss": 0.4341,
+      "step": 4930
+    },
+    {
+      "epoch": 14.002285714285714,
+      "grad_norm": 0.25565865635871887,
+      "learning_rate": 7.975238095238096e-06,
+      "loss": 0.588,
+      "step": 4940
+    },
+    {
+      "epoch": 14.002857142857144,
+      "grad_norm": 0.2059396207332611,
+      "learning_rate": 7.968888888888889e-06,
+      "loss": 0.2607,
+      "step": 4950
+    },
+    {
+      "epoch": 14.003428571428572,
+      "grad_norm": 21.05925941467285,
+      "learning_rate": 7.962539682539683e-06,
+      "loss": 0.676,
+      "step": 4960
+    },
+    {
+      "epoch": 14.004,
+      "grad_norm": 0.20337478816509247,
+      "learning_rate": 7.956190476190477e-06,
+      "loss": 0.235,
+      "step": 4970
+    },
+    {
+      "epoch": 14.00457142857143,
+      "grad_norm": 12.385817527770996,
+      "learning_rate": 7.949841269841272e-06,
+      "loss": 1.1436,
+      "step": 4980
+    },
+    {
+      "epoch": 14.005142857142857,
+      "grad_norm": 169.90371704101562,
+      "learning_rate": 7.943492063492064e-06,
+      "loss": 0.4043,
+      "step": 4990
+    },
+    {
+      "epoch": 14.005714285714285,
+      "grad_norm": 0.9865061044692993,
+      "learning_rate": 7.937142857142857e-06,
+      "loss": 0.5453,
+      "step": 5000
+    },
+    {
+      "epoch": 14.006285714285715,
+      "grad_norm": 0.21574608981609344,
+      "learning_rate": 7.930793650793651e-06,
+      "loss": 0.4185,
+      "step": 5010
+    },
+    {
+      "epoch": 14.006857142857143,
+      "grad_norm": 0.22090749442577362,
+      "learning_rate": 7.924444444444444e-06,
+      "loss": 0.6939,
+      "step": 5020
+    },
+    {
+      "epoch": 14.007428571428571,
+      "grad_norm": 0.02600650116801262,
+      "learning_rate": 7.91809523809524e-06,
+      "loss": 0.8623,
+      "step": 5030
+    },
+    {
+      "epoch": 14.008,
+      "grad_norm": 0.13779287040233612,
+      "learning_rate": 7.911746031746033e-06,
+      "loss": 0.0248,
+      "step": 5040
+    },
+    {
+      "epoch": 14.008571428571429,
+      "grad_norm": 12.554780960083008,
+      "learning_rate": 7.905396825396827e-06,
+      "loss": 0.9988,
+      "step": 5050
+    },
+    {
+      "epoch": 14.009142857142857,
+      "grad_norm": 0.41419580578804016,
+      "learning_rate": 7.89904761904762e-06,
+      "loss": 0.2302,
+      "step": 5060
+    },
+    {
+      "epoch": 14.009714285714285,
+      "grad_norm": 0.16763442754745483,
+      "learning_rate": 7.892698412698412e-06,
+      "loss": 0.8218,
+      "step": 5070
+    },
+    {
+      "epoch": 14.010285714285715,
+      "grad_norm": 120.92338562011719,
+      "learning_rate": 7.886349206349208e-06,
+      "loss": 0.5738,
+      "step": 5080
+    },
+    {
+      "epoch": 14.010857142857143,
+      "grad_norm": 0.15212488174438477,
+      "learning_rate": 7.88e-06,
+      "loss": 0.8477,
+      "step": 5090
+    },
+    {
+      "epoch": 14.01142857142857,
+      "grad_norm": 0.14688020944595337,
+      "learning_rate": 7.873650793650795e-06,
+      "loss": 0.2236,
+      "step": 5100
+    },
+    {
+      "epoch": 14.012,
+      "grad_norm": 0.11865352839231491,
+      "learning_rate": 7.867301587301588e-06,
+      "loss": 0.2869,
+      "step": 5110
+    },
+    {
+      "epoch": 14.012571428571428,
+      "grad_norm": 13.466479301452637,
+      "learning_rate": 7.860952380952382e-06,
+      "loss": 0.4885,
+      "step": 5120
+    },
+    {
+      "epoch": 14.013142857142856,
+      "grad_norm": 0.07300411909818649,
+      "learning_rate": 7.854603174603175e-06,
+      "loss": 0.2464,
+      "step": 5130
+    },
+    {
+      "epoch": 14.013714285714286,
+      "grad_norm": 0.052624981850385666,
+      "learning_rate": 7.848253968253969e-06,
+      "loss": 0.946,
+      "step": 5140
+    },
+    {
+      "epoch": 14.014285714285714,
+      "grad_norm": 0.09969048947095871,
+      "learning_rate": 7.841904761904763e-06,
+      "loss": 0.3992,
+      "step": 5150
+    },
+    {
+      "epoch": 14.014857142857142,
+      "grad_norm": 49.853572845458984,
+      "learning_rate": 7.835555555555556e-06,
+      "loss": 0.0232,
+      "step": 5160
+    },
+    {
+      "epoch": 14.015428571428572,
+      "grad_norm": 0.3982076644897461,
+      "learning_rate": 7.82920634920635e-06,
+      "loss": 0.3344,
+      "step": 5170
+    },
+    {
+      "epoch": 14.016,
+      "grad_norm": 1.8964109420776367,
+      "learning_rate": 7.822857142857143e-06,
+      "loss": 0.4479,
+      "step": 5180
+    },
+    {
+      "epoch": 14.016571428571428,
+      "grad_norm": 13.034674644470215,
+      "learning_rate": 7.816507936507937e-06,
+      "loss": 0.4343,
+      "step": 5190
+    },
+    {
+      "epoch": 14.017142857142858,
+      "grad_norm": 0.1686258316040039,
+      "learning_rate": 7.810158730158732e-06,
+      "loss": 0.4579,
+      "step": 5200
+    },
+    {
+      "epoch": 14.017714285714286,
+      "grad_norm": 0.14098533987998962,
+      "learning_rate": 7.803809523809524e-06,
+      "loss": 0.49,
+      "step": 5210
+    },
+    {
+      "epoch": 14.018285714285714,
+      "grad_norm": 13.026440620422363,
+      "learning_rate": 7.797460317460319e-06,
+      "loss": 0.2029,
+      "step": 5220
+    },
+    {
+      "epoch": 14.018857142857144,
+      "grad_norm": 12.317622184753418,
+      "learning_rate": 7.791111111111111e-06,
+      "loss": 0.7561,
+      "step": 5230
+    },
+    {
+      "epoch": 14.019428571428572,
+      "grad_norm": 0.3012208938598633,
+      "learning_rate": 7.784761904761906e-06,
+      "loss": 0.2343,
+      "step": 5240
+    },
+    {
+      "epoch": 14.02,
+      "grad_norm": 151.4730987548828,
+      "learning_rate": 7.7784126984127e-06,
+      "loss": 0.6278,
+      "step": 5250
+    },
+    {
+      "epoch": 14.02,
+      "eval_accuracy": 0.8552188552188552,
+      "eval_loss": 0.6555138826370239,
+      "eval_runtime": 126.3021,
+      "eval_samples_per_second": 2.352,
+      "eval_steps_per_second": 1.18,
+      "step": 5250
+    },
+    {
+      "epoch": 15.000571428571428,
+      "grad_norm": 0.2991902828216553,
+      "learning_rate": 7.772063492063493e-06,
+      "loss": 0.8981,
+      "step": 5260
+    },
+    {
+      "epoch": 15.001142857142858,
+      "grad_norm": 0.3251695930957794,
+      "learning_rate": 7.765714285714287e-06,
+      "loss": 0.552,
+      "step": 5270
+    },
+    {
+      "epoch": 15.001714285714286,
+      "grad_norm": 0.38806581497192383,
+      "learning_rate": 7.75936507936508e-06,
+      "loss": 0.8149,
+      "step": 5280
+    },
+    {
+      "epoch": 15.002285714285714,
+      "grad_norm": 1.2852102518081665,
+      "learning_rate": 7.753015873015874e-06,
+      "loss": 0.7302,
+      "step": 5290
+    },
+    {
+      "epoch": 15.002857142857144,
+      "grad_norm": 0.33824622631073,
+      "learning_rate": 7.746666666666666e-06,
+      "loss": 0.6803,
+      "step": 5300
+    },
+    {
+      "epoch": 15.003428571428572,
+      "grad_norm": 0.7059460282325745,
+      "learning_rate": 7.74031746031746e-06,
+      "loss": 0.8826,
+      "step": 5310
+    },
+    {
+      "epoch": 15.004,
+      "grad_norm": 0.17795021831989288,
+      "learning_rate": 7.733968253968255e-06,
+      "loss": 0.3081,
+      "step": 5320
+    },
+    {
+      "epoch": 15.00457142857143,
+      "grad_norm": 0.21305586397647858,
+      "learning_rate": 7.727619047619048e-06,
+      "loss": 0.4185,
+      "step": 5330
+    },
+    {
+      "epoch": 15.005142857142857,
+      "grad_norm": 0.458345890045166,
+      "learning_rate": 7.721269841269842e-06,
+      "loss": 0.6545,
+      "step": 5340
+    },
+    {
+      "epoch": 15.005714285714285,
+      "grad_norm": 0.11840073764324188,
+      "learning_rate": 7.714920634920635e-06,
+      "loss": 0.2937,
+      "step": 5350
+    },
+    {
+      "epoch": 15.006285714285715,
+      "grad_norm": 0.17106355726718903,
+      "learning_rate": 7.708571428571429e-06,
+      "loss": 0.2317,
+      "step": 5360
+    },
+    {
+      "epoch": 15.006857142857143,
+      "grad_norm": 0.09697124361991882,
+      "learning_rate": 7.702222222222223e-06,
+      "loss": 0.4099,
+      "step": 5370
+    },
+    {
+      "epoch": 15.007428571428571,
+      "grad_norm": 0.021129153668880463,
+      "learning_rate": 7.695873015873016e-06,
+      "loss": 0.2169,
+      "step": 5380
+    },
+    {
+      "epoch": 15.008,
+      "grad_norm": 0.4217149019241333,
+      "learning_rate": 7.68952380952381e-06,
+      "loss": 0.784,
+      "step": 5390
+    },
+    {
+      "epoch": 15.008571428571429,
+      "grad_norm": 0.17383898794651031,
+      "learning_rate": 7.683174603174603e-06,
+      "loss": 0.2046,
+      "step": 5400
+    },
+    {
+      "epoch": 15.009142857142857,
+      "grad_norm": 0.08642620593309402,
+      "learning_rate": 7.676825396825397e-06,
+      "loss": 0.2429,
+      "step": 5410
+    },
+    {
+      "epoch": 15.009714285714285,
+      "grad_norm": 13.887992858886719,
+      "learning_rate": 7.670476190476192e-06,
+      "loss": 0.9628,
+      "step": 5420
+    },
+    {
+      "epoch": 15.010285714285715,
+      "grad_norm": 0.249220609664917,
+      "learning_rate": 7.664126984126984e-06,
+      "loss": 0.5269,
+      "step": 5430
+    },
+    {
+      "epoch": 15.010857142857143,
+      "grad_norm": 0.07579600065946579,
+      "learning_rate": 7.657777777777779e-06,
+      "loss": 0.4449,
+      "step": 5440
+    },
+    {
+      "epoch": 15.01142857142857,
+      "grad_norm": 0.12848280370235443,
+      "learning_rate": 7.651428571428571e-06,
+      "loss": 0.245,
+      "step": 5450
+    },
+    {
+      "epoch": 15.012,
+      "grad_norm": 13.671841621398926,
+      "learning_rate": 7.645079365079366e-06,
+      "loss": 0.7608,
+      "step": 5460
+    },
+    {
+      "epoch": 15.012571428571428,
+      "grad_norm": 0.19747650623321533,
+      "learning_rate": 7.63873015873016e-06,
+      "loss": 0.3063,
+      "step": 5470
+    },
+    {
+      "epoch": 15.013142857142856,
+      "grad_norm": 0.2382912039756775,
+      "learning_rate": 7.632380952380953e-06,
+      "loss": 0.6737,
+      "step": 5480
+    },
+    {
+      "epoch": 15.013714285714286,
+      "grad_norm": 0.34730061888694763,
+      "learning_rate": 7.626031746031747e-06,
+      "loss": 0.3126,
+      "step": 5490
+    },
+    {
+      "epoch": 15.014285714285714,
+      "grad_norm": 12.084749221801758,
+      "learning_rate": 7.61968253968254e-06,
+      "loss": 0.4176,
+      "step": 5500
+    },
+    {
+      "epoch": 15.014857142857142,
+      "grad_norm": 0.24228954315185547,
+      "learning_rate": 7.613333333333334e-06,
+      "loss": 0.4131,
+      "step": 5510
+    },
+    {
+      "epoch": 15.015428571428572,
+      "grad_norm": 40.641319274902344,
+      "learning_rate": 7.606984126984127e-06,
+      "loss": 0.5213,
+      "step": 5520
+    },
+    {
+      "epoch": 15.016,
+      "grad_norm": 13.184736251831055,
+      "learning_rate": 7.600634920634922e-06,
+      "loss": 0.7655,
+      "step": 5530
+    },
+    {
+      "epoch": 15.016571428571428,
+      "grad_norm": 0.10150767862796783,
+      "learning_rate": 7.594285714285715e-06,
+      "loss": 0.7638,
+      "step": 5540
+    },
+    {
+      "epoch": 15.017142857142858,
+      "grad_norm": 0.16732892394065857,
+      "learning_rate": 7.587936507936509e-06,
+      "loss": 0.2141,
+      "step": 5550
+    },
+    {
+      "epoch": 15.017714285714286,
+      "grad_norm": 0.31903275847435,
+      "learning_rate": 7.581587301587302e-06,
+      "loss": 0.44,
+      "step": 5560
+    },
+    {
+      "epoch": 15.018285714285714,
+      "grad_norm": 0.04827267304062843,
+      "learning_rate": 7.575238095238096e-06,
+      "loss": 0.0074,
+      "step": 5570
+    },
+    {
+      "epoch": 15.018857142857144,
+      "grad_norm": 0.05066690221428871,
+      "learning_rate": 7.56888888888889e-06,
+      "loss": 1.0622,
+      "step": 5580
+    },
+    {
+      "epoch": 15.019428571428572,
+      "grad_norm": 0.14147022366523743,
+      "learning_rate": 7.562539682539683e-06,
+      "loss": 0.5052,
+      "step": 5590
+    },
+    {
+      "epoch": 15.02,
+      "grad_norm": 0.4057610332965851,
+      "learning_rate": 7.556190476190477e-06,
+      "loss": 0.0107,
+      "step": 5600
+    },
+    {
+      "epoch": 15.02,
+      "eval_accuracy": 0.8518518518518519,
+      "eval_loss": 0.6550477147102356,
+      "eval_runtime": 126.4968,
+      "eval_samples_per_second": 2.348,
+      "eval_steps_per_second": 1.178,
+      "step": 5600
+    },
+    {
+      "epoch": 16.00057142857143,
+      "grad_norm": 0.12237854301929474,
+      "learning_rate": 7.54984126984127e-06,
+      "loss": 0.0925,
+      "step": 5610
+    },
+    {
+      "epoch": 16.001142857142856,
+      "grad_norm": 0.1167961061000824,
+      "learning_rate": 7.543492063492064e-06,
+      "loss": 0.6409,
+      "step": 5620
+    },
+    {
+      "epoch": 16.001714285714286,
+      "grad_norm": 0.33305051922798157,
+      "learning_rate": 7.537142857142857e-06,
+      "loss": 0.961,
+      "step": 5630
+    },
+    {
+      "epoch": 16.002285714285716,
+      "grad_norm": 0.10626853257417679,
+      "learning_rate": 7.530793650793652e-06,
+      "loss": 1.2608,
+      "step": 5640
+    },
+    {
+      "epoch": 16.002857142857142,
+      "grad_norm": 67.93856048583984,
+      "learning_rate": 7.524444444444445e-06,
+      "loss": 0.7227,
+      "step": 5650
+    },
+    {
+      "epoch": 16.00342857142857,
+      "grad_norm": 0.09368494153022766,
+      "learning_rate": 7.518095238095239e-06,
+      "loss": 0.0237,
+      "step": 5660
+    },
+    {
+      "epoch": 16.004,
+      "grad_norm": 0.1382751762866974,
+      "learning_rate": 7.511746031746032e-06,
+      "loss": 0.4677,
+      "step": 5670
+    },
+    {
+      "epoch": 16.004571428571428,
+      "grad_norm": 0.31585273146629333,
+      "learning_rate": 7.505396825396826e-06,
+      "loss": 0.8341,
+      "step": 5680
+    },
+    {
+      "epoch": 16.005142857142857,
+      "grad_norm": 0.2557847797870636,
+      "learning_rate": 7.499047619047619e-06,
+      "loss": 0.2347,
+      "step": 5690
+    },
+    {
+      "epoch": 16.005714285714287,
+      "grad_norm": 0.14112916588783264,
+      "learning_rate": 7.492698412698413e-06,
+      "loss": 0.7506,
+      "step": 5700
+    },
+    {
+      "epoch": 16.006285714285713,
+      "grad_norm": 109.31643676757812,
+      "learning_rate": 7.486349206349207e-06,
+      "loss": 0.6344,
+      "step": 5710
+    },
+    {
+      "epoch": 16.006857142857143,
+      "grad_norm": 0.407155305147171,
+      "learning_rate": 7.48e-06,
+      "loss": 0.2471,
+      "step": 5720
+    },
+    {
+      "epoch": 16.007428571428573,
+      "grad_norm": 0.037655897438526154,
+      "learning_rate": 7.473650793650794e-06,
+      "loss": 0.4707,
+      "step": 5730
+    },
+    {
+      "epoch": 16.008,
+      "grad_norm": 0.08762380480766296,
+      "learning_rate": 7.467301587301587e-06,
+      "loss": 0.0635,
+      "step": 5740
+    },
+    {
+      "epoch": 16.00857142857143,
+      "grad_norm": 0.15955060720443726,
+      "learning_rate": 7.460952380952382e-06,
+      "loss": 0.5162,
+      "step": 5750
+    },
+    {
+      "epoch": 16.00914285714286,
+      "grad_norm": 13.772893905639648,
+      "learning_rate": 7.454603174603175e-06,
+      "loss": 0.6237,
+      "step": 5760
+    },
+    {
+      "epoch": 16.009714285714285,
+      "grad_norm": 20.581626892089844,
+      "learning_rate": 7.448253968253969e-06,
+      "loss": 0.7149,
+      "step": 5770
+    },
+    {
+      "epoch": 16.010285714285715,
+      "grad_norm": 0.14489834010601044,
+      "learning_rate": 7.441904761904762e-06,
+      "loss": 0.0079,
+      "step": 5780
+    },
+    {
+      "epoch": 16.010857142857144,
+      "grad_norm": 0.1383170783519745,
+      "learning_rate": 7.435555555555556e-06,
+      "loss": 0.2527,
+      "step": 5790
+    },
+    {
+      "epoch": 16.01142857142857,
+      "grad_norm": 68.98406219482422,
+      "learning_rate": 7.429206349206349e-06,
+      "loss": 0.3456,
+      "step": 5800
+    },
+    {
+      "epoch": 16.012,
+      "grad_norm": 0.11179056763648987,
+      "learning_rate": 7.422857142857144e-06,
+      "loss": 0.0647,
+      "step": 5810
+    },
+    {
+      "epoch": 16.01257142857143,
+      "grad_norm": 0.18071648478507996,
+      "learning_rate": 7.416507936507937e-06,
+      "loss": 0.5227,
+      "step": 5820
+    },
+    {
+      "epoch": 16.013142857142856,
+      "grad_norm": 67.35832214355469,
+      "learning_rate": 7.41015873015873e-06,
+      "loss": 0.3971,
+      "step": 5830
+    },
+    {
+      "epoch": 16.013714285714286,
+      "grad_norm": 0.1837083250284195,
+      "learning_rate": 7.403809523809524e-06,
+      "loss": 0.8512,
+      "step": 5840
+    },
+    {
+      "epoch": 16.014285714285716,
+      "grad_norm": 0.038320187479257584,
+      "learning_rate": 7.397460317460317e-06,
+      "loss": 0.0037,
+      "step": 5850
+    },
+    {
+      "epoch": 16.014857142857142,
+      "grad_norm": 20.66285514831543,
+      "learning_rate": 7.3911111111111125e-06,
+      "loss": 1.123,
+      "step": 5860
+    },
+    {
+      "epoch": 16.015428571428572,
+      "grad_norm": 133.61224365234375,
+      "learning_rate": 7.384761904761906e-06,
+      "loss": 0.3835,
+      "step": 5870
+    },
+    {
+      "epoch": 16.016,
+      "grad_norm": 0.17678341269493103,
+      "learning_rate": 7.3784126984126995e-06,
+      "loss": 0.1001,
+      "step": 5880
+    },
+    {
+      "epoch": 16.016571428571428,
+      "grad_norm": 0.09736228734254837,
+      "learning_rate": 7.372063492063493e-06,
+      "loss": 0.2866,
+      "step": 5890
+    },
+    {
+      "epoch": 16.017142857142858,
+      "grad_norm": 0.09368955343961716,
+      "learning_rate": 7.365714285714286e-06,
+      "loss": 1.0441,
+      "step": 5900
+    },
+    {
+      "epoch": 16.017714285714284,
+      "grad_norm": 0.10775759816169739,
+      "learning_rate": 7.359365079365079e-06,
+      "loss": 0.3626,
+      "step": 5910
+    },
+    {
+      "epoch": 16.018285714285714,
+      "grad_norm": 0.2658800482749939,
+      "learning_rate": 7.353015873015874e-06,
+      "loss": 0.3755,
+      "step": 5920
+    },
+    {
+      "epoch": 16.018857142857144,
+      "grad_norm": 0.24060992896556854,
+      "learning_rate": 7.346666666666668e-06,
+      "loss": 1.028,
+      "step": 5930
+    },
+    {
+      "epoch": 16.01942857142857,
+      "grad_norm": 0.12600265443325043,
+      "learning_rate": 7.340317460317461e-06,
+      "loss": 0.5612,
+      "step": 5940
+    },
+    {
+      "epoch": 16.02,
+      "grad_norm": 0.12872040271759033,
+      "learning_rate": 7.333968253968255e-06,
+      "loss": 0.3013,
+      "step": 5950
+    },
+    {
+      "epoch": 16.02,
+      "eval_accuracy": 0.8484848484848485,
+      "eval_loss": 0.7404991984367371,
+      "eval_runtime": 126.748,
+      "eval_samples_per_second": 2.343,
+      "eval_steps_per_second": 1.176,
+      "step": 5950
+    },
+    {
+      "epoch": 17.00057142857143,
+      "grad_norm": 0.23484548926353455,
+      "learning_rate": 7.327619047619048e-06,
+      "loss": 0.4457,
+      "step": 5960
+    },
+    {
+      "epoch": 17.001142857142856,
+      "grad_norm": 0.14240051805973053,
+      "learning_rate": 7.3212698412698425e-06,
+      "loss": 0.0042,
+      "step": 5970
+    },
+    {
+      "epoch": 17.001714285714286,
+      "grad_norm": 0.15564581751823425,
+      "learning_rate": 7.314920634920636e-06,
+      "loss": 0.6894,
+      "step": 5980
+    },
+    {
+      "epoch": 17.002285714285716,
+      "grad_norm": 127.26374816894531,
+      "learning_rate": 7.3085714285714295e-06,
+      "loss": 1.2778,
+      "step": 5990
+    },
+    {
+      "epoch": 17.002857142857142,
+      "grad_norm": 0.1205485537648201,
+      "learning_rate": 7.302222222222223e-06,
+      "loss": 0.2591,
+      "step": 6000
+    },
+    {
+      "epoch": 17.00342857142857,
+      "grad_norm": 0.14839039742946625,
+      "learning_rate": 7.2958730158730165e-06,
+      "loss": 0.2446,
+      "step": 6010
+    },
+    {
+      "epoch": 17.004,
+      "grad_norm": 0.2482178807258606,
+      "learning_rate": 7.28952380952381e-06,
+      "loss": 1.1471,
+      "step": 6020
+    },
+    {
+      "epoch": 17.004571428571428,
+      "grad_norm": 0.6684224009513855,
+      "learning_rate": 7.283174603174604e-06,
+      "loss": 0.1935,
+      "step": 6030
+    },
+    {
+      "epoch": 17.005142857142857,
+      "grad_norm": 0.31114834547042847,
+      "learning_rate": 7.276825396825398e-06,
+      "loss": 0.374,
+      "step": 6040
+    },
+    {
+      "epoch": 17.005714285714287,
+      "grad_norm": 0.241834819316864,
+      "learning_rate": 7.270476190476191e-06,
+      "loss": 0.932,
+      "step": 6050
+    },
+    {
+      "epoch": 17.006285714285713,
+      "grad_norm": 17.917661666870117,
+      "learning_rate": 7.264126984126985e-06,
+      "loss": 0.5023,
+      "step": 6060
+    },
+    {
+      "epoch": 17.006857142857143,
+      "grad_norm": 0.11860672384500504,
+      "learning_rate": 7.257777777777778e-06,
+      "loss": 0.2441,
+      "step": 6070
+    },
+    {
+      "epoch": 17.007428571428573,
+      "grad_norm": 0.12156573683023453,
+      "learning_rate": 7.251428571428572e-06,
+      "loss": 0.4809,
+      "step": 6080
+    },
+    {
+      "epoch": 17.008,
+      "grad_norm": 32.46079635620117,
+      "learning_rate": 7.245079365079366e-06,
+      "loss": 1.0452,
+      "step": 6090
+    },
+    {
+      "epoch": 17.00857142857143,
+      "grad_norm": 1.8947296142578125,
+      "learning_rate": 7.2387301587301595e-06,
+      "loss": 0.2181,
+      "step": 6100
+    },
+    {
+      "epoch": 17.00914285714286,
+      "grad_norm": 1.0444539785385132,
+      "learning_rate": 7.232380952380953e-06,
+      "loss": 0.2129,
+      "step": 6110
+    },
+    {
+      "epoch": 17.009714285714285,
+      "grad_norm": 9.648567199707031,
+      "learning_rate": 7.2260317460317465e-06,
+      "loss": 0.2156,
+      "step": 6120
+    },
+    {
+      "epoch": 17.010285714285715,
+      "grad_norm": 0.24711275100708008,
+      "learning_rate": 7.21968253968254e-06,
+      "loss": 0.0108,
+      "step": 6130
+    },
+    {
+      "epoch": 17.010857142857144,
+      "grad_norm": 0.22460851073265076,
+      "learning_rate": 7.213333333333334e-06,
+      "loss": 0.645,
+      "step": 6140
+    },
+    {
+      "epoch": 17.01142857142857,
+      "grad_norm": 49.333404541015625,
+      "learning_rate": 7.206984126984128e-06,
+      "loss": 0.6497,
+      "step": 6150
+    },
+    {
+      "epoch": 17.012,
+      "grad_norm": 0.22554242610931396,
+      "learning_rate": 7.200634920634921e-06,
+      "loss": 0.7637,
+      "step": 6160
+    },
+    {
+      "epoch": 17.01257142857143,
+      "grad_norm": 0.2542533576488495,
+      "learning_rate": 7.194285714285715e-06,
+      "loss": 0.6032,
+      "step": 6170
+    },
+    {
+      "epoch": 17.013142857142856,
+      "grad_norm": 0.03277142718434334,
+      "learning_rate": 7.187936507936508e-06,
+      "loss": 0.4751,
+      "step": 6180
+    },
+    {
+      "epoch": 17.013714285714286,
+      "grad_norm": 0.09158936887979507,
+      "learning_rate": 7.181587301587302e-06,
+      "loss": 0.433,
+      "step": 6190
+    },
+    {
+      "epoch": 17.014285714285716,
+      "grad_norm": 0.2807057201862335,
+      "learning_rate": 7.175238095238096e-06,
+      "loss": 0.2971,
+      "step": 6200
+    },
+    {
+      "epoch": 17.014857142857142,
+      "grad_norm": 80.08293914794922,
+      "learning_rate": 7.1688888888888895e-06,
+      "loss": 0.4443,
+      "step": 6210
+    },
+    {
+      "epoch": 17.015428571428572,
+      "grad_norm": 0.19456472992897034,
+      "learning_rate": 7.162539682539683e-06,
+      "loss": 0.5821,
+      "step": 6220
+    },
+    {
+      "epoch": 17.016,
+      "grad_norm": 0.07997258752584457,
+      "learning_rate": 7.1561904761904765e-06,
+      "loss": 1.0668,
+      "step": 6230
+    },
+    {
+      "epoch": 17.016571428571428,
+      "grad_norm": 0.1534426212310791,
+      "learning_rate": 7.14984126984127e-06,
+      "loss": 0.4562,
+      "step": 6240
+    },
+    {
+      "epoch": 17.017142857142858,
+      "grad_norm": 0.36759310960769653,
+      "learning_rate": 7.143492063492064e-06,
+      "loss": 0.2592,
+      "step": 6250
+    },
+    {
+      "epoch": 17.017714285714284,
+      "grad_norm": 0.15657585859298706,
+      "learning_rate": 7.137142857142858e-06,
+      "loss": 0.5768,
+      "step": 6260
+    },
+    {
+      "epoch": 17.018285714285714,
+      "grad_norm": 0.13546155393123627,
+      "learning_rate": 7.130793650793651e-06,
+      "loss": 0.2564,
+      "step": 6270
+    },
+    {
+      "epoch": 17.018857142857144,
+      "grad_norm": 0.14879682660102844,
+      "learning_rate": 7.124444444444445e-06,
+      "loss": 0.987,
+      "step": 6280
+    },
+    {
+      "epoch": 17.01942857142857,
+      "grad_norm": 20.40616226196289,
+      "learning_rate": 7.118095238095238e-06,
+      "loss": 0.8061,
+      "step": 6290
+    },
+    {
+      "epoch": 17.02,
+      "grad_norm": 0.3881203234195709,
+      "learning_rate": 7.111746031746032e-06,
+      "loss": 0.5055,
+      "step": 6300
+    },
+    {
+      "epoch": 17.02,
+      "eval_accuracy": 0.8451178451178452,
+      "eval_loss": 0.6562865376472473,
+      "eval_runtime": 127.2408,
+      "eval_samples_per_second": 2.334,
+      "eval_steps_per_second": 1.171,
+      "step": 6300
+    },
+    {
+      "epoch": 18.00057142857143,
+      "grad_norm": 1.636939287185669,
+      "learning_rate": 7.105396825396826e-06,
+      "loss": 0.9626,
+      "step": 6310
+    },
+    {
+      "epoch": 18.001142857142856,
+      "grad_norm": 1.2696601152420044,
+      "learning_rate": 7.0990476190476195e-06,
+      "loss": 1.1254,
+      "step": 6320
+    },
+    {
+      "epoch": 18.001714285714286,
+      "grad_norm": 31.090442657470703,
+      "learning_rate": 7.092698412698413e-06,
+      "loss": 0.2333,
+      "step": 6330
+    },
+    {
+      "epoch": 18.002285714285716,
+      "grad_norm": 122.07534790039062,
+      "learning_rate": 7.0863492063492065e-06,
+      "loss": 0.3172,
+      "step": 6340
+    },
+    {
+      "epoch": 18.002857142857142,
+      "grad_norm": 0.2747611701488495,
+      "learning_rate": 7.08e-06,
+      "loss": 0.3612,
+      "step": 6350
+    },
+    {
+      "epoch": 18.00342857142857,
+      "grad_norm": 0.3476632535457611,
+      "learning_rate": 7.073650793650795e-06,
+      "loss": 0.8514,
+      "step": 6360
+    },
+    {
+      "epoch": 18.004,
+      "grad_norm": 0.03347006067633629,
+      "learning_rate": 7.067301587301589e-06,
+      "loss": 0.3275,
+      "step": 6370
+    },
+    {
+      "epoch": 18.004571428571428,
+      "grad_norm": 84.09701538085938,
+      "learning_rate": 7.060952380952381e-06,
+      "loss": 0.5278,
+      "step": 6380
+    },
+    {
+      "epoch": 18.005142857142857,
+      "grad_norm": 0.21676389873027802,
+      "learning_rate": 7.054603174603175e-06,
+      "loss": 0.612,
+      "step": 6390
+    },
+    {
+      "epoch": 18.005714285714287,
+      "grad_norm": 160.529541015625,
+      "learning_rate": 7.048253968253968e-06,
+      "loss": 0.4884,
+      "step": 6400
+    },
+    {
+      "epoch": 18.006285714285713,
+      "grad_norm": 0.364534467458725,
+      "learning_rate": 7.041904761904762e-06,
+      "loss": 0.0083,
+      "step": 6410
+    },
+    {
+      "epoch": 18.006857142857143,
+      "grad_norm": 0.11234438419342041,
+      "learning_rate": 7.035555555555557e-06,
+      "loss": 0.0106,
+      "step": 6420
+    },
+    {
+      "epoch": 18.007428571428573,
+      "grad_norm": 1.1077744960784912,
+      "learning_rate": 7.02920634920635e-06,
+      "loss": 0.0066,
+      "step": 6430
+    },
+    {
+      "epoch": 18.008,
+      "grad_norm": 0.13264243304729462,
+      "learning_rate": 7.022857142857144e-06,
+      "loss": 0.4173,
+      "step": 6440
+    },
+    {
+      "epoch": 18.00857142857143,
+      "grad_norm": 0.1175365149974823,
+      "learning_rate": 7.016507936507937e-06,
+      "loss": 0.6565,
+      "step": 6450
+    },
+    {
+      "epoch": 18.00914285714286,
+      "grad_norm": 0.2617458701133728,
+      "learning_rate": 7.01015873015873e-06,
+      "loss": 0.2389,
+      "step": 6460
+    },
+    {
+      "epoch": 18.009714285714285,
+      "grad_norm": 46.34348678588867,
+      "learning_rate": 7.0038095238095235e-06,
+      "loss": 0.4956,
+      "step": 6470
+    },
+    {
+      "epoch": 18.010285714285715,
+      "grad_norm": 153.98577880859375,
+      "learning_rate": 6.997460317460319e-06,
+      "loss": 1.0497,
+      "step": 6480
+    },
+    {
+      "epoch": 18.010857142857144,
+      "grad_norm": 0.19409751892089844,
+      "learning_rate": 6.991111111111112e-06,
+      "loss": 0.412,
+      "step": 6490
+    },
+    {
+      "epoch": 18.01142857142857,
+      "grad_norm": 0.12201213091611862,
+      "learning_rate": 6.984761904761906e-06,
+      "loss": 0.4613,
+      "step": 6500
+    },
+    {
+      "epoch": 18.012,
+      "grad_norm": 0.020613886415958405,
+      "learning_rate": 6.978412698412699e-06,
+      "loss": 0.3496,
+      "step": 6510
+    },
+    {
+      "epoch": 18.01257142857143,
+      "grad_norm": 0.1668790876865387,
+      "learning_rate": 6.9720634920634926e-06,
+      "loss": 0.4584,
+      "step": 6520
+    },
+    {
+      "epoch": 18.013142857142856,
+      "grad_norm": 0.15408121049404144,
+      "learning_rate": 6.965714285714287e-06,
+      "loss": 0.6383,
+      "step": 6530
+    },
+    {
+      "epoch": 18.013714285714286,
+      "grad_norm": 12.476696968078613,
+      "learning_rate": 6.95936507936508e-06,
+      "loss": 0.4078,
+      "step": 6540
+    },
+    {
+      "epoch": 18.014285714285716,
+      "grad_norm": 0.240266352891922,
+      "learning_rate": 6.953015873015874e-06,
+      "loss": 0.4204,
+      "step": 6550
+    },
+    {
+      "epoch": 18.014857142857142,
+      "grad_norm": 0.20230837166309357,
+      "learning_rate": 6.946666666666667e-06,
+      "loss": 0.4396,
+      "step": 6560
+    },
+    {
+      "epoch": 18.015428571428572,
+      "grad_norm": 33.412925720214844,
+      "learning_rate": 6.940317460317461e-06,
+      "loss": 0.7438,
+      "step": 6570
+    },
+    {
+      "epoch": 18.016,
+      "grad_norm": 21.571928024291992,
+      "learning_rate": 6.933968253968254e-06,
+      "loss": 0.4501,
+      "step": 6580
+    },
+    {
+      "epoch": 18.016571428571428,
+      "grad_norm": 0.3260933458805084,
+      "learning_rate": 6.927619047619049e-06,
+      "loss": 0.1881,
+      "step": 6590
+    },
+    {
+      "epoch": 18.017142857142858,
+      "grad_norm": 0.07090619206428528,
+      "learning_rate": 6.921269841269842e-06,
+      "loss": 0.7295,
+      "step": 6600
+    },
+    {
+      "epoch": 18.017714285714284,
+      "grad_norm": 0.2709910571575165,
+      "learning_rate": 6.914920634920636e-06,
+      "loss": 0.0077,
+      "step": 6610
+    },
+    {
+      "epoch": 18.018285714285714,
+      "grad_norm": 0.2425944209098816,
+      "learning_rate": 6.908571428571429e-06,
+      "loss": 0.7675,
+      "step": 6620
+    },
+    {
+      "epoch": 18.018857142857144,
+      "grad_norm": 12.433034896850586,
+      "learning_rate": 6.902222222222223e-06,
+      "loss": 0.5643,
+      "step": 6630
+    },
+    {
+      "epoch": 18.01942857142857,
+      "grad_norm": 0.23004376888275146,
+      "learning_rate": 6.895873015873017e-06,
+      "loss": 0.2302,
+      "step": 6640
+    },
+    {
+      "epoch": 18.02,
+      "grad_norm": 0.1142105832695961,
+      "learning_rate": 6.88952380952381e-06,
+      "loss": 0.0059,
+      "step": 6650
+    },
+    {
+      "epoch": 18.02,
+      "eval_accuracy": 0.8484848484848485,
+      "eval_loss": 0.6916897296905518,
+      "eval_runtime": 126.3823,
+      "eval_samples_per_second": 2.35,
+      "eval_steps_per_second": 1.179,
+      "step": 6650
+    },
+    {
+      "epoch": 19.00057142857143,
+      "grad_norm": 0.010271835140883923,
+      "learning_rate": 6.883174603174604e-06,
+      "loss": 0.158,
+      "step": 6660
+    },
+    {
+      "epoch": 19.001142857142856,
+      "grad_norm": 15.551984786987305,
+      "learning_rate": 6.876825396825397e-06,
+      "loss": 0.7788,
+      "step": 6670
+    },
+    {
+      "epoch": 19.001714285714286,
+      "grad_norm": 0.12954816222190857,
+      "learning_rate": 6.870476190476191e-06,
+      "loss": 0.3656,
+      "step": 6680
+    },
+    {
+      "epoch": 19.002285714285716,
+      "grad_norm": 4.656391620635986,
+      "learning_rate": 6.864126984126984e-06,
+      "loss": 0.689,
+      "step": 6690
+    },
+    {
+      "epoch": 19.002857142857142,
+      "grad_norm": 0.13569696247577667,
+      "learning_rate": 6.857777777777779e-06,
+      "loss": 0.1878,
+      "step": 6700
+    },
+    {
+      "epoch": 19.00342857142857,
+      "grad_norm": 0.16348059475421906,
+      "learning_rate": 6.851428571428572e-06,
+      "loss": 0.8311,
+      "step": 6710
+    },
+    {
+      "epoch": 19.004,
+      "grad_norm": 12.783551216125488,
+      "learning_rate": 6.845079365079366e-06,
+      "loss": 0.46,
+      "step": 6720
+    },
+    {
+      "epoch": 19.004571428571428,
+      "grad_norm": 0.24606812000274658,
+      "learning_rate": 6.838730158730159e-06,
+      "loss": 0.587,
+      "step": 6730
+    },
+    {
+      "epoch": 19.005142857142857,
+      "grad_norm": 0.12452604621648788,
+      "learning_rate": 6.832380952380953e-06,
+      "loss": 0.0415,
+      "step": 6740
+    },
+    {
+      "epoch": 19.005714285714287,
+      "grad_norm": 14.892534255981445,
+      "learning_rate": 6.826031746031747e-06,
+      "loss": 0.7738,
+      "step": 6750
+    },
+    {
+      "epoch": 19.006285714285713,
+      "grad_norm": 0.1208721324801445,
+      "learning_rate": 6.81968253968254e-06,
+      "loss": 0.2166,
+      "step": 6760
+    },
+    {
+      "epoch": 19.006857142857143,
+      "grad_norm": 0.14853811264038086,
+      "learning_rate": 6.813333333333334e-06,
+      "loss": 0.0052,
+      "step": 6770
+    },
+    {
+      "epoch": 19.007428571428573,
+      "grad_norm": 0.3613605201244354,
+      "learning_rate": 6.806984126984127e-06,
+      "loss": 0.2232,
+      "step": 6780
+    },
+    {
+      "epoch": 19.008,
+      "grad_norm": 11.996315956115723,
+      "learning_rate": 6.800634920634921e-06,
+      "loss": 0.4984,
+      "step": 6790
+    },
+    {
+      "epoch": 19.00857142857143,
+      "grad_norm": 11.95859146118164,
+      "learning_rate": 6.794285714285714e-06,
+      "loss": 0.525,
+      "step": 6800
+    },
+    {
+      "epoch": 19.00914285714286,
+      "grad_norm": 0.1372225433588028,
+      "learning_rate": 6.787936507936509e-06,
+      "loss": 0.0046,
+      "step": 6810
+    },
+    {
+      "epoch": 19.009714285714285,
+      "grad_norm": 13.401540756225586,
+      "learning_rate": 6.781587301587302e-06,
+      "loss": 0.2888,
+      "step": 6820
+    },
+    {
+      "epoch": 19.010285714285715,
+      "grad_norm": 0.1333763599395752,
+      "learning_rate": 6.775238095238096e-06,
+      "loss": 0.4673,
+      "step": 6830
+    },
+    {
+      "epoch": 19.010857142857144,
+      "grad_norm": 0.28985071182250977,
+      "learning_rate": 6.768888888888889e-06,
+      "loss": 0.4904,
+      "step": 6840
+    },
+    {
+      "epoch": 19.01142857142857,
+      "grad_norm": 0.5672123432159424,
+      "learning_rate": 6.762539682539683e-06,
+      "loss": 0.38,
+      "step": 6850
+    },
+    {
+      "epoch": 19.012,
+      "grad_norm": 0.18549109995365143,
+      "learning_rate": 6.756190476190476e-06,
+      "loss": 0.4654,
+      "step": 6860
+    },
+    {
+      "epoch": 19.01257142857143,
+      "grad_norm": 0.005712473299354315,
+      "learning_rate": 6.74984126984127e-06,
+      "loss": 0.4693,
+      "step": 6870
+    },
+    {
+      "epoch": 19.013142857142856,
+      "grad_norm": 0.20795473456382751,
+      "learning_rate": 6.743492063492064e-06,
+      "loss": 0.6987,
+      "step": 6880
+    },
+    {
+      "epoch": 19.013714285714286,
+      "grad_norm": 0.2660701274871826,
+      "learning_rate": 6.737142857142857e-06,
+      "loss": 0.3575,
+      "step": 6890
+    },
+    {
+      "epoch": 19.014285714285716,
+      "grad_norm": 12.896868705749512,
+      "learning_rate": 6.730793650793651e-06,
+      "loss": 1.0003,
+      "step": 6900
+    },
+    {
+      "epoch": 19.014857142857142,
+      "grad_norm": 90.52599334716797,
+      "learning_rate": 6.724444444444444e-06,
+      "loss": 0.5481,
+      "step": 6910
+    },
+    {
+      "epoch": 19.015428571428572,
+      "grad_norm": 0.11832074075937271,
+      "learning_rate": 6.7180952380952395e-06,
+      "loss": 0.461,
+      "step": 6920
+    },
+    {
+      "epoch": 19.016,
+      "grad_norm": 0.021740248426795006,
+      "learning_rate": 6.711746031746032e-06,
+      "loss": 0.94,
+      "step": 6930
+    },
+    {
+      "epoch": 19.016571428571428,
+      "grad_norm": 0.14676664769649506,
+      "learning_rate": 6.705396825396826e-06,
+      "loss": 0.2698,
+      "step": 6940
+    },
+    {
+      "epoch": 19.017142857142858,
+      "grad_norm": 0.01525693666189909,
+      "learning_rate": 6.699047619047619e-06,
+      "loss": 0.5851,
+      "step": 6950
+    },
+    {
+      "epoch": 19.017714285714284,
+      "grad_norm": 0.5535984039306641,
+      "learning_rate": 6.692698412698413e-06,
+      "loss": 0.3717,
+      "step": 6960
+    },
+    {
+      "epoch": 19.018285714285714,
+      "grad_norm": 0.43303757905960083,
+      "learning_rate": 6.686349206349206e-06,
+      "loss": 0.5912,
+      "step": 6970
+    },
+    {
+      "epoch": 19.018857142857144,
+      "grad_norm": 0.15641167759895325,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.2061,
+      "step": 6980
+    },
+    {
+      "epoch": 19.01942857142857,
+      "grad_norm": 0.11748301982879639,
+      "learning_rate": 6.673650793650795e-06,
+      "loss": 0.326,
+      "step": 6990
+    },
+    {
+      "epoch": 19.02,
+      "grad_norm": 0.05292058736085892,
+      "learning_rate": 6.667301587301588e-06,
+      "loss": 0.4332,
+      "step": 7000
+    },
+    {
+      "epoch": 19.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.6888241767883301,
+      "eval_runtime": 134.2388,
+      "eval_samples_per_second": 2.212,
+      "eval_steps_per_second": 1.11,
+      "step": 7000
+    },
+    {
+      "epoch": 20.00057142857143,
+      "grad_norm": 0.4945196211338043,
+      "learning_rate": 6.660952380952381e-06,
+      "loss": 0.103,
+      "step": 7010
+    },
+    {
+      "epoch": 20.001142857142856,
+      "grad_norm": 16.183570861816406,
+      "learning_rate": 6.654603174603174e-06,
+      "loss": 0.2379,
+      "step": 7020
+    },
+    {
+      "epoch": 20.001714285714286,
+      "grad_norm": 0.07635916769504547,
+      "learning_rate": 6.6482539682539695e-06,
+      "loss": 0.6419,
+      "step": 7030
+    },
+    {
+      "epoch": 20.002285714285716,
+      "grad_norm": 159.50653076171875,
+      "learning_rate": 6.641904761904763e-06,
+      "loss": 0.5751,
+      "step": 7040
+    },
+    {
+      "epoch": 20.002857142857142,
+      "grad_norm": 0.09001730382442474,
+      "learning_rate": 6.6355555555555565e-06,
+      "loss": 0.254,
+      "step": 7050
+    },
+    {
+      "epoch": 20.00342857142857,
+      "grad_norm": 0.07919283211231232,
+      "learning_rate": 6.62920634920635e-06,
+      "loss": 0.7664,
+      "step": 7060
+    },
+    {
+      "epoch": 20.004,
+      "grad_norm": 280.9998779296875,
+      "learning_rate": 6.6228571428571435e-06,
+      "loss": 0.2373,
+      "step": 7070
+    },
+    {
+      "epoch": 20.004571428571428,
+      "grad_norm": 0.3689022362232208,
+      "learning_rate": 6.616507936507937e-06,
+      "loss": 0.4658,
+      "step": 7080
+    },
+    {
+      "epoch": 20.005142857142857,
+      "grad_norm": 0.10695375502109528,
+      "learning_rate": 6.610158730158731e-06,
+      "loss": 0.3698,
+      "step": 7090
+    },
+    {
+      "epoch": 20.005714285714287,
+      "grad_norm": 0.13924677670001984,
+      "learning_rate": 6.603809523809525e-06,
+      "loss": 0.5151,
+      "step": 7100
+    },
+    {
+      "epoch": 20.006285714285713,
+      "grad_norm": 0.05997047573328018,
+      "learning_rate": 6.597460317460318e-06,
+      "loss": 0.3443,
+      "step": 7110
+    },
+    {
+      "epoch": 20.006857142857143,
+      "grad_norm": 15.691723823547363,
+      "learning_rate": 6.591111111111112e-06,
+      "loss": 0.9827,
+      "step": 7120
+    },
+    {
+      "epoch": 20.007428571428573,
+      "grad_norm": 0.18234623968601227,
+      "learning_rate": 6.584761904761905e-06,
+      "loss": 0.5862,
+      "step": 7130
+    },
+    {
+      "epoch": 20.008,
+      "grad_norm": 0.27075400948524475,
+      "learning_rate": 6.5784126984126995e-06,
+      "loss": 0.1941,
+      "step": 7140
+    },
+    {
+      "epoch": 20.00857142857143,
+      "grad_norm": 1.7042484283447266,
+      "learning_rate": 6.572063492063493e-06,
+      "loss": 0.5991,
+      "step": 7150
+    },
+    {
+      "epoch": 20.00914285714286,
+      "grad_norm": 0.16708026826381683,
+      "learning_rate": 6.5657142857142865e-06,
+      "loss": 0.3161,
+      "step": 7160
+    },
+    {
+      "epoch": 20.009714285714285,
+      "grad_norm": 0.2559516131877899,
+      "learning_rate": 6.55936507936508e-06,
+      "loss": 0.3776,
+      "step": 7170
+    },
+    {
+      "epoch": 20.010285714285715,
+      "grad_norm": 0.19110074639320374,
+      "learning_rate": 6.5530158730158735e-06,
+      "loss": 0.9548,
+      "step": 7180
+    },
+    {
+      "epoch": 20.010857142857144,
+      "grad_norm": 56.03752517700195,
+      "learning_rate": 6.546666666666667e-06,
+      "loss": 0.2375,
+      "step": 7190
+    },
+    {
+      "epoch": 20.01142857142857,
+      "grad_norm": 16.357572555541992,
+      "learning_rate": 6.540317460317461e-06,
+      "loss": 0.5034,
+      "step": 7200
+    },
+    {
+      "epoch": 20.012,
+      "grad_norm": 0.40159788727760315,
+      "learning_rate": 6.533968253968255e-06,
+      "loss": 0.4687,
+      "step": 7210
+    },
+    {
+      "epoch": 20.01257142857143,
+      "grad_norm": 0.009314529597759247,
+      "learning_rate": 6.527619047619048e-06,
+      "loss": 0.4526,
+      "step": 7220
+    },
+    {
+      "epoch": 20.013142857142856,
+      "grad_norm": 0.13374114036560059,
+      "learning_rate": 6.521269841269842e-06,
+      "loss": 0.9903,
+      "step": 7230
+    },
+    {
+      "epoch": 20.013714285714286,
+      "grad_norm": 0.020518776029348373,
+      "learning_rate": 6.514920634920635e-06,
+      "loss": 0.0067,
+      "step": 7240
+    },
+    {
+      "epoch": 20.014285714285716,
+      "grad_norm": 0.15277713537216187,
+      "learning_rate": 6.5085714285714295e-06,
+      "loss": 0.0422,
+      "step": 7250
+    },
+    {
+      "epoch": 20.014857142857142,
+      "grad_norm": 0.10491526871919632,
+      "learning_rate": 6.502222222222223e-06,
+      "loss": 0.0047,
+      "step": 7260
+    },
+    {
+      "epoch": 20.015428571428572,
+      "grad_norm": 15.871146202087402,
+      "learning_rate": 6.4958730158730165e-06,
+      "loss": 0.2825,
+      "step": 7270
+    },
+    {
+      "epoch": 20.016,
+      "grad_norm": 0.11634642630815506,
+      "learning_rate": 6.48952380952381e-06,
+      "loss": 0.0032,
+      "step": 7280
+    },
+    {
+      "epoch": 20.016571428571428,
+      "grad_norm": 0.014390116557478905,
+      "learning_rate": 6.4831746031746035e-06,
+      "loss": 0.6739,
+      "step": 7290
+    },
+    {
+      "epoch": 20.017142857142858,
+      "grad_norm": 0.012950708158314228,
+      "learning_rate": 6.476825396825397e-06,
+      "loss": 1.2164,
+      "step": 7300
+    },
+    {
+      "epoch": 20.017714285714284,
+      "grad_norm": 0.13078835606575012,
+      "learning_rate": 6.470476190476191e-06,
+      "loss": 0.212,
+      "step": 7310
+    },
+    {
+      "epoch": 20.018285714285714,
+      "grad_norm": 0.28521332144737244,
+      "learning_rate": 6.464126984126985e-06,
+      "loss": 0.6134,
+      "step": 7320
+    },
+    {
+      "epoch": 20.018857142857144,
+      "grad_norm": 0.14261774718761444,
+      "learning_rate": 6.457777777777778e-06,
+      "loss": 0.0948,
+      "step": 7330
+    },
+    {
+      "epoch": 20.01942857142857,
+      "grad_norm": 0.2758195400238037,
+      "learning_rate": 6.451428571428572e-06,
+      "loss": 0.1811,
+      "step": 7340
+    },
+    {
+      "epoch": 20.02,
+      "grad_norm": 0.05165260285139084,
+      "learning_rate": 6.445079365079365e-06,
+      "loss": 0.2602,
+      "step": 7350
+    },
+    {
+      "epoch": 20.02,
+      "eval_accuracy": 0.8417508417508418,
+      "eval_loss": 0.7993361353874207,
+      "eval_runtime": 133.8895,
+      "eval_samples_per_second": 2.218,
+      "eval_steps_per_second": 1.113,
+      "step": 7350
+    },
+    {
+      "epoch": 21.00057142857143,
+      "grad_norm": 0.03576577454805374,
+      "learning_rate": 6.438730158730159e-06,
+      "loss": 0.2528,
+      "step": 7360
+    },
+    {
+      "epoch": 21.001142857142856,
+      "grad_norm": 0.05446856468915939,
+      "learning_rate": 6.432380952380953e-06,
+      "loss": 0.0034,
+      "step": 7370
+    },
+    {
+      "epoch": 21.001714285714286,
+      "grad_norm": 50.52811050415039,
+      "learning_rate": 6.4260317460317465e-06,
+      "loss": 1.0569,
+      "step": 7380
+    },
+    {
+      "epoch": 21.002285714285716,
+      "grad_norm": 26.091224670410156,
+      "learning_rate": 6.41968253968254e-06,
+      "loss": 0.2889,
+      "step": 7390
+    },
+    {
+      "epoch": 21.002857142857142,
+      "grad_norm": 0.09547077119350433,
+      "learning_rate": 6.4133333333333335e-06,
+      "loss": 0.0366,
+      "step": 7400
+    },
+    {
+      "epoch": 21.00342857142857,
+      "grad_norm": 0.08614878356456757,
+      "learning_rate": 6.406984126984127e-06,
+      "loss": 0.2737,
+      "step": 7410
+    },
+    {
+      "epoch": 21.004,
+      "grad_norm": 0.03283363953232765,
+      "learning_rate": 6.400634920634921e-06,
+      "loss": 0.0022,
+      "step": 7420
+    },
+    {
+      "epoch": 21.004571428571428,
+      "grad_norm": 4.929975509643555,
+      "learning_rate": 6.394285714285715e-06,
+      "loss": 0.5315,
+      "step": 7430
+    },
+    {
+      "epoch": 21.005142857142857,
+      "grad_norm": 106.62588500976562,
+      "learning_rate": 6.387936507936508e-06,
+      "loss": 0.4519,
+      "step": 7440
+    },
+    {
+      "epoch": 21.005714285714287,
+      "grad_norm": 0.04590607061982155,
+      "learning_rate": 6.381587301587302e-06,
+      "loss": 0.4353,
+      "step": 7450
+    },
+    {
+      "epoch": 21.006285714285713,
+      "grad_norm": 0.10316039621829987,
+      "learning_rate": 6.375238095238095e-06,
+      "loss": 0.2364,
+      "step": 7460
+    },
+    {
+      "epoch": 21.006857142857143,
+      "grad_norm": 28.117801666259766,
+      "learning_rate": 6.368888888888889e-06,
+      "loss": 0.6987,
+      "step": 7470
+    },
+    {
+      "epoch": 21.007428571428573,
+      "grad_norm": 0.1716347187757492,
+      "learning_rate": 6.362539682539683e-06,
+      "loss": 1.1197,
+      "step": 7480
+    },
+    {
+      "epoch": 21.008,
+      "grad_norm": 17.753265380859375,
+      "learning_rate": 6.3561904761904765e-06,
+      "loss": 0.1678,
+      "step": 7490
+    },
+    {
+      "epoch": 21.00857142857143,
+      "grad_norm": 0.021444451063871384,
+      "learning_rate": 6.34984126984127e-06,
+      "loss": 0.1589,
+      "step": 7500
+    },
+    {
+      "epoch": 21.00914285714286,
+      "grad_norm": 0.013452350161969662,
+      "learning_rate": 6.3434920634920635e-06,
+      "loss": 0.4614,
+      "step": 7510
+    },
+    {
+      "epoch": 21.009714285714285,
+      "grad_norm": 0.023600058630108833,
+      "learning_rate": 6.337142857142857e-06,
+      "loss": 0.25,
+      "step": 7520
+    },
+    {
+      "epoch": 21.010285714285715,
+      "grad_norm": 0.13624207675457,
+      "learning_rate": 6.330793650793652e-06,
+      "loss": 0.334,
+      "step": 7530
+    },
+    {
+      "epoch": 21.010857142857144,
+      "grad_norm": 0.15217389166355133,
+      "learning_rate": 6.324444444444446e-06,
+      "loss": 0.3943,
+      "step": 7540
+    },
+    {
+      "epoch": 21.01142857142857,
+      "grad_norm": 0.11624295264482498,
+      "learning_rate": 6.318095238095239e-06,
+      "loss": 0.0209,
+      "step": 7550
+    },
+    {
+      "epoch": 21.012,
+      "grad_norm": 0.5223240256309509,
+      "learning_rate": 6.311746031746033e-06,
+      "loss": 0.4057,
+      "step": 7560
+    },
+    {
+      "epoch": 21.01257142857143,
+      "grad_norm": 0.059054210782051086,
+      "learning_rate": 6.305396825396825e-06,
+      "loss": 0.8236,
+      "step": 7570
+    },
+    {
+      "epoch": 21.013142857142856,
+      "grad_norm": 0.0346602126955986,
+      "learning_rate": 6.299047619047619e-06,
+      "loss": 0.4062,
+      "step": 7580
+    },
+    {
+      "epoch": 21.013714285714286,
+      "grad_norm": 0.2160838097333908,
+      "learning_rate": 6.292698412698414e-06,
+      "loss": 0.7491,
+      "step": 7590
+    },
+    {
+      "epoch": 21.014285714285716,
+      "grad_norm": 0.17393529415130615,
+      "learning_rate": 6.286349206349207e-06,
+      "loss": 0.0512,
+      "step": 7600
+    },
+    {
+      "epoch": 21.014857142857142,
+      "grad_norm": 12.681178092956543,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.6529,
+      "step": 7610
+    },
+    {
+      "epoch": 21.015428571428572,
+      "grad_norm": 0.008576265536248684,
+      "learning_rate": 6.273650793650794e-06,
+      "loss": 0.8273,
+      "step": 7620
+    },
+    {
+      "epoch": 21.016,
+      "grad_norm": 0.006474177818745375,
+      "learning_rate": 6.267301587301588e-06,
+      "loss": 0.0061,
+      "step": 7630
+    },
+    {
+      "epoch": 21.016571428571428,
+      "grad_norm": 0.42665791511535645,
+      "learning_rate": 6.260952380952382e-06,
+      "loss": 0.924,
+      "step": 7640
+    },
+    {
+      "epoch": 21.017142857142858,
+      "grad_norm": 37.20500946044922,
+      "learning_rate": 6.254603174603176e-06,
+      "loss": 0.2282,
+      "step": 7650
+    },
+    {
+      "epoch": 21.017714285714284,
+      "grad_norm": 0.03610096871852875,
+      "learning_rate": 6.248253968253969e-06,
+      "loss": 0.2507,
+      "step": 7660
+    },
+    {
+      "epoch": 21.018285714285714,
+      "grad_norm": 0.17031346261501312,
+      "learning_rate": 6.241904761904763e-06,
+      "loss": 0.483,
+      "step": 7670
+    },
+    {
+      "epoch": 21.018857142857144,
+      "grad_norm": 13.083247184753418,
+      "learning_rate": 6.235555555555556e-06,
+      "loss": 0.3114,
+      "step": 7680
+    },
+    {
+      "epoch": 21.01942857142857,
+      "grad_norm": 0.013108008541166782,
+      "learning_rate": 6.2292063492063496e-06,
+      "loss": 0.9822,
+      "step": 7690
+    },
+    {
+      "epoch": 21.02,
+      "grad_norm": 0.19977082312107086,
+      "learning_rate": 6.222857142857144e-06,
+      "loss": 0.2142,
+      "step": 7700
+    },
+    {
+      "epoch": 21.02,
+      "eval_accuracy": 0.8451178451178452,
+      "eval_loss": 0.7130899429321289,
+      "eval_runtime": 134.5619,
+      "eval_samples_per_second": 2.207,
+      "eval_steps_per_second": 1.107,
+      "step": 7700
+    },
+    {
+      "epoch": 22.00057142857143,
+      "grad_norm": 0.12899892032146454,
+      "learning_rate": 6.216507936507937e-06,
+      "loss": 0.3791,
+      "step": 7710
+    },
+    {
+      "epoch": 22.001142857142856,
+      "grad_norm": 0.06919983774423599,
+      "learning_rate": 6.210158730158731e-06,
+      "loss": 0.4267,
+      "step": 7720
+    },
+    {
+      "epoch": 22.001714285714286,
+      "grad_norm": 0.082898810505867,
+      "learning_rate": 6.203809523809524e-06,
+      "loss": 0.2237,
+      "step": 7730
+    },
+    {
+      "epoch": 22.002285714285716,
+      "grad_norm": 0.08086587488651276,
+      "learning_rate": 6.197460317460318e-06,
+      "loss": 0.3335,
+      "step": 7740
+    },
+    {
+      "epoch": 22.002857142857142,
+      "grad_norm": 0.007148704957216978,
+      "learning_rate": 6.191111111111111e-06,
+      "loss": 0.261,
+      "step": 7750
+    },
+    {
+      "epoch": 22.00342857142857,
+      "grad_norm": 13.431198120117188,
+      "learning_rate": 6.184761904761906e-06,
+      "loss": 0.9781,
+      "step": 7760
+    },
+    {
+      "epoch": 22.004,
+      "grad_norm": 0.16154451668262482,
+      "learning_rate": 6.178412698412699e-06,
+      "loss": 0.0853,
+      "step": 7770
+    },
+    {
+      "epoch": 22.004571428571428,
+      "grad_norm": 0.14190372824668884,
+      "learning_rate": 6.172063492063493e-06,
+      "loss": 0.035,
+      "step": 7780
+    },
+    {
+      "epoch": 22.005142857142857,
+      "grad_norm": 13.651422500610352,
+      "learning_rate": 6.165714285714286e-06,
+      "loss": 0.7239,
+      "step": 7790
+    },
+    {
+      "epoch": 22.005714285714287,
+      "grad_norm": 0.16447778046131134,
+      "learning_rate": 6.15936507936508e-06,
+      "loss": 0.3368,
+      "step": 7800
+    },
+    {
+      "epoch": 22.006285714285713,
+      "grad_norm": 0.36849430203437805,
+      "learning_rate": 6.153015873015874e-06,
+      "loss": 0.2213,
+      "step": 7810
+    },
+    {
+      "epoch": 22.006857142857143,
+      "grad_norm": 0.5472358465194702,
+      "learning_rate": 6.146666666666667e-06,
+      "loss": 0.7931,
+      "step": 7820
+    },
+    {
+      "epoch": 22.007428571428573,
+      "grad_norm": 0.8773165941238403,
+      "learning_rate": 6.140317460317461e-06,
+      "loss": 0.1896,
+      "step": 7830
+    },
+    {
+      "epoch": 22.008,
+      "grad_norm": 0.20025408267974854,
+      "learning_rate": 6.133968253968254e-06,
+      "loss": 0.72,
+      "step": 7840
+    },
+    {
+      "epoch": 22.00857142857143,
+      "grad_norm": 0.0833997055888176,
+      "learning_rate": 6.127619047619048e-06,
+      "loss": 0.0071,
+      "step": 7850
+    },
+    {
+      "epoch": 22.00914285714286,
+      "grad_norm": 0.018751641735434532,
+      "learning_rate": 6.121269841269841e-06,
+      "loss": 0.2319,
+      "step": 7860
+    },
+    {
+      "epoch": 22.009714285714285,
+      "grad_norm": 12.105188369750977,
+      "learning_rate": 6.114920634920636e-06,
+      "loss": 0.7782,
+      "step": 7870
+    },
+    {
+      "epoch": 22.010285714285715,
+      "grad_norm": 0.0960543304681778,
+      "learning_rate": 6.108571428571429e-06,
+      "loss": 0.0327,
+      "step": 7880
+    },
+    {
+      "epoch": 22.010857142857144,
+      "grad_norm": 12.760798454284668,
+      "learning_rate": 6.102222222222223e-06,
+      "loss": 0.5455,
+      "step": 7890
+    },
+    {
+      "epoch": 22.01142857142857,
+      "grad_norm": 0.19047430157661438,
+      "learning_rate": 6.095873015873016e-06,
+      "loss": 0.1856,
+      "step": 7900
+    },
+    {
+      "epoch": 22.012,
+      "grad_norm": 12.85616683959961,
+      "learning_rate": 6.08952380952381e-06,
+      "loss": 0.2649,
+      "step": 7910
+    },
+    {
+      "epoch": 22.01257142857143,
+      "grad_norm": 0.1496925801038742,
+      "learning_rate": 6.083174603174604e-06,
+      "loss": 0.756,
+      "step": 7920
+    },
+    {
+      "epoch": 22.013142857142856,
+      "grad_norm": 0.1137370839715004,
+      "learning_rate": 6.076825396825397e-06,
+      "loss": 1.1056,
+      "step": 7930
+    },
+    {
+      "epoch": 22.013714285714286,
+      "grad_norm": 14.750025749206543,
+      "learning_rate": 6.070476190476191e-06,
+      "loss": 0.5389,
+      "step": 7940
+    },
+    {
+      "epoch": 22.014285714285716,
+      "grad_norm": 0.15537060797214508,
+      "learning_rate": 6.064126984126984e-06,
+      "loss": 0.2203,
+      "step": 7950
+    },
+    {
+      "epoch": 22.014857142857142,
+      "grad_norm": 0.17418305575847626,
+      "learning_rate": 6.057777777777778e-06,
+      "loss": 0.5758,
+      "step": 7960
+    },
+    {
+      "epoch": 22.015428571428572,
+      "grad_norm": 0.3206419050693512,
+      "learning_rate": 6.051428571428571e-06,
+      "loss": 0.0278,
+      "step": 7970
+    },
+    {
+      "epoch": 22.016,
+      "grad_norm": 0.10264712572097778,
+      "learning_rate": 6.045079365079366e-06,
+      "loss": 0.4427,
+      "step": 7980
+    },
+    {
+      "epoch": 22.016571428571428,
+      "grad_norm": 0.08854754269123077,
+      "learning_rate": 6.038730158730159e-06,
+      "loss": 0.1925,
+      "step": 7990
+    },
+    {
+      "epoch": 22.017142857142858,
+      "grad_norm": 0.1728421151638031,
+      "learning_rate": 6.032380952380953e-06,
+      "loss": 0.3517,
+      "step": 8000
+    },
+    {
+      "epoch": 22.017714285714284,
+      "grad_norm": 0.07346148788928986,
+      "learning_rate": 6.026031746031746e-06,
+      "loss": 0.1056,
+      "step": 8010
+    },
+    {
+      "epoch": 22.018285714285714,
+      "grad_norm": 20.421218872070312,
+      "learning_rate": 6.01968253968254e-06,
+      "loss": 0.7583,
+      "step": 8020
+    },
+    {
+      "epoch": 22.018857142857144,
+      "grad_norm": 0.0952727198600769,
+      "learning_rate": 6.013333333333335e-06,
+      "loss": 0.3924,
+      "step": 8030
+    },
+    {
+      "epoch": 22.01942857142857,
+      "grad_norm": 139.4682159423828,
+      "learning_rate": 6.006984126984127e-06,
+      "loss": 0.1965,
+      "step": 8040
+    },
+    {
+      "epoch": 22.02,
+      "grad_norm": 0.15866141021251678,
+      "learning_rate": 6.000634920634921e-06,
+      "loss": 0.5742,
+      "step": 8050
+    },
+    {
+      "epoch": 22.02,
+      "eval_accuracy": 0.797979797979798,
+      "eval_loss": 0.9735172986984253,
+      "eval_runtime": 134.4584,
+      "eval_samples_per_second": 2.209,
+      "eval_steps_per_second": 1.108,
+      "step": 8050
+    },
+    {
+      "epoch": 23.00057142857143,
+      "grad_norm": 0.1783732771873474,
+      "learning_rate": 5.994285714285714e-06,
+      "loss": 0.1797,
+      "step": 8060
+    },
+    {
+      "epoch": 23.001142857142856,
+      "grad_norm": 0.5633581876754761,
+      "learning_rate": 5.987936507936508e-06,
+      "loss": 0.0058,
+      "step": 8070
+    },
+    {
+      "epoch": 23.001714285714286,
+      "grad_norm": 0.052664969116449356,
+      "learning_rate": 5.981587301587301e-06,
+      "loss": 0.3705,
+      "step": 8080
+    },
+    {
+      "epoch": 23.002285714285716,
+      "grad_norm": 1.9416977167129517,
+      "learning_rate": 5.9752380952380965e-06,
+      "loss": 0.8398,
+      "step": 8090
+    },
+    {
+      "epoch": 23.002857142857142,
+      "grad_norm": 2.0996170043945312,
+      "learning_rate": 5.96888888888889e-06,
+      "loss": 0.5296,
+      "step": 8100
+    },
+    {
+      "epoch": 23.00342857142857,
+      "grad_norm": 0.32839661836624146,
+      "learning_rate": 5.9625396825396835e-06,
+      "loss": 0.2443,
+      "step": 8110
+    },
+    {
+      "epoch": 23.004,
+      "grad_norm": 0.007853704504668713,
+      "learning_rate": 5.956190476190476e-06,
+      "loss": 0.0044,
+      "step": 8120
+    },
+    {
+      "epoch": 23.004571428571428,
+      "grad_norm": 12.491561889648438,
+      "learning_rate": 5.94984126984127e-06,
+      "loss": 0.4636,
+      "step": 8130
+    },
+    {
+      "epoch": 23.005142857142857,
+      "grad_norm": 0.10937748849391937,
+      "learning_rate": 5.943492063492063e-06,
+      "loss": 0.664,
+      "step": 8140
+    },
+    {
+      "epoch": 23.005714285714287,
+      "grad_norm": 0.1766849309206009,
+      "learning_rate": 5.937142857142858e-06,
+      "loss": 0.4253,
+      "step": 8150
+    },
+    {
+      "epoch": 23.006285714285713,
+      "grad_norm": 55.246822357177734,
+      "learning_rate": 5.930793650793652e-06,
+      "loss": 0.5592,
+      "step": 8160
+    },
+    {
+      "epoch": 23.006857142857143,
+      "grad_norm": 1.2145743370056152,
+      "learning_rate": 5.924444444444445e-06,
+      "loss": 0.0061,
+      "step": 8170
+    },
+    {
+      "epoch": 23.007428571428573,
+      "grad_norm": 5.3756794929504395,
+      "learning_rate": 5.918095238095239e-06,
+      "loss": 0.2953,
+      "step": 8180
+    },
+    {
+      "epoch": 23.008,
+      "grad_norm": 3.9541385173797607,
+      "learning_rate": 5.911746031746032e-06,
+      "loss": 0.5192,
+      "step": 8190
+    },
+    {
+      "epoch": 23.00857142857143,
+      "grad_norm": 0.25989338755607605,
+      "learning_rate": 5.9053968253968265e-06,
+      "loss": 0.4489,
+      "step": 8200
+    },
+    {
+      "epoch": 23.00914285714286,
+      "grad_norm": 0.16681231558322906,
+      "learning_rate": 5.89904761904762e-06,
+      "loss": 0.5296,
+      "step": 8210
+    },
+    {
+      "epoch": 23.009714285714285,
+      "grad_norm": 0.2039841264486313,
+      "learning_rate": 5.8926984126984135e-06,
+      "loss": 0.5324,
+      "step": 8220
+    },
+    {
+      "epoch": 23.010285714285715,
+      "grad_norm": 0.3426172137260437,
+      "learning_rate": 5.886349206349207e-06,
+      "loss": 0.663,
+      "step": 8230
+    },
+    {
+      "epoch": 23.010857142857144,
+      "grad_norm": 0.007993980310857296,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.2322,
+      "step": 8240
+    },
+    {
+      "epoch": 23.01142857142857,
+      "grad_norm": 0.28963732719421387,
+      "learning_rate": 5.873650793650794e-06,
+      "loss": 0.2086,
+      "step": 8250
+    },
+    {
+      "epoch": 23.012,
+      "grad_norm": 0.18354713916778564,
+      "learning_rate": 5.867301587301588e-06,
+      "loss": 0.0054,
+      "step": 8260
+    },
+    {
+      "epoch": 23.01257142857143,
+      "grad_norm": 0.2161131054162979,
+      "learning_rate": 5.860952380952382e-06,
+      "loss": 0.0039,
+      "step": 8270
+    },
+    {
+      "epoch": 23.013142857142856,
+      "grad_norm": 0.11740400642156601,
+      "learning_rate": 5.854603174603175e-06,
+      "loss": 0.4064,
+      "step": 8280
+    },
+    {
+      "epoch": 23.013714285714286,
+      "grad_norm": 150.37094116210938,
+      "learning_rate": 5.848253968253969e-06,
+      "loss": 0.0248,
+      "step": 8290
+    },
+    {
+      "epoch": 23.014285714285716,
+      "grad_norm": 0.1555628478527069,
+      "learning_rate": 5.841904761904762e-06,
+      "loss": 0.0036,
+      "step": 8300
+    },
+    {
+      "epoch": 23.014857142857142,
+      "grad_norm": 0.13599923253059387,
+      "learning_rate": 5.8355555555555565e-06,
+      "loss": 0.2426,
+      "step": 8310
+    },
+    {
+      "epoch": 23.015428571428572,
+      "grad_norm": 0.1228395327925682,
+      "learning_rate": 5.82920634920635e-06,
+      "loss": 0.8494,
+      "step": 8320
+    },
+    {
+      "epoch": 23.016,
+      "grad_norm": 295.5143127441406,
+      "learning_rate": 5.8228571428571435e-06,
+      "loss": 0.459,
+      "step": 8330
+    },
+    {
+      "epoch": 23.016571428571428,
+      "grad_norm": 14.874744415283203,
+      "learning_rate": 5.816507936507937e-06,
+      "loss": 0.4832,
+      "step": 8340
+    },
+    {
+      "epoch": 23.017142857142858,
+      "grad_norm": 0.10557269304990768,
+      "learning_rate": 5.8101587301587305e-06,
+      "loss": 0.2448,
+      "step": 8350
+    },
+    {
+      "epoch": 23.017714285714284,
+      "grad_norm": 0.11553741991519928,
+      "learning_rate": 5.803809523809524e-06,
+      "loss": 0.1697,
+      "step": 8360
+    },
+    {
+      "epoch": 23.018285714285714,
+      "grad_norm": 0.08351312577724457,
+      "learning_rate": 5.797460317460318e-06,
+      "loss": 0.4602,
+      "step": 8370
+    },
+    {
+      "epoch": 23.018857142857144,
+      "grad_norm": 6.845589637756348,
+      "learning_rate": 5.791111111111112e-06,
+      "loss": 0.3568,
+      "step": 8380
+    },
+    {
+      "epoch": 23.01942857142857,
+      "grad_norm": 0.0753844752907753,
+      "learning_rate": 5.784761904761905e-06,
+      "loss": 0.0102,
+      "step": 8390
+    },
+    {
+      "epoch": 23.02,
+      "grad_norm": 0.12463897466659546,
+      "learning_rate": 5.778412698412699e-06,
+      "loss": 0.2504,
+      "step": 8400
+    },
+    {
+      "epoch": 23.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.8313712477684021,
+      "eval_runtime": 134.2181,
+      "eval_samples_per_second": 2.213,
+      "eval_steps_per_second": 1.11,
+      "step": 8400
+    },
+    {
+      "epoch": 24.00057142857143,
+      "grad_norm": 0.052970390766859055,
+      "learning_rate": 5.772063492063492e-06,
+      "loss": 0.2138,
+      "step": 8410
+    },
+    {
+      "epoch": 24.001142857142856,
+      "grad_norm": 0.05925741419196129,
+      "learning_rate": 5.7657142857142865e-06,
+      "loss": 0.2863,
+      "step": 8420
+    },
+    {
+      "epoch": 24.001714285714286,
+      "grad_norm": 0.0537530817091465,
+      "learning_rate": 5.75936507936508e-06,
+      "loss": 0.3034,
+      "step": 8430
+    },
+    {
+      "epoch": 24.002285714285716,
+      "grad_norm": 0.23769424855709076,
+      "learning_rate": 5.7530158730158735e-06,
+      "loss": 0.4167,
+      "step": 8440
+    },
+    {
+      "epoch": 24.002857142857142,
+      "grad_norm": 19.706758499145508,
+      "learning_rate": 5.746666666666667e-06,
+      "loss": 0.2434,
+      "step": 8450
+    },
+    {
+      "epoch": 24.00342857142857,
+      "grad_norm": 0.2779850363731384,
+      "learning_rate": 5.7403174603174605e-06,
+      "loss": 0.8818,
+      "step": 8460
+    },
+    {
+      "epoch": 24.004,
+      "grad_norm": 0.29331010580062866,
+      "learning_rate": 5.733968253968254e-06,
+      "loss": 0.226,
+      "step": 8470
+    },
+    {
+      "epoch": 24.004571428571428,
+      "grad_norm": 15.611747741699219,
+      "learning_rate": 5.727619047619048e-06,
+      "loss": 0.2733,
+      "step": 8480
+    },
+    {
+      "epoch": 24.005142857142857,
+      "grad_norm": 17.301841735839844,
+      "learning_rate": 5.721269841269842e-06,
+      "loss": 0.2518,
+      "step": 8490
+    },
+    {
+      "epoch": 24.005714285714287,
+      "grad_norm": 0.828689455986023,
+      "learning_rate": 5.714920634920635e-06,
+      "loss": 0.0048,
+      "step": 8500
+    },
+    {
+      "epoch": 24.006285714285713,
+      "grad_norm": 0.07569330185651779,
+      "learning_rate": 5.708571428571429e-06,
+      "loss": 0.2393,
+      "step": 8510
+    },
+    {
+      "epoch": 24.006857142857143,
+      "grad_norm": 0.027848446741700172,
+      "learning_rate": 5.702222222222222e-06,
+      "loss": 1.2843,
+      "step": 8520
+    },
+    {
+      "epoch": 24.007428571428573,
+      "grad_norm": 0.09675378352403641,
+      "learning_rate": 5.6958730158730165e-06,
+      "loss": 0.006,
+      "step": 8530
+    },
+    {
+      "epoch": 24.008,
+      "grad_norm": 0.08907628059387207,
+      "learning_rate": 5.68952380952381e-06,
+      "loss": 0.7083,
+      "step": 8540
+    },
+    {
+      "epoch": 24.00857142857143,
+      "grad_norm": 0.0546133778989315,
+      "learning_rate": 5.6831746031746035e-06,
+      "loss": 0.5869,
+      "step": 8550
+    },
+    {
+      "epoch": 24.00914285714286,
+      "grad_norm": 0.08426523208618164,
+      "learning_rate": 5.676825396825397e-06,
+      "loss": 0.2361,
+      "step": 8560
+    },
+    {
+      "epoch": 24.009714285714285,
+      "grad_norm": 0.1244712769985199,
+      "learning_rate": 5.6704761904761905e-06,
+      "loss": 0.0052,
+      "step": 8570
+    },
+    {
+      "epoch": 24.010285714285715,
+      "grad_norm": 0.15895043313503265,
+      "learning_rate": 5.664126984126984e-06,
+      "loss": 0.4814,
+      "step": 8580
+    },
+    {
+      "epoch": 24.010857142857144,
+      "grad_norm": 0.24632836878299713,
+      "learning_rate": 5.657777777777778e-06,
+      "loss": 0.0043,
+      "step": 8590
+    },
+    {
+      "epoch": 24.01142857142857,
+      "grad_norm": 245.2207794189453,
+      "learning_rate": 5.651428571428572e-06,
+      "loss": 0.7302,
+      "step": 8600
+    },
+    {
+      "epoch": 24.012,
+      "grad_norm": 0.1819450557231903,
+      "learning_rate": 5.645079365079365e-06,
+      "loss": 0.0047,
+      "step": 8610
+    },
+    {
+      "epoch": 24.01257142857143,
+      "grad_norm": 0.07300709933042526,
+      "learning_rate": 5.638730158730159e-06,
+      "loss": 0.6173,
+      "step": 8620
+    },
+    {
+      "epoch": 24.013142857142856,
+      "grad_norm": 0.014348522759974003,
+      "learning_rate": 5.632380952380952e-06,
+      "loss": 0.3765,
+      "step": 8630
+    },
+    {
+      "epoch": 24.013714285714286,
+      "grad_norm": 76.59358978271484,
+      "learning_rate": 5.626031746031746e-06,
+      "loss": 0.7209,
+      "step": 8640
+    },
+    {
+      "epoch": 24.014285714285716,
+      "grad_norm": 0.1954093724489212,
+      "learning_rate": 5.619682539682541e-06,
+      "loss": 0.2304,
+      "step": 8650
+    },
+    {
+      "epoch": 24.014857142857142,
+      "grad_norm": 13.528507232666016,
+      "learning_rate": 5.613333333333334e-06,
+      "loss": 0.6376,
+      "step": 8660
+    },
+    {
+      "epoch": 24.015428571428572,
+      "grad_norm": 12.869405746459961,
+      "learning_rate": 5.606984126984127e-06,
+      "loss": 0.5984,
+      "step": 8670
+    },
+    {
+      "epoch": 24.016,
+      "grad_norm": 0.1572108119726181,
+      "learning_rate": 5.6006349206349205e-06,
+      "loss": 0.0176,
+      "step": 8680
+    },
+    {
+      "epoch": 24.016571428571428,
+      "grad_norm": 0.20609256625175476,
+      "learning_rate": 5.594285714285714e-06,
+      "loss": 0.4308,
+      "step": 8690
+    },
+    {
+      "epoch": 24.017142857142858,
+      "grad_norm": 0.07268011569976807,
+      "learning_rate": 5.587936507936509e-06,
+      "loss": 0.4066,
+      "step": 8700
+    },
+    {
+      "epoch": 24.017714285714284,
+      "grad_norm": 13.543063163757324,
+      "learning_rate": 5.581587301587303e-06,
+      "loss": 0.7049,
+      "step": 8710
+    },
+    {
+      "epoch": 24.018285714285714,
+      "grad_norm": 19.77041244506836,
+      "learning_rate": 5.575238095238096e-06,
+      "loss": 0.4488,
+      "step": 8720
+    },
+    {
+      "epoch": 24.018857142857144,
+      "grad_norm": 0.14645476639270782,
+      "learning_rate": 5.56888888888889e-06,
+      "loss": 0.7006,
+      "step": 8730
+    },
+    {
+      "epoch": 24.01942857142857,
+      "grad_norm": 0.013115695677697659,
+      "learning_rate": 5.562539682539683e-06,
+      "loss": 0.0082,
+      "step": 8740
+    },
+    {
+      "epoch": 24.02,
+      "grad_norm": 0.41332682967185974,
+      "learning_rate": 5.556190476190476e-06,
+      "loss": 0.8514,
+      "step": 8750
+    },
+    {
+      "epoch": 24.02,
+      "eval_accuracy": 0.8417508417508418,
+      "eval_loss": 0.7481423616409302,
+      "eval_runtime": 137.4085,
+      "eval_samples_per_second": 2.161,
+      "eval_steps_per_second": 1.084,
+      "step": 8750
+    },
+    {
+      "epoch": 25.00057142857143,
+      "grad_norm": 0.371246874332428,
+      "learning_rate": 5.549841269841271e-06,
+      "loss": 0.6345,
+      "step": 8760
+    },
+    {
+      "epoch": 25.001142857142856,
+      "grad_norm": 0.7499510645866394,
+      "learning_rate": 5.543492063492064e-06,
+      "loss": 0.1388,
+      "step": 8770
+    },
+    {
+      "epoch": 25.001714285714286,
+      "grad_norm": 0.013739760965108871,
+      "learning_rate": 5.537142857142858e-06,
+      "loss": 0.5583,
+      "step": 8780
+    },
+    {
+      "epoch": 25.002285714285716,
+      "grad_norm": 190.9663543701172,
+      "learning_rate": 5.530793650793651e-06,
+      "loss": 0.7159,
+      "step": 8790
+    },
+    {
+      "epoch": 25.002857142857142,
+      "grad_norm": 256.3753356933594,
+      "learning_rate": 5.524444444444445e-06,
+      "loss": 0.0527,
+      "step": 8800
+    },
+    {
+      "epoch": 25.00342857142857,
+      "grad_norm": 7.645998001098633,
+      "learning_rate": 5.518095238095239e-06,
+      "loss": 0.5753,
+      "step": 8810
+    },
+    {
+      "epoch": 25.004,
+      "grad_norm": 0.10563361644744873,
+      "learning_rate": 5.511746031746033e-06,
+      "loss": 0.632,
+      "step": 8820
+    },
+    {
+      "epoch": 25.004571428571428,
+      "grad_norm": 27.83478546142578,
+      "learning_rate": 5.505396825396826e-06,
+      "loss": 0.3274,
+      "step": 8830
+    },
+    {
+      "epoch": 25.005142857142857,
+      "grad_norm": 0.09404099732637405,
+      "learning_rate": 5.49904761904762e-06,
+      "loss": 0.0059,
+      "step": 8840
+    },
+    {
+      "epoch": 25.005714285714287,
+      "grad_norm": 0.031015774235129356,
+      "learning_rate": 5.492698412698413e-06,
+      "loss": 0.0043,
+      "step": 8850
+    },
+    {
+      "epoch": 25.006285714285713,
+      "grad_norm": 0.20480813086032867,
+      "learning_rate": 5.4863492063492066e-06,
+      "loss": 0.5294,
+      "step": 8860
+    },
+    {
+      "epoch": 25.006857142857143,
+      "grad_norm": 125.27932739257812,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 1.0979,
+      "step": 8870
+    },
+    {
+      "epoch": 25.007428571428573,
+      "grad_norm": 0.12436144798994064,
+      "learning_rate": 5.473650793650794e-06,
+      "loss": 0.4211,
+      "step": 8880
+    },
+    {
+      "epoch": 25.008,
+      "grad_norm": 0.09650959074497223,
+      "learning_rate": 5.467301587301588e-06,
+      "loss": 0.2128,
+      "step": 8890
+    },
+    {
+      "epoch": 25.00857142857143,
+      "grad_norm": 0.09640531986951828,
+      "learning_rate": 5.460952380952381e-06,
+      "loss": 0.0061,
+      "step": 8900
+    },
+    {
+      "epoch": 25.00914285714286,
+      "grad_norm": 0.5898492932319641,
+      "learning_rate": 5.454603174603175e-06,
+      "loss": 0.3704,
+      "step": 8910
+    },
+    {
+      "epoch": 25.009714285714285,
+      "grad_norm": 0.0638417899608612,
+      "learning_rate": 5.448253968253969e-06,
+      "loss": 0.2117,
+      "step": 8920
+    },
+    {
+      "epoch": 25.010285714285715,
+      "grad_norm": 0.08020645380020142,
+      "learning_rate": 5.441904761904763e-06,
+      "loss": 0.4894,
+      "step": 8930
+    },
+    {
+      "epoch": 25.010857142857144,
+      "grad_norm": 0.15978746116161346,
+      "learning_rate": 5.435555555555556e-06,
+      "loss": 0.3888,
+      "step": 8940
+    },
+    {
+      "epoch": 25.01142857142857,
+      "grad_norm": 0.06692436337471008,
+      "learning_rate": 5.42920634920635e-06,
+      "loss": 0.2216,
+      "step": 8950
+    },
+    {
+      "epoch": 25.012,
+      "grad_norm": 0.41444236040115356,
+      "learning_rate": 5.422857142857143e-06,
+      "loss": 0.7492,
+      "step": 8960
+    },
+    {
+      "epoch": 25.01257142857143,
+      "grad_norm": 0.08462107926607132,
+      "learning_rate": 5.4165079365079366e-06,
+      "loss": 0.9321,
+      "step": 8970
+    },
+    {
+      "epoch": 25.013142857142856,
+      "grad_norm": 0.08357837796211243,
+      "learning_rate": 5.410158730158731e-06,
+      "loss": 0.0149,
+      "step": 8980
+    },
+    {
+      "epoch": 25.013714285714286,
+      "grad_norm": 20.435392379760742,
+      "learning_rate": 5.403809523809524e-06,
+      "loss": 0.6908,
+      "step": 8990
+    },
+    {
+      "epoch": 25.014285714285716,
+      "grad_norm": 0.18590255081653595,
+      "learning_rate": 5.397460317460318e-06,
+      "loss": 0.2268,
+      "step": 9000
+    },
+    {
+      "epoch": 25.014857142857142,
+      "grad_norm": 0.013719220645725727,
+      "learning_rate": 5.391111111111111e-06,
+      "loss": 0.198,
+      "step": 9010
+    },
+    {
+      "epoch": 25.015428571428572,
+      "grad_norm": 0.004588362295180559,
+      "learning_rate": 5.384761904761905e-06,
+      "loss": 0.0098,
+      "step": 9020
+    },
+    {
+      "epoch": 25.016,
+      "grad_norm": 0.07363064587116241,
+      "learning_rate": 5.378412698412698e-06,
+      "loss": 0.3217,
+      "step": 9030
+    },
+    {
+      "epoch": 25.016571428571428,
+      "grad_norm": 0.056400805711746216,
+      "learning_rate": 5.372063492063493e-06,
+      "loss": 0.2697,
+      "step": 9040
+    },
+    {
+      "epoch": 25.017142857142858,
+      "grad_norm": 84.72920989990234,
+      "learning_rate": 5.365714285714286e-06,
+      "loss": 0.0159,
+      "step": 9050
+    },
+    {
+      "epoch": 25.017714285714284,
+      "grad_norm": 17.841381072998047,
+      "learning_rate": 5.35936507936508e-06,
+      "loss": 0.3147,
+      "step": 9060
+    },
+    {
+      "epoch": 25.018285714285714,
+      "grad_norm": 17.061647415161133,
+      "learning_rate": 5.353015873015873e-06,
+      "loss": 0.5654,
+      "step": 9070
+    },
+    {
+      "epoch": 25.018857142857144,
+      "grad_norm": 0.8570396304130554,
+      "learning_rate": 5.346666666666667e-06,
+      "loss": 0.5723,
+      "step": 9080
+    },
+    {
+      "epoch": 25.01942857142857,
+      "grad_norm": 58.06008529663086,
+      "learning_rate": 5.340317460317461e-06,
+      "loss": 1.2434,
+      "step": 9090
+    },
+    {
+      "epoch": 25.02,
+      "grad_norm": 207.67483520507812,
+      "learning_rate": 5.333968253968254e-06,
+      "loss": 0.8148,
+      "step": 9100
+    },
+    {
+      "epoch": 25.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.7210359573364258,
+      "eval_runtime": 137.5062,
+      "eval_samples_per_second": 2.16,
+      "eval_steps_per_second": 1.084,
+      "step": 9100
+    },
+    {
+      "epoch": 26.00057142857143,
+      "grad_norm": 0.09444784373044968,
+      "learning_rate": 5.327619047619048e-06,
+      "loss": 0.3524,
+      "step": 9110
+    },
+    {
+      "epoch": 26.001142857142856,
+      "grad_norm": 0.26179543137550354,
+      "learning_rate": 5.321269841269841e-06,
+      "loss": 0.5438,
+      "step": 9120
+    },
+    {
+      "epoch": 26.001714285714286,
+      "grad_norm": 2.8941266536712646,
+      "learning_rate": 5.314920634920635e-06,
+      "loss": 0.4125,
+      "step": 9130
+    },
+    {
+      "epoch": 26.002285714285716,
+      "grad_norm": 0.1553201973438263,
+      "learning_rate": 5.308571428571428e-06,
+      "loss": 0.0053,
+      "step": 9140
+    },
+    {
+      "epoch": 26.002857142857142,
+      "grad_norm": 0.08845008164644241,
+      "learning_rate": 5.302222222222223e-06,
+      "loss": 0.5051,
+      "step": 9150
+    },
+    {
+      "epoch": 26.00342857142857,
+      "grad_norm": 0.11578679084777832,
+      "learning_rate": 5.295873015873016e-06,
+      "loss": 0.2748,
+      "step": 9160
+    },
+    {
+      "epoch": 26.004,
+      "grad_norm": 0.14951874315738678,
+      "learning_rate": 5.28952380952381e-06,
+      "loss": 0.1894,
+      "step": 9170
+    },
+    {
+      "epoch": 26.004571428571428,
+      "grad_norm": 0.0727708712220192,
+      "learning_rate": 5.283174603174603e-06,
+      "loss": 0.6826,
+      "step": 9180
+    },
+    {
+      "epoch": 26.005142857142857,
+      "grad_norm": 0.05796307697892189,
+      "learning_rate": 5.276825396825397e-06,
+      "loss": 0.5291,
+      "step": 9190
+    },
+    {
+      "epoch": 26.005714285714287,
+      "grad_norm": 0.10209472477436066,
+      "learning_rate": 5.270476190476192e-06,
+      "loss": 0.0044,
+      "step": 9200
+    },
+    {
+      "epoch": 26.006285714285713,
+      "grad_norm": 0.008814401924610138,
+      "learning_rate": 5.264126984126985e-06,
+      "loss": 0.5146,
+      "step": 9210
+    },
+    {
+      "epoch": 26.006857142857143,
+      "grad_norm": 0.3405003249645233,
+      "learning_rate": 5.257777777777779e-06,
+      "loss": 0.3465,
+      "step": 9220
+    },
+    {
+      "epoch": 26.007428571428573,
+      "grad_norm": 0.09109217673540115,
+      "learning_rate": 5.251428571428571e-06,
+      "loss": 0.1963,
+      "step": 9230
+    },
+    {
+      "epoch": 26.008,
+      "grad_norm": 0.05053064227104187,
+      "learning_rate": 5.245079365079365e-06,
+      "loss": 0.197,
+      "step": 9240
+    },
+    {
+      "epoch": 26.00857142857143,
+      "grad_norm": 0.01660529151558876,
+      "learning_rate": 5.238730158730158e-06,
+      "loss": 0.4736,
+      "step": 9250
+    },
+    {
+      "epoch": 26.00914285714286,
+      "grad_norm": 0.015270842239260674,
+      "learning_rate": 5.2323809523809535e-06,
+      "loss": 0.4788,
+      "step": 9260
+    },
+    {
+      "epoch": 26.009714285714285,
+      "grad_norm": 0.09976387768983841,
+      "learning_rate": 5.226031746031747e-06,
+      "loss": 0.002,
+      "step": 9270
+    },
+    {
+      "epoch": 26.010285714285715,
+      "grad_norm": 0.19991345703601837,
+      "learning_rate": 5.2196825396825405e-06,
+      "loss": 0.0036,
+      "step": 9280
+    },
+    {
+      "epoch": 26.010857142857144,
+      "grad_norm": 0.12031792104244232,
+      "learning_rate": 5.213333333333334e-06,
+      "loss": 0.6917,
+      "step": 9290
+    },
+    {
+      "epoch": 26.01142857142857,
+      "grad_norm": 0.051638491451740265,
+      "learning_rate": 5.2069841269841274e-06,
+      "loss": 0.7381,
+      "step": 9300
+    },
+    {
+      "epoch": 26.012,
+      "grad_norm": 91.03575897216797,
+      "learning_rate": 5.200634920634922e-06,
+      "loss": 0.2566,
+      "step": 9310
+    },
+    {
+      "epoch": 26.01257142857143,
+      "grad_norm": 0.1020565778017044,
+      "learning_rate": 5.194285714285715e-06,
+      "loss": 0.2807,
+      "step": 9320
+    },
+    {
+      "epoch": 26.013142857142856,
+      "grad_norm": 0.1554216891527176,
+      "learning_rate": 5.187936507936509e-06,
+      "loss": 0.2259,
+      "step": 9330
+    },
+    {
+      "epoch": 26.013714285714286,
+      "grad_norm": 0.09728459268808365,
+      "learning_rate": 5.181587301587302e-06,
+      "loss": 0.4738,
+      "step": 9340
+    },
+    {
+      "epoch": 26.014285714285716,
+      "grad_norm": 0.7819724082946777,
+      "learning_rate": 5.175238095238096e-06,
+      "loss": 0.2462,
+      "step": 9350
+    },
+    {
+      "epoch": 26.014857142857142,
+      "grad_norm": 0.19601747393608093,
+      "learning_rate": 5.168888888888889e-06,
+      "loss": 0.3972,
+      "step": 9360
+    },
+    {
+      "epoch": 26.015428571428572,
+      "grad_norm": 0.21019776165485382,
+      "learning_rate": 5.1625396825396835e-06,
+      "loss": 0.5798,
+      "step": 9370
+    },
+    {
+      "epoch": 26.016,
+      "grad_norm": 0.13152237236499786,
+      "learning_rate": 5.156190476190477e-06,
+      "loss": 0.0422,
+      "step": 9380
+    },
+    {
+      "epoch": 26.016571428571428,
+      "grad_norm": 42.05175018310547,
+      "learning_rate": 5.1498412698412705e-06,
+      "loss": 0.2365,
+      "step": 9390
+    },
+    {
+      "epoch": 26.017142857142858,
+      "grad_norm": 0.06394017487764359,
+      "learning_rate": 5.143492063492064e-06,
+      "loss": 0.2123,
+      "step": 9400
+    },
+    {
+      "epoch": 26.017714285714284,
+      "grad_norm": 0.1293639838695526,
+      "learning_rate": 5.1371428571428574e-06,
+      "loss": 0.2474,
+      "step": 9410
+    },
+    {
+      "epoch": 26.018285714285714,
+      "grad_norm": 0.46589070558547974,
+      "learning_rate": 5.130793650793651e-06,
+      "loss": 0.5852,
+      "step": 9420
+    },
+    {
+      "epoch": 26.018857142857144,
+      "grad_norm": 0.04075726494193077,
+      "learning_rate": 5.124444444444445e-06,
+      "loss": 0.4732,
+      "step": 9430
+    },
+    {
+      "epoch": 26.01942857142857,
+      "grad_norm": 0.08269976824522018,
+      "learning_rate": 5.118095238095239e-06,
+      "loss": 0.1798,
+      "step": 9440
+    },
+    {
+      "epoch": 26.02,
+      "grad_norm": 0.038522034883499146,
+      "learning_rate": 5.111746031746032e-06,
+      "loss": 0.2594,
+      "step": 9450
+    },
+    {
+      "epoch": 26.02,
+      "eval_accuracy": 0.8249158249158249,
+      "eval_loss": 0.9979982972145081,
+      "eval_runtime": 126.233,
+      "eval_samples_per_second": 2.353,
+      "eval_steps_per_second": 1.18,
+      "step": 9450
+    },
+    {
+      "epoch": 27.00057142857143,
+      "grad_norm": 0.03514016792178154,
+      "learning_rate": 5.105396825396826e-06,
+      "loss": 0.6331,
+      "step": 9460
+    },
+    {
+      "epoch": 27.001142857142856,
+      "grad_norm": 0.039175134152173996,
+      "learning_rate": 5.099047619047619e-06,
+      "loss": 0.0031,
+      "step": 9470
+    },
+    {
+      "epoch": 27.001714285714286,
+      "grad_norm": 0.017468813806772232,
+      "learning_rate": 5.0926984126984135e-06,
+      "loss": 0.0028,
+      "step": 9480
+    },
+    {
+      "epoch": 27.002285714285716,
+      "grad_norm": 3.9538137912750244,
+      "learning_rate": 5.086349206349207e-06,
+      "loss": 0.0074,
+      "step": 9490
+    },
+    {
+      "epoch": 27.002857142857142,
+      "grad_norm": 0.040926579385995865,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.7674,
+      "step": 9500
+    },
+    {
+      "epoch": 27.00342857142857,
+      "grad_norm": 0.15430638194084167,
+      "learning_rate": 5.073650793650794e-06,
+      "loss": 0.0668,
+      "step": 9510
+    },
+    {
+      "epoch": 27.004,
+      "grad_norm": 36.84708786010742,
+      "learning_rate": 5.0673015873015875e-06,
+      "loss": 0.4732,
+      "step": 9520
+    },
+    {
+      "epoch": 27.004571428571428,
+      "grad_norm": 0.04013565182685852,
+      "learning_rate": 5.060952380952381e-06,
+      "loss": 0.0228,
+      "step": 9530
+    },
+    {
+      "epoch": 27.005142857142857,
+      "grad_norm": 0.029689166694879532,
+      "learning_rate": 5.054603174603175e-06,
+      "loss": 0.2372,
+      "step": 9540
+    },
+    {
+      "epoch": 27.005714285714287,
+      "grad_norm": 0.0034933576826006174,
+      "learning_rate": 5.048253968253969e-06,
+      "loss": 0.2009,
+      "step": 9550
+    },
+    {
+      "epoch": 27.006285714285713,
+      "grad_norm": 22.289688110351562,
+      "learning_rate": 5.041904761904762e-06,
+      "loss": 0.3468,
+      "step": 9560
+    },
+    {
+      "epoch": 27.006857142857143,
+      "grad_norm": 0.05222174897789955,
+      "learning_rate": 5.035555555555556e-06,
+      "loss": 0.266,
+      "step": 9570
+    },
+    {
+      "epoch": 27.007428571428573,
+      "grad_norm": 0.08454541116952896,
+      "learning_rate": 5.029206349206349e-06,
+      "loss": 0.543,
+      "step": 9580
+    },
+    {
+      "epoch": 27.008,
+      "grad_norm": 15.091740608215332,
+      "learning_rate": 5.0228571428571435e-06,
+      "loss": 0.2849,
+      "step": 9590
+    },
+    {
+      "epoch": 27.00857142857143,
+      "grad_norm": 108.28709411621094,
+      "learning_rate": 5.016507936507937e-06,
+      "loss": 0.5846,
+      "step": 9600
+    },
+    {
+      "epoch": 27.00914285714286,
+      "grad_norm": 37.8171501159668,
+      "learning_rate": 5.0101587301587305e-06,
+      "loss": 0.3388,
+      "step": 9610
+    },
+    {
+      "epoch": 27.009714285714285,
+      "grad_norm": 0.05814701318740845,
+      "learning_rate": 5.003809523809524e-06,
+      "loss": 0.1932,
+      "step": 9620
+    },
+    {
+      "epoch": 27.010285714285715,
+      "grad_norm": 0.018001697957515717,
+      "learning_rate": 4.997460317460318e-06,
+      "loss": 0.1371,
+      "step": 9630
+    },
+    {
+      "epoch": 27.010857142857144,
+      "grad_norm": 0.20318511128425598,
+      "learning_rate": 4.991111111111112e-06,
+      "loss": 0.3923,
+      "step": 9640
+    },
+    {
+      "epoch": 27.01142857142857,
+      "grad_norm": 0.07720087468624115,
+      "learning_rate": 4.984761904761905e-06,
+      "loss": 0.0501,
+      "step": 9650
+    },
+    {
+      "epoch": 27.012,
+      "grad_norm": 19.00177764892578,
+      "learning_rate": 4.978412698412699e-06,
+      "loss": 0.542,
+      "step": 9660
+    },
+    {
+      "epoch": 27.01257142857143,
+      "grad_norm": 0.011382623575627804,
+      "learning_rate": 4.972063492063492e-06,
+      "loss": 0.2808,
+      "step": 9670
+    },
+    {
+      "epoch": 27.013142857142856,
+      "grad_norm": 0.14707712829113007,
+      "learning_rate": 4.965714285714286e-06,
+      "loss": 0.3066,
+      "step": 9680
+    },
+    {
+      "epoch": 27.013714285714286,
+      "grad_norm": 0.0880795419216156,
+      "learning_rate": 4.95936507936508e-06,
+      "loss": 0.3369,
+      "step": 9690
+    },
+    {
+      "epoch": 27.014285714285716,
+      "grad_norm": 91.44270324707031,
+      "learning_rate": 4.9530158730158735e-06,
+      "loss": 0.1431,
+      "step": 9700
+    },
+    {
+      "epoch": 27.014857142857142,
+      "grad_norm": 0.03979627415537834,
+      "learning_rate": 4.946666666666667e-06,
+      "loss": 0.2022,
+      "step": 9710
+    },
+    {
+      "epoch": 27.015428571428572,
+      "grad_norm": 0.06051446869969368,
+      "learning_rate": 4.9403174603174605e-06,
+      "loss": 1.1248,
+      "step": 9720
+    },
+    {
+      "epoch": 27.016,
+      "grad_norm": 0.011192934587597847,
+      "learning_rate": 4.933968253968254e-06,
+      "loss": 0.5604,
+      "step": 9730
+    },
+    {
+      "epoch": 27.016571428571428,
+      "grad_norm": 0.013154531829059124,
+      "learning_rate": 4.9276190476190475e-06,
+      "loss": 0.1641,
+      "step": 9740
+    },
+    {
+      "epoch": 27.017142857142858,
+      "grad_norm": 0.16332918405532837,
+      "learning_rate": 4.921269841269842e-06,
+      "loss": 0.2698,
+      "step": 9750
+    },
+    {
+      "epoch": 27.017714285714284,
+      "grad_norm": 0.12617675960063934,
+      "learning_rate": 4.914920634920635e-06,
+      "loss": 0.5508,
+      "step": 9760
+    },
+    {
+      "epoch": 27.018285714285714,
+      "grad_norm": 0.007707640528678894,
+      "learning_rate": 4.90857142857143e-06,
+      "loss": 0.5326,
+      "step": 9770
+    },
+    {
+      "epoch": 27.018857142857144,
+      "grad_norm": 0.10187384486198425,
+      "learning_rate": 4.902222222222222e-06,
+      "loss": 0.0046,
+      "step": 9780
+    },
+    {
+      "epoch": 27.01942857142857,
+      "grad_norm": 0.09700454026460648,
+      "learning_rate": 4.895873015873016e-06,
+      "loss": 0.4082,
+      "step": 9790
+    },
+    {
+      "epoch": 27.02,
+      "grad_norm": 0.1347123682498932,
+      "learning_rate": 4.88952380952381e-06,
+      "loss": 0.6742,
+      "step": 9800
+    },
+    {
+      "epoch": 27.02,
+      "eval_accuracy": 0.8585858585858586,
+      "eval_loss": 0.7987341284751892,
+      "eval_runtime": 126.3026,
+      "eval_samples_per_second": 2.351,
+      "eval_steps_per_second": 1.18,
+      "step": 9800
+    },
+    {
+      "epoch": 28.00057142857143,
+      "grad_norm": 0.007627937477082014,
+      "learning_rate": 4.8831746031746035e-06,
+      "loss": 0.8211,
+      "step": 9810
+    },
+    {
+      "epoch": 28.001142857142856,
+      "grad_norm": 0.5802029967308044,
+      "learning_rate": 4.876825396825397e-06,
+      "loss": 0.0109,
+      "step": 9820
+    },
+    {
+      "epoch": 28.001714285714286,
+      "grad_norm": 0.045871149748563766,
+      "learning_rate": 4.870476190476191e-06,
+      "loss": 0.0037,
+      "step": 9830
+    },
+    {
+      "epoch": 28.002285714285716,
+      "grad_norm": 17.630939483642578,
+      "learning_rate": 4.864126984126985e-06,
+      "loss": 1.0221,
+      "step": 9840
+    },
+    {
+      "epoch": 28.002857142857142,
+      "grad_norm": 0.20201857388019562,
+      "learning_rate": 4.857777777777778e-06,
+      "loss": 0.3867,
+      "step": 9850
+    },
+    {
+      "epoch": 28.00342857142857,
+      "grad_norm": 19.86452293395996,
+      "learning_rate": 4.851428571428572e-06,
+      "loss": 0.0709,
+      "step": 9860
+    },
+    {
+      "epoch": 28.004,
+      "grad_norm": 24.085710525512695,
+      "learning_rate": 4.845079365079365e-06,
+      "loss": 0.1767,
+      "step": 9870
+    },
+    {
+      "epoch": 28.004571428571428,
+      "grad_norm": 0.37033602595329285,
+      "learning_rate": 4.83873015873016e-06,
+      "loss": 0.3709,
+      "step": 9880
+    },
+    {
+      "epoch": 28.005142857142857,
+      "grad_norm": 0.0676516443490982,
+      "learning_rate": 4.832380952380953e-06,
+      "loss": 0.6114,
+      "step": 9890
+    },
+    {
+      "epoch": 28.005714285714287,
+      "grad_norm": 0.13778136670589447,
+      "learning_rate": 4.826031746031747e-06,
+      "loss": 0.1078,
+      "step": 9900
+    },
+    {
+      "epoch": 28.006285714285713,
+      "grad_norm": 292.9512939453125,
+      "learning_rate": 4.81968253968254e-06,
+      "loss": 0.6608,
+      "step": 9910
+    },
+    {
+      "epoch": 28.006857142857143,
+      "grad_norm": 0.08069964498281479,
+      "learning_rate": 4.8133333333333336e-06,
+      "loss": 0.6372,
+      "step": 9920
+    },
+    {
+      "epoch": 28.007428571428573,
+      "grad_norm": 16.857074737548828,
+      "learning_rate": 4.806984126984127e-06,
+      "loss": 0.3145,
+      "step": 9930
+    },
+    {
+      "epoch": 28.008,
+      "grad_norm": 0.1705380231142044,
+      "learning_rate": 4.800634920634921e-06,
+      "loss": 0.5308,
+      "step": 9940
+    },
+    {
+      "epoch": 28.00857142857143,
+      "grad_norm": 0.19721141457557678,
+      "learning_rate": 4.794285714285715e-06,
+      "loss": 0.2503,
+      "step": 9950
+    },
+    {
+      "epoch": 28.00914285714286,
+      "grad_norm": 0.018095174804329872,
+      "learning_rate": 4.787936507936508e-06,
+      "loss": 0.0216,
+      "step": 9960
+    },
+    {
+      "epoch": 28.009714285714285,
+      "grad_norm": 0.058308668434619904,
+      "learning_rate": 4.781587301587302e-06,
+      "loss": 0.4907,
+      "step": 9970
+    },
+    {
+      "epoch": 28.010285714285715,
+      "grad_norm": 0.18544158339500427,
+      "learning_rate": 4.775238095238095e-06,
+      "loss": 0.2361,
+      "step": 9980
+    },
+    {
+      "epoch": 28.010857142857144,
+      "grad_norm": 0.18984903395175934,
+      "learning_rate": 4.768888888888889e-06,
+      "loss": 0.1339,
+      "step": 9990
+    },
+    {
+      "epoch": 28.01142857142857,
+      "grad_norm": 117.79344177246094,
+      "learning_rate": 4.762539682539683e-06,
+      "loss": 0.496,
+      "step": 10000
+    },
+    {
+      "epoch": 28.012,
+      "grad_norm": 0.006381293758749962,
+      "learning_rate": 4.756190476190477e-06,
+      "loss": 0.0039,
+      "step": 10010
+    },
+    {
+      "epoch": 28.01257142857143,
+      "grad_norm": 321.21435546875,
+      "learning_rate": 4.74984126984127e-06,
+      "loss": 0.5793,
+      "step": 10020
+    },
+    {
+      "epoch": 28.013142857142856,
+      "grad_norm": 0.06357376277446747,
+      "learning_rate": 4.7434920634920636e-06,
+      "loss": 0.0057,
+      "step": 10030
+    },
+    {
+      "epoch": 28.013714285714286,
+      "grad_norm": 0.2022400200366974,
+      "learning_rate": 4.737142857142857e-06,
+      "loss": 0.3293,
+      "step": 10040
+    },
+    {
+      "epoch": 28.014285714285716,
+      "grad_norm": 0.050252072513103485,
+      "learning_rate": 4.730793650793651e-06,
+      "loss": 0.0027,
+      "step": 10050
+    },
+    {
+      "epoch": 28.014857142857142,
+      "grad_norm": 0.07211655378341675,
+      "learning_rate": 4.724444444444445e-06,
+      "loss": 0.0018,
+      "step": 10060
+    },
+    {
+      "epoch": 28.015428571428572,
+      "grad_norm": 0.012892846018075943,
+      "learning_rate": 4.718095238095238e-06,
+      "loss": 0.3652,
+      "step": 10070
+    },
+    {
+      "epoch": 28.016,
+      "grad_norm": 0.07106824219226837,
+      "learning_rate": 4.711746031746033e-06,
+      "loss": 0.6978,
+      "step": 10080
+    },
+    {
+      "epoch": 28.016571428571428,
+      "grad_norm": 131.25204467773438,
+      "learning_rate": 4.705396825396826e-06,
+      "loss": 0.0722,
+      "step": 10090
+    },
+    {
+      "epoch": 28.017142857142858,
+      "grad_norm": 0.002385256579145789,
+      "learning_rate": 4.699047619047619e-06,
+      "loss": 0.0026,
+      "step": 10100
+    },
+    {
+      "epoch": 28.017714285714284,
+      "grad_norm": 0.15431569516658783,
+      "learning_rate": 4.692698412698413e-06,
+      "loss": 0.3684,
+      "step": 10110
+    },
+    {
+      "epoch": 28.018285714285714,
+      "grad_norm": 0.08459474891424179,
+      "learning_rate": 4.686349206349207e-06,
+      "loss": 0.4475,
+      "step": 10120
+    },
+    {
+      "epoch": 28.018857142857144,
+      "grad_norm": 0.012470588088035583,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.0172,
+      "step": 10130
+    },
+    {
+      "epoch": 28.01942857142857,
+      "grad_norm": 19.614744186401367,
+      "learning_rate": 4.673650793650794e-06,
+      "loss": 0.6554,
+      "step": 10140
+    },
+    {
+      "epoch": 28.02,
+      "grad_norm": 0.002492617815732956,
+      "learning_rate": 4.667301587301588e-06,
+      "loss": 0.0063,
+      "step": 10150
+    },
+    {
+      "epoch": 28.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 0.9369211196899414,
+      "eval_runtime": 126.2783,
+      "eval_samples_per_second": 2.352,
+      "eval_steps_per_second": 1.18,
+      "step": 10150
+    },
+    {
+      "epoch": 29.00057142857143,
+      "grad_norm": 0.11610179394483566,
+      "learning_rate": 4.660952380952381e-06,
+      "loss": 0.0046,
+      "step": 10160
+    },
+    {
+      "epoch": 29.001142857142856,
+      "grad_norm": 13.605859756469727,
+      "learning_rate": 4.654603174603175e-06,
+      "loss": 0.5286,
+      "step": 10170
+    },
+    {
+      "epoch": 29.001714285714286,
+      "grad_norm": 0.03343842178583145,
+      "learning_rate": 4.648253968253968e-06,
+      "loss": 0.5034,
+      "step": 10180
+    },
+    {
+      "epoch": 29.002285714285716,
+      "grad_norm": 0.25074559450149536,
+      "learning_rate": 4.641904761904763e-06,
+      "loss": 0.1946,
+      "step": 10190
+    },
+    {
+      "epoch": 29.002857142857142,
+      "grad_norm": 0.0782301276922226,
+      "learning_rate": 4.635555555555556e-06,
+      "loss": 0.0034,
+      "step": 10200
+    },
+    {
+      "epoch": 29.00342857142857,
+      "grad_norm": 0.037485599517822266,
+      "learning_rate": 4.62920634920635e-06,
+      "loss": 0.183,
+      "step": 10210
+    },
+    {
+      "epoch": 29.004,
+      "grad_norm": 0.03207986056804657,
+      "learning_rate": 4.622857142857143e-06,
+      "loss": 0.7524,
+      "step": 10220
+    },
+    {
+      "epoch": 29.004571428571428,
+      "grad_norm": 0.07413322478532791,
+      "learning_rate": 4.616507936507937e-06,
+      "loss": 0.2849,
+      "step": 10230
+    },
+    {
+      "epoch": 29.005142857142857,
+      "grad_norm": 19.703533172607422,
+      "learning_rate": 4.61015873015873e-06,
+      "loss": 1.0924,
+      "step": 10240
+    },
+    {
+      "epoch": 29.005714285714287,
+      "grad_norm": 13.900262832641602,
+      "learning_rate": 4.603809523809524e-06,
+      "loss": 0.6464,
+      "step": 10250
+    },
+    {
+      "epoch": 29.006285714285713,
+      "grad_norm": 0.2331477850675583,
+      "learning_rate": 4.597460317460318e-06,
+      "loss": 0.0042,
+      "step": 10260
+    },
+    {
+      "epoch": 29.006857142857143,
+      "grad_norm": 0.0446133092045784,
+      "learning_rate": 4.591111111111111e-06,
+      "loss": 0.0039,
+      "step": 10270
+    },
+    {
+      "epoch": 29.007428571428573,
+      "grad_norm": 0.03788290172815323,
+      "learning_rate": 4.584761904761905e-06,
+      "loss": 0.4045,
+      "step": 10280
+    },
+    {
+      "epoch": 29.008,
+      "grad_norm": 0.11208215355873108,
+      "learning_rate": 4.578412698412698e-06,
+      "loss": 0.2883,
+      "step": 10290
+    },
+    {
+      "epoch": 29.00857142857143,
+      "grad_norm": 0.05422298610210419,
+      "learning_rate": 4.572063492063493e-06,
+      "loss": 0.0113,
+      "step": 10300
+    },
+    {
+      "epoch": 29.00914285714286,
+      "grad_norm": 0.003405461786314845,
+      "learning_rate": 4.565714285714286e-06,
+      "loss": 0.0028,
+      "step": 10310
+    },
+    {
+      "epoch": 29.009714285714285,
+      "grad_norm": 0.04415423423051834,
+      "learning_rate": 4.55936507936508e-06,
+      "loss": 0.2427,
+      "step": 10320
+    },
+    {
+      "epoch": 29.010285714285715,
+      "grad_norm": 15.168233871459961,
+      "learning_rate": 4.553015873015873e-06,
+      "loss": 0.7346,
+      "step": 10330
+    },
+    {
+      "epoch": 29.010857142857144,
+      "grad_norm": 0.021887382492423058,
+      "learning_rate": 4.546666666666667e-06,
+      "loss": 0.2257,
+      "step": 10340
+    },
+    {
+      "epoch": 29.01142857142857,
+      "grad_norm": 0.03137199580669403,
+      "learning_rate": 4.54031746031746e-06,
+      "loss": 0.249,
+      "step": 10350
+    },
+    {
+      "epoch": 29.012,
+      "grad_norm": 0.06243215128779411,
+      "learning_rate": 4.5339682539682544e-06,
+      "loss": 0.0026,
+      "step": 10360
+    },
+    {
+      "epoch": 29.01257142857143,
+      "grad_norm": 0.0038311234675347805,
+      "learning_rate": 4.527619047619048e-06,
+      "loss": 0.1538,
+      "step": 10370
+    },
+    {
+      "epoch": 29.013142857142856,
+      "grad_norm": 0.01766922138631344,
+      "learning_rate": 4.521269841269841e-06,
+      "loss": 0.273,
+      "step": 10380
+    },
+    {
+      "epoch": 29.013714285714286,
+      "grad_norm": 0.0019730119965970516,
+      "learning_rate": 4.514920634920636e-06,
+      "loss": 0.0014,
+      "step": 10390
+    },
+    {
+      "epoch": 29.014285714285716,
+      "grad_norm": 0.6263951659202576,
+      "learning_rate": 4.508571428571429e-06,
+      "loss": 0.2918,
+      "step": 10400
+    },
+    {
+      "epoch": 29.014857142857142,
+      "grad_norm": 49.63083267211914,
+      "learning_rate": 4.502222222222223e-06,
+      "loss": 0.3073,
+      "step": 10410
+    },
+    {
+      "epoch": 29.015428571428572,
+      "grad_norm": 37.1717643737793,
+      "learning_rate": 4.495873015873016e-06,
+      "loss": 0.3176,
+      "step": 10420
+    },
+    {
+      "epoch": 29.016,
+      "grad_norm": 0.03757132217288017,
+      "learning_rate": 4.48952380952381e-06,
+      "loss": 0.1175,
+      "step": 10430
+    },
+    {
+      "epoch": 29.016571428571428,
+      "grad_norm": 0.095696821808815,
+      "learning_rate": 4.483174603174604e-06,
+      "loss": 0.5835,
+      "step": 10440
+    },
+    {
+      "epoch": 29.017142857142858,
+      "grad_norm": 0.12873250246047974,
+      "learning_rate": 4.4768253968253975e-06,
+      "loss": 0.1662,
+      "step": 10450
+    },
+    {
+      "epoch": 29.017714285714284,
+      "grad_norm": 0.03826398402452469,
+      "learning_rate": 4.470476190476191e-06,
+      "loss": 0.0065,
+      "step": 10460
+    },
+    {
+      "epoch": 29.018285714285714,
+      "grad_norm": 0.027511196210980415,
+      "learning_rate": 4.4641269841269844e-06,
+      "loss": 0.669,
+      "step": 10470
+    },
+    {
+      "epoch": 29.018857142857144,
+      "grad_norm": 0.01353570818901062,
+      "learning_rate": 4.457777777777778e-06,
+      "loss": 0.0017,
+      "step": 10480
+    },
+    {
+      "epoch": 29.01942857142857,
+      "grad_norm": 0.007252034731209278,
+      "learning_rate": 4.451428571428571e-06,
+      "loss": 0.3284,
+      "step": 10490
+    },
+    {
+      "epoch": 29.02,
+      "grad_norm": 0.14840850234031677,
+      "learning_rate": 4.445079365079366e-06,
+      "loss": 0.5186,
+      "step": 10500
+    },
+    {
+      "epoch": 29.02,
+      "eval_accuracy": 0.8148148148148148,
+      "eval_loss": 1.0870999097824097,
+      "eval_runtime": 126.3664,
+      "eval_samples_per_second": 2.35,
+      "eval_steps_per_second": 1.179,
+      "step": 10500
+    },
+    {
+      "epoch": 30.00057142857143,
+      "grad_norm": 0.020524989813566208,
+      "learning_rate": 4.438730158730159e-06,
+      "loss": 0.0009,
+      "step": 10510
+    },
+    {
+      "epoch": 30.001142857142856,
+      "grad_norm": 0.27530866861343384,
+      "learning_rate": 4.432380952380953e-06,
+      "loss": 1.0032,
+      "step": 10520
+    },
+    {
+      "epoch": 30.001714285714286,
+      "grad_norm": 0.06338400393724442,
+      "learning_rate": 4.426031746031746e-06,
+      "loss": 0.0018,
+      "step": 10530
+    },
+    {
+      "epoch": 30.002285714285716,
+      "grad_norm": 0.049876321107149124,
+      "learning_rate": 4.41968253968254e-06,
+      "loss": 0.0032,
+      "step": 10540
+    },
+    {
+      "epoch": 30.002857142857142,
+      "grad_norm": 0.21788761019706726,
+      "learning_rate": 4.413333333333334e-06,
+      "loss": 0.0011,
+      "step": 10550
+    },
+    {
+      "epoch": 30.00342857142857,
+      "grad_norm": 0.006480003707110882,
+      "learning_rate": 4.4069841269841275e-06,
+      "loss": 0.4141,
+      "step": 10560
+    },
+    {
+      "epoch": 30.004,
+      "grad_norm": 0.011641621589660645,
+      "learning_rate": 4.400634920634921e-06,
+      "loss": 0.2498,
+      "step": 10570
+    },
+    {
+      "epoch": 30.004571428571428,
+      "grad_norm": 0.06504371762275696,
+      "learning_rate": 4.3942857142857144e-06,
+      "loss": 0.5437,
+      "step": 10580
+    },
+    {
+      "epoch": 30.005142857142857,
+      "grad_norm": 284.99957275390625,
+      "learning_rate": 4.387936507936508e-06,
+      "loss": 0.734,
+      "step": 10590
+    },
+    {
+      "epoch": 30.005714285714287,
+      "grad_norm": 81.92947387695312,
+      "learning_rate": 4.381587301587301e-06,
+      "loss": 0.6903,
+      "step": 10600
+    },
+    {
+      "epoch": 30.006285714285713,
+      "grad_norm": 0.0021508638747036457,
+      "learning_rate": 4.375238095238096e-06,
+      "loss": 0.0015,
+      "step": 10610
+    },
+    {
+      "epoch": 30.006857142857143,
+      "grad_norm": 0.008287636563181877,
+      "learning_rate": 4.368888888888889e-06,
+      "loss": 0.0204,
+      "step": 10620
+    },
+    {
+      "epoch": 30.007428571428573,
+      "grad_norm": 0.06526491791009903,
+      "learning_rate": 4.362539682539683e-06,
+      "loss": 0.2437,
+      "step": 10630
+    },
+    {
+      "epoch": 30.008,
+      "grad_norm": 0.09128577262163162,
+      "learning_rate": 4.356190476190477e-06,
+      "loss": 0.1999,
+      "step": 10640
+    },
+    {
+      "epoch": 30.00857142857143,
+      "grad_norm": 0.13880805671215057,
+      "learning_rate": 4.34984126984127e-06,
+      "loss": 0.2869,
+      "step": 10650
+    },
+    {
+      "epoch": 30.00914285714286,
+      "grad_norm": 61.90774154663086,
+      "learning_rate": 4.343492063492064e-06,
+      "loss": 0.2872,
+      "step": 10660
+    },
+    {
+      "epoch": 30.009714285714285,
+      "grad_norm": 0.1190374344587326,
+      "learning_rate": 4.3371428571428575e-06,
+      "loss": 1.0817,
+      "step": 10670
+    },
+    {
+      "epoch": 30.010285714285715,
+      "grad_norm": 0.0040014442056417465,
+      "learning_rate": 4.330793650793651e-06,
+      "loss": 0.8608,
+      "step": 10680
+    },
+    {
+      "epoch": 30.010857142857144,
+      "grad_norm": 0.08265765011310577,
+      "learning_rate": 4.324444444444445e-06,
+      "loss": 0.0143,
+      "step": 10690
+    },
+    {
+      "epoch": 30.01142857142857,
+      "grad_norm": 0.08359917998313904,
+      "learning_rate": 4.318095238095239e-06,
+      "loss": 0.0016,
+      "step": 10700
+    },
+    {
+      "epoch": 30.012,
+      "grad_norm": 0.03306671977043152,
+      "learning_rate": 4.311746031746032e-06,
+      "loss": 0.0031,
+      "step": 10710
+    },
+    {
+      "epoch": 30.01257142857143,
+      "grad_norm": 45.719505310058594,
+      "learning_rate": 4.305396825396826e-06,
+      "loss": 0.3334,
+      "step": 10720
+    },
+    {
+      "epoch": 30.013142857142856,
+      "grad_norm": 0.0030138203874230385,
+      "learning_rate": 4.299047619047619e-06,
+      "loss": 0.2682,
+      "step": 10730
+    },
+    {
+      "epoch": 30.013714285714286,
+      "grad_norm": 0.015341831371188164,
+      "learning_rate": 4.292698412698413e-06,
+      "loss": 0.6864,
+      "step": 10740
+    },
+    {
+      "epoch": 30.014285714285716,
+      "grad_norm": 1.9505524635314941,
+      "learning_rate": 4.286349206349207e-06,
+      "loss": 0.3386,
+      "step": 10750
+    },
+    {
+      "epoch": 30.014857142857142,
+      "grad_norm": 15.920595169067383,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.0109,
+      "step": 10760
+    },
+    {
+      "epoch": 30.015428571428572,
+      "grad_norm": 0.03716844692826271,
+      "learning_rate": 4.273650793650794e-06,
+      "loss": 0.0024,
+      "step": 10770
+    },
+    {
+      "epoch": 30.016,
+      "grad_norm": 0.012245182879269123,
+      "learning_rate": 4.2673015873015875e-06,
+      "loss": 0.0497,
+      "step": 10780
+    },
+    {
+      "epoch": 30.016571428571428,
+      "grad_norm": 0.020435450598597527,
+      "learning_rate": 4.260952380952381e-06,
+      "loss": 0.3038,
+      "step": 10790
+    },
+    {
+      "epoch": 30.017142857142858,
+      "grad_norm": 0.7739905714988708,
+      "learning_rate": 4.254603174603175e-06,
+      "loss": 0.0028,
+      "step": 10800
+    },
+    {
+      "epoch": 30.017714285714284,
+      "grad_norm": 116.42400360107422,
+      "learning_rate": 4.248253968253969e-06,
+      "loss": 0.5479,
+      "step": 10810
+    },
+    {
+      "epoch": 30.018285714285714,
+      "grad_norm": 0.060728929936885834,
+      "learning_rate": 4.241904761904762e-06,
+      "loss": 0.1333,
+      "step": 10820
+    },
+    {
+      "epoch": 30.018857142857144,
+      "grad_norm": 2.810554265975952,
+      "learning_rate": 4.235555555555556e-06,
+      "loss": 1.2669,
+      "step": 10830
+    },
+    {
+      "epoch": 30.01942857142857,
+      "grad_norm": 0.007885237224400043,
+      "learning_rate": 4.229206349206349e-06,
+      "loss": 0.8325,
+      "step": 10840
+    },
+    {
+      "epoch": 30.02,
+      "grad_norm": 0.06585021317005157,
+      "learning_rate": 4.222857142857143e-06,
+      "loss": 0.3076,
+      "step": 10850
+    },
+    {
+      "epoch": 30.02,
+      "eval_accuracy": 0.835016835016835,
+      "eval_loss": 0.8930524587631226,
+      "eval_runtime": 126.4231,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 1.179,
+      "step": 10850
+    },
+    {
+      "epoch": 31.00057142857143,
+      "grad_norm": 0.008948814123868942,
+      "learning_rate": 4.216507936507937e-06,
+      "loss": 0.263,
+      "step": 10860
+    },
+    {
+      "epoch": 31.001142857142856,
+      "grad_norm": 0.22963719069957733,
+      "learning_rate": 4.2101587301587305e-06,
+      "loss": 0.0015,
+      "step": 10870
+    },
+    {
+      "epoch": 31.001714285714286,
+      "grad_norm": 0.2196054607629776,
+      "learning_rate": 4.203809523809524e-06,
+      "loss": 0.0021,
+      "step": 10880
+    },
+    {
+      "epoch": 31.002285714285716,
+      "grad_norm": 0.03480219841003418,
+      "learning_rate": 4.1974603174603175e-06,
+      "loss": 0.6281,
+      "step": 10890
+    },
+    {
+      "epoch": 31.002857142857142,
+      "grad_norm": 0.9133780002593994,
+      "learning_rate": 4.191111111111111e-06,
+      "loss": 0.0018,
+      "step": 10900
+    },
+    {
+      "epoch": 31.00342857142857,
+      "grad_norm": 0.1275293380022049,
+      "learning_rate": 4.184761904761905e-06,
+      "loss": 0.4055,
+      "step": 10910
+    },
+    {
+      "epoch": 31.004,
+      "grad_norm": 0.3001004755496979,
+      "learning_rate": 4.178412698412699e-06,
+      "loss": 0.2961,
+      "step": 10920
+    },
+    {
+      "epoch": 31.004571428571428,
+      "grad_norm": 0.08802656084299088,
+      "learning_rate": 4.172063492063492e-06,
+      "loss": 0.537,
+      "step": 10930
+    },
+    {
+      "epoch": 31.005142857142857,
+      "grad_norm": 159.4746551513672,
+      "learning_rate": 4.165714285714287e-06,
+      "loss": 0.3808,
+      "step": 10940
+    },
+    {
+      "epoch": 31.005714285714287,
+      "grad_norm": 0.11664550751447678,
+      "learning_rate": 4.15936507936508e-06,
+      "loss": 0.7977,
+      "step": 10950
+    },
+    {
+      "epoch": 31.006285714285713,
+      "grad_norm": 0.20757505297660828,
+      "learning_rate": 4.1530158730158736e-06,
+      "loss": 0.7499,
+      "step": 10960
+    },
+    {
+      "epoch": 31.006857142857143,
+      "grad_norm": 0.0029254963155835867,
+      "learning_rate": 4.146666666666667e-06,
+      "loss": 0.0042,
+      "step": 10970
+    },
+    {
+      "epoch": 31.007428571428573,
+      "grad_norm": 0.3312826454639435,
+      "learning_rate": 4.1403174603174605e-06,
+      "loss": 0.1451,
+      "step": 10980
+    },
+    {
+      "epoch": 31.008,
+      "grad_norm": 235.10496520996094,
+      "learning_rate": 4.133968253968254e-06,
+      "loss": 0.2038,
+      "step": 10990
+    },
+    {
+      "epoch": 31.00857142857143,
+      "grad_norm": 0.029614897444844246,
+      "learning_rate": 4.127619047619048e-06,
+      "loss": 0.3989,
+      "step": 11000
+    },
+    {
+      "epoch": 31.00914285714286,
+      "grad_norm": 0.00274507119320333,
+      "learning_rate": 4.121269841269842e-06,
+      "loss": 0.0016,
+      "step": 11010
+    },
+    {
+      "epoch": 31.009714285714285,
+      "grad_norm": 0.0025850103702396154,
+      "learning_rate": 4.114920634920635e-06,
+      "loss": 0.2211,
+      "step": 11020
+    },
+    {
+      "epoch": 31.010285714285715,
+      "grad_norm": 0.003770750481635332,
+      "learning_rate": 4.108571428571429e-06,
+      "loss": 0.3602,
+      "step": 11030
+    },
+    {
+      "epoch": 31.010857142857144,
+      "grad_norm": 0.08702629059553146,
+      "learning_rate": 4.102222222222222e-06,
+      "loss": 0.2275,
+      "step": 11040
+    },
+    {
+      "epoch": 31.01142857142857,
+      "grad_norm": 0.03445100784301758,
+      "learning_rate": 4.095873015873017e-06,
+      "loss": 0.004,
+      "step": 11050
+    },
+    {
+      "epoch": 31.012,
+      "grad_norm": 0.07224719971418381,
+      "learning_rate": 4.08952380952381e-06,
+      "loss": 0.1784,
+      "step": 11060
+    },
+    {
+      "epoch": 31.01257142857143,
+      "grad_norm": 0.08405738323926926,
+      "learning_rate": 4.083174603174604e-06,
+      "loss": 0.275,
+      "step": 11070
+    },
+    {
+      "epoch": 31.013142857142856,
+      "grad_norm": 0.12125347554683685,
+      "learning_rate": 4.076825396825397e-06,
+      "loss": 0.0013,
+      "step": 11080
+    },
+    {
+      "epoch": 31.013714285714286,
+      "grad_norm": 1.5644397735595703,
+      "learning_rate": 4.0704761904761905e-06,
+      "loss": 0.2768,
+      "step": 11090
+    },
+    {
+      "epoch": 31.014285714285716,
+      "grad_norm": 0.030137941241264343,
+      "learning_rate": 4.064126984126984e-06,
+      "loss": 0.1936,
+      "step": 11100
+    },
+    {
+      "epoch": 31.014857142857142,
+      "grad_norm": 0.005889153108000755,
+      "learning_rate": 4.057777777777778e-06,
+      "loss": 0.3787,
+      "step": 11110
+    },
+    {
+      "epoch": 31.015428571428572,
+      "grad_norm": 840.96142578125,
+      "learning_rate": 4.051428571428572e-06,
+      "loss": 0.1409,
+      "step": 11120
+    },
+    {
+      "epoch": 31.016,
+      "grad_norm": 0.3630973994731903,
+      "learning_rate": 4.045079365079365e-06,
+      "loss": 0.5644,
+      "step": 11130
+    },
+    {
+      "epoch": 31.016571428571428,
+      "grad_norm": 0.1463097482919693,
+      "learning_rate": 4.038730158730159e-06,
+      "loss": 0.4875,
+      "step": 11140
+    },
+    {
+      "epoch": 31.017142857142858,
+      "grad_norm": 0.19403041899204254,
+      "learning_rate": 4.032380952380952e-06,
+      "loss": 0.0018,
+      "step": 11150
+    },
+    {
+      "epoch": 31.017714285714284,
+      "grad_norm": 0.002485697390511632,
+      "learning_rate": 4.026031746031747e-06,
+      "loss": 0.7305,
+      "step": 11160
+    },
+    {
+      "epoch": 31.018285714285714,
+      "grad_norm": 0.22555981576442719,
+      "learning_rate": 4.01968253968254e-06,
+      "loss": 0.0121,
+      "step": 11170
+    },
+    {
+      "epoch": 31.018857142857144,
+      "grad_norm": 0.02071559987962246,
+      "learning_rate": 4.013333333333334e-06,
+      "loss": 0.5759,
+      "step": 11180
+    },
+    {
+      "epoch": 31.01942857142857,
+      "grad_norm": 0.10109854489564896,
+      "learning_rate": 4.006984126984128e-06,
+      "loss": 0.3115,
+      "step": 11190
+    },
+    {
+      "epoch": 31.02,
+      "grad_norm": 0.07400314509868622,
+      "learning_rate": 4.000634920634921e-06,
+      "loss": 0.1113,
+      "step": 11200
+    },
+    {
+      "epoch": 31.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 1.0014111995697021,
+      "eval_runtime": 126.4347,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 1.178,
+      "step": 11200
+    },
+    {
+      "epoch": 32.000571428571426,
+      "grad_norm": 0.018319344148039818,
+      "learning_rate": 3.994285714285714e-06,
+      "loss": 1.0033,
+      "step": 11210
+    },
+    {
+      "epoch": 32.00114285714286,
+      "grad_norm": 2.0360021591186523,
+      "learning_rate": 3.987936507936508e-06,
+      "loss": 0.1463,
+      "step": 11220
+    },
+    {
+      "epoch": 32.001714285714286,
+      "grad_norm": 11.246064186096191,
+      "learning_rate": 3.981587301587302e-06,
+      "loss": 0.2805,
+      "step": 11230
+    },
+    {
+      "epoch": 32.00228571428571,
+      "grad_norm": 0.21093471348285675,
+      "learning_rate": 3.975238095238095e-06,
+      "loss": 0.0017,
+      "step": 11240
+    },
+    {
+      "epoch": 32.002857142857145,
+      "grad_norm": 26.97494125366211,
+      "learning_rate": 3.96888888888889e-06,
+      "loss": 0.0044,
+      "step": 11250
+    },
+    {
+      "epoch": 32.00342857142857,
+      "grad_norm": 0.3112584352493286,
+      "learning_rate": 3.962539682539683e-06,
+      "loss": 0.2287,
+      "step": 11260
+    },
+    {
+      "epoch": 32.004,
+      "grad_norm": 9.63143539428711,
+      "learning_rate": 3.956190476190477e-06,
+      "loss": 0.9842,
+      "step": 11270
+    },
+    {
+      "epoch": 32.00457142857143,
+      "grad_norm": 0.25314244627952576,
+      "learning_rate": 3.94984126984127e-06,
+      "loss": 0.2338,
+      "step": 11280
+    },
+    {
+      "epoch": 32.00514285714286,
+      "grad_norm": 0.11641905456781387,
+      "learning_rate": 3.943492063492064e-06,
+      "loss": 0.0021,
+      "step": 11290
+    },
+    {
+      "epoch": 32.005714285714284,
+      "grad_norm": 83.00080871582031,
+      "learning_rate": 3.937142857142858e-06,
+      "loss": 1.038,
+      "step": 11300
+    },
+    {
+      "epoch": 32.00628571428572,
+      "grad_norm": 0.11469433456659317,
+      "learning_rate": 3.930793650793651e-06,
+      "loss": 0.0026,
+      "step": 11310
+    },
+    {
+      "epoch": 32.00685714285714,
+      "grad_norm": 0.014426827430725098,
+      "learning_rate": 3.924444444444445e-06,
+      "loss": 0.0023,
+      "step": 11320
+    },
+    {
+      "epoch": 32.00742857142857,
+      "grad_norm": 0.0037542914506047964,
+      "learning_rate": 3.918095238095238e-06,
+      "loss": 0.4315,
+      "step": 11330
+    },
+    {
+      "epoch": 32.008,
+      "grad_norm": 0.004907695110887289,
+      "learning_rate": 3.911746031746032e-06,
+      "loss": 0.2727,
+      "step": 11340
+    },
+    {
+      "epoch": 32.00857142857143,
+      "grad_norm": 0.13894321024417877,
+      "learning_rate": 3.905396825396825e-06,
+      "loss": 0.1903,
+      "step": 11350
+    },
+    {
+      "epoch": 32.009142857142855,
+      "grad_norm": 0.03604894503951073,
+      "learning_rate": 3.89904761904762e-06,
+      "loss": 1.1061,
+      "step": 11360
+    },
+    {
+      "epoch": 32.00971428571429,
+      "grad_norm": 0.0288707222789526,
+      "learning_rate": 3.892698412698413e-06,
+      "loss": 0.0018,
+      "step": 11370
+    },
+    {
+      "epoch": 32.010285714285715,
+      "grad_norm": 41.77061080932617,
+      "learning_rate": 3.886349206349207e-06,
+      "loss": 0.2094,
+      "step": 11380
+    },
+    {
+      "epoch": 32.01085714285714,
+      "grad_norm": 0.09654809534549713,
+      "learning_rate": 3.88e-06,
+      "loss": 0.0016,
+      "step": 11390
+    },
+    {
+      "epoch": 32.011428571428574,
+      "grad_norm": 0.002318542217835784,
+      "learning_rate": 3.873650793650794e-06,
+      "loss": 0.2902,
+      "step": 11400
+    },
+    {
+      "epoch": 32.012,
+      "grad_norm": 0.1287265121936798,
+      "learning_rate": 3.867301587301588e-06,
+      "loss": 0.0023,
+      "step": 11410
+    },
+    {
+      "epoch": 32.01257142857143,
+      "grad_norm": 0.025955529883503914,
+      "learning_rate": 3.860952380952381e-06,
+      "loss": 0.29,
+      "step": 11420
+    },
+    {
+      "epoch": 32.01314285714286,
+      "grad_norm": 0.022403893992304802,
+      "learning_rate": 3.854603174603175e-06,
+      "loss": 0.2088,
+      "step": 11430
+    },
+    {
+      "epoch": 32.013714285714286,
+      "grad_norm": 14.325145721435547,
+      "learning_rate": 3.848253968253968e-06,
+      "loss": 0.2963,
+      "step": 11440
+    },
+    {
+      "epoch": 32.01428571428571,
+      "grad_norm": 0.15538744628429413,
+      "learning_rate": 3.841904761904762e-06,
+      "loss": 0.4815,
+      "step": 11450
+    },
+    {
+      "epoch": 32.014857142857146,
+      "grad_norm": 0.07028108835220337,
+      "learning_rate": 3.835555555555555e-06,
+      "loss": 0.358,
+      "step": 11460
+    },
+    {
+      "epoch": 32.01542857142857,
+      "grad_norm": 26.09341812133789,
+      "learning_rate": 3.82920634920635e-06,
+      "loss": 0.2815,
+      "step": 11470
+    },
+    {
+      "epoch": 32.016,
+      "grad_norm": 0.17105203866958618,
+      "learning_rate": 3.822857142857143e-06,
+      "loss": 0.5399,
+      "step": 11480
+    },
+    {
+      "epoch": 32.01657142857143,
+      "grad_norm": 124.22401428222656,
+      "learning_rate": 3.816507936507937e-06,
+      "loss": 0.3341,
+      "step": 11490
+    },
+    {
+      "epoch": 32.01714285714286,
+      "grad_norm": 0.05575098842382431,
+      "learning_rate": 3.8101587301587306e-06,
+      "loss": 0.0021,
+      "step": 11500
+    },
+    {
+      "epoch": 32.017714285714284,
+      "grad_norm": 0.025988677516579628,
+      "learning_rate": 3.803809523809524e-06,
+      "loss": 0.2548,
+      "step": 11510
+    },
+    {
+      "epoch": 32.01828571428572,
+      "grad_norm": 31.389314651489258,
+      "learning_rate": 3.7974603174603175e-06,
+      "loss": 0.3914,
+      "step": 11520
+    },
+    {
+      "epoch": 32.018857142857144,
+      "grad_norm": 0.030341658741235733,
+      "learning_rate": 3.7911111111111114e-06,
+      "loss": 0.0424,
+      "step": 11530
+    },
+    {
+      "epoch": 32.01942857142857,
+      "grad_norm": 0.18274207413196564,
+      "learning_rate": 3.784761904761905e-06,
+      "loss": 0.4848,
+      "step": 11540
+    },
+    {
+      "epoch": 32.02,
+      "grad_norm": 0.0034605495166033506,
+      "learning_rate": 3.778412698412699e-06,
+      "loss": 0.2201,
+      "step": 11550
+    },
+    {
+      "epoch": 32.02,
+      "eval_accuracy": 0.8484848484848485,
+      "eval_loss": 0.8628104329109192,
+      "eval_runtime": 126.7935,
+      "eval_samples_per_second": 2.342,
+      "eval_steps_per_second": 1.175,
+      "step": 11550
+    },
+    {
+      "epoch": 33.000571428571426,
+      "grad_norm": 0.009115273132920265,
+      "learning_rate": 3.7720634920634923e-06,
+      "loss": 0.0015,
+      "step": 11560
+    },
+    {
+      "epoch": 33.00114285714286,
+      "grad_norm": 0.029986055567860603,
+      "learning_rate": 3.7657142857142858e-06,
+      "loss": 0.1856,
+      "step": 11570
+    },
+    {
+      "epoch": 33.001714285714286,
+      "grad_norm": 0.04707956314086914,
+      "learning_rate": 3.7593650793650797e-06,
+      "loss": 0.3591,
+      "step": 11580
+    },
+    {
+      "epoch": 33.00228571428571,
+      "grad_norm": 4.939572334289551,
+      "learning_rate": 3.753015873015873e-06,
+      "loss": 0.0061,
+      "step": 11590
+    },
+    {
+      "epoch": 33.002857142857145,
+      "grad_norm": 0.019965268671512604,
+      "learning_rate": 3.7466666666666667e-06,
+      "loss": 0.002,
+      "step": 11600
+    },
+    {
+      "epoch": 33.00342857142857,
+      "grad_norm": 0.4821774959564209,
+      "learning_rate": 3.740317460317461e-06,
+      "loss": 0.6994,
+      "step": 11610
+    },
+    {
+      "epoch": 33.004,
+      "grad_norm": 359.18255615234375,
+      "learning_rate": 3.733968253968254e-06,
+      "loss": 0.2732,
+      "step": 11620
+    },
+    {
+      "epoch": 33.00457142857143,
+      "grad_norm": 0.0022106466349214315,
+      "learning_rate": 3.7276190476190475e-06,
+      "loss": 0.0015,
+      "step": 11630
+    },
+    {
+      "epoch": 33.00514285714286,
+      "grad_norm": 0.4608334004878998,
+      "learning_rate": 3.721269841269842e-06,
+      "loss": 0.0016,
+      "step": 11640
+    },
+    {
+      "epoch": 33.005714285714284,
+      "grad_norm": 0.10803355276584625,
+      "learning_rate": 3.7149206349206353e-06,
+      "loss": 0.0874,
+      "step": 11650
+    },
+    {
+      "epoch": 33.00628571428572,
+      "grad_norm": 0.07706461101770401,
+      "learning_rate": 3.7085714285714284e-06,
+      "loss": 0.0007,
+      "step": 11660
+    },
+    {
+      "epoch": 33.00685714285714,
+      "grad_norm": 0.006313271354883909,
+      "learning_rate": 3.7022222222222227e-06,
+      "loss": 0.0019,
+      "step": 11670
+    },
+    {
+      "epoch": 33.00742857142857,
+      "grad_norm": 0.037479300051927567,
+      "learning_rate": 3.695873015873016e-06,
+      "loss": 0.8099,
+      "step": 11680
+    },
+    {
+      "epoch": 33.008,
+      "grad_norm": 0.08825170248746872,
+      "learning_rate": 3.68952380952381e-06,
+      "loss": 0.0036,
+      "step": 11690
+    },
+    {
+      "epoch": 33.00857142857143,
+      "grad_norm": 196.14361572265625,
+      "learning_rate": 3.6831746031746036e-06,
+      "loss": 0.75,
+      "step": 11700
+    },
+    {
+      "epoch": 33.009142857142855,
+      "grad_norm": 0.5264555215835571,
+      "learning_rate": 3.676825396825397e-06,
+      "loss": 0.4639,
+      "step": 11710
+    },
+    {
+      "epoch": 33.00971428571429,
+      "grad_norm": 10.480401039123535,
+      "learning_rate": 3.670476190476191e-06,
+      "loss": 0.947,
+      "step": 11720
+    },
+    {
+      "epoch": 33.010285714285715,
+      "grad_norm": 0.4620245695114136,
+      "learning_rate": 3.6641269841269845e-06,
+      "loss": 0.0035,
+      "step": 11730
+    },
+    {
+      "epoch": 33.01085714285714,
+      "grad_norm": 0.13946720957756042,
+      "learning_rate": 3.657777777777778e-06,
+      "loss": 0.2192,
+      "step": 11740
+    },
+    {
+      "epoch": 33.011428571428574,
+      "grad_norm": 0.10003682225942612,
+      "learning_rate": 3.651428571428572e-06,
+      "loss": 0.236,
+      "step": 11750
+    },
+    {
+      "epoch": 33.012,
+      "grad_norm": 0.2135056108236313,
+      "learning_rate": 3.6450793650793653e-06,
+      "loss": 0.0021,
+      "step": 11760
+    },
+    {
+      "epoch": 33.01257142857143,
+      "grad_norm": 0.05885402113199234,
+      "learning_rate": 3.638730158730159e-06,
+      "loss": 0.2531,
+      "step": 11770
+    },
+    {
+      "epoch": 33.01314285714286,
+      "grad_norm": 0.011742881499230862,
+      "learning_rate": 3.6323809523809527e-06,
+      "loss": 0.4037,
+      "step": 11780
+    },
+    {
+      "epoch": 33.013714285714286,
+      "grad_norm": 0.05668189749121666,
+      "learning_rate": 3.6260317460317462e-06,
+      "loss": 0.0036,
+      "step": 11790
+    },
+    {
+      "epoch": 33.01428571428571,
+      "grad_norm": 104.82915496826172,
+      "learning_rate": 3.61968253968254e-06,
+      "loss": 0.1297,
+      "step": 11800
+    },
+    {
+      "epoch": 33.014857142857146,
+      "grad_norm": 0.22662226855754852,
+      "learning_rate": 3.6133333333333336e-06,
+      "loss": 0.0031,
+      "step": 11810
+    },
+    {
+      "epoch": 33.01542857142857,
+      "grad_norm": 0.0017571650678291917,
+      "learning_rate": 3.606984126984127e-06,
+      "loss": 0.3367,
+      "step": 11820
+    },
+    {
+      "epoch": 33.016,
+      "grad_norm": 0.3255730867385864,
+      "learning_rate": 3.600634920634921e-06,
+      "loss": 0.0302,
+      "step": 11830
+    },
+    {
+      "epoch": 33.01657142857143,
+      "grad_norm": 0.04617001861333847,
+      "learning_rate": 3.5942857142857145e-06,
+      "loss": 0.6351,
+      "step": 11840
+    },
+    {
+      "epoch": 33.01714285714286,
+      "grad_norm": 0.001643249997869134,
+      "learning_rate": 3.587936507936508e-06,
+      "loss": 0.4685,
+      "step": 11850
+    },
+    {
+      "epoch": 33.017714285714284,
+      "grad_norm": 66.71566772460938,
+      "learning_rate": 3.581587301587302e-06,
+      "loss": 0.4042,
+      "step": 11860
+    },
+    {
+      "epoch": 33.01828571428572,
+      "grad_norm": 0.02842831425368786,
+      "learning_rate": 3.5752380952380954e-06,
+      "loss": 0.3481,
+      "step": 11870
+    },
+    {
+      "epoch": 33.018857142857144,
+      "grad_norm": 0.8168884515762329,
+      "learning_rate": 3.568888888888889e-06,
+      "loss": 0.4965,
+      "step": 11880
+    },
+    {
+      "epoch": 33.01942857142857,
+      "grad_norm": 109.5650863647461,
+      "learning_rate": 3.562539682539683e-06,
+      "loss": 0.5878,
+      "step": 11890
+    },
+    {
+      "epoch": 33.02,
+      "grad_norm": 0.12658925354480743,
+      "learning_rate": 3.5561904761904762e-06,
+      "loss": 0.0324,
+      "step": 11900
+    },
+    {
+      "epoch": 33.02,
+      "eval_accuracy": 0.835016835016835,
+      "eval_loss": 0.9971614480018616,
+      "eval_runtime": 126.5059,
+      "eval_samples_per_second": 2.348,
+      "eval_steps_per_second": 1.178,
+      "step": 11900
+    },
+    {
+      "epoch": 34.000571428571426,
+      "grad_norm": 0.027432570233941078,
+      "learning_rate": 3.5498412698412697e-06,
+      "loss": 0.3804,
+      "step": 11910
+    },
+    {
+      "epoch": 34.00114285714286,
+      "grad_norm": 0.004706758074462414,
+      "learning_rate": 3.543492063492064e-06,
+      "loss": 0.2754,
+      "step": 11920
+    },
+    {
+      "epoch": 34.001714285714286,
+      "grad_norm": 0.03468727692961693,
+      "learning_rate": 3.5371428571428575e-06,
+      "loss": 0.0022,
+      "step": 11930
+    },
+    {
+      "epoch": 34.00228571428571,
+      "grad_norm": 0.16161181032657623,
+      "learning_rate": 3.5307936507936514e-06,
+      "loss": 0.5878,
+      "step": 11940
+    },
+    {
+      "epoch": 34.002857142857145,
+      "grad_norm": 0.0015723078977316618,
+      "learning_rate": 3.524444444444445e-06,
+      "loss": 0.0007,
+      "step": 11950
+    },
+    {
+      "epoch": 34.00342857142857,
+      "grad_norm": 0.051679786294698715,
+      "learning_rate": 3.5180952380952384e-06,
+      "loss": 0.0014,
+      "step": 11960
+    },
+    {
+      "epoch": 34.004,
+      "grad_norm": 0.07004716992378235,
+      "learning_rate": 3.5117460317460323e-06,
+      "loss": 0.4407,
+      "step": 11970
+    },
+    {
+      "epoch": 34.00457142857143,
+      "grad_norm": 0.04986123740673065,
+      "learning_rate": 3.5053968253968258e-06,
+      "loss": 0.2698,
+      "step": 11980
+    },
+    {
+      "epoch": 34.00514285714286,
+      "grad_norm": 0.007722809910774231,
+      "learning_rate": 3.4990476190476193e-06,
+      "loss": 0.0011,
+      "step": 11990
+    },
+    {
+      "epoch": 34.005714285714284,
+      "grad_norm": 0.0745702013373375,
+      "learning_rate": 3.492698412698413e-06,
+      "loss": 0.2513,
+      "step": 12000
+    },
+    {
+      "epoch": 34.00628571428572,
+      "grad_norm": 0.043245185166597366,
+      "learning_rate": 3.4863492063492067e-06,
+      "loss": 0.2987,
+      "step": 12010
+    },
+    {
+      "epoch": 34.00685714285714,
+      "grad_norm": 0.09795431792736053,
+      "learning_rate": 3.48e-06,
+      "loss": 0.3349,
+      "step": 12020
+    },
+    {
+      "epoch": 34.00742857142857,
+      "grad_norm": 0.09430580586194992,
+      "learning_rate": 3.473650793650794e-06,
+      "loss": 0.0011,
+      "step": 12030
+    },
+    {
+      "epoch": 34.008,
+      "grad_norm": 0.022420106455683708,
+      "learning_rate": 3.4673015873015875e-06,
+      "loss": 0.267,
+      "step": 12040
+    },
+    {
+      "epoch": 34.00857142857143,
+      "grad_norm": 0.045474231243133545,
+      "learning_rate": 3.4609523809523814e-06,
+      "loss": 0.2568,
+      "step": 12050
+    },
+    {
+      "epoch": 34.009142857142855,
+      "grad_norm": 80.42121887207031,
+      "learning_rate": 3.454603174603175e-06,
+      "loss": 0.3101,
+      "step": 12060
+    },
+    {
+      "epoch": 34.00971428571429,
+      "grad_norm": 20.72455406188965,
+      "learning_rate": 3.4482539682539684e-06,
+      "loss": 0.6013,
+      "step": 12070
+    },
+    {
+      "epoch": 34.010285714285715,
+      "grad_norm": 33.085418701171875,
+      "learning_rate": 3.4419047619047623e-06,
+      "loss": 0.5322,
+      "step": 12080
+    },
+    {
+      "epoch": 34.01085714285714,
+      "grad_norm": 209.3749542236328,
+      "learning_rate": 3.435555555555556e-06,
+      "loss": 0.2576,
+      "step": 12090
+    },
+    {
+      "epoch": 34.011428571428574,
+      "grad_norm": 0.305754691362381,
+      "learning_rate": 3.4292063492063493e-06,
+      "loss": 0.1585,
+      "step": 12100
+    },
+    {
+      "epoch": 34.012,
+      "grad_norm": 19.921072006225586,
+      "learning_rate": 3.422857142857143e-06,
+      "loss": 0.2088,
+      "step": 12110
+    },
+    {
+      "epoch": 34.01257142857143,
+      "grad_norm": 0.02304167114198208,
+      "learning_rate": 3.4165079365079367e-06,
+      "loss": 0.167,
+      "step": 12120
+    },
+    {
+      "epoch": 34.01314285714286,
+      "grad_norm": 0.08073779195547104,
+      "learning_rate": 3.41015873015873e-06,
+      "loss": 0.7815,
+      "step": 12130
+    },
+    {
+      "epoch": 34.013714285714286,
+      "grad_norm": 94.09422302246094,
+      "learning_rate": 3.403809523809524e-06,
+      "loss": 0.0197,
+      "step": 12140
+    },
+    {
+      "epoch": 34.01428571428571,
+      "grad_norm": 0.05119791999459267,
+      "learning_rate": 3.3974603174603175e-06,
+      "loss": 0.0028,
+      "step": 12150
+    },
+    {
+      "epoch": 34.014857142857146,
+      "grad_norm": 15.968472480773926,
+      "learning_rate": 3.391111111111111e-06,
+      "loss": 0.3075,
+      "step": 12160
+    },
+    {
+      "epoch": 34.01542857142857,
+      "grad_norm": 95.5150375366211,
+      "learning_rate": 3.3847619047619053e-06,
+      "loss": 0.5475,
+      "step": 12170
+    },
+    {
+      "epoch": 34.016,
+      "grad_norm": 69.68689727783203,
+      "learning_rate": 3.3784126984126984e-06,
+      "loss": 0.293,
+      "step": 12180
+    },
+    {
+      "epoch": 34.01657142857143,
+      "grad_norm": 0.052636485546827316,
+      "learning_rate": 3.3720634920634927e-06,
+      "loss": 0.2891,
+      "step": 12190
+    },
+    {
+      "epoch": 34.01714285714286,
+      "grad_norm": 22.63859748840332,
+      "learning_rate": 3.3657142857142862e-06,
+      "loss": 0.2413,
+      "step": 12200
+    },
+    {
+      "epoch": 34.017714285714284,
+      "grad_norm": 0.07254651933908463,
+      "learning_rate": 3.3593650793650797e-06,
+      "loss": 0.2965,
+      "step": 12210
+    },
+    {
+      "epoch": 34.01828571428572,
+      "grad_norm": 0.016718924045562744,
+      "learning_rate": 3.3530158730158736e-06,
+      "loss": 0.0566,
+      "step": 12220
+    },
+    {
+      "epoch": 34.018857142857144,
+      "grad_norm": 0.0034701621625572443,
+      "learning_rate": 3.346666666666667e-06,
+      "loss": 0.1736,
+      "step": 12230
+    },
+    {
+      "epoch": 34.01942857142857,
+      "grad_norm": 0.4016928970813751,
+      "learning_rate": 3.3403174603174606e-06,
+      "loss": 0.3631,
+      "step": 12240
+    },
+    {
+      "epoch": 34.02,
+      "grad_norm": 173.18507385253906,
+      "learning_rate": 3.3339682539682545e-06,
+      "loss": 0.4411,
+      "step": 12250
+    },
+    {
+      "epoch": 34.02,
+      "eval_accuracy": 0.835016835016835,
+      "eval_loss": 1.059208869934082,
+      "eval_runtime": 126.2592,
+      "eval_samples_per_second": 2.352,
+      "eval_steps_per_second": 1.18,
+      "step": 12250
+    },
+    {
+      "epoch": 35.000571428571426,
+      "grad_norm": 0.02301918901503086,
+      "learning_rate": 3.327619047619048e-06,
+      "loss": 0.264,
+      "step": 12260
+    },
+    {
+      "epoch": 35.00114285714286,
+      "grad_norm": 0.14874428510665894,
+      "learning_rate": 3.3212698412698414e-06,
+      "loss": 0.0022,
+      "step": 12270
+    },
+    {
+      "epoch": 35.001714285714286,
+      "grad_norm": 0.5483258962631226,
+      "learning_rate": 3.3149206349206354e-06,
+      "loss": 0.1753,
+      "step": 12280
+    },
+    {
+      "epoch": 35.00228571428571,
+      "grad_norm": 0.0009629835840314627,
+      "learning_rate": 3.308571428571429e-06,
+      "loss": 0.1238,
+      "step": 12290
+    },
+    {
+      "epoch": 35.002857142857145,
+      "grad_norm": 0.543134331703186,
+      "learning_rate": 3.3022222222222223e-06,
+      "loss": 0.1245,
+      "step": 12300
+    },
+    {
+      "epoch": 35.00342857142857,
+      "grad_norm": 0.08552040904760361,
+      "learning_rate": 3.2958730158730162e-06,
+      "loss": 0.5037,
+      "step": 12310
+    },
+    {
+      "epoch": 35.004,
+      "grad_norm": 0.015447727404534817,
+      "learning_rate": 3.2895238095238097e-06,
+      "loss": 0.5062,
+      "step": 12320
+    },
+    {
+      "epoch": 35.00457142857143,
+      "grad_norm": 0.004400096833705902,
+      "learning_rate": 3.2831746031746036e-06,
+      "loss": 0.2552,
+      "step": 12330
+    },
+    {
+      "epoch": 35.00514285714286,
+      "grad_norm": 0.018583452329039574,
+      "learning_rate": 3.276825396825397e-06,
+      "loss": 0.3969,
+      "step": 12340
+    },
+    {
+      "epoch": 35.005714285714284,
+      "grad_norm": 0.04393948242068291,
+      "learning_rate": 3.2704761904761906e-06,
+      "loss": 0.2678,
+      "step": 12350
+    },
+    {
+      "epoch": 35.00628571428572,
+      "grad_norm": 0.05897550284862518,
+      "learning_rate": 3.2641269841269845e-06,
+      "loss": 0.0057,
+      "step": 12360
+    },
+    {
+      "epoch": 35.00685714285714,
+      "grad_norm": 0.004284753929823637,
+      "learning_rate": 3.257777777777778e-06,
+      "loss": 0.2586,
+      "step": 12370
+    },
+    {
+      "epoch": 35.00742857142857,
+      "grad_norm": 0.007391383405774832,
+      "learning_rate": 3.2514285714285715e-06,
+      "loss": 0.6294,
+      "step": 12380
+    },
+    {
+      "epoch": 35.008,
+      "grad_norm": 0.05153246968984604,
+      "learning_rate": 3.2450793650793654e-06,
+      "loss": 0.6797,
+      "step": 12390
+    },
+    {
+      "epoch": 35.00857142857143,
+      "grad_norm": 20.67218017578125,
+      "learning_rate": 3.238730158730159e-06,
+      "loss": 0.2888,
+      "step": 12400
+    },
+    {
+      "epoch": 35.009142857142855,
+      "grad_norm": 0.0017453532200306654,
+      "learning_rate": 3.2323809523809523e-06,
+      "loss": 0.0024,
+      "step": 12410
+    },
+    {
+      "epoch": 35.00971428571429,
+      "grad_norm": 0.6204124689102173,
+      "learning_rate": 3.2260317460317462e-06,
+      "loss": 0.3306,
+      "step": 12420
+    },
+    {
+      "epoch": 35.010285714285715,
+      "grad_norm": 0.11453156918287277,
+      "learning_rate": 3.2196825396825397e-06,
+      "loss": 0.6737,
+      "step": 12430
+    },
+    {
+      "epoch": 35.01085714285714,
+      "grad_norm": 116.97684478759766,
+      "learning_rate": 3.213333333333334e-06,
+      "loss": 0.453,
+      "step": 12440
+    },
+    {
+      "epoch": 35.011428571428574,
+      "grad_norm": 0.036407049745321274,
+      "learning_rate": 3.206984126984127e-06,
+      "loss": 0.3485,
+      "step": 12450
+    },
+    {
+      "epoch": 35.012,
+      "grad_norm": 0.003050927072763443,
+      "learning_rate": 3.2006349206349206e-06,
+      "loss": 0.4884,
+      "step": 12460
+    },
+    {
+      "epoch": 35.01257142857143,
+      "grad_norm": 0.2217385321855545,
+      "learning_rate": 3.194285714285715e-06,
+      "loss": 0.5697,
+      "step": 12470
+    },
+    {
+      "epoch": 35.01314285714286,
+      "grad_norm": 4.006773471832275,
+      "learning_rate": 3.1879365079365084e-06,
+      "loss": 0.0041,
+      "step": 12480
+    },
+    {
+      "epoch": 35.013714285714286,
+      "grad_norm": 0.09846015274524689,
+      "learning_rate": 3.1815873015873015e-06,
+      "loss": 0.0028,
+      "step": 12490
+    },
+    {
+      "epoch": 35.01428571428571,
+      "grad_norm": 0.00579382898285985,
+      "learning_rate": 3.175238095238096e-06,
+      "loss": 0.0024,
+      "step": 12500
+    },
+    {
+      "epoch": 35.014857142857146,
+      "grad_norm": 0.08295068144798279,
+      "learning_rate": 3.1688888888888893e-06,
+      "loss": 0.001,
+      "step": 12510
+    },
+    {
+      "epoch": 35.01542857142857,
+      "grad_norm": 0.0009396121604368091,
+      "learning_rate": 3.1625396825396828e-06,
+      "loss": 0.0129,
+      "step": 12520
+    },
+    {
+      "epoch": 35.016,
+      "grad_norm": 0.007435150910168886,
+      "learning_rate": 3.1561904761904767e-06,
+      "loss": 0.0014,
+      "step": 12530
+    },
+    {
+      "epoch": 35.01657142857143,
+      "grad_norm": 0.03919130563735962,
+      "learning_rate": 3.14984126984127e-06,
+      "loss": 0.6772,
+      "step": 12540
+    },
+    {
+      "epoch": 35.01714285714286,
+      "grad_norm": 0.03321586921811104,
+      "learning_rate": 3.1434920634920636e-06,
+      "loss": 0.4575,
+      "step": 12550
+    },
+    {
+      "epoch": 35.017714285714284,
+      "grad_norm": 0.19160214066505432,
+      "learning_rate": 3.1371428571428575e-06,
+      "loss": 0.195,
+      "step": 12560
+    },
+    {
+      "epoch": 35.01828571428572,
+      "grad_norm": 0.0026826318353414536,
+      "learning_rate": 3.130793650793651e-06,
+      "loss": 0.0045,
+      "step": 12570
+    },
+    {
+      "epoch": 35.018857142857144,
+      "grad_norm": 0.038780469447374344,
+      "learning_rate": 3.124444444444445e-06,
+      "loss": 0.0011,
+      "step": 12580
+    },
+    {
+      "epoch": 35.01942857142857,
+      "grad_norm": 0.07216961681842804,
+      "learning_rate": 3.1180952380952384e-06,
+      "loss": 0.0011,
+      "step": 12590
+    },
+    {
+      "epoch": 35.02,
+      "grad_norm": 0.026877250522375107,
+      "learning_rate": 3.111746031746032e-06,
+      "loss": 0.0011,
+      "step": 12600
+    },
+    {
+      "epoch": 35.02,
+      "eval_accuracy": 0.8282828282828283,
+      "eval_loss": 1.0746172666549683,
+      "eval_runtime": 126.6057,
+      "eval_samples_per_second": 2.346,
+      "eval_steps_per_second": 1.177,
+      "step": 12600
+    },
+    {
+      "epoch": 36.000571428571426,
+      "grad_norm": 0.0886671394109726,
+      "learning_rate": 3.105396825396826e-06,
+      "loss": 0.2623,
+      "step": 12610
+    },
+    {
+      "epoch": 36.00114285714286,
+      "grad_norm": 0.012463954277336597,
+      "learning_rate": 3.0990476190476193e-06,
+      "loss": 0.4357,
+      "step": 12620
+    },
+    {
+      "epoch": 36.001714285714286,
+      "grad_norm": 0.026730680838227272,
+      "learning_rate": 3.0926984126984128e-06,
+      "loss": 0.2669,
+      "step": 12630
+    },
+    {
+      "epoch": 36.00228571428571,
+      "grad_norm": 0.07718096673488617,
+      "learning_rate": 3.0863492063492067e-06,
+      "loss": 0.5815,
+      "step": 12640
+    },
+    {
+      "epoch": 36.002857142857145,
+      "grad_norm": 0.2121714949607849,
+      "learning_rate": 3.08e-06,
+      "loss": 0.1494,
+      "step": 12650
+    },
+    {
+      "epoch": 36.00342857142857,
+      "grad_norm": 0.1541670709848404,
+      "learning_rate": 3.0736507936507936e-06,
+      "loss": 0.0016,
+      "step": 12660
+    },
+    {
+      "epoch": 36.004,
+      "grad_norm": 0.030359311029314995,
+      "learning_rate": 3.0673015873015875e-06,
+      "loss": 0.0706,
+      "step": 12670
+    },
+    {
+      "epoch": 36.00457142857143,
+      "grad_norm": 0.07423070073127747,
+      "learning_rate": 3.060952380952381e-06,
+      "loss": 0.234,
+      "step": 12680
+    },
+    {
+      "epoch": 36.00514285714286,
+      "grad_norm": 0.005180784966796637,
+      "learning_rate": 3.054603174603175e-06,
+      "loss": 0.2837,
+      "step": 12690
+    },
+    {
+      "epoch": 36.005714285714284,
+      "grad_norm": 345.4854736328125,
+      "learning_rate": 3.0482539682539684e-06,
+      "loss": 0.3923,
+      "step": 12700
+    },
+    {
+      "epoch": 36.00628571428572,
+      "grad_norm": 0.0064833336509764194,
+      "learning_rate": 3.041904761904762e-06,
+      "loss": 0.0015,
+      "step": 12710
+    },
+    {
+      "epoch": 36.00685714285714,
+      "grad_norm": 0.04605857655405998,
+      "learning_rate": 3.0355555555555562e-06,
+      "loss": 0.0033,
+      "step": 12720
+    },
+    {
+      "epoch": 36.00742857142857,
+      "grad_norm": 0.10902780294418335,
+      "learning_rate": 3.0292063492063493e-06,
+      "loss": 0.0027,
+      "step": 12730
+    },
+    {
+      "epoch": 36.008,
+      "grad_norm": 0.006975686177611351,
+      "learning_rate": 3.0228571428571428e-06,
+      "loss": 0.4695,
+      "step": 12740
+    },
+    {
+      "epoch": 36.00857142857143,
+      "grad_norm": 0.09180530905723572,
+      "learning_rate": 3.016507936507937e-06,
+      "loss": 0.0007,
+      "step": 12750
+    },
+    {
+      "epoch": 36.009142857142855,
+      "grad_norm": 0.007026137318462133,
+      "learning_rate": 3.0101587301587306e-06,
+      "loss": 0.4985,
+      "step": 12760
+    },
+    {
+      "epoch": 36.00971428571429,
+      "grad_norm": 0.005862717051059008,
+      "learning_rate": 3.0038095238095236e-06,
+      "loss": 0.163,
+      "step": 12770
+    },
+    {
+      "epoch": 36.010285714285715,
+      "grad_norm": 0.05073744058609009,
+      "learning_rate": 2.997460317460318e-06,
+      "loss": 0.002,
+      "step": 12780
+    },
+    {
+      "epoch": 36.01085714285714,
+      "grad_norm": 0.017824208363890648,
+      "learning_rate": 2.9911111111111115e-06,
+      "loss": 0.4371,
+      "step": 12790
+    },
+    {
+      "epoch": 36.011428571428574,
+      "grad_norm": 13.799386024475098,
+      "learning_rate": 2.984761904761905e-06,
+      "loss": 0.6038,
+      "step": 12800
+    },
+    {
+      "epoch": 36.012,
+      "grad_norm": 199.8282012939453,
+      "learning_rate": 2.978412698412699e-06,
+      "loss": 0.4218,
+      "step": 12810
+    },
+    {
+      "epoch": 36.01257142857143,
+      "grad_norm": 0.15811072289943695,
+      "learning_rate": 2.9720634920634923e-06,
+      "loss": 0.0011,
+      "step": 12820
+    },
+    {
+      "epoch": 36.01314285714286,
+      "grad_norm": 0.17585258185863495,
+      "learning_rate": 2.9657142857142862e-06,
+      "loss": 0.4932,
+      "step": 12830
+    },
+    {
+      "epoch": 36.013714285714286,
+      "grad_norm": 16.838157653808594,
+      "learning_rate": 2.9593650793650797e-06,
+      "loss": 1.1193,
+      "step": 12840
+    },
+    {
+      "epoch": 36.01428571428571,
+      "grad_norm": 0.04459202662110329,
+      "learning_rate": 2.953015873015873e-06,
+      "loss": 0.184,
+      "step": 12850
+    },
+    {
+      "epoch": 36.014857142857146,
+      "grad_norm": 0.38909366726875305,
+      "learning_rate": 2.946666666666667e-06,
+      "loss": 0.2212,
+      "step": 12860
+    },
+    {
+      "epoch": 36.01542857142857,
+      "grad_norm": 0.1248052716255188,
+      "learning_rate": 2.9403174603174606e-06,
+      "loss": 0.0016,
+      "step": 12870
+    },
+    {
+      "epoch": 36.016,
+      "grad_norm": 0.5679760575294495,
+      "learning_rate": 2.933968253968254e-06,
+      "loss": 0.0029,
+      "step": 12880
+    },
+    {
+      "epoch": 36.01657142857143,
+      "grad_norm": 0.02419651672244072,
+      "learning_rate": 2.927619047619048e-06,
+      "loss": 0.2692,
+      "step": 12890
+    },
+    {
+      "epoch": 36.01714285714286,
+      "grad_norm": 0.2062833309173584,
+      "learning_rate": 2.9212698412698415e-06,
+      "loss": 0.1403,
+      "step": 12900
+    },
+    {
+      "epoch": 36.017714285714284,
+      "grad_norm": 0.21458952128887177,
+      "learning_rate": 2.914920634920635e-06,
+      "loss": 0.4953,
+      "step": 12910
+    },
+    {
+      "epoch": 36.01828571428572,
+      "grad_norm": 0.24665455520153046,
+      "learning_rate": 2.908571428571429e-06,
+      "loss": 0.7257,
+      "step": 12920
+    },
+    {
+      "epoch": 36.018857142857144,
+      "grad_norm": 0.5908558368682861,
+      "learning_rate": 2.9022222222222223e-06,
+      "loss": 0.005,
+      "step": 12930
+    },
+    {
+      "epoch": 36.01942857142857,
+      "grad_norm": 0.19099822640419006,
+      "learning_rate": 2.895873015873016e-06,
+      "loss": 0.0019,
+      "step": 12940
+    },
+    {
+      "epoch": 36.02,
+      "grad_norm": 0.05261155590415001,
+      "learning_rate": 2.8895238095238097e-06,
+      "loss": 0.3917,
+      "step": 12950
+    },
+    {
+      "epoch": 36.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 0.9695500135421753,
+      "eval_runtime": 126.7723,
+      "eval_samples_per_second": 2.343,
+      "eval_steps_per_second": 1.175,
+      "step": 12950
+    },
+    {
+      "epoch": 37.000571428571426,
+      "grad_norm": 0.07244765758514404,
+      "learning_rate": 2.8831746031746032e-06,
+      "loss": 0.0006,
+      "step": 12960
+    },
+    {
+      "epoch": 37.00114285714286,
+      "grad_norm": 0.04542018845677376,
+      "learning_rate": 2.876825396825397e-06,
+      "loss": 0.4526,
+      "step": 12970
+    },
+    {
+      "epoch": 37.001714285714286,
+      "grad_norm": 0.007704177405685186,
+      "learning_rate": 2.8704761904761906e-06,
+      "loss": 0.0015,
+      "step": 12980
+    },
+    {
+      "epoch": 37.00228571428571,
+      "grad_norm": 0.01741478592157364,
+      "learning_rate": 2.864126984126984e-06,
+      "loss": 0.4346,
+      "step": 12990
+    },
+    {
+      "epoch": 37.002857142857145,
+      "grad_norm": 0.020365318283438683,
+      "learning_rate": 2.8577777777777784e-06,
+      "loss": 0.2763,
+      "step": 13000
+    },
+    {
+      "epoch": 37.00342857142857,
+      "grad_norm": 0.10666250437498093,
+      "learning_rate": 2.8514285714285715e-06,
+      "loss": 0.1876,
+      "step": 13010
+    },
+    {
+      "epoch": 37.004,
+      "grad_norm": 0.0016732965596020222,
+      "learning_rate": 2.845079365079365e-06,
+      "loss": 0.0018,
+      "step": 13020
+    },
+    {
+      "epoch": 37.00457142857143,
+      "grad_norm": 0.025395436212420464,
+      "learning_rate": 2.8387301587301593e-06,
+      "loss": 0.0012,
+      "step": 13030
+    },
+    {
+      "epoch": 37.00514285714286,
+      "grad_norm": 2.5686378479003906,
+      "learning_rate": 2.8323809523809528e-06,
+      "loss": 0.197,
+      "step": 13040
+    },
+    {
+      "epoch": 37.005714285714284,
+      "grad_norm": 0.044472586363554,
+      "learning_rate": 2.826031746031746e-06,
+      "loss": 0.0021,
+      "step": 13050
+    },
+    {
+      "epoch": 37.00628571428572,
+      "grad_norm": 0.05117283761501312,
+      "learning_rate": 2.81968253968254e-06,
+      "loss": 0.0008,
+      "step": 13060
+    },
+    {
+      "epoch": 37.00685714285714,
+      "grad_norm": 0.1931295245885849,
+      "learning_rate": 2.8133333333333336e-06,
+      "loss": 0.0012,
+      "step": 13070
+    },
+    {
+      "epoch": 37.00742857142857,
+      "grad_norm": 0.0703316405415535,
+      "learning_rate": 2.8069841269841276e-06,
+      "loss": 0.0013,
+      "step": 13080
+    },
+    {
+      "epoch": 37.008,
+      "grad_norm": 0.03455556929111481,
+      "learning_rate": 2.800634920634921e-06,
+      "loss": 0.0009,
+      "step": 13090
+    },
+    {
+      "epoch": 37.00857142857143,
+      "grad_norm": 0.007254268042743206,
+      "learning_rate": 2.7942857142857145e-06,
+      "loss": 0.906,
+      "step": 13100
+    },
+    {
+      "epoch": 37.009142857142855,
+      "grad_norm": 0.01259919349104166,
+      "learning_rate": 2.7879365079365084e-06,
+      "loss": 0.5051,
+      "step": 13110
+    },
+    {
+      "epoch": 37.00971428571429,
+      "grad_norm": 0.05640314146876335,
+      "learning_rate": 2.781587301587302e-06,
+      "loss": 0.0017,
+      "step": 13120
+    },
+    {
+      "epoch": 37.010285714285715,
+      "grad_norm": 0.0015288791619241238,
+      "learning_rate": 2.7752380952380954e-06,
+      "loss": 0.6821,
+      "step": 13130
+    },
+    {
+      "epoch": 37.01085714285714,
+      "grad_norm": 0.1644653081893921,
+      "learning_rate": 2.7688888888888893e-06,
+      "loss": 0.1161,
+      "step": 13140
+    },
+    {
+      "epoch": 37.011428571428574,
+      "grad_norm": 0.0032219102140516043,
+      "learning_rate": 2.7625396825396828e-06,
+      "loss": 0.0012,
+      "step": 13150
+    },
+    {
+      "epoch": 37.012,
+      "grad_norm": 0.1719381958246231,
+      "learning_rate": 2.7561904761904763e-06,
+      "loss": 0.0014,
+      "step": 13160
+    },
+    {
+      "epoch": 37.01257142857143,
+      "grad_norm": 0.04314820095896721,
+      "learning_rate": 2.74984126984127e-06,
+      "loss": 0.0016,
+      "step": 13170
+    },
+    {
+      "epoch": 37.01314285714286,
+      "grad_norm": 0.05535136163234711,
+      "learning_rate": 2.7434920634920637e-06,
+      "loss": 0.0016,
+      "step": 13180
+    },
+    {
+      "epoch": 37.013714285714286,
+      "grad_norm": 0.004680715035647154,
+      "learning_rate": 2.737142857142857e-06,
+      "loss": 0.0195,
+      "step": 13190
+    },
+    {
+      "epoch": 37.01428571428571,
+      "grad_norm": 0.020550280809402466,
+      "learning_rate": 2.730793650793651e-06,
+      "loss": 0.0009,
+      "step": 13200
+    },
+    {
+      "epoch": 37.014857142857146,
+      "grad_norm": 0.02597636915743351,
+      "learning_rate": 2.7244444444444445e-06,
+      "loss": 0.3391,
+      "step": 13210
+    },
+    {
+      "epoch": 37.01542857142857,
+      "grad_norm": 0.022728653624653816,
+      "learning_rate": 2.7180952380952384e-06,
+      "loss": 0.3329,
+      "step": 13220
+    },
+    {
+      "epoch": 37.016,
+      "grad_norm": 37.18050003051758,
+      "learning_rate": 2.711746031746032e-06,
+      "loss": 1.1542,
+      "step": 13230
+    },
+    {
+      "epoch": 37.01657142857143,
+      "grad_norm": 0.11314905434846878,
+      "learning_rate": 2.7053968253968254e-06,
+      "loss": 0.3948,
+      "step": 13240
+    },
+    {
+      "epoch": 37.01714285714286,
+      "grad_norm": 16.31825065612793,
+      "learning_rate": 2.6990476190476193e-06,
+      "loss": 0.4456,
+      "step": 13250
+    },
+    {
+      "epoch": 37.017714285714284,
+      "grad_norm": 0.0014047048753127456,
+      "learning_rate": 2.6926984126984128e-06,
+      "loss": 0.2791,
+      "step": 13260
+    },
+    {
+      "epoch": 37.01828571428572,
+      "grad_norm": 0.059588722884655,
+      "learning_rate": 2.6863492063492063e-06,
+      "loss": 0.2329,
+      "step": 13270
+    },
+    {
+      "epoch": 37.018857142857144,
+      "grad_norm": 0.005540232639759779,
+      "learning_rate": 2.68e-06,
+      "loss": 0.0019,
+      "step": 13280
+    },
+    {
+      "epoch": 37.01942857142857,
+      "grad_norm": 18.04555892944336,
+      "learning_rate": 2.6736507936507937e-06,
+      "loss": 0.7975,
+      "step": 13290
+    },
+    {
+      "epoch": 37.02,
+      "grad_norm": 0.0029781581833958626,
+      "learning_rate": 2.667301587301587e-06,
+      "loss": 0.7268,
+      "step": 13300
+    },
+    {
+      "epoch": 37.02,
+      "eval_accuracy": 0.8181818181818182,
+      "eval_loss": 1.1061733961105347,
+      "eval_runtime": 126.817,
+      "eval_samples_per_second": 2.342,
+      "eval_steps_per_second": 1.175,
+      "step": 13300
+    },
+    {
+      "epoch": 38.000571428571426,
+      "grad_norm": 0.47502627968788147,
+      "learning_rate": 2.6609523809523815e-06,
+      "loss": 0.3805,
+      "step": 13310
+    },
+    {
+      "epoch": 38.00114285714286,
+      "grad_norm": 0.08994001895189285,
+      "learning_rate": 2.6546031746031745e-06,
+      "loss": 0.003,
+      "step": 13320
+    },
+    {
+      "epoch": 38.001714285714286,
+      "grad_norm": 0.06851596385240555,
+      "learning_rate": 2.648253968253969e-06,
+      "loss": 0.0035,
+      "step": 13330
+    },
+    {
+      "epoch": 38.00228571428571,
+      "grad_norm": 0.1989380419254303,
+      "learning_rate": 2.6419047619047623e-06,
+      "loss": 0.3274,
+      "step": 13340
+    },
+    {
+      "epoch": 38.002857142857145,
+      "grad_norm": 0.04935070872306824,
+      "learning_rate": 2.635555555555556e-06,
+      "loss": 0.4874,
+      "step": 13350
+    },
+    {
+      "epoch": 38.00342857142857,
+      "grad_norm": 0.04223153740167618,
+      "learning_rate": 2.6292063492063497e-06,
+      "loss": 0.2244,
+      "step": 13360
+    },
+    {
+      "epoch": 38.004,
+      "grad_norm": 0.06442692130804062,
+      "learning_rate": 2.6228571428571432e-06,
+      "loss": 0.005,
+      "step": 13370
+    },
+    {
+      "epoch": 38.00457142857143,
+      "grad_norm": 0.04460546746850014,
+      "learning_rate": 2.6165079365079367e-06,
+      "loss": 0.238,
+      "step": 13380
+    },
+    {
+      "epoch": 38.00514285714286,
+      "grad_norm": 0.0043782079592347145,
+      "learning_rate": 2.6101587301587306e-06,
+      "loss": 0.2346,
+      "step": 13390
+    },
+    {
+      "epoch": 38.005714285714284,
+      "grad_norm": 0.05058957263827324,
+      "learning_rate": 2.603809523809524e-06,
+      "loss": 0.4129,
+      "step": 13400
+    },
+    {
+      "epoch": 38.00628571428572,
+      "grad_norm": 0.020779475569725037,
+      "learning_rate": 2.5974603174603176e-06,
+      "loss": 0.154,
+      "step": 13410
+    },
+    {
+      "epoch": 38.00685714285714,
+      "grad_norm": 0.3094046413898468,
+      "learning_rate": 2.5911111111111115e-06,
+      "loss": 0.3172,
+      "step": 13420
+    },
+    {
+      "epoch": 38.00742857142857,
+      "grad_norm": 0.38032662868499756,
+      "learning_rate": 2.584761904761905e-06,
+      "loss": 0.2117,
+      "step": 13430
+    },
+    {
+      "epoch": 38.008,
+      "grad_norm": 0.004349476657807827,
+      "learning_rate": 2.5784126984126984e-06,
+      "loss": 0.2803,
+      "step": 13440
+    },
+    {
+      "epoch": 38.00857142857143,
+      "grad_norm": 0.06518778204917908,
+      "learning_rate": 2.5720634920634924e-06,
+      "loss": 0.2218,
+      "step": 13450
+    },
+    {
+      "epoch": 38.009142857142855,
+      "grad_norm": 0.09332817047834396,
+      "learning_rate": 2.565714285714286e-06,
+      "loss": 0.2499,
+      "step": 13460
+    },
+    {
+      "epoch": 38.00971428571429,
+      "grad_norm": 0.11110708862543106,
+      "learning_rate": 2.5593650793650797e-06,
+      "loss": 0.6162,
+      "step": 13470
+    },
+    {
+      "epoch": 38.010285714285715,
+      "grad_norm": 0.07068169862031937,
+      "learning_rate": 2.5530158730158732e-06,
+      "loss": 0.4614,
+      "step": 13480
+    },
+    {
+      "epoch": 38.01085714285714,
+      "grad_norm": 0.0015391431516036391,
+      "learning_rate": 2.5466666666666667e-06,
+      "loss": 0.3118,
+      "step": 13490
+    },
+    {
+      "epoch": 38.011428571428574,
+      "grad_norm": 0.05324092507362366,
+      "learning_rate": 2.5403174603174606e-06,
+      "loss": 0.0024,
+      "step": 13500
+    },
+    {
+      "epoch": 38.012,
+      "grad_norm": 0.03473285958170891,
+      "learning_rate": 2.533968253968254e-06,
+      "loss": 0.1767,
+      "step": 13510
+    },
+    {
+      "epoch": 38.01257142857143,
+      "grad_norm": 0.05727458372712135,
+      "learning_rate": 2.5276190476190476e-06,
+      "loss": 0.2763,
+      "step": 13520
+    },
+    {
+      "epoch": 38.01314285714286,
+      "grad_norm": 0.04184458777308464,
+      "learning_rate": 2.5212698412698415e-06,
+      "loss": 0.0069,
+      "step": 13530
+    },
+    {
+      "epoch": 38.013714285714286,
+      "grad_norm": 0.003181879874318838,
+      "learning_rate": 2.514920634920635e-06,
+      "loss": 0.001,
+      "step": 13540
+    },
+    {
+      "epoch": 38.01428571428571,
+      "grad_norm": 0.019657937809824944,
+      "learning_rate": 2.5085714285714285e-06,
+      "loss": 0.0081,
+      "step": 13550
+    },
+    {
+      "epoch": 38.014857142857146,
+      "grad_norm": 0.032800447195768356,
+      "learning_rate": 2.5022222222222224e-06,
+      "loss": 0.0018,
+      "step": 13560
+    },
+    {
+      "epoch": 38.01542857142857,
+      "grad_norm": 0.04382657632231712,
+      "learning_rate": 2.495873015873016e-06,
+      "loss": 0.0013,
+      "step": 13570
+    },
+    {
+      "epoch": 38.016,
+      "grad_norm": 0.008959997445344925,
+      "learning_rate": 2.4895238095238097e-06,
+      "loss": 0.0004,
+      "step": 13580
+    },
+    {
+      "epoch": 38.01657142857143,
+      "grad_norm": 0.013616718351840973,
+      "learning_rate": 2.4831746031746037e-06,
+      "loss": 0.0008,
+      "step": 13590
+    },
+    {
+      "epoch": 38.01714285714286,
+      "grad_norm": 27.68400001525879,
+      "learning_rate": 2.4768253968253967e-06,
+      "loss": 0.4851,
+      "step": 13600
+    },
+    {
+      "epoch": 38.017714285714284,
+      "grad_norm": 20.209680557250977,
+      "learning_rate": 2.4704761904761906e-06,
+      "loss": 0.3812,
+      "step": 13610
+    },
+    {
+      "epoch": 38.01828571428572,
+      "grad_norm": 0.13937675952911377,
+      "learning_rate": 2.4641269841269845e-06,
+      "loss": 0.2243,
+      "step": 13620
+    },
+    {
+      "epoch": 38.018857142857144,
+      "grad_norm": 0.0037647627759724855,
+      "learning_rate": 2.457777777777778e-06,
+      "loss": 0.6971,
+      "step": 13630
+    },
+    {
+      "epoch": 38.01942857142857,
+      "grad_norm": 0.033469632267951965,
+      "learning_rate": 2.4514285714285715e-06,
+      "loss": 0.3434,
+      "step": 13640
+    },
+    {
+      "epoch": 38.02,
+      "grad_norm": 0.023815317079424858,
+      "learning_rate": 2.4450793650793654e-06,
+      "loss": 0.3747,
+      "step": 13650
+    },
+    {
+      "epoch": 38.02,
+      "eval_accuracy": 0.835016835016835,
+      "eval_loss": 1.036763072013855,
+      "eval_runtime": 126.6279,
+      "eval_samples_per_second": 2.345,
+      "eval_steps_per_second": 1.177,
+      "step": 13650
+    },
+    {
+      "epoch": 39.000571428571426,
+      "grad_norm": 49.7794075012207,
+      "learning_rate": 2.438730158730159e-06,
+      "loss": 0.003,
+      "step": 13660
+    },
+    {
+      "epoch": 39.00114285714286,
+      "grad_norm": 0.03126109018921852,
+      "learning_rate": 2.4323809523809524e-06,
+      "loss": 0.1622,
+      "step": 13670
+    },
+    {
+      "epoch": 39.001714285714286,
+      "grad_norm": 0.047299765050411224,
+      "learning_rate": 2.4260317460317463e-06,
+      "loss": 0.2872,
+      "step": 13680
+    },
+    {
+      "epoch": 39.00228571428571,
+      "grad_norm": 0.05949264392256737,
+      "learning_rate": 2.4196825396825398e-06,
+      "loss": 0.2092,
+      "step": 13690
+    },
+    {
+      "epoch": 39.002857142857145,
+      "grad_norm": 0.026437293738126755,
+      "learning_rate": 2.4133333333333337e-06,
+      "loss": 0.0031,
+      "step": 13700
+    },
+    {
+      "epoch": 39.00342857142857,
+      "grad_norm": 0.20057931542396545,
+      "learning_rate": 2.406984126984127e-06,
+      "loss": 0.4873,
+      "step": 13710
+    },
+    {
+      "epoch": 39.004,
+      "grad_norm": 0.004649260081350803,
+      "learning_rate": 2.4006349206349206e-06,
+      "loss": 0.2073,
+      "step": 13720
+    },
+    {
+      "epoch": 39.00457142857143,
+      "grad_norm": 0.5119889974594116,
+      "learning_rate": 2.3942857142857145e-06,
+      "loss": 0.4199,
+      "step": 13730
+    },
+    {
+      "epoch": 39.00514285714286,
+      "grad_norm": 0.16033364832401276,
+      "learning_rate": 2.387936507936508e-06,
+      "loss": 0.0135,
+      "step": 13740
+    },
+    {
+      "epoch": 39.005714285714284,
+      "grad_norm": 0.0032101422548294067,
+      "learning_rate": 2.381587301587302e-06,
+      "loss": 0.9572,
+      "step": 13750
+    },
+    {
+      "epoch": 39.00628571428572,
+      "grad_norm": 0.026194965466856956,
+      "learning_rate": 2.3752380952380954e-06,
+      "loss": 1.2531,
+      "step": 13760
+    },
+    {
+      "epoch": 39.00685714285714,
+      "grad_norm": 0.0062312232330441475,
+      "learning_rate": 2.3688888888888893e-06,
+      "loss": 0.2321,
+      "step": 13770
+    },
+    {
+      "epoch": 39.00742857142857,
+      "grad_norm": 0.07098673284053802,
+      "learning_rate": 2.362539682539683e-06,
+      "loss": 0.0025,
+      "step": 13780
+    },
+    {
+      "epoch": 39.008,
+      "grad_norm": 0.0009907495696097612,
+      "learning_rate": 2.3561904761904763e-06,
+      "loss": 0.563,
+      "step": 13790
+    },
+    {
+      "epoch": 39.00857142857143,
+      "grad_norm": 0.0647115707397461,
+      "learning_rate": 2.34984126984127e-06,
+      "loss": 0.8326,
+      "step": 13800
+    },
+    {
+      "epoch": 39.009142857142855,
+      "grad_norm": 0.0811251774430275,
+      "learning_rate": 2.3434920634920637e-06,
+      "loss": 0.6204,
+      "step": 13810
+    },
+    {
+      "epoch": 39.00971428571429,
+      "grad_norm": 0.011888241395354271,
+      "learning_rate": 2.337142857142857e-06,
+      "loss": 0.0029,
+      "step": 13820
+    },
+    {
+      "epoch": 39.010285714285715,
+      "grad_norm": 0.06965713948011398,
+      "learning_rate": 2.330793650793651e-06,
+      "loss": 0.0021,
+      "step": 13830
+    },
+    {
+      "epoch": 39.01085714285714,
+      "grad_norm": 0.0806642472743988,
+      "learning_rate": 2.3244444444444445e-06,
+      "loss": 0.2988,
+      "step": 13840
+    },
+    {
+      "epoch": 39.011428571428574,
+      "grad_norm": 0.007325103506445885,
+      "learning_rate": 2.318095238095238e-06,
+      "loss": 0.0013,
+      "step": 13850
+    },
+    {
+      "epoch": 39.012,
+      "grad_norm": 0.11560201644897461,
+      "learning_rate": 2.311746031746032e-06,
+      "loss": 0.1653,
+      "step": 13860
+    },
+    {
+      "epoch": 39.01257142857143,
+      "grad_norm": 0.033399466425180435,
+      "learning_rate": 2.305396825396826e-06,
+      "loss": 0.0027,
+      "step": 13870
+    },
+    {
+      "epoch": 39.01314285714286,
+      "grad_norm": 1141.728515625,
+      "learning_rate": 2.2990476190476193e-06,
+      "loss": 0.0865,
+      "step": 13880
+    },
+    {
+      "epoch": 39.013714285714286,
+      "grad_norm": 0.0021837761159986258,
+      "learning_rate": 2.292698412698413e-06,
+      "loss": 0.2472,
+      "step": 13890
+    },
+    {
+      "epoch": 39.01428571428571,
+      "grad_norm": 0.008280741050839424,
+      "learning_rate": 2.2863492063492067e-06,
+      "loss": 0.2712,
+      "step": 13900
+    },
+    {
+      "epoch": 39.014857142857146,
+      "grad_norm": 0.00286860391497612,
+      "learning_rate": 2.28e-06,
+      "loss": 0.2084,
+      "step": 13910
+    },
+    {
+      "epoch": 39.01542857142857,
+      "grad_norm": 0.012107732705771923,
+      "learning_rate": 2.2736507936507937e-06,
+      "loss": 0.0006,
+      "step": 13920
+    },
+    {
+      "epoch": 39.016,
+      "grad_norm": 0.0011101092677563429,
+      "learning_rate": 2.2673015873015876e-06,
+      "loss": 0.3675,
+      "step": 13930
+    },
+    {
+      "epoch": 39.01657142857143,
+      "grad_norm": 0.03219223394989967,
+      "learning_rate": 2.260952380952381e-06,
+      "loss": 0.0032,
+      "step": 13940
+    },
+    {
+      "epoch": 39.01714285714286,
+      "grad_norm": 0.0029516194481402636,
+      "learning_rate": 2.254603174603175e-06,
+      "loss": 0.0086,
+      "step": 13950
+    },
+    {
+      "epoch": 39.017714285714284,
+      "grad_norm": 0.030327564105391502,
+      "learning_rate": 2.2482539682539685e-06,
+      "loss": 0.2982,
+      "step": 13960
+    },
+    {
+      "epoch": 39.01828571428572,
+      "grad_norm": 0.010290997102856636,
+      "learning_rate": 2.241904761904762e-06,
+      "loss": 0.2149,
+      "step": 13970
+    },
+    {
+      "epoch": 39.018857142857144,
+      "grad_norm": 0.031805507838726044,
+      "learning_rate": 2.235555555555556e-06,
+      "loss": 0.1666,
+      "step": 13980
+    },
+    {
+      "epoch": 39.01942857142857,
+      "grad_norm": 0.05369047820568085,
+      "learning_rate": 2.2292063492063493e-06,
+      "loss": 0.0689,
+      "step": 13990
+    },
+    {
+      "epoch": 39.02,
+      "grad_norm": 0.07717669010162354,
+      "learning_rate": 2.222857142857143e-06,
+      "loss": 0.5584,
+      "step": 14000
+    },
+    {
+      "epoch": 39.02,
+      "eval_accuracy": 0.8417508417508418,
+      "eval_loss": 1.0148627758026123,
+      "eval_runtime": 126.757,
+      "eval_samples_per_second": 2.343,
+      "eval_steps_per_second": 1.175,
+      "step": 14000
+    },
+    {
+      "epoch": 40.000571428571426,
+      "grad_norm": 0.2859209179878235,
+      "learning_rate": 2.2165079365079367e-06,
+      "loss": 0.001,
+      "step": 14010
+    },
+    {
+      "epoch": 40.00114285714286,
+      "grad_norm": 0.06745479255914688,
+      "learning_rate": 2.2101587301587306e-06,
+      "loss": 0.0017,
+      "step": 14020
+    },
+    {
+      "epoch": 40.001714285714286,
+      "grad_norm": 0.08140210062265396,
+      "learning_rate": 2.203809523809524e-06,
+      "loss": 0.0011,
+      "step": 14030
+    },
+    {
+      "epoch": 40.00228571428571,
+      "grad_norm": 27.934417724609375,
+      "learning_rate": 2.1974603174603176e-06,
+      "loss": 0.1988,
+      "step": 14040
+    },
+    {
+      "epoch": 40.002857142857145,
+      "grad_norm": 0.01658036932349205,
+      "learning_rate": 2.1911111111111115e-06,
+      "loss": 0.5945,
+      "step": 14050
+    },
+    {
+      "epoch": 40.00342857142857,
+      "grad_norm": 0.0029238651040941477,
+      "learning_rate": 2.184761904761905e-06,
+      "loss": 0.001,
+      "step": 14060
+    },
+    {
+      "epoch": 40.004,
+      "grad_norm": 0.5616852045059204,
+      "learning_rate": 2.1784126984126985e-06,
+      "loss": 0.2197,
+      "step": 14070
+    },
+    {
+      "epoch": 40.00457142857143,
+      "grad_norm": 0.005956857465207577,
+      "learning_rate": 2.1720634920634924e-06,
+      "loss": 0.007,
+      "step": 14080
+    },
+    {
+      "epoch": 40.00514285714286,
+      "grad_norm": 0.0020911735482513905,
+      "learning_rate": 2.165714285714286e-06,
+      "loss": 0.2357,
+      "step": 14090
+    },
+    {
+      "epoch": 40.005714285714284,
+      "grad_norm": 0.010295256972312927,
+      "learning_rate": 2.1593650793650793e-06,
+      "loss": 0.2058,
+      "step": 14100
+    },
+    {
+      "epoch": 40.00628571428572,
+      "grad_norm": 390.59674072265625,
+      "learning_rate": 2.1530158730158732e-06,
+      "loss": 0.4373,
+      "step": 14110
+    },
+    {
+      "epoch": 40.00685714285714,
+      "grad_norm": 0.01252991147339344,
+      "learning_rate": 2.1466666666666667e-06,
+      "loss": 0.0004,
+      "step": 14120
+    },
+    {
+      "epoch": 40.00742857142857,
+      "grad_norm": 0.0566883347928524,
+      "learning_rate": 2.1403174603174606e-06,
+      "loss": 0.3251,
+      "step": 14130
+    },
+    {
+      "epoch": 40.008,
+      "grad_norm": 0.0027798376977443695,
+      "learning_rate": 2.133968253968254e-06,
+      "loss": 0.0004,
+      "step": 14140
+    },
+    {
+      "epoch": 40.00857142857143,
+      "grad_norm": 0.049019478261470795,
+      "learning_rate": 2.1276190476190476e-06,
+      "loss": 0.0005,
+      "step": 14150
+    },
+    {
+      "epoch": 40.009142857142855,
+      "grad_norm": 0.25185105204582214,
+      "learning_rate": 2.1212698412698415e-06,
+      "loss": 0.0016,
+      "step": 14160
+    },
+    {
+      "epoch": 40.00971428571429,
+      "grad_norm": 0.05718778818845749,
+      "learning_rate": 2.114920634920635e-06,
+      "loss": 0.2511,
+      "step": 14170
+    },
+    {
+      "epoch": 40.010285714285715,
+      "grad_norm": 0.017743902280926704,
+      "learning_rate": 2.108571428571429e-06,
+      "loss": 0.2719,
+      "step": 14180
+    },
+    {
+      "epoch": 40.01085714285714,
+      "grad_norm": 0.003888359060510993,
+      "learning_rate": 2.1022222222222224e-06,
+      "loss": 0.001,
+      "step": 14190
+    },
+    {
+      "epoch": 40.011428571428574,
+      "grad_norm": 0.1548689901828766,
+      "learning_rate": 2.0958730158730163e-06,
+      "loss": 0.2496,
+      "step": 14200
+    },
+    {
+      "epoch": 40.012,
+      "grad_norm": 0.14938634634017944,
+      "learning_rate": 2.0895238095238098e-06,
+      "loss": 0.1626,
+      "step": 14210
+    },
+    {
+      "epoch": 40.01257142857143,
+      "grad_norm": 0.0049010817892849445,
+      "learning_rate": 2.0831746031746032e-06,
+      "loss": 0.0009,
+      "step": 14220
+    },
+    {
+      "epoch": 40.01314285714286,
+      "grad_norm": 0.06158607453107834,
+      "learning_rate": 2.076825396825397e-06,
+      "loss": 0.6294,
+      "step": 14230
+    },
+    {
+      "epoch": 40.013714285714286,
+      "grad_norm": 0.396128386259079,
+      "learning_rate": 2.0704761904761906e-06,
+      "loss": 0.2451,
+      "step": 14240
+    },
+    {
+      "epoch": 40.01428571428571,
+      "grad_norm": 0.7225034236907959,
+      "learning_rate": 2.064126984126984e-06,
+      "loss": 0.1585,
+      "step": 14250
+    },
+    {
+      "epoch": 40.014857142857146,
+      "grad_norm": 28.033405303955078,
+      "learning_rate": 2.057777777777778e-06,
+      "loss": 0.2748,
+      "step": 14260
+    },
+    {
+      "epoch": 40.01542857142857,
+      "grad_norm": 95.52556610107422,
+      "learning_rate": 2.0514285714285715e-06,
+      "loss": 0.1725,
+      "step": 14270
+    },
+    {
+      "epoch": 40.016,
+      "grad_norm": 75.94400787353516,
+      "learning_rate": 2.045079365079365e-06,
+      "loss": 0.158,
+      "step": 14280
+    },
+    {
+      "epoch": 40.01657142857143,
+      "grad_norm": 0.08160189539194107,
+      "learning_rate": 2.038730158730159e-06,
+      "loss": 0.0047,
+      "step": 14290
+    },
+    {
+      "epoch": 40.01714285714286,
+      "grad_norm": 0.04288149252533913,
+      "learning_rate": 2.032380952380953e-06,
+      "loss": 0.002,
+      "step": 14300
+    },
+    {
+      "epoch": 40.017714285714284,
+      "grad_norm": 0.013937451876699924,
+      "learning_rate": 2.026031746031746e-06,
+      "loss": 0.0041,
+      "step": 14310
+    },
+    {
+      "epoch": 40.01828571428572,
+      "grad_norm": 91.30559539794922,
+      "learning_rate": 2.0196825396825398e-06,
+      "loss": 0.2636,
+      "step": 14320
+    },
+    {
+      "epoch": 40.018857142857144,
+      "grad_norm": 0.051494400948286057,
+      "learning_rate": 2.0133333333333337e-06,
+      "loss": 0.0489,
+      "step": 14330
+    },
+    {
+      "epoch": 40.01942857142857,
+      "grad_norm": 0.0007844906649552286,
+      "learning_rate": 2.006984126984127e-06,
+      "loss": 0.6524,
+      "step": 14340
+    },
+    {
+      "epoch": 40.02,
+      "grad_norm": 19.62409210205078,
+      "learning_rate": 2.0006349206349206e-06,
+      "loss": 0.4637,
+      "step": 14350
+    },
+    {
+      "epoch": 40.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 1.0104492902755737,
+      "eval_runtime": 126.2637,
+      "eval_samples_per_second": 2.352,
+      "eval_steps_per_second": 1.18,
+      "step": 14350
+    },
+    {
+      "epoch": 41.000571428571426,
+      "grad_norm": 0.10992828756570816,
+      "learning_rate": 1.9942857142857146e-06,
+      "loss": 0.1912,
+      "step": 14360
+    },
+    {
+      "epoch": 41.00114285714286,
+      "grad_norm": 7.585694313049316,
+      "learning_rate": 1.987936507936508e-06,
+      "loss": 0.0027,
+      "step": 14370
+    },
+    {
+      "epoch": 41.001714285714286,
+      "grad_norm": 0.027735279873013496,
+      "learning_rate": 1.9815873015873015e-06,
+      "loss": 0.3344,
+      "step": 14380
+    },
+    {
+      "epoch": 41.00228571428571,
+      "grad_norm": 0.02809802256524563,
+      "learning_rate": 1.9752380952380954e-06,
+      "loss": 0.0989,
+      "step": 14390
+    },
+    {
+      "epoch": 41.002857142857145,
+      "grad_norm": 0.0027615518774837255,
+      "learning_rate": 1.968888888888889e-06,
+      "loss": 0.0007,
+      "step": 14400
+    },
+    {
+      "epoch": 41.00342857142857,
+      "grad_norm": 0.009083566255867481,
+      "learning_rate": 1.962539682539683e-06,
+      "loss": 0.0071,
+      "step": 14410
+    },
+    {
+      "epoch": 41.004,
+      "grad_norm": 43.1031379699707,
+      "learning_rate": 1.9561904761904763e-06,
+      "loss": 0.213,
+      "step": 14420
+    },
+    {
+      "epoch": 41.00457142857143,
+      "grad_norm": 0.06005506217479706,
+      "learning_rate": 1.9498412698412698e-06,
+      "loss": 0.0027,
+      "step": 14430
+    },
+    {
+      "epoch": 41.00514285714286,
+      "grad_norm": 0.10170082747936249,
+      "learning_rate": 1.9434920634920637e-06,
+      "loss": 0.3536,
+      "step": 14440
+    },
+    {
+      "epoch": 41.005714285714284,
+      "grad_norm": 0.0068178740330040455,
+      "learning_rate": 1.9371428571428576e-06,
+      "loss": 0.3189,
+      "step": 14450
+    },
+    {
+      "epoch": 41.00628571428572,
+      "grad_norm": 381.4216003417969,
+      "learning_rate": 1.930793650793651e-06,
+      "loss": 0.1193,
+      "step": 14460
+    },
+    {
+      "epoch": 41.00685714285714,
+      "grad_norm": 0.004170392639935017,
+      "learning_rate": 1.9244444444444446e-06,
+      "loss": 0.258,
+      "step": 14470
+    },
+    {
+      "epoch": 41.00742857142857,
+      "grad_norm": 0.059858277440071106,
+      "learning_rate": 1.9180952380952385e-06,
+      "loss": 0.2108,
+      "step": 14480
+    },
+    {
+      "epoch": 41.008,
+      "grad_norm": 0.0008723547798581421,
+      "learning_rate": 1.911746031746032e-06,
+      "loss": 0.2695,
+      "step": 14490
+    },
+    {
+      "epoch": 41.00857142857143,
+      "grad_norm": 44.76935958862305,
+      "learning_rate": 1.9053968253968254e-06,
+      "loss": 0.7032,
+      "step": 14500
+    },
+    {
+      "epoch": 41.009142857142855,
+      "grad_norm": 0.011419898830354214,
+      "learning_rate": 1.8990476190476193e-06,
+      "loss": 0.0005,
+      "step": 14510
+    },
+    {
+      "epoch": 41.00971428571429,
+      "grad_norm": 0.001278414623811841,
+      "learning_rate": 1.892698412698413e-06,
+      "loss": 0.0007,
+      "step": 14520
+    },
+    {
+      "epoch": 41.010285714285715,
+      "grad_norm": 0.011867762543261051,
+      "learning_rate": 1.8863492063492065e-06,
+      "loss": 0.0005,
+      "step": 14530
+    },
+    {
+      "epoch": 41.01085714285714,
+      "grad_norm": 0.014038086868822575,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.6461,
+      "step": 14540
+    },
+    {
+      "epoch": 41.011428571428574,
+      "grad_norm": 32.356666564941406,
+      "learning_rate": 1.873650793650794e-06,
+      "loss": 0.5288,
+      "step": 14550
+    },
+    {
+      "epoch": 41.012,
+      "grad_norm": 0.18130460381507874,
+      "learning_rate": 1.8673015873015874e-06,
+      "loss": 0.7912,
+      "step": 14560
+    },
+    {
+      "epoch": 41.01257142857143,
+      "grad_norm": 0.2205055207014084,
+      "learning_rate": 1.860952380952381e-06,
+      "loss": 0.0011,
+      "step": 14570
+    },
+    {
+      "epoch": 41.01314285714286,
+      "grad_norm": 0.002843776484951377,
+      "learning_rate": 1.8546031746031748e-06,
+      "loss": 0.0006,
+      "step": 14580
+    },
+    {
+      "epoch": 41.013714285714286,
+      "grad_norm": 0.0019020799081772566,
+      "learning_rate": 1.8482539682539685e-06,
+      "loss": 0.0116,
+      "step": 14590
+    },
+    {
+      "epoch": 41.01428571428571,
+      "grad_norm": 0.12416958808898926,
+      "learning_rate": 1.841904761904762e-06,
+      "loss": 0.2979,
+      "step": 14600
+    },
+    {
+      "epoch": 41.014857142857146,
+      "grad_norm": 0.0510811023414135,
+      "learning_rate": 1.8355555555555557e-06,
+      "loss": 0.2946,
+      "step": 14610
+    },
+    {
+      "epoch": 41.01542857142857,
+      "grad_norm": 0.04932905361056328,
+      "learning_rate": 1.8292063492063493e-06,
+      "loss": 0.557,
+      "step": 14620
+    },
+    {
+      "epoch": 41.016,
+      "grad_norm": 6.153399467468262,
+      "learning_rate": 1.8228571428571428e-06,
+      "loss": 0.2424,
+      "step": 14630
+    },
+    {
+      "epoch": 41.01657142857143,
+      "grad_norm": 0.1840139478445053,
+      "learning_rate": 1.8165079365079365e-06,
+      "loss": 0.3622,
+      "step": 14640
+    },
+    {
+      "epoch": 41.01714285714286,
+      "grad_norm": 0.05411524698138237,
+      "learning_rate": 1.8101587301587304e-06,
+      "loss": 0.2168,
+      "step": 14650
+    },
+    {
+      "epoch": 41.017714285714284,
+      "grad_norm": 0.02711281180381775,
+      "learning_rate": 1.8038095238095241e-06,
+      "loss": 0.3268,
+      "step": 14660
+    },
+    {
+      "epoch": 41.01828571428572,
+      "grad_norm": 0.005898493342101574,
+      "learning_rate": 1.7974603174603176e-06,
+      "loss": 0.003,
+      "step": 14670
+    },
+    {
+      "epoch": 41.018857142857144,
+      "grad_norm": 0.08160807937383652,
+      "learning_rate": 1.7911111111111113e-06,
+      "loss": 0.0009,
+      "step": 14680
+    },
+    {
+      "epoch": 41.01942857142857,
+      "grad_norm": 0.0039462801069021225,
+      "learning_rate": 1.784761904761905e-06,
+      "loss": 0.0006,
+      "step": 14690
+    },
+    {
+      "epoch": 41.02,
+      "grad_norm": 1.8989238739013672,
+      "learning_rate": 1.7784126984126985e-06,
+      "loss": 0.0014,
+      "step": 14700
+    },
+    {
+      "epoch": 41.02,
+      "eval_accuracy": 0.8417508417508418,
+      "eval_loss": 1.043727159500122,
+      "eval_runtime": 126.4211,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 1.179,
+      "step": 14700
+    },
+    {
+      "epoch": 42.000571428571426,
+      "grad_norm": 0.014098647981882095,
+      "learning_rate": 1.7720634920634922e-06,
+      "loss": 0.1607,
+      "step": 14710
+    },
+    {
+      "epoch": 42.00114285714286,
+      "grad_norm": 20.41655158996582,
+      "learning_rate": 1.7657142857142859e-06,
+      "loss": 0.7527,
+      "step": 14720
+    },
+    {
+      "epoch": 42.001714285714286,
+      "grad_norm": 0.023175185546278954,
+      "learning_rate": 1.7593650793650796e-06,
+      "loss": 0.2158,
+      "step": 14730
+    },
+    {
+      "epoch": 42.00228571428571,
+      "grad_norm": 0.013949972577393055,
+      "learning_rate": 1.753015873015873e-06,
+      "loss": 0.0014,
+      "step": 14740
+    },
+    {
+      "epoch": 42.002857142857145,
+      "grad_norm": 78.23534393310547,
+      "learning_rate": 1.7466666666666667e-06,
+      "loss": 0.2982,
+      "step": 14750
+    },
+    {
+      "epoch": 42.00342857142857,
+      "grad_norm": 0.06348254531621933,
+      "learning_rate": 1.7403174603174604e-06,
+      "loss": 0.3159,
+      "step": 14760
+    },
+    {
+      "epoch": 42.004,
+      "grad_norm": 0.0164299625903368,
+      "learning_rate": 1.7339682539682543e-06,
+      "loss": 0.0015,
+      "step": 14770
+    },
+    {
+      "epoch": 42.00457142857143,
+      "grad_norm": 0.0010301744332537055,
+      "learning_rate": 1.7276190476190476e-06,
+      "loss": 0.0005,
+      "step": 14780
+    },
+    {
+      "epoch": 42.00514285714286,
+      "grad_norm": 0.11905546486377716,
+      "learning_rate": 1.7212698412698415e-06,
+      "loss": 0.0194,
+      "step": 14790
+    },
+    {
+      "epoch": 42.005714285714284,
+      "grad_norm": 0.05502673611044884,
+      "learning_rate": 1.7149206349206352e-06,
+      "loss": 0.0014,
+      "step": 14800
+    },
+    {
+      "epoch": 42.00628571428572,
+      "grad_norm": 0.005338889546692371,
+      "learning_rate": 1.7085714285714287e-06,
+      "loss": 0.0021,
+      "step": 14810
+    },
+    {
+      "epoch": 42.00685714285714,
+      "grad_norm": 0.021363425999879837,
+      "learning_rate": 1.7022222222222224e-06,
+      "loss": 0.2946,
+      "step": 14820
+    },
+    {
+      "epoch": 42.00742857142857,
+      "grad_norm": 0.0004772421089001,
+      "learning_rate": 1.695873015873016e-06,
+      "loss": 0.3743,
+      "step": 14830
+    },
+    {
+      "epoch": 42.008,
+      "grad_norm": 0.08759511262178421,
+      "learning_rate": 1.6895238095238098e-06,
+      "loss": 0.3515,
+      "step": 14840
+    },
+    {
+      "epoch": 42.00857142857143,
+      "grad_norm": 0.03472195193171501,
+      "learning_rate": 1.6831746031746033e-06,
+      "loss": 0.0011,
+      "step": 14850
+    },
+    {
+      "epoch": 42.009142857142855,
+      "grad_norm": 0.011670676060020924,
+      "learning_rate": 1.676825396825397e-06,
+      "loss": 0.3558,
+      "step": 14860
+    },
+    {
+      "epoch": 42.00971428571429,
+      "grad_norm": 0.013489765115082264,
+      "learning_rate": 1.6704761904761907e-06,
+      "loss": 0.0845,
+      "step": 14870
+    },
+    {
+      "epoch": 42.010285714285715,
+      "grad_norm": 0.005060871131718159,
+      "learning_rate": 1.6641269841269841e-06,
+      "loss": 0.0024,
+      "step": 14880
+    },
+    {
+      "epoch": 42.01085714285714,
+      "grad_norm": 0.08271327614784241,
+      "learning_rate": 1.6577777777777778e-06,
+      "loss": 0.3332,
+      "step": 14890
+    },
+    {
+      "epoch": 42.011428571428574,
+      "grad_norm": 0.0020348222460597754,
+      "learning_rate": 1.6514285714285715e-06,
+      "loss": 0.9036,
+      "step": 14900
+    },
+    {
+      "epoch": 42.012,
+      "grad_norm": 0.020888779312372208,
+      "learning_rate": 1.6450793650793654e-06,
+      "loss": 0.0897,
+      "step": 14910
+    },
+    {
+      "epoch": 42.01257142857143,
+      "grad_norm": 0.030984889715909958,
+      "learning_rate": 1.6387301587301587e-06,
+      "loss": 0.5513,
+      "step": 14920
+    },
+    {
+      "epoch": 42.01314285714286,
+      "grad_norm": 90.73495483398438,
+      "learning_rate": 1.6323809523809526e-06,
+      "loss": 0.3941,
+      "step": 14930
+    },
+    {
+      "epoch": 42.013714285714286,
+      "grad_norm": 0.013540121726691723,
+      "learning_rate": 1.6260317460317463e-06,
+      "loss": 0.0006,
+      "step": 14940
+    },
+    {
+      "epoch": 42.01428571428571,
+      "grad_norm": 0.0985909178853035,
+      "learning_rate": 1.6196825396825398e-06,
+      "loss": 0.3027,
+      "step": 14950
+    },
+    {
+      "epoch": 42.014857142857146,
+      "grad_norm": 0.001703809481114149,
+      "learning_rate": 1.6133333333333335e-06,
+      "loss": 0.3457,
+      "step": 14960
+    },
+    {
+      "epoch": 42.01542857142857,
+      "grad_norm": 0.005111075472086668,
+      "learning_rate": 1.6069841269841272e-06,
+      "loss": 0.2664,
+      "step": 14970
+    },
+    {
+      "epoch": 42.016,
+      "grad_norm": 0.0010470590787008405,
+      "learning_rate": 1.6006349206349209e-06,
+      "loss": 0.0113,
+      "step": 14980
+    },
+    {
+      "epoch": 42.01657142857143,
+      "grad_norm": 0.08761300891637802,
+      "learning_rate": 1.5942857142857144e-06,
+      "loss": 0.327,
+      "step": 14990
+    },
+    {
+      "epoch": 42.01714285714286,
+      "grad_norm": 0.0029850888531655073,
+      "learning_rate": 1.587936507936508e-06,
+      "loss": 0.0013,
+      "step": 15000
+    },
+    {
+      "epoch": 42.017714285714284,
+      "grad_norm": 0.006845708005130291,
+      "learning_rate": 1.5815873015873017e-06,
+      "loss": 0.3322,
+      "step": 15010
+    },
+    {
+      "epoch": 42.01828571428572,
+      "grad_norm": 0.036936238408088684,
+      "learning_rate": 1.5752380952380952e-06,
+      "loss": 0.0012,
+      "step": 15020
+    },
+    {
+      "epoch": 42.018857142857144,
+      "grad_norm": 0.026531461626291275,
+      "learning_rate": 1.568888888888889e-06,
+      "loss": 0.6484,
+      "step": 15030
+    },
+    {
+      "epoch": 42.01942857142857,
+      "grad_norm": 0.04399362578988075,
+      "learning_rate": 1.5625396825396826e-06,
+      "loss": 0.4781,
+      "step": 15040
+    },
+    {
+      "epoch": 42.02,
+      "grad_norm": 0.03520611673593521,
+      "learning_rate": 1.5561904761904763e-06,
+      "loss": 0.6253,
+      "step": 15050
+    },
+    {
+      "epoch": 42.02,
+      "eval_accuracy": 0.8148148148148148,
+      "eval_loss": 1.1686880588531494,
+      "eval_runtime": 126.3263,
+      "eval_samples_per_second": 2.351,
+      "eval_steps_per_second": 1.179,
+      "step": 15050
+    },
+    {
+      "epoch": 43.000571428571426,
+      "grad_norm": 0.002064335159957409,
+      "learning_rate": 1.5498412698412698e-06,
+      "loss": 0.0016,
+      "step": 15060
+    },
+    {
+      "epoch": 43.00114285714286,
+      "grad_norm": 833.5392456054688,
+      "learning_rate": 1.5434920634920635e-06,
+      "loss": 0.1628,
+      "step": 15070
+    },
+    {
+      "epoch": 43.001714285714286,
+      "grad_norm": 0.02547420747578144,
+      "learning_rate": 1.5371428571428574e-06,
+      "loss": 0.4407,
+      "step": 15080
+    },
+    {
+      "epoch": 43.00228571428571,
+      "grad_norm": 0.0005154515383765101,
+      "learning_rate": 1.530793650793651e-06,
+      "loss": 0.0059,
+      "step": 15090
+    },
+    {
+      "epoch": 43.002857142857145,
+      "grad_norm": 0.019299926236271858,
+      "learning_rate": 1.5244444444444446e-06,
+      "loss": 0.1375,
+      "step": 15100
+    },
+    {
+      "epoch": 43.00342857142857,
+      "grad_norm": 0.04400021582841873,
+      "learning_rate": 1.5180952380952383e-06,
+      "loss": 0.1776,
+      "step": 15110
+    },
+    {
+      "epoch": 43.004,
+      "grad_norm": 0.37237548828125,
+      "learning_rate": 1.511746031746032e-06,
+      "loss": 0.0009,
+      "step": 15120
+    },
+    {
+      "epoch": 43.00457142857143,
+      "grad_norm": 27.535037994384766,
+      "learning_rate": 1.5053968253968255e-06,
+      "loss": 0.302,
+      "step": 15130
+    },
+    {
+      "epoch": 43.00514285714286,
+      "grad_norm": 0.016599314287304878,
+      "learning_rate": 1.4990476190476191e-06,
+      "loss": 0.1959,
+      "step": 15140
+    },
+    {
+      "epoch": 43.005714285714284,
+      "grad_norm": 0.002258901484310627,
+      "learning_rate": 1.4926984126984128e-06,
+      "loss": 0.011,
+      "step": 15150
+    },
+    {
+      "epoch": 43.00628571428572,
+      "grad_norm": 21.613733291625977,
+      "learning_rate": 1.4863492063492065e-06,
+      "loss": 0.2254,
+      "step": 15160
+    },
+    {
+      "epoch": 43.00685714285714,
+      "grad_norm": 0.004936868790537119,
+      "learning_rate": 1.48e-06,
+      "loss": 0.0045,
+      "step": 15170
+    },
+    {
+      "epoch": 43.00742857142857,
+      "grad_norm": 0.09461677819490433,
+      "learning_rate": 1.4736507936507937e-06,
+      "loss": 0.3332,
+      "step": 15180
+    },
+    {
+      "epoch": 43.008,
+      "grad_norm": 0.001690115430392325,
+      "learning_rate": 1.4673015873015874e-06,
+      "loss": 0.6643,
+      "step": 15190
+    },
+    {
+      "epoch": 43.00857142857143,
+      "grad_norm": 0.018435359001159668,
+      "learning_rate": 1.4609523809523809e-06,
+      "loss": 0.001,
+      "step": 15200
+    },
+    {
+      "epoch": 43.009142857142855,
+      "grad_norm": 30.231048583984375,
+      "learning_rate": 1.4546031746031746e-06,
+      "loss": 0.2802,
+      "step": 15210
+    },
+    {
+      "epoch": 43.00971428571429,
+      "grad_norm": 0.05070001631975174,
+      "learning_rate": 1.4482539682539685e-06,
+      "loss": 0.001,
+      "step": 15220
+    },
+    {
+      "epoch": 43.010285714285715,
+      "grad_norm": 0.026784028857946396,
+      "learning_rate": 1.4419047619047622e-06,
+      "loss": 0.4937,
+      "step": 15230
+    },
+    {
+      "epoch": 43.01085714285714,
+      "grad_norm": 0.002373154740780592,
+      "learning_rate": 1.4355555555555557e-06,
+      "loss": 0.0013,
+      "step": 15240
+    },
+    {
+      "epoch": 43.011428571428574,
+      "grad_norm": 0.09937410801649094,
+      "learning_rate": 1.4292063492063494e-06,
+      "loss": 0.2796,
+      "step": 15250
+    },
+    {
+      "epoch": 43.012,
+      "grad_norm": 0.026784038171172142,
+      "learning_rate": 1.422857142857143e-06,
+      "loss": 0.4131,
+      "step": 15260
+    },
+    {
+      "epoch": 43.01257142857143,
+      "grad_norm": 0.0008246685029007494,
+      "learning_rate": 1.4165079365079365e-06,
+      "loss": 0.0036,
+      "step": 15270
+    },
+    {
+      "epoch": 43.01314285714286,
+      "grad_norm": 0.013181746006011963,
+      "learning_rate": 1.4101587301587302e-06,
+      "loss": 0.0016,
+      "step": 15280
+    },
+    {
+      "epoch": 43.013714285714286,
+      "grad_norm": 0.23996856808662415,
+      "learning_rate": 1.403809523809524e-06,
+      "loss": 0.2515,
+      "step": 15290
+    },
+    {
+      "epoch": 43.01428571428571,
+      "grad_norm": 0.08559610694646835,
+      "learning_rate": 1.3974603174603176e-06,
+      "loss": 0.6153,
+      "step": 15300
+    },
+    {
+      "epoch": 43.014857142857146,
+      "grad_norm": 0.03630613163113594,
+      "learning_rate": 1.3911111111111111e-06,
+      "loss": 0.2843,
+      "step": 15310
+    },
+    {
+      "epoch": 43.01542857142857,
+      "grad_norm": 0.0006657622870989144,
+      "learning_rate": 1.3847619047619048e-06,
+      "loss": 0.0004,
+      "step": 15320
+    },
+    {
+      "epoch": 43.016,
+      "grad_norm": 0.3452169895172119,
+      "learning_rate": 1.3784126984126985e-06,
+      "loss": 0.0034,
+      "step": 15330
+    },
+    {
+      "epoch": 43.01657142857143,
+      "grad_norm": 2.725701332092285,
+      "learning_rate": 1.372063492063492e-06,
+      "loss": 0.1951,
+      "step": 15340
+    },
+    {
+      "epoch": 43.01714285714286,
+      "grad_norm": 0.00891941599547863,
+      "learning_rate": 1.3657142857142857e-06,
+      "loss": 0.2923,
+      "step": 15350
+    },
+    {
+      "epoch": 43.017714285714284,
+      "grad_norm": 0.0006919241859577596,
+      "learning_rate": 1.3593650793650796e-06,
+      "loss": 0.5058,
+      "step": 15360
+    },
+    {
+      "epoch": 43.01828571428572,
+      "grad_norm": 0.04941738769412041,
+      "learning_rate": 1.3530158730158733e-06,
+      "loss": 0.0011,
+      "step": 15370
+    },
+    {
+      "epoch": 43.018857142857144,
+      "grad_norm": 0.11309830099344254,
+      "learning_rate": 1.3466666666666668e-06,
+      "loss": 0.0771,
+      "step": 15380
+    },
+    {
+      "epoch": 43.01942857142857,
+      "grad_norm": 0.019192902371287346,
+      "learning_rate": 1.3403174603174605e-06,
+      "loss": 0.4352,
+      "step": 15390
+    },
+    {
+      "epoch": 43.02,
+      "grad_norm": 0.0010145236738026142,
+      "learning_rate": 1.3339682539682542e-06,
+      "loss": 0.0009,
+      "step": 15400
+    },
+    {
+      "epoch": 43.02,
+      "eval_accuracy": 0.8417508417508418,
+      "eval_loss": 1.0243438482284546,
+      "eval_runtime": 126.4767,
+      "eval_samples_per_second": 2.348,
+      "eval_steps_per_second": 1.178,
+      "step": 15400
+    },
+    {
+      "epoch": 44.000571428571426,
+      "grad_norm": 0.020503785461187363,
+      "learning_rate": 1.3276190476190478e-06,
+      "loss": 0.2084,
+      "step": 15410
+    },
+    {
+      "epoch": 44.00114285714286,
+      "grad_norm": 0.03143817558884621,
+      "learning_rate": 1.3212698412698413e-06,
+      "loss": 0.0007,
+      "step": 15420
+    },
+    {
+      "epoch": 44.001714285714286,
+      "grad_norm": 0.04019314423203468,
+      "learning_rate": 1.314920634920635e-06,
+      "loss": 0.0006,
+      "step": 15430
+    },
+    {
+      "epoch": 44.00228571428571,
+      "grad_norm": 0.050954557955265045,
+      "learning_rate": 1.3085714285714287e-06,
+      "loss": 0.1784,
+      "step": 15440
+    },
+    {
+      "epoch": 44.002857142857145,
+      "grad_norm": 0.03376142680644989,
+      "learning_rate": 1.3022222222222222e-06,
+      "loss": 0.001,
+      "step": 15450
+    },
+    {
+      "epoch": 44.00342857142857,
+      "grad_norm": 0.012583643198013306,
+      "learning_rate": 1.295873015873016e-06,
+      "loss": 0.0007,
+      "step": 15460
+    },
+    {
+      "epoch": 44.004,
+      "grad_norm": 0.03293371573090553,
+      "learning_rate": 1.2895238095238096e-06,
+      "loss": 0.0009,
+      "step": 15470
+    },
+    {
+      "epoch": 44.00457142857143,
+      "grad_norm": 0.007353159133344889,
+      "learning_rate": 1.2831746031746035e-06,
+      "loss": 0.2371,
+      "step": 15480
+    },
+    {
+      "epoch": 44.00514285714286,
+      "grad_norm": 0.1071564108133316,
+      "learning_rate": 1.2768253968253968e-06,
+      "loss": 0.001,
+      "step": 15490
+    },
+    {
+      "epoch": 44.005714285714284,
+      "grad_norm": 0.03245130181312561,
+      "learning_rate": 1.2704761904761907e-06,
+      "loss": 0.0003,
+      "step": 15500
+    },
+    {
+      "epoch": 44.00628571428572,
+      "grad_norm": 0.0014682890614494681,
+      "learning_rate": 1.2641269841269844e-06,
+      "loss": 0.0009,
+      "step": 15510
+    },
+    {
+      "epoch": 44.00685714285714,
+      "grad_norm": 51.87623596191406,
+      "learning_rate": 1.2577777777777779e-06,
+      "loss": 0.1958,
+      "step": 15520
+    },
+    {
+      "epoch": 44.00742857142857,
+      "grad_norm": 0.010897007770836353,
+      "learning_rate": 1.2514285714285715e-06,
+      "loss": 0.0008,
+      "step": 15530
+    },
+    {
+      "epoch": 44.008,
+      "grad_norm": 0.36192965507507324,
+      "learning_rate": 1.2450793650793652e-06,
+      "loss": 0.7262,
+      "step": 15540
+    },
+    {
+      "epoch": 44.00857142857143,
+      "grad_norm": 0.1297987550497055,
+      "learning_rate": 1.2387301587301587e-06,
+      "loss": 0.0005,
+      "step": 15550
+    },
+    {
+      "epoch": 44.009142857142855,
+      "grad_norm": 0.002931043738499284,
+      "learning_rate": 1.2323809523809526e-06,
+      "loss": 0.0008,
+      "step": 15560
+    },
+    {
+      "epoch": 44.00971428571429,
+      "grad_norm": 0.016998382285237312,
+      "learning_rate": 1.2260317460317461e-06,
+      "loss": 0.0034,
+      "step": 15570
+    },
+    {
+      "epoch": 44.010285714285715,
+      "grad_norm": 0.013441096059978008,
+      "learning_rate": 1.2196825396825398e-06,
+      "loss": 0.0003,
+      "step": 15580
+    },
+    {
+      "epoch": 44.01085714285714,
+      "grad_norm": 0.0027524903416633606,
+      "learning_rate": 1.2133333333333335e-06,
+      "loss": 0.001,
+      "step": 15590
+    },
+    {
+      "epoch": 44.011428571428574,
+      "grad_norm": 0.028492752462625504,
+      "learning_rate": 1.206984126984127e-06,
+      "loss": 0.0005,
+      "step": 15600
+    },
+    {
+      "epoch": 44.012,
+      "grad_norm": 0.011916798539459705,
+      "learning_rate": 1.2006349206349207e-06,
+      "loss": 0.342,
+      "step": 15610
+    },
+    {
+      "epoch": 44.01257142857143,
+      "grad_norm": 0.0014567896723747253,
+      "learning_rate": 1.1942857142857144e-06,
+      "loss": 0.0004,
+      "step": 15620
+    },
+    {
+      "epoch": 44.01314285714286,
+      "grad_norm": 0.029987970367074013,
+      "learning_rate": 1.187936507936508e-06,
+      "loss": 0.0193,
+      "step": 15630
+    },
+    {
+      "epoch": 44.013714285714286,
+      "grad_norm": 0.08593633025884628,
+      "learning_rate": 1.1815873015873018e-06,
+      "loss": 0.0006,
+      "step": 15640
+    },
+    {
+      "epoch": 44.01428571428571,
+      "grad_norm": 0.06858720630407333,
+      "learning_rate": 1.1752380952380955e-06,
+      "loss": 0.0006,
+      "step": 15650
+    },
+    {
+      "epoch": 44.014857142857146,
+      "grad_norm": 0.16518321633338928,
+      "learning_rate": 1.168888888888889e-06,
+      "loss": 0.0037,
+      "step": 15660
+    },
+    {
+      "epoch": 44.01542857142857,
+      "grad_norm": 0.006396264769136906,
+      "learning_rate": 1.1625396825396826e-06,
+      "loss": 0.288,
+      "step": 15670
+    },
+    {
+      "epoch": 44.016,
+      "grad_norm": 0.018989071249961853,
+      "learning_rate": 1.1561904761904763e-06,
+      "loss": 0.2859,
+      "step": 15680
+    },
+    {
+      "epoch": 44.01657142857143,
+      "grad_norm": 0.0026919255033135414,
+      "learning_rate": 1.1498412698412698e-06,
+      "loss": 0.0005,
+      "step": 15690
+    },
+    {
+      "epoch": 44.01714285714286,
+      "grad_norm": 0.07339149713516235,
+      "learning_rate": 1.1434920634920637e-06,
+      "loss": 0.6642,
+      "step": 15700
+    },
+    {
+      "epoch": 44.017714285714284,
+      "grad_norm": 0.07957630604505539,
+      "learning_rate": 1.1371428571428572e-06,
+      "loss": 0.0012,
+      "step": 15710
+    },
+    {
+      "epoch": 44.01828571428572,
+      "grad_norm": 0.0010191805195063353,
+      "learning_rate": 1.130793650793651e-06,
+      "loss": 0.001,
+      "step": 15720
+    },
+    {
+      "epoch": 44.018857142857144,
+      "grad_norm": 0.03606898710131645,
+      "learning_rate": 1.1244444444444446e-06,
+      "loss": 0.8374,
+      "step": 15730
+    },
+    {
+      "epoch": 44.01942857142857,
+      "grad_norm": 0.13875551521778107,
+      "learning_rate": 1.118095238095238e-06,
+      "loss": 0.3241,
+      "step": 15740
+    },
+    {
+      "epoch": 44.02,
+      "grad_norm": 0.023555980995297432,
+      "learning_rate": 1.1117460317460318e-06,
+      "loss": 0.0003,
+      "step": 15750
+    },
+    {
+      "epoch": 44.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 1.0864453315734863,
+      "eval_runtime": 126.4265,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 1.179,
+      "step": 15750
+    },
+    {
+      "epoch": 45.000571428571426,
+      "grad_norm": 0.05806439369916916,
+      "learning_rate": 1.1053968253968255e-06,
+      "loss": 0.0007,
+      "step": 15760
+    },
+    {
+      "epoch": 45.00114285714286,
+      "grad_norm": 0.00057839514920488,
+      "learning_rate": 1.0990476190476192e-06,
+      "loss": 0.1975,
+      "step": 15770
+    },
+    {
+      "epoch": 45.001714285714286,
+      "grad_norm": 23.461442947387695,
+      "learning_rate": 1.0926984126984129e-06,
+      "loss": 0.2195,
+      "step": 15780
+    },
+    {
+      "epoch": 45.00228571428571,
+      "grad_norm": 0.04696512967348099,
+      "learning_rate": 1.0863492063492066e-06,
+      "loss": 0.0007,
+      "step": 15790
+    },
+    {
+      "epoch": 45.002857142857145,
+      "grad_norm": 0.008673385716974735,
+      "learning_rate": 1.08e-06,
+      "loss": 0.4171,
+      "step": 15800
+    },
+    {
+      "epoch": 45.00342857142857,
+      "grad_norm": 0.005190796218812466,
+      "learning_rate": 1.0736507936507937e-06,
+      "loss": 0.2958,
+      "step": 15810
+    },
+    {
+      "epoch": 45.004,
+      "grad_norm": 0.0010960869258269668,
+      "learning_rate": 1.0673015873015874e-06,
+      "loss": 0.0008,
+      "step": 15820
+    },
+    {
+      "epoch": 45.00457142857143,
+      "grad_norm": 0.04663942754268646,
+      "learning_rate": 1.060952380952381e-06,
+      "loss": 0.1735,
+      "step": 15830
+    },
+    {
+      "epoch": 45.00514285714286,
+      "grad_norm": 0.0008961122948676348,
+      "learning_rate": 1.0546031746031748e-06,
+      "loss": 0.2856,
+      "step": 15840
+    },
+    {
+      "epoch": 45.005714285714284,
+      "grad_norm": 0.05738474056124687,
+      "learning_rate": 1.0482539682539683e-06,
+      "loss": 0.0014,
+      "step": 15850
+    },
+    {
+      "epoch": 45.00628571428572,
+      "grad_norm": 0.04210129752755165,
+      "learning_rate": 1.041904761904762e-06,
+      "loss": 0.0007,
+      "step": 15860
+    },
+    {
+      "epoch": 45.00685714285714,
+      "grad_norm": 0.003992922138422728,
+      "learning_rate": 1.0355555555555557e-06,
+      "loss": 0.6006,
+      "step": 15870
+    },
+    {
+      "epoch": 45.00742857142857,
+      "grad_norm": 0.014494026079773903,
+      "learning_rate": 1.0292063492063494e-06,
+      "loss": 0.0008,
+      "step": 15880
+    },
+    {
+      "epoch": 45.008,
+      "grad_norm": 0.09102319180965424,
+      "learning_rate": 1.0228571428571429e-06,
+      "loss": 0.1237,
+      "step": 15890
+    },
+    {
+      "epoch": 45.00857142857143,
+      "grad_norm": 0.02508743479847908,
+      "learning_rate": 1.0165079365079366e-06,
+      "loss": 0.2364,
+      "step": 15900
+    },
+    {
+      "epoch": 45.009142857142855,
+      "grad_norm": 0.002251436933875084,
+      "learning_rate": 1.0101587301587303e-06,
+      "loss": 0.4723,
+      "step": 15910
+    },
+    {
+      "epoch": 45.00971428571429,
+      "grad_norm": 0.007445584516972303,
+      "learning_rate": 1.003809523809524e-06,
+      "loss": 0.2701,
+      "step": 15920
+    },
+    {
+      "epoch": 45.010285714285715,
+      "grad_norm": 0.03148816525936127,
+      "learning_rate": 9.974603174603176e-07,
+      "loss": 0.001,
+      "step": 15930
+    },
+    {
+      "epoch": 45.01085714285714,
+      "grad_norm": 0.0006346903392113745,
+      "learning_rate": 9.911111111111111e-07,
+      "loss": 0.2117,
+      "step": 15940
+    },
+    {
+      "epoch": 45.011428571428574,
+      "grad_norm": 0.0017558577237650752,
+      "learning_rate": 9.847619047619048e-07,
+      "loss": 0.0015,
+      "step": 15950
+    },
+    {
+      "epoch": 45.012,
+      "grad_norm": 0.029404617846012115,
+      "learning_rate": 9.784126984126985e-07,
+      "loss": 0.3177,
+      "step": 15960
+    },
+    {
+      "epoch": 45.01257142857143,
+      "grad_norm": 0.008277475833892822,
+      "learning_rate": 9.720634920634922e-07,
+      "loss": 0.0016,
+      "step": 15970
+    },
+    {
+      "epoch": 45.01314285714286,
+      "grad_norm": 0.002906983019784093,
+      "learning_rate": 9.657142857142857e-07,
+      "loss": 0.0123,
+      "step": 15980
+    },
+    {
+      "epoch": 45.013714285714286,
+      "grad_norm": 0.1758769452571869,
+      "learning_rate": 9.593650793650794e-07,
+      "loss": 0.181,
+      "step": 15990
+    },
+    {
+      "epoch": 45.01428571428571,
+      "grad_norm": 106.2284927368164,
+      "learning_rate": 9.530158730158731e-07,
+      "loss": 0.2344,
+      "step": 16000
+    },
+    {
+      "epoch": 45.014857142857146,
+      "grad_norm": 0.5874964594841003,
+      "learning_rate": 9.466666666666667e-07,
+      "loss": 0.0037,
+      "step": 16010
+    },
+    {
+      "epoch": 45.01542857142857,
+      "grad_norm": 26.586414337158203,
+      "learning_rate": 9.403174603174605e-07,
+      "loss": 0.2546,
+      "step": 16020
+    },
+    {
+      "epoch": 45.016,
+      "grad_norm": 0.08056703954935074,
+      "learning_rate": 9.339682539682541e-07,
+      "loss": 0.3617,
+      "step": 16030
+    },
+    {
+      "epoch": 45.01657142857143,
+      "grad_norm": 0.015912292525172234,
+      "learning_rate": 9.276190476190478e-07,
+      "loss": 0.0014,
+      "step": 16040
+    },
+    {
+      "epoch": 45.01714285714286,
+      "grad_norm": 0.1449512094259262,
+      "learning_rate": 9.212698412698413e-07,
+      "loss": 0.0525,
+      "step": 16050
+    },
+    {
+      "epoch": 45.017714285714284,
+      "grad_norm": 0.06955932825803757,
+      "learning_rate": 9.149206349206349e-07,
+      "loss": 0.0011,
+      "step": 16060
+    },
+    {
+      "epoch": 45.01828571428572,
+      "grad_norm": 0.002916391473263502,
+      "learning_rate": 9.085714285714286e-07,
+      "loss": 0.0006,
+      "step": 16070
+    },
+    {
+      "epoch": 45.018857142857144,
+      "grad_norm": 0.006070619914680719,
+      "learning_rate": 9.022222222222222e-07,
+      "loss": 0.0072,
+      "step": 16080
+    },
+    {
+      "epoch": 45.01942857142857,
+      "grad_norm": 461.6020202636719,
+      "learning_rate": 8.95873015873016e-07,
+      "loss": 0.0531,
+      "step": 16090
+    },
+    {
+      "epoch": 45.02,
+      "grad_norm": 0.0010222607525065541,
+      "learning_rate": 8.895238095238096e-07,
+      "loss": 0.291,
+      "step": 16100
+    },
+    {
+      "epoch": 45.02,
+      "eval_accuracy": 0.8383838383838383,
+      "eval_loss": 1.0647377967834473,
+      "eval_runtime": 126.4165,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 1.179,
+      "step": 16100
+    },
+    {
+      "epoch": 46.000571428571426,
+      "grad_norm": 0.047850172966718674,
+      "learning_rate": 8.831746031746033e-07,
+      "loss": 0.0005,
+      "step": 16110
+    },
+    {
+      "epoch": 46.00114285714286,
+      "grad_norm": 0.11881715804338455,
+      "learning_rate": 8.768253968253969e-07,
+      "loss": 0.0008,
+      "step": 16120
+    },
+    {
+      "epoch": 46.001714285714286,
+      "grad_norm": 1.4675657749176025,
+      "learning_rate": 8.704761904761906e-07,
+      "loss": 0.0016,
+      "step": 16130
+    },
+    {
+      "epoch": 46.00228571428571,
+      "grad_norm": 0.005048373248428106,
+      "learning_rate": 8.641269841269842e-07,
+      "loss": 0.0005,
+      "step": 16140
+    },
+    {
+      "epoch": 46.002857142857145,
+      "grad_norm": 0.008005023933947086,
+      "learning_rate": 8.577777777777778e-07,
+      "loss": 0.6526,
+      "step": 16150
+    },
+    {
+      "epoch": 46.00342857142857,
+      "grad_norm": 0.24868465960025787,
+      "learning_rate": 8.514285714285716e-07,
+      "loss": 0.0015,
+      "step": 16160
+    },
+    {
+      "epoch": 46.004,
+      "grad_norm": 0.000758437963668257,
+      "learning_rate": 8.450793650793652e-07,
+      "loss": 0.2208,
+      "step": 16170
+    },
+    {
+      "epoch": 46.00457142857143,
+      "grad_norm": 0.001206890563480556,
+      "learning_rate": 8.387301587301588e-07,
+      "loss": 0.2555,
+      "step": 16180
+    },
+    {
+      "epoch": 46.00514285714286,
+      "grad_norm": 0.08712995797395706,
+      "learning_rate": 8.323809523809524e-07,
+      "loss": 0.2027,
+      "step": 16190
+    },
+    {
+      "epoch": 46.005714285714284,
+      "grad_norm": 0.021634329110383987,
+      "learning_rate": 8.260317460317461e-07,
+      "loss": 0.0004,
+      "step": 16200
+    },
+    {
+      "epoch": 46.00628571428572,
+      "grad_norm": 0.017216574400663376,
+      "learning_rate": 8.196825396825397e-07,
+      "loss": 0.2594,
+      "step": 16210
+    },
+    {
+      "epoch": 46.00685714285714,
+      "grad_norm": 0.03562033176422119,
+      "learning_rate": 8.133333333333333e-07,
+      "loss": 0.0008,
+      "step": 16220
+    },
+    {
+      "epoch": 46.00742857142857,
+      "grad_norm": 0.06330721825361252,
+      "learning_rate": 8.069841269841271e-07,
+      "loss": 0.3,
+      "step": 16230
+    },
+    {
+      "epoch": 46.008,
+      "grad_norm": 0.006224165204912424,
+      "learning_rate": 8.006349206349207e-07,
+      "loss": 0.0629,
+      "step": 16240
+    },
+    {
+      "epoch": 46.00857142857143,
+      "grad_norm": 34.4676513671875,
+      "learning_rate": 7.942857142857144e-07,
+      "loss": 0.8066,
+      "step": 16250
+    },
+    {
+      "epoch": 46.009142857142855,
+      "grad_norm": 0.00887818168848753,
+      "learning_rate": 7.87936507936508e-07,
+      "loss": 0.0873,
+      "step": 16260
+    },
+    {
+      "epoch": 46.00971428571429,
+      "grad_norm": 0.02603054791688919,
+      "learning_rate": 7.815873015873017e-07,
+      "loss": 0.0004,
+      "step": 16270
+    },
+    {
+      "epoch": 46.010285714285715,
+      "grad_norm": 0.11655969172716141,
+      "learning_rate": 7.752380952380953e-07,
+      "loss": 0.0017,
+      "step": 16280
+    },
+    {
+      "epoch": 46.01085714285714,
+      "grad_norm": 0.09002801775932312,
+      "learning_rate": 7.688888888888891e-07,
+      "loss": 0.2952,
+      "step": 16290
+    },
+    {
+      "epoch": 46.011428571428574,
+      "grad_norm": 0.06367355585098267,
+      "learning_rate": 7.625396825396827e-07,
+      "loss": 0.2261,
+      "step": 16300
+    },
+    {
+      "epoch": 46.012,
+      "grad_norm": 0.002869822084903717,
+      "learning_rate": 7.561904761904762e-07,
+      "loss": 0.004,
+      "step": 16310
+    },
+    {
+      "epoch": 46.01257142857143,
+      "grad_norm": 0.01801781728863716,
+      "learning_rate": 7.498412698412699e-07,
+      "loss": 0.0004,
+      "step": 16320
+    },
+    {
+      "epoch": 46.01314285714286,
+      "grad_norm": 0.034441664814949036,
+      "learning_rate": 7.434920634920635e-07,
+      "loss": 0.3214,
+      "step": 16330
+    },
+    {
+      "epoch": 46.013714285714286,
+      "grad_norm": 0.030220109969377518,
+      "learning_rate": 7.371428571428572e-07,
+      "loss": 0.1798,
+      "step": 16340
+    },
+    {
+      "epoch": 46.01428571428571,
+      "grad_norm": 0.09933434426784515,
+      "learning_rate": 7.307936507936508e-07,
+      "loss": 0.0004,
+      "step": 16350
+    },
+    {
+      "epoch": 46.014857142857146,
+      "grad_norm": 0.06845889985561371,
+      "learning_rate": 7.244444444444446e-07,
+      "loss": 0.3171,
+      "step": 16360
+    },
+    {
+      "epoch": 46.01542857142857,
+      "grad_norm": 0.0007817966397851706,
+      "learning_rate": 7.180952380952382e-07,
+      "loss": 0.3191,
+      "step": 16370
+    },
+    {
+      "epoch": 46.016,
+      "grad_norm": 19.301998138427734,
+      "learning_rate": 7.117460317460318e-07,
+      "loss": 0.4349,
+      "step": 16380
+    },
+    {
+      "epoch": 46.01657142857143,
+      "grad_norm": 0.012864407151937485,
+      "learning_rate": 7.053968253968255e-07,
+      "loss": 0.2711,
+      "step": 16390
+    },
+    {
+      "epoch": 46.01714285714286,
+      "grad_norm": 0.003735810751095414,
+      "learning_rate": 6.990476190476191e-07,
+      "loss": 0.2612,
+      "step": 16400
+    },
+    {
+      "epoch": 46.017714285714284,
+      "grad_norm": 0.7684434652328491,
+      "learning_rate": 6.926984126984128e-07,
+      "loss": 0.2843,
+      "step": 16410
+    },
+    {
+      "epoch": 46.01828571428572,
+      "grad_norm": 0.034190673381090164,
+      "learning_rate": 6.863492063492064e-07,
+      "loss": 0.3447,
+      "step": 16420
+    },
+    {
+      "epoch": 46.018857142857144,
+      "grad_norm": 0.02056187205016613,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.3328,
+      "step": 16430
+    },
+    {
+      "epoch": 46.01942857142857,
+      "grad_norm": 0.0970858708024025,
+      "learning_rate": 6.736507936507936e-07,
+      "loss": 0.753,
+      "step": 16440
+    },
+    {
+      "epoch": 46.02,
+      "grad_norm": 0.03131880611181259,
+      "learning_rate": 6.673015873015874e-07,
+      "loss": 0.4962,
+      "step": 16450
+    },
+    {
+      "epoch": 46.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 1.116576910018921,
+      "eval_runtime": 126.7393,
+      "eval_samples_per_second": 2.343,
+      "eval_steps_per_second": 1.176,
+      "step": 16450
+    },
+    {
+      "epoch": 47.000571428571426,
+      "grad_norm": 0.04706917330622673,
+      "learning_rate": 6.60952380952381e-07,
+      "loss": 0.4978,
+      "step": 16460
+    },
+    {
+      "epoch": 47.00114285714286,
+      "grad_norm": 0.0006357289967127144,
+      "learning_rate": 6.546031746031746e-07,
+      "loss": 0.3942,
+      "step": 16470
+    },
+    {
+      "epoch": 47.001714285714286,
+      "grad_norm": 0.02852693945169449,
+      "learning_rate": 6.482539682539683e-07,
+      "loss": 0.0281,
+      "step": 16480
+    },
+    {
+      "epoch": 47.00228571428571,
+      "grad_norm": 0.01799050346016884,
+      "learning_rate": 6.419047619047619e-07,
+      "loss": 0.3552,
+      "step": 16490
+    },
+    {
+      "epoch": 47.002857142857145,
+      "grad_norm": 0.0006481860764324665,
+      "learning_rate": 6.355555555555556e-07,
+      "loss": 0.001,
+      "step": 16500
+    },
+    {
+      "epoch": 47.00342857142857,
+      "grad_norm": 0.20090606808662415,
+      "learning_rate": 6.292063492063492e-07,
+      "loss": 0.2303,
+      "step": 16510
+    },
+    {
+      "epoch": 47.004,
+      "grad_norm": 0.0003543874772731215,
+      "learning_rate": 6.228571428571429e-07,
+      "loss": 0.1126,
+      "step": 16520
+    },
+    {
+      "epoch": 47.00457142857143,
+      "grad_norm": 0.0005535692907869816,
+      "learning_rate": 6.165079365079366e-07,
+      "loss": 0.0008,
+      "step": 16530
+    },
+    {
+      "epoch": 47.00514285714286,
+      "grad_norm": 0.0009550791000947356,
+      "learning_rate": 6.101587301587302e-07,
+      "loss": 0.0005,
+      "step": 16540
+    },
+    {
+      "epoch": 47.005714285714284,
+      "grad_norm": 0.021146543323993683,
+      "learning_rate": 6.038095238095239e-07,
+      "loss": 0.001,
+      "step": 16550
+    },
+    {
+      "epoch": 47.00628571428572,
+      "grad_norm": 0.007506783120334148,
+      "learning_rate": 5.974603174603176e-07,
+      "loss": 0.3273,
+      "step": 16560
+    },
+    {
+      "epoch": 47.00685714285714,
+      "grad_norm": 0.008311262354254723,
+      "learning_rate": 5.911111111111111e-07,
+      "loss": 0.0786,
+      "step": 16570
+    },
+    {
+      "epoch": 47.00742857142857,
+      "grad_norm": 0.7662389278411865,
+      "learning_rate": 5.847619047619047e-07,
+      "loss": 0.0013,
+      "step": 16580
+    },
+    {
+      "epoch": 47.008,
+      "grad_norm": 0.0042674667201936245,
+      "learning_rate": 5.784126984126984e-07,
+      "loss": 0.0208,
+      "step": 16590
+    },
+    {
+      "epoch": 47.00857142857143,
+      "grad_norm": 0.0019062272040173411,
+      "learning_rate": 5.720634920634921e-07,
+      "loss": 0.3439,
+      "step": 16600
+    },
+    {
+      "epoch": 47.009142857142855,
+      "grad_norm": 0.12979654967784882,
+      "learning_rate": 5.657142857142857e-07,
+      "loss": 0.0246,
+      "step": 16610
+    },
+    {
+      "epoch": 47.00971428571429,
+      "grad_norm": 196.23326110839844,
+      "learning_rate": 5.593650793650794e-07,
+      "loss": 1.4075,
+      "step": 16620
+    },
+    {
+      "epoch": 47.010285714285715,
+      "grad_norm": 0.10944508761167526,
+      "learning_rate": 5.530158730158731e-07,
+      "loss": 0.001,
+      "step": 16630
+    },
+    {
+      "epoch": 47.01085714285714,
+      "grad_norm": 0.17711256444454193,
+      "learning_rate": 5.466666666666667e-07,
+      "loss": 0.0017,
+      "step": 16640
+    },
+    {
+      "epoch": 47.011428571428574,
+      "grad_norm": 0.026470154523849487,
+      "learning_rate": 5.403174603174604e-07,
+      "loss": 0.5992,
+      "step": 16650
+    },
+    {
+      "epoch": 47.012,
+      "grad_norm": 0.04885806888341904,
+      "learning_rate": 5.33968253968254e-07,
+      "loss": 0.0008,
+      "step": 16660
+    },
+    {
+      "epoch": 47.01257142857143,
+      "grad_norm": 0.1748293787240982,
+      "learning_rate": 5.276190476190477e-07,
+      "loss": 0.0011,
+      "step": 16670
+    },
+    {
+      "epoch": 47.01314285714286,
+      "grad_norm": 0.028242330998182297,
+      "learning_rate": 5.212698412698413e-07,
+      "loss": 0.2365,
+      "step": 16680
+    },
+    {
+      "epoch": 47.013714285714286,
+      "grad_norm": 499.1086120605469,
+      "learning_rate": 5.14920634920635e-07,
+      "loss": 0.2298,
+      "step": 16690
+    },
+    {
+      "epoch": 47.01428571428571,
+      "grad_norm": 0.029000846669077873,
+      "learning_rate": 5.085714285714286e-07,
+      "loss": 0.0006,
+      "step": 16700
+    },
+    {
+      "epoch": 47.014857142857146,
+      "grad_norm": 0.018016472458839417,
+      "learning_rate": 5.022222222222222e-07,
+      "loss": 0.1712,
+      "step": 16710
+    },
+    {
+      "epoch": 47.01542857142857,
+      "grad_norm": 0.9705002903938293,
+      "learning_rate": 4.958730158730159e-07,
+      "loss": 0.1886,
+      "step": 16720
+    },
+    {
+      "epoch": 47.016,
+      "grad_norm": 0.011911354027688503,
+      "learning_rate": 4.895238095238096e-07,
+      "loss": 0.0004,
+      "step": 16730
+    },
+    {
+      "epoch": 47.01657142857143,
+      "grad_norm": 0.008706871420145035,
+      "learning_rate": 4.831746031746032e-07,
+      "loss": 0.0008,
+      "step": 16740
+    },
+    {
+      "epoch": 47.01714285714286,
+      "grad_norm": 0.038937151432037354,
+      "learning_rate": 4.7682539682539686e-07,
+      "loss": 0.0004,
+      "step": 16750
+    },
+    {
+      "epoch": 47.017714285714284,
+      "grad_norm": 0.006863133050501347,
+      "learning_rate": 4.704761904761905e-07,
+      "loss": 0.1916,
+      "step": 16760
+    },
+    {
+      "epoch": 47.01828571428572,
+      "grad_norm": 34.81043243408203,
+      "learning_rate": 4.6412698412698414e-07,
+      "loss": 0.1632,
+      "step": 16770
+    },
+    {
+      "epoch": 47.018857142857144,
+      "grad_norm": 0.004815808031708002,
+      "learning_rate": 4.5777777777777784e-07,
+      "loss": 0.3468,
+      "step": 16780
+    },
+    {
+      "epoch": 47.01942857142857,
+      "grad_norm": 0.05895291268825531,
+      "learning_rate": 4.514285714285715e-07,
+      "loss": 0.2842,
+      "step": 16790
+    },
+    {
+      "epoch": 47.02,
+      "grad_norm": 0.0008096261299215257,
+      "learning_rate": 4.450793650793651e-07,
+      "loss": 0.0919,
+      "step": 16800
+    },
+    {
+      "epoch": 47.02,
+      "eval_accuracy": 0.8282828282828283,
+      "eval_loss": 1.1209441423416138,
+      "eval_runtime": 126.6719,
+      "eval_samples_per_second": 2.345,
+      "eval_steps_per_second": 1.176,
+      "step": 16800
+    },
+    {
+      "epoch": 48.000571428571426,
+      "grad_norm": 174.62635803222656,
+      "learning_rate": 4.387301587301588e-07,
+      "loss": 0.008,
+      "step": 16810
+    },
+    {
+      "epoch": 48.00114285714286,
+      "grad_norm": 0.04038200154900551,
+      "learning_rate": 4.323809523809524e-07,
+      "loss": 0.2735,
+      "step": 16820
+    },
+    {
+      "epoch": 48.001714285714286,
+      "grad_norm": 0.012884361669421196,
+      "learning_rate": 4.2603174603174605e-07,
+      "loss": 0.2273,
+      "step": 16830
+    },
+    {
+      "epoch": 48.00228571428571,
+      "grad_norm": 0.11547985672950745,
+      "learning_rate": 4.196825396825397e-07,
+      "loss": 0.2345,
+      "step": 16840
+    },
+    {
+      "epoch": 48.002857142857145,
+      "grad_norm": 0.34670960903167725,
+      "learning_rate": 4.133333333333334e-07,
+      "loss": 0.0018,
+      "step": 16850
+    },
+    {
+      "epoch": 48.00342857142857,
+      "grad_norm": 0.006787777412682772,
+      "learning_rate": 4.06984126984127e-07,
+      "loss": 0.1207,
+      "step": 16860
+    },
+    {
+      "epoch": 48.004,
+      "grad_norm": 0.0013283508596941829,
+      "learning_rate": 4.0063492063492067e-07,
+      "loss": 0.0036,
+      "step": 16870
+    },
+    {
+      "epoch": 48.00457142857143,
+      "grad_norm": 0.0029995145741850138,
+      "learning_rate": 3.9428571428571436e-07,
+      "loss": 0.0003,
+      "step": 16880
+    },
+    {
+      "epoch": 48.00514285714286,
+      "grad_norm": 0.09618420153856277,
+      "learning_rate": 3.87936507936508e-07,
+      "loss": 0.0005,
+      "step": 16890
+    },
+    {
+      "epoch": 48.005714285714284,
+      "grad_norm": 0.033103521913290024,
+      "learning_rate": 3.815873015873016e-07,
+      "loss": 0.0123,
+      "step": 16900
+    },
+    {
+      "epoch": 48.00628571428572,
+      "grad_norm": 0.0023510020691901445,
+      "learning_rate": 3.7523809523809523e-07,
+      "loss": 0.0004,
+      "step": 16910
+    },
+    {
+      "epoch": 48.00685714285714,
+      "grad_norm": 0.02988676354289055,
+      "learning_rate": 3.6888888888888893e-07,
+      "loss": 0.4017,
+      "step": 16920
+    },
+    {
+      "epoch": 48.00742857142857,
+      "grad_norm": 0.014092681929469109,
+      "learning_rate": 3.6253968253968257e-07,
+      "loss": 0.295,
+      "step": 16930
+    },
+    {
+      "epoch": 48.008,
+      "grad_norm": 0.014224675484001637,
+      "learning_rate": 3.561904761904762e-07,
+      "loss": 0.3255,
+      "step": 16940
+    },
+    {
+      "epoch": 48.00857142857143,
+      "grad_norm": 0.09868843853473663,
+      "learning_rate": 3.498412698412699e-07,
+      "loss": 0.2179,
+      "step": 16950
+    },
+    {
+      "epoch": 48.009142857142855,
+      "grad_norm": 0.0006300527020357549,
+      "learning_rate": 3.4349206349206355e-07,
+      "loss": 0.094,
+      "step": 16960
+    },
+    {
+      "epoch": 48.00971428571429,
+      "grad_norm": 0.0006618179613724351,
+      "learning_rate": 3.371428571428572e-07,
+      "loss": 0.0007,
+      "step": 16970
+    },
+    {
+      "epoch": 48.010285714285715,
+      "grad_norm": 0.12803080677986145,
+      "learning_rate": 3.307936507936508e-07,
+      "loss": 0.3201,
+      "step": 16980
+    },
+    {
+      "epoch": 48.01085714285714,
+      "grad_norm": 0.033108729869127274,
+      "learning_rate": 3.2444444444444447e-07,
+      "loss": 0.2339,
+      "step": 16990
+    },
+    {
+      "epoch": 48.011428571428574,
+      "grad_norm": 0.10151008516550064,
+      "learning_rate": 3.180952380952381e-07,
+      "loss": 0.0004,
+      "step": 17000
+    },
+    {
+      "epoch": 48.012,
+      "grad_norm": 0.015769364312291145,
+      "learning_rate": 3.1174603174603176e-07,
+      "loss": 0.3187,
+      "step": 17010
+    },
+    {
+      "epoch": 48.01257142857143,
+      "grad_norm": 0.009230137802660465,
+      "learning_rate": 3.0539682539682545e-07,
+      "loss": 0.4486,
+      "step": 17020
+    },
+    {
+      "epoch": 48.01314285714286,
+      "grad_norm": 0.0004300758882891387,
+      "learning_rate": 2.990476190476191e-07,
+      "loss": 0.049,
+      "step": 17030
+    },
+    {
+      "epoch": 48.013714285714286,
+      "grad_norm": 0.006301034241914749,
+      "learning_rate": 2.9269841269841274e-07,
+      "loss": 0.0007,
+      "step": 17040
+    },
+    {
+      "epoch": 48.01428571428571,
+      "grad_norm": 0.009746396914124489,
+      "learning_rate": 2.863492063492064e-07,
+      "loss": 0.5487,
+      "step": 17050
+    },
+    {
+      "epoch": 48.014857142857146,
+      "grad_norm": 0.009282475337386131,
+      "learning_rate": 2.8e-07,
+      "loss": 0.0008,
+      "step": 17060
+    },
+    {
+      "epoch": 48.01542857142857,
+      "grad_norm": 0.002887505106627941,
+      "learning_rate": 2.736507936507937e-07,
+      "loss": 0.0004,
+      "step": 17070
+    },
+    {
+      "epoch": 48.016,
+      "grad_norm": 0.025261027738451958,
+      "learning_rate": 2.673015873015873e-07,
+      "loss": 0.0006,
+      "step": 17080
+    },
+    {
+      "epoch": 48.01657142857143,
+      "grad_norm": 0.016581503674387932,
+      "learning_rate": 2.6095238095238094e-07,
+      "loss": 0.5268,
+      "step": 17090
+    },
+    {
+      "epoch": 48.01714285714286,
+      "grad_norm": 0.03690231218934059,
+      "learning_rate": 2.5460317460317464e-07,
+      "loss": 0.0583,
+      "step": 17100
+    },
+    {
+      "epoch": 48.017714285714284,
+      "grad_norm": 0.16912665963172913,
+      "learning_rate": 2.482539682539683e-07,
+      "loss": 0.0008,
+      "step": 17110
+    },
+    {
+      "epoch": 48.01828571428572,
+      "grad_norm": 0.01130374800413847,
+      "learning_rate": 2.419047619047619e-07,
+      "loss": 0.0004,
+      "step": 17120
+    },
+    {
+      "epoch": 48.018857142857144,
+      "grad_norm": 0.3978990316390991,
+      "learning_rate": 2.3555555555555556e-07,
+      "loss": 0.001,
+      "step": 17130
+    },
+    {
+      "epoch": 48.01942857142857,
+      "grad_norm": 0.003729065880179405,
+      "learning_rate": 2.2920634920634923e-07,
+      "loss": 0.6757,
+      "step": 17140
+    },
+    {
+      "epoch": 48.02,
+      "grad_norm": 0.0032000578939914703,
+      "learning_rate": 2.228571428571429e-07,
+      "loss": 0.0007,
+      "step": 17150
+    },
+    {
+      "epoch": 48.02,
+      "eval_accuracy": 0.8316498316498316,
+      "eval_loss": 1.1259746551513672,
+      "eval_runtime": 126.1783,
+      "eval_samples_per_second": 2.354,
+      "eval_steps_per_second": 1.181,
+      "step": 17150
+    },
+    {
+      "epoch": 49.000571428571426,
+      "grad_norm": 0.021503791213035583,
+      "learning_rate": 2.1650793650793652e-07,
+      "loss": 0.188,
+      "step": 17160
+    },
+    {
+      "epoch": 49.00114285714286,
+      "grad_norm": 0.009531227871775627,
+      "learning_rate": 2.1015873015873019e-07,
+      "loss": 0.0006,
+      "step": 17170
+    },
+    {
+      "epoch": 49.001714285714286,
+      "grad_norm": 0.02791694551706314,
+      "learning_rate": 2.0380952380952383e-07,
+      "loss": 0.0036,
+      "step": 17180
+    },
+    {
+      "epoch": 49.00228571428571,
+      "grad_norm": 0.014732033014297485,
+      "learning_rate": 1.974603174603175e-07,
+      "loss": 0.4474,
+      "step": 17190
+    },
+    {
+      "epoch": 49.002857142857145,
+      "grad_norm": 0.0033705858513712883,
+      "learning_rate": 1.911111111111111e-07,
+      "loss": 0.001,
+      "step": 17200
+    },
+    {
+      "epoch": 49.00342857142857,
+      "grad_norm": 0.076132632791996,
+      "learning_rate": 1.8476190476190478e-07,
+      "loss": 0.5201,
+      "step": 17210
+    },
+    {
+      "epoch": 49.004,
+      "grad_norm": 0.008379080332815647,
+      "learning_rate": 1.7841269841269842e-07,
+      "loss": 0.0005,
+      "step": 17220
+    },
+    {
+      "epoch": 49.00457142857143,
+      "grad_norm": 0.014781179837882519,
+      "learning_rate": 1.720634920634921e-07,
+      "loss": 0.2197,
+      "step": 17230
+    },
+    {
+      "epoch": 49.00514285714286,
+      "grad_norm": 0.03410143777728081,
+      "learning_rate": 1.657142857142857e-07,
+      "loss": 0.001,
+      "step": 17240
+    },
+    {
+      "epoch": 49.005714285714284,
+      "grad_norm": 0.0006565088406205177,
+      "learning_rate": 1.5936507936507937e-07,
+      "loss": 0.0015,
+      "step": 17250
+    },
+    {
+      "epoch": 49.00628571428572,
+      "grad_norm": 0.014354717917740345,
+      "learning_rate": 1.5301587301587304e-07,
+      "loss": 0.5642,
+      "step": 17260
+    },
+    {
+      "epoch": 49.00685714285714,
+      "grad_norm": 0.09152337163686752,
+      "learning_rate": 1.4666666666666668e-07,
+      "loss": 0.276,
+      "step": 17270
+    },
+    {
+      "epoch": 49.00742857142857,
+      "grad_norm": 0.03440091758966446,
+      "learning_rate": 1.4031746031746032e-07,
+      "loss": 0.0006,
+      "step": 17280
+    },
+    {
+      "epoch": 49.008,
+      "grad_norm": 0.004568720702081919,
+      "learning_rate": 1.3396825396825397e-07,
+      "loss": 0.0073,
+      "step": 17290
+    },
+    {
+      "epoch": 49.00857142857143,
+      "grad_norm": 0.021223975345492363,
+      "learning_rate": 1.2761904761904763e-07,
+      "loss": 0.0004,
+      "step": 17300
+    },
+    {
+      "epoch": 49.009142857142855,
+      "grad_norm": 0.007963555864989758,
+      "learning_rate": 1.2126984126984128e-07,
+      "loss": 0.002,
+      "step": 17310
+    },
+    {
+      "epoch": 49.00971428571429,
+      "grad_norm": 0.0007649322506040335,
+      "learning_rate": 1.1492063492063493e-07,
+      "loss": 0.3511,
+      "step": 17320
+    },
+    {
+      "epoch": 49.010285714285715,
+      "grad_norm": 0.057293448597192764,
+      "learning_rate": 1.0857142857142857e-07,
+      "loss": 0.3396,
+      "step": 17330
+    },
+    {
+      "epoch": 49.01085714285714,
+      "grad_norm": 0.006769323721528053,
+      "learning_rate": 1.0222222222222224e-07,
+      "loss": 0.0022,
+      "step": 17340
+    },
+    {
+      "epoch": 49.011428571428574,
+      "grad_norm": 0.0010334831895306706,
+      "learning_rate": 9.587301587301588e-08,
+      "loss": 0.2595,
+      "step": 17350
+    },
+    {
+      "epoch": 49.012,
+      "grad_norm": 0.015894345939159393,
+      "learning_rate": 8.952380952380954e-08,
+      "loss": 0.2016,
+      "step": 17360
+    },
+    {
+      "epoch": 49.01257142857143,
+      "grad_norm": 0.013702361844480038,
+      "learning_rate": 8.317460317460318e-08,
+      "loss": 0.0008,
+      "step": 17370
+    },
+    {
+      "epoch": 49.01314285714286,
+      "grad_norm": 0.007162360940128565,
+      "learning_rate": 7.682539682539682e-08,
+      "loss": 0.2925,
+      "step": 17380
+    },
+    {
+      "epoch": 49.013714285714286,
+      "grad_norm": 0.009392665699124336,
+      "learning_rate": 7.047619047619048e-08,
+      "loss": 0.007,
+      "step": 17390
+    },
+    {
+      "epoch": 49.01428571428571,
+      "grad_norm": 5.373441219329834,
+      "learning_rate": 6.412698412698413e-08,
+      "loss": 0.2043,
+      "step": 17400
+    },
+    {
+      "epoch": 49.014857142857146,
+      "grad_norm": 0.020008977502584457,
+      "learning_rate": 5.777777777777778e-08,
+      "loss": 0.0742,
+      "step": 17410
+    },
+    {
+      "epoch": 49.01542857142857,
+      "grad_norm": 0.04810080677270889,
+      "learning_rate": 5.142857142857143e-08,
+      "loss": 0.0005,
+      "step": 17420
+    },
+    {
+      "epoch": 49.016,
+      "grad_norm": 0.010244383476674557,
+      "learning_rate": 4.507936507936508e-08,
+      "loss": 0.4077,
+      "step": 17430
+    },
+    {
+      "epoch": 49.01657142857143,
+      "grad_norm": 0.006721619050949812,
+      "learning_rate": 3.873015873015873e-08,
+      "loss": 0.8587,
+      "step": 17440
+    },
+    {
+      "epoch": 49.01714285714286,
+      "grad_norm": 0.0023023684043437243,
+      "learning_rate": 3.238095238095239e-08,
+      "loss": 0.0006,
+      "step": 17450
+    },
+    {
+      "epoch": 49.017714285714284,
+      "grad_norm": 0.19224895536899567,
+      "learning_rate": 2.6031746031746037e-08,
+      "loss": 0.0009,
+      "step": 17460
+    },
+    {
+      "epoch": 49.01828571428572,
+      "grad_norm": 0.04239881783723831,
+      "learning_rate": 1.9682539682539685e-08,
+      "loss": 0.0007,
+      "step": 17470
+    },
+    {
+      "epoch": 49.018857142857144,
+      "grad_norm": 0.0015393183566629887,
+      "learning_rate": 1.3333333333333334e-08,
+      "loss": 0.2342,
+      "step": 17480
+    },
+    {
+      "epoch": 49.01942857142857,
+      "grad_norm": 0.05108032375574112,
+      "learning_rate": 6.9841269841269845e-09,
+      "loss": 0.187,
+      "step": 17490
+    },
+    {
+      "epoch": 49.02,
+      "grad_norm": 0.0034502753987908363,
+      "learning_rate": 6.34920634920635e-10,
+      "loss": 0.0008,
+      "step": 17500
+    },
+    {
+      "epoch": 49.02,
+      "eval_accuracy": 0.835016835016835,
+      "eval_loss": 1.1138967275619507,
+      "eval_runtime": 126.4744,
+      "eval_samples_per_second": 2.348,
+      "eval_steps_per_second": 1.178,
+      "step": 17500
+    },
+    {
+      "epoch": 49.02,
+      "step": 17500,
+      "total_flos": 1.5368592103538688e+20,
+      "train_loss": 0.39184775381689624,
+      "train_runtime": 51190.6894,
+      "train_samples_per_second": 0.684,
+      "train_steps_per_second": 0.342
+    },
+    {
+      "epoch": 49.02,
+      "eval_accuracy": 0.8585858585858586,
+      "eval_loss": 0.7987341284751892,
+      "eval_runtime": 125.2938,
+      "eval_samples_per_second": 2.37,
+      "eval_steps_per_second": 1.189,
+      "step": 17500
+    },
+    {
+      "epoch": 49.02,
+      "eval_accuracy": 0.8585858585858586,
+      "eval_loss": 0.798734188079834,
+      "eval_runtime": 125.2382,
+      "eval_samples_per_second": 2.371,
+      "eval_steps_per_second": 1.19,
+      "step": 17500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 17500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5368592103538688e+20,
+  "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null
 }