diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7680 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1089,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0027548209366391185,
+      "grad_norm": 1624.0,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 18.9199,
+      "step": 1
+    },
+    {
+      "epoch": 0.005509641873278237,
+      "grad_norm": 1640.0,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 18.8697,
+      "step": 2
+    },
+    {
+      "epoch": 0.008264462809917356,
+      "grad_norm": 1584.0,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 17.1905,
+      "step": 3
+    },
+    {
+      "epoch": 0.011019283746556474,
+      "grad_norm": 640.0,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 12.3268,
+      "step": 4
+    },
+    {
+      "epoch": 0.013774104683195593,
+      "grad_norm": 552.0,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 12.8585,
+      "step": 5
+    },
+    {
+      "epoch": 0.01652892561983471,
+      "grad_norm": 374.0,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 12.3047,
+      "step": 6
+    },
+    {
+      "epoch": 0.01928374655647383,
+      "grad_norm": 186.0,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 10.8696,
+      "step": 7
+    },
+    {
+      "epoch": 0.02203856749311295,
+      "grad_norm": 104.5,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 9.2818,
+      "step": 8
+    },
+    {
+      "epoch": 0.024793388429752067,
+      "grad_norm": 75.5,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 8.1257,
+      "step": 9
+    },
+    {
+      "epoch": 0.027548209366391185,
+      "grad_norm": 33.25,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 7.371,
+      "step": 10
+    },
+    {
+      "epoch": 0.030303030303030304,
+      "grad_norm": 19.75,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 6.9551,
+      "step": 11
+    },
+    {
+      "epoch": 0.03305785123966942,
+      "grad_norm": 15.25,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 6.7196,
+      "step": 12
+    },
+    {
+      "epoch": 0.03581267217630854,
+      "grad_norm": 29.375,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 6.5858,
+      "step": 13
+    },
+    {
+      "epoch": 0.03856749311294766,
+      "grad_norm": 19.625,
+      "learning_rate": 8.484848484848486e-05,
+      "loss": 6.5112,
+      "step": 14
+    },
+    {
+      "epoch": 0.04132231404958678,
+      "grad_norm": 7.09375,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 6.3153,
+      "step": 15
+    },
+    {
+      "epoch": 0.0440771349862259,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 6.1753,
+      "step": 16
+    },
+    {
+      "epoch": 0.046831955922865015,
+      "grad_norm": 11.9375,
+      "learning_rate": 0.00010303030303030303,
+      "loss": 6.034,
+      "step": 17
+    },
+    {
+      "epoch": 0.049586776859504134,
+      "grad_norm": 22.25,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 6.0472,
+      "step": 18
+    },
+    {
+      "epoch": 0.05234159779614325,
+      "grad_norm": 8.75,
+      "learning_rate": 0.00011515151515151516,
+      "loss": 5.8011,
+      "step": 19
+    },
+    {
+      "epoch": 0.05509641873278237,
+      "grad_norm": 14.625,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 5.565,
+      "step": 20
+    },
+    {
+      "epoch": 0.05785123966942149,
+      "grad_norm": 60.75,
+      "learning_rate": 0.00012727272727272728,
+      "loss": 5.4288,
+      "step": 21
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 19.125,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 4.9008,
+      "step": 22
+    },
+    {
+      "epoch": 0.06336088154269973,
+      "grad_norm": 21.625,
+      "learning_rate": 0.0001393939393939394,
+      "loss": 4.1849,
+      "step": 23
+    },
+    {
+      "epoch": 0.06611570247933884,
+      "grad_norm": 77.0,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 4.1083,
+      "step": 24
+    },
+    {
+      "epoch": 0.06887052341597796,
+      "grad_norm": 19.0,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 3.0481,
+      "step": 25
+    },
+    {
+      "epoch": 0.07162534435261708,
+      "grad_norm": 17.75,
+      "learning_rate": 0.00015757575757575757,
+      "loss": 2.3526,
+      "step": 26
+    },
+    {
+      "epoch": 0.0743801652892562,
+      "grad_norm": 20.125,
+      "learning_rate": 0.00016363636363636366,
+      "loss": 1.943,
+      "step": 27
+    },
+    {
+      "epoch": 0.07713498622589532,
+      "grad_norm": 5.625,
+      "learning_rate": 0.00016969696969696972,
+      "loss": 1.6061,
+      "step": 28
+    },
+    {
+      "epoch": 0.07988980716253444,
+      "grad_norm": 3.78125,
+      "learning_rate": 0.00017575757575757578,
+      "loss": 1.3966,
+      "step": 29
+    },
+    {
+      "epoch": 0.08264462809917356,
+      "grad_norm": 5.90625,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 1.2738,
+      "step": 30
+    },
+    {
+      "epoch": 0.08539944903581267,
+      "grad_norm": 5.1875,
+      "learning_rate": 0.0001878787878787879,
+      "loss": 1.2299,
+      "step": 31
+    },
+    {
+      "epoch": 0.0881542699724518,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00019393939393939395,
+      "loss": 1.0751,
+      "step": 32
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 1.875,
+      "learning_rate": 0.0002,
+      "loss": 1.0609,
+      "step": 33
+    },
+    {
+      "epoch": 0.09366391184573003,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00019999955747114604,
+      "loss": 0.9572,
+      "step": 34
+    },
+    {
+      "epoch": 0.09641873278236915,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019999822988850082,
+      "loss": 0.9422,
+      "step": 35
+    },
+    {
+      "epoch": 0.09917355371900827,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019999601726381413,
+      "loss": 0.854,
+      "step": 36
+    },
+    {
+      "epoch": 0.10192837465564739,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019999291961666908,
+      "loss": 0.8462,
+      "step": 37
+    },
+    {
+      "epoch": 0.1046831955922865,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001999889369744816,
+      "loss": 0.7722,
+      "step": 38
+    },
+    {
+      "epoch": 0.10743801652892562,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019998406937250034,
+      "loss": 0.7679,
+      "step": 39
+    },
+    {
+      "epoch": 0.11019283746556474,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.00019997831685380642,
+      "loss": 0.7679,
+      "step": 40
+    },
+    {
+      "epoch": 0.11294765840220386,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0001999716794693129,
+      "loss": 0.753,
+      "step": 41
+    },
+    {
+      "epoch": 0.11570247933884298,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019996415727776455,
+      "loss": 0.7336,
+      "step": 42
+    },
+    {
+      "epoch": 0.1184573002754821,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00019995575034573705,
+      "loss": 0.7555,
+      "step": 43
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019994645874763658,
+      "loss": 0.717,
+      "step": 44
+    },
+    {
+      "epoch": 0.12396694214876033,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0001999362825656992,
+      "loss": 0.7194,
+      "step": 45
+    },
+    {
+      "epoch": 0.12672176308539945,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019992522188998994,
+      "loss": 0.678,
+      "step": 46
+    },
+    {
+      "epoch": 0.12947658402203857,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019991327681840218,
+      "loss": 0.7154,
+      "step": 47
+    },
+    {
+      "epoch": 0.1322314049586777,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019990044745665672,
+      "loss": 0.6698,
+      "step": 48
+    },
+    {
+      "epoch": 0.1349862258953168,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0001998867339183008,
+      "loss": 0.696,
+      "step": 49
+    },
+    {
+      "epoch": 0.13774104683195593,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019987213632470717,
+      "loss": 0.6746,
+      "step": 50
+    },
+    {
+      "epoch": 0.14049586776859505,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001998566548050729,
+      "loss": 0.6681,
+      "step": 51
+    },
+    {
+      "epoch": 0.14325068870523416,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001998402894964184,
+      "loss": 0.6078,
+      "step": 52
+    },
+    {
+      "epoch": 0.14600550964187328,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019982304054358614,
+      "loss": 0.6642,
+      "step": 53
+    },
+    {
+      "epoch": 0.1487603305785124,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019980490809923926,
+      "loss": 0.6543,
+      "step": 54
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019978589232386035,
+      "loss": 0.6227,
+      "step": 55
+    },
+    {
+      "epoch": 0.15426997245179064,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019976599338575004,
+      "loss": 0.644,
+      "step": 56
+    },
+    {
+      "epoch": 0.15702479338842976,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019974521146102537,
+      "loss": 0.6395,
+      "step": 57
+    },
+    {
+      "epoch": 0.15977961432506887,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001997235467336184,
+      "loss": 0.6027,
+      "step": 58
+    },
+    {
+      "epoch": 0.162534435261708,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001997009993952744,
+      "loss": 0.6463,
+      "step": 59
+    },
+    {
+      "epoch": 0.1652892561983471,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019967756964555045,
+      "loss": 0.6159,
+      "step": 60
+    },
+    {
+      "epoch": 0.16804407713498623,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019965325769181325,
+      "loss": 0.6143,
+      "step": 61
+    },
+    {
+      "epoch": 0.17079889807162535,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019962806374923764,
+      "loss": 0.6104,
+      "step": 62
+    },
+    {
+      "epoch": 0.17355371900826447,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001996019880408046,
+      "loss": 0.6274,
+      "step": 63
+    },
+    {
+      "epoch": 0.1763085399449036,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019957503079729916,
+      "loss": 0.6072,
+      "step": 64
+    },
+    {
+      "epoch": 0.1790633608815427,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019954719225730847,
+      "loss": 0.6298,
+      "step": 65
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001995184726672197,
+      "loss": 0.5852,
+      "step": 66
+    },
+    {
+      "epoch": 0.18457300275482094,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019948887228121777,
+      "loss": 0.6086,
+      "step": 67
+    },
+    {
+      "epoch": 0.18732782369146006,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001994583913612832,
+      "loss": 0.6177,
+      "step": 68
+    },
+    {
+      "epoch": 0.19008264462809918,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019942703017718975,
+      "loss": 0.5989,
+      "step": 69
+    },
+    {
+      "epoch": 0.1928374655647383,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019939478900650193,
+      "loss": 0.5716,
+      "step": 70
+    },
+    {
+      "epoch": 0.19559228650137742,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019936166813457274,
+      "loss": 0.5617,
+      "step": 71
+    },
+    {
+      "epoch": 0.19834710743801653,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.000199327667854541,
+      "loss": 0.5655,
+      "step": 72
+    },
+    {
+      "epoch": 0.20110192837465565,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019929278846732884,
+      "loss": 0.6169,
+      "step": 73
+    },
+    {
+      "epoch": 0.20385674931129477,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019925703028163892,
+      "loss": 0.5933,
+      "step": 74
+    },
+    {
+      "epoch": 0.2066115702479339,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019922039361395185,
+      "loss": 0.5748,
+      "step": 75
+    },
+    {
+      "epoch": 0.209366391184573,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001991828787885233,
+      "loss": 0.5967,
+      "step": 76
+    },
+    {
+      "epoch": 0.21212121212121213,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019914448613738106,
+      "loss": 0.5905,
+      "step": 77
+    },
+    {
+      "epoch": 0.21487603305785125,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019910521600032227,
+      "loss": 0.5392,
+      "step": 78
+    },
+    {
+      "epoch": 0.21763085399449036,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019906506872491034,
+      "loss": 0.5568,
+      "step": 79
+    },
+    {
+      "epoch": 0.22038567493112948,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001990240446664718,
+      "loss": 0.56,
+      "step": 80
+    },
+    {
+      "epoch": 0.2231404958677686,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001989821441880933,
+      "loss": 0.578,
+      "step": 81
+    },
+    {
+      "epoch": 0.22589531680440772,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019893936766061812,
+      "loss": 0.5757,
+      "step": 82
+    },
+    {
+      "epoch": 0.22865013774104684,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019889571546264335,
+      "loss": 0.5996,
+      "step": 83
+    },
+    {
+      "epoch": 0.23140495867768596,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019885118798051605,
+      "loss": 0.5746,
+      "step": 84
+    },
+    {
+      "epoch": 0.23415977961432508,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019880578560833016,
+      "loss": 0.5497,
+      "step": 85
+    },
+    {
+      "epoch": 0.2369146005509642,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001987595087479229,
+      "loss": 0.5716,
+      "step": 86
+    },
+    {
+      "epoch": 0.2396694214876033,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019871235780887113,
+      "loss": 0.5817,
+      "step": 87
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001986643332084879,
+      "loss": 0.573,
+      "step": 88
+    },
+    {
+      "epoch": 0.24517906336088155,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019861543537181867,
+      "loss": 0.5579,
+      "step": 89
+    },
+    {
+      "epoch": 0.24793388429752067,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019856566473163746,
+      "loss": 0.5799,
+      "step": 90
+    },
+    {
+      "epoch": 0.25068870523415976,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019851502172844317,
+      "loss": 0.5565,
+      "step": 91
+    },
+    {
+      "epoch": 0.2534435261707989,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019846350681045568,
+      "loss": 0.5515,
+      "step": 92
+    },
+    {
+      "epoch": 0.256198347107438,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001984111204336116,
+      "loss": 0.5734,
+      "step": 93
+    },
+    {
+      "epoch": 0.25895316804407714,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019835786306156072,
+      "loss": 0.5379,
+      "step": 94
+    },
+    {
+      "epoch": 0.26170798898071623,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019830373516566146,
+      "loss": 0.5603,
+      "step": 95
+    },
+    {
+      "epoch": 0.2644628099173554,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019824873722497694,
+      "loss": 0.5444,
+      "step": 96
+    },
+    {
+      "epoch": 0.26721763085399447,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019819286972627066,
+      "loss": 0.5374,
+      "step": 97
+    },
+    {
+      "epoch": 0.2699724517906336,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00019813613316400227,
+      "loss": 0.5585,
+      "step": 98
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 0.561,
+      "step": 99
+    },
+    {
+      "epoch": 0.27548209366391185,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001980200548650716,
+      "loss": 0.5279,
+      "step": 100
+    },
+    {
+      "epoch": 0.27823691460055094,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019796071415576925,
+      "loss": 0.5487,
+      "step": 101
+    },
+    {
+      "epoch": 0.2809917355371901,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019790050643761552,
+      "loss": 0.5525,
+      "step": 102
+    },
+    {
+      "epoch": 0.2837465564738292,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019783943224348352,
+      "loss": 0.5848,
+      "step": 103
+    },
+    {
+      "epoch": 0.2865013774104683,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019777749211391502,
+      "loss": 0.514,
+      "step": 104
+    },
+    {
+      "epoch": 0.2892561983471074,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019771468659711595,
+      "loss": 0.5564,
+      "step": 105
+    },
+    {
+      "epoch": 0.29201101928374656,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019765101624895143,
+      "loss": 0.5158,
+      "step": 106
+    },
+    {
+      "epoch": 0.29476584022038566,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001975864816329407,
+      "loss": 0.5225,
+      "step": 107
+    },
+    {
+      "epoch": 0.2975206611570248,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001975210833202524,
+      "loss": 0.552,
+      "step": 108
+    },
+    {
+      "epoch": 0.3002754820936639,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0001974548218896993,
+      "loss": 0.5314,
+      "step": 109
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019738769792773336,
+      "loss": 0.5422,
+      "step": 110
+    },
+    {
+      "epoch": 0.30578512396694213,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019731971202844036,
+      "loss": 0.5293,
+      "step": 111
+    },
+    {
+      "epoch": 0.3085399449035813,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001972508647935347,
+      "loss": 0.5217,
+      "step": 112
+    },
+    {
+      "epoch": 0.31129476584022037,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019718115683235417,
+      "loss": 0.5319,
+      "step": 113
+    },
+    {
+      "epoch": 0.3140495867768595,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019711058876185447,
+      "loss": 0.5627,
+      "step": 114
+    },
+    {
+      "epoch": 0.3168044077134986,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0001970391612066037,
+      "loss": 0.5335,
+      "step": 115
+    },
+    {
+      "epoch": 0.31955922865013775,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.000196966874798777,
+      "loss": 0.5388,
+      "step": 116
+    },
+    {
+      "epoch": 0.32231404958677684,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019689373017815073,
+      "loss": 0.5201,
+      "step": 117
+    },
+    {
+      "epoch": 0.325068870523416,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019681972799209704,
+      "loss": 0.5424,
+      "step": 118
+    },
+    {
+      "epoch": 0.3278236914600551,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.000196744868895578,
+      "loss": 0.5341,
+      "step": 119
+    },
+    {
+      "epoch": 0.3305785123966942,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019666915355113975,
+      "loss": 0.5332,
+      "step": 120
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019659258262890683,
+      "loss": 0.5247,
+      "step": 121
+    },
+    {
+      "epoch": 0.33608815426997246,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019651515680657608,
+      "loss": 0.5323,
+      "step": 122
+    },
+    {
+      "epoch": 0.33884297520661155,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019643687676941068,
+      "loss": 0.5486,
+      "step": 123
+    },
+    {
+      "epoch": 0.3415977961432507,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001963577432102342,
+      "loss": 0.5052,
+      "step": 124
+    },
+    {
+      "epoch": 0.3443526170798898,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001962777568294242,
+      "loss": 0.5039,
+      "step": 125
+    },
+    {
+      "epoch": 0.34710743801652894,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019619691833490643,
+      "loss": 0.5288,
+      "step": 126
+    },
+    {
+      "epoch": 0.349862258953168,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019611522844214813,
+      "loss": 0.5205,
+      "step": 127
+    },
+    {
+      "epoch": 0.3526170798898072,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019603268787415202,
+      "loss": 0.542,
+      "step": 128
+    },
+    {
+      "epoch": 0.35537190082644626,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019594929736144976,
+      "loss": 0.5236,
+      "step": 129
+    },
+    {
+      "epoch": 0.3581267217630854,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001958650576420955,
+      "loss": 0.5205,
+      "step": 130
+    },
+    {
+      "epoch": 0.3608815426997245,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019577996946165937,
+      "loss": 0.5423,
+      "step": 131
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001956940335732209,
+      "loss": 0.5164,
+      "step": 132
+    },
+    {
+      "epoch": 0.36639118457300274,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019560725073736226,
+      "loss": 0.5386,
+      "step": 133
+    },
+    {
+      "epoch": 0.3691460055096419,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019551962172216164,
+      "loss": 0.5246,
+      "step": 134
+    },
+    {
+      "epoch": 0.371900826446281,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001954311473031864,
+      "loss": 0.503,
+      "step": 135
+    },
+    {
+      "epoch": 0.3746556473829201,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001953418282634862,
+      "loss": 0.542,
+      "step": 136
+    },
+    {
+      "epoch": 0.3774104683195592,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019525166539358606,
+      "loss": 0.5124,
+      "step": 137
+    },
+    {
+      "epoch": 0.38016528925619836,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019516065949147943,
+      "loss": 0.5084,
+      "step": 138
+    },
+    {
+      "epoch": 0.38292011019283745,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00019506881136262112,
+      "loss": 0.5375,
+      "step": 139
+    },
+    {
+      "epoch": 0.3856749311294766,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019497612181991998,
+      "loss": 0.5138,
+      "step": 140
+    },
+    {
+      "epoch": 0.3884297520661157,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019488259168373197,
+      "loss": 0.5049,
+      "step": 141
+    },
+    {
+      "epoch": 0.39118457300275483,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001947882217818528,
+      "loss": 0.5096,
+      "step": 142
+    },
+    {
+      "epoch": 0.3939393939393939,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001946930129495106,
+      "loss": 0.4893,
+      "step": 143
+    },
+    {
+      "epoch": 0.39669421487603307,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019459696602935837,
+      "loss": 0.5345,
+      "step": 144
+    },
+    {
+      "epoch": 0.39944903581267216,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019450008187146684,
+      "loss": 0.5522,
+      "step": 145
+    },
+    {
+      "epoch": 0.4022038567493113,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019440236133331674,
+      "loss": 0.5099,
+      "step": 146
+    },
+    {
+      "epoch": 0.4049586776859504,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019430380527979123,
+      "loss": 0.509,
+      "step": 147
+    },
+    {
+      "epoch": 0.40771349862258954,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001942044145831682,
+      "loss": 0.5334,
+      "step": 148
+    },
+    {
+      "epoch": 0.41046831955922863,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019410419012311268,
+      "loss": 0.5144,
+      "step": 149
+    },
+    {
+      "epoch": 0.4132231404958678,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019400313278666902,
+      "loss": 0.5207,
+      "step": 150
+    },
+    {
+      "epoch": 0.41597796143250687,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019390124346825295,
+      "loss": 0.5321,
+      "step": 151
+    },
+    {
+      "epoch": 0.418732782369146,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019379852306964374,
+      "loss": 0.5192,
+      "step": 152
+    },
+    {
+      "epoch": 0.4214876033057851,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001936949724999762,
+      "loss": 0.5055,
+      "step": 153
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001935905926757326,
+      "loss": 0.5222,
+      "step": 154
+    },
+    {
+      "epoch": 0.42699724517906334,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0001934853845207346,
+      "loss": 0.5161,
+      "step": 155
+    },
+    {
+      "epoch": 0.4297520661157025,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019337934896613516,
+      "loss": 0.5383,
+      "step": 156
+    },
+    {
+      "epoch": 0.4325068870523416,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001932724869504101,
+      "loss": 0.5556,
+      "step": 157
+    },
+    {
+      "epoch": 0.43526170798898073,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001931647994193499,
+      "loss": 0.5111,
+      "step": 158
+    },
+    {
+      "epoch": 0.4380165289256198,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019305628732605137,
+      "loss": 0.4922,
+      "step": 159
+    },
+    {
+      "epoch": 0.44077134986225897,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00019294695163090916,
+      "loss": 0.5191,
+      "step": 160
+    },
+    {
+      "epoch": 0.44352617079889806,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019283679330160726,
+      "loss": 0.5117,
+      "step": 161
+    },
+    {
+      "epoch": 0.4462809917355372,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019272581331311047,
+      "loss": 0.536,
+      "step": 162
+    },
+    {
+      "epoch": 0.4490358126721763,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019261401264765573,
+      "loss": 0.488,
+      "step": 163
+    },
+    {
+      "epoch": 0.45179063360881544,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001925013922947434,
+      "loss": 0.5257,
+      "step": 164
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.5125,
+      "step": 165
+    },
+    {
+      "epoch": 0.4573002754820937,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001922736965208125,
+      "loss": 0.5307,
+      "step": 166
+    },
+    {
+      "epoch": 0.46005509641873277,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001921586231150329,
+      "loss": 0.5044,
+      "step": 167
+    },
+    {
+      "epoch": 0.4628099173553719,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019204273405225587,
+      "loss": 0.5166,
+      "step": 168
+    },
+    {
+      "epoch": 0.465564738292011,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019192603035816656,
+      "loss": 0.521,
+      "step": 169
+    },
+    {
+      "epoch": 0.46831955922865015,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019180851306566,
+      "loss": 0.5107,
+      "step": 170
+    },
+    {
+      "epoch": 0.47107438016528924,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019169018321483198,
+      "loss": 0.5087,
+      "step": 171
+    },
+    {
+      "epoch": 0.4738292011019284,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00019157104185297004,
+      "loss": 0.5009,
+      "step": 172
+    },
+    {
+      "epoch": 0.4765840220385675,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019145109003454396,
+      "loss": 0.521,
+      "step": 173
+    },
+    {
+      "epoch": 0.4793388429752066,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019133032882119653,
+      "loss": 0.536,
+      "step": 174
+    },
+    {
+      "epoch": 0.4820936639118457,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019120875928173422,
+      "loss": 0.5197,
+      "step": 175
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019108638249211758,
+      "loss": 0.4989,
+      "step": 176
+    },
+    {
+      "epoch": 0.48760330578512395,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019096319953545185,
+      "loss": 0.5135,
+      "step": 177
+    },
+    {
+      "epoch": 0.4903581267217631,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019083921150197725,
+      "loss": 0.5124,
+      "step": 178
+    },
+    {
+      "epoch": 0.4931129476584022,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019071441948905943,
+      "loss": 0.4891,
+      "step": 179
+    },
+    {
+      "epoch": 0.49586776859504134,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019058882460117974,
+      "loss": 0.4992,
+      "step": 180
+    },
+    {
+      "epoch": 0.4986225895316804,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019046242794992538,
+      "loss": 0.5117,
+      "step": 181
+    },
+    {
+      "epoch": 0.5013774104683195,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00019033523065397972,
+      "loss": 0.5013,
+      "step": 182
+    },
+    {
+      "epoch": 0.5041322314049587,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019020723383911215,
+      "loss": 0.5084,
+      "step": 183
+    },
+    {
+      "epoch": 0.5068870523415978,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00019007843863816842,
+      "loss": 0.4984,
+      "step": 184
+    },
+    {
+      "epoch": 0.509641873278237,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00018994884619106031,
+      "loss": 0.5314,
+      "step": 185
+    },
+    {
+      "epoch": 0.512396694214876,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00018981845764475582,
+      "loss": 0.517,
+      "step": 186
+    },
+    {
+      "epoch": 0.5151515151515151,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00018968727415326884,
+      "loss": 0.4851,
+      "step": 187
+    },
+    {
+      "epoch": 0.5179063360881543,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00018955529687764893,
+      "loss": 0.517,
+      "step": 188
+    },
+    {
+      "epoch": 0.5206611570247934,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00018942252698597113,
+      "loss": 0.5006,
+      "step": 189
+    },
+    {
+      "epoch": 0.5234159779614325,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018928896565332566,
+      "loss": 0.4963,
+      "step": 190
+    },
+    {
+      "epoch": 0.5261707988980716,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00018915461406180731,
+      "loss": 0.5444,
+      "step": 191
+    },
+    {
+      "epoch": 0.5289256198347108,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00018901947340050528,
+      "loss": 0.5151,
+      "step": 192
+    },
+    {
+      "epoch": 0.5316804407713499,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00018888354486549237,
+      "loss": 0.4973,
+      "step": 193
+    },
+    {
+      "epoch": 0.5344352617079889,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018874682965981454,
+      "loss": 0.4932,
+      "step": 194
+    },
+    {
+      "epoch": 0.5371900826446281,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00018860932899348028,
+      "loss": 0.4949,
+      "step": 195
+    },
+    {
+      "epoch": 0.5399449035812672,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001884710440834498,
+      "loss": 0.5052,
+      "step": 196
+    },
+    {
+      "epoch": 0.5426997245179064,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0001883319761536244,
+      "loss": 0.5119,
+      "step": 197
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001881921264348355,
+      "loss": 0.4726,
+      "step": 198
+    },
+    {
+      "epoch": 0.5482093663911846,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001880514961648338,
+      "loss": 0.4884,
+      "step": 199
+    },
+    {
+      "epoch": 0.5509641873278237,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001879100865882784,
+      "loss": 0.4953,
+      "step": 200
+    },
+    {
+      "epoch": 0.5537190082644629,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00018776789895672558,
+      "loss": 0.5015,
+      "step": 201
+    },
+    {
+      "epoch": 0.5564738292011019,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00018762493452861796,
+      "loss": 0.5116,
+      "step": 202
+    },
+    {
+      "epoch": 0.559228650137741,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00018748119456927324,
+      "loss": 0.4906,
+      "step": 203
+    },
+    {
+      "epoch": 0.5619834710743802,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00018733668035087302,
+      "loss": 0.5098,
+      "step": 204
+    },
+    {
+      "epoch": 0.5647382920110193,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018719139315245148,
+      "loss": 0.5101,
+      "step": 205
+    },
+    {
+      "epoch": 0.5674931129476584,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018704533425988423,
+      "loss": 0.5009,
+      "step": 206
+    },
+    {
+      "epoch": 0.5702479338842975,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00018689850496587674,
+      "loss": 0.5054,
+      "step": 207
+    },
+    {
+      "epoch": 0.5730027548209367,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00018675090656995296,
+      "loss": 0.4899,
+      "step": 208
+    },
+    {
+      "epoch": 0.5757575757575758,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.5127,
+      "step": 209
+    },
+    {
+      "epoch": 0.5785123966942148,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00018645340770447595,
+      "loss": 0.4663,
+      "step": 210
+    },
+    {
+      "epoch": 0.581267217630854,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00018630350986795934,
+      "loss": 0.509,
+      "step": 211
+    },
+    {
+      "epoch": 0.5840220385674931,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00018615284819557647,
+      "loss": 0.489,
+      "step": 212
+    },
+    {
+      "epoch": 0.5867768595041323,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00018600142402077006,
+      "loss": 0.4974,
+      "step": 213
+    },
+    {
+      "epoch": 0.5895316804407713,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00018584923868373142,
+      "loss": 0.4829,
+      "step": 214
+    },
+    {
+      "epoch": 0.5922865013774105,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00018569629353138863,
+      "loss": 0.5064,
+      "step": 215
+    },
+    {
+      "epoch": 0.5950413223140496,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00018554258991739452,
+      "loss": 0.4886,
+      "step": 216
+    },
+    {
+      "epoch": 0.5977961432506887,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001853881292021148,
+      "loss": 0.504,
+      "step": 217
+    },
+    {
+      "epoch": 0.6005509641873278,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00018523291275261597,
+      "loss": 0.4898,
+      "step": 218
+    },
+    {
+      "epoch": 0.6033057851239669,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001850769419426531,
+      "loss": 0.4922,
+      "step": 219
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001849202181526579,
+      "loss": 0.4863,
+      "step": 220
+    },
+    {
+      "epoch": 0.6088154269972452,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00018476274276972636,
+      "loss": 0.4867,
+      "step": 221
+    },
+    {
+      "epoch": 0.6115702479338843,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001846045171876065,
+      "loss": 0.4871,
+      "step": 222
+    },
+    {
+      "epoch": 0.6143250688705234,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00018444554280668603,
+      "loss": 0.4823,
+      "step": 223
+    },
+    {
+      "epoch": 0.6170798898071626,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018428582103397997,
+      "loss": 0.492,
+      "step": 224
+    },
+    {
+      "epoch": 0.6198347107438017,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00018412535328311814,
+      "loss": 0.4871,
+      "step": 225
+    },
+    {
+      "epoch": 0.6225895316804407,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00018396414097433277,
+      "loss": 0.5034,
+      "step": 226
+    },
+    {
+      "epoch": 0.6253443526170799,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001838021855344458,
+      "loss": 0.5144,
+      "step": 227
+    },
+    {
+      "epoch": 0.628099173553719,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00018363948839685636,
+      "loss": 0.5138,
+      "step": 228
+    },
+    {
+      "epoch": 0.6308539944903582,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00018347605100152802,
+      "loss": 0.4761,
+      "step": 229
+    },
+    {
+      "epoch": 0.6336088154269972,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00018331187479497603,
+      "loss": 0.4727,
+      "step": 230
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 0.5208,
+      "step": 231
+    },
+    {
+      "epoch": 0.6391184573002755,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00018298131176694378,
+      "loss": 0.5015,
+      "step": 232
+    },
+    {
+      "epoch": 0.6418732782369146,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00018281492787113708,
+      "loss": 0.4731,
+      "step": 233
+    },
+    {
+      "epoch": 0.6446280991735537,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.000182647811015428,
+      "loss": 0.4776,
+      "step": 234
+    },
+    {
+      "epoch": 0.6473829201101928,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018247996267889706,
+      "loss": 0.5096,
+      "step": 235
+    },
+    {
+      "epoch": 0.650137741046832,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00018231138434709893,
+      "loss": 0.4797,
+      "step": 236
+    },
+    {
+      "epoch": 0.6528925619834711,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018214207751204918,
+      "loss": 0.4824,
+      "step": 237
+    },
+    {
+      "epoch": 0.6556473829201102,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018197204367221095,
+      "loss": 0.5107,
+      "step": 238
+    },
+    {
+      "epoch": 0.6584022038567493,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00018180128433248184,
+      "loss": 0.5057,
+      "step": 239
+    },
+    {
+      "epoch": 0.6611570247933884,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001816298010041806,
+      "loss": 0.4754,
+      "step": 240
+    },
+    {
+      "epoch": 0.6639118457300276,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00018145759520503358,
+      "loss": 0.4684,
+      "step": 241
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018128466845916154,
+      "loss": 0.4781,
+      "step": 242
+    },
+    {
+      "epoch": 0.6694214876033058,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001811110222970659,
+      "loss": 0.4812,
+      "step": 243
+    },
+    {
+      "epoch": 0.6721763085399449,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00018093665825561552,
+      "loss": 0.5041,
+      "step": 244
+    },
+    {
+      "epoch": 0.6749311294765841,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00018076157787803268,
+      "loss": 0.5,
+      "step": 245
+    },
+    {
+      "epoch": 0.6776859504132231,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001805857827138798,
+      "loss": 0.4692,
+      "step": 246
+    },
+    {
+      "epoch": 0.6804407713498623,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018040927431904553,
+      "loss": 0.4593,
+      "step": 247
+    },
+    {
+      "epoch": 0.6831955922865014,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00018023205425573096,
+      "loss": 0.5098,
+      "step": 248
+    },
+    {
+      "epoch": 0.6859504132231405,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00018005412409243606,
+      "loss": 0.4892,
+      "step": 249
+    },
+    {
+      "epoch": 0.6887052341597796,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00017987548540394533,
+      "loss": 0.5027,
+      "step": 250
+    },
+    {
+      "epoch": 0.6914600550964187,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001796961397713143,
+      "loss": 0.4778,
+      "step": 251
+    },
+    {
+      "epoch": 0.6942148760330579,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0001795160887818553,
+      "loss": 0.4964,
+      "step": 252
+    },
+    {
+      "epoch": 0.696969696969697,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00017933533402912354,
+      "loss": 0.5048,
+      "step": 253
+    },
+    {
+      "epoch": 0.699724517906336,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001791538771129028,
+      "loss": 0.4973,
+      "step": 254
+    },
+    {
+      "epoch": 0.7024793388429752,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001789717196391916,
+      "loss": 0.4947,
+      "step": 255
+    },
+    {
+      "epoch": 0.7052341597796143,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00017878886322018862,
+      "loss": 0.4908,
+      "step": 256
+    },
+    {
+      "epoch": 0.7079889807162535,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00017860530947427875,
+      "loss": 0.4599,
+      "step": 257
+    },
+    {
+      "epoch": 0.7107438016528925,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00017842106002601856,
+      "loss": 0.5075,
+      "step": 258
+    },
+    {
+      "epoch": 0.7134986225895317,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00017823611650612194,
+      "loss": 0.4909,
+      "step": 259
+    },
+    {
+      "epoch": 0.7162534435261708,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00017805048055144585,
+      "loss": 0.4758,
+      "step": 260
+    },
+    {
+      "epoch": 0.71900826446281,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00017786415380497553,
+      "loss": 0.4698,
+      "step": 261
+    },
+    {
+      "epoch": 0.721763085399449,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00017767713791581026,
+      "loss": 0.4707,
+      "step": 262
+    },
+    {
+      "epoch": 0.7245179063360881,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00017748943453914858,
+      "loss": 0.4895,
+      "step": 263
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001773010453362737,
+      "loss": 0.4671,
+      "step": 264
+    },
+    {
+      "epoch": 0.7300275482093664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00017711197197453878,
+      "loss": 0.4893,
+      "step": 265
+    },
+    {
+      "epoch": 0.7327823691460055,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00017692221612735214,
+      "loss": 0.4913,
+      "step": 266
+    },
+    {
+      "epoch": 0.7355371900826446,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00017673177947416258,
+      "loss": 0.4797,
+      "step": 267
+    },
+    {
+      "epoch": 0.7382920110192838,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00017654066370044436,
+      "loss": 0.486,
+      "step": 268
+    },
+    {
+      "epoch": 0.7410468319559229,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00017634887049768237,
+      "loss": 0.5123,
+      "step": 269
+    },
+    {
+      "epoch": 0.743801652892562,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00017615640156335712,
+      "loss": 0.4724,
+      "step": 270
+    },
+    {
+      "epoch": 0.7465564738292011,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00017596325860092974,
+      "loss": 0.485,
+      "step": 271
+    },
+    {
+      "epoch": 0.7493112947658402,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00017576944331982696,
+      "loss": 0.4708,
+      "step": 272
+    },
+    {
+      "epoch": 0.7520661157024794,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00017557495743542585,
+      "loss": 0.4909,
+      "step": 273
+    },
+    {
+      "epoch": 0.7548209366391184,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00017537980266903866,
+      "loss": 0.4668,
+      "step": 274
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00017518398074789775,
+      "loss": 0.4804,
+      "step": 275
+    },
+    {
+      "epoch": 0.7603305785123967,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001749874934051401,
+      "loss": 0.4815,
+      "step": 276
+    },
+    {
+      "epoch": 0.7630853994490359,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001747903423797921,
+      "loss": 0.4844,
+      "step": 277
+    },
+    {
+      "epoch": 0.7658402203856749,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00017459252941675409,
+      "loss": 0.4798,
+      "step": 278
+    },
+    {
+      "epoch": 0.768595041322314,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00017439405626678496,
+      "loss": 0.508,
+      "step": 279
+    },
+    {
+      "epoch": 0.7713498622589532,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00017419492468648658,
+      "loss": 0.4928,
+      "step": 280
+    },
+    {
+      "epoch": 0.7741046831955923,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0001739951364382884,
+      "loss": 0.4841,
+      "step": 281
+    },
+    {
+      "epoch": 0.7768595041322314,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00017379469329043167,
+      "loss": 0.5048,
+      "step": 282
+    },
+    {
+      "epoch": 0.7796143250688705,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00017359359701695396,
+      "loss": 0.4564,
+      "step": 283
+    },
+    {
+      "epoch": 0.7823691460055097,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001733918493976733,
+      "loss": 0.4736,
+      "step": 284
+    },
+    {
+      "epoch": 0.7851239669421488,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00017318945221817255,
+      "loss": 0.4782,
+      "step": 285
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00017298640726978357,
+      "loss": 0.4577,
+      "step": 286
+    },
+    {
+      "epoch": 0.790633608815427,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001727827163495713,
+      "loss": 0.4815,
+      "step": 287
+    },
+    {
+      "epoch": 0.7933884297520661,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00017257838126031797,
+      "loss": 0.4836,
+      "step": 288
+    },
+    {
+      "epoch": 0.7961432506887053,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00017237340381050703,
+      "loss": 0.4938,
+      "step": 289
+    },
+    {
+      "epoch": 0.7988980716253443,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00017216778581430716,
+      "loss": 0.4921,
+      "step": 290
+    },
+    {
+      "epoch": 0.8016528925619835,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00017196152909155628,
+      "loss": 0.4793,
+      "step": 291
+    },
+    {
+      "epoch": 0.8044077134986226,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0001717546354677455,
+      "loss": 0.4712,
+      "step": 292
+    },
+    {
+      "epoch": 0.8071625344352618,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00017154710677400265,
+      "loss": 0.4825,
+      "step": 293
+    },
+    {
+      "epoch": 0.8099173553719008,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00017133894484707655,
+      "loss": 0.481,
+      "step": 294
+    },
+    {
+      "epoch": 0.8126721763085399,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00017113015152932034,
+      "loss": 0.5066,
+      "step": 295
+    },
+    {
+      "epoch": 0.8154269972451791,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00017092072866867534,
+      "loss": 0.473,
+      "step": 296
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.4743,
+      "step": 297
+    },
+    {
+      "epoch": 0.8209366391184573,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00017050000173832714,
+      "loss": 0.4692,
+      "step": 298
+    },
+    {
+      "epoch": 0.8236914600550964,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001702887013923001,
+      "loss": 0.4613,
+      "step": 299
+    },
+    {
+      "epoch": 0.8264462809917356,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00017007677895070357,
+      "loss": 0.4771,
+      "step": 300
+    },
+    {
+      "epoch": 0.8292011019283747,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00016986423628917346,
+      "loss": 0.4685,
+      "step": 301
+    },
+    {
+      "epoch": 0.8319559228650137,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00016965107528883503,
+      "loss": 0.4753,
+      "step": 302
+    },
+    {
+      "epoch": 0.8347107438016529,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00016943729783628608,
+      "loss": 0.4836,
+      "step": 303
+    },
+    {
+      "epoch": 0.837465564738292,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00016922290582358044,
+      "loss": 0.4545,
+      "step": 304
+    },
+    {
+      "epoch": 0.8402203856749312,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00016900790114821122,
+      "loss": 0.5018,
+      "step": 305
+    },
+    {
+      "epoch": 0.8429752066115702,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00016879228571309377,
+      "loss": 0.4917,
+      "step": 306
+    },
+    {
+      "epoch": 0.8457300275482094,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00016857606142654918,
+      "loss": 0.4691,
+      "step": 307
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00016835923020228712,
+      "loss": 0.4738,
+      "step": 308
+    },
+    {
+      "epoch": 0.8512396694214877,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00016814179395938913,
+      "loss": 0.4974,
+      "step": 309
+    },
+    {
+      "epoch": 0.8539944903581267,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00016792375462229133,
+      "loss": 0.4819,
+      "step": 310
+    },
+    {
+      "epoch": 0.8567493112947658,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001677051141207678,
+      "loss": 0.5076,
+      "step": 311
+    },
+    {
+      "epoch": 0.859504132231405,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00016748587438991303,
+      "loss": 0.4875,
+      "step": 312
+    },
+    {
+      "epoch": 0.8622589531680441,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00016726603737012529,
+      "loss": 0.4792,
+      "step": 313
+    },
+    {
+      "epoch": 0.8650137741046832,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001670456050070889,
+      "loss": 0.4967,
+      "step": 314
+    },
+    {
+      "epoch": 0.8677685950413223,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00016682457925175763,
+      "loss": 0.4549,
+      "step": 315
+    },
+    {
+      "epoch": 0.8705234159779615,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00016660296206033693,
+      "loss": 0.4735,
+      "step": 316
+    },
+    {
+      "epoch": 0.8732782369146006,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00016638075539426677,
+      "loss": 0.4708,
+      "step": 317
+    },
+    {
+      "epoch": 0.8760330578512396,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00016615796122020443,
+      "loss": 0.4614,
+      "step": 318
+    },
+    {
+      "epoch": 0.8787878787878788,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00016593458151000688,
+      "loss": 0.5232,
+      "step": 319
+    },
+    {
+      "epoch": 0.8815426997245179,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016571061824071354,
+      "loss": 0.5013,
+      "step": 320
+    },
+    {
+      "epoch": 0.8842975206611571,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00016548607339452853,
+      "loss": 0.4825,
+      "step": 321
+    },
+    {
+      "epoch": 0.8870523415977961,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001652609489588033,
+      "loss": 0.5196,
+      "step": 322
+    },
+    {
+      "epoch": 0.8898071625344353,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00016503524692601908,
+      "loss": 0.4606,
+      "step": 323
+    },
+    {
+      "epoch": 0.8925619834710744,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00016480896929376907,
+      "loss": 0.4919,
+      "step": 324
+    },
+    {
+      "epoch": 0.8953168044077136,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00016458211806474088,
+      "loss": 0.4736,
+      "step": 325
+    },
+    {
+      "epoch": 0.8980716253443526,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00016435469524669885,
+      "loss": 0.478,
+      "step": 326
+    },
+    {
+      "epoch": 0.9008264462809917,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001641267028524661,
+      "loss": 0.4843,
+      "step": 327
+    },
+    {
+      "epoch": 0.9035812672176309,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00016389814289990694,
+      "loss": 0.4677,
+      "step": 328
+    },
+    {
+      "epoch": 0.90633608815427,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00016366901741190882,
+      "loss": 0.4844,
+      "step": 329
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00016343932841636456,
+      "loss": 0.4593,
+      "step": 330
+    },
+    {
+      "epoch": 0.9118457300275482,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00016320907794615427,
+      "loss": 0.477,
+      "step": 331
+    },
+    {
+      "epoch": 0.9146005509641874,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00016297826803912755,
+      "loss": 0.4878,
+      "step": 332
+    },
+    {
+      "epoch": 0.9173553719008265,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001627469007380852,
+      "loss": 0.4824,
+      "step": 333
+    },
+    {
+      "epoch": 0.9201101928374655,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00016251497809076143,
+      "loss": 0.4805,
+      "step": 334
+    },
+    {
+      "epoch": 0.9228650137741047,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0001622825021498054,
+      "loss": 0.4774,
+      "step": 335
+    },
+    {
+      "epoch": 0.9256198347107438,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00016204947497276345,
+      "loss": 0.4725,
+      "step": 336
+    },
+    {
+      "epoch": 0.928374655647383,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00016181589862206052,
+      "loss": 0.4831,
+      "step": 337
+    },
+    {
+      "epoch": 0.931129476584022,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00016158177516498214,
+      "loss": 0.4687,
+      "step": 338
+    },
+    {
+      "epoch": 0.9338842975206612,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00016134710667365596,
+      "loss": 0.4742,
+      "step": 339
+    },
+    {
+      "epoch": 0.9366391184573003,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00016111189522503359,
+      "loss": 0.4629,
+      "step": 340
+    },
+    {
+      "epoch": 0.9393939393939394,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00016087614290087208,
+      "loss": 0.4617,
+      "step": 341
+    },
+    {
+      "epoch": 0.9421487603305785,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00016063985178771555,
+      "loss": 0.4893,
+      "step": 342
+    },
+    {
+      "epoch": 0.9449035812672176,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00016040302397687665,
+      "loss": 0.4857,
+      "step": 343
+    },
+    {
+      "epoch": 0.9476584022038568,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00016016566156441827,
+      "loss": 0.4803,
+      "step": 344
+    },
+    {
+      "epoch": 0.9504132231404959,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001599277666511347,
+      "loss": 0.4898,
+      "step": 345
+    },
+    {
+      "epoch": 0.953168044077135,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.0001596893413425332,
+      "loss": 0.4965,
+      "step": 346
+    },
+    {
+      "epoch": 0.9559228650137741,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00015945038774881534,
+      "loss": 0.4748,
+      "step": 347
+    },
+    {
+      "epoch": 0.9586776859504132,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00015921090798485832,
+      "loss": 0.4843,
+      "step": 348
+    },
+    {
+      "epoch": 0.9614325068870524,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001589709041701962,
+      "loss": 0.4609,
+      "step": 349
+    },
+    {
+      "epoch": 0.9641873278236914,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00015873037842900137,
+      "loss": 0.4519,
+      "step": 350
+    },
+    {
+      "epoch": 0.9669421487603306,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001584893328900653,
+      "loss": 0.4736,
+      "step": 351
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00015824776968678024,
+      "loss": 0.4806,
+      "step": 352
+    },
+    {
+      "epoch": 0.9724517906336089,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00015800569095711982,
+      "loss": 0.5068,
+      "step": 353
+    },
+    {
+      "epoch": 0.9752066115702479,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0001577630988436206,
+      "loss": 0.4547,
+      "step": 354
+    },
+    {
+      "epoch": 0.977961432506887,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001575199954933627,
+      "loss": 0.4729,
+      "step": 355
+    },
+    {
+      "epoch": 0.9807162534435262,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001572763830579511,
+      "loss": 0.4616,
+      "step": 356
+    },
+    {
+      "epoch": 0.9834710743801653,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0001570322636934964,
+      "loss": 0.4631,
+      "step": 357
+    },
+    {
+      "epoch": 0.9862258953168044,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00015678763956059592,
+      "loss": 0.4784,
+      "step": 358
+    },
+    {
+      "epoch": 0.9889807162534435,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00015654251282431437,
+      "loss": 0.4613,
+      "step": 359
+    },
+    {
+      "epoch": 0.9917355371900827,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00015629688565416478,
+      "loss": 0.4409,
+      "step": 360
+    },
+    {
+      "epoch": 0.9944903581267218,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0001560507602240894,
+      "loss": 0.4585,
+      "step": 361
+    },
+    {
+      "epoch": 0.9972451790633609,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00015580413871244035,
+      "loss": 0.4528,
+      "step": 362
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 0.4772,
+      "step": 363
+    },
+    {
+      "epoch": 1.0,
+      "eval_train_loss": 0.46483051776885986,
+      "eval_train_runtime": 8.4396,
+      "eval_train_samples_per_second": 159.604,
+      "eval_train_steps_per_second": 20.025,
+      "step": 363
+    },
+    {
+      "epoch": 1.002754820936639,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001553094161797631,
+      "loss": 0.4305,
+      "step": 364
+    },
+    {
+      "epoch": 1.0055096418732783,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001550613195373149,
+      "loss": 0.431,
+      "step": 365
+    },
+    {
+      "epoch": 1.0082644628099173,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.000154812735570414,
+      "loss": 0.444,
+      "step": 366
+    },
+    {
+      "epoch": 1.0110192837465566,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00015456366647917207,
+      "loss": 0.4232,
+      "step": 367
+    },
+    {
+      "epoch": 1.0137741046831956,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001543141144679942,
+      "loss": 0.4234,
+      "step": 368
+    },
+    {
+      "epoch": 1.0165289256198347,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00015406408174555976,
+      "loss": 0.4268,
+      "step": 369
+    },
+    {
+      "epoch": 1.019283746556474,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0001538135705248026,
+      "loss": 0.429,
+      "step": 370
+    },
+    {
+      "epoch": 1.022038567493113,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00015356258302289165,
+      "loss": 0.4307,
+      "step": 371
+    },
+    {
+      "epoch": 1.024793388429752,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00015331112146121104,
+      "loss": 0.4068,
+      "step": 372
+    },
+    {
+      "epoch": 1.0275482093663912,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00015305918806534082,
+      "loss": 0.4102,
+      "step": 373
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001528067850650368,
+      "loss": 0.4267,
+      "step": 374
+    },
+    {
+      "epoch": 1.0330578512396693,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00015255391469421128,
+      "loss": 0.4107,
+      "step": 375
+    },
+    {
+      "epoch": 1.0358126721763086,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.000152300579190913,
+      "loss": 0.4245,
+      "step": 376
+    },
+    {
+      "epoch": 1.0385674931129476,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00015204678079730724,
+      "loss": 0.4229,
+      "step": 377
+    },
+    {
+      "epoch": 1.0413223140495869,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00015179252175965633,
+      "loss": 0.4371,
+      "step": 378
+    },
+    {
+      "epoch": 1.044077134986226,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00015153780432829945,
+      "loss": 0.4068,
+      "step": 379
+    },
+    {
+      "epoch": 1.046831955922865,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001512826307576329,
+      "loss": 0.4152,
+      "step": 380
+    },
+    {
+      "epoch": 1.0495867768595042,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00015102700330609,
+      "loss": 0.4284,
+      "step": 381
+    },
+    {
+      "epoch": 1.0523415977961432,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00015077092423612117,
+      "loss": 0.4076,
+      "step": 382
+    },
+    {
+      "epoch": 1.0550964187327825,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00015051439581417406,
+      "loss": 0.4264,
+      "step": 383
+    },
+    {
+      "epoch": 1.0578512396694215,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00015025742031067317,
+      "loss": 0.4253,
+      "step": 384
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4233,
+      "step": 385
+    },
+    {
+      "epoch": 1.0633608815426998,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00014974213716047288,
+      "loss": 0.4097,
+      "step": 386
+    },
+    {
+      "epoch": 1.0661157024793388,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00014948383407432678,
+      "loss": 0.4143,
+      "step": 387
+    },
+    {
+      "epoch": 1.0688705234159779,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00014922509302769298,
+      "loss": 0.4124,
+      "step": 388
+    },
+    {
+      "epoch": 1.0716253443526171,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00014896591631057912,
+      "loss": 0.4227,
+      "step": 389
+    },
+    {
+      "epoch": 1.0743801652892562,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00014870630621684872,
+      "loss": 0.4449,
+      "step": 390
+    },
+    {
+      "epoch": 1.0771349862258952,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00014844626504420088,
+      "loss": 0.396,
+      "step": 391
+    },
+    {
+      "epoch": 1.0798898071625345,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00014818579509415008,
+      "loss": 0.4218,
+      "step": 392
+    },
+    {
+      "epoch": 1.0826446280991735,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001479248986720057,
+      "loss": 0.416,
+      "step": 393
+    },
+    {
+      "epoch": 1.0853994490358128,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00014766357808685158,
+      "loss": 0.4134,
+      "step": 394
+    },
+    {
+      "epoch": 1.0881542699724518,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00014740183565152575,
+      "loss": 0.4133,
+      "step": 395
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001471396736825998,
+      "loss": 0.4277,
+      "step": 396
+    },
+    {
+      "epoch": 1.09366391184573,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00014687709450035837,
+      "loss": 0.4151,
+      "step": 397
+    },
+    {
+      "epoch": 1.0964187327823691,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001466141004287789,
+      "loss": 0.4359,
+      "step": 398
+    },
+    {
+      "epoch": 1.0991735537190084,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00014635069379551055,
+      "loss": 0.4354,
+      "step": 399
+    },
+    {
+      "epoch": 1.1019283746556474,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00014608687693185411,
+      "loss": 0.4238,
+      "step": 400
+    },
+    {
+      "epoch": 1.1046831955922864,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00014582265217274104,
+      "loss": 0.428,
+      "step": 401
+    },
+    {
+      "epoch": 1.1074380165289257,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00014555802185671298,
+      "loss": 0.4207,
+      "step": 402
+    },
+    {
+      "epoch": 1.1101928374655647,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00014529298832590085,
+      "loss": 0.4401,
+      "step": 403
+    },
+    {
+      "epoch": 1.1129476584022038,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00014502755392600442,
+      "loss": 0.4105,
+      "step": 404
+    },
+    {
+      "epoch": 1.115702479338843,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00014476172100627127,
+      "loss": 0.4231,
+      "step": 405
+    },
+    {
+      "epoch": 1.118457300275482,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00014449549191947615,
+      "loss": 0.4295,
+      "step": 406
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00014422886902190014,
+      "loss": 0.4217,
+      "step": 407
+    },
+    {
+      "epoch": 1.1239669421487604,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00014396185467330973,
+      "loss": 0.429,
+      "step": 408
+    },
+    {
+      "epoch": 1.1267217630853994,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00014369445123693596,
+      "loss": 0.4326,
+      "step": 409
+    },
+    {
+      "epoch": 1.1294765840220387,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00014342666107945362,
+      "loss": 0.4187,
+      "step": 410
+    },
+    {
+      "epoch": 1.1322314049586777,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00014315848657096004,
+      "loss": 0.4242,
+      "step": 411
+    },
+    {
+      "epoch": 1.1349862258953167,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00014288993008495452,
+      "loss": 0.4036,
+      "step": 412
+    },
+    {
+      "epoch": 1.137741046831956,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00014262099399831683,
+      "loss": 0.4045,
+      "step": 413
+    },
+    {
+      "epoch": 1.140495867768595,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00014235168069128657,
+      "loss": 0.4246,
+      "step": 414
+    },
+    {
+      "epoch": 1.1432506887052343,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00014208199254744192,
+      "loss": 0.4347,
+      "step": 415
+    },
+    {
+      "epoch": 1.1460055096418733,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00014181193195367858,
+      "loss": 0.4302,
+      "step": 416
+    },
+    {
+      "epoch": 1.1487603305785123,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00014154150130018866,
+      "loss": 0.4143,
+      "step": 417
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00014127070298043947,
+      "loss": 0.4227,
+      "step": 418
+    },
+    {
+      "epoch": 1.1542699724517906,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00014099953939115248,
+      "loss": 0.4179,
+      "step": 419
+    },
+    {
+      "epoch": 1.1570247933884297,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00014072801293228188,
+      "loss": 0.4135,
+      "step": 420
+    },
+    {
+      "epoch": 1.159779614325069,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001404561260069935,
+      "loss": 0.4389,
+      "step": 421
+    },
+    {
+      "epoch": 1.162534435261708,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00014018388102164363,
+      "loss": 0.4126,
+      "step": 422
+    },
+    {
+      "epoch": 1.165289256198347,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00013991128038575741,
+      "loss": 0.4252,
+      "step": 423
+    },
+    {
+      "epoch": 1.1680440771349863,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00013963832651200784,
+      "loss": 0.4259,
+      "step": 424
+    },
+    {
+      "epoch": 1.1707988980716253,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00013936502181619416,
+      "loss": 0.4266,
+      "step": 425
+    },
+    {
+      "epoch": 1.1735537190082646,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00013909136871722067,
+      "loss": 0.4137,
+      "step": 426
+    },
+    {
+      "epoch": 1.1763085399449036,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001388173696370752,
+      "loss": 0.4187,
+      "step": 427
+    },
+    {
+      "epoch": 1.1790633608815426,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00013854302700080776,
+      "loss": 0.3995,
+      "step": 428
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.4146,
+      "step": 429
+    },
+    {
+      "epoch": 1.184573002754821,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001379933207752887,
+      "loss": 0.4207,
+      "step": 430
+    },
+    {
+      "epoch": 1.1873278236914602,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001377179620512544,
+      "loss": 0.4037,
+      "step": 431
+    },
+    {
+      "epoch": 1.1900826446280992,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00013744226950148972,
+      "loss": 0.4294,
+      "step": 432
+    },
+    {
+      "epoch": 1.1928374655647382,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00013716624556603274,
+      "loss": 0.4178,
+      "step": 433
+    },
+    {
+      "epoch": 1.1955922865013775,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001368898926878547,
+      "loss": 0.4272,
+      "step": 434
+    },
+    {
+      "epoch": 1.1983471074380165,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00013661321331283796,
+      "loss": 0.4162,
+      "step": 435
+    },
+    {
+      "epoch": 1.2011019283746556,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00013633620988975464,
+      "loss": 0.4381,
+      "step": 436
+    },
+    {
+      "epoch": 1.2038567493112948,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.000136058884870245,
+      "loss": 0.4238,
+      "step": 437
+    },
+    {
+      "epoch": 1.2066115702479339,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00013578124070879534,
+      "loss": 0.4297,
+      "step": 438
+    },
+    {
+      "epoch": 1.209366391184573,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00013550327986271686,
+      "loss": 0.4095,
+      "step": 439
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00013522500479212337,
+      "loss": 0.4156,
+      "step": 440
+    },
+    {
+      "epoch": 1.2148760330578512,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00013494641795990986,
+      "loss": 0.4268,
+      "step": 441
+    },
+    {
+      "epoch": 1.2176308539944904,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.0001346675218317305,
+      "loss": 0.4088,
+      "step": 442
+    },
+    {
+      "epoch": 1.2203856749311295,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0001343883188759771,
+      "loss": 0.4092,
+      "step": 443
+    },
+    {
+      "epoch": 1.2231404958677685,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00013410881156375684,
+      "loss": 0.4108,
+      "step": 444
+    },
+    {
+      "epoch": 1.2258953168044078,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00013382900236887075,
+      "loss": 0.4333,
+      "step": 445
+    },
+    {
+      "epoch": 1.2286501377410468,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00013354889376779167,
+      "loss": 0.4267,
+      "step": 446
+    },
+    {
+      "epoch": 1.231404958677686,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00013326848823964243,
+      "loss": 0.4164,
+      "step": 447
+    },
+    {
+      "epoch": 1.234159779614325,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00013298778826617365,
+      "loss": 0.4092,
+      "step": 448
+    },
+    {
+      "epoch": 1.2369146005509641,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00013270679633174218,
+      "loss": 0.4175,
+      "step": 449
+    },
+    {
+      "epoch": 1.2396694214876034,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00013242551492328875,
+      "loss": 0.4297,
+      "step": 450
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00013214394653031616,
+      "loss": 0.4221,
+      "step": 451
+    },
+    {
+      "epoch": 1.2451790633608815,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001318620936448672,
+      "loss": 0.4161,
+      "step": 452
+    },
+    {
+      "epoch": 1.2479338842975207,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0001315799587615025,
+      "loss": 0.4319,
+      "step": 453
+    },
+    {
+      "epoch": 1.2506887052341598,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00013129754437727863,
+      "loss": 0.4122,
+      "step": 454
+    },
+    {
+      "epoch": 1.2534435261707988,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00013101485299172588,
+      "loss": 0.4014,
+      "step": 455
+    },
+    {
+      "epoch": 1.256198347107438,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001307318871068261,
+      "loss": 0.426,
+      "step": 456
+    },
+    {
+      "epoch": 1.258953168044077,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001304486492269907,
+      "loss": 0.4305,
+      "step": 457
+    },
+    {
+      "epoch": 1.2617079889807163,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001301651418590384,
+      "loss": 0.4289,
+      "step": 458
+    },
+    {
+      "epoch": 1.2644628099173554,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012988136751217291,
+      "loss": 0.4198,
+      "step": 459
+    },
+    {
+      "epoch": 1.2672176308539944,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00012959732869796103,
+      "loss": 0.4475,
+      "step": 460
+    },
+    {
+      "epoch": 1.2699724517906337,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001293130279303102,
+      "loss": 0.4227,
+      "step": 461
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00012902846772544624,
+      "loss": 0.3967,
+      "step": 462
+    },
+    {
+      "epoch": 1.275482093663912,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00012874365060189117,
+      "loss": 0.4117,
+      "step": 463
+    },
+    {
+      "epoch": 1.278236914600551,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00012845857908044096,
+      "loss": 0.408,
+      "step": 464
+    },
+    {
+      "epoch": 1.28099173553719,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00012817325568414297,
+      "loss": 0.4066,
+      "step": 465
+    },
+    {
+      "epoch": 1.283746556473829,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00012788768293827402,
+      "loss": 0.4183,
+      "step": 466
+    },
+    {
+      "epoch": 1.2865013774104683,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00012760186337031762,
+      "loss": 0.4156,
+      "step": 467
+    },
+    {
+      "epoch": 1.2892561983471074,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.000127315799509942,
+      "loss": 0.4126,
+      "step": 468
+    },
+    {
+      "epoch": 1.2920110192837466,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0001270294938889773,
+      "loss": 0.4305,
+      "step": 469
+    },
+    {
+      "epoch": 1.2947658402203857,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001267429490413935,
+      "loss": 0.4245,
+      "step": 470
+    },
+    {
+      "epoch": 1.2975206611570247,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001264561675032779,
+      "loss": 0.4272,
+      "step": 471
+    },
+    {
+      "epoch": 1.300275482093664,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001261691518128126,
+      "loss": 0.4057,
+      "step": 472
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00012588190451025207,
+      "loss": 0.4025,
+      "step": 473
+    },
+    {
+      "epoch": 1.3057851239669422,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00012559442813790076,
+      "loss": 0.4126,
+      "step": 474
+    },
+    {
+      "epoch": 1.3085399449035813,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00012530672524009036,
+      "loss": 0.4268,
+      "step": 475
+    },
+    {
+      "epoch": 1.3112947658402203,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00012501879836315764,
+      "loss": 0.411,
+      "step": 476
+    },
+    {
+      "epoch": 1.3140495867768596,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00012473065005542155,
+      "loss": 0.4174,
+      "step": 477
+    },
+    {
+      "epoch": 1.3168044077134986,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00012444228286716095,
+      "loss": 0.4145,
+      "step": 478
+    },
+    {
+      "epoch": 1.3195592286501379,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00012415369935059184,
+      "loss": 0.4299,
+      "step": 479
+    },
+    {
+      "epoch": 1.322314049586777,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00012386490205984488,
+      "loss": 0.4223,
+      "step": 480
+    },
+    {
+      "epoch": 1.325068870523416,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00012357589355094275,
+      "loss": 0.4076,
+      "step": 481
+    },
+    {
+      "epoch": 1.327823691460055,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001232866763817775,
+      "loss": 0.4252,
+      "step": 482
+    },
+    {
+      "epoch": 1.3305785123966942,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00012299725311208808,
+      "loss": 0.421,
+      "step": 483
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00012270762630343734,
+      "loss": 0.4116,
+      "step": 484
+    },
+    {
+      "epoch": 1.3360881542699725,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0001224177985191897,
+      "loss": 0.4116,
+      "step": 485
+    },
+    {
+      "epoch": 1.3388429752066116,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00012212777232448838,
+      "loss": 0.4069,
+      "step": 486
+    },
+    {
+      "epoch": 1.3415977961432506,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00012183755028623245,
+      "loss": 0.4101,
+      "step": 487
+    },
+    {
+      "epoch": 1.3443526170798898,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00012154713497305455,
+      "loss": 0.4395,
+      "step": 488
+    },
+    {
+      "epoch": 1.3471074380165289,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00012125652895529766,
+      "loss": 0.4243,
+      "step": 489
+    },
+    {
+      "epoch": 1.3498622589531681,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00012096573480499289,
+      "loss": 0.4087,
+      "step": 490
+    },
+    {
+      "epoch": 1.3526170798898072,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00012067475509583616,
+      "loss": 0.4061,
+      "step": 491
+    },
+    {
+      "epoch": 1.3553719008264462,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00012038359240316589,
+      "loss": 0.4051,
+      "step": 492
+    },
+    {
+      "epoch": 1.3581267217630855,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00012009224930393988,
+      "loss": 0.3838,
+      "step": 493
+    },
+    {
+      "epoch": 1.3608815426997245,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00011980072837671273,
+      "loss": 0.3986,
+      "step": 494
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 0.4179,
+      "step": 495
+    },
+    {
+      "epoch": 1.3663911845730028,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00011921716336031972,
+      "loss": 0.3922,
+      "step": 496
+    },
+    {
+      "epoch": 1.3691460055096418,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00011892512443604102,
+      "loss": 0.4042,
+      "step": 497
+    },
+    {
+      "epoch": 1.3719008264462809,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001186329180134898,
+      "loss": 0.4099,
+      "step": 498
+    },
+    {
+      "epoch": 1.3746556473829201,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00011834054667886145,
+      "loss": 0.405,
+      "step": 499
+    },
+    {
+      "epoch": 1.3774104683195592,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00011804801301981108,
+      "loss": 0.4068,
+      "step": 500
+    },
+    {
+      "epoch": 1.3801652892561984,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00011775531962543036,
+      "loss": 0.4156,
+      "step": 501
+    },
+    {
+      "epoch": 1.3829201101928374,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00011746246908622467,
+      "loss": 0.4173,
+      "step": 502
+    },
+    {
+      "epoch": 1.3856749311294765,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00011716946399409039,
+      "loss": 0.4111,
+      "step": 503
+    },
+    {
+      "epoch": 1.3884297520661157,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001168763069422916,
+      "loss": 0.4234,
+      "step": 504
+    },
+    {
+      "epoch": 1.3911845730027548,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00011658300052543742,
+      "loss": 0.4213,
+      "step": 505
+    },
+    {
+      "epoch": 1.393939393939394,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0001162895473394589,
+      "loss": 0.4104,
+      "step": 506
+    },
+    {
+      "epoch": 1.396694214876033,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00011599594998158602,
+      "loss": 0.4113,
+      "step": 507
+    },
+    {
+      "epoch": 1.399449035812672,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001157022110503249,
+      "loss": 0.4175,
+      "step": 508
+    },
+    {
+      "epoch": 1.4022038567493114,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00011540833314543458,
+      "loss": 0.402,
+      "step": 509
+    },
+    {
+      "epoch": 1.4049586776859504,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00011511431886790407,
+      "loss": 0.4215,
+      "step": 510
+    },
+    {
+      "epoch": 1.4077134986225897,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00011482017081992944,
+      "loss": 0.4189,
+      "step": 511
+    },
+    {
+      "epoch": 1.4104683195592287,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00011452589160489062,
+      "loss": 0.4025,
+      "step": 512
+    },
+    {
+      "epoch": 1.4132231404958677,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00011423148382732853,
+      "loss": 0.3975,
+      "step": 513
+    },
+    {
+      "epoch": 1.4159779614325068,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00011393695009292185,
+      "loss": 0.4094,
+      "step": 514
+    },
+    {
+      "epoch": 1.418732782369146,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00011364229300846417,
+      "loss": 0.4179,
+      "step": 515
+    },
+    {
+      "epoch": 1.421487603305785,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00011334751518184061,
+      "loss": 0.4076,
+      "step": 516
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00011305261922200519,
+      "loss": 0.3997,
+      "step": 517
+    },
+    {
+      "epoch": 1.4269972451790633,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001127576077389572,
+      "loss": 0.41,
+      "step": 518
+    },
+    {
+      "epoch": 1.4297520661157024,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001124624833437186,
+      "loss": 0.4096,
+      "step": 519
+    },
+    {
+      "epoch": 1.4325068870523416,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00011216724864831056,
+      "loss": 0.426,
+      "step": 520
+    },
+    {
+      "epoch": 1.4352617079889807,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00011187190626573052,
+      "loss": 0.3964,
+      "step": 521
+    },
+    {
+      "epoch": 1.43801652892562,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011157645880992902,
+      "loss": 0.4115,
+      "step": 522
+    },
+    {
+      "epoch": 1.440771349862259,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00011128090889578648,
+      "loss": 0.4288,
+      "step": 523
+    },
+    {
+      "epoch": 1.443526170798898,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00011098525913909025,
+      "loss": 0.4084,
+      "step": 524
+    },
+    {
+      "epoch": 1.4462809917355373,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00011068951215651132,
+      "loss": 0.4067,
+      "step": 525
+    },
+    {
+      "epoch": 1.4490358126721763,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00011039367056558103,
+      "loss": 0.4235,
+      "step": 526
+    },
+    {
+      "epoch": 1.4517906336088156,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00011009773698466832,
+      "loss": 0.4003,
+      "step": 527
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001098017140329561,
+      "loss": 0.4104,
+      "step": 528
+    },
+    {
+      "epoch": 1.4573002754820936,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00010950560433041826,
+      "loss": 0.4165,
+      "step": 529
+    },
+    {
+      "epoch": 1.4600550964187327,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00010920941049779667,
+      "loss": 0.4114,
+      "step": 530
+    },
+    {
+      "epoch": 1.462809917355372,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001089131351565776,
+      "loss": 0.4035,
+      "step": 531
+    },
+    {
+      "epoch": 1.465564738292011,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010861678092896881,
+      "loss": 0.4025,
+      "step": 532
+    },
+    {
+      "epoch": 1.4683195592286502,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00010832035043787625,
+      "loss": 0.4211,
+      "step": 533
+    },
+    {
+      "epoch": 1.4710743801652892,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00010802384630688078,
+      "loss": 0.4175,
+      "step": 534
+    },
+    {
+      "epoch": 1.4738292011019283,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00010772727116021513,
+      "loss": 0.3955,
+      "step": 535
+    },
+    {
+      "epoch": 1.4765840220385675,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00010743062762274046,
+      "loss": 0.4233,
+      "step": 536
+    },
+    {
+      "epoch": 1.4793388429752066,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00010713391831992323,
+      "loss": 0.4105,
+      "step": 537
+    },
+    {
+      "epoch": 1.4820936639118458,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00010683714587781209,
+      "loss": 0.4012,
+      "step": 538
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00010654031292301432,
+      "loss": 0.3917,
+      "step": 539
+    },
+    {
+      "epoch": 1.487603305785124,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00010624342208267292,
+      "loss": 0.4105,
+      "step": 540
+    },
+    {
+      "epoch": 1.4903581267217632,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00010594647598444312,
+      "loss": 0.4166,
+      "step": 541
+    },
+    {
+      "epoch": 1.4931129476584022,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001056494772564693,
+      "loss": 0.4642,
+      "step": 542
+    },
+    {
+      "epoch": 1.4958677685950414,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00010535242852736151,
+      "loss": 0.4127,
+      "step": 543
+    },
+    {
+      "epoch": 1.4986225895316805,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00010505533242617254,
+      "loss": 0.4035,
+      "step": 544
+    },
+    {
+      "epoch": 1.5013774104683195,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00010475819158237425,
+      "loss": 0.4136,
+      "step": 545
+    },
+    {
+      "epoch": 1.5041322314049586,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00010446100862583459,
+      "loss": 0.41,
+      "step": 546
+    },
+    {
+      "epoch": 1.5068870523415978,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00010416378618679425,
+      "loss": 0.4264,
+      "step": 547
+    },
+    {
+      "epoch": 1.509641873278237,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0001038665268958433,
+      "loss": 0.4157,
+      "step": 548
+    },
+    {
+      "epoch": 1.512396694214876,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00010356923338389806,
+      "loss": 0.4109,
+      "step": 549
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00010327190828217763,
+      "loss": 0.4196,
+      "step": 550
+    },
+    {
+      "epoch": 1.5179063360881542,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00010297455422218074,
+      "loss": 0.4192,
+      "step": 551
+    },
+    {
+      "epoch": 1.5206611570247934,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00010267717383566246,
+      "loss": 0.4453,
+      "step": 552
+    },
+    {
+      "epoch": 1.5234159779614325,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00010237976975461075,
+      "loss": 0.4115,
+      "step": 553
+    },
+    {
+      "epoch": 1.5261707988980717,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00010208234461122343,
+      "loss": 0.402,
+      "step": 554
+    },
+    {
+      "epoch": 1.5289256198347108,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0001017849010378846,
+      "loss": 0.4132,
+      "step": 555
+    },
+    {
+      "epoch": 1.5316804407713498,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00010148744166714159,
+      "loss": 0.3973,
+      "step": 556
+    },
+    {
+      "epoch": 1.5344352617079888,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00010118996913168144,
+      "loss": 0.4122,
+      "step": 557
+    },
+    {
+      "epoch": 1.537190082644628,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010089248606430775,
+      "loss": 0.4054,
+      "step": 558
+    },
+    {
+      "epoch": 1.5399449035812673,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00010059499509791736,
+      "loss": 0.4052,
+      "step": 559
+    },
+    {
+      "epoch": 1.5426997245179064,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00010029749886547701,
+      "loss": 0.3983,
+      "step": 560
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001,
+      "loss": 0.4236,
+      "step": 561
+    },
+    {
+      "epoch": 1.5482093663911844,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.970250113452301e-05,
+      "loss": 0.4274,
+      "step": 562
+    },
+    {
+      "epoch": 1.5509641873278237,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.940500490208263e-05,
+      "loss": 0.4133,
+      "step": 563
+    },
+    {
+      "epoch": 1.553719008264463,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.910751393569229e-05,
+      "loss": 0.4218,
+      "step": 564
+    },
+    {
+      "epoch": 1.556473829201102,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.881003086831859e-05,
+      "loss": 0.4155,
+      "step": 565
+    },
+    {
+      "epoch": 1.559228650137741,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.851255833285843e-05,
+      "loss": 0.4111,
+      "step": 566
+    },
+    {
+      "epoch": 1.56198347107438,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.821509896211539e-05,
+      "loss": 0.3969,
+      "step": 567
+    },
+    {
+      "epoch": 1.5647382920110193,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.791765538877659e-05,
+      "loss": 0.4197,
+      "step": 568
+    },
+    {
+      "epoch": 1.5674931129476584,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.762023024538926e-05,
+      "loss": 0.4182,
+      "step": 569
+    },
+    {
+      "epoch": 1.5702479338842976,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.732282616433756e-05,
+      "loss": 0.4203,
+      "step": 570
+    },
+    {
+      "epoch": 1.5730027548209367,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.702544577781925e-05,
+      "loss": 0.4083,
+      "step": 571
+    },
+    {
+      "epoch": 1.5757575757575757,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.67280917178224e-05,
+      "loss": 0.4044,
+      "step": 572
+    },
+    {
+      "epoch": 1.5785123966942147,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.643076661610196e-05,
+      "loss": 0.4112,
+      "step": 573
+    },
+    {
+      "epoch": 1.581267217630854,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.613347310415672e-05,
+      "loss": 0.4021,
+      "step": 574
+    },
+    {
+      "epoch": 1.5840220385674932,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.583621381320576e-05,
+      "loss": 0.4215,
+      "step": 575
+    },
+    {
+      "epoch": 1.5867768595041323,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.553899137416545e-05,
+      "loss": 0.402,
+      "step": 576
+    },
+    {
+      "epoch": 1.5895316804407713,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.524180841762577e-05,
+      "loss": 0.3915,
+      "step": 577
+    },
+    {
+      "epoch": 1.5922865013774103,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.494466757382748e-05,
+      "loss": 0.4214,
+      "step": 578
+    },
+    {
+      "epoch": 1.5950413223140496,
+      "grad_norm": 0.375,
+      "learning_rate": 9.464757147263849e-05,
+      "loss": 0.3978,
+      "step": 579
+    },
+    {
+      "epoch": 1.5977961432506889,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.435052274353074e-05,
+      "loss": 0.4008,
+      "step": 580
+    },
+    {
+      "epoch": 1.600550964187328,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.405352401555691e-05,
+      "loss": 0.3899,
+      "step": 581
+    },
+    {
+      "epoch": 1.603305785123967,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.37565779173271e-05,
+      "loss": 0.4085,
+      "step": 582
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.345968707698569e-05,
+      "loss": 0.3965,
+      "step": 583
+    },
+    {
+      "epoch": 1.6088154269972452,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.316285412218795e-05,
+      "loss": 0.4,
+      "step": 584
+    },
+    {
+      "epoch": 1.6115702479338843,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.286608168007678e-05,
+      "loss": 0.396,
+      "step": 585
+    },
+    {
+      "epoch": 1.6143250688705235,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.256937237725958e-05,
+      "loss": 0.4255,
+      "step": 586
+    },
+    {
+      "epoch": 1.6170798898071626,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.227272883978488e-05,
+      "loss": 0.3959,
+      "step": 587
+    },
+    {
+      "epoch": 1.6198347107438016,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.197615369311925e-05,
+      "loss": 0.4098,
+      "step": 588
+    },
+    {
+      "epoch": 1.6225895316804406,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.167964956212378e-05,
+      "loss": 0.4028,
+      "step": 589
+    },
+    {
+      "epoch": 1.6253443526170799,
+      "grad_norm": 0.458984375,
+      "learning_rate": 9.13832190710312e-05,
+      "loss": 0.4148,
+      "step": 590
+    },
+    {
+      "epoch": 1.6280991735537191,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.108686484342241e-05,
+      "loss": 0.3891,
+      "step": 591
+    },
+    {
+      "epoch": 1.6308539944903582,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.079058950220336e-05,
+      "loss": 0.4064,
+      "step": 592
+    },
+    {
+      "epoch": 1.6336088154269972,
+      "grad_norm": 0.375,
+      "learning_rate": 9.049439566958175e-05,
+      "loss": 0.414,
+      "step": 593
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.019828596704394e-05,
+      "loss": 0.4194,
+      "step": 594
+    },
+    {
+      "epoch": 1.6391184573002755,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.99022630153317e-05,
+      "loss": 0.4016,
+      "step": 595
+    },
+    {
+      "epoch": 1.6418732782369148,
+      "grad_norm": 0.380859375,
+      "learning_rate": 8.960632943441896e-05,
+      "loss": 0.4196,
+      "step": 596
+    },
+    {
+      "epoch": 1.6446280991735538,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.931048784348875e-05,
+      "loss": 0.4106,
+      "step": 597
+    },
+    {
+      "epoch": 1.6473829201101928,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.901474086090977e-05,
+      "loss": 0.4151,
+      "step": 598
+    },
+    {
+      "epoch": 1.6501377410468319,
+      "grad_norm": 0.380859375,
+      "learning_rate": 8.871909110421353e-05,
+      "loss": 0.4048,
+      "step": 599
+    },
+    {
+      "epoch": 1.6528925619834711,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.842354119007099e-05,
+      "loss": 0.4107,
+      "step": 600
+    },
+    {
+      "epoch": 1.6556473829201102,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.812809373426951e-05,
+      "loss": 0.391,
+      "step": 601
+    },
+    {
+      "epoch": 1.6584022038567494,
+      "grad_norm": 0.375,
+      "learning_rate": 8.783275135168945e-05,
+      "loss": 0.392,
+      "step": 602
+    },
+    {
+      "epoch": 1.6611570247933884,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.753751665628141e-05,
+      "loss": 0.4218,
+      "step": 603
+    },
+    {
+      "epoch": 1.6639118457300275,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.724239226104281e-05,
+      "loss": 0.408,
+      "step": 604
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.694738077799488e-05,
+      "loss": 0.3977,
+      "step": 605
+    },
+    {
+      "epoch": 1.6694214876033058,
+      "grad_norm": 0.380859375,
+      "learning_rate": 8.665248481815941e-05,
+      "loss": 0.3919,
+      "step": 606
+    },
+    {
+      "epoch": 1.672176308539945,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.635770699153587e-05,
+      "loss": 0.4258,
+      "step": 607
+    },
+    {
+      "epoch": 1.674931129476584,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.606304990707815e-05,
+      "loss": 0.4039,
+      "step": 608
+    },
+    {
+      "epoch": 1.677685950413223,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.57685161726715e-05,
+      "loss": 0.3994,
+      "step": 609
+    },
+    {
+      "epoch": 1.6804407713498621,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.547410839510939e-05,
+      "loss": 0.4064,
+      "step": 610
+    },
+    {
+      "epoch": 1.6831955922865014,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.517982918007059e-05,
+      "loss": 0.4104,
+      "step": 611
+    },
+    {
+      "epoch": 1.6859504132231407,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.488568113209593e-05,
+      "loss": 0.3915,
+      "step": 612
+    },
+    {
+      "epoch": 1.6887052341597797,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.459166685456547e-05,
+      "loss": 0.4006,
+      "step": 613
+    },
+    {
+      "epoch": 1.6914600550964187,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.42977889496751e-05,
+      "loss": 0.4068,
+      "step": 614
+    },
+    {
+      "epoch": 1.6942148760330578,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.400405001841399e-05,
+      "loss": 0.4321,
+      "step": 615
+    },
+    {
+      "epoch": 1.696969696969697,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.371045266054114e-05,
+      "loss": 0.3974,
+      "step": 616
+    },
+    {
+      "epoch": 1.699724517906336,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.34169994745626e-05,
+      "loss": 0.3953,
+      "step": 617
+    },
+    {
+      "epoch": 1.7024793388429753,
+      "grad_norm": 0.375,
+      "learning_rate": 8.312369305770843e-05,
+      "loss": 0.3908,
+      "step": 618
+    },
+    {
+      "epoch": 1.7052341597796143,
+      "grad_norm": 0.375,
+      "learning_rate": 8.283053600590962e-05,
+      "loss": 0.3852,
+      "step": 619
+    },
+    {
+      "epoch": 1.7079889807162534,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.253753091377533e-05,
+      "loss": 0.4014,
+      "step": 620
+    },
+    {
+      "epoch": 1.7107438016528924,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.224468037456969e-05,
+      "loss": 0.392,
+      "step": 621
+    },
+    {
+      "epoch": 1.7134986225895317,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.195198698018896e-05,
+      "loss": 0.4187,
+      "step": 622
+    },
+    {
+      "epoch": 1.716253443526171,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.165945332113857e-05,
+      "loss": 0.422,
+      "step": 623
+    },
+    {
+      "epoch": 1.71900826446281,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.136708198651021e-05,
+      "loss": 0.3927,
+      "step": 624
+    },
+    {
+      "epoch": 1.721763085399449,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.107487556395901e-05,
+      "loss": 0.4004,
+      "step": 625
+    },
+    {
+      "epoch": 1.724517906336088,
+      "grad_norm": 0.400390625,
+      "learning_rate": 8.07828366396803e-05,
+      "loss": 0.4061,
+      "step": 626
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.049096779838719e-05,
+      "loss": 0.3846,
+      "step": 627
+    },
+    {
+      "epoch": 1.7300275482093666,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.01992716232873e-05,
+      "loss": 0.406,
+      "step": 628
+    },
+    {
+      "epoch": 1.7327823691460056,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.990775069606012e-05,
+      "loss": 0.4048,
+      "step": 629
+    },
+    {
+      "epoch": 1.7355371900826446,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.961640759683416e-05,
+      "loss": 0.4172,
+      "step": 630
+    },
+    {
+      "epoch": 1.7382920110192837,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.932524490416385e-05,
+      "loss": 0.3989,
+      "step": 631
+    },
+    {
+      "epoch": 1.741046831955923,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.903426519500714e-05,
+      "loss": 0.3922,
+      "step": 632
+    },
+    {
+      "epoch": 1.743801652892562,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.874347104470234e-05,
+      "loss": 0.3937,
+      "step": 633
+    },
+    {
+      "epoch": 1.7465564738292012,
+      "grad_norm": 0.38671875,
+      "learning_rate": 7.845286502694551e-05,
+      "loss": 0.4166,
+      "step": 634
+    },
+    {
+      "epoch": 1.7493112947658402,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.816244971376757e-05,
+      "loss": 0.4166,
+      "step": 635
+    },
+    {
+      "epoch": 1.7520661157024793,
+      "grad_norm": 0.38671875,
+      "learning_rate": 7.787222767551164e-05,
+      "loss": 0.395,
+      "step": 636
+    },
+    {
+      "epoch": 1.7548209366391183,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.758220148081028e-05,
+      "loss": 0.3938,
+      "step": 637
+    },
+    {
+      "epoch": 1.7575757575757576,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.729237369656269e-05,
+      "loss": 0.4089,
+      "step": 638
+    },
+    {
+      "epoch": 1.7603305785123968,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.700274688791196e-05,
+      "loss": 0.3922,
+      "step": 639
+    },
+    {
+      "epoch": 1.7630853994490359,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.671332361822249e-05,
+      "loss": 0.402,
+      "step": 640
+    },
+    {
+      "epoch": 1.765840220385675,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.642410644905726e-05,
+      "loss": 0.4052,
+      "step": 641
+    },
+    {
+      "epoch": 1.768595041322314,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.613509794015517e-05,
+      "loss": 0.4091,
+      "step": 642
+    },
+    {
+      "epoch": 1.7713498622589532,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.584630064940819e-05,
+      "loss": 0.3989,
+      "step": 643
+    },
+    {
+      "epoch": 1.7741046831955924,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.555771713283907e-05,
+      "loss": 0.4085,
+      "step": 644
+    },
+    {
+      "epoch": 1.7768595041322315,
+      "grad_norm": 0.337890625,
+      "learning_rate": 7.526934994457844e-05,
+      "loss": 0.3883,
+      "step": 645
+    },
+    {
+      "epoch": 1.7796143250688705,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.498120163684239e-05,
+      "loss": 0.4039,
+      "step": 646
+    },
+    {
+      "epoch": 1.7823691460055096,
+      "grad_norm": 0.34765625,
+      "learning_rate": 7.469327475990966e-05,
+      "loss": 0.3922,
+      "step": 647
+    },
+    {
+      "epoch": 1.7851239669421488,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.440557186209926e-05,
+      "loss": 0.3944,
+      "step": 648
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.3925,
+      "step": 649
+    },
+    {
+      "epoch": 1.790633608815427,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.383084818718741e-05,
+      "loss": 0.4092,
+      "step": 650
+    },
+    {
+      "epoch": 1.7933884297520661,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.354383249672212e-05,
+      "loss": 0.3987,
+      "step": 651
+    },
+    {
+      "epoch": 1.7961432506887052,
+      "grad_norm": 0.34765625,
+      "learning_rate": 7.325705095860652e-05,
+      "loss": 0.4096,
+      "step": 652
+    },
+    {
+      "epoch": 1.7988980716253442,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.297050611102272e-05,
+      "loss": 0.3975,
+      "step": 653
+    },
+    {
+      "epoch": 1.8016528925619835,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.268420049005807e-05,
+      "loss": 0.4046,
+      "step": 654
+    },
+    {
+      "epoch": 1.8044077134986227,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.239813662968237e-05,
+      "loss": 0.3886,
+      "step": 655
+    },
+    {
+      "epoch": 1.8071625344352618,
+      "grad_norm": 0.3515625,
+      "learning_rate": 7.211231706172601e-05,
+      "loss": 0.4027,
+      "step": 656
+    },
+    {
+      "epoch": 1.8099173553719008,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.182674431585704e-05,
+      "loss": 0.3948,
+      "step": 657
+    },
+    {
+      "epoch": 1.8126721763085398,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.15414209195591e-05,
+      "loss": 0.404,
+      "step": 658
+    },
+    {
+      "epoch": 1.815426997245179,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.125634939810884e-05,
+      "loss": 0.4014,
+      "step": 659
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.3515625,
+      "learning_rate": 7.097153227455379e-05,
+      "loss": 0.3887,
+      "step": 660
+    },
+    {
+      "epoch": 1.8209366391184574,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.068697206968979e-05,
+      "loss": 0.3937,
+      "step": 661
+    },
+    {
+      "epoch": 1.8236914600550964,
+      "grad_norm": 0.375,
+      "learning_rate": 7.040267130203896e-05,
+      "loss": 0.3945,
+      "step": 662
+    },
+    {
+      "epoch": 1.8264462809917354,
+      "grad_norm": 0.44140625,
+      "learning_rate": 7.011863248782711e-05,
+      "loss": 0.427,
+      "step": 663
+    },
+    {
+      "epoch": 1.8292011019283747,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.983485814096165e-05,
+      "loss": 0.4112,
+      "step": 664
+    },
+    {
+      "epoch": 1.8319559228650137,
+      "grad_norm": 0.375,
+      "learning_rate": 6.955135077300931e-05,
+      "loss": 0.3874,
+      "step": 665
+    },
+    {
+      "epoch": 1.834710743801653,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.92681128931739e-05,
+      "loss": 0.395,
+      "step": 666
+    },
+    {
+      "epoch": 1.837465564738292,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.898514700827417e-05,
+      "loss": 0.4028,
+      "step": 667
+    },
+    {
+      "epoch": 1.840220385674931,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.870245562272138e-05,
+      "loss": 0.4038,
+      "step": 668
+    },
+    {
+      "epoch": 1.84297520661157,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.842004123849752e-05,
+      "loss": 0.3949,
+      "step": 669
+    },
+    {
+      "epoch": 1.8457300275482094,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.813790635513282e-05,
+      "loss": 0.3951,
+      "step": 670
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.785605346968386e-05,
+      "loss": 0.4097,
+      "step": 671
+    },
+    {
+      "epoch": 1.8512396694214877,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.757448507671128e-05,
+      "loss": 0.4104,
+      "step": 672
+    },
+    {
+      "epoch": 1.8539944903581267,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.729320366825784e-05,
+      "loss": 0.3977,
+      "step": 673
+    },
+    {
+      "epoch": 1.8567493112947657,
+      "grad_norm": 0.375,
+      "learning_rate": 6.701221173382636e-05,
+      "loss": 0.3932,
+      "step": 674
+    },
+    {
+      "epoch": 1.859504132231405,
+      "grad_norm": 0.400390625,
+      "learning_rate": 6.673151176035762e-05,
+      "loss": 0.4114,
+      "step": 675
+    },
+    {
+      "epoch": 1.8622589531680442,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.645110623220834e-05,
+      "loss": 0.4022,
+      "step": 676
+    },
+    {
+      "epoch": 1.8650137741046833,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.617099763112929e-05,
+      "loss": 0.4017,
+      "step": 677
+    },
+    {
+      "epoch": 1.8677685950413223,
+      "grad_norm": 0.345703125,
+      "learning_rate": 6.589118843624315e-05,
+      "loss": 0.4046,
+      "step": 678
+    },
+    {
+      "epoch": 1.8705234159779613,
+      "grad_norm": 0.375,
+      "learning_rate": 6.561168112402294e-05,
+      "loss": 0.3818,
+      "step": 679
+    },
+    {
+      "epoch": 1.8732782369146006,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.533247816826948e-05,
+      "loss": 0.4053,
+      "step": 680
+    },
+    {
+      "epoch": 1.8760330578512396,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.505358204009017e-05,
+      "loss": 0.3957,
+      "step": 681
+    },
+    {
+      "epoch": 1.878787878787879,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.477499520787665e-05,
+      "loss": 0.397,
+      "step": 682
+    },
+    {
+      "epoch": 1.881542699724518,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.449672013728318e-05,
+      "loss": 0.4106,
+      "step": 683
+    },
+    {
+      "epoch": 1.884297520661157,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.421875929120469e-05,
+      "loss": 0.3964,
+      "step": 684
+    },
+    {
+      "epoch": 1.887052341597796,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.394111512975504e-05,
+      "loss": 0.4169,
+      "step": 685
+    },
+    {
+      "epoch": 1.8898071625344353,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.366379011024535e-05,
+      "loss": 0.3912,
+      "step": 686
+    },
+    {
+      "epoch": 1.8925619834710745,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.338678668716209e-05,
+      "loss": 0.4056,
+      "step": 687
+    },
+    {
+      "epoch": 1.8953168044077136,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.311010731214533e-05,
+      "loss": 0.4081,
+      "step": 688
+    },
+    {
+      "epoch": 1.8980716253443526,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.283375443396726e-05,
+      "loss": 0.3948,
+      "step": 689
+    },
+    {
+      "epoch": 1.9008264462809916,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.25577304985103e-05,
+      "loss": 0.412,
+      "step": 690
+    },
+    {
+      "epoch": 1.9035812672176309,
+      "grad_norm": 0.3515625,
+      "learning_rate": 6.228203794874562e-05,
+      "loss": 0.407,
+      "step": 691
+    },
+    {
+      "epoch": 1.9063360881542701,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.200667922471132e-05,
+      "loss": 0.3945,
+      "step": 692
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.3955,
+      "step": 693
+    },
+    {
+      "epoch": 1.9118457300275482,
+      "grad_norm": 0.36328125,
+      "learning_rate": 6.145697299919226e-05,
+      "loss": 0.3903,
+      "step": 694
+    },
+    {
+      "epoch": 1.9146005509641872,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.118263036292479e-05,
+      "loss": 0.3874,
+      "step": 695
+    },
+    {
+      "epoch": 1.9173553719008265,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.090863128277938e-05,
+      "loss": 0.3856,
+      "step": 696
+    },
+    {
+      "epoch": 1.9201101928374655,
+      "grad_norm": 0.39453125,
+      "learning_rate": 6.063497818380587e-05,
+      "loss": 0.4032,
+      "step": 697
+    },
+    {
+      "epoch": 1.9228650137741048,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.036167348799219e-05,
+      "loss": 0.394,
+      "step": 698
+    },
+    {
+      "epoch": 1.9256198347107438,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.008871961424258e-05,
+      "loss": 0.3948,
+      "step": 699
+    },
+    {
+      "epoch": 1.9283746556473829,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.9816118978356394e-05,
+      "loss": 0.3953,
+      "step": 700
+    },
+    {
+      "epoch": 1.931129476584022,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.9543873993006496e-05,
+      "loss": 0.4042,
+      "step": 701
+    },
+    {
+      "epoch": 1.9338842975206612,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.9271987067718125e-05,
+      "loss": 0.3912,
+      "step": 702
+    },
+    {
+      "epoch": 1.9366391184573004,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.900046060884753e-05,
+      "loss": 0.4091,
+      "step": 703
+    },
+    {
+      "epoch": 1.9393939393939394,
+      "grad_norm": 0.375,
+      "learning_rate": 5.872929701956054e-05,
+      "loss": 0.3874,
+      "step": 704
+    },
+    {
+      "epoch": 1.9421487603305785,
+      "grad_norm": 0.34765625,
+      "learning_rate": 5.845849869981137e-05,
+      "loss": 0.3867,
+      "step": 705
+    },
+    {
+      "epoch": 1.9449035812672175,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.818806804632142e-05,
+      "loss": 0.379,
+      "step": 706
+    },
+    {
+      "epoch": 1.9476584022038568,
+      "grad_norm": 0.375,
+      "learning_rate": 5.7918007452558085e-05,
+      "loss": 0.3879,
+      "step": 707
+    },
+    {
+      "epoch": 1.950413223140496,
+      "grad_norm": 0.34375,
+      "learning_rate": 5.7648319308713464e-05,
+      "loss": 0.3876,
+      "step": 708
+    },
+    {
+      "epoch": 1.953168044077135,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.73790060016832e-05,
+      "loss": 0.3841,
+      "step": 709
+    },
+    {
+      "epoch": 1.955922865013774,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.711006991504551e-05,
+      "loss": 0.4022,
+      "step": 710
+    },
+    {
+      "epoch": 1.9586776859504131,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.684151342903992e-05,
+      "loss": 0.3988,
+      "step": 711
+    },
+    {
+      "epoch": 1.9614325068870524,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.657333892054643e-05,
+      "loss": 0.4002,
+      "step": 712
+    },
+    {
+      "epoch": 1.9641873278236914,
+      "grad_norm": 0.375,
+      "learning_rate": 5.630554876306407e-05,
+      "loss": 0.388,
+      "step": 713
+    },
+    {
+      "epoch": 1.9669421487603307,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.603814532669032e-05,
+      "loss": 0.395,
+      "step": 714
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.577113097809989e-05,
+      "loss": 0.3799,
+      "step": 715
+    },
+    {
+      "epoch": 1.9724517906336088,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.550450808052388e-05,
+      "loss": 0.3883,
+      "step": 716
+    },
+    {
+      "epoch": 1.9752066115702478,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.5238278993728756e-05,
+      "loss": 0.3942,
+      "step": 717
+    },
+    {
+      "epoch": 1.977961432506887,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.497244607399561e-05,
+      "loss": 0.4083,
+      "step": 718
+    },
+    {
+      "epoch": 1.9807162534435263,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.470701167409917e-05,
+      "loss": 0.3901,
+      "step": 719
+    },
+    {
+      "epoch": 1.9834710743801653,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.4441978143287066e-05,
+      "loss": 0.3921,
+      "step": 720
+    },
+    {
+      "epoch": 1.9862258953168044,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.417734782725896e-05,
+      "loss": 0.4082,
+      "step": 721
+    },
+    {
+      "epoch": 1.9889807162534434,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.39131230681459e-05,
+      "loss": 0.3827,
+      "step": 722
+    },
+    {
+      "epoch": 1.9917355371900827,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.364930620448946e-05,
+      "loss": 0.4046,
+      "step": 723
+    },
+    {
+      "epoch": 1.994490358126722,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.3385899571221175e-05,
+      "loss": 0.3847,
+      "step": 724
+    },
+    {
+      "epoch": 1.997245179063361,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.3122905499641615e-05,
+      "loss": 0.3954,
+      "step": 725
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.286032631740023e-05,
+      "loss": 0.3724,
+      "step": 726
+    },
+    {
+      "epoch": 2.0,
+      "eval_train_loss": 0.4305049180984497,
+      "eval_train_runtime": 8.5141,
+      "eval_train_samples_per_second": 158.207,
+      "eval_train_steps_per_second": 19.849,
+      "step": 726
+    },
+    {
+      "epoch": 2.002754820936639,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.259816434847426e-05,
+      "loss": 0.3496,
+      "step": 727
+    },
+    {
+      "epoch": 2.005509641873278,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.233642191314842e-05,
+      "loss": 0.3539,
+      "step": 728
+    },
+    {
+      "epoch": 2.0082644628099175,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.207510132799436e-05,
+      "loss": 0.3534,
+      "step": 729
+    },
+    {
+      "epoch": 2.0110192837465566,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.1814204905849906e-05,
+      "loss": 0.3523,
+      "step": 730
+    },
+    {
+      "epoch": 2.0137741046831956,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.155373495579911e-05,
+      "loss": 0.3512,
+      "step": 731
+    },
+    {
+      "epoch": 2.0165289256198347,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.1293693783151275e-05,
+      "loss": 0.3482,
+      "step": 732
+    },
+    {
+      "epoch": 2.0192837465564737,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.1034083689420905e-05,
+      "loss": 0.3623,
+      "step": 733
+    },
+    {
+      "epoch": 2.022038567493113,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.0774906972307044e-05,
+      "loss": 0.3445,
+      "step": 734
+    },
+    {
+      "epoch": 2.024793388429752,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.051616592567323e-05,
+      "loss": 0.3555,
+      "step": 735
+    },
+    {
+      "epoch": 2.0275482093663912,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.025786283952709e-05,
+      "loss": 0.3446,
+      "step": 736
+    },
+    {
+      "epoch": 2.0303030303030303,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.3495,
+      "step": 737
+    },
+    {
+      "epoch": 2.0330578512396693,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.9742579689326874e-05,
+      "loss": 0.3586,
+      "step": 738
+    },
+    {
+      "epoch": 2.0358126721763083,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.948560418582598e-05,
+      "loss": 0.3502,
+      "step": 739
+    },
+    {
+      "epoch": 2.038567493112948,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.9229075763878806e-05,
+      "loss": 0.3483,
+      "step": 740
+    },
+    {
+      "epoch": 2.041322314049587,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.8972996693910054e-05,
+      "loss": 0.3464,
+      "step": 741
+    },
+    {
+      "epoch": 2.044077134986226,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.871736924236714e-05,
+      "loss": 0.3538,
+      "step": 742
+    },
+    {
+      "epoch": 2.046831955922865,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.8462195671700574e-05,
+      "loss": 0.3398,
+      "step": 743
+    },
+    {
+      "epoch": 2.049586776859504,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.82074782403437e-05,
+      "loss": 0.3453,
+      "step": 744
+    },
+    {
+      "epoch": 2.0523415977961434,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.795321920269279e-05,
+      "loss": 0.3538,
+      "step": 745
+    },
+    {
+      "epoch": 2.0550964187327825,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.7699420809087044e-05,
+      "loss": 0.3464,
+      "step": 746
+    },
+    {
+      "epoch": 2.0578512396694215,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.7446085305788725e-05,
+      "loss": 0.3458,
+      "step": 747
+    },
+    {
+      "epoch": 2.0606060606060606,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.7193214934963206e-05,
+      "loss": 0.345,
+      "step": 748
+    },
+    {
+      "epoch": 2.0633608815426996,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.694081193465921e-05,
+      "loss": 0.3573,
+      "step": 749
+    },
+    {
+      "epoch": 2.0661157024793386,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.668887853878896e-05,
+      "loss": 0.3493,
+      "step": 750
+    },
+    {
+      "epoch": 2.068870523415978,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.643741697710837e-05,
+      "loss": 0.3369,
+      "step": 751
+    },
+    {
+      "epoch": 2.071625344352617,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.618642947519739e-05,
+      "loss": 0.3507,
+      "step": 752
+    },
+    {
+      "epoch": 2.074380165289256,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.593591825444028e-05,
+      "loss": 0.347,
+      "step": 753
+    },
+    {
+      "epoch": 2.077134986225895,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.5685885532005803e-05,
+      "loss": 0.344,
+      "step": 754
+    },
+    {
+      "epoch": 2.0798898071625342,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.543633352082794e-05,
+      "loss": 0.3525,
+      "step": 755
+    },
+    {
+      "epoch": 2.0826446280991737,
+      "grad_norm": 0.39453125,
+      "learning_rate": 4.518726442958599e-05,
+      "loss": 0.3545,
+      "step": 756
+    },
+    {
+      "epoch": 2.0853994490358128,
+      "grad_norm": 0.375,
+      "learning_rate": 4.493868046268514e-05,
+      "loss": 0.3466,
+      "step": 757
+    },
+    {
+      "epoch": 2.088154269972452,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.4690583820236933e-05,
+      "loss": 0.3598,
+      "step": 758
+    },
+    {
+      "epoch": 2.090909090909091,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.444297669803981e-05,
+      "loss": 0.359,
+      "step": 759
+    },
+    {
+      "epoch": 2.09366391184573,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.419586128755967e-05,
+      "loss": 0.3599,
+      "step": 760
+    },
+    {
+      "epoch": 2.0964187327823693,
+      "grad_norm": 0.39453125,
+      "learning_rate": 4.394923977591059e-05,
+      "loss": 0.3623,
+      "step": 761
+    },
+    {
+      "epoch": 2.0991735537190084,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.370311434583525e-05,
+      "loss": 0.3434,
+      "step": 762
+    },
+    {
+      "epoch": 2.1019283746556474,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.345748717568567e-05,
+      "loss": 0.3437,
+      "step": 763
+    },
+    {
+      "epoch": 2.1046831955922864,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.321236043940411e-05,
+      "loss": 0.3563,
+      "step": 764
+    },
+    {
+      "epoch": 2.1074380165289255,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.296773630650358e-05,
+      "loss": 0.3448,
+      "step": 765
+    },
+    {
+      "epoch": 2.110192837465565,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.272361694204894e-05,
+      "loss": 0.3412,
+      "step": 766
+    },
+    {
+      "epoch": 2.112947658402204,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.248000450663734e-05,
+      "loss": 0.3505,
+      "step": 767
+    },
+    {
+      "epoch": 2.115702479338843,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.223690115637944e-05,
+      "loss": 0.3565,
+      "step": 768
+    },
+    {
+      "epoch": 2.118457300275482,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.19943090428802e-05,
+      "loss": 0.3422,
+      "step": 769
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.17522303132198e-05,
+      "loss": 0.3523,
+      "step": 770
+    },
+    {
+      "epoch": 2.12396694214876,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.15106671099347e-05,
+      "loss": 0.3444,
+      "step": 771
+    },
+    {
+      "epoch": 2.1267217630853996,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.1269621570998665e-05,
+      "loss": 0.3528,
+      "step": 772
+    },
+    {
+      "epoch": 2.1294765840220387,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.10290958298038e-05,
+      "loss": 0.3409,
+      "step": 773
+    },
+    {
+      "epoch": 2.1322314049586777,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.0789092015141714e-05,
+      "loss": 0.3555,
+      "step": 774
+    },
+    {
+      "epoch": 2.1349862258953167,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.054961225118469e-05,
+      "loss": 0.3419,
+      "step": 775
+    },
+    {
+      "epoch": 2.1377410468319558,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.0310658657466816e-05,
+      "loss": 0.3552,
+      "step": 776
+    },
+    {
+      "epoch": 2.1404958677685952,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.007223334886531e-05,
+      "loss": 0.3433,
+      "step": 777
+    },
+    {
+      "epoch": 2.1432506887052343,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.983433843558176e-05,
+      "loss": 0.3716,
+      "step": 778
+    },
+    {
+      "epoch": 2.1460055096418733,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.9596976023123344e-05,
+      "loss": 0.3351,
+      "step": 779
+    },
+    {
+      "epoch": 2.1487603305785123,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.9360148212284475e-05,
+      "loss": 0.3482,
+      "step": 780
+    },
+    {
+      "epoch": 2.1515151515151514,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.9123857099127936e-05,
+      "loss": 0.3607,
+      "step": 781
+    },
+    {
+      "epoch": 2.1542699724517904,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.888810477496645e-05,
+      "loss": 0.3481,
+      "step": 782
+    },
+    {
+      "epoch": 2.15702479338843,
+      "grad_norm": 0.375,
+      "learning_rate": 3.865289332634407e-05,
+      "loss": 0.3511,
+      "step": 783
+    },
+    {
+      "epoch": 2.159779614325069,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.841822483501787e-05,
+      "loss": 0.3659,
+      "step": 784
+    },
+    {
+      "epoch": 2.162534435261708,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.8184101377939476e-05,
+      "loss": 0.3526,
+      "step": 785
+    },
+    {
+      "epoch": 2.165289256198347,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.7950525027236585e-05,
+      "loss": 0.349,
+      "step": 786
+    },
+    {
+      "epoch": 2.168044077134986,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.7717497850194614e-05,
+      "loss": 0.3534,
+      "step": 787
+    },
+    {
+      "epoch": 2.1707988980716255,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.748502190923863e-05,
+      "loss": 0.3456,
+      "step": 788
+    },
+    {
+      "epoch": 2.1735537190082646,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.725309926191479e-05,
+      "loss": 0.3515,
+      "step": 789
+    },
+    {
+      "epoch": 2.1763085399449036,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.7021731960872484e-05,
+      "loss": 0.3419,
+      "step": 790
+    },
+    {
+      "epoch": 2.1790633608815426,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.679092205384574e-05,
+      "loss": 0.3455,
+      "step": 791
+    },
+    {
+      "epoch": 2.1818181818181817,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.6560671583635467e-05,
+      "loss": 0.3436,
+      "step": 792
+    },
+    {
+      "epoch": 2.184573002754821,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.6330982588091186e-05,
+      "loss": 0.347,
+      "step": 793
+    },
+    {
+      "epoch": 2.18732782369146,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.610185710009304e-05,
+      "loss": 0.3509,
+      "step": 794
+    },
+    {
+      "epoch": 2.190082644628099,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.5873297147533915e-05,
+      "loss": 0.3465,
+      "step": 795
+    },
+    {
+      "epoch": 2.1928374655647382,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.5645304753301176e-05,
+      "loss": 0.3552,
+      "step": 796
+    },
+    {
+      "epoch": 2.1955922865013773,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.541788193525913e-05,
+      "loss": 0.3521,
+      "step": 797
+    },
+    {
+      "epoch": 2.1983471074380168,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.519103070623096e-05,
+      "loss": 0.3502,
+      "step": 798
+    },
+    {
+      "epoch": 2.201101928374656,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.496475307398095e-05,
+      "loss": 0.352,
+      "step": 799
+    },
+    {
+      "epoch": 2.203856749311295,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.4739051041196715e-05,
+      "loss": 0.3531,
+      "step": 800
+    },
+    {
+      "epoch": 2.206611570247934,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.45139266054715e-05,
+      "loss": 0.3463,
+      "step": 801
+    },
+    {
+      "epoch": 2.209366391184573,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.428938175928648e-05,
+      "loss": 0.3448,
+      "step": 802
+    },
+    {
+      "epoch": 2.212121212121212,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.406541848999312e-05,
+      "loss": 0.3358,
+      "step": 803
+    },
+    {
+      "epoch": 2.2148760330578514,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.384203877979559e-05,
+      "loss": 0.3404,
+      "step": 804
+    },
+    {
+      "epoch": 2.2176308539944904,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.361924460573325e-05,
+      "loss": 0.3488,
+      "step": 805
+    },
+    {
+      "epoch": 2.2203856749311295,
+      "grad_norm": 0.396484375,
+      "learning_rate": 3.33970379396631e-05,
+      "loss": 0.3549,
+      "step": 806
+    },
+    {
+      "epoch": 2.2231404958677685,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.3175420748242406e-05,
+      "loss": 0.3399,
+      "step": 807
+    },
+    {
+      "epoch": 2.2258953168044076,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.2954394992911083e-05,
+      "loss": 0.355,
+      "step": 808
+    },
+    {
+      "epoch": 2.228650137741047,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.273396262987475e-05,
+      "loss": 0.3513,
+      "step": 809
+    },
+    {
+      "epoch": 2.231404958677686,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.2514125610086955e-05,
+      "loss": 0.3545,
+      "step": 810
+    },
+    {
+      "epoch": 2.234159779614325,
+      "grad_norm": 0.375,
+      "learning_rate": 3.229488587923225e-05,
+      "loss": 0.3513,
+      "step": 811
+    },
+    {
+      "epoch": 2.236914600550964,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.207624537770868e-05,
+      "loss": 0.3367,
+      "step": 812
+    },
+    {
+      "epoch": 2.239669421487603,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.185820604061088e-05,
+      "loss": 0.3468,
+      "step": 813
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.164076979771287e-05,
+      "loss": 0.333,
+      "step": 814
+    },
+    {
+      "epoch": 2.2451790633608817,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.142393857345085e-05,
+      "loss": 0.3651,
+      "step": 815
+    },
+    {
+      "epoch": 2.2479338842975207,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.1207714286906256e-05,
+      "loss": 0.3501,
+      "step": 816
+    },
+    {
+      "epoch": 2.2506887052341598,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.099209885178882e-05,
+      "loss": 0.333,
+      "step": 817
+    },
+    {
+      "epoch": 2.253443526170799,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.077709417641953e-05,
+      "loss": 0.3457,
+      "step": 818
+    },
+    {
+      "epoch": 2.256198347107438,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.056270216371395e-05,
+      "loss": 0.3444,
+      "step": 819
+    },
+    {
+      "epoch": 2.2589531680440773,
+      "grad_norm": 0.375,
+      "learning_rate": 3.0348924711165006e-05,
+      "loss": 0.347,
+      "step": 820
+    },
+    {
+      "epoch": 2.2617079889807163,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.013576371082655e-05,
+      "loss": 0.3365,
+      "step": 821
+    },
+    {
+      "epoch": 2.2644628099173554,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.9923221049296446e-05,
+      "loss": 0.3408,
+      "step": 822
+    },
+    {
+      "epoch": 2.2672176308539944,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.971129860769992e-05,
+      "loss": 0.3423,
+      "step": 823
+    },
+    {
+      "epoch": 2.2699724517906334,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.9499998261672855e-05,
+      "loss": 0.3471,
+      "step": 824
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.3382,
+      "step": 825
+    },
+    {
+      "epoch": 2.275482093663912,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.9079271331324664e-05,
+      "loss": 0.3403,
+      "step": 826
+    },
+    {
+      "epoch": 2.278236914600551,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.8869848470679682e-05,
+      "loss": 0.3394,
+      "step": 827
+    },
+    {
+      "epoch": 2.28099173553719,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.8661055152923456e-05,
+      "loss": 0.358,
+      "step": 828
+    },
+    {
+      "epoch": 2.283746556473829,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.8452893225997346e-05,
+      "loss": 0.3349,
+      "step": 829
+    },
+    {
+      "epoch": 2.2865013774104685,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.8245364532254525e-05,
+      "loss": 0.337,
+      "step": 830
+    },
+    {
+      "epoch": 2.2892561983471076,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.8038470908443714e-05,
+      "loss": 0.3484,
+      "step": 831
+    },
+    {
+      "epoch": 2.2920110192837466,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.7832214185692884e-05,
+      "loss": 0.3402,
+      "step": 832
+    },
+    {
+      "epoch": 2.2947658402203857,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.7626596189492983e-05,
+      "loss": 0.3424,
+      "step": 833
+    },
+    {
+      "epoch": 2.2975206611570247,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.742161873968202e-05,
+      "loss": 0.3502,
+      "step": 834
+    },
+    {
+      "epoch": 2.3002754820936637,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.7217283650428683e-05,
+      "loss": 0.3502,
+      "step": 835
+    },
+    {
+      "epoch": 2.303030303030303,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.7013592730216465e-05,
+      "loss": 0.3415,
+      "step": 836
+    },
+    {
+      "epoch": 2.3057851239669422,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.681054778182748e-05,
+      "loss": 0.3289,
+      "step": 837
+    },
+    {
+      "epoch": 2.3085399449035813,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.6608150602326708e-05,
+      "loss": 0.3464,
+      "step": 838
+    },
+    {
+      "epoch": 2.3112947658402203,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.6406402983046053e-05,
+      "loss": 0.353,
+      "step": 839
+    },
+    {
+      "epoch": 2.3140495867768593,
+      "grad_norm": 0.375,
+      "learning_rate": 2.6205306709568354e-05,
+      "loss": 0.3582,
+      "step": 840
+    },
+    {
+      "epoch": 2.316804407713499,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.6004863561711635e-05,
+      "loss": 0.3436,
+      "step": 841
+    },
+    {
+      "epoch": 2.319559228650138,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.580507531351345e-05,
+      "loss": 0.3421,
+      "step": 842
+    },
+    {
+      "epoch": 2.322314049586777,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.5605943733215042e-05,
+      "loss": 0.3429,
+      "step": 843
+    },
+    {
+      "epoch": 2.325068870523416,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.540747058324593e-05,
+      "loss": 0.3563,
+      "step": 844
+    },
+    {
+      "epoch": 2.327823691460055,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.5209657620207915e-05,
+      "loss": 0.3417,
+      "step": 845
+    },
+    {
+      "epoch": 2.330578512396694,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.501250659485992e-05,
+      "loss": 0.3553,
+      "step": 846
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.4816019252102273e-05,
+      "loss": 0.344,
+      "step": 847
+    },
+    {
+      "epoch": 2.3360881542699725,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.462019733096136e-05,
+      "loss": 0.3448,
+      "step": 848
+    },
+    {
+      "epoch": 2.3388429752066116,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.4425042564574184e-05,
+      "loss": 0.347,
+      "step": 849
+    },
+    {
+      "epoch": 2.3415977961432506,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.4230556680173045e-05,
+      "loss": 0.3585,
+      "step": 850
+    },
+    {
+      "epoch": 2.3443526170798896,
+      "grad_norm": 0.375,
+      "learning_rate": 2.403674139907025e-05,
+      "loss": 0.3375,
+      "step": 851
+    },
+    {
+      "epoch": 2.347107438016529,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.38435984366429e-05,
+      "loss": 0.3456,
+      "step": 852
+    },
+    {
+      "epoch": 2.349862258953168,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.3651129502317647e-05,
+      "loss": 0.3514,
+      "step": 853
+    },
+    {
+      "epoch": 2.352617079889807,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.3459336299555657e-05,
+      "loss": 0.3505,
+      "step": 854
+    },
+    {
+      "epoch": 2.355371900826446,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.3268220525837437e-05,
+      "loss": 0.3349,
+      "step": 855
+    },
+    {
+      "epoch": 2.3581267217630852,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.3077783872647896e-05,
+      "loss": 0.3483,
+      "step": 856
+    },
+    {
+      "epoch": 2.3608815426997243,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.288802802546124e-05,
+      "loss": 0.3359,
+      "step": 857
+    },
+    {
+      "epoch": 2.3636363636363638,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.26989546637263e-05,
+      "loss": 0.3425,
+      "step": 858
+    },
+    {
+      "epoch": 2.366391184573003,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.2510565460851418e-05,
+      "loss": 0.3522,
+      "step": 859
+    },
+    {
+      "epoch": 2.369146005509642,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.2322862084189744e-05,
+      "loss": 0.3366,
+      "step": 860
+    },
+    {
+      "epoch": 2.371900826446281,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.2135846195024513e-05,
+      "loss": 0.3492,
+      "step": 861
+    },
+    {
+      "epoch": 2.3746556473829203,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.194951944855418e-05,
+      "loss": 0.3396,
+      "step": 862
+    },
+    {
+      "epoch": 2.3774104683195594,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.1763883493878057e-05,
+      "loss": 0.3376,
+      "step": 863
+    },
+    {
+      "epoch": 2.3801652892561984,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.157893997398146e-05,
+      "loss": 0.3447,
+      "step": 864
+    },
+    {
+      "epoch": 2.3829201101928374,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.139469052572127e-05,
+      "loss": 0.3403,
+      "step": 865
+    },
+    {
+      "epoch": 2.3856749311294765,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.1211136779811402e-05,
+      "loss": 0.3417,
+      "step": 866
+    },
+    {
+      "epoch": 2.3884297520661155,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.1028280360808407e-05,
+      "loss": 0.3458,
+      "step": 867
+    },
+    {
+      "epoch": 2.391184573002755,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.0846122887097186e-05,
+      "loss": 0.3493,
+      "step": 868
+    },
+    {
+      "epoch": 2.393939393939394,
+      "grad_norm": 0.375,
+      "learning_rate": 2.0664665970876496e-05,
+      "loss": 0.3479,
+      "step": 869
+    },
+    {
+      "epoch": 2.396694214876033,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.0483911218144715e-05,
+      "loss": 0.3508,
+      "step": 870
+    },
+    {
+      "epoch": 2.399449035812672,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.0303860228685724e-05,
+      "loss": 0.3469,
+      "step": 871
+    },
+    {
+      "epoch": 2.402203856749311,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.0124514596054668e-05,
+      "loss": 0.3447,
+      "step": 872
+    },
+    {
+      "epoch": 2.4049586776859506,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.994587590756397e-05,
+      "loss": 0.3457,
+      "step": 873
+    },
+    {
+      "epoch": 2.4077134986225897,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.9767945744269034e-05,
+      "loss": 0.3392,
+      "step": 874
+    },
+    {
+      "epoch": 2.4104683195592287,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.9590725680954513e-05,
+      "loss": 0.3358,
+      "step": 875
+    },
+    {
+      "epoch": 2.4132231404958677,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.941421728612023e-05,
+      "loss": 0.3603,
+      "step": 876
+    },
+    {
+      "epoch": 2.4159779614325068,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.923842212196735e-05,
+      "loss": 0.3297,
+      "step": 877
+    },
+    {
+      "epoch": 2.418732782369146,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.90633417443845e-05,
+      "loss": 0.3533,
+      "step": 878
+    },
+    {
+      "epoch": 2.4214876033057853,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.8888977702934085e-05,
+      "loss": 0.3393,
+      "step": 879
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.8715331540838487e-05,
+      "loss": 0.3468,
+      "step": 880
+    },
+    {
+      "epoch": 2.4269972451790633,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.854240479496643e-05,
+      "loss": 0.3451,
+      "step": 881
+    },
+    {
+      "epoch": 2.4297520661157024,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.837019899581943e-05,
+      "loss": 0.354,
+      "step": 882
+    },
+    {
+      "epoch": 2.4325068870523414,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.8198715667518175e-05,
+      "loss": 0.351,
+      "step": 883
+    },
+    {
+      "epoch": 2.435261707988981,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.8027956327789076e-05,
+      "loss": 0.3438,
+      "step": 884
+    },
+    {
+      "epoch": 2.43801652892562,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.7857922487950874e-05,
+      "loss": 0.3542,
+      "step": 885
+    },
+    {
+      "epoch": 2.440771349862259,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.7688615652901076e-05,
+      "loss": 0.3375,
+      "step": 886
+    },
+    {
+      "epoch": 2.443526170798898,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.7520037321102966e-05,
+      "loss": 0.3515,
+      "step": 887
+    },
+    {
+      "epoch": 2.446280991735537,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.7352188984572026e-05,
+      "loss": 0.3366,
+      "step": 888
+    },
+    {
+      "epoch": 2.449035812672176,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.7185072128862933e-05,
+      "loss": 0.3351,
+      "step": 889
+    },
+    {
+      "epoch": 2.4517906336088156,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.7018688233056258e-05,
+      "loss": 0.358,
+      "step": 890
+    },
+    {
+      "epoch": 2.4545454545454546,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.3571,
+      "step": 891
+    },
+    {
+      "epoch": 2.4573002754820936,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.6688125205023985e-05,
+      "loss": 0.3429,
+      "step": 892
+    },
+    {
+      "epoch": 2.4600550964187327,
+      "grad_norm": 0.375,
+      "learning_rate": 1.6523948998471973e-05,
+      "loss": 0.3408,
+      "step": 893
+    },
+    {
+      "epoch": 2.462809917355372,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.6360511603143648e-05,
+      "loss": 0.3453,
+      "step": 894
+    },
+    {
+      "epoch": 2.465564738292011,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.619781446555422e-05,
+      "loss": 0.357,
+      "step": 895
+    },
+    {
+      "epoch": 2.46831955922865,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.6035859025667265e-05,
+      "loss": 0.3471,
+      "step": 896
+    },
+    {
+      "epoch": 2.4710743801652892,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.587464671688187e-05,
+      "loss": 0.3484,
+      "step": 897
+    },
+    {
+      "epoch": 2.4738292011019283,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.5714178966020066e-05,
+      "loss": 0.3379,
+      "step": 898
+    },
+    {
+      "epoch": 2.4765840220385673,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.5554457193313987e-05,
+      "loss": 0.3482,
+      "step": 899
+    },
+    {
+      "epoch": 2.479338842975207,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.5395482812393514e-05,
+      "loss": 0.3402,
+      "step": 900
+    },
+    {
+      "epoch": 2.482093663911846,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.523725723027365e-05,
+      "loss": 0.3403,
+      "step": 901
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.5079781847342123e-05,
+      "loss": 0.3519,
+      "step": 902
+    },
+    {
+      "epoch": 2.487603305785124,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.4923058057346929e-05,
+      "loss": 0.3507,
+      "step": 903
+    },
+    {
+      "epoch": 2.490358126721763,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.4767087247384059e-05,
+      "loss": 0.3421,
+      "step": 904
+    },
+    {
+      "epoch": 2.4931129476584024,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.4611870797885197e-05,
+      "loss": 0.3679,
+      "step": 905
+    },
+    {
+      "epoch": 2.4958677685950414,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.4457410082605483e-05,
+      "loss": 0.3449,
+      "step": 906
+    },
+    {
+      "epoch": 2.4986225895316805,
+      "grad_norm": 0.375,
+      "learning_rate": 1.430370646861139e-05,
+      "loss": 0.347,
+      "step": 907
+    },
+    {
+      "epoch": 2.5013774104683195,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.415076131626859e-05,
+      "loss": 0.356,
+      "step": 908
+    },
+    {
+      "epoch": 2.5041322314049586,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.3998575979229944e-05,
+      "loss": 0.343,
+      "step": 909
+    },
+    {
+      "epoch": 2.5068870523415976,
+      "grad_norm": 0.375,
+      "learning_rate": 1.3847151804423552e-05,
+      "loss": 0.3509,
+      "step": 910
+    },
+    {
+      "epoch": 2.509641873278237,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.3696490132040652e-05,
+      "loss": 0.3465,
+      "step": 911
+    },
+    {
+      "epoch": 2.512396694214876,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.3546592295524074e-05,
+      "loss": 0.3472,
+      "step": 912
+    },
+    {
+      "epoch": 2.515151515151515,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.347,
+      "step": 913
+    },
+    {
+      "epoch": 2.517906336088154,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.3249093430047088e-05,
+      "loss": 0.3417,
+      "step": 914
+    },
+    {
+      "epoch": 2.5206611570247937,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.3101495034123313e-05,
+      "loss": 0.3546,
+      "step": 915
+    },
+    {
+      "epoch": 2.5234159779614327,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.2954665740115779e-05,
+      "loss": 0.342,
+      "step": 916
+    },
+    {
+      "epoch": 2.5261707988980717,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.280860684754852e-05,
+      "loss": 0.3425,
+      "step": 917
+    },
+    {
+      "epoch": 2.5289256198347108,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.2663319649127026e-05,
+      "loss": 0.336,
+      "step": 918
+    },
+    {
+      "epoch": 2.53168044077135,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.2518805430726787e-05,
+      "loss": 0.351,
+      "step": 919
+    },
+    {
+      "epoch": 2.534435261707989,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.2375065471382064e-05,
+      "loss": 0.3572,
+      "step": 920
+    },
+    {
+      "epoch": 2.537190082644628,
+      "grad_norm": 0.375,
+      "learning_rate": 1.2232101043274436e-05,
+      "loss": 0.3556,
+      "step": 921
+    },
+    {
+      "epoch": 2.5399449035812673,
+      "grad_norm": 0.375,
+      "learning_rate": 1.2089913411721631e-05,
+      "loss": 0.3488,
+      "step": 922
+    },
+    {
+      "epoch": 2.5426997245179064,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.1948503835166191e-05,
+      "loss": 0.3513,
+      "step": 923
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.1807873565164506e-05,
+      "loss": 0.3409,
+      "step": 924
+    },
+    {
+      "epoch": 2.5482093663911844,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.166802384637561e-05,
+      "loss": 0.3395,
+      "step": 925
+    },
+    {
+      "epoch": 2.550964187327824,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.1528955916550188e-05,
+      "loss": 0.3489,
+      "step": 926
+    },
+    {
+      "epoch": 2.553719008264463,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.139067100651976e-05,
+      "loss": 0.3444,
+      "step": 927
+    },
+    {
+      "epoch": 2.556473829201102,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.1253170340185493e-05,
+      "loss": 0.3458,
+      "step": 928
+    },
+    {
+      "epoch": 2.559228650137741,
+      "grad_norm": 0.375,
+      "learning_rate": 1.1116455134507664e-05,
+      "loss": 0.3469,
+      "step": 929
+    },
+    {
+      "epoch": 2.56198347107438,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.0980526599494733e-05,
+      "loss": 0.3608,
+      "step": 930
+    },
+    {
+      "epoch": 2.564738292011019,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.0845385938192686e-05,
+      "loss": 0.3433,
+      "step": 931
+    },
+    {
+      "epoch": 2.567493112947658,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.0711034346674364e-05,
+      "loss": 0.3546,
+      "step": 932
+    },
+    {
+      "epoch": 2.5702479338842976,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.057747301402887e-05,
+      "loss": 0.353,
+      "step": 933
+    },
+    {
+      "epoch": 2.5730027548209367,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.0444703122351095e-05,
+      "loss": 0.3427,
+      "step": 934
+    },
+    {
+      "epoch": 2.5757575757575757,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.0312725846731175e-05,
+      "loss": 0.3409,
+      "step": 935
+    },
+    {
+      "epoch": 2.5785123966942147,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.0181542355244167e-05,
+      "loss": 0.3534,
+      "step": 936
+    },
+    {
+      "epoch": 2.581267217630854,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.0051153808939685e-05,
+      "loss": 0.3339,
+      "step": 937
+    },
+    {
+      "epoch": 2.5840220385674932,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.921561361831599e-06,
+      "loss": 0.3385,
+      "step": 938
+    },
+    {
+      "epoch": 2.5867768595041323,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.792766160887868e-06,
+      "loss": 0.3451,
+      "step": 939
+    },
+    {
+      "epoch": 2.5895316804407713,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.664769346020297e-06,
+      "loss": 0.3578,
+      "step": 940
+    },
+    {
+      "epoch": 2.5922865013774103,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.537572050074618e-06,
+      "loss": 0.3462,
+      "step": 941
+    },
+    {
+      "epoch": 2.5950413223140494,
+      "grad_norm": 0.375,
+      "learning_rate": 9.411175398820271e-06,
+      "loss": 0.3459,
+      "step": 942
+    },
+    {
+      "epoch": 2.597796143250689,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.285580510940594e-06,
+      "loss": 0.3479,
+      "step": 943
+    },
+    {
+      "epoch": 2.600550964187328,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.160788498022777e-06,
+      "loss": 0.3484,
+      "step": 944
+    },
+    {
+      "epoch": 2.603305785123967,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.036800464548157e-06,
+      "loss": 0.3279,
+      "step": 945
+    },
+    {
+      "epoch": 2.606060606060606,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.91361750788241e-06,
+      "loss": 0.3374,
+      "step": 946
+    },
+    {
+      "epoch": 2.6088154269972454,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.791240718265792e-06,
+      "loss": 0.3392,
+      "step": 947
+    },
+    {
+      "epoch": 2.6115702479338845,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.669671178803485e-06,
+      "loss": 0.3455,
+      "step": 948
+    },
+    {
+      "epoch": 2.6143250688705235,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.548909965456065e-06,
+      "loss": 0.3538,
+      "step": 949
+    },
+    {
+      "epoch": 2.6170798898071626,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.428958147029965e-06,
+      "loss": 0.3374,
+      "step": 950
+    },
+    {
+      "epoch": 2.6198347107438016,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.309816785168034e-06,
+      "loss": 0.3486,
+      "step": 951
+    },
+    {
+      "epoch": 2.6225895316804406,
+      "grad_norm": 0.375,
+      "learning_rate": 8.191486934340031e-06,
+      "loss": 0.3423,
+      "step": 952
+    },
+    {
+      "epoch": 2.6253443526170797,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.073969641833445e-06,
+      "loss": 0.3401,
+      "step": 953
+    },
+    {
+      "epoch": 2.628099173553719,
+      "grad_norm": 0.375,
+      "learning_rate": 7.95726594774413e-06,
+      "loss": 0.3359,
+      "step": 954
+    },
+    {
+      "epoch": 2.630853994490358,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.841376884967121e-06,
+      "loss": 0.3478,
+      "step": 955
+    },
+    {
+      "epoch": 2.633608815426997,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.726303479187501e-06,
+      "loss": 0.3456,
+      "step": 956
+    },
+    {
+      "epoch": 2.6363636363636362,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.3355,
+      "step": 957
+    },
+    {
+      "epoch": 2.6391184573002757,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.498607705256588e-06,
+      "loss": 0.3293,
+      "step": 958
+    },
+    {
+      "epoch": 2.6418732782369148,
+      "grad_norm": 0.375,
+      "learning_rate": 7.385987352344282e-06,
+      "loss": 0.3516,
+      "step": 959
+    },
+    {
+      "epoch": 2.644628099173554,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.2741866868895395e-06,
+      "loss": 0.3428,
+      "step": 960
+    },
+    {
+      "epoch": 2.647382920110193,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.163206698392744e-06,
+      "loss": 0.3452,
+      "step": 961
+    },
+    {
+      "epoch": 2.650137741046832,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.053048369090853e-06,
+      "loss": 0.3405,
+      "step": 962
+    },
+    {
+      "epoch": 2.652892561983471,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.943712673948644e-06,
+      "loss": 0.3427,
+      "step": 963
+    },
+    {
+      "epoch": 2.65564738292011,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.835200580650125e-06,
+      "loss": 0.3546,
+      "step": 964
+    },
+    {
+      "epoch": 2.6584022038567494,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.7275130495899175e-06,
+      "loss": 0.3489,
+      "step": 965
+    },
+    {
+      "epoch": 2.6611570247933884,
+      "grad_norm": 0.375,
+      "learning_rate": 6.620651033864844e-06,
+      "loss": 0.3522,
+      "step": 966
+    },
+    {
+      "epoch": 2.6639118457300275,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.514615479265396e-06,
+      "loss": 0.3381,
+      "step": 967
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.409407324267447e-06,
+      "loss": 0.3495,
+      "step": 968
+    },
+    {
+      "epoch": 2.669421487603306,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.3050275000238414e-06,
+      "loss": 0.3355,
+      "step": 969
+    },
+    {
+      "epoch": 2.672176308539945,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.201476930356264e-06,
+      "loss": 0.3351,
+      "step": 970
+    },
+    {
+      "epoch": 2.674931129476584,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.098756531747041e-06,
+      "loss": 0.3415,
+      "step": 971
+    },
+    {
+      "epoch": 2.677685950413223,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.996867213330992e-06,
+      "loss": 0.3412,
+      "step": 972
+    },
+    {
+      "epoch": 2.680440771349862,
+      "grad_norm": 0.388671875,
+      "learning_rate": 5.895809876887326e-06,
+      "loss": 0.3507,
+      "step": 973
+    },
+    {
+      "epoch": 2.683195592286501,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.795585416831828e-06,
+      "loss": 0.3415,
+      "step": 974
+    },
+    {
+      "epoch": 2.6859504132231407,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.696194720208792e-06,
+      "loss": 0.3489,
+      "step": 975
+    },
+    {
+      "epoch": 2.6887052341597797,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.597638666683258e-06,
+      "loss": 0.3528,
+      "step": 976
+    },
+    {
+      "epoch": 2.6914600550964187,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.499918128533155e-06,
+      "loss": 0.339,
+      "step": 977
+    },
+    {
+      "epoch": 2.6942148760330578,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.403033970641647e-06,
+      "loss": 0.3456,
+      "step": 978
+    },
+    {
+      "epoch": 2.6969696969696972,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.306987050489442e-06,
+      "loss": 0.3514,
+      "step": 979
+    },
+    {
+      "epoch": 2.6997245179063363,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.211778218147201e-06,
+      "loss": 0.3339,
+      "step": 980
+    },
+    {
+      "epoch": 2.7024793388429753,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.1174083162680465e-06,
+      "loss": 0.3469,
+      "step": 981
+    },
+    {
+      "epoch": 2.7052341597796143,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.023878180080055e-06,
+      "loss": 0.3561,
+      "step": 982
+    },
+    {
+      "epoch": 2.7079889807162534,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.931188637378914e-06,
+      "loss": 0.3526,
+      "step": 983
+    },
+    {
+      "epoch": 2.7107438016528924,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.839340508520562e-06,
+      "loss": 0.3433,
+      "step": 984
+    },
+    {
+      "epoch": 2.7134986225895315,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.748334606413951e-06,
+      "loss": 0.3222,
+      "step": 985
+    },
+    {
+      "epoch": 2.716253443526171,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.658171736513828e-06,
+      "loss": 0.3655,
+      "step": 986
+    },
+    {
+      "epoch": 2.71900826446281,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.568852696813619e-06,
+      "loss": 0.3383,
+      "step": 987
+    },
+    {
+      "epoch": 2.721763085399449,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.4803782778383816e-06,
+      "loss": 0.3581,
+      "step": 988
+    },
+    {
+      "epoch": 2.724517906336088,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.392749262637752e-06,
+      "loss": 0.3525,
+      "step": 989
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.375,
+      "learning_rate": 4.305966426779118e-06,
+      "loss": 0.3402,
+      "step": 990
+    },
+    {
+      "epoch": 2.7300275482093666,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.220030538340636e-06,
+      "loss": 0.357,
+      "step": 991
+    },
+    {
+      "epoch": 2.7327823691460056,
+      "grad_norm": 0.375,
+      "learning_rate": 4.1349423579045145e-06,
+      "loss": 0.3313,
+      "step": 992
+    },
+    {
+      "epoch": 2.7355371900826446,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.050702638550275e-06,
+      "loss": 0.3541,
+      "step": 993
+    },
+    {
+      "epoch": 2.7382920110192837,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.967312125847999e-06,
+      "loss": 0.3433,
+      "step": 994
+    },
+    {
+      "epoch": 2.7410468319559227,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.884771557851885e-06,
+      "loss": 0.3429,
+      "step": 995
+    },
+    {
+      "epoch": 2.7438016528925617,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.8030816650935776e-06,
+      "loss": 0.3543,
+      "step": 996
+    },
+    {
+      "epoch": 2.746556473829201,
+      "grad_norm": 0.375,
+      "learning_rate": 3.722243170575801e-06,
+      "loss": 0.3393,
+      "step": 997
+    },
+    {
+      "epoch": 2.7493112947658402,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.64225678976583e-06,
+      "loss": 0.3469,
+      "step": 998
+    },
+    {
+      "epoch": 2.7520661157024793,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.5631232305893046e-06,
+      "loss": 0.3415,
+      "step": 999
+    },
+    {
+      "epoch": 2.7548209366391183,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.4848431934239255e-06,
+      "loss": 0.3452,
+      "step": 1000
+    },
+    {
+      "epoch": 2.757575757575758,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.40741737109318e-06,
+      "loss": 0.3447,
+      "step": 1001
+    },
+    {
+      "epoch": 2.760330578512397,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.330846448860259e-06,
+      "loss": 0.3475,
+      "step": 1002
+    },
+    {
+      "epoch": 2.763085399449036,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.2551311044220157e-06,
+      "loss": 0.3577,
+      "step": 1003
+    },
+    {
+      "epoch": 2.765840220385675,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.18027200790294e-06,
+      "loss": 0.347,
+      "step": 1004
+    },
+    {
+      "epoch": 2.768595041322314,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.1062698218492724e-06,
+      "loss": 0.3375,
+      "step": 1005
+    },
+    {
+      "epoch": 2.771349862258953,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.033125201223019e-06,
+      "loss": 0.3459,
+      "step": 1006
+    },
+    {
+      "epoch": 2.7741046831955924,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.9608387933963035e-06,
+      "loss": 0.338,
+      "step": 1007
+    },
+    {
+      "epoch": 2.7768595041322315,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.889411238145545e-06,
+      "loss": 0.3471,
+      "step": 1008
+    },
+    {
+      "epoch": 2.7796143250688705,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.818843167645835e-06,
+      "loss": 0.3461,
+      "step": 1009
+    },
+    {
+      "epoch": 2.7823691460055096,
+      "grad_norm": 0.375,
+      "learning_rate": 2.7491352064653143e-06,
+      "loss": 0.3487,
+      "step": 1010
+    },
+    {
+      "epoch": 2.785123966942149,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.6802879715596585e-06,
+      "loss": 0.3351,
+      "step": 1011
+    },
+    {
+      "epoch": 2.787878787878788,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.612302072266637e-06,
+      "loss": 0.3424,
+      "step": 1012
+    },
+    {
+      "epoch": 2.790633608815427,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.5451781103006944e-06,
+      "loss": 0.3343,
+      "step": 1013
+    },
+    {
+      "epoch": 2.793388429752066,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.4789166797476228e-06,
+      "loss": 0.3424,
+      "step": 1014
+    },
+    {
+      "epoch": 2.796143250688705,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.41351836705932e-06,
+      "loss": 0.3461,
+      "step": 1015
+    },
+    {
+      "epoch": 2.798898071625344,
+      "grad_norm": 0.375,
+      "learning_rate": 2.3489837510485946e-06,
+      "loss": 0.3513,
+      "step": 1016
+    },
+    {
+      "epoch": 2.8016528925619832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.2853134028840594e-06,
+      "loss": 0.3559,
+      "step": 1017
+    },
+    {
+      "epoch": 2.8044077134986227,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.2225078860849904e-06,
+      "loss": 0.3562,
+      "step": 1018
+    },
+    {
+      "epoch": 2.8071625344352618,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.160567756516507e-06,
+      "loss": 0.3552,
+      "step": 1019
+    },
+    {
+      "epoch": 2.809917355371901,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.0994935623844692e-06,
+      "loss": 0.3431,
+      "step": 1020
+    },
+    {
+      "epoch": 2.81267217630854,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.039285844230765e-06,
+      "loss": 0.3501,
+      "step": 1021
+    },
+    {
+      "epoch": 2.8154269972451793,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.9799451349284183e-06,
+      "loss": 0.3336,
+      "step": 1022
+    },
+    {
+      "epoch": 2.8181818181818183,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.921471959676957e-06,
+      "loss": 0.3488,
+      "step": 1023
+    },
+    {
+      "epoch": 2.8209366391184574,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.8638668359977296e-06,
+      "loss": 0.3534,
+      "step": 1024
+    },
+    {
+      "epoch": 2.8236914600550964,
+      "grad_norm": 0.375,
+      "learning_rate": 1.8071302737293295e-06,
+      "loss": 0.3486,
+      "step": 1025
+    },
+    {
+      "epoch": 2.8264462809917354,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.751262775023077e-06,
+      "loss": 0.3549,
+      "step": 1026
+    },
+    {
+      "epoch": 2.8292011019283745,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.6962648343385568e-06,
+      "loss": 0.3465,
+      "step": 1027
+    },
+    {
+      "epoch": 2.8319559228650135,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.642136938439287e-06,
+      "loss": 0.3466,
+      "step": 1028
+    },
+    {
+      "epoch": 2.834710743801653,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.5888795663883904e-06,
+      "loss": 0.3452,
+      "step": 1029
+    },
+    {
+      "epoch": 2.837465564738292,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.5364931895443413e-06,
+      "loss": 0.3332,
+      "step": 1030
+    },
+    {
+      "epoch": 2.840220385674931,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.4849782715568139e-06,
+      "loss": 0.3395,
+      "step": 1031
+    },
+    {
+      "epoch": 2.84297520661157,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.4343352683625411e-06,
+      "loss": 0.3416,
+      "step": 1032
+    },
+    {
+      "epoch": 2.8457300275482096,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.3845646281813507e-06,
+      "loss": 0.3235,
+      "step": 1033
+    },
+    {
+      "epoch": 2.8484848484848486,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.3356667915121025e-06,
+      "loss": 0.3409,
+      "step": 1034
+    },
+    {
+      "epoch": 2.8512396694214877,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.2876421911288905e-06,
+      "loss": 0.351,
+      "step": 1035
+    },
+    {
+      "epoch": 2.8539944903581267,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.240491252077125e-06,
+      "loss": 0.3493,
+      "step": 1036
+    },
+    {
+      "epoch": 2.8567493112947657,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.1942143916698457e-06,
+      "loss": 0.3433,
+      "step": 1037
+    },
+    {
+      "epoch": 2.8595041322314048,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.148812019483958e-06,
+      "loss": 0.3431,
+      "step": 1038
+    },
+    {
+      "epoch": 2.8622589531680442,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.104284537356659e-06,
+      "loss": 0.3519,
+      "step": 1039
+    },
+    {
+      "epoch": 2.8650137741046833,
+      "grad_norm": 0.375,
+      "learning_rate": 1.060632339381873e-06,
+      "loss": 0.3411,
+      "step": 1040
+    },
+    {
+      "epoch": 2.8677685950413223,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.0178558119067315e-06,
+      "loss": 0.3469,
+      "step": 1041
+    },
+    {
+      "epoch": 2.8705234159779613,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.759553335281891e-07,
+      "loss": 0.3529,
+      "step": 1042
+    },
+    {
+      "epoch": 2.873278236914601,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.349312750896455e-07,
+      "loss": 0.3446,
+      "step": 1043
+    },
+    {
+      "epoch": 2.87603305785124,
+      "grad_norm": 0.361328125,
+      "learning_rate": 8.947839996777285e-07,
+      "loss": 0.3469,
+      "step": 1044
+    },
+    {
+      "epoch": 2.878787878787879,
+      "grad_norm": 0.39453125,
+      "learning_rate": 8.555138626189618e-07,
+      "loss": 0.3447,
+      "step": 1045
+    },
+    {
+      "epoch": 2.881542699724518,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.171212114767345e-07,
+      "loss": 0.336,
+      "step": 1046
+    },
+    {
+      "epoch": 2.884297520661157,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.796063860481595e-07,
+      "loss": 0.343,
+      "step": 1047
+    },
+    {
+      "epoch": 2.887052341597796,
+      "grad_norm": 0.375,
+      "learning_rate": 7.429697183610862e-07,
+      "loss": 0.354,
+      "step": 1048
+    },
+    {
+      "epoch": 2.889807162534435,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.072115326711704e-07,
+      "loss": 0.3513,
+      "step": 1049
+    },
+    {
+      "epoch": 2.8925619834710745,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.723321454590092e-07,
+      "loss": 0.3429,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8953168044077136,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.383318654272774e-07,
+      "loss": 0.3525,
+      "step": 1051
+    },
+    {
+      "epoch": 2.8980716253443526,
+      "grad_norm": 0.375,
+      "learning_rate": 6.052109934980843e-07,
+      "loss": 0.3558,
+      "step": 1052
+    },
+    {
+      "epoch": 2.9008264462809916,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.729698228102653e-07,
+      "loss": 0.3477,
+      "step": 1053
+    },
+    {
+      "epoch": 2.903581267217631,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.416086387167951e-07,
+      "loss": 0.3367,
+      "step": 1054
+    },
+    {
+      "epoch": 2.90633608815427,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.111277187822339e-07,
+      "loss": 0.3476,
+      "step": 1055
+    },
+    {
+      "epoch": 2.909090909090909,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.815273327803182e-07,
+      "loss": 0.3449,
+      "step": 1056
+    },
+    {
+      "epoch": 2.911845730027548,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.5280774269154115e-07,
+      "loss": 0.3314,
+      "step": 1057
+    },
+    {
+      "epoch": 2.9146005509641872,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.2496920270085337e-07,
+      "loss": 0.3634,
+      "step": 1058
+    },
+    {
+      "epoch": 2.9173553719008263,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.9801195919541014e-07,
+      "loss": 0.3503,
+      "step": 1059
+    },
+    {
+      "epoch": 2.9201101928374653,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.719362507623614e-07,
+      "loss": 0.3404,
+      "step": 1060
+    },
+    {
+      "epoch": 2.922865013774105,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.467423081867649e-07,
+      "loss": 0.3349,
+      "step": 1061
+    },
+    {
+      "epoch": 2.925619834710744,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.224303544495766e-07,
+      "loss": 0.3448,
+      "step": 1062
+    },
+    {
+      "epoch": 2.928374655647383,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.990006047255967e-07,
+      "loss": 0.3464,
+      "step": 1063
+    },
+    {
+      "epoch": 2.931129476584022,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.764532663816266e-07,
+      "loss": 0.3602,
+      "step": 1064
+    },
+    {
+      "epoch": 2.9338842975206614,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.547885389746485e-07,
+      "loss": 0.3501,
+      "step": 1065
+    },
+    {
+      "epoch": 2.9366391184573004,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.3400661424998194e-07,
+      "loss": 0.3363,
+      "step": 1066
+    },
+    {
+      "epoch": 2.9393939393939394,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.141076761396521e-07,
+      "loss": 0.3449,
+      "step": 1067
+    },
+    {
+      "epoch": 2.9421487603305785,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.9509190076074657e-07,
+      "loss": 0.3475,
+      "step": 1068
+    },
+    {
+      "epoch": 2.9449035812672175,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.7695945641386102e-07,
+      "loss": 0.3474,
+      "step": 1069
+    },
+    {
+      "epoch": 2.9476584022038566,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.5971050358158936e-07,
+      "loss": 0.3409,
+      "step": 1070
+    },
+    {
+      "epoch": 2.950413223140496,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.4334519492711362e-07,
+      "loss": 0.3497,
+      "step": 1071
+    },
+    {
+      "epoch": 2.953168044077135,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.278636752928497e-07,
+      "loss": 0.3364,
+      "step": 1072
+    },
+    {
+      "epoch": 2.955922865013774,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.1326608169920372e-07,
+      "loss": 0.3427,
+      "step": 1073
+    },
+    {
+      "epoch": 2.958677685950413,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.955254334328423e-08,
+      "loss": 0.3459,
+      "step": 1074
+    },
+    {
+      "epoch": 2.9614325068870526,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.672318159782533e-08,
+      "loss": 0.3483,
+      "step": 1075
+    },
+    {
+      "epoch": 2.9641873278236917,
+      "grad_norm": 2.96875,
+      "learning_rate": 7.47781100100875e-08,
+      "loss": 0.441,
+      "step": 1076
+    },
+    {
+      "epoch": 2.9669421487603307,
+      "grad_norm": 0.359375,
+      "learning_rate": 6.37174343008251e-08,
+      "loss": 0.3425,
+      "step": 1077
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.354125236343155e-08,
+      "loss": 0.3562,
+      "step": 1078
+    },
+    {
+      "epoch": 2.9724517906336088,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.424965426298444e-08,
+      "loss": 0.3551,
+      "step": 1079
+    },
+    {
+      "epoch": 2.975206611570248,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.584272223546847e-08,
+      "loss": 0.3579,
+      "step": 1080
+    },
+    {
+      "epoch": 2.977961432506887,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.8320530687098166e-08,
+      "loss": 0.3478,
+      "step": 1081
+    },
+    {
+      "epoch": 2.9807162534435263,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.168314619359624e-08,
+      "loss": 0.3547,
+      "step": 1082
+    },
+    {
+      "epoch": 2.9834710743801653,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.593062749967178e-08,
+      "loss": 0.3494,
+      "step": 1083
+    },
+    {
+      "epoch": 2.9862258953168044,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.1063025518409653e-08,
+      "loss": 0.3469,
+      "step": 1084
+    },
+    {
+      "epoch": 2.9889807162534434,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.080383330915208e-09,
+      "loss": 0.3602,
+      "step": 1085
+    },
+    {
+      "epoch": 2.991735537190083,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.982736185859093e-09,
+      "loss": 0.345,
+      "step": 1086
+    },
+    {
+      "epoch": 2.994490358126722,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.7701114991997003e-09,
+      "loss": 0.3427,
+      "step": 1087
+    },
+    {
+      "epoch": 2.997245179063361,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.4252885396112163e-10,
+      "loss": 0.3469,
+      "step": 1088
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0,
+      "loss": 0.3415,
+      "step": 1089
+    },
+    {
+      "epoch": 3.0,
+      "eval_train_loss": 0.4371190667152405,
+      "eval_train_runtime": 8.4038,
+      "eval_train_samples_per_second": 160.284,
+      "eval_train_steps_per_second": 20.11,
+      "step": 1089
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1089,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4213817871368192e+17,
+  "train_batch_size": 60,
+  "trial_name": null,
+  "trial_params": null
+}