diff --git "a/72b/trainer_state.json" "b/72b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/72b/trainer_state.json"
@@ -0,0 +1,15297 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2181,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 35.6231464997871,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 9.3652,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 35.92330916133144,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 9.3382,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.231722225089904,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 8.8522,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.369954036743097,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 8.8564,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.663503624079658,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 8.4349,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 31.25181061747781,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 8.6229,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 37.33243831299671,
+      "learning_rate": 0.00010606060606060606,
+      "loss": 8.8206,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 24.270798474480543,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 7.3819,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 65.23057542361212,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 6.9387,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 52.07605968519506,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 7.0564,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.079150298164448,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 6.4266,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6688270060011425,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 6.1114,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.050211195613581,
+      "learning_rate": 0.00019696969696969695,
+      "loss": 6.207,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.636321004360315,
+      "learning_rate": 0.00021212121212121213,
+      "loss": 5.8461,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7193781177674055,
+      "learning_rate": 0.00022727272727272727,
+      "loss": 5.7272,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7939980680893302,
+      "learning_rate": 0.00024242424242424245,
+      "loss": 5.3522,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 31.475560112408974,
+      "learning_rate": 0.00025757575757575756,
+      "loss": 5.8448,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 18.95582034095008,
+      "learning_rate": 0.00027272727272727274,
+      "loss": 5.4235,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.860416105519063,
+      "learning_rate": 0.0002878787878787879,
+      "loss": 5.2066,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2909871451343204,
+      "learning_rate": 0.00030303030303030303,
+      "loss": 4.9889,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4575217589545706,
+      "learning_rate": 0.0003181818181818182,
+      "loss": 4.8823,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7934215420950284,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 4.6509,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.153137745543451,
+      "learning_rate": 0.0003484848484848485,
+      "loss": 4.4638,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.590465433445637,
+      "learning_rate": 0.00036363636363636367,
+      "loss": 4.5866,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1028291207140761,
+      "learning_rate": 0.0003787878787878788,
+      "loss": 4.3329,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9443674527308606,
+      "learning_rate": 0.0003939393939393939,
+      "loss": 4.2267,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6131706999245661,
+      "learning_rate": 0.00040909090909090913,
+      "loss": 4.4129,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9931732213719607,
+      "learning_rate": 0.00042424242424242425,
+      "loss": 4.2866,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4956897560760095,
+      "learning_rate": 0.0004393939393939394,
+      "loss": 4.0904,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.363838849342912,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 4.3782,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6873388271271459,
+      "learning_rate": 0.0004696969696969697,
+      "loss": 4.13,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.43150893817963,
+      "learning_rate": 0.0004848484848484849,
+      "loss": 4.241,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.48305076836344624,
+      "learning_rate": 0.0005,
+      "loss": 4.1711,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3350872433005161,
+      "learning_rate": 0.0005151515151515151,
+      "loss": 4.1739,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.35771104058944886,
+      "learning_rate": 0.0005303030303030302,
+      "loss": 4.1007,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.35394352895084363,
+      "learning_rate": 0.0005454545454545455,
+      "loss": 3.9501,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3556968965015149,
+      "learning_rate": 0.0005606060606060606,
+      "loss": 4.12,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5342531256035216,
+      "learning_rate": 0.0005757575757575758,
+      "loss": 4.078,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.40274236021233567,
+      "learning_rate": 0.0005909090909090909,
+      "loss": 4.1505,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.37982336204910594,
+      "learning_rate": 0.0006060606060606061,
+      "loss": 4.1318,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4427249973641121,
+      "learning_rate": 0.0006212121212121212,
+      "loss": 3.9231,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0419164218551258,
+      "learning_rate": 0.0006363636363636364,
+      "loss": 4.0269,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5567583515701285,
+      "learning_rate": 0.0006515151515151515,
+      "loss": 3.9712,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.41262532216581865,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 3.913,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3584993367687472,
+      "learning_rate": 0.0006818181818181818,
+      "loss": 3.9027,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3764530568011376,
+      "learning_rate": 0.000696969696969697,
+      "loss": 4.0029,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.28725402532410044,
+      "learning_rate": 0.0007121212121212122,
+      "loss": 4.0063,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.24639421197145955,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 3.9326,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2861555938572352,
+      "learning_rate": 0.0007424242424242425,
+      "loss": 3.943,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.44017978305923483,
+      "learning_rate": 0.0007575757575757576,
+      "loss": 3.9489,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4840463207066867,
+      "learning_rate": 0.0007727272727272727,
+      "loss": 3.942,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8957230667217483,
+      "learning_rate": 0.0007878787878787878,
+      "loss": 3.9037,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4816467809741688,
+      "learning_rate": 0.000803030303030303,
+      "loss": 3.8997,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.38237712591734363,
+      "learning_rate": 0.0008181818181818183,
+      "loss": 3.9628,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6996121842131515,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 3.9518,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3549458661135047,
+      "learning_rate": 0.0008484848484848485,
+      "loss": 4.0217,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8855649903773636,
+      "learning_rate": 0.0008636363636363636,
+      "loss": 3.9283,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.532703832375215,
+      "learning_rate": 0.0008787878787878789,
+      "loss": 3.9352,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4301527966259608,
+      "learning_rate": 0.000893939393939394,
+      "loss": 3.8933,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.36343822063257797,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 3.899,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4571230800818782,
+      "learning_rate": 0.0009242424242424242,
+      "loss": 3.8235,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4716767686854641,
+      "learning_rate": 0.0009393939393939394,
+      "loss": 3.8761,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44258928637702416,
+      "learning_rate": 0.0009545454545454546,
+      "loss": 3.8516,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5266597950036421,
+      "learning_rate": 0.0009696969696969698,
+      "loss": 3.8353,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.412600079316998,
+      "learning_rate": 0.000984848484848485,
+      "loss": 3.8122,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.36959748396394654,
+      "learning_rate": 0.001,
+      "loss": 3.7723,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.30613521235924535,
+      "learning_rate": 0.0009999994484067654,
+      "loss": 3.7035,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2806990997671294,
+      "learning_rate": 0.0009999977936282788,
+      "loss": 3.8159,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.28574560709629776,
+      "learning_rate": 0.0009999950356681913,
+      "loss": 3.6558,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4246176360205799,
+      "learning_rate": 0.0009999911745325876,
+      "loss": 3.7948,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4276343999050113,
+      "learning_rate": 0.0009999862102299873,
+      "loss": 3.6605,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3427873016989194,
+      "learning_rate": 0.0009999801427713433,
+      "loss": 3.7644,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4607730758750902,
+      "learning_rate": 0.0009999729721700424,
+      "loss": 3.6255,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.26964079915231465,
+      "learning_rate": 0.000999964698441906,
+      "loss": 3.7065,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.26937917391578037,
+      "learning_rate": 0.0009999553216051892,
+      "loss": 3.5876,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3120123704867196,
+      "learning_rate": 0.00099994484168058,
+      "loss": 3.7445,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.32349978172818317,
+      "learning_rate": 0.0009999332586912019,
+      "loss": 3.8054,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.29844230641343156,
+      "learning_rate": 0.0009999205726626108,
+      "loss": 3.5192,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.39790642080834976,
+      "learning_rate": 0.000999906783622797,
+      "loss": 3.5319,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3305412165578488,
+      "learning_rate": 0.0009998918916021842,
+      "loss": 3.7056,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.34563140622786687,
+      "learning_rate": 0.0009998758966336297,
+      "loss": 3.7685,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2825570292873176,
+      "learning_rate": 0.0009998587987524242,
+      "loss": 3.6081,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.22299810473790152,
+      "learning_rate": 0.0009998405979962926,
+      "loss": 3.5665,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3092802999538855,
+      "learning_rate": 0.000999821294405392,
+      "loss": 3.5616,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2576039617678053,
+      "learning_rate": 0.0009998008880223134,
+      "loss": 3.6937,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.21481665405008674,
+      "learning_rate": 0.000999779378892081,
+      "loss": 3.5144,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.25123404483566714,
+      "learning_rate": 0.0009997567670621522,
+      "loss": 3.6283,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2577115241815996,
+      "learning_rate": 0.0009997330525824165,
+      "loss": 3.5051,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3043087260371866,
+      "learning_rate": 0.0009997082355051976,
+      "loss": 3.5543,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3545016683615683,
+      "learning_rate": 0.000999682315885251,
+      "loss": 3.542,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.34386194955491955,
+      "learning_rate": 0.0009996552937797645,
+      "loss": 3.5394,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.37112854656297195,
+      "learning_rate": 0.0009996271692483596,
+      "loss": 3.4994,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5114308982283607,
+      "learning_rate": 0.0009995979423530893,
+      "loss": 3.4647,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2807433631747716,
+      "learning_rate": 0.000999567613158439,
+      "loss": 3.4074,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3306154771619005,
+      "learning_rate": 0.0009995361817313263,
+      "loss": 3.4999,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2957302402988648,
+      "learning_rate": 0.0009995036481411004,
+      "loss": 3.491,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2698580892627919,
+      "learning_rate": 0.0009994700124595429,
+      "loss": 3.5976,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.27196746273699035,
+      "learning_rate": 0.0009994352747608663,
+      "loss": 3.4213,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.30358956210258486,
+      "learning_rate": 0.0009993994351217151,
+      "loss": 3.5215,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2726926376668444,
+      "learning_rate": 0.000999362493621165,
+      "loss": 3.4668,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.32003603958140714,
+      "learning_rate": 0.0009993244503407226,
+      "loss": 3.4924,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2618368497370899,
+      "learning_rate": 0.0009992853053643258,
+      "loss": 3.4169,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.23056619271604348,
+      "learning_rate": 0.0009992450587783426,
+      "loss": 3.3563,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2963111516603701,
+      "learning_rate": 0.000999203710671572,
+      "loss": 3.3747,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.25090411114949296,
+      "learning_rate": 0.0009991612611352438,
+      "loss": 3.3951,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2725259882426756,
+      "learning_rate": 0.0009991177102630173,
+      "loss": 3.3461,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.293535650775447,
+      "learning_rate": 0.0009990730581509817,
+      "loss": 3.3314,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.27592942808116405,
+      "learning_rate": 0.0009990273048976566,
+      "loss": 3.3422,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.30689045930157793,
+      "learning_rate": 0.0009989804506039905,
+      "loss": 3.3833,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.33436394612317977,
+      "learning_rate": 0.0009989324953733614,
+      "loss": 3.3433,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3314279350266806,
+      "learning_rate": 0.0009988834393115766,
+      "loss": 3.3731,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3227977228279974,
+      "learning_rate": 0.000998833282526872,
+      "loss": 3.3655,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.24104148477876433,
+      "learning_rate": 0.0009987820251299122,
+      "loss": 3.2887,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.269439271196603,
+      "learning_rate": 0.00099872966723379,
+      "loss": 3.3846,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.25200135742878493,
+      "learning_rate": 0.0009986762089540266,
+      "loss": 3.2429,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.26918520504524285,
+      "learning_rate": 0.0009986216504085709,
+      "loss": 3.2299,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4157644024379225,
+      "learning_rate": 0.0009985659917177991,
+      "loss": 3.2479,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.35060698689189057,
+      "learning_rate": 0.0009985092330045155,
+      "loss": 3.3372,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3558036195591651,
+      "learning_rate": 0.0009984513743939508,
+      "loss": 3.3206,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.30656831404857576,
+      "learning_rate": 0.0009983924160137626,
+      "loss": 3.216,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.28631288158763357,
+      "learning_rate": 0.000998332357994035,
+      "loss": 3.1817,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2710260040907228,
+      "learning_rate": 0.0009982712004672786,
+      "loss": 3.2832,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39985552897750976,
+      "learning_rate": 0.0009982089435684295,
+      "loss": 3.3005,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2886244988018138,
+      "learning_rate": 0.0009981455874348499,
+      "loss": 3.1814,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2743220990121642,
+      "learning_rate": 0.0009980811322063269,
+      "loss": 3.1812,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3331957968692638,
+      "learning_rate": 0.0009980155780250728,
+      "loss": 3.2038,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40366690185871634,
+      "learning_rate": 0.0009979489250357243,
+      "loss": 3.2163,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2692444778412937,
+      "learning_rate": 0.0009978811733853431,
+      "loss": 3.2061,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.358578429884789,
+      "learning_rate": 0.0009978123232234147,
+      "loss": 3.1957,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.32314805601099256,
+      "learning_rate": 0.000997742374701848,
+      "loss": 3.1299,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.30095652221922026,
+      "learning_rate": 0.0009976713279749754,
+      "loss": 3.1885,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36225961331034895,
+      "learning_rate": 0.0009975991831995528,
+      "loss": 3.1614,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3126475955532988,
+      "learning_rate": 0.0009975259405347581,
+      "loss": 3.1087,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.32382557990622896,
+      "learning_rate": 0.0009974516001421926,
+      "loss": 3.1152,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3400310829439886,
+      "learning_rate": 0.000997376162185878,
+      "loss": 3.1077,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39457268313456995,
+      "learning_rate": 0.0009972996268322594,
+      "loss": 3.1568,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37721772998908687,
+      "learning_rate": 0.0009972219942502017,
+      "loss": 3.1725,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3066023976813978,
+      "learning_rate": 0.0009971432646109918,
+      "loss": 3.046,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.31722667477351196,
+      "learning_rate": 0.0009970634380883365,
+      "loss": 3.0917,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4024486208572313,
+      "learning_rate": 0.0009969825148583627,
+      "loss": 3.0452,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.35978612321237574,
+      "learning_rate": 0.0009969004950996173,
+      "loss": 3.1304,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2958063445213475,
+      "learning_rate": 0.0009968173789930668,
+      "loss": 3.1333,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3047961892188907,
+      "learning_rate": 0.0009967331667220958,
+      "loss": 3.0437,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2963194238982676,
+      "learning_rate": 0.0009966478584725086,
+      "loss": 3.1806,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.27385528529967657,
+      "learning_rate": 0.0009965614544325263,
+      "loss": 3.0404,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3465255342427384,
+      "learning_rate": 0.000996473954792789,
+      "loss": 3.0744,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2606621343708543,
+      "learning_rate": 0.0009963853597463532,
+      "loss": 2.9269,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3065895457630765,
+      "learning_rate": 0.000996295669488693,
+      "loss": 3.0698,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.30958792017026254,
+      "learning_rate": 0.0009962048842176979,
+      "loss": 3.0043,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3668036995288627,
+      "learning_rate": 0.0009961130041336748,
+      "loss": 3.0222,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2830879398649476,
+      "learning_rate": 0.0009960200294393449,
+      "loss": 2.9509,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.30285431864411605,
+      "learning_rate": 0.0009959259603398453,
+      "loss": 3.0342,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2575727626365476,
+      "learning_rate": 0.0009958307970427275,
+      "loss": 3.0521,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3156884570711579,
+      "learning_rate": 0.0009957345397579572,
+      "loss": 2.9831,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2911513452618453,
+      "learning_rate": 0.0009956371886979138,
+      "loss": 2.9508,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45403156425258584,
+      "learning_rate": 0.00099553874407739,
+      "loss": 3.0999,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2942844063625299,
+      "learning_rate": 0.0009954392061135916,
+      "loss": 2.929,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2947085197043086,
+      "learning_rate": 0.0009953385750261364,
+      "loss": 2.8922,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3752940939833083,
+      "learning_rate": 0.0009952368510370538,
+      "loss": 2.8992,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.35022535809651395,
+      "learning_rate": 0.0009951340343707852,
+      "loss": 2.9463,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2800399676379611,
+      "learning_rate": 0.0009950301252541823,
+      "loss": 2.8644,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43931130971195914,
+      "learning_rate": 0.0009949251239165075,
+      "loss": 2.8918,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2587391122033232,
+      "learning_rate": 0.000994819030589433,
+      "loss": 2.7714,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2912233214164248,
+      "learning_rate": 0.00099471184550704,
+      "loss": 2.8757,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.26369976668205913,
+      "learning_rate": 0.0009946035689058189,
+      "loss": 2.8366,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3119861807026255,
+      "learning_rate": 0.0009944942010246681,
+      "loss": 2.8904,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3135685663334647,
+      "learning_rate": 0.0009943837421048942,
+      "loss": 2.9331,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2741790055246914,
+      "learning_rate": 0.0009942721923902106,
+      "loss": 2.897,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.30069323259587705,
+      "learning_rate": 0.0009941595521267377,
+      "loss": 2.9062,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2752725400966257,
+      "learning_rate": 0.0009940458215630017,
+      "loss": 2.8521,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.29833317210850385,
+      "learning_rate": 0.0009939310009499348,
+      "loss": 2.8392,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3010990853723632,
+      "learning_rate": 0.000993815090540874,
+      "loss": 2.886,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.28126011575543086,
+      "learning_rate": 0.000993698090591561,
+      "loss": 2.8285,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4910568871779766,
+      "learning_rate": 0.0009935800013601416,
+      "loss": 2.7636,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8706310528189558,
+      "learning_rate": 0.000993460823107164,
+      "loss": 2.8934,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5705255710078868,
+      "learning_rate": 0.0009933405560955803,
+      "loss": 2.9715,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.33145568768393224,
+      "learning_rate": 0.0009932192005907446,
+      "loss": 2.8063,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39377750214390944,
+      "learning_rate": 0.0009930967568604118,
+      "loss": 2.8612,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.34542992690272417,
+      "learning_rate": 0.000992973225174739,
+      "loss": 2.8684,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.29692469665994997,
+      "learning_rate": 0.0009928486058062827,
+      "loss": 2.8374,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3407979421259312,
+      "learning_rate": 0.0009927228990299999,
+      "loss": 2.8532,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.29920766642270125,
+      "learning_rate": 0.0009925961051232468,
+      "loss": 2.8698,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3582698424662068,
+      "learning_rate": 0.000992468224365778,
+      "loss": 2.871,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3545106726865571,
+      "learning_rate": 0.000992339257039746,
+      "loss": 2.8401,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.34150027994337356,
+      "learning_rate": 0.0009922092034297006,
+      "loss": 2.8161,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.29597427379567576,
+      "learning_rate": 0.0009920780638225891,
+      "loss": 2.7254,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.279054159682378,
+      "learning_rate": 0.0009919458385077538,
+      "loss": 2.85,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2998415316645372,
+      "learning_rate": 0.0009918125277769336,
+      "loss": 2.7565,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2917424714923249,
+      "learning_rate": 0.0009916781319242614,
+      "loss": 2.9422,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.29413397288366033,
+      "learning_rate": 0.0009915426512462646,
+      "loss": 2.7579,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3322563981558645,
+      "learning_rate": 0.0009914060860418644,
+      "loss": 2.7233,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.27603662628058084,
+      "learning_rate": 0.000991268436612374,
+      "loss": 2.7266,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.29898026977242703,
+      "learning_rate": 0.0009911297032614997,
+      "loss": 2.7981,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2877462442144209,
+      "learning_rate": 0.000990989886295339,
+      "loss": 2.916,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2939298284789785,
+      "learning_rate": 0.0009908489860223804,
+      "loss": 2.7377,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3036147319081863,
+      "learning_rate": 0.000990707002753502,
+      "loss": 2.8093,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.31942314453425996,
+      "learning_rate": 0.0009905639368019724,
+      "loss": 2.6946,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2874918418586274,
+      "learning_rate": 0.0009904197884834482,
+      "loss": 2.7245,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2646050865471292,
+      "learning_rate": 0.0009902745581159742,
+      "loss": 2.7087,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.36966099141889186,
+      "learning_rate": 0.0009901282460199829,
+      "loss": 2.7819,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.29404822841447537,
+      "learning_rate": 0.0009899808525182935,
+      "loss": 2.7658,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.30171022116516655,
+      "learning_rate": 0.0009898323779361107,
+      "loss": 2.6611,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3043178613531503,
+      "learning_rate": 0.000989682822601025,
+      "loss": 2.7405,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2937650446577514,
+      "learning_rate": 0.0009895321868430113,
+      "loss": 2.6451,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3166616949683505,
+      "learning_rate": 0.0009893804709944281,
+      "loss": 2.8168,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.25831943272390334,
+      "learning_rate": 0.0009892276753900174,
+      "loss": 2.76,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.26178197011596577,
+      "learning_rate": 0.0009890738003669028,
+      "loss": 2.6926,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44255648226971,
+      "learning_rate": 0.0009889188462645904,
+      "loss": 2.7413,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.30467274112881104,
+      "learning_rate": 0.0009887628134249667,
+      "loss": 2.6813,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.328070305717662,
+      "learning_rate": 0.0009886057021922983,
+      "loss": 2.743,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.27721448591162473,
+      "learning_rate": 0.0009884475129132311,
+      "loss": 2.7179,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6511726190385966,
+      "learning_rate": 0.0009882882459367897,
+      "loss": 2.6842,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4058805133439206,
+      "learning_rate": 0.0009881279016143766,
+      "loss": 2.7386,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3659065615282764,
+      "learning_rate": 0.0009879664802997707,
+      "loss": 2.7264,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3133910835012604,
+      "learning_rate": 0.000987803982349128,
+      "loss": 2.7252,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.27110262432610865,
+      "learning_rate": 0.0009876404081209796,
+      "loss": 2.689,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.27239237136382227,
+      "learning_rate": 0.000987475757976231,
+      "loss": 2.6658,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5423522825845379,
+      "learning_rate": 0.000987310032278162,
+      "loss": 2.7756,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.32554655473300864,
+      "learning_rate": 0.0009871432313924254,
+      "loss": 2.7887,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.25846401899226173,
+      "learning_rate": 0.000986975355687046,
+      "loss": 2.698,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3019388953834801,
+      "learning_rate": 0.0009868064055324204,
+      "loss": 2.7292,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.26745258430936475,
+      "learning_rate": 0.0009866363813013153,
+      "loss": 2.6795,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.24968530117239846,
+      "learning_rate": 0.0009864652833688676,
+      "loss": 2.7136,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.28143839914604074,
+      "learning_rate": 0.0009862931121125836,
+      "loss": 2.761,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.29440843646184306,
+      "learning_rate": 0.000986119867912337,
+      "loss": 2.5947,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36495001194544974,
+      "learning_rate": 0.000985945551150369,
+      "loss": 2.5828,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2943033342906207,
+      "learning_rate": 0.0009857701622112876,
+      "loss": 2.7798,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4448223001789178,
+      "learning_rate": 0.000985593701482066,
+      "loss": 2.67,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3044824976896804,
+      "learning_rate": 0.0009854161693520424,
+      "loss": 2.7579,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3992527378125072,
+      "learning_rate": 0.0009852375662129194,
+      "loss": 2.5987,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.26628358368490856,
+      "learning_rate": 0.0009850578924587613,
+      "loss": 2.7062,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2694631997998857,
+      "learning_rate": 0.000984877148485996,
+      "loss": 2.7057,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.31222163013070225,
+      "learning_rate": 0.000984695334693412,
+      "loss": 2.6277,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2582876938246746,
+      "learning_rate": 0.000984512451482158,
+      "loss": 2.6598,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2980822516680601,
+      "learning_rate": 0.0009843284992557431,
+      "loss": 2.5825,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2629171383394584,
+      "learning_rate": 0.000984143478420034,
+      "loss": 2.6298,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4153204621263423,
+      "learning_rate": 0.0009839573893832563,
+      "loss": 2.6654,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2661578110933496,
+      "learning_rate": 0.000983770232555991,
+      "loss": 2.6443,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.32868480742901773,
+      "learning_rate": 0.0009835820083511765,
+      "loss": 2.6721,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.25464574601357637,
+      "learning_rate": 0.0009833927171841055,
+      "loss": 2.6175,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.32013898257585327,
+      "learning_rate": 0.0009832023594724246,
+      "loss": 2.5438,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2640855958581696,
+      "learning_rate": 0.0009830109356361344,
+      "loss": 2.6358,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2409276372971484,
+      "learning_rate": 0.0009828184460975867,
+      "loss": 2.615,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.30073813431851604,
+      "learning_rate": 0.0009826248912814855,
+      "loss": 2.6969,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.23932315711083457,
+      "learning_rate": 0.0009824302716148847,
+      "loss": 2.6994,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.28363131908120987,
+      "learning_rate": 0.0009822345875271884,
+      "loss": 2.6548,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2843183963888889,
+      "learning_rate": 0.0009820378394501481,
+      "loss": 2.7219,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.23316552112670008,
+      "learning_rate": 0.0009818400278178636,
+      "loss": 2.595,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2503998689419003,
+      "learning_rate": 0.0009816411530667814,
+      "loss": 2.6582,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.27546854396037773,
+      "learning_rate": 0.000981441215635693,
+      "loss": 2.5897,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3218138457570169,
+      "learning_rate": 0.0009812402159657353,
+      "loss": 2.5597,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.22877170074503378,
+      "learning_rate": 0.000981038154500388,
+      "loss": 2.5368,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2857201796053759,
+      "learning_rate": 0.0009808350316854746,
+      "loss": 2.5967,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2626764135266731,
+      "learning_rate": 0.0009806308479691594,
+      "loss": 2.573,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.24004132426734054,
+      "learning_rate": 0.0009804256038019482,
+      "loss": 2.6474,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2838742359912671,
+      "learning_rate": 0.0009802192996366857,
+      "loss": 2.5998,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3339200315347223,
+      "learning_rate": 0.0009800119359285563,
+      "loss": 2.6048,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.31495154258745606,
+      "learning_rate": 0.0009798035131350813,
+      "loss": 2.5904,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25454722173721295,
+      "learning_rate": 0.0009795940317161194,
+      "loss": 2.6457,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2780311528635556,
+      "learning_rate": 0.0009793834921338646,
+      "loss": 2.5425,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.26011588384390677,
+      "learning_rate": 0.0009791718948528457,
+      "loss": 2.6891,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3402631377661696,
+      "learning_rate": 0.0009789592403399252,
+      "loss": 2.6052,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.31514351232795046,
+      "learning_rate": 0.0009787455290642985,
+      "loss": 2.5763,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.33398676237416175,
+      "learning_rate": 0.000978530761497492,
+      "loss": 2.7201,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.296948613783211,
+      "learning_rate": 0.0009783149381133633,
+      "loss": 2.644,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.498072821470756,
+      "learning_rate": 0.0009780980593880992,
+      "loss": 2.603,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2672412775974075,
+      "learning_rate": 0.0009778801258002153,
+      "loss": 2.5773,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2609394048230652,
+      "learning_rate": 0.000977661137830554,
+      "loss": 2.6203,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3066173778930956,
+      "learning_rate": 0.0009774410959622845,
+      "loss": 2.5656,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.24133568694980762,
+      "learning_rate": 0.000977220000680901,
+      "loss": 2.507,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2661567063627454,
+      "learning_rate": 0.000976997852474223,
+      "loss": 2.5056,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7894568711705849,
+      "learning_rate": 0.0009767746518323914,
+      "loss": 2.6361,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.32332952652784674,
+      "learning_rate": 0.0009765503992478704,
+      "loss": 2.5988,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3731253506008722,
+      "learning_rate": 0.0009763250952154449,
+      "loss": 2.648,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.32521929126962584,
+      "learning_rate": 0.0009760987402322195,
+      "loss": 2.5739,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.24380197457646935,
+      "learning_rate": 0.0009758713347976178,
+      "loss": 2.5778,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.28224805722631924,
+      "learning_rate": 0.000975642879413381,
+      "loss": 2.6387,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4667848243338128,
+      "learning_rate": 0.0009754133745835665,
+      "loss": 2.6301,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3344104172851605,
+      "learning_rate": 0.0009751828208145482,
+      "loss": 2.6079,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2990515640519381,
+      "learning_rate": 0.0009749512186150131,
+      "loss": 2.6515,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.22370886225609918,
+      "learning_rate": 0.0009747185684959625,
+      "loss": 2.5512,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.28276566937554515,
+      "learning_rate": 0.000974484870970709,
+      "loss": 2.5587,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2501305986971202,
+      "learning_rate": 0.0009742501265548767,
+      "loss": 2.5496,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.25615455084945,
+      "learning_rate": 0.0009740143357663993,
+      "loss": 2.5386,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2663748719165565,
+      "learning_rate": 0.000973777499125519,
+      "loss": 2.5585,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.24325142243799,
+      "learning_rate": 0.0009735396171547859,
+      "loss": 2.5945,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2752485611807798,
+      "learning_rate": 0.0009733006903790564,
+      "loss": 2.6296,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23736678798362604,
+      "learning_rate": 0.0009730607193254922,
+      "loss": 2.6021,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.27847273881659146,
+      "learning_rate": 0.0009728197045235585,
+      "loss": 2.5228,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23264877118039437,
+      "learning_rate": 0.0009725776465050242,
+      "loss": 2.5551,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.21831592615779596,
+      "learning_rate": 0.0009723345458039594,
+      "loss": 2.5509,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23777138921871704,
+      "learning_rate": 0.000972090402956735,
+      "loss": 2.5391,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.24381865262282584,
+      "learning_rate": 0.0009718452185020212,
+      "loss": 2.5143,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.25072048479148346,
+      "learning_rate": 0.0009715989929807862,
+      "loss": 2.5151,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2784732534240599,
+      "learning_rate": 0.0009713517269362955,
+      "loss": 2.5409,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22926021538286676,
+      "learning_rate": 0.0009711034209141101,
+      "loss": 2.4994,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2436718288733235,
+      "learning_rate": 0.0009708540754620856,
+      "loss": 2.5662,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22506426040064556,
+      "learning_rate": 0.0009706036911303713,
+      "loss": 2.5866,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24942195637059375,
+      "learning_rate": 0.0009703522684714083,
+      "loss": 2.5769,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24500114802222245,
+      "learning_rate": 0.0009700998080399286,
+      "loss": 2.5079,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24448219667617863,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 2.4727,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2771818606173995,
+      "learning_rate": 0.0009695917760897954,
+      "loss": 2.5461,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22524025429937,
+      "learning_rate": 0.0009693362056920501,
+      "loss": 2.4659,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2827412801107414,
+      "learning_rate": 0.0009690795997636015,
+      "loss": 2.5577,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.26121061135901197,
+      "learning_rate": 0.0009688219588706179,
+      "loss": 2.6101,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.30169848264370014,
+      "learning_rate": 0.0009685632835815518,
+      "loss": 2.4885,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.26432657968165396,
+      "learning_rate": 0.0009683035744671367,
+      "loss": 2.5344,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.27123429154544226,
+      "learning_rate": 0.0009680428321003883,
+      "loss": 2.5453,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2408096412142215,
+      "learning_rate": 0.000967781057056601,
+      "loss": 2.4644,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22876028072376958,
+      "learning_rate": 0.0009675182499133485,
+      "loss": 2.5372,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2373888391940621,
+      "learning_rate": 0.0009672544112504813,
+      "loss": 2.5068,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2176608114129179,
+      "learning_rate": 0.0009669895416501257,
+      "loss": 2.4595,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2425916154692314,
+      "learning_rate": 0.0009667236416966833,
+      "loss": 2.5567,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.330554324751859,
+      "learning_rate": 0.0009664567119768281,
+      "loss": 2.5671,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2749751358238836,
+      "learning_rate": 0.0009661887530795067,
+      "loss": 2.4215,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24540581436777884,
+      "learning_rate": 0.0009659197655959365,
+      "loss": 2.3667,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2456429599622594,
+      "learning_rate": 0.000965649750119604,
+      "loss": 2.5299,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40511840050867626,
+      "learning_rate": 0.0009653787072462643,
+      "loss": 2.5858,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2915964636637138,
+      "learning_rate": 0.0009651066375739388,
+      "loss": 2.5886,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23995172980788054,
+      "learning_rate": 0.000964833541702915,
+      "loss": 2.4773,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2290573949683804,
+      "learning_rate": 0.0009645594202357438,
+      "loss": 2.4679,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2217290283106237,
+      "learning_rate": 0.0009642842737772397,
+      "loss": 2.5029,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.25217557779155975,
+      "learning_rate": 0.0009640081029344782,
+      "loss": 2.5546,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.21599048579024788,
+      "learning_rate": 0.0009637309083167956,
+      "loss": 2.4589,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23779490242619933,
+      "learning_rate": 0.0009634526905357859,
+      "loss": 2.6191,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2241675800026173,
+      "learning_rate": 0.000963173450205302,
+      "loss": 2.5681,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2221024752733853,
+      "learning_rate": 0.0009628931879414517,
+      "loss": 2.5149,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.22522388450484887,
+      "learning_rate": 0.0009626119043625983,
+      "loss": 2.577,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2639253545944138,
+      "learning_rate": 0.0009623296000893582,
+      "loss": 2.5457,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.25411540848854947,
+      "learning_rate": 0.0009620462757446,
+      "loss": 2.5138,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2683841375615613,
+      "learning_rate": 0.0009617619319534428,
+      "loss": 2.575,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0199077704410708,
+      "learning_rate": 0.000961476569343255,
+      "loss": 2.5656,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3561368416425003,
+      "learning_rate": 0.0009611901885436529,
+      "loss": 2.6289,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4632827687137865,
+      "learning_rate": 0.0009609027901864996,
+      "loss": 2.6316,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3946513372694328,
+      "learning_rate": 0.0009606143749059029,
+      "loss": 2.5917,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.38841030468445586,
+      "learning_rate": 0.0009603249433382144,
+      "loss": 2.4967,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2531117025642261,
+      "learning_rate": 0.0009600344961220282,
+      "loss": 2.5499,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.26699663949518543,
+      "learning_rate": 0.0009597430338981791,
+      "loss": 2.5438,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3115428999287217,
+      "learning_rate": 0.0009594505573097414,
+      "loss": 2.5238,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.31981913987689775,
+      "learning_rate": 0.0009591570670020277,
+      "loss": 2.6135,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30483462481777157,
+      "learning_rate": 0.0009588625636225871,
+      "loss": 2.5353,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.497471969678511,
+      "learning_rate": 0.0009585670478212036,
+      "loss": 2.5704,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2741563196984564,
+      "learning_rate": 0.0009582705202498956,
+      "loss": 2.5753,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4041368975316352,
+      "learning_rate": 0.0009579729815629133,
+      "loss": 2.4716,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2323256663029621,
+      "learning_rate": 0.0009576744324167379,
+      "loss": 2.6081,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3095996276768797,
+      "learning_rate": 0.0009573748734700804,
+      "loss": 2.5574,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2347644657122629,
+      "learning_rate": 0.0009570743053838796,
+      "loss": 2.6287,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.230129316090552,
+      "learning_rate": 0.0009567727288213005,
+      "loss": 2.5612,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.312513382655126,
+      "learning_rate": 0.0009564701444477337,
+      "loss": 2.5412,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.29306169909258206,
+      "learning_rate": 0.000956166552930793,
+      "loss": 2.5825,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41582562358594316,
+      "learning_rate": 0.0009558619549403147,
+      "loss": 2.5352,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.24212448365671677,
+      "learning_rate": 0.0009555563511483555,
+      "loss": 2.5216,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.22614785600091616,
+      "learning_rate": 0.0009552497422291912,
+      "loss": 2.5255,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.199800912368347,
+      "learning_rate": 0.0009549421288593157,
+      "loss": 2.5728,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.35485161703560264,
+      "learning_rate": 0.0009546335117174385,
+      "loss": 2.6037,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.24453258748195944,
+      "learning_rate": 0.0009543238914844843,
+      "loss": 2.5388,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.403061657350811,
+      "learning_rate": 0.0009540132688435907,
+      "loss": 2.4898,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2587874292120635,
+      "learning_rate": 0.0009537016444801074,
+      "loss": 2.5749,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2961475540504943,
+      "learning_rate": 0.0009533890190815935,
+      "loss": 2.5807,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.22191430055017883,
+      "learning_rate": 0.0009530753933378173,
+      "loss": 2.5329,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24726715187349263,
+      "learning_rate": 0.0009527607679407545,
+      "loss": 2.5156,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2626248686156067,
+      "learning_rate": 0.0009524451435845857,
+      "loss": 2.5396,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2695892022178038,
+      "learning_rate": 0.0009521285209656963,
+      "loss": 2.532,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.28467743945309887,
+      "learning_rate": 0.0009518109007826734,
+      "loss": 2.4994,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24822240456160663,
+      "learning_rate": 0.0009514922837363059,
+      "loss": 2.5485,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.26775712893221276,
+      "learning_rate": 0.0009511726705295817,
+      "loss": 2.516,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.23630246901871552,
+      "learning_rate": 0.000950852061867687,
+      "loss": 2.5466,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2084717698182876,
+      "learning_rate": 0.0009505304584580038,
+      "loss": 2.4778,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2282390971123192,
+      "learning_rate": 0.0009502078610101092,
+      "loss": 2.559,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.22511098057298112,
+      "learning_rate": 0.0009498842702357736,
+      "loss": 2.5442,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.22249997905472296,
+      "learning_rate": 0.0009495596868489587,
+      "loss": 2.4418,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.22681400521284273,
+      "learning_rate": 0.0009492341115658167,
+      "loss": 2.4528,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24216937085357818,
+      "learning_rate": 0.0009489075451046879,
+      "loss": 2.4743,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.20652486824076494,
+      "learning_rate": 0.0009485799881861,
+      "loss": 2.472,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.23630893196470576,
+      "learning_rate": 0.0009482514415327654,
+      "loss": 2.655,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.23962543892445873,
+      "learning_rate": 0.000947921905869581,
+      "loss": 2.4349,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2030928997837339,
+      "learning_rate": 0.0009475913819236248,
+      "loss": 2.4466,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24825214645690102,
+      "learning_rate": 0.0009472598704241561,
+      "loss": 2.5037,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.19707397801968132,
+      "learning_rate": 0.0009469273721026131,
+      "loss": 2.4901,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2338732110706325,
+      "learning_rate": 0.0009465938876926111,
+      "loss": 2.4779,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24189949821344117,
+      "learning_rate": 0.0009462594179299406,
+      "loss": 2.423,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.23190175642179273,
+      "learning_rate": 0.0009459239635525672,
+      "loss": 2.4773,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2325892481960379,
+      "learning_rate": 0.0009455875253006281,
+      "loss": 2.5214,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3465889910349687,
+      "learning_rate": 0.0009452501039164315,
+      "loss": 2.5612,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.22844933911344173,
+      "learning_rate": 0.0009449117001444549,
+      "loss": 2.5201,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.21993552132703634,
+      "learning_rate": 0.0009445723147313433,
+      "loss": 2.4697,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.23502112840926154,
+      "learning_rate": 0.0009442319484259074,
+      "loss": 2.3488,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.252153402717478,
+      "learning_rate": 0.0009438906019791222,
+      "loss": 2.4759,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.21018655238430012,
+      "learning_rate": 0.0009435482761441251,
+      "loss": 2.5721,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.26808481229498776,
+      "learning_rate": 0.000943204971676215,
+      "loss": 2.4398,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.21779778696655663,
+      "learning_rate": 0.0009428606893328493,
+      "loss": 2.4588,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.25508339975782446,
+      "learning_rate": 0.0009425154298736432,
+      "loss": 2.5685,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.22490668125183028,
+      "learning_rate": 0.0009421691940603678,
+      "loss": 2.5775,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2249564364855568,
+      "learning_rate": 0.0009418219826569488,
+      "loss": 2.3819,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.21652201172433794,
+      "learning_rate": 0.0009414737964294635,
+      "loss": 2.5289,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.23087123620886738,
+      "learning_rate": 0.000941124636146141,
+      "loss": 2.4889,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.23545498235108664,
+      "learning_rate": 0.0009407745025773589,
+      "loss": 2.4443,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2170195249154736,
+      "learning_rate": 0.0009404233964956423,
+      "loss": 2.4976,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.259659306426799,
+      "learning_rate": 0.0009400713186756625,
+      "loss": 2.4226,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.24418664020215997,
+      "learning_rate": 0.0009397182698942342,
+      "loss": 2.5256,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.24907251876402392,
+      "learning_rate": 0.0009393642509303149,
+      "loss": 2.5905,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2324191400034502,
+      "learning_rate": 0.0009390092625650023,
+      "loss": 2.4809,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2479198984124518,
+      "learning_rate": 0.0009386533055815332,
+      "loss": 2.4708,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22009000979645943,
+      "learning_rate": 0.0009382963807652813,
+      "loss": 2.417,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22820255895293776,
+      "learning_rate": 0.000937938488903756,
+      "loss": 2.5519,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.24983542487514318,
+      "learning_rate": 0.0009375796307866003,
+      "loss": 2.4596,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.20968411385161706,
+      "learning_rate": 0.0009372198072055888,
+      "loss": 2.3977,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.23522003095493732,
+      "learning_rate": 0.0009368590189546268,
+      "loss": 2.5229,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21867102622388068,
+      "learning_rate": 0.0009364972668297474,
+      "loss": 2.5352,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2427595149806782,
+      "learning_rate": 0.0009361345516291111,
+      "loss": 2.457,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.24624307699752854,
+      "learning_rate": 0.0009357708741530024,
+      "loss": 2.4999,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.20411934583397426,
+      "learning_rate": 0.00093540623520383,
+      "loss": 2.3988,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.20798247995767968,
+      "learning_rate": 0.000935040635586123,
+      "loss": 2.469,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22308432548454912,
+      "learning_rate": 0.0009346740761065305,
+      "loss": 2.4577,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.26893936224953247,
+      "learning_rate": 0.0009343065575738197,
+      "loss": 2.4627,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.23181250257456554,
+      "learning_rate": 0.0009339380807988733,
+      "loss": 2.4481,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4710626042800092,
+      "learning_rate": 0.0009335686465946887,
+      "loss": 2.6139,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.29654075658920276,
+      "learning_rate": 0.0009331982557763754,
+      "loss": 2.3753,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2916196648685602,
+      "learning_rate": 0.0009328269091611537,
+      "loss": 2.5643,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.25237095366661694,
+      "learning_rate": 0.0009324546075683524,
+      "loss": 2.4877,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22584476966297573,
+      "learning_rate": 0.0009320813518194083,
+      "loss": 2.3215,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34199853274773573,
+      "learning_rate": 0.0009317071427378624,
+      "loss": 2.4135,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2366242724909103,
+      "learning_rate": 0.0009313319811493594,
+      "loss": 2.5269,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2064178580144876,
+      "learning_rate": 0.000930955867881646,
+      "loss": 2.3445,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2256422223670338,
+      "learning_rate": 0.0009305788037645681,
+      "loss": 2.5281,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22461738207238655,
+      "learning_rate": 0.0009302007896300697,
+      "loss": 2.4717,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.24049128500155229,
+      "learning_rate": 0.0009298218263121911,
+      "loss": 2.422,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23692615481512005,
+      "learning_rate": 0.0009294419146470668,
+      "loss": 2.5149,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.20675669064975458,
+      "learning_rate": 0.0009290610554729234,
+      "loss": 2.4264,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.376023309762259,
+      "learning_rate": 0.0009286792496300784,
+      "loss": 2.5531,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.20928554921892148,
+      "learning_rate": 0.0009282964979609379,
+      "loss": 2.5273,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2056501954310486,
+      "learning_rate": 0.0009279128013099947,
+      "loss": 2.4772,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23481073302102604,
+      "learning_rate": 0.0009275281605238268,
+      "loss": 2.4741,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.21483835439476132,
+      "learning_rate": 0.0009271425764510953,
+      "loss": 2.4483,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22700853412110927,
+      "learning_rate": 0.0009267560499425423,
+      "loss": 2.4469,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22187936941292152,
+      "learning_rate": 0.0009263685818509895,
+      "loss": 2.4396,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.19866276989777332,
+      "learning_rate": 0.000925980173031336,
+      "loss": 2.4085,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2420081498456651,
+      "learning_rate": 0.0009255908243405567,
+      "loss": 2.4206,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.32196214725802497,
+      "learning_rate": 0.0009252005366376996,
+      "loss": 2.5171,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.656139937938564,
+      "learning_rate": 0.0009248093107838852,
+      "loss": 2.3876,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2292508626104499,
+      "learning_rate": 0.0009244171476423036,
+      "loss": 2.4913,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2514520532393082,
+      "learning_rate": 0.0009240240480782129,
+      "loss": 2.3984,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.24722039656627623,
+      "learning_rate": 0.0009236300129589376,
+      "loss": 2.4797,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23217974052113197,
+      "learning_rate": 0.0009232350431538657,
+      "loss": 2.4446,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2465976110275916,
+      "learning_rate": 0.0009228391395344482,
+      "loss": 2.4833,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2964307924613834,
+      "learning_rate": 0.000922442302974196,
+      "loss": 2.4422,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2958105133486997,
+      "learning_rate": 0.0009220445343486785,
+      "loss": 2.5096,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20742642232919467,
+      "learning_rate": 0.0009216458345355217,
+      "loss": 2.476,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2939197760413958,
+      "learning_rate": 0.0009212462044144061,
+      "loss": 2.3075,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22831308952967008,
+      "learning_rate": 0.0009208456448670648,
+      "loss": 2.3645,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2046184427803563,
+      "learning_rate": 0.0009204441567772816,
+      "loss": 2.5325,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.218976544652746,
+      "learning_rate": 0.0009200417410308888,
+      "loss": 2.4718,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4169027678599138,
+      "learning_rate": 0.0009196383985157656,
+      "loss": 2.4338,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22471421905704042,
+      "learning_rate": 0.000919234130121836,
+      "loss": 2.4738,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22223962907276715,
+      "learning_rate": 0.0009188289367410672,
+      "loss": 2.3882,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24389398730912437,
+      "learning_rate": 0.0009184228192674666,
+      "loss": 2.3694,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21451366449639808,
+      "learning_rate": 0.0009180157785970808,
+      "loss": 2.4673,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21329521493092138,
+      "learning_rate": 0.0009176078156279932,
+      "loss": 2.4373,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21841560134213592,
+      "learning_rate": 0.0009171989312603226,
+      "loss": 2.4582,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21433374431620714,
+      "learning_rate": 0.0009167891263962202,
+      "loss": 2.3777,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21076283005751678,
+      "learning_rate": 0.0009163784019398685,
+      "loss": 2.5632,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2317826379608557,
+      "learning_rate": 0.0009159667587974785,
+      "loss": 2.3418,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21140652198332757,
+      "learning_rate": 0.0009155541978772887,
+      "loss": 2.3636,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21700199296556189,
+      "learning_rate": 0.0009151407200895625,
+      "loss": 2.4332,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21563581553204864,
+      "learning_rate": 0.000914726326346586,
+      "loss": 2.4558,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.27401473087278994,
+      "learning_rate": 0.0009143110175626661,
+      "loss": 2.4574,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.23359146797667518,
+      "learning_rate": 0.0009138947946541291,
+      "loss": 2.4475,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20337007677553431,
+      "learning_rate": 0.0009134776585393181,
+      "loss": 2.3648,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.20548678898069075,
+      "learning_rate": 0.0009130596101385906,
+      "loss": 2.5058,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1983072404050006,
+      "learning_rate": 0.0009126406503743174,
+      "loss": 2.4441,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.19573282475927392,
+      "learning_rate": 0.0009122207801708802,
+      "loss": 2.3784,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.20887133066575367,
+      "learning_rate": 0.0009118000004546689,
+      "loss": 2.3518,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.20451510673611523,
+      "learning_rate": 0.0009113783121540807,
+      "loss": 2.4296,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.23003735727880745,
+      "learning_rate": 0.0009109557161995172,
+      "loss": 2.5437,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.22329661643155108,
+      "learning_rate": 0.0009105322135233828,
+      "loss": 2.4263,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2311749107313663,
+      "learning_rate": 0.0009101078050600821,
+      "loss": 2.4799,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.21077159822014843,
+      "learning_rate": 0.0009096824917460186,
+      "loss": 2.3934,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.23273551705487114,
+      "learning_rate": 0.0009092562745195921,
+      "loss": 2.4765,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.20348622126366023,
+      "learning_rate": 0.0009088291543211967,
+      "loss": 2.3589,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.29917308023971134,
+      "learning_rate": 0.0009084011320932188,
+      "loss": 2.3977,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2240233328471604,
+      "learning_rate": 0.0009079722087800352,
+      "loss": 2.3624,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.264946541041109,
+      "learning_rate": 0.0009075423853280106,
+      "loss": 2.4174,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2067416289958836,
+      "learning_rate": 0.0009071116626854958,
+      "loss": 2.4683,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.21804443835595105,
+      "learning_rate": 0.0009066800418028256,
+      "loss": 2.3821,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.20887622182622434,
+      "learning_rate": 0.0009062475236323168,
+      "loss": 2.4204,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.19760867706820356,
+      "learning_rate": 0.0009058141091282656,
+      "loss": 2.3633,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2596812147923856,
+      "learning_rate": 0.0009053797992469461,
+      "loss": 2.5797,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.30979846882101825,
+      "learning_rate": 0.0009049445949466078,
+      "loss": 2.438,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.21976708697845465,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 2.3821,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.26724957221663764,
+      "learning_rate": 0.0009040715069317382,
+      "loss": 2.349,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20852210950888997,
+      "learning_rate": 0.0009036336251435648,
+      "loss": 2.4259,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20953732860842395,
+      "learning_rate": 0.0009031948527890839,
+      "loss": 2.4767,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.21564146223863298,
+      "learning_rate": 0.000902755190836391,
+      "loss": 2.3925,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.22303540610094452,
+      "learning_rate": 0.0009023146402555442,
+      "loss": 2.5553,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3167420975414526,
+      "learning_rate": 0.0009018732020185624,
+      "loss": 2.4242,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.21488045658744837,
+      "learning_rate": 0.0009014308770994235,
+      "loss": 2.339,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.23487915201224632,
+      "learning_rate": 0.0009009876664740605,
+      "loss": 2.4191,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2478952830933866,
+      "learning_rate": 0.0009005435711203618,
+      "loss": 2.4027,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.21374886931597525,
+      "learning_rate": 0.000900098592018167,
+      "loss": 2.3841,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3361911458848779,
+      "learning_rate": 0.0008996527301492663,
+      "loss": 2.4415,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.24769890503372907,
+      "learning_rate": 0.0008992059864973972,
+      "loss": 2.3923,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20914863750413765,
+      "learning_rate": 0.0008987583620482427,
+      "loss": 2.3953,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.21183630108056073,
+      "learning_rate": 0.0008983098577894292,
+      "loss": 2.4063,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.23564931821882754,
+      "learning_rate": 0.0008978604747105246,
+      "loss": 2.3431,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.24860652518481535,
+      "learning_rate": 0.0008974102138030354,
+      "loss": 2.4204,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2550527418934057,
+      "learning_rate": 0.000896959076060405,
+      "loss": 2.457,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.22306945023809946,
+      "learning_rate": 0.0008965070624780116,
+      "loss": 2.3422,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20084927846012107,
+      "learning_rate": 0.0008960541740531658,
+      "loss": 2.4928,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.19632627513684062,
+      "learning_rate": 0.0008956004117851083,
+      "loss": 2.4219,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2165555879303392,
+      "learning_rate": 0.0008951457766750079,
+      "loss": 2.4041,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2176244634978172,
+      "learning_rate": 0.0008946902697259593,
+      "loss": 2.3943,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3272669909320465,
+      "learning_rate": 0.0008942338919429805,
+      "loss": 2.4134,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20099338364821506,
+      "learning_rate": 0.0008937766443330113,
+      "loss": 2.3729,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23262943610630704,
+      "learning_rate": 0.0008933185279049103,
+      "loss": 2.4557,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.22496695506197298,
+      "learning_rate": 0.0008928595436694532,
+      "loss": 2.4638,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20141524594427854,
+      "learning_rate": 0.0008923996926393305,
+      "loss": 2.2829,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20598645698167217,
+      "learning_rate": 0.0008919389758291449,
+      "loss": 2.4188,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23622830188364277,
+      "learning_rate": 0.0008914773942554098,
+      "loss": 2.4293,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3483381310896749,
+      "learning_rate": 0.000891014948936546,
+      "loss": 2.5452,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.19296938480701972,
+      "learning_rate": 0.0008905516408928804,
+      "loss": 2.4789,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24609843293317585,
+      "learning_rate": 0.0008900874711466434,
+      "loss": 2.3897,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24864621690391547,
+      "learning_rate": 0.0008896224407219666,
+      "loss": 2.4935,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20484951705170515,
+      "learning_rate": 0.0008891565506448804,
+      "loss": 2.4352,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2669855623951663,
+      "learning_rate": 0.0008886898019433122,
+      "loss": 2.2807,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.25340569076951064,
+      "learning_rate": 0.0008882221956470836,
+      "loss": 2.4292,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23151207906462135,
+      "learning_rate": 0.0008877537327879086,
+      "loss": 2.3664,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23649736754179107,
+      "learning_rate": 0.0008872844143993908,
+      "loss": 2.3583,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.21240447920797598,
+      "learning_rate": 0.0008868142415170218,
+      "loss": 2.4167,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2721591028654908,
+      "learning_rate": 0.0008863432151781781,
+      "loss": 2.5031,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.19096215240533407,
+      "learning_rate": 0.0008858713364221195,
+      "loss": 2.3601,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2046198990848414,
+      "learning_rate": 0.0008853986062899868,
+      "loss": 2.3968,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24970964399913645,
+      "learning_rate": 0.0008849250258247986,
+      "loss": 2.4135,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2057778519880038,
+      "learning_rate": 0.0008844505960714503,
+      "loss": 2.3534,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.21465961945404075,
+      "learning_rate": 0.0008839753180767108,
+      "loss": 2.3701,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.25732585858458407,
+      "learning_rate": 0.0008834991928892204,
+      "loss": 2.4429,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.21365043954618054,
+      "learning_rate": 0.000883022221559489,
+      "loss": 2.4283,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.24680465357320902,
+      "learning_rate": 0.0008825444051398934,
+      "loss": 2.399,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.26093165498943477,
+      "learning_rate": 0.0008820657446846745,
+      "loss": 2.4116,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2355947219046093,
+      "learning_rate": 0.000881586241249936,
+      "loss": 2.4024,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3241075278272244,
+      "learning_rate": 0.0008811058958936411,
+      "loss": 2.4542,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1999214090036047,
+      "learning_rate": 0.000880624709675611,
+      "loss": 2.3668,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2737352499451527,
+      "learning_rate": 0.000880142683657522,
+      "loss": 2.3845,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.26357771066534397,
+      "learning_rate": 0.0008796598189029029,
+      "loss": 2.3453,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2757845839473988,
+      "learning_rate": 0.0008791761164771338,
+      "loss": 2.4009,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2262926643916019,
+      "learning_rate": 0.0008786915774474424,
+      "loss": 2.3193,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2681246344456434,
+      "learning_rate": 0.0008782062028829027,
+      "loss": 2.3441,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20248358326103355,
+      "learning_rate": 0.0008777199938544318,
+      "loss": 2.3992,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20862031535845815,
+      "learning_rate": 0.0008772329514347883,
+      "loss": 2.2813,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2184798913741189,
+      "learning_rate": 0.0008767450766985694,
+      "loss": 2.4004,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.18995957948272044,
+      "learning_rate": 0.0008762563707222086,
+      "loss": 2.4303,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20959668134058398,
+      "learning_rate": 0.0008757668345839738,
+      "loss": 2.468,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.27292510406315823,
+      "learning_rate": 0.0008752764693639638,
+      "loss": 2.4601,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.19840487305807933,
+      "learning_rate": 0.0008747852761441078,
+      "loss": 2.3421,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.20055448448728347,
+      "learning_rate": 0.0008742932560081607,
+      "loss": 2.3359,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23286312814488175,
+      "learning_rate": 0.0008738004100417025,
+      "loss": 2.3448,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1976749331690812,
+      "learning_rate": 0.0008733067393321355,
+      "loss": 2.469,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.25797412100545836,
+      "learning_rate": 0.000872812244968681,
+      "loss": 2.4306,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2894067488982873,
+      "learning_rate": 0.0008723169280423783,
+      "loss": 2.4533,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.20442043723960854,
+      "learning_rate": 0.0008718207896460811,
+      "loss": 2.4385,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24388858399066854,
+      "learning_rate": 0.0008713238308744557,
+      "loss": 2.4717,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21292847247630436,
+      "learning_rate": 0.0008708260528239789,
+      "loss": 2.3366,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.19976489392785904,
+      "learning_rate": 0.000870327456592934,
+      "loss": 2.3744,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21206258135815417,
+      "learning_rate": 0.0008698280432814107,
+      "loss": 2.4004,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.20954470296735697,
+      "learning_rate": 0.000869327813991301,
+      "loss": 2.4174,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21800195919666462,
+      "learning_rate": 0.0008688267698262971,
+      "loss": 2.4124,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.23516885549977773,
+      "learning_rate": 0.0008683249118918894,
+      "loss": 2.3878,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2101725173734163,
+      "learning_rate": 0.0008678222412953637,
+      "loss": 2.4054,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.22889224585043877,
+      "learning_rate": 0.0008673187591457987,
+      "loss": 2.5794,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2187492812026622,
+      "learning_rate": 0.0008668144665540639,
+      "loss": 2.4125,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.193578338408776,
+      "learning_rate": 0.0008663093646328167,
+      "loss": 2.4336,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21541402518268304,
+      "learning_rate": 0.0008658034544965003,
+      "loss": 2.454,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2698734535906141,
+      "learning_rate": 0.0008652967372613412,
+      "loss": 2.4178,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.18621398462622735,
+      "learning_rate": 0.0008647892140453466,
+      "loss": 2.395,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2254863568333006,
+      "learning_rate": 0.0008642808859683021,
+      "loss": 2.3298,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2484311273331186,
+      "learning_rate": 0.0008637717541517689,
+      "loss": 2.4352,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21784843563444936,
+      "learning_rate": 0.0008632618197190816,
+      "loss": 2.4143,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.18494274768227145,
+      "learning_rate": 0.0008627510837953458,
+      "loss": 2.4224,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1956293788016938,
+      "learning_rate": 0.0008622395475074355,
+      "loss": 2.3882,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1975937784405231,
+      "learning_rate": 0.0008617272119839903,
+      "loss": 2.3548,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2174944822781471,
+      "learning_rate": 0.0008612140783554136,
+      "loss": 2.2991,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2094381648115037,
+      "learning_rate": 0.0008607001477538696,
+      "loss": 2.3997,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.21491611068237,
+      "learning_rate": 0.0008601854213132807,
+      "loss": 2.4631,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.24484071899162763,
+      "learning_rate": 0.0008596699001693256,
+      "loss": 2.3438,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.197539248230288,
+      "learning_rate": 0.000859153585459436,
+      "loss": 2.3695,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19839716463046012,
+      "learning_rate": 0.0008586364783227949,
+      "loss": 2.346,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2518523582373015,
+      "learning_rate": 0.0008581185799003332,
+      "loss": 2.3956,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.21542938108873624,
+      "learning_rate": 0.0008575998913347283,
+      "loss": 2.3763,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2568553990776338,
+      "learning_rate": 0.0008570804137704004,
+      "loss": 2.3313,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.20235281636101288,
+      "learning_rate": 0.0008565601483535108,
+      "loss": 2.3154,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2295542764071981,
+      "learning_rate": 0.0008560390962319591,
+      "loss": 2.3038,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.23857225207943458,
+      "learning_rate": 0.0008555172585553804,
+      "loss": 2.4039,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.21646702082850616,
+      "learning_rate": 0.0008549946364751435,
+      "loss": 2.3144,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2626300808580288,
+      "learning_rate": 0.0008544712311443475,
+      "loss": 2.3865,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19649068430260827,
+      "learning_rate": 0.0008539470437178196,
+      "loss": 2.3726,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2352309802570764,
+      "learning_rate": 0.000853422075352113,
+      "loss": 2.3757,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2137881470486359,
+      "learning_rate": 0.0008528963272055035,
+      "loss": 2.3862,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.20794025780059436,
+      "learning_rate": 0.0008523698004379877,
+      "loss": 2.3696,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19989483787457923,
+      "learning_rate": 0.00085184249621128,
+      "loss": 2.4657,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2344962326786526,
+      "learning_rate": 0.0008513144156888101,
+      "loss": 2.4402,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2013813602991416,
+      "learning_rate": 0.0008507855600357207,
+      "loss": 2.4327,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19700384967989573,
+      "learning_rate": 0.0008502559304188644,
+      "loss": 2.397,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2549281385265782,
+      "learning_rate": 0.0008497255280068019,
+      "loss": 2.4236,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.21474110151975728,
+      "learning_rate": 0.0008491943539697986,
+      "loss": 2.2786,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2035123052510544,
+      "learning_rate": 0.0008486624094798226,
+      "loss": 2.344,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2251696375490433,
+      "learning_rate": 0.0008481296957105417,
+      "loss": 2.4343,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2419176314896895,
+      "learning_rate": 0.0008475962138373213,
+      "loss": 2.3148,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2195381469097167,
+      "learning_rate": 0.0008470619650372211,
+      "loss": 2.4232,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.26144703785516543,
+      "learning_rate": 0.0008465269504889934,
+      "loss": 2.3801,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.20116454519465898,
+      "learning_rate": 0.0008459911713730799,
+      "loss": 2.3737,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2088106790770107,
+      "learning_rate": 0.0008454546288716089,
+      "loss": 2.4945,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.20234949831885418,
+      "learning_rate": 0.0008449173241683935,
+      "loss": 2.3904,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19560218025364734,
+      "learning_rate": 0.0008443792584489281,
+      "loss": 2.4444,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18390141461863452,
+      "learning_rate": 0.0008438404329003863,
+      "loss": 2.4053,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18761271991611642,
+      "learning_rate": 0.0008433008487116183,
+      "loss": 2.4177,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19536034728683432,
+      "learning_rate": 0.0008427605070731481,
+      "loss": 2.2697,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.211679348289352,
+      "learning_rate": 0.0008422194091771708,
+      "loss": 2.4451,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19397610797730208,
+      "learning_rate": 0.0008416775562175503,
+      "loss": 2.2843,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.21871429229999056,
+      "learning_rate": 0.000841134949389816,
+      "loss": 2.3792,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.204490633497813,
+      "learning_rate": 0.0008405915898911611,
+      "loss": 2.3491,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1971118255914251,
+      "learning_rate": 0.0008400474789204396,
+      "loss": 2.404,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.17751864394272654,
+      "learning_rate": 0.0008395026176781626,
+      "loss": 2.4039,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.18596073969537139,
+      "learning_rate": 0.0008389570073664976,
+      "loss": 2.3819,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1927749157983514,
+      "learning_rate": 0.0008384106491892642,
+      "loss": 2.3509,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2234001284600126,
+      "learning_rate": 0.0008378635443519327,
+      "loss": 2.3851,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22129679675400996,
+      "learning_rate": 0.0008373156940616199,
+      "loss": 2.3779,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2186865444563567,
+      "learning_rate": 0.0008367670995270882,
+      "loss": 2.4054,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.193874218877803,
+      "learning_rate": 0.0008362177619587416,
+      "loss": 2.4078,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2493696364585381,
+      "learning_rate": 0.0008356676825686238,
+      "loss": 2.3454,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1953469196750395,
+      "learning_rate": 0.0008351168625704147,
+      "loss": 2.3811,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.26927494887071446,
+      "learning_rate": 0.0008345653031794292,
+      "loss": 2.4361,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22632974479884757,
+      "learning_rate": 0.0008340130056126125,
+      "loss": 2.4216,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22427260200946117,
+      "learning_rate": 0.0008334599710885394,
+      "loss": 2.3011,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2248139015545977,
+      "learning_rate": 0.0008329062008274098,
+      "loss": 2.3795,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.20702451567581667,
+      "learning_rate": 0.000832351696051048,
+      "loss": 2.4376,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1981608935191815,
+      "learning_rate": 0.000831796457982898,
+      "loss": 2.3904,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2351700413889612,
+      "learning_rate": 0.0008312404878480222,
+      "loss": 2.4509,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.18917376536905353,
+      "learning_rate": 0.0008306837868730979,
+      "loss": 2.3927,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2010016900452529,
+      "learning_rate": 0.0008301263562864152,
+      "loss": 2.3788,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22107828224727413,
+      "learning_rate": 0.0008295681973178737,
+      "loss": 2.3459,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.19525667513251108,
+      "learning_rate": 0.0008290093111989804,
+      "loss": 2.4176,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1843495976820245,
+      "learning_rate": 0.0008284496991628465,
+      "loss": 2.2873,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2331315436960734,
+      "learning_rate": 0.0008278893624441847,
+      "loss": 2.2469,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.18661483815699137,
+      "learning_rate": 0.000827328302279307,
+      "loss": 2.389,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.219937239375505,
+      "learning_rate": 0.0008267665199061211,
+      "loss": 2.223,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1908610471332063,
+      "learning_rate": 0.0008262040165641288,
+      "loss": 2.3493,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2218041733023859,
+      "learning_rate": 0.0008256407934944219,
+      "loss": 2.3966,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2400988486651035,
+      "learning_rate": 0.0008250768519396807,
+      "loss": 2.4234,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.19696602307904285,
+      "learning_rate": 0.0008245121931441706,
+      "loss": 2.3386,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.19581465458040276,
+      "learning_rate": 0.0008239468183537393,
+      "loss": 2.3839,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18191346808001652,
+      "learning_rate": 0.0008233807288158146,
+      "loss": 2.3373,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18648996069853518,
+      "learning_rate": 0.0008228139257794012,
+      "loss": 2.3567,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2173595679994734,
+      "learning_rate": 0.0008222464104950778,
+      "loss": 2.3188,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.23416536777652644,
+      "learning_rate": 0.000821678184214995,
+      "loss": 2.349,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.23933556124199534,
+      "learning_rate": 0.0008211092481928716,
+      "loss": 2.3022,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.33135031664789705,
+      "learning_rate": 0.0008205396036839927,
+      "loss": 2.4384,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.27816027405540256,
+      "learning_rate": 0.0008199692519452069,
+      "loss": 2.4068,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.241217360103072,
+      "learning_rate": 0.0008193981942349224,
+      "loss": 2.3245,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.22571173283869966,
+      "learning_rate": 0.0008188264318131056,
+      "loss": 2.3532,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.36527475366548495,
+      "learning_rate": 0.0008182539659412776,
+      "loss": 2.2805,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.20824281813904175,
+      "learning_rate": 0.0008176807978825118,
+      "loss": 2.3076,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.21618348875387058,
+      "learning_rate": 0.0008171069289014306,
+      "loss": 2.4162,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3277306074095195,
+      "learning_rate": 0.0008165323602642028,
+      "loss": 2.4918,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.21618338001683027,
+      "learning_rate": 0.0008159570932385414,
+      "loss": 2.3178,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.41150585631550446,
+      "learning_rate": 0.0008153811290936999,
+      "loss": 2.3755,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18514624084149653,
+      "learning_rate": 0.0008148044691004698,
+      "loss": 2.3044,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.19826778274700552,
+      "learning_rate": 0.0008142271145311783,
+      "loss": 2.371,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.19800687029254593,
+      "learning_rate": 0.000813649066659685,
+      "loss": 2.46,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2038158634919084,
+      "learning_rate": 0.0008130703267613787,
+      "loss": 2.3798,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2162426368972599,
+      "learning_rate": 0.0008124908961131759,
+      "loss": 2.3736,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20210868759344444,
+      "learning_rate": 0.0008119107759935163,
+      "loss": 2.2218,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2370674538947633,
+      "learning_rate": 0.0008113299676823615,
+      "loss": 2.3402,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.25954801138517875,
+      "learning_rate": 0.0008107484724611911,
+      "loss": 2.4794,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2441431767717557,
+      "learning_rate": 0.0008101662916130006,
+      "loss": 2.3825,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24799002523689026,
+      "learning_rate": 0.0008095834264222979,
+      "loss": 2.3756,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.19606787597209255,
+      "learning_rate": 0.0008089998781751009,
+      "loss": 2.4495,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.18675910769787626,
+      "learning_rate": 0.0008084156481589349,
+      "loss": 2.3685,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20931629068769167,
+      "learning_rate": 0.0008078307376628291,
+      "loss": 2.4268,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20023719464407022,
+      "learning_rate": 0.0008072451479773143,
+      "loss": 2.2848,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.18839254025218666,
+      "learning_rate": 0.0008066588803944195,
+      "loss": 2.4575,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.22735360447256375,
+      "learning_rate": 0.0008060719362076697,
+      "loss": 2.3781,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20011640949650045,
+      "learning_rate": 0.0008054843167120826,
+      "loss": 2.4546,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2409932598725054,
+      "learning_rate": 0.0008048960232041663,
+      "loss": 2.3864,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2086057699858548,
+      "learning_rate": 0.0008043070569819153,
+      "loss": 2.3165,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20379135157872952,
+      "learning_rate": 0.0008037174193448089,
+      "loss": 2.4397,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1858701651450303,
+      "learning_rate": 0.0008031271115938077,
+      "loss": 2.268,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1987946802050316,
+      "learning_rate": 0.0008025361350313505,
+      "loss": 2.2898,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2045174568438363,
+      "learning_rate": 0.0008019444909613523,
+      "loss": 2.4713,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.21540873376267614,
+      "learning_rate": 0.0008013521806892003,
+      "loss": 2.4278,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24698433541838496,
+      "learning_rate": 0.000800759205521752,
+      "loss": 2.271,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.21731184854277974,
+      "learning_rate": 0.0008001655667673318,
+      "loss": 2.4819,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.21353345965209342,
+      "learning_rate": 0.0007995712657357279,
+      "loss": 2.4431,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20339203627152008,
+      "learning_rate": 0.0007989763037381904,
+      "loss": 2.3394,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24603137742451495,
+      "learning_rate": 0.0007983806820874271,
+      "loss": 2.3408,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23677984198331511,
+      "learning_rate": 0.0007977844020976016,
+      "loss": 2.4641,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.19725586455489866,
+      "learning_rate": 0.00079718746508433,
+      "loss": 2.4005,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1963443784909098,
+      "learning_rate": 0.0007965898723646776,
+      "loss": 2.3582,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20979711414374208,
+      "learning_rate": 0.0007959916252571573,
+      "loss": 2.2866,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2052443232619753,
+      "learning_rate": 0.000795392725081725,
+      "loss": 2.3464,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2242005329337713,
+      "learning_rate": 0.000794793173159778,
+      "loss": 2.4148,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.22519323494860072,
+      "learning_rate": 0.0007941929708141513,
+      "loss": 2.3863,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23520972594585451,
+      "learning_rate": 0.0007935921193691153,
+      "loss": 2.4021,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1844359162498499,
+      "learning_rate": 0.0007929906201503722,
+      "loss": 2.38,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.19717614082154636,
+      "learning_rate": 0.0007923884744850536,
+      "loss": 2.2309,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2037077725616735,
+      "learning_rate": 0.0007917856837017176,
+      "loss": 2.4433,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1925404590739173,
+      "learning_rate": 0.0007911822491303452,
+      "loss": 2.2646,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20858132264413534,
+      "learning_rate": 0.0007905781721023382,
+      "loss": 2.2701,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.19363526401138292,
+      "learning_rate": 0.000789973453950516,
+      "loss": 2.4239,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.21832182679668807,
+      "learning_rate": 0.000789368096009112,
+      "loss": 2.3583,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.27138257155868745,
+      "learning_rate": 0.0007887620996137721,
+      "loss": 2.3999,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1809686309066487,
+      "learning_rate": 0.0007881554661015497,
+      "loss": 2.4271,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.22860960447863507,
+      "learning_rate": 0.0007875481968109051,
+      "loss": 2.374,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.18724085451783046,
+      "learning_rate": 0.0007869402930817007,
+      "loss": 2.2868,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.22760141002918394,
+      "learning_rate": 0.0007863317562551987,
+      "loss": 2.3192,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.21009428782855413,
+      "learning_rate": 0.0007857225876740584,
+      "loss": 2.2556,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.23318589408968843,
+      "learning_rate": 0.0007851127886823327,
+      "loss": 2.4696,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.18931431759570835,
+      "learning_rate": 0.0007845023606254658,
+      "loss": 2.3637,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.21371372831041674,
+      "learning_rate": 0.0007838913048502894,
+      "loss": 2.3044,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.23741446025466503,
+      "learning_rate": 0.0007832796227050208,
+      "loss": 2.3959,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1954794363907472,
+      "learning_rate": 0.0007826673155392587,
+      "loss": 2.4036,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19355549139180442,
+      "learning_rate": 0.000782054384703981,
+      "loss": 2.3518,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19058707181452672,
+      "learning_rate": 0.0007814408315515418,
+      "loss": 2.2554,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.18862008679669443,
+      "learning_rate": 0.0007808266574356683,
+      "loss": 2.3749,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2011672418433559,
+      "learning_rate": 0.0007802118637114573,
+      "loss": 2.3588,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20490752082136918,
+      "learning_rate": 0.0007795964517353734,
+      "loss": 2.2891,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19697554112813798,
+      "learning_rate": 0.0007789804228652449,
+      "loss": 2.3216,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20214514347467497,
+      "learning_rate": 0.0007783637784602609,
+      "loss": 2.3624,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.216794108368424,
+      "learning_rate": 0.0007777465198809692,
+      "loss": 2.3486,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1887730859733003,
+      "learning_rate": 0.0007771286484892722,
+      "loss": 2.4041,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2047214032497465,
+      "learning_rate": 0.000776510165648425,
+      "loss": 2.3556,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20502758969992438,
+      "learning_rate": 0.0007758910727230311,
+      "loss": 2.3405,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.20142954830152926,
+      "learning_rate": 0.0007752713710790404,
+      "loss": 2.3886,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1904148106052041,
+      "learning_rate": 0.0007746510620837459,
+      "loss": 2.2503,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19559286284291258,
+      "learning_rate": 0.0007740301471057807,
+      "loss": 2.34,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19435104129637765,
+      "learning_rate": 0.0007734086275151146,
+      "loss": 2.3234,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19637982771274068,
+      "learning_rate": 0.0007727865046830517,
+      "loss": 2.4429,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.20868362721860748,
+      "learning_rate": 0.0007721637799822269,
+      "loss": 2.2851,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19460682529125425,
+      "learning_rate": 0.0007715404547866032,
+      "loss": 2.3141,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1842484478550409,
+      "learning_rate": 0.0007709165304714685,
+      "loss": 2.3279,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1954823941784777,
+      "learning_rate": 0.0007702920084134324,
+      "loss": 2.3628,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2057945409179893,
+      "learning_rate": 0.0007696668899904236,
+      "loss": 2.3612,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.221795770009671,
+      "learning_rate": 0.0007690411765816864,
+      "loss": 2.4244,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1897258456722474,
+      "learning_rate": 0.0007684148695677778,
+      "loss": 2.3541,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.18979243212370528,
+      "learning_rate": 0.000767787970330565,
+      "loss": 2.4033,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2433378278761861,
+      "learning_rate": 0.000767160480253221,
+      "loss": 2.4029,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.18750501388840535,
+      "learning_rate": 0.0007665324007202235,
+      "loss": 2.3978,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2225863064309629,
+      "learning_rate": 0.0007659037331173498,
+      "loss": 2.3345,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.213089444219252,
+      "learning_rate": 0.0007652744788316752,
+      "loss": 2.2527,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.21971372810416237,
+      "learning_rate": 0.0007646446392515692,
+      "loss": 2.3673,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23501992316198161,
+      "learning_rate": 0.000764014215766693,
+      "loss": 2.3157,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1856917542793935,
+      "learning_rate": 0.0007633832097679958,
+      "loss": 2.3439,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.25242396922092497,
+      "learning_rate": 0.0007627516226477122,
+      "loss": 2.2608,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.20923617621552523,
+      "learning_rate": 0.0007621194557993589,
+      "loss": 2.429,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.320184605735471,
+      "learning_rate": 0.0007614867106177319,
+      "loss": 2.2587,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.21760552346946388,
+      "learning_rate": 0.0007608533884989029,
+      "loss": 2.3994,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.25308705870800474,
+      "learning_rate": 0.0007602194908402166,
+      "loss": 2.3485,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2736373068892872,
+      "learning_rate": 0.0007595850190402877,
+      "loss": 2.3523,
+      "step": 756
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18683039056710407,
+      "learning_rate": 0.0007589499744989976,
+      "loss": 2.336,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.23174670383419038,
+      "learning_rate": 0.0007583143586174916,
+      "loss": 2.436,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.22727209616487662,
+      "learning_rate": 0.000757678172798175,
+      "loss": 2.3595,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.19149219724328997,
+      "learning_rate": 0.0007570414184447112,
+      "loss": 2.3684,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2046532281742551,
+      "learning_rate": 0.0007564040969620179,
+      "loss": 2.3544,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20459527147083986,
+      "learning_rate": 0.0007557662097562636,
+      "loss": 2.4206,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.26070828709713334,
+      "learning_rate": 0.0007551277582348658,
+      "loss": 2.2653,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18062158061965738,
+      "learning_rate": 0.0007544887438064862,
+      "loss": 2.4017,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1960319135279008,
+      "learning_rate": 0.0007538491678810294,
+      "loss": 2.3809,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.21023391476072942,
+      "learning_rate": 0.0007532090318696381,
+      "loss": 2.2907,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18803088825275407,
+      "learning_rate": 0.0007525683371846913,
+      "loss": 2.3441,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.21310102160155725,
+      "learning_rate": 0.0007519270852398001,
+      "loss": 2.3438,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18331767472362986,
+      "learning_rate": 0.000751285277449806,
+      "loss": 2.3357,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20360588716768827,
+      "learning_rate": 0.0007506429152307756,
+      "loss": 2.317,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2033600173331542,
+      "learning_rate": 0.00075,
+      "loss": 2.4427,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18489937934985012,
+      "learning_rate": 0.00074935653317599,
+      "loss": 2.3327,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20500140415873963,
+      "learning_rate": 0.000748712516178473,
+      "loss": 2.3208,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2664603500454692,
+      "learning_rate": 0.0007480679504283911,
+      "loss": 2.4536,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20872735960447777,
+      "learning_rate": 0.0007474228373478964,
+      "loss": 2.3758,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40314195957461746,
+      "learning_rate": 0.0007467771783603492,
+      "loss": 2.4367,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21929009051306372,
+      "learning_rate": 0.0007461309748903138,
+      "loss": 2.2901,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.28769864004957624,
+      "learning_rate": 0.0007454842283635562,
+      "loss": 2.2851,
+      "step": 778
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20376323767622917,
+      "learning_rate": 0.0007448369402070404,
+      "loss": 2.2719,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20847468050595735,
+      "learning_rate": 0.0007441891118489254,
+      "loss": 2.4159,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21483670199828028,
+      "learning_rate": 0.0007435407447185622,
+      "loss": 2.3409,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18774575655600578,
+      "learning_rate": 0.0007428918402464908,
+      "loss": 2.4111,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.22257786676712701,
+      "learning_rate": 0.0007422423998644359,
+      "loss": 2.327,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19656173140214175,
+      "learning_rate": 0.0007415924250053055,
+      "loss": 2.3243,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19294087772840215,
+      "learning_rate": 0.0007409419171031865,
+      "loss": 2.2991,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2780291290639327,
+      "learning_rate": 0.0007402908775933419,
+      "loss": 2.3384,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.32131660538566914,
+      "learning_rate": 0.0007396393079122077,
+      "loss": 2.2836,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18014440861514922,
+      "learning_rate": 0.0007389872094973896,
+      "loss": 2.3441,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.17922724952907773,
+      "learning_rate": 0.00073833458378766,
+      "loss": 2.3926,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20876180547026796,
+      "learning_rate": 0.0007376814322229544,
+      "loss": 2.3279,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20086651258069252,
+      "learning_rate": 0.0007370277562443688,
+      "loss": 2.3815,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18149783222286842,
+      "learning_rate": 0.0007363735572941564,
+      "loss": 2.3196,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20689608625505085,
+      "learning_rate": 0.0007357188368157236,
+      "loss": 2.3387,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2051407010325912,
+      "learning_rate": 0.0007350635962536284,
+      "loss": 2.3518,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18497534360929443,
+      "learning_rate": 0.0007344078370535756,
+      "loss": 2.328,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2062021676685693,
+      "learning_rate": 0.0007337515606624148,
+      "loss": 2.3337,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.19292346646592923,
+      "learning_rate": 0.0007330947685281362,
+      "loss": 2.3881,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.19431158318164274,
+      "learning_rate": 0.0007324374620998682,
+      "loss": 2.231,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.17960174500928416,
+      "learning_rate": 0.000731779642827874,
+      "loss": 2.3514,
+      "step": 799
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.18651871599780215,
+      "learning_rate": 0.0007311213121635483,
+      "loss": 2.3136,
+      "step": 800
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22151104341638359,
+      "learning_rate": 0.0007304624715594139,
+      "loss": 2.2776,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20191035823757558,
+      "learning_rate": 0.0007298031224691193,
+      "loss": 2.4019,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.21520051258278164,
+      "learning_rate": 0.0007291432663474339,
+      "loss": 2.3578,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2123667485860328,
+      "learning_rate": 0.0007284829046502467,
+      "loss": 2.3404,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20197111059713233,
+      "learning_rate": 0.0007278220388345619,
+      "loss": 2.3036,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22779217721407607,
+      "learning_rate": 0.0007271606703584958,
+      "loss": 2.2107,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2023083957595873,
+      "learning_rate": 0.000726498800681274,
+      "loss": 2.2484,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1975370132197128,
+      "learning_rate": 0.0007258364312632279,
+      "loss": 2.3137,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4135758179483727,
+      "learning_rate": 0.0007251735635657915,
+      "loss": 2.3741,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2053124658749272,
+      "learning_rate": 0.000724510199051498,
+      "loss": 2.3386,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.28838204621232394,
+      "learning_rate": 0.0007238463391839769,
+      "loss": 2.4388,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.195761835066669,
+      "learning_rate": 0.0007231819854279508,
+      "loss": 2.3712,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22214621901168474,
+      "learning_rate": 0.0007225171392492316,
+      "loss": 2.2705,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20084627881887687,
+      "learning_rate": 0.0007218518021147182,
+      "loss": 2.3459,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20314888013603266,
+      "learning_rate": 0.0007211859754923923,
+      "loss": 2.3749,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2013203444191679,
+      "learning_rate": 0.0007205196608513158,
+      "loss": 2.2786,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.19874411305826045,
+      "learning_rate": 0.0007198528596616272,
+      "loss": 2.3801,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.21376035120377598,
+      "learning_rate": 0.0007191855733945387,
+      "loss": 2.2399,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20544550327052827,
+      "learning_rate": 0.0007185178035223327,
+      "loss": 2.2568,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1805528869228331,
+      "learning_rate": 0.0007178495515183583,
+      "loss": 2.2206,
+      "step": 820
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.17453856200009432,
+      "learning_rate": 0.000717180818857029,
+      "loss": 2.2467,
+      "step": 821
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20836730061901637,
+      "learning_rate": 0.0007165116070138182,
+      "loss": 2.2966,
+      "step": 822
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.19154117460323508,
+      "learning_rate": 0.0007158419174652569,
+      "loss": 2.2179,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18413944919918598,
+      "learning_rate": 0.00071517175168893,
+      "loss": 2.3136,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1855865790021418,
+      "learning_rate": 0.0007145011111634732,
+      "loss": 2.2829,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18286565202728272,
+      "learning_rate": 0.0007138299973685694,
+      "loss": 2.3683,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.19827864677694135,
+      "learning_rate": 0.0007131584117849459,
+      "loss": 2.3483,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.19180813794350873,
+      "learning_rate": 0.0007124863558943713,
+      "loss": 2.3933,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2509763907915549,
+      "learning_rate": 0.0007118138311796514,
+      "loss": 2.3301,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18646497383333407,
+      "learning_rate": 0.0007111408391246262,
+      "loss": 2.3524,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.22084470145995486,
+      "learning_rate": 0.0007104673812141675,
+      "loss": 2.3048,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2924279373206126,
+      "learning_rate": 0.0007097934589341745,
+      "loss": 2.2938,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.17615696957060115,
+      "learning_rate": 0.0007091190737715711,
+      "loss": 2.3244,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.28010122400223514,
+      "learning_rate": 0.0007084442272143026,
+      "loss": 2.2665,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.23895613550242362,
+      "learning_rate": 0.000707768920751332,
+      "loss": 2.3736,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.30638765855342515,
+      "learning_rate": 0.0007070931558726373,
+      "loss": 2.42,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.19740973110442517,
+      "learning_rate": 0.0007064169340692076,
+      "loss": 2.3688,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18614562765778994,
+      "learning_rate": 0.0007057402568330407,
+      "loss": 2.2925,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.19021276884658653,
+      "learning_rate": 0.0007050631256571389,
+      "loss": 2.3176,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.20377023574621192,
+      "learning_rate": 0.000704385542035506,
+      "loss": 2.3721,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.19398576581794946,
+      "learning_rate": 0.000703707507463144,
+      "loss": 2.4265,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.18286379247760692,
+      "learning_rate": 0.0007030290234360505,
+      "loss": 2.2834,
+      "step": 842
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2241367836936832,
+      "learning_rate": 0.0007023500914512139,
+      "loss": 2.3313,
+      "step": 843
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1943041234399729,
+      "learning_rate": 0.0007016707130066116,
+      "loss": 2.3732,
+      "step": 844
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1902692626457823,
+      "learning_rate": 0.0007009908896012055,
+      "loss": 2.3236,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.21815866675174161,
+      "learning_rate": 0.0007003106227349399,
+      "loss": 2.3503,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2356714869033572,
+      "learning_rate": 0.000699629913908737,
+      "loss": 2.2722,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.27328164160288015,
+      "learning_rate": 0.0006989487646244943,
+      "loss": 2.3425,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.38423376026800166,
+      "learning_rate": 0.0006982671763850814,
+      "loss": 2.2623,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.22450150737278143,
+      "learning_rate": 0.0006975851506943359,
+      "loss": 2.355,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1962949145078148,
+      "learning_rate": 0.0006969026890570611,
+      "loss": 2.3988,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.251564593921598,
+      "learning_rate": 0.0006962197929790216,
+      "loss": 2.2942,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.19865063121892915,
+      "learning_rate": 0.0006955364639669409,
+      "loss": 2.312,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.21344764152276532,
+      "learning_rate": 0.0006948527035284978,
+      "loss": 2.2317,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2082542713452519,
+      "learning_rate": 0.0006941685131723225,
+      "loss": 2.316,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.19498036767222396,
+      "learning_rate": 0.0006934838944079943,
+      "loss": 2.3128,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.19265923803500504,
+      "learning_rate": 0.0006927988487460378,
+      "loss": 2.3014,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.21734190277392662,
+      "learning_rate": 0.0006921133776979186,
+      "loss": 2.3098,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.198225655389868,
+      "learning_rate": 0.0006914274827760418,
+      "loss": 2.2765,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1895948243656752,
+      "learning_rate": 0.0006907411654937475,
+      "loss": 2.2727,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25103940569568,
+      "learning_rate": 0.0006900544273653075,
+      "loss": 2.3454,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.20405639697615052,
+      "learning_rate": 0.000689367269905922,
+      "loss": 2.1382,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.19272975071836476,
+      "learning_rate": 0.0006886796946317168,
+      "loss": 2.2369,
+      "step": 863
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.19307218900900847,
+      "learning_rate": 0.0006879917030597397,
+      "loss": 2.3114,
+      "step": 864
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.19365180829284487,
+      "learning_rate": 0.0006873032967079561,
+      "loss": 2.3261,
+      "step": 865
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.18465156347534134,
+      "learning_rate": 0.0006866144770952474,
+      "loss": 2.2731,
+      "step": 866
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.19809557904987649,
+      "learning_rate": 0.0006859252457414067,
+      "loss": 2.3451,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.202695843152827,
+      "learning_rate": 0.0006852356041671351,
+      "loss": 2.278,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.198137078645546,
+      "learning_rate": 0.0006845455538940394,
+      "loss": 2.3089,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21071404489809498,
+      "learning_rate": 0.0006838550964446276,
+      "loss": 2.3682,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.20118375585240011,
+      "learning_rate": 0.0006831642333423067,
+      "loss": 2.2411,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.18620407458597057,
+      "learning_rate": 0.000682472966111378,
+      "loss": 2.2187,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.22012949643674648,
+      "learning_rate": 0.0006817812962770348,
+      "loss": 2.2905,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.23612356969713408,
+      "learning_rate": 0.0006810892253653589,
+      "loss": 2.3812,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.22179099176395964,
+      "learning_rate": 0.0006803967549033167,
+      "loss": 2.3639,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2147968161186134,
+      "learning_rate": 0.0006797038864187564,
+      "loss": 2.2744,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2021288657092707,
+      "learning_rate": 0.0006790106214404043,
+      "loss": 2.3627,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.18415746260523375,
+      "learning_rate": 0.0006783169614978614,
+      "loss": 2.3297,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.20374460850089376,
+      "learning_rate": 0.0006776229081216001,
+      "loss": 2.3115,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.23287018318766442,
+      "learning_rate": 0.0006769284628429611,
+      "loss": 2.3263,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.20027640374845418,
+      "learning_rate": 0.0006762336271941498,
+      "loss": 2.2428,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.19756273154373202,
+      "learning_rate": 0.0006755384027082326,
+      "loss": 2.1943,
+      "step": 882
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1915900624203546,
+      "learning_rate": 0.0006748427909191342,
+      "loss": 2.2723,
+      "step": 883
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2085759904057532,
+      "learning_rate": 0.0006741467933616335,
+      "loss": 2.3204,
+      "step": 884
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.22157961624577513,
+      "learning_rate": 0.0006734504115713604,
+      "loss": 2.2837,
+      "step": 885
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1896462746692905,
+      "learning_rate": 0.0006727536470847932,
+      "loss": 2.3189,
+      "step": 886
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.21588230997983862,
+      "learning_rate": 0.000672056501439254,
+      "loss": 2.329,
+      "step": 887
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2036890927389993,
+      "learning_rate": 0.0006713589761729063,
+      "loss": 2.3204,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.22516307588625917,
+      "learning_rate": 0.0006706610728247508,
+      "loss": 2.2577,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23133588472591893,
+      "learning_rate": 0.0006699627929346227,
+      "loss": 2.3687,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.19229499003414702,
+      "learning_rate": 0.0006692641380431879,
+      "loss": 2.3339,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.21774830527816041,
+      "learning_rate": 0.0006685651096919393,
+      "loss": 2.3177,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.20930569239816452,
+      "learning_rate": 0.0006678657094231944,
+      "loss": 2.3752,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2081323023299026,
+      "learning_rate": 0.0006671659387800909,
+      "loss": 2.3064,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.34002435720731017,
+      "learning_rate": 0.000666465799306584,
+      "loss": 2.3881,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.20198223078033153,
+      "learning_rate": 0.0006657652925474423,
+      "loss": 2.3666,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26007057501897357,
+      "learning_rate": 0.000665064420048245,
+      "loss": 2.2755,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26147757897529245,
+      "learning_rate": 0.0006643631833553785,
+      "loss": 2.3477,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2060910800057873,
+      "learning_rate": 0.000663661584016032,
+      "loss": 2.3614,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26167207038201895,
+      "learning_rate": 0.0006629596235781957,
+      "loss": 2.3654,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.29495660841161697,
+      "learning_rate": 0.0006622573035906556,
+      "loss": 2.2417,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.19310821947391765,
+      "learning_rate": 0.0006615546256029921,
+      "loss": 2.2905,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26893157663279665,
+      "learning_rate": 0.0006608515911655743,
+      "loss": 2.3304,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23075046515978526,
+      "learning_rate": 0.0006601482018295591,
+      "loss": 2.35,
+      "step": 904
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1771148701433882,
+      "learning_rate": 0.0006594444591468851,
+      "loss": 2.2412,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.19489504687864068,
+      "learning_rate": 0.0006587403646702713,
+      "loss": 2.2459,
+      "step": 906
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2714518039398787,
+      "learning_rate": 0.0006580359199532126,
+      "loss": 2.3484,
+      "step": 907
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.19406895693333825,
+      "learning_rate": 0.000657331126549977,
+      "loss": 2.3049,
+      "step": 908
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18414270817475306,
+      "learning_rate": 0.0006566259860156014,
+      "loss": 2.3283,
+      "step": 909
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.17909426935652828,
+      "learning_rate": 0.0006559204999058888,
+      "loss": 2.3302,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18988836140984194,
+      "learning_rate": 0.0006552146697774049,
+      "loss": 2.3368,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2044259510785487,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 2.1915,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18714581792959542,
+      "learning_rate": 0.0006538019836941758,
+      "loss": 2.136,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18502481710818772,
+      "learning_rate": 0.0006530951308563431,
+      "loss": 2.3043,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1995484296661357,
+      "learning_rate": 0.0006523879402335567,
+      "loss": 2.3699,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.201887498415686,
+      "learning_rate": 0.0006516804133861429,
+      "loss": 2.3451,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18885297927682473,
+      "learning_rate": 0.0006509725518751698,
+      "loss": 2.4236,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.19519431856405164,
+      "learning_rate": 0.0006502643572624438,
+      "loss": 2.4622,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.17556897597691706,
+      "learning_rate": 0.0006495558311105064,
+      "loss": 2.2327,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.17800897997761006,
+      "learning_rate": 0.0006488469749826305,
+      "loss": 2.3598,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.169142320627074,
+      "learning_rate": 0.000648137790442817,
+      "loss": 2.3581,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.18464938311431825,
+      "learning_rate": 0.0006474282790557916,
+      "loss": 2.268,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.17079890582372748,
+      "learning_rate": 0.000646718442387001,
+      "loss": 2.2837,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1981558380671666,
+      "learning_rate": 0.0006460082820026094,
+      "loss": 2.3379,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1863631429552079,
+      "learning_rate": 0.0006452977994694959,
+      "loss": 2.2186,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20432748249266522,
+      "learning_rate": 0.0006445869963552496,
+      "loss": 2.4264,
+      "step": 926
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.22605622439992526,
+      "learning_rate": 0.0006438758742281672,
+      "loss": 2.3001,
+      "step": 927
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.18833865740090783,
+      "learning_rate": 0.0006431644346572495,
+      "loss": 2.2868,
+      "step": 928
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.19269999433795965,
+      "learning_rate": 0.0006424526792121974,
+      "loss": 2.2867,
+      "step": 929
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23145182417898588,
+      "learning_rate": 0.0006417406094634089,
+      "loss": 2.254,
+      "step": 930
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2292636450731889,
+      "learning_rate": 0.0006410282269819756,
+      "loss": 2.1581,
+      "step": 931
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.19024776600663415,
+      "learning_rate": 0.0006403155333396787,
+      "loss": 2.345,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.20433117337761977,
+      "learning_rate": 0.0006396025301089863,
+      "loss": 2.3431,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.190969878801415,
+      "learning_rate": 0.0006388892188630493,
+      "loss": 2.2589,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.27387163863468883,
+      "learning_rate": 0.0006381756011756982,
+      "loss": 2.4413,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1880523612724434,
+      "learning_rate": 0.0006374616786214403,
+      "loss": 2.368,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.24568313897734026,
+      "learning_rate": 0.0006367474527754544,
+      "loss": 2.2505,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.19424083008133555,
+      "learning_rate": 0.0006360329252135894,
+      "loss": 2.161,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.18516979864603655,
+      "learning_rate": 0.0006353180975123595,
+      "loss": 2.2522,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1933796940382306,
+      "learning_rate": 0.0006346029712489413,
+      "loss": 2.373,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.19567104200965937,
+      "learning_rate": 0.0006338875480011698,
+      "loss": 2.253,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.18928172635653925,
+      "learning_rate": 0.0006331718293475357,
+      "loss": 2.365,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1883402341287598,
+      "learning_rate": 0.0006324558168671811,
+      "loss": 2.2489,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.18528204655746094,
+      "learning_rate": 0.0006317395121398968,
+      "loss": 2.367,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.17923568960477376,
+      "learning_rate": 0.0006310229167461179,
+      "loss": 2.2457,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.21058161914817144,
+      "learning_rate": 0.0006303060322669214,
+      "loss": 2.237,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.22219026920193016,
+      "learning_rate": 0.0006295888602840214,
+      "loss": 2.2812,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.17728502029969723,
+      "learning_rate": 0.0006288714023797671,
+      "loss": 2.2083,
+      "step": 948
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.21983946784046893,
+      "learning_rate": 0.000628153660137138,
+      "loss": 2.2738,
+      "step": 949
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19393617441958705,
+      "learning_rate": 0.0006274356351397413,
+      "loss": 2.3642,
+      "step": 950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.18198251376069294,
+      "learning_rate": 0.0006267173289718079,
+      "loss": 2.3328,
+      "step": 951
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19840623276474004,
+      "learning_rate": 0.000625998743218189,
+      "loss": 2.3928,
+      "step": 952
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.16490550143095098,
+      "learning_rate": 0.000625279879464353,
+      "loss": 2.253,
+      "step": 953
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2219088542312716,
+      "learning_rate": 0.000624560739296381,
+      "loss": 2.2894,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.21281788014230854,
+      "learning_rate": 0.0006238413243009648,
+      "loss": 2.2371,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.17765650671732197,
+      "learning_rate": 0.000623121636065402,
+      "loss": 2.3553,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.20055443908547071,
+      "learning_rate": 0.0006224016761775933,
+      "loss": 2.3315,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23511185735862003,
+      "learning_rate": 0.0006216814462260386,
+      "loss": 2.3971,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19347813456716173,
+      "learning_rate": 0.0006209609477998338,
+      "loss": 2.2927,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1865685439486773,
+      "learning_rate": 0.0006202401824886674,
+      "loss": 2.3553,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.22144670009934586,
+      "learning_rate": 0.0006195191518828162,
+      "loss": 2.2591,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.22211037280269338,
+      "learning_rate": 0.0006187978575731427,
+      "loss": 2.3444,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1939491948181642,
+      "learning_rate": 0.0006180763011510911,
+      "loss": 2.3641,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19701694503342218,
+      "learning_rate": 0.000617354484208684,
+      "loss": 2.1935,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31511446440932334,
+      "learning_rate": 0.0006166324083385189,
+      "loss": 2.4057,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.21384094713131208,
+      "learning_rate": 0.0006159100751337642,
+      "loss": 2.2941,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.21524731913204212,
+      "learning_rate": 0.0006151874861881565,
+      "loss": 2.3714,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.22198862415603574,
+      "learning_rate": 0.0006144646430959964,
+      "loss": 2.3958,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.26020426652389267,
+      "learning_rate": 0.0006137415474521454,
+      "loss": 2.3187,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2158983488323549,
+      "learning_rate": 0.0006130182008520222,
+      "loss": 2.4328,
+      "step": 970
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.29806179780157543,
+      "learning_rate": 0.000612294604891599,
+      "loss": 2.2996,
+      "step": 971
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19094894769493104,
+      "learning_rate": 0.0006115707611673986,
+      "loss": 2.2549,
+      "step": 972
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2407964857339151,
+      "learning_rate": 0.0006108466712764902,
+      "loss": 2.3891,
+      "step": 973
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.18501491780491267,
+      "learning_rate": 0.0006101223368164858,
+      "loss": 2.3702,
+      "step": 974
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.20377471957296275,
+      "learning_rate": 0.0006093977593855375,
+      "loss": 2.277,
+      "step": 975
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.17517923073942576,
+      "learning_rate": 0.0006086729405823335,
+      "loss": 2.2429,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19493129783515237,
+      "learning_rate": 0.0006079478820060943,
+      "loss": 2.2864,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.20241673364164225,
+      "learning_rate": 0.0006072225852565695,
+      "loss": 2.307,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21947432359092806,
+      "learning_rate": 0.0006064970519340341,
+      "loss": 2.2389,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.20267004419356788,
+      "learning_rate": 0.0006057712836392856,
+      "loss": 2.2574,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2483365290695732,
+      "learning_rate": 0.0006050452819736389,
+      "loss": 2.3306,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.24423742779599047,
+      "learning_rate": 0.000604319048538925,
+      "loss": 2.3142,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1926480109330013,
+      "learning_rate": 0.0006035925849374855,
+      "loss": 2.3787,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19989371937625453,
+      "learning_rate": 0.0006028658927721697,
+      "loss": 2.3171,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21778763716887023,
+      "learning_rate": 0.0006021389736463321,
+      "loss": 2.2416,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2528921633359687,
+      "learning_rate": 0.0006014118291638271,
+      "loss": 2.2518,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1880540989884949,
+      "learning_rate": 0.0006006844609290065,
+      "loss": 2.3139,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1806888423551429,
+      "learning_rate": 0.0005999568705467161,
+      "loss": 2.2504,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2086466381261986,
+      "learning_rate": 0.0005992290596222915,
+      "loss": 2.3673,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19798143853838837,
+      "learning_rate": 0.0005985010297615551,
+      "loss": 2.3874,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19687214659810998,
+      "learning_rate": 0.0005977727825708123,
+      "loss": 2.2746,
+      "step": 991
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.19757455206800495,
+      "learning_rate": 0.0005970443196568478,
+      "loss": 2.3306,
+      "step": 992
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.24065239213636638,
+      "learning_rate": 0.0005963156426269227,
+      "loss": 2.2898,
+      "step": 993
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1813782260020924,
+      "learning_rate": 0.0005955867530887702,
+      "loss": 2.2615,
+      "step": 994
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1966319891318423,
+      "learning_rate": 0.0005948576526505923,
+      "loss": 2.2771,
+      "step": 995
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2009311773025214,
+      "learning_rate": 0.0005941283429210568,
+      "loss": 2.246,
+      "step": 996
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21336161194018574,
+      "learning_rate": 0.0005933988255092926,
+      "loss": 2.2709,
+      "step": 997
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1879526817611693,
+      "learning_rate": 0.0005926691020248874,
+      "loss": 2.1845,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2328901742430851,
+      "learning_rate": 0.0005919391740778833,
+      "loss": 2.3408,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2028350622823925,
+      "learning_rate": 0.0005912090432787736,
+      "loss": 2.239,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2247958121454979,
+      "learning_rate": 0.000590478711238499,
+      "loss": 2.3131,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.18419163699253302,
+      "learning_rate": 0.0005897481795684446,
+      "loss": 2.2877,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22600690682937993,
+      "learning_rate": 0.0005890174498804355,
+      "loss": 2.3341,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1868463333906242,
+      "learning_rate": 0.0005882865237867339,
+      "loss": 2.2543,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22091897929768306,
+      "learning_rate": 0.0005875554029000353,
+      "loss": 2.2889,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2031067202214159,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 2.3111,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21675791690392532,
+      "learning_rate": 0.0005860925832005753,
+      "loss": 2.3156,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1832991047491456,
+      "learning_rate": 0.0005853608876153395,
+      "loss": 2.2029,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23240960393591845,
+      "learning_rate": 0.0005846290036921512,
+      "loss": 2.2621,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2948036650826921,
+      "learning_rate": 0.0005838969330458195,
+      "loss": 2.3628,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.18631865453648605,
+      "learning_rate": 0.0005831646772915651,
+      "loss": 2.3131,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.46756874766931944,
+      "learning_rate": 0.0005824322380450173,
+      "loss": 2.3837,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21758479069621758,
+      "learning_rate": 0.0005816996169222102,
+      "loss": 2.3547,
+      "step": 1013
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2093711128555137,
+      "learning_rate": 0.0005809668155395793,
+      "loss": 2.2432,
+      "step": 1014
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.33866714062915965,
+      "learning_rate": 0.0005802338355139578,
+      "loss": 2.2499,
+      "step": 1015
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23024644829842,
+      "learning_rate": 0.0005795006784625728,
+      "loss": 2.3437,
+      "step": 1016
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.26825840097822606,
+      "learning_rate": 0.0005787673460030423,
+      "loss": 2.2835,
+      "step": 1017
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.20183240016934045,
+      "learning_rate": 0.000578033839753371,
+      "loss": 2.2759,
+      "step": 1018
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.19674374213288,
+      "learning_rate": 0.0005773001613319476,
+      "loss": 2.3313,
+      "step": 1019
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.18085727925907052,
+      "learning_rate": 0.00057656631235754,
+      "loss": 2.268,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2038679592440122,
+      "learning_rate": 0.0005758322944492929,
+      "loss": 2.3487,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1926720797649195,
+      "learning_rate": 0.0005750981092267237,
+      "loss": 2.3543,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.19165106237028234,
+      "learning_rate": 0.0005743637583097183,
+      "loss": 2.2886,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.18209386914076375,
+      "learning_rate": 0.0005736292433185291,
+      "loss": 2.2387,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.17615385670374137,
+      "learning_rate": 0.0005728945658737699,
+      "loss": 2.31,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.17806370038029987,
+      "learning_rate": 0.0005721597275964133,
+      "loss": 2.3343,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.19007782070156048,
+      "learning_rate": 0.0005714247301077865,
+      "loss": 2.2888,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.19564426674892993,
+      "learning_rate": 0.0005706895750295682,
+      "loss": 2.3542,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.20871425165631555,
+      "learning_rate": 0.0005699542639837844,
+      "loss": 2.1775,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1907305098301913,
+      "learning_rate": 0.0005692187985928055,
+      "loss": 2.35,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.18374067062515861,
+      "learning_rate": 0.0005684831804793427,
+      "loss": 2.2276,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1861635101650719,
+      "learning_rate": 0.0005677474112664438,
+      "loss": 2.2546,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2242272596410295,
+      "learning_rate": 0.0005670114925774899,
+      "loss": 2.2026,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2618489426289969,
+      "learning_rate": 0.0005662754260361924,
+      "loss": 2.2997,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.33415021566162034,
+      "learning_rate": 0.0005655392132665884,
+      "loss": 2.3202,
+      "step": 1035
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23644062504153296,
+      "learning_rate": 0.000564802855893038,
+      "loss": 2.2847,
+      "step": 1036
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2501062949591512,
+      "learning_rate": 0.0005640663555402198,
+      "loss": 2.2566,
+      "step": 1037
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.203792001564127,
+      "learning_rate": 0.0005633297138331285,
+      "loss": 2.2547,
+      "step": 1038
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19196674519389317,
+      "learning_rate": 0.0005625929323970705,
+      "loss": 2.3013,
+      "step": 1039
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25791094714748275,
+      "learning_rate": 0.0005618560128576603,
+      "loss": 2.2359,
+      "step": 1040
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20325322131583645,
+      "learning_rate": 0.0005611189568408173,
+      "loss": 2.3657,
+      "step": 1041
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2018379047208872,
+      "learning_rate": 0.0005603817659727619,
+      "loss": 2.3465,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21047171654750876,
+      "learning_rate": 0.0005596444418800121,
+      "loss": 2.3567,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20661382862125846,
+      "learning_rate": 0.0005589069861893798,
+      "loss": 2.2525,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19228367630058127,
+      "learning_rate": 0.0005581694005279673,
+      "loss": 2.3059,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20259043264545934,
+      "learning_rate": 0.0005574316865231637,
+      "loss": 2.2479,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19089153911893383,
+      "learning_rate": 0.0005566938458026411,
+      "loss": 2.271,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23068471072136526,
+      "learning_rate": 0.0005559558799943514,
+      "loss": 2.3517,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21223248162790012,
+      "learning_rate": 0.0005552177907265223,
+      "loss": 2.2576,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.31319663438786005,
+      "learning_rate": 0.000554479579627654,
+      "loss": 2.3121,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20446917400573966,
+      "learning_rate": 0.0005537412483265157,
+      "loss": 2.2647,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2786858949326627,
+      "learning_rate": 0.0005530027984521413,
+      "loss": 2.2764,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21629333620963334,
+      "learning_rate": 0.0005522642316338268,
+      "loss": 2.2398,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.26547945028077696,
+      "learning_rate": 0.0005515255495011259,
+      "loss": 2.3294,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.26634753251535265,
+      "learning_rate": 0.0005507867536838472,
+      "loss": 2.3617,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.29386828296597256,
+      "learning_rate": 0.0005500478458120492,
+      "loss": 2.2953,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19006661747447498,
+      "learning_rate": 0.0005493088275160387,
+      "loss": 2.2812,
+      "step": 1057
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2454462241562612,
+      "learning_rate": 0.0005485697004263657,
+      "loss": 2.3125,
+      "step": 1058
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2399264690887458,
+      "learning_rate": 0.0005478304661738199,
+      "loss": 2.3748,
+      "step": 1059
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1738298340232373,
+      "learning_rate": 0.0005470911263894279,
+      "loss": 2.2223,
+      "step": 1060
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.179827510270877,
+      "learning_rate": 0.0005463516827044491,
+      "loss": 2.3037,
+      "step": 1061
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22965394233230552,
+      "learning_rate": 0.000545612136750372,
+      "loss": 2.3539,
+      "step": 1062
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.18669379906517364,
+      "learning_rate": 0.0005448724901589107,
+      "loss": 2.2044,
+      "step": 1063
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.21985341501033437,
+      "learning_rate": 0.0005441327445620014,
+      "loss": 2.2574,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19425404727092083,
+      "learning_rate": 0.0005433929015917988,
+      "loss": 2.3843,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.25767406485142264,
+      "learning_rate": 0.0005426529628806724,
+      "loss": 2.2817,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.18238472501806316,
+      "learning_rate": 0.0005419129300612029,
+      "loss": 2.2495,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.201570602298331,
+      "learning_rate": 0.000541172804766179,
+      "loss": 2.2961,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.24086221216322745,
+      "learning_rate": 0.0005404325886285927,
+      "loss": 2.3457,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2209726697952291,
+      "learning_rate": 0.000539692283281637,
+      "loss": 2.2352,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.23454298404531085,
+      "learning_rate": 0.0005389518903587017,
+      "loss": 2.3354,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.17445272692232963,
+      "learning_rate": 0.0005382114114933695,
+      "loss": 2.294,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2030831412403649,
+      "learning_rate": 0.0005374708483194132,
+      "loss": 2.2684,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2435087475863543,
+      "learning_rate": 0.000536730202470791,
+      "loss": 2.244,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2709614190995955,
+      "learning_rate": 0.0005359894755816443,
+      "loss": 2.3084,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22929096859745687,
+      "learning_rate": 0.0005352486692862926,
+      "loss": 2.3476,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.23709380146561126,
+      "learning_rate": 0.0005345077852192307,
+      "loss": 2.2751,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.20407291229680802,
+      "learning_rate": 0.0005337668250151254,
+      "loss": 2.2393,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19233547093680542,
+      "learning_rate": 0.0005330257903088111,
+      "loss": 2.3139,
+      "step": 1079
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.21543834936182232,
+      "learning_rate": 0.000532284682735287,
+      "loss": 2.2958,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17978413090947715,
+      "learning_rate": 0.0005315435039297124,
+      "loss": 2.2525,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20171842770879145,
+      "learning_rate": 0.0005308022555274046,
+      "loss": 2.2355,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18658755947304875,
+      "learning_rate": 0.0005300609391638336,
+      "loss": 2.2889,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1851271837008504,
+      "learning_rate": 0.0005293195564746201,
+      "loss": 2.3078,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18869242598405417,
+      "learning_rate": 0.0005285781090955304,
+      "loss": 2.2867,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19400741311691838,
+      "learning_rate": 0.0005278365986624743,
+      "loss": 2.288,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19096776001875135,
+      "learning_rate": 0.0005270950268115001,
+      "loss": 2.3019,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18541987260378812,
+      "learning_rate": 0.0005263533951787919,
+      "loss": 2.227,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18190656042986617,
+      "learning_rate": 0.000525611705400666,
+      "loss": 2.2503,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.26748336928825994,
+      "learning_rate": 0.0005248699591135664,
+      "loss": 2.3488,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17069847335031554,
+      "learning_rate": 0.0005241281579540618,
+      "loss": 2.3185,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1707549629958552,
+      "learning_rate": 0.0005233863035588427,
+      "loss": 2.1952,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1978292021451277,
+      "learning_rate": 0.0005226443975647161,
+      "loss": 2.2348,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23992658415043197,
+      "learning_rate": 0.0005219024416086036,
+      "loss": 2.2473,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1909821484576342,
+      "learning_rate": 0.0005211604373275366,
+      "loss": 2.2342,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19530377327191842,
+      "learning_rate": 0.0005204183863586533,
+      "loss": 2.2779,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.30953001418338605,
+      "learning_rate": 0.0005196762903391951,
+      "loss": 2.3838,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1936213278165242,
+      "learning_rate": 0.0005189341509065023,
+      "loss": 2.2591,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.279486149984322,
+      "learning_rate": 0.0005181919696980112,
+      "loss": 2.3146,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2274874356108612,
+      "learning_rate": 0.0005174497483512506,
+      "loss": 2.2009,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19372626673628057,
+      "learning_rate": 0.0005167074885038374,
+      "loss": 2.2979,
+      "step": 1101
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23207218951252356,
+      "learning_rate": 0.0005159651917934735,
+      "loss": 2.2427,
+      "step": 1102
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18513489727114354,
+      "learning_rate": 0.0005152228598579428,
+      "loss": 2.2296,
+      "step": 1103
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20474673296478796,
+      "learning_rate": 0.000514480494335106,
+      "loss": 2.2801,
+      "step": 1104
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20996337013276892,
+      "learning_rate": 0.0005137380968628983,
+      "loss": 2.2274,
+      "step": 1105
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.17903395378055992,
+      "learning_rate": 0.0005129956690793255,
+      "loss": 2.2409,
+      "step": 1106
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2015642795837232,
+      "learning_rate": 0.0005122532126224601,
+      "loss": 2.2672,
+      "step": 1107
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18309425104212837,
+      "learning_rate": 0.0005115107291304378,
+      "loss": 2.2631,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20530802885289537,
+      "learning_rate": 0.0005107682202414544,
+      "loss": 2.1865,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23239605891290058,
+      "learning_rate": 0.0005100256875937613,
+      "loss": 2.2096,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1893105068635605,
+      "learning_rate": 0.0005092831328256625,
+      "loss": 2.2451,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.21067724947451022,
+      "learning_rate": 0.0005085405575755105,
+      "loss": 2.249,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18714955962932112,
+      "learning_rate": 0.0005077979634817034,
+      "loss": 2.3871,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.181492974892437,
+      "learning_rate": 0.0005070553521826808,
+      "loss": 2.292,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18986047298411007,
+      "learning_rate": 0.00050631272531692,
+      "loss": 2.2914,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1772325310036856,
+      "learning_rate": 0.0005055700845229327,
+      "loss": 2.2242,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.19854487559145773,
+      "learning_rate": 0.000504827431439262,
+      "loss": 2.2306,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20294962941402123,
+      "learning_rate": 0.000504084767704477,
+      "loss": 2.2413,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20062240416716906,
+      "learning_rate": 0.0005033420949571712,
+      "loss": 2.2624,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1905768920984189,
+      "learning_rate": 0.0005025994148359574,
+      "loss": 2.2958,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.21825056218276598,
+      "learning_rate": 0.0005018567289794651,
+      "loss": 2.2078,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1862547344235112,
+      "learning_rate": 0.0005011140390263362,
+      "loss": 2.2051,
+      "step": 1122
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2036333448599533,
+      "learning_rate": 0.0005003713466152218,
+      "loss": 2.2355,
+      "step": 1123
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21524136440080122,
+      "learning_rate": 0.0004996286533847783,
+      "loss": 2.3383,
+      "step": 1124
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21719887869380947,
+      "learning_rate": 0.000498885960973664,
+      "loss": 2.2381,
+      "step": 1125
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21165162016723654,
+      "learning_rate": 0.000498143271020535,
+      "loss": 2.3354,
+      "step": 1126
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.18888347385427734,
+      "learning_rate": 0.0004974005851640428,
+      "loss": 2.2506,
+      "step": 1127
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.19807216773467462,
+      "learning_rate": 0.000496657905042829,
+      "loss": 2.2268,
+      "step": 1128
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21322649681062633,
+      "learning_rate": 0.0004959152322955232,
+      "loss": 2.3084,
+      "step": 1129
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2003205884852552,
+      "learning_rate": 0.0004951725685607382,
+      "loss": 2.2654,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22529761269497126,
+      "learning_rate": 0.0004944299154770673,
+      "loss": 2.3608,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.18678574236074785,
+      "learning_rate": 0.0004936872746830802,
+      "loss": 2.3206,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.17608884330479413,
+      "learning_rate": 0.0004929446478173195,
+      "loss": 2.2501,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.18342957380802533,
+      "learning_rate": 0.0004922020365182968,
+      "loss": 2.2308,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1792246947862169,
+      "learning_rate": 0.0004914594424244897,
+      "loss": 2.2912,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1809363773455921,
+      "learning_rate": 0.0004907168671743376,
+      "loss": 2.2992,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.19986065221721425,
+      "learning_rate": 0.0004899743124062387,
+      "loss": 2.3218,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21152675875515442,
+      "learning_rate": 0.0004892317797585456,
+      "loss": 2.3354,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2005213989352476,
+      "learning_rate": 0.0004884892708695623,
+      "loss": 2.2712,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.17032316773415224,
+      "learning_rate": 0.0004877467873775402,
+      "loss": 2.3343,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2025232049746483,
+      "learning_rate": 0.00048700433092067473,
+      "loss": 2.1061,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.20032447958338767,
+      "learning_rate": 0.0004862619031371019,
+      "loss": 2.2667,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21950388068719603,
+      "learning_rate": 0.0004855195056648942,
+      "loss": 2.2991,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2795589966691644,
+      "learning_rate": 0.00048477714014205734,
+      "loss": 2.3178,
+      "step": 1144
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.19261517425854324,
+      "learning_rate": 0.00048403480820652644,
+      "loss": 2.2733,
+      "step": 1145
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2054961487492141,
+      "learning_rate": 0.0004832925114961629,
+      "loss": 2.3108,
+      "step": 1146
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.23954046960380154,
+      "learning_rate": 0.0004825502516487497,
+      "loss": 2.2358,
+      "step": 1147
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.18611298954807698,
+      "learning_rate": 0.00048180803030198896,
+      "loss": 2.2842,
+      "step": 1148
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.19093372814583423,
+      "learning_rate": 0.0004810658490934979,
+      "loss": 2.2221,
+      "step": 1149
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.19223203769535774,
+      "learning_rate": 0.000480323709660805,
+      "loss": 2.3011,
+      "step": 1150
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22731974388004775,
+      "learning_rate": 0.0004795816136413467,
+      "loss": 2.2097,
+      "step": 1151
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1912017775056782,
+      "learning_rate": 0.00047883956267246353,
+      "loss": 2.3103,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2499409836693643,
+      "learning_rate": 0.00047809755839139657,
+      "loss": 2.3499,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2867112636491996,
+      "learning_rate": 0.0004773556024352841,
+      "loss": 2.3371,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22500797636774908,
+      "learning_rate": 0.00047661369644115754,
+      "loss": 2.3049,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22330440236901491,
+      "learning_rate": 0.0004758718420459383,
+      "loss": 2.2907,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.18147654948271694,
+      "learning_rate": 0.0004751300408864339,
+      "loss": 2.1761,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2032371439142983,
+      "learning_rate": 0.00047438829459933414,
+      "loss": 2.2264,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.21820864234735196,
+      "learning_rate": 0.0004736466048212082,
+      "loss": 2.2077,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.23299194856534536,
+      "learning_rate": 0.0004729049731885002,
+      "loss": 2.2465,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2038406896073209,
+      "learning_rate": 0.000472163401337526,
+      "loss": 2.1641,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1772943850702147,
+      "learning_rate": 0.00047142189090446985,
+      "loss": 2.1578,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.20447740878697107,
+      "learning_rate": 0.0004706804435253802,
+      "loss": 2.2378,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.20052718808675293,
+      "learning_rate": 0.0004699390608361665,
+      "loss": 2.2263,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.23535840606907324,
+      "learning_rate": 0.0004691977444725955,
+      "loss": 2.2469,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.19195892214675228,
+      "learning_rate": 0.0004684564960702877,
+      "loss": 2.2247,
+      "step": 1166
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.22245366059583868,
+      "learning_rate": 0.0004677153172647131,
+      "loss": 2.2232,
+      "step": 1167
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3128015982715286,
+      "learning_rate": 0.00046697420969118894,
+      "loss": 2.2859,
+      "step": 1168
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.18677959306683925,
+      "learning_rate": 0.00046623317498487466,
+      "loss": 2.2198,
+      "step": 1169
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.18287516761933456,
+      "learning_rate": 0.0004654922147807694,
+      "loss": 2.2928,
+      "step": 1170
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.23487836310485485,
+      "learning_rate": 0.00046475133071370757,
+      "loss": 2.3733,
+      "step": 1171
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.236264178735234,
+      "learning_rate": 0.00046401052441835574,
+      "loss": 2.1708,
+      "step": 1172
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.18304796803271958,
+      "learning_rate": 0.000463269797529209,
+      "loss": 2.1492,
+      "step": 1173
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.19650688261061686,
+      "learning_rate": 0.00046252915168058697,
+      "loss": 2.2861,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2153293851553533,
+      "learning_rate": 0.0004617885885066305,
+      "loss": 2.2142,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.19038469718345974,
+      "learning_rate": 0.0004610481096412984,
+      "loss": 2.2386,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2787724107934683,
+      "learning_rate": 0.000460307716718363,
+      "loss": 2.3103,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.20564265023096304,
+      "learning_rate": 0.0004595674113714074,
+      "loss": 2.2908,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.19329286241075597,
+      "learning_rate": 0.0004588271952338212,
+      "loss": 2.2678,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2365834813151521,
+      "learning_rate": 0.00045808706993879714,
+      "loss": 2.3055,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.185809493920274,
+      "learning_rate": 0.00045734703711932767,
+      "loss": 2.2478,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.18810050631567263,
+      "learning_rate": 0.0004566070984082013,
+      "loss": 2.2351,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2467368430148837,
+      "learning_rate": 0.00045586725543799865,
+      "loss": 2.2115,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17927689117548848,
+      "learning_rate": 0.00045512750984108937,
+      "loss": 2.2159,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1846781117772759,
+      "learning_rate": 0.000454387863249628,
+      "loss": 2.3077,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2900751512075014,
+      "learning_rate": 0.00045364831729555096,
+      "loss": 2.2833,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1912324515919627,
+      "learning_rate": 0.0004529088736105721,
+      "loss": 2.2981,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.21196450310910056,
+      "learning_rate": 0.0004521695338261802,
+      "loss": 2.34,
+      "step": 1188
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18125382055607042,
+      "learning_rate": 0.0004514302995736344,
+      "loss": 2.2563,
+      "step": 1189
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.19025315113124708,
+      "learning_rate": 0.0004506911724839613,
+      "loss": 2.2725,
+      "step": 1190
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18721804621277707,
+      "learning_rate": 0.0004499521541879508,
+      "loss": 2.2682,
+      "step": 1191
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18032970095817946,
+      "learning_rate": 0.00044921324631615303,
+      "loss": 2.2253,
+      "step": 1192
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.20585341122256842,
+      "learning_rate": 0.0004484744504988742,
+      "loss": 2.3214,
+      "step": 1193
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1908612281789712,
+      "learning_rate": 0.00044773576836617336,
+      "loss": 2.2477,
+      "step": 1194
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18685817519312983,
+      "learning_rate": 0.0004469972015478588,
+      "loss": 2.2694,
+      "step": 1195
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18094547890158294,
+      "learning_rate": 0.0004462587516734844,
+      "loss": 2.3555,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.19024835084237673,
+      "learning_rate": 0.00044552042037234596,
+      "loss": 2.2099,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.25298620550383893,
+      "learning_rate": 0.00044478220927347774,
+      "loss": 2.2874,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18346510605154787,
+      "learning_rate": 0.00044404412000564875,
+      "loss": 2.2492,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18878340688085313,
+      "learning_rate": 0.000443306154197359,
+      "loss": 2.2732,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24906530126708393,
+      "learning_rate": 0.00044256831347683646,
+      "loss": 2.2715,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1922182814973345,
+      "learning_rate": 0.0004418305994720328,
+      "loss": 2.2154,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.19288009080164145,
+      "learning_rate": 0.0004410930138106203,
+      "loss": 2.2669,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.17549155099447092,
+      "learning_rate": 0.000440355558119988,
+      "loss": 2.2603,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.21255106368592674,
+      "learning_rate": 0.00043961823402723814,
+      "loss": 2.2817,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.21128789558812464,
+      "learning_rate": 0.0004388810431591829,
+      "loss": 2.1811,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.17766707366364953,
+      "learning_rate": 0.0004381439871423398,
+      "loss": 2.1903,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.27827061949316606,
+      "learning_rate": 0.00043740706760292966,
+      "loss": 2.3613,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.19926969181455123,
+      "learning_rate": 0.0004366702861668716,
+      "loss": 2.2602,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18811612085536475,
+      "learning_rate": 0.00043593364445978036,
+      "loss": 2.1927,
+      "step": 1210
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18896764136852554,
+      "learning_rate": 0.0004351971441069622,
+      "loss": 2.2669,
+      "step": 1211
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1938219094816705,
+      "learning_rate": 0.0004344607867334116,
+      "loss": 2.2337,
+      "step": 1212
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18926851477943216,
+      "learning_rate": 0.00043372457396380766,
+      "loss": 2.1689,
+      "step": 1213
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1973034228890134,
+      "learning_rate": 0.00043298850742251013,
+      "loss": 2.1653,
+      "step": 1214
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.19488744185364906,
+      "learning_rate": 0.0004322525887335563,
+      "loss": 2.3052,
+      "step": 1215
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18349572536925124,
+      "learning_rate": 0.00043151681952065734,
+      "loss": 2.2275,
+      "step": 1216
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.20265050580001648,
+      "learning_rate": 0.00043078120140719456,
+      "loss": 2.3454,
+      "step": 1217
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18276899434119848,
+      "learning_rate": 0.0004300457360162158,
+      "loss": 2.2345,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1862332177549386,
+      "learning_rate": 0.0004293104249704319,
+      "loss": 2.2503,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16820538876559463,
+      "learning_rate": 0.00042857526989221355,
+      "loss": 2.2382,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18881251207114616,
+      "learning_rate": 0.00042784027240358674,
+      "loss": 2.2268,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18387922415519586,
+      "learning_rate": 0.0004271054341262301,
+      "loss": 2.1616,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18911087035563762,
+      "learning_rate": 0.000426370756681471,
+      "loss": 2.3294,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.17645987498222496,
+      "learning_rate": 0.0004256362416902817,
+      "loss": 2.1901,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.190222143136341,
+      "learning_rate": 0.00042490189077327637,
+      "loss": 2.2048,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18429446437189342,
+      "learning_rate": 0.00042416770555070703,
+      "loss": 2.2047,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.20509554885670242,
+      "learning_rate": 0.00042343368764246,
+      "loss": 2.3187,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.23090396078407904,
+      "learning_rate": 0.0004226998386680524,
+      "loss": 2.2922,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.18881409479070263,
+      "learning_rate": 0.000421966160246629,
+      "loss": 2.3231,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21068368167770815,
+      "learning_rate": 0.00042123265399695783,
+      "loss": 2.2151,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1927934321189989,
+      "learning_rate": 0.0004204993215374273,
+      "loss": 2.206,
+      "step": 1231
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.20990951449812512,
+      "learning_rate": 0.00041976616448604226,
+      "loss": 2.3327,
+      "step": 1232
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.19463041699272174,
+      "learning_rate": 0.00041903318446042076,
+      "loss": 2.2831,
+      "step": 1233
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18135827752732028,
+      "learning_rate": 0.00041830038307778984,
+      "loss": 2.3148,
+      "step": 1234
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1746691825406842,
+      "learning_rate": 0.0004175677619549828,
+      "loss": 2.2179,
+      "step": 1235
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.21862887106824952,
+      "learning_rate": 0.000416835322708435,
+      "loss": 2.3168,
+      "step": 1236
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2728361671936902,
+      "learning_rate": 0.00041610306695418056,
+      "loss": 2.3114,
+      "step": 1237
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18749376213413793,
+      "learning_rate": 0.0004153709963078488,
+      "loss": 2.2081,
+      "step": 1238
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20172508003082934,
+      "learning_rate": 0.0004146391123846606,
+      "loss": 2.1533,
+      "step": 1239
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1905211734280136,
+      "learning_rate": 0.0004139074167994249,
+      "loss": 2.3602,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.22928757774524552,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 2.2466,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.17774388780265438,
+      "learning_rate": 0.0004124445970999648,
+      "loss": 2.2,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18508717661770255,
+      "learning_rate": 0.00041171347621326627,
+      "loss": 2.2453,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20431750035364474,
+      "learning_rate": 0.00041098255011956465,
+      "loss": 2.3943,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1866450005792133,
+      "learning_rate": 0.00041025182043155547,
+      "loss": 2.1703,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2172384875477202,
+      "learning_rate": 0.000409521288761501,
+      "loss": 2.1623,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20709352285882723,
+      "learning_rate": 0.00040879095672122646,
+      "loss": 2.3205,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.17462529618629322,
+      "learning_rate": 0.0004080608259221167,
+      "loss": 2.2962,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.21031774762766467,
+      "learning_rate": 0.0004073308979751126,
+      "loss": 2.2274,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18354241116167444,
+      "learning_rate": 0.0004066011744907074,
+      "loss": 2.2819,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18975372655131278,
+      "learning_rate": 0.00040587165707894326,
+      "loss": 2.2845,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1797951642734182,
+      "learning_rate": 0.0004051423473494076,
+      "loss": 2.2174,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.19008946430655352,
+      "learning_rate": 0.0004044132469112299,
+      "loss": 2.1995,
+      "step": 1253
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18958866429164678,
+      "learning_rate": 0.00040368435737307733,
+      "loss": 2.1729,
+      "step": 1254
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2224723642152882,
+      "learning_rate": 0.00040295568034315224,
+      "loss": 2.2977,
+      "step": 1255
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1957094952263542,
+      "learning_rate": 0.0004022272174291878,
+      "loss": 2.2346,
+      "step": 1256
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.22023229819181303,
+      "learning_rate": 0.0004014989702384449,
+      "loss": 2.2463,
+      "step": 1257
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1906386449405532,
+      "learning_rate": 0.00040077094037770843,
+      "loss": 2.1761,
+      "step": 1258
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.20287614741675425,
+      "learning_rate": 0.0004000431294532838,
+      "loss": 2.2827,
+      "step": 1259
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2528854170430994,
+      "learning_rate": 0.0003993155390709935,
+      "loss": 2.2153,
+      "step": 1260
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.19241776590473417,
+      "learning_rate": 0.0003985881708361729,
+      "loss": 2.3144,
+      "step": 1261
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1888228793502787,
+      "learning_rate": 0.00039786102635366784,
+      "loss": 2.162,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2234874940035254,
+      "learning_rate": 0.0003971341072278302,
+      "loss": 2.1962,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.18955654291845556,
+      "learning_rate": 0.00039640741506251457,
+      "loss": 2.1761,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23993605003845758,
+      "learning_rate": 0.00039568095146107495,
+      "loss": 2.2399,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.18445708208105568,
+      "learning_rate": 0.00039495471802636096,
+      "loss": 2.2359,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.186414375797195,
+      "learning_rate": 0.0003942287163607145,
+      "loss": 2.3477,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1838374171172832,
+      "learning_rate": 0.0003935029480659658,
+      "loss": 2.2312,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.20425104144258618,
+      "learning_rate": 0.00039277741474343054,
+      "loss": 2.2464,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.21409283388797837,
+      "learning_rate": 0.0003920521179939057,
+      "loss": 2.2854,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.18037460096337335,
+      "learning_rate": 0.00039132705941766644,
+      "loss": 2.2891,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.188793063550782,
+      "learning_rate": 0.0003906022406144624,
+      "loss": 2.2689,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23379910810357254,
+      "learning_rate": 0.0003898776631835143,
+      "loss": 2.2444,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2418846347958164,
+      "learning_rate": 0.00038915332872350994,
+      "loss": 2.2695,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.20898329577923583,
+      "learning_rate": 0.00038842923883260135,
+      "loss": 2.2475,
+      "step": 1275
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.20182748139344864,
+      "learning_rate": 0.00038770539510840093,
+      "loss": 2.2841,
+      "step": 1276
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17670515578860566,
+      "learning_rate": 0.00038698179914797783,
+      "loss": 2.1946,
+      "step": 1277
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.20684398924091532,
+      "learning_rate": 0.0003862584525478545,
+      "loss": 2.4186,
+      "step": 1278
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17770115857670463,
+      "learning_rate": 0.00038553535690400353,
+      "loss": 2.3019,
+      "step": 1279
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2358064511008403,
+      "learning_rate": 0.00038481251381184355,
+      "loss": 2.3173,
+      "step": 1280
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17765235888945596,
+      "learning_rate": 0.00038408992486623584,
+      "loss": 2.1932,
+      "step": 1281
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.18892421840227794,
+      "learning_rate": 0.00038336759166148117,
+      "loss": 2.1715,
+      "step": 1282
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3000085130779585,
+      "learning_rate": 0.0003826455157913159,
+      "loss": 2.2992,
+      "step": 1283
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22045775371586201,
+      "learning_rate": 0.00038192369884890886,
+      "loss": 2.2475,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22013586932891713,
+      "learning_rate": 0.00038120214242685723,
+      "loss": 2.1329,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.340075280439399,
+      "learning_rate": 0.00038048084811718373,
+      "loss": 2.2927,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.23364609073286757,
+      "learning_rate": 0.0003797598175113327,
+      "loss": 2.1574,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2102621086284534,
+      "learning_rate": 0.0003790390522001662,
+      "loss": 2.3669,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3094465808386579,
+      "learning_rate": 0.0003783185537739615,
+      "loss": 2.1568,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.28653852249152384,
+      "learning_rate": 0.00037759832382240697,
+      "loss": 2.132,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.18038992475960816,
+      "learning_rate": 0.00037687836393459826,
+      "loss": 2.2981,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2546417220749466,
+      "learning_rate": 0.0003761586756990354,
+      "loss": 2.2508,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2690586355651456,
+      "learning_rate": 0.0003754392607036191,
+      "loss": 2.2928,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1799824059163454,
+      "learning_rate": 0.0003747201205356472,
+      "loss": 2.1295,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.20952931801550648,
+      "learning_rate": 0.0003740012567818111,
+      "loss": 2.2855,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3001885106631049,
+      "learning_rate": 0.0003732826710281922,
+      "loss": 2.2083,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.24445052518317822,
+      "learning_rate": 0.0003725643648602588,
+      "loss": 2.1802,
+      "step": 1297
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.19661733746944998,
+      "learning_rate": 0.0003718463398628621,
+      "loss": 2.2961,
+      "step": 1298
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18121739168870124,
+      "learning_rate": 0.0003711285976202331,
+      "loss": 2.2257,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18195545427631368,
+      "learning_rate": 0.0003704111397159787,
+      "loss": 2.2414,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2405270900423001,
+      "learning_rate": 0.0003696939677330788,
+      "loss": 2.2563,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18034266293866646,
+      "learning_rate": 0.00036897708325388213,
+      "loss": 2.2153,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1850260387124836,
+      "learning_rate": 0.0003682604878601034,
+      "loss": 2.2158,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2064985826226712,
+      "learning_rate": 0.000367544183132819,
+      "loss": 2.3413,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18712081828786284,
+      "learning_rate": 0.0003668281706524645,
+      "loss": 2.2059,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2734511063203182,
+      "learning_rate": 0.0003661124519988304,
+      "loss": 2.2164,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18255899373613693,
+      "learning_rate": 0.00036539702875105893,
+      "loss": 2.3133,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1879123363188035,
+      "learning_rate": 0.0003646819024876406,
+      "loss": 2.2098,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.180183574488371,
+      "learning_rate": 0.0003639670747864107,
+      "loss": 2.2786,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.19242360692321994,
+      "learning_rate": 0.00036325254722454584,
+      "loss": 2.1864,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1806397753088917,
+      "learning_rate": 0.00036253832137855997,
+      "loss": 2.2608,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18482521161844623,
+      "learning_rate": 0.00036182439882430183,
+      "loss": 2.2817,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2552333068790262,
+      "learning_rate": 0.00036111078113695096,
+      "loss": 2.2366,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.19595820427188101,
+      "learning_rate": 0.0003603974698910139,
+      "loss": 2.2617,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.17740108185580644,
+      "learning_rate": 0.0003596844666603214,
+      "loss": 2.2654,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.23578194280941028,
+      "learning_rate": 0.0003589717730180245,
+      "loss": 2.2537,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2658032910863844,
+      "learning_rate": 0.00035825939053659117,
+      "loss": 2.3486,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.21632284442012603,
+      "learning_rate": 0.00035754732078780273,
+      "loss": 2.3047,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18913978026393974,
+      "learning_rate": 0.00035683556534275076,
+      "loss": 2.1875,
+      "step": 1319
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2844470053866404,
+      "learning_rate": 0.00035612412577183303,
+      "loss": 2.2038,
+      "step": 1320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.18642782790201706,
+      "learning_rate": 0.00035541300364475063,
+      "loss": 2.2436,
+      "step": 1321
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1775893194022604,
+      "learning_rate": 0.0003547022005305043,
+      "loss": 2.1513,
+      "step": 1322
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.20385968479729516,
+      "learning_rate": 0.0003539917179973907,
+      "loss": 2.2503,
+      "step": 1323
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.17595962168765855,
+      "learning_rate": 0.00035328155761299917,
+      "loss": 2.1725,
+      "step": 1324
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.17872433773992158,
+      "learning_rate": 0.0003525717209442085,
+      "loss": 2.2968,
+      "step": 1325
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.19567661799867794,
+      "learning_rate": 0.00035186220955718306,
+      "loss": 2.2611,
+      "step": 1326
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.20273265369321386,
+      "learning_rate": 0.0003511530250173696,
+      "loss": 2.3664,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.20211778955213286,
+      "learning_rate": 0.00035044416888949364,
+      "loss": 2.2597,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.19139268504947132,
+      "learning_rate": 0.0003497356427375562,
+      "loss": 2.2682,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.18447269131526733,
+      "learning_rate": 0.00034902744812483034,
+      "loss": 2.2752,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.19284858252669615,
+      "learning_rate": 0.00034831958661385714,
+      "loss": 2.2985,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.18461881893382576,
+      "learning_rate": 0.0003476120597664434,
+      "loss": 2.2102,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1855244860987682,
+      "learning_rate": 0.00034690486914365704,
+      "loss": 2.2778,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1911956466898806,
+      "learning_rate": 0.00034619801630582435,
+      "loss": 2.2485,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.20465760409312944,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 2.2052,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2801317674455071,
+      "learning_rate": 0.0003447853302225952,
+      "loss": 2.2726,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.18842951958541385,
+      "learning_rate": 0.00034407950009411126,
+      "loss": 2.1703,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1973682529156317,
+      "learning_rate": 0.00034337401398439873,
+      "loss": 2.2267,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1835531916927005,
+      "learning_rate": 0.00034266887345002305,
+      "loss": 2.2262,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2582758447711379,
+      "learning_rate": 0.0003419640800467874,
+      "loss": 2.243,
+      "step": 1340
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.29089237700785897,
+      "learning_rate": 0.0003412596353297288,
+      "loss": 2.1537,
+      "step": 1341
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.20166142977304616,
+      "learning_rate": 0.00034055554085311493,
+      "loss": 2.2695,
+      "step": 1342
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.19518983404552798,
+      "learning_rate": 0.00033985179817044105,
+      "loss": 2.2087,
+      "step": 1343
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2468136038172577,
+      "learning_rate": 0.0003391484088344257,
+      "loss": 2.2733,
+      "step": 1344
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.24428106967888125,
+      "learning_rate": 0.00033844537439700807,
+      "loss": 2.2454,
+      "step": 1345
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2029546107476688,
+      "learning_rate": 0.00033774269640934445,
+      "loss": 2.2247,
+      "step": 1346
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.20522621702789048,
+      "learning_rate": 0.0003370403764218045,
+      "loss": 2.2284,
+      "step": 1347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.18433801924083765,
+      "learning_rate": 0.000336338415983968,
+      "loss": 2.2512,
+      "step": 1348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.19598892253268746,
+      "learning_rate": 0.00033563681664462155,
+      "loss": 2.1926,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.197308707696409,
+      "learning_rate": 0.000334935579951755,
+      "loss": 2.2156,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1793875776965209,
+      "learning_rate": 0.0003342347074525578,
+      "loss": 2.2778,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1811369925364848,
+      "learning_rate": 0.0003335342006934161,
+      "loss": 2.272,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.19513199396328343,
+      "learning_rate": 0.00033283406121990914,
+      "loss": 2.1905,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.20070695255916304,
+      "learning_rate": 0.0003321342905768057,
+      "loss": 2.2238,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.29299169256740776,
+      "learning_rate": 0.00033143489030806086,
+      "loss": 2.2868,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1857749396649075,
+      "learning_rate": 0.00033073586195681227,
+      "loss": 2.3183,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.23241927742011315,
+      "learning_rate": 0.00033003720706537736,
+      "loss": 2.1935,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1871646743527776,
+      "learning_rate": 0.0003293389271752492,
+      "loss": 2.2085,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.20041938281263647,
+      "learning_rate": 0.00032864102382709374,
+      "loss": 2.2322,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.17434633835093435,
+      "learning_rate": 0.000327943498560746,
+      "loss": 2.1981,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.17996352870957807,
+      "learning_rate": 0.00032724635291520694,
+      "loss": 2.2281,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1910484386383553,
+      "learning_rate": 0.00032654958842863967,
+      "loss": 2.2664,
+      "step": 1362
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1863680563344538,
+      "learning_rate": 0.0003258532066383667,
+      "loss": 2.3068,
+      "step": 1363
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2214543974915027,
+      "learning_rate": 0.000325157209080866,
+      "loss": 2.229,
+      "step": 1364
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1996595869906418,
+      "learning_rate": 0.00032446159729176743,
+      "loss": 2.2015,
+      "step": 1365
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18695596002789788,
+      "learning_rate": 0.0003237663728058502,
+      "loss": 2.408,
+      "step": 1366
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18537889814260455,
+      "learning_rate": 0.0003230715371570389,
+      "loss": 2.3043,
+      "step": 1367
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.20161357499972068,
+      "learning_rate": 0.00032237709187839996,
+      "loss": 2.2627,
+      "step": 1368
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.17770813522381917,
+      "learning_rate": 0.0003216830385021388,
+      "loss": 2.1704,
+      "step": 1369
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1882891867328501,
+      "learning_rate": 0.0003209893785595959,
+      "loss": 2.1541,
+      "step": 1370
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18148336221915956,
+      "learning_rate": 0.00032029611358124366,
+      "loss": 2.1277,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22063536784742852,
+      "learning_rate": 0.00031960324509668336,
+      "loss": 2.2815,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.19274576882109398,
+      "learning_rate": 0.0003189107746346412,
+      "loss": 2.274,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1783566472982605,
+      "learning_rate": 0.0003182187037229653,
+      "loss": 2.2153,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18987526139622726,
+      "learning_rate": 0.0003175270338886221,
+      "loss": 2.206,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1918301507595119,
+      "learning_rate": 0.00031683576665769345,
+      "loss": 2.278,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.19405913021661603,
+      "learning_rate": 0.0003161449035553724,
+      "loss": 2.289,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.21268375964761443,
+      "learning_rate": 0.00031545444610596077,
+      "loss": 2.2073,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18786091584820366,
+      "learning_rate": 0.000314764395832865,
+      "loss": 2.2418,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1920977801485012,
+      "learning_rate": 0.0003140747542585934,
+      "loss": 2.2774,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18637431001456997,
+      "learning_rate": 0.00031338552290475266,
+      "loss": 2.2658,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.18751012684114202,
+      "learning_rate": 0.00031269670329204396,
+      "loss": 2.308,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.179257558551878,
+      "learning_rate": 0.0003120082969402604,
+      "loss": 2.1993,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.19745655258571612,
+      "learning_rate": 0.00031132030536828314,
+      "loss": 2.2377,
+      "step": 1384
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.20481796168631677,
+      "learning_rate": 0.00031063273009407805,
+      "loss": 2.1236,
+      "step": 1385
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.17455041855029613,
+      "learning_rate": 0.00030994557263469265,
+      "loss": 2.2097,
+      "step": 1386
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2000083857724696,
+      "learning_rate": 0.0003092588345062526,
+      "loss": 2.3426,
+      "step": 1387
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.22740707474544689,
+      "learning_rate": 0.0003085725172239582,
+      "loss": 2.2301,
+      "step": 1388
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.21101889302825252,
+      "learning_rate": 0.0003078866223020815,
+      "loss": 2.1959,
+      "step": 1389
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18547495425636396,
+      "learning_rate": 0.0003072011512539624,
+      "loss": 2.3096,
+      "step": 1390
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.189292496462095,
+      "learning_rate": 0.00030651610559200574,
+      "loss": 2.3003,
+      "step": 1391
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.19720870200104826,
+      "learning_rate": 0.00030583148682767757,
+      "loss": 2.2529,
+      "step": 1392
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.23025911409106464,
+      "learning_rate": 0.00030514729647150243,
+      "loss": 2.2748,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18967007909460432,
+      "learning_rate": 0.0003044635360330592,
+      "loss": 2.254,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18041237489552786,
+      "learning_rate": 0.00030378020702097845,
+      "loss": 2.3182,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1916160109152058,
+      "learning_rate": 0.000303097310942939,
+      "loss": 2.1872,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1986127504362061,
+      "learning_rate": 0.0003024148493056641,
+      "loss": 2.2702,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2144703637681908,
+      "learning_rate": 0.00030173282361491865,
+      "loss": 2.2711,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1940810374355363,
+      "learning_rate": 0.0003010512353755057,
+      "loss": 2.2136,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1792079359632857,
+      "learning_rate": 0.00030037008609126313,
+      "loss": 2.2427,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.19538946019120856,
+      "learning_rate": 0.0002996893772650602,
+      "loss": 2.1542,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18738676868155263,
+      "learning_rate": 0.0002990091103987945,
+      "loss": 2.259,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1862799861232553,
+      "learning_rate": 0.0002983292869933886,
+      "loss": 2.2193,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18993300120099818,
+      "learning_rate": 0.0002976499085487862,
+      "loss": 2.2386,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1999741007162706,
+      "learning_rate": 0.00029697097656394963,
+      "loss": 2.1155,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.21520654881067242,
+      "learning_rate": 0.00029629249253685595,
+      "loss": 2.2783,
+      "step": 1406
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20289509460833827,
+      "learning_rate": 0.00029561445796449416,
+      "loss": 2.2414,
+      "step": 1407
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20052060521708598,
+      "learning_rate": 0.0002949368743428612,
+      "loss": 2.2342,
+      "step": 1408
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18517905426925613,
+      "learning_rate": 0.0002942597431669593,
+      "loss": 2.1882,
+      "step": 1409
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1748698672646149,
+      "learning_rate": 0.0002935830659307924,
+      "loss": 2.1729,
+      "step": 1410
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20515858688000596,
+      "learning_rate": 0.0002929068441273629,
+      "loss": 2.2973,
+      "step": 1411
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.21035373044550434,
+      "learning_rate": 0.0002922310792486681,
+      "loss": 2.2552,
+      "step": 1412
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18171867688596213,
+      "learning_rate": 0.00029155577278569745,
+      "loss": 2.164,
+      "step": 1413
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19142554860981092,
+      "learning_rate": 0.00029088092622842895,
+      "loss": 2.3058,
+      "step": 1414
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.22585574415724136,
+      "learning_rate": 0.00029020654106582544,
+      "loss": 2.2967,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20839197388943148,
+      "learning_rate": 0.0002895326187858326,
+      "loss": 2.1379,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18963445812626922,
+      "learning_rate": 0.00028885916087537377,
+      "loss": 2.2517,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.17964683700955608,
+      "learning_rate": 0.00028818616882034877,
+      "loss": 2.1998,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18310176037805095,
+      "learning_rate": 0.0002875136441056286,
+      "loss": 2.1636,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19933472339432423,
+      "learning_rate": 0.000286841588215054,
+      "loss": 2.2215,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18944002179010844,
+      "learning_rate": 0.0002861700026314308,
+      "loss": 2.2052,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20397469954379419,
+      "learning_rate": 0.00028549888883652686,
+      "loss": 2.3307,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19102753565566183,
+      "learning_rate": 0.00028482824831107,
+      "loss": 2.2958,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19255815578118618,
+      "learning_rate": 0.000284158082534743,
+      "loss": 2.1523,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.20179182221086048,
+      "learning_rate": 0.00028348839298618177,
+      "loss": 2.247,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.19373487368374542,
+      "learning_rate": 0.0002828191811429709,
+      "loss": 2.2765,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1895313183213899,
+      "learning_rate": 0.00028215044848164164,
+      "loss": 2.1476,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18845367250090436,
+      "learning_rate": 0.00028148219647766747,
+      "loss": 2.2103,
+      "step": 1428
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1914251447677708,
+      "learning_rate": 0.00028081442660546124,
+      "loss": 2.2424,
+      "step": 1429
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1758667348251855,
+      "learning_rate": 0.0002801471403383728,
+      "loss": 2.1285,
+      "step": 1430
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.18453065999171286,
+      "learning_rate": 0.00027948033914868415,
+      "loss": 2.238,
+      "step": 1431
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19511681773065126,
+      "learning_rate": 0.00027881402450760775,
+      "loss": 2.1317,
+      "step": 1432
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.17179269480164785,
+      "learning_rate": 0.00027814819788528165,
+      "loss": 2.1735,
+      "step": 1433
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.21864748351254626,
+      "learning_rate": 0.00027748286075076836,
+      "loss": 2.2219,
+      "step": 1434
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.18213288357270352,
+      "learning_rate": 0.00027681801457204937,
+      "loss": 2.178,
+      "step": 1435
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1892362563238357,
+      "learning_rate": 0.00027615366081602306,
+      "loss": 2.2616,
+      "step": 1436
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2315352583424298,
+      "learning_rate": 0.0002754898009485021,
+      "loss": 2.2967,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.20177778596450854,
+      "learning_rate": 0.0002748264364342085,
+      "loss": 2.1796,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.22340149496506595,
+      "learning_rate": 0.00027416356873677204,
+      "loss": 2.2529,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2078209518914614,
+      "learning_rate": 0.0002735011993187258,
+      "loss": 2.2154,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.20002581821327756,
+      "learning_rate": 0.0002728393296415042,
+      "loss": 2.2803,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2046058389672194,
+      "learning_rate": 0.00027217796116543817,
+      "loss": 2.2288,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19135021347450232,
+      "learning_rate": 0.0002715170953497532,
+      "loss": 2.2098,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.17985280756765962,
+      "learning_rate": 0.00027085673365256614,
+      "loss": 2.2436,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2083308502417597,
+      "learning_rate": 0.00027019687753088075,
+      "loss": 2.1847,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2061242343323679,
+      "learning_rate": 0.00026953752844058597,
+      "loss": 2.2943,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19412827683592032,
+      "learning_rate": 0.0002688786878364516,
+      "loss": 2.2281,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.18666790069883518,
+      "learning_rate": 0.00026822035717212597,
+      "loss": 2.2776,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.20049123909505298,
+      "learning_rate": 0.00026756253790013193,
+      "loss": 2.2565,
+      "step": 1449
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.17328967327141262,
+      "learning_rate": 0.0002669052314718641,
+      "loss": 2.2974,
+      "step": 1450
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1784440959640575,
+      "learning_rate": 0.0002662484393375855,
+      "loss": 2.2682,
+      "step": 1451
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18739598589064754,
+      "learning_rate": 0.00026559216294642446,
+      "loss": 2.2617,
+      "step": 1452
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.17769180545133384,
+      "learning_rate": 0.0002649364037463718,
+      "loss": 2.2972,
+      "step": 1453
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18685724740454043,
+      "learning_rate": 0.0002642811631842764,
+      "loss": 2.2233,
+      "step": 1454
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.20997638800080842,
+      "learning_rate": 0.0002636264427058439,
+      "loss": 2.1861,
+      "step": 1455
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18851332189013073,
+      "learning_rate": 0.00026297224375563123,
+      "loss": 2.3069,
+      "step": 1456
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.21763711466799426,
+      "learning_rate": 0.00026231856777704575,
+      "loss": 2.1716,
+      "step": 1457
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.33604905222996384,
+      "learning_rate": 0.00026166541621234026,
+      "loss": 2.2342,
+      "step": 1458
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.26924801805362825,
+      "learning_rate": 0.00026101279050261045,
+      "loss": 2.2035,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2128950809075568,
+      "learning_rate": 0.00026036069208779247,
+      "loss": 2.2379,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3201951134786103,
+      "learning_rate": 0.0002597091224066581,
+      "loss": 2.2109,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2314934489626402,
+      "learning_rate": 0.00025905808289681365,
+      "loss": 2.1433,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.22030784569205167,
+      "learning_rate": 0.0002584075749946946,
+      "loss": 2.2151,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18558642645304532,
+      "learning_rate": 0.00025775760013556424,
+      "loss": 2.2211,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1942349599815909,
+      "learning_rate": 0.0002571081597535095,
+      "loss": 2.2139,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.304175682701553,
+      "learning_rate": 0.00025645925528143776,
+      "loss": 2.2207,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.26056394424861723,
+      "learning_rate": 0.0002558108881510747,
+      "loss": 2.1943,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.19850786391687902,
+      "learning_rate": 0.00025516305979295963,
+      "loss": 2.2624,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2503943561817623,
+      "learning_rate": 0.0002545157716364439,
+      "loss": 2.2216,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.24051158136163897,
+      "learning_rate": 0.00025386902510968624,
+      "loss": 2.2161,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.29955666371913825,
+      "learning_rate": 0.00025322282163965095,
+      "loss": 2.2801,
+      "step": 1471
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1766186329171908,
+      "learning_rate": 0.00025257716265210384,
+      "loss": 2.1943,
+      "step": 1472
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.20273439610205923,
+      "learning_rate": 0.0002519320495716091,
+      "loss": 2.3143,
+      "step": 1473
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.19405165341480968,
+      "learning_rate": 0.00025128748382152716,
+      "loss": 2.1951,
+      "step": 1474
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2711970556161782,
+      "learning_rate": 0.00025064346682401016,
+      "loss": 2.261,
+      "step": 1475
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3395786486518534,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 2.1434,
+      "step": 1476
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18208708161603418,
+      "learning_rate": 0.0002493570847692246,
+      "loss": 2.2851,
+      "step": 1477
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.22639657759011805,
+      "learning_rate": 0.00024871472255019424,
+      "loss": 2.2174,
+      "step": 1478
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.27989288403569,
+      "learning_rate": 0.00024807291476019994,
+      "loss": 2.259,
+      "step": 1479
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24917959388693353,
+      "learning_rate": 0.00024743166281530877,
+      "loss": 2.1206,
+      "step": 1480
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.19333691959358187,
+      "learning_rate": 0.000246790968130362,
+      "loss": 2.1526,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.22376378951731024,
+      "learning_rate": 0.0002461508321189706,
+      "loss": 2.1425,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.20320233380512162,
+      "learning_rate": 0.00024551125619351385,
+      "loss": 2.2239,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3624227986507542,
+      "learning_rate": 0.00024487224176513453,
+      "loss": 2.2069,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.23020888990948654,
+      "learning_rate": 0.0002442337902437365,
+      "loss": 2.1234,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2120820611283303,
+      "learning_rate": 0.0002435959030379824,
+      "loss": 2.3385,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18619653761939353,
+      "learning_rate": 0.00024295858155528888,
+      "loss": 2.2005,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.25916498473478644,
+      "learning_rate": 0.00024232182720182523,
+      "loss": 2.2734,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24031492633414528,
+      "learning_rate": 0.00024168564138250855,
+      "loss": 2.2926,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18899670681697164,
+      "learning_rate": 0.00024105002550100246,
+      "loss": 2.1873,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18926732100237362,
+      "learning_rate": 0.00024041498095971254,
+      "loss": 2.2124,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.18302231516414166,
+      "learning_rate": 0.0002397805091597835,
+      "loss": 2.193,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1877635024827204,
+      "learning_rate": 0.0002391466115010973,
+      "loss": 2.3009,
+      "step": 1493
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.22776005416556602,
+      "learning_rate": 0.00023851328938226808,
+      "loss": 2.1645,
+      "step": 1494
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18093223572753056,
+      "learning_rate": 0.00023788054420064109,
+      "loss": 2.2668,
+      "step": 1495
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.20672138956261293,
+      "learning_rate": 0.00023724837735228773,
+      "loss": 2.2672,
+      "step": 1496
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.19688890269915277,
+      "learning_rate": 0.00023661679023200422,
+      "loss": 2.1192,
+      "step": 1497
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2112176393348204,
+      "learning_rate": 0.00023598578423330714,
+      "loss": 2.1721,
+      "step": 1498
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2049004971652839,
+      "learning_rate": 0.00023535536074843083,
+      "loss": 2.2084,
+      "step": 1499
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.20996757462415702,
+      "learning_rate": 0.00023472552116832502,
+      "loss": 2.2519,
+      "step": 1500
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.24012963549409447,
+      "learning_rate": 0.0002340962668826503,
+      "loss": 2.1776,
+      "step": 1501
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.23219992976677206,
+      "learning_rate": 0.00023346759927977663,
+      "loss": 2.2641,
+      "step": 1502
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.19027393249576735,
+      "learning_rate": 0.0002328395197467789,
+      "loss": 2.2316,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18114565370265182,
+      "learning_rate": 0.00023221202966943515,
+      "loss": 2.1765,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1963232943881231,
+      "learning_rate": 0.0002315851304322223,
+      "loss": 2.3062,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18620842633782264,
+      "learning_rate": 0.0002309588234183137,
+      "loss": 2.2425,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1814893136825068,
+      "learning_rate": 0.00023033311000957653,
+      "loss": 2.1954,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.19736821813135216,
+      "learning_rate": 0.00022970799158656758,
+      "loss": 2.1966,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1961562580393141,
+      "learning_rate": 0.0002290834695285316,
+      "loss": 2.2678,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18527695410564257,
+      "learning_rate": 0.00022845954521339678,
+      "loss": 2.1979,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.21667467879384777,
+      "learning_rate": 0.0002278362200177732,
+      "loss": 2.2624,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.21346141189985268,
+      "learning_rate": 0.00022721349531694852,
+      "loss": 2.2377,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.19281292972224368,
+      "learning_rate": 0.0002265913724848855,
+      "loss": 2.2496,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1803697689486273,
+      "learning_rate": 0.00022596985289421946,
+      "loss": 2.1063,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1825434056396797,
+      "learning_rate": 0.00022534893791625405,
+      "loss": 2.152,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2600537184465599,
+      "learning_rate": 0.00022472862892095968,
+      "loss": 2.3034,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.19650118922496396,
+      "learning_rate": 0.00022410892727696896,
+      "loss": 2.2242,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2069140068687479,
+      "learning_rate": 0.0002234898343515751,
+      "loss": 2.2107,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20020060944353182,
+      "learning_rate": 0.00022287135151072792,
+      "loss": 2.1635,
+      "step": 1519
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.21591449089316175,
+      "learning_rate": 0.00022225348011903096,
+      "loss": 2.2733,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1972273480033043,
+      "learning_rate": 0.0002216362215397393,
+      "loss": 2.2402,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1851253877143652,
+      "learning_rate": 0.00022101957713475522,
+      "loss": 2.1626,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.19786856956763427,
+      "learning_rate": 0.00022040354826462666,
+      "loss": 2.1804,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1828524978322351,
+      "learning_rate": 0.0002197881362885426,
+      "loss": 2.2766,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2141707036014695,
+      "learning_rate": 0.0002191733425643318,
+      "loss": 2.1837,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2168531970815879,
+      "learning_rate": 0.00021855916844845826,
+      "loss": 2.2433,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20233740037321105,
+      "learning_rate": 0.00021794561529601898,
+      "loss": 2.098,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1822423276508221,
+      "learning_rate": 0.00021733268446074138,
+      "loss": 2.1459,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20588598327158075,
+      "learning_rate": 0.00021672037729497917,
+      "loss": 2.2571,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20405876785711166,
+      "learning_rate": 0.0002161086951497106,
+      "loss": 2.2175,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.20425237961336562,
+      "learning_rate": 0.00021549763937453442,
+      "loss": 2.1977,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.18484668034465665,
+      "learning_rate": 0.00021488721131766736,
+      "loss": 2.3079,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.18195405135316484,
+      "learning_rate": 0.00021427741232594183,
+      "loss": 2.2581,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.19934623738919066,
+      "learning_rate": 0.0002136682437448013,
+      "loss": 2.3151,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1761090674293767,
+      "learning_rate": 0.0002130597069182994,
+      "loss": 2.2707,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.18966203461235204,
+      "learning_rate": 0.0002124518031890948,
+      "loss": 2.1894,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.17705196092563874,
+      "learning_rate": 0.0002118445338984502,
+      "loss": 2.0647,
+      "step": 1537
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19188595382024975,
+      "learning_rate": 0.00021123790038622808,
+      "loss": 2.1992,
+      "step": 1538
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1946295819693327,
+      "learning_rate": 0.0002106319039908879,
+      "loss": 2.2271,
+      "step": 1539
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18664364365603164,
+      "learning_rate": 0.00021002654604948412,
+      "loss": 2.1954,
+      "step": 1540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19524868971213807,
+      "learning_rate": 0.00020942182789766172,
+      "loss": 2.2256,
+      "step": 1541
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.214082147660546,
+      "learning_rate": 0.00020881775086965492,
+      "loss": 2.2055,
+      "step": 1542
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.21404205150775732,
+      "learning_rate": 0.00020821431629828246,
+      "loss": 2.2571,
+      "step": 1543
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18260312053171399,
+      "learning_rate": 0.00020761152551494643,
+      "loss": 2.2115,
+      "step": 1544
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.20599866488769605,
+      "learning_rate": 0.00020700937984962798,
+      "loss": 2.2872,
+      "step": 1545
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18836850248849757,
+      "learning_rate": 0.0002064078806308848,
+      "loss": 2.1471,
+      "step": 1546
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.22931003477469553,
+      "learning_rate": 0.00020580702918584882,
+      "loss": 2.2113,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2667415549362884,
+      "learning_rate": 0.000205206826840222,
+      "loss": 2.26,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2459939185179412,
+      "learning_rate": 0.0002046072749182751,
+      "loss": 2.2062,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1928491364931795,
+      "learning_rate": 0.00020400837474284273,
+      "loss": 2.2544,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1956199814566195,
+      "learning_rate": 0.0002034101276353224,
+      "loss": 2.3059,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19420135338028036,
+      "learning_rate": 0.00020281253491567027,
+      "loss": 2.2213,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.17915577332501773,
+      "learning_rate": 0.0002022155979023984,
+      "loss": 2.1965,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.20189144657011465,
+      "learning_rate": 0.000201619317912573,
+      "loss": 2.2754,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19249308224512263,
+      "learning_rate": 0.00020102369626180962,
+      "loss": 2.2355,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1956005781510566,
+      "learning_rate": 0.0002004287342642721,
+      "loss": 2.1826,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19535225604423562,
+      "learning_rate": 0.00019983443323266824,
+      "loss": 2.2161,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18821594441648345,
+      "learning_rate": 0.00019924079447824805,
+      "loss": 2.2628,
+      "step": 1558
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18541912165572785,
+      "learning_rate": 0.00019864781931079977,
+      "loss": 2.1582,
+      "step": 1559
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.19739640589124663,
+      "learning_rate": 0.00019805550903864773,
+      "loss": 2.2898,
+      "step": 1560
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.17805589213959752,
+      "learning_rate": 0.00019746386496864948,
+      "loss": 2.1451,
+      "step": 1561
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.19098289775592805,
+      "learning_rate": 0.00019687288840619226,
+      "loss": 2.1777,
+      "step": 1562
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2099962397444676,
+      "learning_rate": 0.0001962825806551911,
+      "loss": 2.2387,
+      "step": 1563
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.21564978665286988,
+      "learning_rate": 0.0001956929430180846,
+      "loss": 2.2113,
+      "step": 1564
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.21208826286476679,
+      "learning_rate": 0.00019510397679583374,
+      "loss": 2.2778,
+      "step": 1565
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.23282160200171023,
+      "learning_rate": 0.0001945156832879174,
+      "loss": 2.2405,
+      "step": 1566
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.18909088214296713,
+      "learning_rate": 0.00019392806379233036,
+      "loss": 2.1233,
+      "step": 1567
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.18878499736645987,
+      "learning_rate": 0.00019334111960558065,
+      "loss": 2.2506,
+      "step": 1568
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2234964171369105,
+      "learning_rate": 0.00019275485202268573,
+      "loss": 2.2678,
+      "step": 1569
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20470776775386346,
+      "learning_rate": 0.00019216926233717085,
+      "loss": 2.1672,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.22275309748918465,
+      "learning_rate": 0.00019158435184106498,
+      "loss": 2.259,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.27014707622836,
+      "learning_rate": 0.00019100012182489905,
+      "loss": 2.2463,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.21767159587327195,
+      "learning_rate": 0.00019041657357770226,
+      "loss": 2.2115,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20202068925949876,
+      "learning_rate": 0.00018983370838699943,
+      "loss": 2.211,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20342210861943552,
+      "learning_rate": 0.00018925152753880892,
+      "loss": 2.2409,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1821893125466419,
+      "learning_rate": 0.00018867003231763847,
+      "loss": 2.3182,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20932059582293366,
+      "learning_rate": 0.00018808922400648375,
+      "loss": 2.3418,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20712601550413887,
+      "learning_rate": 0.00018750910388682428,
+      "loss": 2.2584,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.19180017770030586,
+      "learning_rate": 0.00018692967323862125,
+      "loss": 2.185,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.18183780880952385,
+      "learning_rate": 0.00018635093334031517,
+      "loss": 2.2351,
+      "step": 1580
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.20306791528719403,
+      "learning_rate": 0.00018577288546882165,
+      "loss": 2.2009,
+      "step": 1581
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1867125106669833,
+      "learning_rate": 0.00018519553089953023,
+      "loss": 2.2079,
+      "step": 1582
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24180331343026557,
+      "learning_rate": 0.0001846188709063001,
+      "loss": 2.1587,
+      "step": 1583
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19897304441323937,
+      "learning_rate": 0.00018404290676145857,
+      "loss": 2.2069,
+      "step": 1584
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.18028708175954042,
+      "learning_rate": 0.00018346763973579722,
+      "loss": 2.1999,
+      "step": 1585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1788102325546695,
+      "learning_rate": 0.00018289307109856939,
+      "loss": 2.3119,
+      "step": 1586
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.25970919726063385,
+      "learning_rate": 0.0001823192021174882,
+      "loss": 2.1805,
+      "step": 1587
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3230291821157351,
+      "learning_rate": 0.0001817460340587223,
+      "loss": 2.2451,
+      "step": 1588
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.22237760149571476,
+      "learning_rate": 0.00018117356818689445,
+      "loss": 2.2716,
+      "step": 1589
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19861589483443368,
+      "learning_rate": 0.00018060180576507756,
+      "loss": 2.2241,
+      "step": 1590
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2027138907481531,
+      "learning_rate": 0.00018003074805479313,
+      "loss": 2.1104,
+      "step": 1591
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.18223812240588746,
+      "learning_rate": 0.00017946039631600724,
+      "loss": 2.229,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19096538687424766,
+      "learning_rate": 0.00017889075180712837,
+      "loss": 2.2384,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24099476424481772,
+      "learning_rate": 0.00017832181578500512,
+      "loss": 2.1888,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19173945259739875,
+      "learning_rate": 0.0001777535895049221,
+      "loss": 2.1742,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2144360021734371,
+      "learning_rate": 0.0001771860742205988,
+      "loss": 2.2156,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19787805521022947,
+      "learning_rate": 0.00017661927118418525,
+      "loss": 2.27,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19645527472942198,
+      "learning_rate": 0.00017605318164626066,
+      "loss": 2.2824,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1933763195917297,
+      "learning_rate": 0.00017548780685582949,
+      "loss": 2.2063,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.20359093024569735,
+      "learning_rate": 0.00017492314806031922,
+      "loss": 2.3112,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2448882109095324,
+      "learning_rate": 0.00017435920650557806,
+      "loss": 2.2054,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.23499228007922196,
+      "learning_rate": 0.00017379598343587112,
+      "loss": 2.1748,
+      "step": 1602
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1866572012738809,
+      "learning_rate": 0.00017323348009387878,
+      "loss": 2.1825,
+      "step": 1603
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.19465957865531244,
+      "learning_rate": 0.0001726716977206929,
+      "loss": 2.1952,
+      "step": 1604
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.20300670254647896,
+      "learning_rate": 0.00017211063755581525,
+      "loss": 2.187,
+      "step": 1605
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.22383187660444911,
+      "learning_rate": 0.0001715503008371536,
+      "loss": 2.2442,
+      "step": 1606
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.22880211581630674,
+      "learning_rate": 0.0001709906888010196,
+      "loss": 2.1344,
+      "step": 1607
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.19066504318640934,
+      "learning_rate": 0.00017043180268212638,
+      "loss": 2.2903,
+      "step": 1608
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.180144463851334,
+      "learning_rate": 0.00016987364371358481,
+      "loss": 2.1999,
+      "step": 1609
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.19929896838028124,
+      "learning_rate": 0.00016931621312690214,
+      "loss": 2.1486,
+      "step": 1610
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.21491407979435423,
+      "learning_rate": 0.00016875951215197777,
+      "loss": 2.2174,
+      "step": 1611
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.19428066108897044,
+      "learning_rate": 0.00016820354201710214,
+      "loss": 2.2611,
+      "step": 1612
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.17798574500816167,
+      "learning_rate": 0.00016764830394895203,
+      "loss": 2.0659,
+      "step": 1613
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.21506942949471067,
+      "learning_rate": 0.00016709379917259027,
+      "loss": 2.1912,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2207737704747922,
+      "learning_rate": 0.00016654002891146091,
+      "loss": 2.2741,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1953707785874773,
+      "learning_rate": 0.00016598699438738764,
+      "loss": 2.2494,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.17814475841864036,
+      "learning_rate": 0.00016543469682057105,
+      "loss": 2.3114,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.18975162989295133,
+      "learning_rate": 0.00016488313742958526,
+      "loss": 2.2936,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.21606112399882343,
+      "learning_rate": 0.00016433231743137646,
+      "loss": 2.1237,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1964138622007884,
+      "learning_rate": 0.0001637822380412584,
+      "loss": 2.2238,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1768568605076697,
+      "learning_rate": 0.00016323290047291195,
+      "loss": 2.1795,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.17697122000960472,
+      "learning_rate": 0.0001626843059383803,
+      "loss": 2.2457,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.20133139460135735,
+      "learning_rate": 0.00016213645564806752,
+      "loss": 2.2088,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.22371244223071798,
+      "learning_rate": 0.0001615893508107359,
+      "loss": 2.1771,
+      "step": 1624
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.21846873722225582,
+      "learning_rate": 0.00016104299263350252,
+      "loss": 2.2472,
+      "step": 1625
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.193463227000711,
+      "learning_rate": 0.00016049738232183758,
+      "loss": 2.1808,
+      "step": 1626
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2107648637239321,
+      "learning_rate": 0.0001599525210795606,
+      "loss": 2.2614,
+      "step": 1627
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18837238583996224,
+      "learning_rate": 0.00015940841010883889,
+      "loss": 2.1317,
+      "step": 1628
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18255780164595026,
+      "learning_rate": 0.00015886505061018413,
+      "loss": 2.1659,
+      "step": 1629
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19917463614874878,
+      "learning_rate": 0.0001583224437824498,
+      "loss": 2.1712,
+      "step": 1630
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18999035519797716,
+      "learning_rate": 0.0001577805908228293,
+      "loss": 2.1518,
+      "step": 1631
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19678592738515135,
+      "learning_rate": 0.00015723949292685191,
+      "loss": 2.1471,
+      "step": 1632
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18147378911278336,
+      "learning_rate": 0.0001566991512883818,
+      "loss": 2.1206,
+      "step": 1633
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2793719348265906,
+      "learning_rate": 0.00015615956709961378,
+      "loss": 2.133,
+      "step": 1634
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2164462871765308,
+      "learning_rate": 0.00015562074155107215,
+      "loss": 2.163,
+      "step": 1635
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19753690648887595,
+      "learning_rate": 0.0001550826758316068,
+      "loss": 2.2178,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19249246869373612,
+      "learning_rate": 0.00015454537112839122,
+      "loss": 2.1445,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19607124438698734,
+      "learning_rate": 0.00015400882862692033,
+      "loss": 2.1911,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.23964537019996135,
+      "learning_rate": 0.00015347304951100665,
+      "loss": 2.2094,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.17592119810258908,
+      "learning_rate": 0.00015293803496277907,
+      "loss": 2.2175,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1861908918131432,
+      "learning_rate": 0.00015240378616267886,
+      "loss": 2.1775,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.20125455766328215,
+      "learning_rate": 0.00015187030428945843,
+      "loss": 2.1232,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18220011769452754,
+      "learning_rate": 0.0001513375905201776,
+      "loss": 2.2761,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.20401137958201015,
+      "learning_rate": 0.00015080564603020142,
+      "loss": 2.2621,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.20725862344826565,
+      "learning_rate": 0.0001502744719931982,
+      "loss": 2.2654,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.19131166980598346,
+      "learning_rate": 0.00014974406958113558,
+      "loss": 2.1841,
+      "step": 1646
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.18763084871233615,
+      "learning_rate": 0.00014921443996427947,
+      "loss": 2.1816,
+      "step": 1647
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20007590157905134,
+      "learning_rate": 0.0001486855843111901,
+      "loss": 2.2322,
+      "step": 1648
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.18486852230259052,
+      "learning_rate": 0.0001481575037887201,
+      "loss": 2.244,
+      "step": 1649
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.24071422489637626,
+      "learning_rate": 0.00014763019956201253,
+      "loss": 2.2352,
+      "step": 1650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.18850086248691222,
+      "learning_rate": 0.0001471036727944966,
+      "loss": 2.1758,
+      "step": 1651
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20958107722806976,
+      "learning_rate": 0.0001465779246478872,
+      "loss": 2.1242,
+      "step": 1652
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19816103135578358,
+      "learning_rate": 0.00014605295628218045,
+      "loss": 2.2191,
+      "step": 1653
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19767401454404324,
+      "learning_rate": 0.0001455287688556527,
+      "loss": 2.1659,
+      "step": 1654
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20170741661861794,
+      "learning_rate": 0.00014500536352485673,
+      "loss": 2.1583,
+      "step": 1655
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19451970892839868,
+      "learning_rate": 0.00014448274144461965,
+      "loss": 2.1531,
+      "step": 1656
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1983971592885169,
+      "learning_rate": 0.00014396090376804112,
+      "loss": 2.2808,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20036668100112304,
+      "learning_rate": 0.00014343985164648926,
+      "loss": 2.1621,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.17956418346262712,
+      "learning_rate": 0.00014291958622959973,
+      "loss": 2.1108,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19163779240422277,
+      "learning_rate": 0.00014240010866527176,
+      "loss": 2.2189,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1849911819141893,
+      "learning_rate": 0.00014188142009966686,
+      "loss": 2.1158,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.18703705509274776,
+      "learning_rate": 0.0001413635216772053,
+      "loss": 2.1498,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.21083153137799815,
+      "learning_rate": 0.000140846414540564,
+      "loss": 2.146,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20404962877498423,
+      "learning_rate": 0.00014033009983067452,
+      "loss": 2.1202,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1882720781745955,
+      "learning_rate": 0.00013981457868671927,
+      "loss": 2.2111,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2288132013418471,
+      "learning_rate": 0.0001392998522461305,
+      "loss": 2.3476,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.23193204455004215,
+      "learning_rate": 0.00013878592164458635,
+      "loss": 2.1672,
+      "step": 1667
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.18286888327065542,
+      "learning_rate": 0.00013827278801600978,
+      "loss": 2.2505,
+      "step": 1668
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.22110820889936017,
+      "learning_rate": 0.0001377604524925647,
+      "loss": 2.3063,
+      "step": 1669
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18268790456642167,
+      "learning_rate": 0.00013724891620465424,
+      "loss": 2.1805,
+      "step": 1670
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18300284011449935,
+      "learning_rate": 0.0001367381802809185,
+      "loss": 2.0894,
+      "step": 1671
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19340820190300712,
+      "learning_rate": 0.00013622824584823113,
+      "loss": 2.1876,
+      "step": 1672
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19451855855276823,
+      "learning_rate": 0.00013571911403169795,
+      "loss": 2.1924,
+      "step": 1673
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.20061107686542368,
+      "learning_rate": 0.0001352107859546533,
+      "loss": 2.1782,
+      "step": 1674
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18293368740032123,
+      "learning_rate": 0.00013470326273865886,
+      "loss": 2.1969,
+      "step": 1675
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19105161403968232,
+      "learning_rate": 0.00013419654550349985,
+      "loss": 2.2505,
+      "step": 1676
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18329848000966667,
+      "learning_rate": 0.00013369063536718346,
+      "loss": 2.193,
+      "step": 1677
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19918211680274883,
+      "learning_rate": 0.00013318553344593632,
+      "loss": 2.1907,
+      "step": 1678
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18306875277000662,
+      "learning_rate": 0.00013268124085420136,
+      "loss": 2.2271,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1969061388149091,
+      "learning_rate": 0.0001321777587046364,
+      "loss": 2.2799,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1909693595505324,
+      "learning_rate": 0.00013167508810811059,
+      "loss": 2.2479,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2187907272156998,
+      "learning_rate": 0.0001311732301737029,
+      "loss": 2.231,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19151448184451567,
+      "learning_rate": 0.0001306721860086991,
+      "loss": 2.232,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.24237967985760334,
+      "learning_rate": 0.00013017195671858928,
+      "loss": 2.1274,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.22313216538361463,
+      "learning_rate": 0.0001296725434070661,
+      "loss": 2.2988,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1871589346127272,
+      "learning_rate": 0.00012917394717602121,
+      "loss": 2.226,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2031560102603832,
+      "learning_rate": 0.00012867616912554426,
+      "loss": 2.1,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1855331389357105,
+      "learning_rate": 0.00012817921035391882,
+      "loss": 2.2082,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.23430526836707816,
+      "learning_rate": 0.00012768307195762168,
+      "loss": 2.2291,
+      "step": 1689
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18490800513601963,
+      "learning_rate": 0.00012718775503131908,
+      "loss": 2.1937,
+      "step": 1690
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1943216741961921,
+      "learning_rate": 0.0001266932606678646,
+      "loss": 2.2073,
+      "step": 1691
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.18084687237287614,
+      "learning_rate": 0.00012619958995829756,
+      "loss": 2.2264,
+      "step": 1692
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1855015575812618,
+      "learning_rate": 0.0001257067439918394,
+      "loss": 2.15,
+      "step": 1693
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.200991730166067,
+      "learning_rate": 0.00012521472385589234,
+      "loss": 2.164,
+      "step": 1694
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.26668447006870294,
+      "learning_rate": 0.00012472353063603626,
+      "loss": 2.1674,
+      "step": 1695
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.22909325452640183,
+      "learning_rate": 0.0001242331654160263,
+      "loss": 2.1367,
+      "step": 1696
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1833503891920896,
+      "learning_rate": 0.0001237436292777914,
+      "loss": 2.2136,
+      "step": 1697
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.20651070209236647,
+      "learning_rate": 0.00012325492330143061,
+      "loss": 2.2119,
+      "step": 1698
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.19682019786365978,
+      "learning_rate": 0.00012276704856521175,
+      "loss": 2.1082,
+      "step": 1699
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.20335231082644095,
+      "learning_rate": 0.00012228000614556816,
+      "loss": 2.258,
+      "step": 1700
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21082622693188877,
+      "learning_rate": 0.00012179379711709738,
+      "loss": 2.2404,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21016183505928604,
+      "learning_rate": 0.0001213084225525577,
+      "loss": 2.2428,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.19984774299750688,
+      "learning_rate": 0.00012082388352286627,
+      "loss": 2.3315,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.19050267394620785,
+      "learning_rate": 0.00012034018109709716,
+      "loss": 2.2552,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.18778525856213116,
+      "learning_rate": 0.00011985731634247809,
+      "loss": 2.1114,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.19865405908954992,
+      "learning_rate": 0.00011937529032438904,
+      "loss": 2.1821,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1974929876620853,
+      "learning_rate": 0.00011889410410635887,
+      "loss": 2.1329,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21243387820957768,
+      "learning_rate": 0.0001184137587500641,
+      "loss": 2.1182,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1809294113991889,
+      "learning_rate": 0.00011793425531532564,
+      "loss": 2.1128,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.17787120218346783,
+      "learning_rate": 0.00011745559486010671,
+      "loss": 2.2297,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21606891099112033,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 2.2467,
+      "step": 1711
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2258865065213219,
+      "learning_rate": 0.00011650080711077964,
+      "loss": 2.1912,
+      "step": 1712
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19934098623832255,
+      "learning_rate": 0.00011602468192328936,
+      "loss": 2.296,
+      "step": 1713
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19709652821797638,
+      "learning_rate": 0.00011554940392854973,
+      "loss": 2.2448,
+      "step": 1714
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1740995585014064,
+      "learning_rate": 0.00011507497417520146,
+      "loss": 2.2519,
+      "step": 1715
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19590812595424664,
+      "learning_rate": 0.00011460139371001339,
+      "loss": 2.234,
+      "step": 1716
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1755138115956415,
+      "learning_rate": 0.00011412866357788049,
+      "loss": 2.1393,
+      "step": 1717
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.17941813046667845,
+      "learning_rate": 0.00011365678482182207,
+      "loss": 2.0629,
+      "step": 1718
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19125703883185885,
+      "learning_rate": 0.0001131857584829783,
+      "loss": 2.1965,
+      "step": 1719
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.21757703770553918,
+      "learning_rate": 0.0001127155856006093,
+      "loss": 2.2998,
+      "step": 1720
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.23780922613017255,
+      "learning_rate": 0.00011224626721209141,
+      "loss": 2.084,
+      "step": 1721
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.23940348029101957,
+      "learning_rate": 0.0001117778043529164,
+      "loss": 2.2932,
+      "step": 1722
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2727031863292812,
+      "learning_rate": 0.0001113101980566879,
+      "loss": 2.169,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1851528949937712,
+      "learning_rate": 0.00011084344935511958,
+      "loss": 2.1378,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.20585820991495263,
+      "learning_rate": 0.00011037755927803345,
+      "loss": 2.2071,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.17328164649097605,
+      "learning_rate": 0.00010991252885335651,
+      "loss": 2.2355,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2079753273911144,
+      "learning_rate": 0.00010944835910711958,
+      "loss": 2.1974,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.18850172563237755,
+      "learning_rate": 0.00010898505106345396,
+      "loss": 2.2555,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.21488912378705807,
+      "learning_rate": 0.00010852260574459022,
+      "loss": 2.1867,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19873615422040616,
+      "learning_rate": 0.00010806102417085512,
+      "loss": 2.1659,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2661226017715601,
+      "learning_rate": 0.00010760030736066951,
+      "loss": 2.1782,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.21400204753584326,
+      "learning_rate": 0.00010714045633054687,
+      "loss": 2.2344,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.19683992285367743,
+      "learning_rate": 0.00010668147209508971,
+      "loss": 2.1691,
+      "step": 1733
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19587003337868084,
+      "learning_rate": 0.00010622335566698877,
+      "loss": 2.2495,
+      "step": 1734
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2171496905534823,
+      "learning_rate": 0.00010576610805701942,
+      "loss": 2.241,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2093414877701797,
+      "learning_rate": 0.00010530973027404073,
+      "loss": 2.1838,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1782116980049615,
+      "learning_rate": 0.00010485422332499212,
+      "loss": 2.1561,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.18083806183817183,
+      "learning_rate": 0.00010439958821489165,
+      "loss": 2.1607,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.20240197636029783,
+      "learning_rate": 0.00010394582594683428,
+      "loss": 2.1399,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19213255189345904,
+      "learning_rate": 0.0001034929375219884,
+      "loss": 2.106,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1861107618455325,
+      "learning_rate": 0.00010304092393959514,
+      "loss": 2.174,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1822005738759056,
+      "learning_rate": 0.00010258978619696468,
+      "loss": 2.0973,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19943263125110733,
+      "learning_rate": 0.00010213952528947551,
+      "loss": 2.2293,
+      "step": 1743
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.18231295484075816,
+      "learning_rate": 0.00010169014221057089,
+      "loss": 2.095,
+      "step": 1744
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19399413023280057,
+      "learning_rate": 0.00010124163795175734,
+      "loss": 2.2225,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.22718083779593137,
+      "learning_rate": 0.00010079401350260287,
+      "loss": 2.0985,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.18741488304511364,
+      "learning_rate": 0.00010034726985073362,
+      "loss": 2.2973,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2060957117334076,
+      "learning_rate": 9.9901407981833e-05,
+      "loss": 2.2446,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19253235354533865,
+      "learning_rate": 9.94564288796384e-05,
+      "loss": 2.222,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.25337820962425156,
+      "learning_rate": 9.901233352593953e-05,
+      "loss": 2.2028,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2677452623177619,
+      "learning_rate": 9.856912290057668e-05,
+      "loss": 2.1535,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2002660480282296,
+      "learning_rate": 9.812679798143748e-05,
+      "loss": 2.191,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2036564576414624,
+      "learning_rate": 9.768535974445586e-05,
+      "loss": 2.2427,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19910617508886524,
+      "learning_rate": 9.724480916360906e-05,
+      "loss": 2.1171,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.19563108353633346,
+      "learning_rate": 9.68051472109162e-05,
+      "loss": 2.1735,
+      "step": 1755
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20773625682939686,
+      "learning_rate": 9.636637485643529e-05,
+      "loss": 2.2396,
+      "step": 1756
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20881795791125538,
+      "learning_rate": 9.592849306826174e-05,
+      "loss": 2.3078,
+      "step": 1757
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.22115639761056105,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 2.2354,
+      "step": 1758
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.22310616035679062,
+      "learning_rate": 9.505540505339223e-05,
+      "loss": 2.1185,
+      "step": 1759
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20643716715485386,
+      "learning_rate": 9.4620200753054e-05,
+      "loss": 2.2384,
+      "step": 1760
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20577665878935408,
+      "learning_rate": 9.418589087173441e-05,
+      "loss": 2.1033,
+      "step": 1761
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.19116741242747853,
+      "learning_rate": 9.375247636768325e-05,
+      "loss": 2.1755,
+      "step": 1762
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.19018272894363455,
+      "learning_rate": 9.331995819717443e-05,
+      "loss": 2.2542,
+      "step": 1763
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.20118894812462212,
+      "learning_rate": 9.288833731450419e-05,
+      "loss": 2.1996,
+      "step": 1764
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18727501332591692,
+      "learning_rate": 9.245761467198948e-05,
+      "loss": 2.1894,
+      "step": 1765
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.23599146487908126,
+      "learning_rate": 9.20277912199648e-05,
+      "loss": 2.1774,
+      "step": 1766
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24251108155452514,
+      "learning_rate": 9.159886790678123e-05,
+      "loss": 2.3079,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.21508329933928858,
+      "learning_rate": 9.11708456788033e-05,
+      "loss": 2.2564,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.212138063223538,
+      "learning_rate": 9.074372548040793e-05,
+      "loss": 2.2158,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18114125553046279,
+      "learning_rate": 9.031750825398145e-05,
+      "loss": 2.141,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2000891578175025,
+      "learning_rate": 8.98921949399179e-05,
+      "loss": 2.1094,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18884757369029959,
+      "learning_rate": 8.94677864766173e-05,
+      "loss": 2.2201,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.17884144415827705,
+      "learning_rate": 8.904428380048269e-05,
+      "loss": 2.1754,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.18784435817932182,
+      "learning_rate": 8.862168784591929e-05,
+      "loss": 2.255,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.19944156139958769,
+      "learning_rate": 8.819999954533115e-05,
+      "loss": 2.2959,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2328736129968475,
+      "learning_rate": 8.777921982911996e-05,
+      "loss": 2.1767,
+      "step": 1776
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1891135650617138,
+      "learning_rate": 8.735934962568253e-05,
+      "loss": 2.1549,
+      "step": 1777
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2342808209378535,
+      "learning_rate": 8.694038986140945e-05,
+      "loss": 2.0982,
+      "step": 1778
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2420984741543743,
+      "learning_rate": 8.652234146068206e-05,
+      "loss": 2.225,
+      "step": 1779
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20031550440461612,
+      "learning_rate": 8.610520534587086e-05,
+      "loss": 2.2026,
+      "step": 1780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20575513621666017,
+      "learning_rate": 8.568898243733397e-05,
+      "loss": 2.265,
+      "step": 1781
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20281790383411716,
+      "learning_rate": 8.527367365341409e-05,
+      "loss": 2.2274,
+      "step": 1782
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.19099244638056956,
+      "learning_rate": 8.485927991043757e-05,
+      "loss": 2.0307,
+      "step": 1783
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.21005635827262786,
+      "learning_rate": 8.444580212271125e-05,
+      "loss": 2.2305,
+      "step": 1784
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1830157410501735,
+      "learning_rate": 8.403324120252159e-05,
+      "loss": 2.2158,
+      "step": 1785
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.19146743934589475,
+      "learning_rate": 8.362159806013175e-05,
+      "loss": 2.227,
+      "step": 1786
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1887199787218294,
+      "learning_rate": 8.321087360377988e-05,
+      "loss": 2.2483,
+      "step": 1787
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2028962951882763,
+      "learning_rate": 8.280106873967752e-05,
+      "loss": 2.1891,
+      "step": 1788
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20052000910923584,
+      "learning_rate": 8.239218437200679e-05,
+      "loss": 2.1685,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.19132550343799978,
+      "learning_rate": 8.198422140291939e-05,
+      "loss": 2.239,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1898714917050587,
+      "learning_rate": 8.157718073253351e-05,
+      "loss": 2.2275,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.18634586399936065,
+      "learning_rate": 8.117106325893287e-05,
+      "loss": 2.142,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1825308805880104,
+      "learning_rate": 8.076586987816404e-05,
+      "loss": 2.1782,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.21418924241377196,
+      "learning_rate": 8.036160148423449e-05,
+      "loss": 2.3745,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1949175517634401,
+      "learning_rate": 7.995825896911141e-05,
+      "loss": 2.0629,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.18712306672705675,
+      "learning_rate": 7.955584322271853e-05,
+      "loss": 2.1716,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1916650205086422,
+      "learning_rate": 7.915435513293523e-05,
+      "loss": 2.2967,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20178552403826586,
+      "learning_rate": 7.875379558559387e-05,
+      "loss": 2.3119,
+      "step": 1798
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.18446327719664796,
+      "learning_rate": 7.835416546447838e-05,
+      "loss": 2.1926,
+      "step": 1799
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.19612353850654568,
+      "learning_rate": 7.795546565132167e-05,
+      "loss": 2.1325,
+      "step": 1800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.19997504214175774,
+      "learning_rate": 7.755769702580412e-05,
+      "loss": 2.1765,
+      "step": 1801
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1987177620796784,
+      "learning_rate": 7.716086046555193e-05,
+      "loss": 2.1697,
+      "step": 1802
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2176508952062808,
+      "learning_rate": 7.676495684613432e-05,
+      "loss": 2.0908,
+      "step": 1803
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.20728211151429424,
+      "learning_rate": 7.636998704106252e-05,
+      "loss": 2.2785,
+      "step": 1804
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.21682824484302715,
+      "learning_rate": 7.597595192178702e-05,
+      "loss": 2.1959,
+      "step": 1805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.18481565817678483,
+      "learning_rate": 7.558285235769646e-05,
+      "loss": 2.2127,
+      "step": 1806
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.20926846487120232,
+      "learning_rate": 7.519068921611494e-05,
+      "loss": 2.2315,
+      "step": 1807
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1984765497680103,
+      "learning_rate": 7.479946336230047e-05,
+      "loss": 2.2238,
+      "step": 1808
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1928630238687702,
+      "learning_rate": 7.440917565944349e-05,
+      "loss": 2.1736,
+      "step": 1809
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2704599584699775,
+      "learning_rate": 7.4019826968664e-05,
+      "loss": 2.2734,
+      "step": 1810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2304509562758494,
+      "learning_rate": 7.363141814901053e-05,
+      "loss": 2.1236,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1942000925477848,
+      "learning_rate": 7.32439500574577e-05,
+      "loss": 2.1942,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.19554712749801237,
+      "learning_rate": 7.285742354890473e-05,
+      "loss": 2.2288,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1820137605640334,
+      "learning_rate": 7.247183947617325e-05,
+      "loss": 2.1655,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.20015347253316956,
+      "learning_rate": 7.20871986900053e-05,
+      "loss": 2.2659,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.22495034164038688,
+      "learning_rate": 7.170350203906218e-05,
+      "loss": 2.1841,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.27835217168178317,
+      "learning_rate": 7.132075036992158e-05,
+      "loss": 2.1883,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.18648231209692093,
+      "learning_rate": 7.093894452707666e-05,
+      "loss": 2.2003,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1829525060701676,
+      "learning_rate": 7.055808535293334e-05,
+      "loss": 2.1439,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.19777689920103092,
+      "learning_rate": 7.017817368780888e-05,
+      "loss": 2.2089,
+      "step": 1820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.21225748704768785,
+      "learning_rate": 6.979921036993042e-05,
+      "loss": 2.1958,
+      "step": 1821
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.23122050625000368,
+      "learning_rate": 6.942119623543202e-05,
+      "loss": 2.2188,
+      "step": 1822
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.19676928170023078,
+      "learning_rate": 6.904413211835414e-05,
+      "loss": 2.2175,
+      "step": 1823
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1848667586477778,
+      "learning_rate": 6.866801885064056e-05,
+      "loss": 2.2651,
+      "step": 1824
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.18818300533971175,
+      "learning_rate": 6.829285726213769e-05,
+      "loss": 2.1735,
+      "step": 1825
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.21194043528920234,
+      "learning_rate": 6.79186481805918e-05,
+      "loss": 2.158,
+      "step": 1826
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.20049431740385892,
+      "learning_rate": 6.754539243164754e-05,
+      "loss": 2.3047,
+      "step": 1827
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.18561535657623318,
+      "learning_rate": 6.717309083884654e-05,
+      "loss": 2.204,
+      "step": 1828
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.20699383399240484,
+      "learning_rate": 6.680174422362468e-05,
+      "loss": 2.2048,
+      "step": 1829
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.21501953782226746,
+      "learning_rate": 6.643135340531136e-05,
+      "loss": 2.2059,
+      "step": 1830
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.20953027982785097,
+      "learning_rate": 6.606191920112664e-05,
+      "loss": 2.1573,
+      "step": 1831
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2005048894980158,
+      "learning_rate": 6.569344242618036e-05,
+      "loss": 2.1987,
+      "step": 1832
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1927734002521077,
+      "learning_rate": 6.532592389346958e-05,
+      "loss": 2.2697,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.19528504866207952,
+      "learning_rate": 6.495936441387713e-05,
+      "loss": 2.2455,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.19711341754790157,
+      "learning_rate": 6.459376479617013e-05,
+      "loss": 2.1955,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.18752884114031879,
+      "learning_rate": 6.422912584699752e-05,
+      "loss": 2.1222,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2085162032520396,
+      "learning_rate": 6.386544837088904e-05,
+      "loss": 2.2519,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.19750882148801546,
+      "learning_rate": 6.350273317025251e-05,
+      "loss": 2.2485,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.20195687587452138,
+      "learning_rate": 6.314098104537324e-05,
+      "loss": 2.2525,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.18450607629703755,
+      "learning_rate": 6.278019279441122e-05,
+      "loss": 2.0268,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1985112798807352,
+      "learning_rate": 6.242036921339972e-05,
+      "loss": 2.2566,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.19469125691243405,
+      "learning_rate": 6.206151109624402e-05,
+      "loss": 2.2547,
+      "step": 1842
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19039330079737057,
+      "learning_rate": 6.170361923471868e-05,
+      "loss": 2.2077,
+      "step": 1843
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.21939976600894096,
+      "learning_rate": 6.134669441846691e-05,
+      "loss": 2.2381,
+      "step": 1844
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19221216812944528,
+      "learning_rate": 6.099073743499772e-05,
+      "loss": 2.235,
+      "step": 1845
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1836073425587186,
+      "learning_rate": 6.063574906968511e-05,
+      "loss": 2.2427,
+      "step": 1846
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.22577770334124844,
+      "learning_rate": 6.028173010576582e-05,
+      "loss": 2.1423,
+      "step": 1847
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.211336298659514,
+      "learning_rate": 5.9928681324337544e-05,
+      "loss": 2.1811,
+      "step": 1848
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.20619227868507703,
+      "learning_rate": 5.957660350435773e-05,
+      "loss": 2.1179,
+      "step": 1849
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19970617236740604,
+      "learning_rate": 5.922549742264122e-05,
+      "loss": 2.3006,
+      "step": 1850
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19942773250705054,
+      "learning_rate": 5.8875363853859166e-05,
+      "loss": 2.1493,
+      "step": 1851
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.23911792575825705,
+      "learning_rate": 5.852620357053651e-05,
+      "loss": 2.2839,
+      "step": 1852
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.182643237008338,
+      "learning_rate": 5.8178017343051336e-05,
+      "loss": 2.1686,
+      "step": 1853
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.18945587597081381,
+      "learning_rate": 5.783080593963219e-05,
+      "loss": 2.1759,
+      "step": 1854
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.22518719464280804,
+      "learning_rate": 5.748457012635683e-05,
+      "loss": 2.1207,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.20338083766079387,
+      "learning_rate": 5.713931066715078e-05,
+      "loss": 2.2452,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2162505419761062,
+      "learning_rate": 5.679502832378497e-05,
+      "loss": 2.2722,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.20059055268022957,
+      "learning_rate": 5.645172385587482e-05,
+      "loss": 2.1582,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1922801930959447,
+      "learning_rate": 5.6109398020877834e-05,
+      "loss": 2.2056,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.18217671989263112,
+      "learning_rate": 5.576805157409265e-05,
+      "loss": 2.1004,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19023629531242559,
+      "learning_rate": 5.542768526865677e-05,
+      "loss": 2.1139,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.19569367089707743,
+      "learning_rate": 5.508829985554509e-05,
+      "loss": 2.1616,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.18580240173141926,
+      "learning_rate": 5.474989608356856e-05,
+      "loss": 2.1938,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.20821077022850057,
+      "learning_rate": 5.441247469937194e-05,
+      "loss": 2.1402,
+      "step": 1864
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.199332613636617,
+      "learning_rate": 5.407603644743286e-05,
+      "loss": 2.2157,
+      "step": 1865
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19107402508334112,
+      "learning_rate": 5.374058207005944e-05,
+      "loss": 2.1925,
+      "step": 1866
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1998386059360506,
+      "learning_rate": 5.3406112307389066e-05,
+      "loss": 2.1977,
+      "step": 1867
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19577399880479301,
+      "learning_rate": 5.3072627897386926e-05,
+      "loss": 2.0557,
+      "step": 1868
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1914850550675608,
+      "learning_rate": 5.27401295758439e-05,
+      "loss": 2.2003,
+      "step": 1869
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.17950457843660056,
+      "learning_rate": 5.2408618076375315e-05,
+      "loss": 2.218,
+      "step": 1870
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19889337052379386,
+      "learning_rate": 5.207809413041914e-05,
+      "loss": 2.0621,
+      "step": 1871
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19109818430953646,
+      "learning_rate": 5.174855846723459e-05,
+      "loss": 2.2414,
+      "step": 1872
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19748445704293383,
+      "learning_rate": 5.1420011813900104e-05,
+      "loss": 2.1245,
+      "step": 1873
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.26941446052539003,
+      "learning_rate": 5.109245489531211e-05,
+      "loss": 2.2114,
+      "step": 1874
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18661276153060335,
+      "learning_rate": 5.0765888434183446e-05,
+      "loss": 2.1697,
+      "step": 1875
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19972798422298496,
+      "learning_rate": 5.0440313151041364e-05,
+      "loss": 2.1656,
+      "step": 1876
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19824390940594824,
+      "learning_rate": 5.011572976422657e-05,
+      "loss": 2.1265,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18572163346625642,
+      "learning_rate": 4.9792138989890825e-05,
+      "loss": 2.184,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18500496629958407,
+      "learning_rate": 4.9469541541996234e-05,
+      "loss": 2.1129,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2711769701201184,
+      "learning_rate": 4.914793813231305e-05,
+      "loss": 2.2072,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3012206830534826,
+      "learning_rate": 4.882732947041818e-05,
+      "loss": 2.2257,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.24537301154376234,
+      "learning_rate": 4.850771626369416e-05,
+      "loss": 2.0437,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2167949372812504,
+      "learning_rate": 4.818909921732662e-05,
+      "loss": 2.2226,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.194005884542549,
+      "learning_rate": 4.787147903430383e-05,
+      "loss": 2.2402,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.20235480462107255,
+      "learning_rate": 4.755485641541424e-05,
+      "loss": 2.3329,
+      "step": 1885
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.19910654069227096,
+      "learning_rate": 4.723923205924557e-05,
+      "loss": 2.2449,
+      "step": 1886
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.19415300642372504,
+      "learning_rate": 4.6924606662182736e-05,
+      "loss": 2.1628,
+      "step": 1887
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.22299199880414705,
+      "learning_rate": 4.6610980918406596e-05,
+      "loss": 2.131,
+      "step": 1888
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.22716178879800655,
+      "learning_rate": 4.629835551989276e-05,
+      "loss": 2.2413,
+      "step": 1889
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.19141449013198333,
+      "learning_rate": 4.5986731156409224e-05,
+      "loss": 2.1117,
+      "step": 1890
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2178077859840288,
+      "learning_rate": 4.567610851551568e-05,
+      "loss": 2.2135,
+      "step": 1891
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18395662145279898,
+      "learning_rate": 4.536648828256146e-05,
+      "loss": 2.217,
+      "step": 1892
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.24833285193023724,
+      "learning_rate": 4.505787114068433e-05,
+      "loss": 2.3179,
+      "step": 1893
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.20138924696714214,
+      "learning_rate": 4.4750257770808764e-05,
+      "loss": 2.183,
+      "step": 1894
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.19012786203001,
+      "learning_rate": 4.444364885164448e-05,
+      "loss": 2.1483,
+      "step": 1895
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23678039556562408,
+      "learning_rate": 4.413804505968533e-05,
+      "loss": 2.1112,
+      "step": 1896
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.24128247937453084,
+      "learning_rate": 4.3833447069206944e-05,
+      "loss": 2.1589,
+      "step": 1897
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1979738062347433,
+      "learning_rate": 4.352985555226635e-05,
+      "loss": 2.2818,
+      "step": 1898
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18919320079146243,
+      "learning_rate": 4.322727117869951e-05,
+      "loss": 2.2398,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18221378097255525,
+      "learning_rate": 4.29256946161205e-05,
+      "loss": 2.1706,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1978861247834856,
+      "learning_rate": 4.262512652991968e-05,
+      "loss": 2.1463,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.19058448074411846,
+      "learning_rate": 4.2325567583262113e-05,
+      "loss": 2.3173,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2093543645908347,
+      "learning_rate": 4.2027018437086895e-05,
+      "loss": 2.2174,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.19682517595557825,
+      "learning_rate": 4.172947975010449e-05,
+      "loss": 2.1142,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1860850272660553,
+      "learning_rate": 4.143295217879645e-05,
+      "loss": 2.2912,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.20166645416034587,
+      "learning_rate": 4.113743637741296e-05,
+      "loss": 2.3348,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18803886008201887,
+      "learning_rate": 4.084293299797226e-05,
+      "loss": 2.1972,
+      "step": 1907
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.18953121028684525,
+      "learning_rate": 4.054944269025862e-05,
+      "loss": 2.2734,
+      "step": 1908
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1971576034749491,
+      "learning_rate": 4.025696610182095e-05,
+      "loss": 2.2135,
+      "step": 1909
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.20026802939844263,
+      "learning_rate": 3.996550387797187e-05,
+      "loss": 2.121,
+      "step": 1910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.24744569396766025,
+      "learning_rate": 3.9675056661785556e-05,
+      "loss": 2.0837,
+      "step": 1911
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19065586578441535,
+      "learning_rate": 3.9385625094097154e-05,
+      "loss": 2.207,
+      "step": 1912
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.20765971142714085,
+      "learning_rate": 3.909720981350034e-05,
+      "loss": 2.3281,
+      "step": 1913
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2075437632919322,
+      "learning_rate": 3.880981145634704e-05,
+      "loss": 2.2532,
+      "step": 1914
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.21958259656117657,
+      "learning_rate": 3.852343065674507e-05,
+      "loss": 2.1759,
+      "step": 1915
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.20612161030245038,
+      "learning_rate": 3.8238068046557276e-05,
+      "loss": 2.216,
+      "step": 1916
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19204842084921714,
+      "learning_rate": 3.795372425540006e-05,
+      "loss": 2.3331,
+      "step": 1917
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19954221468337327,
+      "learning_rate": 3.76703999106418e-05,
+      "loss": 2.137,
+      "step": 1918
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.20224296754995272,
+      "learning_rate": 3.7388095637401754e-05,
+      "loss": 2.2432,
+      "step": 1919
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19853684804394414,
+      "learning_rate": 3.7106812058548376e-05,
+      "loss": 2.1035,
+      "step": 1920
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.18477678035420592,
+      "learning_rate": 3.682654979469807e-05,
+      "loss": 2.1804,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2024734553471205,
+      "learning_rate": 3.654730946421403e-05,
+      "loss": 2.207,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19686517653073293,
+      "learning_rate": 3.6269091683204466e-05,
+      "loss": 2.2153,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.18510770456825373,
+      "learning_rate": 3.5991897065521693e-05,
+      "loss": 2.1849,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.21388936167274522,
+      "learning_rate": 3.571572622276026e-05,
+      "loss": 2.3079,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19610269827686574,
+      "learning_rate": 3.544057976425619e-05,
+      "loss": 2.1634,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.17990448810000684,
+      "learning_rate": 3.5166458297085146e-05,
+      "loss": 2.1194,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19463739838401434,
+      "learning_rate": 3.489336242606111e-05,
+      "loss": 2.1601,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.19597754550921037,
+      "learning_rate": 3.462129275373577e-05,
+      "loss": 2.1924,
+      "step": 1929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1936132902928904,
+      "learning_rate": 3.4350249880395924e-05,
+      "loss": 2.3264,
+      "step": 1930
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2077776730847013,
+      "learning_rate": 3.408023440406355e-05,
+      "loss": 2.1909,
+      "step": 1931
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1918216632549217,
+      "learning_rate": 3.381124692049331e-05,
+      "loss": 2.1951,
+      "step": 1932
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.18768923670567975,
+      "learning_rate": 3.354328802317197e-05,
+      "loss": 2.2152,
+      "step": 1933
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2098135394092074,
+      "learning_rate": 3.327635830331677e-05,
+      "loss": 2.1604,
+      "step": 1934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.20327429747523748,
+      "learning_rate": 3.3010458349874206e-05,
+      "loss": 2.1177,
+      "step": 1935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.18687694693447232,
+      "learning_rate": 3.2745588749518775e-05,
+      "loss": 2.1616,
+      "step": 1936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.23459461102577694,
+      "learning_rate": 3.248175008665161e-05,
+      "loss": 2.2814,
+      "step": 1937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.19292016642247395,
+      "learning_rate": 3.221894294339911e-05,
+      "loss": 2.2436,
+      "step": 1938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.19431476506648462,
+      "learning_rate": 3.1957167899611836e-05,
+      "loss": 2.2327,
+      "step": 1939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.18265038775289763,
+      "learning_rate": 3.169642553286334e-05,
+      "loss": 2.1112,
+      "step": 1940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.21706147352754185,
+      "learning_rate": 3.143671641844831e-05,
+      "loss": 2.1174,
+      "step": 1941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1957030574443801,
+      "learning_rate": 3.117804112938205e-05,
+      "loss": 2.1653,
+      "step": 1942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.20656869936447403,
+      "learning_rate": 3.092040023639869e-05,
+      "loss": 2.2373,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1912810812060888,
+      "learning_rate": 3.066379430795002e-05,
+      "loss": 2.1595,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.20029621312624993,
+      "learning_rate": 3.040822391020459e-05,
+      "loss": 2.191,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.19008659157006305,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 2.2164,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.20186695443127436,
+      "learning_rate": 2.9900191960071545e-05,
+      "loss": 2.0746,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.21022148165027219,
+      "learning_rate": 2.9647731528591848e-05,
+      "loss": 2.1648,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.18973935150204035,
+      "learning_rate": 2.9396308869628795e-05,
+      "loss": 2.1556,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.23653670039381058,
+      "learning_rate": 2.914592453791448e-05,
+      "loss": 2.2075,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1845112970208968,
+      "learning_rate": 2.8896579085889994e-05,
+      "loss": 2.2772,
+      "step": 1951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.20908254439747587,
+      "learning_rate": 2.86482730637046e-05,
+      "loss": 2.1511,
+      "step": 1952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19286081646580286,
+      "learning_rate": 2.840100701921383e-05,
+      "loss": 2.0803,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.20551856561898202,
+      "learning_rate": 2.8154781497978898e-05,
+      "loss": 2.2413,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.21889998141671516,
+      "learning_rate": 2.7909597043265013e-05,
+      "loss": 2.1405,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2231807640861434,
+      "learning_rate": 2.7665454196040662e-05,
+      "loss": 2.1732,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19956850408106178,
+      "learning_rate": 2.7422353494975905e-05,
+      "loss": 2.1642,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1967234300049274,
+      "learning_rate": 2.7180295476441573e-05,
+      "loss": 2.2617,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1929931859540948,
+      "learning_rate": 2.6939280674508016e-05,
+      "loss": 2.2123,
+      "step": 1959
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19819382593309176,
+      "learning_rate": 2.669930962094358e-05,
+      "loss": 2.2196,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.18880256169158757,
+      "learning_rate": 2.6460382845214126e-05,
+      "loss": 2.1853,
+      "step": 1961
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.22390549165456095,
+      "learning_rate": 2.6222500874481025e-05,
+      "loss": 2.2164,
+      "step": 1962
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.20143490297331865,
+      "learning_rate": 2.5985664233600827e-05,
+      "loss": 2.2549,
+      "step": 1963
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.18776271406355677,
+      "learning_rate": 2.574987344512336e-05,
+      "loss": 2.2014,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.20132125627381,
+      "learning_rate": 2.5515129029290984e-05,
+      "loss": 2.2347,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.18040612108491977,
+      "learning_rate": 2.5281431504037556e-05,
+      "loss": 2.1925,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19986193573422673,
+      "learning_rate": 2.504878138498684e-05,
+      "loss": 2.1117,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19618785250489398,
+      "learning_rate": 2.48171791854519e-05,
+      "loss": 2.2049,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2079422231330272,
+      "learning_rate": 2.4586625416433473e-05,
+      "loss": 2.1231,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.20815911283199579,
+      "learning_rate": 2.435712058661921e-05,
+      "loss": 2.1638,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1847587492213394,
+      "learning_rate": 2.4128665202382327e-05,
+      "loss": 2.1642,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.18821514796089872,
+      "learning_rate": 2.3901259767780515e-05,
+      "loss": 2.1013,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.18355602953655625,
+      "learning_rate": 2.367490478455514e-05,
+      "loss": 2.1657,
+      "step": 1973
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.20535638029339043,
+      "learning_rate": 2.3449600752129597e-05,
+      "loss": 2.2678,
+      "step": 1974
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18867902458849223,
+      "learning_rate": 2.3225348167608685e-05,
+      "loss": 2.2627,
+      "step": 1975
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1914067256373385,
+      "learning_rate": 2.3002147525777118e-05,
+      "loss": 2.195,
+      "step": 1976
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2136669811879824,
+      "learning_rate": 2.2779999319098856e-05,
+      "loss": 2.2534,
+      "step": 1977
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19946708066958158,
+      "learning_rate": 2.255890403771571e-05,
+      "loss": 2.1757,
+      "step": 1978
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19346733206840042,
+      "learning_rate": 2.233886216944614e-05,
+      "loss": 2.0634,
+      "step": 1979
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2000127873713282,
+      "learning_rate": 2.211987419978484e-05,
+      "loss": 2.1263,
+      "step": 1980
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18134249399893168,
+      "learning_rate": 2.1901940611900705e-05,
+      "loss": 2.1627,
+      "step": 1981
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2363422354942448,
+      "learning_rate": 2.168506188663666e-05,
+      "loss": 2.2085,
+      "step": 1982
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1819892336970973,
+      "learning_rate": 2.1469238502507925e-05,
+      "loss": 2.2615,
+      "step": 1983
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18033471780381427,
+      "learning_rate": 2.125447093570154e-05,
+      "loss": 2.0973,
+      "step": 1984
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.211072872722578,
+      "learning_rate": 2.1040759660074793e-05,
+      "loss": 2.1567,
+      "step": 1985
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19425712631528852,
+      "learning_rate": 2.0828105147154273e-05,
+      "loss": 2.199,
+      "step": 1986
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18725057734621575,
+      "learning_rate": 2.061650786613545e-05,
+      "loss": 2.1931,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.23587010700663522,
+      "learning_rate": 2.040596828388058e-05,
+      "loss": 2.2556,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.20548593971297058,
+      "learning_rate": 2.019648686491865e-05,
+      "loss": 2.2556,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1902358075622334,
+      "learning_rate": 1.9988064071443767e-05,
+      "loss": 2.1304,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19612024562925534,
+      "learning_rate": 1.9780700363314253e-05,
+      "loss": 2.2905,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19556413541453066,
+      "learning_rate": 1.957439619805196e-05,
+      "loss": 2.1391,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19209113008189557,
+      "learning_rate": 1.9369152030840554e-05,
+      "loss": 2.2448,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18751244063533964,
+      "learning_rate": 1.916496831452552e-05,
+      "loss": 2.2361,
+      "step": 1994
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.19187763370449215,
+      "learning_rate": 1.8961845499611998e-05,
+      "loss": 2.2454,
+      "step": 1995
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.20455863432545388,
+      "learning_rate": 1.8759784034264925e-05,
+      "loss": 2.1975,
+      "step": 1996
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1853286827226229,
+      "learning_rate": 1.855878436430708e-05,
+      "loss": 2.142,
+      "step": 1997
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19312026744014948,
+      "learning_rate": 1.835884693321871e-05,
+      "loss": 2.1713,
+      "step": 1998
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19252971065582783,
+      "learning_rate": 1.8159972182136386e-05,
+      "loss": 2.1235,
+      "step": 1999
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.33114500484963694,
+      "learning_rate": 1.7962160549851945e-05,
+      "loss": 2.2087,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.21921616924700024,
+      "learning_rate": 1.776541247281177e-05,
+      "loss": 2.1372,
+      "step": 2001
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19012941654090756,
+      "learning_rate": 1.7569728385115224e-05,
+      "loss": 2.0686,
+      "step": 2002
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19655146436767246,
+      "learning_rate": 1.7375108718514665e-05,
+      "loss": 2.1836,
+      "step": 2003
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.20110724467803348,
+      "learning_rate": 1.7181553902413438e-05,
+      "loss": 2.153,
+      "step": 2004
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1942797000691788,
+      "learning_rate": 1.698906436386577e-05,
+      "loss": 2.1477,
+      "step": 2005
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1891071110101527,
+      "learning_rate": 1.679764052757532e-05,
+      "loss": 2.1255,
+      "step": 2006
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2170093693354212,
+      "learning_rate": 1.6607282815894464e-05,
+      "loss": 2.1107,
+      "step": 2007
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19286216104030646,
+      "learning_rate": 1.6417991648823405e-05,
+      "loss": 2.2312,
+      "step": 2008
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.205173395238311,
+      "learning_rate": 1.6229767444008835e-05,
+      "loss": 2.1098,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1981955010333475,
+      "learning_rate": 1.604261061674378e-05,
+      "loss": 2.2001,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2069598551103229,
+      "learning_rate": 1.5856521579965865e-05,
+      "loss": 2.2153,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19260963093953506,
+      "learning_rate": 1.5671500744256938e-05,
+      "loss": 2.0238,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19701925963760566,
+      "learning_rate": 1.5487548517841953e-05,
+      "loss": 2.2616,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.20101777943736646,
+      "learning_rate": 1.530466530658814e-05,
+      "loss": 2.1451,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19404024047867283,
+      "learning_rate": 1.5122851514004054e-05,
+      "loss": 2.1685,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.19723895213031353,
+      "learning_rate": 1.4942107541238703e-05,
+      "loss": 2.1524,
+      "step": 2016
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.21154888915051387,
+      "learning_rate": 1.4762433787080809e-05,
+      "loss": 2.2019,
+      "step": 2017
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1947307519730949,
+      "learning_rate": 1.4583830647957541e-05,
+      "loss": 2.2779,
+      "step": 2018
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20203760047354097,
+      "learning_rate": 1.4406298517934068e-05,
+      "loss": 2.2577,
+      "step": 2019
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.19348372488895035,
+      "learning_rate": 1.4229837788712562e-05,
+      "loss": 2.2309,
+      "step": 2020
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20412188871083298,
+      "learning_rate": 1.4054448849631085e-05,
+      "loss": 2.2145,
+      "step": 2021
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20275106138522248,
+      "learning_rate": 1.3880132087663145e-05,
+      "loss": 2.1992,
+      "step": 2022
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.18848634444128004,
+      "learning_rate": 1.3706887887416419e-05,
+      "loss": 2.1111,
+      "step": 2023
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1950374091505675,
+      "learning_rate": 1.3534716631132316e-05,
+      "loss": 2.124,
+      "step": 2024
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1980610734894286,
+      "learning_rate": 1.3363618698684853e-05,
+      "loss": 2.1667,
+      "step": 2025
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20403693275685497,
+      "learning_rate": 1.3193594467579728e-05,
+      "loss": 2.1501,
+      "step": 2026
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1875024155918712,
+      "learning_rate": 1.3024644312954026e-05,
+      "loss": 2.1538,
+      "step": 2027
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.22049484051797572,
+      "learning_rate": 1.2856768607574564e-05,
+      "loss": 2.15,
+      "step": 2028
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1825459604626624,
+      "learning_rate": 1.2689967721837947e-05,
+      "loss": 2.1531,
+      "step": 2029
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2096288210912088,
+      "learning_rate": 1.2524242023769006e-05,
+      "loss": 2.0841,
+      "step": 2030
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.18540237027527212,
+      "learning_rate": 1.2359591879020526e-05,
+      "loss": 2.206,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2309550219168011,
+      "learning_rate": 1.2196017650872081e-05,
+      "loss": 2.1261,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1860902171012852,
+      "learning_rate": 1.2033519700229367e-05,
+      "loss": 2.2473,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20046705856995362,
+      "learning_rate": 1.1872098385623586e-05,
+      "loss": 2.1723,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20482532684124266,
+      "learning_rate": 1.1711754063210289e-05,
+      "loss": 2.1784,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2187139674223324,
+      "learning_rate": 1.155248708676887e-05,
+      "loss": 2.1451,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.19393108346645146,
+      "learning_rate": 1.1394297807701737e-05,
+      "loss": 2.2221,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2282910523870032,
+      "learning_rate": 1.1237186575033254e-05,
+      "loss": 2.1543,
+      "step": 2038
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.20548067942247925,
+      "learning_rate": 1.1081153735409522e-05,
+      "loss": 2.2179,
+      "step": 2039
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19971225488982752,
+      "learning_rate": 1.0926199633097156e-05,
+      "loss": 2.1512,
+      "step": 2040
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.18659223958085702,
+      "learning_rate": 1.0772324609982787e-05,
+      "loss": 2.1873,
+      "step": 2041
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19303822054027964,
+      "learning_rate": 1.0619529005571893e-05,
+      "loss": 2.1571,
+      "step": 2042
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1884339440639785,
+      "learning_rate": 1.0467813156988748e-05,
+      "loss": 2.1396,
+      "step": 2043
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1918076845717858,
+      "learning_rate": 1.0317177398975031e-05,
+      "loss": 2.1823,
+      "step": 2044
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1879511613649959,
+      "learning_rate": 1.0167622063889326e-05,
+      "loss": 2.2193,
+      "step": 2045
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19511193646995698,
+      "learning_rate": 1.0019147481706625e-05,
+      "loss": 2.1675,
+      "step": 2046
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1912134356420688,
+      "learning_rate": 9.871753980017051e-06,
+      "loss": 2.2016,
+      "step": 2047
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19656688625405103,
+      "learning_rate": 9.725441884025855e-06,
+      "loss": 2.0808,
+      "step": 2048
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.18858274058121444,
+      "learning_rate": 9.580211516551862e-06,
+      "loss": 2.2434,
+      "step": 2049
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2061849735006877,
+      "learning_rate": 9.436063198027589e-06,
+      "loss": 2.1201,
+      "step": 2050
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2080860279532561,
+      "learning_rate": 9.292997246497959e-06,
+      "loss": 2.1285,
+      "step": 2051
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19466607767495836,
+      "learning_rate": 9.151013977619693e-06,
+      "loss": 2.2615,
+      "step": 2052
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2039045666200661,
+      "learning_rate": 9.010113704661038e-06,
+      "loss": 2.1697,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.24723047297594025,
+      "learning_rate": 8.870296738500316e-06,
+      "loss": 2.2283,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1871069223455371,
+      "learning_rate": 8.731563387626095e-06,
+      "loss": 2.1561,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19282128174545868,
+      "learning_rate": 8.59391395813569e-06,
+      "loss": 2.1445,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19370877413321544,
+      "learning_rate": 8.457348753735328e-06,
+      "loss": 2.2292,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.19454351853028562,
+      "learning_rate": 8.321868075738593e-06,
+      "loss": 2.1805,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20540528006645054,
+      "learning_rate": 8.187472223066371e-06,
+      "loss": 2.2358,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1888830282951952,
+      "learning_rate": 8.054161492246136e-06,
+      "loss": 2.1635,
+      "step": 2060
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.25498811129545057,
+      "learning_rate": 7.921936177411049e-06,
+      "loss": 2.2323,
+      "step": 2061
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2113278323148511,
+      "learning_rate": 7.790796570299463e-06,
+      "loss": 2.0849,
+      "step": 2062
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2775332652026939,
+      "learning_rate": 7.660742960254207e-06,
+      "loss": 2.2149,
+      "step": 2063
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.18903080172632977,
+      "learning_rate": 7.531775634222138e-06,
+      "loss": 2.0931,
+      "step": 2064
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1919165792825485,
+      "learning_rate": 7.403894876753192e-06,
+      "loss": 2.1333,
+      "step": 2065
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.20763809115738863,
+      "learning_rate": 7.277100970000061e-06,
+      "loss": 2.192,
+      "step": 2066
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.18955348581369685,
+      "learning_rate": 7.151394193717408e-06,
+      "loss": 2.12,
+      "step": 2067
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.18328044779973232,
+      "learning_rate": 7.026774825261151e-06,
+      "loss": 2.1856,
+      "step": 2068
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1751543879896552,
+      "learning_rate": 6.903243139588233e-06,
+      "loss": 2.0948,
+      "step": 2069
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19257414335085654,
+      "learning_rate": 6.780799409255522e-06,
+      "loss": 2.1193,
+      "step": 2070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2032283100932146,
+      "learning_rate": 6.659443904419637e-06,
+      "loss": 2.365,
+      "step": 2071
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1901335731900858,
+      "learning_rate": 6.539176892836008e-06,
+      "loss": 2.2127,
+      "step": 2072
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1830822410232406,
+      "learning_rate": 6.4199986398585375e-06,
+      "loss": 2.2466,
+      "step": 2073
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19127275389743414,
+      "learning_rate": 6.3019094084388884e-06,
+      "loss": 2.1827,
+      "step": 2074
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.18726510950164066,
+      "learning_rate": 6.18490945912592e-06,
+      "loss": 2.2055,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.22584349296502365,
+      "learning_rate": 6.068999050065249e-06,
+      "loss": 2.203,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19956677967305267,
+      "learning_rate": 5.9541784369983586e-06,
+      "loss": 2.1919,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.20067286258465994,
+      "learning_rate": 5.840447873262433e-06,
+      "loss": 2.2103,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.23129609567199327,
+      "learning_rate": 5.727807609789471e-06,
+      "loss": 2.1047,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.20171431506497148,
+      "learning_rate": 5.616257895105892e-06,
+      "loss": 2.1326,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19486933048017996,
+      "learning_rate": 5.505798975331933e-06,
+      "loss": 2.2657,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.21593627074979774,
+      "learning_rate": 5.396431094181198e-06,
+      "loss": 2.2251,
+      "step": 2082
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23641825745444192,
+      "learning_rate": 5.288154492960107e-06,
+      "loss": 2.241,
+      "step": 2083
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.18691229164519382,
+      "learning_rate": 5.1809694105671155e-06,
+      "loss": 2.2606,
+      "step": 2084
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1921254091290113,
+      "learning_rate": 5.074876083492441e-06,
+      "loss": 2.1476,
+      "step": 2085
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1889631408149205,
+      "learning_rate": 4.96987474581767e-06,
+      "loss": 2.2383,
+      "step": 2086
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.18570516765200648,
+      "learning_rate": 4.865965629214819e-06,
+      "loss": 2.0722,
+      "step": 2087
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.17649910697872934,
+      "learning_rate": 4.763148962946218e-06,
+      "loss": 2.1595,
+      "step": 2088
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2099212550584415,
+      "learning_rate": 4.661424973863681e-06,
+      "loss": 2.1306,
+      "step": 2089
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.18869501286940418,
+      "learning_rate": 4.560793886408398e-06,
+      "loss": 2.2137,
+      "step": 2090
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.20169776238862558,
+      "learning_rate": 4.461255922609986e-06,
+      "loss": 2.2509,
+      "step": 2091
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.21619108989468952,
+      "learning_rate": 4.362811302086267e-06,
+      "loss": 2.2517,
+      "step": 2092
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.20771470044218263,
+      "learning_rate": 4.265460242042885e-06,
+      "loss": 2.1139,
+      "step": 2093
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19972908253861982,
+      "learning_rate": 4.169202957272522e-06,
+      "loss": 2.1752,
+      "step": 2094
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19957327839708133,
+      "learning_rate": 4.074039660154738e-06,
+      "loss": 2.1985,
+      "step": 2095
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19883606882050645,
+      "learning_rate": 3.9799705606551325e-06,
+      "loss": 2.2758,
+      "step": 2096
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.20808595260955495,
+      "learning_rate": 3.886995866325294e-06,
+      "loss": 2.113,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1894819161544858,
+      "learning_rate": 3.795115782302072e-06,
+      "loss": 2.2941,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.18966916999933695,
+      "learning_rate": 3.704330511307197e-06,
+      "loss": 2.1239,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.21555008905562795,
+      "learning_rate": 3.614640253646828e-06,
+      "loss": 2.1663,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19513533608743058,
+      "learning_rate": 3.5260452072110594e-06,
+      "loss": 2.2996,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2523173585943737,
+      "learning_rate": 3.4385455674737498e-06,
+      "loss": 2.1433,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19230225924480368,
+      "learning_rate": 3.3521415274915256e-06,
+      "loss": 2.2118,
+      "step": 2103
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.19152261302246,
+      "learning_rate": 3.2668332779041133e-06,
+      "loss": 2.1712,
+      "step": 2104
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.18578534899639837,
+      "learning_rate": 3.1826210069332838e-06,
+      "loss": 2.1601,
+      "step": 2105
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20699721862798706,
+      "learning_rate": 3.0995049003826324e-06,
+      "loss": 2.1755,
+      "step": 2106
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1992680098851376,
+      "learning_rate": 3.017485141637355e-06,
+      "loss": 2.1091,
+      "step": 2107
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19970589910061495,
+      "learning_rate": 2.9365619116636376e-06,
+      "loss": 2.2035,
+      "step": 2108
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19131702011255383,
+      "learning_rate": 2.856735389008269e-06,
+      "loss": 2.1642,
+      "step": 2109
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2091099076513038,
+      "learning_rate": 2.778005749798307e-06,
+      "loss": 2.1523,
+      "step": 2110
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20417045585748392,
+      "learning_rate": 2.700373167740744e-06,
+      "loss": 2.163,
+      "step": 2111
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20007450741266897,
+      "learning_rate": 2.62383781412201e-06,
+      "loss": 2.1222,
+      "step": 2112
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1987948992709418,
+      "learning_rate": 2.5483998578076373e-06,
+      "loss": 2.0882,
+      "step": 2113
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.18181097162034857,
+      "learning_rate": 2.4740594652418736e-06,
+      "loss": 2.0758,
+      "step": 2114
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20301242155231014,
+      "learning_rate": 2.4008168004472917e-06,
+      "loss": 2.1441,
+      "step": 2115
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2032301846008531,
+      "learning_rate": 2.3286720250246253e-06,
+      "loss": 2.147,
+      "step": 2116
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.20544452243777264,
+      "learning_rate": 2.2576252981520994e-06,
+      "loss": 2.106,
+      "step": 2117
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2068995093228831,
+      "learning_rate": 2.1876767765853233e-06,
+      "loss": 2.1379,
+      "step": 2118
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19580164526399707,
+      "learning_rate": 2.118826614656788e-06,
+      "loss": 2.1396,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.18186058111208828,
+      "learning_rate": 2.051074964275701e-06,
+      "loss": 2.1983,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1838385133269049,
+      "learning_rate": 1.984421974927375e-06,
+      "loss": 2.1144,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1916877246486266,
+      "learning_rate": 1.9188677936731734e-06,
+      "loss": 2.1611,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.18554331794605056,
+      "learning_rate": 1.8544125651501208e-06,
+      "loss": 2.2233,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19093714451069582,
+      "learning_rate": 1.7910564315704035e-06,
+      "loss": 2.1864,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19972965700226353,
+      "learning_rate": 1.7287995327214257e-06,
+      "loss": 2.156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19739820613325956,
+      "learning_rate": 1.6676420059649754e-06,
+      "loss": 2.1547,
+      "step": 2126
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.21596054472451073,
+      "learning_rate": 1.6075839862374485e-06,
+      "loss": 2.1517,
+      "step": 2127
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19737953667045977,
+      "learning_rate": 1.5486256060492366e-06,
+      "loss": 2.2091,
+      "step": 2128
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19934670093801882,
+      "learning_rate": 1.4907669954844495e-06,
+      "loss": 2.1902,
+      "step": 2129
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2057176331862614,
+      "learning_rate": 1.434008282200805e-06,
+      "loss": 2.0593,
+      "step": 2130
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2080553688250285,
+      "learning_rate": 1.3783495914291844e-06,
+      "loss": 2.1453,
+      "step": 2131
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19914143735690726,
+      "learning_rate": 1.3237910459734104e-06,
+      "loss": 2.233,
+      "step": 2132
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.22979860331370242,
+      "learning_rate": 1.270332766210025e-06,
+      "loss": 2.1735,
+      "step": 2133
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19289931128356794,
+      "learning_rate": 1.2179748700879012e-06,
+      "loss": 2.1888,
+      "step": 2134
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.22941647849578664,
+      "learning_rate": 1.1667174731280205e-06,
+      "loss": 2.2537,
+      "step": 2135
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2009315236826002,
+      "learning_rate": 1.1165606884234182e-06,
+      "loss": 2.2183,
+      "step": 2136
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.18355656786725322,
+      "learning_rate": 1.0675046266386268e-06,
+      "loss": 2.196,
+      "step": 2137
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19148086396910793,
+      "learning_rate": 1.019549396009567e-06,
+      "loss": 2.1516,
+      "step": 2138
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19402210604911022,
+      "learning_rate": 9.726951023434348e-07,
+      "loss": 2.2103,
+      "step": 2139
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.18056688474285934,
+      "learning_rate": 9.269418490182591e-07,
+      "loss": 2.1522,
+      "step": 2140
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19810340509218416,
+      "learning_rate": 8.822897369827332e-07,
+      "loss": 2.2411,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.20897103240974127,
+      "learning_rate": 8.387388647561611e-07,
+      "loss": 2.1462,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.21112426223430172,
+      "learning_rate": 7.962893284279016e-07,
+      "loss": 2.1869,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1908637158993216,
+      "learning_rate": 7.549412216574791e-07,
+      "loss": 2.2562,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.22542434179114795,
+      "learning_rate": 7.146946356743067e-07,
+      "loss": 2.131,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19572774300812076,
+      "learning_rate": 6.755496592773524e-07,
+      "loss": 2.2349,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19299751435300574,
+      "learning_rate": 6.375063788349733e-07,
+      "loss": 2.2702,
+      "step": 2147
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.19205107148682682,
+      "learning_rate": 6.005648782848594e-07,
+      "loss": 2.276,
+      "step": 2148
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.22331968774920136,
+      "learning_rate": 5.647252391337565e-07,
+      "loss": 2.1672,
+      "step": 2149
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.18897163342669585,
+      "learning_rate": 5.299875404572441e-07,
+      "loss": 2.1269,
+      "step": 2150
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.20127712926093394,
+      "learning_rate": 4.963518588996796e-07,
+      "loss": 2.0969,
+      "step": 2151
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.20714562316330812,
+      "learning_rate": 4.638182686738657e-07,
+      "loss": 2.223,
+      "step": 2152
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.23590564639677303,
+      "learning_rate": 4.3238684156110543e-07,
+      "loss": 2.1911,
+      "step": 2153
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19070247624896428,
+      "learning_rate": 4.020576469108139e-07,
+      "loss": 2.0987,
+      "step": 2154
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1819939250388134,
+      "learning_rate": 3.7283075164046274e-07,
+      "loss": 2.1798,
+      "step": 2155
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19964014107200853,
+      "learning_rate": 3.4470622023557995e-07,
+      "loss": 2.2356,
+      "step": 2156
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.18279341900127724,
+      "learning_rate": 3.176841147492504e-07,
+      "loss": 2.062,
+      "step": 2157
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1957150414429896,
+      "learning_rate": 2.9176449480244895e-07,
+      "loss": 2.0838,
+      "step": 2158
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.23085452732795825,
+      "learning_rate": 2.6694741758342967e-07,
+      "loss": 2.2414,
+      "step": 2159
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19089795244642885,
+      "learning_rate": 2.432329378478926e-07,
+      "loss": 2.2245,
+      "step": 2160
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.20066195840208897,
+      "learning_rate": 2.2062110791892798e-07,
+      "loss": 2.2919,
+      "step": 2161
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2157511364181122,
+      "learning_rate": 1.9911197768662792e-07,
+      "loss": 2.1778,
+      "step": 2162
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19891805881025382,
+      "learning_rate": 1.7870559460814173e-07,
+      "loss": 2.1806,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.24079538938176967,
+      "learning_rate": 1.5940200370750947e-07,
+      "loss": 2.2058,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.21886722587155213,
+      "learning_rate": 1.4120124757577291e-07,
+      "loss": 2.2762,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.19607844221964868,
+      "learning_rate": 1.2410336637047603e-07,
+      "loss": 2.1298,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.18686611614232976,
+      "learning_rate": 1.081083978159425e-07,
+      "loss": 2.2318,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.18841378167004874,
+      "learning_rate": 9.321637720310915e-08,
+      "loss": 2.2597,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.21712361431867647,
+      "learning_rate": 7.942733738924845e-08,
+      "loss": 2.0964,
+      "step": 2169
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.21578423526770352,
+      "learning_rate": 6.6741308798135e-08,
+      "loss": 2.0722,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.21765132516734073,
+      "learning_rate": 5.5158319419934546e-08,
+      "loss": 2.2093,
+      "step": 2171
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1865658112942751,
+      "learning_rate": 4.4678394810981906e-08,
+      "loss": 2.1088,
+      "step": 2172
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.19971384253638758,
+      "learning_rate": 3.5301558093947527e-08,
+      "loss": 2.1057,
+      "step": 2173
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.20203706154656964,
+      "learning_rate": 2.7027829957559925e-08,
+      "loss": 2.1473,
+      "step": 2174
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1922634552480605,
+      "learning_rate": 1.985722865682771e-08,
+      "loss": 2.2618,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.18535374714409486,
+      "learning_rate": 1.3789770012762048e-08,
+      "loss": 2.16,
+      "step": 2176
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.18746212844333343,
+      "learning_rate": 8.825467412376665e-09,
+      "loss": 2.2552,
+      "step": 2177
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.23552355636141623,
+      "learning_rate": 4.96433180879885e-09,
+      "loss": 2.097,
+      "step": 2178
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1997303195681656,
+      "learning_rate": 2.206371721158451e-09,
+      "loss": 2.1229,
+      "step": 2179
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.21279036854422084,
+      "learning_rate": 5.515932345323549e-10,
+      "loss": 2.3312,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.18951126699640802,
+      "learning_rate": 0.0,
+      "loss": 2.1756,
+      "step": 2181
+    },
+    {
+      "epoch": 1.0,
+      "step": 2181,
+      "total_flos": 209650944909312.0,
+      "train_loss": 2.4349290611874013,
+      "train_runtime": 70604.6117,
+      "train_samples_per_second": 7.905,
+      "train_steps_per_second": 0.031
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2181,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000.0,
+  "total_flos": 209650944909312.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}