diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,61619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9702708996351781,
+  "eval_steps": 50,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 4.0254,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 4.1849,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015,
+      "loss": 4.2386,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 4.1285,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025,
+      "loss": 4.0887,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003,
+      "loss": 4.1409,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00035,
+      "loss": 4.2036,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004,
+      "loss": 4.346,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 4.1549,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0005,
+      "loss": 4.2178,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999999883621518,
+      "loss": 4.3462,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999999534486084,
+      "loss": 4.4665,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999998952593729,
+      "loss": 4.33,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999998137944509,
+      "loss": 4.4873,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999997090538498,
+      "loss": 4.4698,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999995810375794,
+      "loss": 4.401,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999994297456517,
+      "loss": 4.2051,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999992551780808,
+      "loss": 4.3085,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999990573348828,
+      "loss": 4.2592,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999988362160762,
+      "loss": 4.2577,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999985918216817,
+      "loss": 4.444,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999983241517218,
+      "loss": 4.2662,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999980332062218,
+      "loss": 4.1662,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999977189852084,
+      "loss": 4.1995,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499997381488711,
+      "loss": 4.1688,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999970207167612,
+      "loss": 4.304,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999966366693922,
+      "loss": 4.2263,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999962293466402,
+      "loss": 4.2731,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999957987485427,
+      "loss": 4.3407,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999953448751401,
+      "loss": 4.0295,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999948677264746,
+      "loss": 4.3386,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999943673025905,
+      "loss": 4.37,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999938436035345,
+      "loss": 4.5564,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999932966293553,
+      "loss": 4.3882,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999927263801038,
+      "loss": 4.2626,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999921328558333,
+      "loss": 4.2227,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999915160565988,
+      "loss": 4.1267,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999908759824579,
+      "loss": 4.4166,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00049999021263347,
+      "loss": 4.2755,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499989526009697,
+      "loss": 4.3542,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999888161112029,
+      "loss": 4.2336,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999880829380535,
+      "loss": 4.2205,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999873264903174,
+      "loss": 4.2486,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499986546768065,
+      "loss": 4.3493,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999857437713686,
+      "loss": 4.2765,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999849175003031,
+      "loss": 4.3685,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999840679549457,
+      "loss": 4.1998,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499983195135375,
+      "loss": 4.2707,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999822990416727,
+      "loss": 4.293,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499981379673922,
+      "loss": 4.2182,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 4.204286098480225,
+      "eval_runtime": 184.2137,
+      "eval_samples_per_second": 457.1,
+      "eval_steps_per_second": 28.57,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999804370322086,
+      "loss": 4.3972,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999794711166202,
+      "loss": 4.2902,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999784819272467,
+      "loss": 4.3047,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999774694641803,
+      "loss": 4.0996,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999764337275151,
+      "loss": 3.8992,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999753747173477,
+      "loss": 4.114,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999742924337767,
+      "loss": 4.2322,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999731868769026,
+      "loss": 4.2854,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999720580468287,
+      "loss": 4.6661,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999709059436599,
+      "loss": 4.2559,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999697305675034,
+      "loss": 4.3452,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999685319184688,
+      "loss": 4.1544,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999673099966675,
+      "loss": 4.5038,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999660648022135,
+      "loss": 4.3017,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999647963352225,
+      "loss": 4.2749,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999635045958128,
+      "loss": 4.2192,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999621895841046,
+      "loss": 4.4467,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999608513002201,
+      "loss": 4.2274,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999594897442843,
+      "loss": 4.1026,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999581049164237,
+      "loss": 4.1336,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999566968167673,
+      "loss": 4.3004,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999552654454462,
+      "loss": 4.3176,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999538108025937,
+      "loss": 3.9967,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999523328883452,
+      "loss": 4.036,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999508317028382,
+      "loss": 4.1553,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999493072462126,
+      "loss": 4.2617,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999477595186103,
+      "loss": 4.1227,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999461885201753,
+      "loss": 4.2314,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499944594251054,
+      "loss": 4.1835,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999429767113947,
+      "loss": 4.3012,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999413359013481,
+      "loss": 4.2128,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499939671821067,
+      "loss": 4.3455,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999379844707062,
+      "loss": 4.2239,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499936273850423,
+      "loss": 4.2008,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999345399603764,
+      "loss": 3.9634,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499932782800728,
+      "loss": 4.1452,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999310023716414,
+      "loss": 4.1263,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999291986732823,
+      "loss": 4.1302,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999273717058186,
+      "loss": 4.2172,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999255214694205,
+      "loss": 4.4581,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999236479642601,
+      "loss": 4.0654,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999217511905121,
+      "loss": 4.2981,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999198311483529,
+      "loss": 4.3817,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999178878379611,
+      "loss": 4.0603,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499915921259518,
+      "loss": 4.1958,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999139314132064,
+      "loss": 4.0573,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999119182992118,
+      "loss": 4.335,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999098819177213,
+      "loss": 4.2611,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999078222689249,
+      "loss": 4.2342,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999057393530141,
+      "loss": 3.9201,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 4.091300010681152,
+      "eval_runtime": 184.1194,
+      "eval_samples_per_second": 457.334,
+      "eval_steps_per_second": 28.585,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999036331701828,
+      "loss": 4.1719,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999015037206272,
+      "loss": 4.161,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998993510045456,
+      "loss": 4.0996,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998971750221382,
+      "loss": 4.3701,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998949757736078,
+      "loss": 4.0531,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998927532591592,
+      "loss": 4.0008,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998905074789992,
+      "loss": 4.1318,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998882384333367,
+      "loss": 4.0224,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998859461223834,
+      "loss": 4.1142,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998836305463523,
+      "loss": 4.1741,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998812917054592,
+      "loss": 4.1456,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499878929599922,
+      "loss": 3.9141,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998765442299603,
+      "loss": 4.195,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998741355957963,
+      "loss": 4.0699,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998717036976543,
+      "loss": 4.1865,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998692485357607,
+      "loss": 4.048,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499866770110344,
+      "loss": 4.0203,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998642684216351,
+      "loss": 4.0822,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998617434698669,
+      "loss": 4.0472,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998591952552744,
+      "loss": 4.2214,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998566237780947,
+      "loss": 4.2381,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998540290385674,
+      "loss": 3.961,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998514110369342,
+      "loss": 4.055,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998487697734387,
+      "loss": 4.1068,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998461052483266,
+      "loss": 4.2508,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998434174618464,
+      "loss": 3.9818,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998407064142479,
+      "loss": 4.1142,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998379721057838,
+      "loss": 4.2614,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998352145367086,
+      "loss": 4.0893,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998324337072792,
+      "loss": 4.1456,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998296296177541,
+      "loss": 4.3584,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998268022683947,
+      "loss": 3.9565,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998239516594642,
+      "loss": 3.8816,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998210777912279,
+      "loss": 3.9346,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998181806639534,
+      "loss": 4.0294,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998152602779104,
+      "loss": 4.0097,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998123166333708,
+      "loss": 4.0989,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998093497306088,
+      "loss": 4.0331,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998063595699004,
+      "loss": 3.9976,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998033461515242,
+      "loss": 4.199,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998003094757606,
+      "loss": 3.9018,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997972495428924,
+      "loss": 4.1025,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997941663532046,
+      "loss": 4.0761,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499791059906984,
+      "loss": 4.026,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00049978793020452,
+      "loss": 4.0295,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997847772461038,
+      "loss": 4.0301,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997816010320293,
+      "loss": 4.0007,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997784015625919,
+      "loss": 4.0894,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997751788380896,
+      "loss": 4.0053,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997719328588224,
+      "loss": 3.8079,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 4.00947380065918,
+      "eval_runtime": 184.1264,
+      "eval_samples_per_second": 457.316,
+      "eval_steps_per_second": 28.584,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997686636250925,
+      "loss": 4.0394,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997653711372043,
+      "loss": 3.9516,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997620553954645,
+      "loss": 4.0177,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997587164001815,
+      "loss": 3.9605,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997553541516664,
+      "loss": 4.0784,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997519686502321,
+      "loss": 4.0349,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997485598961939,
+      "loss": 3.9866,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997451278898691,
+      "loss": 4.0668,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997416726315772,
+      "loss": 3.8588,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00049973819412164,
+      "loss": 3.9015,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997346923603814,
+      "loss": 4.0536,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997311673481272,
+      "loss": 4.0526,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997276190852057,
+      "loss": 3.9857,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997240475719474,
+      "loss": 3.9856,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997204528086845,
+      "loss": 4.0151,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499716834795752,
+      "loss": 3.8728,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997131935334866,
+      "loss": 4.1646,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997095290222274,
+      "loss": 3.9145,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997058412623154,
+      "loss": 3.9926,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499702130254094,
+      "loss": 3.881,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996983959979088,
+      "loss": 3.9515,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996946384941075,
+      "loss": 3.7699,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996908577430397,
+      "loss": 3.9309,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996870537450576,
+      "loss": 3.9159,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996832265005154,
+      "loss": 4.1335,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996793760097693,
+      "loss": 4.0103,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996755022731778,
+      "loss": 4.0954,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996716052911017,
+      "loss": 3.9624,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996676850639036,
+      "loss": 4.1184,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996637415919486,
+      "loss": 4.0021,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996597748756038,
+      "loss": 4.1113,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996557849152387,
+      "loss": 3.839,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996517717112246,
+      "loss": 4.0812,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499647735263935,
+      "loss": 4.1611,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996436755737461,
+      "loss": 4.0044,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996395926410354,
+      "loss": 4.0748,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996354864661835,
+      "loss": 3.9803,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996313570495724,
+      "loss": 3.8608,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996272043915865,
+      "loss": 4.1062,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996230284926127,
+      "loss": 3.9188,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996188293530397,
+      "loss": 4.0634,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996146069732582,
+      "loss": 4.1937,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996103613536617,
+      "loss": 4.0061,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996060924946452,
+      "loss": 3.9141,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996018003966062,
+      "loss": 3.9331,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995974850599444,
+      "loss": 4.008,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995931464850615,
+      "loss": 3.9823,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995887846723615,
+      "loss": 4.0186,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995843996222503,
+      "loss": 4.0526,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995799913351365,
+      "loss": 3.9802,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 3.9131250381469727,
+      "eval_runtime": 184.1626,
+      "eval_samples_per_second": 457.226,
+      "eval_steps_per_second": 28.578,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995755598114301,
+      "loss": 3.903,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499571105051544,
+      "loss": 3.6821,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995666270558929,
+      "loss": 3.8249,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995621258248936,
+      "loss": 3.5775,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995576013589653,
+      "loss": 3.8041,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995530536585292,
+      "loss": 3.851,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995484827240088,
+      "loss": 3.9172,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995438885558294,
+      "loss": 4.0218,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995392711544189,
+      "loss": 3.7805,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995346305202073,
+      "loss": 3.9137,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995299666536265,
+      "loss": 3.8648,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995252795551106,
+      "loss": 3.9125,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995205692250964,
+      "loss": 3.9111,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499515835664022,
+      "loss": 4.06,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995110788723284,
+      "loss": 3.7837,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995062988504583,
+      "loss": 3.935,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995014955988568,
+      "loss": 3.9268,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994966691179711,
+      "loss": 4.0726,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994918194082506,
+      "loss": 3.9495,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994869464701467,
+      "loss": 4.1226,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994820503041132,
+      "loss": 3.8296,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994771309106059,
+      "loss": 3.7579,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994721882900829,
+      "loss": 3.8448,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994672224430042,
+      "loss": 3.9049,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994622333698322,
+      "loss": 3.8118,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994572210710315,
+      "loss": 3.8781,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994521855470687,
+      "loss": 4.0319,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994471267984125,
+      "loss": 3.8519,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994420448255339,
+      "loss": 3.7218,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994369396289063,
+      "loss": 3.6272,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994318112090048,
+      "loss": 3.8977,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994266595663068,
+      "loss": 3.7845,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994214847012921,
+      "loss": 3.6805,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994162866144425,
+      "loss": 3.7553,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994110653062418,
+      "loss": 4.1188,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994058207771763,
+      "loss": 3.8478,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994005530277342,
+      "loss": 3.8812,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993952620584058,
+      "loss": 3.862,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993899478696839,
+      "loss": 3.9371,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993846104620632,
+      "loss": 3.9546,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993792498360406,
+      "loss": 3.6551,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993738659921154,
+      "loss": 3.793,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993684589307884,
+      "loss": 3.7492,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993630286525634,
+      "loss": 3.8575,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993575751579459,
+      "loss": 3.6834,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993520984474434,
+      "loss": 3.8484,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993465985215662,
+      "loss": 3.8605,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993410753808261,
+      "loss": 3.9291,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993355290257373,
+      "loss": 3.8218,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993299594568163,
+      "loss": 3.9497,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 3.8371810913085938,
+      "eval_runtime": 184.1172,
+      "eval_samples_per_second": 457.339,
+      "eval_steps_per_second": 28.585,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993243666745814,
+      "loss": 3.8581,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993187506795538,
+      "loss": 3.6341,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993131114722559,
+      "loss": 3.901,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993074490532128,
+      "loss": 3.9358,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993017634229519,
+      "loss": 4.0478,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004992960545820024,
+      "loss": 3.6667,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004992903225308958,
+      "loss": 3.7403,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992845672701657,
+      "loss": 3.7197,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992787888003483,
+      "loss": 3.7863,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992729871219811,
+      "loss": 3.9484,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992671622356046,
+      "loss": 3.8581,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992613141417608,
+      "loss": 3.5621,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992554428409944,
+      "loss": 3.6723,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992495483338521,
+      "loss": 3.6985,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992436306208825,
+      "loss": 3.8762,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992376897026367,
+      "loss": 3.9452,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992317255796678,
+      "loss": 3.6146,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992257382525309,
+      "loss": 3.8964,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992197277217837,
+      "loss": 3.822,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992136939879857,
+      "loss": 3.852,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992076370516985,
+      "loss": 4.0157,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992015569134861,
+      "loss": 3.7584,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991954535739147,
+      "loss": 3.8811,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991893270335525,
+      "loss": 3.7871,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991831772929699,
+      "loss": 4.0029,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991770043527393,
+      "loss": 3.9745,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991708082134355,
+      "loss": 3.8977,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991645888756355,
+      "loss": 4.003,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991583463399183,
+      "loss": 3.9365,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000499152080606865,
+      "loss": 3.7859,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000499145791677059,
+      "loss": 3.6961,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991394795510857,
+      "loss": 3.8537,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991331442295331,
+      "loss": 3.9736,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991267857129908,
+      "loss": 3.8854,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991204040020507,
+      "loss": 3.7761,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991139990973072,
+      "loss": 3.9869,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991075709993565,
+      "loss": 3.7911,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000499101119708797,
+      "loss": 4.0526,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990946452262294,
+      "loss": 3.7443,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990881475522565,
+      "loss": 3.5344,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990816266874833,
+      "loss": 4.04,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990750826325168,
+      "loss": 3.9097,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990685153879664,
+      "loss": 3.8687,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990619249544435,
+      "loss": 3.7919,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990553113325615,
+      "loss": 3.8372,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990486745229364,
+      "loss": 3.9524,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000499042014526186,
+      "loss": 3.7484,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990353313429303,
+      "loss": 3.8512,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990286249737917,
+      "loss": 3.7325,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990218954193943,
+      "loss": 3.7474,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.762497901916504,
+      "eval_runtime": 184.1193,
+      "eval_samples_per_second": 457.334,
+      "eval_steps_per_second": 28.585,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990151426803649,
+      "loss": 3.7843,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990083667573321,
+      "loss": 3.9164,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990015676509267,
+      "loss": 3.9578,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498994745361782,
+      "loss": 3.6507,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989878998905327,
+      "loss": 3.7601,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989810312378165,
+      "loss": 3.8629,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989741394042727,
+      "loss": 3.7923,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989672243905432,
+      "loss": 3.731,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989602861972715,
+      "loss": 3.8733,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989533248251037,
+      "loss": 3.9624,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989463402746879,
+      "loss": 3.7685,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989393325466744,
+      "loss": 3.6661,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989323016417157,
+      "loss": 3.9066,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989252475604664,
+      "loss": 4.0844,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989181703035831,
+      "loss": 3.7673,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989110698717248,
+      "loss": 3.8874,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989039462655525,
+      "loss": 3.7354,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988967994857297,
+      "loss": 3.693,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988896295329215,
+      "loss": 3.6098,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988824364077955,
+      "loss": 3.6768,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988752201110214,
+      "loss": 3.7613,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988679806432712,
+      "loss": 3.8889,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988607180052187,
+      "loss": 3.765,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988534321975402,
+      "loss": 3.658,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498846123220914,
+      "loss": 3.9153,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988387910760206,
+      "loss": 3.7043,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988314357635426,
+      "loss": 3.8711,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988240572841649,
+      "loss": 3.6643,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988166556385744,
+      "loss": 3.8915,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988092308274601,
+      "loss": 3.8172,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988017828515134,
+      "loss": 3.9576,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987943117114277,
+      "loss": 3.7745,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987868174078986,
+      "loss": 3.7209,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987792999416239,
+      "loss": 3.793,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987717593133034,
+      "loss": 3.7815,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987641955236391,
+      "loss": 3.9081,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987566085733353,
+      "loss": 3.7843,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987489984630985,
+      "loss": 3.8733,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987413651936369,
+      "loss": 3.6538,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987337087656614,
+      "loss": 3.7336,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987260291798848,
+      "loss": 3.5674,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498718326437022,
+      "loss": 3.6628,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987106005377904,
+      "loss": 3.75,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498702851482909,
+      "loss": 3.7008,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986950792730994,
+      "loss": 3.7785,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986872839090852,
+      "loss": 3.6186,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986794653915923,
+      "loss": 3.8573,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986716237213483,
+      "loss": 3.7398,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986637588990835,
+      "loss": 3.9458,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986558709255302,
+      "loss": 3.7449,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.676128387451172,
+      "eval_runtime": 184.0671,
+      "eval_samples_per_second": 457.463,
+      "eval_steps_per_second": 28.593,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986479598014228,
+      "loss": 3.6677,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986400255274975,
+      "loss": 3.813,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986320681044934,
+      "loss": 3.6978,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986240875331513,
+      "loss": 3.8736,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498616083814214,
+      "loss": 3.5511,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986080569484269,
+      "loss": 3.7988,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986000069365372,
+      "loss": 3.9047,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004985919337792944,
+      "loss": 3.5888,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00049858383747745,
+      "loss": 3.5318,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004985757180317581,
+      "loss": 3.6175,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985675754429744,
+      "loss": 3.5572,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498559409711857,
+      "loss": 3.705,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985512208391663,
+      "loss": 3.7699,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985430088256645,
+      "loss": 3.7144,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985347736721164,
+      "loss": 3.6857,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985265153792886,
+      "loss": 3.5465,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00049851823394795,
+      "loss": 3.369,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985099293788716,
+      "loss": 3.7223,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985016016728264,
+      "loss": 3.5865,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00049849325083059,
+      "loss": 3.6819,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984848768529399,
+      "loss": 3.5468,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984764797406555,
+      "loss": 3.7132,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984680594945186,
+      "loss": 3.7813,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984596161153135,
+      "loss": 3.7986,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984511496038261,
+      "loss": 3.6622,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984426599608445,
+      "loss": 3.667,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984341471871593,
+      "loss": 3.7767,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498425611283563,
+      "loss": 3.501,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984170522508504,
+      "loss": 3.7259,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984084700898182,
+      "loss": 3.501,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983998648012654,
+      "loss": 3.8285,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983912363859935,
+      "loss": 3.7801,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983825848448055,
+      "loss": 3.6693,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983739101785071,
+      "loss": 3.7481,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983652123879057,
+      "loss": 3.2755,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983564914738113,
+      "loss": 3.684,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983477474370358,
+      "loss": 3.472,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983389802783933,
+      "loss": 3.7224,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983301899986999,
+      "loss": 3.4917,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983213765987741,
+      "loss": 3.5853,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983125400794366,
+      "loss": 3.7741,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983036804415098,
+      "loss": 3.6434,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982947976858189,
+      "loss": 3.5856,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982858918131906,
+      "loss": 3.6015,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982769628244543,
+      "loss": 3.7365,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498268010720441,
+      "loss": 3.6484,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982590355019846,
+      "loss": 3.7473,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982500371699203,
+      "loss": 3.8226,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982410157250862,
+      "loss": 3.5541,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982319711683221,
+      "loss": 3.6893,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 3.623260498046875,
+      "eval_runtime": 184.0682,
+      "eval_samples_per_second": 457.461,
+      "eval_steps_per_second": 28.593,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00049822290350047,
+      "loss": 3.7499,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982138127223742,
+      "loss": 3.6531,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498204698834881,
+      "loss": 3.7112,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498195561838839,
+      "loss": 3.7751,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981864017350989,
+      "loss": 3.6975,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981772185245135,
+      "loss": 3.6251,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981680122079378,
+      "loss": 3.8168,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981587827862288,
+      "loss": 3.5461,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498149530260246,
+      "loss": 3.5439,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981402546308508,
+      "loss": 3.6618,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981309558989065,
+      "loss": 3.6898,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981216340652793,
+      "loss": 3.4321,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981122891308368,
+      "loss": 3.9118,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498102921096449,
+      "loss": 3.6051,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980935299629882,
+      "loss": 3.6082,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980841157313287,
+      "loss": 3.8166,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498074678402347,
+      "loss": 3.7267,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980652179769218,
+      "loss": 3.6596,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980557344559339,
+      "loss": 3.4434,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498046227840266,
+      "loss": 3.6444,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980366981308036,
+      "loss": 3.8024,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980271453284335,
+      "loss": 3.5395,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980175694340454,
+      "loss": 3.539,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980079704485309,
+      "loss": 3.6833,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979983483727834,
+      "loss": 3.6486,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979887032076989,
+      "loss": 3.8428,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979790349541753,
+      "loss": 3.8499,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000497969343613113,
+      "loss": 3.59,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979596291854139,
+      "loss": 3.6031,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979498916719828,
+      "loss": 3.722,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979401310737262,
+      "loss": 3.5753,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979303473915526,
+      "loss": 3.5489,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979205406263733,
+      "loss": 3.5153,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979107107791009,
+      "loss": 3.6377,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979008578506509,
+      "loss": 3.664,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978909818419405,
+      "loss": 3.5744,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978810827538892,
+      "loss": 3.5325,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978711605874187,
+      "loss": 3.6944,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978612153434526,
+      "loss": 3.4832,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978512470229171,
+      "loss": 3.4928,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00049784125562674,
+      "loss": 3.6547,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978312411558518,
+      "loss": 3.5741,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978212036111846,
+      "loss": 3.6153,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978111429936731,
+      "loss": 3.5536,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000497801059304254,
+      "loss": 3.542,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977909525438661,
+      "loss": 3.3775,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977808227134502,
+      "loss": 3.7064,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977706698139494,
+      "loss": 3.6554,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977604938463094,
+      "loss": 3.5449,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977502948114771,
+      "loss": 3.518,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 3.556971788406372,
+      "eval_runtime": 184.1927,
+      "eval_samples_per_second": 457.152,
+      "eval_steps_per_second": 28.573,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977400727104024,
+      "loss": 3.5335,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977298275440367,
+      "loss": 3.6111,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977195593133341,
+      "loss": 3.4747,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977092680192506,
+      "loss": 3.634,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976989536627443,
+      "loss": 3.561,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976886162447753,
+      "loss": 3.6533,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976782557663065,
+      "loss": 3.609,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976678722283019,
+      "loss": 3.4062,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976574656317287,
+      "loss": 3.4043,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976470359775556,
+      "loss": 3.4333,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976365832667536,
+      "loss": 3.4599,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976261075002959,
+      "loss": 3.625,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976156086791579,
+      "loss": 3.7779,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497605086804317,
+      "loss": 3.626,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975945418767529,
+      "loss": 3.4872,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975839738974473,
+      "loss": 3.653,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497573382867384,
+      "loss": 3.6776,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975627687875493,
+      "loss": 3.4885,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975521316589312,
+      "loss": 3.4373,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00049754147148252,
+      "loss": 3.4638,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975307882593085,
+      "loss": 3.2983,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975200819902911,
+      "loss": 3.668,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975093526764645,
+      "loss": 3.307,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974986003188277,
+      "loss": 3.6146,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497487824918382,
+      "loss": 3.4834,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974770264761304,
+      "loss": 3.3853,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974662049930783,
+      "loss": 3.4725,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974553604702333,
+      "loss": 3.4121,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974444929086048,
+      "loss": 3.6159,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974336023092048,
+      "loss": 3.4893,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974226886730473,
+      "loss": 3.6623,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974117520011483,
+      "loss": 3.7649,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497400792294526,
+      "loss": 3.7025,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973898095542009,
+      "loss": 3.535,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973788037811954,
+      "loss": 3.4931,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973677749765342,
+      "loss": 3.3132,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973567231412442,
+      "loss": 3.5644,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973456482763542,
+      "loss": 3.3578,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973345503828954,
+      "loss": 3.4934,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973234294619011,
+      "loss": 3.5464,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973122855144066,
+      "loss": 3.475,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973011185414493,
+      "loss": 3.5033,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972899285440692,
+      "loss": 3.6623,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972787155233079,
+      "loss": 3.3105,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972674794802095,
+      "loss": 3.6213,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972562204158199,
+      "loss": 3.3377,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972449383311875,
+      "loss": 3.3521,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972336332273626,
+      "loss": 3.5145,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972223051053979,
+      "loss": 3.6647,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497210953966348,
+      "loss": 3.5812,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 3.4917216300964355,
+      "eval_runtime": 184.1276,
+      "eval_samples_per_second": 457.313,
+      "eval_steps_per_second": 28.583,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971995798112696,
+      "loss": 3.4399,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971881826412218,
+      "loss": 3.5405,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971767624572656,
+      "loss": 3.2547,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971653192604645,
+      "loss": 3.3149,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971538530518835,
+      "loss": 3.4607,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971423638325906,
+      "loss": 3.5429,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497130851603655,
+      "loss": 3.4515,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497119316366149,
+      "loss": 3.5918,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971077581211462,
+      "loss": 3.4458,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970961768697228,
+      "loss": 3.2676,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497084572612957,
+      "loss": 3.3739,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970729453519293,
+      "loss": 3.4312,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970612950877222,
+      "loss": 3.4372,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970496218214205,
+      "loss": 3.4041,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970379255541107,
+      "loss": 3.6581,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000497026206286882,
+      "loss": 3.4133,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970144640208254,
+      "loss": 3.4567,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970026987570342,
+      "loss": 3.4705,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969909104966037,
+      "loss": 3.3829,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969790992406315,
+      "loss": 3.4686,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969672649902171,
+      "loss": 3.6561,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969554077464626,
+      "loss": 3.5318,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969435275104716,
+      "loss": 3.451,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969316242833504,
+      "loss": 3.4354,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969196980662073,
+      "loss": 3.3466,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969077488601525,
+      "loss": 3.4495,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968957766662985,
+      "loss": 3.4876,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968837814857599,
+      "loss": 3.5439,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968717633196537,
+      "loss": 3.4654,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968597221690986,
+      "loss": 3.4573,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968476580352159,
+      "loss": 3.3884,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968355709191284,
+      "loss": 3.6373,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968234608219619,
+      "loss": 3.4147,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968113277448437,
+      "loss": 3.634,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967991716889034,
+      "loss": 3.402,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967869926552727,
+      "loss": 3.4826,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967747906450857,
+      "loss": 3.5228,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967625656594782,
+      "loss": 3.6278,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967503176995885,
+      "loss": 3.4805,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967380467665571,
+      "loss": 3.4043,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967257528615261,
+      "loss": 3.31,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967134359856404,
+      "loss": 3.4862,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967010961400466,
+      "loss": 3.3763,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966887333258935,
+      "loss": 3.5035,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966763475443321,
+      "loss": 3.6563,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966639387965158,
+      "loss": 3.4833,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966515070835997,
+      "loss": 3.5044,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966390524067414,
+      "loss": 3.585,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966265747671001,
+      "loss": 3.497,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966140741658379,
+      "loss": 3.4697,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 3.4366118907928467,
+      "eval_runtime": 184.1361,
+      "eval_samples_per_second": 457.292,
+      "eval_steps_per_second": 28.582,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966015506041183,
+      "loss": 3.2915,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965890040831077,
+      "loss": 3.3788,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965764346039739,
+      "loss": 3.6581,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965638421678871,
+      "loss": 3.4908,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965512267760199,
+      "loss": 3.4104,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965385884295467,
+      "loss": 3.4501,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965259271296442,
+      "loss": 3.3338,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965132428774912,
+      "loss": 3.5431,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004965005356742687,
+      "loss": 3.4741,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964878055211596,
+      "loss": 3.5172,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964750524193494,
+      "loss": 3.5046,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964622763700252,
+      "loss": 3.4484,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964494773743766,
+      "loss": 3.4179,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964366554335951,
+      "loss": 3.4276,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964238105488747,
+      "loss": 3.3172,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964109427214111,
+      "loss": 3.5027,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963980519524022,
+      "loss": 3.1924,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963851382430486,
+      "loss": 3.372,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963722015945522,
+      "loss": 3.4341,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963592420081177,
+      "loss": 3.2188,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963462594849515,
+      "loss": 3.5551,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963332540262623,
+      "loss": 3.4752,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963202256332611,
+      "loss": 3.3831,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963071743071607,
+      "loss": 3.3107,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962941000491764,
+      "loss": 3.3597,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962810028605253,
+      "loss": 3.4294,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962678827424269,
+      "loss": 3.1569,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962547396961026,
+      "loss": 3.4361,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000496241573722776,
+      "loss": 3.4754,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962283848236732,
+      "loss": 3.4104,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004962151730000218,
+      "loss": 3.4316,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000496201938253052,
+      "loss": 3.3896,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000496188680583996,
+      "loss": 3.4906,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961753999940882,
+      "loss": 3.6927,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961620964845648,
+      "loss": 3.4545,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961487700566646,
+      "loss": 3.2612,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961354207116283,
+      "loss": 3.4538,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961220484506987,
+      "loss": 3.3336,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004961086532751209,
+      "loss": 3.5687,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960952351861419,
+      "loss": 3.3951,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960817941850112,
+      "loss": 3.396,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960683302729799,
+      "loss": 3.5702,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960548434513015,
+      "loss": 3.4872,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960413337212321,
+      "loss": 3.5206,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000496027801084029,
+      "loss": 3.4891,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960142455409524,
+      "loss": 3.624,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960006670932643,
+      "loss": 3.5563,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959870657422289,
+      "loss": 3.4533,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959734414891124,
+      "loss": 3.4627,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959597943351835,
+      "loss": 3.4005,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 3.372851848602295,
+      "eval_runtime": 184.1276,
+      "eval_samples_per_second": 457.313,
+      "eval_steps_per_second": 28.583,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959461242817126,
+      "loss": 3.4211,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959324313299723,
+      "loss": 3.274,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959187154812379,
+      "loss": 3.3707,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959049767367859,
+      "loss": 3.4472,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958912150978957,
+      "loss": 3.2566,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958774305658484,
+      "loss": 3.0934,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958636231419275,
+      "loss": 3.4912,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958497928274184,
+      "loss": 3.285,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495835939623609,
+      "loss": 3.5146,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958220635317886,
+      "loss": 3.363,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958081645532495,
+      "loss": 3.4615,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957942426892856,
+      "loss": 3.2183,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495780297941193,
+      "loss": 3.3366,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957663303102702,
+      "loss": 3.4687,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957523397978174,
+      "loss": 3.2872,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957383264051372,
+      "loss": 3.3523,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957242901335344,
+      "loss": 3.3784,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957102309843157,
+      "loss": 3.3856,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956961489587901,
+      "loss": 3.3344,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956820440582686,
+      "loss": 3.3663,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956679162840646,
+      "loss": 3.4402,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956537656374933,
+      "loss": 3.3245,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956395921198721,
+      "loss": 3.4338,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956253957325207,
+      "loss": 3.4348,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956111764767608,
+      "loss": 3.5207,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955969343539161,
+      "loss": 3.5667,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955826693653129,
+      "loss": 3.4396,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955683815122791,
+      "loss": 3.5276,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495554070796145,
+      "loss": 3.4173,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955397372182429,
+      "loss": 3.4642,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955253807799073,
+      "loss": 3.2383,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955110014824749,
+      "loss": 3.3408,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954965993272844,
+      "loss": 3.2831,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954821743156767,
+      "loss": 3.2292,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954677264489948,
+      "loss": 3.4784,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954532557285838,
+      "loss": 3.1743,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954387621557911,
+      "loss": 3.4497,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954242457319659,
+      "loss": 3.2354,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004954097064584598,
+      "loss": 3.2553,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953951443366266,
+      "loss": 3.2734,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953805593678218,
+      "loss": 3.5161,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953659515534035,
+      "loss": 3.3145,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953513208947316,
+      "loss": 3.4657,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953366673931684,
+      "loss": 3.3028,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495321991050078,
+      "loss": 3.2454,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495307291866827,
+      "loss": 3.3635,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952925698447839,
+      "loss": 3.2458,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952778249853191,
+      "loss": 3.239,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952630572898058,
+      "loss": 3.4243,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952482667596186,
+      "loss": 3.202,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 3.3235702514648438,
+      "eval_runtime": 184.101,
+      "eval_samples_per_second": 457.379,
+      "eval_steps_per_second": 28.588,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952334533961346,
+      "loss": 3.5184,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952186172007333,
+      "loss": 3.2326,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952037581747955,
+      "loss": 3.3256,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004951888763197048,
+      "loss": 3.5275,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004951739716368467,
+      "loss": 3.2562,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495159044127609,
+      "loss": 3.3098,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004951440937933815,
+      "loss": 3.2825,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495129120635556,
+      "loss": 3.2279,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004951141246555265,
+      "loss": 3.425,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950991058546893,
+      "loss": 3.3422,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950840642344427,
+      "loss": 3.255,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950689997961869,
+      "loss": 3.1536,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950539125413247,
+      "loss": 3.3539,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950388024712607,
+      "loss": 3.3315,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950236695874016,
+      "loss": 3.4701,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004950085138911564,
+      "loss": 3.3076,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004949933353839361,
+      "loss": 3.2248,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000494978134067154,
+      "loss": 3.4146,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000494962909942225,
+      "loss": 3.6063,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004949476630105669,
+      "loss": 3.2367,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004949323932735991,
+      "loss": 3.3149,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004949171007327432,
+      "loss": 3.3074,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000494901785389423,
+      "loss": 3.1021,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948864472450646,
+      "loss": 3.2404,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948710863010957,
+      "loss": 3.2589,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948557025589467,
+      "loss": 3.1803,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948402960200497,
+      "loss": 3.3883,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948248666858391,
+      "loss": 3.1683,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948094145577515,
+      "loss": 3.3195,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947939396372257,
+      "loss": 3.2256,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947784419257021,
+      "loss": 3.4126,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947629214246237,
+      "loss": 3.2153,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947473781354355,
+      "loss": 3.2456,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947318120595849,
+      "loss": 3.0954,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947162231985208,
+      "loss": 3.2406,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947006115536948,
+      "loss": 3.2444,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00049468497712656,
+      "loss": 3.4792,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946693199185725,
+      "loss": 3.4302,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946536399311897,
+      "loss": 3.3301,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946379371658717,
+      "loss": 3.2881,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946222116240801,
+      "loss": 3.3142,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946064633072795,
+      "loss": 3.4541,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945906922169356,
+      "loss": 3.2304,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945748983545172,
+      "loss": 3.2794,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945590817214944,
+      "loss": 3.2144,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945432423193398,
+      "loss": 3.3857,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945273801495283,
+      "loss": 3.2658,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004945114952135365,
+      "loss": 3.4003,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944955875128437,
+      "loss": 3.1518,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944796570489304,
+      "loss": 3.2481,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 3.2711403369903564,
+      "eval_runtime": 184.2103,
+      "eval_samples_per_second": 457.108,
+      "eval_steps_per_second": 28.571,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944637038232801,
+      "loss": 3.3286,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944477278373781,
+      "loss": 3.2638,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944317290927117,
+      "loss": 3.3588,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944157075907705,
+      "loss": 3.1931,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943996633330462,
+      "loss": 3.3989,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943835963210324,
+      "loss": 3.2189,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943675065562251,
+      "loss": 3.2239,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943513940401223,
+      "loss": 3.4371,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000494335258774224,
+      "loss": 3.2744,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943191007600326,
+      "loss": 3.2448,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943029199990524,
+      "loss": 3.206,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942867164927899,
+      "loss": 3.3416,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942704902427536,
+      "loss": 3.2992,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942542412504543,
+      "loss": 3.4096,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942379695174047,
+      "loss": 3.2201,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942216750451199,
+      "loss": 3.0601,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000494205357835117,
+      "loss": 3.4477,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941890178889149,
+      "loss": 3.0565,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941726552080352,
+      "loss": 3.1112,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941562697940011,
+      "loss": 3.3194,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941398616483382,
+      "loss": 3.4879,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941234307725741,
+      "loss": 3.124,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004941069771682387,
+      "loss": 3.2209,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940905008368638,
+      "loss": 3.231,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940740017799833,
+      "loss": 3.3144,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940574799991335,
+      "loss": 3.445,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940409354958524,
+      "loss": 3.2732,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940243682716805,
+      "loss": 3.3041,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940077783281603,
+      "loss": 3.0482,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939911656668361,
+      "loss": 3.0878,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939745302892549,
+      "loss": 3.3114,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939578721969652,
+      "loss": 3.2027,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939411913915182,
+      "loss": 3.2476,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939244878744668,
+      "loss": 3.4025,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004939077616473662,
+      "loss": 3.1716,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938910127117735,
+      "loss": 3.317,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938742410692482,
+      "loss": 3.3428,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938574467213517,
+      "loss": 3.2785,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938406296696479,
+      "loss": 3.1061,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938237899157021,
+      "loss": 3.3157,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004938069274610823,
+      "loss": 3.158,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004937900423073585,
+      "loss": 3.2348,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004937731344561027,
+      "loss": 3.2497,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000493756203908889,
+      "loss": 3.2452,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004937392506672939,
+      "loss": 3.341,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004937222747328955,
+      "loss": 3.1215,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004937052761072745,
+      "loss": 3.3909,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936882547920136,
+      "loss": 3.1646,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936712107886973,
+      "loss": 3.3852,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936541440989125,
+      "loss": 3.4377,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 3.219867467880249,
+      "eval_runtime": 184.0977,
+      "eval_samples_per_second": 457.387,
+      "eval_steps_per_second": 28.588,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936370547242482,
+      "loss": 3.1954,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936199426662957,
+      "loss": 3.2901,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936028079266477,
+      "loss": 3.1655,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004935856505068998,
+      "loss": 3.2756,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004935684704086495,
+      "loss": 3.3747,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000493551267633496,
+      "loss": 3.073,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004935340421830412,
+      "loss": 3.1629,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004935167940588887,
+      "loss": 3.3325,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934995232626444,
+      "loss": 3.1951,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934822297959161,
+      "loss": 3.1688,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934649136603141,
+      "loss": 3.2935,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934475748574505,
+      "loss": 3.0927,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934302133889397,
+      "loss": 3.141,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004934128292563977,
+      "loss": 3.153,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933954224614435,
+      "loss": 3.2017,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933779930056974,
+      "loss": 3.1219,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933605408907823,
+      "loss": 2.9406,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000493343066118323,
+      "loss": 3.2077,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933255686899464,
+      "loss": 3.1298,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933080486072816,
+      "loss": 3.1962,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004932905058719598,
+      "loss": 3.0294,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004932729404856143,
+      "loss": 3.2156,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004932553524498803,
+      "loss": 3.3942,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004932377417663955,
+      "loss": 3.186,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004932201084367993,
+      "loss": 3.159,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004932024524627336,
+      "loss": 3.3827,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004931847738458423,
+      "loss": 3.3625,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000493167072587771,
+      "loss": 3.2593,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000493149348690168,
+      "loss": 3.1522,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004931316021546834,
+      "loss": 3.2833,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004931138329829693,
+      "loss": 3.261,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930960411766803,
+      "loss": 3.0904,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930782267374726,
+      "loss": 3.4169,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930603896670051,
+      "loss": 3.1319,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930425299669381,
+      "loss": 3.2762,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930246476389348,
+      "loss": 3.027,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930067426846597,
+      "loss": 3.1217,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00049298881510578,
+      "loss": 3.2176,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004929708649039649,
+      "loss": 3.0827,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004929528920808855,
+      "loss": 3.1524,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492934896638215,
+      "loss": 3.0894,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492916878577629,
+      "loss": 3.2148,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492898837900805,
+      "loss": 3.129,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004928807746094226,
+      "loss": 3.0628,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004928626887051636,
+      "loss": 3.214,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004928445801897118,
+      "loss": 3.2448,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004928264490647532,
+      "loss": 3.1993,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004928082953319756,
+      "loss": 3.2212,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004927901189930697,
+      "loss": 3.3408,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004927719200497273,
+      "loss": 3.2781,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 3.173494577407837,
+      "eval_runtime": 184.1307,
+      "eval_samples_per_second": 457.306,
+      "eval_steps_per_second": 28.583,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004927536985036429,
+      "loss": 3.0724,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492735454356513,
+      "loss": 3.316,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004927171876100363,
+      "loss": 3.0209,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926988982659132,
+      "loss": 3.2168,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926805863258467,
+      "loss": 3.4104,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926622517915417,
+      "loss": 3.1951,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926438946647051,
+      "loss": 3.1892,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926255149470461,
+      "loss": 3.3378,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926071126402758,
+      "loss": 3.2316,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004925886877461075,
+      "loss": 3.2881,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004925702402662568,
+      "loss": 3.256,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492551770202441,
+      "loss": 3.0451,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004925332775563798,
+      "loss": 3.412,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004925147623297949,
+      "loss": 3.2687,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924962245244101,
+      "loss": 3.0826,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924776641419512,
+      "loss": 3.2069,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924590811841466,
+      "loss": 3.0912,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924404756527261,
+      "loss": 3.3059,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924218475494221,
+      "loss": 3.1814,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924031968759687,
+      "loss": 2.9895,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004923845236341027,
+      "loss": 3.0084,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004923658278255623,
+      "loss": 3.1009,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004923471094520882,
+      "loss": 3.2916,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004923283685154231,
+      "loss": 3.3034,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492309605017312,
+      "loss": 3.3668,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004922908189595018,
+      "loss": 3.1585,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004922720103437414,
+      "loss": 3.0537,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492253179171782,
+      "loss": 3.1074,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004922343254453768,
+      "loss": 3.0252,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004922154491662813,
+      "loss": 3.0797,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004921965503362528,
+      "loss": 3.3666,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004921776289570507,
+      "loss": 3.295,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004921586850304369,
+      "loss": 3.2816,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492139718558175,
+      "loss": 3.1347,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004921207295420308,
+      "loss": 3.1361,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004921017179837724,
+      "loss": 3.1993,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004920826838851695,
+      "loss": 3.0721,
+      "step": 837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004920636272479946,
+      "loss": 3.1691,
+      "step": 838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004920445480740217,
+      "loss": 3.3731,
+      "step": 839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004920254463650272,
+      "loss": 3.1701,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004920063221227895,
+      "loss": 2.9147,
+      "step": 841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919871753490891,
+      "loss": 3.2938,
+      "step": 842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919680060457086,
+      "loss": 3.149,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919488142144328,
+      "loss": 3.1499,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919295998570485,
+      "loss": 2.8197,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919103629753446,
+      "loss": 2.8719,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000491891103571112,
+      "loss": 3.1645,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004918718216461439,
+      "loss": 3.0197,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004918525172022356,
+      "loss": 3.0766,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004918331902411841,
+      "loss": 3.1309,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 3.135138750076294,
+      "eval_runtime": 184.0835,
+      "eval_samples_per_second": 457.423,
+      "eval_steps_per_second": 28.59,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004918138407647892,
+      "loss": 2.9907,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000491794468774852,
+      "loss": 3.0044,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004917750742731763,
+      "loss": 3.1415,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004917556572615678,
+      "loss": 3.1901,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004917362177418342,
+      "loss": 3.0142,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004917167557157854,
+      "loss": 3.0856,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004916972711852334,
+      "loss": 2.978,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000491677764151992,
+      "loss": 3.34,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004916582346178778,
+      "loss": 3.2043,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004916386825847089,
+      "loss": 3.0525,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004916191080543054,
+      "loss": 3.1111,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00049159951102849,
+      "loss": 3.1213,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004915798915090872,
+      "loss": 3.1659,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004915602494979236,
+      "loss": 3.2623,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004915405849968279,
+      "loss": 3.1852,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000491520898007631,
+      "loss": 3.0088,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004915011885321657,
+      "loss": 3.3292,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914814565722671,
+      "loss": 3.2465,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914617021297723,
+      "loss": 3.1942,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914419252065204,
+      "loss": 3.2578,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914221258043528,
+      "loss": 3.1714,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914023039251127,
+      "loss": 3.2052,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004913824595706457,
+      "loss": 3.3249,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004913625927427996,
+      "loss": 3.0762,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004913427034434236,
+      "loss": 3.078,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004913227916743698,
+      "loss": 3.2499,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004913028574374918,
+      "loss": 3.3072,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004912829007346457,
+      "loss": 3.1702,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004912629215676895,
+      "loss": 3.1088,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004912429199384833,
+      "loss": 3.072,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004912228958488892,
+      "loss": 3.0988,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004912028493007717,
+      "loss": 2.9587,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004911827802959969,
+      "loss": 2.931,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004911626888364337,
+      "loss": 3.1089,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004911425749239523,
+      "loss": 2.8809,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004911224385604255,
+      "loss": 3.194,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000491102279747728,
+      "loss": 3.232,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004910820984877367,
+      "loss": 3.2045,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004910618947823306,
+      "loss": 3.2206,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004910416686333906,
+      "loss": 3.1075,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004910214200427998,
+      "loss": 3.024,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004910011490124436,
+      "loss": 3.2131,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004909808555442091,
+      "loss": 3.2502,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004909605396399855,
+      "loss": 3.0573,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004909402013016647,
+      "loss": 3.0506,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00049091984053114,
+      "loss": 3.1526,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490899457330307,
+      "loss": 2.8793,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004908790517010636,
+      "loss": 3.3011,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004908586236453095,
+      "loss": 3.0275,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004908381731649466,
+      "loss": 3.0013,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 3.0992560386657715,
+      "eval_runtime": 184.1592,
+      "eval_samples_per_second": 457.235,
+      "eval_steps_per_second": 28.579,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004908177002618791,
+      "loss": 3.1838,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004907972049380128,
+      "loss": 3.1041,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490776687195256,
+      "loss": 3.1682,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490756147035519,
+      "loss": 3.3146,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004907355844607142,
+      "loss": 3.362,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004907149994727558,
+      "loss": 3.2354,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906943920735605,
+      "loss": 3.1533,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906737622650469,
+      "loss": 3.2642,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906531100491356,
+      "loss": 3.1863,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906324354277495,
+      "loss": 3.1756,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906117384028133,
+      "loss": 3.0728,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004905910189762542,
+      "loss": 3.2595,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490570277150001,
+      "loss": 3.0497,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490549512925985,
+      "loss": 3.1965,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004905287263061392,
+      "loss": 2.9298,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490507917292399,
+      "loss": 2.9807,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904870858867018,
+      "loss": 2.9911,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904662320909871,
+      "loss": 2.8925,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904453559071963,
+      "loss": 3.2099,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904244573372732,
+      "loss": 3.11,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904035363831635,
+      "loss": 3.2299,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004903825930468149,
+      "loss": 2.9375,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004903616273301773,
+      "loss": 3.0109,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004903406392352027,
+      "loss": 2.9439,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490319628763845,
+      "loss": 3.1074,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902985959180607,
+      "loss": 2.9759,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902775406998076,
+      "loss": 2.8173,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902564631110464,
+      "loss": 3.0322,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902353631537392,
+      "loss": 3.097,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902142408298504,
+      "loss": 3.1125,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004901930961413468,
+      "loss": 2.9736,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004901719290901969,
+      "loss": 2.8839,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004901507396783714,
+      "loss": 3.0597,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000490129527907843,
+      "loss": 3.1005,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004901082937805869,
+      "loss": 3.2361,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004900870372985798,
+      "loss": 3.2612,
+      "step": 936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004900657584638007,
+      "loss": 3.0553,
+      "step": 937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004900444572782309,
+      "loss": 3.1462,
+      "step": 938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004900231337438535,
+      "loss": 2.9031,
+      "step": 939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004900017878626537,
+      "loss": 2.8627,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000489980419636619,
+      "loss": 3.3697,
+      "step": 941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004899590290677387,
+      "loss": 2.9401,
+      "step": 942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004899376161580046,
+      "loss": 2.8733,
+      "step": 943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004899161809094099,
+      "loss": 3.0832,
+      "step": 944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004898947233239506,
+      "loss": 3.2361,
+      "step": 945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004898732434036243,
+      "loss": 3.2303,
+      "step": 946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000489851741150431,
+      "loss": 3.0398,
+      "step": 947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004898302165663725,
+      "loss": 3.0086,
+      "step": 948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004898086696534527,
+      "loss": 3.2773,
+      "step": 949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897871004136779,
+      "loss": 2.9647,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 3.055100202560425,
+      "eval_runtime": 184.016,
+      "eval_samples_per_second": 457.591,
+      "eval_steps_per_second": 28.601,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897655088490561,
+      "loss": 3.1132,
+      "step": 951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897438949615975,
+      "loss": 3.024,
+      "step": 952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897222587533145,
+      "loss": 3.0759,
+      "step": 953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897006002262216,
+      "loss": 2.8295,
+      "step": 954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004896789193823351,
+      "loss": 3.1885,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004896572162236736,
+      "loss": 2.9883,
+      "step": 956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004896354907522578,
+      "loss": 3.1518,
+      "step": 957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004896137429701102,
+      "loss": 3.1378,
+      "step": 958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895919728792558,
+      "loss": 3.1547,
+      "step": 959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895701804817214,
+      "loss": 3.1148,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895483657795358,
+      "loss": 2.993,
+      "step": 961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895265287747302,
+      "loss": 3.15,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895046694693374,
+      "loss": 3.1485,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004894827878653929,
+      "loss": 3.1206,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004894608839649337,
+      "loss": 3.1654,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004894389577699994,
+      "loss": 2.8907,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000489417009282631,
+      "loss": 3.0343,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004893950385048723,
+      "loss": 3.0267,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004893730454387686,
+      "loss": 3.2202,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004893510300863676,
+      "loss": 3.193,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004893289924497191,
+      "loss": 2.9206,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004893069325308747,
+      "loss": 3.0309,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004892848503318883,
+      "loss": 3.3482,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004892627458548159,
+      "loss": 2.8198,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004892406191017154,
+      "loss": 3.1597,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004892184700746468,
+      "loss": 3.1458,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004891962987756724,
+      "loss": 3.0345,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004891741052068563,
+      "loss": 2.8913,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004891518893702648,
+      "loss": 3.0624,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004891296512679662,
+      "loss": 2.9053,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004891073909020311,
+      "loss": 3.2977,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004890851082745319,
+      "loss": 2.9565,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004890628033875431,
+      "loss": 3.173,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004890404762431414,
+      "loss": 2.9383,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004890181268434056,
+      "loss": 3.0766,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004889957551904164,
+      "loss": 2.968,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004889733612862567,
+      "loss": 3.086,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004889509451330114,
+      "loss": 2.8443,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004889285067327675,
+      "loss": 3.116,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004889060460876142,
+      "loss": 3.1185,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004888835631996426,
+      "loss": 2.9097,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004888610580709457,
+      "loss": 3.202,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004888385307036191,
+      "loss": 3.0869,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048881598109976,
+      "loss": 2.9147,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004887934092614679,
+      "loss": 3.079,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004887708151908441,
+      "loss": 3.0847,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004887481988899925,
+      "loss": 2.9887,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004887255603610184,
+      "loss": 3.0142,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004887028996060299,
+      "loss": 2.9507,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004886802166271364,
+      "loss": 3.0833,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 3.0239622592926025,
+      "eval_runtime": 184.1366,
+      "eval_samples_per_second": 457.291,
+      "eval_steps_per_second": 28.582,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048865751142645,
+      "loss": 3.3002,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004886347840060845,
+      "loss": 3.0003,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004886120343681558,
+      "loss": 3.0236,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004885892625147821,
+      "loss": 2.8006,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004885664684480837,
+      "loss": 2.9491,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004885436521701824,
+      "loss": 3.1643,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004885208136832026,
+      "loss": 3.0507,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004884979529892708,
+      "loss": 2.9363,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004884750700905152,
+      "loss": 3.1662,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004884521649890664,
+      "loss": 3.0903,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004884292376870567,
+      "loss": 3.1392,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000488406288186621,
+      "loss": 3.2029,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004883833164898957,
+      "loss": 3.1804,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004883603225990197,
+      "loss": 3.0509,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004883373065161337,
+      "loss": 2.8329,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004883142682433806,
+      "loss": 3.064,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004882912077829053,
+      "loss": 2.928,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048826812513685485,
+      "loss": 2.9288,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004882450203073783,
+      "loss": 3.147,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004882218932966267,
+      "loss": 2.9938,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048819874410675334,
+      "loss": 2.8341,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048817557273991336,
+      "loss": 3.1191,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048815237919826415,
+      "loss": 2.9767,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048812916348396515,
+      "loss": 2.8859,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048810592559917767,
+      "loss": 2.9857,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048808266554606534,
+      "loss": 3.0153,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048805938332679364,
+      "loss": 2.8181,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048803607894353033,
+      "loss": 2.8903,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000488012752398445,
+      "loss": 2.9674,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004879894036937094,
+      "loss": 3.1624,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048796603283149755,
+      "loss": 3.0267,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004879426398139851,
+      "loss": 2.8955,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048791922464335016,
+      "loss": 2.9084,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048789578732177263,
+      "loss": 3.1493,
+      "step": 1034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004878723278514348,
+      "loss": 2.9193,
+      "step": 1035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004878488462345206,
+      "loss": 2.8076,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004878253424732163,
+      "loss": 3.1054,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004878018165697101,
+      "loss": 3.1642,
+      "step": 1038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048777826852619256,
+      "loss": 3.1284,
+      "step": 1039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048775469834485587,
+      "loss": 2.8045,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048773110602789445,
+      "loss": 3.1575,
+      "step": 1041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048770749157750487,
+      "loss": 2.8781,
+      "step": 1042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048768385499588576,
+      "loss": 2.9303,
+      "step": 1043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048766019628523774,
+      "loss": 3.113,
+      "step": 1044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048763651544776336,
+      "loss": 2.9449,
+      "step": 1045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048761281248566756,
+      "loss": 3.0389,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048758908740115703,
+      "loss": 2.9384,
+      "step": 1047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004875653401964407,
+      "loss": 2.9085,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048754157087372945,
+      "loss": 3.0174,
+      "step": 1049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004875177794352363,
+      "loss": 2.9659,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 3.0027263164520264,
+      "eval_runtime": 184.0945,
+      "eval_samples_per_second": 457.396,
+      "eval_steps_per_second": 28.589,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004874939658831764,
+      "loss": 2.4932,
+      "step": 1051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048747013021976666,
+      "loss": 3.0751,
+      "step": 1052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004874462724472264,
+      "loss": 3.0594,
+      "step": 1053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048742239256777673,
+      "loss": 2.958,
+      "step": 1054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000487398490583641,
+      "loss": 2.9436,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004873745664970446,
+      "loss": 2.8123,
+      "step": 1056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004873506203102148,
+      "loss": 3.142,
+      "step": 1057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048732665202538117,
+      "loss": 2.96,
+      "step": 1058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004873026616447752,
+      "loss": 3.2119,
+      "step": 1059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048727864917063034,
+      "loss": 2.7932,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004872546146051825,
+      "loss": 3.1803,
+      "step": 1061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004872305579506691,
+      "loss": 3.0971,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048720647920932994,
+      "loss": 3.2079,
+      "step": 1063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048718237838340687,
+      "loss": 3.1766,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004871582554751437,
+      "loss": 3.0276,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048713411048678635,
+      "loss": 3.1083,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004871099434205829,
+      "loss": 2.8903,
+      "step": 1067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004870857542787832,
+      "loss": 3.0656,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048706154306363936,
+      "loss": 3.0413,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048703730977740556,
+      "loss": 3.0342,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000487013054422338,
+      "loss": 2.9647,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048698877700069495,
+      "loss": 2.9838,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004869644775147366,
+      "loss": 3.1212,
+      "step": 1073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004869401559667253,
+      "loss": 3.0476,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048691581235892555,
+      "loss": 2.8983,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048689144669360375,
+      "loss": 2.8602,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048686705897302843,
+      "loss": 3.0661,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004868426491994702,
+      "loss": 3.113,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048681821737520147,
+      "loss": 2.739,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004867937635024972,
+      "loss": 3.0827,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048676928758363395,
+      "loss": 3.1751,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048674478962089054,
+      "loss": 2.9852,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048672026961654773,
+      "loss": 2.9441,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004866957275728885,
+      "loss": 2.8723,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048667116349219774,
+      "loss": 3.1621,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004866465773767624,
+      "loss": 2.9273,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004866219692288716,
+      "loss": 2.9679,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004865973390508163,
+      "loss": 3.0303,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004865726868448898,
+      "loss": 2.9065,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048654801261338717,
+      "loss": 2.894,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048652331635860564,
+      "loss": 2.8336,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048649859808284456,
+      "loss": 2.9301,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004864738577884053,
+      "loss": 2.9467,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004864490954775912,
+      "loss": 2.8952,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004864243111527076,
+      "loss": 3.1197,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004863995048160622,
+      "loss": 3.1009,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004863746764699644,
+      "loss": 2.9879,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048634982611672577,
+      "loss": 3.0995,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048632495375866005,
+      "loss": 3.1375,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048630005939808286,
+      "loss": 2.9136,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.964057683944702,
+      "eval_runtime": 183.7978,
+      "eval_samples_per_second": 458.134,
+      "eval_steps_per_second": 28.635,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000486275143037312,
+      "loss": 3.1252,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004862502046786671,
+      "loss": 3.0731,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004862252443244702,
+      "loss": 2.8396,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000486200261977045,
+      "loss": 3.0309,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048617525763871753,
+      "loss": 2.8547,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004861502313118157,
+      "loss": 2.7592,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048612518299866956,
+      "loss": 3.0029,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004861001127016111,
+      "loss": 2.9333,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048607502042297466,
+      "loss": 3.0921,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048604990616509615,
+      "loss": 3.0177,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004860247699303139,
+      "loss": 3.0213,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004859996117209682,
+      "loss": 2.8351,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048597443153940124,
+      "loss": 2.9682,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048594922938795736,
+      "loss": 2.7993,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004859240052689831,
+      "loss": 2.9271,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004858987591848268,
+      "loss": 3.0047,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004858734911378389,
+      "loss": 3.1523,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000485848201130372,
+      "loss": 2.7329,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048582288916478066,
+      "loss": 3.1417,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004857975552434214,
+      "loss": 2.9499,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000485772199368653,
+      "loss": 2.9462,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004857468215428361,
+      "loss": 2.8297,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004857214217683336,
+      "loss": 2.9586,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048569600004751,
+      "loss": 3.1029,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048567055638273226,
+      "loss": 2.9691,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004856450907763693,
+      "loss": 3.1339,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048561960323079215,
+      "loss": 2.9504,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004855940937483735,
+      "loss": 2.7444,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004855685623314885,
+      "loss": 2.7535,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004855430089825143,
+      "loss": 3.0818,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048551743370382983,
+      "loss": 3.0392,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004854918364978163,
+      "loss": 2.9654,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004854662173668567,
+      "loss": 2.9016,
+      "step": 1133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004854405763133365,
+      "loss": 2.9242,
+      "step": 1134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004854149133396429,
+      "loss": 2.9941,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048538922844816503,
+      "loss": 2.8206,
+      "step": 1136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004853635216412944,
+      "loss": 2.9792,
+      "step": 1137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048533779292142434,
+      "loss": 2.8435,
+      "step": 1138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048531204229095015,
+      "loss": 3.0903,
+      "step": 1139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048528626975226954,
+      "loss": 2.8562,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048526047530778174,
+      "loss": 2.98,
+      "step": 1141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048523465895988836,
+      "loss": 3.1212,
+      "step": 1142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048520882071099303,
+      "loss": 2.8026,
+      "step": 1143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004851829605635014,
+      "loss": 3.016,
+      "step": 1144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004851570785198211,
+      "loss": 3.1102,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004851311745823616,
+      "loss": 2.8068,
+      "step": 1146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004851052487535349,
+      "loss": 3.0342,
+      "step": 1147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004850793010357547,
+      "loss": 2.9767,
+      "step": 1148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048505333143143676,
+      "loss": 2.7906,
+      "step": 1149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048502733994299895,
+      "loss": 2.9501,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.9362847805023193,
+      "eval_runtime": 184.0542,
+      "eval_samples_per_second": 457.496,
+      "eval_steps_per_second": 28.595,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004850013265728612,
+      "loss": 2.9616,
+      "step": 1151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048497529132344525,
+      "loss": 2.6951,
+      "step": 1152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004849492341971753,
+      "loss": 2.8032,
+      "step": 1153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004849231551964771,
+      "loss": 3.1684,
+      "step": 1154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048489705432377885,
+      "loss": 2.7924,
+      "step": 1155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048487093158151054,
+      "loss": 3.0276,
+      "step": 1156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004848447869721043,
+      "loss": 2.8517,
+      "step": 1157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048481862049799433,
+      "loss": 2.9323,
+      "step": 1158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004847924321616166,
+      "loss": 2.9357,
+      "step": 1159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048476622196540946,
+      "loss": 2.9098,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048473998991181326,
+      "loss": 3.0552,
+      "step": 1161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048471373600326995,
+      "loss": 2.7781,
+      "step": 1162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004846874602422242,
+      "loss": 2.8433,
+      "step": 1163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048466116263112217,
+      "loss": 2.9662,
+      "step": 1164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004846348431724122,
+      "loss": 2.834,
+      "step": 1165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048460850186854486,
+      "loss": 3.0633,
+      "step": 1166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048458213872197244,
+      "loss": 2.9248,
+      "step": 1167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004845557537351496,
+      "loss": 3.042,
+      "step": 1168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004845293469105326,
+      "loss": 3.0226,
+      "step": 1169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048450291825058027,
+      "loss": 2.9602,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048447646775775306,
+      "loss": 2.8651,
+      "step": 1171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004844499954345135,
+      "loss": 2.9485,
+      "step": 1172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004844235012833265,
+      "loss": 2.853,
+      "step": 1173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048439698530665837,
+      "loss": 2.8594,
+      "step": 1174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004843704475069781,
+      "loss": 2.9262,
+      "step": 1175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048434388788675634,
+      "loss": 2.9041,
+      "step": 1176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048431730644846584,
+      "loss": 2.9516,
+      "step": 1177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004842907031945815,
+      "loss": 2.9406,
+      "step": 1178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004842640781275801,
+      "loss": 3.036,
+      "step": 1179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004842374312499405,
+      "loss": 3.0766,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048421076256414356,
+      "loss": 2.8095,
+      "step": 1181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048418407207267224,
+      "loss": 3.0198,
+      "step": 1182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048415735977801156,
+      "loss": 2.9415,
+      "step": 1183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004841306256826484,
+      "loss": 3.0937,
+      "step": 1184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048410386978907187,
+      "loss": 2.8727,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048407709209977303,
+      "loss": 3.1855,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048405029261724486,
+      "loss": 2.857,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004840234713439825,
+      "loss": 2.9457,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048399662828248316,
+      "loss": 2.7817,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048396976343524596,
+      "loss": 2.9927,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004839428768047721,
+      "loss": 3.0008,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004839159683935647,
+      "loss": 2.7393,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004838890382041291,
+      "loss": 3.085,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048386208623897267,
+      "loss": 2.8714,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004838351125006045,
+      "loss": 2.9129,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048380811699153605,
+      "loss": 2.9342,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048378109971428064,
+      "loss": 3.1137,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048375406067135377,
+      "loss": 2.7957,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048372699986527263,
+      "loss": 2.8564,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048369991729855686,
+      "loss": 3.0921,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.914381742477417,
+      "eval_runtime": 183.8767,
+      "eval_samples_per_second": 457.937,
+      "eval_steps_per_second": 28.622,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048367281297372777,
+      "loss": 2.907,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048364568689330894,
+      "loss": 2.6085,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004836185390598259,
+      "loss": 2.7835,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004835913694758061,
+      "loss": 2.9767,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048356417814377915,
+      "loss": 2.7773,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004835369650662767,
+      "loss": 3.0573,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048350973024583225,
+      "loss": 2.7118,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048348247368498157,
+      "loss": 2.9137,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048345519538626217,
+      "loss": 2.5974,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048342789535221384,
+      "loss": 2.8146,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048340057358537824,
+      "loss": 3.0021,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048337323008829916,
+      "loss": 2.7809,
+      "step": 1212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048334586486352226,
+      "loss": 2.9382,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004833184779135954,
+      "loss": 3.0665,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048329106924106834,
+      "loss": 2.8027,
+      "step": 1215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000483263638848493,
+      "loss": 2.6974,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048323618673842307,
+      "loss": 2.9519,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004832087129134145,
+      "loss": 3.0347,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048318121737602523,
+      "loss": 2.887,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004831537001288151,
+      "loss": 3.0672,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048312616117434607,
+      "loss": 2.7205,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000483098600515182,
+      "loss": 2.7573,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000483071018153889,
+      "loss": 2.9179,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000483043414093035,
+      "loss": 2.8944,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048301578833519004,
+      "loss": 2.9748,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048298814088292615,
+      "loss": 2.8769,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048296047173881733,
+      "loss": 2.9335,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004829327809054398,
+      "loss": 2.8778,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004829050683853715,
+      "loss": 2.9158,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048287733418119253,
+      "loss": 2.9798,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048284957829548516,
+      "loss": 2.9174,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004828218007308335,
+      "loss": 2.896,
+      "step": 1232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048279400148982355,
+      "loss": 2.9182,
+      "step": 1233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004827661805750437,
+      "loss": 2.7234,
+      "step": 1234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004827383379890842,
+      "loss": 2.8648,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000482710473734537,
+      "loss": 2.8742,
+      "step": 1236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048268258781399665,
+      "loss": 2.9494,
+      "step": 1237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048265468023005913,
+      "loss": 2.7489,
+      "step": 1238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004826267509853229,
+      "loss": 2.8471,
+      "step": 1239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004825988000823882,
+      "loss": 2.8026,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048257082752385726,
+      "loss": 3.0127,
+      "step": 1241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048254283331233464,
+      "loss": 2.8845,
+      "step": 1242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004825148174504264,
+      "loss": 2.8911,
+      "step": 1243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048248677994074097,
+      "loss": 2.8088,
+      "step": 1244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004824587207858888,
+      "loss": 2.9314,
+      "step": 1245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004824306399884822,
+      "loss": 2.9267,
+      "step": 1246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004824025375511356,
+      "loss": 3.0613,
+      "step": 1247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048237441347646545,
+      "loss": 2.7344,
+      "step": 1248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004823462677670901,
+      "loss": 2.9466,
+      "step": 1249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004823181004256301,
+      "loss": 2.8951,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.8847079277038574,
+      "eval_runtime": 184.1372,
+      "eval_samples_per_second": 457.289,
+      "eval_steps_per_second": 28.582,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048228991145470777,
+      "loss": 2.9102,
+      "step": 1251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004822617008569478,
+      "loss": 3.0138,
+      "step": 1252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004822334686349764,
+      "loss": 2.8456,
+      "step": 1253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004822052147914223,
+      "loss": 2.9283,
+      "step": 1254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004821769393289158,
+      "loss": 2.8224,
+      "step": 1255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004821486422500897,
+      "loss": 2.9867,
+      "step": 1256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004821203235575783,
+      "loss": 3.1252,
+      "step": 1257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048209198325401817,
+      "loss": 2.9127,
+      "step": 1258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048206362134204793,
+      "loss": 2.7777,
+      "step": 1259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004820352378243082,
+      "loss": 2.9192,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004820068327034415,
+      "loss": 2.9872,
+      "step": 1261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004819784059820925,
+      "loss": 2.9366,
+      "step": 1262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048194995766290765,
+      "loss": 3.0385,
+      "step": 1263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048192148774853574,
+      "loss": 2.8384,
+      "step": 1264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004818929962416273,
+      "loss": 3.0328,
+      "step": 1265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000481864483144835,
+      "loss": 2.8951,
+      "step": 1266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004818359484608135,
+      "loss": 2.6742,
+      "step": 1267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048180739219221935,
+      "loss": 2.7761,
+      "step": 1268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048177881434171134,
+      "loss": 3.0702,
+      "step": 1269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048175021491195013,
+      "loss": 2.8411,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004817215939055984,
+      "loss": 2.9851,
+      "step": 1271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048169295132532086,
+      "loss": 2.8646,
+      "step": 1272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004816642871737842,
+      "loss": 2.8547,
+      "step": 1273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000481635601453657,
+      "loss": 2.6881,
+      "step": 1274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048160689416761027,
+      "loss": 2.9381,
+      "step": 1275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004815781653183165,
+      "loss": 2.9884,
+      "step": 1276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004815494149084505,
+      "loss": 2.8638,
+      "step": 1277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048152064294068895,
+      "loss": 2.9391,
+      "step": 1278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004814918494177107,
+      "loss": 2.9855,
+      "step": 1279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004814630343421966,
+      "loss": 2.9166,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004814341977168291,
+      "loss": 3.0547,
+      "step": 1281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004814053395442932,
+      "loss": 2.8231,
+      "step": 1282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048137645982727575,
+      "loss": 2.893,
+      "step": 1283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004813475585684652,
+      "loss": 2.9534,
+      "step": 1284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048131863577055274,
+      "loss": 2.9448,
+      "step": 1285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004812896914362309,
+      "loss": 3.1012,
+      "step": 1286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048126072556819457,
+      "loss": 2.8089,
+      "step": 1287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048123173816914044,
+      "loss": 2.9071,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004812027292417676,
+      "loss": 2.8881,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048117369878877646,
+      "loss": 2.7923,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048114464681287016,
+      "loss": 3.0024,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004811155733167534,
+      "loss": 2.7133,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048108647830313303,
+      "loss": 2.78,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048105736177471783,
+      "loss": 2.8885,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048102822373421864,
+      "loss": 3.0001,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048099906418434834,
+      "loss": 2.8665,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004809698831278217,
+      "loss": 2.6989,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048094068056735563,
+      "loss": 2.7903,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004809114565056689,
+      "loss": 2.8648,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048088221094548246,
+      "loss": 2.8615,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.8680028915405273,
+      "eval_runtime": 184.1301,
+      "eval_samples_per_second": 457.307,
+      "eval_steps_per_second": 28.583,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000480852943889519,
+      "loss": 2.7408,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004808236553405035,
+      "loss": 2.8016,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004807943453011627,
+      "loss": 2.7674,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048076501377422554,
+      "loss": 2.9727,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004807356607624228,
+      "loss": 2.872,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048070628626848734,
+      "loss": 2.806,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048067689029515404,
+      "loss": 2.8904,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004806474728451597,
+      "loss": 2.8284,
+      "step": 1308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004806180339212432,
+      "loss": 2.9378,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048058857352614536,
+      "loss": 2.9125,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000480559091662609,
+      "loss": 3.0018,
+      "step": 1311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048052958833337913,
+      "loss": 2.8521,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004805000635412023,
+      "loss": 2.9442,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004804705172888277,
+      "loss": 2.8086,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048044094957900587,
+      "loss": 3.0773,
+      "step": 1315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004804113604144898,
+      "loss": 2.9204,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004803817497980343,
+      "loss": 2.7513,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004803521177323962,
+      "loss": 3.0496,
+      "step": 1318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004803224642203342,
+      "loss": 2.8247,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004802927892646093,
+      "loss": 2.9231,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048026309286798433,
+      "loss": 2.8448,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004802333750332239,
+      "loss": 2.8472,
+      "step": 1322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004802036357630951,
+      "loss": 2.9439,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004801738750603665,
+      "loss": 2.8633,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000480144092927809,
+      "loss": 2.7689,
+      "step": 1325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048011428936819544,
+      "loss": 2.7849,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048008446438430063,
+      "loss": 2.84,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004800546179789013,
+      "loss": 2.9033,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004800247501547761,
+      "loss": 2.7909,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004799948609147061,
+      "loss": 2.8998,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047996495026147377,
+      "loss": 2.7944,
+      "step": 1331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047993501819786413,
+      "loss": 2.588,
+      "step": 1332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047990506472666384,
+      "loss": 2.9156,
+      "step": 1333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047987508985066163,
+      "loss": 2.9368,
+      "step": 1334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047984509357264823,
+      "loss": 2.9702,
+      "step": 1335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004798150758954164,
+      "loss": 2.9116,
+      "step": 1336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004797850368217609,
+      "loss": 3.0425,
+      "step": 1337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047975497635447843,
+      "loss": 2.8779,
+      "step": 1338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004797248944963677,
+      "loss": 2.821,
+      "step": 1339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004796947912502294,
+      "loss": 2.6453,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004796646666188662,
+      "loss": 2.5651,
+      "step": 1341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004796345206050829,
+      "loss": 2.7922,
+      "step": 1342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047960435321168604,
+      "loss": 2.9062,
+      "step": 1343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004795741644414844,
+      "loss": 2.8965,
+      "step": 1344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047954395429728857,
+      "loss": 2.8543,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047951372278191124,
+      "loss": 2.9121,
+      "step": 1346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047948346989816704,
+      "loss": 2.7625,
+      "step": 1347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047945319564887256,
+      "loss": 2.727,
+      "step": 1348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004794229000368464,
+      "loss": 2.7585,
+      "step": 1349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047939258306490927,
+      "loss": 2.886,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.839277744293213,
+      "eval_runtime": 183.8873,
+      "eval_samples_per_second": 457.911,
+      "eval_steps_per_second": 28.621,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004793622447358837,
+      "loss": 2.6913,
+      "step": 1351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004793318850525943,
+      "loss": 2.9237,
+      "step": 1352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047930150401786755,
+      "loss": 2.9038,
+      "step": 1353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047927110163453214,
+      "loss": 2.7435,
+      "step": 1354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047924067790541846,
+      "loss": 2.7537,
+      "step": 1355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047921023283335924,
+      "loss": 2.7823,
+      "step": 1356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047917976642118887,
+      "loss": 2.8051,
+      "step": 1357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047914927867174383,
+      "loss": 2.8283,
+      "step": 1358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047911876958786273,
+      "loss": 2.7411,
+      "step": 1359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047908823917238596,
+      "loss": 2.9444,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047905768742815603,
+      "loss": 2.7143,
+      "step": 1361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004790271143580174,
+      "loss": 2.8489,
+      "step": 1362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004789965199648165,
+      "loss": 2.8167,
+      "step": 1363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047896590425140176,
+      "loss": 2.6683,
+      "step": 1364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004789352672206235,
+      "loss": 2.7956,
+      "step": 1365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047890460887533415,
+      "loss": 2.8257,
+      "step": 1366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047887392921838826,
+      "loss": 2.7831,
+      "step": 1367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047884322825264195,
+      "loss": 2.7635,
+      "step": 1368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004788125059809536,
+      "loss": 3.0056,
+      "step": 1369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047878176240618377,
+      "loss": 2.7413,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004787509975311945,
+      "loss": 3.0077,
+      "step": 1371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047872021135885026,
+      "loss": 2.9465,
+      "step": 1372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004786894038920172,
+      "loss": 2.8326,
+      "step": 1373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047865857513356363,
+      "loss": 2.7185,
+      "step": 1374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004786277250863599,
+      "loss": 2.7267,
+      "step": 1375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047859685375327806,
+      "loss": 3.0757,
+      "step": 1376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047856596113719243,
+      "loss": 2.7423,
+      "step": 1377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004785350472409792,
+      "loss": 2.9967,
+      "step": 1378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047850411206751644,
+      "loss": 2.8573,
+      "step": 1379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004784731556196844,
+      "loss": 2.9615,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047844217790036516,
+      "loss": 2.8594,
+      "step": 1381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004784111789124429,
+      "loss": 2.9959,
+      "step": 1382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047838015865880364,
+      "loss": 2.8522,
+      "step": 1383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004783491171423354,
+      "loss": 2.7746,
+      "step": 1384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047831805436592846,
+      "loss": 2.8071,
+      "step": 1385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004782869703324746,
+      "loss": 2.6624,
+      "step": 1386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047825586504486795,
+      "loss": 2.7461,
+      "step": 1387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047822473850600447,
+      "loss": 2.9024,
+      "step": 1388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004781935907187821,
+      "loss": 2.84,
+      "step": 1389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047816242168610093,
+      "loss": 2.8566,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047813123141086266,
+      "loss": 2.8443,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004781000198959714,
+      "loss": 2.6186,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004780687871443329,
+      "loss": 2.739,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000478037533158855,
+      "loss": 2.7628,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004780062579424476,
+      "loss": 2.8273,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047797496149802257,
+      "loss": 2.7629,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047794364382849354,
+      "loss": 2.8849,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004779123049367764,
+      "loss": 2.8596,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004778809448257888,
+      "loss": 2.8705,
+      "step": 1399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047784956349845056,
+      "loss": 2.6634,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.8210883140563965,
+      "eval_runtime": 184.0107,
+      "eval_samples_per_second": 457.604,
+      "eval_steps_per_second": 28.602,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047781816095768313,
+      "loss": 2.8323,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004777867372064105,
+      "loss": 2.6888,
+      "step": 1402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004777552922475582,
+      "loss": 2.8175,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047772382608405375,
+      "loss": 2.7413,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047769233871882674,
+      "loss": 2.9077,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047766083015480875,
+      "loss": 2.8838,
+      "step": 1406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047762930039493346,
+      "loss": 2.7398,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004775977494421362,
+      "loss": 2.7814,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047756617729935447,
+      "loss": 2.748,
+      "step": 1409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047753458396952784,
+      "loss": 2.8489,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047750296945559765,
+      "loss": 2.9637,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004774713337605073,
+      "loss": 2.8156,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004774396768872021,
+      "loss": 2.7541,
+      "step": 1413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004774079988386296,
+      "loss": 2.7002,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047737629961773894,
+      "loss": 2.7554,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004773445792274814,
+      "loss": 2.9012,
+      "step": 1416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047731283767081035,
+      "loss": 2.9614,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047728107495068094,
+      "loss": 2.8147,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047724929107005034,
+      "loss": 2.6929,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004772174860318778,
+      "loss": 2.8497,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047718565983912445,
+      "loss": 2.7305,
+      "step": 1421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047715381249475324,
+      "loss": 2.8589,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004771219440017295,
+      "loss": 2.8701,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047709005436302005,
+      "loss": 2.8319,
+      "step": 1424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047705814358159403,
+      "loss": 2.6835,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004770262116604224,
+      "loss": 2.7708,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047699425860247814,
+      "loss": 2.9186,
+      "step": 1427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047696228441073606,
+      "loss": 2.7748,
+      "step": 1428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047693028908817313,
+      "loss": 2.8907,
+      "step": 1429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004768982726377682,
+      "loss": 2.9468,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004768662350625022,
+      "loss": 2.8624,
+      "step": 1431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047683417636535763,
+      "loss": 2.9101,
+      "step": 1432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004768020965493195,
+      "loss": 2.7173,
+      "step": 1433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047676999561737445,
+      "loss": 2.7889,
+      "step": 1434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047673787357251107,
+      "loss": 2.8539,
+      "step": 1435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047670573041772013,
+      "loss": 2.8065,
+      "step": 1436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004766735661559943,
+      "loss": 2.8068,
+      "step": 1437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000476641380790328,
+      "loss": 2.9411,
+      "step": 1438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047660917432371797,
+      "loss": 2.8269,
+      "step": 1439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047657694675916254,
+      "loss": 2.8818,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047654469809966225,
+      "loss": 2.5593,
+      "step": 1441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047651242834821953,
+      "loss": 2.8152,
+      "step": 1442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004764801375078388,
+      "loss": 2.9335,
+      "step": 1443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047644782558152644,
+      "loss": 2.6525,
+      "step": 1444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004764154925722908,
+      "loss": 2.8531,
+      "step": 1445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004763831384831421,
+      "loss": 2.8021,
+      "step": 1446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047635076331709264,
+      "loss": 2.9016,
+      "step": 1447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047631836707715664,
+      "loss": 2.9895,
+      "step": 1448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004762859497663503,
+      "loss": 2.7754,
+      "step": 1449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004762535113876917,
+      "loss": 2.7992,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.8028955459594727,
+      "eval_runtime": 183.8805,
+      "eval_samples_per_second": 457.928,
+      "eval_steps_per_second": 28.622,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047622105194420104,
+      "loss": 2.9302,
+      "step": 1451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047618857143890027,
+      "loss": 2.9567,
+      "step": 1452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004761560698748135,
+      "loss": 2.8921,
+      "step": 1453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047612354725496667,
+      "loss": 2.6128,
+      "step": 1454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047609100358238774,
+      "loss": 2.7448,
+      "step": 1455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004760584388601066,
+      "loss": 2.7767,
+      "step": 1456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047602585309115516,
+      "loss": 2.7268,
+      "step": 1457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004759932462785672,
+      "loss": 2.8623,
+      "step": 1458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047596061842537855,
+      "loss": 2.8801,
+      "step": 1459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047592796953462697,
+      "loss": 2.6925,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004758952996093521,
+      "loss": 2.8133,
+      "step": 1461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004758626086525956,
+      "loss": 2.7691,
+      "step": 1462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004758298966674011,
+      "loss": 2.7866,
+      "step": 1463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047579716365681425,
+      "loss": 2.8498,
+      "step": 1464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047576440962388244,
+      "loss": 2.7752,
+      "step": 1465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004757316345716554,
+      "loss": 3.0245,
+      "step": 1466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047569883850318424,
+      "loss": 2.9426,
+      "step": 1467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004756660214215226,
+      "loss": 2.7106,
+      "step": 1468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047563318332972585,
+      "loss": 2.8198,
+      "step": 1469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004756003242308512,
+      "loss": 2.6782,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000475567444127958,
+      "loss": 2.8727,
+      "step": 1471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047553454302410746,
+      "loss": 2.7967,
+      "step": 1472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004755016209223628,
+      "loss": 2.7425,
+      "step": 1473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004754686778257891,
+      "loss": 2.7427,
+      "step": 1474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047543571373745343,
+      "loss": 2.7978,
+      "step": 1475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004754027286604249,
+      "loss": 2.84,
+      "step": 1476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047536972259777443,
+      "loss": 2.8485,
+      "step": 1477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004753366955525751,
+      "loss": 2.6256,
+      "step": 1478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004753036475279018,
+      "loss": 2.7325,
+      "step": 1479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004752705785268313,
+      "loss": 2.8124,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004752374885524424,
+      "loss": 2.938,
+      "step": 1481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047520437760781596,
+      "loss": 2.7113,
+      "step": 1482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047517124569603473,
+      "loss": 2.67,
+      "step": 1483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004751380928201834,
+      "loss": 2.8154,
+      "step": 1484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004751049189833484,
+      "loss": 2.7512,
+      "step": 1485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047507172418861856,
+      "loss": 2.7352,
+      "step": 1486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047503850843908413,
+      "loss": 2.9132,
+      "step": 1487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004750052717378378,
+      "loss": 2.8089,
+      "step": 1488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047497201408797394,
+      "loss": 2.8373,
+      "step": 1489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000474938735492589,
+      "loss": 2.6315,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004749054359547812,
+      "loss": 2.9288,
+      "step": 1491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004748721154776508,
+      "loss": 2.4851,
+      "step": 1492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004748387740643001,
+      "loss": 2.7034,
+      "step": 1493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004748054117178333,
+      "loss": 2.8128,
+      "step": 1494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047477202844135646,
+      "loss": 2.7105,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047473862423797776,
+      "loss": 2.8046,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004747051991108071,
+      "loss": 2.8342,
+      "step": 1497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004746717530629565,
+      "loss": 2.8038,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004746382860975399,
+      "loss": 2.67,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047460479821767316,
+      "loss": 2.7284,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.78027606010437,
+      "eval_runtime": 184.0991,
+      "eval_samples_per_second": 457.384,
+      "eval_steps_per_second": 28.588,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004745712894264741,
+      "loss": 2.6865,
+      "step": 1501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004745377597270625,
+      "loss": 2.9565,
+      "step": 1502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004745042091225599,
+      "loss": 2.8617,
+      "step": 1503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004744706376160902,
+      "loss": 2.5719,
+      "step": 1504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004744370452107789,
+      "loss": 2.9042,
+      "step": 1505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047440343190975355,
+      "loss": 2.8975,
+      "step": 1506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047436979771614356,
+      "loss": 2.7136,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004743361426330805,
+      "loss": 2.6993,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004743024666636977,
+      "loss": 2.8607,
+      "step": 1509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047426876981113044,
+      "loss": 2.8524,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000474235052078516,
+      "loss": 2.6551,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004742013134689937,
+      "loss": 2.987,
+      "step": 1512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004741675539857046,
+      "loss": 2.7399,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047413377363179183,
+      "loss": 2.6835,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047409997241040037,
+      "loss": 2.5671,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004740661503246774,
+      "loss": 2.7139,
+      "step": 1516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004740323073777716,
+      "loss": 2.7865,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047399844357283395,
+      "loss": 2.8076,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004739645589130174,
+      "loss": 2.6396,
+      "step": 1519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047393065340147644,
+      "loss": 2.6613,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000473896727041368,
+      "loss": 2.8415,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004738627798358506,
+      "loss": 2.7691,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047382881178808486,
+      "loss": 2.8398,
+      "step": 1523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004737948229012332,
+      "loss": 2.829,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004737608131784603,
+      "loss": 2.6935,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047372678262293233,
+      "loss": 2.5431,
+      "step": 1526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047369273123781775,
+      "loss": 2.8524,
+      "step": 1527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047365865902628684,
+      "loss": 2.8328,
+      "step": 1528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004736245659915118,
+      "loss": 2.6564,
+      "step": 1529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004735904521366667,
+      "loss": 2.7468,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047355631746492786,
+      "loss": 2.7718,
+      "step": 1531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047352216197947306,
+      "loss": 2.8201,
+      "step": 1532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004734879856834824,
+      "loss": 2.8841,
+      "step": 1533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047345378858013775,
+      "loss": 2.8473,
+      "step": 1534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047341957067262297,
+      "loss": 2.7575,
+      "step": 1535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047338533196412394,
+      "loss": 2.6818,
+      "step": 1536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047335107245782815,
+      "loss": 2.6209,
+      "step": 1537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047331679215692544,
+      "loss": 2.7127,
+      "step": 1538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047328249106460743,
+      "loss": 2.4853,
+      "step": 1539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004732481691840674,
+      "loss": 2.6223,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004732138265185011,
+      "loss": 2.6844,
+      "step": 1541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004731794630711058,
+      "loss": 2.7382,
+      "step": 1542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047314507884508087,
+      "loss": 2.7263,
+      "step": 1543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004731106738436275,
+      "loss": 2.6332,
+      "step": 1544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004730762480699489,
+      "loss": 2.7659,
+      "step": 1545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004730418015272503,
+      "loss": 2.7931,
+      "step": 1546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047300733421873874,
+      "loss": 2.7278,
+      "step": 1547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047297284614762316,
+      "loss": 2.6467,
+      "step": 1548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047293833731711457,
+      "loss": 2.9704,
+      "step": 1549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004729038077304257,
+      "loss": 2.5383,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.757880449295044,
+      "eval_runtime": 183.9997,
+      "eval_samples_per_second": 457.631,
+      "eval_steps_per_second": 28.603,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004728692573907715,
+      "loss": 2.6962,
+      "step": 1551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004728346863013687,
+      "loss": 2.6346,
+      "step": 1552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004728000944654359,
+      "loss": 2.6784,
+      "step": 1553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047276548188619375,
+      "loss": 2.8152,
+      "step": 1554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047273084856686465,
+      "loss": 2.7796,
+      "step": 1555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004726961945106732,
+      "loss": 2.722,
+      "step": 1556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004726615197208457,
+      "loss": 2.7657,
+      "step": 1557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004726268242006106,
+      "loss": 2.6028,
+      "step": 1558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047259210795319795,
+      "loss": 2.96,
+      "step": 1559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047255737098184016,
+      "loss": 2.6808,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047252261328977114,
+      "loss": 2.6713,
+      "step": 1561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000472487834880227,
+      "loss": 2.6128,
+      "step": 1562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004724530357564457,
+      "loss": 2.6857,
+      "step": 1563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004724182159216672,
+      "loss": 2.7471,
+      "step": 1564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047238337537913326,
+      "loss": 2.8211,
+      "step": 1565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004723485141320877,
+      "loss": 2.699,
+      "step": 1566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047231363218377606,
+      "loss": 2.7558,
+      "step": 1567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000472278729537446,
+      "loss": 2.6793,
+      "step": 1568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004722438061963471,
+      "loss": 2.9331,
+      "step": 1569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004722088621637309,
+      "loss": 2.9083,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004721738974428506,
+      "loss": 2.889,
+      "step": 1571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004721389120369616,
+      "loss": 2.8741,
+      "step": 1572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004721039059493212,
+      "loss": 2.6035,
+      "step": 1573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004720688791831884,
+      "loss": 2.6051,
+      "step": 1574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004720338317418245,
+      "loss": 2.6127,
+      "step": 1575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047199876362849237,
+      "loss": 2.8882,
+      "step": 1576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000471963674846457,
+      "loss": 2.7828,
+      "step": 1577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004719285653989852,
+      "loss": 2.7893,
+      "step": 1578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004718934352893459,
+      "loss": 2.8756,
+      "step": 1579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004718582845208096,
+      "loss": 2.7808,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047182311309664903,
+      "loss": 2.7186,
+      "step": 1581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004717879210201389,
+      "loss": 2.8263,
+      "step": 1582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004717527082945554,
+      "loss": 2.7947,
+      "step": 1583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004717174749231772,
+      "loss": 2.8051,
+      "step": 1584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004716822209092845,
+      "loss": 2.6243,
+      "step": 1585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047164694625615946,
+      "loss": 2.7312,
+      "step": 1586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004716116509670864,
+      "loss": 2.7052,
+      "step": 1587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047157633504535135,
+      "loss": 2.7186,
+      "step": 1588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047154099849424237,
+      "loss": 2.8187,
+      "step": 1589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004715056413170493,
+      "loss": 2.7575,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047147026351706406,
+      "loss": 2.6293,
+      "step": 1591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004714348650975804,
+      "loss": 2.6665,
+      "step": 1592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000471399446061894,
+      "loss": 2.6393,
+      "step": 1593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047136400641330245,
+      "loss": 2.6248,
+      "step": 1594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004713285461551054,
+      "loss": 2.8445,
+      "step": 1595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004712930652906041,
+      "loss": 2.6645,
+      "step": 1596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047125756382310213,
+      "loss": 2.7088,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047122204175590465,
+      "loss": 2.6758,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047118649909231886,
+      "loss": 2.958,
+      "step": 1599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004711509358356539,
+      "loss": 2.9158,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.741515636444092,
+      "eval_runtime": 184.0093,
+      "eval_samples_per_second": 457.607,
+      "eval_steps_per_second": 28.602,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047111535198922084,
+      "loss": 2.6531,
+      "step": 1601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047107974755633265,
+      "loss": 2.7061,
+      "step": 1602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047104412254030413,
+      "loss": 2.5983,
+      "step": 1603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004710084769444521,
+      "loss": 2.8379,
+      "step": 1604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047097281077209524,
+      "loss": 2.8452,
+      "step": 1605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047093712402655427,
+      "loss": 2.6809,
+      "step": 1606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004709014167111516,
+      "loss": 2.7072,
+      "step": 1607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004708656888292117,
+      "loss": 2.9661,
+      "step": 1608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000470829940384061,
+      "loss": 2.6997,
+      "step": 1609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004707941713790278,
+      "loss": 2.7533,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047075838181744213,
+      "loss": 2.6653,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004707225717026363,
+      "loss": 2.6634,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004706867410379442,
+      "loss": 2.6454,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004706508898267018,
+      "loss": 2.7812,
+      "step": 1614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047061501807224706,
+      "loss": 2.7857,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004705791257779195,
+      "loss": 2.7177,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000470543212947061,
+      "loss": 2.7772,
+      "step": 1617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047050727958301505,
+      "loss": 2.5747,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047047132568912717,
+      "loss": 2.9905,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047043535126874473,
+      "loss": 3.0204,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047039935632521714,
+      "loss": 2.6845,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004703633408618955,
+      "loss": 2.7906,
+      "step": 1622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047032730488213303,
+      "loss": 2.6329,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047029124838928485,
+      "loss": 2.8245,
+      "step": 1624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004702551713867078,
+      "loss": 2.9326,
+      "step": 1625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004702190738777608,
+      "loss": 2.7417,
+      "step": 1626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004701829558658046,
+      "loss": 2.835,
+      "step": 1627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047014681735420196,
+      "loss": 2.598,
+      "step": 1628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004701106583463174,
+      "loss": 2.6588,
+      "step": 1629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047007447884551745,
+      "loss": 2.611,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004700382788551705,
+      "loss": 2.7352,
+      "step": 1631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004700020583786468,
+      "loss": 2.5434,
+      "step": 1632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004699658174193189,
+      "loss": 2.6523,
+      "step": 1633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046992955598056053,
+      "loss": 2.6141,
+      "step": 1634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046989327406574794,
+      "loss": 2.6655,
+      "step": 1635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000469856971678259,
+      "loss": 2.6807,
+      "step": 1636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004698206488214737,
+      "loss": 2.7231,
+      "step": 1637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004697843054987737,
+      "loss": 2.7444,
+      "step": 1638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004697479417135426,
+      "loss": 2.6863,
+      "step": 1639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046971155746916603,
+      "loss": 2.7001,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004696751527690315,
+      "loss": 2.5693,
+      "step": 1641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046963872761652837,
+      "loss": 2.4487,
+      "step": 1642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046960228201504793,
+      "loss": 2.8179,
+      "step": 1643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004695658159679833,
+      "loss": 2.7194,
+      "step": 1644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004695293294787297,
+      "loss": 2.6707,
+      "step": 1645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000469492822550684,
+      "loss": 2.7283,
+      "step": 1646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046945629518724513,
+      "loss": 2.5896,
+      "step": 1647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004694197473918139,
+      "loss": 2.7073,
+      "step": 1648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046938317916779304,
+      "loss": 2.773,
+      "step": 1649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004693465905185871,
+      "loss": 2.569,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.7274718284606934,
+      "eval_runtime": 184.07,
+      "eval_samples_per_second": 457.456,
+      "eval_steps_per_second": 28.592,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004693099814476026,
+      "loss": 2.7386,
+      "step": 1651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046927335195824806,
+      "loss": 2.661,
+      "step": 1652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046923670205393353,
+      "loss": 2.8299,
+      "step": 1653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046920003173807145,
+      "loss": 2.7829,
+      "step": 1654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004691633410140759,
+      "loss": 2.6207,
+      "step": 1655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046912662988536283,
+      "loss": 2.7346,
+      "step": 1656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004690898983553501,
+      "loss": 2.8379,
+      "step": 1657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004690531464274576,
+      "loss": 2.7344,
+      "step": 1658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000469016374105107,
+      "loss": 2.6217,
+      "step": 1659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000468979581391722,
+      "loss": 2.9412,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004689427682907279,
+      "loss": 2.6245,
+      "step": 1661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046890593480555234,
+      "loss": 2.6714,
+      "step": 1662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046886908093962443,
+      "loss": 2.5756,
+      "step": 1663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004688322066963755,
+      "loss": 2.946,
+      "step": 1664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004687953120792385,
+      "loss": 2.5923,
+      "step": 1665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004687583970916486,
+      "loss": 2.7786,
+      "step": 1666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046872146173704255,
+      "loss": 2.6225,
+      "step": 1667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004686845060188593,
+      "loss": 2.6382,
+      "step": 1668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004686475299405393,
+      "loss": 2.6164,
+      "step": 1669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004686105335055252,
+      "loss": 2.5813,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046857351671726155,
+      "loss": 2.772,
+      "step": 1671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004685364795791947,
+      "loss": 2.7939,
+      "step": 1672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004684994220947729,
+      "loss": 2.7838,
+      "step": 1673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046846234426744626,
+      "loss": 2.5741,
+      "step": 1674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004684252461006668,
+      "loss": 2.6508,
+      "step": 1675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004683881275978886,
+      "loss": 2.7672,
+      "step": 1676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046835098876256745,
+      "loss": 2.6882,
+      "step": 1677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046831382959816104,
+      "loss": 2.6439,
+      "step": 1678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004682766501081289,
+      "loss": 2.8791,
+      "step": 1679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004682394502959327,
+      "loss": 2.6422,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004682022301650359,
+      "loss": 2.8276,
+      "step": 1681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004681649897189036,
+      "loss": 2.7187,
+      "step": 1682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004681277289610031,
+      "loss": 2.7066,
+      "step": 1683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004680904478948034,
+      "loss": 2.5153,
+      "step": 1684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004680531465237755,
+      "loss": 2.7566,
+      "step": 1685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004680158248513924,
+      "loss": 2.5806,
+      "step": 1686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046797848288112866,
+      "loss": 2.5669,
+      "step": 1687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004679411206164611,
+      "loss": 2.7362,
+      "step": 1688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046790373806086805,
+      "loss": 2.5365,
+      "step": 1689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046786633521783005,
+      "loss": 2.6185,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046782891209082945,
+      "loss": 2.7121,
+      "step": 1691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004677914686833504,
+      "loss": 2.5088,
+      "step": 1692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004677540049988789,
+      "loss": 2.5898,
+      "step": 1693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000467716521040903,
+      "loss": 2.6244,
+      "step": 1694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046767901681291264,
+      "loss": 2.659,
+      "step": 1695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046764149231839946,
+      "loss": 2.7428,
+      "step": 1696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004676039475608571,
+      "loss": 2.7262,
+      "step": 1697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046756638254378115,
+      "loss": 2.6628,
+      "step": 1698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004675287972706689,
+      "loss": 2.6727,
+      "step": 1699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046749119174501976,
+      "loss": 2.639,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.7057242393493652,
+      "eval_runtime": 184.024,
+      "eval_samples_per_second": 457.571,
+      "eval_steps_per_second": 28.6,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046745356597033483,
+      "loss": 2.7683,
+      "step": 1701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004674159199501173,
+      "loss": 2.6812,
+      "step": 1702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046737825368787193,
+      "loss": 2.6424,
+      "step": 1703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004673405671871057,
+      "loss": 2.8988,
+      "step": 1704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046730286045132733,
+      "loss": 2.7319,
+      "step": 1705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004672651334840473,
+      "loss": 2.5946,
+      "step": 1706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046722738628877816,
+      "loss": 2.6617,
+      "step": 1707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004671896188690343,
+      "loss": 2.7266,
+      "step": 1708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046715183122833206,
+      "loss": 2.7727,
+      "step": 1709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004671140233701894,
+      "loss": 2.7236,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004670761952981264,
+      "loss": 2.7015,
+      "step": 1711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004670383470156649,
+      "loss": 2.7173,
+      "step": 1712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004670004785263289,
+      "loss": 2.6359,
+      "step": 1713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004669625898336438,
+      "loss": 2.6853,
+      "step": 1714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004669246809411373,
+      "loss": 2.6142,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004668867518523388,
+      "loss": 2.6295,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046684880257077954,
+      "loss": 2.6984,
+      "step": 1717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004668108330999928,
+      "loss": 2.753,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046677284344351355,
+      "loss": 2.7562,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046673483360487876,
+      "loss": 2.7548,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004666968035876273,
+      "loss": 2.4315,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004666587533952998,
+      "loss": 2.7465,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000466620683031439,
+      "loss": 2.7628,
+      "step": 1723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004665825924995891,
+      "loss": 2.672,
+      "step": 1724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004665444818032966,
+      "loss": 2.481,
+      "step": 1725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046650635094610973,
+      "loss": 2.7939,
+      "step": 1726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004664681999315784,
+      "loss": 2.8907,
+      "step": 1727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004664300287632548,
+      "loss": 2.6098,
+      "step": 1728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046639183744469267,
+      "loss": 2.7634,
+      "step": 1729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004663536259794477,
+      "loss": 2.6176,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004663153943710775,
+      "loss": 2.6902,
+      "step": 1731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004662771426231416,
+      "loss": 2.736,
+      "step": 1732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046623887073920134,
+      "loss": 2.7294,
+      "step": 1733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046620057872281983,
+      "loss": 2.58,
+      "step": 1734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004661622665775622,
+      "loss": 2.4929,
+      "step": 1735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004661239343069956,
+      "loss": 2.5931,
+      "step": 1736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004660855819146886,
+      "loss": 2.8504,
+      "step": 1737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004660472094042121,
+      "loss": 2.5989,
+      "step": 1738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046600881677913853,
+      "loss": 2.8293,
+      "step": 1739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004659704040430425,
+      "loss": 2.7393,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004659319711995003,
+      "loss": 2.4424,
+      "step": 1741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046589351825209016,
+      "loss": 2.7879,
+      "step": 1742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046585504520439203,
+      "loss": 2.771,
+      "step": 1743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000465816552059988,
+      "loss": 2.5774,
+      "step": 1744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004657780388224618,
+      "loss": 2.6624,
+      "step": 1745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004657395054953992,
+      "loss": 2.6662,
+      "step": 1746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046570095208238764,
+      "loss": 2.7953,
+      "step": 1747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004656623785870167,
+      "loss": 2.6869,
+      "step": 1748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004656237850128776,
+      "loss": 2.5733,
+      "step": 1749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004655851713635635,
+      "loss": 2.6051,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.6914498805999756,
+      "eval_runtime": 184.0217,
+      "eval_samples_per_second": 457.576,
+      "eval_steps_per_second": 28.6,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046554653764266943,
+      "loss": 2.5896,
+      "step": 1751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046550788385379233,
+      "loss": 2.7141,
+      "step": 1752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000465469210000531,
+      "loss": 2.5139,
+      "step": 1753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000465430516086486,
+      "loss": 2.7962,
+      "step": 1754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004653918021152599,
+      "loss": 2.7942,
+      "step": 1755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004653530680904571,
+      "loss": 2.6427,
+      "step": 1756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046531431401568383,
+      "loss": 2.6288,
+      "step": 1757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004652755398945481,
+      "loss": 2.7433,
+      "step": 1758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046523674573066,
+      "loss": 2.5776,
+      "step": 1759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004651979315276314,
+      "loss": 2.5829,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004651590972890759,
+      "loss": 2.9294,
+      "step": 1761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004651202430186092,
+      "loss": 2.606,
+      "step": 1762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046508136871984865,
+      "loss": 2.7656,
+      "step": 1763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004650424743964136,
+      "loss": 2.6583,
+      "step": 1764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004650035600519251,
+      "loss": 2.7072,
+      "step": 1765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046496462569000643,
+      "loss": 2.8572,
+      "step": 1766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046492567131428225,
+      "loss": 2.7235,
+      "step": 1767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004648866969283795,
+      "loss": 2.5847,
+      "step": 1768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046484770253592657,
+      "loss": 2.7511,
+      "step": 1769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046480868814055424,
+      "loss": 2.7235,
+      "step": 1770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004647696537458946,
+      "loss": 2.8206,
+      "step": 1771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046473059935558204,
+      "loss": 2.6138,
+      "step": 1772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046469152497325257,
+      "loss": 2.7006,
+      "step": 1773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046465243060254415,
+      "loss": 2.7864,
+      "step": 1774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046461331624709644,
+      "loss": 2.8137,
+      "step": 1775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004645741819105513,
+      "loss": 2.5559,
+      "step": 1776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004645350275965521,
+      "loss": 2.635,
+      "step": 1777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004644958533087443,
+      "loss": 2.6186,
+      "step": 1778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004644566590507751,
+      "loss": 2.4807,
+      "step": 1779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004644174448262935,
+      "loss": 2.6181,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004643782106389506,
+      "loss": 2.8133,
+      "step": 1781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004643389564923992,
+      "loss": 2.4782,
+      "step": 1782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046429968239029384,
+      "loss": 2.6844,
+      "step": 1783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046426038833629127,
+      "loss": 2.5507,
+      "step": 1784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046422107433404956,
+      "loss": 2.6409,
+      "step": 1785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046418174038722926,
+      "loss": 2.7497,
+      "step": 1786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046414238649949237,
+      "loss": 2.6941,
+      "step": 1787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004641030126745028,
+      "loss": 2.5063,
+      "step": 1788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004640636189159263,
+      "loss": 2.7689,
+      "step": 1789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004640242052274307,
+      "loss": 2.8458,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046398477161268544,
+      "loss": 2.4554,
+      "step": 1791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004639453180753619,
+      "loss": 2.5782,
+      "step": 1792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046390584461913334,
+      "loss": 2.7236,
+      "step": 1793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004638663512476748,
+      "loss": 2.4821,
+      "step": 1794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046382683796466324,
+      "loss": 2.7247,
+      "step": 1795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004637873047737775,
+      "loss": 2.6365,
+      "step": 1796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004637477516786982,
+      "loss": 2.622,
+      "step": 1797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004637081786831078,
+      "loss": 2.801,
+      "step": 1798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004636685857906908,
+      "loss": 2.6232,
+      "step": 1799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046362897300513317,
+      "loss": 2.5859,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.671586751937866,
+      "eval_runtime": 184.1712,
+      "eval_samples_per_second": 457.205,
+      "eval_steps_per_second": 28.577,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004635893403301232,
+      "loss": 2.6746,
+      "step": 1801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004635496877693507,
+      "loss": 2.468,
+      "step": 1802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046351001532650746,
+      "loss": 2.5625,
+      "step": 1803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000463470323005287,
+      "loss": 2.8845,
+      "step": 1804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000463430610809385,
+      "loss": 2.7977,
+      "step": 1805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004633908787424985,
+      "loss": 2.7918,
+      "step": 1806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046335112680832694,
+      "loss": 2.675,
+      "step": 1807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046331135501057107,
+      "loss": 2.8372,
+      "step": 1808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000463271563352934,
+      "loss": 2.6182,
+      "step": 1809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046323175183912024,
+      "loss": 2.5845,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046319192047283647,
+      "loss": 2.6791,
+      "step": 1811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004631520692577911,
+      "loss": 2.4706,
+      "step": 1812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004631121981976944,
+      "loss": 2.6267,
+      "step": 1813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004630723072962584,
+      "loss": 2.7431,
+      "step": 1814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046303239655719707,
+      "loss": 2.6122,
+      "step": 1815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046299246598422626,
+      "loss": 2.4815,
+      "step": 1816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046295251558106366,
+      "loss": 2.7027,
+      "step": 1817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004629125453514286,
+      "loss": 2.5592,
+      "step": 1818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004628725552990426,
+      "loss": 2.6535,
+      "step": 1819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004628325454276287,
+      "loss": 2.7667,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000462792515740912,
+      "loss": 2.7982,
+      "step": 1821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004627524662426194,
+      "loss": 2.5513,
+      "step": 1822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004627123969364796,
+      "loss": 2.7158,
+      "step": 1823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046267230782622313,
+      "loss": 2.8177,
+      "step": 1824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004626321989155824,
+      "loss": 2.5192,
+      "step": 1825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046259207020829177,
+      "loss": 2.5041,
+      "step": 1826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046255192170808714,
+      "loss": 2.7023,
+      "step": 1827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046251175341870663,
+      "loss": 2.7611,
+      "step": 1828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004624715653438899,
+      "loss": 2.6015,
+      "step": 1829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046243135748737864,
+      "loss": 2.6122,
+      "step": 1830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004623911298529163,
+      "loss": 2.6713,
+      "step": 1831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046235088244424803,
+      "loss": 2.6339,
+      "step": 1832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004623106152651213,
+      "loss": 2.7173,
+      "step": 1833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004622703283192848,
+      "loss": 2.6749,
+      "step": 1834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046223002161048955,
+      "loss": 2.6774,
+      "step": 1835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046218969514248817,
+      "loss": 2.7144,
+      "step": 1836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000462149348919035,
+      "loss": 2.7893,
+      "step": 1837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004621089829438867,
+      "loss": 2.8523,
+      "step": 1838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004620685972208012,
+      "loss": 2.5753,
+      "step": 1839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004620281917535386,
+      "loss": 2.5756,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004619877665458607,
+      "loss": 2.6582,
+      "step": 1841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004619473216015313,
+      "loss": 2.5533,
+      "step": 1842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004619068569243159,
+      "loss": 2.7996,
+      "step": 1843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004618663725179819,
+      "loss": 2.7339,
+      "step": 1844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004618258683862985,
+      "loss": 2.5536,
+      "step": 1845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004617853445330367,
+      "loss": 2.5158,
+      "step": 1846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004617448009619694,
+      "loss": 2.5475,
+      "step": 1847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004617042376768713,
+      "loss": 2.6401,
+      "step": 1848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000461663654681519,
+      "loss": 2.4883,
+      "step": 1849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004616230519796909,
+      "loss": 2.8165,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 2.65916109085083,
+      "eval_runtime": 184.7148,
+      "eval_samples_per_second": 455.86,
+      "eval_steps_per_second": 28.493,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004615824295751672,
+      "loss": 2.4919,
+      "step": 1851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046154178747173,
+      "loss": 2.8683,
+      "step": 1852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046150112567316313,
+      "loss": 2.5294,
+      "step": 1853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004614604441832523,
+      "loss": 2.7713,
+      "step": 1854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046141974300578523,
+      "loss": 2.7829,
+      "step": 1855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004613790221445511,
+      "loss": 2.6397,
+      "step": 1856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004613382816033412,
+      "loss": 2.5389,
+      "step": 1857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046129752138594874,
+      "loss": 2.5924,
+      "step": 1858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046125674149616834,
+      "loss": 2.644,
+      "step": 1859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000461215941937797,
+      "loss": 2.6385,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000461175122714633,
+      "loss": 2.7189,
+      "step": 1861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000461134283830477,
+      "loss": 2.7481,
+      "step": 1862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046109342528913104,
+      "loss": 2.7618,
+      "step": 1863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004610525470943992,
+      "loss": 2.5656,
+      "step": 1864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004610116492500874,
+      "loss": 2.6865,
+      "step": 1865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004609707317600032,
+      "loss": 2.6854,
+      "step": 1866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046092979462795627,
+      "loss": 2.5077,
+      "step": 1867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046088883785775796,
+      "loss": 2.7118,
+      "step": 1868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046084786145322143,
+      "loss": 2.7782,
+      "step": 1869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004608068654181617,
+      "loss": 2.6318,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004607658497563956,
+      "loss": 2.4322,
+      "step": 1871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046072481447174187,
+      "loss": 2.4943,
+      "step": 1872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004606837595680209,
+      "loss": 2.618,
+      "step": 1873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046064268504905507,
+      "loss": 2.7677,
+      "step": 1874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046060159091866853,
+      "loss": 2.6566,
+      "step": 1875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046056047718068727,
+      "loss": 2.6368,
+      "step": 1876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046051934383893915,
+      "loss": 2.4725,
+      "step": 1877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046047819089725363,
+      "loss": 2.627,
+      "step": 1878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004604370183594624,
+      "loss": 2.4176,
+      "step": 1879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046039582622939854,
+      "loss": 2.8503,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046035461451089714,
+      "loss": 2.5782,
+      "step": 1881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004603133832077953,
+      "loss": 2.6408,
+      "step": 1882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046027213232393167,
+      "loss": 2.5773,
+      "step": 1883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046023086186314677,
+      "loss": 2.6715,
+      "step": 1884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004601895718292831,
+      "loss": 2.6044,
+      "step": 1885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004601482622261848,
+      "loss": 2.6451,
+      "step": 1886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046010693305769795,
+      "loss": 2.6336,
+      "step": 1887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004600655843276704,
+      "loss": 2.8111,
+      "step": 1888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004600242160399519,
+      "loss": 2.6169,
+      "step": 1889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045998282819839376,
+      "loss": 2.5985,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045994142080684955,
+      "loss": 2.6813,
+      "step": 1891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045989999386917413,
+      "loss": 2.4105,
+      "step": 1892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004598585473892247,
+      "loss": 2.5756,
+      "step": 1893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045981708137086,
+      "loss": 2.6954,
+      "step": 1894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004597755958179406,
+      "loss": 2.4001,
+      "step": 1895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045973409073432885,
+      "loss": 2.5581,
+      "step": 1896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004596925661238891,
+      "loss": 2.6218,
+      "step": 1897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004596510219904874,
+      "loss": 2.7344,
+      "step": 1898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004596094583379915,
+      "loss": 2.7228,
+      "step": 1899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045956787517027124,
+      "loss": 2.7914,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 2.64477276802063,
+      "eval_runtime": 184.6135,
+      "eval_samples_per_second": 456.11,
+      "eval_steps_per_second": 28.508,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045952627249119804,
+      "loss": 2.7252,
+      "step": 1901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045948465030464533,
+      "loss": 2.4852,
+      "step": 1902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045944300861448806,
+      "loss": 2.6656,
+      "step": 1903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045940134742460337,
+      "loss": 2.6824,
+      "step": 1904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045935966673886996,
+      "loss": 2.7819,
+      "step": 1905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045931796656116844,
+      "loss": 2.5507,
+      "step": 1906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045927624689538115,
+      "loss": 2.8519,
+      "step": 1907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045923450774539243,
+      "loss": 2.7107,
+      "step": 1908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004591927491150882,
+      "loss": 2.723,
+      "step": 1909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004591509710083563,
+      "loss": 2.5831,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045910917342908644,
+      "loss": 2.492,
+      "step": 1911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045906735638117014,
+      "loss": 2.5439,
+      "step": 1912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004590255198685006,
+      "loss": 2.6073,
+      "step": 1913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004589836638949728,
+      "loss": 2.5744,
+      "step": 1914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045894178846448395,
+      "loss": 2.586,
+      "step": 1915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004588998935809325,
+      "loss": 2.5526,
+      "step": 1916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045885797924821915,
+      "loss": 2.763,
+      "step": 1917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004588160454702462,
+      "loss": 2.7392,
+      "step": 1918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004587740922509177,
+      "loss": 2.7735,
+      "step": 1919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045873211959413977,
+      "loss": 2.6811,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045869012750382004,
+      "loss": 2.5551,
+      "step": 1921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004586481159838682,
+      "loss": 2.6777,
+      "step": 1922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004586060850381956,
+      "loss": 2.6906,
+      "step": 1923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004585640346707153,
+      "loss": 2.7271,
+      "step": 1924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004585219648853426,
+      "loss": 2.7396,
+      "step": 1925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000458479875685994,
+      "loss": 2.788,
+      "step": 1926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045843776707658835,
+      "loss": 2.4055,
+      "step": 1927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045839563906104606,
+      "loss": 2.8365,
+      "step": 1928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004583534916432892,
+      "loss": 2.7574,
+      "step": 1929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045831132482724196,
+      "loss": 2.5463,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045826913861683017,
+      "loss": 2.6909,
+      "step": 1931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004582269330159815,
+      "loss": 2.6143,
+      "step": 1932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004581847080286252,
+      "loss": 2.6158,
+      "step": 1933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045814246365869285,
+      "loss": 2.8448,
+      "step": 1934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004581001999101173,
+      "loss": 2.6789,
+      "step": 1935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045805791678683356,
+      "loss": 2.4515,
+      "step": 1936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045801561429277814,
+      "loss": 2.9431,
+      "step": 1937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004579732924318897,
+      "loss": 2.5692,
+      "step": 1938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004579309512081085,
+      "loss": 2.5665,
+      "step": 1939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045788859062537645,
+      "loss": 2.5846,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004578462106876376,
+      "loss": 2.62,
+      "step": 1941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004578038113988376,
+      "loss": 2.7506,
+      "step": 1942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045776139276292396,
+      "loss": 2.586,
+      "step": 1943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000457718954783846,
+      "loss": 2.8373,
+      "step": 1944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004576764974655546,
+      "loss": 2.5459,
+      "step": 1945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045763402081200295,
+      "loss": 2.7373,
+      "step": 1946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004575915248271456,
+      "loss": 2.4894,
+      "step": 1947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000457549009514939,
+      "loss": 2.5261,
+      "step": 1948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045750647487934157,
+      "loss": 2.6736,
+      "step": 1949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045746392092431334,
+      "loss": 2.8124,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 2.635617733001709,
+      "eval_runtime": 184.7043,
+      "eval_samples_per_second": 455.885,
+      "eval_steps_per_second": 28.494,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004574213476538162,
+      "loss": 2.8757,
+      "step": 1951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045737875507181376,
+      "loss": 2.6228,
+      "step": 1952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004573361431822717,
+      "loss": 2.5173,
+      "step": 1953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045729351198915707,
+      "loss": 2.5277,
+      "step": 1954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045725086149643916,
+      "loss": 2.7968,
+      "step": 1955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045720819170808877,
+      "loss": 2.659,
+      "step": 1956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004571655026280785,
+      "loss": 2.6461,
+      "step": 1957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045712279426038296,
+      "loss": 2.6637,
+      "step": 1958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004570800666089783,
+      "loss": 2.6567,
+      "step": 1959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045703731967784266,
+      "loss": 2.515,
+      "step": 1960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000456994553470956,
+      "loss": 2.504,
+      "step": 1961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004569517679922997,
+      "loss": 2.8999,
+      "step": 1962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004569089632458574,
+      "loss": 2.583,
+      "step": 1963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045686613923561425,
+      "loss": 2.641,
+      "step": 1964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004568232959655574,
+      "loss": 2.5951,
+      "step": 1965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045678043343967554,
+      "loss": 2.7567,
+      "step": 1966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004567375516619594,
+      "loss": 2.5677,
+      "step": 1967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004566946506364013,
+      "loss": 2.6104,
+      "step": 1968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045665173036699554,
+      "loss": 2.5749,
+      "step": 1969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004566087908577381,
+      "loss": 2.6585,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004565658321126267,
+      "loss": 2.5143,
+      "step": 1971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045652285413566097,
+      "loss": 2.54,
+      "step": 1972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004564798569308423,
+      "loss": 2.5316,
+      "step": 1973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004564368405021737,
+      "loss": 2.7378,
+      "step": 1974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004563938048536603,
+      "loss": 2.6465,
+      "step": 1975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045635074998930884,
+      "loss": 2.5914,
+      "step": 1976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004563076759131277,
+      "loss": 2.6251,
+      "step": 1977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004562645826291274,
+      "loss": 2.5336,
+      "step": 1978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004562214701413198,
+      "loss": 2.5799,
+      "step": 1979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045617833845371894,
+      "loss": 2.4962,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045613518757034054,
+      "loss": 2.645,
+      "step": 1981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045609201749520197,
+      "loss": 2.4696,
+      "step": 1982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045604882823232254,
+      "loss": 2.5727,
+      "step": 1983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004560056197857232,
+      "loss": 2.5236,
+      "step": 1984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004559623921594269,
+      "loss": 2.5248,
+      "step": 1985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004559191453574582,
+      "loss": 2.5735,
+      "step": 1986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045587587938384353,
+      "loss": 2.4117,
+      "step": 1987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000455832594242611,
+      "loss": 2.5165,
+      "step": 1988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004557892899377906,
+      "loss": 2.6865,
+      "step": 1989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004557459664734141,
+      "loss": 2.6392,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045570262385351507,
+      "loss": 2.8141,
+      "step": 1991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045565926208212875,
+      "loss": 2.6253,
+      "step": 1992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004556158811632923,
+      "loss": 2.5888,
+      "step": 1993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045557248110104464,
+      "loss": 2.6052,
+      "step": 1994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004555290618994263,
+      "loss": 2.3888,
+      "step": 1995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004554856235624799,
+      "loss": 2.5307,
+      "step": 1996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045544216609424954,
+      "loss": 2.8702,
+      "step": 1997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004553986894987813,
+      "loss": 2.6145,
+      "step": 1998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004553551937801229,
+      "loss": 2.6652,
+      "step": 1999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000455311678942324,
+      "loss": 2.536,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 2.6179487705230713,
+      "eval_runtime": 184.5731,
+      "eval_samples_per_second": 456.21,
+      "eval_steps_per_second": 28.514,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004552681449894359,
+      "loss": 2.5118,
+      "step": 2001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004552245919255117,
+      "loss": 2.4508,
+      "step": 2002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045518101975460646,
+      "loss": 2.4558,
+      "step": 2003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004551374284807767,
+      "loss": 2.5457,
+      "step": 2004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004550938181080809,
+      "loss": 2.5965,
+      "step": 2005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004550501886405795,
+      "loss": 2.6907,
+      "step": 2006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004550065400823343,
+      "loss": 2.7291,
+      "step": 2007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004549628724374092,
+      "loss": 2.4782,
+      "step": 2008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045491918570986976,
+      "loss": 2.4736,
+      "step": 2009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045487547990378336,
+      "loss": 2.5813,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004548317550232192,
+      "loss": 2.4853,
+      "step": 2011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045478801107224796,
+      "loss": 2.5758,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004547442480549425,
+      "loss": 2.6786,
+      "step": 2013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045470046597537736,
+      "loss": 2.5745,
+      "step": 2014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004546566648376286,
+      "loss": 2.3253,
+      "step": 2015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045461284464577424,
+      "loss": 2.6316,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004545690054038941,
+      "loss": 2.7978,
+      "step": 2017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045452514711606987,
+      "loss": 2.5362,
+      "step": 2018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045448126978638465,
+      "loss": 2.6428,
+      "step": 2019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004544373734189237,
+      "loss": 2.6128,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004543934580177738,
+      "loss": 2.59,
+      "step": 2021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004543495235870236,
+      "loss": 2.4522,
+      "step": 2022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004543055701307637,
+      "loss": 2.4278,
+      "step": 2023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045426159765308606,
+      "loss": 2.6808,
+      "step": 2024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004542176061580847,
+      "loss": 2.6713,
+      "step": 2025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045417359564985544,
+      "loss": 2.6742,
+      "step": 2026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045412956613249565,
+      "loss": 2.5211,
+      "step": 2027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045408551761010475,
+      "loss": 2.4219,
+      "step": 2028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004540414500867837,
+      "loss": 2.5266,
+      "step": 2029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004539973635666352,
+      "loss": 2.729,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000453953258053764,
+      "loss": 2.6279,
+      "step": 2031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045390913355227634,
+      "loss": 2.6033,
+      "step": 2032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004538649900662805,
+      "loss": 2.504,
+      "step": 2033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004538208275998861,
+      "loss": 2.4866,
+      "step": 2034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045377664615720497,
+      "loss": 2.665,
+      "step": 2035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004537324457423505,
+      "loss": 2.6175,
+      "step": 2036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045368822635943786,
+      "loss": 2.6663,
+      "step": 2037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045364398801258396,
+      "loss": 2.5262,
+      "step": 2038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004535997307059075,
+      "loss": 2.5311,
+      "step": 2039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004535554544435291,
+      "loss": 2.4212,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045351115922957086,
+      "loss": 2.5539,
+      "step": 2041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004534668450681568,
+      "loss": 2.4897,
+      "step": 2042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004534225119634128,
+      "loss": 2.7326,
+      "step": 2043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004533781599194663,
+      "loss": 2.4317,
+      "step": 2044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045333378894044655,
+      "loss": 2.6347,
+      "step": 2045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004532893990304847,
+      "loss": 2.5239,
+      "step": 2046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004532449901937136,
+      "loss": 2.4661,
+      "step": 2047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004532005624342678,
+      "loss": 2.7065,
+      "step": 2048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004531561157562836,
+      "loss": 2.5203,
+      "step": 2049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045311165016389914,
+      "loss": 2.4832,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 2.6063754558563232,
+      "eval_runtime": 184.7017,
+      "eval_samples_per_second": 455.892,
+      "eval_steps_per_second": 28.495,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045306716566125433,
+      "loss": 2.689,
+      "step": 2051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004530226622524908,
+      "loss": 2.69,
+      "step": 2052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045297813994175186,
+      "loss": 2.6146,
+      "step": 2053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045293359873318275,
+      "loss": 2.5877,
+      "step": 2054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004528890386309303,
+      "loss": 2.6832,
+      "step": 2055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004528444596391433,
+      "loss": 2.7205,
+      "step": 2056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045279986176197204,
+      "loss": 2.5697,
+      "step": 2057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004527552450035688,
+      "loss": 2.7039,
+      "step": 2058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045271060936808745,
+      "loss": 2.6729,
+      "step": 2059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045266595485968377,
+      "loss": 2.5878,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045262128148251514,
+      "loss": 2.5541,
+      "step": 2061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045257658924074083,
+      "loss": 2.4942,
+      "step": 2062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045253187813852185,
+      "loss": 2.454,
+      "step": 2063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004524871481800209,
+      "loss": 2.5551,
+      "step": 2064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045244239936940235,
+      "loss": 2.5175,
+      "step": 2065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045239763171083257,
+      "loss": 2.3818,
+      "step": 2066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045235284520847953,
+      "loss": 2.7049,
+      "step": 2067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000452308039866513,
+      "loss": 2.6043,
+      "step": 2068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045226321568910444,
+      "loss": 2.6546,
+      "step": 2069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004522183726804271,
+      "loss": 2.7868,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045217351084465596,
+      "loss": 2.7868,
+      "step": 2071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045212863018596785,
+      "loss": 2.723,
+      "step": 2072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045208373070854124,
+      "loss": 2.619,
+      "step": 2073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045203881241655643,
+      "loss": 2.6752,
+      "step": 2074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045199387531419545,
+      "loss": 2.4448,
+      "step": 2075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000451948919405642,
+      "loss": 2.7467,
+      "step": 2076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004519039446950817,
+      "loss": 2.7269,
+      "step": 2077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045185895118670175,
+      "loss": 2.658,
+      "step": 2078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045181393888469115,
+      "loss": 2.4024,
+      "step": 2079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045176890779324066,
+      "loss": 2.2788,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045172385791654294,
+      "loss": 2.6134,
+      "step": 2081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045167878925879206,
+      "loss": 2.7048,
+      "step": 2082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004516337018241842,
+      "loss": 2.5189,
+      "step": 2083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004515885956169171,
+      "loss": 2.4506,
+      "step": 2084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045154347064119024,
+      "loss": 2.5058,
+      "step": 2085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045149832690120483,
+      "loss": 2.6625,
+      "step": 2086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045145316440116395,
+      "loss": 2.5436,
+      "step": 2087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004514079831452723,
+      "loss": 2.6216,
+      "step": 2088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004513627831377365,
+      "loss": 2.6963,
+      "step": 2089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045131756438276466,
+      "loss": 2.5712,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045127232688456684,
+      "loss": 2.6343,
+      "step": 2091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004512270706473547,
+      "loss": 2.5418,
+      "step": 2092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004511817956753419,
+      "loss": 2.6964,
+      "step": 2093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045113650197274347,
+      "loss": 2.5844,
+      "step": 2094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004510911895437765,
+      "loss": 2.5305,
+      "step": 2095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045104585839265954,
+      "loss": 2.5302,
+      "step": 2096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004510005085236133,
+      "loss": 2.5442,
+      "step": 2097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004509551399408598,
+      "loss": 2.7704,
+      "step": 2098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000450909752648623,
+      "loss": 2.7125,
+      "step": 2099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045086434665112863,
+      "loss": 2.7004,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 2.589585065841675,
+      "eval_runtime": 184.7147,
+      "eval_samples_per_second": 455.86,
+      "eval_steps_per_second": 28.493,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004508189219526042,
+      "loss": 2.6862,
+      "step": 2101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004507734785572786,
+      "loss": 2.4993,
+      "step": 2102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000450728016469383,
+      "loss": 2.6937,
+      "step": 2103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045068253569315,
+      "loss": 2.4468,
+      "step": 2104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004506370362328139,
+      "loss": 2.5562,
+      "step": 2105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004505915180926108,
+      "loss": 2.6099,
+      "step": 2106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004505459812767787,
+      "loss": 2.5763,
+      "step": 2107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004505004257895572,
+      "loss": 2.6116,
+      "step": 2108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045045485163518754,
+      "loss": 2.7513,
+      "step": 2109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045040925881791277,
+      "loss": 2.3173,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004503636473419779,
+      "loss": 2.6661,
+      "step": 2111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045031801721162926,
+      "loss": 2.5554,
+      "step": 2112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004502723684311153,
+      "loss": 2.6077,
+      "step": 2113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000450226701004686,
+      "loss": 2.675,
+      "step": 2114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004501810149365931,
+      "loss": 2.4624,
+      "step": 2115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004501353102310901,
+      "loss": 2.3852,
+      "step": 2116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004500895868924323,
+      "loss": 2.4081,
+      "step": 2117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004500438449248766,
+      "loss": 2.5455,
+      "step": 2118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004499980843326818,
+      "loss": 2.6709,
+      "step": 2119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004499523051201082,
+      "loss": 2.571,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004499065072914181,
+      "loss": 2.6116,
+      "step": 2121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004498606908508753,
+      "loss": 2.4571,
+      "step": 2122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004498148558027455,
+      "loss": 2.7115,
+      "step": 2123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044976900215129605,
+      "loss": 2.4251,
+      "step": 2124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044972312990079614,
+      "loss": 2.4531,
+      "step": 2125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004496772390555164,
+      "loss": 2.6357,
+      "step": 2126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044963132961972966,
+      "loss": 2.5216,
+      "step": 2127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044958540159770996,
+      "loss": 2.6338,
+      "step": 2128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004495394549937335,
+      "loss": 2.6367,
+      "step": 2129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004494934898120779,
+      "loss": 2.5243,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004494475060570228,
+      "loss": 2.2522,
+      "step": 2131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004494015037328493,
+      "loss": 2.4927,
+      "step": 2132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004493554828438404,
+      "loss": 2.5219,
+      "step": 2133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004493094433942808,
+      "loss": 2.578,
+      "step": 2134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004492633853884569,
+      "loss": 2.4405,
+      "step": 2135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004492173088306567,
+      "loss": 2.7332,
+      "step": 2136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044917121372517016,
+      "loss": 2.7017,
+      "step": 2137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044912510007628887,
+      "loss": 2.6504,
+      "step": 2138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004490789678883062,
+      "loss": 2.4909,
+      "step": 2139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044903281716551693,
+      "loss": 2.6384,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004489866479122181,
+      "loss": 2.7963,
+      "step": 2141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004489404601327081,
+      "loss": 2.7623,
+      "step": 2142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000448894253831287,
+      "loss": 2.6743,
+      "step": 2143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044884802901225696,
+      "loss": 2.6979,
+      "step": 2144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004488017856799216,
+      "loss": 2.7781,
+      "step": 2145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004487555238385862,
+      "loss": 2.649,
+      "step": 2146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004487092434925579,
+      "loss": 2.6764,
+      "step": 2147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044866294464614555,
+      "loss": 2.5508,
+      "step": 2148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004486166273036597,
+      "loss": 2.5438,
+      "step": 2149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044857029146941265,
+      "loss": 2.5611,
+      "step": 2150
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 2.5802812576293945,
+      "eval_runtime": 184.706,
+      "eval_samples_per_second": 455.881,
+      "eval_steps_per_second": 28.494,
+      "step": 2150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044852393714771836,
+      "loss": 2.4908,
+      "step": 2151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044847756434289243,
+      "loss": 2.4909,
+      "step": 2152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044843117305925253,
+      "loss": 2.6182,
+      "step": 2153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004483847633011177,
+      "loss": 2.5497,
+      "step": 2154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004483383350728088,
+      "loss": 2.5326,
+      "step": 2155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044829188837864844,
+      "loss": 2.622,
+      "step": 2156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000448245423222961,
+      "loss": 2.6498,
+      "step": 2157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004481989396100724,
+      "loss": 2.6345,
+      "step": 2158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004481524375443105,
+      "loss": 2.66,
+      "step": 2159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004481059170300047,
+      "loss": 2.4132,
+      "step": 2160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044805937807148623,
+      "loss": 2.6908,
+      "step": 2161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000448012820673088,
+      "loss": 2.5891,
+      "step": 2162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004479662448391446,
+      "loss": 2.6838,
+      "step": 2163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044791965057399243,
+      "loss": 2.55,
+      "step": 2164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004478730378819694,
+      "loss": 2.7557,
+      "step": 2165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004478264067674155,
+      "loss": 2.6859,
+      "step": 2166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044777975723467203,
+      "loss": 2.3622,
+      "step": 2167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004477330892880823,
+      "loss": 2.6173,
+      "step": 2168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044768640293199114,
+      "loss": 2.5877,
+      "step": 2169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044763969817074534,
+      "loss": 2.5819,
+      "step": 2170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000447592975008693,
+      "loss": 2.6637,
+      "step": 2171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004475462334501844,
+      "loss": 2.7558,
+      "step": 2172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004474994734995712,
+      "loss": 2.6332,
+      "step": 2173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044745269516120685,
+      "loss": 2.6651,
+      "step": 2174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044740589843944665,
+      "loss": 2.5898,
+      "step": 2175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004473590833386474,
+      "loss": 2.4504,
+      "step": 2176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004473122498631678,
+      "loss": 2.5971,
+      "step": 2177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044726539801736816,
+      "loss": 2.5641,
+      "step": 2178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044721852780561046,
+      "loss": 2.573,
+      "step": 2179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044717163923225844,
+      "loss": 2.6388,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004471247323016777,
+      "loss": 2.5268,
+      "step": 2181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004470778070182353,
+      "loss": 2.5139,
+      "step": 2182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044703086338630007,
+      "loss": 2.3873,
+      "step": 2183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004469839014102427,
+      "loss": 2.6555,
+      "step": 2184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004469369210944354,
+      "loss": 2.6492,
+      "step": 2185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044688992244325216,
+      "loss": 2.4957,
+      "step": 2186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004468429054610688,
+      "loss": 2.6679,
+      "step": 2187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004467958701522625,
+      "loss": 2.3869,
+      "step": 2188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004467488165212127,
+      "loss": 2.7181,
+      "step": 2189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044670174457229995,
+      "loss": 2.6068,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004466546543099069,
+      "loss": 2.5763,
+      "step": 2191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004466075457384179,
+      "loss": 2.5717,
+      "step": 2192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044656041886221865,
+      "loss": 2.6643,
+      "step": 2193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004465132736856969,
+      "loss": 2.4461,
+      "step": 2194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044646611021324203,
+      "loss": 2.5258,
+      "step": 2195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000446418928449245,
+      "loss": 2.5707,
+      "step": 2196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044637172839809866,
+      "loss": 2.3551,
+      "step": 2197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044632451006419737,
+      "loss": 2.6131,
+      "step": 2198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004462772734519375,
+      "loss": 2.8224,
+      "step": 2199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044623001856571666,
+      "loss": 2.4008,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 2.5668466091156006,
+      "eval_runtime": 184.8533,
+      "eval_samples_per_second": 455.518,
+      "eval_steps_per_second": 28.471,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004461827454099345,
+      "loss": 2.4127,
+      "step": 2201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004461354539889923,
+      "loss": 2.4343,
+      "step": 2202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044608814430729303,
+      "loss": 2.6819,
+      "step": 2203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044604081636924134,
+      "loss": 2.47,
+      "step": 2204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044599347017924357,
+      "loss": 2.5653,
+      "step": 2205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004459461057417078,
+      "loss": 2.5153,
+      "step": 2206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044589872306104384,
+      "loss": 2.635,
+      "step": 2207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044585132214166305,
+      "loss": 2.7375,
+      "step": 2208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004458039029879787,
+      "loss": 2.6137,
+      "step": 2209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044575646560440557,
+      "loss": 2.5949,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044570900999536025,
+      "loss": 2.2453,
+      "step": 2211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044566153616526087,
+      "loss": 2.6069,
+      "step": 2212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044561404411852756,
+      "loss": 2.6045,
+      "step": 2213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004455665338595819,
+      "loss": 2.6914,
+      "step": 2214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004455190053928471,
+      "loss": 2.6463,
+      "step": 2215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004454714587227484,
+      "loss": 2.4947,
+      "step": 2216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004454238938537124,
+      "loss": 2.4285,
+      "step": 2217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004453763107901675,
+      "loss": 2.6406,
+      "step": 2218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044532870953654393,
+      "loss": 2.6649,
+      "step": 2219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044528109009727335,
+      "loss": 2.5055,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044523345247678945,
+      "loss": 2.3364,
+      "step": 2221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004451857966795272,
+      "loss": 2.5945,
+      "step": 2222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004451381227099237,
+      "loss": 2.6098,
+      "step": 2223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004450904305724174,
+      "loss": 2.5382,
+      "step": 2224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044504272027144854,
+      "loss": 2.3977,
+      "step": 2225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044499499181145923,
+      "loss": 2.7623,
+      "step": 2226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044494724519689307,
+      "loss": 2.681,
+      "step": 2227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004448994804321953,
+      "loss": 2.4475,
+      "step": 2228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004448516975218131,
+      "loss": 2.6649,
+      "step": 2229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000444803896470195,
+      "loss": 2.7626,
+      "step": 2230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004447560772817916,
+      "loss": 2.6611,
+      "step": 2231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004447082399610549,
+      "loss": 2.5826,
+      "step": 2232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004446603845124388,
+      "loss": 2.7291,
+      "step": 2233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044461251094039865,
+      "loss": 2.6147,
+      "step": 2234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004445646192493916,
+      "loss": 2.3821,
+      "step": 2235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004445167094438767,
+      "loss": 2.6409,
+      "step": 2236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004444687815283142,
+      "loss": 2.6439,
+      "step": 2237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004444208355071665,
+      "loss": 2.5891,
+      "step": 2238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044437287138489746,
+      "loss": 2.563,
+      "step": 2239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004443248891659727,
+      "loss": 2.6385,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004442768888548595,
+      "loss": 2.6019,
+      "step": 2241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004442288704560268,
+      "loss": 2.584,
+      "step": 2242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044418083397394526,
+      "loss": 2.6137,
+      "step": 2243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044413277941308715,
+      "loss": 2.4911,
+      "step": 2244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044408470677792663,
+      "loss": 2.4122,
+      "step": 2245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004440366160729392,
+      "loss": 2.2682,
+      "step": 2246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004439885073026024,
+      "loss": 2.6833,
+      "step": 2247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044394038047139526,
+      "loss": 2.3965,
+      "step": 2248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044389223558379844,
+      "loss": 2.3616,
+      "step": 2249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044384407264429443,
+      "loss": 2.6482,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 2.5550291538238525,
+      "eval_runtime": 184.5136,
+      "eval_samples_per_second": 456.357,
+      "eval_steps_per_second": 28.524,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044379589165736733,
+      "loss": 2.6492,
+      "step": 2251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004437476926275029,
+      "loss": 2.5009,
+      "step": 2252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044369947555918864,
+      "loss": 2.3658,
+      "step": 2253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044365124045691364,
+      "loss": 2.4342,
+      "step": 2254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004436029873251688,
+      "loss": 2.4149,
+      "step": 2255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044355471616844657,
+      "loss": 2.6946,
+      "step": 2256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004435064269912411,
+      "loss": 2.3485,
+      "step": 2257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004434581197980483,
+      "loss": 2.768,
+      "step": 2258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044340979459336574,
+      "loss": 2.4507,
+      "step": 2259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044336145138169247,
+      "loss": 2.6571,
+      "step": 2260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004433130901675295,
+      "loss": 2.7248,
+      "step": 2261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044326471095537944,
+      "loss": 2.3938,
+      "step": 2262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044321631374974647,
+      "loss": 2.5378,
+      "step": 2263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004431678985551365,
+      "loss": 2.5318,
+      "step": 2264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004431194653760571,
+      "loss": 2.5095,
+      "step": 2265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044307101421701755,
+      "loss": 2.6016,
+      "step": 2266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004430225450825288,
+      "loss": 2.4731,
+      "step": 2267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044297405797710343,
+      "loss": 2.3757,
+      "step": 2268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004429255529052558,
+      "loss": 2.6047,
+      "step": 2269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004428770298715018,
+      "loss": 2.5385,
+      "step": 2270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004428284888803591,
+      "loss": 2.6197,
+      "step": 2271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004427799299363469,
+      "loss": 2.5911,
+      "step": 2272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004427313530439863,
+      "loss": 2.5447,
+      "step": 2273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004426827582077999,
+      "loss": 2.5315,
+      "step": 2274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044263414543231204,
+      "loss": 2.6242,
+      "step": 2275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004425855147220487,
+      "loss": 2.5083,
+      "step": 2276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044253686608153737,
+      "loss": 2.557,
+      "step": 2277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004424881995153076,
+      "loss": 2.5043,
+      "step": 2278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004424395150278903,
+      "loss": 2.6226,
+      "step": 2279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004423908126238181,
+      "loss": 2.6042,
+      "step": 2280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044234209230762533,
+      "loss": 2.2923,
+      "step": 2281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004422933540838481,
+      "loss": 2.5059,
+      "step": 2282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004422445979570239,
+      "loss": 2.5652,
+      "step": 2283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044219582393169223,
+      "loss": 2.4997,
+      "step": 2284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000442147032012394,
+      "loss": 2.5558,
+      "step": 2285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044209822220367184,
+      "loss": 2.6009,
+      "step": 2286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044204939451007014,
+      "loss": 2.4036,
+      "step": 2287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044200054893613484,
+      "loss": 2.6307,
+      "step": 2288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044195168548641367,
+      "loss": 2.411,
+      "step": 2289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044190280416545594,
+      "loss": 2.5689,
+      "step": 2290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004418539049778126,
+      "loss": 2.4809,
+      "step": 2291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044180498792803627,
+      "loss": 2.6411,
+      "step": 2292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044175605302068134,
+      "loss": 2.5749,
+      "step": 2293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044170710026030375,
+      "loss": 2.4707,
+      "step": 2294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004416581296514611,
+      "loss": 2.5119,
+      "step": 2295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004416091411987128,
+      "loss": 2.5091,
+      "step": 2296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044156013490661975,
+      "loss": 2.6214,
+      "step": 2297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004415111107797445,
+      "loss": 2.5203,
+      "step": 2298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004414620688226515,
+      "loss": 2.5796,
+      "step": 2299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004414130090399065,
+      "loss": 2.4384,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 2.5412535667419434,
+      "eval_runtime": 184.7965,
+      "eval_samples_per_second": 455.658,
+      "eval_steps_per_second": 28.48,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004413639314360772,
+      "loss": 2.5306,
+      "step": 2301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004413148360157329,
+      "loss": 2.3727,
+      "step": 2302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044126572278344446,
+      "loss": 2.5066,
+      "step": 2303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004412165917437845,
+      "loss": 2.4372,
+      "step": 2304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044116744290132717,
+      "loss": 2.5332,
+      "step": 2305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044111827626064836,
+      "loss": 2.5648,
+      "step": 2306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044106909182632583,
+      "loss": 2.6764,
+      "step": 2307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004410198896029386,
+      "loss": 2.5333,
+      "step": 2308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004409706695950676,
+      "loss": 2.4618,
+      "step": 2309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044092143180729526,
+      "loss": 2.5428,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004408721762442059,
+      "loss": 2.4895,
+      "step": 2311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044082290291038524,
+      "loss": 2.5086,
+      "step": 2312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044077361181042073,
+      "loss": 2.62,
+      "step": 2313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044072430294890173,
+      "loss": 2.3873,
+      "step": 2314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044067497633041875,
+      "loss": 2.7081,
+      "step": 2315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004406256319595645,
+      "loss": 2.3825,
+      "step": 2316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004405762698409328,
+      "loss": 2.4727,
+      "step": 2317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004405268899791196,
+      "loss": 2.6022,
+      "step": 2318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044047749237872226,
+      "loss": 2.4116,
+      "step": 2319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004404280770443398,
+      "loss": 2.4202,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004403786439805729,
+      "loss": 2.5544,
+      "step": 2321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044032919319202406,
+      "loss": 2.4063,
+      "step": 2322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004402797246832971,
+      "loss": 2.6028,
+      "step": 2323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004402302384589979,
+      "loss": 2.7355,
+      "step": 2324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044018073452373353,
+      "loss": 2.5298,
+      "step": 2325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044013121288211305,
+      "loss": 2.4637,
+      "step": 2326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044008167353874706,
+      "loss": 2.463,
+      "step": 2327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004400321164982478,
+      "loss": 2.6848,
+      "step": 2328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004399825417652292,
+      "loss": 2.5868,
+      "step": 2329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004399329493443067,
+      "loss": 2.5551,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004398833392400977,
+      "loss": 2.4971,
+      "step": 2331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004398337114572208,
+      "loss": 2.6712,
+      "step": 2332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004397840660002967,
+      "loss": 2.7013,
+      "step": 2333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043973440287394737,
+      "loss": 2.5065,
+      "step": 2334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043968472208279663,
+      "loss": 2.4261,
+      "step": 2335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043963502363147,
+      "loss": 2.5067,
+      "step": 2336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043958530752459437,
+      "loss": 2.7223,
+      "step": 2337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004395355737667985,
+      "loss": 2.5648,
+      "step": 2338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004394858223627129,
+      "loss": 2.6289,
+      "step": 2339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043943605331696935,
+      "loss": 2.3892,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004393862666342017,
+      "loss": 2.4233,
+      "step": 2341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000439336462319045,
+      "loss": 2.463,
+      "step": 2342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004392866403761363,
+      "loss": 2.5448,
+      "step": 2343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004392368008101141,
+      "loss": 2.5063,
+      "step": 2344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043918694362561866,
+      "loss": 2.5693,
+      "step": 2345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004391370688272919,
+      "loss": 2.4852,
+      "step": 2346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043908717641977707,
+      "loss": 2.6548,
+      "step": 2347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004390372664077195,
+      "loss": 2.3564,
+      "step": 2348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004389873387957659,
+      "loss": 2.5437,
+      "step": 2349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004389373935885646,
+      "loss": 2.6389,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 2.533193826675415,
+      "eval_runtime": 184.7249,
+      "eval_samples_per_second": 455.835,
+      "eval_steps_per_second": 28.491,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043888743079076564,
+      "loss": 2.4236,
+      "step": 2351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043883745040702084,
+      "loss": 2.6698,
+      "step": 2352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004387874524419834,
+      "loss": 2.5315,
+      "step": 2353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043873743690030833,
+      "loss": 2.308,
+      "step": 2354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043868740378665206,
+      "loss": 2.4462,
+      "step": 2355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000438637353105673,
+      "loss": 2.5241,
+      "step": 2356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043858728486203093,
+      "loss": 2.5187,
+      "step": 2357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043853719906038733,
+      "loss": 2.4551,
+      "step": 2358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043848709570540534,
+      "loss": 2.5562,
+      "step": 2359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004384369748017498,
+      "loss": 2.5597,
+      "step": 2360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004383868363540869,
+      "loss": 2.5913,
+      "step": 2361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043833668036708483,
+      "loss": 2.4693,
+      "step": 2362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004382865068454133,
+      "loss": 2.3476,
+      "step": 2363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043823631579374353,
+      "loss": 2.4142,
+      "step": 2364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043818610721674836,
+      "loss": 2.4851,
+      "step": 2365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004381358811191025,
+      "loss": 2.4465,
+      "step": 2366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000438085637505482,
+      "loss": 2.4059,
+      "step": 2367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004380353763805648,
+      "loss": 2.4613,
+      "step": 2368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004379850977490303,
+      "loss": 2.4706,
+      "step": 2369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004379348016155596,
+      "loss": 2.5076,
+      "step": 2370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043788448798483536,
+      "loss": 2.6186,
+      "step": 2371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000437834156861542,
+      "loss": 2.4272,
+      "step": 2372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004377838082503654,
+      "loss": 2.6295,
+      "step": 2373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004377334421559932,
+      "loss": 2.3925,
+      "step": 2374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004376830585831146,
+      "loss": 2.3441,
+      "step": 2375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004376326575364206,
+      "loss": 2.5971,
+      "step": 2376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043758223902060333,
+      "loss": 2.7343,
+      "step": 2377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043753180304035726,
+      "loss": 2.4309,
+      "step": 2378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004374813496003779,
+      "loss": 2.4625,
+      "step": 2379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004374308787053628,
+      "loss": 2.5633,
+      "step": 2380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043738039036001077,
+      "loss": 2.4436,
+      "step": 2381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004373298845690224,
+      "loss": 2.6361,
+      "step": 2382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004372793613371,
+      "loss": 2.3469,
+      "step": 2383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004372288206689474,
+      "loss": 2.3817,
+      "step": 2384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004371782625692702,
+      "loss": 2.6046,
+      "step": 2385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043712768704277526,
+      "loss": 2.4811,
+      "step": 2386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004370770940941715,
+      "loss": 2.5755,
+      "step": 2387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043702648372816915,
+      "loss": 2.5208,
+      "step": 2388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004369758559494803,
+      "loss": 2.495,
+      "step": 2389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043692521076281834,
+      "loss": 2.517,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004368745481728987,
+      "loss": 2.561,
+      "step": 2391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043682386818443805,
+      "loss": 2.3591,
+      "step": 2392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043677317080215486,
+      "loss": 2.3002,
+      "step": 2393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004367224560307692,
+      "loss": 2.6561,
+      "step": 2394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004366717238750029,
+      "loss": 2.3805,
+      "step": 2395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004366209743395791,
+      "loss": 2.3682,
+      "step": 2396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004365702074292227,
+      "loss": 2.4622,
+      "step": 2397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043651942314866044,
+      "loss": 2.5518,
+      "step": 2398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004364686215026203,
+      "loss": 2.538,
+      "step": 2399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004364178024958321,
+      "loss": 2.33,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 2.521118640899658,
+      "eval_runtime": 184.6665,
+      "eval_samples_per_second": 455.979,
+      "eval_steps_per_second": 28.5,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004363669661330272,
+      "loss": 2.5506,
+      "step": 2401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004363161124189387,
+      "loss": 2.4623,
+      "step": 2402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043626524135830115,
+      "loss": 2.5894,
+      "step": 2403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004362143529558508,
+      "loss": 2.5611,
+      "step": 2404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004361634472163255,
+      "loss": 2.4834,
+      "step": 2405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043611252414446465,
+      "loss": 2.7102,
+      "step": 2406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043606158374500954,
+      "loss": 2.5199,
+      "step": 2407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004360106260227027,
+      "loss": 2.4532,
+      "step": 2408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004359596509822884,
+      "loss": 2.7272,
+      "step": 2409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043590865862851266,
+      "loss": 2.6819,
+      "step": 2410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004358576489661229,
+      "loss": 2.6142,
+      "step": 2411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043580662199986845,
+      "loss": 2.5644,
+      "step": 2412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004357555777344998,
+      "loss": 2.4646,
+      "step": 2413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043570451617476957,
+      "loss": 2.3257,
+      "step": 2414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004356534373254316,
+      "loss": 2.3439,
+      "step": 2415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004356023411912415,
+      "loss": 2.545,
+      "step": 2416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004355512277769564,
+      "loss": 2.5479,
+      "step": 2417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004355000970873352,
+      "loss": 2.4691,
+      "step": 2418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004354489491271383,
+      "loss": 2.4237,
+      "step": 2419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004353977839011276,
+      "loss": 2.4819,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043534660141406685,
+      "loss": 2.5813,
+      "step": 2421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043529540167072125,
+      "loss": 2.5429,
+      "step": 2422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043524418467585756,
+      "loss": 2.5508,
+      "step": 2423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004351929504342444,
+      "loss": 2.5439,
+      "step": 2424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004351416989506517,
+      "loss": 2.7186,
+      "step": 2425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004350904302298511,
+      "loss": 2.4181,
+      "step": 2426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043503914427661596,
+      "loss": 2.5588,
+      "step": 2427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043498784109572097,
+      "loss": 2.4512,
+      "step": 2428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043493652069194277,
+      "loss": 2.3777,
+      "step": 2429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004348851830700593,
+      "loss": 2.4227,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004348338282348504,
+      "loss": 2.4161,
+      "step": 2431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043478245619109725,
+      "loss": 2.524,
+      "step": 2432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043473106694358267,
+      "loss": 2.3444,
+      "step": 2433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043467966049709126,
+      "loss": 2.5315,
+      "step": 2434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043462823685640894,
+      "loss": 2.451,
+      "step": 2435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043457679602632363,
+      "loss": 2.726,
+      "step": 2436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004345253380116244,
+      "loss": 2.4842,
+      "step": 2437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004344738628171023,
+      "loss": 2.5386,
+      "step": 2438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043442237044754965,
+      "loss": 2.5318,
+      "step": 2439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043437086090776067,
+      "loss": 2.452,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000434319334202531,
+      "loss": 2.5073,
+      "step": 2441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043426779033665794,
+      "loss": 2.4433,
+      "step": 2442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004342162293149403,
+      "loss": 2.4054,
+      "step": 2443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043416465114217855,
+      "loss": 2.2352,
+      "step": 2444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043411305582317483,
+      "loss": 2.5248,
+      "step": 2445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004340614433627328,
+      "loss": 2.6706,
+      "step": 2446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043400981376565774,
+      "loss": 2.4099,
+      "step": 2447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004339581670367564,
+      "loss": 2.6042,
+      "step": 2448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043390650318083734,
+      "loss": 2.5141,
+      "step": 2449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004338548222027106,
+      "loss": 2.3862,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 2.5077686309814453,
+      "eval_runtime": 184.7753,
+      "eval_samples_per_second": 455.71,
+      "eval_steps_per_second": 28.483,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004338031241071878,
+      "loss": 2.3476,
+      "step": 2451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043375140889908214,
+      "loss": 2.555,
+      "step": 2452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004336996765832085,
+      "loss": 2.2952,
+      "step": 2453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004336479271643833,
+      "loss": 2.4268,
+      "step": 2454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004335961606474246,
+      "loss": 2.4379,
+      "step": 2455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004335443770371519,
+      "loss": 2.5567,
+      "step": 2456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043349257633838647,
+      "loss": 2.4804,
+      "step": 2457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000433440758555951,
+      "loss": 2.6051,
+      "step": 2458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043338892369467006,
+      "loss": 2.5228,
+      "step": 2459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043333707175936944,
+      "loss": 2.4575,
+      "step": 2460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004332852027548768,
+      "loss": 2.4707,
+      "step": 2461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004332333166860212,
+      "loss": 2.4915,
+      "step": 2462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043318141355763354,
+      "loss": 2.4968,
+      "step": 2463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000433129493374546,
+      "loss": 2.5569,
+      "step": 2464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004330775561415925,
+      "loss": 2.4386,
+      "step": 2465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004330256018636086,
+      "loss": 2.4003,
+      "step": 2466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043297363054543137,
+      "loss": 2.5198,
+      "step": 2467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043292164219189944,
+      "loss": 2.4569,
+      "step": 2468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004328696368078532,
+      "loss": 2.4472,
+      "step": 2469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004328176143981343,
+      "loss": 2.6886,
+      "step": 2470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043276557496758634,
+      "loss": 2.5594,
+      "step": 2471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043271351852105426,
+      "loss": 2.6123,
+      "step": 2472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004326614450633847,
+      "loss": 2.5988,
+      "step": 2473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004326093545994258,
+      "loss": 2.3994,
+      "step": 2474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043255724713402735,
+      "loss": 2.4856,
+      "step": 2475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004325051226720407,
+      "loss": 2.6198,
+      "step": 2476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004324529812183187,
+      "loss": 2.5972,
+      "step": 2477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043240082277771596,
+      "loss": 2.4058,
+      "step": 2478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043234864735508863,
+      "loss": 2.5819,
+      "step": 2479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004322964549552943,
+      "loss": 2.3982,
+      "step": 2480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004322442455831922,
+      "loss": 2.6467,
+      "step": 2481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043219201924364327,
+      "loss": 2.432,
+      "step": 2482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004321397759415099,
+      "loss": 2.4317,
+      "step": 2483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043208751568165596,
+      "loss": 2.4565,
+      "step": 2484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004320352384689471,
+      "loss": 2.5396,
+      "step": 2485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004319829443082506,
+      "loss": 2.4102,
+      "step": 2486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043193063320443503,
+      "loss": 2.4649,
+      "step": 2487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004318783051623707,
+      "loss": 2.4027,
+      "step": 2488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043182596018692963,
+      "loss": 2.4744,
+      "step": 2489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043177359828298526,
+      "loss": 2.678,
+      "step": 2490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043172121945541243,
+      "loss": 2.3979,
+      "step": 2491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000431668823709088,
+      "loss": 2.3637,
+      "step": 2492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043161641104889003,
+      "loss": 2.4422,
+      "step": 2493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043156398147969834,
+      "loss": 2.257,
+      "step": 2494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043151153500639415,
+      "loss": 2.6023,
+      "step": 2495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004314590716338606,
+      "loss": 2.5526,
+      "step": 2496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043140659136698203,
+      "loss": 2.5087,
+      "step": 2497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043135409421064444,
+      "loss": 2.3769,
+      "step": 2498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043130158016973553,
+      "loss": 2.3917,
+      "step": 2499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043124904924914456,
+      "loss": 2.4908,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 2.499448537826538,
+      "eval_runtime": 184.509,
+      "eval_samples_per_second": 456.368,
+      "eval_steps_per_second": 28.524,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004311965014537622,
+      "loss": 2.4428,
+      "step": 2501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043114393678848087,
+      "loss": 2.4272,
+      "step": 2502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004310913552581945,
+      "loss": 2.636,
+      "step": 2503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004310387568677985,
+      "loss": 2.3285,
+      "step": 2504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043098614162218995,
+      "loss": 2.5184,
+      "step": 2505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004309335095262675,
+      "loss": 2.6194,
+      "step": 2506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004308808605849314,
+      "loss": 2.2572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043082819480308335,
+      "loss": 2.4659,
+      "step": 2508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004307755121856266,
+      "loss": 2.3838,
+      "step": 2509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043072281273746617,
+      "loss": 2.5967,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043067009646350854,
+      "loss": 2.3689,
+      "step": 2511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043061736336866166,
+      "loss": 2.4755,
+      "step": 2512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004305646134578351,
+      "loss": 2.6354,
+      "step": 2513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004305118467359401,
+      "loss": 2.4301,
+      "step": 2514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004304590632078894,
+      "loss": 2.349,
+      "step": 2515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004304062628785973,
+      "loss": 2.4975,
+      "step": 2516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004303534457529795,
+      "loss": 2.4956,
+      "step": 2517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043030061183595364,
+      "loss": 2.509,
+      "step": 2518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043024776113243857,
+      "loss": 2.4915,
+      "step": 2519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043019489364735485,
+      "loss": 2.5214,
+      "step": 2520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004301420093856247,
+      "loss": 2.3814,
+      "step": 2521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004300891083521716,
+      "loss": 2.5102,
+      "step": 2522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000430036190551921,
+      "loss": 2.3205,
+      "step": 2523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00042998325598979955,
+      "loss": 2.5302,
+      "step": 2524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00042993030467073555,
+      "loss": 2.6606,
+      "step": 2525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004298773365996591,
+      "loss": 2.5597,
+      "step": 2526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004298243517815016,
+      "loss": 2.4802,
+      "step": 2527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042977135022119597,
+      "loss": 2.4749,
+      "step": 2528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004297183319236769,
+      "loss": 2.4959,
+      "step": 2529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042966529689388066,
+      "loss": 2.4863,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042961224513674473,
+      "loss": 2.5332,
+      "step": 2531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042955917665720854,
+      "loss": 2.5259,
+      "step": 2532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004295060914602128,
+      "loss": 2.7254,
+      "step": 2533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004294529895507,
+      "loss": 2.3148,
+      "step": 2534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000429399870933614,
+      "loss": 2.4943,
+      "step": 2535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004293467356139003,
+      "loss": 2.4943,
+      "step": 2536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042929358359650594,
+      "loss": 2.3943,
+      "step": 2537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004292404148863796,
+      "loss": 2.5383,
+      "step": 2538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004291872294884714,
+      "loss": 2.3541,
+      "step": 2539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042913402740773294,
+      "loss": 2.4629,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004290808086491176,
+      "loss": 2.4758,
+      "step": 2541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042902757321758013,
+      "loss": 2.3432,
+      "step": 2542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000428974321118077,
+      "loss": 2.4873,
+      "step": 2543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000428921052355566,
+      "loss": 2.5435,
+      "step": 2544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042886776693500664,
+      "loss": 2.4416,
+      "step": 2545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042881446486136,
+      "loss": 2.5782,
+      "step": 2546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004287611461395886,
+      "loss": 2.3756,
+      "step": 2547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004287078107746566,
+      "loss": 2.3699,
+      "step": 2548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042865445877152966,
+      "loss": 2.7102,
+      "step": 2549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004286010901351749,
+      "loss": 2.4269,
+      "step": 2550
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 2.491239547729492,
+      "eval_runtime": 184.7793,
+      "eval_samples_per_second": 455.7,
+      "eval_steps_per_second": 28.483,
+      "step": 2550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042854770487056116,
+      "loss": 2.4409,
+      "step": 2551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004284943029826588,
+      "loss": 2.5422,
+      "step": 2552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004284408844764397,
+      "loss": 2.3981,
+      "step": 2553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004283874493568772,
+      "loss": 2.4765,
+      "step": 2554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004283339976289463,
+      "loss": 2.6763,
+      "step": 2555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004282805292976234,
+      "loss": 2.759,
+      "step": 2556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042822704436788674,
+      "loss": 2.5972,
+      "step": 2557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042817354284471575,
+      "loss": 2.4297,
+      "step": 2558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042812002473309165,
+      "loss": 2.4107,
+      "step": 2559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004280664900379971,
+      "loss": 2.4865,
+      "step": 2560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004280129387644164,
+      "loss": 2.5075,
+      "step": 2561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042795937091733515,
+      "loss": 2.4404,
+      "step": 2562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004279057865017408,
+      "loss": 2.2893,
+      "step": 2563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042785218552262227,
+      "loss": 2.3746,
+      "step": 2564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042779856798496983,
+      "loss": 2.3855,
+      "step": 2565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042774493389377544,
+      "loss": 2.4396,
+      "step": 2566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004276912832540327,
+      "loss": 2.3077,
+      "step": 2567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004276376160707365,
+      "loss": 2.5738,
+      "step": 2568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042758393234888336,
+      "loss": 2.5045,
+      "step": 2569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004275302320934716,
+      "loss": 2.4136,
+      "step": 2570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042747651530950073,
+      "loss": 2.5982,
+      "step": 2571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004274227820019718,
+      "loss": 2.3965,
+      "step": 2572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042736903217588786,
+      "loss": 2.6578,
+      "step": 2573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042731526583625286,
+      "loss": 2.4858,
+      "step": 2574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004272614829880728,
+      "loss": 2.3897,
+      "step": 2575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042720768363635475,
+      "loss": 2.2191,
+      "step": 2576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042715386778610794,
+      "loss": 2.5174,
+      "step": 2577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042710003544234256,
+      "loss": 2.5496,
+      "step": 2578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004270461866100705,
+      "loss": 2.4908,
+      "step": 2579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042699232129430535,
+      "loss": 2.5073,
+      "step": 2580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004269384395000622,
+      "loss": 2.3607,
+      "step": 2581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004268845412323573,
+      "loss": 2.2797,
+      "step": 2582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004268306264962091,
+      "loss": 2.4274,
+      "step": 2583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042677669529663686,
+      "loss": 2.3981,
+      "step": 2584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042672274763866196,
+      "loss": 2.3613,
+      "step": 2585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004266687835273071,
+      "loss": 2.5098,
+      "step": 2586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042661480296759624,
+      "loss": 2.4012,
+      "step": 2587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004265608059645554,
+      "loss": 2.5652,
+      "step": 2588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042650679252321166,
+      "loss": 2.4118,
+      "step": 2589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004264527626485939,
+      "loss": 2.5038,
+      "step": 2590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042639871634573255,
+      "loss": 2.2181,
+      "step": 2591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042634465361965926,
+      "loss": 2.1886,
+      "step": 2592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004262905744754075,
+      "loss": 2.5993,
+      "step": 2593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004262364789180123,
+      "loss": 2.6073,
+      "step": 2594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042618236695251,
+      "loss": 2.4662,
+      "step": 2595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004261282385839386,
+      "loss": 2.4428,
+      "step": 2596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042607409381733755,
+      "loss": 2.5953,
+      "step": 2597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000426019932657748,
+      "loss": 2.3535,
+      "step": 2598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004259657551102123,
+      "loss": 2.3847,
+      "step": 2599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004259115611797748,
+      "loss": 2.4281,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 2.4781548976898193,
+      "eval_runtime": 184.7272,
+      "eval_samples_per_second": 455.829,
+      "eval_steps_per_second": 28.491,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004258573508714809,
+      "loss": 2.3417,
+      "step": 2601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042580312419037775,
+      "loss": 2.5768,
+      "step": 2602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004257488811415141,
+      "loss": 2.3743,
+      "step": 2603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004256946217299401,
+      "loss": 2.5329,
+      "step": 2604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004256403459607075,
+      "loss": 2.516,
+      "step": 2605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004255860538388694,
+      "loss": 2.3444,
+      "step": 2606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004255317453694806,
+      "loss": 2.3704,
+      "step": 2607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004254774205575974,
+      "loss": 2.5536,
+      "step": 2608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004254230794082775,
+      "loss": 2.4657,
+      "step": 2609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042536872192658034,
+      "loss": 2.591,
+      "step": 2610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004253143481175667,
+      "loss": 2.2644,
+      "step": 2611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004252599579862989,
+      "loss": 2.5034,
+      "step": 2612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004252055515378408,
+      "loss": 2.3372,
+      "step": 2613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004251511287772579,
+      "loss": 2.2441,
+      "step": 2614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000425096689709617,
+      "loss": 2.5642,
+      "step": 2615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042504223433998664,
+      "loss": 2.4908,
+      "step": 2616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042498776267343665,
+      "loss": 2.4037,
+      "step": 2617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042493327471503854,
+      "loss": 2.558,
+      "step": 2618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004248787704698653,
+      "loss": 2.3554,
+      "step": 2619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042482424994299146,
+      "loss": 2.2294,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042476971313949296,
+      "loss": 2.2024,
+      "step": 2621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004247151600644473,
+      "loss": 2.3567,
+      "step": 2622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042466059072293367,
+      "loss": 2.5674,
+      "step": 2623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004246060051200325,
+      "loss": 2.6459,
+      "step": 2624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000424551403260826,
+      "loss": 2.504,
+      "step": 2625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042449678515039745,
+      "loss": 2.4508,
+      "step": 2626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004244421507938323,
+      "loss": 2.3817,
+      "step": 2627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042438750019621705,
+      "loss": 2.4226,
+      "step": 2628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004243328333626397,
+      "loss": 2.5485,
+      "step": 2629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042427815029819006,
+      "loss": 2.4487,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004242234510079591,
+      "loss": 2.6787,
+      "step": 2631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004241687354970396,
+      "loss": 2.4717,
+      "step": 2632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004241140037705258,
+      "loss": 2.5987,
+      "step": 2633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042405925583351313,
+      "loss": 2.413,
+      "step": 2634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004240044916910989,
+      "loss": 2.4099,
+      "step": 2635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004239497113483819,
+      "loss": 2.3064,
+      "step": 2636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042389491481046226,
+      "loss": 2.3931,
+      "step": 2637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004238401020824416,
+      "loss": 2.4388,
+      "step": 2638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004237852731694233,
+      "loss": 2.477,
+      "step": 2639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042373042807651194,
+      "loss": 2.6613,
+      "step": 2640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004236755668088139,
+      "loss": 2.7318,
+      "step": 2641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004236206893714368,
+      "loss": 2.3698,
+      "step": 2642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004235657957694899,
+      "loss": 2.5526,
+      "step": 2643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042351088600808395,
+      "loss": 2.3204,
+      "step": 2644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004234559600923312,
+      "loss": 2.4471,
+      "step": 2645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042340101802734547,
+      "loss": 2.3149,
+      "step": 2646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042334605981824195,
+      "loss": 2.6211,
+      "step": 2647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004232910854701374,
+      "loss": 2.3409,
+      "step": 2648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042323609498815006,
+      "loss": 2.4906,
+      "step": 2649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042318108837739987,
+      "loss": 2.3384,
+      "step": 2650
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 2.4694766998291016,
+      "eval_runtime": 184.683,
+      "eval_samples_per_second": 455.938,
+      "eval_steps_per_second": 28.497,
+      "step": 2650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004231260656430079,
+      "loss": 2.5097,
+      "step": 2651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042307102679009696,
+      "loss": 2.4279,
+      "step": 2652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004230159718237914,
+      "loss": 2.2936,
+      "step": 2653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004229609007492169,
+      "loss": 2.4127,
+      "step": 2654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004229058135715008,
+      "loss": 2.4712,
+      "step": 2655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004228507102957719,
+      "loss": 2.3665,
+      "step": 2656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042279559092716035,
+      "loss": 2.4669,
+      "step": 2657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042274045547079797,
+      "loss": 2.4755,
+      "step": 2658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042268530393181804,
+      "loss": 2.4935,
+      "step": 2659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004226301363153553,
+      "loss": 2.4915,
+      "step": 2660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004225749526265461,
+      "loss": 2.5638,
+      "step": 2661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000422519752870528,
+      "loss": 2.4148,
+      "step": 2662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004224645370524405,
+      "loss": 2.2232,
+      "step": 2663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004224093051774241,
+      "loss": 2.4206,
+      "step": 2664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042235405725062113,
+      "loss": 2.329,
+      "step": 2665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004222987932771755,
+      "loss": 2.5439,
+      "step": 2666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004222435132622322,
+      "loss": 2.572,
+      "step": 2667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042218821721093803,
+      "loss": 2.5361,
+      "step": 2668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042213290512844125,
+      "loss": 2.46,
+      "step": 2669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004220775770198916,
+      "loss": 2.362,
+      "step": 2670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042202223289044015,
+      "loss": 2.4211,
+      "step": 2671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042196687274523965,
+      "loss": 2.1845,
+      "step": 2672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004219114965894444,
+      "loss": 2.4536,
+      "step": 2673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042185610442820986,
+      "loss": 2.5328,
+      "step": 2674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004218006962666934,
+      "loss": 2.5966,
+      "step": 2675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042174527211005347,
+      "loss": 2.5275,
+      "step": 2676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004216898319634504,
+      "loss": 2.6437,
+      "step": 2677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042163437583204576,
+      "loss": 2.2885,
+      "step": 2678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004215789037210026,
+      "loss": 2.6107,
+      "step": 2679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042152341563548565,
+      "loss": 2.5112,
+      "step": 2680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004214679115806609,
+      "loss": 2.5155,
+      "step": 2681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042141239156169596,
+      "loss": 2.6012,
+      "step": 2682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042135685558375994,
+      "loss": 2.2994,
+      "step": 2683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042130130365202335,
+      "loss": 2.305,
+      "step": 2684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004212457357716583,
+      "loss": 2.4391,
+      "step": 2685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042119015194783815,
+      "loss": 2.5522,
+      "step": 2686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042113455218573814,
+      "loss": 2.7468,
+      "step": 2687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042107893649053456,
+      "loss": 2.4641,
+      "step": 2688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042102330486740554,
+      "loss": 2.2943,
+      "step": 2689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042096765732153044,
+      "loss": 2.4929,
+      "step": 2690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042091199385809017,
+      "loss": 2.4751,
+      "step": 2691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042085631448226726,
+      "loss": 2.4527,
+      "step": 2692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004208006191992455,
+      "loss": 2.4804,
+      "step": 2693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004207449080142104,
+      "loss": 2.4937,
+      "step": 2694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004206891809323488,
+      "loss": 2.3452,
+      "step": 2695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000420633437958849,
+      "loss": 2.312,
+      "step": 2696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042057767909890075,
+      "loss": 2.6752,
+      "step": 2697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042052190435769554,
+      "loss": 2.4153,
+      "step": 2698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004204661137404261,
+      "loss": 2.2408,
+      "step": 2699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042041030725228657,
+      "loss": 2.6808,
+      "step": 2700
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 2.463630437850952,
+      "eval_runtime": 184.6251,
+      "eval_samples_per_second": 456.081,
+      "eval_steps_per_second": 28.506,
+      "step": 2700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042035448489847284,
+      "loss": 2.463,
+      "step": 2701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042029864668418195,
+      "loss": 2.2995,
+      "step": 2702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004202427926146128,
+      "loss": 2.4368,
+      "step": 2703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004201869226949654,
+      "loss": 2.4587,
+      "step": 2704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004201310369304415,
+      "loss": 2.5454,
+      "step": 2705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004200751353262442,
+      "loss": 2.3458,
+      "step": 2706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042001921788757803,
+      "loss": 2.4745,
+      "step": 2707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041996328461964906,
+      "loss": 2.2681,
+      "step": 2708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004199073355276649,
+      "loss": 2.5998,
+      "step": 2709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041985137061683453,
+      "loss": 2.5399,
+      "step": 2710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041979538989236854,
+      "loss": 2.4158,
+      "step": 2711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041973939335947875,
+      "loss": 2.6307,
+      "step": 2712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041968338102337855,
+      "loss": 2.563,
+      "step": 2713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041962735288928306,
+      "loss": 2.5524,
+      "step": 2714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004195713089624085,
+      "loss": 2.4573,
+      "step": 2715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041951524924797276,
+      "loss": 2.4936,
+      "step": 2716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041945917375119514,
+      "loss": 2.4269,
+      "step": 2717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004194030824772964,
+      "loss": 2.4022,
+      "step": 2718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004193469754314988,
+      "loss": 2.5623,
+      "step": 2719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041929085261902617,
+      "loss": 2.4397,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004192347140451036,
+      "loss": 2.4306,
+      "step": 2721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004191785597149577,
+      "loss": 2.6301,
+      "step": 2722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004191223896338167,
+      "loss": 2.5426,
+      "step": 2723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041906620380691015,
+      "loss": 2.3944,
+      "step": 2724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041901000223946907,
+      "loss": 2.4861,
+      "step": 2725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004189537849367261,
+      "loss": 2.4189,
+      "step": 2726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004188975519039151,
+      "loss": 2.5574,
+      "step": 2727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004188413031462716,
+      "loss": 2.5007,
+      "step": 2728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004187850386690324,
+      "loss": 2.3644,
+      "step": 2729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000418728758477436,
+      "loss": 2.3759,
+      "step": 2730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00041867246257672224,
+      "loss": 2.7575,
+      "step": 2731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041861615097213233,
+      "loss": 2.491,
+      "step": 2732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041855982366890916,
+      "loss": 2.6779,
+      "step": 2733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004185034806722969,
+      "loss": 2.486,
+      "step": 2734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004184471219875412,
+      "loss": 2.4214,
+      "step": 2735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041839074761988927,
+      "loss": 2.2586,
+      "step": 2736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004183343575745897,
+      "loss": 2.2583,
+      "step": 2737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041827795185689256,
+      "loss": 2.5763,
+      "step": 2738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041822153047204935,
+      "loss": 2.3569,
+      "step": 2739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004181650934253132,
+      "loss": 2.3319,
+      "step": 2740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041810864072193833,
+      "loss": 2.5504,
+      "step": 2741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004180521723671807,
+      "loss": 2.3554,
+      "step": 2742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004179956883662979,
+      "loss": 2.2705,
+      "step": 2743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004179391887245484,
+      "loss": 2.549,
+      "step": 2744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041788267344719277,
+      "loss": 2.5784,
+      "step": 2745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041782614253949257,
+      "loss": 2.4374,
+      "step": 2746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041776959600671107,
+      "loss": 2.5397,
+      "step": 2747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004177130338541129,
+      "loss": 2.579,
+      "step": 2748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000417656456086964,
+      "loss": 2.3897,
+      "step": 2749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041759986271053213,
+      "loss": 2.4791,
+      "step": 2750
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 2.4529619216918945,
+      "eval_runtime": 184.8202,
+      "eval_samples_per_second": 455.6,
+      "eval_steps_per_second": 28.476,
+      "step": 2750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041754325373008624,
+      "loss": 2.4845,
+      "step": 2751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004174866291508967,
+      "loss": 2.4503,
+      "step": 2752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004174299889782355,
+      "loss": 2.5532,
+      "step": 2753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004173733332173759,
+      "loss": 2.4318,
+      "step": 2754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041731666187359287,
+      "loss": 2.3068,
+      "step": 2755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004172599749521625,
+      "loss": 2.3668,
+      "step": 2756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004172032724583626,
+      "loss": 2.5969,
+      "step": 2757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004171465543974723,
+      "loss": 2.2731,
+      "step": 2758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004170898207747722,
+      "loss": 2.4227,
+      "step": 2759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041703307159554436,
+      "loss": 2.2375,
+      "step": 2760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041697630686507236,
+      "loss": 2.5563,
+      "step": 2761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004169195265886411,
+      "loss": 2.4005,
+      "step": 2762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004168627307715369,
+      "loss": 2.4796,
+      "step": 2763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041680591941904764,
+      "loss": 2.5789,
+      "step": 2764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004167490925364628,
+      "loss": 2.4727,
+      "step": 2765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004166922501290729,
+      "loss": 2.4628,
+      "step": 2766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004166353922021702,
+      "loss": 2.3557,
+      "step": 2767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004165785187610484,
+      "loss": 2.4065,
+      "step": 2768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004165216298110025,
+      "loss": 2.4746,
+      "step": 2769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041646472535732896,
+      "loss": 2.4414,
+      "step": 2770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041640780540532586,
+      "loss": 2.5162,
+      "step": 2771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041635086996029256,
+      "loss": 2.4721,
+      "step": 2772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041629391902753,
+      "loss": 2.3813,
+      "step": 2773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041623695261234025,
+      "loss": 2.5101,
+      "step": 2774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041617997072002723,
+      "loss": 2.409,
+      "step": 2775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000416122973355896,
+      "loss": 2.5775,
+      "step": 2776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004160659605252533,
+      "loss": 2.2465,
+      "step": 2777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041600893223340705,
+      "loss": 2.3459,
+      "step": 2778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004159518884856669,
+      "loss": 2.2348,
+      "step": 2779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041589482928734364,
+      "loss": 2.2284,
+      "step": 2780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004158377546437496,
+      "loss": 2.1761,
+      "step": 2781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004157806645601988,
+      "loss": 2.6736,
+      "step": 2782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004157235590420063,
+      "loss": 2.3734,
+      "step": 2783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004156664380944889,
+      "loss": 2.3147,
+      "step": 2784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041560930172296464,
+      "loss": 2.2521,
+      "step": 2785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041555214993275303,
+      "loss": 2.3668,
+      "step": 2786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041549498272917525,
+      "loss": 2.5298,
+      "step": 2787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004154378001175535,
+      "loss": 2.431,
+      "step": 2788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004153806021032118,
+      "loss": 2.6258,
+      "step": 2789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004153233886914754,
+      "loss": 2.4525,
+      "step": 2790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041526615988767105,
+      "loss": 2.5575,
+      "step": 2791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041520891569712677,
+      "loss": 2.4345,
+      "step": 2792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041515165612517237,
+      "loss": 2.3174,
+      "step": 2793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041509438117713865,
+      "loss": 2.4024,
+      "step": 2794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041503709085835826,
+      "loss": 2.2027,
+      "step": 2795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000414979785174165,
+      "loss": 2.3692,
+      "step": 2796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041492246412989426,
+      "loss": 2.4305,
+      "step": 2797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004148651277308827,
+      "loss": 2.4623,
+      "step": 2798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041480777598246855,
+      "loss": 2.2673,
+      "step": 2799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041475040888999134,
+      "loss": 2.4134,
+      "step": 2800
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 2.442607879638672,
+      "eval_runtime": 184.6756,
+      "eval_samples_per_second": 455.956,
+      "eval_steps_per_second": 28.499,
+      "step": 2800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004146930264587922,
+      "loss": 2.2662,
+      "step": 2801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041463562869421355,
+      "loss": 2.4417,
+      "step": 2802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041457821560159926,
+      "loss": 2.6167,
+      "step": 2803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004145207871862947,
+      "loss": 2.4732,
+      "step": 2804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041446334345364666,
+      "loss": 2.4831,
+      "step": 2805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004144058844090032,
+      "loss": 2.3909,
+      "step": 2806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041434841005771395,
+      "loss": 2.387,
+      "step": 2807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004142909204051299,
+      "loss": 2.3636,
+      "step": 2808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004142334154566036,
+      "loss": 2.467,
+      "step": 2809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041417589521748887,
+      "loss": 2.341,
+      "step": 2810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000414118359693141,
+      "loss": 2.3338,
+      "step": 2811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004140608088889167,
+      "loss": 2.1604,
+      "step": 2812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004140032428101741,
+      "loss": 2.3361,
+      "step": 2813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041394566146227274,
+      "loss": 2.4765,
+      "step": 2814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041388806485057376,
+      "loss": 2.3079,
+      "step": 2815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004138304529804393,
+      "loss": 2.3887,
+      "step": 2816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004137728258572334,
+      "loss": 2.4868,
+      "step": 2817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004137151834863213,
+      "loss": 2.5119,
+      "step": 2818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004136575258730695,
+      "loss": 2.4261,
+      "step": 2819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041359985302284627,
+      "loss": 2.3311,
+      "step": 2820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041354216494102105,
+      "loss": 2.3222,
+      "step": 2821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004134844616329646,
+      "loss": 2.4346,
+      "step": 2822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004134267431040495,
+      "loss": 2.1446,
+      "step": 2823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041336900935964936,
+      "loss": 2.3276,
+      "step": 2824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041331126040513943,
+      "loss": 2.3887,
+      "step": 2825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041325349624589625,
+      "loss": 2.2445,
+      "step": 2826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004131957168872979,
+      "loss": 2.4325,
+      "step": 2827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004131379223347237,
+      "loss": 2.4484,
+      "step": 2828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041308011259355457,
+      "loss": 2.3598,
+      "step": 2829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004130222876691726,
+      "loss": 2.5135,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004129644475669616,
+      "loss": 2.4736,
+      "step": 2831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004129065922923067,
+      "loss": 2.3444,
+      "step": 2832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041284872185059417,
+      "loss": 2.6311,
+      "step": 2833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041279083624721217,
+      "loss": 2.5707,
+      "step": 2834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041273293548754975,
+      "loss": 2.4337,
+      "step": 2835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041267501957699783,
+      "loss": 2.4094,
+      "step": 2836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004126170885209485,
+      "loss": 2.4252,
+      "step": 2837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004125591423247952,
+      "loss": 2.418,
+      "step": 2838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041250118099393296,
+      "loss": 2.3438,
+      "step": 2839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041244320453375817,
+      "loss": 2.4975,
+      "step": 2840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041238521294966856,
+      "loss": 2.3783,
+      "step": 2841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004123272062470633,
+      "loss": 2.4876,
+      "step": 2842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041226918443134297,
+      "loss": 2.4001,
+      "step": 2843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041221114750790964,
+      "loss": 2.4595,
+      "step": 2844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004121530954821666,
+      "loss": 2.304,
+      "step": 2845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004120950283595188,
+      "loss": 2.2523,
+      "step": 2846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041203694614537224,
+      "loss": 2.2123,
+      "step": 2847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004119788488451347,
+      "loss": 2.6257,
+      "step": 2848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041192073646421516,
+      "loss": 2.49,
+      "step": 2849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000411862609008024,
+      "loss": 2.2267,
+      "step": 2850
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 2.4315478801727295,
+      "eval_runtime": 184.9078,
+      "eval_samples_per_second": 455.384,
+      "eval_steps_per_second": 28.463,
+      "step": 2850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041180446648197316,
+      "loss": 2.4019,
+      "step": 2851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004117463088914758,
+      "loss": 2.3511,
+      "step": 2852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041168813624194644,
+      "loss": 2.33,
+      "step": 2853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004116299485388014,
+      "loss": 2.5349,
+      "step": 2854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041157174578745784,
+      "loss": 2.397,
+      "step": 2855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004115135279933346,
+      "loss": 2.2585,
+      "step": 2856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041145529516185223,
+      "loss": 2.6454,
+      "step": 2857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000411397047298432,
+      "loss": 2.3407,
+      "step": 2858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041133878440849717,
+      "loss": 2.4282,
+      "step": 2859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004112805064974722,
+      "loss": 2.4459,
+      "step": 2860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041122221357078274,
+      "loss": 2.346,
+      "step": 2861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004111639056338561,
+      "loss": 2.4763,
+      "step": 2862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041110558269212105,
+      "loss": 2.4345,
+      "step": 2863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041104724475100744,
+      "loss": 2.6201,
+      "step": 2864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041098889181594666,
+      "loss": 2.365,
+      "step": 2865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004109305238923718,
+      "loss": 2.487,
+      "step": 2866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004108721409857168,
+      "loss": 2.4694,
+      "step": 2867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041081374310141753,
+      "loss": 2.4712,
+      "step": 2868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041075533024491063,
+      "loss": 2.2188,
+      "step": 2869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004106969024216348,
+      "loss": 2.5285,
+      "step": 2870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041063845963702983,
+      "loss": 2.5344,
+      "step": 2871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041058000189653676,
+      "loss": 2.4763,
+      "step": 2872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004105215292055982,
+      "loss": 2.4906,
+      "step": 2873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004104630415696582,
+      "loss": 2.4573,
+      "step": 2874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004104045389941621,
+      "loss": 2.4561,
+      "step": 2875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041034602148455656,
+      "loss": 2.319,
+      "step": 2876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041028748904628975,
+      "loss": 2.3784,
+      "step": 2877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004102289416848114,
+      "loss": 2.2666,
+      "step": 2878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041017037940557213,
+      "loss": 2.3214,
+      "step": 2879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041011180221402443,
+      "loss": 2.4798,
+      "step": 2880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000410053210115622,
+      "loss": 2.5777,
+      "step": 2881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040999460311581993,
+      "loss": 2.4334,
+      "step": 2882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004099359812200746,
+      "loss": 2.4872,
+      "step": 2883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040987734443384387,
+      "loss": 2.3438,
+      "step": 2884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004098186927625872,
+      "loss": 2.3977,
+      "step": 2885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000409760026211765,
+      "loss": 2.3741,
+      "step": 2886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040970134478683927,
+      "loss": 2.4222,
+      "step": 2887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040964264849327367,
+      "loss": 2.398,
+      "step": 2888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004095839373365327,
+      "loss": 2.4123,
+      "step": 2889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004095252113220827,
+      "loss": 2.2715,
+      "step": 2890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004094664704553912,
+      "loss": 2.5996,
+      "step": 2891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004094077147419271,
+      "loss": 2.4034,
+      "step": 2892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004093489441871607,
+      "loss": 2.45,
+      "step": 2893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004092901587965638,
+      "loss": 2.4265,
+      "step": 2894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040923135857560944,
+      "loss": 2.2283,
+      "step": 2895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040917254352977204,
+      "loss": 2.5059,
+      "step": 2896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040911371366452755,
+      "loss": 2.5562,
+      "step": 2897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000409054868985353,
+      "loss": 2.2615,
+      "step": 2898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004089960094977272,
+      "loss": 2.4566,
+      "step": 2899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040893713520713006,
+      "loss": 2.4479,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 2.426853895187378,
+      "eval_runtime": 184.8325,
+      "eval_samples_per_second": 455.569,
+      "eval_steps_per_second": 28.474,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040887824611904284,
+      "loss": 2.3504,
+      "step": 2901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004088193422389484,
+      "loss": 2.3467,
+      "step": 2902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040876042357233084,
+      "loss": 2.3943,
+      "step": 2903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040870149012467563,
+      "loss": 2.2543,
+      "step": 2904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040864254190146963,
+      "loss": 2.417,
+      "step": 2905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040858357890820107,
+      "loss": 2.3816,
+      "step": 2906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004085246011503596,
+      "loss": 2.6208,
+      "step": 2907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004084656086334363,
+      "loss": 2.3652,
+      "step": 2908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004084066013629233,
+      "loss": 2.3252,
+      "step": 2909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004083475793443146,
+      "loss": 2.3963,
+      "step": 2910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040828854258310516,
+      "loss": 2.3044,
+      "step": 2911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040822949108479145,
+      "loss": 2.3561,
+      "step": 2912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004081704248548715,
+      "loss": 2.3998,
+      "step": 2913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004081113438988443,
+      "loss": 2.3302,
+      "step": 2914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040805224822221064,
+      "loss": 2.2362,
+      "step": 2915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040799313783047245,
+      "loss": 2.5623,
+      "step": 2916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000407934012729133,
+      "loss": 2.6148,
+      "step": 2917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004078748729236971,
+      "loss": 2.5312,
+      "step": 2918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004078157184196708,
+      "loss": 2.3092,
+      "step": 2919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004077565492225615,
+      "loss": 2.2171,
+      "step": 2920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040769736533787806,
+      "loss": 2.4662,
+      "step": 2921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040763816677113064,
+      "loss": 2.2921,
+      "step": 2922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004075789535278308,
+      "loss": 2.4946,
+      "step": 2923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004075197256134915,
+      "loss": 2.2064,
+      "step": 2924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040746048303362694,
+      "loss": 2.3953,
+      "step": 2925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040740122579375283,
+      "loss": 2.421,
+      "step": 2926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004073419538993862,
+      "loss": 2.3809,
+      "step": 2927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004072826673560454,
+      "loss": 2.5073,
+      "step": 2928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040722336616925015,
+      "loss": 2.5482,
+      "step": 2929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040716405034452164,
+      "loss": 2.3723,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040710471988738217,
+      "loss": 2.4285,
+      "step": 2931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004070453748033557,
+      "loss": 2.3279,
+      "step": 2932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004069860150979675,
+      "loss": 2.4001,
+      "step": 2933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004069266407767439,
+      "loss": 2.4011,
+      "step": 2934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000406867251845213,
+      "loss": 2.3702,
+      "step": 2935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040680784830890405,
+      "loss": 2.5215,
+      "step": 2936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040674843017334753,
+      "loss": 2.7507,
+      "step": 2937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004066889974440757,
+      "loss": 2.3772,
+      "step": 2938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004066295501266216,
+      "loss": 2.3837,
+      "step": 2939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004065700882265202,
+      "loss": 2.5018,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040651061174930747,
+      "loss": 2.1896,
+      "step": 2941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040645112070052087,
+      "loss": 2.4184,
+      "step": 2942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004063916150856991,
+      "loss": 2.2838,
+      "step": 2943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040633209491038237,
+      "loss": 2.4305,
+      "step": 2944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040627256018011215,
+      "loss": 2.4883,
+      "step": 2945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040621301090043127,
+      "loss": 2.6155,
+      "step": 2946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040615344707688406,
+      "loss": 2.5072,
+      "step": 2947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040609386871501583,
+      "loss": 2.3592,
+      "step": 2948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004060342758203737,
+      "loss": 2.5406,
+      "step": 2949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040597466839850595,
+      "loss": 2.4216,
+      "step": 2950
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 2.41066837310791,
+      "eval_runtime": 184.6972,
+      "eval_samples_per_second": 455.903,
+      "eval_steps_per_second": 28.495,
+      "step": 2950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000405915046454962,
+      "loss": 2.4336,
+      "step": 2951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040585540999529303,
+      "loss": 2.4355,
+      "step": 2952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040579575902505117,
+      "loss": 2.2945,
+      "step": 2953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004057360935497902,
+      "loss": 2.4019,
+      "step": 2954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040567641357506525,
+      "loss": 2.3893,
+      "step": 2955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004056167191064325,
+      "loss": 2.5066,
+      "step": 2956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040555701014944966,
+      "loss": 2.5613,
+      "step": 2957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000405497286709676,
+      "loss": 2.2913,
+      "step": 2958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004054375487926718,
+      "loss": 2.4079,
+      "step": 2959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040537779640399885,
+      "loss": 2.507,
+      "step": 2960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004053180295492203,
+      "loss": 2.3555,
+      "step": 2961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004052582482339004,
+      "loss": 2.3949,
+      "step": 2962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004051984524636053,
+      "loss": 2.519,
+      "step": 2963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000405138642243902,
+      "loss": 2.3651,
+      "step": 2964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004050788175803589,
+      "loss": 2.1994,
+      "step": 2965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040501897847854595,
+      "loss": 2.4957,
+      "step": 2966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004049591249440343,
+      "loss": 2.3351,
+      "step": 2967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004048992569823965,
+      "loss": 2.3097,
+      "step": 2968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004048393745992064,
+      "loss": 2.6435,
+      "step": 2969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004047794778000393,
+      "loss": 2.4161,
+      "step": 2970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004047195665904716,
+      "loss": 2.39,
+      "step": 2971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004046596409760814,
+      "loss": 2.2653,
+      "step": 2972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004045997009624477,
+      "loss": 2.1988,
+      "step": 2973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004045397465551513,
+      "loss": 2.3287,
+      "step": 2974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004044797777597741,
+      "loss": 2.449,
+      "step": 2975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040441979458189913,
+      "loss": 2.2121,
+      "step": 2976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040435979702711127,
+      "loss": 2.5275,
+      "step": 2977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004042997851009964,
+      "loss": 2.5788,
+      "step": 2978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040423975880914164,
+      "loss": 2.4802,
+      "step": 2979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004041797181571358,
+      "loss": 2.4622,
+      "step": 2980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004041196631505687,
+      "loss": 2.4394,
+      "step": 2981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040405959379503166,
+      "loss": 2.3219,
+      "step": 2982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040399951009611736,
+      "loss": 2.4813,
+      "step": 2983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040393941205941965,
+      "loss": 2.2684,
+      "step": 2984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040387929969053395,
+      "loss": 2.3934,
+      "step": 2985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004038191729950569,
+      "loss": 2.1429,
+      "step": 2986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004037590319785863,
+      "loss": 2.488,
+      "step": 2987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004036988766467215,
+      "loss": 2.2036,
+      "step": 2988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004036387070050632,
+      "loss": 2.4081,
+      "step": 2989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004035785230592134,
+      "loss": 2.4288,
+      "step": 2990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040351832481477524,
+      "loss": 2.5045,
+      "step": 2991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004034581122773535,
+      "loss": 2.409,
+      "step": 2992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040339788545255405,
+      "loss": 2.4364,
+      "step": 2993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004033376443459842,
+      "loss": 2.3076,
+      "step": 2994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004032773889632525,
+      "loss": 2.3957,
+      "step": 2995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040321711930996895,
+      "loss": 2.483,
+      "step": 2996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004031568353917449,
+      "loss": 2.3225,
+      "step": 2997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004030965372141927,
+      "loss": 2.4575,
+      "step": 2998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040303622478292655,
+      "loss": 2.4242,
+      "step": 2999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040297589810356165,
+      "loss": 2.4072,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 2.399169921875,
+      "eval_runtime": 184.817,
+      "eval_samples_per_second": 455.607,
+      "eval_steps_per_second": 28.477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004029155571817145,
+      "loss": 2.5464,
+      "step": 3001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040285520202300307,
+      "loss": 2.4515,
+      "step": 3002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004027948326330465,
+      "loss": 2.3849,
+      "step": 3003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004027344490174655,
+      "loss": 2.3028,
+      "step": 3004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040267405118188176,
+      "loss": 2.3259,
+      "step": 3005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004026136391319187,
+      "loss": 2.423,
+      "step": 3006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040255321287320067,
+      "loss": 2.4871,
+      "step": 3007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040249277241135364,
+      "loss": 2.3403,
+      "step": 3008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004024323177520047,
+      "loss": 2.4441,
+      "step": 3009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040237184890078246,
+      "loss": 2.4412,
+      "step": 3010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004023113658633166,
+      "loss": 2.6021,
+      "step": 3011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004022508686452384,
+      "loss": 2.2316,
+      "step": 3012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004021903572521802,
+      "loss": 2.405,
+      "step": 3013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004021298316897758,
+      "loss": 2.3523,
+      "step": 3014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004020692919636604,
+      "loss": 2.1971,
+      "step": 3015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040200873807947024,
+      "loss": 2.3287,
+      "step": 3016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004019481700428432,
+      "loss": 2.0583,
+      "step": 3017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004018875878594184,
+      "loss": 2.3305,
+      "step": 3018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040182699153483594,
+      "loss": 2.3616,
+      "step": 3019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040176638107473775,
+      "loss": 2.576,
+      "step": 3020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004017057564847667,
+      "loss": 2.4302,
+      "step": 3021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040164511777056723,
+      "loss": 2.3701,
+      "step": 3022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004015844649377849,
+      "loss": 2.2912,
+      "step": 3023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004015237979920666,
+      "loss": 2.3938,
+      "step": 3024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040146311693906066,
+      "loss": 2.2067,
+      "step": 3025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040140242178441667,
+      "loss": 2.3033,
+      "step": 3026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004013417125337855,
+      "loss": 2.3754,
+      "step": 3027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040128098919281935,
+      "loss": 2.5713,
+      "step": 3028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004012202517671717,
+      "loss": 2.3358,
+      "step": 3029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004011595002624974,
+      "loss": 2.392,
+      "step": 3030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040109873468445256,
+      "loss": 2.3969,
+      "step": 3031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040103795503869464,
+      "loss": 2.4464,
+      "step": 3032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040097716133088247,
+      "loss": 2.3732,
+      "step": 3033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040091635356667607,
+      "loss": 2.1109,
+      "step": 3034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004008555317517367,
+      "loss": 2.4987,
+      "step": 3035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004007946958917272,
+      "loss": 2.2953,
+      "step": 3036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040073384599231143,
+      "loss": 2.3216,
+      "step": 3037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040067298205915473,
+      "loss": 2.2033,
+      "step": 3038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004006121040979238,
+      "loss": 2.4415,
+      "step": 3039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040055121211428633,
+      "loss": 2.3863,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004004903061139118,
+      "loss": 2.3519,
+      "step": 3041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004004293861024705,
+      "loss": 2.5066,
+      "step": 3042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040036845208563446,
+      "loss": 2.3507,
+      "step": 3043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040030750406907655,
+      "loss": 2.3254,
+      "step": 3044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040024654205847145,
+      "loss": 2.3447,
+      "step": 3045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040018556605949475,
+      "loss": 2.5246,
+      "step": 3046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004001245760778235,
+      "loss": 2.5597,
+      "step": 3047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040006357211913606,
+      "loss": 2.3335,
+      "step": 3048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00040000255418911213,
+      "loss": 2.3335,
+      "step": 3049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039994152229343253,
+      "loss": 2.3671,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 2.4000086784362793,
+      "eval_runtime": 184.6529,
+      "eval_samples_per_second": 456.012,
+      "eval_steps_per_second": 28.502,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003998804764377796,
+      "loss": 2.4816,
+      "step": 3051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003998194166278367,
+      "loss": 2.1922,
+      "step": 3052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039975834286928895,
+      "loss": 2.3366,
+      "step": 3053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003996972551678223,
+      "loss": 2.4222,
+      "step": 3054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003996361535291242,
+      "loss": 2.3344,
+      "step": 3055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039957503795888343,
+      "loss": 2.221,
+      "step": 3056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039951390846279,
+      "loss": 2.2554,
+      "step": 3057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003994527650465352,
+      "loss": 2.2855,
+      "step": 3058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003993916077158117,
+      "loss": 2.3901,
+      "step": 3059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039933043647631347,
+      "loss": 2.3086,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003992692513337356,
+      "loss": 2.3057,
+      "step": 3061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039920805229377463,
+      "loss": 2.3494,
+      "step": 3062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039914683936212843,
+      "loss": 2.3131,
+      "step": 3063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039908561254449606,
+      "loss": 2.2318,
+      "step": 3064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003990243718465779,
+      "loss": 2.422,
+      "step": 3065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039896311727407555,
+      "loss": 2.375,
+      "step": 3066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003989018488326921,
+      "loss": 2.4839,
+      "step": 3067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039884056652813183,
+      "loss": 2.3161,
+      "step": 3068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003987792703661001,
+      "loss": 2.2369,
+      "step": 3069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000398717960352304,
+      "loss": 2.5044,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003986566364924515,
+      "loss": 2.2631,
+      "step": 3071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003985952987922521,
+      "loss": 2.0219,
+      "step": 3072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039853394725741656,
+      "loss": 2.3653,
+      "step": 3073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039847258189365666,
+      "loss": 2.241,
+      "step": 3074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039841120270668585,
+      "loss": 2.2192,
+      "step": 3075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003983498097022188,
+      "loss": 2.6285,
+      "step": 3076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003982884028859711,
+      "loss": 2.5181,
+      "step": 3077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039822698226366017,
+      "loss": 2.4472,
+      "step": 3078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039816554784100424,
+      "loss": 2.3901,
+      "step": 3079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039810409962372304,
+      "loss": 2.3938,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003980426376175378,
+      "loss": 2.3112,
+      "step": 3081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039798116182817054,
+      "loss": 2.4919,
+      "step": 3082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039791967226134495,
+      "loss": 2.4731,
+      "step": 3083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039785816892278587,
+      "loss": 2.425,
+      "step": 3084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039779665181821936,
+      "loss": 2.2964,
+      "step": 3085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039773512095337294,
+      "loss": 2.4719,
+      "step": 3086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039767357633397525,
+      "loss": 2.4345,
+      "step": 3087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003976120179657563,
+      "loss": 2.2491,
+      "step": 3088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003975504458544472,
+      "loss": 2.3062,
+      "step": 3089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039748886000578077,
+      "loss": 2.2135,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039742726042549053,
+      "loss": 2.6586,
+      "step": 3091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003973656471193118,
+      "loss": 2.2103,
+      "step": 3092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039730402009298074,
+      "loss": 2.3757,
+      "step": 3093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003972423793522352,
+      "loss": 2.2923,
+      "step": 3094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000397180724902814,
+      "loss": 2.2433,
+      "step": 3095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039711905675045723,
+      "loss": 2.5166,
+      "step": 3096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003970573749009065,
+      "loss": 2.4022,
+      "step": 3097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003969956793599048,
+      "loss": 2.4045,
+      "step": 3098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039693397013319565,
+      "loss": 2.2589,
+      "step": 3099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003968722472265247,
+      "loss": 2.2338,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 2.389512777328491,
+      "eval_runtime": 184.8643,
+      "eval_samples_per_second": 455.491,
+      "eval_steps_per_second": 28.47,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039681051064563845,
+      "loss": 2.4158,
+      "step": 3101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003967487603962847,
+      "loss": 2.3376,
+      "step": 3102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003966869964842126,
+      "loss": 2.2575,
+      "step": 3103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039662521891517256,
+      "loss": 2.4464,
+      "step": 3104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039656342769491627,
+      "loss": 2.4287,
+      "step": 3105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039650162282919653,
+      "loss": 2.3642,
+      "step": 3106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003964398043237677,
+      "loss": 2.4275,
+      "step": 3107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003963779721843852,
+      "loss": 2.2224,
+      "step": 3108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003963161264168057,
+      "loss": 2.3533,
+      "step": 3109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039625426702678736,
+      "loss": 2.3488,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003961923940200893,
+      "loss": 2.2437,
+      "step": 3111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039613050740247227,
+      "loss": 2.3849,
+      "step": 3112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003960686071796978,
+      "loss": 2.2877,
+      "step": 3113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039600669335752923,
+      "loss": 2.3348,
+      "step": 3114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003959447659417309,
+      "loss": 2.3741,
+      "step": 3115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003958828249380683,
+      "loss": 2.4697,
+      "step": 3116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003958208703523083,
+      "loss": 2.2761,
+      "step": 3117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003957589021902191,
+      "loss": 2.4146,
+      "step": 3118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003956969204575701,
+      "loss": 2.2333,
+      "step": 3119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000395634925160132,
+      "loss": 2.1982,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003955729163036768,
+      "loss": 2.2828,
+      "step": 3121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003955108938939774,
+      "loss": 2.306,
+      "step": 3122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003954488579368087,
+      "loss": 2.268,
+      "step": 3123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039538680843794605,
+      "loss": 2.384,
+      "step": 3124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039532474540316667,
+      "loss": 2.4837,
+      "step": 3125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039526266883824863,
+      "loss": 2.2416,
+      "step": 3126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039520057874897156,
+      "loss": 2.3003,
+      "step": 3127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000395138475141116,
+      "loss": 2.4073,
+      "step": 3128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950763580204643,
+      "loss": 2.4814,
+      "step": 3129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950142273927996,
+      "loss": 2.2835,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003949520832639063,
+      "loss": 2.365,
+      "step": 3131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948899256395703,
+      "loss": 2.3767,
+      "step": 3132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039482775452557873,
+      "loss": 2.2796,
+      "step": 3133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003947655699277197,
+      "loss": 2.304,
+      "step": 3134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039470337185178297,
+      "loss": 2.4341,
+      "step": 3135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003946411603035592,
+      "loss": 2.3067,
+      "step": 3136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003945789352888406,
+      "loss": 2.4717,
+      "step": 3137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039451669681342036,
+      "loss": 2.2427,
+      "step": 3138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003944544448830931,
+      "loss": 2.3956,
+      "step": 3139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003943921795036547,
+      "loss": 2.3899,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003943299006809021,
+      "loss": 2.3059,
+      "step": 3141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003942676084206338,
+      "loss": 2.3271,
+      "step": 3142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039420530272864934,
+      "loss": 2.5184,
+      "step": 3143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039414298361074954,
+      "loss": 2.4149,
+      "step": 3144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003940806510727364,
+      "loss": 2.3953,
+      "step": 3145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039401830512041325,
+      "loss": 2.145,
+      "step": 3146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039395594575958483,
+      "loss": 2.43,
+      "step": 3147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039389357299605694,
+      "loss": 2.3992,
+      "step": 3148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003938311868356365,
+      "loss": 2.2919,
+      "step": 3149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000393768787284132,
+      "loss": 2.3592,
+      "step": 3150
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 2.378669261932373,
+      "eval_runtime": 184.8325,
+      "eval_samples_per_second": 455.569,
+      "eval_steps_per_second": 28.474,
+      "step": 3150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039370637434735294,
+      "loss": 2.3382,
+      "step": 3151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039364394803111016,
+      "loss": 2.1927,
+      "step": 3152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039358150834121564,
+      "loss": 2.5552,
+      "step": 3153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039351905528348287,
+      "loss": 2.2935,
+      "step": 3154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039345658886372615,
+      "loss": 2.3067,
+      "step": 3155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039339410908776154,
+      "loss": 2.4317,
+      "step": 3156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003933316159614059,
+      "loss": 2.2633,
+      "step": 3157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039326910949047763,
+      "loss": 2.4229,
+      "step": 3158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039320658968079617,
+      "loss": 2.2509,
+      "step": 3159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003931440565381823,
+      "loss": 2.3787,
+      "step": 3160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039308151006845805,
+      "loss": 2.2516,
+      "step": 3161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003930189502774467,
+      "loss": 2.298,
+      "step": 3162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039295637717097275,
+      "loss": 2.1708,
+      "step": 3163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003928937907548619,
+      "loss": 2.5029,
+      "step": 3164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000392831191034941,
+      "loss": 2.2686,
+      "step": 3165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039276857801703847,
+      "loss": 2.3635,
+      "step": 3166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003927059517069836,
+      "loss": 2.32,
+      "step": 3167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003926433121106072,
+      "loss": 2.3051,
+      "step": 3168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000392580659233741,
+      "loss": 2.4165,
+      "step": 3169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039251799308221834,
+      "loss": 2.5326,
+      "step": 3170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003924553136618736,
+      "loss": 2.5812,
+      "step": 3171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039239262097854233,
+      "loss": 2.3013,
+      "step": 3172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003923299150380615,
+      "loss": 2.2095,
+      "step": 3173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000392267195846269,
+      "loss": 2.4235,
+      "step": 3174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039220446340900437,
+      "loss": 2.3194,
+      "step": 3175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039214171773210805,
+      "loss": 2.3108,
+      "step": 3176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003920789588214219,
+      "loss": 2.3261,
+      "step": 3177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039201618668278893,
+      "loss": 2.4213,
+      "step": 3178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003919534013220535,
+      "loss": 2.4099,
+      "step": 3179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003918906027450609,
+      "loss": 2.2705,
+      "step": 3180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039182779095765796,
+      "loss": 2.407,
+      "step": 3181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039176496596569265,
+      "loss": 2.2541,
+      "step": 3182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003917021277750142,
+      "loss": 2.4169,
+      "step": 3183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003916392763914729,
+      "loss": 2.2235,
+      "step": 3184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039157641182092044,
+      "loss": 2.3853,
+      "step": 3185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039151353406920977,
+      "loss": 2.3773,
+      "step": 3186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039145064314219483,
+      "loss": 2.3342,
+      "step": 3187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000391387739045731,
+      "loss": 2.2106,
+      "step": 3188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039132482178567485,
+      "loss": 2.5172,
+      "step": 3189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039126189136788413,
+      "loss": 2.3326,
+      "step": 3190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003911989477982179,
+      "loss": 2.5678,
+      "step": 3191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039113599108253627,
+      "loss": 2.332,
+      "step": 3192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039107302122670087,
+      "loss": 2.3413,
+      "step": 3193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039101003823657413,
+      "loss": 2.2944,
+      "step": 3194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003909470421180201,
+      "loss": 2.3895,
+      "step": 3195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039088403287690387,
+      "loss": 2.4995,
+      "step": 3196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003908210105190917,
+      "loss": 2.3773,
+      "step": 3197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003907579750504513,
+      "loss": 2.4436,
+      "step": 3198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003906949264768512,
+      "loss": 2.3077,
+      "step": 3199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039063186480416167,
+      "loss": 2.5145,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 2.372828722000122,
+      "eval_runtime": 184.7074,
+      "eval_samples_per_second": 455.878,
+      "eval_steps_per_second": 28.494,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039056879003825383,
+      "loss": 2.2701,
+      "step": 3201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000390505702185,
+      "loss": 2.2887,
+      "step": 3202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039044260125027403,
+      "loss": 2.3147,
+      "step": 3203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039037948723995064,
+      "loss": 2.297,
+      "step": 3204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000390316360159906,
+      "loss": 2.2891,
+      "step": 3205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039025322001601736,
+      "loss": 2.3772,
+      "step": 3206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003901900668141633,
+      "loss": 2.3895,
+      "step": 3207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003901269005602235,
+      "loss": 2.3872,
+      "step": 3208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039006372126007896,
+      "loss": 2.3399,
+      "step": 3209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039000052891961193,
+      "loss": 2.4588,
+      "step": 3210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038993732354470554,
+      "loss": 2.4609,
+      "step": 3211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003898741051412446,
+      "loss": 2.3362,
+      "step": 3212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038981087371511494,
+      "loss": 2.4699,
+      "step": 3213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003897476292722034,
+      "loss": 2.3208,
+      "step": 3214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038968437181839834,
+      "loss": 2.2569,
+      "step": 3215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003896211013595892,
+      "loss": 2.3025,
+      "step": 3216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038955781790166665,
+      "loss": 2.3765,
+      "step": 3217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038949452145052257,
+      "loss": 2.324,
+      "step": 3218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003894312120120499,
+      "loss": 2.1265,
+      "step": 3219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038936788959214295,
+      "loss": 2.1584,
+      "step": 3220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038930455419669747,
+      "loss": 2.5858,
+      "step": 3221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003892412058316098,
+      "loss": 2.2764,
+      "step": 3222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038917784450277814,
+      "loss": 2.3658,
+      "step": 3223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003891144702161015,
+      "loss": 2.4039,
+      "step": 3224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038905108297748015,
+      "loss": 2.3985,
+      "step": 3225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003889876827928156,
+      "loss": 2.5325,
+      "step": 3226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038892426966801076,
+      "loss": 2.4502,
+      "step": 3227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038886084360896934,
+      "loss": 2.3491,
+      "step": 3228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003887974046215967,
+      "loss": 2.5079,
+      "step": 3229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000388733952711799,
+      "loss": 2.5168,
+      "step": 3230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038867048788548394,
+      "loss": 2.3794,
+      "step": 3231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003886070101485601,
+      "loss": 2.231,
+      "step": 3232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038854351950693767,
+      "loss": 2.4875,
+      "step": 3233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003884800159665276,
+      "loss": 2.4969,
+      "step": 3234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038841649953324234,
+      "loss": 2.412,
+      "step": 3235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038835297021299535,
+      "loss": 2.228,
+      "step": 3236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038828942801170154,
+      "loss": 2.4448,
+      "step": 3237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038822587293527676,
+      "loss": 2.3338,
+      "step": 3238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003881623049896382,
+      "loss": 2.4193,
+      "step": 3239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038809872418070425,
+      "loss": 2.5187,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038803513051439444,
+      "loss": 2.3551,
+      "step": 3241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003879715239966294,
+      "loss": 2.2863,
+      "step": 3242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038790790463333124,
+      "loss": 2.3136,
+      "step": 3243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000387844272430423,
+      "loss": 2.2961,
+      "step": 3244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038778062739382895,
+      "loss": 2.1507,
+      "step": 3245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00038771696952947485,
+      "loss": 2.2619,
+      "step": 3246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003876532988432872,
+      "loss": 2.2902,
+      "step": 3247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003875896153411941,
+      "loss": 2.304,
+      "step": 3248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003875259190291245,
+      "loss": 2.1333,
+      "step": 3249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003874622099130087,
+      "loss": 2.6603,
+      "step": 3250
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 2.3650050163269043,
+      "eval_runtime": 184.8386,
+      "eval_samples_per_second": 455.554,
+      "eval_steps_per_second": 28.473,
+      "step": 3250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038739848799877844,
+      "loss": 2.2746,
+      "step": 3251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003873347532923661,
+      "loss": 2.3273,
+      "step": 3252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003872710057997058,
+      "loss": 2.3142,
+      "step": 3253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003872072455267324,
+      "loss": 2.3524,
+      "step": 3254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003871434724793823,
+      "loss": 2.3603,
+      "step": 3255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038707968666359293,
+      "loss": 2.3953,
+      "step": 3256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038701588808530294,
+      "loss": 2.3533,
+      "step": 3257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038695207675045205,
+      "loss": 2.3338,
+      "step": 3258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003868882526649814,
+      "loss": 2.3795,
+      "step": 3259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003868244158348331,
+      "loss": 2.3481,
+      "step": 3260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038676056626595057,
+      "loss": 2.539,
+      "step": 3261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038669670396427843,
+      "loss": 2.4535,
+      "step": 3262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038663282893576235,
+      "loss": 2.1728,
+      "step": 3263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038656894118634935,
+      "loss": 2.3259,
+      "step": 3264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003865050407219875,
+      "loss": 2.4878,
+      "step": 3265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038644112754862613,
+      "loss": 2.3268,
+      "step": 3266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003863772016722158,
+      "loss": 2.2719,
+      "step": 3267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003863132630987081,
+      "loss": 2.3027,
+      "step": 3268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003862493118340559,
+      "loss": 2.1912,
+      "step": 3269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038618534788421323,
+      "loss": 2.267,
+      "step": 3270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038612137125513537,
+      "loss": 2.3592,
+      "step": 3271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038605738195277874,
+      "loss": 2.3452,
+      "step": 3272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003859933799831008,
+      "loss": 2.4282,
+      "step": 3273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003859293653520604,
+      "loss": 2.2993,
+      "step": 3274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038586533806561754,
+      "loss": 2.3706,
+      "step": 3275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038580129812973316,
+      "loss": 2.3065,
+      "step": 3276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038573724555036973,
+      "loss": 2.3017,
+      "step": 3277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003856731803334906,
+      "loss": 2.3927,
+      "step": 3278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038560910248506045,
+      "loss": 2.2882,
+      "step": 3279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003855450120110452,
+      "loss": 2.3584,
+      "step": 3280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003854809089174118,
+      "loss": 2.2781,
+      "step": 3281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038541679321012836,
+      "loss": 2.1839,
+      "step": 3282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003853526648951643,
+      "loss": 2.2919,
+      "step": 3283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003852885239784901,
+      "loss": 2.3506,
+      "step": 3284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003852243704660776,
+      "loss": 2.3568,
+      "step": 3285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003851602043638994,
+      "loss": 2.5491,
+      "step": 3286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003850960256779298,
+      "loss": 2.1662,
+      "step": 3287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038503183441414393,
+      "loss": 2.4093,
+      "step": 3288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000384967630578518,
+      "loss": 2.4752,
+      "step": 3289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003849034141770298,
+      "loss": 2.4373,
+      "step": 3290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038483918521565807,
+      "loss": 2.4073,
+      "step": 3291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003847749437003825,
+      "loss": 2.2784,
+      "step": 3292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038471068963718424,
+      "loss": 2.4659,
+      "step": 3293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003846464230320457,
+      "loss": 2.2203,
+      "step": 3294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038458214389095006,
+      "loss": 2.3998,
+      "step": 3295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038451785221988187,
+      "loss": 2.3558,
+      "step": 3296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038445354802482706,
+      "loss": 2.3941,
+      "step": 3297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038438923131177237,
+      "loss": 2.5296,
+      "step": 3298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000384324902086706,
+      "loss": 2.3432,
+      "step": 3299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000384260560355617,
+      "loss": 2.3985,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 2.357994794845581,
+      "eval_runtime": 184.6508,
+      "eval_samples_per_second": 456.018,
+      "eval_steps_per_second": 28.502,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003841962061244959,
+      "loss": 2.2555,
+      "step": 3301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003841318393993342,
+      "loss": 2.3034,
+      "step": 3302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038406746018612465,
+      "loss": 2.2691,
+      "step": 3303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003840030684908611,
+      "loss": 2.4119,
+      "step": 3304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038393866431953873,
+      "loss": 2.3738,
+      "step": 3305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038387424767815346,
+      "loss": 2.2632,
+      "step": 3306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038380981857270293,
+      "loss": 2.2971,
+      "step": 3307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038374537700918554,
+      "loss": 2.212,
+      "step": 3308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003836809229936011,
+      "loss": 2.3486,
+      "step": 3309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038361645653195025,
+      "loss": 2.4899,
+      "step": 3310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003835519776302352,
+      "loss": 2.5646,
+      "step": 3311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003834874862944591,
+      "loss": 2.1199,
+      "step": 3312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038342298253062605,
+      "loss": 2.3184,
+      "step": 3313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038335846634474183,
+      "loss": 2.1617,
+      "step": 3314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038329393774281286,
+      "loss": 2.1782,
+      "step": 3315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000383229396730847,
+      "loss": 2.4204,
+      "step": 3316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038316484331485324,
+      "loss": 2.2276,
+      "step": 3317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038310027750084163,
+      "loss": 2.4436,
+      "step": 3318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003830356992948234,
+      "loss": 2.3949,
+      "step": 3319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003829711087028111,
+      "loss": 2.3605,
+      "step": 3320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038290650573081814,
+      "loss": 2.181,
+      "step": 3321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038284189038485935,
+      "loss": 2.3059,
+      "step": 3322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038277726267095047,
+      "loss": 2.2133,
+      "step": 3323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038271262259510866,
+      "loss": 2.3213,
+      "step": 3324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000382647970163352,
+      "loss": 2.3609,
+      "step": 3325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038258330538169983,
+      "loss": 2.269,
+      "step": 3326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038251862825617266,
+      "loss": 2.34,
+      "step": 3327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000382453938792792,
+      "loss": 2.3851,
+      "step": 3328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038238923699758075,
+      "loss": 2.2543,
+      "step": 3329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003823245228765628,
+      "loss": 2.1824,
+      "step": 3330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003822597964357632,
+      "loss": 2.3203,
+      "step": 3331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003821950576812081,
+      "loss": 2.2766,
+      "step": 3332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000382130306618925,
+      "loss": 2.1577,
+      "step": 3333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038206554325494223,
+      "loss": 2.4157,
+      "step": 3334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003820007675952896,
+      "loss": 2.3333,
+      "step": 3335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003819359796459978,
+      "loss": 2.2814,
+      "step": 3336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003818711794130988,
+      "loss": 2.4908,
+      "step": 3337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038180636690262563,
+      "loss": 2.3662,
+      "step": 3338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003817415421206126,
+      "loss": 2.3237,
+      "step": 3339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000381676705073095,
+      "loss": 2.3979,
+      "step": 3340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038161185576610947,
+      "loss": 2.1963,
+      "step": 3341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038154699420569354,
+      "loss": 2.1341,
+      "step": 3342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000381482120397886,
+      "loss": 2.2551,
+      "step": 3343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003814172343487269,
+      "loss": 2.1771,
+      "step": 3344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038135233606425716,
+      "loss": 2.4403,
+      "step": 3345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003812874255505191,
+      "loss": 2.2903,
+      "step": 3346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038122250281355605,
+      "loss": 2.3415,
+      "step": 3347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038115756785941235,
+      "loss": 2.5129,
+      "step": 3348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038109262069413387,
+      "loss": 2.4571,
+      "step": 3349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038102766132376727,
+      "loss": 2.1567,
+      "step": 3350
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 2.350057601928711,
+      "eval_runtime": 184.9079,
+      "eval_samples_per_second": 455.383,
+      "eval_steps_per_second": 28.463,
+      "step": 3350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038096268975436044,
+      "loss": 2.4032,
+      "step": 3351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003808977059919623,
+      "loss": 2.3656,
+      "step": 3352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003808327100426233,
+      "loss": 2.4048,
+      "step": 3353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003807677019123944,
+      "loss": 2.3333,
+      "step": 3354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003807026816073284,
+      "loss": 2.2934,
+      "step": 3355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038063764913347844,
+      "loss": 2.3091,
+      "step": 3356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038057260449689956,
+      "loss": 2.2464,
+      "step": 3357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003805075477036476,
+      "loss": 2.3471,
+      "step": 3358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003804424787597793,
+      "loss": 2.503,
+      "step": 3359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003803773976713529,
+      "loss": 2.3505,
+      "step": 3360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003803123044444278,
+      "loss": 2.205,
+      "step": 3361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000380247199085064,
+      "loss": 2.3101,
+      "step": 3362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038018208159932323,
+      "loss": 2.3934,
+      "step": 3363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003801169519932681,
+      "loss": 2.4353,
+      "step": 3364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003800518102729622,
+      "loss": 2.196,
+      "step": 3365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037998665644447063,
+      "loss": 2.2732,
+      "step": 3366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003799214905138592,
+      "loss": 2.3933,
+      "step": 3367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003798563124871952,
+      "loss": 2.2566,
+      "step": 3368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003797911223705467,
+      "loss": 2.4352,
+      "step": 3369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003797259201699833,
+      "loss": 2.254,
+      "step": 3370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003796607058915753,
+      "loss": 2.3019,
+      "step": 3371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037959547954139443,
+      "loss": 2.2898,
+      "step": 3372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003795302411255135,
+      "loss": 2.5624,
+      "step": 3373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037946499065000624,
+      "loss": 2.3782,
+      "step": 3374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003793997281209478,
+      "loss": 2.307,
+      "step": 3375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003793344535444142,
+      "loss": 2.4487,
+      "step": 3376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037926916692648273,
+      "loss": 2.3179,
+      "step": 3377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037920386827323185,
+      "loss": 2.2342,
+      "step": 3378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003791385575907408,
+      "loss": 2.2715,
+      "step": 3379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037907323488509044,
+      "loss": 2.147,
+      "step": 3380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003790079001623623,
+      "loss": 2.173,
+      "step": 3381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037894255342863935,
+      "loss": 2.1697,
+      "step": 3382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003788771946900056,
+      "loss": 2.3953,
+      "step": 3383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037881182395254595,
+      "loss": 2.2713,
+      "step": 3384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037874644122234675,
+      "loss": 2.2715,
+      "step": 3385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037868104650549527,
+      "loss": 2.5233,
+      "step": 3386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003786156398080799,
+      "loss": 2.3701,
+      "step": 3387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003785502211361902,
+      "loss": 2.4074,
+      "step": 3388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003784847904959169,
+      "loss": 2.2797,
+      "step": 3389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003784193478933516,
+      "loss": 2.2905,
+      "step": 3390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037835389333458735,
+      "loss": 2.4471,
+      "step": 3391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037828842682571817,
+      "loss": 2.2827,
+      "step": 3392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003782229483728391,
+      "loss": 2.3637,
+      "step": 3393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037815745798204644,
+      "loss": 2.3581,
+      "step": 3394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003780919556594374,
+      "loss": 2.1665,
+      "step": 3395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037802644141111047,
+      "loss": 2.3597,
+      "step": 3396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003779609152431653,
+      "loss": 2.245,
+      "step": 3397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037789537716170253,
+      "loss": 2.2686,
+      "step": 3398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037782982717282377,
+      "loss": 2.3488,
+      "step": 3399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003777642652826322,
+      "loss": 2.4226,
+      "step": 3400
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 2.3415284156799316,
+      "eval_runtime": 184.8698,
+      "eval_samples_per_second": 455.477,
+      "eval_steps_per_second": 28.469,
+      "step": 3400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003776986914972316,
+      "loss": 2.4764,
+      "step": 3401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003776331058227271,
+      "loss": 2.1655,
+      "step": 3402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000377567508265225,
+      "loss": 2.4898,
+      "step": 3403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037750189883083244,
+      "loss": 2.2254,
+      "step": 3404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037743627752565804,
+      "loss": 2.2988,
+      "step": 3405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037737064435581114,
+      "loss": 2.2105,
+      "step": 3406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003773049993274026,
+      "loss": 2.2595,
+      "step": 3407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037723934244654397,
+      "loss": 2.2989,
+      "step": 3408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037717367371934807,
+      "loss": 2.3872,
+      "step": 3409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037710799315192903,
+      "loss": 2.1835,
+      "step": 3410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037704230075040175,
+      "loss": 2.4457,
+      "step": 3411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037697659652088234,
+      "loss": 2.3147,
+      "step": 3412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037691088046948817,
+      "loss": 2.3835,
+      "step": 3413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037684515260233756,
+      "loss": 2.4535,
+      "step": 3414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037677941292554986,
+      "loss": 2.2792,
+      "step": 3415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003767136614452458,
+      "loss": 2.3689,
+      "step": 3416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037664789816754677,
+      "loss": 2.5113,
+      "step": 3417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037658212309857575,
+      "loss": 2.4186,
+      "step": 3418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037651633624445635,
+      "loss": 2.2338,
+      "step": 3419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003764505376113138,
+      "loss": 2.2426,
+      "step": 3420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003763847272052739,
+      "loss": 2.2839,
+      "step": 3421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003763189050324639,
+      "loss": 2.2175,
+      "step": 3422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000376253071099012,
+      "loss": 2.2794,
+      "step": 3423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037618722541104746,
+      "loss": 2.4277,
+      "step": 3424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003761213679747008,
+      "loss": 2.2788,
+      "step": 3425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037605549879610346,
+      "loss": 2.2492,
+      "step": 3426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375989617881388,
+      "loss": 2.236,
+      "step": 3427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037592372523668826,
+      "loss": 2.1982,
+      "step": 3428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375857820868139,
+      "loss": 2.3215,
+      "step": 3429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375791904781876,
+      "loss": 2.4412,
+      "step": 3430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003757259769840362,
+      "loss": 2.3257,
+      "step": 3431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037566003748075786,
+      "loss": 2.1437,
+      "step": 3432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037559408627818,
+      "loss": 2.5684,
+      "step": 3433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003755281233824428,
+      "loss": 2.2775,
+      "step": 3434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037546214879968776,
+      "loss": 2.3729,
+      "step": 3435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037539616253605716,
+      "loss": 2.1458,
+      "step": 3436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003753301645976945,
+      "loss": 2.2614,
+      "step": 3437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003752641549907445,
+      "loss": 2.3788,
+      "step": 3438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037519813372135277,
+      "loss": 2.2795,
+      "step": 3439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037513210079566595,
+      "loss": 2.3277,
+      "step": 3440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003750660562198321,
+      "loss": 2.1695,
+      "step": 3441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375,
+      "loss": 2.4027,
+      "step": 3442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037493393214231975,
+      "loss": 2.3428,
+      "step": 3443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003748678526529424,
+      "loss": 2.2182,
+      "step": 3444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003748017615380202,
+      "loss": 2.4771,
+      "step": 3445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037473565880370637,
+      "loss": 2.2091,
+      "step": 3446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003746695444561552,
+      "loss": 2.3833,
+      "step": 3447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003746034185015223,
+      "loss": 2.3764,
+      "step": 3448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037453728094596396,
+      "loss": 2.3196,
+      "step": 3449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037447113179563796,
+      "loss": 2.2642,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 2.3362231254577637,
+      "eval_runtime": 184.6571,
+      "eval_samples_per_second": 456.002,
+      "eval_steps_per_second": 28.501,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003744049710567029,
+      "loss": 2.3549,
+      "step": 3451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003743387987353184,
+      "loss": 2.2617,
+      "step": 3452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003742726148376455,
+      "loss": 2.3042,
+      "step": 3453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037420641936984595,
+      "loss": 2.1288,
+      "step": 3454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037414021233808285,
+      "loss": 2.3299,
+      "step": 3455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037407399374852015,
+      "loss": 2.3274,
+      "step": 3456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037400776360732303,
+      "loss": 2.2287,
+      "step": 3457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037394152192065766,
+      "loss": 2.1545,
+      "step": 3458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003738752686946915,
+      "loss": 2.2119,
+      "step": 3459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003738090039355927,
+      "loss": 2.4433,
+      "step": 3460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003737427276495308,
+      "loss": 2.1253,
+      "step": 3461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003736764398426763,
+      "loss": 2.4191,
+      "step": 3462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037361014052120066,
+      "loss": 2.2891,
+      "step": 3463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003735438296912768,
+      "loss": 2.2648,
+      "step": 3464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003734775073590781,
+      "loss": 2.2013,
+      "step": 3465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003734111735307796,
+      "loss": 2.309,
+      "step": 3466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003733448282125571,
+      "loss": 2.3018,
+      "step": 3467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003732784714105876,
+      "loss": 2.1354,
+      "step": 3468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037321210313104894,
+      "loss": 2.2204,
+      "step": 3469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003731457233801203,
+      "loss": 2.2894,
+      "step": 3470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003730793321639818,
+      "loss": 2.2266,
+      "step": 3471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003730129294888146,
+      "loss": 2.2983,
+      "step": 3472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037294651536080116,
+      "loss": 2.4163,
+      "step": 3473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003728800897861245,
+      "loss": 2.4692,
+      "step": 3474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037281365277096933,
+      "loss": 2.4095,
+      "step": 3475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000372747204321521,
+      "loss": 2.4776,
+      "step": 3476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037268074444396607,
+      "loss": 2.2171,
+      "step": 3477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037261427314449206,
+      "loss": 2.3438,
+      "step": 3478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037254779042928765,
+      "loss": 2.2466,
+      "step": 3479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003724812963045426,
+      "loss": 2.4261,
+      "step": 3480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003724147907764478,
+      "loss": 2.3532,
+      "step": 3481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037234827385119497,
+      "loss": 2.2058,
+      "step": 3482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003722817455349771,
+      "loss": 2.1142,
+      "step": 3483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000372215205833988,
+      "loss": 2.2925,
+      "step": 3484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003721486547544229,
+      "loss": 2.3235,
+      "step": 3485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003720820923024778,
+      "loss": 2.4696,
+      "step": 3486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003720155184843498,
+      "loss": 2.3233,
+      "step": 3487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037194893330623724,
+      "loss": 2.2492,
+      "step": 3488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037188233677433924,
+      "loss": 2.3366,
+      "step": 3489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037181572889485626,
+      "loss": 2.1522,
+      "step": 3490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037174910967398956,
+      "loss": 2.4602,
+      "step": 3491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037168247911794174,
+      "loss": 2.3801,
+      "step": 3492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000371615837232916,
+      "loss": 2.2125,
+      "step": 3493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003715491840251172,
+      "loss": 2.3298,
+      "step": 3494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037148251950075076,
+      "loss": 2.2017,
+      "step": 3495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003714158436660233,
+      "loss": 2.2887,
+      "step": 3496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003713491565271427,
+      "loss": 2.2902,
+      "step": 3497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037128245809031765,
+      "loss": 2.4396,
+      "step": 3498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003712157483617578,
+      "loss": 2.3712,
+      "step": 3499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003711490273476743,
+      "loss": 2.2697,
+      "step": 3500
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 2.327763080596924,
+      "eval_runtime": 184.8184,
+      "eval_samples_per_second": 455.604,
+      "eval_steps_per_second": 28.477,
+      "step": 3500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003710822950542788,
+      "loss": 2.2403,
+      "step": 3501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037101555148778435,
+      "loss": 2.3479,
+      "step": 3502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037094879665440506,
+      "loss": 2.272,
+      "step": 3503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037088203056035585,
+      "loss": 2.481,
+      "step": 3504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003708152532118529,
+      "loss": 2.2587,
+      "step": 3505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037074846461511337,
+      "loss": 2.1579,
+      "step": 3506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003706816647763555,
+      "loss": 2.3197,
+      "step": 3507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003706148537017984,
+      "loss": 2.2365,
+      "step": 3508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037054803139766247,
+      "loss": 2.547,
+      "step": 3509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000370481197870169,
+      "loss": 2.4194,
+      "step": 3510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003704143531255405,
+      "loss": 2.4559,
+      "step": 3511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003703474971700003,
+      "loss": 2.1226,
+      "step": 3512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003702806300097729,
+      "loss": 2.0575,
+      "step": 3513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003702137516510838,
+      "loss": 2.4626,
+      "step": 3514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037014686210015956,
+      "loss": 2.4942,
+      "step": 3515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037007996136322784,
+      "loss": 2.0613,
+      "step": 3516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003700130494465173,
+      "loss": 2.1327,
+      "step": 3517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036994612635625745,
+      "loss": 2.2559,
+      "step": 3518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003698791920986792,
+      "loss": 2.4494,
+      "step": 3519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003698122466800142,
+      "loss": 2.1807,
+      "step": 3520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003697452901064954,
+      "loss": 2.2649,
+      "step": 3521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003696783223843564,
+      "loss": 2.0471,
+      "step": 3522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003696113435198324,
+      "loss": 2.5064,
+      "step": 3523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003695443535191591,
+      "loss": 2.3878,
+      "step": 3524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036947735238857345,
+      "loss": 2.3127,
+      "step": 3525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003694103401343136,
+      "loss": 2.3581,
+      "step": 3526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036934331676261836,
+      "loss": 2.1475,
+      "step": 3527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000369276282279728,
+      "loss": 2.1821,
+      "step": 3528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036920923669188344,
+      "loss": 2.3163,
+      "step": 3529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000369142180005327,
+      "loss": 2.4443,
+      "step": 3530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003690751122263016,
+      "loss": 2.307,
+      "step": 3531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003690080333610517,
+      "loss": 2.5378,
+      "step": 3532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003689409434158224,
+      "loss": 2.1543,
+      "step": 3533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003688738423968599,
+      "loss": 2.2786,
+      "step": 3534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036880673031041165,
+      "loss": 2.3222,
+      "step": 3535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036873960716272584,
+      "loss": 2.3345,
+      "step": 3536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003686724729600519,
+      "loss": 2.1199,
+      "step": 3537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003686053277086401,
+      "loss": 2.4478,
+      "step": 3538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036853817141474195,
+      "loss": 2.4193,
+      "step": 3539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036847100408460986,
+      "loss": 2.3553,
+      "step": 3540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003684038257244973,
+      "loss": 2.4453,
+      "step": 3541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003683366363406588,
+      "loss": 2.314,
+      "step": 3542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003682694359393498,
+      "loss": 2.219,
+      "step": 3543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036820222452682696,
+      "loss": 2.4535,
+      "step": 3544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036813500210934765,
+      "loss": 2.2573,
+      "step": 3545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003680677686931707,
+      "loss": 2.0116,
+      "step": 3546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036800052428455555,
+      "loss": 2.2028,
+      "step": 3547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036793326888976297,
+      "loss": 2.1938,
+      "step": 3548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036786600251505447,
+      "loss": 2.1312,
+      "step": 3549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036779872516669293,
+      "loss": 2.3797,
+      "step": 3550
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 2.323045015335083,
+      "eval_runtime": 184.6799,
+      "eval_samples_per_second": 455.946,
+      "eval_steps_per_second": 28.498,
+      "step": 3550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036773143685094197,
+      "loss": 2.2505,
+      "step": 3551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003676641375740662,
+      "loss": 2.3517,
+      "step": 3552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003675968273423315,
+      "loss": 2.2035,
+      "step": 3553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003675295061620047,
+      "loss": 2.2466,
+      "step": 3554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036746217403935344,
+      "loss": 2.4445,
+      "step": 3555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003673948309806466,
+      "loss": 2.2948,
+      "step": 3556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036732747699215403,
+      "loss": 2.1384,
+      "step": 3557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036726011208014655,
+      "loss": 2.4,
+      "step": 3558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036719273625089593,
+      "loss": 2.0839,
+      "step": 3559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003671253495106752,
+      "loss": 2.2769,
+      "step": 3560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003670579518657582,
+      "loss": 2.2764,
+      "step": 3561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036699054332241986,
+      "loss": 2.21,
+      "step": 3562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036692312388693604,
+      "loss": 2.262,
+      "step": 3563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003668556935655837,
+      "loss": 2.2271,
+      "step": 3564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036678825236464074,
+      "loss": 2.3554,
+      "step": 3565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003667208002903863,
+      "loss": 2.5187,
+      "step": 3566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003666533373491002,
+      "loss": 2.2438,
+      "step": 3567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003665858635470634,
+      "loss": 2.2363,
+      "step": 3568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000366518378890558,
+      "loss": 2.4009,
+      "step": 3569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036645088338586694,
+      "loss": 2.4464,
+      "step": 3570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003663833770392744,
+      "loss": 2.2781,
+      "step": 3571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003663158598570652,
+      "loss": 2.2202,
+      "step": 3572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036624833184552535,
+      "loss": 2.2389,
+      "step": 3573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036618079301094214,
+      "loss": 2.222,
+      "step": 3574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036611324335960346,
+      "loss": 2.1678,
+      "step": 3575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036604568289779843,
+      "loss": 2.1831,
+      "step": 3576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036597811163181706,
+      "loss": 2.2113,
+      "step": 3577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003659105295679504,
+      "loss": 2.2781,
+      "step": 3578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003658429367124906,
+      "loss": 2.2414,
+      "step": 3579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003657753330717308,
+      "loss": 2.1608,
+      "step": 3580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036570771865196487,
+      "loss": 2.2848,
+      "step": 3581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036564009345948817,
+      "loss": 2.1446,
+      "step": 3582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036557245750059664,
+      "loss": 2.4248,
+      "step": 3583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003655048107815874,
+      "loss": 2.4485,
+      "step": 3584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003654371533087585,
+      "loss": 2.3389,
+      "step": 3585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003653694850884091,
+      "loss": 2.1401,
+      "step": 3586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003653018061268393,
+      "loss": 2.2781,
+      "step": 3587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003652341164303502,
+      "loss": 2.2311,
+      "step": 3588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036516641600524386,
+      "loss": 2.1366,
+      "step": 3589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036509870485782343,
+      "loss": 2.1398,
+      "step": 3590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000365030982994393,
+      "loss": 2.1695,
+      "step": 3591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036496325042125756,
+      "loss": 2.0758,
+      "step": 3592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036489550714472336,
+      "loss": 2.4955,
+      "step": 3593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036482775317109734,
+      "loss": 2.3668,
+      "step": 3594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003647599885066877,
+      "loss": 2.4973,
+      "step": 3595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036469221315780354,
+      "loss": 2.3802,
+      "step": 3596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003646244271307548,
+      "loss": 2.2657,
+      "step": 3597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003645566304318526,
+      "loss": 2.2982,
+      "step": 3598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036448882306740904,
+      "loss": 2.1626,
+      "step": 3599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003644210050437372,
+      "loss": 2.2473,
+      "step": 3600
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 2.313283920288086,
+      "eval_runtime": 184.0287,
+      "eval_samples_per_second": 457.559,
+      "eval_steps_per_second": 28.599,
+      "step": 3600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036435317636715107,
+      "loss": 2.3523,
+      "step": 3601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036428533704396565,
+      "loss": 2.2034,
+      "step": 3602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000364217487080497,
+      "loss": 2.2967,
+      "step": 3603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003641496264830623,
+      "loss": 2.3476,
+      "step": 3604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036408175525797924,
+      "loss": 2.1645,
+      "step": 3605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003640138734115671,
+      "loss": 2.4748,
+      "step": 3606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003639459809501457,
+      "loss": 2.2205,
+      "step": 3607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003638780778800362,
+      "loss": 2.3776,
+      "step": 3608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036381016420756037,
+      "loss": 2.2945,
+      "step": 3609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036374223993904125,
+      "loss": 2.3728,
+      "step": 3610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036367430508080277,
+      "loss": 2.2034,
+      "step": 3611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003636063596391699,
+      "loss": 2.2144,
+      "step": 3612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036353840362046854,
+      "loss": 2.3268,
+      "step": 3613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003634704370310254,
+      "loss": 2.3123,
+      "step": 3614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003634024598771687,
+      "loss": 2.1881,
+      "step": 3615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036333447216522704,
+      "loss": 2.2851,
+      "step": 3616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036326647390153027,
+      "loss": 2.3898,
+      "step": 3617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036319846509240937,
+      "loss": 2.3521,
+      "step": 3618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036313044574419603,
+      "loss": 2.2927,
+      "step": 3619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000363062415863223,
+      "loss": 2.2749,
+      "step": 3620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036299437545582426,
+      "loss": 2.1112,
+      "step": 3621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036292632452833434,
+      "loss": 2.4888,
+      "step": 3622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003628582630870891,
+      "loss": 2.3058,
+      "step": 3623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003627901911384252,
+      "loss": 2.3077,
+      "step": 3624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036272210868868034,
+      "loss": 2.2602,
+      "step": 3625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003626540157441932,
+      "loss": 2.0509,
+      "step": 3626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036258591231130334,
+      "loss": 2.1928,
+      "step": 3627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003625177983963515,
+      "loss": 2.1848,
+      "step": 3628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003624496740056792,
+      "loss": 2.174,
+      "step": 3629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036238153914562913,
+      "loss": 2.3775,
+      "step": 3630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036231339382254463,
+      "loss": 2.4132,
+      "step": 3631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003622452380427703,
+      "loss": 2.2166,
+      "step": 3632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036217707181265173,
+      "loss": 2.2565,
+      "step": 3633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036210889513853527,
+      "loss": 2.4745,
+      "step": 3634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036204070802676845,
+      "loss": 2.4408,
+      "step": 3635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003619725104836996,
+      "loss": 2.2242,
+      "step": 3636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003619043025156781,
+      "loss": 2.1978,
+      "step": 3637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036183608412905445,
+      "loss": 2.1903,
+      "step": 3638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036176785533017976,
+      "loss": 2.275,
+      "step": 3639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003616996161254065,
+      "loss": 2.3346,
+      "step": 3640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036163136652108774,
+      "loss": 2.0284,
+      "step": 3641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003615631065235779,
+      "loss": 2.1796,
+      "step": 3642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003614948361392321,
+      "loss": 2.4565,
+      "step": 3643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003614265553744064,
+      "loss": 2.424,
+      "step": 3644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003613582642354581,
+      "loss": 2.1915,
+      "step": 3645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003612899627287452,
+      "loss": 2.2505,
+      "step": 3646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003612216508606268,
+      "loss": 2.3341,
+      "step": 3647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003611533286374629,
+      "loss": 2.4456,
+      "step": 3648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003610849960656145,
+      "loss": 2.5346,
+      "step": 3649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036101665315144355,
+      "loss": 2.3887,
+      "step": 3650
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 2.3055033683776855,
+      "eval_runtime": 183.9762,
+      "eval_samples_per_second": 457.69,
+      "eval_steps_per_second": 28.607,
+      "step": 3650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036094829990131297,
+      "loss": 2.1103,
+      "step": 3651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036087993632158666,
+      "loss": 2.4347,
+      "step": 3652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036081156241862935,
+      "loss": 2.1864,
+      "step": 3653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003607431781988069,
+      "loss": 2.3147,
+      "step": 3654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036067478366848623,
+      "loss": 2.239,
+      "step": 3655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003606063788340348,
+      "loss": 2.2698,
+      "step": 3656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003605379637018215,
+      "loss": 2.1822,
+      "step": 3657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036046953827821587,
+      "loss": 2.2939,
+      "step": 3658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00036040110256958845,
+      "loss": 2.2292,
+      "step": 3659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00036033265658231103,
+      "loss": 2.3838,
+      "step": 3660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003602642003227558,
+      "loss": 2.239,
+      "step": 3661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003601957337972964,
+      "loss": 2.4007,
+      "step": 3662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00036012725701230734,
+      "loss": 2.3636,
+      "step": 3663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00036005876997416387,
+      "loss": 2.1764,
+      "step": 3664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035999027268924236,
+      "loss": 2.1932,
+      "step": 3665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035992176516392004,
+      "loss": 2.3196,
+      "step": 3666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035985324740457527,
+      "loss": 2.2002,
+      "step": 3667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003597847194175871,
+      "loss": 2.2712,
+      "step": 3668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035971618120933584,
+      "loss": 2.2627,
+      "step": 3669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035964763278620245,
+      "loss": 2.0689,
+      "step": 3670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003595790741545691,
+      "loss": 2.1069,
+      "step": 3671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003595105053208186,
+      "loss": 2.5731,
+      "step": 3672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003594419262913351,
+      "loss": 2.2468,
+      "step": 3673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035937333707250344,
+      "loss": 2.3453,
+      "step": 3674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003593047376707094,
+      "loss": 2.3169,
+      "step": 3675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003592361280923399,
+      "loss": 2.458,
+      "step": 3676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003591675083437825,
+      "loss": 2.16,
+      "step": 3677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003590988784314261,
+      "loss": 2.5389,
+      "step": 3678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035903023836166015,
+      "loss": 2.4181,
+      "step": 3679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035896158814087545,
+      "loss": 2.3514,
+      "step": 3680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035889292777546326,
+      "loss": 2.382,
+      "step": 3681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003588242572718162,
+      "loss": 2.142,
+      "step": 3682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035875557663632776,
+      "loss": 2.1962,
+      "step": 3683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035868688587539213,
+      "loss": 2.358,
+      "step": 3684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003586181849954047,
+      "loss": 2.3785,
+      "step": 3685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035854947400276163,
+      "loss": 2.3611,
+      "step": 3686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035848075290386024,
+      "loss": 2.2733,
+      "step": 3687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035841202170509856,
+      "loss": 2.2673,
+      "step": 3688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003583432804128757,
+      "loss": 2.1164,
+      "step": 3689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003582745290335917,
+      "loss": 2.3786,
+      "step": 3690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035820576757364744,
+      "loss": 2.235,
+      "step": 3691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035813699603944477,
+      "loss": 2.2917,
+      "step": 3692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003580682144373866,
+      "loss": 2.4429,
+      "step": 3693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035799942277387663,
+      "loss": 2.1299,
+      "step": 3694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003579306210553196,
+      "loss": 2.1896,
+      "step": 3695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035786180928812106,
+      "loss": 2.1602,
+      "step": 3696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035779298747868775,
+      "loss": 2.2555,
+      "step": 3697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000357724155633427,
+      "loss": 2.1422,
+      "step": 3698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003576553137587473,
+      "loss": 2.1017,
+      "step": 3699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003575864618610581,
+      "loss": 2.2165,
+      "step": 3700
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 2.3002214431762695,
+      "eval_runtime": 184.0502,
+      "eval_samples_per_second": 457.506,
+      "eval_steps_per_second": 28.595,
+      "step": 3700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003575175999467695,
+      "loss": 2.2461,
+      "step": 3701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035744872802229296,
+      "loss": 2.2862,
+      "step": 3702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003573798460940405,
+      "loss": 2.4148,
+      "step": 3703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003573109541684252,
+      "loss": 2.2075,
+      "step": 3704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035724205225186133,
+      "loss": 2.1675,
+      "step": 3705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003571731403507635,
+      "loss": 2.4385,
+      "step": 3706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035710421847154794,
+      "loss": 2.2148,
+      "step": 3707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003570352866206312,
+      "loss": 2.17,
+      "step": 3708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003569663448044312,
+      "loss": 2.4184,
+      "step": 3709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035689739302936655,
+      "loss": 2.2613,
+      "step": 3710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035682843130185687,
+      "loss": 2.2871,
+      "step": 3711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003567594596283226,
+      "loss": 2.3996,
+      "step": 3712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003566904780151854,
+      "loss": 2.1961,
+      "step": 3713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035662148646886736,
+      "loss": 2.4005,
+      "step": 3714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035655248499579206,
+      "loss": 2.3449,
+      "step": 3715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035648347360238365,
+      "loss": 2.3294,
+      "step": 3716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003564144522950671,
+      "loss": 2.2633,
+      "step": 3717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035634542108026873,
+      "loss": 2.2682,
+      "step": 3718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035627637996441544,
+      "loss": 2.435,
+      "step": 3719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035620732895393515,
+      "loss": 2.1783,
+      "step": 3720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035613826805525663,
+      "loss": 2.4417,
+      "step": 3721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003560691972748098,
+      "loss": 2.3312,
+      "step": 3722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003560001166190252,
+      "loss": 2.2482,
+      "step": 3723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003559310260943345,
+      "loss": 2.4711,
+      "step": 3724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003558619257071702,
+      "loss": 2.3178,
+      "step": 3725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003557928154639658,
+      "loss": 2.0129,
+      "step": 3726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035572369537115563,
+      "loss": 2.4144,
+      "step": 3727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035565456543517487,
+      "loss": 2.2628,
+      "step": 3728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035558542566245977,
+      "loss": 2.2374,
+      "step": 3729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003555162760594475,
+      "loss": 2.3393,
+      "step": 3730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000355447116632576,
+      "loss": 2.2143,
+      "step": 3731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035537794738828423,
+      "loss": 2.3108,
+      "step": 3732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000355308768333012,
+      "loss": 2.2994,
+      "step": 3733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035523957947320014,
+      "loss": 2.1944,
+      "step": 3734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035517038081529025,
+      "loss": 2.2497,
+      "step": 3735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000355101172365725,
+      "loss": 2.2237,
+      "step": 3736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035503195413094787,
+      "loss": 2.2322,
+      "step": 3737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003549627261174032,
+      "loss": 2.3261,
+      "step": 3738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035489348833153645,
+      "loss": 2.1826,
+      "step": 3739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003548242407797937,
+      "loss": 2.1133,
+      "step": 3740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035475498346862217,
+      "loss": 2.4461,
+      "step": 3741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003546857164044699,
+      "loss": 2.2937,
+      "step": 3742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003546164395937858,
+      "loss": 2.3539,
+      "step": 3743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003545471530430198,
+      "loss": 2.2787,
+      "step": 3744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035447785675862267,
+      "loss": 2.3914,
+      "step": 3745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035440855074704596,
+      "loss": 2.3592,
+      "step": 3746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035433923501474254,
+      "loss": 2.3313,
+      "step": 3747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003542699095681656,
+      "loss": 2.1039,
+      "step": 3748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035420057441376963,
+      "loss": 2.3382,
+      "step": 3749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035413122955801,
+      "loss": 2.2813,
+      "step": 3750
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 2.2885825634002686,
+      "eval_runtime": 184.0456,
+      "eval_samples_per_second": 457.517,
+      "eval_steps_per_second": 28.596,
+      "step": 3750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035406187500734276,
+      "loss": 2.1668,
+      "step": 3751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035399251076822515,
+      "loss": 2.5196,
+      "step": 3752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003539231368471152,
+      "loss": 2.4273,
+      "step": 3753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035385375325047166,
+      "loss": 2.3394,
+      "step": 3754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035378435998475446,
+      "loss": 2.3798,
+      "step": 3755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035371495705642416,
+      "loss": 2.0771,
+      "step": 3756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003536455444719425,
+      "loss": 2.2497,
+      "step": 3757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000353576122237772,
+      "loss": 2.4226,
+      "step": 3758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035350669036037603,
+      "loss": 2.3627,
+      "step": 3759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035343724884621887,
+      "loss": 2.2201,
+      "step": 3760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035336779770176575,
+      "loss": 1.9862,
+      "step": 3761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003532983369334827,
+      "loss": 2.2363,
+      "step": 3762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035322886654783677,
+      "loss": 2.2071,
+      "step": 3763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035315938655129577,
+      "loss": 2.0027,
+      "step": 3764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003530898969503287,
+      "loss": 2.2498,
+      "step": 3765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003530203977514049,
+      "loss": 2.267,
+      "step": 3766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003529508889609952,
+      "loss": 2.2842,
+      "step": 3767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035288137058557095,
+      "loss": 2.017,
+      "step": 3768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003528118426316046,
+      "loss": 2.2836,
+      "step": 3769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003527423051055692,
+      "loss": 2.0507,
+      "step": 3770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035267275801393905,
+      "loss": 2.2649,
+      "step": 3771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035260320136318925,
+      "loss": 2.478,
+      "step": 3772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003525336351597955,
+      "loss": 2.444,
+      "step": 3773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035246405941023474,
+      "loss": 2.1013,
+      "step": 3774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003523944741209847,
+      "loss": 2.2708,
+      "step": 3775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035232487929852385,
+      "loss": 2.2328,
+      "step": 3776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035225527494933174,
+      "loss": 2.1199,
+      "step": 3777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003521856610798887,
+      "loss": 2.3908,
+      "step": 3778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000352116037696676,
+      "loss": 2.263,
+      "step": 3779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003520464048061758,
+      "loss": 2.2574,
+      "step": 3780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000351976762414871,
+      "loss": 2.2415,
+      "step": 3781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003519071105292456,
+      "loss": 2.271,
+      "step": 3782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035183744915578437,
+      "loss": 2.1623,
+      "step": 3783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003517677783009729,
+      "loss": 2.1768,
+      "step": 3784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003516980979712978,
+      "loss": 2.2248,
+      "step": 3785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003516284081732466,
+      "loss": 2.331,
+      "step": 3786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003515587089133075,
+      "loss": 2.3931,
+      "step": 3787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035148900019796965,
+      "loss": 2.4391,
+      "step": 3788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035141928203372326,
+      "loss": 2.2536,
+      "step": 3789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003513495544270592,
+      "loss": 2.2125,
+      "step": 3790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035127981738446927,
+      "loss": 2.2987,
+      "step": 3791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035121007091244635,
+      "loss": 2.3059,
+      "step": 3792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003511403150174838,
+      "loss": 2.0539,
+      "step": 3793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003510705497060762,
+      "loss": 2.1772,
+      "step": 3794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000351000774984719,
+      "loss": 2.3299,
+      "step": 3795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003509309908599083,
+      "loss": 2.2102,
+      "step": 3796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003508611973381412,
+      "loss": 2.2538,
+      "step": 3797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003507913944259157,
+      "loss": 2.1563,
+      "step": 3798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003507215821297306,
+      "loss": 2.3425,
+      "step": 3799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003506517604560857,
+      "loss": 2.2621,
+      "step": 3800
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 2.287198781967163,
+      "eval_runtime": 184.0572,
+      "eval_samples_per_second": 457.488,
+      "eval_steps_per_second": 28.594,
+      "step": 3800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035058192941148146,
+      "loss": 2.2129,
+      "step": 3801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003505120890024195,
+      "loss": 2.2826,
+      "step": 3802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035044223923540207,
+      "loss": 2.1542,
+      "step": 3803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035037238011693234,
+      "loss": 2.1146,
+      "step": 3804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035030251165351446,
+      "loss": 2.4038,
+      "step": 3805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003502326338516534,
+      "loss": 2.2127,
+      "step": 3806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000350162746717855,
+      "loss": 2.3026,
+      "step": 3807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035009285025862575,
+      "loss": 2.2944,
+      "step": 3808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035002294448047345,
+      "loss": 2.1086,
+      "step": 3809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003499530293899064,
+      "loss": 2.3148,
+      "step": 3810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034988310499343387,
+      "loss": 2.23,
+      "step": 3811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000349813171297566,
+      "loss": 2.259,
+      "step": 3812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034974322830881393,
+      "loss": 2.1336,
+      "step": 3813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003496732760336895,
+      "loss": 2.4095,
+      "step": 3814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003496033144787054,
+      "loss": 2.2197,
+      "step": 3815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034953334365037526,
+      "loss": 2.2825,
+      "step": 3816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003494633635552137,
+      "loss": 2.1086,
+      "step": 3817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034939337419973585,
+      "loss": 2.2555,
+      "step": 3818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034932337559045795,
+      "loss": 2.2562,
+      "step": 3819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034925336773389726,
+      "loss": 2.3805,
+      "step": 3820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034918335063657146,
+      "loss": 2.4567,
+      "step": 3821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003491133243049995,
+      "loss": 2.3191,
+      "step": 3822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000349043288745701,
+      "loss": 2.1287,
+      "step": 3823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003489732439651964,
+      "loss": 2.4296,
+      "step": 3824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034890318997000714,
+      "loss": 2.2723,
+      "step": 3825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034883312676665533,
+      "loss": 2.3289,
+      "step": 3826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034876305436166424,
+      "loss": 2.4577,
+      "step": 3827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034869297276155767,
+      "loss": 2.2175,
+      "step": 3828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034862288197286034,
+      "loss": 2.294,
+      "step": 3829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034855278200209813,
+      "loss": 2.3688,
+      "step": 3830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034848267285579734,
+      "loss": 2.284,
+      "step": 3831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034841255454048535,
+      "loss": 2.2647,
+      "step": 3832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003483424270626904,
+      "loss": 2.3136,
+      "step": 3833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003482722904289417,
+      "loss": 2.2064,
+      "step": 3834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003482021446457689,
+      "loss": 2.3545,
+      "step": 3835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034813198971970295,
+      "loss": 2.1832,
+      "step": 3836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034806182565727545,
+      "loss": 2.294,
+      "step": 3837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003479916524650188,
+      "loss": 2.2074,
+      "step": 3838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003479214701494664,
+      "loss": 2.1602,
+      "step": 3839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034785127871715226,
+      "loss": 2.2246,
+      "step": 3840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034778107817461156,
+      "loss": 2.3677,
+      "step": 3841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034771086852838023,
+      "loss": 2.2557,
+      "step": 3842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034764064978499477,
+      "loss": 2.1317,
+      "step": 3843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034757042195099287,
+      "loss": 2.1218,
+      "step": 3844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034750018503291297,
+      "loss": 2.2377,
+      "step": 3845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003474299390372942,
+      "loss": 2.521,
+      "step": 3846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034735968397067675,
+      "loss": 2.0154,
+      "step": 3847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034728941983960154,
+      "loss": 2.236,
+      "step": 3848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034721914665061034,
+      "loss": 2.0188,
+      "step": 3849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003471488644102457,
+      "loss": 2.2801,
+      "step": 3850
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 2.2794532775878906,
+      "eval_runtime": 183.9541,
+      "eval_samples_per_second": 457.745,
+      "eval_steps_per_second": 28.61,
+      "step": 3850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003470785731250514,
+      "loss": 2.1769,
+      "step": 3851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003470082728015713,
+      "loss": 2.2987,
+      "step": 3852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000346937963446351,
+      "loss": 2.3162,
+      "step": 3853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034686764506593616,
+      "loss": 2.142,
+      "step": 3854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003467973176668738,
+      "loss": 2.2581,
+      "step": 3855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003467269812557115,
+      "loss": 2.3769,
+      "step": 3856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034665663583899786,
+      "loss": 2.1313,
+      "step": 3857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034658628142328216,
+      "loss": 2.1706,
+      "step": 3858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034651591801511465,
+      "loss": 2.1863,
+      "step": 3859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034644554562104636,
+      "loss": 2.1604,
+      "step": 3860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034637516424762906,
+      "loss": 2.1922,
+      "step": 3861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003463047739014156,
+      "loss": 2.1897,
+      "step": 3862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034623437458895947,
+      "loss": 2.3726,
+      "step": 3863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003461639663168149,
+      "loss": 2.2854,
+      "step": 3864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034609354909153733,
+      "loss": 2.1834,
+      "step": 3865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034602312291968256,
+      "loss": 2.3722,
+      "step": 3866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003459526878078077,
+      "loss": 2.1913,
+      "step": 3867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003458822437624702,
+      "loss": 2.3005,
+      "step": 3868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034581179079022883,
+      "loss": 2.0441,
+      "step": 3869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003457413288976429,
+      "loss": 2.3035,
+      "step": 3870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003456708580912725,
+      "loss": 2.3059,
+      "step": 3871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034560037837767863,
+      "loss": 2.3663,
+      "step": 3872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034552988976342344,
+      "loss": 2.3267,
+      "step": 3873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034545939225506936,
+      "loss": 2.3105,
+      "step": 3874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034538888585917987,
+      "loss": 2.3082,
+      "step": 3875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003453183705823195,
+      "loss": 2.4618,
+      "step": 3876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034524784643105336,
+      "loss": 2.1376,
+      "step": 3877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034517731341194733,
+      "loss": 2.352,
+      "step": 3878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003451067715315683,
+      "loss": 2.2889,
+      "step": 3879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003450362207964841,
+      "loss": 2.283,
+      "step": 3880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003449656612132629,
+      "loss": 2.06,
+      "step": 3881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003448950927884741,
+      "loss": 2.2671,
+      "step": 3882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034482451552868786,
+      "loss": 2.1711,
+      "step": 3883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003447539294404751,
+      "loss": 2.292,
+      "step": 3884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034468333453040764,
+      "loss": 2.2901,
+      "step": 3885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003446127308050579,
+      "loss": 2.0563,
+      "step": 3886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003445421182709995,
+      "loss": 2.1913,
+      "step": 3887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034447149693480646,
+      "loss": 2.1804,
+      "step": 3888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034440086680305394,
+      "loss": 2.3,
+      "step": 3889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003443302278823178,
+      "loss": 2.3875,
+      "step": 3890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034425958017917464,
+      "loss": 2.2795,
+      "step": 3891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034418892370020204,
+      "loss": 2.1509,
+      "step": 3892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003441182584519783,
+      "loss": 2.1323,
+      "step": 3893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034404758444108255,
+      "loss": 2.2545,
+      "step": 3894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003439769016740947,
+      "loss": 2.452,
+      "step": 3895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003439062101575956,
+      "loss": 2.2707,
+      "step": 3896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003438355098981667,
+      "loss": 2.2211,
+      "step": 3897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003437648009023905,
+      "loss": 2.1889,
+      "step": 3898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034369408317685014,
+      "loss": 2.3118,
+      "step": 3899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003436233567281296,
+      "loss": 2.2543,
+      "step": 3900
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 2.2729978561401367,
+      "eval_runtime": 184.0034,
+      "eval_samples_per_second": 457.622,
+      "eval_steps_per_second": 28.603,
+      "step": 3900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034355262156281387,
+      "loss": 2.324,
+      "step": 3901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034348187768748847,
+      "loss": 2.4022,
+      "step": 3902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003434111251087398,
+      "loss": 1.9955,
+      "step": 3903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034334036383315526,
+      "loss": 2.2444,
+      "step": 3904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003432695938673228,
+      "loss": 2.2064,
+      "step": 3905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034319881521783146,
+      "loss": 2.0589,
+      "step": 3906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003431280278912707,
+      "loss": 2.2883,
+      "step": 3907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003430572318942312,
+      "loss": 2.1809,
+      "step": 3908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003429864272333043,
+      "loss": 2.0571,
+      "step": 3909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034291561391508186,
+      "loss": 2.2233,
+      "step": 3910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000342844791946157,
+      "loss": 2.2419,
+      "step": 3911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003427739613331235,
+      "loss": 2.1768,
+      "step": 3912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003427031220825757,
+      "loss": 2.1171,
+      "step": 3913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000342632274201109,
+      "loss": 2.4089,
+      "step": 3914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003425614176953197,
+      "loss": 2.2477,
+      "step": 3915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003424905525718044,
+      "loss": 2.2986,
+      "step": 3916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003424196788371611,
+      "loss": 2.1562,
+      "step": 3917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034234879649798836,
+      "loss": 2.0165,
+      "step": 3918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003422779055608853,
+      "loss": 2.3118,
+      "step": 3919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034220700603245223,
+      "loss": 2.2072,
+      "step": 3920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003421360979192901,
+      "loss": 2.2001,
+      "step": 3921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034206518122800055,
+      "loss": 2.2086,
+      "step": 3922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003419942559651863,
+      "loss": 2.1656,
+      "step": 3923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034192332213745056,
+      "loss": 2.238,
+      "step": 3924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003418523797513974,
+      "loss": 2.3345,
+      "step": 3925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003417814288136319,
+      "loss": 2.1152,
+      "step": 3926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003417104693307597,
+      "loss": 2.2331,
+      "step": 3927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034163950130938737,
+      "loss": 2.0327,
+      "step": 3928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034156852475612223,
+      "loss": 2.3192,
+      "step": 3929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034149753967757234,
+      "loss": 2.1558,
+      "step": 3930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034142654608034667,
+      "loss": 2.0728,
+      "step": 3931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034135554397105494,
+      "loss": 2.2636,
+      "step": 3932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034128453335630755,
+      "loss": 2.2225,
+      "step": 3933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003412135142427159,
+      "loss": 2.1696,
+      "step": 3934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003411424866368919,
+      "loss": 2.3665,
+      "step": 3935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003410714505454486,
+      "loss": 2.2533,
+      "step": 3936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003410004059749996,
+      "loss": 2.0647,
+      "step": 3937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034092935293215926,
+      "loss": 2.2888,
+      "step": 3938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003408582914235429,
+      "loss": 2.2126,
+      "step": 3939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003407872214557666,
+      "loss": 2.249,
+      "step": 3940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000340716143035447,
+      "loss": 2.35,
+      "step": 3941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003406450561692018,
+      "loss": 2.233,
+      "step": 3942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034057396086364954,
+      "loss": 2.4436,
+      "step": 3943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003405028571254091,
+      "loss": 2.188,
+      "step": 3944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034043174496110065,
+      "loss": 2.5214,
+      "step": 3945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003403606243773448,
+      "loss": 1.9915,
+      "step": 3946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003402894953807631,
+      "loss": 2.4074,
+      "step": 3947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003402183579779781,
+      "loss": 2.2693,
+      "step": 3948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034014721217561243,
+      "loss": 2.0846,
+      "step": 3949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003400760579802903,
+      "loss": 2.3779,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 2.2644312381744385,
+      "eval_runtime": 184.008,
+      "eval_samples_per_second": 457.61,
+      "eval_steps_per_second": 28.602,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003400048953986363,
+      "loss": 2.2048,
+      "step": 3951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003399337244372758,
+      "loss": 2.1561,
+      "step": 3952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033986254510283496,
+      "loss": 2.1974,
+      "step": 3953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000339791357401941,
+      "loss": 2.2903,
+      "step": 3954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033972016134122153,
+      "loss": 2.1225,
+      "step": 3955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003396489569273051,
+      "loss": 2.318,
+      "step": 3956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033957774416682113,
+      "loss": 2.0669,
+      "step": 3957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003395065230663996,
+      "loss": 2.4335,
+      "step": 3958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003394352936326715,
+      "loss": 2.0956,
+      "step": 3959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033936405587226844,
+      "loss": 2.1187,
+      "step": 3960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003392928097918229,
+      "loss": 2.1487,
+      "step": 3961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033922155539796797,
+      "loss": 2.369,
+      "step": 3962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033915029269733775,
+      "loss": 2.221,
+      "step": 3963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000339079021696567,
+      "loss": 2.2599,
+      "step": 3964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00033900774240229123,
+      "loss": 2.2137,
+      "step": 3965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003389364548211466,
+      "loss": 2.0942,
+      "step": 3966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003388651589597703,
+      "loss": 2.2863,
+      "step": 3967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003387938548248003,
+      "loss": 2.2535,
+      "step": 3968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000338722542422875,
+      "loss": 2.1112,
+      "step": 3969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003386512217606339,
+      "loss": 2.4047,
+      "step": 3970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033857989284471714,
+      "loss": 2.2347,
+      "step": 3971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033850855568176564,
+      "loss": 2.2065,
+      "step": 3972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000338437210278421,
+      "loss": 2.2436,
+      "step": 3973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003383658566413258,
+      "loss": 2.2606,
+      "step": 3974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003382944947771232,
+      "loss": 2.4422,
+      "step": 3975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033822312469245717,
+      "loss": 2.2168,
+      "step": 3976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003381517463939726,
+      "loss": 2.1414,
+      "step": 3977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033808035988831483,
+      "loss": 2.0206,
+      "step": 3978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003380089651821302,
+      "loss": 2.3302,
+      "step": 3979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003379375622820658,
+      "loss": 2.14,
+      "step": 3980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033786615119476936,
+      "loss": 2.2575,
+      "step": 3981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003377947319268895,
+      "loss": 2.1416,
+      "step": 3982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003377233044850756,
+      "loss": 2.0723,
+      "step": 3983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033765186887597763,
+      "loss": 2.1655,
+      "step": 3984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003375804251062466,
+      "loss": 2.1844,
+      "step": 3985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033750897318253405,
+      "loss": 2.3325,
+      "step": 3986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003374375131114923,
+      "loss": 2.3065,
+      "step": 3987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003373660448997746,
+      "loss": 2.2322,
+      "step": 3988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003372945685540347,
+      "loss": 2.4646,
+      "step": 3989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003372230840809274,
+      "loss": 2.3271,
+      "step": 3990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033715159148710793,
+      "loss": 2.3473,
+      "step": 3991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003370800907792325,
+      "loss": 2.3077,
+      "step": 3992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003370085819639582,
+      "loss": 2.3023,
+      "step": 3993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033693706504794245,
+      "loss": 2.3099,
+      "step": 3994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033686554003784386,
+      "loss": 2.2612,
+      "step": 3995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003367940069403216,
+      "loss": 2.1851,
+      "step": 3996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003367224657620354,
+      "loss": 2.1219,
+      "step": 3997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033665091650964614,
+      "loss": 2.4409,
+      "step": 3998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003365793591898152,
+      "loss": 2.2201,
+      "step": 3999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003365077938092047,
+      "loss": 2.3853,
+      "step": 4000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 2.2590601444244385,
+      "eval_runtime": 183.9167,
+      "eval_samples_per_second": 457.838,
+      "eval_steps_per_second": 28.616,
+      "step": 4000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003364362203744777,
+      "loss": 2.2351,
+      "step": 4001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003363646388922978,
+      "loss": 2.1032,
+      "step": 4002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033629304936932947,
+      "loss": 2.2822,
+      "step": 4003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033622145181223784,
+      "loss": 2.1617,
+      "step": 4004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033614984622768894,
+      "loss": 2.2259,
+      "step": 4005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003360782326223493,
+      "loss": 2.2402,
+      "step": 4006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003360066110028866,
+      "loss": 2.2007,
+      "step": 4007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033593498137596874,
+      "loss": 2.3536,
+      "step": 4008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003358633437482647,
+      "loss": 2.3031,
+      "step": 4009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033579169812644435,
+      "loss": 2.2561,
+      "step": 4010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003357200445171778,
+      "loss": 2.3492,
+      "step": 4011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003356483829271365,
+      "loss": 2.2759,
+      "step": 4012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003355767133629921,
+      "loss": 2.0877,
+      "step": 4013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003355050358314172,
+      "loss": 2.3099,
+      "step": 4014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003354333503390854,
+      "loss": 2.4537,
+      "step": 4015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033536165689267074,
+      "loss": 2.2365,
+      "step": 4016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000335289955498848,
+      "loss": 2.285,
+      "step": 4017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033521824616429286,
+      "loss": 2.0896,
+      "step": 4018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003351465288956816,
+      "loss": 2.1751,
+      "step": 4019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033507480369969136,
+      "loss": 2.1743,
+      "step": 4020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033500307058299995,
+      "loss": 2.1735,
+      "step": 4021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003349313295522859,
+      "loss": 2.2854,
+      "step": 4022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003348595806142284,
+      "loss": 2.286,
+      "step": 4023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003347878237755078,
+      "loss": 2.0196,
+      "step": 4024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033471605904280443,
+      "loss": 2.0852,
+      "step": 4025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003346442864228,
+      "loss": 2.2587,
+      "step": 4026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033457250592217687,
+      "loss": 2.2006,
+      "step": 4027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003345007175476178,
+      "loss": 2.0086,
+      "step": 4028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033442892130580655,
+      "loss": 2.1884,
+      "step": 4029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003343571172034276,
+      "loss": 2.333,
+      "step": 4030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000334285305247166,
+      "loss": 2.1079,
+      "step": 4031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003342134854437078,
+      "loss": 2.2139,
+      "step": 4032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003341416577997394,
+      "loss": 2.131,
+      "step": 4033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033406982232194837,
+      "loss": 2.0768,
+      "step": 4034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003339979790170227,
+      "loss": 2.3897,
+      "step": 4035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003339261278916512,
+      "loss": 2.2825,
+      "step": 4036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033385426895252347,
+      "loss": 2.3421,
+      "step": 4037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033378240220632973,
+      "loss": 2.1903,
+      "step": 4038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033371052765976097,
+      "loss": 2.1577,
+      "step": 4039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003336386453195088,
+      "loss": 2.2244,
+      "step": 4040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033356675519226597,
+      "loss": 2.2194,
+      "step": 4041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033349485728472535,
+      "loss": 2.2252,
+      "step": 4042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000333422951603581,
+      "loss": 2.2449,
+      "step": 4043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003333510381555274,
+      "loss": 2.1693,
+      "step": 4044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003332791169472601,
+      "loss": 2.163,
+      "step": 4045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033320718798547503,
+      "loss": 1.9653,
+      "step": 4046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033313525127686897,
+      "loss": 2.2205,
+      "step": 4047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003330633068281394,
+      "loss": 1.9497,
+      "step": 4048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003329913546459846,
+      "loss": 2.0853,
+      "step": 4049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033291939473710356,
+      "loss": 2.2737,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 2.252211570739746,
+      "eval_runtime": 183.9489,
+      "eval_samples_per_second": 457.757,
+      "eval_steps_per_second": 28.611,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033284742710819586,
+      "loss": 2.144,
+      "step": 4051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033277545176596204,
+      "loss": 2.2056,
+      "step": 4052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000332703468717103,
+      "loss": 2.0995,
+      "step": 4053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033263147796832064,
+      "loss": 2.1819,
+      "step": 4054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003325594795263176,
+      "loss": 2.0602,
+      "step": 4055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000332487473397797,
+      "loss": 2.2734,
+      "step": 4056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033241545958946286,
+      "loss": 2.2808,
+      "step": 4057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033234343810801994,
+      "loss": 2.1279,
+      "step": 4058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003322714089601735,
+      "loss": 2.1707,
+      "step": 4059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033219937215262977,
+      "loss": 2.4656,
+      "step": 4060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033212732769209545,
+      "loss": 2.2347,
+      "step": 4061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003320552755852783,
+      "loss": 2.3067,
+      "step": 4062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033198321583888636,
+      "loss": 2.2007,
+      "step": 4063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033191114845962865,
+      "loss": 2.1816,
+      "step": 4064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003318390734542149,
+      "loss": 2.1548,
+      "step": 4065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033176699082935546,
+      "loss": 2.1756,
+      "step": 4066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003316949005917614,
+      "loss": 2.0056,
+      "step": 4067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003316228027481445,
+      "loss": 2.2383,
+      "step": 4068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003315506973052174,
+      "loss": 2.0397,
+      "step": 4069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033147858426969313,
+      "loss": 2.0889,
+      "step": 4070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033140646364828574,
+      "loss": 2.1637,
+      "step": 4071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033133433544770983,
+      "loss": 2.0662,
+      "step": 4072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003312621996746808,
+      "loss": 1.9873,
+      "step": 4073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003311900563359146,
+      "loss": 2.2165,
+      "step": 4074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033111790543812803,
+      "loss": 2.242,
+      "step": 4075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033104574698803845,
+      "loss": 2.4094,
+      "step": 4076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003309735809923641,
+      "loss": 2.0638,
+      "step": 4077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003309014074578239,
+      "loss": 2.0863,
+      "step": 4078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003308292263911372,
+      "loss": 2.4022,
+      "step": 4079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003307570377990245,
+      "loss": 2.2894,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003306848416882066,
+      "loss": 2.3351,
+      "step": 4081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003306126380654051,
+      "loss": 2.1272,
+      "step": 4082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033054042693734257,
+      "loss": 2.2051,
+      "step": 4083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033046820831074186,
+      "loss": 2.0107,
+      "step": 4084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003303959821923269,
+      "loss": 2.306,
+      "step": 4085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003303237485888221,
+      "loss": 2.123,
+      "step": 4086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003302515075069524,
+      "loss": 2.1531,
+      "step": 4087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000330179258953444,
+      "loss": 2.1652,
+      "step": 4088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033010700293502316,
+      "loss": 2.2299,
+      "step": 4089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033003473945841723,
+      "loss": 2.1812,
+      "step": 4090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032996246853035416,
+      "loss": 2.1935,
+      "step": 4091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003298901901575625,
+      "loss": 2.2986,
+      "step": 4092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032981790434677167,
+      "loss": 2.0516,
+      "step": 4093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003297456111047116,
+      "loss": 2.2651,
+      "step": 4094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000329673310438113,
+      "loss": 2.0676,
+      "step": 4095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032960100235370724,
+      "loss": 2.4128,
+      "step": 4096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032952868685822643,
+      "loss": 2.2345,
+      "step": 4097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003294563639584034,
+      "loss": 2.2536,
+      "step": 4098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032938403366097153,
+      "loss": 2.3345,
+      "step": 4099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032931169597266507,
+      "loss": 2.3751,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 2.2451629638671875,
+      "eval_runtime": 184.0288,
+      "eval_samples_per_second": 457.559,
+      "eval_steps_per_second": 28.599,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003292393509002187,
+      "loss": 2.2808,
+      "step": 4101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032916699845036816,
+      "loss": 2.3525,
+      "step": 4102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032909463862984946,
+      "loss": 2.0432,
+      "step": 4103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032902227144539963,
+      "loss": 2.277,
+      "step": 4104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032894989690375627,
+      "loss": 2.3058,
+      "step": 4105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032887751501165755,
+      "loss": 2.0599,
+      "step": 4106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003288051257758424,
+      "loss": 2.036,
+      "step": 4107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003287327292030506,
+      "loss": 2.2094,
+      "step": 4108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003286603253000224,
+      "loss": 2.2101,
+      "step": 4109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032858791407349894,
+      "loss": 2.2884,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032851549553022165,
+      "loss": 2.18,
+      "step": 4111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032844306967693305,
+      "loss": 2.2971,
+      "step": 4112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003283706365203762,
+      "loss": 2.1449,
+      "step": 4113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032829819606729476,
+      "loss": 2.2912,
+      "step": 4114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003282257483244332,
+      "loss": 2.2232,
+      "step": 4115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003281532932985365,
+      "loss": 2.1795,
+      "step": 4116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032808083099635063,
+      "loss": 2.2172,
+      "step": 4117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032800836142462175,
+      "loss": 2.3062,
+      "step": 4118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032793588459009725,
+      "loss": 2.2199,
+      "step": 4119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003278634004995248,
+      "loss": 2.2229,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032779090915965284,
+      "loss": 2.1235,
+      "step": 4121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032771841057723064,
+      "loss": 2.0098,
+      "step": 4122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032764590475900784,
+      "loss": 2.1864,
+      "step": 4123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032757339171173505,
+      "loss": 2.177,
+      "step": 4124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032750087144216343,
+      "loss": 2.2404,
+      "step": 4125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032742834395704486,
+      "loss": 2.1417,
+      "step": 4126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003273558092631318,
+      "loss": 2.1727,
+      "step": 4127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003272832673671774,
+      "loss": 2.0811,
+      "step": 4128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003272107182759355,
+      "loss": 2.1439,
+      "step": 4129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003271381619961608,
+      "loss": 2.1927,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032706559853460813,
+      "loss": 2.3455,
+      "step": 4131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003269930278980337,
+      "loss": 2.0301,
+      "step": 4132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003269204500931939,
+      "loss": 2.2962,
+      "step": 4133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032684786512684593,
+      "loss": 2.2953,
+      "step": 4134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032677527300574774,
+      "loss": 2.2851,
+      "step": 4135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003267026737366577,
+      "loss": 2.2289,
+      "step": 4136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032663006732633516,
+      "loss": 2.2557,
+      "step": 4137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003265574537815398,
+      "loss": 2.063,
+      "step": 4138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003264848331090323,
+      "loss": 2.2414,
+      "step": 4139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032641220531557383,
+      "loss": 2.4123,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003263395704079262,
+      "loss": 2.3275,
+      "step": 4141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003262669283928519,
+      "loss": 2.1007,
+      "step": 4142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003261942792771142,
+      "loss": 2.3988,
+      "step": 4143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003261216230674768,
+      "loss": 2.1878,
+      "step": 4144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032604895977070427,
+      "loss": 2.2634,
+      "step": 4145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003259762893935617,
+      "loss": 2.3006,
+      "step": 4146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003259036119428151,
+      "loss": 1.9554,
+      "step": 4147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003258309274252307,
+      "loss": 2.3114,
+      "step": 4148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032575823584757573,
+      "loss": 2.1733,
+      "step": 4149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000325685537216618,
+      "loss": 2.0766,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 2.2410128116607666,
+      "eval_runtime": 184.0187,
+      "eval_samples_per_second": 457.584,
+      "eval_steps_per_second": 28.6,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000325612831539126,
+      "loss": 2.0841,
+      "step": 4151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032554011882186866,
+      "loss": 2.2107,
+      "step": 4152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000325467399071616,
+      "loss": 2.1136,
+      "step": 4153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032539467229513813,
+      "loss": 2.2062,
+      "step": 4154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003253219384992064,
+      "loss": 2.3711,
+      "step": 4155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032524919769059225,
+      "loss": 2.1772,
+      "step": 4156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032517644987606825,
+      "loss": 2.1889,
+      "step": 4157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032510369506240746,
+      "loss": 2.3236,
+      "step": 4158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032503093325638333,
+      "loss": 2.4431,
+      "step": 4159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032495816446477044,
+      "loss": 2.255,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003248853886943436,
+      "loss": 2.2564,
+      "step": 4161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032481260595187847,
+      "loss": 2.3097,
+      "step": 4162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003247398162441513,
+      "loss": 2.419,
+      "step": 4163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032466701957793915,
+      "loss": 2.1231,
+      "step": 4164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032459421596001944,
+      "loss": 2.1614,
+      "step": 4165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032452140539717046,
+      "loss": 2.0354,
+      "step": 4166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032444858789617106,
+      "loss": 2.3559,
+      "step": 4167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032437576346380076,
+      "loss": 2.3534,
+      "step": 4168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032430293210683974,
+      "loss": 2.1803,
+      "step": 4169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032423009383206875,
+      "loss": 2.1345,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003241572486462692,
+      "loss": 2.24,
+      "step": 4171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032408439655622333,
+      "loss": 2.1266,
+      "step": 4172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032401153756871374,
+      "loss": 2.3826,
+      "step": 4173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032393867169052385,
+      "loss": 2.0864,
+      "step": 4174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003238657989284378,
+      "loss": 2.1325,
+      "step": 4175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032379291928924,
+      "loss": 2.1699,
+      "step": 4176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003237200327797159,
+      "loss": 2.2558,
+      "step": 4177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032364713940665147,
+      "loss": 2.328,
+      "step": 4178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003235742391768331,
+      "loss": 2.2926,
+      "step": 4179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003235013320970481,
+      "loss": 2.253,
+      "step": 4180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003234284181740845,
+      "loss": 2.4692,
+      "step": 4181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003233554974147306,
+      "loss": 2.2771,
+      "step": 4182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003232825698257755,
+      "loss": 2.1642,
+      "step": 4183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003232096354140091,
+      "loss": 2.5149,
+      "step": 4184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032313669418622165,
+      "loss": 2.2719,
+      "step": 4185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032306374614920433,
+      "loss": 2.2975,
+      "step": 4186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003229907913097487,
+      "loss": 2.2189,
+      "step": 4187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032291782967464706,
+      "loss": 2.4444,
+      "step": 4188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032284486125069243,
+      "loss": 2.0163,
+      "step": 4189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032277188604467824,
+      "loss": 2.193,
+      "step": 4190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032269890406339875,
+      "loss": 2.1668,
+      "step": 4191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003226259153136489,
+      "loss": 2.2588,
+      "step": 4192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000322552919802224,
+      "loss": 2.3429,
+      "step": 4193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032247991753592015,
+      "loss": 2.2994,
+      "step": 4194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032240690852153417,
+      "loss": 2.1422,
+      "step": 4195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032233389276586325,
+      "loss": 2.336,
+      "step": 4196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032226087027570544,
+      "loss": 2.1565,
+      "step": 4197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003221878410578593,
+      "loss": 2.0898,
+      "step": 4198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003221148051191241,
+      "loss": 2.1485,
+      "step": 4199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032204176246629975,
+      "loss": 2.2456,
+      "step": 4200
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 2.2332441806793213,
+      "eval_runtime": 184.0047,
+      "eval_samples_per_second": 457.619,
+      "eval_steps_per_second": 28.603,
+      "step": 4200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003219687131061865,
+      "loss": 2.2418,
+      "step": 4201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003218956570455857,
+      "loss": 2.1565,
+      "step": 4202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000321822594291299,
+      "loss": 2.464,
+      "step": 4203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032174952485012866,
+      "loss": 2.4521,
+      "step": 4204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003216764487288777,
+      "loss": 2.3117,
+      "step": 4205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003216033659343497,
+      "loss": 2.1616,
+      "step": 4206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032153027647334893,
+      "loss": 2.1952,
+      "step": 4207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032145718035268025,
+      "loss": 2.2244,
+      "step": 4208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003213840775791489,
+      "loss": 2.1306,
+      "step": 4209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003213109681595612,
+      "loss": 2.445,
+      "step": 4210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003212378521007237,
+      "loss": 2.2365,
+      "step": 4211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003211647294094437,
+      "loss": 2.2568,
+      "step": 4212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003210916000925292,
+      "loss": 2.0266,
+      "step": 4213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003210184641567887,
+      "loss": 2.266,
+      "step": 4214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003209453216090314,
+      "loss": 2.1333,
+      "step": 4215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000320872172456067,
+      "loss": 2.1653,
+      "step": 4216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000320799016704706,
+      "loss": 2.2797,
+      "step": 4217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003207258543617593,
+      "loss": 2.1944,
+      "step": 4218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003206526854340385,
+      "loss": 2.2025,
+      "step": 4219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000320579509928356,
+      "loss": 2.1639,
+      "step": 4220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032050632785152444,
+      "loss": 2.3984,
+      "step": 4221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032043313921035745,
+      "loss": 2.2941,
+      "step": 4222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032035994401166895,
+      "loss": 2.0162,
+      "step": 4223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003202867422622737,
+      "loss": 2.2752,
+      "step": 4224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000320213533968987,
+      "loss": 2.3101,
+      "step": 4225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032014031913862466,
+      "loss": 1.9355,
+      "step": 4226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003200670977780033,
+      "loss": 2.1544,
+      "step": 4227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031999386989393995,
+      "loss": 1.9896,
+      "step": 4228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003199206354932523,
+      "loss": 2.0901,
+      "step": 4229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003198473945827588,
+      "loss": 2.3082,
+      "step": 4230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031977414716927833,
+      "loss": 2.2883,
+      "step": 4231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003197008932596303,
+      "loss": 2.0427,
+      "step": 4232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000319627632860635,
+      "loss": 2.278,
+      "step": 4233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031955436597911317,
+      "loss": 2.0496,
+      "step": 4234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003194810926218861,
+      "loss": 2.166,
+      "step": 4235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003194078127957758,
+      "loss": 2.1011,
+      "step": 4236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003193345265076048,
+      "loss": 2.3039,
+      "step": 4237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031926123376419625,
+      "loss": 1.9589,
+      "step": 4238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031918793457237394,
+      "loss": 2.0756,
+      "step": 4239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031911462893896217,
+      "loss": 2.2504,
+      "step": 4240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031904131687078585,
+      "loss": 2.3337,
+      "step": 4241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003189679983746708,
+      "loss": 2.3017,
+      "step": 4242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031889467345744293,
+      "loss": 2.1745,
+      "step": 4243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000318821342125929,
+      "loss": 2.3756,
+      "step": 4244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003187480043869565,
+      "loss": 2.1047,
+      "step": 4245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031867466024735323,
+      "loss": 2.3041,
+      "step": 4246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031860130971394786,
+      "loss": 2.5189,
+      "step": 4247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031852795279356945,
+      "loss": 2.065,
+      "step": 4248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003184545894930477,
+      "loss": 2.4029,
+      "step": 4249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031838121981921303,
+      "loss": 2.1394,
+      "step": 4250
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 2.227912187576294,
+      "eval_runtime": 183.9508,
+      "eval_samples_per_second": 457.753,
+      "eval_steps_per_second": 28.611,
+      "step": 4250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003183078437788964,
+      "loss": 2.3242,
+      "step": 4251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003182344613789291,
+      "loss": 2.2949,
+      "step": 4252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031816107262614356,
+      "loss": 2.3101,
+      "step": 4253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003180876775273722,
+      "loss": 2.2806,
+      "step": 4254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031801427608944843,
+      "loss": 2.3134,
+      "step": 4255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003179408683192061,
+      "loss": 2.1098,
+      "step": 4256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003178674542234797,
+      "loss": 2.201,
+      "step": 4257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031779403380910427,
+      "loss": 2.3161,
+      "step": 4258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003177206070829155,
+      "loss": 2.2078,
+      "step": 4259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031764717405174947,
+      "loss": 2.4502,
+      "step": 4260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031757373472244323,
+      "loss": 2.2807,
+      "step": 4261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031750028910183396,
+      "loss": 2.2898,
+      "step": 4262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031742683719675977,
+      "loss": 2.2148,
+      "step": 4263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003173533790140593,
+      "loss": 2.0867,
+      "step": 4264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031727991456057165,
+      "loss": 2.0333,
+      "step": 4265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031720644384313645,
+      "loss": 2.1205,
+      "step": 4266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003171329668685942,
+      "loss": 2.2158,
+      "step": 4267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031705948364378573,
+      "loss": 2.213,
+      "step": 4268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003169859941755525,
+      "loss": 2.1573,
+      "step": 4269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003169124984707367,
+      "loss": 2.2279,
+      "step": 4270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003168389965361809,
+      "loss": 2.3284,
+      "step": 4271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003167654883787283,
+      "loss": 2.2,
+      "step": 4272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003166919740052228,
+      "loss": 2.1627,
+      "step": 4273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003166184534225087,
+      "loss": 2.0919,
+      "step": 4274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031654492663743116,
+      "loss": 2.04,
+      "step": 4275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00031647139365683545,
+      "loss": 1.9415,
+      "step": 4276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000316397854487568,
+      "loss": 1.9497,
+      "step": 4277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003163243091364752,
+      "loss": 2.2274,
+      "step": 4278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031625075761040455,
+      "loss": 2.1733,
+      "step": 4279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003161771999162038,
+      "loss": 2.2181,
+      "step": 4280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031610363606072143,
+      "loss": 2.0952,
+      "step": 4281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003160300660508064,
+      "loss": 2.2102,
+      "step": 4282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031595648989330826,
+      "loss": 2.0972,
+      "step": 4283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003158829075950773,
+      "loss": 2.0698,
+      "step": 4284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003158093191629641,
+      "loss": 2.1017,
+      "step": 4285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003157357246038199,
+      "loss": 2.1355,
+      "step": 4286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031566212392449667,
+      "loss": 2.0008,
+      "step": 4287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031558851713184684,
+      "loss": 2.1536,
+      "step": 4288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031551490423272334,
+      "loss": 2.0845,
+      "step": 4289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003154412852339798,
+      "loss": 2.0897,
+      "step": 4290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003153676601424703,
+      "loss": 2.1966,
+      "step": 4291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031529402896504963,
+      "loss": 2.3021,
+      "step": 4292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000315220391708573,
+      "loss": 2.1588,
+      "step": 4293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003151467483798961,
+      "loss": 2.0502,
+      "step": 4294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003150730989858756,
+      "loss": 2.3199,
+      "step": 4295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031499944353336823,
+      "loss": 2.3096,
+      "step": 4296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031492578202923165,
+      "loss": 2.1643,
+      "step": 4297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031485211448032394,
+      "loss": 2.1354,
+      "step": 4298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003147784408935037,
+      "loss": 2.3447,
+      "step": 4299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031470476127563017,
+      "loss": 2.0715,
+      "step": 4300
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 2.221841812133789,
+      "eval_runtime": 184.0404,
+      "eval_samples_per_second": 457.53,
+      "eval_steps_per_second": 28.597,
+      "step": 4300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031463107563356316,
+      "loss": 2.1191,
+      "step": 4301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000314557383974163,
+      "loss": 1.97,
+      "step": 4302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003144836863042906,
+      "loss": 2.0819,
+      "step": 4303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003144099826308073,
+      "loss": 2.367,
+      "step": 4304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003143362729605753,
+      "loss": 2.2106,
+      "step": 4305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031426255730045697,
+      "loss": 2.2739,
+      "step": 4306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003141888356573157,
+      "loss": 2.2535,
+      "step": 4307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000314115108038015,
+      "loss": 2.1393,
+      "step": 4308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003140413744494191,
+      "loss": 2.144,
+      "step": 4309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031396763489839293,
+      "loss": 2.2956,
+      "step": 4310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031389388939180167,
+      "loss": 2.1019,
+      "step": 4311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003138201379365114,
+      "loss": 2.0898,
+      "step": 4312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003137463805393885,
+      "loss": 2.4159,
+      "step": 4313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031367261720730005,
+      "loss": 2.0816,
+      "step": 4314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031359884794711357,
+      "loss": 2.1911,
+      "step": 4315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003135250727656972,
+      "loss": 2.2462,
+      "step": 4316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031345129166991957,
+      "loss": 2.2057,
+      "step": 4317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031337750466665,
+      "loss": 2.233,
+      "step": 4318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003133037117627582,
+      "loss": 2.1317,
+      "step": 4319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031322991296511447,
+      "loss": 2.0411,
+      "step": 4320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031315610828058977,
+      "loss": 2.2214,
+      "step": 4321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031308229771605543,
+      "loss": 2.1481,
+      "step": 4322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031300848127838356,
+      "loss": 2.1472,
+      "step": 4323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031293465897444647,
+      "loss": 2.2928,
+      "step": 4324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031286083081111737,
+      "loss": 2.1266,
+      "step": 4325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031278699679526976,
+      "loss": 2.1686,
+      "step": 4326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000312713156933778,
+      "loss": 2.0975,
+      "step": 4327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003126393112335165,
+      "loss": 2.2077,
+      "step": 4328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003125654597013608,
+      "loss": 2.1194,
+      "step": 4329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031249160234418644,
+      "loss": 2.1301,
+      "step": 4330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031241773916886983,
+      "loss": 2.3448,
+      "step": 4331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003123438701822879,
+      "loss": 2.2208,
+      "step": 4332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031226999539131793,
+      "loss": 2.055,
+      "step": 4333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031219611480283796,
+      "loss": 2.2909,
+      "step": 4334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003121222284237265,
+      "loss": 2.0184,
+      "step": 4335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031204833626086246,
+      "loss": 2.3682,
+      "step": 4336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003119744383211256,
+      "loss": 2.2544,
+      "step": 4337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031190053461139583,
+      "loss": 2.3599,
+      "step": 4338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031182662513855384,
+      "loss": 2.1761,
+      "step": 4339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031175270990948097,
+      "loss": 2.2672,
+      "step": 4340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003116787889310587,
+      "loss": 2.0539,
+      "step": 4341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031160486221016936,
+      "loss": 2.1648,
+      "step": 4342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003115309297536958,
+      "loss": 2.2406,
+      "step": 4343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003114569915685213,
+      "loss": 2.2368,
+      "step": 4344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031138304766152965,
+      "loss": 2.2617,
+      "step": 4345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003113090980396053,
+      "loss": 2.1783,
+      "step": 4346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031123514270963316,
+      "loss": 2.2035,
+      "step": 4347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031116118167849865,
+      "loss": 2.2979,
+      "step": 4348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003110872149530878,
+      "loss": 2.0251,
+      "step": 4349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031101324254028713,
+      "loss": 2.2069,
+      "step": 4350
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 2.2148072719573975,
+      "eval_runtime": 183.9749,
+      "eval_samples_per_second": 457.693,
+      "eval_steps_per_second": 28.607,
+      "step": 4350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003109392644469836,
+      "loss": 2.2192,
+      "step": 4351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031086528068006476,
+      "loss": 2.3222,
+      "step": 4352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031079129124641873,
+      "loss": 2.1963,
+      "step": 4353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003107172961529343,
+      "loss": 2.2788,
+      "step": 4354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003106432954065003,
+      "loss": 2.2768,
+      "step": 4355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031056928901400664,
+      "loss": 2.0762,
+      "step": 4356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003104952769823435,
+      "loss": 2.01,
+      "step": 4357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003104212593184017,
+      "loss": 2.3166,
+      "step": 4358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003103472360290721,
+      "loss": 2.2506,
+      "step": 4359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003102732071212469,
+      "loss": 2.0911,
+      "step": 4360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003101991726018182,
+      "loss": 2.3677,
+      "step": 4361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031012513247767875,
+      "loss": 2.3252,
+      "step": 4362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003100510867557221,
+      "loss": 1.9612,
+      "step": 4363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030997703544284193,
+      "loss": 2.1517,
+      "step": 4364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003099029785459328,
+      "loss": 2.4251,
+      "step": 4365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030982891607188946,
+      "loss": 2.1142,
+      "step": 4366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030975484802760737,
+      "loss": 2.1038,
+      "step": 4367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030968077441998257,
+      "loss": 2.0992,
+      "step": 4368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030960669525591136,
+      "loss": 2.0703,
+      "step": 4369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003095326105422908,
+      "loss": 2.4202,
+      "step": 4370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030945852028601835,
+      "loss": 2.0784,
+      "step": 4371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030938442449399205,
+      "loss": 2.1671,
+      "step": 4372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003093103231731105,
+      "loss": 2.1458,
+      "step": 4373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030923621633027255,
+      "loss": 2.0602,
+      "step": 4374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030916210397237794,
+      "loss": 2.2504,
+      "step": 4375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003090879861063266,
+      "loss": 2.1925,
+      "step": 4376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003090138627390193,
+      "loss": 2.0677,
+      "step": 4377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003089397338773569,
+      "loss": 2.2188,
+      "step": 4378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030886559952824104,
+      "loss": 2.1462,
+      "step": 4379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00030879145969857396,
+      "loss": 2.3341,
+      "step": 4380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003087173143952582,
+      "loss": 2.1336,
+      "step": 4381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000308643163625197,
+      "loss": 2.4442,
+      "step": 4382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030856900739529387,
+      "loss": 2.1752,
+      "step": 4383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030849484571245295,
+      "loss": 2.1142,
+      "step": 4384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003084206785835791,
+      "loss": 2.2891,
+      "step": 4385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030834650601557724,
+      "loss": 2.3459,
+      "step": 4386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030827232801535317,
+      "loss": 2.3099,
+      "step": 4387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030819814458981303,
+      "loss": 2.378,
+      "step": 4388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030812395574586357,
+      "loss": 2.5298,
+      "step": 4389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000308049761490412,
+      "loss": 2.1228,
+      "step": 4390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003079755618303658,
+      "loss": 2.3084,
+      "step": 4391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003079013567726334,
+      "loss": 2.2799,
+      "step": 4392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003078271463241234,
+      "loss": 1.9998,
+      "step": 4393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030775293049174497,
+      "loss": 2.1815,
+      "step": 4394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003076787092824078,
+      "loss": 2.2024,
+      "step": 4395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003076044827030222,
+      "loss": 2.1063,
+      "step": 4396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030753025076049877,
+      "loss": 2.2605,
+      "step": 4397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003074560134617488,
+      "loss": 2.1732,
+      "step": 4398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003073817708136839,
+      "loss": 2.0717,
+      "step": 4399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030730752282321635,
+      "loss": 2.1304,
+      "step": 4400
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 2.2079806327819824,
+      "eval_runtime": 183.9744,
+      "eval_samples_per_second": 457.694,
+      "eval_steps_per_second": 28.607,
+      "step": 4400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030723326949725873,
+      "loss": 2.1929,
+      "step": 4401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030715901084272433,
+      "loss": 2.2695,
+      "step": 4402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003070847468665268,
+      "loss": 2.2342,
+      "step": 4403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003070104775755804,
+      "loss": 2.3245,
+      "step": 4404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003069362029767997,
+      "loss": 2.1659,
+      "step": 4405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030686192307709985,
+      "loss": 2.2859,
+      "step": 4406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003067876378833967,
+      "loss": 2.3484,
+      "step": 4407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030671334740260626,
+      "loss": 2.2447,
+      "step": 4408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003066390516416451,
+      "loss": 2.128,
+      "step": 4409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003065647506074306,
+      "loss": 2.1866,
+      "step": 4410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030649044430688016,
+      "loss": 2.3812,
+      "step": 4411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000306416132746912,
+      "loss": 2.3022,
+      "step": 4412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003063418159344448,
+      "loss": 2.2359,
+      "step": 4413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030626749387639756,
+      "loss": 2.2064,
+      "step": 4414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030619316657968994,
+      "loss": 2.353,
+      "step": 4415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000306118834051242,
+      "loss": 2.4857,
+      "step": 4416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003060444962979743,
+      "loss": 2.0644,
+      "step": 4417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003059701533268079,
+      "loss": 2.1092,
+      "step": 4418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030589580514466425,
+      "loss": 2.1167,
+      "step": 4419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030582145175846545,
+      "loss": 2.3681,
+      "step": 4420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003057470931751341,
+      "loss": 2.077,
+      "step": 4421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000305672729401593,
+      "loss": 2.2314,
+      "step": 4422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003055983604447658,
+      "loss": 2.1142,
+      "step": 4423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003055239863115763,
+      "loss": 2.1247,
+      "step": 4424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030544960700894914,
+      "loss": 2.1385,
+      "step": 4425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000305375222543809,
+      "loss": 2.1204,
+      "step": 4426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030530083292308147,
+      "loss": 2.3435,
+      "step": 4427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003052264381536923,
+      "loss": 2.2416,
+      "step": 4428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003051520382425679,
+      "loss": 2.1871,
+      "step": 4429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030507763319663516,
+      "loss": 2.2153,
+      "step": 4430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030500322302282133,
+      "loss": 2.1954,
+      "step": 4431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003049288077280543,
+      "loss": 2.0354,
+      "step": 4432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003048543873192622,
+      "loss": 2.0888,
+      "step": 4433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030477996180337385,
+      "loss": 2.0751,
+      "step": 4434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030470553118731853,
+      "loss": 2.3343,
+      "step": 4435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003046310954780258,
+      "loss": 2.1632,
+      "step": 4436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030455665468242594,
+      "loss": 2.238,
+      "step": 4437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003044822088074496,
+      "loss": 2.2418,
+      "step": 4438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030440775786002785,
+      "loss": 2.0914,
+      "step": 4439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003043333018470923,
+      "loss": 2.0503,
+      "step": 4440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000304258840775575,
+      "loss": 2.1098,
+      "step": 4441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003041843746524085,
+      "loss": 2.2256,
+      "step": 4442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030410990348452574,
+      "loss": 2.3067,
+      "step": 4443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003040354272788603,
+      "loss": 2.1934,
+      "step": 4444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000303960946042346,
+      "loss": 2.1483,
+      "step": 4445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003038864597819174,
+      "loss": 2.0626,
+      "step": 4446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003038119685045093,
+      "loss": 2.1054,
+      "step": 4447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000303737472217057,
+      "loss": 2.1632,
+      "step": 4448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030366297092649644,
+      "loss": 2.151,
+      "step": 4449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003035884646397637,
+      "loss": 2.1161,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 2.1998133659362793,
+      "eval_runtime": 183.9737,
+      "eval_samples_per_second": 457.696,
+      "eval_steps_per_second": 28.607,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003035139533637956,
+      "loss": 2.2392,
+      "step": 4451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003034394371055295,
+      "loss": 2.1045,
+      "step": 4452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000303364915871903,
+      "loss": 2.2454,
+      "step": 4453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000303290389669854,
+      "loss": 2.1609,
+      "step": 4454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030321585850632144,
+      "loss": 2.2297,
+      "step": 4455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003031413223882442,
+      "loss": 2.1378,
+      "step": 4456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003030667813225618,
+      "loss": 2.51,
+      "step": 4457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003029922353162143,
+      "loss": 2.1923,
+      "step": 4458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030291768437614195,
+      "loss": 2.2346,
+      "step": 4459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030284312850928596,
+      "loss": 2.2407,
+      "step": 4460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003027685677225874,
+      "loss": 2.1591,
+      "step": 4461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030269400202298817,
+      "loss": 2.1532,
+      "step": 4462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003026194314174307,
+      "loss": 2.2118,
+      "step": 4463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030254485591285754,
+      "loss": 2.2395,
+      "step": 4464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003024702755162119,
+      "loss": 2.181,
+      "step": 4465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030239569023443754,
+      "loss": 2.197,
+      "step": 4466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030232110007447836,
+      "loss": 2.2725,
+      "step": 4467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003022465050432791,
+      "loss": 2.1756,
+      "step": 4468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003021719051477847,
+      "loss": 2.3554,
+      "step": 4469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003020973003949406,
+      "loss": 2.385,
+      "step": 4470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003020226907916927,
+      "loss": 2.1595,
+      "step": 4471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003019480763449874,
+      "loss": 2.1002,
+      "step": 4472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030187345706177154,
+      "loss": 2.3598,
+      "step": 4473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003017988329489923,
+      "loss": 2.3235,
+      "step": 4474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003017242040135975,
+      "loss": 2.2241,
+      "step": 4475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003016495702625351,
+      "loss": 2.1429,
+      "step": 4476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003015749317027539,
+      "loss": 2.0964,
+      "step": 4477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003015002883412029,
+      "loss": 2.1332,
+      "step": 4478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003014256401848316,
+      "loss": 2.2004,
+      "step": 4479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030135098724059,
+      "loss": 2.3244,
+      "step": 4480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003012763295154284,
+      "loss": 2.137,
+      "step": 4481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003012016670162977,
+      "loss": 2.3197,
+      "step": 4482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003011269997501492,
+      "loss": 2.0893,
+      "step": 4483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003010523277239346,
+      "loss": 2.0999,
+      "step": 4484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030097765094460607,
+      "loss": 2.3435,
+      "step": 4485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003009029694191163,
+      "loss": 2.2929,
+      "step": 4486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030082828315441826,
+      "loss": 2.2188,
+      "step": 4487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003007535921574655,
+      "loss": 2.2086,
+      "step": 4488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003006788964352119,
+      "loss": 2.1321,
+      "step": 4489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000300604195994612,
+      "loss": 2.1438,
+      "step": 4490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003005294908426203,
+      "loss": 2.1092,
+      "step": 4491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030045478098619246,
+      "loss": 1.9162,
+      "step": 4492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003003800664322838,
+      "loss": 2.05,
+      "step": 4493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030030534718785074,
+      "loss": 2.2002,
+      "step": 4494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030023062325984967,
+      "loss": 2.2058,
+      "step": 4495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030015589465523767,
+      "loss": 1.9724,
+      "step": 4496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030008116138097223,
+      "loss": 2.2395,
+      "step": 4497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030000642344401115,
+      "loss": 2.0501,
+      "step": 4498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002999316808513127,
+      "loss": 2.0821,
+      "step": 4499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029985693360983575,
+      "loss": 2.1512,
+      "step": 4500
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 2.1965131759643555,
+      "eval_runtime": 184.2325,
+      "eval_samples_per_second": 457.053,
+      "eval_steps_per_second": 28.567,
+      "step": 4500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029978218172653937,
+      "loss": 2.0988,
+      "step": 4501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029970742520838323,
+      "loss": 2.1329,
+      "step": 4502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002996326640623273,
+      "loss": 2.0704,
+      "step": 4503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029955789829533215,
+      "loss": 2.2302,
+      "step": 4504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002994831279143586,
+      "loss": 2.283,
+      "step": 4505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000299408352926368,
+      "loss": 2.0505,
+      "step": 4506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002993335733383222,
+      "loss": 2.0783,
+      "step": 4507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002992587891571833,
+      "loss": 2.2623,
+      "step": 4508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002991840003899138,
+      "loss": 2.1573,
+      "step": 4509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002991092070434769,
+      "loss": 2.2674,
+      "step": 4510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029903440912483617,
+      "loss": 2.2375,
+      "step": 4511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029895960664095525,
+      "loss": 2.1648,
+      "step": 4512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029888479959879854,
+      "loss": 2.1308,
+      "step": 4513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029880998800533096,
+      "loss": 2.0141,
+      "step": 4514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002987351718675174,
+      "loss": 2.2229,
+      "step": 4515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002986603511923237,
+      "loss": 2.2842,
+      "step": 4516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029858552598671575,
+      "loss": 2.299,
+      "step": 4517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985106962576599,
+      "loss": 2.2924,
+      "step": 4518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029843586201212325,
+      "loss": 2.2945,
+      "step": 4519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029836102325707283,
+      "loss": 2.1442,
+      "step": 4520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029828617999947646,
+      "loss": 2.323,
+      "step": 4521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029821133224630224,
+      "loss": 2.2732,
+      "step": 4522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002981364800045188,
+      "loss": 2.0618,
+      "step": 4523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002980616232810949,
+      "loss": 2.0857,
+      "step": 4524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000297986762083,
+      "loss": 2.0487,
+      "step": 4525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029791189641720385,
+      "loss": 2.1439,
+      "step": 4526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029783702629067674,
+      "loss": 2.0706,
+      "step": 4527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029776215171038923,
+      "loss": 2.203,
+      "step": 4528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029768727268331233,
+      "loss": 2.0848,
+      "step": 4529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029761238921641756,
+      "loss": 2.1579,
+      "step": 4530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029753750131667666,
+      "loss": 1.9626,
+      "step": 4531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000297462608991062,
+      "loss": 2.1281,
+      "step": 4532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029738771224654633,
+      "loss": 2.1823,
+      "step": 4533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029731281109010257,
+      "loss": 2.2361,
+      "step": 4534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002972379055287043,
+      "loss": 2.2061,
+      "step": 4535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002971629955693255,
+      "loss": 2.0938,
+      "step": 4536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002970880812189404,
+      "loss": 2.1553,
+      "step": 4537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002970131624845239,
+      "loss": 2.2239,
+      "step": 4538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029693823937305085,
+      "loss": 1.8822,
+      "step": 4539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002968633118914971,
+      "loss": 2.0759,
+      "step": 4540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002967883800468384,
+      "loss": 2.319,
+      "step": 4541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029671344384605126,
+      "loss": 2.3574,
+      "step": 4542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002966385032961123,
+      "loss": 2.2568,
+      "step": 4543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029656355840399883,
+      "loss": 2.2248,
+      "step": 4544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029648860917668834,
+      "loss": 2.1409,
+      "step": 4545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029641365562115885,
+      "loss": 2.1221,
+      "step": 4546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029633869774438885,
+      "loss": 2.0566,
+      "step": 4547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029626373555335684,
+      "loss": 2.1014,
+      "step": 4548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029618876905504227,
+      "loss": 2.3567,
+      "step": 4549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029611379825642464,
+      "loss": 2.0955,
+      "step": 4550
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 2.1923842430114746,
+      "eval_runtime": 184.0268,
+      "eval_samples_per_second": 457.564,
+      "eval_steps_per_second": 28.599,
+      "step": 4550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029603882316448394,
+      "loss": 2.2594,
+      "step": 4551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002959638437862006,
+      "loss": 2.1926,
+      "step": 4552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029588886012855536,
+      "loss": 1.9976,
+      "step": 4553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002958138721985294,
+      "loss": 2.0328,
+      "step": 4554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002957388800031044,
+      "loss": 2.1284,
+      "step": 4555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002956638835492622,
+      "loss": 2.2955,
+      "step": 4556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029558888284398526,
+      "loss": 2.1969,
+      "step": 4557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955138778942564,
+      "loss": 2.1688,
+      "step": 4558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954388687070587,
+      "loss": 2.12,
+      "step": 4559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029536385528937566,
+      "loss": 2.2205,
+      "step": 4560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002952888376481915,
+      "loss": 2.0638,
+      "step": 4561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029521381579049024,
+      "loss": 2.2044,
+      "step": 4562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029513878972325686,
+      "loss": 2.1308,
+      "step": 4563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002950637594534765,
+      "loss": 2.1661,
+      "step": 4564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002949887249881345,
+      "loss": 2.0535,
+      "step": 4565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002949136863342169,
+      "loss": 2.089,
+      "step": 4566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002948386434987101,
+      "loss": 2.2362,
+      "step": 4567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002947635964886006,
+      "loss": 2.0377,
+      "step": 4568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029468854531087565,
+      "loss": 2.3604,
+      "step": 4569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002946134899725226,
+      "loss": 2.1416,
+      "step": 4570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029453843048052943,
+      "loss": 2.3434,
+      "step": 4571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002944633668418843,
+      "loss": 2.0522,
+      "step": 4572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002943882990635759,
+      "loss": 2.394,
+      "step": 4573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002943132271525931,
+      "loss": 2.1784,
+      "step": 4574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029423815111592554,
+      "loss": 2.2592,
+      "step": 4575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029416307096056285,
+      "loss": 2.138,
+      "step": 4576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002940879866934952,
+      "loss": 2.2184,
+      "step": 4577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029401289832171326,
+      "loss": 2.2994,
+      "step": 4578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002939378058522078,
+      "loss": 2.1885,
+      "step": 4579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002938627092919703,
+      "loss": 2.1527,
+      "step": 4580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029378760864799236,
+      "loss": 2.1361,
+      "step": 4581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029371250392726614,
+      "loss": 2.0626,
+      "step": 4582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029363739513678394,
+      "loss": 2.1609,
+      "step": 4583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002935622822835388,
+      "loss": 2.2181,
+      "step": 4584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002934871653745238,
+      "loss": 1.9591,
+      "step": 4585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029341204441673266,
+      "loss": 2.1619,
+      "step": 4586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002933369194171591,
+      "loss": 2.2319,
+      "step": 4587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029326179038279774,
+      "loss": 2.3914,
+      "step": 4588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002931866573206432,
+      "loss": 2.0827,
+      "step": 4589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002931115202376906,
+      "loss": 2.1343,
+      "step": 4590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029303637914093527,
+      "loss": 2.128,
+      "step": 4591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002929612340373733,
+      "loss": 2.0502,
+      "step": 4592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029288608493400066,
+      "loss": 2.179,
+      "step": 4593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029281093183781406,
+      "loss": 2.1377,
+      "step": 4594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002927357747558105,
+      "loss": 2.2078,
+      "step": 4595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029266061369498733,
+      "loss": 2.308,
+      "step": 4596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000292585448662342,
+      "loss": 2.0763,
+      "step": 4597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000292510279664873,
+      "loss": 2.0681,
+      "step": 4598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002924351067095785,
+      "loss": 2.3341,
+      "step": 4599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002923599298034574,
+      "loss": 2.2466,
+      "step": 4600
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 2.1825172901153564,
+      "eval_runtime": 183.9909,
+      "eval_samples_per_second": 457.653,
+      "eval_steps_per_second": 28.605,
+      "step": 4600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002922847489535087,
+      "loss": 2.2464,
+      "step": 4601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002922095641667322,
+      "loss": 2.3309,
+      "step": 4602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029213437545012774,
+      "loss": 2.2712,
+      "step": 4603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029205918281069546,
+      "loss": 2.2299,
+      "step": 4604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002919839862554362,
+      "loss": 2.2374,
+      "step": 4605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002919087857913508,
+      "loss": 2.1243,
+      "step": 4606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029183358142544075,
+      "loss": 2.1816,
+      "step": 4607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002917583731647077,
+      "loss": 2.1543,
+      "step": 4608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002916831610161539,
+      "loss": 2.035,
+      "step": 4609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029160794498678163,
+      "loss": 2.142,
+      "step": 4610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029153272508359375,
+      "loss": 2.1408,
+      "step": 4611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029145750131359357,
+      "loss": 2.2633,
+      "step": 4612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002913822736837844,
+      "loss": 2.0838,
+      "step": 4613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002913070422011705,
+      "loss": 2.1133,
+      "step": 4614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029123180687275576,
+      "loss": 1.8635,
+      "step": 4615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029115656770554494,
+      "loss": 2.1764,
+      "step": 4616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002910813247065431,
+      "loss": 2.3791,
+      "step": 4617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002910060778827555,
+      "loss": 2.0519,
+      "step": 4618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002909308272411877,
+      "loss": 2.1582,
+      "step": 4619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029085557278884605,
+      "loss": 2.0502,
+      "step": 4620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002907803145327367,
+      "loss": 2.0317,
+      "step": 4621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002907050524798665,
+      "loss": 2.1982,
+      "step": 4622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002906297866372425,
+      "loss": 2.0113,
+      "step": 4623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002905545170118722,
+      "loss": 2.2572,
+      "step": 4624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029047924361076346,
+      "loss": 2.1075,
+      "step": 4625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029040396644092436,
+      "loss": 2.0109,
+      "step": 4626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029032868550936343,
+      "loss": 2.0377,
+      "step": 4627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029025340082308954,
+      "loss": 2.1661,
+      "step": 4628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002901781123891119,
+      "loss": 2.1488,
+      "step": 4629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002901028202144401,
+      "loss": 2.0829,
+      "step": 4630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029002752430608406,
+      "loss": 2.2236,
+      "step": 4631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002899522246710539,
+      "loss": 2.1207,
+      "step": 4632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028987692131636044,
+      "loss": 2.2968,
+      "step": 4633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002898016142490145,
+      "loss": 2.2035,
+      "step": 4634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002897263034760275,
+      "loss": 2.0356,
+      "step": 4635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028965098900441083,
+      "loss": 2.0951,
+      "step": 4636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028957567084117675,
+      "loss": 2.0881,
+      "step": 4637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002895003489933375,
+      "loss": 2.0875,
+      "step": 4638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002894250234679057,
+      "loss": 1.9068,
+      "step": 4639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028934969427189437,
+      "loss": 2.0826,
+      "step": 4640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028927436141231697,
+      "loss": 2.3026,
+      "step": 4641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002891990248961871,
+      "loss": 2.1687,
+      "step": 4642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002891236847305188,
+      "loss": 2.1747,
+      "step": 4643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002890483409223264,
+      "loss": 2.1787,
+      "step": 4644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002889729934786248,
+      "loss": 2.0441,
+      "step": 4645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002888976424064289,
+      "loss": 2.1674,
+      "step": 4646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002888222877127542,
+      "loss": 1.9872,
+      "step": 4647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028874692940461635,
+      "loss": 2.2107,
+      "step": 4648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028867156748903146,
+      "loss": 2.0842,
+      "step": 4649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028859620197301585,
+      "loss": 2.0475,
+      "step": 4650
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 2.1782724857330322,
+      "eval_runtime": 183.9706,
+      "eval_samples_per_second": 457.704,
+      "eval_steps_per_second": 28.608,
+      "step": 4650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028852083286358643,
+      "loss": 2.1412,
+      "step": 4651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028844546016776015,
+      "loss": 2.2327,
+      "step": 4652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028837008389255445,
+      "loss": 1.9762,
+      "step": 4653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000288294704044987,
+      "loss": 2.1509,
+      "step": 4654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000288219320632076,
+      "loss": 2.2661,
+      "step": 4655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028814393366083976,
+      "loss": 1.9452,
+      "step": 4656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002880685431382971,
+      "loss": 2.1574,
+      "step": 4657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002879931490714669,
+      "loss": 2.1855,
+      "step": 4658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028791775146736886,
+      "loss": 1.9036,
+      "step": 4659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002878423503330224,
+      "loss": 2.1108,
+      "step": 4660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002877669456754478,
+      "loss": 2.1375,
+      "step": 4661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002876915375016654,
+      "loss": 2.1905,
+      "step": 4662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028761612581869577,
+      "loss": 2.2093,
+      "step": 4663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028754071063356006,
+      "loss": 2.1076,
+      "step": 4664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002874652919532797,
+      "loss": 2.2387,
+      "step": 4665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028738986978487627,
+      "loss": 2.1313,
+      "step": 4666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002873144441353718,
+      "loss": 1.9915,
+      "step": 4667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002872390150117887,
+      "loss": 2.3458,
+      "step": 4668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002871635824211495,
+      "loss": 2.1986,
+      "step": 4669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028708814637047733,
+      "loss": 2.2729,
+      "step": 4670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028701270686679536,
+      "loss": 2.1992,
+      "step": 4671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002869372639171273,
+      "loss": 2.071,
+      "step": 4672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002868618175284972,
+      "loss": 2.0098,
+      "step": 4673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000286786367707929,
+      "loss": 2.1405,
+      "step": 4674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028671091446244763,
+      "loss": 2.1741,
+      "step": 4675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002866354577990779,
+      "loss": 2.1378,
+      "step": 4676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028655999772484494,
+      "loss": 2.2547,
+      "step": 4677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028648453424677436,
+      "loss": 1.9902,
+      "step": 4678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028640906737189203,
+      "loss": 2.1993,
+      "step": 4679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002863335971072241,
+      "loss": 2.1711,
+      "step": 4680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028625812345979714,
+      "loss": 2.0888,
+      "step": 4681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028618264643663784,
+      "loss": 2.0711,
+      "step": 4682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002861071660447733,
+      "loss": 2.2259,
+      "step": 4683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002860316822912312,
+      "loss": 2.2864,
+      "step": 4684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028595619518303897,
+      "loss": 2.1173,
+      "step": 4685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028588070472722484,
+      "loss": 1.9633,
+      "step": 4686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002858052109308172,
+      "loss": 2.0766,
+      "step": 4687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028572971380084474,
+      "loss": 2.0402,
+      "step": 4688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028565421334433627,
+      "loss": 2.278,
+      "step": 4689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028557870956832135,
+      "loss": 2.2263,
+      "step": 4690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028550320247982934,
+      "loss": 2.1428,
+      "step": 4691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002854276920858904,
+      "loss": 2.1915,
+      "step": 4692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002853521783935346,
+      "loss": 2.1342,
+      "step": 4693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028527666140979255,
+      "loss": 2.0303,
+      "step": 4694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000285201141141695,
+      "loss": 2.3079,
+      "step": 4695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002851256175962732,
+      "loss": 2.1054,
+      "step": 4696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028505009078055854,
+      "loss": 2.0843,
+      "step": 4697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028497456070158283,
+      "loss": 2.2198,
+      "step": 4698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028489902736637807,
+      "loss": 2.0186,
+      "step": 4699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028482349078197666,
+      "loss": 2.1238,
+      "step": 4700
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 2.1722171306610107,
+      "eval_runtime": 184.067,
+      "eval_samples_per_second": 457.464,
+      "eval_steps_per_second": 28.593,
+      "step": 4700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002847479509554113,
+      "loss": 2.195,
+      "step": 4701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002846724078937149,
+      "loss": 2.0018,
+      "step": 4702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002845968616039207,
+      "loss": 2.1172,
+      "step": 4703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028452131209306237,
+      "loss": 2.0228,
+      "step": 4704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028444575936817373,
+      "loss": 2.1027,
+      "step": 4705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028437020343628896,
+      "loss": 2.1943,
+      "step": 4706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002842946443044425,
+      "loss": 2.2421,
+      "step": 4707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002842190819796691,
+      "loss": 2.1968,
+      "step": 4708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028414351646900395,
+      "loss": 2.01,
+      "step": 4709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002840679477794822,
+      "loss": 2.1547,
+      "step": 4710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002839923759181396,
+      "loss": 2.0484,
+      "step": 4711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002839168008920122,
+      "loss": 2.0707,
+      "step": 4712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002838412227081361,
+      "loss": 1.9991,
+      "step": 4713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002837656413735479,
+      "loss": 2.1208,
+      "step": 4714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028369005689528456,
+      "loss": 2.1865,
+      "step": 4715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028361446928038296,
+      "loss": 2.1464,
+      "step": 4716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028353887853588066,
+      "loss": 1.9102,
+      "step": 4717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028346328466881545,
+      "loss": 2.1803,
+      "step": 4718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002833876876862251,
+      "loss": 1.9832,
+      "step": 4719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028331208759514807,
+      "loss": 2.2303,
+      "step": 4720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002832364844026229,
+      "loss": 2.0447,
+      "step": 4721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028316087811568847,
+      "loss": 2.0336,
+      "step": 4722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028308526874138394,
+      "loss": 2.1139,
+      "step": 4723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028300965628674866,
+      "loss": 2.0595,
+      "step": 4724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002829340407588225,
+      "loss": 2.0618,
+      "step": 4725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002828584221646454,
+      "loss": 2.2222,
+      "step": 4726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028278280051125763,
+      "loss": 2.2655,
+      "step": 4727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002827071758056999,
+      "loss": 2.0885,
+      "step": 4728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000282631548055013,
+      "loss": 2.3053,
+      "step": 4729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000282555917266238,
+      "loss": 2.1673,
+      "step": 4730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028248028344641634,
+      "loss": 2.2575,
+      "step": 4731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028240464660259004,
+      "loss": 2.2933,
+      "step": 4732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028232900674180075,
+      "loss": 2.0073,
+      "step": 4733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028225336387109086,
+      "loss": 2.1017,
+      "step": 4734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000282177717997503,
+      "loss": 2.2413,
+      "step": 4735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028210206912807995,
+      "loss": 2.0143,
+      "step": 4736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002820264172698649,
+      "loss": 2.1978,
+      "step": 4737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002819507624299012,
+      "loss": 2.2317,
+      "step": 4738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028187510461523245,
+      "loss": 2.1607,
+      "step": 4739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002817994438329027,
+      "loss": 2.0641,
+      "step": 4740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028172378008995617,
+      "loss": 2.2206,
+      "step": 4741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028164811339343734,
+      "loss": 2.4148,
+      "step": 4742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028157244375039104,
+      "loss": 1.8266,
+      "step": 4743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002814967711678622,
+      "loss": 2.1285,
+      "step": 4744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028142109565289626,
+      "loss": 2.042,
+      "step": 4745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002813454172125389,
+      "loss": 2.0145,
+      "step": 4746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028126973585383577,
+      "loss": 2.2037,
+      "step": 4747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002811940515838331,
+      "loss": 2.1411,
+      "step": 4748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028111836440957745,
+      "loss": 2.0738,
+      "step": 4749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002810426743381153,
+      "loss": 2.2063,
+      "step": 4750
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 2.1654436588287354,
+      "eval_runtime": 184.0447,
+      "eval_samples_per_second": 457.519,
+      "eval_steps_per_second": 28.596,
+      "step": 4750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002809669813764938,
+      "loss": 2.1513,
+      "step": 4751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028089128553176,
+      "loss": 2.1992,
+      "step": 4752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028081558681096154,
+      "loss": 1.8864,
+      "step": 4753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000280739885221146,
+      "loss": 2.3583,
+      "step": 4754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002806641807693617,
+      "loss": 2.0342,
+      "step": 4755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002805884734626566,
+      "loss": 2.0436,
+      "step": 4756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002805127633080795,
+      "loss": 2.1006,
+      "step": 4757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028043705031267907,
+      "loss": 2.1187,
+      "step": 4758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028036133448350455,
+      "loss": 2.1543,
+      "step": 4759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002802856158276052,
+      "loss": 1.9707,
+      "step": 4760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028020989435203065,
+      "loss": 2.0801,
+      "step": 4761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002801341700638307,
+      "loss": 2.19,
+      "step": 4762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002800584429700558,
+      "loss": 1.9311,
+      "step": 4763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000279982713077756,
+      "loss": 2.2055,
+      "step": 4764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002799069803939821,
+      "loss": 2.2117,
+      "step": 4765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000279831244925785,
+      "loss": 2.0856,
+      "step": 4766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002797555066802159,
+      "loss": 2.2426,
+      "step": 4767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002796797656643263,
+      "loss": 2.1632,
+      "step": 4768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002796040218851678,
+      "loss": 2.0031,
+      "step": 4769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027952827534979247,
+      "loss": 2.1661,
+      "step": 4770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002794525260652524,
+      "loss": 2.1937,
+      "step": 4771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027937677403860016,
+      "loss": 2.2575,
+      "step": 4772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002793010192768884,
+      "loss": 2.2987,
+      "step": 4773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027922526178717015,
+      "loss": 2.1514,
+      "step": 4774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002791495015764986,
+      "loss": 2.0366,
+      "step": 4775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027907373865192724,
+      "loss": 2.0115,
+      "step": 4776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002789979730205099,
+      "loss": 2.3217,
+      "step": 4777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002789222046893004,
+      "loss": 2.1853,
+      "step": 4778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027884643366535316,
+      "loss": 2.3488,
+      "step": 4779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002787706599557226,
+      "loss": 2.071,
+      "step": 4780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002786948835674634,
+      "loss": 2.2302,
+      "step": 4781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027861910450763065,
+      "loss": 1.8996,
+      "step": 4782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027854332278327956,
+      "loss": 2.1566,
+      "step": 4783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002784675384014656,
+      "loss": 2.0322,
+      "step": 4784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027839175136924456,
+      "loss": 2.0812,
+      "step": 4785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002783159616936723,
+      "loss": 2.1195,
+      "step": 4786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027824016938180526,
+      "loss": 2.0198,
+      "step": 4787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002781643744406997,
+      "loss": 2.0878,
+      "step": 4788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027808857687741244,
+      "loss": 2.0551,
+      "step": 4789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002780127766990004,
+      "loss": 2.1398,
+      "step": 4790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027793697391252095,
+      "loss": 1.9451,
+      "step": 4791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002778611685250313,
+      "loss": 2.2272,
+      "step": 4792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002777853605435893,
+      "loss": 2.0508,
+      "step": 4793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002777095499752528,
+      "loss": 2.1946,
+      "step": 4794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027763373682708,
+      "loss": 1.849,
+      "step": 4795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027755792110612936,
+      "loss": 2.1447,
+      "step": 4796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002774821028194595,
+      "loss": 1.9945,
+      "step": 4797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027740628197412933,
+      "loss": 1.9905,
+      "step": 4798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027733045857719803,
+      "loss": 2.3903,
+      "step": 4799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027725463263572484,
+      "loss": 1.9169,
+      "step": 4800
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 2.159355401992798,
+      "eval_runtime": 184.0985,
+      "eval_samples_per_second": 457.386,
+      "eval_steps_per_second": 28.588,
+      "step": 4800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002771788041567694,
+      "loss": 2.0476,
+      "step": 4801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027710297314739165,
+      "loss": 2.2179,
+      "step": 4802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027702713961465166,
+      "loss": 2.3077,
+      "step": 4803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027695130356560955,
+      "loss": 2.2791,
+      "step": 4804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027687546500732615,
+      "loss": 2.317,
+      "step": 4805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000276799623946862,
+      "loss": 2.0764,
+      "step": 4806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002767237803912783,
+      "loss": 2.2201,
+      "step": 4807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002766479343476361,
+      "loss": 2.1887,
+      "step": 4808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000276572085822997,
+      "loss": 1.9175,
+      "step": 4809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027649623482442275,
+      "loss": 2.1538,
+      "step": 4810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002764203813589752,
+      "loss": 2.1025,
+      "step": 4811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002763445254337166,
+      "loss": 2.3532,
+      "step": 4812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027626866705570926,
+      "loss": 2.0202,
+      "step": 4813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002761928062320158,
+      "loss": 2.2391,
+      "step": 4814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002761169429696992,
+      "loss": 1.9825,
+      "step": 4815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002760410772758225,
+      "loss": 2.0168,
+      "step": 4816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002759652091574489,
+      "loss": 2.146,
+      "step": 4817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027588933862164193,
+      "loss": 2.0453,
+      "step": 4818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027581346567546555,
+      "loss": 2.1754,
+      "step": 4819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027573759032598365,
+      "loss": 2.3238,
+      "step": 4820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002756617125802603,
+      "loss": 2.3972,
+      "step": 4821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027558583244536005,
+      "loss": 2.2636,
+      "step": 4822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002755099499283475,
+      "loss": 2.1609,
+      "step": 4823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002754340650362876,
+      "loss": 1.8747,
+      "step": 4824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027535817777624545,
+      "loss": 2.1183,
+      "step": 4825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002752822881552862,
+      "loss": 2.2008,
+      "step": 4826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027520639618047564,
+      "loss": 2.0961,
+      "step": 4827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027513050185887927,
+      "loss": 1.9997,
+      "step": 4828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002750546051975632,
+      "loss": 2.2138,
+      "step": 4829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027497870620359365,
+      "loss": 2.1297,
+      "step": 4830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002749028048840369,
+      "loss": 2.2101,
+      "step": 4831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027482690124595974,
+      "loss": 2.2863,
+      "step": 4832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027475099529642886,
+      "loss": 2.1101,
+      "step": 4833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002746750870425114,
+      "loss": 2.1299,
+      "step": 4834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002745991764912746,
+      "loss": 1.9727,
+      "step": 4835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002745232636497859,
+      "loss": 2.2696,
+      "step": 4836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027444734852511316,
+      "loss": 2.0903,
+      "step": 4837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002743714311243241,
+      "loss": 2.0641,
+      "step": 4838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002742955114544869,
+      "loss": 2.0556,
+      "step": 4839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027421958952266996,
+      "loss": 2.0997,
+      "step": 4840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027414366533594174,
+      "loss": 2.1414,
+      "step": 4841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027406773890137104,
+      "loss": 2.1878,
+      "step": 4842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002739918102260268,
+      "loss": 2.1823,
+      "step": 4843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027391587931697824,
+      "loss": 2.1246,
+      "step": 4844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027383994618129467,
+      "loss": 2.2683,
+      "step": 4845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027376401082604564,
+      "loss": 2.1151,
+      "step": 4846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027368807325830113,
+      "loss": 2.1052,
+      "step": 4847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027361213348513094,
+      "loss": 1.9467,
+      "step": 4848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027353619151360543,
+      "loss": 2.302,
+      "step": 4849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027346024735079485,
+      "loss": 2.2551,
+      "step": 4850
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 2.155595064163208,
+      "eval_runtime": 184.0586,
+      "eval_samples_per_second": 457.485,
+      "eval_steps_per_second": 28.594,
+      "step": 4850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027338430100376987,
+      "loss": 2.0131,
+      "step": 4851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027330835247960143,
+      "loss": 1.9345,
+      "step": 4852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027323240178536045,
+      "loss": 2.094,
+      "step": 4853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002731564489281181,
+      "loss": 2.2092,
+      "step": 4854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027308049391494587,
+      "loss": 2.0296,
+      "step": 4855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002730045367529153,
+      "loss": 2.024,
+      "step": 4856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002729285774490984,
+      "loss": 2.1654,
+      "step": 4857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000272852616010567,
+      "loss": 2.0457,
+      "step": 4858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027277665244439334,
+      "loss": 1.9693,
+      "step": 4859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027270068675764995,
+      "loss": 2.0763,
+      "step": 4860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027262471895740946,
+      "loss": 2.118,
+      "step": 4861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027254874905074456,
+      "loss": 1.9897,
+      "step": 4862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002724727770447283,
+      "loss": 1.9228,
+      "step": 4863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000272396802946434,
+      "loss": 2.0322,
+      "step": 4864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002723208267629348,
+      "loss": 2.0237,
+      "step": 4865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002722448485013046,
+      "loss": 2.1249,
+      "step": 4866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027216886816861695,
+      "loss": 2.1369,
+      "step": 4867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002720928857719459,
+      "loss": 2.2852,
+      "step": 4868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002720169013183658,
+      "loss": 2.2116,
+      "step": 4869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027194091481495076,
+      "loss": 2.1471,
+      "step": 4870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002718649262687754,
+      "loss": 2.3867,
+      "step": 4871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002717889356869146,
+      "loss": 2.1323,
+      "step": 4872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027171294307644313,
+      "loss": 2.1135,
+      "step": 4873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002716369484444362,
+      "loss": 2.1648,
+      "step": 4874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027156095179796906,
+      "loss": 2.056,
+      "step": 4875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027148495314411724,
+      "loss": 2.0617,
+      "step": 4876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027140895248995654,
+      "loss": 2.1633,
+      "step": 4877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002713329498425626,
+      "loss": 2.1076,
+      "step": 4878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002712569452090117,
+      "loss": 1.943,
+      "step": 4879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027118093859637997,
+      "loss": 1.9559,
+      "step": 4880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027110493001174387,
+      "loss": 1.9688,
+      "step": 4881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027102891946217993,
+      "loss": 2.1929,
+      "step": 4882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027095290695476517,
+      "loss": 1.9237,
+      "step": 4883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002708768924965763,
+      "loss": 2.118,
+      "step": 4884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027080087609469063,
+      "loss": 2.1225,
+      "step": 4885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002707248577561854,
+      "loss": 2.208,
+      "step": 4886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002706488374881382,
+      "loss": 2.135,
+      "step": 4887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027057281529762673,
+      "loss": 2.2721,
+      "step": 4888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002704967911917289,
+      "loss": 2.0419,
+      "step": 4889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027042076517752264,
+      "loss": 2.1519,
+      "step": 4890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002703447372620863,
+      "loss": 2.1561,
+      "step": 4891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002702687074524982,
+      "loss": 2.0897,
+      "step": 4892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000270192675755837,
+      "loss": 2.2332,
+      "step": 4893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002701166421791815,
+      "loss": 2.1899,
+      "step": 4894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002700406067296105,
+      "loss": 2.1644,
+      "step": 4895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002699645694142032,
+      "loss": 2.1019,
+      "step": 4896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000269888530240039,
+      "loss": 2.2159,
+      "step": 4897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002698124892141971,
+      "loss": 2.2804,
+      "step": 4898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026973644634375733,
+      "loss": 2.0783,
+      "step": 4899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002696604016357994,
+      "loss": 2.2897,
+      "step": 4900
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 2.1474642753601074,
+      "eval_runtime": 184.0828,
+      "eval_samples_per_second": 457.425,
+      "eval_steps_per_second": 28.59,
+      "step": 4900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026958435509740337,
+      "loss": 2.1297,
+      "step": 4901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002695083067356493,
+      "loss": 2.2642,
+      "step": 4902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026943225655761755,
+      "loss": 1.9581,
+      "step": 4903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026935620457038857,
+      "loss": 2.154,
+      "step": 4904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002692801507810431,
+      "loss": 2.0881,
+      "step": 4905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026920409519666174,
+      "loss": 2.1785,
+      "step": 4906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026912803782432564,
+      "loss": 2.2633,
+      "step": 4907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026905197867111607,
+      "loss": 2.1807,
+      "step": 4908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002689759177441141,
+      "loss": 2.1738,
+      "step": 4909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002688998550504014,
+      "loss": 2.2991,
+      "step": 4910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002688237905970595,
+      "loss": 1.9422,
+      "step": 4911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002687477243911703,
+      "loss": 2.1641,
+      "step": 4912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002686716564398157,
+      "loss": 1.9594,
+      "step": 4913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000268595586750078,
+      "loss": 2.1149,
+      "step": 4914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026851951532903926,
+      "loss": 2.0732,
+      "step": 4915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002684434421837821,
+      "loss": 2.0922,
+      "step": 4916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026836736732138906,
+      "loss": 2.2226,
+      "step": 4917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000268291290748943,
+      "loss": 1.9142,
+      "step": 4918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002682152124735268,
+      "loss": 2.1511,
+      "step": 4919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002681391325022237,
+      "loss": 2.2945,
+      "step": 4920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002680630508421166,
+      "loss": 2.0636,
+      "step": 4921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002679869675002894,
+      "loss": 1.9712,
+      "step": 4922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002679108824838253,
+      "loss": 2.1516,
+      "step": 4923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026783479579980805,
+      "loss": 2.1382,
+      "step": 4924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002677587074553218,
+      "loss": 2.029,
+      "step": 4925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002676826174574504,
+      "loss": 2.1008,
+      "step": 4926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000267606525813278,
+      "loss": 1.9417,
+      "step": 4927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000267530432529889,
+      "loss": 2.0556,
+      "step": 4928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002674543376143679,
+      "loss": 2.1671,
+      "step": 4929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002673782410737995,
+      "loss": 2.2019,
+      "step": 4930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002673021429152683,
+      "loss": 2.0576,
+      "step": 4931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002672260431458594,
+      "loss": 2.1066,
+      "step": 4932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026714994177265796,
+      "loss": 2.2132,
+      "step": 4933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026707383880274915,
+      "loss": 2.2563,
+      "step": 4934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002669977342432184,
+      "loss": 2.0926,
+      "step": 4935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002669216281011512,
+      "loss": 2.0294,
+      "step": 4936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026684552038363336,
+      "loss": 2.2236,
+      "step": 4937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002667694110977506,
+      "loss": 2.1525,
+      "step": 4938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000266693300250589,
+      "loss": 2.2121,
+      "step": 4939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002666171878492346,
+      "loss": 2.0692,
+      "step": 4940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026654107390077376,
+      "loss": 2.1863,
+      "step": 4941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026646495841229284,
+      "loss": 1.9239,
+      "step": 4942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002663888413908784,
+      "loss": 2.1111,
+      "step": 4943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002663127228436173,
+      "loss": 2.1497,
+      "step": 4944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026623660277759616,
+      "loss": 2.0706,
+      "step": 4945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002661604811999021,
+      "loss": 1.9733,
+      "step": 4946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026608435811762224,
+      "loss": 2.372,
+      "step": 4947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002660082335378438,
+      "loss": 2.095,
+      "step": 4948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002659321074676542,
+      "loss": 1.9281,
+      "step": 4949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002658559799141411,
+      "loss": 2.0693,
+      "step": 4950
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 2.14253306388855,
+      "eval_runtime": 184.1222,
+      "eval_samples_per_second": 457.327,
+      "eval_steps_per_second": 28.584,
+      "step": 4950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002657798508843921,
+      "loss": 2.2241,
+      "step": 4951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026570372038549507,
+      "loss": 2.0765,
+      "step": 4952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002656275884245379,
+      "loss": 2.0911,
+      "step": 4953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026555145500860864,
+      "loss": 2.0947,
+      "step": 4954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002654753201447956,
+      "loss": 2.0461,
+      "step": 4955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026539918384018725,
+      "loss": 2.1955,
+      "step": 4956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002653230461018719,
+      "loss": 2.1002,
+      "step": 4957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002652469069369384,
+      "loss": 2.1894,
+      "step": 4958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026517076635247525,
+      "loss": 2.181,
+      "step": 4959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026509462435557154,
+      "loss": 2.1348,
+      "step": 4960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002650184809533163,
+      "loss": 2.1732,
+      "step": 4961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026494233615279863,
+      "loss": 1.952,
+      "step": 4962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026486618996110777,
+      "loss": 1.9651,
+      "step": 4963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026479004238533335,
+      "loss": 2.3682,
+      "step": 4964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026471389343256466,
+      "loss": 2.3893,
+      "step": 4965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002646377431098915,
+      "loss": 2.227,
+      "step": 4966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002645615914244038,
+      "loss": 2.0776,
+      "step": 4967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002644854383831914,
+      "loss": 2.0624,
+      "step": 4968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026440928399334426,
+      "loss": 2.2255,
+      "step": 4969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026433312826195266,
+      "loss": 2.1674,
+      "step": 4970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026425697119610687,
+      "loss": 2.2338,
+      "step": 4971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002641808128028974,
+      "loss": 2.1462,
+      "step": 4972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002641046530894147,
+      "loss": 2.2377,
+      "step": 4973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002640284920627495,
+      "loss": 2.1436,
+      "step": 4974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026395232972999274,
+      "loss": 2.0914,
+      "step": 4975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026387616609823507,
+      "loss": 1.9724,
+      "step": 4976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026380000117456773,
+      "loss": 1.9458,
+      "step": 4977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002637238349660819,
+      "loss": 2.1551,
+      "step": 4978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026364766747986873,
+      "loss": 1.983,
+      "step": 4979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026357149872301974,
+      "loss": 2.0358,
+      "step": 4980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026349532870262646,
+      "loss": 2.0024,
+      "step": 4981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002634191574257804,
+      "loss": 1.9455,
+      "step": 4982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002633429848995734,
+      "loss": 1.998,
+      "step": 4983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026326681113109747,
+      "loss": 2.0533,
+      "step": 4984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026319063612744435,
+      "loss": 2.18,
+      "step": 4985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002631144598957063,
+      "loss": 2.1261,
+      "step": 4986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002630382824429756,
+      "loss": 2.1375,
+      "step": 4987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026296210377634434,
+      "loss": 2.1374,
+      "step": 4988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002628859239029052,
+      "loss": 2.2597,
+      "step": 4989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026280974282975064,
+      "loss": 1.9691,
+      "step": 4990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002627335605639733,
+      "loss": 2.2432,
+      "step": 4991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026265737711266603,
+      "loss": 2.0861,
+      "step": 4992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026258119248292165,
+      "loss": 2.0756,
+      "step": 4993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026250500668183326,
+      "loss": 1.8889,
+      "step": 4994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002624288197164939,
+      "loss": 2.0624,
+      "step": 4995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002623526315939968,
+      "loss": 2.1408,
+      "step": 4996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026227644232143526,
+      "loss": 1.9521,
+      "step": 4997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026220025190590285,
+      "loss": 2.1542,
+      "step": 4998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002621240603544929,
+      "loss": 2.0711,
+      "step": 4999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002620478676742993,
+      "loss": 2.2112,
+      "step": 5000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 2.1386890411376953,
+      "eval_runtime": 184.0578,
+      "eval_samples_per_second": 457.487,
+      "eval_steps_per_second": 28.594,
+      "step": 5000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002619716738724155,
+      "loss": 2.1978,
+      "step": 5001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002618954789559356,
+      "loss": 1.9573,
+      "step": 5002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002618192829319535,
+      "loss": 1.9535,
+      "step": 5003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002617430858075632,
+      "loss": 2.0958,
+      "step": 5004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026166688758985884,
+      "loss": 2.1959,
+      "step": 5005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026159068828593483,
+      "loss": 2.1503,
+      "step": 5006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002615144879028854,
+      "loss": 1.9404,
+      "step": 5007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002614382864478051,
+      "loss": 1.9517,
+      "step": 5008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002613620839277885,
+      "loss": 2.2334,
+      "step": 5009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026128588034993015,
+      "loss": 1.9486,
+      "step": 5010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026120967572132496,
+      "loss": 1.9738,
+      "step": 5011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002611334700490678,
+      "loss": 2.1037,
+      "step": 5012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002610572633402535,
+      "loss": 2.0498,
+      "step": 5013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002609810556019772,
+      "loss": 2.096,
+      "step": 5014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026090484684133404,
+      "loss": 2.1197,
+      "step": 5015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002608286370654192,
+      "loss": 2.2297,
+      "step": 5016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026075242628132816,
+      "loss": 2.3447,
+      "step": 5017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002606762144961563,
+      "loss": 2.0944,
+      "step": 5018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002606000017169991,
+      "loss": 2.1557,
+      "step": 5019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026052378795095223,
+      "loss": 2.1558,
+      "step": 5020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002604475732051114,
+      "loss": 2.2473,
+      "step": 5021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002603713574865723,
+      "loss": 2.0045,
+      "step": 5022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026029514080243104,
+      "loss": 1.9105,
+      "step": 5023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002602189231597835,
+      "loss": 2.1511,
+      "step": 5024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026014270456572563,
+      "loss": 2.122,
+      "step": 5025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002600664850273538,
+      "loss": 2.0553,
+      "step": 5026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025999026455176416,
+      "loss": 2.0683,
+      "step": 5027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002599140431460531,
+      "loss": 2.2505,
+      "step": 5028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025983782081731696,
+      "loss": 2.2901,
+      "step": 5029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002597615975726523,
+      "loss": 2.1242,
+      "step": 5030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002596853734191558,
+      "loss": 2.2752,
+      "step": 5031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002596091483639239,
+      "loss": 2.2568,
+      "step": 5032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025953292241405365,
+      "loss": 2.2179,
+      "step": 5033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002594566955766418,
+      "loss": 2.0743,
+      "step": 5034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025938046785878516,
+      "loss": 2.1502,
+      "step": 5035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025930423926758087,
+      "loss": 2.0557,
+      "step": 5036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025922800981012593,
+      "loss": 1.9157,
+      "step": 5037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025915177949351764,
+      "loss": 2.0659,
+      "step": 5038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025907554832485313,
+      "loss": 2.1807,
+      "step": 5039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025899931631122986,
+      "loss": 2.1196,
+      "step": 5040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002589230834597451,
+      "loss": 2.1414,
+      "step": 5041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002588468497774965,
+      "loss": 2.2244,
+      "step": 5042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025877061527158153,
+      "loss": 2.0557,
+      "step": 5043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002586943799490978,
+      "loss": 2.2091,
+      "step": 5044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002586181438171431,
+      "loss": 2.0773,
+      "step": 5045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002585419068828152,
+      "loss": 2.063,
+      "step": 5046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025846566915321194,
+      "loss": 1.9168,
+      "step": 5047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002583894306354314,
+      "loss": 2.0582,
+      "step": 5048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002583131913365714,
+      "loss": 2.0242,
+      "step": 5049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002582369512637302,
+      "loss": 2.1032,
+      "step": 5050
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 2.133467674255371,
+      "eval_runtime": 184.0484,
+      "eval_samples_per_second": 457.51,
+      "eval_steps_per_second": 28.596,
+      "step": 5050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025816071042400587,
+      "loss": 1.9959,
+      "step": 5051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025808446882449664,
+      "loss": 2.2055,
+      "step": 5052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002580082264723009,
+      "loss": 2.2231,
+      "step": 5053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025793198337451694,
+      "loss": 2.0695,
+      "step": 5054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025785573953824323,
+      "loss": 2.1526,
+      "step": 5055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025777949497057833,
+      "loss": 2.1712,
+      "step": 5056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002577032496786207,
+      "loss": 2.1083,
+      "step": 5057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002576270036694691,
+      "loss": 2.1243,
+      "step": 5058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025755075695022223,
+      "loss": 2.1618,
+      "step": 5059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025747450952797883,
+      "loss": 2.1506,
+      "step": 5060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002573982614098378,
+      "loss": 2.2326,
+      "step": 5061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025732201260289804,
+      "loss": 2.0615,
+      "step": 5062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025724576311425844,
+      "loss": 2.0497,
+      "step": 5063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025716951295101816,
+      "loss": 2.2661,
+      "step": 5064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002570932621202763,
+      "loss": 2.1207,
+      "step": 5065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002570170106291319,
+      "loss": 1.8963,
+      "step": 5066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025694075848468434,
+      "loss": 2.0789,
+      "step": 5067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002568645056940328,
+      "loss": 2.2199,
+      "step": 5068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025678825226427667,
+      "loss": 1.9371,
+      "step": 5069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025671199820251534,
+      "loss": 2.1551,
+      "step": 5070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025663574351584824,
+      "loss": 2.095,
+      "step": 5071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002565594882113749,
+      "loss": 2.2267,
+      "step": 5072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002564832322961951,
+      "loss": 1.9774,
+      "step": 5073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002564069757774082,
+      "loss": 2.1372,
+      "step": 5074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000256330718662114,
+      "loss": 2.0856,
+      "step": 5075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002562544609574123,
+      "loss": 2.1775,
+      "step": 5076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025617820267040284,
+      "loss": 2.1882,
+      "step": 5077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002561019438081855,
+      "loss": 2.093,
+      "step": 5078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025602568437786016,
+      "loss": 2.1439,
+      "step": 5079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025594942438652684,
+      "loss": 2.2493,
+      "step": 5080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002558731638412856,
+      "loss": 2.0327,
+      "step": 5081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002557969027492364,
+      "loss": 2.0478,
+      "step": 5082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002557206411174794,
+      "loss": 2.2147,
+      "step": 5083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025564437895311484,
+      "loss": 2.1349,
+      "step": 5084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025556811626324287,
+      "loss": 2.0535,
+      "step": 5085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002554918530549637,
+      "loss": 2.1239,
+      "step": 5086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002554155893353778,
+      "loss": 2.1061,
+      "step": 5087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002553393251115854,
+      "loss": 2.1565,
+      "step": 5088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025526306039068685,
+      "loss": 2.4307,
+      "step": 5089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002551867951797829,
+      "loss": 2.1034,
+      "step": 5090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025511052948597374,
+      "loss": 1.9855,
+      "step": 5091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002550342633163601,
+      "loss": 2.1623,
+      "step": 5092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025495799667804255,
+      "loss": 1.9139,
+      "step": 5093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002548817295781216,
+      "loss": 2.2966,
+      "step": 5094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002548054620236981,
+      "loss": 2.1056,
+      "step": 5095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002547291940218727,
+      "loss": 1.9984,
+      "step": 5096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025465292557974616,
+      "loss": 1.9828,
+      "step": 5097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025457665670441937,
+      "loss": 2.0773,
+      "step": 5098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000254500387402993,
+      "loss": 2.0911,
+      "step": 5099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025442411768256807,
+      "loss": 1.8994,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 2.125892162322998,
+      "eval_runtime": 184.0832,
+      "eval_samples_per_second": 457.424,
+      "eval_steps_per_second": 28.59,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025434784755024536,
+      "loss": 2.0571,
+      "step": 5101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002542715770131261,
+      "loss": 2.0985,
+      "step": 5102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000254195306078311,
+      "loss": 2.061,
+      "step": 5103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025411903475290126,
+      "loss": 2.16,
+      "step": 5104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002540427630439979,
+      "loss": 2.0534,
+      "step": 5105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000253966490958702,
+      "loss": 1.9472,
+      "step": 5106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002538902185041148,
+      "loss": 2.1292,
+      "step": 5107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002538139456873374,
+      "loss": 2.097,
+      "step": 5108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025373767251547097,
+      "loss": 2.0285,
+      "step": 5109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025366139899561695,
+      "loss": 2.1919,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002535851251348764,
+      "loss": 2.2286,
+      "step": 5111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025350885094035073,
+      "loss": 2.0525,
+      "step": 5112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025343257641914125,
+      "loss": 2.1196,
+      "step": 5113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025335630157834936,
+      "loss": 2.1401,
+      "step": 5114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025328002642507647,
+      "loss": 2.0041,
+      "step": 5115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025320375096642394,
+      "loss": 2.0374,
+      "step": 5116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002531274752094933,
+      "loss": 2.2792,
+      "step": 5117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000253051199161386,
+      "loss": 2.0644,
+      "step": 5118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025297492282920356,
+      "loss": 2.1184,
+      "step": 5119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002528986462200475,
+      "loss": 2.3704,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002528223693410195,
+      "loss": 1.9043,
+      "step": 5121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002527460921992209,
+      "loss": 1.9339,
+      "step": 5122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025266981480175354,
+      "loss": 2.1901,
+      "step": 5123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025259353715571895,
+      "loss": 2.2968,
+      "step": 5124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002525172592682189,
+      "loss": 2.2635,
+      "step": 5125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000252440981146355,
+      "loss": 2.1752,
+      "step": 5126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025236470279722887,
+      "loss": 2.2477,
+      "step": 5127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025228842422794236,
+      "loss": 2.1933,
+      "step": 5128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002522121454455972,
+      "loss": 2.065,
+      "step": 5129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025213586645729514,
+      "loss": 2.1211,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002520595872701379,
+      "loss": 2.2663,
+      "step": 5131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025198330789122744,
+      "loss": 2.1881,
+      "step": 5132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002519070283276654,
+      "loss": 2.327,
+      "step": 5133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002518307485865538,
+      "loss": 2.122,
+      "step": 5134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002517544686749944,
+      "loss": 2.2712,
+      "step": 5135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002516781886000891,
+      "loss": 1.9474,
+      "step": 5136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025160190836893964,
+      "loss": 2.1005,
+      "step": 5137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002515256279886481,
+      "loss": 2.2329,
+      "step": 5138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025144934746631635,
+      "loss": 2.1828,
+      "step": 5139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002513730668090464,
+      "loss": 2.0308,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025129678602394003,
+      "loss": 2.1022,
+      "step": 5141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025122050511809923,
+      "loss": 2.0313,
+      "step": 5142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025114422409862607,
+      "loss": 1.8848,
+      "step": 5143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002510679429726224,
+      "loss": 2.2453,
+      "step": 5144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002509916617471903,
+      "loss": 2.244,
+      "step": 5145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002509153804294318,
+      "loss": 2.037,
+      "step": 5146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025083909902644876,
+      "loss": 2.0836,
+      "step": 5147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002507628175453432,
+      "loss": 2.0911,
+      "step": 5148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025068653599321733,
+      "loss": 1.9861,
+      "step": 5149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000250610254377173,
+      "loss": 2.0233,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 2.1222126483917236,
+      "eval_runtime": 184.1175,
+      "eval_samples_per_second": 457.338,
+      "eval_steps_per_second": 28.585,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002505339727043123,
+      "loss": 2.2767,
+      "step": 5151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002504576909817372,
+      "loss": 2.1558,
+      "step": 5152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025038140921654985,
+      "loss": 2.2123,
+      "step": 5153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002503051274158522,
+      "loss": 2.1474,
+      "step": 5154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025022884558674637,
+      "loss": 1.9972,
+      "step": 5155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002501525637363343,
+      "loss": 1.8612,
+      "step": 5156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002500762818717182,
+      "loss": 1.964,
+      "step": 5157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025,
+      "loss": 2.2661,
+      "step": 5158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002499237181282818,
+      "loss": 2.357,
+      "step": 5159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002498474362636657,
+      "loss": 2.1425,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024977115441325375,
+      "loss": 2.1469,
+      "step": 5161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002496948725841479,
+      "loss": 1.9941,
+      "step": 5162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002496185907834502,
+      "loss": 1.9886,
+      "step": 5163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002495423090182628,
+      "loss": 1.9613,
+      "step": 5164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002494660272956878,
+      "loss": 2.1259,
+      "step": 5165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000249389745622827,
+      "loss": 2.1255,
+      "step": 5166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024931346400678273,
+      "loss": 2.3103,
+      "step": 5167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002492371824546568,
+      "loss": 1.9878,
+      "step": 5168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002491609009735513,
+      "loss": 2.1764,
+      "step": 5169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024908461957056825,
+      "loss": 2.159,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024900833825280966,
+      "loss": 2.1432,
+      "step": 5171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024893205702737766,
+      "loss": 2.2513,
+      "step": 5172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000248855775901374,
+      "loss": 1.9503,
+      "step": 5173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002487794948819008,
+      "loss": 2.1254,
+      "step": 5174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024870321397606004,
+      "loss": 1.9825,
+      "step": 5175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024862693319095366,
+      "loss": 1.9866,
+      "step": 5176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002485506525336836,
+      "loss": 2.1315,
+      "step": 5177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024847437201135195,
+      "loss": 2.0799,
+      "step": 5178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002483980916310604,
+      "loss": 2.0095,
+      "step": 5179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000248321811399911,
+      "loss": 2.1926,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002482455313250056,
+      "loss": 2.1258,
+      "step": 5181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024816925141344623,
+      "loss": 2.1129,
+      "step": 5182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002480929716723346,
+      "loss": 1.9456,
+      "step": 5183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024801669210877263,
+      "loss": 2.0155,
+      "step": 5184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024794041272986216,
+      "loss": 2.0833,
+      "step": 5185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002478641335427049,
+      "loss": 2.2442,
+      "step": 5186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024778785455440284,
+      "loss": 2.0697,
+      "step": 5187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024771157577205765,
+      "loss": 2.1429,
+      "step": 5188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024763529720277125,
+      "loss": 2.1835,
+      "step": 5189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002475590188536451,
+      "loss": 2.2489,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024748274073178113,
+      "loss": 2.0706,
+      "step": 5191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024740646284428106,
+      "loss": 2.0184,
+      "step": 5192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002473301851982465,
+      "loss": 2.0904,
+      "step": 5193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024725390780077906,
+      "loss": 2.2061,
+      "step": 5194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002471776306589806,
+      "loss": 2.0783,
+      "step": 5195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002471013537799525,
+      "loss": 2.2637,
+      "step": 5196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024702507717079645,
+      "loss": 1.9949,
+      "step": 5197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000246948800838614,
+      "loss": 1.9694,
+      "step": 5198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002468725247905067,
+      "loss": 2.138,
+      "step": 5199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002467962490335761,
+      "loss": 2.1795,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 2.1163952350616455,
+      "eval_runtime": 184.0914,
+      "eval_samples_per_second": 457.403,
+      "eval_steps_per_second": 28.589,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024671997357492354,
+      "loss": 2.0368,
+      "step": 5201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002466436984216507,
+      "loss": 2.1031,
+      "step": 5202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024656742358085876,
+      "loss": 2.0257,
+      "step": 5203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002464911490596493,
+      "loss": 2.0827,
+      "step": 5204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024641487486512357,
+      "loss": 2.1391,
+      "step": 5205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024633860100438317,
+      "loss": 2.0912,
+      "step": 5206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024626232748452905,
+      "loss": 2.0365,
+      "step": 5207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002461860543126627,
+      "loss": 2.1786,
+      "step": 5208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002461097814958852,
+      "loss": 1.9591,
+      "step": 5209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000246033509041298,
+      "loss": 1.971,
+      "step": 5210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002459572369560021,
+      "loss": 1.9791,
+      "step": 5211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024588096524709885,
+      "loss": 2.0229,
+      "step": 5212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024580469392168906,
+      "loss": 2.0349,
+      "step": 5213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000245728422986874,
+      "loss": 2.2927,
+      "step": 5214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024565215244975466,
+      "loss": 2.1467,
+      "step": 5215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000245575882317432,
+      "loss": 1.9176,
+      "step": 5216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024549961259700707,
+      "loss": 2.1682,
+      "step": 5217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024542334329558075,
+      "loss": 2.1287,
+      "step": 5218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024534707442025385,
+      "loss": 2.3684,
+      "step": 5219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002452708059781273,
+      "loss": 2.0726,
+      "step": 5220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024519453797630186,
+      "loss": 2.0752,
+      "step": 5221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024511827042187836,
+      "loss": 1.9492,
+      "step": 5222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024504200332195757,
+      "loss": 2.1154,
+      "step": 5223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024496573668363996,
+      "loss": 2.0916,
+      "step": 5224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002448894705140263,
+      "loss": 2.142,
+      "step": 5225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024481320482021713,
+      "loss": 2.0792,
+      "step": 5226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002447369396093131,
+      "loss": 1.9907,
+      "step": 5227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024466067488841474,
+      "loss": 2.21,
+      "step": 5228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002445844106646223,
+      "loss": 2.0868,
+      "step": 5229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002445081469450363,
+      "loss": 2.0769,
+      "step": 5230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002444318837367572,
+      "loss": 2.1597,
+      "step": 5231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024435562104688517,
+      "loss": 2.1959,
+      "step": 5232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024427935888252057,
+      "loss": 2.1246,
+      "step": 5233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002442030972507636,
+      "loss": 2.0099,
+      "step": 5234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024412683615871445,
+      "loss": 2.0597,
+      "step": 5235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024405057561347317,
+      "loss": 1.9898,
+      "step": 5236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024397431562213985,
+      "loss": 2.061,
+      "step": 5237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002438980561918145,
+      "loss": 2.1035,
+      "step": 5238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024382179732959717,
+      "loss": 2.0392,
+      "step": 5239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024374553904258782,
+      "loss": 2.0154,
+      "step": 5240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024366928133788608,
+      "loss": 2.0842,
+      "step": 5241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024359302422259188,
+      "loss": 2.1891,
+      "step": 5242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024351676770380495,
+      "loss": 2.1528,
+      "step": 5243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024344051178862503,
+      "loss": 2.0684,
+      "step": 5244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024336425648415185,
+      "loss": 2.1454,
+      "step": 5245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024328800179748475,
+      "loss": 2.0178,
+      "step": 5246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002432117477357234,
+      "loss": 1.9875,
+      "step": 5247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002431354943059672,
+      "loss": 2.2401,
+      "step": 5248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002430592415153157,
+      "loss": 2.2612,
+      "step": 5249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024298298937086807,
+      "loss": 2.04,
+      "step": 5250
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 2.1114773750305176,
+      "eval_runtime": 184.1164,
+      "eval_samples_per_second": 457.341,
+      "eval_steps_per_second": 28.585,
+      "step": 5250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024290673787972375,
+      "loss": 2.0647,
+      "step": 5251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024283048704898188,
+      "loss": 2.088,
+      "step": 5252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024275423688574157,
+      "loss": 2.1838,
+      "step": 5253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000242677987397102,
+      "loss": 2.1345,
+      "step": 5254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024260173859016218,
+      "loss": 2.1243,
+      "step": 5255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024252549047202123,
+      "loss": 1.9929,
+      "step": 5256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024244924304977784,
+      "loss": 2.0648,
+      "step": 5257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024237299633053094,
+      "loss": 2.2383,
+      "step": 5258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024229675032137933,
+      "loss": 2.1815,
+      "step": 5259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024222050502942173,
+      "loss": 2.1257,
+      "step": 5260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024214426046175675,
+      "loss": 2.196,
+      "step": 5261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024206801662548315,
+      "loss": 2.1712,
+      "step": 5262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024199177352769916,
+      "loss": 2.0766,
+      "step": 5263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024191553117550337,
+      "loss": 1.9776,
+      "step": 5264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002418392895759942,
+      "loss": 2.0937,
+      "step": 5265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024176304873626983,
+      "loss": 1.9893,
+      "step": 5266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002416868086634286,
+      "loss": 2.0601,
+      "step": 5267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024161056936456872,
+      "loss": 2.1366,
+      "step": 5268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002415343308467881,
+      "loss": 2.1479,
+      "step": 5269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024145809311718487,
+      "loss": 2.0085,
+      "step": 5270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024138185618285694,
+      "loss": 2.2728,
+      "step": 5271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024130562005090223,
+      "loss": 2.2865,
+      "step": 5272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002412293847284186,
+      "loss": 1.9907,
+      "step": 5273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002411531502225036,
+      "loss": 2.0053,
+      "step": 5274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002410769165402549,
+      "loss": 2.1648,
+      "step": 5275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024100068368877018,
+      "loss": 2.0595,
+      "step": 5276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024092445167514685,
+      "loss": 2.222,
+      "step": 5277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024084822050648237,
+      "loss": 2.1441,
+      "step": 5278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024077199018987414,
+      "loss": 2.1165,
+      "step": 5279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024069576073241922,
+      "loss": 2.001,
+      "step": 5280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024061953214121488,
+      "loss": 2.2293,
+      "step": 5281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002405433044233583,
+      "loss": 2.1961,
+      "step": 5282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024046707758594636,
+      "loss": 2.2723,
+      "step": 5283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002403908516360761,
+      "loss": 2.2639,
+      "step": 5284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024031462658084433,
+      "loss": 2.1072,
+      "step": 5285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024023840242734772,
+      "loss": 2.0088,
+      "step": 5286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024016217918268308,
+      "loss": 2.1498,
+      "step": 5287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024008595685394692,
+      "loss": 2.0828,
+      "step": 5288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024000973544823583,
+      "loss": 2.0961,
+      "step": 5289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002399335149726463,
+      "loss": 1.9807,
+      "step": 5290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023985729543427443,
+      "loss": 1.9969,
+      "step": 5291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023978107684021662,
+      "loss": 2.2173,
+      "step": 5292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000239704859197569,
+      "loss": 2.1547,
+      "step": 5293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002396286425134277,
+      "loss": 2.2398,
+      "step": 5294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002395524267948887,
+      "loss": 2.0436,
+      "step": 5295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023947621204904789,
+      "loss": 2.3301,
+      "step": 5296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023939999828300097,
+      "loss": 2.0947,
+      "step": 5297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023932378550384374,
+      "loss": 2.0069,
+      "step": 5298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023924757371867188,
+      "loss": 2.3887,
+      "step": 5299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002391713629345808,
+      "loss": 2.0926,
+      "step": 5300
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 2.105954647064209,
+      "eval_runtime": 184.0552,
+      "eval_samples_per_second": 457.493,
+      "eval_steps_per_second": 28.595,
+      "step": 5300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023909515315866605,
+      "loss": 2.0078,
+      "step": 5301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002390189443980229,
+      "loss": 2.0738,
+      "step": 5302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023894273665974657,
+      "loss": 2.1924,
+      "step": 5303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023886652995093224,
+      "loss": 2.1377,
+      "step": 5304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000238790324278675,
+      "loss": 2.1601,
+      "step": 5305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002387141196500698,
+      "loss": 2.1051,
+      "step": 5306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023863791607221162,
+      "loss": 2.1489,
+      "step": 5307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023856171355219493,
+      "loss": 2.1527,
+      "step": 5308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023848551209711464,
+      "loss": 2.071,
+      "step": 5309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002384093117140652,
+      "loss": 2.0739,
+      "step": 5310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023833311241014117,
+      "loss": 2.0093,
+      "step": 5311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023825691419243695,
+      "loss": 2.1227,
+      "step": 5312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023818071706804662,
+      "loss": 1.9906,
+      "step": 5313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023810452104406444,
+      "loss": 1.9444,
+      "step": 5314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023802832612758454,
+      "loss": 2.0829,
+      "step": 5315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023795213232570079,
+      "loss": 2.0345,
+      "step": 5316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023787593964550707,
+      "loss": 1.8095,
+      "step": 5317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023779974809409724,
+      "loss": 1.9775,
+      "step": 5318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023772355767856475,
+      "loss": 1.8898,
+      "step": 5319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023764736840600322,
+      "loss": 2.0497,
+      "step": 5320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002375711802835061,
+      "loss": 1.9513,
+      "step": 5321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023749499331816675,
+      "loss": 2.0021,
+      "step": 5322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002374188075170783,
+      "loss": 2.022,
+      "step": 5323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023734262288733408,
+      "loss": 2.0391,
+      "step": 5324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023726643943602677,
+      "loss": 2.1463,
+      "step": 5325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023719025717024945,
+      "loss": 2.1702,
+      "step": 5326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023711407609709485,
+      "loss": 2.0659,
+      "step": 5327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023703789622365565,
+      "loss": 1.8287,
+      "step": 5328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023696171755702454,
+      "loss": 1.9941,
+      "step": 5329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023688554010429372,
+      "loss": 2.0249,
+      "step": 5330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023680936387255564,
+      "loss": 2.0534,
+      "step": 5331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002367331888689026,
+      "loss": 1.9679,
+      "step": 5332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023665701510042656,
+      "loss": 2.0496,
+      "step": 5333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023658084257421958,
+      "loss": 1.9387,
+      "step": 5334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023650467129737366,
+      "loss": 2.2318,
+      "step": 5335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023642850127698033,
+      "loss": 2.1382,
+      "step": 5336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002363523325201313,
+      "loss": 1.9981,
+      "step": 5337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023627616503391814,
+      "loss": 1.951,
+      "step": 5338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023619999882543225,
+      "loss": 1.9622,
+      "step": 5339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000236123833901765,
+      "loss": 2.1512,
+      "step": 5340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002360476702700074,
+      "loss": 2.2041,
+      "step": 5341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002359715079372505,
+      "loss": 1.9555,
+      "step": 5342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023589534691058535,
+      "loss": 2.1201,
+      "step": 5343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023581918719710264,
+      "loss": 1.9961,
+      "step": 5344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023574302880389317,
+      "loss": 2.3698,
+      "step": 5345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023566687173804743,
+      "loss": 2.2085,
+      "step": 5346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002355907160066558,
+      "loss": 1.971,
+      "step": 5347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023551456161680865,
+      "loss": 1.9916,
+      "step": 5348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002354384085755962,
+      "loss": 2.2165,
+      "step": 5349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023536225689010842,
+      "loss": 1.9639,
+      "step": 5350
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 2.1000213623046875,
+      "eval_runtime": 184.0786,
+      "eval_samples_per_second": 457.435,
+      "eval_steps_per_second": 28.591,
+      "step": 5350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023528610656743532,
+      "loss": 2.0611,
+      "step": 5351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002352099576146668,
+      "loss": 2.1385,
+      "step": 5352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023513381003889227,
+      "loss": 2.2335,
+      "step": 5353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023505766384720146,
+      "loss": 1.9234,
+      "step": 5354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023498151904668375,
+      "loss": 2.2114,
+      "step": 5355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023490537564442845,
+      "loss": 2.0785,
+      "step": 5356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023482923364752487,
+      "loss": 1.9497,
+      "step": 5357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023475309306306174,
+      "loss": 1.9508,
+      "step": 5358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023467695389812812,
+      "loss": 2.0934,
+      "step": 5359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023460081615981278,
+      "loss": 2.037,
+      "step": 5360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023452467985520437,
+      "loss": 1.9608,
+      "step": 5361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002344485449913914,
+      "loss": 2.1047,
+      "step": 5362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023437241157546224,
+      "loss": 2.0313,
+      "step": 5363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000234296279614505,
+      "loss": 1.997,
+      "step": 5364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023422014911560793,
+      "loss": 2.214,
+      "step": 5365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023414402008585888,
+      "loss": 2.0163,
+      "step": 5366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023406789253234577,
+      "loss": 2.1257,
+      "step": 5367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002339917664621563,
+      "loss": 1.9012,
+      "step": 5368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023391564188237786,
+      "loss": 1.9696,
+      "step": 5369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023383951880009797,
+      "loss": 2.1675,
+      "step": 5370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002337633972224039,
+      "loss": 2.1811,
+      "step": 5371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023368727715638275,
+      "loss": 2.0409,
+      "step": 5372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023361115860912155,
+      "loss": 2.1043,
+      "step": 5373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023353504158770723,
+      "loss": 2.1789,
+      "step": 5374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002334589260992263,
+      "loss": 2.2049,
+      "step": 5375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023338281215076542,
+      "loss": 2.0528,
+      "step": 5376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023330669974941102,
+      "loss": 2.0804,
+      "step": 5377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023323058890224938,
+      "loss": 1.9745,
+      "step": 5378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023315447961636665,
+      "loss": 2.0696,
+      "step": 5379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023307837189884883,
+      "loss": 2.1275,
+      "step": 5380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023300226575678163,
+      "loss": 2.1447,
+      "step": 5381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002329261611972509,
+      "loss": 1.9882,
+      "step": 5382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023285005822734208,
+      "loss": 2.0657,
+      "step": 5383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002327739568541406,
+      "loss": 2.0056,
+      "step": 5384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023269785708473183,
+      "loss": 2.1684,
+      "step": 5385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023262175892620063,
+      "loss": 2.0931,
+      "step": 5386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002325456623856321,
+      "loss": 2.2837,
+      "step": 5387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000232469567470111,
+      "loss": 2.0761,
+      "step": 5388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023239347418672202,
+      "loss": 2.1729,
+      "step": 5389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023231738254254962,
+      "loss": 1.9599,
+      "step": 5390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023224129254467826,
+      "loss": 2.1418,
+      "step": 5391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023216520420019194,
+      "loss": 1.9711,
+      "step": 5392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002320891175161748,
+      "loss": 2.0189,
+      "step": 5393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023201303249971068,
+      "loss": 2.1352,
+      "step": 5394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002319369491578834,
+      "loss": 2.2373,
+      "step": 5395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023186086749777646,
+      "loss": 2.1403,
+      "step": 5396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023178478752647322,
+      "loss": 2.0737,
+      "step": 5397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023170870925105704,
+      "loss": 1.8666,
+      "step": 5398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023163263267861098,
+      "loss": 2.2391,
+      "step": 5399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023155655781621792,
+      "loss": 2.3427,
+      "step": 5400
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 2.0952579975128174,
+      "eval_runtime": 184.0043,
+      "eval_samples_per_second": 457.62,
+      "eval_steps_per_second": 28.603,
+      "step": 5400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023148048467096075,
+      "loss": 2.0935,
+      "step": 5401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023140441324992212,
+      "loss": 1.9618,
+      "step": 5402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002313283435601843,
+      "loss": 1.9728,
+      "step": 5403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023125227560882972,
+      "loss": 2.3521,
+      "step": 5404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023117620940294047,
+      "loss": 2.1753,
+      "step": 5405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002311001449495986,
+      "loss": 2.0124,
+      "step": 5406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023102408225588586,
+      "loss": 1.9097,
+      "step": 5407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023094802132888405,
+      "loss": 2.1084,
+      "step": 5408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023087196217567435,
+      "loss": 1.7766,
+      "step": 5409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002307959048033383,
+      "loss": 2.0215,
+      "step": 5410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000230719849218957,
+      "loss": 2.0262,
+      "step": 5411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023064379542961147,
+      "loss": 1.9383,
+      "step": 5412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023056774344238252,
+      "loss": 1.8755,
+      "step": 5413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023049169326435075,
+      "loss": 2.0775,
+      "step": 5414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002304156449025967,
+      "loss": 2.1956,
+      "step": 5415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002303395983642006,
+      "loss": 2.1189,
+      "step": 5416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023026355365624268,
+      "loss": 2.245,
+      "step": 5417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023018751078580287,
+      "loss": 1.9723,
+      "step": 5418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023011146975996112,
+      "loss": 1.9755,
+      "step": 5419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002300354305857968,
+      "loss": 2.236,
+      "step": 5420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022995939327038952,
+      "loss": 2.0352,
+      "step": 5421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022988335782081853,
+      "loss": 2.0192,
+      "step": 5422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000229807324244163,
+      "loss": 1.9054,
+      "step": 5423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022973129254750188,
+      "loss": 2.1871,
+      "step": 5424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022965526273791378,
+      "loss": 2.0864,
+      "step": 5425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022957923482247743,
+      "loss": 2.1193,
+      "step": 5426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022950320880827117,
+      "loss": 2.1576,
+      "step": 5427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022942718470237328,
+      "loss": 2.1358,
+      "step": 5428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022935116251186183,
+      "loss": 2.0025,
+      "step": 5429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022927514224381468,
+      "loss": 2.1179,
+      "step": 5430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022919912390530946,
+      "loss": 1.9728,
+      "step": 5431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022912310750342377,
+      "loss": 2.2234,
+      "step": 5432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022904709304523492,
+      "loss": 2.0255,
+      "step": 5433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022897108053782,
+      "loss": 1.9024,
+      "step": 5434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022889506998825625,
+      "loss": 2.144,
+      "step": 5435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002288190614036201,
+      "loss": 2.1687,
+      "step": 5436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022874305479098834,
+      "loss": 2.0719,
+      "step": 5437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002286670501574374,
+      "loss": 2.0655,
+      "step": 5438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002285910475100435,
+      "loss": 1.9454,
+      "step": 5439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022851504685588274,
+      "loss": 2.1551,
+      "step": 5440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022843904820203103,
+      "loss": 2.1198,
+      "step": 5441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002283630515555639,
+      "loss": 2.065,
+      "step": 5442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022828705692355694,
+      "loss": 2.2505,
+      "step": 5443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022821106431308543,
+      "loss": 2.0189,
+      "step": 5444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002281350737312246,
+      "loss": 2.0908,
+      "step": 5445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022805908518504927,
+      "loss": 2.0458,
+      "step": 5446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022798309868163428,
+      "loss": 2.2256,
+      "step": 5447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002279071142280541,
+      "loss": 1.8701,
+      "step": 5448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002278311318313831,
+      "loss": 2.1488,
+      "step": 5449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022775515149869543,
+      "loss": 1.9793,
+      "step": 5450
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 2.0905227661132812,
+      "eval_runtime": 184.1013,
+      "eval_samples_per_second": 457.379,
+      "eval_steps_per_second": 28.588,
+      "step": 5450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022767917323706516,
+      "loss": 2.2639,
+      "step": 5451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022760319705356614,
+      "loss": 2.0472,
+      "step": 5452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022752722295527172,
+      "loss": 2.1284,
+      "step": 5453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022745125094925548,
+      "loss": 1.9439,
+      "step": 5454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022737528104259055,
+      "loss": 2.1951,
+      "step": 5455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022729931324234998,
+      "loss": 1.9309,
+      "step": 5456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022722334755560658,
+      "loss": 2.0636,
+      "step": 5457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002271473839894331,
+      "loss": 2.1036,
+      "step": 5458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022707142255090168,
+      "loss": 2.1285,
+      "step": 5459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002269954632470847,
+      "loss": 2.0027,
+      "step": 5460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002269195060850542,
+      "loss": 2.3531,
+      "step": 5461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022684355107188198,
+      "loss": 1.9091,
+      "step": 5462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022676759821463967,
+      "loss": 2.1528,
+      "step": 5463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002266916475203986,
+      "loss": 1.9293,
+      "step": 5464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022661569899623017,
+      "loss": 2.0851,
+      "step": 5465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022653975264920522,
+      "loss": 2.0909,
+      "step": 5466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022646380848639463,
+      "loss": 2.0422,
+      "step": 5467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022638786651486907,
+      "loss": 2.0856,
+      "step": 5468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000226311926741699,
+      "loss": 1.7349,
+      "step": 5469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002262359891739544,
+      "loss": 2.1192,
+      "step": 5470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022616005381870542,
+      "loss": 1.8633,
+      "step": 5471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002260841206830218,
+      "loss": 1.9859,
+      "step": 5472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022600818977397318,
+      "loss": 1.9004,
+      "step": 5473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022593226109862894,
+      "loss": 2.3596,
+      "step": 5474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022585633466405832,
+      "loss": 2.0682,
+      "step": 5475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022578041047733007,
+      "loss": 1.8542,
+      "step": 5476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022570448854551312,
+      "loss": 2.0406,
+      "step": 5477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002256285688756759,
+      "loss": 2.0594,
+      "step": 5478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002255526514748869,
+      "loss": 2.048,
+      "step": 5479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022547673635021413,
+      "loss": 2.2015,
+      "step": 5480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022540082350872543,
+      "loss": 2.094,
+      "step": 5481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022532491295748866,
+      "loss": 2.1065,
+      "step": 5482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022524900470357118,
+      "loss": 2.0289,
+      "step": 5483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002251730987540403,
+      "loss": 2.1607,
+      "step": 5484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022509719511596305,
+      "loss": 2.0196,
+      "step": 5485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022502129379640642,
+      "loss": 2.2413,
+      "step": 5486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022494539480243683,
+      "loss": 2.0309,
+      "step": 5487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022486949814112076,
+      "loss": 1.9238,
+      "step": 5488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002247936038195244,
+      "loss": 1.9869,
+      "step": 5489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022471771184471375,
+      "loss": 2.0123,
+      "step": 5490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022464182222375464,
+      "loss": 1.9565,
+      "step": 5491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002245659349637124,
+      "loss": 2.0961,
+      "step": 5492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022449005007165249,
+      "loss": 2.0091,
+      "step": 5493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022441416755463996,
+      "loss": 2.064,
+      "step": 5494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022433828741973973,
+      "loss": 2.043,
+      "step": 5495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002242624096740164,
+      "loss": 1.8586,
+      "step": 5496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022418653432453446,
+      "loss": 1.9978,
+      "step": 5497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022411066137835805,
+      "loss": 2.1634,
+      "step": 5498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022403479084255119,
+      "loss": 2.0015,
+      "step": 5499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022395892272417757,
+      "loss": 1.927,
+      "step": 5500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 2.087319850921631,
+      "eval_runtime": 184.1068,
+      "eval_samples_per_second": 457.365,
+      "eval_steps_per_second": 28.587,
+      "step": 5500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022388305703030078,
+      "loss": 2.0574,
+      "step": 5501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022380719376798413,
+      "loss": 2.0606,
+      "step": 5502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022373133294429083,
+      "loss": 2.1219,
+      "step": 5503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002236554745662835,
+      "loss": 2.0751,
+      "step": 5504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022357961864102484,
+      "loss": 2.0041,
+      "step": 5505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022350376517557728,
+      "loss": 2.091,
+      "step": 5506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022342791417700298,
+      "loss": 2.1961,
+      "step": 5507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000223352065652364,
+      "loss": 2.2703,
+      "step": 5508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022327621960872184,
+      "loss": 1.9166,
+      "step": 5509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022320037605313807,
+      "loss": 2.1033,
+      "step": 5510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022312453499267391,
+      "loss": 2.0154,
+      "step": 5511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022304869643439046,
+      "loss": 2.1375,
+      "step": 5512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002229728603853484,
+      "loss": 2.0484,
+      "step": 5513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022289702685260838,
+      "loss": 2.138,
+      "step": 5514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022282119584323066,
+      "loss": 2.2028,
+      "step": 5515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022274536736427525,
+      "loss": 2.0418,
+      "step": 5516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022266954142280204,
+      "loss": 2.0006,
+      "step": 5517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022259371802587066,
+      "loss": 1.9379,
+      "step": 5518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002225178971805406,
+      "loss": 2.01,
+      "step": 5519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022244207889387068,
+      "loss": 1.9586,
+      "step": 5520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022236626317292005,
+      "loss": 1.8566,
+      "step": 5521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022229045002474727,
+      "loss": 2.0413,
+      "step": 5522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022221463945641077,
+      "loss": 2.122,
+      "step": 5523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022213883147496872,
+      "loss": 1.8952,
+      "step": 5524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002220630260874792,
+      "loss": 2.1044,
+      "step": 5525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022198722330099964,
+      "loss": 2.0282,
+      "step": 5526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002219114231225876,
+      "loss": 2.0384,
+      "step": 5527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002218356255593003,
+      "loss": 2.0875,
+      "step": 5528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022175983061819483,
+      "loss": 2.1619,
+      "step": 5529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002216840383063277,
+      "loss": 2.0615,
+      "step": 5530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002216082486307555,
+      "loss": 2.1589,
+      "step": 5531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022153246159853447,
+      "loss": 2.0224,
+      "step": 5532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002214566772167205,
+      "loss": 1.9193,
+      "step": 5533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022138089549236936,
+      "loss": 1.8762,
+      "step": 5534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022130511643253658,
+      "loss": 2.2699,
+      "step": 5535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022122934004427752,
+      "loss": 2.2528,
+      "step": 5536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002211535663346469,
+      "loss": 1.9275,
+      "step": 5537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022107779531069962,
+      "loss": 2.0659,
+      "step": 5538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022100202697949016,
+      "loss": 1.856,
+      "step": 5539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022092626134807275,
+      "loss": 1.9523,
+      "step": 5540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002208504984235014,
+      "loss": 1.8486,
+      "step": 5541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022077473821282995,
+      "loss": 2.0024,
+      "step": 5542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022069898072311166,
+      "loss": 1.9294,
+      "step": 5543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022062322596139985,
+      "loss": 2.042,
+      "step": 5544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022054747393474765,
+      "loss": 1.7586,
+      "step": 5545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022047172465020754,
+      "loss": 1.9733,
+      "step": 5546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022039597811483223,
+      "loss": 2.0002,
+      "step": 5547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022032023433567378,
+      "loss": 2.0595,
+      "step": 5548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022024449331978414,
+      "loss": 1.9964,
+      "step": 5549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022016875507421504,
+      "loss": 1.8329,
+      "step": 5550
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 2.0810322761535645,
+      "eval_runtime": 184.0777,
+      "eval_samples_per_second": 457.437,
+      "eval_steps_per_second": 28.591,
+      "step": 5550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022009301960601796,
+      "loss": 2.1273,
+      "step": 5551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022001728692224405,
+      "loss": 2.0148,
+      "step": 5552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021994155702994433,
+      "loss": 2.0736,
+      "step": 5553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021986582993616926,
+      "loss": 2.1305,
+      "step": 5554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002197901056479694,
+      "loss": 2.0261,
+      "step": 5555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021971438417239483,
+      "loss": 2.0028,
+      "step": 5556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021963866551649548,
+      "loss": 2.243,
+      "step": 5557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021956294968732089,
+      "loss": 2.0458,
+      "step": 5558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002194872366919206,
+      "loss": 2.1243,
+      "step": 5559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021941152653734346,
+      "loss": 2.0935,
+      "step": 5560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021933581923063838,
+      "loss": 2.1854,
+      "step": 5561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021926011477885402,
+      "loss": 2.0481,
+      "step": 5562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021918441318903855,
+      "loss": 1.9976,
+      "step": 5563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021910871446824005,
+      "loss": 1.8243,
+      "step": 5564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002190330186235063,
+      "loss": 2.0316,
+      "step": 5565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021895732566188476,
+      "loss": 2.0496,
+      "step": 5566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002188816355904226,
+      "loss": 1.8806,
+      "step": 5567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002188059484161669,
+      "loss": 2.1642,
+      "step": 5568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021873026414616427,
+      "loss": 1.8316,
+      "step": 5569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021865458278746126,
+      "loss": 1.9688,
+      "step": 5570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021857890434710377,
+      "loss": 2.0207,
+      "step": 5571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021850322883213783,
+      "loss": 2.2217,
+      "step": 5572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021842755624960902,
+      "loss": 1.8657,
+      "step": 5573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021835188660656267,
+      "loss": 1.8495,
+      "step": 5574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021827621991004392,
+      "loss": 1.9063,
+      "step": 5575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021820055616709737,
+      "loss": 2.1496,
+      "step": 5576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002181248953847676,
+      "loss": 1.9937,
+      "step": 5577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021804923757009885,
+      "loss": 2.0103,
+      "step": 5578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002179735827301351,
+      "loss": 1.9808,
+      "step": 5579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021789793087192004,
+      "loss": 2.1196,
+      "step": 5580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021782228200249704,
+      "loss": 2.0454,
+      "step": 5581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021774663612890918,
+      "loss": 1.8666,
+      "step": 5582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021767099325819934,
+      "loss": 2.3651,
+      "step": 5583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021759535339741006,
+      "loss": 2.196,
+      "step": 5584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021751971655358361,
+      "loss": 2.0548,
+      "step": 5585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021744408273376205,
+      "loss": 2.2524,
+      "step": 5586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021736845194498717,
+      "loss": 2.0133,
+      "step": 5587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002172928241943002,
+      "loss": 1.9458,
+      "step": 5588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021721719948874238,
+      "loss": 2.131,
+      "step": 5589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021714157783535464,
+      "loss": 2.3602,
+      "step": 5590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021706595924117747,
+      "loss": 2.1127,
+      "step": 5591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021699034371325138,
+      "loss": 2.2617,
+      "step": 5592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021691473125861612,
+      "loss": 1.8119,
+      "step": 5593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021683912188431154,
+      "loss": 2.2006,
+      "step": 5594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002167635155973771,
+      "loss": 1.8239,
+      "step": 5595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021668791240485194,
+      "loss": 2.2151,
+      "step": 5596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002166123123137749,
+      "loss": 1.7982,
+      "step": 5597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002165367153311847,
+      "loss": 2.2052,
+      "step": 5598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021646112146411937,
+      "loss": 1.8454,
+      "step": 5599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021638553071961705,
+      "loss": 1.9679,
+      "step": 5600
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 2.0751194953918457,
+      "eval_runtime": 184.0799,
+      "eval_samples_per_second": 457.432,
+      "eval_steps_per_second": 28.591,
+      "step": 5600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021630994310471548,
+      "loss": 1.9455,
+      "step": 5601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021623435862645204,
+      "loss": 1.8609,
+      "step": 5602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021615877729186397,
+      "loss": 2.2037,
+      "step": 5603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021608319910798788,
+      "loss": 2.2156,
+      "step": 5604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021600762408186043,
+      "loss": 1.9239,
+      "step": 5605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021593205222051787,
+      "loss": 2.215,
+      "step": 5606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021585648353099614,
+      "loss": 2.0249,
+      "step": 5607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002157809180203309,
+      "loss": 2.3102,
+      "step": 5608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021570535569555757,
+      "loss": 2.1154,
+      "step": 5609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002156297965637111,
+      "loss": 2.1133,
+      "step": 5610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021555424063182628,
+      "loss": 2.0693,
+      "step": 5611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021547868790693764,
+      "loss": 2.0537,
+      "step": 5612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002154031383960793,
+      "loss": 2.0684,
+      "step": 5613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002153275921062851,
+      "loss": 2.083,
+      "step": 5614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002152520490445888,
+      "loss": 2.1774,
+      "step": 5615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021517650921802338,
+      "loss": 2.1626,
+      "step": 5616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021510097263362194,
+      "loss": 2.0699,
+      "step": 5617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002150254392984172,
+      "loss": 2.0706,
+      "step": 5618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021494990921944147,
+      "loss": 2.308,
+      "step": 5619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002148743824037269,
+      "loss": 2.0056,
+      "step": 5620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021479885885830507,
+      "loss": 2.0155,
+      "step": 5621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021472333859020755,
+      "loss": 1.9616,
+      "step": 5622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021464782160646542,
+      "loss": 2.0737,
+      "step": 5623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002145723079141096,
+      "loss": 1.9657,
+      "step": 5624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002144967975201706,
+      "loss": 2.0731,
+      "step": 5625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021442129043167875,
+      "loss": 2.1116,
+      "step": 5626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021434578665566377,
+      "loss": 1.9149,
+      "step": 5627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021427028619915535,
+      "loss": 1.9242,
+      "step": 5628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021419478906918284,
+      "loss": 1.7898,
+      "step": 5629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021411929527277514,
+      "loss": 1.9945,
+      "step": 5630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021404380481696106,
+      "loss": 2.0224,
+      "step": 5631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021396831770876892,
+      "loss": 2.0265,
+      "step": 5632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002138928339552267,
+      "loss": 1.8767,
+      "step": 5633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021381735356336222,
+      "loss": 2.2627,
+      "step": 5634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002137418765402029,
+      "loss": 2.0469,
+      "step": 5635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021366640289277586,
+      "loss": 2.1774,
+      "step": 5636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021359093262810804,
+      "loss": 2.0781,
+      "step": 5637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002135154657532257,
+      "loss": 2.1544,
+      "step": 5638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002134400022751551,
+      "loss": 1.933,
+      "step": 5639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021336454220092214,
+      "loss": 2.0168,
+      "step": 5640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021328908553755233,
+      "loss": 2.0451,
+      "step": 5641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021321363229207096,
+      "loss": 2.1086,
+      "step": 5642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021313818247150295,
+      "loss": 1.8754,
+      "step": 5643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021306273608287271,
+      "loss": 2.2681,
+      "step": 5644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002129872931332047,
+      "loss": 1.9349,
+      "step": 5645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021291185362952274,
+      "loss": 2.173,
+      "step": 5646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021283641757885053,
+      "loss": 2.1449,
+      "step": 5647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021276098498821145,
+      "loss": 2.0547,
+      "step": 5648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021268555586462825,
+      "loss": 2.0015,
+      "step": 5649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002126101302151238,
+      "loss": 2.1962,
+      "step": 5650
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 2.069695472717285,
+      "eval_runtime": 184.0298,
+      "eval_samples_per_second": 457.556,
+      "eval_steps_per_second": 28.599,
+      "step": 5650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002125347080467203,
+      "loss": 2.2566,
+      "step": 5651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002124592893664399,
+      "loss": 1.9102,
+      "step": 5652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021238387418130424,
+      "loss": 2.2884,
+      "step": 5653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021230846249833472,
+      "loss": 2.2112,
+      "step": 5654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021223305432455225,
+      "loss": 2.0333,
+      "step": 5655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002121576496669776,
+      "loss": 2.0919,
+      "step": 5656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021208224853263118,
+      "loss": 1.9324,
+      "step": 5657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021200685092853306,
+      "loss": 2.0797,
+      "step": 5658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021193145686170304,
+      "loss": 2.0918,
+      "step": 5659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021185606633916033,
+      "loss": 2.0477,
+      "step": 5660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021178067936792407,
+      "loss": 2.2034,
+      "step": 5661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021170529595501307,
+      "loss": 2.0828,
+      "step": 5662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021162991610744564,
+      "loss": 2.0278,
+      "step": 5663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002115545398322399,
+      "loss": 2.0587,
+      "step": 5664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021147916713641366,
+      "loss": 2.1034,
+      "step": 5665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021140379802698418,
+      "loss": 1.8618,
+      "step": 5666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002113284325109686,
+      "loss": 1.9368,
+      "step": 5667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021125307059538368,
+      "loss": 2.2186,
+      "step": 5668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021117771228724582,
+      "loss": 2.1669,
+      "step": 5669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021110235759357106,
+      "loss": 1.9839,
+      "step": 5670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021102700652137526,
+      "loss": 1.9407,
+      "step": 5671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021095165907767362,
+      "loss": 2.0474,
+      "step": 5672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002108763152694813,
+      "loss": 2.0069,
+      "step": 5673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021080097510381296,
+      "loss": 2.0196,
+      "step": 5674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002107256385876831,
+      "loss": 2.0725,
+      "step": 5675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021065030572810567,
+      "loss": 1.9105,
+      "step": 5676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021057497653209433,
+      "loss": 1.843,
+      "step": 5677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002104996510066625,
+      "loss": 2.1794,
+      "step": 5678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021042432915882326,
+      "loss": 2.0634,
+      "step": 5679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021034901099558915,
+      "loss": 2.0095,
+      "step": 5680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021027369652397256,
+      "loss": 2.3013,
+      "step": 5681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021019838575098555,
+      "loss": 2.0698,
+      "step": 5682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021012307868363962,
+      "loss": 2.0333,
+      "step": 5683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002100477753289461,
+      "loss": 2.0791,
+      "step": 5684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000209972475693916,
+      "loss": 2.0956,
+      "step": 5685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002098971797855599,
+      "loss": 2.1432,
+      "step": 5686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020982188761088823,
+      "loss": 1.9319,
+      "step": 5687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020974659917691056,
+      "loss": 1.9397,
+      "step": 5688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020967131449063664,
+      "loss": 2.1865,
+      "step": 5689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002095960335590757,
+      "loss": 2.0251,
+      "step": 5690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020952075638923655,
+      "loss": 2.1275,
+      "step": 5691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002094454829881278,
+      "loss": 2.2091,
+      "step": 5692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002093702133627575,
+      "loss": 1.8896,
+      "step": 5693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020929494752013354,
+      "loss": 2.1911,
+      "step": 5694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020921968546726334,
+      "loss": 2.0081,
+      "step": 5695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020914442721115396,
+      "loss": 1.8954,
+      "step": 5696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020906917275881223,
+      "loss": 2.0589,
+      "step": 5697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002089939221172446,
+      "loss": 1.8973,
+      "step": 5698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020891867529345697,
+      "loss": 2.1843,
+      "step": 5699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020884343229445507,
+      "loss": 2.0246,
+      "step": 5700
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 2.0648117065429688,
+      "eval_runtime": 184.0801,
+      "eval_samples_per_second": 457.431,
+      "eval_steps_per_second": 28.591,
+      "step": 5700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002087681931272443,
+      "loss": 1.9799,
+      "step": 5701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020869295779882957,
+      "loss": 1.9543,
+      "step": 5702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002086177263162155,
+      "loss": 2.1972,
+      "step": 5703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020854249868640652,
+      "loss": 2.0299,
+      "step": 5704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020846727491640626,
+      "loss": 2.0831,
+      "step": 5705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020839205501321843,
+      "loss": 1.9504,
+      "step": 5706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020831683898384613,
+      "loss": 2.0644,
+      "step": 5707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020824162683529224,
+      "loss": 2.301,
+      "step": 5708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020816641857455926,
+      "loss": 2.1395,
+      "step": 5709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020809121420864923,
+      "loss": 2.1414,
+      "step": 5710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020801601374456386,
+      "loss": 1.9627,
+      "step": 5711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020794081718930458,
+      "loss": 2.2068,
+      "step": 5712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020786562454987235,
+      "loss": 2.0705,
+      "step": 5713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002077904358332678,
+      "loss": 2.0251,
+      "step": 5714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020771525104649135,
+      "loss": 1.9509,
+      "step": 5715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020764007019654274,
+      "loss": 1.8474,
+      "step": 5716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020756489329042154,
+      "loss": 1.9262,
+      "step": 5717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020748972033512701,
+      "loss": 1.9002,
+      "step": 5718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002074145513376579,
+      "loss": 2.2227,
+      "step": 5719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002073393863050127,
+      "loss": 1.8629,
+      "step": 5720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020726422524418956,
+      "loss": 2.0531,
+      "step": 5721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020718906816218598,
+      "loss": 1.9738,
+      "step": 5722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020711391506599938,
+      "loss": 2.1197,
+      "step": 5723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020703876596262679,
+      "loss": 2.1073,
+      "step": 5724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002069636208590648,
+      "loss": 2.0571,
+      "step": 5725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002068884797623095,
+      "loss": 1.8387,
+      "step": 5726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020681334267935687,
+      "loss": 2.0085,
+      "step": 5727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020673820961720227,
+      "loss": 2.0102,
+      "step": 5728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002066630805828409,
+      "loss": 2.0583,
+      "step": 5729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020658795558326743,
+      "loss": 1.918,
+      "step": 5730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002065128346254762,
+      "loss": 2.0207,
+      "step": 5731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002064377177164613,
+      "loss": 2.0997,
+      "step": 5732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002063626048632161,
+      "loss": 1.9926,
+      "step": 5733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020628749607273396,
+      "loss": 1.9365,
+      "step": 5734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020621239135200768,
+      "loss": 1.8565,
+      "step": 5735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020613729070802973,
+      "loss": 1.7443,
+      "step": 5736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020606219414779218,
+      "loss": 2.0226,
+      "step": 5737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020598710167828686,
+      "loss": 2.095,
+      "step": 5738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020591201330650487,
+      "loss": 1.9433,
+      "step": 5739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020583692903943722,
+      "loss": 1.8595,
+      "step": 5740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020576184888407452,
+      "loss": 2.0504,
+      "step": 5741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020568677284740693,
+      "loss": 2.008,
+      "step": 5742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020561170093642424,
+      "loss": 1.9813,
+      "step": 5743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020553663315811576,
+      "loss": 2.066,
+      "step": 5744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020546156951947064,
+      "loss": 2.1707,
+      "step": 5745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020538651002747745,
+      "loss": 2.2582,
+      "step": 5746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020531145468912436,
+      "loss": 1.9754,
+      "step": 5747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002052364035113994,
+      "loss": 1.8878,
+      "step": 5748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020516135650129,
+      "loss": 1.9089,
+      "step": 5749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020508631366578312,
+      "loss": 1.9195,
+      "step": 5750
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 2.060515880584717,
+      "eval_runtime": 184.0565,
+      "eval_samples_per_second": 457.49,
+      "eval_steps_per_second": 28.594,
+      "step": 5750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020501127501186553,
+      "loss": 2.1302,
+      "step": 5751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020493624054652358,
+      "loss": 1.8855,
+      "step": 5752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020486121027674312,
+      "loss": 2.1235,
+      "step": 5753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020478618420950983,
+      "loss": 2.1919,
+      "step": 5754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020471116235180865,
+      "loss": 2.0308,
+      "step": 5755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020463614471062435,
+      "loss": 1.9074,
+      "step": 5756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002045611312929414,
+      "loss": 1.9192,
+      "step": 5757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020448612210574363,
+      "loss": 1.9579,
+      "step": 5758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020441111715601475,
+      "loss": 1.9831,
+      "step": 5759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020433611645073786,
+      "loss": 2.1017,
+      "step": 5760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020426111999689568,
+      "loss": 2.0511,
+      "step": 5761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020418612780147063,
+      "loss": 1.8898,
+      "step": 5762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020411113987144468,
+      "loss": 1.9162,
+      "step": 5763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002040361562137994,
+      "loss": 2.0251,
+      "step": 5764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020396117683551602,
+      "loss": 2.0033,
+      "step": 5765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020388620174357542,
+      "loss": 1.9988,
+      "step": 5766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002038112309449578,
+      "loss": 2.1102,
+      "step": 5767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020373626444664317,
+      "loss": 2.2875,
+      "step": 5768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020366130225561125,
+      "loss": 2.0482,
+      "step": 5769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020358634437884113,
+      "loss": 2.012,
+      "step": 5770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020351139082331175,
+      "loss": 2.0473,
+      "step": 5771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020343644159600126,
+      "loss": 1.9349,
+      "step": 5772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020336149670388773,
+      "loss": 1.8912,
+      "step": 5773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002032865561539488,
+      "loss": 2.0852,
+      "step": 5774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020321161995316166,
+      "loss": 2.0669,
+      "step": 5775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020313668810850293,
+      "loss": 1.999,
+      "step": 5776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002030617606269492,
+      "loss": 2.2383,
+      "step": 5777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020298683751547618,
+      "loss": 2.272,
+      "step": 5778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020291191878105962,
+      "loss": 1.9774,
+      "step": 5779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002028370044306745,
+      "loss": 2.1706,
+      "step": 5780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002027620944712957,
+      "loss": 2.158,
+      "step": 5781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020268718890989755,
+      "loss": 1.8961,
+      "step": 5782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002026122877534538,
+      "loss": 2.2877,
+      "step": 5783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000202537391008938,
+      "loss": 2.1851,
+      "step": 5784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020246249868332336,
+      "loss": 1.9548,
+      "step": 5785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020238761078358248,
+      "loss": 2.0398,
+      "step": 5786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020231272731668768,
+      "loss": 1.9621,
+      "step": 5787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002022378482896109,
+      "loss": 2.1897,
+      "step": 5788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020216297370932336,
+      "loss": 2.0601,
+      "step": 5789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020208810358279618,
+      "loss": 2.1792,
+      "step": 5790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020201323791700006,
+      "loss": 2.1051,
+      "step": 5791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020193837671890518,
+      "loss": 2.0539,
+      "step": 5792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020186351999548126,
+      "loss": 2.0663,
+      "step": 5793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020178866775369774,
+      "loss": 2.1124,
+      "step": 5794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020171382000052358,
+      "loss": 2.1804,
+      "step": 5795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002016389767429272,
+      "loss": 2.0366,
+      "step": 5796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002015641379878768,
+      "loss": 1.9754,
+      "step": 5797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020148930374234004,
+      "loss": 1.9454,
+      "step": 5798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020141447401328437,
+      "loss": 1.787,
+      "step": 5799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020133964880767634,
+      "loss": 2.0181,
+      "step": 5800
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 2.060652494430542,
+      "eval_runtime": 184.031,
+      "eval_samples_per_second": 457.553,
+      "eval_steps_per_second": 28.598,
+      "step": 5800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020126482813248258,
+      "loss": 1.9947,
+      "step": 5801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020119001199466908,
+      "loss": 1.8774,
+      "step": 5802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002011152004012014,
+      "loss": 1.9619,
+      "step": 5803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020104039335904476,
+      "loss": 1.939,
+      "step": 5804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020096559087516395,
+      "loss": 2.0136,
+      "step": 5805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020089079295652307,
+      "loss": 1.9818,
+      "step": 5806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020081599961008621,
+      "loss": 2.1116,
+      "step": 5807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020074121084281678,
+      "loss": 1.9446,
+      "step": 5808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020066642666167784,
+      "loss": 1.9406,
+      "step": 5809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020059164707363203,
+      "loss": 1.9761,
+      "step": 5810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020051687208564142,
+      "loss": 2.133,
+      "step": 5811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002004421017046679,
+      "loss": 2.1145,
+      "step": 5812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020036733593767274,
+      "loss": 1.9834,
+      "step": 5813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002002925747916168,
+      "loss": 1.7998,
+      "step": 5814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020021781827346064,
+      "loss": 2.0597,
+      "step": 5815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020014306639016435,
+      "loss": 1.9479,
+      "step": 5816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020006831914868735,
+      "loss": 1.9359,
+      "step": 5817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019999357655598894,
+      "loss": 1.9942,
+      "step": 5818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019991883861902778,
+      "loss": 1.9906,
+      "step": 5819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001998441053447623,
+      "loss": 2.0155,
+      "step": 5820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019976937674015028,
+      "loss": 2.0921,
+      "step": 5821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019969465281214932,
+      "loss": 2.0512,
+      "step": 5822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019961993356771622,
+      "loss": 2.2801,
+      "step": 5823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019954521901380763,
+      "loss": 1.9357,
+      "step": 5824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001994705091573797,
+      "loss": 2.0523,
+      "step": 5825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001993958040053881,
+      "loss": 1.8252,
+      "step": 5826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019932110356478815,
+      "loss": 2.1106,
+      "step": 5827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019924640784253452,
+      "loss": 2.1181,
+      "step": 5828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001991717168455818,
+      "loss": 2.103,
+      "step": 5829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019909703058088374,
+      "loss": 2.0739,
+      "step": 5830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019902234905539392,
+      "loss": 1.9416,
+      "step": 5831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001989476722760654,
+      "loss": 1.9369,
+      "step": 5832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019887300024985089,
+      "loss": 1.8778,
+      "step": 5833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019879833298370235,
+      "loss": 2.1763,
+      "step": 5834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001987236704845717,
+      "loss": 1.9688,
+      "step": 5835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019864901275941005,
+      "loss": 2.0089,
+      "step": 5836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001985743598151684,
+      "loss": 1.8827,
+      "step": 5837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001984997116587972,
+      "loss": 2.2507,
+      "step": 5838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019842506829724618,
+      "loss": 2.1617,
+      "step": 5839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019835042973746498,
+      "loss": 1.8669,
+      "step": 5840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001982757959864026,
+      "loss": 2.1992,
+      "step": 5841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019820116705100777,
+      "loss": 2.006,
+      "step": 5842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001981265429382285,
+      "loss": 1.9143,
+      "step": 5843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019805192365501261,
+      "loss": 1.8399,
+      "step": 5844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019797730920830737,
+      "loss": 2.1429,
+      "step": 5845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019790269960505948,
+      "loss": 1.935,
+      "step": 5846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019782809485221532,
+      "loss": 2.151,
+      "step": 5847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001977534949567209,
+      "loss": 1.9836,
+      "step": 5848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001976788999255216,
+      "loss": 1.8569,
+      "step": 5849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019760430976556258,
+      "loss": 1.9007,
+      "step": 5850
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 2.0494749546051025,
+      "eval_runtime": 184.0206,
+      "eval_samples_per_second": 457.579,
+      "eval_steps_per_second": 28.6,
+      "step": 5850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019752972448378815,
+      "loss": 1.9466,
+      "step": 5851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019745514408714253,
+      "loss": 1.903,
+      "step": 5852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019738056858256933,
+      "loss": 1.814,
+      "step": 5853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019730599797701176,
+      "loss": 1.9212,
+      "step": 5854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001972314322774127,
+      "loss": 2.0368,
+      "step": 5855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019715687149071416,
+      "loss": 1.9056,
+      "step": 5856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019708231562385803,
+      "loss": 2.014,
+      "step": 5857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019700776468378577,
+      "loss": 1.9488,
+      "step": 5858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019693321867743824,
+      "loss": 1.8871,
+      "step": 5859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019685867761175583,
+      "loss": 2.1448,
+      "step": 5860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001967841414936786,
+      "loss": 1.9585,
+      "step": 5861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019670961033014603,
+      "loss": 1.9124,
+      "step": 5862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001966350841280971,
+      "loss": 2.1119,
+      "step": 5863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019656056289447048,
+      "loss": 2.221,
+      "step": 5864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019648604663620434,
+      "loss": 2.1584,
+      "step": 5865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019641153536023644,
+      "loss": 1.9132,
+      "step": 5866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963370290735037,
+      "loss": 2.0825,
+      "step": 5867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019626252778294303,
+      "loss": 1.8891,
+      "step": 5868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019618803149549076,
+      "loss": 2.0902,
+      "step": 5869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001961135402180826,
+      "loss": 1.9041,
+      "step": 5870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019603905395765395,
+      "loss": 1.8893,
+      "step": 5871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001959645727211398,
+      "loss": 2.0845,
+      "step": 5872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001958900965154743,
+      "loss": 2.1134,
+      "step": 5873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019581562534759156,
+      "loss": 1.9097,
+      "step": 5874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001957411592244251,
+      "loss": 2.2548,
+      "step": 5875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019566669815290774,
+      "loss": 2.0892,
+      "step": 5876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019559224213997218,
+      "loss": 2.044,
+      "step": 5877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019551779119255042,
+      "loss": 1.9148,
+      "step": 5878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001954433453175741,
+      "loss": 2.1403,
+      "step": 5879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019536890452197424,
+      "loss": 1.9679,
+      "step": 5880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001952944688126815,
+      "loss": 1.926,
+      "step": 5881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019522003819662614,
+      "loss": 2.0763,
+      "step": 5882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019514561268073788,
+      "loss": 2.0071,
+      "step": 5883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001950711922719458,
+      "loss": 1.8181,
+      "step": 5884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019499677697717868,
+      "loss": 2.109,
+      "step": 5885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019492236680336482,
+      "loss": 2.1825,
+      "step": 5886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019484796175743208,
+      "loss": 1.7987,
+      "step": 5887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019477356184630768,
+      "loss": 2.1453,
+      "step": 5888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019469916707691865,
+      "loss": 2.0811,
+      "step": 5889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019462477745619107,
+      "loss": 1.8905,
+      "step": 5890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019455039299105093,
+      "loss": 1.9353,
+      "step": 5891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019447601368842373,
+      "loss": 1.9868,
+      "step": 5892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019440163955523424,
+      "loss": 1.938,
+      "step": 5893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019432727059840704,
+      "loss": 2.027,
+      "step": 5894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019425290682486602,
+      "loss": 2.056,
+      "step": 5895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001941785482415346,
+      "loss": 2.0963,
+      "step": 5896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019410419485533582,
+      "loss": 2.0364,
+      "step": 5897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001940298466731922,
+      "loss": 1.9031,
+      "step": 5898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019395550370202572,
+      "loss": 1.8909,
+      "step": 5899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001938811659487581,
+      "loss": 2.029,
+      "step": 5900
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 2.0456221103668213,
+      "eval_runtime": 184.0607,
+      "eval_samples_per_second": 457.48,
+      "eval_steps_per_second": 28.594,
+      "step": 5900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019380683342031012,
+      "loss": 1.9543,
+      "step": 5901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019373250612360245,
+      "loss": 2.0942,
+      "step": 5902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001936581840655552,
+      "loss": 2.0376,
+      "step": 5903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019358386725308797,
+      "loss": 2.0297,
+      "step": 5904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019350955569311985,
+      "loss": 1.9934,
+      "step": 5905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019343524939256952,
+      "loss": 1.8573,
+      "step": 5906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019336094835835493,
+      "loss": 2.054,
+      "step": 5907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001932866525973938,
+      "loss": 1.9348,
+      "step": 5908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019321236211660336,
+      "loss": 1.8879,
+      "step": 5909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001931380769229001,
+      "loss": 2.0524,
+      "step": 5910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019306379702320036,
+      "loss": 2.143,
+      "step": 5911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001929895224244197,
+      "loss": 1.8899,
+      "step": 5912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001929152531334732,
+      "loss": 1.9996,
+      "step": 5913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019284098915727568,
+      "loss": 1.9185,
+      "step": 5914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019276673050274125,
+      "loss": 1.8903,
+      "step": 5915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001926924771767837,
+      "loss": 1.9252,
+      "step": 5916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019261822918631617,
+      "loss": 1.8691,
+      "step": 5917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019254398653825125,
+      "loss": 1.9486,
+      "step": 5918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019246974923950124,
+      "loss": 2.0806,
+      "step": 5919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001923955172969778,
+      "loss": 2.3281,
+      "step": 5920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001923212907175922,
+      "loss": 2.0595,
+      "step": 5921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019224706950825515,
+      "loss": 2.2654,
+      "step": 5922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001921728536758767,
+      "loss": 2.0213,
+      "step": 5923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019209864322736664,
+      "loss": 2.0,
+      "step": 5924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019202443816963426,
+      "loss": 1.8189,
+      "step": 5925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001919502385095881,
+      "loss": 2.0203,
+      "step": 5926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001918760442541364,
+      "loss": 2.0858,
+      "step": 5927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019180185541018695,
+      "loss": 2.1443,
+      "step": 5928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001917276719846469,
+      "loss": 1.9933,
+      "step": 5929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019165349398442283,
+      "loss": 2.1042,
+      "step": 5930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019157932141642097,
+      "loss": 2.169,
+      "step": 5931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000191505154287547,
+      "loss": 2.0674,
+      "step": 5932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019143099260470625,
+      "loss": 2.1297,
+      "step": 5933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019135683637480308,
+      "loss": 2.0125,
+      "step": 5934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019128268560474182,
+      "loss": 2.0684,
+      "step": 5935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019120854030142605,
+      "loss": 1.8836,
+      "step": 5936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019113440047175898,
+      "loss": 2.0536,
+      "step": 5937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019106026612264316,
+      "loss": 2.0133,
+      "step": 5938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019098613726098085,
+      "loss": 1.933,
+      "step": 5939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001909120138936734,
+      "loss": 2.0237,
+      "step": 5940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019083789602762207,
+      "loss": 1.9798,
+      "step": 5941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001907637836697275,
+      "loss": 1.9611,
+      "step": 5942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019068967682688955,
+      "loss": 2.0829,
+      "step": 5943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019061557550600794,
+      "loss": 2.0188,
+      "step": 5944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019054147971398174,
+      "loss": 1.9398,
+      "step": 5945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019046738945770932,
+      "loss": 1.9761,
+      "step": 5946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019039330474408873,
+      "loss": 1.8726,
+      "step": 5947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019031922558001752,
+      "loss": 1.9813,
+      "step": 5948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019024515197239262,
+      "loss": 2.1429,
+      "step": 5949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019017108392811063,
+      "loss": 1.9925,
+      "step": 5950
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 2.04077410697937,
+      "eval_runtime": 184.1125,
+      "eval_samples_per_second": 457.351,
+      "eval_steps_per_second": 28.586,
+      "step": 5950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019009702145406728,
+      "loss": 1.8995,
+      "step": 5951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019002296455715805,
+      "loss": 2.0179,
+      "step": 5952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018994891324427792,
+      "loss": 1.9116,
+      "step": 5953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898748675223212,
+      "loss": 1.9207,
+      "step": 5954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018980082739818182,
+      "loss": 2.235,
+      "step": 5955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897267928787532,
+      "loss": 2.0329,
+      "step": 5956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018965276397092792,
+      "loss": 1.8196,
+      "step": 5957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018957874068159842,
+      "loss": 2.0476,
+      "step": 5958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001895047230176565,
+      "loss": 1.9765,
+      "step": 5959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001894307109859933,
+      "loss": 1.8999,
+      "step": 5960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001893567045934997,
+      "loss": 2.0902,
+      "step": 5961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018928270384706584,
+      "loss": 2.0283,
+      "step": 5962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018920870875358128,
+      "loss": 2.0586,
+      "step": 5963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018913471931993528,
+      "loss": 2.0347,
+      "step": 5964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018906073555301645,
+      "loss": 2.0404,
+      "step": 5965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018898675745971288,
+      "loss": 1.9414,
+      "step": 5966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018891278504691223,
+      "loss": 2.0642,
+      "step": 5967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018883881832150134,
+      "loss": 1.808,
+      "step": 5968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018876485729036685,
+      "loss": 2.0441,
+      "step": 5969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018869090196039468,
+      "loss": 2.1845,
+      "step": 5970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018861695233847036,
+      "loss": 2.1923,
+      "step": 5971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018854300843147877,
+      "loss": 2.041,
+      "step": 5972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018846907024630424,
+      "loss": 2.0984,
+      "step": 5973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018839513778983068,
+      "loss": 2.066,
+      "step": 5974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001883212110689414,
+      "loss": 1.9136,
+      "step": 5975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001882472900905191,
+      "loss": 1.8659,
+      "step": 5976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001881733748614461,
+      "loss": 2.1097,
+      "step": 5977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018809946538860426,
+      "loss": 2.1016,
+      "step": 5978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001880255616788745,
+      "loss": 1.9316,
+      "step": 5979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018795166373913755,
+      "loss": 1.9601,
+      "step": 5980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018787777157627353,
+      "loss": 2.0428,
+      "step": 5981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018780388519716202,
+      "loss": 1.8984,
+      "step": 5982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018773000460868205,
+      "loss": 2.0627,
+      "step": 5983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001876561298177122,
+      "loss": 1.9943,
+      "step": 5984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018758226083113023,
+      "loss": 1.9445,
+      "step": 5985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001875083976558136,
+      "loss": 2.0768,
+      "step": 5986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018743454029863924,
+      "loss": 1.7991,
+      "step": 5987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018736068876648343,
+      "loss": 1.9794,
+      "step": 5988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001872868430662221,
+      "loss": 1.9323,
+      "step": 5989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018721300320473022,
+      "loss": 2.0908,
+      "step": 5990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018713916918888267,
+      "loss": 1.9957,
+      "step": 5991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001870653410255536,
+      "loss": 1.901,
+      "step": 5992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001869915187216165,
+      "loss": 2.1377,
+      "step": 5993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018691770228394455,
+      "loss": 1.8909,
+      "step": 5994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018684389171941032,
+      "loss": 1.8906,
+      "step": 5995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018677008703488557,
+      "loss": 2.0173,
+      "step": 5996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018669628823724186,
+      "loss": 2.3136,
+      "step": 5997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018662249533335002,
+      "loss": 1.925,
+      "step": 5998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001865487083300804,
+      "loss": 1.9385,
+      "step": 5999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001864749272343028,
+      "loss": 1.936,
+      "step": 6000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 2.0346012115478516,
+      "eval_runtime": 184.1054,
+      "eval_samples_per_second": 457.368,
+      "eval_steps_per_second": 28.587,
+      "step": 6000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001864011520528865,
+      "loss": 1.8621,
+      "step": 6001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001863273827927,
+      "loss": 1.954,
+      "step": 6002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001862536194606115,
+      "loss": 1.9768,
+      "step": 6003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001861798620634886,
+      "loss": 2.0736,
+      "step": 6004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018610611060819828,
+      "loss": 2.0705,
+      "step": 6005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001860323651016072,
+      "loss": 1.8749,
+      "step": 6006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018595862555058094,
+      "loss": 2.0851,
+      "step": 6007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018588489196198505,
+      "loss": 1.8302,
+      "step": 6008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018581116434268435,
+      "loss": 2.1392,
+      "step": 6009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018573744269954299,
+      "loss": 1.8238,
+      "step": 6010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018566372703942468,
+      "loss": 2.0286,
+      "step": 6011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018559001736919274,
+      "loss": 2.0679,
+      "step": 6012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018551631369570948,
+      "loss": 1.9839,
+      "step": 6013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000185442616025837,
+      "loss": 2.3086,
+      "step": 6014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018536892436643682,
+      "loss": 1.9585,
+      "step": 6015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001852952387243698,
+      "loss": 1.9551,
+      "step": 6016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018522155910649638,
+      "loss": 1.8673,
+      "step": 6017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018514788551967615,
+      "loss": 1.8759,
+      "step": 6018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001850742179707684,
+      "loss": 1.9951,
+      "step": 6019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001850005564666318,
+      "loss": 2.0902,
+      "step": 6020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018492690101412447,
+      "loss": 2.0758,
+      "step": 6021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018485325162010392,
+      "loss": 2.2241,
+      "step": 6022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018477960829142714,
+      "loss": 1.954,
+      "step": 6023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001847059710349504,
+      "loss": 1.9229,
+      "step": 6024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018463233985752976,
+      "loss": 1.9328,
+      "step": 6025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018455871476602022,
+      "loss": 2.1123,
+      "step": 6026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018448509576727664,
+      "loss": 2.0342,
+      "step": 6027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018441148286815314,
+      "loss": 1.7543,
+      "step": 6028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001843378760755034,
+      "loss": 2.0873,
+      "step": 6029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018426427539618018,
+      "loss": 1.9214,
+      "step": 6030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018419068083703598,
+      "loss": 2.0749,
+      "step": 6031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018411709240492274,
+      "loss": 1.9954,
+      "step": 6032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001840435101066917,
+      "loss": 1.8017,
+      "step": 6033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001839699339491937,
+      "loss": 1.961,
+      "step": 6034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018389636393927866,
+      "loss": 2.0782,
+      "step": 6035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018382280008379627,
+      "loss": 2.1124,
+      "step": 6036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018374924238959552,
+      "loss": 1.9463,
+      "step": 6037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018367569086352481,
+      "loss": 2.0364,
+      "step": 6038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001836021455124321,
+      "loss": 2.0342,
+      "step": 6039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018352860634316456,
+      "loss": 1.8648,
+      "step": 6040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018345507336256888,
+      "loss": 2.0597,
+      "step": 6041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001833815465774913,
+      "loss": 2.0374,
+      "step": 6042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018330802599477721,
+      "loss": 1.8592,
+      "step": 6043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001832345116212717,
+      "loss": 2.1218,
+      "step": 6044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001831610034638192,
+      "loss": 2.1595,
+      "step": 6045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018308750152926337,
+      "loss": 1.8659,
+      "step": 6046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001830140058244475,
+      "loss": 2.0129,
+      "step": 6047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001829405163562143,
+      "loss": 1.819,
+      "step": 6048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001828670331314058,
+      "loss": 2.0206,
+      "step": 6049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018279355615686353,
+      "loss": 2.0462,
+      "step": 6050
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 2.0309104919433594,
+      "eval_runtime": 183.9432,
+      "eval_samples_per_second": 457.772,
+      "eval_steps_per_second": 28.612,
+      "step": 6050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018272008543942847,
+      "loss": 1.9515,
+      "step": 6051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018264662098594077,
+      "loss": 1.9133,
+      "step": 6052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001825731628032402,
+      "loss": 2.081,
+      "step": 6053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018249971089816602,
+      "loss": 2.0308,
+      "step": 6054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001824262652775568,
+      "loss": 1.8923,
+      "step": 6055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018235282594825054,
+      "loss": 2.0047,
+      "step": 6056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018227939291708462,
+      "loss": 2.0127,
+      "step": 6057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018220596619089574,
+      "loss": 2.1292,
+      "step": 6058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018213254577652036,
+      "loss": 2.0045,
+      "step": 6059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001820591316807939,
+      "loss": 1.9081,
+      "step": 6060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018198572391055158,
+      "loss": 2.0985,
+      "step": 6061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001819123224726279,
+      "loss": 2.0223,
+      "step": 6062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018183892737385656,
+      "loss": 2.0201,
+      "step": 6063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018176553862107086,
+      "loss": 2.0042,
+      "step": 6064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018169215622110363,
+      "loss": 1.8831,
+      "step": 6065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018161878018078693,
+      "loss": 2.072,
+      "step": 6066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018154541050695227,
+      "loss": 1.9933,
+      "step": 6067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018147204720643065,
+      "loss": 2.0531,
+      "step": 6068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813986902860522,
+      "loss": 1.8531,
+      "step": 6069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813253397526468,
+      "loss": 1.9339,
+      "step": 6070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018125199561304356,
+      "loss": 1.9042,
+      "step": 6071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018117865787407102,
+      "loss": 2.0914,
+      "step": 6072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001811053265425572,
+      "loss": 2.0858,
+      "step": 6073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018103200162532926,
+      "loss": 2.0362,
+      "step": 6074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018095868312921416,
+      "loss": 2.1298,
+      "step": 6075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018088537106103792,
+      "loss": 1.9288,
+      "step": 6076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018081206542762613,
+      "loss": 2.2086,
+      "step": 6077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018073876623580376,
+      "loss": 1.9126,
+      "step": 6078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001806654734923953,
+      "loss": 1.887,
+      "step": 6079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018059218720422427,
+      "loss": 1.9814,
+      "step": 6080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018051890737811395,
+      "loss": 1.9826,
+      "step": 6081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018044563402088684,
+      "loss": 2.0067,
+      "step": 6082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000180372367139365,
+      "loss": 1.8251,
+      "step": 6083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001802991067403697,
+      "loss": 2.0409,
+      "step": 6084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001802258528307218,
+      "loss": 2.0309,
+      "step": 6085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018015260541724126,
+      "loss": 2.0472,
+      "step": 6086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018007936450674768,
+      "loss": 2.1575,
+      "step": 6087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018000613010606006,
+      "loss": 2.1345,
+      "step": 6088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017993290222199674,
+      "loss": 1.982,
+      "step": 6089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017985968086137538,
+      "loss": 1.9048,
+      "step": 6090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017978646603101305,
+      "loss": 2.1285,
+      "step": 6091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017971325773772634,
+      "loss": 2.0796,
+      "step": 6092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017964005598833109,
+      "loss": 2.0807,
+      "step": 6093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017956686078964256,
+      "loss": 2.0385,
+      "step": 6094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017949367214847555,
+      "loss": 1.9404,
+      "step": 6095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001794204900716441,
+      "loss": 1.8144,
+      "step": 6096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017934731456596155,
+      "loss": 1.9137,
+      "step": 6097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017927414563824077,
+      "loss": 1.9001,
+      "step": 6098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017920098329529406,
+      "loss": 1.964,
+      "step": 6099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017912782754393296,
+      "loss": 1.9308,
+      "step": 6100
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 2.0266716480255127,
+      "eval_runtime": 184.1297,
+      "eval_samples_per_second": 457.308,
+      "eval_steps_per_second": 28.583,
+      "step": 6100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001790546783909687,
+      "loss": 1.9075,
+      "step": 6101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017898153584321137,
+      "loss": 2.2709,
+      "step": 6102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017890839990747086,
+      "loss": 2.0224,
+      "step": 6103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017883527059055633,
+      "loss": 1.978,
+      "step": 6104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017876214789927633,
+      "loss": 2.0061,
+      "step": 6105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017868903184043886,
+      "loss": 2.1729,
+      "step": 6106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017861592242085115,
+      "loss": 1.9438,
+      "step": 6107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017854281964731984,
+      "loss": 1.8577,
+      "step": 6108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017846972352665108,
+      "loss": 2.0339,
+      "step": 6109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001783966340656503,
+      "loss": 2.0659,
+      "step": 6110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017832355127112232,
+      "loss": 2.1018,
+      "step": 6111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017825047514987132,
+      "loss": 1.9609,
+      "step": 6112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001781774057087011,
+      "loss": 2.0331,
+      "step": 6113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017810434295441433,
+      "loss": 2.2094,
+      "step": 6114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001780312868938135,
+      "loss": 1.977,
+      "step": 6115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017795823753370032,
+      "loss": 2.2302,
+      "step": 6116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001778851948808759,
+      "loss": 1.9319,
+      "step": 6117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001778121589421408,
+      "loss": 2.0497,
+      "step": 6118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017773912972429465,
+      "loss": 1.9479,
+      "step": 6119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017766610723413684,
+      "loss": 1.9114,
+      "step": 6120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001775930914784659,
+      "loss": 2.1713,
+      "step": 6121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017752008246407986,
+      "loss": 2.1469,
+      "step": 6122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000177447080197776,
+      "loss": 2.1166,
+      "step": 6123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017737408468635114,
+      "loss": 2.1142,
+      "step": 6124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017730109593660126,
+      "loss": 2.0417,
+      "step": 6125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001772281139553218,
+      "loss": 2.0534,
+      "step": 6126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017715513874930764,
+      "loss": 2.0088,
+      "step": 6127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017708217032535295,
+      "loss": 1.9276,
+      "step": 6128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001770092086902514,
+      "loss": 1.9568,
+      "step": 6129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017693625385079576,
+      "loss": 2.0406,
+      "step": 6130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017686330581377838,
+      "loss": 2.0237,
+      "step": 6131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017679036458599095,
+      "loss": 2.0866,
+      "step": 6132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017671743017422448,
+      "loss": 1.9643,
+      "step": 6133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017664450258526943,
+      "loss": 1.9232,
+      "step": 6134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017657158182591556,
+      "loss": 2.0979,
+      "step": 6135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001764986679029519,
+      "loss": 1.8866,
+      "step": 6136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017642576082316694,
+      "loss": 1.8339,
+      "step": 6137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001763528605933486,
+      "loss": 2.1187,
+      "step": 6138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017627996722028414,
+      "loss": 1.9242,
+      "step": 6139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017620708071076002,
+      "loss": 2.0985,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017613420107156225,
+      "loss": 1.8405,
+      "step": 6141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017606132830947613,
+      "loss": 2.1204,
+      "step": 6142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017598846243128624,
+      "loss": 1.9911,
+      "step": 6143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017591560344377665,
+      "loss": 2.0175,
+      "step": 6144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017584275135373075,
+      "loss": 1.9782,
+      "step": 6145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017576990616793137,
+      "loss": 1.9454,
+      "step": 6146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017569706789316035,
+      "loss": 2.0596,
+      "step": 6147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001756242365361993,
+      "loss": 2.2211,
+      "step": 6148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017555141210382895,
+      "loss": 2.1046,
+      "step": 6149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017547859460282955,
+      "loss": 1.957,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 2.0216543674468994,
+      "eval_runtime": 184.245,
+      "eval_samples_per_second": 457.022,
+      "eval_steps_per_second": 28.565,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017540578403998054,
+      "loss": 1.8773,
+      "step": 6151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017533298042206094,
+      "loss": 1.8857,
+      "step": 6152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017526018375584873,
+      "loss": 2.0078,
+      "step": 6153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017518739404812157,
+      "loss": 1.8149,
+      "step": 6154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017511461130565644,
+      "loss": 2.0133,
+      "step": 6155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017504183553522962,
+      "loss": 1.989,
+      "step": 6156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017496906674361668,
+      "loss": 2.0615,
+      "step": 6157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017489630493759263,
+      "loss": 1.9365,
+      "step": 6158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017482355012393176,
+      "loss": 2.0608,
+      "step": 6159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017475080230940776,
+      "loss": 1.9134,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017467806150079366,
+      "loss": 1.9208,
+      "step": 6161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017460532770486185,
+      "loss": 2.0676,
+      "step": 6162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017453260092838412,
+      "loss": 2.1814,
+      "step": 6163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017445988117813132,
+      "loss": 1.9677,
+      "step": 6164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017438716846087404,
+      "loss": 1.9026,
+      "step": 6165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017431446278338197,
+      "loss": 1.9924,
+      "step": 6166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017424176415242425,
+      "loss": 1.806,
+      "step": 6167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001741690725747693,
+      "loss": 1.9048,
+      "step": 6168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000174096388057185,
+      "loss": 2.1241,
+      "step": 6169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017402371060643829,
+      "loss": 1.9551,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017395104022929577,
+      "loss": 1.8213,
+      "step": 6171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017387837693252328,
+      "loss": 1.9642,
+      "step": 6172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017380572072288585,
+      "loss": 2.0669,
+      "step": 6173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017373307160714813,
+      "loss": 1.9616,
+      "step": 6174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017366042959207387,
+      "loss": 2.1346,
+      "step": 6175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001735877946844262,
+      "loss": 2.0395,
+      "step": 6176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001735151668909677,
+      "loss": 1.9363,
+      "step": 6177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017344254621846016,
+      "loss": 1.9841,
+      "step": 6178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017336993267366485,
+      "loss": 1.8375,
+      "step": 6179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017329732626334234,
+      "loss": 1.8499,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017322472699425235,
+      "loss": 2.0081,
+      "step": 6181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017315213487315405,
+      "loss": 2.1292,
+      "step": 6182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017307954990680607,
+      "loss": 2.051,
+      "step": 6183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017300697210196627,
+      "loss": 1.9682,
+      "step": 6184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017293440146539193,
+      "loss": 2.0851,
+      "step": 6185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017286183800383937,
+      "loss": 1.9316,
+      "step": 6186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017278928172406456,
+      "loss": 2.1483,
+      "step": 6187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017271673263282263,
+      "loss": 1.7552,
+      "step": 6188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017264419073686826,
+      "loss": 2.0104,
+      "step": 6189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017257165604295512,
+      "loss": 2.1082,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017249912855783655,
+      "loss": 1.8959,
+      "step": 6191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000172426608288265,
+      "loss": 1.9184,
+      "step": 6192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001723540952409922,
+      "loss": 2.0914,
+      "step": 6193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017228158942276942,
+      "loss": 1.8948,
+      "step": 6194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017220909084034714,
+      "loss": 1.9786,
+      "step": 6195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017213659950047528,
+      "loss": 1.9219,
+      "step": 6196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001720641154099028,
+      "loss": 1.917,
+      "step": 6197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017199163857537826,
+      "loss": 1.8509,
+      "step": 6198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017191916900364946,
+      "loss": 1.9557,
+      "step": 6199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017184670670146348,
+      "loss": 1.9303,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 2.0154411792755127,
+      "eval_runtime": 184.149,
+      "eval_samples_per_second": 457.26,
+      "eval_steps_per_second": 28.58,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001717742516755668,
+      "loss": 1.9522,
+      "step": 6201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001717018039327053,
+      "loss": 2.0345,
+      "step": 6202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001716293634796239,
+      "loss": 2.0867,
+      "step": 6203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017155693032306696,
+      "loss": 2.0746,
+      "step": 6204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017148450446977838,
+      "loss": 2.1717,
+      "step": 6205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001714120859265011,
+      "loss": 2.1025,
+      "step": 6206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001713396746999775,
+      "loss": 1.9397,
+      "step": 6207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017126727079694938,
+      "loss": 2.1535,
+      "step": 6208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017119487422415764,
+      "loss": 2.1234,
+      "step": 6209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017112248498834254,
+      "loss": 1.9071,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001710501030962438,
+      "loss": 1.9652,
+      "step": 6211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017097772855460036,
+      "loss": 1.9797,
+      "step": 6212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001709053613701506,
+      "loss": 1.7541,
+      "step": 6213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017083300154963193,
+      "loss": 1.7671,
+      "step": 6214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001707606490997813,
+      "loss": 1.9411,
+      "step": 6215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000170688304027335,
+      "loss": 2.0961,
+      "step": 6216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017061596633902848,
+      "loss": 2.1976,
+      "step": 6217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017054363604159659,
+      "loss": 1.9227,
+      "step": 6218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017047131314177363,
+      "loss": 1.9134,
+      "step": 6219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017039899764629285,
+      "loss": 2.0415,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017032668956188708,
+      "loss": 1.9224,
+      "step": 6221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001702543888952885,
+      "loss": 1.9667,
+      "step": 6222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001701820956532284,
+      "loss": 2.0137,
+      "step": 6223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017010980984243754,
+      "loss": 2.0964,
+      "step": 6224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017003753146964593,
+      "loss": 1.8087,
+      "step": 6225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001699652605415828,
+      "loss": 1.7979,
+      "step": 6226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016989299706497688,
+      "loss": 1.9938,
+      "step": 6227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016982074104655604,
+      "loss": 2.1414,
+      "step": 6228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001697484924930475,
+      "loss": 1.7772,
+      "step": 6229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016967625141117804,
+      "loss": 1.9924,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016960401780767315,
+      "loss": 1.7443,
+      "step": 6231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016953179168925815,
+      "loss": 1.8319,
+      "step": 6232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016945957306265747,
+      "loss": 1.9958,
+      "step": 6233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016938736193459485,
+      "loss": 1.8729,
+      "step": 6234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016931515831179343,
+      "loss": 1.8793,
+      "step": 6235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016924296220097556,
+      "loss": 1.9917,
+      "step": 6236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001691707736088628,
+      "loss": 2.0339,
+      "step": 6237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016909859254217615,
+      "loss": 1.9219,
+      "step": 6238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001690264190076359,
+      "loss": 1.8675,
+      "step": 6239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016895425301196156,
+      "loss": 2.0168,
+      "step": 6240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016888209456187203,
+      "loss": 1.7573,
+      "step": 6241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016880994366408547,
+      "loss": 1.8222,
+      "step": 6242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016873780032531928,
+      "loss": 2.092,
+      "step": 6243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016866566455229015,
+      "loss": 1.8715,
+      "step": 6244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016859353635171427,
+      "loss": 2.0421,
+      "step": 6245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016852141573030688,
+      "loss": 2.0005,
+      "step": 6246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016844930269478272,
+      "loss": 1.9942,
+      "step": 6247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016837719725185557,
+      "loss": 1.9948,
+      "step": 6248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001683050994082387,
+      "loss": 2.0529,
+      "step": 6249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001682330091706446,
+      "loss": 2.0676,
+      "step": 6250
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 2.011756658554077,
+      "eval_runtime": 184.2625,
+      "eval_samples_per_second": 456.978,
+      "eval_steps_per_second": 28.563,
+      "step": 6250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001681609265457851,
+      "loss": 2.0218,
+      "step": 6251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001680888515403714,
+      "loss": 2.1062,
+      "step": 6252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001680167841611137,
+      "loss": 1.9175,
+      "step": 6253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016794472441472176,
+      "loss": 1.831,
+      "step": 6254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016787267230790453,
+      "loss": 2.1637,
+      "step": 6255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016780062784737027,
+      "loss": 1.9519,
+      "step": 6256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016772859103982653,
+      "loss": 2.0601,
+      "step": 6257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016765656189198012,
+      "loss": 1.801,
+      "step": 6258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016758454041053717,
+      "loss": 1.9562,
+      "step": 6259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016751252660220304,
+      "loss": 1.9231,
+      "step": 6260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016744052047368245,
+      "loss": 1.8963,
+      "step": 6261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016736852203167935,
+      "loss": 2.1699,
+      "step": 6262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000167296531282897,
+      "loss": 2.0362,
+      "step": 6263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016722454823403808,
+      "loss": 1.9683,
+      "step": 6264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016715257289180412,
+      "loss": 2.0042,
+      "step": 6265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016708060526289648,
+      "loss": 2.0669,
+      "step": 6266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016700864535401538,
+      "loss": 1.876,
+      "step": 6267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001669366931718606,
+      "loss": 1.9528,
+      "step": 6268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016686474872313115,
+      "loss": 1.7688,
+      "step": 6269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016679281201452506,
+      "loss": 2.0934,
+      "step": 6270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001667208830527399,
+      "loss": 1.915,
+      "step": 6271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001666489618444726,
+      "loss": 1.9962,
+      "step": 6272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016657704839641903,
+      "loss": 1.976,
+      "step": 6273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016650514271527466,
+      "loss": 2.0685,
+      "step": 6274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016643324480773415,
+      "loss": 2.0514,
+      "step": 6275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001663613546804912,
+      "loss": 2.1842,
+      "step": 6276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001662894723402391,
+      "loss": 2.0596,
+      "step": 6277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016621759779367033,
+      "loss": 2.0752,
+      "step": 6278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016614573104747654,
+      "loss": 1.8555,
+      "step": 6279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016607387210834887,
+      "loss": 1.9844,
+      "step": 6280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016600202098297737,
+      "loss": 1.9222,
+      "step": 6281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001659301776780517,
+      "loss": 1.78,
+      "step": 6282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016585834220026062,
+      "loss": 1.859,
+      "step": 6283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016578651455629228,
+      "loss": 2.0896,
+      "step": 6284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000165714694752834,
+      "loss": 1.9744,
+      "step": 6285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016564288279657252,
+      "loss": 1.8787,
+      "step": 6286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016557107869419349,
+      "loss": 1.8777,
+      "step": 6287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016549928245238223,
+      "loss": 1.9179,
+      "step": 6288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001654274940778232,
+      "loss": 2.0416,
+      "step": 6289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016535571357719997,
+      "loss": 1.7842,
+      "step": 6290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016528394095719558,
+      "loss": 2.1952,
+      "step": 6291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016521217622449236,
+      "loss": 2.1093,
+      "step": 6292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016514041938577156,
+      "loss": 1.8609,
+      "step": 6293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016506867044771415,
+      "loss": 1.9415,
+      "step": 6294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016499692941700006,
+      "loss": 1.948,
+      "step": 6295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016492519630030862,
+      "loss": 1.8078,
+      "step": 6296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016485347110431847,
+      "loss": 2.0953,
+      "step": 6297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001647817538357072,
+      "loss": 1.8728,
+      "step": 6298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016471004450115207,
+      "loss": 1.967,
+      "step": 6299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016463834310732932,
+      "loss": 2.0288,
+      "step": 6300
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 2.0064728260040283,
+      "eval_runtime": 184.4754,
+      "eval_samples_per_second": 456.451,
+      "eval_steps_per_second": 28.53,
+      "step": 6300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001645666496609146,
+      "loss": 2.0099,
+      "step": 6301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016449496416858284,
+      "loss": 1.9501,
+      "step": 6302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016442328663700807,
+      "loss": 1.8757,
+      "step": 6303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016435161707286362,
+      "loss": 2.0998,
+      "step": 6304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016427995548282223,
+      "loss": 1.8463,
+      "step": 6305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016420830187355572,
+      "loss": 2.001,
+      "step": 6306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016413665625173525,
+      "loss": 2.0538,
+      "step": 6307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016406501862403132,
+      "loss": 2.0957,
+      "step": 6308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016399338899711352,
+      "loss": 2.0365,
+      "step": 6309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016392176737765069,
+      "loss": 1.7905,
+      "step": 6310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001638501537723111,
+      "loss": 2.0516,
+      "step": 6311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016377854818776217,
+      "loss": 1.9742,
+      "step": 6312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016370695063067054,
+      "loss": 2.0579,
+      "step": 6313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016363536110770228,
+      "loss": 1.8058,
+      "step": 6314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016356377962552238,
+      "loss": 1.9714,
+      "step": 6315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016349220619079534,
+      "loss": 1.9996,
+      "step": 6316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016342064081018484,
+      "loss": 1.9372,
+      "step": 6317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016334908349035387,
+      "loss": 1.9545,
+      "step": 6318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016327753423796463,
+      "loss": 1.9547,
+      "step": 6319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016320599305967852,
+      "loss": 2.0999,
+      "step": 6320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016313445996215615,
+      "loss": 1.7711,
+      "step": 6321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016306293495205757,
+      "loss": 1.9032,
+      "step": 6322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016299141803604184,
+      "loss": 1.964,
+      "step": 6323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016291990922076745,
+      "loss": 1.9364,
+      "step": 6324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016284840851289218,
+      "loss": 1.8463,
+      "step": 6325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001627769159190727,
+      "loss": 1.9902,
+      "step": 6326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016270543144596535,
+      "loss": 2.0102,
+      "step": 6327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016263395510022544,
+      "loss": 1.8106,
+      "step": 6328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016256248688850767,
+      "loss": 1.9987,
+      "step": 6329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001624910268174659,
+      "loss": 2.1326,
+      "step": 6330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016241957489375342,
+      "loss": 1.9114,
+      "step": 6331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016234813112402235,
+      "loss": 1.9481,
+      "step": 6332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001622766955149244,
+      "loss": 1.8048,
+      "step": 6333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016220526807311049,
+      "loss": 1.9248,
+      "step": 6334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016213384880523063,
+      "loss": 2.0203,
+      "step": 6335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016206243771793433,
+      "loss": 1.8211,
+      "step": 6336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016199103481786986,
+      "loss": 1.9559,
+      "step": 6337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016191964011168523,
+      "loss": 1.8563,
+      "step": 6338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001618482536060275,
+      "loss": 1.8459,
+      "step": 6339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016177687530754281,
+      "loss": 1.989,
+      "step": 6340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001617055052228768,
+      "loss": 1.7133,
+      "step": 6341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016163414335867426,
+      "loss": 1.9272,
+      "step": 6342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016156278972157906,
+      "loss": 1.8625,
+      "step": 6343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016149144431823443,
+      "loss": 2.0285,
+      "step": 6344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001614201071552829,
+      "loss": 2.1064,
+      "step": 6345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016134877823936609,
+      "loss": 1.7306,
+      "step": 6346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000161277457577125,
+      "loss": 1.941,
+      "step": 6347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016120614517519977,
+      "loss": 2.0729,
+      "step": 6348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016113484104022968,
+      "loss": 1.8862,
+      "step": 6349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016106354517885345,
+      "loss": 1.833,
+      "step": 6350
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 2.004859685897827,
+      "eval_runtime": 184.1399,
+      "eval_samples_per_second": 457.283,
+      "eval_steps_per_second": 28.582,
+      "step": 6350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016099225759770886,
+      "loss": 2.1592,
+      "step": 6351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016092097830343305,
+      "loss": 2.0653,
+      "step": 6352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016084970730266229,
+      "loss": 1.9549,
+      "step": 6353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016077844460203207,
+      "loss": 2.005,
+      "step": 6354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001607071902081772,
+      "loss": 2.0523,
+      "step": 6355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001606359441277316,
+      "loss": 1.991,
+      "step": 6356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001605647063673285,
+      "loss": 1.8725,
+      "step": 6357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016049347693360038,
+      "loss": 1.9876,
+      "step": 6358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016042225583317899,
+      "loss": 2.0109,
+      "step": 6359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016035104307269494,
+      "loss": 1.8675,
+      "step": 6360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001602798386587785,
+      "loss": 2.045,
+      "step": 6361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000160208642598059,
+      "loss": 2.049,
+      "step": 6362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000160137454897165,
+      "loss": 1.8005,
+      "step": 6363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016006627556272435,
+      "loss": 1.9856,
+      "step": 6364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015999510460136383,
+      "loss": 1.674,
+      "step": 6365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001599239420197098,
+      "loss": 2.1123,
+      "step": 6366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001598527878243876,
+      "loss": 1.8202,
+      "step": 6367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000159781642022022,
+      "loss": 1.9118,
+      "step": 6368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015971050461923686,
+      "loss": 1.8659,
+      "step": 6369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015963937562265523,
+      "loss": 2.1306,
+      "step": 6370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015956825503889938,
+      "loss": 1.7946,
+      "step": 6371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015949714287459094,
+      "loss": 1.8062,
+      "step": 6372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015942603913635052,
+      "loss": 2.1547,
+      "step": 6373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015935494383079814,
+      "loss": 1.9635,
+      "step": 6374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015928385696455298,
+      "loss": 2.0228,
+      "step": 6375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001592127785442335,
+      "loss": 1.9343,
+      "step": 6376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015914170857645715,
+      "loss": 1.9501,
+      "step": 6377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001590706470678408,
+      "loss": 2.0339,
+      "step": 6378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015899959402500047,
+      "loss": 1.9892,
+      "step": 6379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001589285494545514,
+      "loss": 1.9107,
+      "step": 6380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015885751336310817,
+      "loss": 1.9977,
+      "step": 6381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587864857572842,
+      "loss": 2.0317,
+      "step": 6382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587154666436925,
+      "loss": 1.9064,
+      "step": 6383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586444560289451,
+      "loss": 2.0117,
+      "step": 6384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015857345391965326,
+      "loss": 2.1416,
+      "step": 6385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015850246032242764,
+      "loss": 1.8406,
+      "step": 6386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584314752438778,
+      "loss": 2.005,
+      "step": 6387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015836049869061264,
+      "loss": 1.9891,
+      "step": 6388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015828953066924034,
+      "loss": 2.1386,
+      "step": 6389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015821857118636812,
+      "loss": 2.0583,
+      "step": 6390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001581476202486026,
+      "loss": 2.1161,
+      "step": 6391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015807667786254956,
+      "loss": 1.9841,
+      "step": 6392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015800574403481376,
+      "loss": 2.09,
+      "step": 6393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015793481877199946,
+      "loss": 1.8747,
+      "step": 6394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015786390208070994,
+      "loss": 1.8444,
+      "step": 6395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015779299396754775,
+      "loss": 1.7526,
+      "step": 6396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001577220944391147,
+      "loss": 2.1044,
+      "step": 6397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001576512035020118,
+      "loss": 1.8429,
+      "step": 6398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015758032116283894,
+      "loss": 1.8157,
+      "step": 6399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001575094474281956,
+      "loss": 2.0001,
+      "step": 6400
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.9984697103500366,
+      "eval_runtime": 184.1193,
+      "eval_samples_per_second": 457.334,
+      "eval_steps_per_second": 28.585,
+      "step": 6400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001574385823046804,
+      "loss": 2.0223,
+      "step": 6401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015736772579889099,
+      "loss": 1.9688,
+      "step": 6402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001572968779174243,
+      "loss": 1.8558,
+      "step": 6403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015722603866687656,
+      "loss": 1.8456,
+      "step": 6404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000157155208053843,
+      "loss": 1.7764,
+      "step": 6405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015708438608491815,
+      "loss": 1.7754,
+      "step": 6406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015701357276669576,
+      "loss": 1.6929,
+      "step": 6407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015694276810576874,
+      "loss": 1.97,
+      "step": 6408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015687197210872934,
+      "loss": 1.9345,
+      "step": 6409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015680118478216866,
+      "loss": 2.1177,
+      "step": 6410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015673040613267718,
+      "loss": 1.8718,
+      "step": 6411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015665963616684475,
+      "loss": 1.9108,
+      "step": 6412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015658887489126017,
+      "loss": 1.9149,
+      "step": 6413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015651812231251157,
+      "loss": 1.8908,
+      "step": 6414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001564473784371862,
+      "loss": 2.1181,
+      "step": 6415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001563766432718704,
+      "loss": 2.0711,
+      "step": 6416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015630591682314993,
+      "loss": 1.9203,
+      "step": 6417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015623519909760953,
+      "loss": 1.8928,
+      "step": 6418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015616449010183333,
+      "loss": 1.7924,
+      "step": 6419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001560937898424045,
+      "loss": 1.8241,
+      "step": 6420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015602309832590534,
+      "loss": 1.9418,
+      "step": 6421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015595241555891752,
+      "loss": 1.796,
+      "step": 6422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015588174154802174,
+      "loss": 1.9122,
+      "step": 6423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015581107629979797,
+      "loss": 1.8059,
+      "step": 6424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015574041982082534,
+      "loss": 2.0307,
+      "step": 6425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001556697721176823,
+      "loss": 2.0977,
+      "step": 6426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015559913319694612,
+      "loss": 1.8911,
+      "step": 6427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015552850306519358,
+      "loss": 1.8338,
+      "step": 6428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015545788172900056,
+      "loss": 1.9367,
+      "step": 6429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015538726919494205,
+      "loss": 1.8818,
+      "step": 6430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015531666546959237,
+      "loss": 1.7233,
+      "step": 6431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015524607055952494,
+      "loss": 1.9799,
+      "step": 6432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015517548447131215,
+      "loss": 1.9522,
+      "step": 6433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015510490721152594,
+      "loss": 1.8543,
+      "step": 6434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015503433878673713,
+      "loss": 1.9841,
+      "step": 6435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015496377920351596,
+      "loss": 1.9533,
+      "step": 6436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015489322846843167,
+      "loss": 1.8056,
+      "step": 6437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015482268658805268,
+      "loss": 1.8607,
+      "step": 6438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015475215356894673,
+      "loss": 1.8511,
+      "step": 6439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015468162941768048,
+      "loss": 1.8553,
+      "step": 6440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015461111414082011,
+      "loss": 2.0485,
+      "step": 6441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015454060774493068,
+      "loss": 1.8292,
+      "step": 6442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015447011023657665,
+      "loss": 1.8044,
+      "step": 6443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015439962162232135,
+      "loss": 1.9961,
+      "step": 6444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015432914190872756,
+      "loss": 2.0058,
+      "step": 6445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015425867110235717,
+      "loss": 2.1223,
+      "step": 6446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015418820920977118,
+      "loss": 1.8478,
+      "step": 6447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015411775623752985,
+      "loss": 1.905,
+      "step": 6448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001540473121921924,
+      "loss": 1.8756,
+      "step": 6449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015397687708031745,
+      "loss": 1.9148,
+      "step": 6450
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.9955353736877441,
+      "eval_runtime": 184.1318,
+      "eval_samples_per_second": 457.303,
+      "eval_steps_per_second": 28.583,
+      "step": 6450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015390645090846274,
+      "loss": 2.0789,
+      "step": 6451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015383603368318512,
+      "loss": 1.7099,
+      "step": 6452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001537656254110406,
+      "loss": 1.94,
+      "step": 6453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015369522609858443,
+      "loss": 2.1478,
+      "step": 6454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015362483575237097,
+      "loss": 1.8506,
+      "step": 6455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001535544543789537,
+      "loss": 1.8616,
+      "step": 6456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015348408198488536,
+      "loss": 1.9453,
+      "step": 6457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015341371857671783,
+      "loss": 1.8574,
+      "step": 6458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015334336416100226,
+      "loss": 2.0783,
+      "step": 6459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015327301874428856,
+      "loss": 1.8799,
+      "step": 6460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001532026823331263,
+      "loss": 2.1486,
+      "step": 6461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015313235493406388,
+      "loss": 2.0014,
+      "step": 6462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015306203655364906,
+      "loss": 1.7887,
+      "step": 6463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015299172719842864,
+      "loss": 1.9747,
+      "step": 6464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015292142687494873,
+      "loss": 2.1193,
+      "step": 6465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015285113558975429,
+      "loss": 2.0701,
+      "step": 6466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001527808533493897,
+      "loss": 1.8803,
+      "step": 6467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015271058016039848,
+      "loss": 2.0221,
+      "step": 6468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015264031602932326,
+      "loss": 1.8243,
+      "step": 6469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015257006096270581,
+      "loss": 1.9373,
+      "step": 6470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001524998149670871,
+      "loss": 1.8867,
+      "step": 6471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015242957804900714,
+      "loss": 1.9251,
+      "step": 6472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015235935021500524,
+      "loss": 1.9893,
+      "step": 6473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015228913147161984,
+      "loss": 1.9497,
+      "step": 6474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015221892182538837,
+      "loss": 1.6699,
+      "step": 6475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001521487212828478,
+      "loss": 1.935,
+      "step": 6476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015207852985053372,
+      "loss": 1.9821,
+      "step": 6477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015200834753498126,
+      "loss": 1.9687,
+      "step": 6478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015193817434272462,
+      "loss": 1.8229,
+      "step": 6479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015186801028029703,
+      "loss": 1.8506,
+      "step": 6480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015179785535423109,
+      "loss": 2.039,
+      "step": 6481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015172770957105842,
+      "loss": 1.8469,
+      "step": 6482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015165757293730958,
+      "loss": 2.0465,
+      "step": 6483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015158744545951466,
+      "loss": 1.958,
+      "step": 6484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001515173271442027,
+      "loss": 1.888,
+      "step": 6485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015144721799790193,
+      "loss": 1.9778,
+      "step": 6486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001513771180271397,
+      "loss": 2.0105,
+      "step": 6487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001513070272384424,
+      "loss": 2.015,
+      "step": 6488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015123694563833585,
+      "loss": 1.8462,
+      "step": 6489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015116687323334465,
+      "loss": 1.705,
+      "step": 6490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001510968100299929,
+      "loss": 1.8021,
+      "step": 6491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001510267560348036,
+      "loss": 1.9592,
+      "step": 6492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001509567112542991,
+      "loss": 1.8156,
+      "step": 6493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015088667569500053,
+      "loss": 1.9126,
+      "step": 6494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015081664936342858,
+      "loss": 2.0168,
+      "step": 6495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001507466322661028,
+      "loss": 1.9312,
+      "step": 6496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015067662440954203,
+      "loss": 1.6593,
+      "step": 6497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001506066258002642,
+      "loss": 1.963,
+      "step": 6498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015053663644478644,
+      "loss": 1.9094,
+      "step": 6499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015046665634962475,
+      "loss": 1.9895,
+      "step": 6500
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.9917577505111694,
+      "eval_runtime": 184.1263,
+      "eval_samples_per_second": 457.316,
+      "eval_steps_per_second": 28.584,
+      "step": 6500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001503966855212946,
+      "loss": 1.9986,
+      "step": 6501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015032672396631056,
+      "loss": 1.9985,
+      "step": 6502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015025677169118606,
+      "loss": 1.9485,
+      "step": 6503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015018682870243403,
+      "loss": 1.9217,
+      "step": 6504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015011689500656623,
+      "loss": 1.8334,
+      "step": 6505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015004697061009371,
+      "loss": 2.001,
+      "step": 6506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001499770555195266,
+      "loss": 1.884,
+      "step": 6507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014990714974137423,
+      "loss": 1.8998,
+      "step": 6508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014983725328214504,
+      "loss": 1.9884,
+      "step": 6509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014976736614834662,
+      "loss": 2.2017,
+      "step": 6510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014969748834648555,
+      "loss": 1.8097,
+      "step": 6511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014962761988306767,
+      "loss": 1.8867,
+      "step": 6512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014955776076459797,
+      "loss": 1.9482,
+      "step": 6513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014948791099758053,
+      "loss": 2.0151,
+      "step": 6514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014941807058851863,
+      "loss": 2.0345,
+      "step": 6515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001493482395439144,
+      "loss": 2.0158,
+      "step": 6516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014927841787026948,
+      "loss": 2.0825,
+      "step": 6517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014920860557408433,
+      "loss": 2.0007,
+      "step": 6518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014913880266185887,
+      "loss": 2.0147,
+      "step": 6519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014906900914009172,
+      "loss": 1.723,
+      "step": 6520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000148999225015281,
+      "loss": 2.0376,
+      "step": 6521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001489294502939238,
+      "loss": 1.9387,
+      "step": 6522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014885968498251623,
+      "loss": 1.9536,
+      "step": 6523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014878992908755372,
+      "loss": 2.1929,
+      "step": 6524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014872018261553071,
+      "loss": 1.9248,
+      "step": 6525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001486504455729408,
+      "loss": 1.8041,
+      "step": 6526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001485807179662768,
+      "loss": 2.1309,
+      "step": 6527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001485109998020304,
+      "loss": 2.0428,
+      "step": 6528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001484412910866926,
+      "loss": 1.9414,
+      "step": 6529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001483715918267534,
+      "loss": 2.0459,
+      "step": 6530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014830190202870213,
+      "loss": 1.9026,
+      "step": 6531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014823222169902718,
+      "loss": 1.8622,
+      "step": 6532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014816255084421572,
+      "loss": 1.9522,
+      "step": 6533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014809288947075443,
+      "loss": 1.797,
+      "step": 6534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014802323758512904,
+      "loss": 1.9321,
+      "step": 6535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014795359519382428,
+      "loss": 1.9523,
+      "step": 6536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014788396230332401,
+      "loss": 2.0875,
+      "step": 6537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014781433892011131,
+      "loss": 2.1918,
+      "step": 6538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014774472505066832,
+      "loss": 1.9155,
+      "step": 6539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001476751207014762,
+      "loss": 2.0214,
+      "step": 6540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014760552587901536,
+      "loss": 1.9116,
+      "step": 6541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014753594058976524,
+      "loss": 1.815,
+      "step": 6542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014746636484020458,
+      "loss": 1.9241,
+      "step": 6543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014739679863681084,
+      "loss": 1.8926,
+      "step": 6544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014732724198606096,
+      "loss": 1.9711,
+      "step": 6545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001472576948944308,
+      "loss": 1.9758,
+      "step": 6546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014718815736839547,
+      "loss": 1.8141,
+      "step": 6547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014711862941442906,
+      "loss": 1.955,
+      "step": 6548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014704911103900488,
+      "loss": 1.9823,
+      "step": 6549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014697960224859514,
+      "loss": 2.1705,
+      "step": 6550
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 1.986474633216858,
+      "eval_runtime": 184.1427,
+      "eval_samples_per_second": 457.276,
+      "eval_steps_per_second": 28.581,
+      "step": 6550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001469101030496714,
+      "loss": 1.9395,
+      "step": 6551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014684061344870424,
+      "loss": 1.9292,
+      "step": 6552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014677113345216327,
+      "loss": 1.9324,
+      "step": 6553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014670166306651733,
+      "loss": 1.8826,
+      "step": 6554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014663220229823437,
+      "loss": 2.1001,
+      "step": 6555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001465627511537812,
+      "loss": 1.8581,
+      "step": 6556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014649330963962398,
+      "loss": 2.029,
+      "step": 6557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000146423877762228,
+      "loss": 1.8045,
+      "step": 6558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014635445552805746,
+      "loss": 2.0816,
+      "step": 6559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014628504294357593,
+      "loss": 2.0579,
+      "step": 6560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014621564001524568,
+      "loss": 1.8083,
+      "step": 6561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001461462467495284,
+      "loss": 1.9824,
+      "step": 6562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014607686315288486,
+      "loss": 1.8326,
+      "step": 6563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001460074892317748,
+      "loss": 1.8035,
+      "step": 6564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014593812499265723,
+      "loss": 2.1142,
+      "step": 6565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014586877044199016,
+      "loss": 1.7807,
+      "step": 6566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014579942558623043,
+      "loss": 1.8663,
+      "step": 6567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001457300904318345,
+      "loss": 1.9048,
+      "step": 6568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014566076498525755,
+      "loss": 2.0249,
+      "step": 6569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014559144925295397,
+      "loss": 1.8562,
+      "step": 6570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014552214324137745,
+      "loss": 1.7787,
+      "step": 6571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014545284695698026,
+      "loss": 1.8391,
+      "step": 6572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014538356040621426,
+      "loss": 1.8949,
+      "step": 6573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014531428359553018,
+      "loss": 1.9887,
+      "step": 6574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014524501653137787,
+      "loss": 1.8214,
+      "step": 6575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001451757592202063,
+      "loss": 1.8578,
+      "step": 6576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014510651166846366,
+      "loss": 1.9558,
+      "step": 6577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014503727388259685,
+      "loss": 2.0467,
+      "step": 6578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001449680458690522,
+      "loss": 1.9979,
+      "step": 6579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000144898827634275,
+      "loss": 2.111,
+      "step": 6580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014482961918470974,
+      "loss": 1.7913,
+      "step": 6581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014476042052679987,
+      "loss": 1.8814,
+      "step": 6582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014469123166698804,
+      "loss": 1.993,
+      "step": 6583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014462205261171586,
+      "loss": 1.9004,
+      "step": 6584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014455288336742404,
+      "loss": 1.9192,
+      "step": 6585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001444837239405525,
+      "loss": 2.1146,
+      "step": 6586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001444145743375402,
+      "loss": 1.8808,
+      "step": 6587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001443454345648252,
+      "loss": 1.8478,
+      "step": 6588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014427630462884438,
+      "loss": 2.0699,
+      "step": 6589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014420718453603414,
+      "loss": 1.9738,
+      "step": 6590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001441380742928297,
+      "loss": 1.967,
+      "step": 6591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014406897390566552,
+      "loss": 2.033,
+      "step": 6592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001439998833809748,
+      "loss": 2.096,
+      "step": 6593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001439308027251902,
+      "loss": 1.9722,
+      "step": 6594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014386173194474343,
+      "loss": 1.9773,
+      "step": 6595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014379267104606497,
+      "loss": 1.902,
+      "step": 6596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014372362003558465,
+      "loss": 1.8804,
+      "step": 6597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001436545789197313,
+      "loss": 1.8456,
+      "step": 6598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014358554770493298,
+      "loss": 1.9819,
+      "step": 6599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001435165263976165,
+      "loss": 1.8735,
+      "step": 6600
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 1.9826250076293945,
+      "eval_runtime": 184.1203,
+      "eval_samples_per_second": 457.331,
+      "eval_steps_per_second": 28.585,
+      "step": 6600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000143447515004208,
+      "loss": 1.9575,
+      "step": 6601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014337851353113262,
+      "loss": 2.0608,
+      "step": 6602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014330952198481469,
+      "loss": 2.073,
+      "step": 6603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014324054037167735,
+      "loss": 1.7793,
+      "step": 6604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014317156869814325,
+      "loss": 2.0468,
+      "step": 6605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001431026069706335,
+      "loss": 1.9617,
+      "step": 6606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014303365519556882,
+      "loss": 1.9959,
+      "step": 6607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014296471337936877,
+      "loss": 2.0337,
+      "step": 6608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014289578152845207,
+      "loss": 1.8844,
+      "step": 6609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014282685964923643,
+      "loss": 1.8571,
+      "step": 6610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014275794774813876,
+      "loss": 2.1448,
+      "step": 6611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001426890458315748,
+      "loss": 1.9608,
+      "step": 6612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014262015390595955,
+      "loss": 1.9583,
+      "step": 6613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001425512719777071,
+      "loss": 1.8728,
+      "step": 6614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001424824000532305,
+      "loss": 1.8025,
+      "step": 6615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014241353813894202,
+      "loss": 1.9805,
+      "step": 6616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014234468624125274,
+      "loss": 1.986,
+      "step": 6617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000142275844366573,
+      "loss": 2.0247,
+      "step": 6618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014220701252131224,
+      "loss": 1.8457,
+      "step": 6619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014213819071187884,
+      "loss": 2.1336,
+      "step": 6620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014206937894468033,
+      "loss": 2.052,
+      "step": 6621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014200057722612335,
+      "loss": 2.0447,
+      "step": 6622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014193178556261342,
+      "loss": 1.7739,
+      "step": 6623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001418630039605552,
+      "loss": 1.8716,
+      "step": 6624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014179423242635257,
+      "loss": 1.952,
+      "step": 6625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014172547096640836,
+      "loss": 1.9193,
+      "step": 6626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014165671958712428,
+      "loss": 1.8259,
+      "step": 6627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014158797829490145,
+      "loss": 2.0415,
+      "step": 6628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014151924709613985,
+      "loss": 2.0126,
+      "step": 6629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014145052599723846,
+      "loss": 2.0791,
+      "step": 6630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001413818150045954,
+      "loss": 1.9797,
+      "step": 6631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014131311412460796,
+      "loss": 2.0764,
+      "step": 6632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001412444233636724,
+      "loss": 1.9082,
+      "step": 6633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014117574272818386,
+      "loss": 1.7092,
+      "step": 6634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001411070722245368,
+      "loss": 1.918,
+      "step": 6635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014103841185912465,
+      "loss": 1.9826,
+      "step": 6636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001409697616383398,
+      "loss": 2.0346,
+      "step": 6637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014090112156857386,
+      "loss": 1.9925,
+      "step": 6638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014083249165621752,
+      "loss": 1.7307,
+      "step": 6639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014076387190766016,
+      "loss": 2.1524,
+      "step": 6640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001406952623292906,
+      "loss": 2.0332,
+      "step": 6641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014062666292749657,
+      "loss": 2.0517,
+      "step": 6642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014055807370866487,
+      "loss": 1.9461,
+      "step": 6643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014048949467918142,
+      "loss": 1.9519,
+      "step": 6644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000140420925845431,
+      "loss": 1.9548,
+      "step": 6645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014035236721379756,
+      "loss": 1.8388,
+      "step": 6646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001402838187906642,
+      "loss": 1.8773,
+      "step": 6647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014021528058241287,
+      "loss": 1.9905,
+      "step": 6648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014014675259542474,
+      "loss": 2.0062,
+      "step": 6649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014007823483608002,
+      "loss": 1.9762,
+      "step": 6650
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.9780467748641968,
+      "eval_runtime": 184.1255,
+      "eval_samples_per_second": 457.319,
+      "eval_steps_per_second": 28.584,
+      "step": 6650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001400097273107577,
+      "loss": 1.8955,
+      "step": 6651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013994123002583614,
+      "loss": 1.8862,
+      "step": 6652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013987274298769264,
+      "loss": 1.9291,
+      "step": 6653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001398042662027035,
+      "loss": 2.0899,
+      "step": 6654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013973579967724423,
+      "loss": 2.0785,
+      "step": 6655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013966734341768906,
+      "loss": 1.9826,
+      "step": 6656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013959889743041148,
+      "loss": 2.0374,
+      "step": 6657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013953046172178414,
+      "loss": 1.8817,
+      "step": 6658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013946203629817855,
+      "loss": 1.8667,
+      "step": 6659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001393936211659652,
+      "loss": 1.9797,
+      "step": 6660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001393252163315138,
+      "loss": 1.7452,
+      "step": 6661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013925682180119313,
+      "loss": 2.0009,
+      "step": 6662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013918843758137074,
+      "loss": 1.8376,
+      "step": 6663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013912006367841346,
+      "loss": 1.842,
+      "step": 6664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001390517000986871,
+      "loss": 1.9072,
+      "step": 6665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013898334684855646,
+      "loss": 1.7738,
+      "step": 6666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013891500393438562,
+      "loss": 2.0511,
+      "step": 6667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013884667136253716,
+      "loss": 1.6393,
+      "step": 6668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013877834913937327,
+      "loss": 1.906,
+      "step": 6669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001387100372712548,
+      "loss": 1.8699,
+      "step": 6670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013864173576454193,
+      "loss": 1.8973,
+      "step": 6671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013857344462559368,
+      "loss": 1.9231,
+      "step": 6672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000138505163860768,
+      "loss": 1.8825,
+      "step": 6673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013843689347642216,
+      "loss": 1.9959,
+      "step": 6674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013836863347891227,
+      "loss": 2.0886,
+      "step": 6675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013830038387459354,
+      "loss": 1.8921,
+      "step": 6676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001382321446698202,
+      "loss": 1.6725,
+      "step": 6677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001381639158709456,
+      "loss": 1.8929,
+      "step": 6678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013809569748432187,
+      "loss": 1.9076,
+      "step": 6679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001380274895163004,
+      "loss": 1.9927,
+      "step": 6680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013795929197323153,
+      "loss": 2.0339,
+      "step": 6681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013789110486146468,
+      "loss": 1.7184,
+      "step": 6682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001378229281873483,
+      "loss": 1.9396,
+      "step": 6683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013775476195722973,
+      "loss": 1.9209,
+      "step": 6684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013768660617745544,
+      "loss": 2.1519,
+      "step": 6685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001376184608543709,
+      "loss": 1.9105,
+      "step": 6686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013755032599432074,
+      "loss": 2.0143,
+      "step": 6687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013748220160364843,
+      "loss": 1.8815,
+      "step": 6688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013741408768869667,
+      "loss": 1.9152,
+      "step": 6689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013734598425580686,
+      "loss": 1.9412,
+      "step": 6690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001372778913113197,
+      "loss": 2.0147,
+      "step": 6691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001372098088615749,
+      "loss": 1.9482,
+      "step": 6692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013714173691291095,
+      "loss": 2.0333,
+      "step": 6693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001370736754716657,
+      "loss": 1.9269,
+      "step": 6694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013700562454417586,
+      "loss": 1.7113,
+      "step": 6695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013693758413677707,
+      "loss": 1.9944,
+      "step": 6696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001368695542558041,
+      "loss": 1.8757,
+      "step": 6697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013680153490759072,
+      "loss": 1.8645,
+      "step": 6698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001367335260984698,
+      "loss": 1.9515,
+      "step": 6699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013666552783477313,
+      "loss": 1.8613,
+      "step": 6700
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.972604513168335,
+      "eval_runtime": 184.1382,
+      "eval_samples_per_second": 457.287,
+      "eval_steps_per_second": 28.582,
+      "step": 6700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013659754012283144,
+      "loss": 1.9926,
+      "step": 6701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001365295629689746,
+      "loss": 1.8469,
+      "step": 6702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013646159637953153,
+      "loss": 1.9252,
+      "step": 6703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001363936403608301,
+      "loss": 1.8914,
+      "step": 6704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013632569491919718,
+      "loss": 1.9037,
+      "step": 6705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001362577600609588,
+      "loss": 2.0294,
+      "step": 6706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013618983579243967,
+      "loss": 1.8622,
+      "step": 6707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013612192211996386,
+      "loss": 2.1118,
+      "step": 6708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013605401904985426,
+      "loss": 1.8628,
+      "step": 6709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001359861265884329,
+      "loss": 1.7337,
+      "step": 6710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013591824474202082,
+      "loss": 2.0795,
+      "step": 6711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013585037351693783,
+      "loss": 1.8816,
+      "step": 6712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000135782512919503,
+      "loss": 1.7957,
+      "step": 6713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013571466295603438,
+      "loss": 1.6487,
+      "step": 6714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013564682363284897,
+      "loss": 2.0546,
+      "step": 6715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001355789949562628,
+      "loss": 1.7175,
+      "step": 6716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000135511176932591,
+      "loss": 1.6197,
+      "step": 6717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001354433695681474,
+      "loss": 1.9187,
+      "step": 6718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013537557286924523,
+      "loss": 2.1326,
+      "step": 6719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013530778684219647,
+      "loss": 1.7723,
+      "step": 6720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013524001149331223,
+      "loss": 1.8176,
+      "step": 6721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013517224682890267,
+      "loss": 1.8538,
+      "step": 6722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001351044928552767,
+      "loss": 2.0964,
+      "step": 6723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013503674957874245,
+      "loss": 1.9317,
+      "step": 6724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013496901700560705,
+      "loss": 1.8107,
+      "step": 6725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013490129514217664,
+      "loss": 2.0625,
+      "step": 6726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013483358399475617,
+      "loss": 2.0636,
+      "step": 6727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013476588356964983,
+      "loss": 1.9434,
+      "step": 6728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001346981938731608,
+      "loss": 2.0278,
+      "step": 6729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013463051491159096,
+      "loss": 1.8698,
+      "step": 6730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013456284669124158,
+      "loss": 1.8945,
+      "step": 6731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013449518921841266,
+      "loss": 2.139,
+      "step": 6732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013442754249940337,
+      "loss": 1.7699,
+      "step": 6733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001343599065405119,
+      "loss": 2.1229,
+      "step": 6734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013429228134803514,
+      "loss": 1.9443,
+      "step": 6735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001342246669282693,
+      "loss": 1.8424,
+      "step": 6736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001341570632875094,
+      "loss": 1.9144,
+      "step": 6737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013408947043204962,
+      "loss": 1.9211,
+      "step": 6738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013402188836818306,
+      "loss": 1.8178,
+      "step": 6739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001339543171022017,
+      "loss": 1.7221,
+      "step": 6740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013388675664039658,
+      "loss": 1.8076,
+      "step": 6741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013381920698905787,
+      "loss": 1.8566,
+      "step": 6742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013375166815447463,
+      "loss": 2.1102,
+      "step": 6743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013368414014293484,
+      "loss": 1.7711,
+      "step": 6744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001336166229607257,
+      "loss": 1.9101,
+      "step": 6745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013354911661413304,
+      "loss": 2.0059,
+      "step": 6746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000133481621109442,
+      "loss": 1.783,
+      "step": 6747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001334141364529366,
+      "loss": 1.9821,
+      "step": 6748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013334666265089984,
+      "loss": 1.7793,
+      "step": 6749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013327919970961376,
+      "loss": 1.801,
+      "step": 6750
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.968043565750122,
+      "eval_runtime": 184.1199,
+      "eval_samples_per_second": 457.333,
+      "eval_steps_per_second": 28.585,
+      "step": 6750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013321174763535927,
+      "loss": 1.8744,
+      "step": 6751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013314430643441633,
+      "loss": 1.7955,
+      "step": 6752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013307687611306397,
+      "loss": 1.9863,
+      "step": 6753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013300945667758012,
+      "loss": 1.9419,
+      "step": 6754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001329420481342417,
+      "loss": 1.7135,
+      "step": 6755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001328746504893248,
+      "loss": 2.1117,
+      "step": 6756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013280726374910405,
+      "loss": 1.8373,
+      "step": 6757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013273988791985346,
+      "loss": 1.8735,
+      "step": 6758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013267252300784604,
+      "loss": 1.9183,
+      "step": 6759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013260516901935347,
+      "loss": 2.0034,
+      "step": 6760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013253782596064662,
+      "loss": 1.7929,
+      "step": 6761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013247049383799544,
+      "loss": 1.9252,
+      "step": 6762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013240317265766858,
+      "loss": 1.6902,
+      "step": 6763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013233586242593386,
+      "loss": 2.0019,
+      "step": 6764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013226856314905818,
+      "loss": 1.9409,
+      "step": 6765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013220127483330713,
+      "loss": 1.8839,
+      "step": 6766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001321339974849456,
+      "loss": 2.0093,
+      "step": 6767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013206673111023715,
+      "loss": 1.7226,
+      "step": 6768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013199947571544452,
+      "loss": 1.7898,
+      "step": 6769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013193223130682935,
+      "loss": 1.8906,
+      "step": 6770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001318649978906523,
+      "loss": 1.9142,
+      "step": 6771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013179777547317307,
+      "loss": 1.8476,
+      "step": 6772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013173056406065025,
+      "loss": 1.881,
+      "step": 6773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013166336365934124,
+      "loss": 1.8528,
+      "step": 6774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013159617427550268,
+      "loss": 2.057,
+      "step": 6775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013152899591539015,
+      "loss": 1.8458,
+      "step": 6776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013146182858525803,
+      "loss": 1.9416,
+      "step": 6777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013139467229136,
+      "loss": 2.0925,
+      "step": 6778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013132752703994818,
+      "loss": 1.924,
+      "step": 6779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001312603928372742,
+      "loss": 1.9614,
+      "step": 6780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013119326968958836,
+      "loss": 1.8989,
+      "step": 6781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013112615760314003,
+      "loss": 1.9272,
+      "step": 6782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013105905658417755,
+      "loss": 1.9284,
+      "step": 6783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001309919666389483,
+      "loss": 1.6894,
+      "step": 6784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013092488777369837,
+      "loss": 2.0294,
+      "step": 6785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013085781999467303,
+      "loss": 1.9435,
+      "step": 6786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013079076330811652,
+      "loss": 1.8517,
+      "step": 6787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013072371772027198,
+      "loss": 2.1242,
+      "step": 6788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013065668323738167,
+      "loss": 1.996,
+      "step": 6789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013058965986568647,
+      "loss": 1.7723,
+      "step": 6790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001305226476114265,
+      "loss": 1.8909,
+      "step": 6791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013045564648084102,
+      "loss": 2.0114,
+      "step": 6792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013038865648016767,
+      "loss": 2.1003,
+      "step": 6793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013032167761564356,
+      "loss": 1.9081,
+      "step": 6794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013025470989350474,
+      "loss": 2.0186,
+      "step": 6795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013018775331998588,
+      "loss": 1.9876,
+      "step": 6796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001301208079013209,
+      "loss": 1.8299,
+      "step": 6797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001300538736437426,
+      "loss": 2.1178,
+      "step": 6798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001299869505534828,
+      "loss": 1.9997,
+      "step": 6799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012992003863677214,
+      "loss": 1.9148,
+      "step": 6800
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 1.9659554958343506,
+      "eval_runtime": 184.0709,
+      "eval_samples_per_second": 457.454,
+      "eval_steps_per_second": 28.592,
+      "step": 6800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012985313789984048,
+      "loss": 1.9922,
+      "step": 6801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012978624834891628,
+      "loss": 2.086,
+      "step": 6802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012971936999022716,
+      "loss": 1.8612,
+      "step": 6803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012965250282999974,
+      "loss": 1.9245,
+      "step": 6804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001295856468744595,
+      "loss": 1.9484,
+      "step": 6805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012951880212983105,
+      "loss": 1.8966,
+      "step": 6806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012945196860233762,
+      "loss": 1.8392,
+      "step": 6807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001293851462982017,
+      "loss": 1.8724,
+      "step": 6808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012931833522364462,
+      "loss": 2.2001,
+      "step": 6809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012925153538488667,
+      "loss": 1.8996,
+      "step": 6810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001291847467881471,
+      "loss": 2.0098,
+      "step": 6811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012911796943964422,
+      "loss": 1.8515,
+      "step": 6812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000129051203345595,
+      "loss": 1.7501,
+      "step": 6813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012898444851221563,
+      "loss": 1.8117,
+      "step": 6814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012891770494572122,
+      "loss": 2.1008,
+      "step": 6815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001288509726523257,
+      "loss": 1.7841,
+      "step": 6816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012878425163824212,
+      "loss": 1.892,
+      "step": 6817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012871754190968245,
+      "loss": 1.5875,
+      "step": 6818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012865084347285727,
+      "loss": 1.9289,
+      "step": 6819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012858415633397664,
+      "loss": 1.9678,
+      "step": 6820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012851748049924922,
+      "loss": 1.8407,
+      "step": 6821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012845081597488285,
+      "loss": 1.8193,
+      "step": 6822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012838416276708401,
+      "loss": 2.0993,
+      "step": 6823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012831752088205832,
+      "loss": 1.9351,
+      "step": 6824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001282508903260104,
+      "loss": 1.76,
+      "step": 6825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001281842711051438,
+      "loss": 2.1107,
+      "step": 6826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012811766322566077,
+      "loss": 1.9188,
+      "step": 6827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012805106669376283,
+      "loss": 2.1628,
+      "step": 6828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012798448151565028,
+      "loss": 1.9288,
+      "step": 6829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001279179076975223,
+      "loss": 2.0284,
+      "step": 6830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012785134524557717,
+      "loss": 1.9696,
+      "step": 6831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012778479416601201,
+      "loss": 2.0173,
+      "step": 6832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012771825446502297,
+      "loss": 2.0682,
+      "step": 6833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001276517261488051,
+      "loss": 1.8463,
+      "step": 6834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012758520922355227,
+      "loss": 1.9656,
+      "step": 6835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012751870369545737,
+      "loss": 1.9243,
+      "step": 6836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012745220957071239,
+      "loss": 1.8328,
+      "step": 6837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012738572685550798,
+      "loss": 1.9327,
+      "step": 6838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012731925555603397,
+      "loss": 2.0084,
+      "step": 6839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012725279567847908,
+      "loss": 1.8942,
+      "step": 6840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001271863472290307,
+      "loss": 2.1134,
+      "step": 6841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001271199102138755,
+      "loss": 1.8034,
+      "step": 6842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001270534846391989,
+      "loss": 2.0994,
+      "step": 6843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012698707051118536,
+      "loss": 1.7479,
+      "step": 6844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001269206678360182,
+      "loss": 1.7988,
+      "step": 6845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012685427661987975,
+      "loss": 1.8923,
+      "step": 6846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001267878968689511,
+      "loss": 1.842,
+      "step": 6847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012672152858941244,
+      "loss": 2.1313,
+      "step": 6848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012665517178744285,
+      "loss": 1.9147,
+      "step": 6849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012658882646922036,
+      "loss": 2.1368,
+      "step": 6850
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 1.9615252017974854,
+      "eval_runtime": 184.166,
+      "eval_samples_per_second": 457.218,
+      "eval_steps_per_second": 28.577,
+      "step": 6850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001265224926409219,
+      "loss": 1.9882,
+      "step": 6851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012645617030872328,
+      "loss": 1.7092,
+      "step": 6852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012638985947879933,
+      "loss": 1.8202,
+      "step": 6853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001263235601573237,
+      "loss": 1.7584,
+      "step": 6854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001262572723504692,
+      "loss": 1.7731,
+      "step": 6855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012619099606440734,
+      "loss": 1.9599,
+      "step": 6856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012612473130530855,
+      "loss": 1.7952,
+      "step": 6857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001260584780793423,
+      "loss": 1.9623,
+      "step": 6858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012599223639267704,
+      "loss": 1.9653,
+      "step": 6859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012592600625147994,
+      "loss": 1.6725,
+      "step": 6860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012585978766191725,
+      "loss": 1.7859,
+      "step": 6861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012579358063015417,
+      "loss": 1.869,
+      "step": 6862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001257273851623546,
+      "loss": 1.9322,
+      "step": 6863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012566120126468168,
+      "loss": 2.1056,
+      "step": 6864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012559502894329721,
+      "loss": 1.9851,
+      "step": 6865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012552886820436208,
+      "loss": 2.0028,
+      "step": 6866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000125462719054036,
+      "loss": 2.0194,
+      "step": 6867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001253965814984778,
+      "loss": 1.9403,
+      "step": 6868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001253304555438448,
+      "loss": 1.7868,
+      "step": 6869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012526434119629367,
+      "loss": 1.8549,
+      "step": 6870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012519823846197982,
+      "loss": 1.9928,
+      "step": 6871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001251321473470576,
+      "loss": 1.9023,
+      "step": 6872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001250660678576802,
+      "loss": 1.9106,
+      "step": 6873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012500000000000006,
+      "loss": 1.917,
+      "step": 6874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012493394378016795,
+      "loss": 1.8633,
+      "step": 6875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012486789920433406,
+      "loss": 1.9171,
+      "step": 6876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001248018662786473,
+      "loss": 1.7448,
+      "step": 6877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012473584500925547,
+      "loss": 2.0024,
+      "step": 6878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012466983540230554,
+      "loss": 2.0829,
+      "step": 6879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012460383746394288,
+      "loss": 2.0895,
+      "step": 6880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012453785120031228,
+      "loss": 1.8771,
+      "step": 6881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012447187661755716,
+      "loss": 1.8941,
+      "step": 6882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012440591372182,
+      "loss": 2.0127,
+      "step": 6883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001243399625192421,
+      "loss": 1.8408,
+      "step": 6884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001242740230159638,
+      "loss": 2.2014,
+      "step": 6885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012420809521812405,
+      "loss": 2.0969,
+      "step": 6886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012414217913186103,
+      "loss": 1.7813,
+      "step": 6887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012407627476331167,
+      "loss": 1.9541,
+      "step": 6888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000124010382118612,
+      "loss": 2.0699,
+      "step": 6889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012394450120389657,
+      "loss": 1.983,
+      "step": 6890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012387863202529924,
+      "loss": 1.9198,
+      "step": 6891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012381277458895263,
+      "loss": 1.8981,
+      "step": 6892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001237469289009881,
+      "loss": 2.111,
+      "step": 6893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012368109496753617,
+      "loss": 1.9757,
+      "step": 6894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012361527279472611,
+      "loss": 1.9033,
+      "step": 6895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012354946238868632,
+      "loss": 1.8288,
+      "step": 6896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001234836637555437,
+      "loss": 2.0273,
+      "step": 6897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012341787690142437,
+      "loss": 1.8543,
+      "step": 6898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012335210183245327,
+      "loss": 1.859,
+      "step": 6899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012328633855475428,
+      "loss": 1.8184,
+      "step": 6900
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.9569963216781616,
+      "eval_runtime": 184.2442,
+      "eval_samples_per_second": 457.024,
+      "eval_steps_per_second": 28.565,
+      "step": 6900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012322058707445012,
+      "loss": 1.8957,
+      "step": 6901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012315484739766253,
+      "loss": 1.7296,
+      "step": 6902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012308911953051184,
+      "loss": 1.9589,
+      "step": 6903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012302340347911767,
+      "loss": 1.9923,
+      "step": 6904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012295769924959828,
+      "loss": 1.9276,
+      "step": 6905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012289200684807098,
+      "loss": 1.8216,
+      "step": 6906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012282632628065197,
+      "loss": 1.9785,
+      "step": 6907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001227606575534561,
+      "loss": 2.0465,
+      "step": 6908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012269500067259748,
+      "loss": 1.9298,
+      "step": 6909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012262935564418887,
+      "loss": 1.7708,
+      "step": 6910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000122563722474342,
+      "loss": 2.1185,
+      "step": 6911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012249810116916757,
+      "loss": 1.8721,
+      "step": 6912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012243249173477514,
+      "loss": 1.6892,
+      "step": 6913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012236689417727294,
+      "loss": 2.0048,
+      "step": 6914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012230130850276845,
+      "loss": 1.8379,
+      "step": 6915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001222357347173678,
+      "loss": 2.1249,
+      "step": 6916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012217017282717616,
+      "loss": 1.811,
+      "step": 6917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012210462283829753,
+      "loss": 2.0399,
+      "step": 6918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001220390847568347,
+      "loss": 1.9105,
+      "step": 6919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001219735585888895,
+      "loss": 1.7329,
+      "step": 6920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001219080443405626,
+      "loss": 2.2956,
+      "step": 6921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012184254201795364,
+      "loss": 1.8994,
+      "step": 6922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001217770516271609,
+      "loss": 1.7969,
+      "step": 6923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012171157317428183,
+      "loss": 1.7407,
+      "step": 6924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012164610666541262,
+      "loss": 2.0247,
+      "step": 6925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012158065210664848,
+      "loss": 1.8248,
+      "step": 6926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012151520950408324,
+      "loss": 1.9443,
+      "step": 6927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012144977886380987,
+      "loss": 1.9179,
+      "step": 6928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012138436019192017,
+      "loss": 1.9917,
+      "step": 6929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012131895349450489,
+      "loss": 1.8303,
+      "step": 6930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012125355877765334,
+      "loss": 1.869,
+      "step": 6931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012118817604745411,
+      "loss": 1.9644,
+      "step": 6932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012112280530999448,
+      "loss": 2.0924,
+      "step": 6933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012105744657136064,
+      "loss": 1.7756,
+      "step": 6934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012099209983763776,
+      "loss": 2.0367,
+      "step": 6935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012092676511490966,
+      "loss": 2.0074,
+      "step": 6936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012086144240925926,
+      "loss": 1.8163,
+      "step": 6937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012079613172676823,
+      "loss": 1.9634,
+      "step": 6938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012073083307351727,
+      "loss": 1.8657,
+      "step": 6939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012066554645558578,
+      "loss": 1.9368,
+      "step": 6940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012060027187905229,
+      "loss": 1.8502,
+      "step": 6941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012053500934999381,
+      "loss": 1.9901,
+      "step": 6942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012046975887448658,
+      "loss": 1.9157,
+      "step": 6943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012040452045860558,
+      "loss": 1.7311,
+      "step": 6944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012033929410842471,
+      "loss": 1.8277,
+      "step": 6945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012027407983001682,
+      "loss": 1.8445,
+      "step": 6946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012020887762945332,
+      "loss": 2.1936,
+      "step": 6947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012014368751280486,
+      "loss": 1.9617,
+      "step": 6948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001200785094861408,
+      "loss": 1.7821,
+      "step": 6949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012001334355552937,
+      "loss": 1.5963,
+      "step": 6950
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.956361174583435,
+      "eval_runtime": 184.5352,
+      "eval_samples_per_second": 456.303,
+      "eval_steps_per_second": 28.52,
+      "step": 6950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011994818972703772,
+      "loss": 2.0542,
+      "step": 6951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011988304800673198,
+      "loss": 1.6719,
+      "step": 6952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011981791840067676,
+      "loss": 1.841,
+      "step": 6953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011975280091493598,
+      "loss": 2.0231,
+      "step": 6954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011968769555557221,
+      "loss": 1.7495,
+      "step": 6955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011962260232864708,
+      "loss": 1.9117,
+      "step": 6956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011955752124022071,
+      "loss": 1.9558,
+      "step": 6957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011949245229635244,
+      "loss": 1.8212,
+      "step": 6958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001194273955031005,
+      "loss": 1.8367,
+      "step": 6959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011936235086652161,
+      "loss": 2.0007,
+      "step": 6960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011929731839267177,
+      "loss": 1.6675,
+      "step": 6961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011923229808760564,
+      "loss": 1.8974,
+      "step": 6962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011916728995737685,
+      "loss": 1.9336,
+      "step": 6963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011910229400803774,
+      "loss": 1.7998,
+      "step": 6964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011903731024563966,
+      "loss": 1.7251,
+      "step": 6965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011897233867623278,
+      "loss": 1.81,
+      "step": 6966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011890737930586613,
+      "loss": 2.086,
+      "step": 6967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001188424321405876,
+      "loss": 1.862,
+      "step": 6968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011877749718644407,
+      "loss": 1.8372,
+      "step": 6969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011871257444948097,
+      "loss": 1.9541,
+      "step": 6970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011864766393574285,
+      "loss": 1.8907,
+      "step": 6971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011858276565127313,
+      "loss": 1.9401,
+      "step": 6972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011851787960211397,
+      "loss": 1.6895,
+      "step": 6973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011845300579430653,
+      "loss": 1.8592,
+      "step": 6974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011838814423389057,
+      "loss": 1.8452,
+      "step": 6975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000118323294926905,
+      "loss": 1.9266,
+      "step": 6976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001182584578793874,
+      "loss": 1.7284,
+      "step": 6977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011819363309737438,
+      "loss": 2.0085,
+      "step": 6978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011812882058690122,
+      "loss": 1.8257,
+      "step": 6979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001180640203540023,
+      "loss": 1.7976,
+      "step": 6980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011799923240471045,
+      "loss": 1.9471,
+      "step": 6981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011793445674505776,
+      "loss": 1.8475,
+      "step": 6982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000117869693381075,
+      "loss": 1.9966,
+      "step": 6983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011780494231879183,
+      "loss": 1.8967,
+      "step": 6984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011774020356423685,
+      "loss": 1.9807,
+      "step": 6985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011767547712343721,
+      "loss": 1.9582,
+      "step": 6986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011761076300241922,
+      "loss": 1.8123,
+      "step": 6987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011754606120720795,
+      "loss": 2.1368,
+      "step": 6988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011748137174382742,
+      "loss": 1.9831,
+      "step": 6989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001174166946183002,
+      "loss": 1.8923,
+      "step": 6990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011735202983664803,
+      "loss": 1.9698,
+      "step": 6991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011728737740489145,
+      "loss": 2.0085,
+      "step": 6992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011722273732904959,
+      "loss": 1.8926,
+      "step": 6993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011715810961514073,
+      "loss": 1.8642,
+      "step": 6994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001170934942691819,
+      "loss": 1.8178,
+      "step": 6995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011702889129718893,
+      "loss": 1.8433,
+      "step": 6996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011696430070517666,
+      "loss": 1.931,
+      "step": 6997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011689972249915845,
+      "loss": 1.831,
+      "step": 6998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011683515668514682,
+      "loss": 1.7406,
+      "step": 6999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011677060326915304,
+      "loss": 1.8879,
+      "step": 7000
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 1.9505616426467896,
+      "eval_runtime": 184.5197,
+      "eval_samples_per_second": 456.341,
+      "eval_steps_per_second": 28.523,
+      "step": 7000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011670606225718719,
+      "loss": 2.0356,
+      "step": 7001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011664153365525829,
+      "loss": 1.9507,
+      "step": 7002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011657701746937399,
+      "loss": 1.9958,
+      "step": 7003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000116512513705541,
+      "loss": 1.9546,
+      "step": 7004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011644802236976482,
+      "loss": 1.9161,
+      "step": 7005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011638354346804972,
+      "loss": 1.9408,
+      "step": 7006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011631907700639893,
+      "loss": 1.9445,
+      "step": 7007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001162546229908145,
+      "loss": 1.8288,
+      "step": 7008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011619018142729712,
+      "loss": 1.9924,
+      "step": 7009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011612575232184655,
+      "loss": 2.0726,
+      "step": 7010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011606133568046135,
+      "loss": 1.9221,
+      "step": 7011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011599693150913887,
+      "loss": 1.9225,
+      "step": 7012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011593253981387542,
+      "loss": 1.815,
+      "step": 7013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011586816060066585,
+      "loss": 2.0805,
+      "step": 7014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011580379387550413,
+      "loss": 2.0171,
+      "step": 7015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011573943964438299,
+      "loss": 1.9627,
+      "step": 7016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000115675097913294,
+      "loss": 1.7455,
+      "step": 7017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011561076868822756,
+      "loss": 2.0621,
+      "step": 7018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011554645197517297,
+      "loss": 1.9614,
+      "step": 7019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011548214778011809,
+      "loss": 1.9706,
+      "step": 7020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011541785610904995,
+      "loss": 1.9997,
+      "step": 7021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011535357696795438,
+      "loss": 1.9965,
+      "step": 7022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011528931036281576,
+      "loss": 1.95,
+      "step": 7023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011522505629961755,
+      "loss": 2.132,
+      "step": 7024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011516081478434209,
+      "loss": 1.9169,
+      "step": 7025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011509658582297025,
+      "loss": 1.8895,
+      "step": 7026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011503236942148207,
+      "loss": 2.1986,
+      "step": 7027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011496816558585621,
+      "loss": 1.8078,
+      "step": 7028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011490397432207026,
+      "loss": 1.7205,
+      "step": 7029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001148397956361007,
+      "loss": 1.7849,
+      "step": 7030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011477562953392253,
+      "loss": 1.7971,
+      "step": 7031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011471147602150991,
+      "loss": 1.852,
+      "step": 7032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011464733510483574,
+      "loss": 1.9736,
+      "step": 7033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011458320678987167,
+      "loss": 1.7631,
+      "step": 7034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001145190910825882,
+      "loss": 1.7007,
+      "step": 7035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011445498798895485,
+      "loss": 1.9159,
+      "step": 7036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011439089751493959,
+      "loss": 1.7887,
+      "step": 7037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011432681966650946,
+      "loss": 1.8766,
+      "step": 7038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011426275444963033,
+      "loss": 2.0723,
+      "step": 7039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011419870187026687,
+      "loss": 1.7612,
+      "step": 7040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011413466193438259,
+      "loss": 1.9608,
+      "step": 7041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011407063464793965,
+      "loss": 1.9042,
+      "step": 7042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011400662001689926,
+      "loss": 1.8853,
+      "step": 7043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011394261804722131,
+      "loss": 1.8174,
+      "step": 7044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011387862874486463,
+      "loss": 1.6854,
+      "step": 7045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011381465211578674,
+      "loss": 1.9876,
+      "step": 7046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011375068816594416,
+      "loss": 1.8407,
+      "step": 7047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011368673690129198,
+      "loss": 1.8752,
+      "step": 7048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011362279832778421,
+      "loss": 1.8995,
+      "step": 7049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011355887245137384,
+      "loss": 1.9633,
+      "step": 7050
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 1.948055624961853,
+      "eval_runtime": 184.1175,
+      "eval_samples_per_second": 457.338,
+      "eval_steps_per_second": 28.585,
+      "step": 7050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011349495927801246,
+      "loss": 1.7881,
+      "step": 7051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001134310588136506,
+      "loss": 2.0901,
+      "step": 7052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011336717106423766,
+      "loss": 1.6723,
+      "step": 7053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001133032960357216,
+      "loss": 1.9313,
+      "step": 7054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011323943373404941,
+      "loss": 2.0438,
+      "step": 7055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011317558416516696,
+      "loss": 1.792,
+      "step": 7056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011311174733501867,
+      "loss": 1.6574,
+      "step": 7057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011304792324954796,
+      "loss": 2.1072,
+      "step": 7058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011298411191469718,
+      "loss": 1.9609,
+      "step": 7059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011292031333640715,
+      "loss": 1.8865,
+      "step": 7060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011285652752061773,
+      "loss": 2.0768,
+      "step": 7061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011279275447326762,
+      "loss": 1.9413,
+      "step": 7062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011272899420029426,
+      "loss": 1.8357,
+      "step": 7063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011266524670763398,
+      "loss": 1.9666,
+      "step": 7064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011260151200122165,
+      "loss": 2.0427,
+      "step": 7065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001125377900869913,
+      "loss": 1.6392,
+      "step": 7066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011247408097087555,
+      "loss": 2.017,
+      "step": 7067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011241038465880596,
+      "loss": 2.0684,
+      "step": 7068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011234670115671286,
+      "loss": 1.8473,
+      "step": 7069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011228303047052524,
+      "loss": 1.9575,
+      "step": 7070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011221937260617107,
+      "loss": 1.8583,
+      "step": 7071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001121557275695771,
+      "loss": 1.8984,
+      "step": 7072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011209209536666881,
+      "loss": 1.87,
+      "step": 7073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001120284760033706,
+      "loss": 1.9454,
+      "step": 7074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011196486948560567,
+      "loss": 2.0746,
+      "step": 7075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001119012758192958,
+      "loss": 1.8542,
+      "step": 7076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011183769501036178,
+      "loss": 1.8528,
+      "step": 7077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011177412706472321,
+      "loss": 1.8622,
+      "step": 7078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011171057198829843,
+      "loss": 1.9348,
+      "step": 7079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011164702978700458,
+      "loss": 1.8925,
+      "step": 7080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011158350046675772,
+      "loss": 1.7635,
+      "step": 7081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011151998403347244,
+      "loss": 1.7749,
+      "step": 7082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011145648049306237,
+      "loss": 2.0801,
+      "step": 7083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011139298985143983,
+      "loss": 1.7194,
+      "step": 7084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011132951211451605,
+      "loss": 1.962,
+      "step": 7085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011126604728820103,
+      "loss": 1.8408,
+      "step": 7086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011120259537840333,
+      "loss": 2.026,
+      "step": 7087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011113915639103061,
+      "loss": 1.843,
+      "step": 7088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011107573033198933,
+      "loss": 1.9753,
+      "step": 7089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011101231720718442,
+      "loss": 1.9839,
+      "step": 7090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011094891702251991,
+      "loss": 2.0179,
+      "step": 7091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011088552978389863,
+      "loss": 1.8582,
+      "step": 7092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011082215549722191,
+      "loss": 2.0256,
+      "step": 7093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011075879416839022,
+      "loss": 1.8361,
+      "step": 7094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001106954458033026,
+      "loss": 1.9902,
+      "step": 7095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000110632110407857,
+      "loss": 1.9213,
+      "step": 7096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011056878798795023,
+      "loss": 2.0177,
+      "step": 7097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011050547854947756,
+      "loss": 1.9535,
+      "step": 7098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001104421820983334,
+      "loss": 1.9244,
+      "step": 7099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011037889864041082,
+      "loss": 1.8162,
+      "step": 7100
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.9433521032333374,
+      "eval_runtime": 184.3738,
+      "eval_samples_per_second": 456.703,
+      "eval_steps_per_second": 28.545,
+      "step": 7100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011031562818160168,
+      "loss": 1.9465,
+      "step": 7101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011025237072779662,
+      "loss": 1.8655,
+      "step": 7102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001101891262848852,
+      "loss": 1.9907,
+      "step": 7103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011012589485875546,
+      "loss": 2.0075,
+      "step": 7104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001100626764552945,
+      "loss": 1.8796,
+      "step": 7105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010999947108038816,
+      "loss": 1.7132,
+      "step": 7106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010993627873992101,
+      "loss": 1.7389,
+      "step": 7107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010987309943977646,
+      "loss": 1.842,
+      "step": 7108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010980993318583677,
+      "loss": 1.9972,
+      "step": 7109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010974677998398266,
+      "loss": 1.7118,
+      "step": 7110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010968363984009402,
+      "loss": 1.9024,
+      "step": 7111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010962051276004933,
+      "loss": 2.1162,
+      "step": 7112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010955739874972593,
+      "loss": 2.0342,
+      "step": 7113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000109494297815,
+      "loss": 1.8324,
+      "step": 7114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010943120996174618,
+      "loss": 1.7788,
+      "step": 7115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001093681351958383,
+      "loss": 1.7802,
+      "step": 7116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010930507352314872,
+      "loss": 2.04,
+      "step": 7117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010924202494954872,
+      "loss": 2.033,
+      "step": 7118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010917898948090832,
+      "loss": 1.7391,
+      "step": 7119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010911596712309618,
+      "loss": 1.8951,
+      "step": 7120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010905295788197991,
+      "loss": 1.9942,
+      "step": 7121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010898996176342594,
+      "loss": 1.8304,
+      "step": 7122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001089269787732992,
+      "loss": 1.8579,
+      "step": 7123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010886400891746373,
+      "loss": 1.9776,
+      "step": 7124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010880105220178221,
+      "loss": 1.9025,
+      "step": 7125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010873810863211595,
+      "loss": 1.6273,
+      "step": 7126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010867517821432524,
+      "loss": 1.8904,
+      "step": 7127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010861226095426907,
+      "loss": 1.9176,
+      "step": 7128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010854935685780526,
+      "loss": 1.6922,
+      "step": 7129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010848646593079028,
+      "loss": 1.8634,
+      "step": 7130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010842358817907963,
+      "loss": 1.9804,
+      "step": 7131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010836072360852717,
+      "loss": 1.885,
+      "step": 7132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010829787222498585,
+      "loss": 1.817,
+      "step": 7133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010823503403430735,
+      "loss": 1.9397,
+      "step": 7134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010817220904234201,
+      "loss": 1.9232,
+      "step": 7135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001081093972549391,
+      "loss": 1.766,
+      "step": 7136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010804659867794662,
+      "loss": 1.8375,
+      "step": 7137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010798381331721108,
+      "loss": 1.8537,
+      "step": 7138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010792104117857812,
+      "loss": 1.9186,
+      "step": 7139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010785828226789196,
+      "loss": 2.1731,
+      "step": 7140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010779553659099564,
+      "loss": 1.9563,
+      "step": 7141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010773280415373108,
+      "loss": 2.0396,
+      "step": 7142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010767008496193861,
+      "loss": 1.851,
+      "step": 7143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010760737902145768,
+      "loss": 1.9755,
+      "step": 7144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010754468633812639,
+      "loss": 1.9307,
+      "step": 7145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001074820069177816,
+      "loss": 1.6715,
+      "step": 7146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010741934076625894,
+      "loss": 1.7666,
+      "step": 7147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010735668788939287,
+      "loss": 1.9174,
+      "step": 7148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010729404829301642,
+      "loss": 1.744,
+      "step": 7149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010723142198296154,
+      "loss": 1.6854,
+      "step": 7150
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.9387941360473633,
+      "eval_runtime": 184.3855,
+      "eval_samples_per_second": 456.674,
+      "eval_steps_per_second": 28.543,
+      "step": 7150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010716880896505898,
+      "loss": 1.9051,
+      "step": 7151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010710620924513812,
+      "loss": 1.7255,
+      "step": 7152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001070436228290273,
+      "loss": 1.9999,
+      "step": 7153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010698104972255329,
+      "loss": 1.9696,
+      "step": 7154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001069184899315419,
+      "loss": 1.8913,
+      "step": 7155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010685594346181776,
+      "loss": 1.9447,
+      "step": 7156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001067934103192039,
+      "loss": 1.7623,
+      "step": 7157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010673089050952242,
+      "loss": 1.9542,
+      "step": 7158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001066683840385942,
+      "loss": 1.882,
+      "step": 7159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010660589091223854,
+      "loss": 1.8589,
+      "step": 7160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010654341113627388,
+      "loss": 1.8656,
+      "step": 7161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010648094471651722,
+      "loss": 1.9277,
+      "step": 7162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010641849165878436,
+      "loss": 1.9163,
+      "step": 7163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010635605196888987,
+      "loss": 1.8442,
+      "step": 7164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010629362565264713,
+      "loss": 1.8216,
+      "step": 7165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010623121271586805,
+      "loss": 1.7906,
+      "step": 7166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010616881316436353,
+      "loss": 2.0067,
+      "step": 7167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001061064270039431,
+      "loss": 2.0569,
+      "step": 7168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010604405424041513,
+      "loss": 2.1607,
+      "step": 7169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010598169487958679,
+      "loss": 1.9922,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010591934892726371,
+      "loss": 2.04,
+      "step": 7171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010585701638925055,
+      "loss": 1.8827,
+      "step": 7172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010579469727135068,
+      "loss": 1.8701,
+      "step": 7173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010573239157936618,
+      "loss": 1.993,
+      "step": 7174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010567009931909785,
+      "loss": 1.8815,
+      "step": 7175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010560782049634538,
+      "loss": 1.8159,
+      "step": 7176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010554555511690692,
+      "loss": 1.7363,
+      "step": 7177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010548330318657968,
+      "loss": 1.6337,
+      "step": 7178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001054210647111594,
+      "loss": 1.8959,
+      "step": 7179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010535883969644072,
+      "loss": 1.8138,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010529662814821705,
+      "loss": 1.8509,
+      "step": 7181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010523443007228028,
+      "loss": 1.8615,
+      "step": 7182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010517224547442131,
+      "loss": 1.7559,
+      "step": 7183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010511007436042965,
+      "loss": 1.9114,
+      "step": 7184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010504791673609368,
+      "loss": 1.9601,
+      "step": 7185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010498577260720049,
+      "loss": 1.8907,
+      "step": 7186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010492364197953569,
+      "loss": 2.0214,
+      "step": 7187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010486152485888395,
+      "loss": 2.0399,
+      "step": 7188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010479942125102859,
+      "loss": 1.7959,
+      "step": 7189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010473733116175146,
+      "loss": 1.788,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010467525459683341,
+      "loss": 2.0521,
+      "step": 7191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010461319156205395,
+      "loss": 2.0027,
+      "step": 7192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010455114206319143,
+      "loss": 1.9353,
+      "step": 7193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001044891061060226,
+      "loss": 2.0315,
+      "step": 7194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010442708369632334,
+      "loss": 1.9087,
+      "step": 7195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010436507483986803,
+      "loss": 1.8734,
+      "step": 7196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010430307954242993,
+      "loss": 1.7993,
+      "step": 7197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010424109780978103,
+      "loss": 2.085,
+      "step": 7198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010417912964769183,
+      "loss": 1.9289,
+      "step": 7199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010411717506193184,
+      "loss": 2.1203,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 1.9358806610107422,
+      "eval_runtime": 184.2515,
+      "eval_samples_per_second": 457.006,
+      "eval_steps_per_second": 28.564,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010405523405826916,
+      "loss": 1.8775,
+      "step": 7201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010399330664247075,
+      "loss": 1.7846,
+      "step": 7202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010393139282030214,
+      "loss": 1.9146,
+      "step": 7203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010386949259752784,
+      "loss": 1.9934,
+      "step": 7204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001038076059799107,
+      "loss": 1.8047,
+      "step": 7205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010374573297321269,
+      "loss": 1.9659,
+      "step": 7206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010368387358319428,
+      "loss": 2.0907,
+      "step": 7207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001036220278156148,
+      "loss": 1.8683,
+      "step": 7208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010356019567623237,
+      "loss": 1.922,
+      "step": 7209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010349837717080349,
+      "loss": 1.907,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010343657230508377,
+      "loss": 1.8421,
+      "step": 7211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010337478108482742,
+      "loss": 1.917,
+      "step": 7212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010331300351578737,
+      "loss": 1.9536,
+      "step": 7213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010325123960371526,
+      "loss": 2.0585,
+      "step": 7214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010318948935436159,
+      "loss": 1.7957,
+      "step": 7215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001031277527734753,
+      "loss": 1.6168,
+      "step": 7216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010306602986680433,
+      "loss": 1.8184,
+      "step": 7217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010300432064009526,
+      "loss": 1.7864,
+      "step": 7218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010294262509909346,
+      "loss": 1.9645,
+      "step": 7219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010288094324954278,
+      "loss": 1.8991,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010281927509718609,
+      "loss": 1.9575,
+      "step": 7221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010275762064776492,
+      "loss": 1.732,
+      "step": 7222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010269597990701932,
+      "loss": 1.8671,
+      "step": 7223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010263435288068829,
+      "loss": 1.8723,
+      "step": 7224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001025727395745095,
+      "loss": 1.8238,
+      "step": 7225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010251113999421935,
+      "loss": 1.9242,
+      "step": 7226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010244955414555282,
+      "loss": 1.8522,
+      "step": 7227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001023879820342438,
+      "loss": 1.9565,
+      "step": 7228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001023264236660248,
+      "loss": 1.8657,
+      "step": 7229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010226487904662707,
+      "loss": 1.9615,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010220334818178062,
+      "loss": 1.8374,
+      "step": 7231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010214183107721422,
+      "loss": 1.863,
+      "step": 7232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010208032773865512,
+      "loss": 1.8928,
+      "step": 7233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001020188381718295,
+      "loss": 1.9438,
+      "step": 7234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010195736238246223,
+      "loss": 1.9064,
+      "step": 7235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010189590037627689,
+      "loss": 1.9432,
+      "step": 7236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010183445215899583,
+      "loss": 1.8852,
+      "step": 7237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010177301773633993,
+      "loss": 1.9135,
+      "step": 7238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010171159711402892,
+      "loss": 1.981,
+      "step": 7239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010165019029778128,
+      "loss": 1.9144,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010158879729331414,
+      "loss": 1.9173,
+      "step": 7241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010152741810634334,
+      "loss": 2.2308,
+      "step": 7242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010146605274258356,
+      "loss": 1.8129,
+      "step": 7243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010140470120774792,
+      "loss": 1.9358,
+      "step": 7244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010134336350754849,
+      "loss": 1.8051,
+      "step": 7245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000101282039647696,
+      "loss": 1.8054,
+      "step": 7246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010122072963389983,
+      "loss": 2.061,
+      "step": 7247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010115943347186826,
+      "loss": 1.9343,
+      "step": 7248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010109815116730792,
+      "loss": 2.0501,
+      "step": 7249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010103688272592446,
+      "loss": 1.7929,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 1.9337921142578125,
+      "eval_runtime": 184.1356,
+      "eval_samples_per_second": 457.294,
+      "eval_steps_per_second": 28.582,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010097562815342216,
+      "loss": 1.7316,
+      "step": 7251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010091438745550402,
+      "loss": 1.906,
+      "step": 7252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010085316063787162,
+      "loss": 1.9745,
+      "step": 7253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010079194770622538,
+      "loss": 1.8614,
+      "step": 7254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010073074866626442,
+      "loss": 1.9198,
+      "step": 7255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010066956352368661,
+      "loss": 1.809,
+      "step": 7256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010060839228418834,
+      "loss": 1.7749,
+      "step": 7257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010054723495346482,
+      "loss": 1.8647,
+      "step": 7258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010048609153721002,
+      "loss": 2.0264,
+      "step": 7259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010042496204111665,
+      "loss": 1.9137,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010036384647087587,
+      "loss": 1.8404,
+      "step": 7261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010030274483217778,
+      "loss": 1.718,
+      "step": 7262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001002416571307111,
+      "loss": 1.9297,
+      "step": 7263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010018058337216326,
+      "loss": 1.9228,
+      "step": 7264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010011952356222054,
+      "loss": 1.8906,
+      "step": 7265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010005847770656756,
+      "loss": 1.8852,
+      "step": 7266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.999744581088793e-05,
+      "loss": 1.8645,
+      "step": 7267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.993642788086393e-05,
+      "loss": 1.8798,
+      "step": 7268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.98754239221765e-05,
+      "loss": 2.0519,
+      "step": 7269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.981443394050524e-05,
+      "loss": 1.7633,
+      "step": 7270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.975345794152863e-05,
+      "loss": 1.9562,
+      "step": 7271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.969249593092348e-05,
+      "loss": 1.7254,
+      "step": 7272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.96315479143656e-05,
+      "loss": 2.0989,
+      "step": 7273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.957061389752947e-05,
+      "loss": 1.9605,
+      "step": 7274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.950969388608818e-05,
+      "loss": 2.0283,
+      "step": 7275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.944878788571368e-05,
+      "loss": 1.9398,
+      "step": 7276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.938789590207626e-05,
+      "loss": 1.7956,
+      "step": 7277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.932701794084525e-05,
+      "loss": 1.881,
+      "step": 7278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.926615400768857e-05,
+      "loss": 1.887,
+      "step": 7279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.920530410827281e-05,
+      "loss": 1.9423,
+      "step": 7280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.914446824826325e-05,
+      "loss": 1.9152,
+      "step": 7281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.908364643332398e-05,
+      "loss": 2.0402,
+      "step": 7282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.902283866911751e-05,
+      "loss": 1.7969,
+      "step": 7283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.896204496130531e-05,
+      "loss": 1.9843,
+      "step": 7284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.890126531554741e-05,
+      "loss": 1.948,
+      "step": 7285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.884049973750268e-05,
+      "loss": 1.8851,
+      "step": 7286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.877974823282834e-05,
+      "loss": 1.7254,
+      "step": 7287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.871901080718069e-05,
+      "loss": 1.9337,
+      "step": 7288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.865828746621459e-05,
+      "loss": 1.6258,
+      "step": 7289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.859757821558338e-05,
+      "loss": 2.1351,
+      "step": 7290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.853688306093936e-05,
+      "loss": 1.8207,
+      "step": 7291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.847620200793342e-05,
+      "loss": 1.8962,
+      "step": 7292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.841553506221524e-05,
+      "loss": 1.8754,
+      "step": 7293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.835488222943285e-05,
+      "loss": 1.9279,
+      "step": 7294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.82942435152333e-05,
+      "loss": 1.7415,
+      "step": 7295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.823361892526228e-05,
+      "loss": 1.9204,
+      "step": 7296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.817300846516406e-05,
+      "loss": 1.9387,
+      "step": 7297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.811241214058167e-05,
+      "loss": 1.9673,
+      "step": 7298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.805182995715683e-05,
+      "loss": 1.7812,
+      "step": 7299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.799126192052978e-05,
+      "loss": 1.8433,
+      "step": 7300
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.929073452949524,
+      "eval_runtime": 184.0934,
+      "eval_samples_per_second": 457.398,
+      "eval_steps_per_second": 28.589,
+      "step": 7300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.793070803633969e-05,
+      "loss": 1.6736,
+      "step": 7301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.787016831022422e-05,
+      "loss": 1.8236,
+      "step": 7302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.780964274781984e-05,
+      "loss": 2.1654,
+      "step": 7303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.77491313547617e-05,
+      "loss": 1.8768,
+      "step": 7304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.768863413668343e-05,
+      "loss": 1.7952,
+      "step": 7305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.76281510992176e-05,
+      "loss": 1.8862,
+      "step": 7306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.75676822479953e-05,
+      "loss": 1.9237,
+      "step": 7307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.750722758864636e-05,
+      "loss": 1.9125,
+      "step": 7308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.74467871267993e-05,
+      "loss": 1.8566,
+      "step": 7309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.738636086808137e-05,
+      "loss": 1.8904,
+      "step": 7310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.732594881811827e-05,
+      "loss": 1.8874,
+      "step": 7311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.726555098253457e-05,
+      "loss": 1.8144,
+      "step": 7312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.720516736695348e-05,
+      "loss": 1.9358,
+      "step": 7313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.714479797699693e-05,
+      "loss": 1.6989,
+      "step": 7314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.708444281828547e-05,
+      "loss": 1.7762,
+      "step": 7315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.702410189643837e-05,
+      "loss": 1.9659,
+      "step": 7316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.696377521707342e-05,
+      "loss": 1.7818,
+      "step": 7317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.690346278580725e-05,
+      "loss": 2.1028,
+      "step": 7318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.684316460825523e-05,
+      "loss": 2.1038,
+      "step": 7319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.678288069003111e-05,
+      "loss": 1.7831,
+      "step": 7320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.672261103674754e-05,
+      "loss": 1.9565,
+      "step": 7321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.666235565401593e-05,
+      "loss": 1.7472,
+      "step": 7322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.660211454744605e-05,
+      "loss": 1.9118,
+      "step": 7323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.654188772264655e-05,
+      "loss": 2.1012,
+      "step": 7324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.648167518522477e-05,
+      "loss": 2.0448,
+      "step": 7325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.642147694078662e-05,
+      "loss": 2.1071,
+      "step": 7326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.636129299493685e-05,
+      "loss": 2.1062,
+      "step": 7327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.630112335327856e-05,
+      "loss": 1.9503,
+      "step": 7328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.62409680214138e-05,
+      "loss": 1.7387,
+      "step": 7329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.618082700494318e-05,
+      "loss": 1.8356,
+      "step": 7330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.612070030946604e-05,
+      "loss": 2.0064,
+      "step": 7331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.60605879405804e-05,
+      "loss": 1.8958,
+      "step": 7332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.600048990388271e-05,
+      "loss": 1.8288,
+      "step": 7333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.594040620496838e-05,
+      "loss": 1.8148,
+      "step": 7334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.588033684943135e-05,
+      "loss": 1.8628,
+      "step": 7335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.582028184286423e-05,
+      "loss": 2.0237,
+      "step": 7336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.576024119085833e-05,
+      "loss": 1.9464,
+      "step": 7337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.570021489900368e-05,
+      "loss": 1.9471,
+      "step": 7338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.564020297288875e-05,
+      "loss": 1.8976,
+      "step": 7339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.558020541810087e-05,
+      "loss": 1.8525,
+      "step": 7340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.552022224022597e-05,
+      "loss": 1.7806,
+      "step": 7341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.546025344484868e-05,
+      "loss": 1.9365,
+      "step": 7342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.540029903755224e-05,
+      "loss": 2.035,
+      "step": 7343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.534035902391869e-05,
+      "loss": 1.8881,
+      "step": 7344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.528043340952841e-05,
+      "loss": 2.233,
+      "step": 7345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.522052219996072e-05,
+      "loss": 1.9205,
+      "step": 7346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.516062540079356e-05,
+      "loss": 1.9094,
+      "step": 7347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.510074301760347e-05,
+      "loss": 1.748,
+      "step": 7348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.504087505596573e-05,
+      "loss": 1.9775,
+      "step": 7349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.498102152145407e-05,
+      "loss": 2.0861,
+      "step": 7350
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.9252119064331055,
+      "eval_runtime": 184.0646,
+      "eval_samples_per_second": 457.47,
+      "eval_steps_per_second": 28.593,
+      "step": 7350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.49211824196411e-05,
+      "loss": 1.9452,
+      "step": 7351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.486135775609809e-05,
+      "loss": 1.9062,
+      "step": 7352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.480154753639473e-05,
+      "loss": 1.8183,
+      "step": 7353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.474175176609956e-05,
+      "loss": 1.9025,
+      "step": 7354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.468197045077975e-05,
+      "loss": 2.0308,
+      "step": 7355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.462220359600124e-05,
+      "loss": 2.0322,
+      "step": 7356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.456245120732826e-05,
+      "loss": 1.8295,
+      "step": 7357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.450271329032403e-05,
+      "loss": 1.8516,
+      "step": 7358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.444298985055031e-05,
+      "loss": 1.8614,
+      "step": 7359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.438328089356762e-05,
+      "loss": 1.9692,
+      "step": 7360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.432358642493486e-05,
+      "loss": 2.0381,
+      "step": 7361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.42639064502098e-05,
+      "loss": 1.7546,
+      "step": 7362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.420424097494884e-05,
+      "loss": 1.7856,
+      "step": 7363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.414459000470701e-05,
+      "loss": 2.0326,
+      "step": 7364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.408495354503799e-05,
+      "loss": 1.7767,
+      "step": 7365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.402533160149416e-05,
+      "loss": 1.8128,
+      "step": 7366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.396572417962631e-05,
+      "loss": 1.8509,
+      "step": 7367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.390613128498418e-05,
+      "loss": 2.0127,
+      "step": 7368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.384655292311601e-05,
+      "loss": 1.8089,
+      "step": 7369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.378698909956867e-05,
+      "loss": 1.964,
+      "step": 7370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.372743981988782e-05,
+      "loss": 1.8131,
+      "step": 7371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.366790508961767e-05,
+      "loss": 1.8765,
+      "step": 7372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.360838491430093e-05,
+      "loss": 1.7783,
+      "step": 7373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.354887929947914e-05,
+      "loss": 1.7363,
+      "step": 7374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.34893882506925e-05,
+      "loss": 2.0489,
+      "step": 7375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.342991177347976e-05,
+      "loss": 2.0536,
+      "step": 7376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.337044987337843e-05,
+      "loss": 1.9517,
+      "step": 7377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.331100255592437e-05,
+      "loss": 2.1103,
+      "step": 7378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.325156982665245e-05,
+      "loss": 1.9823,
+      "step": 7379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.319215169109599e-05,
+      "loss": 1.9156,
+      "step": 7380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.313274815478697e-05,
+      "loss": 1.8433,
+      "step": 7381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.307335922325605e-05,
+      "loss": 1.8148,
+      "step": 7382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.301398490203256e-05,
+      "loss": 2.0212,
+      "step": 7383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.295462519664427e-05,
+      "loss": 1.8615,
+      "step": 7384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.28952801126178e-05,
+      "loss": 1.89,
+      "step": 7385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.283594965547846e-05,
+      "loss": 1.7196,
+      "step": 7386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.277663383074988e-05,
+      "loss": 1.9291,
+      "step": 7387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.271733264395463e-05,
+      "loss": 1.8851,
+      "step": 7388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.265804610061388e-05,
+      "loss": 1.7554,
+      "step": 7389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.25987742062472e-05,
+      "loss": 1.9801,
+      "step": 7390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.253951696637311e-05,
+      "loss": 2.0747,
+      "step": 7391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.248027438650855e-05,
+      "loss": 1.9157,
+      "step": 7392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.24210464721692e-05,
+      "loss": 2.0264,
+      "step": 7393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.236183322886945e-05,
+      "loss": 1.8878,
+      "step": 7394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.230263466212202e-05,
+      "loss": 1.9297,
+      "step": 7395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.224345077743856e-05,
+      "loss": 1.8464,
+      "step": 7396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.218428158032924e-05,
+      "loss": 1.8697,
+      "step": 7397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.212512707630288e-05,
+      "loss": 1.9038,
+      "step": 7398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.206598727086695e-05,
+      "loss": 1.7053,
+      "step": 7399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.200686216952761e-05,
+      "loss": 2.0828,
+      "step": 7400
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.9235292673110962,
+      "eval_runtime": 184.0923,
+      "eval_samples_per_second": 457.401,
+      "eval_steps_per_second": 28.589,
+      "step": 7400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.194775177778939e-05,
+      "loss": 1.8963,
+      "step": 7401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.18886561011557e-05,
+      "loss": 1.8002,
+      "step": 7402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.182957514512855e-05,
+      "loss": 1.8376,
+      "step": 7403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.177050891520853e-05,
+      "loss": 2.0118,
+      "step": 7404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.171145741689493e-05,
+      "loss": 1.8372,
+      "step": 7405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.165242065568547e-05,
+      "loss": 2.0058,
+      "step": 7406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.159339863707669e-05,
+      "loss": 1.9715,
+      "step": 7407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.153439136656374e-05,
+      "loss": 2.0201,
+      "step": 7408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.147539884964037e-05,
+      "loss": 1.8742,
+      "step": 7409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.14164210917989e-05,
+      "loss": 1.9381,
+      "step": 7410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.135745809853041e-05,
+      "loss": 1.8785,
+      "step": 7411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.12985098753244e-05,
+      "loss": 1.9155,
+      "step": 7412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.123957642766917e-05,
+      "loss": 1.9532,
+      "step": 7413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.118065776105158e-05,
+      "loss": 2.0557,
+      "step": 7414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.112175388095714e-05,
+      "loss": 1.785,
+      "step": 7415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.106286479287001e-05,
+      "loss": 1.7138,
+      "step": 7416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.100399050227282e-05,
+      "loss": 1.9985,
+      "step": 7417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.094513101464697e-05,
+      "loss": 1.8139,
+      "step": 7418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.088628633547255e-05,
+      "loss": 2.0321,
+      "step": 7419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.082745647022797e-05,
+      "loss": 2.1033,
+      "step": 7420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.076864142439057e-05,
+      "loss": 2.3009,
+      "step": 7421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.070984120343626e-05,
+      "loss": 2.004,
+      "step": 7422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.065105581283933e-05,
+      "loss": 1.7473,
+      "step": 7423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.059228525807294e-05,
+      "loss": 1.7764,
+      "step": 7424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.053352954460883e-05,
+      "loss": 1.8661,
+      "step": 7425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.047478867791731e-05,
+      "loss": 1.958,
+      "step": 7426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.041606266346731e-05,
+      "loss": 1.7931,
+      "step": 7427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.035735150672644e-05,
+      "loss": 1.9055,
+      "step": 7428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.029865521316074e-05,
+      "loss": 1.7961,
+      "step": 7429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.02399737882351e-05,
+      "loss": 1.7706,
+      "step": 7430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.018130723741286e-05,
+      "loss": 1.8034,
+      "step": 7431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.012265556615609e-05,
+      "loss": 2.0546,
+      "step": 7432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.006401877992549e-05,
+      "loss": 1.7597,
+      "step": 7433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.000539688418016e-05,
+      "loss": 1.8615,
+      "step": 7434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.994678988437802e-05,
+      "loss": 1.9495,
+      "step": 7435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.988819778597556e-05,
+      "loss": 1.9645,
+      "step": 7436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.982962059442787e-05,
+      "loss": 1.8066,
+      "step": 7437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.977105831518865e-05,
+      "loss": 1.8322,
+      "step": 7438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.971251095371025e-05,
+      "loss": 1.8745,
+      "step": 7439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.96539785154435e-05,
+      "loss": 1.7559,
+      "step": 7440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.959546100583795e-05,
+      "loss": 2.0155,
+      "step": 7441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.95369584303418e-05,
+      "loss": 1.9595,
+      "step": 7442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.947847079440177e-05,
+      "loss": 1.9235,
+      "step": 7443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.94199981034633e-05,
+      "loss": 1.7437,
+      "step": 7444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.936154036297023e-05,
+      "loss": 1.7992,
+      "step": 7445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.930309757836516e-05,
+      "loss": 1.7381,
+      "step": 7446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.924466975508935e-05,
+      "loss": 1.7913,
+      "step": 7447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.918625689858254e-05,
+      "loss": 1.8554,
+      "step": 7448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.912785901428322e-05,
+      "loss": 1.9669,
+      "step": 7449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.906947610762825e-05,
+      "loss": 1.6605,
+      "step": 7450
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.920853614807129,
+      "eval_runtime": 184.1325,
+      "eval_samples_per_second": 457.301,
+      "eval_steps_per_second": 28.583,
+      "step": 7450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.901110818405327e-05,
+      "loss": 1.8203,
+      "step": 7451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.895275524899268e-05,
+      "loss": 1.7018,
+      "step": 7452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.889441730787905e-05,
+      "loss": 1.962,
+      "step": 7453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.883609436614393e-05,
+      "loss": 2.0563,
+      "step": 7454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.87777864292173e-05,
+      "loss": 1.9084,
+      "step": 7455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.871949350252795e-05,
+      "loss": 1.7576,
+      "step": 7456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.866121559150286e-05,
+      "loss": 1.8524,
+      "step": 7457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.860295270156804e-05,
+      "loss": 1.8867,
+      "step": 7458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.854470483814783e-05,
+      "loss": 1.9234,
+      "step": 7459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.848647200666535e-05,
+      "loss": 1.9415,
+      "step": 7460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.842825421254231e-05,
+      "loss": 2.1308,
+      "step": 7461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.837005146119873e-05,
+      "loss": 2.0343,
+      "step": 7462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.831186375805356e-05,
+      "loss": 1.8636,
+      "step": 7463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.825369110852427e-05,
+      "loss": 2.1237,
+      "step": 7464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.819553351802684e-05,
+      "loss": 2.125,
+      "step": 7465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.813739099197595e-05,
+      "loss": 2.0609,
+      "step": 7466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.80792635357849e-05,
+      "loss": 2.078,
+      "step": 7467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.802115115486533e-05,
+      "loss": 1.7266,
+      "step": 7468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.79630538546278e-05,
+      "loss": 1.84,
+      "step": 7469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.790497164048125e-05,
+      "loss": 1.7318,
+      "step": 7470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.784690451783337e-05,
+      "loss": 1.8227,
+      "step": 7471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.778885249209042e-05,
+      "loss": 1.8956,
+      "step": 7472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.773081556865706e-05,
+      "loss": 1.7459,
+      "step": 7473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.767279375293671e-05,
+      "loss": 1.9872,
+      "step": 7474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.761478705033146e-05,
+      "loss": 1.7623,
+      "step": 7475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.755679546624181e-05,
+      "loss": 1.9374,
+      "step": 7476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.7498819006067e-05,
+      "loss": 1.9083,
+      "step": 7477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.744085767520485e-05,
+      "loss": 1.7853,
+      "step": 7478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.738291147905156e-05,
+      "loss": 1.9513,
+      "step": 7479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.732498042300216e-05,
+      "loss": 1.8194,
+      "step": 7480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.72670645124502e-05,
+      "loss": 1.9913,
+      "step": 7481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.720916375278782e-05,
+      "loss": 1.7393,
+      "step": 7482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.715127814940582e-05,
+      "loss": 1.8289,
+      "step": 7483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.709340770769333e-05,
+      "loss": 1.986,
+      "step": 7484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.703555243303834e-05,
+      "loss": 1.8396,
+      "step": 7485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.697771233082744e-05,
+      "loss": 1.5903,
+      "step": 7486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.691988740644551e-05,
+      "loss": 1.8005,
+      "step": 7487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.686207766527632e-05,
+      "loss": 1.7484,
+      "step": 7488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.680428311270219e-05,
+      "loss": 2.0715,
+      "step": 7489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.674650375410378e-05,
+      "loss": 1.7143,
+      "step": 7490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.66887395948606e-05,
+      "loss": 1.8586,
+      "step": 7491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.663099064035065e-05,
+      "loss": 1.9012,
+      "step": 7492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.657325689595053e-05,
+      "loss": 1.8414,
+      "step": 7493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.651553836703541e-05,
+      "loss": 1.8149,
+      "step": 7494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.645783505897908e-05,
+      "loss": 1.936,
+      "step": 7495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.640014697715381e-05,
+      "loss": 1.7458,
+      "step": 7496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.634247412693053e-05,
+      "loss": 2.0056,
+      "step": 7497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.628481651367875e-05,
+      "loss": 2.0132,
+      "step": 7498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.622717414276657e-05,
+      "loss": 1.9358,
+      "step": 7499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.616954701956073e-05,
+      "loss": 1.6951,
+      "step": 7500
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 1.9171736240386963,
+      "eval_runtime": 184.0753,
+      "eval_samples_per_second": 457.443,
+      "eval_steps_per_second": 28.592,
+      "step": 7500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.611193514942634e-05,
+      "loss": 2.0213,
+      "step": 7501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.605433853772726e-05,
+      "loss": 1.6484,
+      "step": 7502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.599675718982594e-05,
+      "loss": 1.9067,
+      "step": 7503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.59391911110833e-05,
+      "loss": 1.8194,
+      "step": 7504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.5881640306859e-05,
+      "loss": 1.7761,
+      "step": 7505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.582410478251118e-05,
+      "loss": 1.9162,
+      "step": 7506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.576658454339643e-05,
+      "loss": 1.7552,
+      "step": 7507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.57090795948701e-05,
+      "loss": 1.8643,
+      "step": 7508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.565158994228609e-05,
+      "loss": 1.9638,
+      "step": 7509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.559411559099681e-05,
+      "loss": 1.8315,
+      "step": 7510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.553665654635342e-05,
+      "loss": 1.9781,
+      "step": 7511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.54792128137053e-05,
+      "loss": 1.9438,
+      "step": 7512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.542178439840073e-05,
+      "loss": 1.8242,
+      "step": 7513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.536437130578647e-05,
+      "loss": 1.9538,
+      "step": 7514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.530697354120781e-05,
+      "loss": 1.944,
+      "step": 7515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.524959111000872e-05,
+      "loss": 1.8192,
+      "step": 7516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.519222401753151e-05,
+      "loss": 1.9243,
+      "step": 7517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.513487226911732e-05,
+      "loss": 1.9727,
+      "step": 7518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.507753587010583e-05,
+      "loss": 2.0389,
+      "step": 7519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.502021482583503e-05,
+      "loss": 1.9978,
+      "step": 7520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.496290914164176e-05,
+      "loss": 2.1967,
+      "step": 7521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.490561882286135e-05,
+      "loss": 2.0283,
+      "step": 7522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.484834387482775e-05,
+      "loss": 1.9963,
+      "step": 7523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.47910843028733e-05,
+      "loss": 1.9287,
+      "step": 7524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.473384011232907e-05,
+      "loss": 1.978,
+      "step": 7525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.467661130852463e-05,
+      "loss": 1.6213,
+      "step": 7526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.461939789678821e-05,
+      "loss": 1.9261,
+      "step": 7527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.456219988244657e-05,
+      "loss": 1.8184,
+      "step": 7528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.450501727082486e-05,
+      "loss": 1.9065,
+      "step": 7529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.444785006724698e-05,
+      "loss": 1.9834,
+      "step": 7530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.43906982770354e-05,
+      "loss": 2.0556,
+      "step": 7531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.433356190551111e-05,
+      "loss": 1.9896,
+      "step": 7532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.427644095799366e-05,
+      "loss": 1.9935,
+      "step": 7533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.421933543980126e-05,
+      "loss": 1.9511,
+      "step": 7534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.416224535625038e-05,
+      "loss": 1.7829,
+      "step": 7535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.410517071265641e-05,
+      "loss": 1.8968,
+      "step": 7536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.404811151433312e-05,
+      "loss": 1.9551,
+      "step": 7537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.39910677665929e-05,
+      "loss": 1.9333,
+      "step": 7538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.393403947474675e-05,
+      "loss": 2.0818,
+      "step": 7539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.387702664410402e-05,
+      "loss": 2.0576,
+      "step": 7540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.382002927997281e-05,
+      "loss": 1.966,
+      "step": 7541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.376304738765977e-05,
+      "loss": 1.6872,
+      "step": 7542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.370608097247004e-05,
+      "loss": 1.8606,
+      "step": 7543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.36491300397074e-05,
+      "loss": 1.8852,
+      "step": 7544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.359219459467415e-05,
+      "loss": 1.9918,
+      "step": 7545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.353527464267105e-05,
+      "loss": 1.8698,
+      "step": 7546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.347837018899754e-05,
+      "loss": 1.9461,
+      "step": 7547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.342148123895157e-05,
+      "loss": 1.8793,
+      "step": 7548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.336460779782982e-05,
+      "loss": 1.9974,
+      "step": 7549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.330774987092712e-05,
+      "loss": 1.9173,
+      "step": 7550
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 1.9149448871612549,
+      "eval_runtime": 184.0487,
+      "eval_samples_per_second": 457.509,
+      "eval_steps_per_second": 28.596,
+      "step": 7550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.325090746353722e-05,
+      "loss": 1.9411,
+      "step": 7551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.319408058095237e-05,
+      "loss": 1.834,
+      "step": 7552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.313726922846319e-05,
+      "loss": 2.0185,
+      "step": 7553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.3080473411359e-05,
+      "loss": 1.9051,
+      "step": 7554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.302369313492766e-05,
+      "loss": 1.8517,
+      "step": 7555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.29669284044557e-05,
+      "loss": 1.6965,
+      "step": 7556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.291017922522787e-05,
+      "loss": 1.7456,
+      "step": 7557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.285344560252778e-05,
+      "loss": 2.0135,
+      "step": 7558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.279672754163745e-05,
+      "loss": 1.7924,
+      "step": 7559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.274002504783754e-05,
+      "loss": 1.8511,
+      "step": 7560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.268333812640716e-05,
+      "loss": 1.7966,
+      "step": 7561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.262666678262416e-05,
+      "loss": 1.8251,
+      "step": 7562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.257001102176459e-05,
+      "loss": 1.9692,
+      "step": 7563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.251337084910335e-05,
+      "loss": 1.9106,
+      "step": 7564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.245674626991381e-05,
+      "loss": 1.7401,
+      "step": 7565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.240013728946785e-05,
+      "loss": 1.9115,
+      "step": 7566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.234354391303605e-05,
+      "loss": 1.8121,
+      "step": 7567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.228696614588722e-05,
+      "loss": 1.8347,
+      "step": 7568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.2230403993289e-05,
+      "loss": 1.826,
+      "step": 7569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.217385746050742e-05,
+      "loss": 2.1708,
+      "step": 7570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.211732655280724e-05,
+      "loss": 1.8579,
+      "step": 7571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.206081127545156e-05,
+      "loss": 2.0089,
+      "step": 7572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.20043116337022e-05,
+      "loss": 2.0076,
+      "step": 7573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.19478276328193e-05,
+      "loss": 2.0037,
+      "step": 7574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.189135927806171e-05,
+      "loss": 1.8281,
+      "step": 7575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.183490657468687e-05,
+      "loss": 2.095,
+      "step": 7576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.177846952795059e-05,
+      "loss": 1.7568,
+      "step": 7577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.172204814310741e-05,
+      "loss": 1.9869,
+      "step": 7578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.166564242541033e-05,
+      "loss": 1.9419,
+      "step": 7579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.160925238011072e-05,
+      "loss": 1.7997,
+      "step": 7580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.155287801245878e-05,
+      "loss": 1.9527,
+      "step": 7581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.149651932770308e-05,
+      "loss": 1.8965,
+      "step": 7582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.144017633109085e-05,
+      "loss": 2.0194,
+      "step": 7583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.138384902786767e-05,
+      "loss": 1.976,
+      "step": 7584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.132753742327778e-05,
+      "loss": 1.684,
+      "step": 7585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.127124152256408e-05,
+      "loss": 2.0045,
+      "step": 7586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.121496133096767e-05,
+      "loss": 1.7723,
+      "step": 7587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.115869685372851e-05,
+      "loss": 1.833,
+      "step": 7588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.110244809608495e-05,
+      "loss": 1.7647,
+      "step": 7589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.1046215063274e-05,
+      "loss": 1.5836,
+      "step": 7590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.098999776053098e-05,
+      "loss": 1.7823,
+      "step": 7591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.093379619308991e-05,
+      "loss": 1.766,
+      "step": 7592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.087761036618335e-05,
+      "loss": 1.7712,
+      "step": 7593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.082144028504232e-05,
+      "loss": 2.0046,
+      "step": 7594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.076528595489651e-05,
+      "loss": 2.136,
+      "step": 7595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.07091473809739e-05,
+      "loss": 1.7984,
+      "step": 7596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.065302456850123e-05,
+      "loss": 2.0676,
+      "step": 7597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.059691752270365e-05,
+      "loss": 1.8964,
+      "step": 7598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.05408262488049e-05,
+      "loss": 2.0154,
+      "step": 7599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.048475075202727e-05,
+      "loss": 1.9349,
+      "step": 7600
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 1.9102898836135864,
+      "eval_runtime": 184.0283,
+      "eval_samples_per_second": 457.56,
+      "eval_steps_per_second": 28.599,
+      "step": 7600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.042869103759157e-05,
+      "loss": 2.1436,
+      "step": 7601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.037264711071699e-05,
+      "loss": 2.0092,
+      "step": 7602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.031661897662143e-05,
+      "loss": 2.029,
+      "step": 7603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.026060664052131e-05,
+      "loss": 2.0361,
+      "step": 7604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.02046101076315e-05,
+      "loss": 1.656,
+      "step": 7605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.014862938316542e-05,
+      "loss": 1.9095,
+      "step": 7606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.009266447233512e-05,
+      "loss": 1.9573,
+      "step": 7607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.003671538035098e-05,
+      "loss": 1.8269,
+      "step": 7608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.998078211242202e-05,
+      "loss": 1.9807,
+      "step": 7609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.992486467375584e-05,
+      "loss": 1.7806,
+      "step": 7610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.986896306955848e-05,
+      "loss": 1.8265,
+      "step": 7611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.981307730503462e-05,
+      "loss": 1.9797,
+      "step": 7612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.975720738538725e-05,
+      "loss": 1.9108,
+      "step": 7613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.970135331581804e-05,
+      "loss": 1.788,
+      "step": 7614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.96455151015272e-05,
+      "loss": 1.769,
+      "step": 7615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.95896927477135e-05,
+      "loss": 2.0141,
+      "step": 7616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.953388625957397e-05,
+      "loss": 1.9023,
+      "step": 7617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.947809564230446e-05,
+      "loss": 1.8183,
+      "step": 7618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.942232090109927e-05,
+      "loss": 1.8118,
+      "step": 7619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.936656204115109e-05,
+      "loss": 1.8397,
+      "step": 7620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.931081906765125e-05,
+      "loss": 1.6588,
+      "step": 7621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.925509198578959e-05,
+      "loss": 2.1598,
+      "step": 7622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.919938080075456e-05,
+      "loss": 1.7386,
+      "step": 7623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.914368551773285e-05,
+      "loss": 1.8415,
+      "step": 7624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.90880061419099e-05,
+      "loss": 1.696,
+      "step": 7625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.903234267846965e-05,
+      "loss": 1.8362,
+      "step": 7626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.897669513259451e-05,
+      "loss": 1.6892,
+      "step": 7627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.892106350946543e-05,
+      "loss": 1.8072,
+      "step": 7628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.886544781426195e-05,
+      "loss": 1.845,
+      "step": 7629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.880984805216184e-05,
+      "loss": 1.7442,
+      "step": 7630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.875426422834176e-05,
+      "loss": 1.6959,
+      "step": 7631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.869869634797663e-05,
+      "loss": 1.8038,
+      "step": 7632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.864314441624005e-05,
+      "loss": 1.85,
+      "step": 7633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.858760843830401e-05,
+      "loss": 1.8219,
+      "step": 7634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.853208841933915e-05,
+      "loss": 1.8608,
+      "step": 7635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.84765843645144e-05,
+      "loss": 1.8034,
+      "step": 7636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.84210962789974e-05,
+      "loss": 1.8201,
+      "step": 7637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.836562416795426e-05,
+      "loss": 1.8323,
+      "step": 7638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.831016803654959e-05,
+      "loss": 1.7888,
+      "step": 7639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.825472788994656e-05,
+      "loss": 1.8661,
+      "step": 7640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.81993037333067e-05,
+      "loss": 1.7135,
+      "step": 7641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.814389557179016e-05,
+      "loss": 1.8523,
+      "step": 7642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.808850341055565e-05,
+      "loss": 1.7648,
+      "step": 7643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.803312725476031e-05,
+      "loss": 1.7306,
+      "step": 7644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.797776710955984e-05,
+      "loss": 1.7617,
+      "step": 7645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.792242298010846e-05,
+      "loss": 1.9247,
+      "step": 7646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.786709487155875e-05,
+      "loss": 1.7062,
+      "step": 7647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.781178278906195e-05,
+      "loss": 1.7901,
+      "step": 7648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.775648673776787e-05,
+      "loss": 1.9147,
+      "step": 7649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.770120672282458e-05,
+      "loss": 1.8001,
+      "step": 7650
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 1.909618854522705,
+      "eval_runtime": 184.0432,
+      "eval_samples_per_second": 457.523,
+      "eval_steps_per_second": 28.597,
+      "step": 7650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.764594274937884e-05,
+      "loss": 1.6447,
+      "step": 7651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.7590694822576e-05,
+      "loss": 1.7587,
+      "step": 7652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.753546294755964e-05,
+      "loss": 1.8374,
+      "step": 7653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.748024712947205e-05,
+      "loss": 1.7843,
+      "step": 7654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.7425047373454e-05,
+      "loss": 1.7783,
+      "step": 7655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.736986368464472e-05,
+      "loss": 1.8556,
+      "step": 7656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.731469606818206e-05,
+      "loss": 1.795,
+      "step": 7657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.725954452920211e-05,
+      "loss": 2.0726,
+      "step": 7658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.72044090728397e-05,
+      "loss": 2.0486,
+      "step": 7659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.714928970422817e-05,
+      "loss": 1.9429,
+      "step": 7660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.709418642849919e-05,
+      "loss": 1.8785,
+      "step": 7661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.703909925078307e-05,
+      "loss": 1.737,
+      "step": 7662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.698402817620865e-05,
+      "loss": 1.8533,
+      "step": 7663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.692897320990305e-05,
+      "loss": 1.7284,
+      "step": 7664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.687393435699213e-05,
+      "loss": 1.9729,
+      "step": 7665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.681891162260016e-05,
+      "loss": 1.8266,
+      "step": 7666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.676390501184988e-05,
+      "loss": 1.8644,
+      "step": 7667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.670891452986264e-05,
+      "loss": 1.7309,
+      "step": 7668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.665394018175812e-05,
+      "loss": 1.9061,
+      "step": 7669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.659898197265456e-05,
+      "loss": 1.9176,
+      "step": 7670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.654403990766878e-05,
+      "loss": 2.0443,
+      "step": 7671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.648911399191605e-05,
+      "loss": 1.9259,
+      "step": 7672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.643420423051009e-05,
+      "loss": 1.8875,
+      "step": 7673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.637931062856323e-05,
+      "loss": 1.6043,
+      "step": 7674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.632443319118612e-05,
+      "loss": 2.0553,
+      "step": 7675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.626957192348802e-05,
+      "loss": 1.9662,
+      "step": 7676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.621472683057668e-05,
+      "loss": 1.8642,
+      "step": 7677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.615989791755834e-05,
+      "loss": 1.8403,
+      "step": 7678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.610508518953779e-05,
+      "loss": 1.8589,
+      "step": 7679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.605028865161809e-05,
+      "loss": 1.7459,
+      "step": 7680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.599550830890107e-05,
+      "loss": 1.7772,
+      "step": 7681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.594074416648689e-05,
+      "loss": 1.9833,
+      "step": 7682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.588599622947432e-05,
+      "loss": 2.0083,
+      "step": 7683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.58312645029604e-05,
+      "loss": 1.8921,
+      "step": 7684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.577654899204089e-05,
+      "loss": 1.8604,
+      "step": 7685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.572184970181004e-05,
+      "loss": 2.0302,
+      "step": 7686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.566716663736034e-05,
+      "loss": 1.851,
+      "step": 7687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.561249980378302e-05,
+      "loss": 2.0057,
+      "step": 7688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.55578492061677e-05,
+      "loss": 1.9171,
+      "step": 7689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.550321484960252e-05,
+      "loss": 1.8748,
+      "step": 7690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.544859673917415e-05,
+      "loss": 1.7861,
+      "step": 7691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.539399487996754e-05,
+      "loss": 2.08,
+      "step": 7692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.533940927706637e-05,
+      "loss": 1.9522,
+      "step": 7693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.528483993555269e-05,
+      "loss": 1.8511,
+      "step": 7694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.523028686050706e-05,
+      "loss": 1.747,
+      "step": 7695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.517575005700863e-05,
+      "loss": 1.9966,
+      "step": 7696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.512122953013473e-05,
+      "loss": 2.0407,
+      "step": 7697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.506672528496147e-05,
+      "loss": 1.7714,
+      "step": 7698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.501223732656337e-05,
+      "loss": 2.0266,
+      "step": 7699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.495776566001336e-05,
+      "loss": 1.9535,
+      "step": 7700
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.9044908285140991,
+      "eval_runtime": 184.1117,
+      "eval_samples_per_second": 457.353,
+      "eval_steps_per_second": 28.586,
+      "step": 7700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.490331029038294e-05,
+      "loss": 1.8598,
+      "step": 7701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.484887122274214e-05,
+      "loss": 1.8266,
+      "step": 7702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.479444846215919e-05,
+      "loss": 1.8313,
+      "step": 7703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.474004201370113e-05,
+      "loss": 1.9762,
+      "step": 7704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.468565188243331e-05,
+      "loss": 1.7953,
+      "step": 7705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.463127807341966e-05,
+      "loss": 1.9411,
+      "step": 7706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.457692059172255e-05,
+      "loss": 1.6945,
+      "step": 7707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.452257944240267e-05,
+      "loss": 1.9776,
+      "step": 7708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.446825463051946e-05,
+      "loss": 1.9332,
+      "step": 7709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.441394616113063e-05,
+      "loss": 1.8177,
+      "step": 7710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.435965403929252e-05,
+      "loss": 1.8103,
+      "step": 7711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.430537827005982e-05,
+      "loss": 1.9245,
+      "step": 7712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.425111885848587e-05,
+      "loss": 1.9825,
+      "step": 7713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.419687580962223e-05,
+      "loss": 1.8236,
+      "step": 7714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.414264912851912e-05,
+      "loss": 1.7866,
+      "step": 7715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.408843882022531e-05,
+      "loss": 1.6946,
+      "step": 7716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.403424488978772e-05,
+      "loss": 1.8307,
+      "step": 7717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.39800673422521e-05,
+      "loss": 2.0538,
+      "step": 7718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.392590618266257e-05,
+      "loss": 1.9928,
+      "step": 7719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.387176141606153e-05,
+      "loss": 1.8179,
+      "step": 7720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.38176330474901e-05,
+      "loss": 1.814,
+      "step": 7721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.376352108198774e-05,
+      "loss": 2.0532,
+      "step": 7722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.370942552459247e-05,
+      "loss": 1.681,
+      "step": 7723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.365534638034083e-05,
+      "loss": 1.9084,
+      "step": 7724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.360128365426754e-05,
+      "loss": 1.9722,
+      "step": 7725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.354723735140609e-05,
+      "loss": 1.6935,
+      "step": 7726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.349320747678834e-05,
+      "loss": 1.8344,
+      "step": 7727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.343919403544461e-05,
+      "loss": 1.8746,
+      "step": 7728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.33851970324037e-05,
+      "loss": 1.805,
+      "step": 7729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.333121647269301e-05,
+      "loss": 1.8218,
+      "step": 7730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.327725236133803e-05,
+      "loss": 1.7069,
+      "step": 7731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.322330470336314e-05,
+      "loss": 2.0436,
+      "step": 7732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.316937350379097e-05,
+      "loss": 1.6562,
+      "step": 7733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.311545876764267e-05,
+      "loss": 1.8592,
+      "step": 7734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.306156049993792e-05,
+      "loss": 1.6575,
+      "step": 7735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.300767870569466e-05,
+      "loss": 1.7102,
+      "step": 7736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.295381338992951e-05,
+      "loss": 2.0563,
+      "step": 7737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.28999645576575e-05,
+      "loss": 1.8032,
+      "step": 7738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.284613221389208e-05,
+      "loss": 1.963,
+      "step": 7739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.279231636364516e-05,
+      "loss": 1.8801,
+      "step": 7740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.27385170119273e-05,
+      "loss": 2.0731,
+      "step": 7741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.268473416374718e-05,
+      "loss": 1.9988,
+      "step": 7742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.263096782411218e-05,
+      "loss": 2.0025,
+      "step": 7743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.257721799802814e-05,
+      "loss": 1.8252,
+      "step": 7744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.25234846904993e-05,
+      "loss": 1.7946,
+      "step": 7745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.246976790652843e-05,
+      "loss": 1.9016,
+      "step": 7746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.241606765111661e-05,
+      "loss": 1.858,
+      "step": 7747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.236238392926353e-05,
+      "loss": 1.8535,
+      "step": 7748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.23087167459674e-05,
+      "loss": 1.8045,
+      "step": 7749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.225506610622457e-05,
+      "loss": 1.7291,
+      "step": 7750
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.902468204498291,
+      "eval_runtime": 184.061,
+      "eval_samples_per_second": 457.479,
+      "eval_steps_per_second": 28.594,
+      "step": 7750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.220143201503018e-05,
+      "loss": 2.0011,
+      "step": 7751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.214781447737781e-05,
+      "loss": 1.7912,
+      "step": 7752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.209421349825924e-05,
+      "loss": 1.7374,
+      "step": 7753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.20406290826649e-05,
+      "loss": 1.7887,
+      "step": 7754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.198706123558369e-05,
+      "loss": 1.9635,
+      "step": 7755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.193350996200293e-05,
+      "loss": 1.9945,
+      "step": 7756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.187997526690838e-05,
+      "loss": 1.8157,
+      "step": 7757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.182645715528436e-05,
+      "loss": 1.8524,
+      "step": 7758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.177295563211336e-05,
+      "loss": 1.9214,
+      "step": 7759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.171947070237663e-05,
+      "loss": 2.0221,
+      "step": 7760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.166600237105378e-05,
+      "loss": 2.0217,
+      "step": 7761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.161255064312283e-05,
+      "loss": 1.9067,
+      "step": 7762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.155911552356037e-05,
+      "loss": 2.058,
+      "step": 7763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.150569701734122e-05,
+      "loss": 2.1014,
+      "step": 7764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.145229512943885e-05,
+      "loss": 1.8602,
+      "step": 7765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.139890986482513e-05,
+      "loss": 2.042,
+      "step": 7766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.13455412284704e-05,
+      "loss": 1.9574,
+      "step": 7767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.129218922534342e-05,
+      "loss": 2.0047,
+      "step": 7768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.123885386041142e-05,
+      "loss": 2.092,
+      "step": 7769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.118553513864002e-05,
+      "loss": 1.9537,
+      "step": 7770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.113223306499337e-05,
+      "loss": 1.7179,
+      "step": 7771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1078947644434e-05,
+      "loss": 1.8502,
+      "step": 7772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.102567888192301e-05,
+      "loss": 1.8327,
+      "step": 7773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.097242678241988e-05,
+      "loss": 1.7971,
+      "step": 7774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.091919135088242e-05,
+      "loss": 1.8737,
+      "step": 7775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.086597259226707e-05,
+      "loss": 1.7568,
+      "step": 7776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.081277051152866e-05,
+      "loss": 1.7949,
+      "step": 7777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.075958511362037e-05,
+      "loss": 1.7227,
+      "step": 7778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.070641640349407e-05,
+      "loss": 1.7158,
+      "step": 7779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.065326438609973e-05,
+      "loss": 1.8753,
+      "step": 7780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.060012906638602e-05,
+      "loss": 1.86,
+      "step": 7781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.054701044930009e-05,
+      "loss": 1.7913,
+      "step": 7782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.049390853978724e-05,
+      "loss": 1.9441,
+      "step": 7783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.044082334279151e-05,
+      "loss": 1.8239,
+      "step": 7784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.038775486325527e-05,
+      "loss": 1.7339,
+      "step": 7785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.033470310611945e-05,
+      "loss": 1.9073,
+      "step": 7786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.02816680763231e-05,
+      "loss": 1.7705,
+      "step": 7787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.022864977880408e-05,
+      "loss": 1.9426,
+      "step": 7788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.017564821849848e-05,
+      "loss": 1.8669,
+      "step": 7789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.01226634003409e-05,
+      "loss": 1.9757,
+      "step": 7790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.00696953292645e-05,
+      "loss": 1.8228,
+      "step": 7791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.001674401020055e-05,
+      "loss": 1.9468,
+      "step": 7792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.996380944807906e-05,
+      "loss": 1.8122,
+      "step": 7793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.991089164782838e-05,
+      "loss": 1.9405,
+      "step": 7794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.985799061437531e-05,
+      "loss": 2.0505,
+      "step": 7795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.980510635264512e-05,
+      "loss": 1.9132,
+      "step": 7796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.975223886756149e-05,
+      "loss": 1.6715,
+      "step": 7797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.96993881640464e-05,
+      "loss": 1.8506,
+      "step": 7798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.964655424702049e-05,
+      "loss": 2.1426,
+      "step": 7799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.959373712140274e-05,
+      "loss": 1.8336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 1.9009184837341309,
+      "eval_runtime": 184.1058,
+      "eval_samples_per_second": 457.368,
+      "eval_steps_per_second": 28.587,
+      "step": 7800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.95409367921106e-05,
+      "loss": 1.9135,
+      "step": 7801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.948815326405994e-05,
+      "loss": 1.7856,
+      "step": 7802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.943538654216494e-05,
+      "loss": 1.7597,
+      "step": 7803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.938263663133839e-05,
+      "loss": 1.8301,
+      "step": 7804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.932990353649149e-05,
+      "loss": 1.966,
+      "step": 7805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.927718726253379e-05,
+      "loss": 1.8831,
+      "step": 7806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.922448781437335e-05,
+      "loss": 1.9267,
+      "step": 7807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.91718051969167e-05,
+      "loss": 1.983,
+      "step": 7808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.911913941506861e-05,
+      "loss": 1.8596,
+      "step": 7809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.906649047373245e-05,
+      "loss": 1.8883,
+      "step": 7810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.901385837781002e-05,
+      "loss": 1.8589,
+      "step": 7811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.896124313220148e-05,
+      "loss": 1.9569,
+      "step": 7812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.890864474180556e-05,
+      "loss": 1.7661,
+      "step": 7813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.885606321151913e-05,
+      "loss": 1.9018,
+      "step": 7814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.880349854623779e-05,
+      "loss": 1.7407,
+      "step": 7815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.875095075085552e-05,
+      "loss": 1.9588,
+      "step": 7816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.86984198302645e-05,
+      "loss": 1.843,
+      "step": 7817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.864590578935562e-05,
+      "loss": 1.758,
+      "step": 7818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.85934086330181e-05,
+      "loss": 1.865,
+      "step": 7819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.854092836613949e-05,
+      "loss": 1.7581,
+      "step": 7820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.848846499360583e-05,
+      "loss": 1.8111,
+      "step": 7821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.843601852030171e-05,
+      "loss": 2.0173,
+      "step": 7822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.838358895111e-05,
+      "loss": 1.9045,
+      "step": 7823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.833117629091201e-05,
+      "loss": 1.8956,
+      "step": 7824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.827878054458761e-05,
+      "loss": 1.7331,
+      "step": 7825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.822640171701486e-05,
+      "loss": 1.7971,
+      "step": 7826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.81740398130704e-05,
+      "loss": 1.9228,
+      "step": 7827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.812169483762928e-05,
+      "loss": 1.6703,
+      "step": 7828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.806936679556502e-05,
+      "loss": 1.9417,
+      "step": 7829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.801705569174951e-05,
+      "loss": 1.9326,
+      "step": 7830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.796476153105294e-05,
+      "loss": 1.8988,
+      "step": 7831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.791248431834412e-05,
+      "loss": 1.991,
+      "step": 7832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.786022405849018e-05,
+      "loss": 1.9985,
+      "step": 7833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.780798075635675e-05,
+      "loss": 2.0514,
+      "step": 7834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.775575441680776e-05,
+      "loss": 1.8975,
+      "step": 7835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.770354504470574e-05,
+      "loss": 1.9189,
+      "step": 7836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.765135264491138e-05,
+      "loss": 1.9715,
+      "step": 7837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.759917722228401e-05,
+      "loss": 1.7775,
+      "step": 7838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.75470187816813e-05,
+      "loss": 2.0032,
+      "step": 7839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.749487732795933e-05,
+      "loss": 1.8696,
+      "step": 7840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.744275286597264e-05,
+      "loss": 1.9049,
+      "step": 7841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.739064540057424e-05,
+      "loss": 2.0106,
+      "step": 7842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.733855493661531e-05,
+      "loss": 2.0318,
+      "step": 7843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.728648147894572e-05,
+      "loss": 1.6991,
+      "step": 7844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.723442503241362e-05,
+      "loss": 1.8499,
+      "step": 7845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.718238560186571e-05,
+      "loss": 1.8465,
+      "step": 7846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.713036319214685e-05,
+      "loss": 1.9473,
+      "step": 7847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.707835780810054e-05,
+      "loss": 1.8828,
+      "step": 7848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.70263694545687e-05,
+      "loss": 1.7321,
+      "step": 7849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.697439813639145e-05,
+      "loss": 2.0697,
+      "step": 7850
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 1.8965145349502563,
+      "eval_runtime": 184.0709,
+      "eval_samples_per_second": 457.454,
+      "eval_steps_per_second": 28.592,
+      "step": 7850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.692244385840754e-05,
+      "loss": 1.8472,
+      "step": 7851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.687050662545405e-05,
+      "loss": 1.9027,
+      "step": 7852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.681858644236655e-05,
+      "loss": 1.8883,
+      "step": 7853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.676668331397884e-05,
+      "loss": 1.9216,
+      "step": 7854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.671479724512327e-05,
+      "loss": 1.8474,
+      "step": 7855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.66629282406306e-05,
+      "loss": 1.7734,
+      "step": 7856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.661107630532998e-05,
+      "loss": 1.9047,
+      "step": 7857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.655924144404906e-05,
+      "loss": 1.9316,
+      "step": 7858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.650742366161364e-05,
+      "loss": 1.9471,
+      "step": 7859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.645562296284818e-05,
+      "loss": 1.7268,
+      "step": 7860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.640383935257546e-05,
+      "loss": 1.8447,
+      "step": 7861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.63520728356167e-05,
+      "loss": 1.7317,
+      "step": 7862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.630032341679149e-05,
+      "loss": 1.9385,
+      "step": 7863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.624859110091791e-05,
+      "loss": 1.7817,
+      "step": 7864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.619687589281226e-05,
+      "loss": 1.8834,
+      "step": 7865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.614517779728944e-05,
+      "loss": 1.8784,
+      "step": 7866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.609349681916265e-05,
+      "loss": 1.7964,
+      "step": 7867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.604183296324359e-05,
+      "loss": 1.8786,
+      "step": 7868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.599018623434225e-05,
+      "loss": 1.9803,
+      "step": 7869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.593855663726722e-05,
+      "loss": 2.0809,
+      "step": 7870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.588694417682517e-05,
+      "loss": 1.8512,
+      "step": 7871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.583534885782144e-05,
+      "loss": 1.7409,
+      "step": 7872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.578377068505972e-05,
+      "loss": 1.897,
+      "step": 7873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.573220966334207e-05,
+      "loss": 2.0309,
+      "step": 7874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5680665797469e-05,
+      "loss": 1.6528,
+      "step": 7875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.56291390922393e-05,
+      "loss": 1.9414,
+      "step": 7876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.55776295524503e-05,
+      "loss": 1.7758,
+      "step": 7877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.552613718289771e-05,
+      "loss": 1.9477,
+      "step": 7878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.547466198837562e-05,
+      "loss": 1.9167,
+      "step": 7879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.542320397367641e-05,
+      "loss": 1.8034,
+      "step": 7880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.537176314359103e-05,
+      "loss": 1.9448,
+      "step": 7881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.532033950290886e-05,
+      "loss": 1.8411,
+      "step": 7882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.526893305641741e-05,
+      "loss": 1.9352,
+      "step": 7883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.521754380890283e-05,
+      "loss": 2.0313,
+      "step": 7884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.516617176514964e-05,
+      "loss": 1.992,
+      "step": 7885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.511481692994075e-05,
+      "loss": 1.9281,
+      "step": 7886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.506347930805731e-05,
+      "loss": 1.7821,
+      "step": 7887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.501215890427908e-05,
+      "loss": 1.9352,
+      "step": 7888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.496085572338415e-05,
+      "loss": 1.9192,
+      "step": 7889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.490956977014892e-05,
+      "loss": 1.7938,
+      "step": 7890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.485830104934831e-05,
+      "loss": 1.662,
+      "step": 7891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.480704956575564e-05,
+      "loss": 1.8343,
+      "step": 7892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.475581532414243e-05,
+      "loss": 1.9088,
+      "step": 7893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.47045983292788e-05,
+      "loss": 1.9948,
+      "step": 7894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.465339858593316e-05,
+      "loss": 1.7082,
+      "step": 7895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.46022160988724e-05,
+      "loss": 1.8427,
+      "step": 7896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.455105087286172e-05,
+      "loss": 1.8149,
+      "step": 7897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.449990291266486e-05,
+      "loss": 1.9869,
+      "step": 7898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.444877222304362e-05,
+      "loss": 1.8635,
+      "step": 7899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.439765880875856e-05,
+      "loss": 1.7562,
+      "step": 7900
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 1.894385576248169,
+      "eval_runtime": 184.2516,
+      "eval_samples_per_second": 457.005,
+      "eval_steps_per_second": 28.564,
+      "step": 7900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.434656267456842e-05,
+      "loss": 2.0498,
+      "step": 7901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.429548382523043e-05,
+      "loss": 1.947,
+      "step": 7902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.42444222655002e-05,
+      "loss": 1.8785,
+      "step": 7903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.419337800013162e-05,
+      "loss": 1.9506,
+      "step": 7904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.414235103387708e-05,
+      "loss": 1.7403,
+      "step": 7905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.409134137148737e-05,
+      "loss": 1.8848,
+      "step": 7906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.404034901771161e-05,
+      "loss": 1.9466,
+      "step": 7907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.398937397729732e-05,
+      "loss": 1.8561,
+      "step": 7908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.39384162549905e-05,
+      "loss": 1.8574,
+      "step": 7909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.388747585553531e-05,
+      "loss": 2.0087,
+      "step": 7910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.383655278367451e-05,
+      "loss": 1.94,
+      "step": 7911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.378564704414922e-05,
+      "loss": 1.9511,
+      "step": 7912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.373475864169892e-05,
+      "loss": 1.6875,
+      "step": 7913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.368388758106134e-05,
+      "loss": 1.7922,
+      "step": 7914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.36330338669728e-05,
+      "loss": 1.7914,
+      "step": 7915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3582197504168e-05,
+      "loss": 1.9583,
+      "step": 7916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.353137849737977e-05,
+      "loss": 1.8699,
+      "step": 7917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.348057685133963e-05,
+      "loss": 1.805,
+      "step": 7918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.342979257077727e-05,
+      "loss": 1.9386,
+      "step": 7919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3379025660421e-05,
+      "loss": 1.8344,
+      "step": 7920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.332827612499716e-05,
+      "loss": 1.8189,
+      "step": 7921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.32775439692308e-05,
+      "loss": 1.7412,
+      "step": 7922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.32268291978452e-05,
+      "loss": 1.9188,
+      "step": 7923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3176131815562e-05,
+      "loss": 1.9169,
+      "step": 7924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.312545182710133e-05,
+      "loss": 1.7528,
+      "step": 7925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.307478923718171e-05,
+      "loss": 1.8445,
+      "step": 7926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.302414405051979e-05,
+      "loss": 1.9304,
+      "step": 7927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.297351627183088e-05,
+      "loss": 2.1312,
+      "step": 7928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.292290590582852e-05,
+      "loss": 1.8519,
+      "step": 7929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.28723129572247e-05,
+      "loss": 1.9482,
+      "step": 7930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.282173743072988e-05,
+      "loss": 1.8711,
+      "step": 7931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.277117933105258e-05,
+      "loss": 1.8146,
+      "step": 7932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.272063866290001e-05,
+      "loss": 1.775,
+      "step": 7933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.267011543097761e-05,
+      "loss": 1.8775,
+      "step": 7934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.261960963998928e-05,
+      "loss": 1.9558,
+      "step": 7935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.256912129463718e-05,
+      "loss": 1.8895,
+      "step": 7936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.251865039962207e-05,
+      "loss": 1.8874,
+      "step": 7937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.246819695964273e-05,
+      "loss": 1.9655,
+      "step": 7938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.241776097939664e-05,
+      "loss": 1.8577,
+      "step": 7939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.236734246357948e-05,
+      "loss": 1.8712,
+      "step": 7940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.231694141688534e-05,
+      "loss": 1.9664,
+      "step": 7941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.226655784400684e-05,
+      "loss": 1.7468,
+      "step": 7942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.22161917496346e-05,
+      "loss": 1.8855,
+      "step": 7943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.216584313845802e-05,
+      "loss": 1.7555,
+      "step": 7944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.21155120151646e-05,
+      "loss": 1.7466,
+      "step": 7945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.206519838444044e-05,
+      "loss": 1.8631,
+      "step": 7946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.20149022509697e-05,
+      "loss": 1.7174,
+      "step": 7947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.196462361943519e-05,
+      "loss": 1.9365,
+      "step": 7948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.191436249451804e-05,
+      "loss": 1.6946,
+      "step": 7949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.186411888089757e-05,
+      "loss": 1.7786,
+      "step": 7950
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 1.893282413482666,
+      "eval_runtime": 184.2059,
+      "eval_samples_per_second": 457.119,
+      "eval_steps_per_second": 28.571,
+      "step": 7950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.181389278325166e-05,
+      "loss": 1.9164,
+      "step": 7951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.176368420625652e-05,
+      "loss": 1.8909,
+      "step": 7952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.171349315458669e-05,
+      "loss": 1.7844,
+      "step": 7953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.166331963291518e-05,
+      "loss": 1.6645,
+      "step": 7954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.161316364591313e-05,
+      "loss": 1.5784,
+      "step": 7955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.15630251982503e-05,
+      "loss": 1.9941,
+      "step": 7956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.151290429459467e-05,
+      "loss": 1.9873,
+      "step": 7957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.146280093961268e-05,
+      "loss": 1.9147,
+      "step": 7958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.141271513796915e-05,
+      "loss": 1.8604,
+      "step": 7959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.136264689432705e-05,
+      "loss": 1.7922,
+      "step": 7960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.131259621334795e-05,
+      "loss": 1.8278,
+      "step": 7961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.126256309969172e-05,
+      "loss": 1.8821,
+      "step": 7962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.121254755801658e-05,
+      "loss": 1.9426,
+      "step": 7963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.116254959297913e-05,
+      "loss": 1.566,
+      "step": 7964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.111256920923436e-05,
+      "loss": 1.9606,
+      "step": 7965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.106260641143546e-05,
+      "loss": 1.8244,
+      "step": 7966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.101266120423416e-05,
+      "loss": 2.0512,
+      "step": 7967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.096273359228052e-05,
+      "loss": 1.8549,
+      "step": 7968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.091282358022293e-05,
+      "loss": 1.8732,
+      "step": 7969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0862931172708217e-05,
+      "loss": 1.8296,
+      "step": 7970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.081305637438136e-05,
+      "loss": 1.7946,
+      "step": 7971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0763199189885934e-05,
+      "loss": 1.8291,
+      "step": 7972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0713359623863735e-05,
+      "loss": 1.8209,
+      "step": 7973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0663537680955035e-05,
+      "loss": 1.8447,
+      "step": 7974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.061373336579834e-05,
+      "loss": 2.1076,
+      "step": 7975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.056394668303064e-05,
+      "loss": 1.7669,
+      "step": 7976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.051417763728711e-05,
+      "loss": 1.9078,
+      "step": 7977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0464426233201446e-05,
+      "loss": 1.7376,
+      "step": 7978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0414692475405706e-05,
+      "loss": 1.7294,
+      "step": 7979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0364976368530086e-05,
+      "loss": 1.8475,
+      "step": 7980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.031527791720337e-05,
+      "loss": 1.9345,
+      "step": 7981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.02655971260527e-05,
+      "loss": 1.7827,
+      "step": 7982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0215933999703376e-05,
+      "loss": 1.8196,
+      "step": 7983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.016628854277922e-05,
+      "loss": 1.7132,
+      "step": 7984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.011666075990235e-05,
+      "loss": 1.9126,
+      "step": 7985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.006705065569329e-05,
+      "loss": 1.9679,
+      "step": 7986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.001745823477092e-05,
+      "loss": 1.7699,
+      "step": 7987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.996788350175228e-05,
+      "loss": 2.0556,
+      "step": 7988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9918326461253e-05,
+      "loss": 1.7976,
+      "step": 7989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.986878711788701e-05,
+      "loss": 1.5965,
+      "step": 7990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.981926547626651e-05,
+      "loss": 1.8803,
+      "step": 7991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.976976154100214e-05,
+      "loss": 1.7061,
+      "step": 7992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9720275316702905e-05,
+      "loss": 1.8758,
+      "step": 7993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.967080680797599e-05,
+      "loss": 1.9952,
+      "step": 7994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9621356019427085e-05,
+      "loss": 1.7456,
+      "step": 7995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.957192295566022e-05,
+      "loss": 1.837,
+      "step": 7996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.952250762127775e-05,
+      "loss": 2.0024,
+      "step": 7997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.947311002088046e-05,
+      "loss": 1.8002,
+      "step": 7998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.942373015906724e-05,
+      "loss": 1.927,
+      "step": 7999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.937436804043558e-05,
+      "loss": 1.8246,
+      "step": 8000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.889865517616272,
+      "eval_runtime": 184.2022,
+      "eval_samples_per_second": 457.128,
+      "eval_steps_per_second": 28.572,
+      "step": 8000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.932502366958123e-05,
+      "loss": 1.8177,
+      "step": 8001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.927569705109828e-05,
+      "loss": 1.7531,
+      "step": 8002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.922638818957918e-05,
+      "loss": 1.9212,
+      "step": 8003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.91770970896148e-05,
+      "loss": 1.6661,
+      "step": 8004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.912782375579412e-05,
+      "loss": 1.9825,
+      "step": 8005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.907856819270471e-05,
+      "loss": 1.9391,
+      "step": 8006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.902933040493241e-05,
+      "loss": 2.0608,
+      "step": 8007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.898011039706136e-05,
+      "loss": 1.8751,
+      "step": 8008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.893090817367419e-05,
+      "loss": 1.8768,
+      "step": 8009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8881723739351606e-05,
+      "loss": 1.7952,
+      "step": 8010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8832557098672886e-05,
+      "loss": 1.7985,
+      "step": 8011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.878340825621556e-05,
+      "loss": 1.7578,
+      "step": 8012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.873427721655561e-05,
+      "loss": 1.9204,
+      "step": 8013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8685163984267156e-05,
+      "loss": 1.7734,
+      "step": 8014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8636068563922815e-05,
+      "loss": 1.9077,
+      "step": 8015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8586990960093597e-05,
+      "loss": 1.9893,
+      "step": 8016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.85379311773486e-05,
+      "loss": 1.8043,
+      "step": 8017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.848888922025553e-05,
+      "loss": 1.9214,
+      "step": 8018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.843986509338029e-05,
+      "loss": 1.8219,
+      "step": 8019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8390858801287194e-05,
+      "loss": 1.9271,
+      "step": 8020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8341870348538936e-05,
+      "loss": 1.7015,
+      "step": 8021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.829289973969631e-05,
+      "loss": 1.6289,
+      "step": 8022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.82439469793187e-05,
+      "loss": 1.7877,
+      "step": 8023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.819501207196376e-05,
+      "loss": 1.9244,
+      "step": 8024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.814609502218743e-05,
+      "loss": 1.7851,
+      "step": 8025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.809719583454415e-05,
+      "loss": 1.7915,
+      "step": 8026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.804831451358636e-05,
+      "loss": 1.9852,
+      "step": 8027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.799945106386517e-05,
+      "loss": 1.8538,
+      "step": 8028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7950605489929895e-05,
+      "loss": 1.8284,
+      "step": 8029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.790177779632816e-05,
+      "loss": 1.6802,
+      "step": 8030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7852967987606006e-05,
+      "loss": 1.9275,
+      "step": 8031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.780417606830782e-05,
+      "loss": 1.869,
+      "step": 8032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.775540204297608e-05,
+      "loss": 1.745,
+      "step": 8033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7706645916151904e-05,
+      "loss": 1.6503,
+      "step": 8034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.765790769237464e-05,
+      "loss": 1.8743,
+      "step": 8035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7609187376181896e-05,
+      "loss": 1.8816,
+      "step": 8036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.756048497210978e-05,
+      "loss": 1.925,
+      "step": 8037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7511800484692426e-05,
+      "loss": 1.7568,
+      "step": 8038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.746313391846264e-05,
+      "loss": 1.8827,
+      "step": 8039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.741448527795137e-05,
+      "loss": 1.6872,
+      "step": 8040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.736585456768797e-05,
+      "loss": 2.063,
+      "step": 8041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.731724179220007e-05,
+      "loss": 1.6868,
+      "step": 8042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.726864695601372e-05,
+      "loss": 1.9853,
+      "step": 8043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.72200700636531e-05,
+      "loss": 1.9375,
+      "step": 8044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.717151111964095e-05,
+      "loss": 1.6566,
+      "step": 8045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7122970128498266e-05,
+      "loss": 2.0017,
+      "step": 8046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.707444709474424e-05,
+      "loss": 1.8165,
+      "step": 8047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.702594202289657e-05,
+      "loss": 1.9056,
+      "step": 8048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.697745491747131e-05,
+      "loss": 1.8448,
+      "step": 8049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.692898578298253e-05,
+      "loss": 1.7964,
+      "step": 8050
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.8870518207550049,
+      "eval_runtime": 184.6533,
+      "eval_samples_per_second": 456.011,
+      "eval_steps_per_second": 28.502,
+      "step": 8050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6880534623942966e-05,
+      "loss": 1.911,
+      "step": 8051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.683210144486356e-05,
+      "loss": 1.7683,
+      "step": 8052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.678368625025354e-05,
+      "loss": 1.9826,
+      "step": 8053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.673528904462061e-05,
+      "loss": 1.8897,
+      "step": 8054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.668690983247052e-05,
+      "loss": 1.9569,
+      "step": 8055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6638548618307585e-05,
+      "loss": 1.8835,
+      "step": 8056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.659020540663434e-05,
+      "loss": 1.8499,
+      "step": 8057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.654188020195172e-05,
+      "loss": 2.0126,
+      "step": 8058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.64935730087589e-05,
+      "loss": 1.8998,
+      "step": 8059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.644528383155351e-05,
+      "loss": 1.6329,
+      "step": 8060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6397012674831246e-05,
+      "loss": 1.7062,
+      "step": 8061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.634875954308638e-05,
+      "loss": 1.6466,
+      "step": 8062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.630052444081138e-05,
+      "loss": 1.7872,
+      "step": 8063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.625230737249709e-05,
+      "loss": 1.8542,
+      "step": 8064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.620410834263273e-05,
+      "loss": 1.9471,
+      "step": 8065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.615592735570563e-05,
+      "loss": 1.9162,
+      "step": 8066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6107764416201605e-05,
+      "loss": 1.7732,
+      "step": 8067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6059619528604775e-05,
+      "loss": 1.8549,
+      "step": 8068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.601149269739758e-05,
+      "loss": 2.0837,
+      "step": 8069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.596338392706077e-05,
+      "loss": 1.8149,
+      "step": 8070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.591529322207345e-05,
+      "loss": 2.0168,
+      "step": 8071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.586722058691285e-05,
+      "loss": 2.049,
+      "step": 8072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.581916602605478e-05,
+      "loss": 2.0327,
+      "step": 8073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.57711295439732e-05,
+      "loss": 1.7541,
+      "step": 8074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5723111145140476e-05,
+      "loss": 1.8319,
+      "step": 8075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.567511083402732e-05,
+      "loss": 1.843,
+      "step": 8076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.562712861510255e-05,
+      "loss": 1.9016,
+      "step": 8077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.557916449283351e-05,
+      "loss": 1.8503,
+      "step": 8078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.553121847168588e-05,
+      "loss": 1.7835,
+      "step": 8079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5483290556123416e-05,
+      "loss": 1.7476,
+      "step": 8080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.543538075060839e-05,
+      "loss": 1.771,
+      "step": 8081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.538748905960145e-05,
+      "loss": 1.7596,
+      "step": 8082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.533961548756128e-05,
+      "loss": 1.6719,
+      "step": 8083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.529176003894509e-05,
+      "loss": 1.7455,
+      "step": 8084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.524392271820841e-05,
+      "loss": 1.7046,
+      "step": 8085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5196103529805e-05,
+      "loss": 1.726,
+      "step": 8086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5148302478186976e-05,
+      "loss": 1.9717,
+      "step": 8087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5100519567804766e-05,
+      "loss": 2.0133,
+      "step": 8088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.505275480310701e-05,
+      "loss": 1.8121,
+      "step": 8089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.500500818854079e-05,
+      "loss": 1.977,
+      "step": 8090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.495727972855144e-05,
+      "loss": 1.8199,
+      "step": 8091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.490956942758263e-05,
+      "loss": 2.0988,
+      "step": 8092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.486187729007638e-05,
+      "loss": 2.0268,
+      "step": 8093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.481420332047285e-05,
+      "loss": 1.6744,
+      "step": 8094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.476654752321064e-05,
+      "loss": 2.0509,
+      "step": 8095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.471890990272666e-05,
+      "loss": 1.7889,
+      "step": 8096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.467129046345609e-05,
+      "loss": 1.9147,
+      "step": 8097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.462368920983249e-05,
+      "loss": 1.8647,
+      "step": 8098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.457610614628766e-05,
+      "loss": 1.7344,
+      "step": 8099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.452854127725163e-05,
+      "loss": 1.9483,
+      "step": 8100
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 1.8859477043151855,
+      "eval_runtime": 184.1245,
+      "eval_samples_per_second": 457.321,
+      "eval_steps_per_second": 28.584,
+      "step": 8100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448099460715289e-05,
+      "loss": 1.7336,
+      "step": 8101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4433466140418134e-05,
+      "loss": 1.6014,
+      "step": 8102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4385955881472407e-05,
+      "loss": 1.7681,
+      "step": 8103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.433846383473906e-05,
+      "loss": 1.6511,
+      "step": 8104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.429099000463983e-05,
+      "loss": 1.7791,
+      "step": 8105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4243534395594456e-05,
+      "loss": 1.7392,
+      "step": 8106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.41960970120213e-05,
+      "loss": 1.5877,
+      "step": 8107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.414867785833691e-05,
+      "loss": 1.8681,
+      "step": 8108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.410127693895619e-05,
+      "loss": 1.9576,
+      "step": 8109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4053894258292185e-05,
+      "loss": 1.8158,
+      "step": 8110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.400652982075643e-05,
+      "loss": 1.8039,
+      "step": 8111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.395918363075866e-05,
+      "loss": 1.8283,
+      "step": 8112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.391185569270701e-05,
+      "loss": 2.0864,
+      "step": 8113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3864546011007733e-05,
+      "loss": 1.7411,
+      "step": 8114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.381725459006553e-05,
+      "loss": 1.9429,
+      "step": 8115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.376998143428344e-05,
+      "loss": 1.7511,
+      "step": 8116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3722726548062597e-05,
+      "loss": 1.7174,
+      "step": 8117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3675489935802625e-05,
+      "loss": 1.7688,
+      "step": 8118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3628271601901365e-05,
+      "loss": 1.8313,
+      "step": 8119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3581071550755034e-05,
+      "loss": 1.6775,
+      "step": 8120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.353388978675808e-05,
+      "loss": 1.983,
+      "step": 8121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.348672631430318e-05,
+      "loss": 1.8932,
+      "step": 8122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.343958113778141e-05,
+      "loss": 1.9645,
+      "step": 8123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.339245426158218e-05,
+      "loss": 1.7553,
+      "step": 8124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.334534569009306e-05,
+      "loss": 1.7497,
+      "step": 8125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3298255427700004e-05,
+      "loss": 1.8697,
+      "step": 8126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3251183478787355e-05,
+      "loss": 1.9115,
+      "step": 8127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.320412984773748e-05,
+      "loss": 1.8257,
+      "step": 8128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3157094538931275e-05,
+      "loss": 1.9203,
+      "step": 8129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3110077556747847e-05,
+      "loss": 2.0278,
+      "step": 8130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3063078905564635e-05,
+      "loss": 2.0355,
+      "step": 8131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.301609858975731e-05,
+      "loss": 1.622,
+      "step": 8132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.296913661369998e-05,
+      "loss": 1.9475,
+      "step": 8133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2922192981764763e-05,
+      "loss": 1.8781,
+      "step": 8134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.287526769832232e-05,
+      "loss": 2.057,
+      "step": 8135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.282836076774153e-05,
+      "loss": 1.9409,
+      "step": 8136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.278147219438953e-05,
+      "loss": 1.8424,
+      "step": 8137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2734601982631915e-05,
+      "loss": 2.0172,
+      "step": 8138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.268775013683222e-05,
+      "loss": 1.9781,
+      "step": 8139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2640916661352595e-05,
+      "loss": 2.0326,
+      "step": 8140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.259410156055336e-05,
+      "loss": 1.7358,
+      "step": 8141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.254730483879311e-05,
+      "loss": 1.6446,
+      "step": 8142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2500526500428855e-05,
+      "loss": 1.7336,
+      "step": 8143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.245376654981562e-05,
+      "loss": 1.9224,
+      "step": 8144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.240702499130698e-05,
+      "loss": 1.7786,
+      "step": 8145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2360301829254746e-05,
+      "loss": 1.7942,
+      "step": 8146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.231359706800887e-05,
+      "loss": 1.9582,
+      "step": 8147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.226691071191772e-05,
+      "loss": 1.641,
+      "step": 8148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.222024276532803e-05,
+      "loss": 1.8443,
+      "step": 8149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.217359323258458e-05,
+      "loss": 1.7607,
+      "step": 8150
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 1.8835762739181519,
+      "eval_runtime": 184.0842,
+      "eval_samples_per_second": 457.421,
+      "eval_steps_per_second": 28.59,
+      "step": 8150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.212696211803061e-05,
+      "loss": 1.8565,
+      "step": 8151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.208034942600764e-05,
+      "loss": 1.6923,
+      "step": 8152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.20337551608554e-05,
+      "loss": 2.0246,
+      "step": 8153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.198717932691199e-05,
+      "loss": 2.1119,
+      "step": 8154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1940621928513816e-05,
+      "loss": 1.7471,
+      "step": 8155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1894082969995344e-05,
+      "loss": 1.6284,
+      "step": 8156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.184756245568953e-05,
+      "loss": 1.9014,
+      "step": 8157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1801060389927604e-05,
+      "loss": 1.7664,
+      "step": 8158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.175457677703901e-05,
+      "loss": 1.6125,
+      "step": 8159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1708111621351526e-05,
+      "loss": 1.9183,
+      "step": 8160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1661664927191235e-05,
+      "loss": 1.8833,
+      "step": 8161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.161523669888235e-05,
+      "loss": 1.7865,
+      "step": 8162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.156882694074749e-05,
+      "loss": 1.6138,
+      "step": 8163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.152243565710754e-05,
+      "loss": 1.7827,
+      "step": 8164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.147606285228171e-05,
+      "loss": 1.856,
+      "step": 8165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.142970853058743e-05,
+      "loss": 1.9362,
+      "step": 8166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.138337269634033e-05,
+      "loss": 1.9277,
+      "step": 8167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.133705535385447e-05,
+      "loss": 1.8005,
+      "step": 8168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.129075650744211e-05,
+      "loss": 1.915,
+      "step": 8169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.124447616141381e-05,
+      "loss": 1.9044,
+      "step": 8170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.11982143200784e-05,
+      "loss": 1.6917,
+      "step": 8171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.115197098774302e-05,
+      "loss": 1.7883,
+      "step": 8172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.110574616871297e-05,
+      "loss": 1.8372,
+      "step": 8173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.105953986729195e-05,
+      "loss": 1.8331,
+      "step": 8174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1013352087781904e-05,
+      "loss": 1.7623,
+      "step": 8175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0967182834483124e-05,
+      "loss": 1.7438,
+      "step": 8176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.092103211169391e-05,
+      "loss": 1.7735,
+      "step": 8177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.087489992371114e-05,
+      "loss": 1.8077,
+      "step": 8178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0828786274829894e-05,
+      "loss": 2.0298,
+      "step": 8179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.078269116934334e-05,
+      "loss": 1.7644,
+      "step": 8180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.073661461154316e-05,
+      "loss": 1.8637,
+      "step": 8181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0690556605719195e-05,
+      "loss": 1.682,
+      "step": 8182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.064451715615962e-05,
+      "loss": 1.9437,
+      "step": 8183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.059849626715071e-05,
+      "loss": 1.8551,
+      "step": 8184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.055249394297723e-05,
+      "loss": 1.9324,
+      "step": 8185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.050651018792213e-05,
+      "loss": 1.7856,
+      "step": 8186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0460545006266565e-05,
+      "loss": 1.7985,
+      "step": 8187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.041459840229007e-05,
+      "loss": 1.695,
+      "step": 8188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.036867038027046e-05,
+      "loss": 1.6765,
+      "step": 8189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.032276094448362e-05,
+      "loss": 1.7148,
+      "step": 8190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.027687009920393e-05,
+      "loss": 2.0138,
+      "step": 8191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.023099784870394e-05,
+      "loss": 1.9323,
+      "step": 8192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.018514419725451e-05,
+      "loss": 1.7363,
+      "step": 8193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.013930914912476e-05,
+      "loss": 1.9224,
+      "step": 8194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.009349270858196e-05,
+      "loss": 1.9737,
+      "step": 8195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.004769487989183e-05,
+      "loss": 1.7677,
+      "step": 8196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.000191566731824e-05,
+      "loss": 1.7939,
+      "step": 8197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9956155075123384e-05,
+      "loss": 1.8431,
+      "step": 8198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9910413107567693e-05,
+      "loss": 1.6914,
+      "step": 8199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.986468976890993e-05,
+      "loss": 1.6808,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.8815739154815674,
+      "eval_runtime": 184.1189,
+      "eval_samples_per_second": 457.335,
+      "eval_steps_per_second": 28.585,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981898506340696e-05,
+      "loss": 1.6047,
+      "step": 8201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977329899531405e-05,
+      "loss": 1.6697,
+      "step": 8202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9727631568884715e-05,
+      "loss": 1.7765,
+      "step": 8203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9681982788370746e-05,
+      "loss": 2.0061,
+      "step": 8204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.963635265802219e-05,
+      "loss": 1.9558,
+      "step": 8205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9590741182087254e-05,
+      "loss": 1.7245,
+      "step": 8206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.954514836481253e-05,
+      "loss": 1.7649,
+      "step": 8207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.949957421044282e-05,
+      "loss": 1.5857,
+      "step": 8208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.945401872322131e-05,
+      "loss": 1.701,
+      "step": 8209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9408481907389196e-05,
+      "loss": 1.7649,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.936296376718616e-05,
+      "loss": 1.7438,
+      "step": 8211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.931746430685002e-05,
+      "loss": 1.6349,
+      "step": 8212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9271983530617046e-05,
+      "loss": 1.9097,
+      "step": 8213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.922652144272141e-05,
+      "loss": 1.7057,
+      "step": 8214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.918107804739588e-05,
+      "loss": 1.9359,
+      "step": 8215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.913565334887135e-05,
+      "loss": 1.9773,
+      "step": 8216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9090247351377044e-05,
+      "loss": 1.9483,
+      "step": 8217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.904486005914027e-05,
+      "loss": 1.7793,
+      "step": 8218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8999491476386745e-05,
+      "loss": 1.8174,
+      "step": 8219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.895414160734046e-05,
+      "loss": 1.7839,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.890881045622358e-05,
+      "loss": 1.9017,
+      "step": 8221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.886349802725662e-05,
+      "loss": 1.7255,
+      "step": 8222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.881820432465819e-05,
+      "loss": 1.8127,
+      "step": 8223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8772929352645296e-05,
+      "loss": 1.885,
+      "step": 8224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.872767311543319e-05,
+      "loss": 1.8024,
+      "step": 8225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8682435617235344e-05,
+      "loss": 2.0092,
+      "step": 8226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8637216862263494e-05,
+      "loss": 1.8188,
+      "step": 8227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8592016854727695e-05,
+      "loss": 1.9627,
+      "step": 8228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8546835598836075e-05,
+      "loss": 1.7328,
+      "step": 8229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.850167309879519e-05,
+      "loss": 1.7523,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.845652935880979e-05,
+      "loss": 1.8371,
+      "step": 8231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.841140438308289e-05,
+      "loss": 1.9031,
+      "step": 8232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.836629817581581e-05,
+      "loss": 1.7693,
+      "step": 8233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.832121074120793e-05,
+      "loss": 1.9617,
+      "step": 8234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.82761420834571e-05,
+      "loss": 1.8555,
+      "step": 8235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.823109220675934e-05,
+      "loss": 1.986,
+      "step": 8236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8186061115308875e-05,
+      "loss": 1.7613,
+      "step": 8237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8141048813298285e-05,
+      "loss": 1.8145,
+      "step": 8238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.809605530491837e-05,
+      "loss": 1.9355,
+      "step": 8239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8051080594358e-05,
+      "loss": 1.952,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.800612468580456e-05,
+      "loss": 1.8696,
+      "step": 8241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.796118758344353e-05,
+      "loss": 1.7918,
+      "step": 8242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.791626929145879e-05,
+      "loss": 1.7371,
+      "step": 8243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.787136981403217e-05,
+      "loss": 1.985,
+      "step": 8244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782648915534407e-05,
+      "loss": 1.8789,
+      "step": 8245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7781627319573e-05,
+      "loss": 1.9307,
+      "step": 8246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.773678431089565e-05,
+      "loss": 1.8494,
+      "step": 8247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.769196013348703e-05,
+      "loss": 1.8206,
+      "step": 8248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7647154791520455e-05,
+      "loss": 1.811,
+      "step": 8249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7602368289167476e-05,
+      "loss": 1.7688,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.8794952630996704,
+      "eval_runtime": 184.2123,
+      "eval_samples_per_second": 457.103,
+      "eval_steps_per_second": 28.57,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.755760063059769e-05,
+      "loss": 1.8384,
+      "step": 8251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.751285181997919e-05,
+      "loss": 1.8535,
+      "step": 8252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.746812186147817e-05,
+      "loss": 1.7907,
+      "step": 8253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7423410759259156e-05,
+      "loss": 1.8062,
+      "step": 8254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.737871851748485e-05,
+      "loss": 1.7318,
+      "step": 8255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.733404514031631e-05,
+      "loss": 2.0463,
+      "step": 8256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.728939063191262e-05,
+      "loss": 1.8617,
+      "step": 8257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.724475499643127e-05,
+      "loss": 1.8487,
+      "step": 8258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.720013823802799e-05,
+      "loss": 1.8866,
+      "step": 8259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.715554036085673e-05,
+      "loss": 1.8009,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7110961369069755e-05,
+      "loss": 1.9346,
+      "step": 8261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.70664012668173e-05,
+      "loss": 1.8706,
+      "step": 8262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7021860058248154e-05,
+      "loss": 1.6662,
+      "step": 8263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.697733774750923e-05,
+      "loss": 1.7661,
+      "step": 8264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.693283433874565e-05,
+      "loss": 1.7148,
+      "step": 8265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6888349836100825e-05,
+      "loss": 1.9103,
+      "step": 8266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.684388424371644e-05,
+      "loss": 1.8117,
+      "step": 8267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.679943756573224e-05,
+      "loss": 1.9244,
+      "step": 8268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6755009806286414e-05,
+      "loss": 1.8542,
+      "step": 8269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6710600969515266e-05,
+      "loss": 1.8908,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.666621105955343e-05,
+      "loss": 1.6449,
+      "step": 8271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.66218400805338e-05,
+      "loss": 1.6711,
+      "step": 8272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.657748803658726e-05,
+      "loss": 1.9375,
+      "step": 8273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.653315493184321e-05,
+      "loss": 1.7567,
+      "step": 8274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.648884077042914e-05,
+      "loss": 1.7839,
+      "step": 8275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6444545556470955e-05,
+      "loss": 1.8407,
+      "step": 8276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6400269294092494e-05,
+      "loss": 2.0669,
+      "step": 8277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6356011987416066e-05,
+      "loss": 2.014,
+      "step": 8278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.631177364056224e-05,
+      "loss": 1.8287,
+      "step": 8279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.626755425764956e-05,
+      "loss": 1.9503,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.622335384279505e-05,
+      "loss": 1.781,
+      "step": 8281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.617917240011393e-05,
+      "loss": 1.7044,
+      "step": 8282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.613500993371955e-05,
+      "loss": 1.8779,
+      "step": 8283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6090866447723676e-05,
+      "loss": 1.7764,
+      "step": 8284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6046741946236036e-05,
+      "loss": 1.8626,
+      "step": 8285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600263643336483e-05,
+      "loss": 1.9356,
+      "step": 8286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.595854991321638e-05,
+      "loss": 1.8749,
+      "step": 8287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.591448238989526e-05,
+      "loss": 1.6887,
+      "step": 8288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.587043386750436e-05,
+      "loss": 1.6064,
+      "step": 8289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.58264043501446e-05,
+      "loss": 1.6247,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5782393841915294e-05,
+      "loss": 1.9093,
+      "step": 8291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.573840234691395e-05,
+      "loss": 2.0253,
+      "step": 8292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.56944298692363e-05,
+      "loss": 1.8044,
+      "step": 8293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5650476412976315e-05,
+      "loss": 2.0772,
+      "step": 8294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.560654198222622e-05,
+      "loss": 1.815,
+      "step": 8295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.556262658107635e-05,
+      "loss": 1.8791,
+      "step": 8296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.551873021361536e-05,
+      "loss": 1.8954,
+      "step": 8297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.547485288393016e-05,
+      "loss": 1.8561,
+      "step": 8298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5430994596105863e-05,
+      "loss": 1.9009,
+      "step": 8299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5387155354225826e-05,
+      "loss": 2.0228,
+      "step": 8300
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 1.8767882585525513,
+      "eval_runtime": 184.1465,
+      "eval_samples_per_second": 457.266,
+      "eval_steps_per_second": 28.581,
+      "step": 8300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.534333516237149e-05,
+      "loss": 1.6694,
+      "step": 8301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5299534024622694e-05,
+      "loss": 1.7412,
+      "step": 8302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.525575194505746e-05,
+      "loss": 1.907,
+      "step": 8303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.521198892775202e-05,
+      "loss": 1.8068,
+      "step": 8304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.516824497678085e-05,
+      "loss": 1.7137,
+      "step": 8305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.512452009621665e-05,
+      "loss": 1.7872,
+      "step": 8306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5080814290130254e-05,
+      "loss": 1.9263,
+      "step": 8307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.503712756259082e-05,
+      "loss": 1.8294,
+      "step": 8308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.499345991766576e-05,
+      "loss": 1.8239,
+      "step": 8309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.494981135942055e-05,
+      "loss": 1.9413,
+      "step": 8310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.490618189191908e-05,
+      "loss": 1.9866,
+      "step": 8311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4862571519223404e-05,
+      "loss": 1.8432,
+      "step": 8312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.481898024539363e-05,
+      "loss": 1.8612,
+      "step": 8313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.477540807448832e-05,
+      "loss": 1.7265,
+      "step": 8314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.473185501056415e-05,
+      "loss": 1.8935,
+      "step": 8315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.468832105767603e-05,
+      "loss": 1.7455,
+      "step": 8316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.464480621987718e-05,
+      "loss": 1.8916,
+      "step": 8317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4601310501218794e-05,
+      "loss": 1.8407,
+      "step": 8318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4557833905750524e-05,
+      "loss": 1.805,
+      "step": 8319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.451437643752013e-05,
+      "loss": 1.9097,
+      "step": 8320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4470938100573675e-05,
+      "loss": 1.812,
+      "step": 8321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.442751889895538e-05,
+      "loss": 1.684,
+      "step": 8322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4384118836707734e-05,
+      "loss": 1.7625,
+      "step": 8323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.434073791787127e-05,
+      "loss": 2.0378,
+      "step": 8324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4297376146484965e-05,
+      "loss": 1.7311,
+      "step": 8325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.425403352658591e-05,
+      "loss": 1.9011,
+      "step": 8326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.421071006220942e-05,
+      "loss": 1.7549,
+      "step": 8327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.416740575738909e-05,
+      "loss": 1.8096,
+      "step": 8328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4124120616156534e-05,
+      "loss": 1.8425,
+      "step": 8329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.408085464254183e-05,
+      "loss": 1.7531,
+      "step": 8330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.403760784057312e-05,
+      "loss": 1.6522,
+      "step": 8331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.399438021427679e-05,
+      "loss": 1.9507,
+      "step": 8332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.395117176767749e-05,
+      "loss": 1.8214,
+      "step": 8333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.390798250479808e-05,
+      "loss": 1.7091,
+      "step": 8334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.386481242965951e-05,
+      "loss": 1.9065,
+      "step": 8335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.382166154628106e-05,
+      "loss": 1.9201,
+      "step": 8336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.377852985868019e-05,
+      "loss": 1.8704,
+      "step": 8337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.373541737087264e-05,
+      "loss": 1.7574,
+      "step": 8338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3692324086872235e-05,
+      "loss": 1.7184,
+      "step": 8339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.36492500106912e-05,
+      "loss": 1.9622,
+      "step": 8340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3606195146339676e-05,
+      "loss": 1.6027,
+      "step": 8341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.356315949782627e-05,
+      "loss": 1.9759,
+      "step": 8342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.35201430691578e-05,
+      "loss": 1.6273,
+      "step": 8343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.34771458643391e-05,
+      "loss": 1.6318,
+      "step": 8344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.343416788737334e-05,
+      "loss": 1.7496,
+      "step": 8345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.339120914226199e-05,
+      "loss": 1.9987,
+      "step": 8346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.334826963300451e-05,
+      "loss": 1.8474,
+      "step": 8347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3305349363598726e-05,
+      "loss": 1.8641,
+      "step": 8348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3262448338040664e-05,
+      "loss": 1.5426,
+      "step": 8349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3219566560324494e-05,
+      "loss": 1.9403,
+      "step": 8350
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 1.875037670135498,
+      "eval_runtime": 184.1497,
+      "eval_samples_per_second": 457.258,
+      "eval_steps_per_second": 28.58,
+      "step": 8350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.317670403444271e-05,
+      "loss": 1.9739,
+      "step": 8351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.31338607643858e-05,
+      "loss": 1.6977,
+      "step": 8352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.309103675414266e-05,
+      "loss": 1.8897,
+      "step": 8353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.304823200770036e-05,
+      "loss": 1.8803,
+      "step": 8354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.300544652904406e-05,
+      "loss": 2.0823,
+      "step": 8355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.296268032215733e-05,
+      "loss": 1.7939,
+      "step": 8356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.291993339102171e-05,
+      "loss": 2.1296,
+      "step": 8357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2877205739617064e-05,
+      "loss": 1.953,
+      "step": 8358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.283449737192149e-05,
+      "loss": 1.8549,
+      "step": 8359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.279180829191126e-05,
+      "loss": 1.7387,
+      "step": 8360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2749138503560854e-05,
+      "loss": 1.998,
+      "step": 8361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.270648801084295e-05,
+      "loss": 1.7204,
+      "step": 8362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.266385681772839e-05,
+      "loss": 1.7817,
+      "step": 8363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.262124492818625e-05,
+      "loss": 2.0218,
+      "step": 8364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.257865234618386e-05,
+      "loss": 1.6624,
+      "step": 8365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.253607907568669e-05,
+      "loss": 1.6197,
+      "step": 8366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.249352512065843e-05,
+      "loss": 1.7398,
+      "step": 8367
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.245099048506102e-05,
+      "loss": 1.9078,
+      "step": 8368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.240847517285446e-05,
+      "loss": 1.8529,
+      "step": 8369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.236597918799709e-05,
+      "loss": 1.6856,
+      "step": 8370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2323502534445393e-05,
+      "loss": 1.7808,
+      "step": 8371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.228104521615406e-05,
+      "loss": 1.9588,
+      "step": 8372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.223860723707609e-05,
+      "loss": 1.9416,
+      "step": 8373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2196188601162425e-05,
+      "loss": 1.8553,
+      "step": 8374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2153789312362386e-05,
+      "loss": 1.7188,
+      "step": 8375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.21114093746236e-05,
+      "loss": 1.8821,
+      "step": 8376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.206904879189158e-05,
+      "loss": 1.8542,
+      "step": 8377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.202670756811028e-05,
+      "loss": 1.9767,
+      "step": 8378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1984385707221875e-05,
+      "loss": 2.0435,
+      "step": 8379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194208321316651e-05,
+      "loss": 1.7409,
+      "step": 8380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1899800089882727e-05,
+      "loss": 1.6631,
+      "step": 8381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.185753634130718e-05,
+      "loss": 1.9518,
+      "step": 8382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.181529197137476e-05,
+      "loss": 1.8372,
+      "step": 8383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.177306698401864e-05,
+      "loss": 1.7564,
+      "step": 8384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.173086138316989e-05,
+      "loss": 1.752,
+      "step": 8385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1688675172758066e-05,
+      "loss": 2.0991,
+      "step": 8386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.16465083567108e-05,
+      "loss": 1.7158,
+      "step": 8387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.160436093895398e-05,
+      "loss": 1.897,
+      "step": 8388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.15622329234116e-05,
+      "loss": 1.7758,
+      "step": 8389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.152012431400601e-05,
+      "loss": 1.878,
+      "step": 8390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.147803511465748e-05,
+      "loss": 1.8652,
+      "step": 8391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1435965329284673e-05,
+      "loss": 1.7876,
+      "step": 8392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.139391496180448e-05,
+      "loss": 1.5674,
+      "step": 8393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1351884016131826e-05,
+      "loss": 1.6745,
+      "step": 8394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1309872496179926e-05,
+      "loss": 1.9683,
+      "step": 8395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.126788040586027e-05,
+      "loss": 1.7196,
+      "step": 8396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.12259077490823e-05,
+      "loss": 1.7793,
+      "step": 8397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.118395452975382e-05,
+      "loss": 1.942,
+      "step": 8398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1142020751780834e-05,
+      "loss": 1.8049,
+      "step": 8399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.110010641906744e-05,
+      "loss": 1.8574,
+      "step": 8400
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.8727118968963623,
+      "eval_runtime": 184.1468,
+      "eval_samples_per_second": 457.266,
+      "eval_steps_per_second": 28.58,
+      "step": 8400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1058211535516086e-05,
+      "loss": 1.8433,
+      "step": 8401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.101633610502717e-05,
+      "loss": 1.7726,
+      "step": 8402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0974480131499484e-05,
+      "loss": 1.7714,
+      "step": 8403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.09326436188299e-05,
+      "loss": 1.9962,
+      "step": 8404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.089082657091353e-05,
+      "loss": 1.818,
+      "step": 8405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.084902899164372e-05,
+      "loss": 1.836,
+      "step": 8406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.080725088491183e-05,
+      "loss": 1.7024,
+      "step": 8407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.076549225460757e-05,
+      "loss": 1.9282,
+      "step": 8408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.072375310461887e-05,
+      "loss": 1.8305,
+      "step": 8409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0682033438831586e-05,
+      "loss": 1.8826,
+      "step": 8410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0640333261130034e-05,
+      "loss": 1.8187,
+      "step": 8411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.059865257539666e-05,
+      "loss": 1.8458,
+      "step": 8412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.055699138551197e-05,
+      "loss": 2.0445,
+      "step": 8413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0515349695354726e-05,
+      "loss": 1.878,
+      "step": 8414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0473727508801964e-05,
+      "loss": 2.0236,
+      "step": 8415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.043212482972877e-05,
+      "loss": 1.821,
+      "step": 8416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.039054166200848e-05,
+      "loss": 1.7836,
+      "step": 8417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.034897800951268e-05,
+      "loss": 1.8737,
+      "step": 8418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.030743387611091e-05,
+      "loss": 1.9244,
+      "step": 8419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.026590926567117e-05,
+      "loss": 1.799,
+      "step": 8420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0224404182059444e-05,
+      "loss": 1.9579,
+      "step": 8421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0182918629140004e-05,
+      "loss": 1.8476,
+      "step": 8422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0141452610775255e-05,
+      "loss": 1.6541,
+      "step": 8423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.01000061308259e-05,
+      "loss": 1.669,
+      "step": 8424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0058579193150535e-05,
+      "loss": 1.9646,
+      "step": 8425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0017171801606246e-05,
+      "loss": 1.8911,
+      "step": 8426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.997578396004817e-05,
+      "loss": 1.821,
+      "step": 8427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9934415672329586e-05,
+      "loss": 1.8839,
+      "step": 8428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9893066942302076e-05,
+      "loss": 1.8433,
+      "step": 8429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.985173777381521e-05,
+      "loss": 1.8891,
+      "step": 8430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.981042817071692e-05,
+      "loss": 1.9486,
+      "step": 8431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9769138136853205e-05,
+      "loss": 1.6475,
+      "step": 8432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972786767606834e-05,
+      "loss": 1.7151,
+      "step": 8433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.968661679220467e-05,
+      "loss": 1.662,
+      "step": 8434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9645385489102876e-05,
+      "loss": 1.8656,
+      "step": 8435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.960417377060152e-05,
+      "loss": 1.7822,
+      "step": 8436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9562981640537647e-05,
+      "loss": 1.7812,
+      "step": 8437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.952180910274633e-05,
+      "loss": 1.8565,
+      "step": 8438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9480656161060865e-05,
+      "loss": 1.6252,
+      "step": 8439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.943952281931273e-05,
+      "loss": 1.9024,
+      "step": 8440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.939840908133146e-05,
+      "loss": 1.8292,
+      "step": 8441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.935731495094494e-05,
+      "loss": 1.8775,
+      "step": 8442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.931624043197918e-05,
+      "loss": 1.7542,
+      "step": 8443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.927518552825821e-05,
+      "loss": 2.0223,
+      "step": 8444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9234150243604413e-05,
+      "loss": 1.7774,
+      "step": 8445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.919313458183837e-05,
+      "loss": 1.8119,
+      "step": 8446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9152138546778624e-05,
+      "loss": 1.8757,
+      "step": 8447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.911116214224208e-05,
+      "loss": 1.9422,
+      "step": 8448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.907020537204373e-05,
+      "loss": 1.9282,
+      "step": 8449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9029268239996806e-05,
+      "loss": 1.8918,
+      "step": 8450
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.8711286783218384,
+      "eval_runtime": 184.4726,
+      "eval_samples_per_second": 456.458,
+      "eval_steps_per_second": 28.53,
+      "step": 8450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8988350749912635e-05,
+      "loss": 1.9275,
+      "step": 8451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.894745290560084e-05,
+      "loss": 1.992,
+      "step": 8452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8906574710869e-05,
+      "loss": 2.0034,
+      "step": 8453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8865716169523e-05,
+      "loss": 2.0441,
+      "step": 8454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8824877285366964e-05,
+      "loss": 1.8568,
+      "step": 8455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.878405806220303e-05,
+      "loss": 1.9639,
+      "step": 8456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8743258503831674e-05,
+      "loss": 2.001,
+      "step": 8457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.870247861405135e-05,
+      "loss": 1.8275,
+      "step": 8458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.86617183966588e-05,
+      "loss": 1.7789,
+      "step": 8459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.862097785544894e-05,
+      "loss": 1.7146,
+      "step": 8460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.85802569942148e-05,
+      "loss": 1.863,
+      "step": 8461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8539555816747646e-05,
+      "loss": 1.7716,
+      "step": 8462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8498874326836906e-05,
+      "loss": 1.8618,
+      "step": 8463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.845821252827003e-05,
+      "loss": 1.971,
+      "step": 8464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.841757042483279e-05,
+      "loss": 1.9736,
+      "step": 8465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8376948020309085e-05,
+      "loss": 1.91,
+      "step": 8466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8336345318480984e-05,
+      "loss": 1.754,
+      "step": 8467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.829576232312876e-05,
+      "loss": 1.6544,
+      "step": 8468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.825519903803068e-05,
+      "loss": 1.8715,
+      "step": 8469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.821465546696337e-05,
+      "loss": 2.1196,
+      "step": 8470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.817413161370156e-05,
+      "loss": 1.7617,
+      "step": 8471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8133627482018096e-05,
+      "loss": 1.835,
+      "step": 8472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.809314307568412e-05,
+      "loss": 2.0081,
+      "step": 8473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8052678398468736e-05,
+      "loss": 1.9946,
+      "step": 8474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8012233454139315e-05,
+      "loss": 2.0451,
+      "step": 8475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.797180824646151e-05,
+      "loss": 1.8065,
+      "step": 8476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7931402779198876e-05,
+      "loss": 1.8427,
+      "step": 8477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7891017056113365e-05,
+      "loss": 1.6412,
+      "step": 8478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7850651080964945e-05,
+      "loss": 1.8779,
+      "step": 8479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7810304857511914e-05,
+      "loss": 1.8093,
+      "step": 8480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7769978389510465e-05,
+      "loss": 1.7239,
+      "step": 8481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7729671680715175e-05,
+      "loss": 1.8284,
+      "step": 8482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.768938473487873e-05,
+      "loss": 1.7893,
+      "step": 8483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.764911755575193e-05,
+      "loss": 1.9696,
+      "step": 8484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7608870147083805e-05,
+      "loss": 1.8349,
+      "step": 8485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.756864251262143e-05,
+      "loss": 1.7535,
+      "step": 8486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.752843465611017e-05,
+      "loss": 1.799,
+      "step": 8487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.748824658129343e-05,
+      "loss": 2.0496,
+      "step": 8488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.744807829191288e-05,
+      "loss": 1.8145,
+      "step": 8489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.740792979170826e-05,
+      "loss": 1.981,
+      "step": 8490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7367801084417615e-05,
+      "loss": 2.0014,
+      "step": 8491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.73276921737769e-05,
+      "loss": 1.8832,
+      "step": 8492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.728760306352042e-05,
+      "loss": 1.8933,
+      "step": 8493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7247533757380604e-05,
+      "loss": 1.6677,
+      "step": 8494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.720748425908796e-05,
+      "loss": 1.9066,
+      "step": 8495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7167454572371334e-05,
+      "loss": 1.6407,
+      "step": 8496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.712744470095744e-05,
+      "loss": 1.8151,
+      "step": 8497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.708745464857138e-05,
+      "loss": 1.8668,
+      "step": 8498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.704748441893638e-05,
+      "loss": 1.6926,
+      "step": 8499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.70075340157737e-05,
+      "loss": 1.7663,
+      "step": 8500
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.8691030740737915,
+      "eval_runtime": 184.9543,
+      "eval_samples_per_second": 455.269,
+      "eval_steps_per_second": 28.456,
+      "step": 8500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6967603442802895e-05,
+      "loss": 1.8436,
+      "step": 8501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.692769270374163e-05,
+      "loss": 1.6516,
+      "step": 8502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6887801802305636e-05,
+      "loss": 1.7033,
+      "step": 8503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.684793074220888e-05,
+      "loss": 2.1272,
+      "step": 8504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6808079527163504e-05,
+      "loss": 1.6637,
+      "step": 8505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.676824816087978e-05,
+      "loss": 2.0796,
+      "step": 8506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.672843664706607e-05,
+      "loss": 1.9045,
+      "step": 8507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.668864498942892e-05,
+      "loss": 1.965,
+      "step": 8508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6648873191673175e-05,
+      "loss": 1.676,
+      "step": 8509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.660912125750152e-05,
+      "loss": 1.7245,
+      "step": 8510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.656938919061509e-05,
+      "loss": 1.8914,
+      "step": 8511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.652967699471299e-05,
+      "loss": 1.8746,
+      "step": 8512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.648998467349263e-05,
+      "loss": 1.6835,
+      "step": 8513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.645031223064935e-05,
+      "loss": 1.8573,
+      "step": 8514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6410659669876834e-05,
+      "loss": 1.7361,
+      "step": 8515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6371026994866826e-05,
+      "loss": 1.6506,
+      "step": 8516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633141420930927e-05,
+      "loss": 1.9913,
+      "step": 8517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6291821316892184e-05,
+      "loss": 1.6476,
+      "step": 8518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625224832130186e-05,
+      "loss": 1.9106,
+      "step": 8519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621269522622256e-05,
+      "loss": 1.6371,
+      "step": 8520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6173162035336796e-05,
+      "loss": 1.9626,
+      "step": 8521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613364875232525e-05,
+      "loss": 1.9524,
+      "step": 8522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.609415538086669e-05,
+      "loss": 1.8581,
+      "step": 8523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.605468192463815e-05,
+      "loss": 1.6969,
+      "step": 8524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.601522838731461e-05,
+      "loss": 1.9105,
+      "step": 8525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.597579477256932e-05,
+      "loss": 1.8809,
+      "step": 8526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.593638108407371e-05,
+      "loss": 1.8484,
+      "step": 8527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.589698732549726e-05,
+      "loss": 1.9778,
+      "step": 8528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5857613500507644e-05,
+      "loss": 1.6201,
+      "step": 8529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.581825961277074e-05,
+      "loss": 1.818,
+      "step": 8530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.577892566595042e-05,
+      "loss": 1.8605,
+      "step": 8531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.573961166370879e-05,
+      "loss": 1.7958,
+      "step": 8532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.570031760970613e-05,
+      "loss": 1.9414,
+      "step": 8533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5661043507600825e-05,
+      "loss": 1.9127,
+      "step": 8534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.562178936104943e-05,
+      "loss": 1.7658,
+      "step": 8535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.558255517370651e-05,
+      "loss": 1.8075,
+      "step": 8536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5543340949224984e-05,
+      "loss": 1.7951,
+      "step": 8537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.550414669125573e-05,
+      "loss": 1.7374,
+      "step": 8538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5464972403447956e-05,
+      "loss": 1.9195,
+      "step": 8539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.542581808944875e-05,
+      "loss": 1.9477,
+      "step": 8540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.538668375290355e-05,
+      "loss": 1.824,
+      "step": 8541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5347569397455894e-05,
+      "loss": 1.7304,
+      "step": 8542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.530847502674747e-05,
+      "loss": 1.7991,
+      "step": 8543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5269400644417996e-05,
+      "loss": 1.6626,
+      "step": 8544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.52303462541054e-05,
+      "loss": 1.8482,
+      "step": 8545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5191311859445795e-05,
+      "loss": 1.7497,
+      "step": 8546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.515229746407347e-05,
+      "loss": 2.0074,
+      "step": 8547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.511330307162061e-05,
+      "loss": 1.6131,
+      "step": 8548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.507432868571778e-05,
+      "loss": 1.7708,
+      "step": 8549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5035374309993624e-05,
+      "loss": 1.7333,
+      "step": 8550
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 1.867645263671875,
+      "eval_runtime": 184.1514,
+      "eval_samples_per_second": 457.254,
+      "eval_steps_per_second": 28.58,
+      "step": 8550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.499643994807486e-05,
+      "loss": 1.9128,
+      "step": 8551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4957525603586504e-05,
+      "loss": 1.7455,
+      "step": 8552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491863128015141e-05,
+      "loss": 1.9574,
+      "step": 8553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.487975698139084e-05,
+      "loss": 2.0311,
+      "step": 8554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.48409027109241e-05,
+      "loss": 2.0808,
+      "step": 8555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.480206847236861e-05,
+      "loss": 1.8672,
+      "step": 8556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4763254269339964e-05,
+      "loss": 1.8036,
+      "step": 8557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4724460105451936e-05,
+      "loss": 1.8846,
+      "step": 8558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.468568598431624e-05,
+      "loss": 1.7407,
+      "step": 8559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4646931909542935e-05,
+      "loss": 1.8185,
+      "step": 8560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.460819788474009e-05,
+      "loss": 1.8947,
+      "step": 8561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4569483913514e-05,
+      "loss": 1.9032,
+      "step": 8562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.453078999946904e-05,
+      "loss": 1.7464,
+      "step": 8563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449211614620768e-05,
+      "loss": 1.8072,
+      "step": 8564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4453462357330574e-05,
+      "loss": 1.711,
+      "step": 8565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4414828636436526e-05,
+      "loss": 1.9151,
+      "step": 8566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43762149871224e-05,
+      "loss": 1.849,
+      "step": 8567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433762141298327e-05,
+      "loss": 1.8632,
+      "step": 8568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.429904791761235e-05,
+      "loss": 1.8799,
+      "step": 8569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4260494504600836e-05,
+      "loss": 1.7847,
+      "step": 8570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.42219611775382e-05,
+      "loss": 1.9492,
+      "step": 8571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.418344794001199e-05,
+      "loss": 1.6878,
+      "step": 8572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4144954795608005e-05,
+      "loss": 1.8346,
+      "step": 8573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.41064817479099e-05,
+      "loss": 1.8465,
+      "step": 8574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.40680288004997e-05,
+      "loss": 1.8547,
+      "step": 8575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.402959595695754e-05,
+      "loss": 1.912,
+      "step": 8576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3991183220861504e-05,
+      "loss": 1.899,
+      "step": 8577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.395279059578798e-05,
+      "loss": 1.6245,
+      "step": 8578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.391441808531143e-05,
+      "loss": 1.816,
+      "step": 8579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.38760656930045e-05,
+      "loss": 1.7978,
+      "step": 8580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.383773342243779e-05,
+      "loss": 1.6933,
+      "step": 8581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.379942127718022e-05,
+      "loss": 1.8039,
+      "step": 8582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.376112926079872e-05,
+      "loss": 1.767,
+      "step": 8583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.37228573768584e-05,
+      "loss": 1.9119,
+      "step": 8584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.368460562892248e-05,
+      "loss": 1.7017,
+      "step": 8585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3646374020552346e-05,
+      "loss": 2.14,
+      "step": 8586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3608162555307384e-05,
+      "loss": 1.8006,
+      "step": 8587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.356997123674524e-05,
+      "loss": 1.9029,
+      "step": 8588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3531800068421603e-05,
+      "loss": 1.8525,
+      "step": 8589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3493649053890325e-05,
+      "loss": 1.7411,
+      "step": 8590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.345551819670345e-05,
+      "loss": 1.855,
+      "step": 8591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.341740750041095e-05,
+      "loss": 1.8843,
+      "step": 8592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.337931696856106e-05,
+      "loss": 1.7254,
+      "step": 8593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.334124660470017e-05,
+      "loss": 1.5834,
+      "step": 8594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.330319641237267e-05,
+      "loss": 1.7943,
+      "step": 8595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.32651663951212e-05,
+      "loss": 1.7641,
+      "step": 8596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3227156556486476e-05,
+      "loss": 1.8986,
+      "step": 8597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3189166900007226e-05,
+      "loss": 1.8752,
+      "step": 8598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.315119742922043e-05,
+      "loss": 2.0124,
+      "step": 8599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.31132481476612e-05,
+      "loss": 1.5394,
+      "step": 8600
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 1.8652031421661377,
+      "eval_runtime": 184.1482,
+      "eval_samples_per_second": 457.262,
+      "eval_steps_per_second": 28.58,
+      "step": 8600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3075319058862654e-05,
+      "loss": 1.8783,
+      "step": 8601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.303741016635614e-05,
+      "loss": 1.8119,
+      "step": 8602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2999521473671136e-05,
+      "loss": 1.7137,
+      "step": 8603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2961652984335065e-05,
+      "loss": 1.9529,
+      "step": 8604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.292380470187365e-05,
+      "loss": 1.7872,
+      "step": 8605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.288597662981069e-05,
+      "loss": 1.8766,
+      "step": 8606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.284816877166802e-05,
+      "loss": 1.8419,
+      "step": 8607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.281038113096568e-05,
+      "loss": 1.9597,
+      "step": 8608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2772613711221894e-05,
+      "loss": 1.6823,
+      "step": 8609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2734866515952757e-05,
+      "loss": 1.978,
+      "step": 8610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2697139548672736e-05,
+      "loss": 1.8441,
+      "step": 8611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2659432812894296e-05,
+      "loss": 1.9439,
+      "step": 8612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2621746312128055e-05,
+      "loss": 1.7155,
+      "step": 8613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2584080049882784e-05,
+      "loss": 1.8942,
+      "step": 8614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.254643402966517e-05,
+      "loss": 1.7951,
+      "step": 8615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.250880825498026e-05,
+      "loss": 1.6087,
+      "step": 8616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2471202729331126e-05,
+      "loss": 1.6789,
+      "step": 8617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.243361745621889e-05,
+      "loss": 1.8859,
+      "step": 8618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.239605243914295e-05,
+      "loss": 1.9336,
+      "step": 8619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2358507681600593e-05,
+      "loss": 1.7612,
+      "step": 8620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.23209831870874e-05,
+      "loss": 1.8876,
+      "step": 8621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2283478959096975e-05,
+      "loss": 1.6466,
+      "step": 8622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.22459950011211e-05,
+      "loss": 1.7618,
+      "step": 8623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.220853131664963e-05,
+      "loss": 1.5746,
+      "step": 8624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2171087909170585e-05,
+      "loss": 1.8163,
+      "step": 8625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2133664782169944e-05,
+      "loss": 1.7496,
+      "step": 8626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.209626193913196e-05,
+      "loss": 1.8354,
+      "step": 8627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.205887938353894e-05,
+      "loss": 1.7148,
+      "step": 8628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.20215171188713e-05,
+      "loss": 1.7295,
+      "step": 8629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.19841751486076e-05,
+      "loss": 1.9871,
+      "step": 8630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1946853476224477e-05,
+      "loss": 1.787,
+      "step": 8631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.190955210519664e-05,
+      "loss": 1.9317,
+      "step": 8632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1872271038996953e-05,
+      "loss": 1.9575,
+      "step": 8633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.183501028109642e-05,
+      "loss": 1.6774,
+      "step": 8634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.17977698349641e-05,
+      "loss": 1.837,
+      "step": 8635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1760549704067246e-05,
+      "loss": 1.7518,
+      "step": 8636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.172334989187106e-05,
+      "loss": 1.6496,
+      "step": 8637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.168617040183896e-05,
+      "loss": 1.9048,
+      "step": 8638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1649011237432576e-05,
+      "loss": 1.7291,
+      "step": 8639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.161187240211139e-05,
+      "loss": 1.8982,
+      "step": 8640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.157475389933317e-05,
+      "loss": 1.9261,
+      "step": 8641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1537655732553766e-05,
+      "loss": 1.7921,
+      "step": 8642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.150057790522717e-05,
+      "loss": 1.8788,
+      "step": 8643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1463520420805334e-05,
+      "loss": 2.0905,
+      "step": 8644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1426483282738475e-05,
+      "loss": 1.5559,
+      "step": 8645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1389466494474824e-05,
+      "loss": 1.7142,
+      "step": 8646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.135247005946082e-05,
+      "loss": 1.6905,
+      "step": 8647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.131549398114081e-05,
+      "loss": 1.7502,
+      "step": 8648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1278538262957454e-05,
+      "loss": 2.0138,
+      "step": 8649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1241602908351405e-05,
+      "loss": 1.899,
+      "step": 8650
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.8644733428955078,
+      "eval_runtime": 184.1622,
+      "eval_samples_per_second": 457.227,
+      "eval_steps_per_second": 28.578,
+      "step": 8650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1204687920761474e-05,
+      "loss": 1.619,
+      "step": 8651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1167793303624514e-05,
+      "loss": 1.7045,
+      "step": 8652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.11309190603756e-05,
+      "loss": 2.0218,
+      "step": 8653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.109406519444771e-05,
+      "loss": 1.7292,
+      "step": 8654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1057231709272074e-05,
+      "loss": 1.8728,
+      "step": 8655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.102041860827804e-05,
+      "loss": 1.811,
+      "step": 8656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.098362589489298e-05,
+      "loss": 1.9675,
+      "step": 8657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.094685357254237e-05,
+      "loss": 1.7937,
+      "step": 8658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.091010164464994e-05,
+      "loss": 1.7621,
+      "step": 8659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.087337011463723e-05,
+      "loss": 1.7988,
+      "step": 8660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.083665898592411e-05,
+      "loss": 1.7939,
+      "step": 8661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.079996826192849e-05,
+      "loss": 2.0338,
+      "step": 8662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0763297946066416e-05,
+      "loss": 1.7512,
+      "step": 8663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.072664804175201e-05,
+      "loss": 1.8153,
+      "step": 8664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.069001855239739e-05,
+      "loss": 1.6335,
+      "step": 8665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.06534094814129e-05,
+      "loss": 1.7745,
+      "step": 8666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.061682083220696e-05,
+      "loss": 2.0291,
+      "step": 8667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0580252608186086e-05,
+      "loss": 1.8729,
+      "step": 8668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.054370481275484e-05,
+      "loss": 1.9123,
+      "step": 8669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0507177449316025e-05,
+      "loss": 1.9061,
+      "step": 8670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.047067052127031e-05,
+      "loss": 1.8772,
+      "step": 8671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.043418403201667e-05,
+      "loss": 1.7622,
+      "step": 8672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0397717984952112e-05,
+      "loss": 1.8062,
+      "step": 8673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.036127238347164e-05,
+      "loss": 1.7949,
+      "step": 8674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0324847230968483e-05,
+      "loss": 1.8224,
+      "step": 8675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0288442530834014e-05,
+      "loss": 1.8309,
+      "step": 8676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.025205828645747e-05,
+      "loss": 1.7091,
+      "step": 8677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0215694501226383e-05,
+      "loss": 1.6536,
+      "step": 8678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0179351178526326e-05,
+      "loss": 1.7533,
+      "step": 8679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0143028321740955e-05,
+      "loss": 1.8358,
+      "step": 8680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0106725934252095e-05,
+      "loss": 1.6436,
+      "step": 8681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0070444019439507e-05,
+      "loss": 1.6888,
+      "step": 8682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0034182580681162e-05,
+      "loss": 1.8263,
+      "step": 8683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9997941621353136e-05,
+      "loss": 1.7769,
+      "step": 8684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9961721144829534e-05,
+      "loss": 1.7326,
+      "step": 8685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9925521154482578e-05,
+      "loss": 1.8974,
+      "step": 8686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.988934165368265e-05,
+      "loss": 1.7555,
+      "step": 8687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9853182645798087e-05,
+      "loss": 1.6574,
+      "step": 8688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9817044134195392e-05,
+      "loss": 1.8663,
+      "step": 8689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9780926122239203e-05,
+      "loss": 2.1006,
+      "step": 8690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.974482861329217e-05,
+      "loss": 1.888,
+      "step": 8691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.970875161071518e-05,
+      "loss": 1.8578,
+      "step": 8692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.967269511786694e-05,
+      "loss": 1.8334,
+      "step": 8693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.963665913810451e-05,
+      "loss": 1.8691,
+      "step": 8694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9600643674782906e-05,
+      "loss": 1.7445,
+      "step": 8695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.956464873125528e-05,
+      "loss": 1.8757,
+      "step": 8696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9528674310872834e-05,
+      "loss": 1.7794,
+      "step": 8697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9492720416985004e-05,
+      "loss": 1.7944,
+      "step": 8698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9456787052939027e-05,
+      "loss": 1.7978,
+      "step": 8699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9420874222080506e-05,
+      "loss": 1.9249,
+      "step": 8700
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.8626952171325684,
+      "eval_runtime": 184.1474,
+      "eval_samples_per_second": 457.264,
+      "eval_steps_per_second": 28.58,
+      "step": 8700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9384981927752963e-05,
+      "loss": 1.7474,
+      "step": 8701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9349110173298137e-05,
+      "loss": 1.7632,
+      "step": 8702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9313258962055804e-05,
+      "loss": 1.7706,
+      "step": 8703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.927742829736371e-05,
+      "loss": 1.9664,
+      "step": 8704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9241618182557827e-05,
+      "loss": 1.8296,
+      "step": 8705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9205828620972265e-05,
+      "loss": 1.8632,
+      "step": 8706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.917005961593902e-05,
+      "loss": 1.8194,
+      "step": 8707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9134311170788324e-05,
+      "loss": 1.8202,
+      "step": 8708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9098583288848484e-05,
+      "loss": 1.7378,
+      "step": 8709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9062875973445812e-05,
+      "loss": 1.711,
+      "step": 8710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9027189227904788e-05,
+      "loss": 1.5354,
+      "step": 8711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8991523055547948e-05,
+      "loss": 1.9242,
+      "step": 8712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8955877459695913e-05,
+      "loss": 1.886,
+      "step": 8713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.892025244366736e-05,
+      "loss": 1.8472,
+      "step": 8714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8884648010779168e-05,
+      "loss": 1.7918,
+      "step": 8715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.88490641643461e-05,
+      "loss": 1.8779,
+      "step": 8716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.881350090768117e-05,
+      "loss": 1.7932,
+      "step": 8717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8777958244095376e-05,
+      "loss": 1.8347,
+      "step": 8718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8742436176897867e-05,
+      "loss": 1.7759,
+      "step": 8719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8706934709395895e-05,
+      "loss": 1.7307,
+      "step": 8720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.867145384489464e-05,
+      "loss": 2.0182,
+      "step": 8721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.863599358669755e-05,
+      "loss": 1.8432,
+      "step": 8722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.860055393810601e-05,
+      "loss": 1.7111,
+      "step": 8723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8565134902419603e-05,
+      "loss": 1.8238,
+      "step": 8724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.852973648293594e-05,
+      "loss": 1.7558,
+      "step": 8725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.849435868295072e-05,
+      "loss": 1.7358,
+      "step": 8726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.845900150575767e-05,
+      "loss": 1.7498,
+      "step": 8727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8423664954648627e-05,
+      "loss": 2.0069,
+      "step": 8728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8388349032913598e-05,
+      "loss": 1.871,
+      "step": 8729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8353053743840535e-05,
+      "loss": 1.9338,
+      "step": 8730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.831777909071559e-05,
+      "loss": 1.7939,
+      "step": 8731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.828252507682283e-05,
+      "loss": 1.7743,
+      "step": 8732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.824729170544457e-05,
+      "loss": 1.8966,
+      "step": 8733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8212078979861137e-05,
+      "loss": 2.0028,
+      "step": 8734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8176886903350907e-05,
+      "loss": 1.6845,
+      "step": 8735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.814171547919042e-05,
+      "loss": 1.7877,
+      "step": 8736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.810656471065415e-05,
+      "loss": 1.9282,
+      "step": 8737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8071434601014772e-05,
+      "loss": 1.841,
+      "step": 8738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.803632515354304e-05,
+      "loss": 1.846,
+      "step": 8739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8001236371507672e-05,
+      "loss": 1.8303,
+      "step": 8740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7966168258175523e-05,
+      "loss": 1.9537,
+      "step": 8741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7931120816811566e-05,
+      "loss": 2.005,
+      "step": 8742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7896094050678888e-05,
+      "loss": 1.758,
+      "step": 8743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7861087963038434e-05,
+      "loss": 1.6204,
+      "step": 8744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.782610255714943e-05,
+      "loss": 1.8089,
+      "step": 8745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7791137836269158e-05,
+      "loss": 1.765,
+      "step": 8746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7756193803652878e-05,
+      "loss": 1.8534,
+      "step": 8747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7721270462554043e-05,
+      "loss": 1.7859,
+      "step": 8748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7686367816224024e-05,
+      "loss": 1.6634,
+      "step": 8749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7651485867912384e-05,
+      "loss": 1.747,
+      "step": 8750
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.8614939451217651,
+      "eval_runtime": 184.1469,
+      "eval_samples_per_second": 457.265,
+      "eval_steps_per_second": 28.58,
+      "step": 8750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7616624620866753e-05,
+      "loss": 1.7588,
+      "step": 8751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7581784078332784e-05,
+      "loss": 1.9398,
+      "step": 8752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.754696424355424e-05,
+      "loss": 1.8415,
+      "step": 8753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7512165119773032e-05,
+      "loss": 1.9971,
+      "step": 8754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7477386710228897e-05,
+      "loss": 2.05,
+      "step": 8755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7442629018159888e-05,
+      "loss": 1.664,
+      "step": 8756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7407892046802023e-05,
+      "loss": 1.9579,
+      "step": 8757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7373175799389417e-05,
+      "loss": 1.6902,
+      "step": 8758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.733848027915431e-05,
+      "loss": 1.995,
+      "step": 8759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7303805489326846e-05,
+      "loss": 1.7358,
+      "step": 8760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7269151433135382e-05,
+      "loss": 1.8353,
+      "step": 8761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.723451811380631e-05,
+      "loss": 1.9209,
+      "step": 8762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7199905534564113e-05,
+      "loss": 1.8946,
+      "step": 8763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.716531369863129e-05,
+      "loss": 1.746,
+      "step": 8764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7130742609228487e-05,
+      "loss": 1.7788,
+      "step": 8765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7096192269574298e-05,
+      "loss": 2.0925,
+      "step": 8766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7061662682885485e-05,
+      "loss": 1.8133,
+      "step": 8767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7027153852376864e-05,
+      "loss": 1.7398,
+      "step": 8768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.699266578126125e-05,
+      "loss": 1.9226,
+      "step": 8769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6958198472749718e-05,
+      "loss": 1.702,
+      "step": 8770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6923751930051087e-05,
+      "loss": 1.8396,
+      "step": 8771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.688932615637252e-05,
+      "loss": 1.7285,
+      "step": 8772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6854921154919192e-05,
+      "loss": 1.8808,
+      "step": 8773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.682053692889422e-05,
+      "loss": 1.7557,
+      "step": 8774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6786173481498895e-05,
+      "loss": 1.7836,
+      "step": 8775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6751830815932604e-05,
+      "loss": 1.7492,
+      "step": 8776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.671750893539268e-05,
+      "loss": 1.5419,
+      "step": 8777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.668320784307457e-05,
+      "loss": 1.7463,
+      "step": 8778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6648927542171853e-05,
+      "loss": 1.8229,
+      "step": 8779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6614668035876115e-05,
+      "loss": 1.9123,
+      "step": 8780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6580429327377004e-05,
+      "loss": 1.6764,
+      "step": 8781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.65462114198623e-05,
+      "loss": 1.81,
+      "step": 8782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6512014316517647e-05,
+      "loss": 1.8413,
+      "step": 8783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6477838020526972e-05,
+      "loss": 1.8184,
+      "step": 8784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6443682535072176e-05,
+      "loss": 1.6425,
+      "step": 8785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6409547863333243e-05,
+      "loss": 1.7275,
+      "step": 8786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.637543400848824e-05,
+      "loss": 1.7179,
+      "step": 8787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6341340973713186e-05,
+      "loss": 1.9273,
+      "step": 8788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6307268762182256e-05,
+      "loss": 1.8784,
+      "step": 8789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.627321737706767e-05,
+      "loss": 1.7508,
+      "step": 8790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6239186821539725e-05,
+      "loss": 1.869,
+      "step": 8791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.620517709876677e-05,
+      "loss": 1.9472,
+      "step": 8792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.617118821191522e-05,
+      "loss": 1.8292,
+      "step": 8793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6137220164149435e-05,
+      "loss": 1.7945,
+      "step": 8794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6103272958632047e-05,
+      "loss": 1.8824,
+      "step": 8795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6069346598523557e-05,
+      "loss": 1.8411,
+      "step": 8796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6035441086982637e-05,
+      "loss": 1.7165,
+      "step": 8797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6001556427166062e-05,
+      "loss": 1.9754,
+      "step": 8798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.596769262222845e-05,
+      "loss": 1.6489,
+      "step": 8799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.593384967532267e-05,
+      "loss": 1.8479,
+      "step": 8800
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.8595343828201294,
+      "eval_runtime": 184.1975,
+      "eval_samples_per_second": 457.14,
+      "eval_steps_per_second": 28.573,
+      "step": 8800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5900027589599584e-05,
+      "loss": 1.7421,
+      "step": 8801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5866226368208178e-05,
+      "loss": 1.9056,
+      "step": 8802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5832446014295435e-05,
+      "loss": 1.7287,
+      "step": 8803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.579868653100631e-05,
+      "loss": 1.9867,
+      "step": 8804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5764947921483982e-05,
+      "loss": 1.9255,
+      "step": 8805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5731230188869608e-05,
+      "loss": 1.8668,
+      "step": 8806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.569753333630237e-05,
+      "loss": 1.84,
+      "step": 8807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5663857366919534e-05,
+      "loss": 1.6752,
+      "step": 8808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5630202283856458e-05,
+      "loss": 1.987,
+      "step": 8809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5596568090246547e-05,
+      "loss": 1.8256,
+      "step": 8810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.556295478922116e-05,
+      "loss": 1.7118,
+      "step": 8811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5529362383909822e-05,
+      "loss": 1.6738,
+      "step": 8812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5495790877440085e-05,
+      "loss": 2.0117,
+      "step": 8813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5462240272937588e-05,
+      "loss": 1.7012,
+      "step": 8814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.542871057352597e-05,
+      "loss": 1.7607,
+      "step": 8815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5395201782326877e-05,
+      "loss": 1.6615,
+      "step": 8816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5361713902460114e-05,
+      "loss": 1.7733,
+      "step": 8817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5328246937043525e-05,
+      "loss": 2.0382,
+      "step": 8818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5294800889192922e-05,
+      "loss": 1.8304,
+      "step": 8819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5261375762022254e-05,
+      "loss": 1.9159,
+      "step": 8820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5227971558643535e-05,
+      "loss": 1.5666,
+      "step": 8821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5194588282166724e-05,
+      "loss": 1.9446,
+      "step": 8822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.516122593569989e-05,
+      "loss": 1.6971,
+      "step": 8823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.512788452234921e-05,
+      "loss": 1.6753,
+      "step": 8824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5094564045218844e-05,
+      "loss": 1.6128,
+      "step": 8825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5061264507411057e-05,
+      "loss": 1.9851,
+      "step": 8826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.502798591202604e-05,
+      "loss": 1.7944,
+      "step": 8827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4994728262162175e-05,
+      "loss": 1.8408,
+      "step": 8828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4961491560915843e-05,
+      "loss": 1.918,
+      "step": 8829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.492827581138149e-05,
+      "loss": 1.7099,
+      "step": 8830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4895081016651556e-05,
+      "loss": 1.7873,
+      "step": 8831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.486190717981665e-05,
+      "loss": 1.7923,
+      "step": 8832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4828754303965223e-05,
+      "loss": 1.765,
+      "step": 8833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4795622392183997e-05,
+      "loss": 1.7464,
+      "step": 8834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.476251144755759e-05,
+      "loss": 1.9054,
+      "step": 8835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4729421473168784e-05,
+      "loss": 1.628,
+      "step": 8836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4696352472098284e-05,
+      "loss": 1.8738,
+      "step": 8837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4663304447424926e-05,
+      "loss": 1.7639,
+      "step": 8838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.463027740222562e-05,
+      "loss": 1.9215,
+      "step": 8839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4597271339575176e-05,
+      "loss": 1.9077,
+      "step": 8840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4564286262546644e-05,
+      "loss": 1.7791,
+      "step": 8841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4531322174210974e-05,
+      "loss": 1.8551,
+      "step": 8842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4498379077637273e-05,
+      "loss": 1.7807,
+      "step": 8843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.446545697589256e-05,
+      "loss": 1.8667,
+      "step": 8844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4432555872042016e-05,
+      "loss": 1.6853,
+      "step": 8845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4399675769148784e-05,
+      "loss": 1.8121,
+      "step": 8846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4366816670274155e-05,
+      "loss": 1.7473,
+      "step": 8847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4333978578477385e-05,
+      "loss": 1.7776,
+      "step": 8848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4301161496815798e-05,
+      "loss": 1.8945,
+      "step": 8849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4268365428344735e-05,
+      "loss": 1.8021,
+      "step": 8850
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.8586759567260742,
+      "eval_runtime": 184.1476,
+      "eval_samples_per_second": 457.264,
+      "eval_steps_per_second": 28.58,
+      "step": 8850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4235590376117582e-05,
+      "loss": 1.7156,
+      "step": 8851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.420283634318579e-05,
+      "loss": 1.8771,
+      "step": 8852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.417010333259889e-05,
+      "loss": 1.6616,
+      "step": 8853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4137391347404474e-05,
+      "loss": 1.8544,
+      "step": 8854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4104700390647966e-05,
+      "loss": 1.9156,
+      "step": 8855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.407203046537307e-05,
+      "loss": 1.8484,
+      "step": 8856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4039381574621433e-05,
+      "loss": 1.9058,
+      "step": 8857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4006753721432794e-05,
+      "loss": 1.8233,
+      "step": 8858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3974146908844835e-05,
+      "loss": 1.619,
+      "step": 8859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3941561139893432e-05,
+      "loss": 1.7304,
+      "step": 8860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3908996417612295e-05,
+      "loss": 1.9559,
+      "step": 8861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.387645274503336e-05,
+      "loss": 1.6266,
+      "step": 8862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3843930125186542e-05,
+      "loss": 1.9898,
+      "step": 8863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3811428561099748e-05,
+      "loss": 1.8935,
+      "step": 8864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.377894805579897e-05,
+      "loss": 1.6658,
+      "step": 8865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3746488612308297e-05,
+      "loss": 1.591,
+      "step": 8866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3714050233649726e-05,
+      "loss": 1.7575,
+      "step": 8867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.368163292284334e-05,
+      "loss": 1.6741,
+      "step": 8868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.364923668290736e-05,
+      "loss": 1.7793,
+      "step": 8869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3616861516857933e-05,
+      "loss": 2.0336,
+      "step": 8870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3584507427709233e-05,
+      "loss": 1.8601,
+      "step": 8871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.355217441847357e-05,
+      "loss": 1.8947,
+      "step": 8872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3519862492161258e-05,
+      "loss": 1.8765,
+      "step": 8873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3487571651780532e-05,
+      "loss": 1.6899,
+      "step": 8874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3455301900337816e-05,
+      "loss": 1.7809,
+      "step": 8875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3423053240837516e-05,
+      "loss": 1.5404,
+      "step": 8876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3390825676282118e-05,
+      "loss": 1.7413,
+      "step": 8877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3358619209672e-05,
+      "loss": 1.8035,
+      "step": 8878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3326433844005734e-05,
+      "loss": 1.9044,
+      "step": 8879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3294269582279843e-05,
+      "loss": 1.6343,
+      "step": 8880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.326212642748893e-05,
+      "loss": 2.0279,
+      "step": 8881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3230004382625654e-05,
+      "loss": 1.8763,
+      "step": 8882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3197903450680568e-05,
+      "loss": 1.7709,
+      "step": 8883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3165823634642398e-05,
+      "loss": 2.0255,
+      "step": 8884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3133764937497886e-05,
+      "loss": 1.8901,
+      "step": 8885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3101727362231763e-05,
+      "loss": 1.8817,
+      "step": 8886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.306971091182686e-05,
+      "loss": 1.6413,
+      "step": 8887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3037715589263985e-05,
+      "loss": 1.9498,
+      "step": 8888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3005741397521928e-05,
+      "loss": 1.7819,
+      "step": 8889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.297378833957761e-05,
+      "loss": 1.7701,
+      "step": 8890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.294185641840596e-05,
+      "loss": 1.9123,
+      "step": 8891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2909945636979956e-05,
+      "loss": 1.8069,
+      "step": 8892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.287805599827053e-05,
+      "loss": 1.6596,
+      "step": 8893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.284618750524678e-05,
+      "loss": 1.8005,
+      "step": 8894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2814340160875635e-05,
+      "loss": 1.8626,
+      "step": 8895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2782513968122226e-05,
+      "loss": 2.1579,
+      "step": 8896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2750708929949654e-05,
+      "loss": 1.6972,
+      "step": 8897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.271892504931905e-05,
+      "loss": 1.8104,
+      "step": 8898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2687162329189687e-05,
+      "loss": 1.8899,
+      "step": 8899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2655420772518586e-05,
+      "loss": 1.7032,
+      "step": 8900
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.8571223020553589,
+      "eval_runtime": 184.143,
+      "eval_samples_per_second": 457.275,
+      "eval_steps_per_second": 28.581,
+      "step": 8900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.262370038226108e-05,
+      "loss": 1.846,
+      "step": 8901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.259200116137039e-05,
+      "loss": 1.8504,
+      "step": 8902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2560323112797874e-05,
+      "loss": 1.8811,
+      "step": 8903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2528666239492735e-05,
+      "loss": 1.7905,
+      "step": 8904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2497030544402387e-05,
+      "loss": 1.8703,
+      "step": 8905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2465416030472224e-05,
+      "loss": 1.7959,
+      "step": 8906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.243382270064556e-05,
+      "loss": 1.834,
+      "step": 8907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.240225055786385e-05,
+      "loss": 1.8956,
+      "step": 8908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2370699605066574e-05,
+      "loss": 1.8919,
+      "step": 8909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.233916984519127e-05,
+      "loss": 1.8419,
+      "step": 8910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2307661281173315e-05,
+      "loss": 1.7926,
+      "step": 8911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.227617391594633e-05,
+      "loss": 1.8421,
+      "step": 8912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.224470775244186e-05,
+      "loss": 1.6587,
+      "step": 8913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2213262793589484e-05,
+      "loss": 1.7462,
+      "step": 8914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.21818390423168e-05,
+      "loss": 1.5356,
+      "step": 8915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.215043650154955e-05,
+      "loss": 1.8155,
+      "step": 8916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2119055174211255e-05,
+      "loss": 1.9579,
+      "step": 8917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2087695063223666e-05,
+      "loss": 1.9798,
+      "step": 8918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2056356171506503e-05,
+      "loss": 1.7197,
+      "step": 8919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2025038501977485e-05,
+      "loss": 1.8827,
+      "step": 8920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.199374205755239e-05,
+      "loss": 1.7687,
+      "step": 8921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1962466841145057e-05,
+      "loss": 1.8125,
+      "step": 8922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.193121285566718e-05,
+      "loss": 1.687,
+      "step": 8923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1899980104028662e-05,
+      "loss": 1.763,
+      "step": 8924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1868768589137333e-05,
+      "loss": 1.637,
+      "step": 8925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.18375783138991e-05,
+      "loss": 2.0274,
+      "step": 8926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1806409281217883e-05,
+      "loss": 1.826,
+      "step": 8927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.177526149399556e-05,
+      "loss": 1.6513,
+      "step": 8928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1744134955132082e-05,
+      "loss": 1.8087,
+      "step": 8929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.171302966752542e-05,
+      "loss": 1.8141,
+      "step": 8930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1681945634071554e-05,
+      "loss": 1.7313,
+      "step": 8931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1650882857664533e-05,
+      "loss": 1.8244,
+      "step": 8932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1619841341196406e-05,
+      "loss": 2.1153,
+      "step": 8933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1588821087557138e-05,
+      "loss": 1.6894,
+      "step": 8934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1557822099634835e-05,
+      "loss": 1.764,
+      "step": 8935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1526844380315658e-05,
+      "loss": 1.9813,
+      "step": 8936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.149588793248361e-05,
+      "loss": 1.7797,
+      "step": 8937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1464952759020855e-05,
+      "loss": 1.706,
+      "step": 8938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.143403886280762e-05,
+      "loss": 1.7517,
+      "step": 8939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.140314624672199e-05,
+      "loss": 1.8574,
+      "step": 8940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.137227491364016e-05,
+      "loss": 1.8133,
+      "step": 8941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1341424866436366e-05,
+      "loss": 1.794,
+      "step": 8942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.131059610798283e-05,
+      "loss": 1.8196,
+      "step": 8943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1279788641149822e-05,
+      "loss": 1.8544,
+      "step": 8944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1249002468805538e-05,
+      "loss": 1.9979,
+      "step": 8945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1218237593816304e-05,
+      "loss": 1.8421,
+      "step": 8946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1187494019046376e-05,
+      "loss": 1.9417,
+      "step": 8947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1156771747358113e-05,
+      "loss": 1.8482,
+      "step": 8948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.11260707816118e-05,
+      "loss": 1.7486,
+      "step": 8949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.109539112466588e-05,
+      "loss": 1.8989,
+      "step": 8950
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.8558391332626343,
+      "eval_runtime": 184.546,
+      "eval_samples_per_second": 456.276,
+      "eval_steps_per_second": 28.519,
+      "step": 8950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1064732779376566e-05,
+      "loss": 1.8093,
+      "step": 8951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1034095748598326e-05,
+      "loss": 1.7921,
+      "step": 8952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1003480035183514e-05,
+      "loss": 2.0481,
+      "step": 8953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0972885641982604e-05,
+      "loss": 1.7573,
+      "step": 8954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.094231257184398e-05,
+      "loss": 1.7356,
+      "step": 8955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0911760827614035e-05,
+      "loss": 1.9018,
+      "step": 8956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.088123041213727e-05,
+      "loss": 1.7666,
+      "step": 8957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.085072132825616e-05,
+      "loss": 1.7159,
+      "step": 8958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.082023357881116e-05,
+      "loss": 1.7955,
+      "step": 8959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.078976716664077e-05,
+      "loss": 1.8827,
+      "step": 8960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0759322094581535e-05,
+      "loss": 2.0618,
+      "step": 8961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.07288983654679e-05,
+      "loss": 1.9984,
+      "step": 8962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.069849598213247e-05,
+      "loss": 1.9021,
+      "step": 8963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0668114947405726e-05,
+      "loss": 1.8678,
+      "step": 8964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0637755264116292e-05,
+      "loss": 1.6738,
+      "step": 8965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0607416935090746e-05,
+      "loss": 1.6305,
+      "step": 8966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.05770999631536e-05,
+      "loss": 1.8673,
+      "step": 8967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.054680435112746e-05,
+      "loss": 1.6055,
+      "step": 8968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0516530101832992e-05,
+      "loss": 1.8367,
+      "step": 8969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0486277218088796e-05,
+      "loss": 1.8094,
+      "step": 8970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.045604570271145e-05,
+      "loss": 1.9808,
+      "step": 8971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0425835558515625e-05,
+      "loss": 1.971,
+      "step": 8972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0395646788313988e-05,
+      "loss": 1.8409,
+      "step": 8973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0365479394917147e-05,
+      "loss": 1.8806,
+      "step": 8974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0335333381133803e-05,
+      "loss": 2.0113,
+      "step": 8975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0305208749770653e-05,
+      "loss": 1.7497,
+      "step": 8976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0275105503632345e-05,
+      "loss": 1.9235,
+      "step": 8977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0245023645521637e-05,
+      "loss": 1.7659,
+      "step": 8978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.021496317823915e-05,
+      "loss": 1.6768,
+      "step": 8979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.018492410458361e-05,
+      "loss": 1.6964,
+      "step": 8980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0154906427351794e-05,
+      "loss": 1.8936,
+      "step": 8981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.01249101493384e-05,
+      "loss": 1.8103,
+      "step": 8982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0094935273336196e-05,
+      "loss": 1.9987,
+      "step": 8983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0064981802135867e-05,
+      "loss": 1.8758,
+      "step": 8984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0035049738526212e-05,
+      "loss": 1.6992,
+      "step": 8985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0005139085293944e-05,
+      "loss": 1.6373,
+      "step": 8986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.997524984522389e-05,
+      "loss": 1.9114,
+      "step": 8987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9945382021098772e-05,
+      "loss": 1.7685,
+      "step": 8988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.991553561569942e-05,
+      "loss": 1.7643,
+      "step": 8989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9885710631804554e-05,
+      "loss": 1.662,
+      "step": 8990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.985590707219098e-05,
+      "loss": 1.7616,
+      "step": 8991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9826124939633512e-05,
+      "loss": 1.9539,
+      "step": 8992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9796364236904924e-05,
+      "loss": 1.7576,
+      "step": 8993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9766624966776088e-05,
+      "loss": 2.0118,
+      "step": 8994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.973690713201573e-05,
+      "loss": 1.7697,
+      "step": 8995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9707210735390695e-05,
+      "loss": 1.6564,
+      "step": 8996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.96775357796658e-05,
+      "loss": 1.9933,
+      "step": 8997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.964788226760386e-05,
+      "loss": 1.7872,
+      "step": 8998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9618250201965697e-05,
+      "loss": 1.5988,
+      "step": 8999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9588639585510214e-05,
+      "loss": 1.8185,
+      "step": 9000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.854606032371521,
+      "eval_runtime": 184.6227,
+      "eval_samples_per_second": 456.087,
+      "eval_steps_per_second": 28.507,
+      "step": 9000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9559050420994122e-05,
+      "loss": 1.718,
+      "step": 9001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9529482711172302e-05,
+      "loss": 1.7917,
+      "step": 9002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9499936458797664e-05,
+      "loss": 1.6733,
+      "step": 9003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.947041166662092e-05,
+      "loss": 1.6648,
+      "step": 9004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9440908337390957e-05,
+      "loss": 1.6967,
+      "step": 9005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9411426473854686e-05,
+      "loss": 1.895,
+      "step": 9006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9381966078756865e-05,
+      "loss": 1.7685,
+      "step": 9007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9352527154840342e-05,
+      "loss": 1.7276,
+      "step": 9008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9323109704846022e-05,
+      "loss": 1.7901,
+      "step": 9009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9293713731512673e-05,
+      "loss": 1.7798,
+      "step": 9010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9264339237577256e-05,
+      "loss": 1.9685,
+      "step": 9011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.923498622577452e-05,
+      "loss": 1.6618,
+      "step": 9012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.920565469883734e-05,
+      "loss": 1.9532,
+      "step": 9013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9176344659496554e-05,
+      "loss": 1.8638,
+      "step": 9014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.914705611048101e-05,
+      "loss": 1.8588,
+      "step": 9015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.911778905451758e-05,
+      "loss": 1.8438,
+      "step": 9016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9088543494331118e-05,
+      "loss": 1.491,
+      "step": 9017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9059319432644413e-05,
+      "loss": 2.0203,
+      "step": 9018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9030116872178316e-05,
+      "loss": 1.6452,
+      "step": 9019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.900093581565171e-05,
+      "loss": 1.8158,
+      "step": 9020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8971776265781393e-05,
+      "loss": 1.6804,
+      "step": 9021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.894263822528225e-05,
+      "loss": 1.7072,
+      "step": 9022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913521696867052e-05,
+      "loss": 1.9033,
+      "step": 9023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8884426683246638e-05,
+      "loss": 1.7692,
+      "step": 9024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.885535318712986e-05,
+      "loss": 1.8854,
+      "step": 9025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8826301211223528e-05,
+      "loss": 1.9339,
+      "step": 9026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.879727075823248e-05,
+      "loss": 1.9654,
+      "step": 9027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8768261830859556e-05,
+      "loss": 1.9209,
+      "step": 9028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.873927443180548e-05,
+      "loss": 1.8447,
+      "step": 9029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8710308563769125e-05,
+      "loss": 1.6475,
+      "step": 9030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.868136422944727e-05,
+      "loss": 1.7311,
+      "step": 9031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8652441431534718e-05,
+      "loss": 1.6989,
+      "step": 9032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8623540172724328e-05,
+      "loss": 1.7301,
+      "step": 9033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8594660455706762e-05,
+      "loss": 1.8278,
+      "step": 9034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.856580228317087e-05,
+      "loss": 1.7537,
+      "step": 9035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8536965657803496e-05,
+      "loss": 1.9577,
+      "step": 9036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8508150582289273e-05,
+      "loss": 1.7471,
+      "step": 9037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.847935705931103e-05,
+      "loss": 1.7708,
+      "step": 9038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8450585091549565e-05,
+      "loss": 1.926,
+      "step": 9039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8421834681683568e-05,
+      "loss": 1.8658,
+      "step": 9040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.839310583238979e-05,
+      "loss": 1.9899,
+      "step": 9041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8364398546342974e-05,
+      "loss": 1.9706,
+      "step": 9042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.833571282621585e-05,
+      "loss": 1.981,
+      "step": 9043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8307048674679138e-05,
+      "loss": 2.0493,
+      "step": 9044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8278406094401623e-05,
+      "loss": 1.8533,
+      "step": 9045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8249785088049893e-05,
+      "loss": 1.8333,
+      "step": 9046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8221185658288673e-05,
+      "loss": 1.9307,
+      "step": 9047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8192607807780677e-05,
+      "loss": 1.9875,
+      "step": 9048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.816405153918657e-05,
+      "loss": 1.7619,
+      "step": 9049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.813551685516507e-05,
+      "loss": 1.6985,
+      "step": 9050
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 1.8531333208084106,
+      "eval_runtime": 184.1462,
+      "eval_samples_per_second": 457.267,
+      "eval_steps_per_second": 28.581,
+      "step": 9050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8107003758372738e-05,
+      "loss": 1.8352,
+      "step": 9051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8078512251464284e-05,
+      "loss": 1.6342,
+      "step": 9052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8050042337092343e-05,
+      "loss": 1.8156,
+      "step": 9053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8021594017907538e-05,
+      "loss": 1.9757,
+      "step": 9054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.799316729655848e-05,
+      "loss": 1.6479,
+      "step": 9055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.796476217569182e-05,
+      "loss": 1.6078,
+      "step": 9056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7936378657952056e-05,
+      "loss": 1.8925,
+      "step": 9057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7908016745981858e-05,
+      "loss": 1.9738,
+      "step": 9058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.787967644242175e-05,
+      "loss": 1.8364,
+      "step": 9059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.785135774991034e-05,
+      "loss": 1.8966,
+      "step": 9060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.782306067108419e-05,
+      "loss": 1.6745,
+      "step": 9061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7794785208577747e-05,
+      "loss": 1.9637,
+      "step": 9062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.77665313650236e-05,
+      "loss": 1.7841,
+      "step": 9063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7738299143052223e-05,
+      "loss": 1.8256,
+      "step": 9064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7710088545292156e-05,
+      "loss": 1.8245,
+      "step": 9065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7681899574369912e-05,
+      "loss": 1.743,
+      "step": 9066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.765373223290989e-05,
+      "loss": 1.8228,
+      "step": 9067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7625586523534576e-05,
+      "loss": 2.0987,
+      "step": 9068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7597462448864455e-05,
+      "loss": 1.9206,
+      "step": 9069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7569360011517848e-05,
+      "loss": 1.7004,
+      "step": 9070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7541279214111278e-05,
+      "loss": 2.0519,
+      "step": 9071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7513220059259067e-05,
+      "loss": 1.9167,
+      "step": 9072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.748518254957371e-05,
+      "loss": 1.8199,
+      "step": 9073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.745716668766545e-05,
+      "loss": 1.6108,
+      "step": 9074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.74291724761427e-05,
+      "loss": 1.864,
+      "step": 9075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7401199917611828e-05,
+      "loss": 1.8086,
+      "step": 9076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.73732490146771e-05,
+      "loss": 1.9384,
+      "step": 9077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7345319769940887e-05,
+      "loss": 1.739,
+      "step": 9078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.731741218600341e-05,
+      "loss": 2.0581,
+      "step": 9079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7289526265462984e-05,
+      "loss": 1.7718,
+      "step": 9080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7261662010915836e-05,
+      "loss": 1.779,
+      "step": 9081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.723381942495625e-05,
+      "loss": 1.8074,
+      "step": 9082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7205998510176403e-05,
+      "loss": 1.8491,
+      "step": 9083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7178199269166584e-05,
+      "loss": 1.7997,
+      "step": 9084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.715042170451486e-05,
+      "loss": 1.7111,
+      "step": 9085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7122665818807477e-05,
+      "loss": 1.9372,
+      "step": 9086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.709493161462855e-05,
+      "loss": 1.7178,
+      "step": 9087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7067219094560244e-05,
+      "loss": 1.7962,
+      "step": 9088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.703952826118266e-05,
+      "loss": 1.7206,
+      "step": 9089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.701185911707387e-05,
+      "loss": 1.9326,
+      "step": 9090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6984211664809956e-05,
+      "loss": 1.7695,
+      "step": 9091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6956585906965e-05,
+      "loss": 1.6741,
+      "step": 9092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6928981846110997e-05,
+      "loss": 1.9329,
+      "step": 9093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6901399484818004e-05,
+      "loss": 1.8204,
+      "step": 9094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.687383882565402e-05,
+      "loss": 1.7715,
+      "step": 9095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.684629987118494e-05,
+      "loss": 1.8165,
+      "step": 9096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6818782623974793e-05,
+      "loss": 1.8152,
+      "step": 9097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6791287086585477e-05,
+      "loss": 1.7076,
+      "step": 9098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6763813261576916e-05,
+      "loss": 1.7501,
+      "step": 9099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6736361151507063e-05,
+      "loss": 1.7797,
+      "step": 9100
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 1.8531150817871094,
+      "eval_runtime": 184.1326,
+      "eval_samples_per_second": 457.301,
+      "eval_steps_per_second": 28.583,
+      "step": 9100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6708930758931652e-05,
+      "loss": 1.8087,
+      "step": 9101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.668152208640461e-05,
+      "loss": 1.801,
+      "step": 9102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6654135136477787e-05,
+      "loss": 1.6176,
+      "step": 9103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6626769911700924e-05,
+      "loss": 1.8645,
+      "step": 9104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659942641462181e-05,
+      "loss": 1.7711,
+      "step": 9105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572104647786245e-05,
+      "loss": 1.78,
+      "step": 9106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654480461373789e-05,
+      "loss": 1.6308,
+      "step": 9107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6517526315018516e-05,
+      "loss": 1.87,
+      "step": 9108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6490269754167757e-05,
+      "loss": 1.8528,
+      "step": 9109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6463034933723335e-05,
+      "loss": 1.9402,
+      "step": 9110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6435821856220832e-05,
+      "loss": 1.7317,
+      "step": 9111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6408630524193945e-05,
+      "loss": 1.7975,
+      "step": 9112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6381460940174147e-05,
+      "loss": 1.7178,
+      "step": 9113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6354313106691083e-05,
+      "loss": 1.8601,
+      "step": 9114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6327187026272254e-05,
+      "loss": 1.7902,
+      "step": 9115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6300082701443204e-05,
+      "loss": 1.6169,
+      "step": 9116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6273000134727405e-05,
+      "loss": 1.8457,
+      "step": 9117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.624593932864632e-05,
+      "loss": 1.8264,
+      "step": 9118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6218900285719373e-05,
+      "loss": 1.8365,
+      "step": 9119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.619188300846397e-05,
+      "loss": 2.0746,
+      "step": 9120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.616488749939554e-05,
+      "loss": 1.6965,
+      "step": 9121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6137913761027382e-05,
+      "loss": 1.6185,
+      "step": 9122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6110961795870906e-05,
+      "loss": 1.9903,
+      "step": 9123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6084031606435345e-05,
+      "loss": 1.8759,
+      "step": 9124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.605712319522795e-05,
+      "loss": 1.9082,
+      "step": 9125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6030236564754054e-05,
+      "loss": 1.9978,
+      "step": 9126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6003371717516813e-05,
+      "loss": 1.9025,
+      "step": 9127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.597652865601745e-05,
+      "loss": 1.7067,
+      "step": 9128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5949707382755168e-05,
+      "loss": 1.7072,
+      "step": 9129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5922907900227018e-05,
+      "loss": 1.6936,
+      "step": 9130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5896130210928118e-05,
+      "loss": 1.7956,
+      "step": 9131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5869374317351583e-05,
+      "loss": 1.9722,
+      "step": 9132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5842640221988503e-05,
+      "loss": 1.7374,
+      "step": 9133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5815927927327777e-05,
+      "loss": 1.9348,
+      "step": 9134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5789237435856503e-05,
+      "loss": 1.9704,
+      "step": 9135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5762568750059604e-05,
+      "loss": 1.7621,
+      "step": 9136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5735921872419955e-05,
+      "loss": 1.8656,
+      "step": 9137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.570929680541852e-05,
+      "loss": 1.9883,
+      "step": 9138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.568269355153415e-05,
+      "loss": 1.9963,
+      "step": 9139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.565611211324372e-05,
+      "loss": 1.8781,
+      "step": 9140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5629552493021946e-05,
+      "loss": 1.9375,
+      "step": 9141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.560301469334166e-05,
+      "loss": 1.8435,
+      "step": 9142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5576498716673605e-05,
+      "loss": 1.687,
+      "step": 9143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.555000456548647e-05,
+      "loss": 1.8713,
+      "step": 9144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.552353224224698e-05,
+      "loss": 1.7323,
+      "step": 9145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5497081749419745e-05,
+      "loss": 1.7306,
+      "step": 9146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.547065308946738e-05,
+      "loss": 1.8847,
+      "step": 9147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.544424626485044e-05,
+      "loss": 1.7103,
+      "step": 9148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.541786127802755e-05,
+      "loss": 1.7976,
+      "step": 9149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5391498131455157e-05,
+      "loss": 1.6903,
+      "step": 9150
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 1.8525229692459106,
+      "eval_runtime": 184.1069,
+      "eval_samples_per_second": 457.365,
+      "eval_steps_per_second": 28.587,
+      "step": 9150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.536515682758782e-05,
+      "loss": 1.8641,
+      "step": 9151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5338837368877896e-05,
+      "loss": 1.6287,
+      "step": 9152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.531253975777583e-05,
+      "loss": 1.6728,
+      "step": 9153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5286263996730027e-05,
+      "loss": 2.0165,
+      "step": 9154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.526001008818681e-05,
+      "loss": 1.5187,
+      "step": 9155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5233778034590501e-05,
+      "loss": 1.6002,
+      "step": 9156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5207567838383424e-05,
+      "loss": 1.8492,
+      "step": 9157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5181379502005737e-05,
+      "loss": 1.8232,
+      "step": 9158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5155213027895687e-05,
+      "loss": 1.6541,
+      "step": 9159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5129068418489433e-05,
+      "loss": 1.6975,
+      "step": 9160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.510294567622114e-05,
+      "loss": 1.6975,
+      "step": 9161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5076844803522921e-05,
+      "loss": 1.6474,
+      "step": 9162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5050765802824772e-05,
+      "loss": 1.818,
+      "step": 9163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5024708676554755e-05,
+      "loss": 1.7276,
+      "step": 9164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4998673427138843e-05,
+      "loss": 1.7793,
+      "step": 9165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4972660057001069e-05,
+      "loss": 2.0197,
+      "step": 9166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4946668568563249e-05,
+      "loss": 1.7754,
+      "step": 9167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4920698964245304e-05,
+      "loss": 1.8024,
+      "step": 9168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.489475124646511e-05,
+      "loss": 1.9117,
+      "step": 9169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4868825417638427e-05,
+      "loss": 1.8608,
+      "step": 9170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4842921480179018e-05,
+      "loss": 1.8623,
+      "step": 9171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.481703943649862e-05,
+      "loss": 1.9136,
+      "step": 9172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4791179289006973e-05,
+      "loss": 1.8499,
+      "step": 9173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.476534104011168e-05,
+      "loss": 1.7344,
+      "step": 9174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4739524692218315e-05,
+      "loss": 2.0004,
+      "step": 9175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4713730247730538e-05,
+      "loss": 1.8806,
+      "step": 9176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4687957709049821e-05,
+      "loss": 1.7935,
+      "step": 9177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4662207078575685e-05,
+      "loss": 1.8731,
+      "step": 9178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.463647835870563e-05,
+      "loss": 2.0123,
+      "step": 9179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4610771551834989e-05,
+      "loss": 1.8544,
+      "step": 9180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4585086660357155e-05,
+      "loss": 1.875,
+      "step": 9181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4559423686663492e-05,
+      "loss": 1.9546,
+      "step": 9182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4533782633143283e-05,
+      "loss": 1.653,
+      "step": 9183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4508163502183786e-05,
+      "loss": 1.8939,
+      "step": 9184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.448256629617023e-05,
+      "loss": 1.6607,
+      "step": 9185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4456991017485738e-05,
+      "loss": 1.8149,
+      "step": 9186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.443143766851146e-05,
+      "loss": 1.7905,
+      "step": 9187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4405906251626494e-05,
+      "loss": 1.7435,
+      "step": 9188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4380396769207881e-05,
+      "loss": 1.762,
+      "step": 9189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4354909223630669e-05,
+      "loss": 1.6435,
+      "step": 9190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4329443617267735e-05,
+      "loss": 1.8845,
+      "step": 9191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4303999952490043e-05,
+      "loss": 1.9533,
+      "step": 9192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4278578231666478e-05,
+      "loss": 1.8465,
+      "step": 9193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.425317845716384e-05,
+      "loss": 1.824,
+      "step": 9194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.422780063134696e-05,
+      "loss": 1.8325,
+      "step": 9195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4202444756578587e-05,
+      "loss": 1.7634,
+      "step": 9196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.417711083521936e-05,
+      "loss": 1.9848,
+      "step": 9197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4151798869628002e-05,
+      "loss": 1.8509,
+      "step": 9198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4126508862161076e-05,
+      "loss": 1.8352,
+      "step": 9199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4101240815173222e-05,
+      "loss": 1.8766,
+      "step": 9200
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 1.8516799211502075,
+      "eval_runtime": 184.0853,
+      "eval_samples_per_second": 457.418,
+      "eval_steps_per_second": 28.59,
+      "step": 9200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4075994731016895e-05,
+      "loss": 1.7324,
+      "step": 9201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4050770612042601e-05,
+      "loss": 1.7405,
+      "step": 9202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4025568460598825e-05,
+      "loss": 1.6044,
+      "step": 9203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4000388279031856e-05,
+      "loss": 1.8771,
+      "step": 9204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3975230069686095e-05,
+      "loss": 1.7145,
+      "step": 9205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3950093834903865e-05,
+      "loss": 1.816,
+      "step": 9206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3924979577025404e-05,
+      "loss": 1.616,
+      "step": 9207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3899887298388897e-05,
+      "loss": 1.8684,
+      "step": 9208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3874817001330504e-05,
+      "loss": 2.0239,
+      "step": 9209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3849768688184356e-05,
+      "loss": 1.91,
+      "step": 9210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3824742361282534e-05,
+      "loss": 1.8787,
+      "step": 9211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3799738022955033e-05,
+      "loss": 1.8212,
+      "step": 9212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3774755675529854e-05,
+      "loss": 1.8078,
+      "step": 9213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3749795321332887e-05,
+      "loss": 1.6215,
+      "step": 9214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3724856962688048e-05,
+      "loss": 1.8689,
+      "step": 9215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3699940601917122e-05,
+      "loss": 1.9105,
+      "step": 9216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3675046241339917e-05,
+      "loss": 1.8273,
+      "step": 9217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3650173883274219e-05,
+      "loss": 1.8385,
+      "step": 9218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3625323530035648e-05,
+      "loss": 2.1014,
+      "step": 9219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3600495183937827e-05,
+      "loss": 1.7511,
+      "step": 9220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3575688847292378e-05,
+      "loss": 1.9315,
+      "step": 9221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3550904522408841e-05,
+      "loss": 1.9618,
+      "step": 9222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3526142211594706e-05,
+      "loss": 1.8309,
+      "step": 9223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.350140191715543e-05,
+      "loss": 1.682,
+      "step": 9224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3476683641394372e-05,
+      "loss": 1.6535,
+      "step": 9225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3451987386612851e-05,
+      "loss": 2.0171,
+      "step": 9226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.34273131551102e-05,
+      "loss": 1.9169,
+      "step": 9227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3402660949183659e-05,
+      "loss": 1.9161,
+      "step": 9228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3378030771128424e-05,
+      "loss": 1.7948,
+      "step": 9229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3353422623237605e-05,
+      "loss": 1.7181,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3328836507802256e-05,
+      "loss": 1.6775,
+      "step": 9231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.330427242711149e-05,
+      "loss": 1.8712,
+      "step": 9232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3279730383452288e-05,
+      "loss": 1.7791,
+      "step": 9233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3255210379109484e-05,
+      "loss": 1.8622,
+      "step": 9234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3230712416366058e-05,
+      "loss": 1.6121,
+      "step": 9235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3206236497502827e-05,
+      "loss": 1.9254,
+      "step": 9236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3181782624798521e-05,
+      "loss": 1.7628,
+      "step": 9237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3157350800529877e-05,
+      "loss": 1.5816,
+      "step": 9238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.31329410269716e-05,
+      "loss": 1.7109,
+      "step": 9239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3108553306396264e-05,
+      "loss": 1.9687,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3084187641074496e-05,
+      "loss": 2.143,
+      "step": 9241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.305984403327473e-05,
+      "loss": 1.7869,
+      "step": 9242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3035522485263462e-05,
+      "loss": 1.6276,
+      "step": 9243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.30112229993051e-05,
+      "loss": 1.8775,
+      "step": 9244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2986945577661975e-05,
+      "loss": 1.858,
+      "step": 9245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2962690222594443e-05,
+      "loss": 1.8481,
+      "step": 9246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.293845693636067e-05,
+      "loss": 1.8325,
+      "step": 9247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2914245721216856e-05,
+      "loss": 1.7751,
+      "step": 9248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2890056579417136e-05,
+      "loss": 1.9675,
+      "step": 9249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.286588951321363e-05,
+      "loss": 1.7187,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.8509694337844849,
+      "eval_runtime": 184.077,
+      "eval_samples_per_second": 457.439,
+      "eval_steps_per_second": 28.591,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2841744524856286e-05,
+      "loss": 1.8683,
+      "step": 9251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2817621616593167e-05,
+      "loss": 1.7058,
+      "step": 9252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2793520790670116e-05,
+      "loss": 1.6504,
+      "step": 9253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2769442049330948e-05,
+      "loss": 1.7913,
+      "step": 9254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2745385394817538e-05,
+      "loss": 1.8908,
+      "step": 9255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2721350829369593e-05,
+      "loss": 1.8558,
+      "step": 9256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2697338355224852e-05,
+      "loss": 1.776,
+      "step": 9257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.267334797461886e-05,
+      "loss": 1.9369,
+      "step": 9258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2649379689785218e-05,
+      "loss": 1.9313,
+      "step": 9259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2625433502955446e-05,
+      "loss": 1.8106,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2601509416358987e-05,
+      "loss": 1.9515,
+      "step": 9261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2577607432223277e-05,
+      "loss": 2.0227,
+      "step": 9262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2553727552773647e-05,
+      "loss": 1.7772,
+      "step": 9263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.252986978023335e-05,
+      "loss": 1.708,
+      "step": 9264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2506034116823634e-05,
+      "loss": 1.8802,
+      "step": 9265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2482220564763668e-05,
+      "loss": 1.8648,
+      "step": 9266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2458429126270542e-05,
+      "loss": 1.8781,
+      "step": 9267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2434659803559318e-05,
+      "loss": 1.9309,
+      "step": 9268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2410912598843e-05,
+      "loss": 1.7939,
+      "step": 9269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2387187514332487e-05,
+      "loss": 1.6463,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2363484552236653e-05,
+      "loss": 1.6554,
+      "step": 9271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2339803714762315e-05,
+      "loss": 1.8579,
+      "step": 9272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2316145004114238e-05,
+      "loss": 1.854,
+      "step": 9273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2292508422495158e-05,
+      "loss": 1.914,
+      "step": 9274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2268893972105593e-05,
+      "loss": 1.9993,
+      "step": 9275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2245301655144203e-05,
+      "loss": 1.8457,
+      "step": 9276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.222173147380748e-05,
+      "loss": 1.7293,
+      "step": 9277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.219818343028986e-05,
+      "loss": 1.7856,
+      "step": 9278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2174657526783733e-05,
+      "loss": 1.8571,
+      "step": 9279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151153765479456e-05,
+      "loss": 1.7547,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.212767214856525e-05,
+      "loss": 1.9106,
+      "step": 9281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2104212678227339e-05,
+      "loss": 1.6174,
+      "step": 9282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2080775356649864e-05,
+      "loss": 1.7101,
+      "step": 9283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2057360186014915e-05,
+      "loss": 1.8827,
+      "step": 9284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2033967168502524e-05,
+      "loss": 1.778,
+      "step": 9285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2010596306290589e-05,
+      "loss": 1.8554,
+      "step": 9286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1987247601555035e-05,
+      "loss": 1.6548,
+      "step": 9287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1963921056469706e-05,
+      "loss": 1.6941,
+      "step": 9288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.194061667320634e-05,
+      "loss": 2.004,
+      "step": 9289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.191733445393467e-05,
+      "loss": 1.7896,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1894074400822353e-05,
+      "loss": 1.9752,
+      "step": 9291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1870836516034877e-05,
+      "loss": 2.0475,
+      "step": 9292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.184762080173582e-05,
+      "loss": 1.8553,
+      "step": 9293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1824427260086618e-05,
+      "loss": 1.5875,
+      "step": 9294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1801255893246682e-05,
+      "loss": 1.6941,
+      "step": 9295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1778106703373315e-05,
+      "loss": 1.7524,
+      "step": 9296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1754979692621708e-05,
+      "loss": 1.8077,
+      "step": 9297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1731874863145143e-05,
+      "loss": 1.6457,
+      "step": 9298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1708792217094672e-05,
+      "loss": 1.6618,
+      "step": 9299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1685731756619438e-05,
+      "loss": 1.7295,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.8499205112457275,
+      "eval_runtime": 184.1395,
+      "eval_samples_per_second": 457.284,
+      "eval_steps_per_second": 28.582,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1662693483866338e-05,
+      "loss": 1.7625,
+      "step": 9301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.163967740098032e-05,
+      "loss": 1.6782,
+      "step": 9302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1616683510104337e-05,
+      "loss": 1.7686,
+      "step": 9303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1593711813379065e-05,
+      "loss": 1.7592,
+      "step": 9304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1570762312943296e-05,
+      "loss": 1.8508,
+      "step": 9305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.154783501093365e-05,
+      "loss": 1.8389,
+      "step": 9306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1524929909484783e-05,
+      "loss": 1.8296,
+      "step": 9307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.150204701072921e-05,
+      "loss": 1.9022,
+      "step": 9308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1479186316797341e-05,
+      "loss": 1.7349,
+      "step": 9309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.145634782981761e-05,
+      "loss": 1.8984,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1433531551916343e-05,
+      "loss": 1.788,
+      "step": 9311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1410737485217788e-05,
+      "loss": 1.935,
+      "step": 9312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1387965631844189e-05,
+      "loss": 1.8251,
+      "step": 9313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1365215993915572e-05,
+      "loss": 1.9104,
+      "step": 9314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1342488573550053e-05,
+      "loss": 1.762,
+      "step": 9315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1319783372863602e-05,
+      "loss": 1.8791,
+      "step": 9316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1297100393970139e-05,
+      "loss": 1.8121,
+      "step": 9317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1274439638981532e-05,
+      "loss": 1.9102,
+      "step": 9318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1251801110007564e-05,
+      "loss": 2.0637,
+      "step": 9319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1229184809155884e-05,
+      "loss": 1.7793,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1206590738532168e-05,
+      "loss": 1.856,
+      "step": 9321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1184018900240012e-05,
+      "loss": 1.9038,
+      "step": 9322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1161469296380899e-05,
+      "loss": 1.7727,
+      "step": 9323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1138941929054264e-05,
+      "loss": 1.821,
+      "step": 9324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1116436800357455e-05,
+      "loss": 1.8336,
+      "step": 9325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1093953912385768e-05,
+      "loss": 1.8393,
+      "step": 9326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1071493267232418e-05,
+      "loss": 1.8026,
+      "step": 9327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1049054866988567e-05,
+      "loss": 1.6435,
+      "step": 9328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1026638713743264e-05,
+      "loss": 1.7157,
+      "step": 9329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1004244809583592e-05,
+      "loss": 1.6468,
+      "step": 9330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.098187315659438e-05,
+      "loss": 1.7231,
+      "step": 9331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0959523756858547e-05,
+      "loss": 1.7819,
+      "step": 9332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0937196612456902e-05,
+      "loss": 1.809,
+      "step": 9333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.091489172546814e-05,
+      "loss": 1.907,
+      "step": 9334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.089260909796888e-05,
+      "loss": 1.8536,
+      "step": 9335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0870348732033769e-05,
+      "loss": 1.7379,
+      "step": 9336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0848110629735231e-05,
+      "loss": 1.7528,
+      "step": 9337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.082589479314372e-05,
+      "loss": 1.8569,
+      "step": 9338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0803701224327612e-05,
+      "loss": 1.8355,
+      "step": 9339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0781529925353167e-05,
+      "loss": 1.8757,
+      "step": 9340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0759380898284654e-05,
+      "loss": 1.7666,
+      "step": 9341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0737254145184145e-05,
+      "loss": 1.7994,
+      "step": 9342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0715149668111712e-05,
+      "loss": 1.8514,
+      "step": 9343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0693067469125323e-05,
+      "loss": 1.6426,
+      "step": 9344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0671007550280942e-05,
+      "loss": 1.8876,
+      "step": 9345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.06489699136324e-05,
+      "loss": 1.7897,
+      "step": 9346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0626954561231472e-05,
+      "loss": 1.948,
+      "step": 9347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0604961495127796e-05,
+      "loss": 1.847,
+      "step": 9348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0582990717369013e-05,
+      "loss": 1.6117,
+      "step": 9349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0561042230000678e-05,
+      "loss": 1.6285,
+      "step": 9350
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 1.8492958545684814,
+      "eval_runtime": 185.1758,
+      "eval_samples_per_second": 454.725,
+      "eval_steps_per_second": 28.422,
+      "step": 9350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0539116035066243e-05,
+      "loss": 1.6951,
+      "step": 9351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.051721213460713e-05,
+      "loss": 1.8221,
+      "step": 9352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0495330530662594e-05,
+      "loss": 1.8324,
+      "step": 9353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0473471225269898e-05,
+      "loss": 1.7799,
+      "step": 9354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0451634220464246e-05,
+      "loss": 1.7513,
+      "step": 9355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0429819518278649e-05,
+      "loss": 1.7672,
+      "step": 9356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0408027120744179e-05,
+      "loss": 1.7883,
+      "step": 9357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0386257029889768e-05,
+      "loss": 1.7965,
+      "step": 9358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0364509247742238e-05,
+      "loss": 1.7256,
+      "step": 9359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.034278377632636e-05,
+      "loss": 1.7708,
+      "step": 9360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0321080617664847e-05,
+      "loss": 1.7415,
+      "step": 9361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0299399773778362e-05,
+      "loss": 1.8897,
+      "step": 9362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0277741246685429e-05,
+      "loss": 1.7422,
+      "step": 9363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.025610503840249e-05,
+      "loss": 1.9009,
+      "step": 9364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0234491150943963e-05,
+      "loss": 1.7412,
+      "step": 9365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0212899586322183e-05,
+      "loss": 2.0273,
+      "step": 9366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0191330346547317e-05,
+      "loss": 1.8118,
+      "step": 9367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0169783433627566e-05,
+      "loss": 1.9739,
+      "step": 9368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0148258849569048e-05,
+      "loss": 1.7607,
+      "step": 9369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0126756596375685e-05,
+      "loss": 1.9586,
+      "step": 9370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0105276676049407e-05,
+      "loss": 1.8595,
+      "step": 9371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0083819090590086e-05,
+      "loss": 1.6916,
+      "step": 9372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0062383841995482e-05,
+      "loss": 1.6393,
+      "step": 9373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.004097093226128e-05,
+      "loss": 1.7123,
+      "step": 9374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0019580363381075e-05,
+      "loss": 1.7956,
+      "step": 9375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.998212137346363e-06,
+      "loss": 1.9204,
+      "step": 9376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.976866256146605e-06,
+      "loss": 1.9698,
+      "step": 9377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.955542721769156e-06,
+      "loss": 1.9129,
+      "step": 9378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.93424153619929e-06,
+      "loss": 2.1255,
+      "step": 9379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.91296270142028e-06,
+      "loss": 1.6916,
+      "step": 9380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.891706219413126e-06,
+      "loss": 1.8553,
+      "step": 9381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.87047209215694e-06,
+      "loss": 1.8313,
+      "step": 9382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.849260321628667e-06,
+      "loss": 1.6074,
+      "step": 9383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.828070909803143e-06,
+      "loss": 1.9269,
+      "step": 9384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.806903858653237e-06,
+      "loss": 1.663,
+      "step": 9385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.785759170149622e-06,
+      "loss": 1.6518,
+      "step": 9386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.764636846260915e-06,
+      "loss": 1.9703,
+      "step": 9387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.743536888953657e-06,
+      "loss": 1.8275,
+      "step": 9388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.722459300192332e-06,
+      "loss": 1.7077,
+      "step": 9389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.701404081939314e-06,
+      "loss": 2.009,
+      "step": 9390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.680371236154895e-06,
+      "loss": 1.8146,
+      "step": 9391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.659360764797342e-06,
+      "loss": 1.8222,
+      "step": 9392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.638372669822732e-06,
+      "loss": 1.8569,
+      "step": 9393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.617406953185137e-06,
+      "loss": 1.8628,
+      "step": 9394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.596463616836498e-06,
+      "loss": 1.7257,
+      "step": 9395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.575542662726755e-06,
+      "loss": 1.7955,
+      "step": 9396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.554644092803654e-06,
+      "loss": 1.9166,
+      "step": 9397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.53376790901292e-06,
+      "loss": 2.0143,
+      "step": 9398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.512914113298188e-06,
+      "loss": 1.7402,
+      "step": 9399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.492082707601047e-06,
+      "loss": 1.6925,
+      "step": 9400
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 1.84866464138031,
+      "eval_runtime": 184.1172,
+      "eval_samples_per_second": 457.339,
+      "eval_steps_per_second": 28.585,
+      "step": 9400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.471273693860887e-06,
+      "loss": 1.8466,
+      "step": 9401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.450487074015108e-06,
+      "loss": 1.7546,
+      "step": 9402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.429722849999046e-06,
+      "loss": 1.8416,
+      "step": 9403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.40898102374585e-06,
+      "loss": 1.8858,
+      "step": 9404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.38826159718667e-06,
+      "loss": 1.8843,
+      "step": 9405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.367564572250519e-06,
+      "loss": 1.7346,
+      "step": 9406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.346889950864384e-06,
+      "loss": 1.9022,
+      "step": 9407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.326237734953141e-06,
+      "loss": 1.654,
+      "step": 9408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.305607926439501e-06,
+      "loss": 1.763,
+      "step": 9409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.285000527244181e-06,
+      "loss": 1.989,
+      "step": 9410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.264415539285837e-06,
+      "loss": 1.8183,
+      "step": 9411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.243852964480937e-06,
+      "loss": 1.6577,
+      "step": 9412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.223312804743922e-06,
+      "loss": 1.8273,
+      "step": 9413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.202795061987207e-06,
+      "loss": 1.7599,
+      "step": 9414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.18229973812093e-06,
+      "loss": 1.7757,
+      "step": 9415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.161826835053344e-06,
+      "loss": 1.6381,
+      "step": 9416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.141376354690506e-06,
+      "loss": 1.6893,
+      "step": 9417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.120948298936422e-06,
+      "loss": 1.7776,
+      "step": 9418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.100542669692985e-06,
+      "loss": 1.9477,
+      "step": 9419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.080159468860095e-06,
+      "loss": 1.7772,
+      "step": 9420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.059798698335341e-06,
+      "loss": 1.8057,
+      "step": 9421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.039460360014489e-06,
+      "loss": 1.9369,
+      "step": 9422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.019144455791023e-06,
+      "loss": 1.6663,
+      "step": 9423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.998850987556456e-06,
+      "loss": 1.7183,
+      "step": 9424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.978579957200167e-06,
+      "loss": 1.8049,
+      "step": 9425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.958331366609424e-06,
+      "loss": 1.7526,
+      "step": 9426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.938105217669412e-06,
+      "loss": 1.8392,
+      "step": 9427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.917901512263239e-06,
+      "loss": 1.851,
+      "step": 9428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.89772025227198e-06,
+      "loss": 1.7566,
+      "step": 9429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.877561439574549e-06,
+      "loss": 1.8724,
+      "step": 9430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.857425076047754e-06,
+      "loss": 1.9874,
+      "step": 9431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.83731116356637e-06,
+      "loss": 2.0886,
+      "step": 9432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.817219704003065e-06,
+      "loss": 1.5006,
+      "step": 9433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.797150699228374e-06,
+      "loss": 1.716,
+      "step": 9434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.777104151110825e-06,
+      "loss": 2.0059,
+      "step": 9435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.757080061516793e-06,
+      "loss": 1.8635,
+      "step": 9436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.73707843231053e-06,
+      "loss": 1.9095,
+      "step": 9437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.7170992653543e-06,
+      "loss": 1.7057,
+      "step": 9438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.697142562508199e-06,
+      "loss": 1.8777,
+      "step": 9439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.677208325630265e-06,
+      "loss": 1.7025,
+      "step": 9440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.657296556576405e-06,
+      "loss": 1.8448,
+      "step": 9441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.637407257200496e-06,
+      "loss": 1.9247,
+      "step": 9442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.617540429354226e-06,
+      "loss": 1.9887,
+      "step": 9443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.597696074887335e-06,
+      "loss": 1.8988,
+      "step": 9444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.577874195647318e-06,
+      "loss": 1.6266,
+      "step": 9445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.55807479347967e-06,
+      "loss": 1.7499,
+      "step": 9446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.538297870227778e-06,
+      "loss": 1.9474,
+      "step": 9447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.51854342773295e-06,
+      "loss": 1.8569,
+      "step": 9448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.498811467834322e-06,
+      "loss": 1.8647,
+      "step": 9449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.479101992369036e-06,
+      "loss": 1.719,
+      "step": 9450
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.8478102684020996,
+      "eval_runtime": 184.1232,
+      "eval_samples_per_second": 457.324,
+      "eval_steps_per_second": 28.584,
+      "step": 9450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.4594150031721e-06,
+      "loss": 1.8155,
+      "step": 9451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.439750502076404e-06,
+      "loss": 1.8106,
+      "step": 9452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.420108490912793e-06,
+      "loss": 1.8484,
+      "step": 9453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.400488971509967e-06,
+      "loss": 1.7887,
+      "step": 9454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.38089194569458e-06,
+      "loss": 1.9965,
+      "step": 9455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.36131741529117e-06,
+      "loss": 1.7667,
+      "step": 9456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.34176538212214e-06,
+      "loss": 1.821,
+      "step": 9457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.322235848007898e-06,
+      "loss": 1.6718,
+      "step": 9458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.302728814766709e-06,
+      "loss": 1.7071,
+      "step": 9459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.283244284214647e-06,
+      "loss": 1.9135,
+      "step": 9460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.26378225816582e-06,
+      "loss": 1.7786,
+      "step": 9461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.244342738432192e-06,
+      "loss": 1.6672,
+      "step": 9462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.22492572682368e-06,
+      "loss": 1.8361,
+      "step": 9463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.205531225148e-06,
+      "loss": 1.7614,
+      "step": 9464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.186159235210877e-06,
+      "loss": 1.826,
+      "step": 9465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.166809758815896e-06,
+      "loss": 1.8337,
+      "step": 9466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.147482797764477e-06,
+      "loss": 1.7691,
+      "step": 9467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.128178353856097e-06,
+      "loss": 1.884,
+      "step": 9468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.10889642888804e-06,
+      "loss": 1.6377,
+      "step": 9469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.089637024655483e-06,
+      "loss": 1.7989,
+      "step": 9470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.070400142951545e-06,
+      "loss": 1.8594,
+      "step": 9471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.051185785567211e-06,
+      "loss": 1.7705,
+      "step": 9472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.031993954291384e-06,
+      "loss": 1.6872,
+      "step": 9473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.012824650910938e-06,
+      "loss": 1.6594,
+      "step": 9474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.99367787721053e-06,
+      "loss": 1.8015,
+      "step": 9475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.974553634972848e-06,
+      "loss": 1.8669,
+      "step": 9476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.955451925978325e-06,
+      "loss": 1.8722,
+      "step": 9477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.9363727520054e-06,
+      "loss": 1.6388,
+      "step": 9478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.91731611483043e-06,
+      "loss": 1.921,
+      "step": 9479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.898282016227638e-06,
+      "loss": 1.7658,
+      "step": 9480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.879270457969184e-06,
+      "loss": 1.7375,
+      "step": 9481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.860281441825018e-06,
+      "loss": 2.0597,
+      "step": 9482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.841314969563084e-06,
+      "loss": 1.6218,
+      "step": 9483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.822371042949277e-06,
+      "loss": 1.8908,
+      "step": 9484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.80344966374727e-06,
+      "loss": 1.7077,
+      "step": 9485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.784550833718707e-06,
+      "loss": 1.7687,
+      "step": 9486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.76567455462318e-06,
+      "loss": 1.8317,
+      "step": 9487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.746820828218038e-06,
+      "loss": 1.9934,
+      "step": 9488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.72798965625865e-06,
+      "loss": 1.7044,
+      "step": 9489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.709181040498253e-06,
+      "loss": 1.8437,
+      "step": 9490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.690394982687976e-06,
+      "loss": 1.8071,
+      "step": 9491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.671631484576891e-06,
+      "loss": 1.9083,
+      "step": 9492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.652890547911878e-06,
+      "loss": 2.0838,
+      "step": 9493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.634172174437793e-06,
+      "loss": 1.8805,
+      "step": 9494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.61547636589735e-06,
+      "loss": 1.781,
+      "step": 9495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.5968031240312425e-06,
+      "loss": 1.8384,
+      "step": 9496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.5781524505779134e-06,
+      "loss": 1.7185,
+      "step": 9497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.559524347273861e-06,
+      "loss": 1.9397,
+      "step": 9498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.540918815853365e-06,
+      "loss": 1.8023,
+      "step": 9499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.522335858048706e-06,
+      "loss": 1.781,
+      "step": 9500
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.8475309610366821,
+      "eval_runtime": 184.1184,
+      "eval_samples_per_second": 457.336,
+      "eval_steps_per_second": 28.585,
+      "step": 9500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.503775475589975e-06,
+      "loss": 1.8724,
+      "step": 9501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.485237670205175e-06,
+      "loss": 1.6945,
+      "step": 9502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.4667224436202604e-06,
+      "loss": 1.702,
+      "step": 9503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.448229797559047e-06,
+      "loss": 1.9217,
+      "step": 9504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.429759733743241e-06,
+      "loss": 1.7567,
+      "step": 9505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.411312253892466e-06,
+      "loss": 2.0501,
+      "step": 9506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.392887359724209e-06,
+      "loss": 1.9541,
+      "step": 9507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3744850529539334e-06,
+      "loss": 1.7645,
+      "step": 9508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.356105335294905e-06,
+      "loss": 1.8426,
+      "step": 9509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.337748208458311e-06,
+      "loss": 1.8454,
+      "step": 9510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.319413674153286e-06,
+      "loss": 1.8901,
+      "step": 9511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3011017340868234e-06,
+      "loss": 1.7572,
+      "step": 9512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.282812389963784e-06,
+      "loss": 1.798,
+      "step": 9513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.2645456434869975e-06,
+      "loss": 1.8959,
+      "step": 9514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.246301496357133e-06,
+      "loss": 1.9216,
+      "step": 9515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.228079950272748e-06,
+      "loss": 1.862,
+      "step": 9516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.209881006930346e-06,
+      "loss": 1.6985,
+      "step": 9517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.191704668024296e-06,
+      "loss": 1.7571,
+      "step": 9518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.173550935246853e-06,
+      "loss": 1.7568,
+      "step": 9519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.155419810288221e-06,
+      "loss": 1.6297,
+      "step": 9520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.137311294836413e-06,
+      "loss": 1.9233,
+      "step": 9521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.119225390577383e-06,
+      "loss": 1.9626,
+      "step": 9522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.10116209919498e-06,
+      "loss": 1.9186,
+      "step": 9523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.083121422370997e-06,
+      "loss": 1.8546,
+      "step": 9524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.065103361785008e-06,
+      "loss": 1.704,
+      "step": 9525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.047107919114587e-06,
+      "loss": 1.8484,
+      "step": 9526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.029135096035117e-06,
+      "loss": 1.6141,
+      "step": 9527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.011184894219952e-06,
+      "loss": 1.7605,
+      "step": 9528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.993257315340312e-06,
+      "loss": 1.7954,
+      "step": 9529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.975352361065307e-06,
+      "loss": 1.9415,
+      "step": 9530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.95747003306188e-06,
+      "loss": 1.6915,
+      "step": 9531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.939610332994978e-06,
+      "loss": 1.8442,
+      "step": 9532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.921773262527409e-06,
+      "loss": 1.6558,
+      "step": 9533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.903958823319789e-06,
+      "loss": 1.664,
+      "step": 9534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.886167017030709e-06,
+      "loss": 1.8064,
+      "step": 9535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.868397845316676e-06,
+      "loss": 1.8095,
+      "step": 9536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.850651309832034e-06,
+      "loss": 1.8371,
+      "step": 9537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.8329274122290175e-06,
+      "loss": 1.7492,
+      "step": 9538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.815226154157778e-06,
+      "loss": 1.9019,
+      "step": 9539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.797547537266358e-06,
+      "loss": 1.9176,
+      "step": 9540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.779891563200663e-06,
+      "loss": 1.8165,
+      "step": 9541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.762258233604546e-06,
+      "loss": 1.9629,
+      "step": 9542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.74464755011972e-06,
+      "loss": 1.7805,
+      "step": 9543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.727059514385764e-06,
+      "loss": 1.7893,
+      "step": 9544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.709494128040145e-06,
+      "loss": 1.7759,
+      "step": 9545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.691951392718332e-06,
+      "loss": 1.8131,
+      "step": 9546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.674431310053519e-06,
+      "loss": 1.7898,
+      "step": 9547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.656933881676985e-06,
+      "loss": 1.5334,
+      "step": 9548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.63945910921765e-06,
+      "loss": 1.9175,
+      "step": 9549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.622006994302543e-06,
+      "loss": 1.7595,
+      "step": 9550
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 1.846955418586731,
+      "eval_runtime": 184.1115,
+      "eval_samples_per_second": 457.353,
+      "eval_steps_per_second": 28.586,
+      "step": 9550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.604577538556506e-06,
+      "loss": 1.9016,
+      "step": 9551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.587170743602238e-06,
+      "loss": 2.0286,
+      "step": 9552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.569786611060391e-06,
+      "loss": 1.9428,
+      "step": 9553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.552425142549501e-06,
+      "loss": 1.6346,
+      "step": 9554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.535086339685886e-06,
+      "loss": 1.9161,
+      "step": 9555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.517770204083895e-06,
+      "loss": 1.843,
+      "step": 9556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.5004767373556815e-06,
+      "loss": 1.9725,
+      "step": 9557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.483205941111348e-06,
+      "loss": 1.4823,
+      "step": 9558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.4659578169588285e-06,
+      "loss": 1.8939,
+      "step": 9559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.448732366503979e-06,
+      "loss": 1.6838,
+      "step": 9560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.431529591350543e-06,
+      "loss": 1.6904,
+      "step": 9561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.41434949310013e-06,
+      "loss": 1.9207,
+      "step": 9562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.397192073352293e-06,
+      "loss": 1.8146,
+      "step": 9563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.380057333704392e-06,
+      "loss": 1.8655,
+      "step": 9564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.3629452757517355e-06,
+      "loss": 1.746,
+      "step": 9565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.345855901087522e-06,
+      "loss": 1.7372,
+      "step": 9566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.328789211302754e-06,
+      "loss": 1.814,
+      "step": 9567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.311745207986469e-06,
+      "loss": 1.7341,
+      "step": 9568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.2947238927254514e-06,
+      "loss": 1.933,
+      "step": 9569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.2777252671044895e-06,
+      "loss": 1.9606,
+      "step": 9570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.260749332706178e-06,
+      "loss": 1.7715,
+      "step": 9571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.243796091111003e-06,
+      "loss": 1.7386,
+      "step": 9572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.226865543897342e-06,
+      "loss": 1.8537,
+      "step": 9573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.209957692641544e-06,
+      "loss": 1.9752,
+      "step": 9574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.1930725389177375e-06,
+      "loss": 1.7903,
+      "step": 9575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.176210084297973e-06,
+      "loss": 1.9137,
+      "step": 9576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.159370330352215e-06,
+      "loss": 1.9061,
+      "step": 9577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.142553278648239e-06,
+      "loss": 1.8186,
+      "step": 9578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.125758930751818e-06,
+      "loss": 1.8374,
+      "step": 9579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.108987288226536e-06,
+      "loss": 1.6138,
+      "step": 9580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.092238352633866e-06,
+      "loss": 1.9114,
+      "step": 9581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0755121255332e-06,
+      "loss": 1.611,
+      "step": 9582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.058808608481792e-06,
+      "loss": 2.0577,
+      "step": 9583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.042127803034758e-06,
+      "loss": 1.7254,
+      "step": 9584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.025469710745163e-06,
+      "loss": 1.97,
+      "step": 9585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.008834333163876e-06,
+      "loss": 1.7204,
+      "step": 9586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.992221671839798e-06,
+      "loss": 1.8052,
+      "step": 9587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.975631728319497e-06,
+      "loss": 2.0741,
+      "step": 9588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.959064504147571e-06,
+      "loss": 1.9434,
+      "step": 9589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.942520000866508e-06,
+      "loss": 1.6202,
+      "step": 9590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.92599822001666e-06,
+      "loss": 1.8066,
+      "step": 9591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.909499163136184e-06,
+      "loss": 1.7761,
+      "step": 9592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.893022831761297e-06,
+      "loss": 1.6028,
+      "step": 9593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.876569227425854e-06,
+      "loss": 1.9733,
+      "step": 9594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.860138351661826e-06,
+      "loss": 1.7872,
+      "step": 9595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.843730205998959e-06,
+      "loss": 1.8449,
+      "step": 9596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.827344791964867e-06,
+      "loss": 2.0093,
+      "step": 9597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8109821110851056e-06,
+      "loss": 2.0144,
+      "step": 9598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.794642164883096e-06,
+      "loss": 1.8476,
+      "step": 9599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.778324954880093e-06,
+      "loss": 2.1111,
+      "step": 9600
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 1.846481442451477,
+      "eval_runtime": 184.107,
+      "eval_samples_per_second": 457.364,
+      "eval_steps_per_second": 28.587,
+      "step": 9600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7620304825952705e-06,
+      "loss": 1.9324,
+      "step": 9601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.745758749545749e-06,
+      "loss": 1.7215,
+      "step": 9602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.729509757246426e-06,
+      "loss": 1.6915,
+      "step": 9603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7132835072101484e-06,
+      "loss": 1.7068,
+      "step": 9604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6970800009475964e-06,
+      "loss": 1.8007,
+      "step": 9605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.680899239967369e-06,
+      "loss": 1.8582,
+      "step": 9606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.664741225775982e-06,
+      "loss": 1.7169,
+      "step": 9607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.648605959877734e-06,
+      "loss": 1.7919,
+      "step": 9608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.632493443774922e-06,
+      "loss": 1.7779,
+      "step": 9609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.616403678967624e-06,
+      "loss": 1.8312,
+      "step": 9610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.600336666953837e-06,
+      "loss": 1.8512,
+      "step": 9611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.584292409229475e-06,
+      "loss": 1.9428,
+      "step": 9612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.568270907288286e-06,
+      "loss": 1.825,
+      "step": 9613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.552272162621913e-06,
+      "loss": 1.7989,
+      "step": 9614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.536296176719912e-06,
+      "loss": 1.874,
+      "step": 9615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5203429510696515e-06,
+      "loss": 2.0802,
+      "step": 9616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5044124871564125e-06,
+      "loss": 1.9157,
+      "step": 9617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.488504786463427e-06,
+      "loss": 1.8388,
+      "step": 9618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.472619850471678e-06,
+      "loss": 1.6795,
+      "step": 9619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.456757680660174e-06,
+      "loss": 1.7316,
+      "step": 9620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.44091827850568e-06,
+      "loss": 1.7355,
+      "step": 9621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.425101645482849e-06,
+      "loss": 1.7895,
+      "step": 9622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.409307783064338e-06,
+      "loss": 1.799,
+      "step": 9623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.393536692720524e-06,
+      "loss": 2.0976,
+      "step": 9624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.37778837591979e-06,
+      "loss": 1.7763,
+      "step": 9625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3620628341283225e-06,
+      "loss": 2.0796,
+      "step": 9626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.346360068810258e-06,
+      "loss": 1.6176,
+      "step": 9627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.330680081427508e-06,
+      "loss": 1.733,
+      "step": 9628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.315022873439962e-06,
+      "loss": 1.9116,
+      "step": 9629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.299388446305342e-06,
+      "loss": 1.7972,
+      "step": 9630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2837768014792065e-06,
+      "loss": 1.92,
+      "step": 9631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2681879404151144e-06,
+      "loss": 1.8323,
+      "step": 9632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.252621864564405e-06,
+      "loss": 1.7455,
+      "step": 9633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.237078575376336e-06,
+      "loss": 1.6063,
+      "step": 9634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.221558074297972e-06,
+      "loss": 1.7488,
+      "step": 9635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.206060362774379e-06,
+      "loss": 1.7428,
+      "step": 9636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.19058544224843e-06,
+      "loss": 1.5514,
+      "step": 9637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.175133314160891e-06,
+      "loss": 1.8594,
+      "step": 9638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.159703979950358e-06,
+      "loss": 2.0246,
+      "step": 9639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1442974410533524e-06,
+      "loss": 1.8224,
+      "step": 9640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.128913698904281e-06,
+      "loss": 1.8356,
+      "step": 9641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.113552754935413e-06,
+      "loss": 1.805,
+      "step": 9642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0982146105769125e-06,
+      "loss": 1.6623,
+      "step": 9643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.082899267256774e-06,
+      "loss": 1.879,
+      "step": 9644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.067606726400914e-06,
+      "loss": 1.7907,
+      "step": 9645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.052336989433082e-06,
+      "loss": 1.948,
+      "step": 9646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.037090057774974e-06,
+      "loss": 1.6777,
+      "step": 9647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0218659328460926e-06,
+      "loss": 1.9231,
+      "step": 9648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0066646160638886e-06,
+      "loss": 1.8997,
+      "step": 9649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.99148610884359e-06,
+      "loss": 1.9001,
+      "step": 9650
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 1.8460966348648071,
+      "eval_runtime": 184.2468,
+      "eval_samples_per_second": 457.017,
+      "eval_steps_per_second": 28.565,
+      "step": 9650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.976330412598374e-06,
+      "loss": 1.6818,
+      "step": 9651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.961197528739303e-06,
+      "loss": 1.7344,
+      "step": 9652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.94608745867528e-06,
+      "loss": 1.8603,
+      "step": 9653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9310002038130685e-06,
+      "loss": 1.9547,
+      "step": 9654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.915935765557405e-06,
+      "loss": 1.7284,
+      "step": 9655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.900894145310753e-06,
+      "loss": 1.7849,
+      "step": 9656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.885875344473545e-06,
+      "loss": 1.6234,
+      "step": 9657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.870879364444108e-06,
+      "loss": 1.9758,
+      "step": 9658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8559062066185765e-06,
+      "loss": 1.6734,
+      "step": 9659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8409558723910005e-06,
+      "loss": 1.7731,
+      "step": 9660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.826028363153323e-06,
+      "loss": 1.7433,
+      "step": 9661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8111236802952925e-06,
+      "loss": 1.8334,
+      "step": 9662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.796241825204633e-06,
+      "loss": 1.8002,
+      "step": 9663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.78138279926682e-06,
+      "loss": 1.8631,
+      "step": 9664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7665466038653015e-06,
+      "loss": 1.7335,
+      "step": 9665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.751733240381389e-06,
+      "loss": 1.5063,
+      "step": 9666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.736942710194231e-06,
+      "loss": 1.899,
+      "step": 9667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.722175014680835e-06,
+      "loss": 1.7968,
+      "step": 9668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.707430155216158e-06,
+      "loss": 1.6697,
+      "step": 9669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.69270813317299e-06,
+      "loss": 1.6725,
+      "step": 9670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678008949921986e-06,
+      "loss": 1.6259,
+      "step": 9671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.663332606831661e-06,
+      "loss": 1.6799,
+      "step": 9672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.648679105268422e-06,
+      "loss": 1.7868,
+      "step": 9673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.63404844659654e-06,
+      "loss": 1.7646,
+      "step": 9674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.619440632178229e-06,
+      "loss": 1.6277,
+      "step": 9675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.604855663373458e-06,
+      "loss": 1.6412,
+      "step": 9676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.590293541540197e-06,
+      "loss": 1.8403,
+      "step": 9677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.575754268034138e-06,
+      "loss": 1.6433,
+      "step": 9678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.561237844208977e-06,
+      "loss": 1.8465,
+      "step": 9679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5467442714161865e-06,
+      "loss": 1.7556,
+      "step": 9680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532273551005217e-06,
+      "loss": 1.8173,
+      "step": 9681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.517825684323323e-06,
+      "loss": 1.8707,
+      "step": 9682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.503400672715624e-06,
+      "loss": 1.8824,
+      "step": 9683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.488998517525128e-06,
+      "loss": 1.4851,
+      "step": 9684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.474619220092707e-06,
+      "loss": 1.6153,
+      "step": 9685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.460262781757124e-06,
+      "loss": 1.7499,
+      "step": 9686
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.445929203855004e-06,
+      "loss": 1.8627,
+      "step": 9687
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.43161848772089e-06,
+      "loss": 1.6125,
+      "step": 9688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4173306346870504e-06,
+      "loss": 1.9373,
+      "step": 9689
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.403065646083809e-06,
+      "loss": 1.7152,
+      "step": 9690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.388823523239216e-06,
+      "loss": 1.7186,
+      "step": 9691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.374604267479293e-06,
+      "loss": 1.8494,
+      "step": 9692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.360407880127898e-06,
+      "loss": 1.8043,
+      "step": 9693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.346234362506724e-06,
+      "loss": 1.5732,
+      "step": 9694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.332083715935381e-06,
+      "loss": 1.9424,
+      "step": 9695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3179559417313445e-06,
+      "loss": 1.9099,
+      "step": 9696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.303851041209922e-06,
+      "loss": 1.812,
+      "step": 9697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.289769015684314e-06,
+      "loss": 1.7591,
+      "step": 9698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.275709866465666e-06,
+      "loss": 1.638,
+      "step": 9699
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.261673594862819e-06,
+      "loss": 1.85,
+      "step": 9700
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 1.8458763360977173,
+      "eval_runtime": 185.298,
+      "eval_samples_per_second": 454.425,
+      "eval_steps_per_second": 28.403,
+      "step": 9700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.247660202182674e-06,
+      "loss": 1.7233,
+      "step": 9701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.233669689729852e-06,
+      "loss": 1.9633,
+      "step": 9702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.219702058806951e-06,
+      "loss": 1.7517,
+      "step": 9703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.205757310714403e-06,
+      "loss": 2.0551,
+      "step": 9704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.191835446750503e-06,
+      "loss": 1.8745,
+      "step": 9705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.17793646821138e-06,
+      "loss": 1.7527,
+      "step": 9706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.164060376391083e-06,
+      "loss": 1.8654,
+      "step": 9707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.150207172581522e-06,
+      "loss": 1.8678,
+      "step": 9708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.136376858072472e-06,
+      "loss": 1.9147,
+      "step": 9709
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.122569434151568e-06,
+      "loss": 1.8748,
+      "step": 9710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.10878490210434e-06,
+      "loss": 1.7859,
+      "step": 9711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.095023263214121e-06,
+      "loss": 1.7039,
+      "step": 9712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.081284518762163e-06,
+      "loss": 1.856,
+      "step": 9713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0675686700276386e-06,
+      "loss": 1.6086,
+      "step": 9714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.05387571828747e-06,
+      "loss": 1.7202,
+      "step": 9715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.040205664816554e-06,
+      "loss": 1.845,
+      "step": 9716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.026558510887568e-06,
+      "loss": 1.6675,
+      "step": 9717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.012934257771134e-06,
+      "loss": 1.7835,
+      "step": 9718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.999332906735709e-06,
+      "loss": 1.7372,
+      "step": 9719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9857544590475856e-06,
+      "loss": 1.8428,
+      "step": 9720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.972198915970976e-06,
+      "loss": 1.8439,
+      "step": 9721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.958666278767953e-06,
+      "loss": 1.8718,
+      "step": 9722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.945156548698398e-06,
+      "loss": 1.8499,
+      "step": 9723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.931669727020137e-06,
+      "loss": 1.7705,
+      "step": 9724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.918205814988834e-06,
+      "loss": 1.831,
+      "step": 9725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9047648138580135e-06,
+      "loss": 1.9016,
+      "step": 9726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8913467248790635e-06,
+      "loss": 1.8662,
+      "step": 9727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.877951549301234e-06,
+      "loss": 1.854,
+      "step": 9728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.864579288371695e-06,
+      "loss": 1.7461,
+      "step": 9729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.851229943335394e-06,
+      "loss": 1.8213,
+      "step": 9730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.837903515435226e-06,
+      "loss": 1.9076,
+      "step": 9731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.824600005911893e-06,
+      "loss": 1.7883,
+      "step": 9732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8113194160039886e-06,
+      "loss": 1.725,
+      "step": 9733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.798061746947995e-06,
+      "loss": 1.9345,
+      "step": 9734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7848269999782303e-06,
+      "loss": 1.9671,
+      "step": 9735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.771615176326876e-06,
+      "loss": 1.8561,
+      "step": 9736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7584262772239752e-06,
+      "loss": 1.9815,
+      "step": 9737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7452603038974908e-06,
+      "loss": 1.9721,
+      "step": 9738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.73211725757322e-06,
+      "loss": 1.847,
+      "step": 9739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.718997139474767e-06,
+      "loss": 1.9575,
+      "step": 9740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7058999508236544e-06,
+      "loss": 1.693,
+      "step": 9741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6928256928393246e-06,
+      "loss": 1.7931,
+      "step": 9742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.679774366738969e-06,
+      "loss": 1.8003,
+      "step": 9743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6667459737377286e-06,
+      "loss": 1.8321,
+      "step": 9744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.653740515048576e-06,
+      "loss": 1.8723,
+      "step": 9745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6407579918823484e-06,
+      "loss": 2.0314,
+      "step": 9746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.627798405447774e-06,
+      "loss": 1.733,
+      "step": 9747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6148617569514154e-06,
+      "loss": 1.9373,
+      "step": 9748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6019480475976975e-06,
+      "loss": 1.8044,
+      "step": 9749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.589057278588964e-06,
+      "loss": 1.9016,
+      "step": 9750
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.8454468250274658,
+      "eval_runtime": 184.3297,
+      "eval_samples_per_second": 456.812,
+      "eval_steps_per_second": 28.552,
+      "step": 9750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.576189451125339e-06,
+      "loss": 1.7107,
+      "step": 9751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5633445664048635e-06,
+      "loss": 1.8796,
+      "step": 9752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5505226256234145e-06,
+      "loss": 1.8313,
+      "step": 9753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.537723629974815e-06,
+      "loss": 1.8534,
+      "step": 9754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.524947580650639e-06,
+      "loss": 1.9485,
+      "step": 9755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5121944788403525e-06,
+      "loss": 1.689,
+      "step": 9756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4994643257313386e-06,
+      "loss": 1.7356,
+      "step": 9757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4867571225088156e-06,
+      "loss": 1.9207,
+      "step": 9758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4740728703558377e-06,
+      "loss": 1.9566,
+      "step": 9759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4614115704533768e-06,
+      "loss": 1.8419,
+      "step": 9760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4487732239801563e-06,
+      "loss": 1.7056,
+      "step": 9761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4361578321129296e-06,
+      "loss": 1.866,
+      "step": 9762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4235653960262012e-06,
+      "loss": 1.7917,
+      "step": 9763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41099591689234e-06,
+      "loss": 1.9162,
+      "step": 9764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.398449395881603e-06,
+      "loss": 1.8053,
+      "step": 9765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3859258341621123e-06,
+      "loss": 1.8165,
+      "step": 9766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3734252328998796e-06,
+      "loss": 1.817,
+      "step": 9767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3609475932586674e-06,
+      "loss": 1.9258,
+      "step": 9768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.348492916400242e-06,
+      "loss": 1.8079,
+      "step": 9769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3360612034841476e-06,
+      "loss": 1.7529,
+      "step": 9770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.323652455667847e-06,
+      "loss": 1.9133,
+      "step": 9771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3112666741065834e-06,
+      "loss": 1.7277,
+      "step": 9772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.298903859953517e-06,
+      "loss": 1.8891,
+      "step": 9773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2865640143596444e-06,
+      "loss": 1.773,
+      "step": 9774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2742471384738794e-06,
+      "loss": 1.8098,
+      "step": 9775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2619532334429435e-06,
+      "loss": 1.9513,
+      "step": 9776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.249682300411422e-06,
+      "loss": 1.6533,
+      "step": 9777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2374343405217886e-06,
+      "loss": 1.6855,
+      "step": 9778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2252093549143546e-06,
+      "loss": 1.6699,
+      "step": 9779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.213007344727292e-06,
+      "loss": 1.7059,
+      "step": 9780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2008283110966374e-06,
+      "loss": 1.7604,
+      "step": 9781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1886722551563164e-06,
+      "loss": 1.8736,
+      "step": 9782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1765391780380902e-06,
+      "loss": 1.9039,
+      "step": 9783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.164429080871556e-06,
+      "loss": 1.7701,
+      "step": 9784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1523419647842276e-06,
+      "loss": 1.8307,
+      "step": 9785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1402778309014278e-06,
+      "loss": 1.8381,
+      "step": 9786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1282366803463412e-06,
+      "loss": 1.708,
+      "step": 9787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.116218514240099e-06,
+      "loss": 1.676,
+      "step": 9788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1042233337015835e-06,
+      "loss": 1.4671,
+      "step": 9789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.092251139847568e-06,
+      "loss": 1.866,
+      "step": 9790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0803019337926895e-06,
+      "loss": 1.8795,
+      "step": 9791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.068375716649502e-06,
+      "loss": 1.8446,
+      "step": 9792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.056472489528367e-06,
+      "loss": 1.886,
+      "step": 9793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.044592253537426e-06,
+      "loss": 1.8574,
+      "step": 9794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.03273500978285e-06,
+      "loss": 1.8582,
+      "step": 9795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.020900759368561e-06,
+      "loss": 1.8783,
+      "step": 9796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.009089503396345e-06,
+      "loss": 1.8777,
+      "step": 9797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.99730124296585e-06,
+      "loss": 1.6736,
+      "step": 9798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.985535979174614e-06,
+      "loss": 1.8393,
+      "step": 9799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9737937131180393e-06,
+      "loss": 1.9073,
+      "step": 9800
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.8452401161193848,
+      "eval_runtime": 184.513,
+      "eval_samples_per_second": 456.358,
+      "eval_steps_per_second": 28.524,
+      "step": 9800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9620744458893066e-06,
+      "loss": 1.7999,
+      "step": 9801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9503781785795713e-06,
+      "loss": 1.6535,
+      "step": 9802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.93870491227774e-06,
+      "loss": 1.8123,
+      "step": 9803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.927054648070665e-06,
+      "loss": 1.8595,
+      "step": 9804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.915427387043007e-06,
+      "loss": 1.7081,
+      "step": 9805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.903823130277289e-06,
+      "loss": 1.6988,
+      "step": 9806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.892241878853896e-06,
+      "loss": 1.7304,
+      "step": 9807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.880683633851078e-06,
+      "loss": 1.5519,
+      "step": 9808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.869148396344945e-06,
+      "loss": 1.8428,
+      "step": 9809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.857636167409472e-06,
+      "loss": 1.9092,
+      "step": 9810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.846146948116468e-06,
+      "loss": 2.0253,
+      "step": 9811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.834680739535578e-06,
+      "loss": 1.8325,
+      "step": 9812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8232375427343915e-06,
+      "loss": 1.8944,
+      "step": 9813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8118173587782513e-06,
+      "loss": 1.7633,
+      "step": 9814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8004201887304735e-06,
+      "loss": 1.8906,
+      "step": 9815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7890460336520984e-06,
+      "loss": 1.8393,
+      "step": 9816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7776948946021396e-06,
+      "loss": 1.9667,
+      "step": 9817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.766366772637391e-06,
+      "loss": 1.7288,
+      "step": 9818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.755061668812564e-06,
+      "loss": 1.9046,
+      "step": 9819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7437795841801503e-06,
+      "loss": 1.7276,
+      "step": 9820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7325205197905876e-06,
+      "loss": 1.963,
+      "step": 9821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.721284476692093e-06,
+      "loss": 2.033,
+      "step": 9822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7100714559308015e-06,
+      "loss": 1.892,
+      "step": 9823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.698881458550656e-06,
+      "loss": 1.6038,
+      "step": 9824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6877144855934622e-06,
+      "loss": 1.8139,
+      "step": 9825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6765705380989436e-06,
+      "loss": 1.9841,
+      "step": 9826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6654496171046037e-06,
+      "loss": 1.7106,
+      "step": 9827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.654351723645809e-06,
+      "loss": 2.0808,
+      "step": 9828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6432768587558442e-06,
+      "loss": 1.7915,
+      "step": 9829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6322250234658006e-06,
+      "loss": 1.7262,
+      "step": 9830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6211962188046335e-06,
+      "loss": 1.8177,
+      "step": 9831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.610190445799132e-06,
+      "loss": 1.8309,
+      "step": 9832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.599207705473977e-06,
+      "loss": 1.7222,
+      "step": 9833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5882479988517394e-06,
+      "loss": 1.9619,
+      "step": 9834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.577311326952714e-06,
+      "loss": 1.8659,
+      "step": 9835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.566397690795197e-06,
+      "loss": 1.7195,
+      "step": 9836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5555070913952374e-06,
+      "loss": 1.8317,
+      "step": 9837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.544639529766829e-06,
+      "loss": 1.8022,
+      "step": 9838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5337950069217464e-06,
+      "loss": 1.8646,
+      "step": 9839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5229735238696263e-06,
+      "loss": 1.913,
+      "step": 9840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5121750816179956e-06,
+      "loss": 1.9458,
+      "step": 9841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5013996811722174e-06,
+      "loss": 1.92,
+      "step": 9842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4906473235355444e-06,
+      "loss": 1.7519,
+      "step": 9843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.479918009708981e-06,
+      "loss": 1.8435,
+      "step": 9844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.469211740691535e-06,
+      "loss": 1.7248,
+      "step": 9845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4585285174799355e-06,
+      "loss": 1.8241,
+      "step": 9846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4478683410688317e-06,
+      "loss": 1.6987,
+      "step": 9847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.437231212450708e-06,
+      "loss": 2.0784,
+      "step": 9848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.426617132615938e-06,
+      "loss": 1.8937,
+      "step": 9849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.416026102552732e-06,
+      "loss": 1.7666,
+      "step": 9850
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.8451992273330688,
+      "eval_runtime": 185.1211,
+      "eval_samples_per_second": 454.859,
+      "eval_steps_per_second": 28.43,
+      "step": 9850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4054581232470786e-06,
+      "loss": 1.7778,
+      "step": 9851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3949131956829405e-06,
+      "loss": 2.049,
+      "step": 9852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3843913208420607e-06,
+      "loss": 1.7952,
+      "step": 9853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3738924997040723e-06,
+      "loss": 1.9419,
+      "step": 9854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.363416733246415e-06,
+      "loss": 1.8038,
+      "step": 9855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3529640224444758e-06,
+      "loss": 1.6511,
+      "step": 9856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3425343682713374e-06,
+      "loss": 1.7666,
+      "step": 9857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3321277716980836e-06,
+      "loss": 1.4855,
+      "step": 9858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.321744233693607e-06,
+      "loss": 1.8745,
+      "step": 9859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.311383755224633e-06,
+      "loss": 1.806,
+      "step": 9860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3010463372557243e-06,
+      "loss": 1.8485,
+      "step": 9861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.290731980749361e-06,
+      "loss": 1.7185,
+      "step": 9862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.28044068666583e-06,
+      "loss": 1.8435,
+      "step": 9863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2701724559632544e-06,
+      "loss": 1.7675,
+      "step": 9864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2599272895976465e-06,
+      "loss": 1.7325,
+      "step": 9865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2497051885228823e-06,
+      "loss": 1.6684,
+      "step": 9866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2395061536906737e-06,
+      "loss": 1.998,
+      "step": 9867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.22933018605051e-06,
+      "loss": 1.9044,
+      "step": 9868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.219177286549884e-06,
+      "loss": 1.9219,
+      "step": 9869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2090474561340113e-06,
+      "loss": 1.6695,
+      "step": 9870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.198940695745999e-06,
+      "loss": 1.7193,
+      "step": 9871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.188857006326872e-06,
+      "loss": 1.6629,
+      "step": 9872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.178796388815407e-06,
+      "loss": 1.6775,
+      "step": 9873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.168758844148272e-06,
+      "loss": 1.7156,
+      "step": 9874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1587443732599964e-06,
+      "loss": 1.6606,
+      "step": 9875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1487529770829737e-06,
+      "loss": 2.0149,
+      "step": 9876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1387846565474044e-06,
+      "loss": 1.6192,
+      "step": 9877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1288394125813794e-06,
+      "loss": 1.5776,
+      "step": 9878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.118917246110852e-06,
+      "loss": 1.9676,
+      "step": 9879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.109018158059556e-06,
+      "loss": 1.9678,
+      "step": 9880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0991421493491426e-06,
+      "loss": 1.8288,
+      "step": 9881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0892892208990987e-06,
+      "loss": 1.7863,
+      "step": 9882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0794593736267744e-06,
+      "loss": 2.0286,
+      "step": 9883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069652608447353e-06,
+      "loss": 1.7673,
+      "step": 9884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0598689262738556e-06,
+      "loss": 2.0168,
+      "step": 9885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.050108328017164e-06,
+      "loss": 1.9019,
+      "step": 9886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0403708145860513e-06,
+      "loss": 1.7891,
+      "step": 9887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.03065638688707e-06,
+      "loss": 1.9449,
+      "step": 9888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0209650458246907e-06,
+      "loss": 1.769,
+      "step": 9889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.011296792301165e-06,
+      "loss": 1.7127,
+      "step": 9890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0016516272166886e-06,
+      "loss": 1.9346,
+      "step": 9891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.99202955146921e-06,
+      "loss": 1.8715,
+      "step": 9892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.982430565954596e-06,
+      "loss": 1.7457,
+      "step": 9893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.972854671566493e-06,
+      "loss": 1.9689,
+      "step": 9894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9633018691964932e-06,
+      "loss": 2.0452,
+      "step": 9895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9537721597339966e-06,
+      "loss": 1.8356,
+      "step": 9896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9442655440662093e-06,
+      "loss": 1.6502,
+      "step": 9897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9347820230782296e-06,
+      "loss": 1.722,
+      "step": 9898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9253215976530183e-06,
+      "loss": 1.8295,
+      "step": 9899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9158842686713416e-06,
+      "loss": 1.8695,
+      "step": 9900
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.8449455499649048,
+      "eval_runtime": 184.3251,
+      "eval_samples_per_second": 456.823,
+      "eval_steps_per_second": 28.553,
+      "step": 9900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9064700370118592e-06,
+      "loss": 1.7949,
+      "step": 9901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8970789035510638e-06,
+      "loss": 1.903,
+      "step": 9902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.887710869163284e-06,
+      "loss": 1.8049,
+      "step": 9903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8783659347207382e-06,
+      "loss": 1.7097,
+      "step": 9904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8690441010934256e-06,
+      "loss": 1.6797,
+      "step": 9905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8597453691492628e-06,
+      "loss": 1.6537,
+      "step": 9906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8504697397539738e-06,
+      "loss": 1.6923,
+      "step": 9907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.841217213771146e-06,
+      "loss": 2.0954,
+      "step": 9908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8319877920622285e-06,
+      "loss": 1.9122,
+      "step": 9909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.822781475486507e-06,
+      "loss": 2.0522,
+      "step": 9910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8135982649011007e-06,
+      "loss": 1.7275,
+      "step": 9911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8044381611609929e-06,
+      "loss": 1.8361,
+      "step": 9912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.795301165119001e-06,
+      "loss": 1.9466,
+      "step": 9913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7861872776258615e-06,
+      "loss": 1.7769,
+      "step": 9914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7770964995300343e-06,
+      "loss": 1.8731,
+      "step": 9915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.768028831677926e-06,
+      "loss": 1.8725,
+      "step": 9916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7589842749137774e-06,
+      "loss": 1.7662,
+      "step": 9917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7499628300796376e-06,
+      "loss": 1.6966,
+      "step": 9918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.740964498015446e-06,
+      "loss": 1.9555,
+      "step": 9919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7319892795589486e-06,
+      "loss": 1.944,
+      "step": 9920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.723037175545783e-06,
+      "loss": 1.7479,
+      "step": 9921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.714108186809421e-06,
+      "loss": 1.6866,
+      "step": 9922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.70520231418117e-06,
+      "loss": 1.8747,
+      "step": 9923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6963195584901724e-06,
+      "loss": 1.9708,
+      "step": 9924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6874599205634612e-06,
+      "loss": 1.7008,
+      "step": 9925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6786234012258761e-06,
+      "loss": 1.8983,
+      "step": 9926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6698100013001206e-06,
+      "loss": 1.7599,
+      "step": 9927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6610197216067601e-06,
+      "loss": 1.666,
+      "step": 9928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.652252562964196e-06,
+      "loss": 1.965,
+      "step": 9929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6435085261886917e-06,
+      "loss": 1.7079,
+      "step": 9930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6347876120942906e-06,
+      "loss": 2.0264,
+      "step": 9931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6260898214929543e-06,
+      "loss": 1.7127,
+      "step": 9932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6174151551945071e-06,
+      "loss": 1.7799,
+      "step": 9933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6087636140065532e-06,
+      "loss": 1.9025,
+      "step": 9934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6001351987345592e-06,
+      "loss": 1.8636,
+      "step": 9935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5915299101818825e-06,
+      "loss": 1.9521,
+      "step": 9936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5829477491496879e-06,
+      "loss": 1.6182,
+      "step": 9937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.574388716437003e-06,
+      "loss": 1.7813,
+      "step": 9938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.565852812840718e-06,
+      "loss": 1.8482,
+      "step": 9939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5573400391555025e-06,
+      "loss": 1.804,
+      "step": 9940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5488503961739454e-06,
+      "loss": 1.7159,
+      "step": 9941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5403838846864693e-06,
+      "loss": 1.8819,
+      "step": 9942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.531940505481305e-06,
+      "loss": 2.142,
+      "step": 9943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.523520259344574e-06,
+      "loss": 1.7473,
+      "step": 9944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5151231470602045e-06,
+      "loss": 1.8409,
+      "step": 9945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5067491694100155e-06,
+      "loss": 1.6558,
+      "step": 9946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4983983271736334e-06,
+      "loss": 1.7503,
+      "step": 9947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4900706211285199e-06,
+      "loss": 1.8854,
+      "step": 9948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4817660520500542e-06,
+      "loss": 1.8635,
+      "step": 9949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4734846207113961e-06,
+      "loss": 1.6799,
+      "step": 9950
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.8448765277862549,
+      "eval_runtime": 184.3616,
+      "eval_samples_per_second": 456.733,
+      "eval_steps_per_second": 28.547,
+      "step": 9950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4652263278835954e-06,
+      "loss": 1.8955,
+      "step": 9951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4569911743354536e-06,
+      "loss": 1.8648,
+      "step": 9952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4487791608337464e-06,
+      "loss": 2.171,
+      "step": 9953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4405902881430289e-06,
+      "loss": 1.7173,
+      "step": 9954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4324245570256633e-06,
+      "loss": 1.792,
+      "step": 9955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4242819682419582e-06,
+      "loss": 1.8336,
+      "step": 9956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4161625225499741e-06,
+      "loss": 1.7621,
+      "step": 9957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4080662207056894e-06,
+      "loss": 1.8323,
+      "step": 9958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.399993063462862e-06,
+      "loss": 1.8469,
+      "step": 9959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3919430515731412e-06,
+      "loss": 2.1061,
+      "step": 9960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3839161857859828e-06,
+      "loss": 1.8801,
+      "step": 9961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3759124668487056e-06,
+      "loss": 2.0302,
+      "step": 9962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3679318955065467e-06,
+      "loss": 1.7286,
+      "step": 9963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3599744725024399e-06,
+      "loss": 1.8169,
+      "step": 9964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3520401985772647e-06,
+      "loss": 1.6965,
+      "step": 9965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3441290744697632e-06,
+      "loss": 1.5745,
+      "step": 9966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.33624110091643e-06,
+      "loss": 1.9629,
+      "step": 9967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.328376278651705e-06,
+      "loss": 1.8396,
+      "step": 9968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3205346084077808e-06,
+      "loss": 2.013,
+      "step": 9969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3127160909147673e-06,
+      "loss": 2.0972,
+      "step": 9970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3049207269005826e-06,
+      "loss": 1.89,
+      "step": 9971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2971485170910069e-06,
+      "loss": 1.8998,
+      "step": 9972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2893994622096284e-06,
+      "loss": 1.9558,
+      "step": 9973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2816735629779253e-06,
+      "loss": 1.8046,
+      "step": 9974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2739708201152111e-06,
+      "loss": 1.6314,
+      "step": 9975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2662912343386067e-06,
+      "loss": 1.7857,
+      "step": 9976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2586348063630959e-06,
+      "loss": 1.5993,
+      "step": 9977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2510015369015527e-06,
+      "loss": 2.0333,
+      "step": 9978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2433914266646306e-06,
+      "loss": 1.676,
+      "step": 9979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2358044763608744e-06,
+      "loss": 1.8132,
+      "step": 9980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2282406866966079e-06,
+      "loss": 1.82,
+      "step": 9981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.220700058376073e-06,
+      "loss": 1.6948,
+      "step": 9982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2131825921012918e-06,
+      "loss": 1.6847,
+      "step": 9983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.205688288572232e-06,
+      "loss": 1.7004,
+      "step": 9984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.198217148486558e-06,
+      "loss": 1.8917,
+      "step": 9985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1907691725398794e-06,
+      "loss": 1.7607,
+      "step": 9986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1833443614256424e-06,
+      "loss": 1.7144,
+      "step": 9987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1759427158350988e-06,
+      "loss": 1.7033,
+      "step": 9988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1685642364573645e-06,
+      "loss": 2.0989,
+      "step": 9989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.16120892397939e-06,
+      "loss": 1.8021,
+      "step": 9990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1538767790859883e-06,
+      "loss": 1.91,
+      "step": 9991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.146567802459808e-06,
+      "loss": 1.9526,
+      "step": 9992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1392819947813326e-06,
+      "loss": 1.6584,
+      "step": 9993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1320193567288528e-06,
+      "loss": 1.8187,
+      "step": 9994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1247798889785777e-06,
+      "loss": 1.7378,
+      "step": 9995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1175635922045235e-06,
+      "loss": 1.7899,
+      "step": 9996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1103704670785419e-06,
+      "loss": 1.8002,
+      "step": 9997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1032005142703194e-06,
+      "loss": 1.7837,
+      "step": 9998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0960537344474332e-06,
+      "loss": 1.7413,
+      "step": 9999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0889301282752117e-06,
+      "loss": 1.9409,
+      "step": 10000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.844720482826233,
+      "eval_runtime": 184.2959,
+      "eval_samples_per_second": 456.896,
+      "eval_steps_per_second": 28.557,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 10306,
+  "num_train_epochs": 1,
+  "save_steps": 50,
+  "total_flos": 5.22584064e+16,
+  "trial_name": null,
+  "trial_params": null
+}