diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7411 @@
+{
+  "best_metric": 2.9153051376342773,
+  "best_model_checkpoint": "checkpoints-mistral-300M-FA2-3/checkpoint-12000",
+  "epoch": 7.5649913344887345,
+  "eval_steps": 500,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999998940072441,
+      "loss": 9.6899,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999992462742786,
+      "loss": 8.3056,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029999800969575163,
+      "loss": 7.6302,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999618427651757,
+      "loss": 7.3792,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999377002374244,
+      "loss": 7.2716,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029999076694690364,
+      "loss": 7.1973,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999871750577904,
+      "loss": 7.1332,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999829943705031,
+      "loss": 7.0465,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029997822490145375,
+      "loss": 6.9584,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999728666693657,
+      "loss": 6.8312,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996691969527337,
+      "loss": 6.7483,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996038400252265,
+      "loss": 6.643,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999532596167704,
+      "loss": 6.57,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994554656598456,
+      "loss": 6.4822,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999372448804439,
+      "loss": 6.42,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992835459273796,
+      "loss": 6.3358,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029991887573776695,
+      "loss": 6.269,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990880835274167,
+      "loss": 6.1999,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998981524771831,
+      "loss": 6.133,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998869081529227,
+      "loss": 6.0829,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029987507542410165,
+      "loss": 6.0279,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029986265433717133,
+      "loss": 5.9759,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029984964494089266,
+      "loss": 5.9344,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998360472863359,
+      "loss": 5.8819,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029982186142688103,
+      "loss": 5.8322,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029980708741821673,
+      "loss": 5.7839,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029979172531834064,
+      "loss": 5.7294,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997757751875591,
+      "loss": 5.7153,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029975923708848695,
+      "loss": 5.6566,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299742111086047,
+      "loss": 5.6366,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002997243972474701,
+      "loss": 5.5897,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002997060956422947,
+      "loss": 5.5639,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996872063423667,
+      "loss": 5.5376,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996677294218388,
+      "loss": 5.5143,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029964766495717087,
+      "loss": 5.487,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000299627013027129,
+      "loss": 5.4502,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996057737127858,
+      "loss": 5.4142,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029958394709751925,
+      "loss": 5.3994,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995615332670133,
+      "loss": 5.3587,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029953853230925697,
+      "loss": 5.3274,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000299514944314544,
+      "loss": 5.3148,
+      "step": 410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029949076937547297,
+      "loss": 5.2685,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029946600758694623,
+      "loss": 5.2593,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029944065904617006,
+      "loss": 5.2311,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029941472385265427,
+      "loss": 5.2198,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029938820210821134,
+      "loss": 5.1929,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993610939169567,
+      "loss": 5.1693,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029933339938530775,
+      "loss": 5.1466,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002993051186219837,
+      "loss": 5.1009,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029927625173800513,
+      "loss": 5.1178,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 5.128925323486328,
+      "eval_runtime": 362.8452,
+      "eval_samples_per_second": 209.828,
+      "eval_steps_per_second": 3.748,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029924679884669347,
+      "loss": 5.0697,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002992167600636708,
+      "loss": 5.0626,
+      "step": 520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002991861355068589,
+      "loss": 5.0449,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991549252964794,
+      "loss": 5.0268,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002991231295550528,
+      "loss": 5.0151,
+      "step": 550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029909074840739835,
+      "loss": 5.0032,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990577819806332,
+      "loss": 4.971,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029902423040417236,
+      "loss": 4.9582,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002989900938097277,
+      "loss": 4.938,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989553723313078,
+      "loss": 4.9201,
+      "step": 600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989200661052174,
+      "loss": 4.8989,
+      "step": 610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988841752700564,
+      "loss": 4.8805,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029884769996672004,
+      "loss": 4.8578,
+      "step": 630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988106403383978,
+      "loss": 4.8442,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987729965305731,
+      "loss": 4.8243,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029873476869102255,
+      "loss": 4.803,
+      "step": 660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002986959569698156,
+      "loss": 4.8042,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029865656151931375,
+      "loss": 4.7661,
+      "step": 680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029861658249416995,
+      "loss": 4.764,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985760200513281,
+      "loss": 4.7372,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029853487435002254,
+      "loss": 4.7294,
+      "step": 710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002984931455517771,
+      "loss": 4.7145,
+      "step": 720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984508338204048,
+      "loss": 4.6965,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002984079393220069,
+      "loss": 4.6641,
+      "step": 740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002983644622249725,
+      "loss": 4.6533,
+      "step": 750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002983204026999778,
+      "loss": 4.6306,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982757609199853,
+      "loss": 4.6291,
+      "step": 770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982305370602433,
+      "loss": 4.6066,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981847312982853,
+      "loss": 4.585,
+      "step": 790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029813834381392875,
+      "loss": 4.5656,
+      "step": 800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029809137478927525,
+      "loss": 4.5646,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029804382440870895,
+      "loss": 4.5447,
+      "step": 820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002979956928588963,
+      "loss": 4.5335,
+      "step": 830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002979469803287853,
+      "loss": 4.5072,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978976870096047,
+      "loss": 4.4861,
+      "step": 850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978478130948629,
+      "loss": 4.4802,
+      "step": 860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000297797358780348,
+      "loss": 4.4456,
+      "step": 870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029774632426412634,
+      "loss": 4.4483,
+      "step": 880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029769470974654174,
+      "loss": 4.419,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002976425154302152,
+      "loss": 4.4263,
+      "step": 900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002975897415200437,
+      "loss": 4.3959,
+      "step": 910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002975363882231994,
+      "loss": 4.384,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029748245574912915,
+      "loss": 4.3722,
+      "step": 930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029742794430955325,
+      "loss": 4.3491,
+      "step": 940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000297372854118465,
+      "loss": 4.347,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029731718539212945,
+      "loss": 4.3346,
+      "step": 960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000297260938349083,
+      "loss": 4.3087,
+      "step": 970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029720411321013226,
+      "loss": 4.3229,
+      "step": 980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002971467101983531,
+      "loss": 4.3127,
+      "step": 990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002970887295390902,
+      "loss": 4.3028,
+      "step": 1000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 4.317672252655029,
+      "eval_runtime": 232.0131,
+      "eval_samples_per_second": 328.15,
+      "eval_steps_per_second": 5.862,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029703017145995566,
+      "loss": 4.2698,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969710361908284,
+      "loss": 4.2621,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002969113239638532,
+      "loss": 4.2548,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029685103501343973,
+      "loss": 4.2319,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029679016957626173,
+      "loss": 4.2401,
+      "step": 1050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002967287278912561,
+      "loss": 4.2409,
+      "step": 1060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966667101996218,
+      "loss": 4.2014,
+      "step": 1070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029660411674481884,
+      "loss": 4.1969,
+      "step": 1080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029654094777256776,
+      "loss": 4.1922,
+      "step": 1090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002964772035308483,
+      "loss": 4.1694,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002964128842698984,
+      "loss": 4.1732,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029634799024221345,
+      "loss": 4.1665,
+      "step": 1120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002962825217025451,
+      "loss": 4.161,
+      "step": 1130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002962164789079004,
+      "loss": 4.1565,
+      "step": 1140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029614986211754066,
+      "loss": 4.1453,
+      "step": 1150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029608267159298064,
+      "loss": 4.1329,
+      "step": 1160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002960149075979871,
+      "loss": 4.1231,
+      "step": 1170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029594657039857836,
+      "loss": 4.1111,
+      "step": 1180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002958776602630228,
+      "loss": 4.0985,
+      "step": 1190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002958081774618379,
+      "loss": 4.103,
+      "step": 1200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957381222677894,
+      "loss": 4.1013,
+      "step": 1210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029566749495588986,
+      "loss": 4.0805,
+      "step": 1220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955962958033978,
+      "loss": 4.08,
+      "step": 1230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029552452508981687,
+      "loss": 4.0518,
+      "step": 1240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029545218309689407,
+      "loss": 4.0406,
+      "step": 1250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002953792701086193,
+      "loss": 4.0475,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029530578641122395,
+      "loss": 4.0357,
+      "step": 1270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029523173229317973,
+      "loss": 4.0333,
+      "step": 1280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002951571080451978,
+      "loss": 4.0324,
+      "step": 1290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002950819139602272,
+      "loss": 4.008,
+      "step": 1300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002950061503334543,
+      "loss": 4.0081,
+      "step": 1310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002949298174623009,
+      "loss": 4.0138,
+      "step": 1320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029485291564642377,
+      "loss": 3.9971,
+      "step": 1330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000294775445187713,
+      "loss": 4.0074,
+      "step": 1340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002946974063902911,
+      "loss": 3.9783,
+      "step": 1350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002946187995605115,
+      "loss": 3.9844,
+      "step": 1360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002945396250069576,
+      "loss": 3.9594,
+      "step": 1370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944598830404417,
+      "loss": 3.9639,
+      "step": 1380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002943795739740033,
+      "loss": 3.9662,
+      "step": 1390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002942986981229082,
+      "loss": 3.9468,
+      "step": 1400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002942172558046473,
+      "loss": 3.9549,
+      "step": 1410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029413524733893514,
+      "loss": 3.9311,
+      "step": 1420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002940526730477088,
+      "loss": 3.9261,
+      "step": 1430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002939695332551267,
+      "loss": 3.9248,
+      "step": 1440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000293885828287567,
+      "loss": 3.9043,
+      "step": 1450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002938015584736267,
+      "loss": 3.9027,
+      "step": 1460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029371672414412034,
+      "loss": 3.906,
+      "step": 1470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002936313256320782,
+      "loss": 3.9054,
+      "step": 1480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002935453632727456,
+      "loss": 3.8934,
+      "step": 1490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002934588374035812,
+      "loss": 3.9062,
+      "step": 1500
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 3.932741165161133,
+      "eval_runtime": 231.8207,
+      "eval_samples_per_second": 328.422,
+      "eval_steps_per_second": 5.867,
+      "step": 1500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029337174836425593,
+      "loss": 3.8889,
+      "step": 1510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002932840964966515,
+      "loss": 3.8737,
+      "step": 1520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002931958821448589,
+      "loss": 3.8813,
+      "step": 1530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002931071056551776,
+      "loss": 3.8551,
+      "step": 1540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930177673761135,
+      "loss": 3.8772,
+      "step": 1550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029292786765837815,
+      "loss": 3.867,
+      "step": 1560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928374068548868,
+      "loss": 3.857,
+      "step": 1570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002927463853207577,
+      "loss": 3.8386,
+      "step": 1580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029265480341331005,
+      "loss": 3.8578,
+      "step": 1590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000292562661492063,
+      "loss": 3.8435,
+      "step": 1600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002924699599187341,
+      "loss": 3.8417,
+      "step": 1610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000292376699057238,
+      "loss": 3.8288,
+      "step": 1620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002922828792736847,
+      "loss": 3.8254,
+      "step": 1630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002921885009363786,
+      "loss": 3.8248,
+      "step": 1640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002920935644158166,
+      "loss": 3.8104,
+      "step": 1650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919980700846869,
+      "loss": 3.8048,
+      "step": 1660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029190201831786743,
+      "loss": 3.792,
+      "step": 1670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002918054094924246,
+      "loss": 3.7815,
+      "step": 1680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002917082439876113,
+      "loss": 3.7978,
+      "step": 1690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029161052218486613,
+      "loss": 3.7963,
+      "step": 1700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029151224446781124,
+      "loss": 3.781,
+      "step": 1710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002914134112222513,
+      "loss": 3.7736,
+      "step": 1720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002913140228361717,
+      "loss": 3.7792,
+      "step": 1730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000291214079699737,
+      "loss": 3.774,
+      "step": 1740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002911135822052899,
+      "loss": 3.7752,
+      "step": 1750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029101253074734885,
+      "loss": 3.7709,
+      "step": 1760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029091092572260726,
+      "loss": 3.7504,
+      "step": 1770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029080876752993163,
+      "loss": 3.7567,
+      "step": 1780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029070605657035995,
+      "loss": 3.7666,
+      "step": 1790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002906027932471002,
+      "loss": 3.7396,
+      "step": 1800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002904989779655287,
+      "loss": 3.7363,
+      "step": 1810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002903946111331886,
+      "loss": 3.7365,
+      "step": 1820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029028969315978846,
+      "loss": 3.7414,
+      "step": 1830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002901842244572,
+      "loss": 3.7447,
+      "step": 1840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00029007820543945726,
+      "loss": 3.7083,
+      "step": 1850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028997163652275446,
+      "loss": 3.7141,
+      "step": 1860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028986451812544465,
+      "loss": 3.719,
+      "step": 1870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028975685066803786,
+      "loss": 3.6981,
+      "step": 1880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002896486345731996,
+      "loss": 3.7213,
+      "step": 1890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002895398702657491,
+      "loss": 3.7031,
+      "step": 1900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028943055817265775,
+      "loss": 3.697,
+      "step": 1910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002893206987230472,
+      "loss": 3.6925,
+      "step": 1920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002892102923481881,
+      "loss": 3.6913,
+      "step": 1930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002890993394814979,
+      "loss": 3.694,
+      "step": 1940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002889878405585395,
+      "loss": 3.6847,
+      "step": 1950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028887579601701935,
+      "loss": 3.6871,
+      "step": 1960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002887632062967859,
+      "loss": 3.6958,
+      "step": 1970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028865007183982776,
+      "loss": 3.6716,
+      "step": 1980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000288536393090272,
+      "loss": 3.6803,
+      "step": 1990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002884221704943823,
+      "loss": 3.6796,
+      "step": 2000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 3.7106287479400635,
+      "eval_runtime": 231.452,
+      "eval_samples_per_second": 328.945,
+      "eval_steps_per_second": 5.876,
+      "step": 2000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028830740450055747,
+      "loss": 3.6717,
+      "step": 2010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028819209555932943,
+      "loss": 3.671,
+      "step": 2020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028807624412336135,
+      "loss": 3.6582,
+      "step": 2030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002879598506474463,
+      "loss": 3.6543,
+      "step": 2040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002878429155885052,
+      "loss": 3.6701,
+      "step": 2050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002877254394055849,
+      "loss": 3.6581,
+      "step": 2060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002876074225598565,
+      "loss": 3.6328,
+      "step": 2070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028748886551461367,
+      "loss": 3.6513,
+      "step": 2080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028736976873527074,
+      "loss": 3.635,
+      "step": 2090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028725013268936076,
+      "loss": 3.6369,
+      "step": 2100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002871299578465338,
+      "loss": 3.6442,
+      "step": 2110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000287009244678555,
+      "loss": 3.6478,
+      "step": 2120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002868879936593029,
+      "loss": 3.6244,
+      "step": 2130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002867662052647674,
+      "loss": 3.6349,
+      "step": 2140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028664387997304777,
+      "loss": 3.6069,
+      "step": 2150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865210182643514,
+      "loss": 3.6206,
+      "step": 2160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002863976206209909,
+      "loss": 3.6199,
+      "step": 2170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028627368752738325,
+      "loss": 3.6152,
+      "step": 2180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002861492194700472,
+      "loss": 3.6227,
+      "step": 2190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028602421693760163,
+      "loss": 3.6011,
+      "step": 2200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002858986804207636,
+      "loss": 3.6167,
+      "step": 2210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028577261041234627,
+      "loss": 3.58,
+      "step": 2220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00028564600740725745,
+      "loss": 3.6072,
+      "step": 2230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002855188719024969,
+      "loss": 3.5932,
+      "step": 2240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028539120439715504,
+      "loss": 3.5748,
+      "step": 2250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028526300539241065,
+      "loss": 3.5769,
+      "step": 2260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028513427539152906,
+      "loss": 3.5949,
+      "step": 2270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002850050148998601,
+      "loss": 3.59,
+      "step": 2280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002848752244248361,
+      "loss": 3.5819,
+      "step": 2290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002847449044759699,
+      "loss": 3.5742,
+      "step": 2300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028461405556485283,
+      "loss": 3.5698,
+      "step": 2310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028448267820515286,
+      "loss": 3.5552,
+      "step": 2320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028435077291261244,
+      "loss": 3.5663,
+      "step": 2330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028421834020504646,
+      "loss": 3.5596,
+      "step": 2340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028408538060234023,
+      "loss": 3.5519,
+      "step": 2350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002839518946264476,
+      "loss": 3.559,
+      "step": 2360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028381788280138864,
+      "loss": 3.5509,
+      "step": 2370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002836833456532478,
+      "loss": 3.5612,
+      "step": 2380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002835482837101718,
+      "loss": 3.538,
+      "step": 2390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002834126975023674,
+      "loss": 3.536,
+      "step": 2400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028327658756209956,
+      "loss": 3.5435,
+      "step": 2410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002831399544236892,
+      "loss": 3.5472,
+      "step": 2420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002830027986235111,
+      "loss": 3.5376,
+      "step": 2430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028286512069999187,
+      "loss": 3.5466,
+      "step": 2440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002827269211936078,
+      "loss": 3.5272,
+      "step": 2450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002825882006468827,
+      "loss": 3.5275,
+      "step": 2460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028244895960438584,
+      "loss": 3.5263,
+      "step": 2470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002823091986127298,
+      "loss": 3.536,
+      "step": 2480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028216891822056817,
+      "loss": 3.5218,
+      "step": 2490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002820281189785937,
+      "loss": 3.5212,
+      "step": 2500
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 3.564971685409546,
+      "eval_runtime": 231.4549,
+      "eval_samples_per_second": 328.941,
+      "eval_steps_per_second": 5.876,
+      "step": 2500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002818868014395359,
+      "loss": 3.516,
+      "step": 2510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002817449661581589,
+      "loss": 3.513,
+      "step": 2520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028160261369125937,
+      "loss": 3.5122,
+      "step": 2530
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002814597445976642,
+      "loss": 3.4902,
+      "step": 2540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00028131635943822846,
+      "loss": 3.5033,
+      "step": 2550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000281172458775833,
+      "loss": 3.5199,
+      "step": 2560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028102804317538255,
+      "loss": 3.5023,
+      "step": 2570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028088311320380305,
+      "loss": 3.4981,
+      "step": 2580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028073766943003997,
+      "loss": 3.4965,
+      "step": 2590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002805917124250556,
+      "loss": 3.5009,
+      "step": 2600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000280445242761827,
+      "loss": 3.499,
+      "step": 2610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002802982610153439,
+      "loss": 3.4942,
+      "step": 2620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028015076776260605,
+      "loss": 3.4787,
+      "step": 2630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002800027635826214,
+      "loss": 3.4983,
+      "step": 2640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002798542490564037,
+      "loss": 3.4939,
+      "step": 2650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027970522476696984,
+      "loss": 3.4799,
+      "step": 2660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027955569129933805,
+      "loss": 3.461,
+      "step": 2670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027940564924052553,
+      "loss": 3.4817,
+      "step": 2680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00027925509917954577,
+      "loss": 3.4738,
+      "step": 2690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002791040417074067,
+      "loss": 3.4764,
+      "step": 2700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027895247741710813,
+      "loss": 3.4696,
+      "step": 2710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002788004069036394,
+      "loss": 3.4622,
+      "step": 2720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027864783076397723,
+      "loss": 3.4669,
+      "step": 2730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002784947495970831,
+      "loss": 3.4712,
+      "step": 2740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002783411640039012,
+      "loss": 3.4702,
+      "step": 2750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002781870745873558,
+      "loss": 3.4626,
+      "step": 2760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002780324819523491,
+      "loss": 3.444,
+      "step": 2770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002778773867057587,
+      "loss": 3.4455,
+      "step": 2780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027772178945643537,
+      "loss": 3.4417,
+      "step": 2790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027756569081520047,
+      "loss": 3.4368,
+      "step": 2800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002774090913948437,
+      "loss": 3.4543,
+      "step": 2810
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002772519918101206,
+      "loss": 3.4426,
+      "step": 2820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00027709439267775026,
+      "loss": 3.4516,
+      "step": 2830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002769362946164128,
+      "loss": 3.4368,
+      "step": 2840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00027677769824674696,
+      "loss": 3.4482,
+      "step": 2850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002766186041913476,
+      "loss": 3.4555,
+      "step": 2860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002764590130747634,
+      "loss": 3.4459,
+      "step": 2870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027629892552349434,
+      "loss": 3.4408,
+      "step": 2880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027613834216598916,
+      "loss": 3.4288,
+      "step": 2890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027597726363264306,
+      "loss": 3.4429,
+      "step": 2900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002758156905557951,
+      "loss": 3.4363,
+      "step": 2910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002756536235697256,
+      "loss": 3.4278,
+      "step": 2920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002754910633106541,
+      "loss": 3.4199,
+      "step": 2930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027532801041673624,
+      "loss": 3.4193,
+      "step": 2940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002751644655280618,
+      "loss": 3.416,
+      "step": 2950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027500042928665187,
+      "loss": 3.4273,
+      "step": 2960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002748359023364565,
+      "loss": 3.4276,
+      "step": 2970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00027467088532335187,
+      "loss": 3.4175,
+      "step": 2980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002745053788951383,
+      "loss": 3.4092,
+      "step": 2990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002743393837015371,
+      "loss": 3.4045,
+      "step": 3000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 3.44966721534729,
+      "eval_runtime": 231.419,
+      "eval_samples_per_second": 328.992,
+      "eval_steps_per_second": 5.877,
+      "step": 3000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002741729003941885,
+      "loss": 3.4111,
+      "step": 3010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027400592962664884,
+      "loss": 3.4155,
+      "step": 3020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000273838472054388,
+      "loss": 3.403,
+      "step": 3030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000273670528334787,
+      "loss": 3.4067,
+      "step": 3040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002735020991271352,
+      "loss": 3.3993,
+      "step": 3050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002733331850926279,
+      "loss": 3.3928,
+      "step": 3060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027316378689436357,
+      "loss": 3.4282,
+      "step": 3070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002729939051973415,
+      "loss": 3.3903,
+      "step": 3080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002728235406684588,
+      "loss": 3.4059,
+      "step": 3090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002726526939765082,
+      "loss": 3.386,
+      "step": 3100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002724813657921752,
+      "loss": 3.3977,
+      "step": 3110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002723095567880354,
+      "loss": 3.3909,
+      "step": 3120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002721372676385521,
+      "loss": 3.3833,
+      "step": 3130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002719644990200733,
+      "loss": 3.383,
+      "step": 3140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002717912516108292,
+      "loss": 3.3826,
+      "step": 3150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002716175260909298,
+      "loss": 3.3754,
+      "step": 3160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002714433231423619,
+      "loss": 3.3547,
+      "step": 3170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027126864344898633,
+      "loss": 3.3545,
+      "step": 3180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00027109348769653575,
+      "loss": 3.3543,
+      "step": 3190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027091785657261153,
+      "loss": 3.3571,
+      "step": 3200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002707417507666811,
+      "loss": 3.3657,
+      "step": 3210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002705651709700755,
+      "loss": 3.3608,
+      "step": 3220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027038811787598636,
+      "loss": 3.3607,
+      "step": 3230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027021059217946335,
+      "loss": 3.3625,
+      "step": 3240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002700325945774115,
+      "loss": 3.3365,
+      "step": 3250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026985412576858823,
+      "loss": 3.3552,
+      "step": 3260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00026967518645360086,
+      "loss": 3.3539,
+      "step": 3270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026949577733490377,
+      "loss": 3.3554,
+      "step": 3280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026931589911679556,
+      "loss": 3.3367,
+      "step": 3290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002691355525054163,
+      "loss": 3.3529,
+      "step": 3300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000268954738208745,
+      "loss": 3.3458,
+      "step": 3310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002687734569365965,
+      "loss": 3.3611,
+      "step": 3320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002685917094006189,
+      "loss": 3.3554,
+      "step": 3330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002684094963142906,
+      "loss": 3.3453,
+      "step": 3340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002682268183929176,
+      "loss": 3.332,
+      "step": 3350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002680436763536308,
+      "loss": 3.3356,
+      "step": 3360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026786007091538297,
+      "loss": 3.3248,
+      "step": 3370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000267676002798946,
+      "loss": 3.3346,
+      "step": 3380
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026749147272690825,
+      "loss": 3.3365,
+      "step": 3390
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026730648142367136,
+      "loss": 3.3292,
+      "step": 3400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00026712102961544773,
+      "loss": 3.3306,
+      "step": 3410
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002669351180302575,
+      "loss": 3.3376,
+      "step": 3420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002667487473979258,
+      "loss": 3.3219,
+      "step": 3430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002665619184500797,
+      "loss": 3.3314,
+      "step": 3440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026637463192014555,
+      "loss": 3.3386,
+      "step": 3450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026618688854334596,
+      "loss": 3.3298,
+      "step": 3460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002659986890566971,
+      "loss": 3.3246,
+      "step": 3470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002658100341990055,
+      "loss": 3.328,
+      "step": 3480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002656209247108654,
+      "loss": 3.3225,
+      "step": 3490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026543136133465566,
+      "loss": 3.3238,
+      "step": 3500
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 3.364104986190796,
+      "eval_runtime": 231.353,
+      "eval_samples_per_second": 329.086,
+      "eval_steps_per_second": 5.878,
+      "step": 3500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026524134481453716,
+      "loss": 3.3292,
+      "step": 3510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00026505087589644946,
+      "loss": 3.3076,
+      "step": 3520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002648599553281082,
+      "loss": 3.3122,
+      "step": 3530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002646685838590021,
+      "loss": 3.3057,
+      "step": 3540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026447676224038966,
+      "loss": 3.2999,
+      "step": 3550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026428449122529687,
+      "loss": 3.3088,
+      "step": 3560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026409177156851377,
+      "loss": 3.2952,
+      "step": 3570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002638986040265915,
+      "loss": 3.3009,
+      "step": 3580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026370498935783954,
+      "loss": 3.3222,
+      "step": 3590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026351092832232266,
+      "loss": 3.3066,
+      "step": 3600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000263316421681858,
+      "loss": 3.3035,
+      "step": 3610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026312147020001163,
+      "loss": 3.3059,
+      "step": 3620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026292607464209635,
+      "loss": 3.2897,
+      "step": 3630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000262730235775168,
+      "loss": 3.3011,
+      "step": 3640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002625339543680228,
+      "loss": 3.3175,
+      "step": 3650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026233723119119414,
+      "loss": 3.2998,
+      "step": 3660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002621400670169497,
+      "loss": 3.3112,
+      "step": 3670
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002619424626192884,
+      "loss": 3.2962,
+      "step": 3680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002617444187739372,
+      "loss": 3.2828,
+      "step": 3690
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002615459362583484,
+      "loss": 3.2919,
+      "step": 3700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002613470158516962,
+      "loss": 3.2742,
+      "step": 3710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026114765833487374,
+      "loss": 3.2876,
+      "step": 3720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002609478644904903,
+      "loss": 3.2987,
+      "step": 3730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000260747635102868,
+      "loss": 3.2979,
+      "step": 3740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026054697095803853,
+      "loss": 3.2827,
+      "step": 3750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002603458728437406,
+      "loss": 3.2907,
+      "step": 3760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002601443415494164,
+      "loss": 3.2716,
+      "step": 3770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002599423778662085,
+      "loss": 3.2919,
+      "step": 3780
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025973998258695715,
+      "loss": 3.2793,
+      "step": 3790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002595371565061966,
+      "loss": 3.2895,
+      "step": 3800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002593339004201526,
+      "loss": 3.2629,
+      "step": 3810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025913021512673863,
+      "loss": 3.281,
+      "step": 3820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002589261014255534,
+      "loss": 3.2647,
+      "step": 3830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002587215601178771,
+      "loss": 3.2834,
+      "step": 3840
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002585165920066688,
+      "loss": 3.27,
+      "step": 3850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00025831119789656296,
+      "loss": 3.2528,
+      "step": 3860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025810537859386634,
+      "loss": 3.2714,
+      "step": 3870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002578991349065549,
+      "loss": 3.2773,
+      "step": 3880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025769246764427054,
+      "loss": 3.2661,
+      "step": 3890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000257485377618318,
+      "loss": 3.2672,
+      "step": 3900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025727786564166174,
+      "loss": 3.2664,
+      "step": 3910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002570699325289225,
+      "loss": 3.2444,
+      "step": 3920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025686157909637425,
+      "loss": 3.2565,
+      "step": 3930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002566528061619412,
+      "loss": 3.2536,
+      "step": 3940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002564436145451941,
+      "loss": 3.2727,
+      "step": 3950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025623400506734757,
+      "loss": 3.2585,
+      "step": 3960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002560239785512564,
+      "loss": 3.2558,
+      "step": 3970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025581353582141255,
+      "loss": 3.2512,
+      "step": 3980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002556026777039422,
+      "loss": 3.2573,
+      "step": 3990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002553914050266017,
+      "loss": 3.2587,
+      "step": 4000
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 3.2957873344421387,
+      "eval_runtime": 231.6662,
+      "eval_samples_per_second": 328.641,
+      "eval_steps_per_second": 5.871,
+      "step": 4000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002551797186187751,
+      "loss": 3.2439,
+      "step": 4010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025496761931147075,
+      "loss": 3.2466,
+      "step": 4020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002547551079373176,
+      "loss": 3.244,
+      "step": 4030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002545421853305625,
+      "loss": 3.2359,
+      "step": 4040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002543288523270663,
+      "loss": 3.2601,
+      "step": 4050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002541151097643014,
+      "loss": 3.2408,
+      "step": 4060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002539009584813476,
+      "loss": 3.2462,
+      "step": 4070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002536863993188894,
+      "loss": 3.2459,
+      "step": 4080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002534714331192124,
+      "loss": 3.2358,
+      "step": 4090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00025325606072620015,
+      "loss": 3.2436,
+      "step": 4100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025304028298533076,
+      "loss": 3.2471,
+      "step": 4110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025282410074367357,
+      "loss": 3.246,
+      "step": 4120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025260751484988583,
+      "loss": 3.224,
+      "step": 4130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002523905261542095,
+      "loss": 3.2352,
+      "step": 4140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00025217313550846763,
+      "loss": 3.2328,
+      "step": 4150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002519553437660614,
+      "loss": 3.2436,
+      "step": 4160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002517371517819664,
+      "loss": 3.2193,
+      "step": 4170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025151856041272946,
+      "loss": 3.2364,
+      "step": 4180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025129957051646525,
+      "loss": 3.2247,
+      "step": 4190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00025108018295285305,
+      "loss": 3.2206,
+      "step": 4200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00025086039858313307,
+      "loss": 3.2104,
+      "step": 4210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025064021827010327,
+      "loss": 3.2223,
+      "step": 4220
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00025041964287811613,
+      "loss": 3.2228,
+      "step": 4230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00025019867327307475,
+      "loss": 3.2256,
+      "step": 4240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002499773103224301,
+      "loss": 3.2202,
+      "step": 4250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002497555548951771,
+      "loss": 3.2146,
+      "step": 4260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002495334078618515,
+      "loss": 3.2248,
+      "step": 4270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002493108700945263,
+      "loss": 3.2263,
+      "step": 4280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024908794246680837,
+      "loss": 3.2199,
+      "step": 4290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002488646258538351,
+      "loss": 3.2165,
+      "step": 4300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002486409211322708,
+      "loss": 3.2257,
+      "step": 4310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00024841682918030347,
+      "loss": 3.213,
+      "step": 4320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00024819235087764123,
+      "loss": 3.2143,
+      "step": 4330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002479674871055089,
+      "loss": 3.2147,
+      "step": 4340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002477422387466443,
+      "loss": 3.2156,
+      "step": 4350
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002475166066852954,
+      "loss": 3.1976,
+      "step": 4360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00024729059180721613,
+      "loss": 3.2241,
+      "step": 4370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024706419499966335,
+      "loss": 3.2114,
+      "step": 4380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002468374171513932,
+      "loss": 3.2148,
+      "step": 4390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024661025915265775,
+      "loss": 3.2067,
+      "step": 4400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024638272189520124,
+      "loss": 3.2173,
+      "step": 4410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002461548062722569,
+      "loss": 3.2083,
+      "step": 4420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002459265131785433,
+      "loss": 3.202,
+      "step": 4430
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024569784351026073,
+      "loss": 3.2152,
+      "step": 4440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002454687981650877,
+      "loss": 3.1965,
+      "step": 4450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002452393780421779,
+      "loss": 3.2036,
+      "step": 4460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002450095840421558,
+      "loss": 3.2003,
+      "step": 4470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024477941706711385,
+      "loss": 3.2016,
+      "step": 4480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002445488780206086,
+      "loss": 3.1901,
+      "step": 4490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002443179678076572,
+      "loss": 3.2023,
+      "step": 4500
+    },
+    {
+      "epoch": 2.84,
+      "eval_loss": 3.2410736083984375,
+      "eval_runtime": 232.4819,
+      "eval_samples_per_second": 327.488,
+      "eval_steps_per_second": 5.85,
+      "step": 4500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024408668733473388,
+      "loss": 3.205,
+      "step": 4510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002438550375097664,
+      "loss": 3.2041,
+      "step": 4520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024362301924213257,
+      "loss": 3.2007,
+      "step": 4530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024339063344265641,
+      "loss": 3.1913,
+      "step": 4540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024315788102360477,
+      "loss": 3.1823,
+      "step": 4550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024292476289868378,
+      "loss": 3.1869,
+      "step": 4560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002426912799830353,
+      "loss": 3.1804,
+      "step": 4570
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000242457433193233,
+      "loss": 3.1816,
+      "step": 4580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024222322344727922,
+      "loss": 3.1914,
+      "step": 4590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024198865166460097,
+      "loss": 3.1816,
+      "step": 4600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024175371876604656,
+      "loss": 3.1705,
+      "step": 4610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000241518425673882,
+      "loss": 3.1754,
+      "step": 4620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024128277331178714,
+      "loss": 3.18,
+      "step": 4630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024104676260485228,
+      "loss": 3.1763,
+      "step": 4640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024081039447957451,
+      "loss": 3.1827,
+      "step": 4650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024057366986385392,
+      "loss": 3.1719,
+      "step": 4660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024033658968699017,
+      "loss": 3.1723,
+      "step": 4670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002400991548796786,
+      "loss": 3.1831,
+      "step": 4680
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023986136637400686,
+      "loss": 3.1744,
+      "step": 4690
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023962322510345086,
+      "loss": 3.1742,
+      "step": 4700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023938473200287166,
+      "loss": 3.1726,
+      "step": 4710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023914588800851128,
+      "loss": 3.1819,
+      "step": 4720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023890669405798924,
+      "loss": 3.1574,
+      "step": 4730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023866715109029885,
+      "loss": 3.1578,
+      "step": 4740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002384272600458037,
+      "loss": 3.1794,
+      "step": 4750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023818702186623365,
+      "loss": 3.1638,
+      "step": 4760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023794643749468135,
+      "loss": 3.1713,
+      "step": 4770
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023770550787559846,
+      "loss": 3.1672,
+      "step": 4780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000237464233954792,
+      "loss": 3.1633,
+      "step": 4790
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023722261667942057,
+      "loss": 3.1574,
+      "step": 4800
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023698065699799065,
+      "loss": 3.1555,
+      "step": 4810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002367383558603529,
+      "loss": 3.1487,
+      "step": 4820
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023649571421769852,
+      "loss": 3.154,
+      "step": 4830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002362527330225552,
+      "loss": 3.1571,
+      "step": 4840
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023600941322878382,
+      "loss": 3.1508,
+      "step": 4850
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023576575579157432,
+      "loss": 3.1512,
+      "step": 4860
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002355217616674422,
+      "loss": 3.1492,
+      "step": 4870
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002352774318142247,
+      "loss": 3.1457,
+      "step": 4880
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002350327671910769,
+      "loss": 3.1452,
+      "step": 4890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023478776875846815,
+      "loss": 3.1326,
+      "step": 4900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023454243747817827,
+      "loss": 3.1484,
+      "step": 4910
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0002342967743132936,
+      "loss": 3.1389,
+      "step": 4920
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023405078022820344,
+      "loss": 3.1407,
+      "step": 4930
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023380445618859612,
+      "loss": 3.143,
+      "step": 4940
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023355780316145535,
+      "loss": 3.1251,
+      "step": 4950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023331082211505624,
+      "loss": 3.1261,
+      "step": 4960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023306351401896158,
+      "loss": 3.1344,
+      "step": 4970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002328158798440181,
+      "loss": 3.1344,
+      "step": 4980
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023256792056235254,
+      "loss": 3.1245,
+      "step": 4990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023231963714736802,
+      "loss": 3.1294,
+      "step": 5000
+    },
+    {
+      "epoch": 3.15,
+      "eval_loss": 3.191751718521118,
+      "eval_runtime": 231.6631,
+      "eval_samples_per_second": 328.645,
+      "eval_steps_per_second": 5.871,
+      "step": 5000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023207103057374003,
+      "loss": 3.1464,
+      "step": 5010
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002318221018174125,
+      "loss": 3.1494,
+      "step": 5020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023157285185559446,
+      "loss": 3.1321,
+      "step": 5030
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023132328166675561,
+      "loss": 3.1534,
+      "step": 5040
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023107339223062297,
+      "loss": 3.1463,
+      "step": 5050
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023082318452817663,
+      "loss": 3.1359,
+      "step": 5060
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023057265954164614,
+      "loss": 3.1323,
+      "step": 5070
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023032181825450668,
+      "loss": 3.1299,
+      "step": 5080
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00023007066165147503,
+      "loss": 3.146,
+      "step": 5090
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022981919071850577,
+      "loss": 3.1275,
+      "step": 5100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022956740644278754,
+      "loss": 3.1321,
+      "step": 5110
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022931530981273903,
+      "loss": 3.1333,
+      "step": 5120
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022906290181800494,
+      "loss": 3.1364,
+      "step": 5130
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002288101834494526,
+      "loss": 3.1311,
+      "step": 5140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022855715569916745,
+      "loss": 3.1292,
+      "step": 5150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022830381956044974,
+      "loss": 3.1364,
+      "step": 5160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002280501760278101,
+      "loss": 3.1342,
+      "step": 5170
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022779622609696607,
+      "loss": 3.1139,
+      "step": 5180
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022754197076483781,
+      "loss": 3.1168,
+      "step": 5190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022728741102954458,
+      "loss": 3.1144,
+      "step": 5200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022703254789040056,
+      "loss": 3.1203,
+      "step": 5210
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002267773823479109,
+      "loss": 3.1184,
+      "step": 5220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022652191540376793,
+      "loss": 3.115,
+      "step": 5230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022626614806084725,
+      "loss": 3.116,
+      "step": 5240
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022601008132320372,
+      "loss": 3.1191,
+      "step": 5250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022575371619606738,
+      "loss": 3.1217,
+      "step": 5260
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022549705368583975,
+      "loss": 3.1143,
+      "step": 5270
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022524009480008977,
+      "loss": 3.1103,
+      "step": 5280
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022498284054754984,
+      "loss": 3.0979,
+      "step": 5290
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00022472529193811187,
+      "loss": 3.0993,
+      "step": 5300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00022446744998282328,
+      "loss": 3.1084,
+      "step": 5310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00022420931569388315,
+      "loss": 3.1025,
+      "step": 5320
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002239508900846381,
+      "loss": 3.1086,
+      "step": 5330
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00022369217416957834,
+      "loss": 3.106,
+      "step": 5340
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00022343316896433378,
+      "loss": 3.0939,
+      "step": 5350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00022317387548566995,
+      "loss": 3.1205,
+      "step": 5360
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00022291429475148413,
+      "loss": 3.1047,
+      "step": 5370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00022265442778080112,
+      "loss": 3.1108,
+      "step": 5380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00022239427559376946,
+      "loss": 3.1116,
+      "step": 5390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00022213383921165737,
+      "loss": 3.1005,
+      "step": 5400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00022187311965684873,
+      "loss": 3.0941,
+      "step": 5410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00022161211795283903,
+      "loss": 3.1077,
+      "step": 5420
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00022135083512423134,
+      "loss": 3.1011,
+      "step": 5430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00022108927219673244,
+      "loss": 3.0966,
+      "step": 5440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002208274301971485,
+      "loss": 3.0986,
+      "step": 5450
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002205653101533815,
+      "loss": 3.1077,
+      "step": 5460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00022030291309442472,
+      "loss": 3.1003,
+      "step": 5470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002200402400503589,
+      "loss": 3.1052,
+      "step": 5480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021977729205234828,
+      "loss": 3.0985,
+      "step": 5490
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002195140701326365,
+      "loss": 3.1086,
+      "step": 5500
+    },
+    {
+      "epoch": 3.47,
+      "eval_loss": 3.1522881984710693,
+      "eval_runtime": 231.5362,
+      "eval_samples_per_second": 328.825,
+      "eval_steps_per_second": 5.874,
+      "step": 5500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021925057532454242,
+      "loss": 3.1143,
+      "step": 5510
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021898680866245629,
+      "loss": 3.0888,
+      "step": 5520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021872277118183546,
+      "loss": 3.103,
+      "step": 5530
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002184584639192005,
+      "loss": 3.0973,
+      "step": 5540
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021819388791213094,
+      "loss": 3.1052,
+      "step": 5550
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021792904419926145,
+      "loss": 3.0999,
+      "step": 5560
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021766393382027758,
+      "loss": 3.1019,
+      "step": 5570
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002173985578159116,
+      "loss": 3.0909,
+      "step": 5580
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021713291722793866,
+      "loss": 3.0935,
+      "step": 5590
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021686701309917258,
+      "loss": 3.0937,
+      "step": 5600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021660084647346164,
+      "loss": 3.1037,
+      "step": 5610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021633441839568476,
+      "loss": 3.0861,
+      "step": 5620
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00021606772991174708,
+      "loss": 3.0893,
+      "step": 5630
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00021580078206857606,
+      "loss": 3.0782,
+      "step": 5640
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002155335759141173,
+      "loss": 3.0919,
+      "step": 5650
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00021526611249733052,
+      "loss": 3.0914,
+      "step": 5660
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00021499839286818527,
+      "loss": 3.0919,
+      "step": 5670
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00021473041807765688,
+      "loss": 3.0852,
+      "step": 5680
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00021446218917772249,
+      "loss": 3.0811,
+      "step": 5690
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00021419370722135665,
+      "loss": 3.0896,
+      "step": 5700
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00021392497326252737,
+      "loss": 3.0909,
+      "step": 5710
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00021365598835619201,
+      "loss": 3.087,
+      "step": 5720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00021338675355829297,
+      "loss": 3.0814,
+      "step": 5730
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00021311726992575362,
+      "loss": 3.0856,
+      "step": 5740
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00021284753851647428,
+      "loss": 3.0843,
+      "step": 5750
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00021257756038932783,
+      "loss": 3.077,
+      "step": 5760
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002123073366041558,
+      "loss": 3.0701,
+      "step": 5770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00021203686822176404,
+      "loss": 3.0914,
+      "step": 5780
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002117661563039185,
+      "loss": 3.0674,
+      "step": 5790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00021149520191334132,
+      "loss": 3.071,
+      "step": 5800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00021122400611370645,
+      "loss": 3.0797,
+      "step": 5810
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00021095256996963557,
+      "loss": 3.0826,
+      "step": 5820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002106808945466937,
+      "loss": 3.0722,
+      "step": 5830
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00021040898091138544,
+      "loss": 3.0675,
+      "step": 5840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00021013683013115035,
+      "loss": 3.0702,
+      "step": 5850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020986444327435896,
+      "loss": 3.0713,
+      "step": 5860
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020959182141030866,
+      "loss": 3.0685,
+      "step": 5870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020931896560921927,
+      "loss": 3.0672,
+      "step": 5880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020904587694222908,
+      "loss": 3.0757,
+      "step": 5890
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020877255648139035,
+      "loss": 3.074,
+      "step": 5900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002084990052996655,
+      "loss": 3.0793,
+      "step": 5910
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020822522447092262,
+      "loss": 3.0852,
+      "step": 5920
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020795121506993124,
+      "loss": 3.0712,
+      "step": 5930
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020767697817235817,
+      "loss": 3.0675,
+      "step": 5940
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020740251485476345,
+      "loss": 3.0627,
+      "step": 5950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002071278261945958,
+      "loss": 3.0644,
+      "step": 5960
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020685291327018866,
+      "loss": 3.073,
+      "step": 5970
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020657777716075589,
+      "loss": 3.0741,
+      "step": 5980
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020630241894638727,
+      "loss": 3.0745,
+      "step": 5990
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020602683970804474,
+      "loss": 3.0731,
+      "step": 6000
+    },
+    {
+      "epoch": 3.78,
+      "eval_loss": 3.115403652191162,
+      "eval_runtime": 231.5784,
+      "eval_samples_per_second": 328.766,
+      "eval_steps_per_second": 5.873,
+      "step": 6000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020575104052755782,
+      "loss": 3.0535,
+      "step": 6010
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002054750224876194,
+      "loss": 3.0738,
+      "step": 6020
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020519878667178167,
+      "loss": 3.0598,
+      "step": 6030
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020492233416445152,
+      "loss": 3.0621,
+      "step": 6040
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020464566605088666,
+      "loss": 3.0625,
+      "step": 6050
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020436878341719116,
+      "loss": 3.0692,
+      "step": 6060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020409168735031124,
+      "loss": 3.0621,
+      "step": 6070
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002038143789380309,
+      "loss": 3.0681,
+      "step": 6080
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020353685926896784,
+      "loss": 3.054,
+      "step": 6090
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020325912943256898,
+      "loss": 3.0648,
+      "step": 6100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002029811905191064,
+      "loss": 3.0569,
+      "step": 6110
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002027030436196729,
+      "loss": 3.0654,
+      "step": 6120
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00020242468982617775,
+      "loss": 3.0641,
+      "step": 6130
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002021461302313423,
+      "loss": 3.0408,
+      "step": 6140
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00020186736592869606,
+      "loss": 3.0616,
+      "step": 6150
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00020158839801257192,
+      "loss": 3.0528,
+      "step": 6160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00020130922757810222,
+      "loss": 3.06,
+      "step": 6170
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00020102985572121427,
+      "loss": 3.0514,
+      "step": 6180
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002007502835386261,
+      "loss": 3.0604,
+      "step": 6190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002004705121278421,
+      "loss": 3.0602,
+      "step": 6200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00020019054258714887,
+      "loss": 3.0492,
+      "step": 6210
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019991037601561074,
+      "loss": 3.0482,
+      "step": 6220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019963001351306552,
+      "loss": 3.054,
+      "step": 6230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019934945618012016,
+      "loss": 3.0552,
+      "step": 6240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019906870511814643,
+      "loss": 3.0339,
+      "step": 6250
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001987877614292767,
+      "loss": 3.0507,
+      "step": 6260
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019850662621639941,
+      "loss": 3.0447,
+      "step": 6270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.000198225300583155,
+      "loss": 3.0439,
+      "step": 6280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019794378563393135,
+      "loss": 3.0676,
+      "step": 6290
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019766208247385943,
+      "loss": 3.0559,
+      "step": 6300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001973801922088093,
+      "loss": 3.0542,
+      "step": 6310
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019709811594538528,
+      "loss": 3.0367,
+      "step": 6320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019681585479092208,
+      "loss": 3.0419,
+      "step": 6330
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019653340985348005,
+      "loss": 3.0448,
+      "step": 6340
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019625078224184113,
+      "loss": 3.0384,
+      "step": 6350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019596797306550425,
+      "loss": 3.0378,
+      "step": 6360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019568498343468124,
+      "loss": 3.0153,
+      "step": 6370
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019540181446029223,
+      "loss": 3.0367,
+      "step": 6380
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019511846725396145,
+      "loss": 3.0358,
+      "step": 6390
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019483494292801284,
+      "loss": 3.0304,
+      "step": 6400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001945512425954655,
+      "loss": 3.0338,
+      "step": 6410
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019426736737002964,
+      "loss": 3.0216,
+      "step": 6420
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019398331836610193,
+      "loss": 3.0279,
+      "step": 6430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019369909669876136,
+      "loss": 3.0267,
+      "step": 6440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019341470348376453,
+      "loss": 3.0294,
+      "step": 6450
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019313013983754165,
+      "loss": 3.0223,
+      "step": 6460
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019284540687719196,
+      "loss": 3.028,
+      "step": 6470
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001925605057204793,
+      "loss": 3.0214,
+      "step": 6480
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019227543748582786,
+      "loss": 3.0319,
+      "step": 6490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019199020329231767,
+      "loss": 3.0129,
+      "step": 6500
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 3.0856213569641113,
+      "eval_runtime": 231.6823,
+      "eval_samples_per_second": 328.618,
+      "eval_steps_per_second": 5.87,
+      "step": 6500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019170480425968034,
+      "loss": 3.0334,
+      "step": 6510
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019141924150829445,
+      "loss": 3.0273,
+      "step": 6520
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001911335161591814,
+      "loss": 3.0207,
+      "step": 6530
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019084762933400083,
+      "loss": 3.0216,
+      "step": 6540
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019056158215504634,
+      "loss": 3.0218,
+      "step": 6550
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019027537574524097,
+      "loss": 3.016,
+      "step": 6560
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001899890112281328,
+      "loss": 3.0082,
+      "step": 6570
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001897024897278907,
+      "loss": 3.0206,
+      "step": 6580
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00018941581236929975,
+      "loss": 3.024,
+      "step": 6590
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018912898027775683,
+      "loss": 3.0257,
+      "step": 6600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018884199457926633,
+      "loss": 3.0314,
+      "step": 6610
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001885548564004356,
+      "loss": 3.0197,
+      "step": 6620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018826756686847059,
+      "loss": 3.0035,
+      "step": 6630
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001879801271111714,
+      "loss": 3.0193,
+      "step": 6640
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001876925382569279,
+      "loss": 3.009,
+      "step": 6650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018740480143471518,
+      "loss": 3.0146,
+      "step": 6660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018711691777408926,
+      "loss": 3.0116,
+      "step": 6670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018682888840518256,
+      "loss": 3.0063,
+      "step": 6680
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018654071445869965,
+      "loss": 3.0291,
+      "step": 6690
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001862523970659124,
+      "loss": 3.0157,
+      "step": 6700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018596393735865606,
+      "loss": 3.0168,
+      "step": 6710
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001856753364693244,
+      "loss": 3.0159,
+      "step": 6720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018538659553086546,
+      "loss": 3.0221,
+      "step": 6730
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018509771567677704,
+      "loss": 3.0041,
+      "step": 6740
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018480869804110236,
+      "loss": 3.0101,
+      "step": 6750
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018451954375842543,
+      "loss": 3.0115,
+      "step": 6760
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018423025396386672,
+      "loss": 3.0159,
+      "step": 6770
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018394082979307868,
+      "loss": 3.0139,
+      "step": 6780
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018365127238224134,
+      "loss": 2.995,
+      "step": 6790
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018336158286805763,
+      "loss": 3.0203,
+      "step": 6800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018307176238774924,
+      "loss": 3.0184,
+      "step": 6810
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018278181207905184,
+      "loss": 3.009,
+      "step": 6820
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018249173308021089,
+      "loss": 3.0187,
+      "step": 6830
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018220152652997693,
+      "loss": 3.011,
+      "step": 6840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018191119356760136,
+      "loss": 3.0173,
+      "step": 6850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0001816207353328316,
+      "loss": 3.0101,
+      "step": 6860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018133015296590717,
+      "loss": 3.0029,
+      "step": 6870
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001810394476075546,
+      "loss": 3.0139,
+      "step": 6880
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018074862039898338,
+      "loss": 3.0091,
+      "step": 6890
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018045767248188138,
+      "loss": 3.0094,
+      "step": 6900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001801666049984102,
+      "loss": 3.0028,
+      "step": 6910
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001798754190912009,
+      "loss": 2.9853,
+      "step": 6920
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00017958411590334944,
+      "loss": 3.0048,
+      "step": 6930
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00017929269657841212,
+      "loss": 3.0061,
+      "step": 6940
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00017900116226040127,
+      "loss": 2.9846,
+      "step": 6950
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001787095140937805,
+      "loss": 2.9917,
+      "step": 6960
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00017841775322346043,
+      "loss": 3.0093,
+      "step": 6970
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001781258807947941,
+      "loss": 2.9916,
+      "step": 6980
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001778338979535725,
+      "loss": 2.999,
+      "step": 6990
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00017754180584602015,
+      "loss": 3.0212,
+      "step": 7000
+    },
+    {
+      "epoch": 4.41,
+      "eval_loss": 3.0574326515197754,
+      "eval_runtime": 231.6039,
+      "eval_samples_per_second": 328.729,
+      "eval_steps_per_second": 5.872,
+      "step": 7000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00017724960561879026,
+      "loss": 2.9958,
+      "step": 7010
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00017695729841896073,
+      "loss": 2.9982,
+      "step": 7020
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001766648853940293,
+      "loss": 3.0031,
+      "step": 7030
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00017637236769190914,
+      "loss": 3.0053,
+      "step": 7040
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00017607974646092437,
+      "loss": 2.9969,
+      "step": 7050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00017578702284980555,
+      "loss": 3.0035,
+      "step": 7060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017549419800768498,
+      "loss": 2.9902,
+      "step": 7070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017520127308409263,
+      "loss": 2.991,
+      "step": 7080
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017490824922895116,
+      "loss": 2.9895,
+      "step": 7090
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017461512759257172,
+      "loss": 2.9887,
+      "step": 7100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017432190932564916,
+      "loss": 2.9906,
+      "step": 7110
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017402859557925782,
+      "loss": 2.9885,
+      "step": 7120
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017373518750484678,
+      "loss": 2.993,
+      "step": 7130
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017344168625423552,
+      "loss": 2.9871,
+      "step": 7140
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001731480929796091,
+      "loss": 3.0084,
+      "step": 7150
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017285440883351406,
+      "loss": 2.9945,
+      "step": 7160
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001725606349688535,
+      "loss": 2.9958,
+      "step": 7170
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001722667725388828,
+      "loss": 3.003,
+      "step": 7180
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017197282269720504,
+      "loss": 2.9912,
+      "step": 7190
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017167878659776637,
+      "loss": 2.989,
+      "step": 7200
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001713846653948517,
+      "loss": 2.9867,
+      "step": 7210
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001710904602430798,
+      "loss": 2.9781,
+      "step": 7220
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017079617229739922,
+      "loss": 2.9936,
+      "step": 7230
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017050180271308333,
+      "loss": 2.98,
+      "step": 7240
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017020735264572622,
+      "loss": 2.9854,
+      "step": 7250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00016991282325123778,
+      "loss": 2.9879,
+      "step": 7260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00016961821568583919,
+      "loss": 2.9999,
+      "step": 7270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001693235311060588,
+      "loss": 3.0009,
+      "step": 7280
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00016902877066872707,
+      "loss": 2.986,
+      "step": 7290
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00016873393553097236,
+      "loss": 2.9756,
+      "step": 7300
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001684390268502162,
+      "loss": 2.9828,
+      "step": 7310
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00016814404578416884,
+      "loss": 2.9746,
+      "step": 7320
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00016784899349082472,
+      "loss": 2.9849,
+      "step": 7330
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00016755387112845792,
+      "loss": 2.9925,
+      "step": 7340
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00016725867985561748,
+      "loss": 2.9901,
+      "step": 7350
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00016696342083112308,
+      "loss": 2.9781,
+      "step": 7360
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001666680952140603,
+      "loss": 2.9703,
+      "step": 7370
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00016637270416377617,
+      "loss": 2.9757,
+      "step": 7380
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00016607724883987456,
+      "loss": 2.9868,
+      "step": 7390
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001657817304022117,
+      "loss": 2.9959,
+      "step": 7400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00016548615001089155,
+      "loss": 2.9855,
+      "step": 7410
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016519050882626127,
+      "loss": 2.9812,
+      "step": 7420
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016489480800890665,
+      "loss": 2.9826,
+      "step": 7430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00016459904871964767,
+      "loss": 2.9794,
+      "step": 7440
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001643032321195338,
+      "loss": 2.9787,
+      "step": 7450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00016400735936983945,
+      "loss": 2.975,
+      "step": 7460
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00016371143163205947,
+      "loss": 2.9893,
+      "step": 7470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0001634154500679047,
+      "loss": 2.9887,
+      "step": 7480
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00016311941583929706,
+      "loss": 2.9831,
+      "step": 7490
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00016282333010836543,
+      "loss": 2.9627,
+      "step": 7500
+    },
+    {
+      "epoch": 4.73,
+      "eval_loss": 3.032374858856201,
+      "eval_runtime": 231.6607,
+      "eval_samples_per_second": 328.649,
+      "eval_steps_per_second": 5.871,
+      "step": 7500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00016252719403744072,
+      "loss": 2.9821,
+      "step": 7510
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00016223100878905153,
+      "loss": 2.9728,
+      "step": 7520
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00016193477552591952,
+      "loss": 2.9809,
+      "step": 7530
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00016163849541095472,
+      "loss": 2.9836,
+      "step": 7540
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00016134216960725126,
+      "loss": 2.9723,
+      "step": 7550
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00016104579927808256,
+      "loss": 2.9781,
+      "step": 7560
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016074938558689674,
+      "loss": 2.9796,
+      "step": 7570
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016045292969731233,
+      "loss": 2.9788,
+      "step": 7580
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016015643277311327,
+      "loss": 2.9822,
+      "step": 7590
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001598598959782449,
+      "loss": 2.9757,
+      "step": 7600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00015956332047680874,
+      "loss": 2.9686,
+      "step": 7610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00015926670743305856,
+      "loss": 2.9649,
+      "step": 7620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00015897005801139528,
+      "loss": 2.979,
+      "step": 7630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00015867337337636278,
+      "loss": 2.9611,
+      "step": 7640
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00015837665469264306,
+      "loss": 2.9606,
+      "step": 7650
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00015807990312505186,
+      "loss": 2.9722,
+      "step": 7660
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00015778311983853402,
+      "loss": 2.9699,
+      "step": 7670
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001574863059981588,
+      "loss": 2.9647,
+      "step": 7680
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00015718946276911555,
+      "loss": 2.9704,
+      "step": 7690
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00015689259131670887,
+      "loss": 2.9684,
+      "step": 7700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00015659569280635418,
+      "loss": 2.9748,
+      "step": 7710
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00015629876840357322,
+      "loss": 2.9708,
+      "step": 7720
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00015600181927398918,
+      "loss": 2.9732,
+      "step": 7730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00015570484658332249,
+      "loss": 2.9672,
+      "step": 7740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00015540785149738596,
+      "loss": 2.9743,
+      "step": 7750
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00015511083518208053,
+      "loss": 2.9541,
+      "step": 7760
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00015481379880339012,
+      "loss": 2.9664,
+      "step": 7770
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00015451674352737768,
+      "loss": 2.9756,
+      "step": 7780
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00015421967052018032,
+      "loss": 2.965,
+      "step": 7790
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001539225809480047,
+      "loss": 2.9597,
+      "step": 7800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00015362547597712247,
+      "loss": 2.9646,
+      "step": 7810
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00015332835677386587,
+      "loss": 2.9423,
+      "step": 7820
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0001530312245046228,
+      "loss": 2.97,
+      "step": 7830
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001527340803358326,
+      "loss": 2.9503,
+      "step": 7840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00015243692543398133,
+      "loss": 2.9649,
+      "step": 7850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001521397609655972,
+      "loss": 2.9552,
+      "step": 7860
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0001518425880972459,
+      "loss": 2.9635,
+      "step": 7870
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001515454079955261,
+      "loss": 2.9696,
+      "step": 7880
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00015124822182706484,
+      "loss": 2.9638,
+      "step": 7890
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001509510307585131,
+      "loss": 2.9675,
+      "step": 7900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000150653835956541,
+      "loss": 2.957,
+      "step": 7910
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00015035663858783336,
+      "loss": 2.9649,
+      "step": 7920
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00015005943981908498,
+      "loss": 2.959,
+      "step": 7930
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001497622408169963,
+      "loss": 2.9612,
+      "step": 7940
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00014946504274826854,
+      "loss": 2.9554,
+      "step": 7950
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00014916784677959934,
+      "loss": 2.9646,
+      "step": 7960
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00014887065407767814,
+      "loss": 2.9496,
+      "step": 7970
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00014857346580918144,
+      "loss": 2.9502,
+      "step": 7980
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00014827628314076837,
+      "loss": 2.9422,
+      "step": 7990
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00014797910723907603,
+      "loss": 2.9347,
+      "step": 8000
+    },
+    {
+      "epoch": 5.04,
+      "eval_loss": 3.01143479347229,
+      "eval_runtime": 232.2812,
+      "eval_samples_per_second": 327.771,
+      "eval_steps_per_second": 5.855,
+      "step": 8000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001476819392707152,
+      "loss": 2.9407,
+      "step": 8010
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00014738478040226524,
+      "loss": 2.9393,
+      "step": 8020
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00014708763180026985,
+      "loss": 2.9287,
+      "step": 8030
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001467904946312325,
+      "loss": 2.9391,
+      "step": 8040
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00014649337006161176,
+      "loss": 2.9454,
+      "step": 8050
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00014619625925781665,
+      "loss": 2.9597,
+      "step": 8060
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00014589916338620233,
+      "loss": 2.9465,
+      "step": 8070
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001456020836130651,
+      "loss": 2.9496,
+      "step": 8080
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00014530502110463825,
+      "loss": 2.9388,
+      "step": 8090
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00014500797702708725,
+      "loss": 2.9439,
+      "step": 8100
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001447109525465052,
+      "loss": 2.9382,
+      "step": 8110
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00014441394882890824,
+      "loss": 2.9319,
+      "step": 8120
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001441169670402311,
+      "loss": 2.9512,
+      "step": 8130
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00014382000834632227,
+      "loss": 2.9437,
+      "step": 8140
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00014352307391293973,
+      "loss": 2.9514,
+      "step": 8150
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00014322616490574614,
+      "loss": 2.9437,
+      "step": 8160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00014292928249030434,
+      "loss": 2.9427,
+      "step": 8170
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00014263242783207284,
+      "loss": 2.9398,
+      "step": 8180
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001423356020964011,
+      "loss": 2.9319,
+      "step": 8190
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00014203880644852518,
+      "loss": 2.9489,
+      "step": 8200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00014174204205356287,
+      "loss": 2.9451,
+      "step": 8210
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001414453100765094,
+      "loss": 2.9535,
+      "step": 8220
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00014114861168223255,
+      "loss": 2.9362,
+      "step": 8230
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00014085194803546846,
+      "loss": 2.9405,
+      "step": 8240
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00014055532030081682,
+      "loss": 2.943,
+      "step": 8250
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00014025872964273627,
+      "loss": 2.9429,
+      "step": 8260
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00013996217722553995,
+      "loss": 2.9638,
+      "step": 8270
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001396656642133909,
+      "loss": 2.938,
+      "step": 8280
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00013936919177029742,
+      "loss": 2.931,
+      "step": 8290
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00013907276106010858,
+      "loss": 2.9352,
+      "step": 8300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00013877637324650961,
+      "loss": 2.9381,
+      "step": 8310
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00013848002949301736,
+      "loss": 2.9373,
+      "step": 8320
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00013818373096297567,
+      "loss": 2.9279,
+      "step": 8330
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00013788747881955096,
+      "loss": 2.936,
+      "step": 8340
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00013759127422572736,
+      "loss": 2.9489,
+      "step": 8350
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00013729511834430252,
+      "loss": 2.9451,
+      "step": 8360
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00013699901233788276,
+      "loss": 2.9402,
+      "step": 8370
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001367029573688786,
+      "loss": 2.9519,
+      "step": 8380
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00013640695459950038,
+      "loss": 2.9457,
+      "step": 8390
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00013611100519175326,
+      "loss": 2.9177,
+      "step": 8400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001358151103074331,
+      "loss": 2.928,
+      "step": 8410
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00013551927110812162,
+      "loss": 2.9448,
+      "step": 8420
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00013522348875518207,
+      "loss": 2.9362,
+      "step": 8430
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00013492776440975437,
+      "loss": 2.9273,
+      "step": 8440
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00013463209923275085,
+      "loss": 2.9254,
+      "step": 8450
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001343364943848515,
+      "loss": 2.9433,
+      "step": 8460
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001340409510264995,
+      "loss": 2.9455,
+      "step": 8470
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00013374547031789672,
+      "loss": 2.9362,
+      "step": 8480
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001334500534189989,
+      "loss": 2.9327,
+      "step": 8490
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00013315470148951146,
+      "loss": 2.9371,
+      "step": 8500
+    },
+    {
+      "epoch": 5.36,
+      "eval_loss": 2.9945642948150635,
+      "eval_runtime": 231.6664,
+      "eval_samples_per_second": 328.641,
+      "eval_steps_per_second": 5.871,
+      "step": 8500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001328594156888847,
+      "loss": 2.9253,
+      "step": 8510
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00013256419717630932,
+      "loss": 2.9306,
+      "step": 8520
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00013226904711071191,
+      "loss": 2.9248,
+      "step": 8530
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00013197396665075035,
+      "loss": 2.9175,
+      "step": 8540
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001316789569548092,
+      "loss": 2.9245,
+      "step": 8550
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001313840191809953,
+      "loss": 2.9269,
+      "step": 8560
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00013108915448713313,
+      "loss": 2.9211,
+      "step": 8570
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00013079436403076026,
+      "loss": 2.9288,
+      "step": 8580
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001304996489691229,
+      "loss": 2.9245,
+      "step": 8590
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001302050104591712,
+      "loss": 2.9152,
+      "step": 8600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00012991044965755482,
+      "loss": 2.9415,
+      "step": 8610
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00012961596772061845,
+      "loss": 2.9324,
+      "step": 8620
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00012932156580439705,
+      "loss": 2.9423,
+      "step": 8630
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001290272450646115,
+      "loss": 2.9338,
+      "step": 8640
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00012873300665666398,
+      "loss": 2.9298,
+      "step": 8650
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00012843885173563358,
+      "loss": 2.9212,
+      "step": 8660
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001281447814562715,
+      "loss": 2.9332,
+      "step": 8670
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00012785079697299676,
+      "loss": 2.9262,
+      "step": 8680
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001275568994398916,
+      "loss": 2.9117,
+      "step": 8690
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00012726309001069678,
+      "loss": 2.9185,
+      "step": 8700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001269693698388073,
+      "loss": 2.9324,
+      "step": 8710
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00012667574007726778,
+      "loss": 2.918,
+      "step": 8720
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001263822018787679,
+      "loss": 2.9052,
+      "step": 8730
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00012608875639563786,
+      "loss": 2.9041,
+      "step": 8740
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00012579540477984395,
+      "loss": 2.9162,
+      "step": 8750
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001255021481829839,
+      "loss": 2.9222,
+      "step": 8760
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012520898775628236,
+      "loss": 2.9139,
+      "step": 8770
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012491592465058683,
+      "loss": 2.9275,
+      "step": 8780
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00012462296001636224,
+      "loss": 2.9076,
+      "step": 8790
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001243300950036873,
+      "loss": 2.9258,
+      "step": 8800
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001240373307622495,
+      "loss": 2.9246,
+      "step": 8810
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001237446684413408,
+      "loss": 2.9109,
+      "step": 8820
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00012345210918985306,
+      "loss": 2.9298,
+      "step": 8830
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00012315965415627347,
+      "loss": 2.9193,
+      "step": 8840
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001228673044886801,
+      "loss": 2.9327,
+      "step": 8850
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001225750613347374,
+      "loss": 2.9271,
+      "step": 8860
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00012228292584169173,
+      "loss": 2.9148,
+      "step": 8870
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00012199089915636677,
+      "loss": 2.922,
+      "step": 8880
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00012169898242515907,
+      "loss": 2.9128,
+      "step": 8890
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00012140717679403355,
+      "loss": 2.9194,
+      "step": 8900
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001211154834085189,
+      "loss": 2.9179,
+      "step": 8910
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001208239034137033,
+      "loss": 2.9243,
+      "step": 8920
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001205324379542297,
+      "loss": 2.9277,
+      "step": 8930
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001202410881742915,
+      "loss": 2.9141,
+      "step": 8940
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00011994985521762797,
+      "loss": 2.9172,
+      "step": 8950
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001196587402275197,
+      "loss": 2.9157,
+      "step": 8960
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00011936774434678416,
+      "loss": 2.9087,
+      "step": 8970
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00011907686871777146,
+      "loss": 2.9169,
+      "step": 8980
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00011878611448235945,
+      "loss": 2.9167,
+      "step": 8990
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00011849548278194941,
+      "loss": 2.9293,
+      "step": 9000
+    },
+    {
+      "epoch": 5.67,
+      "eval_loss": 2.9773881435394287,
+      "eval_runtime": 231.746,
+      "eval_samples_per_second": 328.528,
+      "eval_steps_per_second": 5.868,
+      "step": 9000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00011820497475746169,
+      "loss": 2.9055,
+      "step": 9010
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001179145915493311,
+      "loss": 2.9215,
+      "step": 9020
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00011762433429750244,
+      "loss": 2.8982,
+      "step": 9030
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00011733420414142608,
+      "loss": 2.9057,
+      "step": 9040
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00011704420222005342,
+      "loss": 2.9034,
+      "step": 9050
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00011675432967183248,
+      "loss": 2.9133,
+      "step": 9060
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001164645876347034,
+      "loss": 2.9116,
+      "step": 9070
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00011617497724609403,
+      "loss": 2.9133,
+      "step": 9080
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00011588549964291528,
+      "loss": 2.9252,
+      "step": 9090
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00011559615596155694,
+      "loss": 2.9056,
+      "step": 9100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00011530694733788294,
+      "loss": 2.9046,
+      "step": 9110
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00011501787490722712,
+      "loss": 2.9103,
+      "step": 9120
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00011472893980438864,
+      "loss": 2.905,
+      "step": 9130
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00011444014316362751,
+      "loss": 2.9042,
+      "step": 9140
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00011415148611866027,
+      "loss": 2.9129,
+      "step": 9150
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00011386296980265537,
+      "loss": 2.9154,
+      "step": 9160
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001135745953482289,
+      "loss": 2.9139,
+      "step": 9170
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00011328636388743995,
+      "loss": 2.908,
+      "step": 9180
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00011299827655178641,
+      "loss": 2.9102,
+      "step": 9190
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00011271033447220016,
+      "loss": 2.8991,
+      "step": 9200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00011242253877904306,
+      "loss": 2.9127,
+      "step": 9210
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00011213489060210219,
+      "loss": 2.9128,
+      "step": 9220
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00011184739107058563,
+      "loss": 2.9016,
+      "step": 9230
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00011156004131311779,
+      "loss": 2.8966,
+      "step": 9240
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00011127284245773531,
+      "loss": 2.8996,
+      "step": 9250
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001109857956318822,
+      "loss": 2.8952,
+      "step": 9260
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00011069890196240588,
+      "loss": 2.901,
+      "step": 9270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00011041216257555238,
+      "loss": 2.9066,
+      "step": 9280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00011012557859696219,
+      "loss": 2.9108,
+      "step": 9290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001098391511516656,
+      "loss": 2.9177,
+      "step": 9300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00010955288136407848,
+      "loss": 2.9016,
+      "step": 9310
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00010926677035799775,
+      "loss": 2.9062,
+      "step": 9320
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00010898081925659703,
+      "loss": 2.894,
+      "step": 9330
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00010869502918242223,
+      "loss": 2.9177,
+      "step": 9340
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00010840940125738706,
+      "loss": 2.8976,
+      "step": 9350
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00010812393660276874,
+      "loss": 2.9123,
+      "step": 9360
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00010783863633920356,
+      "loss": 2.9176,
+      "step": 9370
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00010755350158668245,
+      "loss": 2.9107,
+      "step": 9380
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00010726853346454653,
+      "loss": 2.9117,
+      "step": 9390
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00010698373309148293,
+      "loss": 2.8976,
+      "step": 9400
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00010669910158552012,
+      "loss": 2.8997,
+      "step": 9410
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00010641464006402368,
+      "loss": 2.9049,
+      "step": 9420
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00010613034964369196,
+      "loss": 2.8854,
+      "step": 9430
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00010584623144055157,
+      "loss": 2.8868,
+      "step": 9440
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001055622865699531,
+      "loss": 2.9141,
+      "step": 9450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001052785161465666,
+      "loss": 2.9023,
+      "step": 9460
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001049949212843774,
+      "loss": 2.9132,
+      "step": 9470
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00010471150309668156,
+      "loss": 2.9008,
+      "step": 9480
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00010442826269608168,
+      "loss": 2.9089,
+      "step": 9490
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00010414520119448234,
+      "loss": 2.9053,
+      "step": 9500
+    },
+    {
+      "epoch": 5.99,
+      "eval_loss": 2.962033987045288,
+      "eval_runtime": 232.0169,
+      "eval_samples_per_second": 328.144,
+      "eval_steps_per_second": 5.862,
+      "step": 9500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00010386231970308587,
+      "loss": 2.8956,
+      "step": 9510
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00010357961933238784,
+      "loss": 2.9039,
+      "step": 9520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00010329710119217296,
+      "loss": 2.8843,
+      "step": 9530
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00010301476639151046,
+      "loss": 2.8909,
+      "step": 9540
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00010273261603874986,
+      "loss": 2.8955,
+      "step": 9550
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00010245065124151664,
+      "loss": 2.8872,
+      "step": 9560
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00010216887310670784,
+      "loss": 2.8843,
+      "step": 9570
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00010188728274048766,
+      "loss": 2.8933,
+      "step": 9580
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00010160588124828328,
+      "loss": 2.8857,
+      "step": 9590
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00010132466973478038,
+      "loss": 2.8785,
+      "step": 9600
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001010436493039188,
+      "loss": 2.9006,
+      "step": 9610
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00010076282105888835,
+      "loss": 2.8947,
+      "step": 9620
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00010048218610212434,
+      "loss": 2.8959,
+      "step": 9630
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00010020174553530328,
+      "loss": 2.895,
+      "step": 9640
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.992150045933858e-05,
+      "loss": 2.8819,
+      "step": 9650
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 9.96414519743763e-05,
+      "loss": 2.8811,
+      "step": 9660
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.93616011797906e-05,
+      "loss": 2.8857,
+      "step": 9670
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.908194917417967e-05,
+      "loss": 2.8937,
+      "step": 9680
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 9.880249705536131e-05,
+      "loss": 2.8767,
+      "step": 9690
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.852324592036866e-05,
+      "loss": 2.8863,
+      "step": 9700
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.82441968654458e-05,
+      "loss": 2.8859,
+      "step": 9710
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 9.796535098604359e-05,
+      "loss": 2.8737,
+      "step": 9720
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 9.768670937681522e-05,
+      "loss": 2.8882,
+      "step": 9730
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.740827313161197e-05,
+      "loss": 2.8795,
+      "step": 9740
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.713004334347913e-05,
+      "loss": 2.8915,
+      "step": 9750
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.685202110465122e-05,
+      "loss": 2.8816,
+      "step": 9760
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.657420750654819e-05,
+      "loss": 2.8877,
+      "step": 9770
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.629660363977089e-05,
+      "loss": 2.8866,
+      "step": 9780
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.60192105940968e-05,
+      "loss": 2.8851,
+      "step": 9790
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 9.574202945847584e-05,
+      "loss": 2.9005,
+      "step": 9800
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 9.546506132102602e-05,
+      "loss": 2.8955,
+      "step": 9810
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.518830726902917e-05,
+      "loss": 2.8784,
+      "step": 9820
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 9.491176838892672e-05,
+      "loss": 2.8888,
+      "step": 9830
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 9.463544576631543e-05,
+      "loss": 2.893,
+      "step": 9840
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.435934048594305e-05,
+      "loss": 2.8771,
+      "step": 9850
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.408345363170418e-05,
+      "loss": 2.8894,
+      "step": 9860
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.380778628663594e-05,
+      "loss": 2.8968,
+      "step": 9870
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.353233953291368e-05,
+      "loss": 2.8816,
+      "step": 9880
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 9.325711445184682e-05,
+      "loss": 2.8863,
+      "step": 9890
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 9.29821121238746e-05,
+      "loss": 2.895,
+      "step": 9900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 9.270733362856175e-05,
+      "loss": 2.8737,
+      "step": 9910
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 9.24327800445944e-05,
+      "loss": 2.8894,
+      "step": 9920
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 9.215845244977561e-05,
+      "loss": 2.8835,
+      "step": 9930
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 9.188435192102137e-05,
+      "loss": 2.8817,
+      "step": 9940
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 9.161047953435637e-05,
+      "loss": 2.8659,
+      "step": 9950
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 9.133683636490959e-05,
+      "loss": 2.8942,
+      "step": 9960
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.106342348691015e-05,
+      "loss": 2.8824,
+      "step": 9970
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.079024197368319e-05,
+      "loss": 2.8895,
+      "step": 9980
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.05172928976456e-05,
+      "loss": 2.8905,
+      "step": 9990
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.024457733030174e-05,
+      "loss": 2.8819,
+      "step": 10000
+    },
+    {
+      "epoch": 6.3,
+      "eval_loss": 2.95068359375,
+      "eval_runtime": 232.6252,
+      "eval_samples_per_second": 327.286,
+      "eval_steps_per_second": 5.846,
+      "step": 10000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 8.997209634223937e-05,
+      "loss": 2.881,
+      "step": 10010
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 8.969985100312532e-05,
+      "loss": 2.8876,
+      "step": 10020
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 8.942784238170134e-05,
+      "loss": 2.8745,
+      "step": 10030
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 8.915607154577993e-05,
+      "loss": 2.8599,
+      "step": 10040
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 8.888453956224013e-05,
+      "loss": 2.8831,
+      "step": 10050
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 8.861324749702332e-05,
+      "loss": 2.8706,
+      "step": 10060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 8.834219641512906e-05,
+      "loss": 2.8843,
+      "step": 10070
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 8.807138738061083e-05,
+      "loss": 2.8908,
+      "step": 10080
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 8.780082145657197e-05,
+      "loss": 2.8741,
+      "step": 10090
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 8.75304997051615e-05,
+      "loss": 2.8586,
+      "step": 10100
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 8.726042318756976e-05,
+      "loss": 2.8802,
+      "step": 10110
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 8.699059296402449e-05,
+      "loss": 2.8618,
+      "step": 10120
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 8.672101009378648e-05,
+      "loss": 2.8803,
+      "step": 10130
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 8.64516756351457e-05,
+      "loss": 2.8695,
+      "step": 10140
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 8.618259064541675e-05,
+      "loss": 2.8637,
+      "step": 10150
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 8.591375618093491e-05,
+      "loss": 2.8873,
+      "step": 10160
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 8.564517329705207e-05,
+      "loss": 2.885,
+      "step": 10170
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 8.537684304813239e-05,
+      "loss": 2.8809,
+      "step": 10180
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 8.510876648754847e-05,
+      "loss": 2.8691,
+      "step": 10190
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 8.484094466767672e-05,
+      "loss": 2.8832,
+      "step": 10200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 8.457337863989376e-05,
+      "loss": 2.8705,
+      "step": 10210
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 8.430606945457198e-05,
+      "loss": 2.8769,
+      "step": 10220
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 8.403901816107552e-05,
+      "loss": 2.8736,
+      "step": 10230
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 8.37722258077561e-05,
+      "loss": 2.8717,
+      "step": 10240
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 8.350569344194892e-05,
+      "loss": 2.8683,
+      "step": 10250
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 8.32394221099686e-05,
+      "loss": 2.8738,
+      "step": 10260
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 8.297341285710497e-05,
+      "loss": 2.8846,
+      "step": 10270
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 8.270766672761908e-05,
+      "loss": 2.8783,
+      "step": 10280
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 8.244218476473907e-05,
+      "loss": 2.879,
+      "step": 10290
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 8.217696801065596e-05,
+      "loss": 2.8674,
+      "step": 10300
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 8.191201750651978e-05,
+      "loss": 2.8727,
+      "step": 10310
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 8.164733429243518e-05,
+      "loss": 2.874,
+      "step": 10320
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 8.13829194074576e-05,
+      "loss": 2.8679,
+      "step": 10330
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 8.111877388958924e-05,
+      "loss": 2.8904,
+      "step": 10340
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 8.08548987757747e-05,
+      "loss": 2.8785,
+      "step": 10350
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 8.059129510189713e-05,
+      "loss": 2.8733,
+      "step": 10360
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 8.032796390277408e-05,
+      "loss": 2.8868,
+      "step": 10370
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 8.006490621215347e-05,
+      "loss": 2.8696,
+      "step": 10380
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.980212306270948e-05,
+      "loss": 2.8869,
+      "step": 10390
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.953961548603865e-05,
+      "loss": 2.884,
+      "step": 10400
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.927738451265551e-05,
+      "loss": 2.8794,
+      "step": 10410
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.901543117198893e-05,
+      "loss": 2.8789,
+      "step": 10420
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.875375649237784e-05,
+      "loss": 2.8817,
+      "step": 10430
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.849236150106718e-05,
+      "loss": 2.8697,
+      "step": 10440
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.823124722420402e-05,
+      "loss": 2.8961,
+      "step": 10450
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.797041468683338e-05,
+      "loss": 2.8773,
+      "step": 10460
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.770986491289426e-05,
+      "loss": 2.8721,
+      "step": 10470
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.74495989252157e-05,
+      "loss": 2.8515,
+      "step": 10480
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.71896177455126e-05,
+      "loss": 2.8621,
+      "step": 10490
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.692992239438188e-05,
+      "loss": 2.8776,
+      "step": 10500
+    },
+    {
+      "epoch": 6.62,
+      "eval_loss": 2.939117908477783,
+      "eval_runtime": 232.8018,
+      "eval_samples_per_second": 327.038,
+      "eval_steps_per_second": 5.842,
+      "step": 10500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.667051389129841e-05,
+      "loss": 2.8717,
+      "step": 10510
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.641139325461076e-05,
+      "loss": 2.8705,
+      "step": 10520
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.615256150153781e-05,
+      "loss": 2.8668,
+      "step": 10530
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.589401964816412e-05,
+      "loss": 2.8688,
+      "step": 10540
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.563576870943629e-05,
+      "loss": 2.8642,
+      "step": 10550
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.537780969915886e-05,
+      "loss": 2.8552,
+      "step": 10560
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.512014362999039e-05,
+      "loss": 2.8675,
+      "step": 10570
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.48627715134394e-05,
+      "loss": 2.8693,
+      "step": 10580
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.460569435986052e-05,
+      "loss": 2.8556,
+      "step": 10590
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.434891317845051e-05,
+      "loss": 2.8581,
+      "step": 10600
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.4092428977244e-05,
+      "loss": 2.8718,
+      "step": 10610
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.383624276311e-05,
+      "loss": 2.8615,
+      "step": 10620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.358035554174769e-05,
+      "loss": 2.8594,
+      "step": 10630
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.332476831768248e-05,
+      "loss": 2.8669,
+      "step": 10640
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.306948209426205e-05,
+      "loss": 2.8654,
+      "step": 10650
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.281449787365257e-05,
+      "loss": 2.8794,
+      "step": 10660
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.255981665683453e-05,
+      "loss": 2.881,
+      "step": 10670
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.2305439443599e-05,
+      "loss": 2.8784,
+      "step": 10680
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.20513672325436e-05,
+      "loss": 2.88,
+      "step": 10690
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.179760102106864e-05,
+      "loss": 2.8688,
+      "step": 10700
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.15441418053732e-05,
+      "loss": 2.8732,
+      "step": 10710
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.129099058045114e-05,
+      "loss": 2.8554,
+      "step": 10720
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.10381483400873e-05,
+      "loss": 2.8797,
+      "step": 10730
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.078561607685356e-05,
+      "loss": 2.8754,
+      "step": 10740
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.053339478210489e-05,
+      "loss": 2.8756,
+      "step": 10750
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.028148544597552e-05,
+      "loss": 2.8688,
+      "step": 10760
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.002988905737509e-05,
+      "loss": 2.868,
+      "step": 10770
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.977860660398465e-05,
+      "loss": 2.8689,
+      "step": 10780
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.952763907225288e-05,
+      "loss": 2.8753,
+      "step": 10790
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.927698744739217e-05,
+      "loss": 2.8586,
+      "step": 10800
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.902665271337484e-05,
+      "loss": 2.8814,
+      "step": 10810
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 6.877663585292903e-05,
+      "loss": 2.8571,
+      "step": 10820
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.852693784753517e-05,
+      "loss": 2.8664,
+      "step": 10830
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.827755967742196e-05,
+      "loss": 2.8642,
+      "step": 10840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.802850232156247e-05,
+      "loss": 2.8713,
+      "step": 10850
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.777976675767046e-05,
+      "loss": 2.874,
+      "step": 10860
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 6.753135396219633e-05,
+      "loss": 2.8779,
+      "step": 10870
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.72832649103235e-05,
+      "loss": 2.8744,
+      "step": 10880
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.703550057596441e-05,
+      "loss": 2.8613,
+      "step": 10890
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 6.678806193175684e-05,
+      "loss": 2.8682,
+      "step": 10900
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.654094994905998e-05,
+      "loss": 2.8658,
+      "step": 10910
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.629416559795065e-05,
+      "loss": 2.8592,
+      "step": 10920
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.604770984721951e-05,
+      "loss": 2.8659,
+      "step": 10930
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.580158366436729e-05,
+      "loss": 2.8727,
+      "step": 10940
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 6.555578801560087e-05,
+      "loss": 2.8689,
+      "step": 10950
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.531032386582965e-05,
+      "loss": 2.8724,
+      "step": 10960
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.50651921786616e-05,
+      "loss": 2.8648,
+      "step": 10970
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.482039391639965e-05,
+      "loss": 2.868,
+      "step": 10980
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.457593004003771e-05,
+      "loss": 2.8673,
+      "step": 10990
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 6.433180150925708e-05,
+      "loss": 2.8621,
+      "step": 11000
+    },
+    {
+      "epoch": 6.93,
+      "eval_loss": 2.929156541824341,
+      "eval_runtime": 232.09,
+      "eval_samples_per_second": 328.041,
+      "eval_steps_per_second": 5.86,
+      "step": 11000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.408800928242264e-05,
+      "loss": 2.857,
+      "step": 11010
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.38445543165789e-05,
+      "loss": 2.8705,
+      "step": 11020
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.360143756744655e-05,
+      "loss": 2.8602,
+      "step": 11030
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 6.335865998941852e-05,
+      "loss": 2.868,
+      "step": 11040
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.311622253555625e-05,
+      "loss": 2.8562,
+      "step": 11050
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.287412615758597e-05,
+      "loss": 2.8695,
+      "step": 11060
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.263237180589501e-05,
+      "loss": 2.862,
+      "step": 11070
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.239096042952794e-05,
+      "loss": 2.8519,
+      "step": 11080
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.2149892976183e-05,
+      "loss": 2.8517,
+      "step": 11090
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.190917039220828e-05,
+      "loss": 2.8547,
+      "step": 11100
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.166879362259802e-05,
+      "loss": 2.8427,
+      "step": 11110
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 6.142876361098892e-05,
+      "loss": 2.8593,
+      "step": 11120
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.118908129965642e-05,
+      "loss": 2.8549,
+      "step": 11130
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.0949747629511036e-05,
+      "loss": 2.8383,
+      "step": 11140
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.07107635400946e-05,
+      "loss": 2.8571,
+      "step": 11150
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.0472129969576645e-05,
+      "loss": 2.8464,
+      "step": 11160
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 6.023384785475065e-05,
+      "loss": 2.8512,
+      "step": 11170
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.9995918131030436e-05,
+      "loss": 2.8595,
+      "step": 11180
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.975834173244644e-05,
+      "loss": 2.8364,
+      "step": 11190
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.9521119591642065e-05,
+      "loss": 2.8371,
+      "step": 11200
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.928425263987001e-05,
+      "loss": 2.8571,
+      "step": 11210
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.9047741806988705e-05,
+      "loss": 2.8545,
+      "step": 11220
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.881158802145838e-05,
+      "loss": 2.8447,
+      "step": 11230
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.857579221033784e-05,
+      "loss": 2.8493,
+      "step": 11240
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.8340355299280496e-05,
+      "loss": 2.8389,
+      "step": 11250
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.810527821253084e-05,
+      "loss": 2.8658,
+      "step": 11260
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.787056187292087e-05,
+      "loss": 2.8495,
+      "step": 11270
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7636207201866364e-05,
+      "loss": 2.8539,
+      "step": 11280
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.740221511936332e-05,
+      "loss": 2.849,
+      "step": 11290
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.716858654398433e-05,
+      "loss": 2.8382,
+      "step": 11300
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.6935322392875115e-05,
+      "loss": 2.8427,
+      "step": 11310
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.6702423581750553e-05,
+      "loss": 2.8463,
+      "step": 11320
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.646989102489148e-05,
+      "loss": 2.8453,
+      "step": 11330
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.623772563514092e-05,
+      "loss": 2.8443,
+      "step": 11340
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.600592832390051e-05,
+      "loss": 2.8633,
+      "step": 11350
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.5774500001126926e-05,
+      "loss": 2.8373,
+      "step": 11360
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.5543441575328366e-05,
+      "loss": 2.8572,
+      "step": 11370
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.5312753953560894e-05,
+      "loss": 2.8572,
+      "step": 11380
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.5082438041424956e-05,
+      "loss": 2.8412,
+      "step": 11390
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.485249474306175e-05,
+      "loss": 2.8455,
+      "step": 11400
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.462292496114978e-05,
+      "loss": 2.8576,
+      "step": 11410
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.4393729596901265e-05,
+      "loss": 2.8539,
+      "step": 11420
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.4164909550058424e-05,
+      "loss": 2.8403,
+      "step": 11430
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.393646571889031e-05,
+      "loss": 2.8586,
+      "step": 11440
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.370839900018898e-05,
+      "loss": 2.85,
+      "step": 11450
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.348071028926611e-05,
+      "loss": 2.8455,
+      "step": 11460
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.325340047994945e-05,
+      "loss": 2.8512,
+      "step": 11470
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.302647046457927e-05,
+      "loss": 2.8383,
+      "step": 11480
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.279992113400489e-05,
+      "loss": 2.8568,
+      "step": 11490
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.257375337758134e-05,
+      "loss": 2.8486,
+      "step": 11500
+    },
+    {
+      "epoch": 7.25,
+      "eval_loss": 2.922297716140747,
+      "eval_runtime": 232.8422,
+      "eval_samples_per_second": 326.981,
+      "eval_steps_per_second": 5.841,
+      "step": 11500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.2347968083165605e-05,
+      "loss": 2.8533,
+      "step": 11510
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.212256613711317e-05,
+      "loss": 2.8399,
+      "step": 11520
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.189754842427475e-05,
+      "loss": 2.8422,
+      "step": 11530
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.1672915827992675e-05,
+      "loss": 2.8355,
+      "step": 11540
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.144866923009742e-05,
+      "loss": 2.8508,
+      "step": 11550
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.122480951090417e-05,
+      "loss": 2.8484,
+      "step": 11560
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.100133754920935e-05,
+      "loss": 2.8483,
+      "step": 11570
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.077825422228717e-05,
+      "loss": 2.8538,
+      "step": 11580
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.0555560405886253e-05,
+      "loss": 2.8633,
+      "step": 11590
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.0333256974226065e-05,
+      "loss": 2.8479,
+      "step": 11600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.0111344799993585e-05,
+      "loss": 2.8392,
+      "step": 11610
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.9889824754339845e-05,
+      "loss": 2.8392,
+      "step": 11620
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.966869770687658e-05,
+      "loss": 2.8422,
+      "step": 11630
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.944796452567259e-05,
+      "loss": 2.8465,
+      "step": 11640
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.9227626077250625e-05,
+      "loss": 2.8607,
+      "step": 11650
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.900768322658383e-05,
+      "loss": 2.8398,
+      "step": 11660
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.8788136837092345e-05,
+      "loss": 2.8458,
+      "step": 11670
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.856898777063996e-05,
+      "loss": 2.8606,
+      "step": 11680
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.8350236887530624e-05,
+      "loss": 2.8371,
+      "step": 11690
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.8131885046505346e-05,
+      "loss": 2.846,
+      "step": 11700
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.791393310473846e-05,
+      "loss": 2.8548,
+      "step": 11710
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.7696381917834535e-05,
+      "loss": 2.845,
+      "step": 11720
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.7479232339824786e-05,
+      "loss": 2.86,
+      "step": 11730
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.726248522316395e-05,
+      "loss": 2.8588,
+      "step": 11740
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.704614141872684e-05,
+      "loss": 2.8491,
+      "step": 11750
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.6830201775805e-05,
+      "loss": 2.8517,
+      "step": 11760
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.661466714210333e-05,
+      "loss": 2.8444,
+      "step": 11770
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.639953836373685e-05,
+      "loss": 2.8409,
+      "step": 11780
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6184816285227295e-05,
+      "loss": 2.8288,
+      "step": 11790
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.597050174949986e-05,
+      "loss": 2.8427,
+      "step": 11800
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.575659559787986e-05,
+      "loss": 2.855,
+      "step": 11810
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.554309867008943e-05,
+      "loss": 2.8484,
+      "step": 11820
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.533001180424425e-05,
+      "loss": 2.8502,
+      "step": 11830
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.511733583685027e-05,
+      "loss": 2.8428,
+      "step": 11840
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.490507160280022e-05,
+      "loss": 2.8495,
+      "step": 11850
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.469321993537074e-05,
+      "loss": 2.8366,
+      "step": 11860
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.448178166621872e-05,
+      "loss": 2.8526,
+      "step": 11870
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.427075762537822e-05,
+      "loss": 2.8478,
+      "step": 11880
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.406014864125728e-05,
+      "loss": 2.851,
+      "step": 11890
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.384995554063444e-05,
+      "loss": 2.8435,
+      "step": 11900
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.36401791486557e-05,
+      "loss": 2.8464,
+      "step": 11910
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.343082028883115e-05,
+      "loss": 2.8501,
+      "step": 11920
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.32218797830319e-05,
+      "loss": 2.8499,
+      "step": 11930
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.3013358451486554e-05,
+      "loss": 2.8398,
+      "step": 11940
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.280525711277835e-05,
+      "loss": 2.8377,
+      "step": 11950
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.259757658384172e-05,
+      "loss": 2.8344,
+      "step": 11960
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.239031767995913e-05,
+      "loss": 2.8476,
+      "step": 11970
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.218348121475794e-05,
+      "loss": 2.848,
+      "step": 11980
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.19770680002071e-05,
+      "loss": 2.8589,
+      "step": 11990
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.1771078846614056e-05,
+      "loss": 2.8466,
+      "step": 12000
+    },
+    {
+      "epoch": 7.56,
+      "eval_loss": 2.9153051376342773,
+      "eval_runtime": 233.3059,
+      "eval_samples_per_second": 326.331,
+      "eval_steps_per_second": 5.829,
+      "step": 12000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 15860,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 2.007955407686861e+19,
+  "trial_name": null,
+  "trial_params": null
+}